advertisement: compare things at compare-stuff.com! |
Since the main feature of most globular protein domains is the hydrophobic
core, the distributions of hydrophobic amino acids along the sequences of
remote homologues and analogues ought to be similar. But is this pattern
recognition already achieved by standard alignment techniques using
log-odds matrices derived from observed substitutions in sequence
alignments[Dayhoff et al.,
1978, for example]? Most of these matrices
cluster polar and apolar amino acids separately. Analyses of conservation
in multiple sequence alignments[Han & Baker, 1995,Fiser et al.,
1996,Ladunga & Smith, 1997] have
also identified the amino acids which commonly substitute with each other
and rationalise them in terms of structure and/or amino acid properties.
Han and Bakerhan:sequence found that the most common clusters
of single column amino acid profiles were predominantly hydrophobic or
polar in nature. Consecutive segments of up to 13 residues were also often
best explained in terms of hydrophobicity. Fiser et
al.fiser:conservation looked at the conservation of
amino acids with respect to structure and concluded that hydrophobicity was
well conserved in the core. Ladunga and
Smithladunga:substitution employed a robust all-or-nothing
binary profile analysis on a number of large sequence databases and also
reported common substitutions between hydrophobic or polar residues.
There is now sufficient data to allow the generation of substitution matrices from structural alignments. Johnson and Overingtonjohnson:jmb93 and Russell et al.russell:jmb97 performed this analysis with homologues of low sequence identity and found a similar clustering of polar and apolar amino acids to the Dayhoff matrix. Using a set of analogous pairs (same fold, different SCOP superfamily), Russell et al.russell:jmb97 obtained a matrix which appeared to exhibit a broader split between polar and apolar characteristics, with less insistence on the maintenance of `finer' properties such as size and charge. These studies suggest that at greater evolutionary distances, hydrophobicity is the most conserved amino acid property.
Since hydrophobic residues are generally buried, patterns of hydrophobicity should roughly mirror those of solvent-accessibility. Russell et al.russell:analysis1,russell:jmb97 found, however, that only around 60% of residues at structurally equivalent positions in pairs of remote homologues and analogues had similar solvent accessibility. Predicted solvent accessibility (which depends mostly on the hydrophobicity of the predicted residues, utilising little short-range information) was slightly worse. These results indicate that the pattern of hydrophobicity is not completely conserved between distantly related structures.