Systemic lupus erythematosus (SLE) is a chronic, life-threatening autoimmune disorder, leading to multiple organ pathologies and kidney destruction. Analyses of numerous murine models of spontaneous SLE have revealed a critical role for endosomal TLRs in the production of autoantibodies and development of other clinical disease manifestations. Nevertheless, the corresponding TLR9-deficient autoimmune-prone strains consistently develop more severe disease pathology. Injection of BALB/c mice with 2,6,10,14-tetramethylpentadecane (TMPD), commonly known as pristane, also results in the development of SLE-like disease. We now show that Tlr9−/− BALB/c mice injected i.p. with TMPD develop more severe autoimmunity than do their TLR-sufficient cohorts. Early indications include an increased accumulation of TLR7-expressing Ly6Chi inflammatory monocytes at the site of injection, upregulation of IFN-regulated gene expression in the peritoneal cavity, and an increased production of myeloid lineage precursors (common myeloid progenitors and granulocyte myeloid precursors) in the bone marrow. TMPD-injected Tlr9−/− BALB/c mice develop higher autoantibody titers against RNA, neutrophil cytoplasmic Ags, and myeloperoxidase than do TMPD-injected wild-type BALB/c mice. The TMP-injected Tlr9−/− mice, and not the wild-type mice, also develop a marked increase in glomerular IgG deposition and infiltrating granulocytes, much more severe glomerulonephritis, and a reduced lifespan. Collectively, the data point to a major role for TLR7 in the response to self-antigens in this model of experimental autoimmunity. Therefore, the BALB/c pristane model recapitulates other TLR7-driven spontaneous models of SLE and is negatively regulated by TLR9.
Systemic lupus erythematosus (SLE) is a systemic autoimmune disorder promoted by a combination of genetic susceptibilities and environmental factors. Numerous spontaneous models of SLE that have been used to explore the genetic basis for this disease and to further analyze SLE pathogenesis. Systemic autoimmunity can also be induced in non–autoimmune-prone mice with a single injection of the proinflammatory hydrocarbon oil 2,6,10,14-tetramethylpentadecane (TMPD), also known as pristane, and these mice exhibit many features of human lupus (1). Immunostimulatory hydrocarbon oils such as squalene have been used as vaccine adjuvants, and in some instances these vaccines have been associated with the induction of autoimmunity in both animals (2) and humans (3–6). Squalene by itself can also induce autoantibody production (4). Therefore, the TMPD model has relevance to human disease.
As in other models of murine SLE, disease onset and progression in TMPD-injected mice is highly dependent on endosomal TLRs. In C57BL/6J (B6) mice, the i.p. injection of TMPD has been found to induce the rapid and persistent influx of both Ly6Chi inflammatory monocytes and Ly6CmedLy6G+ granulocytes into the peritoneal cavity, and the extravasation of Ly6Chi inflammatory monocytes detected at 14 d postinjection is dependent on both type I IFNs and TLR7 (7). Moreover, in contrast to TMPD-injected wild-type (WT) B6 mice, TMPD-injected Tlr7−/− B6 mice failed to make autoantibodies reactive with Smith Ag/ribonucleoproteins, argonaute 2, and other RNA-binding proteins, even though Abs against dsDNA, as detected in a Crithidia luciliae kinetoplast immunofluorescent assay, were still produced. Additionally, much less IgG was deposited in the kidneys of these Tlr7−/− B6 mice, and as a result they developed less severe nephritis and exhibited markedly improved survival rates (7, 8). Overall, TLR7 deficiency clearly protects B6 mice from TMPD-induced inflammation and autoimmunity.
The impact of TLR9 deficiency on TMPD-injected mice is less clearcut. As predicted from in vitro studies (9, 10), TLR9-deficient autoimmune-prone mice do not make anti-dsDNA or anti-nucleosome Abs, as detected by homogeneous nuclear and mitotic plate staining of HEp-2 cells. Shlomchik and colleagues (11) demonstrated that loss of TLR9 expression in MRL/lpr mice correlated with loss of binding to chromatin and nucleosomes, but not dsDNA. However, quite unexpectedly, in essentially all spontaneous models of SLE (including MRL/lpr, B6/lpr, Ali5 B6, Nba2.Yaa, B6.Nba2, and WASp-deficient B6 mice), Tlr9−/− mice produce significantly increased titers of autoantibodies directed against RNA-associated autoantigens and invariably develop more severe renal disease (11–17). In contrast, TMPD Tlr9−/− B6 mice have been reported to develop less severe peritoneal inflammation than does the TLR-sufficient control group (7), and also less severe disease pathology (18). However, TMPD-injected BALB/c mice more closely mimic the renal complications and other clinical manifestations of human disease than do TMPD-injected B6 mice (1). Therefore, we decided to reexamine the role of TLR9 in the TMPD-treated BALB/c model of lupus. We show in the present study that TMPD-treated Tlr9−/− BALB/c exhibit a more rapid progression to lupus nephritis and decreased survival compared with TLR-sufficient TMPD-treated control groups. Disease severity is preceded by an early increase in the number of Ly6Chi inflammatory monocytes, a stronger IFN signature, and increased TMPD-driven myelopoiesis. Therefore, TMPD-injected Tlr9−/− BALB/c mice do in fact develop exacerbated autoimmune disease and provide a useful model for evaluating the negative regulatory role of TLR9 in murine SLE.
