Context Is Key: Delineating the Unique Functions of IFNα and IFNβ in Disease

Type I interferons (IFNs) are critical effector cytokines of the immune system and were originally known for their important role in protecting against viral infections; however, they have more recently been shown to play protective or detrimental roles in many disease states. Type I IFNs consist of IFNα, IFNβ, IFNϵ, IFNκ, IFNω, and a few others, and they all signal through a shared receptor to exert a wide range of biological activities, including antiviral, antiproliferative, proapoptotic, and immunomodulatory effects. Though the individual type I IFN subtypes possess overlapping functions, there is growing appreciation that they also have unique properties. In this review, we summarize some of the mechanisms underlying differential expression of and signaling by type I IFNs, and we discuss examples of differential functions of IFNα and IFNβ in models of infectious disease, cancer, and autoimmunity.


INTRODUCTION
Interferons (IFNs) are cytokines that were originally discovered and named for their ability to interfere with viral replication (1). IFNs are grouped into three classes according to the receptor that mediates their effects: type I IFNs (the focus of this review), type II IFN (IFNg), and type III IFNs (IFNls) (2,3). Broadly speaking, each IFN class signals through receptor-associated Janus kinases (JAKs), which activate various Signal Transducer and Activator of Transcription (STAT)-signaling pathways. Type I IFNs signal through the heterodimeric IFN-a/b receptor 1 (IFNAR1) and IFNAR2, which are associated with the JAKs tyrosine kinase 2 (TYK2) and JAK1, respectively (4). Canonically, activation of TYK2 and JAK1 leads to the formation of the IFN-stimulated gene (ISG) factor 3 (ISGF3) complex, composed of STAT1, STAT2, and interferon regulatory factor 9 (IRF9). The ISGF3 complex then translocates to the nucleus to regulate the expression of hundreds of IFN-stimulated genes. Type I IFN signaling can activate other STAT complexes, often in a celltype dependent manner. Additionally, alternative signaling cascades, including the mitogenactivated protein kinase p38 pathway and the phosphatidylinositol 3-kinase pathway, are also required for optimal generation of type I IFN responses (4).
Type I IFNs have broad, pleiotropic effects that include antiviral activity, antiproliferative effects, and immunomodulatory properties. There is growing evidence that the overall outcome of type I IFN responses can be beneficial or detrimental for the host depending on the timing, magnitude, and source of IFN production, as well as the specific biological context (5). Moreover, despite signaling through a shared receptor, type I IFN subtypes possess important functional differences, both in vitro and in vivo. The purpose of this review is to summarize the current understanding of differential type I IFN properties, focusing on the role of human and mouse IFNa and IFNb in infectious disease, cancer, and autoimmunity. In particular, we seek to highlight the few examples that demonstrate or suggest differential activities for type I IFN subtypes in vivo.

TYPE I IFNS: A MULTIGENE FAMILY
Type I IFNs exist as a multigene family across many species ( Figure 1) (6). IFNas, IFNb, IFNϵ, IFNk, and IFNw are found in many species, whereas IFNd and IFNt are only found in pigs and cattle (7). In humans (HuIFN), the type I IFN genes are located on chromosome 9 and encode 13 IFNa subtypes and single forms of IFNb, IFNϵ, IFNk, and IFNw (7). Type I IFNs in mice (MuIFN) are located on chromosome 4, and likewise, consist of multiple genes with some differences compared to human. MuIFNs include 14 IFNa subtypes, IFNb, IFNϵ, IFNk, an IFN-like cytokine IFNz (also known as limitin), but lack a functional IFNw, which is present as a pseudogene (8).
Phylogenetic analyses reveal that the type I IFN subtypes form clades consistent with mammalian speciation (7,9,10). For the most part, placental mammals possess single copies of the genes encoding IFNk, IFNb, and IFNϵ, and these unduplicated subtypes represent the first major clade within mammalian IFNs (11). IFNk is the first subtype to diverge within mammalian type I IFNs and forms an outgroup, possibly the result of a unique evolutionary route for IFNk relative to IFNb and IFNϵ (11). IFNk is additionally distinctive as the only A B FIGURE 1 | Type I IFNs are a closely related family of related cytokines. (A) Depicted is a summary of existing phylogenetic analyses of the type I IFNs. The branches are not drawn to scale. IFNk, IFNb, and IFNϵ are mostly present in placental mammals as single copies and the first subtypes to diverge from the other type I IFNs. IFNb and IFNϵ are especially similar and can be found within the same clade in some analyses. IFNd and IFNz are the next subtypes to diverge and are only found in pigs and mice, respectively. IFNt and IFNw are closely related, despite their differences in function and distribution-IFNt is only expressed in placental tissues of ungulate species and involved in pregnancy, whereas IFNw is found in many species and possesses the more canonical antiviral and immunomodulatory functions. IFNw and IFNa loci are expanded to include many subtypes in a number of species. (B) The chromosomal locations of human (top) and murine (bottom) IFNk, IFNb, and IFNϵ genes are depicted. The arrow direction indicates on which strand the gene is encoded: a left-to-right arrow depicts the forward or positive strand and a right-to-left arrow indicates the reverse or negative strand. IFNk is the only subtype to contain an intron and is situated further away from the other type I IFNs, though its positioning relative to the other IFNs is different in mice and humans. IFNb and IFNϵ roughly form the boundaries of the type I IFN locus, with the other type I IFNs falling between the two genes. mammalian type I IFN that contains an intron, and for many species, the gene encoding IFNk is situated further away from the IFN locus (7,9,11). Depending on the analysis, IFNb or IFNϵ is the next subtype to diverge from mammalian type I IFNs, and in some analyses IFNb and IFNϵ fall within the same clade, suggesting that these subtypes might be more closely related to each other than the other type I IFN subtypes (7,9,11,12). The genes encoding IFNb and IFNϵ are situated at the "beginning" and "end" of the type I IFN locus across many species, which is relatively conserved across mammalian species. IFNd and IFNz (limitin) are the next type I IFNs to diverge within mammalian IFNs and are only found in pigs and mice, respectively (7). However, recent identification of a putative HuIFNd gene calls this into question (11).
The last subtypes to diverge are the IFNas, IFNws, and IFNts. These subtypes are thought to be exclusively found in placental mammals and are usually situated between the IFNϵ and IFNb genes within the type I IFN locus. IFNw and IFNt are closely related, even though they possess different functions (7,11). IFNt is only found in placental tissues of ungulate species, is involved in pregnancy, and may have arisen from an IFNw subtype (10,13). In contrast, IFNw is an antiviral and immunomodulatory molecule, like IFNa, and functional copies have been identified in humans and other animal groups including felines, pigs, cattle, serotine bats, and others but are not present in canines or mice (14). Notably, humans have only one IFNw, but there is evidence that IFNw is still expanding and diversifying in many species, including bats and pigs (15)(16)(17). Lastly, the genes encoding IFNa are found in all placental mammals and form species-specific clades, with some exceptions for closely related organisms (e.g. chimpanzees, humans, and gorillas); a combination of gene duplication and gene conversion events likely gave rise to the expanded IFNa genes present in many mammals (6). Of note, a recent study found that for some IFNa subtypes, such as HuIFNa6, a8, a13, and a14, amino acid-altering variation was more constrained in the human population, suggesting that they might perform non-redundant functions in host responses (18).
As sequenced genomes of other species become available, the phylogenetic clustering of some type I IFNs may change. However, the key point is that the multigene nature of type I IFNs is conserved across many species. Both IFNa and IFNw subtypes expanded independently and multiple times, suggesting that it is advantageous for the host to possess a large repertoire of at least several type I IFN subtypes. Unfortunately, the fact that type I IFNs expanded multiple times complicates directly applying results of IFN studies from animal models to clinical settings, and caution is warranted in drawing conclusions about specific human IFNa subtypes from studies of murine IFNa subtypes.

MOLECULAR MECHANISMS UNDERLYING DISTINCT FUNCTIONS OF TYPE I IFNS
Though type I IFNs possess many overlapping functions, it is now appreciated that the individual subtypes have different potencies of their shared functions and some unique functions in vitro. An important early example demonstrating this was the finding that HuIFNb was 100-fold more potent than HuIFNa2 in inhibiting osteoclastogenesis through its ability to preferentially induce the chemokine CXCL11 (19). Since this observation, it is now appreciated that the pleiotropic activities ascribed to different type I IFN subtypes are the product of distinct patterns and kinetics of expression, as well as signaling differences that arise from differential binding affinities and susceptibility to negative feedback loops (20,21). The ability of the type I IFN receptor to have fine-tuned responses to many ligands is likely advantageous considering the array of pathogens that have co-evolved alongside humans, mice, and other animals.

