Autophagy Induction with RAD001 Enhances Chemosensitivity and Radiosensitivity through Met Inhibition in Papillary Thyroid Cancer

This article is featured in Highlights of This Issue, p. 1173

Thyroid cancer is the most common endocrine malignancy and is responsible for the majority of deaths from endocrine cancer (1). Papillary thyroid cancer (PTC) accounts for approximately 80% of thyroid cancers. Although PTC is generally associated with favorable outcomes after surgical resection and radioactive iodine ablation, it is the most common histologic subtype in patients with locally advanced disease (2, 3). Doxorubicin is the only Food and Drug Administration–approved chemotherapy agent for advanced PTC that does not trap radioactive iodine; however, intrinsic resistance to doxorubicin results in limited efficacy and it is rarely used for these patients (4).

RAD001 (everolimus, Afinitor), a rapamycin analogue, is an oral mammalian target of rapamycin (mTOR) inhibitor. mTOR is an intracellular protein that acts as a central regulator of multiple signaling pathways induced by insulin-like growth factor, epidermal growth factor, platelet-derived growth factor, vascular endothelial growth factor, and amino acids that mediate abnormal growth, proliferation, and angiogenesis (5). RAD001 was recently approved for the treatment of papillary renal carcinoma (6), a malignancy driven by an activating Met mutation (7). Although we are now testing the therapeutic efficacy of RAD001 in advanced PTC, its mechanism of action in this context is unknown.

RAD001 stimulates autophagy (8, 9), a highly conserved process that entails the degradation of intracellular components through the lysosomal machinery to regenerate metabolites for energy and growth (10, 11). In many contexts, autophagy promotes cell survival under stressful conditions such as nutrient and growth factor deprivation (12, 13). However, excess or prolonged autophagy could promote an alternative mechanism that leads to programmed cell death, which is particularly relevant in cancers that are defective in apoptosis (14, 15). Prior work has shown that PTC is intrinsically resistant to cytotoxic chemotherapies because of defects in apoptosis (16).

We previously observed that inhibition of autophagy decreased PTC sensitivity to doxorubicin and external beam radiation (17). This result led us to hypothesize that RAD001 could enhance cancer cell response to these therapies through the activation of autophagy. In the current study, we found that the combined effects of RAD001 and doxorubicin together exceeded the sum of their individual effects. To understand the complementary activity of RAD001, we used kinome-wide analysis to determine how RAD001 affects signaling pathways. We found that RAD001 and autophagic activation inhibit Met phosphorylation, an event that imparts the effects of RAD001 on chemosensitivity and radiosensitivity.

Mouse anti-human microtubule-associated protein 1 light chain 3 (LC3) antibody was purchased from MBL International Cooperation. Mouse anti-actin antibody, pan Ab-5 was purchased from Neomarker. Rabbit anti-human Atg-5, Src, phosphorylated Src Tyr⁴¹⁶, Met, phosphorylated Met Tyr¹²³⁴, cleaved caspase-3 Asp¹⁷⁵, phosphorylated histone H3 Ser¹⁰, antibodies were purchased from Cell Signaling. Mouse anti-human Ki-67 antibody was obtained from BD Transduction Laboratories. Rabbit anti-human p62 antibody was from Abgent. Secondary horseradish peroxidase–conjugated mouse and rabbit antibodies were obtained from Vector Laboratories. 3-Methyladenine (3-MA), a class III phosphoinositide-3-kinase inhibitor widely used as autophagy inhibitor reported to act specifically on the autophagic/lysosomal pathway of degradation (13), was purchased from Calbiochem. Doxorubicin was obtained from Fluka. RAD001, an mTOR inhibitor that induces autophagy (18), was obtained from Novartis Pharmaceutical. PHA665752, a selective small molecule inhibitor of Met (19), was obtained from Pfizer, Inc. Dasatinib obtained from Broad Institute was synthesized as previously described (20).

