Sequence analysis of V_4–34-encoded antibodies from single B cells of two patients with systemic lupus erythematosus (SLE)

Abstract

SLE is an autoimmune disease characterized by the presence of autoantibodies against double-stranded (ds)DNA. A large proportion (approx. 40%) of patients with lupus also have increased levels of serum immunoglobulin encoded by the V_4–34 heavy chain gene, which often fluctuate with disease activity, and this gene is utilized by a subset of anti-dsDNA antibodies. In order to probe the nature of the V_4–34-encoded immunoglobulin, B cells were isolated from the blood of two patients with active disease, using the 9G4 MoAb specific for the immunoglobulin gene product. Following cell picking, single-cell polymerase chain reaction (PCR) amplification of cDNA was used to investigate both V_H and V_L genes. Sequences were obtained from B cells synthesizing IgM (n = 10), IgG (n = 4) and IgA (n = 1). For V_H, all were derived from V_4–34 as expected, and the isotype-switched sequences and 2/6 IgM sequences were somatically mutated. In contrast, V_L (12 κ and 3 λ) showed a low level of mutation, possibly indicating secondary rearrangements. The three most highly mutated V_H sequences were associated with unmutated V_L sequences. Analysis of the distribution of mutations revealed only minor clustering in complementarity-determining regions (CDRs) characteristic of antigen selection. The CDR3 lengths of V_H ranged from five to 19 amino acids, and in 3/15 there was evidence of an excess of positively charged amino acids, compared with the normal expressed repertoire. Basic amino acids were also found at the V_L–J_L junctions in 4/15. These findings provide insight into the V_4–34–V_L gene combinations used by B cells in patients with SLE which might have clinical relevance.

INTRODUCTION

SLE is an autoimmune rheumatic disease in which most patients have serum antibodies against double-stranded (ds)DNA. These autoantibodies appear to be involved in the pathological manifestations of disease in skin and kidney [1], and it is generally considered that antibodies of the IgG isotype are the most damaging [2]. There has been considerable interest in analysing the structure of the autoantibodies, initially at the level of predominant idiotypic determinants [3], and subsequently by sequencing the V-genes involved. In order to analyse V-genes, it has been necessary to isolate hybridomas secreting anti-dsDNA antibodies, and, because of the technical difficulties involved, relatively few human antibodies have been described [2,4–6].

Although a variety of V_H genes is clearly used to encode anti-DNA antibodies, a significant proportion appears to use the V_4–34 gene [5,6]. This gene is a member of the V_H4 family, and has been extensively studied, largely due to the availability of a MoAb (9G4) specific for immunoglobulin encoded by the gene [7]. One interesting feature is that the V_4–34 gene is mandatory for encoding IgM cold agglutinins of anti-I/i specificity [8], and that binding to the erythrocyte antigen is mediated via the framework region 1 (FWR1) [9]. For anti-DNA activity, this appears not to be the case, and sequences of basic amino acids in the CDR3 of the V_4–34-derived sequences are more likely to be involved in recognition of DNA [6,10]. Conversely, V_4–34-encoded anti-DNA MoAbs are poor cold agglutinins, indicating that certain CDR3 sequences can inhibit binding via FWR1 [6,9]. In fact, this influence of CDR3 has been formally demonstrated for V_4–34 by interchange of sequences in immunoglobulins expressed in insect cells [10].

