Effector memory T cells induce innate inflammation by triggering DNA damage and a non-canonical STING pathway in dendritic cells

SUMMARY

Cognate interaction between CD4⁺ effector memory T (T_EM) cells and dendritic cells (DCs) induces innate inflammatory cytokine production, resulting in detrimental autoimmune pathology and cytokine storms. While T_EM cells use tumor necrosis factor (TNF) superfamily ligands to activate DCs, whether T_EM cells prompt other DC-intrinsic changes that influence the innate inflammatory response has never been investigated. We report the surprising discovery that T_EM cells trigger double-strand DNA breaks via mitochondrial reactive oxygen species (ROS) production in interacting DCs. Initiation of the DNA damage response in DCs induces activation of a cyclic guanosine monophosphate (GMP)-AMP synthase (cGAS)-independent, non-canonical stimulator of interferon genes (STING)-TNF receptor-associated factor 6 (TRAF6)-nuclear factor κB (NF-κB) signaling axis. Consequently, STING-deficient DCs display reduced NF-κB activation and subsequent defects in transcriptional induction and functional production of interleukin-1β (IL-1β) and IL-6 following their interaction with T_EM cells. The discovery of T_EM cell-induced innate inflammation through DNA damage and a non-canonical STING-NF-κB pathway presents this pathway as a potential target to alleviate T cell-driven inflammation in autoimmunity and cytokine storms.

In brief

Meibers et al. show that effector memory CD4⁺ T cells induce double-strand DNA breaks in dendritic cells during cognate interactions through production of mitochondrial ROS. Initiation of the DNA damage response pathway results in activation of the STING-TRAF6-NF-κB axis, resulting in production of innate cytokines and subsequent inflammation and pathology.

Graphical Abstract

INTRODUCTION

The innate immune system functions as a first line of defense against invading pathogens and orchestrates initiation of pathogen-specific adaptive immune responses. Coordinated interactions between the innate and adaptive arms of the immune system are critical for robust and tailored immune responses^1,2 that result in elimination of the microbial threat. Classically, information flows from the innate immune system to the adaptive immune system when antigen-presenting cells (APCs) sense microbial non-self via pattern recognition receptors (PRRs)^1,3 and subsequently prime naive CD4⁺ T cells against microbial antigens. This is primarily accomplished by dendritic cells (DCs) that provide critical signals to naive T cells through major histocompatibility complex (MHC)-peptide complexes, co-stimulatory molecules, and priming cytokines,^4,5 resulting in activation and differentiation of pathogen-specific T cells. While effector T cells are critical for clearing the primary infection, protection from secondary infections is facilitated by generation of long-lived central and effector memory T cells. Effector memory T (T_EM) cells not only respond rapidly when they interact with DCs that present cognate peptide-MHC complexes⁶ but also have the potential to induce innate inflammatory responses.⁷ This is a result of a reciprocal flow of information from T_EM cells that, upon recognition of cognate antigen, directly instruct myeloid cells to secrete proinflammatory innate cytokines, including interleukin-1β (IL-1β), IL-6, and IL-12.^8,9

While innate inflammation resembles the responses induced by microbial stimuli,⁹ memory CD4⁺ T cell-induced innate inflammation is independent of classic PRR signaling.^10,11 For example, cognate interactions between antigen-specific CD4⁺ memory T cells and DCs bypass the need for PRR or interferon α/β receptor (IFNAR) signaling during early innate immune responses against secondary influenza responses.⁷ Others have demonstrated detrimental effects of CD4⁺ T cell-mediated innate inflammation during autoimmunity.^12–14 Our previous work revealed that autoreactive CD4⁺ T_EM cells rapidly activate myeloid cells presenting self-antigens, leading to an innate cytokine storm that drives pathology during autoimmune inflammation.^8,9 T_EM cells induce DC activation through ligation of several tumor necrosis factor (TNF) receptor superfamily (TNFRSF) members. Specifically, T cell-derived TNF and FasL engage TNFR and Fas on DCs to induce synthesis and cleavage of IL-1β.⁸ In parallel, TNF and CD40L on T_EM cells engage TNFR and CD40 on DCs, resulting in a broad transcriptional program that drives secretion of innate inflammatory cytokines such as IL-6 and IL-12.⁹ Blockade of TNF-TNFR and CD40L-CD40 interactions mitigates the T_EM cell-driven innate cytokine storm and autoimmune pathogenesis, but it does not completely abrogate innate cytokine synthesis, suggesting that additional mechanisms are contributing to T_EM cell-induced innate inflammation.

While ligands of the TNF superfamily have a critical role in dictating T_EM cell-induced innate inflammation, it is unclear whether T_EM cells drive other DC-intrinsic changes that determine the magnitude of this innate response. Examination of the transcriptional profile of DCs following their interaction with T_EM cells indicated specific upregulation of the gene Tmem173, which encodes for stimulator of interferon genes (STING). Canonical activation of STING depends on upstream detection of DNA by cyclic guanosine monophosphate (GMP)-AMP synthase (cGAS).¹⁵ cGAS catalyzes the synthesis of cyclic GMP-AMP (cGAMP), which binds to STING in the endoplasmic reticulum (ER), leading to downstream activation of TANK-binding kinase 1 (TBK1), interferon regulatory factor 3 (IRF3), and nuclear factor kB (NF-κB).¹⁵ This results in predominant production of type I interferon (IFN-I) genes and, to a lesser extent, inflammatory cytokines that are dependent on NF-κB activation.¹⁶ STING can also function independent of cGAS during anti-viral responses, such as with the enveloped RNA virus influenza A, which can interact with STING through its hemagglutinin fusion peptide, triggering an IFN-I response.¹⁷ Also independent of cGAS, a STING-IRF7 pathway induced by DNA vaccines leads to innate cytokine production and downstream antigen-specific T and B cell responses.¹⁸ In addition, disruption of ER-related cofactors known to activate STING can trigger its prolonged activation, giving rise to sterile inflammatory diseases such as STING-associated vasculopathy with onset in infancy (SAVI).¹⁹ While cGAS-independent STING has been shown to induce IFN-I-mediated responses, the mechanisms and its role in driving NF-κB-mediated proinflammatory cytokines such as IL-1β and IL-6 remain largely unexplored.

Here, we investigated the role of DC-intrinsic STING in mediating innate inflammation following T_EM cell-facilitated cognate interactions. Absence of STING in DCs led to significant abrogation of NF-κB-mediated innate cytokine production induced by interacting T_EM cells. Perplexingly, cGAS-deficient DCs did not affect T_EM cell-instructed innate inflammation. Instead, we discovered that T_EM cells induce double-strand breaks (DSBs) and a subsequent DNA damage response (DDR) in DCs that was partially driven by mitochondrial reactive oxygen species (ROS). This interaction rapidly activated a cGAS-independent STING-TRAF6-NF-κB axis in DCs, which led to synthesis of the proinflammatory cytokines IL-1β and IL-6. Absence of STING led to significantly diminished chromatin accessibility and transcription of NF-κB-dependent target genes such as Il1b and Il6. Furthermore, STING-deficient animals displayed defects in production of innate inflammatory cytokines in vivo following polyclonal T cell activation. Our study has thus discovered a critical role of DC-intrinsic STING in sensing a T_EM cell-induced DDR to drive innate inflammation.

RESULTS

STING promotes myeloid cell production of IL-1β and IL-6 following T_EM cell-instructed activation

We have reported previously that the cognate interaction between DCs and T_EM cells leads to induction of transcriptional changes in DCs that resembles stimulation by a microbial ligand.⁹ This RNA sequencing dataset (GEO: GSE184608) includes CD11c⁺ bone marrow-derived DCs (BMDCs) that were purified by fluorescence-activated cell sorting (FACS) following 3 h of culture with T_EM cells in the presence or absence of soluble anti-CD3ε antibody (anti-CD3). FACS-sorted BMDCs treated with lipopolysaccharide (LPS), a Toll-like receptor 4 (TLR4) ligand, for 3 h were included for comparison of classic PRR-mediated DC activation. While there is an overlapping inflammatory transcriptional profile between BMDCs activated via LPS and T_EM cells⁹ (Figure 1A), we focused on delineating the differentially expressed genes between BMDCs activated by LPS versus T_EM cells. Interestingly, ER-resident transcripts related to cholesterol and sterol biosynthesis pathways were uniquely upregulated following T_EM cell-instructed DC activation (Figure S1A). While the finding of cholesterol biosynthesis genes being uniquely expressed by DCs following T_EM cell-instructed activation is intriguing, we were perplexed by the notable upregulation of Tmem173, which encodes the STING protein (Figures 1A–1C). STING plays a well-known role downstream of cGAS in cytosolic DNA sensing, inducing TBK1 and IRF3 phosphorylation, leading to an anti-viral IFN-I response.^15,16 There have also been reports showing a role of STING in NF-κB activation,^20,21 IL-1β and IL-6 production during metabolic dysregulation^22,23 and autoimmune inflammation caused by deficiency in the Ca²⁺ sensor stromal interaction molecule,¹⁹ or in patients with gain-of-function STING mutations leading to SAVI.²⁴

Figure 1. DC-intrinsic STING promotes innate cytokine production following interaction with T_EM cells.

(A and B) Heatmap and reads per kilobase per million mapped reads (RPKM) values of Tmem173 and genes up-regulated in FACS-sorted BMDCs after 3 h of culture with LPS (+LPS; 100 ng/mL) or T_EM cells with anti-CD3 (+T_EM +αCD3; 200 ng/mL) compared with unstimulated (BMDC) or co-culture with T_EM cells in the absence of anti-CD3 (+T_EM). Heatmap gene selection bias was determined by upregulated genes discovered by Gene Ontology enrichment analysis related to innate inflammatory pathways between DCs activated by LPS versus T_EM cells with anti-CD3.

(C) Relative expression of Tmem173 (qPCR) from sorted BMDC populations.

(D) Western blotting for pro-IL-1β in sorted WT or STING^gt/gt BMDCs following 1- or 3-h culture with WT T_EM cells and anti-CD3.

(E and F) Quantified %pro-IL-1β+ WT or STING^gt/gt BMDCs following culture with WT T_EM cells in the presence of anti-CD3 for 3 h.

