Mer tyrosine kinase (Mer) signaling maintains immune tolerance by clearing apoptotic cells (ACs) and inducing immunoregulatory signals. We previously showed that Mer-deficient mice (Mer−/−) have increased germinal center (GC) responses, T cell activation, and AC accumulation within GCs. Accumulated ACs in GCs can undergo necrosis and release self-ligands, which may influence the outcome of a GC response and selection. In this study, we generated Mer−/− mice with a global MyD88, TLR7, or TLR9 deficiency and cell type–specific MyD88 deficiency to study the functional correlation between Mer and TLRs in the development of GC responses and autoimmunity. We found that GC B cell–intrinsic sensing of self-RNA, but not self-DNA, released from dead cells accumulated in GCs drives enhanced GC responses in Mer−/− mice. Although self-ligands directly affect GC B cell responses, the loss of Mer in dendritic cells promotes enhanced T cell activation and proinflammatory cytokine production. To study the impact of Mer deficiency on the development of autoimmunity, we generated autoimmune-prone B6.Sle1b mice deficient in Mer (Sle1b.Mer−/−). We observed accelerated autoimmunity development even under conditions where Sle1b.Mer−/− mice did not exhibit increased AC accumulation in GCs compared with B6.Sle1b mice, indicating that Mer immunoregulatory signaling in APCs regulates B cell selection and autoimmunity. We further found significant expansion, retention, and class-switching of autoreactive B cells in GCs under conditions where ACs accumulated in GCs of Sle1b.Mer−/− mice. Altogether, both the phagocytic and immunomodulatory functions of Mer regulate GC responses to prevent the development of autoimmunity.
Systemic lupus erythematosus (SLE) is an autoimmune disease characterized by the production of antinuclear Abs (ANAs) and systemic inflammation that leads to pathology in several organ systems. Homeostatically, apoptotic cells (ACs) are generated every day in a healthy individual. Inefficient clearance and accumulation of these ACs is an important factor in SLE development (1, 2). If not efficiently cleared, accumulated ACs undergo cell membrane disintegration in a process called secondary necrosis. During secondary necrosis, intracellular contents are released into the surrounding microenvironment, leading to self-antigen exposure (1–3). Autoreactive B cells are activated by self-antigen and break tolerance, leading to the secretion of autoantibodies that deposit in peripheral organs and induce myeloid cell recruitment, which drives inflammation. SLE patients exhibit AC accumulation in secondary lymphoid organs, specifically within germinal centers (GCs) (4, 5), which data suggest is due to inefficient clearance by CD68+ macrophages (4, 6). Further, functional studies have shown that monocyte-derived macrophages from lupus patients exhibit a deficit in AC clearance in vitro (7). These functional deficits correlate with the presence of polymorphisms in phagocytic receptors and their adaptor molecules in SLE patients (8–11).
AC clearance is mediated by many different phagocytic receptors, which are differentially expressed among immune cell subsets (12). In particular, deficiency of the TAM (Tyro3, Axl, and Mer tyrosine kinase [Mer]) family of receptors leads to systemic autoimmunity (13). The Tyro3, Axl, and Mer family of receptors are expressed on macrophages and dendritic cells (DCs) (14, 15), and dually function to phagocytose ACs and prevent sterile inflammation through the induction of SOCS1 and SOCS3, which dampen TLR and cytokine receptor signaling (15–18). Our previous studies have indicated that Mer expression on tingible body macrophages has implications for the regulation of the GC microenvironment (6, 19). Mice with a deficiency of Mer (Mer−/−) accumulate ACs predominantly within the GCs (6, 19). This AC accumulation in GCs correlates with multiple immune activation phenotypes, including enhanced GC B cell, T follicular helper (Tfh), Th1, and IgG2 Ab responses, events that are associated with the development of lupus-like autoimmunity (19).
A variety of factors are released from ACs, including nuclear Ags. Immune complexes, which contain self-RNA and DNA released from dead cells, are abundant in the serum and glomeruli in SLE (20, 21). These immune complexes stimulate TLRs 7 and 9, respectively, in innate immune cells and B cells in vitro (22–27). In vitro, activation of autoreactive B cells is dependent on the dual engagement of the BCR and TLR7 or TLR9 (25, 26, 28). Further, self-nucleic acid activation of TLRs in myeloid cells promotes an SLE-driving environment by inducing type I IFN responses (22, 27, 29). A number of in vivo studies on the RNA sensor TLR7 have demonstrated its requirement for enhanced spontaneous GC responses and autoantibody production in SLE-prone mice (30–38). Conversely, data on the deficiency of the DNA sensor TLR9 in lupus mouse models are more complex, demonstrating a regulatory function in spontaneous GC response but a requirement for the production of antiDNA Abs (33–35, 38–41).
Although TLRs are clearly involved in SLE development, the source of stimulatory factors for these receptors in vivo and the impact on self-antigen–induced GC responses and pathogenic autoantibody production are undefined. ACs are suggested as potential sources of TLR ligands; however, direct in vivo experimental studies have not been performed. Using our Mer−/− model in which ACs accumulate in GCs, we determined the innate sensors and cell types involved in the response to self-RNA and self-DNA released from dead cells, and the resultant impact of autoantigen availability in GCs on the GC response, T cell activation, and autoimmunity development. Accordingly, we generated TLR7- and TLR9-deficient mice on the Mer−/− background, in addition to global and cell-type–specific knockouts of the TLR signaling adaptor, MyD88, on the Mer−/− background. Because Mer deficiency alone does not drive robust autoimmunity, we generated mice with Mer deficiency on the autoimmune B6.Sle1b background (referred to as Sle1b.Mer−/−), which possesses the disease-causing SLAM family gene variants derived from the NZM2410 model (42, 43). Depending on the immunization conditions used, we were able to induce (or not) enhanced AC accumulation in these mice compared with their B6.Sle1b control counterparts, allowing us to isolate the contribution of Mer immunoregulatory signaling to the development of autoimmunity.
