We recently demonstrated that >30% of the cases with splenic marginal-zone lymphoma (SMZL) express B-cell receptors (BcRs) that carry a single polymorphic variant of the IGHV1-2 gene (IGHV1-2*04).1 The IGHV1-2*04 BcRs in SMZL are characterized by long complementarity-determining region-3 (CDR3) and exhibit a low impact of somatic hypermutation (SHM), leading to limited changes, which cluster in certain positions of the VH domain. Altogether, these features argue for selection by antigen in the pathogenesis of SMZL and suggest heavy chain (HC) dominance in the clonogenic immunoglobulin (IG) receptors of SMZL.1
That notwithstanding, the possibility that IG light chains (IG LC) might also have an important role in SMZL ontogeny exists and needs to be investigated. This is supported by studies demonstrating that IG LCs may be critically implicated in recognition of and selection by antigen. Examples include: (i) significant LC sequence restrictions among estradiol-specific antibodies;2 (ii) impaired responses to Haemophilus influenzae in individuals lacking a specific allelic variant of the IGKV2D-29 gene; and, (iii) biased IG LC gene usage in autoimmunity (for example, IGKV1-17 and IGLV1-47 in lupus,3, 4 IGKV1-39/1D-39 in Graves disease5). Furthermore, the BcR specificity may drastically change through genetic processes affecting LCs, including secondary rearrangements in the context of receptor editing6 or distinctive SHM patterns.7 Finally, marked LC gene repertoire biases exist in various B-cell lymphomas, strongly indicating that LCs can contribute significantly toward shaping antigen reactivity of the clonotypic BcRs.8
In order to gain insight into the role of IG LCs in SMZL we explored the IG LC gene repertoire in 107 SMZL cases diagnosed as described previously1 and carrying at least one productive LC rearrangement. The analysis was performed on spleen specimens or peripheral blood samples as described previously.8, 9, 10 Sequences were analyzed using the IMGT databases and the IMGT/V-QUEST tool (version 3.2.17, Université Montpellier 2, CNRS, Montpellier, France; www.imgt.org). All sequences, including 41 sequences previously reported from our groups,9, 10 were evaluated following the novel strategy detailed in our recent publications.1, 8
Overall, 113 productive IG LC rearrangements were amplified from 107 cases. Six cases carried two productive IG LC rearrangements along with a single productive HC rearrangement, effectively ruling out the possibility of biclonal population. Notably, four of the six cases with double-productive rearrangements were found to carry one each IGKV-IGKJ and IGLV-IGLJ rearrangement, hence raising the possibility that an initially expressed IGKV-IGKJ rearrangement was subsequently edited through a process involving inactivation of the IGK locus by a rearrangement involving the kappa-deleting element, thus, ensuring allelic exclusion.6, 8
The IG LC gene repertoire was biased (Supplementary Table 1) with only six genes (IGKV3-20, IGKV4-1, IGKV1-5, IGKV1-8, IGKV1-39, IGLV2-14), accounting for 72/113 rearrangements (64%). In particular, the IGKV3-20 and IGLV2-14 genes predominated by far within IGKV-IGKJ and IGLV-IGLJ rearrangements (20/82 cases, 24%; and 10/31 cases, 32%, respectively). Hence, the IG gene repertoire in SMZL is restricted with regards to not only IG HCs but LCs as well, strongly indicating selection by antigen.
On the basis of the percentage of IGKV/IGLV gene identity to the germline (GI), 21/113 sequences (18%) were assigned to a ‘truly unmutated’ subgroup (100% GI), whereas the remaining sequences (92/113, 82%) exhibited a variable impact of SHM, ranging from minimal to pronounced (Supplementary Table 2). For statistical comparisons, sequences with 97–99.9% GI were classified as ‘borderline/minimally mutated’ (n=61, 54%), whereas those with <97% GI as ‘significantly mutated’ (n=31, 28%). Rearrangements of most genes did not differ regarding the impact of SHM. The IGKV1-5 gene constituted an exception as it was predominantly utilized in ‘significantly mutated’ rearrangements (P=0.01 for comparison with other IGKV genes; Supplementary Figure 1). This is noteworthy as it has not been seen in other B-cell malignancies, for example, chronic lymphocytic leukemia.8 No statistical differences were observed regarding SHM load between IGKV-IGKJ or IGLV-IGLJ rearrangements. Hence, overall, the distribution of IG LC sequences with regards to SHM status recapitulated what we observed in SMZL HCs.1
Assessment of paired heavy and light variable domains (VH and VK/VL) with regards to SHM status in 100 of the 107 cases with available data revealed two general patterns (Supplementary Table 3): (i) both VH and VK/VL with a similar impact of SHM: 61 cases (6 cases with truly unmutated VH+VK/VL, 30 cases with borderline/minimally mutated VH+VK/VL, 25 cases with significantly mutated VH+VK/VL); (ii) VH and VK/VL with discordant mutational status, that is assigned to a different SHM subgroup; in this latter category, 33 cases carried more heavily mutated VH versus VK/VL domains, whereas the opposite was observed in the remaining 6 cases, which carried more heavily mutated VK/VL versus VH domains.
