Intracellular sensing of DNA in autoinflammation and autoimmunity

Abstract

DNA has emerged as a Pathogen Associated Molecular Pattern (PAMP), posing unique challenges in the discrimination between endogenous (self) and foreign DNA. This challenge is highlighted by certain autoinflammatory diseases that arise from monogenic mutations and result in periodic flares of inflammation, typically in the absence of autoantibodies or antigen-specific T lymphocytes. A number of autoinflammatory diseases arise due to mutations in genes that normally prevent the accrual of endogenous DNA or due to mutations that cause activation of intracellular DNA sensing pathway components. Experimental evidence from genetically-modified murine models further support the ability of endogenous DNA and DNA sensing to drive disease pathogenesis, prompting the question of whether endogenous DNA can also incite inflammation in human autoimmune diseases. Here we review the current understanding of intracellular DNA sensing and downstream signaling pathways as they pertain to autoinflammatory disease, including the development of monogenic disorders such as STING-associated vasculopathy with onset in infancy (SAVI) and Aicardi-Goutieres Syndrome (AGS). In addition, we discuss systemic rheumatic diseases, including certain forms of systemic lupus erythematosus, familial chilblain lupus and others with established links to intracellular DNA sensing pathways, and highlight the lessons learned from these examples as they apply to the development of therapies targeting these pathways.

Introduction

The immune system has evolved to rapidly detect and eradicate invading pathogens. Inappropriate activation of the immune system also leads to inflammatory disease. Examination of rare monogenic diseases arising from dysregulated immunity has greatly informed our mechanistic understanding of innate and adaptive disease mechanisms and has educated pharmaceutical design. Recent years have seen an explosion of studies focusing on autoinflammatory diseases, establishing this burgeoning field. In particular, a wealth of clinical and basic insights have established DNA, and the sensors that are activated by DNA, as key drivers of several autoinflammatory diseases.

Mammalian DNA is typically contained within the nucleus or mitochondria, and DNA accrual in the cytosol is perceived as a danger signal for cells. Accordingly, humans have an extensive network of enzymes and pathways dedicated to the rapid clearance of endogenous (self) DNA from multiple cellular compartments. Detection of cytosolic DNA in particular is a trigger to mount rapid and effective host defenses. Clinical and experimental evidence link activation of the innate immune intracellular DNA sensing pathways to a subset of autoinflammatory diseases. Autoinflammatory diseases are distinct from autoimmune diseases in that they present in periodic flares of inflammation, usually in the absence of autoantibodies or antigen-specific T lymphocytes^1,2, although the distinction between autoinflammatory and autoimmune diseases is sometimes difficult due to overlapping features.

Patients with autoinflammatory disease typically experience fever and a cluster of organ and disease-specific manifestations. These diseases arise from disruptions in the innate immune system and are subdivided into four main mechanistic categories^1,2: 1) inflammasomeopathies and the IL-1β-mediated autoinflammatory diseases (including Familial Mediterranean Fever (FMF), Cryopyrin-Associated Periodic Syndrome (CAPS), Hyperimmunoglobulinemia D syndrome (HIDS)/Mevalonate Kinase Deficiency (MKD), Deficiency of Interleukin-1 Receptor Antagonist (DIRA) and others); 2) the autoinflammatory Type-I Interferonopathies (including Chronic Atypical Neutrophilic Dermatosis with Lipodystrophy and Elevated Temperature (CANDLE)/Proteasome-Associated Autoinflammatory Syndrome (PRAAS), STING-Associated Vasculopathy with onset in Infancy (SAVI), Aicardi-Goutieres Syndrome (AGS) and others) that result in excessive type 1 interferon signaling; 3) autoinflammatory diseases mediated by the NF-kappaB pathway and/or aberrant TNF activity; and 4) autoinflammatory diseases mediated by other pathways including the complement pathway. In this review, we focus on the subset of autoinflammatory Type 1 Interferonopathies and other inflammatory diseases that are associated with DNA sensing and type I interferon responses. We use these disease states to discuss the principles of the pathways involved, describe how rare mutations lead to disease, and review what we have learned about these pathways that may be relevant in more common rheumatic diseases. In addition, we describe potential therapeutic targets in these pathways that might be leveraged for treatment.

