Toll-Like Receptor 9 Deficiency Breaks Tolerance to RNA-Associated Antigens and Up-Regulates Toll-Like Receptor 7 Protein in Sle1 Mice.

OBJECTIVE: Toll-like receptors (TLRs) 7 and 9 are important innate signaling molecules with opposing roles in the development and progression of systemic lupus erythematosus (SLE). While multiple studies support the notion of a dependency on TLR-7 for disease development, genetic ablation of TLR-9 results in severe disease with glomerulonephritis (GN) by a largely unknown mechanism. This study was undertaken to examine the suppressive role of TLR-9 in the development of severe lupus in a mouse model.
METHODS: We crossed Sle1 lupus-prone mice with TLR-9-deficient mice to generate Sle1TLR-9-/- mice. Mice ages 4.5-6.5 months were evaluated for severe autoimmunity by assessing splenomegaly, GN, immune cell populations, autoantibody and total Ig profiles, kidney dendritic cell (DC) function, and TLR-7 protein expression. Mice ages 8-10 weeks were used for functional B cell studies, Ig profiling, and determination of TLR-7 expression.
RESULTS: Sle1TLR-9-/- mice developed severe disease similar to TLR-9-deficient MRL and Nba2 models. Sle1TLR-9-/- mouse B cells produced more class-switched antibodies, and the autoantibody repertoire was skewed toward RNA-containing antigens. GN in these mice was associated with DC infiltration, and purified Sle1TLR-9-/- mouse renal DCs were more efficient at TLR-7-dependent antigen presentation and expressed higher levels of TLR-7 protein. Importantly, this increase in TLR-7 expression occurred prior to disease development, indicating a role in the initiation stages of tissue destruction.
CONCLUSION: The increase in TLR-7-reactive immune complexes, and the concomitant enhanced expression of their receptor, promotes inflammation and disease in Sle1TLR9-/- mice.

The pattern‐recognition receptors Toll‐like receptor 7 (TLR‐7) and TLR‐9 are intracellular sensors of single‐stranded RNA and double‐stranded DNA (dsDNA), respectively 1. Upon binding microbial nucleic acids, signaling cascades are activated to rapidly induce an inflammatory immune response designed to clear the pathogen 1. Several mechanisms prevent the activation of TLR‐7 and TLR‐9 by self nucleic acids, including endosomal localization, tight regulation of their trafficking, and proteolytic processing prior to receptor activation 2, 3. In systemic lupus erythematosus (SLE), however, a unique combination of self nucleic acid overload and the presence of antinuclear autoantibodies (ANAs) leads to aberrant TLR‐7/9 activation 4, 5. The immune complexes (ICs) formed between ANAs and their respective antigens (RNA/DNA and RNA/DNA‐associated proteins) deposit in tissues where they cause chronic inflammation and ultimately lead to irreversible organ damage 6. Due to the similarities between TLR‐7 and TLR‐9 expression and shared signaling pathways, it has been predicted that the deletion of either of these receptors would be beneficial in terms of disease progression. Unexpectedly, TLR‐9 deletion caused disease exacerbation in multiple models of spontaneous and induced lupus, while TLR‐7 proved to be essential for disease development 7, 8, 9, 10, 11, 12, 13, 14. These data suggest that the system is more complex than originally thought; however, the means by which TLR‐9 prevents progression to severe disease are largely unknown.

Supporting the notion of their divergent in vivo roles, analysis of TLR‐7/9 messenger RNA (mRNA) expression reveals a separation in regulatory mechanisms. TLR‐7 is rapidly induced in human immune cells following stimulation with bacteria, viruses, lipopolysaccharide (LPS), CpG, interferon‐α (IFNα), and SLE patient serum 15, 16, 17, 18, 19. However, TLR‐9 can be induced in B cells, but is either down‐regulated or unchanged in other cell types 16, 17, 18, 19, 20. We have previously shown this in human dendritic cell (DC) subsets, where plasmacytoid DCs (pDCs), BDCA‐1+, and BDCA‐3+ DCs all up‐regulated TLR‐7 mRNA in response to IFNα or influenza, but TLR‐9 mRNA was down‐regulated in pDCs 15. These differences suggest a separation of TLR‐7/9 receptor roles, which might have important implications in SLE. Abundant immune complex–bound TLR‐7/9 ligands and inflammatory mediators such as IFNα in SLE serum can chronically interfere with the expression of TLR‐7 and TLR‐9, disrupting homeostatic mechanisms to retain tolerance. Studies involving TLR‐9– and/or TLR‐7–deficient autoimmune mice can significantly contribute to the understating of these mechanisms, since TLRs are highly conserved in vertebrates 21 and humans and mice with lupus share the same autoantibody repertoire 22.

In this study we set out to examine the cellular mechanisms by which TLR‐9 deficiency results in severe lupus nephritis. We bred B6.NZM2410‐derived Sle1 congenic mice, which develop mild autoimmune traits, with TLR‐9–deficient mice to generate Sle1TLR‐9−/− mice 23, 24. By ages 4–6.5 months, female Sle1TLR9−/− mice developed severe autoimmunity, characterized by splenomegaly and kidney disease, similar to findings in TLR‐9–deficient MRL and Nba2 mice 8, 10. Glomerulonephritis (GN) was associated with DC infiltration, and upon extraction, Sle1TLR‐9−/− mouse renal DCs were more efficient at TLR‐7–dependent antigen presentation than Sle1 mouse controls. A comprehensive analysis of intracellular TLR‐7 protein expression revealed an increase in TLR‐7 expression in renal DCs and macrophages, which positively correlated with their recruitment into the kidney. Importantly, this increase in TLR‐7 occurred prior to disease development, indicating a role in the initiation stages of tissue destruction. Additionally, our data show that, in the absence of TLR‐9, Sle1 mouse B cells are primed to produce more class‐switched antibodies, and the autoantibody repertoire is skewed toward RNA‐containing antigens. In summary, this study provides a unique understanding of the protective role TLR‐9 plays in the development of autoimmunity and identifies the TLR‐7 pathway as a critical instigator of disease development.

