Modeling Structural and Genomic Constraints in the Evolution of Proteins - Computational Modeling of Biological Systems - page 339

sequence identity 0:15, where structure divergence explodes. Interestingly, the

explosion of structural divergence seems to take place only for proteins performing
different functions, whereas proteins that share exactly the same function can
diverge in structure only up to a limiting value of divergence, so that functional
conservation imposes strong constraints on sequence and structure [ 71 ].
The simplest explanation for the explosion of structural divergence is that, below
the crossover, sequence identity does not allow for estimation of the evolutionary
divergence time, so that protein pairs with identity below the crossover may have
diverged for a time much longer than what is inferred from their sequence identity.
This simple explanation is supported by the fact that the sequence identity at
the crossover decreases with protein length. Nevertheless, it is interesting that a
qualitatively similar explosion of structural diversity has been found in a recent
study of protein sequence design [ 72 ]. In this study, protein sequences were
designed by optimizing the folding stability of a target structure. It was found that,
when the target structure and the reference structure in the PDB are very similar, the
designed sequence has a rather large identity with the reference sequence. However,
when the target and the reference structure are more different, as it would be in
the case of selection for new function, the designed and reference sequences only
share very low identity, on the order of 20%, i.e., slightly more than the average
identity of unrelated protein pairs. Therefore, the plot of sequence divergence versus
structure divergence of designed protiens shows a crossover very much reminiscent
of the one that we observed for evolved proteins and it may help to rationalize it:
When two proteins perform the same function, natural selection targets very similar
structures, determining sequence and structure conservation, whereas for proteins
with significantly different function, natural selection targets different structures,
whose typical sequence identities are below the crossover region. This interpretation
is consistent with the findings, reported above, that protein function influences
evolution by limiting the extent of sequence and structure divergence in the case
of function conservation, and by accelerating structure divergence with respect to
sequence divergence in the case of function change.
We also found that structure evolution is accelerated upon function change, since
protein pairs with different functions diverge in structure at a rate significantly larger
than those with the same function even before the explosion of structural divergence.
Although not unexpected, this is an interesting result, since it demonstrates a
quantitative influence of protein function on the sequence to structure relationship.
Moreover, it suggests possible improvements to protein function prediction. In fact,
it is known that very small changes in sequence and structure are sufficient to modify
protein function, so that sequence and structure conservation are not a sufficient
indication of function conservation. Our observation that function change modifies
quantitatively the sequence to structure relationship suggests that this information
could be used in order to predict function conservation more reliably.
Besides the quantification of the rate of structure divergence, another interesting
result concerning protein structure evolution was the observation that protein
structures tend to diverge along directions that overlap with the normal modes of
low frequency [ 73 ]. This observation has been subsequently rationalized using one