Type I IFNs (IFN-Is) are powerful cytokines. They provide remarkable protection against viral infections, but their indiscriminate production causes severe self-inflicted damage that can be lethal, particularly in early development. In humans, inappropriately high IFN-I levels caused by defects in the regulatory mechanisms that control IFN-I production and response result in clinical conditions known as type I interferonopathies. In essence, type I interferonopathies define the upper limit of safe, IFN-related inflammation in vivo. Conversely, the loss of IFN-I responsiveness increases susceptibility to viral infections, but, surprisingly, most affected individuals survive despite these inborn errors of immunity. These findings suggest that too much IFN-I early in life is toxic, but that insensitivity to IFN-I is perhaps not the death sentence it was initially thought to be. Human genetic analyses have suggested that seemingly insignificant levels of IFN-regulated gene activity may be sufficient for most of the antiviral defenses used by humans in natura.
Classically, the detection of pathogen-associated molecular patterns (PAMPs) by a diverse network of cytoplasmic and membrane-bound pattern recognition receptors (PRRs) is seen as the primary trigger for the production of a heterogeneous group of soluble cytokines responsible for activating proinflammatory and antiviral effectors that “interfere” with viral replication. The appropriately named IFNs form three families of cytokines: type I IFNs (IFN-Is) (α/β), type II IFNs (IFN-IIs) (γ), and type III IFNs (IFN-IIIs) (λ). The IFN-I family consists of 13 partially homologous α subtypes, one β subtype, and six other gene products (IFN-ε, IFN-τ, IFN-κ, IFN-ω, IFN-δ, and IFN-ζ) that, together with the four IFN-III subtypes, orchestrate an effective, fast-acting, and powerfully amplified weapon in immune responses to the invasion of cells by microbes (1, 2). IFN-Is engage in both autocrine and paracrine signaling via a canonical JAK/STAT signaling pathway to induce highly effective antiproliferative and antiviral effectors. However, these responses are not without negative consequences for the host, and the IFN system is, therefore, rigorously regulated. Human genetic studies have shown how defective regulation of the IFN-I system can lead to autoinflammatory complications of various severities known collectively as type I interferonopathies (3, 4). Conversely, impairments in IFN-I induction or the downstream signaling cascade render individuals susceptible to certain viral infections, including those not typically associated with disease. This dichotomy highlights the crucial balancing act required for the IFN-driven inflammatory response to restrict infection safely without significant toxicity. Therefore, the IFN-I “Goldilocks” zone, as defined by human genetics, likely exists at impressively low levels of IFN-related activity. By factoring variable penetrance associated with these conditions into the equation, this review explores the mechanisms that compensate for defects and dysregulation of the IFN-I pathway. The knowledge gleaned may not only guide the treatment of these syndromes but may also influence the design of a new generation of antiviral and anti-inflammatory therapeutics (Fig. 1).
History and background of IFN-I
The discovery of IFN-Is as antiviral agents in the mid- to late 1950s revolutionized our understanding of how we fight viral infections (5, 6). The purification of IFN-I cytokines in the late 1970s and early 1980s (7–11) and the generation of IFN-I (α/β) receptor (IFNAR) knockout mice in the 1990s solidified the robust antiviral function of the cytokine previously described in vitro (12). The yin to this yang was the gradual awareness that IFN-I could be detrimental to the host. Experiments performed in mice in the 1970s and 1980s indicated that IFN-I was potentially neurotoxic in developing mice and might also cause renal damage (13, 14). Additional side effects were later confirmed in humans with the implementation of IFN-I as a therapeutic agent for hepatitis C and multiple sclerosis (15, 16).
