Evolution of irreversible somatic differentiation

  1. Yuanxiao Gao
  2. Hye Jin Park
  3. Arne Traulsen
  4. Yuriy Pichugin

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    This paper will be of interest to researchers working on a broad range of questions in evolutionary biology, from the evolution of multicellularity to senescence and cancer. With their model, the authors study an often-neglected aspect of cellular differentiation and division of labour. While the model is relatively simple, the premise and the findings are thought-provoking and this study can potentially provide the groundwork for further investigation.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1, Reviewer #2 and Reviewer #3 agreed to share their names with the authors.)

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

A key innovation emerging in complex animals is irreversible somatic differentiation: daughters of a vegetative cell perform a vegetative function as well, thus, forming a somatic lineage that can no longer be directly involved in reproduction. Primitive species use a different strategy: vegetative and reproductive tasks are separated in time rather than in space. Starting from such a strategy, how is it possible to evolve life forms which use some of their cells exclusively for vegetative functions? Here, we develop an evolutionary model of development of a simple multicellular organism and find that three components are necessary for the evolution of irreversible somatic differentiation: (i) costly cell differentiation, (ii) vegetative cells that significantly improve the organism’s performance even if present in small numbers, and (iii) large enough organism size. Our findings demonstrate how an egalitarian development typical for loose cell colonies can evolve into germ-soma differentiation dominating metazoans.

Article activity feed

  1. Version published to 10.7554/elife.66711 on eLife
  2. eLife

    Author Response:

    Reviewer #1:

    In this paper, the authors study one of the understudied aspects of the evolutionary transition to multicellularity: the evolution of irreversible somatic differentiation of germ cells. Division of labour via functional specialisation of cells to perform different tasks is pervasive across the tree of life. Various studies assume that the differentiation of reproductive cells ("germ-role cells" in this manuscript) into a non-reproducing cell type ("soma-role cells") is irreversible. In reality, the conditions that promote the evolution of this irreversible transition are unclear. Here, the authors set out to fill in this knowledge gap. They model a population of organisms that grow from a single germ-role cell and find the optimal developmental strategy in terms of differentiation probabilities, under different scenarios. Under their model assumptions, they show that irreversible somatic differentiation can evolve when 1) cell differentiation is costly, 2) somatic cells' contribution to growth rate is large, 3) organismal body size is large.

    Overall, I think the authors identified an interesting and neglected aspect of cellular differentiation and division of labour. I enjoyed reading the paper; I thought the writing was clear and the modelling approach was adequate to address the authors' question.

    Thank you for a detailed and constructive review.

    Some aspects that can be improved:

    1. Throughout the manuscript, I was somewhat confused about what system the authors have in mind: a colony with division of labour or a multicellular organism? While their model can potentially capture both, their Introduction and Discussion seem to be geared towards colonies at the transition to multicellularity, whereas the Results section gives the impression that the authors have multicellular organisms in mind (e.g. very large body sizes).

    We are interested in the transition from a colonial life, where tasks are distributed in time, to multicellular organisms, where tasks are divided between cells. As such, our model covers these scenarios as two limit cases. In the context of our study, we discuss examples from the nature where this transition is observed – e.g. among Volvocales algae. For the purpose of the necessary colony/organism size, we do not need to go further than 2^6 = 64 cells. However, to infer the patterns of the composition effect Fcomp (Fig.3 C,D), we consider organisms doing four more rounds of cell divisions before reproduction, leading to maturity size of 2^10=1024 cells. There, irreversible somatic differentiation can occur at a wide range of differentiation costs (see Fig.4 A). Also, smaller sizes put stronger restrictions on the composition effect Fcomp, so the distribution of parameters presented at Fig.3C,D taken at the n=6 instead of 10, would have much less data points and this could obfuscate the pattern found in this study. Overall, the scale of about 1000 cells, for which we report most of our modeling results, features entities with very diverse complexity: from undifferentiated colonies (ocean algae Phaeocystis antarctica), to intermediary life forms (slime molds slugs), to paradigm multicellular organisms (higher Volvocales and C. elegans). We think that the chosen range of the organism size is adequate to the comparison of entities with undifferentiated and differentiated cells. In the updated manuscript, we extend the exposition of organism size to reflect this aspect.

