CAMP EVOLUTION VI - EVOLUTION ON
FITNESS LANDSCAPES
Fitness landscape is a unifying
concept of evolutionary biology. We will use it to consider a wide range of
seemingly unrelated evolutionary phenomena at all levels of organization.
1) Complex fitness landscapes
Fitness landscapes,
maps from the space of genotypes or phenotypes into fitness, is the key conceptual tool for studying
evolution of life. Usually, fitness landscapes are complex, due to epistasis,
deviation of fitness of a genotype from the product of contributions of the
constituent alleles. On top of generic fitness landscapes, it often makes sense
to consider special cases of epistasis: one-dimensional (fitness is an
arbitrary function of a simple, scalar characteristic of a genotype),
single-peak (adaptive evolution from any initial conditions will lead to the
same genotype), and monotonic (one of two alleles confers a higher fitness on
any genetic background). There is a specific fitness landscape for every
environment, and families of fitness landscapes are characterized by
bifurcations.
Whitlock MC (1995) Multiple fitness
peaks and epistasis. Ann. Rev. Ecol. Syst. 26, 601-629
de Visser
JAGM, Krug J (2014) Empirical fitness landscapes and the predictability of
evolution. Nature Reviews Genetics 15, 480-490
2. Fitness landscapes of molecules
2a) Compensated pathogenic deviations and ongoing expansion of protein
universe
Homologous sites of homologous
proteins (and of functional RNAs) may be under different modes of selection. In
particular, an amino acid which is pathogenic, at a particular site of a
particular protein, for humans, may be occupy the
homologous site of the wild-type homologous protein in another organism. This
phenomenon, known as compensated pathogenic deviation, can be used to study
fitness landscapes of proteins. Analogously, ongoing divergence of homologous
proteins even after they accumulated several neutral allele replacements per
site would be impossible without epistasis, and can reveal some of its
quantitative properties.
Povolotskaya IS, Kondrashov FA (2010) Sequence space and the ongoing
expansion of the protein universe. Nature 465, 922-926
Jordan DM et al. (2015)
Identification of cis-suppression of human disease mutations by comparative
genomics. Nature
524, 225-229
Storz F (2018) Compensatory mutations and
epistasis for protein function. Current Opinion in Structural Biology 50, 18-25
2b) Empirical fitness landscapes, experimental
evolution, and protein design
Existing proteins and
other biological objects cover only a tiny proportion of the space of
genotypes. Thus, complete fitness landscapes are hidden in the darkness of
non-existence. The only way to get them out of there is to create more
proteins. This can be done by taking the existing protein as a point of
departure and analyzing some volume of the space of genotypes around it. Such
studies revealed pervasive narrowing epistasis. Alternatively, fitness
landscape can be explored by artificial evolution of a protein directed toward
some new function. Information about fitness landscapes will facilitate
designing of proteins with novel functions.
Philip A. Romero PA, Arnold
FH (2009) Exploring protein fitness landscapes by directed evolution. Nature
Reviews Molecular Cell Biology 10, 866-876
Sarkisyan KS et al. (2016) Local fitness
landscape of the green fluorescent protein. Nature 533, 397-401
Wrenbeck EE (2017) Deep sequencing methods
for protein engineering and design. Current Opinion in Structural Biology 45, 36-44
3. Fitness landscapes of cells
3a) Evolution of functional units within a cell
Functioning of a living
cell is governed by complex networks of interactions between genes, proteins,
and small molecules. Evolution of a gene or a protein is restricted by
interactions in which it participates, and these interactions can be understood
only in the context of the fitness landscape of the cell. This paradigm
produces new insights into evolution of individual genes and proteins. When
this context evolves, even the key property of a gene, its essentiality, can
also evolve. Landscapes of multiple regulatory genes which perform similar
functions can be also very similar to each other. Promiscuous interactions
facilitate the origin of new regulatory functions.
Haldane A et al. (2014) Biophysical fitness landscapes for transcription
factor binding sites. PLoS Comput.
Biol. 10(7): e1003683
Friedlander
T et al. (2016) Evolution of new regulatory functions on biophysically
realistic fitness landscapes. Nature Comm. 8: 216
Rancati G et al. (2017) Emerging and
evolving concepts in gene essentiality. Nature Reviews Genetics doi:10.1038/nrg.2017.74
3b) Global properties of fitness
landscapes of cells
Fitness landscape of even
the simplest cell must be incredibly complex. It has longed being suspected
that they are very "rugged" (possess multiple peaks), and recent data
on adaptive evolution confirm this. Networks of interactions within cells are
known to be modular, but the reasons for this remain obscure. One possibility
is that modularity emerges as a result of selection for higher computational efficiency,
if the fitness landscape often undergoes bifurcations. The global fitness
landscape of a cell can be very fragile and undergo many radical changes after
a single amino acid substitution within an important regulatory protein.
