YAP1-Mediated CDK6 Activation Confers Radiation Resistance in Esophageal Cancer – Rationale for the Combination of YAP1 and CDK4/6 Inhibitors in Esophageal Cancer

Esophageal cancer ranks eighth in incidence and sixth in mortality among all types of cancers worldwide (1, 2). A total of 17,290 new cases and 15,850 deaths were likely to occur in 2018 in the United States (3, 4). Preoperative chemoradiotherapy is an accepted standard approach for some localized esophageal cancer patients (5, 6). However, most patients do not benefit, and their 5-year survival rate is ≤40% (7). Thus, resistance to therapy and metastatic potential are common in the clinic (8). Therefore, defining key mediators of resistance to therapy is critical to improve the outcome of patients.

Molecular mechanisms that mediate resistance to chemoradiation therapy in esophageal cancer remain unclear, and multiple pathways appear to be engaged by cancer cells. One widely accepted notion is the role of cancer stem cells (CSCs). CSCs are defined as a small subpopulation of undifferentiated cells that have the self-renewal capacity, and they produce progenies to maintain tumor progression. We previously reported on the role of the Hippo pathway transcriptional coactivator Yes-associated protein (YAP1) on CSC phenotype that can endow CSC properties to cells including high capacity to form tumor spheres, increased proliferation, and progression by regulating its target SOX9 (9). We reported that YAP1′s upregulation of the EGFR pathway plays an important role in conferring chemotherapy resistance to esophageal cancer cells. (10) Taken together, these results suggest that YAP1 may be a key player in chemoradiation resistance in esophageal cancer.

Amplification of cyclin-dependent kinases (CDK6) has been reported in esophageal cancer (11). CDK6 is a key regulator of the cell cycle and plays a pivotal role in the transition from G₁ to S phase in cancer cells (12) through phosphorylation of retinoblastoma-associated protein 1 (Rb1). Overexpression and amplification of CDK6 together with its homologue CDK4 were associated with poor survival in esophageal adenocarcinoma (11). A previous report suggested that YAP1/Tead-mediated transcription controls cellular senescence by CDK6, implying a crosstalk between the Hippo/YAP1 and CDK6 pathways (13). Both YAP1 and CDK6 are highly relevant to esophageal cancer progression; however, their role and mechanisms involved in radiation resistance remain unclear. We hypothesized that YAP1 regulates the sustained CDK6 overexpression and its activation, which confers resistance in esophageal cancer. Therefore, targeting both YAP1 and CDK6 might be more advantageous in overcoming resistance in esophageal cancer.

In this study, we demonstrated that overexpression of YAP1 was positively associated with CDK6 expression in posttreatment (radiation-resistant) esophageal cancer tissues. Induced YAP1 in esophageal cancer cells upregulated CDK6 expression, increased transcription, and induced radiation resistance, but blocking YAP1 and CDK6 by the YAP1 inhibitor, CA3, and the CDK6 inhibitor, LEE001, significantly suppressed esophageal cancer cell growth and CSC properties particularly in radiation-resistant cells. The combination of LEE001 and CA3 had the highest antitumor effects in radiation-resistant cells with high YAP1 and CDK6 in vitro and in vivo. Furthermore, the combined inhibition of YAP1 and CDK6 sensitized resistant tumors to irradiation in vivo. Our data imply that a crosstalk between YAP1 and CDK6 seems to play a pivotal role in conferring radiation resistance, and targeting both YAP1 and CDK6 could be a useful therapeutic strategy to treat esophageal cancer.

The normal human esophageal epithelial cell line HET-1A; esophageal adenocarcinoma cell (EAC) lines Flo-1, SKGT-4, BE3, OE33, JHESO, and OACP; plus esophageal squamous carcinoma (ESCC) cells Yes-6, KATO-TN (TN), TE-2, TE3, TE-7, TE-8, and TE-12 were used in this study (14–16). All human cell lines were tested and authenticated in the Characterized Cell Line Core facility at The U.T.M D Anderson Cancer Center (Houston, TX). LEE001 was purchased from the United States Pharmacopeia. Doxycycline hyclate (DOX) was purchased from Sigma-Aldrich. The antibodies against YAP1, CDK4, CDK6, and phosphorylated Rb were purchased from Cell Signaling Technology. CD133-APC and CD44-PE were purchased from Miltenyi Biotec Inc. ALDH1 labeling and sorting were performed using an ALDEFLUOR kit from STEMCELL Technologies Canada Inc. DNA plasmids that encoded wild-type human YAP1 (hYAP1, CMV-YAP1) or a mutant protein that can no longer be phosphorylated at Ser127 (hYAP1 S127A, CMV-S127A-YAP1) were purchased from Addgene (9). We have previously reported on the doxycycline-inducible YAP1 lentivirus expression plasmid (PIN20YAP1) and lentiCRISPR to knock down YAP1 (17).

