Sustained IRF7 activity induces the kinase MAP3K8, which phosphorylates the proline-rich hinge of IRF3 and redirects it from homodimers into IRF3 and IRF7 heterodimers, broadening the antiviral transcriptome and scaling the cellular response to the persistence of the viral threat.
Schmid S, Sachs D, tenOever BR. Mitogen-activated Protein Kinase-mediated Licensing of Interferon Regulatory Factor 3/7 Reinforces the Cell Response to Virus. Journal of Biological Chemistry. 2014. Volume 289, issue 1, pages 299-311.
DOI 10.1074/jbc.m113.519934. PMID 24275658. PMCID PMC3879553.
Sustained IRF7 activity induces the kinase MAP3K8, which phosphorylates the proline-rich hinge of IRF3 and redirects it from homodimers into IRF3 and IRF7 heterodimers, broadening the antiviral transcriptome and scaling the cellular response to the persistence of the viral threat.
The mammalian antiviral response must be strong enough to clear a replicating virus and restrained enough not to shut down the cell when the trigger is minor. The authors ask how a cell scales its response to the persistence of the threat. Their entry point is MAP3K8, a kinase they had previously identified as one of the few genes induced specifically by IRF7 rather than by IRF3 or by type I interferon signaling. Working in human cells and in fibroblasts from Map3k8 knockout and wild type mice, with reporter assays, gel shift assays, coimmunoprecipitation, phosphorylation mapping and messenger RNA deep sequencing, they establish that MAP3K8 is transcribed and activated downstream of IRF7 during viral infection and that it then acts back on IRF3. MAP3K8 suppresses reporter and endogenous gene output that depends on IRF3 alone, an effect that added IRF7 reverses. It does not block IRF3 activation, nuclear entry or association with the coactivator CBP. Instead it drives phosphorylation within the proline-rich hinge of IRF3 and promotes formation of IRF3 and IRF7 heterodimers. Cells lacking MAP3K8 support higher virus titers and fail to induce roughly seventy infection-induced genes, including three SP100 family members, and expressing SP100B or SP100C raises interferon beta transcription and lowers virus titers. The authors propose a feed forward loop that upgrades the transcription factor in use once infection persists.
Interferon regulatory factors share a consensus binding element but differ in how strictly they require it. IRF3 needs an eight-nucleotide match, while IRF7 tolerates up to three deviations, so IRF7 reaches many more promoters and is often described as the master regulator of the response. IRF3 acts first, from a preexisting pool, and induces small amounts of interferon beta. Interferon beta signaling then raises IRF7, which amplifies the response. IRF3 and IRF7 were known to form both homodimers and heterodimers, and the heterodimer was known to be important for robust interferon beta induction, but what governs which dimer forms was not known. MAP3K8, also called TPL2 or COT, was known to be held inactive by the NF-kappaB1 precursor p105, to be released upon IKK activation, and to phosphorylate MEK and thereby activate ERK. Its characterized roles in innate immunity had come almost entirely from bacterial stimulation models in macrophages and dendritic cells, and the authors note that work with viral pathogens was lacking. Their own earlier comparison of the IRF3, IRF7 and ISGF3 transcriptomes had placed MAP3K8 in the small set of genes unique to IRF7, which motivated this study.
Does the IRF7-specific induction of MAP3K8 constitute a feedback arm of the antiviral response, and if so, how does this kinase alter the transcriptional output of the IRF3 and IRF7 system in a way that strengthens defense against a virus that has not been cleared?
