TLR-Induced Murine Dendritic Cell (DC) Activation Requires DC-Intrinsic Complement

This article is featured in In This Issue, p.1

Toll-like receptors initiate innate immune activation, which amplifies T and B cell immune responses. Surface-expressed TLRs (1–6) and intracellular TLRs (7–10) recognize conserved molecular patterns derived from pathogens (pathogen-associated molecular patterns [PAMPs]) (1–3, 7) and, potentially, some damaged/necrotic cells (e.g., HMGB1 which signals through TLR4) (4). TLR-initiated signaling is transmitted through linker proteins, which include TRIF (TICAM1), to activate gene-expression profiles driven by NF-κB, AP-1, and members of the IFN regulatory factor (IRF) family (3, 5, 8, 9) to induce dendritic cell (DC) maturation (3, 10). Important among processes associated with DC maturation are upregulation of MHC class II (MHCII) expression, induction of CD80, CD86, and CD40, and production of “innate” cytokines (e.g., type 1 IFN, IL-1, IL-6, IL-12, TNF-α) (2, 5, 6, 8, 9, 11). TLR activation educates DCs to initiate effector T cell (T_eff) differentiation (e.g., Th1, Th2, Th17) and expansion while limiting Foxp3⁺ T regulatory cell (T_reg) generation, function, and stability (12, 13), together providing protective, pathogen-reactive T cell immunity (3, 14).

Nonphysiological TLR activation can overcome normal immunological homeostasis and result in pathological immune injury. For example, HMGB1/TLR4 activation is a crucial mediator of ischemia reperfusion injury (15, 16), and experimental CpG/TLR9 ligation can promote Th1 and Th17 cell activation in rodent models of autoimmunity (17–20). In solid organ and hematopoietic cell transplant systems, TLR9-initiated signals accelerate rejection or graft-versus-host-disease (GVHD) (13, 21–25).

Conventionally, the complement system has been connected with innate immunity. In previous work (26–28), we found that, early during interaction of murine DCs with cognate T cells, both partners locally generate C3a and C5a, which establish autocrine C3ar1/C5ar1 signaling loops with DC- and T cell–expressed receptors (C3ar1 and C5ar1). This process occurs in concert with downregulation of the cell surface C3/C5 convertase regulator decay accelerating factor (DAF; or CD55). The resultant C3ar1/C5ar1 signaling in the DCs and cognate T cell partners plays an integral role in DC-induced T_eff activation, including during GVHD (27–36).

The observations that C3ar1/C5ar1 signaling and TLR signals each cause changes in DCs that promote their ability to induce T_eff responses prompted the hypothesis that the two processes are linked (i.e., that TLR-enhancing effects on T_eff immunity are dependent on C3ar1/C5ar1 signal transduction in DCs). We use analyses of DC-interacting T cells, along with murine models of alloimmunity, to demonstrate that C3ar1/C5ar1 signaling in DCs is a requisite process connecting TLR stimulation with DC maturation and T_eff activation at threshold PAMP concentrations that model physiological TLR signaling.

C57BL/6 (B6), B6 CD45.1, B6 C3^−/−, OTII, B6 Myd88^−/−, B6 TRIF^−/−, and BALB/c mice were originally purchased from The Jackson Laboratory and maintained at the Mount Sinai Center for Comparative Medicine or at Case Western Reserve University. Factor D^−/− (fD^−/−) mice were a kind gift from Y. Xu (University of Alabama at Birmingham, Birmingham, AL). B10.D2 Hc⁰ mice (C5 deficient [C5^def]; The Jackson Laboratory) and BALB/c C3^−/− mice (backcrossed for >12 generations from B6 C3^−/−mice) were crossed together to produce C3^−/−C5^def mice. B6 C3ar1^−/−C5ar1^−/−, C3ar1^−/−C5ar1^−/− Foxp3-GFP mice were produced as previously described (28, 37–39). Rosa(dTomato)×Foxp3CreERT2-GFP mice (40) were obtained from A. Rudensky (Sloan-Kettering Institute, New York, NY) and were crossed with B6 C3ar1^−/−C5ar1^−/− mice to produce C3ar1^−/−C5ar1^−/−(dTomato)×Foxp3CreERT2-GFP mice. TEa mice (41) were a gift of J. Bromberg (University of Maryland). All mice were housed in the Icahn School of Medicine at Mount Sinai Center for Comparative Medicine or at Case Western Reserve University in accordance with guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care International. All experiments were performed using animals that were littermates or were maintained in the same room and/or were cohoused within the same cages to limit the potential effects of microbiome differences.