Microbial nucleic acid sensing pathways

The innate immune system relies on Pattern Recognition Receptors (PRRs) expressed in different cellular compartments that detect Pathogen Associated Molecular Patterns (PAMPS). Engagement of the PRRs by PAMPs triggers protective anti-microbial defenses and subsequent induction of an adaptive immune response³. Accurate discrimination of pathogens from self is critical to avoiding unintended activation of these immune pathways. Over the past few decades, the identification of PRRs and other receptors that sense microbial nucleic acids has been critical to defining key principles of host defense. In mammalian cells, single and double stranded RNA and DNA, RNA-DNA hybrids and cyclic dinucleotides are all recognized as foreign, alerting the immune system to the presence of a wide array of microbial pathogens including viruses, bacteria, parasites and fungi^3,4.

Although certain features of nucleic acids (e.g. TLR9 recognition of hypomethylated CpG) differ between foreign and endogenous DNA, both foreign and endogenous nucleic acids can activate intracellular sensors. Accordingly, the mammalian immune system has developed several mechanisms to prevent the accumulation of endogenous nucleic acids. One such mechanism is through degradation of nucleic acids by ribonucleases (RNases) or deoxyribonucleases (DNases). For DNA, the DNases are either secreted (DNase I and DNase 1L3), lysosomal (DNase II) or cytosolic (DNase III, also known as TREX1) (Figure 1, orange panel). A second mechanism relates to localization of the nucleic acid PRRs to either the endolysosome, where the pathogen genome is exposed to low pH, or to the cytosol, taking advantage of the fact that mammalian DNA is typically restricted to the nucleus or to mitochondria.

Figure 1: Schematic representation of the intracelluar DNA sensing pathways.

Intracellular DNA is recognized by several major pathways that include the cGAS-STING (green box) and AIM2 (blue box) pathways. DNA binding to cGAS leads to production of cGAMP (GA), the ligand for STING. cGAMP binding to STING causes homodimerization and transfer of STING to the Golgi in a process that uses the COPII complex. Active STING signals through TBK1 and IRF3 to stimulate Type I Interferons, resulting in downstream activation of the IFN-stimulated genes (ISGs) that signal through IFNAR1. STING also mediates activation of NF-kappaB, resulting in the production of TNF and IL-6. Activation of STING can lead to pathogen clearance, autophagy and/or programmed cell death, depending upon the context. STING is returned to the ER to downregulate its signaling through the COPA complex, or is targeted for lysosomal degradation. To ensure that cytosolic DNA is effectively eliminated, several DNA degrading enzymes (the DNases) regulate levels of DNA. These include TREX1/DNase III and the lysosomal DNase II (orange box). Detection of DNA by AIM2 leads to activation of an ASC inflammasome and caspase I-mediated activation of IL-18 and IL-1β (blue box).

The ability of RNA to elicit immune activation has been reviewed extensively elsewhere and occurs in both the endolysosome (by Toll Like Receptors (TLRs)) and cytosol (by the RIG-I-like receptors RIG-I and MDA5)³. The endolysosomal TLR9 was the first described sensor of microbial DNA, recognizing unmethylated CG dinucleotides present in bacterial and viral DNA (Reviewed in⁵). Studies of TLR9-deficient mice highlighted the central role for TLR9 in controlling type I IFN production from plasmacytoid dendritic cells, but key studies also revealed that TLR9 activation accounted for only a portion of the DNA sensing response⁶. The ability of dsDNA delivered to the cytosol of macrophages to unleash a type I IFN response formally established the existence of a TLR9-independent intracellular DNA sensing pathway^6,7. This review focuses on DNA as a potent immune stimulator and on intracellular DNA sensing pathways.

