WEE1 inhibition in pancreatic cancer cells is dependent on DNA repair status in a context dependent manner

Pancreatic ductal adenocarcinoma (PDA) is a lethal disease, in part, because of the lack of effective targeted therapeutic options. MK-1775 (also known as AZD1775), a mitotic inhibitor, has been demonstrated to enhance the anti-tumor effects of DNA damaging agents such as gemcitabine. We evaluated the efficacy of MK-1775 alone or in combination with DNA damaging agents (MMC or oxaliplatin) in PDA cell lines that are either DNA repair proficient (DDR-P) or deficient (DDR-D). PDA cell lines PL11, Hs 766T and Capan-1 harboring naturally selected mutations in DNA repair genes FANCC, FANCG and BRCA2 respectively, were less sensitive to MK-1775 as compared to two out of four representative DDR-P (MIA PaCa2 and PANC-1) cell lines. Accordingly, DDR-P cells exhibit reduced sensitivity to MK-1775 upon siRNA silencing of DNA repair genes, BRCA2 or FANCD2, compared to control cells. Only DDR-P cells showed increased apoptosis as a result of early mitotic entry and catastrophe compared to DDR-D cells. Taken together with other recently published reports, our results add another level of evidence that the efficacy of WEE1 inhibition is influenced by the DNA repair status of a cell and may also be dependent on the tumor type and model evaluated.


MK-1775 is more effective against DDR-proficient PDA cells compared to DDR-deficient
and Supplementary Tables S1 and S2), a short-term cell survival assay was performed with increasing concentrations of MK-1775 for 7 days. As a control, a non-transformed pancreatic cell line HPNE was also included in the analysis ( Supplementary Fig. S1A). Hs 766T and PL11 cells, defective in FANCG and FANCC respectively 36 , were less sensitive to MK-1775 compared to the DDR-proficient (DDR-P) cell lines MIA PaCa2 and PANC-1 ( Fig. 1A and Table 1). Capan-1 cells, which harbor a BRCA2 mutation 41 , were more sensitive (2.2 fold change) to MK-1775 compared to Hs 766T cells ( Fig. 1A and Table 1), but consistently more resistant (4.3 and 1.8 fold change) compared to the MIA PaCa2 and PANC-1 cell lines, respectively. Surprisingly, HPNE was sensitive to MK-1775 similar to DDR-P cell lines MIA PaCa2 and PANC-1 ( Supplementary Fig. S1A and Supplementary Table S1). Of note, SU.86.86 and BxPC3 cells that are DNA repair-proficient were also resistant to MK-1775 (Fig. 1A, Table 1 and Supplementary Table S2). Wang et al. also showed that BxPC3 are similar in sensitivity to MK-1775 compared to MIA PaCa2 and PANC-1 cells 15 Table S3). In line with our overall observations, neither did a pre-treatment with low dose of MMC sensitize SU.86.86 and BxPC3 cells to MK-1775, nor a pre-treatment with a low dose of MK-1775  Fig. 1A,B). In lieu of the above results, we performed subsequent experiments in five selected, representative PDA cells lines namely MIA PaCa2 and PANC-1 (DNA repair proficient, DDR-P) and Capan-1, Hs 766T and PL11 (DNA repair deficient, DDR-D) ( Table 1 and Supplementary Table S2), all of which also have deficient p53 status. Based on FA biology and the sequence of the signaling cascade, FANCD2 foci are not expected in the FANCC (cell line PL11) and FANCG (cell line Hs 766T) deficient cells, but should be detectable in FA proficient (MIA PaCa2 and PANC-1) and BRCA2 deficient cells (Capan-1) 42 . To confirm the integrity of our DDR-deficient PDA lines, all five PDA cell lines were screened for FANCD2 foci formation by immunofluorescence assay (Supplementary Fig. S1D). Additionally, we validated previously published reports that cell lines with defects in the FA pathway are sensitive to inter-strand crosslinking agents such as mitomycin C (MMC) 35 (Fig. 1B) and oxaliplatin ( Supplementary Fig. S1E). Dose response data with MK-1775, MMC and oxaliplatin are summarized in Table 1 and Supplementary Tables  S1 and S4. To validate the results obtained in the endogenous repair deficient cell lines, we transiently transfected the DDR-P cells (MIA PaCa2) with siRNA oligos against FANCD2 and BRCA2 (Fig. 1C inset). Consistent with the above results, silencing either FANCD2 or BRCA2 induced resistance to MK-1775 as compared to control  Table S5). FANCD2 or BRCA2 silencing sensitizes the cells to MMC (Fig. 1D), in agreement with previous studies 35 . Interestingly, despite the phenotypic differences observed in cell survival, all five PDA cell lines respond mechanistically to WEE1 inhibition (through MK-1775 treatment) as evidenced by a decrease in WEE1 protein expression and downstream phosphorylation of CDK1 (Fig. 1E), as also reported by other studies 14,43 . These data suggest that endogenous genetic defects occurring in PDA cells influence their sensitivity to MK-1775.

