Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2014 May 15.
Published in final edited form as: Cancer Res. 2013 Apr 18;73(10):3087–3096. doi: 10.1158/0008-5472.CAN-12-3256

Cytotoxic activity of Tivantinib (ARQ 197) is not due solely to MET inhibition

Ryohei Katayama 1,2,3,*, Aki Aoyama 2,4,*, Takao Yamori 5,6, Jie Qi 1, Tomoko Oh-hara 2, Youngchul Song 1, Jeffrey A Engelman 1,3, Naoya Fujita 1,2
PMCID: PMC3759033  NIHMSID: NIHMS456298  PMID: 23598276

Abstract

The receptor tyrosine kinase c-MET is the high-affinity receptor for the hepatocyte growth factor (HGF). The HGF/c-MET axis is often dysregulated in tumors. c-MET activation can be caused by MET gene amplification, activating mutations, and auto- or paracrine mechanisms. Thus, c-MET inhibitors are under development as anti-cancer drugs. Tivantinib (ARQ 197) was reported as a small molecule c-MET inhibitor and early clinical studies suggest anti-tumor activity. To assess if the anti-tumor activity of tivantinib was due to inhibition of c-MET, we compared the activity of tivantinib to other c-MET inhibitors in both c-MET addicted and non-addicted cancer cells. As expected, other c-MET inhibitors, crizotinib and PHA-665752, suppressed the growth of c-MET addicted cancers, but not the growth of cancers that are not addicted to c-MET. In contrast, tivantinib inhibited cell viability with similar potency in both c-MET addicted and non-addicted cells. These results suggest that tivantinib exhibits its antitumor activity in a manner independent of c-MET status. Tivantinib treatment induced a G2/M cell cycle arrest in EBC1 cells similarly to vincristine treatment, whereas PHA-665752 or crizotinib treatment markedly induced G0/G1 cell cycle arrest. To identify the additional molecular target of tivantinib, we performed COMPARE analysis, an in silico screening of a database of drug sensitivities across 39 cancer cell lines (JFCR39), and identified microtubule as a target of tivantinib. Tivantinib treated cells demonstrated typical microtubule disruption similar to vincristine and inhibited microtubule assembly in vitro. These results suggest that tivantinib inhibits microtubule polymerization in addition to inhibiting c-MET.

Keywords: c-MET, Tivantinib, microtubule, tubulin, Oncogene addiction

Introduction

The met proto-oncogene (c-MET) was originally identified from N-methyl-N'-nitro-N-nitrosoguanidine (MNNG)-treated human osteosarcoma cell lines. c-MET is an activated oncogene encoding a receptor tyrosine kinase (RTK) for hepatocyte growth factor (HGF), also called scatter factor (SF) (1). The HGF/c-MET signaling pathway is frequently dysregulated in human cancer (2). Aberrant activation of c-MET can be due to gene amplification, transcriptional upregulation, activating mutations or HGF-mediated auto- or paracrine stimulation. Activation of c-MET pathway by co-expression of HGF and c-MET was shown to drive tumorigenesis and metastasis in xenograft models and in transgenic mouse models (3). Although HGF/c-MET axis has been associated with metastasis and migration of cancer cells (3, 4), recent studies have demonstrated that some cancers are addicted to the pathway for their growth and survival. In particular, cancers with amplification of c-MET have been shown to be highly sensitive to c-MET kinase inhibitors in cell lines and in the clinic (5–7). In addition, HGF/c-MET pathway was associated with the acquired resistance to inhibitors to epidermal growth factor receptor (EGFR) in EGFR mutant non-small cell lung cancers (NSCLC) (8–11). Thus, inhibitors of c-MET have been pursued as therapeutic interventions in oncology. Many low-molecular inhibitors of c-MET and monoclonal antibodies against c-MET and to HGF are now entering clinical trials.

