Regulating STING in health and disease

Cellular stresses or infections lead to the release of DNA molecules into the cytoplasm which may threaten the stability of the host genome [1]. The intracellular appearance of naked DNA molecules triggers a double-stranded (ds)DNA sensing mechanism which consequently induces innate immune responses including the production and release of type I interferons (IFN-I). This response is central to the resolution of DNA-induced cellular stress [2‐4] and prevents the emergence of autoimmunity [4, 5]. A recently described protein named STING (Stimulator of Interferon Genes, also known as TMEM173, ERIS, MITA and MPYS) is a critical regulator of these innate immune responses [6‐9].

STING is an endoplasmic reticulum (ER)-resident transmembrane protein and was first recognised as part of the ER translocon system [6, 10]. Suppression of components of the translocon-associated protein (TRAP) complex such as TRAP-β and Sec61 β have been found to impair DNA-induced type I interferon (IFN-I) signals downstream of STING [11]. The TRAP complex has been shown to be involved in two of the ER’s major responses: protein N-glycosylation [12] and endoplasmic reticulum-associated degradation (ERAD) [13]. Although the functional relevance of STING in the ER translocon system has not yet been fully elucidated, it has been proposed that STING can interact with TRAP component Ssr2/TRAPβ to enable its migration from the ER to perinuclear membranes, a process key to IFN-β promoter activation [6, 14].

Recent reports have demonstrated that cytoplasmic DNA released by microbes and viruses can trigger dsDNA-sensing pathways which activate STING [6, 15‐17]. STING then signals to the TANK binding kinase 1 (TBK1) / interferon regulatory factor-3 (IRF3) axis to upregulate type I interferon production [11, 18]. As an IKK (IκB kinase) –related kinase, TBK1 can also interact with IκB kinases to induce phosphorylation and thus degradation of IκB, thereby liberating NF-κB (nuclear factor kappa B) subunits allowing their nuclear translocation resulting in upregulation of type I interferon and other pro-inflammatory cytokines such as IL-6 (interleukin-6), CXCL10 (C-X-C motif chemokine 10), CCL5 (C-C motif chemokine ligand 5) and CCL2 [19]. Table 1 summarises the pathogens that have been shown to activate STING [6, 20‐37]. Of note, STING knockout mice generated by Ishikawa and Barber were highly susceptible to infection by the single stranded RNA viruses vesicular stomatitis virus (VSV) and Sendai virus [6], suggesting STING activation pathways may overlap with RNA sensing mechanisms or reverse transcription of viral RNA [6, 24, 38].

Whilst the STING-mediated dsDNA-sensing mechanism is critical for successful cellular protection against infections and disease progression, dysregulated STING activity leads to the excessive production of inflammatory mediators with potentially detrimental effects on surrounding cells and tissues. Recent studies revealed some important functions for STING in autoinflammatory diseases [39‐41], cancer [41‐44] and lipid regulations [45, 46], highlighting the importance of this protein in health and disease. Here we review the recent insights into STING function in human pathologies and discuss the potential of STING-targeted therapies which are of considerable scientific and clinical interest.

The sequence and topology of STING has been studied in parallel to its function. It is known that STING is a 379 amino acid long ER transmembrane protein encoded by the human TMEM173 gene (Accession NP_938023, XP_291127) and homologous genes in other mammalian species. The STING structure is highly conserved between mammalian species, with the N-terminal forming a putative multi membrane-spanning region, a middle CDN-recognition domain, and a cytoplasmic tail (Figure 3).

It is understood that the N-terminal 130 amino acids of STING form four transmembrane helices [92, 93] that are mainly responsible for membrane anchorage and inter-organelle trafficking [94] (Figure 3). An additional helix, named helix α1, was previously considered as the fifth transmembrane domain [6, 8, 95, 96] but has more recently been proposed to form a distinct domain with different functions [97]. Helix α1 is formed between residues 153–177 and has been reported to play an essential role in protein folding and dimerisation. Compromising the integrity of helix α1 causes STING precipitation in the cytoplasm and abolishes the homo-dimeric structure due to the loss of abundant hydrophobic interactions between dimers [97]. Ouyang’s group reported that helix α1 also supports strong hydrogen force between STING dimers and cyclic di-GMP, suggesting its importance in multiple STING functions [98].