Materials and Methods

Mice. Mice were bred at the Biomedical Resource Center (Singapore) or the University of Texas Southwestern Medical Center. The derivations of the B6.Sle1 (Sle1), TLR‐9−/−, and TLR‐7−/− mouse strains have been described previously 23, 24, 25. TLR‐9−/− and TLR‐7−/− mice (backcrossed to B6 more than 10 generations) and OT‐II–transgenic mice were bred with Sle1 mice (defined by the microsatellite markers D1Mit17, D1Mit113, and D1Mit202). SLE disease traits were evaluated in 4.5–6.5‐month‐old female mice, and functional cellular assessments were conducted using 8–10‐week‐old female mice. The care and use of laboratory animals conformed to the National Institutes of Health guidelines, and all experimental procedures were conducted according to an Institutional Animal Care and Use Committee–approved animal protocol.

Pathologic assessment of mouse kidneys. Proteinuria was assessed using Albustix (Bayer). Blood urea nitrogen (BUN) was assessed using a QuantiChrom Urea Assay Kit (BioAssay Systems). For evaluation of GN, mouse kidneys were fixed in formalin and embedded in paraffin, and 3‐μm sections were stained with hematoxylin and eosin and with periodic acid–Schiff. Microscopic morphologic analysis was performed by an independent pathologist (TPT) according to the International Society of Nephrology/Renal Pathology Society 2003 criteria for the classification of lupus nephritis 26.

Autoantibody enzyme‐linked immunosorbent assays (ELISAs). Serum autoantibodies were measured using ELISAs to detect antinucleosomes (histones and dsDNA), anti‐dsDNA, anti–U1 small nuclear RNP (anti–U1 snRNP), or anti‐RNA as previously described 27, 28. Bound IgG was detected with alkaline phosphatase–conjugated anti‐mouse IgG (Jackson ImmunoResearch) using paranitrophenyl phosphate as a substrate (Sigma). Absorbance was measured at 405/410 nm. Results are shown as arbitrary units (AU) that were calculated as absorbance at 405 nm (sample minus blank). For anti‐RNA, serial dilutions of pooled serum from diseased mice were used to construct a standard curve.

ANA Luminex assay. An AtheNA Multi‐Lyte ANA III Test System (Zeus Scientific) was used to measure 10 analytes (autoantibodies to SSA 52, SSA 60, SSB, Sm, RNP, Scl‐70, Jo‐1, centromere B, ribosomal P, and dsDNA) according to the recommendations of the manufacturer, with a goat polyclonal secondary antibody to mouse IgG heavy and light chains (Dylight 550; Abcam). Samples were run on a Luminex 200 system using Luminex 100 IS software and analyzed using AtheNA Multi‐Lyte Test System data analysis software (Zeus Scientific). Unit values reported are IU/ml for dsDNA and AU/ml for the remaining analytes.

Ig isotyping assays. Ig subtypes (IgA, IgG1, IgG2a/c, IgG2b, IgG3, and IgM) were measured using a mouse Ig isotyping bead panel (EMD Millipore), according to the recommendations of the manufacturer. This panel is designed to detect IgG2a (from BALB/c mice), which cross‐reacts with IgG2c from mice on the B6 background, which we have labeled as IgG2a/c 29. Luminex plates were read on a Flexmap 3D System (Luminex) with Bio‐Plex Manager version 6.0 software (Bio‐Rad). IgM concentrations from cell culture supernatants were analyzed with an IgM ELISA (eBioscience) according to the recommendations of the manufacturer.

Microscopy. ANA screening was performed with NOVA Lite HEp‐2 slides and the Crithidia luciliae indirect immunofluorescence test (CLIFT) using NOVA Lite dsDNA Crithidia luciliae substrate slides (both from Inova Diagnostics) according to the recommendations of the manufacturer. Sera were diluted 200‐fold for HEp‐2 and 40‐fold for CLIFT, and a goat anti‐mouse IgG DyLight 488 secondary antibody (Abcam) was used for detection. CLIFT slides were counterstained with DAPI. All images were obtained using a Zeiss LSM 800 upright confocal microscope with Zeiss Zen (Blue edition) software at 100× and 200× magnification for HEp‐2 and CLIFT, respectively. HEp‐2 staining patterns were evaluated by 2 independent investigators according to the International Consensus on Antinuclear Antibody Patterns. Confocal images of splenic germinal centers were obtained with an Olympus FV1000 confocal laser scanning microscope and were processed with FluoView (Olympus).

