H.B. Greenberg Departments of Microbiology and Immunology, Department of Medicine, Stanford University, School of Medicine, Stanford VA Palo Alto Health Care System, Palo Alto, CA, United States
Hosts have evolved highly sophisticated mechanisms to prevent and control viral infections. These include the innate immune response, hardwired into most eukaryotic cells and triggered soon after virus entry, as well as the more specialized adaptive immune response, defined by aspects of antigen specificity and memory. Following host cell infection, most viruses trigger one or more pattern recognition receptors, which have evolved to recognize virus-specific signatures, or pathogen associated molecular patterns (eg, 5′-triphosphate RNA), and upon ligand binding, trigger conserved signaling pathways that culminate in the functional activation of critical host transcription factors involved in the initiation of various antiviral responses. Several virus stress-induced genes (vSIGs) are transcribed as a result of this early activation, including those encoding the secretory type I and III interferons (IFNs). Expression and secretion of IFNs result in their binding to cognate surface receptors on cells in both autocrine and paracrine patterns, followed by ligand-stimulated activation of the JAK-STAT signaling cascade. Activated STATs, together with other accessory factors, then translocate to the nucleus and initiate a second wave of transcription—resulting in the expression of hundreds of genes that encode antiviral proteins targeting multiple aspects of viral replication, assembly, maturation, and spread. This second phase of the IFN response is also crucial for ensuring that initial IFN expression is amplified through positive feedback, thus ensuring robust establishment of an antiviral state in different host cell types, both infected and uninfected. Coevolution of hosts and their viruses [also known as the Red Queen Hypothesis (Muraille, 2013)] likely resulted not only in the present complexity of the host innate response, but also in the emergence of viral strategies to block the innate response at multiple steps in order to ensure evolutionary advantage (Medzhitov and Janeway, 1997). Accordingly, most pathogenic viruses, including the rotaviruses, encode factors that target redundant steps of the IFN induction and amplification signaling pathways.
pathogen associated molecular patterns
pattern recognition receptors
virus stress-induced genes
interferon regulatory factors
single cell analysis
Hosts have evolved highly sophisticated mechanisms to prevent and control viral infections. These include the innate immune response, hardwired into most eukaryotic cells and triggered soon after virus entry as well as the more specialized adaptive immune response, defined by aspects of antigen specificity and memory. Following host cell infection, most viruses trigger one or more pattern recognition receptors (PRRs), which have evolved to recognize virus-specific signatures, or pathogen associated molecular patterns (PAMPs) (eg, 5′-triphosphate RNA), and upon ligand binding, trigger conserved signaling pathways that culminate in the functional activation of critical host transcription factors (TFs) involved in the initiation of various antiviral responses. Several virus stress-induced genes (vSIGs) are transcribed as a result of this early activation, including those encoding the secretory type I and III interferons (IFNs). Expression and secretion of IFNs result in their binding to cognate surface receptors on cells in both autocrine and paracrine patterns, followed by ligand-stimulated activation of the JAK-STAT signaling cascade. Activated STATs, together with other accessory factors, then translocate to the nucleus and initiate a second wave of transcription—resulting in the expression of hundreds of genes that encode antiviral proteins targeting multiple aspects of viral replication, assembly, maturation, and spread. This second phase of the IFN response is also crucial for ensuring that initial IFN expression is amplified through positive feedback, thus ensuring robust establishment of an antiviral state in different host cell types, both infected and uninfected. Coevolution of hosts and their viruses [also known as the Red Queen Hypothesis (Muraille, 2013)] likely resulted not only in the present complexity of the host innate response, but also in the emergence of viral strategies to block the innate response at multiple steps in order to ensure evolutionary advantage (Medzhitov and Janeway, 1997). Accordingly, most pathogenic viruses, including the rotaviruses (RVs), encode factors that target redundant steps of the IFN induction and amplification signaling pathways. Since many of the basic aspects of RV structure, replication, and pathophysiology are reviewed elsewhere in this book (Chapters 2.1–2.4 and 2.6), they will not be covered here. Certain facets of RV biology may however be specifically pertinent to the host innate immune response to viruses. Rotaviruses encapsidate an 11-segmented dsRNA genome and during replication capped mRNA is generated that serves as a template for nonconservative genome replication. Together, such RV nucleic acids represent extremely potent stimuli for components of the host