Lupus Susceptibility Genes May Breach Tolerance to DNA by Impairing Receptor Editing of Nuclear Antigen-Reactive B Cells¹

The line that demarcates tolerance from autoimmunity is indeed fine, being tightly patrolled by a series of key checkpoints (1). Although it is well accepted that systemic autoimmunity resulting in the production of antinuclear autoantibodies (ANAs)⁴ is the end result of a breach in immune tolerance, the molecules and mechanisms responsible for breaching this delicate balance have remained obscure. Recent genetic studies in murine models of lupus have uncovered a spate of loci that confer susceptibility to systemic autoimmunity. Although the molecular identities of most of these loci remain to be elucidated, we have gained a glimpse of how these loci may be functioning through congenic dissection studies (2, 3, 4, 5, 6, 7).

Sle2^z is a particularly interesting NZM2410-derived locus on mid-chromosome 4, which impacts the production of autoantibodies and lupus nephritis (8). Other studies have also identified a disease susceptibility locus on chromosome 4 arising from the New Zealand Black strain as reviewed (9, 10). When introgressed onto the normal C57BL/6 (B6) background, the Sle2^z locus leads to age-dependent hypergammaglobulinemia and generalized B cell hyperactivity accompanied by an expansion of B1a cells in the peripheral lymphoid organs. Despite these cellular and serological phenotypes, glomerulonephritis is not prominent in this congenic strain (2). In contrast to these phenotypes, the epistatic interaction of Sle2^z with additional lupus susceptibility loci such as Sle1^z and Sle3^z recreates full-blown lupus nephritis (6). To understand whether and how Sle2^z might breach B cell tolerance, the Sle2^z locus was bred to two anti-DNA Ig H chain (HC) transgene (Tg) models, B6.3H9 and B6.56R.

The “3H9” and “56R” anti-DNA Ig Tg models have been very useful in defining the checkpoints that maintain B cell tolerance to DNA (11, 12, 13, 14) and in delineating the impact of various autoimmune genomes and molecules on B cell tolerance (15, 16, 17, 18). Whereas high avidity anti-dsDNA B cells are centrally edited or deleted (11, 12, 14, 19), low-avidity anti-ssDNA B cells may be developmentally arrested or rendered anergic and excluded from the B cell follicles, effectively precluding them from differentiating into memory or plasma cells (13). It has also become evident that the strain background on which the anti-DNA Ig Tg is expressed could potentially influence the degree to which antiself B cells are censored. The initial studies with the 3H9 and 56R anti-DNA Ig Tgs have been conducted on the BALB/c background. On this background, tolerance to nuclear Ags was effectively maintained through a variety of mechanisms as discussed above. In contrast, when these Tgs were bred onto a lupus-prone genome such as the MRL.lpr, tolerance was breached, resulting in the production of anti-DNA Abs encoded by the Tg HC paired with specific L chain (LC) partners, with a significant degree of somatic mutation (15, 16, 17, 18).

For the purpose of the present study, the 3H9 and 56R Tgs were backcrossed onto the B6 background to facilitate breeding to the B6.Sle2^z congenics. In addition to uncovering the impact of Sle2^z on B cell tolerance, these studies also reveal that the B6 background itself might contribute to loss of B cell tolerance. This is of particular importance given the observation that a couple of antinuclear autoimmunity-promoting loci have also been mapped to the B6 genome even though this strain does not develop lupus spontaneously (20, 21, 22, 23, 24, 25). The introgression of Sle2^z onto these Tg mice was noted to further breach B cell tolerance apparently by thwarting receptor editing.

B6.Sle2^{NZM2410/NZM2410} (referred to in this manuscript as B6.Sle2^z) mice are congenic homozygotes for the 95% confidence interval flanking Sle2^z, which is derived from the NZM2410 lupus strain with termini at D4mit17 and D4mit12 on chromosome 4. The B cell and serological phenotypes of this congenic strain have previously been reported (2). The 3H9 and 56R anti-DNA HC-only, site-directed Tgs were backcrossed from the BALB/c background onto the B6 background for more than seven generations. “B6.3H9” mice are hemizygous for the 3H9 anti-DNA Ig HC Tgs, whereas “B6.56R” mice are hemizygous for the 3H9^D56R anti-DNA Ig HC Tgs, both on the B6 background. Importantly, the single residue change in the CDR2 region of the latter HC Tg, which replaces an aspartate residue with an arginine residue, greatly augments the avidity for DNA (14, 19). No Ig LC Tg constructs were used in this study. B6.Sle2^z.3H9 and B6.Sle2^z.56R mice are B6 mice that are homozygous for the Sle2^z locus but hemizygous for the respective anti-DNA Tgs. The primers used to confirm the presence of the NZM2410 allele of the Sle2^z locus as well as that of the HC Tg have been reported previously (2, 19). All mice used for this study were bred and housed in a specific pathogen free colony, at the University of Texas Southwestern Medical Center Department of Animal Resources (Dallas, TX). Comparable numbers of male and female mice were pooled for all experiments, as no significant sex differences were noted in the phenotypes studied.

