Combined CDK4/6 and Pan-mTOR Inhibition Is Synergistic Against Intrahepatic Cholangiocarcinoma Free

Intrahepatic cholangiocarcinoma (ICC) is the second most frequent form of primary liver cancer after hepatocellular carcinoma (HCC; refs. 1–3). Epidemiologic data indicate a steady increase of ICC incidence over the past decades worldwide (1). Most ICC patients do not show early symptoms and are often diagnosed at advanced stages, thus precluding potentially effective therapies such as surgical resection or liver transplantation (4). The median survival time of patients with unresectable ICC without treatment is shorter than 5 months, and it can be expanded up to approximately 1 year by systemic chemotherapy (5). The combination of gemcitabine and platin-based drugs is the standard first line treatment for patients with advanced or metastatic ICC (2). However, this regimen has very limited therapeutic value, with drug resistance rapidly developing (2). Thus, new and effective therapeutic strategies are urgently needed for this lethal malignancy.

Deregulation of cell cycle induces unconstrained cell division, which leads to continuous proliferation and represents a crucial driver of carcinogenesis (6–8). The cyclin/cyclin-dependent kinase (CDK)/retinoblastoma (Rb) axis is an important regulator of the cell cycle (9). In physiologic conditions, unphosphorylated (active) Rb protein (encoded by the Rb1 gene) blocks cell-cycle progression by sequestrating the E2F1 transcription factor through physical interaction (10). In cancer cells, mitogens and cytokines trigger the upregulation of the Cyclin D1 (CCND1) protein. Subsequently, CCND1, in complex with CDK4/6, inactivates the Rb protein via phosphorylation at multiple serine residues (10, 11). Consequently, the ability of Rb to repress E2F1 is impaired, resulting in the induction of E2F target genes, which are responsible for G₁–S transition and, thus, unconstrained cell proliferation (10). In light of this body of evidence, it is not surprising that cell-cycle inhibitors have been developed and shown to be effective against various cancer types (12). Among these inhibitors, palbociclib is a highly selective and orally active CDK4/6 suppressor able to induce cell-cycle arrest in vitro and in vivo (6, 13). To date, palbociclib has demonstrated significant antineoplastic activity on multiple tumor entities, including human breast, colon, lung, and bladder cancers, as well as hepatocellular carcinoma and leukemia, which are all characterized by an intact Rb1 gene (14–16). The importance of palbociclib as a clinically relevant antitumor drug is underscored by its approval from the FDA for the treatment of hormone receptor-positive/HER2-negative advanced breast cancer (17). Although ICC is rarely affected by genetic alterations at the Rb1 gene and its locus (18), the therapeutic significance of palbociclib or other CDK4/6 inhibitors for ICC treatment has not been determined so far.

PI3K/v-AKT murine thymoma viral oncogene homolog (AKT)/mammalian target of rapamycin (mTOR) cascade is a critical pathway regulating various cellular processes including cell proliferation, survival, and metabolism, and is implicated in cancer development and/or progression (19). Aberrant activation of the PI3K/AKT/mTOR cascade has been detected in most human ICCs (20, 21). MLN0128 is a second-generation pan-mTOR inhibitor, possessing significant anti-cancer growth activity on multiple tumor types (22). MLN0128 is currently under evaluation in several Phase I and II clinical trials as a single agent or in combination therapy (https://clinicaltrials.gov/). In a recent study, using a murine ICC model induced by activated/oncogenic forms AKT and Yap (AKT/YapS127A), we found that MLN0128 treatment results in a stable disease (21). Mechanistically, MLN0128 induced elevated cell apoptosis in AKT/YapS127A cholangiocellular lesions, while only marginally affecting their proliferation properties (21).

