Cell intrinsic expression of TLR9 in autoreactive B cells constrains BCR/TLR7-dependent responses

Abstract

Endosomal Toll-like receptors (TLRs) play an important role in systemic autoimmune diseases such as SLE, where DNA- and RNA-associated autoantigens activate autoreactive B cells through TLR9- and TLR7-dependent pathways. Nevertheless, TLR9-deficient autoimmune prone mice develop more severe clinical disease, while TLR7-deficient and TLR7/9-double deficient autoimmune-prone mice develop less severe disease. To determine whether the regulatory activity of TLR9 is B cell intrinsic, we have now directly compared the functional properties of autoantigen activated WT, TLR9-deficient and TLR7-deficient B cells, in an experimental system where proliferation depends on BCR/TLR co-engagement. In vitro, TLR9-deficient cells are less dependent on survival factors for a sustained proliferative response than either WT or TLR7-deficient cells. The TLR9-deficient cells also preferentially differentiate toward the plasma cell lineage, as indicated by expression of CD138, sustained expression of IRF4, and other molecular markers of plasma cells. In vivo, autoantigen-activated TLR9-deficient cells give rise to greater numbers of autoantibody producing cells. Our results identify distinct roles for TLR7 and TLR9 in the differentiation of autoreactive B cells that explain the capacity of TLR9 to limit, and TLR7 to promote, the clinical features of SLE.

Introduction

Many of the autoantigens targeted during systemic autoimmune diseases act as autoadjuvants by associating with macromolecular complexes that stimulate innate immune receptors. In B cells, nucleic acid-associated autoantigens need to be bound by the BCR and transported to a TLR-associated compartment where TLR detection of DNA or RNA provides a second signal that promotes B cell activation. This paradigm, whereby BCR-delivered TLR agonists promote autoreactive B cell activation, initially emerged from in vitro studies (1) and has been supported by numerous in vivo observations. Thus, TLR7-deficient autoimmune prone mice fail to make autoantibodies reactive with RNA-associated autoantigens, and TLR9-deficient autoimmune prone mice fail to make autoantibodies reactive with dsDNA or chromatin (2). Moreover, autoimmune prone mice lacking only TLR7 have markedly attenuated disease (2), while overexpression of TLR7 results in exacerbated clinical symptoms and accelerated mortality (3, 4). However, quite paradoxically, autoimmune prone mice that lack functional TLR9 invariably develop more severe clinical disease and have shortened lifespans (5–9).

Remarkably little is known about the differential outcomes of TLR7 versus TLR9 engagement, or how TLR9, but not TLR7, mitigates systemic autoimmunity. In mice, both TLR7 and TLR9 are expressed by B cells, dendritic cells (DCs), macrophages, and even neutrophils, and therefore any of these cell types could negatively regulate disease onset through a TLR9-dependent mechanism. However, the growing appreciation that B cells play a pivotal role in the etiology of systemic autoimmune diseases (10, 11), led us to monitor the direct effects of BCR/TLR7 and BCR/TLR9 co-engagement on B cell differentiation. We utilized BALB/c mice expressing an IgG2a-specific site-directed transgene encoded receptor, AM14, derived from an approximately 6-months old Fas-deficient MRL/lpr mouse (12–14). These rheumatoid factor (RF) B cells bind IgG2a with sufficiently low affinity that they survive tolerance checkpoints and persist in BALB/c mice as resting naïve follicular (FO) B cells, even in the presence of (monomeric) serum IgG2a (15). In fact, only IgG2a immune complexes (IC) which incorporate endogenous nucleic acids, capable of engaging either TLR7 or TLR9, can induce these RF B cells to proliferate in vitro (16). RF B cell responses to DNA-associated ICs are TLR9-dependent and inhibited by the addition of DNase I to the culture medium, while responses to RNA-associated ICs are TLR7 dependent and inhibited by the addition of RNase to the culture medium (1, 17). Stimulatory ICs include defined ligands, such as IgG2a-bound CG-rich dsDNA fragments (16, 18), as well as IgG2a autoantibodies that bind cell debris or surface bound autoantigens, present in the primary B cell cultures (1, 17).

The availability of autoantibodies reactive with DNA and/or RNA-associated autoantigens, together with TLR-deficient RF B cells, make it possible to directly compare the downstream effects of BCR/TLR7 and BCR/TLR9 engagement. We found that in vitro activation of RF B cells, through a mechanism dependent on the BCR and TLR7, promotes the extended survival of RF B cells and their differentiation into CD138⁺ plasmablasts. BCR/TLR7 and BCR/TLR9 activation pathways also have distinct functional outcomes in vivo, where again RF B cells activated through the BCR/TLR7 pathway, and not the BCR/TLR9 pathway, preferentially differentiate into antibody producing cells.

Materials and Methods

Mice

AM14, AM14 Tlr9^−/−, AM14 Tlr7^−/− mice have been described previously (13, 15, 19, 20). FcγR2b-deficient BALB/c mice and CD45.1 BALB/c mice were obtained from Jackson Lab. AM14 Tlr9^−/− and AM14 Tlr7^−/− mice were intercrossed to generate AM14 TLR7/9 double KO mice (Tlr7^−/−Tlr9^−/−). All mice were bred and maintained at the Department of Animal Medicine of the University of Massachusetts Medical School in accordance with the regulations of the American Association for the Accreditation of Laboratory Animal Care.

