“All In Transcription-modus assembly”

- a new regulatory model for the eukaryotic cell cycle.

Author: Bjørn Sponberg M.Sc. (Stockholm University)
Abstract

The detection of weakly expressed mRNA's in time-series eukaryotic cell cycle experiments have been a challenge with micro-array technology.
Transcription factors belongs to the genre of weakly expressed genes.
Due to the detection problem in micro-array experiments it becomes difficult to map the regulatory network of transcription factors into their
respective cell cycle phases - from in vivo experiments.
To compensate for this, the temporal placement of cell cycle involved transcription factors is achieved by ChIP-chip experiments. In vivo ChIP-
chip experiments reveals cell cycle transcription factors target genes – which are not weakly expressed during the cell cycle.
By implementing the assumption of “Just-In-Time synthesis” during the eukaryotic cell cycle, results from these ChIP-chip experiments are used
to position transcription factors expression profiles along the cell cycle. Hence the transcription factors expression profiles are interpolated to the
same cell cycle phases as their target genes based on a “Just-In-Time synthesis” behavior. However, believing without seeing causes a restless
struggle to truly detect transcription factor synthesis during the cell cycle - in vivo.

One attempt to prove “Just-In-Time synthesis” activity in cell cycle has been to create customized micro-arrays. These micro-arrays eliminates
strongly expressed cell cycle genes from its probe array, while keeping the weakly expressed probes. This to better detect the weakly expressed
genes in the Yeast cell cycle. This work confirmed key transcription factors to show an oscillating behavior during the Yeast cell cycle.
I downloaded and processed their published micro-array signal intensities stemming from the control dataset. This dataset consisted of 105 genes
known in prior not to be cell cycle regulated. The result from this analysis showed that the control dataset oscillated in a similar manner as the
average cell cycle expression profile does. This indicates that noise could be a contributing factor and weaken some of their conclusions performed
on their sample dataset. Some of these conclusions was devastating to the All-In-G1 hypothesis (Now: All In Transcription-modus – see under).

One other cell cycle experiment on Fission yeast combines time-series gene expression data with flow cytometry data. This approach measures
DNA content distribution at the same time points as mRNA expression during the fission yeast cell cycle.
When comparing the two data sources a pattern emerged. Condensed clusters of transcripts was detected immediately after sister chromosome
separation in late M phase.
Figure 1 (Over). Montage of fluorescence micrographs Figure 2 (Under). All In Transcription-modus
The reason for this behavior is hypothesized to be caused by the first principal argument described in the All In Transcription-modus regulatory showing the general organization of chromosomes and illustration. Three alternative sources of transcription
model. microtubules during mitosis in Eukaryotic cells. Hela cells during the eukaryotic cell cycle is suggested. Correct
In addition, their raw micro array data shows a characteristic signal splitting pattern in the genomes division modus which could support the (human cervical cancer cells) were stained with anti- assembly of the transcriptional machinery complex
acetylated a-tubulin (red, microtubules) and with Hoechst needs all three regulatory levels: 1) DNA level, 2)
hypothesis. (blue, DNA). Green line during the cycle illustrates when Chromosome level and 3) Nucleus level [Modified
the genome is in a transcription modus. Red line from 1].
The old hypothesis title [3] changed during this work. According to the hypothesis the eukaryotic genome must have two possible regulatory illustrates when the genome is in a division modus.
[Picture credits: Tomasz Szul/Visuals Unlimited, Inc. (This
modus-es during the cell cycle. 1) A transcription modus that lasts from chromosome separation in late M phase - to beginning of S phase. 2) A picture is modified by the posters author)]
division modus lasting from beginning of S phase to chromosome separation in late M phase. This new approach has changed the hypothesis name
from “All-In-G1 assembly” to “All In Transcription-modus assembly”. The new name is a more precise formulation of what the hypothesis tries to
describe in vivo.

Conclusion of the poster: The “All In Transcription-modus assembly” model postulates that the eukaryotic genome has two possible regulatory
modus-es during the cell cycle: 1) A transcription modus lasting from chromosome separation in late M phase to the beginning of S phase, in
which all transcription machinery assembly processes during the cell cycle takes place, and 2) A division modus lasting from the beginning of S
phase to chromosome separation in late M phase, in which the genome loses its transcription regulatory levels to fully support division regulation.

The All In Transcription-modus assembly model

First principal argument (Directional selection):

.Each individual cells genetic contribution to its progeny cells will be awarded relative to natures definition of fitness. Consequently, the first eukaryotic lifeforms 1 billion years ago
were forced to compete by mechanisms that increased cell division speed, since all eukaryotes was unicellular. The All In Transcription-modus regulatory model would have won over
the current “just-in-time synthesis” model in these evolutionary battles. This since it optimizes cell division speed better than “just-in-time synthesis”.

Second principal argument (Genome robustness):

The current “just-in-time synthesis” model would be forced to combine two genome processes (Or modus-es): The transcription modus and the division modus. This is a more
complex solution. In cases were such mixed moduses would be able to compete with cycle-speed against All In Transcription-modus, the mixed-model would lose due to lower
genome robustness. Thus, lower numbers of progeny cells would survive compared to cells using All In Transcription-modus. The net result would be lower fitness.