In addition to helix α1, the rest of the cytoplasmic tail also contributes to STING dimerisation [7], cyclic dinucleotide recognition [99], and TBK1 and IRF3 recruitment [18, 100] (Figure 3). In the absence of ligands, STING dimers show an open structure susceptible to cyclic dinucleotides [98]. Upon recognition of cyclic di-GMP, residues 139–344 are rearranged to expose a docking site for TBK1 [98], enabling TBK1 to phosphorylate STING at serine 358 [8]. This phosphorylation in turn enhances TBK1-STING attachment. Tanaka’s group found that a 39-residue fragment at the carboxyl end of the STING protein was sufficient to activate IRF3 signalling in response to DNA challenge [18], and the loss of this tail region, encoded by exon 7, creates a dominant negative STING isoform for TBK1-IRF3 signalling [101]. Further truncation of the STING C-terminal fragment revealed that S366 and L374 are the key residues for IRF3 recruitment and activation [18]. Therefore, TBK1 and IRF3 are recruited to STING in a 20 amino acid –spanning region to facilitate their interaction by close proximity.

Crystallisation of the STING protein has revealed that key residue substitutions or segment deletions significantly alter its structure, resulting in a dysfunctional protein [6, 7, 53, 97, 98, 102]. Mutations of single or multiple amino acids of STING have been found to influence its dimerisation capability [39, 103‐105], ligand binding capacity [39, 104‐106], or the ability to be post-translational modified by regulatory proteins [102, 107]. Some of these mutations occur naturally in humans to cause lethal autoinflammatory diseases [39, 105, 108], and some were generated experimentally in order to understand the structure-function relationship of STING [53, 96, 97, 102].

As a critical coordinator of the innate immunity, STING is tightly regulated by a variety of signalling molecules. Except that STING is post-translationally modified to enable dimerisation and activation, some regulators are essential for the prevention of constitutive type I interferon signalling which have been shown to cause autoimmunity both in animal models [6, 112] and in human [39, 108]. Negative regulation of STING signalling is necessary for the resolution of inflammatory responses post infection (Figure 4).

It is evident that many viruses can evade STING signalling to establish a biological niche in mammalian cells, and that those within the same family tend to adopt similar evasive strategies [11, 62, 63, 65, 67, 68, 137‐148] (Table 3). For instance, several members of the Gammaherpesviridae family were found to express ORF52 protein homologs to disrupt cGAS activities, preventing subsequent production of 2′-3′ cGAMP [48]. Species of the Flaviviridae and Coronaviviridae families tend to secrete proteases that directly cleave or block STING [140]; typical examples include the NS2B/3 and NS4B proteases released by flaviviridae Dengue virus and hepatitis C virus, respectively, which can both degrade hSTING. In particular, NS4B protease shows strong homology to the ER-embedded N-terminal domains of STING, leading to colocalisation and direct protein-protein interactions with STING [11]. It has also been suggested that tumour associated viruses such as human papillomavirus and adenovirus could potentially release oncogenic proteins to block cGAS/STING interactions with tumour suppressors, hence compromising innate immunity and supporting cancer progression [142].

The impairment or absence of interferon responses often seen in HIV infection has been proposed as one of the mechanisms by which this virus is capable of suppressing host immunity [148‐151]. Recent research suggests that HIV-1 enhances the action of NLRX1 to dampen STING activity [67] and recruits host 3′ exonuclease TREX1 to degrade excessively produced, reverse-transcribed, viral DNA thereby avoiding detection by the cGAS/STING pathway [68, 152]. The capsid of HIV also regulates its association with host protein cyclin A, which controls the masking of viral cDNA from cGAS recognition in the cytoplasm and its exposure in the nucleus to facilitate genome integration [146]. Mutations in the HIV-1 capsid sequence enhance its binding to cyclin A prematurely in the cytoplasm enabling DC sensing of double-stranded DNA and a potent innate immune response against viral infection [146].

Type I interferons are key cytokines induced by antimicrobial and antiviral immunity. This family of cytokines consists of the predominantly produced interferon-α and interferon-β, and the less abundantly expressed subtypes such as IFN-ε, −κ, −τ, and –ζ [153]. Type I interferons are ubiquitously expressed by a variety of cells including macrophages, lymphocytes, dendritic cells, fibroblasts and haematopoietic plasmacytoid dendric cells, with a widespread role in cellular biology [153, 154].

