Analytical Biochemistry 462 (2014) 2931

Contents lists available at ScienceDirect

Analytical Biochemistry
journal homepage: www.elsevier.com/locate/yabio

Notes & Tips

Quantitative pH assessment of small-volume samples using a universal

pH indicator
Jeffrey D. Brown a,, Nathaniel Bell a,1, Victoria Li b,1, Kevin Cantrell b
a
b

Department of Biology, University of Portland, Portland, OR 97203, USA

Department of Chemistry, University of Portland, Portland, OR 97203, USA

a r t i c l e

i n f o

Article history:
Received 12 December 2013
Received in revised form 26 May 2014
Accepted 2 June 2014
Available online 11 June 2014
Keywords:
Isoelectric focusing
pH indicator
pH determination
Wnt

a b s t r a c t
We developed a hue-based pH determination method to analyze digital images of samples in a 384-well
plate after the addition of a universal pH indicator. The standard error of calibration for 69 pH standards
was 0.078 pH units, and no sample gave an error greater than 0.23 units. We then used in-solution isoelectric focusing to determine the isoelectric point of Wnt3A protein in conditioned medium and after
purication and applied the described method to assess the pH of these small-volume samples. End users
may access our standard to assay the pH of their own samples with no additional calibration.
2014 Elsevier Inc. All rights reserved.

Although pH determination using standard electrodes is useful

for many applications, electrode fouling, small sample volume, and
a desire to not contaminate an electrode with surfactant-containing analytes can limit the utility of this approach. In such cases,
universal pH indicator solutions may be used. These multichromatic pH indicators consist of a dye cocktail that undergoes a continuous change in color as the solution pH changes. pH assessment
relies on a qualitative comparison between the solution color and
a color reference card. Due to the subjective nature of such assessments and variability among users, pH determination using
universal indicators can be unreliable; the resolution, reproducibility, and precision of such determinations are subject to substantial
and variable user errors that limit the condence of pH determination. Other methods rely on spectrophotometric measurements of
the absorbance at a particular concentration of pH indicator (e.g.,
see Ref. [1]). These methods, although accurate, require expensive
equipment, large sample volume, and extensive sample preparation. To address these shortcomings, we developed a quantitative
hue-based method to assay the pH of small (100 ll) samples in a
384-well plate after the addition of a universal pH indicator.
Ubiquitous digital cameras or scanners are used to capture RGB
(redgreenblue)2 color intensities in TIFF or JPG digital images.
Corresponding author. Fax: +1 503 943 7784.
E-mail address: brownje@up.edu (J.D. Brown).
These authors contributed equally to this work.
Abbreviations used: RGB, redgreenblue; IEF, isoelectric focusing; pI, isoelectric
point; BSA, bovine serum albumin.
1
2

http://dx.doi.org/10.1016/j.ab.2014.06.001
0003-2697/ 2014 Elsevier Inc. All rights reserved.

RGB values are converted into HSV or HSL color spaces, which quantify and separate the color, or hue, from the other parameters such as
saturation and lightness. The hue is measured on a cylindrical coordinate system (a color wheel) and, due to its insensitivity to concentration and illumination variations, is a robust measure of the optical
properties of color-changing indicators [2]. The relationship between
the easily measured hue and the pH is quite stable for a particular pH
indicator cocktail, and once established the hue values from any digital image can be transformed into a good estimate of pH. Investigators employing the same commercial pH indicator solution used in
our work can apply our hue calibration data with no additional
manipulation to assess the pH of their small-volume samples. Due
to the reliable nature of the hue parameter, this approach provides
a transferable, rapid, inexpensive, and easily scalable means to
determine the pH of numerous small fractions using only a consumer-grade digital camera or scanner and data processing using
free and/or common software.
To generate a standard curve that relates hue and pH, we prepared a buffer solution containing a (1:50 dilution of) universal
pH indicator (Fluka, cat. no. 36828, SigmaAldrich, St. Louis, MO,
USA) and monitored pH with a Ross micro pH electrode (Thermo
Scientic, cat. no. 8220BNWP, ThermoFisher Scientic, Waltham,
MA, USA) as we titrated the solution with acid or base. Images of
the buffer solution were acquired approximately every 0.1 pH unit
with a Canon EOS Rebel T1i digital SLR camera. Other work demonstrated the efcacy of various cameras or scanners in generating
the images used in the hue-based approach [3]. Images of these
standards were processed in MATLAB and used to calculate the

