Toll-like receptor 9 signaling protects against murine lupus




Hypomethylated CpG–containing DNA, which is recognized by Toll-like receptor 9 (TLR-9), has been strongly implicated in the pathogenesis of autoantibody-mediated diseases such as systemic lupus erythematosus. This study was undertaken to determine the role of TLR-9 in the MRL/+ and MRL/lpr models of murine lupus.


TLR-9–deficient MRL mice were generated by backcrossing a TLR-9–deficient allele against the MRL backgrounds by a speed congenic technique. Parameters of murine lupus were examined by routine methods. Regulatory T cell activity was assessed by autologous mixed lymphocyte reaction (AMLR), an in vitro assay for autoreactivity.


Surprisingly, TLR-9–deficient animals of both the MRL/+ and the MRL/lpr backgrounds developed more severe lupus, as judged by anti-DNA and rheumatoid factor autoantibodies, total serum Ig isotypes, lymphadenopathy, inflammatory infiltrates in the salivary gland and kidney, proteinuria, and mortality, in comparison with their TLR-9–sufficient littermates. In vitro, regulatory T cells from TLR-9–deficient animals were impaired in their ability to suppress the AMLR.


In the MRL model of murine lupus, TLR-9 signaling plays a protective role, perhaps by modulating the activity of regulatory T cells. These results contrast with findings of recent studies that implicate TLR-9 in the pathogenesis of anti-DNA responses, based in part on investigations in incompletely backcrossed TLR-9–deficient MRL/lpr mice in vivo or transgenic B cells in vitro. The present results highlight the need for caution in the assessment of disease paradigms based on the study of isolated cell populations in vitro, as well as in vivo studies of knockout animals involving non-ideal genetic models.

In humoral autoimmune diseases such as systemic lupus erythematosus, Toll-like receptor 9 (TLR-9), a receptor for hypomethylated CpG–containing DNA, has emerged as a prominent pathogenic target. Anti-DNA IgG autoantibodies can form complexes with such DNA, binding both to rheumatoid factor B cells via the B cell receptor and to TLR-9 via co-complexed DNA, resulting in costimulation in vitro (1, 2). In addition, CpG DNA can break lymphocyte tolerance (3) and can skew immune responses toward a Th1-like phenotype, promoting the production of pathogenic cytokines such as interferon-γ (IFNγ) and interleukin-12 (IL-12) (4), as well as pathogenic IgG isotypes (5–7).

Supporting this notion, Christensen et al have recently reported that TLR-9–deficient lpr mice of a mixed MRL/Mp–C57BL/6–129 background are resistant to the production of anti–double-stranded DNA autoantibodies (8). However, that study utilized TLR-9+/+ versus TLR-9−/− intercross progeny of mixed 129 × C57BL/6 TLR-9−/− mice backcrossed only once or twice (F1 or F2) against the MRL/Mp background. Such a strategy could generate animals with ∼50% or 75% MRL genes, respectively, but could have easily generated animals with significantly fewer. Indeed, inheritance of the lupus phenotype, including autoantibody production, is polygenic, with at least 24 loci in MRL mice proposed to date (9–12). Therefore, it remains unclear whether the results described by Christensen et al reflect direct effects of TLR-9 deficiency or confounding effects of disease-protective loci derived from 129 or C57BL/6, perhaps some related to chromosome 9 (on which TLR-9 resides), or a more complex interplay between the role of TLR-9 in B cells versus immune cells of other lineages, such as regulatory T cells (13, 14). The aim of the present study was to elucidate further the role of TLR-9 in murine lupus.


We generated congenic MRL/Mp-CD95+/+ (MRL/+) and MRL/Mp-CD95-lpr/lpr (MRL/lpr) mice deficient in TLR-9 by backcrossing the TLR-9–deficient allele (15) against the MRL/lpr background at least 6 times until all 24 proposed MRL disease susceptibility loci (refs.9–12 and Wu X, Peng SL: unpublished observations), as well as H-2 (16) and CD95 (17), had been bred to MRL homozygosity using sequence tagged site marker–assisted genotyping. Then TLR-9+/− MRL/lpr animals were crossed against MRL/+ animals to generate TLR-9+/− MRL-CD95+/lpr animals. Animals generated in this manner were homozygotic for MRL chromosome 9 markers D9Mit201 (65 cM; 9E4) and D9Mit151 (72 cM; 9F3), indicating the development of a ≤7 cM congenic interval containing TLR-9 (at 9F1). These animals were then intercrossed to generate TLR-9+/+ (wild-type [WT]) or TLR-9−/− (knockout [KO]) MRL/+ or MRL/lpr animals for use in this study. As expected, splenocytes of TLR-9–KO animals failed to proliferate in response to CpG oligonucleotides in vitro (results not shown). Animals were then assessed over time for the development of lupus by established protocols, including blinded assessment of histopathologic features (18).