Tivantinib (ARQ 197) was initially reported as a c-MET selective inhibitor in 2010 (12) and entered into clinical trials (13–18). In the initial report, tivantinib inhibited recombinant human c-MET with a calculated inhibitory constant (Ki) of ~355 nmol/L and had weak inhibitory effects on p21-activated kinase 3 (PAK3), vascular endothelial growth factor receptor-3 (VEGFR-3/Flt4), calmodulin-dependent kinase II (CAMKII)-delta and Pim-1. Tivantinib did not inhibit any of the other 225 human kinases tested, including the Ron kinase which belongs to the c-MET family of RTKs. The crystal structure of the tivantinib in complex with the c-MET kinase domain revealed that tivantinib binds to the inactive form of c-MET, suggesting that it inhibits c-MET through a non-ATP-competitive mechanism (19). This suggested inhibitory mode of action is different from the disclosed c-MET inhibitors under preclinical and clinical development. Recent clinical trial results suggest that tivantinib may be active in KRAS mutant lung cancers, which is not a cancer type identified in other preclinical studies to be dependent on c-MET signaling (16). In addition, a recent study found that tivantinib was equally potent against MKN-45 cells (with MET amplification) and NCI-H460 cells (KRAS mutation and no MET amplification) (12), although a different study found that another c-MET inhibitor PHA-665752 was effective only in the MKN-45 cells (7). In this study, we aimed to determine if the toxicity of tivantinib is due solely to inhibition of c-MET, and found that this was not the case. Thus, we sought to determine if tivantinib inhibits additional target molecules or pathways in the cells. We previously established the COMPARE analysis which consists of the sensitivity data of a panel of 39 cancer cell lines (termed JFCR39) against numerous drugs (20–24). The COMPARE analysis enables us to putatively identify the molecular target of a test compound by comparing the growth inhibitory patterns (fingerprints) of JFCR39 with those of the known anticancer drug or compounds (24–27). Here, we utilized COMPARE in silico screening to identify additional molecular targets of tivantinib that may lend insights into its indiscriminatory activity.

Materials and Methods

Cell lines and reagents

EBC1, MKN45, SNU638, A549, NCI-H460, HCC827, SNU-5, BT-474 and SKBR3 were cultured in RPMI 1640 medium with 10% FBS (RPMI growth medium). SNU638 subclones, SR-A1 and SR-C1 cells, were cultured in RPMI growth medium containing 1 µmol/L of a c-MET inhibitor PHA-665752 as described previously (28). SR-A1 and SR-C1 cells were cultured at least one week in drug-free RPMI growth medium before experiments. Tivantinib (ARQ 197) and crizotinib were purchased from ChemieTek. PHA-665752 was purchased from Tocris Biosciences. Vincristine and paclitaxel were purchased from Sigma. Compounds were dissolved in DMSO to a final concentration of 10 mmol/L and stored at −20°C until use.

Survival assays

Assessment of cell viability was performed as follows. 2000 to 3000 cells were seeded in 96-well plates in sextuplet. On the following day, cells were treated with increasing concentrations of the indicated drugs and incubated for another 72 hr. Cell viability was determined by adding the CellTiter-Glo assay reagent (Promega) for 10 min and luminescence was measured using a Centro LB 960 luminometer (Berthold Technologies). The data were graphically displayed using GraphPad Prism version 5.0 (GraphPad Software). IC50 value was determined by a nonlinear regression model with a sigmoidal dose response in GraphPad.

Immunoblot analysis

Lysates were prepared as described previously (9, 29). Equal amounts of lysates were electrophoresed and immunoblotted with the antibodies against phospho-c-MET (Tyr1234/1235) (3D7), c-MET (25H2), phospho-p42/44 ERK/MAPK (extracellular-signal-regulated kinase/mitogen-activated protein kinase) (Thr202/Tyr204), p42/44 ERK/MAPK, phospho-AKT (Ser473) (D9E), AKT1 (C73H10), β-actin (13E5) (Cell Signaling Technology), and α-Tubulin (DM1A) (Sigma).

Infection of shRNA

Cells were seeded into 96-well plates. After incubation for 24 h, cells were infected with MET shRNAs (MET KD 345 shRNA and MET KD 4571 shRNA) or scramble shRNA (9, 28). Media was changed the following day. Three days later, cell viability was assessed by CellTiter-Glo assay as described above. Titers of virus were used that gave the same results in the absence and presence of puromycin.

