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.
tenOever BR, Ng SL, Chua MA, McWhirter SM, García-Sastre A, Maniatis T. Multiple Functions of the IKK-Related Kinase IKKε in Interferon-Mediated Antiviral Immunity. Science, 2007, volume 315, issue 5816, pages 1274-1278.
DOI 10.1126/science.1136567. PMID 17332413.
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.
Type I interferon induction and type I interferon signalling had been treated as two largely separate arms of the antiviral response, with the IKK-related kinases TBK1 and IKKε assigned to the induction arm as activators of IRF3 and IRF7. This study asked what IKKε actually does in a whole animal. Mice deficient in Ikbke were generated and challenged with influenza A/WSN/33. They were more susceptible to infection at sublethal doses and carried higher pulmonary viral loads, yet their production of interferon-β, other cytokines and virus-specific antibody was not detectably impaired. Transcriptional profiling of infected lung and of primary embryonic fibroblasts instead showed that a defined subset of interferon-stimulated genes, including Adar1, Ifit3 and Ifi203, was poorly induced. Loss of ADAR1 induction had a measurable consequence, since adenosine-to-guanosine editing of influenza matrix mRNA was largely absent in the knockout animals. Treating fibroblasts with recombinant interferon-β reproduced the defect, placing IKKε in the signalling arm rather than the induction arm. Gel shift and chromatin immunoprecipitation assays showed that ISGF3 failed to occupy the affected promoters, and rescue required IKKε kinase activity. Interferon-β activated IKKε, and recombinant IKKε phosphorylated STAT1 at Ser708, Ser744 and Ser747 in vitro. A STAT1 Ser708Ala mutant selectively lost binding at an IKKε-dependent element while retaining binding at an IKKε-independent one. The work reframes IKKε as a determinant of promoter selectivity within the interferon response.
Virus infection triggers pattern recognition receptors that activate latent kinase complexes, among them the IKK complex and the IKK-related kinases TBK1 and IKKε, which assemble the interferon-β enhanceosome through ATF2 and cJun, NF-κB, IRF3 and IRF7. Secreted interferon-β then engages its receptor, activates TYK2 and JAK1, and drives tyrosine phosphorylation of STAT1 and STAT2, which associate with IRF9 to form ISGF3 and bind interferon-stimulated response elements across a large family of interferon-stimulated genes. Work in cultured cells before this study had placed virus-inducible IKKε in a role redundant with the ubiquitously expressed TBK1 in activating IRF3 and IRF7, a conclusion the paper attributes to earlier reports from several groups. What IKKε contributes in an intact animal, and whether its only function lies upstream of interferon production, was unresolved.
What is the in vivo function of IKKε in antiviral immunity, and is that function confined to the induction of interferon-β or does it extend into the interferon signalling pathway itself?
The design begins with loss of function in a whole animal rather than in a cell line, because a kinase reported to be redundant with TBK1 ex vivo may nonetheless carry a non-redundant role in vivo. Ikbke-null mice were generated and challenged intranasally with influenza A/WSN/33 across a dose range, allowing separation of a survival phenotype from a lethality ceiling. Cytokine output and antibody responses were measured to test the expected induction defect, and unbiased microarray profiling of infected lung was used to look for defects the candidate assays would miss. Primary embryonic fibroblasts from the same animals then allowed the phenotype to be moved into a controlled setting, where recombinant interferon-β could be supplied exogenously. That step is the pivot of the design, since a defect that persists when interferon is provided from outside the cell cannot be an interferon induction defect. Promoter occupancy was then interrogated directly by gel shift and chromatin immunoprecipitation, kinase requirement by reconstitution with wild-type and catalytically inactive IKKε, and substrate identity by in vitro phosphorylation with mass spectrometry followed by phosphosite mutants expressed in Stat1-null fibroblasts.
Ikbke-null mice were more susceptible to influenza A/WSN/33 than wild-type littermates at sublethal doses, continuing to lose body mass below 80 percent of starting weight after inoculation with 700 plaque-forming units, whereas doses above 7000 plaque-forming units killed both cohorts (Fig. 1A and 1B). The observation establishes a non-redundant requirement for IKKε in host survival of influenza infection within this dose window.
Knockout mice carried elevated pulmonary viral load by quantitative PCR for viral nucleocapsid mRNA and by plaque assay (Fig. 2A). Lung histopathology showed an inflammatory infiltrate of lymphocytes, macrophages and neutrophils.
