Dosage of X-linked Toll-like Receptor 8 determines gender differences in the development of systemic lupus erythematosus

Abstract

Systemic lupus erythematosus (SLE) is an autoimmune disease with a high incidence in females and a complex phenotype. Using 564Igi mice, a model of SLE with knock-in genes encoding an autoreactive anti-RNA antibody, we investigated how expression of Toll-like receptors (TLRs) in B cells and neutrophils affects pathogenesis. We established that TLR signaling through MyD88 is necessary for disease. Autoantibody was produced in mice with single deletions of Tlr7, Tlr8 or Tlr9 or combined deletions of Tlr7 and 9. Autoantibody was not produced in the combined absence of Tlr7 and 8, indicating that TLR8 contributes to the break in tolerance. Furthermore, TLR8 was sufficient for the loss of B cell tolerance, the production of class-switched autoantibody, heightened granulopoiesis, and increased production of type I interferon (IFN-I) by neutrophils as well as glomerulonephritis and death. We show that dosage of X-linked Tlr8 plays a major role in the high incidence of disease in females. In addition, we show that the negative regulation of disease by TLR9 is exerted primarily on granulopoiesis and IFN-I production by neutrophils. Collectively, we suggest that individual TLRs play unique roles in the pathogenesis of SLE, suggesting new targets for treatment.

Introduction

SLE is an autoimmune disorder characterized by autoantibodies against nuclear self-antigens [1]. It is a disease that primarily affects females [2] and is strongly associated with genes such as Irf5, Irf7, Irak1, Tnfaip3, Tnip1, Ifh1, Tyk2 and C1q [3]. SLE patients typically exhibit up-regulation of Ifn-I and the signature genes of heightened granulopoiesis. [4] [5, 6] However, it is unclear how these factors contribute to the pathogenesis of disease. Our study investigates the cellular and molecular mechanisms that mediate increased granulopoiesis, heightened production of IFN-I, autoantibody and a predilection for females in a mouse model of SLE.

In several mouse models of autoimmune disease the activation of self-reactive B cells resulted when endogenous nucleic acid antigens synergistically engaged B cell receptors (BCR) and TLRs [7] [8]. The TLRs that recognize nucleic acids are TLR3 (double stranded (ds) RNA), TLR7 (single stranded (ss) RNA), TLR8 (ssRNA) and TLR9 (un-methylated CpG and dsDNA). TLR7 and TLR9 have both been shown to be involved in SLE autoantibody production in mouse models [9] [10] [11] [12] [13] [14] [15]. The role of TLR7 in SLE pathogenesis was first revealed when FcγRIIb deficient C57BL/6 (B6.RIIb^−/−) and Sle-1 congenic mice were crossed to mice bearing the Y chromosome linked autoimmune-accelerator (Yaa). Male offspring developed autoantibody to nucleoli and accelerated disease [10, 11]. Because the Yaa mutation is a translocation of the telomeric end of the X-chromosome that includes Tlr7 and Tlr8 onto the Y-chromosome, this observation suggested that these genes contribute to the phenotype. Further evidence that Tlr7 is partially responsible for the autoimmune phenotype came with the observation that mice transgenic for multiple copies of Tlr7 developed severe autoimmunity [12].

The belief that the phenotype of Yaa is attributed solely to Tlr7 duplication [10] [11] was put into question by a report that the Yaa phenotype is not completely abrogated by the deletion of Tlr7 [13] [14]. Further, in MRL/lpr mice, another model of SLE, deficiency of Tlr7 had no effect on anti-DNA antibodies but prevented the appearance of anti-Sm autoantibodies while Tlr9 deletion resulted in diminished anti DNA-antibody but augmented hypergammaglobulinemia, lymphocyte activation, and glomerulonephritis [16]. Subsequent studies confirmed that Tlr9-deficient MRL/lpr mice had more severe disease [17] [18] [19] suggesting that TLR9 might act as a negative regulator. Further studies revealed that MyD88 deficiency totally abrogated autoantibody production in autoimmune MRL/lpr mice [15]. MyD88 is an adaptor protein that is utilized by most TLRs and importantly, specifically mediates signals transduced by TLR7, 8 and 9 binding of nucleic acid antigens. Since MyD88 is critical for autoantibody production of MRL/lpr mice and TLR7 and TLR9 are not responsible for all the features of SLE, it would be reasonable to ask if TLR8 plays a role in SLE pathogenesis

In order to further elucidate the mechanisms involved in the development and pathogenesis of SLE and the role of TLR8 in this disease, we have utilized the 564Igi mouse model, which was created in our laboratory and previously described [9]. In brief, 564Igi is a knock-in mouse in which rearranged heavy chain and light chain genes from the 564 hybridoma (derived from an autoimmune SWR X NZB F₁ mouse) were introduced into the IgH and IgL loci of a C57BL/6 mouse. Antibodies purified from a 564 hybridoma are pathogenic as their injection into young (pre-autoimmune) female F₁ (SWRxNZB) mice accelerated the appearance of glomerulonephritis [20]

564Igi mice have auto-reactive B-cells that carry the 564Igi B-cell receptor (BCR) and have IgG2a and IgG2b autoantibodies in their sera. These autoantibodies bind nucleoli and cytoplasmic antigens suggesting that they bind RNA or RNA associated proteins. The production of autoantibodies in 564Igi is partially dependent on TLR7, which recognizes ssRNA. Deletion of Tlr7 in 564Igi significantly reduces autoantibody; however, it does not completely eliminate it [9]. These results suggest that another nucleic acid sensing TLR such as TLR8 and/or another molecule might be involved in the activation of B cells. We hypothesized that TLR8 was an excellent candidate since it also sensed ssRNA, and its gene is a part of the Yaa translocation (Pisitkum 2006).

Increased type I interferon (IFN-I) production has been found in SLE patients [21, 22] [23, 24]. The involvement of IFN-I in SLE is further supported by the observation that a subset of patients with SLE with severe disease expressed an IFN-I inducible gene signature [4] [5]. In addition, genome-wide association studies provide strong evidence that IFN-1 is an important SLE risk factor [3]. Because IFN-I production is a key feature of SLE, the characterization of its cellular sources may be crucial for the development of new therapeutic strategies aimed at controlling the early stages of disease.

We reported recently that 564Igi mice develop increased granulopoiesis, have increased Ifn-I production and increased Ifn-I signature genes in B cells and neutrophils as do SLE patients [27]. Here we have asked whether the expression of specific TLRs contribute to these disease factors. We report that while TLR7 and TLR9 are important for autoantibody production as previously shown [28], TLR8 can also drive the production of autoantibody. Furthermore, we define a novel role for TLR8 as the major contributor to increased granulopoiesis and neutrophil Ifn-I production in the 564Igi mouse model of SLE. In addition, dosage of Tlr8 on the X-chromosome is responsible for the increased frequency of disease in females.

