Evolution of multicellularity, bet hedging and
We evolve simple multicellular organisms in the lab
The evolution of multicellular organisms from unicellular ancestors was critical to the
evolution of large, complex organisms. While multicellularity has evolved more than 25
times independently during the last 3.5 billion years on Earth, the first steps in this
transition remain poorly understood. By evolving novel multicellularity in the lab, we find
that the transition to multicellularity may be less constrained than previously thought.
We exposed the unicellular yeast Saccharomyces cerevisiae to selection for rapid settling
through liquid media, favoring strains that evolved greater size. In all 10 replicate
populations, ‘snowflake’ yeast clusters (see left picture) evolved. Unlike flocculating yeast,
snowflake yeast adhere to one another only through mother-daughter cell adhesion,
resulting in genetic uniformity among the cells in the cluster.
Snowflake yeast display a novel multicellular life history: they have determinate growth,
reproduce by making multicellular propagules, and these propagules are functionally
juvenile (they have to grow to their parent’s size before
they can reproduce themselves).
We found that once snowflake clusters evolve, whole clusters become the unit of selection
(they either make it to the bottom of the tube and survive, or fail to do so and perish). As a
result, we see the evolution of multicellular traits, like changes in cluster morphology that
speed settling and growth, and even a simple form of division of labor through
programmed cell death (PCD). Our large-cluster forming snowflake yeast strains evolve
higher rates of PCD. Dead cells act as break-points within the cluster, allowing snowflake
yeast to regulate the number and size of offspring they produce. We demonstrated that
this can be adaptive in these two modeling papers (one and two). See some criticisms of
our 2012 PNAS paper and my responses here.
In our 2013 Evolution paper, we showed that cluster-level
adaptation takes place in several discrete phases. First, clusters
evolve to settle faster simply by increasing the number of cells per group. Next, we see an increase
in the size of individual cells (increasing cluster size), and the evolution of elevated programmed
cell death. Finally, we see the evolution of more hydrodynamic clusters that settle more efficiently.
We believe that this pattern reflects a key constraint: large clusters grow slowly (diffusion limitation),
favoring traits that increase settling speed without reducing growth rates.
So how hard is it really to get a snowflake yeast cluster? It turns out to be remarkably easy! In our
2015 Nature Communications paper, we show that knocking out a single gene (ACE2) is sufficient
to transform a unicellular yeast into a snowflake. ACE2 regulates daughter-cell separation after
mitosis, so this mutation causes daughter cells to remain attached to mother cells after budding. This simple
microevolutionary change turns out to have profound macroevolutionary
consequences. Snowflake yeast clusters are geometrically elegant. In fact, the
distribution of cells within a cluster follows predictions from Pascal’s triangle
(figure to the right). This simple growth form has a few important properties.
First, it results in clonal development, because snowflake yeast clusters
undergo unicellular genetic bottlenecks at reproduction, and the only way that
cells can join a cluster is to be born into them. This limits the potential for
genetic conflict to erode multicellular complexity. Second, it increases the
heritability of multicellular traits, exposing the cluster-level effects of every de
novo mutation to selection. Thus, knocking out a single gene both results in the
formation of clusters and also provides the basis upon which the ability to
evolve as a multicellular individual is founded. Indeed, preliminary experiments
suggest that snowflake yeast have a substantial advantage over yeast that form
chimeric clusters through aggregation.
One previously unexpected consequence of multicellular
what Eric Libby and I have termed the ‘ratcheting’ hypothesis. As we describe in our 2014
Perspective in Science, traits evolved by cells early in the transition to multicellularity that
are adaptive in clusters, but maladaptive to single cells, may entrench the lineage in a
multicellular way of life. Cutting off the unicellular escape route may facilitate the evolution
of increased multicellular complexity by maintaining strong selection on multicellular traits.
One example of ratcheting in snowflake yeast may be elevated rates of programmed cell
death (left), which can increase the fitness of cells in snowflake clusters, but should be
costly to single cells not living in clusters.
In our 2016 Phil. Trans. Biol. follow-up, we examine how ratcheting can stabilize the transition to multicellularity in more
detail. We show that ratcheting can evolve in two ways: cells can evolve traits in a multicellular context that are costly to
isolated cells (like PCD), or cells in groups can evolve traits that directly limit the rate at which they revert to unicellularity.
One of the neatest aspects of the paper is that we show that these two traits can interact synergistically- the presence of
either kind of ratcheting trait selects for the other. This has a pretty intuitive explanation: if cells in multicellular groups evolve
traits that are costly in a single-cell context (type 1 ratcheting), then evolving reversion-limiting traits (type 2 ratcheting) is
good. And in the opposite case, if reversion is rare, then it frees up the lineage to evolve traits in a multicellular context that
are very costly to isolated cells. The take-away message from this paper is that it's pretty easy, at least in theory world, for a
lineage that switches between uni and multicellular states to evolve traits that lock them into a multicellular lifestyle- an
important step for evolving complex multicellular phenotypes.
In collaboration with Matthew Herron, Frank Rosenzweig and Michael Travisano, we have been
pursuing similar experiments in the unicellular green alga Chlamydomonas reinhardii. This alga is
closely related to the multicellular volvocine algae, allowing us to compare experimentally-evolved
traits to those that occur in nature. In a pilot experiment (Ratcliff et al, 2014), we observe the
evolution of a novel multicellular life cycle. Cells form clusters by producing a viscous extracellular
matrix. After transfer to fresh medium, motile cells burst from the cluster, then settle down and form
new clusters. This demonstrates that a unicellular genetic bottleneck, a trait that strongly facilitates
multicellular-level adaptation, can arise early in this evolutionary transition and in the absence of
selection for among-cell conflict mediation. We are following up on this work using predators (i.e.,
Paramecium and rotifers) to select for increased size, a more ecologically-realistic selective agent
Now that we have a model system spanning the transition from single-celled to simple multicellular organisms, we’re uniquely
suited to research several aspects of this evolutionary process. For example, we’re studying how the transition to
multicellularity affects the evolution of aging, the evolutionary costs and benefits of sex, how cells evolve into parts of a novel
multicellular individual, and how multicellular development can evolve from scratch. We are also working to determine the
genetic basis of emergent multicellular traits, like apoptosis.
Snowflake with red dead cells
Snowflake yeast reproduction