Breast Cancer–Derived Bone Metastasis Can Be Effectively Reduced through Specific c-MET Inhibitor Tivantinib (ARQ 197) and shRNA c-MET Knockdown

Breast cancer, along with prostate, thyroid, and kidney cancer, displays a remarkable predilection to metastasize to bone. At least 80% of patients with metastatic breast cancer will develop bone metastases during the course of their disease (1, 2). The pathologic manifestations of osteolytic lesions can have devastating effects, including pain, pathologic fractures, spinal compression, and hypercalcemia. An increased understanding of the cellular and molecular mechanisms that contribute to bone metastasis is necessary for improving clinical management. Breast cancer cells secrete various factors that stimulate osteoblasts, osteoclasts, and other cells of the bone microenvironment favoring bone resorption; bone cells, in turn, secrete factors that stimulate tumor cell growth. These interactions between tumor cells and the bone microenvironment result in a vicious cycle that increases both bone destruction and tumor burden, nurturing the development and propagation of bone metastasis (3–8). Therapeutic targeting of tumor–bone interaction is under intensive investigation (9). A potential target candidate is c-MET, the tyrosine kinase receptor for the hepatocyte growth factor (HGF). Primarily expressed on epithelial cells, c-MET drives different intracellular signaling pathways that are essential for the development and progression of many human cancers (refs. 10–14; additional information available at: http://www.vai.org/met). The HGF/c-MET pathway regulates diverse biological activities, ranging from proliferation, motility, and invasion to survival and angiogenesis, many of which are hallmarks of cancer (15). Aberrant signaling of the c-MET pathway, identified in a wide variety of human malignancies, has been associated with a poor prognosis, aggressive phenotype, increased metastasis, and shortened patient survival (10).

However, the role of c-MET signaling in human breast cancer bone metastasis has scarcely been investigated. Recently, we have reported that the c-MET receptor acts as an important mediator of the cross-talk between epithelial breast cancer cells and mesenchymal cells of the bone microenvironment, contributing to progression of osteolytic bone metastases in vivo (16). Given the importance and the potential therapeutic role of the c-MET receptor in bone metastasis progression, we examined the effects of inhibiting c-MET using both a specific c-MET inhibitor (tivantinib) and RNA interference technology in an in vivo murine model of breast cancer bone metastasis (17). Tivantinib is a novel, orally available, small-molecule, non–ATP-competitive c-MET inhibitor that is highly specific for the c-MET receptor (18–21). Here we show that treatment with different concentrations of tivantinib affected bone metastasis progression in a dose-dependent manner. Moreover, treatment with tivantinib in combination with the silencing of c-MET protein expression by specific short hairpin RNA (shRNA) led to an even greater reduction in bone metastasis progression and cancer-induced bone destruction as well as an increase in overall survival.

Tivantinib [(-)-trans-3-(5,6-dihydro-4H-pyrrolo [3,2,1-ij] quinolin-1-yl)-4(1H-indol-3-yl) pyrrolidine-2, 5-dione; ref. 22] was synthesized and separated by chiral purification at ArQule, Inc. Tivantinib, also known as ARQ 197, has been shown to selectively inhibit c-MET (K_i = 355 nmol/L) via a non–ATP-competitive mechanism and to have minimal inhibitory activity against the majority of a panel of 229 human kinases (20, 21). Tivantinib also showed antiproliferative activity in several human cancer cell lines expressing c-MET (GI₅₀ values ranging from 0.30 to 0.66 μmol/L) and showed antitumor activity in murine xenograft models of colon, gastric, and breast cancer (21).

The human MDA-MB-231 breast cancer cell line was obtained from the American Type Culture Collection. The bone-seeking clone wild-type (1833) and retrovirally transfected with the triple reporter construct (1833/TGL; ref. 23), and the parental MDA-MB-231 genetically modified to express TGL reporter protein (MDA-MB-231/TGL; ref. 17) were obtained from Professor J. Massagué (Memorial Sloan-Kettering Cancer Center, New York, NY). Cell line authentication was not carried out by the authors within the last 6 months. Cells were cultured in Dulbecco's Modified Eagle's Medium supplemented with 10% FBS (Sigma-Aldrich) and 1% l-glutamine.

