advertisement: compare things at compare-stuff.com!

Sequence conservation

The hydrophobic core residues of proteins are more conserved than non-core residues due to the evolutionary constraints of sidechain packing in the core. Conservation, therefore, is also a sequence feature which clusters in three dimensions in protein structures. The measure of conservation $g_i$, introduced in Section 4.3.3, can be mapped in the same manner as hydrophobicity in the previous section.

As mentioned previously, however, functional constraints restrict the freedom of amino acid substitutions (in the active site of a protein for example), and hence affect conservation. In Figure 5.11 we investigate the distribution of conserved residues in the structure of domain 1knb01[Bruns & Karplus, 1995] from spinach ferredoxin reductase. Figure 5.11(a) shows the active site residues (from the SITE records in the PDB file) in yellow. Figures 5.11(b) and (c) show the raw and mapped conservation measure $g_i$, respectively. Compared to the mapping of hydrophobicity shown in (d), the distribution of conserved residues is more a consequence of function than structure. Since function is not necessarily conserved between structures with similar folds, the use of sequence conservation measures as described here may not help to detect analogues. The identification of active sites[Zvelebil et al., 1987,Peters et al., 1996,Laskowski et al., 1996,Cabral et al., 1996] using this technique remains an interesting possibility.

Figure 5.11: Mapping sequence conservation and structure for domain 1knb01 (spinach ferredoxin reductase). (a) Active site residues in yellow. (b) Raw sequence conservation, $g_i$, shown by colour scale (blue=least unconserved, red=most conserved). (c) Mapped conservation, using a weighting of 1.5 for $g_i$. (d) Mapped Kyte and Doolittle hydrophobicity (see Section 5.4.1). The mapping (b) of sequence conservation has located the cluster of conserved amino acids which form the FAD (flavin-adenine dinucleotide) binding site for this enzyme; this is not so clear in the raw data in (a). The mapped hydrophobicity in (c) shows an overlapping distribution, but is less centred around the active site.