The Kabat Numbering Scheme
The Kabat numbering scheme is a widely adopted standard for numbering the residues in an antibody in a consistent manner. However the scheme has problems! First, since the numbering scheme was developed from sequence data (a fairly limited set), the position at which insertions occur in CDR-L1 and CDR-H1 does not match the structural insertion position. Thus topologically equivalent residues in these loops do not get the same number.
Second, the numbering adopts a rigid specification. For example in the potentially very long CDR-H3, insertions are numbered between residue H100 and H101 with letters up to K (i.e. H100, H100A ... H100K, H101). If there are more residues than that, there is no standard way of numbering them. Such situations occur at other positions too.
The Chothia Numbering Scheme
The Chothia numbering scheme is identical to the Kabat scheme, but places the insertions in CDR-L1 and CDR-H1 at the structurally correct positions. This means that topologically equivalent residues in these loops do get the same label (unlike the Kabat scheme).
There are two disadvantages: first, the Kabat scheme is so widely used that some confusion can arise; second, Chothia et al. changed their numbering scheme as of their 1989 Nature paper such that insertions in CDR-L1 are placed after residue L31 rather than L30. Examining the conformations of the loops shows that L30 is the correct position.
Note That in their latest paper (Al-Lazikani et al., (1997) JMB 273,927-948), Chothia's group returns to using residue L30 as the insertion site in CDR-L1!
Martin (Enhanced Chothia) Numbering Scheme
The only differences between the Chothia and Kabat numbering schemes are in the sites of indels in CDR-L1 and CDR-H1.
Our 'Martin' (Enhanced Chothia) scheme also considers the structurally correct locations for indels in the framework regions. Thus the numbering scheme is identical to the Chothia in most regards but positions of framework indels have been refined.
The most important of these is the insertion which present in the majority of antibodies at H82a,b,c has been moved to the structurally correct location of H72a,b,c.
We have also introduced an indel site at L52 in CDR-L2. All structures have the standard length of 7 residues and the conformations are relatively conserved. However sequences of varying lengthare seen and analysis of the structure suggests this is the correct location. It also corresponds with the AHo numbering scheme.
How to identify the CDRs by looking at a sequence
The following set of rules will allow you to find the (Kabat or Chothia) CDRs in an antibody sequence. Note that the word 'always' should always be treated with care! There are rare examples where these virtually constant features do not occur (for example the human heavy chain sequence EU does not have Trp-Gly after CDR-H3). The Cys residues are the best conserved feature.
CDR-L1
Start Approx residue 24
Residue before always a Cys
Residue after always a Trp. Typically Trp-Tyr-Gln, but also, Trp-Leu-Gln, Trp-Phe-Gln, Trp-Tyr-Leu
Length 10 to 17 residues
CDR-L2
Start always 16 residues after the end of L1
Residues before generally Ile-Tyr, but also, Val-Tyr, Ile-Lys, Ile-Phe
Length always 7 residues (except NEW (7FAB) which has a deletion in this region)
CDR-L3
Start always 33 residues after end of L2 (except NEW (7FAB) which has the deletion at the end of CDR-L2)
Residue before always Cys
Residues after always Phe-Gly-XXX-Gly
Length 7 to 11 residues
CDR-H2
Start always 15 residues after the end of Kabat / AbM definition) of CDR-H1 Residues before typically Leu-Glu-Trp-Ile-Gly, but a number of variations
Residues after Lys/Arg-Leu/Ile/Val/Phe/Thr/Ala-Thr/Ser/Ile/Ala
Length Kabat definition 16 to 19 residues;AbM (and recent Chothia) definition ends 7 residues earlier
CDR-H3
Start always 33 residues (sometimes 30 residues) after the end of CDR-H2 (always 3 after a Cys)
Residues before always Cys-XXX-XXX (typically Cys-Ala-Arg)
Residues after always Trp-Gly-XXX-Gly
Length 3 to 25(!) residues - and sometimes much longer, particularly in bovine antibodies
