Insights into the regulation of immunoglobulin light chain gene rearrangements via analysis of the κ light chain locus in λ myeloma

Abstract

Accumulating evidence indicates that B cells may undergo sequential rearrangements at the light chain loci, despite already expressing light chain receptors. This phenomenon may occur in the bone marrow and, perhaps, in germinal centers. As immunoglobulin (Ig)κ light chains usually rearrange before Igλ light chains, we analysed, by polymerase chain reaction, the Igκ locus of bone marrow mononuclear cells from 29 patients with Igλ myeloma to identify earlier recombinations in marrow plasma cells. The results demonstrated that Igκ alleles were inactivated via the kappa-deleting element, presumably prior to Vκ-Jκ rearrangement, in many cases. Eighteen alleles (16 myeloma clones, 55%) showed Vκ–Jκ rearrangements, with increased utilization of 5‵ distant Vκ and 3′ distant Jκ gene segments (Jκ4, 56%), an indication of multiple sequential rearrangements. In-frame, potentially functional Vκ–Jκ rearrangements were found in approximately one-third of available rearrangements (as expected by chance), each one in different myeloma clones: three were germline encoded, while one had several nucleotide substitutions, suggesting inactivation after the onset of somatic hypermutation. Three of four potentially functional Vκ–Jκ rearrangements involved Vκ4–1, a segment considered to be associated with autoimmunity. These findings provide insights into the regulation of light chain rearrangements and support the view that B cells may occasionally undergo sequential light chain rearrangements after the onset of somatic hypermutation.

Introduction

Immunoglobulin variable (V) regions are generated via somatic recombination of variable (V), diversity (D) and joining (J) segments for the heavy chain and V and J segments for the light chains. This process is developmentally regulated by a recombinase complex that includes three lymphoid-specific enzymes [recombination activating gene 1 (RAG1), RAG2 and terminal deoxynucleotidyl transferase], and which is used to generate diversity in antibody V regions.¹ Two loci are present for light chains – the immunoglobulin (Ig)κ light chain gene (IGK, at 2p11.2) and the Igλ light chain gene (IGL, 22q11.2) loci – and usually mature B cells express either Igκ (60%) or Igλ (40%) light chains (isotype exclusion) and only one of the two isotypic alleles (allelic exclusion).

Recent evidence, however, documents that the immunoglobulin genes of B cells can undergo secondary rearrangements, even though the cells already express a functional antibody receptor.² As a consequence, new V regions can be formed. In the bone marrow, developing B cells can change receptors with autoantibody activity, thus contributing to sustaining self-tolerance, a process known as ‘receptor editing’ (secondary rearrangement in maturing lymphocytes).^2–5 Emerging data indicate that a similar phenomenon could occur in germinal centres (GCs),6–11 but the process appears to be slightly different in this site and is referred to as ‘receptor revision’ (secondary rearrangement in the periphery).^2,12,13 It has been proposed that receptor revision participates in the antibody affinity-maturation process, by offering new V regions to B cells that have acquired deleterious immunoglobulin mutations (e.g. stop codons, frameshifts) or residues that significantly decrease the affinity for the antigen, thus providing new substrates for somatic hypermutation.^2,12–14 However, the existence/relevance of this phenomenon is disputed: it is rare or difficult to detect in normal B cells,¹⁵ its contribution to the humoral response is doubtful, and it may be limited to immature B cells whose immigration from the bone marrow is facilitated by clonal expansion, such as during chronic inflammation or autoimmune states.^12,16,17 Inadequate antigen binding¹¹ would drive antigen-activated cells to secondary rearrangements in peripheral lymphoid organs. It is acknowledged¹² that the strongest evidence supporting the existence of receptor revision in humans comes from B cells that have undergone somatic mutation and concurrent receptor revision, but examples in normal lymphoid cells are just a few^18–20 and it is unclear whether the putative revised B cells may ultimately undergo plasma cell differentiation. On the other hand, evidence of receptor revision is accumulating in B-cell lymphomas^21–24 and autoimmune conditions,^25–27 giving rise to speculations on the role of late RAG activity in neoplastic transformation and self-reactivity.

