One of the most remarkable things about multicellular organisms is the differentiation of genetically identical cells into functionally specialized cell types. It’s difficult to say exactly how many cell types a given species has, since we would first have to say how different two cells need to be to count as different types. Nevertheless, it’s clear that there’s a wide range among different multicellular groups. Within animals, for example, placozoa have around five cell types, mammals over a hundred.
Amazingly, all of these very different cell types share a genome: your liver cells are pretty much genetically identical to your brain cells (and your skin cells, your kidney cells, your muscle cells…). The dramatic differences in form and function among all these cell types are mainly a result of differences in gene expression.
Volvox has just two cell types: a dozen or so big cells that are responsible for reproduction and one or two thousand smaller cells that bear the flagella that colonies use to swim:
Figure 1A from Matt & Umen 2017. Micrographs of an intact adult Volvox carteri spheroid with fully mature somatic and gonidial cells (left), isolated somatic cell (top right), and isolated gonidial cell (bottom right).
This was one of the main attractions for the researchers who developed Volvox as a model organism. With only two cell types, Volvox retains something close to its original form of cellular differentiation, making questions about how such differentiation evolved much more tractable.
We have long known about a handful of the differences in gene expression between Volvox gonidia and somatic cells, largely through the work of David Kirk and members of his lab at Washington University in St. Louis. In the ’80s and ’90s, when most of this work was done, comparing expression levels of even a handful of genes required an immense amount of work. Comparing the expression levels of ALL the genes in the genome was nearly inconceivable.
Advances in sequencing technologies have made it possible to do just that. Two recent papers, from two independent groups, have compared levels of gene expression across essentially the whole genome between Volvox gonidia and somatic cells. Interestingly, both papers were submitted to their respective journals within one day of each other (September 7 and 8) and published only a week apart (November 28 and December 5). This explains why neither paper cites the other despite covering very similar ground: the authors of the second paper couldn’t have seen the first before the final version of their own paper was submitted.
The first paper is from Benjamin Klein, Daniel Wibberg and Armin Hallmann at the University of Bielefeld in Germany (Whole transcriptome RNA-Seq analysis reveals extensive cell type-specific compartmentalization in Volvox carteri). The second is from Gavriel Matt and Jim Umen from Washington University in St. Louis and the Donald Danforth Plant Science Center, respectively (Cell-type transcriptomes of the multicellular green alga Volvox carteri yield insights into the evolutionary origins of germ and somatic differentiation programs). Gavriel Matt presented this work at the Volvox meeting back in 2015 (Volvox 2015: cell differentiation). Klein et al. is open access; Matt & Umen is not in an open access journal, but the online early version is not paywalled (for now).
Both studies used mechanical disruption (basically crushing Volvox colonies) to separate gonidia and somatic cells, then extracted RNA from each cell type separately. Both used RNA-Seq to quantify gene expression levels and compared them between cell types. RNA was extracted from a similar time point in both studies, shortly before the onset of cell division in the gonidia.
Figure 1B from Matt & Umen 2017. Diagram of the Volvox vegetative (asexual) life cycle. Center, two-day diurnal regime used to synchronize Volvox development with alternating 16hr light (open bars) and 8hr dark (closed bars) periods. Stages of the asexual life cycle are depicted proceeding clockwise from upper left (mature adult stage, black arrows indicate gonidial and somatic cell types). Red line indicates the period in the diurnal cycle when spheroids were harvested.
Broadly speaking, the main results of the two studies are similar. Both groups found that around half of all Volvox genes are differentially expressed between reproductive and somatic cells (49% in Matt & Umen, 54% in Klein et al.). However, Klein et al. found that these were roughly split between genes overexpressed in gonidia and those overexpressed in somatic cells, while Matt and Umen found about 1.7 times as many genes overexpressed in gonidia than in soma.
Some of Matt and Umen’s results are particularly interesting in the context of the evolution of multicellularity. They performed a phylostratigraphic analysis, that is, they identified Volvox gene homologs in other species to estimate the evolutionary age of the genes. The idea here is that a gene that is shared between Chlamydomonas and Volvox, but not with other green algae, is probably at least as old as the divergence between Chlamydomonas and Volvox, but not as old as the divergence between volvocine algae and other green algae. On the other end of the spectrum, Volvox nuclear genes that are shared with bacteria are probably very old, at least as old as the divergence between bacteria and eukaryotes.
In short, they found that
Compared to the entire transcriptome, gonidial genes were enriched for those with ancient origins (genes conserved across all cellular organisms) and those originating within green eukaryotes (Viridiplantae). In contrast, somatic genes showed a reciprocal phylostratigraphic pattern with a strong enrichment for lineage-specific genes that were found only in Volvox or only in the volvocine algae.
Figure 4A from Matt & Umen 2017. Phylostratigraphic profiling of gonidial and somatic genes. Bar graphs show the fraction of genes from each Volvox cell-type expression classification group in each phylostratum. Color key on right shows nested phylostrata progressing from the most recent phylostratum of Volvox-specific genes on the bottom to the most ancient phylostratum of genes found across all cellular organisms at the top. Vca = Volvox carteri; Gpe = Gonium pectorale; Cre = Chlamydomonas reinhardtii. Asterisk indicates significant enrichment compared to all 3 expressed genes (* false discovery rate < 0.01).
The overrepresentation of lineage-specific genes (Volvox-specific and volvocine-specific) in somatic cells suggests the importance both co-option of genes present in the unicellular ancestors of Volvox and genes that originated after the divergence of Volvox and Gonium:
Our germ-soma transcriptome data suggests that lineage-specific genes did play a large role in the evolution of cell type specialization in Volvox…Category 1 somatic cell genes (those exclusive to Volvox) evolved during the evolution of germ-soma differentiation, a trait found in two volvocine genera, Volvox and Pleodorina…Category 2 somatic cell genes (those that are pan-volvocine) presumably arose as genetic specializations in a unicellular ancestor and were subsequently coopted or re-deployed for somatic cell differentiation.
