Report

Autoreactive B Cell Responses to RNA-Related Antigens Due to TLR7 Gene Duplication

See allHide authors and affiliations

Science  16 Jun 2006:
Vol. 312, Issue 5780, pp. 1669-1672
DOI: 10.1126/science.1124978

Abstract

Antibodies against nuclear self-antigens are characteristic of systemic autoimmunity, although mechanisms promoting their generation and selection are unclear. Here, we report that B cells containing the Y-linked autoimmune accelerator (Yaa) locus are intrinsically biased toward nucleolar antigens because of increased expression of TLR7, a single-stranded RNA-binding innate immune receptor. The TLR7 gene is duplicated in Yaa mice because of a 4-Megabase expansion of the pseudoautosomal region. These results reveal high divergence in mouse Y chromosomes and represent a good example of gene copy number qualitatively altering a polygenic disease manifestation.

The analysis of genetic modifiers of autoimmune disease provides insight into mechanisms underlying disease susceptibility and possible therapeutic approaches. The Yaa genetic modifier, for example, increases severity of systemic lupus erythematosus (SLE) in male mice (1). Yaa was first identified from a cross between a C57BL/6 (B6) female and a SB/Le male that produced the BXSB hybrid line. BXSB mice develop SLE with much higher incidence and with earlier onset in males compared with females, whereas mice from the reciprocal cross (SB/Le female and B6 male) do not show the same acceleration of disease in males. Thus, this autoimmune-enhancing effect in BXSB males was attributed to the Y chromosome derived from the SB/Le strain and was named Yaa, which stands for Y-linked autoimmune accelerator (2). In addition to the BXSB strain, the Yaa Y chromosome exacerbates disease in a number of lupus-prone strains, but it has not been found to induce autoimmune disease in wild-type B6 male mice (35).

We have analyzed the effect of the Yaa genetic modifier in mice deficient in the antibody-binding inhibitory receptor FcγRIIB. With the B6 background, FcγRIIB–/– mice develop spontaneous SLE-like disease, characterized by the presence of autoantibodies against chromatin (antichromatin) and development of lethal glomerulonephritis (6). Addition of the Yaa modifier to the FcγRIIB-deficient model by genetic crossing not only aggravates the autoimmune pathology, but it also leads to a switch of specificity from chromatin to nucleolar autoantibodies (7). Distinct autoantibody specificities against nuclear antigens are often used in the diagnosis and prognosis of systemic autoimmune disease (8). Thus, in humans, chromatin antibodies are considered characteristic of SLE, whereas the presence of nucleolar antibodies often corresponds with systemic sclerosis (9). In this study, we aimed to identify the Yaa gene product and to understand how Yaa modifies autoantibody specificity toward nucleolar antigens.

To determine which cell type instructs the Yaa nucleolar antibody specificity, we generated bone marrow chimeras with an equal mix of FcγRIIB–/– Yaa and FcγRIIB–/– cells. For this experiment, the FcγRIIB mutation was bred into a congenic B6.IgH a strain so that antibodies produced by FcγRIIB–/– B cells (a allotype) could be distinguished from the antibodies derived from FcγRIIB–/– Yaa B cells (b allotype) by using allotype-specific immunofluorescence staining. RAG–/– mice reconstituted with the mixed bone marrow contained equivalent numbers of Yaa [immunoglobulin Mb+ (IgMb+)] and non-Yaa (IgMa+) B cells in peripheral blood 1 month after the transfer (Fig. 1A) and in spleen 4 months after the transfer (table S1). Equal amounts of spontaneous activation of both IgMa+ and IgMb+ B cells were detected by the expression of the activation marker CD69 (Fig. 1, B and C, and table S1). When tested for antigen specificity by using a standard assay for nuclear antibodies (ANA) on Hep-2 cells (10), serum from the mixed chimeras yielded a homogeneous nuclear pattern typical of antichromatin specificities when detecting a-allotype immunoglobulins G (IgGs) (Fig. 1D). In contrast, detection of b-allotype IgGs gave rise to a speckled nucleolar pattern, characteristic of RNA-related specificities (Fig. 1E). This differential pattern was not apparent when detecting a and b allotype–specific IgGs in the control experiment, for which an equal mix of non-Yaa FcγRIIB–/– IgH a and FcγRIIB–/– IgH b was used to reconstitute irradiated RAG–/– mice (Fig. 1, F and G). This suggests that Yaa B cells have an intrinsic bias toward the production of antibodies with nucleolar specificity in the context of a general loss of tolerance promoted by genetic factors such as the FcγRIIB–/– background. These results validate and further expand the notion, previously suggested by Izui's group, that Yaa can activate autoreactive B cells depending on the nature of autoantigens (11, 12).

Fig. 1.