PRR machinery, including RIG-I, MDA-5, and TLR3. However replication, transcription, and assembly steps in the RV life cycle (according to the current paradigm) occur within subviral particles and specialized sites of assembly within the cell (viroplasms) and likely preclude any significant exposure of the viral dsRNA to innate sensors in the cytoplasm, particularly early during infection (Arnold et al., 2013a). Interferon induction by RVs in several but not all cell types in vitro requires viral replication (Sen et al., 2009; Deal et al., 2010), and the sites of RV assembly (viroplasms) may play a protective role by sequestering not only potential viral PAMPs (Sen et al., 2011; Uzri and Greenberg, 2013), but also critical host innate response factors, such as NF-kB (Holloway et al., 2009). Although in vitro cell culture models continue to be vital in deciphering RV interactions with the innate immune response, crucial differences exist between in vitro cell culture systems and replication in vivo due to unique aspects of RV tropism and the heterogeneity of innate immune responses to RV in different intestinal cell types. Specifically, RVs predominantly infect mature villous enterocytes in the small intestine and replicate poorly, if at all, in intestinal immune cells. However, shortly after infection with RV both intestinal epithelial and immune cells mount innate responses, and recent studies (Frias et al., 2012; Sen et al., 2012; Deal et al., 2013; Feng et al., 2013; Pane et al., 2014) indicate crucial roles for both cell types in restricting the replication of certain strains of RV.
2. Homologous and heterologous rotaviruses: contrasting paradigms of innate immune regulation
Rotavirus infection of the immunocompetent host follows an acute course and provided the host survives severe dehydration, is resolved in a matter of days. Early work established that both in vitro and in vivo, although RV is a potent inducer of IFN, the replication of several RV strains is efficient and relatively insensitive to IFN’s effects (La Bonnardiere and Laude, 1983; De Boissieu et al., 1993). The lack of a substantial antiviral effect of IFN on RV replication in vivo is reflected by its highly contagious nature; remarkably, a single rotaviral infectious particle in cell culture can constitute the minimal infectious dose in several host species (Graham et al., 1987; Burns et al., 1995). In contrast to these findings for homologous RVs (ie, RV frequently isolated from host species where they cause disease), the replication (and infectivity) of rotaviruses in a heterologous host species (eg, of simian RV or bovine RV in suckling mice) is severely restricted (Arnold et al., 2013a; Sen et al., 2012; Feng et al., 2008; Greenberg and Estes, 2009; Angel et al., 2012). This natural “species barrier,” or host range restriction (HRR), has been exploited for the development of several attenuated rotavirus vaccines and vaccine candidates, and previous studies from this laboratory established a prominent role for the RV nonstructural protein NSP1 and the host IFN response in the HRR phenomenon (Feng et al., 2008, 2013; Burns et al., 1995; Greenberg and Estes, 2009; Angel et al., 2012; Bass et al., 1992; Franco et al., 1996a,b; Rose et al., 1998; Franco and Greenberg, 1999). In the suckling mouse model of RV infection, infection with heterologous RVs (such as simian or bovine RV) results in poor viral replication, limited disease when the inoculum is not very large, and poor transmissibility. Genetic ablation of the receptor specific for type I IFN (IFNAR1−/−) results in a modest (but not statistically significant) increase in intestinal replication of RRV in IFNAR1 mice (Feng et al., 2008). Combined knockouts of the type I and II IFN receptors (IFNAGR−/−) result in significantly more prolonged and extensive intestinal replication of RRV compared to WT mice. Similar gains in replication can be seen in mice lacking STAT1, which likely disables cumulative IFN-α/β/γ- and IFN-λ-dependent responses (Sen et al., 2012; Feng et al., 2008). In addition to higher intestinal replication, RRV replicates substantially better at systemic sites in mice lacking types I and II IFN receptors, or STAT1. Notably, in the IFNAGR−/− or STAT1−/− mice RRV infection leads to the development of a severe systemic disease, which results in morbidity or mortality of nearly 85–90% of animals (Feng et al., 2008). A milder self-limiting illness was observed in IFNAR−/− mice infected with RRV, but was not seen in the IFNGR−/− mice (in contrast to IFNAGR−/− mice as noted earlier). The combined effect of IFNAR1 and IFNGR1, and the absence of systemic illness in IFNGR-deficient mice, is interesting and in line with findings that during RV infection, intestinal immune cells (which are the major IFNGR1 responders) secrete significant type I IFNs (Sen et al., 2012). Studies examining the nature of IFN receptor cross talk found that in MEFs from IFNAR1- and IFNGR1-deficient mice, STAT1-mediated responses to IFN-γ are impaired, while STAT1-responses to IFN-α in IFNGR1−/− cells are not affected (Takaoka et al., 2000). Therefore, crosstalk between the two types of IFNs appears to be unidirectional in that IFN-γ-signaling is dependent on IFN-α/β signaling, but not vice-versa.
