Methods in

Molecular Biology 1603

Paula Meleady Editor

Heterologous
Protein
Production
in CHO Cells
Methods and Protocols
Methods in Molecular Biology

Series Editor
John M. Walker
School of Life and Medical Sciences
University of Hertfordshire
Hatfield, Hertfordshire, AL10 9AB, UK

For further volumes:

http://www.springer.com/series/7651
 Heterologous Protein
Production in CHO Cells

Methods and Protocols

Edited by

Paula Meleady
National Institute for Cellular Biotechnology, Dublin City University, Dublin, Ireland
Editor
Paula Meleady
National Institute for Cellular Biotechnology
Dublin City University
Dublin, Ireland

ISSN 1064-3745     ISSN 1940-6029 (electronic)

Methods in Molecular Biology
ISBN 978-1-4939-6971-5    ISBN 978-1-4939-6972-2 (eBook)
DOI 10.1007/978-1-4939-6972-2

Library of Congress Control Number: 2017935545

© Springer Science+Business Media LLC 2017

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction
on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation,
computer software, or by similar or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and
regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to
be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty,
express or implied, with respect to the material contained herein or for any errors or omissions that may have been made.
The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Cover Illustration: The front cover image, kindly provided by Alan Costello (National Institute for Cellular Biotechnology,
Dublin City University), shows Chinese hamster ovary (CHO) cells with inducible green fluorescent protein (GFP)
expression (from Chapter 6).

Printed on acid-free paper

This Humana Press imprint is published by Springer Nature

The registered company is Springer Science+Business Media LLC
The registered company address is: 233 Spring Street, New York, NY 10013, U.S.A.
Preface

Since their introduction into the market over 20 years ago, biotherapeutics have consti-
tuted a large and growing percentage of the total pharmaceutical market, as well as approxi-
mately 25% of the R&D pipeline in industry. These biotherapeutics are having a huge
global impact on the treatment of challenging and previously untreatable chronic disease.
Currently biopharmaceuticals generate global revenues of $163 billion, making up about
20% of the pharma market, and predicted to grow to over $320 billion by 2020. The num-
ber of approved products in Europe and the USA has steadily increased to 2016 in 2014,
of which 37 have “blockbuster” status, i.e., sales over $1 billion per year, with monoclonal
antibodies (Mabs) representing the most lucrative single product class [1]. Most signifi-
cantly, nearly 50% of these biopharmaceutical products are produced in a single production
host, i.e., Chinese hamster ovary (CHO) cells. Improving the efficiency of production of
these biologics will be critical in controlling costs to healthcare systems as more of these
drugs come to market.
There has been considerable success in developing high-producing CHO cell culture
processes using approaches such as optimization of media formulation, improvements in
expression vector design, and also improvements in the design of bioreactors. The next
generation of improvements is expected to be made via genetic engineering of the host
(CHO) cell itself to increase or decrease the expression of endogenous genes depending on
the desired outcome, in order to improve the efficiency of the production of therapeutic
protein product. In order to enhance the production capabilities and efficiency of the host
cell line, an increased understanding of cellular physiology of CHO cells is of critical impor-
tance. There are substantial research efforts in progress focusing on the ‘omic analysis and
systems biology of CHO cells to understand CHO cell physiology. The publication of the
draft CHO-K1 genome in 2011 represented a major milestone in CHO systems biology.
This information has been supplemented further with the publication of draft genomes for
Chinese hamster and the CHO-S, CHO DG44 and CHO DXB11 cell lines. Availability of
the genome sequence will facilitate the interpretation and analysis of transcriptomic and
proteomic data to assess the physiological state of the cells under different growth and pro-
duction systems. Combining all levels of regulation through systems biology models will
unveil the underlying complexity inherent in CHO cell biology and will ultimately enhance
and accelerate CHO productive capabilities in the coming decades.
This book includes reviews and protocols for genetic manipulation of CHO cells for
recombinant protein production, including “difficult-to-express” therapeutics. A method is
also included on the use of the recently described genome editing tool, CRISPR/Cas9, and
how this can be applied to CHO cells. The book also includes a review and protocols for
characterization of CHO cells using ‘omic approaches and how these methods can be used
to improve efficiency of recombinant protein production during cell line development.
Analytical methods for characterization of recombinant protein product, such as glycosyl-
ation and host cell protein analysis, are also described in this book.

v
vi Preface

I am deeply grateful to all authors for giving up their valuable time and for contributing
to the book. I would also like to thank the series editor, Prof. John Walker, for help and
guidance during the process of getting the book to publication.

Dublin, Ireland Paula Meleady

Reference

1. Walsh G (2014) Biopharmaceutical benchmarks 2014. Nat Biotechnol 32(10):992–1000

Contents

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

  1 Strategies and Considerations for Improving Expression of “Difficult

to Express” Proteins in CHO Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Christina S. Alves and Terrence M. Dobrowsky
  2 Glycoengineering of CHO Cells to Improve Product Quality . . . . . . . . . . . . . . 25
Qiong Wang, Bojiao Yin, Cheng-Yu Chung, and Michael J. Betenbaugh
  3 Large-Scale Transient Transfection of Chinese Hamster Ovary Cells
in Suspension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Yashas Rajendra, Sowmya Balasubramanian, and David L. Hacker
  4 Cloning of Single-Chain Antibody Variants by Overlap-­Extension PCR
for Evaluation of Antibody Expression in Transient Gene Expression . . . . . . . . 57
Patrick Mayrhofer and Renate Kunert
  5 Anti-Apoptosis Engineering for Improved Protein Production
from CHO Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Eric Baek, Soo Min Noh, and Gyun Min Lee
  6 Conditional Knockdown of Endogenous MicroRNAs in CHO Cells
Using TET-ON-SanDI Sponge Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Alan Costello, Nga Lao, Martin Clynes, and Niall Barron
  7 Application of CRISPR/Cas9 Genome Editing to Improve
Recombinant Protein Production in CHO Cells . . . . . . . . . . . . . . . . . . . . . . . . 101
Lise Marie Grav, Karen Julie la Cour Karottki, Jae Seong Lee,
and Helene Faustrup Kildegaard
  8 Improved CHO Cell Line Stability and Recombinant Protein Expression
During Long-Term Culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Zeynep Betts and Alan J. Dickson
  9 Selection of High-Producing Clones Using FACS for CHO
Cell Line Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Clair Gallagher and Paul S. Kelly
10 The ‘Omics Revolution in CHO Biology: Roadmap to Improved
CHO Productivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Hussain Dahodwala and Susan T. Sharfstein
11 A Bioinformatics Pipeline for the Identification of CHO Cell Differential
Gene Expression from RNA-Seq Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
Craig Monger, Krishna Motheramgari, John McSharry, Niall Barron,
and Colin Clarke

vii
viii Contents

12 Filter-Aided Sample Preparation (FASP) for Improved Proteome

Analysis of Recombinant Chinese Hamster Ovary Cells . . . . . . . . . . . . . . . . . . 187
Orla Coleman, Michael Henry, Martin Clynes, and Paula Meleady
13 Phosphopeptide Enrichment and LC-MS/MS Analysis to Study the
Phosphoproteome of Recombinant Chinese Hamster Ovary Cells . . . . . . . . . . 195
Michael Henry, Orla Coleman, Prashant, Martin Clynes,
and Paula Meleady
14 Engineer Medium and Feed for Modulating N-Glycosylation
of Recombinant Protein Production in CHO Cell Culture . . . . . . . . . . . . . . . . 209
Yuzhou Fan, Helene Faustrup Kildegaard, and Mikael Rørdam Andersen
15 Glycosylation Analysis of Therapeutic Glycoproteins Produced
in CHO Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
Sara Carillo, Stefan Mittermayr, Amy Farrell, Simone Albrecht,
and Jonathan Bones
16 Characterization of Host Cell Proteins (HCPs) in CHO Cell Bioprocesses . . . . 243
Catherine E.M. Hogwood, Lesley M. Chiverton, and C. Mark Smales

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
Contributors

Simone Albrecht  •  National Institute for Bioprocessing Research and Training (NIBRT),

Dublin, Ireland
Christina S. Alves  •  Biogen Inc., Cambridge, MA, USA
Mikael Rørdam Andersen  •  Department of Systems Biology, Technical University
of Denmark, Kgs. Lyngby, Denmark
Eric Baek  •  Department of Biological Sciences, KAIST, Daejeon, Republic of Korea
Sowmya Balasubramanian  •  Laboratory of Cellular Biotechnology (LBTC), École
Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland
Niall Barron  •  National Institute for Cellular Biotechnology, Dublin City University,
Dublin, Ireland
Michael J. Betenbaugh  •  Department of Chemical and Biomolecular Engineering,
Johns Hopkins University, Baltimore, MD, USA
Zeynep Betts  •  Faculty of Science and Literature, Department of Biology, Kocaeli
University, Izmit, Kocaeli, Turkey
Jonathan Bones  •  National Institute for Bioprocessing Research and Training (NIBRT),
Dublin, Ireland
Sara Carillo  •  National Institute for Bioprocessing Research and Training (NIBRT),
Dublin, Ireland
Lesley M. Chiverton  •  Industrial Biotechnology Centre and School of Biosciences,
University of Kent, Canterbury, Kent, UK
Cheng-Yu Chung  •  Department of Chemical and Biomolecular Engineering,
Johns Hopkins University, Baltimore, MD, USA
Colin Clarke  •  National Institute for Bioprocessing Research and Training (NIBRT),
Dublin, Ireland
Martin Clynes  •  National Institute for Cellular Biotechnology, Dublin City University,
Dublin, Ireland
Orla Coleman  •  National Institute for Cellular Biotechnology, Dublin City University,
Dublin, Ireland
Alan Costello  •  National Institute for Cellular Biotechnology, Dublin City University,
Dublin, Ireland
Hussain Dahodwala  •  Vaccine production program (VPP), VRC/NIAID/NIH,
Gaithersburg, MD, USA; SUNY Polytechnic Institute, Albany, NY, USA
Alan J. Dickson  •  Faculty of Life Sciences, The University of Manchester, Manchester, UK
Terrence M. Dobrowsky  •  Biogen Inc., Cambridge, MA, USA
Yuzhou Fan  •  Department of Systems Biology, Technical University of Denmark, Kgs.
Lyngby, Denmark; The Novo Nordisk Foundation Center for Biosustainability, Technical
University of Denmark, Hørsholm, Denmark
Amy Farrell  •  National Institute for Bioprocessing Research and Training (NIBRT),
Dublin, Ireland
Clair Gallagher  •  National Institute for Cellular Biotechnology, Dublin City University,
Dublin, Ireland

ix
x Contributors

Lise Marie Grav  •  The Novo Nordisk Foundation Center for Biosustainability, Technical

University of Denmark, Hørsholm, Denmark
David L. Hacker  •  Laboratory of Cellular Biotechnology (LBTC), École Polytechnique
Fédérale de Lausanne (EPFL), Lausanne, Switzerland; Protein Expression Core Facility
(PECF), École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland
Michael Henry  •  National Institute for Cellular Biotechnology, Dublin City University,
Dublin, Ireland
Catherine E.M. Hogwood  •  Industrial Biotechnology Centre and School of Biosciences,
University of Kent, Canterbury, Kent, UK
Paul S. Kelly  •  National Institute for Cellular Biotechnology, Dublin City University,
Dublin, Ireland
Karen Julie la Cour Karottki  •  The Novo Nordisk Foundation Center for
Biosustainability, Technical University of Denmark, Lyngby, Denmark
Helene Faustrup Kildegaard  •  The Novo Nordisk Foundation Center for
Biosustainability, Technical University of Denmark, Lyngby, Denmark
Renate Kunert  •  Department of Biotechnology, Vienna Institute of BioTechnology,
University of Natural Resources and Life Sciences-Vienna, Vienna, Austria
Nga Lao  •  National Institute for Cellular Biotechnology, Dublin City University, Dublin,
Ireland
Gyun Min Lee  •  Department of Biological Sciences, KAIST, Daejeon, Republic of Korea
Jae Seong Lee  •  The Novo Nordisk Foundation Center for Biosustainability, Technical
University of Denmark, Lyngby, Denmark
Patrick Mayrhofer  •  Department of Biotechnology, Vienna Institute of BioTechnology,
University of Natural Resources and Life Sciences-Vienna, Vienna, Austria
John McSharry  •  National Institute for Bioprocessing Research and Training (NIBRT),
Dublin, Ireland
Paula Meleady  •  National Institute for Cellular Biotechnology, Dublin City University,
Dublin, Ireland
Stefan Mittermayr  •  National Institute for Bioprocessing Research and Training
(NIBRT), Dublin, Ireland
Craig Monger  •  National Institute for Bioprocessing Research and Training (NIBRT),
Dublin, Ireland; National Institute for Cellular Biotechnology, Dublin City University,
Dublin, Ireland
Krishna Motheramgari  •  National Institute for Bioprocessing Research and Training
(NIBRT), Dublin, Ireland; National Institute for Cellular Biotechnology, Dublin City
University, Dublin, Ireland
Soo Min Noh  •  Department of Biological Sciences, KAIST, Daejeon, Republic of Korea
Prashant  •  National Institute for Cellular Biotechnology, Dublin City University, Dublin,
Ireland
Yashas Rajendra  •  Laboratory of Cellular Biotechnology (LBTC), École Polytechnique Fédérale
de Lausanne (EPFL), Lausanne, Switzerland; Biotechnology Discovery Research, Lilly
Research Laboratories, Eli Lilly and Company, Lilly Corporate Center, Indianapolis, IN, USA
Susan T. Sharfstein  •  SUNY Polytechnic Institute, Albany, NY, USA
C. Mark Smales  •  Industrial Biotechnology Centre and School of Biosciences, University of
Kent, Canterbury, Kent, UK
Qiong Wang  •  Department of Chemical and Biomolecular Engineering, Johns Hopkins
University, Baltimore, MD, USA
Bojiao Yin  •  Department of Chemical and Biomolecular Engineering, Johns Hopkins
University, Baltimore, MD, USA
 Chapter 1

Strategies and Considerations for Improving Expression

of “Difficult to Express” Proteins in CHO Cells
Christina S. Alves and Terrence M. Dobrowsky

Abstract
Despite substantial advances in the field of mammalian expression, there are still proteins that are characterized
as difficult to express. Determining the expression bottleneck requires troubleshooting techniques specific
for the given molecule and host. The complex array of intracellular processes involved in protein expres-
sion includes transcription, protein folding, post-translation processing, and secretion. Challenges in any
of these steps could result in low protein expression, while the inherent properties of the molecule itself
may limit its production via mechanisms such as cytotoxicity or inherent instability. Strategies to identify
the rate-limiting step and subsequently improve expression and production are discussed here.

Key words Productivity, Difficult to express, Vector design, Cell engineering, Process optimization

1  Introduction

CHO cells have been utilized extensively for recombinant protein

expression; however, not all proteins are expressed at high levels in
this host system. There are many reasons why a protein may be
“difficult to express” and require an alternative strategy to standard
platform workflows for CHO cell production. Although there are
clearly monoclonal antibodies (mAb) that can be challenging to
express at industry standard levels of 5 g/L or more, productivity
improvements for non-mAb therapeutic proteins have lagged
behind [1, 2]. It is more difficult to define productivity levels that
constitute low expression for non-mAb products and a molecule
may be difficult to express not only because of intracellular chal-
lenges but also due to its biophysical properties. Determining the
expression bottleneck requires troubleshooting techniques specific
for the given molecule and has historically focused on transcrip-
tion, protein folding, post-translational processing, and secretion
[3–6]. Alternatively, challenges in producing unique or difficult to
express proteins may have solutions in the bioprocessing space
such as operating parameters or media and feed o ­ ptimization.

Paula Meleady (ed.), Heterologous Protein Production in CHO Cells: Methods and Protocols, Methods in Molecular Biology,
vol. 1603, DOI 10.1007/978-1-4939-6972-2_1, © Springer Science+Business Media LLC 2017

1
2 Christina S. Alves and Terrence M. Dobrowsky

Often, business, regulatory, or biological limitations may require

the introduction of additional process steps or modifications to
reach yield demands with an existing cell line. Previous bioprocess
strategies for improved protein production include chemically
induced specific productivity increases, affecting secreted protein
stability or toxicity in culture, continuous removal of protein from
culture, and general increases to culture biomass. The diversity
that exists among the different CHO host lineages can also be lev-
eraged to improve expression of problematic proteins and antibod-
ies [7, 8].
Engineering CHO cells and their production bioprocess to
better express difficult proteins requires a two-step approach by
which you first determine what is the rate-limiting function and
then develop a strategy to alleviate it. This chapter will outline vari-
ous strategies that can be used to determine the expression bottle-
neck and consequently to improve protein expression.

2  Strategy and Methods

2.1  Determining In order to design a comprehensive set of experiments to increase

the Bottleneck the production of a difficult to express protein, a fundamental
understanding of the biophysical and biochemical properties of the
protein is essential. The process by which a DNA sequence is con-
verted to a fully folded protein product is complex with steps that
include transcription, translation, post-translational modification,
protein folding, and ultimately secretion (Fig. 1). Any of these indi-
vidual steps could limit protein expression and may be attributed to
either a poorly designed molecule or suboptimal DNA coding
sequence. Inherent properties of the molecule can also result in the
protein being prone to degradation, aggregation, and other unfa-
vorable inter-protein interactions that can lead to cytotoxicity. Due
to these factors, knowledge of the biology of the protein will aid in
narrowing the scope and focus of troubleshooting efforts.

2.2  Integration Transfection of a gene into a cell is followed by the integration of

of the Gene of Interest that gene into the host cell’s genome. This has historically been a
random event in standard cell line development protocols. Efficient
expression of the transgene is highly dependent on both the num-
ber of gene copies integrated into the genome as well as the sites of
integration. The latter is greatly influenced by positional effect
variation, which is affected by the local permissiveness of the site as
well as the proximity and interaction with local and distal enhancers.
Although it has been shown that high copy numbers of transgenes
do not always correlate with high cellular productivity [9, 10], the
number of integrated transgenes is an important parameter to
measure as it has been shown to affect expression in some cell lines
[4, 11]. Several methods can be used to measure transgene copy
 Optimizing ‘Difficult to Express’ Proteins 3

DNA
Transcripon

NUCLEUS
RNA
CYTOPLASM Secreon
Translaon

Protein

Protein Folding

Post-
Translaonal
Modificaon

Fig. 1 Summary of protein synthesis. RNA is transcribed in the nucleus and then transported to the cytoplasm
and translated by the ribosomes. The proteins become bound to the rough ER, where they undergo folding and
processing before moving to the golgi. Soluble proteins undergo post-translational modifications and are sub-
sequently processed through the secretory pathway

number including analysis by southern blot [4], qPCR [11], and

digital droplet PCR [12]. In the situations where gene copy num-
ber does not correlate with product expression it is possibly a result
of the transgene being integrated in a suboptimal location in the
genome. Random genome insertion could result in the transgene
benefiting from a location in a highly transcribed genetic region of
euchromatin (referred to as a “hot spot”) or possibly suffering
from the effects of epigenetic gene silencing. By using a targeted
integration approach whereby the gene of interest is inserted into
a predetermined loci in the genome, one can circumvent such
issues by integrating the gene into a known hot spot where a region
of euchromatin and high gene-expression have already been estab-
lished. Determination of the desired hot spot is often the most
challenging part of developing a targeted integration system. One
approach is to utilize a screen of high expressing cell lines to deter-
mine whether expression is driven by a single copy of the trans-
gene. The location of the single integration site may be in an area
that naturally drives high expression and which can be utilized for
other genes of interest. Elaborate systems to determine permissive
loci for integration have been used such as transfecting CHO cells
with a plasmid containing a FRT-tag to specifically screen for single
integration loci with high transcriptional activity [13]. In this work,
4 Christina S. Alves and Terrence M. Dobrowsky

fluorescence in situ hybridization (FISH) was used to locate the

integration site of the FRT sequence or the antibody genes in the
chromosomes. Given the substantial advances in the field of CHO
‘omics, it is now feasible to use next generation sequencing (NGS)
to determine hot spots for integration. Advancements in this tech-
nology have increased the speed and throughput of whole genome
(DNA-seq) and transcriptome (RNA-Seq) sequencing such that it
is now feasible to screen clones for the location of genes that have
a high level of expression. A more refined method is targeted
sequencing where the genome is fragmented, incubated with
probes specific for the transgene, and then enriched via a wash
step. This enables sequencing of just the genes with high expres-
sion to elucidate their location in the euchromatin. Although it has
not yet been demonstrated, it may be possible to screen early in the
cell line development process and identify clones that display a pre-
defined ‘omics profile that is predictive of productivity using RNA-
seq [14].
Once a desired site has been elucidated, several methods exist
to insert a gene of interest into a specific location. These methods
which include site-specific recombinases, integrases, or transposases
for the integration of the expression cassettes are summarized in
Table 1. Integrases and transposases allow for multiple integrations
with higher copy numbers at various recognition sites within the
genome [15]. Phage integrases such as PhiC31 integrase rely on
unmodified, native acceptor (attP) and donor (attB) sites, but the
quantity of these sites in the CHO genome may be limiting. On
the other hand, site-specific recombinases have a higher specificity
of integration into a predetermined single site [16]. Flp recombi-
nases have been used in combination with Flp recognition target

Table 1
Methods for targeted integration

Method Benefits Disadvantages References

Integrases and Multiple integrations with higher
transposases copy numbers using native
donor and receptor sites
Site-specific Higher specificity of integration
recombinases (Flp, into a predetermined single site
Cre/LoxP)
TALENs Easy to design for knock in/out,
target DNA sequences using
proteins
CRISPR/Cas9 Target-specific DNA sequences
system using RNA, inexpensive, and
able to screen many sites quickly
Integrate randomly, limited [15]
number of sites in CHO
genome
Only can support one [18–20]
insertion
High frequency of insertion-­ [21]
deletion mutations,
expensive, and time
consuming to develop
Can have off target effects, [22, 23]
IP landscape is undefined
 Optimizing ‘Difficult to Express’ Proteins 5

sites (FRT) for targeted integration of transgenes into mammalian

cells with a high specificity of integration and low off target effects
[17]. This is accomplished either by using Flp-in or Flp recombinase-­
mediated cassette exchange (RMCE) strategies. RMCE uses a set
of hetero-specific FRT sites to direct a gene of interest to a prede-
termined and tagged locus that has been characterized to yield
high protein expression [18]. A binary RMCE expression system
has been used to co-express multiple proteins with different com-
binations of expression levels [17]. The Cre/loxP system for site-­
specific DNA recombination has also been used as a tool for
transgene integration in CHO cells [19]. Recent work has demon-
strated the ability to insert multiple transgenes into a targeted site
of the CHO cell genome using Cre recombinase-incorporating
integrase-defective retroviral vectors [20]. More recently, site-­
specific gene insertion in CHO cells has been performed using
transcription activator-like effector nucleases (TALENs) [21] or
CRISPR/Cas9 RNA-guided nucleases [22]. Precise insertion of a
gene expression cassette at a defined loci in CHO cells has been
accomplished using the CRISPR/Cas9 system following a simple
drug-selection methodology that resulted in homogeneous trans-
gene expression [23].

2.3  Transcription Traditionally, transcription has been considered the dominant fac-
tor in controlling protein production.
In a particular study to elucidate the mechanisms and processes-­
limiting gene expression in CHO cells, transcription appeared to
be the primary limitation for low- and medium-producing cell
lines, whereas in high-producing cell lines post-translational limita-
tions tended to dominate [6]. Within the process of transcription
the rate-limiting step is likely to be initiation. Due to the highly
condensed nature of DNA into chromatin structures, transcription
complexes often have trouble accessing certain regions of
DNA. Chromatin remodeling, the rearrangement of chromatin
structure by various remodeling complexes, is therefore required
for activation of transcription. Additionally, chromatin as well as
other proteins involved in transcriptional control can be altered by
methylation, acetylation, phosphorylation, and other modifica-
tions to affect whether a gene is active or inactive [24]. The syner-
gistic effect of these modifications, known as the “histone code,”
adds complexity in the form of epigenetic regulation of genes.
Specific methods to affect these features are detailed here.
Unstable protein expression has been observed in CHO cells
where mRNA decreases despite constant transgene copy numbers
[11], which suggests that either the mRNA is degrading or that the
promoter is being silenced. The expression level of mRNA tran-
script for the gene of interest can be determined by quantitative
real-time reverse transcription-PCR (qRT-PCR). There have been
mixed results on the correlation between productivity of a cell line
6 Christina S. Alves and Terrence M. Dobrowsky

and mRNA levels in CHO cells. Some studies showing that high
mRNA levels and gene copy numbers in methotrexate amplified
cells correspond to high specific productivities [4], whereas others
have seen no correlation between mRNA levels and expression
[25]. Despite these conflicting reports, it may still be valuable to
assess mRNA levels of the protein of interest to ensure that the
sequence of interest is being adequately transcribed. Additionally, in
the case of antibodies, using qPCR to determine mRNA levels can
be a useful diagnostic tool for determining the ratio of heavy to
light chain which may be important to ensure assembly of the mAb.
Studies have indicated that it is advantageous to have an excess of
light chain in relation to the heavy chain for optimal antibody pro-
duction [26, 27]. The ratio of heavy to light chain can be influ-
enced by optimizing the quantities of DNA transfected or by the
vector design. If a two-plasmid system is utilized, where the heavy
and light chains are located on different vectors, the mixture of
DNA used at the point of transfection can be used to modulate the
ratio. Alternatively, one can use a single-­plasmid system that employs
IRES-mediated bi- or tri-cistronic vectors that enable control of
heavy to light chain expression at different ratios.

2.3.1  Methylation DNA methylation has been reported to repress gene expression,
whereas hypomethylation of DNA in the promoter region can ele-
vate gene transcription activity [28, 29]. Enzymatic methylation of
cytosine at carbon 5 is well known as a fundamental epigenetic
mechanism that results in gene silencing [30]. DNA methylation
often occurs at CpG dinucleotides sites within promoter regions
which subsequently renders the promoter transcriptionally inactive.
Bisulfite treatment of DNA can be used to differentiate between
methylated and unmethylated CpG sites. In this method, sodium
bisulfite converts cytosine residues to uracil residues via deamination
at C4, while 5-methylcytosine remains unaffected [31]. Subsequent
amplification of the region by PCR allows for further analysis via
DNA sequencing [31] or microarray analysis [32]. Methylation-
specific real-time qPCR is a highly sensitive measurement of pro-
moter methylation and has been utilized to correlate hCMV-IE
methylation with unstable protein expression in recombinant CHO
cell lines [28]. Chemical compounds exist that can affect the degree
of DNA methylation, specifically a class of molecules known as DNA
methyltransferase inhibitors (iDNMTs). These compounds, which
include azacytidine, RG-108, and hydralazine, have been tested in
CHO cells for their capacity to increase cellular productivity in tran-
sient gene expression systems with some success [33].

2.3.2  Acetylation Acetylation of histones typically plays a role in transcriptional control

of active genes [34]. Histone acetyltransferases (HATs) and histone
deacetylases (HDACs) control the enhancement of transcription by
modifying histone acetylation. The most commonly used mechanism
 Optimizing ‘Difficult to Express’ Proteins 7

to control acetylation in CHO cell cultures is the use of HDAC

inhibitors to prevent deacetylation. Several studies have demonstrated
that sodium butyrate [35, 36] and/or valproic acid [37] can be used
to enhance mRNA transcription and increase specific productivities.
However, these compounds can also have adverse negative effects on
cell growth due to cytotoxicity and induction of apoptosis [38].
Their appropriateness must be evaluated to determine the optimal
concentration for enhanced productivity, but they are commonly
used over short production durations.

2.3.3  Vector Design The design of vectors to promote active transcription by creating a
Elements favorable chromatin environment around the transgene has been
extensively reviewed [39]. The available methods either alter the epi-
genetic environment of the DNA surrounding the transgene or pre-
vent the surrounding environment from affecting transcription of the
gene of interest [39]. A list of vector design elements for CHO cells
is shown in Table 2. In order to enhance gene transcription and
reduce transgene expression dependence on the surrounding chro-
matin, strong cellular enhancers such as the Locus Control Region
(LCR) have been utilized [40]. The LCR is a cis-­acting DNA ele-
ment that controls the expression of human β-globin locus genes.
Unfortunately, these enhancers only function in certain cell lines and
cannot be used as general regulatory elements in all mammalian cells.

Table 2
Vector design elements to enhance transcription

Category Specific Elements Description Reference

Locus control human β-globin, Can lead to stable high expression of [40]
regions HGH transgene in a copy number dependent
manner. Limited usefulness in mammalian
cell lines
Insulators cHS4 Block the positive action of enhancers, can [41]
modestly increase transgene expression in
CHO, but may not be universally effective
Matrix Attachment Chicken lysozyme Bind to the nuclear matrix and affect the [42, 43]
Regions (MARs) MAR, human arrangement of chromatin into loops.
β-globin MAR, X Have shown some positive effects on
MAR transgene expression in CHO
Ubiquitous HNRPA2B1, CBX3, Derived from promoters of housekeeping [41, 44]
Chromatin TBP and PSNB1 genes that are transcriptionally active.
Opening Elements Large increases in gene expression were
(UCOEs) observed but elements are typically large
(~16 kb), promoter dependent
Antirepressor or STAR7, STAR44, Small elements (< 2 kb) that block [45, 46]
STAR (stabilizing STAR67 chromatin-associated repressors. Convey
and antirepressor) copy number-dependent stable expression
8 Christina S. Alves and Terrence M. Dobrowsky

Regulatory elements that block interactions between the enhancer

and promoter while not directly affecting their individual activity are
referred to as insulators. Insulators such as the chicken β-globin 5′
hypersensitive site 4 (cHS4) have been used to control the effects of
the surrounding chromatin environment on the transgene [41].
Matrix-attachment regions (MARs) are DNA elements that
bind to the nuclear matrix and are believed to influence gene
expression by affecting the arrangement of chromatin into loops.
MARs, such as chicken lysozyme MAR, human β-globin MAR,
and X MAR, can associate with euchromatin and act as boundary
or insulator elements, and hence create an independent chromatin
structure from the surroundings [42]. Although the specific mech-
anisms by which MARs function in the cell are not entirely under-
stood, they have been effective in enhancing the expression of
target proteins in mammalian cell cultures. MARs can be integrated
into expression vectors that may increase the percentage of high-­
producer cells in a population to reduce the number of clones that
need to be screened. Protocols are available that describe how to
incorporate MARs into vectors that can then be transfected into
CHO cells for increased transgene expression [43].
Other elements that have been shown to protect a transgene from
silencing and convey higher transgene expression are ubiquitous chro-
matin opening elements (UCOEs), which are derived from the pro-
moters of housekeeping genes that are typically transcriptionally active
[44]. Some well-characterized UCOE pairs include HNRPA2B1 and
CBX3 or TBP and PSNB1, which are DNA regions that contain a pair
of divergent gene promoters that are transcriptionally active in all cells
of an organism. Large UCOEs of up to 16 kb have been used to gen-
erate high-level and stable transgene expression for cells in extended
culture by increasing the efficiency of the CMV promoter. Because
UCOEs directly affect transcriptional regulation that is dependent on
the promoter and its activity, these elements have variable effects on
expression of a target protein in CHO cells and need to be tested for
specific host and vector strategies [41].
Antirepressor or STAR (stabilizing and antirepressor) are DNA
elements that block chromatin-associated repressors and have been
used to flank transgenes in mammalian expression vectors. These
elements affect the spread of methylation and histone deacetylation
from the adjacent chromatin environment into the transgene
region. They can enhance protein expression as well as overcome
genetic instability caused by positional effects, epigenetic silencing,
or loss of gene copy number [45]. The positive effects of STAR
elements are most pronounced when high selection stringency is
used to develop stable clones in CHO cells [46].

2.4  Translation The process of translation consists of initiation, elongation, termi-

nation, and recycling. The initiation of mRNA translation is an
essential precursory step that influences cell growth and protein
 Optimizing ‘Difficult to Express’ Proteins 9

synthesis via the coordination of numerous initiation factors [47].

The secondary structure of the mRNA can affect translational effi-
ciency. Formation of a closed loop structure consisting of mRNA,
a number of eukaryotic initiation factors (eIFs), and ribosomal
proteins can potentially increase global translation efficiency by
promoting re-initiation of translation. High-producing cell lines
have been shown to maintain appropriate levels of these translation
initiation factors [48]. Use of cell engineering approaches to main-
tain the levels of these initiation factors may allow for generation of
new host cell lines with high growth and recombinant protein pro-
ductivity. Another target to improve translation is the global meta-
bolic sensor and processing protein mammalian target of rapamycin
(mTOR). The treatment of CHO cell cultures with adenosine
results in growth arrest but also increases productivity. The ade-
nosine contributes to high ATP levels which increase mTOR activ-
ity, inhibiting the key translation initiation repressor 4E-BP1 [49].
mTOR has also been shown to influence ribosomal protein synthe-
sis, translation initiation, and translation elongation in addition to
other cellular functions. Its overexpression in CHO cells has
resulted in increased specific antibody productivity [50] making it
an attractive engineering target for difficult to express proteins.

2.4.1  Codon Optimization Because it is often the case that human proteins are being express-
ing in CHO cells and synonymous codons are used with different
frequencies in different organisms (known as codon bias) [51], it is
important to ensure that the transgene sequence is optimized. By
optimizing the DNA sequences for expression in CHO cells, one
can ensure that certain preferred codons are translated more accu-
rately and/or efficiently. Poorly optimized sequences can adversely
affect protein translation, and subsequently protein expression, by
preventing the host from efficiently translating the rare codons.
Codon optimization has been used to increase protein expression
in multiple studies [52, 53] and there are several websites and ser-
vices that will perform codon optimization for expression in CHO
cells of a given amino acid sequence. A list of codon usage for
CHO cells is shown in Table 3.
Another important consideration is the translation initiation
sequence located upstream of the start codon (AUG). The efficient
consensus sequence GCCACC(AUG)G, known as the Kozak
sequence [54], yields high fidelity and efficiency of initiation and is
typically used at the start of the coding sequence.

2.4.2  Splice Sites Splice sites are located between an exon and an intron. The splice
site upstream of an intron is referred to as the donor splice site
(5′–3′ direction), while the one downstream of an intron is
the acceptor splice site (3′–5′ direction). The acceptor splice site
corresponds to the end of an intron (AG) and the donor splice site
corresponds to the beginning of an intron (GT). Splice sites can
10 Christina S. Alves and Terrence M. Dobrowsky

Table 3
Codon usage in Chinese hamster genes

Amino Relative Amino Relative Amino Relative

acid. Codon frequency acid Codon frequency acid Codon frequency
Ala GCT 22.4 His CAT 10.2 Ser TCA 10.3
GCA 16.3 CAC 12.9 AGT 11.4
GCC 25.9 Leu TTG 14.1 TCC 16.5
GCG 5.0 CTC 18.4 AGC 16.4
Arg AGA 10.1 CTG 38.8 TCT 16.0
CGA 7.2 CTA 7.6 TCG 3.4
CGG 10.1 CTT 13.2 Thr ACT 14.1
AGG 10.2 TTA 6.4 ACA 15.7
CGC 9.3 Ile ATT 17.4 ACC 20.3
CGT 5.6 ATC 24.8 ACG 4.5
Asn AAT 17.4 ATA 6.9 Trp TGG 13.1
AAC 21.2 Lys AAG 38.4 Tyr TAT 13.1
Asp GAT 24.6 AAA 24.6 TAC 16.4
GAC 28.1 Met ATG 23.0 Val GTA 7.8
Cys TGT 9.1 Phe TTC 22.0 GTT 11.6
TGC 10.3 TTT 19.6 GTG 30.1
Gln CAA 10.3 Pro CCA 15.7 GTC 15.7
CAG 33.4 CCC 17.0 Stop TGA 1.2
Glu GAA 28.4 CCT 16.7 TAA 0.6
GAG 41.1 CCG 4.3 TAG 0.5
Gly GGA 15.8
GGG 13.4
GGT 12.8
GGC 21.3
The amino acid abbreviation is shown adjacent to the codon and the relative frequency in identified genes of the
Chinese hamster (Cricetulus griseus). The source of these data is http://www.kazusa.or.jp/codon/. These records were
a snapshot of usage as of March 2016. A total of 331 genes and 153,527 codons contributed to this data set.

also unintentionally exist in a coding sequence. As possible accep-

tor and donor splice sites, every AG and GT in a DNA sequence
needs to be evaluated as either a real splice site or a pseudo splice
site to ensure that the sequence is not compromised during transla-
tion. In addition to the sequences immediately adjacent to the
 Optimizing ‘Difficult to Express’ Proteins 11

Table 4
Programs available to identify potential splice sites in a DNA sequence

Program Website Reference

Gene splicer https://ccb.jhu.edu/software/genesplicer/ [107]
NetGene2 http://genome.cbs.dtu.dk/services/NetGene2/ [108]
HSPL http://genomic.sanger.ac.uk/ [109]
NNSplice http://www.fruitfly.org/seq_tools/splice.html [110]
GENIO splice site and http://biogenio.com/splice/
exon predictor
SpliceView http://l25.itba.mi.cnr.it/~webgene/wwwspliceview.html [111]

splice event, distal sequences also contribute to the probability of

splicing. Several programs that are summarized in Table 4 exist
online to help evaluate a sequence and the probability that donor
and acceptor splice sites are present.

2.5  Protein Folding Proper protein folding is essential for adequate expression of a
and Processing molecule. The ER is responsible for ensuring that proteins are
properly processed and folded and as such there are specific quality
control systems to aid in the efficiency of folding and eliminate
misfolded proteins. When a protein is misfolded in the ER it is
proteolytically destroyed via the ER-associated degradation
(ERAD) pathway. Similarly, the unfolded protein response (UPR),
a signal cascade that protects cells from aggregated protein by
restoring ER function, can be triggered by intracellular accumula-
tion of misfolded protein. Several chaperones and cofactors are
involved in the process of protein folding and assembly and can be
modulated to enhance protein expression.

2.5.1  Chaperones Molecular chaperones are proteins that assist the folding and assem-
bly of intracellular proteins which may be good targets for cellular
engineering to improve protein expression. Heat shock proteins
(HSPs) function as molecular chaperones and are ­primarily respon-
sible for protein folding, assembly, translocation, and degradation
under cellular stress. Chaperones also prevent newly synthesized
polypeptide chains from aggregating into defective proteins. BiP is
a HSP70 molecular chaperone that binds newly synthesized pro-
teins as they are translocated into the ER, and preserves them in a
state suitable for subsequent folding. Protein disulfide isomerase
(PDI) is an enzyme in the ER that catalyzes the formation and
breakage of disulfide bonds to assist in protein folding [55].
Cyclophilin B (CypB) interacts with other proteins in the ER
including BIP and PDI to form chaperone complexes that facilitate
protein folding. Some work has been done on expressing molecular
12 Christina S. Alves and Terrence M. Dobrowsky

chaperones in CHO cells to improve productivity of difficult to

express proteins. Specifically, co-expression of CypB with a difficult
to express antibody improved cell growth but had no effect on cell
specific productivity whereas, co-expression with BIP increased the
productivity but reduced cell growth [56]. In another CHO cell
line expressing a fusion protein, co-transfection of CypB followed
by addition of chemical chaperones at the start of stationary phase
increased cell-specific production and eliminated protein aggrega-
tion [57]. The disparity in these findings suggests that the engineer-
ing of molecular chaperones for increased protein expression may
be product and cell line dependent.
Chemical chaperones are a group of compounds that can improve
the folding capacity of the ER, facilitate protein folding in the ER,
and enhance the secretion of protein. Chaperones can be added to
cell culture to potentially improve expression of a difficult to express
protein, especially if misfolding or aggregation is occurring intracel-
lularly. This approach provides a simpler alternative to overexpression
of molecular chaperones given their uncertain effect on productivity.
PBA (Sodium 4-phenylbutyrate) has been used to promote the secre-
tion of a mutated protein C from CHO cells by utilizing an uncon-
ventional GRASP55-dependent pathway that restores normal
intracellular trafficking through the ER and golgi [58]. Treatment of
mammalian cells with PBA has been shown to suppress ER stress by
chemically enhancing the ER capacity to cope with the expression of
misfolded protein, ­preventing intracellular aggregates by facilitating
protein degradation [59]. Osmotically active chaperones such as
DMSO, glycerol, and proline have been used to increase specific pro-
ductivity, but also can have a negative impact on cell growth.
Additionally, DMSO and proline can reduce protein aggregate for-
mation in culture supernatants by an undefined mechanism [60]. To
counteract the negative effect on cell growth and viability, the addi-
tion of DMSO [61] and glycerol [62] in a two-staged approach has
been utilized to increase the specific productivity of CHO cell lines.
Similarly, a combination of PBA and glycerol has been added at the
start of stationary phase alongside expression of the molecular chap-
erone CypB to maximize cell specific production and eliminate pro-
tein aggregation [57]. Analogous to the overexpression of molecular
chaperones, the effect of these chemicals may be cell line and protein
specific and their suppression of cell growth requires that their con-
centration and dosing strategy be carefully considered.

2.5.2  Bioprocess In addition to enabling or improving chaperone protein function,

Modifications aggregation can be prevented through bioprocess modifications.
Altering the cellular redox potential by supplementing media with
the antioxidant glutathione can reduce aggregation [63], while
media optimization of components such as cysteine or glycerol can
reduce aggregation [64, 65]. Additionally, a reduction in tempera-
ture has been shown to reduce aggregation and positively affect
 Optimizing ‘Difficult to Express’ Proteins 13

protein processing by reducing the mis-/unfolded protein des-

tined for degradation via the ERAD pathway [64–66].
While protein folding is typically a primary driver for proper
function, other post-translational modifications in CHO cultures
that could affect protein structure and function include glycosyl-
ation, oxidation of methionine, deamidation of asparagine and glu-
tamine, hydroxylation, and sulfation [67]. It is possible that in
trying to achieve the specific ranges or combinations of these mod-
ifications, the expression of the protein may be compromised.
The addition of sodium butyrate has been used regularly
used to improve specific protein production, but it has also been
shown to affect post-translational modifications of the protein of
interest by altering histone modification, chaperones, lipid
metabolism, and protein processing [68]. These changes in post-
translational modifications can result in increased microheteroge-
neity and reduced sialylation which may decrease in vivo activity
[69]. However, a combination of sodium butyrate treatment with
lower production culture temperatures has been shown to miti-
gate these risks [70].

2.6  Secretion Translocation of a nascent protein from ribosomes through the

cytosol into the endoplasmic reticulum is mediated by its signal
peptide and is an essential stage in protein secretion. The efficient
secretion of recombinant proteins from CHO cells is strongly
dependent on the signal peptide used, which makes identifying the
optimal signal sequence for each target protein an important step
in maximizing the efficiency of protein secretion [71]. In mam-
malian cells, a signal peptide that ranges from 5 to 30 amino acids
at the N-terminal end of nascent proteins is recognized by the sig-
nal recognition particle (SRP) in the cytosol as the protein is being
synthesized on the ribosome. The SRP then transfers the complex
consisting of the SRP and ribosome-nascent chain to a receptor on
the endoplasmic reticulum (ER) membrane, where it is eventually
translocated to the lumen of the ER and the signal peptide is
cleaved by a signal peptide peptidase [72]. The translocation of
proteins into the ER lumen is considered a bottleneck of the secre-
tory pathway and has motivated further investigation into enhanc-
ing the capacity of signal peptides for recombinant protein
expression. Several studies have shown positive effects with native
signal peptides, natural signal peptides derived from human albu-
min and human azurocidin, as well as optimized signal sequences
[71, 73, 74] indicating the importance of carefully evaluating sig-
nal sequences for the expression of a given protein. There are sev-
eral resources online (Table 5) that assess the probability that a
peptide is a suitable signal sequence as well as how efficiently the
sequence will be cleaved from the protein [75, 76].
Secretion of antibodies has been affected by improper cleavage
of the light chain from the signal peptide due to a dysfunctional SRP
14 Christina S. Alves and Terrence M. Dobrowsky

Table 5
Websites that offer free signal peptide prediction algorithms
SignalP 4.1 Server (http://www.cbs.dtu.dk/services/SignalP/)
PrediSi: Prediction of Signal peptides (http://www.predisi.de/)
Signal-BLAST Signal Peptide Prediction (http://sigpep.services.came.
sbg.ac.at/signalblast.html)

complex, which results in its precipitation in an insoluble cellular

fraction [3]. Western blotting of intracellular fractions can be used to
determine whether light chain is precipitating in the cells. This can
be achieved by using standard lysis techniques to create protein
extracts, followed by blotting with light chain-specific antibodies. If
this inadequate cleavage of the light chain is affecting expression,
proper processing and secretion can be restored by over-expressing
SRP proteins such as the signal recognition protein, SRP14 [3].

2.6.1  Russell Bodies Russell bodies are intracellular aggregates of immunoglobulins

stored in the endoplasmic reticulum that can form during protein
biosynthesis. The formation of Russell bodies depends on the phys-
iochemical properties of the protein coded by the variable regions of
the heavy and light chains as well as extrinsic factors such as stressful
cell culture conditions [77]. Immunofluorescent microscopy can be
used to determine whether Russell bodies are forming intracellularly
for a given protein. Cells must be fixed, permeabilized, and stained
with fluorescently conjugated antibodies to the IgG of interest fol-
lowed by fluorescent imaging and quantification [78]. A frequency
of Russell body phenotype can be calculated by determining the
number of Russell bodies observed and normalizing to the overall
number of cells in the image. This value can be compared to an
alternative cell line that produces an easy to express molecule to
determine whether intracellular aggregation is resulting in reduced
protein expression. Recent studies suggest that there are IgG anti-
body sequences with intrinsically high condensation/aggregation
propensities that are more prone to form Russell bodies in the ER
lumen [78]. This implies that if a protein is not being expressed due
to the formation of these intracellular aggregates the sequence may
need to be altered to enable better expression.

2.7  Protein Toxicity Another mechanism that leads to insufficient protein production is
the inherent toxicity of the protein being expressed. Limiting cel-
lular exposure to high concentrations of the protein or adapting
cell lines specifically to be resistant to the toxic protein can improve
growth and subsequently yield. Toxicity of the protein of interest
may diminish a cell’s ability to recover after transfection and selec-
tion. Ultimately, this toxicity will result in the unintentional selec-
tion of low-producing cells from the population. A dose-response
 Optimizing ‘Difficult to Express’ Proteins 15

study with purified protein and host cells is often the most direct
determination of cytotoxicity, wherein the toxicity of the purified
protein and buffer to the naive culture is assessed. If the quantity
of purified protein is limiting, similar results can be confirmed by a
less ideal but easy-to-execute experiment utilizing spent media
from a transient transfection. Clarified culture supernatant from a
transiently transfected culture will likely contain a toxic level of
protein but sufficient unprocessed metabolites for continued
growth. Performing a dose-response study using this supernatant
incrementally blended with fresh media may yield similar results to
dosing purified protein. However, toxicity will be confounded
with other supernatant components such as metabolic waste prod-
ucts and the highest concentration available for testing will be
limited.
Protein toxicity can be mitigated by multiple methods. One
approach, likely the most extreme, is to alter the protein itself to
reduce its cytotoxicity. Modifying the protein to produce a more
stable, less toxic form while retaining its intended biological func-
tion can be difficult but possible with extensive knowledge of the
structure function relationship [79]. Introducing stabilizing agents
that can be degraded later can be effective as long as a more com-
plicated downstream purification process is acceptable. N-terminal
tags such as a small ubiquitin-like modifier (SUMO) can be used to
create “dormant fusion” proteins with decreased toxicity that are
capable of being cleaved downstream [80]. An alternative option is
to modify the promoter in the transfected vector rather than the
protein of interest. When the optimum environment for cell
growth varies significantly from that of protein production, an
inducible expression system may be appropriate. An inducible
expression system can enable high cell densities to be achieved
prior to protein production and subsequently alleviate the effects
of toxicity or degradation [81]. Inducible promoter systems are
commercially available [82] for direct implementation. In general,
it can be difficult to ensure that transgene expression is entirely
inhibited prior to the addition of the inducing agent [83].
Therefore, most industrially relevant systems utilize promotor-
transactivator combinations. In these systems, the activity of a con-
stitutively expressed transactivator is controlled via supplementation
of some complementary ligand [84]. The tetracycline (Tet) induc-
ible system, often referred to as Tet-on, allows for the expression of
the protein of interest in the presence of tetracycline. Alternatively,
protein expression could be repressed in the presence of tetracy-
cline, the Tet-off system, and activated by complete media
exchange. The Tet-on system is often preferred for recombinant
protein production as it is relatively straightforward to supplement
culture with tetracycline while removing it would require signifi-
cant liquid handling at scale [85, 86]. Other applications for induc-
ible systems include increasing specific productivity by arresting
16 Christina S. Alves and Terrence M. Dobrowsky

cellular proliferation [87]. In this case, protein production is not

induced directly, but rather through the subsequent reactions to
decreased cell growth such as increased mitochondrial mass and
activity. The selection of stable, clonal cell lines is more involved
with inducible systems and may increase development time.
Cell cultures can often be adapted to new growth environ-
ments to suit productivity needs. Adapting cultures from serum
containing to serum free, chemically defined medium for example
or from adherent to suspension growth environments are common­
place with commercially available materials and protocols [88, 89].
If the protein of interest is determined to be toxic, one option for
improving production is adapting host cells. Naive host cells cul-
tured in the presence of low concentrations of the protein can
decrease sensitivity of the cell line over time. The resulting host
culture can then be transfected and ideally recover with the ability
to survive higher productivity than before adaptation [8].
Concerns over protein toxicity or degradation may also be
mitigated by reducing contact of the protein with producing cul-
ture. This can be performed through chemical supplementation,
wherein the toxic protein is competitively inhibited from interact-
ing with the cells through an antagonist [8]. Alternatively, this can
be achieved through the use of perfusion growth systems where
culture supernatant is removed continuously via filtration and cell
mass is retained. Perfusion growth systems are also commercially
available and enable high cell densities and increased volumetric
productivity [90, 91]. However, these can be technically challeng-
ing to implement even at smaller scales. If small protein require-
ments are desired for non-industrial purposes, complete media
exchange through centrifugation is a simple alternative [92].

2.8  Protein-Cell In addition to a protein’s ability to limit cell culture growth, the
Adhesion cell culture may limit the availability or stability of protein in the
and Consumption supernatant. Adhesion to lipid head groups or lipoproteins in the
cell membrane can accelerate protein degradation and reabsorp-
tion into the producing cells themselves. One example of such pro-
tein loss is the generation of recombinant Factor VIII (rFVIII)
[93]. Here, 90% of the secreted rFVIII in serum-free conditions
were determined to be bound to cells. This effect was limited by
supplementing culture with a complimentary protein, von
Willibrand Factor (vWF) capable of competitive inhibition [93].
Co-expressing vWF, rather than supplementation, was also found
to have a stabilizing effect [94]. Co-expression of a complex pro-
tein antagonist is not typically the most efficient mitigation strat-
egy and can complicate cell line selection. Chemical inhibition may
be possible and substantially easier to implement. In this current
example, rFVIII can be prevented from binding to host cell culture
by supplementation of o-phospho-l-serine (OPLS) [93, 95].
Implementation of these methods will require specific knowledge
 Optimizing ‘Difficult to Express’ Proteins 17

of the protein being produced, its likely binding partners, and their
protein sequence or commercial availability.

2.9  Effects The bioprocess applied to protein expressing cultures will affect
of Bioprocessing the amount and quality of the product. Cell culture parameters
on Protein Expression known to have an effect on specific productivity include pCO2,
osmolality, temperature, dissolved oxygen, and pH [96, 97]. In
addition to specific productivity, the culture environment has been
shown to have a direct effect on post-transcriptional modifications
and impurity profiles [66]. The effect of certain parameters can be
cell line and protein specific which makes general recommenda-
tions difficult. Indeed, process settings can have alternate effects
depending on how other parameters are controlled. For example,
pH control can alter specific productivity of erythropoietin (EPO)
producing culture at one temperature differently than another
[98]. Because these process parameter sensitivities will often be cell
specific they are difficult to predict without in-depth experience
with the cell line of interest.
Defining a well-understood operating space for all process
parameters using Design of Experiment (DOE) methodologies is
recommended and may reveal configurations that allow for suffi-
cient protein expression [99, 100]. Alternatively, shifts in process
parameters during production to temporarily increase productivity
can be useful [101]. While this often results in decreased overall
cell mass, that in turn may be compensated for by maintaining high
process temperature during the growth phase of the culture. A net
increase in productivity with lower temperatures may be the result
of either increased specific productivity or decreased protein deg-
radation at reduced temperatures. Growth arrest and subsequent
increases in specific productivity can also be accomplished through
hyperosmolality or chemical treatment with agents such as sodium
butyrate or DMSO (as discussed earlier) [101, 102]. Addition of a
small molecule inhibitor of cyclin-dependent kinases (CDK) 4/6
mid-way through production can mediate G0/G1 growth arrest
without impacting G2/M phase. This resulted in sustained cell
mass and increased specific productivity without negatively impact-
ing product quality [103]. While improvements in specific produc-
tivity may be obtained through process condition shifts or media
supplementation, the medium formulation itself may be optimized
for a more consistent increase to protein production [104, 105].
Also, temporary increases to specific productivity will likely increase
product titer before cell death becomes a concern, it usually comes
at the expense of product quality such as increased heterogeneity,
decreased biological activity, altered acidic isoforms, and inconsis-
tent sialylated species [69]. Modifications to your bioprocess can
reduce protein loss by limiting these mechanisms. Adjusting your
medium formulation, operating parameters, or harvest procedure
can significantly increase your product yield [106].
18 Christina S. Alves and Terrence M. Dobrowsky

3  Summary

Determining the source of low protein production can be chal-

lenging given the large number of potential causes. Successful
troubleshooting of low productivity requires a fundamental
­understanding of the protein as well as a methodical approach to
investigating the inhibiting factors. In the end, there are three fun-
damental approaches to maximizing a difficult to express protein:
host cell engineering, improved vector design, and the optimiza-
tion of the cell culture process. This chapter has covered many, but
not all of the currently known methods to improve protein expres-
sion in CHO cells. A single solution may result in improved expres-
sion or alternatively, a protein may require a synergistic approach
where multiple strategies are combined to ultimately increase pro-
ductivity. In either case, the solution may be both cell line and
protein specific, thus requiring a well-designed set of experiments
to discover and relieve the bottleneck.

References
1. Ohya T, Hayashi T, Kiyama E et al (2008)
Improved production of recombinant human
antithrombin III in Chinese hamster ovary cells
by ATF4 overexpression. Biotechnol Bioeng
100:317–324. doi:10.1002/bit.21758
2. Novo JB, Morganti L, Moro AM et al (2012)
Generation of a Chinese hamster ovary cell line
producing recombinant human glucocerebrosi-
dase. J Biomed Biotechnol 2012:875383.
doi:10.1155/2012/875383
3. Le Fourn V, Girod P-A, Buceta M et al (2014)
CHO cell engineering to prevent polypeptide
aggregation and improve therapeutic protein
secretion. Metab Eng 21:91–102.
doi:10.1016/j.ymben.2012.12.003
4. Jiang Z, Huang Y, Sharfstein ST (2006)
Regulation of recombinant monoclonal anti-
body production in chinese hamster ovary
cells: a comparative study of gene copy num-
ber, mRNA level, and protein expression.
Biotechnol Prog 22:313–318
5. Nishimiya D, Mano T, Miyadai K et al
(2013) Overexpression of CHOP alone and
in combination with chaperones is effective
in improving antibody production in mam-
malian cells. Appl Microbiol Biotechnol
97:2531–2539. doi:10.1007/s00253-012-
4365-9
6. Mead EJ, Chiverton LM, Smales CM, von
der Haar T (2009) Identification of the limi-
tations on recombinant gene expression in
CHO cell lines with varying luciferase pro-
duction rates. Biotechnol Bioeng 102:1593–
1602. doi:10.1002/bit.22201
7. Hu Z, Guo D, Yip SSM et al (2013) Chinese
hamster ovary K1 host cell enables stable cell
line development for antibody molecules which
are difficult to express in DUXB11-­ derived
dihydrofolate reductase deficient host cell.
Biotechnol Prog 29:980–985. doi:10.1002/
btpr.1730
8. Alves C, Gilbert A, Dalvi S et al (2015)
Integration of cell line and process develop-
ment to overcome the challenge of a difficult
to express protein. Biotechnol Prog:1–11.
doi:10.1002/btpr.2091
9. Lattenmayer C, Trummer E, Schriebl K et al
(2007) Characterisation of recombinant
CHO cell lines by investigation of protein
productivities and genetic parameters.
J Biotechnol 128:716–725. doi:10.1016/j.
jbiotec.2006.12.016
10. Reisinger H, Steinfellner W, Stern B et al
(2008) The absence of effect of gene copy
number and mRNA level on the amount of
mAb secretion from mammalian cells. Appl
Microbiol Biotechnol 81:701–710.
doi:10.1007/s00253-008-1701-1
11. Chusainow J, Yang YS, Yeo JHM et al (2009)
A study of monoclonal antibody-producing
CHO cell lines: what makes a stable high pro-
ducer? Biotechnol Bioeng 102:1182–1196.
doi:10.1002/bit.22158
12. Hindson BJ, Ness KD, Masquelier D a., et al.
(2011) High-throughput droplet digital PCR
system for absolute quantitation of DNA copy
number. Anal Chem 83:8604–8610. doi:
10.1021/ac202028g
 Optimizing ‘Difficult to Express’ Proteins 19

13. Huang Y, Li Y, Wang YG et al (2007) An effi- 25. Reinhart D, Sommeregger W, Debreczeny M
cient and targeted gene integration system for et al (2013) Characterization of recombinant
high-level antibody expression. J Immunol IgA producing CHO cell lines by qPCR. BMC
Methods 322:28–39. doi:10.1016/j.jim.2007. Proc 7:P114. doi:10.1186/1753-6561-7-S6-
01.022 P114
14. Wright C, Estes SD (2014) Pharmaceutical 26. Schlatter S, Stansfield SH, Dinnis DM et al
next-generation bioprocess: an industry per- (2005) On the Optimal Ratio of Heavy to
spective of how the ‘ omics era will affect Light Chain Genes for Efficient Recombinant
future biotherapeutic development. Pharm Antibody Production by CHO Cells.
Bioprocess 2:371–375 Biotechnol Prog 21:122–133
15. Büssow K (2015) Stable mammalian pro- 27. Ho SCL, Koh EYC, van Beers M et al (2013)
ducer cell lines for structural biology. Curr Control of IgG LC:HC ratio in stably trans-
Opin Struct Biol 32:81–90. doi:10.1016/j. fected CHO cells and study of the impact on
sbi.2015.03.002 expression, aggregation, glycosylation and con-
16. Turan S, Galla M, Ernst E et al (2011) formational stability. J Biotechnol 165:157–
Recombinase-mediated cassette exchange 166. doi:10.1016/j.jbiotec.2013.03.019
(RMCE): traditional concepts and current chal- 28. Osterlehner A, Simmeth S, Göpfert U (2011)
lenges. J Mol Biol 407:193–221. doi:10.1016/j. Promoter methylation and transgene copy
jmb.2011.01.004 numbers predict unstable protein production
17. Baser B, Spehr J, Büssow K, van den Heuvel in recombinant Chinese hamster ovary cell
J (2015) A method for specifically targeting lines. Biotechnol Bioeng 108:2670–2681.
two independent genomic integration sites doi:10.1002/bit.23216
for co-expression of genes in CHO cells. 29. Wippermann A, Rupp O, Brinkrolf K et al
Methods 95:3–12. doi:10.1016/j.ymeth. (2015) The DNA methylation landscape of
2015.11.022 Chinese hamster ovary (CHO) DP-12 cells.
18. Turan S, Zehe C, Kuehle J et al (2013) J Biotechnol 199:38–46. doi:10.1016/j.
Recombinase-mediated cassette exchange jbiotec.2015.02.014
(RMCE) - a rapidly-expanding toolbox for 30. De Carvalho DD, You JS, Jones PA (2011)
targeted genomic modifications. Gene 515:1– DNA methylation and cellular reprogram-
27. doi:10.1016/j.gene.2012.11.016 ming. Trends Cell Biol 20:609–617.
19. Inao T, Kawabe Y, Yamashiro T et al (2015) doi:10.1016/j.tcb.2010.08.003.DNA
Improved transgene integration into the 31. Clark SJ, Harrison J, Paul CL et al (1994)
Chinese hamster ovary cell genome using the High sensitivity mapping of methylated cyto-
Cre-loxP system. J Biosci Bioeng 120:99– sines. Nucleic Acids Res 22:2990–2997
106. doi:10.1016/j.jbiosc.2014.11.019 32. Adorján P, Distler J, Lipscher E et al (2002)
20. Kawabe Y, Shimomura T, Huang S et al Tumour class prediction and discovery by
(2016) Targeted transgene insertion into the microarray-based DNA methylation analysis.
CHO cell genome using Cre recombinase-­ Nucleic Acids Res 30:e21. doi:10.1016/
incorporating integrase-defective retroviral S0959-8049(01)80570-0
vectors. Biotechnol Bioeng:1600–1610. 33. Backliwal G, Hildinger M, Kuettel I et al (2008)
doi:10.1002/bit.25923 Valproic acid: a viable alternative to sodium
21. Sakuma T, Takenaga M, Kawabe Y et al butyrate for enhancing protein expression in
(2015) Homologous recombination-­ mammalian cell cultures. Biotechnol Bioeng
independent large gene cassette knock-in in 101:182–189. doi:10.1002/bit.21882
CHO cells using TALEN and MMEJ-directed 34. Kouzarides T (2000) Acetylation: a regula-
donor plasmids. Int J Mol Sci 16:23849– tory modification to rival phosphorylation?
23866. doi:10.3390/ijms161023849 EMBO J 19:1176–1179. doi:10.1093/
22. Bachu R, Bergareche I, Chasin L a. (2015) emboj/19.6.1176
CRISPR-Cas targeted plasmid integration 35. Jiang Z, Sharfstein ST (2008) Sodium butyr-
into mammalian cells via non-homologous ate stimulates monoclonal antibody over-­
end joining. Biotechnol Bioeng 112:2154– expression in CHO cells by improving gene
2162. doi: 10.1002/bit.25629 accessibility. Biotechnol Bioeng 100:189–
23. Lee JS, Kallehauge TB, Pedersen LE, 194. doi:10.1002/bit.21726
Kildegaard HF (2015) Site-specific integration 36. Kyoung MIN (2007) Correlation between
in CHO cells mediated by CRISPR/Cas9 and enhancing effect of sodium butyrate on spe-
homology-directed DNA repair pathway. Sci cific productivity and mRNA transcription
Rep 5:8572. doi:10.1038/srep08572 level in recombinant chinese hamster ovary
24. Peterson CL, Laniel M (2004) Histones and cells producing antibody. J Microbiol
histone modifications. Curr Biol 14:546–551 Biotechnol 17:1036–1040
20 Christina S. Alves and Terrence M. Dobrowsky
37. Yang WC, Lu J, Nguyen NB et al (2014)
Addition of valproic acid to CHO cell fed-­batch
cultures improves monoclonal antibody titers.
Mol Biotechnol 56:421–428. doi:10.1007/
s12033-013-9725-x
38. Kim NS, Lee GM (2002) Inhibition of
sodium butyrate-induced apoptosis in recom-
binant Chinese hamster ovary cells by consti-
tutively expressing antisense RNA of
caspase-3. Biotechnol Bioeng 78:217–228.
doi:10.1002/bit.10191
39. Barnes LM, Dickson AJ (2006) Mammalian
cell factories for efficient and stable protein
expression. Curr Opin Biotechnol 17:381–
386. doi:10.1016/j.copbio.2006.06.005
40. Kwaks THJ, Otte AP (2006) Employing epi-
genetics to augment the expression of thera-
peutic proteins in mammalian cells. Trends
Biotechnol 24:137–142. doi:10.1016/j.
tibtech.2006.01.007
41. Maksimenko O, Gasanov NB, Georgiev P,
Lines TPC (2015) Regulatory elements in vec-
tors for efficient generation of cell Lines produc-
ing target proteins. Acta Naturae 7:15–26
42. Harraghy N, Gaussin A, Mermod N (2008)
Sustained transgene expression using MAR
elements. Curr Gene Ther 8:353–366.
doi:10.2174/156652308786071032
43. Harraghy N, Buceta M, Regamey A et al
(2012) Using matrix attachment regions to
improve recombinant protein production.
Methods Mol Biol 801:93–110.
doi:10.1007/978-1-61779-352-3_7
44. Betts Z, Croxford AS, Dickson AJ (2015)
Evaluating the interaction between UCOE
and DHFR-linked amplification and stability
of recombinant protein expression. Biotechnol
Prog 31:1014–1025. doi:10.1002/btpr.2083
45. Kwaks THJ, Barnett P, Hemrika W et al
(2003) Identification of anti-repressor ele-
ments that confer high and stable protein pro-
duction in mammalian cells. Nat Biotechnol
21:553–558. doi:10.1038/nbt814
46. Otte AP, Kwaks THJ, Van Blokland RJM et al
(2007) Various expression-augmenting DNA
elements benefit from STAR-select, a novel
high stringency selection system for protein
expression. Biotechnol Prog 23:801–807.
doi:10.1021/bp070107r
47. Mead EJJ, Smales CMM (2011) mRNA
translation and recombinant gene expression
from mammalian cell expression systems.
Compr Biotechnol 1:403–409. doi:10.1016/
B978-0-08-088504-9.00043-X
48. Mead EJ, Masterton RJ, Feary M et al (2015)
Biological insights into the expression of
translation initiation factors from recombi-
nant CHOK1SV cell lines and their relation-
ship to enhanced productivity. Biochem
J 472:261–273. doi:10.1042/BJ20150928
49. Chong WPK, Sim LC, Wong KTK, Yap MGS
(2009) Enhanced IFN c production in
adenosine-­treated CHO cells: a mechanistic
study. Biotechnol Prog 25:866–873.
doi:10.1021/bp.118
50. Dreesen IA, Fussenegger M (2011) Ectopic
expression of human mTOR increases viabil-
ity, robustness, cell size, proliferation, and
antibody production of chinese hamster ovary
cells. Biotechnol Bioeng 108:853–866.
doi:10.1002/bit.22990
51. Gustafsson C, Govindarajan S, Minshull
J (2004) Codon bias and heterologous pro-
tein expression. Trends Biotechnol 22:346–
353. doi:10.1016/j.tibtech.2004.04.006
52. Welch M, Villalobos a., Gustafsson C,
Minshull J (2009) You’re one in a googol:
optimizing genes for protein expression. J R
Soc Interface 6:S467–S476. doi: 10.1098/
rsif.2008.0520.focus
53. Chung BK-S, Yusufi FNK, Mariati, et al.
(2013) Enhanced expression of codon opti-
mized interferon gamma in CHO cells.
J Biotechnol 167:326–333. doi: 10.1016/j.
jbiotec.2013.07.011
54. Kozak M (1991) Structural features in eukary-
otic mRNAs that modulate the initiation of
translation. J Biol Chem 266:19867–19870
55. Nishimiya D (2014) Proteins improving
recombinant antibody production in mamma-
lian cells. Appl Microbiol Biotechnol 98:1031–
1042. doi:10.1007/s00253-013-5427-3
56. Pybus LP, Dean G, West NR et al (2013)
Model-directed engineering of “difficult-to-­
express” monoclonal antibody production by
Chinese hamster ovary cells. Biotechnol
Bioeng xxx:1–14. doi:10.1002/bit.25116
57. Johari YB, Estes SD, Alves CS et al (2015)
Integrated cell and process engineering for
improved transient production of a
“difficult-­ to-­
express” fusion protein by
CHO cells. Biotechnol Bioeng 112:2527–
2542. doi:10.1002/bit.25687
58. Chollet ME, Skarpen E, Iversen N et al
(2015) The chemical chaperone sodium
4-phenylbutyrate improves the secretion of
the protein CA267T mutant in CHO-K1 cells
trough the GRASP55 pathway. Cell Biosci
5:57. doi:10.1186/s13578-015-0048-4
59. De Almeida SF, Picarote G, Fleming JV et al
(2007) Chemical chaperones reduce endoplas-
mic reticulum stress and prevent mutant HFE
aggregate formation. J Biol Chem 282:27905–
27912. doi:10.1074/jbc.M702672200
 Optimizing ‘Difficult to Express’ Proteins
60. Hwang S-J, Jeon C-J, Cho SM et al (2011)
Effect of chemical chaperone addition on pro-
duction and aggregation of recombinant flag-­
tagged COMP-angiopoietin 1 in Chinese
hamster ovary cells. Biotechnol Prog 27:587–
591. doi:10.1002/btpr.579
61. Liu C-H, Chen L-H (2007) Promotion of
recombinant macrophage colony stimulating
factor production by dimethyl sulfoxide addi-
tion in Chinese hamster ovary cells. J Biosci
Bioeng 103:45–49. doi:10.1263/jbb.103.45
62. Liu C-H, Chen L-H (2007) Enhanced recom-
binant M-CSF production in CHO cells by
glycerol addition: model and validation.
Cytotechnology54:89–96.doi:10.1007/s10616-
007-9078-z
63. Chakravarthi S, Jessop CE, Bulleid NJ (2006)
The role of glutathione in disulphide bond for-
mation and endoplasmic-reticulum-­generated
oxidative stress. EMBO Rep 7:271–275.
doi:10.1038/sj.embor.7400645
64. Jing Y, Borys M, Nayak S et al (2012)
Identification of cell culture conditions to con-
trol protein aggregation of IgG fusion proteins
expressed in Chinese hamster ovary cells.
Process Biochem 47:69–75. doi:10.1016/j.
procbio.2011.10.009
65. Rezaei M, Zarkesh-Esfahani SH, Gharagozloo
M (2013) The effect of different media com-
position and temperatures on the production
of recombinant human growth hormone by
CHO cells. Res Pharm Sci 8:211–217
66. Vergara M, Berrios J, Martínez I et al (2015)
Endoplasmic reticulum-associated rht-PA pro-
cessing in CHO cells: Influence of mild hypo-
thermia and specific growth rates in batch and
Chemostat cultures. PLoS One 10:e0144224.
doi:10.1371/journal.pone.0144224
67. Walsh G, Jefferis R (2006) Post-translational
modifications in the context of therapeutic
proteins. Nat Biotechnol 24:1241–1252.
doi:10.1038/nbt1252
68. De Leon GM, Wlaschin KF, Nissom PM et al
(2007) Comparative transcriptional analysis
of mouse hybridoma and recombinant
Chinese hamster ovary cells undergoing
butyrate treatment. J Biosci Bioeng 103:82–
91. doi:10.1263/jbb.103.82
69. Sung YH, Song YJ, Lim SW et al (2004)
Effect of sodium butyrate on the production,
heterogeneity and biological activity of human
thrombopoietin by recombinant Chinese
hamster ovary cells. J Biotechnol 112:323–
335. doi:10.1016/j.jbiotec.2004.05.003
70. Kantardjieff A, Jacob NM, Yee JC et al (2010)
Transcriptome and proteome analysis of Chinese
hamster ovary cells under low temperature and
21
butyrate treatment. J Biotechnol 145:143–159.
doi:10.1016/j.jbiotec.2009.09.008
71. Knappskog S, Ravneberg H, Gjerdrum C et al
(2007) The level of synthesis and secretion of
Gaussia princeps luciferase in transfected
CHO cells is heavily dependent on the choice
of signal peptide. J Biotechnol 128:705–715.
doi:10.1016/j.jbiotec.2006.11.026
72. Hegde RS, Bernstein HD (2006) The surpris-
ing complexity of signal sequences. Trends
Biochem Sci 31:563–571. doi:10.1016/j.
tibs.2006.08.004
73. Kober L, Zehe C, Bode J (2013) Optimized
signal peptides for the development of high
expressing CHO cell lines. Biotechnol Bioeng
110:1164–1173. doi:10.1002/bit.24776
74. Haryadi R, Ho S, Kok YJ et al (2015)
Optimization of heavy chain and light chain
signal peptides for high level expression of
therapeutic antibodies in CHO cells. PLoS
One 10:e0116878. doi:10.1371/journal.
pone.0116878
75. Petersen TN, Brunak S, von Heijne G, Nielsen
H (2011) SignalP 4.0: discriminating signal pep-
tides from transmembrane regions. Nat Methods
8:785–786. doi:10.1038/nmeth.1701
76. Frank K, Sippl MJ (2008) High-performance
signal peptide prediction based on sequence
alignment techniques. Bioinformatics 24:2172–
2176. doi:10.1093/bioinformatics/btn422
77. Stoops J, Byrd S, Hasegawa H (2012) Russell
body inducing threshold depends on the vari-
able domain sequences of individual human
IgG clones and the cellular protein homeosta-
sis. Biochim Biophys Acta 1823:1643–1657.
doi:10.1016/j.bbamcr.2012.06.015
78. Hasegawa H, Woods CE, Kinderman F et al
(2014) Russell body phenotype is preferen-
tially induced by IgG mAb clones with high
intrinsic condensation propensity: relations
between the biosynthetic events in the ER
and solution behaviors in vitro. MAbs
6:1518–1532. doi:10.4161/mabs.36242
79. Marotta NP, Lin YH, Lewis YE et al (2015)
O-GlcNAc modification blocks the aggrega-
tion and toxicity of the protein α-synuclein
associated with Parkinson’s disease. Nat
Chem 7:913–920. doi:10.1038/nchem.2361
80. Peroutka RJ, Elshourbagy N, Piech T, Butt TR
(2008) Enhanced protein expression in mam-
malian cells using engineered SUMO fusions:
secreted phospholipase A2. Protein Sci 17:1586–
1595. doi:10.1110/ps.035576.108.of
81. Kaufmann H, Fussenegger M (2003)
Metabolic engineering of mammalian cells for
higher protein yield. New Compr Biochem
38:457–469
22 Christina S. Alves and Terrence M. Dobrowsky
82. Inducible Protein Expression - T-REx™ System.
http://www.thermofisher.com/us/en/home/
references/protocols/proteins-­e xpression-
isolation-and-analysis/protein-­e xpression-­
protocol/inducible-protein-expression-using-
the-trex-system.html.
83. Gossen M, Bujard H (1992) Tight control of
gene expression in mammalian cells by
tetracycline-­responsive promoters. Proc Natl
Acad Sci U S A 89:5547–5551. doi:10.1073/
pnas.89.12.5547
84. Gossen M, Freundlieb S, Bender G et al
(1995) Transcriptional activation by tetracy-
clines in mammalian cells. Science 268:1766–
1769. doi:10.1126/science.7792603
85. Zhang Y, Katakura Y, Ohashi H, Shirahata S
(1997) Efficient and inducible production of
human interleukin 6 in Chinese hamster ovary
cells using a novel expression system. Cyto­
technology 25:53–60
86. Fussenegger M, Bailey JE, Hauser H, Mueller
PP (1999) Genetic optimization of recombi-
nant glycoprotein production by mammalian
cells. Trends Biotechnol 17:35–42.
doi:10.1016/S0167-7799(98)01248-7
87. Bi JX, Shuttleworth J, Al-Rubeai M (2004)
Uncoupling of cell growth and proliferation
results in enhancement of productivity in
p21CIP1-arrested CHO cells. Biotechnol
Bioeng 85:741–749. doi:10.1002/bit.20025
88. Sinacore MS, Drapeau D, Adamson SR
(2000) Adaptation of mammalian cells to
growth in serum-free media. Mol Biotechnol
15:249–257. doi:10.1385/MB:15:3:249
89. Costa AR, Rodrigues ME, Henriques M et al
(2011) Strategies for adaptation of mAb-­
producing CHO cells to serum-free medium.
BMC Proc 5(Suppl 8):P112. doi:10.1186/
1753-6561-5-S8-P112
90. Langer ES (2011) Trends in perfusion biore-
actors. Bioprocess Int 9:18–22
91. Li L, Shi M, Song Y et al (2009) A single-use,
scalable perfusion bioreactor system.
Bioprocess Int:46–54
92. Kompala DS, Ozturk SS (2006) Optimization
of high cell density perfusion bioreactors. Cell
culture technologies for pharmaceutical and
cell-based therapies. CRC Press, Taylor and
Francis Group, Boca Raton
93. Kolind MP, Nørby PL, Berchtold MW,
Johnsen LB (2011) Optimisation of the fac-
tor VIII yield in mammalian cell cultures by
reducing the membrane bound fraction.
J Biotechnol 151:357–362. doi:10.1016/j.
jbiotec.2010.12.019
94. Kaufman RJ, Wasley LC, Davies MV et al
(1989) Effect of von Willebrand factor coex-
pression on the synthesis and secretion of
factor VIII in Chinese hamster ovary cells.
­
Mol Cell Biol 9:1233–1242. doi:10.1128/
MCB.9.3.1233
95. Spiegel PC, Kaiser SM, Simon JA, Stoddard
BL (2004) Disruption of protein-membrane
binding and identification of small-molecule
inhibitors of coagulation factor VIII. Chem
Biol 11:1413–1422. doi:10.1016/j.chembiol.
2004.08.006
96. Trummer E, Fauland K, Seidinger S et al
(2006) Process parameter shifting: Part
I. Effect of DOT, pH, and temperature on
the performance of Epo-Fc expressing CHO
cells cultivated in controlled batch bioreac-
tors. Biotechnol Bioeng 94:1033–1044.
doi:10.1002/bit
97. Zhu MM, Goyal A, Rank DL et al (2005)
Effects of elevated pCO 2 and osmolality on
growth of CHO cells and production of
antibody-­ fusion protein B1: a case study.
Biotechnol Prog 21:70–77
98. Yoon SK, Choi SL, Song JY, Lee GM (2005)
Effect of culture pH on erythropoietin pro-
duction by Chinese hamster ovary cells grown
in suspension at 32.5 and 37.0 degrees
C. Biotechnol Bioeng 89:345–356.
doi:10.1002/bit.20353
99. Mandenius CF, Brundin A (2008) Bioprocess
optimization using design-of-experiments
methodology. Biotechnol Prog 24:1191–
1203. doi:10.1002/btpr.67
100. Jiang Z, Droms K, Geng Z et al (2012) Fed-­
batch cell culture process optimization.
Bioprocess Int 10:40–45
101. Sunley K, Butler M (2010) Strategies for the
enhancement of recombinant protein produc-
tion from mammalian cells by growth arrest.
Biotechnol Adv 28:385–394. doi:10.1016/j.
biotechadv.2010.02.003
102. Min Lee G, Koo J (2010) Osmolarity effects,
Chinese hamster ovary cell culture. In:
Encyclopedia of Industrial Biotechnology.
John Wiley & Sons, Inc., pp 1–8
103. Du Z, Treiber D, McCarter JD et al (2015) Use
of a small molecule cell cycle inhibitor to control
cell growth and improve specific productivity
and product quality of recombinant proteins in
CHO cell cultures. Biotechnol Bioeng
112:141–155. doi:10.1002/bit.25332
104. Ganne V, Mignot G (1991) Application of sta-
tistical design of experiments to the optimiza-
tion of factor VIII expression by CHO cells.
Cytotechnology 6:233–240. doi:10.1007/
BF00624762
105. Torkashvand F, Vaziri B, Maleknia S et al (2015)
Designed amino acid feed in improvement of

production and quality targets of a therapeutic
monoclonal antibody. PLoS One 10:e0140597.
­doi:10.1371/journal.pone.0140597
106. Ellert A, Vikström C (2014) Design of experi-
ments with small-scale bioreactor systems for
efficient bioprocess development and optimiza-
tion. Bioprocess Int 12(5):10–13
107.
Pertea M, Lin X, Salzberg S (2001)
GeneSplicer: a new computational method
for splice site prediction. Nucleic Acids Res
29:1185–1190. doi:10.1093/nar/29.5.1185
108. Brunak S, Engelbrecht J, Knudsen S (1991)
Prediction of human mRNA donor and accep-
tor sites from the DNA sequence. J Mol Biol
220:49–65. doi: 0022-2836(91)90380-O [pii]
Optimizing ‘Difficult to Express’ Proteins 23
109. Solovyev VV, Salamov AA, Lawrence CB
(1994) Predicting internal exons by oligonu-
cleotide composition and discriminant analy-
sis of spliceable open reading frames. Nucleic
Acids Res 22:5156–5163. doi:10.1093/
nar/22.24.5156
110. Reese MG, Eeckman FH, Kulp D, Haussler D
(1997) Improved splice site detection in
genie. J Comput Biol 4:311–323. doi:10.1089/
cmb.1997.4.311
111. Rogozin IB, Milanesi L (1997) Analysis of
donor splice sites in different eukaryotic
organisms. J Mol Evol 45:50–59.
doi:10.1007/PL00006200
 Chapter 2

Glycoengineering of CHO Cells to Improve Product Quality

Qiong Wang, Bojiao Yin, Cheng-Yu Chung, and Michael J. Betenbaugh

Abstract
Chinese hamster ovary (CHO) cells represent the predominant platform in biopharmaceutical industry for
the production of recombinant biotherapeutic proteins, especially glycoproteins. These glycoproteins
include oligosaccharide or glycan attachments that represent one of the principal components dictating
product quality. Especially important are the N-glycan attachments present on many recombinant glyco-
proteins of commercial interest. Furthermore, altering the glycan composition can be used to modulate
the production quality of a recombinant biotherapeutic from CHO and other mammalian hosts. This
review first describes the glycosylation network in mammalian cells and compares the glycosylation pat-
terns between CHO and human cells. Next genetic strategies used in CHO cells to modulate the sialylation
patterns through overexpression of sialyltransfereases and other glycosyltransferases are summarized. In
addition, other approaches to alter sialylation including manipulation of sialic acid biosynthetic pathways
and inhibition of sialidases are described. Finally, this review also covers other strategies such as the glyco-
sylation site insertion and manipulation of glycan heterogeneity to produce desired glycoforms for diverse
biotechnology applications.