Cell culture

Splenic B cells were positively selected and then cultured in RPMI/5% heat-inactivated FCS as described previously (15) with the following ligands: 1 μg/ml CpG 1826 (s-oligodeoxynucleotide, kindly provided by Idera Pharmaceuticals), 0.1–1.0 μg/ml CL097 (InvivoGen), 15 μg/ml goat anti-mouse IgM F(ab′)₂ (Jackson ImmunoResearch), or the mAbs PL2-3 (1 μg/ml), PA4 (0.3 μg/ml), and BWR4 (10 μg/ml) (21–23). The ligands recognized by the monoclonal antibodies are derived from cell debris generated in culture and therefore the monoclonal autoAbs spontaneously form ICs. Better defined ICs were formed by combining a biotinylated-CG-rich ds DNA fragment (18) with streptavidin (SA) and an IgG2a anti-SA mAb, and used at a final concentration of 0.5 μg/ml DNA, 0.13 μg/ml SA and 0.5 μg/ml anti-SA mAb. In certain experiments, the BWR4 cultures were supplemented with IFN-β (300 U/ml, PBL). B cell proliferation was assessed by ³H-thymidine incorporation at the times indicated or by fluorescent dye dilution at 72 hr. BLyS, provided by Human Genome Sciences, was added to selected experimental groups maintained for 72 hr at a final concentration of 50 ng/ml. The TLR9 inhibitory ODN (3′ CCT GGA TGG GAA CTT ACC GCT GCA 5′) has been described (24).

Flow Cytometry

B cell subsets were identified with CD22.2-FITC, CD138-PE, CD45.2-PE, CD45.1-APC (BD Biosciences), CD45R/B220-eFluor450, CD44-eFluor780 (eBioscience). RF B cells were detected with biotinylated-4G7 in combination with streptavidin-PerCP-Cy5.5. IRF-4 was detected using an IRF-4 antibody (clone M-17, Santa Cruz) and anti-goat IgG Alexa Fluor 647 (Jackson ImmunoResearch). IRF4-PE and IRF8-PercP-efluor710 (eBioscience) were used to co-stain IRF8 and IRF4. B cell proliferation was assessed by CFSE dilution (Life Technologies) (15) or VPD450 dilution (BD). Dead cells were distinguished with TO-PRO-3 (Life Technologies). To analyze TLR7 expression levels, unstimulated purified B cells, or B cells stimulated for 24 hr, were fixed and permeabilized using the Foxp3 Fix/Perm kit (ebioscience). TLR7 protein was detected using a biotinylated mouse TLR7 specific mAb, A94 (25), in combination with SA-PE. Flow cytometric analysis was carried out using a BD LSR II with Diva Software (BD) and analysis was conducted with FlowJo software (Tree Star).

Gene expression

Total RNA was extracted using the RNeasy Minikit (Qiagen). Reverse transcribed DNA (Quanta) was analyzed by qPCR using Taqman probes for bcl-6, pax5 and prdm1 (Life Technologies). Samples were normalized to GAPDH and analyzed using the ΔΔCT method. For microarrays, RNA was prepared by the Trizol method (Invitrogen), purified using RNeasy Mini columns (Qiagen), and used on Agilent mouse 6x80K arrays with a control pool of B cell RNAs from all genotypes (unstimulated) which served as a reference (Cy3) for each genotype’s sample over a time course of stimulation with PL2-3 (Cy5). The Cy5/Cy3 ratio of gene expression was captured and normalized to the ratio values of the WT at 0 hr (unstimulated) array. Gene expression data has been deposited under accession GSE58756 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE58756).

ELIspot assay

Antibody forming cells were measured by ELISpot assay. Filter plates (Millipore) were coated with antibodies specific for mouse IgG2a (SouthernBiotech) and IgG1 (Jackson Immunoresearch). Bound antibody was detected with biotinylated anti-clonotype 4G7 (for IgG1) or 4-44 (for IgG2a), and streptavidin-AP (BD Bioscience) (12). Spots were developed with BCIP/NBT substrate (Sigma-Aldrich) and counted using an ImmunoSpot reader (C.T.L.).

In vivo activation of AM14 B cells

B220 purified splenic B cells (15 x 10⁶) were injected i.v. into CD45.1 BALB/c mice on day 0 together with 50 μg of PL2-3. Mice received additional i.v. injections of PL2-3 on day 3 or on days 3, 7, and 10. Spleens were harvested on day 6 or 13. To assess proliferation, purified B cells were labeled with 3.5μM VPD450 (BD) for 5 min prior to injection.

Statistical analyses

Statistical analyses were conducted with the Graphpad Prism6 software. Comparisons between two groups were performed by Students t-test for normally distributed data. Two-way ANOVA including Bonferroni post-test or Tukey’s multiple comparisons test was used for multiple group comparisons. A P value <0.05 was considered significant. P values are denoted as follows: * P ≤ 0.05, ** P ≤ 0.005, *** P ≤ 0.0005.