We found that in Mer−/− mice, MyD88 signaling was necessary for the enhanced magnitude of GC responses in vivo. Specifically, TLR7 but not TLR9 was essential for the enhanced GC responses observed in Mer−/− mice, highlighting the importance of self-RNA sensing. Interestingly, direct sensing of ligands through MyD88-dependent signaling pathways in GC B cells was critical for these responses, but no requirement for this sensing in macrophages was observed. Further promoting a permissive environment, Mer−/− DCs exhibited both enhanced proinflammatory cytokine production and Ag presentation, leading to increased activation and proliferation of T cells. Sle1b.Mer−/− mice exhibited accelerated lupus onset, even in the absence of AC accumulation, suggesting a potential cytokine-mediated control of B cell selection by Mer immunoregulatory signaling. Under conditions when ACs accumulated in GCs of Sle1b.Mer−/− mice, autoreactive GC B cell retention, expansion, and differentiation into class-switched IgG-producing cells was robust compared with B6.Sle1b mice, suggesting that AC accumulation also drives aberrant B cell selection processes. Collectively, our results indicate that Mer prevents the development of autoimmunity by multiple mechanisms. First, the clearance of ACs prevents aberrant exposure to self-RNA, which can be sensed by GC B cells to drive enhanced GC responses. Second, Mer immunoregulatory signaling in DCs limits Ag presentation, proinflammatory cytokine production, and T cell activation. Both Mer-mediated AC clearance and immunoregulatory signaling cascades are crucial for preventing autoreactive B cells from escaping tolerance.
Materials and Methods
Breeding pairs of MerTK-deficient mice (stock number 011122) on a B6/129F2 mixed background (referred to as Mer−/−), B6/129SF2/J wild-type (WT) mice (stock number 101045), MyD88-deficient (stock number 009088; MyD88−/−), LysMCre (stock number 004781), Ighg1Cre (stock number 010611; further referred to as GC-BCre), and MyD88 floxed (stock number 008888) mice were originally purchased from The Jackson Laboratory (Bar Harbor, ME) and bred in house. TLR7-deficient (TLR7−/−) and TLR9-deficient (TLR9−/−) mice were generated as previously described (44, 45). Mice were crossed to generate Mer−/−MyD88−/−, Mer−/−TLR7−/−, and Mer−/−TLR9−/− mice for our studies. Floxed and Cre mice were crossed to Mer−/− mice to generate Mer−/−LysMcre/+MyD88fl/fl and Mer−/−GCBCre/+MyD88fl/fl mice and littermate controls for our studies. B6.Mer−/− mice were generated by backcrossing Mer−/− mice (originally generated on a B6129F1 hybrid background) for 10 generations to a pure B6 background (46) and were bred in house. These mice have been verified to contain ∼98% B6 genome following backcrossing. B6 mice congenic for the Sle1b sublocus (B6.Sle1b) and p-azophenyl arsonate (Ars) and DNA dual-reactive H chain knockin (B6.HKIR) mice were generated as described previously (47–49). B6.HKIR mice were crossed to B6.SJL (CD45.1) mice (stock number 002014), generating congenically labeled heterozygotes for use in our studies. B6.Sle1b mice were crossed to B6.Mer−/− mice to generate B6.Sle1b.Mer−/− mice (referred to as Sle1b.Mer−/−). OT-II.Thy1.1 breeding pairs were generously provided by Dr. A. Lukacher (Pennsylvania State University College of Medicine, Hershey, PA) and bred in house. All mice were housed in a specific pathogen-free facility and all procedures were performed in accordance with Pennsylvania State University Institutional Animal Care and Use Committee guidelines.
The 4-hydroxy-3-nitrophenol–OVA immunization protocol
Mice 7–9 wk old were used for immunization studies. The T-dependent Ag (TD-Ag), 4-hydroxy-3-nitrophenol–OVA (NP-OVA), with a conjugation ratio between 14 and 17 (InvivoGen), was premixed for 1 h in a 1:1 mixture of Imject alum (Thermo Fisher Scientific) to PBS prior to i.p. injection. Mice received two doses of Ag: 200 μg on day 0 and a 150 μg booster dose on day 7. Spleens and serum were collected and analyzed at 60 or 90 d postimmunization (dpi) for independent experiments, as indicated.
Epicutaneous imiquimod treatment
A thin layer of Aldara 5% imiquimod (IMQ) cream (Meda, Solna, Sweden) was applied to ear tissue beginning at 6–8 wk of age three times per week for 8 wk as previously described (50).
Flow cytometry and Abs
Single-cell suspensions were prepared from harvested spleens by mechanical disruption. RBCs were lysed by incubation with Tris Ammonium Chloride. Single-cell splenocyte suspensions were stained with combinations of the following Abs: anti–B220-PacBlue (RA3-6B2), GL7-FITC, anti–CD95-PeCy7 (Jo2), anti–CD4–Alexa Fluor 700 (RM4-5), anti–CXCR5-biotin (2G8), and anti–PD-1-Pe (29F.1A12) (BioLegend). Biotinylated Abs were detected by staining with streptavidin-PeCy5. Dead cells were excluded from analysis by staining with Fixable Viability Dye (e780) (eBioscience). Data were acquired on an LSR II flow cytometer (BD Biosciences). Data were analyzed using FlowJo software (Tree Star, San Carlos, CA).