Using the 9G4 MoAb, we showed that the V_4–34 gene is utilized by a high proportion (3–11%) of normal immunoglobulin-expressing B cells [11], and this has been confirmed by others [12]. Curiously, serum levels of V_4–34-encoded immunoglobulin in normal serum are not correspondingly raised, accounting for only 0.2% and 0.4% of IgM and IgG, respectively [7]. However, serum levels are strikingly increased in patients with SLE, and not in other autoimmune rheumatic conditions, such as rheumatoid arthritis (RA) or Sjögren's syndrome (SS) [6,13]. Increases are considerably greater than any increase in total immunoglobulin, suggesting a specific stimulation of B cells expressing this immunoglobulin [6]. Furthermore, in some patients, levels fluctuate with disease activity [13]. The association of the raised levels of V_4–34-encoded immunoglobulin with anti-DNA activity is less clear. Although 6–17% of the anti-dsDNA antibodies in patients are derived from the gene, the antibody specificities of the remaining 9G4⁺ immunoglobulin remain unknown [6,13]. Interestingly, similar raised levels of V_4–34-encoded immunoglobulin were evident transiently in patients following certain infections, such as Epstein–Barr virus (EBV) and Mycoplasma pneumoniae [14].

There have been extensive studies of the V_4–34 gene in cold agglutinins [8,15], and in the few hybridomas secreting anti-DNA antibodies established from patients with SLE and other diseases [5,6]. Some of the IgM anti-DNA MoAbs have the ability to kill target B cells in a direct non-complement-dependent manner [16], which could be relevant to the lymphopenia observed in patients. Since the gene appears to be specifically activated in SLE, and to encode antibodies with possible clinical importance, it is desirable to analyse the antibodies at a clonal level. However, human hybridoma technology is too limited, and phage libraries can generate non-physiological pairing of V_H and V_L [17]. In order to gain insight into the V_H–V_L combinations used by antibodies in patients with SLE, and to set the scene for subsequent expression in vitro, we have isolated and analysed the V_4–34 gene, with its accompanying V_L, in single cells of two patients with active SLE. The approach used allowed isolation of RNA, which facilitates analysis of the functional gene, and enables identification of the isotype involved.

PATIENTS AND METHODS

Clinical background and expression of V_4–34-encoded immunoglobulin

Patient 1 (JK), a Caucasian female, presented at age 29 years with arthritis, and subsequently developed a photosensitive rash and pleurisy. By age 36 she also had WHO grade IV glomerular nephritis. Her disease has been generally active and her serological profile included anti-dsDNA, anti-Ro and anti-Sm antibodies. Patient 2 (KC), a Caucasian female, presented at age 16 years with what appeared to be idiopathic thrombocytopenia, for which she had a splenectomy. Three years later she developed fever, arthralgia and lymphopenia, and was found to have a strongly positive anti-nuclear antibody (ANA). In three further years of follow up, she has had persistent serum anti-dsDNA antibodies. The two normal healthy adults were aged 36 years (M) and 52 years (F). Sera were tested for the presence of V_4–34-encoded immunoglobulin by inhibition ELISA using the MoAb (9G4) which is specific for immunoglobulin encoded by this gene segment [7–9]. Levels are expressed as percentage inhibition of binding of the 9G4 antibody to a standard V_4–34-encoded IgM, using sera diluted 1:30 000. The percentage inhibition is expressed as U/ml [6]. Peripheral blood mononuclear cells (PBMC) were isolated from 50 ml of whole blood of patients or normal controls using Lymphoprep (Nycomed, Oslo, Norway), incubated at 37°C for 30 min to remove bound immunoglobulin, and frozen in medium with 10% DMSO until required. For assessment of expression of V_4–34-encoded immunoglobulin by B cells, PBMC were thawed rapidly, washed and exposed to PE-labelled anti-CD19 (FMC63) and biotinylated 9G4, or an isotype-matched rat MoAb control (MC10) followed by FITC–streptavidin. Analysis was carried out in the FACScan.

Isolation of the 9G4⁺ B cell population

Thawed cells (10⁶/ml) were washed and resuspended in PBS with 1% bovine serum albumin (BSA) and exposed to biotinylated 9G4 (30 μg/ml) for 30 min at 4°C. Following further washes with PBS–BSA, cells were incubated with 1 μl of washed Dynabeads M280 Streptavidin (Dynal AS, Oslo, Norway) per 2 × 10⁶ cells for 15 min at 4°C with mixing. Bound cells were washed ×4 with PBS–BSA using a Dynal MPC-1 magnet, resuspended in 10 μl, and kept on ice.

