Differential use of immunoglobulin light chain genes and B lymphocyte expansion at sites of disease in rheumatoid arthritis (RA) compared with circulating B lymphocytes

Abstract

The presence of germinal centre-like structures and clonotypic expansion of lymphocytes in RA synovia may indicate a site-specific immune response to local antigens, rather than passively entrapped immune cells, that sustains synovial inflammation. In this study we compare the nature of immunoglobulin light chain variable region gene use in the synovium of RA patients with peripheral B cells to determine the nature of the synovial immune response. Using Vλ and Vκ gene fingerprinting, which relies on differences in CDR3 length, we demonstrate differences in the pattern of Vλ and Vκ use and clonotypic expansion of B cells between the synovium and peripheral blood of RA patients. Further, we show that some synovial rearrangements with long CDR3 are selectively expanded. These longer than usual CDR3 were generated by a number of mechanisms including N-additions. However, the observed differences were not uniform in different patients. These observations suggest that local synovial antigens drive significant numbers of T and B lymphocytes selected from an existing repertoire shaped by genetic and environmental factors. Further, the data argue against passive retention of most B cells in the synovium of RA patients.

INTRODUCTION

Joint pathology in RA is characterized by infiltration of macrophages, T and B lymphocytes and plasma cells actively secreting IgM, IgG and IgA antibodies, some with binding specificity for IgG (rheumatoid factors (RF)) [1–3]. In long-standing RA, an increase in the total number of synovial plasma cells is associated with increased severity of disease [4] and chronic synovitis [5]. B cell aggregates forming structures which resemble germinal centres of secondary lymphoid tissues have also been observed in the synovium of ≈ 20% of these patients [6]. These structures appear to support clonal expansion and somatic mutation in immunoglobulin variable region (IgV) genes [7,8]. However, it is unclear yet whether the synovial immune response is site-specific or reflects trafficking of cells activated elsewhere.

Recently, Lee et al. observed that B cells expressing the Humkv325 κ light chain gene, which is frequently associated with RF, were preferentially expanded in the synovium but not blood of an RA patient [9]. Sequences of the rearranged genes revealed evidence for somatic mutations, suggesting that an antigen-driven immune response was responsible for the B cell expansion. In addition, transcripts from these cells possessed longer than expected complementarity determining region 3 (CDR3) which had N-additions in the Vκ–Jκ junction. Further studies indicated that although N-additions were occasionally found in κ transcripts from peripheral blood lymphocytes (PBL) of healthy neonates and adults [10] and RA patients, N-additions were found more frequently in κ transcripts isolated from synovial fluid (SF) of RA patients [11,12]. These observations may suggest selective oligoclonal expansion of B cells expressing these unusual gene rearrangements in the synovium.

To determine whether this unusual pattern of IgV gene expression in the synovium is a common feature in RA and related to the disease process rather than reflecting peculiarity of individual patients or genes studied, we have studied the nature of light chain gene diversity within the SF and PBL of three RA patients using a new approach. The approach capitalizes on the diverse pattern of V–J recombination and size variation within the CDR3, to detect clonotypic expansion in RA synovia. Here we show that the spectrotypes of Vλ and Vκ rearrangements in SF B cells are different from those in matched PBL, and that specific light chain V gene rearrangements can be found enriched in the SF. Isolation of individual size-separated Vλ transcripts revealed that some of the SF-enriched rearrangements had evidence of somatic mutation and longer than expected CDR3 due to frequent N-additions. The results also showed variation between individual patients in the use of IgV genes from different light chain gene families.

PATIENTS AND METHODS

Patients

SF and matched blood from three long-standing RA patients (C, D and F) were used in the study. Patient D was an 85-year-old woman, patient C a 55-year-old woman, and patient F a 61-year-old man. All three patients fulfilled the ARA criteria for RA [13].