Immunohistology, FLICA staining, and Abs
Spleens and kidneys were flash-frozen in OCT compound and 6 μm sections were prepared on a cryostat. For histology staining, sections were fixed in cold acetone and stained as previously described (51). The following Abs were used on spleen sections: GL7-FITC, anti–CD4-PE (GK1.5), anti–IgD-allophycocyanin (11-26c.2a), anti–Ki67-PE (16A8), anti–CD169-APC (3D6.112), goat anti-mouse IgG-FITC (Jackson ImmunoResearch), anti–E4-biotin [generated in house by hybridoma kindly provided by Dr. U. Hammerling, Sloan-Kettering Memorial Hospital, New York, NY (52)], streptavidin-PE, and streptavidin-AF633. The following Abs were used on kidney sections: anti–C3-FITC (ICL Laboratory; polyclonal) and anti–IgG-PE (Abcam; polyclonal). To detect AC accumulation, the SR-FLICA Poly Caspase Assay kit was purchased from ImmunoChemistry Technologies and the protocol was performed according to the manufacturer’s instructions. Briefly, unfixed 6 μm spleen sections were blocked in 3% BSA in TBS and incubated with working concentration of SR-FLICA reagent, washed, and then subsequently stained with GL7-FITC and anti–IgD-APC. Images were acquired on a Leica DM4000 fluorescent microscope. Image color intensity was consistently enhanced among all images using Adobe Photoshop while maintaining the integrity of the data. Magnification for specific experiments is indicated in independent figures. GCs were quantitated by counting how many there were per 10× field in five randomly selected fields and by measuring the area of 10 randomly selected GCs per spleen section (or the highest number present on the section) for the number of mice indicated in independent figures. For colocalization studies, images were acquired on a C2+ confocal microscope system (Nikon) in 0.3 micron optical sections. Data were acquired using Nikon NIS elements software and object colocalization was calculated using SlideBook software (3i, version 5.0).
Ex vivo B cell proliferation assay
B cells were isolated from mice by anti-CD43 (Ly-48) negative selection MACS purification (Miltenyi Biotec). Isolated B cells were labeled with 3 μM CFSE and subsequently cultured with 20 μg/ml anti-CD40 (University of California, San Francisco) and 25 μg/ml anti-IgM (Jackson ImmunoResearch; Polyclonal Fab Fragment) for 72 h. Unstimulated controls were maintained in parallel. CFSE dilution was detected by flow cytometry.
For general Ab subtype analysis, Immulon 4HBX high binding plates (Thermo Fisher Scientific) were coated with IgG capture Ab (Invitrogen; polyclonal) for the assessment of total IgG, IgG1, IgG2b, and IgG2c titers. For ANA ELISAs, plates were coated with a 1:10 dilution of poly-l-lysine (Sigma), followed by coating with salmon sperm dsDNA, histone, or Smith ribonucleoprotein (SmRNP). Plates were blocked with 4% milk in PBS and incubated with diluted serum samples. Detection of Ab subtypes was performed using the following combinations of primary and secondary detection Abs: primary Abs were anti–IgG-biotin (Jackson ImmunoResearch; polyclonal), anti–IgG1-biotin (Invitrogen; polyclonal), anti–IgG2b-biotin (Southern Biotech; polyclonal), and anti–IgG2c-alkaline phosphatase (Southern Biotech; polyclonal); the secondary Ab was streptavidin-alkaline phosphatase (Vector Laboratories). Plates were developed with p-nitrophenyl phosphate, disodium salt (Thermo Fisher Scientific) substrate, and quantitation was performed as previously described (6).
OT-II T cell–DC coculture, Ag presentation, and cytokine analysis
Bone marrow was harvested from WT and Mer−/− mice. Bone marrow–derived DCs (BMDCs) were generated in vitro by supplementing 10% DMEM with 20 ng/ml of recombinant murine GM-CSF and 20 ng/ml Flt3L (PeproTech) for 2 d, followed by supplementation with 10 ng/ml GM-CSF and 10 ng/ml Flt3L for an additional 6 d. Resultant DCs (phenotypically verified by flow cytometry) were stimulated with 1 μg/ml LPS for 24 h and subsequently pulsed with 10 μg/ml OVA peptide (323–339) for 6–8 h. OT-II CD4+ transgenic T cells were purified from OT-II transgenic mouse splenocyte suspension using the Pan T Cell isolation kit (Miltenyi Biotec), labeled with CFSE (Sigma-Aldrich), and cocultured with activated DCs (53). CFSE dilution was assessed by flow cytometry to measure proliferation at the indicated time points. Culture supernatants were collected from coculture at 96 h and assessed for cytokine levels. Cytokine profile was assessed by flow cytometry using the mouse Th1/Th2/Th17 cytokine bead array kit (BD Biosciences) according to the manufacturer’s instructions. Data were fit to a standard curve for the determination of cytokine concentrations in individual samples. BD FCAP array software was used for analysis.
BMDC stimulation and cytokine analysis
BMDCs were generated as detailed in DC–OT-II coculture. BMDCs were stimulated for 24 h with 1 μg/ml LPS, 1 μg/ml IMQ, or 1 μg/ml CpG-C. Supernatants were collected from the culture and assessed for cytokine production by cytokine bead array analysis as detailed in DC–OT-II coculture.
Human epithelial-2 ANA assay
Human epithelial type 2 (HEp-2) diagnostic slides were purchased from Abs Inc. The assay protocol was followed according to the manufacturer’s instructions. Briefly, serum was diluted 1:50 in PBS in 2% BSA and incubated on HEp-2 slides. Detection of binding was performed by incubation with anti-κ-FITC (H139-52.1) and slides were imaged on a Leica DM4000 fluorescent microscope. Samples were characterized as exhibiting negative, nuclear, cytoplasmic, or nuclear and cytoplasmic staining patterns.
Kidney pathology analysis
Kidneys were fixed in 10% formalin solution, stored in 70% EtOH, and subsequently embedded in paraffin for sectioning. Kidney sections 3 μm thick were stained with periodic acid-Schiff reagent and imaged with an Olympus BX51 microscope fixed with a DP71 digital camera. One section per mouse was imaged and scored by a pathologist in a blinded manner. Individual glomeruli were assessed at 400× for glomeruli size and lobulation, karyorrhectic nuclear debris, thickening of the capillary basement membrane, mesangial matrix expansion, and mesangial cell proliferation on a scale of 0 (normal) to 4 (severe pathology).