Peter L. Freddolino PL et al. (2012) Fitness landscape transformation
through a single amino acid change in the rho terminator. PLoS
Genetics 8(5): e1002744
Van Clevea
J, Weissman DB (2015) Measuring ruggedness in fitness
landscapes. PNAS 112, 7345-7346
Tosh CR (2016) Can
computational efficiency alone drive the evolution of
modularity in neural networks? Scientific Reports 6:31982
4. Fitness landscapes of organisms
4a) Mutation load and the evolution of sex
Genomic deleterious
mutation rate in humans is ~10, and, as long as fitness landscape is
non-epistatic, this must lead to an intolerable genetic load. Some data
indicate that narrowing epistasis is present, however,
more information is needed. Whatever forces are responsible for generation of
linkage disequilibria and for overrepresented genotypes to have reduced fitnesses - the two conditions for the advantage of sex and
recombination - they cannot cause a substantial advantage of a
recombination-increasing modifier unless selection differential is comparable
to the standard deviation of fitness potential. Such strong, directional
selection is not out of the question; however, it is not clear why it should be
pervasive among eukaryotes.
Barton NH (2009) Why sex
and recombination? Cold Spring Harbor Symp.
Quant. Biol. 54, 1-9
Kondrashov AS (2017) Crumbling Genome, Wiley.
Sohail M et al. (2017) Negative selection
in humans and fruit flies involves synergistic epistasis. Science 356, 539-542
4b) Punctuated equilibrium revisited
Over 30 years ago, a
debate raged among evolutionary biologists on whether evolution is gradual or
consists of short bursts alternating with periods of stasis. As usual in our
field, the subject was eventually quietly abandoned, without any clear resolution.
However, the only codimension 1 bifurcation of a
fitness landscape, "fold catastrophe", leads to an abrupt
disappearance of a fitness peak and, thus, appears to be capable of inducing a
rapid adaptive walk. Recent phylogenetic data made it possible to study such
walks in evolution of individual proteins. So far, the results are mixed, but the
subject is worth revisiting.
Dodson
MM, Hallam A (1977) Allopatric speciation and the fold catastrophe. American Naturalist 111, 415-433
Richter H, Engelbrecht A, eds. (2014) Recent Advances in the Theory
and Application of Fitness Landscapes. Springer, chapter 11
Pennell MW (2014). Is
there room for punctuated equilibrium in macroevolution? Trends in Ecology and
Evolution 29, 23-32
5. Fitness landscapes of populations
5a) Frequency-dependent selection, multilevel selection,
evolution of cooperation
Fitness landscape of the
population can depend on its state, which readily leads to extremely complex
dynamics, especially when the population is structured. There is no general
theory of such dynamics; instead, specific biological problems can be
considered. Some data indicate that in the course of human history negative
frequency-dependent selection favored rare facial features. Pairwise
interactions between individuals can lead to evolution of indirect reciprocity,
based on reputation. Group selection can favor strong emotional bound between
individuals who together shared a negative experience.
Nowak
MA, Sigmund K (2005) Evolution of indirect reciprocity. Nature 437, 1291-1298
Sheehan MJ, Nachman MW (2014). Morphological and
population genomic evidence that human faces have evolved to signal individual
identity. Nature Communications 5:4800
Harvey Whitehouse H et
al. (2017) The evolution of extreme cooperation via
shared dysphoric experiences. Scientific Reports 7: 44292
5b) Red Queen dynamics, Darwinian
extinction, community assembly
Fitness landscape of the
population can also depend on populations of other species in the same
community. Again, the resulting dynamics can be very complex, and no general
theory exists. In many cases ecological interactions between coevolving
populations lead to endless changes of their fitness landscapes, known as Red
Queen dynamics. A number of nonlinear effects can cause extinction of a
population, as a result of its adaptive evolution. The process of community
assembly depends critically on coevolution of interacting populations.
Webb C
(2003) A complete classification of Darwinian extinction in ecological interactions. American Naturalist 161, 181-205
Nemergut DR et al. (2013) Patterns and processes
of microbial community assembly. Microbiol. Mol. Biol. Rev. 77, 342-356
Richter H, Engelbrecht A, eds. (2014) Recent Advances in the Theory
and Application of Fitness Landscapes. Springer, chapters 11 and 12