Mechanism-based screen establishes signalling framework for DNA damage-associated G1 checkpoint response

Abstract

DNA damage activates checkpoint controls which block progression of cells through the division cycle. Several different checkpoints exist that control transit at different positions in the cell cycle. A role for checkpoint activation in providing resistance of cells to genotoxic anticancer therapy, including chemotherapy and ionizing radiation, is widely recognized. Although the core molecular functions that execute different damage activated checkpoints are known, the signals that control checkpoint activation are far from understood. We used a kinome-spanning RNA interference screen to delineate signalling required for radiation-mediated retinoblastoma protein activation, the recognized executor of G(1) checkpoint control. Our results corroborate the involvement of the p53 tumour suppressor (TP53) and its downstream targets p21(CIP1/WAF1) but infer lack of involvement of canonical double strand break (DSB) recognition known for its role in activating TP53 in damaged cells. Instead our results predict signalling involving the known TP53 phosphorylating kinase PRPK/TP53RK and the JNK/p38MAPK activating kinase STK4/MST1, both hitherto unrecognised for their contribution to DNA damage G1 checkpoint signalling. Our results further predict a network topology whereby induction of p21(CIP1/WAF1) is required but not sufficient to elicit checkpoint activation. Our experiments document a role of the kinases identified in radiation protection proposing their pharmacological inhibition as a potential strategy to increase radiation sensitivity in proliferating cancer cells.

Figure 1. siRNA screen for gene products involved in IR-associated RB1 activation.

A) Loss of RB1 phosporylation in IR exposed cells. HCT116 cells exposed to 5 Gy ionising radiation (IR) or untreated (C) were analysed at the time indicated. Levels of RB1 with phosphorylation on Ser608 (RB1-PS608), total RB1 (RB1) and TP53 were established by immunoblotting. B) Loss of RB1 phosphorylation is TP53-dependent. TP53 positive (+/+) and isogenic TP53 null (−/−) HCT116 cells exposed to 5 Gy ionizing radiation (IR) or untreated (C) were analysed 24 hours post irradiation. Levels of RB1-PS608, total RB1 and p21^CIP1/WAF1 were established by immunoblotting. C) IR affects RB1 phosphorylation at multiple sites. Immunoblots probing for levels of RB1-PS608, RB1-PS780, RB1-PS795 in irradiated (IR) or untreated (C) HCT116. To document RB1 specificity of the signal cells transfected with siRNA duplexes targeting RB1 (RB(1) and RB(2)) or nontargeting control siRNA (NT) were analysed in parallel. Cells were irradiated and harvested at 24 hours following IR. Actin was used as a loading control. D) siRNA screening strategy. HCT116 were reverse transfected with siRNA library pools in a 96 well format, and irradiated, fixed and stained using anti RB1-PS780 antibody and Hoechst 33342 dye, with timelines as indicated. Plates were analysed using an IN Cell Analyzer 3000 high content platform (GE) with sequential blue and green laser excitation. A set number of cell objects per well were analyzed for nucleus-associated antibody fluorescence (green channel). Hoechst 3342 DNA staining (blue channel) was used for object and compartment identification. Intensity profiles were generated and automatically gated to determine the percentage of cells with sub-normal antibody fluorescence (POS-LoRBPS780) in individual wells. E) Radio-resistant RB1 phosphorylation in cells with siRNA-mediated TP53-signalling knockdown. Assay set up was as described in D, siRNA pools for TP53, p21^CIP1/WAF1 or a non-targeting oligonucleotide (nt) were used for transfection. Error bars relate to variance in POS-LoRBPS780 values from triplicate wells. F) Primary screen outcome. Z-score distribution for target screened. Z-scores were calculated for the mean POS-LoRBPS780 observed in triplicate wells and are plotted in ranked order. Hits are shown colour-coded according to hit class within the Z-score distribution.

Figure 2. Hit gene-ontology and pathway associations.