Single-cell picking and RNA isolation

Cells were picked by eye using a micropipette and binocular microscope as described [18]. They were deposited individually onto 1 mm² squares of messenger-affinity paper (MAP) (Amersham, Aylesbury, UK), and processed for reverse transcriptase-polymerase chain reaction (RT-PCR). Following lysis with 4 m guanidinium thiocyanate, they were washed twice with 0.5 m NaCl. After three more washes with NaCl and an 80% EtOH wash, the mRNA bound to the MAP paper was then stored under 80% EtOH at −80°C for up to 1 week.

Preparation of cDNA and amplification and sequencing of V-genes

The EtOH was removed by aspiration, the paper dried, and 20 μl of first-strand buffer containing hexanucleotide primers were added for preparation of cDNA, as described [18]. One quarter of the cDNA was then used for a nested PCR amplification. For V_H, the primary amplification used the 5′-V_H4 leader primer [14] with either the 3′-Cγ100 [14] or 3′-Cμ300 (AGG CAC GTT GTC GAC TTT GTT GCC GTT GGG GTG CTG GAC TTT GCA) primers. For V_L, either the 5′-Vκ leader mix primers [19] with a 3′-Cκ69 (GTT GTG TGC CTG CTG AAT AAC T) primer, or the 5′-Vλ leader mix primers [19] with a 3′-Cλ33 (TTG GCT TGA AGC TCC TCA GAG GA) primer were used, mixed or separately. Conditions were: 95°C for 5 min, 40 cycles of 95°C for 1 min, 55°C for 1 min, 72°C for 1 min, followed by 72°C for 20 min. The secondary amplification for V_H used a 5′-V_4–34 primer [20] with either Cγ10 [20] or Cμ [20]. For V_L, either a 5′-Vκ framework region primer mix [21] with a 3′-Cκ27 (CAA CTG CTC ATC AGA TGG CGG GAA) primer, or a 5′-Vλ primer mix [22] with a 3′-Jλ primer [22] were used separately, with the same PCR conditions apart from a reduction to 30 cycles. Amplified products were gel-purified, cloned by ligation into pGEM-T vector, and transfected into JM109 competent bacteria (Promega, Madison, WI) [22]. Nucleotide sequence analysis was by the dideoxy chain termination method, and alignment was made to EMBL/GenBank and V-BASE [23] sequence directories, using MacVector 4.0 sequence analysis software (IBI, New Haven, CT). At least two independent PCR amplifications were performed on each sample, and a minimum of three sequences analysed.

Analysis of somatic mutational patterns

The number of expected replacement (R) mutations in the CDRs was calculated on a codon by codon basis according to the method of Chang & Casali [24]. The binomial distribution model was used to calculate whether the excess or scarcity of R mutations, in CDRs or FWRs, respectively, were significantly different from those due to chance [24].