Subgroups of cases employing the six most frequent IGKV or IGLV genes (IGKV3-20, IGKV4-1, IGKV1-5, IGKV1-8, IGKV1-39, IGLV2-14) were studied for the existence of recurrent (shared) amino-acid (AA) replacements introduced by SHM (Supplementary Table 4). Recurrent replacements were identified especially among cases utilizing the IGKV3-20, IGKV1-5, IGKV1-39 and IGLV2-14 genes. The most notable change was identified in IGLV2-14 rearrangements with codon CDR2-56 standing out, in that 5/10 mutated IGLV2-14 sequences (50%) carried an identical, conservative substitution of an acidic for an acidic residue (aspartic acid for glutamic acid; E-to-D) (Figure 1) due to a G-to-T substitution at the third position of codon CDR2-56. The possibility that this substitution represents an as yet unidentified polymorphism cannot be a priori excluded; that notwithstanding, it is noteworthy that recurrent AA changes were identified in other positions of the IGLV2-14 rearrangements (Supplementary Table 4), which strengthens the argument for selection of certain changes leading to distinctive physicochemical attributes.
Among mutated IGKV-IGKJ gene rearrangements (64 cases) the most frequent ‘hotspot’ for change was codon VK CDR1-37, where 28/64 (44%) cases exhibited AA substitutions. In 14 such cases, utilizing different IGKV genes, this change concerned a substitution of serine (S) by asparagine (N). Given that N is the germline-encoded residue for this position in seven other IGKV genes used in this cohort, it could be argued that in this particular case the SHM process induced sequence convergence rather than divergence; similar examples were observed in other VK codons as well. Hence, germline-encoded differences in SMZL rearrangements utilizing different IGKV genes can be evened out by SHM, alluding to functional selection.
Analysis of the CDR3 was possible in all 82 IGKV-IGKJ and 31 IGLV-IGLJ rearrangements. The molecular characteristics of the VK or VL CDR3 were generally comparable to normal PB IgM+ B cells with regards to length, number of N-nucleotides, and 5′ and 3′ exonuclease activity at the V and J genes, respectively. Among IGKV-IGKJ rearrangements, the IGKJ1 and IGKJ2 genes collectively accounted for 53/82 cases; the remaining cases (29/82) utilized distal IGKJ genes (IGKJ3-5); among IGLV-IGLJ rearrangements, the IGLJ2 and IGLJ3 genes predominated (14 and 9 cases, respectively) (Supplementary Table 5).
Focusing on the subgroup of 20 IGHV1-2*04 cases with available IG LC sequence information (22 productive rearrangements), we obtained evidence for biased usage of three LC genes, namely IGKV3-20 (8/20 cases, 40%), IGKV1-8 (5/20, 25%) and IGLV2-14 (3/20 cases, 15%) (Figure 2). In all these pairs, as for the HC (IGHV1-2*04), also the LC genes (IGKV3-20, IGKV1-8, IGLV2-14) were ‘borderline/minimally mutated’. Interestingly, in the case of IGKV3-20 sequences, recurrent somatic mutations were identified, with codons CDR1-37 and FR3-66 emerging as ‘hotspots’ for recurrent, conservative AA changes.
According to the WHO 2008 classification, the postulated normal counterpart of SMZL is a B cell of ‘unknown differentiation stage’.11 That notwithstanding, immunogenetic studies by us and others,1, 12 including the present one, strongly implicate interactions with antigen in the immune pathway(s) leading to SMZL. The precise timing and microenvironmental location of such interactions as well as the nature of the selecting antigens are currently largely unknown. Interestingly, a recent study demonstrated that IGHV1-2*04 recombinant antibodies from SMZL patients are poly- and self-reactive,13 reminiscent of natural antibodies.14 In humans, a major source of natural antibodies are splenic marginal-zone (SMZ) B cells,15 most of which carry somatic mutations in their BcR,15 leading to the assumption that they are memory B cells that developed during a T-dependent antigen interaction inside a germinal center (GC). However, the presence of cells with an identical immunophenotype to SMZ B cells and mutated BcRs in CD40L-deficient patients that could not assemble classical GCs has prompted speculations that SMZ cells may diversify and acquire somatic mutations also in a GC-independent fashion.15
The present results further support the argument that antigen selection could be implicated in SMZL immunopathogenesis. In fact, we demonstrate that the immunogenetic profile of SMZL is distinctive also with regards to IG LCs, as evidenced by: (i) restrictions in the IGKV and IGLV gene repertoire; (ii) biased associations of the IGHV1-2*04 with the IGKV3-20, IGKV1-8 and IGLV2-14 genes; (iii) the existence of recurrent mutations in rearrangements of the predominant IGKV and IGLV genes; (iv) sequence convergence induced by SHM. Overall, these findings may reflect an antigen-driven immune pathway to lymphoma development and support our recent claims for the existence of distinct SMZL subtypes defined by immunogenetic analysis.
This work was supported in part by the ENosAI project (code 09SYN-13-880) co-funded by the EU and the Hellenic General Secretariat for Research and Technology to ND and KS; and the Cariplo Foundation (Milan, Italy) to PG and KS.
About this article
Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)
Philosophical Transactions of the Royal Society B: Biological Sciences (2015)