Materials and Methods
Wild-type BALB/c mice were purchased from The Jackson Laboratory. Tlr7−/− and Tlr9−/− mice, provided by Dr. S. Akira (Osaka University), were backcrossed 10 generations to the BALB/c background. All mice were bred and maintained at the Department of Animal Medicine of the University of Massachusetts Medical School in accordance with the regulations of the American Association for the Accreditation of Laboratory Animal Care, and all protocols were approved by the Institutional Animal Care and Use Committee.
For long-term experiments (5–6 mo) 8- to 12-wk-old female mice received a single i.p. injection of 0.5 ml 2,6,10,14-tetramethylpentadecane (TMPD, Sigma-Aldrich) (19). For short-term experiments (4 or 14 d) age-matched female or male mice, as indicated in the figure legends, were injected at 6–8 wk of age.
Single-cell suspensions obtained from the peritoneal cavity, spleen, kidney, and bone marrow (BM) were stained with the following Abs: anti-CD11b (M1/70), anti-CD86 (GL1), and anti-Ly6G (1A8) from BD Biosciences, anti-CD138 (281-2) and anti-GL7 (eBioscience), and anti-Ly6C (ER-MP20, Serotec). Hematopoietic subsets (hematopoietic stem progenitor cells, common myeloid progenitors [CMPs], and granulocyte myeloid progenitors [GMPs]) were stained as described in Murphy et al. (20). For TLR protein expression, cells were fixed with 2% PFA, permeabilized with 0.1% saponin, and then stained with the biotinylated monoclonal anti-mTLR7 A94B10. A fluorescently labeled streptavidin tetramer (SouthernBiotech) was added as detecting reagent. Flow cytometric analysis was carried out using a BD LSR II with FACSDiva software (BD Biosciences) and analysis was conducted with FlowJo software (Tree Star).
RNA from total peritoneal lavage cells 4 d after TMPD injection was extracted using the RNeasy mini kit (Qiagen) and cDNA was made by reverse transcription using the iScript cDNA synthesis kit (Biorad). Quantitative RT-PCR was performed using SYBR Green PCR Master Mix (Bio-Rad) as described previously (19). IFN-γ–inducible protein (IP)10, IFN-stimulated gene (ISG)15, CCL5, MX dynamin-like GTPase 1 (Mx1), IFN regulatory factor (IRF7), and IFN-activated gene (IFI)204 expression are presented relative to β-actin expression. Primer sequences are as follows: IP10, 5′-GCTGCCGTCATTTTCTGC, 3′-TCTCACTGGCCCGTCATC; ISG15, 5′-AGTCGACCCAGTCTCTGACTCT, 3′-CCCCAGCATCTTCACCTTTA; CCL5, 5′-TGCAGAGGACTCTGAGACAGC, 3′-GAGTGGTGTCCGAGCCATA; Mx1, 5′-GAGCAAGTCTTCTTCAAGGATC, 3′-GGGAGGTGAGCTCCTCAGT; IRF7, 5′-AGCGTGAGGGTGTGTCCT, 3′-TCTTCGTAGAGACTGTTGGTGCT; IFI204, 5′-GCCAGCCCTAAGATCTGTGAT, 3′-TCTTTCGGTTCACTGTTTTCTTG; and β-actin, 5′-CTAAGGCCAACCGTGAAAAG, 3′-ACCAGAGGCATACAGGGACA.