Intracellular DNA Sensing Pathways

Detection of DNA by AIM2:

Delivery of dsDNA to the cytosol via lipofection was an approach used to identify the receptors for cytosolic DNA. This delivery resulted in induction of type I IFN gene transcription as well as activation of a second pathway involving caspase-1-dependent processing of the pro-inflammatory cytokines interleukin (IL)-1ß and IL-18. The receptor responsible for this latter pathway was first defined as Absent in Melanoma (AIM2) (Reviewed in⁸) (Figure 1, blue panel). AIM2 forms a caspase-1-activating inflammasome that controls the proteolytic maturation of IL-1ß and IL-18 and processing of the pore forming protein gasdermin D, an executioner of pyroptotic cell death. AIM2 is important for protection from DNA viruses and cytosolic bacterial pathogens but does not drive the type I IFN response to dsDNA. While AIM2-dependent responses are important, induction of type I IFN and IFN-stimulated genes (ISGs) is a dominant response elicited by cytosolic dsDNA.

cGAS/STING activation by DNA:

Cyclic GMP-AMP synthase (cGAS) was identified as a DNA binding protein controlling the type I IFN response. cGAS is a nucleotidyl transferase enzyme that is activated upon binding to dsDNA. Many viral pathogens, including herpes simplex viruses (HSV1 and 2), vaccinia virus and cytomegalovirus, all activate cGAS (Reviewed in³). In addition, bacterial pathogens including Mycobacterium tuberculosis, Legionella, Listeria, Shigella, Francisella, Chlamydia, Neisseria and group B streptococcus engage cGAS through sensing of pathogen DNA³. cGAS recognizes short dsDNA segments in a sequence independent manner. DNA binding leads to the dimerization and activation of cGAS (Reviewed in⁹) (Figure 1, green panel). Active cGAS converts GTP and ATP to the novel second messenger cyclic GMP-AMP (cGAMP)¹⁰. cGAMP binds to the endoplasmic reticulum (ER)-localized adaptor protein STimulator of INterferon Genes (STING, also known as MITA, ERIS, and MPYS). Upon cGAMP binding, STING homodimerizes and is trafficked to the Golgi through the ER-Golgi intermediate compartment using the cytoplasmic coat protein complex II (COPII). In the Golgi, STING interacts with the IKK-related kinase TANK-binding kinase 1 (TBK1)^11,12. STING is phosphorylated by TBK1, facilitating the binding and phosphorylation of IRF3 by TBK1¹³. Phospho-IRF3 dimerizes and moves to the nucleus to induce transcription of IFNß¹⁴, which in turn leads to induction of multiple interferon-stimulated genes (ISGs).

Although induction of Type I IFNs is the best studied response to STING activation, additional downstream responses include NF-kappaB activation of the pro-inflammatory cytokines TNF and IL-6, induction of autophagy, and apoptotic, necroptotic or pyroptotic cell death¹⁵ (Figure 1, green panel). STING activation has been shown to lead to antigen presentation in macrophages, and to B cell apoptosis and T cell death^16,17. How these mechanisms contribute to human disease is a topic of active investigation. Deactivation of STING signaling occurs by retrograde trafficking of STING back to the ER in a process dependent upon the cytoplasmic coat protein complex A (COPA), or via lysosomal degradation or STING ubiquitination¹⁵. Initially cGAS was proposed to be localized exclusively in the cytosol where it senses DNA that gains access to that compartment. However, recent studies have shown that cGAS localizes to the nucleus where it is tethered to nucleosomes, the chromatin subunit of DNA packaged around histones^18,19. This binding to nucleosomes is proposed to restrain cGAS activity, preventing its activation²⁰. Whether cGAS recognizes endogenous DNA in the nucleus and whether cGAS activity is limited to the cytosol is an area of ongoing research. Much remains to be learned therefore about how cGAS is activated by cytosolic DNA.

A growing number of Type I Interferonopathies have been defined, several of which are associated with DNA sensing pathways (e.g. SAVI) or with mutations in DNases or RNases (e.g AGS) (Table 1). These diseases reveal that perturbations in DNA recognition can also result in distinct end organ damage. In many cases, murine models of these diseases recapitulate the organ-specificity of the human disease (Table 1). These observations imply that the innate immune system maintains coordinated tissue- and cell-specific responses, the nature of which are the subject of ongoing inquiry. With the mounting evidence supporting the role of DNA sensing in the pathogenesis of several autoinflammatory diseases, the longstanding question again arises as to whether endogenous DNA in the wrong context also contributes to autoimmune diseases and indeed whether these pathways can be manipulated for therapy.

