Specific bile acids can elicit the type-I interferon response through the cGAS-STING pathway

Abstract

Bile acids are metabolites crucial to lipid metabolism and immune regulation, yet their biological functions and mechanistic underpinnings remain largely elusive. In this study, we demonstrate that specific bile acids DCA, CDCA and LCA can trigger the type-I interferon response (IFN-I) in various cells through the cytosolic DNA-sensing cGAS-STING pathway. Phosphoproteomics indicates that bile acids can elicit a wide array of changes across numerous signaling pathways, culminating in the downregulation of Bcl-2 and p-BAD, resulting in the formation of Bax/Bak pore for the cytosolic release of mitochondrial DNA. The induction of the IFN-I response also depends on inter-organelle interactions among the endolysosome, ER, and mitochondria, leading to calcium flux and mitochondrial dysfunction, which also contribute to mtDNA release. Further, while systemic administration of bile acid DCA can trigger the STING-dependent IFN-I response in various tissues and bloodstream, tissue-restricted application of DCA can exert antiviral and antitumor effects. Together, these findings identify the cGAS-STING pathway as a mechanistic underpinning of specific bile acids and provide new insights into harnessing bile acids for future therapy.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12964-026-03036-2.

Introduction

The primary bile acids are synthesized in the liver and released to the small intestine to facilitate the uptake of dietary lipids and fat-soluble vitamins. While most of the primary bile acids are reabsorbed in the small intestine, about 10% of the primary bile acids are retained in the colon and undergo extensive modification by microbiota. Commensal microbes are endowed with a vast array of bile acid-modifying enzymes, thereby responsible for the synthesis of a large pool of secondary bile acids [1–4]. The composition and function of such an enormously complex pool of secondary bile acids have just begun to be unveiled. In addition to their canonical roles as mediators and regulators of lipid uptake, bile acids have been increasingly recognized as pleiotropic signaling molecules that engage a myriad of signaling pathways crucially involved in intestinal homeostasis, inflammation and tumorigenesis [5–7]. The characterization of their novel physiological and pathological functions and the elusive underlying mechanisms has become an intensively investigated area, holding the promise of deciphering the complex interplay between microbiota and host.

In addition to their conventional roles in metabolism, emerging evidence indicates that bile acids function as signaling molecules, broadly involved in development, homeostasis and immunity. Both primary and secondary bile acids have been implicated in maintaining the delicate balance of CD4 regulatory T cells (Treg) and Th17 subsets in the gut, possibly acting through the nuclear receptors FXR, VDR or Rorγt, respectively [8–10]. While bile acids’ immune regulatory functions have been largely linked to their conventional receptors, including nuclear receptors FXR, VDR and LXR, as well as membrane-bound GPCRs such as TGR5 and S1PR2, various unconventional mechanisms have emerged. The secondary bile acids lithocholic acid (LCA) derivatives 3-oxoLCA and isoalloLCA can induce mitochondrial ROS production, promoting Treg development [8]. Also, a subset of rarely characterized bile acids has been recently identified as novel antagonists of the androgen receptor (AR) with the ability to promote the stemness of antitumor CD8 T cells [5]. DCA and CDCA act as versatile immune regulators engaging the calcium channel PMCA in CD8 T cells to dampen antitumor immunity [7], or triggering the type I interferon response in macrophages to promote Paneth cell dysfunction [11]. On the other hand, DCA produced by Clostridium Scindens appears to be important for the expression of interferon-stimulated genes (ISGs) required for host defense against virus CHKIV [12]. Although bile acids, particularly DCA and CDCA, have been implicated in facilitating the antiviral IFN-I and IFN-III responses, they appear to act through TGR5 and FXR rather than the well-characterized innate immune receptors [13, 14]. It is conceivable that bile acids may execute their multifaceted functions through a variety of molecular mechanisms.

The IFN-I response is mounted by nucleic acid-sensing pattern recognition receptors composed of TLRs, RLRs and cGAS. While pathogen-derived RNAs and DNAs are well-established triggers for IFN-I production [15–18], host-derived RNAs and DNAs have also been implicated in mounting IFN-I responses, contributing to autoimmunity and antitumor immunity [19–21]. In particular, both mitochondrial and nuclear genomic DNAs can be released into the cytosol during ageing, radiation or chemotherapy. The activation of the cytosolic DNA sensor cGAS can lead to the production of cGAMP, a second messenger and immunotransmitter that directly engages the ER-residing STING for a robust IFN-I response [15, 21]. The activation of the cGAS-STING pathway by self-DNAs has been linked to neurodegeneration, ageing and metabolic disorders [22, 23]. In this study, we provide compelling genetic and pharmacologic evidence that the cGAS-STING pathway mediates IFN-I induction by a subset of bile acids, comprising DCA, LCA and CDCA. Our findings implicate that DCA can engage multiple molecular and cellular events that converge on mitochondrial dysfunction and mtDNA release through the Bax/Bak pore. These findings open a new avenue for further exploration of the rapidly expanding bile acid family, both mechanistically and therapeutically.

Results

A subset of bile acids can trigger the IFN-I response independent of cell death pathways

We screened a small library of bile acids, including conjugated and unconjugated primary and secondary bile acids for their ability to trigger the IFN-I response in murine colon cancer cell MC38 and bone marrow-derived macrophages (BMDM). Following stimulating MC38 or BMDM with various bile acids for 6 h, qPCR assays revealed that taurine- and glycine-conjugated primary bile acids (TCA, TCDCA, GCA and GCDCA) were unable to elicit an IFN-I response, whereas unconjugated primary bile acid CDCA rather than CA triggered a prominent induction of Ifnb1 (Fig. 1a). Also, the secondary bile acid DCA elicited a strong IFN-I response in both MC38 and BMDM (Fig. 1a). These observations are in agreement with previous reports implicating a role of DCA and CDCA in the IFN-I response [11–14], and further revealed secondary bile acid LCA as a previously unappreciated inducer of the IFN-I response in both colon epithelium and macrophage (Fig. 1a). Of particular interest, the secondary bile acids LCA and DCA are enriched in the gut lumen with an average concentration of 500 µM and their abundances often fluctuate with dynamic changes in specific microbial compositions [24–26].

Fig. 1.

Bile acids induce the type-I interferon response independent of the cell death pathways. a BMDMs or MC38were treated with 200 µM of bile acids for 6 h, and the expression of Ifnb1 was quantified by RT-PCR. b BMDMs were treated with 200 µM of DCA or CDCA for 6 h, and the expression of various genes was quantified by RT-PCR and quantified as relative expression (indicated by log₂ fold change) (n = 3). c HMDM or HT29 were treated with 200 µM of bile acids for 6 h, and the expression of Ifnb1 was quantified by RT-PCR. d-f BMDMs were not pretreated (d), or pre-treated with Nec-1s (60 µM), or zVAD (20 µM) (e), or with Trolox (100 µM) or Fer-1 (10 µM) (f) for 30 min, and then treated with 200 µM of DCA or CDCA for 6 h. The expression of Ifnb1 and Cxcl10 were quantified by RT-PCR (n = 3). Data presented as means ± SD were analyzed by paired two-tailed Student’s t test

We noted that these three bile acids triggered the IFN-I response in a dose- and time-dependent manner. 50–300 µM of LCA, DCA or CDCA were capable of inducing the expression of Ifnb1 in a dose-dependent manner (Fig. S1a). 200 µM of each bile acid elicited the expression of Ifnb1 following stimulation for 3–18 h (Fig. S1b). When the treatment was extended to 48 h, lower dose of bile acids (20 µM) could also induce a modest IFN-I response (Fig. S1c). Subsequently, we used 200 µM of bile acids to stimulate BMDM for 6 h in the following studies unless indicated otherwise. Through this regimen, we validated that DCA and CDCA induced the expression of α and β-interferons, proinflammatory cytokines Tnfa and Il6, as well as chemokines Cxcl1, Cxcl10 and Ccl5 in BMDM (Fig. 1b). Of note, Rsad2 along with Cxcl10 and Ccl5 are regarded as interferon-stimulated genes (ISGs) whose induction is likely due to the autocrine or paracrine of type-I interferons. Furthermore, 200 µM of DCA, CDCA and LCA induced a prominent IFN-I response in human monocyte-derived macrophages (HMDM) and human colon cancer cell line HT-29 cells (Fig. 1c), indicating the IFN-I-inducing effect of bile acids is conserved in mouse and human species.

Of note, stimulation with 200 µM DCA or CDCA for 6 h did not cause a discernible cell death as assessed by permeation of Sytox-green and Propidium Iodide (PI) dyes (Fig. S1d, e), albeit prolonged stimulation for 18–24 h caused considerable cell death (Fig. S1d), likely due to necroptosis or/and pyroptosis as previously suggested [27, 28]. Considering cell death and IFN-I response can be intertwined, we further interrogated the role of cell death pathways in bile acid-elicited IFN-I response. We found that DCA and CDCA led to comparable Ifnb1 and Cxcl10 expression in BMDMs deficient in GsdmD or GsdmE (Fig. 1d). Disrupting Zbp1 or inhibiting RIPK1 with Nec1s did not have an impact on Ifnb1 induction, whereas blockade of pan-caspases via zVAD instead increased the IFN-I response (Fig. 1d, e), corroborating reported inhibitory roles of zVAD on the IFN-I response [29, 30]. These results excluded the involvement of pyroptosis and necroptosis. Also, inhibiting ferroptosis via Ferrostatin-1 (Fer-1) or Trolox did not alter DCA and CDCA-induced Ifnb1 (Fig. 1f). Together, these data indicate that a subset of bile acids can elicit a robust IFN-I response independent of the cell death pathways.

