CRISPR-Cas9 screening identifies genes whose deficiency sensitizes pancreatic cancer cells to chemotherapy.
Currently, weak treatment response represents a crucial factor in restricting the survival rate of pancreatic cancer. To explore new mediators of chemotherapy sensitivity, we performed CRISPR-Cas9 knockout screening in pancreatic cancer cell line TB32047 derived from the KPC mouse model. We transduced TB32047-cas9 cells with a protein kinase library with additional sgRNAs enriched for PDAC associated genes containing 3438 small-guide RNAs (sgRNA) that target 920 murine genes and 110 non-targeting control sgRNAs. Transfected cells were selected by puromycin treatment, followed by continuous exposure to vehicle control or gemcitabine/paclitaxel chemotherapy for 21 days (Fig. 1A). We experimentally determined the optimal concentration of the drugs in TB32047 cells (Supplementary Fig. S1A and S1B). Continuously treatment of transfected TB32047 cells with gemcitabine at IC90 and paclitaxel at IC70 resulted in a strong cell proliferation inhibition and cell death. Subsequently, genomic DNA was isolated from transfected and puromycin-selected cells treated with vehicle control, gemcitabine or paclitaxel, and referred to as TB-C3W and TB-G3W/TB-P3W, respectively. sgRNAs were quantified by next-generation sequencing (Supplementary Fig. S1C, Supplementary Table S2).
We then analyzed the sgRNA counts of each group and scored genes targeted by sgRNA in TB-C3W versus TB-G3W/TB-P3W. We prioritized targets according to commonalities among sgRNAs, meaning that genes were selected which had multiple sgRNAs significantly reduced in TB-C3W compared with TB-G3W/TB-P3W. We ranked individual sgRNAs based on log2 fold-change (FC)< -1 and adjusted P< 0.05, and finally identified 33 genes as candidates in gemcitabine drug screening (Fig. 1B; Supplementary Table S4), while 47 genes were determined in paclitaxel drug screening (Fig. 1C; Supplementary Table S4). Gene ontology (GO) terms biological process analysis (DAVID Bioinformatics Resources 6.8) indicated that these genes were involved in cell division, protein phosphorylation, cell cycle, mitotic nuclear division, DNA repair, apoptotic process, and cellular response to DNA damage stimulus (Supplementary Fig. S1D and S1E).
We took the intersection of the top 20 genes in the two drug screenings, with a total of four identical gene hits, namely Cdk7, Wee1, Fgfrl1 and Havcr2 (Fig. 1D), whose loss of function by specific sgRNAs led to lethality in presence of gemcitabine and paclitaxel. Furthermore, Cdk7 was took second and first place in screening for gemcitabine and paclitaxel, respectively (Supplementary Table S4). Interestingly, CDK7 is involved in a wide spectrum of biological processes, including cell-cycle regulation and DNA damage response . The majority of CDK7-targeting sgRNA standardized read counts were considerably decreased following treatment with gemcitabine and paclitaxel (Fig. 1E). Data from TCGA-PDAC indicated that increased CDK7 expression is statistically associated with poor overall survival in PDAC, and CDK7 is upregulated in pancreatic cancer as compared to normal tissues (Fig. 1F and G). Furthermore, CDK7 is expressed in PC cell lines of different subtypes (Fig. 1H). Therefore, we focused our subsequent analyses on the role of CDK7 in the cellular response to chemotherapy.
Knockout of CDK7 combined with gemcitabine and paclitaxel chemotherapy enhances the antiproliferative effects and apoptosis in pancreatic cancer cells.