The protein extraction and Western blot analyses were performed as previously described, and immunoreactive bands were visualized by chemiluminescence detection (18).

The LentiCrispr/Cas9 system was used to knock out YAP1 or CDK6 in JHESO, Flo-1 XTR cells using the GeCKO LentiCrispr resource tool (MIT, http://genome-engineering.org/gecko/). gRNAs were designed using the MIT's online webpage (http://crispr.mit.edu/). pLentiCrispr v1 was used for cloning gRNAs. Briefly, the vector included Cas9 gene and was cut with BsmB1, the longer fragment was ligated with the gRNA pairing duplexes, and resulting clones were verified by sequencing. Lentiviruses were made using pLentiCripr-gRNA, packaging plasmids pCMV.Dr8.2, and pCMV.VSV.G in 10:10:1 ratio in the 6-well plate of HEK293T. Lentiviral supernatant was used to transduce cell lines in the 6-well plate in the presence of 8 μg/mL polybrene. Cells were then selected in puromycin at proper concentrations for 1 to 2 weeks. Cells were then propagated and verified for knockout efficiency by Western blots.

SKGT-4 (pIN20YAP1) with (DOX+) or without (DOX-) or radiation-resistant Flo-1 XTR and its parental cells were treated with 0.1% dimethyl sulfoxide (control), CA3, LEE001, or the combination (CA3 plus LEE001) for 3 and 6 days. Cell viability was then assessed by using the CellTiter 96 aqueous nonradioactive cell proliferation assay (MTS) according to the instructions of the manufacturer (Promega). The results have been presented as the percentage of control and repeated at least 3 times.

To establish 5-FU–resistant subclones, SKGT-4 parental cells were cultured with various concentrations of 5-FU for 3 to 5 weeks, and surviving cells were collected. This procedure was repeated 4 times. The establishment of these 5-FU–resistant subclones took 3 to 6 months, and newly derived 5-FU–resistant clones were designated SKGT4-RF. To establish radiation-resistant subclones, Flo-1 and SKGT-4 esophageal adenocarcinoma parental cells were irradiated with 2 GY, 4 times, once every week for 8 weeks. Flo-1 and SKGT-4 cells were made resistant to radiation and designated Flo-1 XTR or SKGT-4 XTR.

To quantify changes in the YAP1 and CDK6 mRNA levels, real-time RT-PCR was performed using the ABI 7500 fast system (Applied Biosystems). Total RNA from cell cultures was extracted by using Trizol (Ambion), and concentrations of RNAs were measured by Nanodrop 1000 (Nanodrop). The first strand of cDNA was synthesized by reverse transcription PCR using the Invitrogen's Superscript III kit (Invitrogen). Quantitative PCR measuring mRNA expression levels was performed by the ABI 7500 Sequence Detection System 2.2 software that automatically determined the fold change for YAP1 and/or CDK6 in each sample by using the δδCt method with 95% confidence using primers listed below. Relative quantitation was calculated by using RE = 2^(−ΔΔCt). Primers for reference gene GAPDH were hGAPDH-5 5′ ACCCAGAAGACTGTGGATGG 3′; hGAPDH-3 5′ TCTAGACGGCAGGTCAGGTC-3′. Primers for YAP1 were hYAP1.E3.F 5′ CTGTCCCAGATGAACGTCAC-3′ and hYAP1. R 5′-TTCTCTGGTTCATGGCAAAA-3′. Primers for CDK6 were hCDK6. F 5′ GGAGACCTTCGAGCACC 3′ and hCDK6.mRNA.R 5′ CACTCCAGGCTCTGGAACTT 3′.

Analysis of the cell cycle in esophageal adenocarcinoma cells using flow cytometry was performed as described previously (19). In brief, SKGT-4 and JHESO cells were seeded onto the 6-well plates (1 × 10⁵ cells/well) in the DMEM and cultured for 24 hours to allow for cell attachment. Cells were then treated with 0.1% dimethyl sulfoxide (control) or LEE001 at different concentrations as indicated for 48 hours. Next, cells were harvested, fixed with methanol, washed, treated with RNase A, stained for DNA with propidium iodide (Sigma), and their DNA histograms and the cell-cycle phase distributions were analyzed using flow cytometry with the FACS Calibur instrument (Becton Dickinson).