(G and H) IL-1β or IL-6 was measured by ELISA in the supernatants of WT or STING^gt/gt BMDCs cultured with WT T_EM cells in the presence of anti-CD3.

(I and J) Quantified %pro-IL-1β+ WT or STING^gt/gt BMDCs cultured with OT-II TCR-transgenic T_EM cells in the presence of absence of OVA_323–339 peptide for 3 h.

(K and L) IL-1β or IL-6 was measured by ELISA in the supernatants of WT or STING^gt/gt BMDCs cultured with OT-II T_EM cells and OVA protein.

Error bars indicate SEM. In (A)–(C), n = 3 biological replicates; in (D), 3 independent experiments; in (E)–(H) and (I)–(L), n = 4–12 and 3–4 independent experiments, respectively; in (B) and (C), ordinary one-way ANOVA; in (F)–(H) and (J)–(L), unpaired t test.

While upregulation of STING (Figures 1A–1C) is not necessarily an indicator of its functional role, STING’s known role in driving NF-κB-dependent cytokines prompted us to investigate whether DC-intrinsic STING played a role in regulating the T_EM cell-induced innate inflammatory response. Using a previously described in vitro system of cognate interaction between T_EM cells and DCs, we compared wild-type (WT) with STING^gt/gt (Tmem173 Goldenticket mouse; contains an I199N missense mutant allele of the STING gene where no protein is detected²⁵) BMDCs and their ability to produce IL-1β. Polyclonal effector memory T helper 0 (Th0) T_EM cells were cultured with BMDCs in the presence or absence of anti-CD3. Cognate interaction induced by anti-CD3 led to robust synthesis of pro-IL-1β in WT DCs, which was defective in STING^gt/gt BMDCs (Figures 1D–1F). The two bands seen for pro-IL-1β are consistent with previous reports and potentially reflect post-translational modification.^26–28 Additionally, secreted mature IL-1β and IL-6 in the supernatants of the STING^gt/gt BMDC cultures at later time points were significantly reduced compared with supernatants from WT BMDC cultures (Figures 1G and 1H). Of note, we have shown previously that cleavage and release of mature IL-1β is dependent on Fas-dependent activation of caspase-8 in DCs during T_EM cell-mediated interactions.⁸ Additionally, we found no differences in the STING^gt/gt BMDCs’ capacity to synthesize pro-IL-1β or secrete IL-1β after LPS stimulation, indicating that the STING^gt/gt BMDCs are not inherently defective in producing inflammatory cytokines in response to other cell-extrinsic stimuli (Figures S1B–S1D).

STING^gt/gt spleen-derived DCs (sDCs) activated by T_EM cells phenocopy BMDCs and exhibit defective inflammatory responses compared with WT sDCs in response to T_EM cell interaction. (Figures S1E–S1H). To test the importance of STING in driving innate inflammation induced by antigen-specific T_EM cells, we cultured ovalbumin (OVA)-specific OT-II T cell receptor (TCR) transgenic²⁹ T_EM cells with WT or STING^gt/gt BMDCs in the presence or absence of endotoxin-free OVA protein or OVA_323–339 peptide. Moreover, optimal IL-1β and IL-6 production by DCs following cognate MHC-TCR interaction with antigen-specific T_EM cells also requires functional STING (Figures 1I–1L). Overall, these data demonstrate that STING-deficient DCs are defective in their ability to produce and secrete innate pro-inflammatory cytokines following interaction with T_EM cells.

STING regulates innate cytokine production independent of cGAS during T_EM cell-mediated DC activation

During canonical, viral DNA sensing, STING is activated by cGAMP that is produced by the upstream sensor cGAS.¹⁵ To test the potential role of cGAS in STING-dependent innate cytokine production during T_EM cell-instructed activation, we compared cGAS knockout (cGAS^−/−) with WT BMDCs cultured with T_EM cells and anti-CD3. Surprisingly, cGAS^−/− BMDCs and WT BMDCs expressed comparable levels of pro-IL-1β and secreted equivalent levels of IL-1β and IL-6 (Figures S2A–S2D), suggesting a cGAS-independent role of STING. In addition, expression of the cGAS gene (Cgas) in the previously described RNA sequencing (RNA-seq) was unchanged in T_EM cell-activated DCs, while it was significantly upregulated following LPS activation (Figure S2E), as reported previously.³⁰ Interestingly, cGAS^−/− BMDCs exhibited a slight but significant defect in expression of pro-IL-1β but secreted comparable amounts of IL-1β and IL-6 following activation by LPS compared with WT BMDCs (Figures S2F–S2I).

STING trafficking and subsequent degradation occurs during cGAS-dependent activation to evade excessive production of IFN-I.^31,32 We investigated whether DC-intrinsic STING degrades following T_EM cell-mediated activation. While canonical activation of STING by 5,6-dimethylxanthenone-4-acetic acid (DMXAA) stimulation (Figure S2J) led to its rapid degradation, western blotting displayed no signs of STING degradation in BMDCs activated by T_EM cells, even after 6 h of cognate interaction (Figure S2J). Lack of degradation suggests that STING does not translocate away from the ER-Golgi apparatus,³³ as reported following cGAMP mediated activation. Absence of cGAS involvement, in addition to these data, suggested an alternate, non-canonical role of STING during T_EM cell-induced innate inflammation.

T_EM cell-mediated activation induces nuclear DSBs and a subsequent DDR in DCs

As described previously, canonical STING activation involves detection of cytosolic double-stranded DNA via cGAS.³⁴ Because cGAS was dispensable in STING-dependent induction of inflammatory cytokines, we aimed to uncover which upstream pathway was responsible for this STING-dependent T_EM cell-mediated innate response. Several reports discuss upstream mitochondrial (mtDNA) or nuclear DNA damage that leads to STING-NF-κB activation dependent and independent of cGAS activation. For example, mtDNA damage tends to activate a cGAS-dependent STING-NF-κB pathway because of leakage of mtDNA into the cytosol.^35,36 A recent study also described an IFN-independent signaling activity for the cGAS-STING pathway in driving DDR signaling following genotoxic insult.³⁷ Alternatively, during etoposide treatment of human cancer cell lines, a nuclear double-strand DDR orchestrated by ataxia telangiectasia mutated (ATM) serine/threonine kinase led to initiation of a STING-TRAF6 complex.²² Thus, we sought to investigate whether T_EM cells induce nuclear DNA damage and a subsequent DDR in DCs. To this end, we measured DNA damage in DCs using a neutral comet assay, a well-characterized assay to quantify DNA damage in individual cells.³⁸ Following stimulation, sorted BMDCs were permeabilized and denatured, followed by electroporation to separate supercoiled DNA from fragmented DNA, resulting in a comet tail that can be captured by microscopy. This assay led to the exciting finding that cognate interactions between BMDCs with T_EM cells led to DNA damage in individual DCs (Figures 2A–2C). Importantly, DCs stimulated with LPS produced few to no comet tails, indicating that T_EM cells, but not classic PRRs, have the unique ability to induce DSBs in interacting DCs (Figures 2B and 2C).

Figure 2. T_EM cells induce nuclear DSBs and a subsequent DDR in DCs.

(A and B) Comet assay visualization (A) and DNA fragmentation (B) denoted by comet tail length in FACS-sorted BMDCs after being cultured with T_EM cells with or without anti-CD3 for 3 h or BMDCs with or without LPS for 3 h. Scale bar, 100 μm.

(C) Number of cells with a comet tail length greater than 35 μm per 100 cells.

(D and E) H2AX phosphorylation in BMDCs with or without LPS or BMDCs cultured with T_EM cells with or without anti-CD3 for 3 h.

(F and G) H2AX phosphorylation in WT BMDCs cultured with naive CD4⁺ T cells (+nCD4 T) sorted from spleens of 6– 8-week-old C57BL/6J WT mice, expanded CD4⁺ T_Eff (+T_Eff ) cells, or expanded and rested CD4⁺ T_EM cells (+T_EM) with or without anti-CD3 for 3 h.

(H and I) ATM phosphorylation in BMDCs cultured with T_EM cells with or without anti-CD3 for 1 h.

(J) Western blotting for DDR proteins in sorted BMDCs following stimulation with T_EM cells with or without anti-CD3 for 1 h.

(K and L) H2AX phosphorylation in BMDCs cultured with T_EM cells and anti-CD3 for 3 h. BMDCs were pretreated with an inhibitor for ATM or ATR (ATRi) prior to addition of T_EM cells and anti-CD3.

Error bars indicate SEM. In (A)–(C), 2 independent experiments; in (E) and (I), n = 9–10; in (G), (K), and (L), n = 3–7; in (J), 3 independent experiments; in (B), (C), and (E), ordinary one-way ANOVA; in (G), two-way ANOVA; in (I), (K), and (L), one-tailed paired t test.

DNA damage generated by multiple upstream stimuli results in activation of a DDR that subsequently triggers either DNA repair or cell death.³⁹ Histone H2AX phosphorylation at Ser 139 (γH2AX)^40,41 is a highly sensitive marker for detecting nuclear DSBs and the subsequent repair of the DNA lesions. We therefore examined induction of γH2AX in DCs following their interaction with T_EM cells and discovered rapid H2AX phosphorylation that peaks within 3 h in BMDCs (Figures 2D, 2E, S3A, and S3B) and ex vivo-derived sDCs (Figures S3C and S3D). γH2AX was also found in BMDCs activated by TCR transgenic LCMV-specific (SMARTA)⁴² CD4⁺ T_EM cells in the presence of GP_61–80 peptide, albeit at a lesser intensity (Figures S3E and S3F), possibly because of a lower avidity of interaction. Importantly, initiating the DDR in DCs was unique to T_EM cells because neither naive (CD62L⁺ CD44⁻) nor effector (CD44⁺) CD4⁺ T cells had the ability to induce γH2AX in DCs (Figures 2F, 2G, and S3G). This further highlights the central role of T_EM cells as critical drivers of innate cytokine storms and autoimmune pathology.^8,9 Additionally, H2AX is not phosphorylated in BMDCs activated by LPS (Figures 2D and 2E) or other TLR-activating ligands, such as Pam3CSK4 (TLR2) and CpG (TLR9) (Figure S3H).