MK-1775 is more effective against non-pancreatic DDR-deficient cancer cells.
To further validate the observed resistance in DDR-D cell lines, we evaluated the efficacy of MK-1775 in isogenic cell culture models of recombinantly modified FANCC-null human colon cancer cell line (RKO) and BRCA2-null human colon cancer cell line (DLD1) (Supplementary Fig. S2A and Supplementary Table S6) 32,44 . As previously published, both FANCC-and BRCA2-deficient cells were more sensitive to MMC treatment (Supplementary Fig. S2B and Supplementary Table S6) 32,44,45 . Surprisingly, we observed that BRCA2 (+ /− ), BRCA2 (− /− ) and FANCC (− /− ) clones were more sensitive to MK-1775 than parental clones ( Supplementary Fig. S2A). Of note, Aarts et al. performed a similar assay using recombinantly modified BRCA2 in DLD1 cell lines and failed to observe any differences in the MK-1775 sensitivity between BRCA2-proficient and -deficient cells. Moreover, siRNAmediated BRCA2 knockdown did not significantly increase MK-1775 sensitivity (cell viability > 50%) compared to vehicle treated cells 40 . However, Kausar et al. observed that BRCA2 wild-type isogenic cells were sensitive to MK-1775 (AZD-1775) upon gemcitabine-radiation treatment as compared to artificially knocked out BRCA2 isogenic cells 39 . This suggests that MK-1775 sensitivity to cells with genetically disrupted FA genes is both context (e.g., MSI vs CIN) and cell lineage dependent (e.g., colon cancer cells vs. pancreatic cancer cells) 23 .  Fig. 2A,B). However, FANCG-deficient Hs 766T cells, FANCC-deficient PL11, and BRCA2-deficient Capan-1 cells did not exert the same effect when treated with MK-1775 ( Fig. 2A,B). Importantly, co-treatment with MMC also failed to enhance cell death in DDR-D cells ( Fig. 2A,B and Supplementary Table S7). We further validated these results in siRNA FANCD2 and/or BRCA2 transfected DDR-P MIA PaCa2 cell lines ( Supplementary Fig. S2D). In the control transfected (si-scrambled-control) cells, MK-1775 treatment alone as well as in combination with MMC induced significant cell death compared to untreated cells ( Table S7) suggesting that there were indeed more apoptotic cells in these DDR-P cell lines. Taken together, these data demonstrate that MK-1775 induces more cell death in FA pathway proficient cells as compared to DDR-D cells.

MK-1775 treatment induces abnormal nuclear morphology and replication stress in DDR-proficient pancreatic cancer cells.
To further evaluate the mechanism of action of MK-1775, all five of our selected cell lines (MIA PaCa2, PANC-1, Hs 766T, Capan-1 and PL11) were either treated with MMC or MK-1775 alone or in combination for 24 hours before immunofluorescence was performed. MMC treatment induced γ H2AX foci formation in all cell lines (Fig. 3A,B). However, WEE1 inhibition through MK-1775 treatment induced pan-nuclear γ H2AX staining without visible foci consistent with the previous findings ( Fig. 3A,B) 46 . Interestingly, MK-1775 treatment caused significant nuclear abnormalities forming multi-nucleated phenotype in DDR-P MIA PaCa-2 and PANC-1 cells suggesting that the cells undergo mitotic catastrophe which was not very evident in the DDR-D cell lines (Fig. 3A,C). We further validated the results by transfecting MIA PaCa2 cells with siRNA oligos against FANCD2 or BRCA2. Results demonstrated that control transfected cells showed more abnormal nuclear morphology, γ H2AX foci and pan-nuclear γ H2AX staining after treatment with MK-1775 than siFANCD2 or BRCA2 transfected cells (Fig. 3D-F). These results suggest that WEE1 inhibition with MK-1775 alone results in a higher degree of replication stress and induced DNA damage in DDR-P cells as compared to DDR-D cells, which is further enhanced upon co-treatment with MMC.