Here, we determined whether palbociclib administration possesses antiproliferative proliferative activity toward ICC cells in vitro and in vivo. In addition, we evaluated whether palbociclib synergizes with the mTOR inhibitor MLN0128 to induce ICC regression. For this purpose, palbociclib and MLN0128, either alone or in combination, were administered to ICC cell lines and AKT/YapS127A mice. Noticeably, although treatment with palbociclib alone slowed ICC proliferation, the combination therapy had a striking effect in inhibiting tumor growth. At the cellular level, the enhanced growth constraint achieved by palbociclib/MLN0128 combination was mainly due to strong inhibition of tumor cell proliferation. To our knowledge, this is the first body of evidence demonstrating the important therapeutic effect of palbociclib and, particularly, palbociclib/MLN0128 combination, on ICC growth in vitro and in vivo. Thus, the present study underlines the antineoplastic efficacy of palbociclib/MLN0128 treatment in experimental models of ICC, envisaging its usefulness as a new therapeutic option for the human disease.

The human KKU-M213, huCC-T1, SNU1196, and MzChA-1 ICC cell lines, after validation (Genetica DNA Laboratories, Burlington, NC), were used in the study. Cells were grown in a 5% CO₂ atmosphere, at 37°C, in RPMI Medium supplemented with 10% FBS (Gibco) and penicillin/streptomycin (Gibco). All experiments were repeated at least three times.

The constructs used for mouse injection, including pT3-EF1α, pT3-EF1α-HA-myr-AKT (mouse), pT3-EF1α-YapS127A (human), and pCMV/sleeping beauty transposase (SB) have been described previously (23–25). Plasmids were purified using the Endotoxin-free Maxi Prep Kit (Sigma-Aldrich) before injection. Palbociclib (LC Laboratories) was formulated in 0.5% Tween 80 and 0.5% carboxymethylcellulose (CMC) in purified water to a concentration of 20 mg/mL and stored at −20°C. MLN0128 (LC Laboratories) was dissolved in NMP (1-methyl-2-pyrrolidinone; Sigma-Aldrich) to generate a stock solution of 20 mg/mL. It was diluted 1:200 into 15% PVP/H₂O (PVP: polyvinylpyrrolidone K 30, Sigma-Aldrich; diluted in H₂O at a 15.8:84.2 w/v ratio) before administration to the mice. Palbociclib and MLN0128 were dissolved in DMSO for in vitro experiments.

Eight human ICC cell lines (KKU-M213, huCC-T1, SNU1196, MzChA-1, RBE, TGBC, OCUG-1 and KMCH) were used for in vitro studies. Cell lines were maintained as monolayer cultures in Dulbecco's modified Eagle medium supplemented with 10% FBS (Gibco), 100 U/mL penicillin, and 100 g/mL streptomycin (Gibco). For colony formation assay, cells were plated in 6-well culture plates at a density of 0.5–1 × 10³ cells/well when cells reached 70% to 80% confluency in 60 × 15-mm culture dishes and treated with indicated doses of palbociclib. Equivalent amounts of DMSO-treated cells were used as control. Two to three weeks later, colonies were stained with crystal violet and then counted for quantification and colony formation IC₅₀ determination. For IC₅₀ determination in short-term palbociclib treatment, cells were seeded in 96-well plates and treated with increasing doses of palbociclib in triplicate for 72 hours. Subsequently, cells were tested by BrdUrd Cell Proliferation kit (Merck Millipore). OD was measured at 450 nm with the BioTek ELx808 Absorbance Microplate Reader (Thermo Fisher Scientific).

For the bromodeoxyuridine (BrdUrd) incorporation assay, control or drug-treated cells were incubated with BrdUrd for 1 hour and the assay was performed using the FITC BrdUrd Flow Kit (BD Biosciences), following the manufacturer's instructions. Briefly, the cells were fixed after removing the medium with BrdUrd. Then, DNase was used to expose incorporated BrdUrd. Next, the anti-BrdUrd antibody was added and bound to newly synthesized cellular DNA, which is labeled with BrdUrd. 7-AAD was used for the total DNA staining. Measurement of cell-cycle parameters was performed with the Becton Dickinson LSRII Flow Cytometer (BD Biosciences) and data processed using the FlowJo 10 software (FlowJo, LLC).

