Role of TLR9 in Anti-Nucleosome and Anti-DNA Antibody Production in lpr Mutation-Induced Murine Lupus1

Systemic lupus erythematosus is characterized by the production of autoantibodies directed against nuclear Ags, including nucleosome and DNA. TLR9 is thought to play a role in the production of these autoantibodies through the capacity of nuclear immunogenic particles to interact both with BCR and TLR9. To determine the role of TLR9 in SLE, C57BL/6-lpr/lpr-TLR9−/− and TLR9+/+ mice were analyzed. The abrogation of TLR9 totally impaired the production of anti-nucleosome Abs, whereas no difference was observed in the frequency of anti-dsDNA autoantibodies whose titer was strikingly higher in TLR9−/− mice. In addition a higher rate of mesangial proliferation was observed in the kidney of TLR9-deficient animals. These results indicate that in C57BL/6-lpr/lpr mice, TLR9 is absolutely required for the anti-nucleosome Ab response but not for anti-dsDNA Ab production which is involved in mesangial proliferation.

S ystemic lupus erythematosus (SLE) 3 is a nonorgan-specific autoimmune disease characterized by the production of autoantibodies directed against self-nuclear Ags among which anti-nucleosome and anti-dsDNA Abs are thought to play a major role in disease pathogenesis, in particular nephritis (1,2). The molecular and cellular mechanisms responsible for the production of these autoantibodies still remain unprecised. It is well established that B (3) and T lymphocytes (4,5) participate in disease pathogenesis, but the sustained activation of autoreactive B cells constitutes the common and major characteristic of lupus diseases observed in both humans and mice. The activation of lupusassociated autoreactive B cells is Ag driven and some studies suggest that nucleosome, an ubiquitous endogenous product of cell apoptosis, could be the primary and major immunogen leading to the production of both anti-nucleosome and anti-dsDNA Abs (6). Others also showed that exogenous Ags such as bacterial DNA, could induce an anti-dsDNA Ab response (7). It is also demonstrated that pathogen-associated molecular patterns can costimulate autoreactive B cells through their interaction with TLR and participate in the breakage of B cell tolerance. Indeed, TLR9, which binds bacterial DNA via unmethylated CpG sequences, is thought to play a role in the B cell response to DNA (8 -10) and the pathogenesis of SLE (11,12,13). Apoptotic-generated DNA fragments can also bind to TLR9 and activate B cells by dual engagement of BCR and TLR9 (14,15). These observations invited us to determine the role of TLR9 in the production of lupus-related au-toantibodies and the spontaneous development of murine lupus. For this purpose, we compared the phenotype, i.e., lymphadenopathy, splenomegaly, autoantibody production and renal injury, of C57BL/6 (B6) mice which do not spontaneously develop lupus, C57BL/6-lpr/lpr (B6-lpr/lpr) mice which present a mild type of lupus related to Fas deficiency and C57BL/6-lpr/lpr-TLR9 Ϫ/Ϫ (B6-lpr/lpr-TLR9 Ϫ/Ϫ ) mice.

Flow cytometry analysis and mAbs
Splenocytes and lymph node (LN) cells from the inguinal sites were purified using Lympholyte-M (Cliniscience). Cells at the interface were washed with PBS and incubated for 20 min on ice with optimal amounts of FITC-or PE-conjugated primary Abs diluted in PBS. The following Abs were obtained from BD PharMingen: CD3 (CD3 -chain), CD4 (L3T4), CD8a (Ly-2), CD19, and CD45R/B220 (RA3-6B2) and used at pretitrated dilutions. After washes, cell staining was analyzed using an Epics XL (Beckman Coulter).
DNA-stripped nucleosome (Euroimmun) as the Ags on the solid phase. For anti-cardiolipin (CL), anti-ssDNA Ab and rheumatoid factor detection, plates were incubated, respectively, with 10 g/ml bovine heart CL, 1 g/ml calf thymus ssDNA (Sigma-Aldrich), or 10 g/ml rabbit IgG (Jackson ImmunoResearch Laboratories). Mouse sera were diluted 1/100 and incubated for 2 h at room temperature. Bound IgG or IgM were detected either with alkaline phosphatase-conjugated goat anti-mouse IgG or antimouse IgM (Rockland). The cutoff values were 0.5 for anti-ssDNA, 0.276 for anti-dsDNA, 0.22 for anti-nucleosome, 0.109 for anti-CL Abs, and 0.268 for rheumatoid factor.

