Idiotype-Specific Th Cells Support Oligoclonal Expansion of Anti-dsDNA B Cells in Mice with Lupus

Systemic lupus erythematosus (SLE) is a debilitating systemic autoimmune disease that is associated with genetic and environmental influences, with involvement of multiple organs, such as the skin and kidneys (1, 2). SLE is marked by a Th cell–dependent B cell hyperresponsiveness, with frequent germinal center reactions, hypergammaglobulinemia, and high levels of highly mutated, affinity-matured IgG autoantibodies (1–6). Anti- dsDNA Abs are a serological hallmark of SLE that mediate, at least in part, the nephritis that marks the course of this disease in both humans and mice (1, 2).

In other mouse models of lupus, MRL/lpr and BWF1 mice spontaneously develop anti-dsDNA and nephritis; however, anti-dsDNA autoantibodies also can be induced [e.g., by vaccinating non–lupus-prone strains with dsDNA and adjuvants (7, 8)]. In both spontaneous and induced models, the anti-dsDNA Ig variable H chain (IgVH) regions often express basic amino acids like arginine (R) and lysine (K), or asparagine (N), in Ag-binding CDR and FR3 regions (9). The presence of several arginines in CDR3 is relatively rare (10, 11) but is important for DNA binding; the positively charged arginine can bind to the phosphodieseter backbone, as well as donate up to five H bonds (9–12).

Autoreactive B cells experience sustained signaling through their BCR, become anergic, and eventually undergo apoptosis, as was described for anti-dsDNA B cells (13). Nevertheless, such cells may be rescued, activated, and expanded if provided with Th cell help (13), a feature that also was found for anergic B cell responses of other autospecificities (14, 15). However, in contrast to B cells, it is unclear what specificity Th cells may have in cognate interactions with dsDNA-specific B cells. Ab Ids are a plausible candidate because both lupus-prone mice (16–19) and SLE patients (20–22) have clearly measurable Th cell responses toward pathogenic anti-dsDNA Abs, suggesting that Id determinants play a role in pathogenesis. In such experiments, APCs, such as dendritic cells, present somatically mutated V-region determinants (Id peptides) to Th cells. Further, anti-Id Th cell responses increased with disease severity, and disease was aggravated by injection of Id peptide (17).

It has been clear for some time that somatically mutated Id peptides can serve as cognate Ags for Th cells (23, 24) and that this responsiveness is restricted by tolerance to germline V regions (25, 26). Moreover, it was shown that individual B cells can present endogenous Id peptide on MHC class II molecules to Th cells and that such Id⁺ B cells can collaborate with Id-specific Th cells (27–30). In this interaction, B cells can undergo the germinal center reaction, provided that ligands bind the BCR (31). Such Id-driven Th cell–B cell collaboration can also cause expansion of autoreactive B cells, secretion of autoantibodies (26, 31), and autoimmune disease (32, 33). In the latter studies, mice were double transgenic (DTG): B cells expressed a transgenic (TG) λ2 L chain (Id⁺), and T cells expressed an MHC class II (I-E^d), Id peptide–restricted TCR transgene. In these mice, we observed a marked, but incomplete, negative selection of Id-specific thymocytes, a progressive expansion of low-frequency Id-specific T cells, an ongoing collaboration between Id-specific Th cells and Id⁺ B cells, hypergammaglobulinemia, and autoantibodies, including high titers of anti-nuclear Abs (ANAs). In this study, we show clonal expansion of anti-dsDNA B cells in such mice, finding a marked oligoclonality, with clonal expansions of B cells with mutated VH regions and arginine-rich CDR3 regions.

Mice were TG for both the λ2³¹⁵ Ig L-chain derived from the MOPC315 myeloma, as well as an αβ TCR transgene specific for the Id(λ2³¹⁵) peptide presented on I-E^d MHC class II molecules (32, 33). The TCR specificity is toward the λ2³¹⁵CDR3 motif that includes three replacement mutations (CALWFRNHFVFGG) (23, 24).