Flow cytometric analysis and cell sorting. Single‐cell suspensions of the mouse spleen and kidneys were obtained as previously described with additional collagenase digestion for splenic DC analysis 30. Cells were blocked for 15 minutes in staining buffer (phosphate buffered saline with 1% fetal calf serum and 15 mM HEPES) containing 20% 2.4G2 hybridoma supernatant and incubated on ice for 30 minutes with the antibodies listed in Supplementary Table 1 (available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.40535/abstract). When a biotin‐conjugated antibody was used, cells were washed and incubated for an additional 30 minutes with a streptavidin‐conjugated fluorophore. Red blood cell lysis was achieved using BD FACS lysing solution, followed by washing and resuspension in 1% paraformaldehyde. For sorting of live kidney cells, ACK lysis buffer (Lonza) was used, and cells were resuspended in staining buffer. For intracellular staining with anti–TLR‐7, the BD Cytofix/Cytoperm fixation/permeabilization solution kit was used according to the recommendations of the manufacturer. Acquisition and analysis were completed using a BD Canto II, BD Fortessa, BD LSR II, or BD Symphony system, with FlowJo 10 for Windows (Tree Star). Mouse kidney cells were sorted using BD Aria II, Aria IV, and Aria V.

In vitro B cell stimulation. Mouse splenocytes stained with 5,6‐carboxyfluorescein succinimidyl ester (CFSE) were resuspended in complete RPMI media, consisting of RPMI 1640 (Gibco Life Technologies) supplemented with 10% fetal bovine serum (HyClone; ThermoFisher Scientific), 15 mM HEPES, 100 μg/ml streptomycin, 100 units/ml penicillin, 200 μM l‐glutamine,10 μM nonessential amino acids, 100 μM sodium pyruvate (all from Gibco), and 45 μM 2‐mercaptoethanol (Sigma‐Aldrich). They were plated at 1.5 × 106 cells/ml in round‐bottomed 96‐well plates in the presence of either CpG‐B (ODN 1826), R848, or LPS‐EB Ultrapure (all from InvivoGen) at the indicated concentrations. Activation was measured by CD69 and CD86 up‐regulation at 24 hours, and proliferation at 72 hours was measured by CFSE dilution using flow cytometry. Ig subtype analysis in supernatants was measured 96 hours after stimulation.

Renal DC antigen presentation assay. Mouse kidneys cells were sorted, and cells were seeded in a 96‐well plate at 1 × 104 cells per well. Where indicated, cells sorted from multiple mice were pooled to achieve sufficient cell numbers. Cells were stimulated overnight with or without 1 μg/ml R848 (InvivoGen), and then incubated with 10 μg/ml ovalbumin (Sigma‐Aldrich). After 4 hours, medium was replaced with CFSE‐stained Sle1OT‐II mouse splenocytes at 2 × 105 cells per well. After 5 days of incubation, T cell proliferation was assessed by CFSE dilution.

Statistical analysis. Data were analyzed using GraphPad Prism 7.01 for Windows. Normal distribution was assessed using the Kolmogorov‐Smirnov test. Gaussian data were analyzed using Student's t‐test for 2 comparisons and ordinary one‐way analysis of variance with post hoc Bonferroni adjustment for multiple comparisons for 3 or more comparisons. Nonparametric data were assessed using the Mann‐Whitney test for 2 comparisons and the Kruskal‐Wallis test with post hoc Dunn's multiple comparisons test for 3 or more comparisons. Where indicated, a multiple t‐test corrected with the Holm‐Sidak method was used.

Results

Severe autoimmune disease in mice with genetic ablation of TLR‐9. Cohorts of 4.5–6.5‐month‐old female Sle1TLR‐9−/− mice and controls were analyzed for pathologic traits associated with common models of SLE. Genetic ablation of TLR‐9 in Sle1 mice resulted in splenomegaly and the expansion of all major leukocyte subsets (Figures 1A and B). There was a significant increase in CD11b+ cell numbers in Sle1 mice versus wild‐type C57BL/6J (B6) mice and in TLR‐9−/− mice versus wild‐type mice, indicating that both genetic components contribute to splenic myeloid cell expansion (Figure 1B). We detected an expansion of germinal center B cells in Sle1TLR‐9−/− mice compared to Sle1 controls, similar to observations in Sle1bTLR‐9−/− and MRL/+TLR‐9−/− mice (Figures 1C and D and Supplementary Table 2, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.40535/abstract) 9, 31. We also observed increases in CD138+B220− plasma cell numbers (Supplementary Table 2) and a decreased frequency of marginal zone B cells (Supplementary Figure 1A, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.40535/abstract), which are characteristics of lupus models overexpressing TLR‐7 32, 33. CD4+ and CD8+ T cells from Sle1TLR‐9−/− mice had a more activated phenotype, with higher percentages of CD62LlowCD44high effector memory cells and CD4+ follicular helper T cells, and increased programmed death 1 and inducible costimulator expression, consistent with observations in Sle1bTLR‐9−/− mice and in Sle1 mouse models overexpressing TLR‐7, such as Sle1Tg7 15, 31, 32 (Figure 1E, Supplementary Figures 1B–D, and Supplementary Table 2).