Key words Chinese hamster ovary (CHO), N-linked glycosylation, Glycoengineering, Sialylation,
Glycosylation site insertion, Heterogeneity

1  Introduction

Therapeutic glycoproteins represent a rapidly growing segment of

the biopharmaceutical industry with total sales of many tens of bil-
lion dollars annually [1]. These products include several protein
classes such as enzymes, hormones, cytokines, growth factors, clot-
ting factors, as well as monoclonal antibodies and Ig-Fc-Fusion
proteins [2–4]. The increasing demand of biotherapeutics for the
treatments of diseases, such as cancer, immune disorders, infec-
tious diseases, genetic disorders, and ailments such as Alzheimer’s
and Parkinson’s, are the main drivers for the development of gly-
coprotein therapeutics [1].
Glycosylation is a critical posttranslational modification found
on most of these biotherapeutics. What is unique about glycosyl-
ation compared to other posttranslational processing events is the

Paula Meleady (ed.), Heterologous Protein Production in CHO Cells: Methods and Protocols, Methods in Molecular Biology,
vol. 1603, DOI 10.1007/978-1-4939-6972-2_2, © Springer Science+Business Media LLC 2017

25
26 Qiong Wang et al.

structural variety and functional diversity present, in which the gly-

cosylation can vary widely even for a single protein and also from
organism to organism. Glycosylation characteristics can play a major
role in modulating a protein’s stability, folding, targeting/traffick-
ing, immunogenicity, biological activity, and especially circulatory
half-life [5]. Oligosaccharides are attached cotranslationally through
glycosidic linkages on specific asparagine (N-linked) or serine/thre-
onine (O-linked) residues. While N-glycans are the most common
modification on biotherapeutics including monoclonal antibodies
and will be the focus of the current review, several therapeutic gly-
coproteins such as erythropoietin (EPO) and etanercept (Enbrel)
also include O-glycan modifications [6]. N-glycans are linked to the
Asn of the Asn-X-Ser/Thr consensus sequence in which X denotes
any amino acid except proline [7]. A consensus sequence for
O-linked glycosylation has yet to be identified [5]. Given its non-
template-driven nature, heterogeneity of glycosylation arises both
from variations in glycosylation site occupancy and in the diversity
of final glycan structures attached to glycoproteins emerging from
the cellular secretory compartments. As a result of the stochastic
nature of interactions between enzymes and oligosaccharide sub-
strates and the variety of enzymes that can act on any one glycan
substrate, a wide range of different glycans are generated for most
proteins as these polypeptides traverse through the endoplasmic
reticulum (ER) and various Golgi compartments [8, 9].
In particular, the N-linked glycosylation pathway in mammalian
cells involves a highly complex and interconnected reaction network
catalyzed by glycosidases and glycosyltransferases contained within
different compartments of the ER and Golgi apparatus, depicted in
the schematic of Fig. 1. The biosynthesis of mammalian N-glycans
initiates at the cytoplasmic face of the ER membrane with the transfer
of GlcNAc-P from UDP-GlcNAc to the dolichol phosphate (Dol-P)
lipid carrier to generate dolichol pyrophosphate N-acetylglucosamine
(Dol-P-P-GlcNAc) [10]. Then 14 sugars are sequentially added to
Dol-P-P-GlcNAc to form an oligosaccharide precursor (Glc3Man9
GlcNAc2) [10]. Next, oligosaccharyltransferase (OST) selects Asn-X-
Ser/Thr sequons in a nascent polypeptide and proceeds with an en
bloc transfer of Glc3Man9GlcNAc2 to the side chain amide of aspara-
gine and releasing Dol-P-P in the process [11]. The glucose residues
on the precursor are sequentially trimmed by ER α-glucosidase I and
II to form monoglucosylated glycan, which is a key intermediate in
the ER lectin chaperones calnexin/calreticulin-associated glycopro-
tein folding control cycle [12]. Once correctly folded, the precursor
is trimmed by ER α-mannosidase I to yield Man8GlcNAc2-protein
before exiting ER. After translocation into the cis-Golgi, the
Man8GlcNAc2 glycoform is further trimmed by Golgi α-mannosidases
I to give Man5GlcNAc2, a key intermediate along the pathway to
form hybrid and complex N-glycans and sometimes found as a final
glycan product.
 CHO Glycoengineering 27

Fig. 1 Schematic of N-glycosylation biosynthesis pathway in CHO cell

28 Qiong Wang et al.

As shown in Fig.1, biosynthesis of hybrid and complex

N-glycans begins in the medial-Golgi by the action of an
N-acetylglucosaminyltransferase (GnT-1 or Mgat1), which adds a
GlcNAc to Man5GlcNAc2 [10]. Then the majority of N-glycans
are trimmed by Golgi α-mannosidase II removing two mannoses
from GlcNAcMan5GlcNAc2 to yield GlcNAcMan3GlcNAc2.
Hybrid N-glycans result when a structure such as GlcNAcMan3
GlcNAc2 either undergoes no further extension or trimming to
remove exposed mannose residues resulting in structures with one
or two terminal Man residue. In addition, sometimes another
GlcNac can be added to the innermost Man group by the enzyme
β1,4-N-acetylglucosaminyltransferase III (GnT-III or Mgat3) in
the medial Golgi, resulting in bisecting GlcNAc structures that can
also alter the capacity for other downstream enzymes to act on the
glycan structure.
Next, the enzyme β-1,2-N-acetylglucosaminyltransferase II
(GnT-II or Mgat2) adds a GlcNAc to the GlcNAcMan3GlcNAc2
structure to generate the glycan product GlcNAc2Man3GlcNAc2,
which is the precursor for all multiantennary complex N-glycans. Tri-
and tetra-antennary branches can be achieved by adding GlcNAc at
α(1,3)-mannose site by N-acetylglucosaminyltransferase IV (GnT-IV
orMgat4)andatα(1,6)-mannosesitebyN-acetylglucosaminyltransferase
V (GnT-V or Mgat 5). Additional modifications of complex and
hybrid N-glycans can occur in the trans-Golgi and include the addi-
tion of core α(1,6)-fucose to the GlcNAc adjacent to Asn at the
N-glycan sites by α-(1,6)-fucosyltransferase and the branch elongation
by the addition of a β-linked galactose residue to GlcNAc by galacto-
syltransferase to produce Galβ1-4GlcNAc, or poly-acetyl-lactosamine
(poly-­LacNAc) sequences. Finally, these terminal Gal residues can
serve as acceptors for several sialyltransferases, leading to even more
complex structures [10].
Chinese hamster ovary (CHO) cells are widely used for pro-
duction of many commercial and clinical biopharmaceuticals due
to their capacity to produce glycoforms that are, with exceptions,
accepted by the human immune system [2, 13]. Alternative mam-
malian cell lines also used in the production of biopharmaceuticals
include baby hamster kidney (BHK21), murine myeloma and
hybridoma cell lines (NS0 and Sp2/0), and, to a lesser extent,
human host cell lines such as human embryonic kidney (HEK293)
and human retinal cells (PER. C6) [2, 3, 14].
Two nonhuman glycans—terminal Galα1,3-Gal linkages
(alpha-Gal) and N-glycolylneuraminic acid (Neu5Gc) residues—
exist in nonhuman mammalian cells and could elicit adverse
­immunogenic reactions in humans [2, 15]. Mouse cells have an
α1,3-galactosyltransferase enzyme that produces glycans contain-
ing the alpha-Gal linkage [16]. The second potential immunogenic
epitope Neu5Gc is common in all non-primate mammalian cells
[2] due to the presence of the enzyme, N-acetylneuraminic acid
 CHO Glycoengineering 29

hydroxylase, which coverts CMP-Neu5Ac to CMP-Neu5Gc in all

mammals other than old-world primates [17]. Furthermore, the
presence of a circulating polyclonal anti-Neu5Gc antibody response
has been detected in humans [2, 15]. In contrast to the alpha-Gal
epitope, Neu5Gc can even be taken up from the media as a metab-
olite by all mammalian cells, including human cells, and then meta-
bolically incorporated onto cell surface glycoconjugates, While all
mammalian cells have the potential for immunogenic epitopes,
mouse myeloma cells (NS0 and Sp2/0) tend to express higher
levels of both of these epitopes compared to hamster (CHO and
BHK), making recombinant products from murine cells a higher
likelihood for being immunogenic in humans. This potential
immunogenicity can be especially concerning when the therapeu-
tic glycoproteins are administered repeatedly in large doses for
chronic diseases [17–19].
Even without these two nonhuman immunogenic epitopes,
the glycosylation patterns of proteins expressed in CHO and
human cell lines are likely to differ [20]. CHO cells typically do
not express N-acetylglucosaminyltransferase III (GnT-III) and
thus lack bisecting GlcNAc residues in their glycoforms, which can
impact antibody efficacy [21]. Human cells contain GnT III and
can produce glycans with bisecting GlcNAc, while antibodies pro-
duced in mouse myeloma cells also contain a fraction of glycans
with bisecting GlcNAc residues [22].
The glycosylation of biotherapeutics has been identified as a
critical quality attribute [23] because each biotherapeutic requires
defining glycosylation characteristics to maintain consistent quality
parameters such as solubility, thermal stability, protease resistance
[24], aggregation [2, 3], serum half-life [25], immunogenicity
[5], and efficacy [26]. Thus, in order to tailor the glycosylation
structures produced by CHO cells, a number of researchers have
undertaken metabolic glycoengineering strategies to alter the final
glycan profiles and distribution in CHO. In this review, we will
document recombinant protein N-linked glycoengineering studies
in CHO cells and evaluate the impact on the N-glycosylation pat-
terns attached to proteins used across the biotechnology industry.
Given the diversity of structures possible, this review will focus on
glycoengineering primarily for non-antibody motifs and briefly dis-
cuss the glycoengineering approaches in antibodies.

2  Glycoengineering Strategies in CHO Cells

2.1  Sialylation Among the numerous sugar moieties found in glycans, the terminal
sialic acid (Neu5Ac) is considered particularly important for the
lifespan of glycosylated proteins. As an electro-negatively charged
acidic 9-carbon moiety, sialic acid is α-glycosidically linked on the
C3- or the C6-hydroxyl group of the terminal galactose in humans,
30 Qiong Wang et al.

through the action of α2,3-sialyltransferases (ST3) or the α2,6-

sialyltransferase-1 (ST6) [27–29]. Terminal sialic acid residues can
alter protein properties including biological activity and in vivo cir-
culatory half-life. Serving as a biological mask, the distal sialic acid
can shield galactose residues that when exposed prompt a fast
removal of the protein from blood circulation due to the endocyto-
sis-mediated uptake by asialoglycoprotein receptors on hepatocytes
[29–31]. Therefore, in mammalian cells, it is generally desirable to
maximize the distal sialic acid content of a glycoprotein to ensure its
quality and consistency as an effective therapeutic [12].
However, the sialic acid content of glycoproteins expressed in
CHO cells can sometimes be incomplete, which is due to two oppos-
ing cellular processes. The first process consists of two steps—the
biosynthesis of cytidine monophospho-sialic acid (CMP-SA) sub-
strate and the transfer of sialic acid from this substrate onto a glycan
catalyzed by a sialyltransferase. The second process is the extracellu-
lar removal of sialic acid by sialidase cleavage [32]. Both these path-
ways are targets for genetic engineering. Hence, in the next sections,
we discuss genetic manipulation of the sialylation process, and divide
it into three parts: genetic engineering of sialylation pathways, over-
expression of N-acetylglucosaminyltransferase (GnT) genes, and
inhibition of sialidase activity and present a table to summarize the
achievements of glycoengineering to improve protein sialylation
(Table 1).

2.1.1  Genetic Genetic engineering of sialyltransferase enzymes is probably the

Engineering of Sialylation most straightforward method to alter sialylation content in terms
Pathways of modifying the oligosaccharide biosynthesis reaction networks.
Sialyltransferases are ultimately responsible for introducing a
Neu5Ac residue to the penultimate galactose residue.
Six β-galactoside α2,3-sialyltransferases (ST3GAL1–6) and two
β-galactoside α2,6-sialyltransferases (ST6GAL1–2) are responsible for
forming these terminal sialic acids in mammalian cells. Human glyco-
proteins bear sialic acid residues in both α2,3- and α2,6-linkages,
whereas only α2,3-terminal sialic acids are found in glycoproteins from
CHO and BHK cells. A report from our group revealed that three
genes from the α2,3-sialyltransferase family (ST3GAL3, ST3GAL4,
and ST3GAL6) are responsible for α2,3-­sialylation in CHO cells using
siRNA knockdown approaches, among which ST3GAL4 plays the
critical role in dictating glycoprotein α2,3-sialylation [33]. ST6GAL1
appears to prefer the Galβ1-4GlcNAc disaccharide sequence linked to
a protein, whereas ST6GAL2 shows a preference for free disaccharide
Galβ1-4GlcNAc substrate in humans [34] .
The overexpression of heterologous α2,6-sialyltransferase
with or without recombinant α2,3-sialyltransferase serves to
introduce linkages similar to those found in human cells and has
been adapted to elevate the amounts of sialic acid on recombinant
proteins [29]. Since the first introduction of ST6GAL1 in CHO cells
Table 1
Summary of Glycoengineering strategies in CHO cells to improve protein sialylation

Increase biotherapeutic
Enzyme name Enzyme function properties Target protein Reference
ST6GAL1 (+) α2,6-sialyltransferase 1 Capping Gal residues with Increase terminal sialylation to t-PA [36]
α2,6 sialic acid further extend protein half-life EPO [37]
IFN-­ɤ [38–40, 61]
IgG [38, 39]
ST3GAL4 (+) α2,3-sialyltransferase 4 Capping Gal residues with Increase terminal sialylation to EPO [43, 46]
α2,3 sialic acid further extend protein half-life
β1,4GALT1(+) β1,4-galactosyltransferase 1 Adding Gal to GlcNac Increase galactosylation which EPO [43]
can indirectly increase
sialylation
CMP-SAS (+) CMP-sialic acid synthase Synthesize the CMP-sialic Overexpress the enzymes in EPO [46, 54,
acid in the nucleus sialylation pathway to further 55]
increase sialylation
GNE/ MNK (+) UDP-GlcNAc 2-epimerase/ Epimerization of GlcNAc Overexpress the enzymes in EPO [46]
ManNAc kinase to MAnNAc/ sialylation pathway to further
phosphorylation of increase sialylation
ManNAc
CMP-SAT (+) CMP-sialic acid transporter Transport CMP-SA from Overexpress the enzymes in INF-­ɤ [55, 56]
Cytosol to Golgi sialylation pathway to further EPO
increase sialylation
GnT-IV (+) α-1,3-d-mannoside Adding GlcNac to α1-3 Create tri and tetra antennary INF-­ɤ [59–61]
GnT-V β-1,4-N-­ mannose residue glycans and allow more EPO
(+) acetylglucosaminyltransferase Adding GlcNac to α 1-6 complex and sialylated glycan
α-1,6-d-mannoside mannose residue
CHO Glycoengineering

β-1,6-N-­
acetylglucosaminyltransferase

(continued)
31
Table 1
32

(continued)

Increase biotherapeutic
Enzyme name Enzyme function properties Target protein Reference
GnT-I (+) N-acetylglucosaminyltransferase I Transfer UDP-­GlcNAc to Increase the first reaction of EPO [63–65]
the terminal α-1,3-linked synthesis of complex glycan
Mannose further increase sialylation
Qiong Wang et al.

Neu 3 (-) Neuraminidase 3 (membrane Release of sialic acid from Reduce activity of sialidase INF-­ɤ [76]
sialidase) the Galactose residue further inhibit cleavage of
sialylation
Bcl-xL (+) An anti-apoptotic member of the Anti-apoptosis protein Delay the extracellular Fc-fusion [80]
Bcl-2 family prevents cell death accumulation of sialidase
30Kc19 (+) Cell-penetrating protein Anti-apoptosis Delay the extracellular EPO [82, 83]
accumulation of sialidase
activity further increase in
sialylation
Note: “+” indicates overexpression or introcution, “-” indicates knockdown or inhibition
 CHO Glycoengineering 33

in 1989 [35], rat or human ST6GAL1-expressing CHO and BHK

cells were successively generated and tested for various thera-
peutic glycoproteins production [36, 37]. Expressing rat
ST6GAL1 in a recombinant human tissue plasminogen activator
(tPA)-expressing CHO cell line significantly increased the α2,6
sialylation level [36]. A similar protocol was applied to modify
recombinant human interferon-ɤ (IFN-ɤ) and tissue-inhibitor of
metalloproteinases-­1 in CHO and human erythropoietin (EPO)
in BHK-21A cells. Analysis of the IFN-ɤ showed about 40% content
of α2,6-linked sialic acid for engineered CHO expressing recom-
binant STGAL1 when compared to non-detectable levels of α2,6
sialic acid for wild-type IFN-ɤ produced by CHO cells [38]. In all
cases, a mixture of α2,6- and α2,3-linked sialic acids was observed
[38–40]. These findings indicate a competition between the
endogenous α2,3 sialyltransferase and exogenous α2,6 sialyltrans-
ferase for the same sialic acid donors and acceptors [41].
Meanwhile, the step prior to sialylation for N-glycans typically
involves the addition of galactose onto the branched N-glycan
chains (see Fig. 1) and insufficient or inconsistent galactosylation
can also result in an unsatisfactory sialylation level [42]. Thus, over-
expressing both the human β1,4-galactosyltransferase (β1,4-­GalT)
and α2,3-sialyltransferase (α2,3-ST) was applied in the synthesis of
glycoprotein products with a greater and consistent proportion of
fully sialylated N-glycans [42]. The resulting oligosaccharides
showed greater homogeneity compared with control cell lines, in
which ≥90% of available branches were capped with sialic acid [42].
Compared to α2,3-ST expression alone, co-­expression of β1,4-GalT
and α2,3-ST in a CHO-producing EPO cell line achieved a higher
sialic acid content and more trisialyated glycans [43].
Aside from sialyltransferase level, the availability of nucleotide
sugar substrates and the transport of them into the Golgi can also
affect the extent of protein sialylation [3]. The biosynthesis of sialic
acid in mammalian cells takes place in the cytosol, and then is com-
pleted in the nucleus followed by the transport into the Golgi, as
shown in Fig. 2. In eukaryotes, uridine diphosphate (UDP)-
GlcNAc is initially epimerized to N-acetylmannosamine (ManNAc)
by UDP-GlcNAc 2-epimerase (GNE), and ManNAc is phosphory-
lated to ManNAc-6-phosphate by ManNAc kinase (MNK). These
two enzymes are integrated into a single bifunctional enzyme
(encoded by GNE/MNK ) [44–46]. As a rate-limiting enzyme,
GNE is regulated by feedback inhibition of the level of cytoplasmic
free CMP-Neu5Ac. A genetic disease called “sialuria” arises from
the absence of feedback regulation of this enzyme, leading to
excessive synthesis of free sialic acid, which is accumulated in cyto-
plasm and secreted into urine [47, 48].
Previous studies on the sialic acid pathway have enlightened
researchers about new approaches to increase sialylation of therapeu-
tic proteins. As a direct precursor of sialic acid, N-acetylmannosamine
34 Qiong Wang et al.

Hexosamine Pathway

UDP-GlcNAc
Feedback
UDP-GlcNAc 2-epimerase/
ManNAc kinase inhibition
(GNE/MNK)
ManNAc
UDP-GlcNAc 2-epimerase/
ManNAc kinase CMP-sialic acid
(GNE/MNK) transporter
CMP-Neu5Ac Golgi
ManNAc-6P
N-AcetyIneuraminic
Acid Synthase (NANS)
Neu5Ac-9P
N-AcetyIneuraminic
Acid Phosphatase CMP-Neu5Ac
(NANP) CMP-sialic acid
synthetase
Neu5Ac Neu5Ac

Nucleus

Fig. 2 Schematic representation of sialic acid biosynthesis pathway in mammalian cells

(ManNAc) supplementation has long been investigated for its effect

on sialylation. Numerous reports demonstrated that ManNAc sup-
plementation can significantly increase CMP-sialic acid in the intra-
cellular pool up to 12-fold, but only improve protein sialylation to a
very limited extent of about 10–20% increase [49–53]. A similar issue
exists with the overexpression of CMP-SAS and sialuria-mutated
GNE/MNK in the biosynthesis pathway, which can increase
sialylation to a limited extent [46, 54, 55]. All these approaches can
increase the intracellular pool of CMP-sialic acid, but the next step-
transporting CMP-­sialic acid to the Golgi is hampered by the ineffi-
ciency of the CMP-sialic aid transporter (CMP-SAT) on the Golgi
membrane, thereby causing the reduced availability of CMP-sialic
acid substrate for sialylation. Previous researchers overexpressed
CMP-SAT alone in CHO IFN-ɤ cell line and resulted in a 4–16%
increase in site sialylation of IFN-ɤ [56]. In addition, several groups
have implemented combinatorial engineering to apply multiple genes
in the pathway to improve sialic acid content in the intracellular pool
and the availability of sialic acid substrates in the Golgi. One group
overexpressed α2,3-ST, CMP-SAS, and CMP-SAT in a CHO recom-
binant EPO cell line, and a corresponding increase in the sialylation
was observed compared to the co-expression of α2,3-ST and CMP-
SAS [55]. Another group introduced a sialuria-mutated rat GNE/
 CHO Glycoengineering 35

MNK, Chinese hamster CMP-SAT, and human α2,3-­sialyltransferase

(α2,3-ST) simultaneously into recombinant human EPO-producing
CHO cells and found the sialic acid content of rhEPO produced
from engineered cells was 43% higher than that of control cells. The
fraction of tetra-sialylated glycans for rhEPO produced from engi-
neered cells increased ∼32%, and fraction of asialo- and mono-
sialylated glycans decreased ∼50% compared with controls [46].

2.1.2  Overexpression Overexpression of branching genes can also be applied to increase

of N-acetylg­ sialylation acceptor sites. For human proteins, bi-, tri-, and tetra-­
lucosaminyltransferase antennary structures are produced with complex-type N-glycans con-
(GnT) Genes sisting predominantly of the disaccharide Galβ1,4-GlcNAc capped by
a terminal sialic acid. As illustrated in Fig. 1, tri- and t­ etra-antennary
complex N-glycans are controlled by UDP-N-­acetylglucosamine:
α-1,3-d-mannoside β-1,4-N-­acetylglucosaminyl­transferase (GnT-IV
or Mgat4) and UDP-N-acetyl­gluco­samine:α-1,6-d-mannoside
β-1,6-N-­ acetylglucosaminyltransferase (GnT-V or Mgat5). These
branched structures are associated with various biological functions,
including cellular proliferation, cell surface signaling [21], cancer
metastasis, and regulation of T-cell activation [57] and also affect
therapeutic proteins’ clearance rate by the glomeruli of the kidneys
[58]. In one study, only a small fraction of glycoproteins produced
in a CHO cell line contained the GlcNAcβ1-6 branch product
of GnT-V [59]. Thus, more extensive modifications to glycoform
distribution can occur if genetic modulations are introduced in
the branching pathway of N-oligosaccharide biosynthesis, thereby
potentially increasing the number of sialylation acceptor sites.
In order to tailor the multi-antennary glycoforms of recombi-
nant proteins, overexpression of GnT-IV and V was used in CHO
cells producing human IFN-ɤ and EPO cell lines [59, 60]. In both
cases, tri- and tetra-antennary sugar chains increased significantly,
representing ≥50% of the total sugar chains and almost all N-glycans
were in tri- or tetra-antennary glycoforms [60]. At the same time, an
increase in poly-N-acetyllactosamine (Galβ1-4GlcNAcβ1-3) was also
observed [59, 60]. However, the increase in sialic acid content was
not equal to the increase of available sialylation a­ cceptor sites because
of incomplete sialylation. In another study, mouse ST3 and/or rat
ST6 were introduced into CHO IFN-ɤ cell lines stably transfected
with GnT-V, reaching 61.2% sialylation in α 2,3- and α 2,6-linked
sialic acid content [61]. Furthermore, the coordinated overexpres-
sion of GnT-IV, GnT-V, and ST6GAL1 genes in a CHO EPO cell
line by our group enhanced sialic acid content approximately 45%
compared to control CHO EPO expressing cells [60].
Another approach to enhance sialylation is to restore missing
functions in CHO-deficient mutants. For example, GnT I-deficient
mutants are generated either through the lectin Ricinus communis
agglutinin I (RCA-I) selection or by genetic modulations. Treating
36 Qiong Wang et al.

the CHO cells with the cytotoxic lectin Ricinus communis agglutinin
I (RCA-I) was designed to select mutants with defects in the
N-glycosylation pathway upstream of galactose addition as this lectin
was reported to be specific for terminal β1,4-linked galactose [62].
Unexpectedly, genetic analysis of RCA-I-resistant CHO mutants
showed that they are all the same type of mutants with genetic muta-
tions in the GnT-I gene [63], similar to Stanley’s Lec1 mutant. A
plausible explanation is that RCA-I is not specific for terminal β1,4-
linked galactose but possibly binds many glycan structures except for
Man5GlcNAc2 [64]. Without functional GnT-­ I, the cells fail to
­transfer N-acetylglucosamine to Man5GlcNAc2 glycan (Fig. 1). Sur­
pri­singly, the restoration of functional GnT-I in these mutants led to
an increase in the sialylation of recombinant proteins both in tran-
sient expression and in stably transfected clones [63]. While the
molecular mechanism for this phenomenon remains unknown [65],
recombinant EPO generated in this RCA-1 mutant line displayed
30% greater sialylation compared with the control EPO producing
CHO clone cultured under the same conditions [66]. Moreover,
HPAEC-PAD and MALDI-TOF MS analyses showed that EPO pro-
duced by the GnT I-restored CHO-GnT I-deficient cells also con-
tained a higher content of tri- and tetra-antennary glycans [66].

2.1.3  Inhibition Any attempt to maximize sialic acid content of a therapeutic pro-
of Sialidase Activity tein should also consider the sialidase activity because the glyco-
protein is subject to desialylation and degradation during prolonged
cell culture [29, 67]. Sialidases (neuraminidases, N-acylneuraminosyl
glycohydrolases, EC 3.2.1.18) are exoglycosidases catalyzing the
hydrolytic removal of sialic acid from sialoglycoconjugates (glyco-
proteins, polysaccharides, gangliosides) [68]. The resulting asialo-
glycoprotein product would then be rapidly cleared from the
plasma by asialoglycoprotein receptors in the liver. There are four
sialidases (Neu 1–4) identified in human, mouse, rat, and CHO
cells and their activity has been localized to different subcellular
compartments: Neu1 is located in the lysosome, Neu2 is a ­cytosolic
protein, Neu3 is located in the plasma membrane, and Neu4 is a
second lysosomal sialidase [67–69]. The functions of these siali-
dases vary in part due to different substrate specificities and subcel-
lular locations [29]. These sialidases can be crucial to various
biological functions such as growth control and differentiation,
tumorigenesis, T-cell activation and immune cell interactions, neu-
ronal differentiation, and genetic defects [68, 70–73]. Therefore,
in mammalian cells, it is often desirable to lower the cellular siali-
dase activity to ensure product quality and consistency for secreted
biotherapeutic glycoproteins.
In mammalian cell culture, some extracelluar sialidase origi-
nates from the cytosol of the CHO cells and is released to the cell
culture supernatant as a result of cell lysis [74]. Gene manipulation
techniques can be applied to inhibit the sialidase’s activity in CHO
 CHO Glycoengineering 37

cells and prevent the enzyme from being released into the culture
medium. When gene expression of CHO Neu2 was knocked down
to 40% by homologous recombination or RNA interference
(RNAi), the sialic acid content of the recombinant glycoprotein was
improved but only when cells were in the death phase [67, 75]. In
another study, CHO cells overexpressing recombinant human
interferon gamma (IFN-ɤ) were treated using short interfering
RNA (siRNA) and short-hairpin RNA (shRNA) to reduce expres-
sion of the Neu1 and Neu3 sialidase genes [76]. By knocking down
expression of Neu3, a 98% reduction in Neu3 sialidase activity was
achieved in CHO cells. Accordingly, the sialic acid content on
recombinant IFN-ɤ was found to be increased 33% and 26% for
samples from the cell stationary phase and death phase, respectively,
as compared to corresponding controls [76]. Interestingly, when
using the same siRNA technique to knock down both genes indi-
vidually, Neu3 (located in the plasma membrane) knockdown
almost silenced sialidase activity, while Neu2 (located in the cyto-
plasm) knockdown only reduced sialidase activity to 40%. Unlike
Neu2 knockdown effects that acted exclusively in the death phase,
protein sialylation was enhanced in the whole cell process after
knocking down Neu3 expression, suggesting different mechanisms
of protein sialylation regulation by Neu2 and Neu3 [29].
In addition to silencing the genes for sialidases, other
approaches have focused on inhibiting glycan degradation. Bcl-xL,
an antiapoptotic protein that inhibits the release of proapoptotic
molecules from mitochondria, is well documented for its role in
extending culture longevity by suppressing apoptotic cell death
and improved glycoprotein production [77–79]. Overexpression
of Bcl-xL can enhance the sialylation of glycoproteins produced
from CHO cell lines by reducing cell lysis and delaying the extra-
cellular accumulation of sialidase activity during prolonged cell
culture [80]. Likewise, the investigation of anti-apoptotic
components of silkworm hemolymph revealed Bombyx mori
­
30Kc19 gene expression can also enhance recombinant protein
production and sialylation in CHO [81, 82]. Stable expression of
the 30Kc19 gene in a CHO cell line producing recombinant
human EPO increased the EPO production and sialylation by
102.6% and 87.1%, respectively [82]. Moreover, with the intro-
duction of 30Kc19 gene the host suspension cells produced recom-
binant human EPO with more complex glycan structures and a
larger content of sialic acid and fucose [83]. The 30Kc19 protein
is able to maintain the activity of glycotransferases involved in the
glycosylation process [83].

2.2  Introduction In addition to modifying the oligosaccharide structures at specific

of Additional glycan sites (microheterogeneity), glycoengineering can also be
Glycosylation Sites applied to control the glycan site occupancy of a target protein
(macroheterogeneity) by altering the N-glycan consensus sequence
38 Qiong Wang et al.

as well as the number and position of the glycosylation sites on the

nascent peptide chain using site-directed mutagenesis.
Asn-X-Ser/Thr, where X is any amino acid except proline, is
the preferred N-glycan consensus sequence of choice [84]. The
presence of Pro at the X position completely blocked the glycosyl-
ation at that site, while Glu, Trp, Asp, and Leu showed inefficient
glycosylation [7, 84–87]. Moreover, the sequon of Asn-X-Thr is
more likely to be glycosylated than Asn-X-Ser. Studies on rabies
virus glycosprotein (RGP) showed that using site-directed muta-
genesis on an Asn-X-Ser sequon at Asn37 site, to substitute of Thr
for Ser at position 39 dramatically increased core glycosylation effi-
ciency of Asn37 in both membrane-anchored and secreted forms
of RGP; whereas substitution of Cys for Ser blocked the core gly-
cosylation [88]. Thus, the glycosylation of the target protein was
enhanced when threonine was present instead of serine at the
hydroxy position of N-glycan sites [84]. In addition to which
amino acid is at the X position and whether the hydroxy amino
acid in the sequence is serine or threonine, the efficiency of core
glycosylation on the asparagine residue at the consensus sequence
is also dependent on the accessibility of the consensus sequon for
the active site of the OST complex and proper transfer of the oli-
gosaccharide moiety from its lipid-linked carrier [84].
Based on previous studies of N-glycan consensus sequences,
introduction of additional N-glycan target sites into desired posi-
tions on the protein backbone by genetic mutation has been used
to create glycoproteins with enhanced levels of glycosylation and
consequently sialylation, leading to extended serum half-life and
improved in vivo activity [29, 89, 90]. These manipulations have
produced hyperglycosylated recombinant protein analogues for use
as biotherapeutics. For example, through the selection of ­several
dozen analogues of recombinant EPO containing one or two amino
acid mutations, two additional oligosaccharide-attachment sites at
asparagine residues 30 and 88 have been incorporated into EPO,
creating darbepoetin alfa (also called novel erythropoiesis stimulat-
ing protein, NESP), with a total of five N-linked oligosaccharides, a
threefold longer serum half-life, increased in vivo potency and phar-
macokinetics (PK), and less frequency of administration to obtain
the same biological response [89, 90]. In another study, additional
N-linked glycosylation sites have been added to the follicle-stimu-
lating hormone (FSH) molecule through N-terminal extensions.
The resulting FSH1208 variant was found to have a three- to four-
fold increased serum half-life compared with wild-type recombi-
nant FSH [91]. Introduction of N-glycosylation sequons onto the
flanking linker and a C-terminal extension on a recombinant anti-
body has also been shown to yield prolonged circulation time [92].
However, in producing rHuACHE from HEK-293 (HEK) cells,
the decisive factor in determining the clearance rate was related to
the number of N-glycan termini which are not occupied by sialic
 CHO Glycoengineering 39

acid residues, rather than the absolute amount of N-glycan units

[93]. Thus, the N-glycosylation load, terminal N-glycan sialylation,
and subunit oligomerization act together in determining the ulti-
mate residence time of a biotherapeutic [94]. These results clearly
suggest that a multifactorial mechanism is involved and that multi-
ple factors exert their influence in a hierarchical manner on protein
clearance. Terminal N-glycan sialylation is the governing factor in
this hierarchy, since totally desialylated forms of AChE are cleared
rapidly, and equally as well, from the circulation within minutes,
regardless of their oligomerization state and their number of
appended N-glycans [95, 96]. In this case, increasing the number
of N-glycans on the enzyme surface resulted in an increase in the
number of terminal Gal residues, which serve as highly potent clear-
ance epitopes. Thus, for glycosylation mutants of rHuAChE pro-
duced in the HEK cell system, addition of N-glycan sites had a clear
adverse pharmacokinetic effect, owing to the increase of pharmaco-
kinetically unfavorable uncapped glycan termini [96].

2.3  Heterogeneity Another issue with N-glycans of therapeutic proteins is the genera-
of Glycans tion of heterogeneous glycoforms, which present challenges in
protein purification, product consistency, and lot-to-lot reproduc-
ibility, resulting in variable therapeutic efficacy. This diversity can
sometimes adversely affect drug potency and pharmacokinetics
[97, 98]. However, N-glycans can also be crucial for protein fold-
ing, so these difficulties cannot necessarily be overcome by remov-
ing N-glycosylation sites [99]. Heterogeneity is attributed to the
lack of 100% efficiency for each step in mammalian N-glycan
­biosynthesis due to variability in enzyme levels, substrate concen-
tration, intracellular location, and the competition of different
enzymes for the same substrates [99].
In order to provide homogenous glycoforms, Zhang et al.
conducted a comprehensive Zinc-finger nuclease knockout screen
of 19 glycosyltransferase genes and identified the key genes that
control decisive steps in N-glycosylation in CHO [100]. The
authors stacked knockouts of GnT-IV-A/GnT-IV-B/GnT-V to
produce almost homogenous biantennary N-glycans [100].
Subsequently, the introduction of ST6Gal-I in CHO ST3Gal4/6
knockout cells produced a normal range of N-glycans with only
α2,6-sialylation, and when combined with a GnT-IV-A/GnT-­
IV-­B/GnT-V knockout, homogenous biantennary N-glycans
capped by α2,6-sialic acid residues were generated [100].

3  Conclusion

This review has highlighted the role of glycosylation as a critical

quality attribute in biotherapeutic production, and more impor-
tantly how these glycans can be manipulated in CHO expression
40 Qiong Wang et al.

systems through cell engineering, as summarized in Table 1.

Mammalian cell lines such as CHO can produce valuable recombi-
nant proteins that can be accepted by humans as therapeutics.
However, subtle differences between glycosylation in human and
other mammals exist and understanding these differences requires
knowledge of the physiological characteristics of each cell type.
Moreover, these differences can help to direct efforts toward gly-
can reengineering to make a wider selection of glycan moieties in
CHO cells. Efforts to exert control over protein glycosylation in
CHO cells have been demonstrated through several success stories
for maximizing terminal sialylation such as overexpression of sialyl-
transferases and other glycosyltransferases, inhibition of sialidases,
and manipulation of nucleotide sugar substrate levels. The advent
of advanced technologies such as CRISPR Cas9, TALE nucleases,
RNA interference tools, as well as the combination of next genera-
tion of sequencing with systems biotechnology will further facili-
tate the enhancements in cell glycosylation processing. These tools
will enable cell engineers to make even more highly refined and
targeted modifications to the processing capability of these cells to
meet the demand for diverse and highly effective biotherapeutic
glycoproteins for future health care needs.

Acknowledgment

This work was supported by the National Science Foundation

(grant no. 1512265).

References
1. Aggarwal RS (2014) What’s fueling the bio-
tech engine-2012 to 2013. Nat Biotechnol
32(1):32–39
2. Ghaderi D et al (2012) Production platforms
for biotherapeutic glycoproteins. Occurrence,
impact, and challenges of non-human
sialylation. Biotechnol Genet Eng Rev 28:
147–175
3. Hossler P, Khattak SF, Li ZJ (2009) Optimal
and consistent protein glycosylation in mamma-
lian cell culture. Glycobiology 19(9):936–949
4. Lepenies B, Seeberger PH (2014) Simply bet-
ter glycoproteins. Nat Biotechnol 32(5):
443–445
5. Walsh G, Jefferis R (2006) Post-translational
modifications in the context of therapeutic
proteins. Nat Biotechnol 24(10):1241–1252
6. Palomares LA, Estrada-Mondaca S, Ramirez
OT (2004) Production of recombinant pro-
teins: challenges and solutions. Methods Mol
Biol 267:15–52
7. Gavel Y, Vonheijne G (1990) Sequence differ-
ences between glycosylated and nonglycosyl-
ated Asn-X-Thr Ser acceptor sites - implications
for protein engineering. Protein Eng
3(5):433–442
8. An HJ et al (2003) Determination of
N-glycosylation sites and site heterogeneity in
glycoproteins. Anal Chem 75(20):5628–5637
9. Stavenhagen K et al (2013) Quantitative
mapping of glycoprotein micro-heterogeneity
and macro-heterogeneity: an evaluation of
mass spectrometry signal strengths using syn-
thetic peptides and glycopeptides. J Mass
Spectrom 48(6):i
10. Varki A, Cummings RD, Esko JD et al (eds)
(2009) Essentials of glycobiology, 2nd edn.
Cold Spring Harbor Laboratory Press, New
York
11. Aebi M (2013) N-linked protein glycosyl-
ation in the ER. Biochim Biophys Acta
1833(11):2430–2437

12. Aebi M et al (2010) N-glycan structures: rec-
ognition and processing in the ER. Trends
Biochem Sci 35(2):74–82
13. Butler M, Meneses-Acosta A (2012) Recent
advances in technology supporting biophar-
maceutical production from mammalian cells.
Appl Microbiol Biotechnol 96(4):885–894
14. Swiech K, Picanco-Castro V, Covas DT
(2012) Human cells: new platform for recom-
binant therapeutic protein production.
Protein Expr Purif 84(1):147–153
15. Padler-Karavani V, Varki A (2011) Potential
impact of the non-human sialic acid
N-glycolylneuraminic acid on transplant
rejection risk. Xenotransplantation 18(1):1–5
16. Bosques CJ et al (2011) Chinese hamster
ovary cells can produce galactose-alpha-1,
3-galactose antigens on proteins (vol 28, pg
1153, 2010). Nat Biotechnol 29(5):459–459
17. Muchmore EA et al (1989) Biosynthesis of
N-glycolyneuraminic acid. The primary site of
hydroxylation of N-acetylneuraminic acid is
the cytosolic sugar nucleotide pool. J Biol
Chem 264(34):20216–20223
18. Chung CH et al (2008) Cetuximab-induced
anaphylaxis and IgE specific for galactose-­
alpha-­1,3-galactose. N Engl J Med 358(11):
1109–1117
19. Butler M, Spearman M (2014) The choice of
mammalian cell host and possibilities for gly-
cosylation engineering. Curr Opin Biotechnol
30:107–112
20. Croset A et al (2012) Differences in the gly-
cosylation of recombinant proteins expressed
in HEK and CHO cells. J Biotechnol
161(3):336–348
21. Zhao Y et al (2008) Branched N-glycans reg-
ulate the biological functions of integrins and
cadherins. FEBS J 275(9):1939–1948
22. Raju TS, Jordan RE (2012) Galactosylation
variations in marketed therapeutic antibodies.
MAbs 4(3):385–391
23. Spearman M, Butler M (2015) Glycosylation
in cell culture. Anim Cell Culture 9:237–258
24. Sareneva T et al (1995) N-Glycosylation of
human interferon-gamma - glycans at Asn-25
are critical for protease resistance. Biochem
J 308:9–14
25. Wright A, Morrison SL (1997) Effect of gly-
cosylation on antibody function: implications
for genetic engineering. Trends Biotechnol
15(1):26–32
26. Sola RJ, Griebenow K (2010) Glycosylation
of therapeutic proteins: an effective strategy
to optimize efficacy. BioDrugs 24(1):9–21
27. Angata T, Varki A (2002) Chemical diversity
in the sialic acids and related alpha-keto acids:
CHO Glycoengineering 41
an evolutionary perspective. Chem Rev
102(2):439–469
28. Harduin-Lepers A et al (2001) The human sial-
yltransferase family. Biochimie 83(8):727–737
29. Wang Q et al (2015) Strategies for engineering
protein N-glycosylation pathways in mamma-
lian cells. Methods Mol Biol 1321:287–305
30. Ashwell G, Harford J (1982) Carbohydrate-­
specific receptors of the liver. Annu Rev
Biochem 51:531–554
31. Cole ES et al (1993) Invivo clearance of tissue
plasminogen-activator - the complex role of
sites of glycosylation and level of sialylation.
Fibrinolysis 7(1):15–22
32. Schauer R (2004) Sialic acids: fascinating sug-
ars in higher animals and man. Zoology (Jena)
107(1):49–64
33. Chung CY et al (2015) Assessment of the
coordinated role of ST3GAL3, ST3GAL4
and ST3GAL6 on the alpha 2,3 sialylation
linkage of mammalian glycoproteins. Biochem
Biophys Res Commun 463(3):211–215
34. Krzewinski-Recchi MA et al (2003)
Identification and functional expression of a
second human beta-galactoside alpha2,6-­
sialyltransferase, ST6Gal II. Eur J Biochem
270(5):950–961
35. Lee EU, Roth J, Paulson JC (1989) Alteration
of terminal glycosylation sequences on
N-linked oligosaccharides of Chinese hamster
ovary cells by expression of beta-galactoside
alpha 2,6-sialyltransferase. J Biol Chem
264(23):13848–13855
36. Minch SL, Kallio PT, Bailey JE (1995) Tissue
plasminogen activator coexpressed in Chinese
hamster ovary cells with alpha(2,6)-sialyl-
transferase contains NeuAc alpha(2,6)Gal
beta(1,4)Glc-N-AcR linkages. Biotechnol
Prog 11(3):348–351
37. Schlenke P et al (1997) Expression of human
α2, 6-Sialyltransferase in BHK-21A cells
increases the sialylation of coexpressed human
erythropoietin: NeuAc-transfer onto
GalNAc(βl-4)GlcNAc-R motives. In: Carrondo
MT, Griffiths B, Moreira JP (eds) Animal Cell
Technology. Springer, The Netherlands,
pp 475–480
38. Bragonzi A et al (2000) A new Chinese ham-
ster ovary cell line expressing alpha
2,6-­sialyltransferase used as universal host for
the production of human-like sialylated
recombinant glycoproteins. BBA-Gen
Subjects 1474(3):273–282
39. Jassal R et al (2001) Sialylation of human
IgG-Fc carbohydrate by transfected rat
alpha2,6-sialyltransferase. Biochem Biophys
Res Commun 286(2):243–249
42 Qiong Wang et al.
40. Monaco L et al (1996) Genetic engineering of
alpha2,6-sialyltransferase in recombinant CHO
cells and its effects on the sialylation of recom-
binant interferon-gamma. Cytotechnology
22(1–3):197–203
41. Donadio-Andrei S et al (2012) Glycoengi­
neering of protein-based therapeutics.
Carbohydrate Chemistry: Chemical and
Biological Approaches 38:94–123
42. Weikert S et al (1999) Engineering Chinese
hamster ovary cells to maximize sialic acid
content of recombinant glycoproteins. Nat
Biotechnol 17(11):1116–1121
43. Jeong YT et al (2008) Enhanced sialylation of
recombinant erythropoietin in CHO cells
by human glycosyltransferase expression.
J Microbiol Biotechnol 18(12):1945–1952
44. Stasche R et al (1997) A bifunctional enzyme
catalyzes the first two steps in
N-acetylneuraminic acid biosynthesis of rat
liver. Molecular cloning and functional
expression of UDP-N-acetyl-glucosamine
2-epimerase/N-acetylmannosamine kinase.
J Biol Chem 272(39):24319–24324
45. Reinke SO et al (2009) Regulation and patho-
physiological implications of UDP-GlcNAc
2-epimerase/ManNAc kinase (GNE) as the
key enzyme of sialic acid biosynthesis. Biol
Chem 390(7):591–599
46. Son YD et al (2011) Enhanced sialylation of
recombinant human erythropoietin in
Chinese hamster ovary cells by combinatorial
engineering of selected genes. Glycobiology
21(8):1019–1028
47. Ferreira H et al (1999) Sialuria in a Portuguese
girl: clinical, biochemical, and molecular char-
acteristics. Mol Genet Metab 67(2):131–137
48. Enns GM et al (2001) Clinical course and
biochemistry of sialuria. J Inherit Metab Dis
24(3):328–336
49. Hauser H, Wagner R (eds) (2014) Animal cell
biotechnology. Biologics production. De
Gruyter, Berlin, Boston
50. Baker KN et al (2001) Metabolic control of
recombinant protein N-glycan processing in
NS0 and CHO cells. Biotechnol Bioeng
73(3):188–202
51. Hills AE et al (2001) Metabolic control
of recombinant monoclonal antibody
N-glycosylation in GS-NS0 cells. Biotechnol
Bioeng 75(2):239–251
52. Wong NS et al (2010) An investigation of
intracellular glycosylation activities in CHO
cells: effects of nucleotide sugar precursor
feeding. Biotechnol Bioeng 107(2):321–336
53. Gu X, Wang DI (1998) Improvement of
interferon-gamma sialylation in Chinese
hamster ovary cell culture by feeding of
­ of sialidase improves glycoprotein sialic acid
N-acetylmannosamine. Biotechnol Bioeng
58(6):642–648
54. Lawrence SM et al (2001) Cloning and
expression of human sialic acid pathway genes
to generate CMP-sialic acids in insect cells.
Glycoconj J 18(3):205–213
55. Jeong YT et al (2009) Enhanced sialylation of
recombinant erythropoietin in genetically
engineered Chinese-hamster ovary cells.
Biotechnol Appl Biochem 52(Pt 4):283–291
56. Wong NS, Yap MG, Wang DI (2006)
Enhancing recombinant glycoprotein sialylation
through CMP-sialic acid t­ ransporter over
expression in Chinese hamster ovary cells.
Biotechnol Bioeng 93(5):1005–1016
57. Demetriou M et al (2001) Negative regula-
tion of T-cell activation and autoimmunity by
Mgat5 N-glycosylation. Nature 409(6821):
733–739
58. Misaizu T et al (1995) Role of antennary
structure of N-linked sugar chains in renal
handling of recombinant human erythropoi-
etin. Blood 86(11):4097–4104
59. Fukuta K et al (2000) Remodeling of sugar
chain structures of human interferon-gamma.
Glycobiology 10(4):421–430
60. Yin B et al (2015) Glycoengineering of Chinese
hamster ovary cells for enhanced erythropoie-
tin N-glycan branching and sialylation.
Biotechnol Bioeng 112(11):2343–2351
61. Fukuta K et al (2000) Genetic engineering of
CHO cells producing human interferon-­
gamma by transfection of sialyltransferases.
Glycoconj J 17(12):895–904
62. Chan KF, Goh JSY, Song Z (2014) Improving
sialylation of recombinant biologics for
enhanced therapeutic efficacy. Pharmaceutical
Bioprocessing 2(5):363–366
63. Goh JS et al (2010) RCA-I-resistant CHO
mutant cells have dysfunctional GnT I and
expression of normal GnT I in these mutants
enhances sialylation of recombinant erythro-
poietin. Metab Eng 12(4):360–368
64. Iskratsch T et al (2009) Specificity analysis
of lectins and antibodies using remodeled
glycoproteins. Anal Biochem 386(2):
133–146
65. Goh JS et al (2014) Producing recombinant
therapeutic glycoproteins with enhanced
sialylation using CHO-gmt4 glycosylation
mutant cells. Bioengineered 5(4):269–273
66. Goh JS et al (2014) Highly sialylated recom-
binant human erythropoietin production in
large-scale perfusion bioreactor utilizing
CHO-gmt4 (JW152) with restored GnT I
function. Biotechnol J 9(1):100–109
67. Ngantung FA et al (2006) RNA interference

content consistency. Biotechnol Bioeng
95(1):106–119
68. Burg M, Muthing J (2001) Characterization
of cytosolic sialidase from Chinese hamster
ovary cells: part I: cloning and expression of
soluble sialidase in Escherichia coli. Carbohydr
Res 330(3):335–346
69. Munzert E et al (1997) Production of
recombinant human antithrombin III on
20-L bioreactor scale: correlation of super-
natant neuraminidase activity, desialylation,
and decrease of biological activity of recom-
binant glycoprotein. Biotechnol Bioeng
56(4):441–448
70. Hinek A et al (2006) Lysosomal sialidase (neur-
aminidase-1) is targeted to the cell surface in a
multiprotein complex that facilitates elastic fiber
assembly. J Biol Chem 281(6):3698–3710
71. Kakugawa Y et al (2002) Up-regulation of
plasma membrane-associated ganglioside sial-
idase (Neu3) in human colon cancer and its
involvement in apoptosis suppression. Proc
Natl Acad Sci U S A 99(16):10718–10723
72. Seyrantepe V et al (2004) Neu4, a novel
human lysosomal lumen sialidase, confers nor-
mal phenotype to sialidosis and galactosialido-
sis cells. J Biol Chem 279(35):37021–37029
73. de Geest N et al (2002) Systemic and neuro-
logic abnormalities distinguish the lysosomal
disorders sialidosis and galactosialidosis in
mice. Hum Mol Genet 11(12):1455–1464
74. Gramer MJ et al (1995) Removal of sialic acid
from a glycoprotein in CHO cell culture super-
natant by action of an extracellular CHO cell
sialidase. Biotechnology (N Y) 13(7):692–698
75. Ferrari J et al (1998) Chinese hamster ovary
cells with constitutively expressed sialidase
antisense RNA produce recombinant DNase
in batch culture with increased sialic acid.
Biotechnol Bioeng 60(5):589–595
76. Zhang M et al (2010) Enhancing glycopro-
tein sialylation by targeted gene silencing in
mammalian cells. Biotechnol Bioeng 105(6):
1094–1105
77. Figueroa B Jr et al (2003) A comparison of
the properties of a Bcl-xL variant to the wild-­
type anti-apoptosis inhibitor in mammalian
cell cultures. Metab Eng 5(4):230–245
78. Reed JC et al (1996) Structure-function anal-
ysis of Bcl-2 family proteins. Regulators of
programmed cell death. Adv Exp Med Biol
406:99–112
79. Kim R (2005) Unknotting the roles of Bcl-2
and Bcl-xL in cell death. Biochem Biophys
Res Commun 333(2):336–343
80. Lee JH, Kim YG, Lee GM (2015) Effect
of Bcl-xL overexpression on sialylation of
CHO Glycoengineering 43
Fc-fusion protein in recombinant Chinese
hamster ovary cell cultures. Biotechnol Prog
31(4):1133–1136
81. Rhee WJ, Lee EH, Park TH (2009) Expression
of Bombyx mori 30Kc19 protein in Escherichia
coli and Its anti-apoptotic effect in Sf9 Cell.
Biotechnol Bioproc Eng 14(5):645–650
82. Wang Z et al (2011) Enhancement of recom-
binant human EPO production and sialylation
in chinese hamster ovary cells through Bombyx
mori 30Kc19 gene expression. Biotechnol
Bioeng 108(7):1634–1642
83. Park JH et al (2012) Enhancement of recom-
binant human EPO production and glycosyl-
ation in serum-free suspension culture of
CHO cells through expression and supple-
mentation of 30Kc19. Appl Microbiol
Biotechnol 96(3):671–683
84. Jones J, Krag SS, Betenbaugh MJ (2005)
Controlling N-linked glycan site occupancy.
Biochim Biophys Acta 1726(2):121–137
85. Bause E (1983) Structural requirements of
N-glycosylation of proteins. Studies with pro-
line peptides as conformational probes.
Biochem J 209(2):331–336
86. Mononen I, Karjalainen E (1984) Structural
comparison of protein sequences around
potential N-glycosylation sites. Biochimica et
Biophysica Acta (BBA)—Protein Structure
and Molecular Enzymology 788(3):364–367
87. Shakin-Eshleman SH, Spitalnik SL, Kasturi L
(1996) The amino acid at the X position of an
Asn-X-Ser sequon is an important determi-
nant of N-linked core-glycosylation efficiency.
J Biol Chem 271(11):6363–6366
88. Kasturi L et al (1995) The hydroxy amino acid
in an Asn-X-Ser/Thr sequon can influence
N-linked core glycosylation efficiency and the
level of expression of a cell surface glycopro-
tein. J Biol Chem 270(24):14756–14761
89. Egrie JC, Browne JK (2001) Development
and characterization of novel erythropoiesis
stimulating protein (NESP). Br J Cancer
84(Suppl 1):3–10
90. Egrie JC et al (2003) Darbepoetin alfa has a
longer circulating half-life and greater in vivo
potency than recombinant human erythro-
poietin. Exp Hematol 31(4):290–299
91. Perlman S et al (2003) Glycosylation of an
N-terminal extension prolongs the half-life
and increases the in vivo activity of follicle
stimulating hormone. J Clin Endocrinol
Metab 88(7):3227–3235
92. Stork R et al (2008) N-glycosylation as novel
strategy to improve pharmacokinetic properties
of bispecific single-chain diabodies. J Biol
Chem 283(12):7804–7812
44 Qiong Wang et al.
93. Kronman C et al (1995) Involvement of
oligomerization, N-glycosylation and
sialylation in the clearance of cholinesterases
from the circulation. Biochem J 311(Pt
3):959–967
94. Chitlaru T et al (2002) Overloading and
removal of N-glycosylation targets on human
acetylcholinesterase: effects on glycan compo-
sition and circulatory residence time. Biochem
J 363(Pt 3):619–631
95. Kronman C et al (2000) Hierarchy of post-­
translational modifications involved in the circu-
latory longevity of glycoproteins. Demonstration
of concerted contributions of glycan sialylation
and subunit assembly to the pharmacokinetic
behavior of bovine acetylcholinesterase. J Biol
Chem 275(38):29488–29502
96. Chitlaru T et al (2001) Effect of human
acetylcholinesterase subunit assembly on its
­ coproteins. Nat Biotechnol 33(8):842–844
circulatory residence. Biochem J 354(Pt 3):
613–625
97. Ferrara C et al (2006) Modulation of therapeu-
tic antibody effector functions by glycosylation
engineering: influence of Golgi enzyme local-
ization domain and co-­expression of heterolo-
gous beta1, 4-N-acetylglucosaminyltransferase
III and Golgi alpha-mannosidase II. Biotechnol
Bioeng 93(5):851–861
98. Elliott S et al (2004) Control of rHuEPO
biological activity: the role of carbohydrate.
Exp Hematol 32(12):1146–1155
99. Meuris L et al (2014) GlycoDelete engi­
neering of mammalian cells simplifies
N-glycosylation of recombinant proteins. Nat
Biotechnol 32(5):485–489
100. Yang Z et al (2015) Engineered CHO cells
for production of diverse, homogeneous gly-
 Chapter 3

Large-Scale Transient Transfection of Chinese Hamster

Ovary Cells in Suspension
Yashas Rajendra, Sowmya Balasubramanian, and David L. Hacker

Abstract
We describe a one-liter transfection of suspension-adapted Chinese hamster ovary (CHO-DG44) cells
using polyethyleneimine (PEI) for DNA delivery. The method involves transfection at a high cell density
(5 × 106 cells/mL) by direct addition of plasmid DNA (pDNA) and PEI to the culture and subsequent
incubation at 31 °C with agitation by orbital shaking. We also describe an alternative method in which 90%
of the pDNA is replaced by nonspecific (filler) DNA, and the production phase is performed at 31 °C in
the presence of 0.25% N, N-dimethylacetamide (DMA).

Key words CHO cells, Transfection, Polyethyleneimine, Orbital shaking, Recombinant protein

1  Introduction

Transient gene expression (TGE) is an established method for the

rapid production of recombinant proteins for various research
applications. Recently, significant improvements in volumetric
TGE productivities of secreted recombinant proteins have been
achieved with yields reaching up to 1 g/L for human embryonic
kidney (HEK-293E) cells and up to 2 g/L for Chinese hamster
ovary (CHO) cells using linear 25 kDa polyethyleneimine (PEI)
for DNA delivery [1, 2]. TGE methods described for CHO cells
often require the use of engineered host cells, TGE-specific expres-
sion vectors, and proprietary cell culture media [1, 3]. These
requirements severely limit the availability of some published
methods. However, all of the materials necessary to execute the
methods described here are commercially available.
In addition, the pDNA requirement for CHO transfections
remains high, restricting the scalability of TGE. We have previously
reported that the amount of transgene-bearing pDNA can be
reduced by replacing some of it with nonspecific (filler) DNA with
only a moderate loss in volumetric productivity [4]. When using
reduced pDNA amounts, the transient protein yield from ­transfected

Paula Meleady (ed.), Heterologous Protein Production in CHO Cells: Methods and Protocols, Methods in Molecular Biology,
vol. 1603, DOI 10.1007/978-1-4939-6972-2_3, © Springer Science+Business Media LLC 2017

45
46 Yashas Rajendra et al.

CHO cells can be enhanced by use of polar solvents such as N,

N-dimethylacetamide (DMA) [5, 6].
Here, we describe the procedure for one-liter transient trans-
fection of CHO-DG44 cells in an orbitally shaken 5 L glass bottle.
The method is scalable and can be performed in other containers,
including disposable TubeSpin® bioreactor 50 and 600 tubes
(TPP, Trasadingen, Switzerland), Erlenmeyer flasks, and Nalgene
carboys [7–10]. The production of a human monoclonal antibody
from a bicistronic plasmid carrying the full-length cDNAs of the
IgG light and heavy chain genes is described, but the method is
suitable for the production of most other mammalian proteins for
research use. The basic transfection method is described along with
a modified version that requires less pDNA without a substantial
reduction in protein yield.

2  Materials

2.1  Cell Culture 1. CHO-DG44 cells adapted to cultivation in serum-free

suspension.
2. Cylindrical and square-shaped glass bottles with nominal vol-
umes of 100 mL to 5 L (Schott Glass, Mainz, Germany).
3. ProCHO5 medium (Lonza AG, Verviers, Belgium) without
l-glutamine, hypoxanthine, thymidine, and phenol red (Sigma-­
Aldrich, St. Louis, MO).
4. 50× l-glutamine and phenol red solution. A stock solution
with 200 mM glutamine and 250 μg/mL phenol red is made
by dissolving 29.23 g glutamine (Applichem GmbH,
Darmstadt, Germany) and 250 mg phenol red in 800 mL
water. After complete dissolution, the volume is adjusted to
1 L by further addition of water. The solution is sterilized by
filtration through a 0.2-μm Steritop bottle-top filter with a
neck size of 45 mm (Merck & Cie, Schaffhausen, Switzerland).
The solution is transferred into sterile 50 mL centrifuge tubes
(TPP, Trasadingen, Switzerland) and kept frozen at −20 °C. For
each liter of ProCHO5 medium, 20 mL of the stock solution
is added.
5. 50× HT solution. A stock solution is made by dissolving
680 mg of hypoxanthine and 194 mg of thymidine in 800 mL
of water. After complete dissolution, the volume is adjusted to
1 L by further addition of water. The solution is sterilized by
filtration and frozen as aliquots as explained in the previous
step. For each liter of ProCHO5 medium, 20 mL of the stock
solution is added.
6. Trypan blue solution (0.4%) (Sigma-Aldrich).
7. 500 mL centrifuge bottles (Costar, Corning, New York).
 Large-Scale Transient Transfection of CHO Cells 47

8. Inverted phase contrast microscope (Telaval 31, Carl Zeiss

AG, Feldbach, Switzerland).
9. Standard tabletop centrifuge (e.g., Labofuge 200, Heraeus
AG).
10. Orbital shaker (model ISF-4-W with a rotational diameter of
5 cm; Kühner AG, Birsfelden, Switzerland).
11. Double-sided adhesive transfer tape (3M Corp., Minneapolis,
MN, USA).

2.2  Plasmids 1. pXLGCHO-A3 expressing the anti-Rhesus D IgG1 with the

heavy and light chain cDNAs cloned in separate expression cas-
settes in a head-to-head orientation [11] (see Note 1).

2.3  Plasmid DNA 1. LB agar plates with 100 μg/mL ampicillin (Applichem).

Preparation 2. LB medium (Invitrogen AG, Basel, Switzerland) with 100 μg/mL
ampicillin.
3. NucleoBond A× 500 anion exchange chromatography column
(Macherey-Nagel, Düren, Germany). The kit includes all the
necessary buffers whose compositions are given below.
4. Resuspension buffer S1: 50 mM Tris–HCl, 10 mM EDTA,
100 μg/mL RNase A (Macherey-Nagel), pH 8.0.
5. Lysis buffer S2: 200 mM NaOH, 1% SDS.
6. Neutralization buffer S3: 2.8 M potassium acetate, adjusted to
pH 5.1 with acetic acid.
7. Equilibration buffer N2: 100 mM Tris–HCl, 15% ethanol,
900 mM KCl, 0.15% Triton X-100, adjusted to pH 6.3 with
H3PO4.
8. Wash buffer N3: 100 mM Tris–HCl, 15% ethanol, 1.15 M
KCl, adjusted to pH 6.3 with H3PO4.
9. Elution buffer N5: 100 mM Tris–HCl, 15% ethanol, 1 M KCl,
adjusted to pH 8.5 with H3PO4.
10. 95% and 70% ethanol (Applichem).
11. TE (10 mM Tris–HCl, pH 7.4; 1 mM EDTA), sterilized by
autoclaving.
12. NanoDrop 2000 (Thermo Fisher Scientific AG, Reinach,
Switzerland).

2.4  Transfection 1.
Linear 25 kDa polyethyleneimine (PEI) (Polysciences,
Eppenheim, Germany) is dissolved in water at 1 mg/mL at a
pH of 7.0. When dissolving, lower the pH with 1 N HCl.
When the PEI is in solution, increase the pH to 7.0 with 1 N
NaOH. ­Filter sterilize the solution, aliquot into sterile 50-mL
tubes, and store at −20 °C. It can be stored frozen for years as
long as repeated freeze–thaw cycles are avoided (see Note 2).
48 Yashas Rajendra et al.

2. ProCHO5 medium without l-glutamine, hypoxanthine, thy-

midine, and phenol red (see Subheading 2.1).
3. 50× l-glutamine–phenol red solution (see Subheading 2.1).
4. 50× HT solution (see Subheading 2.1).
5. Sheared herring sperm DNA (Thermo Fisher Scientific AG) to
be used as filler DNA (see Note 3).
6. N, N-Dimethyl acetamide (DMA) (Sigma-Aldrich).

2.5  ELISA 1. 96-well ELISA microtiter plates with flat bottom (Becton-­
Dickinson AG, Basel, Switzerland).
2. Blocking buffer: 0.5% casein hydrolysate (Applichem) and
0.05% Tween 20 (Sigma-Aldrich) in PBS (pH 7.1).
3. Capture antibody: Goat anti-human kappa light chain (AbD
Serotec, Dusseldorf, Germany).
4. Coating solution: For each 96-well plate, 11 μL of capture
antibody is mixed with 11 mL of PBS (pH 8.0).
5. Washing buffer: PBS (pH 8.0) with 2% Tween 20.
6. Detection antibody: Alkaline phosphatase-conjugated goat
anti-human gamma heavy chain (Biosource).
7. Standard: Human IgG, whole molecule (ChromPure, Jackson
ImmunoResearch Europe Ltd., Suffolk, UK). Dilute the stan-
dard in blocking buffer to 100 ng/mL and then serially dilute
1:2 with blocking buffer.
8. Substrate buffer: Add 97 mL diethanolamine (Sigma-Aldrich)
to 700 mL H2O and adjust pH to 9.8 with 2 M HCl. Add
0.5 mL 1 M MgCl2 and 2 g NaN3. Adjust volume to 1 L.
9. Substrate: Dissolve 4-nitrophenyl phosphate disodium salt
(NPP) (Applichem) in substrate buffer to 1.5 mg/mL.
10. Stop solution: 3 M NaOH.

11. Microplate reader (e.g., SPECTRAmax™340; Molecular
Devices, Palo Alto, CA, USA).

3  Methods

3.1  Plasmid 1. Transform E. coli strain DH5α with pXLGCHO-A3 by the stan-
Purification dard CaCl2 method and spread onto LB agar plates with
100 μg/mL ampicillin (see Note 4).
2. Incubate the plates overnight (16 h) at 37 °C.
3. With a sterile loop or a pipette tip, transfer a single colony from
the transformed plate to a sterile round-bottom, polypropyl-
ene 14-mL culture tube containing 3 mL LB broth with
100 μg/mL ampicillin.
 Large-Scale Transient Transfection of CHO Cells 49

4. Incubate at 37 °C for 4–6 h with agitation at 220 rpm.

5. Use the 2 × 1.5 mL culture to inoculate 2 × 5 L Erlenmeyer
flasks each containing 1 L of LB broth with 100 μg/mL ampi-
cillin (see Note 4).
6. Incubate for 12–16 h at 37 °C with agitation at 220 rpm.
7. Transfer the culture to four 500-mL centrifuge bottles.
8. Centrifuge at 4500–6000 × g for 20 min at 4 °C and decant
the medium into an Erlenmeyer flask. Retain the cell pellets
and safely dispose of the medium.
9. Resuspend each cell pellet in 120 mL of buffer S1 from the
NucleoBond A× 500 kit. Completely resuspend the cells by
pipetting with a 25 mL pipette.
10. Transfer the resuspended cells into a 1-L glass bottle.
11. Add 120 mL of buffer S2 to the suspension. Close the cap and
mix gently by inverting the bottle six to eight times.
12. Incubate the mixture at room temperature (20–25 °C) for
2–3 min. Do not vortex, as this will release chromosomal DNA
from the cellular debris.
13. Add 120 mL of prechilled (4 °C) buffer S3 to the suspension.
Close the cap and mix gently by inverting the tube six to eight
times until a homogeneous suspension containing an off-white
flocculate is formed. Let the bottle stand in ice for 5 min.
14. Place NucleoBond® folded filter (type 1) into a folded filter
(type 2) and place the combination in a funnel. Wet the filters
with a few drops of Buffer N2 and load the bacterial lysate into
the wet filter. Collect the lysate in a separate 1-L glass bottle.
15. Attach the NucleoBond A× 10000 column to a support stand
and equilibrate the column with 100 mL of buffer N2. Allow
the column to empty by gravity flow and discard the
flow-through.
16. Load the cleared lysate supernatant from step 14 onto the
equilibrated NucleoBond column. Allow the column to empty
by gravity flow.
17. Wash the column twice with 100 mL of buffer N3. Collect the
flow-through in a beaker and then discard.
18. Elute the plasmid DNA with 100 mL of buffer N5. Collect the
eluate in a 250-mL glass bottle.
19. Add 70 mL of isopropanol at room temperature to precipitate
the plasmid DNA. Mix well and transfer 45 mL each into four
50 mL centrifuge tubes. Centrifuge at 4000 × g for 1 h at 4 °C.
20. In a laminar flow hood, carefully remove the supernatant.
21. To the pellet, add 15 mL of 70% ethanol. Vortex briefly and
centrifuge at 4000 × g for 20 min at room temperature.
50 Yashas Rajendra et al.

22. In a laminar flow hood, carefully decant the 70% ethanol.

Allow the pellet to air dry in the hood at room temperature.
23. To the pellet add 5 mL of sterile TE and incubate at 37 °C for
2–3 h on an orbital shaker.
24. Measure the absorbance of the plasmid DNA at 260 nm (A260)
and 280 nm (A280) by UV spectrometry (e.g., NanoDrop
2000) to determine the concentration. Only DNA prepa­
rations with an A260/A280 ratio ≥1.8 should be used for
transfection.

3.2  Routine Cell 1. Subcultivate CHO-DG44 cells every 3–4 days (see Note 5) by
Culture inoculation in 100 mL ProCHO5 medium (when used for cell
culture, the medium contains l-glutamine, hypoxanthine, thy-
midine, and phenol red as indicated in Subheading 2.1) (see
Note 6) in a 250 mL square-shaped glass bottle at an initial
cell density of 0.3 × 106 cells/mL.
2. Determine the cell density and viability by Trypan blue stain-
ing using a Neubauer hemocytometer and an inverted phase
contrast microscope (100× magnification).
3. After cell counting, transfer 3 × 107 cells into a 50-mL centri-
fuge tube and centrifuge at 500 × g for 3 min in a standard
tabletop centrifuge.
4. Remove the medium by aspiration or decanting. Resuspend
the cell pellet in 10 mL of ProCHO5 medium and transfer to
a 250-mL square-shaped bottle containing 90 mL of pre-
warmed ProCHO5 medium.
5. Attach the bottle to a platform mounted on an orbital shaker
using double-sided adhesive transfer tape and agitate at
110 rpm at 37 °C in a 5% CO2 atmosphere without humidity.
Keep the cap of the bottle opened about one quarter of a turn.
Alternatively, vented caps can be used.

