Casein kinase II promotes target silencing by miRISC through direct phosphorylation of the DEAD-box RNA helicase CGH-1

Abstract

MicroRNAs (miRNAs) play essential, conserved roles in diverse developmental processes through association with the miRNA-induced silencing complex (miRISC). Whereas fundamental insights into the mechanistic framework of miRNA biogenesis and target gene silencing have been established, posttranslational modifications that affect miRISC function are less well understood. Here we report that the conserved serine/threonine kinase, casein kinase II (CK2), promotes miRISC function in Caenorhabditis elegans. CK2 inactivation results in developmental defects that phenocopy loss of miRISC cofactors and enhances the loss of miRNA function in diverse cellular contexts. Whereas CK2 is dispensable for miRNA biogenesis and the stability of miRISC cofactors, it is required for efficient miRISC target mRNA binding and silencing. Importantly, we identify the conserved DEAD-box RNA helicase, CGH-1/DDX6, as a key CK2 substrate within miRISC and demonstrate phosphorylation of a conserved N-terminal serine is required for CGH-1 function in the miRNA pathway.

Keywords: CGH-1; casein kinase II; miRISC; microRNA; phosphorylation.

Fig. 1.

CK2 genetically interacts with the miRNA pathway. (A and B) CK2 depletion results in phenotypes associated with reduced miRNA pathway function. (A) RNAi of kin-3, kin-10, or alg-1 significantly increases defective alae versus empty vector control (two-tailed Fisher’s exact test P < 0.0001). (B) RNAi of kin-3, kin-10, nhl-2, alg-1, and ain-1 in animals expressing a seam cell gfp reporter (Pscm::gfp) all exhibit significant seam cell hyperplasia post-L4 versus empty vector control (two-tailed Student’s t test *P < 0.001, mean and SD plotted, n = 50). (C–J) CK2 depletion enhances miRNA mutant defects. (C) RNAi of kin-3, kin-10, nhl-2, alg-1, and ain-1 all exhibit significant enhancement of adult animals in lethargus versus empty vector control (two-tailed Student’s t test of biological replicates *P < 0.05, mean and SD plotted, n ≥ 50 per replicate). Adult animals inappropriately entering lethargus were scored every 2 h from 60 h to 72 h post-L1. (D) kin-3 and kin-10 RNAi enhance Rup of let-7, let-7 family, and miRISC factor mutants at 72 h post-L1 (20 °C). Heatmap represents mean percent Rup of biological replicates (n ≥ 50 per replicate). (E) Rup in homozygous kin-10 deletion mutants is significantly greater than hT2[qIs48] balanced kin-10/+ siblings (two-tailed Fisher’s exact test P ≤ 0.003). (F and G) kin-3 and kin-10 RNAi Rup enhancement in let-7 and mir-48 mutants is dependent on their targets. (F) Rup enhancement of kin-3 and kin-10 RNAi in let-7(mg279) is decreased in the lin-41(ma104) background (mean and SD of biological replicates plotted, n ≥ 37 per replicate). (G) Rup enhancement of kin-3 and kin-10 RNAi in mir-48(n4097) is decreased in the hbl-1(mg285) background (mean and SD of biological replicates, n ≥ 56 per replicate). (H) RNAi of kin-3, kin-10, nhl-2, and alg-1 all significantly enhance Muv versus empty vector control (two-tailed Student’s t test of biological replicates *P < 0.05, mean and SD plotted, n ≥ 94 per replicate). (I) kin-3 and kin-10 RNAi increase the penetrance of ASEL misspecification in neuronal RNAi-competent lsy-6(ot150) strain indicated by the lack of Plim-6::gfp expression in ASEL (Fisher’s exact test P < 0.0001 for kin-3 and alg-1 RNAi, P = 0.0002 for kin-10 RNAi). (J) kin-3 and kin-10 RNAi enhance embryonic lethality in mir-35-41(nDf50) (two-tailed Fisher’s exact test *P < 0.001, n ≥ 145).

Fig. 2.