Differential Dependence on IRF3 and IRF7 for Transcription
Before examining the signaling and functional properties of IFN subtypes, it should be noted that type I IFNs are differentially induced downstream of pattern recognition receptor (PRR) signaling, except for IFNϵ, which is hormonally regulated (see below). PRR signaling converges on the phosphorylation and activation of the transcription factors IRF3 and IRF7, though other IRFs can be involved in IFN-dependent antiviral responses (22,23). For most cell types IRF3 is constitutively expressed, whereas IRF7 is induced downstream of type I IFN signaling to then amplify and diversify the type I IFN response (22). The exception to this rule is plasmacytoid dendritic cells (pDCs), which constitutively express IRF7 and are thus poised to rapidly secrete large amounts of type I IFN (24). The promoters of specific type I IFN genes differ in their requirement of IRF3 or IRF7 binding for maximal transcription. Thus, the temporal regulation of IRFs dictates the expression of IFN subtypes.
Early in a response, IRF3 activation first induces transcription of MuIFNb and MuIFNa4 via unique IRF3 binding sites within their promoters (25)(26)(27)(28)(29)(30)(31). For the most part, the other MuIFNa subtypes require both IRF3 and IRF7 for maximal transcription, and so they depend on type I IFN-mediated upregulation of IRF7 (32)(33)(34). Similar to mice, IRF3 also initiates human type I IFN responses by upregulating transcription of HuIFNb and HuIFNa1, while the other HuIFNA genes require both IRF3 and IRF7 (35,36). Altogether, these findings demonstrate that for most cell types, activation of constitutive IRF3 by PRR signaling initiates a first wave of HuIFNb and HuIFNa1 (or MuIFNb and MuIFNa4 for mice). Subsequently, a second, amplified wave of diverse IFNa subtypes follows that is IRF7dependent. As the ratio of IRF3 to IRF7 or other IRFs changes over time, the repertoire of IFN subtypes expressed changes as well.
There are several intriguing deviations from this paradigm. First, the IFNb promoter has additional response elements that make it responsive to NF-kB signaling through activating transcription factor 2 (ATF-2) and c-Jun, which allows other signaling pathways to augment IFNb production (29,37,38). This unique promoter feature also permits IRF3-independent basal expression of low amounts of IFNb in the absence of infection, which can have significant impact on mounting successful innate immune responses against a variety of infections (39)(40)(41)(42)(43)(44)(45)(46)(47). IFNk may have somewhat restricted expression, as it was named for its high expression in keratinocytes; however, other cell types, including immune cells and lung epithelial cells, can upregulate IFNk expression (48)(49)(50). Further characterization is needed to determine which cells are capable of expressing IFNk in different contexts. Lastly, IFNϵ is the most notable exception to the IRF-mediated IFN induction paradigm, as it is not regulated at all by PRR signaling and IRF3/7. Instead, it is constitutively expressed in the epithelium of reproductive organs and hormonally regulated, and this is reflected in its unique promoter (51)(52)(53).

Differential Binding Affinity Determines Signaling and Function
All type I IFNs bind to and signal through the heterodimeric receptor IFNAR1 and IFNAR2 to activate canonical JAK/STAT signaling pathways (4). A unique feature of type I IFN signaling is that the signaling outcome can vary depending on the cell type, specific ligand, and concentration of the type I IFN subtype. The molecular mechanisms that underlie the plasticity of type I IFN signaling have been extensively reviewed elsewhere, so only key features will be outlined in this review (20,54,55).
In general, IFNAR2 is the primary ligand binding receptor subunit and binds type I IFNs with high affinity (typically nanomolar affinity); IFNAR1 is subsequently recruited to the receptor-ligand complex and binds with relatively lower affinity (approximately micromolar affinity) (54). HuIFNb has the highest natural binding affinity to the type I IFN receptors with picomolar affinity for IFNAR2 and nanomolar affinity for IFNAR1, whereas HuIFNa2 possesses nanomolar affinity for IFNAR2 and micromolar affinity for IFNAR1 (56)(57)(58). This higher affinity interaction may enable IFNb to uniquely signal through IFNAR1 in an IFNAR2-independent manner, but further work is needed to corroborate this finding and to determine if other receptors are involved in this phenomenon (59,60). Engineered IFNa2 and IFNw mutants that mimic the range of affinities for the receptor complex have demonstrated that type I IFN signaling outcomes can be directly linked to IFN affinity to the receptor complex. Hence, type I IFN mutants that acquire IFNb-like affinity acquire IFNb-like potency (61,62).
In line with these findings for IFNa, IFNb, and IFNw, recent work showed that HuIFNϵ and HuIFNk bound IFNAR2 with particularly weak affinity and demonstrated approximately 1000fold decreased potency in ISGF3-mediated gene expression compared to HuIFNa2, whereas their affinity for IFNAR1 was comparable to other type I IFN subtypes (63). HuIFNϵ and HuIFNk also bound the poxvirus antagonist B18R with weaker affinity relative to the other IFN subtypes, perhaps suggesting a fitness advantage for the host to have some weaker binding IFN subtypes in order to avoid virus inhibition (63). In influenza A virus (IAV) infection, HuIFNk, but not IFNa or IFNb, relied on chromodomain helicase DNA binding protein 6 (CHD6) to efficiently suppress viral replication (50). Moreover, induction of CHD6 was not dependent on STAT1, but rather, IFNk signaled through the mitogen-activated protein kinase (MAPK) p38 and the transcription factor c-Fos to mediate its antiviral effects. Altogether, these findings suggest that in addition to having unique expression patterns, IFNϵ and IFNk may possess additional biochemical and signaling features that grant unique properties in vivo.

Differential Sensitivity to Feedback Loops
The affinity of individual subtypes, as outlined above, is a key component in determining the signaling outcome from IFNAR1/ 2 engagement, but negative feedback loops are an additional level of regulation and fine-tuning. IFNAR1/2 surface abundance is typically quite low, and modulating the surface receptor expression is one means of regulating type I IFN signaling after type I IFN induction (64). Manipulation of a cell line's IFNAR expression demonstrated that the antiproliferative and proapoptotic activities induced by HuIFNb are less sensitive to decreased receptor levels than those induced by HuIFNa2 (65,66). The physiological relevance of receptor expression influencing type I IFN signaling is demonstrated in the number of IFN-dependent mechanisms that downregulate IFNAR1 and IFNAR2 levels. We will outline a few examples.
First, protein kinase D2 (PKD2) is a negative regulator activated downstream of IFN signaling. It phosphorylates IFNAR1, enabling interaction with a ubiquitin E3 ligase, and subsequent ubiquitination leads to endocytosis of the IFN signaling complex (67,68). Endosomes with short-lived receptor-ligand complexes formed by lower affinity IFNas are more likely to be recycled to the cell surface; endosomes with longer-lived complexes formed by higher affinity IFNb ultimately fuse with the lysosome, but signaling can continue to take place as trafficking progresses through the endosomal compartment (69)(70)(71)(72). Second, Suppressor of Cytokine Signaling 1 (SOCS1) can directly dampen the type I IFN response by interacting with TYK2 to disrupt TYK2-STAT signaling, but it also decreases surface levels of IFNAR1, which requires TYK2 for stability at the cell surface (73). Lastly, ubiquitin-specific peptidase 18 (USP18) can bind the cytoplasmic domain of IFNAR2 and interfere with IFNAR1 recruitment and ternary receptor complex formation without decreasing surface IFNAR2 levels (74,75). The USP18-IFNAR2 interaction makes it so that only higher affinity ligands such as IFNb are able to recruit IFNAR1 into the receptor complex, making the cell less responsive to weaker affinity type I IFNs (76,77). subtypes may be able to signal through alternative pathways, in spite of or, more likely, because of possessing lower binding affinity. Understanding the molecular mechanisms underlying differential signaling by IFNs is an active area of research and how the differential activities of IFNa and IFNb impact disease will be explored in the remaining sections.

INFECTIOUS DISEASES
Type I IFNs have been extensively studied in the context of infectious diseases, and this body of work includes most of the studies that have directly compared the functions of IFNa and IFNb in vivo. In the following subsections we highlight key findings from animal models and human studies that have contributed to understanding the mechanisms of differential properties of IFNa and IFNb in viral, bacterial, and parasitic infections.

Viral Infections
The important role that viral infections have served in helping us understand type I IFN biology cannot be understated. Viral infections were key instruments in the discovery of the antiviral properties of type I IFNs (1). It is now widely appreciated that type I IFNs play a much larger role in coordinating protective immunity beyond directly eliciting an antiviral state, including their role in DC maturation, augmenting antibody production by B cells, and improving cytolytic T cell effector functions (5). Intriguingly, type I IFNs can also play a detrimental role in certain contexts, such as persistent viral infections. Given their key roles in disease outcome, viral systems also include some of the clearest examples of differential functions of IFNa and IFNb in vivo ( Table 1). The following viral models collectively highlight that differential functions of IFNas and IFNb can profoundly influence disease pathogenesis and that the mechanisms underlying differential functions vary depending on the biological context.

Lymphocytic Choriomeningitis Virus
Lymphocytic choriomeningitis virus (LCMV) is a nonlytic, negative-strand RNA virus and a prototypic member of the Arenaviridae family, which are causative agents of hemorrhagic fevers in humans (100). The host genetics, viral strain, dose, and inoculation route all have profound impacts on host responses and disease outcome, and this remains true for the role of type I IFN responses during LCMV pathogenesis (101). LCMV infection serves as an excellent example of the pathogenic potential of type I IFNs.
LCMV-Clone-13 (Cl-13), which differs from its parent strain LCMV-Armstrong (Arm) by just three amino acids, causes a persistent infection, whereas LCMV-Arm is acutely and effectively cleared by immunocompetent mice (102). A clear pathogenic role for type I IFNs during persistent LCMV-Cl-13 infection has been established (78,79,(103)(104)(105). Loss of IFNAR1 caused increased viral loads early during infection but ultimately restored splenic organization, decreased expression of the negative immune regulators IL-10 and programmed deathligand 1 (PD-L1), increased protective adaptive immune responses, and accelerated clearance of persistent virus (78,79,105). While both LCMV-Arm and LCMV-Cl-13 infection led to high IFNa levels in the serum, only LCMV-Cl-13 induced significant serum IFNb (79). In a seminal study, Ng and colleagues showed that the pathogenic activity of type I IFNs in persistent LCMV infection could be ascribed to just one subtype-IFNb. Using monoclonal antibody (mAb) blockade and genetic deletion, they showed that IFNb was dispensable for controlling early LCMV-Cl-13 viral loads, suggesting that IFNa or other subtypes mediate these antiviral responses (80). Instead, blockade of IFNb but not IFNa improved splenic architecture, decreased infection of CD8a − DC, and enhanced antiviral T cell responses that led to clearance of persistent virus, mimicking many of the effects seen with IFNAR1 blockade. Altogether, persistent LCMV-Cl-13 infection serves as an important example that the type I IFN subtypes can have distinct properties in vivo that have profound impacts on viral pathogenesis.
As discussed above, LCMV-Cl-13 infection causes persistent infection in certain mouse strains (C57BL/6, BALB/C, C3H, or SWR/J); however, LCMV-Cl-13 infection of other strains (NZB, SJL/J, PL/J, NZO, or FVB/N mice) causes type I IFN-and CD8 T cell-dependent severe vascular leakage and death by about 6-8 days post infection (dpi) (81,82,106,107). NZB.Ifnar1 −/− but not NZB.Ifnb −/− mice were protected from LCMV-Cl-13 induced lethal vascular leakage, suggesting that IFNb is dispensable for the detrimental effects of type I IFN in this model and that other subtypes like IFNa may drive this phenotype (81). However, this is challenged by the fact that blockade of IFNb alone, pan-IFNa (a1, a4, a5, a11, and a13) alone, or combined pan-IFNa/b did not replicate the protection provided by anti-IFNAR1 treatment in FVB/N mice (82). The inability of IFNb or IFNa blockade to phenocopy IFNAR1 blockade could be due to dosing issues, as the serum levels of IFNa were severely elevated (roughly 18-fold over IFNb levels), involvement of IFNa subtypes not blocked by the mAb, or involvement other type I IFN subtypes altogether could be responsible for the lethal phenotype. Nevertheless, type I IFNs are clearly important host determinants of lethal LCMV infection, and the individual IFN subtype(s) responsible remains an open question.