30

Notes & Tips / Anal. Biochem. 462 (2014) 2931

coefcients of a sigmoidal curve t that relates the measured hue

to the pH of the solution, as shown in Fig. 1. This standard curve
was used in subsequent analyses and is transferable to other digital
devices and applications using the same indicator cocktail. The
freely available application ImageJ (http://rsbweb.nih.gov/ij) can
be used to manually select a region of interest in an image and calculate the mode of the hue values in the selection as described in
the supplementary protocol (see online supplementary material).
A simple Excel spreadsheet (http://wordpress.up.edu/phindicator/
excel) can be used to calculate the pH based on this hue using
the pre-established curve t parameters that are usable with any
image of the Fluka indicator cocktail. A MATLAB script that identies all of the lled wells in an image and automatically calculates
the pH in each one from a single image is available on request.
To validate this pH assessment method, we analyzed fractions
generated by a MicroRotofor free-solution, carrier ampholytebased isoelectric focusing (IEF) system (Bio-Rad Laboratories,
Hercules, CA, USA) to characterize the isoelectric point (pI) of
the bovine serum albumin (BSA), Wnt3A, and a soluble Frizzled8CRD-IgFc fusion protein [4] under native conditions. The
MicroRotofor yields 10 approximately 250-ll fractions, and the
pH of these samples must be assessed in order to understand
the pI of the proteins that focus to the fractions. Although this
apparatus is not well suited to high-resolution pI determination,
the 3-ml focusing chamber can accommodate a relatively large
analyte volume, and recovered fractions are suitable for subsequent analyses, including functional assays. IEF analysis can indicate post-translational modications that alter a proteins charge
[5] or biomolecule complex formation [6].
Initial assays were performed using broad-range ( pH 3.59.5)
Bio-Rad BioLyte 3/10 ampholytes (cat. no. 163-1112), and pH was
monitored with a Ross micro pH electrode. As shown in Fig. 2, puried Wnt3A focused to approximately pH 9.0 fractions, consistent
with a predicted pI of 8.26 (http://web.expasy.org/compute_pi;
see Ref. [7]), whereas unpuried Wnt3A protein in conditioned
medium from expressing cells focused to an approximately pH
5.0 fraction. Post-translational modication by anionic groups,
including protein phosphorylation [8] and sialic acid-terminated

1.2

Hue

0.8

0.6

0.4

0.2

10

pH
Fig.1. Hue/pH standard curves. A beaker containing universal pH indicator solution
and a buffer salt (either tris or potassium hydrogen phthalate) was titrated with HCl
(tris) or NaOH (KHP) addition. A Ross micro pH electrode was used to monitor
buffer pH, and an image was captured using a digital SLR camera approximately
every 0.1 pH unit. A second data set, shown in gray, was generated in the presence
of 1% Chaps detergent. Hue values were determined with the origin of the
cylindrical coordinate system set to the blue primary (rather than traditional red) to
preserve a continuous function over the range of colors observed. The curve t is a
sigmoidal t to a Boltzmann equation with 10 adjustable parameters (3 inection
points).

Fig.2. Wnt3A and Frizzled8CRD-IgFc isoelectric focusing (IEF). Western blot

analysis for Wnt3A in IEF fractions using broad-range Bio-Rad BioLyte 3/10
ampholytes revealed that Wnt3A protein in conditioned medium samples (CM)
focuses to a low-pH fraction (fraction 3, pH 5.0), whereas puried Wnt3A in the
presence of Chaps detergent (pure)required to maintain puried Wnt3A solubilityaccumulates in approximately pH 9.0 fractions. Including Chaps with Wnt3A
conditioned medium results in both low-pI and high-pI pools of Wnt3A
(CM + detergent). Bio-Rad BioLyte 5/7 ampholytes were used to rene the low-pI
fraction of Wnt3A in conditioned medium (Wnt3A CM [pH 5.07] by the hue-based
method) and a soluble Frizzled8CRD-IgFc fusion protein (Fz8CRD-Ig [pH 4.93] by
the hue-based method).