In marked contrast to the findings with the simian RRV strain, infection of IFN-receptor/STAT1 knockout suckling mice with the homologous murine EW RV is not substantially more efficient or pathogenic from that seen in WT mice, and IFNAGR- or STAT1-deficient mice show no discernible systemic illness following murine RV infection (Sen et al., 2012; Feng et al., 2008; 2013; Greenberg and Estes, 2009). Thus, susceptibility of homologous and heterologous RV strains to the effects of IFN is markedly different in suckling mice. An important distinction was also observed in these studies between simian RRV and the heterologous bovine and porcine strains (NCDV and OSU), which failed to replicate to detectable levels even in the IFN-receptor or STAT1 deficient mice (Feng et al., 2008, 2013). The basis for these differences may be related to replication restrictions mediated by VP4-mediated viral entry, in addition to IFN-based restrictions, as discussed later. In contrast, another heterologous simian strain (SA11) was similar to RRV, and showed ∼103-fold increased intestinal replication at 5 dpi in mice that lacked IFN-α/γ responses (Feng et al., 2008). More recently, in other studies we have reproduced and extended these earlier findings on the substantial differential IFN susceptibility (>104) of murine EW and simian RRV intestinal replication, using highly quantitative (and comparable) measurement of viral loads by qRT-PCR (Sen et al., 2012; Feng et al., 2013). Intestinal RRV replication, which is quite efficient compared to several other heterologous viruses in suckling mice, was still severely restricted (more than 10,000-fold) when compared to the homologous murine EW strain. The restricted replication of RRV could be rescued by a factor of ∼103 in STAT1−/− mice. In contrast, EW RV replication did not vary substantially between WT and STAT1-deficient mice (less than 10-fold). The RRV VP4 protein, when expressed on a murine RV genetic background, only restricted viral replication by ∼1-log (Sen et al., 2012; Feng et al., 2013), whereas substitution of the murine NSP1 encoding gene with its RRV counterpart resulted in RRV-like replication restriction (Feng et al., 2013), indicating that the IFN sensitivity of RRV is not related to VP4-associated entry requirements but instead depends crucially on NSP1. Other comparisons revealed that unlike RRV VP4, the VP4 gene from the heterologous bovine UK RV strain severely restricts replication in the murine intestine, suggesting that the heterologous host may also be refractory to infection by certain RVs due to VP4-mediated inefficient viral entry (Feng et al., 2013). Thus, considerable evidence exists to support the conclusion that in the suckling mouse model, homologous murine RVs are potent inhibitors of IFN-mediated antiviral responses, including all STAT1-dependent responses (types I, II, and III IFNs). In contrast, mouse intestinal cell entry-competent heterologous RVs such as the simian RRV and SA11 strains are highly sensitive to the antireplication effects of IFN, and the roles of different IFNs (particularly of IFN-λ) in restricting their replication will need to be carefully examined in future studies. In a recently published study (Lin et al., 2016) using suckling mice deficient in IFNAR1, IFNLR, or both IFNALR we found that the absence of the IFNAR or IFNLR singly or in combination did not significantly enhance the replication of murine EW RV. Similar studies with heterologous simian RV infection reveal IFNAR1- and IFNLR-dependent viral replication, and enable a more detailed comparison of the role of these IFNRs in regulating heterologous RV replication in vivo (Lin et al., 2016).
