Type I Interferons
Investigating the general phenomenon of “viral interference,” where the prior infection by one nonlethal virus might interfere with a subsequent infection of cells by a second virus, Isaacs, Lindenmann, and colleagues published a series of articles in 1957 showing that an inactivated influenza virus, reacting with cells of chick allantoic membrane, can elicit the secretion of a proteinaceous substance, “the interferon,” into the medium that can, when incubated with fresh cells, confer protection against viral infection on these cells (Isaacs and Lindenmann 1957). The next 20 years saw explorations of what types of cells could secrete “interferon,” what substances – viruses or chemical – could induce interferons and initial explorations of biological and therapeutic function. Heterogeneity of “interferon” was initially indicated by the recognition that different cells might produce different types of interferons, distinguished by immune sera and by physicochemical properties, such as pH sensitivity. By 1980, intensive work in several laboratories led to the purification of the proteins and then the cDNA cloning of the genes for IFN-β and the complex family of closely related IFN-αs and IFN-ω, the most widely expressed Type I IFNs. The late 1980s also saw the description of the canonical Jak-Stat pathway of Type I IFN signaling (reviewed in Stark and Darnell 2012) and a large expansion of the knowledge of the functional roles of Type I IFNs, including the identification, characterization, and mechanistic studies of various expressed antiviral proteins. Receptor studies lagged until 1990 when the low-affinity IFNAR1 subunit of the receptor was cloned, followed in 1994 by the identification and cloning of the high-affinity IFNAR2 subunit (reviewed in de Weerd et al. 2007). However, it was not until the early 2000s when the genes for IFN-κ, then IFN-ε, the last of the human Type I IFNs, were discovered and their proteins initially characterized. Throughout this period, our knowledge of the intricacies of the molecular and immune functions of the Type I IFN system increased rapidly. Beyond the scope of this short review are the diverse medical and therapeutic aspects of the Type I IFNs, including their applications as antiviral and antitumor agents. [In Memoriam: Readers are referred to a brief personal description of the purification and cloning of the IFNs and their receptors and their therapeutic applications by Sidney Pestka, MD, a leader in molecular interferon research from the late 1970s until his recent death in 2016, and this author’s mentor (Pestka 2007)].
The Type I Interferon Family
The seriousness of the threat to vertebrates from viruses is evidenced by the development of highly sophisticated antiviral defense mechanisms, emerging at least during the evolution of fish, before the divergence of tetrapods. The human Type I interferon family (IFN-I) is a closely related group of 17 expressed proteins, encoded by intron-less genes clustered on Chromosome 9 (mouse Chr4). These ligands all signal through a heterodimeric receptor, IFNAR, composed of IFNAR1 and IFNAR2, encoded in a cytokine receptor gene cluster on human Chromosome 21q22.1 (Murine Chr16). This entry will focus on the human IFN-I, with occasional reference to mouse or other species (see also Capobianchi et al. 2015). Type II IFN in humans and most species consists solely of a single IFN-γ protein, which signals through a dedicated heterodimeric cell-surface receptor (IFNGR). Although it has modest direct antiviral activity, IFN-γ is primarily involved in immunomodulatory activities. The Type III IFNs (IFN-λs; IL28/29), only discovered in 2003, are now recognized as an important antiviral complement to the IFN-I (Kotenko and Durbin 2017). Type III IFNs share many activities with the IFN-I family and seem to be transduced through the same signaling pathways. However, compared to the IFN-Is, the IFN-λs have different gene structures and very low amino acid sequence similarity, and they signal through a unique heterodimeric receptor composed of a unique subunit, IFN-λR1 (=IL-28RA) and IL-10R2, which is also used in the receptor of several IL-10-related cytokines. All three IFN types fall within the family of Class II cytokines, which also includes interleukin-10 (IL-10) and its related cytokines (in humans: IL-20, IL-22, IL-24, IL-26); the corresponding cell-surface receptors of the Class II cytokines also show evolutionary relatedness and are classified as the Class II cytokine receptor family (CRF2) (Langer et al. 2004).