oligosaccharides [9], can decrease protein pI; however, such pI

shifts tend to be on the order of <0.2 pH units. Wnt proteins bind
to the anionic glycosaminoglycan heparin [10], but pre-incubation
of puried Wnt3A protein with heparin did not cause the protein
to focus to low-pH fractions (not shown). The addition of 1% Chaps
detergent to Wnt3A conditioned medium during IEF resulted in a
substantial shift from an approximately pH 5.0 fraction (fraction
3) to approximately pH 9.0 fractions where puried Wnt3A
focuses. Gross and coworkers [11] reported that secreted Wnt proteins are associated with extracellular membrane structures
known as exosomes. Taken together, these data are consistent with
the hypothesis that exosome-associated Wnt3A focuses to low-pH
fractions and the addition of detergent results in extraction of
Wnt3A from the exovesicles, allowing the protein to focus to
high-pH fractions. The Rotofor system has been reported to fractionate exosomes and their associated proteins [12]; thus, this
strategy may prove to be generally useful for distinguishing
between exosome-associated and soluble pools of other proteins.
We then used a BioLyte 5/7 narrow-range carrier ampholyte
mixture (Bio-Rad Laboratories, cat. no. 163-1153) to fractionate
Wnt3A in conditioned medium as well as a puried soluble fusion
protein of the Frizzled8 Wnt receptor (Frizzled8CRD-IgFc) [13]. pH
was monitored with a micro pH electrode before and after the
addition of 10 ll Fluka universal pH indicator solution (Fluka, cat.
no. 36828) to 150 ll of each fraction. Indicator addition resulted
in a 60.02-unit shift in solution pH. Samples (100 ll) were loaded
into the wells of a 384-well plate (Nalge Nunc, cat. no. 242757,
Peneld, NY, USA) and were imaged with a Canon EOS Rebel T1i
DSLR camera. pH determinations by universal indicator hue and
Ross micro pH electrode were in good agreement. Peak fraction
pH for Wnt3A was determined to be 5.07 (hue method) or 4.87
(Ross pH electrode), and the Frizzled8CRD-IgFc fusion protein peak
fraction was 4.93 (hue method) or 4.88 (Ross pH electrode). The
standard error of pH determination using the described hue
method was 0.2 pH units in both of these assays when all 10 fractions were considered. The addition of the universal indicator did
not affect protein activity; after pH assessment, the Frizzled8CRD-IgFc fractions were active as assayed by Wnt3A binding.

Notes & Tips / Anal. Biochem. 462 (2014) 2931

We also subjected BSA to MicroRotofor fractionation. Fraction pH

was monitored with the Ross pH electrode and the universal pH
indicator as described. Electrode and hue methods were in good
agreement, and BSA focused to approximately pH 5.0 fractions
(data not shown), consistent with the reported pI of the native protein [13].
We noticed a substantial depression in hue-based pH calculation for pH > 7.0 samples from MicroRotofor runs that included
1% Chaps detergent. This deviation was not observed using the
Ross micro pH electrode, indicating that the universal indicator
suffers from Chaps-dependent error in this range. The deviation
may be analogous to the protein error exhibited by some pH indicator dyes used for protein quantitation assays. Suzuki [14]
reported that inclusion of non-ionic detergents in such assays
results in an upward shift in the pH required for a chromatic
change of the protein-associated dye molecule. A similar phenomenon affecting a dye in the universal indicator in the presence of
Chaps could result in the observed depression in the hue-based
pH determination. To quantify this surfactant effect, a second standard curve was generated in the presence of 1% Chaps (Fig. 1) and
reveals a substantial shift relative to the surfactant-free curve.
Although the technique that we described here can be applied
to most aqueous samples, deeply colored analytes (e.g., chromoproteins) and carrier ampholytes can be a source of signicant
errors; however, the Bio-Rad BioLyte ampholytes used in our
experiments have little color at working concentration, and the
standard error of pH assessment using the hue method was only
0.2 pH units for analyzed fractions. In addition, hue-based pH
assessment of BSAwhich has a modest reddish-brown huedid
not suffer from any greater error from electrode-based pH assessment than other proteins.
In summary, we developed a quantitative method to assess the
pH of small-volume samples that is inexpensive, rapid, and nondestructive and that scales easily to numerous analytes. Although
our individual experiments generated only 10 fractions, we routinely stored frozen fractions and prepared several dozen samples
for hue-based pH assessment in a single plate. This method will be
of particular utility to assay large numbers of fractions generated
by techniques that include chromatofocusing and free-ow IEF.
We then used a Bio-Rad MicroRotofor IEF apparatus to document
a dramatic pI shift in Wnt3A protein during purication, a difference that may correspond to the inclusion of Wnt3A in exosomes
before purication in the presence of detergents. Finally, we
uncovered a substantial Chaps-dependent error in pH assessment
using a universal indicator in solutions of pH > 7.0 and recommend
caution when using universal indicators in the presence of
surfactants.

