Which kinases convert pathogen sensing into interferon gene induction?
By the start of the 2000s the interferon beta enhancer was understood as a composite element requiring nuclear factor kappa B, ATF-2 with c-Jun, and the interferon regulatory factors acting together. Two of those three routes had been traced to their kinases. The third had not. IRF-3 and IRF-7 were known to be activated by phosphorylation of a serine and threonine cluster in their C-terminal regulatory region, which permits dimerisation, nuclear translocation and DNA binding, but the responsible enzyme was known only by an operational name, the virus-activated kinase. Sharma 2003 states the gap in those terms and calls it a critical missing link.
A second problem sat behind the first. If two related kinases carry out the same reaction on the same substrates, what is each of them for. Answering that requires moving out of reconstituted biochemistry and into an animal, where a kinase described as redundant may turn out not to be.
Both publications carry the contribution character training period, and neither belongs to the independent laboratory. Sharma 2003 came from doctoral training in the Hiscott laboratory at the Lady Davis Institute and McGill University, with John Hiscott and Rongtuan Lin as corresponding authors and tenOever second of three authors marked as contributing equally. tenOever 2007 was carried out with the Maniatis laboratory at Harvard together with the García-Sastre laboratory at Mount Sinai, with tenOever as first author and Tom Maniatis as corresponding author. The work is recorded here because it is where the questions pursued later actually begin.
Sharma 2003 surveyed the IKK family against recombinant IRF-3 and IRF-7 C termini and found that IKKepsilon and TBK1, and not IKKalpha or IKKbeta, phosphorylate the regulatory cluster. An alanine-substituted substrate was not phosphorylated and a catalytically inactive IKKepsilon blocked the reaction. In cells the same kinases produced Ser396-phosphorylated IRF-3, drove nuclear accumulation of both factors and generated IRF-containing complexes on DNA. Only these two activated the IFNA4 and IFNB reporters while every family member tested activated a nuclear factor kappa B reporter, which is the paper's argument for a division of labour inside the family. Silencing both kinases in A549 cells blocked virus-induced IRF-3 phosphorylation and ISG56 induction, and IKKepsilon restricted vesicular stomatitis virus in an IRF-3 dependent manner. How virus detection reaches the kinases was not addressed, and the paper says so.
tenOever 2007 then asked what IKKepsilon does in a whole animal and revised the redundancy model rather than confirming it. Mice lacking Ikbke were more susceptible to influenza A/WSN/33 at sublethal doses and carried higher pulmonary virus, yet made normal interferon beta, other cytokines and virus-specific antibody. What failed was induction of roughly a third of interferon-stimulated genes, including Adar1, Ifit3 and Ifi203, and the failure persisted when recombinant interferon beta was supplied from outside the cell. That single design choice moves the lesion from the induction arm into the signalling arm. ISGF3 did not occupy the affected elements, rescue required IKKepsilon catalytic activity, interferon itself activated IKKepsilon at Thr501, and recombinant IKKepsilon phosphorylated STAT1 at Ser708, a residue required for occupancy at IKKepsilon-dependent but not IKKepsilon-independent elements. Loss of ADAR1 induction had a measurable consequence on the virus, since adenosine to guanosine editing of influenza matrix mRNA fell from more than 30 percent of transcripts to under 5 percent.
The two papers move in opposite directions along one pathway. Sharma 2003 works upstream of interferon and assigns an activity to enzymes. tenOever 2007 works downstream of interferon and shows that one of those enzymes has a second job there, determining which genes the response reaches. The structural basis of the Ser708 requirement was proposed rather than solved, and resolving it is what Ng 2011 takes up under transcription factor selectivity.
Both studies carry real boundaries. Sharma 2003 rests heavily on ectopic expression and recombinant fragments, with necessity established from partial knockdown in one cell line and two RNA viruses. tenOever 2007 uses a germline constitutive knockout, one virus and pooled microarray material, tests Ser708 by ectopic expression of mutant STAT1 in Stat1-null cells rather than in a knock-in animal, and does not show that endogenous STAT1 Ser708 phosphorylation rises after interferon treatment in an IKKepsilon-dependent way.
The Ser708 result opens directly into transcription factor selectivity, where Ng 2011 explains the modification as a block on STAT1 homodimer formation. Treating the interferon-stimulated gene set as a structured output rather than one readout recurs throughout the area and underlies both homeostatic repression and calibration of the response in vivo. ADAR1 enters the corpus here as an IKKepsilon-dependent effector measured directly on viral RNA, an early point of contact with editing biology that reappears in work on viral populations and evolution. Blanco-Melo 2020 cites Sharma 2003 for TBK1 as the kinase driving type I and type III interferon expression, which is the one explicit textual link from this training-period work into the pandemic-era corpus.
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.
The IKK-related kinases IKKepsilon and TBK1 are identified as components of the virus-activated kinase activity that phosphorylates the C-terminal serine cluster of IRF-3 and IRF-7, linking pathogen sensing to type I interferon gene induction and to establishment of an antiviral state.