We read with interest the report by Zhu et al,1 who examined the generation of potential N-glycosylation sites in the clonogenic immunoglobulin (Ig) heavy chain (VH) genes of patients with follicular lymphoma (FL).1 Using their own and a number of sequences derived from the published literature, these authors have investigated the frequency of potential N-glycosylation sites introduced into functional VH genes as a consequence of somatic mutation. FL cells were compared with normal memory B cells or plasma cells matched for similar levels of mutation. Strikingly, novel sites were detected in almost 80% of patients with FL, compared with 9% in the normal B-cell population (P < .001), whereas diffuse large B-cell lymphoma (DLCL) showed an intermediate frequency (41%). Multiple myeloma (MM) and the mutated subset of B-cell chronic lymphocytic leukemia (B-CLL) showed frequencies similar to those of normal cells in 8% and 13% of patients, respectively. The authors concluded that N-glycosylation of the VH region may be a common event in FL and in a subset of DLCL, with the novel sites predominantly located in complementarity-determining regions (CDRs), the major target of the somatic hypermutation machinery. Sequences of nonfunctional VH genes contained few sites, arguing for positive selection for novel glycosylation sites in functional FL VH sequences. Moreover, the authors suggested that the added carbohydrate in the VH region might contribute to interaction with antigenic elements in the germinal center (GC) environment, and that this common feature of FL may be critical for tumor behavior.
We conducted a similar analysis for novel glycosylation sites in our published series of VH and Vκ sequences from FL2 and MM patients.3 Novel glycosylation sites were identified in 4 out of 10 FL VH genes; there were 2 novel sites in the CDR2 and one each in the FWR1 and FWR2 regions. In MM Vκ and Vλ genes, novel glycosylation sites were detected in only 2 out of 17 cases (11.7%); both sites were located in Vκ CDR1 region. Our findings are generally in accordance with those by Zhu et al1regarding MM. However, markedly different results were obtained from analysis of FL Vκ genes; Zhu et al reported an overall incidence of almost 59% (10/17 cases; Table 1 of their article), whereas no novel glycosylation sites were identified in our published sequences.
FL can be considered as the neoplastic counterpart of a certain stage in the intra-germinal center maturation of B cells.4 At this stage, normal B cells that are specific for antigen (with surface Ig [sIg] exhibiting an affinity above the germ line potential) are subject to positive selection.5 Sequence analysis of rearranged VH and Vκ genes of FL cells carried out by our group demonstrated that whereas rearranged VH genes were hypermutated in their majority, with clustering of mutations in the CDRs, the rearranged Vκgenes of the same neoplastic clones varied significantly with regard to their mutational profile; indeed, very few or no mutations were observed in a significant number of cases (5/10; 50%) of the analyzed Vκ sequences.3 A similar pattern is obtained from the study by Zhu et al;1 although the Vκ genes of that study were, on average, more heavily mutated than the corresponding genes in our study, the median “mutation load” in Vκ sequences was still significantly smaller than in the same cohort's VH genes (median homology to germ line was 89% for VH genes versus 96% for Vκ genes). The fact that clonogenic Vκ genes are less frequently mutated indicates that the somatic hypermutation machinery might have ceased to operate in the Vκ locus at the time when neoplastic transformation had occurred. Perhaps this is a reflection of the generally smaller potential contribution of Vκ genes (compared with VH genes) in antigen selection of the clonogenic B cells in FL.3 Similar conclusions regarding the role of Ig light chains in antigen recognition (evidenced by the mutation status of the corresponding V region genes) have been reached from single-cell studies in the normal peripheral B-cell repertoire, indicating a more limited mutational load both in the expressed as well as nonfunctional Vκ genes compared with their partner VH genes in IgM+/CD5− B cells.6
In conclusion, the study of Zhu et al,1 as well as analysis of our previously published series, refocuses the interest of antigen selection in human lymphomagenesis to the potential contribution of Ig V region glycosylation at novel sites as a result of active somatic hypermutation. However, until these results are accepted as definite, further confirmatory studies in larger numbers of patients are awaited.
High incidence of novel N-glycosylation sites in the immunoglobulin variable region genes of follicular lymphoma
Our finding of a high frequency of potential glycosylation sites in the variable regions of cases of follicular lymphoma (FL) has implications for the understanding of the pathogenesis of this tumor.1-1 The presence of oligosaccharides in the binding site of the B-cell receptor of FL cells may influence interactions with stromal elements in the germinal center and could contribute to tumor growth or survival. Clearly the data on 70 cases in our paper need to be extended, and Belessi et al also make this point. The sites are largely acquired by somatic mutation and can be found in either VH or VL. One problem in extending the data is that generally only VH sequences are available from the databases, leaving open the question of the contribution of sites in VL. Our paper included matched VH and VL sequences from 17 cases from this laboratory, and we have now added a further 11 matched cases, all of which had sites. In summary, we now have 26 of 28 cases with sites in VH and 14 of 28 with sites in VL, and all our cases had at least one site. An interesting exception was a case initially diagnosed as FL, but with no sites evident. We asked a pathologist (Professor Dennis Wright) to give a second opinion, and he made the diagnosis of mantle cell lymphoma, subsequently confirmed by staining for cyclin D1. While this does not yet allow us to conclude that all cases of FL will have sites, it does indicate that diagnostic criteria must be met. Disturbingly, Belessi et al have added 10 cases, and only 4 of these have sites, all of which were in VH. While this is a small number, it does not fit with our observations, and we await further findings from our cases and from the community.
Belessi et al then focus on events occurring in VK, where they have sequence data on 10 cases. They make the point that levels of somatic mutation are lower in VL than VH, an observation that we confirm in our set of sequences, which had 89% and 96% homology for VH and VK, respectively.1-1 This discrepancy has been reported in normal B cells and in other B-cell malignancies, eg, in Burkitt lymphoma.1-2 In relation to acquisition of glycosylation sites, it is obvious that they will be less likely to accumulate in VL, and, in fact, 5 of 10 of the cases analyzed by Belessi et al were close to germ line sequence. However, a low mutational rate may be sufficient since, in one of our cases with 99% homology in VK, a site had been acquired.1-1 Belessi et al speculate on the reason for the differential mutational rate in VH and VL, and they suggest that “the somatic hypermutation machinery might have ceased to operate in the VK locus at the time when neoplastic transformation had occurred.” In our view this is unlikely, given that intraclonal variation continues to accumulate in VK sequences in FL (Zhu et al1-3 and our unpublished observations, May 2002). An alternative explanation for the discrepancy in mutational level may be a lower frequency of sequence motifs that surround mutational hot spots,1-4 or a difference in the elements outside the coding region that influence somatic mutation.1-5 The role of germ line VL sequence in antigen recognition has been assumed to be less than that of VH, due to the lack of D-segment genes. The relatively low level of somatic mutation would support this lesser role. However, crystallographic analysis clearly implicates VL in antigen recognition,1-6 and the degree of involvement is likely to vary for different antibodies. Generalizations about the immensely variable structures of the binding site of antibodies may be difficult, but the presence of oligosaccharides, largely avoided by normal B cells, could be telling us something about B-cell tumors. Since there may be consequences for new treatment options, we await data from more matched sequences of FL with considerable interest.
This feature is available to Subscribers Only
Sign In or Create an Account Close Modal