Organisms have learnt how to coexist with invading pathogens by developing barriers that could counteract the invasion and spread of pathogen. The first step for an invading pathogen is to overcome chemical and physical barriers such as the skin, mucous membranes, antimicrobial enzymes, and peptides, which are acidic or basic environments that limit pathogen invasion [1]. However, in some cases, these barriers are not sufficient to counteract the invasion and pathogens can attack the organism, gaining access to the main organism’s delivery system, the blood flux, through which they reach all the body districts. At the same time, they also come into contact with the intrinsic cellular defense system, where they are detected, triggering the activation of the organisms’ countermeasures [1]. The pathogens’ detection occurs through specific proteins called pattern recognition receptors (PRRs), which bind pathogens’ structures such as lipopolysaccharide (LPS), nucleic acids, and pathogens’ proteins; more in general, the so-called pathogen-associated molecular patterns (PAMPs) [2,3]. Different PRRs are involved in the pathogen recognition depending on the PAMP detected, among them are the toll-like receptors (TLRs) and retinoic-acid-inducible gene (RIG)-I-like receptors (RLRs), as well as NOD-like receptors (NLRs) and cyclic GMP-AMP synthase (cGAS) [4,5,6,7] (Figure 1). TLRs are transmembrane receptors located on the cellular surface or on the membrane of intracellular vesicles such as endosomes and lysosomes that detect a wide variety of PAMPs such as lipoproteins, LPS, flagellin, C–phosphate–G (CpG)-DNA, viral ssRNA, and dsRNA. Differently, RLRs, NLRs, and cGAS are all located in the cytoplasm of the cells. RLRs detect long and short dsRNA and 5′triphosphate RNA, inducing pTBK1-dependent IFN-I transcription; NLRs detect peptidoglycan component (iE-DAP) and intracellular muramyl dipeptide (MDP); cGAS recognizes viral dsDNA and tumor-derived DNA, producing cyclic dinucleotides that bind the stimulator of interferon genes (STING), inducing IFN-I transcription [4,5,6,7,8,9] (Figure 1).

PAMP detection is the starting point of the host organism innate immune response that involves the activation of phosphorylating cascades and leads to the production of cytokines and chemokines. Cytokines are low-molecular-weight (15–20 kDa) proteins or glycoproteins [8]. Among the cytokines, there are specific ones principally involved in the response to viral infections, called interferons (IFNs) for their ability to interfere with viral replication [9]. IFNs can be divided into type I, type II, and the more recently identified type III, each of which triggers a different response interacting with specific receptors [9], [10]. IFN-I comprehends 13 IFN-α isoforms and only one IFN-β, as well as IFN-ε, κ, δ, τ, ω, and ζ. [10]. IFN-II has only one member γ [10], while IFN-III includes λ1, 2, 3, and 4 [11,12]. IFNs, thanks to their potency in inhibiting viral replication, have been studied and used as potential medicines to counteract viral infections such as hepatitis C virus (HCV) and hepatitis B virus (HBV) infection [13]. However, treatment with IFNs is not complication-free: cases of autoimmune disorders such as psoriasis, vitiligo, rheumatoid arthritis, and autoimmune hepatitis have been reported in patients treated with IFNα, indicating that direct exposure to INFs can be comparable to an uncontrolled cytokine expression incompatible with physiologic conditions where the innate immune response is tightly regulated [14].

In the fight for spreading, viruses have evolved strategies to evade the innate immune response at different levels. A number of virus-specific mechanisms through which viruses mask themself to prevent PRR detection have been identified [15]. Viral evolution has indeed occurred by selecting strains possessing proteins capable of interacting with cellular proteins, preventing the transcription factors’ activation or their translocation in the nucleus, impeding the transcription of IFNs or IFN stimulated genes (ISGs) [15,16,17]. Moreover, viral enzymes were shown to be able to degrade cellular proteins by direct cleavage or by regulating the cellular-proteasomal-mediated degradation [18,19,20].

The search for new targets and antiviral agents is continuously growing, especially with broad-spectrum activity. In fact, even though vaccines are the elective way to limit diseases and possibly eradicate pathogens, they are not always effective, particularly in people with compromised immune defenses, and they can also lose activity against rapidly evolving pathogens. Hence, the identification of novel antiviral agents is a priority for health systems [15], and the number of approved antiviral drugs is increasing yearly; up to today, over one hundred antivirals have been approved [21,22]. The most common strategies to develop antiviral drugs are based on the identification of molecules targeting viral proteins and blocking viral replication as direct acting agents. Although successful, this strategy must consider that most viruses are easily capable of selecting drug-resistant strains [22]. Hence, a novel approach for the identification of potential broad-spectrum antivirals is to target cellular proteins inducing an innate immune response, thus avoiding the high mutagenesis rate occurring for viral proteins [23,24,25].