Third principal argument:

All molecular mechanisms predicted in All In Transcription-modus have shown to be possible in wetlab experiments. In particular microRNA mediated transport of mRNA for
storage in cytosol. Thus, the All In Transcription-modus regulatory model have been an option to the cells during evolution [1-3].
 Result

In 2004 Lichtenberg et al [4] used their own customized micro array probes to
detect weakly transcribed genes. Probes that represented known strongly
expressed cell cycle regulated genes was removed from the array.
Probes was constructed to capture known and putative weakly expressed cell
cycle genes in Yeast cells (Saccharomyces cerevisiae). In this paper the
oscillating pattern to many transcription factors was reported to peak while the
genome was in its division modus. According to All In Transcription-modus
this would not be possible since transcription factors should not be needed
until the end of the cycle. In particular; some reported unique discoveries of
transcription factor transcripts in mid S phase is suspected to be a result of
noise. Eg. SWI6, ZMS1 and REB1.

In the same year Rustici et al [5] performed a similar cell cycle experiment as
in Lichtenberg et al. This time the micro-array design was not custom made,
and the model organism was fission Yeast rather than budding Yeast. Unique
to this experiment was that DNA content measurement was performed in
parallel with the time-series gene expression profiles (Figure 4 – left). Since
we were interested in proving/disproving gene transcription activity in the
genomes division modus this paper was scrutinized. From figure 4 it seems
that a transcription burst emerge immediately after chromosome separation
begins in M phase. To follow the principal arguments in All In Transcription-
modus there should exist a certain time point in Anaphase (Chromosome Figure 3. By taking the average signal intensities for all the 105 control Figure 4. It is interesting to observe that transcription activity in the cell cycle (Right) correlates with separation of
separation takes place) in which the genome exits its division modus to enter genes (Not cell cycle regulated) the average cell cycle genes oscillation sister chromosomes (Left). Transcription seems to take place only in the genomes transcript modus – except for
its transcription modus. This since after chromosome separation the purpose pattern emerged. This could indicate that noise is driving much of the data cluster 4. Consequently the All In Transcription-modus model predicts these genes to be so called “ecin genes”
of chromosome condensation and organization disappears. The main job is reported for the “never seen oscillating weakly expressed genes” in (Genes that are transported to cytosol for storage). Left figure: DNA content in grey [Modified from 5].
done – to separate the two sister chromosomes in a clean and ordered
Lichtenberg et al.
manner. Hence, the genome can very quickly fall back into a transcription
modus, since it now goes from order-to-chaos in contrast to - chaos-to-order.
The genome can thus quickly regain all three regulatory levels. After
chromosome separation the genome can once again focus solely on
regulating transcription. That is, the genome has exited its division modus and
entered its transcription modus. In conclusion, the data from Rustici et al in
figure 4 is in accordance to the principal arguments of the proposed All In
Transcription-modus regulatory model.

Figure 6. Weblogo representing 100 individual 3'UTR regions in suspected ecin genes (Upper) and not ecin
genes (Lower). Both datasets are cut 100 bases upstream of the 3'UTR end.

Methods

The control dataset in the Lichtenberg paper was downloaded from:

http://www.cbs.dtu.dk/cellcycle/yeast_validation/deLichtenbergetal_control_genes.txt

The micro array dataset was opened in a spread sheet. Averages of each 16 sample values (One sample per 20 minutes per gene) was taken. The graph was generated in the same spread sheet.

The suspected ecin and not ecin datasets were downloaded from Ensemble biomart:

http://www.ensembl.org/biomart/martview/b31952e04962a02bded1f8544711dd74

GO annotation codes for genes in cell cycle mechanisms typically for G2 function were downloaded and were referred to as the ecin dataset. Go annotation codes for mechanisms typically for G1/S function was downloaded
and referred to as the not-ecin dataset.

The download process was done in two steps:

The GO annotation codes was first found with AmiGO here: http://www.geneontology.org/ . Search words for ecin genes were typically G2, prometaphase, metaphase. Search words for ecin genes was typically DNA
replication, G1/S phase. The GO codes was pasted into the “Gene ontology” field in biomarts “Filters”. In Biomarts “Attributes” field: Sequences and 3'UTR were selected.

To construct the weblogo a BioPython script was used. It was constructed by first cutting the 3'UTR regions 100 bases upstream of 3'UTR end. These subsequence datasets were the source for weblogo generation.

References

1. Björn Sponberg. Regulatory motifs in cell cycle regulated genes. Unpublished 2009.

2. Richard E. Lenski#, Judith A. Mongold, Paul D. Sniegowski1, Michael Travisano, Farida Vasi, Philip J. Gerrish & Thomas M. Schmidt. Evolution of competitive fitness in experimental populations of E. coli: What
makes one genotype a better competitor than another? Antonie van Leeuwenhoek 73: 35–47, 1998.

3. Björn Sponberg. http://www.scribd.com/doc/30189257/Hypothesis-All-In-G1-assembly

4. Ulrik de Lichtenberg, Rasmus Wernersson , Thomas Skøt Jensen, Henrik Bjørn Nielsen,
Anders Fausbøll, Peer Schmidt, Flemming Bryde Hansen, Steen Knudsen and Søren Brunak.
New weakly expressed cell cycle-regulated genes in yeast. Yeast 2005; 22: 1191–1201.

5. Gabriella Rustici, Juan Mata, Katja Kivinen, Pietro Lió, Christopher J Penkett, Gavin Burns,
Jacqueline Hayles, Alvis Brazma, Paul Nurse & Jürg Bähler. Periodic gene expression program of the fission yeast cell cycle Nat Genet. 2004 Aug;36(8):809-17. Epub 2004 Jun 13.

Figure 5. Raw microarray data from the Rustici et al experiment. An unmistakable pattern emerges in mid G2 phase. Is the mRNA detected twice?
First on their way to cytosol storage and secondly after cytosol release (Thank you to Erik Sonnhammer for helping with this detection) [Modified
from 4].