To further confirm activation of the DDR, we investigated phosphorylation of the ATM/CHK2 and ataxia telangiectasia and Rad3 related (ATR)/CHK1 axes, which are known to aid in nuclear repair following DSBs and single-strand breaks (SSBs), respectively.^43,44 While we detected phosphorylation of ATM and its substrate CHK2 in BMDCs activated by T_EM cells (Figures 2H–2J), phosphorylation of the ATR substrate CHK1 was absent (Figure 2J), strongly supporting the idea that activation of the DDR in T_EM cell-activated DCs was driven by formation of DSBs rather than SSBs. Additionally, experiments using specific chemical inhibitors of ATM or ATR activation revealed that ATM, but not ATR, aided in overall H2AX phosphorylation (Figures 2K and 2L). These data overall indicate that T_EM cell-mediated activation of DCs is fundamentally different from classic PRR activation pathways induced by microbial ligands.

A previous report suggests the possibility that STING itself can induce mitochondrial and nuclear damage,⁴⁵ indicating that STING may be playing a role upstream of DNA damage. To address this, we compared H2AX and ATM phosphorylation in WT and STING^gt/gt DCs and observed no difference in their capacity to phosphorylate H2AX or ATM (Figures S3I and S3J). These data allow us to conclude that the T_EM cell-mediated DDR is upstream of STING activation.

STING forms a complex with TRAF6 to promote NF-κB activation during T_EM cell-mediated DC activation

Because cGAS does not play a role in T_EM cell-mediated DC activation, and T_EM cells induce DSBs leading to activation of the DDR, we hypothesized that DC-intrinsic STING was being activated in a non-canonical fashion. Prior evidence examining the downstream effects of a chemically induced DDR in human epithelial cell lines suggested that non-canonical STING pathways can activate the transcription factor NF-κB following interactions with the ubiquitin E3 ligase TNF receptor-associated factor 6 (TRAF6),²² a protein that normally functions downstream of the TLR and IL-1R family of receptors,⁴⁶ to drive NF-κB-driven innate inflammatory cytokines.

To determine whether DC-intrinsic STING complexes to TRAF6 during T_EM cell-mediated DC activation, we pulled down DC-intrinsic STING via co-immunoprecipitation (coIP) assay and probed for TRAF6. Of note, STING-deficient CD4⁺ T_EM cells were used to ascertain that we were examining STING in DCs. This led to the clear finding that TRAF6 was indeed bound to DC-intrinsic STING following T_EM cell-mediated activation but not when STING was canonically activated by DMXAA (Figures 3A and S4A). We subsequently inspected STING-mediated activation of NF-κB by examining phosphorylation of the upstream kinases IKKα/β.⁴⁷ During T_EM cell-mediated interactions, we identified phosphorylation of IKKα/β in WT BMDCs, which was measurably compromised in STING^gt/gt BMDCs (Figures 3B and 3C). This was consistent in ex vivo-derived sDCs (Figures 3D and 3E), further supporting a role of STING-induced activation of NF-κB downstream of the DDR during T_EM cell-instructed DC activation.

Figure 3. T_EM cells induce a non-canonical STING-TRAF6-NF-κB pathway in DCs.

(A) TRAF6 and STING protein expression following co-immunoprecipitation (coIP) of STING in WT BMDCs cultured with STING^gt/gt T_EM cells and anti-CD3 for 1 or 3 h. BMDCs stimulated with DMXAA for 1 h were used as a control for canonical STING activation.

(B–E) Histograms and median fluorescence intensity (MFI) of phosphorylated IKKα/β in BMDCs (B and C) and sDCs (D and E) after 1 h of T_EM with or without anti-CD3.

(F) Quantified %pro-IL-1β+ WT or STING^gt/gt BMDCs with or withoutTRAF6 sgRNA transfection following culture with WT T_EM cells in the presence of anti-CD3 for 3 h.

(G and H) IL-1β or IL-6 was measured by ELISA in the supernatants of WT or STING^gt/gt BMDCs with or without TRAF6 sgRNA transfection cultured with WT T_EM cells in the presence of anti-CD3.

Error bars indicate SEM. In (A)–(H), 3 independent experiments; in (B)–(H), n = 5–8; two-way ANOVA.

Because STING and TRAF6 appear to be interacting during cGAS-independent activation, and the known role of TRAF6 upstream of NF-κB activation,⁴⁶ we next examined the functional role of TRAF6 in DCs following T_EM cell-mediated interactions. To this end, we utilized CRISPR-Cas9 gene editing⁴⁸ to delete TRAF6 in WT and STING^gt/gt bone marrow precursor cells prior to expanding them into BMDCs (Figure S4B). We found that TRAF6-deficient DCs exhibited similar decreases in pro-IL-1β staining as well as IL-1β production, as seen in STING^gt/gt BMDCs (Figures 3F and 3G). Deletion of TRAF6 in STING^gt/gt BMDCs (TRAF6-STING double knockout) caused no additional decrease in IL-1β production (Figures 3F and 3G), suggesting that STING and TRAF6 participate in the same signaling pathway downstream of T_EM cell-mediated DC activation. Of note, IL-6 secretion was compromised in TRAF6 knockout cells in WT and STING^gt/gt DCs (Figure 3H), which is likely due to the additional role of TRAF6 downstream of CD40 signaling.⁴⁹ Additionally, H2AX phosphorylation was unchanged when TRAF6 was knocked out (Figure S4C), suggesting that STING and TRAF6 play a role downstream of DNA damage.

STING-directed NF-κB binding in open chromatin promotes transcriptional regulation of Il1b and Il6 following T_EM cell-mediated activation

Given the functional defect in cytokine synthesis and secretion in STING^gt/gt DCs, we asked how transcriptional regulation of DCs was affected in the absence of STING at an early time point following T_EM cell-mediated activation. Transcriptional profiling of WT and STING^gt/gt BMDCs via RNA-seq was performed on FACS-sorted CD45.2⁺ CD11c⁺ BMDCs following 1 h of T_EM cell-mediated cognate interaction (Figure S5A). We compared the transcriptional profile of WT and STING^gt/gt DCs cultured in the presence of T_EM cells either with or without the cognate anti-CD3 antibody. Principal-component analysis (PCA) revealed clear separation between control (+T_EM) and treatment (+T_EM +αCD3) groups as well as noticeable differences between WT and STING^gt/gt-activated DCs (Figure S5B). The transcriptional profile comparison between activated WT and STING^gt/gt DCs revealed elevated expression of 54 genes in WT DCs compared with STING^gt/gt, including several key proinflammatory cytokines (Il1b, Il6, Il1a, Il12b, and Tnf), an NF-κB subunit (Nfkbia), and migration-related genes (Cxcl1 and Ccr8) (Figures 4A–4C). Defective Il1b and Il6 transcripts in STING^gt/gt DCs was also confirmed through quantitative PCR (qPCR) after 1 h of interaction with T_EM cells (Figures 4D and 4E). These data coincide with the lack of pro-IL-1β protein seen in STING^gt/gt DCs after 1 h of stimulation (Figure 1D), suggesting that STING plays a critical early role in synthesis of innate proinflammatory cytokines during T_EM cell-instructed activation. Notably, WT DCs cultured with T_EM cells in the absence of anti-CD3 also revealed higher fragments per kilobase million (FPKM) values and relative expression of Il1b and Il6 transcripts compared with STING^gt/gt DCs, suggesting that steady-state interactions between DCs and T_EM cells might induce STING-dependent inflammatory gene expression.

Figure 4. Enrichment of NF-κB binding motifs and transcriptional regulation of Il1b and Il6 is directed by DC-intrinsic STING.

(A–E) Heatmap of differentially expressed genes and list of select genes upregulated in sorted WT BMDCs compared with sorted STING^gt/gt BMDCs after 1 h of cognate interaction with WT T_EM cells with or without anti-CD3. Genes shown have differential adjusted p values of less than 0.05 and Log₂FC greater than 1.0. FPKM values (RNA-seq) (B and C) or relative expression (qPCR) (D and E) of Il1b and Il6.

(F) Differentially accessible chromatin sites in WT versus STING^gt/gt DCs (ATAC-seq). Each row represents one genomic locus. Color bars indicate the strength of the ATAC-seq signal.

(G–I) Comparison of genome-wide transcription factor motif enrichment analysis in various categories of BMDC ATAC-seq peaks: unstimulated (+T_EM) versus 1-h-stimulated (+αCD3) WT (G), unstimulated versus 1-h-stimulated STING^gt/gt (H), and 1-h-stimulated WT versus STING^gt/gt (I). Each dot represents a motif from the indicated transcription factor family.

Error bars indicate SEM of biological triplicates. In (B)–(E), unpaired t test.

To further understand the differences in early gene expression, we performed ATAC-seq (assay for transposase-accessible chromatin with sequencing) to evaluate changes in chromatin accessibility of WT and STING^gt/gt DCs after 1 h of T_EM cell-instructed activation. Pairwise differential chromatin accessibility analysis between conditions revealed 1,139 regions with significantly stronger ATAC-seq peaks in activated WT DCs compared with the control, while STING^gt/gt DCs exhibited only a minor increase compared with its respective control (136 peaks) (Figures S5C and S5D). In accordance with steady-state differences seen in RNA levels of inflammatory cytokines, unstimulated (+T_EM) WT DCs had 109 regions with significantly stronger ATAC-seq peaks compared with unstimulated STING^gt/gt DCs (Figure S5C), suggesting that early activation of STING leads to opening of NF-κB target genes and subsequent inflammatory responses in DCs interacting with T_EM cells.

