B Cell–Intrinsic STING Signaling Triggers Cell Activation, Synergizes with B Cell Receptor Signals, and Promotes Antibody Responses

Cyclic dinucleotides (CDNs), classified as alarmins, function in the innate immune response to host damage and infection. CDNs can be pathogen derived or synthesized in metazoans in response to injury (1–4). Previous findings have shown that CDNs have immunomodulatory activity. When administered via mucosal routes, CDNs significantly increase Ag-specific immune responses and provide protection in bacterial disease models (5–7). Cyclic-di-GMP (CDG), a common bacterially derived CDN, has been shown to have adjuvant activity, promoting balanced Th1, Th2, and Th17 responses, and robust mucosal and systemic Ab responses in mice (8–12).

The most well-characterized intracellular sensor of CDNs is the stimulator of IFN genes (STING), also referred to as MYPS, mediator of IRF3 activation (MITA), and Tmem173 (13–16). STING functions in the relay of signals generated upon cytosolic DNA sensing, coupling DNA detection to downstream production of proinflammatory cytokines (17). Subsequent to the CDN binding, STING becomes activated, leading to its translocation from the endoplasmic reticulum to the endoplasmic reticulum–Golgi intermediate compartments and engagement of TANK-binding kinase 1 (TBK1) and downstream IFN regulatory factor 3 (IRF3) and NF-κB pathways, leading to the production of proinflammatory cytokines such as type I IFN, IL-6, and TNF-α (15, 17–20).

The STING signaling pathway has been shown to be indispensable for CDN enhancement of Ag-specific Ab responses in vivo (12). STING function in this context is reportedly independent of type I IFN but is partially dependent on TNF-α (12, 21). The mechanisms by which STING signaling enhances Ab responses are still unclear. A recent study suggests that the answer may lie in part in its ability to promote Ag uptake by CD11c⁺ cells. Deletion of STING in CD11c⁺ cells resulted in reduced Ab responses following CDN/Ag immunization (22).

STING is highly expressed in B cells (13). In work described in this study, we examined the direct effects of CDNs on B cells and the role of STING in the responses observed. We further explored the importance of B cell–intrinsic STING in in vivo Ab responses promoted by CDNs. Results demonstrate that B cells are activated by CDNs in vitro and in vivo, and this response is STING dependent. Although largely cell intrinsic and cytokine independent, responses can be modulated by type I IFN. Importantly, Ag- and CDN-induced signals act synergistically to stimulate B cell activation. Finally, B cell–intrinsic STING is required for optimal CDG adjuvant effects on in vivo Ab responses to thymus-dependent Ags.

Eight- to twelve-week-old mice were used for all experiments. STING^flox/flox mice were generated by deleting the neo cassette from Tmem173^<tm1Camb> mice (17). This was achieved by crossing Tmem173^<tm1Camb> with an FLP1 recombinase line [B6; SJL-Tg(ACTFLPe)9205Dym/J]. Total STING knockout (KO) mice were generated by crossing STING^flox/flox mice with a Cre delete line [B6.C-Tg(CMV-cre)1Cgn/1]. B cell–targeted STING KO mice were generated by crossing STING^flox/flox mice with mb-1 Cre mice (23). C57BL/6 mice were used as wild-type (WT) controls during CDN/OVA immunization studies. Separate experiments showed that STING^flox/flox mice yield similar responses to CDN/OVA immunization as C57BL/6 mice (Supplemental Fig. 1). In some instances, MD4 B cell (anti–hen egg lysozyme [HEL]) Ag receptor–transgenic mice (24) were used as B cell donors. MD4 mice were crossed with total STING KO mice to generate MD4 STING KO mice. Congenically marked CD45.1 C57BL/6 (B6.SJL-Ptprc^aPep^b/BoyJ) B cells were used as WT controls in coculture experiments. Mice deficient in the TNF-α receptor (TNFAR) (B6.129S-Tnfrsf1a^tm1lmx Tnfrsf1b^tm1mx/J) and mice deficient in the receptor for type I IFNs (IFNAR) [B6(Cg)-Ifnar1^tm1.2Ees/J] were used to study cytokine dependence, and IFNAR KO mice were crossed with total STING KO mice to generate IFNAR STING double KO (DKO) mice for in vitro culture experiments. All mice, other than Tmem173^<tm1Camb>, were originally purchased from The Jackson Laboratory. Mice were housed and bred in the Animal Research Facility at the University of Colorado Anschutz Medical Campus and National Jewish Health. All experiments were performed in accordance with the regulations and approval of University of Colorado and National Jewish Health Institutional Animal Care and Use Committees.