Splenocytes were depleted of RBC using Tris ammonium chloride and single-cell suspensions were prepared for culture or flow cytometric analysis as described previously (3, 4, 5). Flow sorting was conducted as described previously (7). Sera and mAb culture supernatants were screened for Ab reactivity to ssDNA, dsDNA, and histone/DNA complexes by ELISA as described previously (3, 5). Raw OD was converted to micrograms per milliliter by using defined concentrations of IgM or IgG mAbs to construct a standard curve or converted to units per milliliter by using a positive control serum sample derived from a B6.Sle1^z.Sle3^z mouse (which bears the IgM^b/IgG2a^b allotype) or a serum sample derived from a seropositive B6.Sle1^z.56R mouse (which bears the IgM^a/IgG2a^a allotype), arbitrarily setting the reactivity of a 1/100 dilution of this serum sample to be 10 U/ml. Sera with reactivities stronger than the test standard were diluted further and reassayed. The glomerular-binding ELISA was performed as described previously (5) using glomerular sonicate as a substrate. For the allotype-specific ELISA, the above assays were repeated with one modification; instead of using anti-IgM or anti-IgG second-step Abs, biotin-coupled anti-IgM^a, anti-IgM^b, anti-IgG2a^a, or anti-IgG2a^b Abs were used followed by avidin-coupled alkaline phosphatase.

Tissue sections were processed as previously described (26). Frozen sections were stained with MOMA-1 (rat IgG2a, specific for mouse splenic metallophilic macrophages; a gift from Dr. G. Kraal, Viride Universiteit, Amsterdam, The Netherlands) developed with goat anti-rat IgG-Alexa 647 (Molecular Probes/Invitrogen Life Technologies), blocked with normal rat serum (Pell-Freeze), and washed and stained with Alexa Fluor 555-RS3.1 (Anti-IgM^a). T cells were stained with a mixture of FITC anti-CD4 and anti-CD8 (BD Biosciences). Images were obtained on a Leica DMRB microscope equipped with a Hamamatsu C4742-95 charge-coupled device camera using Openlab imaging software (Improvision) to recolor anti-IgM in green and anti-CD4/CD8 in red.

Mice were monitored at 3, 6, and 9–12 mo of age for evidence of clinical nephritis. Urine was collected over a 24-h period using metabolic cages and the total amount of urinary protein was assayed using a Coomassie-based assay (Pierce). Typically, normal B6 mice do not excrete >1.2 mg of urinary protein per day. In addition the level of blood-urea-nitrogen (BUN) was measured using a commercially available kit (Sigma-Aldrich). Typically, normal B6 mice exhibit <30 mg/dl BUN. Mice were sacrificed at 9–12 mo of age and their kidneys were fixed, sectioned, and stained with H&E, and periodic acid-Schiff. For the analysis of kidney sections, at least 100 glomeruli were examined per section by light microscopy for evidence of hypertrophy, proliferative changes, crescent formation, hyaline deposits, fibrosis/sclerosis, and basement membrane thickening and graded semiquantitatively.

Spleens were removed aseptically from 9- to 12-mo-old seropositive Tg mice. A quarter of the splenocytes were stimulated with LPS (20 μg/ml for 48 h), whereas the rest were not. Both the LPS-prestimulated and the unstimulated splenocytes were separately fused to the SP2/0 fusion partner. All wells with single colonies were tested for Ab production 7–10 days postfusion. Cells from positive wells were subcloned once more. Ab concentration was determined using a Coomassie Plus protein assay kit (Pierce) and an isotype-specific ELISA as described above.

All Abs in the ANA and non-ANA databases were “blasted” against the public IgBLAST database of mouse Ig sequences at National Institutes of Health/National Center of Biotechnology Information (NCBI) (www.ncbi.nlm.nih.gov/igblast/) to determine the closest germline gene of origin. Comparing these sequences to the deposited repertoire of germline genes allowed us to identify residues that were potentially somatically mutated. The CDR position and numbering scheme adopted matched that used by the NCBI IgBLAST database. The usage frequencies of individual genes were compared using χ² tests. Where applicable, the Fisher’s exact test was applied. Intergroup comparisons of phenotypes between the two Tg strains were conducted using Student’s t test. All statistical comparisons were performed using SigmaStat (Jandel Scientific).

Surprisingly, in the process of constructing the B6.Sle2^z.56R Tg mouse we observed that when the 56R (but not the 3H9) anti-DNA Ig Tg was expressed on the B6 background tolerance was spontaneously lost even in the absence of any introgressed lupus susceptibility loci, effectively leading to significant levels of IgM and IgG autoantibodies (Fig. 1). Similar observations have been reported by other investigators (27, 28). When Sle2^z was introgressed onto these Tg models this locus clearly augmented serum anti-DNA Ab levels further. Both B6.Sle2^z.3H9 and B6.Sle2^z.56R mice exhibited significantly higher levels of IgM anti-dsDNA Abs (p < 0.0001), and IgG anti-dsDNA Abs (p < 0.008) than age-matched B6.3H9 and B6.56R mice. Similar differences were seen with IgM and IgG anti-ssDNA and anti-histone/DNA Abs (Fig. 1). Because the highest levels of anti-DNA Abs were observed consistently in B6.Sle2^z.56R mice (Fig. 1), further studies were focused on B6.Sle2^z.56R mice and age-matched B6.56R controls.