Meanwhile, the molecular mechanisms and functions of IFN-Is were gradually being deciphered. By the late 1980s and early 1990s, the canonical IFN-I signaling pathway had been defined. Infection exposes host cells to PAMPs that stimulate PRRs located on the plasma membrane, in endosomal compartments, and in the cytoplasm. These PRRs in turn initiate the processes of the innate and, indirectly, the adaptive immune systems by inducing a number of cytokines, including IFNs. The RNA helicases retinoic acid–inducible gene I (RIG-I or DDX58) and melanoma differentiation–associated gene 5 (MDA5 or IFIH1) are the primary RNA sensors in the cytosol, whereas the cytosolic GAMP synthase (cGAS) is responsible for the detection of DNA in the cytoplasm, where it should not exist outside of mitosis. Various TLRs on the cell surface and in endosomes detect viral or bacterial PAMPs (LPS by TLR4, dsRNA by TLR3, ssRNA by TLR7 and TLR8, and unmethylated CpG DNA by TLR9) (1, 17). Signaling cascades from all these PRRs ultimately converge on the principal regulators of IFN expression, the IFN regulatory factor (IRF) family of transcription factors, of which IRF3 and IRF7 are cardinal (1). Several other PRR pathways converge one step upstream from the IRFs. TANK-binding kinase 1 (TBK1) and the IκB kinase-ε (IKKε) form a complex that directly phosphorylates IRF3 and IRF7. TBK1 and IKKε are activated by both cytosolic RNA sensors (RIG-I and MDA5), the cytosolic DNA-sensing cGAS, the endosomal dsRNA sensor TLR3, and the LPS-detecting TLR4. RIG-I and MDA5 are dependent on the mitochondrial antiviral signaling protein (MAVS) for the activation of IRF3 via TBK1/IKKε, whereas molecular messages from cGAS, TLR3, and TLR4 reach TBK1/IKKε and ultimately IRF3 independently of MAVS. cGAS generates a cyclic dinucleotide signaling molecule called cyclic di–GMP-AMP (cGAMP) in response to stimulation with cytosolic DNA. cGAMP is recognized by the stimulator of IFN genes (STING or TMEM173). At the endoplasmic reticulum, STING recruits TBK1 and IRF3 for the phosphorylation of IRF3 by TBK1 in a manner similar to that observed for the recruitment of these proteins to the mitochondrial membrane by MAVS (18). Ultimately, these various means of IRF phosphorylation lead to IFN-I production and immune activation via IFNAR.
Cytokine binding to IFNAR triggers dimerization of the two receptor subunits, IFNAR1 and IFNAR2. Once the receptors are in close proximity, the Janus kinases JAK1 and TYK2 transphosphorylate each other as well as sites on the receptor subunits at which STAT proteins bind to the complex. The JAKs then phosphorylate STAT1 and STAT2, which then dimerize and bind to IRF9 to form the phosphorylated transcription factor complex IFN-stimulated gene factor 3 (ISGF3). ISGF3 is translocated to the nucleus, where it binds to IFN-stimulated response element (consensus sequence 5′-TTTCNNTTTC-3′) enhancer sites associated with the promoters of more than 380 target genes known as IFN-stimulated genes (ISGs) (19). The proteins encoded by these genes antagonize viral replication and initiate inflammatory processes (1, 17, 20). However, the functions of IFN-I extend beyond pathogen resistance.
With the gradual elucidation of IFN-I antiviral function and the molecular mechanisms of rapid induction in response to acute infection, an important role for almost undetectable baseline levels of IFN-I also emerged. Despite the tiny amounts of cytokine involved, tonic IFN-I expression has a marked impact on several key developmental and immune processes, ranging from maintenance of the hematopoietic stem cell compartment to immune cell function. Constitutively produced IFN in healthy tissue is thought to act by maintaining the stoichiometry of signaling proteins that are themselves ISGs (e.g., STAT1) in response to a broad network of cytokines (21–24). Ratios of these intermediates can dictate how a cell responds to a particular cytokine or infectious insult (22). For example, low levels of STAT3 lead to an IL-6 response that more closely resembles the IFN-II response (25) and boosts the antiviral activity triggered by IFN-I, as STAT1 is preferentially phosphorylated in both contexts (26). In addition, stem cells constitutively express a cell type–specific subset of ISGs, although they do not respond to IFN stimulation. This intrinsic ISG expression protects against viral infection but is gradually lost during terminal differentiation as sensitivity to IFN cytokines is inversely acquired (27). There are several reported examples of IFN-I–independent ISG induction, including direct binding of IRFs to many ISG promoters (17, 20) and an unphosphorylated form of ISGF3 that facilitates expression of select ISGs long after the cessation of cytokine-triggered receptor activity (28–30). Therefore, the occurrence of IFN-I–independent ISG expression on such a scale further highlights the importance of baseline ISG activity in various contexts and raises new questions about the mechanisms regulating the IFN system.