HKIR adoptive cell transfer, immunization, and analysis
Experimental recipient mice were immunized on day 0 with 500 μg of Ars-keyhole limpet hemocyanin (KLH) (i.p.) in a 1:1 mixture of CFA/PBS. On day 7, B cells were isolated from B6.HKIR.SJL heterozygous mice by anti-CD43 (Ly-48) MACS negative selection (Miltenyi Biotec). Two million purified B cells were transferred via the tail vein into recipient mice. At the time of transfer, recipient mice were boosted with 250 μg of Ars-KLH in a 1:1 mixture of IFA/PBS. On day 12, spleens were harvested from mice for histological analysis.
Comparisons between multiple groups were performed by one-way ANOVA with multiple comparisons analysis, Tukey test. Comparisons between two groups were performed by Student t test with Mann–Whitney analysis. A p value ≤0.05 was considered significant and significance was assigned according to the following breakdown: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Graph Pad Prism 6 was used for all statistical analysis.
Self-RNA and not DNA released from accumulated ACs in GCs promotes elevated GC responses in Mer-deficient mice
We previously showed that B6.129-Mertktm1Grl/J F2 mice (designated Mer−/−) exhibit enhanced B and T cell activation, GC responses, and Ab class-switching compared with B6.129F2 (WT) mice following immunization with TD-Ag. These elevated responses correlated with the prolonged accumulation of apoptotic and late necrotic cells within the GCs in Mer−/− mice (19). In this study, we first determined whether the enhanced GC responses observed in Mer−/− mice are attributed to TLR stimulation by self-ligands aberrantly available due to local AC accumulation in GCs. We immunized Mer−/− mice and WT control mice with the TD-Ag NP-OVA (described in 2Materials and Methods). We assessed responses at 60 dpi whereby the foreign Ag response has resolved but a robust accumulation of ACs is still observed in the GCs of Mer−/− mice. At 60 dpi, GC B cell percentages (Fig. 1A, 1B) and the total magnitude of the GC response (Fig. 1C–E) were 2–3 fold higher in Mer−/− mice compared with WT. However, loss of MyD88, an adaptor molecule for all TLRs but TLR3, restored both the enhanced GC B cell percentage and magnitude of the GC response in Mer−/− mice back to WT levels (Fig. 1A–E). Both self-RNA (TLR7 ligand) and self-DNA (TLR9 ligand) released over time from ACs undergoing secondary necrosis may contribute to the MyD88-dependent enhancement of the GC response in Mer−/− mice.
To determine whether MyD88 deficiency could be phenocopied by TLR7 or TLR9 deficiency in Mer−/− mice, we generated Mer−/−TLR7−/− and Mer−/−TLR9−/− mice. To establish similar genetic backgrounds in Mer−/−MyD88−/−, Mer−/−TLR7−/−, and Mer−/−TLR9−/− mice, we crossed B6/129F2 Mer−/− mice in a similar manner to MyD88−/−, TLR7−/−, and TLR9−/− mice that were on the B6 background. When Mer−/− mice were deficient for TLR7, the enhanced GC B cell percentage and the magnitude of the GC response returned to a WT level at 60 dpi with NP-OVA (Fig. 1A–E). Conversely, Mer−/−TLR9−/− mice did not show any significant difference in these responses when compared with Mer−/− mice, indicating that self-RNA sensing through TLR7 but not self-DNA sensing through TLR9 drives the enhanced GC responses observed in Mer−/− mice (Fig. 1A–E). Mer−/−, Mer−/−MyD88−/−, Mer−/−TLR7−/−, and Mer−/−TLR9−/− mice exhibited elevated levels of AC accumulation when normalized to GC area compared with WT mice as determined by histological SR-FLICA staining, which detects active caspases involved in apoptosis induction (Fig. 1C, top panels, Fig. 1F). Interestingly, Mer−/−TLR9−/− mice exhibited a slight reduction in AC accumulation in GCs compared with Mer−/− mice (Fig. 1C, top panels, Fig. 1F).
Titers of IgG1 are primarily driven by response to immunization with NP-OVA in this system and no difference in IgG1 response is observed between WT and Mer−/− mice (data not shown). However, interestingly, in Mer−/− mice the serum titer of total IgG was increased compared with WT mice, which was reflected in the IgG2b and IgG2c titers (Fig. 2A–C). This correlated with enhanced GC responses. Mer−/−MyD88−/− mice exhibited a return to the WT level of IgG and IgG2 subtypes in the serum, correlating with the reduced magnitude of GC response in these mice (Fig. 2A–C). Further, TLR7 deficiency also resulted in a reduction in these responses to WT level, whereas TLR9 deficiency did not affect these Ab responses (Fig. 2A–C). These results indicate that the TLR7-MyD88 axis is required for the production of IgG2 Abs generated through self-antigen derived from AC accumulation in GCs (Fig. 2A–C). In addition to class-switching, B cells undergo rapid proliferation in GCs to drive clonal expansion. B cells derived from Mer−/− mice and further stimulated ex vivo with anti-IgM and anti-CD40 exhibited an enhanced rate of proliferation compared with WT B cells, which was abrogated when Mer−/− B cells lacked MyD88 (Fig. 2D, 2E). This indicates that MyD88-dependent self-ligand sensing is required for B cell hypersensitivity to proliferation signals following derivation from a cytokine- and AC-rich Mer-deficient microenvironment.
To completely eliminate the potential influence of the genetic background difference on the GC responses and confirm our results, we generated another set of experimental mice in which we could compare Mer−/−TLR7−/− and Mer−/−TLR9−/− mice with littermate control Mer−/− mice. We found similar results, which showed a reduced GC and Ab response in Mer−/−TLR7−/− mice compared with their littermate controls, whereas Mer−/−TLR9−/− mice did not exhibit any difference in GC or Ab response compared with littermate control mice (data not shown).