We next compared activated WT DCs with their STING^gt/gt DC counterparts and found 1,491 regions showing significantly increased accessibility in WT DCs (Figure 4F), suggesting that STING may play an important role in the downstream transcriptional response to T_EM cell activation. To better understand the role of STING in transcriptomic regulation, we used the Hyper-geometric Optimization of Motif EnRichment (HOMER) software package⁵⁰ to scan each set of differentially accessible peaks for enriched mouse transcription factor binding site motifs obtained from the Cis-BP database.⁵¹ Remarkably, we discovered significant enrichment of NF-κB motifs in activated WT DCs (+αCD3) compared with unstimulated (+T_EM) controls (Figure 4G), which was severely defective in the activated STING^gt/gt DCs compared with unstimulated controls (Figure 4H). Notably, there are substantial differences in the enrichment levels (−log₁₀ p value) of NF-κB motifs in WT (Figure 4G) versus STING^gt/gt DCs (Figure 4H). Direct comparisons between activated WT and STING^gt/gt DCs likewise revealed significant enrichment of NF-κB motifs in WT DCs that was completely absent in activated STING^gt/gt DCs (Figure 4I). There were no measurable differences in accessibility of activator protein-1 (AP-1), signal transducer and activator of transcription (STAT), total IRF, or IRF3-specific motifs in WT or STING^gt/gt DCs (Figures 4G–4I). Paucity of IRF3-accessible peaks may signify lack of IRF3 activation during T_EM cell-mediated DC activation, indicating an NF-κB-dominant response as a result of non-canonical STING activation. Together, these results indicate that DC-intrinsic STING enables NF-κB-controlled transcription via increased chromatin accessibility of NF-κB-bound regulatory elements during T_EM cell-mediated interactions.

Myeloid-intrinsic mitochondrial ROS (mROS) production activates the DDR following T_EM cell-mediated DC activation

Because of STING’s role as an early activator of T_EM cell-induced NF-κB and innate proinflammatory cytokines in DCs, we sought to determine which pathways upstream of STING-TRAF6 and the DDR could be initiating this T_EM cell-mediated response. To gain insights into factors initiating this response, we analyzed the top Gene Ontology Biological Process annotations related to NF-κB activation that were upregulated in both WT and STING^gt/gt activated DCs. We found several similar annotations between activated WT and STING^gt/gt DCs, including cellular response to LPS, negative regulation of cytokine production, c-Jun N-terminal kinase (JNK) cascade, and response to oxidative stress (Figure S5E). Based on our previous work, there is a large overlap in gene expression in DCs activated by T_EM cells versus LPS,⁹ suggesting why this biological process is highly upregulated. Following activation of DCs, negative regulation of cytokine production prevents excessive secretion of cytokines.⁵² Several studies have examined how STING activates the NF-κB axis following treatment with DNA-damaging agents.^22,53,54 DSBs are induced by many agents, including chemicals, toxins, UV light/radiation, and oxidative stress.⁵⁵ Because DCs in our experiments are not treated with chemicals or radiation, and there is enrichment for oxidative stress-NF-κB-related pathways (Figure S5E), we asked whether these DCs were undergoing oxidative stress induced by production of mROS.

ROS are highly reactive compounds that can induce localized cellular injury in the form of DNA damage.^56–58 Therefore, we examined whether T_EM cell-mediated activation induces DC-intrinsic mROS. Following T_EM cell-mediated activation, we observed significant mROS in WT BMDCs, which were absent in BMDCs responding to LPS stimulation (Figures 5A and 5B). To ensure that STING was not playing an upstream role in mROS production,⁵⁹ we compared mROS in WT and STING^gt/gt BMDCs and observed no differences (Figure 5C), suggesting a potential role of mROS upstream of non-canonical STING activation. To determine whether mROS triggered a DC-intrinsic DDR, we scavenged ROS via N-acetyl-coenzyme A (CoA) (NAC) and discovered a significant drop in γH2AX (Figures 5D and 5E) with no change in DC or T_EM cell death (Figures S6A–S6D). Farther downstream, ROS scavenging abrogated IL-1β and IL-6 secretion (Figures 5F and 5G), indicating that ROS promotes induction of a DDR and subsequent proinflammatory cytokine secretion in BMDCs activated by T_EM cells.

Figure 5. DC-intrinsic mROS production activates a DDR following T_EM cell-mediated activation.

(A and B) MitoSOX Red (mROS) staining in BMDCs stimulated with WT T_EM cells with or without anti-CD3 or BMDCs with or without LPS for 1 or 3 h.

(C) Quantification for MitoSOX Red staining in WT or STING^gt/gt BMDCs following stimulation with WT T_EM cells with or without anti-CD3 for 3 h.

(D and E) H2AX phosphorylation in BMDCs stimulated with WT T_EM cells with or without anti-CD3 for 3 h. BMDCs were treated with NAC prior to interaction with T_EM cells in the presence of anti-CD3.

(F and G) IL-1β and IL-6 measured by ELISA in the supernatants of WT BMDCs cultured with NAC 20 min prior to culture with WT T_EM cells with or without anti-CD3.

(H–K) MitoSOX Red (H and I) and H2AX phosphorylation (J and K) in BMDCs stimulated with WT T_EM cells with or withour anti-CD3 for 3 h. BMDCs and T_EM cells were pretreated with the neutralizing antibodies anti-TNF-α or anti-CD40L prior to addition of anti-CD3.

Error bars indicate SEM. In (A)–(I), n = 4–8 and 2–3 independent experiments, respectively; in (C), (I), and (K), two-way ANOVA; in (B), ordinary one-way ANOVA; in (E)–(G), one-tailed paired t test.

Because of only partial abrogation of H2AX phosphorylation following ROS scavenging, we decided to investigate other pathways known to play a role in T_EM cell-instructed DC activation. We have previously reported a requirement for TNF and CD40L on T_EM cells to activate TNFR and CD40 on DCs, driving secretion of innate inflammatory cytokines such as IL-1β,⁸ IL-6 and IL-12.⁹ Blocking TNF-TNFR and CD40L-CD40 interactions mitigates T_EM cell-driven innate cytokine storm and autoimmune pathogenesis. However, it does not completely abrogate the innate cytokine response. To determine whether the TNFR and CD40 pathway is working upstream of mROS, the DDR, and subsequent cytokine production, we examined the ability of TNFR and CD40 to induce mROS and γH2AX. With use of neutralizing antibodies, we found that blocking CD40-CD40L and TNF-TNFR interactions between T_EM cells and DCs results in no loss in mROS production (Figures 5H and 5I) or the DDR in DCs (Figures 5J and 5K). Although there are suggested roles for TNF in inducing ROS,^60–62 our data argue that TNF plays no role in initiation of the DDR by T_EM cells and downstream STING-related innate inflammation.

Absence of STING in vivo ameliorates T cell-induced innate inflammation

We have so far provided several molecular insights into how DC-intrinsic STING is activated in a non-canonical fashion during T_EM cell-induced innate inflammation in vitro. Therefore, we aimed to uncover the role of STING in vivo during innate inflammation by utilizing models of systemic, T_EM cell-mediated sterile inflammation. Administration of anti-CD3 is a simple but powerful way of investigating T cell-induced innate inflammation.^8,9,63 WT or STING^gt/gt mice were injected intravenously with anti-CD3 and then sacrificed 4 h later to assess pro-IL-1β production in myeloid cell populations in the spleen as well as IL-6 in the peripheral blood. We found a significant defect in pro-IL-1β synthesis in several myeloid cell populations in STING^gt/gt mice compared with WT mice (Figures 6A, 6B, and S7A–S7C). Additionally, we detected a significant defect in circulating IL-6 in STING^gt/gt mice compared with WT mice (Figure 6C), indicating an important role of STING in systemic, T cell-mediated inflammation in vivo.

Figure 6. Absence of STING in vivo ameliorates T cell-induced innate inflammation and autoimmune pathology.

(A) Quantification of pro-IL-1β in live, CD11c⁺, CD11b⁻, Ly6G⁻, CD90.2⁻, and CD19⁻ sDCs of WT or STING^gt/gt mice following intravenous anti-CD3 (2.5mg/kg) for 4 h.

(B) Quantified %pro-IL-1β+ CD11c⁺ DCs from (A) and other independent experiments.

(C) IL-6 was measured by ELISA from the serum of WT and STING^gt/gt mice following intravenous anti-CD3.

(D) Timeline of DT injections in WT FoxP3-DTR or FoxP3-DTR^ΔSTING mice for early markers of innate inflammation.

(E and F) MFI of γH2AX in splenic CD11b⁺ myeloid populations following 3 days of Treg cell depletion.

(G) Il1b and Il6 transcripts were analyzed via qPCR in splenocytes taken from WT FoxP3-DTR and FoxP3-DTR^ΔSTING mice on day 3 post DT injections.

(H) Timeline of DT injections for histopathology in WT FoxP3-DTR and FoxP3-DTR^ΔSTING mice.

(I and J) Splenic cell counts (I) and H&E staining (J) of lung sections and injury scores of WT FoxP3-DTR and FoxP3-DTR^ΔSTING mice on day 10 post DT injections. An arrow indicates neutrophil infiltration and alveolar septal thickening. Scale bar, 50 μm.

Error bars indicate SEM. In (A)–(C) and (E)–(G), n = 3–8 per group; in (I) and (J), n = 3–7 mice per group. Two independent experiments, unpaired t tests.

Finally, we sought to delineate the role of STING in a more defined systemic autoimmune mouse model. We have previously used the inducible Foxp3^ΔTR/GFP (FoxP3-DTR) mouse model to evaluate T cell-driven innate pathology by simply depleting FOXP3⁺ CD4⁺ regulatory T (Treg) cells.⁹ Depletion of FOXP3⁺ Treg cells initiates widespread activation of autoreactive T cells and infiltration of myeloid populations, resulting in systemic inflammation and subsequent tissue damage.⁶⁴ We first aimed to examine whether self-reactive T cell-mediated autoimmune inflammation leads to a DDR in myeloid cells in vivo. WT FoxP3-DTR mice were given 2 injections of diphtheria toxin (DT) to ensure Treg cell depletion and then sacrificed on day 3 (Figure 6D). This early time point allows autoreactive T cells to begin expanding and activating cells of the innate immune system. We found that activated CD11b⁺ myeloid populations express γH2AX 3 days post Treg cell depletion (Figures 6E and 6F). Examination of another major population of antigen-presenting cells, CD19⁺ B cells, revealed no H2AX phosphorylation (Figure S7D), indicating that this T cell-mediated DDR is restricted to myeloid cell populations.