The organization of the immunoglobulin light-chain genes offers unique opportunities to undergo secondary rearrangements¹² and it is therefore not surprising that secondary rearrangements have been studied and documented mainly in light chains: Vκ genes are arranged in both sense and antisense transcriptional orientations (in the latter case, rearrangements occur via inversion of the DNA between joined Vκ–Jκ segments, thereby conserving Vκ genes for further rearrangements); and there are no D segments, but multiple J genes, so that a 5′ V-gene segment can recombine with a 3′ J-gene segment to remove the original V–J rearrangement and replace it, and this can theoretically occur in both alleles and light chain loci. As IGL usually rearranges after IGK, Igλ⁺ cells have both Igκ alleles rearranged and inactivated to prevent further recombination and expression. Inactivation occurs via a RAG-dependent deletion mechanism involving a downstream element, the kappa-deleting element (Kde).²⁸ There are two major types of Kde recombinations (see the legend to Fig. 1 for an explanation), and one retains the Vκ–Jκ rearrangement that was rearranged just before Kde inactivation (Fig. 1a).²⁸ Hence, one approach to study sequential light chain rearrangements via polymerase chain reaction (PCR),²⁹ and to document possible receptor editing/revision, is to analyse the IGK locus for prior Vκ–Jκ rearrangements in terminally differentiated Igλ-expressing B cells. In this report, we employed Igλ⁺ multiple myeloma as a ‘single-cell model’ of marrow plasma cells, i.e. of cells that have completed the various phases of B-cell differentiation.

Figure 1.

Schematic diagram of the most frequently observed recombination activating gene (RAG)-mediated rearrangements that may occur at the IGK locus. The kappa-deleting element (Kde), situated downstream of the kappa constant region (Cκ), can rearrange to the intronic heptamer recombination signal sequence (Intron RSS) (a) or, alternatively, to the complete heptamer-nonamer of a Vκ region segment (b). In the first instance (a), the Vκ–Jκ join (adjacent to Cκ) is retained, allowing analysis of the Vκ region assembled just before Kde rearrangement (in this case a secondary rearrangement). The approximate position of the primers (→) employed in the PCR stategies used to study Kde/(primers A+B) and Vκ–Jκ rearrangements (primers A+C) are illustrated. Modified from Beishuizen et al.³⁰

Materials and methods

Clinical material

Ficoll-separated bone marrow mononuclear cells from 29 patients with Igλ⁺ light chain myeloma (14 IgGλ, seven IgAλ, eight BJλ) and plasma cell infiltration of ≥ 15% were used in our studies. An identification letter was attributed to each patient. Bone marrow was taken during clinical evaluation and after obtaining written, informed consent. Plasma cell infiltration was evaluated by morphology on May–Grünwald–Giemsa-stained marrow slides. Bone marrow DNA from two donors was used as a negative control.

Monoclonal rearrangements at the IGK locus

Monoclonal rearrangements at the IGK locus were analysed following previously described DNA PCR strategies.^30–32 Requirements for PCR amplification were that rearranged genes from normal polyclonal lymphocytes should not produce an amplification signal that is visible on an ethidium bromide-stained agarose gel and that, similarly, no product should be obtained from DNA in germline configuration because of excessive distance between the primers.^30–32 Briefly, 1 µg of DNA extracted from bone marrow mononuclear cells, and 0·2 U of DyNazyme I thermostable DNA polymerase (Finnzymes Oy, Espoo, Finland) were used for amplification in a final volume of 50 µl consisting of 1 × PCR buffer (10 mm Tris–HCl, pH 8·8, 1·5 mm MgCl₂, 50 mm KCl, 0·1% Triton-X-100), 200 nm dNTP mixture and 7·5 pmol of primers. PCR cycles consisted of 45 seconds at 94°, 90 seconds at 60° and 2 min at 72° for Kde rearrangement analysis; and 45 seconds at 94°, 45 seconds at 52° and 1 min at 72° for amplification of Vκ–Jκ rearrangements. Thirty cycles and a final extension step at 72° (60 min) were applied. The PCR products were run on a 1·5% agarose gel in 90 mm Tris-borate 2 mm EDTA (TBE) pH 8·0 and visualized by staining with ethidium bromide. Under these conditions, distinct amplification signals were obtained only in the presence of monoclonal B cells comprising ≥ 15% of the total cell population; this was demonstrated by dilution experiments in which DNA from the Daudi cell line was mixed in different dilutions with DNA from normal peripheral blood lymphocytes or from fibroblasts, as previously shown.³¹ No amplification was observed with DNA from two normal bone marrow donors. Amplified alleles were designated with the patient=s identification letter (followed by *1 or *2 as required to identify the two alleles).