DTG F1 mice were from TCR transgenic (homozygous females) × Id⁺ transgenic (homozygous males); all offspring are DTG. Alternatively, to generate littermate controls (Fig. 1), offspring from TCR transgenic (hemizygous females) × Id⁺ transgenic (hemizygous males) were analyzed. Both TG strains are on a BALB/c background (>20 backcrosses). The Norwegian Animal Research Authority approved the experiments.

Mice were weighed and monitored for disease, and hind leg saphenous vein blood samples were drawn every 10 d for autoantibody analysis. Diluted serum samples were analyzed for the presence of anti-dsDNA Abs or other autoantibodies, as previously described (32, 33), HEp-2 cells (Immuno Concepts) and Crithidia luciliae (Binding Site) served as substrates; bound autoantibodies were stained with Alexa Fluor–488/546 goat anti-mouse IgG [F(ab′)₂], Alexa Fluor–488/546 goat anti-mouse IgG2a, and Alexa Fluor–488/546 goat anti-mouse IgG1 (Molecular Probes). Hybridomas were screened by an ELISA-based measurement of Abs toward confluent ethanol-fixed HEp-2 cell monolayers. Bound Abs were detected by anti-mouse IgG (Fc-specific) peroxidase conjugate (Sigma-Aldrich).

Fetal calf DNA (Sigma-Aldrich) was digested into ≈500-bp fragments (DNase) and labeled with Alexa Fluor 488 with the ULYSIS Nucleic Acid Labeling Kit (Molecular Probes), according to the manufacturer’s protocol. CpG 2006–fluorescein (tlrl-2006f) with human TLR9-binding motifs was from InvivoGen. Cell suspensions were incubated for 10 min at 37°C with DNase I from bovine pancreas (Sigma-Aldrich), washed four times, stained with DNA, and counterstained with fluorochrome-coupled anti-B220 (RA3-6B2; Southern Biotech), anti-CD19 (1D3), and anti-CD4 (RM4-5; both from BD Pharmingen). DNA–Alexa Fluor 488 and CpG 2006–fluorescein were tested on murine anti-dsDNA lymphoma (positive control; A. Funderud, K. Aas-Hanssen, B. Bogen, and L.A. Munthe, unpublished observations) and A20 murine B cell lymphoma cells (negative control, ATCC TIB-208). Cells were fixed with paraformaldehyde before acquisition. DTG splenocytes or BALB/c controls were cultured with Th2 cells from Id-specific TCR-TG SCID mice, as described (31). BrdU was added on day 3, and cells were stained on day 5 with the BrdU-APC Staining Kit (BD), including surface stains for B220, CD4, and CD19. Cells were acquired on a FACSCalibur (BD) and analyzed with FlowJo X software (TreeStar). The Mann–Whitney U test was used to compare groups.

Spleens, lymph nodes, kidneys, and blood were collected from euthanized mice, and cell suspensions and serum samples were made as previously described (32, 33). Hybridomas were generated by mixing splenocytes with OURI cells (variant of X63-Ag8.653), followed by the drop-wise addition of polyethylene glycol (Roche). Hybridomas were selected with hypoxanthine/aminopterin/thymidine-supplemented RPMI 1640 medium (Sigma-Aldrich) and cloned by limiting dilutions. Clones were screened for autoantibodies, as above. Sixteen hybridomas were further subcloned.