Chikungunya and West Nile Viruses
Chikungunya virus (CHIKV) is a mosquito-transmitted, reemerging alphavirus that causes outbreaks of acute fever, rash, polyarthritis, arthralgia, and myositis (108). West Nile virus (WNV) is a mosquito-transmitted flavivirus that can cause encephalitis in severe cases (109). It is helpful to consider these models together because both models utilize a peripheral route of infection by inoculating the footpad subcutaneously (s.c.), and type I IFNs are essential for controlling both CHIKV and WNV, as Ifnar1 −/− mice rapidly succumb to a severe, disseminated infection with either virus (83,84,88,89). The collective evidence from these models suggest that IFNa and IFNb play nonredundant protective roles.
loads in both peripheral and central nervous system (CNS) tissues compared to WT animals (34,85). Similarly, Irf7 −/− mice infected with CHIKV developed worse clinical disease (foot swelling) and sustained high viral loads at the site of infection and sites of dissemination (90)(91)(92). The poor clinical outcome of Irf7 −/− mice during WNV and CHIKV infection may be the result of decreased IFNa activity in the serum (85,86,91,92). This postulation is supported by the observation that Irf7 −/− mice produce little to no systemic IFNa activity when infected with a number of viruses, including Dengue virus (DENV), herpes simplex virus 1 (HSV-1), and encephalomyocarditis virus (EMCV), and this loss of systemic IFNa activity correlated with increased susceptibility to those infections (34,110,111). Pan-IFNa mAb blockade closely mimicked the clinical and virologic phenotype of Irf7 −/− mice in CHIKV infection and phenocopied the lethality observed in WNV infection (86,90). Altogether, these findings suggest that an important protective function of IRF7 is the production and amplification of IFNa responses and that IFNas are important for controlling viral replication and dissemination.
In contrast with IFNa, the role of IFNb in vivo is more varied and dependent on the biological context. Ifnb −/− mice are more susceptible than WT mice to WNV infection, and this increased lethality was accompanied with elevated viral burden in some but not all tissues (87). Specifically, WT and Ifnb −/− mice similarly controlled WNV replication in the spleen and serum, consistent with IFNa subtypes dominating serum IFN activity. WNV did replicate to a larger extent in the brain, spinal cord, and the draining lymph in Ifnb −/− mice compared to WT mice (87). An antiviral role for IFNb has also been described for vaccinia virus and IAV infections (93,94). In contrast to WNV infection, loss of IFNb exacerbated CHIKV-induced disease but with minimal impact on viral burden at the inoculation site or distant tissues, suggesting that IFNb may be important in restricting viral replication within certain but not all tissues (90). Rather, the increased disease severity of CHIKV-infected Ifnb −/− mice correlated with increased neutrophil accumulation at the site of infection, and depletion of neutrophils in Ifnb −/− mice reversed the disease exacerbation to WT levels. Altogether, these data from CHIKV and WNV infections point to the particular importance of IFNa subtypes in restricting viral replication and spread and highlight that the primary role of IFNb varies depending on the specific context.

Human Immunodeficiency Virus 1 and Friend Retrovirus
Human immunodeficiency virus 1 (HIV-1) is a highly pathogenic retrovirus that leads to acquired immunodeficiency syndrome (AIDS). The relationship between type I IFNs and HIV-1 pathogenesis is complex, and it is outside the scope of the this review to cover all the protective and pathogenic functions, which have been extensively reviewed elsewhere (112)(113)(114). The purpose of reviewing HIV and Friend retrovirus (FV) infection is not to delve into whether type I IFNs have a net protective or pathogenic role, but rather, we seek to underscore that the IFNa subtypes are not equivalent in their antiviral or immunomodulatory properties in vivo.
Harper and colleagues evaluated the mRNA expression of specific IFNa subtypes in human pDCs following HIV-1 exposure (115). Intriguingly, they found an inverse relationship between the subtypes induced and their antiviral potency. HuIFNa1/13 and HuIFNa2 were highly expressed, but they demonstrated weaker antiviral activity in vitro, whereas HuIFNa6, a8, and a14 represented a smaller fraction of the IFNa subtypes induced but demonstrated the highest antiviral activity against HIV-1. Likewise, a study from Lavender and colleagues showed that therapeutic administration of HuIFNa14 was more beneficial than administration of HuIFNa2 in controlling HIV-1 replication in a humanized mouse model (116). The efficacy of IFNa14 was associated with increased ability to stimulate intrinsic immune responses including expression of tetherin and Mx2 as well as a greater frequency of TRAIL + natural killer (NK) cells. Conversely, IFNa2 was superior in increasing the frequency of CD8 + T cells. An additional study used humanized mice that lack pDCs (Hu-PBL mice) and do not express much endogenous type I IFN during acute HIV-1 infection to study the impact of IFNa subtypes. They performed a single hydrodynamic injection of plasmid encoding different type I IFN subtypes (HuIFNa2, a6, a8, a14, or b) into Hu-PBL mice prior to HIV-1 infection (117). The authors found that all subtypes tested limited HIV-1 replication and prevented HIV-induced CD4 + T cell depletion by 10 dpi, but only HuIFNa14-and HuIFNb-expressing mice demonstrated this protective effect out to 40 dpi. Altogether these findings demonstrate nonredundant functions of IFNa subtypes, with HuIFNa14 emerging as an intriguing subtype for further studies during HIV-1 infection.
Distinct properties of murine IFNa subtypes have also been observed in FV infection, a commonly used murine retrovirus model. A protective role for type I IFNs in controlling FV infection in vivo was demonstrated with Ifnar1 −/− and Ifnb −/− mice both having increased viral loads in the spleen. However, only Ifnar1 −/− mice showed a significant increase in viremia (95). These findings suggest that both IFNa and IFNb protect against FV infection, but IFNa may be more important for controlling systemic infection and dissemination. Different potencies among IFNa subtypes have also been revealed. Ex vivo stimulation of FV-specific CD8 + T cells demonstrated differential activities among the IFNa subtypes. IFNa4, a6, and a9 had the strongest effects on CD8 + T cells, including inhibiting proliferation, stimulating cytokine production, and enhancing cytotoxicity (118). Treatment of FV-infected mice with MuIFNa1, a4, or a9, but not a6, significantly decreased viral loads, and subtype effectiveness was associated with different mechanisms (96). Only IFNa1 treatment correlated with activated FV-specific CD8 + T cells in the spleen, whereas NK cell activation was observed after treatment with all examined IFNa subtypes. Another study demonstrated that prophylactic administration of MuIFNa11, but not a2 or a5, significantly reduced viral loads by activating NK cells and ultimately provided long-term protection (6 weeks) (97). Together with the HIV-1 studies, retroviruses have proven to be effective tools for probing the diverse functions IFNa subtypes.

Hepatitis B and Hepatitis C Viruses
Hepatitis B (HBV) and hepatitis C viruses (HCV) are drastically distinct pathogens from a virological perspective-HBV is a double-stranded DNA virus belonging to the Hepadnaviridae family, whereas HCV is a positive-strand RNA virus and a member of Flaviviridae. However, both viruses display tropism for hepatocytes, and chronic infection with either virus can lead to liver failure, cirrhosis, and hepatocellular carcinoma (119). Beginning in the 1980s, derivatives of recombinant HuIFNa2 were used to treat chronic HBV and HCV, but treatment was successful in a limited subset of patients and severe side effects were common [reviewed in reference (120)]. These issues have led to the phasing out of type I IFN-based therapeutics in favor of direct-acting antiviral drugs (120). Though HuIFNa2-based therapeutics are the only approved type I IFN therapies for HCV or HBV treatment, pilot studies of IFNb therapy in IFNanonresponding HBV or HCV patients suggest some beneficial effects of IFNb as well (121)(122)(123). These findings suggest that other IFN subtypes in addition to IFNa2 may offer protective effects against hepatitis viruses.
Indeed, one study with the HBV hydrodynamic injection model demonstrated that prophylactic treatment with MuIFNa4 or a5 was more effective than other IFNa subtypes in decreasing HBV replication in vivo, and both a4 or a5 also increased effector NK and CD8 + T cell frequencies in the liver and spleen (98). Hydrodynamic injection of plasmids expressing MuIFNa4, a5, or combined a4 and a5 was more effective than treatment with the respective recombinant proteins, highlighting the importance of long-lasting endogenous IFNa expression in the liver during HBV infection. Another study directly showed differential effects of IFNa4 and IFNb in the hydrodynamic injection HBV model (99). Co-injection of a plasmid encoding MuIFNa4 with HBV DNA decreased HBV serum markers, elevated liver ISG expression, and reduced HBV + cells in the liver, whereas co-injection of an IFNb-expressing plasmid demonstrated weaker inhibition of HBV and surprisingly led to a transient increase in HBV + hepatocytes. This increase in HBV + hepatocytes was not observed if the IFNb plasmid was injected 14 dpi instead of co-injected with HBV (99). Even as the currently approved type I IFN therapies are being phased out of clinical use, these findings add to the accumulating evidence of distinct potencies and functions of IFNa and IFNb subtypes in mouse models of relevant human pathogens.