Separation of mononuclear cell populations

SF and blood were collected in heparin. SF was treated with hyaluronidase (1 U/ml) at 37°C for 1 h and mononuclear cells (MNC) from blood and SF separated on Ficoll–Paque (Pharmacia, UK). Monocytes and natural killer (NK) cells were eliminated using l-leucine methyl ester and the lymphocyte-enriched MNC used for mRNA extraction. Lymphocyte-enriched MNC from the SF and blood of patient C were further fractionated into B cells expressing membrane immunoglobulin with RF activity (RF⁺ B cells), and cells without RF activity (RF⁻ B cells) using magnetic beads (Dynal, Wirral, UK) coated with a human IgG4 paraprotein. A human IgG4 paraprotein with known ability to act as a good target antigen for RF [14] was used for coating magnetic beads to avoid cell separation due to Fcγ–receptor interaction (IgG4 protein was kindly provided by Professor R. Jefferis, University of Birmingham Medical School, UK). To determine the efficiency of the separation procedure, different fractions of B cell-enriched MNC from the blood of three RA patients (not included in the molecular studies) were cultured according to the protocol originally described by Banchereau et al. [15] prior to spectrotype analysis. Briefly, enriched B cells (10⁴/ml) were cultured, in the presence of anti-CD40 MoAb (clone G28-5; 0.5 μg/ml), 1 U of human IL-4 and IL-10 (final concentration of 5 ng/ml each), on a monolayer of irradiated mouse L cells transfected with the human FcγRII (CD32) in a 96-well culture plate. Supernatants were harvested after 10 days of culture and tested for total immunoglobulin and RF isotypes by ELISA. Ninety-six-well microtitre ELISA plates were coated either with MoAbs specific for the immunoglobulin isotypes (clone AF6 anti-IgM, clone 8a4 anti-IgG or clone 2D7 anti-IgA; kindly provided by Professor R. Jefferis) or with human IgG4 or IgG1 paraprotein to detect RF isotypes (25 μg/ml in PBS at 37°C for 2 h), washed three times in PBS/Tween, incubated (37°C, 2 h) with cell culture supernatants (50 μl). Bound immunoglobulins of the different isotypes were revealed with horseradish peroxidase (HRP)-conjugated goat F(ab)₂ anti-human μ, α, γ or light chains (when detecting IgG RF activity bound to the IgG-Fc-coated plates).

mRNA extraction and cDNA preparation

mRNA was extracted using a mRNA extraction kit (Invitrogen, Leek, The Netherlands). cDNA was synthesized using SuperScript II RNase H-reverse transcriptase (Gibco-BRL, Paisley, UK) and a poly-dT primer. The cDNA was suspended in TE buffer (10 mm Tris–HCl, 1 mm EDTA, pH 8.0), before the addition of a homopolymeric G tail to the 3′ end using terminal deoxynucleotidyl transferase (TdT; Pharmacia).

Anchored-polymerase chain reaction

Poly G-tailed cDNA (5 μl) was amplified using a primer specific for the poly G tail (anch2pc: 5′ ACGAATTCTAGAGTCGACCCC CCCCCCCCCC 3′) with a consensus primer for λ constant regions (CL1: 5′ AAAAATTCACACCAGTGTGGCCTTG 3′) or κ constant region (CK1: 5′ AACAGAGGCAGTTCCAGAGTT 3′). PCR mixture (50 μl), containing 5 μl cDNA, 5 μl 10× polymerase chain reaction (PCR) buffer (Perkin-Elmer, Beaconsfield, UK), 100 ng anch2pc and 175 μm dNTP, was incubated at 96°C for 5 min, 60°C for 5 min before adding 1 U of Taq polymerase. The anch2pc was allowed to extend for 15 min at 70°C and 100 ng of CL1 or CK1 were added. The PCR was carried out for 30 cycles of 94°C/1 min, 60°C/1 min, 72°C/1 min for each cycle, with a final extension of 72°C/10 min.