3.3  Cell Expansion 1. Count the cells prepared as in Subheading 3.2 after reaching a
for Transfection density of about 5 × 106 cells/mL.
2. Transfer 100 mL of CHO-DG44 cells from the 250-mL bottle
into the 2-L bottle with 400 mL of pre-warmed ProCHO5
medium to reach a cell density of about 1 × 106 cells/mL.
3. Incubate the culture at 37 °C with agitation as described in
Subheading 3.2, step 3, until reaching a cell density of about
5 × 106 cells/mL (approximately 2 days).
4. On the day before transfection, transfer the culture into two
sterile 500-mL centrifuge bottles.
5. Centrifuge for 5 min at 500 × g at room temperature.
 Large-Scale Transient Transfection of CHO Cells 51

6. Remove the medium by aspiration and gently resuspend each

cell pellet in 50 mL of pre-warmed ProCHO5 medium with a
25-mL pipette (see Note 7).
7. Transfer the cells into a 5-L cylindrical glass bottle with 900 mL
of pre-warmed ProCHO5 medium. The starting cell density of
the culture is 2.0–2.5 × 106 cells/mL.
8. Transfer the 5-L bottle onto an orbital shaker and incubate at
37 °C overnight (at least 16 h) with agitation as described in
Subheading 3.2, step 6.

3.4  Transfection 1. On the day of the transfection, count the cells as described in
Subheading 3.2.
3.4.1  Standard Method
2. Transfer a total of 5 × 109 cells from the overnight culture into
two sterile 500-mL centrifuge bottles (Corning) and centri-
fuge at 500 × g for 5 min at room temperature.
3. Remove the medium by aspiration and resuspend the cells from
the two centrifuge bottles in a total volume of 50 mL by addition
of pre-warmed ProCHO5 medium. Transfer the cells into a 5-L
cylindrical glass bottle with 950 mL of pre-warmed ProCHO5
medium. The cell density is 5 × 106 cells/mL (see Note 8).
4. Add 3 mg of plasmid DNA to the culture and mix immediately
by swirling the bottle (see Note 9).
5. Add 15 mL of the linear 25 kDa PEI solution at 1 mg/mL to
the culture and mix immediately by swirling the bottle (see
Note 10).
6. Place the bottle on an orbital shaker as described in Subheading
3.2, step 6 and incubate at 31 °C in 5% CO2 and 85% humidity
with agitation at 110 rpm (see Notes 11 and 12). Keep the
bottle caps open one quarter of a turn.
7. For secreted proteins, at day 7 post-transfection, harvest the
culture by centrifuging at 4000 × g for 15 min (see Note 13).
8. A schematic diagram of the transfection method is shown in
Fig. 1 (left side).

3.4.2  Low DNA Method 1. On the day of the transfection, count the cells as described in
Subheading 3.2.
2. Transfer a total of 5 × 109 cells from the overnight cultures
into two 500-mL centrifuge bottles and centrifuge at 500 × g
for 4–5 min at room temperature.
3. Remove the medium by aspiration and resuspend the cells from
the two centrifuge bottles in a total volume of 50 mL by the
addition of pre-warmed ProCHO5 medium. Transfer this 50 mL
into a 5-L cylindrical glass bottle with 950 mL of pre-­warmed
ProCHO5 medium. The cell density after resuspension is
5 × 106 cells/mL.
52 Yashas Rajendra et al.

Fig. 1 Schematic representation of the standard transfection method and the low pDNA transfection method

4. Premix 0.3 mg of plasmid DNA and 2.7 mg of herring sperm

DNA (see Note 14). Add this to the culture and mix immedi-
ately by swirling the bottle.
5. Add 15 mL of the linear 25 kDa PEI solution at 1 mg/mL to
the culture and mix immediately by swirling the bottle.
6. Add 2.5 mL of DMA (0.25%) to the culture and mix immedi-
ately by swirling the bottle (see Notes 15 and 16).
7. Place the bottle on an orbital shaker as described in Subheading
3.2, step 6, and incubate at 31 °C in 5% CO2 and 85% humid-
ity with agitation at 110 rpm. Keep the bottle caps open one
quarter of a turn.
8. For secreted proteins, at day 7 post-transfection, harvest the
culture by centrifuging at 4000 × g for 15 min.
9. A schematic representation of the transfection method is
shown in Fig. 1 (right side).

3.5  Analysis 1. To measure recombinant antibody accumulation over time,

of Antibody Production 100 μL aliquots of the culture can be taken daily during the
production phase. After centrifugation to remove cells, the
antibody concentration in each sample is measured by sand-
wich ELISA.
2. Coat a 96-well ELISA plate overnight at 4 °C with 100 μL of
goat anti-human kappa light chain IgG diluted in PBS (coating
solution) as described in Subheading 2.5, item 4.
3. Remove the coating solution with a multichannel pipettor and
wash each well three times with 200 μL of washing buffer (see
Subheading 2.5, item 5). The final wash is performed just
before the culture samples are loaded into the wells. After the
final wash, tap the plate on a paper towel to remove any remain-
ing wash solution.
 Large-Scale Transient Transfection of CHO Cells 53

4. Samples from the culture are diluted 1:10 in blocking buffer,

and then 200 μL of each is loaded in triplicate onto the plate.
Two serial two-fold dilutions in blocking buffer (100 μL sam-
ple + 100 μL blocking buffer) are done directly on the plate
(see Note 17).
5. Load the antibody standard as serial 1:2 dilutions (see
Subheading 2.5, item 7) in triplicate on the plate.
6. Incubate the plate for 1 h at 37 °C and then remove the liquid
from each well using a multichannel pipette.
7. Wash each well three times with 200 μL washing buffer as in
step 3.
8. Add AP-conjugated goat anti-human gamma chain IgG diluted
1000-fold in PBS to each well. For each plate, dilute 11 μL of
antibody in 11 mL of PBS and add 100 μL to each well.
9. Incubate the plate for 1 h at 37 °C.
10. Remove the liquid from each well as in step 6 and wash each
well three times in 200 μL washing buffer as in step 3.
11. Add 100 μL substrate solution to each well and cover the plate
with aluminum foil.
12. Incubate the plate for 15 min at room temperature with gentle
agitation.
13. Stop the reaction by the addition of 100 μL of 3 M NaOH to
each well.
14. Measure the absorbance at 405 nm using a microplate reader.
15. Determine the antibody concentration in each sample after the
generation of the standard curve from the absorbance of the
standard samples.

4  Notes

1. For pXLGCHO-A3, both the IgG light and heavy chain genes
are expressed from the human cytomegalovirus (HCMV)
major immediate early promoter/enhancer. This is generally
the most favorable promoter for TGE in either CHO or HEK-­
293 cells.
2. Once thawed, the PEI solution in 50 mL tubes can be ali-
quoted into 15 mL tubes and either used for transfection that
day or refrozen.
3. Sheared herring sperm DNA can be diluted to a desirable con-
centration in either sterile deionized water or TE (10 mM
Tris–HCl, 1 mM EDTA, pH 7.5).
4. The expression vector used here has a high copy number origin
of replication. This is an important point because a significant
amount of pDNA is necessary for TGE at large scale. The LB
54 Yashas Rajendra et al.

culture volume (2 L) is sufficient for a Giga Prep yielding

6–10 mg of pDNA. If a low copy number plasmid is used, its
propagation may require a larger culture volume to obtain a
sufficient pDNA yield.
5. To maintain the transfectability of the cells, it is best to keep
cells in culture for no longer than 3 months (20–25 passages).
We also recommend maintaining the cells in exponential
growth phase at all times.
6. ProCHO5 medium contains plant-derived peptone hydroly-
sates that may be a source of lot-to-lot variation of the medium.
We have observed that different lots of ProCHO5 medium can
result in differences in the percentage of transfected cells follow-
ing PEI-mediated gene delivery. Therefore, each new lot of the
medium should be tested for its support of cell cultivation and
transfection before purchase, if feasible. The CHO cells should
be adapted to any new medium lot for at least four passages
(about 2 weeks) prior to testing cell growth and transfection.
7. It is important to passage the cells into fresh medium on the
day prior to transfection to ensure exponential growth as this
is optimal for PEI-mediated transfection.
8. The TGE method is applicable to the combination of
CHO-­DG44 cells and ProCHO5 medium described here. If
using a different CHO strain and/or a different medium, it is
necessary to optimize the amounts of DNA PEI, and DMA
added as well as the cell density at the time of transfection. Some
commercially available media inhibit PEI-mediated transfection
due to presence of known components such as dextran sulfate,
heparin sulfate, ferric ammonium citrate, and certain hydroly-
sates or other unknown components. Hence, it is essential to
choose a medium that supports PEI-mediated transfection.
9. If two or more plasmids are being co-transfected, for multi-­
protein complex formation as an example, then the optimal
plasmid ratio needs to be empirically determined.
10. The method described here does not involve pre-complex for-
mation with DNA and PEI prior to addition to the culture. It
is very important to minimize the time delay between addition
of pDNA and PEI and to mix the culture well after each com-
ponent is added.
11. For transfections at volumes other than 1 L, the amounts of
DNA and PEI added remain 3 and 15 μg, respectively, for each
mL of culture volume. For a 10 mL transfection, for example,
resuspend 5 × 107 cells in 10 mL of ProCHO5 and add 30 μg
pDNA and 150μL PEI (1 mg/mL stock solution).
12. The method described here is optimal when performed at
31 °C. However, it may be beneficial to test temperatures between
30 and 33 °C to obtain the best yields for the production of any
given protein.
 Large-Scale Transient Transfection of CHO Cells 55

13. Although the supernatant was harvested on day 7 post-­

transfection, the culture can be extended further if its viability
is high (>80%) at that time. Preferably, the viability of the cells
at harvest should be at least 50%.
14. The efficiency of transfection is typically 60–70% for the
method described here as determined by the percentage of
fluorescent cells following transfection with a vector expressing
a fluorescent protein.
15. The reduced pDNA method described here used 10% pDNA
and 90% filler DNA with the addition of 0.25% DMA. The
ratio of pDNA to filler DNA can be protein dependent, and it
may be necessary to optimize the ratio to obtain the best yield.
It may also be necessary to optimize the amount of DMA
added to the culture.
16. The reduced pDNA method can also be utilized in the absence
of DMA, if so desired. However, note that the protein yield is
expected to be several folds lower in the absence of DMA than
in its presence.
17. The dilution must be determined empirically for each recom-
binant protein.

References
1. Daramola O, Stevenson J, Dean G, Hatton D,
Pettman G, Holmes W, Field R (2014) A high-­
yielding CHO transient system: coexpression
of genes encoding EBNA-1 and GS enhances
transient protein expression. Biotechnol Prog
30:132–141
2. Backliwal G, Hildinger M, Chenuet S,
Wulhfard S, De Jesus M, Wurm FM (2008)
Rational vector design and multi-pathway
modulation of HEK 293E cells yield recombi-
nant antibody titers exceeding 1 g/l by tran-
sient transfection under serum-free conditions.
Nucleic Acids Res 36:e96
3. Cain K, Peters S, Hailu H, Sweeney B, Stephens
P, Heads J, Sarkar K, Ventom A, Page C,
Dickson A (2013) A CHO cell line engineered
to express XBP1 and ERO1-Lα has increased
levels of transient protein expression. Biotechnol
Prog 29:697–706
4. Rajendra Y, Kiseljak D, Manoli S, Baldi L,
Hacker DL, Wurm FM (2012) Role of non-­
specific DNA in reducing coding DNA require-
ment for transient gene expression with CHO
and HEK-293E cells. Biotechnol Bioeng
109:2271–2278
5. Rajendra Y, Balasubramanian S, Kiseljak D, Baldi
L, Wurm FM, Hacker DL (2015) Enhanced
plasmid DNA utilization in transiently trans-
fected CHO-DG44 cells in the presence of polar
solvents. Biotechnol Prog 31:1571–1578
6. Rajendra Y, Hougland MD, Schmitt MG,
Barnard GC (2015) Transcriptional and post-­
transcriptional targeting for enhanced transient
gene expression in CHO cells. Biotechnol Lett
37:2379–2386
7. Monteil DT, Tontodonati G, Ghimire S, Baldi
L, Hacker DL, Bürki CA, Wurm FM (2013)
Disposable 600-mL orbitally shaken bioreactor
for mammalian cell cultivation in suspension.
Biochem Eng J 76:6–12
8. Muller N, Girard P, Hacker DL, Jordan M,
Wurm FM (2005) Orbital shaker technology for
the cultivation of mammalian cells in suspension.
Biotechnol Bioeng 89:400–406
9. Klockner W, Buchs J (2012) Advances in shaking
technologies. Trends Biotechnol 30:307–314
10. Klockner W, Diederichs S, Buchs J (2014)
Orbitally shaken single-use bioreactors. Adv
Biochem Eng Biotechnol 138:45–60
11. Rajendra Y, Kiseljak D, Baldi L, Hacker DL,
Wurm FM (2011) A simple high-yielding pro-
cess for transient gene expression in CHO cells.
J Biotechnol 153:22–26
 Chapter 4

Cloning of Single-Chain Antibody Variants

by Overlap-­Extension PCR for Evaluation
of Antibody Expression in Transient Gene Expression
Patrick Mayrhofer and Renate Kunert

Abstract
Single-chain fragment variable–fragment crystallizable antibody constructs (scFv-Fc) are homodimeric
proteins representing valuable alternatives to heterotetrameric full-length IgG molecules to study protein
properties and product-dependent cellular behavior. In contrast to naturally occurring antibodies, these
artificial molecules are assembled from functional antibody domains to reduce molecule complexity and
enhance antibody expression levels. The scFv-Fc format retains critical antibody functions such as antigen
binding affinity and antibody effector functions. Here, we present a protocol to convert the full-length
anti-HIV-1 IgG1 antibody 2F5 into a scFv-Fc construct. Variable and constant regions are amplified by
conventional PCR reactions and assembled by a single overlap-extension PCR reaction. The amplified
product is then cloned into a mammalian expression vector suitable for high-titer transient gene expression.
This workflow can be applied to any antibody sequence by adapting the specific primer sequences to the
antibody of choice.

Key words HEK293, Monoclonal antibodies, Transient gene expression, Anti-HIV-1 2F5

1  Introduction

Since their description as magic bullets, monoclonal antibodies

(mAbs) have been continuously developed to represent the fastest
growing class of biotherapeutic proteins with US sales reaching
$24.6 billion in 2012 [1].
Without doubt, the primary function of these biomolecules is
the affinity to their respective antigen. High affinity antigen bind-
ing is mediated by the specific amino acid sequence of the
complementarity-­determining regions (CDR) present in the vari-
able regions of an antibody molecule. The correct conformation of
the CDR loops is supported by flanking framework (FR) regions.
The constant regions connected to the C-terminus of variable
regions define the isotype of an antibody. Naturally occurring iso-
types can be assigned to subclasses IgA, IgD, IgE, IgG, and IgM

Paula Meleady (ed.), Heterologous Protein Production in CHO Cells: Methods and Protocols, Methods in Molecular Biology,
vol. 1603, DOI 10.1007/978-1-4939-6972-2_4, © Springer Science+Business Media LLC 2017

57
58 Patrick Mayrhofer and Renate Kunert

determined by the different and rather invariable (“constant”)

regions of a full-length antibody molecule. For an IgG1 heavy
chain this includes the constant heavy 1 (CH1), constant heavy 2
(CH2), and constant heavy 3 (CH3) regions. For the light chain,
the constant part is established by either a kappa (κ) or lambda (λ)
constant region. The molecular structure of an antibody can fur-
ther be classified by the functional regions. The fragment crystal-
lizable (Fc) was shown to force crystallization of mAbs in X-ray
crystallographic studies and consists of regions CH2 and CH3.
The antigen binding fragment (Fab) contains the variable frag-
ment (Fv) region, formed by the heavy (vH) and light chain (vL)
variable domain together with the CH1 and light chain constant
region. Distinct regions in the Fc part are responsible for antibody
receptor-mediated effector functions. For example, within the
human body the Fc part interacts with the molecules of the comple­
ment system or Fc-gamma receptors (FcγR) to mediate complement-
dependent cytotoxicity (CDC) or antibody-­ dependent cellular
cytotoxicity (ADCC), respectively [2]. In addition to complete
mAbs, corresponding antibody fragments or fusion proteins
thereof are applied in clinical applications with increasing attention
[3, 4]. For various reasons, smaller molecules are preferable in
human application (e.g., to cross the blood-brain barrier) which
led to the definition of scFv molecules representing a single-chain
protein of the two variable antibody regions connected by a syn-
thetic linker peptide. The linker enables flexible interaction of the
vH and the vL sequence to form the antigen binding pocket. Such
scFv fragments can be produced in prokaryotic expression systems
but they tend to be unstable with limited in vivo half-life.
Alternatively, single-chain fragment variable–fragment crystal-
lizable (scFv-Fc) constructs represent homodimeric biomolecules
with reduced complexity compared to heterotetrameric IgG1 anti-
body molecules but maintain critical antibody functions, such as
antigen binding and receptor-mediated effector functions. Another
advantage of these molecules in technological aspects is that only
one chain has to be expressed within the mammalian host to
assemble a fully functional scFv-Fc molecule by disulfide-bridge
dimerization of two identical chains containing leader-vH-
linker-vL-hinge-CH2-CH3.
In this protocol we describe the construction of scFv-Fc anti-
bodies based on plasmid templates encoding full-length IgG1
sequences (Fig. 1). Critical domains (leader, vH, vL, hinge-CH2,
CH3) are amplified by conventional PCR reactions using primers
containing linker sequences (Table 1) and assembled by overlap-­
extension PCR. The PCR constructs are then ligated into a com-
mercially available mammalian expression plasmid and can then be
transfected into HEK293 cells for efficient transient expression of
the scFv-Fc molecules.
 scFv-Fc Antibody Cloning 59

Fig. 1 Assembly of scFv-Fc fragments. Three individual fragments (i) leader-vH-linker, (ii) linker-vL-hinge,
and (iii) hinge-CH2-CH3 are amplified from heavy- and light chain 2F5IgG plasmid templates with primer
pairs 1 + 2, 3 + 4, or 5 + 6, respectively (a). The three fragments are then assembled to a single open
reading frame by overlap-extension PCR and amplified with primers 1 + 6 (b). The single amplicon is then
cloned into a suitable expression vector using flanking KpnI and XhoI restriction sites (c) for transient
expression of the 2F5scFv-Fc homodimer (d). Primer sequences depicted as small arrows can be found in
Table 1
60 Patrick Mayrhofer and Renate Kunert

Table 1
Primer sequences used for the construction of 2F5scFv-Fc. Overlapping regions, RESTRICTION SITES,
2F5-SPECIFIC SEQUENCES

No Name
1 KpnI_HC-Leader_for
2 2F5vH+linker_rev
3 2F5vL+linker_for
4 2F5vL_hinge_rev
5 mutFc_for
6 XhoI_CH3_rev
7 screen_seq_CMV_for
8 screen_2F5vL_rev
Sequence
ttGGTACCgccaccatggactggacctg
agatccaccacctccgctaccgcctcccccagatcctccgccgcc
GCTGCTGATGGTCACGGT
gaggcggtagcggaggtggtggatctGCTCTGCAGCTGA
CCCAGA
gctgctcttgggctcCCTCACGTCCACCCTGGTC
gagcccaagagcagcgacaagacccacac
taCTCGAGctatcacttgccgggggac
atcaacgggactttccaaaa
GATGGTCAGGGTGAACTCG

2  Materials

2.1  Molecular 1. Use sterile pipette tips, tubes, and autoclaved ultrapure deion-
Biological Transgene ized water (dH2O).
Manipulation 2. Plasmids: Commercially available pCEP4 vector (Invitrogen),
a high-copy number plasmid containing an Epstein-Barr virus
nuclear antigen 1 (EBNA-1) expression cassette and origin of
replication (oriP) required for episomal plasmid replication
and propagation to daughter cells after cell division. High-­level
transcription is mediated by a cytomegalovirus (CMV) imme-
diate early enhancer/promoter. For 2F5 template sequences
any plasmid containing heavy- or light chain of the IgG1 anti-
HIV-1 antibody 2F5 can be used.
3. LB-amp medium and agar plates: 10 g/L tryptone, 5 g/L
yeast extract, 170 mM NaCl, pH 7.0, 1.5% agar-agar (Merck)
for plates, 100 μg/mL ampicillin.
4. Plasmid Miniprep Kit I peqGOLD (Peqlab).
5. NucleoBond Xtra Midi EF (Macherey-Nagel).
6. Thermoshaker incubator.
7. Nanodrop 2000 (Thermo Scientific).
8. Bromophenol Blue/Xylene Cyanol FF (BX) DNA loading dye
(6×, Thermo Fisher Scientific).
9. Generuler DNA ladder mix (Thermo Fisher Scientific).
10. Tris-Acetate-EDTA (TAE) Buffer (pH 8.3): 4.84 g/L Tris
base, 1.14 mL glacial acetic acid, 2 mL EDTA (0.5 M, pH 8.0).
 scFv-Fc Antibody Cloning 61

11. Agarose gels: 1% (w/v) agarose (Peqlab) in Tris-Acetate-­EDTA

(TAE) buffer and 0.3 μg/mL ethidium bromide (EtBr).
12. Gel-electrophoresis chamber (Bio-Rad).
13. Gel Doc illumination chamber (Bio-Rad).
14. KAPA HiFi PCR kit (Peqlab) containing KAPA HiFi DNA
Polymerase (1 U/μL), dNTP mix (40 mM) and fidelity reaction
buffer (5×).
15. C1000 thermal cycler (Bio-Rad).

16. Taq DNA polymerase kit (NEB) containing Taq DNA
Polymerase (5 U/μL) and Thermopol reaction buffer (10×),
dNTP mix (40 mM).
17. Oligonucleotide primer sequences (Table 1) ordered from
Sigma-Aldrich (see Note 4).
18. Solution for DNA purification by isopropanol precipitation:
3 M sodium acetate adjusted to pH 5.2, isopropanol, 70%
(v/v) ethanol.
19. Wizard Plus SV Gel and PCR clean-up system (Promega).
20. High-fidelity (HF) variant of KpnI (NEB) and XhoI (NEB).
21. Restriction buffer CutSmart (NEB).
22. Ligation buffer: T4 DNA Ligase (400 U/μL, NEB), T4 DNA
ligase reaction buffer (10×, NEB).
23. SOC-medium: 20 g/L tryptone, 5 g/L yeast extract, 10 mM
NaCl, 3 mM KCl, 10 mM MgCl2, 10 mM MgSO4, 20 mM
glucose.
24. Electrocompetent E. coli Top10 (Invitrogen).
25. Gene Pulser Xcell electroporator (Bio-Rad).
26. Thermomixer Comfort (Eppendorf).
27. Glycerol (Merck).

2.2  Transient 1. HEK293-6E cells (National Research Council, Canada) stably

Transfection expressing the EBNA-1 protein for episomal replication of
plasmids containing oriP sequences.
2. Freestyle F17 expression medium (Invitrogen) supplemented
with 4 mM l-glutamine (Biochrom), 0.1% Kolliphor P188
(Sigma-Aldrich), 15 mg/L phenol red (Sigma-Aldrich), 25 μg/
mL G418 (Biochrom).
3. Linear polyethylenimine “PEIMax” (40 kDa, Polysciences)
solution at 1 mg/mL in dH2O.
4. Tryptone N1 (TekniScience) 20% (w/v) stock solution in
Freestyle F17 supplemented with 8 mM l-glutamine, 0.1%
Kolliphor P188, and 15 mg/L phenol red.
5. Valproic acid (VPA) 500 mM stock solution in dH2O.
62 Patrick Mayrhofer and Renate Kunert

6. 50 mL conical cell culture tubes (Corning).

7. Climo-shaker ISF1-XC (Kuhner).
8. Hemocytometer (Neubauer).

3  Methods

In a first step three individual fragments are amplified from differ-

ent plasmids containing heavy- or light chain variable and constant
(hinge-CH2-CH3) sequences of the IgG1 antibody 2F5. These
fragments comprise (i) leader-vH-linker, (ii) linker-vL-hinge, and
(iii) hinge-CH2-CH3 (Fig. 1a). The three fragments are then
combined in one reaction for overlap-extension PCR (Fig. 1b).
Afterward the 2F5 scFv-Fc fragment is cloned (Fig. 1c) into the
commercially available mammalian expression vector pCEP4
(see Note 1). The established mammalian expression vector con-
taining the scFv-Fc construct is then subjected to transient gene
expression in mammalian cells to yield the fully functional antibody
variant (Fig. 1d).

3.1  Generation It is recommended to generate and purify template plasmid DNA

of 2F5 Fragments 1–3 for fragment amplification using a “Miniprep” purification kit
from IgG Template according to manufacturer’s protocol (see Note 2).
Plasmids 1. Inoculate 10 mL LB-amp medium with a cryopreserved E. coli
3.1.1  Template Plasmid stock, containing the heavy- or light chain sequence of the
DNA Generation 2F5 IgG1 molecule, using a sterile pipette tip.
and Purification 2. Incubate for 12–16 h at 37 °C and 200 rpm.
3. Pellet the cell suspension by centrifugation for 10 min at
5000 × g and 4 °C.
4. Discard the culture supernatant.
5. Resuspend the pellet in 250 μL solution I (complemented with
RNase A) and transfer everything into a microcentrifuge tube.
6. Vortex vigorously to disrupt any cell clumps (critical step to
enhance efficiency of cell lysis).
7. Add 250 μL alkaline solution II to initiate cell lysis and mix by
inverting the tube six times.
8. Incubate for 2 min at room temperature.
9. Neutralize the clear lysate by adding 350 μL solution III and
mix by inverting the tube six times.
10. Centrifuge the tube for 10 min at 10,000 × g and 4 °C to
remove precipitated proteins and genomic DNA.
11. Transfer the supernatant onto a silica membrane column and
centrifuge for 1 min at 10,000 × g and 4 °C.
 scFv-Fc Antibody Cloning 63

12. Discard the flow-through and wash the bound plasmid DNA
with 500 μL “PW” plasmid kit-buffer and 750 μL DNA wash
kit-buffer by centrifugation for 1 min at 10,000 × g and 4 °C
after each washing step.
13. Elute the purified plasmid DNA from the column by adding
100 μL autoclaved deionized water (dH2O).
14. Determine the concentration and quality of the plasmid DNA
solution by Nanodrop (see Note 3). This procedure usually
gives yields higher than 10 μg for a 10 mL E. coli suspension
carrying the pCEP4 vector.
15. Adjust plasmid DNA concentration to 100 ng/μL.

3.1.2  Isolation of Specific
PCR Amplicons
10. Measure DNA concentrations and adjust concentration to
1. To prepare the primer stock solutions resuspend all lyophilized
primer samples (Table 1) in dH2O according to manufacturer’s
instructions to give a 100 μM primer master stock solution
that is routinely stored at −20 °C (see Note 4). A 10 μM primer
working stock solution is then prepared and used for setting up
the PCR reactions.
2. For amplifying the three 2F5 fragments in three separate PCR
reactions, a master mix (3×) containing all substances but
primers and plasmid DNA is prepared. First, mix 50.3 μL
dH2O with 15 μL fidelity reaction buffer, 2.3 μL dNTP mix,
and 1.5 μL KAPA DNA polymerase (see Note 5).
3. Mix, vortex, and spin down by centrifugation for 5 s at 10,000 × g.
4. Prepare three PCR tubes, each containing 0.75 μL of one for-
ward primer (primers 1, 3, or 5), 0.75 μL of one reverse primer
(primers 2, 4, or 6), and 0.5 μL plasmid template DNA (50 ng
total per reaction, two tubes containing the 2F5 heavy chain
sequence with primer pair 1 + 2 or 5 + 6 and one tube contain-
ing the 2F5 light chain plasmid with primer pair 3 + 4).
5. Add 23 μL of PCR master mix to each tube.
6. To start the PCR reaction use the thermocycler program 1
(Table  2) to amplify each fragment from the 2F5 template
plasmids.
7. After finishing the PCR reaction, add 5 μL of BX-buffer and
load the amplicons individually on a 1% agarose gel (+0.3 μg/
mL EtBr) for gel-electrophoresis in TAE buffer (+0.3 μg/mL
EtBr) at 100 V for approx. 1 h (see Note 6).
8. To isolate the specific PCR amplicons cut the PCR amplicons
showing the correct size (fragment 1:512 bp, fragment 2:362 bp,
and fragment 3:710 bp) under UV-light (see Note 8).
9. Extract and purify the DNA from the sliced agarose gel using
commercial extraction kits (see Note 9). Elute purified frag-
ments with dH2O.
10 ng/μL (see Note 10).
64 Patrick Mayrhofer and Renate Kunert

3.2  Overlap-­ To generate the full-length insert by overlap-extension PCR, a

Extension PCR total of 50 ng DNA template (fragments 1–3) is used in the PCR
reactions. Equimolar amounts for each individual fragment are
adjusted based on the fragment length of each fragment. As a first
PCR step, 10 cycles without any primers are initiated by using
thermocycler program 2 (Table 2) to generate full-length tem-
plates based on overlapping regions of the three fragments. A sec-
ond PCR is initiated with the same solution by adding outer
primers 1 and 6 followed by PCR program 3 (Table 2).
1. PCR without primers: Mix 12.5 μL dH2O, 5 μL fidelity reac-
tion buffer, 0.75 μL dNTP, 1.6 μL fragment 1 (16 ng), 1 μL
fragment 2 (10 ng), 2.2 μL fragment 3 (22 ng), and 0.5 μL
KAPA DNA Polymerase. Vortex and spin down. Start the PCR
cycle program 2 (Table 2).
2. To amplify full-length template using outer primers, add
0.75 μL primer 1 and 0.75 μL primer 6 to the PCR reaction
solution and start PCR cycle program 3 (Table 2).
3. After the PCR reaction has finished, add 5 μL of BX-buffer and
load the solution loaded onto a 1% agarose gel (+0.3 μg/mL
EtBr) for gel-electrophoresis in TAE buffer (+0.3 μg/mL
EtBr) at 100 V for approx. 1 h.
4. To isolate the specific PCR amplicon cut the PCR amplicon
showing the correct size (1543 bp) under UV-light.
5. Purify the DNA from the agarose gel using the Wizard Plus SV
Gel and PCR clean-up system. Elute the purified fragment
with 50 μL dH2O (see Note 9).

Table 2
Thermocycler programs used for PCR amplification

Program 1 Program 2 Program 3 Program 4

Step 1 (initial denaturation) 95 °C for 180 s 95 °C for 180 s 95 °C for 180 s 95 °C for 300 s
Step 2 (denaturation) 98 °C for 20 s 98 °C for 20 s 98 °C for 20 s 95 °C for 30 s
Step 3 (annealing, see Note 7) 65 °C for 25 s 65 °C for 15 s 65 °C for 15 s 60 °C for 30 s
Step 4 (extension) 72 °C for 15 s 72 °C for 45 s 72 °C for 45 s 68 °C for 60 s
Repeat (steps 2–4) 20 cycles 10 cycles 30 cycles 30 cycles
Step 5 (final extension) 72 °C for 300 s 72 °C for 300 s 72 °C for 300 s 68 °C for 300 s
Primers fragment 1: 1 + 2 none 1 + 6 7 + 8
fragment 2: 3 + 4
fragment 3: 4 + 5
Amplicon: fragment 1: 512 bp 1543 bp 1543 bp 862 bp
fragment 2: 362 bp
fragment 3: 710 bp
 scFv-Fc Antibody Cloning 65

3.3  Restriction 1. To generate the sticky ends by KpnI/XhoI double-digestion of

and Cloning the PCR amplicon, add 5 μL CutSmart buffer, 0.5 μL KpnI,
into pCEP4 and 0.5 μL XhoI to the eluted PCR amplicon solutions and
Expression Vector incubate for 3 h at 37 °C. Heat inactivate for 20 min at 65 °C
(see Note 11).
3.3.1  Purification
of Restricted PCR Amplicon 2. Precool centrifuge to 4 °C.
and pCEP4 Vector 3. Add 5  μL of 3 M sodium acetate (pH 5.2) and 35 μL
Backbone isopropanol.
4. Pellet the DNA precipitate by centrifugation at 15,000 × g for
30 min.
5. Remove the supernatant by carefully inverting the tube with-
out disturbing the pellet (hardly visible).
6. Wash the pellet with 1 mL 70% (v/v) ethanol.
7. Centrifuge at 15,000 × g for 15 min.
8. Decant the supernatant and dry the pellet for 5–20 min.
Dissolve in 15 μL dH2O.
9. Measure DNA concentrations and adjust concentration to
10 ng/μL.
10. To carry out restriction of the pCEP4 vector backbone digest
3 μg pCEP4 plasmid DNA in a total of 50 μL containing 5 μL
CutSmart buffer, 0.5 μL KpnI, and 0.5 μL XhoI in dH2O.
Incubate for 3 h at 37 °C. Heat inactivate for 20 min at 65 °C
(see Note 12).
11. To purify the digested pCEP4 vector backbone add 10 μL BX-­
buffer to 50 μL inactivated restriction solution and load onto a
1% agarose gel (+0.3 μg/mL EtBr) for gel-electrophoresis in
TAE buffer (+0.3 μg/mL EtBr) at 100 V for approximately 1 h.
12. Cut the linearized plasmid DNA showing the correct size
(10,157 bp) under UV-light.
13. Extract and purify the DNA from the agarose gel using the
Wizard Plus SV Gel and PCR clean-up system.
14. Elute the linearized vector with 50 μL dH2O.
15. Measure DNA concentrations and adjust to 10 ng/μL.

3.3.2  Ligation To find the optimal molar insert: vector ratio and to check for liga-
tion efficiency three parallel reactions are set up.
1. Prepare a master mix (3×) containing 29.2 μL dH2O, 15 μL
linearized pCEP4 vector (50 ng per reaction), 6 μL T4 DNA
ligase reaction buffer, 3 μL T4 DNA ligase.
2. Transfer 17.7 μL of the master mix into each of three tubes.
Add to tube 1: 2.3 μL dH2O, to tube 2: 0.8 μL digested
2F5scFv-Fc insert (8 ng) and 1.5 μL dH2O, and to tube 3:
2.3 μL digested 2F5scFv-Fc insert (23 ng), resulting in a molar
vector: insert ratios of 1:0, 1:1, and 1:3, respectively.
66 Patrick Mayrhofer and Renate Kunert

3. Incubate for 10 min at room temperature or overnight at

16 °C.
4. Heat inactivate by incubation at 65 °C for 10 min.
5. Chill on ice.

3.3.3  Transformation 1. Chill electroporation cuvettes on ice

into  Electro-­Competent 2. Gently thaw 40 μL TOP10 aliquots on ice.
TOP10 E. coli
3. Add 3  μL of ligation mixture to 40 μL electrocompetent
TOP10 E. coli aliquots and apply an electric pulse (1.8 kV,
25 μF, 200 Ω).
4. Immediately, add 250 μL of SOC medium to regenerate the
cells by incubation for 1 h at 37 °C and 400 rpm on a ther-
moshaker incubator to induce the ampicillin resistance gene.
5. Plate 50 and 100 μL of the cell suspension onto LB-amp selec-
tive agar (100 μg/mL ampicillin).
6. Incubate overnight at 37 °C.

3.3.4  Colony PCR 1. To prepare a master mix (10×) use 256.8 μL dH2O, 30 μL
ThermoPol buffer, 6 μL dNTP mix, 3 μL forward primer 7
(screen_seq_CMV_for), 3 μL reverse primer 8 (screen_2F5vL_
rev), and 1.2 μL Taq DNA polymerase (see Note 13).
2. Aliquot 30 μL into PCR tubes.
3. Prepare microcentrifuge tubes containing 50 μL LB-amp
medium and one LB-amp agar plate.
4. By using sterile toothpicks or pipette tips transfer a single col-
ony of the overnight LB-amp agar plate into the colony PCR
solution and stir.
5. Transfer the same toothpick or pipette tip into the tube with
liquid LB-amp medium and then streak onto the LB-amp plate
to inoculate the liquid culture and the LB-agar plate as a
backup.
6. Repeat this step for ten individual colonies (see Note 14).
7. Incubate the LB-amp agar backup plate overnight at 37 °C
and the liquid LB-amp tubes at 37 °C and 400 rpm in the
thermomixer.
8. Propagate positive clones after PCR screening in liquid over-
night cultures by adding 50 μL cell suspension to 10 mL fresh
LB-amp medium and incubate at 37 °C and 200 rpm in the
thermoshaker incubator.
9. The next day a cryostock and DNA isolation (“Miniprep”) is
carried out. For cryopreservation of positive colonies mix
625 μL of exponential growth cultures with 375 μL 80% glyc-
erol solution to yield a 30% glycerol cryostock that is stored at
−80 °C.
 scFv-Fc Antibody Cloning 67

10. Purify plasmid DNA by commercial miniprep kits as described

in Subheading 3.1.1.
11. Check the integrity of the sequence by restriction enzyme
digestion control and DNA Sanger-sequencing. For control
digestion use 0.5 μg plasmid DNA in a total of 50 μL contain-
ing dH2O, 5 μL CutSmart buffer, 0.5 μL restriction enzyme 1,
and 0.5 μL restriction enzyme 2. Restriction enzymes 1 and 2
should be chosen according to the specific antibody sequence
to ideally cut the plasmid once in the vector sequence and once
within the antibody insert. For the pCEP4_2F5scFv-Fc plas-
mid presented here, this can be done with AgeI and EcoRV to
yield two fragments of 2.5 kb and 9.2 kb in length.
12. Incubate for 3 h at 37 °C.
13. Add 10 μL BX-Buffer and load 12 μL (100 ng DNA) onto a
1% agarose gel.
14. Run the gel at 120 V for approximately 1 h.
15. For the DNA Sanger-sequencing reaction premix 1.2 μg plas-
mid DNA and 3 μL of 10 μM primer stock 7 (screen_seq_
CMV_for) in a total volume of 15 μL and send to a sequencing
service (see Note 15).
16. Plasmid preparation: Purify enough material for transient gene
expression using commercial DNA purification kits (see Note 16).

3.4  Transient 1. HEK293-6E routine cultures are grown and passaged every
Transfection 3–4 days in Freestyle F17 medium supplemented with 4 mM
of Mammalian l-glutamine, 0.1% Kolliphor P188, 15 mg/L phenol red, and
HEK293-6E Cells 25 μg/mL G418.
2. One day before transfection passage cells at 1:2 ratio in fresh
culture medium.
3. For a 15 mL transfection resuspend 15 × 106 cells in exponen-
tial growth phase at high viability (>95%) in 12 mL F17
medium supplemented with 8 mM l-glutamine, 0.1% Kolliphor
P188, and 15 mg/L phenol red in a 50 mL shaker tube
(see Note 17).
4. Transfer 15 μg plasmid DNA in 1.5 mL F17 medium without
supplements and incubate for 3 min.
5. Transfer 15 μL PEIMax-reagent in 1.5 mL F17 medium without
supplements and incubate for 3 min.
6. Transfer 1.5 mL PEIMax/F17 solution to 1.5 mL DNA/
F17 solution and incubate for 3 min to induce polyplex
formation.
7. Add the 3 mL polyplex-solution dropwise under continuous
swirling to the 12 mL cell suspension and incubate the shaker
tube in the climo-shaker at 220 rpm, 50 mm shaking ampli-
tude, 37 °C, 7% CO2 and 85% humidity.
68 Patrick Mayrhofer and Renate Kunert

8. After 2 days feed cultures to a final concentration of 5 mM

valproic acid (VPA) and 0.5% (w/v) Tryptone N1.
9. Culture cells for 7 days or until viability drops below 80%.

4  Notes

1. pCEP4 can be obtained from Invitrogen. Other mammalian

expression vectors might be used as well.
2. The procedure here is described for Plasmid Miniprep Kit I,
peqGOLD (Peqlab). Other commercial miniprep kits might
be suitable as well.
3. The quality of purified plasmid DNA is usually assessed by
absorbance ratios at 260 nm/280 nm and 260 nm/230 nm.
Low protein or phenol contamination of purified plasmid
DNA preparations usually gives 260 nm/280 nm ratios of
about 1.8. 260 nm/230 nm ratios of 2.0–2.2 are generally
obtained from pure preparations free of EDTA, phenols, and
carbohydrates.
4. Primers in our lab are routinely ordered from Sigma-Aldrich
using a synthesis scale of 0.025 μmol, desalted as purifica­
tion method and dried: https://www.sigmaaldrich.com/­
configurator/servlet/DesignTool?prod_type=STANDARD.
Primers ordered from other suppliers might be suitable as well.
5. Use high-fidelity DNA polymerases such as KAPA HiFi DNA
polymerase (error rate: 2.8 × 10−7). Other high-fidelity enzymes
such as Phusion DNA polymerase (NEB, error rate: 4.4 × 10−7)
might be used as well using modified PCR cycle conditions.
6. EtBr is strongly mutagenic. Working with EtBr requires per-
sonal protective equipment such as protective goggles, labora-
tory coat, and nitrile gloves. Use separate space for procedures
involving EtBr. If possible, use a separate room.
7. Optimal annealing temperature for each primer pair can be deter-
mined using a temperature gradient for the annealing step.
8. For working under UV-light use protective equipment to
cover skin and eyes. Exposure time of preparative DNA samples
to UV-light should be minimized to prevent degradation and
introduction of random mutations.
9. In our lab we routinely use Wizard Plus SV Gel and PCR clean-up
system (Promega) for purifying PCR reaction or agarose gels.
Products from other suppliers might be used as well.
10. For concentrating DNA samples isopropanol precipitation or
vacuum evaporation might be used. Individual fragments can
be sequenced at this point to check for errors.
 scFv-Fc Antibody Cloning 69

11. Only XhoI can be heat-inactivated. KpnI is removed by

­isopropanol precipitation or gel-electrophoresis.
12. To check for restriction efficiency we suggest including following
controls: (i) negative control containing no restriction enzyme,
(ii) KpnI only control, (iii) XhoI only control. To prevent
co-purification of single-cut plasmids an additional step for
dephosphorylation of linearized plasmids with calf intestinal
alkaline phosphatase (CIP) might be included following iso-
propanol purification to remove active KpnI.
13. For this qualitative screen a low fidelity DNA polymerase is
sufficient. We routinely use the Taq DNA Polymerase (NEB).
14. Success rate for finding positive colonies containing the gene
of interest correctly integrated into the plasmid DNA depends
on the ligation and transformation efficiency. More colonies
should be screened at lower cloning efficiencies.
15. This protocol is used for a Barcode Economy Run at Microsynth
AG. Conditions might change for other sequencing services.
16. Possible suppliers: Macherey-Nagel NucleoBond Xtra Midi EF
or Maxi EF or QIAGEN Plasmid Midi Kit, Maxi kit, Mega kit,
or Giga kit.
17. For high transfection efficiencies it is important to have single-­
cell suspensions with minor aggregate formation. A short and
gentle vortex step can be included for resuspension of the cell
pellet in fresh transfection medium.

References

1. Aggarwal S (2014) What’s fueling the biotech
engine-2012 to 2013. Nat Biotechnol 32:32–39
2. Hansel TT, Kropshofer H, Singer T, Mitchell
JA, George AJT (2010) The safety and side
effects of monoclonal antibodies. Nat Rev Drug
Discov 9:325–338
3. Beck A, Wurch T, Bailly C, Corvaia N (2010)
Strategies and challenges for the next gene­
ration of therapeutic antibodies. Nat Rev
Immunol 10:345–352
4. Carter PJ (2006) Potent antibody therapeutics
by design. Nat Rev Immunol 6:343–357
 Chapter 5

Anti-Apoptosis Engineering for Improved Protein

Production from CHO Cells
Eric Baek, Soo Min Noh, and Gyun Min Lee

Abstract
Improving the time integral of viable cell concentration by overcoming cell death, namely apoptosis, is one
of the widely used strategies for efficient production of therapeutic proteins. By establishing stable cell lines
that overexpress anti-apoptotic genes or down-regulate pro-apoptotic genes, the final product yields can
be enhanced as cells become more resistance to environmental stresses. From the selection of high-­
expressing clones to verification of anti-apoptotic activity, the method to construct a stable anti-apoptotic
cell line is discussed in this chapter.

Key words Chinese hamster ovary cells, Therapeutic proteins, Apoptosis, Transfection, Selection,
Suspension adaptation, Western blot analysis, Overexpression, siRNA, Cell culture

1  Introduction

Since the 1980s, mammalian cell lines have been used for ­large-­scale
commercial production of therapeutic proteins, including mono-
clonal antibodies and fusion proteins. In order to satisfy the
fast-growing economic demand of biopharmaceuticals, major
­
improvements have been made to establish high and stable pro-
duction of therapeutic proteins. As a result, a more than 100-fold
increase in product titers from Chinese hamster ovary (CHO) cells
has been achieved [1]. Due to several advantages over other mam-
malian cell lines, CHO cells have been the most widely used host
cell line for therapeutic protein production. CHO cells are known
to be a safe host which is efficient for a gene amplification system
with higher specific productivity and adequate for a serum-free sus-
pension culture [2].
For the cost-effective production of therapeutic proteins with
recombinant CHO (rCHO) cell lines, numerous techniques and
approaches have been used to maximize the production of high-­
quality therapeutic proteins. These approaches include the optimiza-
tion of the culture medium and feeding supplements [3], selection of

Paula Meleady (ed.), Heterologous Protein Production in CHO Cells: Methods and Protocols, Methods in Molecular Biology,
vol. 1603, DOI 10.1007/978-1-4939-6972-2_5, © Springer Science+Business Media LLC 2017

71
72 Eric Baek et al.

high-producing clones [4], manipulation of transcriptional activity

via vector engineering [5], and introduction or silencing of genetic
characteristics through cell line engineering [6], etc.
Improving the time integral of the viable cell concentration by
increasing the culture longevity, namely overcoming cell death, has
been one of the main strategies to enhance product yields. During
cell culture, rCHO cells are exposed to various kinds of stresses
such as nutrient deprivation, accumulation of waste products, and
shear stress. When cells become intolerant to such stresses, they
eventually undergo programmed cell death (PCD). Apoptosis, one
of the PCDs along with autophagy, has been widely studied in
rCHO cells using such tactics as delaying or inhibiting apoptosis
for prolonged culture longevity [7].
Apoptosis causes a structural breakdown as well as cell lysis as
an outcome of the caspase-dependent cascade. Briefly, environ-
mental stresses induce mitochondria to release cytochrome c to
recruit procaspases by forming a quaternary protein structure
called an apoptosome. Activated procaspases then trigger an apop-
tosis cascade. Together with caspases, the executioners of apopto-
sis, pro-, and anti-apoptotic members of the Bcl-2 family also have
an important role in the fate of apoptosis by regulating the release
of cytochrome c in the mitochondrial membrane [8]. Thus, many
studies have primarily been focusing on the down-regulation of
caspases and overexpression of anti-apoptotic protein members of
the Bcl-2 family to inhibit or attenuate apoptosis [9, 10].
In this chapter, the methods to construct a stable anti-­apoptotic
cell line will be discussed. In order to construct a stable apopto­
tic cell line, it is important to understand the pathway of apoptosis
and to identify targets to be manipulated. Among the anti-­apoptotic
genes, the overexpression of Bcl-2 or Bcl-xL has shown precise and
effective inhibition of apoptosis in rCHO cell cultures by enhanc-
ing the culture longevity and endurance to environmental stresses.
Consequently, the overexpression of Bcl-2 or Bcl-xL yields greater
therapeutic protein production, which is a definite economic
advantage in biopharmaceutical industry [9, 11]. In a similar con-
cept, down-regulation of caspases, such as capspase-8 and -9, and
knockout of pro-apoptotic genes, such as Bax and Bak, enhances
the viability in both batch and fed-batch cultures [10, 12, 13].
Down-regulation of such genes can be done with various genome-­
editing techniques, such as small interfering RNA (siRNA), zinc-­
finger nucleases (ZFNs), transcription activator-like effector
nucleases (TALENs), and the recently developed clustered regu-
larly interspaced short palindromic repeats (CRISPR)/CRISPR-­
associated (Cas) system [6, 14].
At the very beginning, one must decide which apoptosis-­
related gene to manipulate in the rCHO cell line. The most widely
targeted anti-apoptotic genes are Bcl-2 and Bcl-xL. However,
there are also reports of successfully enhanced culture longevity by
 Construction of Anti-Apoptotic Cell Line 73

overexpressing the following genes; Mcl-1, Aven, XIAP, and CrmA

[11, 15–17]. The down-regulation of pro-apoptotic genes, such as
caspase-3 and -7 via siRNA, and deletion of Bax and Bak using
zinc-finger nucleases or the CRISPR/Cas9 system were also
­successful in enhancing cell viability and improving resistance to
apoptosis [6, 13, 18, 19]. Next, one should design an expression
vector, in which the chosen apoptosis-related gene should be
incorporated, with the proper antibiotic resistance gene for selec-
tion of stable cell lines. After a stable transfection with the designed
vector, individual clones will survive against the selection marker
only if the expression vector was correctly introduced into the
genome. Among the clones that show high resistance to the selec-
tion marker, clones will be further narrowed down by selecting
clones with the highest expression of the anti-apoptotic gene or
the lowest expression of the pro-apoptotic gene. Throughout these
steps, the cells are cultured under adherent conditions for conve-
nient and precise selection; thus, the clones need to be adapted to
serum-­free suspension cultures as biopharmaceuticals commonly
require in the large-scale production of therapeutic proteins. Once
the clones are adapted to a serum-free suspension culture, one
needs to verify if both the product and apoptotic genes are prop-
erly expressed. Finally, one can carry out the desired cultures under
various kinds of conditions to check for cell viability, apoptotic
expression, and product titers.

2  Materials

2.1  Cell Culture 1. rCHO cell line (see Note 1).

2. Iscove’s modified Dulbecco’s medium (IMDM) (Invitrogen,
Carlsbad, CA) (see Note 2).
3. Dialyzed fetal bovine serum (dFBS) (Invitrogen) (see Note 3).
4.
Serum-free suspension media: SFM4CHO™ (Hyclone,
Logan, UT).
5. Hypoxanthine-thymidine (HT) supplement (Invitrogen).
6. 96-, 48-, 24-, 6-well sterile tissue culture plates (Nunc, Roskilde,
Denmark).
7. 25-, 75 cm2 T-flask (T25-, T75-flask) (Nunc).
8. 125 mL Erlenmeyer flask (Corning, Corning, NY).
9. Cell counter: Countess II® FL automated cell counter
(Invitrogen).
10. Humidified CO2 incubator (Forma Scientific, Marietta, OH).
11. Climo-shaking CO2 incubator (Adolf Kuhner AG, Birsfelden,
Switzerland, Cat. # ISF1-X).
74 Eric Baek et al.

12. Phosphate-buffered saline (PBS) [pH 7.4]: 0.01 M Phosphate

buffer, 0.0027 M potassium chloride, and 0.137 M sodium
chloride.
13. Trypsin (Invitrogen).

2.2  Transfection 1. Expression vector: for overexpression, pcDNA3.1/Hygro(+)

and Selection (Invitrogen); for down-regulation, pSilencer2.1-U6 hygro
(Invitrogen). It is important to select an antibiotic-resistance
gene that the original cell line does not contain, because we
want the selected antibiotic-resistance gene to be a selection
marker only for the gene of interest (see Note 4).
2. Hygromycin (Clontech, Mountain View, CA): the selection
marker should be selected according to the antibiotic-resistance
gene in the expression vector.
3. Lipofectamine® 2000 (Invitrogen).
4. Opti-MEM® I reduced serum medium (Invitrogen).

2.3  Verification 1. Lysis buffer: 20 mM Tris [pH 7.4], 5 mM ethylenediaminetet-

of Protein Expression raacetic acid (EDTA), 10 mM Na4P2O7, 100 mM NaF, 2 mM
Na3VO4, 1% NP-40, 1 mM PMSF, and 1× Roche protease
inhibitor cocktail (Roche, Indianapolis, IN).
2. NuPAGE LDS sample buffer (Invitrogen).
3. 1 M dithiothreitol (DTT) (Sigma-Aldrich, St. Louis, MO).
4. Heating block.
5. 4–12% Bis-Tris NuPAGE gel (Invitrogen).
6. NuPAGE MES SDS running buffer (Invitrogen).
7. Electrophoresis equipment to run gel and to transfer proteins
to membrane.
8. iBlot® 2 Dry blotting system (Invitrogen).
9.
Polyvinylidene difuoride membrane (PVDF) (Biorad,
Hercules, CA).
10. 3MM chromatography blotting papers (GE Healthcare, Little
Chalfont, UK).
11. Western blot transfer buffer: 0.025 M Tris–HCl, 0.192 M
­glycine, and 20% methanol.
12. Antibodies for immunoblotting: anti-human-Bcl-2, anti-β-­
actin (Sigma-Aldrich), anti-caspase-3 (Cell Signaling Tech­
nology, Danvers, MA), and HRP-conjugated anti-mouse IgG
(Koma Biotechnology, Seoul, South Korea).
13. PBS Tween 20 (PBS-T): add 1 mL Tween 20 (Sigma-Aldrich)
to 1000 mL PBS.
14. ECL Western blotting detection reagents (GE Healthcare).
 Construction of Anti-Apoptotic Cell Line 75

15. 20.3 × 24.5 cm medical X-ray film blue (Agfa HealthCare,

Mortsel, Belgium).
16. X-ray film cassette.

3  Methods

3.1  Transfection For overexpression of anti-apoptotic genes, one must insert the
target gene sequence into the expression vector pcDNA3.1/
Hygro(+) in the proper orientation (see Note 5). In a similar way,
the target sequence of the siRNA for a pro-apoptotic gene should
be designed and inserted into pSilencer2.1-U6 hygro. The
OligoEngine Workstation 2 program (OligoEngine, Seattle, WA)
is a useful tool for designing the target sequence (see Note 6).
From transfection to selection, the rCHO cell lines should be
maintained in an adherent state. The general overview of cell line
construction is illustrated in Fig. 1. In addition, the rCHO cell
lines should be cultivated in the media supplemented with 7–10%
dFBS in a humidified incubator.
1. One day prior to the transfection, seed the cells at a concentra-
tion of 0.5 × 106 cells/mL in T25-flasks. On the day of trans-
fection, the cells should be at 80–90% confluence.
2. Using Lipofectamine® 2000, transfect the rCHO cell line with
the expression vector according to the manufacturer’s ­protocol.
Briefly, prepare DNA diluted 8 μg in 500 μL of Opti-MEM
medium and 20 μL of Lipofectamine® 2000 in 500 μL of Opti-­
MEM medium, separately. After 5 min of incubation, combine
the diluted DNA and Lipofectamine® 2000 and incubate for
20 min at room temperature. Add the combined mixture of
DNA- Lipofectamine® 2000 to the T25-flask containing
the cells.
3. After 24–48 h of cultivation, discard the media into a waste
bottle.
4. After washing the cells with PBS, detach cells by trypsinization
using 1× trypsin in PBS.
5. Resuspend the cells with fresh IMDM and count the cells to
conduct pool selection.

3.2  Selection 1. Conduct pool selection by seeding the cells at a concentration

of 0.1 × 106 cells/mL in IMDM with the proper concentration
of antibiotic, hygromycin (see Note 7).
2. Exchange the media with fresh media every 3–4 days until the
cells reach confluence. When the cells reach 90–100% conflu-
ence, the pool selection is done (see Note 8).
76 Eric Baek et al.

Fig. 1 The general overview of cell line construction

3. Discard the media, wash the cells with PBS, and count the cells
after trypsinization.
4. Conduct limited dilution by seeding the cells at a concentra-
tion of 0.2–1.0 cells/well onto 96-well plates.
5. With a microscope, observe single-colony formation after
10–14 days; and if single colonies are observed and reach
 Construction of Anti-Apoptotic Cell Line 77

40–50% confluence, transfer the colonies to 48-well plates

(see Notes 9 and 10).
6. As the cells reach 90–100% confluence, expand the cells step-
wise (48 well → 24 well → 6 well) (see Note 11).
7. When the cells are ready to cultivate in the 6-well plates,
expand the cultivation into two wells; one for the Western blot
analysis to check for the expression level of the manipulated
gene and the other for maintenance. The methods for Western
blot analysis will be described in the verification step.
8. Expand the overexpressing clones (10–15 clones) in 6-well
plates into T25-flasks.
9. Do batch cultures. Determine the cell growth and confirm that
the manipulated gene is stably expressed as expected through-
out the culture (see Note 12).
10. Generate a frozen cell bank.

3.3  Serum-Free 1. Seed the cells at a concentration of 0.5 × 106 cells/mL to

Suspension 50 mL of suspension media in a 125 mL Erlenmeyer flask.
Adaptation 2. Count the cells on day 3 of cultivation.
3. If the viable cell concentration is greater than 1 × 106 cells/
mL, passage the cells with a concentration of 0.5 × 106 cells/
mL two more times. After two passages, if the viable cell con-
centration is maintained greater than 1 × 106 cells/mL when
counted on day 3, reduce the seeding concentration to
0.3 × 106 cells/mL.
If the viable cell concentration is less than 1 × 106 cells/mL,
passage the cells with a concentration of 0.5 × 106 cells/mL until
the viable cell concentration is maintained greater than 1 × 106
cells/mL when counted on day 3.
If the viable cell concentration is less than 0.5 × 106 cells/mL,
collect all the cells with centrifugation at 300 × g for 5 min. Discard
the supernatant media and seed the cells in fresh 50 mL of
­suspension media. If it is still difficult to maintain a viable cell con-
centration at more than 0.5 × 106 cells/mL when counted on day
3, add 7% dFBS to the suspension media and gradually decrease
the concentration of dFBS between passages until the growth is
stabilized without the addition of dFBS.
1. The seeding concentration can be reduced to 0.2 × 106 cells/
mL. Keep passaging the cells until the growth is stabilized.
2. When the cells are fully adapted to the suspension culture, do
batch cultures with daily sampling and monitor the growth
and expression level of the manipulated gene.

3.4  Verification The newly constructed serum-free adapted cell line is ready for
of Anti-­Apoptotic Gene testing its ability to inhibit apoptosis in various environments.
In addition, it is important to verify that the engineered gene is
78 Eric Baek et al.

correctly expressed in a suspension culture. A convenient method

to check overexpression or down-regulation is to detect the amount
of protein by Western blot analysis.

3.4.1  Sample 1. Count the cells and collect 1 × 106 cells.

Preparation 2. Pellet the cells by centrifugation at 14,000 × g and aspirate off
the media.
3. Wash the cell pellet with cold PBS and aspirate off the superna-
tant after centrifuging at 14,000 × g.
4. Lyse the cell pellet by gently resuspending in lysis buffer. One
milliliter of lysis buffer is used to lyse 107 cells (see Note 13).
5. Incubate the cells on ice for 30 min.
6. Centrifuge the cells at 4 °C for 10 min at 14,000 × g.
7. Transfer the supernatant into a fresh tube and discard the
pellet. Cell lysates can be stored at −70 °C until needed
­
(see Note 14).
8. Reduce the cell lysates by mixing the sample with 4× sample
buffer and 1 M DTT by a ratio of 6.5:2.5:1 (sample: sample
buffer: DTT).
9. Incubate the samples at 95 °C for 7 min.
10. Cool the samples on ice. Samples can be stored at −70 °C.

3.4.2  Gel Loading 1. Assemble the equipment for gel electrophoresis according to
the manufacturer’s instructions.
2. Load equal amounts of the samples into the SDS-PAGE gel
(see Note 15).
3. Perform gel electrophoresis running for 2 h at 100 V
(see Note 16).

3.4.3  Gel Transfer
to the Membrane
1. When gel electrophoresis is done, remove the gel from the
plastic cover.
2. Soak the gel in transfer buffer.
3. Prior to transfer, activate the PVDF membrane by soaking it in
methanol for 5 min.
4. Stack the membrane and the gel in a cassette as shown in
Fig. 2.
5. Perform electrophoresis for protein transfer for 2 h at 100 V
(see Note 17).

3.4.4  Antibody Binding 1. When the transfer is done, take out the membrane and soak in
and Detection PBS-T. The membrane can be stored at 4 °C until needed.
2. Block the membrane in blocking buffer for 1–2 h at room tem-
perature. Place it on the shaker at 30 rpm (see Note 18).
 Construction of Anti-Apoptotic Cell Line 79

Fig. 2 Gel and membrane assembly for electrophoresis transfer. Sandwiched

between blotting papers and sponges, PVDF membrane should be located on
anode side (positive) and SDS-PAGE gel should be located on cathode side
(negative)

3. Wash the membrane with PBS-T three times on the shaker at

60 rpm, 10 min each.
4. Incubate the membrane with the diluted primary antibody on
the shaker at 30 rpm overnight at 4 °C (see Note 19).
5. Wash the membrane with PBS-T three times on the shaker at
60 rpm, 10 min each.
6. Incubate the membrane with conjugated secondary antibody,
which is diluted in blocking buffer at a 1:2000 or 1:5000
ratio for 40 min to 1 h on the shaker at 30 rpm at room
temperature.
7. Wash the membrane with PBS-T three times on the shaker at
60 rpm, 10 min each.
8. Soak the membrane in 1 mL of ECL Western blotting detec-
tion reagent.
9. Place the membrane in Western blot cassette.
10. In dark room, expose the membrane to x-ray film and acquire
the image (see Note 20).
The expected outcome of Western blot analysis for Bcl-2 over-
expressed cell line is illustrated in Fig. 3. While the expression level
of β-actin is the same between clones and a negative control, Bcl-2
is highly overexpressed in the Bcl-2 overexpressed cell line.
It is important to check the expression level of the manipulated
protein; however, it is also noteworthy to check the expression
level of other apoptotic-related proteins, such as poly (ADP-ribose)
polymerase (PARP) and caspase-3, which are key apoptotic mak-
ers. In an anti-apoptotic cell line, a reduced level of apoptosis
should be reflected by the delayed cleavage of PARP and caspase-3.
In addition, Annexin V affinity assay and chromosomal DNA
80 Eric Baek et al.

Fig. 3 The expected outcome of Western blot analysis for Bcl-2 overexpressed
clones. Compared to the negative control (N), Bcl-2 is highly overexpressed in the
Bcl-2 overexpressed clones

f­ ragmentation are good methods to measure the level of apoptosis.

Along with Western blot analysis, these kinds of assays should be
done in order to verify that constructed cell line is apoptosis-
resistant.
Of course, the culture performance of the anti-apoptotic cell
line should be observed. The expected viable cell concentration,
viability, and product yield of the anti-apoptotic cell line are shown
in Fig. 4. Under various culturing conditions, including batch, fed-­
batch, and apoptosis stimulating conditions, the cell line should be
compared to a negative control cell line. Commonly used apoptosis-­
stimulating agents are hyperosmotic conditions by the addition of
salts or sugars to the medium and the addition of sodium butyrate
(NaBu). Both of these agents are known to increase the specific
protein productivity in rCHO cell lines; however, they have the
drawbacks of suppressing cell growth and inducing apoptosis [9, 20].
In these stressful environments, the beneficial effects of anti-­
apoptosis should be reflected by enhancements in cell growth,
viability, and protein titers.

4  Notes

1. CHO-K1, CHO-DUKX, and CHO-DG44 are the most

widely used CHO cell lines. CHO-K1 was derived from the
original Chinese hamster; and CHO-DUKX and CHO-DG44
are derivative forms of the original CHO cell line. CHO-­
DUKX lacks one dihydrofolate reductase (dhfr) allele, while
CHO-DG44 lacks two alleles. DHFR lacking cell lines require
glycine, hypoxanthine, and thymidine for growth [2].
The cell line of interest does not have to be a recombinant
therapeutic protein-producing cell line. However, if the cell
line is a producing cell line, either dhfr- or the glutamine syn-
thetase- (gs) system is used to amplify the expression of the
product gene. In the DHFR-amplification system, methotrex-
ate (MTX) (Sigma-Aldrich, Cat. # A6770) should be added as
 Construction of Anti-Apoptotic Cell Line 81

Fig. 4 The expected viable cell concentration, viability, and product yield of the
selected anti-apoptotic clones. Compared to the negative control (N), anti-­apoptotic
cell lines show improved viable cell concentration, viability, and product titer
82 Eric Baek et al.

an inhibitor of the dhfr gene. Likewise, in the GS-­amplification

system, in addition to selection in the medium without gluta-
mine, methionine sulphoximine (MSX) (Sigma-Aldrich, Cat.
# M5397) should be added as an inhibitor of the gs gene.
2. Make sure high glucose and l-glutamine are included in
IMDM. IMDM and dialyzed FBS are used for the CHO-­DG44
and CHO-DUKX cell lines, while Dulbecco’s modified Eagle
medium (DMEM) and FBS are used for the CHO-K1 cell line.
3. Dialysis reduces small molecules, which can cause interference
during the selection of rCHO cells using the DHFR system.
4. Neomycin, zeocin, and blasticidin are other selection markers
that are widely used. If one wants to co-overexpress or co-­
down-­regulate multiple genes of interest, different selection
markers should correspond to each gene.
5. Map of the pcDNA3.1 vectors are provided at http://www.
thermofisher.com; target gene sequences can be searched for at
the NCBI site (http://www.ncbi.nlm.nih.gov) or because the
CHO-K1 and Chinese hamster genome sequences are now
well characterized, target mRNA sequences can be more easily
searched for at the CHO genome site (http://chogenome.org).
6. In order to evaluate the expression of the gene of interest in an
engineered cell line, it is important to have a negative control
cell line by transfecting a null vector. The null vector should
only contain the antibiotic-resistance gene without the gene of
interest.
7. The optimal concentration of antibiotics should be decided by
drawing a kill curve. As shown in Fig. 5, various concentrations
of antibiotics should be tested on untransfected cells, and the
minimal concentration that inhibits the growth of the cells is
the optimal concentration for selection. For CHO cell lines,
the optimal concentration approximately ranges from 250 to
500 μg/mL for hygromycin, from 300 to 500 μg/mL for zeo-
cin, from 400 to 600 μg/mL for G418, and from 5 to 10 μg/
mL for blasticidin.
8. It may take 2–3 weeks to reach 100% confluence. If the flask
becomes confluent within 1 week, it is more likely that the
antibiotic concentration is too low, and the target gene may
not be expressed. In this case, the concentration of antibiotic
should be changed.
9. During cultivation in 96-well plates, it is important to observe
and mark the single colonies before they become too much
confluence. Otherwise, it is hard to differentiate the single
colonies from double or triple colonies.
10. If too many wells show colonies of double, triple, or more,
decrease the seeding concentration in step 4. If too many wells
 Construction of Anti-Apoptotic Cell Line 83

Fig. 5 A kill curve to optimize the concentration of antibiotics on CHO cells. The minimal concentration that
inhibits the growth of the cells is the optimal concentration for selection. In this graph, 300 μg/mL of antibiotics
is the optimal concentration for selection

show no colonies, increase the seeding concentration. The

seeding concentration that yields 10–20 single colonies per
one plate is an appropriate condition; seeding 3–5 plates
according to the optimized seeding concentration should be
enough for the selection process.
11. Depending on the stability of the antibiotics, the media should
be exchanged several times during the limiting dilution pro-
cess. In such a case, it is important not to disturb and detach
the colonies from the plate so that single colonies can be
identified.
12. If the gene of interest is not stably expressed throughout the
culture, the process must be redone starting from the initial
transfection step.
13. It is important to keep everything, including the cell pellet and
lysis buffer, on ice to reduce degradation of the proteins.
84 Eric Baek et al.

14. Although the expression level of β-actin, a housekeeping gene,

is used as an internal control, it is important to load equal
amounts of proteins between the clones and negative controls.
Thus, it is necessary to measure the concentration of proteins
at this step. Next, determine how much protein to load and
add an equal volume of protein by diluting with lysis buffer.
15. The maximum loading volume is 25 and 37 μL for a 10-well
gel with a thickness of 1.0 and 1.5 mm, respectively. It is 15
and 25 μL for a 15-well gel with a thickness of 1.0 and 1.5 mm,
respectively.
16. Two hours of electrophoresis for a thin gel may be too much,
and the proteins may run out of the gel; thus, depending
on the thickness of the gel, the running time should be
optimized.
17. The transfer can also be done in dry condition using iBlot®
two-dry blotting system. Stack the gel and the membrane and
run the device according to the manufacturer’s instruction.
The running time usually takes 7 min.
18. It is important to optimize the blocking buffer to avoid the
nonspecific binding of antibodies. Typically, 3–5% bovine
serum albumin (BSA) or 3–5% skim milk in PBS-T is used
as a blocking buffer. Usually, BSA works better for
phosphoproteins.
19. The primary antibody is usually diluted in blocking buffer at
1:500 or 1:1000 ratio. The primary antibody diluted in block-
ing buffer can be reused multiple times. Most antibodies, in
our case, can support up to 8–10 cycles of freeze and thaw.
It is important to store the primary antibody at −20 °C.
20. The exposure time must be optimized for a clear detection of
signals. For the detection of housekeeping proteins like β-actin,
an exposure time of 10–15 s is enough. However, for apop-
totic-proteins like Bcl-2, the optimal exposure time ranges
from 3 to 10 min.

References

1. Kim JY, Kim YG, Lee GM (2012) CHO cells antibody-­dependent cellular cytotoxicity.
in biotechnology for production of recombi- Biotechnol Bioeng 87:614–622
nant proteins: current state and further poten- 3. Hacker D, Jesus MD, Wurm FM (2009)
tial. Appl Microbiol Biotechnol 93:917–930 25 Years of recombinant proteins from reactor-­
2. Yamane-Ohnuki N, Kinoshita S, Inoue-­ grown cells – where do we go from here?
Urakubo M, Kusunoki M, Lida S, Nakano R, Biotechnol Adv 27:1023–1027
Wakitani M, Niwa R, Sakurada M, Uchida K, 4. Borth N, Zeyda M, Katinger H (2000) Efficient
Shitara K, Satoh M (2004) Establishment of selection of high-producing subclones during
FUT8 knockout Chinese hamster ovary gene amplification of recombinant Chinese
cells: an ideal host cell line for producing com- hamster ovary cells by flow cytometry and cell
pletely defucosylated antibodies with enhanced sorting. Biotechnol Bioeng 71:266–273

5. Kim JD, Yoon Y, Hwang HY, Park JS, Yu S,
Lee J, Baek K, Yoon J (2005) Efficient selec-
tion of stable Chinese hamster ovary (CHO)
cell lines for expression of recombinant pro-
teins by using human interferon beta SAR ele-
ment. Biotechnol Prog 21:933–937
6. Grav LM, Lee JS, Gerling S, Kallehauge TB,
Hansen AH, Kol S, Lee GM, Pedersen LE,
Kildegaard HF (2015) One-step generation of
triple knockout CHO cell lines using CRISPR/
Cas9 and fluorescent enrichment. Biotechnol
J 10:1446–1456
7. Hwang SO, Lee GM (2008) Nutrient depriva-
tion induces autophagy as well as apoptosis in
Chinese hamster ovary cell culture. Biotechnol
Bioeng 99:678–685
8. Arden N, Betenbaugh MJ (2004) Life and
death in mammalian cell culture: strategies for
apoptosis inhibition. Trends Biotechnol 22:
174–180
9. Kim NS, Lee GM (2000) Overexpression of
bcl-2 inhibits sodium butyrate-induced apop-
tosis in Chinese hamster ovary cells resulting in
enhanced humanized antibody production.
Biotechnol Bioeng 71:184–193
10. Yun CY, Liu S, Lim SF, Wang TW, Chung
BYF, Teo JJ, Chuan KH, Soon ASC, Goh KS,
Song Z (2007) Specific inhibition of caspase-8
and -9 in CHO cells enhances cell viability in
batch and fed-batch cultures. Metab Eng
9:406–418
11. Figueroa B, Chen S, Oyler GA, Hardwick JM,
Betenbaugh MJ (2004) Aven and Bcl-xL
enhance protection against apoptosis for mam-
malian cells exposed to various culture condi-
tions. Biotechnol Bioeng 85:589–600
12. Lim SF, Chuan KH, Liu S, Loh SOH, Chung
BYF, Ong CC, Song Z (2006) RNAi suppres-
sion of Bax and Bak enhances viability in
­fed-­batch cultures of CHO cells. Metab Eng
8:509–522
Construction of Anti-Apoptotic Cell Line 85
13. Cost GJ, Freyvert Y, Vafiadis A, Santiago Y,
Miller JC, Rebar E, Collingwood TN, Snowden
A, Gregory PD (2010) Bak and Bax deletion
using zinc-finger nucleases yields apoptosis-­
resistant CHO cells. Biotechnol Bioeng
105:330–340
14. Silva G, Poirot L, Galetto R, Smith J, Montoya
G, Duchateau P, Paques F (2011) Mega­
nucleases and other tools for targeted genome
engineering: perspectives and challenges for
gene therapy. Curr Gene Ther 11:11–27
15. Reynolds JE, Yang T, Qian L, Jenkinson JD,
Zhou P, Eastman A, Craig RW (1994) Mcl-1, a
member of the Bcl-2 family, delays apoptosis
induced by c-Myc overexpression in Chinese
hamster ovary cells. Cancer Res 54:6348–6352
16. Majors BS, Betenbaugh MJ, Pederson NE,
Chiang GG (2009) Mcl-1 overexpression leads
to higher viabilities and increased production
of humanized monoclonal antibody in Chinese
hamster ovary cells. Biotechnol Prog 25:
1161–1168
17. Sauerwald TM, Oyler GA, Betenbaugh MJ
(2003) Study of caspase inhibitors for limiting
death in mammalian cell culture. Biotechnol
Bioeng 81:329–340
18. Kim NS, Lee GM (2002) Inhibition of sodium
butyrate-induced apoptosis in recombinant
Chinese hamster ovary cells by constitutively
expressing antisense RNA of caspase-3.
Biotechnol Bioeng 78:217–228
19. Sung YH, Lee JS, Park SH, Koo J, Lee GM
(2007) Influence of co-down-regulation of
caspase-3 and caspase-7 by siRNAs on sodium
butyrate-induced apoptotic cell death of
Chinese hamster ovary cells producing throm-
bopoietin. Metab Eng 9:452–464
20. Han YK, Kim YG, Kim JY, Lee GM (2010)
Hyperosmotic stress induces autophagy and
apoptosis in recombinant Chinese hamster ovary
cell culture. Biotechnol Bioeng 105:1187–1192
 Chapter 6

Conditional Knockdown of Endogenous MicroRNAs in CHO

Cells Using TET-ON-SanDI Sponge Vectors
Alan Costello, Nga Lao, Martin Clynes, and Niall Barron

Abstract
MicroRNAs (miRNAs) are small, noncoding RNAs of about 22 nucleotides in length and have proven to
be useful targets for genetic modifications for desirable phenotype in the biotech industry. The use of
constitutively expressed “miRNA sponge” vectors in which multiple, tandem miRNA binding sites con-
taining transcripts are transcriptionally regulated by a constitutive promoter for down regulating the levels
of endogenous microRNAs in Chinese hamster ovary (CHO) cells has shown to be more advantageous
than using synthetic antisense oligonucleotides. The application of miRNA sponges in biotechnological
processes, however, could be more effective, if expression of miRNA sponges could be tuned. In this chap-
ter, we present a method for the generation of stable CHO cell lines expressing a TET-ON-SanDI-­
miRNA-sponge that is in theory expressed only in the presence of an inducer.