Proteinuria and histopathology
Urine was collected for proteinuria determination at different ages using reagent strips (Multistix; Bayer). Mice were sacrificed at 6 mo of age and renal examination was performed. Half kidneys were fixed in 4% neutral formaldehyde, sectioned, and stained with trichrome green. We analyzed the presence of interstitial lymphoid infiltration and glomerulonephritis including mesangial cell proliferation. A nephritis index was determined based on the mesangial proliferation (0, Ͻ3 cells per glomeruli; 1, mild with 3-10 cells per glomeruli; 2, moderate with 10 -15 cells per glomeruli; and 3, severe with Ͼ15 cells per glomeruli). Frozen kidney sections were stained with FITC-conjugated goat anti-mouse IgG (Sigma-Aldrich). The topography and intensity of glomerular Ig deposits were assessed by two renal pathologists (P.C. and A.F.).

Statistical analysis
Weight of lymphoid organ and OD measured in mouse sera were compared using the Mann-Whitney U test. Mesangial proliferation and proteinuria observed in B6-lpr/lpr-TLR9 Ϫ/Ϫ and B6-lpr/lpr-TLR9 ϩ/ϩ mice were compared using Fisher's exact test. Comparison of OD measured in mouse sera at different age was performed using the Friedman test.

Anti-chromatin autoantibodies in TLR9-deficient mice
Sera from B6-lpr/lpr-TLR9 Ϫ/Ϫ and B6-lpr/lpr mice were collected from 1 to 6 mo of age and examined by indirect immunofluorescence (IIF) analysis on HEp-2 cells. Although 10 of 11 B6-lpr/lpr sera exhibited an homogenous staining of the nucleus, none of 14 B6-lpr/lpr-TLR9 Ϫ/Ϫ sera presented this IIF pattern (Fig. 3A) since 11 of 14 bound to nucleoli ( p Ͻ 0.01). The dramatic change of the nuclear staining pattern in TLR9 Ϫ/Ϫ mice indicates that the abrogation of TLR9 expression either impaired the production of the autoantibody population giving the homogenous nuclear staining, demasking an antinucleolar activity, or eventually changed the autoantigen binding properties of autoantibodies. Since a homogenous nuclear staining is given by autoantibodies directed against chromatin and/or its components, we performed an autoantibody analysis of wild-type and TLR9-deficient B6-lpr/lpr mouse sera by using dsDNA-and DNA-stripped nucleosome ELISAs. No significant difference in the frequency of anti-dsDNA autoantibodies was observed between B6-lpr/lpr and B6-lpr/lpr-TLR9 Ϫ/Ϫ mice: they were detected in 7 of 11 B6-lpr/lpr and 13 of 16 B6-lpr/lpr-TLR9 Ϫ/Ϫ mice at 3 mo of age ( Fig. 3B; p ϭ 0.160) and in 9 of 11 B6-lpr/lpr and 11 of 12 B6-lpr/lpr-TLR9-deficient mice at 6 mo of age ( p ϭ 0.466). Interestingly, the titer of IgG anti-dsDNA Abs was higher in B6-lpr/lpr-TLR9 Ϫ/Ϫ ( p Ͻ 0.0021; Fig. 4, A and B), whereas the titer of IgM anti-dsDNA Abs was not significantly different between these two strains of mice ( p ϭ 0.46). To confirm these results, sera were tested by IIF on Crithidia luciliae, a substrate commonly used for the detection of anti-dsDNA in SLE (18). Among 11 sera collected from 6-mo-old B6-lpr/lpr mice, 8 contained Abs staining the C luciliae kinetoplast. Similarly, 9 of 12 six-mo-old B6-lpr/lpr-TLR9 Ϫ/Ϫ mice displayed the same IIF pattern (Fig. 4C) at dilutions 1/5-1/320 (Fig. 4D). In contrast, when the nucleosome-binding activity was examined, none of the 3-mo-old B6-lpr/lpr-TLR9 Ϫ/Ϫ mice had anti-nucleosome Abs (n ϭ 16) that were detected at high levels in 10 of 11 B6-lpr/lpr mice ( Fig. 3B; p Ͻ 0.0001). Although two sera from 6-mo-old TLR9-deficient mice weakly bound to nucleosome, the difference in nucleosome-binding activity between the two groups was again highly significant ( p Ͻ 0.0001). Moreover, anti-histone Abs were not detected in TLR9-deficient sera (data not shown). Thus, in B6-lpr/lpr mice, TLR9 deficiency dramatically impaired antichromatin Ab production and dissociated the anti-dsDNA from the anti-nucleosome Ab response. These results indicate that TLR9 plays an essential role in the B cell response against nucleosomeassociated proteins.