Immunohistochmistry

EBC1 or A549 cells were fixed in 4% paraformaldehyde for 15 min at room temperature, followed by 0.5% Triton X-100 for 5 min at room temperature. Fixed and permeabilized cells were incubated with 1% bovine serum albumin in phosphate-buffered saline for 60 min at room temperature. Then, cells were incubated with anti-α-tubulin (1:1000) as the primary antibody over night at 4 °C and Alexa Fluor 488-conjugated anti-mouse IgG (1:1000) as the secondary antibody with Hoechst 33342 and Alexa Fluor 568 conjugated pharoidine for 1 h at room temperature. The cells were observed under an Olympus IX71 fluorescence microscopy equipped with a charge-coupled device camera.

COMPARE analysis

Based on a series of GI50 values of a panel of 39 human cancer cell lines (termed JFCR39) described previously, fingerprints are presented in the graphic profiles of relative growth inhibition within JFCR39. To analyze the correlation between the fingerprints of drug A and drug B, we exploited the COMPARE computer algorithm as described previously (23, 27, 30). The Pearson correlation coefficient between the fingerprints of drug A and drug B was calculated.

Microtubule polymerization assay

Porcine brain tubulin polymerization assay was performed in 55ul volumes at 30°C using the Tubulin Polymerization Assay Kit (Cytoskeleton, Denver) according to the manufacturer’s instructions. Tubulin preparation was incubated at 30 °C in the presence of inhibitors (tivantinib, vincristine, paclitaxel, crizotinib or PHA-665752). Fluorescence emission was read at excitation (355nm), emission (460nm) with a TriStar LB941 plate reader (Berthold Technologies) every minute right after starting incubation.

Flow cytometric analysis

Cells were treated with indicated concentration of tivantinib, vincristine, PHA-665752 or crizotinib for 24 hr. After incubation, the cells were harvested and fixed with 70% ice-cold ethanol for 30 min at 4°C. The cells were washed with PBS and then incubated with RNaseA (0.25mg/mL, Sigma) for 15 min at 37°C, add propidium iodide (final concentration is 50 µg/mL, Sigma) and incubate for 30min at 4°C. Analyses were performed using a Cytomics 500 flow cytometer (Beckman Coulter) with a Cytomics RXP and FlowJo (Treestar) software.

Apoptosis assay

1×105 cells were seeded in 6-well plates. On the following day, cells were treated with the indicated concentration of drugs and incubated for another 48 or 72 hr. After incubation, cells were collected and stained with FITC labeled AnnexinV and 5 µg/ml PI for 10 minutes. Cells were then assayed using a Cytomics 500 flow cytometer (Beckman Coulter), and the data was analyzed using FlowJo software (Treestar).

Results

Tivantinib inhibits the growth of both c-MET-dependent and -independent cancer cells

To examine if the anti-proliferative activity of tivantinib was due to inhibition of c-MET, we examined cancer cell lines previously shown to be addicted to c-MET and sensitive to c-MET inhibitors (NSCLC cell line EBC1 and gastric carcinoma cell lines MKN45 and SNU638) and NSCLC cell lines previously shown to be resistant to c-MET inhibitors, A549, NCI-H460 (H460) and HCC827 (9, 10, 28). As shown in supplementary table S1, EBC-1 and MKN45 have MET gene amplification. The c-MET addicted SNU638 cells do not have MET amplification, but express levels of c-MET equivalent to those cell lines with MET gene amplification (28). The resistant A549 and H460 cells harbor KRAS mutations, and HCC827 cells have an EGFR exon 19 deletion. As expected, the expression levels of Tyr1234/Tyr1235-phosphorylated and total c-MET were higher in the EBC1, MKN45 and SNU638 compared A549, H460 and HCC827 cells (Supplementary Fig. S1A). Of note, the EGFR-addicted HCC827 cell line has high expression of c-MET as well (Supplementary Fig. S1A), but they are driven by EGFR signaling and are resistant to c-MET inhibitors (9). Consistent with the previous data with c-MET kinase inhibitors, we found that the viability EBC-1, MKN-45 and SNU638 cells was impaired by MET knockdown, but the viability A549, H460 and HCC827 cells was unaffected (Supplementary Fig. S1B and S1C). These results are consistent with the expected finding that EBC1, MKN45 and SNU638 require c-MET, but A549, H460 and HCC827 do not.