Although both strains exhibited significant levels of anti-DNA Abs at 3 mo of age, the serological differences between these two strains only became apparent after 6 mo (Fig. 2,A). Although the contribution of the B6 genome to breaching B cell tolerance is apparently early, the impact of Sle2^z is clearly age-dependent, becoming most pronounced at 9 to 12 mo of age (Figs. 1 and 2,A). As noted in Fig. 1 and reported previously (29), antinuclear Abs are not particularly prominent in Tg-negative B6.Sle2^z congenics, indicating that the presence of the anti-DNA Tg is critical for this serological phenotype in B6.Sle2^z.56R Tg mice. Homozygosity for Sle2^z was associated with higher Ab levels, but the differences between the homozygous and heterozygous mice failed to attain statistical significance (Fig. 2 B). Any apparent sex differences were also not statistically significant (data not shown). Finally, no differences were noted in the IgG subclass distribution of the serum ANAs, with IgG2a and IgG2b Abs dominating in both strains (data not shown).

Next, allotype-specific reagents were used to ascertain whether the antinuclear Abs in these mice were Tg encoded or otherwise. Reagents specific for the Igh“a” allotype were used to detect Tg-encoded Abs, and reagents specific for the Igh“b” allotype were used to detect Abs encoded by the endogenous HC locus. Because IgM and IgG2a Abs were prominent in both these strains and because allotype-specific reagents for these isotypes or subclasses were readily available, these Abs were specifically studied. As is clear from Fig. 2, C–F, B6.Sle2^z.56R mice exhibited increased levels of IgM^a anti-dsDNA (p < 0.001), IgG2A^a anti-dsDNA (p < 0.006), and IgM^a anti-ssDNA ANAs (p < 0.0002), relative to the levels seen in B6.56R mice.

Because the serological differences between the two strains were most marked in 9- to 12-mo-old mice, the cellular phenotypes in 9- to 12-mo-old B6.56R and B6.Sle2^z.56R mice were next examined. B6.56R and B6.Sle2^z.56R mice did not differ in the relative frequencies of immature or mature B cells in the bone marrow (BM) (data not shown). However, interesting differences were seen in the periphery. B6.Sle2^z.56R spleens exhibited increased total cell numbers with interesting differences in cell composition as detailed in Table I. As anticipated, the splenic B cell compartments of these mice consisted of a mixture of IgM^a-expressing and IgM^b-expressing B cells. Sle2^z did not affect this distribution, as both B6.56R and B6.Sle2^z.56R spleens exhibited three times as many IgM^a–bearing Tg B cells compared with IgM^b-bearing B cells (Table I). However, the absolute numbers of IgM^a and IgM^b B cells in B6.Sle2^z.56R spleens were both elevated (almost twice) relative to the corresponding numbers in B6.56R spleens (Table I). Intriguingly, Sle2^z led to augmented splenic CD4:CD8 ratios (p < 0.00003), arising mainly from increased CD4 T cell numbers (p < 0.00001) bearing a more activated phenotype (p < 0.0002). Hence, the increased cell numbers seen in B6.Sle2^z.56R spleens were composed of expanded populations of CD4⁺ T cells as well as IgM^a- and/or IgM^b-expressing B cells.

Compared with the IgM^a-bearing B cells from B6.56R spleens, IgM^a-bearing B cells from B6.Sle2^z.56R spleens were skewed toward being more marginal zone (MZ)-like (i.e., CD21^highCD23^low) and plasmacytoid in phenotype (i.e., CD21^lowCD23^lowB220^dimIgM^dimCD43⁺syndecan-1⁺) relative to the proportion of (CD23^high) follicular B cells (Fig. 3,A and data not shown). This difference in phenotype between B6.56R and B6.Sle2^z.56R was not seen among IgM^b B cells (data not shown). The above differences were also noted histologically. As can be seen from Fig. 3,B, B6.Sle2^z.56R spleens exhibited more B cells, including MZ B cells, compared with B6.56R spleens. Although MZ B cells from these mice were capable of secreting more Abs than the follicular B cells, the CD21^lowCD23^lowsyndecan-1⁺“plasmacytoid” B cells were the most prolific Ab secretors in culture (Fig. 3,C). Sle2^z-bearing Ig Tg mature B cells were also larger in size with higher surface levels of the activation markers CD86/B72 and I-A^b (Fig. 3,D). In addition, B6.Sle2^z.56R Tg B cells showed increased frequencies of peanut agglutinin^high germinal center type B cells (Fig. 3,D). Because one of the hallmarks of Sle2^z is the expansion of peritoneal and splenic B1a cells (2), these cells were studied next. Interestingly, in the spleens and particularly in the peritoneal cavity Sle2^z led to a prominent skewing toward the B1a B cell phenotype, but this was seen only among the Tg-negative (i.e., IgM^b-positive but lacking IgM^a) B cells (Table I).

