A cross-section of the fitness landscape of dihydrofolate reductase.

In vitro molecular evolution is regarded as a hill-climbing on a fitness landscape in sequence space, where the 'fitness' is a quantitative measure of a certain physicochemical property of a biopolymer. We analyzed a 'cross-section' of the enzymatic activity landscape of dihydrofolate reductase (DHFR) by using a method of analysis of a fitness landscape. We limited the sequence space of interest to the five-dimensional sequence space, where the coordinate corresponds to the 1st, 16th, 20th, 42nd and 92nd site in the DHFR sequence. Thirty six mutants mapped into the limited sequence space were taken in the analysis. As a result, the cross-section is of the rough Mt Fuji type based on the mutational additivity. The ratio of the mean slope to the roughness is 2.8 and the Z-score of the original ratio against a distribution of random references is 7.0, which indicates a large statistical significance. The existence of such a cross-section was discussed in terms of the occurrence probability of sets of five sites distantly separated from each other on the DHFR 3D structure. Our results support the effectiveness of the evolution strategy which exploits the accumulation of advantageous single point mutations in such a cross-section.     

From Amino Acid Landscape to Protein Landscape: Analysis of Genetic Codes in Terms of Fitness Landscape

Assigning the values of a certain physicochemical property for individual amino acids to the corresponding codons, we can make an amino acid property "landscape" on a four valued three dimensional sequence space from a genetic code table. Eleven property landscapes made from the standard genetic code (SGC) were analyzed. The evaluation of correlation for each landscape is done by theta value, which represents the ratio of the mean slope (as an additive term) to the degree of roughness (as a nonadditive term). The theta-values for hydropathy indices, polarity, specific heat, and beta-sheet propensity were considerably large with respect to SGC. This implies that the additivity of the contribution from each letter holds for these properties. To clarify the meaning of the so-called mutational robustness of SGC, we next examined correlations between the amino acid property and the actual "site fitnesses" of a protein. The site fitnesses were derived from a set of binding preference scores of amino acid residues at every site in MHC class I molecule binding peptides (Udaka et al. in press). We found that the SGC's theta value for an amino acid property is correlated with the significance of the property in the protein function. Adaptive walk simulation on fitness (= affinity) landscapes in a base sequence space for these model peptides confirmed better evolvability due to the introduction of SGC.     

Thermodynamical interpretation of an adaptive walk on a Mt. Fuji-type fitness landscape: Einstein relation-like formula holds in a stochastic evolution

We have theoretically studied the statistical properties of adaptive walks (or hill-climbing) on a Mt. Fuji-type fitness landscape in the multi-dimensional sequence space through mathematical analysis and computer simulation. The adaptive walk is characterized by the "mutation distance" d as the step-width of the walker and the "population size" N as the number of randomly generated d-fold point mutants to be screened. In addition to the fitness W, we introduced the following quantities analogous to thermodynamical concepts: "free fitness" G(W) is identical with W+T x S(W), where T is the "evolutionary temperature" T infinity square root of d/lnN and S(W) is the entropy as a function of W, and the "evolutionary force" X is identical with d(G(W)/T)/dW, that is caused by the mutation and selection pressure. It is known that a single adaptive walker rapidly climbs on the fitness landscape up to the stationary state where a "mutation-selection-random drift balance" is kept. In our interpretation, the walker tends to the maximal free fitness state, driven by the evolutionary force X. Our major findings are as follows: First, near the stationary point W*, the "climbing rate" J as the expected fitness change per generation is described by J approximately L x X with L approximately V/2, where V is the variance of fitness distribution on a local landscape. This simple relationship is analogous to the well-known Einstein relation in Brownian motion. Second, the "biological information gain" (DeltaG/T) through adaptive walk can be described by combining the Shannon's information gain (DeltaS) and the "fitness information gain" (DeltaW/T).     