Table 1: Clinical features of Type I interferonopathies are determined by genetic alterations.

Murine models of Type I interferonopathies often, but not always, match clinical manifestation of human disease.

Disease	Mutation		Clinical Findings	Mouse Model	Murine Phenotype
FCL	Trex1 p.D18N, p.F17S, others	Autosomal dominant, inactivating mutation	Cutaneous form of SLE22,29	Trex1 D18N	SLE like disease, spontaneous production of autoantibodies26
SLE	Trex1 p.D18N	Defective catabolism of ssDNA and nicked dsDNA	Splenomegaly, vasculitis, renal disease25
AGS	Trex1 Loss of function		Encephalopathy, seizures, mimic of congenital viral infections21	Trex1 Knockout	Tissue-specific inflammation targeting heart, skeletal muscle, tongue, skin and stomach32,33
RCVL	Trex1 frameshift	Carboxy-terminus mutations preserving Trex1 activity	Three families identified, non-autoimmune retinal vasculopathy23	V235fs and D272fs	Autoantibody response to non-nuclear targets37
	DNase II Loss of function	Two families, 3 affected	Anti-DNA antibodies, anemia, nephritis, arthropathy39	DNase II-deficient	Lethal anemia43
				DNase II-deficient macrophages	Sufficient to induce polyarthritis in Rag2-deficient mice40
				DNase II /IFNAR Double Knockout	Polyarthritis, ANA, splenomegaly41,42
SAVI	STING Gain of function mutations	Dimerization domain mutations: V147L. V147M, N154S, V155M and G166E	SAVI: pulmonary inflammation, cutaneous vasculopathy SAVI: pulmonary inflammation, cutaneous vasculopathy47,87	STING N153S Gain of function	Hypercytokinemia, skin ulcerations, pulmonary disease, lymphopenia, lymph node collapse, colitis55,57,58
				STING V154M Gain of function	More severe than N153N, Pulmonary inflammation and fibrosis, lymphopenia, lymph node collapse, hypogammaglobulinemia55,56
		cGAMP binding site mutations: C206Y, R281Q, R284G	SAVI: pulmonary inflammation, cutaneous vasculopathy48,50
		cGAMP binding site mutation: G207E	Cutaneous vasculopathy and Thyroid dysfunction4
COPA Syndrome	COPA mutations	Loss of trafficking of STING out of Golgi, leading to hyperactivation	Clinically similar to SAVI syndrome61–63	COPAE241K point mutant	Spontaneous T cell activation, and T cell mediated lung disease which is STING-dependent61,88
AGS8/9	Bi-allelic mutations in LSM11 and RNU7-1	cGAS/STING activation	Encephalopathy, seizures, mimic of congenital viral infections28

Disease abbreviations are as follows: Familial Chilblain Lupus (FCL), Systemic Lupus Erythematosus (SLE), Aicardi-Goutieres Syndrome (AGS), retinal vasculopathy with cerebral leukodystrophy (RVCL), STING-Associated Vasculopathy with onset in Infancy (SAVI), and COatomer Protein complex subunit A (COPA) syndrome.

Lessons learned from monogenic mutations in the intracellular DNA sensing pathway

Diseases involving TREX1 inactivation:

The potential for DNA as a disease causing agent was first revealed by mutations in TREX1 (DNase III), which lead to disorders including AGS²¹, Familial Chilblain Lupus (FCL)²², Retinal Vasculopathy with Cerebral Leukodystrophy (RVCL)^23,24 and certain patients with Systemic Lupus Erythematosus (SLE)²⁵ (Figure 1, orange panel. Table 1). TREX1 is an ER-associated 3’−5’ exonuclease with high affinity both for single-stranded DNA and dsDNA²⁶. TREX1 mutations are frequently found in familial cases of AGS, which are often misdiagnosed as a congenital viral infection. Patients with AGS suffer from encephalopathy and severe developmental delay. TREX1 loss of function mutations account for a number of the AGS subsets²¹, although this complex disease can also be initiated by mutations in other genes including, but not limited to, RNase H²⁷ or the histone processing proteins LSM11 and RNU7-1, which result in inefficient nucleosome packaging, failure to tether cGAS to nucleosomes and downstream cGAS-STING-mediated activation of ISGs²⁸. The AGS-causing TREX1 mutations cluster within one of the three exonuclease domains of TREX1 or result in frame shifts that abolish TREX1 catalytic activity²⁷, strengthening the role of loss of DNase activity in driving AGS. FCL is a cutaneous form of lupus that presents with ulcerating lesions in the skin in early childhood. Patients are affected by skin changes and acral ischemia that are induced by cold exposure. A number of inactivating mutations in TREX1 (frequently affecting Asp18) have been identified in patients with FCL^22,29. Non-catalytic mutations in TREX1 have been shown to result in Retinal Vasculopathy with Cerebral Leukodystrophy (RVCL, OMIM 192315), a fatal disease of the microvasculature that presents with reduced visual acuity and commonly with neurologic, hepatic, and renal manifestations^30,31. The TREX1 mutations that give rise to RVCL are primarily frame shift mutations in the carboxy-terminus of TREX1, which preserve DNase activity but are proposed to disrupt the interaction of TREX1 with the SET complex, an ER-associated DNA repair complex containing three DNases that is critical for TREX1 translocation to the nucleus following oxidative stress²³.

Unlike the human diseases resulting from TREX1 mutations, TREX1-deficient mice succumb to tissue-specific responses in sites including heart, skeletal muscle, skin and stomach, while leaving other organs unaffected, suggesting that the downstream effectors of DNA detection are highly context-dependent^32,33. The myocarditis observed in the TREX1-deficient mouse is characterized by leukocyte infiltration in the myocardium, increased IFNß levels, and production of autoantibodies to the abundant, heart-specific targets of Myh6 and Jhp2^33,34. The stimulus for IFN production in the heart, however, remains unclear and presents little phenotypic overlap with human TREX1-associated diseases (Table 1). However, fitting with the concept that DNA drives this inflammatory response, TREX1-deficient cells demonstrate accumulation of endogenous, extranuclear ssDNA, leading to chronic DNA damage and checkpoint activation³⁵. In mice, deficiency of STING or cGAS, but not of MAVS, a key downstream mediator of RNA sensing, rescued the TREX-deficient lethality^33,36, strengthening the link between intracellular DNA and activation of the cGAS-STING pathway in disease pathogenesis.

Murine knock-in models of the human point mutants in TREX1 have led to important mechanistic insights into the pathways of disease specificity. The autosomal dominant mutation TREX1 D18N results in SLE in mice, with spontaneous autoantibody formation to nuclear material, while 3’ frameshift mutations (V235fs and D272fs) exhibit serologic responses to non-nuclear targets³⁷. Indeed, the TREX1 D18N knock-in mouse, as well as the TREX1-deficient mouse, exhibit other hallmarks of SLE, including splenomegaly, vasculitis and renal disease^26,33,38. The disease manifestations imparted by the TREX1 D18N mutations were reversed when mutant mice were crossed with mice deficient in cGAS, including reversal of the inflammation, IFN signature and aberrant T-cell activation, again supporting a pathogenic link to the cGAS-STING pathway³⁸. Thus, inactivation of TREX1, resulting in super-physiologic amounts of DNA, highlights a role for excess DNA in initiating autoinflammatory disease. However, the tissue-specific reactions leading to heterogeneous downstream disease manifestations is an area needing further investigation.

DNase II deficiency in mice promotes polyarthritis:

Like mutations in or deficiency of TREX1, depletion of the lysosomal-associated enzyme DNase II also results in tissue-specific pathology. Patients with mutations in DNase II that result in loss of endonuclease activity develop anti-DNA antibodies, liver fibrosis, glomerulonephritis and deforming arthropathy³⁹, presumably due to the entry of undegraded DNA into the cytosol and the subsequent activation of intracellular DNA sensing pathways. These patients also exhibit an increased type I interferon signature and mild anemia³⁹. This human disease is closely phenocopied in DNase II-deficient mice, including autoantibody production, liver fibrosis, extra-medullary hematopoiesis and anemia^40–42, but these mice also develop spontaneous inflammatory arthritis. The anemia in DNase II-deficient mice leads to embryonic lethality, coincident with the onset of definitive erythropoiesis, which occurs in the liver and requires ejection of nuclei from the developing red blood cells. Presumably, DNase II-deficient macrophages are unable to adequately metabolize DNA, stimulating IFNß production in the liver. Lethality from DNase II deficiency is reversed in DNase II/IFNAR1 double deficient mice⁴³, which survive to adulthood. The progressive, spontaneous polyarthritis that these mice develop is driven by the proinflammatory cytokines TNF, IL-6, IL-1 and IL-18^40–42.