    1. From the point of view of someone who works on topics related to cancer and senescence, I think these fields are very much connected to the evolution of multicellularity. Maybe because I had multicellular organisms in mind rather than colonies with division of labour (above), I thought the manuscript missed this connection. Damage accumulation is key to Weismann and Kirkwood's theories of germ-soma divide and disposable soma, respectively, whereas dysregulated differentiation is one of the important aspects of tumour development (e.g. Aktipis et al. 2015). Making these links could also be relevant to discuss some of the model assumptions. For instance, the authors assume that fast growth comes with no cost in terms of cell damage, which may not always be the case (e.g. Ricklefs 2006) and reversibility of somatic differentiation can come at a cost of increased risk of somatic "cheaters" or cancerous cell lines.

    Thank you for this suggestion. Indeed, the aspect of cancer risk has not been considered in the initially submitted manuscript. In the updated manuscript, we introduce a model where differentiation is linked to the risk of an organism for death instead of a delay in development. The results with this model exhibit very similar pattern, see Fig.5. Hence, the term “cost of differentiation” can be interpreted more broadly than just cell division delay suggested by our main model.

    1. The authors assume the differentiation strategy (D) does not change over the lifetime (which equates to ontogenesis in their model, i.e. they do not consider mature lifespan). I wonder if this is really the case, or whether organisms/cells can respond to the composition of cells they perceive. For instance, at least in some animal tissues, a small number of stem cells are kept to replenish differentiated tissue cells when needed. I understand that making D plastic can make the model really complicated, but maybe it is worth talking about what strategy would evolve if D was not stable through ontogenesis (and mature lifespan). My initial guess is that if differentiation probabilities can change through life and if one considers cellular damage accumulation, senescence and cancer (as above), the conditions that favour irreversible somatic differentiation would expand.

    Indeed, we assume the differentiation strategy to be constant in our model. We do not know whether it is true at the brink of multicellularity and, for sure, once evolution makes a species complex enough, this assumption will become inadequate. Yet, when we consider a dynamic differentiation strategy, we find a very efficient but unrealistic solution: at the very beginning of a life cycle a germ-role cell gives rise to two soma-role cells, then these soma-role cells produce only soma-role cells and finally, at the very last round of cell division, they give rise to as many germ cells as possible. This scenario is the most efficient in terms of the rate of the organism development (100% of useful soma-role cells during growth), amount of offspring produced (every cell becomes a germ at the end of the day), and differentiation costs/risks (differentiation occurs only twice in a life time). Still, it is unrealistic. There must be some constraints on the flexibility of the dynamic differentiation strategy. We think that the exploration of the space of dynamical differentiation strategies and their constraints goes beyond the scope of the current study. Nevertheless, we are very interested to explore this topic further in following projects.

    Reviewer #2:

    This works seeks to determine the conditions in which simple multicellular groups can evolve irreversibly somatic cells, that is: a replicating cell lineage that provides cooperative benefits as the group grows and cannot de-differentiate into reproductive germ cells.

    This question is addressed with a well-constructed model that is easy to understand and provides intuitive results. Groups are composed of germ and soma cells that replicate synchronously until the group has reached a maximal size. When each type of cell divides, they may have different probabilities of producing daughter cells of each type, and the analysis determines the optimal differentiation probabilities for each type of cell depending on a variety of factors. Critically, irreversible somatic differentiation arises when the optimal probability for soma cells is to produce only soma cells.

    The elegance of the model means that the predictions are easy to interpret. First, when there is a higher cost for soma cells to produce germ cells, then a dedicated lineage of somatic cells is more favourable. Second, when soma cells produce only soma cells and germ cells can produce both types, the proportion of soma cells in the group will increase with each division. Consequently, for irreversible somatic cells to be optimal, germ cells must produce a small number of soma cells and these few must provide large benefits. Third, larger group sizes are required for a small number of soma cells to arise and provide sufficient benefits to the group.

    Inevitably, there is a trade-off between the benefits of a simple model and the costs of idealised assumptions.