Key words miRNA, Sponge, Knockdown, Tetracycline, TET-on, Inducible, CHO

1  Introduction

Chinese Hamster Ovary (CHO) cells have been the workhorse of

the biopharmaceutical industry since their development in the
1950s [1]. Process optimization is a major area of interest in the
CHO community. This includes numerous factors of the fed-batch
process, such as vessel design, media composition, temperature
shift, online monitoring of process variables (pO2, CO2, nutrient,
and glucose concentration) to name a few. On a cellular level the
standard industry approach is to screen a large panel of clones for
desired traits. However, the process of single-cell cloning is both
time consuming and gives no information on why one clone is
outperforming the others.
MiRNAs with their ability to impact on and regulate complex
cellular pathways and alter phenotypes have become a popular
genetic engineering tool [2]. Traditionally, miRNA loss of func-
tion studies would be carried out using chemically modified oligo-
nucleotides (antagomirs or anti-miRs) that inhibit endogenous

Paula Meleady (ed.), Heterologous Protein Production in CHO Cells: Methods and Protocols, Methods in Molecular Biology,
vol. 1603, DOI 10.1007/978-1-4939-6972-2_6, © Springer Science+Business Media LLC 2017

87
88 Alan Costello et al.

miRNAs by directly binding miRNAs, hence rendering them

­inactive. However, application of anti-miRs in biotechnological
processes is impractical, as the effect is transient. MiRNA sponges,
abundantly transcribed transcripts containing several copies of the
miRNA-binding sites developed nearly a decade ago [3] as an
effective method for scavenging endogenous miRNA in vivo have
been shown to be as effective as the present antisense technology,
and their activities are specific to miRNA seed families. Further
development of the miRNA sponges by the use of the restriction
enzyme SanDI that generates non-palindromic overhangs upon
digestion allowing the one-step generation of miRNA sponges
containing large numbers of miRNA-binding sites. A high number
of miRNA binding sites has been shown to be more efficient in
silencing of endogenous target miRNAs [4]. Constitutive stable
expression of miRNA sponges in CHO cells has been successful
and demonstrates that they can be used as potential tools to gener-
ate desirable phenotypes in CHO cells [5, 6]. However, the levels
of specific miRNAs are tightly regulated in response to the cell’s
needs. MiRNAs that positively impact on specific productivity or
product quality but have a negative effect with regard to growth in
a constitutive expression system are of no real benefit in recombi-
nant protein production processes. Therefore by mimicking nature
and timely tuning of the level of specific miRNAs at specific stages
of culture, analogous to the use of inducible promoter, one could
potentially improve a product yield or growth beneficial pheno-
type, more efficiently respond to biotechnological processes.
The need for the tight regulation of expression of specific
genes has been around for decades. Gene function studies would,
in principle, benefit greatly from not only a controllable on/off
expression but also controlled expression at a defined level. The
inducible control of gene activity by means of heat shock, heavy
metal ions [7, 8], or hormones [9] has been described; however,
many of these systems are limited by their “leakiness” [8] and ple­
io­tropic effects [10]. A tetracycline-responsive promoter was
developed by Gossen and Bujard in 1992 [11]. This system is
based on the Tn10-specified tetracycline-resistance operon of
Escherichia coli (E. coli) [12], wherein the transcription of resis-
tance genes is negatively regulated by the tetracycline repressor
(tetR). In the presence of tetracycline, tetR is unable to bind to the
promoter region of the operon, allowing transcription of the resis-
tance genes. By attaching the tetR to the C-terminal of VP16 from
herpes simplex virus (HSV), a hybrid trans-activator was created
that stimulates minimal promoters fused with a tetracycline opera-
tor (tetO), a promoter was generated that is silent in the presence
of even very low levels of tetracycline, due to the prevention of the
tetracycline-controlled trans-activator (tTA) binding to the (tetO),
but can be induced ~1000 fold. This is referred to as the TET-off
or tTA-dependent system. In 1995, Gossen et al. [13] explored
 TET-ON-SanDI-miRNA Sponge 89

the possibility of reversing the effect of tetracycline and its

­derivatives on the tetO. Random mutagenesis of the Tn10 tetR
gene was used to identify the sequence responsible for tetR bind-
ing to tetO in the presence of tetracycline that was named reverse
tTA (rtTA) [14]. This resulted in the TET-ON system, an induc-
ible promoter, tuneable with respect to doxycycline (Dox) concen-
tration. The leaky expression of TET inducible systems, however,
is well known [11]. Efforts to limit the extent of basal expression
have included: titration of vector, multi-vector approaches, rtTA
promoter optimization [16], and incorporation of a transcriptional
repressor/silencer [17].
For the conditional knockdown of endogenous miRNA exp­
ression, we take advantage of the “TET-ON” system described
above. In this chapter, we outline methods to generate TET-ON-
SanDI-­miRNA sponge constructs containing a desired number of
miRNA-­binding sites and to characterize stable CHO cells express-
ing the inducible miRNA-sponges that are expressed in the pres-
ence of doxycycline inducer, a tetracycline derivative.

2  Materials

2.1  Cloning and PCR 1. Oligos: purchase from Integrated DNA Technology (IDT) or
Screening MWG Eurofins.
2. E. coli DH5α for routine subcloning kit (Invitrogen).
3. SOC medium.
4. Ampicillin, sodium salt: Prepare 100 mg/mL stock by dissolve
1 g of ampicillin in sterilized deionized water to a final volume
of 10 mL. Store aliquots at −20 °C. One mL of 100 mg/mL
stock is used for 1 L of medium to achieve a final concentration
100 μg/mL.
5. Luria Bertani (LB): Dissolve 25 g of powder LB in deionized
water to a final volume of 1 L. Autoclave.
6. Luria Bertani-Agar (LB): Dissolve 25 g of powder LB in deion-
ized water to a final volume of 1 L. Add 15 g agar. Autoclave.
Allow to cool before adding antibiotic.
7. Restriction enzymes (Thermo Fisher Scientific or New England
BioLabs), store at −20 °C.
8. Fast Alkaline Phosphatase (Thermo Fisher Scientific).
9. T4 Polynucleotide Kinase (New England BioLabs).
10. T4 ligase (Roche).
11. 10× T4 Ligation Buffer (New England BioLabs).
12. MyTaq Red DNA Polymerase (Bioline).
13. Petri Dishes.
90 Alan Costello et al.

14. Plasmid Mini-prep kit and Midi-prep kit.

15. Qiagen Quick Gel Extraction kit.
16. PCR cleaning kit.
17. 50× Tris-Acetate-EDTA (TEA): For routine electrophoresis,
dilute 10 mL with 490 mL water (1× TEA).
18. Agarose Gel; for a 0.8% gel mix 0.8 g of agarose powder with
100 mL of 1× TEA buffer. Microwave for 1–2 min. Cool down
the solution for 5 min and add 8 μL Safe View Nucleic Stain
(NBS Biologicals). Pour solution into the gel-casting tray and
insert comb. Allow the gel to solidify.
19. DNA Gel Loading Dye (6×).
20. 100 bp DNA Ladder.
21. Heat block.
22. Water bath.
23. PCR machine.
24. Orbital shaker.
25. Incubator.
26. Gel running unit.
27. Trans-illuminator.
28. NanoDrop 2000 (Thermo Fisher Scientific).

2.2  Stable CHO Cell 1. TET-ON-SanDI-sponge vector (in house made) or basic vec-
Line Development tors can be purchased from Addgene and modified.
2. CHO cells: ATCC.
3. Medium for transfection: Dulbecco’s Modified Eagle’s Medium
(DMEM)/Nutrient Mixture F12-Ham and CHO-S-­ SFM
(Thermo Fisher Scientific).
4. Standard Fetal Bovine Serum (FBS).
5. 100× EDTA: add 5.3 mL 0.5 M EDTA to 44.7 mL water.
6. PBS buffer: Dissolve 8 g NaCl, 0.2 g KCl, 1.44 g Na2HPO4,
and 0.24 g KH2PO4 in 800 mL water, adjust pH to 7.4. Adjust
volume to 1 L with water. Autoclave.
7. 10× Trypsin.
8. Trypsin-EDTA solution: For 500 mL: add 50 mL 10× Trypsin,
and 5 mL 100 X EDTA into 445 mL PBS.
9. Doxycycline hyclate (DOX). To make 50 mg/mL solution:
Dissolve 1 g in 20 mL sterilized water. Store at −80 °C. To
make 100 μg/mL solution, add 1 mL 50 mg/mL solution to
49 mL sterilized water. Store 1 mL aliquots at −20 °C.
10. Polyvinyl alcohol (PVA) (Sigma Aldrich) stock solution: weigh
12.6 g, add water to make up to 100 mL. Autoclave. Use 2 mL
for 100 mL of SFM.
 TET-ON-SanDI-miRNA Sponge 91

11. Transfection reagent: TransIT-X2® Dynamic Delivery System.

12. Guava EasyCyte (EMD Millipore).
13. Kuhner shaker (Climo-Shaker IDF1-XC) at 170 rpm, 80%
humidity, 5% CO2.
14. Fluorescence-Activated Cell Sorter (FACS).

3  Methods

3.1  Cloning The SanDI sponge oligos contain two miRNA-binding sites
and Screening (MBS) with SanDI overhangs for cloning and a 4–5 nucleotide
of E. coli (nt) spacer between (Fig. 1). The nucleotide composition of the
Transformant Clones spacer can be altered as required. The MBS is a sequence comple-
mentary to the mature miRNA of interest with a 3 bp mismatch
3.1.1  Design SanDI and one nucleotide deletion starting at base 9–12 from the 5′ end
Sponge Oligos for Cloning of the miRNA. This creates a bulge, inhibiting AGOII, a compo-
into the Backbone Vector nent of the RISC complex from degrading the transcript. Mature
miRNA sequences can be found on miRNA databases such as miR-
Base (http://www.mirbase.org/) or in the literature [15].
1. Obtain the mature miRNA sequence from miRBase (http://
www.mirbase.org/) or from the literature.
2. Generate the reverse complement to obtain one MBS at
http://www.bioinformatics.org/sms/rev_comp.html.
3. Manually modify to add mismatched nucleotides, avoiding the
eight nucleotides of the seed region.
4. Add a 4–5 nucleotide spacer and a second MBS unit. This is
the sense sponge oligo.
5. Generate the reverse complement of the sense sponge oligo to
obtain the antisense sponge oligo.
6. Add overhang for cloning into SanDI site.
7. Optional: Input the sense-bulged sponge oligo into the online
miRNA prediction tool http://genie.weizmann.ac.il/pubs/
mir07/mir07_prediction.html or STarMir ohttp://sfold.
wadsworth.org/cgi-bin/starmir.pl. In both web sites, a lower
ddG or ∆G total is expected for the perfectly matched MBS
than the bulged MBS.
8. Oligos can be ordered from any companies that provide oligos
for standard polymerase chain reaction (PCR). Input the
sequence in the 5′–3′ orientation.
9. The same procedure is performed to design negative sponge
oligos using a scrambled (non-miRNA targeting) sequence
(Fig. 1b).
Fig. 1 Cloning oligos into TET-ON-SanDI-sponge vector. (a) Illustrates the Tetracycline inducible sponge vector
with a Tetracycline inducible promoter (tetO) transcriptionally regulates expression of an unstable green fluores-
cent protein (d2GFP) reporter gene. The transcriptional factor rtTA3 presents in the same vector and constitu-
tively expressed. An antibiotic marker (HYG) allows for the selection of stable CHO cell lines. Primers for screening
of sponge insert are red arrow heads. (b) SanDI site with its overhang bases is in red color. (c) An example of a
sponge oligo duplex design with scramble nucleotides (capital letters) for NC (top two lines) or two binding sites
for CHO miR-204 (cgr-miR-204) (bottom three lines). Overhang bases in the duplex for cloning into vector are in
red color, and the spacer sequences between each binding site are in green. The blue turquoise color letters are
modified nucleotides for imperfect pairing of a miRNA and the miRNA sponge at the site of AGOII cleavage.
(d) PCR screening for sponge inserts of different size. Amplicons from different transformant E. coli clones (lanes
2–9), 100 bp DNA ladder (lanes 1 and 10)
 TET-ON-SanDI-miRNA Sponge 93

3.1.2  Cloning SanDI-­ For Oligo Annealing and Phosphorylation (see Note 1).
Sponge Oligos
1. Resuspend the oligos in nuclease-free water to give a 100 μM
into Backbone
stock solution.
TET-ON-­SanDI-  Vector
2. In a clean 0.2 mL PCR tube make up the following reaction:
1 μL 100 μM Oligo 1
1 μL 100 μM Oligo 2
1 μL 10× T4 Ligation buffer
6.5 μL Nuclease-free water
0.5 μL T4 Polynucleotide Kinase (PNK)

Incubate at 37 °C for 30 min.

3. On a preheated heat block, incubate tubes at 100 °C for 5 min.
Spin down tubes to bring condensation in the lids to the bot-
tom of the tubes.
4. Return tubes to the heat block, turn off the heat block, and
allow the reaction to cool slowly in the heat block for at least
2 h or until the temperature in the heat block drops below
40 °C.
5. Store at −20 °C.
6. The oligo duplex should be diluted 1:3, to a concentration of
300–400 ng/μL, with nuclease-free water prior to its use in a
ligation reaction.
For vector digestion and de-phosphorylation
7. In a 0.2 mL PCR tube make up the following reaction:
5 μL TET-ON-SanDI-HYG vector (5 μg)
10 μL Fast Digest buffer
5 μL Fast Digest SanDI
5 μL Fast Alkaline phosphatase
75 μL Nuclease-free water

8. Incubate for 1 h at 37 °C.

9. Clean the digested vector using a PCR cleaning kit, according
to the supplier’s instructions.
10. Verify the concentration of the digested vector using a
Nanodrop. Store at −20 °C.
94 Alan Costello et al.

11. In a 0.2 mL PCR tube make up the ligation reaction as

follows:
1 μL TET-ON-SanDI-HYG digested above (50 ng)
1 μL Oligo Duplex
1 μL 10× T4 Ligase Buffer
1 μL T4 Ligase
7 μL Nuclease-free water

12. Incubate overnight at 16 °C in a water bath.

13. Store at −20 °C or use immediately for transformation in the
following step.
14. To a 1.7 mL microcentrifuge tube add 50 μL of competent
DH5α subcloning efficiency and 5 μL of the ligation mix from
above.
15. Incubate for 30 min on ice.
16. Heat shock the cells for 30 s at 42 °C (water bath should be
warmed in advance).
17. Incubate on ice for 2 min.
18. Add 500 μL of SOC.
19. Revive the cells by incubating at 37 °C for 1 h with shaking
(220 rpm).
20. Pellet the cells by centrifuging at 4000 × g for 3 min.
21. Decant all but 50 μL of supernatant.
22. Resuspend the pellet and plate the total 50 μL of cells on an
LB agar plate containing ampicillin.
23. Incubate plates upside down overnight at 37 °C.

3.1.3  Identification To identify colonies positive for oligo inserts a PCR-screening

and Verification method is used. This provides a fast method of screening a large
of the Transformant Clones number of clones.
1. On an ampicillin-containing LB agar plate draw a grid of 8–16
squares and number each section 1–8/16.
2. Add 3 μL of nuclease-free water to 250 μL PCR tubes.
3. Pick single colonies from the LB plates. Suspend each single
colony in 3 μL nuclease-free water. Take 2 μL of this and spread
it in one square of the gridded plate. Each PCR reaction cor-
responds to a square on the gridded plate. Incubate plates
overnight at 37 °C.
 TET-ON-SanDI-miRNA Sponge 95

4. Perform PCR reaction with 1 μL cells. Make a PCR master mix
with the following components per reaction:
5 μL 2× MyTaq reaction mix
0.5 μL 10 μM forward primer
0.5 μL 10 μM reverse primer
3 μL Nuclease-free water

Mix well by gently pipetting.

5. Add 9 μL of master mix to each reaction tube that contain
1 μL of cell suspension.
6. Perform PCR as follows:
Initial denaturation 94 °C, 1 min
25 cycles 94 °C, 15 s; 55 °C, 15 s; and 72 °C, 1 min
Hold 4 °C, ∞

7. Run all 10 μL of PCR reactions on a 0.8% agarose gel (Fig. 1c).

8. Miniprep the corresponding clones containing the number of
MBS desired (based on fragment length on gel), according to
the kit provided.
9. Quantify the concentration of the plasmid DNA using a
Nanodrop.
10. Verify the number of oligo inserts in plasmid miniprep by
sequencing (see Notes 2 and 3).
11. Store plasmid at −20 °C.

3.2  Generation The choice of transfection methods and transfection reagents

of Stable CHO Cell depends on the CHO cell line. For most commonly used CHO
Lines Expressing lines, TransIT-X2® Dynamic Delivery System is satisfactory using
the  TET-ON-­SanDI- the protocol provided by the supplier. Transfection is done with
miRNA Sponge duplicates for both negative control (NC) and miRNA sponge vec-
tors using 6-well plates, with parental cells treated with transfec-
3.2.1  Generation tion reagent as a negative control as follows:
of Transfected CHO Cells
1. Plate cells at a density of 1.5 × 105 cells per well a day in advance
on DMEM/Ham’s F12 medium containing 1% (v/v) FBS.
The confluence of cells at the time of transfection should be
about 60%.
2. Perform transfection as recommended by the supplier using
1 μg DNA: 1.5 μL transfection reagent ratio for each well (this
ratio may require optimization depending on the CHO cell
line).
96 Alan Costello et al.

3. Optional: Estimate transfection efficiency using epifluorescence

microscope 16–24 h post transfection of cells transfected with
a positive control CMV-GFP vector.
4. Replace with fresh medium containing selective antibiotic 48 h
post transfection (see Note 4).
5. Replace with selective medium twice per week, allowing the
antibiotic resistant cells to expand.
6. Expand resistant cells into T25 flask. Store one half of the pool
of transfected cells.
7. Adapt the second half of the pool of transfected cells into sus-
pension culture conditions if desired by following previously
published protocol [18].

3.2.2  Induction Testing
(See Note 5)
DOX is quite toxic to CHO cell viability and unstable at 37 °C;
therefore, a range of concentrations of DOX (from 10 ng to
2000 ng/mL) should be tested to determine the concentration of
DOX suitable for the induction of the CHO cell line of interest.
This is carried out in 24-well plates containing a 1 mL volume of
media per well as follows:
1. Dilute DOX to concentrations of 0, 10, 100, 1000 ng per
10 μL and 2000 ng per 20 μL.
2. Add 10 μL or 20 μL of each Dox concentration to triplicate
wells on days 0, 2, and 4 of CHO cell culture at a density of
2 × 105 cells/mL.
3. Take cell samples each day up to day 8 to measure cell growth
and viability using the Guava Easycyte “Viacount” programme
(Fig. 2a).
4. Induction testing should be performed using the CHO cell
pools expressing the NC. Carry out the experiment as described
above. Induction of the total mixed population reflected by
the Mean Fluorescence Intensity (MFI) value is read by the
GFP_Plus program (Guava EasyCyte) (Fig. 2b). Flow Cyto­
metry (FACS) is used to obtain sub-pools for further charac-
terization in Subheading 3.2.3, with CHO parental cells used
for gating of negative GFP CHO population.

Fig. 2  (continued) induction in CHO DP12 cells. Stable negative control (NC) sponge and miR-204-sponge
expressing cell lines were tested in parallel for induction by the addition of Dox at concentrations of 0, 10, 100,
and 1000 ng/mL, to growth media on day 2, 4, 6 of culture. The % of total GFP-positive population and the mean
fluorescence intensity (MFI) of NC-sponge cell line and miR204-sponge cell line are shown (third panel and
bottom panel, respectively). Statistics were carried out by a two-tailed homoscedastic Student t-test, (*p ≤ 0.05,
**p ≤ 0.01, ***p ≤ 0.001). Experimental replicates (n = 3)
 TET-ON-SanDI-miRNA Sponge 97

a 1E+07
9E+06 0 ng/mL
10 ng/mL
8E+06
100 ng/mL
7E+06
1000 ng/mL
6E+06
VCD /ml 2000 ng/mL
5E+06
4E+06 **
3E+06
2E+06
** *** ***
1E+06
*** *** ***
0E+00
0 2 4 6 8 10
Time (days)
110
100
90
80
70
% Viability

60 **
50
0 ng/mL ***
40
10 ng/mL
30
100 ng/mL
20
1000 ng/mL
10
2000 ng/mL
0
0 2 4 6 8 10
b Time (days)
100
NC-spg Day2
% of Population GFP(+)

80 204-spg Day2
NC-spg Day4
204-spg Day4 **
60
NC-spg Day6
204-spg Day6 ** *
40
**
20

0
0 ng/mL 10 ng/mL 100 ng/mL 1000 ng/mL
[Doxycycline]
600
NC-spg Day2 **
500 204-spg Day2
NC-spg Day4
400 204-spg Day4
300 NC-spg Day6
MFI

204-spg Day6
200
100 **
*
0
0 ng/mL 10 ng/mL 100 ng/mL 1000 ng/mL
[Doxycycline]

Fig. 2 Evaluating the effect of Doxycycline on CHO DP12 cell growth and viability and induction testing.
(a) DP12 cells were grown in media supplemented with different concentrations of Doxycycline 0, 10, 100,
1000, and 2000 ng/mL at day 0 of culture at a density of 2 × 105 cells/mL. The effects of these doxycycline
concentrations on cell growth reflected by Viable Cell Density (VCD) value (top panel) and viability (second panel)
are shown. Statistics were carried out by a two-tailed homoscedastic student t-test (*p ≤ 0.05, **p ≤ 0.01,
***p ≤ 0.001). Experimental replicates (n = 2). (b) The TET-ON-SanDI-miRNA sponge vector was tested for
98 Alan Costello et al.

3.2.3  Induction The sub-pools of transfected cells obtained using FACS can ­initially
for Phenotyping be characterized at the molecular level to determine the level of
and Molecular Analysis endogenous miRNAs and GFP. These characterizations should be
carried out in parallel for both sets of pools, i.e., miRNA-sponge
and NC sponge, in the absence and presence of inducer. Induction
of TET-ON systems can be achieved with as little as 10 ng/mL
DOX (Fig. 2b) which is not detrimental to CHO cells. Experiments
can be set up in 5 mL of suspension cells in a 50 mL reactor tube
for convenience. However, this experiment can also be used in
attached cells if desired. Each cell pool (miRNA sponge and NC)
and condition (induced and un-induced) should be done in trip­
licates. Total RNA can be prepared using the standard Triazol
reagent and protocol which is followed by the reverse transcription
reaction using the TagMan® MicroRNA Reverse Transcription kit.
The levels of endogenous miRNAs can then be determined using
the miRNA specific Taqman assays. The choice of endogenous
controls is extremely important to avoid variations among samples;
therefore, commonly used RNA references (e.g., 5S RNA, U6
small noncoding RNA) should be tested before use to normalize
the signals in the miRNA specific Taqman assays. Either β-actin
or Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) can be
used a reference for endogenous genes. Subsequently, a wide range
of bioprocess-related phenotypes of CHO cell pools or sub-pools
expressing the inducible sponge vectors can be further character-
ized such as cell proliferation, life span of the CHO culture, bio-
pharmaceutical product quality, or stress response.
Functional characterization of single, stable clones expressing
the TET-ON miRNA sponge can be isolated by FACS and ana-
lyzed in similar way (see Note 6).

4  Notes

1. Molecular cloning and bacteria-related work are done on the

bench. Standard care should be taken including cleaning the
bench and pipettes with 70% industrial methanol solution.
Molecular reagents are stored at −20 °C freezer, kept on ice
when carrying out the procedure and return immediately to
the freezer after use.
2. It is essential that the number of MBS of the transformant
clones is verified by sequencing before use to generate stable
transfected CHO cell lines.
3. The number of MBS inserts for specific miRNA and scramble
sequence should be the same for accurate assessment. We
found that 300–400 ng of oligos duplex is the optimal concen-
tration in a ligation reaction that consistently resulted in large
inserts (more than 10 MBS). This concentration gives a ratio
of ~1:1000 (vector: oligos).
 TET-ON-SanDI-miRNA Sponge 99

4. The optimal concentration of antibiotics used for the selection

of stable CHO cell lines should be determined in advance
using the antibiotic killing curve approach. The concentrations
of antibiotics for selection of most commonly used CHO cells
are 300 μg/mL (hygromycin), 1 mg/mL (G418), 10 μg/mL
(puromycin), and 5 μg/mL (blasticidin).
5. Induction should be tested first with the backbone vector
before use for cloning. We found that DOX is quite toxic to
CHO cells; however at the low concentration at 10 ng/mL
(which is needed to be determined based on the CHO strain
used) the negative effect of DOX on CHO cell viability is neg-
ligible, and at the same time sufficient to use for induction.
The addition of DOX to growth media in this study did not
initiate transcription of the miRNA sponge; however, it
enhanced it in a conditional fashion.
6. The number of independent stable clones expressing the
miRNA sponge or NC for functional characterization should
not be so small, as variations among these are well documented
[11], due to the differences in the numbers and the insertion
sites of the transgene.

Acknowledgments

This work was supported by the Scientific Foundation of Ireland

(SFI) grants numbers 13/IA/1963 and 13/IA/1841.

References
1. Lewis NE, Liu X, Li Y, Nagarajan H, Yerganian
G, O'Brien E, Bordbar A, Roth AM,
Rosenbloom J, Bian C, Xie M, Chen W, Li N,
Baycin-Hizal D, Latif H, Forster J, Betenbaugh
MJ, Famili I, Xu X, Wang J, Palsson BO (2013)
Genomic landscapes of Chinese hamster ovary
cell lines as revealed by the Cricetulus griseus
draft genome. Nat Biotechnol 31:759–765.
doi:10.1038/nbt.2624
2. Jadhav V, Hackl M, Druz A, Shridhar S, Chung
CY, Heffner KM, Kreil DP, Betenbaugh M,
Shiloach J, Barron N, Grillari J, Borth N
(2013) CHO microRNA engineering is grow-
ing up: recent successes and future challenges.
Biotechnol Adv 31:1501–1513. doi:10.1016/j.
biotechadv.2013.07.007
3. Ebert MS, Neilson JR, Sharp PA (2007)
MicroRNA sponges: competitive inhibitors of
small RNAs in mammalian cells. Nat Methods
4:721–726
4. Kluiver J, Gibcus JH, Hettinga C, Adema A,
Richter MK, Halsema N, Slezak-Prochazka I,
Ding Y, Kroesen BJ, van den Berg A (2012)
Rapid generation of microRNA sponges for
microRNA inhibition. PLoS One 7:e29275.
doi:10.1371/journal.pone.0029275
5. Sanchez N, Kelly P, Gallagher C, Lao NT,
Clarke C, Clynes M, Barron N (2014) CHO
cell culture longevity and recombinant protein
yield are enhanced by depletion of miR-7 activ-
ity via sponge decoy vectors. Biotechnol
J 9:396–404. doi:10.1002/biot.201300325
6. Kelly PS, Breen L, Gallagher C, Kelly S, Henry
M, Lao NT, Meleady P, O'Gorman D, Clynes
M, Barron N (2015) Re-programming CHO
cell metabolism using miR-23 tips the balance
towards a highly productive phenotype.
Biotechnol J 10:1029–1040. doi:10.1002/
biot.201500101
7. Brinster RL, Chen HY, Warren R, Sarthy A,
Palmiter RD (1982) Regulation of metallo-
thionein--thymidine kinase fusion plasmids
injected into mouse eggs. Nature 296:39–42
8. Mayo KE, Warren R, Palmiter RD (1982) The
mouse metallothionein-I gene is transcription-
ally regulated by cadmium following trans­
fection into human or mouse cells. Cell 29:
99–108
100 Alan Costello et al.
9. Lee F, Mulligan R, Berg P, Ringold G (1981)
Glucocorticoids regulate expression of dihy-
drofolate reductase cDNA in mouse mammary
tumour virus chimaeric plasmids. Nature 294:
228–232
10. Lee SW, Tsou AP, Chan H, Thomas J, Petrie K,
Eugui EM, Allison AC (1988) Glucocorticoids
selectively inhibit the transcription of the inter-
leukin 1 beta gene and decrease the stability of
interleukin 1 beta mRNA. Proc Natl Acad Sci
U S A 85:1204–1208
11. Gossen M, Bujard H (1992) Tight control
of gene expression in mammalian cells by
tetracycline-­responsive promoters. Proc Natl
Acad Sci U S A 89(12):5547–5551
12. Tovar K, Hillen W (1989) Tet repressor bind-
ing induced curvature of tet operator DNA.
Nucleic Acids Res 17:6515–6522
13. Gossen M, Freundlieb S, Bender G, Müller G,
Hillen W, Bujard H (1995) Transcriptional
activation by tetracyclines in mammalian cells.
Science 268(5218):1766–1769
14. Hecht B, Muller G, Hillen W (1993) Non­
inducible Tet repressor mutations map from
the operator binding motif to the C terminus.
J Bacteriol 175:1206–1210
15. Hackl M, Jakobi T, Blom J, Doppmeier D,
Brinkrolf K, Szczepanowski R, Bernhart SH,
Höner Zu Siederdissen C, Bort JA, Wieser M,
Kunert R, Jeffs S, Hofacker IL, Goesmann A,
Pühler A, Borth N, Grillari J (2011) Next-­
generation sequencing of the Chinese hamster
ovary microRNA transcriptome: identification,
annotation and profiling of microRNAs as tar-
gets for cellular engineering. J Biotechnol
153(1–2):62–75
16. Zabala M, Wang L, Hernandez-Alcoceba R,
Hillen W, Qian C, Prieto J, Kramer MG (2004)
Optimization of the Tet-on system to regulate
interleukin 12 expression in the liver for the
treatment of hepatic tumors. Cancer Res
64:2799–2804
17. McGee Sanftner LH, Rendahl KG, Quiroz D,
Coyne M, Ladner M, Manning WC, Flannery
JG (2001) Recombinant AAV-mediated deliv-
ery of a tet-inducible reporter gene to the rat
retina. Mol Ther 3:688–696. doi:10.1006/
mthe.2001.0308
18. Sinacore MS, Drapeau D, Adamson SR (2000)
Adaptation of mammalian cells to growth
in serum-free media. Mol Biotechnol 15:
249–257
 Chapter 7

Application of CRISPR/Cas9 Genome Editing to Improve

Recombinant Protein Production in CHO Cells
Lise Marie Grav, Karen Julie la Cour Karottki, Jae Seong Lee,
and Helene Faustrup Kildegaard

Abstract
Genome editing has become an increasingly important aspect of Chinese Hamster Ovary (CHO) cell line
engineering for improving production of recombinant protein therapeutics. Currently, the focus is directed
toward expanding the product diversity, controlling and improving product quality and yields. In this
chapter, we present our protocol on how to use the genome editing tool Clustered Regularly Interspaced
Short Palindromic Repeat (CRISPR)/CRISPR-associated protein 9 (Cas9) to knockout engineering tar-
get genes in CHO cells. As an example, we refer to the glutamine synthetase (GS)-encoding gene as the
knockout target gene, a knockout that increases the selection efficiency of the GS-mediated gene amplifica-
tion system.

Key words Chinese Hamster Ovary Cells, CRISPR/Cas9, Genome editing, Glutamine synthetase,
Knockout, Recombinant protein production

1  Introduction

Chinese Hamster Ovary (CHO) cells are extensively used as a host

cell system for the production of recombinant protein therapeu-
tics. The progress and success of CHO cell culture technology has
long been depending on large-scale screening of highly productive
cell lines and process optimization. Despite the established success
of CHO cells, there are increasing demands to expand product
diversity, control and improve product quality, and improve cellu-
lar production capacities. Advances in this area have been made by
genetic engineering approaches including the repression or knock-
out of disadvantageous genes.
Knockout of genes has improved production of recombinant
proteins in CHO cells and the performance of CHO production
cells on several levels. Prolonging cell cultures by targeting pro-­
apoptotic genes, increasing product quality by targeting genes
involved in glycosylation of proteins, and enabling amplification of

Paula Meleady (ed.), Heterologous Protein Production in CHO Cells: Methods and Protocols, Methods in Molecular Biology,
vol. 1603, DOI 10.1007/978-1-4939-6972-2_7, © Springer Science+Business Media LLC 2017

101
102 Lise Marie Grav et al.

genes by targeting the dihydrofolate reductase (DHFR) gene are

among the achievements using gene knockouts. An overview of
genes targeted using gene knockout for improvement of CHO
production cells was recently presented by Fischer et al. [1].
The bacterial clustered regularly interspaced short palindromic
repeat (CRISPR)/CRISPR-associated protein 9 (Cas9) gene edit-
ing tool has drastically improved the knockout of genes—­making
it easier, cheaper, and more efficient. CRISPR/Cas9 is a relatively
simple genome editing system comprised only of the nuclease Cas9
and a single guide RNA (sgRNA), which has shown to be highly
applicable for genome editing in CHO cells [2]. In this system
target recognition is enabled by a 20-nt target complementary
sequence in the sgRNA, and a protospacer adjacent motif (PAM)
sequence directly downstream of the target sequence [3]. The
PAM sequence may differ between Cas9 orthologs [2]. The sgRNA
guides the Cas9 nuclease to the target DNA, where it introduces a
double strand break (DSB) [4]. The DSB can be repaired by the
cells’ own DNA repair system. In CHO cells the most frequently
used DNA repair pathway is the error-prone non-­homologous end
joining (NHEJ). Through the NHEJ pathway both dissected
DNA ends are ligated directly without DNA-end resection, which
may cause insertion or deletion of one or more base pairs—also
known as indel formation. Indel formation can cause a frameshift
in the coding region of genes, disrupting their translation and leav-
ing the genes dysfunctional. The Cas9 protein can be programmed
to target any DNA sequence of interest that is followed by a PAM
sequence, simply by changing the 20-nt target complementary
sequence in the sgRNA.
In this chapter, we describe a general and robust platform
for generating a single-gene knockout in CHO cells using the
CRISPR/Cas9 genome editing tool. As an example, we refer to
the gene glul, encoding glutamine synthetase (GS), as the target
gene. However, the methods are general and can be used to knock-
out other genes or even to multiplex knockouts [5]. GS is the
enzyme that converts glutamate and ammonia to glutamine, and
without glutamine in the growth medium GS is essential for cell
survival [6]. The widely used GS expression system ™ (Lonza)
exploits the glutamine metabolism in mammalian cells, by using a
transfected GS-encoding gene as a selectable marker as it permits
growth in medium lacking glutamine. The system works well in
cell lines that do not express sufficient GS to survive. In cell lines
such as CHO cells, which express sufficient endogenous GS, the
addition of the GS inhibitor methionine sulfoximine (MSX) is
required to inhibit excess GS activity. Increasing the MSX concen-
tration can result in gene amplification and increased productivity
[7]. It is desirable to eliminate the endogenous GS-encoding gene
from CHO cell lines to improve the selection stringency and effi-
ciency of the GS system in CHO cells [8].
 Knockout Generation in CHO Cells using CRISPR/Cas9 103

2  Materials

2.1  sgRNA 1. Target sequence analysis software, e.g., CRISPy tool available
Expression Plasmid for free online.
Construction 2. Glycerol stock of E. coli transformed with sgRNA expression
plasmid (from Ronda et al. [9]).
3. 2× YT medium.
4. Kanamycin.
5. 500 mL baffled Erlenmeyer shake flask.
6. Sterile pipette tips.
7. Incubator with shaker, 37 °C, 250 rpm.
8. Plasmid midi- or maxiprep kit (Machery-Nagel).
9. Sterile Milli-Q water.
10. NanoDrop 2000 (Thermo Scientific).
11. PCR primers for amp­lification of sgRNA backbone (for design
instructions see Subheading 3.2.2).
12. Primers containing the sgRNA sequence (for design instruc-
tions, see Subheading 3.2.3).
13. Phusion U polymerase (Thermo Scientific).
14. 5× HF Buffer (Thermo Scientific).
15. dNTPs.
16. PCR tubes.
17. Thermocycler.
18. Fast Digest DpnI enzyme (Thermo Scientific).
19. 10× Green Buffer (Thermo Scientific).
20. 1 kb DNA ladder.
21. 1% agarose gel: 1 g agarose powder (Bio-Rad) dissolved in
100 mL 1× TAE buffer (Sigma).
22. Gel chamber and power source.
23. PCR and gel purification kit (Machery-Nagel).
24. 10× NEBuffer 4 (New England Biolabs).
25. Heat block.
26. USER enzyme (New England Biolabs).
27. 10× BSA (New England Biolabs).
28. 1.5 mL eppendorf tubes.
29. Mach1 competent E. coli cells (Thermo Scientific).
30. Heat block, 37 °C, 300 rpm.
31. Table top centrifuge.
32. Sterile spatula.
104 Lise Marie Grav et al.

33. LB-kanamycin plates: 15 g/L Agar, 10 g/L Tryptone, 10 g/L

NaCl, 5 g/L Yeast Extract, 50 μg/mL kanamycin.
34. 10 mL bacterial culture tubes.
35. Plasmid miniprep kit (Machery-Nagel).
36. Access to Sanger sequencing facility.
37. Sequencing primers.
38. Sequence analysis software (e.g., CLC Main Workbench).

2.2  Prepare GFP 2A 1. Glycerol stock with E. coli transformed with GFP 2A peptide-­
Peptide-Linked Cas9 linked Cas9 expression plasmid (from Grav et al. [5]).
Expression Plasmid 2. Ampicillin.
3. LB-ampicillin agar plates: 15 g/L Agar, 10 g/L Tryptone,
10 g/L NaCl, 5 g/L Yeast Extract, 60 μg/mL ampicillin.

2.3  Transfection 1. CHO-S cells (Life Technologies).

of CHO-S Cells 2. NucleoCounter NC-200 Cell counter (ChemoMetec).
3. Growth medium: CD CHO medium (Life technologies) sup-
plemented with 8 mM l-glutamine (Lonza).
4. 15 or 50 mL centrifuge tubes.
5. 6-well plate, flat bottom (Corning #351146).
6. Humidified incubator, 37 °C, 5% CO2, 120 rpm.
7. OptiPro™ SFM reduced serum medium (Life Technologies).
8. FreeStyle™ MAX reagent (Life Technologies).
9. sgRNA expression plasmid generated in Subheading 3.2.
10. GFP 2A peptide-linked Cas9 expression plasmid prepared in
Subheading 3.3.

2.4  Analysis of Indel 1. Quick extract (Epicentre).

Generation: T7 2. 2× Phusion Master Mix (Thermo Scientific).
Endonuclease Assay
3. Primers (for design see Subheading 3.5).
4. 10× NEBuffer 2 (New England Biolabs).
5. T7 endonuclease (New England Biolabs).
6. 4% E-Gel (Invitrogen #G501804) or equivalent.
7. Mother E-base (Invitrogen).

2.5  Generation 1. Fluorescence-activated cell sorter.

of Clonal Cell Lines: 2. 384-well plates, flat bottom (Corning #3542).
Fluorescence-­
3. FACS sorting medium: CD CHO medium (Life Techno­
Activated Cell Sorting
logies) supplemented with 8 mM l-glutamine (Lonza), 1%
(FACS) Antibiotic-­Antimycotic 100× (Gibco), and 1.5% HEPES (Life
Technologies).
 Knockout Generation in CHO Cells using CRISPR/Cas9 105

4. FACS tubes.
5. 30 μm cell strainer.
6. Celigo cytometer or microscope.
7. Humidified incubator, 37 °C, 5% CO2, no shake.
8. 96-well plates, flat bottom (Corning #351172).
9. Clone expansion medium: CD CHO medium (Life Techno­
logies) supplemented with 8 mM l-glutamine (Lonza), 1%
Antibiotic-­
Antimycotic 100× (Gibco), and 1 μL/mL Anti-­
clumping agent (Life Technologies #0010057AE).
10. 96-well plates, V-Shaped (Greiner bio-one #651161).
11. Breathable plastic bag.

2.6  Analysis of Gene 1. Primers from Subheading 3.5.

Modifications: Sanger 2. 2× Phusion Master Mix (Thermo Scientific).
Sequencing
3. DNA quick extract from Subheading 3.6.

2.7  Expansion 1. 12-well plates, flat bottom (Corning #351143).

of Clones 2. 125 mL shake flask (Corning #431143).
3. DMSO (Sigma-Aldrich #472301).
4. Cryotubes.

3  Methods

The following section is a general protocol that we use for single

knockouts in our lab. To make the protocol easier to follow, we
refer to the GS-encoding gene as an example of a specific target.
The protocol can be used to target other genes, and can easily be
adapted to multiple knockouts [5]. An overview of the protocol is
shown in Fig. 1.

3.1  Identification 1. Go to http://staff.biosustain.dtu.dk/laeb/crispy/.

of Target Site 2. Search for your target gene by name, id or symbol, e.g., the
and sgRNA Primer GS-­encoding gene glul. Be aware of that there can be multiple
Design genes annotated to encode one protein (see Note 1).
3. A selection of different target sequences will be displayed. You
can sort them according to the number of exact matches (pref-
erably one) and see where the different target sequences are
located in relation to exons and introns (see Note 2). The target
sequence is followed by a PAM sequence (5′-NGG-3′) in the
target gene, as shown in Fig. 2a.
106 Lise Marie Grav et al.

Identify target sequence Construct sgRNA Transfect GFP_2A_Cas9

in gene of interest expression plasmid and sgRNA

target sgRNA
U6 sequence scaffold GFP
5’-GNNNNNNNNNNNNNNNNNNNNGG-3’

2A
promoter termination Cas9
signal

sgRNA plasmid

DSB by CRISPR Cas9

and GFP expression

Make a cell bank Single cell sort for GFP

Expand clones
of clones positive cells using FACS

Sample & extract + -

Verify genomic DNA
knockouts

Analyse gene WASTE SINGLE CELLSORT

modifications

Fig. 1 Schematic outline of the experimental setup for the method described in this chapter

4. Design sgRNA primers using your selected target sequence

from Subheading 3.1. The target sequence minus the PAM
sequence is the only varying region when designing primers
and constructing the sgRNA plasmids for different target
sequences, as described in Fig. 2b.

3.2  sgRNA Plasmid 1. Request the sgRNA plasmid from Ronda et al. [9] and gener-
Construction ate a bacterial glycerol stock (or prepare it from scratch follow-
ing the method described in the publication).
3.2.1  Prepare sgRNA
Backbone Plasmid 2. Use the tip of a sterile pipette tip and scrape the bacterial stock,
add the pipette tip to 100–200 mL of 2× YT medium supple-
mented with 50 μg/mL kanamycin, and incubate overnight at
37 °C with shaking at 250 rpm.
3. Isolate plasmid DNA using a plasmid midi- or maxiprep kit,
resuspend at 1 μg/μL in Milli-Q water. Measure the concen-
tration using Nanodrop 2000.

3.2.2  Amplify sgRNA 1. Use the sgRNA plasmid map, and design and order uracil-­
Backbone containing primers to amplify the sgRNA backbone. The prim-
ers should amplify the sgRNA backbone so that it acquires
overhangs after Uracil-Excision Specific Reagent (USER)
treatment that is compatible for USER fusion with the over-
hangs of the annealed primers from Subheading 3.2.3, as
described in Fig. 2b. Alternatively, apply primers previously
published [9].
 A Target sequence PAM
N (19) + G NGG
5’ -NNNNNNNNN GNNNNNNNNNNNNNNNNNNNAGGNNNNNNNNNNN- 3’
3’ -NNNNNNNNN CNNNNNNNNNNNNNNNNNNN TCC NNNNNNNNNNN- 5’

Target complementary
B sequence G+N (19)
Forward sgRNA primer NNNNNNNNGNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN
Reverse sgRNA primer NNNNNNNNNNNNNNNNNNNNNNNNCNNNNNNNNNNNNNNN

Anneal primers

Annealed NNNNNNNNGNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN
sgRNA primers NNNNNNNNNNNNNNNCNNNNNNNNNNNNNNNNNNNNNNNN

+
sgRNA
scaffold
U6 N
NNNN
NN
NNNN NNU
NN
NNN
NN NNN N
NNNN termination
promoter
N
UNN
signal

PCR amplified
backbone

USER treatment

target sgRNA
sequence scaffold
U6 termination
promoter signal

sgRNA plasmid

Fig. 2 An outline of the target gene and target sequence, and a schematic overview of how to construct the
sgRNA expression plasmid. (a) An outline of the target gene showing the target sequence (5′-G+N(19)-3′) and
the PAM sequence (5′-NGG-3′) in relation to each other. (b) Schematic overview of the sgRNA plasmid con-
struction described in Subheading 3.2. A simple way to construct your sgRNA plasmid is by ordering your
target complementary sequence as primers. You can keep your sgRNA “constant” and just exchange the target
complementary sequence. The primers for the target sequence are designed to anneal and give rise to over-
hangs that match overhangs generated after USER treatment of the amplified sgRNA backbone. In this case,
the following uracil containing primers would be used for amplifying the backbone: Fwd: AGCTAGAAA
UAGCAAGTTAAAATAAGGC and Rev.: ACAAGATAUATAAAGCCAAGAAATCGA. After assembly of the annealed
primers and the amplified sgRNA backbone upon USER enzyme treatment, you will attain the complete sgRNA
expression construct
108 Lise Marie Grav et al.

Table 1
PCR program to amplify the sgRNA backbone

Temperature (°C) Time (min) Number of cycles

98 00:30  1
98 00:10 35
57 00:30
72 01:15
72 10:00  1
4 ∞  1

2. Mix the following components in a PCR tube:

–– 10 μL 5× HF buffer.
–– 1 μL dNTPs (10 mM).
–– 2.5 μL Primer forward (10 μM).
–– 2.5 μL Primer reverse (10 μM).
–– 0.5 μL Phusion U polymerase.
–– 1  μL sgRNA plasmid template (1.7 ng/uL) (prepared in
Subheading 3.2.1).
–– 32 μL Milli-Q water.
3. Place the PCR tube in thermocycler, and run the following
program (as shown in Table 1).
4. Treat the sgRNA backbone amplicon with DpnI enzyme to
remove methylated DNA by mixing the following components:
–– 44 μL of the sgRNA backbone PCR reaction mixture.
–– 5 μL 10× Green buffer.
–– 1 μL Fast Digest DpnI enzyme.
5. Incubate the mixture at 37 °C for 1 h, this will degrade meth-
ylated DNA.
6. Run the PCR product on a 1% agarose gel alongside a 1 kb
DNA ladder.
7. Cut out the band at approximately 4.2 kb and purify the PCR
product using a PCR and gel purification kit.

3.2.3  Annealing Primers 1. Mix the following components in an Eppendorf tube:

for sgRNA Construct –– 10 μL 10× NEBuffer 4.
–– 10 μL sgRNA Forward primer (100 μM).
–– 10 μL sgRNA Reverse primer (100 μM).
–– 70 μL Milli-Q water.
 Knockout Generation in CHO Cells using CRISPR/Cas9 109

Table 2
Components of reaction for assembly of backbone and sgRNA insert

Component Negative control, μL sgRNA reaction, μL

Backbone 1 1
Annealed sgRNA primers – 7
10× BSA 0.5 0.5
NEBuffer 4 0.5 0.5
USER enzyme 1 1
Milli-Q water 7 –

2. Incubate the mixture at 95 °C for 5 min on a heat block, turn

off the heat block, and leave the mixture on the heat block
overnight for gradual cooling. Store at −20 °C.

3.2.4  Assembly 1. Mix components in a PCR tube according to Table 2, including

of Backbone and sgRNA a negative control (backbone only).
Insert 2. Incubate the mixed reactions at 37 °C for 40 min, and 25 °C
for 30 min. Store at −20 °C.

3.2.5  Transformation
of sgRNA Plasmid in E. coli
1. Add 1.5 μL of USER reaction to 15 μL competent E. coli cells
in an Eppendorf tube and incubate on ice for 30 min.
2. Heat shock at 42 °C for 30 s.
3. Return to ice and keep it there for 1 min.
4. Add 1 mL 2× YT medium to the Eppendorf tube and incubate
the mixture at 37 °C for 1 h at 300 rpm shake.
5. Pellet the cells at 2000 × g for 3 min.
6. Remove the supernatant, resuspend the pellet in 100 μL 2× YT
medium, and plate it using a sterile spatula on a pre-warmed
(37 °C) kanamycin agar plate.
7. Incubate the plates upside down at 37 °C overnight.

3.2.6  Analyze 1. Pick a colony using a pipette tip and transfer it to 4 mL 2× YT
and Prepare sgRNA medium supplemented with 50 μg/mL kanamycin in a 10 mL
Plasmid bacterial culture tube.
2. Shake cells at 250 rpm at 37 °C for 5 h.
3. Isolate plasmid DNA using a plasmid miniprep kit and use this
for sanger sequencing. Use a primer for sequencing that
anneals before the U6 promoter and covers the target sequence,
to make sure your sgRNA expression cassette does not contain
any mutations using a sequence analysis software, e.g., CLC
Main Workbench.
110 Lise Marie Grav et al.
3.3  Prepare the GFP
2A Peptide-­Linked
Cas9 Expression
Plasmid
3.4  Transfection
of CHO-S Cells
3.4.1  Day 0: Washing
and Seeding Cells
for Transfection
3.4.2  Day 1: Transfection
4. Grow a midi- or maxiprep culture of the correct transformant
by inoculating 100–200 mL of 2× YT medium supplemented
with 50 μg/mL kanamycin with 1 mL culture from step 2.
Incubate overnight at 37 °C with shaking at 250 rpm.
5. Isolate plasmid DNA using a midi- or maxiprep kit. Resuspend
plasmid DNA at approximately 1 μg/μL in sterile Milli-Q
water. Measure the concentration using Nanodrop 2000. Use
this product for transfection.
1. Request the GFP 2A peptide-linked Cas9 expression plasmid
(GFP_2A_Cas9) plasmid from Grav et al. [5], and generate a
bacterial glycerol stock (or prepare it from scratch following
the method described in the publication).
2. Use the tip of a sterile pipette tip and scrape the bacterial stock,
add the pipette tip to 100–200 mL of 2× YT medium supple-
mented with 60 μg/mL ampicillin, and incubate overnight at
37 °C with shaking at 250 rpm.
3. Isolate plasmid DNA using a plasmid midi- or maxiprep kit,
resuspend 1 μg/μL in sterile Milli-Q water. Measure the
concentration using Nanodrop 2000. Use this product for
­
transfection.
1. Use healthy (above 95% viability) CHO-S cells at a low
passage.
2. Count cells using a NucleoCounter.
3. Harvest 1.5–2 × 106 cells (for a single transfection in 1 × 6
well), spin down cells at 200 × g for 5 min, and remove the
supernatant.
4. Wash cells with preheated CD CHO medium supplemented
with 8 mM glutamine (no anti-clumping agent), spin down at
200 × g for 5 min, and remove the supernatant.
5. Inoculate cells at 5–6 × 105 cells/mL in 1 × 6 well with pre-
heated growth medium.
6. Incubate cells at 37 °C, 5% CO2, and shake at 120 rpm for
1 day.
1. Count cells and inoculate cells at 1 × 106 cells/mL in 3 mL
preheated growth medium in a 6-well plate.
2. Use a total amount of 3.75 μg of plasmid (1:1 (w/w) of sgRNA
plasmid and Cas9 expression plasmid). If you want to use mul-
tiple sgRNAs, see Note 3.
3. Gently mix plasmids with 60 μL OptiPRO™ SFM reduced
serum medium.
4. Dilute 3.75 μL FreeStyle™ MAX reagent in 60 μL OptiPRO™
SFM reduced serum medium. Mix gently and add to plasmid
premix (step 3).
 Knockout Generation in CHO Cells using CRISPR/Cas9 111

5. Incubate for 5 min.

6. Gently add transfection mix to the cells from step 1.
7. Incubate cells at 37 °C, 5% CO2 and shake at 120 rpm for
2 days.

3.5  Optional: T7 1. Harvest 50 μL cells 2 days after transfection, spin down at
Endonuclease Assay 1000 × g, remove the supernatant, and add 20 μL quick extract
to Check if Your and incubate at 65 °C for 15 min followed by 95 °C for 5 min.
sgRNA Works Store at −20 °C.
2. Design primers using, e.g., NCBI Primer-BLAST tool (http://
www.ncbi.nlm.nih.gov/tools/primer-blast/) so that a prod-
uct between 600 and 1000 bp will be amplified. This product
should span the selected target sequence in your target gene.
The T7 endonuclease will cleave the product where an indel is
present when hybridized to wild-type sequence. Design your
primers so that the PCR product after cleavage will give bands
of different sizes that are separable on an agarose gel.
3. Mix the following components in a PCR tube (prepare one for
quick extract of the transfected pool of cells and one for quick
extract from CHO-S cells):
–– 10 μL 2× Phusion Master Mix.
–– 1 μL Primer Forward (10 μM).
–– 1 μL Primer Reverse (10 μM).
–– 1 μL DNA template (quick extract from step 1).
–– 7 μL Milli-Q water.
4. Place the PCR tube in a thermocycler, and run the following
program as outlined in Table 3:
5. Run 5 μL of the PCR product next to a 1 kb DNA ladder on a
1% agarose gel. There should be only one clear band (if not
redo and/or troubleshoot your PCR).
6. Transfer 10 μL to a new PCR tube.
7. Place the PCR tube in a thermocycler and run the following
T7 endonuclease annealing program as outlined in Table 4.
8. Divide the PCR product into two PCR tubes (5 μL in each
tube).
9. Mix the components for the following two reactions (T7+ and
T7−) as outlined in Table 5.
10. Incubate the PCR tubes for 30 min at 37 °C.
11. Load the products on a 4% E-Gel or equivalent, and run for
approximately 30 min. Your results should be similar to what is
shown in Fig. 3a.
112 Lise Marie Grav et al.

Table 3
PCR program for T7 endonuclease assay

Temperature (°C) Time (min) Number of cycles

98 00:30  1
98 00:10 10
68 00:30
−1 per cycle
72 00:30
98 00:10 20
58 00:30
72 00:30
72 05:00  1
4 ∞  1

Table 4
T7 endonuclease annealing program

Temperature (°C) Time (min) Number of cycles

95 05:00  1
95 00:01  5
−2 s −1

85 00:01 300
−0.2 s−1
4 ∞

Table 5
Components of T7+ and T7− reactions

Component T7+, μL T7– (negative control), μL

10× NEBuffer 2 1 1
T7 endonuclease 0.5 –
PCR product 5 5
Milli-Q water 3.5 4
 Knockout Generation in CHO Cells using CRISPR/Cas9 113

Fig. 3 Analysis of genome modifications. (a) T7 endonuclease assay of a pool of cells 2 days after transfection,
showing that a sgRNA complementary sequence to a selected target site is capable of generating indels. The
samples are analyzed on a 4% E-Gel, where the two first lanes (1 and 2) are not treated with T7 endonuclease,
and the two last lanes (3 and 4) are treated with T7 endonuclease. Samples in lane 2 and 4 are transfected
with GFP_2A_Cas9 and a sgRNA, while the other two samples are un-transfected. After T7 endonuclease
treatment, the un-transfected sample shows two bands, while the transfected sample shows two additional
bands at expected sizes, designated by arrows. A result like this shows that a selected sgRNA is capable of
generating indels at the target site. The percentage of indels generated in this case is estimated to 9.3, by
using ImageJ software. (b) An example of sanger sequencing analysis of a target sequence region. The align-
ment shows that out of four sequences, in this case, there are one sequence with an indel of −10, one with an
indel of +1, and two that still have the wild-type sequence (no indel)

3.6  Generation 1. Two to three days after transfection, prepare wanted number
of Clonal Cell Lines of 384-well flat-bottom plates (see Note 4) with 30 μL FACS
Using FACS sorting medium.
2. Strain cells through a 30 μm cell strainer into a FACS tube.
3. Using a FACS, sort for GFP-positive single cells into one or
more pre-warmed (37 °C) 384-well plates. If a FACS is not
available, an alternative method can be used (see Note 5).
4. Spin plates at 200 × g for 5 min to make sure cells reach the
medium.
5. Place cells in a breathable plastic bag (to limit evaporation),
and incubate cells at 37 °C, 5% CO2, no shake for 10 days.
6. Check for surviving cells using a microscope or Celigo cytom-
eter. Cell count should preferably be around >1000 in a well or
confluency >50%.
114 Lise Marie Grav et al.

7. Carefully pipette up and down three times and transfer cells to

a 96-well flat-bottom plate with 180 μL of clone expansion
medium.
8. After 4 days, check plates in microscope. When the clones have
a confluency >50%, carefully pipette up and down three times
and transfer 50 μL cell suspension to a 96-well V-shaped plate.
9. Spin down the V-shaped 96-well plate at 1000 × g for 5 min,
remove the supernatant, add 20 μL quick extract, resuspend
the pellets, and move them to PCR tubes or plates. Incubate at
65 °C for 15 min and 95 °C for 5 min. Store at −20 °C.

3.7  Analysis of Gene 1. Mix the following components in a PCR tube (per clone you
Modifications: Sanger have generated):
Sequencing –– 10 μL 2× Phusion Master Mix.
–– 1 μL Primer Forward (10 μM).
–– 1 μL Primer Reverse (10 μM).
–– 1 μL DNA template (quick extract from Subheading 3.6).
–– 7 μL Milli-Q water.
2. Place the PCR tube in thermocycler, and run the following
program as shown in Table 6.
3. Run the PCR product on a 1% agarose gel, cut out the band
with the expected amplicon size, and purify it using a gel
and PCR purification kit. Measure the concentration using
Nanodrop 2000.

Table 6
PCR program for analysis of gene modifications

Temperature (°C) Time (min) Number of cycles

98 00:30  1
98 00:10 10
68 00:30
−1 per cycle
72 00:30
98 00:10 20
58 00:30
72 00:30
72 05:00  1
4 ∞  1
 Knockout Generation in CHO Cells using CRISPR/Cas9 115

4. Sequence the product using the forward primer designed for

the T7 endonuclease assay (see Note 6). Mix primer and puri-
fied PCR product according to the instruction provided by the
sequencing service you use. We discuss application of alterna-
tive methods to analyze gene modifications in Subheading 4
(see Note 7).
5. Analyze your sequencing results using a sequence analysis soft-
ware, e.g., CLC Main Workbench and align the results to the
wild-type target sequence (e.g., the GS sequence). The results
should be similar to what is shown in Fig. 3b.

3.8  Expansion 1. Select clones with indels that lead to a frameshift, which indi-
of Clone Candidates cates that you have rendered the gene dysfunctional. Even if
the analysis shows there is a frameshift, it is important to verify
that it is a real knockout, e.g., by western blotting (see Note 8)
and/or a functional assay (see Note 9).
2. Move the selected clones from the 96-well plate when >90%
confluent to a 12-well flat-bottom plate, maintain in the
12-well plate until confluent, and then move to a 6-well flat-
bottom plate, when confluent seed in a 125 mL shake flask at
3 × 105 cells/mL.
3. Take out the samples you need and bank the clones, using 107
cells in 1 mL conditioned medium with 5–10% DMSO. Freeze
in a Styrofoam box at −80 °C the first 24 h before moving to
permanent storage at −180 °C.

4  Notes

1. When selecting a sgRNA sequence to target the gene of inter-

est, it is important to be aware that there can be several genes
annotated for the same protein that may be isoforms or pseu-
dogenes. When this is the case you could either select a sgRNA
specific for each of the genes, select a sgRNA that matches all
of the genes, or select sgRNAs based on expression levels of
the gene variants.
2. Examples of sgRNAs in CRISPy with exact match in each of the two
glul gene variants in the CHO-K1 genome (GCF_000223135.1)
are “GGCCCAGGGAAGCCATCGGAAGG” (GeneID:100,
689,337) and “GGCCTCCTCGATGTGCCTGGTGG”
(GeneID:100,764,163), or that matches both genes (in the
3′-end) are “GAGAAGGCATGTGCGGACGATGG.” When
selecting your target sequence, you should consider the vari-
ous exons and splice variants for the gene of interest, and which
target position will most likely make a break that leaves the
gene dysfunctional. If opportunity allows it, it is a good idea
116 Lise Marie Grav et al.

to test a minimum of two different sgRNAs. Always r­ emember

that the annotation of some genomes can be incomplete at
the time of analysis, and it is always important to validate your
knockout, either by a Western Blot (see Note 8) and/or a
functional assay (see Note 9). The CRISPy tool takes the pres-
ence of a 5′-NGG-3′ PAM sequence directly downstream of
the target sequence for Streptococcus pyogenes Cas9 (spCas9)
into consideration. If you are using a different Cas9 ortholog,
there might be different requirements to your PAM sequence
that you need to consider.
3. When using multiple sgRNAs, the weight ratio between the
Cas9 expression plasmid and sgRNA plasmids is still (1:1). The
weight ratio between the different sgRNAs is also (1:1). For
instance, if you use three different sgRNAs the weight ratio
should be (1:1:1). Another recommendation when multiplex-
ing gene knockouts is to use deep sequencing when analyzing
the gene modifications, as explained in Note 6.
4. The number of 384-well plates necessary for sorting depends
on your FACS sorting efficiency. It is important to use strin-
gent settings so that you only sort one cell per well. Depending
on your sorting efficiency (how many wells get a single cell
sorted into it) and survival (how many single cells survive), you
can adjust the number of plates you sort. If your sorting effi-
ciency is around 30%, two 384-well plates should give a suffi-
cient number of knockout cells to choose among.
5. Limiting dilution is an alternative way to single cell sort your
cells, if you do not have access to a FACS [10]. In this case you
will need to screen more clones, as you cannot enrich for
GFP_2A_Cas9 expressing cells. When applying limiting dilu-
tion it is sufficient to use a plasmid expressing Cas9 that is not
linked to GFP with a 2A peptide.
6. If your coverage is not good when only using the forward
primer, you can also use the reverse primer for sequencing, or
design new sequencing primers. The Sanger sequencing
method is very simple, but not optimal. You might experience
some unclear sequencing result, due to, e.g., mixed indels on
different alleles. Select only the clones that show clear sequenc-
ing results indicating a frameshift; otherwise, it is necessary to
use one of the alternative methods described in Note 7. When
potential functional knockout clones are selected, we recom-
mend applying an additional primer set for verification.
7. TOPO™ cloning-based sequencing and deep sequencing are
alternative methods that can be used to analyze gene modifica-
tions in greater detail. A description of how to perform these
analyses can be found in Ronda et al. [9]. Both methods can
reveal if you have an indel on one or more alleles of your
targeted gene. However, TOPO cloning-based sequencing
­
 Knockout Generation in CHO Cells using CRISPR/Cas9 117

putative GS KO

CHO-S
Anti-Vinculin

Anti-GS

1 2

Fig. 4 Knockout validation. An example of a western blot of cell lysates from a

putative GS knockout (KO) clone in lane 1 and wild-type CHO-S in lane 2, which
were analyzed with anti-GS antibody (Abcam ab 49873) and anti-vinculin anti-
body (Sigma-Aldrich V9131). Anti-vinculin is used as a loading control. The puta-
tive GS knockout in lane 1 shows no band at the expected size of GS (45 kDa),
while the wild-type shows a band at the expected size of GS, indicating that GS
is not expressed in this putative GS knockout

requires a much higher extent of screening than deep sequencing.

It is therefore recommended to use deep sequencing analysis
when analyzing gene modifications from multiplexed
­knockouts, as it minimizes the work load and gives a higher
coverage. When potential functional knockout clones are
selected, we recommend applying an additional primer set for
verification.
8. Even though you see clear gene modifications in your analysis,
the protein could still be expressed in an intact or truncated ver-
sion. A simple way to ensure that you have disrupted the protein
expression is to perform a Western Blot of your putative clones,
using an antibody against target gene. In the case of a GS knock-
out, you can use an antibody against GS as shown in Fig. 4.
9. In addition to ensuring that you have lost the protein expres-
sion, it is important to ensure that you have acquired the
wanted phenotypic change. As in the case of a GS knockout, a
simple test is to grow the cells in the presence and absence of
glutamine. If the cells die in the absence of glutamine (and not
in the presence), you likely have a functional knockout.

Acknowledgments

We thank Karen Kathrine Brøndum and Johnny Arnsdorf for opti-

mizing and setting up FACS sorting, and Nachon Charanyanonda
Petersen for help with the transfection and FACS sorting. This
work was supported by the Novo Nordisk Foundation.
118 Lise Marie Grav et al.
References
1. Fischer S, Handrick R, Otte K (2015) The art
of CHO cell engineering: a comprehensive ret-
rospect and future perspectives. Biotechnol
Adv 33(8):1878–1896
2. Lee JS, Grav LM, Lewis NE, Faustrup Kildegaard
H (2015) CRISPR/Cas9-mediated genome
engineering of CHO cell factories: application
and perspectives. Biotechnol J 10:979–994
3. Jinek M, Chylinski K, Fonfara I, Hauer M et al
(2012) A programmable dual-RNA-guided
DNA endonuclease in adaptive bacterial immu-
nity. Science 337:816–821
4. Garneau JE, Dupuis MÈ, Villion M, Romero
DA et al (2010) The CRISPR/Cas bacterial
immune system cleaves bacteriophage and plas-
mid DNA. Nature 468:67–71
5. Grav LM, Lee JS, Gerling S, Kallehauge TB,
Hansen AH, Kol S, Lee GM, Pedersen LE,
Kildegaard HF (2015) One-step generation of
triple knockout CHO cell lines using CRISPR/
Cas9 and fluorescent enrichment. Biotech­
nol J 10:1446–1456
6. Wurm FM (2004) Production of recombinant
protein therapeutics in cultivated mammalian
cells. Nat Biotechnol 22:1393–1398
7. Jun SC, Kim MS, Hong HJ, Lee GM (2006)
Limitations to the development of humanized
antibody producing Chinese hamster ovary
cells using glutamine synthetase-mediated gene
amplification. Biotechnol Prog 22:770–780
8. Fan L, Kadura I, Krebs LE, Hatfield CC, Shaw
MM, Frye CC (2012) Improving the efficiency
of CHO cell line generation using glutamine syn-
thetase gene knockout cells. Biotechnol Bioeng
109:1007–1015
9. Ronda C, Pedersen LE, Hansen HG,
Kallehauge TB, Betenbaugh MJ, Nielsen AT,
Kildegaard HF (2014) Accelerating genome
editing in CHO cells using CRISPR Cas9
and CRISPy, a web-based target finding
tool. Biotechnol Bioeng 111:1604–1616
10. Freshney RI (2010) Cloning and selection in:
culture of animal cells, 6th edn. Wiley,
New York, NY
 Chapter 8

Improved CHO Cell Line Stability and Recombinant Protein

Expression During Long-Term Culture
Zeynep Betts and Alan J. Dickson

Abstract
Therapeutic proteins require proper folding and posttranslational modifications to be effective and
­biologically active. Chinese hamster ovary (CHO) cells are by far the most frequently used host for com-
mercial production of therapeutic proteins. However, an unpredictable decrease in protein productivity
during the time required for scale up impairs process yields, time, finance, and regulatory approval for the
desired product. Therefore, it is important to assess cell lines at stages throughout the period of long-term
culture in terms of productivity and various molecular parameters including plasmid and mRNA copy
numbers and location of the plasmid on the host cell chromosome. Here, we describe methods, which are
frequently used to analyze stability of the recombinant CHO cells over long-term culture. These proce-
dures include the following; western blotting, ELISA to evaluate protein production, real-time PCR to
analyze plasmid and mRNA copy numbers, and fluorescent in situ hybridization (FISH) to assess the loca-
tion of the inserted plasmid on host cell chromosomes.

Key words CHO cells, Recombinant protein production, Western blotting, ELISA, Real-time PCR,
FISH

1  Introduction

Chinese hamster ovary (CHO) cells are the most widely used host
cell platform for the production of recombinant proteins that
require complex post-translational modifications. One of the prob-
lems often encountered, however, is that these cell lines are highly
unpredictable and display variable levels of recombinant protein
expression. In addition, it is often observed that cell lines display a
decrease in recombinant protein production during long periods
of culture. It is important for commercial production to obtain cell
lines that maintain stable production over the long-term culture
(i.e., retention of 70% of the starting value by 60 generations, a
period required to scale up to manufacture) [1–3]. If a recombi-
nant cell line fails to retain stability during prolonged culture, it
can create problems for process yield, effective use of time and

Paula Meleady (ed.), Heterologous Protein Production in CHO Cells: Methods and Protocols, Methods in Molecular Biology,
vol. 1603, DOI 10.1007/978-1-4939-6972-2_8, © Springer Science+Business Media LLC 2017

119
120 Zeynep Betts and Alan J. Dickson

money, protein quality, and regulatory approval [2]. Different

­processes contribute to the recombinant protein production and
overall production stability of CHO cell lines including changes to
transgene chromosomal location [4–6], epigenetic regulation of
promoter environment [7, 8], recombinant gene copy number [9],
and recombinant mRNA expression [10, 11]. Therefore, to improve
CHO cell line stability and recombinant protein expression during
long-term culture, exact molecular mechanisms (genomic, epigen-
etic, post-transcriptional, post-translational, chromosomal, etc.)
leading to instability need to be defined by conducting analysis at
different stages of culture.
Here, we demonstrate different methods to analyze molecular
parameters that might contribute to overall productivity and insta-
bility of production of CHO cells over long-term culture. First, we
will describe the methods to measure the level of recombinant pro-
tein production. Western blot analysis can be used to detect and
quantify individual proteins in samples to assess changes in protein
expression level from a population of cells. With this technique
proteins are separated based on their molecular weight through gel
electrophoresis. These results are then transferred to a solid sup-
port (nitrocellulose membrane). The membrane is then incubated
using a proper primary (specific to the protein of interest) and sec-
ondary antibody to visualize the result [12].
Antibody-sandwich ELISA is another useful method for quan-
tifying proteins in solution [13]. To detect the protein of interest
the wells of microtiter plates are coated with specific (capture) anti-
body that binds an epitope on the target protein. This step is fol-
lowed by adding a sample solution containing the protein of
interest. The immobilized antibody will capture any target protein
that is present in the sample and nonspecifically bound material
will be washed out. Subsequently, incubation with a detection anti-
body followed by a secondary antibody, usually conjugated to an
enzyme that allows detection by chromogenic or chemilumines-
cent methods, is performed. Unbound conjugate is washed out
and plates are incubated with the substrate. The degree of sub-
strate hydrolysis is measured, which is proportional to the amount
of protein in the sample [14].
Flow cytometry is also a useful tool to determine the percent-
age of the number of positive cells in a population as well as for
providing information on recombinant protein production when a
fluorescence protein encoding gene is incorporated in the expres-
sion construct.
Second, we will explain how to extract DNA and RNA from
cell lines and analyze these samples by using real-time PCR respec-
tively to assess changes in gene copy number and mRNA copies
over long-term culture. An adapted version of the protocol detailed
by Blin and Stafford [15] is used to extract DNA from CHO cells.
 CHO Cell Line Stability Analysis During Long-Term Culture 121

In summary, cell lysates are placed in a solution of proteinase K,

RNAse A, and sarcosyl and incubated to degrade the cellular
­protein. The digest is then subsequently deproteinized by phenol/
chloroform/isoamyl alcohol extractions. DNA is then recovered
by ethanol precipitation. The single-step RNA isolation protocol
was performed by Chomczynski and Sacchi [16, 17]. The principle
is to separate RNA from DNA using an acidic solution containing
guanidinium thiocyanate, sodium acetate, phenol, and chloroform.
Total RNA remains in the aqueous phase while most proteins and
DNA remain in the interphase or lower organic phase. Total RNA
can be recovered by isopropanol precipitation. The DNA and RNA
samples can be analyzed by using the quantitative PCR method.
The inclusion of standard curves in real-time PCR reactions allows
the program to calculate amplification efficiency and the relative
quantity of the target sequence in each sample. The accuracy is
increased by standardizing the abundance of the target sequence to
the abundance of a housekeeping gene (i.e., beta actin). For the
absolute quantification of gene copy number a standard curve of
various amounts of the plasmid DNA mixed with parental genomic
DNA (20 ng/μL) is included in every run. The values obtained
from these standard samples with known plasmid copy numbers
per reaction are used to create a standard curve and allow us to
subsequently estimate the number of plasmid copies in each
sample.
After investigating cell lines over long-term culture in terms of
recombinant protein productivity and changes in plasmid and
RNA copy numbers, it is important to extend these findings to
define chromosomal localization of recombinant genes. This
approach should be developed to provide a molecular analysis of
the manner in which the location of amplified genes might relate
to the differential productivity and stability of recombinant CHO
cell lines. Therefore, we will explain methodology for fluorescent
in situ hybridization (FISH). In situ hybridization experiments
were first introduced by Gall and Pardue [18]. The procedure
involved hybridization of endogenous single-stranded RNA or
DNA in the cell and a complementary single-stranded RNA or
DNA probe. These probes were labeled radioactively and the
hybrid was detected using autoradiography [18]. Today, mainly
fluorescent probes are used to detect DNA sequences due to their
greater safety and ease of detection [19–23]. Prior to hybridiza-
tion, the DNA probe is labeled via nick translation or various other
methods (i.e., random primed labeling, pCR) [24, 25]. The DNA
probe can be labeled indirectly (with modified nucleotides that
contain a hapten) or directly (using nucleotides that contain a fluo-
rophore). In the next step, both the target and the probe sequences
need to be denatured using heat or chemicals to allow hydrogen
bonds to form between the probe and the target sequences during
122 Zeynep Betts and Alan J. Dickson

hybridization step. The probe and target sequences are mixed

together subsequently and the probe hybridizes its complementary
sequence on the chromosome. After hybridization, hybrid mole-
cules formed between the probes and the target sequences can be
visualized under a fluorescent microscope [26].

2  Materials

Prepare all solutions in miliQ water (ddH2O) unless otherwise

stated. Make all solutions used in the processing of RNA in 0.05%
(v/v) diethylpyrocarbonate (DEPC)-treated ddH2O. Sterilize
solutions by autoclaving or by filtration through a 0.2 μm filter
where autoclaving is not appropriate. Reagents should be prepared
and stored at room temperature, unless indicated otherwise.
Carefully follow all waste disposal regulations when disposing
waste materials.

2.1  Western Blot
2.1.1  Protein Extraction
1. RIPA buffer: 0.5% (w/v) Sodium deoxycholate, 0.2% (w/v)
SDS, 1.0% (v/v) Triton X-100, 125 mM Sodium cloride,
10 mM Sodium fluoride, 10 mM Sodium orthovanadate,
10 mM Sodium pyrophosphate, 25 mM HEPES, pH 7.5. Add
about 50 mL water to a glass beaker or measuring cylinder.
Weigh 0.5 g Sodium deoxycholate, 0.2 g SDS, 1 g Triton
X-100, 730 mg sodium chloride, 42 mg sodium fluoride,
1184 mg sodium orthovanadate, 446 mg sodium pyrophos-
phate, and 596 mg HEPES. Add water to a volume of
90 mL. Mix and adjust pH and make up to 100 mL with water.
Store at 2–8 °C.
2. PMSF (10 mg/mL): Dissolve 10 mg of PMSF per 1 mL of
Isopropyl alcohol. Shake well to dissolve and store at −20 °C.
3. Aprotinin: 1 mg/mL (see Note 1). Store at −20 °C.
4. Leupeptin: 1 mg/mL (see Note 2). Store at −20 °C.
5. Trypsin solution (1× EDTA in Hank’s Balanced Salt Solution
(HBSS) without calcium or magnesium).

2.1.2  Bradford Assay 1. BSA standard: 100 μg/mL (see Note 3). Store at 2–8 °C.
2. Bio-rad Protein Assay.

2.1.3  SDS-­
Polyacrylamide Gel
1. Separating gel buffer: 1.5 M Tris, 14 mM SDS, pH 8.8. Weigh
45.4 g Tris–HCl and 1 g SDS and transfer to a glass beaker or
a measuring cylinder. Add 200 mL water and adjust the
pH. Make up the final solution to 250 mL by adding water.
Store at 2–8 °C.
2. Stacking gel buffer: 0.5 M Tris, 14 mM SDS, pH 6.8. Weigh
15 g Tris–HCl and 1 g SDS and prepare a 250 mL solution as
in the previous step. Store at 2–8 °C.
 CHO Cell Line Stability Analysis During Long-Term Culture 123

3. Ammonium persulfate: Dissolve 100 mg of ammonium

­persulfate in 1 mL water (see Note 4).
4. SDS PAGE Electrode running buffer: 10 mM Tris, 80 mM
glycine, 1.4 mM SDS (see Note 5).
5. SDS PAGE Sample buffer (2×): 20% (v/v) glycerol, 7 mM
SDS, 0.025% (w/v) bromophenol blue. Weigh 0.76 g Tris–
HCl, 2 g SDS, 5 mg Bromophenol blue and mix with 40 mL
water and 10 mL Glycerol. On the day of use add 18 μL of
β-mercaptoethanol per 1 mL of buffer (see Note 6).
6. Bromophenol Blue solution: Dissolve 0.1 g Bromophenol
blue in 100 mL water.

2.1.4  Protein Transfer 1. Transfer pads or thick filter paper.

and Western Blotting 2. Nitrocellulose membrane.
3. Blotting buffer: 25 mM Tris, 190 mM glycine, 20% (v/v)
methanol. Weigh 3 g Tris–HCl and 14.4 g Glycine. Mix and
make it up to 800 mL with water. Adjust pH to 7.4 and then
add 200 mL Methanol.
4. Ponceau Stain: 0.5% (w/v) in 1% (v/v) glacial acetic acid
(see Note 7).
5. Blocking buffer: 3% (w/v) milk powder in PBS.
6. Wash buffer: 3% (v/v) Tween-20 in PBS.
7. TBS-Tween: Weigh 2.42 g Tris–HCl, 16.35 g NaCl and
­measure 2 mL Tween. Make it up to 2 L with water. Adjust pH
to 7.4.
8. Stripping buffer: 0.1 M glycine, pH 2.5. Weigh 3.75 g Glycine
and mix with 400 mL water. Adjust pH to 2.5 and then make
the solution up to 500 mL with water.