We have previously described the 5 × -UAS-luciferase reporter and Gal4-TEAD4 constructs (20). Transient cotransfection 293T and SKGT4 dox+ esophageal adenocarcinoma cells with 5 × -UAS-luciferase reporter and Gal4-TEAD4 with a CMV-β-gal construct were performed as described previously (21).

Esophageal cancer cells and esophageal cancer tissues were subjected to indirect immunofluorescence staining with YAP1 (1:100) and CDK6 (1:100) primary antibodies and then labeling with Alex-488 (for CDK6) and Alex-555 (for YAP1) as described elsewhere (18). Fluorescence was assayed by the confocal microscope (FluoView FV500; Olympus) and analyzed by the CellQuest PRO software (BD Biosciences).

Sphere cultures were performed as described previously (21). Briefly, parental Flo-1 and Flo-1 XTR cells were seeded in triplicate onto the 6-well ultra-low attachment plates (800 cells/well; Corning Life Sciences) in the serum-free combined Dulbecco's modified essential and F-12 media supplemented with 20 ng/mL epidermal growth factor, 5 μg/mL insulin, 0.5 μg/mL hydrocortisone, and 2% B27 supplement without vitamin A and 1% N2 Supplement (Invitrogen, Life Technologies). CA3, LEE001, or their combination was added at the time the cells were seeded. After 10 to 20 days in culture, the tumor spheres (diameter >100 μm) were counted.

IHC staining for YAP1 and CDK6 was performed on human esophageal cancer tissues and mouse xenografts using the antibodies against CDK6 (1:100), YAP1 (1:100), and KI67 (1:100) as described previously (21).

Esophageal adenocarcinoma cells were seeded in the 6-well plates for different radiation doses to allow for an approximately equal number of resultant colonies, and the optimal number of cells was determined to be 800 per well. The following day, cells were irradiated using a high-dose-rate ¹³⁷Cs irradiator (4 Gy or 5 Gy/min) and cultured for 10 to 14 days to allow for colony formation. Cells were then fixed in a 3% crystal violet/10% formalin solution. Colonies of more than 50 cells were then counted, and survival fraction was determined. The quantification of colony was also assessed by using the fluostar omega microplate reader (BMG LABTECH Inc.) at wavelength of 590 nm after dissolving colonies using 10% acetate acid. All treatments were performed in triplicate or higher.

In vivo experiments were conducted in accordance with the guidelines of the MD Anderson Institutional Animal Care and Use Committee. Nude mice were inoculated subcutaneously with Flo-1 cells and Flo-1 XTR–resistant cells (5 × 10⁶ cells and n = 5/group). After 15 days, the mice bearing Flo-1 XTR xenografts underwent intraperitoneal injection of CA3 at 1 mg/kg/mouse, LEE001 at 30 mg/kg/mouse, or a combination of them, 3 times a week for total 3 weeks. The control group was given PBS at 100 μL/mouse. The mice tumor volumes, tumor weights, and body weights were measured as described previously (22). All measurements were compared using the unpaired Student t test.

Differences between the groups were assessed using the Student t test or the Fisher exact test. The IHC expression analyses were performed using the appropriate nonparametric test (χ² test and Spearman correlation test). Univariate survival analysis was performed using the Kaplan–Meier method. A multivariate Cox regression model with the backward stepwise method was used to detect the independent prognosticators of survival. Two-tailed P values of less than 0.05 were considered significant. All statistical analyses were performed using the SPSS_20.0 software program (IBM Corporation).