RESULTS

Expression of 9G4 idiotope in the blood of patients

The expression of 9G4 idiotope by B cells, and the levels of secreted idiotope in sera, are summarized for patients 1 and 2 in Table 1. For patient 1, two blood samples 2 months apart were obtained (JK1 and JK2). Both patients had low numbers of lymphocytes (normal range 1.5–4.0 × 10⁹/l). Double staining with anti-CD19 and 9G4 revealed a high percentage of 9G4⁺ B cells (46% and 30%) in both patients, with the expected levels (6% and 9%) in normals [11]. The profile of CD19/9G4 staining of PBMC is illustrated for patient 1 in Fig. 1. Although the control antibody (Fig. 1a) stained a few (presumably dead) cells, there was a clear shift to the right with the 9G4 MoAb (Fig. 1b), indicating that approx. 46% of the CD19 population was positive. In contrast, for the normal control (Fig. 1c), the shift with the 9G4 MoAb (Fig. 1d) was less, indicating that 9% of the B cells were positive. Serum levels of V_4–34-encoded immunoglobulin were also high in the two patients (Table 1). To see if this could influence the phenotype of the PBMC by non-specific binding, a sample of 9G4⁻ PBMC was incubated with serum from patient 1, washed, and again analysed for expression of 9G4 idiotope. No increase in expression was observed (data not shown), indicating that the increased percentage of 9G4⁺ B cells is likely to arise from B cells synthesizing immunoglobulin encoded by the V_4–34 gene. However, the raised percentages of 9G4⁺ B cells appeared to be associated with an increase in serum levels, and the slight shift from the ordinate in Fig. 1b suggests that some 9G4⁺ immunoglobulin may have bound to both CD19⁺ and CD19⁻ cells in patient 1's blood.

Table 1.

Expression of V_4–34-encoded immunoglobulin on B cells and in serum

Fig. 1.

Expression of the V_4–34-associated idiotope (9G4) by B lymphocytes from patient 1 (JK) with SLE, and from a normal adult (N). Peripheral blood mononuclear cells (PBMC) from patient 1 (a,b) or a normal individual (N) (c,d) were treated with PE–anti-CD19 together with biotinylated 9G4 (b,d) or MC10 control (a,c) and FITC–streptavidin. Antibody-coated cells, gated on live lymphocytes by size and granularity, were enumerated by flow cytometric analysis. The control antibody for 9G4 was an isotype-matched rat MoAb, MC10, of irrelevant specificity. 9G4⁺ B cells are in the upper right quadrant. For patient 1, a few cells staining with anti-CD19 and the MC10 control MoAb have been subtracted from the 9G4⁺ population. Percentages of total gated lymphocytes are shown in each quadrant. Percentages of 9G4⁺ cells in the CD19⁺ population are in parentheses.

Analysis of single 9G4⁺ cells

For each patient, cells were observed which were surrounded by magnetic beads, indicating reactivity with the 9G4 MoAb. Individual cells were picked from the centre of the clusters, and PCR amplification for V_H and V_L was carried out. From 60 single cells, 15 V_H/V_L pairs were identified. The 15 pairs all had productively rearranged V_H and V_L genes, possibly reflecting the use of RNA as a source material. Nucleotide sequences have been deposited in the EMBL database (Y17927–Y17956). Deduced amino acid sequences from the two patients are shown in Fig. 2 (patient 1) and Fig. 3 (patient 2). All the V_H sequences were derived from V_4–34, confirming the specificity of the 9G4 MoAb, and the selection process. Light chains differed, with most of the pairs (12/15) derived from Vκ genes, and only 3/15 from Vλ (Figs 2 and 3).

Fig. 2.

Deduced amino acid sequences of the V_H (V_4–34) and V_L pairs in single cells from patient 1 (JK). Comparisons have been made with the V_4–34 germ-line gene, or with the closest V_L germ-line gene. Upper case, replacement mutations; lower case, silent mutations. Replacement mutations in J sequences are underlined. Arginine (R) and lysine (K) residues in CDR3 or J sequences are in bold.

Fig. 3.

Deduced amino acid sequences of the V_H (V_4–34) and V_L pairs in single cells from patient 2 (KC). Comparisons have been made with the V_4–34 germ-line gene, or with the closest V_L germ-line gene. Upper case, replacement mutations; lower case, silent mutations. Replacement mutations in J sequences are underlined. Arginine (R) and lysine (K) residues in CDR3 or J sequences are in bold.

V_H gene analysis

V_H sequences revealed that 10/15 were derived from IgM, 4/15 from IgG, and 1/15 from IgA (Figs 2 and 3). The IgG constant region sequences were from either IgG1 or IgG2. For patient 1, it was quite surprising to find that 50% of the sequences were obtained from B cells which had undergone isotype switching (Fig. 2), since IgG⁺ B cells might be expected to be in only low numbers in blood lymphocytes.