The cGAS-STING pathway pivots bile acids-induced IFN-I response

Considering TLRs, RIG-I/MDA5 and cGAS are the well-established innate immune receptors responsible for mounting the IFN-I response in settings of pathogen infection, sterile inflammation and cancer therapy, we sought to determine the role of these IFN-I-inducing innate immune signaling pathways in bile acids’ action. Similar to WT BMDM, Myd88^−/− and Trif^−/− BMDMs mounted a robust IFN-I response to DCA and CDCA, indicative of a dispensable role for TLRs (Fig. 2a). Also, Mavs^−/− BMDMs responded to DCA and CDCA just like WT BMDM (Fig. 2a), excluding the involvement of RIG-I or MDA5. Remarkably, BMDMs deficient in cGAS or STING almost completely lost their responses to DCA and CDCA (Fig. 2a), hence revealing a crucial role of the cGAS-STING pathway. Also, DCA and CDCA elicited a robust IFN-I response in murine embryonic fibroblasts (MEFs), and ablating Cgas or Sting1 abolished such an effect (Fig. 2b and Fig. S2a). Further, pharmacological inhibition of cGAS or STING with RU. 521 or H151, respectively, also abolished DCA and CDCA-induced Ifnb1 expression in BMDM and MC38 (Fig. 2c, d). These genetic and chemical evidences demonstrate that the cGAS-STING pathway is the major innate immune pathway that pivots specific bile acids-elicited IFN-I response.

Fig. 2.

Bile acids-induced IFN-I response depends on the cGAS-STING pathway. a-b WT or various knockout BMDMs (a) and MEFs (b) were treated with 200 µM of DCA or CDCA for 6 h. c-d BMDMs (c) or MC38 cells (d) were pretreated with H151 (1 µM) or RU.521 521 (10 µM) for 30 min and then treated with 200 µM of DCA or CDCA for 6 h. The expression of Ifnb1 and Cxcl10 were quantified by RT-PCR. e BMDMs were treated with 200 µM of DCA for indicated times and the activation of STING signaling was assessed by immunoblotting. f BMDMs were treated with various stimuli (200 µM DCA (6 h), 20 µM DMXAA (3 h) or transfected with 1 µg dsDNA (6 h)) as indicated. The fluorescence images were collected on a laser-capture confocal microscope and about 45 cells were quantified. Scale bars, 10 μm. Quantification of the percentages of the nuclear IRF3 positive cells over the total cells. N.D., not detected (right). g-h BMDMs were pretreated with BX795 (10 µM) for 30 min and then stimulated with 200 µM of indicated bile acids (g); WT or Irf3^−/− BMDMs were treated with 200 µM of indicated bile acids for 6 h (h), and the expression of Ifnb1 and Cxcl10 were quantified by RT-PCR. All the experiments were repeated 3 times with similar results. Data presented as means ± SD were analyzed by paired two-tailed Student’s t test

Biochemical analysis demonstrated that DCA stimulation for 6 h triggered the phosphorylation of STING (Fig. 2e), a hallmark of STING activation. Correspondingly, DCA also induced the phosphorylation of TBK1 (Fig. 2e), a key protein kinase downstream of STING. Immunofluorescence further revealed pronounced nuclear translocation of IRF3 in WT but not Cgas^−/− or Sting1^−/− BMDMs and MEFs upon DCA and CDCA stimulation (Fig. 2f and Fig. S2b), indicating the activation of transcriptional factor IRF3 downstream of the cGAS-STING pathway. Of note, pharmacological inhibition of TBK1 with BX795 or genetical disruption of IRF3 rendered macrophages unable to mount a notable IFN-I response to DCA and CDCA (Fig. 2g, h), supporting a crucial role for TBK1 and IRF3 in this process. Collectively, these data indicate that bile acids could trigger the activation of the cGAS-STING-TBK1-IRF3 pathway.

cGAS senses the mtDNA released into the cytosol following DCA treatment

Considering that cGAS is a cytosolic DNA sensor, we wondered whether these IFN-I-inducing bile acids could trigger the release of genomic or mitochondrial DNAs into the cytosol. Indeed, immunofluorescence staining with J2 antibody that specifically recognizes dsDNA revealed pronounced accumulation of cytosolic DNAs in BMDMs treated with DCA or CDCA (Fig. 3a). The presence of genomic DNAs (gDNA) and mitochondrial DNAs (mtDNA) in the cytosol can be assessed by quantitative PCR, quantifying the abundance of representative mitochondrial/nuclear genome-encoded genes in the cytosol. While the nuclear genome-encoded Tert1 and Polg1 were not enriched in the cytosol, mitochondrial genome-encoded Cox1 and Nd2 were highly present in the cytosol upon DCA or CDCA treatment (Fig. 3b). This data implicates that DCA and CDCA likely promote the release of mtDNA into the cytosol. To interrogate the role of mtDNA in bile acid-induced IFN-I response, BMDMs were cultured with ethidium bromide (EtBr) for 7 consecutive days for mtDNA depletion prior to bile acid treatment. Remarkably, DCA and CDCA failed to induce Ifnb1 and Cxcl10 expression in EtBr-treated BMDMs (Fig. 3c). Therefore, bile acids-triggered cytosolic release of mtDNA is crucial for the IFN-I induction.

Fig. 3.

Bile acids trigger IFN-I depends on mtDNA release and calcium flux. a MEFs were treated with 200 µM of indicated bile acids for various times and the fluorescence images were collected on a laser-capture confocal microscope. Scale bars, 10 μm. Quantification the percentages of cells presenting cytosolic dsDNA over the total cells (n = 3). b The expression of cytosolic gDNA (Tert, Polg1) and mtDNA (Cox1, Nd2) were quantified by qPCR in BMDMs treated with 200 µM of bile acids for 1 h (n = 3). c The induction of IFN-I by bile acids in BMDMs pretreated with 200 ng/mL EtBr for one weeks (n = 3). d-g BMDMs were transfected with indicated siRNAs for 72 h or pretreated with various inhibitors (PMCA inhibitor Lacl3 (60 µM), BAPTA-AM (0, 1, 5, 10 µM) or 2-APB (0, 10, 50, 100 µM)) for 30 min, and then treated with 200 µM DCA or CDCA for 6 h (n = 3). c-g The expression of Ifnb1 and Cxcl10 were quantified by RT-PCR. Data were presented as mean ± SD and analyzed by paired two-tailed Student’s t test

It is important to note that while cGAS and STING are required for DCA and CDCA-elicited Ifnb1 expression (Fig. 2a), they are not involved in mtDNA release per se (Fig. S3a). As in WT cells (Fig. 3b), Cgas^−/− and Sting1^−/− BMDMs demonstrated a prominent accumulation of the mitochondrial genes Cox1 and Nd2 in the cytosol upon DCA and CDCA treatment (Fig. S3a). Conceivably, the aberrant accumulation of dsDNA in the cytosol could act as a direct signal for cGAS activation. We then asked how bile acids might incite mtDNA release. To this end, we investigated whether any of the previously identified bile acid receptors, such as TGR5, S1PR2, PMCA and FXR [31], is involved in this process. Of note, knockdown of Gpbar1 which encodes TGR5 or S1pr2, did not impact the effect of DCA or CDCA on Ifnb1 (Fig. 3d and Fig. S3b). Also, inhibiting PMCA with LaCl3 had no impact on bile acid-induced IFN-I response (Fig. 3e). Genetic ablation of Fxr only resulted in a subtle reduction in Ifnb1 induction by DCA or CDCA (Fig. 3f). Thus, all those receptors are dispensable for DCA and CDCA’s effect on IFN-I response. As DCA has been shown to impact the intracellular calcium pool [7, 28], we interrogated the role of calcium in this process. Remarkably, depleting the intracellular calcium pool with BAPTA-AM or 2-APB nearly abolished Ifnb1 induction by DCA and CDCA (Fig. 3g). While PMCA is a calcium efflux channel and designated target for DCA executing an inhibitory role on CD8 T cells [7], inhibiting it with LaCl3 did not influence DCA’s ability to induce Ifnb1 in BMDM (Fig. 3e). Despite BAPTA being a general calcium chelator, 2-APB is regarded as an ER-specific IP3R blocker [32]. Hence, these data collectively suggests that ER-derived calcium flux might be one of the upstream signals responsible for mtDNA release.