Results

IgG anti-nuclear antibody production was MyD88 dependent

As previously shown [9], 564Igi mice produced IgG antibodies that recognize self nucleic acids. These antibodies were often class switched to IgG2a and IgG2b. Sera from 564Igi stained the cytoplasm and nucleoli of HEp-2 cells (Fig. 1A and B). This suggests that at least some of the IgG antibodies produced in these mice were reactive to RNA or RNA associated macromolecules.

Figure 1. 564Igi IgG anti-nuclear antibody production is MyD88 dependent; TLR7 contributes to but is not required for autoantibody production.

(A) IgG antinuclear antibody (ANA) staining of HEp-2 cells using sera C57BL/6, 564 Igi, 564Igi Tlr7−/− and 564Igi MyD88−/− at 100X magnification. Each figure is representative of studies of at least three mice. (B) Intensity in pixels of the fluorescent signals from ANA staining of Hep-2 cells (C) Serial dilution of sera from 564Igi, 564Igi Tlr7−/−, 564Igi Myd88−/− and C57BL/6 mice tested for binding to yeast RNA. Anti-IgG2a, IgG2b and IgM alkaline phosphatase (AP) labeled secondary antibodies were used to detect bound antibody of different isotypes (D). IgG2a anti-RNA antibodies from females and males 564Igi and 564Igi Tlr7−/− are measured. (C,D) Data are shown as means +/− SEM of the number of sera tested indicated on the graph and are representitive of many independent experiments. Mice were tested at 3 to 4 months of age. A two tailed Student’s t-test was run, statistically significance at p<.05. * p< .05, ** p< .01, *** p< .001 and **** p< .0001.

A monoclonal antibody that was raised against the heavy and light chain knock-in of the 564Igi mouse allowed us to measure the presence of 564 idiotype positive (Id+) antibodies of various isotypes in mouse sera. As previously shown, 564Igi mice had high levels of Id+ IgG2a and IgG2b (Supporting Information Fig. 1A; [9]) however, there was very little circulating anti RNA IgM antibody (Fig. 1C). These results were consistent with previous findings that although most 564 anti RNA-specific IgM+ B cells in the periphery were anergic, IgG and IgA Id+ auto-reactive antibodies were produced [9].

Tlr7 was critical for 564Igi to produce high levels of IgG Id+ antibodies (Supporting Information Fig. 1A). However, both female and male 564Igi deficient in Tlr7 had some autoantibody (Fig. 1A–E; [9]). Therefore, it is possible that other factors, including another TLR, contributes to the production of autoantibodies.

To investigate whether TLRs other thanTLR7 mediate the production of autoantibodies, a MyD88 deficient mouse was crossed to 564Igi. The 564Igi MyD88−/−had no circulating IgG that stained the nuclei of HEp-2 cells (Fig. 1A and B) and no anti RNA IgG2a and IgG2b antibodies (Fig. 1C) nor Id+ antibodies (Supporting Information Fig. 1A). This indicated that additional TLRs might contribute to the generation of autoantibodies. Further, MyD88 is also utilized by the IL-1 receptor type 1 signaling pathway and for stabilizing Ifn-γ mRNA [29]. We found that there were high levels of IgG autoantibody in sera from 564Igi Il1r −/− mice (Supporting Information Fig. 2A). Therefore, while IL-1R signaling is dependent on MyD88, it does not play a necessary role in autoantibody production in 564Igi mice. In order to investigate whether other TLRs play a role in autoantibody production we created additional Tlr knockouts on the 564Igi background.

564Igi Tlr9−/−, and Tlr7/9−/− produced IgG autoantibody

To see if TLR9 could compensate for the loss of TLR7 to allow autoantibody production in 564Igi mice we created 564Igi Tlr9−/− mice. We found that female (Fig. 2A–C) and male (data not shown) 564Igi Tlr9−/− mice produced high levels of 564 Id+ antibodies that stained the nucleoli of HEp-2 cells. In females and males the sera from 564Igi Tlr9−/− mice had no more IgG2 anti RNA antibodies than 564Igi mice (Fig. 2C). 564Igi Tlr9−/− mice have in the sera antibodies that bind nucleoli of HEp-2 cells (Fig. 2A–B). This indicated that TLR9 was not critical for production of anti-RNA antibody. Similarly, TLR8 was not critical for autoantibody production (Fig. 2D).

Figure 2. Tlr9 deficiency has no effect on autoantibody production; female 564Igi Tlr7/9−/− produce more autoantibody than males.

(A) IgG ANA staining of HEp-2 cells using sera from 564 Igi, 564 Igi Tlr9−/−, 564Igi Tlr8−/− and 564 Igi Tlr7/9−/− at 100X magnification. Representative data from at least three mice are shown. (B) Intensity of fluorescent signal from ANA staining of Hep-2 cells. Outliers (p<.05) were removed according to Grubb’s outlier test. (C, D, E) Anti RNA IgG2a and IgG2b antibodies detected in the sera of mice by ELISA using yeast RNA as capture antigen and detecting binding with AP conjugated anti-IgG2a or anti-IgG2b measures the amount of RNA binding. Sera from 564Igi, 564 Igi Tlr9−/−, 564Igi Tlr8−/− and 564 Igi Tlr7/9−/− (both female and male) and C57BL/6 mice were tested. (C–E) Data are shown as the means +/− SEM of the number of sera tested indicated on the graph and are representative of many independent experiments. Mice were tested at 3 to 4 months of age. A two tailed Student’s t-test was run, statistically significance at p<.05. * p< .05, ** p< .01, *** p< .001 and **** p< .0001. F= Female; M=Male.

Tlr7 and Tlr9 double knockouts were crossed onto the 564Igi background in order to see if the combined absence of these receptors would abolish autoantibody production, as did deficiency of Myd88. Surprisingly, female 564Igi mice lacking Tlr7 and Tlr9 produced IgG anti RNA antibodies detected through ANA staining and ELISA (Fig. 2A–B and E). Some male mice had detectable levels of Id+ antibodies (Supporting Information Fig. 1D); however, they were no RNA reactive antibodies in the sera from male 564Igi Tlr7/9−/− mice (Fig. 2E). This suggests that some form of receptor editing must have made these Id+ antibodies non-reactive to RNA. The lack of autoantibodies in female 564Igi Myd88−/− and the presence of autoantibodies in female 564Igi Tlr7/9−/−mice suggested that signaling through another TLR must be important for autoantibody production. TLR8 was a possible candidate because of its ability to detect viral or self-RNA and signal through MyD88. Thus, 564Igi Tlr7/9−/− is a mouse model of SLE that mimics the propensity for clinical disease to arise in females more often than in males. Note that there is no observed difference in autoantibody production between male and female 564Igi or 564Igi Tlr7^−/− mice (Fig. 1D).