IFNw Subtype Differences
IFNw is understudied compared to IFNa/b subtypes likely because mice lack a functional IFNw, but there is much in vitro evidence that it signals and functions similarly to IFNa/b (61,124). Humans have only one IFNw subtype, but several species possess an expansion of IFNw genes (15-17, 125, 126). A number of these IFNw subtypes have been cloned from several species and have been demonstrated to be functional type I IFNs (127)(128)(129). Just as there is growing appreciation that expanded IFNa subtypes provide an evolutionary advantage beyond redundancy, it stands to reason that the expansion of IFNw genes likewise imparts a fitness advantage for those species.
Indeed, a recent study compared two different IFNw subtypes from Rousettus aegyptiacus bats and found that IFNw9 displayed more effective antiviral activity against several RNA viruses in vitro compared to IFNw4 (130). Additionally, differences in expression and activity of porcine IFNw subtypes have also been demonstrated, with IFNw7 demonstrating the best antiviral activity in vitro (131). Several of these animals with expanded IFNw subtypes represent important reservoirs and transmitters of relevant human pathogens, so IFNw functional studies may provide valuable information on understanding the interactions between pathogens and their natural hosts.

Remarks on Viral Infections
When type I IFNs act on the proper cell type at the opportune time, they can induce an antiviral state, promote apoptosis of virally infected cells, coordinate recruitment of immune cells, enhance activation of antigen-presenting cells, and augment protective B and T cell responses. Not all IFNs are equal in their ability to induce these protective effects, and exploring this idea in vivo is an active area of research. Studies from infection with LCMV, WNV, and CHIKV have made it evident that endogenous IFNa subtypes are particularly important for limiting viremia and viral spread, likely due to their abundant activity in the serum in a number of viral infections. In peripheral tissues, IFNas and IFNb can exert important antiviral or immunomodulatory activity. Whether a particular subtype emerges as more important than others is likely going to depend on its biochemical properties, the cellular tropism of the virus, the source and magnitude of its induction, how long its expression is sustained, and the specific cell types responding to IFN.
If type I IFN signaling is sustained too long, immunosuppression and viral persistence can occur through the upregulation of negative immune regulators, like IL-10 and PD-L1. LCMV infection is a good example of this scenario, and strikingly, IFNb was critical in promoting many detrimental features of type I IFN signaling in this model. We did not have space to discuss the growing evidence that type I IFNs can promote tissue damage during acute viral infections by promoting excessive inflammation and cell death [discussed in references (132,133)]. This has been observed for mouse strains highly susceptible to influenza or coronavirus infection (134)(135)(136). The mechanisms responsible for these detrimental effects of type I IFN are an active area of research, but initial observations suggest that excessive or delayed IFN induction may play a role. It is also unknown whether specific IFN subtypes are responsible for these effects. Future studies exploring this possibility could have an important impact on human disease.

Bacterial Infections
Type I IFNs can play a pathogenic or protective role during bacterial infection depending on the pathogen. The mechanisms underlying the beneficial or detrimental roles during bacterial infection remain poorly understood and warrant further study. Below we explore some of the properties of type I IFNs during models of bacterial infections ( Table 2). However, compared to the examples from viral infections, few of these studies directly compare the functions of IFNa and IFNb. We draw attention to a few instances in which specific subtypes have been examined and highlight areas where this may be an interesting avenue to explore.

Mycobacterium Tuberculosis
Mycobacterium tuberculosis (Mtb) causes the disease tuberculosis and represents a global health burden. This intracellular pathogen primarily infects the lungs, and it can enter latency if it is not eliminated, persisting in granulomas (154). The actions of type I IFNs during Mtb infections are complex, and there are numerous examples of contradictory findings. Overall, there is strong evidence that type I IFNs are detrimental to the host, but depending on the timing of IFN induction, the bacterial strain, and host genetics, IFNs may occasionally benefit the host during infection [reviewed in reference (155)].
Numerous studies have shown a type I IFN-inducible transcriptional profile in blood isolated from patients with active tuberculosis, but this signature is typically absent in patients with latent infection or patients who have undergone successful treatment (156)(157)(158). Concordantly, infection with hypervirulent Mtb laboratory strains showed increased recruitment of type I IFN-producing pDCs and classical DCs and elevated expression of IFNa or IFNb in the lung, depending on the study (138,139,(159)(160)(161)(162). Multiple studies with human and mouse models have shown that type I IFNs are associated with impaired IFNg-mediated antibacterial effects, decreased expression of IL-1a and IL-1b, decreased production of prostaglandin E2 (PGE2), and upregulation of IL-10 (138-142, 159, 162-165). Type I IFNs are also associated with increased cell death of macrophages and increased recruitment of myeloid cells permissive to Mtb infection (137,143). Limited work has addressed the pathogenic potential of individual type I IFNs, but one recent study found that in vitro blockade of IFNa (subtypes unspecified), but not IFNb blockade, significantly decreased intracellular Mtb bacterial load in a macrophage cell line (166). It remains to be determined if a similar effect could be observed in vivo. Despite all of the evidence pointing to detrimental effects of type I IFNs in Mtb infection, type I IFNs may play a beneficial role in particular circumstances. First, several case reports have suggested that coadministration of IFNa with antimycobacterial therapy decreased bacterial burden in individuals who failed to respond to antimycobacterial therapy alone (167)(168)(169)(170). However, these studies were employed before the pathogenic effects of type I IFNs were appreciated, and the mechanisms driving the apparent protection remain elusive. Second, in agreement with the findings that the detrimental effects of type I IFNs are largely due to inhibition of IFNg, type I IFNs appear to be protective in contexts of IFNg deficiency. Mice lacking both type I and type II IFN receptors displayed increased mortality and pathology compared to mice lacking only the type II IFN receptor in Mtb infection (171,172). Mechanistically, type I IFNs may dampen recruitment of Mtb-permissible macrophages and suppress macrophages from entering an alternative activation state. In accord with these mice studies, administration of IFNa2b combined with antimycobacterial chemotherapy had beneficial effects in Mtb-infected children with underlying IFNg signaling deficiencies (173,174). It is unclear whether IFNb can induce these effects as well. Further head-to-head comparison studies of IFNa and IFNb are needed to determine if this protective effect of type I IFNs is unique to IFNa.
Type I IFNs may also benefit the host in infection with less virulent Mycobacterium strains, such as the bacille Calmette-Gueŕin (BCG) vaccine derived from M. bovis (175,176). Administration of IFNa at the time of BCG vaccination (s.c.) in mice followed by intramuscular IFNa boosts (subtype not disclosed) promoted production of IFNg, tumor necrosis factor (TNF), and IL-12, thus slightly increasing the protection seen upon re-challenge with Mtb intranasal (i.n.) compared to immunization with BCG alone (175). Moreover, the bacterial ESX-1 secretion system promotes type I IFN induction, and its recombinant expression in the BCG vaccine better protected against Mtb infection than other versions of the vaccine (176)(177)(178)(179). In vitro data also highlight the complexity of type I IFN functions, as pretreatment of permissible cells with IFN before Mycobacterium infection can promote bacterial growth or increase immune activation, depending on the cell type and bacterial strain (180,181). Thus, type I IFNs may play a protective role in vaccination with weaker Mycobacterium strains.

Salmonella enterica Serovar Typhimurium
Salmonella is a common, pathogenic genus of bacteria that causes acute gastroenteritis. Type I IFNs largely play a pathogenic role in Salmonella infection by promoting necroptosis and suppressing protective innate cell recruitment and proinflammatory responses. Deletion of IFNAR1 increased survival of adult mice infected (i.v.) with S. enterica serovar Typhimurium (S. Typhimurium) and decreased splenic bacterial loads (144). Additionally, splenic macrophages in Ifnar1 −/− mice were resistant to S. Typhimurium-induced necroptosis ex vivo, and a follow-up mechanistic study further determined that type I IFN signaling impaired antioxidative stress responses to S. Typhimurium infection of bone marrow-derived macrophages (144,182). IFNb may be the dominant type I IFN subtype driving this necroptosis phenotype, as blockade of IFNb, but not IFNa, prevented necroptosis and enhanced survival of bone marrow-derived macrophages during S. Typhimurium infection in vitro (144). It is unclear how many IFNa subtypes the antibody used blocks (clone: RMMA-1), so it is premature to rule out a contribution of IFNa. A role for IFNb was further demonstrated in a separate study which showed that Ifnb −/− mice were more resistant to oral infection of S. Typhimurium, which was characterized by decreased bacterial burden, dampened expression of IL-10, and increased levels of CXCL2 and myeloperoxidase activity in the liver (145). Altogether, these findings suggest that IFNb may play a detrimental role in S. Typhimurium infection by negatively regulating protective immune responses, but further studies are needed to rule out the involvement of other type I IFN subtypes.