MK-1775 induces mitotic entry in DDR-proficient pancreatic cancer cells. It was previously pub-
lished that WEE1 inhibition via MK-1775 abrogates G2/M cell cycle arrest and enhances early mitotic entry 43 .
To further validate this mechanism of action in PDA cells, we performed cell cycle kinetics by pulse-labeling the cells with bromodeoxyuridine (BrdU) followed by MK-1775 and/or MMC treatment for 24 hours. The cell cycle distribution was monitored by the progression of the BrdU-labeled cells. Our results demonstrate that DDR-P cells showed a higher percentage of cells arrested in the G2/M phase of the cell-cycle than DDR-D cells after combination treatment of MK-1775 and MMC, with the exception of PL11 cells that induce a higher percentage of cells in G2/M phase (Fig. 4A). We further determined the mitotic index of cells by immunofluorescence assay. All cell lines were either treated with MMC or MK-1775 alone or in combination for 24 hours and then stained with phospho-histone 3 (pH3), a marker of mitotic entry. Results showed that DDR-P cells induce mitotic entry upon MK-1775 exposure which was further enhanced by combination treatment compared to control cells (Fig. 4B,C). Interestingly, DDR-P PANC-1 cells after combinational treatment showed a higher percentage of cells undergoing mitosis than MK-1775 alone treatment suggesting that there was indeed a significant difference in cell-cycle kinetics which was not evident in BrdU experiment (Fig. 4A,C). In comparison, DDR-D Hs 766T and Capan-1 cells did not show enhanced mitotic entry after MK-1775 alone or in combination with MMC treatments (Fig. 4B,C). However, though DDR-D PL11 cells showed significant mitotic entry compared to Hs 766T and Capan-1 cells, but not significant when compared to untreated control cells (Fig. 4B,C).
We further analyzed the cell cycle kinetics after treatments using pH3 antibody by FACS. Consistent with previous findings, we observed a higher percentage of pH3 positive cells in DDR-P MIA PaCa2 and PANC-1 cells upon combination treatment compared to DDR-D Hs 766T and Capan-1 cells (Fig. 4D,E). Surprisingly, we observed more cells arrested in mitosis in FANCC-deficient PL11 cells in response MK-1775 alone or combination treatments (Fig. 4D,E). However, an apoptotic assay indicated less cell death in PL11 cells upon MK-1775 alone or combination treatment compared to DDR-P cells ( Fig. 2A). This suggests that PL11 cells were arrested in G2/M phase but did not undergo mitotic catastrophe as observed in DDR-P MIA PaCa2 and PANC-1 cells. A similar trend was observed in all cell lines with oxaliplatin alone or in combination with MK-1775 treatments ( Supplementary Fig. S3A-D). Taken together with the cell cycle analyses, an FA-repair deficiency in PDA cells is likely to cause G2/M arrest even with MK-1775 inhibition.

WEE1 inhibition in combination with a DNA damaging agent induces caspase-dependent cell death in DDR-proficient pancreatic cancer cell lines.