Validated Gene Expression Assays for human Rb1 (ID # Hs01078066_m1), CDK4 (ID # Hs00364847_m1), CDK6 (ID # Hs01026371_m1), E2F1 (ID # Hs00153451_m1), Cyclin E/CCNE1 (ID # Hs01026536_m1), and β-Actin (ID # 4333762T) genes were purchased from Applied Biosystems. PCR reactions were performed with 100 ng of cDNA of the collected samples or cell lines, using an ABI Prism 7000 Sequence Detection System with TaqMan Universal PCR Master Mix (Applied Biosystems). Cycling conditions were: denaturation at 95°C for 10 minutes, 40 cycles at 95°C for 15 seconds, and then extension at 60°C for 1 minutes. Quantitative values were calculated by using the PE Biosystems Analysis software and expressed as N target (NT). NT = 2^−ΔCt, wherein ΔC_t value of each sample was calculated by subtracting the average C_t value of the target gene from the average C_t value of the β-actin gene.

The nature of the combinatory effects of palbociclib and MLN0128 was quantitatively evaluated by the Chou–Talalay method (26, 27). Cells were exposed to various doses of palbociclib and/or MLN0128. The overall suppressive actions were used to calculate the combination Index (CI) using the following formula:

where |{\rm{CA}},{\rm{\ x}}$| and |{\rm{CB}},{\rm{\ x}}$| are the doses of each compound alone used in combination to produce x% inhibitory action and |{\rm{ICx}},{\rm{\ A}}$| and|{\rm{\ ICx}},{\rm{\ and\ B}}$| are the doses for each compound alone to produce the same action. CI < 1 indicates a synergistic effect by the combination, whereas CI = 1 means an additive effect. If CI is greater than 1, it suggests antagonism produced by the combination.

Female wild-type (WT) FVB/N mice were obtained from Charles River Laboratories. Hydrodynamic injection was performed as described previously in detail (28). To generate the ICC model, 20 μg pT3-EF1α-HA-myr-AKT and 30 μg pT3-EF1α-YapS127A and 2 μg pCMV/SB were injected in FVB/N mice. MLN0128 (0.5 mg/kg/d), palbociclib (100 mg/kg/d), MLN0128 + palbociclib or vehicle were orally administered via gavage. For early-stage MLN0128 therapy model, therapy began 3.5 weeks postinjection for 3 consecutive weeks, and mice were sacrificed 6.5 weeks after hydrodynamic injection. For histopathologic analysis, liver specimens were fixed in 4% paraformaldehyde and embedded in paraffin. Sections were done at 5 μm in thickness. Preneoplastic and neoplastic liver lesions were assessed by two board-certified pathologists and liver experts (M. Evert and K. Utpatel). Mice were housed, fed, and monitored in accordance with protocols approved by the Committee for Animal Research at the University of California, San Francisco (San Francisco, CA).

Antigen unmasking was achieved by placing the slides in a microwave oven on high for 10 minutes either in 10 mmol/L sodium citrate buffer (pH 6.0) or 1 mmol/L ethylenediaminetetraacetic acid (EDTA; pH 8.5), followed by a 20-minutes cool down at room temperature. After a blocking step with the 5% goat serum and Avidin-Biotin blocking kit (Vector Laboratories), the slides were incubated with primary antibodies overnight at 4°C. Slides were then subjected to 3% hydrogen peroxide for 10 minutes to quench endogenous peroxidase activity and, subsequently, the biotin-conjugated secondary antibody was applied at a 1:500 dilution for 30 minutes at room temperature. The primary antibodies against CK19 (Abcam; 1:150), CCNE1 (Abcam; 1:100); phosphorylated/inactivated Rb (Ser807/8011; Cell Signaling Technology; 1:50); CDK4, cleaved caspase 3 (Cell Signaling Technology; 1:100), and Ki-67 (Thermo Fisher Scientific; 1:150) were used. Immunoreactivity was visualized with the Vectastain Elite ABC kit (Vector Laboratories) and 3,3′-diaminobenzidine as the chromogen. Slides were counterstained with hematoxylin. Proliferation and apoptosis indices were determined in mouse ICC lesions by counting Ki-67 and Caspase 3–positive cells, respectively, on at least 3,000 tumor cells per mouse sample.