Collectively, these results indicate that the self-ligand sensing through TLR7-MyD88 signaling during GC responses in Mer−/− mice drives multiple GC events including B cell proliferation and Ab class-switching. Based on these results, we further tested if overstimulation of TLR7 could recapitulate this GC phenotype in our system in the absence of initial antigenic stimulation. We treated mice by epicutaneous application of the TLR7 agonist, IMQ, as previously described (50). Even though the total percentage of GC B cells was similar between IMQ-treated WT and Mer−/− mice (Fig. 3A, 3B), larger GC structures were observed in Mer−/− mice compared with WT mice (Fig. 3C, 3D). Following IMQ treatment, even though not statistically significant, GCs in Mer−/− exhibited a trend toward greater AC accumulation normalized to GC area compared with IMQ-treated WT mice (Fig. 3C, top panels, Fig. 3E). As expected, Mer−/−MyD88−/− and Mer−/−TLR7−/− mice did not exhibit a response to IMQ treatment and the size of GCs was reduced in Mer−/−MyD88−/− and Mer−/−TLR7−/− mice (Fig. 3A–D). However, AC accumulation was apparent in Mer−/−MyD88−/− and Mer−/−TLR7−/− GCs despite their small size (Fig. 3C, top panels, Fig. 3E). Consistent with our observations in Mer−/− mice immunized with TD-Ag (Figs. 1, 2), the IgG2 titers were 2–3 fold higher in IMQ-treated Mer−/− mice compared with IMQ-treated WT mice (Fig. 3G, 3H), but total IgG responses were comparable between IMQ-treated WT and Mer−/− mice (Fig. 3F). Enhanced IgG2 responses were dependent on TLR7 and MyD88 signaling, with a more substantial deficit upon MyD88 deficiency (Fig. 3G, 3H), whereas total IgG responses in Mer−/− and WT mice were dependent on MyD88 but not TLR7 signaling alone (Fig. 3F). Despite robust GC responses induced following IMQ treatment, autoantibodies were undetectable in the serum of both Mer−/− and WT mice at this age (Fig. 3I).
GC B cell–intrinsic MyD88 signaling is required for elevated GC responses in Mer−/− mice
Although our data indicated that the requirement of self-RNA sensing through TLR7-MyD88 signaling is critical for elevated GC responses in Mer−/− mice, the cell type–specific contribution of self-RNA sensing to this phenotype remained unclear. In contrast to prior studies that used bone marrow–chimeric approaches to demonstrate the B cell–intrinsic requirement of TLR7 signaling in the initiation of spontaneous GCs (33, 34), in this study we examined the impact of direct sensing of self-RNA by GC B cells on driving enhanced GC responses following GC initiation, when self-antigen is locally available in Mer−/− mice (6, 19). Based on the phenotypic similarity between Mer−/−MyD88−/− and Mer−/−TLR7−/− mice at 60 dpi, we used a GC B cell–specific Cre-loxP system in which MyD88 is excised following the induction of class-switching, thus eliminating TLR signaling in GC B cells following differentiation.
At 60 dpi with NP-OVA, Mer−/−MyD88fl/flGC-BCre/+ mice exhibited a 2-fold reduction in the GC B cell percentage and magnitude of GC response compared with Mer−/−MyD88fl/fl control mice (Fig. 4A–E). AC accumulation within GCs was similar between Mer−/−MyD88fl/fl control mice and Mer−/−MyD88fl/flGC-BCre/+ mice when normalized to GC area (Fig. 4B top panel, 4F). The reduction in GC response resulted in a reduced serum titer of total IgG, IgG1, IgG2b, and IgG2c in Mer−/−MyD88fl/flGC-BCre/+ mice compared with Mer−/−MyD88fl/fl control mice (Fig. 4G–J), mirroring the response observed in global MyD88 knockout mice. GC-Tfh response was also reduced in Mer−/−MyD88fl/flGC-BCre/+ mice, but Tfh (CD4+CXCR5intPD-1int) were unaffected (Fig. 4K–M).
Macrophage-intrinsic MyD88 signaling does not impact the magnitude of the GC response in Mer−/− mice
GC B cell–intrinsic MyD88 signaling exhibits a critical role in the enhanced GC responses observed in Mer−/− mice, but Mer-expressing macrophages are primarily responsible for AC clearance. As such, these macrophages might also indirectly contribute to these responses through promoting T cell activation and/or altering the local cytokine environment. Therefore, we generated Mer−/−MyD88fl/flLysMCre mice to study the impact of TLR signaling primarily in Mer−/− macrophages on the enhanced GC and Tfh responses observed in Mer−/− mice. Deficiency of MyD88 in macrophages had no effect on the magnitude of the GC response (Fig. 5A–E), the total IgG, IgG2, and IgG1 titers (Fig. 5G–J), or the amount of GC-Tfh and Tfh (Fig. 5K–M). AC accumulation in GCs did not differ between Mer−/−MyD88fl/fl control mice and Mer−/−MyD88fl/flLysMCre/+ mice (Fig. 5F). These data indicate that self-ligand stimulation of TLRs in macrophages is not required for the enhanced GC response in Mer−/− mice, nor is it required for the enhanced magnitude of T cell responses observed in Mer−/− mice.
Mer−/− DCs contribute to proinflammatory cytokine production and promote T cell activation to establish a permissive microenvironment
Although MyD88 deficiency in macrophages did not impact the magnitude and quality of GC responses in Mer−/− mice, we sought to determine whether the cytokine environment generated by Mer−/− DCs is conducive to promoting enhanced immune responses in Mer-deficient mice. To determine the differences in cytokine profile between WT and Mer−/− DCs, and the dependence of this profile on MyD88 signaling, we stimulated BMDCs generated from WT, Mer−/−, and Mer−/−MyD88−/− mice with TLR4 ligand (LPS), TLR7 ligand (IMQ), and TLR9 ligand (CpG), and assessed the inflammatory cytokine profile generated 24 h after stimulation. Unstimulated DCs did not show any TNF or IL-6 production (Fig. 6A, 6B). Following stimulation with LPS or IMQ, Mer−/− DCs exhibited increased production of both TNF (Fig. 6A) and IL-6 (Fig. 6B) compared with WT DCs. Production of these cytokines was almost completely abrogated when Mer−/− DCs were also deficient for MyD88 (Fig. 6A, 6B). Following stimulation with CpG, similar trends were observed, although differences in TNF production were not statistically significant (Fig. 6A, 6B). Neither IFN-γ nor IL-17 was produced by any genotype of DC following stimulation with LPS, IMQ, or CpG (data not shown).