    Among other assumptions, the model assumes that germ cells and soma cells replicate synchronously and at the same rate, and that soma cells provide benefits throughout the growth of the group, but do not increase the fecundity of germ cells in the last generation. Consequently, it is not clear to what extent the predictions of the model apply to the notable empirical cases where these assumptions do not hold. For instance, in the often-cited Volvocine algae, soma cells do not provide any benefits until the last generation of the group life cycle. This may help to explain why many Volcocine species have a very large number of somatic cells, counter to the second prediction of the model.

    Overall, this analysis is targeted and provides clear predictions within the bounds of its assumptions. Thus, these results provide a compelling framework or stepping-stone against which future models of germ-soma differentiation in alternate scenarios can be compared and evaluated.

    Thank you for the kind words and the well-thought review. Indeed, our model takes a number of simplifying assumptions. In the revised manuscript, we consider the model, in which the strongest of our simplifications – of simultaneous cell divisions - is violated. This asynchronous cell division model shows that irreversible differentiation may evolve, at least, under asymmetric differentiation costs. However, its evolution is observed less often than in a synchronous model.

    We absolutely agree that the design of our model does not replicate the details of Volvocine life cycles. However, our work is not aimed to be a model of germ-soma differentiation in Volvocales. Instead, we developed a simplistic model implementing features from a diverse range of organisms. While in higher Volvocales young colonies develop within a maternal organism, there is a wide range of colonial organisms, which grow from independently living single cell, e.g. colonial diatoms, Haptophytes Phaeocystis antarctica, and amoebazoan Phalansterium. We agree that the protection by maternal organism should play a major role in Volvocales and we are looking forward to investigate a follow-up model taking this factor into account.

    Reviewer #3:

    This paper provides a theoretical investigation of the evolution of somatic differentiation. While many studies have considered this broad topic, far fewer have specifically modelled the evolutionary dynamics of the reversibility of somatic differentiation. Within this subset, the conditions that select for irreversible somatic differentiation have appeared conspicuously restrictive. This paper suggests that an overly simplified fitness function (mapping the soma-germline composition of an organism to its growth rate) may be partly to blame. By allowing for a more complex fitness function (that captures the effect of upper and lower bounds for the contribution of somatic cells to organism fitness) the authors are able to identify three conditions for the evolution of irreversible somatic differentiation: costly cell differentiation (particularly for the redifferentiaton of soma-cell lineages to germ line); a high/near maximal organismal growth advantage imbued by a small proportion of soma cells; a large maturity size for the organism (typically greater than 64 cells).

    The model presented is simple and elegant, and succeeds in its aim of providing biologically feasible conditions for the evolution of irreversible somatic differentiation. Although the observation arising from the first condition (that high costs to reversible somatic differentiation promote the evolution of irreversible somatic differentiation) is perhaps unsurprising, the remaining conditions on the fitness function and the organism maturity size are interesting and initially non-obvious. Particularly tantalising is the prospect of testing these conditions, either against available empirical data, or in an experimental setting.

    The model does however make a number of simplifying assumptions, the effects of which may limit the broad applicability of the results.

    The first is to assume that cell division is synchronous, so that the costs of cell differentiation can be straight-forwardly averaged across the organism at each division. While the authors present a convincing biological justification for this assumption for algae such as Eudorina illinoiensis and Pleodorina californica, it is not immediately that this assumption should hold more widely.

    The second is to assume that the development strategy (i.e. the rates of differentiation between somatic and germ-line cell types) is constant throughout the organism's growth. For instance, there may be a growth advantage in the current model (aside from the advantages with respect to reduced mutation accumulation) of producing more germ cells early in the developmental programme, before transitioning to producing more soma cells in later development.

    Exploring such extensions to this model presents a seam of potential avenues for investigation in future theoretical studies.

    Thank you for the kind assessment of our findings. In the updated manuscript, we in addition investigated a model with asynchronous cell divisions. However, due to computational limitations, we are unable to fully replicate the investigation protocol of the original synchronous model. The execution time of the synchronous model scales linearly with the number of generations (n) and it still takes about a week to compute a single map like Fig.2A on a 2000-node cluster. The asynchronous model, in turn scales linearly with number of cell divisions, and hence, exponentially with generation time (as 2^n), which results in calculations taking much more time. For instance, the map in Fig.2A requires about 160 times more computer time with the asynchronous model. Nevertheless, we were able to implement this model for smaller organisms, with less statistics. There, we found that asynchronous model allows an evolution of irreversible somatic differentiation. However, it is suppressed comparing with the synchronous model – the fraction of Fcomp profiles promoting irreversible differentiation is much smaller and the organism size restriction is higher.