These cell lines were examined for sensitivity to three c-MET inhibitors, tivantinib, PHA-665752 or crizotinib. Unexpectedly, tivantinib inhibited cell viability in all of the cell lines examined (Fig. 1A). In contrast, the other two c-MET inhibitors, PHA-665752 and crizotinib, inhibited cell viability specifically in the cancer cell lines that were shown to be c-MET-dependent in the shRNA experiments (Fig. 1B and C). Similar analyses were performed on the MET amplified gastric carcinoma cell line SNU-5 and the HER2 amplified BT-474 and SKBR3 breast cancer cell lines (supplementary table S1 and Fig. S2). As shown in Fig. 1D, tivantinib is equally potent at diminishing cell viability in cancers cell lines regardless of their dependence on c-MET, whereas the other two c-MET inhibitors demonstrate substantially greater potency in the cancers driven by c-MET (Fig. 1D; supplementary table S2). These data suggest that the toxicity of tivantinib against cancer cell lines may be independent of inhibition of c-MET. Importantly, the IC50 values obtained with tivantinib in these cell lines are equivalent to those reported in earlier publications (12).

Figure 1. Antitumor activity of tivantinib on c-MET-addicted and –independent cancer cell lines.

Figure 1

Open in a new tab

A–C, Cells were treated with the increasing concentrations of tivantinib (A), PHA-665752 (B) or crizotinib (C) for 72 hr. Viable cells were assessed by CellTiter-Glo assay and were graphed relative to untreated cells. Experiments were carried out in sextuplet. The average values and SDs are shown. D, IC50 values of each cell lines to tivantinib, PHA-665752 or crizotinib were shown. Repeated experiments gave similar results.

In previous studies, we had developed clones of SNU638 with acquired resistance to PHA-665752 (28). Both of these clones demonstrated resistance to both PHA-665752 and crizotinib (Fig. 2A and B). One of these clones, SR-A1, developed a mutation, Y1230H, that was the cause of resistance. Another clone, SR-C1, was resistant due to increased activation of EGFR resulting from increased expression of transforming growth factor α (28). In these cells, c-MET was still inhibited by PHA-665752, but EGFR signaling maintained cell viability. Furthermore, the viability of these cells were unaffected by MET knockdown (28). We observed that both of the c-MET inhibitor-resistant clones showed the same sensitivity to tivantinib (Fig. 2C and D; supplementary table S2). This was surprising since viability of the SR-C1 clone is not dependent on c-MET (28). This data is consistent with studies in Figure 1 suggesting that sensitivity to tivantinib is independent of a cell’s dependence on c-MET.

Figure 2. Sensitivity to tivantinib in c-MET inhibitor PHA-665752-resistant SNU638 subclones.

Figure 2

Open in a new tab

A–C, parental SNU638 and resistant subclones SR-A1 and SR-C1 were treated with the increasing concentrations of PHA-665752 (A), crizotinib (B), or tivantinib (C), for 72 hr. Viable cells were assessed by CellTiter-Glo assay and were graphed relative to untreated cells. Experiments were carried out in sextuplet. The average values and SDs are shown. D, IC50 values of each cell lines to tivantinib, PHA-665752 or crizotinib were shown. Repeated experiments gave similar results.

Tivantinib does not suppress c-MET phosphorylation and downstream signaling

To examine the effect of tivantinib on c-MET signaling, MKN45, EBC1 and A549 cells were treated with the increasing concentrations of tivantinib or crizotinib (Fig. 3A). The first two cell lines are addicted to c-MET signaling, whereas the A549 cells are not (Fig. 1 and Supplementary Fig. S1). After exposure to 1 µmol/L of crizotinib for 6 h, there was a marked downregulation of phosphorylation of c-MET and downstream AKT and ERK in MKN45 and EBC1 cells (Fig. 3A). Previous studies have shown that the PI3K-AKT and MEK-ERK signaling are downstream of c-MET in these sensitive cells (7, 9). However, phospho-c-MET was undetectable in the A549 cells and crizotinib did not suppress ERK or AKT phosphorylation. In contrast, tivantinib, used at doses up to 10 µmol/L, failed to impair c-MET, AKT, or ERK phosphorylation in the EBC1 or MKN45 cells. Of the note, the IC50 of tivantinib was 300–400 nmol/L in these cells (supplementary table S2), despite the lack of effect on c-MET phosphorylation or downstream signaling.

