Myeloma V_L and V_H Gene Sequences Reveal a Complementary Imprint of Antigen Selection in Tumor Cells

MULTIPLE MYELOMA is a malignant tumor involving plasma cells. Neoplastic cells are found in the bone marrow (BM), and typically secrete a monoclonal Ig of IgG or IgA isotype. Although the major identifiable tumor population consists of plasma cells, there has been a great deal of debate concerning the nature of the malignant cell, with some early indications that this may be a less mature B cell capable of “feeding” the plasma cell compartment.1,2 The advent of genetic technology aimed at Ig genes has allowed a more incisive investigation of the characteristics of myeloma clones. In fact, there have now been reports of a total of more than 50 sequences of V_H genes used by tumor cells from patients' BM biopsies, and these have revealed common features.3-5 

One conclusion is that usage of V_H genes from the available repertoire appears to reflect no striking bias, with predominance of the large V_H3 family in line with serological analysis of myeloma proteins.6 However, at the level of individual V_H genes there may be some asymmetry in usage. For example, one gene, V_4-34 , commonly used by normal B cells, and mandatory for encoding IgM autoanti–red blood cell antibodies of I/i specificity in patients with cold agglutinin disease,7 has so far not been found to be used by tumor cells in myeloma.5,8 In all cases of myeloma, the V_H genes have been found to be somatically hypermutated.3-5 A further common feature is the lack of intraclonal variation in sequence, a finding that contrasts with the heterogeneity found in B-cell tumors of the germinal center.9,10 This leads to the conclusion that the final tumorigenic event in myeloma has occurred at a postfollicular stage, when the cell is no longer influenced by the somatic hypermutation mechanism.11 It argues against the concept that there is a “feeder” B cell, unless that cell has escaped the mutator before isotype switching.

However, IgM⁺ B cells with V_H sequences indicating a clonal relationship with the neoplastic plasma cells have been detected in some cases of myeloma.12,13 Although there is some controversy about their frequency,14 it appears that such cells do exist and presumably continue to proliferate. There is uncertainty as to their contribution to malignancy, and it is possible that these cells have undergone some, but not all, of the events leading to malignant behavior.4 

A further problem in understanding development of myeloma lies in the role of antigen in selecting the V_H sequences of the tumor cell. If the cell of origin has been through somatic hypermutation, and antigen selection, before neoplastic transformation, this experience should be reflected in the V-gene sequences. For V_H regions of antibody molecules, it is known that recognition of antigen can involve several sites, with CDR3 having a major influence.15 However, replacement mutations in CDR1 and CDR2 have a significant role in affinity maturation.11,16 For B-cell tumors, where the putative antigen is generally unknown, it is difficult to estimate involvement of CDR3 in recognition. In contrast, the possible clustering of replacement mutations in CDR1 and CDR2 which could be involved in affinity maturation can be analyzed. Rules to assess the significance of apparent clustering of replacement mutations compared with silent mutations have been developed.17 When these rules were applied to the large panel of V_H sequences from myeloma cells, only a minority of cases (10 of 52) showed statistically significant clustering in CDRs.5 However, the antigen-binding site is known to involve both V_H and V_L ,18 and we have investigated V_L sequences from a group of 15 patients both to extend our knowledge of Vκ and Vλ gene usage in myeloma, and to assess the role of V_L in the selection of the cell of origin by antigen.

Patients and cell preparation.Heparinized BM aspirates from unselected patients with multiple myeloma at different stages of disease from the Hematology (UK) or Immunology (Germany) clinics were taken for investigation. All patient material was obtained with consent, and with permission from local Ethical Committees. Clinical and laboratory features are shown in Table 1. All patients had an identifiable monoclonal Ig in serum (13 IgG, 2 IgA) of the same light chain type (9κ, 6λ) as the major plasma cell population. Mononuclear cells (MNC) were isolated by centrifugation on Ficoll-Hypaque (Pharmacia, Uppsala, Sweden), and plasma cell involvement was assessed by direct immunofluorescent staining for surface CD38 and cytoplasmic κ or λ light chains using the FACS-SCAN (Becton Dickinson, CA).19 In some cases, cytocentrifuged MNC preparations were stained with the same reagents, and assessed by fluorescent microscopy.