mRNA isolation was performed on washed pelleted cells with a Dynabeads mRNA DIRECT Kit (Life Technologies), according to the manufacturer’s recommendations. Reverse transcription was performed using a First-Strand cDNA Synthesis Kit (Amersham Biosciences), according to the manufacturer’s protocol. Reverse transcription was performed using the Not I-d(T)₁₈ primer (First-Strand cDNA Synthesis Kit; Amersham Biosciences). cDNA was amplified by PCR using PfuTurbo DNA Polymerase (Stratagene) and a mixture of 5′ H-chain FR1 region degeneracy primers, together with a mixture of 3′ H-chain constant (C)-region primers (34) (Sigma-Genosys). The PCR products were run on a 1.5% agarose gel, and gel-purified products of predicted size (≈400 bp) (QIAquick Gel Extraction Kit; QIAGEN) were ligated into pGEM-T Easy vector (Promega) and used to transform One Shot chemically competent E. coli TOP10 cells (Life Technologies). Plasmid DNA was prepared from overnight cultures (Wizard Plus SV Minipreps DNA Purification System; Promega), and colonies found to contain an insert were sequenced using a primer to the T7 promoter of the pGEM-T Easy vector (GATC Biotech). At least four colonies were sequenced for each sample.

The following control data sets were downloaded from the National Center for Biotechnology Information (NCBI) Web site (http://www.ncbi.nlm.nih.gov/nuccore/): control VH sequences from BALB/c mice, retrieved with the search term “V region immunoglobulin heavy chain Balb”; sequences derived from splenic L2-TG mice IgG⁺ B cells, as deposited (35); anti-DNA hybridomas from BWF1 mice, as deposited (36); and neonatal liver B cell IgVH sequences, as deposited (37).

Sequences were analyzed with the International ImMunoGeneTics (IMGT)/HighV-QUEST version 1.1.2 or IMGT/HighV-QUEST version 3.2.30 tools (http://imgt.cines.fr) and compared with the IMGT/V-QUEST reference directory release: 201310-4 (March 14, 2013) (38). IgVH region family identification and clonality analysis were performed using the statistics module of IMGT/HighV-QUEST. Statistics were reported only for unique sequences (38). Translated amino acid sequences were analyzed further in Excel (Microsoft); amino acids in individual positions were counted with the “countif” function. Phylogenetic trees were constructed and visualized with the iTOL program (http://itol.embl.de). Data sets were compared with the two-sided Fisher exact test, two-sided χ² test, or the Wilcoxon signed-rank test.

Clonally related sequences in the green and blue families of mouse 21 were tested for evidence of Ag-driven selection in IgVH sequences with Baseline Version 1.1, focused selection statistics (http://selection.med.yale.edu/baseline/) as described (39).

We analyzed sera from DTG mice (32, 33) and found IgG1 and IgG2a anti-dsDNA autoantibodies from age 6 wk (Fig. 1A). High-titer autoantibodies stained nuclei with a predominantly homogenous pattern, and anti-dsDNA autoantibodies were found toward the kinetoplast of C. luciliae (Fig, 1B, 1C). In addition, DTG mice had variable autoantibodies of lower titer toward a range of other autoantigens, including cytosolic Ags (Fig. 1B, 1C) and extracellular Ags (data not shown) (32, 33). This feature resembles that found in SLE patients who often have autoantibodies with other autospecificities, in addition to high-titer anti-dsDNA. Thus, an interindividual variance was seen in addition to the dominant anti-dsDNA reactivity, a finding that was reminiscent of human SLE serology.

We proceeded to investigate whether B cells in DTG mice could bind DNA. B cells from spleens and lymph nodes of >115-d-old DTG mice with symptoms of disease were isolated in the presence of DNase, washed, and tested for binding to Alexa Fluor 488–conjugated dsDNA or CpG 2006–fluorescein (the latter with human TLR9-binding motifs). An increased frequency of splenic and lymph node B cells from DTG mice bound dsDNA (≈4 and 7% respectively) or CpG 2006 DNA (≈17%) (Fig. 2A, 2B, Supplemental Fig. 1). Id-specific Th cells stimulated the blastogenesis and proliferation of Id⁺ B cells from DTG mice (Fig. 2C, Supplemental Fig. 1), including the proliferation of the dsDNA-specific subset of B cells (Fig. 2D).