Abs against CD4, CD8, CD11c, and fixable viability dye (eBioscience), CD55 (BD Biosciences), CD88 (AbD Serotec), CD80, CD86, and MHCII (Miltenyi Biotec, Auburn, CA) were used for flow cytometry. CFSE was obtained from eBioscience. MR1 (anti-mouse CD40L) was purchased from Bio X Cell (West Lebanon, NH). Peptides were synthesized by GenScript (Piscataway, NJ).

Spleen was passed through a 40-μm strainer (BD Falcon) and lysed with RBC lysis buffer (Life Technologies/Thermo Scientific, Waltham MA). T cell depletion from bone marrow (BM) suspensions and isolation of murine splenic naive CD4⁺ T cells and total T cells/APCs were accomplished using magnetic beads and an autoMACS Pro Separator (Miltenyi Biotec). For splenic DC isolation, single-cell suspensions of spleen cells were treated with Collagenase D (Roche) for 30 min at 37°C and incubated with CD11c MicroBeads (Miltenyi Biotec), following the manufacturer’s protocol. The purity of DCs after isolation was 85–95%. Isolated DCs were stimulated with CpG ODN 1826 or LPS (both from InvivoGen) for 18 h in 96-well plates, if needed.

Splenic DCs from untreated mice were stimulated in vitro with TLR ligands overnight, washed with PBS three times, and then cocultured with naive allogeneic CD4⁺ T cells or unfractionated T cells labeled with CFSE for MLRs. Analogous studies were performed using splenic DCs isolated 4 h after i.v. injections of CpG. On day 4, CFSE dilution was assessed for cellular proliferation, and live cell numbers were counted in the well using flow cytometry. Cells were incubated in complete medium (RPMI 1640 + 10% FCS + l-glutamine + sodium pyruvate + nonessential amino acids + Pen/Strep + 2-ME) at 37°C. Splenic DCs were phenotyped for surface markers by flow cytometry. In some experiments, cells were harvested at 18 h for analysis of cytokine gene expression by quantitative PCR (qPCR).

Cytokine ELISPOT assays were performed using spleen cells cocultured with BALB/c APCs on IFN-γ capture plates for 24 h and then analyzed as previously described (42).

Data were collected on a FACSCanto II (BD Biosciences) and analyzed using FlowJo software (TreeStar, Ashland, OR) or Cytobank (Cytobank, Santa Clara, CA). To measure recall immune responses posttransplant, spleen cells from heart transplant recipients were stimulated with donor cells overnight and then analyzed for intracellular IFN-γ within the CD4 or CD8 gate by flow cytometry (32).

Heterotopic heart transplants were performed as previously described by our laboratory (32, 43–45). For graft survival experiments, recipients were treated with anti-CD40L (anti-CD154) mAb MR1 (1 mg on day 0 and 500 μg on days 7 and 14, i.p.) with or without CpG ODN 1826 (100 μg on day 1 and 50 μg on days 3 and 5, i.p.). Heart graft function was monitored every other day by palpation; rejection was defined as the day on which a palpable heartbeat was no longer detectable and was confirmed by histology.

RNA isolation was performed using TRIzol Reagent (Thermo Fisher), and cDNA was reverse transcribed using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems), as per the manufacturer’s instructions. Real-time PCR (TaqMan probes; Applied Biosystems) was performed using a CFX96 Real-Time PCR Detection System (Bio-Rad Laboratories). All mouse PCR primers were purchased from Life Technologies. PCR products were normalized to the control gene (GAPDH) and expressed as the fold increase compared with unstimulated cells using the ΔΔCt method.

Splenic APCs were cultured in serum-free HL-1 medium with allogeneic or syngeneic splenic T cells, with or without CpG (10 μg), in 48-well plates for 48 h. Culture supernatant fluids were concentrated with an Amicon Ultra-0.5, normal m.w. limit of 10 kDa (Millipore), and tested for C5a with a Mouse Complement Component C5a DuoSet ELISA (R&D Systems, Minneapolis, MN), as per the manufacturer’s instructions.

Six- to eight-week-old male B6 or BALB/c mice were fasted for 24 h prior to irradiation. On day 0, recipients were irradiated with 650 rad twice, with a ≥3-h interval between treatments. Once irradiated, mice received adoptive transfer of T cell–depleted BM cells isolated from the various donors. Eight to ten weeks later, the percentage of chimerism was assessed by flow cytometry.

Tamoxifen (Sigma-Aldrich) was dissolved in olive oil (Fluka) to a final concentration of 20 mg/ml by shaking overnight at 37°C in a light-blocking vessel. The dose of tamoxifen was determined by weight: ∼75 mg/kg body weight of a mouse.