Analysis of V_L genes.For preparation of cDNA, total RNA (2 to 10 μg) was isolated from the MNC fraction (1 to 6 × 10⁶ cells) using RNAzol B (Cinna Biotecx Labs Inc, Houston, TX). Reverse transcription was carried out with an oligo dT primer, using a first-strand cDNA synthesis kit (Pharmacia). A sample of the cDNA (1/3 to 1/5) was then amplified by polymerase chain reaction (PCR) using a mixture of 5′ oligonucleotide FWR1 primers specific for the expressed Vκ or Vλ families together with a mixture of downstream 3′ primers specific for Jκ or Jλ genes as appropriate (Table 2). Amplification conditions were as described,4,20 except that annealing temperature was 65°C for 1 minute. Amplified products were cloned and sequenced as described⁴; alignment was made to current EMBL/GenBank and V-BASE sequence directories21 using MacVector 4.0 sequence analysis software (International Biotechnologies Inc, New Haven, CT). At least two independent PCR amplifications were performed from each sample.

V_L gene usage by tumor cells.The V_L genes used by the tumor cells were identified as repeated identical V_L-J_L sequences obtained after cloning of PCR products.4 Remaining sequences, presumably from normal B cells in the aspirates,²⁰ were different from each other. Repeated sequences were seen in all cases, at variable frequency, as indicated in Tables 3 and 4 for Vκ and Vλ, respectively. The profile of Vκ genes used by the 9 κ-positive tumors indicates that 5 of 9 use Vκ1 and 4 of 9 VκIII frequencies in line with normal B cells.22,23 Among the Vκ1 group, 4 of 5 use the O8/18 gene, which is commonly rearranged in B-cell tumors,24 and 3 of 4 of the VκIII group use the A27 gene, found to be used frequently in chronic lymphocytic leukemia.25 The Vλ genes (Table 4) used three different families. There appeared to be no preferential use of particular J_L genes for either light chain type.

V-J joining region.Clonality of tumor-derived sequences was confirmed by analysis of the V-J junction (Fig 1), which showed intraclonal identity. In 9 of 15 sequences, there were base additions at the junction which were not encoded by the V or J genes. In some cases, these appeared to be derived from flanking regions of the genes, and could therefore be accounted for by imprecision at the joint. In 5 of 15, there were additional nucleotides which may represent N-region additions, contributed by TdT activity. In a majority (11 of 15) of cases, nucleotides had been lost by trimming from either V or J genes.

Somatic mutation.Nucleotide sequences of all V_L genes have been submitted to the EMBL/GenBank database (accession numbers Z70253-255; Z70258-261; Z70263-264; Z75558; X98894-898). To assess the degree of somatic hypermutation, comparison with the closest counterparts in the database of germline sequences has been made. This does not take into account any polymorphisms in V_L , but these are known to be insignificant in Vκ.26 Less information is available for Vλ genes, but again suggests only limited polymorphic variation.27 The V_L sequences obtained deviated significantly from the closest germline genes in the database (Tables 3 and 4), with a mean percent mutation of 5.3 for Vκ and 6.2 for Vλ. There was evidence for block mutations, involving two or more adjacent nucleotides, in both light-chain types. For Vκ, the high incidence of block mutations (6 of 9 sequences [67%]) compares with the reported figure of ∼50%.28 The numerical distribution of the mutations in FWRs and CDRs, and the ratio of replacement to silent mutations are shown in Tables 3 and 4. Deduced amino acid sequences are shown in Figs 2 and 3. In all cases, somatic mutations were identified in the V_L sequences, with 9 of 15 having additional identifiable mutations in J_L , even though events at the 3′ end of J_L are obscured by the primer sites. Several sequences derived from the same V_L family member were available for the Vκ genes O8/18 and A27. Comparison of these showed no evidence for common sites or “hot-spots” of mutational activity. Analysis of the distribution of somatic mutations in each sequence (Table 5) has been carried out by the method of Chang and Casali.17 In this method, each V_L or V_H gene sequence is assessed codon by codon for significance of deviation from germline sequence. A modification of the binomial distribution model is then used to calculate whether the probability (P in Tables 5 and 6) of an excess (in CDRs) or scarcity (in FWRs) of replacement mutations resulted by chance alone.17 For the FWRs, there were generally fewer replacement (R) mutations than expected due to chance, with significant (P < .05) conservation of sequence in 10 of 15 sequences, a feature commonly seen for V_H .14 For the CDRs, there were more R mutations than expected in 14 of 15 sequences, with significant (P < .05) clustering indicative of antigen selection in 4 of 15 (3Vκ and 1Vλ).