Figure 1

Severe disease in Toll‐like receptor 9 (TLR‐9)–deficient mice. Female B6, TLR‐9−/−, Sle1, and Sle1TLR‐9−/− mice ages 4.5–6.5 months were analyzed for lupus disease features. A, Left, Representative spleens from the indicated mouse strains. Right, Cumulative splenic weights for the indicated mouse strains. Symbols represent individual mice; bars show the mean (n = 30–65 mice/group). B, Numbers of leukocytes of each subset in the spleens of mice of the indicated strains. My = myeloid cells. C, Representative images of germinal center (GC) staining in the spleens of Sle1 and Sle1TLR‐9−/− mice. PNA = peanut agglutinin. Bars = 200 μm; original magnification × 100. D, GC frequency measured by flow cytometric analysis of GL‐7+Fas+ cells and presented as the percentage of B cells (CD19+B220+). E, Flow cytometric analysis of naive (CD62LhighCD44low) and effector memory (CD62LlowCD44high) CD4+ T cells in the spleens of mice of the indicated strains. F, Left, Representative photomicrographs of an Sle1 mouse glomerulus with mild segmental mesangial proliferation and an Sle1TLR‐9−/− mouse glomerulus with global endocapillary proliferation. Periodic acid–Schiff stained. Bars = 20 μm; original magnification × 600. Right, Kidney glomerulonephritis (GN) class (I–IV represented as 1–4) in Sle1 and Sle1TLR‐9−/− mice. G and H, Functional assessment of mouse kidneys. Blood urea nitrogen (BUN) (G) and urinary protein levels (measured by Albustix) (H) were determined. I, Infiltration of CD45+ leukocytes into the mouse kidney, assessed by flow cytometry. All flow cytometry data (in B, D, E, and I) are from 2–5 separate cohorts with a total of 6–17 mice per group. In B, D, E, and I, bars show the mean ± SEM. In F–H, circles represent individual mice; horizontal lines and error bars show the mean ± SEM. * = P < 0.05; ** = P < 0.01; *** = P < 0.001; **** = P < 0.0001, by one‐way analysis of variance (with Bonferroni adjustment for multiple comparisons) for parametric data or Kruskal‐Wallis test (with Dunn's multiple comparison test) for nonparametric data, taking into consideration all 4 groups; # = P < 0.05; ## = P < 0.01, by Student’s t‐test for parametric data or Mann‐Whitney test for nonparametric data, taking into consideration only 2 groups.

Analysis of the splenic CD11b+ myeloid lineage revealed increases in the numbers of Gr1high polymorphonuclear leukocytes (PMNs) and SSC‐Ahigh eosinophils in Sle1TLR‐9−/− mice compared to controls (Supplementary Table 2 and Supplementary Figure 1E). Gr1−/low cells represented the majority of the expanding myeloid population (Supplementary Figure 1E) and were further analyzed for CD11c and major histocompatibility complex (MHC) class II expression (Supplementary Figure 1F). Conventional CD11b+ DCs (cDCs) expressing MHC class II, and a possible precursor, which lacks MHC class II, were increased; both were previously characterized in the Sle1Tg7 model (Supplementary Figure 1F and Supplementary Table 2) 15. Splenic F4/80+CD64+CD11bintermediate macrophage, CD8+ DC, and pDC numbers, but not frequencies, were increased due to increased cellularity in TLR‐9–deficient Sle1 mice (Supplementary Table 2).

Pathologic analysis confirmed that the majority of Sle1TLR‐9−/− mice developed severe GN (class III–IV), characterized by segmental to global endocapillary proliferation of the glomeruli (Figure 1F). Increased serum BUN levels and urinary protein levels confirmed these findings (Figures 1G and H). GN was associated with an infiltration of CD45+ leukocytes into the mouse kidney, consisting mainly of T cells and CD11b+ myeloid cells (Figure 1I and Supplementary Table 2). CD11b+ cells were mostly Gr1−/low, as determined previously in Sle1Tg7 mice (Supplementary Figure 1G) 15. B cells and pDCs comprised minor populations in the kidney infiltrates and were not significantly increased in the absence of TLR‐9 (Supplementary Table 2).

TLR‐9 deficiency in mice skews the autoantibody profile toward RNA‐associated autoantigens. Previous studies have shown that TLR‐9 deletion on the MRL mouse background results in a shift from homogenous nuclear to cytoplasmic ANA HEp‐2 staining patterns and the loss of binding to mitotic chromatin 7, 8, 9. This coincides with lower levels of antinucleosome antibodies, but not with lower anti‐dsDNA antibody titers as measured by ELISA 8. Similarly, elimination of TLR‐9 decreased antinucleosome/chromatin antibodies in B6.Fas/lpr, Plcg2Ali5, and B6Nba2 mice, while anti‐dsDNA levels measured by ELISA were increased or unchanged 10, 13, 34.

Next, we examined the autoantibody profile in Sle1TLR‐9−/− mice, using a variety of methods for clarity. Microscopy of HEp‐2 cells revealed that serum autoantibodies from Sle1TLR‐9−/− mice bound primarily to the cytoplasm, with some nucleolar specificity, in contrast to Sle1 mouse sera, which bound primarily to the nucleus (Figures 2A–C). All Sle1TLR‐9−/− mouse sera tested lost the ability to bind mitotic chromatin (Figures 2A and D). Consistent with this finding, the levels of antinucleosome autoantibodies were significantly decreased in Sle1TLR‐9−/− mice compared to Sle1 controls (Figure 2E). However, we detected increased levels of anti‐dsDNA antibodies using an in‐house ELISA and a commercial Luminex‐based assay (Figure 2E and Supplementary Figure 2F, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.40535/abstract). In clinical practice, dsDNA ELISAs have shown low specificity and poor correlation with superior detection assays, such as CLIFT 35, 36. We therefore tested Sle1TLR‐9−/− samples from Figures 2A–D with the CLIFT assay, and 6 of 8 serum samples showed positive kinetoplast staining, confirming the presence of anti‐dsDNA antibodies despite negative chromatin staining (Supplementary Figures 2A and B). High levels of anti‐dsDNA as determined by anti‐dsDNA ELISA did not correlate with positive CLIFT findings (Supplementary Figure 2C), confirming low specificity of the ELISA. Discrepancies between ELISA and CLIFT for anti‐dsDNA measurement in TLR‐9−/− lupus models have also been observed by other groups and are due to a number of factors, including the origin and purity of the DNA antigen used for ELISA 8, 13.