We isolated splenic CD11c⁺ DCs (using Miltenyi Biotec magnetic beads) from wild-type (WT) or C3ar1^−/−C5ar1^−/− mice 4 h after injection with CpG (100 μg) or vehicle control. The cells were immediately placed in TRIzol Reagent and sent to the State University of New York Albany Center for Functional Genomics. Total RNA was isolated by standard techniques (>150 pg RNA was obtained per sample). After quality-control testing, the samples were processed using standard WT Pico protocols and hybridized to Mouse Gene 2.0 ST Arrays, and the chips were scanned using a GeneChip Scanner 7G (all from Affymetrix). The intensity data at the probe set level were extracted and normalized with the RMA algorithm (46), and data quality was assessed using Expression Console (Affymetrix). The Affymetrix control probe sets or the probe sets with low intensity across all samples were excluded from downstream analysis. The limma test (47) was performed on normalized data between comparison groups, and the differentially expressed genes with p < 0.05 were identified and visualized using a heat map. Gene ontology enrichment analysis using a Fisher exact test was performed on differentially expressed genes to investigate their associated biological functions or pathways. The microarray data presented in this article have been submitted to the National Center for Biotechnology Information Gene Expression Omnibus under accession number GSE98315 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE98315).

Statistical significance was determined using a Student t test (unpaired, two-tailed), two-way ANOVA (with a Bonferroni posttest to compare replicate means), or a log-rank (Mantel–Cox) test with GraphPad Prism 5 or Prism 6, with a significance threshold p value < 0.05. All experiments were repeated at least twice. Data are presented as mean values with SD.

We first studied the effects of DC-produced complement on TLR3, TLR4, and TLR9 induction of IFN-γ production by OVA_323–339+I-A^b–specific CD4⁺ OT-II T cells. To exclude the effects of systemic complement on signaling through each of the TLRs, we performed in vitro studies using DCs as APCs in the absence of serum complement. We titrated OVA_323–339 and TLR ligand concentrations to establish threshold conditions for detection of IFN-γ by the responding cells (Fig. 1A). A total of 0.1 μg/ml OVA gave ∼25% of the maximal IFN-γ response, and 0.1 μg/ml of polyinosinic-polycytidylic acid (poly I:C), CpG, or LPS did not generate IFN-γ–producing cells in the absence of OVA_323–339 (Fig. 1A). In the presence of Ag, the number of IFN-γ–producing OT-II cells increased in a dose-dependent manner with the addition of increasing amounts of each PAMP (Fig. 1A), such that the effect of TLR signaling on endogenous complement production, and vice versa, could be interrogated.

Using the above conditions, with or without threshold amounts of each PAMP, we determined the effects of TLR stimulation on complement component gene expression by cells within the culture (noting that DCs produce ∼1000-fold more complement than T cells) (28). The inclusion of each PAMP increased C3, factor B (fB), and factor D (fD) mRNA expression levels from OT-II cultures (Fig. 1B) over the levels induced by OVA_323–339 without a PAMP. Substitution of C3^−/−, fD^−/−C3ar1^−/−, or C5ar1^−/− DCs for WT DCs abolished the TLR-induced upregulation (Fig. 1B), connecting autocrine C3ar1/C5ar1 signaling with TLR signaling. Studies with DCs from MyD88^−/− or Ticam1^−/− (Trif^−/−) mice showed that poly I:C–induced C3/fD (Fig. 1C) mRNA upregulation is mediated via Ticam1, that LPS-induced DC C3/fD (Fig. 1D) gene upregulation is mediated via MyD88, and that CpG-induced C3 expression required MyD88 (data not shown), together confirming that the PAMP/TLR-induced effects are transmitted through established canonical signaling pathways for each of the TLRs (48).

To test whether the TLR-induced increases in T cell IFN-γ production and autocrine C3ar1/C5ar1 signaling in DCs are linked, we cultured OT-II cells with the defined threshold concentrations of Ag and PAMPs, using WT DCs or DCs from C3^−/−, fD^−/−, C3ar1^−/−, C5ar1^−/−, or C3ar1^−/−C5ar1^−/− mice (Fig. 2A). These assays showed that the absence of DCs reduced IFN-γ production by OT-II cells by >80%. The same pattern was observed for IL-2 RNA production by LPS-stimulated OT-II cells (Fig. 2B).