B cell intrinsic ANA specificity in FcγRIIB–/– Yaa mice. (A to C) Flow cytometric analysis of B cells from the blood of RAG–/– mice 1 month after reconstitution with FcγRIIB–/– (IgH a) and FcγRIIB–/– Yaa (IgH b) bone marrow cells. (D to G) ANA production in serum from the mixed bone marrow chimera 4 to 6 months after reconstitution (n = 6).

Besides the intrinsic Yaa effect on antibody specificity, we also observed a B cell extrinsic effect of Yaa in terms of the extent of B cell activation and autoantibody production. Thus, the presence of the Yaa bone marrow in the mixed chimera resulted in increased production of IgG of the a allotype (non-Yaa) and elevated numbers of activated cells as compared with the control chimeras where no Yaa bone marrow was transferred (table S1).

Hyperactivity of B cells in Yaa has been previously suggested by the observation that Yaa B cells show increased spontaneous IgM secretion and lack marginal zone B cells (MZBs) (13). Such an inverse relationship between the strength of B cell receptor signaling and MZB development has been suggested previously by studies of B cells deficient in Bruton's tyrosine kinase, Btk, which are generally hyporesponsive to antigen stimulation and show increased representation of MZBs (14, 15). We therefore reasoned that the MZB defect in B6.Yaa mice might be reversed by mutations that reduce overall B cell responsiveness. To test this, we crossed B6.Yaa mice with Btk-deficient mice to generate B6.Yaa Btk mice and control B6.Yaa Btk+ littermates. We observed that, whereas B6.Yaa Btk+ mice lack MZBs in the spleen, the B6.Yaa Btk double mutant phenotypically resembles B6.Btk mice with restoration of MZBs (Fig. 2, A and B). This revealed a genetic and functional interaction between the Yaa and Btk genes. Moreover, addition of the Btk mutation to FcγRIIB–/– Yaa eliminated the autoimmune phenotype: It resulted in reduced spontaneous B cell activation (Fig. 2C), normal spleen weight (Fig. 2D), and eliminated autoantibody production (Fig. 2E) as well as tissue damage in both kidney and lung (Fig. 2F). Thus, the Btk signaling pathway is required for the loss of tolerance to nucleolar antigens in FcγRIIB–/– Yaa mice, correlating with what was shown in Btk-deficient human patients (16).

Fig. 2.

The Yaa phenotype is Btk-dependent. (A and B) Btk deficiency restores the MZB population in B6.Yaa spleens. (A) Flow cytometric analysis of B220+ splenocytes shows percentage of MZBs. (B) Immunofluorescence staining of spleen sections using antibodies to Moma-1 and B220 to identify marginal zone macrophages and B cells, respectively. (C to F) Btk deficiency prevents autoimmune disease in FcγRIIB–/– Yaa mice. (C) Flow cytometric analysis shows percentage of activated B cells in the spleen. (D) Btk deletion alleviates splenomegaly in FcγRIIB–/– Yaa mice. ** P < 0.005. (E) Titers of serum ANAs in 6-month-old mice. *** P < 0.001. Error bars indicate ± SD. (F) Hematoxylin and eosin staining of kidney and lung sections showing membranoproliferative glomerulonephritis and lung infiltration in FcγRIIB–/– Yaa but not in FcγRIIB–/– Yaa Btk mice (n ≥ 3).

In attempt to identify candidate genes for the Yaa locus, we compared gene expression in B6 versus B6.Yaa by microarray analysis on follicular B cell RNA. Only 26 genes were substantially up-regulated (>1.5-fold) in B6.Yaa B cells compared with B6 B cells in three out of four independent experiments (table S2). Of the genes that showed a twofold increase in expression in Yaa B cells, four (Msl31, TLR7, Tmsb4x, and Rab9) are located in the most distal part of chromosome X, with sequences that are consecutive in the genome. This suggested to us that the genomic region of chromosome X containing these four genes might be duplicated in the Yaa genome. To validate this hypothesis, we performed quantitative polymerase chain reaction (PCR) on genomic DNA from B6 and B6.Yaa. Probes detecting genes Msl31, TLR7, TLR8, Tmsb4x, and Rab9 showed double the quantity of DNA in B6.Yaa male and B6 female cells as compared with wild-type B6 males (table S3). Fluorescence in situ hybridization (FISH) using bacterial artificial chromosome (BAC) probes that span a 4-Mb region at the distal end of the X chromosome (Fig. 3, A to C, and fig. S1) revealed hybridization in the B6. Yaa Y chromosome, confirming the genomic duplication and determining its location in the genome. One of these probes (154O12) expands the region that had been reported to include the Mus musculus pseudoautosomal region (PAR) (17). Indeed, this probe hybridized with the X and Y chromosomes in both B6.Yaa and B6 cells, although the Y chromosome signal in B6 was notably weaker in intensity compared with both the X chromosome and the B6.Yaa Y chromosome (Fig. 3C). Six other probes located in this distal region hybridized with the chromosome X and Y in B6.Yaa cells but only with the chromosome X in B6 cells. Probe 239H22 showed lower amounts of hybridization on the Yaa Y chromosome compared with the X chromosome, indicating that the duplicated region begins somewhere within the sequence represented by this BAC clone. Overall this analysis uncovers a 4-Mb expansion of the PAR in Yaa males that results in the duplication of at least 13 known and four unknown genes that in B6 wild-type mice are X chromosome–specific. Duplication of the TLR7 gene was also confirmed in the BXSB and SB strains from which the Yaa Y chromosome was derived (table S3).

