compensatory evolution - senecio7の日記

Povolotskaya IS & Kondrashov FA 2010 Sequence space and the ongoing expansion of the protein universe. Nature 465:922-926.

~98 per cent of sites cannot accept an amino-acid substitution at any given moment
a vast majority of all sites may eventually be permitted to evolve when other, compensatory, changes occur

~3.5 × 10⁹ yr has not been enough to reach the limit of divergent evolution of proteins

the rate of evolution may be inhibited by two factors
non-epistatic negative selection, where the fitness impact of every allele is independent of genetic context and environmental factors

or a high prevalence of sign epistasis
sign epistasis is a property of fitness landscapes such that an amino acid at a specific site confers high fitness in one genetic background and low fitness in another
if such situations are common, the fitness landscape can be described as rugged
the order in which substitutions accumulate is restricted
with the result that only a small fraction of the substitutions that are possible in at least one genetic background can occur at any given moment
the rate of evolution will be slow
because fitness ridges will consist of fewer and longer evolutionarily accessible paths in sequence space
evolving proteins may meander through sequence space on such a rugged fitness landscape for a long time before reaching the limit of their divergence

the limit of sequence divergence for proteins under non-epistatic negative selection is much closer, in terms of protein distance, than that for neutrally evolving protein sequences
such selection reduces the time to reach this limit
the ongoing divergence of ancient proteins (Fig. 3) is inconsistent with non-epistatic negative selection being the main factor responsible for their slow evolution

as first recognized by John Maynard Smith
"functional proteins must form a continuous network which can be traversed by unit mutational steps without passing through nonfunctional intermediates"

although some aspects of the protein fitness landscape (that is, the network) have been probed, we remain largely ignorant of its global structure spanning sequence space
our data indicate high incidence of sign epistasis, such that selection associated with a substitution at one site depends on other sites
only a small fraction of all possible sequences confer high fitness, leading to slow divergence of ancient proteins
compensatory interactions between different sites in the protein structure are a well-documented phenomenon^{11–14,16–21} that is expected to result in complex, multidimensional sign epistasis and to contribute substantially to the ruggedness of protein fitness landscapes

ridges of high fitness corresponding to specific ancient proteins occupy a tiny fraction of the entire volume of the sequence space
these ridges are long and thin and can be more accurately visualized as a wide-mesh net spanning a large part of sequence space, rather than as a small volume within the space
such fitness ridges imply that sign epistasis and compensatory evolution in ancient proteins must be common
our data show that >90% of the sites in any protein can eventually accept a substitution given the right combination of amino acids at other sites
although it is not clear whether such substitutions are predominantly neutral or beneficial

regardless of the importance of positive selection in protein divergence, it seems that many sites are conserved because there has not been enough time to create the right combination of amino acids at other sites to allow them to evolve, which may take billions of years

protein fitness landscapes cannot be described outside the framework of multidimensional epistasis
which can reflect non-trivial compensatory interactions
simple forms of epistasis, such as synergistic or antagonistic epistasis, do not provide an adequate theoretical framework for understanding protein evolution