The role of cGAS/STING in intestinal immunity

The gastrointestinal tract is a highly complex microenvironment under constant interaction with potentially harmful pathogens. Inflammatory bowel disease (IBD) is an archetypical inflammatory disease, in which the intestinal epithelium, defective autophagy, endoplasmic reticulum stress and dysbiosis play a key role.


Introduction
The gastrointestinal tract represents a major entry point for pathogens and forms a vital first line of defense due to its constant exposure to foreign material and opportunistic commensal microbes [1]. The intestinal mucosa, consisting of a single line of polarized intestinal epithelial cells (IECs) and the subepithelial lamina propria, separates the submucosal tissue from the outside world, that is, the gastrointestinal lumen, and represents the primary site of inflammation in inflammatory bowel disease (IBD). While forming a tight barrier towards such threats, the intestinal epithelium also performs nutrient and water absorption and, importantly, actively engages in gastrointestinal immunity, that is, host-microbe crosstalk and effective immune defense against pathogens. Throughout the past decade, evidence has emerged to support the importance of type I IFN in intestinal defense and homeostasis, as well as implicating dysregulation of type I IFN responses in various immune-mediated disorders including IBD [2][3][4]. Type I IFNs, including its two major constituents IFN-α and IFN-β, are a large family of IFNs exerting a broad range of biological effects including antiviral and distinct antibacterial function [5]. Additionally, they have also been shown to act as a potent mediator of inflammation in autoimmunity [6,7]. Moreover, immunosurveillance of genome integrity has been identified as an important type I IFN-mediated mechanism during DNA damage and cancerogenesis [8][9][10]. Concerning the intestinal mucosa, type I IFNs appear to be involved in balancing between Figure 1. Activation modes of STING signaling in the intestinal tract. STING signaling results as a consequence of dsDNA sensing of the upstream adaptor molecule cGAS. dsDNA can derive from non-self-origins such as bacteria and viruses. In addition, dsDNA can also be of self-origin either as mitochondrial DNA or genomic DNA as a result of genotoxic stress. Activated STING signaling induces secretion of type I IFNs, which in turn signal in a para-and anticrime manner and mediate antibacterial and antiviral immunity, cellular differentiation, and participate in barrier function regulation.
immune tolerance and defense, as well as in orchestrating proliferation, differentiation, and overall barrier function, suggesting an important role of type I IFNs in maintaining a healthy, but alert mucosa [5,[11][12][13].
A growing number of studies indicate an important role of cGAS/STING signaling (cyclic GMP-AMP synthase/stimulator of interferon genes), a main inducer of type I IFNs in response to cytosolic DNA or bacterial cyclic dinucleotides (CDNs), in gastrointestinal homeostasis [14]. However, it remains to be elucidated, whether and to which extent the cGAS/STING/type I IFN axis contributes to the pathophysiology of IBD. In this review, we first introduce the cGAS/STING pathway and research pointing toward the involvement of STING in human IBD, furthermore highlighting recent studies investigating cGAS/STING-mediated type I IFN signaling in the gastrointestinal tract. Next, we demonstrate shared features between epithelial IBD pathophysiology and STING signaling and subsequently speculate on so far unknown aspects of STING function to put forward a concept of how cGAS/STING signaling might be involved in intestinal epithelial biology and IBD.

Origins of cGAS/STING activity and type I IFNs in the intestine
General nucleic acid sensing and the pathogen recognition receptors (PRRs) resulting in type I IFN release, including the cGAS/STING pathway, has recently been extensively reviewed elsewhere [14][15][16]. Therefore, we will only briefly cover dsDNA sensing mechanisms and proceed to review the implications of the cGAS/STING pathway in mucosal immunity in more detail.
