132 A.E. Kister et al.
7.3.3 Structurally Based Sequence Alignment
The essential element of our method is that it involves alignment not of whole
sequences, but of corresponding strands in their respective proteins. In con-
trast to the analysis of homology proteins in protein family, the determination
of the corresponding strands in a group of strongly varied proteins, such as
a collection of sandwich-like proteins from very diverse superfamilies with a
dissimilar arrangement of strands and variable numbers of strands, could be a
complicated problem. However, the delineation of an invariant supersecondary
substructure made it possible to identify and align the corresponding strands.
It follows from the rule of interlocked pairs of strands that four strands
i, i+1,k,andk+ 1 with similar structural properties were found in all sand-
wich proteins. Thus in our procedureistrands from all structures were aligned
with each other, then alli+ 1 strands and so forth. The alignment of strands
carried out in this way maximizes the number of positions occupied by struc-
turally similar residues. It is important to note that no “gaps” within strands
are allowed, since strands are viewed as indivisible structure units. Adjacent
residues within a strand are always assigned sequential position numbers.
However, gaps between strands are a common occurrence. The advantage of
this structurally based approach is that it makes possible a common system of
numbering for sequences from different superfamilies. It allowed us to compare
nonhomologous SP. Details of the structure-based sequence multialignment
and a list of the conserved positions was presented in our works [32, 33].
7.3.4 Sequence Characteristics of thei, i+1,k,andk+ 1 Strands
Analysis of the structurally aligned sequences revealed 12 positions, which are
occupied by residues with structurally similar properties in their respective SP
structures. We suppose that residues that have the same structural properties
across all SPS are their structural determinants. The structural determinants
lie at the center of the interface between theβ-sheets and form the common
geometrical core of SP structures.
Inspection of amino acid frequencies in these 12 positions showed that
eight positions are the conserved hydrophobic positions of SP. Residues at
these eight positions are termed the SP sequence determinants. Eighty percent
of all SP sequence determinants areV, L, I, and F residues.
7.3.5 Structural Features of the Sequence Determinants
Relative positions of the eight sequence determinants in i, i+1,k,and
k+ 1 strands are common in all SP structures that contain interlocked pairs
(Fig. 7.2). The strandsiandkare oriented toward each other in such a way
that residues at position 6 in strandialways form main chain hydrogen bonds
with residues at position 8 in strandk, and side chains of both these residues
look inside the hydrophobic interior of SP. Orientation of strandsi+1 and