Surveying a local fitness landscape of a protein with epistatic sites for the study of directed evolution

We present a method for analysis of a fitness landscape of a biopolymer with significantly epistatic sites. The analysis is based on a quasi-additive fitness model. The fitness model is constructed with additive terms conducted by "site-fitness" and epistatic terms conducted by "pair-fitness," where the site-fitness is a fitness contribution from an independent residue and the pair-fitness is a fitness contribution from a pair of epistatic residues. As a case study, we analyzed the sequence-fitness data for 45 clones of thermostable prolyl endopeptidase mutants. They were generated by a mutation scrambling method, which can accumulate advantageous mutations. The fitness contributions from 14 single-point mutations including E67Q and Q656R were identified by the analysis. As a result, we found that the fitness model with a significant epistatic term by a pair of the 67th site and 656th site was in good agreement with the experimental data and that the explored landscape in the binary 14-dimensional sequence space is still a mountainous landscape with twin peaks. The validity was supported by the analysis of mutant fitness distributions derived from another mutation scrambling experiment and by (3D) structural data.     

Thermodynamical interpretation of evolutionary dynamics on a fitness landscape in an evolution reactor,II

In our previous report [Aita, T., Morinaga, S., Hosimi, Y., 2004. Thermodynamical interpretation of evolutionary dynamics on a fitness landscape in an evolution reactor I. Bull. Math. Biol. 66, 1371–1403], an analogy between thermodynamics and adaptive walks on a Mt. Fuji-type fitness landscape in an artificial selection system was presented. Introducing the ‘free fitness’ as the sum of a fitness term and an entropy term and ‘evolutionary force’ as the gradient of free fitness on a fitness coordinate, we demonstrated that the adaptive walk (=evolution) is driven by the evolutionary force in the direction in which free fitness increases. In this report, we examine the effect of various modifications of the original model on the properties of the adaptive walk. The modifications were as follows: first, mutation distance d was distributed obeying binomial distribution; second, the selection process obeyed the natural selection protocol; third, ruggedness was introduced to the landscape according to the NK model; fourth, a noise was included in the fitness measurement. The effect of each modification was described in the same theoretical framework as the original model by introducing ‘effective’ quantities such as the effective mutation distance or the effective screening size.     

Theory of evolutionary molecular engineering through simultaneous accumulation of advantageous mutations

We examined the effectiveness of an "adaptive leap" strategy using the "mutation scrambling" method as an efficient optimization technique (Uchiyama, 2000;J. Biochem.128, 441-447) for cases where mutational (rough) additivity holds in fitness. The mutation scrambling method is composed of the following three processes: (1) preliminary selection of several advantageous single-point mutations introduced in a wild-type sequence; (2) preparation of various multiple-point mutants incorporating the advantageous mutant residue or wild-type residue at each of the selected sites, by scrambling the mutant residues and wild-type residues (this process is called mutation scrambling); and (3) selection of the fittest through screening of the mutant pool. The fitness distribution in the mutant pool is controlled by the mixing ratio of the mutant residues to the wild-type residues. We focused on the mutant fitness distribution and obtained the optimal mixing ratio which efficiently generates superior multiple-point mutants with high fitnesses. As a result, we found that the optimal ratio lies between 7/3 and 9/1 in realistic cases. Particularly, this strategy works well in cases where the number of component mutations is large and the size of the population to be screened is small. Analysis of the mutant fitness distributions with various mixing ratios is also useful to explore local fitness landscapes.     

Thermodynamical interpretation of evolutionary dynamics on a fitness landscape in a evolution reactor,I

A theory for describing evolution as adaptive walks by a finite population with M walkers (M ≥ 1) on an anisotropic Mt. Fuji-type fitness landscape is presented, from a thermodynamical point of view. Introducing the ‘free fitness’ as the sum of a fitness term and an entropy term and ‘evolutionary force’ as the gradient of free fitness on a fitness coordinate, we demonstrate that the behavior of these theoretical walkers is almost consistent with the thermodynamical schemes. The major conclusions are as follows: (1) an adaptive walk (=evolution) is driven by an evolutionary force in the direction in which free fitness increases; (2) the expectation of the climbing rate obeys an equation analogous to the Einstein relation in Brownian motion; (3) the standard deviation of the climbing rate is a quantity analogous to the mean thermal energy of a particle, kT (×constant). In addition, on the interpretation that the walkers climb the landscape by absorbing ‘fitness information’ from the surroundings, we succeeded in quantifying the fitness information and formulating a macroscopic scheme from an informational point of view.     