Kde-mediated rearrangements were analysed using a set of seven framework 1 family-specific 5′ primers (VκI to VκVII, Fig. 1a, primer A) and a 3′ primer annealing to Kde (Fig. 1, primer B).³⁰ In order to identify the Vκ family employed, separate PCR reactions for each 5′ primer were used. Primers VκI to VκIV were from the V BASE Sequence Directory (I. Tomlinsonet al., MRC Centre for Protein Engineering, Cambridge, UK; http://www.mrc-cpe.cam.ac.uk) and primers VκV to VκVII were from Beishuizen et al.³⁰ Vκ–Jκ/intronic recombination signal sequence (IRSS)–Kde rearrangements generated fragments in the range of 1·5–2·8 Kb (according to the Jκ employed, Fig. 1a), and Vκ–Kde of ≈ 450–500 bp (Fig. 1b).

Vκ–Jκ rearrangements were analysed using the same set of Vκ family specific primers (Fig. 1, primer A) and an antisense oligonucleotide specific for the constant region proximal Jκ gene segment, Jκ5 (Fig. 1a, primer C). This technique permits amplification of the last Vκ–Jκ rearrangement attempt before IRSS–Kde inactivation (Fig. 1a), as previous Vκ–Jκ rearrangements, which may be retained in case of rearrangement by inversion, are not amplifiable by this approach. PCR fragments ranged from ≈ 350–1650 bp for rearrangement to Jκ5 or Jκ1, respectively (Fig. 1a). Because of the long distance between Vκ segments and Jκ5, germline DNA generated no product.³¹ PCR fragments were directly sequenced from both ends using the appropriate 5′ primer, a Jκ consensus oligonucleotide [5′-GAT(A/C)TCCA(G/C)(T/C)TTGGTCCC-3′] and an automated DNA sequencer. Although direct sequencing of PCR products mainly circumvents the problem of Taq DNA polymerase-introduced errors in sequence analyses, sequences were performed in at least two independent PCR products to minimize possible errors. A single, unambiguous sequence was obtained in each case.

Cloning of myeloma Vλ regions

Isolation of the expressed Vλ regions from myeloma was obtained by an inverse reverse transcription–polymerase chain reaction (RT–PCR)-based procedure, previously described in detail.³³

Data analysis and identification of mutations in myeloma light chain V regions

Somatic mutations in monoclonal light chain V regions were identified by sequence alignment to the closest germline gene present in the current releases of the EMBL-GenBank and V-BASE (V BASE Sequence Directory, I. Tomlinsonet al., MRC Centre for Protein Engineering) sequence directories using blast and dnaplot (H.-H. Althaus, University of Cologne, Cologne, Germany) search tools, respectively. Standard nomenclature was used for germline genes (http://www.ebi.ac.uk/imgt). Relevant nucleotide sequences are available in the DDBJ/EMBL/GenBank databases under accession numbers: AY271357–62.

Results

Rearrangements at the IGK locus in Igλ⁺ myeloma

The status of the IGK locus was analysed by PCR of bone marrow mononuclear cells from 29 patients with Igλ⁺ myeloma. A Kde-mediated deletion was present in all cases, i.e. no Igκ light-chain allele (n = 58, 29 clones) was in germline configuration. Thirteen myeloma clones (45%) had both alleles with Vκ–Kde (Fig. 1b), 14 (48%) with one Vκ–Kde and one Vκ–Jκ/IRSS–Kde (Fig. 1a), while only two clones (7%) had both alleles with Vκ–Jκ/IRSS–Kde rearrangements. Thus, slightly more than two-thirds (69%) of Kde rearrangements were to Vκ segments (Fig. 1b).

Vκ–Kde and Vκ–Jκ\IRSS–Kde rearrangements in Igλ⁺ myeloma

Figure 2 shows the frequency distribution of Vκ families involved in Vκ–Kde rearrangements in Igλ⁺ myeloma. The pattern was remarkably similar to that observed in Vκ–Kde rearrangements in B-chronic lymphocytic leukemia (CLL)³⁰ and in expressed Vκ regions from myeloma cells^31,34–37 or normal peripheral blood B cells (Fig. 2).^38–40 The VκI family was, to a large extent, that most frequently observed (≈ 55%, Fig. 2).

Figure 2.