The expression of Tlr7 and Tlr8 was critical for IgG anti-nuclear antibody production

In order to investigate whether TLR8 is required for the production of autoantibodies in 564Igi Tlr7/9−/− mice, 564Igi Tlr7/8−/− mice were developed. Sera from these mice contained no detectable anti-nuclear IgG antibodies by HEp-2 cell staining (Fig. 3A and B). There were also no anti RNA IgG2a or IgG2b nor were there Id+ antibodies in 564Igi Tlr7/8−/− mice (Fig. 3C and Supporting Information Fig. 1E). However, an ELISA that detects any IgG2a and IgG2b antibody regardless of specificity revealed that 564Igi Tlr7/8−/− mice had the ability to class switch to IgG2a and IgG2b encoding genes (Fig. 3D). In 564Igi IgG1 antibodies are abundant, but IgG1antibody does not bind RNA (Supporting Information Fig. 2B). 564IgiTlr7/8−/− mice have high amounts of IgG1 in their sera (Supporting Information Fig. 2C) demonstrating the ability for B cells to class switch genes despite the combined absence of Tlr7 and Tlr8.

Figure 3. Expression of Tlr7 or Tlr8 is critical for IgG anti-nuclear antibody production in 564Igi.

(A) IgG ANA staining of HEp-2 cells using sera from C57BL/6, 564Igi, and 564Igi Tlr7/8−/− at 100X magnification. Data representative of studies of at least three mice are shown. (B) ANA staining intensity of Hep-2 cells. (C) Anti RNA IgG2a and IgG2b antibodies detected in the sera of mice by ELISA using yeast RNA as capture antigen and detecting binding with AP conjugated anti-IgG2a or anti-IgG2b measures the amount of RNA binding. Sera from C57BL/6, 564Igi, and 564Igi Tlr7/8−/− mice assayed by ELISA (D) Total IgG2a and IgG2b detected in the sera of mice by ELISA using anti-IgG2a or anti-IgG2b capture antibodies and detecting binding with AP conjugated anti-IgG2a or anti-IgG2b antibodies, respectively. (E) Immune complexes containing IgG2a antibody detected in the sera of mice by ELISA with C1q as a capture molecule. (C–E) Data are shown as the means +/− SEM of the number of sera tested indicated on the graph and are representative of many independent experiments. Mice were tested at 3 to 4 months of age. A two tailed Student’s t-test was run, statistically significance at p<.05. * p< .05, ** p< .01, *** p< .001 and **** p< .0001. F) Tlr7 and Tlr8 are up-regulated in B220⁺ B cells from the bone marrow of 564 Igi mice. Relative gene expression of Tlr7, Tlr8 and Tlr9 to B-actin in BM B220+ cells was measured by real-time PCR (qPCR) from C57BL/6 (n=3), 564Igi-B6 (n=3), 564Igi Tlr7/8−/− and 564Igi Tlr7/9−/− (both female and male), 564Igi Tlr9−/− and 564Igi Tlr8−/−. 3 mice tested for each strain. 3 dilutions of cDNA tested from each mouse, each dilution was tested in triplicate. (F) Data are shown as the mean +/− SEM of n=3 samples and are pooled from 3 independent experiments A two tailed Student’s t-test was run, statistically significance at p<.05. * p< .05, ** p< .01, *** p< .001 and **** p< .0001. Mice were tested at 3 to 4 months of age.(G) Western blots for TLR7 and TLR8 in B cells from the BM. TLR7: and TLR8:β-Actin densitometry ratios are shown.

FACs analysis revealed that B cells in the bone marrow of 564Igi mice expressed Id+ surface IgM and IgG (Supporting Information Fig. 3). 564Igi Tlr7/8−/− mice did not have Id+ IgG2a+ (Supporting Information Fig. 3), thus confirming that Tlr7 or Tlr8 was necessary for the production of class switched pathogenic antibodies in 564Igi mice.

There were IgM+ Id+ B cells in both 564Igi and 564Igi Tlr7/8−/−. In 564Igi mice there were B cells that were high and low membrane (m) IgM and that all the low mIgM were Id+. In the Tlr7/8−/− mice the B cells were all Id+ and all had low mIgM (Supporting Information Fig. 3). However, 564Igi Tlr7/8−/−, like 564Igi mice (Fig. 1C), had very little circulating, secreted Id+ IgM, indicating that they were anergic.

The relationship between autoantibody production by B cells and Tlr expression is summarized in Table 1. Unlike B cells from C57BL/6 mice, those from 564Igi produced anti-RNA, Id+, class switched antibodies. The production of these autoantibodies leads to the formation of immune complexes detected in the sera (Fig. 3E). The loss of Tlr7 in 564Igi mice did not completely deplete antibody production; however, combined deficiencies of Tlr7 and Tlr8 did. It is important to note that B cells from female Tlr7/Tlr9 deficient 564Igi mice still had IgG RNA binding antibodies. In 564Igi Tlr7^−/− and female Tlr7/9−/− there were immune complexes, but not in 564Igi Tlr7/8−/− or male Tlr7/9−/− (Fig. 3E). The data suggest that in Tlr7/9−/− females TLR8 is sufficient for production of class-switched autoantibodies that lead to pathogenic immune complexes.

Table 1.

Summary of the correlation between the expression of the Tlrs with antibody production, granulopoiesis, IFN-I expression and pathology.

B cells	C57BL/6	564Igi	564Igi Tlr7−/−	564Igi Tlr8−/−	564Igi Tlr9−/−	564Igi Tlr7/9−/−		564Igi Tlr7/8−/−
	Female	Female	Female	Female	Female	Female	Male	Female
* Anti RNA IgG antibodies	−	+++	+	+++	+++	+	−	−
Tlr7 expression in B cells	+	+++	−	+	++	−	−	−
Tlr8 expression in B cells	+	+++	++	−	++	++++	++	−
Tlr9 expression in B cells	++	+	+	+	−	−	nd**	+
Neutrophils	C57BL/6	564Igi	564Igi Tlr7−/−	564Igi Tlr8−/−	564Igi Tlr9−/−	564Igi Tlr7/9−/−		564Igi Tlr7/8−/−
Granulopoiesis	Minimal	+++	+++	+++	++++	+++	+	Minimal
Ifn-I expression in Neutrophils	+	+	+++	+	+++	++++	+	+
Tlr7 expression in Neutrophils	+	++++	−	+	−	−	−	−
Tlr8 expression in Neutrophils	+	++	+	−	+	+	+	−
Tlr9 expression in Neutrophils	+	++++	−	+	−	−	−	+
Pathology
Glomerulonephritis	−	++	++++	nd**	++++	++++	nd**	−

^*anti-RNA as measured by Hep-2 staining and ELISA

^**nd= not done

Tlr7 and Tlr8 were up regulated in B cells from the bone marrow of 564 Igi mice

The expression of Tlr7 in auto-reactive B cells was found to be critical for B cell activation by self-RNA [8]. It has also been shown, that Tlr7 over expression leads to severe autoimmunity [10] [13]. There were higher levels of Tlr7 and Tlr8 transcripts in B cells from 564Igi mice compared to B cells from C57BL/6 mice (Fig. 3F and Table 1). Western Blots of cell lysates from B cells showed an increase in TLR7 and TLR8 protein levels (Fig. 3G). In contrast, Tlr9 was not up-regulated in 564Igi B cells (Fig. 3F)