Vλ and Vκ gene family-specific fingerprinting

These were performed by PCR using 4 μl of anchored PCR (a-PCR) product, 1 μl 10× PCR buffer, 10% dimethyl sulfoxide, 1.5 mm dNTP, 50 ng of a primer specific for one of the light chain V gene families and 50 ng of γ-³²P-labelled nested constant region primer 3′ of CL1 and CK1, CL2 and CK2, respectively. Sequences of the primers used were: VλI (5′ CAGTCTGTVBTG ACKCAGCCRC 3′), VλII (5′ CASDYTRYVSTGACTCARSM KSC 3′), VλIII (5′ TCCTATGWGCTGACWCARSHMY 3′); VλVI (5′ AATTTTATGCTGACTCAGCCCC 3′), VλVII (5′ CA GGCTGTGGTGACTCAGGAGC 3′), VλVIII (5′ CAGCTTGTG CTGACTCAATCGC 3′), VλIX (5′ CAGCCTGTACTGACTCA GCCAC 3′); CL2 (5′ AAGGATCCGGGCGGGAACAGAGTGA CCGAGGG 3′); VκI (5′ GMCATCCRGWTGACCCAG 3′), VκII (5′ TCTCCACTCTCCCTGCCCGTC 3′), VκIII (5′ WGTGWTGA CGCAGTCTCCA 3′); VκIV (5′ ACTGACTACAGGTGCCTACG 3′), CK2 (5′ TGCTTCGGATCCGAAGATGAAGACAGATGGT GC 3′) (degenerate positions Y = C or T, R = A or G, W = A or T, S = G or C, K = T or G, M = C or A, V = not T, H = not G, B = not A, D = not C). The mixtures were heated at 95°C/7 min and 1 U Taq added. The PCR were run for 25 cycles: 90°C/30 s, 60°C/30 s, and 72°C/2 min for the Vλ1, VI, VII, VIII and IX and the VκI, II, III and IV families, or 25 cycles: 90°C/1 min, 64°C/1 min, and 72°C/2 min for VλII genes. Amplification conditions for VλIII genes were 25 cycles: 90°C/1 min, 62°C/1 min, and 72°C/2 min. In all cases a final extension of 10 min at 72°C was performed. The primers were designed to anneal to the first framework regions (FR1) on the basis of published sequences and sequences from Kabat et al. [16,17]. To confirm results obtained from selected experiments, different mRNA aliquots were used to re-generate cDNA and reproduce the fingerprinting profiles.

Separation of PCR products into fingerprinting spectrotypes

PCR products were separated on a 6% polyacrylamide gel. For calibration of the VλI, II, III, VκI, II, III and IV spectrotypes, PCR products of rearrangements of known length from established B lymphocyte hybridomas were used as size markers on each gel (Table 1). Each gel was dried onto Whatman 3MM filter paper without fixing and exposed to an x-ray film. DNA bands of interest were cut out and rehydrated in double-distilled H₂O (ddH₂O). The DNA was cloned using the TA cloning vector (pCRII) (Invitrogen) and plasmid DNA alkali-denatured, ethanol-precipitated and sequenced using Sequenase T7 DNA Polymerase (USB Corp., Cleveland, OH). Spectrotype bands used for DNA extraction and sequences isolated from these bands were named according to the following formula:

Table 1.

Predicted sizes of light chain gene fingerprinting polymerase chain reaction (PCR) products involving rearrangements with and without the loss/addition of nucleotides within the CDR3

where X = patient identity C, D or F, J = cell population used as source of mRNA, i.e. S for SF and P for PBL, RF for RF⁺ B cells, neg for RF⁻ B cells, N = light chain V gene family, e.g. 1 for VλI, 2 for VλII, etc., t/m/b = position of the band within the lane, i.e. top, middle or bottom, Y = number of the recombinant clone sequenced. Sequence information was processed and compared with existing sequences in Genbank and EMBL data bases using Lasergene software (DNAstar Inc., Madison, WI). The principles followed in assigning additional nucleotides in the V–J junctions were: nucleotides were assigned as part of the non-coding ends of the germ-line gene by comparison with the available germ-line gene sequences. P-additions were assigned as originally described [18]. Nucleotides were assigned as N-additions when they could not be assigned to the non-coding nucleotides 3′ or 5′ of the germ-line Vλ or Jλ genes, respectively, the associated heptamer/nonamer signal sequences or P-additions.

