It all starts from a cooperative community

From the Shou lab, Basic Sciences Division

Systems consist of parts, even in biology; for instance, ecological communities are made up of interacting species, and the human body is comprised of different cell types. The ability to predict quantitative traits and behaviors of a biological system can be immensely powerful; it enables researchers to manipulate these systems to harness their activities and predict outcomes of perturbations to individual parts.

Basic Sciences Associate member Dr. Wenying Shou has been interested in understanding how interacting “parts” can generate quantitative properties of the “whole” through mathematical modeling since she was a postdoc about 17 years ago. She tells the story of her scientific journey in a blog post titled "The devil in the closet". At that time, biological phenomena such as gene regulation networks and the cell division cycle had been modeled, with some models matching experimental data perfectly. However, Dr. Shou pointed out that the renowned mathematician and computer scientist von Neumann once stated “with four parameters I can fit an elephant, and with five I can make him wiggle his trunk.” In other words, given enough “free parameters” where one can freely choose parameters rather than only using measured parameters that are readily available, a model can be made to fit any data. Although a model that fits can explain data from a particular state of the system, it does not mean that the model is correct or can predict new data.

Therefore, to eschew the free parameter problem, Dr. Shou set out to model a much simpler system. “In this system, I should know exactly how parts interact with each other, and thus, which parameters to measure.  After much deliberation, I decided to engineer a highly simplified yeast cooperative community where each of the two strains supplied the other strain with an essential metabolite.  Because of the co-dependence, the two strains coexist at a fixed ratio, and so I do not need to worry about losing one of the strains.” explained Dr. Shou.

However, despite distilling the model to one that can be described with four parameters and doing systematic studies, the model prediction did not agree with experimental results. Although Dr. Shou moved on with different projects, she still had a burning desire to solve the problem. Contributions by several different lab members over the years led to an answer. Published in a recent issue of the journal PLOS Biology, Dr. Shou and members of her lab describe how they resolved the many challenges that came with quantitative modeling in a simplified community of interacting cells.


Comparison of model predictions and experiments. Cooperation that is Synthetic and Mutually Obligatory (CoSMO) consists of two differentially fluorescent, non-mating haploid S. cerevisiae strains used in this study. Figure from Dr. Shou

To enable phenotype measurements within a community-like environment, lab member David Skelding built devices called “chemostats”, where nutrients were supplied at a small dose (in small drops) but frequently (every tens of seconds), to mimic the partner strain’s slow but constant metabolite release rate. Sam Hart, then a research technician in the lab, used the chemostats to measure strain metabolite release and consumption phenotypes in community-like environments. The growth rate could be controlled by regulating how fast nutrients were supplied. However, the authors found that a nutrient-limited environment allowed cells to evolve rapidly from their original state. Indeed, metabolite release and consumption phenotypes were sensitive to nutrient availability, resulting in strikingly different chemostat and batch culture measurements. By taking these environment-sensitive parameters into account, the authors were able to accurately predict steady state growth rate in each yeast strain, as well as which cells evolved to be more generous.

The authors demonstrate that despite the many obstacles, quantitative modeling of cell communities is achievable. Their approach is generalizable for modeling communities of interacting cells where genetic or chemical information of interaction mechanisms is available. This ability to successfully predict quantitative traits of a biological system has the potential to greatly accelerate complex biological research; sometimes, going back to a simple cooperative community is what is needed.

Hart SFM, Mi H, Green R, Xie L, Pineda JMB, Momeni B, Shou W. 2019. Uncovering and resolving challenges of quantitative modeling in a simplified community of interacting cells. PLoS Biol, 17(2), e3000135.

This work was supported by the National Institutes of Health, Fred Hutchinson Cancer Research Center, the W.M Keck Foundation, the National Science Foundation and the Richard and Susan Smith Family Foundation.