Unfortunately, tonic IFN activity has seldom been discussed in the literature, a victim of its own subtlety and probably of technical limitations to its study in the past. We are, understandably, drawn to the IFN-induced spectacle that follows infection, in which massive changes in gene expression and cellular activity are abundant, and the system can be simplified in the context of innate immunity. Our reliance on serum-based determinations of IFN-I without an ability to measure IFN-I in tissue, where larger amounts of cytokine may be released more frequently, has proved a major obstacle to its study. However, improvements in our ability to detect vanishingly small amounts of cytokine have opened a new window. A single-molecule array digital ELISA was recently reported to be able to detect attomolar concentrations (10−18 mol/l) of IFN-α in healthy individuals, viral infection, and complex or monogenic interferonopathies (31). In the absence of this sensitivity, detection of IFN activity has relied on signal amplification in the form of mRNA from ISGs to serve as a proxy for cytokine detection (32, 33). Either way, the various functions of tonic expression at infinitesimal quantities of IFN highlight not only the importance but also the sensitivity of the system downstream from the cytokine and demonstrate how even slight dysregulation can have substantial repercussions.
The upper limit: type I interferonopathies
Systemic lupus erythematosus was the first disease in which inappropriately high levels of IFN-I activity were implicated in pathogenesis (4). Type I interferonopathies constitute an expanding group, with many genetic drivers related to deficits in nucleic acid processing or detection leading to hyperactive IFN-I production. The issue of IFN dysregulation was first raised by Dr. I. Gresser at the Institut de Recherches Scientifiques sur le Cancer in Villejuif, France. Across a series of experiments spanning the mid-1970s to the early 1980s, he and his collaborators demonstrated the effects of IFN on mortality, growth, and tissue damage as a function of developmental stage. His forward thinking envisioned that virus-induced disease might actually result, to some extent, from IFN-mediated inflammation rather than the cytopathic effect of the virus itself (13, 14), a principle since validated by several studies that have shown IFN-related inflammation to be a significant driver of pathology during infection (34–36).
Meanwhile, his work on mice with differing susceptibility to influenza in the context of exogenous IFN helped establish a genetic etiology behind variable susceptibility to infection between individuals and its subjugation to relative IFN sensitivity (37).
The prototypic example of IFN dysregulation is Aicardi–Goutières syndrome (AGS), a Mendelian systemic autoinflammatory condition characterized by high levels of IFN-related activity in the absence of infection (38, 39). This condition has classical and nonclassical presentations. Beneath this distinction, the classical phenotype has been subdivided further into early- and late-onset forms, although both typically manifest within the first year of life. In rare cases of early onset, complications have been noted at birth, indicating that disease progression can actually begin in utero. Patients display neurologic dysfunction in the forms of progressive microcephaly, spasticity, and psychomotor retardation, all of which can be accompanied by intracranial calcification, white matter alterations, and abnormally high leukocyte counts in cerebrospinal fluid. Roughly 35% of these individuals die in early childhood. This presentation of AGS closely resembles the sequelae observed in infants suffering from transplacental infection, thus the distinction as a pseudo-TORCH syndrome. More commonly, the onset of AGS occurs at several months of age. Symptoms are similar to the early-onset form but are potentially less severe. Curiously, the driving factor determining early or late onset is the genetic cause (38, 40). There are seven genetic etiologies behind AGS: TREX1 (41), RNASEH2A, RNASEH2B, RNASEH2C (together encode the RNase H2 enzyme complex) (42), SAMHD1 (43), ADAR1 (44), and IFIH1 (45), all of which are involved in nucleic acid metabolism or detection. A loss-of-function mutation in the RNA processors could cause a buildup of nucleic acid, whereas a gain-of-function in the PRRs has the potential to create a hypersensitivity to otherwise healthy levels of nucleic acid. Early onset is associated with mutations in TREX1, RNASEH2A, and RNASEH2C, whereas the more common and slower disease progression is tied to mutations in RNASEH2B, SAMHD1, and ADAR1 (38, 40).