The 17 human Type I IFNs include 13 highly related (80–98% sequence identity) IFN-αs and 1 each of IFN-β, IFN-ω, IFN-κ, and IFN-ε, with the non-IFN-αs having lower (25–30%) amino acid sequence similarity to the IFN-αs and each other. IFN-Is have 165–180 amino acids in the mature sequence. Most IFN-Is are not glycosylated, though several have N-linked or O-linked glycosylation. However, the proteins fold into a strongly conserved, three-dimensional α-helical bundle, and all bind and signal through the heterodimeric Type I IFN receptor (IFNAR), consisting of the high-affinity binding subunit IFNAR-2 and the low-affinity subunit IFNAR-1. This receptor is found on all or almost all nucleated cells in mammals, making them, in principle, responsive to IFN action. Within the family of 13 human IFN-αs, and also IFN-ω, there is variation in biological specific activity, and the various IFN-αs do not always have the same rank order of potency in different assays. IFN-β also has high antiviral activity, but has stronger antiproliferative, antitumor, and immunomodulatory effects than the IFN-αs, and may be involved in basal regulation of the IFN response (see below, also, Ng et al. 2016). The more highly diverged IFN-Is have specialized functions while maintaining their antiviral and related activities. Thus, human IFN-κ is expressed in keratinocytes, during skin infections and in some pathologic skin conditions. It seems important for antiviral defense in the skin, but IFN-κ expression can also be repressed during successful HPV infections. Even more divergent is IFN-ε, whose expression is primarily in the female reproductive system, where its transcription is under hormonal regulation, rather than viral regulation. (As another evolutionary oddity, ruminants, such as cows, sheep, etc., have a specialized IFN-I, IFN-τ, produced at very high quantities by the blastocyst under reproductive hormonal control during early pregnancy, when it is critical for implantation of the conceptus, although it retains the usual antiviral and immunomodulatory activities.).
With potency in the picomolar range, IFN-I expression is under tight genetic control, primarily at the transcriptional level (Honda et al. 2006; Levy et al. 2011). The IFN-αs are generally co-induced, often with IFN-ω, from leukocytes as well as from most other nucleated cells. Importantly, the IFN-αs, IFN-β and IFN-ω are co-induced at high levels in stimulated plasmacytoid dendritic cells (pDC), which are considered “professional IFN-producing cells.” While not the focus of this entry, the signaling pathways from pathogen to induction of interferon are critical and intricate (Schneider et al. 2014). There are several pathways leading to IFN-I induction, depending on the cell type and the nature of the stimulus. Thus, viral nucleic acids or other viral or microbial components may be recognized by various pattern recognition receptors (PRRs), located on the cell membrane, on the endosomal membrane, or in the cytoplasm, triggering IFN induction. Alternatively, IFNs themselves or other signals may initiate IFN production through activation of IRF3 and IRF7, two cytoplasmic members of the interferon response factor (IRF) family (Honda et al. 2006; Levy et al. 2011; Schneider et al. 2014). Dysregulation of this pathway, particularly when leading to chronic IFN expression, has been implicated in the pathogenesis of several autoimmune and autoinflammatory conditions (Crow et al. 2015).
The Type I IFNs are highly pleiotropic in their cellular and physiological functions. Following induction and secretion, the IFNs can feed back on producing (infected or stimulated) cells (autocrine action) through the cell-surface IFNAR receptor or may act on neighboring cells or enter blood or other fluids to act on distal cells. Binding to the receptor leads to the transcriptional activation or enhancement of a large number (>300) of genes (interferon-stimulated genes (ISGs)), and repression of others, to create an “antiviral state,” i.e., an environment that, in multiple ways, slows or block viruses at one or more stages of their life cycle. The proteins of some ISGs are well studied and their functions understood; others are not (Schneider et al. 2014; Mostafavi et al. 2016). In addition to direct antiviral effects, IFNs have antiproliferative effects on many cell types, and they can induce apoptosis in a number of cells types and, under some conditions, when acting on virus-infected primary cells. Antitumor activity of IFN-Is seems to be a combination of antiproliferative or proapoptotic effects and immune activation. Type I IFNs act on all or most cells of the immune system, eliciting cell type-specific responses that help coordinate innate and adaptive immune responses to viruses. Among these responses are activation of natural killer cells, differentiation/maturation and activation of antigen-presenting dendritic cells, various effects on T-cell subsets, and promoting class switching from IgM to IgG for B cells. Thus, in addition to their direct cellular antiviral action, Type I IFNs activate and coordinate innate and adaptive responses to viruses and other microbes.
Signaling in Type I Interferon Action
As noted, the diverse effects of IFN-Is are accomplished by the induction and repression of a large number of interferon-stimulated genes (ISGs), mainly encoding proteins, but also regulatory RNAs, including microRNAs (Forster et al. 2015). This is perhaps to be expected, considering the large and diverse array of viruses and viral lifestyles. Tabulated over various cell types, IFN-Is can upregulate >1000 genes. A recent transcriptomic experiment where murine IFN-α was injected into C57BL/6 mice found 975 genes induced in at least one of 11 cell types of immune origin, with a core group of 166 genes induced in all cell types examined (Mostafavi et al. 2016). While functions and mechanisms of action in viral control have been determined for some of these genes, the role of many interferon-induced proteins is not understood. Among the better understood antiviral proteins are the Mx proteins, RNase L, PKR, and 2′-5′ oligoadenylate synthase (OAS), as well as the upregulation of Class I MHC, which is crucial for displaying viral peptides on infected cells, marking them for attack by cytotoxic T cells. The ISGs also include IFNs, proteins involved in IFN induction (e.g., IRF3 and IRF7), acting thereby as a feed-forward circuit, and proteins that negatively control IFN responses, such as SOCS1, SOCS3, and USP18, which plays a major role in long-term desensitization of IFN responsiveness (Schneider et al. 2014).
