Session H49: Evolutionary and Ecological Dynamics - II

Stochastic population dynamics induced by antibiotic treatment

Frequent antibiotic failure is a serious threat to public health. To cope with this threat, it is critical that we better understand the population dynamics of bacteria exposed to antibiotics. Previous studies have extensively characterized the homogenous dynamics of large bacterial populations exposed to antibiotics, establishing the deterministic framework of pharmacodynamics. However, the outcome of antibiotic treatment is typically far from being deterministic. Here, characterizing small bacterial populations, we demonstrate the stochastic nature of bacterial clearance using antibiotics. We found that bactericidal drugs induce population fluctuations, leading to stochastic population dynamics. Consequently, bacterial clearance does not follow a deterministic course but is highly probabilistic. The probability of the clearance was well captured by the birth-death Markov model. The model also predicted an increase in the probability of clearance with a decrease in growth rate. We experimentally tested this prediction. Our study reveals the stochastic population dynamics induced by antibiotics and how this stochasticity may be manipulated to facilitate bacterial clearance.   

High dimensional microbiome interactions shape the host fitness landscape

Koch’s postulates established the field of infectious disease on the concept of one pathogen, one disease. Here we show that interactions between gut bacteria are as important as the bacteria themselves in determining fly physiology. We performed a complete combinatorial dissection of the fly gut microbiome using germ free flies colonized with each possible combination of the 5 core species of bacteria, forming a 5 dimensional cube. We then measured the microbial community abundances and fly fitness traits, including (i) development, (ii) reproduction, and (iii) aging. Notably, microbial diversity accelerates development, reproduction, and aging. From these measurements we calculated the impact of microbial interactions on fly fitness as the triangulations of the 5-cube, following the combinatorial geometry approach of Beerenwinkel-Pachter-Sturmfels. Single species cannot predict pairs, contradicting Koch’s postulates, and higher order interactions are widely prevalent. Important for evolution, we find a positive feedback between the microbial community and host fitness that poises a population for divergence of hosts and the emergence of host-specific microbiomes.   

Stabilizing Target Microbial Community-Level Behaviors via External Modulation of Quorum-Sensing Communication Networks

Efforts are underway to develop quantitative approaches for predictively controlling the activity level of complex microbial communities. One strategy involves targeting microbial chemical communication networks, which often serve as hubs for regulating global gene expression. A well-known example of such a network is quorum sensing (QS), whereby cells use small molecules called autoinducers (AI) to regulate density-dependent gene expression. Recent work (Yusufaly and Boedicker, Physical Biology 14, 046002 (2017)) has shown that multispecies QS circuits, with multiple chemically distinct AI molecules, can be mapped onto a Hopfield neural network. Here, we extend this result, and show that, by artificially pumping excess AI molecules into a community, we can define a perceptron, where the external AI current serves as a tunable input that drives the community to a certain gene expression output. This perceptron can be interpreted as a control system, allowing the methods of control theory to be applied to rationally design input current strategies that drive a community to ‘target’ output gene expression levels. This formalism is observed to be especially useful for analyzing the tradeoffs between performance and robustness for different strategies.   

Disentangling bacterial invasiveness from lethality in an experimental host-pathogen system

Understanding virulence remains a central problem in human health, pest control, disease ecology and evolutionary biology. Bacterial virulence is typically quantified by phenomenological indicators such as the LT50 (i.e. the time taken to kill 50% of an infected population). However, virulence emerges as a result of complex processes that occur at different stages: the pathogen needs to breach the primary host defenses, find a suitable environment to replicate, and finally express the virulence factors that cause lethality. It is well-known that pathogens exhibit a very broad spectrum of strategies to accomplish these three tasks, yet, phenomenological indicators such as the LT50 cannot distinguish the ability of the pathogen to invade the host from its ability to kill the host. Here, we propose a physical host-pathogen theory that shows how to disentangle colonization, growth, and pathogen lethality from the survival kinetics of a host population. Preliminary experimental data from C. elegans nematodes exposed to various pathogens shows that host mortality becomes severe only once the pathogen population has reached its carrying capacity within the host. In the talk, I will discuss various model predictions and compare them against experimental data.   

Mathematical model of influenza A evolution with HA protein stability

Influenza A H3N2 displays a tremendous capacity to adapt and evolve in response to pressure from host immune systems. The mutations that aid in immune system evasion are thought to occur largely in the viral antigen hemagglutinin (HA). Nucleoprotein (NP) is another protein of influenza A that is under significantly less selection pressure than HA. It has been shown that substitutions in NP are constrained by NP stability. That is, certain substitutions that decrease stability, but otherwise increase fitness, only occur in tandem with others that increase NP stability. In this project, we formulate a phylodynamics model of viral evolution for influenza A that incorporates HA protein stability. Our model tracks host individuals that are, respectively, infected, have strain-transcending immunity, or have lifelong antibodies against each specific strain. Our formulation also includes salient biological features such as cross-immunity between closely related strains. We show that important novel features emerge due to the inclusion of HA protein stability in our model. We characterize these properties through multiple approaches, illustrating the significance of protein stability in influenza A models.   