A collection of frozen ICC samples (n = 32) was used in the present study. The clinicopathologic features of patients with liver cancer are summarized in Supplementary Table S1. ICC specimens were collected at the Medical Universities of Greifswald (Greifswald, Germany) and Sassari (Sassari, Italy). Institutional review board approval was obtained at the local ethical committee of the Medical Universities of Greifswald and Sassari. Informed consent was obtained from all subjects.

GraphPad Prism version 6.0 (GraphPad Software Inc.) was used to evaluate statistical significance by the Tukey–Kramer test. Data are presented as Means ± SD. Comparisons between two groups were performed with two-tailed unpaired t test. P values < 0.05 were considered statistically significant.

First, we evaluated the importance of the CDK4/6 pathway in human ICC specimens. Thus, we assessed the levels of Rb1, CDK4, CDK6 as well as of E2F1 and the E2F1-specific target Cyclin E (CCNE1) in a collection of human ICC samples and corresponding noncancerous surrounding livers by quantitative reverse transcription real-time polymerase chain reaction (qRT-PCR; Fig. 1A). Of note, a strong upregulation of CDK4/6, E2F1, and CCNE1 expression was detected in ICC specimens when compared with nonneoplastic counterparts. In addition, Rb1 levels were significantly higher in tumors when compared with nontumorous livers. Furthermore, higher levels of CDK4/6, Rb, phosphorylated/inactivated Rb, and Cyclin E were observed at the protein level using Western blotting in the same sample collection (Fig. 1B; Supplementary Fig. S1). The results were further confirmed by immunohistochemical staining on the same specimens. Indeed, increased nuclear immunoreactivity for phosphorylated/inactivated Rb, CDK4, and CCNE1 was detected in 81.2%, 78.1%, and 87.5% ICC samples, respectively, when compared with corresponding nontumorous liver tissues (n = 32; Fig. 1C). Subsequently, the Rb1, CDK4, CDK6, E2F1, and CCNE1 mRNA expression data were retrieved from The Cancer Genome Atlas (frozen on 2/25/2015; and consisting of surgically resected 36 ICCs and 59 surrounding nontumorous liver tissues (29, 30) and NCI (from NCBI GEO website: GSE26566; consisting of 104 ICC and 59 non-neoplastic liver tissues; (29, 30) datasets. Both datasets showed a significant upregulation of CDK4, E2F1, and CCNE1 mRNA levels in human ICC specimens compared with surrounding noncancerous livers (Supplementary Fig. S2B, S2D, and S2E). Rb1 mRNA levels were also slightly upregulated in human ICC specimens when compared with nontumorous counterparts (Supplementary Fig. S2A), whereas the expression of CDK6 in ICC was weakly upregulated only in the NCI database (Supplementary Fig. S2C). The present findings emphasize the almost ubiquitous activation of the CDK4/6 pathway in ICC patients, thus potentially representing an effective target in this disease.

Next, we used palbociclib, a CDK4/6 inhibitor, to investigate whether inhibition of the CDK4/6 pathway affects ICC cell growth in vitro. For this purpose, eight ICC cell lines were treated with various concentrations of palbociclib for 72 hours followed by BrdU-based cell proliferation analysis. Noticeably, most ICC cells tested were rather sensitive to palbociclib, with IC₅₀ ranging between 0.11 and 1.34 μmol/L (Supplementary Fig. S3). As palbociclib has been shown to inhibit tumor growth via inducing cell-cycle arrest (6, 13), we investigated the expression of cell cycle–related proteins in response to palbociclib treatment in the ICC cell lines (Fig. 2A; Supplementary Fig. S4A). Palbociclib effectively suppressed the levels of phosphorylated/inactivated retinoblastoma protein (p-Rb) in all cell lines tested. In addition, protein levels of cell cycle–related proteins, including Cyclin A, B, E, and PCNA, were repressed in most cell lines, whereas CCND1 increased in palbociclib-treated cells. Cell cycle–negative regulators, including p21 and p53, did not show consistent changes of expression in response to palbociclib (Fig. 2A; Supplementary Fig. S4A).

Next, we investigated whether palbociclib modulates the expression of major pathways regulating tumor growth, specifically, the PI3K/AKT/mTOR and Ras/MAPK cascades (Fig. 2B; Supplementary Fig. S4B). No consistent changes in protein expression patterns in the two signaling cascades were observed in palbociclib-treated ICC cells, thus suggesting that these pathways do not represent major targets of palbociclib in this tumor type.