We have previously reported that the overexpression of YAP1 in esophageal cancer plays an important role in cell proliferation and acquisition of the CSC properties (refs. 9 and 22). In The Cancer Genome Atlas (TCGA) genomic sequence data, we found that CDK6 is frequently amplified in esophageal cancers, actually the second highest rate of amplification across all tumor types indicating its importance in esophageal cancer (cbioportal.com; Fig. 1A). To determine whether both YAP1 and CDK6 expressions were associated with esophageal cancer, immunoblotting was performed on HET1A cells, 6 esophageal adenocarcinoma cell lines, and 7 ESCC cell lines. The results in Fig. 1B show that the expression of both YAP1 and CDK6 increased in most of esophageal adenocarcinoma and ESCC cell lines compared with HET1A cells, and there was a correlation between overexpression of YAP1 and CDK6. The positive association of YAP1 and CDK6 was further validated in 151 esophageal cancer samples including 71 esophageal adenocarcinoma and 77 ESCC as shown in Fig. 1C. When we analyzed ESCC and esophageal adenocarcinoma separately, we found a higher correlation between YAP1 and CDK6 in ESCC than in esophageal adenocarcinoma (Supplementary Fig. S1). We have previously reported that YAP1 mediates chemoresistance (10). To further explore if both YAP1 and CDK6 were associated with therapy resistance, we measured the expression of YAP1 and CDK6 by IHC in 25 treated residual resistant esophageal cancer tissues (P2, posttreated esophageal cancer tissue with ≥50% residual tumor cells) compared with the relatively sensitive tumors (P0/P1, esophageal cancer tissue with 0% or <50% residual cells). YAP1 and CDK6 were highly expressed in resistant esophageal cancers (P2) compared with that in somewhat sensitive esophageal cancers (P1; Fig. 1D; Table 1), and both YAP1 and CDK6 were highly correlated (Table 2) in the resistant esophageal cancer tissues. Immunofluorescence further showed that coexpression of both YAP1 and CDK6 existed in 1 representative resistant case (Supplementary Fig. S2A). Sixty percent to 86% of resistant esophageal cancers (P2) had strong staining for both YAP1 and CDK6, respectively, whereas only 20% and 30% relatively sensitive esophageal cancers (P1) had a weak-to-medium staining for YAP1 and CDK6 (Table 1; Supplementary Fig. S2B). These data support the notion that both YAP1 and CDK6 were highly correlated and associated with therapy resistance.

To determine whether CDK6 expression was regulated by YAP1 in esophageal cancer and other cell types, transfection of human embryonic kidney (HEK293T) cells with constitutively active mutant YAP1S127A cDNA or with wild type YAP1 was performed. As shown in Fig. 2A, the protein expression of CDK6 dramatically increased with the transfection of wild-type YAP1cDNA or mutant YAP1S127A cDNA (an activating mutation that keeps YAP1 active) which was concomitant with an increase in Rb phosphorylation. Correspondingly, the CDK6 mRNA levels increased significantly in HEK293T cells stably transfected with YAP1 by real-time Q-PCR as shown in Fig. 2B, right, which is in concert with the increased YAP1 mRNA levels (Fig. 2B, left). Next, we transduced esophageal cancer cell lines SKGT-4, KATO-TN, and YES-6 with the doxycycline-inducible human flag-tagged YAP1S127A cDNA (PIN20 YAP1S127A). A successful YAP1 induction in SKGT-4, YES-6, and KATO-TN cells by doxycycline at 1 μg/mL increased expression of CDK6 that correlated with increased YAP1 expression (Fig. 2C, left). In contrast, the LentiCRISPR/Cas9-mediated knock down of YAP1 in JHESO cells with constitutively high YAP1 greatly reduced CDK6 expression (Fig. 2C, right). Immunofluorescence analyses further showed that the induction of YAP1 by doxycycline at 1 μg/mL increased nuclear expression of CDK6 that correlated with YAP1 expression in SKGT-4 cells, whereas the knock down of YAP1 in JHESO cells decreased nuclear expression of both YAP1 and CDK6 (Fig. 2D, bottom). To further determine if YAP1 regulated CDK6 at the level of transcription, quantitative real-time RT-PCR was performed in esophageal cancer cells with genetically altered YAP1 levels. As shown in Fig. 2E, the CDK6 mRNA levels were significantly increased by stably YAP1-induced SKGT-4, Yes-6, and KATO-TN cells by doxycycline (DOX+), which were consistent with the YAP1 mRNA levels. In contrast, the knock down of YAP1 in JHESO cells significantly decreased the CDK6 mRNA levels in 2 different clones (Fig. 2F). These data confirmed that YAP1 upregulated CDK6 overexpression and transcription in esophageal cancer cells.