The study moves from transcriptional regulation of the kinase, to its effect on IRF-dependent output, to the biochemical basis of that effect, and finally to consequence for virus replication and for the transcriptome. Promoter-level specificity is established with a luciferase construct driven by the human MAP3K8 promoter compared against an interferon-stimulated response element from ISG15, tested against IRF3, IRF7 and reconstituted ISGF3, and confirmed by gel shift with a 27-nucleotide element identified in the MAP3K8 promoter. Kinase activity is read out through ERK phosphorylation, with siRNA against MAP3K8 to establish that the ERK signal requires it, and with an attenuated influenza A virus carrying two substitutions in the double-stranded RNA-binding domain of NS1, chosen because it induces the innate response rather than suppressing it. Effects on output are probed with reporters for the IRF-binding region of the interferon beta promoter and for ISG15, driven by upstream activators including MDA5, constitutively active RIG-I, MAVS and synthetic double-stranded RNA, in each case with and without MAP3K8 and with IRF3 or IRF7 added back. Mechanism is dissected by asking in turn whether MAP3K8 blocks nuclear import, coactivator recruitment or dimer choice, using microscopy of an IRF3-GFP fusion, coimmunoprecipitation of CBP, and coimmunoprecipitation of tagged IRF3 and IRF7. The modification is localized by phosphatase treatment, by metabolic labeling with radiolabeled ATP, by a truncation series across the known domain structure of IRF3, and by alanine substitution of five residues in the hinge. Function is tested with an attenuated vesicular stomatitis virus that cannot block host messenger RNA export, comparing Map3k8 knockout and wild type fibroblasts by titer, by protein levels and by messenger RNA deep sequencing.
The MAP3K8 promoter is engaged specifically by IRF7. Only activated IRF7 induced the MAP3K8 promoter reporter, while IRF3, IRF7 and ISGF3 all induced the ISG15 element reporter (Figure 1A), and in gel shift only activated IRF7 bound the newly identified IRF-binding element from the MAP3K8 promoter, while all three factors bound the comparable ISG15 element (Figure 1B,C).
IRF7 activation drives ERK phosphorylation through MAP3K8. Activated IRF7, but not activated IRF3, produced robust ERK phosphorylation (Figure 1D), and this was abolished by a pool of three siRNAs against MAP3K8 with total ERK unchanged (Figure 1E,F). Cells stably expressing IRF7 induced MAP3K8 transcripts and phosphorylated ERK by 6 and 8 hours after infection with the NS1-mutant influenza A virus (Figure 1G,H).
MAP3K8 suppresses output that depends on IRF3 and this is reversed by IRF7. MAP3K8 inhibited the interferon beta promoter reporter induced by double-stranded RNA, MDA5, active RIG-I or MAVS, and inhibited endogenous interferon beta transcripts (Figure 2A,C). Added IRF7 restored the response while added IRF3 did not (Figure 2B,D,E). The inhibition was dose-dependent and present at what the authors describe as physiological levels of the kinase (Figure 2F). The same pattern held for the ISG15 element (Figure 2G).
The block is not at activation, import or coactivator recruitment. MAP3K8 did not prevent nuclear accumulation of an IRF3-GFP fusion driven by IKKε or MAVS, with roughly 80 percent of positive cells showing nuclear signal regardless of the kinase (Figure 3A,B), and did not disrupt the association between active IRF3 and CBP (Figure 3C).
MAP3K8 promotes heterodimer formation. Low levels of IRF3 and IRF7 heterodimer were detectable without stimulus and were lost upon IRF3 activation, which the authors attribute to dominant homodimer formation. MAP3K8 strongly increased heterodimer recovery after activation, and CBP recovery fell in parallel (Figure 3D). The explanation offered for the reduced CBP, that the heterodimer presents only a single serine 396 phosphorylation event, is an interpretation drawn from prior literature rather than a measurement made here.
MAP3K8 causes phosphorylation of IRF3 in the proline-rich hinge. IRF3 migrated more slowly in the presence of the kinase, a shift removed by calf intestinal alkaline phosphatase (Figure 4A), and metabolic labeling with radiolabeled ATP confirmed increased phosphorylation of IRF3 (Figure 4B). A truncation series placed the target region between residues 112 and 270, and comparison of constructs spanning residues 151 to 357 against 198 to 357 narrowed it to the proline-rich hinge (Figure 4C,D).
Hinge residues are required for the induced interaction with IRF7. Substituting serine 173, serine 175, threonine 180 and serine 188 together with the associated cluster to alanine markedly reduced recovery of IRF7 with IRF3 in the presence of MAP3K8, while wild type IRF3 gave robust heterodimer (Figure 4E). The authors note that this region is distinct from the residues previously implicated in IRF7 binding at positions 306 to 357 but that published IRF dimer structures place the hinge near the dimer interface, making an intermolecular contribution plausible. They state explicitly that MAP3K8 may act directly or indirectly on IRF3, so the paper does not establish that IRF3 is a direct substrate.