Similar phenotypic differences were also noted in younger mice. Thus, 4- to 6-mo-old B6.Sle2^z.56R spleens exhibited twice as many MZ B cells compared with age-matched B6.56R mice (27.1 ± 8.5 vs 12.4 ± 3.5%, n = 5 each; p < 0.07). In both the B6.56R and B6.Sle2^z.56R spleens it was also interesting to observe a significant expansion of IgM^a+b+ double-expressor B cells particularly among the MZ B cells; hence, 19% of MZ B cells in B6.56R and 22% of MZ B cells in B6.Sle2^z.56R spleens (n = 5 each) were IgM^a+b+ double expressers (data not plotted). These frequencies were significantly higher than the frequencies of double expressers among follicular B cells in both strains (which ranged from 4.6 to 5.5% on the average).

To ascertain whether Sle2^z has the capacity to influence the molecular structure of anti-DNA Abs, monoclonal hybridomas were isolated from a seropositive B6.56R mouse and an equally seropositive B6.Sle2^z.56R mouse, both aged 9–12 mo. The mAbs isolated from the fusions without LPS prestimulation were studied first. The supernatant from all of the wells with singleton clones were screened for the presence of IgM or IgG Abs as well as for reactivity to ssDNA, dsDNA, and histone-DNA complexes. The mAbs derived fell into two broad categories: Abs that reacted strongly with ssDNA (and also to dsDNA and histone/DNA complexes, albeit to a lesser degree) on the one hand and Abs that failed to react with any of these three nuclear Ags on the other. These two types of Abs are referred to as “ANAs” and “non-ANAs”, respectively, in the rest of this report. Abs that demonstrated exclusive reactivity to ssDNA but not to dsDNA or exclusive reactivity to dsDNA but not ssDNA were not observed. Of the 211 B6.56R-derived singleton wells producing Abs, 117 (i.e., 56%) were ANA-positive; likewise, 142 of 205 (i.e., 69%) B6.Sle2^z.56R-derived singleton mAbs were ANA-positive. Ninety-seven mAbs from B6.56R mice and 86 mAbs from B6.Sle2^z.56R mice were randomly selected for sequence analysis without any reference to their specificity, isotype, or allotype. Of these, ∼93% of the mAbs isolated from B6.56R and 97% of the mAbs isolated from the B6.Sle2^z.56R mouse were IgM in isotype; the rest were IgG2a or IgG2b.

Given that the B6 background itself was sufficient to breach B cell tolerance in the 56R model (i.e., without any contribution from Sle2^z), it became important to ascertain the molecular makeup of the autoantibodies in this strain. Interestingly, a third of the anti-DNA B cells in B6.56R mice were encoded by an endogenous HC, not the Tg HC, in partnership with several different LCs. A large spectrum of somatically mutated endogenous HC genes were used to encode these Abs, including VH57-1M, VH36–60, Vox-1, VGK1B, VGK1A, J558k, J558i, J558.6, J558.15, J558.18, J558.29, 119.12, V23, and VH186.2, paired with different LC genes. Minimal evidence of clonal expansion was noted among the ANAs that used endogenous HC genes, although it remains possible that any underlying clonal expansion might have been masked by expeditious editing and re-editing or by the rapid introduction of additional mutations. The remaining two-thirds of the anti-DNA Abs isolated from B6.56R spleens expressed the Tg HC paired with a restricted subset of LC partners, including ba9 (V_k9), 19-23 (V_k19), and gj38c (V_k38), as displayed in Fig. 4.

The ANAs isolated from the B6.Sle2^z.56R strain were significantly different in their molecular makeup. In contrast to the B6.56R-derived ANAs, endogenous HC were infrequently used to encode ANAs (<5%; p < 0.001, χ² test). Instead, ANAs in B6.Sle2^z.56R mice were primarily generated through the pairing of an unmutated 56R HC Tg with particular LC partners (Fig. 4). Although several different LCs were noted to participate in conferring DNA reactivity, four LC genes dominated: 19-23 (V_k19), bt20 (V_k20), V_k21 (21-12), and gj38c (V_k38). It was evident that several clones were identical, being composed of the Tg HC (with no mutations) and paired with the same germline-encoded V_k gene; however, it was unclear whether these mAbs were truly clonally related or represented descendants of independent founders.