Results

Monoclonal autoantibodies can activate RF B cells through TLR7- or TLR9-dependent pathways

The monoclonal autoAbs PL2-3, PA4, and BWR4 have been reported to recognize chromatin, DNA, and RNA, respectively (21–23) and activate RF B cells in vitro through TLR-dependent mechanisms. Wild-type (WT), TLR7-deficient (Tlr7^−/−), TLR9-deficient (Tlr9^−/−) and TLR7/9 double-deficient (Tlr7^−/−Tlr9^−/−) RF B cells were stimulated with the monoclonal autoantibodies and the responses were compared to small molecule ligands for TLR9 (CpG ODN 1826) and TLR7 (CL097). Proliferation was determined by ³H-thymidine incorporation (Fig. 1A, left panel). The PA4 response was entirely TLR9-dependent as only TLR9-sufficient cells could respond. Tlr7^−/− and Tlr7^−/−Tlr9^−/− B cells mounted comparably low responses to BWR4 compared to WT B cells, indicating a critical role for TLR7, but not TLR9, in this response. By contrast, both the Tlr7^−/− and the Tlr9^−/− cells responded to PL2-3 significantly better than the Tlr7^−/−Tlr9^−/− cells; the relatively modest response of the Tlr9^−/− population was further increased at day 2 (Fig. 1A, right panel). Therefore the PA4 response is TLR9-dependent, the BWR4 response is TLR7-dependent, and the PL2-3 response can be driven by both TLR9 and TLR7. These data suggest that PL2-3 binds autoantigen-associated complexes that incorporate both DNA and RNA.

Fig. 1.

Monoclonal autoantibodies activate RF B cells through either BCR/TLR9-, BCR/TLR7- or both BCR/TLR7- and BCR/TLR9-dependent pathways. (A) Splenic B cells from RF WT (black bar), Tlr7^−/−(grey bar), Tlr9^−/− (white bar) and Tlr7^−/−Tlr9^−/− (hatched bar) mice were activated with the indicated ligands or monoclonal autoantibodies for 30 h (left) or 40h (right) and proliferation was measured by ³H-thymidine uptake. Data represent the mean +/− SEM of 3 independent experiments. (B) RF WT B cells were stimulated with BWR4 in the presence (open squares) or absence (filled circles) of IFN-β for the indicated times. (C) RF WT (open symbols) and RF Tlr9^−/− (closed symbols) B cells were stimulated with PA4 (left) or PL2-3 (right) for the indicated times. For (B) and (C), data is representative of 3 independent experiments. Proliferation was assessed by ³H-thymidine incorporation over the last 6 h of time point. Statistical analysis was performed using unpaired Student’s t test (* P ≤ 0.5, ** P ≤ 0.05, *** P ≤ 0.005).

We have previously shown that type I IFN markedly enhanced the initial response of RF B cells to BWR4, as detected 30 h after the addition of ligand (17). Based on the somewhat delayed PL2-3 response of the Tlr9^−/− cells, we decided to also monitor the PA4 and BWR4 responses over a more extended time period. In the absence of type I IFN, the BWR4-stimulated WT B cells responded well at later time points (Fig. 1B). Notably, the TLR7-driven component of the PL2-3 response (Tlr9^−/− cells) recapitulated the BWR4 kinetics, while the TLR9-driven PL2-3 and PA4 responses peaked and declined at an earlier time point (Fig. 1C). Together these data indicate that BCR/TLR7 co-engagement promotes a slightly delayed but more sustained response than BCR/TLR9 co-engagement. The delay may be at least partially due to a TLR7-dependent induction of type I IFN and subsequent upregulation of TLR7 expression (19). Further studies were performed without the addition of type I IFNs.

RNA-associated ICs promote the prolonged survival of RF B cells

To more precisely monitor both proliferation and death, RF WT B cells were labeled with CFSE and stimulated with the same set of ligands for 72 hr. Cell division was assessed by CFSE dilution and dead cells were identified with the cell-permeable DNA stain TO-PRO-3. Both PA4 and PL2-3 BCR/TLR9 co-engagement of WT B cells induced several rounds of division followed by a synchronous post-proliferative cell death (Fig. 2A, upper panel). In both cases, cells could be rescued by the addition of the B cell survival factor BLyS (Fig. 2A, lower panel). This extent of cell death was not observed in cells stimulated with either the TLR9 ligand CpG 1826 or the TLR7 ligand CL097 (Fig. S1A), indicating that co-engagement of the BCR and TLR9 resulted in a functional phenotype distinct from that elicited by TLR9 alone. Importantly, under the same conditions, BWR4-activated cells divided up to 3 times and remained viable, even in the absence of BLyS (Fig. 2A).

Fig. 2.

RNA-containing ICs induce BLyS-independent survival. (A) CFSE-labeled RF WT B cells were stimulated with the indicated ligands in the absence (top) or presence (bottom) of 50 ng/ml BLyS for 72 h. Proliferation was measure by CFSE dilution and cell death by uptake of TO-PRO-3. The quadrants depict the following: top left - dead divided cells (TOP-RO-3⁺ CFSE diluted); top right – dead undivided (TOP-RO-3⁺ CFSE undiluted); bottom right- live undivided (TO-PRO-3⁻ CFSE undiluted); bottom left – live divided (TO-PRO-3⁻ CFSE diluted). Division numbers are indicated by red arrows underneath the flow plots. Flow plots are representative of >10 experiments. (B, C) B cells from RF WT, Tlr7^−/−, Tlr9^−/− and Tlr7^−/−Tlr9^−/− mice were stimulated as in (A). (B) Representative flow plots of B cells from RF WT, Tlr7^−/−, Tlr9^−/− and Tlr7^−/−Tlr9^−/− mice stimulated with PL2-3. (C) B cells from RF WT, Tlr7^−/−, and Tlr9^−/− mice were stimulated as in (A) with PA4, BWR4, and PL2-3 in the presence or absence of BLyS. Data shown represents % of the recovered cells that had divided and remained alive +/− SEM, n=9. Statistical analysis was performed using 2-way ANOVA and Bonferroni post-test (* P ≤ 0.05, ** P ≤ 0.005, *** P ≤ 0.0005). (D) RF WT B cells were stimulated with PL2-3 in the presence or absence of BLyS, with or without a TLR9 inhibitor for 72 h. (E) RF WT B cells were stimulated with IgG2a IC that incorporated a defined CG-rich DNA-fragment for 72 h. (D,E) Representative flow plots of 3 independent experiments.