Figure 2

Toll‐like receptor 9 (TLR‐9) deletion in mice breaks tolerance to RNA‐associated antibodies. A, Representative microscopy images of nuclear (n) and cytosolic (c) HEp‐2 staining patterns of serum autoantibodies from B6, Sle1, and Sle1TLR‐9−/− mice ages 4.5–6.9 months. White arrows show mitotic chromosome staining; orange arrows indicate the absence of mitotic chromosome staining. B and C, Analysis of nuclear (B) and cytoplasmic (C) staining patterns determined by HEp‐2 staining and microscopy. D, Percentage of Sle1 and Sle1TLR‐9−/− mice with mitotic chromatin positivity. Serum for HEp‐2 staining was obtained from 3 independent cohorts with a total of 6 Sle1 mice and 8 Sle1TLR‐9−/− mice. E, Levels of antinucleosome (double‐stranded DNA [dsDNA]/histone), anti‐dsDNA, anti–small nuclear RNP (anti‐snRNP), and anti‐RNA autoantibodies in Sle1 and Sle1TLR‐9−/− mice. Serum dilutions were 1:100 for anti‐RNA and 1:200 for all other autoantibodies. AU represents absorbance at 405 nm (sample minus blank). Values for anti‐RNA antibodies are represented as logarithmic values of dilutions calculated from a standard curve. F, Levels of IgG subtypes in sera from Sle1 and Sle1TLR‐9−/− mice analyzed by Luminex. In E and F, mouse sera were from 6 independent cohorts. Circles represent individual mice; horizontal lines and error bars show the mean ± SEM. All data presented are from 4.5–6.9‐month‐old mice. Parametric data were assessed by one‐way analysis of variance (with Bonferroni adjustment for multiple comparisons) or Student's t‐test, and nonparametric data were assessed by Kruskal‐Wallis test (with Dunn's multiple comparison test) or Mann‐Whitney test. Significance in D was determined by Fisher's exact test. * = P < 0.05; ** = P < 0.01; *** = P < 0.001.

The cytoplasmic HEp‐2 staining pattern of Sle1TLR‐9−/− mouse serum is characteristic of RNA‐reactive antibodies 33, 37 (Figures 2A and C), which we confirmed by anti‐RNA ELISA (Figure 2E). Moreover, cytoplasmic staining intensity positively correlated with anti‐RNA titers (Supplementary Figures 2D and E). Additionally, anti‐snRNP autoantibody levels were higher in Sle1TLR‐9−/− mice than Sle1 mice (Figure2E). A Luminex autoantigen assay confirmed this change and detected significant increases in several other RNA‐associated autoantibodies, including anti‐SSA 52 and 60, anti‐SSB, anti‐Sm, and anti–ribosomal P (Supplementary Figure 2F). Analyses of serum Ig levels by Luminex showed that TLR‐9 deficiency resulted in significantly higher concentrations of total IgG2a/c, IgG2b, and IgM antibodies compared to Sle1 controls, while IgA concentrations were decreased (Figure 2F).

Systemic up‐regulation of TLR‐7 in diseased TLR‐9 mice. Increased TLR‐7 expression leads to an augmentation in autoimmunity 15, 27, 33. However, it is unknown if TLR‐9−/− mice express higher levels of TLR‐7 protein. Bossaller et al recently showed an infiltration of Ly6GhighLy6Chigh cells with higher levels of TLR‐7 in the peritoneal cavity of TLR‐9−/−BALB/c mice with pristane‐induced lupus 14. Therefore, we assessed TLR‐7 expression by flow cytometry using the same antibody clone (A9410). We verified binding specificity using TLR‐7–deficient B6 mice (TLR‐7−/−) and a fluorescence minus one control. Our analyses showed that TLR‐7 levels were significantly higher in B cells, pDCs, CD11b+ DCs, F4/80+ macrophages, and in the CD11c+MHCII− precursor population in Sle1TLR‐9−/− mice than in Sle1 controls (Figure 3A). TLR‐7 expression positively correlated with the expansion of CD11b+ DCs and CD11c+ MHCII− subsets, but not with pDC frequencies (Figures 3B–D). TLR‐7 was expressed in splenic PMNs and at significantly higher levels in TLR‐9–deficient Sle1 mice (Supplementary Figure 3A, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.40535/abstract). No differences were observed in splenic CD11b+Gr‐1intermediate cells, which had highly variable TLR‐7 expression (Supplementary Figure 3B). As expected, TLR‐7 expression was not detected in CD3+ T cells or CD8+ DCs (Supplementary Figures 3C and D).

Figure 3

Regulation of Toll‐like receptor 7 (TLR‐7) protein expression in the mouse spleen by TLR‐9 deficiency. A, Representative histograms (top) and median fluorescence intensity (MFI) data (bottom) for TLR‐7 expression, measured by intracellular flow cytometry, in the indicated spleen cell subsets in Sle1 and Sle1TLR‐9−/− mice ages 5.7–6.4 months. Results are from 3 independent experiments each conducted with a TLR‐7−/− and fluorescence minus one (FMO) control (n = 8 Sle1 mice, 8 Sle1TLR‐9−/− mice, 3 TLR‐7−/− controls, and 3 FMO controls). Symbols represent individual samples; horizontal lines and error bars show the mean ± SEM. Mϕ = macrophage. B–D, Correlation of leukocyte expansion in Sle1 and Sle1TLR‐9−/− mouse spleens with TLR‐7 expression in plasmacytoid dendritic cells (pDCs) (B), CD11b+ DCs (C), and CD11c+ major histocompatibility complex class II (MHCII)− cells (D). Parametric data were assessed by Student's t‐test, and nonparametric data were assessed by Mann‐Whitney test. Correlations were determined using Spearman's rank correlation for nonparametric data and Pearson's correlation for parametric data. ** = P < 0.01; *** = P < 0.001; **** = P < 0.0001.