The major or canonical pathway for these effects is the Jak-Stat pathway. The main elements were worked out by the early 1990s primarily by the laboratories of James Darnell, Ian Kerr, George Stark and their alumni, using a combination of biochemical, genetic, and cloning methods (reviewed in Stark and Darnell 2012). These studies led to the discovery of the Jak (“Janus kinase”) family of tyrosine kinases(in humans, Jak1, Jak2, Jak3; Tyk2) and the Stat (Signal Transducers and Activators of Transcription) family of transcription factors (in humans: Stat1, Stat2, Stat3, Stat4, Stat5A, Stat5B, Stat6). The studies of IFN signaling provided a prototype for understanding Jak-Stat signaling in other cytokine and non-cytokine-stimulated signaling systems (Stark and Darnell 2012).
Beyond this simple framework are numerous subtleties and alternative signaling pathways (Ivashkiv and Donlin 2014; Fish and Platanias 2014). In addition to the pStat1/pStat2 complex, a pStat1 homodimer can form, leading to transcriptional activation of genes controlled by the GAS DNA sequences, many of which are pro-inflammatory and are also part of the transcriptional response to IFN-γ. Furthermore, depending on the cell type and its complement of Stat proteins, other Stats, particularly Stat3, may be phosphorylated and activated, driving the transcription of genes that suppress inflammation. Beyond the Jak/Stat pathway, there is convincing evidence that other, more widely used pathways, such as through mitogen-activated protein kinase (p38 MAPK) and phosphoinositide 3-kinase (PI3K), mTORC1, mTORC2, and others can be activated by Type I IFNs. Most reports regarding these pathways implicate them in antiproliferative, antiapoptotic, and/or antitumor effects (Fish and Platanias 2014). Considering the work of Schreiber and Piehler (2015; see below), it is possible that these pathways are engaged by IFN-β, by high local concentrations of other Type I IFNs or signals, or by prolonged IFN exposure (Schreiber and Piehler 2015). Thus, the various pathways that are integrated into the overall physiological response likely result from complex temporal and spatial factors, combined with the specific cytokine milieu and the state of the target cells.
What other factors control the strength and duration of IFN-I signaling? As noted above, IFN-Is are under tight transcriptional control. When mice are injected with IFN, most ISGs are induced rapidly and synchronously, peaking at 2 h post-stimulus and then declining rapidly, in common across cell types (Mostafavi et al. 2016). This pattern reflects the necessity of modulating or ending the potent IFN response, and the required suppression of IFN responses occurs at several levels. Endocytosis and receptor downregulation can be the direct result of IFN binding to IFNAR, although, in more complex physiological situations, these processes can also be stimulated by heterologous pathways, including IL-1, TLRs, or stress pathways (Ivashkiv and Donlin 2014). Importantly, several ISGs induced by IFN-Is encode proteins that negatively feed back into the canonical pathway (Schneider et al. 2014). One is SOCS1, the suppressor of cytokine signaling 1, which can bind phosphorylated Tyk2, thereby inhibiting its activity and other interactions; SOCS1, through a destabilizing effect on Tyk2, may also lead to decreased stability of IFNAR1 (Ivashkiv and Donlin 2014). Another important suppressive ISG is USP18 (“ubiquitin-specific protease 18,” “UBP43”), which can bind to the cytoplasmic domain of IFNAR2, blocking the binding of Jak1 and thereby decreasing IFN-induced Jak/Stat signaling. This binding and inhibitory activity is independent of USP18’s enzymatic activity of de-ubiquinating substrates modified by the ISG protein, ISG15, a ubiquitin-like protein. USP18, as with several of these proteins, functions in a number of cellular processes beyond the Type I IFN system (Honke et al. 2016). The physiological importance of most of these molecules is demonstrated by the observation that most or all “successful” viruses – those that cause effective infections – have one or more genes that abrogate the antiviral protection of the IFN system by blocking, modifying, or destabilizing one or another molecule in the IFN signaling pathway, starting with virus-encoded receptor mimics to compete with IFNAR for IFN binding (Schneider et al. 2014).