To verify that the link between TLR-induced T cell responses and DC-derived complement has relevance in a polyclonal system, we performed studies with B6 DCs and allogeneic BALB/c T cells, using T cell proliferation (CFSE dilution) and expansion (live cell number) as readouts (Fig. 2C–E). We determined that pretreatment with 0.3 μg/ml CpG was the minimum threshold dose required for WT DCs to increase alloreactive T cell proliferation and expansion (data not shown) in MLRs. Unstimulated allogeneic DCs induced proliferation of allogeneic T cells: 13–16% of the T cells in the culture underwent more than four cell divisions on day 4, consistent with previous observations (27, 29, 35, 37). Although CpG pretreatment of WT DCs augmented alloreactive T cell proliferation and expansion (Fig. 2C–E), CpG pretreatment of C3ar1^−/−C5ar1^−/− DCs had no effect. Effects paralleling those with C3ar1^−/−C5ar1^−/− DCs occurred using C3/C5-deficient DCs (Fig. 2F, 2G). In the case of WT DCs, CpG induced stronger allogeneic T cell proliferation/expansion of purified CD4⁺ T cells (note reversal of responder/stimulator strains). For BALB/c H-2^d C3^−/−C5^def DCs pretreated with CpG and cultured with purified C3^−/− T cells (removing all C3 and DC-derived C5), proliferation was severely blunted, and T cell expansion was fully abrogated (Fig. 2F, 2G). The same was true for purified CpG-induced enhancements of CD8⁺ T cell responses and for LPS-induced enhancements of T cell proliferation/expansion (data not shown). Thus, for monoclonal T cells and polyclonal T cells responding to DCs in two different genetic backgrounds, TLR augmentation of T cell immunity in vitro was dependent on DC complement synthesis and C3ar1/C5ar1 signals.

To establish that the above interpretation applies in a disease-relevant context, we used a monoclonal TCR-transgenic system germane to alloimmune responses [i.e., TEa CD4⁺ cells (41) (reactive to I-A^b + I-E^dα_52–68), which have been used in transplant models]. We incubated TEa T cells with allogeneic (bxd)F1 APCs (that express I-A^b + I-E^dα_52–68) or syngeneic APCs as controls, with or without added CpG. (Fig. 2H). C5a was detectable in 48-h supernatants of cultures containing allogeneic (bxd)F1 APCs (which undergo cognate interactions with TEa cells). The amounts of C5a increased in cultures containing added CpG. In contrast, no C5a was detected in cultures containing control B6 APCs, with or without CpG. Flow cytometric analyses showed that CpG caused ∼30–50% downregulation of DAF on (bxd)F1 APCs and TEa CD4⁺ compared with cells incubated with media alone (Fig. 2I). Together with previous findings using polyclonal T cells (26), these data show that TLR-induced immune cell C5a production is connected to repression of DAF expression.

We next tested whether the interposition of C3ar1/C5ar1 signaling between TLR signals and DC maturation applies in vivo. We used TLR9 ligation, because its ligation by CpG has been extensively studied in GVHD and solid organ transplant systems (13, 22–25). TLR9 stimulation also augments T_eff induction following immunization (49), as well as heightens pathology in several autoimmune models (18, 19). To define the threshold concentrations of CpG required to induce DC maturation markers and allogeneic T cell stimulatory activity in vivo, we injected (i.v.) various amounts of CpG (0–100 μg) into WT mice. We isolated splenic CD11c⁺ DCs 4 h later and analyzed DC surface phenotypes and tested DC allostimulatory function by coculturing the DCs with allogeneic CFSE-labeled T cells (Fig. 3A, 3B). These assays demonstrated that 100 μg of CpG was the minimum concentration that consistently upregulated CD80, CD86, and MHCII on DC surfaces and enhanced proliferation and expansion of cocultured allogeneic T cells. We next injected groups of B6 WT and C3ar1^−/−C5ar1^−/− mice with 100 μg of CpG, isolated splenic CD11c⁺ DCs 4 h later, and compared DC surface phenotypes and function ex vivo between the two groups (Fig. 3C–G). These experiments showed that the absence of C3ar1/C5ar1 prevented CpG-induced upregulation of CD80/CD86 (Fig. 3C, 3D) and MHCII (WT: 40 ± 2.8% increase versus no CpG; C3ar1^−/−C5ar1^−/−: 21 ± 5.0%, p < 0.05 versus WT, data not shown), as well as abrogated CpG-induced augmentation of allogeneic T cell proliferation and expansion (Fig. 3E, 3F). CD40 expression levels were not altered by CpG administration in WT or C3ar1^−/−C5ar1^−/− DCs (data not shown). For WT DCs, CpG administration augmented mRNA expression of IL-1β, TNF-α, IL-12p40, and IL-6. In contrast, DCs from C3ar1/C5ar1 mice showed diminished TNF-α, IL-12p40, and IL-6 and fully restrained upregulation of IL-1β (Fig. 3G).