Comparison with V_H genes.For 6 cases (patients 1, 3, 6, 8, 14, and 15), tumor-derived V_H gene sequences were known.4,29 V_H sequences from the remaining 9 patients were obtained as described4 and deduced amino acid sequences are shown in Fig 4. Nucleotide sequences have been submitted to EMBL/GenBank database (accession numbers: Z70256-257; Z75556-5557; X98899-99003). Although the closest germline gene has been obtained from the database, rather than from the patients' DNA, it appears that in general polymorphism in V_H is not sufficient to require this approach.21 In fact, where we4 and others5 have analyzed the patients' germline V_H genes, the sequence has been found in the majority of cases to correspond exactly to that obtained from the database. However, in 1 of 9 cases of myeloma, a 2-bp difference from the published sequence of a VII-5 germline gene was found also in the patient's germline sequence, indicating that this was probably caused by a polymorphism.5 The distribution of somatic mutations in V_H of 6 of 15 of these cases indicated a significant clustering in CDRs (Table 6). Comparison of patterns in V_H with those in V_L (Table 6) showed that clustering in CDRs of V_H was not paralleled by clustering in CDRs of V_L . In addition, the clustering in V_L observed in 4 of 15 cases was not paralleled by clustering in V_H . Therefore, from the 10 cases where clustering was evident, it was localized in either V_H or V_L , but not both. However, in 5 of 15 cases, there was no significant clustering in either V_H or V_L .

Analysis of V-genes used by neoplastic B cells is extending our understanding of the origin and progression of B-cell tumors. Now that a complete map of the V_H gene germline repertoire is available,21,30 it is possible to compare a V_H sequence from a tumor cell to the germline gene of origin with confidence. Although some nucleotide changes may reflect polymorphic variation, particularly for certain V_H3 genes,31 it can be assumed that the majority of deviations from germline sequence in V_H genes of a B cell represent somatic mutations.21 In some cases this has been proved by comparing the tumor-derived sequence with the germline counterpart in the patient.4,5 Accumulation of such mutations indicate that the cell of origin has been exposed to the hypermutation mechanism in the germinal center.11,18,32 Heterogeneity of mutations within the tumor clone indicates that the tumor cell is still under the influence of the mutation mechanism, subsequent to neoplastic transformation.9,10 Finally, concentration of replacement mutations in CDRs can suggest a role for antigen in selection of the B cell.17 33 

In the case of multiple myeloma, V_H gene analyses from several laboratories have shown that the malignant cell has undergone extensive somatic hypermutation.3-5 There is further wide agreement that there is no intraclonal heterogeneity among the tumor cell population,3-5 and there is evidence that the V_H sequence is stable from diagnosis through plateau phase.34 These findings strongly suggest that the malignant cell has exited from the germinal center, and is no longer susceptible to the mutation mechanism.4 5 

