GSK-3 is an RNA polymerase II phospho-CTD kinase

Abstract

We have previously found that UV-induced DNA damage causes hyperphosphorylation of the carboxy terminal domain (CTD) of RNA polymerase II (RNAPII), inhibition of transcriptional elongation and changes in alternative splicing (AS) due to kinetic coupling between transcription and splicing. In an unbiased search for protein kinases involved in the AS response to DNA damage, we have identified glycogen synthase kinase 3 (GSK-3) as an unforeseen participant. Unlike Cdk9 inhibition, GSK-3 inhibition only prevents CTD hyperphosphorylation triggered by UV but not basal phosphorylation. This effect is not due to differential degradation of the phospho-CTD isoforms and can be reproduced, at the AS level, by overexpression of a kinase-dead GSK-3 dominant negative mutant. GSK-3 inhibition abrogates both the reduction in RNAPII elongation and changes in AS elicited by UV. We show that GSK-3 phosphorylates the CTD in vitro, but preferentially when the substrate is previously phosphorylated, consistently with the requirement of a priming phosphorylation reported for GSK-3 efficacy. In line with a role for GSK-3 in the response to DNA damage, GSK-3 inhibition prevents UV-induced apoptosis. In summary, we uncover a novel role for a widely studied kinase in key steps of eukaryotic transcription and pre-mRNA processing.

Figure 1.

An alternative splicing fluorescent reporter system to study the transcriptional response to UV-induced DNA damage. (A) Diagram of the enhanced alternative splicing fluorescent reporter used to stably transfect HeLa cells by the Flp-In techonology (HeLa eRGi cells). Inclusion of the 28 nt alternative exon (cTNT E5) results in an mRNA that is translated to GFP, while the mRNA resulting from its skipping is translated to dsRed*. (B) HeLa eRGi cells were UV irradiated (or not) with 15 J/m² and tetracycline was immediately added to induce the transcription of the reporter. Three hours after irradiation, total RNA was prepared and AS patterns of the alternative splicing fluorescent minigene were assessed by radioactive RT-PCR with specific primers. The ratio inclusion/skipping is shown. (C) Western blot of the reporter-encoded proteins was performed with an anti-Flag antibody. HeLa eRGi cells were treated as in (B) and harvested in Laemmli buffer 24 hours after irradiation. (D) Flow cytometry of HeLa eRGi cells treated as in (B), with different doses of UV light, 24 h after irradiation. The left panel is a histogram of GFP/dsRed* fluorescence signal at each UV dose. The right panel shows the quantification of the mean GFP/dsRed* signal of 3 independent experiments. The r² of the linear regression is shown. (E) Epifluorescence microscopy representative fields of HeLa eRGi cells treated as in (B), 24 hours after irradiation. The figure shows GFP signal, dsRed* signal and their merge. (F) HeLa eRGi cells were transfected with WT, A2A5 and C4 α-amanitin resistant Rpb1 expression constructs. Twenty-four hours after transfection, α-amanitin was added at 100 μg/ml to induce the degradation of the endogenous Rpb1. Twenty-four hours later, cells were treated as in (B). Three hours after irradiation, total RNA was prepared and AS patterns were assessed by radioactive RT-PCR. (G) HeLa cells stably expressing the marsupial CPD photolyase were transfected with the alternative splicing fluorescent reporter construct shown in (A). Cells were treated as in (B) followed by 2 h under white light (or not as a control) to induce the repair of the CPDs. Three hours after irradiation, total RNA was prepared and AS patterns were assessed by radioactive RT-PCR.

Figure 2.

Screening of protein kinase inhibitors unveils GSK-3 as a kinase involved in the transcriptional response to UV-induced DNA damage (A) Hits resulting from the screening of the PKIS2 library with the alternative splicing fluorescent reporter. (B) Structure of the commercial selective GSK-3 inhibitors used. (C–E) HeLa cells were UV irradiated (or not) with 15 J/m² and the inhibitors GW8057578X, GW806290, CHIR, AR-A or DMSO as vehicle control were added immediately. Three hours after irradiation, total RNA was prepared and AS patterns of the alternative exons E2 of TMEM188 (C), E9a of EED (D) and E2a of TBX3 (E) were assessed by radioactive RT-PCR with specific primers.