Reciprocal sign epistasis between frequently experimentally evolved adaptive mutations causes a rugged fitness landscape

The fitness landscape captures the relationship between genotype and evolutionary fitness and is a pervasive metaphor used to describe the possible evolutionary trajectories of adaptation. However, little is known about the actual shape of fitness landscapes, including whether valleys of low fitness create local fitness optima, acting as barriers to adaptive change. Here we provide evidence of a rugged molecular fitness landscape arising during an evolution experiment in an asexual population of Saccharomyces cerevisiae. We identify the mutations that arose during the evolution using whole-genome sequencing and use competitive fitness assays to describe the mutations individually responsible for adaptation. In addition, we find that a fitness valley between two adaptive mutations in the genes MTH1 and HXT6/HXT7 is caused by reciprocal sign epistasis, where the fitness cost of the double mutant prohibits the two mutations from being selected in the same genetic background. The constraint enforced by reciprocal sign epistasis causes the mutations to remain mutually exclusive during the experiment, even though adaptive mutations in these two genes occur several times in independent lineages during the experiment. Our results show that epistasis plays a key role during adaptation and that inter-genic interactions can act as barriers between adaptive solutions. These results also provide a new interpretation on the classic Dobzhansky-Muller model of reproductive isolation and display some surprising parallels with mutations in genes often associated with tumors.     

Multiple adaptive substitutions during evolution in novel environments

We consider an asexual population under strong selection-weak mutation conditions evolving on rugged fitness landscapes with many local fitness peaks. Unlike the previous studies in which the initial fitness of the population is assumed to be high, here we start the adaptation process with a low fitness corresponding to a population in a stressful novel environment. For generic fitness distributions, using an analytic argument we find that the average number of steps to a local optimum varies logarithmically with the genotype sequence length and increases as the correlations among genotypic fitnesses increase. When the fitnesses are exponentially or uniformly distributed, using an evolution equation for the distribution of population fitness, we analytically calculate the fitness distribution of fixed beneficial mutations and the walk length distribution.     

A sharp minimum on the mean number of steps taken in adaptive walks

It was recently conjectured by H.A. Orr that from a random initial point on a random fitness landscape of alphabetic sequences with one-mutation adjacency, chosen from a larger class of landscapes, no adaptive algorithm can arrive at a local optimum in fewer than on average e-1 steps. Here, using an example in which the mean number of steps to a local optimum equals (A-1)/A, where A is the number of distinct "letters" in the "alphabet" from which sequences are constructed, it is shown that as originally stated, the conjecture does not hold. It is also demonstrated that (A-1)/A is a sharp minimum on the mean number of steps taken in adaptive walks on fitness landscapes of alphabetic sequences with one-mutation adjacency. As the example that achieves the new lower bound has properties that are not often considered as potential attributes for fitness landscapes-non-identically distributed fitnesses and negative fitness correlations for adjacent points-a weaker set of conditions characteristic of more commonly studied fitness landscapes is proposed under which the lower bound on the mean length of adaptive walks is conjectured to equal e-1.     

Dominance and overdominance of mildly deleterious induced mutations for fitness traits in Caenorhabditis elegans

We estimated the average dominance coefficient of mildly deleterious mutations (h, the proportion by which mutations in the heterozygous state reduce fitness components relative to those in the homozygous state) in the nematode Caenorhabditis elegans. From 56 worm lines that carry mutations induced by the point mutagen ethyl methanesulfonate (EMS), we selected 19 lines that are relatively high in fitness and estimated the viabilities, productivities, and relative fitnesses of heterozygotes and homozygotes compared to the ancestral wild type. There was very little effect of homozygous or heterozygous mutations on egg-to-adult viability. For productivity and relative fitness, we found that the average dominance coefficient, h, was approximately 0.1, suggesting that mildly deleterious mutations are on average partially recessive. These estimates were not significantly different from zero (complete recessivity) but were significantly different from 0.5 (additivity). In addition, there was a significant amount of variation in h among lines, and analysis of average dominance coefficients of individual lines suggested that several lines showed overdominance for fitness. Further investigation of two of these lines partially confirmed this finding.     