Listeria monocytogenes
Listeria monocytogenes is an intracellular, pathogenic bacteria that causes sepsis and meningitis in immunocompromised and pregnant individuals (183). Many groups have shown that type I IFN signaling is detrimental to the host in systemic L. monocytogenes infection, but not in all routes of infection (150-153, 184, 185). Despite the important role that type I IFNs play in L. monocytogenes pathogenesis, the contribution of individual subtypes remains unknown. Irf3 −/− mice displayed increased resistance to L. monocytogenes infection (60% survival), which almost phenocopied the resistance seen in Ifnar1 −/− mice (80% survival) (152). Additionally, C57BL/6ByJ mice, which have a polymorphism in Irf3 causing inefficient splicing of its mRNA, demonstrated reduced IFNb induction and increased resistance to Mtb infection (186). These observations may suggest an important role for IFNb in susceptibility to L. monocytogenes infection. However, these studies did not assess IFNa induction, and characterization of Ifnb −/− mice is needed to confirm this hypothesis. Mechanistically, loss of type I IFN attenuated Listeria-induced cell death in myeloid cells and lymphocytes in vivo and ex vivo (150,152,187,188). Antigenstimulated T cells were more sensitive to lysteriolysin O (LLO)induced apoptosis after exposure to IFNa compared to cells only treated with LLO (150). Thus, a role for IFNa subtypes should not be discounted. Altogether, it is impossible to draw firm conclusions about the roles of individual type I IFNs in L. monocytogenes infection with the currently available information. Studies that specifically block IFNa or IFNb in Listeria infection might yield important insight into the functions of type I IFN subtypes.

Streptococci Species
Streptococci species often colonize mucosal surfaces and skin of healthy individuals without causing disease, but they can cause a variety of serious diseases in immunocompromised individuals or newborns (189). Type I IFNs appear to play a protective role during infection with a variety of Streptococci species (146)(147)(148)(149). S. pneumoniae, an alpha-hemolytic species commonly known as pneumococcus, is an opportunistic pathogen that colonizes the mucosal surfaces of the upper respiratory tract and is a leading bacterial cause of otitis media, pneumonia, sepsis, and meningitis (190). Type I IFNs play a beneficial role during pneumococcal infection, though the route of infection matters (147,148). Loss of IFNAR1 increased lung permeability by decreasing tight junction protein expression, which is consistent with increased bacterial titer in the blood if S. pneumoniae was inoculated via an i.n. route but not via an intraperitoneal (i.p.) route (148). IFNb played a role in mediating these protective effects because pre-treatment of mice with recombinant IFNb i.n. significantly increased survival following S. pneumoniae challenge and decreased blood bacterial titer. However, IFNa subtypes likely provide beneficial effects as well since a separate study showed that prophylactic or therapeutic administration (i.n.) of an adenoviral vector expressing IFNa enhanced survival after pneumococcal infection and decreased lung and spleen bacterial burden (149). It is unclear which IFNa subtype was used in this study, so more work is needed to determine if some IFNa subtypes are more potent than others.
A protective role of type I IFNs was also demonstrated in infection with the beta-hemolytic species S. pyogenes (group A streptococcus, GAS) and S. agalactiae (group B streptococcus, GBS) (146,147). In GBS i.v. challenge, IFNb transcript was more robustly induced in the spleen compared to IFNa4, and Ifnb −/− mice demonstrated increased lethality compared to WT mice (147). Additionally, in vitro GBS infection poorly activated peritoneal macrophages from Ifnar1 −/− or Ifnb −/− mice compared to WT controls, suggesting that IFNb may function to augment macrophage antibacterial properties. However, carefully controlled experiments need to be performed in order to determine if IFNb is directly modulating macrophage activation or if IFNb acts indirectly by influencing bacterial loads. The role of specific subtypes was not evaluated in GAS infection; however, macrophages and DCs were found to induce IFNb downstream of unique pathways. Macrophages required IRF3, STING, TBK1, MyD88, and stimulation with streptococcal DNA, whereas DCs depended on MyD88, IRF5, and streptococcal RNA (146). It might be interesting to evaluate Irf3 −/− , Irf5 −/− , and Ifnb −/− mice in S. pyogenes infection to determine if the cellular source of IFN affects pathogenesis. Additionally, better characterization of the IFNa subtypes induced and their role in GAS and GBS is needed.

Remarks on Bacterial Infections
Similar to viral infections, type I IFNs can be either detrimental or beneficial to the host during bacterial infections, depending on the specific pathogen. The mechanisms underlying these divergent outcomes share many features with viral infections.
The ability of type I IFNs to regulate cell death, suppress protective IFNg responses, and/or upregulate IL-10 can account for the detrimental functions of type I IFNs during Mtb, Salmonella, and L. monocytogenes infection. These activities are reminiscent of the type I IFN-driven increases in IL-10 and PD-L1 observed in LCMV, as well as the increased cell death observed in acute influenza infection (132,135). Even though a detrimental role for type I IFNs is well documented in Mtb infection, in special contexts type I IFNs may be able to serve a protective function. Of particular interest is the possibility of type I IFN serving as an adjuvant with certain, less virulent Mycobacterium vaccination strains. As is the case with some viral infections, the timing, magnitude, and cellular source of type I IFNs underlie these distinct outcomes. In the future it will be interesting to explore if these divergent phenomena are also due to differential induction or functions of type I IFN subtypes.
There are also examples of type I IFNs having a protective role in bacterial infections, such as with several Streptococcus species. This net beneficial effect may reflect many of the functions commonly observed in viral infections, such as coordinating protective immune cell recruitment and activation and promoting the right level of inflammation needed to clear the bacterial infection. The exact mechanisms underlying these protective effects are understood at a very general level and questions remain. Which cells do IFNs signal on to mediate these protective effects? What ISGs are responsible for mediating protection, and are they different from those acting in viral infections? Importantly, do specific type I IFN subtypes drive particular protective functions? We are only beginning to grasp how type I IFNs contribute to protective antibacterial immune responses, and there are many interesting avenues to explore relevant to human health.

Parasitic Infections
Parasites include single-cellular protozoa (e.g. Plasmodium and Leishmania species) and multicellular helminths, which include flatworms (e.g. Schistosoma species) and roundworms (e.g. Ascaris species) (191)(192)(193)(194). Previously, parasite-host interaction studies have not investigated the functions of type I IFNs, but recent studies in malaria have identified both protective and pathogenic properties of IFNa/b [reviewed in references (195,196)]. Below we explore the roles of IFNa and IFNb during Plasmodium infection, the causative agent of malaria ( Table 3).

Plasmodium Overview
Malaria initially presents as a wide variety of symptoms, including periodic fever, chills, headache, malaise, and muscle and joint aches, but as disease progresses severe anemia, blood acidosis, splenomegaly, acute respiratory distress syndrome, and spread to the brain are possible, which can be fatal (210). Infected mosquitoes transmit Plasmodium sporozoites to humans during a blood meal. The sporozoites initially infect hepatocytes, where they replicate as merozoites (liver stage), and eventually, merozoites enter the blood stream to infect red blood cells, where they begin asexual reproduction (blood stage) (191). Symptoms in humans usually begin developing several days after release of parasites into the blood. P. falciparum and P. vivax are the most common species responsible for malaria disease in humans, and several Plasmodium species (P. berghei, P. yoelii, P. chabaudi, and P. vinckei) infect rodents and recapitulate various stages of human disease (210).

Liver-Stage Malaria
Two important studies recently revealed a protective role for type I IFNs in controlling liver-stage Plasmodium infection. First, Liehl and colleagues showed that all of the early upregulated genes in the liver from mice infected with P. berghei (ANKA) were classified as IFN-stimulated genes or linked to the type I IFN signaling pathway (197). Similarly, Miller et al. also uncovered an early type I IFN signature in the liver of mice infected with P. yoelii (Py17XNL) (198). Upon global IFNAR1 deficiency or conditional deletion of IFNAR1 on hepatocytes (Albumin-Cre), mice failed to control parasite replication in the liver (197,198). These studies suggest that type I IFN signaling protects against malaria infection by controlling early parasite replication in the liver. Further characterization revealed that Irf3 −/− mice, but not Irf7 −/− mice, showed a similar early increase in liver parasite burden as Ifnar1 −/− mice following P. yoelii (Py17XNL) infection (198). This is consistent with the observation that Irf3 −/− mice demonstrated a more severe decrease in early liver ISG induction compared to Irf7 −/− mice following P. berghei (ANKA) infection (197). Given that IRF3 is a key regulator of IFNb induction, these findings could suggest that endogenous IFNb is more important than IFNa subtypes for controlling parasite burden in liver stage malaria. Additional studies are needed to confirm this hypothesis.
decreased early parasitemia, and the authors proposed that this was due to IFNa-dependent inhibition of reticulocyte (immature red blood cell) development, as opposed to direct antiplasmodium effects (202,211). Moreover, deletion of inflammasome components or some intracellular PRR sensing components decreased parasitemia and increased resistance to lethal P. yoelii infection through alleviation of SOCS1-mediated suppression of type I IFN responses (201,212).
Other studies have demonstrated that type I IFNs might play a detrimental role during blood-stage malaria. First, a group showed that Ifnar1 −/− and Irf7 −/− mice better controlled parasitemia in non-lethal P. chabaudi infection compared to WT controls (199). Additionally, Sebina and colleagues showed that IFNAR1 deletion in P. yoelii (Py17XNL) infection increased pathogen-specific antibody titers and decreased parasitemia late in infection (17-21 dpi) (200). Mechanistically, type I IFN signaled on DCs to limit their activation of T follicular helper cells in an inducible T cell co-stimulator (ICOS) signalingdependent manner, and this interaction ultimately influenced downstream germinal center B cell responses (200). However, it should be noted that IFNAR1 deletion in the Sebina et al. study also trended toward increased parasitemia early in infection (6-11 dpi), suggesting that these findings are not completely incongruous with the studies that found a protective role for type I IFNs. Altogether, type I IFNs might be detrimental in the blood stage malaria by impeding humoral immunity later in infection, but the Plasmodium strain and timing of IFN action may influence the overall effect of type I IFNs on disease outcome. It would be interesting to determine if this effect is dependent on certain type I IFN subtypes.