The expression of Tlr8 mRNA was highly up-regulated in B cells from female 564Igi Tlr7/9−/− mice (Fig. 3F and Table 1). These mice had activated auto-reactive B cells that produced class switched IgG anti nuclear antibodies in the absence of both Tlr7 and Tlr9 (Fig. 2A–D and Table 1). These results indicated that the overexpression of Tlr8 in B cells is crucial for the production of RNA binding autoantibodies in 564Igi. There are much lower levels of Tlr8 transcripts in B cells from male 564Igi Tlr7/9−/− (Fig. 3F). On the other hand, while there was not much up-regulation of Tlr7 nor Tlr9 in B cells of 564Igi Tlr8^−/−, nevertheless they produce as much IgG anti-RNA antibodies as 564Igi mice (Fig. 2D), suggesting that Tlr8 contributes little to autoantibody production compared to Tlr7.

Granulopoiesis in 564Igi mice was driven by TLR8

564Igi mice have been shown to have increased numbers of granulocytes and increased production of IFN-I similar to that found in SLE patients [27] [4]. We therefore investigated how deficiency of Tlr7, 8 or 9 affected granulopoiesis and IFN-I expression.

The total number of cells recovered from the bone marrow of C57BL/6, 564Igi and all of the Tlr-deficient mice were similar (data not shown), therefore, the percentage of cells shown in Figure 4A reflects the absolute number of neutrophils in the bone marrow. In the bone marrow of 564Igi a higher percentage of nucleated cells were Cd11b+/Ly-6c^med (neutrophils [30]) than in C57BL/6 mice (Fig 4A and B and [27]). Furthermore, 564Igi MyD88−/− and C57BL/6 mice had similar percentages of neutrophils (Fig. 4B), indicating that neutrophil increases in 564Igi is dependent on TLR signaling.

Figure 4. Granulopoiesis is mediated by TLR signaling.

(A) Flow cytometric profile of neutrophils (CD11b hi Ly-6C intermediate [30]) in the BM of representative age-matched C57BL/6 and 564Igi mice. The fraction of viable (propidium iodide⁻) CD11b⁺Ly-6C⁺ cells from the BM of the indicated mice was determined by flow cytometry. To determine the fold increase of neutrophils in each experiment the percentage of BM cells staining for neutrophil (CD11b high, Ly-6C low) of control C57BL/6 mice was determined and set as 1.0. Each symbol represents ratio with respect to C57BL/6″ BM cells from one mouse. (B) Compiled results of the ratio with respect to CD11b⁺Gr-1^hi and CD11b⁺Ly-6C^hi cells in BM of multiple age-matched mice analyzed as shown in (A): C57BL/6, 564Igi, and 564Igi Tlr7, 8 and 9 deficient mice. Significance determined by two tailed Student’s t-test. * p< .05, ** p< .01, *** p< .001 and **** p< .0001. (C) Tlr7 and Tlr8 are up-regulated in bone marrow neutrophils (Ly-6G+) from 564 Igi mice. Relative gene expression of Tlr7, Tlr8 and Tlr9 to B-actin in BM Ly-6G⁺ cells measured by real-time PCR (qPCR) from C57BL/6 (n=3) or 564Igi-B6 (n=3), 564Igi Tlr7/8−/− and 564Igi Tlr7/9−/− (both female and male), 564Igi Tlr7−/−, 564Igi Tlr9−/− and 564Igi Tlr8−/−. 3 mice tested for each strain. 3 dilutions of cDNA tested from each mouse, each dilution was tested in triplicate. Arithmetic means from each dilution are presented. A two tailed Student’s t-test was run, statistically significance at p<.05. * p< .05, ** p< .01, *** p< .001 and **** p< .0001. (C) Data are shown as the mean +/− SEM of n=3 samples and are pooled from 3 independent experiments. Mice were tested at 3 to 4 months of age. Western blot for TLR7 and TLR8 in (D) BM neutrophils Tlr7: and Tlr8:β-Actin densitometry ratios are shown.

An increase in neutrophils was observed in female 564Igi Tlr7/9−/− double knockout mice, and there was no bone marrow granulopoiesis in 564Igi Tlr7/8−/− (Fig. 4B). This demonstrated that Tlr8 was sufficient for a MyD88 dependent increase in neutrophils. Indeed, 564Igi Tlr8−/− mice had much fewer neutrophils than 564IgiTlr7^−/−mice. Male 564Igi Tlr7/9−/− had less neutrophils than females, but slightly more than C57BL/6 (Fig. 4B). Both Tlr7−/− and Tlr9−/− 564Igi mice had a large increase of neutrophils in the bone marrow (Fig. 4B), suggesting that neither TLR7 nor TLR9 can be solely responsible for increased granulopoiesis.

Tlr7, Tlr8 and Tlr9 were up-regulated in neutrophils of 564Igi mice

MyD88 dependent granulopoiesis in 564Igi could be due to the activation of neutrophils mediated by TLRs. We therefore wanted to investigate the expression of Tlr7, Tlr8, and Tlr9 in 564Igi neutrophils. An increase of Tlr7, Tlr8 and Tlr9 mRNA transcripts were observed in Ly-6G+ neutrophils in the bone marrow of 564Igi mice compared to C57BL/6 (Fig. 4C). Western blots reveal that neutrophils had more TLR7 in 564Igi bone marrow compared to C57BL/6, while TLR8 seems to be expressed at similar levels in neutrophils in both 564Igi and C57BL/6 (Fig. 4D).

C57BL/6 and 564Igi Tlr7/9−/− neutrophils had similar levels of Tlr8 mRNA (Fig 4C). Female 564Igi Tlr7/9−/− mice produce auto-reactive antibodies (Fig 2A–D). Unlike in B cells, the neutrophils in these mice express less Tlr8 mRNA than 564Igi mice (Fig. 4C). However, Tlr7/9−/− mice do not express Tlr9 and they have the striking autoimmune phenotype including heightened granulopoiesis (Fig. 4B) and Ifn-I expression (Fig. 5A). These results suggest that increased granulopoiesis and Ifn-I expression can be achieved with Tlr8 in the absence of Tlr7 or Tlr9 expression (Table 1).

Figure 5. Tlr9 suppresses Ifn-I production by neutrophils from 564Igi mice.