Generation of BM–derived macrophages and dendritic cells
M-CSF– and GM-CSF–derived BM macrophages (BMDMs) and dendritic cells (BMDCs) were induced by culturing with 40 ng/ml M-CSF (R&D Systems) or 20 ng/ml GM-CSF (Immunotools), respectively, in 10% FCS, 1% ciprofloxin containing RPMI 1640 medium for 7 d. Differentiated BMDMs and BMDCs were seeded on 96-well cell culture plates (5 × 104 cells/well) and stimulated with the doses indicated in Fig. 4 of LPS (from Escherichia coli 0111:B4), Pam3CSK4 (Sigma-Aldrich), R848, RNA40, or CpG– oligodeoxynucleotide 1826 (InvivoGen) at 37°C in a 5% CO2 atmosphere for 24 h before collecting supernatants. IL-6 concentrations in the supernatants were determined by ELISA (eBioscience) according to the manufacturer’s instructions.
IgG2a specific Ab-forming cells of cells isolated from spleens of mice 5 mo after pristane injection were measured by ELISPOT assay as previously described (21).
Ab and autoantibody titers
Total serum IgG1, IgG2a, IgG2b, IgG3, IgM, and κ L chain were measured by ELISA as described previously (22). Anti-nuclear Abs (ANAs) were detected by immunofluorescence on HEp-2 slides (Antibodies) (19). Anti-neutrophil cytoplasmic Abs (ANCAs) were detected by immunofluorescence on ethanol-fixed granulocytes (AESKUSLIDES) with serum diluted at 1:50 and using FITC-conjugated rat anti-mouse κ (SouthernBiotech) in Vectashield antifade mounting medium (Vector Laboratories), and images were captured on a Leica TCS SP2 microscope (Leica Microsystems, Buffalo Grove, IL) and processed in Adobe Photoshop. Abs against myeloperoxidase (MPO) were detected by ELISA by using recombinant human MPO Ag-coated wells (Ala49-Ser745; R&D Systems), and a goat anti-mouse IgG HRP secondary Ab (Sigma-Aldrich A8924) with TMB substrate (Dako) as detecting reagent. RNA was detected by ELISA using poly-l-lysine–coated wells (Sigma-Aldrich P4832) coated with yeast RNA (Sigma-Aldrich R6750), as described previously, and reported as microgram equivalent of the anti-RNA autoantibody BWR4 (23).
Renal pathology and cell infiltration
Albuminuria was measured by ELISA as described (19). Sections of paraffin-embedded kidneys were stained with H&E and scored in a blinded fashion to determine a glomerular and interstitial inflammation score as described (19). For immunofluorescence, kidneys were fixed in 4% formalin solution overnight and frozen in Tissue-Tek embedding medium (Sakura OCT compound) after dehydration in 30% sucrose solution. Total immune complex deposition was detected with Alexa Fluor 488 F(ab′)2 fragment goat anti-mouse IgG (H+L) Ab (Life Technologies), and IgG2a or IgG2b glomerular immune complexes were detected with FITC-conjugated goat F(ab′)2 anti-mouse IgG2a, or goat anti-mouse IgG2b (γ2b chain specific) Ab (SouthernBiotech). Infiltrating neutrophils were detected by using a monoclonal rat anti-mouse Ly6G Alexa Fluor 647–conjugated Ab (BioLegend). Nuclei were counterstained with DAPI (Invitrogen). Images were captured on a Nikon E600 inverted light microscope or on a Leica TCS SP8 microscope (Leica Microsystems, Buffalo Grove, IL) and processed in Adobe Photoshop. Using the NIS-Elements imaging software BR3.10 (Nikon), 15–20 randomly picked glomeruli per mouse were circled and analyzed for the mean fluorescence intensity emitted in the green channel. Additionally, 20–30 randomly picked glomeruli per mouse were circled and the percentage of Ly6G+ glomeruli of the total number of glomeruli counted was determined.
For flow cytometry on renal single-cell suspensions, kidneys were cut into small pieces, transferred to gentleMACS tubes (Miltenyi Biotec, Bergisch-Gladbach, Germany) and digested in a collagenase type I (10 mg/ml), DNase (200 U/ml), HBSS mixture for 30 min at room temperature. Renal tissue fragments were then processed using the gentleMACS program D for 30 s, incubated for further 15 min at room temperature within the digestion buffer, and processed a second time using the gentleMACS program D for 30 s. Cells were filtered through a 70-μm filter, washed twice with cold HBSS, and RBCs were lysed. Single-cell suspensions were counted and stained for flow cytometric analysis.
All data were analyzed by a non–parametric Mann–Whitney U test or Kruskal–Wallis test, as appropriate, using GraphPad Prism software (GraphPad Software, San Diego, CA). Results are reported as mean ± SEM. A p value <0.05 was considered significant. Multiple comparisons were analyzed by one-way ANOVA, followed by a Bonferroni multicomparison test or Kruskal–Wallis test.