The same CFSE/TOPRO criteria were also used to monitor the 72 h response of Tlr7^−/− and Tlr9^−/− B cells. This analysis again showed that the PA4 response was entirely dependent on TLR9 for proliferation and on BLyS for survival, while the BWR4 response was TLR7-dependent and BLyS independent (Fig. S1B). By contrast, both the Tlr7^−/− and Tlr9^−/−, but not the Tlr7^−/−Tlr9^−/− RF cells proliferated in response to PL2-3 (Fig. 2B). Moreover, in the absence of BLyS, the Tlr9^−/− cells responded better to PL2-3 than either the WT or the Tlr7^−/− cells (Fig. 2B, C). The effect of TLR9-deficiency could be recapitulated with a TLR9-specific inhibitor (Fig. S1C); WT cells stimulated by PL2-3 in the presence of the inhibitor were less dependent on BLyS for survival (Fig. 2D). We also evaluated the response to defined ICs that incorporated a biotinylated CG-rich dsDNA fragment (18), bound by streptavidin, and delivered to the RF BCR via an IgG2a anti-streptavidin antibody. The response elicited by such an anti-SA/SA/Bio-DNA IC was entirely TLR9-dependent (Fig. S1D) and recapitulated the postproliferative cell death/BLyS rescue observed for PA4 and PL2-3 activated RF B cells (Fig. 2E). Thus, TLR9 mediates a dominant proliferation/post-proliferative cell death response and when the TLR9 component is eliminated, either genetically or by the use of an inhibitor, the cells that respond to PL2-3 through TLR7 maintain a more sustained response.

RNA-associated ICs drive B cell differentiation to CD138+ plasmablasts

To determine whether BCR/TLR9 and BCR/TLR7 engagement drive comparable programs of differentiation under conditions where both populations survive and divide, WT RF B cells were stimulated for 3 days with CpG 1826, CL097, PA4, BWR4 or PL2-3, in the presence of BLyS, and then analyzed by flow cytometry for markers of B cell differentiation. Plasma cells (PC) are defined by the surface expression of CD22^lo CD44^hi and CD138⁺ (26). Remarkably, only the BWR4 stimulated cells acquired the CD138 marker, indicative of differentiation toward the PC lineage (Fig. 3A). In contrast to these RNA ICs, cells stimulated with CL097 did not become CD138⁺ cells, again demonstrating the difference between ICs that bind both the BCR and TLR7 compared to small molecules that only engage TLR7.

Fig. 3.

BCR/TLR7 coengagement promotes plasma cell differentiation. (A) RF WT B cells were stimulated with the indicated ligands or monoclonal autoantibodies and stained for CD22, CD44 and CD138 to determine the frequency of CD22^lo CD44^hi CD138⁺ plasmablasts. n=4. (B) RF WT, Tlr7^−/−, Tlr9^−/− and Tlr7^−/−Tlr9^−/− B cells were activated with PL2-3 for 72 h and stained as in (A), n=4. (C) RF WT B cells were activated with PL2-3 in the presence of a TLR9 specific inhibitor for 72 h and stained as in (A), n=3. (D) Purified RF FcγR2b^−/− B cells were activated with the indicated ICs for 96 h in the presence of 50 ng/ml BLyS. mRNA levels for bcl-6, Pax5 and prdm1 were determined by qPCR. Data is normalized to media control using the ΔΔCT method and analyzed by Student’s t test; * P value of ≤ 0.5, n=4. (E,F) Expression of B cell differentiation-associated genes in RF WT, Tlr7^−/−, and Tlr9^−/− cells after stimulation with PL2-3. Gene expression changes are reported as ratios relative to RF WT 0 h (unstimulated) as measured by gene expression arrays.

While these data point to a distinct functional outcome of BCR/TLR7 compared to BCR/TLR9 activated cells, under the conditions of this assay, PA4 and BWR4 both bind undefined endogenous autoantigens and therefore could potentially form ICs that differ in size as well as content. Such differences could theoretically impact the extent of BCR crosslinking and/or interactions with additional PRRs. To address these concerns, we took advantage of the fact that PL2-3 activates B cells through both TLR9- and TLR7-dependent pathways, and that the TLR9 component can be removed by using Tlr9^−/− B cells while the TLR7 component can be removed by using Tlr7^−/− B cells. Other than TLR9 or TLR7, PL2-3 should engage the BCR and any other receptor to a comparable extent in all RF B cells. Therefore WT, Tlr7^−/−, Tlr9^−/− and Tlr7^−/−Tlr9^−/− RF B cells were stimulated for 3 days with PL2-3 in the presence of BLyS. Importantly, and as expected, the WT, Tlr7^−/− and Tlr9^−/− B cells divided and survived comparably (Fig. 2C), but only the PL2-3 activated Tlr9^−/− B cells differentiated into CD138⁺ plasmablasts (Fig. 3A, B). CD138⁺ cells were also detected in cultures stimulated with PL2-3 in the presence of the TLR9 inhibitor (Fig. 3C). These data demonstrate that BCR/TLR7 engagement alone, as compared to BCR/TLR9 engagement alone, is more likely to promote cell survival and differentiation toward the PC lineage. Furthermore simultaneous BCR/TLR9 co-engagement interferes with this process.