Once again, there was a dramatic difference in the molecular makeup of B6.56R-derived and B6.Sle2^z.56R-derived non-ANAs. Approximately 71% of the non-ANAs from B6.56R mice (as opposed to 17% of the non-ANAs isolated from B6.Sle2^z.56R mice) used an endogenous HC. Among the B6.56R-derived non-ANA mAbs that used an endogenous HC, five were noted to be clonally related, using a VGK1A HC and a V_k 23–39 LC partner. Although most of the mAbs that used endogenous V_H genes appeared to have used the nontargeted HC locus, several examples of editing at the targeted Tg HC locus at the 3′ heptamer located within framework region 3 were also observed (data not shown). These included examples of VD→VDJ and V→VDJ replacements as has been observed before in this Tg model system (12). Because few of the mAbs isolated from B6.Sle2^z.56R used endogenous HC, it was perhaps not surprising that no such secondary HC replacement events were evident among the mAbs isolated from this strain.

In both Tg strains, a second “solution” that was frequently adopted to “avoid” nuclear Ag reactivity was the use of certain LC partners (paired with the Tg HC) that apparently “vetoed” DNA binding, as is evident from Fig. 4. In both strains, the preferred “vetoing” LC partner used was a V_k20 family member, bt20. Hence, 50% of the non-ANA mAbs from the B6.56R fusion that possessed the (unmutated) 56R Tg HC used V_k bt20; the corresponding value for the B6.Sle2^z.56R-derived non-ANA mAbs was ∼47% (Fig. 4).

Next, we examined the J_k usage of these mAbs. As displayed in Fig. 5, among the hybridomas isolated from B6.56R mice there was a strong suggestion of receptor editing (and possibly, re-editing) as supported by the increased usage of J_k4 or J_k5 genes. In particular, the LC of non-ANAs demonstrated greater skewing to J_k4 and J_k5. Remarkably, 16 of 25 mAbs of B6.56R origin that had precluded DNA binding by using an endogenous HC gene bore LC partners that were recombined to J_k5. In contrast, skewing to the more distal J_k genes was significantly less prominent among the mAbs from B6.Sle2^z.56R spleens. Thus, although the B6.Sle2^z.56R-derived non-ANA mAbs exhibited heavier skewing toward J_k4, compared with the ANAs derived from the same strain, none of the non-ANAs used J_k5 (Fig. 5). A similar difference was also seen among the ANAs; whereas two-thirds of the ANAs isolated from the B6.56R spleen that had used the Tg HC bore V_k partners that were recombined to J_k4 or J_k5 (Fig. 5,B), only a third of the ANAs drawn from the B6.Sle2^z.56R spleen used the more distal J_k genes (p < 0.013; Fig. 5 C). To summarize: 1) in all groups of Abs (from both strains), non-ANAs exhibited greater usage of the more distal J_ks and, hence, non-ANAs appeared more likely to have been the products of receptor editing; and 2) this skewing was significantly stronger among the B6.56R-derived Abs, indicative perhaps of a more vigorous receptor editing program in the absence of Sle2^z.

The above findings bolster the prevailing notion that the LC partner can greatly influence the DNA-reactivity of Abs encoded by the HC Tg (12, 14, 18). Thus, for example, the V_k 19–23 LC strongly facilitates DNA binding when paired with the 56R Tg HC (Fig. 4), whereas none of the 56R Tg-bearing non-ANAs possessed the V_k 19-23 LC. For yet other LC partners it appeared that different germline genes of the same V_k family may be impacting nuclear Ag reactivity differently despite being paired with the same 56R HC Tg. For instance, among the 56R HC/Vk4 LC-expressing mAbs isolated from B6.Sle2^z.56R mice (Fig. 4), the kh4 V_k4 germline gene facilitated DNA binding but the 4–50 V_k4 germline gene vetoed DNA binding (data not shown). V_k21 constitutes another example. Whereas V_k21D (21-4) was previously reported to be a good vetoing LC, we find that most of the V_k21 LC that facilitated DNA binding in B6.Sle2^z.56R mice (Fig. 4) were V_k21-12, which is significantly less anionic than V_k21-4 because the former possesses an arginine residue at L92 (CDR3) but lacks an aspartate at L94 (CDR3).

It was somewhat perplexing that the LC that turned out to be the most potent “vetoer” of DNA-binding in both strains, V_k bt20, also appeared to have the potential to permit DNA-reactivity in some instances (Fig. 4). To explore how the same LC partner complexed with the identical unmutated 56R Tg could potentially have given rise to strikingly different Ag reactivity patterns, we examined these particular hybridomas for evidence of other potential molecular contributors. As is evident from Table II, it appears likely that the vetoing LC, V_k bt20, may still be able to confer DNA-binding (when paired with the 56R Tg) if it also possessed a V_λ LC or certain J_k2-encoded junctional residues (e.g., Tyr at position L96). Although these two differences appear to adequately account for the specificity differences observed among most of the 56R HC/bt20 LC mAbs, one of the isolated mAbs (italicized in Table II) clearly challenges this explanation. It is presently not clear whether these differences may relate to additional LC partners that we might have failed to amplify.