Rapid progression of lupus nephritis and decreased survival in TMPD-treated Tlr9−/− BALB/c mice
In spontaneous models of murine SLE, Tlr9−/− mice consistently develop exacerbated renal disease. To directly compare TMPD-injected BALB/c mice to these other models, WT and Tlr9−/− mice were injected with either PBS or TMPD and urine samples were collected at monthly intervals and analyzed for proteinuria. By 116–137 d after TMPD injection, >30% of TMPD-injected Tlr9−/− mice, and none of the TMPD-injected WT BALB/c mice, showed elevated levels of albuminuria. The frequency of proteinuric mice as well as urine albumin concentrations increased with time, and by 160–180 d after TMPD injection >70% of the surviving Tlr9−/− mice had urine albumin levels that were at least a 100-fold over physiologic urine protein levels. A much lower frequency of the TMPD-injected WT mice developed proteinuria, and the urine albumin levels in these mice were <10-fold over baseline (Fig. 1A, Supplemental Fig. 1A). None of the PBS injected WT or Tlr9−/− mice developed proteinuria during this time course. At 5 mo after TMPD injection, five mice from each group were euthanized and kidney sections were compared with PBS-injected mice for histological evidence of renal disease. The Tlr9−/− BALB/c kidneys showed more severe glomerulonephritis associated with crescent formation and sclerosis (Fig. 1B, 1C). The TMPD-injected WT BALB/c kidneys exhibited less pathology. Consistent with severe nephritis, the TMPD-injected Tlr9−/− mice showed a markedly reduced survival rate compared with the TMPD-injected WT BALB/c mice (Fig. 1D). Importantly, even our oldest (>6 mo) untreated WT and Tlr9−/− mice failed to exhibit any sign of renal disease, reduced survival rates, or immune activation (Fig. 1B–D, Supplemental Fig. 1). Therefore, TLR9 deficiency on the BALB/c background alone does not predispose to systemic autoimmunity. Rather, TLR9 expression exerts a protective function in the context of TMPD-induced autoimmunity in BALB/c mice.
Increased numbers of TLR7-expressing Ly6Chi inflammatory monocytes in Tlr9−/− TMPD-treated mice
Prior studies suggested that TLR9-deficient B6 mice developed an attenuated acute response to TMPD (7, 18). Considering the apparent negative regulatory role of TLR9 on both nephritis and survival in TMPD-injected BALB/c mice, we decided to re-examine the initial inflammatory response elicited by TMPD in BALB/c mice. BALB/c WT, BALB/c Tlr9−/−, and BALB/c Tlr7−/− mice were injected i.p. with TMPD and cells collected from the peritoneal cavity were examined 4 d later. Compared to PBS injected WT BALB/c mice, all three TMPD-injected strains showed a strong increase in the total number of peritoneal exudate cells, as well as CD11b+ myeloid lineage cells. Although not statistically significant, the Tlr9−/− mice tended to have more, not fewer, total and CD11b+ cells (Fig. 2A, 2B). Untreated WT and Tlr9−/− mice had comparably low numbers of total and CD11b+ peritoneal washout cells.
The CD11b+ cells were further analyzed for the expression of Ly6C and Ly6G. Prior studies showed that CD11b+Ly6ChiLy6G− cells represent a population of inflammatory monocytes, whereas the CD11b+Ly6G+ cells represent inflammatory granulocytes (19, 24). There was a significant increase in the relative and total number of Ly6Chi inflammatory monocytes in TMPD-injected Tlr9−/− mice when compared with either the TMPD-injected WT or Tlr7−/− mice (Fig. 2A, 2B). Reeves and colleagues have previously shown that the number of inflammatory monocytes is decreased in B6 TMPD-injected Tlr7−/− mice compared with WT mice at day 14 postinjection (24), and the same is true for BALB/c Tlr7−/− mice (data not shown). To further examine the role of TLR7 in this model, we stained the day 4 peritoneal cell subsets for TLR7 (Fig. 2C). Compared to the Tlr7−/− negative control, both WT and Tlr9−/− mice inflammatory Ly6Chi monocytes expressed high levels of TLR7, with the Tlr9−/− cells tending toward slightly higher levels.