Molecular markers of plasmablast differentiation

To further verify the PC-skewed phenotype, RNA was isolated from WT B cells 4 days after BWR4, PA4, or PL2-3 activation and analyzed by qPCR for the expression of molecular markers of B cell differentiation. Consistent with the expression of CD138, BWR4-activated WT B cells expressed higher levels of Prdm1 and lower levels of bcl-6 and Pax5 (27), than any of the populations activated by BCR/TLR9 engagement (Fig. 3D). Overall gene expression patterns were further examined by microarray analysis of PL2-3 activated WT and TLR-deficient cells. Overall, the gene expression profiles of WT and Tlr7^−/− B cells were remarkably similar. Both populations only upregulated genes associated with PC differentiation at early time points followed by downregulation of these genes at later time points (Fig 3E). In contrast, Tlr9^−/− B cells upregulated and maintained expression of the PC transcription factors, Prdm1 and Irf4 (Fig 3E) (28, 29). Moreover, expression of B cell transcription factors such as Bcl6, Pou2f2, SpiB, and Ebf1, known targets of Prdm1, were more strongly repressed in the Tlr9^−/− cells at later time points (Fig. 3F) (30). Coordinately, expression of PC-related markers like syndecan-1 (Sdc1, Cd138), Egr1, and Csf1 were expressed at a higher level in the absence of TLR9, while expression of typical B cell markers (MHCII, Cxcr5, and Cd20) was trending downward at 42hrs (Fig. 3F) (30, 31). Finally, during PC differentiation, expression of BCR signaling components is lost, and cells redirect biosynthesis to increase the size and number of organelles like the ER, Golgi, and lysosomes to deal with the increased secretory load required for immunoglobulin secretion (30, 32). These changes are seen clearly in Tlr9^−/− B cells as reflected by a downward trend in the expression of BCR signaling components (Cd79a, Btk, Syk) (Fig. 3F), and increased expression of gene products required in the ER (Kdelr3, Uap1) and lysosomes (Lamp2, Fig. 3E). Together these data further support the idea that BCR/TLR7 co-engagement in the absence of BCR/TLR9 co-engagement, favors differentiation of FO B cells toward the PC lineage.

Regulation by IRF4 and IRF8

Recent studies have shown that high concentrations of IRF4 result in preferential binding to IFN sequence motifs, found in the promoter regions of genes associated with PC differentiation (33). Therefore IRF4 protein levels were monitored by flow cytometry in WT and TLR-deficient RF B cells stimulated with PA4 or PL2-3 (Fig. 4A); IRF4 protein was upregulated in all activated populations at 24 h post-activation (grey line), but high levels of IRF4 protein were only sustained at 72 h in the PL2-3 activated Tlr9^−/− cells (black line). A similar trend in mRNA levels at 42 hr, as detected by microarray, confirmed the sustained expression of IRF4 in PL2-3 activated Tlr9^−/− cells (Fig. 4B, left).

Fig. 4.

BCR/TLR7-dependent activation leads to the prolonged expression of IRF4 and BCR/TLR9-dependent activation leads to early expression of IRF8. (A) RF WT, Tlr7^−/−, Tlr9^−/− and Tlr7^−/−Tlr9^−/− B cells were stimulated with indicated ligands for either 24 h (dotted line) or 72 h (black line), permeabilized and stained for IRF4. The filled area represents the isotype control. Representative plots of 3 independent experiments are shown. (B) RF WT (black circles), Tlr7^−/− (grey squares) or Tlr9^−/− (open triangles) B cells were stimulated with PL2-3 for the indicated times and gene expression for IRF4 (left) and IRF8 (right) determined by microarray. (C) RF Tlr7^−/− (black line) and Tlr9^−/− (dotted line) B cells were stimulated with PL2-3 for 14 h (left panel) and 72 h (right panel) in the presence of 50 ng/ml BLyS. Expression of IRF8 (top) and IRF4 (bottom) was measured by flow cytometry, n=2.

In addition to IRF4, the transcription factor IRF8 is also important for B cell development and differentiation. However, IRF8 is upregulated in activated and germinal center B cells, but not required for PC differentiation (34). In contrast to IRF4, analysis of the microarray RNA expression levels revealed an early upregulation of IRF8 at 6 hr in PL2-3-stimulated WT and Tlr7^−/− cells (Fig. 4B, right), and less induction of IRF8 in Tlr9^−/− cells. The RNA expression levels correlated with upregulation of IRF8 protein at 14 hr in the Tlr7^−/− cells, as measured by flow cytometry (Fig. 4C). Together the data suggest that the differentiation of BCR/TLR7 activated cells is determined by an IRF4-dependent regulatory network while the differentiation of BCR/TLR9 activated cells is restrained by a counter-acting IRF8 network early during activation. Notably, IRF8 protein levels were increased at 72 hrs after both BCR/TLR7 and BCR/TLR9 coengagement, indicative of delayed expression of IRF8 in the BCR/TLR7 activated cells. However, late expression of IRF8 did not seem to promote PC differentiation.