Next, we sought to determine whether T cell interaction with Mer−/− DCs was further able to establish a proinflammatory environment. To assess this, we used a BMDC–OT-II T cell coculture system whereby we preactivated DCs with LPS to prime them for Ag presentation and subsequently added OT-II cells, in the presence or absence of OVA peptide. LPS was selected as it provides a more robust activation of DCs, including their MHC class II and costimulatory molecule expression, compared with IMQ or CpG. IL-17, IFN-γ, TNF, and IL-6 were increased in Mer−/− cocultures compared with WT cocultures when OVA peptide was added, indicating that when Mer−/− DCs interact with T cells, the production of these proinflammatory cytokines is upregulated (Fig. 6C–F). Even in the absence of OVA peptide, IL-6, IFN-γ, and IL-17 production was upregulated in Mer−/− DC cocultures compared with WT DC cocultures, detailing some cytokine production that is occurring in an Ag presentation-independent manner (Fig. 6C–F). When Mer-deficient DCs were also deficient for MyD88, the production of TNF returned to a WT level or lower, whereas IL-6 production was completely absent, indicating that the production of these inflammatory cytokines is MyD88 dependent (Fig. 6C, 6D). MyD88 deficiency also resulted in a partial reduction in IFN-γ production in the presence of OVA peptide, although some IFN-γ was produced in a MyD88-independent manner (Fig. 6E). In contrast, enhanced IL-17 production was driven primarily in a MyD88-independent manner as no difference in IL-17 production was observed when Mer-deficient DCs were also deficient in MyD88, except in the absence of peptide (Fig. 6F). We found that the relative production of these cytokines in macrophage–OT-II coculture was low in comparison with DCs, although similar trends were observed (data not shown).
In addition to assessing for increases in proinflammatory cytokine production in coculture, we assessed T cell activation marker expression and proliferation to determine whether Mer−/− DCs were better Ag-presenting cells than WT DCs. We found that OT-II T cells cultured with Mer−/− DCs exhibited an enhanced rate of activation, as shown by increases in both CD25 (Fig. 6G) and CD44 (Fig. 6H) expression. Further, OT-II T cells cultured with Mer−/− DCs exhibited enhanced proliferation compared with OT-II T cells cultured with WT DCs (Fig. 6I).
Loss of Mer immunoregulatory signaling accelerates autoimmunity independent of the magnitude of the GC response and AC accumulation in GCs
As we have described above, Mer-deficiency alone does not result in the overt and clinically detectable development of autoimmunity in younger (4–5 mo old) mice, indicating that the possession of genetic susceptibility loci is required for the loss of tolerance in these mice. Therefore, to study the impact of Mer deficiency on the development of autoimmunity, we crossed Mer-deficient mice on a B6 background (B6.Mer−/−) to autoimmune-prone B6.Sle1b mice to generate Sle1b.Mer−/− mice. B6.Sle1b mice possess the Sle1b genetic susceptibility locus derived from the autoimmune-prone NZM2410 strain and, by crossing with B6.Mer−/− mice, we were able to maintain the Sle1b susceptibility locus on a pure B6 background in combination with Mer deficiency. B6.Sle1b mice have enhanced GC responses, aberrant selection processes, and a loss of GC tolerance (42).
B6.Sle1b mice start to develop mild autoantibody titers by 3–5 mo of age with a peak in autoantibody production at 7–8 mo of age (33, 42, 43, 47). This leads to mild kidney pathology that is detectable at 9 mo of age (54). Therefore, we assessed whether Sle1b.Mer−/− mice exhibited the acceleration of autoimmunity development by assessing responses at 90 dpi with TD-Ag NP-OVA, at which time the mice were a total of 5 mo of age. At 90 dpi (5 mo age total), B6 and B6.Mer−/− mice did not exhibit clinically detectable levels of autoantibody production, as determined by clinical HEp-2 analysis (Fig. 7A, 7B). At this age, half of the B6.Sle1b mice exhibit a low level of autoantibody production, primarily to nuclear Ags (Fig. 7A, 7B). However, when crossed to B6.Mer−/− mice, the resultant Sle1b.Mer−/− mice exhibit more robust autoantibody production by 5 mo of age and an increased penetrance of autoimmunity (Fig. 7A, 7B). To assess autoantibody specificity and titers for different autoantigens in Sle1b.Mer−/− mice, we measured the serum titers against three common autoantigens, representing RNA, histone, and DNA reactivity. We found that Sle1b.Mer−/− mice exhibited a significant increase in IgG2c class-switched serum autoantibody titers against SmRNP (Fig. 7C), histone (Fig. 7D), and dsDNA (Fig. 7E) compared with B6.Sle1b mice. The increase in autoantibody production correlated with the early onset of kidney pathology in Sle1b.Mer−/− mice (Fig. 7F–H). Whereas B6, B6.Mer−/−, and B6.Sle1b mice exhibit negligible IgG and complement deposition in the kidneys at 5 mo of age, Sle1b.Mer−/− mice have large and organized IgG and complement deposits in glomeruli (Fig. 7F), which drives early stages of glomerular nephritis (Fig. 7G, 7H). Glomerular nephritis is not observed in B6.Sle1b mice at this age (Fig. 7G, 7H). This indicates that the loss of Mer signaling accelerates the onset of autoimmunity in genetically susceptible mice.
Despite the acceleration of autoimmunity in Sle1b.Mer−/− mice, the magnitude of the GC response and GC-Tfh responses were similar between B6.Mer−/− mice and B6 mice, and similar between Sle1b.Mer−/− and B6.Sle1b mice following immunization with TD-Ag in alum (Supplemental Fig. 1A–G). Interestingly, to our surprise, we did not observe a robust accumulation of ACs in the GCs of B6.Mer−/− or Sle1b.Mer−/− mice under these conditions (Supplemental Fig. 1C, top panel). These results further strengthen the notion that AC accumulation is tied strongly to the magnitude of the GC response under conditions of Mer-deficiency. However, as selection is still impaired despite the lack of AC accumulation in Mer-deficient mice, the impact of Mer signaling in APCs on B cell selection processes is evident.