Vκ families involved in Vκ–kappa-deleting element (Kde) and Vκ–Jκ/intronic recombination signal sequence (IRSS)–Kde rearrangements of Igλ⁺ myeloma. Family usage in Vκ–Kde rearrangements was similar to that observed in the Vκ–Kde rearrangements of B-chronic lymphocytic leukemia (B-CLL)³⁰ in expressed Vκ regions of myeloma³¹,³⁴–³⁷ and normal peripheral blood B cells.³⁸–⁴⁰ The VκI family was the most frequently involved. In contrast, over-representation of the VκII family was observed in Vκ–Jκ/IRSS–Kde rearrangements of Igλ⁺ myeloma. Vκ family usage is expressed as percentage of total Vκ (vertical axis).

By contrast, amplified Vκ–Jκ rearrangements (available only in the context of Vκ–Jκ/IRSS–Kde; one for each allele) revealed skewed usage of the VκII family (nine of 18 cases; 50%) (Fig. 2). Rearrangements to VκV–VκVII families were not found in the samples studied.

Analysis of Jκ segment usage, deduced from PCR fragment length, showed preferential utilization of the distal 3′ Jκ elements (Jκ4 accounted for 56% of Vκ–Jκ/IRSS–Kde rearrangements Fig. 3). Only one of 18 rearrangements (6% each) were to Jκ1 and Jκ2, segments that are frequently employed in Vκ regions from normal B cells^41,42 and Igκ⁺ myeloma (Fig. 3).^31,34–37

Figure 3.

Jκ usage in Vκ–Jκ/intronic recombination signal sequence–kappa-deleting element (IRSS–Kde) rearrangements of Igλ⁺ myeloma. Preferential utilization of Jκ distal segments was observed. Comparison is shown with Jκ usage in normal B cells,⁴¹,⁴² and Igκ⁺ myeloma.³¹,³⁴–³⁷. Over-representation of Jκ4 was noted. Jκ usage is expressed as percentage of total Jκ (vertical axis).

Sequence analysis of Vκ–Jκ regions within Vκ–Jκ\IRSS–Kde rearrangements

DNA from Igλ⁺ myeloma with Vκ–Jκ/IRSS–Kde rearrangements (16/29 patients, 55%, for a total of 18 alleles) was PCR amplified using the Vκ family-specific primers and a Jκ5 primer (Fig. 1a, primers A + C) to isolate the monoclonal, inactivated Vκ region located upstream of the IRSS–Kde rearrangement. This second PCR confirmed the Vκ families and Jκ segments observed in the analysis of Kde rearrangements (see above). Sequencing showed that 11 of 40 (27%) available Vκ genes were involved (Table 1), and Vκ2–30 was that most frequently observed (four of 18 rearrangements, 22%, Table 1). Jκ-distal germline genes were frequently used: Vκ2–30 is ≈ 380 kb distal from Jκ1, and six of 18 (33%) Vκ genes were Vκ2–30 distal; four of 18 (22%) Vκ genes were part of the Jκ distal cassette,⁴³ which is infrequently rearranged in the normally expressed repertoire (< 10%).^41,42 On the other hand, Vκ4–1, the single member of the VκIV family and the most proximal gene segment to the Jκ cluster, was frequently found to be rearranged (three cases, Table 1).

Table 1.

Vκ–Jκ/intronic recombination signal sequence–kappa-deleting element (IRSS–Kde) rearrangements in Igλ⁺ multiple myeloma

Vκ family	Alleles	Vκ	Jκ	NS	Frameshift at the Vκ–Jκ junction
Vκ1 (22%)	I	1–5	1	2	+
	Q	1–27	4	–	+
	P	1–33/1D-33	4	1	+
	A*1	1–33/1D-33	3	–	+
Vκ2 (50%)	C	2–28/2D-28	5	2	+
	B*1	2–29	3	–	+
	L	2–30	3	–	+
	M	2–30	2	–	+
	A*2	2–30	4	–	+
	B*2	2–30	4	1	–
	N	2D-30	4	–	+
	H	2D-29	4	–	+
	D	2–26(b)	4	–	–
Vκ3 (17%)	G	3–20	5	–	+
	F	3D-20	4	–	+
Vκ4 (11%)	O	4–1	4	–	–
		E	4–1	4	–	–
	R	4–1	3	10	–

Alleles were named with the patient=s identification letter followed by *1 or *2 when both presented V_κ–J_κ/IRSS–Kde rearrangements (as in patients A and B) NS, nucleotide substitutions.

V_κ2–26 is a pseudogene (frameshift caused by an extra adenine at codon +39).