The human PTC cell lines TPC-1 and 8505-C were cultured as described previously (21). TPC-1 cells were obtained from Dr. Orlo H. Clark (University of California at San Francisco, San Francisco, CA) and 8505-C cells were obtained from Dr. Sareh Parangi (Massachusetts General Hospital, Boston, MA) within the last 5 years and tested within the last 6 months by morphology check by microscope and growth curve analysis according to the Cell Line Verification Test Recommendations (American Type Culture Collection, technical bulletin no. 8, 2008). TPC-1 and 8505-C cells were used throughout passages 12 to 18 and 28 to 35 in this study. Cells were subcultured at 80% confluency by trypsinization [in a 0.5% (v/v) trypsin solution supplemented with 0.2% (v/v) EDTA]. Cells were transfected with scrambled short interfering RNA (siRNA) or Atg-5 prevalidated siRNA duplex (the target sequences were no. 1, AACCTTTGGCCTAAGAAGAAA; no. 2, AAGACTTACCGGACCACTGAA; Qiagen) using LipofectAMINE 2000 reagent (Invitrogen), according to the instructions of the manufacturer. All transfections were done in six-well plates and cells were processed 48 hours after transfection.

Autophagosome formation was also confirmed by the appearance of punctate fluorescence from exogenous GFP-fused LC3, a specific lysosomal/autophagic biochemical vacuole marker (10). Cells were transfected with EGFP-LC3 plasmid (Addgene) using LipofectAMINE 2000 reagent. Forty-eight hours after transfection, cells were treated with DMSO and RAD001 (20 nmol/L) for 24 hours. The fluorescence of EGFP-LC3 was observed under a fluorescence microscope.

Human PTC tissues and normal adjacent thyroid tissue were obtained from thyroidectomy specimens resected from surgery patients. The specimens were immediately frozen in liquid nitrogen upon removal from the patients and the samples were stored frozen until the time of protein extraction. Tissue lysates were prepared using radioimmunoprecipitation assay buffer containing protease inhibitor cocktail as described previously (22). The Institutional Review Board of Brigham & Women's Hospital in Boston, MA approved this protocol (protocol number: 1999-P-001331/5).

Cell lysates were prepared and analyzed by Western blotting as described previously (21).

Cultured PTC tissue specimens were washed with serum-free DMEM (Invitrogen) followed by a brief wash in 1 mol/L of cacodylate buffer and then fixed with 2.5% glutaraldehyde with 0.1 mol/L of sodium cacodylate buffer for 1 hour. Fixed cells and PTC tissue specimens were processed at the Harvard Medical School Transmission Electron Microscopy core facility by using protocols described previously (23) using a JEOL-60kV microscope (JEOL). Morphometric analysis was done as previously described (24). For each control or treatment group, 100 cells were randomly chosen from the fields for quantification of autophagic features. Closed vacuoles with double membranes containing cytosol and/or organelles were considered autophagic vacuoles (11). Cells with ≥10 autophagic vacuoles were determined as autophagic cells.

A piece of each human tissue specimen was excised using sterile techniques. Vital specimens (0.5 mm³) were cultured in Millicell cell culture inserts (Millipore) containing 600 μL of DMEM/F12 medium with 15 mmol/L of HEPES (Sigma-Aldrich) supplemented with GlutaMAX I (Invitrogen) at 37°C and 5% CO₂ for 8 hours in the absence or presence of doxorubicin (1 μmol/L). Immunohistochemical analysis of Ki-67, p62, and cleaved caspase-3 was done as described previously (25). p62, a ubiquitin-binding scaffold protein, is degraded by autophagy and is a marker of autophagy (26).