Although the total number of Ly6G+ granulocytes was unaltered between the three different groups 4 d after TMPD injection (Fig. 2A, 2B), we detected a distinct Ly6ChiCD86hi subpopulation within the Ly6Ghi granulocyte gate in the Tlr9−/− TMPD-treated mice. This population was less apparent in the cells collected from TMPD-treated WT BALB/c and Tlr7−/− mice (Fig. 2A, 2B). Although the Ly6Cint granulocytes expressed barely detectable levels of TLR7, the Ly6G+Ly6Chi cells expressed significantly more TLR7, with the Tlr9−/− cells again expressing more than the WT cells (Fig. 2C). They also expressed higher levels of CD86 than did the Ly6Cint granulocytes (Fig. 2D).
Collectively, these data indicate that TLR9 deficiency results in an early increase in the number of inflammatory monocytes accumulating in the peritoneal cavity, as well as the appearance of a distinct subpopulation of activated Ly6G+ cells, both of which express high levels of TLR7. The accumulation of inflammatory Ly6Chi monocytes in the peritoneal cavity at day 14 after TMPD has been shown to be highly dependent on type I IFN (7). Therefore, the early increase in this population in the TMPD-injected Tlr9−/− mice suggested a stronger type I response in the absence of TLR9. We next tested the day 4 expression levels of six ISGs, including Mx1, IRF7, IFI204, IP10, CCL5, and ISG15. ISG expression levels trended higher in the Tlr9−/− mice than in either the TLR-sufficient or Tlr7−/− mice (Fig. 2E), consistent with the notion of increased IFN activity.
TLR9 deficiency exacerbates TMPD-driven myelopoiesis
TLR7 transgenic mice exhibit a peripheral expansion of inflammatory monocytes and neutrophils that is dependent on increased myelopoiesis in the BM secondary to TLR7 signaling and type I IFN production (25). To determine whether TLR9 deficiency affects the production of myeloid progenitor subsets in the BM following TMPD injection, BM cells were collected 14 d after TMPD injection and the lineage− cells were evaluated for the frequency of myeloid precursor subsets. Importantly, the Tlr9−/− mice consistently had an increased frequency of myeloid precursors, with a significant increase in the percentage of the two myeloid progenitor subsets, CMPs and GMPs, in the BM of the Tlr9−/− compared with WT BALB/c mice 4 d after pristane injection (Fig. 3A, 3B). In contrast, the frequency of myeloid progenitor subsets in untreated WT and Tlr9−/− mice was indistinguishable. Taken together, these results are consistent with an increase in type I IFN production in TMPD-injected Tlr9−/− mice but not untreated Tlr9−/− mice.
TLR9-deficient BMDMs, but not BMDCs, produce more IL-6 in response to TLR ligands
To further explore the effect of TLR9 deficiency on TLR7-driven responses, we stimulated M-CSF–generated BMDMs and GM-CSF-generated BM-derived dendritic cells (DC) with increasing concentrations of the small molecule TLR7 ligand R848 or with increasing concentrations of the stimulatory RNA fragment RNA 40. The extent of activation was then quantified by measuring the amount of IL-6 produced over the next 24 h. Tlr9−/− macrophages (Mϕ) produced more IL-6 than did WT macrophages, and the response was entirely TLR7-dependent, as no IL-6 was produced by Tlr7−/− Mϕ (Fig. 4A). Responses to TLR2 and TLR4 ligands were comparable. These data suggest that TLR9 deficiency results in higher TLR7 activity. However, over the entire titration curve, Tlr9−/− DCs did not respond better to the TLR7 ligands than did WT DCs (Fig. 4B). Taken together, the data show that the capacity of TLR9 to negatively regulate TLR7 responses is highly cell type–dependent.
TLR9-deficient TMPD-injected mice produce autoantibodies specific for RNA-associated autoantigens
Similar to other SLE-prone strains, TMPD-injected mice routinely develop hypergammaglobulinema and autoantibody production. Tlr9−/− autoimmune-prone mice routinely show a particular increase in IgG2 complement-fixing Abs (15). To determine whether TLR9 deficiency affected Ab production in TMPD-injected mice, the frequency of Ab-producing cells in the spleen at 5 mo after pristane injection was determined by ELISPOT. We found that the spleens from the Tlr9−/− mice had a 3-fold greater frequency of IgG2a-producing cells than did spleens from the TLR-sufficient BALB/c mice (Fig. 5A), even though we could not detect a statistically significant difference in circulating IgG2a titers (Supplemental Fig. 1D).