RESULTS

Vλ and Vκ spectrotypes reveal the expansion of B cell clones with different patterns of V–J rearrangements in SF and PBL

Most of the PCR products consisted of 1–7 bands, depending on the V gene family, with each band differing in length by a minimum of three nucleotides, consistent with the expected codon variation (increments of three nucleotides). The Vλ spectrotypes were generally more complex than the Vκ spectrotypes (except the VκIII spectrotype), suggesting greater variation in length of the coding region within a family and/or Vλ CDR3 (Figs 1 and 2). To confirm reproducibility, different cDNA preparations from the same cell population were used in the PCR reactions on at least two separate occasions. The spectrotypes obtained for individual Vλ or Vκ rearrangements were reproducible in all cases. However, because of the small cell yield, it was not possible to use the same cell population to test different RNA preparations on separate occasions. To determine efficiency of separating RF⁺ B cells, supernatants from total unfractionated B cell-enriched MNC, RF⁺ and RF⁻ B cells from the blood of three RA patients cultured for 10 days were assayed for total immunoglobulin and RF isotypes. Of all immunoglobulin-producing lines from the unfractionated MNC, 48% contained RF activity. In contrast, 6.5% of the fractionated RF⁻ B cells and 81% of the RF⁺ B cell populations had RF activity. This suggested that the B lymphocytes which could secrete RF in the CD40/IL-4/IL-10 culture system had been significantly enriched in the RF⁺ B cell fraction.

Fig. 1.

Immunoglobulin gene fingerprinting spectrotypes of synovial fluid (SF) and peripheral blood lymphocyte (PBL) Vλ rearrangements from patients D, C and F. Family origin of the spectrotypes is given on the left of, and source of B cells below, the autoradiographs. The first letter identifies the patient. Tissue source of the lymphocytes is indicated as P for PBL and S for SF. Rearrangements from RF⁺ are indicated as RF, while those from RF⁻ B cells are indicated as neg. The numbers given on the left of each autoradiograph indicate length (in bp) of the marked band. The arrows indicate bands used to extract DNA for cloning and sequencing.

Fig. 2.

Immunoglobulin gene fingerprinting spectrotypes of synovial fluid (SF) and peripheral blood lymphocyte (PBL) Vκ rearrangements.

The SF light chain gene spectrotypes generally differed, either in number of bands, pattern or intensity of individual bands, from those of PBL (Figs 1 and 2). However, these differences were not uniform in different patients. The VλI and VλII spectrotypes of patient D were particularly interesting in that rearrangements with CDR3 longer than the typical 11 amino acid residues were present in the SF but not the matching PBL (Fig. 1; for comparison of the expected CDR3 length see Table 1). The SF VλIII spectrotype of patient D had a band of longer CDR3 than the common dominant band (present in all samples) that was absent in the PBL. Similar rearrangements with long CDR3 were present in the VλI and III spectrotypes of patient C (and VλIII spectrotype of patient F), but not confined to the SF. A few spectrotypes had abnormally large bands. The most obvious of these were the large bands in the VλI and VI spectrotypes of PBL RF⁺ and RF⁻ B cells, and the VλIX spectrotype of SF RF⁺ B cells of patient C. Extraction and sequencing of DNA revealed the insertion of in-frame nucleotides into the FR1 of these rearrangements (presented below). The VλIX spectrotypes had faint bands (marked with stars) smaller than the common band, but subsequent sequencing demonstrated rearrangements within this band to be VλII genes cross-amplified by the VλIX and CL2 primers. This was the only example encountered in the study where non-specific amplification was observed. Nevertheless, caution should be exercised in interpreting results of PCR using family-specific primers without the availability of nucleotide sequences. The VλVI and VII spectrotypes were more diverse than the VλVIII and IX families, especially for RF⁻ B cells in patient C, although all four families were small and comprised similar numbers of genes.