To further evaluate that WEE1 inhibition by MK-1775 induces mitotic cell death in DDR-P cells, we conducted immunofluorescence (IF) experiments using pH3 and cleaved caspase 3 (CSP3) to simultaneously assess mitotic entry and cell death, respectively. Our results demonstrate that upon MK-1775 and combination treatments, DDR-P cells (MIA PaCa2 and PANC-1) that stained positive for CSP3 also co-localize with pH3 suggesting that cells are undergoing mitotic catastrophe (Fig. 5A). On the contrary, DDR-D Capan-1 and PL11 cell lines demonstrated mitotic entry as indicated by the increased expression of pH3 upon MK-1775 or combination treatment (Fig. 5A). However, Capan-1 and PL11 cells stained negative for CSP3 suggesting that cells failed to undergo apoptosis (Fig. 5A). As expected, Hs 766T cells were negative for both CSP3 and pH3 staining indicating that MK-1775 failed to induce mitotic entry or cell death (Fig. 5A). These results indicate that MK-1775 in combination with MMC enhances mitotic entry regardless of enormous DNA damage induced by MMC and promotes cell death in DDR-P cells. Whereas, DDR-D (Hs 766T, Capan-1 and PL11) cells demonstrate induction of DNA damage upon MMC treatment (Fig. 3A,B) but MK-1775 failed to induce significant mitotic entry (Fig. 4B-E) or cell death ( Fig. 2A,B and Fig. 5) compared to DDR-P cells suggesting that these cells are arrested in the G2/M phase of cell cycle.

Discussion
Several studies have been conducted to design therapeutic strategies incorporating MK-1775 either as a monotherapy or in combination with chemotherapeutic drugs in different cancers 7,8,10,14,23 . The G1/S checkpoint is functionally inactive in most of the cancers mainly because of mutation in TP53 gene, therefore abrogating the G2/M checkpoint via WEE1 inhibition followed by DNA damage offers a promising therapeutic opportunity to kill cancer cells. There is compelling evidence to support this strategy that MK-1775 induces forced mitotic entry and combination treatment with DNA damaging agents or radiation promotes apoptosis and reduced tumor growth 10,12 . Targeting WEE1 in an effort to enhance therapy in PDA cells is especially thought provoking in light of our recent study that DNA damaging agents can rapidly and potently induce an HuR-dependent WEE1 upregulation 18 .
In this report, we validated previous work that MK-1775 has a cytotoxic effect in a panel of pancreatic cancer cell lines ( Supplementary Fig. S1A), which of note, is a different panel of cell lines used in a recent study to define genes involved in WEE1 inhibition ( Table 2) 40 . Complementary to our work, Kausar et al. concluded that MK-1775 can sensitize HR-proficient PDA cells to gemcitabine chemoradiation and they propose that PDA tumors without underlying HR-defects would respond best to this combination strategy 39 . Expanding on this line of investigation, we previously found that upregulation of WEE1 in PDA cells via HuR occurred in a variety of PDA lines with DDR-proficient and deficient genetic backgrounds 18 . Our results showed that MIA PaCa2 and PANC-1 cells (DDR-P) were significantly more sensitive to MK-1775 treatment compared to Hs 766T, PL11 and Capan-1 pancreatic cancer cell lines (DDR-D) (Fig. 1A). Whereas drug efficacy, as measured by a cell survival assay, depends on DDR function; on a molecular level, PDA cells are not dependent on hallmark DDR functions (Fig. 1A,E). We do note that in our extensive evaluation of a diverse genotype of PDA cell lines (n = 9), this association between DDR-D lines and MK-1775 did not correlate 100% (e.g., BxPC3 and SU.86.86 cells) (Supplementary Table S2). Limitations to making a connection between DDR status and MK-1775 in these cell lines and in all PDA tumors may be related to several factors intrinsic to PDA biology (i.e., other confounding molecular alterations such as PKMYT1 expression which phosphorylates and inhibits WEE1 targets such as CDK1 47 , along with potential unknown drug transporter/metabolism deficiencies within these cell lines).