ER-mitochondria interplay leads to Bax/Bak pore formation

ER and the mitochondria are often physically linked through MAMs and constantly crosstalk via calcium flux. For instance, ER-derived calcium overload can cause mitochondrial dysfunction and membrane leakage [20]. In particular, VDAC, the mitochondria permeability transition pore (mPTP), and Bax/Bak pores located on the mitochondrial outer and inner membranes have been implicated in the release of mtDNAs into the cytosol [20, 23, 33]. However, blockade of VDAC with VBIT-4 or mPTP with cyclosporine A (CsA) had no impact on DCA or CDCA’s effect on Ifnb1 (Fig. 4a). Instead, blocking Bax oligomerization with BAI1 abolished DCA and CDCA-induced Ifnb1 expression (Fig. 4a). Moreover, genetic knockdown of Bax and Bak in BMDM also diminished DCA and CDCA-induced Ifnb1 expression (Fig. 4b and Fig. S4a). Immunofluorescence assay confirmed the formation of oligomerized Bax foci upon stimulation with DCA or CDCA, but not with CA (Fig. 4c). Further, inhibiting Bax foci with BAII abolished DCA and CDCA-induced mtDNA release to the cytosol (Fig. S4b). These results indicate a crucial role for bile acid-induced Bax/Bak pore in mtDNA release and IFN-I induction. To gain insight into mechanisms underlying calcium flux and Bax/Bak pore formation, we conducted a phosphoproteomic analysis to provide a global view of the signaling events. Following DCA stimulation for only 3 h, substantial phosphorylation and dephosphorylation events across a large number of signaling pathways occurred (Fig. 4d and S4c, d). GSEA analyses indicated DCA stimulation led to substantial changes on processes involved in cholesterol transport, calcium import and release, oxidative stress, apoptosis and senescence, cGAMP-mediated signaling, IκB phosphorylation and NFκB activation, viral carcinogenesis, autophagy and macropinocytosis, and cellular transport and trafficking (Fig. 4e). The phosphorylation states on multiple calcium channels such as Stim1/2 and Trpv2, Trpm7, TPCN1 were drastically altered (Fig. 4d and S4c), and these channels’ activity has been known to be subject to regulation by phosphorylation [34]. Consistently, calcium-sensitive kinases, such as Camk2d and PKCβ’s phosphorylation were also highly induced upon DCA treatment (Fig. 4d and S4c). Correspondingly, NFAT family members such as NFATC1, NFATC2 and NFATC3, whose phosphorylation is subjected to regulation by the calcium-sensitive phosphatase calcineurin, demonstrated universal dephosphorylation by DCA (Fig. 4d). Thus, DCA appeared to trigger a potent calcium influx to impact CaMKII, PKC and Calcineurin-NFAT pathways. Also, broad changes in the phosphorylation of ER stress regulators PERK, ATF4 and DDIT3 implicated the involvement of the ER stress response. The PI3K-ATK-mTOR, cAMP-PKA-CREB/ATF2 and MAPKs pathways are comprehensively influenced by DCA (Fig. 4d, e, and S4c). Western blotting confirmed the downregulation of AKT-mTOR pathway by DCA (Fig. 4f). Corroborating with a crucial role of AKT and PKA on Bad phosphorylation [35–37], we found BAD phosphorylation was downregulated, including p-S155, indicative of its compromised ability to antagonizing Bax (Fig. 4d). Moreover, we also found that Bcl-xL was downregulated by DCA (Fig. 4f). Conceivably, reduction in both Bcl-xL abundance and BAD phosphorylation might have contributed to the activation of Bax and the assembly of Bax/Bak pore on the mitochondria, leading to mtDNA release and cGAS activation. Consistent with this notion, DCA stimulation led to the phosphorylation of TBK1, IRF3, IKK-γ, RELB, as well as several ISGs, such as Ifit1 and Zbp1 (Fig. 4d, e and S4d), supporting the activation of the cGAS-STING pathway and the induction of the IFN-I response.

Fig. 4.

Bile acids dampen the AKT-mTOR pathway and promote the Bax/Bak pore formation. a, b BMDMs were pretreated with indicated inhibitors (CsA (10 µM), VBIT-4 (10 µM) or BAI1 (10 µM) for 30 min or with siRNAs for 72 h and then treated with 200 µM of bile acids for 6 h. The expression of Ifnb1 and Cxcl10 were quantified by RT-PCR (n = 3). c MEFs were treated with 200 µM of various bile acids for 3 h. Fluorescence images were collected on a laser-capture confocal microscope. Scale bars, 10 μm. Quantification the percentages of BAX foci positive cells over the total cells. N.D., not detected (right) (n = 3). d-f BMDMs treated with 200 µM of DCA or vehicle for 3 h. d Scatter plot showing log₁₀ baseline intensity versus log₂ fold change (DCA vs. NT) for all protein phosphosites. Each dot represents a peptide phosphosite colored by –log₁₀ (p-value). Dashed lines indicate upregulation and downregulation thresholds (log₂FC = log₂1.5 and log₂0.7). e GSEA bubble plots showing significantly enriched KEGG and GO Biological Process terms (p < 0.25). Bubble size represents the number of core enrichment genes (Count), and color denotes the normalized enrichment score (NES). f Immunoblotting on various signaling pathways in BMDMs treated with 200 µM of DCA. This experiment has been conducted three times with similar results. Data were presented as mean ± SD and analyzed by paired two-tailed Student’s t test

Mitochondrial dysfunction and endocytosis are associated with bile acid-induced IFN-I response

Calcium overload can lead to mitochondrial dysfunction. Immunofluorescence indicated that the mitochondria became shortened and fragmented after DCA or CDCA, but not CA treatment (Fig. 5a). Compared to control macrophages, DCA or CDCA-treated macrophages had lower TMRM, indicative of dissipated membrane potential (Fig. 5b). These characteristics are the hallmark of mitochondria fission. Indeed, blocking Drp1 and mitochondrial fission with Midiv abrogated DCA or CDCA-induced IFN-I response (Fig. 5c). Concordantly, knockdown of Drp1 also blunted Ifnb1 induction by DCA or CDCA (Fig. 5d and Fig. S5a). On the contrary, knockdown of mitochondrial fusion regulators Oma1 or Pink1 did not impact on bile acid-induced Ifnb1 expression (Fig. 5d and Fig. S5a), further supporting mitochondrial fission rather than fusion is involved in IFN-I response. Of note, treatment with oligomycin or FCCP, which are capable of inducing mitochondrial membrane potential dissipation and fission, did not induce Ifnb1 expression (Fig. S5b), suggesting membrane potential dissipation alone was insufficient to trigger IFN-I response. Thus, mitochondrial dysfunction, characterized by mitochondrial fission and dissipation of the mitochondrial membrane potential, is one of the factors that promote bile acid-induced IFN-I response.

Fig. 5.

Bile acids promote mitochondrial dysfunction and endocytosis for Bax/Bak pore formation. a MEFs were treated with 200 µM of various bile acids for 3 h. Fluorescence images were collected on a laser-capture confocal microscope. Scale bars, 10 μm. Quantification of the percentages of cells with mitochondrial fragmentation over the total cells. N.D., not detected (right) (n = 3). b BMDMs were treated with 200 µM of bile acids for 6 h or 10 µM of FCCP for 40 min, then stained with TMRM. The fluorescence density of TMRM was visualized by a fluorescence microscope. Scale bars, 100 μm. Quantification of the cellular mean fluorescence density of TMRM (n = 3). c-f BMDMs were pretreated with various inhibitors (Mdivi-1 (10 µM), MβCD (500 µM), Genistin (200 µM), Dynsore (100 µM)) for 30 min or with siRNAs for 72 h, and then stimulated with 200 µM of bile acids for 6 h. The expression of Ifnb1 and Cxcl10 were quantified by RT-PCR (n = 3). g MEFs were pretreated with 500 µM of MβCD for 30 min and then stimulated with 200 µM of various bile acids for 3 h. Fluorescence images were collected on a laser-capture confocal microscope. Scale bars, 10 μm. Quantification of the percentages of BAX foci positive cells over the total cells. N.D., not detected (right) (n = 3). Data were presented as mean ± SD and analyzed by paired two-tailed Student’s t test

We then tested whether bile acids must enter the cytosol to induce mitochondrial dysfunction. To this end, we permeabilized the cell membrane with digitonin, allowing bile acids to bypass the membrane barrier and rapidly enter the cytosol. As shown previously [38], digitonin-mediated membrane permeabilization enabled much more efficient delivery of extracellular cGAMP into the cytosol, resulting in much higher Ifnb1 expression than in mock-treated cells (Fig. S5c). In contrast, digitonin treatment did not increase bile acid-induced IFN-I response (Fig. S5c), although it boosted bile acid’s entry as quantified by liquid phase-coupled mass spectrometry (Fig. S5d). These results suggest that DCA and CDCA do not need to enter the cytosol to elicit the IFN-I response. Given that disrupting TGR5 or S1PR2, or blocking PMCA did not influence bile acid-induced Ifnb1 expression (Fig. 3d, e), we also excluded the possibility that bile acids act in the extracellular compartment through their transmembrane receptors. Therefore, we surmised that these bile acids might undergo endocytosis to promote mitochondrial dysfunction and the ensuing IFN-I response. Indeed, blocking endocytosis with Genistein or MβCD abolished DCA or CDCA-induced Ifnb1 expression (Fig. 5e), suggesting that caveolae and cholesterol-dependent endocytosis are crucial. Knockdown of Rab5a or Rab5c, which are essential components of endocytosis, also ablated bile acid-induced IFN-I response, further supporting a critical role for endocytosis (Fig. 5f and Fig. S5e). Moreover, upon concomitant MβCD treatment, DCA or CDCA were unable to elicit mitochondrial fragmentation or Bax foci formation (Fig. 5g and S5f). Of note, permeabilizing the endolysosome membrane with LLOM, a lysosome membrane rupture peptide [39], to promote the release of endocytosed DCA to the cytosol led to much higher induction of Fgf15, a well-known target gene of DCA’s receptor Fxr, but failed to increase DCA-induced Ifnb1 expression that requires mtDNA release and cGAS activation (Fig. S5g). These results demonstrate that the cytosolic DCA is unable to trigger the mtDNA release and cGAS activation, thus supporting the possibility that transfer of bile acids from endolysosome to the ER or/and mitochondrial compartments and inter-organelle crosstalk may be a prerequisite for mitochondrial dysfunction and Bax/Bak pore formation.