31

Acknowledgments
The authors thank the University of Portland for supporting this
work. This work was funded in part by M.J. Murdock Life Sciences
(Grants 2008354 and 2010196). The authors thank Brianna Brown,
Calli VanderWilde, Michelle Thomas, Blair Pearson, and Keri Jackson for their assistance in these investigations.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.ab.2014.06.001.
References
[1] H. Yamazaki, R.P. Sperline, H. Freiser, Spectrophotometric determination of pH
and its application to determination of thermodynamic equilibrium constants,
Anal. Chem. 64 (1992) 27202725.
[2] K. Cantrell, M.M. Erenas, I. de Orbe-Pay, L.F. Capitn-Vallvey, Use of the hue
parameter of the hue, saturation, value color space as a quantitative analytical
parameter for bitonal optical sensors, Anal. Chem. 82 (2010) 531542.
[3] M.M. Erenas, K. Cantrell, J. Ballesta-Claver, I. de Orbe-Pay, L.F. CapitnVallvey, Use of digital reection devices for measurement using hue-based
optical sensors, Sens. Actuators, B 174 (2012) 1017.
[4] J.-C. Hsieh, A. Rattner, P.M. Smallwood, J. Nathans, Biochemical
characterization of Wnt-Frizzled interactions using a soluble, biologically
active vertebrate Wnt protein, Proc. Natl. Acad. Sci. U.S.A. 96 (1999) 3546
3551.
[5] S.Q. Tia, K. Brown, D. Chen, A.E. Herr, Protein post-translational modication
analyses using on-chip immunoprobed isoelectric focusing, Anal. Chem. 85
(2013) 28822890.
[6] P. Kempn, R. Ciollone, L. Villacorta, R. Ricciarelli, J.-M. Zingg, Isoelectric point
mobility shift assay for rapid screening of charged and uncharged ligands
bound to proteins, IUBMB Life 55 (2003) 103107.
[7] B. Bjellqvist, B. Basse, E. Olsen, J.E. Celis, Reference points for comparisons of
two-dimensional maps of proteins from different human cell types dened in a
pH scale where isoelectric points correlate with polypeptide compositions,
Electrophoresis 15 (1994) 529539.
[8] S. Barrabs, A. Sarrats, E. Fort, R. De Llorens, P.M. Rudd, R. Peracaula, Effect of
sialic acid content on glycoprotein pI analyzed by two-dimensional
electrophoresis, Electrophoresis 31 (2010) 29032912.
[9] K. Zhu, J. Zhao, D.M. Lubman, F.R. Miller, T.J. Barder, Protein pI shifts due to
posttranslational modications in the separation and characterization of
proteins, Anal. Chem. 77 (2005) 27452755.
[10] K. Willert, J.D. Brown, E. Danenberg, A.W. Duncan, I.L. Weissman, T. Reya, J.R.
Yates III, R. Nusse, Wnt proteins are lipid-modied and can act as stem cell
growth factors, Nature 423 (2003) 448452.
[11] J.C. Gross, V. Chaudhary, K. Bartscherer, M. Boutros, Active Wnt proteins are
secreted on exosomes, Nat. Cell Biol. 14 (2012) 10361045.
[12] M.W. Graner, O. Alzate, A.M. Dechkovskaia, J.D. Keene, J.H. Sampson, D.A.
Mitchell, D.D. Bigner, Proteomic and immunologic analyses of brain tumor
exosomes, FASEB J. 23 (2009) 15411557.
[13] K. Wallevik, SS-interchanged and oxidized isomers of bovine serum albumin
separated by isoelectric focusing, Biochim. Biophys. Acta 420 (1976) 4256.
[14] Y. Suzuki, Protein error of pH indicators in the presence of detergents, Anal.
Sci. 23 (2007) 733738.