Virus-induced cytokine output was not detectably impaired. Interferon-α, interferon-β, interferon-γ, RANTES and IL2 in lung and serum, and virus-specific antibody amounts, were comparable between genotypes (Fig. 2B). This is the finding that separates the phenotype from the expected interferon induction defect.
Microarray profiling of pooled infected lung identified a subset of interferon-stimulated genes, including Ifit3, Ifi203 and Adar1, that were poorly induced without IKKε, confirmed by RT-PCR (Fig. 2B).
The failure to induce ADAR1 had a functional readout on the virus. Sequencing of the influenza matrix mRNA stem loop from infected lung showed that more than 30 percent of transcripts carried at least one adenosine-to-guanosine transition in control mice, against less than 5 percent in knockout mice (Fig. 2C, 96 transcripts per cohort). The authors read this as evidence that IKKε is required for induction of a subset of interferon-stimulated genes that includes an effector with antiviral activity in vivo.
Primary embryonic fibroblasts reproduced the phenotype. Replication rates were comparable at 2 days post infection but titres continued to rise in knockout cells thereafter (Fig. 2D), with comparable interferon-β induction but reduced induction of the same interferon-stimulated gene subset. Low amounts of virus-inducible Adar1 mRNA were still seen, whereas Ifi203 induction was completely blocked (Fig. 2E).
Treating fibroblasts with recombinant interferon-β reproduced the defect. About 30 percent of interferon-inducible genes were poorly induced in knockout cells (Fig. 3A and 3B). Adar1, Ifit3 and Ifi203 were affected while Irf7, Prkra and Stat1 were not. Because interferon was supplied exogenously, this places the lesion in the signalling arm.
The affected elements share a consensus resembling characterized ISGF3 binding sites, and gel shift assays showed an interferon-inducible DNA-protein complex that failed to form on IKKε-dependent elements from Oas1b, Mx1 and Adar1 in extracts from knockout cells (Fig. 3C). Antibodies to IRF9, STAT1 and STAT2 each disrupted the complex, identifying it as ISGF3 (Fig. 3D).
Reintroducing wild-type IKKε into knockout fibroblasts restored interferon-induced ISGF3 binding, while the catalytically inactive Lys38Ala mutant did not (Fig. 3E). Occupancy therefore depends on IKKε kinase activity and not on the protein as a scaffold.
IKKε is activated by interferon itself. A phosphospecific antibody to Thr501 detected rapid IKKε phosphorylation in interferon-treated cells (Fig. 4A). The paper notes that TBK1 is not activated by interferon treatment and lacks the equivalent residue. The suggestion that p38 signalling mediates IKKε activation is presented by the authors as speculation, not as a demonstrated step.
Recombinant IKKε phosphorylated STAT1 at Ser708, Ser744 and Ser747 in vitro, mapped by mass spectrometry. Expressed in Stat1-null fibroblasts, the Ser708Ala mutant markedly reduced ISGF3 binding and occupancy at the IKKε-dependent Adar1 element but not at the IKKε-independent Irf7 element, by both gel shift and chromatin immunoprecipitation, whereas a C-terminal deletion removing Ser744 and Ser747 had no effect at either element (Fig. 4D and 4E).
Interferon-γ-stimulated gene induction was normal in bone marrow-derived macrophages from knockout mice, and IKKε-independent elements showed normal ISGF3 binding, bounding the defect to a subset of the type I interferon response.
The study establishes the central link, that IKKε is activated downstream of the interferon receptor and phosphorylates STAT1 at Ser708, and that Ser708 is required for ISGF3 occupancy at a promoter subset. It does not establish the structural basis of that requirement, and the authors state that the structural consequences of the phosphorylation remain to be determined.
The model the authors propose is that phosphorylation of STAT1 Ser708 favours STAT1 and STAT2 heterodimer formation over STAT1 homodimers, drawing on a published crystal structure from another group rather than on structural data generated here. Because STAT2 is thought to tether STAT1 and IRF9 to the element, heterodimer formation would be required for productive ISGF3 assembly. Computational comparison of IKKε-dependent and IKKε-independent elements identified a purine-rich region upstream of the IKKε-independent elements, which the authors suggest could act as an additional STAT2 contact. In their reading, elements lacking that purine tract would need a more stable STAT1, STAT2 and IRF9 interaction, and it is that stabilization that Ser708 phosphorylation supplies. This account is author interpretation built on sequence analysis and on structures from the literature, and the paper does not test it directly.
A second proposal in the paper, that STAT1 is subject to IKKε-dependent processing or degradation, rests on the observation of discrete STAT1 breakdown products in wild-type but not knockout lung and on higher STAT1 amounts in knockout fibroblasts. The authors explicitly present this as a suggestion that pointed them toward STAT1 as a candidate substrate, not as an established mechanism.
