Interleukin 6 Dependence of Anti-DNA Antibody Production: Evidence for Two Pathways of Autoantibody Formation in Pristane-induced Lupus

Mice were generated by five backcrosses of B6/129SV IL-6^−/− with BALB/cAn. Mice heterozygous for the IL-6 mutation were intercrossed to create IL-6^−/− and IL-6^+/+ mice and were genotyped by Southern blotting (10). Female BALB/cAn IL-6^−/− (n = 28) and IL-6^+/+ (n = 26) mice, age 10–12 wk, were injected once intraperitoneally with 0.5 ml of pristane (2,6,10,14-tetramethylpentadecane; Sigma Chemical Co., St. Louis, MO) (1). Control mice (IL-6^−/−, n = 10 and IL-6^+/+, n = 10) received 0.5 ml of PBS intraperitoneally. Sera were collected from the tail vein before treatment, at 3 wk, and at 3, 5, and 8 mo after treatment. Mice were housed under specific pathogen–free conditions.

Analysis of autoantibody specificities by immunoprecipitation was carried out as described previously (1).

Antigen capture ELISAs for anti-nRNP/Sm and anti-Su antibodies were performed as described, using mouse sera at a dilution of 1:500 and goat anti–mouse IgG second antibodies (γ chain–specific, from Southern Biotechnology Associates, Birmingham, AL) (11, 12). Anti-ssDNA ELISA was performed as previously described (13), and data were analyzed using the Mann Whitney test.

Anti-dsDNA antibody ELISA was carried out as previously described with minor modifications (14). In brief, calf thymus DNA (Sigma Chemical Co.) was extracted with phenol/chloroform/isoamyl alcohol and digested with S1 nuclease (GIBCO BRL, Gaithersburg, MD). Dynex polystyrene microtiter plates (Dynastar, Dynatech, Chantilly, VA) were coated with 100 μl/ well of 5 μg/ml DNA in SSC, pH 8.0, and incubated at 37°C for 12 h. Sera were tested at a 1:160 dilution followed by alkaline phosphatase–conjugated goat anti–mouse IgG antibodies. Substrate was added, and A_{405 nm} was determined. Absorbances from blank wells (no serum added) were subtracted.

IgG antichromatin antibodies were measured as previously described (15). In brief, polyvinyl chloride microtiter plates were coated with 10 μg/ml chicken erythrocyte chromatin in borate-buffered saline. Sera were tested at a 1:500 dilution, followed by biotinylated goat anti–mouse IgG (pFc′-specific, 1:4,000 dilution, from Dr. P.L. Cohen, University of North Carolina, Division of Rheumatology and Immunology), avidin-conjugated alkaline phosphatase (Zymed Labs, South San Francisco, CA), and p-nitrophenyl phosphate substrate. The A_{405 nm} of a 1:1,000 dilution of a high titer MRL/lpr reference serum was assigned a value of 2,048 units, and a 2¹¹-fold dilution of this standard a value of 2 units (15). IgG antichromatin activity in units for each sample was determined as in reference 12.

Anti-dsDNA antibodies were measured by the Crithidia luciliae kinetoplast staining assay (16) at a serum dilution of 1:20 according to the manufacturer's instructions (The Binding Site, Birmingham, UK). Second antibody was FITC-conjugated goat anti–mouse IgG (1:40 dilution; Southern Biotechnology Associates). All positive sera were titered (1:40, 1:80, 1:160, 1:320, and 1:640 dilutions).

IL-6 has been implicated in both anti-DNA antibody production and the pathogenesis of nephritis in (NZB/ W)F1 mice (6, 7), and in the development of autoantibodies in patients with cardiac myxomas (17, 18). It also is essential for the growth of plasmacytomas in pristane-treated mice (8, 19–21). This study was undertaken to evaluate the role of this cytokine in the induction of autoantibodies by pristane in BALB/c mice.

The production of IgM anti-ssDNA antibodies 2–3 wk after treating BALB/c mice with pristane (2) appears to be thymus-independent (reference 2 and Richards, H.B., M. Satoh, J.C. Jennette, T. Okano, Y.S. Kanwar, and W.H. Reeves, manuscript submitted for publication). Because IL-6 can stimulate T cell–independent Ig production (22, 23), the induction of IgM anti-ssDNA antibodies by pristane in BALB/cAn IL-6^−/− and IL-6^+/+ mice was investigated.

As shown in Fig. 1, IgM anti-ssDNA antibody levels were similar in IL-6^−/− versus IL-6^+/+ mice 3 wk after pristane treatment. In contrast, IgG anti-ssDNA antibodies were detected at a high frequency 8 mo after pristane treatment only in IL-6^+/+ mice (P < 0.05 versus IL-6^−/− by Mann Whitney test).