Prior to intranasal (i.n.) immunization, mice were anesthetized with isoflurane in an E-Z Anesthesia system (Euthanex, Palmer, PA). Mice received three biweekly i.n. immunizations administered dropwise at 40 μl/nostril endotoxin-low OVA (20 μg/mouse) in PBS with or without CDG (5 μg/mouse, CDG; BioLog Life Science Institute, Bremmen, Germany). Endotoxin was removed (<0.1 endotoxin unit) from OVA (Sigma-Aldrich, St. Louis, MO) as described (25), and in some cases endotoxin-low OVA was purchased (BioVendor, Asheville, NC). Serum was collected at the indicated time points, and nasal passages were lavaged at the final time point with 1 ml of ice-cold PBS. To examine the general ability of mice to mount Ab responses, mice were immunized i.p. with OVA–NP_4.5 (20 μg/mouse) in PBS alone or in Alhydrogel (200 μg alum vaccine adjuvant; InvivoGen, San Diego, CA), with Ficoll-NP₄₀ (25 μg/mouse; Biosearch Technologies, Novato, CA), LPS–NP_0.6 (5 μg/mouse; Biosearch Technologies), or 5 × 10⁷ SRBC (Colorado Serum, Denver, CO). Serum was collected at the indicated time points.

For detection of OVA-specific Abs, microtiter plates were coated overnight with 10 μg/ml OVA in PBS at 4°C and blocked with 2% BSA in PBS–0.05% Tween 20 for >1 h at room temperature. For detection of NP-specific Abs, microtiter plates were coated with 10 μg/ml BSA–NP₂ or BSA–NP₁₇ in PBS and blocked with 2% BSA in PBS–0.05% Tween 20. Serial dilutions of mouse serum in PBS were added and incubated overnight at 4°C. IgG Abs were detected with goat anti-mouse IgG–HRP (SouthernBiotech, Birmingham, AL), IgM Abs were detected with goat anti-mouse IgM–HRP (Southern Biotech), IgA Abs were detected with goat anti-mouse IgA–HRP (Southern Biotech), IgG1 Abs were detected with goat anti-mouse IgG1–HRP (Southern Biotech), and IgG2a/c Abs were detected with goat anti-mouse IgG2a/c–HRP (Southern Biotech). Between all steps, the plates were washed four times with PBS–0.05% Tween 20. The ELISA was developed with TMB Single Solution (Invitrogen, Carlsbad, CA), and the reaction was stopped with 1N H₂PO₄ (Sigma-Aldrich). The OD was determined at 450 nm using a VersaMax Microplate Reader (Molecular Devices, Sunnyvale, CA) and the data were analyzed with Softmax software (Molecular Devices). The half-maximum reciprocal of dilution was calculated by using the reciprocal of the dilution factor at a set half-maximum value for each group at each time point. In order to normalize the assays performed on different days, a single serum sample collected on day 35 from the WT mice that were immunized with OVA in the presence of CDG was run with each group.

Serum was serially diluted (2-fold) in PBS in V-bottom microtiter plates in a 50-μl volume. Fifty microliters of 0.5% SRBC suspension in PBS was then added to each well, and plates were incubated at 37°C for 1 h. The hemagglutination titer was defined as the reciprocal value of the highest serum dilution at which hemagglutination of SRBC was detected.

Single-cell suspensions of splenic cells were prepared, and in some experiments, RBCs were lysed using ammonium chloride Tris. In most experiments, live cells were isolated using a Lympholyte-M kit (Cedarlane, Burlington, ON, Canada), and B cells were purified by CD43 negative selection using MACS Miltenyi MicroBeads (Miltenyi Biotec, San Diego, CA). Resultant populations were routinely >97% B cells, based on B220 staining and FACS analysis. Ex vivo B cells and splenocytes were cultured in IMDM supplemented with 10% FBS, sodium pyruvate (1 mM), l-glutamine (2 mM), 1% penicillin–streptomycin, 2-ME (50 μM), HEPES buffer (10 mM), and 1% nonessential amino acids.

