Evolution of multicellularity by collective integration of spatial information

  1. Enrico Sandro Colizzi
  2. Renske MA Vroomans
  3. Roeland MH Merks

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

At the origin of multicellularity, cells may have evolved aggregation in response to predation, for functional specialisation or to allow large-scale integration of environmental cues. These group-level properties emerged from the interactions between cells in a group, and determined the selection pressures experienced by these cells. We investigate the evolution of multicellularity with an evolutionary model where cells search for resources by chemotaxis in a shallow, noisy gradient. Cells can evolve their adhesion to others in a periodically changing environment, where a cell’s fitness solely depends on its distance from the gradient source. We show that multicellular aggregates evolve because they perform chemotaxis more efficiently than single cells. Only when the environment changes too frequently, a unicellular state evolves which relies on cell dispersal. Both strategies prevent the invasion of the other through interference competition, creating evolutionary bi-stability. Therefore, collective behaviour can be an emergent selective driver for undifferentiated multicellularity.

Article activity feed

  1. Version published to 10.7554/elife.56349 on eLife
  2. eLife

    ###This manuscript is in revision at eLife

    The decision letter after peer review, sent to the authors on May 18, 2020, follows.

    Summary

    In this paper, the authors consider the problem of evolutionary transitions to multicellularity, and in particular the case in which aggregation drives the process. Inspired by the life cycle of Dictyostelium, they consider a model in which cells (moving on a grid) search for resources and can adhere to each other based on the match between ligand and receptors on their surfaces. All of this takes place in the context of a chemotactic march towards a local chemoattractant within one temporal "season", after which fitness-dependent reproduction occurs, the population is culled back to its starting size, and the environmental conditions are changed.

    The reviewers all are of the opinion that this work provides an interesting perspective on a possible mechanistic basis of 'collective-level' function, that stems from physical interactions among cells in the absence of explicitly modelled costs and benefits of single cell's choices. At the same time, the reviewers were clear that there are many aspects of the model and the modelling approach that are not clear, unnecessarily complicated or not well justified. In light of these, major revisions to the paper will be necessary, as explained below.

    Essential Revisions

    1. Considering the paper as a whole, there are far too many things happening at once to draw any meaningful conclusions. There is the complexity of adhesion, the nature of the chemotaxis, the temporal switching between seasons, and the reproduction process. Each of these is explored to a limited extent, and it is unclear which are absolutely crucial to the conclusions reached and how sensitive the conclusions are to the assumptions made.

    2. Regarding the definition of the model itself, the reviewers find it inappropriate to relegate so much of that explanation to the Methods section. The very large number of parameters (18) in Table 1 needs to be made clear (and that table should be referenced - it does not appear to be at present). Please explain more of the model in the body of the paper.

    3. The reviewers are supportive of abstract models, but inasmuch as the authors have set up a physical/biological scenario with familiar processes (chemotaxis, adhesion) it would have been very helpful to have justified the kinds of dimensionless parameters that characterize the model in terms of real physical and biological features.

    4. The essence of a Monte Carlo simulation is the definition of an energy function and a temperature, which together yield a Boltzmann factor that is used to decide if an attempted step is taken. The authors do not make clear in the main body of the text that they are performing a Monte Carlo calculation (that is only specified in Section 4, after the Discussion). They refer to MCS (Monte Carlo Steps) in the body of the paper without defining that term. But the larger question is why this kind of nonequilibrium biological system should have such an energy, and what would be the biological significance of the temperature? In addition, of course, the "steps" taken are those of Monte Carlo algorithm and have no direct interpretation in terms of real time.

    5. The presentation of the model and the main results lack clarity in some key aspects: a. the relation between cell-cell and cell-medium adhesion and surface tension (line 136) is not explained, so it is not really clear what negative surface tension means. b. as surface tension pools two different kinds of adhesion, does it mean that in a certain sense adhesion to the surface can be traded off against adhesion between cells? This is important to know in connection to experiments. c. since the measure of sequence complementarity is symmetric, why does one need to suppose the existence of both a ligand and a receptor? Would it change anything if cells were characterized by only one sequence? If yes, it would be interesting to know if at the end of the numerical experiment ligand and receptor evolve to be the same or if 'molecular' diversity is maintained. d. the process of cell division/regrowth and the fact that cells do not change position from one season to the next should be more clearly explained in the main text. e. what is the initial spatial distribution of cells at the beginning of every season, and if this matters (many models assume aggregation-dispersal cycles, that does not seem to be the case here), should be specified or repeated in the evolutionary section. f. Figure 5 should depict a case of bistability: now it is not clear that different evolutionary outcomes are associated to differences in the initial surface tension, rather than in the initial cell configuration. It would by the way be interesting to see if the second also gives rise to bistability.