To investigate the clonal relationship of the observed anti-dsDNA B cells, we selected anti-dsDNA hybridomas and compared IgVH sequences of hybridomas with those derived from lymphoid organs from the same mice. A total of 55 hybridomas was generated by conventional means from three mice undergoing systemic autoimmunity, selecting IgG1 and IgG2a producers that gave bright signals. The majority of these hybridomas secreted ANA autoantibodies with a homogenous nuclear pattern (data not shown). Sixteen hybridomas derived from mouse numbers 21, 15, and 5 were subcloned, and the VH was sequenced. IgVH sequences were obtained from lymph nodes and spleens of mice 21 and 5. Only a few sequences were obtained from mouse 15.

Altogether, >600 IgVH sequences were obtained; 176 of these were unique. Analyzing the V regions (excluding CDR3), we found that these 176 sequences had an average of 9.6 mutations (Fig. 3A). We compared these sequences with data sets downloaded from the NCBI Web site: IgG IgVH from an Id⁺λ2 TG strain (on the BALB/c background) generated by another group (35), called L2-TG in this article; productive IgVH sequences from BALB/c mice; and IgVH derived from IgG⁺ anti-DNA hybridomas from BWF1 mice, as published (36). We found that the mutation rate in the DTG sequences was increased and significantly different from L2-TG, BALB/c, and anti-DNA hybridomas (Fig. 3A, p < 0.0001, Wilcoxon signed-rank test).

To investigate clonality, we plotted VH family usage in DTG and L2-TG mice and found that DTG mice had a skewed VH family usage (Fig. 3B). These families also were increased compared with the BALB/c dataset [note that distribution of L2-TG differs from the distribution of BALB/c, presumably related to the preponderance of marginal zone B cells in L2-TG mice (35, 37)].

These results suggested that the sequences from DTG mice could be oligoclonal. To confirm this, a clonality analysis was performed (IMGT/V-high, statistics module). This module analyzes the sequences that can only be attributed to single alleles (excluding sequences that can belong to more than one allele). A total of 47% of the 137 DTG single-allele sequences that were analyzed in this statistics module were clonally related; these VDJ sequences belonged to sets that had identical CDR3 sequences but differed in the V regions (Fig. 3C). Neonatal L2-TG sequences were chosen as negative controls (representing sequences prior to any Ag selection and clonal evolution); this data set, as well as the unbiased BALB/c sequences, had minimal clonal relatedness.

The analysis above demonstrated that nearly half of the sequences in the DTG data set could be grouped into a few clonally related families of B cells. In terms of the anti-dsDNA specificity, positively charged amino acids and, especially, arginines (Rs) are frequent in anti-dsDNA mAbs, representing a hallmark for these Abs (9). CDR3 junctions most often begin with the amino acids cysteine (C), alanine (A), and arginine (R) (CAR): AR are the first two CDR3 amino acids. Discounting this first arginine (in position 2), a single additional R in CDR3 may be sufficient for binding to dsDNA, and multiple substitutions to R were shown to have an additive effect on affinity (9). Disregarding this first CAR arginine, the DTG CDR3 sequence was plotted from position 3 in Wu-Kabat–like plots (Fig. 4A). The anti-dsDNA hybridomas expressed one to four additional arginines (average 2.5) in their CDR3s. This feature was not found in the negative-control data sets (Fig. 4A). When analyzing the sequences from the lymph node and spleen, similar results were found; the sequences had, on average, 1.1 additional CDR3 arginines compared with 0.3 arginines in BALB/c sequences (Fig. 4B, data not shown). These CDR3 arginines had similar CDR3 positional distributions as did the hybridomas from the same mice, a phenomenon that was especially clear in mouse 5. The BALB/c control sequences had one additional R in CDR3 of every third sequence, whereas 2 Rs were found in <0.1% of the sequences (2/2231). In comparison, two or more CDR3 arginines were found in 34% (54/161) of the DTG lymph node/spleen/kidney sequences (Fig. 4C, p < 0.0001, two-sided χ² test).

