Department of Physics • University of Cologne

Given a list of candidate microbial species, which community should one form to optimize a target function? To answer this question, we must develop predictive quantitative models of community function, but doing so has been difficult due to the complex web of interactions that take place in most communities. To address this challenge, we are drawing inspiration from the theory of quantitative genetics, in particular the concept of global epistasis on fitness landscapes. We find that similar to genetics, community function follows simple, predictable patterns as a result of widespread interactions. These simple patterns allow us to learn the quantitative map between community composition and function in synthetic consortia and rationally optimize their function. Our work seeks to place the goals of engineering biological function across scales, from the molecular to the organismal to the ecological, under a common theoretical and methodological framework that can lead to the flow of ideas and methodologies across fields.

Viral evolution is largely defined by the fitness of novel variants of the virus. This fitness, and therefore future evolution, is in principle predictable. However, to make such predictions, we need to address two properties of the viral fitness landscape: multidimensionality and variability. Using my work on human pathogenic viruses including HIV-1, influenza virus and SARS-CoV-2, I will illustrate how analysis of big genomic data allows to address these challenges, revealing interactions between sites and alternating modes of selection. I will show how this approach allowed to understand emergence of drug resistance in influenza, and describe a previously uncharacterized escape from T-cell immune pressure in SARS-CoV-2.

We aimed at investigating the evolutionary path towards resistance to microtubule hyper-stabilization. We evolved in the lab for over 190 generations yeast cells carrying the hyper- stabilizing tub2-150 mutation in the beta-tubulin gene. We found that tub2-150 mutants were able to evolve and recover the initial growth defect by two temporally distinct solutions. The first evolutionary solution was the selection of recurrent aneuploidies driven by the beneficial duplication of specific genes. In the long - term evolution, aneuploidy was substituted by fitter point mutations in TUB genes. Some representative TUB mutations, when re-introduced in the mutant ancestor, rescued growth of tub2-150 cells by recovering microtubule dynamics. Our results provide new insights on adaptive evolution following the perturbation of an essential component such as microtubules, and potentially lead to the development of new strategies to counteract the emergence of resistance to drugs which stabilize microtubules.

Variation, i.e. the molecular and phenotypic changes caused by random mutations, is a crucial component of evolutionary processes. One biologically relevant example for which variation can be modelled computationally is the RNA genotype-phenotype (GP) map. This GP map links RNA sequences to their folded secondary structures and thus allows us to identify structural changes after sequence mutations. In this talk, I will describe some recent progress on the biophysics of this GP map, compare the effect of short insertions/deletions (indels) to that of substitutions and discuss what the properties of this GP map mean for evolutionary processes.

Many viral genomes consist of different segments, but some viruses individually package these segments into separate virus particles. For these so-called multipartite viruses to propagate, each genome segment must be transmitted between hosts, as the incomplete virus can no longer replicate. This apparently inefficient genome organization is common among plant viruses, raising the question of whether it confers any benefits. We first used a combination of theoretical and experimental work to consider whether the predicted cost to infection associated with multipartitism is real. Next, we explore hypotheses on the benefits associated with multipartitism, finding support for the hypothesis that rapid change in the frequency of viral genome segments is adaptive. Finally, there is a growing awareness that some animal viruses with segmented genomes produce virus particles missing genome segments. We show that these incomplete virus particles can cooperatively spread infection, and consider the implications for infectivity and virus evolution.

The DNA sequencing revolution has revealed enormous microbial diversity. Less well known is that, for many species, this extends down to much finer scales with strains, sub-strains, and sub-sub-strains coexisting, directly competing, and continually evolving. Why doesn’t “survival of the fittest” drive most of these extinct? After introducing some striking examples, the possibility that such diversity is a general consequence of the complexity of biology will be explored via simple models and statistical physics approaches.

Extensive quantitative experiments in the past decades have established simple empirical laws of resource allocation obeyed by exponentially growing bacteria subjected to different environmental or genetic perturbations. Combinations of these laws, together with their kinetic extensions, have led to quantitative account of a number of long-standing phenomena in microbiology. In this talk, I will briefly recount how we embarked on the path of top-down phenomenological studies, exhibit the striking universality of the law of ribosome allocation across diverse bacterial lineages, and describe a recent study establishing how this law is implemented molecularly via simple regulatory processes. The findings provide a rare view of “dimensional reduction” by a living cell, i.e., how a cell manages to collapse the complex, high-dimensional dynamics of metabolic reactions underlying cell growth to quantitatively “perceive” the growth rate, and allocate resources in accordance to the growth rate. Overall, these studies showcase how the basic methodology of classical physics can be used to discover simple organizing principles of living systems and construct quantitative, predictive theories linking molecules to cellular behaviors.