For cell survival determination, standard clonogenic survival assays were done as previously described (27), with minor modifications. Eight hundred cells were plated and then coincubated with doxorubicin (final concentrations of 200, 100, 50, 20, 10, 5, or 1 nmol/L) or medium without doxorubicin and cultured in a humidified CO₂ incubator at 37°C for 10 to 14 days. Cells were fixed for 15 minutes with a 3:1 methanol/acetic acid solution and stained for 15 minutes with 0.5% crystal violet (Sigma-Aldrich) in methanol. After staining, colony numbers were counted. IC₅₀ (a concentration of the compound causing 50% inhibition) values were determined. The mean ± SEM of triplicate samples was determined for each treatment. The data was analyzed by plotting the clonogenic survival rate (number of colonies in treated plates/number of colonies in controls) versus the doxorubicin concentration.

The MTS [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium] assay (Celltiter96 AQ, Promega) was used to assess RAD001 and doxorubicin effects on cell viability as previously described (21).

For radiation experiments, cells were trypsinized, replated, and incubated for an additional 24 hours under standard conditions. Eight hundred cells were plated and then irradiated with doses ranging from 0 to 6 Gy (Gammacell 1000 apparatus, 200 kV, 15 mA, 1 Gy/min; Atomic Energy of Canada, Ottawa, Canada). Forty-eight hours after irradiation, the medium was replaced with fresh medium and then subjected to clonogenic survival assay. The mean ± SD of triplicate samples was determined for each treatment. The data was analyzed by plotting the log of the surviving fraction (number of colonies) versus radiointensity.

We did a bead-based profiling of tyrosine kinase phosphorylation using individual Luminex xMAP microspheres (Luminex) in an immunosandwich assay, as previously described (28). In brief, an antibody-based kinase phosphorylation assay for a panel of 71 unique tyrosine kinases was used to comprehensively evaluate the effects of Atg-5 knockdown and RAD001 treatment. Scrambled or Atg-5 siRNA–transfected cells were washed in ice-cold PBS and harvested in 1× cell lysis buffer (Cell Signaling) supplemented with a cocktail of protease inhibitors (Sigma-Aldrich). The data was acquired with a Luminex FlexMAP 3D instrument (Luminex) according to the instructions of the manufacturer. Subtracting the sample and antibody background signals normalized raw values. Normalized values were considered positive if they were ≥10. The preprocessed data were converted into .gct files and analyzed with GenePattern 3.0 (the Broad Institute, Cambridge, MA).

Differences between treatment groups were evaluated with a two-tailed independent Student's t test. Each assay was done in triplicate, and a value of P < 0.05 was considered statistically significant. We defined positive treatment as “synergistic” when agents had a combined effect on the inhibition of proliferation greater than the sum of their individual effects. To display such activity, we determined the minimum concentration of each agent necessary to achieve a significant effect (IC_min) and then determined the combined activity at the IC_min concentrations of both agents. The correlation between autophagic cell numbers in DMSO- and RAD001-treated cells was analyzed by χ² testing.

To determine if RAD001 induces autophagy in PTC, we treated 8505-C and TPC-1 cells with RAD001 at doses previously shown to inhibit mTOR signaling in cell cultures (18). Then, we assayed the expression level of the LC3-II protein, which is generated when microtubule-associated protein 1A/1B LC3 undergoes autosomal breakdown. Conjugation of ubiquitous cytosolic LC3-I to phosphatidylethanolamine produces the LC3-phosphatidylethanolamine conjugate (LC3-II), which is recruited to autophagosomal membranes and is a reliable marker for autophagic activity. Immunoblotting revealed that 20 nmol/L of RAD001 was required to significantly increase the expression level of the LC3-II protein (Fig. 1A).

We next determined if the ectopic expression of LC3 affected the action of RAD001. PTC cells were transfected with a GFP-labeled LC3 plasmid; we then observed a significant increase in punctate fluorescence in RAD001-treated cells, which reflects the formation of autophagosomes (Fig. 1B). We also determined how siRNA knockdown of Atg-5, which is required for autophagy (29), affected RAD001 activity. Therefore, we used two distinct siRNAs that targeted nonoverlapping regions of the Atg-5 mRNA for this experiment. We found that Atg-5 depletion by the siRNAs, which are designated as nos. 1 and 2, abrogated the RAD001-mediated increase in LC3-II (Fig. 1C).