Altogether, the data demonstrate that palbociclib inhibits ICC cell growth in vitro by inducing cell-cycle arrest and reactivating/dephosphorylating the Rb protein.

Interestingly, the in vitro data revealed an upregulation of Cyclin D1 (CCND1) in palbociclib-treated ICC cell lines (Fig. 2A). Previous studies showed that palbociclib treatment may lead to increased CCND1 protein levels by stabilization of CDK4/CCND1 complex (31, 32). Our intriguing observation prompted us to hypothesize that induction of CCND1 might be a key mechanism whereby ICC cells continue to grow despite CDK4/CDK6 inhibition. Thus, we determined whether silencing of CCND1 enhances the sensitivity of human ICC cell lines to palbociclib. For this purpose, the CCND1 gene was suppressed via specific siRNA in KKU-M213 and SNU1196 cells, randomly selected from the panel of ICC cells (Supplementary Fig. S5). Of note, concomitant silencing of CCND1 and palbociclib administration resulted in an increased anti-growth effect in both cell lines, whereas inhibition of CCND1 expression alone did not significantly modify the growth rate of the same cells (Supplementary Fig. S5).

The present findings suggest that CCND1 induction might represent a major mechanism of resistance by ICC cells to the growth inhibitory effects of palbociclib.

Subsequently, we tested the hypothesis that Rb expression may be used as a biomarker for palbociclib sensitivity. Thus, we analyzed Rb expression in the 8 ICC cell lines (Fig. 3A). A significant negative correlation between Rb protein levels and IC₅₀ against palbociclib was detected (Supplementary Fig. S6). As the growth inhibition assay was a short-term assay, with cell growth rate being analyzed 72 hours after addition of palbociclib to the growth medium, we hypothesized that Rb levels might also affect the long-term growth inhibitory activity of palbociclib. To validate this hypothesis, we performed colony formation assay, and the number of colonies was analyzed approximately 2 weeks after palbociclib or solvent (DMSO) treatment (Fig. 3B; Supplementary Fig. S7). Consistent with BrdUrd-based short-term proliferation assay, all 3 cell lines with low Rb expression (TGBC, OCUG-1, and KMCH) showed higher IC₅₀ against palbociclib, whereas the 5 cell lines with high Rb expression (KKU-M213, huCC-T1, SNU1196, MzChA-1, and RBE) showed low IC₅₀ (Fig. 3C and D) in colony formation assays.

To further investigate whether Rb levels determine long-term growth sensitivity to palbociclib, we silenced the Rb1 gene in four cell lines (KKU-M213, huCC-T1, SNU1196, and MzChA-1) with a previously validated shRNA against Rb (shRb; ref. 33). As expected, Rb1 expression was efficiently inhibited in all shRb-transfected cells. In colony formation assays, Rb suppression conferred strong resistance to palbociclib in the 4 cell lines tested (Fig. 3E; Supplementary Fig. S8). For instance, in KKU-M213 cells, palbociclib treatment effectively inhibited cell proliferation of shRNA control (shCtr) but not shRb-treated cells (Fig. 3E).

In summary, the present data suggest that depletion of Rb renders ICC cells resistant to palbociclib.

Previously, we demonstrated that mTOR inhibitors, such as MLN0128, are valuable drugs for restraining ICC cell growth. However, MLN0128 was found to be effective in promoting cell apoptosis while displaying limited antiproliferative capacity (21). Because palbociclib is a strong cell-cycle inhibitor (6, 13), we hypothesized that it might synergize with MLN0128 to suppress ICC cell growth. Five ICC cell lines were treated with MLN0128 and palbociclib, either alone or in combination. Intriguingly, concomitant treatment of ICC cells with MLN0128 and palbociclib resulted in stronger growth inhibitory activity (Fig. 4A; Supplementary Fig. S9). Subsequently, the Chou–Talalay method was used to calculate the combination index (CI; refs. 26, 27). Strikingly, the CI value was lower than 1 in the four ICC cell lines when MLN0128 and palbociclib were concomitantly administered (Fig. 4B; Supplementary Fig. S10). These data indicate a synergistic, antiproliferative effect of combining palbociclib with MLN0128 on ICC cells.