Differential Activation by Type I IFNs
While many Type I IFNs have high antiviral activity, they display much more variation in their induction of other physiological/biological activities. Thus, both IFN-α2 (a high-activity IFN-α, widely used therapeutically) and IFN-β have high antiviral activity and high receptor affinity for IFNAR, but IFN-β has 20- to-50-fold higher antiproliferative activity and antitumor activity in animal models. Understanding this differential has been a central question for many years, with implications for differential signaling, possibly providing the ability to design novel IFNs with selective activities for clinical or research purposes. It was recently demonstrated that these functional differences primarily arise from differential receptor interactions (reviewed in Schreiber and Piehler 2015). While most IFN-Is bind to IFNAR2 with high affinity (nanomolar), they bind to IFNAR-1 with lower, and more varied, affinity (micromolar for IFN-αs; 50 nM for IFN-β). Furthermore, IFN receptor subunits are generally expressed at low levels (100–1000 copies per cell) and are likely to be monomeric on the cell surface. Precise biophysical measurements or receptor-ligand interactions, coupled to mutagenesis studies, activity measurements in vitro and in vivo, and modeling led to the current model that provides an understanding of the observed differential activity of Type I IFNs. In this model, IFNs bind first to IFNAR2, followed by recruitment of IFNAR1 to form the ternary signaling complex. While the different affinities of IFN-Is for IFNAR2 are relatively small, the high affinity for IFN-β for IFNAR1 is the major factor conferring its high antiproliferative and antitumor activity relative to the IFN-αs (reviewed in Schreiber and Piehler 2015). The higher affinity of IFN-β for IFNAR1 leads to a stronger, more stable ternary complex of IFNAR2/IFN-β/IFNAR1 than occurs with IFN-α. The functional consequences flow from the differential stability of these complexes. On the one hand, the primary antiviral response to IFN-Is is generated rapidly and at low concentrations of most IFN-Is, dominated by the binding affinity to IFNAR2. The antiviral response activates the Jak-STAT pathway and results in the transcriptional expression of a subset of ISGs, termed “robust” genes that are activated rapidly at relatively low concentrations of IFN-Is. On the other hand, the high affinity of IFN-β for IFNAR1 leads to a more stable ternary complex, likely providing a more sustained, active signaling complex. This, or sustained treatment of cells at high concentrations of IFN-αs, may engage additional signaling pathways, resulting in the transcriptional activation of additional ISGs, denoted “tunable genes,” and leading to antiproliferative, antitumor, and cell-specific responses (Schreiber and Piehler 2015). Differential antiviral, antimicrobial, and therapeutic effects of both human and murine IFN-β and the IFN-αs were recently reviewed in the context of this model of differential receptor engagement, signaling, and cell specificity (Ng et al. 2016).
The Type I IFNs have served as a reference system for investigating cytokines, for the development of cytokine therapeutics, and for the understanding of cytokine signaling. With 60 years of IFN research, many fundamental questions remain. Among these are the following. While the broad outlines of non-Jak/Stat signaling are generally recognized, the questions of how, and in what circumstances, these pathways are employed require additional tools and investigation. The recent discovery of Type III IFNs (IFN-λs; IL-28A, IL-28-B and IL-28-29) as critical actors in antiviral and some antibacterial responses in epithelial cells and mucosal immunity will lead to a reexamination of the role of Type I IFNs in viral infections of the mucosa and the development of an understanding of the complementary and/or overlapping roles of the Type I and Type III IFNs (Kotenko and Durbin 2017). The recent reports that some individuals with primary immune deficiencies (PIDs) in IFNAR2, Stat2, and Tyk2 show, in some affected individuals, a surprisingly narrow antiviral deficiency (Casanova et al. 2012; Hambleton et al. 2013; Duncan et al. 2015), suggest important gaps as we extrapolate from rodent models to the details of Type I IFNs in human biology. Increasing our understanding of IFN signaling in various model animals and in humans should also contribute to continued efforts to harness Type I IFNs and perhaps novel interventions in IFN signaling for their therapeutic value. In addition, there are increasing examples wherein dysregulation of Type I IFN expression, or genetic variants in proteins within the signaling pathways of IFN induction or action, contributes centrally to autoimmune and inflammatory conditions. Studying these conditions in humans and developing validated animal models will contribute both to our understanding of the complexities of the IFN-I system and to the development of new therapeutics for their treatment.
- Honke N, Shaabani N, Zhang D-E, et al. Multiple functions of USP18. Cell Death and Disease. 2016;7;e2444. doi: 10.1038/cddis.2016.326. PMID: 27809302.
- Kotenko SV, Durbin, JE. Contribution of Type III interferons to antiviral immunity: Location, location, location. J Biol Chem. 2017 (epub ahead of publication) doi: 10.1074/jbc.R117.777102