In the intestine, like in most tissues, levels of type I IFNs at steady state are low [32][33][34]. However, in the case of infection or cell damage, type I IFN induction is a rapidly initiated process [35]. Secreted type I IFN in turn signals in a para-and autocrine manner via binding to its heterodimeric type I IFN receptor (IFNAR), comprising an IFNAR1 and IFNAR2 subunit, which results in the induction of IFN-stimulated genes (ISGs) via JAK/STAT-signaling [11,36]. The downstream effects of type I IFN secretion are ultimately mediated by the expression of a large subset of immunomodulatory, antiviral, or antiproliferative ISGs ( Fig. 1) [37-39]. Recently, a study focusing on lung immunity has depicted microbially triggered constitutive type I IFNs as an important mechanism to maintain general immune alertness in mice and facilitate protection against Influenza virus infection [40]. Similar effects have been described for the intestine both in mice and humans, where type I IFNs can evoke a beneficial role regulating adaptive immunity, including regulatory T cells, for example, by maintaining FoxP3 expression or regulating cytokine production by T cells [34,41,42]. Interestingly, in mucosal inflammation, a lack of type I IFN signaling resulted in higher cytokine release in effector T cells, suggestive of an altered adaptive immune response in human IBD as a consequence of altered type I IFN levels [42]. In the intestinal mucosa, DCs and mononuclear macrophages localized in the lamina propria are predominantly responsible for type I IFN release during steady state [34, 43,44]. Conversely, intraepithelial lymphocytes have been shown to produce type I IFN following TCR activation during norovirus infection in mice [45].
Importantly, little is known about the dimension and implication of type I IFNs derived from the intestinal epithelium. Paneth cells are particularly critical for intestinal homeostasis and play a key role in gastrointestinal inflammatory processes and might be of interest as they are not only affected by, but also seem to express type I IFNs and ISGs in mice and humans [13,33,46,47]. Likewise, while some studies indicate a potential role of STING in epithelial cells [2, 48,49], dimension and implication of intestinal STING expression, including function, are essentially unknown and remain subject to speculation, regardless if in the colon or small intestine. STING can directly bind to bacterial cGAS-synthesized CDNs, which are used as second messengers in bacteria, establishing a role for STING as an independent pattern recognition receptor ( Fig. 1) [50][51][52][53]. Such CDNs are derived from intracellular bacteria that escape into the cytosol, for example Listeria monocytogenes. In line with this, type I IFN production is a hallmark of L. monocytogenes infection in both human and murine cells [53][54][55].
In host viral defense, cGAS/STING signaling is essential for proper type I IFN response and subsequent ISG induction, which in turn prevents replication, assembly, and release of DNA viruses ( Fig. 1) [56]. Such a critical role for STING in viral infection has been described for a broad number of viruses in both mice and humans, for example the Herpes-simplex virus (HSV), and many others [14,18,[57][58][59][60][61]. Thus, mice deficient for cGAS or STING show severely impaired ability to induce a proper type I IFN response during viral infection and thereby suffer from heightened infection susceptibility and aggravated disease course.
In addition, cGAS/STING signaling can be activated not only by microbial DNA, such as DNA from viruses or intracellular bacteria, but also by extracellular, mitochondrial, and nuclear DNA that is sensed within the cytosol (Fig. 1) [14]. Cytosolic dsDNA levels increase in response to mitotic stress in cancer, genomic instability, radiation therapy, or during genetically driven autoimmune disease resulting in IFN-driven inflammation, commonly summarized as interferonopathies, illustrating the proinflammatory potential of type I IFNs [6]. The cGAS/STING pathway has been linked mechanistically to one of these interferonopathies, namely Aicardi-Goutières Syndrome (AGS) [62][63][64][65]. AGS is a genetically and clinically heterogeneous disorder associated with neurological and dermatological inflammatory phenotypes [6]. Among multiple mutations potentially triggering AGS, loss-offunction mutations in the exonuclease three prime repair exonuclease 1 (TREX1) or endonuclease RNASEH2A result in dsDNA leaking from the nucleus and thereby trigger a cGAS/STING/type I IFN-mediated inflammatory reaction [6, 62,65]. Accordingly, mouse models harboring deletions of either Cgas or Tmem173 rescue Trex1-deficient mice from a lethal autoimmune phenotype [62,63,66]. Along the line, RNase H2-deficiency-mediated perinatal death is also alleviated by additional Tmem173 KO [65], illustrating the disastrous proinflammatory potential of this pathway. Interestingly, AGS patients can also present with IBD-like symptoms, and we have recently shown that mice with an epithelial deletion of Rnaseh2b, resulting in a defective RNase H2 endonuclease complex, phenocopy such IBD-like intestinal inflammation [67,68].