Because we found γH2AX in WT FoxP3-DTR mice, we crossed FoxP3-DTR mice to STING^gt/gt mice (FoxP3-DTR^ΔSTING) to delete Treg cells in STING-deficient mice. These mice allow us to test the hypothesis that STING-deficient, FOXP3⁺ Treg cell-deficient mice would have significantly decreased innate proinflammatory cytokine production and overall tissue damage compared with WT mice lacking FOXP3⁺ Treg cells. We have demonstrated previously that autoreactive T cell-driven myeloid cell activation is one of the main causes of inflammatory innate cytokine production and subsequent organ pathology.⁹ Therefore, to determine the early-stage innate cytokine transcriptional profile, WT FoxP3-DTR and FoxP3-DTR^ΔSTING mice were given 2 injections of DT to ensure Treg cell depletion and then sacrificed on day 3 (Figure 6D). Accordingly, Il1b and Il6 transcripts were upregulated in the splenocytes of WT FoxP3-DTR mice, and there was a significant reduction of these transcripts in spleens of Treg cell-depleted FoxP3-DTR^ΔSTING mice (Figure 6G). Importantly, we found similar proportions of CD4⁺ and CD8⁺ naive and effector T cell populations between WT FoxP3-DTR and FoxP3-DTR^ΔSTING mice before and after 3 days of Treg cell depletion (Figure S7E), suggesting that STING deficiency does not affect activation and expansion of self-reactive T cells following Treg cell depletion. These data support an early role of STING in inducing innate cytokine synthesis. On day 10 post continuous Treg cell depletion (Figure 6H), we examined pathology and cell infiltration in various tissues and organs. We discovered significant increases in splenocyte cell numbers in WT FoxP3-DTR mice that was mitigated in FoxP3-DTR^ΔSTING mice (Figure 6I). H&E staining of the lungs uncovered significant protection from immune cell infiltration in FoxP3-DTR^ΔSTING mice (Figure 6J), suggesting that absence of STING protected the host from autoreactive T cell-induced innate inflammation. Overall, these results provide compelling evidence to establish STING as a major driver of T cell-induced inflammation in DCs, and these results have important implications for targeting STING for autoimmune and autoinflammatory diseases.

DISCUSSION

We have reported previously that the transcriptional profile of DCs interacting with T_EM cells largely overlaps with proinflammatory gene expression of DCs activated with the TLR4 ligand LPS,⁹ including genes like Il1b, Il6, and Il12b; NF-κB subunits; and many other inflammatory cytokines and chemokines. Interestingly, T_EM cell-derived signals upregulate a unique subset of genes in DCs that are not regulated in LPS-activated DCs. This subset contains ER-resident genes, including several involved in cholesterol biosynthesis pathways and, very surprisingly, Tmem173 (STING). Transcriptional regulation of STING in correlation with its function is not well understood. It has been suggested that STING is an IFN-stimulated gene (ISG) and has STAT1 binding sites in its promoter.⁶⁵ The transcription factors cAMP response element-binding protein (CREB) and c-Myc have also been reported to induce STING expression.⁶⁶ However, lack of transcriptional upregulation of STING by LPS makes it unlikely that it is an ISG or can be upregulated by NF-κB in DCs. While we did not fully investigate the mechanisms by which T_EM cell-derived signals induce transcriptional induction of these genes, we discovered a functional role behind the transcriptional upregulation of STING. This is consistent with previous studies linking STING upregulation to its functional role in several inflammatory states, including a mouse model for Parkinson’s disease,⁶⁷ metabolic dysregulation,⁶⁸ and even traumatic brain injuries.⁶⁹ Conversely, downregulation of STING because of epigenetic programming in tumor cells^70,71 or reduced mRNA stability⁷² diminishes STING function during anti-viral sensing and anti-tumor immunity. The cholesterol biosynthesis pathway has also been implicated in limiting STING signaling during viral infection.^73–75 Parallel induction of the ER-resident cholesterol biosynthesis pathway in T_EM cell-activated DCs suggests potential negative regulation of STING-induced inflammation, which warrants future studies. It is also notable that T_EM cell-induced STING plays an extensive role in early-response inflammation by myeloid cells. However, more work needs to be done to understand the relationship between T_EM cell-mediated upregulation of ER-resident genes and Tmem173 in myeloid cells.

The cGAS-STING pathway is canonically activated by double-stranded DNA of bacterial or viral origin⁷⁶ but can also be activated by host-derived damaged mtDNA⁷⁷ and other genotoxic stressors, prompting cell repair or senescence.⁷⁸ Because cGAS in DCs had no role in dictating T_EM cell-induced inflammatory responses, this prompted us to explore other mechanisms of STING activation that regulate NF-κB driven cytokines. One possibility is that T_EM cells make cGAMP that is taken up by DCs, subsequently inducing DC-intrinsic STING activation. This has been demonstrated in the case of tumor cells, where tumor-derived cGAMP activates STING in non-tumor cells to induce anti-tumor immunity.⁷⁹ However, lack of degradation of STING in DCs excludes this possibility. Additionally, there are no reports of cGAMP production by T_EM cells following TCR activation. A previous report discovered a role of etoposide, an exogenous DNA-damaging agent, in inducing a cGAS-independent DDR pathway in human epithelial cell lines. This pathway then leads to activation of a STING-NF-κB axis.²² In the current study, we found that T_EM cells are capable of inducing mROS and subsequent DNA damage in myeloid cells during cognate interactions. We hypothesize that T_EM cell-mediated DNA damage in DCs is activating a cGAS-independent pathway of STING activation, triggering innate inflammatory cytokines. We found that neither TNFR or CD40 ligation is required for ROS or DDR induction, which suggests convergence of TNFRSF signaling and non-canonical STING activation pathways in driving innate inflammation. Additional work is required to understand which signals from T cells induce mROS in DCs, resulting in DSBs and non-canonical STING activation.

Various exogenous and endogenous insults can induce DNA damage, resulting in either DSBs or SSBs.⁸⁰ We find that T_EM cell-mediated interactions specifically lead to DSBs, as visualized by neutral comet assay.⁸¹ It is remarkable that cognate interaction of DCs with naive or effector CD4 T cell populations does not result in DSBs. While we have not fully explored the mechanisms behind these differences, we speculate that it could be because of differences in expression of certain cell-surface molecules. While naive CD4 T cells do not express CD44, high expression of CD44 on T cells has been found to promote clustering with DCs⁸² as well as overall stability of DC-T cell interactions,⁸³ suggesting enhanced interactions between effector CD4⁺ T (T_EFF) and T_EM cells with DCs. However, we find very high expression of PD-1 on effector T cells^84,85 but not T_EM cells (data not shown), suggesting the possibility that PD-1 signaling might mitigate effector T cell-driven DNA damage in interacting DCs. This needs to be further investigated. The DC-intrinsic DSBs are further bolstered by our observations of ATM, CHK2, and H2AX phosphorylation, which indicated a DSB-specific repair response⁴⁴ instead of apoptosis. It is well known that, when DSBs are unrepaired (or improperly repaired), they can give rise to neurodegenerative diseases and predisposition to cancer.⁸⁰ Patients with autoimmunity have a predisposition to multiple myeloma and other cancers,^86–88 suggesting a potential implication of our findings where unwarranted or persistent interaction of T_EM cells with myeloid cells during autoimmunity could result in myeloid-specific disorders.^89,90 It also remains to be examined whether repeated exposure to a pathogen leads to memory T cell-driven DNA damage in DCs and other myeloid cells, thus predisposing to myeloid leukemia.^91,92

Activation of myeloid cells, particularly macrophages, by TLR ligands has been shown to induce ROS through a variety of mechanisms.⁹³ This includes the activation of nicotinamide adenine dinucleotide phosphate oxidase (NOX) enzymes, which are responsible for generating cellular ROS.⁹⁴ ROS have been shown to be very critical in inducing NLR family pyrin domain containing 3 (NLRP3) inflammasome activation and subsequent IL-1β production.⁹⁵ It has also been suggested that TLR activation in macrophages leads to production of mROS.⁹⁶ Excessive or prolonged ROS production following TLR activation has been linked to tissue inflammation^97,98 and detrimental outcomes following inflammasome activation.⁹⁹ While we saw no induction of mROS by LPS-activated DCs at early time points, T_EM cell interactions with DCs led to rapid induction of mROS, alluding that there is a fundamental difference in mROS production between DCs sensing microbial ligands versus DCs interacting with T_EM cells. Scavenging of ROS with the selective scavenger NAC led to a significant decrease in H2AX phosphorylation, suggesting that T_EM cell-induced ROS play a major role in driving DSBs in interacting DCs. T cell-derived TNF and CD40L play an important role in T_EM cell-mediated activation of DCs, but blocking these pathways did not prevent mROS production or subsequent DDR.

Excessive mROS production has indeed been linked to detrimental innate cytokine production¹⁰⁰ and chronic inflammatory diseases, including Crohn disease, diabetes, neurodegeneration, and cardiac dysfunction.^101–104 Most of these diseases have a clear involvement of self-reactive T cells, and emerging literature links STING to similar chronic inflammatory diseases.^105–107 We posit here that autoreactive T cell-driven mROS and alternative activation of STING in myeloid cells could be involved in driving the chronic state of inflammation in these diseases. While we have not identified the upstream T_EM cell-derived signals that induce DC-intrinsic mROS, it is also important to note that scavenging ROS does not fully reverse the DDR or completely abolish inflammatory cytokine production by DCs, suggesting that additional upstream pathways causing DNA damage still need to be uncovered.