TLR7 expression in WT and Tlr9^−/− RF B cells

Both TLR7 and TLR9 depend on Unc93b1 to acquire functional activity. The D34A mutant of Unc93b1 preferentially binds TLR7 and gene targeted mice that express Unc93B1 D34A develop lethal systemic inflammation (35). These data point to a critical balance between Unc93B1 and its capacity to bind TLR7 and TLR9 in the regulation of TLR-dependent responses. To determine whether the unique functional activity of Tlr9^−/− B cells, described above, simply reflected amplified TLR7 expression, we compared TLR7 protein levels in unstimulated and stimulated WT and Tlr9^−/− RF B cells by flow cytometry. In the unstimulated B cells, TLR7 levels were low but comparable between WT and Tlr9^−/− cells, and slightly higher than Tlr7^−/− or unstained cells (Figure 5A). Stimulation for 24 hr with CL097, led to significantly increased, but again comparable levels of TLR7 expression in the WT and Tlr9^−/− B cells. Stimulation with 1826 only led to increased expression of TLR7 in the WT cells as the Tlr9^−/− cells were not activated and the Tlr7^−/− cells did not express TLR7. Thus it appears that in both unstimulated and stimulated cells, WT and Tlr9^−/− B cells express the same amount of TLR7. These expression levels are completely consistent with the overlapping dose response curves of WT and Tlr9^−/− RF B cells in response to increasing concentrations of CL097 (Figure 5B). These values reflected relative mRNA levels, as determined by qPCR (data not shown).

Fig. 5.

RF WT and Tlr9^−/− B cells express comparable levels of TLR7. (A) Unactivated purified WT (grey line), Tlr7^−/− (black line) and Tlr9^−/− (dashed line) RF B cells or B cells activated with the indicated ligands for 24 hr, were permeabilized and stained for TLR7 expression. Unstained cells are indicated by the filled histogram. Representative plots of 3 experiments are shown. Data from the 3 experiments is summarized as MFI of the TLR7 staining intensities over background in the graph on the right. (B) RF WT, Tlr7^−/− and Tlr9^−/− splenic B cells were activated with the indicated concentrations of the synthetic TLR7 ligand CL097 for 30 h and proliferation was measured by ³H-thymidine uptake. Data represent the mean +/− SEM of 8 independent experiments.

RNA-associated ICs induce AFC in vivo

It was important to determine whether the preferential survival and plasmablast differentiation of BCR/TLR7 activated B cells, apparent in vitro, extended to in vivo responses. To ensure that activation conditions were as comparable as possible, we again stimulated TLR-sufficient and deficient RF B cells with PL2-3. WT, Tlr9^−/−, Tlr7^−/−, or Tlr7^−/−Tlr9^−/− RF B cells were labeled with VPD450 and injected i.v., together with 50 μg of PL2-3 or PBS, into BALB/c recipients. To accurately track the injected cells, CD45.2 RF B cells were injected into CD45.1 hosts. The mice were given a second injection of PL2-3 or PBS on day 3 and spleens were harvested on day 6. All genotypes engrafted comparably as similar numbers of RF cells were recovered from the PBS injected control groups (Figs. 6A top panel). RF Tlr7^−/−Tlr9^−/− B cells showed a minimal response, reiterating the critical role for BCR/TLR coengagement in the response to PL2-3 (Figs. 6A bottom panel, 6B). However, in contrast to the survival pattern observed in vitro where BCR/TLR9 activation induced post-proliferative cell death, the PL2-3 stimulated WT, Tlr9^−/− and Tlr7^−/− B cells all divided multiple times (Fig. 6A bottom panel, 6B), although the Tlr7^−/− cells did undergo fewer divisions than the Tlr9^−/− cells. There was also a trend toward fewer divisions in the WT group. The inability of BCR/TLR9 engagement to more effectively limit cell expansion of the transferred WT and Tlr7^−/− cells most likely reflects rescue through steady-state or induced B cell survival factors. Whether constitutive levels of BLyS are sufficient to maintain the survival of these cells is difficult to determine as it is known that PL2-3 IC activation of host DCs and neutrophils promotes the further production of BLyS (36, 37).

Fig. 6.

BCR/TLR7 activation promotes AFC differentiation in vivo. (A) BALB/c CD45.1 recipients were injected i.v. with 15 x 10⁶ WT, Tlr7^−/−, Tlr9^−/− and Tlr7^−/−Tlr9^−/− VPD450-labeled RF B cells and 50 μg PL2-3 on day 0 and injected again with PL2-3 on day 3. Spleens were harvested on day 6. B cell engraftment was ascertained by CD45.2 staining and proliferation was measured by dilution of VPD450. Representative plots of 3 independent experiments are shown. (B) The number of cell divisions based on the data in (A) was calculated for each mouse and the mean value +/− SEM are shown, n=3 mice/group. Statistical analysis was performed using Student’s t test, * P ≤ 0.05. (C) BALB/c CD45.1 recipients were injected with AM14 B cells as above but injected with PL2-3 on day 0, 3, 7 and 10. Additional BALB/c CD45.1 mice were only injected with PL2-3, and not RF B cells (none). Spleens were harvested on day 13; the number of clonotype positive IgG+ AFCs was measured by ELISpot, data is compiled from 4 independent experiments, One-way ANOVA including Tukey’s multiple comparison test was used for statistical analysis; ** P ≤ 0.005, *** P ≤ 0.0005.