Further stressing the proinflammatory potential of STING signaling, gain-of-function mutations in TMEM173 are responsible for a rare autoinflammatory disorder named SAVI (STINGassociated vasculopathy with onset in infancy) [7]. SAVIassociated TMEM173 mutations result in constitutive STING activation without the presence of any ligands, leading to permanent STING ER-exit, trafficking, and activation [27]. Hyperactivation of STING as a consequence of these gain-of-function mutations and subsequent type I IFN production then results in sterile inflammation, mainly affecting the skin and the lungs [7,69]. In opposition to AGS, despite severe systemic inflammation, neither gastrointestinal lesions have been described in the original publication, nor have gastrointestinal affections been evaluated and described in any mouse models investigating SAVI [7, [70][71][72][73].
Differing from Toll-like-receptor 9 (TLR9), which senses dsDNA delivered into the cell by phagocytosis, STING requires an intracellular cytosolic trigger to be activated, which is commonly provided via DNA-sensing-derived CDNs (Fig. 1) [74,75]. In addition to these triggers, recent studies have extended the spectrum of CDN delivery possibilities. Gap junctions, consisting of poreforming connexins, allow intercellular transfer of CDNs in human and murine cells and thereby are able to multiply STING activation within tissue [76]. Furthermore, specific carriers importing CDNs from the extracellular compartment into the cytosol have been identified, namely SLC19A1 in human cells (Soluble Carrier Family 19 Member 1) and LRRC8/VRAC (Leucine Rich Repeat Containing 8 Volume Regulated Anion Channel) in mice [77][78][79]. These novel CDN delivery methods could be of interest in the context of the intestinal epithelial physiology, in which a single layered epithelial line is densely connected by gap junctions. In this context, gap junction-mediated delivery of CDN could potentially enable a synchronized STING activation in response to cellular entry of harmful pathogens.
The import of CDNs from extracellular could likewise be of great interest, for example, during proinflammatory cell death, where CDNs might be released by neighboring dying cells.

A possible role for STING signaling in the pathophysiology of IBD?
IBD, comprising the two main entities ulcerative colitis (UC) and Crohn's disease (CD), are chronic-relapsing inflammatory diseases affecting mainly the gastrointestinal tract with varying degrees of extraintestinal manifestations. IBD usually progresses with alternations of acute episodes with symptom-free intervals [80,81]. IBD pathophysiology is characterized by a complex interaction of genetic predisposition, environmental risk factors, disturbed microbial ecology, and immune dysregulation [82], which remains to be fully understood. IBD pathophysiology is commonly thought to involve a dysregulated, overshooting immune response, that is, inflammation, in response to commensal or pathogenic microbes enabled by individual genetic susceptibility leading to defective stress resolution, thus, lower resilience, triggered and exacerbated by environmental stress factors [83].
What direct evidence is there suggesting a causal implication of STING and type I IFN signaling? While TMEM173 variants leading to excessive STING activity have been linked to autoinflammatory disorders [6,7], genome-wide association study (GWAS) has not found TMEM173 variants associated to IBD. However, TMEM173 has been found to be hypomethylated in the intestinal epithelium in a cohort of pediatric IBD patients, suggesting an overexpression (and possibly overactivity) of STING in the epithelium of human IBD patients [48]. Matching this idea, IBD patients display a signature of IFN-regulated genes and higher IFN signatures are associated with a lack of therapy response [2-4]. In addition, the IFN-regulated gene ISG15 (Interferon-stimulated gene 15) is overexpressed in IBD patients with active inflammation, along with a broad signature of ISGs [3].
Next, a growing number of studies have investigated STING function in different mouse models involving gastrointestinal inflammation or injury, providing further clues supporting the idea that a potential overexpression of STING in IBD might translate to a causally involved clinical phenotype. This will be discussed in the section "cGAS/STING, type I IFNs, and the microbiota in mouse models of gastrointestinal perturbations".
Lastly, STING function and regulation and IBD pathophysiology share a considerable amount of overlapping features. There is extensive knowledge regarding perturbations in ER-stress, autophagy, and cell death in IBD and numerous studies connecting cGAS/STING signaling with the unfolded protein response (UPR), autophagy, and cell death induction. Therefore, it becomes increasingly interesting to study the role of (epithelial) STING in IBD and speculate on potential connections. These rather hypothetical aspects will be discussed in the sections "Interplay of endoplasmic reticulum stress, autophagy, and STING" and "Role of STING in cell death and survival".