3. Detection of rotavirus infection: host innate sensors involved in Rotavirus recognition
Virus entry and replication result in the appearance of several virus-specific PAMPs in the host intestinal epithelial cell including viral glycoprotein, genomic RNA/DNA, viral transcripts, and nucleic acid replication intermediates. Host cells express a repertoire of structurally related PRRs that enable detection of infection and signal transduction of IFN induction pathways. Eukaryotic PRRs include the cell-intrinsic RIG-I-like receptors (RLRs) and NOD-like receptors (NLRs), which are activated purely within infected cells, as well as cell-extrinsic Toll-like receptors (TLRs) whose expression on cell surfaces and vesicular membranes extends their functionality to uninfected bystander cells, such as dendritic cells and macrophages. Several studies have examined interactions between RV PAMPs and host PRRs (Sen et al., 2011; Broquet et al., 2011; Qin et al., 2011; Pott et al., 2012; Nandi et al., 2014), leading to a better understanding of this component of host innate immunity to RV.
In contrast, a cell type where RV dsRNA is a potent PAMP is the plasmacytoid dendritic cell (pDC), although the cognate PRR(s) involved is not known (Deal et al., 2010; 2013; Douagi et al., 2007; Gonzalez et al., 2010; Lopez-Guerrero et al., 2010). Highly purified human primary pDCs that are exposed to inactivated RV strongly express IFNα as early as 6 h postexposure while pDCs exposed to virus-like particles containing VP2, VP6, VP4, and VP7 but no dsRNA do not express type 1 interferons (Deal et al., 2010). Thus, early dsRNA-mediated activation of IFN may not be particularly relevant in intestinal epithelial cells supporting RV replication, but could be quite important for the IFN induction that occurs in vivo in nonreplicating compartments, such as intestinal immune cells (Sen et al., 2012), and possibly in TLR3 expressing IECs from adult mice infected with RV (Pott et al., 2012).
RV replication by-products: Early activation of PRRs in infected cells is dependent on viral replication. Studies on the bovine UK RV in murine embryonic fibroblasts (MEFs) revealed that despite encoding a full complement of both structural and nonstructural viral proteins, UK infection of MEFs resulted in robust IFNβ secretion (Sen et al., 2009). It is worth noting that in several other cell lines, such as COS7 and HT29, UK RV is very capable of efficiently inhibiting IFNβ responses (Sen et al., 2009), demonstrating that RV regulation of IFN expression is contextual rather than absolute—an important factor to consider when classifying RV IFN phenotypes in single cell types. In MEFs, early IFN induction was abrogated with UV-inactivated UK virus, indicating that a by-product(s) of replication acted as the PAMP(s) for early PRR activation (Sen et al., 2009). Subsequent biochemical and enzymatic characterization of RV RNAs from infected cells and in vitro transcription systems revealed that RV mRNA species with exposed 5′-phosphate groups and those with incompletely 5′-O-methylated “cap” structures were particularly effective at stimulating PRRs (Uzri and Greenberg, 2013) (Fig. 2.8.1). Using genetic knockout MEFs lacking different PRRs, their relative contributions to early IFN activation by RVs have been determined (Sen et al., 2011; Broquet et al., 2011). From such studies, it is likely that RV transcription triggers the mitochondrion-associated RLRs RIG-I (retinoic acid induced gene I) and MDA-5 (melanoma differentiation- associated antigen 5) early after infection. In accordance with the function of RLRs, activation of RIG-I and MDA-5 by RV is completely dependent on the mitochondrial adaptor MAVS. While absence of MAVS results in a complete loss of RV recognition (and subsequent IFN secretion later in infection) by the host cell, this defect is not fully recapitulated by the removal of either RIG-I or MDA5 alone, indicating that both PRRs function redundantly in sensing the presence of RV, perhaps by targeting 5′-phosphate moieties and incompletely 5′-O-capped ends of viral transcripts, respectively. Interestingly, the RLR LGP2 (laboratory of genetics and physiology 2), which lacks the caspase activation and recruitment (CARD) domains required for signaling to MAVS, is important for IFN induction by RV (Broquet et al., 2011), although the mechanism involved is not understood. Apart from viral nucleic acid, RV also encodes a nonstructural glycoprotein, NSP4 that is secreted into the extracellular milieu and exhibits certain adjuvant-like properties. Recent evidence indicates that NSP4 may represent a RV PAMP in macrophages and trigger pro-inflammatory cytokine expression by a pathway that utilizes the cell-extrinsic PRR TLR2 (Ge et al., 2013) (see Chapter 2.4). Regardless of their specific IFN antagonistic abilities, all RVs appear to be recognized by the host PRR machinery [eg, the following strain pairs of IFN inducers and suppressors, respectively—UK and RRV in MEFs (Sen et al., 2009), SA11-5S and SA11-4F in HT29 cells (Arnold and Patton, 2011)]. PAMP-mediated activation results in recruitment of adaptor TRAF molecules by MAVS—a crucial step in MAVS-dependent activation of both IRF3 and NF-κB (Liu et al., 2013). Interestingly, both the RV outer capsid protein VP4 and the nonstructural interferon antagonist protein NSP1 regulate TRAF signaling (Bagchi et al., 2013; LaMonica et al., 2001), although whether this leads to perturbed MAVS function is unknown. Therefore, based on the available evidence, RVs do not appear to regulate early PAMP recognition, which may be an inevitable consequence of RV infection.