Arginine-rich CDR3 sequences are hallmarks of anti-dsDNA specificity, and very few (<0.1%) of the BALB/c sequences had two or more additional CDR3 arginines. Therefore, we chose at least two additional CDR3 arginines as a cutoff to identify probable anti-dsDNA–reactive families, which were verified by clonal relatedness to anti-dsDNA hybridomas. Hence, we identified clonally related families with anti-dsDNA–binding properties. Phylogenetic trees were generated for all unique sequences in the mice. We identified two large families in mouse 21 (Fig. 5A, Supplemental Fig. 2). The first, colored blue (blue family), had 22 members with CDR3 coding two or three arginines (mean 2.1), and it included six hybridomas (with verified anti-dsDNA specificity) in addition to the sequences from the lymphoid organs. The second family (green family in Fig. 5) had 18 members and did not include hybridomas, but it had as many as two or three arginines (average 2.8) in the CDR3 sequences, strongly suggesting anti-dsDNA reactivity. Some of these members also were reidentified [i.e., found repeatedly (n = 17)], suggesting clonal expansion of B cells with identical VH. These two families (green + blue) represented 33% of the unique (40/120) sequences, thus demonstrating a marked clonal expansion. In addition, we found five other sequences with more than two arginines, resulting in an anti-dsDNA total of 37% (45/120) in this mouse (Fig. 5B, Supplemental Fig. 2). Further, mouse 21 had a clonally related family (19/120), with the conserved CDR3 motif YYGSS associated with anti-DNA/nucleosome binding (40). This family expressed YYYGSSR (Fig. 5A, left panel; Supplemental Fig. 2).

In mouse 5, a family called red contained 16 members (including three anti-dsDNA hybridomas) and lymph node/spleen sequences with three or four arginines in CDR3 (Fig. 5B).

We found similarities between CDR3 of our anti-dsDNA VH sequences and previously published sequences from lupus-prone mice. Examples are shown in Fig. 5C.

The clonal relationships of sequences within the blue, green, and red families were confirmed by VDJ analysis, where members had identical D and J genes and had the same CDR3 length (Supplemental Table I). The blue and red families had nontemplated nucleotide additions, allowing the D gene segment to be read in an altered reading frame, introducing one arginine to CDR3 (both with arginine in position 5), an event that probably was acquired in the bone marrow during VDJ rearrangement. Somatic mutation accounted for a second arginine in position 7 in all members of the blue family. Members of the red family had two additional arginines (position 4 and 6) due to somatic mutation (NCBI submission #1691092).

Using the hybridomas to identify anti-dsDNA B cells in DTG mice, we found that these IgVH sequences were characterized by increased numbers of arginines in CDR3. On average, these B cells had 1.7 arginines caused by somatic mutation (72 mutations to arginines compared with 88 mutations to other amino acids, Fig. 6A). This very high mutation rate to arginine was significantly different from the mutation to arginine in CDR3 of BALB/c controls (101 mutations to R, 1247 mutations to other amino acids; BALB/c versus DTG, p < 0.0001, two-sided Fisher exact test).

It was of interest whether the B cells with arginine-rich CDR3 also expressed mutations to arginine or other amino acids in CDR1 and CDR2. Fig. 6 shows the frequency of amino acid mutations in the anti-dsDNA DTG data set compared with BALB/c control (NCBI Web site). In general, lower mutation rates were observed for CDR1 and CDR2, but mutations to arginines also were increased compared with BALB/c in CDR1 and CDR2. To test whether these mutations were evidence of selection within the V region (disregarding CDR3), we used Bayesian estimations of Ag-driven selection in Ig sequences (39). We found that clonally related members of the blue and green families had evidence of significant positive selection, showing higher than expected replacement mutations in CDR1 and CDR2 compared with FR regions (p = 0.03 for both). The red family, with fewer members, had the same tendency, but the data did not reach significance.