    To study a dynamic differentiation strategy would be wonderful. Early on, we considered studying this scenario. The crucial factor here is how flexible can the strategy be. In a naïve situation with a complete flexibility between every cell generation, the most successful strategy would be all cells of an organism first completely turn into soma-role to gain the maximal benefits, and then at the last step, they all convert back to germ to produce the maximal number of offspring. This is not observed in natural species; hence the flexibility of dynamic differentiation program must be constrained. We are curious to study what kind of constraints can lead to irreversible soma, but this task is beyond the scope of the current study. Our work with a constant differentiation program is the beginning of the future line of research. We are already looking forward to explore the space of dynamic differentiation programs in later projects.

  3. eLife

    Evaluation Summary:

    This paper will be of interest to researchers working on a broad range of questions in evolutionary biology, from the evolution of multicellularity to senescence and cancer. With their model, the authors study an often-neglected aspect of cellular differentiation and division of labour. While the model is relatively simple, the premise and the findings are thought-provoking and this study can potentially provide the groundwork for further investigation.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1, Reviewer #2 and Reviewer #3 agreed to share their names with the authors.)

  4. eLife

    Reviewer #1 (Public Review):

    In this paper, the authors study one of the understudied aspects of the evolutionary transition to multicellularity: the evolution of irreversible somatic differentiation of germ cells. Division of labour via functional specialisation of cells to perform different tasks is pervasive across the tree of life. Various studies assume that the differentiation of reproductive cells ("germ-role cells" in this manuscript) into a non-reproducing cell type ("soma-role cells") is irreversible. In reality, the conditions that promote the evolution of this irreversible transition are unclear. Here, the authors set out to fill in this knowledge gap. They model a population of organisms that grow from a single germ-role cell and find the optimal developmental strategy in terms of differentiation probabilities, under different scenarios. Under their model assumptions, they show that irreversible somatic differentiation can evolve when 1) cell differentiation is costly, 2) somatic cells' contribution to growth rate is large, 3) organismal body size is large.

    Overall, I think the authors identified an interesting and neglected aspect of cellular differentiation and division of labour. I enjoyed reading the paper; I thought the writing was clear and the modelling approach was adequate to address the authors' question.

    Some aspects that can be improved:

    1. Throughout the manuscript, I was somewhat confused about what system the authors have in mind: a colony with division of labour or a multicellular organism? While their model can potentially capture both, their Introduction and Discussion seem to be geared towards colonies at the transition to multicellularity, whereas the Results section gives the impression that the authors have multicellular organisms in mind (e.g. very large body sizes).

    2. From the point of view of someone who works on topics related to cancer and senescence, I think these fields are very much connected to the evolution of multicellularity. Maybe because I had multicellular organisms in mind rather than colonies with division of labour (above), I thought the manuscript missed this connection. Damage accumulation is key to Weismann and Kirkwood's theories of germ-soma divide and disposable soma, respectively, whereas dysregulated differentiation is one of the important aspects of tumour development (e.g. Aktipis et al. 2015). Making these links could also be relevant to discuss some of the model assumptions. For instance, the authors assume that fast growth comes with no cost in terms of cell damage, which may not always be the case (e.g. Ricklefs 2006) and reversibility of somatic differentiation can come at a cost of increased risk of somatic "cheaters" or cancerous cell lines.

    3. The authors assume the differentiation strategy (D) does not change over the lifetime (which equates to ontogenesis in their model, i.e. they do not consider mature lifespan). I wonder if this is really the case, or whether organisms/cells can respond to the composition of cells they perceive. For instance, at least in some animal tissues, a small number of stem cells are kept to replenish differentiated tissue cells when needed. I understand that making D plastic can make the model really complicated, but maybe it is worth talking about what strategy would evolve if D was not stable through ontogenesis (and mature lifespan). My initial guess is that if differentiation probabilities can change through life and if one considers cellular damage accumulation, senescence and cancer (as above), the conditions that favour irreversible somatic differentiation would expand.