Following synthesis of viral proteins during infection, the viral NSP1 protein appears to reduce the activity of two PRR activation components, RIG-I (Qin et al., 2011) and MAVS (Nandi et al., 2014) by directing their degradation. The consequence of this inhibition (occurring ∼8 h or later during infection), which is preceded by viral transcription and synthesis of NSP1, may be to dampen the magnitude of IFN induction within infected cells later in infection. NSP1 proteins from both OSU and SA11 strains have been reported to interact with RIG-I in a transient overexpression system, and this interaction does not require the NSP1 C-terminal (∼170-aa) IRF3-binding domain (Qin et al., 2011). Expression of NSP1 also led to inhibition of RIG-I mediated IFNβ expression, and a decrease in RIG-I protein expression by a proteasome-independent pathway (Qin et al., 2011). Recently it was also reported that NSP1s from porcine OSU and human Wa, DS-1, and KU RV strains are able to degrade MAVS in a proteasome-dependent manner (Nandi et al., 2014). NSP1 interacted with MAVS, and the C-terminal 395-aa alone was sufficient to mediate this interaction, although MAVS degradation required the full-length protein (Nandi et al., 2014). Although these observations are interesting, several key questions should be resolved in future studies. Whether NSP1-mediated inhibition of RIG-I occurs during RV infection is not known presently. Somewhat paradoxically, although OSU-NSP1 protein is able to degrade both RIG-I and MAVS, infection with the porcine RV OSU (and the related RV SB1A) has been shown to lead to a robust and sustained activation of IRF3, presumably directed by MAVS (Graff et al., 2009; Sen et al., 2014). Similarly, UK RV infection leads to IRF3 S396 activation during infection of 3T3 cells that persists until later times (12–16 hpi) in a MAVS-dependent manner (Sen et al., 2009, 2011). Finally, definitive evidence is lacking on whether NSP1 interacts with either RIG-I or MAVS directly, and similar to most other potential NSP1 interactions, this interaction has not been analyzed using the relevant purified binary components. NSP1 may instead target a common component of this signaling complex, leading to the secondary observed associations of NSP1 with the other members (through indirect interactions). Such indirect interactions with NSP1 are a possibility since multiple NSP1 host partners that have already been identified are a part of common signaling complexes.
4. Intermediaries: innate factors relaying interferon induction during rotavirus infection
Following pathogen detection, PRRs undergo conformational changes, resulting in the recruitment of adaptors that then recruit and activate protein kinases and downstream transcription factors (TFs) (Arnold et al., 2013a). In the case of the RLRs RIG-I and MDA-5, activation leads to conformational and higher-order rearrangements, a process in which RNA binding is followed by PRR interaction with unanchored K63-ubiquitin chains. Such activated PRRs rapidly induce prion-like oligomerization of the mitochondrial adaptor MAVS, which then recruits TRAFs-2, 5, and 6, and two kinase complexes (IKK-α/β/γ and TBK1-IKKi), resulting in activation of the TFs NF-κB and IRF3 (Fig. 2.8.1). Seminal studies by Hardy and coworkers concluded that, depending on the RV strain studied, the RV NSP1 protein targets IRF3 (Graff et al., 2002) or β-TrCP (an essential cofactor for NF-κB activation) (Graff et al., 2009) in the IFN induction cascade. Several studies have uncovered details of these interactions resulting in fascinating insights into RV regulation of the IFN response (Sen et al., 2009; 2011; 2012; Arnold et al., 2013b; Douagi et al., 2007; Arnold and Patton, 2011; Barro and Patton, 2005; 2007; Graff et al., 2007; Feng et al., 2009). In the context of infection, two main approaches have proved useful in deciphering the component intermediate factors involved: the use of mutant RV strains encoding truncated NSP1 that lack IFN antagonistic function, and the use of wild type RV strains encoding full-length NSP1, which induce IFN in some cell types (but not in others). It is important to note that these 2 approaches are likely to query distinct NSP1 functions during RV infection.