Spectrotypes of VκI, II and IV families were restricted (1–3 bands, except patient C, Fig. 2) and relatively similar in PBL and SF of patients C and F, but with an apparent enrichment of VκIV rearrangements in the SF of patient D and significantly in PBL RF⁺ B cells of patient C (CPRF). The inherent size diversity present in VκIII germ-line genes means that the large bands observed in the SF VκIII spectrotype of patient D can represent a CDR3 of either 11 or 10 residues, depending on the VκIII gene (Humkv325 is three nucleotides shorter than Humkv328; Table 1). Nevertheless, size of the bands indicates that at least six nucleotides may have been inserted in some rearrangements (DS and top band of CPneg3).

N-additions contribute to generating Vλ rearrangements with long CDR3 in SF B cells of patient D

Sequencing of DNA from common and ‘restricted’ bands in the PBL and SF VλI, II and III spectrotypes of patient D revealed that size variation in the Vλ–Jλ junctions in different Vλ families was generated by a range of mechanisms, including exonucleotide nibbling, inaccurate splicing, and P- and N-additions (Table 2). Seven rearrangements (DS1-1, DS1-2, DS1-3, DS1-4, DS2-T4, DS2-T1, and DS3-T4) from the single SF VλI band, or the longer bands in the VλII and III spectrotypes had additional nucleotides in the Vλ–Jλ junction consistent with P-additions. A total of eight rearrangements had putative N-additions, involving 1–3 nucleotides (Table 2). The added nucleotides were predominantly G/C, consistent with TdT activity.

Table 2.

Identification and assignment of nucleotides found at the Vλ–Jλ junction of rearranged Vλ genes isolated from specific spectrotype bands in the synovial fluid (SF) and blood of patient D

All six rearranged VλI genes (DS1-1 to DS1-4, DS1-9 and DS1-10) from the 378-bp band in the SF spectrotype had three additional nucleotides in the Vλ–Jλ junction, and aligned with the Humlv1042 germ line. All six sequences shared identical Vλ–Jλ junctions and some nucleotide changes in the CDR and FR (Fig. 3a), suggesting that they originated from the same progenitor B cell. The only rearranged Humlv1042-related gene isolated from the PBL spectrotype did not have additions, but instead had a deletion of three nucleotides from the V–J junction. Six VλI rearrangements isolated from the 372-bp band of the PBL spectrotype of patient D aligned with the highest similarity (91.5–98.5%) to the DPL5 (DP1-1, DP1-3, DP1-5, DP1-9) and Humlv111 (DP1-2, DP1-8) germ-line genes (Fig. 3b, c). None of these sequences had additional junctional nucleotides.

Fig. 3.

(See previous page.) Nucleotide sequences of VλI rearrangements from individual spectrotype bands of patient D. Designation of the sequences is explained in Patients and Methods. In most cases only codon positions with nucleotide changes from the related germ-line genes are shown. Identity with the germ-line genes is given as dots and changes resulting in amino acid replacements are given in capital letters, while silent changes are indicated as small letters. Sequences that are related to the germ-line genes Humlv1042, DPL5 and Humlv318 are given in a, b and c, respectively. The germ-line genes are given in bold. Numbering is according to Kabat et al. [17]. The predicted amino acids are given above the nucleotide sequence. Assignment of the additional nucleotides observed at the Vλ–Jλ junctions is explained in Table 2.

The difference in length of the Vλ rearrangements from the upper 378-bp band in the SF VλII spectrotype of patient D, and the lower 372-bp band in the PBL and SF spectrotypes were also confirmed as six nucleotides. The four Vλ rearrangements DS2-T1, DS2-T2, DS2-T4, and DS2-T5, isolated from the 378 band in the SF spectrotype, each had three additional nucleotides. However, in contrast to the rearrangements from the VλI 378-bp band, these rearrangements were not clonally related (Fig. 4).

Fig. 4.