Ex vivo B cells or splenocytes were seeded (3 × 10⁵ cells/100 μl/well) in a 96-well flat-bottom plate and stimulated with various concentrations of CDG for the indicated time. Additional stimuli were used at the following concentrations: zymosan (Sigma-Aldrich) (10 μg/100 μl), IL-4 (supernatant from a J558L culture, 1:200 dilution), HEL (Sigma-Aldrich) (1 μg/100 μl), F(ab′)₂ goat anti-mouse IgM (Jackson ImmunoResearch, West Grove, PA) (1 μg/100 μl), LPS (1 μg/100 μl), 2′,3′ c-GMP–AMP (cGAMP) (c[G(2′,5′)pA(3′,5′)p]; BioLog Life Science Institute) (3 μg/100 μl), 3′, 3′ cGAMP [c-(ApGp), BioLog Life Science Institute] (3 μg/100 μl), CDA (ML RR-S2 CDA sodium salt; Sigma-Aldrich) (3 μg/100 μl). In coculture experiments, 3 × 10⁵ of each cell type was added per well in a total volume of 100 μl. In experiments using z-vad-fmk (Z-VAD-FMK; Invivogen), z-vad-fmk was added 1 h prior to stimulation at the indicated concentrations.

Mice were injected i.p. with CDG (125 μg/mouse) in 200 μl PBS or PBS alone. After 18 h, mice were sacrificed, spleens and peritoneal lavages were collected, and cells were assessed for upregulation of CD86. In some experiments, nonirradiated STING KO or WT mice were cotransferred with purified WT MD4 and MD4 STING KO B cells at a 1:1 ratio. Donor B cells were labeled with CellTrace Violet (Molecular Probes) at 5 μM (MD4 WT) or 1 μM (MD4 STING KO) for 4 min at room temperature prior to transfer. A total of 2–4 × 10⁶ B cells in 200 μl PBS were adoptively transferred by i.v. injection. Twenty-four hours posttransfer, recipient mice were injected i.p. with CDG (125 μg/mouse), HEL (20 μg/mouse) or CDG and HEL in 200 μl PBS, or with PBS alone. After 18 h, mice were sacrificed, and splenic cells were assessed for upregulation of CD86. In other experiments, peritoneal B cells were enriched from peritoneal lavage from WT and STING KO mice by staining cells with a mixture of biotinylated anti-F4/80 (BM8; eBioscience), anti-CD3ε (500A2; BD Biosciences, San Jose, CA), and anti–Gr-1(RB6-8C5; BD Biosciences) and purified via negative selection using anti-biotin magnetic bead separation (Miltenyi Biotec). These B cells were labeled differentially with CellTrace Violet and mixed 1:1, and 1 × 10⁶ cells were adoptively transferred into total STING KO mice by i.p. injection. One hour posttransfer, the mice were injected i.p. with CDG (125 μg/mouse) in 200 μl PBS or PBS alone. After 18 h, mice were sacrificed and peritoneal lavages were collected, and cells were assessed for upregulation of CD86.

Single-cell suspensions of splenic cells were prepared, and RBCs were lysed using ammonium chloride Tris. Cells were resuspended in PBS containing 1% BSA and 0.02% sodium azide, or cells were taken directly out of cell culture and incubated with optimal amounts of Abs. For analysis of cell surface markers, cells were stained with Abs directed against the following molecules: B220 (RA3-6B2; BD Biosciences), CD86 (GL1; BD Biosciences), CD3ε (145-2C11; BD Biosciences), Gr-1 (RB6-8C5; BD Biosciences), CD23 (B3B4; BD Biosciences), CD93 (AA4.1; BD Biosciences), and CD45.1 (A20; BD Biosciences). Follicular B cells were defined as B220⁺CD23⁺CD93⁻, marginal zone cells were defined as B220⁺CD23⁻CD93⁻, and transitional B cells were defined as B220⁺CD93^med. Dead cells were gated out by excluding annexin V⁺–stained cells. FACS was performed using a BD LSRFortessa X-20 Cytometer (BD Biosciences) and analyzed using FlowJo software (Flowjo, Ashland, OR).

The statistical significance was determined using the unpaired Student t test. Two-way ANOVA was used to compare Ab responses (GraphPad Prism 5.0; GraphPad).