In this study, we demonstrate a marked clonal expansion of prototypical anti-dsDNA B cells in DTG mice developing a systemic autoimmune disease with SLE-like features. Nearly 50% of B cells, identified by IgVH sequencing, were clonally related. Anti-dsDNA B cells accounted for up to 30% of the B cells (detected by flow cytometry) and nearly 40% of the IgVH (clonally related families with more than two arginines in CDR3). Anti-dsDNA B cells were stimulated to proliferate by Id-specific Th cells. The B cells expressed mutated isotype-switched IgG with high levels of arginines in CDR3, and evidence was found for Ag-driven selection of V-region sequences. CDR3 had sequence homologies to anti-DNA CDR3 sequences from classical mouse models of lupus.

Arginine amino acids in CDR3 characterizes anti-dsDNA Abs (10, 11). The arginine side chain can form bidentate interactions with G–C bp (12), as well as electrostatic interactions with phosphate groups. One additional CDR3 arginine can be sufficient for DNA binding (11). The results obtained demonstrate that Id-driven T cell–B cell collaboration could support clonal expansion of arginine-rich classical cationic anti-dsDNA Ab–expressing B cells by two mechanisms: selection of B cells with a priori arginines in CDR3 and further expansion of such B cells, followed by somatic hypermutation. The clonal families either had alternative reading frames of the D region, resulting in a CDR3 arginine (red and green families), or used a D region coding for one arginine (blue family). Clonal expansion was accompanied by further mutations to additional arginines.

Our previous studies demonstrated that B cell Ag presentation (of Id) in itself is insufficient for the generation of the germinal center reaction: B cells also require BCR ligands for the response (31). Contingent upon the presence of DNA, naive anti-dsDNA B cells with one or two CDR3 arginines could be helped by the low-frequency Id-specific Th cells in the mice to isotype switch to IgG and accumulate one or two additional CDR3 arginines by somatic hypermutation. Reciprocally, it is possible that such Th cell–B cell collaboration could play a central role in the expansion of the Id-specific Th cells. The early debut (6 wk) of anti-dsDNA Abs may be explained by such Th cell–B cell collaboration. In comparison, MRL/lpr mice also develop anti-dsDNA responses from week 6 (average 9–10 wk) (41), whereas BWF1 mice seroconvert later, from 4 mo of age (7).

SLE has been associated with incomplete clearance of apoptotic cells. In fact, the level of nucleosomes in the circulation is significantly increased in patients with active disease. Hence, nucleosomal material, including dsDNA, is accessible for BCR ligation (42). However, B cells that are stimulated through the BCR in the absence of T cell help develop into anergic cells and apoptose (reviewed in Ref. 43). Ag-specific Th cells can negate anergy and support conventional immune responses, including germinal center reactions, development of plasma cells, and autoantibody secretion (13–15, 44). Hence, anti-dsDNA B cell responses are dependent on help from Th cells, and both Th2 cells and follicular Th cells have been directly linked to SLE (5, 6, 44).

Yet, because Th cells cannot be activated in a cognate manner by DNA, the drive for this Th cell–B cell interaction requires an additional level of complexity. Conceptually, anti-dsDNA B cells can either present peptide derived from endocytosed DNA-bound nuclear self-proteins or present endogenously derived (B cell–intrinsic) peptides to Th cells. A third possibility is a combination of these two concepts: Th cells could cross-react to cationic peptides derived from DNA-binding proteins, regardless of derivation, including both nuclear self-proteins and anti-DNA Ids (see later discussion).