The average affinity of specific antibodies increases dramatically over the course of an immune response. This increase is the result of a Darwinian process in which B lymphocytes undergo iterative cycles of random hypermutation of their immunoglobulin genes, followed by selective proliferation of clones bearing affinity- enhancing mutations. This evolutionary process takes place in highly dynamic microanatomical structures known as germinal centers, which arise within secondary lymphoid organs upon infection or immunization. Our work combines intravital multiphoton microscopy with mouse genetics to study how the dynamics of B and T lymphocytes within germinal centers shapes the evolution of the high-affinity antibodies that are crucial to protection from infectious disease.

Parallels between biological evolution and statistical physics are well-known and have led to fruitful exchanges and concepts. One such concept is the fitness landscape which, analogous to the energy landscape of a disordered system, is an abstract representation of the environment in which a biological population evolves. Evolution in changing environments is still poorly understood. In this talk we consider a recently introduced model for antibiotic resistance evolution in bacteria. The fitness landscape changes with the antibiotic concentration, giving rise to tradeoffs between adaptation to low and high antibiotic concentrations. We show that the adaptive evolution under slowly changing antibiotic concentration can be mapped to the dynamics of a class of athermal driven disordered systems, which in turn are paradigmatic models for memory formation in matter.

Microbes often colonize spatially-constrained habitats, such as pores in the skin or crypts in the colon. The resulting micro-communities can be very stable and contribute to the long-term function of our microbiomes. Due to a lack of dynamical observations, it is however unclear how these communities and their ecological functions arise. By monitoring and modeling microbial populations in microfluidic channels of systematically varied size, we find a rich spectrum of scale-dependent dynamical patterns that are controlled by the competition between density-dependent outflow and population growth. These results elucidate how the injection of degrees of freedom, driven by cell proliferation, can drive a non-equilibrium phase transition (different from MIPS) and suggest that the mechanics jammed cellular packings can influence the evolutionary dynamics of dense microbial populations

Presentation of projects A1-A5, B2, B6-B8

Presentation of projects A6, B1, B3, B4, B5

Some of the most well-known examples of adaptive evolution involve single mutations with large beneficial effects. However, predicting the evolutionary impact and fate of a new mutation remains difficult. To simplify this problem, we often bin mutations into categories that are expected to have distinct fitness impacts (e.g. synonymous vs. non-synonymous; coding vs. non-coding). However, there is little empirical evidence for differential fitness effects of mutational types across the genome, because prior work has largely focused on the distribution of fitness effects (DFE) for beneficial mutations in single genes. Thus, the general importance of mutation type relative to other factors such as genetic background and environment remains unclear. In this talk, I will describe our efforts to address this problem using bacterial strains, each with a single mutation in its genome that either occurred under strong selection or under genetic drift. Briefly, I suggest that the power and promise of new mutations depends more on the genetic background on which they occur, and very little on the mutational category to which they belong.

Bacterial evolution of resistance to antibiotics is a major public health concern. Once prevalent, resistant strains can transmit among patients and become difficult to eradicate. However, ultimately, the resistant strains must originate from a de novo mutation or horizontal gene transfer event in a single bacterial cell. Using high-replicate experiments with Pseudomonas aeruginosa facing streptomycin as a model system, we quantified the probability that a single resistant cell establishes a large surviving lineage. Strikingly, this probability drops to near zero well below the standardly measured minimum inhibitory concentration (MIC) of the antibiotic. We show that this result can be explained by a simple model in which the antibiotic has partial inhibitory effects below the MIC, and each cell has an independent chance of establishment.

The rates at which bacterial populations grow have been studied for many decades, and have led to the formulation of deterministic growth laws that describe “macroscopic” properties of populations. However, this ignores heterogeneities within populations of cells. Current methods allow us to look at individual cells, and they reveal interesting distributions and lineage correlations of growth variables that present new explanatory challenges. Further, when populations of growing bacteria are exposed to antibiotics, cells may die and divide at comparable rates, and it becomes important to understand how cells in an identical environment suffer different fates. We present empirical observations both at the population and single-cell level, and propose stochastic models of intracellular processes to explain the observations.