Regulation of IgG production by TLR‐9 in lupus‐prone mice. Given the essential role of humoral immunity in the development of SLE, we assessed B cell functional responses to TLR‐7 in young prediseased mice. We stimulated mouse splenocytes with the TLR‐7 ligand R848 and TLR‐4 ligand LPS as a TLR‐7–independent signal. We used flow cytometry assays to measure B cell survival, activation, and proliferation and did not detect any significant differences between Sle1 and Sle1TLR‐9−/− mice within a wide range of R848 concentrations, in contrast to the findings of an earlier study in B6.TLR‐9−/− CD19+B220+ mouse splenocytes 38 (Figures 4A and B and Supplementary Figures 4A–C, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.40535/abstract). Consistent with the lack of increased activation upon TLR‐7 ligation, we did not detect an increase in TLR‐7 protein in B cells in young mice (Figure 4C). However, analyses of supernatants collected on day 4 revealed that Sle1TLR‐9−/− mouse cultures produced significantly higher amounts of IgG upon stimulation with R848 or LPS (Figure 4D). Moreover, in the absence of any stimulation, Sle1TLR‐9−/− B cells spontaneously released IgG2b and IgG3, in contrast to their Sle1 counterparts (Figure 4D).

Figure 4

Regulation of antibody production by Toll‐like receptor 9 (TLR‐9) in Sle1 mice. A and B, B cell (B220+) activation (A) and proliferation (B) in splenocytes from young (8–10‐week‐old) Sle1 and Sle1TLR‐9−/− mice that were left untreated (media), stimulated with R848 (0.01 μg/ml), or stimulated with lipopolysaccharide (LPS; 1 μg/ml). B cell activation was measured by flow cytometry after 24 hours of stimulation and is shown as the percentage of CD69+ cells. B cell proliferation was measured according to the 5,6‐carboxyfluorescein succinimidyl ester dilution after 72 hours of stimulation. C, TLR‐7 expression, measured by intracellular flow cytometry, in mouse splenic B220+ B cells. TLR‐7–deficient Sle1 mice (Sle1TLR‐7−/−) (n = 4) and fluorescence minus one (FMO) samples were used as negative controls. MFI = median fluorescence intensity. D, Levels of IgG subtypes, measured by Luminex, in Sle1 and Sle1TLR‐9−/− mouse culture supernatants collected after 96 hours of incubation. Cultures were left untreated, stimulated with R848, or stimulated with LPS. E, Expression of surface IgG (sIgG) (IgG1/IgG2a/IgG2b/IgG3) on freshly isolated B220+CD19+ splenocytes from Sle1 and Sle1TLR‐9−/− mice. F and G, TLR‐7 expression and frequencies of splenic CD138+ plasma/plasmablasts (F) and CD11b+ dendritic cells (DCs) (G) in Sle1 and Sle1TLR‐9−/− mice. In A, B, and D, bars show the mean ± SEM from 2–3 independent experiments (n = 9–15 mice per group). Data were assessed by multiple t‐tests, and statistical significance was corrected using the Holm‐Sidak method. In C, E, F, and G, data are from 1 representative experiment with 8–10‐week‐old mice (n = 8 Sle1 and 5 Sle1TLR‐9−/− mice). Circles represent individual mice; horizontal lines and error bars show the mean ± SEM. Significance was determined by Student's t‐test. * = P < 0.05; ** = P < 0.01; *** = P < 0.001; **** = P < 0.0001.

No differences were detected between Sle1TLR‐9−/− mice and Sle1 mice in IgM and IgGA production, suggesting an increase in IgG isotype switching (Supplementary Figure 4D). We thus stained freshly isolated mouse splenocytes with a cocktail of anti‐IgG1/IgG2a/IgG2b/IgG3 antibodies and analyzed them by flow cytometry (gating strategy is shown in Supplementary Figure 4E). We confirmed that Sle1TLR‐9−/− mouse B cells expressed significantly more surface IgG than their Sle1 mouse counterparts (Figure 4E). Additionally, the frequencies of CD138+ plasma/plasmablast cells were increased and they expressed significantly higher TLR‐7 protein levels in the absence of TLR‐9 (Figure 4F and Supplementary Figure 4F). Extrafollicular plasmablast responses and antibody switching in lupus have been attributed to DCs 39. We detected increased frequencies of splenic CD11b+ DCs with higher TLR‐7 expression in Sle1TLR‐9−/− mice (Figure 4G). This early stage expansion of TLR‐7–high DCs might play a role in the increase in CD138+ plasma/plasmablasts and IgG‐switched B cells in Sle1TLR‐9−/− mice.