We next performed two sets of studies to exclude any contribution of C3ar1/C5ar1 or MyD88 signaling in non-BM cells. We transplanted BM from B6 WT or C3ar1^−/−C5ar1^−/− mice into congenic lethally irradiated B6 MyD88^−/− recipients. Because CpG/TLR9 signaling is MyD88 dependent (7, 19), in these chimeras only the transplanted BM-derived cells are capable of responding to CpG through MyD88. Nine weeks later (after confirming chimerism, Fig. 4A), we injected groups of chimeras with 100 μg of CpG and analyzed DC phenotypes and function (Fig. 4B–F). These assays demonstrated that the added CpG induced maturation of WT DCs, as assessed by CD80/CD86 surface expression, the ability to stimulate alloreactive T cells, and the ability to upregulate proinflammatory cytokine gene expression, whereas it had much lesser effects on C3ar1^−/−C5ar1^−/− DCs.

In the second approach, we tested whether the TLR effects conform to our prior findings (26–29, 36) that T_eff responses (without TLR stimulation) depend upon immune cell–derived complement rather than serum/liver-derived complement. We constructed WT BM→C3^−/−C5^def, C3^−/−C5^def BM→WT, and control WT→WT and C3^−/−C5^def→C3^−/−C5^def BM chimeras (H-2^d) and verified >90% donor chimerism 10 wk later (Fig. 5A). We then injected groups of chimeras with CpG or vehicle, isolated splenic DCs 4 h later, and analyzed them as above (Fig. 5B–E). These assays showed that, although DCs from CpG-treated WT BM chimeras upregulated surface expression of CD80/86, these costimulatory molecules remained unchanged on DCs from CpG-treated chimeras possessing C3^−/−C5^def BM, regardless of host C3/C5 expression (Fig. 5B). Culturing of the isolated DCs with allogeneic T cells (Fig. 5C, 5D) showed that in vivo CpG administration augmented ex vivo T cell proliferation/expansion when DCs from mice with WT BM were used, but not when DCs from mice with C3^−/−C5^def BM were used, regardless of serum complement. The absence of C3/C5 in BM cells likewise blunted CpG-induced upregulation of IL-1β, TNF-α, IL-12p40, and IL-6 RNA (Fig. 5E).

To broadly assess the role of C3ar1/C5ar1 signaling in TLR-induced DC maturation in vivo, we compared the effects of in vivo CpG injection on genes expressed by splenic DCs from WT and C3ar1^−/−C5ar1^−/− mice using microarrays (Fig. 6). Previous array studies by other investigators performed using in vitro TLR9-stimulated DCs showed that 0.1–10 μg/ml CpG upregulates DC expression of proinflammatory cytokines, type 1 IFNs, and costimulatory signals and that many effects are NF-κB (RelB) dependent (50–52). Other work on murine spleen cell gene expression following in vivo administration of high-dose (400 μg) CpG showed that maximal responses were detected 3–4 h postinjection and resulted in IFN-γ– and TNF-α–initiated inflammatory processes (53, 54).

Building upon these findings and using our above-identified threshold CpG doses for inducing DC maturation (Fig. 3A), our new analyses show that, at 4 h, 100 μg CpG induces upregulation of ∼1200 WT DC genes (Fig. 6A). These genes mapped to the GO terms immune system process, response to virus, innate immune response, defense response to virus, inflammatory responses, immune response, negative regulation of viral genome replication, cellular responses to interferon-beta, defense response to protozoan, and responses to LPS (data not shown). Included genes are related to the type 1 IFN pathway (IFNb and multiple IRFs), various cytokines and chemokines (e.g., CCL17, CCL22, CCL3, CCL4, CCL5, CXCL1, CXCL10, CXCL11, CXCL13, CXCL3, CXCL9, IL-10, IL-15, IL-27, IL-6, IL12A, IL12B, TNF), signaling pathway genes known to be downstream of TLRs (NFKBIA, CD40, LY96, IKBKE, MYD88, CD80, CD86, MAP3K8, PIK3R5, PIK3R1), and complement components (C3, CFB). Of the 1198 upregulated genes, 608 of them (51%) were induced by TLR9 in DCs from WT mice but not in DCs from CpG-treated C3ar1^−/−C5ar1^−/− mice (Fig. 6A). These TLR9-induced C3ar1/C5ar1-dependent genes mapped to the terms innate immunity, immune systems processes, inflammatory response, and defense response to virus (Fig. 6B, 6D). Although the remaining 590 of 1198 genes (49%) were upregulated in WT and C3ar1^−/−C5ar1^−/− mice (including cytokine pathway–related genes, TLR pathway–related genes, TNF pathway–related genes, and Jak-STAT pathway–related genes, Fig. 6C, 6E, 6F), greater increases for the majority of them occurred in WT DCs, indicating that C3ar1/C5ar1 signaling amplifies the induction of these other TLR9-induced DC genes. One hundred and fifty-three genes were uniquely upregulated in DCs from CpG-treated C3ar1^−/−C5ar1^−/− mice. In contrast to the upregulated genes unique to CpG-treated WT DCs, the majority of the upregulated genes in CpG-treated C3ar1^−/−C5ar1^−/− DCs mapped to noninflammatory pathways, including regulation of myeloid differentiation, regulation of protein kinase cascade, negative regulation of cell differentiation, and regulation of MAPPKKK cascade. Together with the phenotyping and functional data delineated above, the findings support the conclusion that autocrine C3ar1/C5ar1 signaling is a crucial intermediary required for TLR-induced DC activation.