Emergent evolutionary dynamics in dense cellular populations

Evolutionary dynamics are controlled by a number of driving forces, such as natural selection, random genetic drift and dispersal. While these forces are usually modeled at the population level, it is often unclear how they emerge from the stochastic and deterministic behavior of individual cells. I discuss how even the most basic mechanical interactions between neighboring cells can couple evolutionary outcomes of otherwise unrelated individuals, thereby weakening natural selection and enhancing random genetic drift. Using microbial examples of varying degrees of complexity, I highlight a feedback loop between ecological and evolutionary dynamics, which is particularly pronounced in pattern-forming systems. Understanding this feedback loop could be key to predicting and potentially steering evolutionary processes, and requires extending the systems biology approach from the cellular to the population scale.   

Cheaters Impart Robustness to Cooperative Yeast Populations in Changing Environments

Natural as well as synthetic microbial communities are continually subjected to changes in environmental factors such antibiotics, temperature, pH, and nutrient availability, which directly impact their composition and function. While designing robust communities that maintain their composition and functionality in fluctuating environments is highly desirable, little is known about the nature of ecological interactions that render populations robust or sensitive to environmental changes. Here, using a phenomenological model of population dynamics we show that the population size of public good producing microbes (cooperators) can be rendered insensitive to changes in mortality rate via interactions with non-producing cheaters. Furthermore, we verify this prediction experimentally by demonstrating that the population size of glucose producing yeast cells is robust with respect to changes in the dilution rate in the presence of glucose non-producing cheater cells. Remarkably, our experiments reveal that in addition to maintaining a constant co-operator population size, the presence of cheaters also suppresses the variation in glucose concentration with dilution rate.   

Quorum sensing control of Vibrio cholerae aggregation

Bacteria coordinate their gene expression through the cell-cell communication process called quorum sensing. Quorum sensing (QS) involves the production, release, and group-wide detection of extracellular signal molecules called autoinducers, which allow cells to coordinate gene expression based on cell density. In Vibrio cholerae, QS controls the formation of biofilms: dynamic surface-adhered communities in which the cells are bound to each other and to the surface by the structural matrix extracellular polymeric substances (EPS). In V. cholerae the canonical biofilm program is upregulated in a low-cell-density (LCD) QS-state and the transition to a high-cell-density QS-state has been presumed to play a role largely in dispersal from biofilms. We find that V. cholerae can, in a HCD QS-state, robustly form aggregates in liquid suspension and known EPS constituents are dispensable for their formation. Unlike in traditional V. cholerae biofilms, cell division is not required for the formation of suspended biofilm communities. Instead, cells rapidly aggregate together during stationary phase. These data reveal new mechanisms by which bacteria control the formation of complex multicellular communities.   

Pairwise and Multi-Species Interactions Among Larval Zebrafish Gut Microbiota

The microbial communities resident in animal intestines play important roles in health and disease, and are composed of dozens to hundreds of interacting species. The determinants of this composition, which may include physical characteristics of bacterial groups as well as biochemical interactions between species, remain largely unknown. More generally, it is unclear for many multi-species consortia whether their species-level makeup can be predicted based on an understanding of pairwise species interactions, or whether higher-order interactions are necessary to explain community assembly.

To address this, we consider commensal intestinal microbes in larval zebrafish, initially raised germ-free to allow introduction of controlled combinations of bacterial species. Using a combination of conventional dissection and plating assays and three-dimensional live imaging, we demonstrate the existence of stable multi-species communities and test whether outcomes from two-species competitions contain enough information to allow prediction of multi-species abundances and interactions. We also compare in vitro competitions of the same commensal species, in which the spatial, temporal and nutritional heterogeneity of the intestine is absent.   

Using pair-wise competitive outcomes to understand microbial communities

Explaining the origin of species diversity is a major challenge in ecology. Temporal fluctuations and spatial structure in the environment likely play a key role, but it has also been suggested that the structure of interactions within the community may act as a stabilizing force for species diversity. In particular, if competitive interactions are non-transitive as in the classic rock-paper-scissors game, they can contribute to the maintenance of species diversity. Here, we investigate the network of pairwise competitive interactions in a model community consisting of 20 strains of naturally co-occurring soil bacteria. We find that the interaction network is strongly hierarchical and lacks significant non-transitive motifs, a result that is robust across multiple environments. Moreover, in agreement with recently proposed community assembly rules, the full 20-strain competition resulted in extinction of all but three of the most highly competitive strains, indicating that higher order interactions do not play a major role in structuring this community. The lack of non-transitivity and higher order interactions in vitro indicates that other factors, such as temporal or spatial heterogeneity, must be at play in enabling these strains to coexist in nature.   

Microbial Community Assembly Rules Emergent from the Stable Marriage Problem

A central debate in microbial ecology asks whether resource habitat or species interactions are more important in determining community composition. Recent surveys point towards habitat filtering as the key factor: naturally co-occurring microbial species tend to metabolically compete with each other. How does this competition not result in the exclusion of all but the most dominant species? We show that modeling microbes as sequential consumers of resources in the vein of the well-studied stable marriage problem can help resolve this apparent paradox. Complementarity in resource prioritization (hierarchical diauxic shifts) in our model allows multiple metabolic competitors to robustly coexist with each other. In addition to this, our model exhibits several interesting properties common to microbial consortia: specifically, multiple stable states and community restructuring in response to external perturbations. Our results are relevant to the understanding of polysaccharide utilization patterns by microbes residing in the human gut.