A recent report demonstrated that STING expression is required for CDG enhancement of Ab responses (12). It is unclear whether the bacterial CDN adjuvant, CDG, directly stimulates B cells and whether this requires STING. To test this, B cells were isolated from WT and total STING KO mice and stimulated with increasing concentrations of CDGs for 18 h before CD86 expression was measured. WT B cells exhibited upregulation of CD86 expression that was dose dependent, whereas increasing concentrations of CDGs had no effect on activation of STING KO B cells (Fig. 1A, 1B). Kinetic analysis revealed a significant fold change in CD86 expression as early as 13 h poststimulation (data not shown). To exclude the possibility that STING KO B cells have an inherent defect in the ability to upregulate CD86 following stimulation, B cells were stimulated with the STING-independent IL-4, and CD86 expression was assessed. The resultant responses did not require STING (Fig. 1A, 1C). Taken together, these data demonstrate that B cells have the ability to become activated upon direct stimulation with CDGs in vitro and that the STING signaling pathway is essential for this activation.

We observed that stimulation of B cells in vitro with CDGs led to a loss of viability of STING-sufficient B cells but not in STING-deficient B cells. STING has been shown to promote cell death in several biological systems, and CDNs have also been shown to promote cell death, including in B cells (26–31). The loss of cell viability increased with increased CDG concentration and occurred only in WT B cells (Fig. 1D). Inhibition of caspase activation by preincubation with z-vad-fmk before CDG stimulation inhibited cell death in a dose-dependent manner (Fig. 1E). Surprisingly, z-vad-fmk prevented CD86 upregulation in a dose-dependent manner (Fig. 1F), demonstrating that both STING-mediated cell death and B cell activation are dependent on caspase activation. The observed inhibition of CDG-induced CD86 upregulation by z-vad-fmk appears specific to STING-mediated upregulation of CD86, and not an off-target effect of the inhibitor, because at none of the z-vad-fmk concentrations used was the IL-4–induced response inhibited (Fig. 1G, 1H).

CDN-induced STING-mediated induction of B cell death seemed inconsistent with a positive role for STING in B cell responses. We hypothesized that perhaps CDN and BCR signals might be complementary, resulting in enhanced cell activation and no cell death. To test this possibility, HEL-specific B cells were isolated from STING-sufficient or -deficient BCR transgenic MD4 mice. B cells from MD4 STING–sufficient (WT MD4) and MD4 STING–deficient mice (STING KO MD4) were stimulated with HEL or CDGs separately or in combination. Whereas both STING-sufficient and STING-deficient MD4 B cells upregulated CD86 after Ag (HEL) exposure, only MD4 STING–sufficient B cells became activated in response to CDG stimulation, and this response was STING dependent (Fig. 2A). Stimulation of STING-sufficient MD4 B cells and not STING deficient MD4 cells with CDG in the presence of their Ag resulted in synergistic upregulation of CD86 (Fig. 2A). We also observed a partial protection of CDG-induced cell death in these experiments (Fig. 2B). Comparable results were obtained when we stimulated WT and STING KO B cells with CDGs or anti-IgM alone or in combination for 18 h. We found that in STING-sufficient B cells, CDGs and F(ab′)₂ anti-IgM stimulation were synergistic in the induction of CD86 upregulation. This did not occur in STING KO B cells (Fig. 2C). To test whether this is unique to CDG or if it is a general feature of STING biology, we tested three additional STING ligands, including the cGAMP synthase (cGAS) product 2′,3′ cGAMP, for their ability to activate B cells and synergize with BCR stimulation. All three STING ligands induced significant CD86 upregulation and synergized with BCR stimulation to further upregulate CD86 expression (Supplemental Fig. 2). Together, these data suggest that the BCR and STING signaling pathways work cooperatively to induce B cell activation and that BCR stimulation can provide alternative signals that mitigate death signals induced by activation of the STING signaling pathway.