The above studies have been based on examining a single B6.56R mouse and a single B6.Sle2^z.56R mouse. The HC and LC usage profiles noted in B6.56R-derived B cells have been reproduced by others (30, 31, 32) as discussed further below. To establish the reproducibility of the Ig repertoire profiles noted in B6.Sle2^z.56R mice, we performed two additional fusions using spleens from seropositive 9- to 12-mo-old B6.Sle2^z.56R mice. As summarized in Table III, all of the derived hybridomas were noted to use the Tg HC (IgM^a) paired to different LC partners. Consistent with the observation made in the first fusion, the two repeat fusions also showed that B6.Sle2^z.56R derived anti-DNA B cells as well as non-ANA B cells exhibited no use of endogenous HC and more frequent usage of J_k1 or J_k2 (Table III). In terms of the specific editor LC used to veto DNA binding, there was clearly evidence of mouse to mouse variation, although individual non-ANA clones in both of the additional fusions continued to use the V_k20 gene bt20 as one of the vetoing LC partner. In addition, V_k21 played a more prominent role as a vetoing LC partner in B6.Sle2^z.56R non-ANA B cells compared with the first fusion (Table III).

A further panel of mAbs was isolated from B6.56R and B6.Sle2^z.56R spleens following prestimulation with LPS. The productive wells with singleton clones from both fusions were all ANA-positive; i.e., all mAbs reacted strongly with ssDNA but with varying grades of reactivity to dsDNA and histone/DNA complexes (data not shown). In both groups of hybridomas the use of endogenous HC and somatic mutations were rare. Almost all ANAs were encoded by the unmutated 56R Tg HC paired with a spectrum of different LC partners. Interestingly though, there was a strong overrepresentation of a single LC; 22 of 38 (i.e., 58%) LPS-rescued mAbs from B6.56R and 17 of 34 (i.e., 50%) LPS-rescued mAbs from B6.Sle2^z.56R spleens were encoded by the 56R HC Tg paired with an unmutated V_k 19–23 LC partner. Indeed, this HC/LC pair also represented the single most prevalent HC/LC combination observed among the ANAs isolated from both Tg strains without LPS prestimulation, as is evident from Fig. 4 and Table III. Because somatic mutations were infrequent, any potential clonal interrelationship between these isolated mAbs could not be readily discerned. Hence, the use of LPS to rescue and study mAbs from these Tg mice did not yield any additional insight.

Finally, the extent of renal disease in these Tg mice was assessed to determine whether the increased circulating levels of anti-DNA Abs was associated with end organ damage. Interestingly, nephritis was not obvious in any of these Tg strains as gauged by their low urinary protein excretion levels, low BUN, negligible renal pathology, and minimal glomerular Ig and complement deposition (data not shown). This may relate to the fact that although the 3H9/56R Tg-encoded ANAs in these mice may bind well to DNA, they may not be nephrophilic. Indeed, compared with B6.Sle1^z.lpr mice (which exhibit high levels of nephrophilic Abs and renal disease), the Tg mice analyzed in this study exhibited minimal serological reactivity to glomerular Ags (data not shown). Hence, the anti-DNA Abs that had originated as a consequence of the breach in B cell tolerance in these Tg models may simply have been incapable of targeting the kidneys.

The 56R anti-DNA Ig HC Tg has turned out to be a particularly informative tool to dissect out how different genetic backgrounds and disease loci may influence B cell tolerance to nuclear Ags. The 56R Tg differs from the 3H9 Tg in bearing an arginine (R) residue in place of an aspartate (D) residue at H56. It is clear that this single residue can dramatically alter the reactivity and affinity of the Ab to DNA (12, 33). This also appears to be true in the case of other independently derived ANAs based on a completed comparative study of HCs from ANAs and non-ANAs (34). It is now clear that the extent to which B cells bearing the site-directed 56R anti-DNA Tgs are tolerized is dependent on the genetic background (14, 18, 27, 30, 31) (Table IV). Hence, on the BALB/c background these B cells are fairly well tolerized with negligible or modest levels of 56R-encoded serum autoantibodies and anti-DNA B cells (14, 27) (Table IV). In most tolerant B cells from these mice the Tg HC is paired with an “editor” LC. There is minimal evidence of secondary rearrangement at the untargeted HC locus among both the ANAs and the non-ANAs derived from BALB.56R mice.

In contrast, when the 56R HC Tg is bred onto the B6 background an age-dependent increase in Tg-encoded anti-DNA Abs and B cells emerge, as noted in the current study (Table IV) and recent reports (27, 28, 32). Remarkably, both the anti-DNA as well as the non-ANA B cells from B6.56R mice show signs of vigorous receptor editing or secondary rearrangement at both the HC and LC loci. Indeed, a third of the ANAs in these mice are encoded by endogenous V_H genes. Likewise, both the DNA-binding and non-ANA B cells from these mice exhibit a J_k usage profile that is skewed toward the more distal J_k genes. Despite these vigorous editing attempts, a modest but substantial number of B cells appear to remain autoreactive because Tg-encoded autoantibodies surface with age in B6.56R mice. Hence, compared with the BALB/c genome the B6 genome appears to harbor tolerance-breaching, autoantibody-promoting genes. Indeed, several such disease susceptibility loci have been mapped to the B6 genome in independent studies (20, 21, 22, 23, 24), although the culprit genes within these loci remain to be elucidated. In contrast to unmanipulated B6.56R mice, the delivery of allogenic T cell help was noted to significantly augment the breach in tolerance (30, 31, Table IV). Once again, it is intriguing to observe that most of the anti-DNA Abs in the latter graft-vs-host disease model were not Tg encoded, but represented the end product of secondary rearrangement at the endogenous HC locus. The same pattern was also noted among the non-ANAs, >90% of which were encoded by endogenous HC genes.

