Comparison of the different numbering schemes for immunoglobulin variable domains:

In their compilation of "Sequences of Proteins of Immunological Interest", of which several printed editions had appeared, Kabat et al., (1991) collected and aligned the sequences of different members of the immunoglobulin superfamily. They proposed numbering schemes for many of the different protein families, which make up the immunoglobulin superfamily. The placement of sequence gaps was based on sequence variability rather than on the spatial structure, as far fewer structures were known at the time. Chothia & Lesk (1987) corrected the positioning of CDR L1 and CDR H1 sequence length variability in the antibody variable domains to better fit their actual position in the three-dimensional structure. In 1989, Chothia and colleagues changed the insertion point in CDR L1 (Chothia et al., 1989), but returned to the old definition in 1997 (Al-Lazikani et al., 1997). The different families of immunoglobulin domains are treated separately by both Kabat and Chothia despite the high degree of sequence and structural homology. Gelfand and colleagues (Gelfand et al., 1998a; Gelfand et al., 1998a; Gelfand et al., 1996) studied in detail the structurally invariant core of antibody VL and VH domains, but they basically restricted their analysis to a set of residues which is almost identical to the core residues we use for least-squares superpositions and ignored the less conserved positions. They identified the residues by their secondary structure position, resulting in a complex nomenclature indicating the b-strand in which a residue is located and its position in this strand. Lefranc and colleagues (IMGT, Lefranc et al., 1999, Ruiz et al., 2000) proposed a unified numbering scheme for immunoglobulin variable domain germ line sequences (Lefranc, 1997) including antibody lambda and kappa light and heavy chain variable domains as well as T-cell receptor alpha, beta, gamma and delta chain variable domains. However, since they deal only with germ-line sequences, their numbering scheme only reaches into CDR 3 and does not address the residues in CDR 3 and framework 4. An important difference to the numbering scheme presented here (AHo) is that in the IMGT scheme insertions and deletions "grow" unidirectionally, as in the original Chothia definition (Chothia & Lesk, 1987), while in the AHo scheme, insertions and deletions are placed symmetrically around the key position marked in yellow. Furthermore, length variations in CDR 1 and CDR 2 are represented by a single gap in IMGT and by one or two gaps in AHo.

AHonegger, June 6, 2001