Bile acid triggers the STING-dependent IFN-I response in vivo

Excessive bile acids have been implicated in multiorgan inflammation and various pathological conditions, such as cholestasis, metabolic dysfunction- and alcohol-associated liver diseases, type-2 diabetes, and inflammatory bowel diseases [40, 41], we wondered whether the STING and IFN-I response are involved. Intraperitoneal administration of various doses of DCA to C57BL/6 wild type (WT) mice led to varied induction of Ifnb1 and Cxcl10 in multiple tissues in 3 days (Fig. 6a). Except for the kidney, DCA induced a dose-dependent IFN-I response in various tissues, including colon, lung, heart, liver, spleen. Among them, heart, lung and spleen were highly sensitive, and responded to the lower DCA challenge (~ 0.25 mg/mouse) (Fig. 6a). ELISA also detected a prominent presence of Cxcl-10 protein in the bloodstream following various DCA treatments (Fig. 6b). Remarkably, DCA was unable to mount an IFN-I response in Sting1^−/− mice, suggesting DCA also impinges on STING in vivo (Fig. 6c). Injection of one dose of DCA at approximately 1.25 mg/mouse also resulted in significant morbidity and mortality in both WT and Sting1^−/− mice (Fig. 6d, e). While both mice exhibited rapid weight loss, WT mice had markedly severe weight loss and all scummed by day 5. In contrast, the majority of Sting1^−/− mice survived after day 5, and 35% even survived throughout (Fig. 6d, e). Overall, the lessened morbidity and mortality in Sting1^−/− mice correlated with the blunted IFN-I response elicited by DCA. These results indicate the cGAS-STING pathway may be responsible for bile acid-induced systemic inflammation.

Fig. 6.

DCA induces STING-dependent IFN-I in vivo and promotes antitumor and antiviral immunity. a-b WT mice were i.p. injected with one dose of 10, 30, or 50 mg/kg DCA for 3 d. The expression of Ifnb1 and Cxcl10 were quantified by RT-PCR (a) or ELISA (b) (n ≥ 4 mice). c-e WT or Sting1^−/− Mice were i.p. infected with 50 mg/kg DCA, the expression of Ifnb1 and Cxcl10 were quantified by RT-PCR on day 3 (n = 3) (c); the survival and body weight of the treated mice were documented everyday (n = 8–12) (d, e). f MC38 tumor cells (1 × 10⁶) were subcutaneously injected into WT mice. The tumors (~100 ± 20 mm³ indicated by black arrow) were injected with DCA (125 µg/mouse) or PBS every 2 days. The tumor volumes ((length×width²)/2) were measured by a Vernier scale every 2 days (n = 4). g WT mice were pretreated with PBS or DCA (125 µg/mouse) for two consecutive days, then intranasally infected with IVA (PR8, 250000 PFU/mouse) for 3 days. The expression of IVA genes was quantified by RT-PCR (n = 3). Data are presented as mean ± SEM and analyzed by two-way ANOVA with Bonferroni’s multiple comparisons test (d-f) or mean ± SD and analyzed by paired Student’s t-test (a-c, g)

The IFN-I response is crucial for mounting antitumor and antiviral immunity [15, 16, 21]. Regarding DCA’s ability to induce the IFN-I response in colon cancer MC38 cells (Fig. 1a), we surmised that intratumoral injection of DCA might boost the antitumor immune response while avoiding systemic inflammation. We first inoculated MC38 cells subcutaneously into the flank of WT C57BL/6 mice. On day 6, when tumors became visible, we started to inject a low dose of DCA (0.125 mg/mouse) intratumorally every other day for 5 times. It is evident that DCA injection led to prominent accumulation of immune cells in the tumor microenvironment, particularly macrophages, B cells, NKT and CD4 and CD8 T cells (Fig. S6a-b). While mock-treated tumors continued to grow, DCA treatment halted tumor growth and prevented progression over 25 days (Fig. 6f). These results demonstrate that local administration of low-dose DCA can promote antitumor immune responses.

We then tested whether locally administering DCA could prevent viral infection by RNA virus influenza virus A (IVA). After intranasal infection with IVA for 3 days, prominent expression of several viral genes was detected in the lung tissue (Fig. 6g). Also, cytokines and interferons were highly induced (Fig. S6c), corroborating the activation of RNA-sensing innate immune receptors TLR3 and RIG-I, as previously reported [42]. Interestingly, pretreatment with a low dose of DCA (0.125 mg/mouse) intranasally efficiently constrained IVA infection (Fig. 6g). We only detected very low levels of viral gene expression in the lungs pretreated with DCA (Fig. 6g). As expected, lack of viral propagation also led to reduced immune responses in DCA-treated lung tissues (Fig. S6c). Therefore, administering DCA to the respiratory tract can exert protection against IVA infection.

Discussion

Bile acids have emerged as versatile signaling molecules regulating a diversity of immune processes through both canonical and noncanonical mechanisms. For instance, DCA and CDCA have been implicated in shaping the gut immune homeostasis and antiviral immunity through the canonical receptors FXR and TGR5 [11, 13]. In this study, we first demonstrate that in addition to DCA and CDCA, LCA can also elicit the IFN-I response, thereby expanding the spectrum of the IFN-I-inducing bile acid family. We further provide compelling evidence demonstrating that the cGAS-STING pathway is responsible for all three specific bile acids-induced IFN-I response, adding a noncanonical mechanism to bile acids’ action. Further, our results indicate that these bile acids can elicit endolysosome-ER-mitochondria inter-organelle crosstalk, in which calcium flux acts as a secondary messenger promoting mitochondrial dysfunction and mtDNA release (Fig. S7). These findings collectively provide novel insights into bile acids’ far-reaching functions and underlying mechanisms.

While bile acids have traditionally been linked to lipid metabolism, emerging evidence suggests they can also play key roles in immune regulation. Nevertheless, the molecular and cellular mechanisms underpinning bile acids’ diverse immune regulatory roles remain an area of active exploration. While DCA and CDCA have been linked to the IFN-I response previously [11, 13], it is unclear which innate immune receptor is involved. Thorough a thorough systemic genetic analysis, we identified the cytosolic DNA sensor cGAS as the sole innate immune receptor pivoting bile acids’ IFN-I response. Further pharmacological and biochemical studies pinpointed that these specific bile acids, namely DCA, CDCA and LCA, induce the IFN-I response through the cGAS-STING-TBK1-IRF3 axis. This new finding reinforces the long-standing view that the IFN-I response must be initiated by TLRs, RIG-I/Mda5 or cGAS, whether in settings of antiviral defense, sterile inflammation or antitumor immunity [43, 44]. We validated the involvement of cGAS rather than TLRs or RIG-I/MDA5 in bile acids’ IFN-I response in at least three types of cells, including transformed gut epithelium, primary macrophage and immortalized embryonic fibroblast in both human and mouse. Linking the cGAS-STING pathway to a subset of bile acids is supported not only by rigorous genetic evidence, but also by orthogonal pharmacological and biochemical data. Therefore, our findings will undoubtedly open a new avenue for future studies on bile acids’ ever-expanding immune regulatory roles.

Both genomic and mitochondrial DNAs have been implicated in activating cGAS during sterile inflammation. Our study indicated that bile acids primarily promote mtDNA release to engage cGAS in the cytosol. Searching for the underlying mechanism reveals a rather intriguing and complex inter-organelle interplay that drives mitochondrial dysfunction. Permeabilizing the plasma membrane to facilitate the cytosolic entry of bile acids did not enhance the IFN-I induction, hence excluding the possibility that bile acids may directly cross the plasma membrane to act on the mitochondria. Instead, blocking cholesterol-dependent endocytosis blunted IFN-I induction, suggesting that endocytosis is required not only for bile acid uptake but also for their function. The translocation of cholesterol from the endolysosome to the ER is a well-established mechanism by which cells sense cholesterol abundance and fine-tune de novo lipid synthesis [45]. We envisage the endocytosed bile acids may follow a similar transport route to the ER, leading to ER membrane damage and calcium release. As the ER and mitochondria are often in close proximity, it is plausible that mitochondria then take up calcium, and calcium overload leads to various aberrant phenotypes such as mitochondrial fission and dissipation of membrane potential. Indeed, ER-derived calcium has been shown to be the primary cause of mitochondrial dysfunction and mtDNA release into the cytosol [20, 28]. This study offers a unique paradigm that highlights endolysosme-ER-mitochondria inter-organelle interplay.

Our phosphoproteomic analysis also revealed bile acids’ rather unexpected impact on a wide range of protein kinase pathways, including Akt, mTOR, PKA, PKC and CamKII. Conceivably, perturbations in these signaling pathways could have profound ramifications for cell growth, survival and function. Lysosomal sensing of cholesterol has been linked to mTOR activation [46]. In contrast, we found that DCA treatment led to mTOR inactivation. One possible scenario is that exuberant bile acids might have interfered with lysosomal mTOR sensing of cholesterol and lipids. Corroborating mTOR’s impact on Bcl-2 family members [47], we found that DCA also downregulated Bcl-2 expression, thereby promoting Bax activation. Our results also suggest that ER stress and mitochondrial dysfunction, in conjunction with calcium flux might have a broad impact on protein kinase pathways, particularly on the phosphorylation of AKT, PKA, PKC and CamKII, and PERK. Of particular interest, AKT and PKA have been implicated in mediating BAD phosphorylation, thereby preventing Bax/Bak pore formation [36, 37]. In this regard, AKT inactivation and Bcl-2 downregulation might have contributed to bile acid-induced Bax/Bak pore formation on the mitochondrial membrane. Although VDAC and mPTP can also mediate mtDNA release [19, 33], our results implicate that the Bax/Bak pore is specifically involved in DCA- and CDCA-mediated mtDNA release in macrophages. Together, our results support a scenario that calcium flux and a dysregulated Bcl2-Bad-Bax/Bak axis might have collaborated to promote mtDNA release and cGAS activation in the bile acid-induced IFN-I response.