    6. Cell migration (lines 394-404) is defined in terms of the actual direction of the cell over the past steps. This seems to build in persistence, and would appear to have a profound effect on the dynamics. Is this the case?

    7. In general, it would be useful if statements like "In our case, aggregation leads to a highly efficient search strategy, guided by long-range, albeit noisy, gradients." (lines 272-273) could be made more quantitative. For instance, one would like to get a sense of whether the conclusions are robust to changes in (at least a few important) parameters. One would expect so from results in active matter physics, but it would be useful of the authors could argument it and indicate when they expect different conclusions to hold. Moreover, what is the role of the particular gradient chosen here in 'focalizing' the formation of multicellular groups (would an essentially 1-D gradient, where isolines are parallel, do the job?) and of its intensity/spatial variation (in the movie, one sees that the center of the gradient changes among four positions, does it matter?).

    8. The authors claim that, in contrast to previous work, the increased fitness of the aggregates (better ability to perform chemotaxis) is an emergent property. The reviewers struggled to find a physical/mathematical explanation as to why such a relationship exists in the model but it appears that lines 424-427 contain the mechanism. The text speaks of the "center of mass of the perceived gradient". Unless we are mistaken, such a quantity averages over the individual constituent's contributions in such a way that larger cells will have more accurate measurements of the gradient. This is just the law of large numbers. If this is the case, then this feature is not an emergent property at all, but is part of the definition of the model. Please clarify. If the above critique is correct, then why bother with the complex model? The authors could just use the fact that larger aggregates are better at chemotaxis for the reason given and proceed from there.

    The above suggests that the authors have basically put the answer in from the beginning. The model has the explicit feature that those that peform chemotaxis better reproduce more. So of course that will be reinforced. But multicellularity has costs and benefits, and the model does not appear to contain any costs associated with multicellularity. In real biological examples there are many - the increased metabolic cost of the structures that hold cells together, greater need for regulatory genetic networks, etc.

    1. The referencing of the text to Figure 3 is all mixed up, leaving both text and figure hard to follow. -The authors should revise this section and make sure that they clearly state if higher chemotactic performance arises due to longer persistence of cell clusters only or due to longer persistence and higher chemotactic accuracy of whole cell cluster. Varennes et al PRL (2017) 119:188101 and manuscripts citing this work give measures for chemotactic accuracy within cell populations. - Fig 3d should show error bars. Annotation of Fig 3 f should be detailed, what is bar{X}? Is this the local gradient including noise or averaged on which scale.

    2. The assessment time scale emerges as a decisive factor - it appears as a theoretical construct right now. What could it correspond to in the real world? Please discuss.

    3. As for the particular details of the model, it is left unsaid in the main text but stated in the Methods section that there is a preferred cell size A_T and a harmonic energy around that size. As the target size is (Table 1) some 50 pixels, we are confused, as it seems that each "cell" occupies one lattice size. This energy would then clearly bias the system to aggregate already. Please clarify. The use of the term "pixel" for a lattice site is confusing.

    4. The literature overview appears limited - please revise and consider recent work for example but not limited to Varennes et al PRL (2017) 119:188101; Jacobeen et al (2018) Phys. Rev. E 97, 050401(R). The authors should also discuss Guttal & Couzin 'Social interactions, information use, and the evolution of collective migration' PNAS 2010. And they should acknowledge relevant literature exploring, for example, similar issues in the Volvocales; "Multicellularity and the Functional Interdependence of Motility and Molecular Transport", C.A. Solari, S. Ganguly, J.O. Kessler, R.E. Michod, and R.E. Goldstein, PNAS 103, 1353-1358 (2006); "A General Allometric and Life-History Model for Cellular Differentiation in the Transition to Multicellularity", C.A. Solari, J.O. Kessler and R.E. Goldstein American Naturalist 181, 369-380 (2013).

  3. Version published to 10.1101/2020.02.20.957647v1 on bioRxiv