Among the proteins possibly used as drug targets, a recently discovered one is STING—a downstream actor in the detection of non-self cytosolic nucleic acids related to viral infections and tumor conditions [26,27,28,29,30]. STING plays a pivotal role in counteracting viral infections, independently from whether the viral genome is DNA or RNA, mounting a strong innate immune response driven principally by IFN-I [31,32]. On the one side, when cytosolic DNA is detected, cGAS produces cyclic GMP-AMP (2′3′ cGAMP) that directly binds STING, determining TBK1 phosphorylation (pTBK1) and hence transcription of IFN-I [28,32,33]. On the other side, when the RIG-I pathway is activated in response to viral RNA detection, STING interacts with activated mitochondrial antiviral signaling protein (MAVS), determining pTBK1-mediated IFN-I production [32,34].

STING protein is the product of Tmem173 gene; it is a transmembrane protein identified in the first decade of the new century by different research groups simultaneously, starting from a previously uncharacterized molecule supposed to be an inducer of the INF-I response [32,35,36]. STING itself cannot be considered a PRR; in fact, its activation is an event downstream of DNA detection in the cytosol by cGAS that determines the production of 2′3′ cGAMP; hence, cGAS is the actual PRR in the cascade. Cytosolic DNA represents a red flag for the cells. During DNA damage reparation, it commonly happens that short ssDNA fragments are released in the cytosol; DNA repair and replication factors are responsible for pulling back these ssDNA fragments to the nucleus, where they are degraded by endonucleases [37,38]. When this mechanism is altered, the ssDNA accumulation in the cytoplasm induces cGAS activation, leading to the production of cyclic dinucleotides (CDNs) [38]. Of note, only dsDNA with a minimum length of 36 nucleotides can effectively induce cGAS, as short dsDNA fragments bind to cGAS in a manner not stable enough to induce the formation of CDNs [37,38].

The cGAS-produced CDN in turn directly binds STING, leading to its dimerization and phosphorylation (Figure 2). Once activated, STING determines TBK-1-mediated IRF3 phosphorylation; pIRF3 dimerizes, and the dimer translocates in the nucleus, establishing the antiviral state [39].

Comparative analysis between wild-type cells expressing STING and STING^-/- cells demonstrated that knocking out of STING reduced IFN-β as well as pro-inflammatory cytokines’ production, resulting in higher susceptibility to viral and bacterial infections [40,41,42]. The same pathway was also found to be activated during infections with L. Monocytogenes, an intracellular pathogen that expresses di-adenylate cyclase (DAC), which produces the CDN 3′3′ cGAMP, detected by STING, inducing IFN-β production. Interestingly, this enzyme was identified through bioinformatic analysis in bacteria and archaea including Staphylococci, Streptococci, Mycobacteria, Chlamydia, and Mycoplasma spp., possibly suggesting that all of these pathogens could induce a STING-mediated immune response [43].

The cGAS-mediated pathway is not the only possible STING activation pathway. In fact, several studies reported that, when viral RNA in the cytosol is detected by RIG-I or MDA5, depending on whether the RNA is short or long, respectively [4], STING may contribute to the RIG-I-mediated innate immune response through direct interaction with RIG-I, but not MDA5, as STING is not able to interact with MDA5 [44]. Of note, as confirmation of the importance of STING in infection with RNA viruses, RNA viruses have evolved strategies to target the STING protein in order to overcome the innate immune response [32,45,46,47].

In addition, it was also demonstrated that long-term exposure to the genotoxin etoposide induces the activation of DNA damage inside the nucleus, leading to a non-canonical STING activation with neither cGAS activation nor cGAMP production. In this case, the proteins ATM (ataxia telangiectasia mutated) and PARP1 (poli-ADP-ribose polymerase 1) are responsible for p53 phosphorylation, which leads to the interaction with IFN-inducible factor 16 (IFI16) in the nucleus, leading to the interaction with the cytosolic E3 ubiquitin ligase tumor-necrosis-factor-receptor-associated factor 6 (TRAF6). This complex, assembled straddling the nuclear membrane, interacts with STING located on the ER surface, leading to its TRAF6-mediated K-63-linked poli-ubiquitination [48,49]. The non-canonically activated STING preferably activates NF-kB rather than IRF3, inducing a different gene profile transcription with respect to the canonical STING activation [48].

STING is composed of 379 amino acids (aa) that can be divided into nine domains (Figure 3A). Starting from the ammino-terminal portion, STING presents four transmembrane domains (TMD) responsible for the localization in the organelles’ membrane, endoplasmic reticulum, and mitochondrion. The main domain, comprehending aa 139–379, is the cytosolic domain (CTD), located in the carboxyl-terminal region in which reside the ligand-binding domain (LBD) and the highly conserved dimerization motif GXXXS required for STING dimerization; the remaining domains are considered linker domains [50]. STING exerts its function through homodimerization by adopting the α+ β fold, involving five-stranded sheets in the center and four helices in the periphery [50] (Figure 3B). The dimerization motif localized in the transmembrane region identified as LBDα1 is essential for STING activity, as demonstrated through mutagenesis studies [51]. The resulting homodimer is the functional element able to bind the signaling nucleotide 2′3′ cGAMP. Its binding site is located in the interface between the two chains, where the Tyr167 in the LBDα1 is responsible for the binding with 2′3′ cGAMP (Figure 3C).