2.2  ELISA 1. Coating buffer: Prepare 100× stock solution by dissolving

7.15 g sodium bicarbonate and 1.58 g sodium carbonate in
100 mL water. The pH should fall in the range 8.6–9.2. Add
0.1 g phenol red powder and stir for 12–24 h at 2–8 °C. Filter
through Whatman paper, aliquot into 50 mL amounts, and
store at 2–8 °C for up to 6 months. Dilute this stock solution
1:100 and use within 8 h. Store at 2–8 °C.
2. Dilution buffer: 1% BSA in PBS-Tween (0.1% [v/v] Tween
20, 1× PBS).
3. ELISA plate wash buffer: Prepare 0.05% (v/v) Tween-20,
0.05% (v/v) phenol red solution (from a 1% w/v stock), in
sterile PBS.
4. ELISA blocking solution: 25% [w/v] BSA, 0.4% [w/v] sodium
azide in sterile PBS. Mix 0.2 g sodium azide and 12.5 g BSA
in 500 mL sterile PBS. Store for a month at 2–8 °C.
124 Zeynep Betts and Alan J. Dickson

5. ELISA development solution: Dissolve two TMB (3,3′,5,5′

Tetramethyl Benzidine Chromogen) tablets and 5 μL 30%
(v/v) hydrogen peroxide in 12 mL TMB substrate solution
(10 mM sodium acetate and 10 mM sodium citrate, pH 5.5,
see Note 8).
6. 0.2 M sulfuric acid: Make a 500 mL solution with 9.8 mL sul-
furic acid and 490.2 mL water (see Note 9).

2.3  Genomic DNA 1. EDTA-Sarcosine solution: 0.1 M EDTA, pH 8.0, containing

Extraction 0.5% (w/v) N-Lauroyl-Sarcosine. Weigh 2.92 g EDTA and
0.5 g N-Lauroyl-Sarcosine dissolve in 100 mL water.
2. Proteinase K: Make 2 mL of 10 mg/mL stock solution in
water and store at −20 °C.
3. RNase A: Make 2 mL of 10 mg/mL stock solution in water
and store at −20 °C.
4. 3 M sodium acetate (pH 5.2). Weigh 49.2 g sodium acetate
and transfer to a beaker or measuring cylinder. Add 150 mL of
water and adjust pH to 5.2. Make it up to 200 mL by adding
more water.

2.4  RNA Extraction 1. Denaturing solution: 4 M guanidinium thiocyanate, 25 mM

sodium citrate (pH 7.0), 0.5% N-laurosylsarcosine (w/v), and
0.1 M 2-mercaptoethanol. Prepare stock solutions of 0.75 M
sodium citrate and 10% N-laurosylsarcosine. Weigh 250 g gua-
nidinium thiocyanate and dissolve in 293 mL water at 65 °C.
Then add 17.6 mL of 0.75 M sodium citrate and 26.4 mL of
10% N-laurosylsarcosine (w/v). This will be a stock-­denaturing
solution that can be stored up to 3 months at room tempera-
ture. To prepare working denaturing solution, add 0.36 mL of
98% 2-mercaptoethanol to 50 mL of stock denaturing solu-
tion. This working solution can be stored up to a month at
room temperature (see Note 10).
2. 2 M sodium acetate pH 4.0: Weigh 16.42 g sodium acetate
and add to 40 mL water and 35 mL glacial acetic acid. Adjust
pH to 4.0 using glacial acetic acid and make the volume up to
100 mL with DEPC-treated water. Store up to 1 year at room
temperature (see Note 11).
3. Water-saturated phenol: Dissolve 100 g nucleic acid grade
phenol crystals in water at 65 °C. Aspirate the upper water
phase and store up to 1 month at 2–8 °C (see Note 12).
4. Chloroform:isoamyl alcohol (49:1 v/v): Mix 49 mL of chloro-
form with 1 mL of isoamyl alcohol. Prepare in a fume hood
just before use each time.
 CHO Cell Line Stability Analysis During Long-Term Culture 125

5. 75% ethanol: Add 75 mL absolute ethanol to 25 mL DEPC-­

treated water.
6. DEPC-treated water (0.05%): Add 0.5 mL of diethylpyrocar-
bonate (DEPC) to 1 L of water. Loosely screw the lid of the
bottle and leave in the fume hood for 12–24 h. Sterilize by
autoclaving to inactivate DEPC. Store at 15–25 °C.

2.5  Fluorescence 1. Hypotonic solution (75 mM KCl): Weigh 2.8 g KCl and dis-
In Situ Hybridization solve in 500 mL of water.
(FISH) 2. Fixative: Methanol:Acetic acid (3:1). Measure 300 mL of
methanol and transfer to a glass bottle. Add 100 mL of acetic
acid. Mix well and store at −20 °C.
3. 0.5 M EDTA (pH 8.0): Weigh 14.612 g EDTA and dissolve in
70 mL water. Adjust pH to 8.0 and bring the volume up to
100 mL with water.
4. TBE buffer (0.09 M Tris, 0.09 M orthoboric acid and 0.2 mM
EDTA, pH 8.0): Prepare 10× concentrated stock TBE buffer
and dilute the solution to 1× before use (1:10, see Note 13).
5. 2% (w/v) agarose: Add 1 g agarose in 50 mL TBE buffer.
Dissolve agarose by heating in a microwave. Cool it down to
55 °C and add 5 μL EtBr.
6. Loading buffer 10× concentrated (100 mM EDTA [pH 8.0],
100 mM Tris–HCl [pH 7.4], 100 mM Tris–HCl, ~0.5% [w/v]
bromophenol blue, and 25% [w/v] Ficoll): Weigh 2.5 g
Ficoll-400. Measure 1 mL 1 M Tris–HCl (see Note 14) and
2 mL 0.5 M EDTA. Bring the solution to 10 mL with water,
heating to 65 °C. Add 25–50 mg bromophenol blue dye.
7. Pepsin/HCl solution (0.05% [w/v] pepsin, 0.01 M HCl):
Prepare 5% w/v pepsin with water and store in aliquots before
use (see Note 15).
8. 10% FBS/PBS (v/v): Add 10 mL FBS into 90 mL PBS.
9. FISH probe hybridization buffer (10% dextran sulfate (v/v),
50% formamide (v/v), 2× SSC [0.3 M NaCl, 0.03 M tri-­
sodium citrate, pH 7.0]): Dilute 20× SSC (see Note 16) into
4× SSC (i.e., 0.4 mL 20× SSC with 1.6 mL water). Add 20 μL
Tween-20 (~3 drops). Add 0.2 g Dextran sulfate and vortex
until dissolved. Place 1 mL of this solution into a clean tube.
Add 1 mL of high grade formamide (carry out in a fume hood).
Mix the solution (see Note 17).
10. 50% formamide/2× SSC (v/v): Mix 50 mL formamide, 10 mL
20× SSC, and 40 mL water. Store at −20 °C in aliquots of
50 mL or make up fresh on the day of use.
11. 1% (v/v) FBS/PBS: Mix 1 mL of FBS with 99 mL of PBS.
126 Zeynep Betts and Alan J. Dickson

3  Methods

3.1  Detection 1. Harvest anchorage-dependent cells by removing the medium

of Recombinant and washing the cell layer with 5 mL PBS. Add 3 mL trypsin
Protein Expression solution and agitate by hand until the cells are detached. Stop
by Western Blot trypsinization by adding equal amount of growth medium.
Transfer the cells to a 50 mL tube and centrifuge at 100 × g for
3.1.1  Protein Extraction
5 min. For suspension cells simply centrifuge at 100 × g for
5 min. Discard the supernatant and resuspend the cells in
10 mL of PBS for counting.
2. Wash the cell pellet with 5 mL of 1 × PBS and centrifuge at
100 × g for 5 min.
3. Resuspend the pellet in RIPA buffer using 300 μL for every
1 × 107 viable cells.
4. Add protease inhibitors, PMSF (10 mg/mL stock), Aprotinin
(1 mg/mL stock), and Leupeptin (1 mg/mL stock). Add
10 μL of each protease inhibitor for every 1 mL of RIPA buffer
used per sample.
5. Sheer the extracts by passing through a syringe and 21 G nee-
dle. Then add 3.5 μL PMSF (10 mg/mL stock) per 1 × 107
cells, incubate cell lysates on ice for 30 min, and centrifuge at
11,000 × g at 4 °C for 30 min.
6. Transfer supernatant to fresh tubes and aliquot into 100 μL
amounts before storage at −80 ° C.

3.1.2  Protein 1. Prepare standard BSA solution (100 μg/mL) and dilute the
Quantitation by Bradford cell lysates to an appropriate degree for assessment against the
Assay standard.
2. For the generation of standard curves, add BSA standard and
water to the wells in the range of 5–60 μL up to a total volume
of 60 μL in each well.
3. Add 1 μL of diluted cell lysates and 59 μL water to the other
wells. Analyze cell lysates and standards in duplicate.
4. Dilute Bio Rad protein assay reagent in 1:3 and add 60 μL of
this diluted reagent to all wells.
5. Measure the absorbance at 570 nm in a plate reader after
10–15 min.
6. Calculate the protein amount in cell lysates by comparing to
the standard curve generated with BSA.

3.1.3  SDS-­ The system consists of a 12.5% (w/v) separating gel overlaid by a
Polyacrylamide Gel 4% (w/v) stacking gel.
1. Prepare separating gel by mixing 6.2 mL Protogel solution
(30% [w/v] Acrylamide), 3.75 mL separating buffer, and
5.05 mL ddH2O in a 50 mL conical tube.
 CHO Cell Line Stability Analysis During Long-Term Culture 127

2. Prepare stacking gel by mixing 1.6 mL Protogel solution,

2.5 mL stacking buffer, and 6 mL ddH2O in a 50 mL conical
tube.
3. Prepare the gel cassette and just before casting the gel add
150  μL ammonium persulfate (100 mg/mL) and 15 μL
TEMED to the separating mixture to initiate polymerization.
Mix by inverting the tube.
4. Immediately cast the separating gel. Allow space for the stack-
ing gel and leave the gel to set for 10 min (see Note 18).
5. Once the separating gel has set, pour the excess water and then
add 100 μL ammonium persulfate (100 mg/mL) and 15 μL
TEMED to the stacking gel mixture and mix by inversion.
Pour the stacking gel onto the separating gel until it reaches
the top of the cast. Insert the gel comb immediately without
introducing air bubbles (see Note 19). Leave the gel to set.
6. Place the gel construct into a designated tank and fill the tank
with running buffer.
7. Prepare the protein samples as follows: Calculate the amount
of sample that contains 20 μg of protein and make this volume
up to 15 μL in ddH2O. Add 15 μL of 2× sample buffer (1:1).
Add β-mercaptoethanol at a final concentration of 1.75% (v/v)
to the sample buffer just before use.
8. Boil the samples at 100 °C for 5 min and centrifuge briefly to
bring down the condensate.
9. Load the marker into first and last lanes and samples into the
rest of the lanes (see Note 20).
10. Electrophorese at 60 V until the bromophenol blue dye reaches
the separating gel and then at 200 V until the dye reaches the
bottom of the gel.

3.1.4  Protein Transfer 1. After the separation, remove the gels from stand and soak them
and Western Blotting in blotting buffer for 20 min.
2. Cut nitrocellulose membrane to the approximate gel size and
place the membrane and two pieces of transfer pad per gel into
blotting buffer for 10 min prior to use.
3. Place a soaked transfer pad onto the lower plate of a Semi-Dry
electroblotter (see Note 21). Place the membrane onto the
pad, and carefully place the gel onto the membrane. Finally,
place the second piece of transfer pad on the top of the gel
(see Fig. 1).
4. Secure the lid and transfer the gel at 15 V for 45–60 min.
5. To assess whether proteins are transferred successfully, stain
the membrane with Ponceau stain. To remove the stain, add a
small amount of TBS-Tween and shake for a few minutes.
128 Zeynep Betts and Alan J. Dickson

Fig. 1 Gel-membrane assembly for the transfer of proteins from SDS-PAGE gels to nitrocellulose membranes

6. Block the membrane in 3% (w/v) milk powder in PBS

­overnight at 4 °C with shaking.
7. Dilute the primary antibodies in the blocking buffer and incu-
bate the membranes with a primary antibody for either 30 min
at room temperature or overnight at 4 °C with shaking.
8. Wash the membranes twice with PBS, twice with 3% (v/v)
Tween-20 in PBS and twice with PBS again for 5 min.
9. Incubate the membranes for 30–60 min with relevant second-
ary antibody at room temperature with shaking. Repeat the
washing steps as above.
10. Once the last wash has been drained from the membrane, pro-
tein bands can be detected with an enhanced chemilumines-
cence (ECL) system.
11. Expose the membranes to Kodak X-ray film and assess band
densities by densitometric analysis using ImageJ software.
12. When the films have developed wash the membrane in TBS-­
Tween at room temperature for 15 min and strip by using a
mild stripping buffer for 30 min at room temperature with agi-
tation. After stripping, wash membranes briefly with PBS,
block and incubate with another primary antibody as above
(see Note 22).

3.2  Detection 1. Prepare serial dilution of the protein of interest as standard.

of Proteins
by  Enzyme-­Linked
Immunosorbent Assay 2. Perform an initial ELISA by using samples diluted to a range
(ELISA)
3.2.1  Preparation
of Standard Curve
and Samples for ELISA
Dilutions are made with dilution buffer to give a final concen-
tration of 0–8000 pg/mL.
of dilutions, between 1:1000 and 1:50,000 in the dilution buf-
fer. This is done to assess the dilution that would fall within the
range of the standard curve. Once suitable sample dilutions are
established, diluted samples and standards can be loaded in
duplicate/triplicate and mean values can be taken to quantify
protein of interest content of each sample.

3.2.2  ELISA 1. Coat the wells of a PVC immunoassay plate with the capture
antibody specific for a protein of interest at a final concentra-
tion of 1–10 μg/mL. Load 100 μL per well.
 CHO Cell Line Stability Analysis During Long-Term Culture 129

2. Cover the plate and incubate overnight at 4 °C.

3. Discard the coating solution the following day and wash the
plate three times by forcefully filling each well with 250 μL
wash buffer. Remove the solutions or washes by flicking the
plate over a sink. The remaining drops are removed by patting
the plate on a paper towel.
4. Block the remaining protein binding sites by adding 220 μL
blocking solution per well and incubating for 1 h at room
temperature.
5. Discard the blocking solution and add 100 μL diluted stan-
dards and samples to each well (see Note 23). Incubate plates
for 2 h at 37 °C.
6. Remove standards/samples and wash the wells three times
with wash buffer.
7. Add 100  μL diluted detection antibody to each well
(see Note 24).
8. Incubate plates at room temperature for 2 h. After the incuba-
tion, wash wells three times with wash buffer and blot dry.
9. Add 100 μL conjugated secondary antibody diluted in block-
ing buffer immediately before use and incubate at room
­temperature for 90 min.
10. Wash the plate subsequently with wash buffer three times.
11. Prepare the ELISA development solution immediately before
use and add 100 μL development solution to each well and
incubate for 20–30 min at room temperature.
12. Stop the reaction by adding 100 μL 0.2 M sulfuric acid to each
well. Measure the OD of each well at 450 nm.
13. Prepare a standard curve from the serial dilutions and interpo-
late the concentration of the sample from this standard curve.

3.3  Plasmid and RNA All procedures should be carried out at room temperature unless
Copy Number Analysis otherwise stated.
by Real-Time PCR
1. Harvest anchorage-dependent cells by removing the medium
3.3.1  Genomic DNA and washing the cell layer with 5 mL PBS. Add 3 mL trypsin
Extraction solution and agitate by hand until the cells are detached. Stop
trypsinization by adding an equal amount of growth medium.
Transfer the cells to a 50 mL tube and centrifuge at 100 × g for
5 min. For suspension cells simply centrifuge at 100 × g for
5 min. Discard the supernatant and resuspend the cells in
10 mL of PBS for counting.
2. Use approximately 2 × 107 cells. Wash the cell pellet three
times in 1× PBS with centrifugation at 100 × g for 10 min
between each wash.
130 Zeynep Betts and Alan J. Dickson

3. Resuspend the final pellet in 100 μL of 1 × PBS and then add

3 mL of EDTA-Sarcosine solution to the pellet in a dropwise
fashion with continuous gentle mixing.
4. Add 60 μL of Proteinase K (10 mg/mL) and 10 μL of RNase
A (10 mg/mL) to the mixture.
5. Incubate the mixture at 55 °C for 2 h and mix by inversion
every 15 min.
6. Add an equal volume of (3 mL) phenol/chloroform/isoamyl
alcohol (25:24:1) and mix the solution by inversion for 5 min.
7. Centrifuge the mixture at 13,000 × g for 10 min at room
­temperature and remove the upper aqueous phase into a fresh
tube. Repeat the phenol/chloroform/isoamyl extraction on
the aqueous phase for three rounds.
8. Add 4 volumes of ddH2O and 0.5 volumes of 3 M sodium
acetate (pH 5.2) to the final aqueous phase.
9. Precipitate DNA by adding 3 volumes of 100% ethanol, mix
the solution by inversion, and centrifuge the mixture at
13,000 × g for 10 min. Remove the supernatant and discard.
10. Wash the pellet with 70% ethanol and re-centrifuge as above.
Air-dry the final DNA pellet for 5–10 min and resuspend in
300 μL of ddH2O (or TE buffer).
11. Determine the DNA concentration using UV spectrophotom-
eter. The purity can be assessed by using the A260nm/A280nm
ratio, where a ratio between 1.6 and 2.0 is considered pure.

3.3.2  Preparation 1. Calculate the molecular mass of plasmid as outlined in Table 1.

of Standard Curve 2. Dilute parental genomic DNA (from non-transfected cells) to
and DNA Samples a concentration of 10 ng/μL. This is used in serial dilutions of
plasmid DNA to ensure the efficiency of the PCR reaction is
the same for all samples.
3. Dilute plasmid DNA to a final concentration of 10,000,00
copies per reaction (since every reaction is 5 μL; final dilution
is 200,000 copies per μL). To achieve this concentration per-
form 1:10 serial dilutions using the diluted parental DNA for
the final two dilutions.
4. Prepare serial dilutions of plasmid DNA (using parental DNA
as diluent) to give a range of 1,000,000–457 copies per 5 μL
reaction. Use 1:3 seven consecutive dilutions to achieve this
(100 μL of final dilution of plasmid DNA into 200 μL diluted
parental DNA).
5. Make 20  μL aliquots of these stock dilutions and store at
−80 °C, for use for future assays. Preparing a stock of stan-
dards results in better repeatability of the standard curves
between assays performed at different times.
 CHO Cell Line Stability Analysis During Long-Term Culture 131

Table 1
Overview of how to calculate the molecular mass of plasmid

Number of bases
Base in plasmida Molecular weight of bases (g/mol) Calculation
A a 331.2 = a × 331.2
C c 307.2 = c × 307.2
G g 347.2 = g × 347.2
T t 322.2 = t × 322.2
Total molecular weight of plasmid Sum of above
(total Mw)
Avogadro constant 6.022 × 1023 molecules/mol
Mass of one plasmid copy (g) Total Mw/6.022 × 1023
Number of bases must include sense and anti-sense strands
a

6. Dedicate one genomic DNA sample as a “check” sample and

dilute this sample to a final concentration of 20 ng/μL. Further
dilute check sample to 10 ng/μL and 5 ng/μL using ddH2O,
and prepare aliquots to store at −80 °C for future assays.
7. Dilute all other samples to be tested to a final concentration of
10 ng/μL in ddH2O.

3.3.3  Real-Time PCR 1. Add the following reagents to each well:

Reaction ●● 5 μL of appropriately diluted sample.
●● 2.5 μL of forward primer (10 μM).
●● 2.5 μL of reverse primer (10 μM).
●● 10 μL of 2× SYBR® Green real-time PCR master mix.
2. Seal the wells and centrifuge at 900 × g.
3. Analyze the samples and standards in triplicate. Additionally,
prepare triplicates of 5 μL ddH2O and non-reverse transcrip-
tase treated sample as negative controls.
4. Primers specific for β-actin or other housekeeping genes can be
used as loading control.
5. Set up the amplification parameters depending on the primers
and the thermal cycler used. It may be necessary to optimize
the system for individual primers.

3.3.4  Data Analysis
1. After the cycles are complete, the SYBR green fluorescence of
each sample can be visualized by using appropriate software.
After a successful real-time PCR experiment, the SYBR green
fluorescence is plotted against the number of cycles, creating
the initial lag, the exponential increase, and the plateau phases.
Set up the threshold within the exponential phase start and
132 Zeynep Betts and Alan J. Dickson

0.25
Plateu

0.2
Fluorescence

0.15

Ct value
0.1
Threshold

0.05

0
10 20 30
Cycle

Fig. 2 Diagram shows the principle of real-time PCR using three representative samples

determine the Ct values for each sample (the cycle value at the
point where the sample line crossed the threshold, see Fig. 2).
The inclusion of standard curves allows the program to calcu-
late the amplification efficiency and the relative quantity of the
target sequence in every sample.
2. To increase accuracy the value of the abundance of each target
sequence can be standardized to the abundance of the house-
keeping gene.
3. Assess the melting curve to check the quality of the amplified
product where a single peak at between 80 and 90 °C indicates
a pure product.

3.3.5  RNA Extraction
All tubes, tips, and solutions should be RNase free.
1. For anchorage-dependent cells, remove medium from culture
and lyse cells by adding of 1 mL of denaturing solution per
10 cm2 and passing the cell lysate several times through a
pipette (see Notes 25 and 26).
2. For cells in suspension, pellet cells by centrifugation. Determine
the amount of viable cells and add 1 mL denaturing solution
per 1 × 107 cells to cell pellets directly (see Note 27). Resuspend
the lysate with a disposable 1 mL pipette by pipetting up and
down at least ten times.
3. Aliquot the cell lysate into a 4 mL polypropylene tube.
4. Add 0.1 mL of 2 M sodium acetate to 1 mL of lysate and mix
thoroughly by inversion.
 CHO Cell Line Stability Analysis During Long-Term Culture 133

5. Add 1 mL water-saturated phenol again mix by inverting

­thoroughly (see Note 28).
6. Add 0.2 mL of chloroform/isoamyl alcohol (49:1), shake vig-
orously for about 10 s (see Note 29).
7. Cool the samples on ice for 15 min.
8. Centrifuge the samples at 10000 × g for 20 min at 2–8 °C.
This will separate the mixture into a lower red phase (phenol-­
chloroform), a white interphase, and a colorless upper aqueous
phase, which contains RNA. Transfer very carefully the upper
aqueous phase to a fresh tube.
9. Precipitate the RNA by adding 1 mL Isopropyl Alcohol, mix,
and incubate samples for 1 h at −20 °C (see Note 30).
10. Centrifuge the samples at 10000 × g for 20 min at 2–8 °C. The
RNA precipitate is invisible before centrifugation and forms a
gel-like pellet at the bottom and side of the tube (see Note 31).
11. Carefully discard the supernatant without losing the pellet
(see Note 32). Wash the RNA pellet with 0.5–1 mL 75% etha-
nol (v/v in DEPC-treated water). Mix the sample by vortexing
(see Note 33).
12. Incubate samples for 10–15 min at room temperature to
­dissolve possible residual traces of guanidinium.
13. Centrifuge at 7500 × g for 5 min at 2–8 °C and carefully
­discard the supernatant.
14. Air-dry the pellet for 5–10 min but do not overdry as this
will decrease RNA solubility. Dissolve the RNA in 30 μL
­DEPC-­treated water mixing by pipetting gently. Incubate the
samples at 55–60 °C for 10–15 min to ensure complete solu-
bilization. Samples can be stored at −80 °C until required for
analysis
15. Determine purity and concentration of each sample using UV
spectrophotometer at wavelengths of 260 and 280 nm. Pure
preparations of RNA have an A260/A280 ratio of between 1.8
and 2.0.

3.3.6  DNAse 1. Prepare the following reagents in a 0.5 mL microcentrifuge

I Treatment of RNA tube:
●● 1 μg of RNA in 8 μL DEPC-treated ddH2O.
●● 1 μL of 10× Reaction buffer.
●● 1 μL of DNase I enzyme (1 Unit/μL).
2. Incubate the reaction at room temperature for 15 min and add
1 μL of stop solution after incubation.
3. Heat the solution at 70 °C for 10 min to denature both the
DNase I and RNA. Chill on ice.
134 Zeynep Betts and Alan J. Dickson
3.3.7  cDNA Synthesis
3.3.8  Preparation
of Standard Curve and RNA
Samples
Subheadings 3.3.3 and 3.3.4.
3.4  Plasmid Location
Analysis
by Fluorescence
In Situ Hybridization
(FISH)
3.4.1  Preparation
of Metaphase Spreads
1. Add 1 μL of Oligo (dT)18 and 1 μL of 10 mM dNTP to DNase
I-treated samples (on ice) and incubate at 65 ° C for 10 min.
Place on ice for 2 min.
2. Prepare the following reaction mix per sample on ice:
●● 4 μL of 5 × RT buffer.
●● 1 μL of RNase inhibitor.
●● 0.25 μL of Reverse Transcriptase (200 U/μL).
●● Up to 10 μL DEPC-treated ddH2O.
3. Add 10 μL of the reaction mix to the tube containing the
­sample. Mix the reaction and incubate at 42 °C for 1 h and
terminate the reaction by heating to 70 °C for 15 min.
4. Reactions can be stored at −20 °C until needed.
1. Dedicate one sample as the “standard” sample and run on all
plates to allow internal comparison of the mRNA content of
the samples.
2. Dilute the cDNA reaction from the standard sample 1:5 in
ddH2O, to give the 100% standard. Prepare serial dilutions
from the 100% standard in ddH2O to give 10% and 1% final
concentrations.
3. Dilute all other samples at a ratio of 1:7 with ddH2O.
Carry out real-time PCR reaction and data analysis as in
1. Grow cells until they are approximately 50% confluent and
then add colcemid solution at a final concentration of 130 ng/
mL (w/v). Incubate for 16–20 h, at 37 °C with 5% CO2.
2. Take out the medium and add trypsin/EDTA solution enough
to cover the surface. After approximately 5 min incubation
inactivate the enzyme by adding culture medium or serum to
the cells. Harvest cells by centrifugation at 100 × g for 5 min.
Resuspend the cell pellets in approximately 100 μL of fresh
growth medium by gently tapping the tube.
3. Add 10 mL of hypotonic solution (75 mM KCl) dropwise,
with gentle mixing, to the resuspended cells.
4. Incubate cells in the hypotonic solution at room temperature for
10 min and then centrifuge at 220 × g for 5 min (see Note 34).
5. Remove supernatant and resuspend the cell pellet in approxi-
mately 100 μL of fresh hypotonic solution.
6. Add 5 mL of ice cold methanol:acetic acid (3:1) to the resus-
pended cells. Prepare the fixative solution fresh.
7. Centrifuge the cells at 220 × g for 5 min and remove the
supernatant.
 CHO Cell Line Stability Analysis During Long-Term Culture 135

8. Repeat steps 6 and 7 three times.

9. After this, resuspend the cells in 100 μL of ice-cold
methanol:acetic acid (3:1) for fixation (see Note 35).
10. Add approximately 10 μL of this cell suspension onto a pre-
cleaned (wiped with acetic acid and allowed to evaporate) glass
slides.
11. Immediately expose the slides, face up, into the steam of hot
water (90 °C) for 30 s to cause the cells to blow up (see Note 36).
12. Look at the spreads under a phase contrast microscope to
check if the concentration and the cell distribution are good.
13. Leave the spreads to age (at least overnight and up to 4 weeks)
at room temperature before examination.

3.4.2  Preparation 1. Prepare FISH probes with plasmid DNA through incorpora-
of Probes for FISH tion of modified dUTPs via nick translation.
Analyses 2. For each reaction, resuspend 1 μg of plasmid DNA in 16 μL
ddH2O.
3. Add 4 μL digoxigenin (DIG)-nick translation mix to the reac-
tion mixture and incubate whole mixture at 15 °C for approxi-
mately 3 h.
4. Halt the nick translation reaction by transferring the reaction
tube on ice.
5. Separate 5 μL of the reaction product on a 2% agarose gel to
confirm that the plasmid DNA has been reduced to under
300 bp in size. If plasmid size is above 300 bp, resume the
reaction by incubating the reaction mix at 15 °C until plasmid
is the optimal size.
6. When the correct probe length is achieved stop the reaction by
adding 1 μL 0.5 M EDTA (pH 8.0) per 20 μL reaction volume
and heat to 65 °C for 10 min.
7. Nick translated probes can be stored at −20 °C until needed.

3.4.3  Agarose Gel 1. Prepare agarose gel by dissolving 2% (w/v) agarose in TBE
Electrophoresis buffer by boiling in a microwave.
2. Once the gel has cooled to less than 55 °C add ethidium bro-
mide to a final concentration of 0.25 μg/mL.
3. Set the gel and run in horizontal electrophoresis tank with
TBE as running buffer.
4. Mix the samples at a 5:1 ratio with loading buffer and load into
wells. Also load 5 μL of DNA Hyperladder I as a reference.
5. Separate the DNA fragments by electrophoresis at 70 V for
45 min – 1 h, and visualize by UV light.
136 Zeynep Betts and Alan J. Dickson

3.4.4  Hyridization 1. Dehydrate metaphase spreads on glass slides through sequential

Protocol incubation in increasing concentrations of ethanol solution
(70%, 90%, 100% [v/v]) for three times, 3 min each.
2. Air-dry the slides and then incubate in 0.05% (w/v) pepsin in
0.01 M HCl solution at 37 °C for 20 min (see Note 37).
3. Immerse the slides in 10% FBS/PBS (v/v) solution to quench
pepsin digestion and then dehydrate again, as described in
step 1.
4. For each hybridization reaction, mix 25 μL of nick-translated
probe with 5 μL herring sperm DNA and precipitate with
ethanol.
5. Wash the resultant DNA pellet with 70% ethanol (v/v) and
resuspend in 30 μL FISH probe hybridization buffer.
6. Apply 15 μL of this hybridization solution onto each slide, and
then cover the area of application with a 22 mm × 22 mm cov-
erslip. Seal the edges of the coverslip with nail varnish.
7. Incubate the slides on a heat block at 70 °C for 2 min for dena-
turation, followed by incubation in a humidified chamber at
37 °C for 16 h.
8. After hybridization, remove coverslips and wash the slides
three times for 3 min each in 50% formamide/2 × SSC (v/v)
at 37 °C.
9. Wash the slides again with 2 × SSC three times for 3 min each
and allow to dry at room temperature.

3.4.5  Antibody Detection The remainder of this procedure must be performed in the dark.
1. Dilute Rhodamine-conjugated Fab fragments 1:10 in 1% (v/v)
FBS/PBS.
2. Apply 25 μL of the diluted antibody to each slide.
3. Cover with a coverslip and incubate in a humidified chamber at
37 °C for 30 min.
4. Wash the slides three times in 2× SSC for 3 min each at 37 °C.
5. Dip the slides quickly dipped in ddH2O and allow to air dry in
the dark.
6. Fix the slides with Prolong anti-fade Gold Dapi and cover with
a 22 mm × 22 mm coverslip.
7. After overnight incubation at room temperature, seal the slides
with nail varnish.
8. Observe and collect the images on a fluorescent microscope
using specific band pass filter sets for FITC and DAPI
(see Fig. 3).
 CHO Cell Line Stability Analysis During Long-Term Culture 137

Fig. 3 Metaphase spreads of recombinant CHO cell lines (engineered to express

GFP) after in situ hybridization with a Digoxigenin labeled probe against the
p1010-GFP plasmid. White arrow indicates plasmid integration site. White scale
bar (bottom right) equals 10 μm

4  Notes

1. Initially add 1 mL water to the 10 mg bottle of Aprotinin to

make a 10 mg/mL stock solution. Then prepare solutions of
1 mg/mL by diluting 20 μL Aprotinin (10 mg/mL) with
180 μL water. Divide this into 20 μL aliquots.
2. Initially, add 100 μL of water to the 1 mg bottle of Leupeptin
to make a 10 mg/mL stock solution. Then prepare solutions
of 1 mg/mL by diluting 20 μL Leupeptin (10 mg/mL) with
180 μL water. Divide this into 20 μL aliquots.
3. Weigh 100 mg BSA and dissolve in 10 mL water to make
10 mg/mL BSA solution as stock. Then prepare solutions of
100 μg/mL by diluting 100 μL BSA (10 mg/mL) with 9.9 mL
water.
4. It is recommended to prepare fresh each time.
5. Prepare 5× native buffer (10 mM Tris, 80 mM glycine, 1.4 mM
SDS) as stock. Weigh 60.1 g Tris–HCl, 288.4 g Glycine, and
20 g SDS mix and make it to 2 L with water. Dilute 200 mL of
5× native buffer by adding 800 mL water.
6. Make aliquots once prepared (a few mLs to use) and store the
one currently used at 2–8 °C. Store the other aliquots at
−20 °C. SDS precipitates at 2–8 °C. Therefore, the solution
needs to be heated prior to use. Do not keep reheating.
138 Zeynep Betts and Alan J. Dickson

7. Mix 5 g of Ponceau S, 10 mL glacial acetic acid and 90 mL

water to prepare 10× Ponceau stain. Dilute this to make 10 mL
of 1× to use (mix 1 mL stain with 9 mL water). This diluted
solution can be reused a few times.
8. Prepare 0.1 M sodium acetate by weighing 82 mg sodium ace-
tate and dissolving in 10 mL of water. Also prepare 0.1 M
sodium citrate by dissolving 258 mg sodium citrate in 10 mL
of water. Then mix 1.2 mL of each solution with 9.6 mL of
water. Adjust pH to 5.5.
9. Do not add water onto acid. Prepare by adding 50 mL water
into a glass erlenmeyer container and then transfer 9.8 mL sul-
furic acid to the water swirling the glass container under a run-
ning tap to cool it down. Make it up to 500 mL by topping up
with water.
10. To minimize handling of guanidinium thiocyanate, dissolve
directly in the manufacturer’s bottle. The 2-mercaptoethanol
should be handled under a fume hood.
11. Glacial acetic acid should be handled in a fume hood.
12. Phenol should be handled in a fume hood.
13. To prepare 10× TBE buffer weigh following into a glass bea-
ker; 27 g TRIS-base, 13.7 g Orthoboric acid, 0.925 g EDTA.
Dissolve the reagents in 200 mL water and adjust the pH to
8.0. Make the volume up to 250 mL by adding water.
14. Dissolve 121.1 g Tris–base in 700 mL water. Add concentrated
HCl to desired pH 7.4. Bring the volume up to 1 L with water.
Filter and autoclave if desired. Store at room temperature.
15. Add 500 μL of 5% pepsin to 50 mL of 10 mM HCl in a Coplin
jar.
16. In order to prepare 20× SSC weigh 70.125 g sodium chloride
and 35.3 g sodium citrate. Add 350 mL DEPC-treated water
and check pH. Adjust pH to 7.0. Autoclave before use.
17. This volume is enough for a large number of slides, so the vol-
ume can be scaled down if necessary.
18. Gently add a few drops of water or isobutanol to the top layer
of the gel to assist in setting the gel with a straight line.
19. Add a little more gel down the side of the comb to allow for
shrinkage while the gel sets.
20. If there are blank lanes load a small amount of sample buffer
into these lanes to ensure the gel runs level.
21. Gently roll out air bubbles by rolling with a pipette.
22. By stripping you can remove primary and secondary antibodies
from a western blot membrane. Stripping is useful when one
wants to investigate more than one protein on the same blot,
 CHO Cell Line Stability Analysis During Long-Term Culture 139

for instance a protein of interest and a loading control. When

probing for multiple targets, stripping and re-probing a single
membrane instead of running and blotting multiple gels has
the advantage of saving samples, materials, and time.
23. Always compare signal of unknown samples against those of a
standard curve. Run standards and blank with each plate.
Ensure concentration of standards spans the most dynamic
detection range of antibody binding. You may need to opti-
mize the concentration range to obtain a suitable standard
curve. Always run samples and standards in duplicate or
triplicate.
24. Check that the detection antibody recognizes a different
­epitope on the target protein to the capture antibody. This
prevents interference with antibody binding. Use a tested
matched pair whenever possible.
25. Cell cultures should be processed immediately after the removal
from the incubator. Isolate RNA from cells after 24–48 h of
subculture during the exponential phase of growth at the start
and end of long-term cultures.
26. Do not wash the cells before adding the reagent as this may
cause mRNA degradation.
27. Carry out in the fume hood.
28. Never use buffered phenol as acidic pH is the critical factor to
separate RNA from DNA and proteins. Ensure mixing thor-
oughly the organic phase and the aqueous phase.
29. Make sure that the lids are securely closed before mixing and
shaking.
30. Samples can be stored at −20 °C at this point for additional
time and the rest of the procedure can be carried out later.
31. Take special care when you pipette the aqueous phase so as to
not to disturb the interphase and lower organic phase rich in
DNA and proteins. The volume of the aqueous phase you
should retrieve will be almost equal to the initial volume of
denaturing solution.
32. You can improve the removal of DNA by carrying out a second
precipitation step. In order to do this, dissolve the RNA pellet
in 0.3 mL denaturing solution and transfer to a 1.5 mL micro-
centrifuge tube. Add 0.3 mL isopropyl alcohol and incubate
for at least 30 min at −20 °C. Centrifuge at 10000 × g for
10 min at 2–8 °C and discard the supernatant.
33. You can also store your sample either at 2–8 °C for up to
1 week or at −20 °C for up to 1 year and complete the proce-
dure later.
140 Zeynep Betts and Alan J. Dickson

34. At this point the hypotonic solution will cause an increase on

the cellular volume and help to untangle the chromosomes.
The time of incubation is important to acquire good chromo-
some spreads. If the timing is too long, the cell membrane may
burst too early and chromosomes are lost. If too short, it may
be difficult to obtain chromosome spreads because the cell
membrane may not disrupt.
35. Fixed metaphase nuclei preparations can be stored at −20 °C
until required.
36. At this stage, the methanol from the fixative solution evapo-
rates; therefore, acetic acid concentration increases. This stim-
ulates the chromosome spreading.
37. Add 500 μL of 5% pepsin to 50 mL of 10 mM HCl in a Coplin
jar at 37 °C.

References
1. Bailey LA, Hatton D, Field R, Dickson AJ
(2012) Determination of Chinese hamster
ovary cell line stability and recombinant anti-
body expression during long-term culture. Bio­
technol Bioeng 109:2093–2103. doi:10.1002/
bit.24485/abstract
2. Barnes LM, Bentley CM, Dickson AJ (2003)
Stability of protein production from recombi-
nant mammalian cells. Biotechnol Bioeng
81:631–639. doi:10.1002/bit.10517
3. Betts Z, Croxford AS, Dickson AJ (2015)
Evaluating the interaction between UCOE and
DHFR-linked amplification and stability of
recombinant protein expression. Biotechnol
Prog 31:1014–1025. doi:10.1002/btpr.2083
4. Kim SJ, Lee GM (1999) Cytogenetic analysis
of chimeric antibody-producing CHO cells
inthe course of dihydrofolate reductase-­
mediated gene amplification and their stability
in the absence of selective pressure. Biotechnol
Bioeng 64:741–749
5. Derouazi M, Martinet D, Besuchet Schmutz N
et al (2006) Genetic characterization of CHO
production host DG44 and derivative recom-
binant cell lines. Biochem Biophys Res
Commun 340:1069–1077. doi:10.1016/j.
bbrc.2005.12.111
6. Betts Z, Dickson AJ (2016) Ubiquitous chro-
matin opening elements (UCOEs) effect on
transgene position and expression stability in
CHO cells following methotrexate (MTX)
amplification. Biotechnol J 11:554–564.
doi:10.1002/biot.201500159
7. Kim M, O'Callaghan PM, Droms KA, James
DC (2011) A mechanistic understanding of pro-
duction instability in CHO cell lines expressing
recombinant monoclonal antibodies. Biotechnol
Bioeng 108:2434–2446. doi:10.1002/bit.
23189
8. Osterlehner A, Simmeth S, Göpfert U (2011)
Promoter methylation and transgene copy
numbers predict unstable protein production
in recombinant chinese hamster ovary cell
lines. Biotechnol Bioeng 108:2670–2681.
doi:10.1002/bit.23216
9. Kim NS, Kim SJ, Lee GM (1998) Clonal vari-
ability within dihydrofolate reductase-­mediated
gene amplified Chinese hamster ovary cells:
stability in the absence of selective pressure.
Biotechnol Bioeng 60:679–688
10. Chusainow J, Yang YS, Yeo JHM et al (2009)
A study of monoclonal antibody-producing
CHO cell lines: what makes a stable high pro-
ducer? Biotechnol Bioeng 102:1182–1196.
doi:10.1002/bit.22158
11. Barnes LM, Bentley CM, Dickson AJ
(2001) Characterization of the stability of
recombinant protein production in the
GS-NS0 expression system. Biotechnol Bioeng
73:261–270
12. Mahmood T, Yang P-C (2012) Western blot:
technique, theory, and trouble shooting.
N Am J Med Sci 4:429–434. doi:10.4103/
1947-2714.100998
13. Feit C, Bartal AH, Tauber G et al (1983) An
enzyme-linked immunosorbent assay (ELISA)
for the detection of monoclonal antibodies rec-
ognizing surface antigens expressed on viable
cells. J Immunol Methods 58:301–308
14. Hornbeck P, Winston SE, Fuller SA (2001)
Enzyme-linked immunosorbent assays (ELISA).
Curr Protoc Mol Biol 11:112
 CHO Cell Line Stability Analysis During Long-Term Culture
15. Blin N, Stafford DW (1976) A general method
for isolation of high molecular weight DNA from
eukaryotes. Nucleic Acids Res 3:2303–2308
16. Chomczynski P, Sacchi N (1987) Single-step
method of RNA isolation by acid guanidinium
thiocyanate-phenol-chloroform extraction. Anal
Biochem 162:156–159. doi:10.1006/abio.
1987.9999
17. Chomczynski P, Sacchi N (2006) The single-­
step method of RNA isolation by acid guanidin-
ium thiocyanate–phenol–chloroform extraction:
twenty-something years on. Nat Protoc 1:581–
585. doi:10.1038/nprot.2006.83
18. Gall JG, Pardue ML (1969) Formation and
detection of Rna-Dna hybrid molecules in
cytological preparations. Proc Natl Acad Sci
U S A 63:378
19. Lichter P, Cremer T, Borden J et al (1988)
Delineation of individual human chromosomes
in metaphase and interphase cells by in situ
suppression hybridization using recombinant
DNA libraries. Hum Genet 80:224–234
20. Rudkin GT, Stollar BD (1977) High resolu-
tion detection of DNA-RNA hybrids in situ by
141
indirect immunofluorescence. Nature 265:
472–473
21. Trask BJ (2002) Human cytogenetics: 46
chromosomes, 46 years and counting. Nat Rev
Genet 3:769–778. doi:10.1038/nrg905
22. Speicher MR, Carter NP (2005) The new cyto-
genetics: blurring the boundaries with mole­
cular biology. Nat Rev Genet 6:782–792.
doi:10.1038/nrg1692
23. Rens W, Fu B, O'Brien PCM, Ferguson-Smith
M (2006) Cross-species chromosome painting.
Nat Protoc 1:783–790. doi:10.1038/nprot.
2006.91
24. Rigby PW, Dieckmann M, Rhodes C,
Berg P (1977) Labeling deoxyribonucleic acid
to high specific activity in vitro by nick transla-
tion with DNA polymerase I. J Mol Biol 113:
237–251
25. Cox WG, Singer VL (2004) Fluorescent DNA
hybridization probe preparation using amine
modification and reactive dye coupling. Bio­
techniques 36:114–122
26. O'connor C (2008) Fluorescence in situ hybri­
dization (FISH). Nat Educ 1:171
 Chapter 9

Selection of High-Producing Clones Using FACS

for CHO Cell Line Development
Clair Gallagher and Paul S. Kelly

Abstract
Cell line development aims to generate and select clones with desirable characteristics. One of the most
important parameters for biopharmaceutical cell selection is cell-specific productivity (Qp) or the quantity
of product produced per cell per day. Fluorescence-activated cell sorting (FACS) is a powerful, high-­
throughput technique that facilitates multiparametric characterization and isolation of individual cell
clones from heterogeneous populations. Here, we describe a FACS-based method for section of high-­
producing CHO cell clones.

Key words FACS, Flow cytometry, Sort, Cell line development, Biopharmaceutical, Productivity

Abbreviations
FSC Forward scatter
SSC Side scatter
FSC-W Forward scatter-width
FSC-H Forward scatter-height

1  Introduction

Chinese hamster ovary (CHO) cells are one of the most commonly
used systems for the production of recombinant biotherapeutic
proteins. These cells allow complex posttranslational modifications
and protein folding; however, cell productivity is often a limiting
factor for development and large-scale production [1]. During
development, cells are transfected with the recombinant gene of
interest and drug applied to select those that have been stably
transfected. Within this heterogenous cell pool, clones must be
evaluated and those with the highest productivity identified and
isolated. The proportion of high producers within such heteroge-
neous populations is low and these desirable clones tend to be

Paula Meleady (ed.), Heterologous Protein Production in CHO Cells: Methods and Protocols, Methods in Molecular Biology,
vol. 1603, DOI 10.1007/978-1-4939-6972-2_9, © Springer Science+Business Media LLC 2017

143
144 Clair Gallagher and Paul S. Kelly

outgrown by lower productivity cells whose energies are expended

on growth rather than production of product [2, 3]. It is therefore
important to isolate clones of interest and a number of screening
and selection methods have been developed accordingly. Cell clon-
ing by serial dilution is a blind test that requires clones to be iso-
lated without prior knowledge of productivity. While straightforward
to perform, the technique is laborious, requiring significant post-­
isolation testing of predominantly low to medium producing
clones. Low toxicity reporter genes such as GFP may be incorpo-
rated into the vector cassette to inform post-isolation clone selec-
tion and reduce the number of clones which need to be assessed.
As the GFP transgene is retained in the final production line how-
ever, additional steps must be introduced into the production pro-
cess to ensure that the secreted GFP is removed [3].
The emergence of FACS technology has allowed protocols to
be developed which significantly stream-line the process of clone
selection. FACS is a powerful technique that allows rapid, high-­
throughput analysis of individual cells within heterogeneous popu-
lations. The productivity of each cell is first analyzed and this
knowledge determines whether each particular cell may be cloned.
Only cells that reach a predefined threshold will be cloned into
96-well plates, significantly reducing unnecessary post-sort testing
of lower productivity cells. A number of methods have been devel-
oped to allow cells productivity assessment by FACS. Cell surface
markers such as CD20, not produced naturally by CHO cells, may
be incorporated into the vector for co-expression with the antibody
sequence. Fluorescently labeled anti-CD20 may then be used to
bind CD20 epitopes and allow proximal quantification of the prod-
uct. Other methods take more direct approaches capturing secreted
product on the cell surface using porous matrices or cold capture
techniques [2, 4]. The product itself can then be labeled with fluo-
rescently labeled anti-human antibody and the highest fluorescing
cells (i.e., the highest producers) only are cloned for follow-up
[2, 4, 5]. Reiterative sorting of high productivity subclones has
been shown to dramatically reduce the number of screening assays
required by more than tenfold and cell line development duration
by over 50% [5]. The effect on productivity is also stark with
increases in specific productivity of 25- to 120-fold reported [5].
Here, we describe a FACS analysis method for isolation of high-
producing CHO clones from a heterogeneous population of mono-
clonal antibody-producing cells using cold capture. Relevant CHO
cell culture considerations, conditioned media preparation, live cell
immunolabeling and staining, instrument preparation and FACS
analysis templates and sort considerations are all described. This
method is applicable to a broad range of monoclonal antibodies and
fusion proteins and may also be applied directly to other nonhuman
host cells. When the host cells are human in origin (i.e., PER.C6),
the fluorescently labeled anti-human IgG antibody may simply be
exchanged for a fluorescently labeled product-specific antibody.
 Selection of High-Producing Clones by FACS 145

2  Materials

Prepare all solutions using sterile reagents. Adhere to health and

safety and waste disposal regulations when disposing of biological
and chemical waste.

2.1  CHO Cell Culture 1. Single-use 50 mL bioreactor tubes (Thermo Fisher Scientific).
2. Appropriate CHO cell culture media (i.e., CD OptiCHO
(Thermo Fisher Scientific), CHO-S-SFM II (Thermo Fisher
Scientific), or an alternative CHO line-specific medium.
3. Penicillin-Streptomycin solution (Thermo Fisher Scientific).
4. A panel of appropriate CHO cell populations (see Note 1 and
Fig. 1).

Fig. 1 Schematic describing required FACS antibody controls. In order to establish appropriate FACS setting
and accurately assess antibody secretion, the following controls are recommended: (a) an unlabeled CHO cell
sample to establish appropriate forward/side scatter settings and negative fluorescence. (b) Non-secretor CHO
cells labeled with FITC-conjugated anti-human IgG to allow identification of nonspecific background signal and
to position negative “non-producer” FITC gates. A labeled producer cell sample should also be prepared to act
as a positive control (c and d)
146 Clair Gallagher and Paul S. Kelly

a) Non-producers (biological negative control).

b) High producers (positive control).
c) CHO development cell line to be assessed and sorted.

2.2  Immunolabeling 1. Dulbecco’s Phosphate-Buffered Saline (DPBS).

and Staining 2. Trypan Blue solution (0.4%) (Sigma Aldrich).
3. Hemocytometer.
4. 4′,6-Diamidino-2-Phenylindole, Dihydrochloride (DAPI)
(Molecular Probes).
5. Anti-Human IgG F(ab’)2-Fluorescein antibody produced in
goat (Sigma Aldrich), see Note 2.

2.3  FACS 1. Absolute ethanol.

Decontamination 2. Sodium azide (Sigma Aldrich).

2.4  FACS Settings 1. Shaker incubator (i.e., Kuhner Shaker Climo-Shaker ISF1-XC).
and Clone Isolation 2. 96-well plates (Corning).
3. FACS tubes (BD Falcon).
4. FACS with appropriate lasers (405 nm, 488 nm) for excitation
and detection of DAPI (450 nm) and FITC (530 nm).
Appropriate FACS instruments may include the BD FACSAria
or FACSJazz, depending on specifications. See Note 3.

3  Methods

3.1  CHO Cell Culture 1. Culture a panel of appropriate CHO suspension cell lines at
37 °C, 5% CO2, with rotation (170 rpm) (see Table 1 for fur-
ther details).
2. Media requirements and seeding densities may differ by cell
line; please select appropriate media and conditions for cell
lines in use.
3. Growth curves should be determined for each cell line in
advance so that samples may be taken at appropriate times for
the preparation of conditioned media (exponential), immuno-
labeling (exponential), and DAPI staining (late exponential
and death). See Note 4.

3.2  Conditioned 1. Prepare 10 mL of conditioned media per 96-well collection plate.
Media Preparation 2. Condition the media by seeding cells taken during the expo-
nential growth phase in pre-warmed fresh media at
1 × 106 cells/mL.
3. Incubate the cells in media for 24 h in a shaker incubator at
170 rpm, 37 °C, 80% humidity, 5% CO2.
4. Pellet cells by centrifugation at 800 × g for 5 min.
 Selection of High-Producing Clones by FACS 147

Table 1
FACS samples required for isolation of High Qp clones for CHO cell development line

Anti-IgG-­
DAPI fluorophore
Sample ID Function Growth phase stained labeled Sort
Unstained/unlabeled Establishing basic FSC, SSC Late exponential No No No
instrument settings and
identification of non-­
fluorescent populations
Healthy-DAPI stained Gating for identification and Late exponential Yes No No
exclusion of dead cells
Stressed-DAPI stained Gating for identification and Death Yes No No
exclusion of dead cells
Stained non-producer For detection of nonspecific Late exponential Yes Yes No
cell line fluorophore background
signal. Used to set negative
fluorescence gate
Stained High-Ab Positive control for producer Late exponential Yes Yes No
producer cell line cells. Used to set positive
fluorescence gate
Development cell line Isolation of high Qp clones Late exponential Yes Yes Yes

5. Aspirate and retain media in sterile collection tubes

6. Add 100 μL Penicillin Streptomycin per 10 mL of conditioned
media (final Penicillin Streptomycin concentration of 1%).
7. Add 100 μL per well of a 96-well plate.
8. Freeze the plate to remove any remaining live (and potentially
contaminating) cells (see Notes 5 and 6).
9. Warm the conditioned media to 37 °C and equilibrate in an
incubator for 2 h immediately prior to use (see Note 7).

3.3  Laboratory 1. Wipe down benches, pipettes, and nearby surfaces using 70%
and FACS methanol (see Note 8).
Decontamination 2. Prepare sterile sheath fluid.
3. Decontaminate FACS and prepare for aseptic sort using
instrument-­specific instructions.

3.4  Preparation DAPI solutions should be handled and processed in sterile condi-
of DAPI Staining tions to reduce the possibility of bacterial or fungal infection.
Solutions Prepare 5 mg/mL DAPI stock solution by dissolving 10 mg DAPI
in 2 mL of deionized water. Aliquot and store at −20 °C. Prepare
1 μg/mL DAPI working solution by adding 1 μL of DAPI stock
solution (5 mg/mL) to 5 mL DPBS and protect from light
(see Note 9 for further details regarding safe preparation and stor-
age of DAPI solutions).
148 Clair Gallagher and Paul S. Kelly

3.5  Immunolabeling Cells are immunolabeled against humanized antibody to allow

and Staining Cells quantitation of secreted protein and stained with DAPI for live/
dead cell identification. Immunolabeling and staining for all sam-
ple and controls required are outlined in Table 1. All samples and
reagents should be handled and processed in sterile conditions to
reduce the possibility of bacterial or fungal contamination. All
steps should be performed at 4 °C.
1. Seed cells at 2 × 105/mL in single-use bioreactor tubes.
2. Collect aliquots of exponential phase cells (~48 h) and per-
form live cell counts using a hemocytometer and trypan blue.
Add 40 μL Trypan Blue (0.4%) to a 200 μL cell suspension.
Incubate at room temperature for 5 min. Load stained cell
suspension onto the hemocytometer and perform live cell
count. Dead cells will stain blue, viable cells will remain
unstained for 15 min.
3. Aliquot 1 × 106 viable, unstained cells per sample.
4. Wash the cells with DPBS, centrifuge at 200 × g for 5 min, and
discard the supernatant.
5. Repeat the previous step.
6. Resuspend cell pellets in 198 μL of DPBS.
7. Add 2 μL of either goat anti-human-IgG-(AF488-conjugate)
or DPBS to samples (see Table 1 for further details). Effective
antibody dilutions vary depending on the antibody used and
recommendations may be available from the manufacturer.
Generally higher antibody concentrations (dilutions of 1:10–
1:100) are effective for this particular application.
8. Incubate for 30 min at 4 °C with rotation, protected from
light.
9. Wash the cells with DPBS, centrifuge at 200 × g for 5 min, and
discard the supernatant.
10. Repeat the previous step.
11. Resuspend the cell pellet in 200 μL of either DPBS or DAPI
working solution supplemented with 2 μL Penicillin-­
Streptomycin (see Table 1 for further details).
12. Incubate for 5 min at 4 °C.
13. Transfer samples on ice to FACS lab for immediate analysis.

3.6  FACS Analysis 1. Set up FACS instrument according to manufacturer’s

and Single Cell Sorting instructions.
2. Establish data acquisition settings, dot plots, histograms,
and statistics to allow identification of cell populations (FSC
versus SSC), doublet discrimination (FSC-W versus FSC-H),
DAPI discrimination, and FITC-labeled antibody secretion
quantification.
 Selection of High-Producing Clones by FACS 149

a
Live cells Dead cells

750 1,000
Count
500
250
0 102 103 104 105
Pacific Blue-A
b
750 1,000 1,250

Live cells Dead cells

Count
250 500
0

102 103 104 105

Pacific Blue-A

Fig. 2 FACS histograms displaying DAPI-stained healthy (a) and stressed (b) con-
trol cell samples. Dead cells display increased fluorescence signals and gates
may be applied so that only live cells are sorted. The DAPI stain should also be
used to assess the health of the development line to be sorted as suboptimal
health can greatly reduce post-sort clone survival

3. Use Unstained/unlabeled sample to set appropriate forward

scatter (FSC) and side scatter (SSC) voltages and gains.
4. Identify and gate dead cells using Healthy DAPI-stained and
Stressed DAPI-stained control samples (see Table 1). Dead cells
will display DAPI-positive signals, while healthy cells will
remain unstained (Fig. 2). See Note 10.
5. Analyze Stained nonproducer cell line (see Table 1), to set FITC
voltages and gains and position the negative antibody secre-
tion gate (Fig. 3). This gate should be a sub-gate of DAPI-­
negative (live) cells only.
6. Analyze Stained High-Ab producer cell line (see Table 1), to
confirm immunolabeling staining has been successful and to
set positive fluorescence gates.
150 Clair Gallagher and Paul S. Kelly

a
Negative FITC Positive

2 2.5 3 3.5
Count %
0.5 1 1.5
0 102 103 104 105
FITC-A
b
3

Negative FITC Positive

2.5
Count %
2
1.5
1
0.5
0

102 103 104 105

FITC-A

Fig. 3 FACS histograms displaying FITC-conjugated anti-human IgG stained non-­

producer (a) and human IgG producer CHO cell line (b). Antibody-secreting cells
display FITC-positive fluorescent signals. In this example, there is clear separa-
tion between non-producer and producer lines with approximately 20% of the
producer population falling within the FITC-positive gate. Such clear separation
is not always evident, however, comparing the statistics of both populations can
reveal whether there has been a subtle increase in fluorescence (FITC mean,
median, mode) indicative of antibody secretion

7. Once appropriate settings have been finalized begin sample

acquisition.
8. Sort FITC-positive cells into collection plate (see Note 11); 1
cell per well. Sort settings should favor purity rather than yield.
See individual instrument instructions for further details.
9. It may also be advisable to sort a small number of low-­
fluorescence, low producer clones into a separate plate for
comparison.
10. Once cells have been cloned, incubate at 37 °C, 5% CO2, with
rotation (170 rpm). (See Note 12).
11. Inspect wells daily for contamination and cell growth. (See
Note 13). Assess and passage CHO clones as appropriate.
 Selection of High-Producing Clones by FACS 151

12. Serial FACS sorting of high producer clones may further

increase specific productivity of clones.
13. Specific productivity of clones may be validated by Enzyme-­
Linked Immunosorbent Assay of conditioned media.
14. Additional assays including growth curves may further assist
development of CHO cell production lines.

4  Notes

1. CHO cell lines should produce the same category of product

where possible (i.e., chimeric mAb, humanized mAb, or fusion
proteins). Alternate categories of product may have alternate
anti-human antibody-binding requirements.
2. Care should be taken to ensure that the anti-human antibody
selected does not contain sodium azide. Sodium azide has
known cytotoxic and metabolic inhibitory effects that may
result in poor survival of FACS-sorted clones [6, 7]. If your
antibody does contain sodium azide, this may be removed by
filtration. Centrifugal filtration devices suitable for small anti-
body volumes may be purchased from Merck Millipore
(Amicon Ultra-0.5 mL Centrifugal Filters, 3 kDa NMWL, cat
no: UFC500308).
3. Alternate live/dead cell stains (Propidium iodide, SYTOX stains)
or reporter fluorophores (R-phycoerythrin, Allophycocyanin
etc) may be used in place of those described. Online tools such
as BD Biosciences Fluorescence spectrum viewer (http://www.
bdbiosciences.com/us/s/spectrumviewer) or Fluorescence
SpectraViewer (https://www.thermofisher.com/ie/en/home/
life-science/cell-analysis/labeling-­chemistry/fluorescence-spec-
traviewer.html) from Thermo Fisher Scientific allow users to
identify stains and fluorophores to suit their specific instrument
and application needs.
4. It is advisable where possible to select CHO cell controls that
display similar growth rates to the heterogenous development
line. If this is not possible, seed cells so that they are at the
same exponential phase of growth following 48 h incubation.
5. If using the conditioned media immediately, it may be more
convenient to remove remaining cells or particles by filtration
using a 0.4 μm filter.
6. Conditioned media may be stored at −20 °C for 1 month or
−80 °C for 6 months until ready for use.
7. Conditioned media storage vessels should be filled to approxi-
mately 70% total volume before freezing to minimize head-
space and potential pH changes caused by CO2 displacement.
If this is not possible conditioned media should be equilibrated
overnight before use.
152 Clair Gallagher and Paul S. Kelly

8. Materials that encourage bacterial or fungal cultures (such as

cardboard and paper) should not be kept in a FACS lab. Remove
any potential sources of contamination and replace with materi-
als that may be readily decontaminated (such as plastics).
9. DAPI stock solution may be stored at −20 °C for 1 year. DAPI
working solution may be stored at 4 °C, for up to 6 months
and must be protected from light at all times. DAPI is a known
mutagen and should be handled and disposed of in accordance
with safely regulations.
10. If >5% of cells are DAPI positive, it may indicate that the cell
population is under stress. Using cells in this condition can result
in poor clone survival and growth. Recheck growth curves and
seeding densities to ensure that cells are in good health before
initiating immunolabeling procedure and FACS sorting.
11. It may be tempting to select only those events that display the
highest fluorescent signals; however, this is not always the best
real-world strategy. Highest producers may also exhibit
unusual characteristics that are undesirable for biopharmaceu-
tical production such as slow growth, cell density limitations,
short lifetime, propensity toward apoptosis, high glutamine
requirements, or sensitivity to ammonia. It is therefore advis-
able to isolate a range of FITC-positive clones; as such cells
may demonstrate characteristics that compensate for slightly
lower specific productivity (Qp).
12. Antibiotics are not routinely used during CHO cell culture;
however, Penicillin-Streptomycin may be applied to isolated
clones at 1% for 1 week post-sort to reduce the possibility of
bacterial contamination.
13. It can be difficult to observe a single cell in suspension. Cells are
more easily observed after a number of days in culture (~7 days).

References
1. Li F, Vijayasankaran N, Shen AY et al (2010) course of the cold capture antibody secretion
Cell culture processes for monoclonal antibody assay. J Biotechnol. doi:10.1016/j.jbiotec.
production. MAbs 2:466–479 2009.03.001
2. Borth N, Zeyda M, Kunert R, Katinger H 5. Brezinsky SC, Chiang G, Szilvasi A et al (2003)
Efficient selection of high-producing subclones A simple method for enriching populations of
during gene amplification of recombinant transfected CHO cells for cells of higher specific
Chinese hamster ovary cells by flow cytometry productivity. J Immunol Methods 277:141–
and cell sorting. Biotechnol Bioeng 71:266–273 155. doi:10.1016/S0022-1759(03)00108-X
3. Caron AW, Nicolas C, Gaillet B et al (2009) 6. Slamenová D, Gabelová A (1980) The effects
Fluorescent labeling in semi-solid medium for of sodium azide on mammalian cells cultivated
selection of mammalian cells secreting high-­ in vitro. Mutat Res 71:253–261
levels of recombinant proteins. BMC Biotechnol 7. Ishikawa T, Zhu B-L, Maeda H (2006) Effect
9:42. doi:10.1186/1472-6750-9-42 of sodium azide on the metabolic activity of
4. Pichler J, Hesse F, Wieser M et al (2009) A cultured fetal cells. Toxicol Ind Health 22:337–
study on the temperature dependency and time 341. doi:10.1177/0748233706071737
 Chapter 10

The ‘Omics Revolution in CHO Biology: Roadmap

to Improved CHO Productivity
Hussain Dahodwala and Susan T. Sharfstein

Abstract
Increased understanding of Chinese hamster ovary (CHO) cell physiology has been ushered in upon availability
of the parental CHO-K1 cell line genome. Free and openly accessible sequence information has comple-
mented transcriptomic and proteomic studies. The previous decade has also seen an increase in sensitivity
and accuracy of proteomic methods due to technology development. In this genomic era, high-­throughput
screening methods, sophisticated informatic tools, and models continually drive major innovations in cell
line development and process engineering. This review describes the various achievements in ‘omics
techniques and their application to improve recombinant protein expression from CHO cell lines.

Key words CHO cell engineering, CHO genome, Proteomics, Transcriptomics, CHO bioinformatics,
Mass spectrometry

1  Introduction

Chinese hamster ovary (CHO) cells are the preferred hosts for
biotherapeutic manufacturing because of the cells’ robust nature,
adaptability to suspension, growth in serum-free media, and ability
to perform human-like posttranslation modifications of recombi-
nant proteins [1, 2]. Since the approval of the first CHO recombi-
nant protein, tissue plasminogen activator (tPA) [3], more than
160 recombinant products have been successfully expressed in
CHO cells [4]. In addition to monoclonal antibodies and hor-
mones, diverse therapeutic molecules such as heparin [5] have
been successfully expressed in CHO cells. The current market
share of biopharmaceuticals, an estimated $160 billion annually
worldwide, has far exceeded previous predictions [1]. Keeping up
with the rise in demand, volumetric productivity from CHO cells
has risen from 0.05 g/L to >10 g/L in the past 30 years [6, 7].
However, these successes are mostly driven by labor-intensive and
time-consuming empirical processes [8, 9], which are case specific
and not easily reproduced for a new campaign process. In order to

Paula Meleady (ed.), Heterologous Protein Production in CHO Cells: Methods and Protocols, Methods in Molecular Biology,
vol. 1603, DOI 10.1007/978-1-4939-6972-2_10, © Springer Science+Business Media LLC 2017

153
154 Hussain Dahodwala and Susan T. Sharfstein

consistently enhance recombinant protein production efficiency in

CHO hosts, a holistic understanding of cellular physiology is
essential. A cell’s final phenotype is influenced by its environment
but is determined by the sum total of all the myriad interactions of
the cellular components. The term “'Omics” refers to the large-­
scale functional and relationship characterization of various bio-
logical molecules in a defined system. In the field of biology, the
three most prominent ‘omics disciplines are:
1. Genomics: the set of genetic information encompassed in the
DNA sequence.
2. Transcriptomics: the total RNA produced in the cells and its
impact on cell physiology.
3. Proteomics: the entire protein composition of cells. This may
include information about protein location, function, interac-
tion, and relationship with other cellular macromolecules.
For a rational inquiry into cellular physiology and application of
a systems-level approach for improving cellular performance, the
first step is to compile all the available ‘omics data. Publication of
the CHO-K1 genome [10] has rapidly accelerated data-driven
‘omics research on CHO cells. According to PubMed statistics,
more than 1000 studies involving CHO ‘omics were reported since
the publication of the CHO genome. In stark contrast, pre-­2009,
only 173 such studies are available [11]. Despite being newly estab-
lished, ‘omics technologies have been rapidly implemented in
biotechnology to identify the transcriptomic and proteomic com-
positions of higher producing cells [11–15]. The use of ‘omics
technologies will continue to gain prevalence among academic and
industrial mammalian cell research, resulting in more detailed data-
sets and further insights into cellular engineering avenues (Fig. 1)
[16]. Summaries of various genomic techniques [17–21], transcrip-
tomic approaches [22–27], and proteomic studies [28–33] that
have paved the way for CHO systems biology are listed in Table 1.

2  CHO ‘Omics

2.1  CHO Genomics The public availability of the CHO genome [10] has been a great
boon to the scientific and industrial community alike. Like the
human genome endeavor, the effort to complete the CHO genome
was the end result of academic and industrial collaboration on an
international scale. This initiative has enabled a range of systems
biology research [34]. Following the availability of CHO-K1
genome, the Chinese hamster genome was also sequenced [35],
creating a universal reference genome for all CHO cells. Most cell
lines used for recombinant protein expression make use of the
mutagenized dhfr-deficient CHO DG44, suspension adapted
CHO-S, or CHO-DXB11 cell lines. Though most of these cells
 CHOmics for Improved Productivity 155

Enhanced
cell growth

Increase in specific
Genomic data productivity

Reduced cell
apoptosis and
necrosis
CHO cells

Transcriptomic Reduced
data cell stress

Improved
product quality

Proteomic data

Fig. 1 Multiple ‘omics data generated for CHO cells can be used to guide cell line and product quality
attributes

Table 1
Representative ‘omics technologies used to generate CHO-specific datasets

Technique Summary of experiment Ref

Genomic Generation of CHO genome scaffold by comparative genomics approach [17]
Genomic First publicly available draft sequence of the CHO-K1 genome. [10]
Genomic Six draft genomes of CHO cell lines derived from CHO-K1, DG44, and [18]
CHO-S lineages. Genomic landscape of CHO revealed. Focuses on
glycosylation genes.
Genomic Genome sequence of the CHO-DXB11 genome sequenced to a depth [19]
of 33X
Genomic The DNA methylation landscape of CHO DP-12 cells [20]
Genomic Chromosome rearrangements between CHO DG44, CHO-K1, and [21]
primary Chinese hamster cells
Transcriptomic First measurement of CHO translation efficiency in mAb-producing cell [22]
line.
Transcriptomic Comparative transcriptome analysis to unveil genes affecting recombinant [14]
protein productivity in mammalian cells
Transcriptomic Profile glycosylation gene expression in CHO cell lines [23, 24]
Transcriptomic Analysis of CHO cells under different processes and impact on titer [25]
(continued)
156 Hussain Dahodwala and Susan T. Sharfstein

Table 1
(continued)

Technique Summary of experiment Ref

Transcriptomic Transcriptome analysis of various progeny cell lines derived from same [26]
parental clone, to gain insight into molecular changes during cell line
selection
Transcriptomic Analysis of microarray data from 295 cell samples to determine [27]
transcriptional mediators of growth and productivity
Proteomic Proteomic analysis of CHO cells with varying lactate profiles [28]
Proteomic Proteomic analysis of 17 high-producing CHO cells to identify universal [29]
markers indicative of productivity
Proteomic Impact of miR7 overexpression on CHO cell protein expression to [30]
understand potential mechanism of increases in productivity
Proteomic Proteomic profiling of high-producing cells in culture [31]
Proteomic First proteomic study performed using the CHO genome, identifying [32]
6164 proteins in CHO-K1 cells
Proteomic Identified secreted proteins in DG44 and CHO-S cell lines [33]

are derived from a common hamster parental cell line, m ­ utagenesis,

adaptation to suspension, and the inherent chromosomal instabil-
ity of CHO lines make the genomes differ to an extent [21, 36].
Lewis, et al. have sequenced the DG44 and CHO-S cell lines and
made available the various genomes [18]. Pyrosequencing for
CHO-­DXB11 was also performed [37] and sequence deposited in
the NCBI database. Further studies were performed to enable
accurate transcriptome profiling [12]. Current database query
returns a transcriptome count of 29,392 genes that represent the
complete CHO genome coverage http://www.ncbi.nlm.nih.gov/
gene/?term=%22Cricetulus+griseus%22[porgn%3A__txid10029]
(Accessed May 2, 2016). The database availability has permitted
studies of productivity using next generation transcriptomics (i.e.,
RNA seq) [38] and specially designed PCR arrays [39]. The recent
availability of genome-editing techniques provides new opportuni-
ties for cell line modification [40] that can be fully exploited with
access to a mapped genome and availability of the transcriptome.
Application of these tools can aid rational design for genome engi-
neering and permit investigation of site-directed integration into
genome hot spots [41].

2.2  CHO Sequencing the CHO-K1 genome has complemented and acceler-
Transcriptome ated transcriptome elucidation. Previously, the use of expressed
sequence tags (ESTs) [42] and proprietary microarrays [43]
limited the transcriptome search to a few hundred genes. With the
 CHOmics for Improved Productivity 157

publication of the CHO genome, it is now possible to design

microarrays that will simultaneously investigate expression of
>11,000 genes from a given sample [15]. To further assist in gene
and pathway regulation within CHO cell lines, PCR arrays devel-
oped from transcriptome analysis are available that cover CHO-­
specific pathways to investigate the role of a set of genes in a given
cell line [39]. Comparative transcriptomics is a powerful tool that
gives an insight into changes in expression patterns. CHO-cell
expression varies greatly between clones, even those derived from
the same parental cell line. mRNA levels in CHO cells having dif-
ferential recombinant protein expression from two to greater than
forty picograms per cell per day were analyzed [14]. It was hypoth-
esized that low productivity cell lines would exhibit a different
metabolic profile than their high productivity counterparts. Insight
into mRNA levels confirmed that expression of genes linked to
energy production, cellular oxidation, and glycolysis is elevated in
the high-producing cell lines.
The power of microarray profiling is in its application as a
high-throughput investigative tool. Clarke et al. performed an
unsupervised, weighted gene co-expression network analysis of
proprietary microarrays profiling the transcriptome of 295 samples
from different cell lines under different process conditions [27].
The influence of culture condition and inherent cell line differ-
ences gave rise to different clusters of upregulated and downregu-
lated genes. Similarly, a panel of cell lines, all arising from the same
parental clone, were subjected to comparative transcriptome pro-
filing [26], providing insight into the changing transcriptome as a
result of adaptation and selection. In the rapidly growing clones,
cell-cycle regulatory genes were overexpressed. Interestingly,
miRNA regulation was implicated in the “fast growing” genes,
suggesting new possibilities for control of cellular regulation.
Extensive bioinformatics investigation and subsequent gene engi-
neering efforts will help identify gene clusters most relevant to higher
productivity phenotypes. Such experiments will help narrow the
desirable gene profiles and choices of platform cell lines. It is expected
that these findings will be incorporated in the bioprocess workflow
by employing cellular and process engineering approaches.