  5. eLife

    Reviewer #2 (Public Review):

    This works seeks to determine the conditions in which simple multicellular groups can evolve irreversibly somatic cells, that is: a replicating cell lineage that provides cooperative benefits as the group grows and cannot de-differentiate into reproductive germ cells.

    This question is addressed with a well-constructed model that is easy to understand and provides intuitive results. Groups are composed of germ and soma cells that replicate synchronously until the group has reached a maximal size. When each type of cell divides, they may have different probabilities of producing daughter cells of each type, and the analysis determines the optimal differentiation probabilities for each type of cell depending on a variety of factors. Critically, irreversible somatic differentiation arises when the optimal probability for soma cells is to produce only soma cells.

    The elegance of the model means that the predictions are easy to interpret. First, when there is a higher cost for soma cells to produce germ cells, then a dedicated lineage of somatic cells is more favourable. Second, when soma cells produce only soma cells and germ cells can produce both types, the proportion of soma cells in the group will increase with each division. Consequently, for irreversible somatic cells to be optimal, germ cells must produce a small number of soma cells and these few must provide large benefits. Third, larger group sizes are required for a small number of soma cells to arise and provide sufficient benefits to the group.

    Inevitably, there is a trade-off between the benefits of a simple model and the costs of idealised assumptions.

    Among other assumptions, the model assumes that germ cells and soma cells replicate synchronously and at the same rate, and that soma cells provide benefits throughout the growth of the group, but do not increase the fecundity of germ cells in the last generation. Consequently, it is not clear to what extent the predictions of the model apply to the notable empirical cases where these assumptions do not hold. For instance, in the often-cited Volvocine algae, soma cells do not provide any benefits until the last generation of the group life cycle. This may help to explain why many Volcocine species have a very large number of somatic cells, counter to the second prediction of the model.

    Overall, this analysis is targeted and provides clear predictions within the bounds of its assumptions. Thus, these results provide a compelling framework or stepping-stone against which future models of germ-soma differentiation in alternate scenarios can be compared and evaluated.

  6. eLife

    Reviewer #3 (Public Review):

    This paper provides a theoretical investigation of the evolution of somatic differentiation. While many studies have considered this broad topic, far fewer have specifically modelled the evolutionary dynamics of the reversibility of somatic differentiation. Within this subset, the conditions that select for irreversible somatic differentiation have appeared conspicuously restrictive. This paper suggests that an overly simplified fitness function (mapping the soma-germline composition of an organism to its growth rate) may be partly to blame. By allowing for a more complex fitness function (that captures the effect of upper and lower bounds for the contribution of somatic cells to organism fitness) the authors are able to identify three conditions for the evolution of irreversible somatic differentiation: costly cell differentiation (particularly for the redifferentiaton of soma-cell lineages to germ line); a high/near maximal organismal growth advantage imbued by a small proportion of soma cells; a large maturity size for the organism (typically greater than 64 cells).

    The model presented is simple and elegant, and succeeds in its aim of providing biologically feasible conditions for the evolution of irreversible somatic differentiation. Although the observation arising from the first condition (that high costs to reversible somatic differentiation promote the evolution of irreversible somatic differentiation) is perhaps unsurprising, the remaining conditions on the fitness function and the organism maturity size are interesting and initially non-obvious. Particularly tantalising is the prospect of testing these conditions, either against available empirical data, or in an experimental setting.

    The model does however make a number of simplifying assumptions, the effects of which may limit the broad applicability of the results.

    The first is to assume that cell division is synchronous, so that the costs of cell differentiation can be straight-forwardly averaged across the organism at each division. While the authors present a convincing biological justification for this assumption for algae such as Eudorina illinoiensis and Pleodorina californica, it is not immediately that this assumption should hold more widely.

    The second is to assume that the development strategy (i.e. the rates of differentiation between somatic and germ-line cell types) is constant throughout the organism's growth. For instance, there may be a growth advantage in the current model (aside from the advantages with respect to reduced mutation accumulation) of producing more germ cells early in the developmental programme, before transitioning to producing more soma cells in later development.

    Exploring such extensions to this model presents a seam of potential avenues for investigation in future theoretical studies.

  7. Version published to 10.1101/2021.01.18.427219v1 on bioRxiv