Cerebral Malaria
Similar to the blood stage, the role of type I IFNs during cerebral malaria remains controversial. Several independent groups have demonstrated that Ifnar1 −/− mice are either completely or partially protected from lethal experimental cerebral malaria (P. berghei ANKA sporozoite infection), demonstrating a net pathogenic effect for type I IFNs in this context (199,(203)(204)(205)(206)(207). Loss of type I IFN signaling may increase IFNg-producing CD4 + T cells, reduce pathogenic CD8 + T cell recruitment and/or activation in the brain, improve DC priming of CD4 + T cell responses, or some combination thereof (199,(204)(205)(206)(207). Irf7 −/− mice only partially recapitulated the decreased brain pathology and protection from P. berghei (ANKA) lethality observed in Ifnar1 −/− mice, but loss of IRF7 perfectly phenocopied the decreased parasitemia observed in Ifnar1 −/− mice (199). These findings may suggest IFNas are more important in promoting parasitemia, whereas IFNb and IFNa might both contribute to brain pathology, but specific antibody blockade of type I IFN subtypes would confirm this hypothesis.
Paradoxically, a few groups have shown that systemically administering either recombinant IFNb or hybrid IFNa1/a8 concurrently with infection alleviated cerebral malaria (P. berghei ANKA) (208,209). Both IFN treatments reduced parasite burden in the brain and decreased infiltrating CD8 + T cells in the brain compared to control mice, but only IFNa1/ a8 treatment decreased blood parasitemia (208,209). A more recent study identified receptor transporter protein 4 (RTP4) as a positive regulator of type I IFN responses, and Rtp4 −/− mice were completely protected from P. berghei (ANKA) lethality and brain pathology (213). This protection in Rtp4 −/− mice correlated with increased type I IFN responses in microglia isolated from the brain, suggesting a protective role for IFNs, but blockade of type I IFN signaling in Rtp4 −/− mice is needed to confirm a causal link (213). Overall, an issue of magnitude and timing of IFN response might underlie these apparent discrepancies with the protective phenotypes of Ifnar1 −/− mice (discussed below). Indeed, antibody blockade of IFNAR1 as late as 5 dpi was almost as protective as Ifnar1 −/− mice, suggesting that the detrimental effects of type I IFNs occurred during priming of adaptive immune responses (199).

Remarks on Parasitic Infections
It is clear that the role of type I IFNs in malaria is complex and depends on the stage of Plasmodium life cycle. Type I IFNs seem to play a protective role during the liver stage, but there are contradictory findings from various models of blood-stage and cerebral malaria. Perhaps infection with some strains of Plasmodium yields suboptimal type I IFN production very early in infection, ultimately leading to delayed and higher levels later in infection when parasite burden is not effectively controlled. Proper intervention at either step would benefit the host, and this could explain why loss of IFN signaling or exogenous IFN treatment can both be protective. The contribution of individual IFN subtypes remains unclear, though divergent phenotypes in Irf3 −/− and Irf7 −/− mice suggest this could be an interesting question to explore. Importantly, genetic variants in IFNAR1 have been associated with either greater or lower risk of severe malaria disease (205,(214)(215)(216)(217). The impact of each genetic variant on IFNAR1 expression and function still need to be determined, but these findings suggest that type I IFNs are important regulators of malaria disease in humans.
Overall, parasitic pathogens are biologically very diverse, so data from other parasitic infection models are needed to begin drawing broad conclusions. A recent study demonstrated that the TLR4-IRF1-IFNb axis played a protective role in mice infected with Leishmania infantum by dampening proinflammatory pathways and IFNg production by CD4 + T cells (218). RNA sequencing analysis of human samples revealed that upregulation of TLR4 and type I IFN pathways was associated with asymptomatic individuals compared to patients with visceral leishmaniasis (218). Another group found that Ifnar1 −/− mice were more susceptible to Toxoplasma gondii infection (219). It would be interesting to know if IFNs are generally more important in single-cellular parasitic infections. That said, the multicellular helminth Schistosoma mansoni can induce a systemic type I IFN signature in mice and activate TLR3 in DCs in vitro, suggesting that a role for type I IFNs in parasitic worm infections is certainly possible (220,221). Continued work to delineate the cellular sources and functions of type I IFNs in malaria and other Frontiers in Immunology | www.frontiersin.org December 2020 | Volume 11 | Article 606874 parasitic diseases may reveal novel opportunities for therapeutic intervention and help uncover novel functions of type I IFNs.

CANCER
The majority of reports from animal models and the clinic demonstrate that type I IFNs play an important protective role in enhancing anti-tumor immune responses and restricting tumor growth [reviewed in (222,223)]. However, similar to persistent viral infections, the functions of type I IFNs in cancer can change throughout disease course, and there is evidence that, in certain contexts, IFN might act as a barrier to efficacious checkpoint-blockade therapy [reviewed in (224)]. Below we discuss the actions of endogenous IFNa/b and IFN-based therapies in animal models and clinical studies ( Figure 2).

Animal Studies: Endogenous Type I IFN Activity
An early study showed that mice transplanted with human tumors and treated with neutralizing antibodies to type I IFNs demonstrated exacerbated tumor growth and metastasis compared to controls, suggesting a protective role for endogenous type I IFN activity (225). Since this finding, we now know that endogenous type I IFN can mediate tumor rejection through signaling on immune cells or tumor cells.
A seminal paper showed that type I IFN signaling on host hematopoietic cells was crucial for development of anti-tumor immune responses (226). Using conditional IFNAR1 deletion, bone marrow chimeras, and adoptive transfer experiments, a number of studies have shown that type I IFN signaling on several types of immune cells is important for immunity in FIGURE 2 | Summary of the Properties of IFNa and IFNb in cancer and autoimmunity. Type I IFNs display both unique and overlapping properties in various disease states. In cancer, depending on the tumor and degree of metastases, both IFNa and IFNb can contribute to tumor rejection by directly limiting tumor cell proliferation (depicted) but also through modulation of antitumor immune responses (not depicted). In certain cases, type I IFNs can induce PD-L1 expression on tumor cells, suppressing immune-mediated killing of the tumor. The factors that cause type I IFNs to exert detrimental effects remain poorly understood. In T1D, there is evidence that IFNa subtypes play an important role in pathogenesis. Forced expression of IFNa by pancreatic b-cells accelerated the onset and severity of T1D in a mouse model, and patients receiving IFNa therapy for treatment of other diseases have a higher incidence of T1D. Similarly, immune complex-driven activation of pDCs induces robust IFNa production, which may participate in initiation of SLE. Finally, IFNb-derived therapeutics have well-established efficacy for treating MS patients. Though still largely debated, the mechanism of protection mediated by IFNb is complex and possibly includes limiting cytokine production from pathogenic CD4 + T cells and augmenting IL-10 production in a number of cell types. b-cell, pancreatic b-cell; DC, dendritic cell; IL, interleukin; Mj, macrophage; MS, multiple sclerosis; pDC, plasmacytoid dendritic cell; Rx, prescription drug; SLE, systemic lupus erythematosus; T1D, type I diabetes. cancer. For instance, type I IFN signaling on DCs, but not granulocytes or macrophages, was required for rejection of highly immunogenic tumors (227). Additionally, Itgax-Cre + Ifnar1 fl/fl (CD11c-Cre) mice showed diminished cross presentation by DCs to CD8 + T cells, which likely contributed to their failed tumor rejection (227,228). In an NK cell sensitive tumor model, endogenous type I IFN was required for NK cellmediated tumor rejection and homeostasis (229).
Other studies have shown that type I IFN signaling on tumor stromal cells may be important for controlling tumor burden. In vivo, both IFNa and IFNb have antiangiogenic activity via signaling on vascular endothelial cells to downregulate growth factors such as vascular endothelial growth factor (230,231). Stromal cells such as mesenchymal stem cells may play a role in controlling tumor growth by producing IFNa in order to enhance NK and CD8 + T cell responses (232). However, extended low level IFN signaling on tumor cells may render them resistant to apoptosis and immune-mediated killing (233,234). These differences highlight the complexities of type I IFN actions and the need to delineate cell-type specific IFN signaling and consequent gene regulation.
Limited studies have directly compared the endogenous functions of individual IFNa/b subtypes in cancer models, but there have been a few studies conducted with IFNb-deficient mice. Ifnb −/− mice showed expedited tumor growth, enhanced angiogenesis, and increased neutrophil infiltration to the tumor compared to WT mice (235)(236)(237)(238). These findings demonstrate that endogenous IFNb is important for the host anti-tumor response, but the specific signaling pathways downstream of IFNb and cell types mediating these effects remain unclear. The direct contributions of endogenous IFNa remain uninvestigated, so much work is needed to fully characterize the contribution of endogenous IFN in tumor rejection.

Animal Studies: Type I IFN-Based Therapies
The possibility that IFNs might be therapeutically useful in cancer was first shown in the early 1970s, when crude preparations of were administered to mice with syngeneic tumors increased their survival compared to untreated mice (239,240). IFN therapies have been quite effective against hematological cancers, including hairy cell leukemia and chronic myelogenous leukemia but vary in efficacy against solid tumors, such as melanoma [reviewed in (222,223,241,242)]. Below we discuss various therapeutic strategies derived from either IFNa or IFNb subtypes. Collectively, these studies show that IFNa and IFNb are able to promote a similar range of immunomodulatory and antitumor effects, so studies that directly compare the activities of IFNas and IFNb are needed to discern if there are bona fide differential properties.