Mer signaling regulates B cell selection at the GC tolerance checkpoint
Although accelerated autoantibody production is observed in Sle1b.Mer−/− mice compared with B6.Sle1b mice, it is unclear whether dysregulated B cell selection is occurring at the level of the GC. To assess the integrity of the GC selection checkpoint, we used a transgenic B cell system termed HKIR (48, 49). HKIR B cells exhibit dual-reactivity of the BCR to Ars and dsDNA. Following transfer into mice immunized with Ars-KLH, these B cells are recruited into the GC reaction due to their affinity for the foreign Ag. In mice that have intact GC selection processes (i.e., B6 mice), HKIR B cells are prevented from expansion in the GC due to their reactivity to dsDNA (42). However, under conditions of aberrant GC selection, they are able to persist and their anatomical location can be detected by histology. Following the transfer of HKIR B cells into B6.Sle1b and Sle1b.Mer−/− mice immunized with Ars-KLH in CFA, we observed a moderate amount of HKIR B cells in B6.Sle1b GCs, which had vastly expanded in Sle1b.Mer−/− GCs (Fig. 8A, 8B). Under these immunization conditions, we also observed robust AC accumulation in Sle1b.Mer−/− mice (Fig. 8C, 8D). E4+ HKIR GC B cells in Sle1b.Mer−/− mice were positive for Ki67 expression, indicating robust proliferation within the GC, whereas the majority of proliferating GC B cells in B6.Sle1b mice were not E4+ HKIR cells (Fig. 8E). To determine whether Sle1b.Mer−/− mice had more class-switched E4+ HKIR plasmablasts as a result of enhanced HKIR E4+ GC responses, we stained for E4+ HKIR B cells expressing IgG and looked for clusters of these cells in the spleen. As defined by CD169 staining of marginal zone metallophilic macrophages (Fig. 8F) and determined by confocal imaging and quantitation, we found a greater overlay of IgG staining on E4+ cells in the splenic red pulp of Sle1b.Mer−/− mice as compared with that of B6.Sle1b mice (Fig. 8F, 8G). The amount of total IgG staining in the red pulp of B6.Sle1b and Sle1b.Mer−/− spleens showed no statistically significant difference, in accordance with a similar total GC response to immunization with Ars-KLH in CFA (data not shown). Our results using this transgenic system suggest that the loss of B cell tolerance can be driven by both the loss of the phagocytic and immunoregulatory functions of the Mer receptor. Particularly, Mer deficiency and resultant AC accumulation can promote the persistence and proliferation of autoreactive B cells in the GC, leading to their exit as class-switched autoantibody-secreting cells.
Mer signaling has important roles in the maintenance of immunological tolerance (3, 15). We have previously shown that Mer deficiency drives alterations in the GC microenvironment, which correlates with the local accumulation of ACs. Although AC clearance mediated through the activation of Rac1 signaling is a critical function of the Mer receptor (17, 55, 56), immunoregulatory signaling cascades that feedback on TLR and cytokine receptor signaling pathways are also initiated upon receptor engagement (16, 18), detailing two key functions of this receptor that may impact the magnitude and quality of GC responses. In our studies of Mer−/− mice on a mixed B6 and 129 background, we noted robust accumulation of ACs in GCs. However, comparing our B6.Sle1b mice with Sle1b.Mer−/− mice, both of which are on a B6 background, we did not see differences in accumulation of ACs and the GC responses unless more robust GC responses were induced through the use of a stronger adjuvant. These data strengthen the notion that ACs drive the enhanced magnitude of GC response in Mer−/− mice on a mixed B6 and 129 background through TLR-mediated self-ligand sensing, with a lesser contribution from the loss of Mer-induced immunoregulation. This difference in response also allowed us to determine that the loss of Mer-induced immunoregulation in DCs and/or macrophages contributes significantly to the altered selection of autoreactive B cells in Sle1b.Mer−/− mice, as B cells do not express Mer protein (data not shown). Our transgenic system further revealed that AC accumulation in GCs can further promote dysregulated selection processes. Thus, both GC B cell– and APC-intrinsic effects of Mer signaling contribute to the maintenance of tolerance.
Although previous studies have demonstrated the critical role of TLR7 in spontaneous GC responses, the source of the ligand that drives these GC responses is still unclear (33, 34, 38). Previous in vitro studies have indicated that autoreactive B cells stimulated with self-RNA– or self-DNA–containing immune complexes require both BCR and TLR7 or TLR9 stimulation, respectively, to attain optimal activation (25, 26, 28). This suggests that the sensing of RNA and/or DNA debris may drive GC responses in vivo. In this study, we show that despite the likelihood that robust amounts of self-DNA are released from accumulated ACs in GCs over time, the sensing of self-RNA by TLR7 in GC B cells is predominantly responsible for driving the enhanced GC responses in Mer−/− mice. Whereas another study has reported enhanced AC accumulation in autoimmune-prone BWF1 mice, which develop enlarged spontaneous GC structures (57), our findings suggest that RNA-based self-antigens may be one of the important AC-derived factors that contribute to the induction of spontaneous GC responses.