The germline repertoire of V_L genes has also been mapped,27,35,36 but there have been fewer studies of usage in B-cell tumors. In myeloma, using DNA as a source, 7 Vκ sequences were obtained from 29 cases, with 4 of 7 potentially functional.37 Sequences were somatically mutated, with a hint of antigen selection from R:S ratios.37 A second study investigated Vκ-gene usage in 3 cases of myeloma.24 Together, these studies showed that 3 of 7 functional genes were derived from the O8/18 gene,24,37 and we have confirmed this incidence (4 of 9 cases). Although the Vκ1 family is often used by normal B cells,22,23 the level of usage of the O8/18 gene appears high in myeloma. However, frequency of this gene in other B-cell tumors has also been reported to be high, and it is not yet clear if there is a difference among the tumor categories.24 

The current results have focused on functional V_L genes, obtained from RNA. Identification of repeated sequences in the cloned PCR product supports the derivation from tumor cells, which can be a problem otherwise. We have analyzed the pattern of both Vκ and Vλ sequences. These confirm the high level of somatic hypermutation, with the level of 5.8% mutation for V_L being comparable with that of 8.2% for V_H .5 There is also a lack of intraclonal heterogeneity in V_L of all patients, again confirming findings in V_H , and supporting the concept that the final event in malignant transformation has occurred at a postfollicular stage.3,4,14 In contrast, the V_H genes in the benign counterpart of myeloma (monoclonal gammopathy of undetermined significance or MGUS), showed intraclonal heterogeneity in 3 of 7 cases.29 This could indicate that the clonal plasma cell in MGUS is less mature, and may have undergone some, but not all the events leading to malignant behavior.29 

If the final neoplastic event is late in maturation of the B cell, it might be expected that the myeloma precursor cell will have been subjected to the same processes of development as a normal B cell. Even if there is an IgM⁺ clonal precursor, which has undergone some neoplastic event, the few cases available for analysis have indicated that it has a homogeneous V_H gene sequence identical to the isotype-switched plasma cell.12,13,38 This would argue that neoplastic transformation in myeloma begins in a mature B cell immediately before isotype-switch. Because a B cell would have reached this point following antigen selection, the imprint of this procedure should remain as a clustering of mutations in CDRs of V-gene sequences.11,17,33 In fact, analysis of the stable sequences in myeloma should be particularly useful, because the selected sequence will not be obliterated by continuing posttransformation mutations. However, analysis of V_H sequences in myeloma has given mixed results, with only 21% of the tumor-derived sequences from a large series showing significant clustering in CDRs.5,14 This leaves open the question of the clonal history of the tumor cells in the remaining 79% of the cases. Because V_L sequence is also known to be involved in recognition of antigen,18 deductions from V-gene sequences that relate to a role for antigen in selection should be strengthened by including analysis of V_L .

In this study, significant clustering of mutations in CDRs of V_L was seen in 4 of 15 sequences. In contrast, a preliminary report of a study of Vκ sequences in 9 cases of myeloma has indicated that clustering of replacement mutations in CDRs occurred in all cases, but more details of the analysis are required.39 In our cases, comparison with V_H sequence in the same cell showed that clustering was in either V_H or V_L , but not both. This suggests that a role for antigen might be more common than estimated from V_H alone, reaching 67% in our study. There have been insufficient studies of normal human B cells to know if this is a typical finding. Investigation of the classical murine response to the hapten phenyl oxazolone has shown that affinity maturation is accompanied by somatic mutations in the CDRs of both V_H and Vκ.40 This is likely to be the case in human antibodies, and might suggest that a role for antigen is more common than estimated from V_H alone. For myeloma, the findings support the idea that the cell of origin has undergone the process of conventional antigen selection. However, there remains a minority of sequences which appear to have no clustering in either V-region. Even this pattern does not rule out a role for antigen selection, because optimal binding may occur via CDR3. Clearly we need more information on how normal human B cells generate antibody, but this study would suggest that deductions concerning a role for antigen in the clonal history of neoplastic B cells should take into account mutational events in both V_H and V_L .

We thank Dr D.G. Oscier for providing patient material and for helpful comments.