To assess the role of TLR7 and TLR9 in PL2-3-induced PC differentiation, CD45.1 BALB/c mice were injected with RF B cells and PL2-3, and then given additional injections of PL2-3 on days 3, 7, and 10. On day 13, the total number of RF B cells in the spleen was determined by flow cytometry and the number of IgG1 and IgG2a antibody forming cells (AFC) was determined by a clonotype-specific ELIspot assay. At this time point we recovered comparable numbers of RF WT and Tlr7^−/− B cells, approximately twice as many RF Tlr9^−/− B cells, and very few RF Tlr7^−/−Tlr9^−/− B cells (data not shown). These data are consistent with the day 6 results and further support the premise that TLR9 expression limits expansion of autoreactive B cells. Importantly, Tlr9^−/− B cell injected mice had almost 10-fold more IgG AFC than Tlr7^−/− B cell injected mice, consistent with the propensity of the in vitro activated Tlr9^−/− cells to acquire PC markers (Fig. 6C). The mice injected with WT cells also had more AFC (2.5 fold) than Tlr7^−/− injected mice, as might be predicted by greater division. These data demonstrate that BCR/TLR7 B cell activation, in the absence of BCR/TLR9, preferentially induces autoreactive B cells to differentiate into isotype-switched AFC, as compared to BCR/TLR9 activated cells, as well as promoting improved survival.

Discussion

The analysis of TLR9-deficient murine models of SLE has given paradoxical results. Even though these mice fail to make autoantibodies reactive with dsDNA, as determined by the immunofluorescent staining of mitotic plates in ANA assays, they invariably develop more severe SLE associated with a decreased lifespan (2, 6, 38, 39). B cells have been shown to play a critical role in SLE, both through the production of autoantibodies that form pathogenic immune complexes, and through their capacity to activate autoreactive B cells. A previous study found that Tlr9^−/− B cells obtained from 2 mon old autoimmune prone Nba2 Yaa mice expressed higher levels of TLR7 mRNA and responded better to the TLR7 ligand Imiquimod than B cells obtained from age-matched TLR-sufficient Nba2 Yaa mice (8). However, these Tlr9^−/− Nba2 Yaa mice develop a hyperaccelerated autoimmune disease and survival is already compromised by 3 months of age. Therefore it is hard to determine whether the enhanced TLR7 response of the Tlr9^−/− Nba2 Yaa B cells reported in this study was due to the loss of TLR9 expression per se, or the fact that these B cells had already been activated in vivo by the autoimmune disease process, as we now document upregulation of TLR7 protein levels in response to TLR activation. In a separate report, purified B cells obtained from Tlr9^−/− and WT non-autoimmune prone C57BL/6 mice were compared and shown to produce comparable amounts of cytokine in response to the TLR7 ligand R848 (40). We now show that purified B cells from non-autoimmune prone mice responded comparably to increasing concentrations of a TLR7 ligand, and that both unstimulated and stimulated WT and Tlr9^−/− B cells expressed comparable levels of TLR7 as detected by flow cytometry with a TLR7-specific antibody (Figure 5). Therefore, TLR9-deficiency in B cells does not seem to impact the TLR7 signaling threshold. Nevertheless, as shown in the current study, BCR/TLR7 activation and BCR/TLR9 activation can lead to distinct functional outcomes, especially with regard to autoantibody production.

We found that isolated in vitro-activated BCR/TLR7-activated Tlr9^−/− B cells are more likely to differentiate toward the PC lineage than BCR/TLR9-activated Tlr7^−/− B cells, and that the BCR/TLR7-activated Tlr9^−/− B cells preferentially give rise to IgG autoantibody producing cells in vivo. Moreover, BCR/TLR9 activation can at least partially block the BCR/TLR7-driven response. Consistent with the studies of Oropallo et al. on non-Tg B cells (manuscript submitted), BCR/TLR9 co-engagement in vitro induced post proliferative cell death, even in cells co-activated by BCR and TLR7. Only Tlr9^−/− RF B cells could sustain an extended BLyS-independent response to PL2-3, an autoantibody that binds both DNA- and RNA-associated ligands. The addition of BLyS to the in vitro activated TLR9-sufficient WT RF B cells prevented post-proliferative cell death, but these cells still did not show the same capacity to differentiate toward the PC lineage as Tlr9^−/− RF B cells. The capacity of TLR9 to constrain the survival of PL2-3 activated RF WT cells in vivo was less apparent, perhaps due to either basal levels of BLyS, or to PL2-3 IC activation of pDCs (or other TLR-sufficient antigen presenting cells that express FcγRs) and the ensuing production of pro-survival factors. This is a limitation of the use of PL2-3 ICs. In the context of a (non-transgenic) polyclonal repertoire, DNA-reactive (or other autoreactive) B cells would be exposed to DNA-associated autoantigens directly, prior to the production of ICs that could engage FcγR⁺ cells. In fact, in a steady state, MRL.Fas^lpr Tlr9^−/− 3H9λ1⁺ DNA-reactive B cells have been shown to have a longer half-life in vivo than MRL.Fas^lpr Tlr9^+/+ 3H9λ1⁺ DNA-reactive B cells (41), consistent with the premise that the lifespan of naïve B cells responding to DNA-associated autoantigens is curtailed through a TLR9-dependent mechanism. However, once RNA-associated ICs are present in the circulation, and able to elicit the production of survival factors, BCR/TLR9 co-engagement of DNA- or chromatin-reactive cells, may no longer lead to cell death but rather activation.