In addition to the previous deficit in knowledge with regard to the source of TLR7 ligand in these responses, the role of TLR7 in GC responses following GC initiation was unclear. Previous studies of B cell–intrinsic TLR7 signaling in the study of spontaneous GC responses have used bone marrow–chimeric approaches. Using bone marrow chimeras, we and others previously showed that mice with a B cell–intrinsic deficiency of TLR7 failed to form spontaneous GCs (33, 34). Although this chimeric approach allowed us to determine the B cell–intrinsic role of TLR7 signaling in spontaneous GC induction without knowing the source of self-ligand, we were unable to investigate the role of TLR7 signaling in the maintenance of spontaneous GC response. As TLR7 deficiency phenocopies MyD88 deficiency in GC responses in Mer−/− mice, in the current study we were able to use a Cre-loxP system whereby MyD88 expression is lost following the induction of class-switching in GC B cells after GCs were formed. As discussed, Mer−/−GC-BCreMyD88fl/fl mice exhibited reduced GC response compared with Mer−/−MyD88fl/fl mice. Because TLR7 signaling is tightly linked to B cell proliferation capacity (33), the reduced magnitude of the GC response in Mer−/−GC-BCreMyD88fl/fl mice may be attributed to the limited capacity of recycling GC B cells to undergo additional proliferative events.
Unlike the role of TLR7 in promoting spontaneous GC responses, TLR9 is known to negatively regulate these responses in a B cell–intrinsic manner (33, 34). Interestingly, in Mer−/− mice, we found that although TLR9 was not required for enhanced GC responses in this model, it did not regulate them either. These data indicate that much like how the role of TLR9 has been suggested to be context dependent in the development of autoimmunity, it is also context dependent in the initiation of GC responses in Mer−/− mice. Although TLR9 is not required in elevated GC responses in Mer−/− mice, cytosolic DNA pattern recognition receptors such as cGAS-STING may contribute to enhanced GC responses observed in Mer−/− mice (58). However, STING has also been shown to have regulatory functions (59). Cell surface sensing of nucleosome or HMGB1 complexes by TLR2 may also be involved in this process and recently Mer signaling has been shown to negatively regulate TLR2-induced responses (60–62).
Our data indicate that sensing of self-RNA through TLR7 is critical for multiple aspects of B cell and GC responses driven by accumulated ACs in GCs of Mer−/− mice. Although TLR7-mediated sensing is required for ongoing response by Sle1b.Mer−/− GC B cells, the aberrant selection of B cells in Sle1b.Mer−/− mice can be driven significantly by deficits in Mer immunoregulatory signaling, as we observed an acceleration in the onset of autoimmunity in Sle1b.Mer−/− mice even when they did not exhibit greater AC accumulation or magnitude of GC response than B6.Sle1b mice. Using our HKIR transfer system, we observed that AC accumulation in GCs of autoimmune prone mice may further promote aberrant selection, specifically within the GC microenvironment. Interestingly, we did not detect significant autoantibody production without possession of genetic susceptibility. However, another study has reported autoantibody production in nonautoimmune prone mice in response to repeated exposure to large numbers of ACs (63).
Apart from GC B cell–intrinsic sensing of self-ligands, Mer deficiency in DCs could lead to enhanced T cell activation and an increase in proinflammatory cytokines available in the local microenvironment, both of which may impact the quality of GC selection (64). Mer is known to regulate TLR function, which directly and indirectly impacts the production of a variety of cytokines that promote autoreactive GC B cell survival (16, 65). We show that DCs derived from Mer−/− mice exhibit enhanced proinflammatory cytokine production and Ag presentation, leading to increases in T cell activation. T cell hyperactivity and elevated levels of IFN-γ in vivo may promote autoreactive GC B cell survival when they would otherwise undergo cell death. When this alteration in the cytokine profile synergizes with AC accumulation and sensing, autoreactive GC B cells expand greatly and exit the GC as class-switched plasmablasts secreting high-affinity autoantibodies.
Phenotypes observed in B6.Mer−/− and Sle1b.Mer−/− mice immunized with TD-Ag in alum were not associated with differences in AC accumulation and correspondingly there was no difference in the magnitude of the GC response. These results seemingly contradicted our previous data where we observed increased GC responses in B6.Mer−/− mice compared with B6 mice, which correlated with the accumulation of ACs (6). Interestingly, when we immunized with a stronger adjuvant (CFA) and a different TD-Ag (Ars-KLH), we observed 3–4 fold higher GC responses than following immunization with TD-Ag alum and in turn, a robust accumulation of ACs in Sle1b.Mer−/− mice that was reminiscent of previous responses obtained in B6.Mer−/− immunized with TD-Ag in alum. Throughout the time that the previous and current experiments have been performed, our mice have been housed in different barrier facilities. The differences in our results using the same strain of mouse from different mouse facilities may highlight a potential role for differences in the microbiota, which is in line with recent studies that suggest that the microbiota contributes to the quality of immune responses in cancer, infection, and autoimmunity (66–68).
Collectively, our results indicate that Mer prevents the development of autoimmunity by multiple mechanisms. First, efficient clearance of ACs in GCs prevents the GC B cell–intrinsic sensing of self-RNA that is capable of activating autoreactive GC B cells to expand and exit as class-switched autoantibody-secreting cells. Second, Mer immunoregulatory signaling dampens Ag presentation and the production of proinflammatory cytokines by APCs, events that can drive T cell activation and GC responses. Lastly, Mer immunoregulatory signaling in DCs and macrophages regulates B cell selection processes. Altogether, the function of Mer has important implications for the maintenance of peripheral B cell tolerance and loss of Mer drives the development of systemic autoimmunity.
We thank Dr. Shizuo Akira for generously providing the TLR7−/− and TLR9−/− mouse strains; the Pennsylvania State University Hershey Medical Center Department of Comparative Medicine for mouse care and housing; the Pennsylvania State University Hershey Medical Center Flow Cytometry core facility for assistance; and Dr. Sathi Babu Chodisetti from the laboratory for assistance and suggestions. Finally, we thank Tracy Krouse and Dr. Chris Norbury for guidance with confocal imaging and analysis.
This work was supported by Department of Defense Congressionally Directed Medical Research Program PR130012 (to Z.S.M.R.).
The online version of this article contains supplemental material.
Abbreviations used in this article:
bone marrow–derived dendritic cell
human epithelial type 2
keyhole limpet hemocyanin
Mer tyrosine kinase
systemic lupus erythematosus
T follicular helper
The authors have no financial conflicts of interest.