Rearrangements showed no frameshifts at the V_κ–J_κ junction in five of 18 alleles (16 clones). Alleles are grouped according to the V_κ family.

Vκ–Jκ remnants with productively rearranged CDR3 constituted approximately one-third of cases (five of 18, 28%), and each was present in a different myeloma clone. A pseudogene was involved in one instance (Table 1, patient D; Vκ2–26 pseudogene with a frameshift at codon 39); hence, four of 18 were the potentially functional/expressed Vκ–Jκ rearrangements (Table 1).

Sequence comparisons with the most homologous germline genes in databases revealed that most Vκ–Jκ rearrangements (13/18, 72%) were identical to germline sequences and that four (22%) had just one or two nucleotide changes (Table 1). Among the potentially functional Vκ–Jκ remnants, three were germline encoded, while one (Table 1, patient R) presented 10 nucleotide substitutions distributed throughout the VL region (Fig. 4a), indicating somatic hypermutation. A comparison of both nucleotide and deduced amino acid sequences with those present in databases revealed two non-conservative amino acid changes in framework region (FR) positions that are almost invariant in humans and mice, both in Igκ and Igλ light chains, and which have never been found in VκIV light chains. Both substitutions are located in turns between β sheets: Pro40Ala, between the end of CDR1 and FR2; and Glu81Gly, which interrupts a conserved sequence of two negatively charged amino acids located in FR3 (Glu81–Asp82) (Fig. 4a, positions 40 and 81).

Figure 4.

Deduced amino acid sequences of the aborted Vκ–Jκ region (a) (Vκ R) and expressed Vλ region (b)(Vλ R) from the Igλ⁺ myeloma clone R. Comparisons are made with the most homologous germline gene segment, namely Vκ4–1 (a) and Vλ1–44 (b). The most relevant amino acid substitutions (Pro40Ala and Glu81Gly), found in the aborted Vκ–Jκ region, are outlined (a) (Vκ R: positions 40 and 81). Replacing mutations, upper case letters; silent mutations, lower case letters.

Nucleotide sequencing of the expressed Vλ region from this same myeloma patient (R) revealed numerous nucleotide changes (n = 27) (Fig. 4b), consistent with somatic hypermutation activity and involvement in a GC reaction.

These findings suggest that the original myeloma B cell was actually expressing the Vκ–Jκ rearrangement during the GC reaction, but subsequently inactivated the allele via Kde rearrangement and rearranged λ light chain genes, thus giving rise to a somatically mutated Igλ⁺ light chain clone.

Discussion

To study rearrangements at the light chain locus, we used multiple myeloma as a ‘single cell-model’ of terminally differentiated B cells. We provide evidence of sequential rearrangements at the IGK loci, some of these compatible with receptor editing and one with revision. These findings favour the view that RAG1/2 activities may be present at both early and late phases of B-cell differentiation.

Inactivation of both Igκ alleles was found in all Igλ⁺ myeloma, consistent with the hierarchical model of ordered immunoglobulin light chain gene rearrangement (Igκ before Igλ) and with stringent isotype exclusion. Alhough it cannot be formally excluded that differentiating B cells rearranged the Igλ locus first and then proceeded to inactivate Igκ alleles, the extensive sequential Vκ to Jκ rearrangement we observed is more readily compatible with earlier Igκ light-chain recombination.

Kde was more frequently rearranged to a Vκ gene (Fig. 1b, 69% of cases) than to IRSS (Fig. 1a). This phenomenon could be ascribed to differences in recombination signals. The RAG complex may recognize the complete canonical heptamer-nonamer RSS of Vκ segments more efficiently than the incomplete heptamer of IRSS. Similar results were observed in precursor-B-acute lymphoblastic leukemia (B-ALL) by Beishuizen et al.³⁰ (Vκ–Kde, 75%).