To further confirm the presence of autophagy, we did electron microscopy to visualize autophagosomes (Fig. 2). Morphometric analysis showed that RAD001 significantly increased the formation of autophagic vacuoles, which are characterized as double-membrane vacuoles containing cytoplasmic contents (24). This further shows that RAD001 induces autophagy in PTC (Table 1).

We next determined if RAD001 induces autophagic activity in human PTC specimens. To that end, we harvested thyroid specimens from patients undergoing resection of thyroid cancer and treated these with RAD001. Autophagy was evaluated by immunoblotting for the LC3-II protein and by electron microscopy. We found that RAD001 induced LC3-II protein expression in three consecutive tumors, demonstrating the ability of RAD001 to induce autophagy in human specimens (Fig. 3A). Likewise, RAD001-treated human ex vivo specimens displayed a substantial increase in autophagosomes that were visualized by electron microscopy, providing further evidence that autophagy is induced by RAD001 (Fig. 3B). RAD001-treated specimens showed a significant decrease in endogenous p62 expression, another marker for autophagy (26), further demonstrating the activation of autophagy in these human operative specimens (Fig. 3A and C). RAD001 did not change caspase-3 cleavage in PTC tissue, indicating that it had no significant effects on apoptosis (Fig. 3D). Together, our in vitro and ex vivo results indicate that RAD001 effectively activates autophagy in PTC.

To determine the effect of RAD001 on doxorubicin chemosensitivity, we first determined a baseline IC₅₀ of doxorubicin of 98.61 and 52.97 nmol/L in 8505-C and TPC-1 cells, respectively (Supplementary Fig. S1). Treatment with 20 nmol/L of RAD001 decreased the IC₅₀ by 48% and 38% in 8505-C and TPC-1 cells, respectively. The effect of RAD001 on chemosensitivity was abrogated with Atg-5 siRNAs, suggesting that the effects of RAD001 on doxorubicin chemosensitivity are mediated through autophagy (Supplementary Fig. S1).

Using the MTS and clonogenic survival assays, we determined that the minimum inhibitory concentration (IC_min) of doxorubicin was 20 and 50 nmol/L in 8505-C and TPC-1 cells, respectively (Fig. 4A and C, left). The IC_min of RAD001 was 5 and 10 nmol/L in 8505-C and TPC-1 cells, respectively (Fig. 4A and C, right). In an MTS assay, the combination of doxorubicin and RAD001, at their IC_min, decreased cell survival by 293% and 299%, in comparison with doxorubicin alone in 8505-C and TPC-1 cells, respectively. The combination of both agents exceeded the additive effects of each individual agent (Fig. 4B). In a clonogenic assay, the combination of doxorubicin with RAD001 at their IC_min decreased cell survival by 319% and 348%, in comparison with doxorubicin alone in 8505-C and TPC-1 cells, respectively (Fig. 4C). Consistent with synergistic activity, the efficacy of both agents exceeded the additive effects of the agents (Fig. 4D).

Because external beam radiation is an important therapeutic modality for patients with locally advanced PTC, we determined if RAD001 could improve radiosensitivity in PTC cells. We treated 8505-C and TPC-1 cells with DMSO (control) or 20 nmol/L of RAD001 and then exposed them to radiation doses ranging from 0 to 6 Gy (Fig. 5A). PTC cells treated with RAD001 had a significant improvement in sensitivity to various doses of radiation. This effect was reversed when cells were treated with Atg-5 siRNA, suggesting that the radiosensitizing effects of RAD001 are mediated through autophagy.

In ex vivo experiments of human PTC, RAD001 activated autophagy, but did not seem to activate apoptosis (Fig. 5B). RAD001, in combination with doxorubicin or radiation, exhibited additive effects on the accumulation of LC3-II and the inhibition of histone H3 phosphorylation. No basal or additive caspase-3 cleavage signals were observed with RAD001 treatment, even when the specimens were treated with doxorubicin or external radiation. Consistent with our in vitro results, RAD001 activated autophagy and potentiated the autophagic activation caused by doxorubicin or external beam radiation in ex vivo specimens.