To define the pathways leading to the development of polyarthritis, the DNase II/IFNAR-deficient mice were selectively bred to mice lacking functional intracellular DNA-dependent pathways. Specifically, these DNase II/IFNAR1-deficient mice were bred to mice deficient in AIM2^42,44, STING^45,46, or the endosomal TLR chaperone Unc93B1⁴². Arthritis is reduced in AIM2 triple deficient mice (DNase II/IFNAR1/AIM2-deficient), due at least in part to reduced levels of IL-18 downstream of the AIM2 inflammasome. In the STING triple deficient mice (DNase II/IFNAR1/STING-deficient) there was almost complete abrogation of joint inflammation^42,44, due to reduced TNF and likely also IL-6 levels, both of which are produced upon activation of the STING pathway. Furthermore, STING deficiency in the context of DNase II-deficiency (STING/DNase II-deficient mice) was sufficient to restore survival and prevent the development of arthritis⁴⁶, demonstrating the importance of the STING pathway in disease pathogenesis in these mice. In contrast, deficiency in Unc93B1 did not protect the DNase II/IFNAR-deficient mice from arthritis. Antinuclear antibody development, however, was found to be dependent entirely on Unc93B1⁴², but not on AIM2 or STING. Thus, AIM2 and STING are both dispensable for the production of ANAs but contribute to arthritis pathogenesis. These data in mice highlight distinct pathways leading to specific manifestations of DNA-stimulated autoimmune disease that are independent of Type I IFNs. In addition, these models highlight the role of endogenous DNA, and presumably the downstream activation of intracellular DNA sensing, in promoting certain manifestations of autoimmunity, and prompt the question of the potential role of DNA as an endogenous danger signal to elicit human autoimmune disease.

STING-associated vasculopathy with onset in infancy (SAVI):

In the first report of this autoinflammatory disease, six children with similar early-onset systemic inflammation primarily affecting cutaneous blood vessels and pulmonary vessels and parenchyma were determined to have autosomal dominant mutations in TMEM173, which encodes the STING protein⁴⁷. Since that time, more than 50 patients with mutations in STING have been described⁴⁸. These mutations occur either in the STING dimerization domain (V147L, V147M, N154S, V155M and G166E), the cGAMP binding domain (C206Y, R281Q, R284G), or in the transmembrane linker region (H72N)⁴⁹, and result in constitutive activation of STING⁴⁸. SAVI patients present with pulmonary fibrosis, cutaneous vasculopathy and vasculitis in many tissues, but the type and extent of other organ involvement is heterogeneous. Some patients exhibit autoantibody production, while others have increased susceptibility to infection and/or neurologic or hepatic involvement⁵⁰. A newly described class of SAVI patients presenting in adulthood further broadens the spectrum of the SAVI syndrome⁵¹. These data underscore the complexity of the STING pathway and the potential importance of other STING driven responses in disease pathogenesis. In accordance with these findings, treatment of SAVI patients with JAK inhibitors has been only partially successful, highlighting the importance of IFN-independent disease pathways^52,53.

Individual mutations in STING described in both patients and animal models appear to lead to differences in the degree of STING activation or might modulate downstream signaling in cell- and context-dependent fashions to result in a spectrum of clinical manifestations. For instance, activating mutations, such as G207E, that lie in the cGAMP binding domain, result predominantly in skin and thyroid manifestations⁵⁴. The crystal structure of STING, along with the confirmational changes observed during STING activation, advised the hypothesis that manifestations of disease resulted in distinct activation modes by STING⁴⁹. This relationship may explain the milder lung phenotypes seen in certain patients but offers an incomplete explanation as to the heterogeneity of disease. Although it is clear that activation of STING leads to robust autoinflammation, understanding the precise mechanism by which each particular mutation results in clinical disease manifestations has remained a challenge.