Another interesting finding is strong similarities between genes differentially expressed between Volvox cell types and those differentially expressed in light versus darkness in Chlamydomonas:
…we found extensive evidence for a general pattern of temporal-to-spatial expression cooption based on comparisons of Chlamydomonas cell cycle/diurnal cycle genes (likely similar to an ancestral unicellular program) and Volvox cell type genes… Importantly, the Volvox gonidial program was most closely associated with light-phase regulons in Chlamydomonas while the Volvox somatic program was most closely associated with dark phase regulons in Chlamydomonas—even though our cell-type transcriptome samples were taken during the light phase of a diurnal cycle.
Figure 9 from Matt & Umen 2017. Cell-type expression patterns of Volvox orthologs of diurnally expressed Chlamydomonas genes. Stacked bar plots show fractions of Volvox genes from each cell type expression classification whose corresponding orthologs in Chlamydomonas belong to the indicated diurnal cluster (all genes, c1-c18, or unclustered (uc) (Zones et al. 2015). Clusters are arranged in temporal order with peak expression within the light, light-dark transition, or dark phase of the diurnal cycle indicated below groups of clusters. Significant Volvox cell-type enrichment within each cluster is indicated as follows: ** false discovery rate < 0.01, * < 0.05.
This supports and even extends Aurora Nedelcu’s ideas about the importance of phenotypic plasticity in a unicell being co-opted for cellular differentiation in its multicellular descendants:
…a life-history gene (i.e., a gene that benefits survival while detracting from immediate reproduction—with the effect over the life cycle being beneficial) can become altruistic in the context of a multicellular group, if the beneficial effect of this gene is also beneficial to the group and if the cell as a group member expends more effort on this beneficial effect than would be optimal for its own survival and reproduction. As shown in figure 1G, this can be realized by shifting this gene’s expression pattern from a temporal context (within the same cell) into a spatial context (between soma and germ). In this way, a life-history gene that trades-off survival and reproduction in a unicellular individual can become an altruistic gene in a multicellular group and create the conditions for individuality at the higher level to emerge.
Figure 1G from Nedelcu & Michod 2006. Schematic representation of the change in expression pattern from a temporal context (environmentally induced) into a spatial context (developmentally induced) of a life-history gene in a unicellular individual as it becomes an altruistic gene in a multicellular individual.
I’ve only scratched the surface here; both papers have extensive analyses I haven’t covered. Armin Hallmann, the senior author of Klein et al., also has a blog post about that paper. Altogether, these new results represent a substantial step forward in understanding cellular differentiation in Volvox. The handful of genes previously known to be differentially expressed between cell types has been increased by about 200-fold. Furthermore, this new work raises new questions: of the thousands of genes now known to be differentially expressed between cell types, which ones are actually causing the differences between cell types, and which expression differences are caused by those differences? Of the ones that do cause phenotypic and functional differences, what are they doing? Of course, these questions are potentially important as new avenues of research.
Stable links:
Klein, B., Wibberg, D. and Hallmann, A. 2017. Whole transcriptome RNA-Seq analysis reveals extensive cell type-specific compartmentalization in Volvox carteri. BMC Biol., 15: 111. doi: 10.1186/s12915-017-0450-y.
Matt, G.Y. and Umen, J.G. 2017. Cell-type transcriptomes of the multicellular green alga Volvox carteri yield insights into the evolutionary origins of germ and somatic differentiation programs. G3 Genes, Genomes, Genet., doi: 10.1534/g3.117.300253.
Nedelcu, A.M. and Michod, R.E. 2006. The evolutionary origin of an altruistic gene. Mol. Biol. Evol., 23: 1460–1464. doi: 10.1093/molbev/msl016.
Thanks for highlighting our work Matt.
Besides the results you note in your blog, there are a couple additional things that readers may find useful in our manuscript (https://doi.org/10.1534/g3.117.300253).
—> Easy access to expression estimates for all Volvox genes and selected subsets of genes in our Supplemental Datasets
—> Statistical analysis to support conclusions for functional enrichment between cell types (e.g. Tables S3, S4)
—> Detailed investigation of cell-type expression for curated sets of flagella/basal bodies genes, photosynthetic genes, central carbon metabolism genes, ECM genes, and selected secretory pathway genes.
You mention in your blog post that Klein et al found an even split between numbers of somatic and gonidial genes while we observed a skew with more gonidial than somatic genes. Although expression estimates for most of the Klein data were not provided in their manuscript, we noticed two things that suggest a possible reason for the differences with our findings. First, comparing the whole transcriptome MA plots (fold changes on y axis, total expression level on x axis; Fig. S2 in our study, Fig. 3 in Klein et al) there is a more constrained dynamic range in the Klein data than in ours (64-fold max somatic/gonidia and 256-fold max gonidia/somatic in Klein et al; versus 4000-fold max ratio for both somatic/gonidia and gonidia/somatic in our data). Second, Fig. 5 in Klein et al shows generally weaker somatic biases in the RNA-seq expression ratios than in our data, or in the prior qRT-PCR study from the Hallmann lab that was used for comparison. For example, using the regA gene as a marker for cell-type specificity Klein et al found a 10-fold somatic expression bias (Fig. 5), whereas we found a >1000-fold expression bias (Supplementary Dataset S2), and a 150-fold expression bias was found in the prior study using qRT-PCR. A possible cause for these discrepancies is the presence of significant somatic cell contamination in the gonidial samples of Klein et al that dampened the expression ratios between cell types and may have somewhat altered the distributions/classifications of cell-type genes.