Previous studies have shown that B cell upregulation of costimulatory molecule expression following exposure to pathogen-associated molecular patterns, such as TLR7 ligands, can be indirect, resulting from autocrine stimulation by type I IFN (32). To test if CDG-stimulated B cells secrete cytokines that can cause upregulation of CD86 on B cells, we cocultured STING-deficient B cells with STING-sufficient B cells or STING-sufficient splenocytes (Fig. 3A). Although CDG did not induce upregulation of CD86 by STING-deficient B cells cultured alone, when these cells were cocultured with WT B cells or WT splenocytes, a modest but significant upregulation of CD86 on STING-deficient B cells was seen. Whereas the cocultured WT B cells died after exposure to CDG, we did not observe a change in viability of the STING-deficient B cells. These findings demonstrate CDG induces production of a factor, possibly type I IFN, capable of stimulating upregulation of CD86 expression by B cells.

Type I IFNs have been shown to influence B cell survival, activation, and proliferation (33) and, therefore, we examined the role of type I IFN in CDG-induced B cell activation by using B cells from IFNAR KO mice. Because TNF-α has previously been shown to mediate, in part, the CDG adjuvant effect (12), we also assayed B cells from TNFAR KO mice. Purified B cells from WT, STING KO, IFNAR KO, and TNFAR KO mice were stimulated for 18 h with CDG and CD86 expression, and B cell survival (Fig. 3B) was assayed. Although TNFAR KO B cells respond slightly better to CDG than WT B cells, CD86 upregulation was significantly lower in IFNAR KO B cells. B cells that were unable to respond to type I IFN or TNF-α died with a similar frequency as WT B cells following CDG exposure. These results suggest that STING-mediated type I IFN production is partly responsible for the observed B cell activation following CDG stimulation, acting in an autocrine or paracrine manner, but these cytokines are not involved in CDG induction of cell death. To further examine this, we set out to determine whether type I INF is solely responsible for the observed CD86 upregulation by STING KO B cells following CDG stimulation in the coculture experiments. First, we verified that STING KO B cells do not have an inherent defect in their ability to respond to type I IFN. WT and STING KO B cells that were stimulated with IFN-β exhibited similar levels of upregulation of CD86 expression (Fig. 3C). Next, we generated STING and IFNAR DKO mice and compared responses of STING/IFNAR DKO B cells and STING KO B cells under similar coculture conditions as in Fig. 3A. When cocultured with WT B cells, the CD86 upregulation that was observed on STING KO B cells following CDG stimulation was almost completely absent in DKO B cells (Fig. 3D). When cocultured with WT splenocytes and stimulated, DKO B cells underwent a reduced CD86 upregulation response compared with STING KO B cells, although significant CD86 upregulation was detected. These results show that whereas B cell–intrinsic effects are mainly responsible for the observed increase in costimulatory molecule expression, type I IFN makes a significant contribution to this upregulation. Additional soluble factors produced by non–B cells may also play a minor role in promoting upregulation.

Because type I IFN was found to contribute, albeit modestly, to the observed CDG-induced B cell activation in vitro, we wanted to determine whether the observed synergistic effect of BCR stimulation with CDG stimulation on costimulatory molecule expression is a B cell–intrinsic effect of STING signaling or if it is the consequence of type I IFN signaling. Although IFNAR KO B cells respond to CDG stimulation with reduced CD86 upregulation, the synergistic effect of BCR stimulation with CDG stimulation was comparable to that observed in WT B cells (Fig. 3E). Thus, this synergism is a consequence of cooperative actions of STING and BCR signaling pathways within the B cell.

As demonstrated above, CDG activation of isolated B cells in vitro is largely type I IFN independent. However, the relative roles of the B cell–intrinsic versus cytokine-mediated responses in in vivo activation of B cells is unclear. Because many cell types are likely to produce type I IFN in response to CDG in vivo, cytokines seem likely to play a more prominent role there. To determine whether B cells become activated in vivo following peritoneal CDG administration and whether this response is STING dependent, WT and STING KO mice were injected i.p. with 125 μg of CDGs, and splenic B cell upregulation of CD86 was assessed. In STING-sufficient mice injected with CDGs, B cells exhibited significant upregulation of CD86, whereas no change in expression of CD86 was seen in STING KO mice (Fig. 4A, 4B). Thus, STING is required at the organismal level to support CDG-induced B cell activation in vivo.

It is well known that B cell subpopulations respond differently to innate immune stimuli. To determine which B cell subpopulations respond to CDG stimulation in vivo, we assessed CD86 expression on splenic follicular, marginal zone, and transitional B cells following CDG treatment. We observed similar upregulation of CD86 on all B cell subpopulations analyzed (Fig. 4A–E).

