Nonadaptive onset of complex multicellularity

Many bacteria and eukaryotes (the latter having a nucleus and organelles in their cells) are exclusively or predominantly unicellular, but a considerable fraction of both empires show facultative multicellularity with moderate division of labor among cells. There are two main ways to arrive at a multicellular state: either by aggregation of solitary cells (as, for example, in slime molds) or by developing a multicellular body from one cell (as in a human embryo) by cell division and lack of separation of the offspring cells (1, 2). The latter is the clonal path to multicellularity, a term that also reflects on the maximal genetic relatedness of constituent cells (3). The emergence of multicellularity, especially in its complex forms (with several differentiated cell types) is among the major transitions in evolution, characterized by three main features: increase in complexity, the establishment of higher-level evolutionary units from lower-level ones (with the capacity to multiply, have heredity and variability), and the evolution of novel ways to store, transmit, and use hereditary information (4, 5). Although when one counts all instances of the origins of all kinds of multicellularity it is in the range of dozens, complex forms of multicellularity are much rarer, represented by plants (including green algae), animals, fungi, brown, and red algae. All of them are eukaryotes with clonal development. Why is there no bacterial analog? In PNAS, Bingham and Ratcliff (6) offer a stimulating hypothesis to help explain this conundrum.

Eukaryotes have cells that are considerably more complex (by any account) than those of prokaryotes. It is an obvious thought that at least some cellular features could have played an important causal role in the evolution of multicellular complexity. Eukaryotic genes sit on elaborate chromosomes, their regulation (i.e., being on and off) is complex, and their transcription is separated from translation (the synthesis of proteins) by the nuclear membrane. The latter fact allows for many gimmicks, often of high regulatory significance, at the transcribed RNA level that would be impossible in a bacterial cell. Is that all? Lynch suggested an answer resting on population genetics (7, 8). While many of the changes that increase genetic complexity can be harnessed by natural selection later, the onset of complexity increase can often be detrimental (as disturbance), unless population size is sufficiently low. The reason is that with lower population size, chance begins to dominate over selection, and mildly deleterious genetic changes are not “seen” by natural selection anymore, and are, therefore, not eliminated. In other words, the onset of complexity increase is nonadaptive. On average, eukaryotic population sizes are below those of prokaryotes, and multicellular eukaryotes enjoy even smaller population sizes. So, some of the cellular and genetic features are likely to have originated without any immediate function.

The PNAS paper (6) develops these ideas further. First, there is the observation that at low population size, prokaryotic genomes tend to shrink and eukaryotic genomes are more likely to keep their genome size or even expand it (9, 10). Of course, the latter helps complexity increase. The other factor is that clonal development decreases population size considerably and all complex multicellular lineages are clonal! So, eukaryotic, clonal multicellular lineages are expected to be more likely to step into the complex realm than other lineages (Fig. 1). Bingham and Ratcliff illustrate their idea by a simulated population (6). They assume, for the sake of the argument, a best-case scenario for the evolution of multicellularity under the assumptions that increased genome size is adaptive, it correlates with organismal size, and larger organisms enjoy more benefit from larger genome size than smaller organism. The ratio of genome deletion to expansion was taken to be 5 and 0.8 for prokaryotes and eukaryotes, respectively. They assume that the carrying capacity is fixed in terms of the total number of cells. Hence bigger organisms will be represented by fewer individuals in the population. The outcome is that prokaryotes reach a complexity ceiling at markedly smaller genome and organism size than eukaryotes. Note that this simulation does not deal with the frequent non-adaptive (likely mildly deleterious) consequences of the original deletions and insertions, as discussed above, and this aspect warrants further modeling.

Fig. 1.

Gene loss and gain, organism and population size, and developmental complexity in eukaryotic evolution toward complex multicellularity. Left of the green line we see the onset of complex multicellularity. The combination of decreasing population size, the propensity for gene gain (G) with clonal development (not shown) favors the fixation of increasing developmental complexity in certain lineages. Subsequent evolution (right of the green line) can happen without net gene gain and even with rampant gene loss. Colors of the filled circles indicate cell differentiation.

Eukaryotic, clonal multicellular lineages are expected to be more likely to step into the complex realm than other lineages.

A caveat is that we do not yet have a full causal understanding of the difference in the deletion: insertion bias in the two empires of life. Another complication is that the realized deletion:insertion ratio is far from constant, according to the phylogenetic analysis of protein-coding genes. Consider animals (11, 12). The node leading to the Metazoa shows an almost balanced incidence of gains and losses of protein homology groups (both over 2000). The lineage leading to the Bilateria (Metazoa without cnidarians, sponges, comb jellies, and “flat animals”) shows many more gains than losses. Spectacular gene group loss happened with the origin of Ecdysozoa (nematodes, tardigrades, and arthropods) and Deuterostomes (a clade that includes also the vertebrates). The pattern of gain and loss in Fungi is no less exciting. Crudely, there are those fungi with macroscopic bodies (Dikarya) and the rest. Duplications of genes happened in five large bursts, whereas gene loss seems to be continuous (13). This quasi-monotonous loss of ancestral groups explains why non-Dikarya fungi show more resemblance to their related protists than early-branching animals show to theirs, suggesting that Fungi have retained more surviving clades from early evolutionary experimentation in multicellularity than animals. The branch leading to chytrids and other fungi was characterized by a high insertion:deletion ratio, and the origin of the Dikarya clade shows more losses than gain. Balanced gene turnover (when deletions and insertions are approximately equally frequent) as well as net gene loss can thus also be constructive. “Losses are compensated by gene gains and duplications in each lineage independently, and thus the size of the gene repertoire is not necessarily altered. Consequently, an increase in morphological complexity must result from other evolutionary phenomena, such as co-option following gene duplication, integration of genes in functional modules, or through the potential implication of non-coding sequences leading to an increase of genomic regulatory complexity” (14, p. 530). It is important to bear in mind that experimental gene loss followed by compensatory mutations of resident genes (i.e., without gene gain) has led to the emergence of morphological novelties in budding yeast, including multicellularity (15).

Phylogenetic information thus suggests that a refined model with non-uniform genome expansion rates is welcome, but it seems that close to the root of the complex multicellular clades gene gain and turnover (close to the assumptions of Bingham and Ratcliff) was more common than in later phylogeny, when rampant gene loss happened multiple times. Clearly, the latter case (Fig. 1) is not covered by the simple model under discussion.

A major remaining task is to assess the relative contributions of enabling molecular–cellular constraints (sensu Kauffman, ref. 16) entailed by the eukaryotic condition (such as the nucleus and the cytoplasm being separated by the nuclear membrane, and a complicated cytoskeleton) versus the population-level phenomena discussed in this commentary. Finally, although a high propensity for gene loss is demonstrated for prokaryotes, horizontal gene transfer can reload genes, which is hypothesized to be one of the selective advantages in favor of bacterial sexuality (17). Since bacteria are still around, gene loss must be compensated by gene gain at the population level in the long run. Could it be that in the lab combination of low population numbers with unnaturally massive genetic transformation (leading to net gene gain and recombination) might promote an increase in complexity in already multicellular, clonal bacteria (such as cyanobacteria)? Or is the eukaryotic way of sex (meiosis and fusion of gametes) in some way also necessary for the emergence of complex multicellularity? The excitement about multicellular origins will surely prevail.