cGAS/STING, type I IFNs, and the microbiota in mouse models of gastrointestinal perturbations
Physiologically, type I IFNs fulfil antiviral, antiproliferative, and immunomodulatory functions ( Fig. 1) [11,38,84]. Identification of the rs2284553 single nucleotide polymorphism linked to IBD, predicted to result in a loss-of-function mutation [85], might affect the type I IFN receptor gene IFNAR1 and thus suggests a causal implication of type I IFN signalling for the development of inflammation and IBD [13]. Furthermore, ISG expression in IBD patients correlates with disease activity and impaired therapy response [2,4], suggesting that uncontrolled IFN production underlies therapy-refractory intestinal inflammation. While abrogation of type I IFN signaling in Ifnar1 -/mice impairs disease course in dextran sulfate sodium (DSS)-induced colitis, Ifnar1 IEC mice, however, are not more affected by DSS treatment than Ifnar1 fl/fl mice [13,86]. Instead, abrogation of epithelial IFNAR1 signaling results in epithelial hyperproliferation, leads to increased Paneth cell and Goblet cell numbers, and ultimately, to increased intestinal carcinogenesis [13] (Fig. 2) [38]. Of note, disruption of type I IFN signaling in Ifnar1 IEC mice also leads to an altered microbial composition, suggesting a crucial role of type I IFNs in shaping the microbial environment (Fig. 2) [13]. While this study does not elucidate the source of type I IFNs during intestinal inflammation, it highlights the complex properties of type I IFNs on mucosal immune instruction. Bearing in mind the proinflammatory potential of dysregulated type I IFN signaling in interferonopathies and the detrimental effect of type I IFNs during some bacterial infections, one would have expected to observe a beneficial effect in the context of Ifnar1 deficiency. Instead, Ifnar1 deficiency impaired DSS colitis outcome, while abrogation of only epithelial Ifnar1 did not [13,86]. This points towards the idea that the intestinal epithelium, although possibly a source of intestinal type I IFN, might not be the mediator of type I IFN-induced beneficial effects during DSS colitis. On Putative molecular interaction points between STING signaling and IBD. ER-stress: ER-stress may manifest as a molecular consequence of STING signaling in response to the intracellular invasion of bacteria. In the context of chemical stress (e.g. ethanol), unresolved ER-stress instigates STING activation. In T-cells, STING activity disrupts ER calcium homeostasis and thereby predisposes for ER-stress. In IBD, several genes involved in ER-stress and the unfolded protein response have been identified as risk genes (e.g. ORMDL3, AGR2, XBP1). Autophagy: Autophagy plays a dual role in coordinating STING functions in the intestinal mucosa. Autophagy might provide a mechanism to shut down STING signaling and thereby limit STING-dependent type I IFN production. Autophagy is also a crucial executer of STING signaling, mediating pathogen clearance. Various IBD risk genes are related to autophagy such as ATG16L1, IRGM, LRRK2, or ATG7. Barrier function and microbiota interaction: Tmem173 −/− mice display increased colitis severity, which includes an altered intestinal microbiota. In several models of intestinal injury or perturbation, STING displays both favorable and detrimental effects. Cell death: STING signaling can induce various types of cell death including apoptosis, pyroptosis, and necroptosis. Whether a direct connection to other cell death modalities such as ferroptosis exist or not, is currently unknown. the other hand, subepithelial immune cells might be crucially dependent on beneficial type I IFN and IFNAR1 signaling, which would translate to greater suffering during DSS colitis [86]. Supporting the idea that type I IFNs maintain proper immune alertness, a recent study in mice has shown that constitutive type I IFN signaling is maintained by the microbiota and required for basal alertness of DCs. This process subsequently involved IFN-I dependent CD8 + T-cell instruction [87]. Although this study has not explicitly examined its consequences on intestinal immunity with regard to intestinal inflammation (e.g. experimental colitis models), this study illustrates a mechanistic role of the intesitinal microbiota to prime the mucosal immune system via IFN-I. STING activation requires a strict balance of positive and negative regulatory mechanisms to prevent inflammation as a consequence of excessive signaling. While the effects of uncontrolled STING activation have been extensively studied in cases of human autoinflammatory disease [7, 69,[88][89][90], several mouse models of intestinal injury have been employed to reveal STING function in intestinal homeostasis [49,[91][92][93][94]. Although to date, there is no data available investigating STING function in specific cell types, multiple studies on Tmem173 -/and Tmem173 gt mice (harboring a loss-of-function mutation ["goldenticket"] within the Tmem173 gene leading to a functional KO of STING [52]) support a critical role for STING in maintaining intestinal homeostasis as well as in induction and propagation of intestinal inflammation. However, the exact factors unravelling beneficial or detrimental aspects of STING signaling yet remain to be determined.