Of note, bile acids have been implicated in the regulation of cell survival, inflammasome and T cell functions. While various nuclear receptors such as FXR, VDR and LXR and ROS have been implicated in these processes [2, 7, 9, 41], our study provides intriguing new perspectives. For example, DCA-mediated inactivation of Akt and mTOR may contribute to DCA-induced necroptosis and pyroptosis [27, 41]. mTOR-regulated cell metabolism may contribute to bile acid-induced CD4 Treg and Th17 function [8, 9]. Also, ER stress and calcium influx may contribute to the induction of ROS by DCA and CDCA in Treg cells [8, 48]. We also anticipate that the transcriptional factors CREB/ATF2 may collaborate with nuclear receptors or IRF3 and NFκB to impact the expression of a variety of immune regulatory genes including Ifna/b [1, 40]. In this regard, the signaling pathways affected by bile acids in macrophages may also be involved in other immune cells. Further studies are needed to explore these aspects in the future.

Bile acids have been implicated in cell death, inflammation, antiviral and antitumor immunity [1, 40]. The link of specific bile acids to the cGAS-STING pathway prompted us to investigate the involvement of this pathway in the myriad functions of bile acids. Interestingly, we found that ablation of cGAS or STING did not affect bile acid-induced cell death, suggesting that other mechanisms such as Apaf1 [27] may operate independently of the cGAS-STING pathway in cell death. Remarkably, DCA induced a prominent IFN-I response in a variety of tissues in a STING-dependent manner. Also, DCA-triggered morbidity and mortality were ameliorated in Sting1^−/− mice. Hence, the cGAS-STING pathway seems to be involved in bile acid-induced systemic inflammation. While systemic administration of high doses of DCA could promote overt inflammation, we found local administration of a low dose of DCA can have antitumor and antiviral effects. Intratumoral injection of DCA to implanted MC38 colon cancer promoted NKT and CD4 and CD8 T cell infiltration, contributing to the control of tumor growth. Intranasal administration of DCA significantly prevented influenza virus infection, expanding DCA’s antiviral spectrum. Overall, this study not only reveals novel mechanisms underlying bile acids’ diverse functions but also provides new insights into harnessing bile acids for future therapies against cancer and infectious diseases.

Materials and methods

Mice

The Sting1^−/−, Cgas^−/−, Mavs^−/−, Trif^−/−, MyD88^−/−, Irf3^−/−, Fxr^−/−, Gsdmd^−/− and Gsdme^−/− mice on C57B/6J background were purchased from the Jackson Laboratory. Zbp1^−/− mice generated by CRISPR-Cas9 were kindly provided by Dr. Jiahuai Han and described previously [49]; All the mice were bred and maintained in a pathogen-free animal facility at the Shanghai Institute of Materia Medica, Chinese Academy of Sciences. All procedures were conducted in compliance with a protocol approved by the Institutional Animal Care and Use Committee of the Shanghai Institute of Materia Medica, Chinese Academy of Sciences.

Cells culture

MC38 (from Dr. Liufu Deng’s lab) cells, MEFs and HT29 were cultured in DMEM (catalog no. SH30243.01; HyClone Laboratories) supplemented with 10% FBS (catalog no. SH30084.03; HyClone Laboratories), 100 U/mL penicillin, and 100 µg/mL streptomycin (catalog no. 6016ES76; Yeasen Biotechnology).

BMDMs were isolated from the femurs and tibia of 6- to 10-week-old mice, and the RBCs were lysed using ACK lysis buffer (catalog no. C3702; Beyotime). Bone marrow cells were cultured in RPMI 1640 (catalog no. SH30809.01; HyClone Laboratories) supplemented with 10% FBS, 100 U/mL penicillin, and 100 µg/mL streptomycin (catalog no. 6016ES76; Yeasen Biotechnology), M-CSF (20 ng/mL; catalog no. 315-03; Peprotech). On day 3, nonadherent cells were removed and fresh medium was added. On day 5–8, adherent macrophages were detached by trypsin digestion and replated for further use.

Peripheral blood mononuclear cells (PBMCs) were isolated from healthy donors via density gradient centrifugation using Ficoll-Paque (catalog no. 07801; STEMCELL). 1 × 10⁷ PBMCs were collected and cultured in RPMI-1640 medium supplemented with 10% FBS and 20 ng/mL recombinant human macrophage colony-stimulating factor (human M-CSF) (20 ng/mL; catalog no. 300 − 25; Peprotech) for 7 days, with medium being replenished on day 3. On day 7, the differentiated macrophages (HMDMs) were detached by trypsin digestion and replated for DCA treatment.

All these cells were checked regularly by mycoplasma contamination using a commercial PCR protocol (MycoAlert, catalog no. LT07-418; Lonza).

Reagents

The following reagents were purchased from the manufacturers as noted:

CA (Sodium Cholate) (catalog no. S9875; Sigma-Aldrich), DCA (Sodium deoxycholate) (catalog no. D6570; Sigma-Aldrich), CDCA (Sodium chenodeoxycholate) (catalog no. C8261; Sigma-Aldrich), LCA (Lithocholic acid) (catalog no. L6250; Sigma-Aldrich), UDCA (Ursodeoxycholic acid) (catalog no. U5127; Sigma-Aldrich), TCA (Sodium taurocholate) (catalog no. T4009; Sigma-Aldrich), TDCA (Sodium taurodeoxycholate) (catalog no. T0875; Sigma-Aldrich), TCDCA (Sodium taurochenodeoxycholate) (catalog no. T6260; Sigma-Aldrich), GCA (Glycocholic acid) (catalog no. G2878; Sigma-Aldrich), GDCA (Sodium glycodeoxycholate) (catalog no. G9910; Sigma-Aldrich), GCDCA (Sodium glycochenodeoxycholate) (catalog no. G0759; Sigma-Aldrich), 2’3’-cGAMP (5 µM; catalog no. tlrl-nacga23-5; InvivoGen), DMXAA (20 µM; catalog no. S1537; Selleck Chemicals), Poly(I: C) (catalog no. P1530; Sigma-Aldrich), dsDNA (catalog no. 31149; Sigma-Aldrich), Sytox Green (1 µM; catalog no. S7020; Thermo Fisher Scientific), Propidium iodide (2 µg/mL catalog no. C1022; Thermo Fisher Scientific), Hoechst 33,342 (5 µg/mL catalog no. P3566; Beyotime), TMRM (Tetramethylrhodamine methyl ester) (100 nM; catalog no. 22221; AAT Bioquest), Mito-Tracker Deep Red FM (5 µg/mL; catalog no. HY-D1783; MCE), DiD (5 µM; catalog no. C1039; Beyotime), Antifade Mounting Medium with DAPI (catalog no. HY-K1047; MCE), 4% paraformaldehyde (catalog no. P0099; Beyotime), Triton X-100 (catalog no. P0096; Beyotime), Digitonin (10 µg/mL; catalog no. D141 Sigma-Aldrich), cOmplete mini (catalog no. 11836153001; Roche), LLOM (L-Leucyl-L-Leucine methyl ester hydrochloride) (600 µM; catalog no. HY-129905; MCE).

TRIzol (catalog no. DP424; TIANGEN), Hifair^® III 1st Strand cDNA Synthesis SuperMix for qPCR (catalog no. 11141ES10; YEASEN), Hieff^® qPCR SYBR Green Master Mix (catalog no. 11202ES03; YEASEN), INVI DNARNA Transfection Reagent (catalog no. IV1216075; Invigentech).

VBIT-4 (10µM; catalog no. HY-129122; MCE), CsA (Cyclosporin A) (10 µM; catalog no. HY-B0579; MCE), BAI1 (10 µM; catalog no. HY-103269; MCE), MβCD (β-Cyclodextrin) (500 µM; catalog no. HY-107201; MCE), Genistin (200 µM; catalog no. HY-N0595; MCE), Dynasore (100 µM; catalog no. HY-15304; MCE), 2-APB (2-Aminoethyl diphenylborinate) (100 µM; catalog no. HY-W009724; MCE), Lacl₃ (Lanthanum (III) chloride) (60 µM; catalog no. HY-W076030; MCE), Mdivi-1 (10 µM; catalog no. HY-15886; MCE), zVAD (20 µM; catalog no. T88862; TargetlMol), Nec-1s (60 µM; catalog no. 2263-5; BioVision), H151 (1 µM; catalog no. inh-h151; InvivoGen), RU.521 (10 µM; catalog no. HY-114180; MCE), BAPTA-AM (0–10 µM; catalog no. B4758; APExBIO), BX795 (10 µM; catalog no. tlrl-bx7; InvivoGen), Trolox (100 µM; catalog no. 238813; Sigma-Aldrich), Fer-1 (Ferrostatin-1) (10 µM; catalog no. S7243; Selleck), FCCP (10 µM; catalog no. HY-100410; MCE), Oligomycin (2 µM; catalog no. HY-N6782; MCE), EtBr (ethidium bromide) (200 ng/mL; catalog no. E7637; Sigma-Aldrich), Cisplatin (50 µM; catalog no. S1166; Selleck).