Because our results suggested that RAD001 has a distinct and complementary mechanism of action with doxorubicin, we determined the effects of RAD001 on cell signaling pathways. We used a kinome array technique, based on the multiplexed coupling of tyrosine kinase–specific antibodies to polystyrene microspheres beads, to simultaneously determine the phosphorylation status of 71 unique tyrosine kinases. In this analysis, Src and Met were the most differentially phosphorylated in scrambled versus Atg-5 siRNA–transfected cells, and in control versus RAD001-treated cells (Fig. 6A). We validated the kinome array results by immunoblotting; this confirmed that Atg-5 siRNA decreased Src but increased Met activity, whereas RAD001 increased Src but decreased Met activity (Fig. 6B).

Because autophagic activation modulated Src and Met, we determined if either or both were essential for the augmentation in chemosensitivity and radiosensitivity observed during autophagy. We treated PTC cell lines with dasatinib, a Src inhibitor, to determine if it reversed the effects of RAD001. RAD001 decreased the IC₅₀ by 36% and 31%, whereas RAD001 plus dasatinib decreased the IC₅₀ by 62% and 52% in 8505-C and TPC-1 cells, respectively. Furthermore, the addition of dasatinib to RAD001 also significantly improved radiosensitivity (Supplementary Fig. S3). Therefore, these results show that dasatinib enhanced, rather than attenuated, the anticancer effects of RAD001.

In contrast, Atg-5 siRNA increased the IC₅₀ by 81% and 71%, whereas Atg-5 siRNA in combination with the Met inhibitor PHA665752 increased the IC₅₀ by only 3% and 6% in 8505-C and TPC-1 cells, respectively. Therefore, PHA665752 abrogated the effects of Atg-5 knockdown on chemosensitivity. Furthermore, RAD001 and PHA665752 induced similar effects on chemosensitivity and radiosensitivity with no significant additive effects in combination (Supplementary Figs. S3A and S4A). Together, these results suggest that the effects of autophagic activation on chemosensitivity and radiosensitivity are mediated largely through Met.

RAD001 potently activates autophagy in PTC cell lines and human PTC specimens maintained ex vivo. RAD001 synergistically enhances both chemosensitivity to doxorubicin and radiosensitivity, and seems to do so through autophagy. That RAD001 elicits these effects is of considerable interest as it inhibits mTOR, a signaling protein that is a convergence point for many signaling pathways. Furthermore, mTOR elicits the cellular response to extracellular stimuli by mediating changes of cellular protein synthesis. That RAD001 acts through autophagy is supported by the observation that Atg-5 knockdown completely abrogates the effect of RAD001 on chemosensitivity. Because Atg-5 is an essential component of the autophagosome with no other known function, our results suggest that RAD001 enhancement of chemosensitivity is mediated through autophagy.

Because autophagy is often considered a prosurvival phenomenon, our findings might be somewhat unexpected. One explanation for our observations is that RAD001 might promote autophagy-mediated programmed cell death, a nonapoptotic escape pathway important in cells that are defective or deficient in apoptosis (30). PTC is intrinsically resistant to cytotoxic chemotherapies, partly due to defects in apoptosis (16). This is relevant to a variety of malignancies, many with known deficiencies in apoptosis that are resistant to conventional cytotoxic chemotherapy, such as gemcitabine-refractory metastatic pancreatic cancer (31–37).