TMPD-injected WT BALB/c mice routinely produce autoantibodies that show a predominantly speckled nuclear HEp-2 staining pattern, often associated with autoantibodies reactive with SmRNP (Fig. 5B). A smaller percentage of the TMPD-injected Tlr9−/− mice gave a speckled nuclear pattern, and this tendency correlated with a loss of nuclear SmRNP reactivity, as determined by an SmRNP ELISA (Fig. 5C). Instead, they shifted to a more prominent cytoplasmic staining pattern, characteristic of anti-RNA autoantibodies, and developed higher RNA autoantibody titers, as detected by an RNA ELISA (Fig. 5D). Both speckled nuclear and cytoplasmic staining patterns are thought to be TLR7-dependent (15). Neither the WT or Tlr9−/− control groups produced autoantibodies as detected by ANAs or RNA ELISA. Taken together, the data indicate that loss of TLR9 results in a shift in autoantibody specificity as well as an increase in autoantibody production.
Increased frequency of neutrophil-specific autoantibodies and glomerular granulocytic infiltrates associated with accelerated renal disease in the Tlr9−/− mice
The increased frequency of CMPs and GMPs in the BM of the TMPD-injected Tlr9−/− mice indicated an increased turnover of granulocytes and the potential for an increase in the release of granulocyte-derived cell debris that could trigger autoantibody production. ANCAs are frequently observed in vasculitis patients (26) but have also been described in animal models (27, 28) and in SLE patients with severe renal disease (29–31). Sera from TMPD-injected Tlr9−/− BALB/c and TLR-sufficient BALB/c mice were therefore assayed for ANCA reactivity by immunofluorescence staining of fixed neutrophils. Approximately half the BALB/c sera stained neutrophils, and those sera that were positive showed a predominantly nuclear staining pattern. However, a much higher frequency of the TMPD-injected Tlr9−/− sera showed strong ANCA reactivity with cytoplasmic Ags, perinuclear Ags, and/or atypical ANCA staining patterns (Fig. 5E, 5F). Because a major antigenic specificity of ANCAs is MPO (32), we further tested the sera in a MPO ELISA. In this study, again a higher frequency of the Tlr9−/− sera were positive (69%) compared with a lower level MPO reactivity of most TLR-sufficient BALB/c TMPD-treated mice (34%) (Fig. 5G).
Owing to the association between ANCA Abs and nephritis scores, we decided to compare kidneys obtained from TMPD-injected Tlr9−/− and WT mice at 5 mo after treatment for evidence of autoantibody deposition and granulocyte infiltration. Even though we could not detect a significant increase in the circulating levels of total IgG or IgG2a in the Tlr9−/− mice compared with WT mice (Supplemental Fig. 1D), we found a dramatic increase in the percentage of glomeruli with IgG, IgG2a, and IgG2b deposits as determined by immunofluorescent staining (Fig. 6A). The average mean fluorescence intensity signal per glomerulus was also increased in the TMPD-treated Tlr9−/− mice (Fig. 6A).
We further examined renal cellular composition 5 mo after TMPD treatment between the different groups. Kidney sections were stained with Ly6G, and infiltrating granulocytes were detected by immunofluorescence. Substantial numbers of granulocytes were only detected in the TMPD-injected Tlr9−/− mice (Fig. 6B, Supplemental Fig. 2). Kidneys of untreated and TMPD-treated mice were also digested with a collagenase/DNase mixture to obtain single-cell suspensions, which were stained for lymphoid (B cells, T cells) and myeloid (monocytes/Mϕ, neutrophils) cells and analyzed by flow cytometry. We did not detect a difference in either the number of lymphoid cells or the formation of ectopic lymphoid tissue within kidneys of the TMPD-treated groups (data not shown). Numbers of migrating inflammatory monocytes (Ly6Chi) and renal Mϕ were comparable in the WT and Tlr9−/− mice. However, there was a strong increase in the percentage of infiltrating neutrophils, as detected by Ly6G, only in the TMPD-treated Tlr9−/− mice and not in the TMPD-treated WT mice (Fig. 6C). Neither the untreated WT or untreated Tlr9−/− mice showed any evidence of IgG deposition or granulocyte infiltration. Taken together, these data indicate that TLR9 deficiency in TMPD-induced autoimmunity leads to severe kidney damage associated with the deposition of IgG immune complexes and granulocyte recruitment.