In contrast to the above models, the introgression of the Sle2^z lupus susceptibility interval onto the B6.56R model augments the breach in B cell tolerance to DNA with little evidence of receptor editing. As is true with spontaneous lupus in non-Tg mouse models, the breach in B cell tolerance in this model also appears to be age-dependent, becoming prominent beyond 6 mo. The age dependence of this phenotype does not appear to reflect the need to accumulate “sufficient” numbers of somatic mutations, because the mAbs isolated from aged, seropositive mice were largely unmutated. It is possible, however, that age-dependent aberrations in intrinsic B cell function, T cell help, or other yet to be defined triggers may have the potential to breach B cell tolerance in this model. In this context, the expanded pool of activated CD4-positive T cells seen in B6.Sle2^z.56R mice may be responsible, in part, for the observed phenotype. However, it is also possible that the expanded CD4-positive T cell compartment may be secondary to the prominent expansion and activation of Tg B cells seen in these mice. Indeed, the latter scenario would be consistent with recent reports demonstrating how activated B cells could drive the subsequent activation and expansion of a CD4-positive T cell in other model systems (35, 36, 37).

Based on previous reports (2), it was perhaps not surprising that Sle2^z led to an activated B cell phenotype with increased numbers of germinal center B cells. However, it was interesting to note the emergence of two distinct pools of B cells in B6.Sle2^z.56R mice, depending on whether or not the HC Tg was used by a given B cell. As indicated in Fig. 3, the presence of Sle2^z skewed a substantial percentage of the Tg B cells toward the MZ phenotype or toward a preplasma cell phenotype. For instance, each 9- to 12-mo-old B6.Sle2^z.56R spleen harbored 11.7 × 10⁶ IgM^a MZ B cells, as opposed to 8.2 × 10⁶ in B6.56R spleens. These B cells are clearly capable of producing high levels of Tg-encoded ANAs based on the ELISA results depicted in Figs. 2 and 3. In contrast, the vast majority of the B cells that had used endogenous V_H genes (i.e., of IgM^b allotype) were skewed toward being B1a in phenotype, particularly in aging mice (Table I), akin to the phenotypes noted in Tg-negative, B6.Sle2^z mice (2). These B cells are not prolific producers of ANAs based on the allotype-specific ELISA shown in Fig. 2, consistent with previous reports that Sle2^z-associated Abs are polyreactive rather than nuclear Ag focused (29).

MZ B cells and B1 cells share several properties, including their ontogeny and in vivo function (38, 39, 40, 41). Several B cell intrinsic signaling molecules that have been demonstrated to influence lupus pathogenesis can also impact the development of B1 and MZ B cells (39, 40, 41, 42). Importantly, the BCR Ag specificity, which could directly impact the nature and level of signaling from autoreactive triggers, greatly influences the degree of skewing to B1 cells (43, 44, 45). Indeed, it appears that newly emergent naive B cells may be recruited into these different peripheral compartments upon exposure to self-Ags (46, 47, 48). Based on the above reports, we posit that Sle2^z may serve to augment the recruitment of nascent B cells (that had escaped central tolerance) into the B1 and MZ B cell compartments, respectively, depending on avidity and the Ag specificity of their BCR. As proposed in Fig. 6, incessant autostimulation by low-avidity self-Ags may induce the up-regulation of CD5, perhaps in an attempt to dampen cell signaling (49, 50), leading to the B1a phenotype in the context of Sle2^z. In contrast, higher avidity triggers (such as those perceived by a 56R-encoded BCR with specificity for DNA) may drive these cells to become MZ B cells and, eventually, plasmablasts as modeled in Fig. 6. In contrast, in the absence of Sle2^z most of the nascent B cells may simply be recruited into the recirculating pool of follicular B cells (Fig. 6).