Kidney disease in TLR‐9 mice is characterized by infiltrating renal cDCs that overexpress TLR‐7. To further understand kidney pathogenesis in Sle1TLR‐9−/− mice, we characterized renal leukocyte infiltrates, using a similar flow cytometry strategy as previously described for Sle1 mice overexpressing TLR‐7 (Sle1Tg7) 15. The majority of infiltrating cells in Sle1TLR‐9 mouse kidneys were CD11b+Gr‐1−/low (Supplementary Figure 1G). These consisted of 3 CD11c+ populations: MHCII−, F4/80+ macrophages, and F4/80−/low DCs (Figures 5A and B). As in Sle1Tg7 mice, the proportions of MHCII− and F4/80−/low DCs were increased in the kidneys of Sle1TLR‐9−/− mice compared to Sle1 mice, while the proportions of macrophages did not change significantly 15 (Figure 5B).

Figure 5

Infiltration of dendritic cells (DCs) with increased Toll‐like receptor 7 (TLR‐7) protein expression into Sle1TLR‐9−/− mouse kidneys. A, Analysis of renal CD45+CD11b+Gr‐1−/low subsets using antibodies to CD11c, major histocompatibility complex class II (MHCII), F4/80, and CD11b. The magenta boxed area includes the CD11c+MHCII− precursor population. The black boxed area includes the CD11c+MHCII+ cells. CD11c+MHCII+ cells were further subdivided into F4/80+ macrophages (Mϕ; green encircled area) and F4/80−/low DCs (blue encircled area). B, Cumulative frequencies of renal CD11b+Gr‐1− subsets identified in A in Sle1 and Sle1TLR‐9−/− mice. Bars show the mean ± SEM. Data are from 3 independent cohorts of 4.5–6.5‐month‐old mice with a total of 9–16 mice per group. C, Proliferation of CD4+TCRvα2+ Sle1OT‐II cells exposed to ovalbumin‐pulsed F4/80+ macrophages or F4/80−/low DCs, with or without prior stimulation with R848. Circles represent cells sorted from an individual mouse (n = 6), except for Sle1 F4/80−/low DCs that were pooled to a total of 3 samples due to low cell numbers. D, Analysis of TLR‐7 protein expression, using flow cytometry, in the CD11b+Gr‐1− subsets identified in A in Sle1 and Sle1TLR‐9−/− mice. Symbols represent individual samples; horizontal lines and error bars show the mean ± SEM. E, Correlation of TLR‐7 expression in Sle1 and Sle1TLR‐9−/− mice with percentages of renal macrophage and DC infiltration. Data in D and E are from 3 independent experiments with a total of 8 mice (ages 5.7–6.4 months) per group. Each experiment was conducted with a TLR‐7−/− and a fluorescence minus one (FMO) control (a total of 3 each). Parametric data were assessed by one‐way analysis of variance (with Bonferroni adjustment for multiple comparisons) or Student's t‐test, and nonparametric data were assessed by Kruskal‐Wallis test (with Dunn's multiple comparison test) or Mann‐Whitney test. Correlations were determined using Spearman's rank correlation for nonparametric data and Pearson's correlation for parametric data. * = P < 0.05; ** = P < 0.01. NS = not significant; MFI = median fluorescence intensity.

We then assessed the ability of DCs to present antigen and induce T cell proliferation by purifying renal DCs, or macrophages as a control, and exposing them to ovalbumin before culturing with Sle1OT‐II T cells 15. Unstimulated DCs from Sle1TLR‐9−/− mouse kidneys induced more T cell proliferation than those from Sle1 controls, which was enhanced with the TLR‐7 ligand R848 (Figure 5C and Supplementary Figure 5A, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.40535/abstract). Renal macrophages did not show augmented T cell proliferation with R848, consistent with earlier data from bone‐marrow derived and peritoneal macrophages 40. Both renal DCs and macrophages had increased TLR‐7 protein expression in the absence of TLR‐9, and the levels positively correlated with the percentage of infiltrating renal DCs and macrophages (Figures 5D and E). Additionally, TLR‐7 expression was increased in TLR‐9–deficient Sle1 kidney Gr1intermediate CD11b+ cells, but not in pDCs and B cells (Supplementary Figure 5B). We did not detect TLR‐7 expression in kidney PMNs and T cells (data not shown). To assess whether the increases in TLR‐7 occurred prior to the development of GN, we analyzed younger Sle1TLR‐9−/− mice with less severe disease, as confirmed by splenic weight and a lack of significant CD45+ kidney infiltration (Supplementary Figure 5C). Expression of TLR‐7 was increased in Sle1TLR‐9−/− mouse kidney F4/80+ macrophages and F4/80−/low DCs, but not in MHCII− cells, similar to their older counterparts (Supplementary Figure 5D).

Discussion

Over the last decade, multiple studies have supported the notion of a fundamental role of TLR‐7 in SLE disease development; however, little progress has been made to elucidate the regulatory role of TLR‐9 7, 8, 9, 10, 12, 13, 14, 27, 31, 32, 33, 34, 41. Moreover, the lack of data from human studies has reduced enthusiasm for understanding the cellular and molecular roles in TLR‐9–deficient mice. In the present study, we have shown for the first time that TLR‐7 is increased at the protein level prior to disease onset in TLR‐9–deficient Sle1 mice. Our data also show that in B cells, TLR‐9 maintains tolerance to RNA and RNA‐associated antigens, prevents antibody switching and antibody production, and regulates B cell maturation. Thus, in the absence of TLR‐9, there is an accumulation of TLR‐7–reactive RNA/anti‐RNA–associated ICs that can activate the increased TLR‐7 in DCs and B cells (Figure 6). This essentially yields an identical kidney inflammation phenotype to that observed in Sle1Tg7 mice, which have a modest increase in TLR‐7 expression 15.