TRAFs: In addition to VP4-encoded TRAF binding motifs, NSP1 has also been identified as an inhibitor of TRAF2, an important component of MAVS and noncanonical NF-kB signaling (Bagchi et al., 2013). The NSP1 protein from different RV strains interacts with and targets TRAF2 for degradation. Interestingly, both IRF3 degrading (simian SA11) and NF-kB inhibiting (porcine OSU, bovine A5-13) RV strains degrade TRAF2. The degradation of TRAF2 by NSP1 was shown to inhibit the noncanonical NF-kB pathway triggered by exogenous IFN (Bagchi et al., 2013). Specifically, NSP1 blocks the nuclear translocation of p52 (which, unlike “canonical” p65, is a repressor lacking a transcription activation domain, and is nonessential for IFN induction) in response to exogenous IFN stimulation. Thus far, NSP1-directed TRAF2 degradation does not appear to be important for IFN induction, but may instead play a role in p52-dependent regulation of cytokine expression triggered by IFN (or other cytokines), as well as in sensitizing cells to potential p52-mediated pro-apoptotic effects of IFNβ, although these possibilities have not been directly examined.
NF-kB: Timely and robust induction of type I IFNs requires formation of a complex that includes IRF3 and NF-kB, both of which are critical for gene induction with an important caveat. Specifically, once viruses activate IRF3/IRF7, whether NF-kB is essential for IFNβ induction remains controversial, and NF-kB has alternately been proposed as essential for maintaining basal autocrine expression of IFNβ and ISGs in the absence of infection (or in uninfected bystander cells) (Wang et al., 2010). Of note, infection with RV strains such as NCDV and OSU that inhibit NF-kB activation results in efficient NSP1-dependent blockage of IFNβ induction and secretion, although IRF3 is activated and nuclear (Holloway et al., 2009; Graff et al., 2009). In addition, IECs infected with murine RV in vivo induce several IRF3-dependent transcripts, but neither NF-kB-dependent or IFNβ transcripts (Sen et al., 2012). Thus RV-mediated activation of IRF3 per se is not sufficient to induce IFNβ, and specifically targeting of NF-kB may lead to effective IFN inhibition as well.
The RV NSP1 protein from certain strains (primarily porcine and human strains) targets β-TrCP, an F-box protein and essential NF-kB activating factor, for degradation (Graff et al., 2009; Di Fiore et al., 2015; Morelli et al., 2015). In the canonical pathway, NF-kB subunits are held in an inactive configuration by inhibitory IκB-α molecules. Upon RLR stimulation, IκB-α is phosphorylated within a phosphodegron motif (DSGxS, where x is generally a hydrophobic residue) by IKKs, and is subsequently recognized and degraded by the F-box protein β-TrCP which is the substrate-directing component of the E3 ubiquitin ligase complex Skp1-Cul1-F box protein-Rbx1 [(SCF), the key E3 ligase enzymatic component of this complex being Rbx1], thereby releasing NF-kB for transcriptional duties. The NSP1 protein from porcine and human RV strains contains an IκB-α-like phosphodegron sequence (called PDL, or phosphodegron-like motif) at the C-terminus—effectively constituting a “β-TrCP trap” (Di Fiore et al., 2015; Morelli et al., 2015). The PDL motif DSGIS, occurs within the carboxyl-8-aa region of NSP1 from human and porcine strains, and is absent in other NSP1 proteins, including the bovine strains. Notably, not all PDL-containing human and porcine NSP1s degrade β-TrCP, although they are all potent inhibitors of NF-kB, indicating that β-TrCP interaction (rather than degradation) is critical for RV inhibition of NF-kB (Morelli et al., 2015). The sequence determinants that predict whether NSP1-mediated β-TrCP degradation is likely to occur are presently not known but are unlikely to reside in the NSP1 RING domain, which is highly conserved. Mutation of two predicted casein kinase II phosphorylation serines in the NSP1 phosphodegron motif (positions 480 and 483 in OSU NSP1) resulted in complete abrogation of NF-kB inhibition, indicating an essential requirement for priming phosphorylation of NSP1 for subsequent β-TrCP interaction (Morelli et al., 2015). These findings are exciting as they reveal the determinants of NSP1’s ability to block NF-kB- dependent IFN responses, by mimicking a cellular motif. However, certain puzzling questions about RV-mediated NF-kB regulation vis-à-vis IFN induction remain unanswered. Bovine RV NCDV, whose NSP1 does not contain the PDL motif is still able to degrade β-TrCP in a proteasome-dependent manner, and actively inhibits p-IκB-α degradation and NF-kB function during infection