Loss of MAP3K8 increases virus replication. Fibroblasts lacking Map3k8 infected with the matrix-mutant vesicular stomatitis virus at low multiplicity produced significantly higher titers over 48 hours and higher viral glycoprotein at 12 hours (Figure 5A,B).
The transcriptome is narrowed in the absence of the kinase. Messenger RNA sequencing at 10 hours after infection showed roughly seventy genes induced to higher levels in wild type than in knockout fibroblasts, including interferon alpha subtypes and IRF7-associated genes such as Oas2, Oas3, Herc6 and Gbp6. Three SP100 family members, Sp100, Sp110 and Sp140, were about fivefold higher in wild type cells, confirmed by quantitative PCR (Figure 6A,B).
SP100 family promoters are IRF7 targets. Only activated IRF7 bound the previously identified IRF-binding elements from the SP100 and SP110 promoters in gel shift (Figure 6C), and activated IRF7 induced two SP100 splice variants and SP110 approximately fortyfold (Figure 6D).
SP100B and SP100C contribute antiviral activity. Neither induced interferon beta on its own, but either raised MAVS-driven interferon beta transcripts roughly fivefold (Figure 7A). Expressing either reduced vesicular stomatitis virus titers by nearly a log and reduced viral glycoprotein while raising STAT1 (Figure 7B,C). The paper itself claims priority for the observation that SP100B and SP100C increase interferon beta expression during the antiviral response, and that claim is reported here as the paper's own.
The model advanced is a feed forward loop with a transcription factor upgrade at its center. IRF3, activated early by pattern recognition signaling, produces limited interferon beta because the enhanceosome is not optimized for IRF3 homodimers. Interferon beta raises IRF7. If pathogen-associated molecular patterns persist so that upstream kinases keep IRF7 active, IRF7 induces MAP3K8. MAP3K8 then phosphorylates or causes phosphorylation of the IRF3 hinge, which disfavors the IRF3 homodimer and favors the IRF3 and IRF7 heterodimer. Because IRF7 tolerates degenerate binding elements, the resulting complex reaches promoters the homodimer cannot, including those of the SP100 family, whose products in turn raise interferon beta output. MAP3K8 simultaneously activates ERK, which has been linked by others to assembly of promyelocytic leukemia nuclear bodies, the structures SP100 proteins occupy.
Several steps are directly demonstrated. The IRF7 specificity of MAP3K8 induction, the requirement for MAP3K8 in virus-induced ERK activation, the shift in dimer composition, the localization of the phosphorylation to the hinge, the requirement for hinge residues in heterodimer formation, the transcriptome narrowing and the virus replication phenotype are all supported by experiment. Other steps are proposals. The paper does not establish that MAP3K8 phosphorylates IRF3 directly and says so. It does not show that hinge phosphorylation occurs on endogenous IRF3 during infection, nor identify which residues are modified as opposed to which are required. The connection from ERK to nuclear body assembly is drawn from other laboratories and is offered as a suggestion about how the arms of the pathway may be interconnected. The proposal that the loop scales the response to the duration of pathogen-associated molecular pattern production is a framing consistent with the data rather than a measured relationship.
The work supplies a candidate answer to a standing question about IRF biology, namely what determines whether IRF3 acts as a homodimer or with IRF7. By locating a regulatory input in the proline-rich hinge, distinct from the C-terminal regulatory domain that receives the canonical activating phosphorylation and distinct from the mapped IRF7 contact residues, it separates dimer partner choice from activation and nuclear entry as independently controlled steps. It also reframes an apparently inhibitory effect. MAP3K8 reduces IRF3-only output, which in isolation reads as suppression of innate immunity, but in the presence of IRF7 the same activity broadens the response. That reading makes the phenotype of MAP3K8 loss interpretable and predicts that the kinase should matter only once IRF7 is available. Finally it brings SP100 family proteins into the interferon beta induction circuit as positive regulators, connecting the IRF7 and MAP3K8 arm to nuclear body biology.