Prior work showed that costimulatory blockade with anti-CD40L mAb (MR1) delays T cell–mediated cardiac allograft rejection and can promote T_reg-dependent allograft tolerance (44, 55, 56). CpG administration reverses the effects of MR1, inhibits T_reg function, augments alloreactive T_eff expansion, and causes rapid transplant rejection (13, 23). To test the importance of the above-described connection between TLR activation and DC-derived complement activation in a clinically relevant system, we used this transplantation model. We transplanted groups of B6 WT and C3ar1^−/−C5ar1^−/− recipients with allogeneic BALB/c hearts and treated them with MR1, with or without CpG (Fig. 7). Consistent with previous reports (13, 23), allografts transplanted into MR1-treated WT recipients survived for >90 d, whereas posttransplant CpG administration caused cessation of graft heartbeat (with histological cellular rejection, data not shown), with a median survival time (MST) of 15 d (p < 0.01 versus MR1-treated WT controls). In contrast, allografts transplanted into MR1+CpG-treated C3ar1^−/−C5ar1^−/− recipients survived longer, with an MST of 45 d (p < 0.01 versus MR1+CpG WT controls, Fig. 7A). In the absence of MR1, all WT and C3ar1^−/−C5ar1^−/− recipients rejected their grafts by day 9 (data not shown).

We repeated the transplant experiments and quantified donor-reactive CD8⁺ T cells on day 14 (all allografts beating). Although CpG administration to MR1-treated WT recipients augmented the frequencies and total numbers of splenic donor-reactive IFN-γ–producing CD8⁺ T cells (Fig. 7B, 7C), CpG administration had no effect on the low frequencies or total numbers of donor-reactive T cells detected in the MR1-treated C3ar1^−/−C5ar1^−/− recipients. No donor-reactive Abs were detected in the sera of any MR1-treated recipients (data not shown).

To test the hypothesis that the prolonged allograft survival of CpG-treated C3ar1^−/−C5ar1^−/− mice is causally linked to the interconnection of TLR signaling and C3ar1/C5ar1 signaling in immune cells, we performed transplant studies in (CD45.2 H-2^b) C3ar1^−/−C5ar1^−/− BM→(CD45.1) WT chimeras and H-2^b WT→WT controls. Ten weeks after documenting ≥90% donor chimerism (data not shown), we transplanted the chimeras with allogeneic hearts and treated them with MR1, with or without CpG (Fig. 7D). Heart grafts transplanted into the chimeras with C3ar1^−/−C5ar1^−/− BM survived longer than did those transplanted into chimeras with WT BM (MST of 51 and 21 d, respectively, p < 0.01). Similar results (Fig. 7E) were obtained in MR1-treated, H-2^d C3^−/−C5^def recipients (MST of 56 d, p < 0.01 versus WT control) compared with WT recipients+MR1/CpG (MST of 21 d) and in corresponding BM chimeras (WT BM→WT CpG+MR1 [MST of 34 d] versus C3^−/−C5^def BM→WT CpG+MR1 [MST of 50 d], p < 0.05, n = 3–4 per group, data not shown), ascribing the CpG reversal of MR1 costimulatory blockade to heightened immune cell C3ar1/C5ar1 signaling.