An alternative factor that is not mutually exclusive relates to potential differences in T cell help. Whereas the B1a cells may have developed following exclusive T-independent BCR cross-linking (i.e., with relatively little T cell help), the 56R-encoded anti-DNA B cells may have potentially received T cell help based on the observation that 56R-encoded B cells are capable of reacting with DNA/histone (and possibly other DNA/protein) complexes. This is supported in part by the observation that whereas B6.Sle2^z mice (which exhibit an expansion of B1a cells) have a normal T cell compartment (2), B6.Sle2^z.56R mice (which possess an expanded MZ compartment) exhibit prominent CD4-positive T cell activation and expansion (Table I). Hence, the extent of T:B cross-talk may also be responsible in part for driving the development of different B cell subsets with differing functional properties and surface phenotypes (i.e., CD5 vs CD21 up-regulation). It has been previously noted that the expanded MZ B cells in 56R Tg models may coexpress κ and λ LC (14). However, we did not find this to be the case in the expanded MZ B cell compartment of B6.Sle2^z.56R mice, although they were enriched for B cells bearing dual BCRs, one encoded by the Tg IgM^a HC and the other by an endogenous IgM^b HC.

A key question relates to how Sle2^z may be functioning to breach B cell tolerance. Because we could detect no difference in the cellular phenotypes of BM cells from these mice, there is insufficient evidence that Sle2^z may be impacting the central deletion of autoreactive B cells. It is also clear that Sle2^z was not impacting somatic mutation rates, based on the Ab sequence analyses. However, Sle2^z did appear to be “dampening” receptor editing and/or revision in the context of the 56R Tg. Thus, whereas the B cells isolated from B6.56R mice demonstrated high usage frequencies of endogenous V_H genes (to encode both ANAs and non-ANAs), this was relatively rare among the B cells isolated from B6.Sle2^z.56R mice. As reviewed in Table IV, the frequency of endogenous HC usage among B6.Sle2^z.56R anti-DNA-positive hybridomas was 3, 0, and 0%, respectively (average of 1% as listed in Table IV), compared with 7, 29, 71, and 10%, respectively, in B6.56R fusions (Table IV). More strikingly, the frequencies of endogenous HC usage among non-ANA hybridomas was 17, 0, and 0%, respectively, among B6.Sle2^z.56R-derived hybridomas (average of 9% as listed in Table IV) as opposed to 60, 70, and 91%, respectively, among non-ANAs from B6.56R mice (Table IV). Whereas the use of endogenous V_H genes in B6.56R mice appeared to involve expression from the non-targeted HC locus primarily, several examples of secondary rearrangement into the targeted Tg locus were also evident in this study. In contrast, these were infrequent in B6.Sle2^z.56R mice.

Likewise, both the ANAs and the non-ANAs demonstrated a high usage frequency of the more distal J_k genes in the absence of Sle2^z (Fig. 5). Whereas ∼50–60% of ANAs and non-ANAs derived from B6.Sle2^z.56R B cells were recombined to J_k1/J_k2 (Fig. 5 and Tables III and IV; n = 3 independent fusions), the corresponding values in B6.56R spleens varied from 20 to 30% based on at least four independent fusions drawn from the present (Table IV, third data column from the left) and previous reports (Table IV, second, fifth, and sixth data columns from the left). Hence, Sle2^z-bearing B cells appear to be relatively “tolerant” of their “primary” V_H/V_k combination (i.e., 56R Tg paired with a V_k LC recombined to J_k1/J_k2) despite the fact this BCR may have been overtly autoreactive! Both the HC and LC usage profiles in these mice suggest that the presence of Sle2^z may somewhat lessen the pressure to edit an initial BCR that may have been autoreactive. Because the self-Ag DNA is ubiquitously expressed, including in the BM milieu, the shift in repertoire away from the primary HC Tg/LC combination in B6.56R B cells is likely to have commenced in the BM during early B cell development. Therefore, we believe that preimmune receptor editing as well as postimmune receptor revision may both be contributing to the observed repetoire differences. This study, taken together with recent observations revolving around another lupus susceptibility locus, Sle1^z (51), suggest that impairment of receptor editing may be a major mechanism through which antinuclear Abs arise in lupus.

One may posit that the degree of autoreactivity seen in the various 56R-based Tg models may be a function of the genetic load of autoimmunity-facilitating loci. Whereas the BALB/c genome may be relatively innocuous in this respect, the B6 genome may harbor a small but significant number of autoimmunity promoting genes (20, 21, 22, 23, 24). When the NZM2410 allele of Sle2^z is added to this genetic background an incremental breach in tolerance ensues, apparently due to impaired receptor editing. Likewise, the engineering of FcRIIb deficiency on the B6 background also facilitated a similar breach in tolerance marked by a significant expansion of IgG self-reactive plasma cells (32). Finally, at the other end of the spectrum profound autoimmunity ensues when the 56R Tg is bred onto the MRL.lpr genome. This is likely to be a consequence of the multitude of lupus susceptibility genes present in that genome (52, 53, 54, 55) that may be acting in concert to breach several B cell tolerance checkpoints. The challenge ahead is to unravel the genetic players within the B6 genome as well as the NZM2410-derived lupus susceptibility intervals that regulate B cell tolerance and to ascertain the molecular mechanisms through which these genes work.

We thank Drs. Nina Luning Prak, Ward Wakeland, and Anne Satterthwaite for helpful feedback on this work and Lisa Jia for technical assistance.

The authors have no financial conflict of interest.