The nonclassical form of AGS is characterized by a complete or partial absence of the disease phenotype, despite the presence of mutations in genes known to be associated with AGS. Even within families affected by AGS, variable penetrance results in discordant disease progression (40, 45), and this reflects a broader picture in the type I interferonopathy field. That is to say, the field is not done characterizing the factors that contribute to disease beyond the presence of associated mutations. Within AGS, the syndromic phenotype has expanded with the reporting of nonclassical, less-typical presentations in individuals with mutations in various AGS genes (46–51), whereas other genetic drivers of type I interferonopathies have been attributed to mutations in a handful of genes scattered across the IFN-I regulatory battery. De novo heterozygous gain-of-function mutations in TMEM173, which encodes the adaptor protein STING that receives cGAMP from cGAS in the cytosolic DNA-sensing pathway, have been tied to STING-associated vasculopathy with onset in infancy, an autoinflammatory vasculopathy characterized by ulcerating acral skin lesions from which many patients develop interstitial lung conditions and fever (4, 52). Other mutations linked to type I interferonopathies, including rare cases of Mendelian systemic lupus erythematosus, have been reported in the ACP5 gene (which encodes the tartrate-resistant acid phosphatase type 5 and leads to spondyloenchondrodysplasia with immune dysregulation) (53) and in DNases (DNASE1 or DNASE1L3) that are responsible for the degradation of extracellular DNA (4, 54–56).
The discovery of individuals with biallelic loss-of-expression mutations in ISG15 and the elucidation of the mechanism underlying the associated type I interferonopathy–like symptoms were significant breakthroughs in the field. Although originally documented in 2012 in a study focusing on these individuals because of their hypersusceptibility to mycobacterial infection due to a related defect of IFN-II production (57, 58), it was not until 2015 that the cause of autoinflammation in these patients was determined (59). Up to this point, all previously identified causes of IFN overactivation stemmed from defects in the machinery upstream from cytokine production, and yet here was a scenario in which the loss of an ISG, a downstream component of the system, was somehow driving high levels of IFN-I activity. It turned out that one of the functions of ISG15 is to stabilize the potent negative regulator of IFN-signaling ubiquitin-specific peptidase 18 (USP18) (59–61). Thus, the loss of ISG15 leads to increased USP18 turnover that effectively lowers the concentration of the negative regulator in the cell. Lower concentrations of USP18 result in a less-efficient negative feedback loop in IFN signaling. Unsurprisingly, an impaired ability to shut down IFN-I signaling produces a similar, but less severe, phenotype to that of classical AGS: basal ganglia calcifications and an elevated IFN-I signature in the blood (59). Of the ISG15-null patients identified to date, the vast majority are living relatively normal lives well into their early twenties. Thus, loss of ISG15 is compatible with life.
Unfortunately, complete deficiencies of USP18 are lethal. Unlike in an ISG15 deficiency, USP18-null individuals cannot negatively regulate IFN-I signaling downstream from the receptor, and disease progresses rapidly as a result. All five reported patients with biallelic loss-of-expression mutations in the USP18 gene developed a pseudo-TORCH–like phenotype similar to that of severe AGS, with innate immune inflammation, brain calcifications, and polymicrogyria, before dying within 22 d of birth (62).