(A) Gene expression of Tlr9, Ifn-β1, and Ifn-α6 relative to Actb was determined by qPCR with cDNA from BM Ly-6G+ cells. 3 mice tested for each strain. 3 dilutions of cDNA tested from each mouse, each dilution was tested in triplicate. Arithmetic means from each dilution are presented. Significance was determined by two tailed Student’s t-test. * p< .05, ** p< .01, *** p< .001 and **** p< .0001. n=number of mice tested. (A) Data are shown as the mean +/− SEM of n=3 samples and are pooled from 3 independent experiments Mice were tested at 3 to 4 months of age. (B) In the absence of TLR7 and TLR9, TLR8 mediates glomerulonephritis in 564Igi mice. Kidneys were harvested from various mouse strains, sectioned and then stained with Periodic Acid Schiff (PAS) (left panels) and Alexa 488 conjugated anti 564-idiotype antibody (right panels). Mice were aged 10 to 16 months and then were sacrificed and organs were harvested.

Tlr8 mediated and Tlr9 inhibited Ifn-I expression in neutrophils

IFN-1 is a critical factor for SLE pathology[3]. We therefore investigated the expression of Ifn-I in neutrophils from 564Igi and Tlr deficient 564Igi mice. Neutrophils from 564Igi and C57BL/6 mice expressed Ifn-α6 at similar levels and there was an increase in Ifn-β1 (Fig. 5A) in 564 Igi. There were more neutrophils in 564Igi than C57BL/6 (Fig. 4B). Therefore, neutrophils could be responsible for more IFN-I production in 564Igi mice, even without a large increase in Ifn-I expression on a per cell basis. Female 564Igi mice deficient in Tlr7 and Tlr9 have high levels of Ifn-α6 and -β1 mRNA (Fig. 5A). These results suggest that Tlr8 can mediate an increase of Ifn-I expression in female 564Igi Tlr7/9−/− despite low levels of Tlr8 expression, but not in males (Fig. 5A Table 1).

High levels of neutrophils and autoantibody production in 564Igi Tlr9−/− mice suggest that TLR9 is a negative regulator of these disease indicators. The expression of Tlr9 in 564Igi neutrophils is higher than in C57BL/6 neutrophils (Fig. 4A), but not in B cells (Fig. 3F). Interestingly, neutrophils from 564Igi Tlr9−/− mice had an increase Ifn-I expression (Fig. 5A). This suggests that TLR9 may be regulating disease by suppressing Ifn-I expression. The upregulation of Ifn-I does not correlate with an upregulation of either Tlr7 or Tlr8 in these neutrophils (Fig. 4C). High levels of Tlr9 expression in 564Igi neutrophils (Fig. 4C) may be responsible for the relative low levels of Ifn-I in these cells (Fig. 5A).

Surprisingly, Tlr7−/− mice also had high Ifn-I expression in neutrophils (Fig. 5A). This may be attributed to the striking downregulation of Tlr9 in these neutrophils (Fig. 4C). Tlr7−/− neutrophils expressed low levels of Tlr8 (Fig. 4C) in the absence of significant Tlr9 transcription (Fig. 4C). The expression of Ifn-I in neutrophils of 564Igi Tlr7−/− and Tlr7/9 −/− were similar (Fig. 5A and Table 1).

Female Tlr7/9 deficient 564Igi mice had increased disease pathology compared to 564Igi mice

To investigate whether the antibodies produced in 564Igi Tlr7/9−/− leads to SLE–like pathology, we examined kidney histology of age-matched female mice. 564Igi Tlr7/9−/− mice had significant levels of kidney damage and IgG ICs are observed in their kidneys (Fig. 5B). The kidneys were stained by PAS and lupus nephritis was scored based on the International Society of Nephrology/Renal pathology Society (ISN/RPS) classification of lupus nephritis (2003) (Table 2). 564Igi Myd88−/− mice had no Id+ autoantibodies in their sera or deposits in their kidneys and their glomeruli had no signs of nephritis (Fig 5B). It is noteworthy that 564Igi deficient in Tlr9, Tlr7 orTlr7/9 all had accelerated disease (Fig. 5B and Table 2). High levels of disease in the kidneys from Tlr7, Tlr9 and Tlr7/9 deficient 564Igi mice correlated with an increase of neutrophils, high Ifn-I expression and low Tlr9 expression in neutrophils (Table 1). Young 564Igi Tlr7/8−/− mice had healthy glomeruli and no Id+ ICs in their kidneys. Therefore TLR8 can mediate kidney pathology in 564Igi mice.

Table 2.

Degree of Lupus Nephritis In 564Igi Mouse Lineages

Genotype	Class
C57BL/6	Negative
564 Igi	Class II – IVS
564 Myd88 KO	Negative
564 Tlr7 KO	Class III – IVG
564 Tlr9 KO	Class IV – S
564 Tlr7/Tlr9 DKO	Class III+ - IVG
564 Tlr7/Tlr8 DKO	Negative-Class II

When aged, several 564Igi Tlr7/8−/− mice older than 12 months died and had signs of inflammation including enlarged spleens and some developed lymphoma (data not shown). We PAS stained the kidneys of aged mice to reveal any pathology, including neutrophil infiltration. Few mice had Id+ antibody in their kidneys (data not shown). Other 564Igi Tlr7/8−/− mice that died of causes other than glomerulonephritis, had no Id+ antibody or kidney damage.

A single copy of Tlr8 564Igi Tlr7/9^−/− mice is not sufficient for autoantibody and Ifn-I expression

Tlr7 and Tlr8 are on the X chromosome and therefore males have one copy of each gene. The observed difference between male and female 564Igi Tlr7/9−/− in autoantibody production, granulopoiesis and Ifn-1 expression could be explained by hormonal differences between females and males. Another alternative is that these differences are due to Tlr8 copy number. In order to test the latter possibility we generated 564Igi females with a single copy Tlr8 gene on a Tlr7/9 deficient background by breeding (Fig. 6A).

Figure 6. A single copy of Tlr8 in female 564Igi Tlr7/9^−/− mice is not sufficient to produce autoantibody, granulopoiesis or Ifn-1 expression.

(A) Generation of female 564Igi mice with a single copy of Tlr8. Tlr7/9−/− female mice were bred to a Tlr7/8/9−/− male, generating females with one copy of Tlr8. (B) Anti RNA ELISA testing of sera from females with a single copy of Tlr8 on a background of 564Igi Tlr7/9−/−. Data are shown as means +/− SEM of the number of sera tested indicated on the graph and are pooled from 1 independent experiment (C) ANA staining intensity on Hep-2 cells. (D) Relative gene expression of Tlr8 to Actb determined by qPCR with cDNA from BM Ly-6G+ cells 3 mice tested for each strain. 3 dilutions of cDNA tested from each mouse, each dilution was tested in triplicate. (D) Data are shown as the means +/− SEM of n=3 and are pooled from of 3 independent experiments (E) Relative gene expression of Ifn-a6 to Actb determined by qPCR with cDNA from BM Ly-6G+ cells 2 mice tested for each strain. 3 dilutions of cDNA tested from each mouse, each dilution was tested in triplicate. Data are shown as the means +/− SEM of n=3 and are pooled from of 2 independent experiments (F) Relative percentage of BM CD11b hi Ly-6C intermediate neutrophils compared to the percentage of BM CD11b hi Ly-6C intermediate neutrophils in C57BL/6 mice (B–E) Significance was determined by two tailed Student’s t-test. * p< .05, ** p< .01, *** p< .001 and **** p< .0001. Mice were tested at 3 to 4 months of age.

