Ku suppresses RNA-mediated innate immune responses in human cells to accommodate primate-specific Alu expansion

Ku70 and Ku80 form Ku , a ring-shaped protein that initiates the non-homologous end-joining (NHEJ) DNA repair pathway. Specifically, Ku binds to double-stranded DNA (dsDNA) ends and recruits other NHEJ factors (e.g., DNA-PKcs and LIG4). While Ku binds to double-stranded RNA (dsRNA) and traps mutated-DNA-PKcs on ribosomal RNA in vivo, the physiological significance of Ku-dsRNA interactions in otherwise wild-type cells remains elusive. Intriguingly, while dispensable for murine development, Ku is essential in human cells. Despite similar genome sizes, human cells express ~100-fold more Ku than mouse cells, implying functions beyond NHEJ, possibly through a dose-sensitive interaction with dsRNA, which is 10~100 times weaker than with dsDNA. While investigating the essentiality of Ku in human cells, we found that depletion of Ku - unlike LIG4 - induces profound interferon (IFN) and NF-kB signaling via dsRNA-sensor MDA5/RIG-I and adaptor MAVS. Prolonged Ku-degradation also activates other dsRNA-sensors, e.g. PKR that suppresses protein translation, and OAS/RNaseL that cleaves rRNAs and eventually induces growth arrest and cell death. MAVS, RIG-I, or MDA5 knockouts suppressed IFN signaling and, like PKR knockouts, all partially rescued Ku-depleted human cells. Ku-irCLIP analyses revealed that Ku binds to diverse dsRNA, predominantly stem-loops in primate-specific Alu elements at anti-sense orientation in introns and 3'-UTRs. Ku expression rose sharply in higher primates tightly correlating with Alu-expansion (r = 0.94/0.95). Together, our study identified a vital role of Ku in accommodating Alu-expansion in primates by mitigating a dsRNA-induced innate immune response, explaining the rise of Ku levels and its essentiality in human cells. Overall design: To investigate the function of Ku above NHEJ in human cells, we generated AID-tagged Ku80 HCT116 cell lines (3 independent clones, #1, #7, #9) and MAVS knockout clones (3 independent clones, #1, #13, #15) on the base of AID-Ku80. To confirm our results in different Ku80 depletion system or cell lines, we used the Ku86flox/- HCT116 cells with one deleted and one floxed Ku80 allele (sometimes referred to as Ku86 in human cells) to allow the 4-hydroxy tamoxifen (4OHT)-induced CreERT2 translocation to excise the remaining Ku80 allele (Ku86flox/-:CreERT2), and a recently published Ku70-KO HEK293 cells with Dox-inducible ectopic Ku70 (3 independent clones, Sa11, SB and TI). We treated the AID cells with IAA&Dox to degrade Ku80 with DMSO as the control, flox cells with 4OHT to deplete Ku80 with EtOH as the control, and HEK293 cells without Dox to deplete Ku70 with Dox as the control. We then performed gene expression profiling analysis using data obtained from RNA-seq of the independent clones (AID HCT116 cells and HEK293 cells) or samples from two independent experiments (Ku86flox/- and Ku86flox/-:CreERT2 cells).