To determine whether type I IFN is required for systemic CDG induction of B cell activation in vivo, IFNAR KO mice were also injected with CDG, and upregulation of CD86 expression was assessed. We observed no upregulation of CD86 by B cells in IFNAR KO mice (Fig. 4A, 4B). Thus, under the conditions used, type I IFN is required for CDG-induced activation of splenic B cells in vivo.

To directly test whether there is a requirement for STING expression in B cells for systemic CDG induction of B cell activation in vivo, we generated B cell–targeted STING KO mice by crossing STING^fl/fl mice to mb1-Cre mice. Consistent with a role of STING-dependent type I IFN production by cells in the peritoneum and indirect activation of B cells by type I IFN, CD86 expression was significantly increased on B cells from B cell– targeted STING KO mice (Fig. 4F).

It seems likely that inability to activate B cells in vivo via the type I IFN–independent pathway is the consequence of inability to achieve stimulatory local concentrations of CDG. Work by others has shown that rapid diffusion of small molecules from the site of injection into the blood limits their direct potency but causes systemic responses (21). To further test this possibility, we explored the ability of i.p. injection of CDG to induce CD86 expression by the cytokine-independent pathway, reasoning that by this approach, we could achieve high local CDG concentrations. Peritoneal B cells from WT and STING KO mice were isolated, loaded with different concentrations of CellTracker Violet to allow later interrogation of different populations, and cotransferred at a 1:1 ratio in STING KO mice by i.p. injection. One hour later, mice were given an i.p. injection of CDG or PBS. After 18 h, we observed upregulation of CD86 on WT but not STING KO peritoneal B cells in STING KO mice that received CDGs (Fig. 4G), demonstrating that B cells can be activated by CDG in vivo by mechanisms that do not require type I IFN signaling. Taken together, these data suggest that both B cell–intrinsic and –extrinsic STING-mediated responses contribute to B cell activation following CDG stimulation in vivo.

Based on our previous observations utilizing coculture, we sought to determine the requirements for B cell–intrinsic and –extrinsic STING in activation of Ag-specific B cells following in vivo CDG exposure. To test this, differentially labeled MD4 STING–sufficient and MD4 STING KO B cells were isolated and cotransferred at a 1:1 ratio i.v. into both WT and STING KO recipient mice (Fig. 4H). These mice were then i.p. injected with PBS, CDG, HEL, or a combination, and changes in CD86 expression by B cells were assessed. STING-sufficient MD4 B cells transferred into WT recipients exhibited upregulation of CD86 after separate CDG and Ag exposure, and a synergistic increase in CD86 expression was seen after exposure to both Ag and CDG (Fig. 4I–K). Cotransferred STING-deficient MD4 B cells underwent similar responses. When transferred into STING KO mice, both MD4 STING–sufficient and –deficient B cells exhibited significant increases in upregulation of CD86 expression following exposure to Ag. In contrast, exposure of MD4 STING–sufficient B cells to CDG did not result in significant upregulation of CD86; however, we did observe superadditive responses to Ag and CDGs by MD4 STING–sufficient B cells. Combined with our in vitro observations, these results suggest that cytokines that are produced in vivo upon CDG exposure, likely by other cell types, can activate B cells and act synergistically with Ag stimulation. However, the synergistic responses of STING-sufficient B cells in STING KO recipients suggest that cytokine independent responses to CDG can cooperate with BCR signals to promote B cell activation.

Before testing the requirement of B cell–intrinsic STING in the Ab response following CDG immunization, we wanted to test the ability of STING-deficient mice to mount Ab responses to different Ags. As shown in Fig. 5A, significantly reduced Ab responses were observed in STING KO mice after immunization with Ficoll-NP₄₀, whereas responses against a model type 1 T-independent Ag, LPS–NP_0.6, was normal (Fig. 5B). STING-deficient mice also responded equally well to a T-dependent Ag, OVA–NP_4.5 in alum (Fig. 5D–H). This included processes such as affinity maturation (Fig. 5F) and class switching (Fig. 5G–H) associated with a germinal center reaction. Interestingly, when we used a different T-dependent Ag, SRBC, we did observe a statistically significant reduction in the ability of STING KO to mount Ab responses (Fig. 5C). This may be due to the repetitive nature of the polysaccharides on SRBC, which may partly act as a type 2 T-independent Ag.