The sera of female 564Igi Tlr7/9−/− mice with a single copy of Tlr8 had no detectable IgG autoantibody as revealed by an anti RNA ELISA (Fig. 6B) and ANA Hep-2 cell staining (Fig. 6C). These results are similar to those found with sera from male 564Igi Tlr7/9−/− mice (Fig. 2E), which also only have one copy of Tlr8 and do not produce autoantibody. The B cells of female 564Igi Tlr7/9−/− mice with a single copy of Tlr8 had lower levels of Tlr8 transcript than female 564Igi Tlr7/9−/− B cells; however, female 564Igi Tlr7/9−/− mice with a single copy of Tlr8 B cells had higher Tlr8 transcript levels compared to male 564Igi Tlr7/9−/− (Fig. 6D). Furthermore, Ifn-I expression and granulopoiesis were similar to that observed in male 564Igi Tlr7/9−/− mice (Fig. 6E and F). These results are highly suggestive that copy number of Tlr8 contributes to the sex difference in 564Ig Tlr7/9−/− mice.

Discussion

In previous studies, we found that in 564Igi mice IgG anti-RNA autoantibody production was partially dependent on TLR7 [9]. In addition, we reported that 564Igi mice showed increased granulopoiesis and increased IFN-I production, characteristics frequently seen in SLE patients [27]. To better understand the role of TLRs in murine lupus and their role in clinical lupus, we generated and analyzed mice deficient in the signaling adaptor MyD88, which is required for signaling through Tlr7, Tlr8 and Tlr9 sensing nucleic acid as well as for signaling via the members of the Il1r family of cytokine receptors. In addition, we generated mice simultaneously deficient for Tlr7 and Tlr8 as well as mice simultaneously deficient in Tlr7 and Tlr9 along with single mutant Tlr7, Tlr8 and Tlr9- deficient mice all on the 564Igi background for direct comparison with 564Igi mice.

MyD88-deficient 564Igi mice clearly showed no circulating autoantibody (Fig. 1A–C), no increased granulopoiesis (Fig. 4B) nor increased IFN-I expression (Fig. 5A). In addition to its role in TLR signaling, Myd88 is also required for signal transduction downstream of the Il1r family of cytokine receptors. However, 564Igi mice deficient in Il1r did not show any differences in autoantibody production (Supporting Information Fig. 2). The demonstration of the MyD88 dependency of autoantibody production, granulopoiesis and IFN-1 expression suggests a role of a TLR other than TLR7. The likely candidates in addition to TLR7 would be endosomal TLRs 8 and 9.

We show in the present study that TLR8 is crucial for the heightened granulopoiesis and IFN-I expression seen in the 564Igi mouse model of SLE. Thus, Tlr8-deficient mice produce as much IgG2a and IgG2b autoantibodies as 564Igi mice (Fig. 2D), however, they have decreased granulopoiesis (Fig. 4B) and lower IFN-I expression by neutrophils (Fig. 5A). Therefore, production of autoantibodies is not sufficient for granulopoiesis nor for increased IFN-I expression. On the other hand, deficiency in Tlr7 results in low levels of autoantibody (Fig. 1A–C) but strikingly high increases in the population of granulocytes (Fig. 4B) and the increased expression of IFN-I (Fig. 5A). Furthermore, double deficient Tlr7/Tlr9 564Igi female mice produced low levels of autoantibody (Fig. 2E) and strikingly increased granulopoiesis (Fig. 4B), IFN-I expression (Fig. 5A) and lupus pathology with glomerulonephritis (Fig. 5B and Tables 1 and 2).

Therefore, TLR8 can strongly activate granulocytes to expand and express IFN-I. These results are important and novel and are validated by the recent finding that a Tlr8 polymorphism is associated with SLE patients [31]. A previous report [15] also found that male Tlr7/Tlr9-deficient MRL/lpr mice did not produce anti-nuclear autoantibodies consistent with our findings. However, they concluded that TLR8 did not play a role in the MRL/lpr model of SLE. However, data on autoantibodies in females was not shown.

These results also establish that glomerulonephritis is associated with the combination of autoantibody production, granulopoiesis and increased type 1 interferon. Neither high levels of anti-RNA antibody nor increased granulopoiesis and interferon production were sufficient.

In analogy to the requirement of dual BCR/TLR signaling for B cell activation, one could postulate that a similar dual FcRγ/TLR is necessary for neutrophil activation resulting in granulopoiesis and production of IFN-I. Indeed, we found previously that neutrophils from 564Igi mice show upregulated IFN-I signature genes, upregulated Fcγ RIV and increased circulating immunocomplexes (ICs) [27]. Immune complexes containing anti-nuclear antigens activate cells through a combination of TLR and activating FcγR recognition [32] [33]. Neutrophils express activating FcγRs (I, III, IV) [34, 35]. Thus, activating FcγRs and TLR8 must combine to recognize ICs that stimulate granulocytes towards expansion and IFN-I production in 564Igi Tlr7−/− or Tlr7/9−/− mice. The relevance of FcγRs in SLE is validated by the finding that a polymorphism at the Fcγr2a region has been found associated with SLE in a cohort of females [36]

Although SLE is more frequent in females, there are few mouse models that recapitulate this aspect of disease. [37]. We found that female 564Igi Tlr7/9−/− mice have low autoantibody production, increased granulopoiesis, and severe disease, features that are not found in males. As such, this is the first mouse model of SLE demonstrating a clear genetic variation that reflects sexual differences, similar to what is observed in human disease [38] [1].

We propose that because Tlr8 is on the X chromosome and females have two potentially active copies of Tlr8, there is an increased incidence of SLE in women. Indeed, one copy of the Tlr8 gene is not sufficient in male 564igi Tlr7/9−/− or female 564Igi Tlr8^+/− Tlr7/9^−/− mice for autoantibody production (Fig. 6A and B). This shows that the disparity of disease between male and female mice may not be due to hormonal differences but to the difference in copy number of a functional Tlr8 gene. These results suggest that Tlr8 on the X-chromosome might not be inactivated. Experiments to determine whether Tlr8 can escape X-chromosome inactivation are now underway. Interestingly, in vitro activation of PBL from normal individuals with R848 a TLR agonist revealed a striking female bias in the expression of IFN-I [39]. The possibility that Tlr7 might escape X-inactivation was investigated and the anticipated silencing of one allele in females was confirmed. However, the model employed used the TLR7/8 agonist R848 to stimulate. Therefore, the sex bias observed may have been due to activation of TLR8 rather than TLR7.