Rheumatoid factor, anti-ssDNA, and anti-CL in TLR9-deficient mice
We also looked for the presence of other autoantibody populations frequently detected in lupus mice (Fig. 3B). Rheumatoid factors were present in 93% of 3-mo-old B6-lpr/lpr-TLR9 Ϫ/Ϫ and 90% of B6-lpr/lpr mice, and their titers were not different between the two groups. Anti-CL were detected in 37.5 and 18%, respectively, of B6-lpr/lpr-TLR9 Ϫ/Ϫ and B6-lpr/lpr sera ( p ϭ 0.405), indicating that anti-phospholipid Ab production is not controlled by TLR9 as reported by others (19). Similarly, TLR9 deficiency did not affect the production of anti-ssDNA Abs that were present in 81 and 72.7% of sera from B6-lpr/lpr-TLR9 Ϫ/Ϫ and B6-lpr/lpr, respectively. Thus, in contrast to the anti-nucleosome Ab response, the production of autoantibodies directed against IgG, CL, and ssDNA was not impaired in TLR9-deficient mice.

Activation of TLR9 with CpG in lpr/lpr mice
To determine the role of TLR9 stimulation in autoantibody production, we immunized B6-lpr/lpr and B6-lpr/lpr-TLR9 Ϫ/Ϫ mice from 2 mo of age (before the appearance of IgG autoantibodies) with hypomethylated CpG-DNA, twice a month and determined the levels of anti-dsDNA and anti-nucleosome Abs. Fig. 5 shows the time course of anti-dsDNA and anti-nucleosome Ab production in B6-lpr/lpr and B6-lpr/lpr-TLR9 Ϫ/Ϫ immunized mice. Antinucleosome Ab response did not significantly differ from that observed during the spontaneous development of the disease of TLR9 ϩ/ϩ mice but was blocked in TLR9-deficient mice (Fig. 5A). Thus, direct activation of TLR9 with hypomethylated CpG is not able per se to induce an anti-nucleosome Ab response in TLR9 ϩ/ϩ or TLR9 Ϫ/Ϫ mice, as has been demonstrated in normal mice (20). We found that the absence of TLR9 did not impair the production of anti-DNA autoantibodies and immunization with CpG even induced a slight but not significant ( p ϭ 0.46) increase of anti-DNA in B6-lpr/lpr TLR9 ϩ/ϩ and in B6-lpr/lpr TLR9 Ϫ/Ϫ mice (Fig. 5B). Thus, TLR9 may not be the unique receptor of CpG able to induce anti-DNA response.