2.3  CHO Proteome The CHO cell genome remains fairly constant throughout the
stages of growth and in response to changes in bioprocess condi-
tions, but the proteome is in constant flux. The CHO genome
[44] is predicted to have 24,383 genes [10], but these can be
translated into many times that number of proteins due to isoforms
and alternate splicing, as well as the inclusion of posttranslational
modifications (PTM) [45]. Therefore, investigation of the pro-
teome has the potential to offer complementary and more dynamic
knowledge about factors that will influence higher productivity
phenotype in cells. Despite the lack of genomic information and
158 Hussain Dahodwala and Susan T. Sharfstein

the inherent technical challenges in proteome studies, protein

expression and function is paramount to a cell function. Efforts to
characterize CHO proteome have long preceded the availability of
the genome [46]. Post-genomics and the availability of the “blue
print” of all possible genetic products, we can design experiments
to investigate hypothetical proteins or changes to protein sequence
compared with the expected sequence from the genome. Availability
of sequence information allows more accurate peptide identifica-
tion and functional prediction within a given sample. Recent
improvements in protein identification technologies, namely
advances in mass spectrometry, labeling techniques, software, and
instrumentation [42, 47], have enabled investigative and compara-
tive analysis of proteins present in CHO cells. Monitoring the pro-
teome can give an insight into the thousands of proteins that play
key roles in governing CHO cell protein production. Identification
of the proteome will enable users to better understand and exploit
changes in host cell productivity as well as predict its capabilities
for performing specific protein-processing functions, such as gly-
cosylation, that will determine the protein quality [48, 49].
With the availability of genomic datasets, many proteomic
efforts have been streamlined due to ease of target identification
and improved mapping tools. A few prominent efforts are listed in
Table 2. In addition, select examples that best utilize advances in

Table 2
Methods employed for CHO proteomic studies

Application to CHO
Technique proteome Ref Major advantage Disadvantage
2D gel Proteomic expression [58] Ease of visualization of Labor intensive,
electro-­ profiling in various [59] comparative highly semiquantitative
phoresis CHO cells expressed samples
iTRAQ Protein profiling in [12] Stains after cell lysis and is Longer MS run needed
high-producing cells not limited by
incorporation of stable
isotopes
SILAC Studying cellular [64] Greater proteome coverage, Costs more compared to
secretory capacity of captures extracellular, chemical labeling
IgG-producing cells and LMW proteins
MALDI-­ Identification of O- and [68] Small sample size, precise Requires very rigid sample
TOF N-glycans in molecular weight preparation. Low run
glycosylation mutants to run reproducibility
ESI Detailed structural [70] High accuracy, fast, couples Sensitive to salts, complex
assignments of IgG with LC spectra
produced in CHO cells
SELDI-­ Characterization of CHO Soft ionization technique Low mass resolution
TOF secretome amenable to studying
low molecular weight
proteins and PTMs
 CHOmics for Improved Productivity 159

techniques, complement the available datasets, and contribute to

bioprocess optimization are briefly discussed below.
Two-dimensional (2D) gel electrophoresis has been a mainstay
in protein identification. Comparative 2D electrophoresis helps
visualize protein intensity and can quickly ascertain the absence or
presence of spots. Software developments have increased accuracy
in protein quantitation. In one such study, proteomic analysis using
2D gel electrophoresis and mass spectrometry was performed on
two monoclonal antibody-expressing cell lines [50]. Using 2D gel
identification software, differentially expressed proteins between
the two cell lines were identified. Mass spectrometry (MS) analysis
identified a number of differentially expressed spots on 2D gels,
including several belonging to the protein chaperone family.
Moreover, 2D gel electrophoresis data can serve as a database for
spot identification. Advances in in-gel digestion methods and
­subsequent label-free liquid chromatography-mass spectrometry
(LC-­ MS) techniques have increased the depth of protein data
available. Using advances in LC-MS, high and low productivity
CHO-GS cell lines were analyzed for differential proteomic expres-
sion [51]. Twelve proteins belonging to the families of cytoskele-
ton rearrangement, protein synthesis, cell metabolism, and cell
growth were identified for further investigation. In yet another
application using proteomics and transcriptome data, two biophar-
maceutical production processes were compared [52]. Data analy-
sis and pathway interpretation demonstrated that the lipid
biosynthesis pathway was affected by changes in batch conditions.
This effort helped optimize batch media by identifying a need for
lipid supplementation. Supplemented media led to a 20% increase
in productivity. In all cases described above, integration of advanced
informatics algorithms improved identification and genome assign-
ment, thereby increasing the information from proteomic efforts.
No doubt increased data availability will help identify “universal
markers” for higher productivity cell phenotypes. Integration of
proteomics and transcriptomics can further be applied to achieve
bioprocess optimization.
A recent study explored in-depth the secreted proteome of
CHO cells. This study combined quantitative LC-MS results with
gene ontology (GO) and protein sequence analyses using multiple
bioinformatics tools to elucidate a data set characteristic of the sec-
retome from CHO cells [53]. It has been reported that CHO host
cell proteins (HCP) co-purify with therapeutic proteins and HCP
impurities change over the life of a batch culture [54]. Qualitative
and quantitative information regarding the secreted proteins in
CHO cell culture will greatly aid downstream bioprocessing.
Identified proteins can be screened to eliminate chances of con-
tamination, antigenicity, or aggregation of desired product. Further
cellular engineering might help elimination of undesirable host cell
proteome.
160 Hussain Dahodwala and Susan T. Sharfstein

3  Advances in Proteomic Techniques

Genome sequence availability has revolutionized approaches to

research and development in all of the life sciences, particularly in
drug discovery and pharmaceutical-based research. However, pro-
teomics still remains a difficult quest. Protein identification remains
challenging due to the complexity of the samples and the wide
dynamic range of protein concentrations [55, 56]. However,
improvements in protein labeling and fractionation technologies
have led to improved protein identification, including very low
abundance proteins [57]. Here, we will discuss recent advances in
proteomic techniques and applications of these techniques to
understanding the CHO proteome.
From the mid-1970s, proteomics was pursued with 2D gel elec-
trophoresis. These gel electrophoresis-based analyses have contrib-
uted substantially to understanding CHO proteomics [58, 59] but
still have many limitations. For analysis of hydrophobic proteins, very
low molecular-­weight proteins, and most importantly, for larger data-
sets, biological MS has risen as the technological basis of most current
proteomics studies. MS has come into prominence with the develop-
ment of the electrospray and soft ionization techniques such as matrix-
assisted laser desorption/ionization (MALDI). These advances made
biological molecules readily amenable to mass spectrometry and
garnered the Chemistry Nobel Prize in 2002. A number of break-
throughs followed that allowed smaller sample size, direct analysis
of labeled and unlabeled proteins, and the ability to map spectral
peaks to peptide fragments in databases that would allow for rapid
detection [60, 61]. Most recently, high-­performance robust MS
instruments have further increased the power of MS-based pro-
teomics. Methods for labeling samples prior to mass spectrometry
analysis have allowed rapid and simpler sample analysis.
Stable isotope labeling by amino acids in cell culture (SILAC)
is a simple and straightforward process for labeling proteins in liv-
ing cells. Cells are cultivated in media that has a “light” or “heavy”
form of the amino acid (e.g., 13C, 15 N). As cells metabolize the
media, they incorporate the labeled substrate into the proteins.
These labeled proteins are amenable to analysis by MS-based meth-
ods. Many groups have successfully implemented SILAC to com-
pare the levels of thousands of proteins in different cellular states
[55, 62]. With the availability of newer bioinformatics approaches,
many different cell samples can be rapidly analyzed for identifying
protein feature differences [63]. SILAC has been used to demon-
strate the IgG secretion kinetics in CHO cells [64].
Another labeling technique that has increased the application
of proteomic analysis is isobaric tagging for relative and absolute
quantification (iTRAQ). In this method, samples are labeled post
 CHOmics for Improved Productivity 161

lysis, which reduces the dependence on stable isotope incorporation

within cells [65]. Since multiple iTRAQ reagents having different
reporter group masses exist, proteins from several samples can be
analyzed simultaneously [66]. iTRAQ is a powerful technique to
monitor relative changes in protein and PTM abundance across
perturbed biological systems [67].
Use of iTRAQ has been successfully demonstrated in proteome
identification in CHO cells. Up to 80 proteins were identified to have
been differentially regulated in higher producing cell lines [12].
Electrospray ionization (ESI) is a technique used in mass spec-
trometry to produce ions using an electrospray in which a high
voltage is applied to a liquid to create an aerosol. Important pro-
tein quality attributes like glycosylation are determined by the host
cell genome [68] as well as bioprocessing conditions. From a regu-
latory standpoint, these attributes must be demonstrated. ESI is
one of the powerful methods that can exhibit accurate, sensitive
data on quality attributes of proteins [69]. Using ESI LC-MS, cell-­
line-­dependent differences in recombinant proteins were analyzed
[70]. Another ionization technique, surface-enhanced laser desorp-
tion/ionization time-of-flight (SELDI-TOF) mass spectrometry,
has recently come into prominence [71, 72]. SELDI-TOF has
been described as a combination of chromatography and mass
spectrometry. In this method, proteins of interest are deposited
onto a chip; then lasers are used to excite the sample ions, which
can then be analyzed by MS [73]. Use of soft ionization makes the
sample preparation gentler, and thereby, the technique is more
amenable to study smaller peptides, lower molecular weight pro-
teins, and delicate posttranslational modifications like phosphory-
lation. SELDI-TOF MS has been used to generate a unique dataset
of extracellular secreted CHO proteins [74]. A recent study in
CHO cells identified 24 differentially expressed proteins in the
very low molecular-weight range, thereby presenting a unique
method to understand the HCP in CHO by utilizing extremely
small samples [75]. This technique could potentially open identify
never before seen targets in the search for biomarkers for elevated
productivity phenotypes.
Labeling techniques have aided in simpler experimental design
and easier identification of unknown targets. As advances in pro-
teomics continue, a new era of label-free proteomic analysis has been
initiated [76]. Employing sophisticated algorithms and normaliza-
tion methods, this technique has been employed to identify targets of
miRNA 23 in CHO cells [77]. These identified targets demonstrate
a route to increase productivity without affecting growth. Such newer
methods complement the more classical approaches using 2-D gel
electrophoresis that have contributed immensely to understanding
the CHO proteome.
162 Hussain Dahodwala and Susan T. Sharfstein

Table 3
List of resources for ‘omics research

Resource Function Reference

http://chogenome.org/ Online database to collect, curate, and distribute [78]
http://www.chogenome.org/ CHO genomic information
proteome_new.php
http://www.cgcdb.org/ CHO gene co-expression database (CGCDB) [86]
http://cho-epigenome.boku.ac.at/ CHO epigenome [80]
ftp://ftp.cebitec.uni-bielefeld.de/ Bielefeld-BOKU-CHO cell protein database [79]
pub/supplements/BB-CHO/
http://web.cos.gmu. KEGG Pathway painter: Generates pathway maps [87]
edu/~gmanyam/kegg/examples. from microarray and RT-PCR data for visualizing
html co-expressed genes on metabolic pathway maps
http://omics.pnl.gov/software/ Online software for proteomics and metabolic data [82]
multialign analysis
Mathematical models Cell line model that predicts productivity from [83]
CHO-specific gene expression

4  ‘Omics Databases

The purpose of a database for CHO ‘omics information is primarily

to serve as a central depository for all the collected information [15].
Table  3 lists many CHO-specific databases that serve as primary
repositories. Genomic information from all the sequencing efforts
can be found at the Chinese hamster genome database (choge-
nome.org) [78]. This website serves as both a repository for the
sequence information as well as a curated public resource for
accessing information in a searchable format. The website is search-
able via accession number, gene name, GO term, or symbol.
Currently, it hosts the CHO-K1 and hamster genomes. In ­addition,
there is the complete nucleotide sequence of CHO-K1 mitochon-
drial DNA. There is also a section of the website devoted to the
CHO proteome (http://www.chogenome.org/proteome_new.
php) where one can find the CHO 2D–PAGE reference gels and
proteome data from high-throughput proteomic analysis of CHO
cells [32]. In addition to this proteomic resource, other initiatives
will carry out further curation of genomic data. The CHO gene
co-expression database (CGCDB) is an online system that collects
and correlates gene cluster expression, building upon a sample set
of 295 CHO expression microarrays. It is a searchable database
where users can browse for co-expressed genes for investigating
relationships to genes of interest. Users can also submit microarray
data to be represented in the database. The database aims to
 CHOmics for Improved Productivity 163

provide access to gene co-expression patterns derived from the

microarray analysis of transcript expression during industrial CHO
cell culture. Another resource aims to expand and improve the ref-
erence databases available for CHO proteomic identification [79].
The latest addition to the ‘omics dataset is the CHO epigenome
[80] that provides for a comprehensive genome and epigenome
characterization of CHO cells in response to evolutionary pressure
and over time.
Many different tools exist to aid in data interpretation and
visualization of information generated from genomics, transcrip-
tomics and proteomics (particularly MS data). Most of these are
exhaustively described elsewhere [81]. MultiAlign is one such
application that is used to analyze LC-MS proteomic and metabo-
lomic datasets [82]. It provides advanced visualization and manip-
ulation capabilities for LC-MS datasets acquired on high-resolution
mass spectrometers. MultiAlign allows researchers to visualize
overlaid 2D plots, alignment plots, normalizations, and basic
statistical comparisons for the ease of informational processing.
The ultimate goal of gathering all biological data from CHO
cells is to gain insight into the cellular mechanisms, thereby offering
researchers insights into molecular mechanisms of higher productiv-
ity phenotypes when performing cellular engineering research.
Toward this end, models have been developed to correlate gene
expression with productivity [83–85]. In one study, a transcriptomic
data set from 70 stationary-phase, temperature-­shifted CHO pro-
duction cell line samples was collected, and their specific productiv-
ity was used to guide the algorithm to learn the relationship between
the gene expression profile and productivity [83]. The algorithm
was successful in predicting specific productivity (qP) of ten addi-
tional samples after analyzing their microarray profiles.
Collectively, these computational systems and tools provide
the foundation for efficient data analysis and annotation. Useful
databases have been developed for CHO genome and transcrip-
tome data representation [86, 87], and we are now seeing evolu-
tion of other ‘omics datasets resulting from higher quality and
quantity of proteomics data. As bioinformatics technologies are
improved, so will the abilities to interpret data. The end goals of all
these endeavors will culminate in predictive algorithms that will
lead to rational bioprocess development

5  Conclusions and Future Directions

CHO cell lines are currently the leading recombinant, therapeutic

protein production system and will likely continue in this role for
the foreseeable future. In the past 30 years, vast improvements in
cell titers, productivity, and product quality have been achieved
through various cell line development and bioprocess strategies.
164 Hussain Dahodwala and Susan T. Sharfstein

Fig. 2 Genome, transcriptome, and proteome mapping of CHO cell lines. Genome, transcriptome, and proteome
of CHO cell lines are mapped to all metabolic pathways in the Kyoto Encyclopedia of Genes and Genomes (KEGG).
The genes found in the CHO genomic level are represented by yellow. Orange represents the genes expressed at
the mRNA level for exponentially grown in CHO-K1, whereas green shows the genes expressed at the proteome
level. Genes identified as expressed at both proteome and mRNA levels are represented in gray

The initiation of ‘omics data collection and distribution has

improved our understanding of the cell lines and led to further
developments in recombinant protein production. The availability
of a CHO genome sequence has complemented and advanced
transcriptomic and proteomic research. There have been volumes
of data generated in the post-genomic age, evaluating CHO cell
behavior in various conditions and with varying phenotypes.
The information gained has, in turn, served to improve produc-
tion strategies. Although the majority of the bioprocess industry
still relies on empirical development techniques, ‘omics technolo-
gies have been increasingly incorporated into bioprocess develop-
ment. Advances in systems biology may speed rational design of
CHO cell lines to improve desirable bioprocess attributes.
Integration of genomic, transcriptomic, proteomic, and metabolo-
mic techniques (see Fig. 2) may uncover the fundamental intrica-
cies in CHO cell biology, revealing future targets for genome-editing
strategies. This systems biology approach may provide clues for the
creation of improved host cell lines and may enhance CHO pro-
duction capabilities by revealing rational strategies for cell line and
process development.

References
1. Walsh G (2014) Biopharmaceutical bench-
marks 2014. Nat Biotechnol 32:992–1000.
doi:10.1038/nbt0910-917
2. Kim JY, Kim Y-GG, Lee GM (2012) CHO
cells in biotechnology for production of recom-
binant proteins: current state and further
potential. Appl Microbiol Biotechnol 93:917–
930. doi:10.1007/s00253-011-3758-5
3. Kaufman RJ, Wasley LC, Spiliotes AJ et al
(1985) Coamplification and coexpression of
human tissue-type plasminogen activator and
murine dihydrofolate reductase sequences in
Chinese hamster ovary cells. Mol Cell Biol
5:1750–1759. doi:10.1128/MCB.5.7.1750.
Updated
4. Butler M, Meneses-Acosta A (2012) Recent
advances in technology supporting biopharma-
ceutical production from mammalian cells.
Appl Microbiol Biotechnol 96:885–894.
doi:10.1007/s00253-012-4451-z
5. Baik JY, Dahodwala H, Oduah E et al (2015)
Optimization of bioprocess conditions
improves production of a CHO cell-derived,
bioengineered heparin. Biotechnol J 10:1067–
1081. doi:10.1002/biot.201400665
6. Datta P, Linhardt RJ, Sharfstein ST (2013) An
omics approach towards CHO cell engineer-
ing. Biotechnol Bioeng 110:1255–1271.
doi:10.1002/bit.24841
7. Dietmair S, Nielsen LK, Timmins NE (2012)
Mammalian cells as biopharmaceutical produc-
tion hosts in the age of omics. Biotechnol
J 7:75–89. doi:10.1002/biot.201100369
8. Hacker DL, De Jesus M, Wurm FM (2009)
25 Years of recombinant proteins from reactor-
grown cells - where do we go from here?
Biotechnol Adv 27:1023–1027. doi:10.1016/j.
biotechadv.2009.05.008
9. Matasci M, Hacker DL, Baldi L, Wurm FM
(2008) Recombinant therapeutic protein pro-
duction in cultivated mammalian cells: current
status and future prospects. Drug Discov
Today Technol 5:e37–e42. doi:10.1016/j.
ddtec.2008.12.003
10. Xu X, Nagarajan H, Lewis NE et al (2011) The
genomic sequence of the Chinese hamster
ovary (CHO)-K1 cell line. Nat Biotechnol
29:735–741. doi:10.1038/nbt.1932
11. Dietmair S, Hodson MP, Quek L-E et al (2012)
Metabolite profiling of CHO cells with different
growth characteristics. Biotechnol Bioeng
109:1404–1414. doi:10.1002/bit.24496
12. Nissom PM, Sanny A, Kok YJ et al (2006)
Transcriptome and proteome profiling to
understanding the biology of high productivity
CHOmics for Improved Productivity 165
CHO cells. Mol Biotechnol 34:125–140.
doi:10.1385/MB:34:2:125
13. Vishwanathan N, Le H, Jacob NM et al (2014)
Transcriptome dynamics of transgene amplifica-
tion in Chinese hamster ovary cells. Biotechnol
Bioeng 111:518–528. doi:10.1002/bit.25117
14. Yee JC, Gerdtzen ZP, Hu WS (2009)
Comparative transcriptome analysis to unveil
genes affecting recombinant protein produc-
tivity in mammalian cells. Biotechnol Bioeng
102:246–263. doi:10.1002/bit.22039
15. Wuest DM, Harcum SW, Lee KH (2012)
Genomics in mammalian cell culture biopro-
cessing. Biotechnol Adv 30:629–638.
doi:10.1016/j.biotechadv.2011.10.010
16. Jayapal KP, Wlaschin KF, Hu W-SH, Yap MGS
(2007) Recombinant protein therapeutics
from cho cells - 20 years and counting. CEP
Mag:40–47
17. Wlaschin KF, Hu WS (2007) A scaffold for the
Chinese hamster genome. Biotechnol Bioeng
98:429–439. doi:10.1002/bit.21430
18. Lewis NE, Liu X, Li Y et al (2013) Genomic
landscapes of Chinese hamster ovary cell lines
as revealed by the Cricetulus griseus draft
genome. Nat Biotechnol 31:759–767.
doi:10.1038/nbt.2624
19. Kaas CS, Kristensen C, Betenbaugh MJ,
Andersen MR (2015) Sequencing the CHO
DXB11 genome reveals regional variations in
genomic stability and haploidy. BMC Genomics
16:160. doi:10.1186/s12864-015-1391-x
20. Wippermann A, Rupp O, Brinkrolf K et al
(2015) The DNA methylation landscape of
Chinese hamster ovary (CHO) DP-12 cells.
J Biotechnol 199:38–46. doi:10.1016/j.
jbiotec.2015.02.014
21. Cao Y, Kimura S, Itoi T et al (2012) Construction
of BAC-based physical map and analysis of chro-
mosome rearrangement in Chinese hamster
ovary cell lines. Biotechnol Bioeng 109:1357–
1367. doi:10.1002/bit.24347
22. Courtes FC, Lin J, Lim HL et al (2013)
Translatome analysis of CHO cells to identify
key growth genes. J Biotechnol 167:215–224.
doi:10.1016/j.jbiotec.2013.07.010
23. Könitzer JD, Müller MM, Leparc G et al
(2015) A global RNA-seq-driven analysis of
CHO host and production cell lines reveals dis-
tinct differential expression patterns of genes
contributing to recombinant antibody glyco-
sylation. Biotechnol J 10:1412–1423.
doi:10.1002/biot.201400652
24. Wong DCF, Wong NSC, Goh JSY et al (2010)
Profiling of N-glycosylation gene expression in
CHO cell fed-batch cultures. Biotechnol
Bioeng 107:516–528. doi:10.1002/bit.22828
166 Hussain Dahodwala and Susan T. Sharfstein
25. Schaub J, Clemens C, Kaufmann H, Schulz
TW (2012) Advancing biopharmaceutical
process development by system-level data
analysis and integration of omics data. Adv
Biochem Eng Biotechnol 127:133–163.
doi:10.1007/10_2010_98
26. Doolan P, Clarke C, Kinsella P et al (2013)
Transcriptomic analysis of clonal growth rate
variation during CHO cell line development.
J Biotechnol 166:105–113. doi:10.1016/j.
jbiotec.2013.04.014
27. Clarke C, Doolan P, Barron N et al (2011)
Large scale microarray profiling and coexpres-
sion network analysis of CHO cells identifies
transcriptional modules associated with growth
and productivity. J Biotechnol 155:350–359.
doi:10.1016/j.jbiotec.2011.07.011
28. Luo J, Vijayasankaran N, Autsen J et al (2012)
Comparative metabolite analysis to understand
lactate metabolism shift in Chinese hamster
ovary cell culture process. Biotechnol Bioeng
109:146–156. doi:10.1002/bit.23291
29. Kang S, Ren D, Xiao G et al (2014) Cell line
profiling to improve monoclonal antibody pro-
duction. Biotechnol Bioeng 111:748–760.
doi:10.1002/bit.25141
30. Meleady P, Gallagher M, Clarke C et al (2012)
Impact of miR-7 over-expression on the pro-
teome of Chinese hamster ovary cells.
J Biotechnol 160:251–262. doi:10.1016/j.
jbiotec.2012.03.002
31. Carlage T, Hincapie M, Zang L et al (2009)
Proteomic profiling of a high-producing
Chinese hamster ovary cell culture. Anal Chem
81:7357–7362. doi:10.1021/ac900792z
32. Baycin-Hizal D, Tabb DL, Chaerkady R et al
(2012) Proteomic analysis of Chinese hamster
ovary cells. J Proteome Res 11:5265–5276.
doi:10.1021/pr300476w
33. Slade PG, Hajivandi M, Bartel CM, Gorfien SF
(2012) Identifying the CHO secretome using
mucin-type O-linked glycosylation and click-­
chemistry. J Proteome Res 11:6175–6186.
doi:10.1021/pr300810f
34. Brown AJ, James DC (2015) Precision control
of recombinant gene transcription for CHO
cell synthetic biology. Biotechnol Adv.
doi:10.1016/j.biotechadv.2015.12.012
35. Brinkrolf K, Rupp O, Laux H et al (2013)
Chinese hamster genome sequenced from
sorted chromosomes. Nat Biotechnol 31:694–
695. doi:10.1038/nbt.2645
36. Kim M, O’Callaghan PM, Droms KA, James
DC (2011) A mechanistic understanding of
production instability in CHO cell lines
expressing recombinant monoclonal antibod-
ies. Biotechnol Bioeng 108:2434–2446.
doi:10.1002/bit.23189
37. Becker J, Hackl M, Rupp O et al (2011)
Unraveling the Chinese hamster ovary cell line
transcriptome by next-generation sequencing.
J Biotechnol 156:227–235. doi:10.1016/j.
jbiotec.2011.09.014
38. Martin JA, Wang Z (2011) Next-generation
transcriptome assembly. Nat Rev Genet
12:671–682. doi:10.1038/nrg3068
39. Edros R, McDonnell S, Al-Rubeai M (2014)
The relationship between mTOR signalling
pathway and recombinant antibody productiv-
ity in CHO cell lines. BMC Biotechnol 14:15.
doi:10.1186/1472-6750-14-15
40. Dahodwala H, Sharfstein ST (2014) Role of
epigenetics in expression of recombinant pro-
teins from mammalian cells. Pharm Bioprocess
2:403–419. doi:10.4155/pbp.14.47
41. Ronda C, Pedersen LE, Hansen HG et al
(2014) Accelerating genome editing in CHO
cells using CRISPR Cas9 and CRISPy, a web-­
based target finding tool. Biotechnol Bioeng
111:1604–1616. doi:10.1002/bit.25233
42. Wei Y-YC, Naderi S, Meshram M et al (2011)
Proteomics analysis of Chinese hamster ovary
cells undergoing apoptosis during prolonged
cultivation. Cytotechnology 63:663–677.
doi:10.1007/s10616-011-9385-2
43. Shen D, Kiehl TR, Khattak SF et al (2009)
Transcriptomic responses to sodium chloride-­
induced osmotic stress: a study of industrial
fed-batch CHO cell cultures. Biotechnol Prog
26:1104–1115
44. Hammond S, Kaplarevic M, Borth N et al
(2012) Chinese hamster genome database: an
online resource for the CHO community at.
Biotechnol Bioeng 109:1353–1356.
doi:10.1002/bit.24374
45. Kim M-S, Pinto SM, Getnet D et al (2014) A
draft map of the human proteome. Nature
509:575–581. doi:10.1038/nature13302
46. Lee JS, Park HJ, Kim YH, Lee GM (2010)
Protein reference mapping of dihydrofolate
reductase-deficient CHO DG44 cell lines using
2-dimensional electrophoresis. Proteomics
10:2292–2302. doi:10.1002/pmic.200900430
47. Meleady P, Doolan P, Henry M et al (2011)
Sustained productivity in recombinant Chinese
hamster ovary (CHO) cell lines: proteome
analysis of the molecular basis for a process-­
related phenotype. BMC Biotechnol 11:78
48. Hossler P, Khattak SF, Li ZJ (2009) Optimal
and consistent protein glycosylation in mam-
malian cell culture. Glycobiology 19:936–949.
doi:10.1093/glycob/cwp079
 CHOmics for Improved Productivity 167

49. Lingg N, Zhang P, Song Z, Bardor M (2012) cellular structures. Nat Rev Mol Cell Biol
The sweet tooth of biopharmaceuticals: 6:702–714. doi:10.1038/nrm1711
Importance of recombinant protein glycosyl- 61. Tyers M, Mann M (2003) From genomics to
ation analysis. Biotechnol J 7:1462–1472. proteomics. Nature 422:193–197.
doi:10.1002/biot.201200078 doi:10.1038/nature01510
50. Pascoe DE, Arnott D, Papoutsakis ET et al 62. Graumann J, Hubner NC, Kim JB et al (2008)
(2007) Proteome analysis of antibody-­ Stable isotope labeling by amino acids in cell
producing CHO cell lines with different meta- culture (SILAC) and proteome quantitation of
bolic profiles. Biotechnol Bioeng 98:391–410. mouse embryonic stem cells to a depth of
doi:10.1002/bit.21460 5,111 proteins. Mol Cell Proteomics 7:672–
51. Dorai H (2013) Proteomic analysis of biore- 683. doi:10.1074/mcp.M700460-MCP200
actor cultures of an antibody expressing 63. Cravatt BF, Simon GM, Yates JR (2007) The
CHOGS cell line that promotes high produc- biological impact of mass-spectrometry-based
tivity. J Proteom Bioinform 06:99–108. proteomics. Nature 450:991–1000.
doi:10.4172/jpb.1000268 doi:10.1038/nature06525
52. Schaub J, Clemens C, Schorn P et al (2010) 64. Kantardjieff A, Jacob NM, Yee JC et al (2010)
CHO gene expression profiling in biopharma- Transcriptome and proteome analysis of Chinese
ceutical process analysis and design. Biotechnol hamster ovary cells under low temperature and
Bioeng 105:431–438. doi:10.1002/bit.22549 butyrate treatment. J Biotechnol 145:143–159.
53. Kumar A, Baycin-Hizal D, Wolozny D et al doi:10.1016/j.jbiotec.2009.09.008
(2015) Elucidation of the CHO super-ome 65. Aggarwal K, Choe LH, Lee KH (2006)
(CHO-SO) by proteoinformatics. J Proteome Shotgun proteomics using the iTRAQ isobaric
Res 14:4687–4703. doi:10.1021/acs. tags. Brief Funct Genom Proteom 5:112–120.
jproteome.5b00588 doi:10.1093/bfgp/ell018
54. Valente KN, Lenhoff AM, Lee KH (2015) 66. Sachon E, Mohammed S, Bache N, Jensen ON
Expression of difficult-to-remove host cell pro- (2006) Phosphopeptide quantitation using
tein impurities during extended Chinese ham- amine-reactive isobaric tagging reagents and
ster ovary cell culture and their impact on tandem mass spectrometry: application to pro-
continuous bioprocessing. Biotechnol Bioeng teins isolated by gel electrophoresis. Rapid
112:1232–1242. doi:10.1002/bit.25515 Commun Mass Spectrom 20:1127–1134.
55. Cox J, Mann M (2007) Is proteomics the new doi:10.1002/rcm.2427
genomics? Cell 130:395–398. doi:10.1016/j. 67. Ross PL, Huang YN, Marchese JN et al (2004)
cell.2007.07.032 Multiplexed protein quantitation in
56. Peleg Y, Unger T (2012) Chemical genomics Saccharomyces cerevisiae using amine-reactive
and proteomics. Methods Mol Biol 800:173– isobaric tagging reagents. Mol Cell Proteomics
186. doi:10.1007/978-1-61779-349-3 3:1154–1169. doi:10.1074/mcp.M400129-
57. Chandramouli K, Qian P-Y (2009) Proteomics: MCP200. M400129-MCP200 [pii]
challenges, techniques and possibilities to over- 68. North SJ, Huang HH, Sundaram S et al
come biological sample complexity. Hum (2010) Glycomics profiling of Chinese ham-
Genom Proteom 2009:22. ster ovary cell glycosylation mutants reveals
doi:10.4061/2009/239204 N-glycans of a novel size and complexity.
58. Doolan P, Meleady P, Barron N et al (2010) J Biol Chem 285:5759–5775. doi:10.1074/
Microarray and proteomics expression profil- jbc.M109.068353
ing identifies several candidates, including the 69. Ho CS, Lam CWK, Chan MHM et al (2003)
valosin-containing protein (VCP), involved in Electrospray ionisation mass spectrometry:
regulating high cellular growth rate in produc- principles and clinical applications. Clin
tion CHO cell lines. Biotechnol Bioeng Biochem 24:3–12
106:42–56. doi:10.1002/bit.22670 70. Stadlmann J, Pabst M, Kolarich D et al (2008)
59. Baik JY, Ha TK, Kim YH, Lee GM (2011) Analysis of immunoglobulin glycosylation by
Proteomic understanding of intracellular LC-ESI-MS of glycopeptides and oligosaccha-
responses of recombinant chinese hamster rides. Proteomics 8:2858–2871. doi:10.1002/
ovary cells adapted to grow in serum-free sus- pmic.200700968
pension culture. Biotechnol Prog 27:1680– 71. Tang N, Tornatore P, Weinberger SR (2004)
1688. doi:10.1002/btpr.685 Current developments in SELDI affinity tech-
60. Yates JR, Gilchrist A, Howell KE, Bergeron nology. Mass Spectrom Rev 23:34–44.
JJM (2005) Proteomics of organelles and large doi:10.1002/mas.10066
168 Hussain Dahodwala and Susan T. Sharfstein
72. De Bock M, De Seny D, Meuwis MA et al
(2010) Challenges for biomarker discovery in
body fluids using SELDI-TOF-MS. J Biomed
Biotechnol. doi:10.1155/2010/906082
73. Seibert V, Wiesner A, Buschmann T, Meuer
J (2004) Surface-enhanced laser desorption
ionization time-of-flight mass spectrometry
(SELDI TOF-MS) and ProteinChip® technol-
ogy in proteomics research. Pathol Res Pract
200:83–94. doi:10.1016/j.prp.2004.01.010
74. Kumar N, Gammell P, Meleady P et al (2008)
Differential protein expression following low
temperature culture of suspension CHO-K1
cells. BMC Biotechnol 8:42. doi:10.1186/
1472-6750-8-42
75. Tait AS, Hogwood CEM, Smales CM,
Bracewell DG (2012) Host cell protein dynam-
ics in the supernatant of a mAb producing
CHO cell line. Biotechnol Bioeng 109:971–
982. doi:10.1002/bit.24383
76. Bantscheff M, Schirle M, Sweetman G et al (2007)
Quantitative mass spectrometry in proteomics: a
critical review. Anal Bioanal Chem 389:1017–
1031. doi:10.1007/s00216-007-1486-6
77. Kelly PS, Breen L, Gallagher C et al (2015)
Re-programming CHO cell metabolism using
miR-23 tips the balance towards a highly pro-
ductive phenotype. Biotechnol J 10:1029–1040.
doi:10.1002/biot.201500101
78. Kremkow BG, Baik JY, MacDonald ML, Lee
KH (2015) CHOgenome.org 2.0: genome
resources and website updates. Biotechnol J
10:931–938. doi:10.1002/biot.201400646
79. Meleady P, Hoffrogge R, Henry M et al (2012)
Utilization and evaluation of CHO-specific
sequence databases for mass spectrometry
based proteomics. Biotechnol Bioeng
109:1386–1394. doi:10.1002/bit.24476
80. Feichtinger J, Hernández I, Fischer C et al
(2016) Comprehensive genome and epig-
enome characterization of CHO cells in
response to evolutionary pressures and over
time. Biotechnol Bioeng. 113(10):2241–
2253n/a–n/a. doi:10.1002/bit.25990
81. Lewis AM, Abu-Absi NR, Borys MC, Li ZJ
(2016) The use of ‘Omics technology to ratio-
nally improve industrial mammalian cell line
performance. Biotechnol Bioeng 113:26–38.
doi:10.1002/bit.25673
82. LaMarche BL, Crowell KL, Jaitly N et al (2013)
MultiAlign: a multiple LC-MS analysis tool for
targeted omics analysis. BMC Bioinform 14:49.
doi:10.1186/1471-2105-14-49
83. Clarke C, Doolan P, Barron N et al (2011)
Predicting cell-specific productivity from CHO
gene expression. J Biotechnol 151:159–165.
doi:10.1016/j.jbiotec.2010.11.016
84. Naderi S, Nikdel A, Meshram M et al (2014)
Modeling of cell culture damage and recovery
leads to increased antibody and biomass produc-
tivity in CHO cell cultures. Biotechnol J 9:
1152–1163. doi:10.1002/biot.201300287
85. Selvarasu S, Ho YS, Chong WPK et al (2012)
Combined in silico modeling and metabolo-
mics analysis to characterize fed-batch CHO
cell culture. Biotechnol Bioeng 109:1415–
1429. doi:10.1002/bit.24445
86. Clarke C, Doolan P, Barron N et al (2012)
CGCDB: a web-based resource for the investi-
gation of gene coexpression in CHO cell cul-
ture. Biotechnol Bioeng 109:1368–1370.
doi:10.1002/bit.24416
87. Manyam G, Birerdinc A, Baranova A (2015)
KPP: KEGG pathway Painter. BMC Syst Biol
9(Suppl 2):S3. doi:10.1186/1752-0509-
9-S2-S3
 Chapter 11

A Bioinformatics Pipeline for the Identification of CHO Cell

Differential Gene Expression from RNA-Seq Data
Craig Monger, Krishna Motheramgari, John McSharry, Niall Barron,
and Colin Clarke

Abstract
In recent years, the publication of genome sequences for the Chinese hamster and Chinese hamster ovary
(CHO) cell lines has facilitated study of these biopharmaceutical cell factories with unprecedented resolu-
tion. Our understanding of the CHO cell transcriptome, in particular, has rapidly advanced through the
application of next-generation sequencing (NGS) technology to characterize RNA expression (RNA-Seq).
In this chapter, we present a computational pipeline for the analysis of CHO cell RNA-Seq data from the
Illumina platform to identify differentially expressed genes. The example data and bioinformatics workflow
required to run this analysis are freely available at www.cgcdb.org/rnaseq_analysis_protocol.html.

Key words Transcriptomics, RNA-Seq, Differential gene expression, Chinese hamster ovary cells,
Biopharmaceutical manufacture, Systems biotechnology

1  Introduction

Our understanding of Chinese hamster ovary (CHO) cell biology

has dramatically improved in recent years bringing the promise of
rational genetic engineering to enhance biopharmaceutical pro-
duction closer to reality. The catalyst for these rapid advances has
undoubtedly been the publication of genome sequences for mul-
tiple CHO cell lines and the Chinese hamster [1–3]. These data
have had a broad impact on the field, revealing CHO cell line
genomic heterogeneity [2, 4], improving proteomic characteriza-
tion [5], and enabling the use of genome-editing technologies
such as CRISPR-Cas9 [6]. The availability of genomic data has also
improved the accuracy and decreased the cost of next-­generation
sequencing-based transcriptomics (RNA-Seq). The alignment of
reads to a closely related species (i.e., mouse) or the deep RNA
sequencing required to accurately reconstruct the transcriptome de
novo is no longer necessary for CHO cell RNA-Seq studies.

Paula Meleady (ed.), Heterologous Protein Production in CHO Cells: Methods and Protocols, Methods in Molecular Biology,
vol. 1603, DOI 10.1007/978-1-4939-6972-2_11, © Springer Science+Business Media LLC 2017

169
170 Craig Monger et al.

Fig. 1 RNA-Seq bioinformatics protocol overview. (a) Quality assessment of raw sequencing data and prepro-
cessing of reads to correct potential issues including low base quality. (b) Alignment of reads to the Chinese
hamster reference sequence and calculation of global mapping quality. (c) Counting reads aligned to each
protein-coding gene. (d) Differential expression analysis

In this chapter, we present an in-silico protocol for ­differential

mRNA expression analysis from Illumina RNA-Seq data utilizing
the Chinese hamster genome as a reference sequence. The typical
stages of a bioinformatics workflow are outlined (Fig. 1) as well the
commands required to perform each operation. To facilitate the
reproduction of this analysis, we have made both the example data
and computer code freely available (www.cgcdb.org/rnaseq_anal-
ysis_protocol.html). The pipeline begins by illustrating the detec-
tion of common issues in raw sequencing data using FASTQC [7]
and correcting those issues using Trim­momatic [8]. Preprocessed
reads are aligned to the Chinese h­ amster reference genome (C_gri-
seus_v1.0, RefSeq Assembly accession: GCF_000419365.1) with
HISAT2 [9] a fast splice-aware alignment algorithm (a prebuilt
HISAT2 genome index is provided for this purpose). RNASeqQC
[10] is used to determine the effectiveness of read mapping to the
reference genome. Finally, the number of reads aligning to anno-
tated protein-coding genes in the Chinese hamster genome is
determined using HTSeq [11] and imported into the R statistical
software environment where differential expression ­ analysis is
accomplished using the DESeq2 [12] Bioconductor package.

2  Materials

2.1  Software This bioinformatics pipeline is configured for the Linux operating
Installation system to ensure compatibility with widely used RNA-Seq data
analysis software. Ubuntu 16.04 LTS (http://www.ubuntu.com)
has been tested for the analysis described in this chapter and is rec-
ommended for users unfamiliar with Linux due to its Windows-
like desktop. The analysis workflow initially utilizes standalone
software (Table 1), while the final stage is carried out within the R
statistical software environment and the Bioconductor DESeq2
package is used for differential expression analysis (see Note 1).
A Bash script (install_software.sh) has been developed to automati-
cally create the required directories, download and install each
 An in-Silico CHO Cell RNA-Seq Data Analysis Protocol 171

Table 1
List of software. Standalone software written in Java, Python, and C is required for this analysis as well
as the R statistical software environment and the DESeq2 Bioconductor package. The install_software.
sh script automates the process of installing each program, while the DESeq2 package is installed
during the execution of the differential_expression. R script. The purpose of each program is provided
with respect to the protocol described here (see Note 1)

Software Purpose URL Citation

FastQC Read quality http://www.bioinformatics.babraham.ac.uk/ [7]
control projects/fastqc/
Trimmomatic Read preprocessing http://www.usadellab.org/cms/?page=trimmomatic [8]
HISAT2 Read alignment to https://ccb.jhu.edu/software/hisat2/manual.shtml [9]
a reference
sequence
Samtools Manipulation of https://github.com/samtools [13]
SAM/BAM files
RNA-­SeQC Alignment quality http://www.broadinstitute.org/cancer/cga/rna-seqc [10]
assessment
Picard RNA-Seq read http://broadinstitute.github.io/picard/ [14]
deduplication
HTSeq Counting #reads http://www-huber.embl.de/users/anders/HTSeq/ [11]
aligned to
genome features
R Differential https://cran.r-project.org/ NA
Expression
DESeq2 Differential https://bioconductor.org/packages/release/bioc/ [12]
Expression html/DESeq2.html

programme as well as any dependencies (see Note 2). This script

ensures that the installation of each component of this protocol is
straightforward and avoids compatibility issues that may occur in
later stages of the analysis. Administrative privileges (i.e., sudo) are
required to successfully run the analysis and the user will be
prompted to provide their password when required. Users can
download and execute the installation script by typing the follow-
ing commands in the Linux terminal (see Notes 3 and 4):

# Script must be run from the user’s home directory

cd $HOME
# Download the installation script
wget -N www.cgcdb.org/rnaseq_protocol/install_software.sh
# Execute the install script
bash install_software.sh
172 Craig Monger et al.

Table 2
RNA-Seq data and additional resources. The download_data.sh script creates the required directories
and automatically downloads data required for this protocol including the downsampled raw FASTQ
format files and HISAT2 index files for alignment as well as GFF annotation files specifying the location
of features in the Chinese hamster genome. Supplementary files are also provided for RNASeqQC
analysis and annotation of DESeq2 results

Filename
rnaseq_raw_data.tar.gz
hisat_index.tar.gz
C_griseus_v1.0.genomic.tar.gz
Supplementary_data.tar.gz
Contents
• Raw RNA-Seq data for forward and reverse reads. 3 biological
replicates were sequenced for the CHO-K1 and CCL39 samples.
• Prebuilt HISAT2 index files for the Chinese hamster genome.
• Chinese hamster genome FASTA file.
• Chinese hamster GFF annotation file for protein coding genes.
• RNASeqQC sample information file.
• Chinese hamster GTF annotation file for protein-coding genes.
• CSV file for annotation of differentially expressed genelist.

2.2  Data Download The example data utilized for this protocol originates from a
­previously published study focused on the identification of recep-
tors for TLQP-21, a peptide that affects energy metabolism and
stress responses [15]. RNA-Seq was utilized to compare gene
expression differences between CHO-K1 and CCL39 cells (derived
from Chinese hamster lung tissue), which are responsive and non-
responsive to TLQP-21 respectively. Total RNA sequencing on an
Illumina HiSeq 2000 configured to acquire 76 bp paired end reads
was performed for three biological replicates of each cell line.
These data are available for download on the NCBI Sequence Read
Archive (accession: SRA096825). To enable the protocol described
in this chapter to be carried out on a desktop computer, the origi-
nal data has been reduced (downsampled) to 2 million reads per
sample (see Note 5). A Bash script (download_data.sh) is provided
to download the RNA-Seq data for each sample along with addi-
tional resources required. The script automatically stores data in
the required directories for analysis (Table 2).

# Script must be run from the user’s home directory

cd $HOME
# Obtain the data download script
wget -N www.cgcdb.org/rnaseq_protocol/download_data.sh
# Execute the install script
bash download_data.sh
 An in-Silico CHO Cell RNA-Seq Data Analysis Protocol 173

3  Method

Once the required software and data have been downloaded, a

further two scripts are provided to automate each analysis stage in
the pipeline. The first script (run_analysis.sh) sequentially executes
each program required for data preprocessing, read mapping to the
Chinese hamster genome, alignment quality assessment and counts
the number of reads aligning to features in the genome. The sec-
ond script imports the read counts into the R software environ-
ment and performs differential expression analysis. Once complete,
a PCA plot showing the global separation of samples based on their
gene expression profiles as well as a file containing differentially
expressed genes can be found in the “$HOME/rnaseq_analysis/
DESeq2_results” directory. The main commands executed during
the run_analysis.sh (Subheadings 3.1–3.4) and differential_expres-
sion. R (Subheading 3.5) scripts are outlined below. The analysis
scripts can be downloaded and executed as follows.

# Script must be run from the user’s home directory

cd $HOME
# Obtain the data analysis script
wget -N www.cgcdb.org/rnaseq_protocol/run_analysis.sh
# Execute pre-processing, reference genome alignment, alignment QC and counting
bash run_analysis.sh
# Obtain the differential expression R script
wget -N www.cgcdb.org/rnaseq_protocol/differential_expression_analysis.r
# Execute DESeq2 workflow
sudo Rscript differential_expression_analysis.r

3.1  RNA-Seq Raw 1. Assess the quality of raw data using FastQC. The command
Data QC below analyses all (specified by the “*” character) FASTQ files
and Preprocessing in the raw directory folder and writes the result to the output
folder specified by the “--outdir” flag (see Note 6).
2. Preprocess the reads using Trimmomatic. The command below

$fastqc_directory/fastqc \
--outdir $HOME/rnaseq_analysis/FASTQC_output/raw_fastqc "$raw_data_directory"/*

carries out trimming based on the quality of each read in the

sample (see Note 7) using a sliding window of four to assess
the average Q score beginning at the 5′ end of the read. If the
average score falls below 20, the remainder of the read to the
3′ end is removed. Following the trimming phase those reads
with a minimum length <25 nucleotides are also removed (see
174 Craig Monger et al.

Notes 8 and 9). Trimmomatic outputs preprocessed reads for

the forward and reverse reads where read pairs are retained as
well those where only one of the read pairs survived. The pro-
portion of reads remaining in each sample is recorded in the
­trimmomatic.log file (see Note 10). Only paired reads are uti-
lized for downstream stages of this pipeline.

java -jar $trimmotatic_directory/trimmomatic-0.36.jar PE \

"$raw_data_directory"/"$sampleName"_1.fq.gz \
"$raw_data_directory"/"$sampleName"_2.fq.gz \
"$preprocessed_data_directory"/paired/"$sampleName"_1.fq.gz \
"$preprocessed_data_directory"/unpaired/"$sampleName"_1.fq.gz \
"$preprocessed_data_directory"/paired/"$sampleName"_2.fq.gz \
"$preprocessed_data_directory"/unpaired/"$sampleName"_2.fq.gz \
SLIDINGWINDOW:4:20 MINLEN:25 >> "$preprocessed_data_directory"/trimmomatic.log

3. Assess the quality of preprocessed data using FastQC. This stage

provides confirmation that issues have been corrected. Figure 2
illustrates the improvement in base quality scores at the 3′ end
of reads following Trimmomatic preprocessing.

$fastqc_directory/fastqc \
--outdir $HOME/rnaseq_analysis/FASTQC_output/raw_fastqc "$raw_data_directory"/*

3.2  Reference 1. Align sequence reads to the Chinese hamster genome. The pre-
Genome Alignment processed data, where both read pairs have been retained, is
aligned using HISAT2 and the prebuilt C_griseus_v1.0
HISAT2 index. Alignments are outputted in the Sequence
Alignment/Map (SAM) format. The “-x” option specifies the
HISAT2 index, “-1” and “-2” are the input forward and
reverse reads respectively. The “-S” option specifies the output
file in the SAM format (see Note 11).

hisat2 -x $hisat2_index/C_griseus_v1.0 --rg-id 1 --rg SM:Pool1 \

-1 $preprocessed_reads_directory/"$sampleName"_1.fq.gz \
-2 $preprocessed_reads_directory/"$sampleName"_2.fq.gz \
-S $mapped/"$sampleName".sam;

2. Sort the SAM file and convert to BAM. Each SAM format file
produced during alignment is sorted based on location within
each scaffold of the Chinese hamster genome and converted to
its equivalent binary format BAM file. For the Samtools view
command the “-bS” option specifies that the input is SAM
format (-S) and that output should be BAM (-b). The output
of Samtools view is then transferred to the Samtools sort
 An in-Silico CHO Cell RNA-Seq Data Analysis Protocol 175

a
40
38
36
34
32
30
28
26
Quality Score

24
22
20
18
16
14
12
10
8
6
4
2
0
1 2 3 4 5 6 7 8 9 12-13 18-19 24-25 30-31 36-37 42-43 48-49 54-55 60-61 66-67 72-73 76
Position in read (bp)

b 40
38
36
34
32
30
28
26
Quality Score

24
22
20
18
16
14
12
10
8
6
4
2
0
1 2 3 4 5 6 7 8 9 12-13 18-19 24-25 30-31 36-37 42-43 48-49 54-55 60-61 66-67 72-73 76
Position in read (bp)

Fig. 2 Base quality preprocessing. (a) FastQC boxplots of base qualities scores show a significant portion of
reads have quality values falling below 20 at all nucleotide positions in forward reads for the CHO-K1_1
sample. For each nucleotide the boxes and whiskers show where 25–75% and 10–90% of the quality scores
lie respectively, with the red horizontal line indicating the median quality value. (b) Trimming and filtering using
the Trimmomatic tool significantly improves base qualities across the reads
176 Craig Monger et al.

command using the “|” character, with “-o” specifying the

output filename.

samtools view -bS $mapped/"$sampleName".sam | samtools sort - \

-o $mapped/"$sampleName".bam

3. Remove the SAM file. Once the BAM file is generated the SAM
file created during HISAT2 alignment is no longer required
and deleted to reduce storage requirements.

rm $mapped/"$sampleName".sam

3.3  Reference 1. Create a FASTA index for the Chinese hamster genome.
Genome Alignment QC A FASTA index file enables rapid access to sequence within
Chinese hamster FASTA file.

samtools faidx $genome/GCF_000419365.1_C_griseus_v1.0_genomic.fasta

2. Create a sequence dictionary for the Chinese hamster reference

sequence. “R=” specifies the FASTA sequence file from which
to create dictionary, while “O=” specifies the output file. The
resulting .dict file contains a list of names and sizes for each
scaffold in the Chinese hamster genome.

java -jar $picard_directory/CreateSequenceDictionary.jar \

R=$genome/GCF_000419365.1_C_griseus_v1.0_genomic.fasta \
O=$genome/GCF_000419365.1_C_griseus_v1.0_genomic.dict

3. Deduplicate preprocessed reads for RNASeqQC analysis. The

Picard MarkDuplicates program identifies duplicate reads in
each dataset and retains only one of the duplicated reads for
RNASeqQC. The “I” and “O” options specify input and
output BAM files, respectively, while the “VALIDATION_
­
STRINGENCY=SILENT” suppresses warning messages. A
summary of the duplication process can be found in the “.metric.
txt” file.

java -jar $picard_directory/MarkDuplicates.jar \

I=$mapped/"$sampleName".bam \
O=$dedup_directory/"$sampleName".dup.bam \
M=$dedup_directory/"$sampleName".metric.txt VALIDATION_STRINGENCY=SILENT
 An in-Silico CHO Cell RNA-Seq Data Analysis Protocol 177

4. Calculate RNASeq metrics using RNASeqQC. The RNASeqQC

software calculates metrics including depth of coverage and
GC bias. The “-s” option specifies a list of sample files and type
to be analyzed. A GTF annotation file is also supplied follow-
ing the “-t” option that specifies the genomic locations of each
mRNA (see Note 12). Table 3 illustrates selected RNASeqQC
metrics utilized to evaluate the effectiveness of RNA-Seq.

java -jar $rnaseqqc_directory/RNA-SeQC_v1.1.8.jar \

-s $supplementary_files_directory/rnaseq_qc_sample_list.txt \
-t $
supplementary_files_directory/ \
GCF_000419365.1_C_griseus_v1.0_genomic.­protein.coding.gtf \
-r $genome/GCF_000419365.1_C_griseus_v1.0_genomic.fasta \
-o $rnaseq_qc_output

3.4  Calculation 1. Count the number of reads mapping to each gene in the Chinese
of Raw Counts hamster genome. The htseq-count program is utilized to count
the number of reads for each BAM file (the input file format is
specified by the “-f” flag) mapping to each feature present in a
GFF format annotation file (see Note 13). Those features
counted within the GFF file are specified by the “-i” flag, in this
case those with “gene” specified. The “>” character writes the
output to a text file for further processing (see Note 14).

htseq-count -f bam -i gene -s no $mapped/"$sampleName".bam \

$genome/GCF_000419365.1_C_griseus_v1.0_genomic.­protein.coding.gff \
> $count_directory/"$sampleName".counts

3.5  Differential Gene The remaining stages of this pipeline are executed within R and
Expression Analysis utilize the DESeq2 Bioconductor package.
1. Import the count data and create a DESeq2 object. The count
file location for each sample within the “HTSeq_counts” direc-
tory along with the cell type (e.g., CHO-K1 or CCL39) is
placed in an R data frame. The data frame is utilized as input to
the DESeqDataSetFromHTSeqCount function, which imports
the count data and constructs a DESeq object. The cell type
information is converted to a R factor variable for downstream
sample comparisons.
 178

Table 3
Read alignment quality metrics. The RNA-SeQC program outputs a selection of measures of alignment quality for each RNA-Seq sample. For the
Craig Monger et al.

example data utilized in this protocol, an average of 91% of the total reads were aligned to the Chinese hamster genome sequence (mapping rate). The
number of mapped reads that spanned an exon-exon junction is shown (split reads). Over 80% of aligned reads assigned to genes (intragenic rate) and
over 74% to exons (exonic rate) enabling the detection of over 9000 genes and 18,000 mRNAs in each sample. An average profiling efficiency (sequenced
reads vs. exon mapped reads) of 68.3% was achieved

Split Mapping Genes Transcripts Intragenic Exonic Intronic Intergenic Expression Profiling
Sample Mapped Reads Rate (%) Detected Detected Rate (%) Rate (%) Rate (%) Rate (%) Efficiency (%)
CCL39_1 2,263,342 491,468 89.0 9,342 18,138 80.8 74.4 6.4 19.2 66.2
CCL39_2 2,355,486 534,102 93.1 9,340 18,088 82.3 77.1 5.2 17.7 71.7
CCL39_3 2,267,567 494,761 89.7 9,376 18,215 80.8 74.3 6.4 19.2 66.7
CHO-­K1_1 2,283,769 512,589 90.4 9,251 18,125 81.4 75.7 5.8 18.5 68.4
CHO-­K1_2 2,314,711 505,472 91.7 9,292 18,173 81.1 74.3 6.9 18.9 68.1
CHO-­K1_3 2,428,589 535,162 92.2 9,293 18,163 81.3 74.5 6.9 18.6 68.7
 An in-Silico CHO Cell RNA-Seq Data Analysis Protocol 179

# determine the names of the HTSeq count files

count_file_names <- grep("counts",list.files("HTSeq_counts"),value=T)

# set sampleConditions and sampleTable for experimental conditions

cell_type <- c("CCL39","CCL39","CCL39", "CHO-K1","CHO-K1","CHO-K1")

sample_information <- data.frame(sampleName = count_file_names,

fileName = count_file_names,
condition = cell_type)

# create a deSeq2 data object by reading the count files and assign cell type
DESeq_data <- DESeqDataSetFromHTSeqCount(sampleTable = sample_information,
directory = "HTSeq_counts",
design = ~condition)

# convert cell type to an R factor variable

colData(DESeq_data)$condition <- factor(colData(DESeq_data)$condition,
levels = c('CCL39','CHO-K1'))

2. Principal components analysis. Principal components analysis

(PCA) is a useful means to determine if the biological hypoth-
esis underlying the experimental design is reflected in the
global expression profiles before progressing to differential
expression analysis. The raw count data is first transformed
using the regularized logarithm before the DESeq2’s plotPCA
wrapper function is used to carry out PCA, generate a scatter
plot of the first vs the second principal component, and label
the points based on the cell type. The PCA plot for the
RNA-Seq data utilized in this protocol illustrates clear separation
of the CHO-K1 and CCL39 samples (Fig. 3).

# transform the count data for PCA

rld <- rlogTransformation(DESeq_data, blind=T)

# create a PDF to save the figure

pdf("DESeq2_results/CHO-K1_v_CCL39.pdf")

# plot principal component 1 v principal 2, label based on cell type

plotPCA(rld, intgroup="condition")

# save plot to PDF file

dev.off()

3. Differential expression analysis using DESeq2. Before conducting

differential expression analysis, a comparator sample is set using
the relevel function, in this case the TLQP-21 nonresponsive
CCL39 cell line. The DESeq wrapper function is called on the
DESeq object containing the raw counts. The DESeq2 method
conducts size factor normalization, removes genes with very
low counts, calculates differential expression and adds the
180 Craig Monger et al.

5.0

2.5

0.0
PC2: 3% variance

group
CCL39
CHO–K1

–2.5

–5.0

–7.5

–20 –10 0 10 20

PC1: 95% variance

Fig. 3 Principal component analysis. PCA provides a global overview of the transcriptomic data, identifying
­outlying samples and confirming that the biological hypothesis underlying the experimental design can be
observed. In this example, the CHO-K1 and CCL39 samples clearly separate following PCA

results to the existing object. The differential expression results

are filtered based on fold change (≤−2 or ≥2), as well as a
Bonferroni-­ adjusted p-value of <0.05 (see Note 15) before
ordering the results based on the degree of up or downregula-
tion in CHO-­K1 when compared to CCL39.
 An in-Silico CHO Cell RNA-Seq Data Analysis Protocol 181

# set the comparator condition

DESeq_data$condition<-relevel(DESeq_data$condition, "CCL39")

# calculate differential expression using the DESeq wrapper function

DESeq_data <- DESeq(DESeq_data)

#set differential expression criteria

de_results<-results(DESeq_data, pAdjustMethod="bonferroni",
lfcThreshold=0, independentFiltering=T)

# order results by padj value (most significant to least)

sig_de_results <- subset(de_results, abs(log2FoldChange)> 1 & padj < 0.05)
sig_de_results <- sig_de_results[order(sig_de_results$log2FoldChange,
decreasing=T),]

4. Annotate differentially expressed genes. An annotation file is

provided in the supplementary data that, when imported into
the R environment using the read.csv function, can be used to
add the gene name and GenBank ID to the differentially
expressed genes identified by DESeq2.

# import annotation information

annotation_file<-
"supplementary_files/GCF_000419365.1_C_griseus_v1.0_genomic.­
protein.coding.csv"
annotation_info<-read.csv(annotation_file, row.names=1, header=T)

# identify annotation information for differentially expressed genes

sig_de_annotations <- annotation_info[rownames(sig_de_results),]

#combine annotation information and DESeq2 output

sig_de_results<-cbind(sig_de_annotations, as.data.frame(sig_de_results))

5. Export differentially expressed genes to a CSV file. Annotated

significantly differentially expressed genes are written to a CSV
file using the write.csv R function. Table 4 illustrates the final
output of this pipeline for the ten up and downregulated genes
in CHO-K1 cells when compared with CCL39 cells.

write.csv(sig_de_results, row.names=T, file="DESeq2_results/CHO-K1_v_CCL39.csv",)

4  Notes

1. The pipeline in this chapter utilizes selected software packages.

Readers should note that there is an array of different
­bioinformatics software available. Alternatives to each stage of
this pipeline are discussed in a recent review from our ­laboratory
[16].
Table 4
182

DESeq2 differential expression output. A total of 572 upregulated and 924 downregulated protein-­coding genes were identified. The ten most up and
downregulated genes are shown here. The corresponding symbol, name, and GenBank ID are shown for each gene. DESeq2 outputs including the fold
change observed in log2 scale as well as the p-value the Bonferroni adjusted p-value

Gene Base Log2 Fold Adjusted

Symbol Gene Name GenBank ID Mean Change P-value P-value
Slc45a1 solute carrier family 45 100750542 277.79 8.28 6.79 × 10−48 7.13 × 10−44
member 1
Craig Monger et al.

Ccnd2 cyclin D2 100771544 431.30 8.22 9.86 × 10−66 1.04 × 10−61

Eng Endoglin 100757433 641.12 8.10 6.57 × 10−92 6.91 × 10−88
Mmp9 matrix metallopeptidase 9 100770707 166.28 7.38 3.30 × 10−40 3.47 × 10−36
Trim44 tripartite motif 100757190 175.97 7.23 5.61 × 10−43 5.9 × 10−39
containing 44
Dcaf12l2 DDB1 and CUL4 100764386 99.35 6.71 1.42 × 10−31 1.49 × 10−27
associated factor
12-like 2
Fam134b family with sequence 100766605 73.70 6.61 1.24 × 10−26 1.30 × 10−22
similarity 134
member B
Ppp1r36 protein phosphatase 1 100765040 108.81 6.40 3.93 × 10−35 4.13 × 10−31
regulatory subunit 36
Tbc1d16 TBC1 domain family 100754815 56.70 6.28 2.28 × 10−23 2.39 × 10−19
member 16
Hoxd8 homeobox D8 100758537 70.42 6.26 2.45 × 10−26 2.58 × 10−22
Gene Base Log2 Fold Adjusted
Symbol Gene Name GenBank ID Mean Change P-value P-value
Thbs2 thrombospondin 2 100752022 261.34 –7.89 1.44 × 10−52 1.51 × 10−48
Slc25a4 solute carrier family 25 100751779 222.01 -7.91 4.51 × 10−48 4.74 × 10−44
member 4
Cdh2 cadherin 2 100689204 185.05 -7.93 1.70 × 10−42 1.78 × 10−38
Cdh11 cadherin 11 100767551 196.15 -8.00 1.11 × 10−43 1.17 × 10−39
Rspo3 R-spondin 3 100755164 250.24 -8.21 1.12 × 10−45 1.18 × 10−41
Grb10 growth factor receptor 100760784 532.50 -8.45 7.24 × 10−73 7.61 × 10−69
bound protein 10
Col1a2 collagen type I alpha 2 100766269 617.51 -9.04 1.35 × 10−75 1.42 × 10−71
Ftl ferritin-light polypeptide 100753846 1049.17 -9.07 3.44 × 10−59 3.61 × 10−55
Fabp4 fatty acid-binding 100760812 584.03 -9.16 1.54 × 10−70 1.62 × 10−66
protein 4
Bgn biglycan 100771022 901.38 -9.20 4.42 × 10−93 4.64 × 10−89
An in-Silico CHO Cell RNA-Seq Data Analysis Protocol
183
184 Craig Monger et al.

2. Bash scripts are provided for convenience and to ensure com-

patibility. Readers are encouraged to examine the contents of
these files using a text-editor such as gedit or by printing the
contents to the terminal using the following command: more
$HOME/install_software.sh.
3. Readers should be aware that when copying and pasting com-
mands directly from this article into the Linux terminal charac-
ters can be altered which will result in an error (e.g., single and
double quotes as well as the “-” character).
4. To aid in understanding of the computer code used in this pro-
tocol, the following color scheme is used: calls to software & R
functions (blue), software options/flags/R function a­ rguments
(green). Variables (purple) are preceded by the “$” character and
are used here as references or “shortcuts” to directories and sam-
ple filenames. Comments lines (gray) are preceded by the “#”
character (ignored by Bash and R) and the “\” character specifies
a new line within a Linux command (used for code readability).
5. Downsampling of the original data comprising of ~750 million
paired-end reads to 2 million reads per sample is necessary to
run this protocol on a standard desktop computer. Down­
sampling was carried out using BBmap [17]. The protocol has
been tested on a 4GB computer with the downsampled data
provided. ~50GB free hard drive space is required. Total run-
time: ~10 h (1 processor, 4GB RAM).
6. FastQC provides a user-friendly HTML-based output that can
be viewed in an Internet browser such as Firefox. The run_
analysis.sh script will automatically open the browser and a new
tab for each sample to display the respective FastQC output.
7. The run_analysis.sh script uses a series of while loops to iterate
over a list of sample names that are imported from a text file.
Upon each iteration of the loop the $sample_name variable speci-
fies the current file name in the list. The while loop completes
when processing of each of the six samples has been carried out. In
this chapter we focus on the commands within the loops, for those
readers wishing to understand how a while loop is constructed in
Bash should see the run_analysis.sh script (see Note 1).
8. The Trimmomatic quality threshold of 20 used here should be
modified as appropriate, for instance, less stringent parameters
can be used with data from newer sequencing instruments that
yield higher quality reads.
9. The presence of adapter sequences is commonly encountered in
raw RNA-Seq data. The example data utilized here did not have
a significant degree of adapter contamination and therefore no
adapter trimming was carried out. Readers should note, however,
that Trimmomatic can also be used to remove adapter sequences
as well as low quality bases. Adapter trimming can be specified
 An in-Silico CHO Cell RNA-Seq Data Analysis Protocol 185

using the ILLUMINACLIP option in conjunction with a FASTA

file of the adapter sequences used for library preparation.
10. gedit can be used to view a summary of Trimmomatic or more
"$preprocessed_data_directory"/trimmomatic.log
can be used to print the log file to the terminal.
11. The read group parameters (--rg-id 1 --rg SM:Pool1) are not
required by HISAT2. In this protocol, arbitrary read groups
are added to the SAM file to ensure compatibility with Picard
and RNA-SeQC during later stages of the pipeline. These
parameters can be removed during alignment if users do not
wish to use RNA-SeQC during their analysis.
12. Conversion of the NCBI GFF format annotation file was
achieved using the gffread utility (http://ccb.jhu.edu/soft-
ware/stringtie/gff.shtml). This file is supplied for convenience.
13. The GFF file utilized here contains 14,781 protein-coding
regions selected from the original GCF_000419365.1_C_gri-
seus_v1.0_genomic.gff file and is provided here for demon-
stration purposes.
14. The final five lines of each sample count file contain the overall
metrics for the counting process including the number of reads
that could not be assigned to a genome feature. The following
command can be used to print the counting metrics: tail –n
5 $count_directory/"$sampleName".counts.
15. In this protocol, we utilize the Bonferroni p-value adjustment
for demonstration purposes; however, in certain cases, it can
be an overly conservative measure of statistical significance.
Readers should note that a number of alternative less stringent
adjustment methods are available for the DESeq2 package, e.g.,
Benjamini-Hochberg.

Acknowledgments

The authors gratefully acknowledge funding from Science

Foundation Ireland (grant refs: 13/SIRG/2084 and 13/
IA/1963) and the eCHO systems Marie Curie ITN programme
(grant ref.: 642663).

References

1. Brinkrolf K, Rupp O, Laux H, Kollin F et al
(2013) Chinese hamster genome sequenced
from sorted chromosomes. Nat Biotechnol 31:
694–695
2. Lewis NE, Liu X, Li Y, Nagarajan H et al (2013)
Genomic landscapes of Chinese hamster ovary
cell lines as revealed by the Cricetulus griseus
draft genome. Nat Biotechnol 31:759–765
3. Xu X, Nagarajan H, Lewis NE, Pan S et al
(2011) The genomic sequence of the Chinese
hamster ovary (CHO)-K1 cell line. Nat Bio­
technol 29:735–741
4. Kaas CS, Kristensen C, Betenbaugh MJ, Andersen
MR (2015) Sequencing the CHO DCB11
genome reveals regional variations in genomic
stability and haploidy. BMC Genomics 16:160
186 Craig Monger et al.
5. Meleady P, Hoffrogge R, Henry M, Rupp O
et al (2012) Utilization and evaluation of
CHO-specific sequence databases for mass
spectrometry based proteomics. Biotechnol
Bioeng 109:1386–1394
6. Ronda C, Pedersen LE, Hansen HG,
Kallehauge TB et al (2014) Accelerating
genome editing in CHO cells using CRISPR
Cas9 and CRISPy, a web-based target finding
tool. Biotechnol Bioeng 111:1604–1616
7. FASTQC, http://www.bioinformatics.babra-
ham.ac.uk/projects/fastqc/
8. Bolger AM, Lohse M, Usadel B (2014) Trim­
momatic: a flexible trimmer for Illu­ mina
sequence data. Bioinformatics 30:2114–2120
9. Kim D, Langmead B, Salzberg SL (2015)
HISAT: a fast spliced aligner with low mem-
ory requirements. Nat Methods 12:357–360
10. DeLuca DS, Levin JZ, Sivachenko A, Fennell T
et al (2012) RNA-SeQC: RNA-seq metrics for
quality control and process optimization.
Bioinformatics 28:1530–1532
11. Anders S, Pyl PT, Huber W (2015) HTSeq—
a Python framework to work with high-­
throughput sequencing data. Bioinformatics
(Oxford, England) 31:166–169
12. Love MI, Huber W, Anders S (2014) Mode­
rated estimation of fold change and dispersion
for RNA-seq data with DESeq2. Genome Biol
15:550
13. Li H, Handsaker B, Wysoker A, Fennell T
et al (2009) The sequence alignment/map
format
­ and SAMtools. Bioinformatics
25:2078–2079
14. Picard, http://broadinstitute.github.io/picard/.
15. Hannedouche S, Beck V, Leighton-Davies J,
Beibel M et al (2013) Identification of the C3a
receptor (C3AR1) as the target of the VGF-­
derived peptide TLQP-21 in rodent cells.
J Biol Chem 288:27434–27443
16. Monger C, Kelly PS, Gallagher C, Clynes M
et al (2015) Towards next generation CHO
cell biology: bioinformatics methods for RNA-­
Seq-­based expression profiling. Biotechnol
J 10:950–966
17. BBMap—Bushnell B.—­sourceforge.net/proj-
ects/bbmap/
 Chapter 12

Filter-Aided Sample Preparation (FASP) for Improved

Proteome Analysis of Recombinant Chinese Hamster
Ovary Cells
Orla Coleman, Michael Henry, Martin Clynes, and Paula Meleady

Abstract
Chinese hamster ovary (CHO) cells are the most commonly used mammalian host cell line for biopharma-
ceutical production because of their ability to correctly fold and posttranslationally modify recombinant
proteins that are compatible with human use. Proteomics, along with other ‘omic platforms, are being
used to understand the biology of CHO cells with the ultimate aim of enhancing CHO cell factories for
more efficient production of biopharmaceuticals. In this chapter, we will describe an efficient protocol
called Filter Aided Sample Preparation (FASP) for the extraction of proteins from CHO cells for proteomic
studies. FASP uses a common ultrafiltration device whereby the membrane pores are small enough to
allow contaminating detergents to pass through, while proteins are too large and are retained and concen-
trated in the filter unit. This method of sample preparation and protein digestion is universally applicable
and can be easily employed in any proteomics facilities as standard everyday laboratory reagents and equip-
ment are used.

Key words Proteomics, Protein solubilization, Filter-aided sample preparation, FASP, Chinese
­hamster ovary, LC-MS/MS

1  Introduction

Chinese hamster ovary (CHO) cells are the most commonly used
host cell line for the production of recombinant biotherapeutics.
A greater understanding of the biology of these cells is required to
improve the efficiency of production with the overall aim of
­reducing the costs of therapeutics. Systems biology approaches,
including proteomics, are being increasingly used to characterize
recombinant CHO cells to achieve this goal. This chapter outlines
a recently described method, Filter-Aided Sample Preparation
(FASP) [1], which enables protein purification using a standard
ultrafiltration device for efficient detergent removal prior to down-
stream proteomic analyses.

Paula Meleady (ed.), Heterologous Protein Production in CHO Cells: Methods and Protocols, Methods in Molecular Biology,
vol. 1603, DOI 10.1007/978-1-4939-6972-2_12, © Springer Science+Business Media LLC 2017

187
188 Orla Coleman et al.

Detergents are the reagents of choice for solubilization of cells

and tissue; however, detergents used for proteomic applications
can suppress enzymatic digestions, contaminate HPLC instrumen-
tation, interfere with column binding and column elution, and
because they are readily ionizable can dominate mass spectra. In
any proteomics experiment the choice of detergent is a balance of
the need to solubilize and stabilize the protein without deleteri-
ously affecting downstream processing and analysis. The detergent
should not interfere with downstream techniques but often dena-
turing detergents used for protein extractions do; therefore, their
removal is required prior to further processing.
In this chapter, we describe a recently introduced method,
FASP, which can be used to purify protein samples [1]. This
method cleverly uses a common ultrafiltration device for detergent
removal to efficiently purify the proteome for subsequent analyses.
The use of a filter device removes the need for protein precipitation
as a means of detergent removal which often results in significant
sample loss. Sample solubilization is achieved using SDS; following
this the protein is retained, washed, and concentrated in the filter
device. The filter device then acts as a “proteomic reactor” for
detergent removal, buffer exchange, chemical modification, and
protein digestion. The underlying principle of this method is that
low-molecular-weight detergents and impurities are small enough
to pass through the filter membrane, while high-molecular-weight
substances such as proteins are retained. After protein digestion,
the digested peptides are small enough to pass through the filter
membrane and be collected for mass spectrometry analysis. The
efficient removal of impurities and detergents allows for a more
complete coverage of the proteome. The FASP method has been
successfully applied to various diverse proteomic analyses such as
FFPE cancer tissues, brain phosphoproteome, Escherichia coli, and
modified versions for glycoproteomic studies [2–6]. It has also
been used to maximize protein recovery from CHO cells where
6164 proteins were identified from both glycoproteome and pro-
teome analysis [7]. This chapter provides a step-by-step compre-
hensive protocol for the improved proteome analysis of recombinant
CHO cells using a combination of FASP preparation and strong
cation exchange (SCX) for efficient fractionation and identification
of the complex proteome.

2  Materials

All solvents and water used must be LC-MS grade. All chemicals
must be of the highest purity.

2.1  Equipment 1. Microcon-10 kDa Centrifugal Filter Unit with Ultracel-10

membrane (Merck Millipore).
 FASP for CHO Proteomic Analysis 189

2. Microcentrifuge capable of spinning 1.7 mL microcentrifuge

tubes at 14,000 × g at 20 °C.
3. Pierce® C18 100 μL tips (Thermo Fisher Scientific).
4. Sonicating probe.
5. Microplate Spectrophotometer reading 96-well plate format at
595 nm (e.g., Multiskan™ GO, Thermo Fisher Scientific).
6. SpeedVac Vacuum Dryer (Thermo Fisher Scientific) or
alternative.
7. Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific).
8. Pierce™ Strong Cation Exchange Spin Column, Mini (Thermo
Fisher Scientific).
9. Dry Heating Block.
10. pH meter.

2.2  Reagents 1. TRIZMA® HCl.

2. TRIZMA® Base.
3. Sodium dodecyl sulfate (SDS).
4. DL-Dithiothreitol (DTT).
5. Iodoacetamide.
6. Urea.
7. LC-MS grade water.
8. Sequencing-grade modified trypsin (e.g., Promega, Sigma
Aldrich).
9. ProteaseMax™ Surfactant Trypsin Enhancer (Promega).
10. Phosphate-Buffered Saline (sterile).
11. Trifluoroacetic acid (TFA).
12. Acetonitrile (ACN).
13. Sodium Chloride (NaCl).
14. Ammonium Bicarbonate.
15. Potassium Phosphate Monobasic (KH2PO4).
16. Potassium Chloride (KCl).

2.3  Buffer 1. Tris–HCl stock solutions: The following Tris–HCl mixing

Preparation for FASP table (see Table 1) is used for the preparation of 0.1 M stock
solutions of varying pH’s in 50 mL, which is used to prepare
the lysis and wash buffers for FASP. The precise blend of
Trizma HCl and Trizma Base, as outlined in Table 1, will bring
the solutions to the desired pH; however, this must be con-
firmed using a pH meter (see Note 1).
2. Lysis Buffer: 4% SDS, 0.1 M DTT solution. To make a 10 mL
solution of the lysis buffer add 0.4 g of high-purity SDS and
190 Orla Coleman et al.

Table 1
Preparation of Tris–HC1 solutions with varying pH

pH Trizma-HCl (g) Trizma Base (g) LC-MS water (mL)

7.6 0.606 0.139 50
7.9 0.488 0.230 50
8.5 0.221 0.436 50

0.0154 g of DTT to 10 mL of 0.1 M Tris–HCl stock solution

at pH 7.6. Take caution when weighing SDS by using a fume
hood or a dust mask. Use 1 mL of this lysis buffer per 2 × 106
CHO cells.
3. Wash buffer 1: A solution of 8 M urea in 0.1 M Tris–HCl
pH 8.5. Add 0.4 g of urea to 1 mL of the 0.1 M Tris–
HCl pH 8.5 stock solution.
4. Wash buffer 2: A solution of 8 M urea in 0.1 M Tris–HCl
pH 7.9. Add 0.4 g of urea to 1 mL of the 0.1 M Tris–
HCl pH 7.9 stock solution.
5. Alkylation solution: 0.05 M Iodoacetamide in 8 M Urea,
0.1 M Tris–HCl, pH 8.5. To prepare 1 mL of this solution add
0.4 g of urea and 0.009 g of iodoacetamide to 1 mL of the
0.1 M Tris–HCl pH 8.5 stock solution.
6. Rinse solution: Add 0.029 g of NaCl to 1 mL of LC-MS grade
water.

2.4  Protein Digestion 1. 50 mM Ammonium Bicarbonate (must be freshly prepared

and used within 24 h).
2. Trypsin solution: use sequence grade trypsin (see Note 2).
3. ProteaseMax™ Surfactant Enhancer (see Note 3).

2.5  Peptide 1. Sample acidification buffer: prepare a solution of 2.5% TFA in

Purification LC-MS grade water.
2. Wetting solution: prepare 50% ACN in LC-MS grade water.
3. Equilibration buffer: 0.1% TFA, 2% ACN.
4. Rinse solution: 0.1% TFA, 2% ACN.
5. Elution solution: 0.1% TFA, 70% ACN.