Inducible models of SLE serve an important role in SLE research, as they provide a system for exploring the early events in the onset of disease without the need for extensive backcrossing to strains with additional risk alleles. TMPD-induced systemic autoimmunity has often been used in this context. However, the effect of TMPD is strain-dependent, and most of the murine studies involving TMPD have used B6 mice. TMPD-treated B6 mice often develop hemorrhagic lesions in the lung and exhibit little evidence of glomerulonephritis or other forms of renal disease (33, 34). Moreover, in contrast to spontaneous models of SLE, TMPD-induced autoimmunity was found to be attenuated in Tlr9−/− B6 mice (18). TMPD-treated BALB/c mice do not develop lung lesions and do eventually develop mild nephritis. We now show that TMPD-treated Tlr9−/− BALB/c mice develop more severe renal disease than do TLR-sufficient mice, along with other clinical manifestations of disease, and exhibit a reduced lifespan. TMPD-injected BALB/c mice therefore resemble genetically programmed models of SLE (11, 12, 14–17), and they can be used to explore the basis for the negative regulatory role of TLR9 in murine SLE.
Because the time of disease induction is clearly delineated by the day of TMPD injection, this model allows us to monitor the very early events leading to systemic autoimmunity. Importantly, within 4 d of TMPD injection, flow cytometry revealed clear differences between the WT and Tlr9−/− mice with regard to the quality and number of inflammatory cells extravasating into the peritoneal cavity. The total number of inflammatory monocytes that accumulated in the peritoneal cavity was increased 2-fold in the Tlr9−/− mice. Reeves and colleagues (24) have reported that these inflammatory monocytes are an early source of type I IFN, at least at the RNA level, and we have reported that they produce many additional proinflammatory cytokines, including IL-6 (19). In the context of TMPD autoimmunity, IL-6 is clearly important because SLE-like disease is ameliorated in IL-6–deficient mice (35). We now show that inflammatory Ly6Chi monocytes express very high levels of TLR7 protein, and that Tlr9−/− BMDMs produce more IL-6 in response to TLR7 ligands than do WT BMDCs, despite that the WT BMDCs also express very high levels of TLR7. These data are consistent with the notion that TLR9 and TLR7 compete for binding to the chaperone protein Unc93B1, and in the absence of TLR9, TLR7 can respond more effectively (36, 37). However, we also found that Tlr7−/− BMDMs produced more IL-6 in response to a small molecule TLR9 ligand than do WT BMDCs, leaving the distinction between TLR7 and TLR9 uncertain. Furthermore, in contrast to BMDMs, Tlr9−/− and WT BMDCs produced comparable levels of IL-6. Therefore, the capacity of TLR9 to negatively regulate TLR7 responses is highly cell type–dependent.
Several studies have pointed to a key role for B cells in the exacerbated disease exhibited by TLR9-deficient MRL/lpr and WASp−/− mice (13, 23), pointing to distinct functional outcomes of TLR9- versus TLR7-activated B cells. Nevertheless, WT and Tlr9−/− B cells express similar levels of TLR7, and R848 induces comparable levels of proliferation in WT and Tlr9−/− B cells (21). Purified WT and Tlr9−/− B cells have also been reported to make comparable levels of IL-6 in response to a TLR7 ligand (38). Clearly, function does not correlate with protein levels, and additional studies involving a more extensive analysis of TLR-expressing primary cell types will be required to fully understand the molecular basis for the apparent negative regulatory role of TLR9 in SLE.
The TMPD-injected Tlr9−/− mice also showed indications of a stronger IFN signature, as reflected by increased expression of IFN-inducible genes, including Mx1 and IRF7. Both MyD88-deficient and IFNaR-deficient TMPD-injected mice fail to produce either Mx1 or IRF-7, indicating that these genes are induced through a TLR-dependent pathway and dependent on a type I IFN (7). TMPD-injected Tlr9−/− mice, but not uninjected Tlr9−/− mice, also had an increased frequency of CMPs and GMPs in the BM than did their WT counterparts. This apparent IFN signature is consistent with the dramatically elevated IFN-α serum titers previously reported for Tlr9−/− MRL/lpr mice (39). TMPD has been shown to induce cell death both in vitro and in vivo, and the cell debris generated by TMPD exposure could contribute to an inflammatory reaction (19, 40). B cells and DCs appear to be particularly sensitive to pristane (40). In response to this inflammatory stimulus, it is then possible that TLR7 more effectively activates IRF5, or other transcription factors that lead to the production of IFN and/or IFN-inducible genes, than does TLR9. We propose that the elevated type I titers lead to increased myelopoiesis/granulopoiesis, a more rapid turnover of neutrophil lineage cells, the increased production of anti-neutrophil Abs, and a much more extensive granulocyte glomerular infiltrate, which collectively promote glomerulonephritis, renal sclerosis, and crescent formation.