Surprisingly, TLR-9–KO animals of both the MRL/+ and the MRL/lpr backgrounds developed more severe lupus in comparison with their TLR-9–WT counterparts. They exhibited higher levels of anti-DNA and rheumatoid factor autoantibodies, as well as increased hypergammaglobulinemia, including elevated levels of the “pathogenic” complement-fixing isotypes IgG2a, IgG2b, and IgG3 (P < 0.001 for all comparisons between TLR-9–WT and –KO mice, except for IgG3 in MRL/+ animals) (Figure 1). Furthermore, they developed accelerated end-organ disease, especially of the kidney and salivary gland, associated with more intense renal IgG deposits and proteinuria (Figure 2 and Table 1), and with increased mortality (at 24 weeks of age, 4 of 12 WT MRL/lpr mice [33%] versus 12 of 14 KO MRL/lpr mice [86%] had died; P < 0.001). Additionally, TLR-9–KO MRL/lpr animals developed more severe lymphadenopathy than their WT counterparts (mean ± SD spleen weight 187 ± 40 mg versus 432 ± 75 mg in WT and KO mice, respectively, and lymph node weight 291 ± 176 mg versus 1,338 ± 474 mg in WT and KO mice, respectively [n = 5 in each group]; P < 0.001 for both comparisons). This was associated with a dramatically increased accumulation of “double-negative” CD3+,CD4−,CD8−,B220+ lpr T cells (mean ± SD 43.7 ± 16.1 million per spleen versus 397 ± 82.1 million per spleen in WT and KO MRL/lpr mice, respectively; P < 0.0001) (Figure 3), which reflects autoreactive and conventionally activated T cells that are unable to complete activation-induced cell death due to functional Fas deficiency (17). Thus, by all parameters examined, TLR-9 was shown to play a protective role in MRL autoimmunity. The analogous findings in mice of both the MRL/+ and the MRL/lpr background suggested that this effect is independent of Fas.

Figure 1.

Serologic and histopathologic development of lupus in Toll-like receptor 9 (TLR-9)–deficient MRL mice. Sera from 12-week-old TLR-9–wild-type (WT) or -deficient (knockout [KO]) animals of MRL/+ and MRL/lpr backgrounds were assessed by A, enzyme-linked immunosorbent assay (ELISA) for total serum Ig isotypes, B, indirect immunofluorescence for the presence of antinuclear antibodies (ANAs), and C, ELISA for anti–single-stranded DNA (anti-ssDNA), anti–double-stranded DNA (anti-dsDNA), and κ chain–specific rheumatoid factor (RF) autoantibody specificities at 1:100 dilution, as previously described (18). Micrographs (original magnification × 200) show representative ANA patterns in sera obtained from TLR-9–WT or –KO MRL/lpr mice. Arrowheads indicate metaphase cells with chromatin staining. Dashed lines indicate thresholds for positivity (3 SD above the mean optical density [OD] of sera from non-autoimmune BALB/c mice). Red circles indicate sera that tested positive for anti-dsDNA antibodies by Crithidia luciliae immunofluorescence (n = 6 mice in the WT MRL/+, WT MRL/lpr, and KO MRL/lpr groups and 10 mice in the KO MRL/+ group).

Figure 2.