Vκ family use in Vκ–Kde rearranged genes (from Igλ⁺ myeloma or B-CLL³⁰) was remarkably similar to the expressed Vκ regions from myeloma^31,34–37 and normal^38–40 B cells (Fig. 2), suggesting comparable germline gene usage. Indeed, analysis of Vκ genes in Vκ–Kde from precursor-B-ALL and B-CLL³⁰ revealed analogous biases as in Vκ–Jκ rearranged genes from the expressed repertoire. Taken together, these findings indicate that a similar group of Vκ segments is involved in Vκ–Kde and Vκ–Jκ rearrangements, and that certain Vκ genes are more prone to recombine independently of the type of rearrangement (Vκ–Jκ or Vκ–Kde). As early Vκ–Jκ recombination would make the most proximal and frequently used Vκ genes unavailable for further rearrangement, these findings indicate that Vκ–Kde recombination commonly takes place prior to regular Vκ–Jκ rearrangements. Moreover, the high prevalence of Vκ–Kde rearrangements we observed suggests that Kde recombination to Vκ segments may occur with efficiencies that are in the same order of magnitude as that of standard Vκ–Jκ rearrangements, and because many Igλ⁺ myeloma cells have both Igκ alleles inactivated by Vκ–Kde recombination, Igλ⁺ cells can be generated without prior Vκ–Jκ attempts.

By constrast, Vκ–Jκ rearranged genes, in the context of IRSS-Kde rearrangements, presented preferential utilization of 5′ distant Vκ and 3′ distant Jκ segments, segments that are infrequently employed in the normal repertoire. Similar findings were recently documented by Bräuninger et al.,⁵ in fluorescence-activated cell sorter (FACS)-sorted naïve Igλ-expressing B cells. As positive selection cannot explain this phenomenon (70% of alleles were non-productively rearranged), we considered the skewing towards distal Vκ and Jκ segments as evidence of multiple rearrangement attempts. Hence, Kde rearrangement to IRSS occurred late, often after completion of possible rearrangements.

Based on the above findings, we propose different regulations of Kde-mediated deletions: while rearrangement of Kde to Vκ segments probably occurs on loci that are in germline configuration, Kde rearrangement to IRSS preferentially occurs on chromosomes that have already undergone multiple sequential Vκ–Jκ recombinations. Given the scarcity of Igλ⁺ clones with both alleles showing IRSS-Kde rearrangement (7%), our results suggest that Igλ⁺ plasma cells can be generated either with few, if any, prior Vκ–Jκ attempts (approximately 50% of cases, both alleles with Vκ–Kde), or with multiple sequential Vκ–Jκ rearrangements occurring on just a single allele (50% of cases, one Vκ–Kde and one IRSS–Kde inactivation). Our reasoning is in line with computer simulation experiments carried out by Mehr et al.,⁴⁴ showing that the κ/λ ratio of 3 : 2 and allelic exclusion could be explained if light chain rearrangement is a highly ordered, rather than a random, process. The proposed model⁴⁴ comprises ordered rearrangement and repetitive, sequential rearrangements occurring preferentially on the same Igκ light-chain allele, with sequentiality of Jκ segment usage.

Receptor editing and revision replace functionally rearranged light chains expressed in the bone marrow and GCs, respectively. While editing would occur on unmutated V regions, revision is expected to replace mutated light chains, if revision occurs after the onset of hypermutation. Sequencing analysis of Vκ–Jκ rearranged genes (n = 18) in Igλ⁺ myeloma showed that productive Vκ–Jκ rearrangements were present in approximately one-third of cases (28%). This frequency, close to that theoretically expected for a random recombination, was also observed by Bräuninger et al.⁵ in peripheral blood pre-GC B cells.

In-frame sequences were then analysed for nucleotide substitutions with respect to the closest germline gene present in databases. As Igκ enhancers are necessary for the full development of somatic hypermutation,⁴⁵ the finding of productively rearranged, hypermutated Vκ–Jκ rearranged genes in post-GC Igλ⁺ plasma cells would indicate that precursor B cells expressed, mutated and subsequently inactivated Igκ light chains, events that are consistent with receptor revision occurring after the onset of somatic hypermutation. Two functionally rearranged Vκ–Jκ rearranged genes were germline encoded (Table 1, alleles O and E), while one had just a single nucleotide variation (Table 1, allele B*2), and could be ascribed either to editing in the bone marrow² or to revision before the onset of somatic mutation in low antigen-affinity GC clones. On the contrary, one productively rearranged Vκ–Jκ remnant presented 10 nucleotide substitutions distributed throughout the Vκ region (Table 1, allele R), a number of changes that is consistent with light chain participation in the T-cell-dependent immune responses in GCs.^31,34–37,^42,46 As the mutated Vκ–Jκ rearrangement was, most probably, only subsequently inactivated via Kde-mediated enhancer deletion, a genetic event involving RAG complex activity (Kde rearrangement) occurred very late in B-cell differentiation, after the onset of somatic hypermutation. Furthermore, this finding demonstrates that cells which underwent late RAG activity can be selected to participate in antibody immune responses and to differentiate into antibody-producing plasma cells. We can envisage two different interpretations for this example: either the original B cell was an Igκ producer that inactivated the Igκ allele at times of GC reaction and subsequently rearranged the Igλ allele, which underwent new rounds of antigenic selection and somatic hypermutation, or the original B cell was one of the few κ/λ dual-receptor positive B cells that are reported to occur rarely in the B-cell repertoire.⁴⁷ This cell could co-express and co-mutate Igκ and Igλ light chains in the GC, subsequently inactivate the Igκ light chain, and eventually differentiate into Igλ-only plasma cells. At present, both interpretations may be valid. However, it is conceivable that double producers, which may be generated via extensive editing (as recently shown in a mouse model)⁴⁸ would possess the RAG activity needed to perform Kde inactivation in the periphery.