The link between these two disorders is the lack of USP18. Although the relative amounts of USP18 determine the severity of clinical outcome, the lower levels of USP18 in both conditions relative to those in healthy individuals is what allows for IFN-mediated autoinflammation sustained by persistent ISG expression (59, 63). Nevertheless, it remains unclear how the loss of a negative regulator, either effectively (ISG15 null) or actually (USP18 null), can yield elevated levels of IFN-I cytokines in the blood. The working answer has several layers of complexity because of the somewhat circular nature of signaling cascades. A prototypic biological signaling cascade induces the expression of its own regulatory machinery to amplify and/or attenuate the magnitude of its effect. In this sense, IFN signaling is no different. Several of the ISGs persistently expressed in ISG15- and USP18-deficient individuals encode PRRs and signaling intermediates that activate and augment both the amounts of IFN-I produced and the sensitivity of the response. These mechanisms together establish a state of hyperresponsiveness/hypersensitivity to both PAMPs and IFN-I within the cell, but they still do not explain how the activation switch is tripped. In AGS, defects in nucleic acid processing lead to an abundance of RNA in the cytoplasm, which may or may not need to be damaged to be detected (the need for such damage is a question beyond the scope of this review). Because a buildup of damaged nucleic acids is unlikely to be the stimulus in the case of ISG15- and USP18-related inflammation, as there is no known problem with nucleic acid processing in these individuals, a hypersensitive state could facilitate IFN activation in response to what would normally be inconsequential levels of PAMPs in a healthy human. It is reasonable to assume that we are exposed daily to infectious and immunostimulatory agents at concentrations that are innocuous in healthy individuals. However, in individuals with little or no USP18 who are hypersensitive to PAMP stimulation, these events become fuel for a self-perpetuating loop of IFN-I activation, although this has yet to be proved experimentally. A similar argument could also be made for homeostatic levels of damaged nucleic acids as a driver of persistent inflammation by replacing PAMPs with undesirable nucleic acid products in the same context.
ISG15 and USP18 losses of function also raise questions about the relative ease of ISG induction. In this model, chromatin landscape along with the quantity and quality of ISGF3 binding to the promoter of a particular ISG may influence the relative readiness of that gene to be induced by IFN-I. This theory may explain why only a select few of the ∼400 canonical ISGs are upregulated in patient-derived, ISG15-null dermal fibroblasts 36 h after stimulation for 12 h with exogenous IFN-α2b relative to WT cells treated in the same way (59). Actually, ISG15 deficiency is far from the only example of this ISG expression pattern. Relative sensitivity to IFN-I stimulation probably applies to a wide range of phenomena related to ISG regulation. This may likely span from tonic IFN activity, to the above described intrinsic, and cell type specific, ISG expression in stem cells despite their refractory response to IFN-I stimulation (27), to the subset of canonical ISGs that are induced by an unphosphorylated form of ISGF3 (28–30).
In vitro study of ISG15 deficiencies has begun to define the lower limit of IFN-I–mediated inflammation effective against viral infection. The few persisting ISGs described above are expressed at only 1% of peak levels, but are orders of magnitude better at controlling viral infection than those in similarly treated WT controls, in which ISG levels have reverted essentially to baseline (63). This late time point after exposure to IFN recapitulates the breadth and magnitude of the inflammation that occurs naturally in these individuals in vivo, and the enhanced ability to restrict viral replication observed in vitro points strongly to a gain-of-function phenotype. In other words, the persistent, low-level IFN I–driven inflammation exhibited by ISG15-null individuals strengthens their ability to combat acute viral infection and, if harnessed, could have a revelatory impact on the design of novel antiviral therapies.
The lower limit: IFN-I insensitivity
Quirks in the system, such as tonic IFN expression, IFN-independent ISG expression, and built-in redundancies across the three IFN families, which we will discuss briefly below, allow the IFN-I system to function sufficiently at activity levels much lower than those observed in a healthy person during acute infection. They are also the mechanisms most likely to compensate for defects in the IFN-I response machinery.