We also consistently found an increased frequency of Ly6ChiLy6G+CD86hi cells in the Tlr9−/− mice. Ly6C expression in neutrophils is associated with stringent cell migration and activation (ROS production, release of proinflammatory cytokines) by a potential interaction with the Src family kinase Fgr (41–43). Additionally, the more activated phenotype of the Ly6Chi granulocyte subpopulation might be linked to the acquisition of an APC-like phenotype and subsequent T cell activation, as has previously been described in a mouse model of chronic colitis (44). The upregulation of CD86 on these Tlr9−/− Ly6G+ cells is significant, as activated neutrophils have been shown to upregulate MHC class II, prime T cells, and activate B cells (45), and even exert B cell helper function (46). This population exhibits greater side scatter than do the inflammatory monocytes or granulocytes for reasons that remain to be determined. Further characterization of this cell subset will be a focus of future studies.
In contrast to previous studies with Tlr9−/− B6 mice (38), we did not detect any evidence of spontaneous disease in Tlr9−/− BALB/c mice, as indicated by the absence of splenomegaly, autoantibody production, increased frequency of myeloid progenitors, or histological indications of renal disease. These differences may reflect shifts in the microbiome between colonies. Alternatively, it is possible that the propensity of Tlr9−/− B6 mice to develop modest SLE-like clinical manifestations reflects the increased signal strength of the TLR9 allele expressed by B6 mice (47).
Type I IFNs have been definitively implicated in granulopoiesis, and the increase in myelopoietic progenitors in the BM in the absence of TLR9 is well in line with the recently observed emergency myelopoiesis resulting from chronic TLR7 stimulation (25). Because neutrophils have a short lifespan, the increase in neutrophil numbers must lead to an increase in neutrophil death and a potential source of neutrophil-associated autoantigens. In fact, we found increased titers of ANCA autoantibodies including anti-MPO. These ANCAs not only recognize Ags that are contained within neutrophils but also within lysosomes of monocytes (48). It is also possible that glomerular neutrophils in the TMPD-treated TLR9 knockout mice promote renal injury through the production of neutrophil extracellular traps (49, 50), respiratory burst, and degranulation (reviewed in Ref. 32). Nucleic acid–associated immune complexes have been shown to promote neutrophil extracellular trap formation, especially in neutrophils exposed to high levels of IFN (51). Neutrophil extracellular traps may also serve as a source of autoantigen in an inflammatory setting. Importantly, autoantibodies against MPO are associated with a so called pauci-immune form of vasculitis (48). Strikingly, these anti-MPO autoantibodies are highly pathogenic and can cause glomerular necrosis and crescent formation even in the absence of T and B cells (28).
Exactly why there is a strong difference in IgG deposition between WT and TLR9−/− mice in the kidneys is not clear. One possibility is that the shift in the autoantibody repertoire to RNA-associated Abs in TLR9−/− mice favors the formation of circulating immune complexes; however, we could not detect any difference in circulating immune complex levels using a standard C1q binding assay (data not shown). Alternatively, in the absence of TLR9, autoantibodies may acquire physical properties that enhance binding to FcgR+ cells present in the kidney. Alternatively, renal disease may directly reflect the capacity of ANCA or other autoantibodies to directly bind neutrophils or other damaged cells in the kidney, thereby promoting an immune complex feed-forward process of renal inflammation. Neutrophil migration to kidneys has been used to define active lupus nephritis, and the presence of neutrophil-specific proteins in the urine of SLE patients serves as a surrogate marker for disease activity (52). This study now identifies a useful model, TMPD-injected BALB/c mice, for rigorously examining the connections between TLR9 deficiency, type I IFN production, the increased production of neutrophils, anti-neutrophil Abs, and the development and progression of renal disease. It may also help decipher the pathogenesis of hydrocarbon oil–induced autoimmunity in predisposed individuals and contribute to the development of safer vaccines.
This work was supported by National Institutes of Health Grants AR066808 (to A.M.-R.), DK090558 (to R.G.B.), and HL093262 (to E.L.), funding from the Lupus Research Institute (to A.M.-R.), and by German Research Foundation Grants SFB TR 57 (to E.L.) and DFG BO 4325/1-1 (to L.B.).
The online version of this article contains supplemental material.
Abbreviations used in this article:
anti-neutrophil cytoplasmic Ab
bone marrow–derived dendritic cell
bone marrow–derived macrophage
common myeloid progenitor
granulocyte monocyte progenitor
IFN regulatory factor
MX dynamin-like GTPase 1
peritoneal exudate cell
systemic lupus erythematosus
The authors have no financial conflicts of interest.