Discussion
Our study shows that B6-lpr/lpr mice develop a lupus-like autoimmune disease characterized by a massive lymphoproliferation, anti-DNA autoantibodies, immune complex deposition, and glomerulonephritis and that TLR9 deficiency does not dramatically affect this phenotype. TLR9-deficient B6-lpr/lpr mice exhibited increased immune activation, as demonstrated by an augmentation of LN and splenic weights and hypergammaglobulinemia that is in agreement with previous results obtained in (TLR9 Ϫ/Ϫ ϫ MRL/ Mp lpr/lpr ) mice (19,23). These observations indicate that the TLR9 signaling pathway could also exert a negative control of lymphocyte proliferation and accumulation in the context of sustained cell activation by endogenous TLR9 ligands (24,25).
despite the persistence of anti-dsDNA. These findings are in agreement with those recently observed by Wu and Peng (23) but are different from that previously reported in (TLR9 Ϫ/Ϫ ϫ MRL/ Mp lpr/lpr ) mice whose anti-nuclear Ab profile switched from a homogenous to a speckled pattern and corresponded to the abrogation of both anti-dsDNA and anti-nucleosome Abs (19). An explanation for the discrepancy between Christensen's data and ours, might be the different genetic background of TLR9-deficient mice (lupus-prone MRL/Mp vs C57BL/6) used in the two studies. Lupus is a polygenic disease and different lupus susceptibility loci have been described to be involved in the development of the disease (26), among them the Sle1 locus triggered the formation of anti-histone/DNA Abs (27). In lupus-prone mice of the MRL/lpr background, anti-nucleosome autoantibodies appear early in disease and precede the emergence of anti-dsDNA Abs, suggesting that anti-dsDNA are produced in response to nucleosome (6,28). This transition does not occur in the B6.Sle1 strain, suggesting that other additional mechanisms are responsible for the activation of anti-dsDNA B cells in these mice (27). Our results clearly demonstrate that the autoimmune response against dsDNA in B6-lpr/ lpr mice could emerge from a TLR9-independent mechanism. Nucleosomes released from apoptotic cells in lupus are somehow enriched in self-hypomethylated DNA sequences (29), suggesting that the anti-nucleosome response could be strictly dependent on TLR9, whereas the anti-DNA could not. The existence of the TLR9-independent pathways in the anti-DNA response has been suggested in several studies. DNA or ODN that do not contain CpG motifs and vertebrate genomic DNA can effectively synergize with specific Ag to costimulate murine B cells, probably via a TLR9-independent way (30,31). Mammalian DNA-Ig complexes, that provide a potent activation of B cells via TLR9, can activate dendritic cells by both TLR9-dependent and -independent pathways (32). Moreover, macrophage activation by endogenous DNA, escaping to lysosomal degradation, requires TLR9-dependent and -independent pathways (33) and is not abrogated in MyD88-deficient mice (34). Other TLR, such as TLR3, TLR7, and TLR8, involved in the recognition of nucleic acids of viral origin (dsRNA and ssRNA) (35,36) have been shown to play an important role in the development of autoantibodies present in lupus (37,38). In the absence of TLR9, these receptors or others that also recognized nucleic acids, could be involved in anti-DNA Ab production. In any case, the common feature of TLR9 Ϫ/Ϫ -deficient lupus mice expressing different genetic backgrounds is the impairment of the anti-nucleosome Ab response that clearly shows the specific requirement of TLR9 for this response.
The assessment of the renal disease in B6-lpr/lpr-TLR9 Ϫ/Ϫ mice showed glomerular deposits and mesangial proliferation similarly to TLR9-deficient MRL-lpr/lpr mice that also developed substantial glomerular protein deposition (19), despite the absence of anti-dsDNA and anti-nucleosome Abs. The authors suggest that other specificities, anti-Sm for example, may play a pathogenic role in their models. In our study, the persistence and even the increased production of anti-dsDNA Abs could explain the elevated mesangial proliferation in TLR9-deficient mice since it has been clearly demonstrated that the level of anti-DNA Abs is associated with the importance of nephritis (21,22).
Our data, as do the other recent reports concerning TLR9-deficient lupus mice (23), provide clear evidence for a protective role of TLR9 in the development of the lupus disease since all TLR9 Ϫ/Ϫ lupus mice present more intense renal deposits, proteinuria, and lymphadenopathy than their normal counterparts. Paradoxically, TLR9 inhibition by synthetic ODN can also have beneficial effects on the disease progression in lupus nephritis (39,40). Whereas the mechanisms responsible for this ambivalent role of TLR9 remain unclear, it should be noted that TLR9 activation could modulate the immune response from a Th1 to a Th2 pattern. Indeed, CpG-ODN have been demonstrated to induce IL-10 in lupus B cells, particularly those with the marginal zone B cell phenotype (41,42) and then to down-regulate proinflammatory cytokines (42). IL-10-producing B cells are known to regulate autoimmunity (43) and IL-10-deficient MRL-Fas lpr mice developed severe lupus (44). Interestingly, these mice present an exacerbated disease phenotype with early appearance of skin lesions, more severe renal disease and augmented production of IgG2a anti-dsDNA autoantibodies, a phenotype resembling that observed in our TLR9-deficient lupus mice.
Taken together, our results indicate that the engagement of TLR expressed at the surface of B cells (or other cells of the immune system) by endogenous or exogenous ligands participate in the expression of systemic autoimmunity by selectively inducing certain autoantibody populations and lymphocyte proliferation. Interestingly, TLR engagement may also lead to receptor inhibition (39,  , ϫ40). C, The nephritis index indicates that renal lesions were more severe in B6-lpr/lpr-TLR9 Ϫ/Ϫ mice. Histological renal disease was assessed in B6-lpr/lpr (n ϭ 11) and B6-lpr/ lpr-TLR9 Ϫ/Ϫ mice (n ϭ 12) at 6 mo of age. Glomerular disease was scored from 0 to 3 for each mouse. The statistical comparison was performed using the Mann-Whitney U test. 40), which, along with our observations and those previously reported by others (19,23), allows us to consider TLR as a potential target of new therapeutic agents of SLE (10).