A recent study by Martin et al. investigated severity of inflammation using a DSS colitis model on Tmem173 gt mice [91]. In line with the possibly detrimental effects of uncontrolled STING signaling and type I IFN hypersecretion, the Tmem173-mutant mice were protected from colitis compared to the control mice. Accordingly, treatment with the STING agonist DMXAA (5,6dimethylxanthenone-4-acetic acid) in the STING-proficient mice demonstrated strong proinflammatory capacity of STING signaling [91]. Conversely, in another study by Canesso et al., STING was reported as a protective factor during colonic inflammation [94]. In their model (using Tmem173 -/mice), STING-deficient mice were highly susceptible to DSS colitis, and also presented with a greater proinflammatory microbial composition, indicating that this protective role of STING is possibly due to licensing of the resident intestinal microbiota (Fig. 2) [94]. Importantly, and in opposition to the study by Martin et al., Canesso et al. Tmem173 -/and C57BL/6 Wt mice were cohoused for 4 weeks before beginning the DSS colitis, which might account for the different outcome of the colitis severity. Given the dense reciprocal crosstalk of STING and the microbiota, cohousing conditions and thereby a reciprocal exchange of intestinal microbiota via coprophagy might very well mediate higher disease severity in the Tmem173 -/mice compared to the beneficial effect of defective STING signaling in the study by Martin et al.
Another potentially protective role of STING signaling in mucosal immunity is shown in experimental models of graft-versus-host disease, in which Tmem173 gt mice displayed increased mortality compared to control mice [49]. This phenotype coincided with dsDNA-mediated improval of epithelial barrier function in vivo and intestinal organoid growth in vitro, respectively. Hence, these data point toward a protective role of type I IFN signaling in intestinal inflammation. Although the responsibility of epithelial STING for the in vivo phenotype remains elusive, the data obtained from organoids point toward a hypothetical role for epithelial STING in regulating epithelial barrier growth, regeneration, and function (Fig. 2) [49].
A proinflammatory role for STING has been reported by Hu et al. using a model of intestinal injury-induced sepsis [92]. Here, STING promoted microbial translocation, gut permeability, and intestinal epithelial cell death (Fig. 2). Tmem173 KO accordingly alleviated sepsis symptoms, while additional pharmaceutical stimulation of STING with a STING agonist further aggravated the inflammatory response [92]. In another study using a genetic colitis model (IL-10 KO mice), additional deletion of STING resulted in protection from spontaneous colitis (Fig. 2) [93]. The authors concluded that STING was engaged in microbial sensing and the transmission of proinflammatory signals [93], supporting the concept that STING might rather promote than restrict intestinal inflammation.
As discussed above, STING can be activated by DNA derived from commensal microbes [94], and STING deficiency results in an intestinal microbiota associated with higher susceptibility to inflammation. Taken together, these studies imply (1) that STINGmediated transmission of microbial signals is key for proper surveillance of the intestinal microbiota, and (2) that excessive mucosal STING signaling rather contributes to disease exacerbation and tissue destruction via type I IFN (Fig. 2). In line with this, a recent study from our group demonstrated that unrestrained STING signaling due to impaired autophagy-mediated termination of STING signaling induced IL-22-dependent cell death and intestinal inflammation, which is discussed in more detail below [2].
These apparently conflicting results [91,94] underscore the necessity for a deeper understanding of STING function during inflammatory responses in the gut. They also highlight the complexity of both beneficial or detrimental effects of type I IFNs and the necessity to disentangle the role of type I IFNs from STING function in the intestine. Given the dense crosstalk involving STING, type I IFNs and the microbiota [13,94], it seems highly likely that shifts in microbial composition and function could act as a switch determining the fate of STING signaling and possibly type I IFN effects (Fig. 2).