Contrary to recent publications 23,40 , we observed that genetically modified colon cancer cell lines for BRCA2 and FANCC were more sensitive to MK-1775 ( Supplementary Fig. S2A). Aarts et al. did not observe any such difference in the MK-1775 sensitivity in their genetically modified colon cancer cell lines for BRCA2-proficient and -deficient cells 40 . While Kausar et al. also reported that BRCA2-null isogenic colon cell lines were not further sensitized to gemcitabine-radiation by MK-1775 while the complementary wild-type BRCA2 cells were significantly more sensitive (Table 2) 39 . Previously, it was demonstrated that TP53 interacts with FANCC to regulate apoptosis and tumorigenesis upon MMC or radiation exposure in FANCC-deficient cells 48 . However, Rosselli et al. 49 reported reduced-while Kruyt et al. 50 and Ridet et al. 51 showed normal-induction of TP53 in FA cells. All pancreatic DDR-D cell lines (Hs 766T, PL11 and Capan-1) used in this study are TP53-deficient and showed less sensitivity to MK-1775 compared to DDR-P (MIA PaCa2 and PANC-1) cells. However, targeted disruption by homologous recombination RKO and DLD1 cells, which are TP53-proficient, were sensitive to MK-1775  Fig. S2A,B and Supplementary Table S6). Moreover, transfecting the RKO and DLD1 cells with siRNA oligos against TP53 did not alter the MK-1775 efficacy in these cells (data not shown). These data suggest that MK-1775 sensitivity to cells with genetic disruption of DDR genes in both pancreatic and colon cancer cell lines could be context dependent (Table 2). Additional differences between this study and the recently reported findings are that: 1) we evaluated MK-1775 efficacy in naturally disrupted FANCC or BRCA2 cell lines, and 2) different identified FA genes were validated in each instance (Table 2) 23,40 . Importantly, a recent report of a Phase I study in patients with refractory solid tumors included 6 BRCA-mutated patients 23 . Two of the 6 patients had partial responses (an ovarian and a head and neck tumor). Only one patient with pancreatic cancer was in this subgroup, but withdrew to go on standard-of-care treatment. These data perhaps support the notion that WEE1 inhibition targeted against BRCAness may be context dependent. We certainly do not dispute the data from other laboratories ( Table 2) 23 , in fact, we hope that future studies may help determine which tumor systems that harbor BRCA/FA-homologous repair gene defects may cause resistance or sensitivity to WEE1 inhibition.
Although a trend towards decreased sensitivity to WEE1 inhibition in DDR-D cells is apparent from this study, the mechanism of resistance is unclear. One possible explanation of this resistance mechanism would be that upon MK-1775 treatment, alternative checkpoints are activated in DDR-D cell lines, compensating for    23,39,40 .
G2/M checkpoint inactivation and allowing cells to go under efficient DNA repair. Apoptotic assays suggest that MK-1775 did not synergize with MMC to promote cell death in DDR-D cells (Fig. 2A). In addition, cell cycle kinetics also suggests that cells are primarily arrested in G2 phase of the cell cycle in these cells and failed to induce mitotic entry either alone or in combination with DNA damaging agents MMC or oxaliplatin ( Fig. 4 and Supplementary Fig. S3). It seems perplexing that there is a decrease in γ H2AX positive cells in DDR-D cell lines suggests that these cell lines adapt an alternative G2/M checkpoint mechanism upon MK-1775 treatment to repair the damage. Incorporating data from the drug sensitivity assays, apoptosis and immunofluorescence with cleaved caspase 3 assays, and with the exception of PL11 cells that showed increased mitotic entry with FACS assay (Figs 1, 2 and 5), these findings indicate that DDR-D cells are taking an alternative route to activate G2/M checkpoint. Therefore, it would be more informative to test the efficacy of MK-1775 when combined with other G2/M checkpoint inhibitors. For instance, recent studies also showed that combined inhibition of checkpoint kinase 1 (CHK1), another inducer of G2/M phase arrest, and WEE1 increased therapeutic efficacy and reduced tumor growth. Wang et al. also reported that the CHK1 selective inhibitor (LY2603618) enhanced MK-1775 sensitivity in pancreatic cancer cell lines 15 . Mak et al. demonstrated that inhibiting more than one of the components of the ATR-CHK1-WEE1 pathway can overcome the pharmacological limitation of these inhibitors 52 , therefore supporting a potentially efficacious strategy to combine MK-1775 and CHK1 inhibitor in DDR-deficient PDA cells.