At the cellular level, using FACS analysis, palbociclib elicited cell-cycle arrest, whereas MLN0128 had limited effect on cell cycle (Fig. 4C; Supplementary Fig. S11). Strikingly, combined palbociclib and MLN0128 treatment resulted in nearly complete cell-cycle arrest.

At the molecular level, palbociclib/MLN0128 combination induced an additional decline of Rb phosphorylation, and decreased the levels of cyclin A, cyclin B, and PCNA proteins when compared with palbociclib alone (Fig. 4D; Supplementary Fig. S12A). Noticeably, palbociclib treatment alone triggered CCND1 induction, whereas combined palbociclib/MLN0128 administration prevented this phenomenon. Moreover, CDK inhibitor proteins p21 and p16 were upregulated in the palbociclib/MLN0128 combination treatment group in several ICC cell lines. As concerns the Ras/MAPK pathway, palbociclib and MLN0128, either alone or in combination, did not show consistent effects on phosphorylated/activated (p-)ERK levels (Fig. 4E; Supplementary Fig. S12B). In contrast, MLN0128 alone was effective against the AKT/mTOR cascade, an effect that was augmented by the combination treatment (as evidenced by the strong decline of phosphorylated/activated (p-)RPS6 and phosphorylated/inactivated (p-)4EBP1) in ICC cell lines.

In conclusion, the data indicate that palbociclib synergizes with MLN0128 to constrain ICC cell growth in vitro.

Recently, we established a new murine ICC model driven by activated AKT and Yap, two oncoproteins frequently activated in human ICC (AKT/YapS127A; ref. 21). In particular, we found that the standard chemotherapy for human ICC, consisting of gemcitabine plus oxaliplatin, had limited therapeutic value in AKT/YapS127A ICC lesions, whereas treatment of MLN0128 resulted in stable disease in these mice (21). Here, we tested the therapeutic potential of palbociclib, either alone or in combination with MLN0128, in AKT/YapS127A mice. Thus, AKT/YapS127A mice were monitored for tumor growth until 3.5 weeks postinjection, when the tumor burden was moderate (average liver weight ∼3.5 g; Fig. 5A and C). Subsequently, AKT/YapS127A mice were randomly separated into 5 cohorts. The first cohort was harvested as “pre-treatment” group, while the remaining mice were treated with vehicle, palbociclib, MLN0128, or palbociclib/MLN0128 for additional 3 weeks. Using mouse total body weight as measurement of overall health stature of mice, we tested different combination concentration ratio and found that 100 mg/kg palbociclib (16) plus 0.5 mg/kg MLN0128 was well-tolerated and this dose was used for successive in vivo studies. It should be underlined that the dose of MLN0128 was lower (0.5 mg/kg) in the combination therapy than that in our previous report when MLN0128 was used as monotherapy (1 mg/kg; ref. 21). In the vehicle-treated cohort, all mice developed lethal burden of tumor and had to be euthanized by 4.7 to 6.5 weeks postinjection (Fig. 5B). Liver tumors in palbociclib or MLN0128-treated cohorts grew slowly, and only a portion of mice developed high tumor burden and were euthanized before the end of experiments. Still, approximately 60% and 40% of mice survived after 3 weeks of treatment in palbociclib and MLN0128 monotherapy cohorts, respectively (Fig. 5B). In contrast, all palbociclib/MLN0128-treated mice showed good health, and only small tumors developed (Fig. 5B). Using total liver weight as the measurement of tumor burden, the vehicle cohort had the highest tumor burden. Palbociclib and MLN0128 monotherapy cohorts showed significantly lower tumor burden than vehicle-treated mice. Yet, mice subjected to single treatment with palbociclib or MLN0128 exhibited higher tumor burden compared with the pre-treatment cohort, indicating that tumors subjected to monotherapy continued to grow despite the treatment (Fig. 5C). Palbociclib/MLN0128-treated mice displayed the lowest tumor burden, even lower than that in the pre-treatment cohort (Fig. 5C). Histopathologic analysis revealed that liver tumors in all cohorts were pure ICC (Fig. 5D), as confirmed by diffuse immunoreactivity for the biliary marker CK19 (Fig. 5E). As expected, liver CK19-positive areas were smallest in the palbociclib/MLN0128 group (Fig. 5E).