To confirm CDK7 as a potential therapeutic target for enhancing the impact of chemotherapy, we produced CDK7 stable knockout (KO) subclones by transfecting TB32047 and MIA PaCa-2 cells with four sgRNAs (Fig. 2A; Supplementary Table S4). Western blotting confirmed that sgRNA1, sgRNA2, sgRNA4 sequences in TB32047 and sgRNA2, sgRNA3, sgRNA4 sequences in MIA PaCa-2 were able to effectively knockout CDK7. Therefore, we selected these three subclones for follow-up experiments (Fig. 2A). Dose-dependent viability assays and clonogenic assays indicated that CDK7-deficient cell lines were significantly more sensitive to both GEM and PTX than cells transduced with scrambled control sgRNA (NC) (Fig. 2B; Supplementary Fig. S2A). Consistently, CDK7 knockout also dramatically induced apoptosis in the presence of GEM and PTX (Fig. 2C-F). Flow cytometry-based apoptotic analysis revealed that in the presence of chemotherapy, the early and late apoptosis of CDK7 knockout cells were significantly enhanced compared with non-targeting control (NC). Additionally, CDK7 knockout cell lines demonstrated higher caspase 3/7 activity (Fig. 2G and H), suggesting that CDK7 knockout induced more considerable cell damage and led to more apoptosis in GEM and PTX treated cells. Interestingly our results showed that CDK7 knockout also slightly increased cell apoptosis without chemotherapy (Fig. 2D and F), which might be related to the role of CDK7 in transcriptional control of DNA repair networks [28, 29].
Following that, we investigated the mechanism through which CDK7 deficiency results in chemosensitivity. CDK7 is required for RNAPII-mediated transcription and also functions as an integral component of the transcription factor TFIIH, which is involved in transcription initiation and DNA repair [28, 30]. Therefore, we performed western blot analysis of apoptosis and DNA repair associated proteins in TB32047 and MIA PaCa-2 KO cell lines treated with vehicle or GEM and PTX. Interestingly, we observed a substantial down-regulation of phosphorylated STAT3 protein expression in the absence of CDK7, which was more pronounced in the presence of GEM and PTX (Fig. 2I). The phosphorylation level of STAT3 in CDK7 knockout cells decreased more drastically after chemotherapy. Meanwhile, we also detected a decrease in the protein level of MCL1 after CDK7 deletion. Similarly, we determined the amount of expression of CHK1, an MCL1 regulatory target involved in cell cycle control and DNA damage repair. CHK1 consistently decreased in the presence of CDK7 knockout, and CHK1 was significantly decreased following 24 hours therapy. According to the foregoing findings, inactivating CDK7 sensitizes PDAC cells to GEM and PTX chemotherapy via the STAT3-MCL1-CHK1 axis (Fig. 6C).
Targeted inhibition of CDK7 enhanced gemcitabine and paclitaxel chemotherapy response in pancreatic cancer in vitro.
Next, we sought to confirm this effect by using the CDK7-specific inhibitor THZ1. Dose-response curves performed on a panel of human PDAC cell lines (MIA PaCa-2, AsPC-1, SUIT-2, BxPC-3, Mayo4636, TKCC-10) and TB32047 indicated that THZ1 inhibited PDAC cell viability in a dose-dependent manner, with the IC50 ranging from 26.08nM to 423.7nM (Fig. 3A; Supplementary Fig. S2B).
To analyze whether CDK7 inhibitor could work synergistically with GEM and PTX in suppressing PDAC growth, we used THZ1 at a dose inhibiting proliferation by less than 20% (IC20) in the aforesaid PDAC cells in conjunction with a concentration gradient of GEM and PTX for 72 hours. As expected, GEM and PTX reduced PDAC cell viability in a dose-dependent manner, while the combination of THZ1 with GEM/PTX generated a substantial synergistic effect (CI < 1), resulting in a more severe reduction of cell viability than the GEM/PTX regimen (Fig. 3B and C; Supplementary Fig. S2C-S2G). Clonogenic assays revealed that the combination treatment group had considerably fewer colonies than groups treated with THZ1 or GEM/PTX alone (Fig. 3D and F). According to the caspase 3/7 activity assay, the combined treatment group exhibited significantly higher caspase 3/7 activity than either chemotherapy or THZ1 alone, suggesting that the cytotoxicity generated by combination treatment was more effective (Fig. 3E and G). Further, the flow cytometry analysis confirmed that the combination treatment was superior to single agents, manifested by a more pronounced activation of apoptotic cell death induced by combined THZ1 and GEM/PTX (Fig. 3H and I; Supplementary Fig. S2H). In conclusion, these results confirmed that the targeted inhibition of CDK7 amplifies the toxicity of chemotherapy, rendering PDAC cells more vulnerable to GEM/PTX chemotherapy.