While these studies provide first insights into the range of phenotypes possibly associated with altered STING function, well known key aspects of IBD pathophysiology display a large overlap with regulatory features involved in STING signaling. Therefore, by highlighting shared characteristics, we now proceed to speculate on how STING might be affected by IBD pathophysiology and what consequences this might have for inflammatory phenotypes and therapy.

Interplay of endoplasmic reticulum stress, autophagy, and STING
STING is an ER-resident protein with four N-terminal transmembrane domains. Upon oligomerization, STING exits the ER towards the Golgi, which crucially depends on canonical COPIIcoat complex-dependent anterograde transport [14]. The transport to the Golgi is facilitated via the ER-Golgi intermediate compartment (ERGIC) [14]. It seems therefore conceivable that disrupted homeostasis of the ER might affect STING signaling events in the intestinal epithelium. ER stress refers to an unresolved accumulation of unfolded proteins in the ER, which can be triggered by both environmental factors, for example, smoking [95] or infection [96][97][98][99], and internal factors, for example, defects in the UPR or autophagy themselves [46,[99][100][101][102], and specifically impairs cell types with high secretory activity [102][103][104]. Likewise, cells with high secretory activity, such as Paneth or Goblet cells, are more prone to suffer from accumulating proteins [102,[105][106][107][108]. Unresolved ER stress results in inflammation via proinflammatory cytokine release, impaired antimicrobial defense, or induction of cell death [46,100,101,104]. Heightened chaperone expression is found in the mucosa of IBD patients, indicative of a stressed ER [100,106,109]. Correspondingly, disorganized secretory granules and elevated chaperone levels in Paneth cells in the context of IBD or crippled autophagy are in line with the identified risk genes and suggest a causative involvement of ER stress and the UPR in IBD pathophysiology [47,99,[110][111][112]. Various polymorphisms in genes involved in ER stress responses (i.e. the UPR) have been identified to mediate risk for IBD development [100,[113][114][115]. Both the UPR and autophagy are engaged upon such accumulation to resolve ER stress via a broad variety of mechanisms, altogether aiming to relieve the ER from accumulated proteins and further protein biosynthesis [46,116,117]. Most prominently, rare variants of X-box binding protein 1 (XBP1) have been associated with both UC and CD [100].
Although not formally shown in IBD, a growing body of evidence suggests a tight interplay of STING signaling and ER stress that might contribute to disease pathophysiology in IBD. As shown in murine T cells, STING gain-of-function mutations lead to chronic activation of ER stress and the UPR, which ultimately resulted in apoptotic cell death (Fig. 2) [118]. Mechanistically, excessive STING signaling disrupted calcium homeostasis, rendering T-cells hyperresponsive to ER-stress induction [118]. Vice versa it was also shown that ER stress can be a strong elicitor of STING signaling in itself, as ethanol-induced ER stress in hepatocytes triggered the association of IRF3 with STING and led to subsequent IRF3 phosphorylation (Fig. 2) [119]. The STING-ER-stress nexus also plays a vital role in pathogen response, as vita-pathogen-associated molecular pattern-induced STING signaling elicited ER stress and thereby promoted immunity and survival after infection in macrophages (Fig. 2) [97]. Importantly, these studies underscore the interplay of ER stress and autophagy induction as critical for STING-induced pathogen clearance. It is therefore interesting to speculate to which extent chronic ER stress, as commonly seen in the intestinal epithelium of IBD patients, could be a direct consequence of constitutive cGAS/STING activation. In this regard, excessive cGAS/STING activation could simply reflect a futile mechanism to cope with cellular stress, for example, elicited by invading bacteria in an inflamed epithelium.