Anti-dsDNA antibodies are highly specific for lupus and are implicated in the pathogenesis of lupus nephritis. However, they were not detected previously in the 6-mo sera from pristane-treated BALB/c mice, a time when severe renal lesions already were apparent (2). In contrast, sera from 10 out of 26 BALB/cAn (IL-6^+/+) mice (38%) 8 mo after pristane treatment were strongly positive for anti- dsDNA antibodies by Crithidia luciliae kinetoplast staining (Table 1) at titers ranging from 1:80 to 1:640 (Table 2). Anti-dsDNA antibodies were not detected in pristane-treated IL-6^−/− mice or PBS-treated IL-6^+/+ or IL-6^−/− mice (Table 1). Although the Crithidia luciliae kinetoplast staining assay is highly specific for anti-dsDNA activity, occasionally antihistone antibodies give a false positive reaction. To exclude that possibility, the sera were tested for IgG anti-dsDNA antibodies by ELISA. The ELISA was somewhat less sensitive than the fluorescence assay, since only sera with titers >1:640 gave a significantly stronger reaction by ELISA than the Crithidia-negative group. An anti-ssDNA mAb (6/O2 576-2; from Dr. D. Pisetsky, Duke University, Durham, NC) also showed weak ELISA reactivity. Remarkably, none of the sera from pristane-treated IL-6^−/− mice bound to S1 nuclease-treated DNA at 8 mo (Fig. 1).

Together, these results indicate that pristane-treated BALB/cAn mice develop high levels of IgG anti-ssDNA and anti-dsDNA antibodies, and that these autoantibodies are IL-6 dependent. However, the appearance of these autoantibodies is delayed until 8 mo after treatment, which is after the onset of renal disease. Some of the pristane-treated BALB/cAn mice developed IgG anti-dsDNA antibodies at levels comparable to those in MRL/lpr mice (Table 2, Fig. 1). Although it has been suggested that anti-dsDNA antibody production can result from polyclonal B cell activation (24), the marked polyclonal hypergammaglobulinemia seen in both IL-6^+/+ and IL-6^−/− mice 3–5 mo after pristane treatment was not accompanied by an anti-dsDNA response, suggesting that other mechanisms may be more important in generating these autoantibodies. These may include effects of IL-6 on terminal B cell differentiation (17, 25).

Like anti-dsDNA, antichromatin antibodies are found at high levels in MRL/lpr and other lupus mice (15, 26). Anti-DNA responses are linked to the immune response to chromatin, since T cells specific for histones can help B cells producing anti-DNA antibodies (27). Thus, it was of interest to see if antichromatin and anti-DNA antibodies exhibit similar IL-6 dependence. Pristane induced high levels of IgG antichromatin antibodies within 3–5 mo, but only in BALB/cAn IL-6^+/+ mice (Figs. 2 and 3, C and D). Interestingly there was low level spontaneous production of these autoantibodies in aged IL-6^+/+ PBS-treated mice (Fig. 2), which was absent in IL-6^−/− animals. Thus, IL-6 may be important for the production of pristane-induced, as well as spontaneous, IgG antichromatin autoantibodies.

Since IL-6 plays an important role in the differentiation of antigen plus CD40 coactivated B cells into antibody- secreting cells (28), the lack of IgG anti-DNA/chromatin autoantibodies in IL-6-deficient mice may reflect an absolute requirement for IL-6 in the differentiation of germinal center B cells into anti-DNA/chromatin secreting plasma cells. This interpretation is consistent with previous observations that IL-6 is crucial for the development of anti-DNA/chromatin antibodies in spontaneous murine lupus models (6, 7) and that the production of these autoantibodies also is dependent on CD40–CD40 ligand interactions (29, 30). Rheumatoid factor is not produced in CD40 ligand–deficient MRL/lpr mice either, whereas anti-nRNP/Sm antibodies are (30), suggesting that the latter may be regulated differently than anti-DNA and rheumatoid factor. Like anti-dsDNA, rheumatoid factor was induced by pristane in IL-6^+/+ but not IL-6^−/− mice (data not shown), suggesting that similar regulatory mechanisms may exist in pristane-induced lupus.

The IL-6 dependence of IgG anti-DNA and chromatin autoantibodies is consistent with previous observations in the (NZB/ W)F1 lupus model (6, 7). However, (NZB/W)F1 do not produce other autoantibodies characteristic of lupus, such as anti-nRNP/Sm. Thus, it is not known whether or not the latter also are IL-6 dependent. This question was addressed in the pristane-induced lupus model, in which autoantibodies to both RNA-protein and DNA-protein autoantigens are produced.

In contrast to the absence of IgG anti-dsDNA and antichromatin autoantibodies and in IL-6^−/− mice (Figs. 1 and 2, Table 1), the frequencies of anti-nRNP/Sm autoantibodies were comparable in the IL-6^−/− and IL-6^+/+ mice (Fig. 3,A, Table 1), although levels were lower in the IL-6^−/− mice (Fig. 3,B). Similarly, the frequencies of anti-Su autoantibodies in IL-6^+/+ versus IL-6^−/− mice were comparable (Table 1). In agreement with the ELISA data, both anti-nRNP/Sm and anti-Su autoantibodies were detectable in the sera of IL-6^−/− by immunoprecipitation of ³⁵S-labeled cell extracts. We conclude that the induction of anti-nRNP/Sm and Su autoantibodies is IL-6 independent, whereas their levels are influenced by IL-6. Since IgG2a was the predominant isotype of the anti-nRNP/Sm and Su, as well as anti-dsDNA and -chromatin autoantibodies (not shown), it is unlikely that the differential effects of IL-6 on their induction reflects merely an altered ability to produce antibodies of a particular isotype or isotypes. Furthermore, analysis of total Ig levels revealed marked increases in IgG2a, IgG2b, and IgG3 from 1 to 3 mo after pristane treatment both in IL-6^+/+ and IL-6^−/− mice (data not shown). This was only slightly less dramatic in the IL-6^−/− group than in the IL-6^+/+ mice.