indicating that PDL-motif independent mechanisms may exist (Graff et al., 2009). Indeed, several lines of evidence point to additional RV strategies to usurp NF-kB functions related to the innate response (Holloway et al., 2009; Sen et al., 2012; Graff et al., 2009; Arnold and Patton, 2011; Holloway et al., 2014). The bovine RV A5-16, which encodes a severely truncated ∼50-aa long NSP1 lacking both the PDL motif and the RING finger, is able to sequester NF-kB p65 within viroplasms at 6 hpi. Interestingly, inhibition of p65 following TNF-α-mediated activation has also been reported during infection with RRV and Wa RV strains (which have contrasting effects on β-TrCP) as early as 6 hpi, and may represent a alternate β-TrCP-independent mechanism by which RV strains block NF-kB function (Holloway et al., 2009). In infected cells, p65 is held in viroplasms instead of translocating into the nucleus, although the viral proteins underlying this interaction remain unknown (Holloway et al., 2009, 2014). Murine EW RV, whose encoded NSP1 also does not contain the β-TrCP interacting motif, is associated with perturbation of NF-kB function (Sen et al., 2012). Intestinal lysates from mice infected with murine EW (but not with simian RRV strain) exhibit accumulation of IκB-α protein at 16hpi, which is accompanied in virus infected IECs by an increase in IRF3- but not NF-kB-dependent ISG transcripts, and within bystander (virus-negative) IECs by significantly reduced basal transcription of NF-kB target genes such as Peli1 and A20 (Sen et al., 2012). It is thus likely that other RV proteins, and/or secreted cellular factors, lead to NF-kB inhibition during infection.
IRF3 and other IRFs: Activation of PRRs, and subsequent assembly of adaptor- kinase complexes leads to phosphorylation of IRF3 at multiple serine/threonine residues and to conformational changes that mediate IRF3 dimerization. Dimeric IRF3 then translocates to the nucleus, where it participates in the expression of several virus stress-induced genes (vSIGs) and IFN. The first indication that RV blocks early IRF3-dependent IFN responses was from findings that RV NSP1 interacts with IRF3, and mediates its proteasomal degradation (Graff et al., 2002; Barro and Patton, 2005). Using yeast 2-hybrid analysis, recombinant GST-IRF3 pull-downs from infected cell lysates, and immunoprecipitation of endogenous IRF3 from infected cells, it was shown that RV NSP1 interacts with IRF3. Mutagenesis approaches have determined that this interaction requires the NSP1 carboxy-terminal, and mutants containing only the last 326-aa, a region rich in helical content, can interact with IRF3 (Graff et al., 2007). A conserved N-terminal RING finger motif in NSP1 is also important, but not sufficient, for the interaction. The NSP1–IRF3 interaction is likely a direct interaction due to its initial identification in a binary yeast 2-hybrid screen, although it is important to note that IRF3 regulation (ie, degradation) itself could involve other unidentified essential intermediary factors. The IFN-inducing ability of a phospho-mimetic mutant of IRF3 in which five critical S/T residues are mutated to aspartic acid resulting in constitutive activation and dimerization is efficiently inhibited by NSP1, indicating that NSP1 targets the dimeric form of IRF3 (Sen et al., 2009). This notion is strengthened by analysis of different IRF mutants (Arnold et al., 2013b), which demonstrates that degradation of IRFs by NSP1 requires the conserved carboxyl IRF-dimerization domain. We have observed previously that NSP1-mediated degradation of IRF3 is in fact more efficient when IRF3 is activated (Sen et al., 2009), pointing to the possibility that the initial PAMP-mediated stimulation of IRF3 during RV infection may serve to generate a more suitable target for degradation by NSP1. This also offers an interesting explanation for the activation of IRF3 (at least as inferred from its transcriptional activity) at 16 hpi within RV-infected intestinal villous epithelial cells in vivo (Sen et al., 2012). Certain RV strains, notably those isolated from porcine and human hosts encode NSP1 proteins that are unable to degrade IRF3, at least in the cell types examined so far (Arnold and Patton, 2011). Infection with such strains generally results in robust and sustained IRF3 activation during the course of infection, although how such strains cope with IRF3- induced antiviral vSIGs is still unknown (Sen et al., 2009; Graff et al., 2009). As discussed, such NSP1 proteins nevertheless inhibit IFN induction by targeting NF-kB activation in the IFN induction pathway.