Anti-p-STING (Western blot [WB]: 1:1000; catalog no. 72971; Cell Signaling Technology), anti-STING (WB: 1:1000; catalog no. 13647; Cell Signaling Technology), Anti-cGAS (WB: 1:1000; catalog no. 31659; Cell Signaling Technology), anti-p-TBK1 (WB: 1:1000; catalog no. 5483; Cell Signaling Technology), anti-TBK1 (WB: 1:1000; catalog no. 3504; Cell Signaling Technology), anti-AKT (WB: 1:1000; catalog no. 9272; Cell Signaling Technology), anti-p-AKT (WB: 1:1000; catalog no. 4060; Cell Signaling Technology), anti-p-AKT (WB: 1:1000; catalog no. 13038; Cell Signaling Technology), anti-Bcl-xL (WB: 1:1000; catalog no. 2764; Cell Signaling Technology), anti-Bcl-2 (WB: 1:1000; catalog no. ET1702-53; HUABIO), anti-p-S6 (WB: 1:1000; catalog no. 4858; Cell Signaling Technology), anti-4E-BP1 (WB: 1:1000; catalog no. 9644; Cell Signaling Technology), anti-p-4E-BP1 (WB: 1:1000; catalog no. 2855; Cell Signaling Technology), anti-β-Actin (WB: 1:2000; catalog no. AC026; ABclonal), anti-IRF3 (IF: 1:250; catalog no. 4302; Cell Signaling Technology), anti-dsDNA (IF: 1:250; catalog no. ab270732; Abcam), anti-BAX (IF: 1:250; catalog no. A20227; ABclonal), Alexa Flour 488 goat anti-rabbit IgG (IF: 1:1000; catalog no. A11001; Invitrogen), Alexa Flour 555 goat anti-rabbit IgG (IF: 1:1000; catalog no. A32732; Invitrogen).

Anti-CD45 (FC: 1:500; catalog no. 103173, 30-F11; BioLegend), anti-CD3 (FC: 1:500; catalog no. 100271, 17A2; BioLegend), anti-Ly6G (FC: 1:500; catalog no. 127613, 1A8; BioLegend), anti-CD11c (FC: 1:500; catalog no. 117305, N418; BioLegend), anti-F4/80 (FC: 1:500; catalog no. 123113, BM8; BioLegend), anti-NK1.1 (FC: 1:500; catalog no. 563220, PK136, 1:500, BD), anti-mouse CD16/32 (FC: 1:500; catalog no. 156603, S17011E; BioLegend), anti-CD8 (FC: 1:500; catalog no. 100741, 53 − 6.7; BioLegend), anti-CD4 (FC: 1: 500; catalog no. 116022, RM4-4; BioLegend), anti-CD19 (FC: 1:500; catalog no. 115573, 6D5; BioLegend), anti-CD11b (FC: 1:500; catalog no. 563168, M1/70; BD), anti-MHC-II (FC: 1:500; catalog no. 107670, M5/114.15.2; BioLegend), anti-CD80 (FC: 1:500; catalog no. 104725, 16-10A1; BioLegend), Live/dead dye (FC: 1:200; catalog no. 423101, BioLegend).

RNA interference assay

Control non-specific or targeted short interfering RNAs (siRNAs; 20 nM) were transfected into 2.5 × 10⁵ BMDMs in 24-well plates with 1.5 µL of INVI DNARNA Transfection Reagent (catalog no. IV1216075; Invigentech) at a ratio of 1:1 with siRNAs for 3 days before stimulation. Control siRNA sequences were 5’-UUCUCCGAACGUGUCACGUTT-3’ and 5’-ACGUGACACGUUCGGAGAATT-3’; Rab5a siRNA sequences were 5’-UGGUCAAGAACGGUAUCAUTT-3’ and 5’-AUGAUACCGUUCUUGACCATT-3’; Rab5c siRNA sequences were 5’-GACGAUACAACGGUCAAGUTT-3’ and5’-ACUUGACCGUUGUAUCGUCTT-3’; Bax siRNA sequences were 5’-CCGGCGAAUUGGAGAUGAATT-3’ and5’-UUC AUCUCCAAUUCGCCGGTT-3’; Bak siRNA sequences were 5’-GCCUAUUUAAGAGUGGCAUTT-3’ and5’-AUGCCACUCUUAAAUAGGCTT-3’ ; Gpbar1 siRNA sequences were 5’-CCUACCUCUACCUGGAAGUUUTT-3’ and 5’-AAACUUCCAGGUAGAGGUAGGTT-3’; S1pr2 siRNA sequences were 5’-ACCGACAUUUCUGGAGGGUAATT-3’ and5’-UUACCCUCCAGAAAUGUCGGUTT-3’; Pink1 siRNA sequences were 5’-GCCAGACAGGGAAUGAAGUTT-3’ and 5’-ACUUCAUUCCCUGUCUGGCTT-3’; Oma1 siRNA sequences were 5’-GUAGGACUCUCAAUCACAATT-3’ and 5’-UUGUGAUUGAGAGUCCUACTT-3’; Drp1 siRNA sequences were 5’-ACGUUUCACCAGAAGAUAATT-3’ and 5’-UUAUCUUCUGGUGAAACGUTT-3’.

Sytox green and propidium iodide staining

BMDMs (3 × 10⁵ cells) were plated in a 24-well plate overnight and treated with 200 µM bile acids for various times as indicated. The cells were stained with 1 µM Sytox Green (catalog no. S7020; Thermo Fisher Scientific) for 10 min or 2 µg/mL Propidium Iodide (catalog no. P3566; Thermo Fisher Scientific) for 30 min, and visualized with a fluorescence microscope (Olympus IX 73). The image data were analyzed by ImageJ software.

RNA quantification by the real-time PCR

Total RNAs were extracted using the TRIzol Reagent (catalog no. DP424; TIANGEN) according to the manufacturer’s instruction. cDNAs were reversely transcribed from 500 ng total RNA by Hifair^® III 1st Strand cDNA Synthesis SuperMix for qPCR (catalog no. 11141ES10; YEASEN). RT-PCR was carried out with primer pairs listed below on ABI QuantStudio 6 flex RT-PCR System. The abundance of the indicated mRNA was normalized to that of Actb by the change-in-cycling-threshold (ΔΔCT) method. The data shown were the relative expression levels of specific genes normalized against the untreated or WT control groups.

The primers were as follows: mActb primers, 5’-CCAGCCTTCCTTCTTGGGTAT-3’ and 5’-AGAGGTCTTTACGGATGTCAACG-3’; mIl6 primer sequences, 5’-AGATAAGCTGGAGTCACAGAAGGAG-3’ and 5’-CGCACTAGGTTTGCCGAGTAG-3’; mCxcl10 primer sequences, 5’-CTCATCCTGCTGGG TCTGAGT-3’ and 5’-CCTATGGCCCTCATTCTCACTG-3’; mTnfa primer sequences, 5’-GTCCCCAAAGGGATGAGAAGTT-3’ and 5’-GTT TGCTACGACGTGGGCTACA-3’; mIfnb1 primer sequences, 5’-GCACT GGGTGGAATGAGACTATTG-3’ and 5’-TTCTGAGGCATCAACTGACAGGTC-3’; mCcl5 primer sequences, 5’-GTGCCCACGTCAAGGAGTATTT-3’ and 5’-CTTCxTCTGGGTTGGCACACACT-3’; mCxcl1 primer sequences, 5’-GCTGGGATTCACCTCAAGAA-3’ and 5’-CTTGGGGACACCTTTTAGCA-3’; mRsad2 primer sequences, 5’-TGGGGATGCTGGTGCCCACT-3’ and 5’-ACCCCGGACCTGTGGCTGTT-3’; mIfna4 primer sequences, 5’-CTTTCCTCATGATCCTGGTAATGAT-3’ and 5’-AATCCAAAATCCTTCCTGTCCTTC-3’; mIfna6 primer sequences, 5’-ACCCAGGATGGTCTTCAGAGA-3’ and 5’-GCCTAGCCATGGTGAGTGTT-3’; mGpbar1 primer sequences, 5’-CTGTGTGAGATCCGCCGAC-3’ and 5’-CGACGCTCATAGGCCAAGA-3’; mS1pr2 primer sequences, 5’-ATGGGCGGCTTATACTCAGAG-3’ and 5’-GCGCAGCACAAGATGATGAT-3’; mDrp1 primer sequences, 5’-TTACGGTTCCCTAAACTTCACG-3’ and 5’-GTCACGGGCAACCTTTTACGA-3’; mPink1 primer sequences, 5’- CACACTGTTCCTCGTTATGAAGA-3’ and 5’-CTTGAGATCCCGATGGGCAAT-3’; mOma1 primer sequences, 5’-TCTCTGGAGTGAATAACCTGGC-3’ and 5’-GCACTTGAGAGGCATCTTGATT-3’; mRab5a primer sequences, 5’-TGGTTCTTCGCTTTGTGAAAGG-3’ and 5’-CACAGTTTGGGTTAGAAAGGCA-3’; mRab5c primer sequences, 5’-TGGTCCTCCGCTTTGTCAAG-3’ and 5’-TGACCGTTGTATCGTCTAAGCA-3’; mBax primer sequences, 5’-CCGGCGAATTGGAGATGAACT-3’ and 5’-CCAGCCCATGATGGTTCTGAT-3’; mBak primer sequences, 5’-GCCCTGTACGTCTACCAGC-3’ and 5’-TGGCGATGTAATGATGCAGTATG-3’; hIfnb1 primer sequences, 5’- AGCTGAAGCAGTTCCAGAAG-3’ and 5’-AGTCTCATTCCAGCCAGTGC-3’, hGapdh primer sequences, 5’-CACCAGGTGGTCTCCTCTGACT-3’ and 5’- CCCTGTTGCTGTAGCCAAATTC-3’; IVA-M primer sequences, 5’-GACCRATCCT GTCACCTCTGAC-3’ and 5’-AGGGCATTYTGGAC AAAKCGTCTA-3’; IVA-NA primer sequences, 5’-TGTTGATGGAGCAAACGGAGTA-3’ and 5’-CTCAAACCCATGTCTGGAACTG-3’; IVA-HA primer sequences, 5’-GGCCCAACCACAACACAAAC-3’ and 5’-AGCCCTCCTTCTCCGTCAGC-3’.