With the goal of clarifying mechanisms, mice harboring the SAVI mutations present in humans have been generated. Using CRISPR technology, three research teams created mice with the two most common SAVI mutations (V154M or N153S)^55–57. Homozygosity for either of these SAVI mutations results in embryonic lethality, but SAVI heterozygotes recapitulate many aspects of the human disease including immune abnormalities and lung inflammation. These mice also develop colitis, a phenotype not described to date in humans⁵⁸. All SAVI mutant mouse strains develop lung inflammation, expansion of monocyte and granulocyte populations in hematopoietic tissues, severe lymphopenia and lymph node collapse, as well as some degree of elevated ISGs^55–57,59. Surprisingly, despite the IFN response being a key feature of this class of diseases and the best studied response in STING signaling, crossing a SAVI mutant mouse with either a mouse deficient for the receptor for Type I interferon (IFNAR), or IRF3, a critical transducer of IFN production, reduced the production of most ISGs^55–57,59; however deficiency of IFNAR or IRF3 failed to rescue lymphocyte dysfunction or the widespread inflammation and fibrosis in the lung^55–57,59.

The clinical manifestations in SAVI overlap with those observed in COPA syndrome, a rare monogenic disease arising from mutations in the WD40 domain of the COatomer Protein complex subunit A (COPA). COPA is a subunit of the COat Protein complex I (COPI) which mediates retrograde trafficking from the Golgi to the ER (Figure 1, green panel). Like patients with SAVI, COPA patients exhibit autoinflammatory disease, including interstitial lung disease and an increased expression of ISGs⁶⁰, which prompted several independent groups to seek a mechanism linking STING and COPA pathways^61–64. Retention of STING in the Golgi is sufficient for its activation, and these recent studies demonstrate that loss of function COPA mutations lead to accumulation of STING in the Golgi and subsequent production of type I interferon and ISGs. These novel findings should lead to further understanding of the homeostatic regulation of STING.

The potential role of intracellular DNA sensing pathways in autoimmune disease

Due to the strong associations of intracellular DNA sensing with autoinflammatory diseases, researchers have begun to investigate the role of endogenous DNA in autoimmune disease more broadly. It is of note that over 60% of patients with SLE demonstrate a type I IFN “signature”⁶⁵, and type I IFNs have been implicated in disease pathogenesis⁶⁶. While TLR activation likely drives this response, elevated levels of cGAMP have been detected in peripheral blood mononuclear cells of 15% of patients with SLE⁶⁷, but not in healthy subjects or in patients with rheumatoid arthritis. Other studies have shown that sera from SLE patients strongly evoke an IFN response in macrophages, a phenomenon proposed to be due to the presence of DNA in apoptotic microvessels⁶⁸. DNase activity was also found to be reduced in sera from SLE patients^69,70 and a subset of SLE patients have been identified who are carriers of mutations in Dnase I^69,70 or Dnase IL3^71,72. The initially described DNase I-deficient mouse developed SLE-like disease, including autoantibody production and kidney damage in the autoimmune susceptible (129 × C57BL/6)F2 background, but not on a pure C57BL/6 background^73,74. In this mouse there was unintentional disruption of TNF Receptor Associated Protein 1/Heat Shock Protein 75 (TRAP1/HSP75). An independently generated DNase I-deficient mouse that targets only Dnase I and does not disrupt TRAP1/HSP75, however, spontaneously develops an SLE-like disease and recapitulates the female bias seen in the human SLE population⁷⁵, underscoring the potential contribution of DNA in SLE pathogenesis.