Tumor cells were seeded in 130 μL of medium at 3,500 cells per well in 96-well microplates. After 24 hours, different concentrations of tivantinib were added for 72 hours. At the end of the treatment period, MTS reagent (Promega) was added to each well and the plates were incubated at 37°C for 4 hours in 5% CO₂. Cell proliferation was evaluated by measuring the absorbance at 490 nm using an infinite M200 Microplate Reader (Tecan). A nonlinear regression method was used to calculate GI₅₀ values (compound concentration required to produce 50% growth inhibition).

An inducible downregulation of c-MET was developed by using pSuperior retrovirus-based vector system (OligoEngine) expressing inducible shRNA. Details and sequences used are available in Supplementary Methods.

Thirty micrograms of total cellular proteins were separated on a 6% to 8% SDS-PAGE gel and then transferred to a nitrocellulose membrane (Whatman). The membranes were incubated with the different antibodies using the dilutions and conditions as specified in Supplementary Methods.

The shMET D and MutB 1833 cells (2 × 10⁵ cells per well) were plated on a 12-well culture plate (Becton Dickinson) in the presence or absence of both doxycycline (2 mg/mL) and tivantinib (0.2 μmol/L). Upon reaching confluence, a single wound was created in the center of the cell monolayers by gentle removal of the attached cells with a sterile plastic pipette tip. The plate was placed under an IX81 motorized inverted microscope (Olympus) fitted with an incubator to maintain 37°C, 5% CO₂, and 60% humidity (Okolab). Cells were followed for 24 hours, and every 30 minutes an image was captured using an ORCA-ER CCD camera (Hamamatsu). Phase-contrast pictures were analyzed by ImageJ software (Wayne Rasband). The extent of wound closure was determined by calculating the ratio between the surface area of the wound for each time point and the surface of the initial wound. These data were then expressed as the percentage of wound closure and plotted against the hours after wounding. The experiment was conducted 3 times independently.

Female 4-week-old athymic nude (nu/nu) mice were used for all experiments. Mice were obtained from Harlan-Italy and maintained under specific pathogen-free conditions with food and water provided ad libitum. Procedures involving animals and their care were conducted in conformity with institutional guidelines that are in compliance with national (Legislative Decree 116 of January 27, 1992, Authorization n.169/94-A issued December 19, 1994, by Ministry of Health) and international laws and policies (EEC Council Directive 86/609, OJ L 358. 1, December 12, 1987; Standards for the Care and Use of Laboratory Animals, United States National Research Council, Statement of Compliance A5023-01, November 6, 1998).

A total of 7.5 × 10⁶ tumor cells in 200 μL 1:1 medium/Matrigel (BD Biosciences) were injected subcutaneously into the left flank of mice; when tumors reached approximately 100 mm³, mice were treated with tivantinib (orally at the dose of 120 mg/kg in a volume of 10 mL/kg of body weight daily until the end of the experiment) or with vehicle [PEG 400:20% vitamin E TPGS solution (60:40)]. Tumor diameters were measured with a caliper twice weekly until the animals were sacrificed. Tumor weight was derived from tumor volume (assuming a density of 1) calculated by the formula: (length × width²)/2. Body weights were measured weekly during the treatment period.