Having established that YAP1 induction increased CDK6 expression in esophageal cancer cells, we questioned whether increased YAP1 and CDK6 in esophageal cancer cells mediated radiation resistance. As shown in Fig. 3A, we exposed SKGT4 DOX⁺ and Yes6 DOX⁺ (with YAP1 induction) and DOX⁻ (without YAP1 induction) cells to different dosages of irradiation (2, 4, and 6 Gy). Both SKGT4 (SK4) and Yes6 YAP1^high cells with DOX⁺ were significantly resistant to radiation compared with DOX⁻ cells and in a dose-dependent manner (Fig. 3A, left and middle plots). In contrast, the knock down of YAP1 in JHESO cells significantly sensitized cells to irradiation (Fig. 3A, right plot) indicating increased YAP1- and CDK6-mediated radiation resistance. Further, we used our established radiation-resistant esophageal cancer cell lines (Flo-1 XTR and SKGT-4 XTR; ref. 23) and found that both YAP1 and CDK6 are highly upregulated in both of these resistant cells compared with their parental counterparts (Fig. 3B, left). The immunofluorescent staining (Fig. 3B, right) further confirmed an increase in the nuclear expression of both CDK6 and YAP1 in radiation-resistant XTR cells. To develop an effective way to overcome radiation resistance, we developed a novel YAP1 inhibitor, CA3, based on the inhibition on YAP1/Tead transcriptional activity (24). We found that CA3 significantly inhibited YAP1 expression in both SKGT-4 and JHESO cells. Interestingly, CA3 dramatically suppressed CDK6 expression and Rb phosphorylation, which represents the downstream CDK6 activity (Fig. 3D). CA3 significantly sensitized radiation-resistant Flo-1 XTR cells to irradiation (Fig. 3E). Most importantly, CA3 preferentially sensitized induced YAP1^high SKGT-4 (DOX⁺) cells to irradiation, whereas there was only minimum effect on SKGT-4 (DOX⁻) induction (Fig. 3F and G).

LEE001 is a small-molecule inhibitor of CDK4/6 that is known to have clinical activity in several tumor types. As demonstrated in Fig. 4A, LEE001 dramatically reduced expression of YAP1 and Rb phosphorylation in SKGT-4, JHESO, and Flo-1cells, whereas LEE001 did not affect the protein level of CDK6 (only exerted its activity by reduced Rb phosphorylation). Moreover, the reduction of YAP1 and Rb phosphorylation was in a dose-dependent manner (Fig. 4B). To determine whether the growth inhibition observed in esophageal cancer cells was associated with specific changes in the cell-cycle distributions, we analyzed the cell-cycle stages using flow cytometry. The cell-cycle phase distributions were analyzed upon treatment of SKGT-4 and JHESO cells with LEE001 at 1 and 5 μmol/L for 48 hours. The results in Fig. 4C and D show that LEE001 increased the G₀–G₁ phase cells but dramatically decreased the S-phase cells in both SKGT-4 and JHESO.

To further confirm that YAP1′s upregulation of CDK6 mediates radiation resistance, we knocked down CDK6 in radiation-resistant Flo-1 XTR cells using the LentiCRISPR/Cas9 system (Fig. 4E, insert). We found that the knock down of CDK6 in radiation-resistant XTR cells also suppressed YAP1 expression (Fig. 4E, insert) but also dramatically decreased colony formation and sensitized tumor cells to irradiation implying that CDK6 can positively crosstalk to YAP1 to play a critical role in YAP1-mediated radiation resistance (Fig. 4E). In addition, as indicated in Supplementary Fig. S2C, YAP1 and CDK6 expressions also increased in the established 5-FU–resistant SKGT-4-RF and Flo-1-RF cells, suggesting that similar mechanisms may apply in both radiation- and chemoresistance. Although in this study, our focus is on radiation resistance of esophageal cancer. These data suggested that YAP1 induction of CDK6 was associated with radiation resistance in esophageal cancer cells.

Having shown that LEE001 or CA3 can inhibit YAP1, CDK6, and Rb phosphorylation, we then aimed to explore whether these 2 inhibitors can synergize. As shown in Fig. 5A, the expression of YAP1, CDK6, and phosphorylated Rb dramatically decreased with the combination at very low doses compared with treatments by single agent in both cell lines (Fig. 5A). Similarly, the YAP/TEAD luciferase activity in both 293T and SKGT-4 DOX⁺ (YAP1-induced cells) was synergistically suppressed by the combination of CA3 and LEE001 as assessed by luciferase activity in cells cotransfected with Gal4-Tead and 5XUAS-luciferase plasmids which represents YAP-1 transcriptional activity (ref. 20; Fig. 5B).