“Basal” expression of type I interferons is regulated via an autocrine mechanism [155], whereas the activation of interferon-inducing regulators such as STING can significantly boost their expression by activating the transcription factor IRF3. Type I interferons are released extracellularly for detection by self or nearby interferon receptors, IFNαRs, which are coupled to JAK1 (Janus kinase 1) and TyK2 (tyrosine kinase 2) [86, 87]. This activation further promotes the formation of STAT1-STAT2 heterodimers [23, 89, 90] and the subsequent recruitment of IRF9 to assemble the transcription complex ISGF3 to upregulates the expression of a series of interferon-stimulated genes (ISGs) [154]. A broad range of ISGs have been found to control chemotaxis, cell migration, apoptosis, cell proliferation, and to regulate immune detection and defense against infection; many of which have been thoroughly reviewed previously [156‐160] (Table 4). Thus, dysregulation of type I interferon signalling can cause an excessive production of ISGs, in turn over-activating the immune system.

The persistent or excessive presence of cytoplasmic DNA is one of the major causes for chronic inflammation and autoimmune diseases. Chronic production of type I interferons, termed “type I interferonopathy”, is a key indication of immune dysregulation predominantly associated with DNA-induced autoimmunity. The overactive IFN-I response alerts cytotoxic immune cells systemically via ISG production. This in turn promotes sustained release of proinflammatory cytokines including IL-1α/β, IL-12, and TNF-α, causing excessive inflammation and tissue damage [161‐163]. Autoimmunity also induces aberrant cell death, which releases cellular components to T and B lymphocytes and leads to the production of self-reacting antibodies that congest in capillaries [164]. Unresolved B cell activation predisposes individuals to the development of systemic lupus erythematosus (SLE) that is clinically challenging to treat [152, 165, 166].

Systemic lupus erythematosus is a systemic chronic autoimmune disease. [152, 165, 166] (Figure 5). SLE is often diagnosed by the accumulation of serological antinuclear antibodies (ANA) against nucleic acids released from dead cells, which cause multiple tissue and organ damage [167]. Chronic activation of DNA and RNA sensing pathways triggered by infection and cell death can contribute to the type I interferonopathy that predisposes individuals to SLE [166]. Mutations in nucleic acid sensors (such as endosomal Toll-like receptors [168], RIG-I [169], and DAI [170]), interferon regulatory factors [171] and DNases [172, 173] can also increase SLE susceptibility. Recently, Ding and colleagues proposed that the cGAS-STING axis could be another pathway potentially exacerbating SLE, via the upregulation of type I interferon production downstream of cytoplasmic DNA sensing cascades [174]. However, STING deficiency in macrophages in fact renders hyper-responsiveness to endosomal TLR ligands, and STING knockout mice have shown accelerated lymphocyte accumulation and expansion of an IFN-α -responding cell population [175]. This therefore suggests inhibitory roles for STING in SLE development. Sharma et al. also showed that STING suppression can restrict the expression of regulatory T cell activation factor IDO-1 and TLR negative regulators such as A20, SOCS1 and SOCS3, contributing to uncontrolled systemic inflammation [175]. Since this effect is not seen in cells lacking IRF3, the transcription factor mediating most of DNA sensing responses downstream of the STING-TBK1 axis, it is possible that STING is immunosuppressive in inflammatory pathways independent of cytoplasmic DNA recognition. Therefore, autoimmune therapies targeting the STING pathway should be considered with caution and an awareness of the resultant STING down-regulation, which may have opposing effects in certain diseases.

AGS (Aicardi-Goutières Syndrome) is another genetically-based autoimmune disease, characterised by DNA-triggered type I interferonopathy [176] (Figure 5). Patients with AGS often carry mutations in DNA restriction factors including the 3′ exonuclease TREX1 [177, 178], dNTP restriction factor SAMHD (SAM domain and HD domain) [179, 180], RNase H2 (ribonuclease H2) [181, 182], dsRNA sensor IFIH1 (IFN-induced helicase C domain containing protein 1) [183, 184], and the dsRNA-specific adenosine deaminase ADAR1 [185]. These regulators maintain a balance between the production and degradation of nucleic acids, providing intrinsic protection against immune activation due to “self-recognition”. Mutations of DNA or RNA restricting factors cause nucleic acids to accumulate in the cytoplasm leading to SLE and AGS. Recent studies show that the type I interferonopathy associated with SLE and AGS is potentially cGAS-STING dependent, and that the aberrant IFN-I response can be suppressed by the loss of DNA sensor or STING in cells or animals expressing mutated Trex1 or Samhd [180, 186].

