Why do different infections induce different subsets of interferon-stimulated genes?
The interferon response is usually described as a switch. Virus is detected, interferon is made, interferon signals, and a few hundred interferon-stimulated genes come on. That account cannot explain why the induced set differs between infections, cell types and host genotypes, and it treats a gene annotated as interferon-stimulated as though its induction required interferon.
The problem is one of allocation rather than activation. A small number of related transcription factors, IRF3, IRF7 and the STAT-containing complexes ISGF3 and GAF, bind partially overlapping DNA elements and draw on shared subunits. What determines which complex assembles, which element it reaches, and therefore which part of the antiviral genome is transcribed.
Two of the four publications are lab-led, Schmid 2010 and Schmid 2014, both with tenOever as senior and corresponding author. Ng 2011 is co-led, first-authored from the Maniatis laboratory with Maniatis and tenOever as joint corresponding authors. tenOever 2007 is training-period work from the Maniatis and García-Sastre laboratories and is treated here as the observation the later work explains.
tenOever 2007 supplied the phenomenon. Interferon-stimulated response elements sorted into IKKepsilon-dependent and IKKepsilon-independent classes, and the substrate distinguishing them was STAT1 Ser708. Ng 2011 resolved why that residue matters. STAT1 is shared between ISGF3, where it partners STAT2 and IRF9 at interferon-stimulated response elements, and GAF, where it homodimerises at gamma-activated sequences. Ser708 sits in the homodimer interface of the published tyrosine-phosphorylated STAT1 structure. Phosphorylation blocks the activated homodimer while leaving the STAT1 and STAT2 interaction intact, so available STAT1 is driven into ISGF3. Loss of IKKepsilon lowered ISGF3 assembly and element binding while raising GAF assembly, adenoviral IKKepsilon produced the mirror-image shift, and recombinant STAT1 phosphorylated by both a JAK kinase and IKKepsilon failed to bind a gamma-activated sequence while still supporting ISGF3. Genome-wide expression and STAT1 occupancy in macrophages tracked a continuous score of relative type I and type II responsiveness. The structural explanation is the authors' proposal built on an existing homodimer structure, and they state that the consequences within ISGF3 are unknown.
Schmid 2010 attacked the same problem from the DNA side. Mice lacking both the type I and type III interferon receptors still induced a large block of interferon-stimulated genes after infection with an NS1-deficient influenza A virus, so those genes do not require interferon signalling. Systematic mutagenesis of the ISG15 element, read against reconstituted IRF7 and reconstituted ISGF3, separated positions conferring IRF7 specificity, ISGF3 specificity or both, with ISGF3 requiring contacts beyond the consensus core and IRF7 tolerating more variation because favourable upstream bases can offset a poor core contact. Activating IRF7 in STAT1-deficient U3A cells reproduced about 80 percent of the gene set induced in infected knockout lung, and individual promoter elements sorted as predicted, with MxA restricted to ISGF3 and CXCL10 restricted to IRF7. The consensus sequences derived there are generalisations from derivatives of one element.
Schmid 2014 followed one gene out of that comparison. MAP3K8 had appeared as a target unique to IRF7, and this study shows it is transcribed from an IRF7-selective element, activated during infection, and that it acts back on IRF3. MAP3K8 suppressed output depending on IRF3 alone, an inhibition reversed by adding IRF7, without blocking IRF3 activation, nuclear entry or CBP association. Instead it drove phosphorylation within the proline-rich hinge of IRF3, mapped by truncation, and hinge residues were required for the induced pairing with IRF7. Fibroblasts lacking MAP3K8 supported more vesicular stomatitis virus and failed to induce roughly seventy genes including Sp100, Sp110 and Sp140, and expressing SP100B or SP100C raised interferon beta and lowered titre. The paper states that it does not establish MAP3K8 as a direct kinase for IRF3, and does not show hinge phosphorylation on endogenous IRF3 during infection.
The line runs from an unexplained promoter class in 2007, to a mechanism of subunit allocation in 2011, to the DNA-level grammar that makes allocation matter in 2010, and to a second allocation step among the IRFs in 2014. Read together, which is synthesis rather than a claim any one paper makes, the same design principle appears twice. A kinase acting at or near a dimer interface decides which of two competing complexes forms, and promoter element architecture decides which genes notice. In 2011 the interface is the STAT1 homodimer contact. In 2014 it is the IRF3 hinge.
The thread did not continue as a mechanistic program. After 2014 the laboratory's transcription-factor work moves toward whole-tissue and whole-animal readouts, and the structural basis of both the Ser708 and hinge effects remains proposed rather than solved.
The interferon-independent arm mapped in Schmid 2010 is what makes the constitutively active IRF7 of Eggenberger 2019 usable in cells that cannot respond to interferon, linking this theme to interferon and cell identity. The demonstration that activating factors alone do not determine output is completed from the other direction by Manivasagam 2025 under homeostatic repression, where a repressor at a distinct motif sets whether a promoter can respond at all. Benitez 2015 uses the Schmid 2010 framework to interpret why MDA5 loss costs Irf7 induction without costing interferon beta, connecting selectivity to sensing aberrant RNA. The NS1-deficient influenza A virus used here to unmask host transcription recurs as an instrument across the area and into influenza genome regulation.
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.
Phosphorylation of STAT1 serine 708 by IKKε blocks formation of the STAT1 homodimer that constitutes GAF while leaving the STAT1 and STAT2 heterodimer of ISGF3 intact, biasing the shared STAT1 pool and the interferon-stimulated transcriptome toward the type I response.
IRF7 and ISGF3 engage overlapping interferon-stimulated response elements and drive largely overlapping antiviral transcriptomes, so that a substantial interferon-like gene program is still induced when type I and type III interferon signaling are both absent.
Mice lacking IKKε produce normal interferon-β but fail to induce roughly a third of interferon-stimulated genes, because interferon activates IKKε, which phosphorylates STAT1 at Ser708 and thereby determines whether ISGF3 occupies a subset of response elements.