Extended Data Fig. 6: SILAC identifies Ser90 as an additional IRF6 phosphorylation site that is essential for its activation in vitro.

a, Schematic of the SILAC experiment that was performed to identify additional RIPK4-dependent phosphorylation sites on IRF6. b, Western blots of 293T cells expressing the IRF6 phosphorylation mutants that were used in luciferase reporter assays. Representative of three independent experiments. c, Schematic of the human IRF6 locus. Exons are shown as rectangles, introns as interconnecting lines and untranslated regions are shaded in grey. The DNA-binding domain (DNA BD), IRF-association domain (IRF AD) and C-terminal domain (CTD) are highlighted in blue, green and orange, respectively. The relative positions of two patient mutations, S90G (which gives rise to VWS) and S424L (which gives rise to PPS) are displayed in red. The locus is drawn approximately to scale. d, Extracted ion chromatograms of phosphorylated peptides and their unmodified counterparts at pS(413,416) (SFDSGSVR) (i), pS424 (short peptide: LQISTPDIK) (ii), pS424 (long peptide: LQISTPDIKDNIVAQLK) (iii) and pS90 (SREFNLMoxYDGTK) (iv). Recombinant full-length IRF6 was incubated with the RIPK4 kinase domain (either wild type or T184I (a BPS mutation that produces a kinase-dead version of RIPK4) for 5 and 30 min. No phosphorylation was observed when IRF6 was incubated with RIPK4(T184I). Representative of two experiments. e, Proposed model for IRF6 regulation by RIPK4. RIPK4 (or kinase(s) X) phosphorylates IRF6 at Ser413 and Ser424, which act as ‘priming’ sites. Priming enhances the phosphorylation of IRF6 by RIPK4 at an additional site that is essential for IRF6 activation, Ser90. This allows normal skin differentiation and development. In the first scenario, in which Irf6 is mutated to Irf6^{S413,S424A/S413A,S424A}, RIPK4 (or other kinases) cannot phosphorylate Ser413 and Ser424. IRF6 is thus non-functional, so the Irf6^{S413A,S424A/S413A,S424A} knock-in mouse phenocopies the Irf6^−/− mouse. In the second scenario, in which Irf6 is mutated to Irf6^{S413E,S424E/S413E,S424E}, the Glu residues at Ser413 and Ser424 mimic priming and allow RIPK4 to phosphorylate IRF6 at Ser90. Thus, IRF6 is functional, and Irf6^S413E,S424E resembles the wild type. In the third, double-mutant scenario (Irf6^{S413E,S424E/S413E,S424E}Ripk4^D161N/D161N), despite effective IRF6 priming at Ser413 and Ser424 as a result of Glu substitutions, RIPK4 is kinase-dead and therefore cannot phosphorylate IRF6 at Ser90 and activate it. Thus, IRF6 is non-functional and this double mutant phenocopies the Irf6^−/− mouse. In the final scenario (Irf6^{+/S413E.S424E}Ripk4^D161N/D161N), one wild-type allele of Irf6 is present (sufficient for normal IRF6 function); however, there is no RIPK4 kinase activity. Therefore, these mice phenocopy Ripk4^D161N/D161N mice.