Because of the potent synergistic anticancer effects of RAD001 with doxorubicin, we characterized the effects of RAD001 on cell signaling. The most prominent alterations detected from autophagy blockade with Atg-5 knockdown were Met activation and Src inactivation. Consistent with this observation, the most prominent alterations during autophagy activation by RAD001 were Met inactivation and Src activation. Our results suggest that the chemosensitizing and radiosensitizing effects of autophagic activation are mediated largely through Met. If the chemosensitizing effects of RAD001 were mediated through Src activation, Src inhibition would at least partially abrogate these effects. However, a Src inhibitor potentiated, rather than reversed, the chemosensitization from RAD001. Of note, a recent report showed that Src inhibition activated autophagy in glioma cells (38). This report, and the results of our current study, suggest that combination treatment with a Src inhibitor might increase RAD001 efficacy.

In contrast, Atg-5 knockdown increased doxorubicin chemoresistance and Met phosphorylation. The addition of a Met inhibitor to Atg-5 knockdown cells abrogated this effect. This suggests that the effects of autophagy on chemosensitivity are mediated mostly through Met. This is the first study to directly associate autophagic activation with Met activation. A recent study showed that a cyclooxygenase-2 inhibitor downregulated both mTOR expression and Met activation in hepatocellular carcinoma cells (39). Although this was associated with autophagic activation, the association between autophagy and Met phosphorylation was not identified. Interestingly, RAD001 has excellent activity in papillary renal carcinoma, a malignancy driven by a constitutively active Met mutation (6, 7).

Ionizing radiation induces autophagy in various cell lines (40–43). However, it is unclear if autophagy promotes or prevents cell death in cancer cells exposed to ionizing radiation. In mouse embryonic fibroblasts, RAD001 promotes autophagy and reduces radioresistance (41). Furthermore, autophagy inhibition by Atg-5 siRNA or 3-MA increases the radiosensitivity of human lung cancer cells (42). In contrast, treatment with 3-MA or bafilomycin A1, another autophagy inhibitor, radiosensitizes malignant glioma cells, suggesting that autophagy may provide protection from ionizing radiation (40). Our data show that RAD001 by itself increases the radiosensitivity of PTC (Fig. 5). These observations imply that the autophagic response to radiation exposure might be a prosurvival or prodeath mechanism, depending on cellular context.

There are several issues that remain to be addressed. First, the concept of autophagy-driven programmed cell death is controversial. Some cancer cell lines that lack the machinery to execute apoptosis could still die in a nonnecrotic manner in a process associated with autophagy (44), which can be abrogated by knocking down Atg-5 (45). In the end, it is difficult to determine if autophagy is the mechanism of cell death, or if it is a survival response triggered by stresses that occur near death (46).

Here, we used RAD001 to induce autophagy. Because mTOR is an important regulator of several effector pathways, it is unclear how much of its anticancer activity could be attributed to autophagy. However, our experiments using Atg-5 knockdown suggest an autophagy-specific effect for RAD001. Lastly, our work did not include in vivo studies, and as such, we did not investigate issues related to efficacy and toxicity in animals. However, RAD001 trials are already under way for other malignancies, and the toxicity profiles are acceptable (47). We have already initiated a RAD001 trial for patients with advanced PTC resistant to radioactive iodine treatment (Dana-Farber Cancer Institute protocol no. 09-049).

In summary, the induction of autophagy in combination with cytotoxic chemotherapy is a promising therapeutic strategy for patients with refractory PTC. The synergistic effects of RAD001 and doxorubicin suggest that improvements in therapeutic efficacy could be achieved with lower overall toxicity. RAD001 improves sensitivity to doxorubicin and ionizing radiation through the attenuation of Met. Future studies should determine the efficacy and safety of combining RAD001 with conventional cytotoxic chemotherapy and radiation therapy for malignancies that are resistant to conventional therapy due to intrinsic resistance to apoptosis.

No potential conflicts of interest were disclosed.

We thank Dr. Orlo H. Clark (University of California at San Francisco, San Francisco, CA) for kindly providing TPC-1 cells; Dr. Sareh Parangi (Massachusetts General Hospital, Boston, MA) for providing 8505-C cells; and Adelaide Carothers for her assistance with this work.

Grant Support: Osteen Junior Faculty Research Grant from the Department of Surgery at Brigham and Women's Hospital.

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.