Considering all that is known about the potency of IFN-I, the efficacy with which it restricts acute viral infection, and evasion and antagonism by viruses, it seems likely that an inability to respond to IFN-I signals would dangerously compromise the immune system. Yet, humans without expression of STAT1, STAT2, and IFNAR2 exist, and the consequences of these deficiencies are not always as catastrophic as expected. That is not to say that these individuals are healthy but rather that human genetics can help us to understand how the IFN system balances itself across all three families by exploiting the above-mentioned quirks.
Loss of IFNAR is effectively equivalent to a complete loss of IFN-I signaling. In one example, two siblings with a homozygous truncation mutation in IFNAR2 displayed no overt susceptibility to common, naturally occurring respiratory viral infections during the first year of life. Sadly, the older sibling developed fatal encephalitis following measles, mumps, rubella (MMR) vaccination at the age of 13 mo, but his younger brother was not vaccinated and remained unaffected when these cases were published (64).
The simplest explanation here is that IFN-III picks up the slack left by a broken IFNAR to the extent that it protected both siblings against natural infections but not against MMR in the case of the vaccinated proband. Although primarily studied in mice, IFN-III, also known as IFN-λ, signals through the same pathway as IFN-I but uses a different receptor and induces many of the same genes as IFN-I. Like its type I counterparts, IFN-λ also has immunomodulatory effects on the adaptive immune system as a promoter of macrophage and dendritic cell differentiation (2). In particular, epithelial cells are the principal cells that respond to IFN-λ, which makes evolutionary sense as most infections begin at the epithelium. New evidence in mice has shown that IFN-λ is actually secreted before the type I cytokines during influenza infection and that the induced cassette of IFN-λ–stimulated genes restricts infection without initiating inflammation (65). This notion was supported by a recent study on an individual who harbors digenic homozygous deficiencies in IFNAR1 (loss of a stop codon led to an extended protein with a hypomorphic response to the cytokine) and the IFN-γ receptor (limited expression and no signaling) and, as a result, experiences disseminated mycobacteria, Streptococcus viridans bacteremia, and exceptionally high CMV viremia. Notably, and unlike the loss-of-function IFNAR2 sibling described above (64), this patient does not suffer from complications following MMR vaccination (66), probably because the patient’s IFN-I responsiveness, although hypomorphic, is sufficient to control infection and/or demonstrates the compensatory capability of IFN-λ.
In the absence of STAT2, neither IFN-I nor IFN-λ can function through its canonical pathways, but, similar to IFNAR2 deficiencies, viral susceptibility remains highly inconsistent. Even within kindreds, reports consistently describe situations in which one or several of the affected members develop life-threatening to lethal infections whereas the others do not. A homozygous splicing mutation in intron 4 that leads to a complete STAT2 deficiency was reported in five members of a single family. One died of an uncontrolled viral infection as an infant, and although all three surviving children were hospitalized for severe infections, they are all now, like the affected adult in this kindred, generally healthy with normal development (67). In another instance, two siblings both survived febrile illness triggered by MMR vaccination, but only one developed severe neurologic impairments (68). Thus, the absence of STAT2 can have catastrophic effects in some individuals, but the apparently normal functioning observed in others points to redundancies in innate immunity, potentially due to overlaps in transcriptional targets across IFN families. It is well established that some ISGs have IFN-γ–activating sites (GAS) in their promoters and are, therefore, promiscuously turned on by either signal (17, 69). These GAS elements are engaged by the IFN-γ–activating factor (GAF), which consists of a STAT1 homodimer, to activate transcription in response to IFN-II and, to a lesser extent, IFN-I (70–72). Although there is substantially less target overlap between IFN-I and IFN-II than IFN-I and IFN-III (λ) (65, 73, 74), the observation that not all STAT2-deficient individuals suffer from severely compromised viral immunity suggests that redundancy within the system can provide sufficient protection against acute infection. This conclusion is supported by work from our group that showed improved viral resistance by only a small fraction of canonical ISGs at low expression levels (1% of the putative peak) (59, 63). Meanwhile, the tendency toward early penetrance indicates that the timing of viral exposure is an important determinant of disease outcome.