Figure 6

Schematic overview of the proposed mechanisms leading to severe disease in the absence of Toll‐like receptor 9 (TLR‐9). 1 B cells from prediseased Sle1TLR‐9−/− mice are primed to produce more IgG‐switched antibodies, most likely due to the presence of increased numbers of CD11b+ dendritic cells (DCs) with increased TLR‐7 protein expression. 2 In the absence of TLR‐9, the autoantibody repertoire shifts toward RNA/RNA‐associated antigens (Ags), which leads to the generation of RNA‐containing immune complexes (ICs). These complexes act as TLR‐7 ligands through Fc receptor (FcR) and B cell receptor (BCR)–mediated uptake and induce TLR‐7 protein expression through a positive feedback loop at the systemic 3 and organ 4 levels. A simultaneous increase in the levels of TLR‐7 and its ligands leads to ongoing inflammation and tissue destruction through various mechanisms, including T cell proliferation and inflammatory cytokine generation. Anti‐snRNP = anti–small nuclear RNP; pDCs = plasmacytoid DCs; PMNs = polymorphonuclear leukocytes; Mϕ = macrophages.

We and others have proposed a multistep model of lupus pathogenesis whereby there is an initial loss of tolerance to self, which, together with an additional immune alteration, leads to the progression of severe disease 42, 43, 44. We propose that the loss of tolerance to RNA plays a key role in the transition to active disease. Surprisingly, anti‐RNA antibodies have not been studied extensively, despite showing high specificity for SLE 28. However, autoantibodies to RNA‐associated molecules, such as snRNP, Sm, SSA/Ro, and SSB/La, are well characterized, and the presence of anti‐Sm antibodies is part of the immunologic criteria for SLE diagnosis 45.

BWR4, an anti‐RNA antibody derived from (NZB × NZW)F1 mice, forms ICs in vitro by binding RNA present in the tissue culture supernatant 37. These RNA‐containing ICs then activate cells through a B cell receptor (BCR)– and TLR‐7–dependent mechanism. We showed that Sle1TLR‐9−/− mouse serum predominantly produces a cytoplasmic HEp‐2 staining pattern, similar to the anti‐RNA antibody BWR4 37. The increase in the cytoplasmic HEp‐2 staining pattern and/or increased anti‐RNA antibody levels are common to the few TLR‐9–deficient model systems that have been examined 8, 9, 14. Increased levels of RNA‐associated autoantibodies, such as anti‐Sm, U1 snRNP‐1, PM/Scl‐100, and/or anti–ribosomal P, were also observed in TLR‐9−/− models that develop severe kidney disease 8, 9, 10. On the other hand, when TLR‐9 was eliminated on a milder autoimmune background, such as Sle1b, there was no increase in anti‐Sm/RNP IgG 31. This might have important implications for human SLE, since there is evidence that anti‐Sm and anti‐snRNP antibodies are present immediately preceding the manifestation of clinical symptoms 42.

Aside from the autoantibody switch to RNA specificity, our data also suggest that the concomitant increase in TLR‐7 plays a crucial role in the progression to end‐organ damage. Since SLE serum and the TLR‐7 agonist R848 have been shown to increase TLR‐7 expression 19, 46, the RNA‐associated ICs in the circulation can increase TLR‐7 through a positive feedback loop. This is supported by clinical data that shows a preferentially higher TLR‐7 expression in SLE patients with an anti‐RNA–associated antibody profile 47. Additionally, it has recently been shown that RNA‐associated antibodies are more prone to form circulating ICs compared to antibodies to dsDNA 48. U1 RNA, which forms the U1 snRNP complex together with Sm and RNP proteins 49, is elevated in the circulation of SLE patients and correlates with disease activity 50. Thus, while anti‐RNA/RNA‐associated antibodies might be less common than anti‐dsDNA in SLE patients 28, 49, they can form ICs more easily and enhance disease progression through TLR‐7 binding and up‐regulation (Figure 6).

In our earlier investigations we proposed that renal cDCs play a fundamental role in disease progression given their location, their increased inflammatory properties, and the correlation of their expansion with disease progression 15. Consistent with the findings of this study, Sle1TLR‐9−/− renal cDCs have increased levels of TLR‐7 protein and have an increased ability to stimulate T cells following TLR‐7 ligation. Furthermore, TLR‐7 in these populations correlated with leukocyte infiltration. Importantly, our analyses in younger mice indicated that the increase in renal cDCs occurred prior to significant leukocyte infiltration, suggesting that cDCs play a crucial role in disease progression.

In summary, we have identified multiple roles for the innate dsDNA receptor, TLR‐9, in preventing the systemic inflammation and progression to severe disease in a lupus‐prone mouse strain. In B cells, TLR‐9 prevents excessive Ig production and the switch to RNA‐reactive antibody production. In addition, TLR‐9 controls TLR‐7 at the protein level, most likely in an indirect manner by preventing TLR‐7 ligands from being generated and by favoring DNA sensing over RNA sensing in the endosome 51. In the absence of TLR‐9, the concomitant increase in the levels of TLR‐7 and its ligands leads to ongoing inflammation and tissue destruction.

Author Contributions

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Drs. Celhar and Fairhurst had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design

Celhar, Lim, Akira, Wakeland, Connolly, Fairhurst.

Acquisition of data

Celhar, Yasuga, Lee, Zharkova, Tripathi, Thornhill, Lu, Au, Lim, Thamboo.

Analysis and interpretation of data

Celhar, Au, Thamboo, Fairhurst.

Click here for additional data file.