Having demonstrated that YAP1 regulates CDK6 to induce therapy resistance and inhibition of both (YAP1/CDK6) results in the highest decrease in YAP1/CDK6 expression as well as in reduced Rb phosphorylation, next, we sought to assess if the combined inhibitors had higher antitumor effects in vitro especially in YAP1 high and radiation-resistant cells. SKGT-4 (PIN20YAP) cells with YAP induction (DOX⁺) or without YAP induction (DOX⁻) were treated with CA3 and LEE011 either alone or in combination at the concentration indicated for 3 and 6 days. The cell growth inhibition was measured using the MTS assay. We observed that the combination had much higher inhibitory effect in SKGT-4 cells on both day 3 (Supplementary Fig. S3A) and day 6 (Fig. 5C, left plot). Similar results were observed in Flo-1 parental cells and radiation-resistant Flo-1 XTR cells (Supplementary Fig. S3B; Fig. 5C, right plot). More importantly, we observed that the combination significantly decreased colony formation in radiation-resistant XTR cells (Fig. 5D). These data indicated that inhibition of both YAP1 and CDK6 could provide the best antitumor effects in resistant esophageal cancer cells.

We first observed that the radiation-resistant XTR cells were enriched with the CSC properties by increased capacity to form larger tumor spheres but also in greater numbers than did the parental counterparts (Supplementary Fig. S3C) and have higher fraction of ALDH1⁺ cells than the parental counterparts (23). The combination of CA3 and LEE001 dramatically reduced tumor sphere formation in XTR Flo-1 cells compared with either treatment alone, whereas parental Flo-1 cells do not form tumor spheres (Supplementary Fig. S3D). CD133, ALDH1, and CD44 are established CSC markers, thus the combination of drugs markedly decreased CSC cells as determined by assessing the fractions of CD133⁺ (Fig. 5E), ALDH1⁺ (Fig. 5F), and CD44⁺ cells (Supplementary Fig. S3E) in radiation-resistant XTR cells. This indicated that combined suppression of YAP1/CDK6 had the strongest effect in reducing the CSC population that is especially enriched in radiation-resistant cells.

Initially, we implanted both parental Flo-1 and Flo-1 XTR cells into nude mice, and Flo-1 XTR xenografts formed successfully, whereas the parental Flo-1 cells formed smaller and slowly growing xenografts (Fig. 6A) consistent with their behaviors in tumor sphere formation assays (Supplementary Fig. S3C), indicating that Flo-1 XTR radiation-resistant cells were enriched with CSCs (tumor initiation cells). Thus, we used Flo-1 XTR cells to assess the effect of combined LEE001/CA3 treatment on tumor growth in vivo. Figure 6B shows the treatment schema using CA3, LEE011, and combination in the Flo-1 XTR cell-derived xerograph (CDX) model. In nude mice bearing Flo-1 XTR xenografts, tumors were divided randomly into 4 groups and then treated with control (PBS), CA3, LEE001, and combination (CA3/LEE001) for 3 weeks. At the end of the 3 weeks, xenografts weights and volumes were measured. The mice treated with either LEE001 or CA3 had reduced tumor weights/volumes, whereas the xenografts after LEE001 plus CA3 had the lowest weights/volumes compared with CA3 or LEE001 alone (Fig. 6B–D). In addition, the level of YAP1, CDK6, and Ki67 in mice xenografts diminished dramatically after the combination (Fig. 6E). To determine if inhibition of YAP1, CDK6, or both can further sensitize radiation-resistant, XTR, cells in vivo to radiation, we treated Flo-1 XTR xenografts as follows: (1) the control (PBS) group, (2) irradiation (10 Gy, 1 time) alone, (3) CA3 and irradiation, (4) LEE001 and irradiation, and (5) CA3, LEE011, and irradiation (Supplementary Fig. S4). The results showed that irradiation alone resulted in marginal inhibition of tumor growth, but the combination of irradiation and CA3 had a higher antitumor effect (P < 0.05). Significantly, the combination of irradiation, CA3, and LEE011 resulted in the maximum antitumor effect (Supplementary Fig. S4B and S4C) but without affecting the mice body weight (Supplementary Fig. S4D).

Here, we demonstrated, for the first time, that both YAP1 and CDK6 are overexpressed in resistant esophageal cancer tissues and are associated with therapy resistance in esophageal cancer tissues and cell lines. YAP1 increased CDK6 expression and transcription in esophageal cancer cells and resulted in activation of both YAP1 and CDK6 to confer radiation resistance in esophageal cancer, whereas depletion of CDK6 reversed YAP1-mediated radiation resistance. Importantly, both pathways positively regulate each other. The combined inhibition of YAP1 and CDK6 produced the highest level of antitumor activity in YAP1^high and therapy-resistant esophageal cancer cells as well as sensitized resistant cells to irradiation (Fig. 6F). Our data demonstrated that CDK6-targeted therapeutics could be a promising strategy when combined with YAP1-targeted agents to achieve maximal effect in patients with esophageal cancer. This notion is consistent with the current approaches in the clinic where single agent in most circumstances has limited value. Cancer cells reprogram frequently, and the dual pathway inhibition may turn out to be a better strategy. Patient safety will remain a concern with the combination approach.