Our in vivo data employing anti-CD3 injection or Treg cell depletion support a major role of STING in inducing NF-κB genes and subsequent innate inflammation. Treg cell depletion has been shown to lead to profound activation of self-reactive T cells.⁶⁴ While activated T cells engage myeloid cells through TNFR and CD40 to induce innate inflammation,⁹ we were able to detect a DDR in select populations of myeloid cells at very early time points. Synthesis of Il1b and Il6 was compromised in STING-deficient mice, providing critical in vivo evidence implicating non-canonical activation of STING by T cell-derived signals. In vitro, we found that absence of STING considerably dampens early NF-κB activation, which is reflected in defective chromatin accessibility and transcription of pro-inflammatory genes, including Il1b, Il6, and Il12b. The regulation of early chromatin accessibility of NF-κB targets by STING is especially striking and suggests that STING-dependent signaling might be activating a pioneer transcription factor capable of specifically opening inflammatory gene-regulatory loci for transcription by NF-κB; however, this needs further investigation. It is thus likely that T_EM-DC interactions lead to rapid induction of mROS and additional signals that induce DDR and initiate STING-mediated NF-κB activation in DCs. We propose that STING is a primary activator of NF-κB during T_EM cell-mediated interactions, paving the way for additional amplification by TNFR and CD40 signaling. Given early activation of STING and its role in regulating chromatin accessibility of NF-κB targets, non-canonical activation of STING could precede engagement of the TNFR and CD40 signaling pathways, which needs further investigation.

STING is a widely expressed protein, and its diverse functional roles in multiple cell types pose a challenge for therapeutic targeting. Nonetheless, many STING antagonists and agonists are being tested for protection against hyper-inflammation as well as initiating anti-tumor responses.^108,109 It would be especially beneficial to preserve the anti-viral activity of STING and specifically target its ability to complex with TRAF6 to prevent T_EM cell-induced inflammation. Furthermore, because STING likely plays a role in inducing IL-1β and IL-6 prior to the TNFR and CD40 signaling pathways, targeting STING-TRAF6 complexes in addition to blocking TNFR-CD40 pathways may be an effective way of mitigating pathology in T cell-driven autoimmune diseases.

Limitations of the study

The study primarily uses in vitro-generated BMDCs and polyclonal T_EM cells, which may not be generalizable to ex vivo-derived DC subsets and other memory T cell populations.

Molecular assays related to DCs cultured with T_EM cells are challenging because of the requirement of sorting DCs following T_EM cell-mediated interactions. These sorting strategies can induce stress on cells, leading to variation among individual experiments.

We do not formally test STING translocation from the ER to the ER-Golgi apparatus and, thus, are not able to fully rule out every aspect of canonical STING activation.

Our in vivo mouse models compare WT with whole-body STING-deficient mice. While there are no inherent baseline differences in T cell populations within either genotype, it is important to address the importance of STING in a cell-specific manner. This includes generation of STING^fl/fl CSF1R-iCre mice for future studies.

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Chandrashekhar Pasare (Chandrashekhar.Pasare@cchmc.org).

Materials availability

Mouse lines generated in this study are available upon request from the lead contact.

Data and code availability

Bulk RNA-seq (GEO: GSE228682) and ATAC-seq (GEO: GSE229204) data generated during this study have been deposited at GEO and are publicly available as of the date of the publication. Previously published bulk RNA-seq data can be found in GEO: GSE184608. No code was generated for this study. Open-source published packages were used for sequencing studies as described in the STAR Methods section. Original immunoblot images and microscopy images reported in this paper will be shared by the lead contact upon request. Additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mice

Age- and sex-matched mice between 6 and 16 weeks were used for all experiments. Both males and females were used for all experiments. All mice were bred and maintained under specific pathogen–free conditions at Cincinnati Children’s Hospital Medical Center in accordance with protocols approved by Institutional Animal Care and Use Committee. cGAS−/− mice were a gift from Dr. Daniel Stetson at the University of Washington. Mice with Foxp3-DTR mutation have been described.⁶⁴ FoxP3-DTR mice were crossed with STING^gt/gt mice to create FoxP3-DTR^ΔSTING mice, with genotyping done by Transnetyx (Cordova, TN) to ensure STING missense mutation on exon 6. For short-term Treg depletion (to determine innate pro-inflammatory cytokine synthesis), WT Foxp3-DTR or FoxP3-DTR^ΔSTING mice were given diphtheria toxin (DT) diluted in PBS (25 μg/kg) on day 0 and 10 μg/kg intraperitoneally on day 1 and sacrificed on day 3. For extended Treg depletion (to determine organ-specific pathology), WT Foxp3-DTR and FoxP3-DTR^ΔSTING mice were given DT diluted in PBS (25 μg/kg) on day 0 and then 10 μg/kg intraperitoneally every 2 days and sacrificed on day 10. For anti-CD3 experiments, WT or STING^gt/gt mice were given intravenous anti-CD3 antibody (2.5 mg/kg) for 4hrs followed by flow cytometry and serum enzyme-linked immunosorbent assays (ELISAs).

METHOD DETAILS

Generation of BMDCs

After red blood cell (RBC) lysis, bone marrow cells were plated at 0.75 × 10⁶ cells/mL in BMDC medium (5% FCS containing complete RPMI +20 ng/mL recombinant GM-CSF). The medium was replaced on days 2, 4, and 6, and then cells were harvested on day 7 by gently flushing each well. For generation of TRAF6 deficient BMDCs, mouse bone marrow progenitor cells were electroporated using the Lonza 4D-Nucleofector X instrument. Cas9 (IDT) was used to electroporate cells with or without TRAF6 single guide RNA (sgRNA) 5′- GAAACTCAGAGTATGTACGT-3′ (Synthego), following the protocol from Nussing et al.⁴⁸ Following electroporation, bone marrow progenitor cells were plated at 0.75 × 10⁶ cells/mL in BMDC medium (5% FCS containing complete RPMI +20 ng/mL recombinant GM-CSF). The medium was replaced on days 2, 4, and 6, and then cells were harvested on day 7 by gently flushing each well.

Isolation of splenic DCs

Ex vivo-derived CD11b⁺ CD11c⁺ sDCs (>95% purity) were isolated from the spleen of B16-Flt3L melanoma¹¹⁰ injected mice. After RBC lysis, single cell suspensions were incubated with Fc Shield, then cells were stained with anti-mouse CD90.2, NK1.1, CD19, Ter119, Ly6G, F4/80, and CD317 biotinylated antibodies for 30 min. Cells were washed and subsequently incubated with MojoSort streptavidin nanobeads. Unstained cells were obtained using MojoSort kit negative selection protocol.

Isolation and differentiation of CD4⁺ T cells

Spleen and mesenteric lymph nodes were harvested and made into single-cell suspensions. After RBC lysis, naive CD4⁺ T cells (CD62L^hi, CD44^lo) were isolated according to the MojoSort Kit Protocol. Cell culture plates were coated with anti-CD3 (5 μg/mL) and anti-CD28 (5 μg/mL) in PBS for 2 h at 37C. Naive CD4⁺ T cells were plated in coated wells at 0.5 × 10⁶ cells/mL for 5 days with IL-2 (50 U/mL) in 10% FCS containing complete RPMI to assume a Th0 polarized cell phenotype. After differentiation, Th0 cells were replated at 1 × 10⁶ cells/mL in the presence of IL-2 (10 U/mL) in 10% FCS containing complete RPMI and rested for two additional days to allow differentiation into T_EM cells.

In vitro co-cultures

BMDCs and T_EM cells were harvested and plated at a ratio of 1:4 (DC:T cells) in 10% FCS containing complete RPMI. Where indicated, cells were preincubated with anti-TNF (20ug/mL), anti-CD40L (10ug/mL), ATMi (5uM), ATRi (10uM) or N-acetyl-CoA (NAC, 10mM) for 20 min before addition of anti-CD3 (100 ng/mL), OVA_323–339 peptide (1uM), EndoFit OVA protein (100 μg/ml), or LCMV GP_61–80 peptide (1uM). Where indicated, DCs were cultured with LPS (100 ng/mL), CpG (100 ng/mL), Pam3CSK4 (100 ng/mL) or DMXAA (10ug/mL). Cells and supernatants were harvested between 1 and 18 h later for downstream assays.

ELISA and serum ELISA

Capture antibodies for IL-6 and IL-1β were diluted and used to coat 96-well flat bottom plates overnight at 4C. Plates were blocked with PBS containing 1% BSA (IL-6) or 10% FCS (IL-1β) for 2 h at room temperature. Samples were diluted in blocking buffer and loaded in duplicate and then incubated overnight at 4C. Detection antibodies for IL-6 and IL-1β were diluted and used according to standard procedure. Protein concentration was quantified using OPD colorimetric assay using a standard curve of known concentrations for IL-1β and IL-6. All in vitro ELISAs were performed 18hrs following BMDC culture with T_EM cells or LPS stimulation. Blood was collected from mice via heart puncture and left to clot at room temperature for 1hr. For serum ELISAs, serum was isolated from coagulated blood after centrifugation at 12000g for 10 min at 10C. Undiluted or diluted serum samples were loaded in duplicate, and ELISA was performed as described above.

Flow cytometry and cell sorting

After RBC lysis, cells were washed with PBS containing 2mM EDTA and 1% BSA. For surface marker staining, cells were blocked with Fc Shield (anti-mouse CD16/32) for 10min and then incubated with antibodies of interest for 30min. For cytokine staining, cells were cultured in the indicated conditions for 1–3 h. Cells were immediately harvested on ice, stained with fixable Zombie Yellow, then fixed with 1% paraformaldehyde (PFA), and stained with anti-mouse pro–IL-1β, anti-mouse γH2AX, anti-mouse phospho-IKKa/b, or anti-mouse phospho-ATM for 30 min. DCs were gated on single cells, live, CD90.2⁻, CD11b⁺ CD11c⁺ populations, unless otherwise noted. Surface and intracellular antibodies can be found in the STAR Methods. Samples were analyzed using ACEA NovoCyte 2001 Flow cytometer (Agilent Technologies Inc). Cells were gated on singlets, and dead cells were excluded. Data were analyzed using FlowJo software (BD Biosciences). For cell sorting, samples were stained, washed, incubated with DAPI, and sorted on the Sony MA900 Cell Sorter (maintained by the Research Flow Cytometry Core in the Division of Rheumatology at Cincinnati Children’s Hospital Medical Center) directly into 20% FCS–containing PBS.