Nucleotide sequences of VλII rearrangements from spectrotype bands of patient D. Sequences that are related to the germ-line genes DPL11 and DPL10 are given in a and b, respectively.

The relative complexity of the VλIII spectrotypes made their analysis more difficult. However, a 369-bp band was found only in the SF of patient D. The three Vλ rearrangements from this band (DS3-T1, DS3-T3 and DS3-T4) each possessed three additions in the V–J junction, resulting in a CDR3 length of 12 amino acids (Table 2 and Fig. 5). Consistent with the spectrotype data, none of the seven Vλ rearrangements from the 366-bp bands of SF and PBL of patient D had any nucleotide deletions or additions within the Vλ–Jλ junction.

Fig. 5.

Nucleotide sequences of VλIII rearrangements from spectrotype bands of patient D. Sequences that are related to the germ-line genes Hsiggll150, Hsiggll295, Humlv318 and Humlv418 are given in a, b, c and d, respectively.

Somatic mutations are frequently observed in the Vλ genes

One of the prime aims of the study, in addition to identifying clonal expansion, was to examine the process that underlies expansion of B cells in the synovium. To achieve this, primers that anneal to sequences within the FR1 were used. One possible problem that could complicate spectrotype analyses is differences in length of the coding region of individual genes. This problem, however, was taken into consideration when interpreting the data because of the availability of the sequencing data. All Vλ rearrangements from SF and PBL spectrotypes had evidence of nucleotide changes compared with the germ-line gene of closest homology (Figs 3–5). Although, generally, more nucleotide changes were observed in the CDR, the changes were not selectively targeted to these regions and substantial numbers of changes were also seen in the FR (Table 3). Further, the calculated R:S values indicated that only six Vλ genes (DP2-4, DS2-B1, DS2-T5, DS2-T1, DP3-3 and DS3-T4) had high values in the CDR, and low in the framework region (FWR), indicative of a classical antigen-driven selection (Table 3).

Table 3.

Average number and frequency of nucleotide changes observed in VλI, II and III rearrangements of similar CDR3 length isolated from synovial fluid and blood spectrotypes of patient D

PCR artefacts

All sequenced Vλ genes were potentially functional. However, 2/32 clones appeared to contain PCR crossover artefacts (Fig. 5). DS3-T1 and DS3-T3 were hybrids of two different genes: Humlv418 and Humlv1042. Both Vλ genes aligned with 85% similarity to Humlv418. However, sequence similarity to position 64 was 95% similar to Humlv418, and from position 64 onwards was identical to the DS1 clones (Fig. 3). It is likely that this is due to a PCR crossover artefact, which occurs when an incomplete product of one amplification cycle serves as a primer for a related sequence [19].

The VλI spectrotype of PBL RF⁺ B cells of patient C consisted of four bands, 375, 378, 381 and 390 bp in length. The 390-bp band was not present in the other VλI spectrotypes (Fig. 1). Vλ rearrangements isolated from this band (CPRF1-T2, and CPRF1-T3) had an insert of 15 nucleotides in FR2. These inserts appeared to be repeats of the adjacent 15 nucleotides in the putative germ-line gene 5′ of the insertion. Nucleotide changes present in this upstream region of the Vλ rearrangement were not present in the inserted segment. This suggests that duplication of the 15 nucleotides occurred before somatic diversification. It is also possible that the insert represents a PCR artefact. An unusually large band, 418 bp in size (compared with the 405 bp length of all other VλIX spectrotypes), was also observed in the SF RF⁺ VλIX (CSRF9) spectrotype of patient C. Sequencing identified the addition of 13 nucleotides in the FR1, rendering the rearrangements non-functional. The inserted sequence showed 100% homology to sequences belonging to a zinc finger protein, a cell surface antigen LAR, and the HLA-Bw57 signal peptide (all members of the immunoglobulin supergene family). The possibility that this insert represents a PCR artefact cannot be excluded, but is unlikely in view of the apparent origin of the insert sequence from an unrelated gene. Pascual et al. [20] suggested the possibility of such gene conversion events in the heavy chain.