Fig. 3.

Expansion of the pseudoautosomal region in the Yaa genome. (A and B) FISH analysis in B6 and B6.Yaa embryonic fibroblasts. BAC clone RP23-74J22 (X centromeric–specific, pink) was used with BAC clones listed in (C). (A) Whole chromosomal spread in B6.Yaa and B6 cells stained with 139P21 and 125I7 probes. (B) Magnified view of X and Y chromosomes using the indicated fluorescently labeled BAC probes. For purposes of visualization, the Y chromosome signal intensity for clone 239H22 was enhanced. (C) Diagram shows genomic location on the X chromosome of BAC DNAs used for hybridization. Hybridization signals are ++ for strong; +, weak; and –, negative.

Among the genes included in the genomic duplication of Yaa, TLR7 was of most interest to us because it shares several characteristics predicted for the Yaa locus: It is expressed in B cells (18), it binds Btk, and its ligand is single-stranded RNA (19, 20), which is also a likely component of nucleolar antigens. Quantitative PCR on cDNA and Western blot analysis confirmed that the genomic duplication in Yaa resulted in increased TLR7 mRNA and protein in B cells (Fig. 4A). Functionally, stimulation of Yaa splenocytes with the TLR7 agonist imiquimod resulted in enhanced phosphorylation of the transcriptional regulator IκBα and increased proliferation compared with wild-type responses (Fig. 4, B and C). Naïve follicular Yaa B cells showed a small, but significant, increase in proliferation upon imiquimod stimulation compared with wild-type B6 cells (Fig. 4D). This increased proliferation in response to imiquimod in B cells and total splenocytes was dependent on Btk, again confirming the Btk requirement for the Yaa phenotype revealed in the genetic analysis. In vivo administration of imiquimod was also sufficient to alter the homogenous nuclear pattern seen in sera from FcgRIIB–/– mice to a nucleolar pattern (Fig. 4, E and F), demonstrating that enhancing TLR7 signaling induces nucleolar-specific responses. It is nevertheless possible that genes other than TLR7 included in the PAR expansion promote the acceleration of autoimmune disease observed in Yaa males.

Fig. 4.

Increased expression and function of TLR7 in Yaa mice. (A) TLR7 mRNA and protein levels measured by quantitative PCR on cDNA and Western blot in CD43CD9 B cells. (B) B6 or B6.Yaa splenocytes were stimulated with imiquimod for the indicated times and IκBα and pIκBα measured by Western blot. ZAP-70 is a loading control. Proliferative responses of splenocytes (C) and follicular B cells (D) upon stimulation with imiquimod (1 μg/ml), CpGs (0.1 μg/ml for splenocytes and 1μg/ml for follicular B cells), and control oligodeoxynucleotides (ODNs). Black bar, B6; white bar, B6. Yaa; gray bar, B6. Btk. ***P < 0.001, ** P < 0.005, * P < 0.05. Error bars indicate ± SD. (E and F) ANA test of serum from B6.FcγRIIB–/– mice before or after treatment with imiquimod. Data are representative of at least three independent experiments (n ≥ 3).

Just as TLR9 has been suggested to promote B cell activation by DNA-containing self-antigens (21), our data indicate that TLR7 may contribute to autoimmunity by inducing activation of B cells by RNA-containing antigens of nucleolar origin. In vitro B cell activation by RNA-associated autoantigens has been reported (22, 23), but the best confirmation of the model would come from the study of the effect of this pathway on the development of spontaneous autoimmune disease. The results presented here provide strong evidence that naturally occurring differences in expression of the TLR7 gene as well as environmental factors that induce TLR7 responses may result in increased B cell sensitivity to RNA-containing self-antigens. Thus, the Yaa locus represents an excellent example in which qualitative phenotypic differences in disease pathology are derived from a copy number polymorphism, a genetic event that has been shown to be common both in mice and humans (24).

Supporting Online Material

www.sciencemag.org/cgi/content/full/1124978/DC1

Materials and Methods

Fig. S1

Table S1 to S3

References

References and Notes

View Abstract

Navigate This Article