A total of 5 × 10⁵ tumor cells per 100 μL of PBS were injected into the left cardiac ventricle of 3% isoflurane-anesthetized mice. Animals were randomized into groups of 10 mice each. Tivantinib was administered orally at the doses of 30, 60, or 120 mg/kg. Control animals were treated with vehicle. In the preventive and therapeutic protocols, treatments started 2 or 12 days after implant, respectively, and, in both cases, continued daily until the end of the experiment. Tetracycline (2 mg/mL; Sigma) was provided in the drinking water and replaced every 2 days, starting from the day of cell implant. Body weights were measured weekly. Cancer cell dissemination to bone and extent of bone lesions were monitored by weekly analysis of bioluminescence imaging (BLI) technology and micro-computed tomography (micro-CT), respectively, as described in Supplementary Methods. Animals were euthanized when controls started to show signs of suffering. After sacrifice, hindlimb specimens (tibia and femora) were removed during autopsy and collected for histologic and immunohistologic analysis.

Bones were fixed for 90 minutes at room temperature and decalcified in Mielodec (Bio-Optica) for 4 days, dehydrated, embedded in paraffin, and cut into 5 μm sections. Bone slides were stained with Mayer's hematoxylin and eosin (H&E) by standard protocols. Bone sections were deparaffinized, rehydrated, and subjected to antigen retrieval by water-bath heating (95°C, 20 minutes) in antigen unmasking solution (Vector Laboratories) at pH = 6. After blocking endogenous peroxidase activity with 1% H₂O₂, slides were probed with different antibodies as specified in Supplementary Methods.

Differences in bioluminescence intensity between the control and treatment groups were evaluated by 1-way ANOVA followed by Bonferroni's multiple comparison test using the GraphPad Prism software package (version 5.03). All data were expressed as the mean ± SE and differences were considered statistically significant at a level of P < 0.05.

The effect of tivantinib (Fig. 1A) on the viability of the parental MDA-MB-231/TGL and the derived bone-seeking clone 1833/TGL human breast cancer cell lines is shown in Fig. 1B. Treatment of cells with tivantinib resulted in a comparable concentration-dependent decrease in cell viability with GI₅₀ values of 1.2 and 3.7 μmol/L obtained in MDA-MB-231/TGL and 1833/TGL cells, respectively.

The efficacy of tivantinib to inhibit bone metastasis progression is shown in Fig. 2A. The appearance of cancer cells in the leg bones varied from 11 to 14 days after cell implant and increased over time, both in control and tivantinib-treated animals. In particular, the signal from the hindlimbs of tivantinib-treated (30 mg/kg) mice was very similar to that of control mice, although treatment with tivantinib at the doses of 60 and 120 mg/kg induced a dose-dependent delay and a reduction of bone metastatic growth. In fact, tivantinib at a dose of 60 mg/kg caused a temporary reduction of the tumor-related luciferase signal starting from 17 to 21 days after cell implant. From day 21 to the end of the experiment, the mean value of the BLI signal intensity from this group was comparable with that of controls. On the contrary, the dose of 120 mg/kg of tivantinib significantly inhibited BLI signal and, consequently, tumor burden in the bone of treated animals compared with the controls, starting from 14 to 21 days after cell injection. This inhibition was maintained until the end of the experiment. These BLI results were further confirmed by micro-CT data, underlining that increasing doses of tivantinib were well correlated with a decrease in the number and the extent of osteolytic lesions. Representative images of ventral views (Fig. 2B) from optical imaging and micro-CT showed that treatment of 1833/TGL-injected mice with increasing doses of tivantinib resulted in a dose-dependent decrease in bioluminescence (Fig. 2B, top panels) and in tumor-induced osteolysis (Fig. 2B, bottom panels), representing a reduction in bone metastasis progression compared with control mice.