Combined CDK7 inhibition with standard chemotherapy suppresses PDAC tumor growth in vivo.
Next, we evaluated the in vivo efficacy of CDK7 inhibitor combined with GEM and nab-PTX chemotherapy. To establish an orthotopic mouse model of pancreatic cancer, we used TB32047 organoids. Tumor-bearing mice were treated with GEM (50mg/kg) and nab-PTX (5mg/kg) once every 4 days and THZ1 (5mg/kg) 5 consecutive days a week. Notably, the dose of chemotherapy and THZ1 used in our study were below clinically achievable concentrations. As a consequence, none of the combinations caused substantial weight loss in mice, and no additional toxicity was observed during the therapy (Supplementary Fig. S2I).
Treatment with GEM and nab-PTX slightly reduced the tumor volume, which confirmed the effectiveness of the chemotherapy regimen. Although THZ1 alone had no effect on tumor volume, the combination treatment considerably increased chemotherapy effectiveness in an orthotopic pancreatic cancer mice model by significantly decreasing tumor size (Fig. 4A), volume and weight (Fig. 4B and C) when compared to the vehicle group. Hence, the combination of CDK7 inhibitor together with the standard care GEM + nab-PTX exhibited a superior anti-tumorigenic activity compared to chemotherapy alone. The high synergistic impact of THZ1 was consistent with our in vitro results (Fig. 2 and Fig. 3). We performed western blots to evaluate the expression levels of cell cycle and apoptosis-related proteins in tumor samples from different treatment groups to confirm the better efficacy of the combination therapy. Combination treatment for PDAC tumors resulted in lower WEE1 and CHK1 expression than monotherapy. The expression of MCL1 in PDAC tumors was reduced following the combination of treatments, suggesting an increase in apoptosis (Fig. 4D). Additionally, hematoxylin and eosin (H&E) staining demonstrated that the combination of THZ1 with GEM and nab-PTX dramatically decreased the tumor cell content compared to the single components and control group (Fig. 4E). Briefly summarized, our in vivo results validate that combined CDK7 inhibition is an effective strategy to promote standardized treatment and overcome chemoresistance of PDAC.
CDK7 inhibition induces cell-cycle arrest, apoptotic cell death and DNA damage through the STAT3-MCL1-CHK1 axis.
To investigate the potential molecular mechanism of chemotherapy and CDK7 inhibition in combination, we first interrogated the Gene Expression Omnibus (GEO) datasets and performed gene set enrichment analysis (GSEA) on the RNA-sequencing data of BXPC-3, MIA PaCa-2, and PANC-1 cells treated with THZ1. Mechanistic investigations demonstrated that CDK7 inhibition significantly decreased gene transcription, with a preference for mitotic cell cycle and NF-κB signaling-related transcripts preferentially repressed . Simultaneously, GSEA analysis revealed that THZ1 substantially decreased the gene features of DNA damage repair (DDR), the G2-M checkpoint, and regulate apoptosis (Fig. 5A). Notably, CDK7 is also strongly positively correlated with the expression of key genes involved in cell cycle and DNA repair through the analysis of TCGA PDAC patient transcriptomic data, in particular CDC25C, CCNB1, CDC25B, CDK1, NHEJ1, CHK1 and RAD51, with Pearson's correlation coefficients being 0.42, 0.51, 0.31, 0.41, 0.48, 0.24 and 0.24, respectively (Supplementary Fig. S3A).