A) Pathway representation within hit pool. Hits were analysed for pathway association using the DAVID functional annotation tool (http://david.abcc.ncifcrf.gov/). B) Enrichment for gene ontology. Pathway association was analysed for hits and input using DAVID. Pathway representation within hits is plotted against that for input targets. C) Hit validation. Hits were assessed using individual oligonucleotides represented within the pool. The number of active oligonucleotides and level of response is indicated. Hit classification was as for the screen.

Figure 3. Effect of target knockdown on IR-mediated p21^CIP1/WAF1 induction.

A) Effect of target knockdown on p21^CIP1/WAF1 positivity. HCT116 cells transfected with siRNA as indicated were irradiated (IR) or left untreated (control). Cells were assessed for p21^CIP1/WAF1 positivity 16 hrs post IR. The percentage of cells with p21^CIP1/WAF1 positivity relative to Mock-treatment (Lipid) is shown. Error bars represent the variance of the mean of three biological replicates, run in triplicate. B) Modulation of RB1 phosphorylation by target knockdown. POS-LoRBPS780 analysis was performed in parallel plates. Data points represent the means of triplicate technical replicates and are evaluated using hit classification as for the screen. C) Statistical analysis. Paired t-tests results for data shown in A. D) Treatment interaction test. Targets that yielded significant impairment of p21^CIP1/WAF1 positivity were tested for evidence of interaction between radiation and target knockdown. Values indicate the degree of antagonism experienced in IR exposed cells.

Figure 4. Effect of target knock down on G1 checkpoint activation.

A) Effect of target knockdown on relative G1 positivity. HCT116 cells transfected with siRNA as indicated were irradiated (IR) or left untreated (control). Cells were fixed 16 hours later and assessed for the proportion of cells in G1. The degree of G1 positivity relative to Mock-treated (Lipid) cells is shown. Error bars represent the variance from the mean of three biological replicates, run in triplicate. B) Modulation of RB1 phosphorylation by target knockdown. POS-LoRBPS780 analysis performed in parallel to A). Data points represent the means of triplicate technical replicates. C) Statistical analysis. Paired t-tests for data shown in A. D) Treatment interaction test. Data were assessed for evidence of a interaction between radiation and target knockdown. Values indicate the degree of antagonism experienced in IR exposed cells.

Figure 5. Effect of target knockdown on radiation survival.

A–I) Target knockdown affects survival of IR exposed cells. HCT116 cells transfected with target siRNA alone or in combination with siRNA targeting CHK1. Cells were Mock-irradiated or irradiated with 2 Gy or 5 Gy and viable cells quantified 5 days later. Data plotted are normalized to the respective untreated controls. Filled squares = combined target and CHK1 knockdown (Target/CHK1), open squares = CHK1 only (Li/CHK1), filled triangles = Target only (Target/NT), open triangles = Mock (Li/NT). Error bars represent the variance from the mean from three technical replicates. K) Modulation of RB1 phosphorylation by target knockdown. Parallel POS-LoRBPS780 analysis was used to verify siRNA performance. L) Statistical analysis. Student t-test for cell viability data shown in A–I, (Li/NT vs Targ/NT) probing for a significant effect of target knockdown in unperturbed cells, (Li/CHK1 vs Targ/CHK1) probing for a significant effect of target knockdown in CHK1-perturbed cells, Target/NT vs Target/CHK1 probing for a significant effect of CHK1 knockdown in target-perturbed cells. M, N) Treatment interaction. Assessment for evidence of interaction between radiation and target knockdown. Values represent the degree of net synergism between target knockdown and IR in NT (M) or CHK1-perturbed (N) cell background.

Figure 6. Data consolidation.

A) Data clustering analysis. Unsupervised clustering based on numerical observations. B) Signalling model. Hits split into groups according to their role in the accumulation of the TP53 target p21^CIP1/WAF1. Canonical double strand signalling components (ATM, ATR, CHK1/2) affecting the TP53/p21^CIP1/WAF1 axes in conjunction with S/G2 checkpoint activation do not affect G1 checkpoint function.