We have previously reported on YAP1′s role in chemotherapy resistance, and we noted that TCGA data showed frequent amplification/overexpression of CDK6 that allowed us to further explore these 2 distinct pathways. We have discovered the crosstalk between the 2 pathways in esophageal cancer. It was recently reported that YAP1 downregulation increased senescence in a p53- and p21-dependent manner in colorectal carcinoma cell line, HCT116 (25). Xie and colleagues reported that YAP1 cooperated with TEAD transcription factors to control CDK6 that leads to senescence (13). Our results suggest that YAP1–CDK6 crosstalk in esophageal cancer cells might be the driver for constitutive or acquired radiation resistance. First, the esophageal cancer cells with high YAP1 and CDK6 were more invasive and more resistant to irradiation (Fig. 3A). Second, both YAP1 and CDK6 were upregulated in posttreatment-resistant esophageal cancer tissues (P2) compared with relatively sensitive esophageal cancer tissues (P0/P1). Induced radiation-resistant Flo-1 XTR and SKGT-4 XTR cells had high YAP1 and CDK6 expression compared with their relatively sensitive parental counterparts. Third, we observed that the YAP1-mediated resistance by introducing YAP1 into esophageal cancer cells made cells more aggressive when treated with the ionized radiation (Fig. 3A). In contrast, pharmacologic and/or genetic knock down of YAP1 or CDK6 in YAP1^high or XTR-resistant esophageal cancer cells significantly decreased tumor growth and increased sensitivity to irradiation (10, 26). In addition, although we focused more on YAP1′s upregulation of CDK6 using genetic and pharmacologic tools in this study, we observed that the inhibition of CDK6 by LEE001 or the genetic knock down of CDK6 in esophageal cancer cells decreased YAP1 expression indicating a feed-forward–positive regulation of YAP1 by CDK6 that requires further investigation (Fig. 6F).

In a previous study, we have reported on another YAP inhibitor Verteporfin (VP) inhibited YAP1 protein levels, sensitized cells to cytotoxics, and overcame chemoresistance (10, 27). In the present study, we investigated CA3 (a compound similar to CIL56 according to analysis of its structure; ref. 24) strongly inhibited both YAP1 and CDK6 protein levels, CSC properties, and tumor sphere formation in radiation-resistant esophageal cancer cells. Similar effects can also be observed with the orally available LEE011 targeting CDK4/6 activity, thereby inhibiting Rb protein phosphorylation and inducing G₁ arrest in esophageal cancer cells while not affecting expression of CDK6. Interestingly, LEE001 inhibited YAP1 overexpression, the YAP1/Tead transcriptional activity, and CA3 strongly suppressed CDK6 overexpression and reduced Rb phosphorylation, further confirming the YAP1–CDK6 crosstalk in esophageal cancer cells.

The clinical use of selective CDK4/6 inhibitor combined with a targeted agent or paclitaxel has been proven to be efficacious in advanced-stage ER⁺ breast cancer and other tumor types (ref. 28; refs. 12, 29). In this study, we showed that YAP1 upregulates CDK6 and keeps CDK6 active (increased Rb phosphorylation). We propose that dual inhibition of both YAP1 and CDK6 may improve outcome.

The mounting evidence suggests that CSCs are particularly resistant to chemo/radiation therapy and may therefore contribute to treatment failure (30). In esophageal cancer, the CSC population can be efficiently evaluated by stem cell markers including CD133 (31), CD44, and ALDH1 labeling (32, 33). Zimmerer and colleagues (34) reported that tumor formation in NOD/SCID mice only needs as few as 500 CD133⁺ melanoma cells; in contrast, 100,000 CD133⁻ cells failed to form tumors in these mice. Our previous report showed that ALDH1 is a reliable CSC marker in upper gastrointestinal track tumors (32). This suggests that the CD133⁺ or ALDH1⁺ cells are likely to be the CSCs. Consistent with these results, we found that the radiation-resistant Flo-1 XTR cells with high CD133⁺ and high ALDH1⁺ populations could easily form tumor sphere in vitro and easily generate xenografts in mice, whereas the Flo-1 parental cells with low CD133⁺ and low ALDH1⁺ cells did not. These data imply that CSC properties could be the seeds for tumor growth and the key factors for therapy resistance. Our previous finding has indicated that YAP1 is the major contributor to CSCs, and targeting YAP1 could be an effective method to target CSCs (22, 24). In the present study, we propose that targeting the YAP1–CDK6 axis could be of value. Inducing radiation resistance in cells enriches them with CSCs, whereas the combination of CA3 and LEE001 reduced number of CSCs reflected by lowered tumor sphere formation and reduced fraction of cells labeling for CD133⁺, CD44⁺, and/or ALDH1⁺.

