Correspondence: Dr MC Cook, Department of Immunology, The Canberra Hospital, Australian National University Medical School, Level 3 Building 10, PO Box 11, Woden, Canberra, Australian Capital Territory 2606, Australia. E-mail: Matthew.Cook@anu.edu.au

Many B-cell malignancies are classified by correlating the neoplastic population with possible counterparts that appear during normal B-cell ontogeny.¹ Prognosis is often related to whether the neoplastic transformation takes place before or after a B cell has traversed a germinal centre (GC). While cell surface markers provide important clues, unequivocal evidence of GC or post-GC origin (and antigen experience) depends on analysis of the immunoglobulin genes for the presence or absence of somatic mutations. In this issue of Immunology and Cell Biology, Collins and co-workers² contest the accuracy of the principal repository of reference germline immunoglobulin sequences.

Antibodies are available in advance for every conceivable antigen, even those not devised by nature. The extent of antibody diversity represented a conundrum for which a solution emerged from analysis of paraproteins obtained from patients with multiple myeloma. Dreyer and Bennett³ observed that diversity was confined to one end of the immunoglobulin light chain, and speculated 'a kind of genetic scrambling', a solution that represented a radical violation of the one gene–one polypeptide rule. Specification of complete immunoglobulin genes does not reside in the germline, but only after stochastic recombination takes place early in the ontogeny of every B cell. 'Genetic scrambling' by V(D)J recombination of antigen receptor genes represents the defining innovation of adaptive immunity. Tremendous diversification is feasible because of the numerous alleles available at each locus, and is extended by more than one open reading frame for D genes, splicing infidelity, and N-region diversity.

Depending on the haplotype, there are between 170 and 176 IGH genes per haploid human genome. This includes 123–129 IGHV gene segments located within 957 kb at the telomeric extremity of the long arm of chromosome 14 (14q32.33),⁴ of which 40 are structurally functional. Phylogenetic analysis has assigned 82–88 IGHV genes to three main clusters and seven families or subgroups. All reported alleles within the IGHV locus have been retrieved from the EMBL nucleotide sequence database and manually curated into the ImMunoGeneTics (IMGT) information system database, which can be accessed via http://imgt.cines.fr/. Here, nomenclature of immunoglobulin genes identifies the group (heavy or light chain, V, D, J or C) and subgroup to which the gene belongs, the gene segment within the subgroup and the allele.⁵ For example, IGHV3-15^*02 identifies the first allelic variant (relative to ^*01) of the IGHV gene, which is a member of subgroup 3.

The immunoglobulin repertoire is also diversified by the introduction of point mutations (somatic hypermutation, SHM), which occurs after V(D)J rearrangement.⁶ In birds, this takes place in the Bursa of Fabricius, and in some mammals, in specialized gut-associated lymphoid tissue, whereas in other mammals, SHM is mainly confined to secondary lymphoid organs such as lymph nodes and spleen. The nucleotide targets of the SHM machinery are enriched in regions defining residues that make contact with antigen.⁷ These motifs appear to have coevolved with the mutator apparatus such that they are unusually susceptible to mutation.⁸ Whether SHM is deployed for diversification of primary or secondary repertoires varies in different species. Humans exemplify how SHM appears to have been co-opted for diversifying the postantigenic repertoire, where mutations introduced in proliferating centroblasts provide the variation on which selection is imposed to cause affinity maturation.

Germline variation in regions encompassing immunoglobulin V genes has been subject to relatively frequent reorganization. Since the last common ancestor of mice and humans (75mya), there appears to have been seven duplication events in the chromosomal region encompassing the immunoglobulin heavy-chain variable segments.⁴ In addition, numerous polymorphisms have been identified within each segment. Germline variants are subject to selection at the level of the individual, and any individual carries a maximum of two allelic variants at any locus. By contrast, for SHM, the unit of selection is the B cell, and numerous allelic variants emerge within 7–10 days of immunization.⁹

Distinguishing IGHV allelic variants is problematic, first because of generic technical problems that arise in identifying any polymorphisms, such as errors introduced during PCR, sequencing or cloning. In addition, contamination of non-lymphoid tissue with B cells could result in false identification of somatic mutants as germline polymorphisms. The criteria for inclusion of a new allele in the IMGT database are stipulated to guard against such errors, but Wang and co-workers have provided an analysis of the catalogue of allelic variants, and conclude that the diversity of germline alleles might have been overestimated. The case is substantiated by examples where more than two allelic variants have been obtained from one individual.

The accuracy of the IMGT database is of immediate relevance to laboratories concerned with determining the prognosis of B-cell malignancy. B-cell lymphomas and chronic lymphocytic leukaemia are heterogeneous with regard to therapeutic response and prognosis. Neoplastic B cells can be assessed by surface phenotype, but definitive evidence that distinguishes pre-GC from GC/post-GC B cells depends on the presence or absence of somatic mutations. Based on the analysis provided by Collins and co-workers, some germline allelic variants of IGHV genes have been contaminated with either erroneous sequences or somatic variants, and they suggest that the IGMT database requires revision. These findings are unlikely to alter the established prognostic significance of presence or absence of SHM in B-cell malignancies. Nevertheless, the findings suggest that further curatorial attention is warranted, since the IGMT is considered the definitive reference list of germline variants.