Evolutionary biology is rapidly developing from a mainly historic and descriptive into an experimental science with an increasingly detailed and quantitative understanding of evolutionary processes. This fosters the exciting question whether we may once predict future evolution. Motivated by this prospect, we try to understand the role of two fundamental evolutionary determinants: (i) the topography of the fitness landscape (‘what is possible’) and (ii) population dynamic parameters, such as population size, mutation rate and recombination rate (‘which solution will be realized’). We study these factors in systems of varying complexity, including single enzymes, bacteria and unicellular and multicellular fungi, using two general approaches: (1) systematic analyses of small-scale real fitness landscapes, (2) laboratory evolution experiments to gauge the dynamics and repeatability of evolution and its determinants. Our focus is on finding ‘organizing principles’ by combining analyses of molecular details of evolutionary change with those of statistical patterns across experimental systems and conditions. We collaborate closely with theoreticians, chemists, biophysicists, bioinformaticians and engineers, who provide us with models to design and interpret the experiments, with molecular details of the underlying mechanisms and with the tools to study phenotypic and genotypic changes at an increasing level of detail.

  1. Empirical fitness landscapes
Figure 2
Example of an empirical fitness landscape of 4 mutations in the gene for TEM-1 beta-lactamase that individually increase resistance to cefotaxime. From de Visser & Krug 2014 Nat Rev Genet 15: 480-490.

Determining the relationship between genotypes, phenotypes and fitness – the GPF map — is a central aim in biology. Evolutionary biologists are interested in the GPF map, since it determines the shape of the fitness landscape, with fundamental consequences for – among others — speciation, genetic robustness, the evolution of sex and recombination [6] and the predictability of evolution [1,3]. Simply put: it presents the possible evolutionary pathways and the likelihood that they are used by natural selection. At the smallest scale, the local fitness landscape is described by the distribution of fitness effects of beneficial mutations of a given genotype, which we determined long ago for a strain of E. coli [7], and more recently for the enzyme TEM-1 β-lactamase, where we estimated that ~3% of all mutations are beneficial, including several synonymous mutations and some with exceptionally large benefits [5]. At larger scales, the topography of the fitness landscape is determined by the epistatic interactions between mutations. Years ago, we have examined the nature of interactions among deleterious mutations in the alga Chlamydomonas moewusii using a statistical approach [8] and fungus Aspergillus niger by analyzing all 28 combinations of 8 mutations [6], to test for weak negative epistasis, which was required by a model for the maintenance of sex. We found no support for weak negative epistasis, but rather strong and variable epistasis – called sign epistasis — in the A. niger data set, leading to a rugged fitness landscape and disadvantages for recombination in simulations of adaptation [6]. More recently, we have analysed epistasis among beneficial mutations in TEM-1, which showed strong and prevailing negative interactions, particularly among large-effect mutations that were not chosen for their collective benefit [4]. We showed that this pattern of interactions could be explained from a nonlinear dependence of resistance on underlying phenotypes. A similar pattern of diminishing-returns epistasis was observed among beneficial mutations in the fungus A. nidulans, and also there an adapted version of Fisher’s geometric model, relating fitness to underlying phenotypes, could well explain the pattern of epistasis [2].

Key publications:

  • Gorter, F.A., J. Bobula, M.G.M. Aarts, B.J. Zwaan, R. Korona and J.A.G.M. de Visser. 2018. Local Fitness Landscapes Predict Yeast Evolutionary Dynamics in Directionally Changing Environments. Genetics, 208: 307-322.
  • Schoustra, S.E., S. Hwang, J. Krug and J.A.G.M. de Visser. 2016. Diminishing-returns epistasis among random beneficial mutations in a multicellular fungus. Proceedings of the Royal Society London B 283: 20161376.
  • De Visser, J.A.G.M. and J. Krug. 2014. Empirical fitness landscapes and the predictability of evolution. Nature Reviews Genetics 15: 480-490.
  • Schenk, M.F., I.G. Szendro, M.L.M. Salverda, J. Krug and J.A.G.M. de Visser. 2013. Patterns of epistasis between beneficial mutations in an antibiotic resistance enzyme. Molecular Biology and Evolution 30: 1779-1787.
  • Schenk, M.F., I.G. Szendro, J. Krug and J.A.G.M. de Visser. 2012. Quantifying the adaptive potential of an antibiotic resistance enzyme. PLoS Genetics 8: e1002783.
  • de Visser, J.A.G.M., S.-C. Park and J. Krug. 2009. Exploring the effect of sex on empirical fitness landscapes. The American Naturalist 174: S15-S30.
  • Rozen, D.E., J.A.G.M. de Visser and P.J. Gerrish. 2002. Fitness effects of fixed beneficial mutations in microbial populations. Current Biology 12: 1040-1045.
  • De Visser, J.A.G.M., R.F. Hoekstra and H. van den Ende. 1997. An experimental test for synergistic epistasis and its application in Chlamydomonas. Genetics 145: 815-819.

2. Dynamics and repeatability of evolution

Figure 3
Results from simulations of evolving asexual populations of different size on an empirically characterized 8-locus fitness landscape of the fungus Aspergillus niger. Arrows indicate mutation paths on this fitness landscape from a low-fitness genotype to higher-fitness genotypes; d = Hamming distance (H) to global optimum, genotypes with equal H are spread across the x-axis. Pathway repeatability, reflected by thickness of the arrows, increases non-monotonically with population size. From Szendro et al. 2013 PNAS 110: 571-576.

The systematic analyses mentioned above provide first glimpses of the properties of real fitness landscapes, but these glimpses are likely biased by the choice of the small set of mutations involved. As a complementary approach, we perform laboratory evolution experiments, where all possible mutations are allowed to contribute to evolution, and the combined impact of fitness landscape and population dynamic parameters on the dynamics and repeatability of evolution is studied. We do this using two different methods: (i) in vitro evolution with the antibiotic resistance enzyme TEM-1 β-lactamase using protocols for directed evolution, and (ii) in vivo evolution of E. coli, yeast and Aspergillus, using serial-transfer protocols. One major focus in these experiments has been on the effect of mutation supply rate (i.e. product of population size and mutation rate) in asexual populations, where clonal interference is expected to increase predictability [9,12,13,15,16], which may provide adaptive benefits to small populations by avoiding ‘local traps’ on a rugged fitness landscape [9,15]. We are also interested in possible benefits of recombination on such fitness landscapes, which we currently explore using in vitro recombination of TEM-1 [10]. Two other sources of potential constraints with potential relevance for predictability are being explored: (i) pleiotropic constraints from the simultaneous or alternating application of multiple antibiotics [12], and the rate of environmental change, studied by exposing yeast to different concentrations of heavy metals [11].

Key publications:

  • Salverda, M.L.M., J. Koomen, B. Koopmanschap, M.P. Zwart and J.A.G.M. de Visser. 2017. Adaptive benefits from small mutation supplies in an antibiotic resistance enzyme. Proceedings of the National Academy of Sciences USA 114: 12773-12778.
  • Pesce, D., N. Lehman and J.A.G.M. de Visser. 2016. Sex in a test tube: benefits of in vitro recombination. Philosophical Transactions of the Royal Society 371: 20150529.
  • Gorter, F.A., M.G.M. Aarts, B.J. Zwaan and J.A.G.M. de Visser. 2016. Dynamics of adaptation in experimental yeast populations exposed to gradual and abrupt change in heavy metal concentration. The American Naturalist 187: 110-119.
  • Schenk, M.F., S. Witte, M.L.M. Salverda, B. Koopmanschap, J. Krug and J.A.G.M. de Visser. 2015. Role of pleiotropy during adaptation of TEM-1 β-lactamase to two novel antibiotics. Evolutionary Applications 8: 248-260.
  • Szendro, I.G., J. Franke, J.A.G.M. de Visser and J. Krug. 2013. Predictability of evolution depends non-monotonically on population size. Proceedings of the National Academy of Sciences USA 110: 571-576.
  • Salverda, M.L.M., E. Dellus, F.A. Gorter, A.J.M. Debets, J. van der Oost, R.F. Hoekstra, D.S. Tawfik and J.A.G.M. de Visser. 2011. Initial mutations direct alternative pathways of protein evolution. PLoS Genetics 7: e1001321.
  • Rozen, D.E., M.G.J.L. Habets, A. Handel and J.A.G.M. de Visser. 2008. Heterogeneous adaptive trajectories of small populations on complex fitness landscapes. PLoS One 3: e1715.
  • De Visser, J.A.G.M. and D.E. Rozen. 2006. Clonal interference and the periodic selection of new beneficial mutations in Escherichia coli. Genetics 172: 2093-2100.