Figure 3. Effect of Tivantinib on c-MET phosphorylation and reactivation.

Figure 3

Open in a new tab

A, Cells were treated with the indicated concentrations of tivantinib or crizotinib for 6 hr. Cell lysates were electrophoresed and immunoblotted with the indicated antibodies. B, MKN45 cells were pretreated with 100 nmol/L of crizotinib or 100 nmol/L of crizotinib plus 10 µmol/L of tivantinib for 6 hr. Then cells were washed three times and further incubated with RPMI growth medium containing 100 nmol/L of crizotinib, 10 µmol/L of tivantinib or no drug (Fresh medium). After the indicated time points, cells were harvested. Cell lysates were electrophoresed and immunoblotted with the indicated antibodies. For positive control, MKN45 cells were incubated with RPMI growth medium for 6 hr, washed three times with RPMI growth medium containing no drugs and then harvested. Repeated experiments gave similar results.

When the cells were treated with tivantinib or crizotinib for 24 hr, 1 µmol/L crizotinib potently suppressed c-MET, AKT, and ERK phosphorylation in the MKN45, EBC-1, and SNU-638 cells. In contrast, tivantinib moderately decreased the phospho-c-MET level in MKN45 and EBC1 cells but not in SNU638 cells (Supplementary Fig. S3). Not surprisingly, there was no suppression of AKT in these cell lines and a modest suppression of ERK in the EBC1 cells treated with tivatinib (up to 10 µmol/L). However, all of these cell lines, including the A549 cells, had submicromolar sensitivity to the tivantinib (supplementary table S2). Similar to the results in Fig. 1 and 2, these results are most consistent with the notion that the effect of tivantinib on cell viability is not solely due to inhibition of c-MET.

A recent report suggests that tivantinib selectively binds to the inactive or unphosphorylated form of c-MET, preventing it from being activated (19). To test whether tivantinib keeps c-MET in inactive form in cells, MKN45 cells were pretreated with 100 nmol/L of crizotinib for 6 hr to suppress c-MET phosphorylation. Pretreatment with crizotinib substantially downregulated the phospho-c-MET level within 6 hr (Fig. 3B). Washout of the crizotinib with fresh medium led to reactivation of c-MET within 15 min. Co-incubation of tivantinib (up to 10 µmol/L) plus 100 nmol/L of crizotinib for 6 hr also suppressed the phosphorylation of c-MET in MKN45 cells. Surprisingly, washout and continued incubation of the cells with medium containing 10 µmol/L of tivantinib failed to prevent reactivation of c-MET phosphorylation. The reactivation of c-MET was almost the same as observed after wash and additional incubation in the medium containing none (fresh media). When A549 cells, which express c-MET but not phospho-c-MET (Supplementary Fig S1A), were treated with HGF ligand, phospho-c-MET level was increased and downstream AKT/ERK signaling was also significantly activated. We did find observe that tivantinib pre-treatment moderately inhibited HGF dependent c-MET activation in A549 cells (Suplementary Fig. S4). These results suggest that tivantinib may have the capacity to mitigate c-MET activation following HGF activation. However, the results in Figures 1–3 demonstrate that this activity is not the major mechanism underlying the inhibition of cell viability.

Tivantinib treatment increases G2/M phase cells

To elucidate the mechanism by which tivantinib inhibits cell growth, we performed cell cycle analyses after tivantinib treatment. MET amplified EBC1 cells were treated with 1 µmol/L of tivantinib, PHA-665752, and crizotinib for 24 hr and the cell cycles were examined by PI staining. To our surprise, tivantinib markedly increased the number of G2/M phase cells, whereas the other two c-MET inhibitors, crizotinib and PHA-665752, induced G0/G1 arrest (Fig 4A). The effect on cell cycle of tivantinib was more similar to vincristine, a microtubule polymerization inhibitor (31). When MET amplified EBC-1 or MKN45 cells were treated with tivantinib, vincristine or crizotinib for 48 to 72 hrs, induction of apoptosis was observed (Supplementary Fig. S5). The A549 or H460 cells, both of which are resistant to c-MET inhibitors and c-MET knockdown, were treated with 1 µmol/L of tivantinib, PHA-665752, crizotinib and vincristine for 24 hr and the cell cycles were analyzed. Tivantinib treatment markedly increased the number of G2/M phase cells and sub-G1 population, whereas the other two MET inhibitors, crizotinib and PHA-665752 did not affect the cell cycle (Supplementary Fig. S6A and B).