IFNa-Based Therapies
Derivatives of IFNa2b have long been used in the clinic, but toxicity issues are associated with systemic administration and persistent use. Consequently, many groups have sought ways to increase IFNa expression with more precision. An influential study developed RNA-lipoplexes encoding neoantigens or endogenous self-antigens, which yielded rapid and robust IFNa production by macrophages and DCs (IFNb induction was not determined) (243). Importantly, these RNA-lipoplex vaccines were able to mediate rejection of several different types of aggressive tumors in mice (243). Another group developed a method to genetically modify human hematopoietic stem cells (HSCs) to express HuIFNa2b, but only in differentiated monocytes (244). The engineered HSCs were able to repopulate immunodeficient mice and effectively inhibit tumor progression in a murine breast cancer model (244). AcTakines (Activity-on-Target), which are optimized cytokines that only act on cells for which they are targeted, represent another interesting alternative to traditional IFN therapies. Indeed, CD20-targeted IFNa2b-derived AcTaferon reduced lymphoma and melanoma tumors engineered to express CD20 (245,246). Increasing tumor cell production of IFNa is another approach, and a very recent study demonstrated that IFNa subtypes are not all equal in their antitumor properties. B16 melanoma cells were engineered to overexpress IFNa2, a4, a5, a6, or a9, but only IFNa2-and a9expressing tumors were effectively controlled in an adaptiveimmunity dependent manner (247). Other studies have used a variety of genetic engineering methods to augment IFNa production in the tumor microenvironment and improve antitumor immunity (248)(249)(250)(251).

IFNb-Based Therapies
Derivatives of IFNa2 have been the focus of most IFN-based therapies, but several studies have explored the effect of IFNb during various models of cancer. IFNb treatment of transformed human mammary epithelial cells in vitro led to a less aggressive state (252). Another group showed that treating mice with an anti-tumor antibody fused to IFNb increased clearance of antibody-resistant tumor cells by increasing cross presentation by tumor-infiltrating DCs and activation of CD8 + T cells (253). Unfortunately, this treatment also upregulated the inhibitory molecule PD-L1 on tumor cells, but this negative effect was overcome with co-administration of anti-PD-L1 antibody (253). Another group transduced induced pluripotent stem cell (iPSC)-derived myeloid cells with an IFNb-encoding lentivirus to treat disseminated gastric cancer (254). When injected into immunocompromised mice, the modified myeloid cells accumulated in the tumors and inhibited growth of the peritoneally disseminated cancer (254). Lastly, intratumoral injection of an mRNA encoding a fusion protein consisting of IFNb and the ectodomain of transforming growth factor-b receptor II enhanced DC activation of CD8 + T cells in vitro and promoted rejection of the TC-1 tumor cell line in vivo (255).

Human Studies
The antitumor and immunomodulatory effects of IFNa therapy have been demonstrated in the treatment of a variety of cancers, and here we present a few representatives. IFNa-derived  (241)]. This is because there is evidence that IFNa therapy is able to target and sensitize the rare CML stem cell population to subsequent killing by chemotherapy, whereas imatinib is more effective against more differentiated CML progenitors (256,257). Lastly, an analysis of matched primary breast cancer tumors and bone metastases revealed that primary tumor cells expressed IRF7, whereas metastases consistently demonstrated downregulation of IRF7 expression (258). This may suggest that IRF7-mediated IFNa production in primary tumors is an important factor for limiting metastases, but further studies are needed to determine if this is an IFNaspecific effect or if there is also a role for IFNb. Fewer clinical studies have been conducted with IFNb-derived therapies, but there is evidence that IFNb also plays a protective role in tumor rejection. Increased IFNb mRNA expression significantly correlated with improved survival in patients with triplenegative breast cancer, though the mechanism is undetermined (252). In vitro studies have shown that IFNb is more potent in inducing apoptosis in melanoma cells compared to IFNa (259). The relevance of this differential potency has yet to be extensively explored in vivo.

Detrimental Effects of Type I IFNs in Cancer
Despite all the evidence that type I IFNs can facilitate protective antitumor immune responses, IFNs can also impede cancer therapies. We provide just a few mechanistic examples. Persistent type II IFN signaling on tumors can result in PD-L1-dependent and PD-L1-independent resistance to immune checkpoint blockade, and the authors identified a role for type I IFNs in maintaining PD-L1-independent resistance (233). Radiation and chemotherapy stimulate immune-mediated destruction of tumor cells partly through induction of type I IFNs (260)(261)(262)(263)(264). However, recent work showed that conditional deletion of IFNAR1 on tumor cells enhanced responsiveness to radiation therapy through increased susceptibility to CD8 + T cell-mediated killing (265). Lastly, oncolytic viruses can preferentially kill cancer cells, but tumor responsiveness to type I IFN activity confers resistance to this therapeutic method. One study showed that IFNa and IFNb differ in their ability to confer resistance to oncolytic virus treatment in vitro. Exogenous IFNb more effectively prevented oncolysis of human head and neck squamous cell carcinoma cells by vesicular stomatitis virus compared to IFNa, but differential effects were not observed for normal keratinocytes or endothelial cells (266).

Remarks on Cancer Studies
Collectively, this large body of cancer studies has shown that the roles of type I IFNs are complex and likely context specific. The extensive use of IFNa-derived therapies to treat a number of cancers in the clinic has greatly increased our understanding of the range of IFNa properties in vivo. Cancer models are uniquely advantageous for studying protective immunomodulatory effects of IFNs compared to infection models because pathogen load is not a confounding factor. Despite the large body of work suggesting the benefits of type I IFN signaling in cancer, the actions of specific IFN subtypes, for the most part, remain undefined. The beneficial effects of indirect activators of type I IFNs, such as the RNA-lipoplexes (discussed above) or STING agonists, may be due to their ability to induce multiple IFN subtypes with either overlapping or unique functions (222,244). The heterogeneity of cancer makes it all the more important to appropriately stratify patients to ensure a beneficial effect of treatment.

AUTOIMMUNITY
Type I IFNs have emerged as critical mediators of autoimmunity.
Patients with a variety of autoimmune diseases display serum type I IFN signatures, and IFN treatments for other diseases have correlated with the development of autoimmunity. These observations have led to the assumption that type I IFNs may contribute to autoimmunity pathogenesis. However, IFNbderived therapeutics have been used to treat multiple sclerosis, highlighting that caution is warranted in attempting to summarize the mechanisms of autoimmune disorders. Below we outline the current understanding of the roles of IFNa and IFNb during systemic lupus erythematosus, type 1 diabetes, and multiple sclerosis ( Figure 2). This is not an exhaustive analysis of autoimmune disorders, and active research is exploring the function of type I IFNs in other disorders, such as rheumatoid arthritis and Sjögren's syndrome (267,268).

Systemic Lupus Erythematosus
Systemic lupus erythematosus (SLE) is an autoimmune disease that affects organs such as the skin, joints, kidneys, and CNS (269). A type I IFN gene signature in the blood of SLE patients is well established (270)(271)(272). Additionally, a number of genetic risk factors for SLE are associated with type I IFN production or signaling, including IRF5, IRF7, IRAK1, and TYK2 [reviewed in reference (273) (276). They also found that type I IFN protected T follicular helper cells from NK cell-mediated death, thus further promoting B cell responses (276). Other studies suggest that IFNa may further drive SLE by increasing production of multiple TNF family members, such as BAFF and APRIL, which promote B cell survival and can drive SLE pathogenesis (279)(280)(281). Due to the apparent pathogenic role of IFNa during SLE, attempts have been made to neutralize type I IFNs in SLE patients (282)(283)(284)(285)(286)(287). Both anti-IFNa and, more recently, anti-IFNAR1 therapies have been tested (282)(283)(284)(285)(286)(287). Both treatment strategies showed disparate efficacy in patients, so further work is needed to clarify if this type of therapeutic intervention would be beneficial for patients. It might be that IFNa only plays a key role in the initiation and early stages of disease, so the disease stage may be important in stratifying patients [reviewed in reference (288)].

Type 1 Diabetes
Type 1 diabetes (T1D) is a chronic, autoimmune disease caused by the immune-mediated destruction of pancreatic b-cells that leads to insulin deficiency and hyperglycemia (289). A blood type I IFN signature in T1D patients precedes the development of autoantibodies and disease (290)(291)(292)(293). One study detected a significant increase in expression of IFNa subtypes, but not IFNb, in postmortem pancreas specimens from T1D patients compared to control subjects (290). Moreover, many genetic polymorphisms associated with T1D are involved in the type I IFN response such as MDA5 and TYK2 (294)(295)(296). Altogether, these findings suggest a detrimental role for type I IFNs in T1D. A role for type I IFNs in the development of T1D is supported in animal models. An early study showed that forced constitutive IFNa expression by pancreatic b-cells in mice resulted in hypoinsulinemic diabetes and pancreatic inflammation (297). Additionally, non-obese diabetic (NOD) mice, a common model for T1D, showed elevated IFNinducible transcripts in the pancreatic islets prior to disease onset, and treatment of young NOD mice with anti-IFNAR1 mAb delayed the onset and decreased the occurrence of T1D (298,299). Collectively, these findings suggest that type I IFN signaling, especially in the pancreas, may play a key role in initiating T1D.
LCMV can be employed as a viral model of T1D, in which mice transgenically express LCMV glycoprotein (GP) under the control of the rat insulin promoter (Rip-LCMV) (300). Development of Rip-LCMV T1D is dependent on type I IFN (301,302). Recent work showed that anti-IFNAR1 mAb treatment reduced blood glucose to normal levels and prevented destruction of pancreatic islets (302). Importantly, they also showed that pan-IFNa (a1, a4, a5, a11, and a13) mAb blockade, but not IFNb blockade, was able to recapitulate the anti-IFNAR1 phenotype, demonstrating a distinct role for IFNa subtypes in promoting pathogenesis in the Rip-LCMV T1D model. A similar detrimental role for IFNa is suggested in human disease. IFNa therapy for HCV in individuals genetically predisposed to T1D induced or exacerbated the development of T1D (303). Moreover, a recent study showed that a subset of AIRE-deficient patients who developed autoantibodies specific for IFNa, especially IFN-a1/13, IFN-a5, and IFN-a14, were less likely to develop T1D, whereas patients who failed to generate these antibodies developed T1D (304). Altogether, animal and human studies suggest a detrimental role of type I IFNs in T1D, and IFNa subtypes appear to play a dominant role in disease development and pathogenesis.