To demonstrate that THZ1 therapy resulted in significant alterations in cell cycle-related proteins, TB32047 and MIA PaCa-2 cells were treated with THZ1 in concentration gradient and time gradient to ascertain the effects of CDK7 inhibition on cell cycle regulation. Importantly, p-CTDSer7 progressively decreased in a time- and dose-dependent manner, which was related to the role of CDK7 in RNAPII-mediated transcription and confirmed the effectiveness of THZ1. Meanwhile, WEE1, phosphorylated WEE1, phosphorylated CDK1, and phosphorylated CDK2 all reduced dose- and time-dependently. Simultaneously, when the dose or time was increased, the cleaved PARP was up-regulated, indicating that THZ1 also regulated apoptosis in a time and dose-dependent manner (Supplementary Fig. S3C and S3D). Through cell cycle analyses, we revealed increased G2-M arrest in a dose-dependent way in the aforementioned two cells (Supplementary Fig. S3E and S3F).
We have established that deletion of CDK7 results in cell cycle arrest and increased apoptosis through the STAT3-MCL1-CHK1 pathway, rendering PDAC susceptible to GEM/PTX therapy. As a result, we were interested in determining if inhibiting CDK7 may improve chemotherapeutic effectiveness through the same mechanism. In order to determine the interaction between CDK7 and the above pathway proteins, we conducted endogenous co-immunoprecipitations and found that CDK7 was immunoprecipitated with STAT3, MCL1 and CHK1 in the reciprocal experiment in TB32047 cells, respectively (Supplementary Fig. S3B). Next, we treated TB32047 and MIA PaCa-2 cells for 24 hours with THZ1 at a dosage that has no impact on cell proliferation in conjunction with chemotherapy. Consistent with CDK7 deletion, targeted inhibition of CDK7 with GEM/PTX chemotherapy also decreased the amount of transcription factor STAT3 and its phosphorylation, despite the fact that a single medication had no effect on their expression. Following combination therapy, the cell cycle checkpoint protein CHK1 and its Ser296 phosphorylation site were also considerably suppressed, showing that the severity of cell cycle arrest and DNA damage increased. The level of the anti-apoptotic protein MCL1 was also considerably decreased, demonstrating that apoptosis was induced, which was compatible with the apoptosis experiment results (Fig. 5B). Immunofluorescence experiments confirmed that the combined treatment significantly increased the number of γH2AX foci compared to the THZ1 or GEM and PTX treatment group (Fig. 5C), revealing that the combined treatment resulted in the highest percentage of DNA double-strand breaks (DSB) that caused increased DNA damage and cell death.
In conclusion, our findings suggest that inhibiting CDK7 enhances the efficacy of standard chemotherapy in generating cell cycle arrest, DNA damage, and ultimately apoptosis in PDAC.
Targeted inhibition of CDK7 reversed chemoresistance in pancreatic cancer cells.
In the following step, we investigated whether CDK7 contributes to resistance to GEM and PTX. To simulate acquired resistance in patients, we produced gemcitabine- and paclitaxel-resistant cells (TBGPR and MIAGPR) by chronically treating TB32047 and MIA PaCa-2 cells with stepwise incremental doses of GEM and PTX. The dose response curve displayed a 5-10-fold increase in IC50 of chemoresistant cells compared with parental cells (Fig. 3B and C; Fig. 6B). It is sufficient to provide a model of drug resistance in refractory pancreatic cancer for subsequent studies.
Interestingly, CDK7, STAT3, P-STAT3Tyr705, CHK1, and MCL1 signal activation were clearly increased in chemoresistant cells compared to parental cells (Fig. 6A). Notably, consistent with the results of chemotherapy on parental cells, the addition of THZ1 restored GEM/PTX sensitivity in TBGPR and MIAGPR cells. The cell viability assay revealed a high synergistic effect between THZ1 and GEM/PTX in the two resistant cell lines (Fig. 6B). Surprisingly, chemoresistant cell lines demonstrated increased susceptibility to THZ1 (Supplementary Fig. S4A and S4B), mainly due to the persistently high CDK7 expression in TBGPR and MIAGPR. These results indicated that CDK7 contributes to chemoresistance and THZ1 abolished this process, rendering refractory PDAC cells susceptible to GEM/PTX chemotherapy (Fig. 6C).