Prior work by other investigators showed that costimulatory blockade with MR1 induces donor-reactive T_regs, which prolong cardiac allograft survival/tolerance (55), and that TLR9 disrupts Foxp3 stability in T_regs (12, 13). In view of these findings, we tested the hypothesis that TLR9 signaling impairs Foxp3 stability through a C3ar1/C5ar1-dependent mechanism. We used B6 Rosa(dTomato)×Foxp3CreERT2-GFP (40) fate-mapping mice, in which Foxp3⁺ cells constitutively express GFP, and tamoxifen treatment induces expression of dTomato under the Foxp3 promoter (GFP⁺dTomato⁺). Thereby, dTomato expression without Foxp3 expression (i.e., GFP^negdTomato⁺ phenotype) identifies T cells that were previously Foxp3⁺ but have lost Foxp3 expression (termed ex-T_regs) (40) (Fig. 8A).

We transplanted allogeneic BALB/c hearts into groups of B6 WT and C3ar1^−/−C5ar1^−/− T_reg fate-mapping recipient mice. We treated both groups with posttransplant MR1 (to induce T_regs and prolong allograft survival) and with tamoxifen to induce Foxp3-GFP⁺dTomato⁺ T_regs. We then administered CpG or vehicle to parallel groups of heart graft recipients and analyzed their spleen cells 14 d posttransplantation (Fig. 8A). CpG administration increased ex-T_reg formation from ∼16 to 25% in WT recipients (p < 0.05, Fig. 8B, 8C). In contrast, ex-T_reg formation in C3ar1^−/−C5ar1^−/− fate-mapping mice did not increase following CpG treatment (p = not significant).

To test whether T_reg extrinsic, systemic C3ar1/C5ar1 signaling mediates in vivo T_reg instability following CpG stimulation, we induced dTomato⁺Foxp3-GFP⁺ T_regs in WT B6 fate-mapping mice by injecting them with allogeneic BALB/c spleen cells plus MR1 in the presence of tamoxifen and IL-2 (Fig. 8D). Seven days later, we sorted the Foxp3-GFP⁺dTomato⁺ WT T_regs and adoptively transferred them into WT or C3ar1^−/−C5ar1^−/− hosts. We then administered BALB/c spleen cells, with or without CpG DNA, to the adoptive hosts and analyzed the splenic T cells on day 14. These assays showed that CpG significantly increased ex-T_reg formation in WT adoptive hosts but had no effect on ex-T_reg formation in C3ar1^−/−C5ar1^−/− adoptive recipients (Fig. 8E).

Our findings demonstrate that autocrine C3ar1/C5ar1 signaling in DCs is a requisite process in TLR-mediated DC maturation that is required for induction and amplification of T_eff responses. TLR3, TLR4, and TLR9 ligations cause MyD88- and/or TICAM1-dependent complement gene (i.e., C3 and fB) upregulation in splenic DCs, in concert with downregulating cell surface DAF, which together augment T cell proliferation and survival. The latter is likely mediated in part by C3ar1/C5ar1-initiated signals in T cells that we previously linked to PI-3Kγ−dependent alterations in Bcl-2, caspase-3, and Fas expression (27, 28, 31).

Using in vitro, ex vivo, and in vivo analyses, our results uniquely demonstrate that this TLR-initiated complement production and C3ar1/C5ar1 signaling by immune cells are required for TLR-induced increases in DC costimulatory molecules, proinflammatory cytokine secretion by DCs, upregulation of DC gene pathways broadly related to inflammation and immune responses, DC-dependent augmentation of T_eff proliferation/expansion, and T_reg instability. Observations that CpG administration induced DC maturation in WT→MyD88^−/− BM chimeras, but not in C3ar1^−/−C5ar1^−/−→MyD88^−/− BM chimeras, indicate that the in vivo effects of TLR9 operate via direct effects on immune cells.

We document profound and expansive effects of autocrine C3ar1/C5ar1 signaling on TLR9-induced DC maturation in vivo by comparative gene-expression analyses. We attempted to model physiological concentrations of CpG in vivo by titrating to threshold effects and by performing studies in MyD88 chimeras. Under the tested conditions, we found that 50% of the upregulated genes were fully dependent on C3ar1/C5ar1 signaling, and increases in the remainder of the genes were impaired in the absence of C3ar1/C5ar1 signaling. Although the latter result suggests that some CpG-induced changes may be complement independent, at least in part, it is possible that exogenous (i.v.) CpG administration at any dose is superphysiological and induces gene-expression changes that are partially dependent on complement, whereas DC maturation in response to physiological TLR ligation by a pathogen is fully dependent on complement. The findings by other investigators, that CpG administration at similar doses to those used in this study induces curative antitumor immunity (57, 58), suggest that the observed CpG-induced antitumor processes may also require autocrine C3ar1/C5ar1 signaling in DCs.