The binding requires charge–charge interaction between the phosphates in the 2′3′ cGAMP and the protein residues and determines the conformational change from an open to a closed active form [51,54,55,56,57]. Of note, bacterial CDN, 3′3′ cGAMP, is also detected by STING; however, it can only form a weak interaction with STING in the binding site, suggesting lower effectiveness in the activation of STING with respect to the interaction with the 2′3′ cGAMP, indicating that STING is not the major detector of bacterial infection, but is still involved in building the innate immune response against bacteria too [43,58,59]. STING interacts with TBK1 through its cytosolic domain, more specifically, with the LBD, in a constitutive manner. The binding with 2′3 cGAMP induces STING multimerization, leading to trans-autophosphorylation of TBK1 [60]. pTBK1 attracts interferon regulatory factor 3 (IRF3), forming the STING–TBK1–IRF3 complex. Although the complex formation dynamics are still unclear, the interaction leads to IRF3 dimerization and phosphorylation mediated by pTBK1 [50]. Then, the active IRF3 dimer translocates in the nucleus, where it binds the IFN-I promoter, leading to IFN-I production [36,61,62].

Mitochondria are cellular organelles with a significant role in cellular metabolism, as they are the center of ATP production and serve as metabolic hubs. They are factories for the biosynthesis of macromolecules as lipids, proteins, and nucleotides [63], and they also play a pivotal role in redox homeostasis responding to cell’s stressors, both cellular and environmental, as well as in programmed cell death, responding to pathogenic conditions leading to instauration of a pro-apoptotic state or triggering the activation of the innate immune response [63,64,65].

One of the most important mitochondrial proteins involved in the innate immune response is MAVS, a transmembrane protein located in the outer membrane (OM) of the mitochondria. MAVS is an effector of the RIG-I pathway, involved when RNA viruses are detected. MAVS aggregates and interacts with STING to induce TBK1 phosphorylation and then IFN-I transcription [63] (Figure 4A). MAVS is also modulated by bioproducts of the metabolism. In fact, it is associated in the OM with the glycolytic enzyme hexokinase 2 (HK2), which prevents MAVS multimerization in normal conditions when RLRs are not activated [63,66].

Viruses have developed strategies to manipulate mitochondria metabolism to inhibit IFN production. An interesting example is the small molecule lactate [63], a bioproduct of the lactate dehydrogenase complex (LDH) that generates it, starting from acetyl-CoA in absence of oxygen. This production can actually occur even in presence of oxygen because of the aerobic glycolysis, a condition called the Warburg effect, which is frequently promoted during viral infections and in cancer cells [62,67,68,69]. Lactate then binds MAVS, preventing its aggregation and the interaction with the downstream effectors of the cascade involved in IFN-I production, TBK1 and STING [63] (Figure 4B).

Another important mitochondrial protein that interacts with the STING signaling pathway is the nucleotide binding oligomerization domain (NOD)-like receptor 1 (NLRX1). It belongs to the NLR family; is ubiquitously expressed; possesses the NACHT domain, responsible for hetero-dimerization, and the leucine-rich repeats domain (LRR) in the C-terminus; and is the only NLR localized on the OM owing to the presence of a mitochondrial targeting sequence in the N-terminus [4,70]. This localization was reported to be important for the interaction with proteins involved in innate immunity, such as the mitochondrial MAVS and STING, as well as cytoplasmic proteins such as TRAF6 and IKK complex [4,71].

The interaction of NLRX1 with the effectors of the innate immune response occurring through the NACHT domain promotes their ubiquitination and, consequently, proteasomal degradation, resulting in the inhibition of NF-kB and IRF signaling following viral infection [71] (Figure 4C).

In fact, in cells infected with Sendai virus, influenza A virus, and HCV, the RIG-I/MAVS signaling was negatively affected by NLRX1 [70,71]. This inhibition is related to glucose levels; high glucose levels activate NLRX1, which binds poly(rC) binding protein 2 (PCBP2), which drives MAVS, unable to aggregate owing to the presence of a lactate covalent bond, as previously described, to ubiquitination and degradation [70,71]. A very similar mechanism was reported for STING in cells infected with human immunodeficiency virus type-1 (HIV-1), in which reverse-transcribed viral DNA induces IFN-I transcription through cGAS-STING; in human papilloma virus 16 (HPV16)-infected cells, the viral protein E7 promotes NLRX1-mediated STING degradation [72,73].