Altogether, the data indicate that palbociclib or MLN0128 monotherapy has moderate therapeutic efficacy against AKT/YapS127A ICC, whereas combined palbociclib/MLN0128 treatment leads to tumor regression.

Because the in vitro findings indicate that combined palbociclib/MLN0128 treatment strongly inhibits ICC cell proliferation, we investigated whether the tumor inhibitory activity induced by the combined therapy is driven by the same mechanism in vivo. Thus, we analyzed the proliferation indices in the 5 cohorts using Ki-67 immunohistochemistry. MLN0128 treatment had no impact on cell proliferation rate, in accordance with previous findings (21), whereas palbociclib alone mildly (although significantly) inhibited tumor cell proliferation at the dose used in the study (Fig. 5F). In striking contrast, combined palbociclib/MLN0128 treatment profoundly decreased tumor cell proliferation (Fig. 5F). No synergistic effect on apoptosis was observed in the palbociclib/MLN0128-treated group (Supplementary Fig. S13).

At the molecular level, consistent with the in vitro data, p-Rb levels were significantly downregulated in palbociclib-treated liver tumors, being virtually undetectable in the palbociclib/MLN0128-treated lesions (Fig. 6A). Palbociclib and palbociclib/MLN0128 treatment also led to decreased expression of PCNA and increased levels of p16. Importantly, palbociclib monotherapy triggered the induction of CCND1, which was instead prevented by palbociclib/MLN0128 treatment (Fig. 6A; Supplementary Fig. S14A). As concerns apoptosis-related proteins (Fig. 6B; Supplementary Fig. S14B), palbociclib and MLN0128, alone or in combination, triggered the increase of activated/cleaved-Caspase 3 and 7 levels. Palbociclib and MLN0128 treatment did not affect Ras/MAPK activity, whereas the PI3K/AKT/mTOR cascade was efficiently inhibited by MLN0128 and palbociclib/MLN0128 (Fig. 6C; Supplementary Fig. S14C). As in the in vitro studies, the combination treatment resulted in the most pronounced inhibition of p-RPS6 in AKT/YapS127A ICC tumor cells.

In summary, our findings demonstrate that combined palbociclib/MLN0128 treatment strongly inhibits tumor cell proliferation, leading to tumor regression, in AKT/YapS127A mice.

ICC is a highly fatal tumor with rising incidence and lacking effective therapies. Thus, novel and successful therapeutic approaches against ICC are urgently needed. The molecular pathogenesis of ICC is complex, involving multiple signaling pathways and molecular alterations that contribute to tumor cell proliferation, survival, migration, and metastasis (34, 35). Given the complexity of the oncogenic mechanisms responsible for ICC development and progression, it is likely that combination therapies simultaneously targeting different key molecules might possess more pronounced antitumor efficacy. In agreement with this hypothesis, combination therapies have emerged as a promising antineoplastic strategy and are now widely used at the clinical level (36–38). In particular, combination therapies targeting multiple pathways might overcome molecular mechanisms of drug resistance in tumor cells when compared with monotherapy, and the lower dose used might reduce the drug toxicity (38).

As a possible targetable signaling cascade, our analysis shows that the CDK4/6 axis is activated in most ICC patients. In particular, Rb is hyperphosphorylated (a reversible event), but only rarely downregulated, in human ICC. This important evidence, together with previous data showing very rare genetic alterations of the Rb1 gene in human ICC (18), indicate that the large majority of ICC patients could positively respond to treatments aimed at suppressing the CDK4/6 proteins.