Previously, we showed that autoantibody production in 564Igi mice was largely dependent on TLR7 [9]. 564Igi Tlr7−/− male and female mice do produce low levels of autoantibody (Fig. 1A–D). However, in male 564Igi Tlr7/9^−/− mice a single copy of Tlr8 is insufficient for autoantibody production. Therefore, TLR9 may be responsible for the low levels of autoantibody produced in 564Igi Tlr7−/− male mice. Activation of mature B lymphocytes and autoantibody production by TLR9 has been recently suggested [28]

The presence of only low levels of autoantibody in the sera of female 564Igi Tlr7/9−/− mice (Fig. 2) and the absence of such antibodies in the sera of 564Igi Tlr7/8−/−mice (Fig. 3) reveals that TLR8 can also drive the production of autoantibody. Even with this low level of autoantibody female 564Igi Tlr7/9−/− mice have high granulopoiesis and IFN-I production that leads to severe glomerulonephritis. Thus, we suggest that TLR8 can sense low levels of RNA in ICs (autoantibody/RNA) on granulocytes leading to granulopoiesis and IFN-I expression. Recently, a Tlr8 polymorphism was found that to be associated with SLE [31]. This suggests that TLR8 may be important for the development of SLE in humans as it is in mice. In our SLE model TLR7 plays an important role in autoantibody production by B cells, while TLR8 is responsible for high levels of granulopoiesis and IFN-I production. While our data suggest that Tlr 8 might play a role in human disease, which would have therapeutic implications, caution is warranted, as the one clinical study of Tlr8 expression in patients saw elevations of 7 and 9 but not 8 in peripheral blood leukocytes [40].

Finally, in this report, we show for the first time that the negative regulatory activity of TLR9 is seen in the suppression of granulopoiesis and IFN-I production as evidenced by the dramatic increase in granulopoiesis and IFN-I expression and pathology in Tlr9 deficient mice, or by low Tlr9 expression in Tlr7 deficient 564Igi mice.

An increase in the populations of monocytes and granulocytes was previously described in a chemically induced (pristane) lupus mouse model [41]. In this model neutrophils were increased in Tlr7 deficient and Tlr9 deficient mice, in agreement with our results, suggesting a role for Tlr8 in the increase of neutrophils. Interestingly, an increase in monocytes was shown to be dependent on TLR7 but not on FcγR [41].

In contrast to our study of 564Igi Tlr8−/− mice, an earlier study showed that in otherwise unmanipulated C57BL/6 mice loss of Tlr8 alone resulted in autoimmunity. This was attributed to an upregulation of Tlr7 in these mice, particularly in pDC [42]. We did not find autoantibodies in the sera of 3 months old C57B/6 Tlr8−/− mice, nor any increase in total IgG (Supporting Information Fig. 4). With 564Igi Tlr8−/− we did not observe upregulation of Tlr7 (Fig. 4C), however these mice produced amounts of IgG anti-RNA autoantibodies similar to that observed in 564Igi mice, indicating that Tlr8−/− is not crucial for autoantibody production (Fig. 2A–B and D). On the other handTlr8−/−seems crucial for granulopoiesis and Ifn-I expression (Fig. 4B and 5A). Since there is a strong correlation between high granulopoiesis and high expression of IFN-I with SLE-like pathology, we would expect that 564Igi Tlr8-deficient mice would not develop SLE-like symptoms. This contrasts to the earlier report [42]. A plausible explanation for the development of autoimmunity-like disease in C57BL/6 Tlr8−/− mice in some laboratories but not others would be environmental exposure to infection causing inflammation and autoantibody production in these mice[42]. Indeed, we found that some Tlr7/8 double deficient 564Igi mice older than 12 months succumb from infections (data not shown) and some of these have autoantibodies in the sera (data not shown). Thus, the induction of autoimmunity like disease in Tlr8 deficient mice might be due to activation by environmental pathogens and not by self-nuclear antigens. Our results with the 564Igi Tlr7/8−/− also raise concerns regarding the susceptibility of these compromised mice to infection.

Our results suggest that TLR7 and TLR8 antagonists would have potential therapeutic benefits in SLE patients. Blocking the function of TLR7 and TLR8 together has the potential to dampen inflammatory responses to RNA associated self-antigen and thus alleviate disease drivers such as IFN-I production, granulopoiesis and autoantibody production. These results also suggest that the blocking of TLR9, even in the context of blocking of TLR7 and 8, might exacerbate disease.

Materials and methods

Mice

All experiments with mice were performed in accordance to the regulations of and with the approval of theTufts/TMC IACUC. The creation of 564Igi mice was previously described [9]. 564Igi WT mice were bred in house to create homozygous, Tlr7 deficient [43], Tlr9 deficient [44], Tlr7/Tlr9 double deficient, Tlr7/Tlr8 double deficient [45], MyD88 deficient [46] and Tlr8 deficient [45]. C57BL/6 and BALB/c mice were purchased from Jackson laboratories. Aicda −/− (AID−/−) were obtained from Dr. Janet Stavnezer with permission from Dr. Tasuke Honjo [47] Experiments were performed with female and male offspring.

564 idiotype, Anti-RNA, and C1q immune complex binding ELISAs

ELISAs were run to determine the level of 564 IgG2a, IgG2b and IgM idiotype+ antibodies in the sera of various mouse lines. Detection of IgG1 on anti-idiotype coated wells was not performed because the anti-idiotype is of the IgG1 isotype. However, 564Igi IgG1 positive antibodies were not found in 564Igi mice on a RAG-deficient background. Therefore, 564Igi mice do not produce IgG1 564 Igi Id+ nor RNA binding antibodies, however, they do produce Id- IgG1 antibodies. The detailed protocol for this assay is described in reference [9]. The data were acquired with a Spectra Max 340 ELISA plate reader (Molecular Devices) at an optical density of 405 nm (OD₄₀₅). To determine the amount of immune complexes in the sera of mice, plates were coated with C1q protein (Complement Tech #A099) performed according to Dr. Rosane De Oliveira (UMASS Medical School). The anti RNA ELISA was carried out according to the protocol of Dr. Keith Elkon (University of Washington, Seattle, Washington). Nunc MaxiSorb flat botton (eBiosciences) ELISA plates were coated with 50mg/ml poly-L-Lysine (Sigma Aldrich) diluted in double distilled H₂0 overnight at 4 degrees Celsius. After washing the plates three times with 1xPBS/0.05% Tween 20, 10mg of yeast RNA (Ambion/Invitrogen) in PBS were used to coat the plates overnight at 4 degrees Celsius. ELISA plates were then blocked in 1xPBS/0.05%Tween/5% goat serum. The RNA ELISA was conducted as for the 564 idiotype ELISA. Detection of 564 IgG2a, IgG2b, IgM was previously described[9] Sera from 564Igi mice and the TLR deficient strains were typed for IgG2a isotype with an goat anti-IgG2a antibody used as a capture and detecting antibody when testing for total IgG2a serum antibodies. BALB/c sera was used as positive control in total IgG2a antibody content. Detection of RNA binding IgG2a was done using RNA as capture antigen and AP conjugated goat anti-IgG2a antibodies. In these assays, BALB/c sera is used as negative control. The goat anti-IgG2a antibodies from Southern Biotech reacts poorly with IgG2a from C57BL/6 mice. Detection of total IgG2b in the sera of any mouse strains was done by goat anti-IgG2b antibodies used as capture antibody and AP-conjugated goat anti-IgG2b as detecting reagent. In RNA binding assays RNA is used as detecting antigen and AP-conjugated goat anti-IgG2b is used as detecting reagent. C57BL/6 sera is used as positive control for total IgG2b, and negative control for RNA binding assays. IgG2a antibodies in the sera of C57BL/6 or of mice on C57BL/6 Tlr8−/− background were tested using goat anti-IgG2c antibodies