Gain-of-function mutations in STING have been identified in infants who suffer from severe and chronic vasculopathy and pulmonary inflammation, a condition known as STING-associated vasculopathy with onset in infancy (SAVI) [39] (Figure 5). Mutations of STING V147 L, N154S, V155 M, and V155R were found to direct it to an active conformation enhancing dimerisation and inducing TBK1-IRF3 signalling (Table 2). This results in an excessive IFN-I response in fibroblasts, keratinocytes and immune cells to attract and amass proinflammatory cells and regulators in capillaries and tissues, ultimately causing lesions in these regions. Sustained IFN-I signals activate interferon receptors and promote expression of interferon-stimulated genes via JAK1-Tyk2 signalling and STAT1-STAT2 dimers. In vitro experiments and pioneering clinical studies suggest that JAK adaptor inhibition effectively dampens STING-mediated IFN-I over-activity. For instance, the elevated IFN-I levels in biopsy samples from SAVI patients can be restored close to that of the normal controls with treatment using the JAK inhibitor, tofacitinib [39]. Further investigation is required to examine the potential adverse effects of interferon suppression, which is likely to increase host susceptibility to infection.

Gain-of-function mutations of STING (Table 2) have also been linked to the autoimmune disease familial chilblain lupus (FCL) [105, 108], a rare hereditary form of SLE commonly associated with cytoplasmic DNA accumulation in monogenic mutations of exonucleases TREX1 [187, 188] or SAMHD [189] (Figure 5). A recent discovery reports FLC in five members of a four-generation family sharing the same TMEM173 (STING) variant that encodes a single polymorphism of G166E. Structural analysis of mutated STING dimers reveals strong hydrogen attractions between E166 on one monomer and two threonine residues on the associating monomer, hence leading to enhanced adaptor dimerization and constitutive IFN-I -activated signalling [105]. Although limited treatment data is currently available for FCL, the authors showed that continuous administration of the JAK inhibitor tofacitinib can markedly ameliorate type I interferonopathy and associated symptoms in two patients. A similar therapeutic strategy was previously proposed by Liu and colleagues for treatment of STING-associated vasculopathy with onset in infancy (SAVI) [39]. Therefore, this therapeutic approach, based on mechanistic data, could be adapted to treat type I interferonopathy found in various diseases.

Studies on STING regulatory pathways provide novel insights into antiviral and anti-inflammatory therapies. Activating STING-dependent pathways has been developed therapeutically for antiviral and, more recently, anti-tumour benefit. The predominant approach taken has been to introduce STING ligand cyclic dinucleotides to promote the IFN-I response, to combat infection or to prevent tumour progression. In contrast, type I interferonopathy associated with STING over-activity represents another set of pathologies underlying autoimmunity. Counteracting these disease processes requires potent suppression of STING signalling; attenuation of the interferon receptor adaptor JAK is used as a current target, whilst inhibiting immune cell activation may also help ameliorate symptoms.

Cytoplasmic DNA has been implicated in many human pathologies, many associated with chronic inflammation. Research has revealed that the ER transmembrane protein STING is a crucial player in dsDNA pathogen -sensing pathways whose dysregulation contributes to the development of several diseases. By responding to DNA sensors and cyclic dinucleotides, STING induces IRF3- and NF-κB –dependent pathways to elicit proinflammatory responses against infection and cancer progression. STING also directly crosstalks with several other regulators to modulate critical biological processes including autophagy and cholesterol biosynthesis. It is evident that activating STING results in the type I interferon response to protect against infection and tumour formation, while dysregulated gain-of-function STING mutations lead to detrimental consequences of autoimmunity. Understanding the molecular signalling mechanisms of STING activation has provided new insights to advance therapeutic strategies in treating infection, cancer and autoimmune diseases. It has also prompted the development of intracellular delivery systems to administer STING agonists. Despite the fact that STING has only been studied for a decade, this adaptor protein will continue to attract attention in immunology research and clinical practice into the future.