Representative end-organ inflammation in the salivary glands and kidneys of 12-week-old TLR-9–WT versus –KO MRL/+ and MRL/lpr mice. Note the development of a significant mononuclear infiltrate in the periacinar region of the salivary gland in WT MRL/lpr mice. This was also observed in KO MRL/+, but not WT MRL/+, animals, and was much more severe in KO MRL/lpr animals. Note also moderate-to-severe hypercellularity and mesangial thickening of renal glomeruli, associated with moderate perivasculitis, in WT MRL/lpr mouse kidneys. These features were largely absent in the kidneys of WT MRL/+ mice, present in those of KO MRL/+ mice, and more severe in those of KO MRL/lpr mice. Similarly, note the presence of renal IgG deposition, assessed by direct immunofluorescence, in the renal glomeruli of WT MRL/lpr mice, which was largely absent in WT MRL/+ mice, present in KO MRL/+ mice, and more severe in KO MRL/lpr mice. Samples are representative of at least 5 animals examined per genotype. See Figure 1 for definitions (original magnification × 200).

Table 1. End-organ disease in Toll-like receptor-9 (TLR-9)–deficient MRL mice*
TLR-9FasAge, weeksRenal diseaseSalivary gland involvementProteinuria
GlomerularVascularIg deposits
  • *

    For renal and salivary gland disease, scores were assessed by light microscopic assessment of glomeruli, perivascular, and/or periacinar regions; Ig deposition was assessed by direct immunofluorescence of IgG at the glomeruli and tubular nuclei (each assessed under blinded conditions on a − to ++++ scale, as previously described [18]). Proteinuria was rated as − (<30 mg/ml), + (30–100 mg/ml), or ++ (>100 mg/ml). Data shown, from 5 animals per genotype/age group, are representative of findings in 10–12 animals examined per group.

Figure 3.

Flow cytometry of T cell subsets in TLR-9–KO MRL/lpr mice. Twelve-week-old TLR-9–WT versus –KO MRL/lpr animals were assessed by flow cytometry for the indicated T cell populations, including “double-negative” lpr T cells (CD3+,CD4−,CD8−,B220+) and regulatory T cells (CD4+,CD25+,CD62L+). CD62L = CD62 ligand (see Figure 1 for other definitions).

Findings of recent studies have suggested that, at least in some contexts, TLR-9 signaling may in fact play a protective role in autoimmunity by inducing activity of regulatory T cells (19–21). We therefore sought to determine whether defective regulatory T cell activity might account, at least in part, for our findings. We assessed the ability of CD4+,CD25+ regulatory T cells from TLR-9–KO MRL mice to inhibit proliferation of MRL CD4+ T helper cells by autologous mixed lymphocyte reaction (AMLR), an in vitro assay for autoreactivity. Although TLR-9–KO animals did not appear to possess a defect in regulatory T cell development, at least as assessed by markers of CD4, CD25, and CD62 ligand positivity in flow cytometry (Figure 3 and data not shown), these cells were consistently less effective (by ∼25%) in KO mice than in their WT counterparts at suppressing AMLR activity (Figure 4). Although this difference (less than an order of magnitude) does not initially seem particularly significant, recent studies indicate that the progressive, acquired loss of regulatory T cell function and/or sensitivity to regulatory T cell suppression underlies the development of autoimmunity in MRL mice (22), such that a consistent and perceptible difference in regulatory T cell function, amplified over weeks to months of autoreactive T cell activity, could significantly tip the clinical environment in favor of disease progression and amplification. Such findings therefore raise the intriguing possibility that TLR-9 is required in vivo for the proper generation of regulatory T cell effector function, such that its absence results in progressively uncontrolled T cell activation, leading to enhanced accumulation of double-negative lpr T cells and autoantibodies in MRL mice—perhaps a reflection of the “hygiene hypothesis” (23).

Figure 4.

Defective regulatory T cell activity in the absence of TLR-9. Regulatory T cell activities in autologous mixed lymphocyte reactions were assessed by incubating regulatory T cells (purified CD4+,CD25+) from TLR-9–WT or –KO MRL/+ mice with bulk T helper cells (CD4+,CD25−) from TLR-9–WT MRL/+ mice at the indicated ratios, with or without an equal number of irradiated autologous splenocytes (antigen-presenting cells [APCs]). Two days later, proliferation was assessed by bromodeoxyuridine (BrdU) incorporation (28). Note the ability of CD4+ MRL mouse T cells to proliferate spontaneously, to some degree by themselves but particularly well when incubated with autologous APCs; under both conditions, proliferation was suppressed in the presence of increasing numbers of purified CD4+,CD25+ T cells from WT MRL mice. At maximal suppressive ratios, CD4+,CD25+ T cells from KO mice were ∼25% less efficient than their WT counterparts (maximal suppression 60% versus 80%). Data in the lower panel were calculated from the data in the upper panel, with % inhibition reflecting the ratio of the OD at each data point relative to maximum OD (CD4 plus APC, no CD4+,CD25+ cells). Values are the mean and SD from 5 animals. See Figure 1 for other definitions.