Quantitative PCR

Following stimulation, cells were pelleted, washed once with PBS, then lysed using TRIzol and stored at −80C. Total RNA was isolated using Qiagen RNeasy Mini extraction kit according to the manufacturer’s protocol. Complementary DNA (cDNA) synthesis was performed with M-MLV reverse transcriptase. Quantitative polymerase chain reaction (qPCR) was performed using the QuantStudio 7 Flex Real-Time PCR system. Primers used in this study are noted in key resources table. All cDNA normalized to Hprt1.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
CD11b BV785 (M1/70)	Biolegend	101243; RRID:AB_2561373
CD11c PE/Cy7 (N418)	Biolegend	117318; RRID:AB_493568
Ly-6C BV650 (HK1.4)	Biolegend	128049; RRID:AB_2800630
Ly-6G FITC (1A8)	Biolegend	127606; RRID:AB_1236494
Ly-6G APC/Cy7 (1A8)	Biolegend	127623; RRID:AB_10645331
MHCII (IA/IE) AF700 (M5/114.15.2)	Biolegend	107622; RRID:AB_493727
CD90.2 Pac Blue (30-H12)	Biolegend	140306; RRID:AB_10641693
CD90.2 BV785 (30-H12)	Biolegend	105331; RRID:AB_2562900
CD4 BV421 (GK1.5)	Biolegend	100443; RRID:AB_2562557
CD4 APC (GK1.5)	Biolegend	100412; RRID:AB_312697
CD8 PE/Cy7 (53-6.7)	Biolegend	100722; RRID:AB_312761
CD25 BV711 (PC61)	Biolegend	102049; RRID:AB_2564130
CD44 BV650 (IM7)	Biolegend	103049; RRID:AB_2562600
CD44 APC/Fire 750 (IM7)	Biolegend	103062; RRID:AB_2616727
CD62L AF700 (MEL-14)	Biolegend	104426; RRID:AB_493719
CD62L FITC (MEL-14)	Biolegend	104406; RRID:AB_313093
CD19AF700 (1D3/CD19)	Biolegend	152403; RRID:AB_2629812
CD45.1 PE (A20)	BD Bioscience	553776; RRID:AB_395044
CD45.2 APC (104)	Biolegend	109814; RRID:AB_389211
FoxP3 AF488 (MF-14)	Biolegend	126406; RRID:AB_1089113
Anti-phospho-H2AX (Ser139) (JBW301)	Millipore	05-636; RRID:AB_309864
Phospho-H2AX FITC	Millipore	16-202A; RRID:AB_568825
Pro-IL1b APC (NJTEN3)	Thermo	17-7114-80; RRID:AB_10670739
Pro-IL1b PE (NJTEN3)	Thermo	12-7114-82; RRID:AB_10732630
phospho-IKKa/b (Ser176/180) (16A6)	CST	14938; RRID:AB_2798653
phospho-ATM (Ser1981) eFluor 660 (10H11.E12)	Invitrogen	50-9046-41; RRID:AB_2574312
IL6 capture (ELISA)	Biolegend	504501; RRID:AB_315335
IL6 biotin (ELISA)	Biolegend	504601; RRID:AB_2127458
Anti-mouse CD45.1 biotin (A20)	Biolegend	110704; RRID:AB_313493
Anti-mouse CD90.2 biotin (53-2.1)	Biolegend	140314; RRID:AB_10643274
Anti-mouse Ly6G biotin (1A8)	Biolegend	127604; RRID:AB_1186108
Anti-mouse NK1.1 biotin (PK136)	Biolegend	108704; RRID:AB_313391
Anti-mouse Ter119 biotin (Ter119)	Biolegend	116204; RRID:AB_313705
Anti-mouse F4/80 biotin (BM8)	Biolegend	123105; RRID:AB_893499
Anti-mouse CD317 biotin (927)	Biolegend	127006; RRID:AB_2028466
Anti-mouse CD19 biotin (6D5)	Biolegend	115504; RRID:AB_313639
STING (D2P2F) Rabbit mAb	CST	13647; RRID:AB_2732796
TRAF6 (D-10) Mouse mAb	Santa Cruz	Sc-8409; RRID:AB_628391
Histone 3 Rabbit polyclonal Ab	CST	9715; RRID:AB_331563
Beta-tubulin Rabbit polyclonal Ab	CST	2146; RRID:AB_2210545
CHK1 (2G1D5) Mouse mAb	CST	2360; RRID:AB_2080320
Phospho-CHK1 (Ser345) (133D3) Rabbit mAb	CST	2348; RRID:AB_331212
CHK2 Rabbit polyclonal Ab	CST	2662; RRID:AB_2080793
Phospho-CHK2 (Thr68) Rabbit polyclonal Ab	CST	2661; RRID:AB_331479
Phospho-H2AX (Ser139) (20E3) Rabbit mAb	CST	9718; RRID:AB_2118009
IL1β/IL-1F2 Goat polyclonal Ab	R&D	AF-401; RRID:AB_416684
Goat anti-mouse HRP	BioRad	1706516; RRID:AB_11125547
Goat anti-rabbit HRP	BioRad	1706515; RRID:AB_11125142
Donkey anti-goat HRP	Thermo	PA1-28664; RRID:AB_10990162
Anti-CD3e in vitro (145-2C11)	Biolegend	100340; RRID:AB_11149115
Anti-CD3e in vivo (145-2C11)	BioXCell	BP0001-1
Anti-CD28 (37.51)	Tonbo	40-0281; RRID:AB_2621445
Fc shield CD16/CD32	Tonbo	70-0161; RRID:AB_2621487
Anti-TNFa (XT3.11)	BioXCell	BP0058
Anti-CD40L (MR-1)	BioXCell	BE0017-1
Chemicals, peptides, and recombinant proteins
ATM inhibitor (AZD0156)	Selleck Chem	S8375
ATR inhibitor (AZD6738)	Selleck Chem	S7693
2-Mercaptoethanol (BME)	Sigma	M6250
4’,6-diamidino-2-phenylindole (DAPI)	Biolegend	422801
Paraformaldehyde aqueous solution (PFA)	Fisher	15710
Fetal calf serum (FCS)	Sigma	F0926
RBC lysis buffer	Sigma	R7757
PBS	Thermo	14190250
Protease Inhibitor Cocktail Tablets, EDTA free	Roche	11873580001
PhosSTOP phosphatase inhibitor tablets	Sigma	4906837001
Hyclone RPMI-1640 with L-Glutamine	Fisher Sci	SH3002701
Bovine Serum Albumin (BSA)	Sigma	A7906
RIPA lysis buffer	Thermo	89900
MojoSort streptavidin nanobeads	Biolegend	480016
TRIzol Reagent	Invitrogen	15596026
5,6-dimethylxanthenone-4-acetic acid (DMXAA)	Invitrogen	tlrl-dmx
N-acetyl-L-cysteine (NAC)	Sigma	A9165
LPS E. coli O55:B5	Sigma	L2880
LCMV GP61-80 peptide	AnaSpec	AS-64851
OVA323-339 peptide	Invivogen	vac-isq
Ovalbumin Endofit	Invivogen	vac-pova
Recombinant IL-6 (ELISA standard)	PeproTech	AF-216-16
Recombinant IL-2	Biolegend	575404
Recombinant GM-CSF	Biolegend	576306
Protein G Dynabeads	Thermo	10004D
SYBR Green 2x master mix	Thermo	A25777
SYBR Gold Nucleic Acid Gel Stain	Invitrogen	S11494
Streptavidin-HRP	VWR	016-030-084
o-Phenylenediamine dihydrochloride (OPD)	Sigma	P6787
Diphtheria Toxin (DT)	Sigma	D0564
Janelia Fluor 646, SE	R&D	6148/1
M-MLV reverse transcriptase	Invitrogen	28025013
Alt-R S.p. Cas9 Nuclease V3	IDT	1081059
Critical commercial assays
Zombie Yellow Fixable Viability Kit	Biolegend	423104
FoxP3 transcription factor perm buffer	Invitrogen	00-5523-00
MitoSOX Red Mitochondrial Superoxide Indicator	Thermo	M36008
Comet Assay kit	R&D	4250-050-K
Il-1b/IL-1F2 DuoSet ELISA kit	R&D	DY401
MojoSort Naive CD4 T cell isolation kit	Biolegend	480040
RNeasy Mini Kit	Qiagen	74106
West Pico Plus Chemiluminescent Substrate	Thermo	34578
BCA Protein Assay kit	ThermoFisher	23227
NuPAGE™ LDS Sample Buffer (4X)	Invitrogen	NP0007
Nextera DNA Library Prep kit	Illumina	FC-121-1030
MinElute Reaction Cleanup kit	Qiagen	28206
AMPure XP Beads	Beckman Coulter	A63881
Lonza Walkersville P3 Primary Cell 4D-Nucleofector X Kit	Thermo	NC0545312
Deposited data
RNA seq	GEO	GSE228682
ATAC seq	GEO	GSE229204
Experimental models: Organisms/strains
Mouse: C57BL/6J	Jackson laboratory	JAX:000664
Mouse: C57BL/6J-Sting1gt/J	Jackson laboratory	JAX:017537
Mouse: B6.129(Cg)-Foxp3tm3(DTR/GFP) Ayr/J	Jackson laboratory	JAX:016958
Mouse: STING1gt Foxp3DTR/GFP	This study	N/A
Mouse: B6(C)-Cgastm1d(EUCOMM)Hmgu/J	Dan Stetson Lab	JAX:026554
Mouse: B6.Cg-Tg(TcraTcrb)425Cbn/J (OT-II)	Jackson laboratory	JAX:004194
Mouse: B6.Cg-Ptprca Pepcb Tg(TcrLCMV)1Aox/PpmJ (SMARTA-1)	Jackson laboratory	JAX:030450
Experimental models: Cell lines
B16-FLT3 mouse melanoma cells	Mach et al.,110	RRID:CVCL_IJ12
Oligonucleotides
STING Fwd 5’-AGCGGAAGTCTCTGCAGTCT-3’	(Sauer et al., 2018)	N/A
STING Rev 5’-GGAGCCCTGGTAAGATCAAC-3’	(Sauer et al., 2018)	N/A
IL1b Fwd 5’-TGTGCTCTGCTTGTGAGGTGCTG-3’	Jain et al.,8	N/A
IL1b Rev 5’-CCCTGCAGCTGGAGAGTGTGGA-3’	Jain et al.,8	N/A
IL6 Fwd 5’-CCGGAGAGGAGACTTCACAG-3’	McDaniel et al.,9	N/A
IL6 Rev 5’-GGAAATTGGGGTAGGAAGGA-3’	McDaniel et al.,9	N/A
Hprt1 Fwd 5’-CAGTCCCAGCGTCGTGATTA-3’	Jain et al.,8	N/A
Hprt1 Rev 5’-TGGCCTCCCATCTCCTTCAT-3’	Jain et al.,8	N/A
Traf6 sgRNA 5’-GAAACTCAGAGTATGTACGT-3’	Synthego	N/A
Software and algorithms
FlowJo 10	https://www.flowjo.com	N/A
GraphPad Prism 9	https://www.graphpad.com	N/A
Adobe Illustrator CC 2022	https://www.adobe.com/	N/A
R v4.2.0	https://www.r-project.org	N/A
DESeq2	https://doi.org/10.1186/s13059-014-0550-8	N/A
DiffBind v3.6.4	http://bioconductor.org/packages/release/bioc/vignettes/DiffBind/inst/doc/DiffBind.pdf	N/A
HOMER	Heinz et al., 50	N/A
DeepTools v3.5.0	(Ramirez et al., 2016)	N/A
Other
Novocyte 2001	Agilent	N/A
Synergy LX Multi-Mode Plate Reader	BioTek	N/A
QuantStudio 7 Flex Real-Time PCR system	Thermo	N/A
MA900 Cell Sorter	Sony	N/A
NIS-Elements Viewer	Nikon	N/A
Nikon Eclipse Ti inverted microscope	Nikon	N/A
Nikon Olympus BX51 microscope	Nikon	N/A
EasySep EasyEight Magnet	Stemcell Technologies	NC0651008
Bolt 8% Bis-Tris 1.0mm Mini Protein Gels	Thermo	NW00085BOX
Polyvinylidene fluoride (PVDF) membranes	BioRad	1620177
HyBlot Autoradiography Film	ThomasSci	1141J52