Fitness landscapes for coupled map lattices

Our goal is to match some dynamical aspects of biological systems with that of networks of coupled logistic maps. With these networks we generate sequences of iterates, convert them to symbol sequences by coarse-graining, and count the number of times combinations of symbols occur. Comparison of this with the number of times these combinations occur in experimental data—a sequence of interbeat intervals for example—is a measure of the fitness of each network to describe the target data. The most fit networks provide a cartoon that suggests a decomposition of the experimental data into a component that may be produced by a simple dynamical subsystem, and a residual component, the result of detailed, particular characteristics of the system that generated the target data. In the space of all network parameters, each point corresponds to a particular network. We construct a fitness landscape when we assign a fitness to each point. Because the parameters are distributed continuously over their ranges, and because fitnesses are estimated numerically, any plot of the landscape involves a finite sample of parameter values. We’ll investigate how the local landscape geometry changes when the array of sample parameters is refined, and use the landscape geometry to explore complex relations between local fitness maxima.     

Adaptive walks by the fittest among finite random mutants on a Mt. Fuji-type fitness landscape II. Effect of small non-additivity

We examined properties of adaptive walks by the fittest on “rough Mt. Fuji-type” fitness landscapes, which are modeled by superposing small uncorrelated random component on an additive fitness landscape. A single adaptive walk is carried out by repetition of the evolution cycle composed of (1) mutagenesis process that produces random d-fold point mutants of population size N and (2) selection process that picks out the fittest mutant among them. To comprehend trajectories of the walkers, the fitness landscape is mapped into a (x, y, z)-space, where x, y and z represent, respectively, normalized Hamming distance from the peak on the additive fitness landscape, scaled additive fitness and scaled non-additive fitness. Thus a single adaptive walk is expressed as the dynamics of a particle in this space. We drew the “hill-climbing” vector field, where each vector represents the most probable step for a walker in a single step. Almost all of the walkers are expected to move along streams of vectors existing on a particular surface that overlies the (x, y)-plane, toward the neighborhood of a characteristic point at which a mutation-selection-random drift balance is reached. We could theoretically predict this reachable point in the case of random sampling search strategy. Received: 1 March 2000 / Published online: 3 August 2000     

The fitness effect of mutations across environments: Fisher's geometrical model with multiple optima

When are mutations beneficial in one environment and deleterious in another? More generally, what is the relationship between mutation effects across environments? These questions are crucial to predict adaptation in heterogeneous conditions in a broad sense. Empirical evidence documents various patterns of fitness effects across environments but we still lack a framework to analyze these multivariate data. In this article, we extend Fisher's geometrical model to multiple environments determining distinct peaks. We derive the fitness distribution, in one environment, among mutants with a given fitness in another and the bivariate distribution of random mutants’ fitnesses across two or more environments. The geometry of the phenotype‐fitness landscape is naturally interpreted in terms of fitness trade‐offs between environments. These results may be used to fit/predict empirical distributions or to predict the pattern of adaptation across heterogeneous conditions. As an example, we derive the genomic rate of substitution and of adaptation in a metapopulation divided into two distinct habitats in a high migration regime and show that they depend critically on the geometry of the phenotype‐fitness landscape.     

Adaptation in Tunably Rugged Fitness Landscapes: The Rough Mount Fuji Model

Much of the current theory of adaptation is based on Gillespie’s mutational landscape model (MLM), which assumes that the fitness values of genotypes linked by single mutational steps are independent random variables. On the other hand, a growing body of empirical evidence shows that real fitness landscapes, while possessing a considerable amount of ruggedness, are smoother than predicted by the MLM. In the present article we propose and analyze a simple fitness landscape model with tunable ruggedness based on the rough Mount Fuji (RMF) model originally introduced by Aita et al. in the context of protein evolution. We provide a comprehensive collection of results pertaining to the topographical structure of RMF landscapes, including explicit formulas for the expected number of local fitness maxima, the location of the global peak, and the fitness correlation function. The statistics of single and multiple adaptive steps on the RMF landscape are explored mainly through simulations, and the results are compared to the known behavior in the MLM model. Finally, we show that the RMF model can explain the large number of second-step mutations observed on a highly fit first-step background in a recent evolution experiment with a microvirid bacteriophage.     