Although DNA binding is critical for the activation of a subset of anti-Sm B cells (31), our data argue that the induction of anti-DNA and anti-Sm responses may be at least partially independent processes. This also appears true in CD40 ligand–deficient mice (30). Whereas anti-DNA/chromatin is CD40 and IL-6 dependent, anti-nRNP/Sm antibody production may be enhanced via a T cell–dependent and CD40- and IL-6–independent mechanism (Richards, H.B., M. Satoh, J.C. Jennette, T. Okano, Y.S. Kanwar, and W.H. Reeves, et al., manuscript submitted for publication). Certain CD40 ligand–negative T cell clones retain the capacity to induce plasma cell differentiation (32). Recently, it was shown that the CD27–CD70 signaling pathway may promote B cell differentiation into IgG-secreting plasma cells, whereas CD40 ligand–CD40 signaling may be more important for maintaining B cell proliferation and differentiation into memory cells (33, 34). The role of this or other pathways of CD40 ligand– independent B cell activation in the generation of anti-nRNP/Sm and Su antibodies remains to be determined.

IL-6 is thought to influence the development of spontaneous human and murine lupus. It may play a role in the induction of autoantibodies (17) and nephritis (35, 36), as well as in polyclonal B cell activation (37, 38). Polyclonal hypergammaglobulinemia in lupus may reflect both the overexpression of IL-6 receptors on B cells (37, 39) and, in MRL/lpr mice, increasing IL-6 gene expression and serum levels with age (40). Although increased IL-6 production in MRL mice is related to the presence of the lpr gene, it is not produced by the abnormal double negative T cells (40), and may be derived from the expanded macrophage subset of these mice (6, 41).

The goal of this study was to examine the production of anti-dsDNA antibodies, long considered to be a marker of lupus nephritis (4, 5), in pristane-induced lupus. Anti-dsDNA antibodies were produced late in the course of pristane-induced lupus and were strictly IL-6 dependent, possibly reflecting the IL-6 overproduction characteristic of pristane-treated mice (8, 9). However, their onset at 8–11 mo after pristane treatment was well after the establishment of renal disease (2), arguing that the renal lesions may not be caused directly by anti-dsDNA antibodies. Nevertheless, renal disease in pristane-treated B6 IL-6^−/− mice is milder than in IL-6^+/+ controls (Richards, H.B., M. Satoh, M. Shaw, C. Libert, V. Poli, and W.H. Reeves, unpublished data), consistent with the idea that IL-6 itself may contribute to the development of nephritis (42). Interestingly, although the levels of anti-nRNP/Sm antibodies also were influenced by IL-6, their frequency was not, suggesting that they may be regulated differently from anti-DNA antibodies.

The different cytokine requirements for anti-dsDNA and anti-Sm antibody production may help to explain differences in the patterns of their expression seen clinically. For example, low levels of antichromatin antibodies developed spontaneously in aged BALB/cAn^+/+ mice but not in IL-6^−/− mice (Fig. 2), an observation that may be relevant to the increased frequency of antinuclear antibodies (43) and higher IL-6 production (44) in elderly people.

The extreme variability characteristic of anti-dsDNA antibody levels (5) contrasts with the fact that the levels of anti-Sm generally remain relatively stable over time. We speculate that, as in pristane-induced lupus, anti-dsDNA antibody production in human SLE may be IL-6 dependent. Thus, anti-dsDNA/chromatin antibody levels might be a surrogate marker for increased IL-6 production, which is associated with active SLE (37). The induction of anti-DNA antibodies in mice by lipopolysaccharide (45), a potent inducer of IL-6 (46), raises the possibility that microbial stimulation could trigger both IL-6 production and anti-DNA autoantibody production. Indeed, pristane- induced lupus is milder in specific pathogen–free than in conventionally housed mice (12). In view of the importance of anti-DNA antibodies and IL-6 in lupus nephritis, it may be of interest to examine the role of microbial stimulation of IL-6 production in the pathogenesis of lupus flares.

We are grateful to Drs. Philip Cohen for providing MRL/lpr mouse sera, David Pisetsky for providing mAb 6/02 576-2, Paul Stewart for assistance with statistical analysis, and Gary Gilkeson for advice on the anti- dsDNA ELISA. We also thank Mr. Robert L. Cheek for assistance with the antichromatin ELISA.

This work was supported by research grants from the Lupus Foundation of America and the United States Public Health Service (AR-42573, AR-44731, and T32-AR-7416). Dr. Richards is an Arthritis Foundation Postdoctoral Fellow.