On the basis of these results, further studies were conducted with the 120 mg/kg dose of tivantinib. Results from 2 separate experiments showed an exponential increase of photon emission associated with an increase in tumor burden, clearly evident from day 14 onward, in the hindlimbs of control mice (Fig. 3A). Treatment with tivantinib (120 mg/kg) resulted in a significant reduction of bioluminescence signal starting from 14 days after implant (Fig. 3A). As evident in Fig. 3B, animals treated with tivantinib exhibited a significant inhibition of photon counts and, therefore, of tumor growth in the leg bones compared with vehicle-treated group. The body weight of animals also showed the effects of tivantinib in injected mice. The relative body weight curve of controls started to decrease at day 17 (Fig. 3C), coincident with the rapid progression of metastatic bone disease. In animals receiving tivantinib (120 mg/kg), weight loss started from day 21 and was less than that observed in controls. The effect of tivantinib on 1833/TGL-induced osteolysis was assessed by micro-CT analysis (Fig. 3D). Three-dimensional (3D) reconstructed micro-CT images of the hindlimbs of representative control animals revealed severe bone destruction on day 21 following cancer cell transplantation. In contrast, bones from tivantinib-treated (120 mg/kg) mice were apparently intact or showed only limited signs of osteolysis at day 21 (arrows). At the end of the experiment, histologic examination of representative tibial and femoral sections from control and treated mice confirmed the presence of metastatic 1833/TGL breast cancer cells within the bone marrow cavity (Supplementary Fig. S1).

The same bone metastatic clone 1833/TGL was injected subcutaneously into the left flanks of athymic nude mice. Chronic treatment with vehicle or tivantinib (120 mg/kg) was initiated when tumors were established and reached a defined size of 100 mm³. Daily dosing for 25 days of tivantinib did not affect the growth of subcutaneously growing tumors compared with controls (Fig. 4A). Daily drug administration was well tolerated by the animals, as there was no significant difference between the body weight of the tivantinib-treated and control mice (Fig. 4B). Similar results were obtained with the bone metastatic model when tivantinib 120 mg/kg administration started 12 days after tumor cell implant, when metastasis were already well established (Supplementary Fig. S2).

To further investigate the role of the c-MET receptor in bone metastatic breast cancer cells, c-MET protein expression was silenced (in a doxycycline-inducible manner) by transfection of cells with shRNA. The results showed that the presence of doxycycline induced a decrease in the expression of c-MET receptor in shMET D 1833 cells (Fig. 5A, top left panels). In these cells, reduction in c-MET protein was found to be time dependent, as indicated by densitometric analysis (Fig. 5A, bottom left panels). By 96 hours after doxycycline administration, the c-MET protein levels were reduced by 90% in shMET D 1833 cells. On the contrary, in control Mut B 1833 cells carrying mutated shRNA, c-MET protein expression was not affected by doxycycline addition (Fig. 5A top right panels). shRNA-mediated downregulation of c-MET protein expression had no effect on either cell growth of shMET D 1833 (Fig. 5B) or Mut B 1833 cells (Supplementary Fig. S3).

To study the effects of c-MET silencing on HGF-mediated signaling events downstream of c-MET activation, specific phosphorylation of c-MET, AKT, and ERK1 proteins was analyzed (Fig. 5C). In the absence of HGF, no phosphorylation of c-MET was observed in both shMET D and Mut B 1833 cells (Supplementary Fig. S4). HGF-mediated induction of tyrosine phosphorylation (pY1234/1235) of c-MET receptor was evident in Mut B 1833 cells both in presence and absence of doxycycline and in shMET D 1833 cells without doxycycline. Upon HGF stimulation, the doxycycline-induced c-MET downregulation correlated with a reduction of c-MET tyrosine phosphorylation status in shMET D 1833 cells. Moreover, in Mut B 1833 cells, phosphorylation of ERK (Tyr204) and AKT (Thr308) was markedly induced after HGF stimulation, independent of the presence of doxycycline. Phosphorylation of Tyr204 of ERK1 and Thr308 of AKT was evident in response to HGF in doxycycline-untreated shMET D 1833 cells; although in the same cells, doxycycline induced moderate decreases in the phosphorylation and, presumably, the activation state of these downstream c-MET–related effectors. Neither HGF stimulation nor doxycycline treatment affected the total steady state of ERK1 and AKT levels.