RNA and ATAC sequencing

CD11b⁺ CD11c⁺ cells were sorted from CD45.2⁺ WT or STING^gt/gt BMDCs using MojoSort streptavidin nanobeads. Sorted BMDCs were then cultured with CD45.1⁺ T_EM cells, or CD45.1⁺ T_EM cells + anti-CD3 (100 ng/mL) for 1hr. Live CD45.2⁺, CD90.2⁻, CD11c⁺ BMDCs were then FACS-sorted directly into 20% FCS-containing PBS. For RNA-sequencing, sorted cells were washed with cold PBS and immediately lysed in TRIzol. Purified RNA was sent to Novogene (San Diego, CA) to generate a cDNA library (150-bp pair-end reads) using HiSeq Illumina platform. With the help of the Bioinformatics Collaborative Services Core at the Cincinnati Children’s Hospital Medical Center, RNA-seq reads in FASTQ format were first subjected to quality control (FastQC v0.11.7) to assess the need for trimming of adapter sequences or bad quality segments (Trim Galore! v0.4.2 and cutadapt v1.9.1). Trimmed reads were aligned to the reference mouse genome version mm10 using STAR v2.6.1e. Aligned reads were stripped of duplicate reads with sambamba (v0.6.8). Gene-level expression was assessed by counting features for each gene, as defined in the NCBI’s RefSeq database.¹¹¹ Read counting was done using featureCounts (v1.6.2) from the Rsubread package.¹¹² Raw counts were normalized as Fragments Per Kilobase Million (FPKM). Differential gene expression (p-value ≤ 0.05, and log₂fold change (log₂FC) of ≥ 1 or ≤ −1) between normalized groups was evaluated using DESeq2 (v1.36.0). Heatmaps of differentially expressed genes were generated using ggplot2 (v3.4.1).

Sorted cells used for ATAC-seq were flash frozen in liquid nitrogen then stored at −80C prior to processing using the OMNI-ATAC protocol.¹¹³ In brief, cells were resuspended in cell lysis buffer and nuclei were isolated by centrifugation. Nuclei were then resuspended in a buffer containing Tn5 transposase for 30 min and DNA was subsequently extracted with a MinElute Reaction Cleanup Kit. Transposed DNA was amplified using PCR according to the protocol’s recommendations and libraries were purified with AMPure XP beads. Purified libraries were sequenced at 150 bases paired end on an Illumina NovaSeq 6000 at the CCHMC DNA Sequencing and Genotyping Core Facility. Raw FASTQ sequences were processed and aligned to the mm10 genome using the ENCODE ATAC-seq pipeline^114,115 (v2.0.0). Peaks were called within the pipeline using MACS2.¹¹⁶ Differential chromatin accessibility analysis was performed on the “conservative overlap” peak sets in R (v4.2.1) with DiffBind¹¹⁷ (v3.8.3). Peaks were considered differentially accessible if the FDR was less than 0.05. A modified version of HOMER⁵⁰ utilizing a log base 2 likelihood scoring system was used to calculate motif enrichment statistics for a large library of mouse position weight matrix binding site models contained in build 2.0 of the CisBP database.⁵¹ DeepTools¹¹⁸ (v2.0.0) was used to generate heatmaps of signal tracks across differentially accessible chromatin for each set of comparisons. GREAT¹¹⁹ analysis was performed using the rGREAT (v2.1.3) package.¹²⁰

Co-Immunoprecipitation

Cells were pelleted, washed once with PBS, then lysed in Mammalian Cell Lysis Buffer (50mM Tris-Cl pH 8, 1mM EDTA, 1mM EGTA, 1mM sodium orthovanadate, 50mM sodium fluoride, 5mM sodium pyrophosphate, 10mM sodium-B-glycerophosphate, 270mM sucrose, 10uL/mL aprotinin, 1% v/v Triton X-100, 0.1% v/v 2-Mercaptoethanol, 1 tablet 50X cOmplete protease inhibitor cocktail) for 1hr on ice and then cleared of debris by centrifugation at 15,000 RPM for 10 min. Protein was incubated with anti-STING rabbit monoclonal antibody (1:200) rotating overnight at 4C. Protein G beads were added to lysates for 2 h at 4C followed by 3 washes with Mammalian Cell Lysis Buffer. Bound proteins were eluted by boiling in reduced 1X NuPAGE LDS Samples buffer at 70C for 10 min. Immunoblotting was performed downstream to visualize protein complexes. A small portion for input controls were retained from whole cell lysates.

Immunoblotting

Following stimulation, CD11c⁺ CD45.2⁺ BMDCs were sorted away from CD45.1⁺ T_EM cells before being pelleted, washed once with PBS, then lysed in 1X RIPA buffer including protease and phosphatase inhibitors. Protein was quantified with Pierce BCA Protein Assay kit according to manufacturers’ protocol. Cell lysates were boiled in reduced 1X NuPAGE LDS Samples buffer at 70C for 10min. Cell lysates were separated by SDS-PAGE and transferred onto PVDF membranes. Blots were blocked in 5% BSA in Tris-Buffered Saline containing 0.1% Tween 20 (TBST) for 1hr. Blots were then incubated with indicated primary antibodies overnight, rocking at 4C. Blots were washed in TBST before being incubated with goat anti-rabbit HRP, goat anti-mouse HRP or donkey anti-goat HRP secondary antibodies in blocking buffer, rocking at room temperature for 1hr. Protein was visualized by incubating blots in West Pico PLUS chemiluminescent substrate for 3min, then enhanced chemiluminescence was recorded on autoradiography films using a Kodak developer located in Cincinnati Children’s Hospital Medical Center.

Comet assay to measure DNA damage

Following stimulation, CD11c⁺ CD45.2⁺ BMDCs were sorted away from CD45.1⁺ T_EM cells then cells were immediately evaluated using the CometAssay Single Cell Gel Electrophoresis assay (RnD systems). In short, sorted BMDCs were combined with 37C agarose at 1:10 ratio (5uL cells +45uL agarose) onto a provided 2-well slide. Slides containing lysed cells were electrophoresed to separate intact from damaged DNA, then stained with SYBR Gold prior to imaging. Image acquisition on 4 quadrants of each slide were performed using a Nikon Eclipse Ti inverted microscope. All imaging was performed under identical conditions using NIS Elements Advanced Research (AR) microscope imaging software. Comet tails were quantified using NIS Elements Viewer software in the Cincinnati Children’s Hospital Medical Center Confocal Imaging Core. Comet assay images used for visualization were enhanced by increasing exposure equally for each image.

Isolation and histopathological scoring of lung

Animals were first sacrificed using CO₂ and then perfused with PBS. Lungs were inflated at with 4% paraformaldehyde then fixed in the same solution overnight. Fixed lungs were paraffin embedded, then 10μm sections were stained with hematoxylin and eosin (H&E) before being scored in a blinded fashion. Lung injury scores were quantified using criteria published by the American Thoracic Society (ATS).¹²¹ Lung injury was assessed on a scale of 0–2 for each of the following criteria: (i) neutrophils in the alveolar space, (ii) neutrophils in the interstitial space, (iii) number of hyaline membranes, (iv) number of proteinaceous debris, and (v) extent of alveolar septal thickening. The final injury score was derived from the following calculation: score = (20 × i + 14 × ii + 7 × iii +7 × iv + 2 × v)/(number of fields 3 100). Lungs were imaged on a Nikon Olympus BX51 at 20x magnification.

QUANTIFICATION AND STATISTICAL ANALYSES

Statistical analyses to determine significance were performed in Prism 9 GraphPad and are shown in figure legends, with corresponding p values. Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns, not significant, as indicated in figure legends.

Highlights.

• CD4⁺ T_EM cells induce mitochondrial ROS and double-strand DNA damage in interacting DCs

• CD4⁺ T_EM cell interaction with DCs results in inflammation through STING-TRAF6-NF-κB pathway

• Non-canonical STING activation in DCs promotes accessibility of NF-κB motifs

• Autoreactive T cell-mediated innate inflammation in vivo involves STING-NF-κB signaling axis

INCLUSION AND DIVERSITY

We support inclusive, diverse, and equitable conduct of research.