IFN regulation by NSP1 is likely to be more complex than a framework where all NSP1 proteins are either IRF3 or NF-kB inhibitors, and such functions can be highly dependent on the cell type studied. Wild type RV strains (as opposed to mutant strains encoding truncated NSP1 proteins) have been identified that induce IRF3-dependent IFNβ secretion in certain cell types and degrade IRF3 efficiently in others (Sen et al., 2009). Bovine RV UK, or its encoded NSP1 expressed singly, can target endogenous IRF3 for degradation in simian COS7 cells (and in human 293/HT29 cells). However, UK RV infection or expression of UK NSP1 does not result in IRF3 degradation in murine 3T3 fibroblasts (or in primary MEFs), and instead triggers IFNβ secretion. Interestingly, UK NSP1 efficiently targets recombinant simian IRF3 exogenously expressed in 3T3 cells for proteasomal degradation. Conversely, recombinant murine IRF3 expressed in COS7 cells is still refractory to UK NSP1-mediated degradation (Sen et al., 2009). Thus, the ability of NSP1 to degrade IRF3 depends on the host cell, and a comparison of different NSP1s in 3T3 cells would likely identify UK NSP1 as “non-IRF3 degrading.” Alternate NSP1 degradation-independent mechanisms for inhibiting IRF3, which depend on the nature of PRR stimulation, are also likely to exist. Specifically, in 3T3 cells, UK NSP1 (or UK RV infection) fails to inhibit activation of an IRF3-responsive (PRDIII) luciferase reporter stimulated by overexpression of murine IRF3. However, UK NSP1 efficiently inhibits PRDIII activity in the absence of IRF3 degradation when the pathway is stimulated with liposome-complexed intracellular poly(I:C) instead (Sen et al., 2009). Thus, NSP1 can also inhibit IRF3 function despite lack of degradation under specific contexts of PRR stimulation, which is likely relevant in vivo where stimulation can be cell type- and timing-specific. Adding to this complexity, lack of IRF3 inhibition by an “IRF3-degrading” NSP1, as reported for UK NSP1 in vitro—has also been noted in vivo (Sen et al., 2012). Measurement of IRF3-dependent transcription in isolated murine IECs 16 h after infection with murine RV EW in vivo indicates that IRF3-dependent vSIGs are significantly upregulated. The IRF3 activity is likely pathway-specific, as it is not accompanied by simultaneous increases in NF-kB targets, or of the IFN genes themselves in these single cells. One possibility is that during infection in vivo, RV inhibits NF-kB (and consequently its transcriptional target IFN) earlier than IRF3, leading to the transcriptional changes observed. An alternate possibility is that such transcripts are a result of EW NSP1 failure to inhibit IRF1 and/or IRF7, both of which have transcriptional “footprints” largely overlapping with that of IRF3. Clearly, more studies are needed on NSP1–IRF interactions, particularly in the context of in vivo infection where IRF1/IRF3/IRF7 signaling states are likely influenced by crosstalk from immune cells in the intestine and where IRFs may be regulated temporally, in order to deepen our understanding.