Among the IgM-derived sequences, 7/10 were in germ-line configuration, with the replacement (Arg (R) to Ser (S)) amino acid at the V–D junction in KC13H due to N-addition. The remaining 3/10 sequences from IgM showed a low level of somatic mutation of between 0.7% and 1.9%. In contrast, the level of mutation in the isotype-switched sequences was higher, ranging from 1.5% in JK3H to 6.8% in KC17H (Figs 2 and 3). Analysis of the distribution of mutations indicated no significant clustering of replacement amino acids in CDR1 and CDR2, characteristic of antigen selection (Table 2). However, one sequence (JK10H) did show more R mutations in CDRs than would be expected by chance.

Table 2.

Analysis of distribution of somatic mutations in V_Hand V_Lgene segments of patients 1 (JK) and 2 (KC)

The CDR3 sequences varied from five to 19 amino acids, and confident assignment to D-segment genes, based on identity of at least 10 sequential nucleotides [25], was possible only for 6/15 sequences. Assignment is shown in Table 3, with the reading frames indicated. All the identified D-segment genes have been found to be used commonly in normal B cells, in those reading frames [25]. In 3/15 sequences (KC25H, JK19H and JK28H) there was an excess of basic amino acids (Arg and Lys) over acidic amino acids (Asp and Glu) in CDR3 (Figs 2 and 3). Only one of the basic amino acids (Arg (R), at position 6 in the CDR3 of JK19H) arose from the D-segment gene. The remainder were derived either from an unassignable D-segment gene (KC25H), or from N-additions (Arg at position 11 in JK19H, and both Arg residues in the CDR3 of JK28H). In addition, there was an increased positive charge arising from a change from Gln to Arg in the J_H sequences of JK47H and JK10H, the latter due to usage of the J_H2 gene.

Table 3.

D-segment gene usage in single V_4–34-positive B cells

V_L gene analysis

V_L sequences were also all potentially functional, and were derived mainly from Vκ (12/15), with only 3/15 from Vλ. Deduced amino acid sequences for the two patients (Figs 2 and 3) indicated that the Vκ families used were Vκ1(6), Vκ3(4) and Vκ4(2). Individual gene segments involved were as expected from the normal expressed repertoire: for Vκ1, 5/6 were derived from the commonly used O12/02 gene, and 1/6 from the less common A20 gene [26]. For Vκ3, 2/3 used the common A27 gene, and 1/3 the Vg gene. The Vκ4 gene involved was the relatively common B3 gene. For Vλ, the families involved were the common Vλ1(2) and Vλ2(1), and the gene segments within the families were Vλb and Vλ2a2, both well represented in the normal expressed repertoire [27]. It appears therefore that V_L gene usage by the B cells reflects that expected of normal B cells.

Most of the sequences (12/15) were in germ-line configuration, and a further sequence (JK28L) had only a single nucleotide substitution. Significant levels of somatic mutation were only evident in two sequences, JK6L and JK47L, reaching 2.5% in the latter. Both these were IgM⁺, and JK47 was also mutated in V_H (Fig. 2). Surprisingly, all the isotype-switched cells had unmutated V_L sequences, indicating no correlation with patterns in V_H. In one case only (JK47), mutations in V_L were clustered in CDR1 and CDR3, with a pattern indicative of antigen selection. In contrast, the V_H sequence of JK47 had mutations only in FWR3 of V_H (Fig. 2).

The CDR3 sequences of the V_L genes showed evidence of N-additions in 7/15 sequences. There was apparent generation of basic amino acid residues at the CDR3/Jk junction, either by N-addition or mutation in JK3L, JK19L, JK28L, KC8L and KC3L (Figs 2 and 3).