Estimation of PCR artefacts showed that an average of 1.4 possible nucleotide changes (0.3%) could have been introduced into the V regions, approximately the predicted error rate of Taq polymerase (1 in 5 × 10⁵). This does not significantly affect our interpretations.

DISCUSSION

The diseased joints in RA patients are infiltrated by lymphocytes, some of which form germinal centre-like cell aggregates that can support proliferation and IgV gene hypermutation [6,7]. Evidence exists for a preferential enrichment in the synovium of B cells expressing κ light chains with longer than expected CDR3 [9,21], leading to suggestions that synovial B cells may be abnormally regulated or expanded by restricted local antigen(s). These studies were limited to randomly selected transcripts encoded by Vκ genes and have not provided a broad insight into the synovial repertoire. In this study we have used a new approach to examine the incidence of light chain gene enrichment to determine whether selective clonal expansion of B cells is widely seen in the diseased synovia. The protocol is based on separating rearrangements on the basis of CDR3 length and study of DNA sequences isolated directly from high resolution gels.

Spectrotypes obtained for rearrangements from SF and matching PBL of three RA patients demonstrated a greater degree of length diversity in the Vλ than Vκ transcripts. Subsequent sequencing revealed that a variety of mechanisms including N-additions was involved in generating Vλ junctional diversity. Junctional variation involving addition and deletion of nucleotides has been observed in Vλ–Jλ junctions in a number of other studies. However, in cases where additional nucleotides were evident within the junction, N-additions were not thought to be involved [22,23]. N-additions observed in this study were predominantly G/C nucleotides, consistent with the activity of TdT, the enzyme responsible for N-addition in heavy chain VDJ junctions. However, TdT is detected in B lineage cells prior to the production of functional heavy chains and not during subsequent light chain gene rearrangements [24]. Possible explanations for the presence of non-germ-line encoded nucleotides at the V–J junctions of light chains may include N-additions being added by below detection levels of TdT, but sufficient for N-additions [25], or a point during the transition from the pro-B cell to the pre-B cell stage at which time there is TdT in amounts sufficient to introduce N-additions into light chain junctions [26]. It has also been suggested that light chain rearrangement could precede heavy chain rearrangement in some pro-B cells [27]. Finally, a TdT-like enzyme, rather than TdT itself, may be responsible for N-addition in light chains [10,28].

Particularly interesting spectrotypes were seen in the SF of patient D. Vλ and Vκ rearrangements from all families in this patient had distinct spectrotypes in the SF, with genes from some families having longer than expected CDR3. SF VλI rearrangements appeared to represent an expanded population of clonally related B cells with long CDR3. The transcripts shared identical Vλ–Jλ junctions (with N-additions), and a number of similar nucleotide changes in the CDR and FR regions. The SF VλII spectrotype also had evidence of rearrangements with long CDR3 and N-additions, but the rearrangements were not clonally related. Enrichment of κ transcripts with unusually long CDR3 in the SF of an RA patient was previously observed [9]. These authors found that 5% of the κ-chain transcripts obtained from the synovial tissue were derived from the same progenitor B cell, but were not due to neoplastic growth. The transcripts studied by Lee et al. [9] belonged to Humkv325, a member of the VκIII family. Interestingly, the synovial fluid VκIII family spectrotype of patient D possessed a unique upper band consistent with N-additions.