In a previous investigation, we reported that the dual mTOR inhibitor MLN0128 displays a significant growth inhibitory activity in both human ICC cell lines and the AKT/YapS127A ICC mouse model, primarily via inducing apoptosis (21). However, MLN0128 failed to repress the sustained proliferation characteristic of AKT/YapS127A mice, thus resulting finally in decreased growth pace but not regression of ICC lesions. In light of these findings, we postulated that the persistent proliferation should be hampered to achieve effective tumor regression in AKT/YapS127A mice. Thus, to hinder AKT/YapS127A-dependent proliferation, we applied the CDK4/6 inhibitor palbociclib, a strong cytostatic agent (6, 13–16), in association with MLN0128. As hypothesized, we show here that combined treatment with palbociclib and MLN0128 induces a more pronounced ICC growth inhibition compared with monotherapies, both in vitro and in vivo. On the basis of the results from the present report, it can be envisaged that this combination treatment might represent a potential new therapy for ICC. To the best of our knowledge, this is the first study showing a therapeutic effect of palbociclib (alone and in combination with MLN0128) in ICC.

The importance of the functional crosstalk between CDK4/6 and mTOR inhibitors in cancer is underscored by a body of experimental evidence. For instance, it has been shown that mTOR activation is sufficient to render cells resistant to CDK4/6 inhibitors, whereas mTOR suppression greatly improves the palbociclib anti-growth properties in pancreatic cancer (39). In particular, palbociclib was found to sensitize pancreatic cancer cells to mTOR inhibitors as well as to enhance its own anti-growth function via a positive feedback loop (39). Likewise, here we show that MLN0128 administration further augments the cytostatic function over ICC cells by palbociclib that, in parallel, promotes MLN0128 anti-mTOR activity. Indeed, although palbociclib alone has no obvious suppressive effects on mTOR in ICC cells, the present data indicate that palbociclib sensitizes ICC cells to mTOR inhibition by MLN0128, with consequent more pronounced downregulation of mTOR downstream effectors. This effect might increase the antiproliferative capacity of palbociclib in ICC cells. In breast cancer cell lines, it has been found that they rapidly adapt to CDK4/6 inhibition by induction of CCND1. Once again, such a resistance was successfully prevented by concomitant administration of PI3K/AKT/mTOR inhibitors (40). Noticeably, our present data indicate that CCND1 upregulation is a mechanism of resistance to CDK4/6 blockade also in ICC cells, and this event can be circumvented by mTOR inhibition. Thus, it is tempting to speculate that induction of CCND1 and/or the PI3K/AKT/mTOR pathway might be a major mechanism of resistance in multiple cancer types that should be considered (and prevented) when administering CDK4/6 inhibitors to patients.

In summary, the present investigation indicates that the CDK4/6 axis and the mTOR pathway are concurrently induced and functionally interact in ICC to sustain unrestrained proliferation of cancer cells. Combined suppression of the two pathways is highly detrimental for ICC growth in vitro and in vivo and prevents acquired tumor cell resistance to monotherapy. Thus, our data strongly support further clinical evaluation of combination therapies with CDK4/6 and mTOR inhibitors for the treatment of human ICC.

No potential conflicts of interest were disclosed.

Conception and design: J.D. Gordan, R.M. Pascale, D.F. Calvisi, X. Chen

Development of methodology: S. Zhang, R.M. Pascale, H. Hu, X. Chen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H. Wang, J. Wang, A. Cigliano, M.G. Pilo, M. Serra, S. Zhang, A. Cossu, A. Porcu, R.M. Pascale, M. Evert

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): X. Song, X. Liu, Y. Qiao, K. Utpatel, S. Ribback, J.D. Gordan, R.M. Pascale

Writing, review, and/or revision of the manuscript: X. Song, Y. Qiao, K. Utpatel, J.D. Gordan, R.M. Pascale, F. Dombrowski, D.F. Calvisi, X. Chen

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Wang, L. Che, H. Hu, M. Evert

Study supervision: R.M. Pascale, F. Dombrowski, X. Chen

The authors would like to thank Dr. Gregory J. Gores at Mayo Clinic (Rochester, MN) for providing the huCC-T1 cell line. This work is supported by NIH grants R01CA136606 and R01CA190606 (to X. Chen); P30DK026743 for UCSF Liver Center; China Scholarship Council (201506350124; to X. Song); and National Natural Science Foundation of China (NSFC; 31671945; to H. Hu).

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.