In summary, our in vitro data support the notion that WEE1 inhibition alone may not provide a clinical advantage for PDA patients with mutations specifically in FANCC, FANCG, and BRCA2. We were able to validate these findings by silencing FANCD2 and BRCA2 in DDR-proficient PDA cells. These pre-clinical in vitro studies provide a rationale to select tumors based on DDR-gene proficiency, and assess efficacy of WEE1 inhibition in patients according to their specific mutational status. Although this work ultimately needs to be supported by either retrospective analysis of a clinical trial and/or a prospective, biomarker driven clinical trial; for now, our data suggest that molecularly identified DDR-deficient PDAs (i.e., FANCC, FANCG, and BRCA2) should not be treated with MK-1775-based therapies. Additionally, this work adds to the literature that WEE1 inhibition may be context dependent and this line of investigation may guide the selection and timing of other therapies to be used in combination with MK-1775.
All cells were treated with the IC 50 values of the DNA damaging agent MMC (mitomycin C; (Sigma, St. Louis, MO) and oxaliplatin (Sigma) as previously described 18 by adding directly into the culture medium. MK-1775 was purchased from Selleckchem, Houston, TX.
Immunofluorescence. Approximately, 50,000 cells per well were plated on coverslips in 24-well plate.
After treatments, cells were washed 2 times with PBS, fixed with 4% paraformaldehyde, permeabilized with 0.2% Triton-X, blocked with 5% goat serum, and incubated with pH3, CSP3 or γ H2AX antibodies, as previously described 18 . Cell nuclei were stained with DAPI and coverslips were mounted with DAPI ProLong Gold Antifade (Invitrogen) for analysis with a Zeiss LSM-510 Confocal Laser Microscope. Approximately, 300-500 cells were used for counting the number of pH3 and γ H2AX positive nuclei using ImageJ 1.47a software (NIH, Bethesda, MD, USA; http://imagej.nih.gov/ij/). FACS for pH3 positive cells. Cells were seeded at 10 6 cell density in 10 cm dishes and treated with MMC (150 nM/L) and MK-1775 (400 nM/L) for 24 hours. Cells were harvested using cellstripper (Corning), fixed in ice-cold 70% ethanol for 2 hours, permeabilized with 0.25% Triton-X on ice for 10 minutes, incubated with anti-phospho (Ser10)-Histone H3 Ab (Clone 3H10) antibody (1:1,000, Cell Signaling) for 1-2 hours at 4 °C followed by Alexa488-conjugated antibody (1:2,000) for 1 hour in dark at 4 °C, washed and re-suspended in 20 μ g/ml propidium iodide supplemented with 100 μ g/ml RNase for 15 minutes at RT in the dark. Samples were analyzed on flow cytometry BD LSRII (BD Biosciences, San Jose, CA). 10,000 events were recorded for each sample. Each cell line has its own compensation controls and gating was done according to each cell line unstained/untreated sample. Data was analyzed in FlowJo (FlowJo LLC.).
Scientific RepoRts | 6:33323 | DOI: 10.1038/srep33323 Drug Sensitivity and Apoptosis Assays. Cells were seeded at 1,000 cells per well in 96-well plates in triplicates and treated after 24 hours. After 5-7 days of treatment (only 1 dose exposed to cells after the cells are plated for 24 h), cells were washed twice with PBS and lysed with deionized-water for 1 hour at 37 °C. Cells were stained with PicoGreen (Invitrogen), a fluorescent dye that selectively binds double stranded DNA, for 2 hours in dark at RT, as previously described 18 . The intensity of the fluorescent signal correlates to the number of viable or surviving cells. Results were analyzed using GraphPad Prism (GraphPad Software Inc.), La Jolla, CA.
For apoptoic assay, cells were seeded at 10 6 cell density in 10 cm dishes and treated with MMC (150 nM/L) and/ or MK-1775 (400 nM/L) for 24 hours. Annexin V13242 labeling kit (Invitrogen) was used following manufacturer's protocol to measure the apoptotic cells. Samples were analyzed on flow cytometry BD LSRII (BD Biosciences, San Jose, CA) and 50,000 events were recorded for each sample in Fig. 2A and 10,000 events were recorded for each sample in Fig. 2C. Each cell line has its own compensation controls and gating was done according to each cell line unstained/untreated sample. Gating strategy is shown in Supplementary Fig. S2C. Only V+ /PI− cells are considered as apoptotic cells and plotted in the graph. Data was analyzed in FlowJo (FlowJo LLC.).
Statistical Analysis. All p-values were calculated in GraphPad using paired T-Test function.