An in vitro wound healing assay was done to assess the effects of c-MET silencing alone or in combination with tivantinib on the migration of both shMET D 1833 and Mut B 1833 cells. Figure 5D (top panel) shows representative photographs of cells migrating into scratch wounds. Quantitative analysis revealed that the combination of doxycycline-induced c-MET downregulation and tivantinib treatment significantly decreased the migration potential of shMET D 1833 (Fig. 5D, bottom panel). As expected, the presence or absence of doxycycline did not affect the migration of Mut B 1833 cells (data not shown).

The effects of c-MET silencing, alone or in combination with tivantinib treatment, was also investigated in the in vivo breast cancer–derived bone metastasis model. Animals transplanted with shMET D 1833 cells were treated with either vehicle alone (shMET D 1833 TET⁻ group), vehicle plus administration of tetracycline (shMET D 1833 TET⁺ group), tivantinib (120 mg/kg) alone (shMET D 1833 tivantinib⁺ group), or tivantinib (120 mg/kg) plus tetracycline (shMET D 1833 TET⁺/tivantinib⁺ group).

Bone metastases in the hindlimbs became detectable at day 27 after tumor cell injection (Fig. 6A). c-MET silencing induced a reduction in bone metastasis growth compared with controls. The antimetastatic efficacy of tivantinib (120 mg/kg) was also confirmed in this bone metastatic model. Interestingly, the combination of tivantinib and shRNA-mediated c-MET silencing induced a significant decrease in tumor burden in the femoral/tibial bones of treated mice compared with control mice starting from 35 days after tumor implant. Representative bioluminescence ventral images taken 38 days after cell implant indicated that c-MET silencing in combination with inhibition of c-MET activity by tivantinib may strongly reduce skeletal tumor burden (Fig. 6B, top panels). It is noteworthy that BLI photon counts from hindlimbs of mice treated with tetracycline plus tivantinib remained constant or increased slowly over time (Fig. 6A). Control animals had to be sacrificed at day 38 because of considerable reductions in body weight and generalized deterioration of their health status. The group of animals treated with the tivantinib/shMET combination survived until day 58 after implantation. As in the 1833/TGL-derived model, bone destruction assessed by micro-CT was again mainly evident at the ends of distal femora and proximal tibiae, and the analysis of skeletal changes 38 days after tumor cell implant closely correlated with the bioluminescence results (Fig. 6B; bottom panels). In vehicle-treated mice at day 38 after implantation, bone destruction was massive, with sizable holes in the tumor-bearing bones. In the tetracycline-treated group, the extent of osteolysis was comparable with controls, even when the number of bone lesions was reduced. In contrast, osteolytic lesions in tivantinib-treated animals were significantly reduced as compared to both control and c-MET–silenced animals. Finally, bones from shMET/tivantinib-treated mice were intact, with only early signs of lysis. BLI and micro-CT data were confirmed by histologic and immunohistochemical examinations of tumors in the bone sections at day 38 after cancer cell inoculation (Fig. 6C). As revealed by H&E and pan-cytokeratin immunostaining, bone marrow of distal femora from control and tetracycline-administered groups was completely invaded by metastatic tumor cells. Otherwise, only small colonies of cytokeratin-positive 1833 cells were detectable in the bones of tivantinib-treated and shMET/tivantinib-treated mice.

In vivo downregulation of c-MET activation was evaluated through immunohistochemical staining. The presence of phospho-c-MET positive cells was clear in the bone metastatic sections from vehicle-treated animals (Fig. 6D). Phospho-c-MET positive cells were also detected in shMET D 1833 TET⁺-derived sections, although the number of positive cells was greatly reduced as compared with the vehicle-treated group, in agreement with a reduction of shRNA-induced total c-MET protein expression. Otherwise, no phospho-c-MET–positive cells were seen in bone sections from tivantinib-treated mice. These findings highlighted the in vivo efficacy of tivantinib treatment in inhibiting c-MET phosphorylation, and thus its functional activation.