A key question regarding role of endogenous DNA in stimulating autoimmune disease is the source of the pathogenic, endogenous DNA. Many of the studies discussed thus far involve mutations leading to altered DNA sensing or accrual of endogenous DNA. Recent insights suggest several possible sources of endogenous DNA, including endogenous retroelements, leaked mitochondrial DNA, or uptake of extracellular DNA following tissue damage¹⁵. Evidence that cGAS localizes to the nucleus raises the possibility that nuclear DNA may be a source for cGAS activation. Reduced clearance of defective mitochondria via mitophagy was observed in red blood cells from SLE patients⁷⁶. When engulfed by myeloid cells, mitochondrial DNA from SLE-derived red blood cells stimulated type I IFN production in a cGAS-STING-dependent fashion⁷⁶. In contrast, delivery of DNA or cGAMP through microparticles offers protection against disease in the animal models of Experimental Autoimmune Encephalomyelitis (EAE) and type 1 diabetes^77,78. These studies suggest that although endogenous DNA may be a contributing factor in the pathogenesis of certain autoimmune diseases, further studies will be needed to better understand its specific role in autoimmunity and the pathways involved and to determine the exact sources of the endogenous DNA at play.

Therapeutic strategies targeting STING activation

The important role that DNA detection plays in host defenses cannot be understated. Therapeutic strategies leading to activation or inhibition of the STING pathway are in development, including those targeting cGAS, STING and TBK1. As these agents have been thoroughly reviewed elsewhere^79,80, we highlight a few intriguing STING-targeting agents. The potent immune stimulatory activity of DNA and cyclic dinucleotides is being leveraged for the treatment of viral infections. For example, two recent studies used small molecule STING agonists to elicit antiviral immunity to SARS-CoV2 infection^81,82. Increasingly, studies support the potential role of DNA as a driver of the anti-tumor immune response. Accordingly, cyclic dinucleotides as well as small molecule STING agonists are in clinical trials for the treatment of melanoma and other cancers⁸³.

The growing body of literature implicating aberrant sensing of endogenous DNA in the pathogenesis of autoinflammatory and autoimmune diseases raises the interesting possibility that limiting the activation of DNA sensing pathways could be a therapeutic approach for the treatment of these diseases. Several small molecules have been identified that block STING signaling. A nitrofuran derivative H-151 covalently binds STING and blocks STING palmitoylation, a critical signal needed for STING trafficking⁸⁴. H-151 blocks IFN responses in TREX1-deficient mice⁸⁴. Another small molecule inhibitor of STING, SN-011, blocks inflammation and autoimmune disease manifestations in TREX1-deficient mice⁸⁵. A peptide inhibitor of STING has also recently been described that acts in a STIM1-dependent manner to block STING trafficking from the ER to the Golgi. This peptide, ISD017, inhibits known STING downstream activities and has no overt toxic effects on cells. Interestingly, ISD017 blocks STING activation in mice and has efficacy in a mouse model of lupus. Lastly, ISD017 also reduced IFN and cytokine responses in peripheral blood mononuclear cells (PBMCs) from lupus patients⁸⁶. Beyond STING, studies are also underway to identify and study cGAS inhibitors in both monogenic and more complex diseases. Collectively these observations highlight the therapeutic potential for targeting the cGAS-STING pathway for the treatment of several diseases, and the need for further investigation to specifically harness these therapeutic approaches.

Conclusion

Studies in autoinflammatory diseases have improved our understanding of the complexities of the innate immune system and have demonstrated how alterations in innate immune pathways can drive human disease. In particular, these diseases highlight that detection of intracellular DNA is both a potent activator of the immune system in response to pathogens and has the power to elicit autoinflammation. Despite the ubiquity of intracellular DNA sensing in diverse tissues and cells, perturbation in these pathways can yield unique disease manifestations in terms of organ involvement, as highlighted by the heterogeneity of tissue involvement in the family of diseases caused by alterations in TREX1. Moreover, site-specific mutations yield unique pathologies, as demonstrated by the SAVI activating mutations. It follows that a network of mechanisms must exist to regulate pathogen- and tissue-specific responses. Pharmaceutical agents targeting activation or inhibition of STING are early examples of harnessing this powerful pathway for the clinical management of disease. It is likely that future drug design will leverage other members of intracellular DNA sensing pathways to generate more selective therapies (e.g. cGAS). Several mechanistic questions remain to be explored in this fascinating and growing field that will further refine these therapies. Further research will likely lead to an improved understanding of the regulatory mechanisms that refine the immune response to intracellular DNA and address whether the intracellular DNA sensing pathways are foundational in other rheumatic diseases.