Next, we tested whether B cell–intrinsic STING is required for CDN-enhanced Ag-specific Ab production. We immunized WT, STING KO, and B cell–targeted STING KO mice with OVA or OVA plus CDG i.n. Mice were boosted twice, 2 wk apart after the first immunization, and bled periodically for the duration of the experiment (Fig. 6A). WT mice produced robust levels of anti-OVA IgG throughout the duration of the experiment with peak Ab levels seen at day 42 (Fig. 6B). As expected, there was no detectable production of anti-OVA IgG Ab by the total STING KO mice at any time point throughout the experiment (Fig. 6B). When we assessed the requirement for B cell expression of STING in CDN enhancement of Ab production, we found that mice with B cells that lacked STING produced less anti-OVA IgG (Fig. 6B). This was reflected in all isotypes but most pronounced in the IgG1 anti-OVA response (Fig. 6C, 6D). Multiple groups have reported that CDNs promote potent systemic STING-dependent, Ag-specific, mucosal responses that can result in protective immunity in several infection models (6, 22, 34). To determine the potential role of B cell–intrinsic STING expression in CDN-enhanced mucosal response, we assessed differences in OVA-specific IgA Ab titers in the nasal lavage on day 59 postimmunization between WT, STING KO, and the STING B cell–targeted KO mice. As expected, we found that mice lacking STING expression in all tissues failed to produce OVA-specific IgA Abs (12) (Fig. 6E). IgA anti-OVA responses in mice that lacked STING in their B cells was reduced by ∼75%. Taken together, these data demonstrate that STING is required in the B cell compartment for optimal CDN-enhanced Ab production. Based on previous experiments, this must reflect a requirement for a cytokine-independent STING function in B cells.

The reduced IgG1 anti-OVA response observed in B cell–targeted STING KO mice immunized with OVA/CDG from day 28 onwards (Fig. 6C) suggests that in these mice there is a change in B cell–T cell interaction. However, because total IgG anti-OVA and all isotypes are significantly reduced at day 59, we wanted to exclude the possibility that this may reflect a defect in B cell–targeted STING KO mice in their ability to maintain plasma cell responses. B cell–targeted STING KO mice and control mice were immunized with OVA–NP_4.5/alum, and total IgG and IgG1 anti-NP responses were assessed throughout the duration of the experiment (up to 63 d postimmunization). We found no significant differences in the magnitude of the IgG and IgG1 anti-OVA Ab response (Fig. 6F–G). These results suggest that the reduction in Ag-specific Ab responses observed in B cell–targeted STING KO mice is specific to CDG adjuvant activity and not because of a defect inherent in all B cell Ab responses.

Although the importance of the STING signaling pathway has been well characterized in immune cell types such as macrophages and dendritic cells (12, 17, 35), there is little known about the role of STING signaling in B cell biology and function. Our results clearly demonstrate that CDNs directly activate B cells, and this response is STING dependent, causing upregulation of CD86 and cell death (Fig. 1). In addition, this activation was partially the consequence of a positive forward-feeding loop involving production of type I IFN (Fig. 3B) and possibly smaller contributions of other cytokines (Fig. 3D). In our in vitro experiments, direct activation by CDNs was responsible for the majority of the observed activation, as evident from experiments using IFNAR KO B cells (Fig. 3B) and cocultures comparing STING IFNAR DKO B cells with STING KO B cells (Fig. 3A, 3D). In vivo, under the conditions examined, secondary B cell activation by type I IFN (33) was mainly observed, presumably because many tissues were responding to CDNs by IFN production in vivo (Fig. 4A–F). As previously reported, the rapid dissipation of CDNs in vivo likely resulted in reduced local exposure of B cells (21). Consistent with this line of reasoning, when we conducted experiments assessing activation of peritoneal B1 and B2 B cells at the site of injection, we observed B cell activation dependent on B cell–intrinsic STING activity (Fig. 4G). Importantly, we observed that even when CDG did not reach levels sufficient in vivo to activate STING-competent B cells, there was clear cooperativity between BCR and STING signaling (Fig. 4H–K), suggesting that there are lower CDN threshold requirements for this synergism.