Regardless of the specific mechanisms by which TLR-9 regulates lupus, the present results clearly indicate that the TLR-9 signaling pathway is not required for the pathogenesis of lupus-related autoantibodies and autoimmunity, at least in the MRL mouse model, and indeed, that it plays a protective role. These findings suggest that previous results reported by Christensen et al (8), from studies with a partially backcrossed TLR-9 allele, were confounded by the incomplete genetics of the backcross, which likely allowed protective genes associated with the 129 and/or C57BL/6 backgrounds to segregate with TLR-9 deficiency.

Alternatively, our findings may reflect differences in housing conditions, with different bacterial or other microorganismal flora, diets, etc. possibly altering ongoing Th1/Th2 cytokine or other inflammatory responses, as has been suggested to explain the differences in penetrance of the 16/6 antiidiotype murine lupus model as well as models of other autoimmune diseases such as NOD diabetes, in different environments (24, 25). As such, the potential influence of bacterial flora and other environmental factors in the activity and effects of CpG-containing DNA stimuli and/or TLR-9 signaling in murine lupus may be of interest, e.g., via the study of TLR-9–KO lupus-prone animals in appropriate gnotobiotic settings. Alternatively, TLR-9 may indeed play a role in the production of anti-DNA autoantibodies by B cells, but its role in regulatory T cells may be dominant in the MRL background, reflecting complex interplays between the actions of TLR-9 in different immune cell lineages as well as genetic backgrounds (13, 14).

A likely confounding variable in the study by Christensen et al is the 129 chromosome 9, where TLR-9 resides. Although to our knowledge no known loci on chromosome 9 have been implicated in promoting autoimmunity in MRL mice, most MRL mapping studies have focused on specific end-organ disease manifestations, such as lymphadenopathy (11), coronary vasculitis (9), sialadenitis (10), glomerulonephritis (12), or arthritis (26)—as opposed to autoantibodies, which have been suggested to be more relevant with regard to TLR-9 (1, 8). Vidal et al sought genetic modifiers of anti-DNA production, but examined interactions between the C57BL/6 and MRL genomes (11) as opposed to interactions between the 129 and MRL genomes, which would be relevant to the mutant TLR-9 allele used in the present study, which was generated on the 129 strain (15).

The present results suggest that TLR-9 may, directly or indirectly, also regulate cellular autoimmunity. In this regard, it is interesting to note that Christensen and colleagues observed diminished levels of anti-DNA autoantibodies but almost no effect on glomerulonephritis and perivasculitis in TLR-9–KO mice (8). These seemingly dichotomous findings might, on the one hand, reflect reduced autoantibody titers as a result of confounding disease susceptibility loci, which ordinarily would confer reduced renal disease, but on the other hand, they might arise from accentuated cellular autoimmunity due to TLR-9 deficiency—perhaps related to defective regulatory T cell activity and resulting in an end-organ disease phenotype indistinguishable from that in WT counterparts. Such a model is consistent with analogous findings in the types I and II IFN systems in MRL mice, where MRL animals that are double-deficient in the receptors for both types I and II IFN resemble their WT counterparts because of opposing roles of type I versus type II IFN (18).

In this sense, the role of CpG-containing DNA and TLR-9 signaling in autoimmunity seems to be context dependent, but is predominantly disease-protective in the MRL/+ and MRL/lpr lupus models. Therefore, as is the case with interpretation of findings regarding type I IFN and IL-18, caution is warranted in the assessment of disease paradigms that rely heavily on the study of isolated cell populations in vitro, and results of studies that utilize non-ideal genetic intercrosses to assess the role of specific mutant loci in polygenic diseases (18, 27).


We thank Ling Lin for technical assistance.