2.6  Peptide 1. 10 mM KH2PO4 in 25% acetonitrile pH 3.0 stock (diluent)

Fractionation Using solution: To prepare 1 L of this stock (diluent) solution, add
SCX Spin Cartridges 1.36 g of KH2PO4 and 250 mL of acetonitrile to 750 mL of
LC-MS grade water and adjust the pH to 3.0.
2. Prepare the following 10 mL Potassium Chloride (KCl) elu-
tion buffers as outlined in Table 2 using the 10 mM KH2PO4
in 25% acetonitrile pH 3.0 diluent solution.
 FASP for CHO Proteomic Analysis 191

Table 2
Overview of preparation of KC1 elution buffers

Elution buffer KCl (g) 10 mM KH2PO4 in 25% ACN (mL) KCl concentration (mM)
Elution buffer step 1 0.007 10  10
Elution buffer step 2 0.019 10  25
Elution buffer step 3 0.037 10  50
Elution buffer step 4 0.056 10  75
Elution buffer step 5 0.075 10 100
Elution buffer step 6 0.093 10 125
Elution buffer step 7 0.112 10 150
Elution buffer step 8 0.149 10 200
Elution buffer step 9 0.224 10 300
Elution buffer step 10 0.373 10 500

3  Methods

3.1  Cell Lysis 1. Harvest CHO cells and wash with sterile Phosphate-Buffered
Saline three times. Centrifuge at 1000 × g for 5 min and remove
the supernatant. Snap freeze the cell pellet in a microcentrifuge
tube using liquid nitrogen (take extreme caution when doing
this). Store at −80 °C until cell lysis is performed.
2. Lyse cell pellets corresponding to 2 × 106 cells (approximately
1 mg pellet of cells) with 1 mL of lysis buffer. Suspend the cell
pellet in the lysis buffer and mix the solution well using a
pipette. Further disrupt the cells using a sonicating probe while
maintaining the cell lysate at 4 °C.
3. Heat the lysate for 20 min at 56 °C using a heating block to
denature the protein.
4. Determine the protein concentration using a BCA assay as per
manufacturer’s instructions.
5. Transfer 100 μg protein lysate aliquots into microcentrifuge
tubes and freeze at −80 °C until required (see Note 4).

3.2  Filter Aided 1. Mix 100 μg of protein lysate with 200 μL of 8 M Urea.
Sample Preparation 2. Transfer the solution to a Microcon-10 kDa Centrifugal Filter
(FASP) Unit and spin using a microcentrifuge set to 14,000 × g at
20 °C for 40 min.
3. Discard the filtrate (flow-through) and dilute the concentrate
with 200 μL of Wash buffer 1 and spin using a microcentrifuge
set at 14,000 × g at 20 °C for 40 min.
192 Orla Coleman et al.

4. Discard the filtrate and alkylate the concentrate with 100 μL of

the alkylation solution.
5. Centrifuge the device at 14,000 × g at 20 °C for 40 min.
6. Discard the filtrate and dilute the concentrate with 100 μL of
Wash buffer 2 and centrifuge at 14,000 × g at 20 °C for 40 min.
7. Repeat step 6.
8. Proceed directly to protein digestion in Subheading 3.3.

3.3  Protein Digestion 1. Dilute the concentrate with 100 μL of 50 mM Ammonium
Bicarbonate.
2. Add 5 μg of sequence-grade trypsin (1:20 enzyme:protein).
3. Add 1 μL of 1% ProteaseMax™ Surfactant (see Note 5).
4. Digest at 37 °C for 3 h, see Note 6.
5. Transfer the Microcon-10 kDa Centrifugal Filter Unit to a
new collection tube and centrifuge at 14,000 × g at 20 °C for
40 min. Retain the filtrate in the collection tube.
6. The FASP ultrafiltration device must be rinsed post-digestion to
elute residual peptides that may be bound to the filter membrane
or walls of the device. Rinse the filter unit with 50 μL of the 0.5 M
NaCl rinse solution and spin at 14,000 × g at 20 °C for 20 min.
7. Store the filtrate that contains the peptides at −80 °C until
required.

3.4  Peptide 1. Add 6 μL of sample acidification buffer to the peptide filtrate
Purification to give a final concentration of 0.1% TFA.
2. Aspirate the 100 μL C18 tip using 100 μL of wetting solution
and discard the solvent flow-through (see Note 7).
3. Repeat step 2.
4. Equilibrate the C18 tip using 100 μL of the equilibration buf-
fer and discard the solvent flow-through.
5. Repeat step 4.
6. Aspirate and dispense the peptide sample into the C18 tip ten
times to allow peptide binding (see Note 8).
7. Wash the C18 tip using 100 μL of the rinse solution and
­discard rinse wash.
8. Repeat step 7.
9. Slowly elute the peptide sample using 50 μL of elution buffer
into a new microcentrifuge tube.
10. Repeat step 9 to give a final volume of 100 μL.
11. Concentrate the peptide sample to dryness in a SpeedVac
­vacuum dryer.
12. Freeze peptide sample at −80 °C or proceed directly to Strong
Cation Exchange preparation.
 FASP for CHO Proteomic Analysis 193

3.5  Strong Cation 1. Condition SCX spin column with 400 μL of 10 mM KH2PO4
Exchange (SCX) Using in 25% ACN pH 3.0 and centrifuge at 2000 × g for 5 min.
Pierce™ Mini Spin Discard the flow-through.
Columns 2. Resuspend the dried peptides from the final step of Subheading
3.4 in 200 μL of 10 mM KH2PO4 in 25% ACN pH 3.0. Apply this
to the conditioned SCX column, centrifuge at 2000 × g for 5 min.
3. Add 200 μL of elution buffer step 1, centrifuge at 2000 × g for
5 min, and collect the flow-through.
4. Repeat this procedure to elute the peptides in a step-wise man-
ner using elution buffer step 2 through to elution buffer step
10, collecting each flow-through.
5. Proceed directly to desalting using peptide purification in
Subheading 3.4 or freeze at −80 °C. Once desalting is com-
plete the peptide samples are ready for LC-MS analysis.

4  Notes

1. For long-term storage of the 0.1 M Tris–HCl stock solutions,

store at 4 °C when not in use.
2. Use sequencing-grade or mass spectrometry grade trypsin.
Reconstitute the lyophilized trypsin in 50 mM acetic acid to
give a final concentration of 1 μg/μL and store aliquots at
−20 °C until required. In our lab we use Promega’s Trypsin
Gold Mass Spectrometry grade or Pierce™ Trypsin Protease
MS grade for protein digestion.
3. Add 100 μL of 50 mM ammonium bicarbonate to a 1 mg vial
of ProteaseMax™ Surfactant to give a 1% solution. Store ali-
quots at −20 °C until required. Ammonium Bicarbonate must
be freshly prepared and used within 24 h.
4. From our use of this method we advise to prepare the protein
lysate and determine the protein concentration on day 1.
Perform FASP and digestion the following day and peptide
purification and fractionation on day 3.
5. We use ProteaseMax™ Surfactant Trypsin Enhancer to ensure
efficient protein digestion with trypsin during the 3 h diges-
tion period.
6. In our experience, to reduce sample loss due to evaporation
during sample digestion we recommend securing the lids of the
Microcon-10 kDa Centrifugal Filter Units using Parafilm wrap.
7. For peptide purification using C18 tips, ensure that air is not
drawn into the tip and the C18 sorbent does not dry during
sample processing.
8. For larger volumes of peptide sample when purifying using the
C18 tips, repeat Subheading 3.4, step 6 as many times as neces-
sary to process the entire sample and ensure peptide binding.
194 Orla Coleman et al.
Acknowledgments
The authors acknowledge funding from Science Foundation
Ireland (grant ref. 13/1A/1841).
References
1. Wisnieskie JR, Zougman A, Nagaraj N, Mann
M (2009) Universal sample preparation method
for proteome analysis. Nat Methods 6(5):
359–362
2. Wiśniewski JR, Ostasiewicz P, Mann M (2011)
High recovery FASP applied to the proteomic
analysis of microdissected formalin fixed paraffin
embedded cancer tissues retrieve known
colon cancer markers. J Proteome Res 10(7):
3040–3049
3. Wiśniewski JR, Nagaraj N, Zougman A, Gnad
F, Mann M (2010) Brain phosphoproteome
obtained by a FASP-based method reveals
plasma membrane protein topology. J Proteome
Res 9(6):3280–3289
4. Erde J, Ogorzalek Loo RR, Loo JA (2014)
Enhanced FASP (eFASP) to increase proteome
coverage and sample recovery for quantitative
proteomic experiments. J Proteome Res 13(4):
1885–1895
5. Deeb SJ, Cox J, Schmidt-Supprian M, Mann M
(2014) N-linked glycosylation enrichment for
in-depth cell surface proteomics of diffuse large
B-cell lymphoma subtypes. Mol Cell Proteomics
13(1):240–251
6. Hecht ES, McCord JP, Muddiman DC (2016)
A quantitative glycomics and proteomics com-
bined purification strategy. J Vis Exp 109:
e53735. doi:10.3791/53735
7. Baycin-Hizal D, Tabb DL, Chaerkady R, Chen
L, Lewis NE, Nagarajan H, Sarkaria V, Kumar
A, Wolozny D, Colao J, Jacobsen E, Tian Y,
O’Meally RN, Krag SS, Cole RN, Palsson BO,
Zhang H, Betenbaugh M (2012) Proteomic
analysis of Chinese hamster ovary cells.
J Proteome Res 11(11):5265–5272
 Chapter 13

Phosphopeptide Enrichment and LC-MS/MS Analysis

to Study the Phosphoproteome of Recombinant Chinese
Hamster Ovary Cells
Michael Henry, Orla Coleman, Prashant, Martin Clynes,
and Paula Meleady

Abstract
The reversible phosphorylation of proteins on serine, threonine, and tyrosine residues is one of the most
important post-translational modifications that regulates many biological processes. The phosphopro-
teome has not been studied in any great detail in recombinant Chinese hamster ovary (CHO) cells to date
despite phosphorylation playing a crucial role in regulating many molecular and cellular processes relevant
to bioprocess phenotypes including, for example, transcription, translation, growth, apoptosis, and signal
transduction. In this chapter, we provide a protocol for the phosphoproteomic analysis of Chinese hamster
ovary cells using phosphopeptide enrichment with metal oxide affinity chromatography (MOAC) and
immobilized metal affinity chromatography (IMAC) techniques, followed by site-specific identification of
phosphorylated residues using LC-MS (MS2 and MS3) strategies.

Key words Phosphoproteomics, Phosphopeptide enrichment, Site specific phosphorylation, Chinese

hamster ovary cells

1  Introduction

The Chinese hamster ovary (CHO) cell is the most commonly used
mammalian host cell line for the production of recombinant bio-
therapeutics [1]. This dominance is likely to continue for the fore-
seeable future due to their ability to correctly fold proteins that are
compatible with human use, their ability to produce proteins with
human-like post-translational modifications, their safety record,
and their track record in industry [2]. There has been considerable
success in developing high-producing CHO cell culture processes
through media optimization and bioreactor design [3, 4]; how-
ever, the bottlenecks in the cellular machinery for the efficient pro-
duction of recombinant proteins are poorly understood. In order
to improve the efficiency of production with the overall aim of

Paula Meleady (ed.), Heterologous Protein Production in CHO Cells: Methods and Protocols, Methods in Molecular Biology,
vol. 1603, DOI 10.1007/978-1-4939-6972-2_13, © Springer Science+Business Media LLC 2017

195
196 Michael Henry et al.

reducing the costs of therapeutics, a fundamental understanding of

the biology underpinning growth and productivity of CHO cells is
required. Systems biology approaches (including transcriptomics,
proteomics, metabolomics, etc.) are being increasingly used to
characterize recombinant CHO cells and will be instrumental in
metabolic and cell engineering interventions to enhance CHO cell
lines for more efficient production of therapeutics [5, 6].
In the past CHO systems biology has been hampered by
the lack of genomic data on CHO cells. The recently sequenced
genomes of CHO cells [7–10] have opened up a new era involving
analysis of CHO cells on a systems biology level. Examining the
transcriptome can provide insight into the regulation of cellular
processes, but mRNA levels are not necessarily correlated with
protein expression levels for their protein products [11, 12]. In
addition, proteins are subjected to a variety of posttranslational
modifications (PTMs) that affect biological activity; phosphoryla-
tion being one of most widely studied PTMs, especially in disease
states such as cancer. The vast majority of ‘omic based studies in
CHO cells to date have missed out on the posttranslational level of
regulation despite their pivotal role in the regulation of growth,
cell cycle arrest, apoptosis, transcription, signal transduction, etc.
and hence are likely to be central to understanding and controlling
bioprocess-relevant phenotypes. The phosphoproteome has not
been assessed in any detail in CHO [13] though some research has
been done in CHO cells, based on knowledge of phosphorylation
events in cancer cells described in the literature.
PTMs increase chemical diversity and complexity of the
­proteome, and can determine protein function by altering activity,
cellular location, turnover, and interaction with other proteins [14].
The complexity of the cellular processes that rely upon phosphory-
lation is demonstrated in that approximately 30% of all eukaryotic
proteins can become phosphorylated [15, 16], though some of
these phosphorylation events may be physiologically irrelevant [17].
Almost half of the tyrosine kinases of the human “kinome” are
implicated in human cancers as well as numerous serine-­threonine
kinases, including Akt and mTOR [18]. Phosphatases such as
PTEN and PP2A are also known to play important roles in cancer
[19, 20]. Phosphorylation of nuclear proteins has been widely rec-
ognized as a key chemical modification necessary for the control of
mRNA transcription [21].
While some proteins are constitutively phosphorylated, most
are only transiently phosphorylated following key cellular cues,
thus altering the chemical state of a protein in subtle ways that
are not easily detected using standard profiling techniques [22].
Phosphorylation sites in most proteins are sub-stoichiometric and
hence are often present in low abundance compared to their native
counterparts. As a result, phosphopeptide enrichment using
immobilized metal ion (Fe3+ or Ga3+) affinity chromatography
 Phosphoproteomic Analysis of CHO Cells 197

(IMAC) and metal oxide affinity chromatography (MOAC) using

TiO2 or ZrO2 prior to LC-MS/MS analysis is widely used in phos-
phoproteomic studies [23]. Both these materials have different
binding specificities, resulting in increased phosphoproteome cov-
erage [24].
In this chapter, we describe methods for phosphopeptide
enrichment using TiO2 and IMAC strategies from CHO cells
followed by LC-MS analyses (MS2 and MS3) for site-specific
­
determination of phosphorylated serine, threonine, or tyrosine
residues. The methods described are also amenable to differential
phosphoproteomic studies.

2  Materials

2.1  Equipment 1. Pierce® Fe-NTA (IMAC) Phosphopeptide Enrichment Kit

(Thermo Fisher Scientific).
2. Pierce® TiO2 Phosphopeptide Enrichment Kit (Thermo Fisher
Scientific).
3. Pierce™ Graphite Spin Columns (Thermo Fisher Scientific).
4. Pierce™ Strong Cation Exchange (SCX) Spin Column, Mini
(Thermo Fisher Scientific).
5. Microcentrifuge capable of spinning 1.7 mL microcentrifuge
tubes at 14,000 × g at 4 °C.
6. Microcentrifuge polypropylene 1.7 mL collection tubes.
7. Sonicating probe.
8. Vortex.
9. Fume Hood. Microplate Spectrophotometer reading 96-well
plate format at 595 nm.
10. Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific).
11. Dry Heating Block.
12. Vacuum Evaporator/Lyophilizer (e.g., SpeedVac™, Thermo
Fisher Scientific).
13. Ultimate® 3000 RSLCnanoLC system (Thermo Fisher
Scientific).
14. LTQ Orbitrap XL (Thermo Fisher Scientific).
15. SilicaTip™ Standard Coating Tubing OD/ID 360/20 μm
Tip, ID 10 μm, length 5 cm (New Objective).
16. Analytical column: Acclaim®PepMap100 75 μm × 50 cm,
nanoViper C18, 3 μm, 100 Å (Thermo Fisher Scientific).
17. Trap column: C18 PepMap, 300 μm ID × 5 mm, 5 μm particle
size, 100 Å pore size; Thermo Fisher Scientific).
18. Xcalibur software, version 2.0.7 (Thermo Fisher Scientific).
198 Michael Henry et al.

19. Proteome Discover 2.1 (Thermo Fisher Scientific) with
SEQUEST HT and phosphoRS 3.1 [25] and suitable
Cricetulus griseus protein database in fasta format [26].

2.2  Reagents Ensure that all solvents and water are LC-MS grade.
1. DL-Dithiothreitol (DTT).
2. Iodoacetamide.
3. Urea.
4. LC-MS grade water.
5. HEPES.
6. Sequencing-grade modified trypsin (e.g., Promega, Thermo
Fisher Scientific, Sigma Aldrich).
7. ProteaseMax™ Surfactant Trypsin Enhancer (Promega).
8. Trifluoroacetic acid (TFA): use a fume hood when preparing
solutions with this acid.
9. Acetonitrile (ACN).
10. Ammonium Bicarbonate.
11. Halt™ Protease Inhibitor Cocktail (Thermo Fisher Scientific).
12. Halt™ Phosphatase Inhibitor Cocktail (Thermo Fisher Scientific).
13. Ammonium Hydroxide (NH4OH) ACS Reagent (Sigma Aldrich).
14. Potassium Phosphate Monobasic (KH2PO4).
15. Potassium Chloride (KCl).

2.3  Cell Lysis, 1. 50 mM HEPES buffer: Add 1.19 g of HEPES to 100 mL of
Protein Extraction, LC-MS water.
and Protein 2. Lysis Buffer: 0.2% ProteaseMax solution in 8 M Urea, 50 mM
Quantification Ammonium Bicarbonate.
3. 1 mg of recombinant Chinese hamster ovary (CHO) cells
(approximately 1 × 107 cells).

2.4  Protein Digestion 1. 50 mM Ammonium Bicarbonate (must be freshly prepared

and used within 24 h).
2. 0.5 M DTT in 50 mM Ammonium Bicarbonate (must be
freshly prepared).
3. 0.55 M Iodoacetamide in 50 mM Ammonium Bicarbonate
(must be freshly prepared).
4. Trypsin solution: use sequence grade trypsin (see Note 1).
5. ProteaseMax™ Surfactant Enhancer (see Note 2).
 Phosphoproteomic Analysis of CHO Cells 199

Table 1
Overview of preparation of KC1 elution buffers

Elution buffer KCL (g) 10 mM KH2PO4 in 25% ACN (mL) KCl concentration (mM)
Elution buffer step 1 0.007 10  10
Elution buffer step 2 0.037 10  50
Elution buffer step 3 0.075 10 100
Elution buffer step 4 0.373 10 500

2.5  Peptide 1. Diluent solution: To prepare 1 L add 1.36 g of KH2PO4 and
Fractionation Using 250 mL of acetonitrile to 750 mL of LC-MS grade water and
Strong Cation adjust the pH to 3.0.
Exchange (SCX) Spin 2. Prepare the following 10 mL Potassium Chloride (KCl) elu-
Cartridges tion buffers as outlined in Table 1 using the 10 mM KH2PO4
in 25% acetonitrile pH 3.0 diluent.

2.6  Fe-NTA (IMAC) Wash buffers A and B must be prepared prior to starting the pro-
Phosphopeptide tocol. 300 μL of each wash buffer is required per spin column, i.e.,
Enrichment Buffer sample.
Preparation 1. Wash buffer A: Add 150 μL of the 2× wash buffer stock sup-
plied in the kit to 150 μL of LC-MS grade water.
2. Wash buffer B: Add 150 μL of the 2× wash buffer stock sup-
plied in the kit and 30 μL of acetonitrile to 120 μL of LC-MS
grade water.

2.7  TiO2 The following buffer preparations are for the enrichment of one to
Phosphopeptide six samples, if more samples are being processed scale up the buffer
Enrichment Buffer volumes accordingly or see manufacturer’s protocol.
Preparation 1. Buffer A: Add 401 μL of ACN and 2 μL of TFA to 100 μL of
LC-MS water.
2. Buffer B: Add 400 μL of 90% lactic acid (provided in the kit)
to 1 mL of Buffer A.
3. Elution Buffer 1: Add 20 μL of 30% Ammonium Hydroxide to
380 μL of LC-MS water.
4. Elution Buffer 2: Add 20 μL of Pyrrolidine (provided in the
kit) to 380 μL of LC-MS water.

2.8  Peptide 1. Column preparation buffer (1 M NH4OH): Add 0.688 mL of

Purification Ammonium hydroxide solution ACS reagent, 28.0–30.0% to
with Pierce™ Graphite 2.5 mL of LC-MS water and adjust the final volume of the
Spin Columns solution to 10 mL with LC-MS water. Store at 4 °C until
required.
200 Michael Henry et al.

2. Equilibration and rinse solution: 1% TFA. Add 100 μL of TFA

to 10 mL of LC-MS water.
3. Elution solution: 0.1% TFA, 70% ACN. Add 100 μL of TFA to
3 mL of LC-MS water containing 7 mL of ACN.

2.9  Reverse Phase 1. Solvent A: 2% ACN in LC-MS grade water containing 0.1%
Chromatography formic acid. Prepare 100 mL. Use a fume hood to prepare this
Buffers solution.
2. Solvent B: 80% ACN, 20% LC-MS grade water containing
0.8% formic acid. Prepare 100 mL.

3  Methods

3.1  Cell Lysis 1. Harvest cells and wash with prechilled 50 mM HEPES buffer
three times. Centrifuge at 1000 × g for 5 min and remove the
supernatant. Snap freeze the cell pellet in a microcentrifuge
tube using liquid nitrogen (take extreme caution when doing
this). Store at −80 °C until cell lysis is performed.
2. Lyse cell pellets corresponding to 1 × 107 cells with 1 mL of
lysis buffer containing 1× Halt™ Protease Inhibitor cocktail
and 1× Halt™ Protease Inhibitor cocktail. Suspend the cell
pellet in the lysis buffer and mix the solution by vigorous vor-
texing for 5 min.
3. Disrupt the cells using a sonicating probe while maintaining
the lysate at 4 °C.
4. Centrifuge the sample at 14,000 × g for 15 min at 4 °C. Collect
supernatant into a fresh microcentrifuge tube.
5. Determine protein concentration using a BCA assay according
to manufacturer’s instructions.
6. Transfer 500 μg of protein lysate aliquots into microcentrifuge
tubes and freeze at −80 °C until required.
7. Prior to digestion, dilute sample with 50 mM ammonium
bicarbonate to achieve a final urea concentration of 1 M.

3.2  In Solution 1. Adjust the volume of 500 μg protein lysate to 84 μL with
Protein Digestion 50 mM ammonium bicarbonate.
2. Add 1 μL of 0.5 M DTT and heat the lysate at 56 °C for
20 min to reduce the protein.
3. Add 2.7 μL of 0.55 M iodoacetamide and incubate at room
temperature in the dark for 20 min to allow for alkylation.
4. Add 12.5  μL of 1 μg/μL sequence-grade trypsin (1:40
enzyme:protein).
 Phosphoproteomic Analysis of CHO Cells 201

5. Digest at 37 °C for 3 h in a dry heating block under gentle

rotation.
6. Centrifuge at 14,000 × g at 20 °C for 20 min and transfer the
supernatant to a new microcentrifuge tube.
7. Add TFA to a final concentration of 0.1% to inactivate the
trypsin.
8. Proceed directly to peptide concentration and desalting
(Subheading 3.3) or store the sample at −80 °C until required.

3.3  Peptide 1. Remove the top and bottom caps from the graphite spin col-
Concentration umn (see Note 3) and place the column in a 1.7 mL collection
and Desalting tube. Centrifuge at 2000 × g for 1 min and discard the
with Graphite Spin flow-through.
Columns 2. Prime the Graphite spin column using 100 μL of 1 M NH4OH.
Centrifuge at 2000 × g for 1 min and discard the flow-
through.
3. Activate the graphite spin column using 100 μL of ACN.
Centrifuge at 2000 × g for 1 min and discard the flow-
through.
4. Apply the peptide sample from Subheading 3.2 to the graphite
resin and allow sample binding for 10 min with occasional
vortexing.
5. Centrifuge at 1000 × g for 3 min and discard the flow-
through.
6. Wash the graphite spin column using 100 μL of the rinse solu-
tion, centrifuge at 2000 × g for 1 min and discard rinse wash.
7. Repeat step 6.
8. Place the graphite spin column in a new collection tube and
elute the peptide sample using 100 μL of elution buffer into a
new microcentrifuge tube.
9. Repeat step 8 three more times to give a final volume of
400 μL.
10. Concentrate the peptide sample to dryness in a SpeedVac vac-
uum dryer.
11. Freeze peptide sample at −80 °C or proceed directly to Strong
Cation Exchange preparation.

3.4  Strong Cation 1. Condition a SCX spin column with 400 μL of 10 mM KH2PO4
Exchange (SCX) Using in 25% ACN pH 3.0 and centrifuge at 2000 × g for 5 min.
Pierce™ Mini Spin Discard the flow-through.
Columns 2. Resuspend the dried peptides from the final step of Subheading
3.4 in 200 μL of 10 mM KH2PO4 in 25% ACN pH 3.0. Apply
this to the conditioned SCX column and centrifuge at 2000 × g
for 5 min.
202 Michael Henry et al.

3. Add 200 μL of elution buffer step 1 (see Subheading 2.4) and

centrifuge at 2000 × g for 5 min and collect the flow-
through.
4. Repeat this procedure to elute the peptides in a step-wise man-
ner using elution buffer step 1 through to elution buffer step
4, collecting each flow-through generating five samples (initial
flow through, 10, 50, 100, and 500 mM salt fractions).
5. Eluted peptides are desalted and concentrated again using
Graphite spin columns (go to Subheading 3.3).
6. Following graphite spin column desalting and concentration,
the SCX fractions peptide samples can either be phosphopep-
tide enriched using Fe-NTA (IMAC) (go to Subheading 3.7)
or TiO2 (go to Subheading 3.8).

3.5  IMAC 1. Resuspend the dried sample from the SCX fractionation proto-
Phosphopeptide col in 200 μL of Binding buffer. Add the sample to a Fe-NTA
Enrichment spin column and incubate for 20 min at room temperature on
a rotator.
2. Remove the bottom cap from the column and place the col-
umn into a microcentrifuge tube and centrifuge at 1000 × g for
1 min. Transfer the column to a new tube.
3. Add 100 μL of wash buffer A to the column and pipette well
to mix contents. Centrifuge the tube at 1000 × g for 1 min.
Transfer column to a new tube.
4. Repeat step 3.
5. Add 100 μL of wash buffer B to the column and pipette well
to mix contents. Centrifuge the tube at 1000 × g for 1 min.
Transfer the column to a new tube.
6. Repeat step 5.
7. Equilibrate the column by adding 100 μL of LC-MS water to
the column and mix contents by pipetting. Centrifuge the col-
umn at 1000 × g for 1 min. Transfer the column to a new tube.
8. Elute the phosphopeptides by adding 50 μL of Elution buffer
to the column and incubate for 5 min at room temperature.
Centrifuge the column at 1000 × g for 1 min and retain the
eluate (flow-through).
9. Repeat step 8 and pool the elution fractions.
10. Acidify the elution fraction by adding 200 μL of 2.5% TFA to
the sample. Proceed to Subheading 3.3.

3.6  TiO2 1. Place a centrifuge column adaptor in a collection tube and

Phosphopeptide insert a TiO2 spin tip into the adaptor. Add 20 μL of Buffer A
Enrichment and centrifuge at 3000 × g for 2 min. Discard the flow-
through.
 Phosphoproteomic Analysis of CHO Cells 203

2. Add 20 μL of Buffer B and centrifuge at 3000 × g for 2 min.

Discard the flow-through.
3. Resuspend the dried sample from the SCX fractionation proto-
col in 150 μL of Buffer B. Add the sample to the spin tip and
centrifuge at 1000 × g for 10 min.
4. Collect the flow-through sample and reapply it to the spin tip
and centrifuge at 1000 × g for 10 min.
5. Wash the column by adding 20 μL of Buffer B and centrifug-
ing at 3000 × g for 2 min.
6. Wash the column by adding 20 μL of Buffer A and centrifuge
at 3000 × g for 2 min. Repeat this step twice so that the col-
umn has been washed three times.
7. Place the spin tip in a new collection tube and add 50 μL of
Elution Buffer 1. Centrifuge at 1000 × g for 5 min.
8. Using the same collection tube, add 50 μL of Elution Buffer 2.
Centrifuge at 1000 × g for 5 min.
9. Acidify the elution fraction by adding 100 μL of 2.5% TFA to
the sample. Proceed to Subheading 3.3.

3.7  LC-MS/MS 1. Resuspend the dried phosphopeptide samples in 25 μL of

Analysis LC-MS grade water with 0.1% TFA and 2% ACN.
2. Carry out nano LC–MS/MS analysis using, for example, an
Ultimate 3000 RSLCnano system coupled to a hybrid linear
ion trap/Orbitrap mass spectrometer. Use SilicaTip™ Standard
Coating Tubing as emitter tips for nano electrospray as the
formation of phosphopeptide metal ion complexes can cause a
mass spectrum signal intensity decrease of protonated phos-
phopeptides [27] (see Note 4).
3. Load each sample onto a C18 trap column. Desalt the sample
for 3 min using a flow rate of 25 μL/min in 0.1% TFA contain-
ing 2% ACN (Loading Buffer). Switch the trap column online
with the analytical column using a column oven at 35 °C and
elute the peptides with the following binary gradients of:
Mobile Phase Buffer A and Mobile phase buffer B: 0–25% sol-
vent B in 120 min and 25–50% solvent B in a further 60 min,
where solvent A consisted of 2% acetonitrile (ACN) and 0.1%
formic acid in water and solvent B consisted of 80% ACN and
0.08% formic acid in water. Set the column flow rate to
300 nL/min. Acquire the data with Xcalibur software.
4. Operate the LTQ Orbitrap XL in data-dependent neutral loss
mode and externally calibrate the instrument. Acquire survey
MS scans in the Orbitrap in the 400–1800 m/z range with the
resolution set to a value of 30,000 at m/z 400. Fragment up
to three of the most intense ions (1+, 2+, and 3+) per scan
using CID in the linear ion trap followed by a Neutral loss
204 Michael Henry et al.

MS3 event. MS3 is activated if a neutral loss of phosphoric acid

is detected indicated by a –98, −49, or −32.7 Da loss from
the parent precursor. Enable dynamic exclusion with a repeat
count of 1, repeat duration of 30 s, exclusion list size of 500
and exclusion duration of 40 s. Set the minimum signal to
3000. Collect all tandem mass spectra using a normalized col-
lision energy of 32%, an isolation window of 2 m/z with an
activation time of 30 ms.

3.8  Mass 1. Search MS spectra using Proteome Discoverer 2.1 against

Spectrometry Data Sequest HT with phosphoRS 3.1 against a suitable Cricetulus
Searches Griseus protein fasta database (see Note 5).
for Phosphorylation 2. The processing workflow used through Proteome Discoverer
Analysis 2.1 is shown in Fig. 1 (see Note 6).

Fig. 1 Processing workflow for MS2 and MS3 data using Proteome Discoverer 2.1
 Phosphoproteomic Analysis of CHO Cells 205

3. Scan Event Filter (11) and (12) (see Fig. 1) selects for MS2
and MS3 data respectively.
4. From Scan Event Filter (11) into Search algorithm SEQUEST
HT (2) (see Fig. 1), apply the following search parameters for
all MS2 data for protein identification: (1) set peptide mass
tolerance to 20 ppm, (2) set MS/MS mass tolerance to 0.6 Da,
(3) allow up to two missed cleavages, (4) set carbamidometh-
ylation as a fixed modification (i.e., Carboxymethyl (C) means
that all calculations will use 161 Da as the mass of cysteine),
(5) set methionine oxidation as a variable modification (where
the methionine residue may or may not be oxidized (Met
+16 Da)), and (6) set serine, threonine, and tyrosine phos-
phorylation as variable modifications (where the serine, threo-
nine, and tyrosine residues may have the addition of a phosphate
group HPO3 (Ser +80 Da, Thr +80 Da, Tyr +80 Da)) (see
Fig. 2a and 2b).
5. From Scan Event Filter (12) into Search algorithm SEQUEST
HT (15) (see Fig. 1), set the following search parameters for all
MS3 (neutral loss) data for protein identification: (1) set pep-
tide mass tolerance to 0.6 Da, (2) set MS/MS mass tolerance
to 0.6 Da, (3) allow up to two missed cleavages, (4) set carb-
amidomethylation as a fixed modification, (5) set methionine
oxidation as variable modification, (6) set serine, threonine,
and tyrosine phosphorylation as variable modifications, and
(7) set neutral loss of phosphoric acid H3PO4 (−98 Da) from
phosphorylated serine, threonine, and tyrosine phosphoryla-
tion as variable modifications (see Fig. 2c).
6. Process the output from SEQUEST through Percolator
(a semi-supervised learning and decoy database search strategy
to distinguish between correct and incorrect identifications
accepting a 1% FDR) [28].
7. Accept only peptides with XCorr scores >1.5 for singly charged
ions, >2.0 for doubly charged ions, and >3.0 for triply charged
ions.
8. Determine the phosphorylation site location in the peptide
sequence through phosphoRS (ptmRS) in Proteome Disco­
verer 2.1 [25]. Apply a site probability cut-off score of 75% or
greater for S, T, or Y amino acids [25, 29, 30].

4  Notes

1. Use sequencing-grade or mass spectrometry grade trypsin.

Reconstitute the lyophilized trypsin in 50 mM acetic acid to
give a final concentration of 1 μg/μL and store aliquots at
−20 °C until required. In our lab we use Promega’s Trypsin
206 Michael Henry et al.

Fig. 2 (a) MS/MS of DGQVINTSQHDDLE showing sequence ladder matching glutamine (Q) to serine (S) and
has a mass difference of 87 Da confirming unmodified serine. (b) MS/MS of DGQVINTS*QHDDL showing a
sequence ladder matching glutamine (Q) to serine (S) and has a mass difference of 167 Da (addition of phos-
phate group (HPO3) confirming phosphorylated serine (+80 ΔM on Serine). (c) MS/MS/MS of DGQVINTS*QHDDL
showing a neutral loss event on serine where the sequence ladder matching glutamine (Q) to serine (S) has a
mass ­difference of −18 Da which corresponds to the loss of phosphoric acid from the phosphorylated residue
to produce a di-dehydroamino acid

Gold Mass Spectrometry grade or Pierce™ Trypsin Protease

MS grade for protein digestion.
2. Add 100 μL of 50 mM ammonium bicarbonate to a 1 mg vial
of ProteaseMax™ Surfactant to a give a 1% solution. Store ali-
quots at −20 °C until required. Ammonium Bicarbonate must
be freshly prepared and used within 24 h.
3. For peptide purification use graphite spin columns instead of
C18 as graphite binds hydrophilic peptides more efficiently
than C18.
 Phosphoproteomic Analysis of CHO Cells 207

4. The emitter tip connecting the LC to the MS should not be

stainless steel. Applying a voltage across a stainless emitter tip
will cause irreversible binding of phosphopeptides.
5. For protein/peptides identification using Chinese Hamster
Ovary cells, we recommend using either protein fasta databases
from Bielefeld BOKU CHO expressed Protein Database
(BBCHO) [26] or from Cricetulus griseus fasta sequences
available from https://www.ncbi.nlm.nih.gov/protein.
6. Using spectrum selector filter in Proteome Discoverer to sepa-
rate MS2 and MS3 data allows for significantly shorter search
times as MS2 streams can be searched with variable phosphor-
ylation modifications on S, T, and Y, while MS3 streams can be
searched with variable phosphorylation modifications on S, T,
and Y along with dehydration as a variable modification on S,
T, and Y (see Fig. 1).

Acknowledgments

The authors acknowledge funding from Science Foundation

Ireland (grant ref. 13/1A/1841) and the Horizon 2020 Marie
Curie ITN programme – eCHO systems (grant ref.: 642663).

References
1. Walsh G (2014) Biopharmaceutical bench-
marks. Nat Biotechnol 32(10):992–1000
2. Wurm FM (2004) Production of recombinant
protein therapeutics in cultivated mammalian
cells. Nat Biotechnol 22(11):1393–1398
3. Altamirano C, Paredes C, Cairó JJ, Gòdia F
(2000) Improvement of CHO cell culture
medium formulation: simultaneous substitu-
tion of glucose and glutamine. Biotechnol
Prog 16:69–75
4. Prentice HL, Ehrenfels BN, Sisk WP (2007)
Improving performance of mammalian cells in
fed-batch processes through “bioreactor evolu-
tion”. Biotechnol Prog 23:458–464
5. Kildegaard HF, Baycin-Hizal D, Lewis NE,
Betenbaugh MJ (2013) The emerging CHO
systems biology era: harnessing the 'omics rev-
olution for biotechnology. Curr Opin
Biotechnol 24(6):1102–1107
6. Gutierrez JM, Lewis NE (2015) Optimizing
eukaryotic cell hosts for protein production
through systems biotechnology and genome-­
scale modeling. Biotechnol J 10(7):939–949
1 0. Kaas CS, Kristensen C, Betenbaugh MJ,
7. Lewis N, Liu X, Li Y, Nagarajan H, Yerganian
G, O'Brien E, Bordbar A, Roth AM,
Rosenbloom J, Bian C, Xie M, Chen W, Li N,
Baycin-Hizal D, Latif H, Forster J, Betenbaugh
MJ, Famili I, Xu X, Wang J, Palsson BO
(2013) Genomic landscapes of Chinese ham-
ster ovary cell lines as revealed by the Cricetulus
griseus draft genome. Nat Biotechnol 31(8):
759–765
8. Xu X, Nagarajan H, Lewis NE, Pan S, Cai Z,
Liu X, Chen W, Xie M, Wang W, Hammond S,
Andersen MR, Neff N, Passarelli B, Koh W,
Fan HC, Wang J, Gui Y, Lee KH, Betenbaugh
MJ, Quake SR, Famil I, Palsson BO, Wang
J (2011) The genomic sequence of the Chinese
hamster ovary (CHO)-K1 cell line. Nat
Biotechnol 29(8):735–741
9. Brinkrolf K, Rupp O, Laux H, Kollin F, Ernst
W, Linke B, Kofler R, Romand S, Hesse F,
Budach WE, Galosy S, Muller D, Noll T,
Wienberg J, Jostock T, Leonard M, Grillari J,
Tauch A, Goesmann A, Helk B, Mott JE,
Puhler A, Borth N (2013) Chinese hamster
genome sequenced from sorted chromosomes.
Nat Biotechnol 31(8):694–695
Andersen MR (2015) Sequencing the CHO
DXB11 genome reveals regional variations in
genomic stability and haploidy. BMC Genomics
16:160
208 Michael Henry et al.
11. Baycin-Hizal D, Tabb DL, Chaerkady R, Chen
L, Lewis NE, Nagarajan H, Sarkaria V, Kumar
A, Wolozny D, Colao J, Jacobson E, Tian Y,
O'Meally RN, Krag SS, Cole RN, Palsson BO,
Zhang H, Betenbaugh M (2012) Proteomic
analysis of Chinese hamster ovary cells.
J Proteome Res 11(11):5265–5276
12. Guo Y, Xiao P, Lei S, Deng F, Xiao GG, Liu Y,
Chen X, Li L, Wu S, Chen Y, Jiang H, Tan L,
Xie J, Zhu X, Liang S, Deng H (2008) How
is mRNA expression predictive for protein
expression? A correlation study on human cir-
culating monocytes. Acta Biochim Biophys Sin
40(5):426–436
13. Farrell A, McLoughlin N, Milne JJ, Marison
IW, Bones J (2014) Application of multi-omics
techniques for bioprocess design and optimiza-
tion in Chinese hamster ovary cells. J Proteome
Res 13(7):3144–3159
14. Choudhary C, Mann M (2010) Decoding sig-
nalling networks by mass spectrometry-based
proteomics. Nat Rev Mol Cell Biol 11(6):
427–439
15. Cohen P (2001) The role of protein phosphor-
ylation in human health and disease. The Sir
Hans Krebs Medal Lecture. Eur J Biochem
268(19):5001–5010
16. Olsen JV, Vermeulen M, Santamaria A, Kumar
C, Miller ML, Jensen LJ, Gnad F, Cox J, Jensen
TS, Nigg EA, Brunak S, Mann M (2010)
Quantitative phosphoproteomics reveals wide-
spread full phosphorylation site occupancy dur-
ing mitosis. Sci Signal 3(104):ra3
17. Dephoure N, Gould KL, Gygi SP, Kellogg DR
(2013) Mapping and analysis of phosphoryla-
tion sites: a quick guide for cell biologists. Mol
Biol Cell 24(5):535–542
18. Martini M, De Santis MC, Braccini L, Gulluni
F, Hirsch E (2014) PI3K/AKT signaling path-
way and cancer: an updated review. Ann Med
5:1–12
19. Hopkins BD, Hodakoski C, Barrows D, Mense
SM, Parsons RE (2014) PTEN function: the
long and the short of it. Trends Biochem Sci
39(4):183–190
20. Farrell AS, Allen-Petersen B, Daniel CJ, Wang X,
Wang Z, Rodriguez S, Impey S, Oddo J, Vitek
MP, Lopez C, Christensen DJ, Sheppard B, Sears
RC (2014) Targeting inhibitors of the tumor
suppressor PP2A for the treatment of pancreatic
cancer. Mol Cancer Res 12(6):924–939
21. Whitmarsh AJ, Davis RJ (2000) Regulation of
transcription factor function by phosphoryla-
tion. Cell Mol Life Sci 57(8–9):1172–1183
22. Simon GM, Cravatt BF (2008) Challenges for
the 'chemical-systems' biologist. Nat Chem
Biol 4(11):639–642
23. Angel TE, Aryal UK, Hengel SM, Baker ES,
Kelly RT, Robinson EW, Smith RD (2012)
Mass spectrometry-based proteomics: existing
capabilities and future directions. Chem Soc
Rev 41(10):3912–3928
24. Thingholm TE, Jensen ON, Robinson PJ,
Larsen MR (2008) SIMAC (sequential elution
from IMAC), a phosphoproteomics strategy
for the rapid separation of monophosphory-
lated from multiply phosphorylated peptides.
Mol Cell Proteomics 7(4):661–671
25. Taus T, Köcher T, Pichler P, Paschke C,
Schmidt A, Henrich C, Mechtler K (2011)
Universal and confident phosphorylation site
localization using phosphoRS. J Proteome Res
10(12):5354–5362
26. Meleady P, Hoffrogge R, Henry M, Rupp O,
Bort JH, Clarke C, Brinkrolf K, Kelly S, Müller
B, Doolan P, Hackl M, Beckmann TF, Noll T,
Grillari J, Barron N, Pühler A, Clynes M, Borth
N (2012) Utilization and evaluation of CHO-­
specific sequence databases for mass spect­
rometry based proteomics. Biotechnol Bioeng
109(6):1386–1394
27. Liu S, Zhang C, Campbell JL, Zhang H, Yeung
KK, Han VK, Lajoie GA (2005) Formation of
phosphopeptide-metal ion complexes in liquid
chromatography/electrospray mass spectrom-
etry and their influence on phosphopeptide
detection. Rapid Commun Mass Spectrom
19(19):2747–2756
28. Spivak M, Weston J, Bottou L, Käll L, Noble
WS (2009) Improvements to the percolator
algorithm for peptide identification from shot-
gun proteomics data sets. J Proteome Res
8(7):3737–3745
29. Alpert AJ, Hudecz O, Mechtler K (2015)
Anion-exchange chromatography of phospho-
peptides: weak anion exchange versus strong
anion exchange and anion-exchange chroma-
tography versus electrostatic repulsion-­
hydrophilic interaction chromatography. Anal
Chem 87(9):4704–4711
30. Roitinger E, Hofer M, Köcher T, Pichler P,
Novatchkova M, Yang J, Schlögelhofer P,
Mechtler K (2015) Quantitative phosphopro-
teomics of the ataxia telangiectasia-mutated
(ATM) and ataxia telangiectasia-mutated and
rad3-related (ATR) dependent DNA damage
response in Arabidopsis thaliana. Mol Cell
Proteomics 14(3):556–571
 Chapter 14

Engineer Medium and Feed for Modulating N-Glycosylation

of Recombinant Protein Production in CHO Cell Culture
Yuzhou Fan, Helene Faustrup Kildegaard, and Mikael Rørdam Andersen

Abstract
Chinese hamster ovary (CHO) cells have become the primary expression system for the production of
complex recombinant proteins due to their long-term success in industrial scale production and generating
appropriate protein N-glycans similar to that of humans. Control and optimization of protein
N-glycosylation is crucial, as the structure of N-glycans can largely influence both biological and physico-
chemical properties of recombinant proteins. Protein N-glycosylation in CHO cell culture can be con-
trolled and tuned by engineering medium, feed, culture process, as well as genetic elements of the cell. In
this chapter, we will focus on how to carry out experiments for N-glycosylation modulation through
medium and feed optimization. The workflow and typical methods involved in the experiment process will
be presented.

Key words Chinese hamster ovary cells, N-Glycosylation, Medium and feed optimization, Fed-batch
culture

1  Introduction

Chinese hamster ovary (CHO) cell is one of the most predominant

workhorses for recombinant protein production in the pharmaceu-
tical industry. It has a proven track record in industrial scale-up
applications and in generating human-like protein glycosylation.
N-glycosylation of recombinant proteins, especially of those as
drug substances, is highly important in drug development and
approval, as it will largely affect protein stability, efficacy, clearance
rate, and immunogenicity [1–3]. It is therefore highly desired to
be able to control and modulate N-glycosylation of recombinant
therapeutic proteins in CHO cells.
Modifying medium and feed, optimizing culture process, and
engineering genetic elements are three major approaches for fine-­
tuning N-glycosylation patterns of recombinant proteins [4].
Modifying medium and feed has its advantage of achieving
successful N-glycosylation modulation in a relatively fast and easy

Paula Meleady (ed.), Heterologous Protein Production in CHO Cells: Methods and Protocols, Methods in Molecular Biology,
vol. 1603, DOI 10.1007/978-1-4939-6972-2_14, © Springer Science+Business Media LLC 2017

209
210 Yuzhou Fan et al.

process independent of cell line genetics. It is therefore often

favored to be applied in the later stage of industrial bioprocess
development when a production cell line has already been devel-
oped [5]. Classical culture components, for example glucose [6–11],
galactose [12–16], amino acids [8, 10, 17–19], NH4+ [10, 17, 18,
20], and many feed additives, such as manganese [14–16, 19, 21,
22], sodium butyrate [23–26], and nucleotide and nucleotide
sugar precursors [12, 14–16, 27–30], have been shown to be able
to modulate protein N-glycosylation in CHO cell culture.
Additionally, impact of process parameters including dissolved
oxygen [31–34], CO2 level [35, 36], pH [26, 37–39], tempera-
ture [26, 40–42], agitation rate [43], and culture duration [7, 11,
44, 45] on N-glycosylation has also been well studied in CHO
cell culture. However, process engineering without compromising
process robustness usually needs to be operated within a predefined
operating window, and thus it could be less effective at changing
N-glycan structures than medium and feed engineering. It is also
difficult to control N-glycosylation by using subtle changes in pro-
cess parameters, as most of the glycosylation responses to culture
parameter modulations remain uncharacterized. Furthermore, in
most cases, the response depends on cell line, equipment, and facil-
ity used [46].
Genetic engineering, although it has shown many promising
results in customizing protein N-glycosylation [47–63], is a time-­
consuming process and due to regulatory issues, it is normally only
applicable at the early stages of bioprocess development (i.e., pro-
duction cell line generation and selection).
In this chapter, we will focus on the typical workflow and
methods used for controlling and modulating recombinant pro-
tein N-glycosylation in CHO cell culture through medium and
feed optimization (Fig. 1).

2  Materials

2.1  Equipment 1. An automated hemocytometer, e.g., nucleocounter NC-200

(Chemometec, Denmark) or Vi-CELL (Beckman Coulter,
Brea, CA). A reagent kit containing a viability dye, e.g., Trypan
Blue, DAPI is supplied with relevant equipment.
2. Bioreactors (1 L maximum working volume) coupled with
sensor modules (with pH and O2 sensors), gassing system,
multipump module, and temperature and agitation control
unit (Dasgip, Germany).
3. Bioprofile 100plus (Nova BioMedical, Waltham, MA).
4. Octet QK384 equipped with specific protein biosensors
(ForteBio, Menlo Park, CA).
 Medium and Feed Related N-Glyco-Engineering 211

Fig. 1 Typical workflow and methods used for modulating recombinant protein N-glycosylation in CHO cell
culture through medium and feed engineering

5. NanoDrop ND-1000 (Thermo Scientific, Waltham, MA).

6. Dionex Ultimate 3000 RSLC System (Dionex, Sunnyvale, CA).
7. Ultimate 3000 RS fluorescence detector (Dionex, Sunnyvale, CA).
8. ACQUITY UPLC BEH Glycan, 130Å, 1.7 μm, 2.1 mm ×
150 mm (Waters, Milford, MA).
9. Optionally, the following equipment may also be considered.
212 Yuzhou Fan et al.

10.
Agilent 2100 Bioanalyzer (Agilent Technologies, Santa
Clara, CA).
11. Total RNA nano chip (Agilent Technologies).
12. Illumina Hiseq 2000 system (Illumina, San Diego, CA).

2.2  Cell Culture 1. Recombinant protein production cell line (1–2 × 107 cells/
vial) that is cryopreserved in culture medium with 5–10%
DMSO (Sigma, St. Louis, MO) from a working cell bank
stored at −150 °C or lower.
2. Culture medium in this experiment setup is a serum-free and
chemically defined medium. Depending on cell line and culture
process, different media can be used. Here are examples of some
commercially available culture media that are typically used for
CHO cell culture: CD CHO, CD OptiCHO™, CD FortiCHO™
(Life Technologies, Carlsbad, CA), Ex-Cell™ CD CHO
(Sigma), ProCHO™5, PowerCHO-2 CD (Lonza, Switzerland),
BalanCD™ CHO Growth A (Irvine Scientific, Santa Ana, CA),
Cellvento™ CHO-100 (Millipore, Billerica, MA), and
CDM4CHO SFM4CHO (GE Healthcare, Fairfield, CA),
ActiCHO™ P (GE Healthcare). If the medium is l-glutamine
free, additional l-glutamine (Life Technologies) is normally sup-
plemented (usually add up to 4–8 mM) for non-glutamine syn-
thetase-engineered cell lines. Optionally, 1% MEM Non-­Essential
Amino Acids (100×) (Life Technologies) and 0.1–1% Anti-
clumping agent (Life Technologies) can also be added to the
­culture medium.
3. Base: autoclaved 0.5–2 M NaOH or NaHCO3 or Na2CO3
(Sigma Aldrich).
4. 45% d-(+)-Glucose solution (Sigma Aldrich).
5. 10% ADCF antifoam-irradiated Solution (GE Healthcare).
6. Feed supplement medium (various commercial products avail-
able): feed supplement medium provides cell culture nutrients
such as lipids, amino acids, vitamins, and growth factors.
7. Feed additives: feed additives can be dissolved in culture
medium and/or feed supplement medium to an appropriate
concentration. The amount of feed additives in culture medium,
feed supplement medium, and the final concentration achieved
in the cell culture can be continuously optimized according to
the results of the experiment.

2.3  Analytical 1. Regeneration buffer: 10 mM Glycine-HCl, pH = 1.5.

Techniques 2. Formulation buffer: A solution of 10 mM Citrate and 150 mM
NaCl at pH = 6.0.
3. 0.9% w/v aqueous NaCl solution.
 Medium and Feed Related N-Glyco-Engineering 213

4. Neutralization solution: culture medium supplemented with

10% (v/v) feed supplement medium.
5. The hydration solution: culture medium supplemented with
10% (v/v) feed supplement medium.
6. Solvent A: 100% acetonitrile.
7. Solvent B: 100 mM ammonium formate at pH = 4.5
8. 50% v/v aqueous acetonitrile.
9. MabSelect SuRe protein A resin slurry (GE Healthcare).
10. 1× PBS solution.
11. 0.1 M citrate solution at pH = 3.5.
12. GlykoPrep® InstantABTM kit (Prozyme, Hayward, CA).
13. TRIzol (Life Technologies).
14. RNeasy Cleanup kit (Qiagen, Hilden, Germany).
15. SDS-lysis buffer: 2% SDS (w/v), 1 mM EDTA, and 0.1 mM
phenylmethylsulfonyl fluoride, adjust pH = 8 using triethylam-
monium bicarbonate (Sigma Aldrich).
16. BCA protein assay kit (Thermo Scientific).
17. 4.5 mM Tris-(2-carboxyethyl) phosphine solution.
18. 8.3 mM methyl methanethiosulfonate solution.
19. 9 M Sequanal grade urea (Thermo Scientific).
20. 0.22 μm low protein-binding filter (Millipore).
21. 10 kDa cutoff 0.5 mL filters (Amicon).
22. iTRAQ® Reagent—8PLEX One Assay Kit (AB Sciex).

3  Methods

3.1  Thaw Cells 1. Thaw a vial of cells from working cell bank at 37 °C and trans-
and Cell Culture fer the cells to a 125 mL shake flask.
Maintenance 2. Wash the cells with 20 mL of pre-warmed culture medium at
(See Note 1) 37 °C. Spin down the cells at 200 × g, and discard the super-
natant to remove DMSO in the culture.
3. Add 25 mL pre-warmed culture medium to the shake flask and
maintain the culture at 37 °C, 5% CO2, 150 rpm in a humidi-
fied incubator.
4. Perform cell count and viability measurement using a hemocy-
tometer after 1 h.
5. Maintain cell in shake flasks with a 2 or 3-day passage for 1–3
weeks before starting the fed-batch culture. Perform cell count
and viability measurement before each passage (see Note 2).
Seed 0.5 × 106 viable cells/mL or 0.3 × 106 viable cells/mL
before a 2-day or a 3-day passage, respectively.
214 Yuzhou Fan et al.

6. Expand the cell culture in shake flasks until it is possible to

inoculate fed-batch cultivation in bioreactors with a desired
viable cell density.

3.2  Parameters Bioreactor systems should be assembled according to the manufac-

for Fed-Batch Culture turer’s instructions and autoclaved properly. All pumps and sensors
in Bioreactors should be well calibrated before starting fed-batch culture in the
bioreactors.
The fed-batch production is controlled in terms of the follow-
ing parameters:
1. Culture medium: the same culture medium as used in cell cul-
ture maintenance is used. Feed additives can be supplemented
to achieve a desired concentration.
2. Use a seeding density of 4 × 105 viable cells/mL (see Note 3).
3. Maintain the temperature at 37 °C along the culture
(see Note 4).
4. Maintain the pH at 7.0 ± 0.2 by the sparging of CO2 or the
addition of base (see Note 5).
5. Set the agitation speed to 150 rpm (see Note 6).
6. Set the dissolved oxygen (DO) at 30% (see Note 7).
7. Maintain the flow rate at 600 mL/h (see Note 7).
8. Set the initial working volume to 420 mL (see Note 8). This
can be done for example by adding about 370–330 mL culture
medium and then inoculating about 50–90 mL cell culture
from the preculture to the bioreactor.
9. Antifoam (optional): add the appropriate amount of 10%
ADCF antifoam to reduce formed foam in the bioreactor
system.
10. Feeding with supplement medium and feed additives: Add the
feed to the culture once a day or every second day. Based on
the frequency of feeding, the volume of feed added every time
is normally about 5–20% of the initial culture volume.
11. Feeding with glucose: spike glucose into the bioreactor to
obtain 6–12 mM every time the level decreases below 3 mM in
the culture (see Note 9).
12. Feeding with glutamine: spike glutamine into the bioreactor to
obtain 4–6 mM every time the level decreases below 2 mM in
the culture (see Note 9).
13. End of the fed-batch culture: The culture is normally harvested
before the viability becomes lower than 60%. Typical fed-­batch
culture duration is around 12–16 days.

3.3  Sampling Sampling is typically carried along the fed-batch process to ­monitor
Strategies cell culture physiology (e.g., growth, metabolism), product quan-
 Medium and Feed Related N-Glyco-Engineering 215

tity, and quality (e.g., titer and N-glycosylation), and even to

obtain complex omics information (e.g., RNA-seq, proteomics,
intracellular metabolites).
1. Sampling for monitoring cell culture physiology is often car-
ried out once a day during the entire cultivation process.
0.6–1 mL/sample is needed for cell count and viability
­measurement using a Nucleocounter or Vi-CELL. 1.5 mL
supernatant of cell culture per sample is needed for measuring
extracellular metabolites using Bioprofile 100plus.
2. Sampling for product titer and N-glycosylation analysis can be
carried out only at the end of fed-batch process or along with
the process with desired frequency. The sampling volume for
measuring product titer is both product and assay dependent.
For example, for IgG-producing cell lines, sampling 100–
500  μL cell culture is needed for measuring IgG titer using
Octet QK384 equipped with Protein A biosensors. The sam-
pling volume for N-glycosylation analysis is often large (10–
15 mL), and therefore it is necessary to plan carefully in
advance to avoid oversampling. It is critical that sampling only
takes a small percentage of the culture (ideally 1% or less), to
avoid that culture conditions (in particular aeration) are not
changed by the sampling.
3. The equivalent volumes of culture for 5 × 106 cells (RNA-seq
or proteomic analysis) and 1 × 107 cells (intracellular metabo-
lites analysis) are usually needed when sampling for complex
omics analysis. The sampling time could be representative time
points at exponential growth phase, stationary (production)
phase, or death phase, which can be designed according to
specific research question.

3.4  Analytical Samples from cell culture can be further processed using various
Techniques analytical techniques to obtain relevant data that support
N-glycosylation modulation. Here, some of the most commonly
used analytical methods are shown. However, many of them have
detailed procedures themselves, as described in relevant manufac-
turer’s instructions or publications; therefore, only brief descrip-
tions are presented here to give readers an overview of how these
analytical methods work.

3.4.1  Cell Count 1. Load sufficient amount of cell culture in sample vial (for Vi-­
and Viability Analysis CELL) or Via1-Cassette (for nucleocounter NC-200).
2. Carry out total cell count and dead cell count using Vi-CELL
or Via1-Cassette. Viable cell count and viability can also be
calculated automatically.

3.4.2  Product Titer 1. High-throughput (96-well format) quantitation of label-free

Analysis recombinant protein (e.g., monoclonal antibody, erythropoi-
216 Yuzhou Fan et al.

etin, Factors VII, VIII, IX, and X) in supernatant samples from

cell culture can be performed by direct measurement of the
binding rate of the recombinant protein in question to a cor-
responding protein sensor using Octet QK384, a biolayer
interferometry-­based system.
2. Supernatant samples are diluted properly with culture medium
(typically 5–10 times) to ensure the concentrations of recom-
binant protein in the samples are within the measurable range
of the equipment.
3. A typical setup for a full 96-well black, flat-bottom plate sam-
ple plate (12 column × 8 well):
For column 1–9: add 200 μL of each sample per well (nor-
mally 70 unknown diluted samples and 2 control samples are
added. If fewer samples are to be measured, fill up remaining
wells with 200 μL of 1× PBS).
For column 10: add 200 μL of each standard per well. The
standards should be prepared from purified recombinant pro-
tein (same or very similar to the recombinant protein in the
samples) with known concentration. It should be diluted
properly in series (e.g., for IgG standards, they should be
diluted in the range of 25 ng/mL to 300 μg/mL) with the
same or very similar culture medium as in the samples.
For column 11: add 200 μL of regeneration buffer per well to
all 8 wells.
For column 12: add 200 μL of neutralization solution, Set
instrument assay temperature (i.e., plate temperature) to
30 °C. Two sets of eight sensors, which are enough for mea-
suring all samples, controls, and standards located in columns
1–10, are hydrated in a hydration tray in a 96-well plate with
200 μL of hydration solution in the wells for 10 min at 30 °C
before running the assay.
4. Recombinant protein quantification can be then done based
on a calculated standard curve (measured binding rate of the
standards versus concentrations of the standards).

3.4.3  Extracellular 1. Extracellular metabolites including glucose, lactate, glutamine,

Metabolic Profiles glutamate, NH4+, Na+, K+, pH, osmolality can be quantified
from supernatant samples using Bioprofile 100plus.
2. Free amino acid analysis is sometimes used in media and feed
optimization, especially when amino acid concentrations are
altered. Several HPLC-based methods are available for quan­
tifying the concentrations of free amino acids in cell culture
[10, 64] as well as multiple commercial services.
 Medium and Feed Related N-Glyco-Engineering 217

3.4.4  Intracellular Intracellular nucleotide sugar quantifications offer the opportunity

Nucleotide Sugar of accessing N-glycosylation at the level of building blocks of
Quantification N-glycans. In brief, it can be carried out as follows:
1. Pellet 1× 107 cells (0 °C, 1000 × g, 1 min) from cell culture
samples and wash two times with 2 mL ice-cold 0.9% w/v
aqueous NaCl.
2. Flash freeze the cell pellets in liquid nitrogen and store at
−80 °C before acetonitrile extraction.
3. Resuspend and incubate the frozen cell pellets in ice-cold 50%
v/v aqueous acetonitrile on ice for 10 min.
4. Spin at 0 °C, 18,000 × g, for 5 min and collect supernatant
into a new 1.5 mL tube.
5. Dry the supernatant in a SpeedVac, resuspend in 240 μL water,
and store at −80 °C.
6. Run HPLC-based high-performance anion-exchange (HPAEC)
analysis as described in [65]. HPAEC method although requires
harsh buffer conditions (high pH), provides high resolution
with few unresolved peaks.

3.4.5  N-Glycosylation N-Glycosylation profiling of recombinant protein consists of

Profiles four major steps: (1) purify the recombinant protein, (2) cleave off
N-glycans from the purified protein by enzyme digestion, (3) label
the N-glycan with a fluorescent dye and cleanup, (4) separate and
quantify the labeled N-glycans using a HPLC coupled with fluo-
rescence detector. Here, we present a standard procedure of
N-glycosylation analysis using IgG as a model protein:
1. Samples from cell culture are centrifuged at 4500 × g for
20 min.
2. Filter the supernatant through a 0.22 μm low protein-binding
filter and then apply onto a MabSelect SuRe ProteinA column,
into which 200 μL of MabSelect SuRe protein A resin slurry is
packed.
3. Equilibrate the column with 2 mL of 1× PBS.
4. Load the filtered supernatant into the column and repeat for at
least two times to ensure sufficient binding of IgG to the
column.
5. Wash the column with 4 mL of 1× PBS.
6. Elute IgG by 500 μL of 0.1 M citrate pH = 3.5.
7. Apply the elution immediately on a NAP-5 column, which has
pre-equilibrated by a formulation buffer containing 10 mM
citrate and 150 mM NaCl pH = 6.0.
8. Elute purified IgG from NAP-5 to a 1.5 mL microcentrifuge
tube by the addition of 1 mL of the formulation buffer.
218 Yuzhou Fan et al.

9. Quantify purified IgG concentration using a NanoDrop

ND-1000.
10. Subject 50  μg of purified IgG to GlykoPrep® InstantAB™ kit
according to the manufacturer’s instructions. Thereby, N-­glycans
cleaved from IgG are labeled.
11. Labeled glycans are dissolved in 50% v/v aqueous acetonitrile.
12. A normal phase HPLC analysis is carried out using Dionex
Ultimate 3000 RSLC System coupled with Ultimate 3000 RS
fluorescence detector. 5 μL of the prepared sample was injected
into the system. HPLC parameters:
Stationary phase: ACQUITY UPLC BEH Glycan 1.7 μm,
2.1 × 150 mm column. Mobile phase: 100% acetonitrile as
solvent A and 100 mM ammonium formate with pH = 4.5 as
solvent B.
Gradient and flow rate: T0 min = 25% B and flow rate = 0.5 min/
mL, T5 min = 25% B and flow rate = 0.5 min/mL, T51.5 min = 40%
B and flow rate = 0.5 min/mL, T53 min = 100% B and flow rate
= 0.25 min/mL, T58 min = 100% B and flow rate = 0.25 min/
mL, T60 min = 25% B and flow rate = 0.5 min/mL, T70 min = 25%
B and flow rate = 0.5 min/mL, T75 min = 25% B and flow
rate = 0.5 min/mL.
Fluorescence detection: excitation = 278 nm and emission =
344 nm.
Peak assignment is done according to retention time of known
standards. Chromeleon software is used to analyze
chromatograms.

3.4.6  RNA-Seq Method for RNA sequencing of samples from CHO cell culture
and data analysis pipeline has been well developed [66]. In brief,
the major steps include:
1. RNA from 5 × 106 cells is extracted using TRIzol and RNeasy
Cleanup kit.
2. RNA concentration is quantified using NanoDrop ND-1000.
3. RNA integrity is analyzed using total RNA nano chip on
Agilent 2100 Bioanalyzer.
4. cDNA library generation is performed using TruSeq RNA
sample Preparation Kit v2.
5. Paired-end sequencing using Illumina Hiseq 2000 system.
6. Sequencing data treatment includes: (1) adaptamers removal
and low quality ends trimming, using FASTX Toolkit, (2)
quality check by FASTQC, (3) align the reads to CHO-K1
genome using Tophat2, (4) count reads using HTseq.

3.4.7  Proteomics Method for proteomics analysis in CHO cell culture is reported
Analysis in [57]. Briefly, the key procedure is as follows:
 Medium and Feed Related N-Glyco-Engineering 219

1. Pellet 5 × 106 cells (0 °C, 1000 × g, 1 min) from cell culture

samples and wash two times with 1× PBS (Life Technologies).
2. Flash freeze the cell pellets in liquid nitrogen and store at
−80 °C.
3. Lyse cells in SDS-lysis buffer in combination with 3× 30 s soni-
cation pulses on ice.
4. Measure total protein concentration in the lysate using the
BCA protein assay kit.
5. Reduce lysates in 4.5 mM Tris-(2-carboxyethyl) phosphine at
60 °C for 1 h.
6. Alkylate lysates in 8.3 mM methyl methanethiosulfonate at
room temperature in the dark for 30 min using a filter-aided
sample preparation (FASP) described by Wisniewski et al. [67].
7. Dilute 90 μg of treated protein with 9 M sequanal grade urea,
incubate at room temperature in the dark for 1 h.
8. Remove <10 kDa substances with ultracentrifugation using a
10 kDa cutoff 0.5 mL filter and perform LysC digestion and
trypsin digestion.
9. Dry the digested peptides in SpeedVac and perform iTRAQ
labeling using iTRAQ® Reagent—8PLEX One Assay Kit.
10. Fractionate peptides using reversed-phase liquid chromatogra-
phy (bRPLC) and analyze using LC-MS/MS. More detailed
procedures are described in [11].

3.5  Comments Cell culture components, including glucose, amino acids, NH4+
on Medium and Feed etc., can be directly modulated by changing culture medium, feed
Engineering supplement medium, and feeding strategy (e.g., feeding amount
and frequency) and consequently N-glycosylation patterns can be
altered accordingly. Therefore, benchmarking different combina-
tions of culture media and feed supplement media with appropriate
feeding strategies may be applied for medium and feed engineering
as a starting point. One or more current systems biology approaches,
including transcriptomic, proteomic, glycomic, and metabolomic
analysis, may also be implemented in this screening process for
selecting and further improving culture medium and feed supple-
ment medium for a specific production cell line. In particular,
extracellular and intracellular metabolic analysis is very commonly
used in exploring metabolic bottlenecks in cell culture and pro-
vides guidance for further optimizing the medium and feed.
When culture medium and feed supplement medium are
­chosen, modification of these media can be done by addition of
different amounts of feed additives or feed additive combinations,
including, but not limited to, glucose, glutamine, other amino
acids, galactose, sodium butyrate, uridine, manganese, and other
nucleotide sugar precursors. Reported effects of various modifica-
tions in media and feed on recombinant protein N-glycosylation
 220
Yuzhou Fan et al.

Table 1
Effect of media and feed modifications on recombinant protein N-glycosylation

Media and feed

modifications Effect on protein N-glycosylation
Glucose Glucose starvation: decreased site occupancy, galactosylation, and sialylation of mAb [6]
Glucose starvation: increased high mannose and decreased sialylation of IFN-γ [7]
Glucose starvation: decreased site occupancy of IFN-γ [8]
Glucose starvation at stationary phase of fed-batch: Improved the maturation of glycans of mAb [11]
Glucose addition: increased galactosylation of mAb [9]
High glucose consumption rate improved the maturation of glycans of mAb [10]
Galactose Galactose feeding: increased CTLA4Ig fusion protein galactosylation [13]
Galactose feeding: increased mAb galactosylation [12]
Glutamine Glutamine limitation: decreased site occupancy of IFN-γ [8]
Glutamine limitation: increased mAb Man5 glycan [10]
Glutamine addition: decreased TNFR-IgG fusion protein galactosylation and sialylation [18]
Amino acids Addition of depleted amino acids (cysteine, isoleucine, leucine, tryptophan, valine, asparagine, aspartate, and glutamate) in cell
culture: increased sialylation of EPO [19]
High amino acid consumption rate: increased mAb Man5 glycan [10]
NH4+ High NH4+ concentration: decreased TNFR-IgG fusion protein galactosylation and sialylation [18]
High NH4+ concentration: low GlcNAc occupancy and high Man5 glycan of mAb [10]
Mn2+ Addition of Mn2+: increased site occupancy and galactosylation of EPO [19]
 Nucleotide and Addition of N-acetylmannosamine (ManNAc): increased sialylation of IFN-γ [27]
nucleotide sugar Addition of glucosamine: decreased sialylation of EPO [28]
precursors Addition of glucosamine and uridine: increased antenarity, decreased sialylation of TIMP-1 [29]
Feeding galactose + Mn2+ + uridine (GMU): increased galactosylation of mAb [14]
Feeding galactose ± uridine, glucosamine ± uridine and ManNAc ± cytidine: increased sialylation of IFN-γ [30]
Sodium butyrate Addition of sodium butyrate: increased sialylation of IFN-γ [23]
Addition of sodium butyrate: decreased galactosylation of mAb [24]
Addition of sodium butyrate: increased site occupancy at Asn-184 of tPA [25]
Addition of sodium butyrate: decreased N-glycolylneuraminic acid (NGNA) content of a recombinant fusion protein [26]
Glycerol Addition glycerol: increased sialylation of IFN-β [68]
Monensin Addition of monesin: increased high mannose glycan of mAb [69]
Lipids Lipids increased site occupancy of IFN-γ [70]
Lipoprotein addition: increased fully glycosylated IFN-γ [71]
DMSO DMSO addition: decreased sialylation of IFN-β [68]
mAb monoclonal antibody, IFN-β interferon beta, IFN-γ interferon gamma, EPO erythropoietin, CTLA4Ig fusion protein of cytotoxic T-lymphocyte-associated protein 4 and
antibodies, TIMP-1 tissue inhibitor of metalloproteinase 1, TNFR-IgG fusion protein of tumor necrosis factor receptor and IgG, tPA tissue plasminogen activator
Medium and Feed Related N-Glyco-Engineering
221
222 Yuzhou Fan et al.

are summarized in Table 1. Although many of these effects may be

cell line and cultivation process dependent, they can still be used to
guide the practice of media and feed engineering. Statistical con-
trol such as replication and randomization need to be carefully
considered when designing experiments for testing the effect of
feed additives. Furthermore, applying a DoE approach (e.g., facto-
rial design) can streamline the investigation of the impact of muti-­
factors, muti-variables, and their interactions on N-glycosylation.