A minimum on the mean number of steps taken in adaptive walks

I consider the adaptation of a DNA sequence when mutant fitnesses are drawn randomly from a probability distribution. I focus on "gradient" adaptation in which the population jumps to the best mutant sequence available at each substitution. Given a random starting point, I derive the distribution of the number of substitutions that occur during adaptive walks to a locally optimal sequence. I show that the mean walk length is a constant:L = e-1, where e approximately 2.72. I argue that this result represents a limit on what is possible under any form of adaptation. No adaptive algorithm on any fitness landscape can arrive at a local optimum in fewer than a mean of L = e-1 steps when starting from a random sequence. Put differently, evolution must try out at least e wild-type sequences during an average bout of adaptation.     

Environmental change exposes beneficial epistatic interactions in a catalytic RNA

Natural selection drives populations of individuals towards local peaks in a fitness landscape. These peaks are created by the interactions between individual mutations. Fitness landscapes may change as an environment changes. In a previous contribution, we discovered a variant of the Azoarcus group I ribozyme that represents a local peak in the RNA fitness landscape. The genotype at this peak is distinguished from the wild-type by four point mutations. We here report ribozyme fitness data derived from constructing all possible combinations of these point mutations. We find that these mutations interact epistatically. Importantly, we show that these epistatic interactions change qualitatively in the three different environments that we studied. We find examples where the relative fitness of a ribozyme can change from neutral or negative in one environment, to positive in another. We also show that the fitness effect of a specific GC-AU base pair switch is dependent on both the environment and the genetic context. Moreover, the mutations that we study improve activity at the cost of decreased structural stability. Environmental change is ubiquitous in nature. Our results suggest that such change can facilitate adaptive evolution by exposing new peaks of a fitness landscape. They highlight a prominent role for genotype-environment interactions in doing so.     

Male-female interactions in Drosophila melanogaster: model with one-locus and three-alleles

We develop a fertility model of fitness that is general in that it does not assume that the fitnesses of the mating combinations are symmetrical or that they are additive or multilicative (i. e., that they can be inferred from fitnesses of the two genotypes involved in a mating). %he model considers one locus with three alleles. An experimental test with Drosophila rnelanogaster confirms that the fitnesses of the mating types depart from both additivity (or multiplicativity) and symmetry although this last property is of no consequence for the development of analytical models). urnerical simulations yield the same, or very nearly the same, equilibrium freuencies as the analytical model, independently of whether or not Hardy-Weinberg equilibrium Trequencies are assumed at the beginning of each selection cycle.     

The fitness effect of mutations across environments: a survey in light of fitness landscape models

The fitness effects of mutations on a given genotype are rarely constant across environments to which this genotype is more or less adapted, that is, between more or less stressful conditions. This can have important implications, especially on the evolution of ecological specialization. Stress is thought to increase the variance of mutations' fitness effects, their average, or the number of expressed mutations. Although empirical evidence is available for these three mechanisms, their relative magnitude is poorly understood. In this paper, we propose a simple approach to discriminate between these mechanisms, using a survey of empirical measures of mutation effects in contrasted environments. This survey, across various species and environments, shows that stress mainly increases the variance of mutations' effects on fitness, with a much more limited impact on their average effect or on the number of expressed mutations. This pattern is consistent with a simple model in which fitness is a Gaussian function of phenotypes around an environmentally determined optimum. These results suggest that a simple, mathematically tractable landscape model may not be quantitatively as unrealistic as previously suggested. They also suggest that mutation parameter estimates may be strongly biased when measured in stressful environments.