To what degree BCR–STING signal cooperativity contributes to CDG enhancement of Ab responses is unclear. Several groups (12, 21) have demonstrated that type I IFN signaling is dispensable for the adjuvant effect of CDG, suggesting that on the B cell level, CDG acts directly. Recently, Blaauboer et al. (22) provided compelling evidence that CD11c⁺ dendritic cells take up and present Ag more efficiently after CDG/Ag immunization and induce more potent T cell responses. Accordingly, targeted deletion of STING in CD11c⁺ cells resulted in reduced but not absent Ab responses after CDG/Ag immunization. B cell–targeted deletion of STING resulted in a significant reduction in Ab responses after CDG/Ag immunization (Fig. 6B–E). This could be due to a loss of the synergistic effects of CDG and Ag on CD86 expression on Ag-specific B cells that may make cognate B cell–T cell interaction less efficient.

Although STING is important for optimal Ab responses to type 2 T independent–Ags (Fig. 5A), we found no evidence that the quantity or quality of T cell–dependent responses were affected upon loss of STING expression (Figs. 5D–H, 6F, 6G), suggesting that the observed changes in Ab responses to CDG/Ag immunization of B cell–targeted STING KO mice are specifically due to differences in the response to CDG adjuvant activity. The fact that STING signaling and BCR signaling directly cooperate in B cell signaling (Fig. 2A, 2C, Supplemental Fig. 1) could have effects on B cell behavior beyond regulation of CD86 expression. In T cells, STING signaling in conjunction with TCR signaling was recently shown to negatively affect T cell proliferation (26, 36). To what extent STING signaling further alters Ag-driven B cell responses is subject of further investigation. Much like has been described for TLR9 and BCR signaling (37), BCR and STING signaling could cooperate to promote activation of autoreactive B cells and infected B cells (e.g., EBV-infected B cells).

Another effect of CDN exposure on B cells is STING- and caspase-dependent death (31). We confirmed these findings (Fig. 1D) and report for the first time, to our knowledge, that coordinating BCR signals could partially rescue CDG-induced cell death (Fig. 2B). Although we had a consistent partial rescue in experiments using MD4 B cells (Fig. 2B), only in two out of four experiments using WT B cells stimulated with anti-IgM F(ab′)₂ and CDG did we observe a similar partial protection against CDG-induced death (data not shown). Although we are unsure what caused this discrepancy, this potential protective effect could be of importance to consider when treating certain forms of B cell lymphomas with CDNs, as some may still receive BCR signals because of self-antigen interaction (38) or expression of EBV Ags (39). Previous work suggested that activation of the IRE–1/XBP-1 pathway could partially protect against CDN-induced apoptosis (31), although activation of this pathway often does not occur until 24 h post-BCR stimulation (40), suggesting that alternative pathways may exist. Cell death was mainly mediated by caspase activation, as demonstrated by the ability of z-vad-fmk to largely inhibit CDG-induced cell death (Fig. 1E). Unexpectedly, B cell activation was also inhibited in a dose-dependent manner by z-vad-fmk (Fig. 1F), whereas B cell activation by IL-4 was unaffected under these condition (Fig. 1G, 1H), suggesting that caspase activation may be part of STING-mediated cell activation. Caspases have previously been implicated in lymphocyte activation (41).

In this study, we describe requirements for B cell–intrinsic STING in CDG enhancement of Ag-specific Ab responses in vivo. Based on the importance of the STING signaling pathway in various immune responses, further evaluation will be important to assess the biological impact of expressing polymorphisms of STING in the human population, as is evident from initial studies (42). In this regard, polymorphisms of STING that confer hypofunctionality might lead individuals to infectious disease susceptibilities and poor Ab responses to CDN-based adjuvanted vaccines and poor antitumor immunity, whereas polymorphisms that confer hyperfunctionality might predispose individuals to development of autoimmunity and autoinflammatory diseases.

We thank Dr. Soojin Kim for managing our mouse colony, Vivian Duarte for technical help, Dr. Ross Kedl’s laboratory for the generous gift of the endotoxin-low OVA, Dr. Laurel Lenz’s laboratory for giving the IFNAR KO breeding pair and rIFN-β, and Dr. David Riches for the TNFAR KO breeding pair. We also thank the University of Colorado Immunology and Microbiology Department Flow Cytometry Core.

The authors have no financial conflicts of interest.