It was somewhat surprising to find that the small molecule ligands 1826 and CL097 did not induce CD138 expression, as others have found that TLR ligands can drive B cells to become CD138⁺ (42–44). Most of these studies used a mixed population of FO and marginal zone (MZ) B cells. As we have reported previously (15), BALB/c AM14 sdTg mice lack B1, MZ and MZ precursor B cell compartments, and therefore our studies were carried out on a highly enriched population of naïve FO B cells. These RF B cells express very low levels of TLR7 prior to activation, and may not completely recapitulate the response of a polyclonal population. However, our data is consistent with the findings of Genestier and colleagues, who reported that TLR ligation predominantly induces MZ B cells and B1 B cells to differentiate into CD138⁺ plasmablasts and AFCs (45). BCR/TLR9 engagement of non-Tg polyclonal B cells also preferentially induces MZ B cells to differentiate into AFC (Oropallo et al., manuscript submitted).

The association between BCR/TLR9 activation and early upregulation of IRF8 is consistent with the phenotype of Irf8^−/− mice. B cell conditional Irf8^−/− mice have twice the number of mature B cells as well as greater numbers of MZ and B1 cells (34). In addition, they spontaneously produce anti-dsDNA autoantibodies by 3 months of age (46). Also, in contrast to MD4 x sHEL mice, Irf8^−/− x MD4 x sHEL B cells differentiate to a more mature phenotype and spontaneously produce anti-HEL antibodies (46). Together the data point to a major role of IRF8 in the maintenance of B cell tolerance, and therefore IRF8 expression by PL2-3 activated B cells may partly account for the negative regulatory role of TLR9.

An interesting comparison can be made between our in vivo experiments and a previous report that involved the day 7 in vivo PL2-3 response of autoimmune-prone MRL.Fas^lpr AM14 B cells (26). Even though this study found that TLR-sufficient mice appeared to have more AFCs than either the Tlr7^−/− or Tlr9^−/− mice, as detected by the number of ELIspot+ cells, the Tlr9^−/− mice had a higher percentage of plasmablasts than the Tlr7^−/− or WT mice, as determined by the phenotype CD22^lo CD138⁺. By contrast, in the current study, at day 13, the number of IgG⁺ AFCs produced by the WT B cells was significantly lower than the number produced by the Tlr9^−/− B cells. The MRL/lpr study may be somewhat confounded by the accelerated disease of Tlr9^−/− MRL/lpr mice, and subsequent changes in total spleen cell number and cell subset distribution. Nevertheless, the data suggest that the CD138⁺ cells are not full-fledged AFCs, but rather a distinct subset preferentially generated by BCR/TLR7 engagement, but only in the absence of BCR/TLR9 engagement. Additional studies will be necessary to further elucidate the direct impact of TLR7 and TLR9 on the long term survival of RF B cells as well as their capacity to move into specific short lived and long lived PC compartments.

Several labs have produced 80% μMT (or JhD^−/−) + 20% Tlr9^−/− mixed chimeras, in which TLR9-deficiency is predominantly limited to the B cell lineage (9, 41, 47). These B cell Tlr9^−/− mice invariably develop more severe clinical features including higher autoantibody titers, more extensive isotype switching of the autoantibody producing cells, and increased activation of potentially autoreactive T cells. However the potential contribution of other TLR9-expressing cell types cannot be completely ruled out because a significant proportion (20%) of the myeloid lineage could also be TLR9-deficient. Nevertheless, the 80% μMT + 20% Tlr9^−/− chimeric mice invariably developed greater numbers of autoantibody producing plasma cells as well as effector/memory T cells, more extensive ectopic follicles, and more severe renal disease (9), consistent with the notion that Tlr9^−/− B cells are more likely to differentiate into autoantibody producing plasmablasts and also more effectively activate autoreactive T cells. Exacerbated disease in these chimeras could also be attributed to the absence of TLR9-expressing cells that make protective antibodies required for the clearance of apoptotic debris (47), or to cytokines produced by residual Tlr9^−/− myeloid cells. The current study clearly shows that BCR/TLR9 and BCR/TLR7 co-engagement lead to distinct functional phenotypes. Our findings are strengthened by a recent publication that documents opposing roles for TLR7 and TLR9 in the formation of spontaneous GCs (48). Overall the data point to a unique B cell intrinsic role for TLR9 in the constraint of autoantibody production. Intriguingly, TLR8 has also been reported to negatively regulate murine SLE (40). These effects may reflect increased activity of TLR7, as a result of improved access to Unc93b (49), or alternatively, distinct downstream components of the relevant TLR signaling cascades. These possibilities will be addressed in upcoming studies. It will also be important to determine whether TLR9 plays a similar role in the regulation of human autoimmunity as a better understanding of the regulatory activity of the individual TLRs is likely to have implications for the optimal design of TLR-based therapeutics.

Abbreviations used in this paper

AFC

antibody forming cell

autoAbs

auto-antibodies

autoAgs

auto-antigens

BLyS

B Lymphocyte Stimulator

DC

dendritic cell

FO

follicular

IC

immune-complex

MZ

marginal zone

PC

plasma cell

SLE

systemic lupus erythematosus

RF

rheumatoid factor