CK2 is required for miRNA target silencing. (A) Abridged table of miRNAs and their target mRNAs examined in this study. (B) kin-3 and kin-10 RNAi attenuate silencing of let-7 family targets lin-41 and daf-12: lin-41 mRNA levels are significantly elevated in kin-3 and kin-10 RNAi versus vector RNAi (44 h post-L1, 20 °C) (Left); daf-12 mRNA levels are significantly elevated in kin-3 and kin-10 RNAi versus vector RNAi (40 h post-L1, 20 °C) (Right) (one-tailed Student’s t test of two biological replicates for lin-41 and three biological replicates for daf-12, *P < 0.050, mean and SD plotted). (C) kin-3 and kin-10 RNAi attenuates silencing of lin-4 target lin-14. LIN-14 is up-regulated in kin-3 and kin-10 RNAi versus empty vector control in L2 (16 h to 20 h post-L1). γ-Tubulin was used as a loading control. (D and E) kin-3 and kin-10 RNAi attenuate silencing of miR-1 target mef-2 in a miR-1–dependent manner. (D) Reporter constructs shown with wild-type and scrambled miR-1 sites (Left). Arrows indicate vulva in representative images (Right). (E) mef-2 reporter quantification. RNAi of kin-3, kin-10, and alg-1 all significantly increase GFP signal of the wild-type reporter versus vector (P < 0.05, mean and SD plotted, n > 10), but not the scrambled reporter (P ≥ 0.60).

Fig. 3.

CK2 promotes miRISC target binding. (A and B) CK2 does not affect miRNA levels. (A) kin-3 RNAi does not affect global mature miRNA abundances as quantified by deep sequencing [reads per million (RPM) mapped reads]. (B) Global analysis in A is supported by Northern blotting: kin-3 and kin-10 RNAi do not substantially alter levels of precursor or mature miR-48 in wild-type or alg-1(tm369) animals. (C) CK2 RNAi does not considerably affect miRISC factor levels. Western analysis of core miRISC proteins in kin-3 and kin-10 RNAi are similar to empty vector control. (D–F) CK2 affects miRISC binding to target mRNAs. (D) Schematic and Western analysis of GFP::ALG-1 RNA immunoprecipitation (RIP) from L4 (48 h post-L1, 20 °C) lysates of empty vector versus kin-3 and kin-10 RNAi; gfp RNAi controls for RIP; γ-tubulin controls for protein input. (E) Levels of let-7 and miR-48 associated with GFP::ALG-1 are not significantly different in kin-3 and kin-10 RNAi versus empty vector. (F) Levels of lin-41 and daf-12 mRNA associated with GFP::ALG-1 are decreased in kin-3 and kin-10 RNAi versus empty vector. RIP RNAs were normalized to spiked-in Firefly Luciferase mRNA (one-tailed Student’s t test of biological replicates *P < 0.05, mean and SD plotted).

Fig. 4.

CK2 phosphorylates miRISC factor CGH-1 at serine 2. (A) KIN-3 associates with CGH-1 in the soma. CK2 catalytic subunit, KIN-3, coimmunopurifies with CGH-1::GFP in L4 stage animals fed glp-1 RNAi to prevent germ-line proliferation. (B) CGH-1 is phosphorylated by CK2 in vitro. Autoradiogram of GST-purified wild-type CGH-1 incubated with GST-purified CK2 composed of wild-type or ATP-binding site mutant KIN-3 (K67M). TBB, CK2 inhibitor (4,5,6,7-tetrabromobenzotriazole). (C) CGH-1 harbors four CK2 recognition motifs (sites 1–4). (D) Site 1 (serine 2) is phosphorylated by CK2 in vitro. GST-tagged CGH-1 peptides composed of 20 residues flanking each putative CK2 phosphorylation site were incubated in vitro with CK2 and analyzed by SDS/PAGE and autoradiography. ³²P, autoradiogram; CB, Coomassie Blue stained gel. (E and F) Genetic analysis of CGH-1 serine 2 (S2) phosphovariants in cgh-1(tn691)ts; let-7(mg279) suggests phosphorylation of CGH-1 at S2 promotes miRISC function (E) Representative images of Protruding vulva (Pvul). Adult vulva indicated by arrow (120 h post-L1, 25 °C). (F) Quantitation of Pvul. Compared with cgh-1(tn691)ts; let-7(mg279) with no transgene the cgh-1::gfp (S2) transgene significantly rescues Pvul. S2A mutants (lines 1 and 2) significantly enhance Pvul (two-tailed Fisher’s exact test *P < 0.0001 for S2 variants versus “no TG,” mean and SD plotted for three technical replicates within a single experiment, n ≥ 95). (G) S2 and S2E significantly rescue alae defects of cgh-1(tn691)ts; let-7(mg279). S2A mutants significantly enhance (line 1) or have no significant effect (line 2) on defects (two-tailed Fisher’s exact test *P < 0.0001 for S2 variants versus “no TG”). TG, transgene; S2A lines 1 and 2, phosphodefective Ser-to-Ala mutation; S2D/E, phosphomimic Ser-to-Asp/Glu mutations.