Autophagy represents another cellular pathomechanism that potentially links IBD pathophysiology with STING signaling (Fig. 2). Among a variety of conditions of cellular stress, including ER stress, autophagy is engaged upon intracellular infection (a process termed xenophagy, from the Greek word "xénos" for "strange, foreign") and helps to trap and kill intracellular pathogens. This process is explicitly important for antibacterial defense in the gut mucosa [120][121][122][123], suggesting a possible molecular link between autophagy and IBD. In IBD, various genetic risk variants have been identified in the ATG16L1 gene, of which the rs2241880 SNP is the most common diseaseassociated polymorphism and encodes for a missense mutation in ATG16L1, the T300A (Threonine to alanine at position 300) variant of ATG16L1 (ATG16L1 T300A ) [124]. Unlike the normal ATG16L1 protein, the T300A variant is subjected to increased caspase-mediated cleavage leading to lower protein levels available for ATG12-5-16L1 complex formation and subsequent autophagy/xenophagy induction [125][126][127][128]. As a consequence, defective autophagy in patients carrying the T300A polymorphism of ATG16L1 fails to restrict proinflammatory NOD2 (nucleotide-binding oligomerization domain-containing protein 2) signaling upon bacterial infection, which sheds light on how microbe sensing and autophagy synergize to maintain mucosal homeostasis and, thus, prevent IBD susceptibility [127,129].
Induction of autophagy is also a major consequence of STING signaling (Fig. 2). STING-induced autophagy is a conserved mechanism [130]. It has been shown that upon cGAS binding, STING translocates to the ERGIC, in which STING-containing ERGIC serves as a membrane source for LC3 (also known as microtubuleassociated protein 1A/1B light chain 3B) lipidation and subsequent autophagosome formation in human and murine cells [130]. This study reports a mechanism independently of TBK1 activation and IFN production, providing an effective clearing system of cytosolic DNA and viruses [130]. Interestingly, STING-induced autophagy also plays a role in pathogen response of vertebrates, for example, HSV-1-induced STING activation results in dual induction of autophagy and immune response in human and murine cells [131]. In addition, it was recently shown in mice, that a STING variant, which was unable to bind to IRF3 (STING S365A), was still able to induce autophagy and protect from HSV-1 infection. This process was independent of IRF3 binding and type I IFNs induction, but required TBK1 instead (Fig. 2) [132,133]. These mouse models could be of interest for elucidating type I IFN-independent effects of STING in gastrointestinal inflammation.
Even though these findings underscore the necessity of autophagy in executing STING-dependent pathogen containment and subsequent dampening of mucosal inflammation, impaired autophagy in itself can also amplify STING-dependent type I IFN signaling with deleterious consequences (Fig. 2). STING degradation following activation occurs through autophagy in a pathway that is mediated by p62 [31]. In the absence of p62, STING trafficking to autophagy-associated vesicles was impaired and led to an elevated type I IFN production in murine and human cells (Fig. 2) [31]. Also, our group recently showed that the barrier protective cytokine IL-22 induces STING-dependent type I IFN signaling in mice, which was strongly augmented in the context of defective autophagy due to deletion of Atg16l1 (Fig. 2) [2]. Intestinal epithelial cells with a deletion of Atg16l1 showed heightened type I IFN expression, which, in the context of ongoing IL-22 stimulation, results in a dramatic increase of proinflammatory TNF-α secretion and subsequent necroptotic cell death (Fig. 2). In vivo treatment with recombinant IL-22 in Atg16l1 IEC mice, but not Atg16l1 fl/fl mice, led to the manifestation of small intestinal inflammation with excessive cell death, which could be completely abrogated by type I IFN neutralization using anti-IFNAR antibodies [2]. Hence, these data implicate that the absence of autophagy-guided pathogen clearance culminates in overactive STING signaling. This mechanism could potentially provide a second layer of defense, in which excessive type I IFN and TNF-α secretion and subsequent cell death provide a rescue pathway to fight pathogen invasion. This indeed might particularly play a role in the context of viral infections in which autophagy is inhibited to facilitate viral replication [134]. However, under conditions of chronic intestinal inflammation and genetically driven impaired autophagy, excessive STING signaling could potentially amplify tissue damage and inflammation (Fig. 2).