It is well established that the sensitivity to irradiation is cell-cycle phase dependent (35–37). Radiation-induced lesions such as single-strand breaks and inter-strand crosslinks that are more toxic in the S phase but relatively nontoxic in the G₁ phase (38). The intrinsic resistance to DNA-damaging agents and an increased rate of DNA repair result from G₁ arrest that may enhance radioresistance (39, 40). We showed that selective pharmacologic inhibition of CDK6 induced by LE001 in esophageal cancer leads to G₁ arrest, which is consistent with the prior reports (41). Further, downregulation of CDK6 by lentiCrispr/Cas9 sensitized the YAP1-induced radiation-resistant esophageal cancer cells.

Whether the YAP1/CDK6 axis is operative in both major histologic phenotypes of esophageal cancer remains unclear. ESCC and esophageal adenocarcinoma phenotypes of esophageal cancer are genomically distinct in many aspects as noted in the TCGA analyses (42). Clinically, ESCC appears to be more sensitive to chemotherapy and chemoradiation than esophageal adenocarcinoma, but there are no distinct initial therapies available for these 2 phenotypes; however, the modern clinical trials no longer combine these 2 histologies. Acknowledging considerable genetic diversity in the 2 phenotypes, our results suggest that the YAP-CDK6 crosstalk may be operative in some cases of both EAC and ESCC since the regulation of YAP1 on CDK6 had similar (preclinical) outcomes in both phenotypes when we genetically overexpressed knockdown of YAP1 in both (EAC = SKGT-4 and JHESO and ESCC = KATO-TN and YES-6). We would emphasize a biomarker-based rational clinical trial (using YAP1 and CDK6 overexpression to enrich patients). We would also like to highlight that the empiric therapeutic strategies that are so prevalent in the clinics could be reduced by in-depth molecular analyses of esophageal cancer.

In conclusion, our data demonstrated that both YAP1 and CDK6 are often overexpressed in resistant esophageal cancer tissues (P2) compared with relatively sensitive esophageal cancer tissues (P1), and YAP1 upregulated the CDK6 overexpression and transcription in esophageal cancer cells. Positive crosstalk of the YAP1 and CDK6 pathways mediated radiation resistance, but the combined inhibition of YAP1 and CDK6 provided best antitumor effect in vitro and in vivo, particularly in radiation-resistant cells. Thus, our results provide a strong rationale for a clinical trial for patients with resistant esophageal cancer with an activated YAP1–CDK6 axis while enriching the patient population based on these biomarkers.

No potential conflicts of interest were disclosed.

Conception and design: S. Song, J.A. Ajani

Development of methodology: F. Li, W. Zhao, M. Pool Pizzi, M. Xie, H.D. Skinner, S. Song, J.A. Ajani

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F. Li, Y. Xu, B. Liu, P.K. Singh, J. Jin, A.W. Scott, X. Dong, Y. Wang, K. Harada, H.D. Skinner, S. Krishnan, R.L. Johnson, J.A. Ajani

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): F. Li, G. Han, L. Huo, L. Ma, Y. Li, M. Xie, L. Wang, S. Krishnan, S. Song, J.A. Ajani

Writing, review, and/or revision of the manuscript: F. Li, P.K. Singh, M. Xie, H.D. Skinner, S. Krishnan, R.L. Johnson, S. Song, J.A. Ajani

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B. Liu, X. Dong, Y. Li, S. Ding, S. Song

Study supervision: S. Song, J.A. Ajani

We would like to acknowledge Donald R. Norwood, in the Department of Scientific Publications of UT MD Anderson Cancer Center for her English revision on this article. This work was supported by Public Health Service Grant DF56338 which supports the Texas Medical Center Digestive Diseases Center (S. Song); an MD Anderson Institutional Research Grant (3-0026317, to S. Song); grants from Department of Defense (CA160433, to S. Song); and NIH (CA129906, CA138671, and CA172741, to J.A. Ajani). This work was also supported by NIH/NCI under award number P30CA016672 and used the Flow Cytometry and Cellular Imaging Core Facility.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.