The study grows directly out of the 2010 comparison by Schmid, Mordstein, Kochs, García-Sastre and tenOever of the IRF3, IRF7 and ISGF3 transcriptomes, which supplied both the identification of MAP3K8 as an IRF7-unique target and several of the reagents and reporters used here. The recurring question across those two papers, and shared with the IKKε work from the same laboratory, is how a small set of related transcription factors with overlapping binding preferences is allocated so that the transcriptional output matches the situation rather than simply maximizing. In the IKKε papers the allocation is between STAT1-containing complexes and is set by a kinase acting on a dimer interface. Here the allocation is between IRF-containing complexes and is set by a kinase acting near a dimer interface. Reading those results as one recurring design principle is category 3 synthesis across the 2010, 2011 and 2014 papers and is offered as such.
Much of the mechanistic work rests on transient overexpression in HEK293T cells, where levels of IRF3, IRF7, MAP3K8 and the upstream activators are set by transfection rather than by infection, and the paper's own statement that the inhibitory effect is evident at physiological levels of MAP3K8 is based on a dose titration in that system. Phosphorylation of the IRF3 hinge is shown for tagged IRF3 in the presence of exogenous MAP3K8 and is not demonstrated on endogenous IRF3 during infection, no phosphosite is directly identified by mass spectrometry, and whether MAP3K8 acts on IRF3 directly is left open. The alanine substitution experiment shows that those residues are required for the induced interaction but does not show that they are the modified residues. Virus work uses two attenuated viruses chosen for their inability to suppress the host response, an influenza A virus with a mutated NS1 and a vesicular stomatitis virus with a matrix substitution, so the findings speak to the behavior of the pathway rather than to the outcome of wild type infection. The Map3k8 knockout is constitutive and whole-cell, and the transcriptome comparison is a single time point at 10 hours in duplicate. The SP100 experiments rely on overexpression of two isoforms and do not test loss of function, so SP100B and SP100C are shown to be sufficient to raise interferon beta and lower titer, not to be necessary for the MAP3K8 phenotype. No animal experiments are reported, and the ERK to nuclear body link is inferred from the literature rather than tested.
A kinase induced late in infection modifies IRF3 so that it pairs with IRF7 instead of itself, widening the set of antiviral genes a cell can reach.
Cells must match the strength of an antiviral response to the size of the threat. The authors describe a loop that does this. Persistent infection keeps IRF7 active, IRF7 induces the kinase MAP3K8, and MAP3K8 causes phosphorylation of IRF3 in a flexible hinge region. The modified IRF3 pairs with IRF7 rather than with itself, and the resulting complex reaches promoters the IRF3 pair cannot. Cells lacking the kinase induce fewer antiviral genes and support more virus.
IRF3 requires a stringent binding element while IRF7 tolerates degenerate ones, so which dimer forms determines how much of the antiviral genome is accessible. Building on their earlier finding that MAP3K8 is induced specifically by IRF7, the authors show that MAP3K8 is transcribed from an IRF7-selective element in its promoter, is activated during infection, and drives ERK phosphorylation. MAP3K8 suppresses interferon beta and ISG15 output that depends on IRF3 alone, an inhibition reversed by IRF7. It does not block IRF3 activation, nuclear entry or CBP binding. Instead it produces phosphorylation within the proline-rich hinge of IRF3, mapped by truncation, and hinge residues are required for the induced pairing with IRF7. Fibroblasts lacking MAP3K8 support higher vesicular stomatitis virus titers and fail to induce about seventy genes, among them Sp100, Sp110 and Sp140, and expressing SP100B or SP100C raises interferon beta and lowers titer. Direct phosphorylation of IRF3 by MAP3K8 is not established.
lab-led for Schmid 2010 and Schmid 2014, co-led with the Maniatis laboratory for Ng 2011, and training period for tenOever 2007, which was carried out with the Maniatis and García-Sastre laboratories and is not the independent program's work