Figure 3.

GSK-3 is an RNAPII phospho-CTD kinase and its activity is necessary for RNAPII hyperphosphorylation and AS response after UV-induced DNA damage (A) HeLa cells were UV irradiated (or not) with 15 J/m² and the inhibitors GW8057578X, GW806290, CHIR, AR-A or DMSO as vehicle control were added immediately. For DRB treatment, cells were pretreated for 2 h with DRB before UV irradiation. Two hours after irradiation, cells were harvested in Laemmli buffer and RNAPII phosphorylation patterns were assessed by western blot of total Rpb1. (B) HeLa cells were pre-treated (or not) with the proteasome inhibitor MG132 for 1 hour prior to irradiation. Subsequent treatment and sample analysis were performed as in (A). (C) Diagram of GSK-3 consensus site and putative GSK-3 target sites in a diheptad of the RNAPII CTD depending on which residue is previously phosphorylated. (D) Recombinant GST-CTD purified from bacteria was incubated with Cdk9 immunopurified from HEK293T cells in the presence of ATP. Cdk9 with no substrate and GST-CTD with no kinase were incubated in the presence of ATP as a control. After thirty minutes (or one hour), kinase assays were stopped by addition of Laemmli buffer and samples were analyzed by western blot with antibodies raised against CTD P-Ser2, CTD P-Ser5, GST and Cdk9. (E) GSK-3β was immunopurified from HEK293T cells. The kinase was incubated in the presence of ATP with GST-CTD purified from bacteria or with GST-CTD pre-phosphorylated (P-CTD) by Cdk9 as in (C). GSK-3 with no substrate was incubated in the presence of ATP as a control. Additionally, GSK-3 was incubated with (P-CTD) in the presence of ATP and CHIR. After one hour, kinase assays were stopped by addition of Laemmli buffer and samples were analyzed by western blot with antibodies raised against CTD P-Ser2, GST and GSK-3β. The diagram below depicts the experiment outcome. (F) HA-tagged WT or kinase dead (KD) GSK-3β were transfected in HEK293T cells and immunopurified with anti HA antibody. GST-CTD purified from bacteria or P-CTD were incubated with WT GSK-3β or KD GSK-3β in the presence of ATP. After one hour, kinase assays were stopped by addition of Laemmli buffer and samples were analyzed by western blot with antibodies raised against CTD P-Ser2, GST and HA. (G) HeLa GSK-3α null cells were co-transfected with WT GSK-3β (GSK-3β WT) or the kinase-dead dominant negative K85A GSK-3β mutant (GSK-3β KD) and a TMEM188 exon E2 AS reporter minigene. Cells were UV irradiated (or not) with 15 J/m² and tetracycline was added immediately to induce the AS reporter. Three hours after irradiation, total RNA was prepared and AS patterns of the minigene were assessed by radioactive RT-PCR with specific primers. (H) HeLa GSK-3α null cells were treated as in G, but in this case co-transfecting a TBX3 E2a AS reporter minigene.

Figure 4.

GSK-3 inhibition prevents the UV-induced reduction in RNAPII elongation rates (A) HeLa cells were UV irradiated (or not) with 15 J/m² and the inhibitor CHIR or DMSO as vehicle control were added immediately. Two hours later, cells were harvested for RNAPII ChIP. RNAPII density was measured along the TBX3 model gene with the amplicons shown in the diagram. (B) HeLa cells stably transfected with the marsupial CPDs photolyase were UV irradiated (or not) and incubated 2 h exposed to white light (or not) to activate the CPD photolyases. Three hours after irradiation, total RNA was prepared and the quantity of long and total mRNA isoforms of histones H3B and H2AG were determined by RT-qPCR. The ratio long/total isoform is shown. (C) Quantification of histone H3B and histone H2AG long/total mRNA isoforms in response to UV-induced DNA damage in cells treated or not with the inhibitors CHIR and AR-A, three hours after irradiation.

Figure 5.

GSK-3 inhibition prevents UV-induced apoptosis (A–D) HeLa cells were UV irradiated with 30 J/m² to induce apoptosis, and immediately treated (or not) with CHIR. Twenty-four hours after UV irradiation, cells were stained with propidium iodide and FITC-annexin V to identify live, early apoptotic, late apoptotic and necrotic cells.