In the first situation, endocytosed DNA-binding nuclear proteins, such as histone determinants, could serve as Th cell Ag (45). In the second situation, Th cell help of anti-dsDNA B cells would be dependent upon presentation of endogenous secretory pathway Ags. Examples include self-proteins derived from the secretory pathway, such as unique Ids of the B cell clone (27, 28, 46), as shown in this study, presentation of peptides derived from viral proteins synthesized by B cells (47), or alloantigens in chronic graft-versus-host models of SLE that stimulated oligoclonal expansion of arginine-rich anti-dsDNA B cells (48). In this study, anti-DNA B cells bind DNA while receiving allospecific T cell help.

The current results demonstrate that Id-specific Th cells can reproduce classical anti-dsDNA B cell responses. However, it is interesting to note that DNA-associated proteins and anti-dsDNA Abs (CDR3) both have positively charged DNA-binding motifs. It is possible that the seemingly dissimilar Th cell specificities for cationic CDR3 Ids and cationic DNA-binding peptides could constitute networks of molecular mimics for Th cells in SLE. For example, histone H4(71-94) was described to be a Th cell epitope in lupus mice and in SLE patients. This peptide contains the following DNA-associated motif: HAKRKTVTAMD (45). The sequence is comparable to CDR3 from several previously described anti-DNA CDR3 peptides: for example CARRRTGTAYY [Lx-163, submitted by (49)]. If Th cells respond to cationic Id peptides of such B cells, it could be hypothesized that established Th cell responses potentially cross-react to peptide mimics from other DNA-binding proteins, including histones.

In terms of this hypothesis, it is notable that most DNA-binding proteins are ubiquitously expressed; T cell tolerance mechanisms are likely to exert a strong tolerizing pressure to delete histone/DNA-binding protein-specific T cells. In contrast, Ids are unique, private proteins of a particular B cell clone. Thus, bone marrow emergent B cells with anti-dsDNA–binding V regions and positively charged CDR3 could fortuitously encounter rare Th cells specific for the unique Id of that B cell. B cells express relatively high levels of Id⁺ Ig; Id peptides were among the first to be eluted from MHC class II molecules of B cells (46). Hence, B cells process and present their endogenous BCR as short Id peptides on MHC class II to T cells that may recognize V-region Id peptides (27–30). It was shown that V regions derived from anti-dsDNA B cells of normal mice (50), as well as BWF1 mice (17, 18), contain Id CD4⁺ T cell epitopes. Further, Id-specific CD4⁺ T cells have been found in humans suffering from SLE (20–22).

The results demonstrate that Id peptides have the potential to constitute Th cell Ag in anti-dsDNA B cell responses in lupus. Another line of evidence implicated Id-specific Th cells in lupus: it was suggested that pathogenic autoreactive Id⁺ anti-DNA Abs are regulated and counteracted by anti-Id Abs. In fact, early studies demonstrated that Id⁺ anti-DNA Abs in healthy individuals were blocked by anti-Id Abs and that titers of such anti-Id Abs correlated negatively with disease progression (51–53). A similar phenomenon was described more recently for anti-GAD65 autoantibodies in type 1 diabetes (54). How are such results related to Id-specific Th cells? It is increasingly clear that Abs are conventional Ags for Th cell and B cell immune responses (55, 56). Thus, anti-Id B cells that can bind a particular Ab (e.g., an Id⁺ anti-dsDNA Ab) internalize this Id⁺ Ab and may elicit help from Id-specific Th cells, the same Th cells that initially could have delivered Id-dependent help to the anti-dsDNA Id⁺ B cell (55, 56). In these terms, pathogenic Abs are counter-balanced by Id-driven help of anti-Id B cells. Further studies into such circuits are warranted in SLE patients.

We thank Hilde Omholt and Peter O. Hofgaard for technical help and Marianne Frøyland for advice on methods; the Department of Comparative Medicine, Oslo University Hospital, for animal care; and Inger Øynebråten and Peter O. Hofgaard for critical reading of the manuscript.

The authors have no financial conflicts of interest.