DISCUSSION

The raised levels of V_4–34-encoded antibodies in sera of patients with SLE pose the question of the relevance to the disease process. It is difficult to approach this at the serological level, especially since the encoded immunoglobulin can bind to a range of antigens, in some cases via the FWR1, the site of the 9G4 idiotope [8,9]. At the cellular level, it is clear from our phenotypic analysis that, although the total numbers of B cells were decreased in the patients, the proportion of remaining B cells which utilize the V_4–34 gene was vastly increased. To analyse the profile of V_4–34-encoded immunoglobulin, we used the technology of single-cell picking, using the 9G4 MoAb for sorting of the target population [7,11]. We opted to analyse RNA so that we could probe for isotype-switched antibodies, which are more likely to have a major role in the pathology of SLE. The approach allowed us to identify the natural V_H–V_L pairs in V_4–34-encoded immunoglobulin in B cells of patients. One surprise was that 5/15 sequences had undergone isotype switching. Although transcripts of V_4–34-Cγ had been identified in normal blood cells, they were infrequent and appeared to derive from a few expanded clones [28]. Also, in tonsil cells, analysis of the V_H4 family had indicated that usage of V_4–34 declines during class switching [29]. In SLE, analysis of serum has revealed that a proportion of the raised V_4–34-encoded IgM and IgG has anti-DNA activity [6]. In addition, the V_4–34 gene was found to be utilized by several cloned anti-DNA antibodies established from patients [5,6]. Sequence analysis revealed that both IgM⁺ and IgG⁺ MoAbs have basic amino acids in the CDR3 sequences, providing candidates for interaction with DNA [5,6]. In one IgG⁺ case (D5), the Fab was expressed in phage, and manipulation of the construct revealed that the CDR3 was essential for recognition of DNA, whereas the contribution of somatic mutations in V_H was insignificant [10]. Unexpectedly, somatic mutations in V_L were found to have a major influence on binding [10].

It appears therefore that basic amino acids, which are not found in CDR3 sequences of normal IgM⁺ B cells in blood [30], may be involved in recognition of DNA. If so, it is possible that our 3/15 sequences (JK19H, JK28H and KC25H), each of which have two Arg residues in CDR3 with no acidic residues, may have anti-DNA activity. The contribution of these candidate sequences to antibody activity is being analysed using expression in phage [10]. However, anti-DNA antibody activity does not account for all the gene product [6], and the nature and function of the remaining V_4–34-derived immunoglobulin are not yet known.

With regard to V_L, although the numbers investigated so far are small, it is clear that, whereas Vκ (12/15) may be preferred to Vλ, there is no particular preference for individual V_L gene segments. The spectrum of V_L gene usage is similar to that seen in the normal expressed repertoire and also in monoclonal IgM cold agglutinins [8]. In some V_L sequences, there are Arg residues at the CDR3-J_L boundaries, but these have also been observed in V_L from cold agglutinins [8], and are therefore unlikely to be involved in anti-DNA specificity.

In terms of somatic mutations, the observation that most of the IgM-derived sequences were unmutated, but that a minority had quite extensive mutational levels, has been reported for normal B cells [31]. It is likely that the mutated sequences are memory cells [31]. One unexpected feature was the asymmetry of somatic mutations, with V_H of IgG generally highly mutated, but the paired V_L sequences having few or no mutations. This might suggest that the V_L sequences have arisen by a secondary recombination event, a mechanism known as ‘receptor editing’ [32]. Evidence for selection of mutated sequences by antigen relies on identification of clustered replacement mutations in the CDR2. Since it is not possible to include the complex CDR3 in the analysis, it is necessarily limited. In our sequences, there was no clear indication of such clustering in V_H, and only one V_L sequence showed this feature. Clearly, features of single antibodies in patients with SLE need to be compared with those of normal individuals, and there is insufficient information available so far. For functional relevance, it is essential to combine this approach either with single-cell culture or with expression of synthetic antibody fragments [10]. The technology is now available to gain an overview of autoantibody structure and function.