The work moves IKKε out of a purely redundant position upstream of interferon production and places it inside interferon signalling, where its activity determines which interferon-stimulated genes are transcribed. That makes the interferon-stimulated gene set divisible into IKKε-dependent and IKKε-independent classes distinguishable by their response elements, which gives a handle for asking why a cell would regulate effector genes and signalling-machinery genes separately. The authors advance the interpretation that IKKε-dependent genes act as direct antiviral effectors while IKKε-independent genes serve the signalling machinery that integrates innate and adaptive immunity, and they frame the phenotype as a failure of the local response despite an intact systemwide one. The identification of a serine phosphosite on STAT1 that governs promoter selectivity rather than overall pathway activation also supplies a mechanism by which a transcription factor complex can be tuned rather than switched.
This paper predates the independent tenOever laboratory and was carried out with the Maniatis laboratory at Harvard together with the García-Sastre laboratory at Mount Sinai, with tenOever as first author. Its recurring concerns carry forward into the later corpus. The use of influenza A virus infection in a defined host genetic background as the assay for an innate immune pathway, and the treatment of the interferon-stimulated gene set as a structured output rather than a single readout, both recur.
Category 3 synthesis, visible only across papers. ADAR1 enters this paper as an IKKε-dependent effector whose editing activity is measured directly on viral RNA. ADAR1 and the boundary between double-stranded RNA sensing and RNA-directed processes remain active subjects in the laboratory's later work, so this paper marks an early point of contact with that theme. Establishing the continuity requires the later papers and is not supported by this paper alone.
No relationship to the other papers in this batch is asserted, since none of them cite or extend this work directly.
The in vivo work uses a single virus, influenza A/WSN/33, in mice, and the survival phenotype is confined to a narrow dose window, since doses above 7000 plaque-forming units killed both genotypes. The knockout is germline and constitutive, so cell-intrinsic and systemic contributions are not separated, and no conditional or cell-type-restricted allele was used. Microarray profiling of lung was performed on pooled samples, which limits statistical treatment of the transcriptional differences. The editing readout is based on 96 sequenced matrix transcripts per cohort from pooled material. Phosphorylation of STAT1 by IKKε is shown with recombinant protein in vitro, and the functional test of Ser708 uses ectopic expression of mutant STAT1 in Stat1-null fibroblasts rather than a knock-in animal, so the requirement is established for reconstituted cells rather than for endogenous STAT1 in vivo. The paper does not show that Ser708 phosphorylation of endogenous STAT1 rises after interferon treatment in an IKKε-dependent manner. The structural interpretation of Ser708, the purine tract model for IKKε-independent elements, and the p38 link to IKKε activation are all proposals rather than demonstrations. Interferon-γ responses and IKKε-independent elements are unaffected, so the conclusions do not extend to the type II interferon response or to the interferon-stimulated gene set as a whole.
Mice lacking IKKε make normal interferon but cannot switch on a third of interferon-response genes, because IKKε phosphorylates STAT1 and directs it to particular promoters.
IKKε was thought to act only in producing interferon-β. Mice lacking it made interferon normally yet still succumbed to influenza, because a defined subset of interferon-stimulated genes was never induced. The defect persisted when interferon was added to cells directly, placing IKKε inside interferon signalling. Interferon activates IKKε, which phosphorylates STAT1 at serine 708, and that residue is required for the ISGF3 complex to occupy the affected promoters but not others.
The IKK-related kinases TBK1 and IKKε had been assigned to interferon-β induction, with IKKε treated as redundant. Ikbke-null mice challenged with influenza A/WSN/33 showed higher viral loads and reduced survival at sublethal doses despite normal interferon, cytokine and antibody responses. Transcriptional profiling of infected lung and of primary fibroblasts showed that roughly 30 percent of interferon-stimulated genes, including Adar1, Ifit3 and Ifi203, were poorly induced, and adenosine-to-guanosine editing of influenza matrix mRNA was largely lost in the knockout, giving the ADAR1 defect a functional readout. The defect persisted when recombinant interferon-β was supplied, locating it in signalling. ISGF3 failed to occupy the affected response elements, and rescue required IKKε catalytic activity. Interferon activated IKKε at Thr501, and recombinant IKKε phosphorylated STAT1 at Ser708, a residue required for ISGF3 occupancy at IKKε-dependent but not IKKε-independent elements. The structural basis of that requirement is proposed rather than demonstrated.
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