Role of STING in cell death and survival
Beyond cytokine induction, STING signaling also plays a role in cell death pathways such as apoptosis, pyroptosis, and necroptosis (Fig. 2). Bearing in mind that STING is essentially involved in antiviral immunity, induction of cell death might resemble an ultimatum to prevent viral replication when the viral burden is overwhelming. In human monocytes, STING has been found to initiate proinflammatory pyroptotic cell death via indirect activation of the NLR family pyrin domain containing 3 (NLRP3) inflammasome (Fig. 2) [135]. Unlike monocytes, in adaptive immune cells, that is, T-and B cells, STING activation can result in apoptosis (Fig. 2) [136][137][138][139]. A more recent study investigating STING function in murine T cells revealed that STING disrupts calcium homeostasis, thereby predisposing T cells to receptor-activation-induced ER stress and in turn leading to apoptotic cell death (Fig. 2) [118]. Besides apoptosis and pyroptosis, STING and type I IFNs have recently been linked to necroptosis, another highly proinflammatory form of cell death. In murine macrophages, DNA sensing by cGAS/STING leads to Receptor-interacting serine/threonine-protein kinase 3 (RIPK3)dependent necroptosis (Fig. 2), requiring both type I IFN and TNF-α signaling [140]. Additionally, STING-dependent type I IFNs have also been found to be critical for maintaining mixed lineage kinase domain-like (MLKL) protein expression in bone marrow derived macrophages (BMDMs), pointing toward a potential role of STING in directing MLKL-dependent necroptosis [141]. Necroptosis is intimately connected with IBD pathophysiology, for example, Atg16l1-dependent autophagy protects from necroptotic intestinal epithelial cell death (Fig. 2) [2, 142,143]. Although poorly understood, an interesting connection between cell death and STING signaling has been established in cellular ferroptosis [144,145]. Ferroptosis is a form of regulated cell death characterized by lethal iron-dependent accumulation of lipid hydroxyperoxides (Fig. 2) [145]. Glutathion peroxidase 4 (GPX4), a known genetic risk factor for IBD, prevents lipid peroxidation and thereby ferroptosis [85,145]. In fact, GPX4 deletion in the intestinal epithelium fuels a dietary lipid-dependent enteritis [145]. Furthermore, it has been very recently shown in murine cells that accumulation of lipid hydroxyperoxides also inhibits STING function in a mechanism involving protein carbonylation [144]. Carbonylation is a process in which covalent lipid-protein binding leads to protein conformational changes [144]. STING carbonylation led to the abrogation of the ER to Golgi transport and thereby impeded the cGAS-induced STING-type I IFN response in murine cells [144]. Altogether, a growing body of evidence suggests a crucial interplay between STING-dependent type I IFN and cell death functions, possibly contributing to the pathophysiology of IBD (Fig. 2).

Conclusions and implications for diagnostics and therapy of IBD
In conclusion, studies on human autoinflammatory disease involving STING and studies employing mouse models of intestinal injury or barrier disruption have revealed an important role for STING in gut homeostasis (section "cGAS/STING, type I IFNs, and the microbiota in mouse models of gastrointestinal perturbations"). Similarly to the duality of type I IFN effects, perturbations in the tightly monitored balance and resolution of STING signaling have been demonstrated to either give rise to infection and inflammation susceptibility, or to protect from microbially induced inflammation. IBD is a currently incurable dis-ease with severe long-term complications. Current therapeutic avenues aim to dampen the immune response by acting on specific proinflammatory immune pathways (TNF-α, IL-12/23, JAK-STAT) [146,147]. Although the cGAS/STING pathway does not present a monocausal immune pathway contributing to the pathophysiology of IBD, we have outlined several potential molecular links that warrant further investigation of the potential causal link between cGAS/STING and IBD (sections "Interplay of endoplasmic reticulum stress, autophagy, and STING" and "Role of STING in cell death and survival").
Regarding clinical translation, we and others have previously shown that high levels of type I IFN responsive genes are closely linked to therapy failure of TNF-α antagonists [2,4]. Likewise, several JAK inhibitors, which among other potential targets also inhibit type I IFN signaling, are in phase III clinical trials of IBD (NCT02914522) or are already approved and show sustained clinical efficacy in a limited, but not yet molecularly defined subset of IBD patients [148]. Future research should thus aim to disentangle the individual heterogeneity of STING-dependent signaling (type I IFN, autophagy, NFκB) at cell-type-specific resolution. Such an individualized understanding of STING signals in the intestinal mucosa could lead to complementary diagnostic tests for existing therapies, as well as paving the way for STING modulators as a therapeutic principle in IBD. 15 Bartok, E. and Hartmann, G., Immune sensing mechanisms that discriminate self from altered self and foreign nucleic acids. Immunity. 2020.