Biallelic STAT1 deficiencies provide a stark contrast by showing that it is not possible to compensate for a complete absence of IFN-mediated activity. Put another way, STAT1 deficiencies are the exception that proves the rule: IFN-I is not necessary if IFN-II and IFN-λ can provide sufficient antiviral activity. Affected individuals have a near-zero capacity to respond to any of the three IFN cytokine families through the classical signaling pathways, and these individuals effectively define the lower limit of survivable IFN-mediated function. As a result, the phenotype of a STAT1 deficiency is the most severe of the three known genetic defects in humans that impair the IFN response. Homozygous STAT1 loss-of-function mutations have been shown to cause Mendelian susceptibility to mycobacterial disease and inadequate control of viral infection in two unrelated infants (75).
Unfortunate as these cases are, they unequivocally demonstrate that a complete loss of all three IFN signals leaves innate immunity critically compromised and is not compatible with life. However, autosomal-dominant STAT1 defects (V266I, K201N, L706S, Q463H, E320Q) support the notion that very small levels of IFN activity are required for a relatively healthy life. These individuals also present with Mendelian susceptibility to mycobacterial disease from diminished IFN-II activity, but they do not typically succumb to viral infection because of a relatively intact IFN-I response (76–79). This suggests that the amount of STAT1 required for effective IFN-I–mediated ISG induction is less than what is required for ISG induction by IFN-II with these genetic variations. Further, mRNA from a small number of ISGs and pSTAT3–GAS element interactions were detected in Epstein–Barr virus–transformed B cells derived from both of the STAT1-null infants discussed above (75). This ISG activity was not enough to keep them alive, but these results again point to a potential role that noncanonical ISG regulation might have in less-defective IFN systems.
These experiments of nature demonstrate the upper and lower bounds of survivable and beneficial IFN-related function. Through the redundancies, signaling cross-talk, and remarkable efficiency at extremely low levels of activity, we are beginning to refine our perception of the IFN system in both the steady and infected states. A more thorough understanding of the contextual difference between the complete and sufficient IFN systems should help to improve therapeutic approaches. As an antiviral agent, the above compendium suggests that IFN-driven inflammation is easily saturated and that adequate restriction of viral infection can be obtained even with significantly compromised signaling machinery. Human genetics studies have also highlighted the severe consequences associated with too much or too little IFN. Therefore, regardless of whether a treatment employs IFN-like activity as the active antiviral agent, its design ought to consider restrained modalities that specifically drive low-grade, sustained, and targeted ISG expression. Crafted by millions of years of evolution and informed by human genetics, the proverbial Goldilocks zone constitutes a dynamic and challengingly small target at which IFN-related therapy should aim.
We thank colleagues Jennie Altman, Sofija Buta, Conor Gruber, Louise Malle, Marta Martin-Fernandez, and Xueer Qiu for input.
This work was supported by National Institute of Allergy and Infectious Diseases/National Institutes of Health Grants R01AI127372, R21AI134366, and R21AI129827, and the March of Dimes.
Abbreviations used in this article:
cytosolic GAMP synthase
type I IFN
type II IFN
type III IFN
IFN-I (α/β) receptor
IFN regulatory factor
IFN-stimulated gene factor 3
mitochondrial antiviral signaling protein
melanoma differentiation–associated gene 5
measles, mumps, rubella
pathogen-associated molecular pattern
pattern recognition receptor
retinoic acid–inducible gene I
TANK-binding kinase 1
ubiquitin-specific peptidase 18.
D.B. is the founder of Lab11 Therapeutics, a company developing antiviral therapies. J.T. has no financial conflicts of interest.