Flow Cytometry

Cells were stained for flow cytometry according to standard procedures [9]. Propidium iodide (PI) was added just prior to analysis on a FACScalibur flow cytometer (BD Biosciences). B6-256 anti-Id was generated as described [9] and coupled to Alexa 488 or Alexa 647 according to the manufacturer’s instructions (Invitrogen Molecular Probes). Granulocytes in the BM or spleen were stained with Alexa 488 labeled anti-CD11b, and PE labeled anti-Ly6-C or with anti-Ly6-G (Biolegend). CD11b_Ly-6Cin or CD11b _Ly-6G+ represents mainly neutrophils [30]. B cells were stained with Fluorescent Anti-IgM, anti- IgG2a, anti-B220 antibodies (Southern Biotech) and anti-Idiotype Alexa 647 B6-256. Other fluorochrome-conjugated antibodies were from BD Biosciences or Southern Biotech and were generally used at 1 μg/ml.

Antinuclear antibody (ANA) testing

Fixed human HEp-2 cells (Antibodies Inc.) were stained with mouse sera from three-month old mice according to the manufacturer’s instructions, except that the secondary antibody was Alexa 488 tagged Goat-anti-mouse IgG2 (Invitrogen). Slides were mounted with ProLong Gold anti-fade reagent (Invitrogen) and digitally photographed with a Nikon E400 fluorescence microscope at 100X magnification. Pixel Fluorescent intensity was measured by MetaXpress® software.

Cell purification

To purify neutrophils, bone marrow cells were treated with biotinylated anti-Ly-6G (Biolegend #127604) then separated by a magnet after conjugation to Dynabeads^® Biotin Binder (Invitrogen #110.47). The remaining cells were treated with biotinylated anti-B220 (Biolegend #103204) and B cells were extracted. Tryzol (Ambion/invitrogen) was added to purified cells and RNA purification performed subsequently.

Quantitative Real Time Taqman PCR

All qPCR experiments were carried out and analyzed using first-strand cDNA synthesis for 3 dilutions of purified RNA. Triplicates for each cDNA reaction were used for amplification with a predesigned mouse Ifn-β1, Ifn-α6, Tlr7, Tlr8, Tlr9- or β-actin FAM primer probe (Life Technologies) in a Biorad IQ5 quantitative PCR system. Relative quantification was determined after establishing standard curves for mRNA expression using serially diluted Raw cell cDNA. qPCR data represent 3–4 mice ± SD. Significance was determined by the Student t-test.

Western Blots

Purified neutrophils and B220+ cells were re-suspended in 5 volumes of lysis buffer. Cell lysates were made by slowly rocking the cells for 45 min at 4 degrees Celsius. After centrifugation at 4 degrees Celsius for 15 min, the supernatants were collected and boiled at 100 degrees Celsius prior to determination of protein concentration utilizing the 1ml of lysate in a Bradford assay. Lysates were combined with 1x Laemmli loading buffer, and run overnight at 46 volts on a 10% acrylamide gel. The protein was transferred to a nitrocellulose membrane (Thermo Scientific #Z613657- A83) using a semidry transfer apparatus for 3hrs at 150 volts. After transfer, the membrane was blocked with 5% milk in TBS (20mM Tris, 0.14M NaCl, 0.1% Tween 20 pH 7.6–8.0) for 1 hr, and then washed three times with TBST for 5 min. The membrane was then incubated for 1–2 hrs at room temperature with rabbit polyclonal anti-TLR7 (Abcam #ab45371) or anti-TLR8 antibody (Abcam #ab24185) and with mouse anti-β-Actin monoclonal antibody (Sigma #A3441). The membrane was washed 3 times for 5 min each wash in TBST. The membrane was incubated for 1hr with goat anti-mouse IgG-HRP (Thermo Scientific #31432) antibody diluted 1:5000 in TBS and anti goat secondary IgG-HRP (Santa Cruz Biotechnology) antibody diluted 1:5000 in TBS and then washed for 15 min 3 times in TBST. West Pico Chemiluminescent substrate (Thermo Scientific #34079) was placed on the membrane for 5 sec then the membrane was wrapped in plastic. Autoradiography film (Denville Scientific #E3018) was placed in a cassette with the membrane for less then 1min and developed on a Kodak M35A X-OMAT processor.

Immunofluorescence, and Light Microscopy Analysis of Kidney Sections

Fresh kidney samples for immunofluorescence studies were frozen in OCT and 4 μm sections cut on a cryostat and mounted on glass slides. The sections were air-dried for 1 hr, dehydrated in PBS, and incubated with Alexa 488 anti-idiotype antibody solution for 1 hr at room temperature. The sections were rinsed in PBS, mounted in Aquamount, and examined and photographed with a Zeiss fluorescence microscope.

For light microscopy, fresh kidneys were fixed in 10% buffered formalin, and embedded in paraffin. 5 μm paraffin sections were stained with periodic acid-Schiff (PAS) and evaluated by light microscopy in a blind manner as previously described [20].

To score the PAS sections for morphologic evidence of lupus, the clinical scale of the international Society of Nephrology/Renal pathology Society (ISN/RPS-2003) was used. Classification of Lupus Nephritis was used (1 through 6). Class 1 indicated minimal mesangial lupus, with normal glomeruli as seen with light microscopy but with mesangial immune deposits revealed by Immunofluorescence. Class 2 is indicated by mesangial proliferative lupus nephritis; class 3 is focal lupus nephritis, class 4 is diffuse lupus nephritis, class 5 is defined as membranous lupus nephritis and class 6 indicated advanced sclerotic nephritis where greater than 90% of the glomeruli are globally sclerosed [48]. IC deposits were evaluated by immunofluorescence and scaled from 1 to 3, 3 being the highest degree of deposits. At least 10 glomeruli were examined for negative samples [48].

Statistical methods

GraphPad Prism v5 was used for all two-tailed Student’s t tests.