Analysis of amino acid changes in the aborted Vκ–Jκ region of allele R revealed two non-conservative substitutions (Pro40Ala and Glu81Gly) in almost invariant framework positions. Database searches showed that no Igκ light chains present substitutions in these positions simultaneously. In particular, we can predict that replacement of Pro40 with Ala could significantly affect the protein stability. Pro40, in cis configuration, is in fact located in a β turn and, as documented for other proteins,^49,50 the mutation Ala to Pro in β turns has a strong positive effect on the stability of protein structure. Hence, it is predictable that these amino acid changes could, in some way, hinder protein folding or stability, leading to reduced expression/binding, features that are considered to drive Igκ light chain inactivation and revision.⁵¹

The present analysis is limited to those successful immunoglobulin recombinations leading to light-chain isotype switching, a very particular event, and it does not provide an estimation of receptor editing/revision prevalence in B-cell differentiation. Theoretically, receptor editing/revision could occur in both heavy and light chains (in both alleles and light chain isotypes), and cases of heavy chain revision have been detected¹⁸ (although it is more difficult in heavy chains because of D segments and presumably mediated by cryptic embedded heptamers).⁵² However, as regards receptor revision, a recent comprehensive analysis conducted using a different PCR strategy and cell population (peripheral blood post-GC B cells), concluded that such events are rare or mostly counterselected, arguing against a major contribution of revision to the affinity-maturation process.¹⁵ Our analysis presents just a single, but clear-cut, example of such recombinations, suggesting that, at least in a few instances, this phenomenon takes place.^18–20,26 The neoplastic nature of the plasma cell population analysed might have favoured identification of such gene recombinations via selection, given the accumulating evidence of revision in B-cell lymphoma.^21–24

Analysis of aborted Vκ–Jκ rearrangements, besides showing preferential use of distal infrequent Vκ and Jκ segments, also suggested increased representation of Vκ4–1 (VκIV family, Table 1), a gene that rearranges by inversion and that is the closest gene to the Jκ cluster.⁴³ In agreement with multiple sequential recombination events, Vκ4–1 was found to be rearranged to 3′ distal Jκ segments (once to Jκ3, twice to Jκ4; Table 1). The Vκ4–1 germline gene has been linked to autoimmune disorders, such as lupus erythematosus, and is thought to encode auotoantibodies to DNA.⁵³ Increased representation of Vκ4–1 light chains, as a consequence of extensive receptor editing, has also been observed in a small subset of peripheral B cells that expressed RAG mRNA, displayed an unusual heavy and light chain repertoire consistent with self-reactivity, and appeared to accumulate in the joints of patients with rheumatoid arthritis.⁵⁴ Extensive receptor editing was also involved in the enrichment of Vκ4–1-encoded Vκ regions in the productive repertoire compared to non-productive Vκ–Jκ rearrangements of a lupus patient.⁵³ Remarkably, Vκ4–1 was found in nearly all potentially functional Vκ–Jκ rearranged genes we observed (three of four, Table 1), and all Vκ4–1-encoded Vκ-Jκ rearrangements (three of three) were productively rearranged. As discussed above, unfavourable amino acid substitutions probably drove revision in the case of the Vκ4–1 light chain of patient R, an event that occurred in the periphery where checks for self-reactivity are not expected to be involved. On the contrary, the above evidence suggests that the other two unmutated Vκ–Jκ rearrangements (Table 1, alleles O and E) could have been edited in the bone marrow because of self-reactivity, in line with the notion that receptor editing is involved in autoimmunity control.