Two possibilities could explain the frequent finding of rearrangements with longer than expected CDR3 and N-additions. First, regulation of N-addition may be abnormal, so that light chains with unusually long CDR3 are generated at rates higher than expected [9]. This may reflect abnormality, or polymorphism, in enzymes involved in V(D)J recombination. Such enzymes include the lymphoid-specific recombination-activating gene protein 1 and 2 (RAG1 and RAG2), which confer directionality on the recombination of V(D)J genes, and ligase enzymes that are involved in joining the coding ends. Recent studies of V(D)J recombination have suggested that retaining RAG1/RAG2 after the initiation of recombination leads to stimulation of coding end junction production [29]. Thus, coupling to joining ends allows RAG1/2 complex to direct the joining of appropriate ends and provides the basis for the different processing events. Considering the requirement for RAG1/2 in late stages of V(D)J joining, a key to end joining may be dictated by protein–protein interactions which can be influenced by polymorphism [29]. However, if polymorphism, or diversity, in these proteins is to lead to excessive N-additions in patient D then similar rearrangements would be found in the PBL. This clearly was not the case, suggesting that the enrichment of rearrangements with N-additions in the SF reflects selection in the synovial microenvironment. Second, the presence of numerous nucleotide changes in the rearranged Vλ genes suggests that antigen selection is important in this process. Local expansion of B cell clones in the synovium was first indicated by a study by Munthe et al. [2], in which IgG subclasses in plasma cells of RA synovial membranes were not evenly distributed throughout the tissues, but showed ‘clonal’ distribution in certain villi. More recently, Sclröder et al. [8] directly isolated B cells from lymphoid clusters in the synovium of two RA patients and showed that the isolated sequences were highly mutated and distinct from those expressed in peripheral blood. On the basis of the patterns of mutations and clonal expansion, these authors predicted specific antigen selection. However, the presence of a joint-specific (auto)antigen capable of inducing such expansion has not been verified to date. Moreover, the autoantibody most commonly associated with RA is RF, which recognizes IgG, a ubiquitous autoantigen [30], making it difficult to explain the joint-localized expansion of certain B cells. A previous study associated the presence of N-additions in κ-chains with RF specificity. Martin et al. [11] found that the κ-chains in 55% of RF⁺ B cell hybridomas isolated from RA patients had an addition of a proline and/or glycine residue at the Vκ–Jκ junction, which could only be attributed to N-additions. It was therefore interesting to note that the N-additions in 2/4 VλII and III (DS2-T5, DS3-T4) and 1/4 VλI rearrangements (including the four clonally related sequences DS1-1–4) isolated from the SF of patient D coded for a proline or a glycine residue, respectively. However, CDR3 with N-additions were not restricted to RF specificity in patient C.

The selective recruitment of B cells on the basis of V_H gene expression irrespective of specificity into long-lived peripheral B cells in the mouse was suggested to be mediated through positive selection by internal and/or external antigens [31], in a manner akin to superantigen expansion of T cells on the basis of TCRβ gene expression [32]. Because of the absence of immunoglobulin class switching and somatic hypermutation, this process appeared to be different from the selection of memory B cells in T cell-dependent immune responses. The selective enrichment of rearrangements in the synovium but not blood suggests that the nature of the synovial microenvironment, or the range of self antigens expressed in the synovium naturally, or as a consequence of the inflammation, may lead to the expansion of such B cells. In this respect it was also interesting to note the clear differences between light chain spectrotypes obtained for the three patients. Previous studies, both from this laboratory and others, have shown variations in the repertoire of heavy chain IgV genes in different RA patients [33,34]. These studies suggested that genetic factors may control V_H gene repertoire formation, and that exposure to exogenous or endogenous antigen or RA disease processes does not normally skew the basic inherited pattern of IgV genes [34]. One possible explanation for these findings may be that duplication of V gene as a result of polymorphism dictates expression frequency of the corresponding gene in PBL [35]. Thus, differences in V gene family expression among individuals could derive from insertion/deletion and duplicate polymorphism of genes or gene expression could have been skewed by polymorphic internal antigens [31]. By inference, enrichment of specific B cells in the synovium of RA patients may reflect site-specific selection by self antigens.

In conclusion, we have shown that CDR3 length is highly variable in λ-chains isolated from RA patients. N-additions play a significant role in generating this variation. Furthermore, we provided evidence that several λ-expressing B cell clones possessing longer than expected CDR3 were enriched in the SF of one patient. Further studies are necessary to assess the factors that influence the development of repertoires in the synovium of RA patients, and the specificity of the encoded antibodies.