Detection of cytosolic DNA

BMDMs (1 × 10⁷ cells) were harvested and divided into two equal aliquots after stimulation. One aliquot was resuspended in 300 µL of 50 mM NaOH and boiled for 30 min to solubilize DNA. 30 µL of 1 M tris-HCl (pH 8.0) was added to neutralize the pH and then centrifuged at 12,000 rpm for 10 min to pellet the cells. In addition, these extracts served as normalization controls for total gDNA and mtDNA. The second aliquots were resuspended in 300 µL of buffer containing 150 mM NaCl, 50 mM HEPES (pH 7.4), and digitonin (15 to 25 mg/mL; catalog no. D141 Sigma-Aldrich). The homogenates were incubated for 10 min on ice to allow selective plasma membrane permeabilization and then centrifuged at 980 g for 3 min to pellet the cells. Last, the cytosolic supernatants were transferred to fresh tubes and spun at 17,000 g for 10 min to pellet any remaining cellular debris, and the cytosolic fractions were free of nuclear, mitochondrial, and endoplasmic reticulum.

Quantitative real-time PCR was performed on the DNAs from both whole cell extracts and cytosolic fractions using gDNA primers (Tert: 5’-CTAGCTCATGTGTCAAGACCCTCTT-3’ and 5’-GCCAGCACGTTTCTCTCGTT-3’; Polg1: 5’-GATGAATGGGCCTACCTTGA-3’ and 5’-TGGGGTCCTGTTTCTACAGC-3’) and mtDNA primers (Cox1: 5’-GCCCCAGATATAGCATTCCC-3’ and 5’-GTTCATCCTGTTCCTGCTCC-3’; Nd2: 5’-CCATCAACTCAATCTCACTTCTATG-3’ and 5’-GAATCCTGTTAGTGGTGGAAGG-3’), and the CT values for whole-cell extracts served as normalization controls for the values obtained from the cytosolic fractions.

Confocal fluorescence microscopy

The stimulated cells were fixed with 4% paraformaldehyde (catalog no. P0099; Beyotime) for 30 min at room temperature, permeabilized with 0.1% Triton X-100 (catalog no. P0096; Beyotime) for 10 min and blocked with 5% BSA for 1 h at room temperature. Subsequently, the fixed cells were probed with primary antibodies overnight at 4 ℃, stained with the secondary antibodies for 1 h at room temperature and incubated with Antifade Mounting Medium with DAPI (catalog no. P0131; Beyotime).

For the measurement of mitochondrial morphology, the cells were stained with Mito-Tracker Deep Red FM (catalog no. C1032; Beyotime) (5 µg/mL) for 30 min before fixed with 4% paraformaldehyde.

The fluorescence images were collected with a laser capture confocal microscope (Olympus SpinSR10 Ixplore) using separate laser excitation to avoid any cross-interference between different fluorophores. The data were analyzed by ImageJ software.

Measurement of mitochondrial membrane potential

Mitochondrial membrane potential was analyzed with specific fluorescent probes for TMRM (Tetramethylrhodamine methyl ester). BMDMs (2 × 10⁵ cells) were plated in 24-well plate overnight and treated with 200 µM DCA or CDCA for 6 h, or with 10 µM FCCP for 40 min as the positive control. Subsequently, the cells were treated with 100 nM TMRM fluorescence probe (catalog no. 22221; AAT Bioquest) for 30 min at 37 ℃. The fluorescence density of TMRM was visualized by a fluorescence microscope (Olympus IX 73) at Ex/Em = 552/574 nm. The data were analyzed by ImageJ software.

Cell lysate preparation and immunoblotting

2 × 10⁶ cells BMDMs were plated on 6-cm dishes overnight, then treated with 200 µΜ DCA for various times as indicated. The treated cells were washed twice with cold 1⋅PBS and lysed in the lysis buffer [50 mM Tris (pH 7.4), 150 mΜ NaCl, 1% Triton X-100, and 1mΜ EDTA (pH 8.0)] supplemented with protease inhibitor complete Protease Inhibitor Cocktail (catalog no. 11836153001; Roche) and 1 mM Phenylmethylsulfonyl fluoride (PMSF), 1 mM Na3VO4, and 1 mM NaF for 30 min on ice, and cell debris was removed by centrifugation at 13,000 rpm for 25 min. After being boiled at 95 °C for 5 min, the samples were separated by 10% SDS-PAGE and transferred onto a PVDF membrane and then probed with corresponding antibodies.

Protein harvest and digestion for phospho-proteomics

2 × 10⁶ cells BMDMs were plated on 6-cm dishes overnight, and then treated with 200 µM DCA or vehicle for 3 h. Cells were washed with cold PBS and lysed in the lysis buffer contains 4% (wt/vol) SDC and 100 mM Tris-HCl (pH 8.5) at 4 ℃ to achieve a protein concentration of ~ 2–5 mg/mL. 200 µg protein lysate were used for digestion. Proteins were reduced and alkylated using 10 mM tris-(2-carboxethyl)-phosphine and 20 mM 2-chloroacetamide for 5 min at 45 ℃ with shaking by 600 rpm. The samples were cooled to room temperature and added Lys-C and trypsin at an enzyme-to-substrate ratio of 1:50 (wt/wt), and digested overnight at 37 ℃ with shaking by 600 rpm, and stopped using 5% formic acid. The generated peptides were desalted using the Waters Sep-Pak WAT054960 and the phosphorylated peptides were enriched using High-Select (Thermo Scientific, A32993). Elution was collected for vacuum centrifuge to dryness.

Mass spectrometric analysis (LC-MS/MS) of phosphopeptides

All nano-LC-MS/MS experiments were performed on an Orbitrap Eclipse (Thermo Scientific) equipped with a Vanquish Neo UPLC system (Thermo Scientific). Peptides were loaded onto a reversed-phase C18 analytical column (20 cm in length, 100 μm inner diameter (id)) for separation. The peptides bound to the analytical column were eluted using a 90-minute linear gradient. Phase A consisted of an aqueous solution containing 0.1% formic acid, and Phase B consisted of an aqueous solution containing 80% acetonitrile (with 0.1% formic acid). The gradient elution program was as follows: At a flow rate of 450 nL/min: Phase B was increased from 1% to 4% for 0.2 min, then further increased from 4% to 10% for 24.8 min; from 10% to 24% for 42 min, from 24% to 35% for 16 min, from 35% to 55% for 1 min; At a flow rate of 500 nL/min: Phase B was increased from 55% to 99% for 1 min, then maintained at 99% for 5 min.

Mass spectrometry analysis was conducted using an Orbitrap Eclipse mass spectrometer (Thermo Scientific). The parameters of the nano-electrospray ionization (nano-ESI) source were set as follows: spray voltage of 2.2 kV; no sheath gas; ion transfer tube temperature of 320 ℃. Data-independent acquisition (DIA) mode was adopted. MS (First-order mass spectrometry) parameters: Scan range of 380–985 m/z; resolution of 60,000 (at m/z 200); automatic gain control (AGC) target of 4.00E + 05; maximum injection time of 50 ms. MS/MS (Tandem mass spectrometry) parameters: Orbitrap was used as the mass analyzer; the scan was divided into 60 fixed isolation windows, with an isolation window width of 10 m/z and a window overlap of 1 m/z; MS/MS scan range of 145–1450 m/z; resolution of 30,000 (at m/z 200); higher-energy collisional dissociation (HCD) collision energy of 30%; AGC target of 2.50E + 05; maximum injection time of 100 ms.

Data processing and bioinformatic analysis of the phospho-proteomics

Data acquired on the Orbitrap Eclipse were processed with DIA-NN (v. 1.8.2 beta 39) in a library-free mode. An in silico spectral library was generated using Trypsin/P specificity (max. 2 missed cleavages), with carbamidomethylation of cysteine as a fixed modification and phosphorylation of serine, threonine, tyrosine, oxidation of methionine and protein N-terminal acetylation as variable modifications. Peptides were restricted to 7–30 amino acids, precursor charges 1–4, and m/z 300–1000. In the main search, mass accuracies and scan windows were automatically determined. Key parameters included cross-run normalization, and enabled match-between-runs (MBR). Proteins were inferred and grouped at the gene level using heuristics. The final protein group report was filtered to a q-value < 0.01, based on a global precursor-level FDR of 1%.