Management of bone metastasis is an important issue for the improvement of quality of life and survival in breast cancer patients. Therapeutic strategies aimed at interrupting the vicious cycle, impeding the development and progression of bone metastases, and improving bone integrity are under intensive investigation (24). A vast amount of compelling in vitro and in vivo evidence supporting the role of the HGF/c-MET pathway in tumor progression and metastasis suggests that this pathway could represent an attractive therapeutic target for metastatic breast cancer (25–29).

On the basis of these considerations, our bone metastatic model established with 1833 cells represents an optimal tool to assess the antimetastatic activity of tivantinib. Our results show that the c-MET inhibitor tivantinib exhibits dose-dependent activity and, at a daily dose of 120 mg/kg, significantly delays the formation and progression of bone metastases and cancer-induced osteolysis. Interestingly, this dose was ineffective in inhibiting growth of the same cells implanted subcutaneously, suggesting that the effects of tivantinib against bone metastasis in this model were not due to a direct cytotoxic effect of the compound. It could be hypothesized that the observed antimetastatic effects by tivantinib occurred in the bone microenvironment. The results obtained with cells, in which c-MET has been downregulated using shRNA, further corroborate these data and strongly support the evidence that tivantinib exerts its antimetastatic action through c-MET inhibition. c-MET silencing in 1833 cells did not affect cell growth in vitro but strongly reduced migration potential. The pharmacodynamic effects of c-MET silencing in 1833 cells clearly showed that c-MET phosphorylation is reduced and that the function of downstream pathways involved in c-MET–induced cell proliferation (i.e., ERK, MAPK, JNKs, NF-kB, and PI3K/AKT; ref. 30) were only marginally affected, in agreement with the lack of in vitro cell growth inhibitory activity. Overall, these in vitro results provide evidence for a potential role of c-MET in breast cancer metastasis to bone, assuming a role in cellular motility without affecting cell proliferation, and could explain the data obtained in vivo. In our bone metastasis model, c-MET silencing showed high antimetastatic activity. It is difficult to determine a direct comparison of activity of c-MET silencing compared with that induced by tivantinib. Notably, both strategies proved to be effective, and in our experiments, we used lower doses of tivantinib compared with those used in other studies (21), in an effort to show activity possibly not linked to a direct cytotoxic effect. Furthermore, the combination of pharmacologic c-MET inhibition (through tivantinib) and genetic downregulation of c-MET protein expression (through shRNA) seemed to produce an enhanced inhibition of bone metastasis progression. These data are in strong agreement with the observed inhibition of in vitro migration of cells simultaneously treated with tivantinib and c-MET shRNA. Taken together, our results revealed that c-MET inhibition is effective in delaying the onset and progression of tumor growth in bone and in greatly diminishing osteolytic lesions. Thus, inhibition of c-MET may offer significant benefits for the prevention of bone metastasis development in patients with breast malignancies. Finally, our finding that tivantinib is active as an antimetastatic agent at well-tolerated noncytotoxic doses suggests that tivantinib may be a good candidate for combination studies with cytotoxic agents.

In summary, this study showed that treatment with tivantinib inhibited bone metastasis progression and cancer-induced bone destruction, with an increase in the survival of treated mice. These findings suggest that tivantinib has significant therapeutic potential for the management of metastasis and provide further evidence that c-MET represents a promising therapeutic target for prevention of bone metastases from breast cancer.

G. Abbadessa and D.S. France are employed by ArQule, Inc. that is developing tivantinib.

The authors thank Jeffrey Riegel of Accuverus, Beachwood, OH, for medical editorial support and Drs. Enrico Radaelli and Marco Losa from the Fondazione Filarete (Milan, Italy) for their expert advice with immunostaining.

Grant support was provided to M. Broggini by the Italian Association for Cancer Research (AIRC), and S. Previdi is recipient of a fellowship from Monzino Foundation, Milan, Italy.

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.