Multiple Sclerosis
Multiple sclerosis (MS) is a chronic, autoimmune disease of the CNS in which immune cells target and destroy the myelin sheath surrounding neurons, leading to neurodegeneration (305). Similar to other autoimmune conditions, MS patients can show a serum type I IFN signature, but this signature is relatively low when compared to SLE patients (306,307). However, in strong contrast to SLE and T1D, type I IFNs, do not appear to play a detrimental role. In fact, IFNb was the first FDA-approved therapy for MS (308)(309)(310)(311). However, due to its flu-like side effects and the availability of more effective treatments, it is no longer the preferred therapy for MS patients (312). Even though IFNb treatment is currently less preferred in clinical use, animal models and clinical studies (discussed below) have revealed important insight into the properties of IFNb in vivo.

In Vitro and Animal Studies
Experimental autoimmune encephalomyelitis (EAE), a mouse model of MS, has provided mechanistic insight into the protective actions of IFNb (313). Mice lacking IFNb, IFNAR1, or IRF7 showed exacerbated clinical EAE compared to WT mice, perhaps due to greater T cell infiltration and increased proinflammatory cytokine production in the CNS (314)(315)(316). Unexpectedly, mice that lack IRF3 showed significantly lessened clinical disease compared to WT mice, and this seemed to be due to a cell-intrinsic defect in the development of T helper type 17 (T H 17) cells (317). Indeed, T H 17 versus T H 1 skewing can drastically influence the impact of IFNb treatment in EAE. IFNb treatment was effective in reducing EAE severity in T H 1induced EAE but worsened disease in T H 17-induced EAE (318). Thus, depending on the skewing of the T helper responses and method of induction of EAE, IFNb may be protective or pathogenic.
Many cell types respond to IFNb therapy in EAE. Deletion of Ifnar1 on myeloid cells including macrophages, monocytes, granulocytes, and microglia, but not neuroectodermal cells, resulted in increased severity of EAE symptoms, suggesting that IFNb mediates its protective effects, in part, by acting on these cells (315). Mice treated with TLR3 or TLR7 agonists display reduced disease severity associated with increased type I IFN production by pDCs and other antigen presenting cells  (319,320). Other reports have also suggested that IFNb signaling on T cells curbs their pathogenicity (321,322). Furthermore, type I IFN signaling on conventional DCs limited their migration to the CNS and prevented their activation of T H 17 cells during EAE (323,324). The tissue resident antigen presenting cells in the CNS, microglia, may also play a role in the type I IFN response during EAE. Type I IFN signaling on microglia promoted clearance of myelin debris by increasing their phagocytic activity (325,326). Finally, a study identified a role for type I IFN signaling on astrocytes to suppress CNS inflammation during EAE (327). Clearly IFNb is able to induce protective effects during EAE, and a recent report demonstrated that sustained low-dose IFNa1 delivery via an adeno-associated viral system prevented the onset of disease in EAE (328). This therapeutic effect was associated with regulatory T cell expansion, and myelin-specific effector T cells displayed reduced proliferative capacity, decreased proinflammatory cytokine production, and increased expression of IL-10 and PD-1 (programmed cell death protein 1) (328). Another study showed that a systemic high dose of MuIFNa11 was able to initially delay EAE in mice but ultimately caused significant toxicity and mortality; however, when IFNa activity was targeted to DCs (Clec9A-targeted AcTaferon), they found efficient protection from EAE (329). These findings suggest that IFNb might not be unique in its ability to confer protection in EAE, but more work is needed to determine what factors cause IFNa treatments to yield detrimental effects or protective effects.

Human Studies
IFNb was the first FDA-approved therapy for MS (308)(309)(310)(311). However, due to its flu-like side effects and the availability of more effective treatments, it is no longer the preferred therapy for MS patients (312). Observations from patients suggest that IFNb therapy likely acts through multiple mechanisms, such as influencing immune cell recruitment and activation. First, IFNb treatment correlated with decreased new brain lesions and increased soluble VCAM-1 in patient serum, suggesting that modulating immune cell entry to the CNS is one potential mechanism of IFNb therapy (330). In addition to impacting cell recruitment, IFNb treatment may also regulate survival of immune cells since an increase in proapoptotic genes was observed in peripheral immune cells isolated from IFNbtreated patients (331,332).
Pathogenic T H 1 and/or T H 17 cells likely play an important role in MS, and IFNb therapy may limit the proliferation of pathogenic T cells and modulate their cytokine production (332,333). IFNb therapy is likely more effective in individuals with a T H 1 driven disease, since high serum IL-17F levels correlated with a poor response to IFNb therapy (318). A number of cell types are likely involved in protective IFNb treatment. For example, IFNb treatment of MS patients can induce IL-10 production by myeloid cells, but treatment can also suppress production of granulocyte-macrophage colony-stimulating factor (GM-CSF), IFNg, and TNF by effector T cells (334)(335)(336)(337)(338)(339)(340). Additionally, in patients that responded to IFNb therapy, treatment induced T regulatory cells that produced IL-10 and expressed PD-L1 (341,342). Altogether, the protective mechanisms that underlie IFNb therapeutic effects likely involve direct or indirect actions on effecter T cells. A better understanding of these mechanisms would likely reveal important information about the functional capacity of IFNb in vivo.

Remarks on Autoimmune Studies
A large proportion of patients with SLE or T1D show a type I IFN signature in their blood, and many studies have shown that type I IFNs promote pathogenesis in these autoimmune disorders. There is strong evidence implicating the IFNa subtypes in initiation and progression of SLE and T1D, but at this time, a role for IFNb cannot be entirely ruled out-direct functional comparisons of IFNa versus IFNb would be needed to draw that conclusion. Altogether, the specific pathogenic functions of type I IFNs during autoimmune disorders are likely tissue specific. A recent study performed geneexpression profiling of structural cells from 12 different tissues and found that the responses of the cells to stimuli were tissue-specific, thus identifying the stroma as an important regulator of tissuespecific immune responses (343). While there is clear evidence that type I IFNs can modulate pathogenic autoimmune responses, it is important to know how systemic IFNa activity might promote celltype specific effects in diseased versus nondiseased tissues in disorders like T1D that target a particular tissue, but also in diseases like SLE that have multi-organ effects.
In contrast, blood from MS patients do not display as robust a type I IFN signature as SLE or T1D patients, and many studies have demonstrated that IFNb treatment has therapeutic properties in animal models of MS and in affected individuals. The protective functions of IFNb are complex and likely include modulating immune cell recruitment and activation directly through action on immune cells and indirectly through action on brain resident cells. The functions of IFNas in MS are less clear. There might be conditions, such as very low doses or when targeted to a specific cell type, in which IFNa subtypes are also protective. Careful comparison of IFNb versus IFNa dose responses in EAE might uncover novel mechanisms for differential functions among type I IFNs in vivo.

CONCLUDING REMARKS
Whether type I IFNs have a net beneficial or detrimental effect on disease outcome depends on a variety of factors including the timing and magnitude of induction relative to disease onset, the duration of expression, the specific subtypes induced, the cell types responding, and likely other factors. Progress is needed in understanding the spatiotemporal induction of the various type I IFN subtypes in vivo, as well as the cell types responsible for type I IFN production. A lack of tools to differentiate between different subtypes has hindered progress in this area. Quantitative reverse transcription polymerase chain reaction has been a useful technique for quantifying specific IFN subtypes, and singlemolecule array (Simoa) digital ELISA technology was demonstrated to detect IFN in blood with high sensitivity (344). However, there is a need for licensed antibodies against individual subtypes that are able to neutralize in animal models and reliably stain tissue sections to more accurately determine the timing of expression at the tissue level. Transcriptomic approaches have successfully differentiated type I and type III ISG signatures in organoid cultures (345). Because the effects of type I IFN are pleiotropic, there is a need to delineate the ISGs responsible for the protective and pathogenic functions of type I IFN subtypes in a given context and to understand how cell-type specificity might affect expression of those genes. A recent report profiled gene-expression networks of fibroblasts, endothelial, and epithelial cells isolated from multiple tissues and revealed tissuespecific signaling networks (343). A similar approach or spatial transcriptomics, which yields gene expression profiles in intact tissue sections, would be powerful tools to unravel the cell typespecific responses to different type I IFN subtypes in vivo (346).
Lastly, given that many type I IFN subtypes have expanded independently after mammalian speciation, there is a great need for tools to allow the study of human type I IFN subtypes in animal models. Immune-humanized mice and hybrid IFNAR (HyBNAR) mice, which transgenically encode variants of IFNAR1/2 that contain the human extracellular domains fused to the transmembrane and cytoplasmic segments of murine IFNAR, have both been used to study HuIFN in mice (347). These two systems are helpful in contexts where immune cells are the predominant sources of and responders to type I IFN or in studies administering exogenous HuIFN, but they do not permit loss-of-function studies, exclude the impact of endogenous IFN expression by stromal cells, and IFNAR1/2 transgenes are likely more highly expressed than endogenous IFNAR1/2. Overall, a concerted effort to address this lack of tools will go a long way toward increasing our ability to directly compare the expression and functions of distinct type I IFN subtypes, which will undoubtedly generate new strategies to augment or dampen the type I IFN pathway for biomedical purposes.

AUTHOR CONTRIBUTIONS
LEF and MCL conceptualized and drafted the manuscript, as well as created the figures and tables. LEF, MCL, and DJL all reviewed and edited the manuscript. All authors contributed to the article and approved the submitted version.