Raw phosphor-proteomic intensities from untreated or DCA-treated samples (two biological replicates per condition) were preprocessed to remove instrument-derived floor effects. Each sample was median-normalized to minimize technical variation. Differential phosphorylation levels were calculated as log₂ fold change (log₂FC), and p-values were computed using Welch’s t-test.

A scatter plot was generated to visualize all quantified phosphosites after floor-value filtering. The x-axis represents log₁₀-transformed baseline intensity (mean of untreated replicates), and the y-axis represents log₂FC (DCA-treated vs. untreated). Each dot corresponds to a phosphosite, and color indicates –log₁₀ (p-value). Two vertical dashed lines mark the fold-change thresholds for upregulation and downregulation (log₂FC = log₂1.5 and log₂0.7), serving as visual reference boundaries. This plot illustrates the global phosphorylation distribution independent of statistical filtering.

A Z-score normalized heatmap was constructed using the ComplexHeatmap package (R). Columns were arranged with untreated replicates on the left and DCA-treated replicates on the right. Phosphosites showing opposite replicate patterns (e.g., NT-1 high/NT-2 low, DCA-1 high/DCA-2 low) were automatically removed. The remaining phosphosites were scaled by row (Z-score), clustered using Euclidean distance, and visualized to reveal consistent group-level changes.

For biological interpretation, Gene Set Enrichment Analysis (GSEA) was performed using the fgsea algorithm based on KEGG and Gene Ontology (GO Biological Process) databases. Pathways with p < 0.25 were considered significant. Enrichment results were visualized as bubble plots, where bubble size represents the number of core enrichment genes (Count), color denotes the normalized enrichment score (NES), and the x-axis shows the GeneRatio (core genes / total genes in each pathway).

In vivo DCA treatment

8- to 12-week-old C57J/B6 WT mice were intraperitoneally injected with various does of DCA (10, 30 or 50 mg/kg). In 3 days, the mice were euthanized and various tissues were removed, rinsed and homogenized for RNA preparation by TRIzol.

8- to 12-week-old sex-matched wild-type (WT) and Sting1^−/− mice were intraperitoneally injected with one dose of DCA (50 mg/kg), and their weight change and survival were monitored and documented every day.

LC-MS/MS quantification of DCA

BMDMs (1 × 10⁶ cells) were pretreated with digitonin (10 µg/mL) or not for 30 min and then stimulated with 200 µM DCA for 6 h. After that, cells were washed twice with cold PBS and harvested. The cells were lysed in 600 µL of lysis buffer (80% methanol and 20% ddH₂O) pre-chilled to 80 °C and vortexed for 30 s. Following centrifugation at 132,000 rpm for 30 min, the supernatants were collected and dried by a speed vacuum. After reconstitution with 30 µL anhydrous methanol by vertexing thoroughly, the DCA-containing samples were analyzed via a UHPLC-ESI-MS/MS system consisting of a TSQ Quantis mass spectrometer (Thermo Scientific, USA) with an electrospray ionization source (ESI) and an VanquishTM Flex UHPLC System (Thermo Scientific, USA). Data acquisition and processing were performed using Trace Finder software (version 5.1, Thermo Scientific, USA). The UHPLC separation was carried out on a Thermo Scientific™ Hypersil GOLD™ C18 100 × 2.1 mm, 1.9 μm column with a flow rate of 0.35 mL/min at 40 °C.

The analysis was performed using a gradient solvent system consisting of 0.1% Formic acid in water (Phase A) and an acetonitrile/methanol mixture (1:1) (Phase B). The column was eluted from 0 to 6 min with a linear gradient of 40–55% solvent B, from 6 to 15 min using a linear gradient of 55–70% solvent B, from 15 to 16 min with a linear gradient of 70–98% solvent B, and from 16.1 to 20 min using a linear gradient of 98–40% solvent B. Mass spectrometry detection was performed in negative electrospray ionization mode. The mass transition for DCA was set to m/z 391.28 > 391.28, 343.321, 345.321. Negative ions were acquired in the MRM mode with a vaporizer temperature of 350 °C and an ion transfer tube temperature of 325 °C. Spray voltage was 3200 V. Sheath gas and aux gas were set to 35 and 15 Arb, respectively.

ELISA assay

8- to 12-week-old C57J/B6 WT mice were intraperitoneally injected with various doses of DCA (10, 30 or 50 mg/kg). In 3 days, the mice were euthanized, and the sera were collected. The levels of IFN-β and CXCL-10 in serum were assessed using ELISA kits according to the manufacturers’ recommendations (IFN-β catalog no. 439407; BioLegend and CXCL10 catalog no. 900-K153; PeproTech).

Tumor implantation, treatment and documentation

8–10 weeks old WT male mice were subcutaneously inoculated with 1 × 10⁶ MC38 in 100 µL phosphate-buffered saline (PBS catalog no. SH30256.01; HyClone). Tumors reached about 100 mm³ ((length*width²)/2) on day 6. Mice were then randomly grouped and injected intratumorally with 100 µL of DCA (125 µg) or PBS every 2 days. The tumor volumes were measured by a Vernier scale every 2 days and analyzed by two-way ANOVA (about 30 days).

Tumor immune cell analysis

Tumors were cut into small pieces with ophthalmic forceps, and digested with 1 mg/mL collagenase IV (catalog no. 17104019; Gibico) and 0.2 mg/mL deoxyribonuclease (DNase) I (catalog no. DN25; Sigma-Aldrich) for 30 min, and filtered with 70 μm cell strainers to obtain the single-cell suspensions. Tumor-derived single-cell suspensions were separated by Percoll and density gradient centrifugation. Enriched immune cells were collected and counted for live cell numbers by a Cell Counter (BIO-RAD). For surface marker staining, immune cells prepared as above were incubated with anti-mouse CD16/32 (catalog no. 156603, S17011E; BioLegend) at 4 °C for 15 min, washed with fluorescence-activated cell sorting (FACS) buffer (2% FBS in PBS), and then incubated at 4 °C for 30 min with the antibody cocktail containing the Live/dead dye (catalog no. 423101, BioLegend), anti-CD45, anti-CD3, anti-CD8, anti-CD4, anti-CD19, anti-MHC-II, anti-CD11c, anti-F4/80, anti-CD11b, anti-NK1.1, anti-Ly6G and anti-CD80. The data were acquired by a Multicolor flow cytometry (Aurora, Cytek) and analyzed with FlowJo software (Tree Star).

Influenza A virus (PR8) intranasal infection

For intranasal infection, mice were anesthetized via Avertin at a standard dosage (250 mg/kg body weight). Before IVA infection, mice were intranasally inoculated with DCA (125 µg) or PBS for two consecutive days. After the second DCA or PBS treatment, mice were intranasally infected with IVA (PR8, 250000 PFU/mouse). On day 3 post-infection, lung tissues were collected for RNA preparation by TRIzol.

Quantification and statistical analysis

All data are shown as mean ± SD or SEM, and error bars represent the SD or SEM of biological repeats. For statistical comparisons, the data were analyzed by two-way ANOVA with Bonferroni’s multiple comparisons test or paired Student’s t-test (*p < 0.05; **p < 0.01; ***p < 0.001; ns, not significant).

Abbreviations

CA

Sodium Cholate

DCA

Sodium deoxycholate

CDCA

Sodium chenodeoxycholate

LCA

Lithocholic acid

UDCA

Ursodeoxycholic acid

TCA

Sodium taurocholate

TDCA

Sodium taurodeoxycholate

TCDCA

Sodium taurochenodeoxycholate

GCA

Glycocholic acid

GDCA

Sodium glycodeoxycholate

GCDCA

Sodium glycochenodeoxycholate

ER

Endoplasmic Reticulum

FXR

Farnesoid X Receptor

VDR

Vitamin D Receptor

LXR

Liver X Receptor

TGR5

Takeda G protein-coupled receptor 5

PMCA

Plasma Membrane Ca²⁺-ATPase

S1PR2

Sphingosine-1-Phosphate Receptor 2

AR

Androgen receptor

MEF

Murine embryonic fibroblasts

BMDM

Bone marrow-derived macrophages

HMDM

Human monocyte-derived macrophages

MC38

Murine carcinoma 38

gDNA

Genomic DNAs

mtDNA

Mitochondrial DNAs

EtBr

Ethidium Bromide

dsDNA

Double-stranded DNA

MAM

Mitochondria-associated endoplasmic reticulum membrane

VDAC

Voltage-dependent anion channel

mPTP

The mitochondria permeability transition pore

CsA

Cyclosporine A

TMRM

Tetramethylrhodamine methyl ester

FCCP

Carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone

ROS

Reactive Oxygen Species

MβCD

β-Cyclodextrin

2-APB

2-Aminoethyl diphenylborinate

Lacl₃

Lanthanum (III) chloride

Fer-1

Ferrostatin-1

Authors’ contributions

H. X. designed and supervised the overall experimental design and execution. J.N. H. performed the experiments and analyzed the data with assistance from Z.Y. H., C.W. X., Z.C. H., H. Y., B. S., X. L., Y. C., X. C., H.B. L. and L.M. C. H. X., J.N. H. and L.M. C. wrote the manuscript . All authors commented on the manuscript.

Funding

This study was supported by the National Key Research and Development Program of China (2021YFA1301400); the National Natural Science Foundation of China grants (92574110; 32030040); the Strategic Priority Research Program of the Chinese Academy of Sciences (XDB0940102).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.