4  Notes

1. As a general rule, appropriate aseptic techniques should always

be applied for all cell culture procedures.
2. The viability should be above 80% for continuing the passage,
except for day 1 after thawing, the criterion is 75%.
3. Depending on cell line used, the common range of cell density
at inoculation is 2–8 × 105 viable cells/mL. However, even
higher seeding density can be used for specific purposes (e.g.,
if the production phase is of more interest to focus on or a
short culture period is desired).
4. Temperature shift from 37 °C to a lower temperature (typically
at 31–35 °C) can be applied to the fed-batch culture at late
experiential growth phase. The temperature shift typically
starts from between day 3 and day 7 to the end of the
fed-batch.
5. Typical base used in a fed-batch CHO cell bioreactor culture
could be NaOH or NaHCO3 or Na2CO3 with a concentration
ranging from 0.5 to 2 M. The optimal base and concentration
are normally selected based on the performance of different
bioreactor systems and cultivation processes.
6. The agitation rate is usually in the range of 100–200 rpm.
7. Dissolved oxygen normally set at 30–50%. DO can be main-
tained using air, N2, O2 and CO2 operated at either a constant
gas flow rate (for 1 L bioreactor 100–600 mL/h), or in a
gradient.
8. For 1 L bioreactor, the common working volume is within the
range of 200–800 mL.
9. Different cell lines may have very different demands for ­glucose
and glutamine. The glucose and glutamine concentrations in
the cell culture can be controlled by adjusting the feeding
criterion, amount, and frequency. Therefore, a Design of
­
Experiment (DoE) strategy is often needed for continuously
optimizing these variables and finally determining the optimal
operation window.
 Medium and Feed Related N-Glyco-Engineering
References
1. Berger M, Kaup M, Blanchard V (2012)
Protein glycosylation and its impact on bio-
technology. Adv Biochem Eng Biotechnol
127:165–185. doi:10.1007/10_2011_101
2. Butler M (2006) Optimisation of the cellular
metabolism of glycosylation for recombinant
proteins produced by Mammalian cell systems.
Cytotechnology 50(1–3):57–76. doi:10.1007/
s10616-005-4537-x
3. Costa AR, Rodrigues ME, Henriques M,
Oliveira R, Azeredo J (2013) Glycosylation:
impact, control and improvement during
therapeutic protein production. Crit Rev
­ 12. Kildegaard HF, Fan Y, Sen JW, Larsen B,
Biotechnol 34(4):281–299. doi:10.3109/073
88551.2013.793649
4. Hossler P, Khattak SF, Li ZJ (2009) Optimal
and consistent protein glycosylation in mam-
malian cell culture. Glycobiology 19(9):936–
949. doi:10.1093/glycob/cwp079
5. Bruhlmann D, Jordan M, Hemberger J, Sauer
M, Stettler M, Broly H (2015) Tailoring
recombinant protein quality by rational media
design. Biotechnol Prog 31(3):615–629.
doi:10.1002/btpr.2089
6. Liu B, Spearman M, Doering J, Lattova E,
Perreault H, Butler M (2014) The availability
of glucose to CHO cells affects the intracellular
lipid-linked oligosaccharide distribution, site
occupancy and the N-glycosylation profile of a
monoclonal antibody. J Biotechnol 170:17–
27. doi:10.1016/j.jbiotec.2013.11.007
7. Chee Furng Wong D, Tin Kam Wong K, Tang
Goh L, Kiat Heng C, Gek Sim Yap M (2005)
Impact of dynamic online fed-batch strategies
on metabolism, productivity and N-glycosy­
lation quality in CHO cell cultures. Biotechnol
Bioeng 89(2):164–177. doi:10.1002/bit.
20317
8. Nyberg GB, Balcarcel RR, Follstad BD,
Stephanopoulos G, Wang DI (1999) Metabolic
effects on recombinant interferon-gamma gly-
cosylation in continuous culture of Chinese
hamster ovary cells. Biotechnol Bioeng 62(3):
336–347
9. Nahrgang S, Kkagten E, De Jesus M, Bourgeois
M, Déjardin S, Von Stockar U, Marison IW
(2002) The effect of cell line, transfection proce-
dure and reactor conditions on the glycosylation
of recombinant human anti-rhesus D IgGl. In:
Bernard A, Griffiths B, Noé W, Wurm F (eds)
Animal cell technology: products from cells, cells
as products. Springer, The Netherlands, pp 259–
261. doi:10.1007/0-306-46875-1_59
10. Fan Y, Jimenez Del Val I, Muller C, Wagtberg
Sen J, Rasmussen SK, Kontoravdi C, Weilguny
D, Andersen MR (2015) Amino acid and
223
­lucose metabolism in fed-batch CHO cell
g
­culture affects antibody production and glyco-
sylation. Biotechnol Bioeng 112(3):521–535.
doi:10.1002/bit.25450
11. Fan Y, Jimenez Del Val I, Muller C, Lund AM,
Sen JW, Rasmussen SK, Kontoravdi C, Baycin-­
Hizal D, Betenbaugh MJ, Weilguny D,
Andersen MR (2015) A multi-pronged investi-
gation into the effect of glucose starvation and
culture duration on fed-batch CHO cell cul-
ture. Biotechnol Bioeng 112(10):2172–2184.
doi:10.1002/bit.25620
Andersen MR (2016) Glycoprofiling effects of
media additives on IgG produced by CHO
cells in fed-batch bioreactors. Biotechnol
Bioeng 113(2):359–366. doi:10.1002/bit.
25715
13. Schilling BM, Gangloff S, Kothari D, Leister
K, Matlock L, Zegarelli SG, Joosten CE, Basch
JD, Sakhamuri S, Lee SS (2008) Production
quality enhancements in mammalian cell cul-
ture process for protein production. US Patent
7,332,303
14. Gramer MJ, Eckblad JJ, Donahue R, Brown J,
Shultz C, Vickerman K, Priem P, van den
Bremer ET, Gerritsen J, van Berkel PH (2011)
Modulation of antibody galactosylation
through feeding of uridine, manganese chlo-
ride, and galactose. Biotechnol Bioeng 108(7):
1591–1602. doi:10.1002/bit.23075
15. St Amand MM, Tran K, Radhakrishnan D,
Robinson AS, Ogunnaike BA (2014) Con­
trollability analysis of protein glycosylation
in CHO cells. PLoS One 9(2):e87973.
doi:10.1371/journal.pone.0087973
16. St Amand MM, Radhakrishnan D, Robinson
AS, Ogunnaike BA (2014) Identification of
manipulated variables for a glycosylation con-
trol strategy. Biotechnol Bioeng 111(10):1957–
1970. doi:10.1002/bit.25251
17. Chen P, Harcum SW (2005) Effects of amino
acid additions on ammonium stressed
CHO cells. J Biotechnol 117(3):277–286.
doi:10.1016/j.jbiotec.2005.02.003
18. Gawlitzek M, Ryll T, Lofgren J, Sliwkowski
MB (2000) Ammonium alters N-glycan struc-
tures of recombinant TNFR-IgG: degradative
versus biosynthetic mechanisms. Biotechnol
Bioeng 68(6):637–646
19. Crowell CK, Grampp GE, Rogers GN, Miller
J, Scheinman RI (2007) Amino acid and man-
ganese supplementation modulates the glyco-
sylation state of erythropoietin in a CHO
culture system. Biotechnol Bioeng 96(3):538–
549. doi:10.1002/bit.21141
224 Yuzhou Fan et al.
20. Chen P, Harcum SW (2006) Effects of elevated
ammonium on glycosylation gene expression
in CHO cells. Metab Eng 8(2):123–132.
doi:10.1016/j.ymben.2005.10.002
21. Slade PG, Caspary RG, Nargund S, Huang CJ
(2016) Mannose metabolism in recombinant
CHO cells and its effect on IgG glycosylation.
Biotechnol Bioeng 7(113):1468–1480.
doi:10.1002/bit.25924
22. Zupke C, Brady LJ, Slade PG, Clark P, Caspary
RG, Livingston B, Taylor L, Bigham K, Morris
AE, Bailey RW (2015) Real-time product attri-
bute control to manufacture antibodies with
defined N-linked glycan levels. Biotechnol
Prog 31(5):1433–1441. doi:10.1002/btpr.
2136
23. Lamotte D, Buckberry L, Monaco L, Soria M,
Jenkins N, Engasser JM, Marc A (1999)
Na-butyrate increases the production and
alpha2,6-sialylation of recombinant interferon-­
gamma expressed by alpha2,6- sialyltransferase
engineered CHO cells. Cytotechnology 29(1):
55–64. doi:10.1023/A:1008080432681
24. Hong JK, Lee SM, Kim KY, Lee GM (2014)
Effect of sodium butyrate on the assembly,
charge variants, and galactosylation of antibody
produced in recombinant Chinese hamster
ovary cells. Appl Microbiol Biotechnol 98(12):
5417–5425. doi:10.1007/s00253-014-5596-8
25. Andersen DC, Bridges T, Gawlitzek M, Hoy C
(2000) Multiple cell culture factors can affect
the glycosylation of Asn-184 in CHO-­
produced tissue-type plasminogen activator.
Biotechnol Bioeng 70(1):25–31
26. Borys MC, Dalal NG, Abu-Absi NR, Khattak
SF, Jing Y, Xing Z, Li ZJ (2010) Effects of cul-
ture conditions on N-glycolylneuraminic acid
(Neu5Gc) content of a recombinant fusion
protein produced in CHO cells. Biotechnol
Bioeng 105(6):1048–1057. doi:10.1002/
bit.22644
27. Gu X, Wang DI (1998) Improvement of
interferon-­gamma sialylation in Chinese
hamster ovary cell culture by feeding of
­ tion by Chinese hamster ovary cells grown in
N-acetylmannosamine. Biotechnol Bioeng
58(6):642–648
28. Yang M, Butler M (2002) Effects of ammonia
and glucosamine on the heterogeneity of
erythropoietin glycoforms. Biotechnol Prog
18(1):129–138. doi:10.1021/bp0101334
29. Baker KN, Rendall MH, Hills AE, Hoare M,
Freedman RB, James DC (2001) Metabolic
control of recombinant protein N-glycan pro-
cessing in NS0 and CHO cells. Biotechnol
Bioeng 73(3):188–202
30. Wong NS, Wati L, Nissom PM, Feng HT, Lee
MM, Yap MG (2010) An investigation of
intracellular glycosylation activities in CHO
cells: effects of nucleotide sugar precursor feed-
ing. Biotechnol Bioeng 107(2):321–336.
doi:10.1002/bit.22812
31. Kunkel JP, Jan DC, Jamieson JC, Butler M
(1998) Dissolved oxygen concentration in
serum-free continuous culture affects N-linked
glycosylation of a monoclonal antibody. J Bio­
technol 62(1):55–71
32. Chotigeat W, Watanapokasin Y, Mahler S, Gray
PP (1994) Role of environmental conditions
on the expression levels, glycoform pattern and
levels of sialyltransferase for hFSH produced by
recombinant CHO cells. Cytotechnology 15
(1–3):217–221
33. Lin AA, Kimura R, Miller WM (1993)
Production of tPA in recombinant CHO cells
under oxygen-limited conditions. Biotechnol
Bioeng 42(3):339–350. doi:10.1002/bit.
260420311
34. Hossler P (2012) Protein glycosylation control
in Mammalian cell culture: past precedents
and contemporary prospects. Adv Biochem
Eng Biotechnol 127:187–219. doi:10.1007/
10_2011_113
35. Zanghi JA, Mendoza TP, Schmelzer AE, Knop
RH, Miller WM (1998) Role of nucleotide
sugar pools in the inhibition of NCAM polysi-
alylation by ammonia. Biotechnol Prog
14(6):834–844. doi:10.1021/bp9800945
36. Kimura R, Miller WM (1997) Glycosylation of
CHO-derived recombinant tPA produced
under elevated pCO2. Biotechnol Prog
13(3):311–317. doi:10.1021/bp9700162
37. Muthing J, Kemminer SE, Conradt HS, Sagi
D, Nimtz M, Karst U, Peter-Katalinic J (2003)
Effects of buffering conditions and culture pH
on production rates and glycosylation of clini-
cal phase I anti-melanoma mouse IgG3 mono-
clonal antibody R24. Biotechnol Bioeng
83(3):321–334. doi:10.1002/bit.10673
38. Yoon SK, Choi SL, Song JY, Lee GM (2005)
Effect of culture pH on erythropoietin produc-
suspension at 32.5 and 37.0 degrees C.
Biotechnol Bioeng 89(3):345–356. doi:10.1002/
bit.20353
39. Borys MC, Linzer DI, Papoutsakis ET (1993)
Culture pH affects expression rates and glyco-
sylation of recombinant mouse placental lacto-
gen proteins by Chinese hamster ovary (CHO)
cells. Biotechnology 11(6):720–724
40. Trummer E, Fauland K, Seidinger S, Schriebl
K, Lattenmayer C, Kunert R, Vorauer-Uhl K,
Weik R, Borth N, Katinger H, Muller D (2006)
Process parameter shifting: Part I. Effect of
DOT, pH, and temperature on the ­performance
 Medium and Feed Related N-Glyco-Engineering
of Epo-Fc expressing CHO cells cultivated in
controlled batch bioreactors. Biotechnol
Bioeng 94(6):1033–1044. doi:10.1002/bit.
21013
41. Yoon SK, Song JY, Lee GM (2003) Effect of
low culture temperature on specific productiv-
ity, transcription level, and heterogeneity of
erythropoietin in Chinese hamster ovary cells.
Biotechnol Bioeng 82(3):289–298. doi:10.1002/
bit.10566
42. Agarabi CD, Schiel JE, Lute SC, Chavez BK,
Boyne MT 2nd, Brorson KA, Khan MA, Read
EK (2015) Bioreactor process parameter
screening utilizing a Plackett-Burman design
for a model monoclonal antibody. J Pharm Sci
104(6):1919–1928. doi:10.1002/jps.24420
43. Senger RS, Karim MN (2003) Effect of shear
stress on intrinsic CHO culture state and gly-
cosylation of recombinant tissue-type plasmin-
ogen activator protein. Biotechnol Prog
19(4):1199–1209. doi:10.1021/bp025715f
44. Robinson DK, Chan CP, Yu Lp C, Tsai PK,
Tung J, Seamans TC, Lenny AB, Lee DK,
Irwin J, Silberklang M (1994) Characterization
of a recombinant antibody produced in the
course of a high yield fed-batch process. Bio­
technol Bioeng 44(6):727–735. doi:10.1002/
bit.260440609
45. Pacis E, Yu M, Autsen J, Bayer R, Li F (2011)
Effects of cell culture conditions on antibody
N-linked glycosylation—what affects high
mannose 5 glycoform. Biotechnol Bioeng
108(10):2348–2358. doi:10.1002/bit.23200
46. Sha S, Agarabi C, Brorson K, Lee DY, Yoon S
(2016) N-glycosylation design and control of
therapeutic monoclonal antibodies. Trends
Biotechnol 34(10):835–846. doi:10.1016/j.
tibtech.2016.02.013
47. Yamane-Ohnuki N, Kinoshita S, Inoue-­
Urakubo M, Kusunoki M, Iida S, Nakano R,
Wakitani M, Niwa R, Sakurada M, Uchida K,
Shitara K, Satoh M (2004) Establishment of
FUT8 knockout Chinese hamster ovary cells:
an ideal host cell line for producing completely
defucosylated antibodies with enhanced
antibody-­dependent cellular cytotoxicity. Bio­
technol Bioeng 87(5):614–622. doi:10.1002/
bit.20151
48. Mori K, Kuni-Kamochi R, Yamane-Ohnuki N,
Wakitani M, Yamano K, Imai H, Kanda Y,
Niwa R, Iida S, Uchida K, Shitara K, Satoh M
(2004) Engineering Chinese hamster ovary
cells to maximize effector function of produced
antibodies using FUT8 siRNA. Biotechnol
Bioeng 88(7):901–908. doi:10.1002/bit.
20326
49. Weikert S, Papac D, Briggs J, Cowfer D, Tom
S, Gawlitzek M, Lofgren J, Mehta S, Chisholm
225
V, Modi N, Eppler S, Carroll K, Chamow S,
Peers D, Berman P, Krummen L (1999)
Engineering Chinese hamster ovary cells to
maximize sialic acid content of recombinant
glycoproteins. Nat Biotechnol 17(11):1116–
1121. doi:10.1038/15104
50. Zhang X, Lok SH, Kon OL (1998) Stable
expression of human alpha-2,6-sialyltransferase
in Chinese hamster ovary cells: functional con-
sequences for human erythropoietin expression
and bioactivity. Biochim Biophys Acta 1425(3):
441–452
51. Jassal R, Jenkins N, Charlwood J, Camilleri P,
Jefferis R, Lund J (2001) Sialylation of human
IgG-Fc carbohydrate by transfected rat
alpha2,6-sialyltransferase. Biochem Biophys
Res Commun 286(2):243–249. doi:10.1006/
bbrc.2001.5382
52. Ferrari J, Gunson J, Lofgren J, Krummen L,
Warner TG (1998) Chinese hamster ovary cells
with constitutively expressed sialidase antisense
RNA produce recombinant DNase in batch
culture with increased sialic acid. Biotechnol
Bioeng 60(5):589–595
53. Wong NS, Yap MG, Wang DI (2006) Enhan­
cing recombinant glycoprotein sialylation
through CMP-sialic acid transporter over
expression in Chinese hamster ovary cells.
Biotechnol Bioeng 93(5):1005–1016.
doi:10.1002/bit.20815
54. Chenu S, Gregoire A, Malykh Y, Visvikis A,
Monaco L, Shaw L, Schauer R, Marc A,
Goergen JL (2003) Reduction of CMP-N-­
acetylneuraminic acid hydroxylase activity in
engineered Chinese hamster ovary cells using
an antisense-RNA strategy. Biochim Biophys
Acta 1622(2):133–144
55. Maszczak-Seneczko D, Olczak T, Jakimowicz
P, Olczak M (2011) Overexpression of UDP-­
GlcNAc transporter partially corrects galacto-
sylation defect caused by UDP-Gal transporter
mutation. FEBS Lett 585(19):3090–3094.
doi:10.1016/j.febslet.2011.08.038
56. Sealover NR, Davis AM, Brooks JK, George HJ,
Kayser KJ, Lin N (2013) Engineering Chinese
hamster ovary (CHO) cells for p ­roducing
recombinant proteins with simple glycoforms
by zinc-finger nuclease (ZFN)-mediated gene
knockout of mannosyl (alpha-1,3-)-glycopro-
tein beta-1,2-N-­acetylglucosaminyltransferase
(Mgat1). J Biotechnol 167(1):24–32.
doi:10.1016/j.jbiotec.2013.06.006
57. Kanda Y, Imai-Nishiya H, Kuni-Kamochi R,
Mori K, Inoue M, Kitajima-Miyama K, Okazaki
A, Iida S, Shitara K, Satoh M (2007) Esta­
blishment of a GDP-mannose 4,6-­dehydratase
(GMD) knockout host cell line: a new strategy
for generating completely non-fucosylated
226 Yuzhou Fan et al.
recombinant therapeutics. J Biotechnol
130(3):300–310. doi:10.1016/j.jbiotec.2007.
04.025
58. Imai-Nishiya H, Mori K, Inoue M, Wakitani
M, Iida S, Shitara K, Satoh M (2007) Double
knockdown of alpha1,6-fucosyltransferase
(FUT8) and GDP-mannose 4,6-dehydratase
(GMD) in antibody-producing cells: a new
strategy for generating fully non-fucosylated
therapeutic antibodies with enhanced ADCC.
BMC Biotechnol 7:84. doi:10.1186/
1472-6750-7-84
59. Davies J, Jiang L, Pan LZ, LaBarre MJ,
Anderson D, Reff M (2001) Expression of
GnTIII in a recombinant anti-CD20 CHO
production cell line: Expression of antibodies
with altered glycoforms leads to an increase in
ADCC through higher affinity for FC gamma
RIII. Biotechnol Bioeng 74(4):288–294
60. Umana P, Jean-Mairet J, Moudry R, Amstutz
H, Bailey JE (1999) Engineered glycoforms of
an antineuroblastoma IgG1 with optimized
antibody-dependent cellular cytotoxic activity.
Nat Biotechnol 17(2):176–180. doi:10.1038/
6179
61. North SJ, Huang HH, Sundaram S, Jang-Lee
J, Etienne AT, Trollope A, Chalabi S, Dell A,
Stanley P, Haslam SM (2010) Glycomics pro-
filing of Chinese hamster ovary cell glycosyl-
ation mutants reveals N-glycans of a novel size
and complexity. J Biol Chem 285(8):5759–
5775. doi:10.1074/jbc.M109.068353
62. von Horsten HH, Ogorek C, Blanchard V,
Demmler C, Giese C, Winkler K, Kaup M,
Berger M, Jordan I, Sandig V (2010) Pro­duction
of non-fucosylated antibodies by co-expression
of heterologous GDP-6-deoxy-­ D-lyxo-4-hexu­
lose reductase. Glycobiology 20(12):1607–
1618. doi:10.1093/glycob/cwq109
63. Yang Z, Wang S, Halim A, Schulz MA, Frodin
M, Rahman SH, Vester-Christensen MB,
Behrens C, Kristensen C, Vakhrushev SY,
Bennett EP, Wandall HH, Clausen H (2015)
Engineered CHO cells for production of diverse,
homogeneous glycoproteins. Nat Biotechnol
33(8):842–844. doi:10.1038/nbt.3280
64. Hanko VP, Heckenberg A, Rohrer JS (2004)
Determination of amino acids in cell culture
and fermentation broth media using anion-­
exchange chromatography with integrated
pulsed amperometric detection. J Biomol Tech
15(4):317–324
65. Jimenez Del Val I, Kyriakopoulos S, Polizzi
KM, Kontoravdi C (2013) An optimized
method for extraction and quantification of
nucleotides and nucleotide sugars from mam-
malian cells. Anal Biochem 443(2):172–180.
doi:10.1016/j.ab.2013.09.005
66. Kaas CS, Bolt G, Hansen JJ, Andersen MR,
Kristensen C (2015) Deep sequencing reveals
different compositions of mRNA transcribed
from the F8 gene in a panel of FVIII-producing
CHO cell lines. Biotechnol J 10(7):1081–
1089. doi:10.1002/biot.201400667
67. Wisniewski JR, Zougman A, Nagaraj N, Mann
M (2009) Universal sample preparation
method for proteome analysis. Nat Methods
6(5):359–362. doi:10.1038/nmeth.1322
68. Rodriguez J, Spearman M, Huzel N, Butler M
(2005) Enhanced production of monomeric
interferon-beta by CHO cells through the con-
trol of culture conditions. Biotechnol Prog
21(1):22–30. doi:10.1021/bp049807b
69. Pande S, Rahardjo A, Livingston B, Mujacic M
(2015) Monensin, a small molecule ionophore,
can be used to increase high mannose levels on
monoclonal antibodies generated by Chinese
hamster ovary production cell-lines. Biotechnol
Bioeng 112(7):1383–1394. doi:10.1002/
bit.25551
70. Castro PM, Ison AP, Hayter PM, Bull AT
(1995) The macroheterogeneity of recombi-
nant human interferon-gamma produced by
Chinese-hamster ovary cells is affected by the
protein and lipid content of the culture
medium. Biotechnol Appl Biochem 21(Pt 1):
87–100
71. Jenkins N, Castro P, Menon S, Ison A, Bull A
(1994) Effect of lipid supplements on the pro-
duction and glycosylation of recombinant
interferon-gamma expressed in CHO cells.
Cytotechnology 15(1–3):209–215
 Chapter 15

Glycosylation Analysis of Therapeutic Glycoproteins

Produced in CHO Cells
Sara Carillo, Stefan Mittermayr, Amy Farrell, Simone Albrecht,
and Jonathan Bones

Abstract
In the last decades, the number of approved therapeutic proteins drugs is increasing exponentially and a
large number of new therapeutic entities are progressing through clinical trials, solidifying biologics as the
most promising class of pharmaceuticals on the market. Several cell lines are available for biopharmaceuti-
cal processes but mammalian cells are preferred since they give fewer problems for immunogenicity as they
produce human-like post-translational modifications (PTMs). Glycosylation is the most common and
complex (for both bioprocess engineering and quality control) of these modifications. Obtaining the
desired glycosylation pattern is crucial for therapeutic proteins as it can impact significantly stability, half-
life and safety as well as driving molecular processes, modifying the way drug interacts with patients’ cells.
As a consequence, glycosylation (like other PTMs) needs to be regulated and accurately analyzed during
biopharmaceutical production. Herein we describe and discuss the analytical approaches for glycosylation
analysis of therapeutic ­glycoproteins produced in CHO (Chinese Hamster Ovary) cells. This chapter will
describe glycoprotein purification after separation from producing cell lines, N-glycan release and their
variants fine structural characterization through mass spectrometry techniques.

Key words Glycan analysis, Monoclonal antibodies, CHO cells, Mass spectrometry, Liquid
chromatography

1  Introduction

In the past decades, recombinant protein technology has become

one of the most promising fields in the development of drugs
against many severe diseases. The success in this research field is
reflected by the exponential growth of the biopharmaceutical sec-
tor, with a CAGR (Compound Annual Growth Rate) of 9.4% in the
period of 2014–2020 and rising from US$ 162 Bn in 2014 to US$
278 Bn by 2020 [1]. More than 200 biopharmaceutical products
have achieved market approval for the treatment of diseases including
cancer, chemotherapy-induced neutropenia, diabetes mellitus,

Paula Meleady (ed.), Heterologous Protein Production in CHO Cells: Methods and Protocols, Methods in Molecular Biology,
vol. 1603, DOI 10.1007/978-1-4939-6972-2_15, © Springer Science+Business Media LLC 2017

227
228 Sara Carillo et al.

hepatitis, multiple sclerosis, rheumatoid arthritis, and more than

350 additional products are currently in clinical trials.
Production of recombinant proteins requires the use of a host
organism to produce proteins using recombinant DNA; in modern
biopharmaceutical facilities the most commonly used host organ-
isms are bacteria or mammalian cells. Mammalian cells are pre-
ferred over bacterial cells for the production of many therapeutic
proteins as they produce human-like post-translational modifica-
tions (PTMs) [2]. Chinese hamster ovary (CHO) cell lines are
widely used due to an established regulatory track record, their
ability to grow in suspension culture under chemically defined
media conditions and their safety record [3]. CHO cells produce
human like glycosylation, however, also potentially some non-­
human epitopes. One of the disadvantages related to CHO cell
lines usage is that protein glycosylation may vary from lot to lot
based upon slight changes in environment conditions encountered
by cells during the bioprocess [3, 4]. This natural variation creates
the requirement for the characterization of the glycans present on
CHO derived proteins at all stages of bioprocessing.
As a consequence, therapeutic protein quality needs to be reg-
ulated by central authorities like the European Medicines Agency
(EMA) and the United States Food and Drug Administration
(FDA). As an example, characterization of monoclonal antibodies
(mAb) includes the determination of physicochemical and
immunoche­ mical properties, purity, biological activity, residual
host cell impurities and quantity, in line with the International
Conference on Harmonization (ICH) guideline, ICH Q6B. In
particular, ICH Q6B describes the specification that should be met
by the product, paying attention on the concept of heterogeneity.
Indeed the desired product can be a mixture of predicted post-
translational forms. The manufacturer should define the pattern of
heterogeneity of the desired product that should match with that
of the lots used in preclinical and clinical studies. If a consistent
pattern of product heterogeneity is demonstrated, further analysis
on each individual forms may not be necessary. On the contrary,
when process changes and degradation products result in hetero-
geneity patterns that differ from the desired ones, the impact on
product safety and quality of these alterations should be evaluated
(according to ICH Q5E guidelines) [5–7]. As for glycoproteins,
N-glyco­sylation is one of the most important post-translational
modifications. N-glycans analysis should consider carbohydrate
content (neutral sugars, amino sugars, and sialic acid), the struc-
ture of the carbohydrate chains, the oligosaccharide pattern (anten-
nary profile), and the glycosylation sites on the polypeptide.
Indeed, all these parameters can affect biological functions such as
complement-­dependent cytotoxicity (CDC), antibody-dependent
cellular cytotoxicity (ADCC), binding to various Fc receptors, and
binding to C1q protein [8–11]. For example, non-galactosylated
 Glycosylation Analysis of Therapeutic Glycoproteins 229

antibodies (G0 glycoforms) are able to interact with mannose-

binding protein present in serum and activate the alternative
complement cascade [12].
In this chapter, N-glycosylation analysis of CHO cells derived
monoclonal antibodies and related techniques is discussed. MAbs
are released in the culture media during cells growth and need to
be purified from culture media components and other proteins
secreted by host cells (HCPs). This is achieved during downstream
processing through multiple orthogonal purification steps, com-
prising protein A affinity chromatography, in which elution buffers
can change the contaminants amount outcome (Fig. 1a) [13].
Once mAbs have been purified, N-glycans are released via
­enzymatic digestion with peptide-N4-(N-acetyl-β-glucosaminyl)-
asparagine amidase (PNGase F). This step is generally followed by
labeling of the glycans at the free reducing end through reductive
amination: a specific fluorescent tag is chosen to increase detection
sensitivity and/or ionization efficiency, depending on the detection
method of choice (Fig. 1b) [14]. Subsequently, liquid
chromatography-­ mass spectrometry (LC-MS)-based techniques
are applied due to their excellent accuracy and reproducibility for
both structural and quantitative analyses. A panel of exoglycosidase
digestions may be used to establish the monosaccharides present
with linkage and positional specificity to simplify structural identi-
fication [15]. Additional information can be obtained by tandem
mass spectrometry (MS/MS) techniques. Moreover, application of
stable isotopologues of reagents commonly used for reducing ter-
minal oligosaccharide derivatization enables simplified quantitative
comparability analysis based on the relative response of light and
heavy isotopically labeled samples [16].

2  Materials

2.1  Protein 1. 0.45 and 0.2 μm bottle top filter with SFCA (surfactant-free
A Purification cellulose acetate) membranes (Thermo Scientific™ Nalgene™).
2. 1 mL HiTrap Protein A column (GE Healthcare).
3. Phosphate-buffered saline (PBS) tablets.
4. 0.5 M acetic acid: 28.71 mL glacial acetic acid, ≥99.7%, make
up to 1 L (see Note 1).
5. 1.5 M Tris–HCl buffer: 181.65 g Tris base, adjust to pH 6.8
with HCl, make up to 1 L.
6. 10 kDa molecular weight cutoff (MWCO) ultra-centrifugal
filter units (Amicon), 4 mL capacity.
7. Bradford assay kit (Biorad protein assay dye), bovine gamma
globulin standard (Biorad).
230 Sara Carillo et al.

a)
IgG containing
supernatant
cell pellet
ÄKTA

clarification

PNGase F

1) 2 mM DTT
2) 10 mM IAA Spin
b) 3) Buffer exchange down
50mM ABC

N-glycans

c)
Reductive
amination

Fig. 1 N-glycan analysis from cell culture to fluorescent labeling. As described in

Subheading 3.1, (a) to separate IgG from cell culture several filtration and
­centrifugation steps need to be carried out before IgG purification through affinity
chromatography with protein A columns. (b) mAbs are treated chemically and
enzymatically to allow N-glycan release and (c) the last step is a reductive ami-
nation where the label of choice is attached to glycans to facilitate glycan sepa-
ration and enable sensitive detection using fluorescent detection
 Glycosylation Analysis of Therapeutic Glycoproteins 231

2.2  Preparation 1. 10 kDa MWCO filters, regenerated cellulose membrane,

of Glycans 500 μL capacity.
for Fluorescent 2. Endoglycosidase digestion buffer: 50 mM ammonium bicar-
Labeling bonate (ABC buffer). Add 0.39 g of ammonium bicarbonate
in 100 mL of water, pH 7.8, does not need to be adjusted.
3. UA solution (buffer for increased protein denaturation): 8 M
urea in 0.1 M Tris–HCl buffer pH 8.5. 24 g urea in 50 mL of
Tris–HCl buffer, pH 8.5.
4. Reduction and alkylation agents. 4 mM DL-dithiothreitol
(DTT, 0.616 mg/mL), ≥98% and 20 mM iodoacetamide
(IAA, 3.70 mg/mL), ≥99% prepared in UA solution.
5. N-glycosidase F (PNGase F), glycerol free, recombinant (New
England Biolabs, Ipswich, MA, USA).
6. 1% formic acid solution.

2.3  Preparation 1. 2-aminobenzamide (2-AB) (Sigma Aldrich).

of Labeling Agents 2. Anthranilic acid (2-AA) (Sigma Aldrich).
(See Note 2)
3. 12C6 aniline. (Sigma Aldrich).
4.
Dimethyl sulfoxide (DMSO), sodium cyanoborohydride
(NaBH3CN), ≥95% and glacial acetic acid, ≥99.7%.

2.4  Separation 1. Hydrophilic interaction liquid chromatography (HILIC).

Buffers Mobile phase A: 50 mM formic acid, 1.925 mL of formic acid,
≥98% (Sigma Aldrich), make up to 1 L after adjusting pH
to 4.4 using ammonium hydroxide (Sigma Aldrich, LC-MS
grade). Mobile phase B: Acetonitrile (LC-MS grade). Mobile
phase C: Water.

2.5  Exoglycosidases 1. Enzymes:

Digestion (a)
Arthrobacter ureafaciens sialidase, ABS (Prozyme).
(b) Bovine testis galactosidase, BTG (Prozyme).
(c) Bovine kidney fucosidase, BKF (Prozyme).
2. Incubation buffer: 50 mM ammonium acetate, pH 5.5.
3. Ice-cold ethanol for enzymes precipitation or 10 kDa MWCO
filters, regenerated cellulose membrane, 500 μL capacity.

2.6  Equipment 1. Vacuum pump.

2. ÄKTA protein purification system (GE Healthcare) and column(s).
3. Perkin Elmer Victor X3 2030 Multilabel Reader.
4. Refrigerated centrifuge for medium capacity applications
(Eppendorf, Hamburg, Germany).
232 Sara Carillo et al.

5. Refrigerated desk centrifuge (Eppendorf).

6. Thermo mixer (Eppendorf).
7. Centrifugal vacuum evaporator (Thermo, Waltham, MA,
USA).
8. Frontal HILIC-FLR. Thermo Scientific Ultimate 3000
UHPLC system (San Jose, CA, USA) and a 50 × 2.1 mm ID,
1.7 μm Waters BEH (ethylene bridged hybrid) Glycan amide
column.
9. LC-FLR-QToF-MS. Waters Acquity UPLC system equipped
with an online fluorescence (FLR) detector. Column: 150 ×
1.0 mm ID, 1.7 μm Waters BEH Glycan amide.
10. The column temperature is maintained at 60 °C. Excitation
and emission wavelengths for fluorescence detection are:
(a) λex = 330 nm and λem = 420 nm for 2-AB label;
(b) λex = 350 nm and λem = 425 nm for 2-AA label;
(c) λex = 250 nm and λem = 340 nm for aniline label.
 he column outlet is connected to a Waters Xevo G2 QToF-MS
T
through an electrospray ionization interface. Capillary voltage
1.80 kV, ion source and nitrogen desolvation gas temperatures
at 120 °C and 400 °C, respectively. Desolvation gas flow rate
600 L/h, cone voltage at 50 V. MSE data acquisition in nega-
tive ionization mode over the range of 500–2000 m/z and
cone voltage ramped from 20 V to 90 V.
11. GlycoWorkBench software (see Note 3).

3  Methods

3.1  Protein 1. Harvest culture supernatants by centrifugation at 1000 × g

A Purification for 5 min and separate mAbs-containing supernatants from the
precipitated cell pellet.
2. Clarify the supernatant by sequential filtration through 0.45
and 0.2 μm cup filter membranes (see Note 4).
3. Purify samples via protein A chromatography using a 1 mL
HiTrap Protein A column; a flow rate of 1 mL/min is used.
Purification is obtained with the following steps: (1) wash sam-
ples with PBS to ensure complete removal of all unretained
material; (2) elute mAb from the affinity chromatography
­column using 0.5 M acetic acid (see Note 1, Fig. 1a); (3) neu-
tralize the pH of the eluate using 1.5 M Tris (100 μL per mL
of eluate) and buffer-exchange into PBS (4000 × g for 20 min),
using Amicon Ultra centrifugal filter units with a 10 kDa c­ utoff
membrane.
4. Establish mAbs concentration through Bradford microassay.
 Glycosylation Analysis of Therapeutic Glycoproteins 233

3.2  N-Glycan 1. Buffer-exchange 100 μg of Protein A purified mAb in 10 kDa

Release cutoff spin filters with UA solution, to a final volume of 50 μL
(10 min, 14,000 × g, 22 °C) (see Note 5).
2. Reduce proteins through the addition of 50 μL of 4 mM DTT
in UA solution (final concentration 2 mM) for 30 min at room
temperature, spin down (see Note 6), and wash three times
with 150 μL of UA solution, to a final volume of 50 μL to
remove residual reducing agent.
3. Alkylate reduced proteins by incubation with 10 mM IAA at
room temperature in the dark for 30 min (add 50 μL of 20 mM
IAA in UA solution), spin down, and wash three times with
150 μL of 50 mM ammonium bicarbonate buffer, pH 7.8 to
remove excess alkylating reagent and have the right buffer for
following enzymatic digestion.
4. Release N-glycans enzymatically by overnight incubation
with 500 units of PNGase F (see Note 7, Fig. 1b). Following
digestion, collect the N-glycans by centrifugation through
the filter membrane and reduce to dryness via vacuum
centrifugation.

3.3  Fluorescent 1. Treat released glycans with 50 μL of 1% formic acid (room
Labeling temperature, 20 min), and reduce sample again to dryness
(see Note 8).
2. Derivatize N-glycans via reductive amination using 10 μL of
0.37 M 2-AB and 0.95 M sodium cyanoborohydride in 30%
v/v acetic acid in dimethyl-sulfoxide (see Note 2) at 65 °C for
2 h (Fig. 1c). For comparative simultaneous analysis it is pos-
sible to use a duplex stable isotope labeling approach [16].

3.4  Excess Dye 1. Upon completion of the labeling reaction, add 5 μL of water
Removal from Labeled and 85 μL of acetonitrile to the sample (see Note 9). Remove
Samples excess fluorophore by HILIC chromatography using a Thermo
Ultimate 3000 RS HPLC equipped with an Acquity BEH
Glycan analytical column, 1.7 μm, 2.1 mm × 50 mm (Waters)
(see Note 10). Gradient conditions are described in Table 1.
2. Collect the fraction eluting off the column in high aqueous
conditions (80% water, approximate retention time 3–6 min)
and concentrate to dryness in a vacuum-centrifuge. Sample
­separation from unreacted labeling agent and targeted fraction
collection can be monitored using fluorescence detection.

3.5  Exoglycosidase 1. Transfer an aliquot of the labeled glycan pool into a fresh
Digestion tube and dry down. Add 1 μL of 10× 50 mM ammonium ace-
tate, pH 5.5, the required enzyme or array of enzymes, and
H2O to make up to 10 μL. Incubate overnight (16–18 h) at
37 °C (see Note 11). A typical array of enzymatic digestion is
234 Sara Carillo et al.

Table 1
Gradient conditions for excess dye removal via HLIC chromatography:
eluent A is ammonium formate buffer (pH 4.4), B is acetonitrile, and C is
water

Minutes %A %B %C
Initial 15 85  0
2.5 15 85  0
2.51  0 20 80
5  0 20 80
5.01 15 85  0
8 15 85  0

shown in Fig. 2 for CHO cell produced (immunoglobulin G1)

IgG1 glycans.
2. Precipitate enzymes with 100 μL of ice-cold ethanol and spin
down for 10 min at 14,000 × g, 4 °C. Alternatively, dilute sam-
ple to 50 μL, transfer it in 10 kDa cutoff spin filters, and spin
it down to elute digested N-glycans. Wash twice with 50 μL of
water and discard retained enzymes.
3. Collect supernatant (or eluate) and reduce it to dryness using
a Speed-vac.
4. Perform LC-FLR-MS/MS analysis (see Subheading 3.6).

3.6  Sample Analysis 1. Reconstitute 2-AB-derivatized N-glycan samples in 50 μL of

Using LC-FLR-MS/MS 80% acetonitrile.
2. Inject 10 μL into a Waters Acquity I-Class UPLC instrument
with fluorescence detection (see Subheading 2.6) equipped
with a BEH Amide 1.7 μm, 1.0 mm × 150 mm analytical col-
umn using flow rate at 0.15 mL/min and the gradient reported
in Table 2.

3.7  MS/MS Data 1. Using Glycoworkbench (see Note 3), draw glycan structure,
Analysis of N-Glycans mass options, and derivatization. In the structure tab, basic
structures from different glycan classes can be inserted and
modified/extended. Change properties of highlighted struc-
tures (right-click, mass options), such as derivatization, reducing
terminal labeling agent, ionization mode, and adducts.
2. Find glycans with a given m/z value. For an experimentally
obtained precursor, a list of possible glycan structures can be
populated based on a database search and provided parameters
by selecting Tools, Profiler, Find all structures with a given
m/z. Search, e.g., for the G0 glycan with a monoisotopic m/z
 Glycosylation Analysis of Therapeutic Glycoproteins 235

FA2
FA2[6]G1 Monosaccharide Symbols
N-Acetyl Glucosamine
Mannose
Galactose (G)
Fucose (F)
Sialic Acid (S)
A2

FA2[3]G1
FA2G2
FA2G2S1

Man5

(1) Undigested

(2) + ABS

(3) + ABS + BTG

(4) + ABS + BTG + BKF

min 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Fig. 2 Annotation of 2-AB-labeled anti-IL-8 IgG1 N-glycan profiles following exoglycosidase digestion. The top
profile (1) shows undigested pool of N-glycan, followed by a series of exoglycosidase digestions with (2)
Arthrobacter ureafaciens sialidase, ABS, (3) ABS and Bovine testis galactosidase, BTG and (4) ABS, BTG and
Bovine kidney fucosidase, BKF

Table 2
Gradient conditions for LC-FLR-MS/MS glycan analysis (column, flow rate,
and MS settings as reported in Subheading 2.6). Eluent A is ammonium
formate (pH 4.4), B is acetonitrile

Minutes %A %B
Initial 28 72
1.0 28 72
31.0 43 57
32.0 70 30
36.0 70 30
37.0 28 72
40.0 28 72
236 Sara Carillo et al.

of 790.2993 (2-AB labeled, −2H, negative mode) using the

appropriate mass accuracy (instrument dependent but gener-
ally 5–25 ppm). Search returns all matching structures for
dHex1HexNAc4Hex3.
3. Generate list of possible MS/MS fragment ions. For a high-
lighted structure, select Tools, Fragments, Compute fragments
for selected structures and select fragment options (see Note 12)
[17]. A list of theoretical fragment ions and respective m/z
values is populated (see Note 13).
4. Import mass spectrum and identify fragment ions. Using the
mass spectrometer software, export a spectrum list containing
relative intensities and respective m/z values (supported for-
mats are .txt, .xml, .mzxml, .mzdata, .t2d). Mass spectrum can
be loaded in Spectrum, Open Spectra. Left-click select spec-
trum peaks, once all are selected, right-click on one of the
peaks, and select “Find all fragments of the selected structures
matching the peaks.” A report containing the annotated spec-
trum can be generated by selecting Tools, Reporting, Create a
report of the annotations.

3.8  Method The described method can be performed for structural identifica-
Applicability Examples tion of any oligosaccharide mixture released from mAbs. As an
example, here we report the 2-AA labeled N-glycan analysis from
IgG1 monoclonal antibody produced by CHO DP12 cell line.

3.8.1  IgG1 N-Glycan 20 mL of culture media was transferred to a 50 mL plastic tube.
Analysis IgG1 was purified as reported in Subheading 3.1. Bradford micro-
assay revealed a protein concentration of 0.154 mg/mL, making
up a total of 3.08 mg of IgG1 material.
100 μg of IgG1 was transferred to a 10 kDa cutoff spin filter
and is treated as reported in Subheadings 3.2–3.5 using 2-AA as
the labeling reagent. Three technical replicates were prepared and
on each sample UPLC-FLR-MS analysis was performed in tripli-
cate. Figure 3 shows the HILIC-MS base-peak intensity (BPI)
chromatogram for IgG1 produced by CHO-DP12 cell line.
N-glycan structures were assigned based on a combination of all
described techniques (exoglycosidases digestion, mass spectrome-
try, and mass loss analysis).

4  Notes

1. Elution buffers used for downstream purification can affect

IgG quality attributes and HCPs (host cell protein) content.
100 mM sodium acetate, 100 mM arginine, 100 mM citrate,
or 100 mM glycine (pH 3.5 for all) are the most frequently
used in industry [13].
 Glycosylation Analysis of Therapeutic Glycoproteins 237

2 4 6 8 10 12 14 16 18 20 min

RT (min) Glycan Structure Symbolic Representation Monoisotopic mass + 2-AA Experimental [M-2H]2- (ppm)

6.16 FA1 1437.5393 689.2507 (1.3)

6.52 A2 1380.5178 717.7583 (5.7)

7.45 FA2 1583.5972 790.7913 (1.0)

8.41 MAN5 1355.4862 676.7405 (6.9)

8.57 FA1[6]G1 1542.5706 770.2864 (10.9)

8.63 A2[6]G1 1599.5921 798.7952 (8.0)

8.96 A2[3]G1 1599.5921 798.801 (15.3)

9.67 FA2[6]G1 1745.6500 871.8160 (1.9)

10.03 FA2[3]G1 1745.6500 871.8160 (1.9)

12.42 FA2G2 1907.7028 952.8459 (1.9)

14.98 FA2G2S1 2198.7983 1098.3748 (15.6)

Fig. 3 UPLC-MS BPI chromatogram from 2-AA labeled N-glycans released by IgG1 produced in CHO-DP12 cell
line; MS data for each glycan are listed in embedded table
238 Sara Carillo et al.

2. Several fluorescent labels have been used for N-glycan

derivatization [18]. For example, 2-aminobenzoic acid (2-AA),
inherently bearing a negative charge (pKa 1 = 2.18 and pKa 2 =
4.84), is recommended for glycans bearing two or less addi-
tional negative charges (e.g. sialic acid residues) to facilitate the
elution off chromatographic columns with cationic character-
istics. 2-AB, aniline, or other neutral labels are recommended
with highly charged glycans. The 2-AB label is of more general
use (the dextran ladder system is based on 2-AB label and also
facilitates data searching within the GlycoBase database). For
labeling solution preparation, dissolve 50 mg of 2-AB or 2-AA
or aniline and 60 mg of sodium cyanoborohydride in 1 mL of
30:70 acetic acid in DMSO. Make labels up in DMSO (solu-
tion can be heated to promote dissolution) and add the 30%
acetic acid in only at the very last step. Labeling reactions for
2-AA and aniline have longer incubation time, 5 and 16 h are
recommended, respectively.
3. MS/MS fragmentation is a key technique for the structural
determination of glycan molecules. Glycan fragments are com-
monly named according to a well-recognized nomenclature
established by Domon and Costello [17]. Fragments resulting
from the cleavage of a single intermolecular glycosidic bond,
such as the reducing-end containing fragments Y and Z and
their counterparts B and C, are most frequently observed in
glycan MS/MS spectra (Fig. 4). Additional cross-ring fragmen-
tation resulting from the cleavage of two intramolecular glyco-
sidic bonds, i.e., the reducing-end containing fragment X and its
counterpart A, may be present depending on fragmentation
conditions. The manual interpretation of glycan fragmentation
data is time-consuming and laborious. GlycoWorkbench is a
freely available software tool for the semi-automated annotation
of glycan fragmentation spectra which was developed as part of
the EUROCarbDB initiative to support the interpretation of
glycan MS data [19, 20]. Essentially, GlycoWorkbench performs
in-silico fragmentation of the proposed structures defined by the
user and compares the resulting fragment lists with the experi-
mental MS/MS spectral data. The graphical interface of
GlycoWorkbench is based on GlycanBuilder [21], a user-friendly
tool for the manual input of putative glycan structures. The tool
allows for the specification of monosaccharide subunits, their
linkage and anomeric type, and their modifications such as per-
methylation or acetylation using various symbolic notations
including CFG and Oxford symbol nomenclature as well as the
specification of reducing end derivatizations such as fluores-
cent labels. It further allows the definition of experimental and
instrument-­specific parameters that are of relevance for the
spectral interpretation, such as ion adducts (H+, Na+, K+, Li+),
charge state, and ionization mode.
 B2α C4α
MSe of aniline-labeled FA2G2S2 B1α C2α B3α
B4α

B1α/β Z3α FA2G2S2

B5 [1221.93]-2
[290.09]-1 Y1α
100
Aniline
142.05 B2β C4β
B1α C2β B3β
Y1β
B4β
Z3β

424.15 1223.94
179.05
Y1β C4α/β

%
[443.14]-1 [835.28]-1
291.09

202.07 306.12 B2α/β Y1α

262.09 B3α/β [1148.91]-2
[452.14]-1
[655.22]-1 B5Z3α/β
B4α/β [961.30]-1
C2α/β [817.27]-1
244.08
[470.15]-1
372.09 1224.44
1149.92
311.17 1233.45
534.15
312.07 1150.41 1233.97
715.24 737.22 1132.89
350.11 383.13
1059.87 1095.89
738.23 836.28 939.31
586.20

0
m/z 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000 1050 1100 1150 1200 1250
Glycosylation Analysis of Therapeutic Glycoproteins

Fig. 4 MSE spectrum of aniline-labeled FA2G2S2 glycan. Signals in the spectrum are assigned, according to Domon and Costello nomenclature, with the help of
GlycoWorkBench software
239
240 Sara Carillo et al.

4. Sequential filtration is preferable as direct filtration with 0.2 μm

membrane will result in a slower filtration and possibly filter
blockage.
5. To enable further protein relaxation and accessibility of poten-
tially shielded glycosylation sites, UA solution is recommended
as reaction buffer.
6. Monitor volumes in spin filters. Don’t concentrate beyond
20 mg/mL to avoid overconcentration and associated poten-
tial formation of aggregates.
7. Other vendors can refer to different activity units. Check the
conversion between vendors unit and IUB (International Union
of Biochemistry) unit.
8. Formic acid treatment assures conversion of released glycosyl-
amines to reducing terminal sugars and also promotes opening
of the reducing end sugar ring, thereby maximizing labeling
efficiency.
9. Interference of free dye with glycan peaks and nonselective
injection to LC instruments may occur if excess fluorescent
agent is not completely removed. Purification step needs to
be non-selective, i.e., prevent the loss of specific glycan classes
(e.g., small versus large or neutral versus charged glycans). It is
recommended to reach a final volume of 100 μL of 85% aceto-
nitrile in water; acetonitrile should be added to the sample just
before subjection to HILIC purification due to potential pre-
cipitation effects in high solvent concentrations.
10. If excess fluorescent dye is not completely removed detector over-
saturation may occur, so it is advised to disable detection from 0
to 2.5 min while the free dye is eluting off the HILIC column.
11. Monitor the loss of monosaccharides based on enzyme speci-
ficity. Either with retention time shift and or with mass loss
between digestions.
12. Produced fragments depend on multiple factors such as
­occupancy of reducing terminal end and instrument-depen-
dent dissociation method.
13. The charge state of the theoretical fragments predicted by
Glycoworkbench will match the precursor charge state set by
the user. However, experimental fragment ions generally have a
lower charge state than their precursor (e.g. -2 vs -1). The pre-
cursor charge state therefore needs to be adjusted to the charge
state of the fragment ions.

Acknowledgments

This publication has emanated from research conducted with the

financial support of Science Foundation Ireland (SFI) under Grant
Numbers SFI/11/SIRG/B107 and SFI/13/CDA/2196.
 Glycosylation Analysis of Therapeutic Glycoproteins
References
1. Pouech C, Lafay F, Wiest L, Baudot R, Léonard
D, Cren-Olivé C (2014) Monitoring the
extraction of additives and additive degrada-
tion products from polymer packaging into
solutions by multi-residue method including
solid phase extraction and ultra-high perfor-
mance liquid chromatography-tandem mass
spectrometry analysis. Anal Bioanal Chem
406(5):1493–1507
2. Durocher Y, Butler M (2009) Expression sys-
tems for therapeutic glycoprotein production.
Curr Opin Biotechnol 20(6):700–707.
doi:10.1016/j.copbio.2009.10.008
3. Planinc A, Bones J, Dejaegher B, Van
Antwerpen P, Delporte C (2016) Glycan char-
acterization of biopharmaceuticals: updates
and perspectives. Anal Chim Acta 921:13–27.
doi:10.1016/j.aca.2016.03.049
4. Farrell A, McLoughlin N, Milne JJ, Marison
IW, Bones J (2014) Application of Multi-­
Omics Techniques for Bioprocess Design and
Optimization in Chinese Hamster Ovary
Cells. J Proteome Res 13(7):3144–3159.
doi:10.1021/pr500219b
5. Rathore A (2014) Defining critical quality
attributes for monoclonal antibody therapeu-
tic products—process development Forum.
Bio­pharm Int 27 http://www.biopharmin-
ternational.com/defining-critical-
quality-attributes-monoclonal-antibody-ther-
apeuticproducts
6. ICH - Quality Guidelines (2014) http://www.
ich.org/products/guidelines/quality/article/
quality-guidelines.html
7. EMA (2009) Guideline on development, pro-
duction, characterisation and specifications
for monoclonal antibodies and related prod-
ucts—draft http://www.ema.europa.eu/
docs/en_GB/document_library/Scientific_
guideline/2009/09/WC500003074.pdf
8. Wright A, Morrison SL (1997) Effect of glyco-
sylation on antibody function: implications for
genetic engineering. Trends Biotechnol 15(1):
26–32. doi:10.1016/S0167-7799(96)10062-7
9. Jefferis R (2005) Glycosylation of recombinant
antibody therapeutics. Biotechnol Prog
21(1):11–16. doi:10.1021/bp040016j
10. Shields RL, Lai J, Keck R, O’Connell LY,
Hong K, Meng YG, Weikert SH, Presta
LG (2002) Lack of fucose on human IgG1
N-linked oligosaccharide improves binding
to human FcγRIII and antibody-dependent
cellular toxicity. J Biol Chem 277(30):
26733–26740
241
11. Umaña P, Jean-Mairet J, Moudry R, Amstutz
H, Bailey JE (1999) Engineered glycoforms of
an antineuroblastoma IgG1 with optimized
antibody-dependent cellular cytotoxic activity.
Nat Biotechnol 17(2):176–180
12. Malhotra R, Wormald MR, Rudd PM, Fischer
PB, Dwek RA, Sim RB (1995) Glycosylation
changes of IgG associated with rheumatoid
arthritis can activate complement via the
mannose-­ binding protein. Nat Med 1(3):
237–243
13. Farrell A, Mittermayr S, Morrissey B, Mc
Loughlin N, Navas Iglesias N, Marison IW,
Bones J (2015) Quantitative host cell protein
analysis using two dimensional data inde­pendent
LC–MSE. Anal Chem 87(18):9186–9193
14. Ruhaak L, Zauner G, Huhn C, Bruggink C,
Deelder A, Wuhrer M (2010) Glycan labeling
strategies and their use in identification and
quantification. Anal Bioanal Chem 397(8):
3457–3481
15. Royle L, Radcliffe CM, Dwek RA, Rudd PM
(2006) Detailed structural analysis of N-glycans
released from glycoproteins in SDS-PAGE gel
bands using HPLC combined with exoglyco­
sidase array digestions. Method Mol Biol
347:125–143
16. Atwood JA, Cheng L, Alvarez-Manilla G,
Warren NL, York WS, Orlando R (2008)
Quantitation by isobaric labeling: applications
to glycomics. J Proteome Res 7(1):367–374.
doi:10.1021/pr070476i
17. Domon B, Costello CE (1988) A systematic
nomenclature for carbohydrate fragmentations
in FAB-MS/MS spectra of glycoconjugates.
Glycoconj J 5(4):397–409. doi:10.1007/
bf01049915
18. Harvey DJ (2011) Derivatization of carbohy-
drates for analysis by chromatography; electro-
phoresis and mass spectrometry. J Chromatogr
B 879(17–18):1196–1225. d ­oi:10.1016/j.
jchromb.2010.11.010
19. EUROCarbDB. Tools to analyse MS spectra:
GlycoWorkbench
20. Ceroni A, Maass K, Geyer H, Geyer R, Dell A,
Haslam SM (2008) GlycoWorkbench: a tool
for the computer-assisted annotation of mass
spectra of glycans. J Proteome Res 7(4):1650–
1659. doi:10.1021/pr7008252
21. Ceroni A, Dell A, Haslam SM (2007) The
GlycanBuilder: a fast, intuitive and flexible soft-
ware tool for building and displaying glycan
structures. Source Code Biol Med 2:3–3.
doi:10.1186/1751-0473-2-3
 Chapter 16

Characterization of Host Cell Proteins (HCPs) in CHO

Cell Bioprocesses
Catherine E.M. Hogwood, Lesley M. Chiverton, and C. Mark Smales

Abstract
Host cell protein content during bioprocessing of biotherapeutic proteins generated from cultured Chinese
hamster ovary (CHO) cells is typically measured using immunological and gel-based methods. Estimation
of HCP concentration is usually undertaken using Enzyme-Linked ImmunoSorbent Assays (ELISA),
while estimation of HCP clearance/presence can be achieved by comparing 2D-PAGE images of samples
and by undertaking western blotting of 2D-PAGE analyzed samples. Here, we describe the analyses of
HCP content using these methodologies.

Key words Host cell proteins (HCPs), ELISA, 2D-PAGE, Western blotting

1  Introduction

The manufacturing of biotherapeutic proteins in Chinese hamster

ovary (CHO) cells results in the presence of not only the target
protein of interest being present in the final cell culture harvest
material, but also impurities in the form of host cell proteins
(HCPs), DNA, RNA, lipids, and other cellular material [1]. HCPs
present a potential risk to the patient if not mitigated against [2]
and hence the amounts must be reduced to acceptable concentra-
tions in the final biotherapeutic preparation. Although each prod-
uct is currently treated on a case-by-case basis, general guidelines
suggest that the upper limit of HCPs in a biotherapeutic prepara-
tion should be 100 ng/mL [3]. As such, HCPs are often moni-
tored as part of bioprocess design, but also during manufacturing
to ensure these are being cleared to acceptable amounts. HCPs are
usually removed via the use of multiple purification steps, in most
cases chromatography based, to deliver a produce with acceptable
amounts of HCPs [4]. Furthermore, the abundance and clearance
of HCPs during a bioprocess is considered a critical quality attri-
bute that can be used as a benchmark of process robustness [5].

Paula Meleady (ed.), Heterologous Protein Production in CHO Cells: Methods and Protocols, Methods in Molecular Biology,
vol. 1603, DOI 10.1007/978-1-4939-6972-2_16, © Springer Science+Business Media LLC 2017

243
244 Catherine E. Hogwood et al.

A comprehensive review of methodologies for host cell protein

analysis has been published by Tscheliessnig et al. [6]. The current
“work horse” for the analysis of HCPs in biotherapeutic proteins is
the use of Enzyme-Linked ImmunoSorbent Assays (ELISA) [7]
and 2D-gel-based [6] comparative analyses including western blot-
ting using anti-CHO HCP immunodetection of 2D-PAGE ana-
lyzed samples. The immune-detection ELISA-based approach has
high selectively and can be used in a high-throughput manner.
This approach traditionally relies on the generation of a polyclonal
antibody pool to CHO proteins via the immunization of animals
with a null host cell line to determine the concentration of HCPs
in a sample using a standard curve [1]. 2D-PAGE analysis, and
2D-DIGE (two-dimensional fluorescence difference gel electro-
phoresis), is used to compare the final biotherapeutic preparation
with either the culture harvest material or another reference sam-
ple to determine the number of HCPs that have been cleared and
to map those remaining [8]. This can also be subjected to western
blotting to allow the comparison of the immunodetected HCPs to
those observed via 2D-PAGE or 2D-DIGE. He we described the
analysis of HCPs in a biotherapeutic protein sample for each of
these methods.

2  Materials

Prepare any buffers or reagents with Milli-Q quality water. All

reagents should be of analytical grade at least.

2.1  Recovery of Host 1. Wash Buffer: 25 mM sodium phosphate, 10 mM sodium chlo-
Cell Proteins ride pH 7.5.
for Analysis 2. General protease inhibitor cocktail.
3. Benzonase nuclease.
4. Centrifugation-based membrane concentrators with molecular
weight cutoff of 5000 Da.
5. Commercially available 2-D Clean Up Kit (GE Healthcare).
6. 2D-PAGE resolubilization buffer: 7 M urea, 2 M thoiurea, 4%
CHAPS, 40 mM DTT, and 0.5% pH 3–10 pharmalytes.

2.2  ELISA Analysis 1. Commercially available CHO host cell protein ELISA kit (e.g.,
of Host Cell Protein Cygnus Technologies HCP ELISA Kit for CHO, 3G Catalogue
Content number P6787-50TAB).

2.3  2D-PAGE 1. Immobine DryStrip, pH 3–10 Non-Linear 7 cm IPG strips.

Analysis of Host Cell 2. Ettan IPGphor II system (GE Healthcare).
Proteins
3. Immobiline PlusOne Drystrip cover fluid (GE Healthcare).
 CHO HCP Analysis 245

4. SDS equilibration buffer: 6 M urea, 50 mM Tris–HCl pH 8.8,

29.3% glycerol, 2% (w/v) SDS.
5. Dithiothreitol (DTT) and iodoacetamide (IAA).
6. Reducing SDS equilibration buffer: 6 M urea, 50 mM Tris–
HCl pH 8.8, 29.3% glycerol, 2% (w/v) SDS, and 1% (w/v)
DTT.
7. Alkylating SDS equilibration buffer: 6 M urea, 50 mM Tris–
HCl pH 8.8, 29.3% glycerol, 2% (w/v) SDS, and 2.5% (w/v)
iodoacetamide.
8. A few grains of bromophenol blue.
9. Standard 12% Tris-Glycine gels and running buffer.
10. Agarose sealing solution: 25 mM Tris base, 192 mM glycine,
0.1% SDS and 0.5% agarose.
11. Bio-Safe Coomassie G-250 Stain.

2.4  2D-PAGE 1. PVDF membrane.

Immunoblotting 2. TBST solution: Tris buffered Saline (TBS, 0.15 M NaCl,
Analysis of Host Cell 3 mM KCl, 25 mM Tris) containing 0.05% Tween 20.
Proteins
3. Anti-CHO host cell protein antibody.
4. Secondary antibody appropriate for recognizing primary anti-
body with HRP label.

3  Methods

3.1  Recovery of Host Select the method of recovery of HCPs to be used based upon the
Cell Proteins subsequent analysis methodology that will be applied to the sam-
ples (see Notes 1 and 2).

3.1.1  Recovery If the method of subsequent analysis is ELISA based, samples from
for ELISA Analysis any stage of cell culture or downstream processing may be recov-
ered directly from the material to be analyzed. For ELISA analysis
of harvest material from cell culture, remove cellular material by
centrifugation to pellet cells and cell debris and then remove super-
natant material for HCP analysis. Store at −20 °C.

3.1.2  Recovery 1. Recover cell culture supernatant material by centrifugation to

of Material for Gel-Based remove cell debris. Sufficient material should be recovered for
Proteomic Analysis the analysis intended. For ELISA 50–100 μL may be sufficient,
20 mL of culture material is sufficient for multiple analyses by
1D- or 2D-PAGE (see Note 3). No further processing is
required for ELISA samples.
2. For 2D-PAGE samples, wash the cell culture supernatant
­sample with 20 mL of Wash Buffer, while concentrating a
­minimum of fivefold using centrifugation-based concentrators
(see Note 4).
246 Catherine E. Hogwood et al.

- pH 3 pH11 + - pH 3 pH11 +

A B High
High

MW
MW

Low
Low

Fig. 1 2D-PAGE analysis of the cell culture supernatant of an antibody-producing cell line at harvest when
stained with Coomassie blue (a) or Sypro Ruby (b). The antibody heavy and light chains are the most prominent
bands on the right-hand side of the image. Host cell proteins are also present

3. Add a general protease inhibitor to the sample and 100 units

of benzonase nuclease per mL of sample for 1 h on ice
(see Note 5).
4. Estimate the total protein concentration in the recovered sam-
ples using an appropriate protein assay (see Note 6).
5. Prior to 2D-PAGE-based analysis the protein should be desalted,
concentrated by precipitation, and resolubilized in a suitable
buffer (see Note 7). Take 100 μg of total protein (as determined
in step 4 above) and subject to the 2-D Clean Up Kit process
following the manufacturer’s instructions (see Note 8).
6. Resolubilize the protein pellet in 125 μL (see Note 9) of
2D-PAGE resolubilization buffer and either use immediately
for analysis or store at −80 °C.

3.2  ELISA Analysis ELISA analysis requires appropriate anti-CHO host cell protein
of Host Cell Protein antibodies and standards. In most cases, these will not be directly
Content available for the analyst and commercial reagents will need to be
purchased (see Note 10). In this case, the directions of the manu-
facturer should be followed precisely.

3.3  2D-PAGE The most common approach taken for visualizing total HCP con-
Analysis of Host Cell tent in a sample and comparing HCP content between samples is
Proteins to use a combination of 2D-PAGE and 2D-PAGE western blot-
ting. This allows an assessment of the immunocoverage using a
specific polyclonal anti-HCP reagent. The stained 2D-PAGE gel
(see Fig. 1 for example) is then directly compared to the western
blot analysis (see Note 11).
1. For each sample to be analyzed, an Immobine DryStrip, pH
3–10 Non-Linear 7 cm IPG strip is overlaid onto the 125 μL
sample (see Note 12).
 CHO HCP Analysis 247

2. The strip and sample is then covered with Immobiline PlusOne

Drystrip cover fluid.
3. Isoelectric focusing (IEF) is then performed to a total of
8 kVh. We recommend the following protocol to achieve this
if using an Ettan IPGphor II system; Rehydration 20 °C for
1 h; Step 1 step-n-hold at 30 V for up 12 h; Step 2 gradient
from 200 V over 45 min; Step 3 step-n-hold to 500 V for
45 min; Step 4 step-n-hold to 1000 V for 45 min; Step 5
­gradient to 8000 V over 30 min; Step 6 hold at 8000 V until a
total of 8000 kVh is achieved.
4. Once a total of 8000 kVh is achieved remove the strips, rinse
with Milli-Q water, and either store at −80 °C or subject
immediately to second-dimension SDS-PAGE analysis.
5. Reduce IPG strips in Reducing SDS equilibration buffer for
15 min with gentle agitation.
6. Pour off the reducing buffer, rinse the strips with Milli-Q
water, and then place in Alkylating SDS equilibration buffer
with the addition of a few grains of bromophenol blue (to
form a dye front when running the SDS-PAGE) for 15 min
with gentle agitation.
7. Wash the strips with the SDS-PAGE Tris-glycine buffer to be
used for running the SDS-PAGE system.
8. Seal each of the equilibrated strips on the top of a 12% Tris-­
Glycine gel without a stacking gel with agarose sealing
solution.
9. Run the second dimension as appropriate for the SDS-PAGE
system until the dye front reaches the bottom of the gel.
10. Stain the 2D-gel with Bio-Safe Coomassie G-250 Stain.
11. Capture the gel images at 600 dpi, preferably using 16 bit
color depth in grayscales and RGB using an appropriate
scanner.
12. Analyze the 2D-PAGE images using appropriate 2D-PAGE
software (see Note 13).

3.4  2D-PAGE For 2D-PAGE western blotting for HCPs, the initial analysis is
Immunoblotting undertaken up until step 9 as described in Subheading 3.2. After
Analysis of Host Cell this follow the protocol as below.
Proteins 1. Remove the 2D-PAGE gel from the tank and transfer to PVDF
membrane following an appropriate protocol.
2. Remove the membrane from the transfer apparatus and block
with nonfat dried milk powder (5% w/v) in TBST solution for
1 h at room temperature with gentle agitation.
3. Pour off the block solution and wash the membrane with
TBST before discarding this wash.
248 Catherine E. Hogwood et al.

4. Dilute the commercial anti-CHO HCP antibody (primary

antibody) as recommended by the manufacturer in TBST and
pour onto the membrane. Incubate overnight at 4 °C with
gentle agitation.
5. Pour off the primary antibody and wash the membrane five
times with excess TBST.
6. Add an appropriate secondary antibody diluted in TBST and
incubate for 1 h.
7. Use an appropriate reagent (e.g., ECL chemiluminescent
reagent) and imagining system to capture the western blot.
8. 2D-western blot images for HCPs can then be compared
to 2D-PAGE images generated for total protein content in
Subheading 3.2 above using an appropriate software package.

4  Notes

1. The method of recovery of HCPs from extracellular samples

(e.g., cell culture harvest material, during downstream pro-
cessing) can influence both those proteins recovered and the
subsequent analysis methods that can be applied to these sam-
ples. For proteomic gel-based analyses (e.g., 1D and 2D-PAGE)
different considerations and processes will be required than
those for non-gel-based mass spectrometry approaches. A
com­p­arison of suitable methods for these types of HCP analy-
ses has been published by Valente et al. [9].
2. For the analysis of HCPs throughout a downstream process or
in the drug substance/drug product, the method of analysis
must consider that the HCP content should be in the range of
1–100 ng/mL while the drug substance may be at a con­
centration in the order of g/L. Approaches such as 1D and
2D-PAGE are unlikely to be appropriate under such condi-
tions unless used for immunoblotting. Under such conditions,
when loading a set amount of protein onto gels for analysis,
the drug substance predominates and can interfere with the
running or observation of HCPs that may be present.
3. For gel-based analysis of samples from downstream processed
samples the amount of material for analysis will need to be
judged on a case-by-case basis and consider the concentration
of drug substance and anticipated concentration of HCPs. The
concentration of HCPs can be determined before recovery for
gel-based analysis by ELISA to aid in determining the amount
of sample to recover HCPs from.
4. We recommend using concentrators that have a molecular
weight cutoff of 5000 Da which should be sufficient to ensure
the majority of recombinant protein and HCPs are retained.
 CHO HCP Analysis 249

Users should consider the membrane used and if this is likely

to result in any loss of proteins (e.g., due to proteins “sticking”
to the membrane). Alternatively, a precipitation-based method
can be used to precipitate the protein and then resolubilize this
is an appropriate volume of buffer to generate a smaller volume
with a higher concentration of material.
5. Although it is not absolutely necessary to add an inhibitor, we
recommend this to ensure that protein degradation does not
occur due to the presence of any proteases. If samples are to be
used to investigate protease activity/presence, protease inhibi-
tors should not be included. The addition of benzonase nucle-
ase should result in the degradation of any DNA and/or RNA
present. Again, if the samples are to be used for host cell DNA
or RNA analysis this should be excluded.
6. Protein concentration can be readily determined by compari-
son to a standard curve using a method such as that described
by Bradford (see [10]). For samples containing recombinant
proteins recovered from CHO cell culture supernatants, it
should be remembered that it is likely that the majority of pro-
teins in a sample are the recombinant protein and not host cell
protein.
7. A review of methods for HCP precipitation is provided in [9].
8. A commercial kit does not have to be used or purchased. One
of the precipitation methods reported in ref. 9 can be reliably
used instead.
9. A total of 125 μL of sample is loaded when undertaking
2D-PAGE with 7 cm Immobiline Drystrip IPG isoelectric
focusing strips. If longer strips or a greater concentration of
total protein is to be used, this should be adjusted as
appropriate.
10. A number of commercial CHO HCP ELISA assay kits are
available which may or may not offer the appropriate ­sensitivity
and coverage for a particular CHO host and/or recombinant
cell line.
11. 2D Fluorescence Difference Gel Electrophoresis (2D-DIGE)
is often used to directly compare two samples for HCP con-
tent/amount. A good description of this methodology for
the HCP analysis is provided by Grzeskowiak et al. [11]. This
approach allows you to run two samples simultaneously on the
same gel and directly compare those proteins/spots present
and their relative abundance.
12. We recommend using a nonlinear pH range 3–10 strip for ini-
tial investigations. However, if investigation of a narrower pH
range is desired then strips with a range of different pH ranges
are available.
250 Catherine E. Hogwood et al.

13. A number of software packages are available for the analysis of

2D-PAGE gels. Recently, some of these have been developed
to allow the comparison of 2D-PAGE gels with the corres­
ponding immunoblots specifically for HCP analysis (for exam-
ple, SpotMap from TotalLab).

References

1. Bracewell DG, Francis R, Smales CM (2015) in therapeutic protein bioprocessing—methods

The future of host cell protein (HCP) identifi-
cation during process development and manu-
facturing linked to a risk based management
for their control. Biotechnol Bioeng 112:
1727–1737
2. Bailey-Kellogg C, Gutiérrez AH, Moise L,
Terry F, Martin WD, De Groot AS (2014)
CHOPPI: a web tool for the analysis of immu-
nogenicity risk from host cell proteins in CHO-­
based protein production. Biotechnol Bioeng
111:2170–2182
3. Chon JH, Zarbis-Papastoitsis G (2011)
Advances in the production and downstream
processing of antibodies. N Biotechnol 28:
458–463
4. Hogwood CE, Tait AS, Koloteva-Levine N,
Bracewell DG, Smales CM (2013) The dynam-
ics of the CHO host cell protein profile during
clarification and protein A capture in a platform
antibody purification process. Biotechnol
Bioeng 110:240–251
5. Tait AS, Hogwood CE, Smales CM, Bracewell
DG (2012) Host cell protein dynamics in the
supernatant of a mAb producing CHO cell
line. Biotechnol Bioeng 109:971–982
6. Tscheliessnig AL, Konrath J, Bates R,
Jungbauer A (2013) Host cell protein analysis
and applications. Biotechnol J 8:655–670
7. Zhu-Shimoni J, Yu C, Nishihara J, Wong RM,
Gunawan F, Lin M, Krawitz D, Liu P, Sandoval
W, Vanderlaan M (2014) Host cell protein
testing by ELISAs and the use of orthogonal
methods. Biotechnol Bioeng 111:2367–2379
8. Jin M, Szapiel N, Zhang J, Hickey J, Ghose S
(2010) Profiling of host cell proteins by two-­
dimensional difference gel electrophoresis
(2D-DIGE): Implications for downstream
­process development. Biotechnol Bioeng 105:
306–316
9. Valente KN, Schaefer AK, Kempton HR,
Lenhoff AM, Lee KH (2014) Recovery of
Chinese hamster ovary host cell proteins for
proteomic analysis. Biotechnol J 9:87–99
10. Ramagli LS (1999) Quantifying protein in 2-D
PAGE solubilization buffers. In: Link AJ (ed)
Methods in molecular biology: 2-D proteome
analysis protocols. Humana Press Inc., Totowa,
NJ, pp 99–103
11. Grzeskowiak JK, Tscheliessnig A, Toh PC,
Chusainow J, Lee YY, Wong N, Jungbauer A
(2009) 2-D DIGE to expedite downstream
process development for human monoclonal
antibody purification. Protein Expr Purif 66:
58–65
 Index

A Fluorescent in situ hybridization (FISH).����������� 4, 121, 125,

Aggregation
Russell bodies���������������������������������������������������������������14
Apoptosis���������������������������������������������7, 72, 77, 80, 152, 196
B Glutamine synthetase (GS)�������������������������������� 80, 102, 212
Biopharmaceutical��������������� 25, 28, 71–73, 87, 152, 169, 227
C
Cell culture�������������������������������������������12, 14, 16–18, 36, 37,
45–47, 50, 62, 72–74, 96, 101, 139, 144–146,
152, 195, 210, 212–219, 222, 230, 243, 245, 246,
248, 249
Cell engineering����������������������������������������������������� 9, 40, 196
Cell line development����������������������������2, 4, 90–91, 143–152
Cell line stability�������������������������������������������������������119–140
Chaperones������������������������������������������������������������� 11–13, 26
Chinese hamster ovary (CHO)������������������������� 1, 45, 71–84,
119, 169, 187, 195, 209, 243
CHO cell engineering������������������������������������������������������196
CHO cells������������������������������� 46, 71, 87, 101, 120, 209, 228
CHO genome����������������������������������������������������������������4, 82
Codon optimization�������������������������������������������������������������9
CRISPR/Cas9���������������������������������������4, 5, 10, 73, 101–117
D Liquid chromatography���������������������������� 159, 219, 229, 231
Design of Experiment (DOE)����������������������������������� 17, 222
Differential gene expression�������������������������������������169–186
Difficult to express protein���������������������������������������������1–18
DNA methylation����������������������������������������������������������������6
2D-PAGE����������������������������������������������������������������244–250
E Monoclonal antibodies (mAb)�������������������������� 1, 26, 46, 57,
Enzyme-Linked Immunosorbent Assay (ELISA)���������������48,
52, 120, 123–124, 128–129, 244–246, 248, 249
Fed-batch culture��������������������������������������� 72, 213, 214, 222
Filter-aided sample preparation (FASP)���������� 187–193, 219
Flow cytometry����������������������������������������������������������������120
Fluorescence-Activated Cell Sorting (FACS)������������� 91, 98,
104–105, 113–114, 116, 144, 146–151
134–137
G
Genome-editing������������������������������������������������� 72, 102, 169
Glycan analysis������������������������������������������������� 230, 235, 236
Glycoengineering���������������������������������������������������������25–40
Glycosylation site insertion�������������������������������������� 228, 240
H
Heterogeneity������������������������������������������17, 26, 39, 169, 228
Histone acetylation���������������������������������������������������������������6
Host Cell Proteins (HCPs)��������������������������������������243–250
Human embryonic kidney (HEK293)�������������� 28, 58, 61, 67
I
Inducible������������������������������������������������������15, 16, 88, 89, 98
K
Knockdown������������������������������������������������������������ 32, 87–99
Knockout������������������������������� 39, 72, 101, 102, 105, 115–117
L
LC-MS/MS������������������������������197, 203–204, 219, 229, 231
M
Mass Spectrometry����������������������������188, 204–205, 236, 248
Medium and feed optimization����������������������������������������210
miRNA������������������������������������������������������������� 87–89, 91, 99
miRNA sponge vector�������������������������������������������� 89, 95–98
71, 144, 216, 228, 229, 236
N
Next generation sequencing (NGS)�����������������������������4, 169
N-glycans������������������������������������������������������������������ 229, 233
N-glycosylation������������������������������������������27, 29, 36, 38, 39,
209–211, 215, 217, 219–221, 228, 229

Paula Meleady (ed.), Heterologous Protein Production in CHO Cells: Methods and Protocols, Methods in Molecular Biology,
vol. 1603, DOI 10.1007/978-1-4939-6972-2, © Springer Science+Business Media LLC 2017

251
 Heterologous Protein Production in CHO Cells: Methods and Protocols
252  Index
N-linked glycosylation��������������������������������������������������26, 38
P Sialylation��������������������������������������������������������������� 13, 29–37
PCR
overlap-extension PCR������������������������� 58, 59, 62, 64–65
real-time PCR������������������������������������������������������������121
Phosphopeptide enrichment
IMAC�������������������������������������������������������������������������197
TiO2, 199
Phosphoproteomic�����������������������������������������������������������197
Polyethyleneimine (PEI)����������������������������������������������45, 47
Process optimization��������������������������������������������������� 87, 159
Productivity�����������������������������������80, 88, 102, 120, 121, 143
Protein folding�����������������������������������������1, 2, 11, 13, 39, 143
Protein translation����������������������������������������������������������������9
Protein solubilization��������������������������������������������������������188
Proteomics����������������� 154, 159, 160, 187, 196, 215, 218–219
R
Real-Time PCR��������������������������������������� 120, 121, 129–134
Recombinant protein production����������������������������� 209, 212
Recombinant proteins��������������������������������������������������������13
RNA-Seq�����������������������������������������������������������������169–186
S
Secretion��������������������������������������������������������� 2, 12, 148, 149
Selection����������������������������������������������������������� 14, 35, 71, 73
Short interfering RNA (siRNA)���������������������� 30, 37, 73, 75
Site specific phosphorylation��������������������������������������������197
Sort�����������������������������105, 113, 116, 144, 147, 150, 152, 176
Suspension adaptation��������������������������������������������������������77
Systems biotechnology�������������������������������������������������������40
T
TET-ON���������������������������������������������������� 15, 89, 90, 93–98
Tetracycline (Tet)��������������������������������������������������� 15, 88, 89
Therapeutic proteins������������������� 1, 33, 35, 36, 39, 71, 73, 80
Transcriptomic������������������������������������������������� 180, 196, 219
Transfection�����������������������������������������������2, 6, 12, 14, 45–55
Transient gene expression (TGE)����������������������� 6, 45, 62, 67
V
Vector design������������������������������������������������������������������6, 18
Vector design elements
Matrix associated regions (MARs)�����������������������������7, 8
ubiquitous chromatin opening elements
(UCOEs)��������������������������������������������������������������7, 8
W
Western blotting�������������������������������������14, 74, 79, 115, 123,
127–128, 244, 246