Our data confirm TLR9-induced upregulation of pathways linked to DC maturation, antiviral immunity, and type 1 IFNs, among others, in WT mice. They also show that the absence of C3ar1/C5ar1 entirely prevented or severely blunted these changes shown to be essential for induction of effective immune responses to pathogens, model Ags, and autoantigens (50–52). TLR-initiated signals alter gene expression by inducing upregulation and intranuclear transport of various transcription factors that include several IRFs, AP-1, and NF-κB (5, 6, 13, 59), the last of which has been independently linked to transcriptional regulation of fB (60). Our microarray data suggest that, although many effects of TLR9 are fully dependent upon autocrine C3ar1/C5ar1 signaling, other signals initiated by C3ar1, C5ar1, and TLR9 may interact additively or synergistically to optimally induce upregulation and/or intranuclear transport of transcription factors required for DC maturation. One candidate target for this signaling nexus is NF-κB, because it is required for DC maturation (51, 61), and because NF-κB is upregulated following ligation of C3ar1 (62, 63), C5ar1 (64, 65), or TLR ligation alone (66) in DC, among other cell types.

The 153 TLR9-induced genes uniquely upregulated in DCs from C3ar1^−/−C5ar1^−/− mice mapped to regulatory/inhibitory pathways, including regulation of myeloid leukocyte differentiation, regulation of protein kinase cascade, and negative regulation of cell differentiation (Database for Annotation, Visualization and Integrated Discovery (DAVID) analysis, data not shown). The smaller number of genes downregulated by TLR9 in WT DCs (∼250) or C3ar1^−/−C5ar1^−/− DCs (∼140, data not shown) mapped to basic pathways related to DNA repair, cell cycle, and cell division (DAVID analysis), possibly implicating a role for C3ar1/C5ar1 signaling in regulating these fundamental cellular processes.

Although past studies have linked complement activation with TLR effects on APCs or T cells (67–70), they have not mechanistically connected the linkage with complement production by APCs themselves. These previously described effects have been attributed to “cross-talk” among systemic complement activation fragments and immune cells. Examples of reported linkages between complement and TLR signaling in APCs include the synergistic effects of C5a and TLR4 signaling on IL-8/IL-6 production by human monocytes (71), C5a augmentation of LPS-induced IL-10 production by human monocytes (72), synergism of C5a and TLR4 signaling in cytokine production by monocytes (71), and the ability of C5a and TLR ligands to substitute for each other in enabling immune complexes to evoke IL-10 production by human monocytes (72). Examples of reported linkages between complement and APC TLR effects on elicited T cell responses include attenuation of T cell priming/migration in C3^−/− mice in response to influenza virus (73), impairment of CD8 T cell expansion in C3^−/− mice during viral infection (74), more IL-6, TNF-α, and IL-1β (70) in response to TLR4 and TLR9 ligations in Daf1^−/− mice (26), and increased cytokine responses derive from increased locally produced C5a/C3a and potentiated DC C5ar1/C3ar1 signal transduction (27, 28, 31). Our findings suggest that reinterpretation of these observations is required and implicate TLR-induced DC-complement–dependent mechanisms.

TLR stimulation or bacterial infection (13, 22, 23) at the time of transplantation prevents costimulatory blockade (MR1)-induced transplant tolerance. Our results provide mechanistic insight by showing that CpG-induced acceleration of allograft rejection is mediated by TLR-induced immune cell autocrine C3ar1/C5ar1 signaling that augments T_effs and downregulates Foxp3 expression in T_regs. Our finding that TLR9-induced T_reg inhibition is C3ar1/C5ar1 dependent also explains previous reports that TLR9-induced enhancements of T cell responses in vivo reflect MyD88-dependent blockade of T_reg suppression (12). Our results also explain other reports of diminished effects of TLR ligations in mice deficient in complement components (75, 76). Because emerging evidence suggests that the sterile inflammation that occurs early posttransplant is triggered by DAMP/TLR ligations (15, 77), our findings raise the possibility that peritransplant blockade of C3ar1/C5ar1 signaling could attenuate the deleterious effects that early inflammation has on alloimmunity and, thereby, improve graft survival and function.

In summary, our data position immune cell–derived C3a/C3ar1 and C5a/C5ar1 interactions as crucial downstream intermediary steps between TLR stimulation and DC maturation that are required for the development and amplification of T cell immune responses. In addition to this fundamental insight, the data support the need for testing complement inhibitors (78) for human TLR-driven disease.

We thank Peter Boros, Jinhua Liu, and Yansui Li (Mount Sinai) for microsurgery support, Robert Fairchild (Cleveland Clinic) for scientific advice, and Kevin Kelley and the Mouse Genetics and Gene Targeting Core at the Icahn School of Medicine at Mount Sinai for mouse rederivation and production.

The authors have no financial conflicts of interest.