3. Eco-evolutionary feedback

Figure 1
Coevolution between the killing ability (a) of a yeast killer strain (K), harboring cytoplasmic toxin-producing RNA virus, and the toxin sensitivity (b) of an initially toxin-sensitive strain (S). Solid line, coevolution of both K and S; dashed line, asymmetric coevolution of K or S; dotted line, monoculture controls of K or S. From Pieczynska et al. 2016 Evolution 70: 1342-1353.

A fundamental problem for the predictability of evolution is the dependence of fitness on environmental conditions. Even in laboratory evolution experiments we cannot fully control environmental conditions, which may change during evolution, e.g. when mutants appear that produce a metabolite on which subsequent mutants may specialize [19]. Such eco-evolutionary feedback mechanisms introduce dependencies of fitness on the biotic environment, and may introduce frequency-dependent [21] and non-transitive fitness interactions [22], which complicate measurements of adaptation. We have studied the role of eco-evolutionary feedback in spatially-structured environments, allowing self-organization of populations, in E. coli populations evolving under resource competition [21] and yeast populations where interference competition plays a role via the production of anti-competitor toxins encoded by cytoplasmic viruses [18,20]. These studies showed that spatial environmental structure provides ample opportunities for frequency-dependent relationships to evolve. Presently, we study eco-evolutionary feedback also in the context of antibiotic resistance, where bacteria expressing β-lactamase lower the antibiotic concentration in the environment, creating opportunities for ‘beneficiary’ mutants (with lower resistance, but high fitness) to invade. We study these direct and indirect fitness consequences of β-lactamase production, among others, in small droplet cultures using millifluidic technology [17]. A related aim is to apply this technology to perform highly-replicated evolution experiments. Finally, we study fitness dependence on the biotic environment also in A. nidulans, where we aim to understand how this primitive multicellular fungus handles cancer-like mutants.

Key publications:

  • Cottinet, D., F. Condamine, N. Bremond, A.D. Griffiths, P.B. Rainey, J.A.G.M. de Visser, J. Baudry and J. Bibette. 2016. Lineage tracking for probing heritable phenotypes at single-cell resolution. PLoS One, 11: e0152395.
  • Pieczynska, M., D. Wloch-Salamon, R. Korona and J.A.G.M. de Visser. 2016. Rapid multiple-level coevolutionary dynamics in experimental populations of yeast killer and non-killer strains. Evolution 70: 1342-1353.
  • Rozen, D.E., N. Philippe, J.A. de Visser, R.E. Lenski and D. Schneider. 2009. Death and cannibalism in a seasonal environment facilitate bacterial coexistence. Ecology Letters 12: 34-44.
  • Wloch-Salamon, D.M., D. Gerla, R.F. Hoekstra and J.A.G.M. de Visser. 2008. Effect of dispersal and nutrient availability on the competitive ability of toxin producing yeast. Proceedings of the Royal Society London B 275: 535-541.
  • Habets, M.G.J.L., D.E. Rozen, R.F. Hoekstra and J.A.G.M. de Visser. 2006. The effect of population structure on the adaptive radiation of microbial populations evolving in spatially structured environments. Ecology Letters 9: 1041-1048.
  • De Visser, J.A.G.M. and R.E. Lenski. 2002. Long-term experimental evolution in Escherichia coli. XI. Rejection of non-transitive interactions as cause of declining rate of adaptation. BMC Evolutionary Biology 2: 19.