Figure 4. Tivantinib induces a G2/M arrest and exhibits similar activity as tubulin inhibitors across the JFCR39 cell line panel.

Figure 4

Open in a new tab

A, EBC1 cells were treated with 1 µmol/L of tivantinib, vincristine, PHA-665752 or crizotinib for 24 hr. Cells were trypsinized, fixed and stained with propidium iodide, and the cell cycle was analyzed by flow cytometry. The histogram shows cell distribution versus DNA content. B, Growth inhibition against a panel of 39 human cancer cell lines. The mean graph was produced by computer processing of the 50% growth inhibition (GI50) values as described under “Materials and Methods”. The x-axis represents the logarithm of difference between the mean of GI50 values for 39 cell lines and the GI50 value for each cell line. MG-MID, the mean of log GI50 values for 39 cell lines; Delta, the logarithm of difference between the MG-MID and the log GI50 of the most sensitive cell line; Range, the logarithm of difference between the log GI50 of the most resistant cell line and the log GI50 of the most sensitive cell line. Quantification of the GI50 value was represented as the mean of four different experiments. Br, breast; CNS, central nervous system; Co, colon; Lu, lung; Me, melanoma; Ov, ovarian; Re, renal; St, stomach; xPg, prostate. C, The three compounds (out of 1805 compounds) which have high Pearson correlation coefficient (r) with tivantinib, and the results of two c-MET inhibitors (Crizotinib and PHA-665752) are shown in the table. Each experiment was performed as described in “Materials and Methods”.

To identify the genuine target of tivantinib, we performed COMPARE analysis using the JFCR39 cell line panel. The JFCR39 panel is effective for in silico screening of compounds with specific pharmacological activities. There is an extensive database assessing the activity of over 1000 compounds, including anti-cancer drugs and known inhibitors of various biological pathways, across the JFCR39 cell line panel. Using this JFCR39 panel fingerprint, which is created by a series of GI50 value among the JFCR 39 cell lines, we can determine potential targets of tests molecules. We have successfully identified the target of new compounds, such as a new telomerase inhibitor (FJ5002), an inhibitor of the Golgi system (AMF-26) and a new phosphatidylinositol 3-kinase (PI3K) inhibitor (ZSTK474) (25–27). From the results of COMPARE analysis, we observed that the fingerprint of tivantinib is similar to that of E7010 and vincristine (Fig. 4B and Supplementary Figure S7). In contrast, the fingerprints of other two c-MET inhibitors PHA-665752 and crizotinib are mutually similar, but different from tivantinib, E7010 or vincristine. The COMPARE analysis of the tivantinib fingerprint identified the tubulin polymerization inhibitor E7010 (32) as the compound with a highest correlation coefficient (r = 0.74) (1805 tests, more than 1000 compounds). Vincristine and paclitaxel ranked, 2nd and 3rd respectively, also had high correlation coefficients (Fig. 4C). All of these compounds target microtubule function. These results suggest that tivantinib may disrupt microtubule and to suppress growth of the tumor cells.

Tivantinib disrupts microtubule by inhibiting tubulin polymerization

To test whether tivantinib alters microtubule dynamics, c-MET non-addicted A549 cells and c-MET addicted EBC1 cells were treated with tivantinib or vincristine for 2 hr or 16 hr, and the effect on microtubules in those cells was determined by immunofluorescent staining of α-tubulin. Tivantinib treatment led to a loss of microtubules in both A549 and EBC1, similar to the vincristine treated cells (Fig. 5A and Supplementary Fig. S8 and S9).

Figure 5. Inhibition of tubulin polymerization by tivantinib.

Figure 5

Open in a new tab

A, Tivantinib treatment disrupted microtubule in A549 cells. Control, tivantinib (10 µM, 16 hr) or vincristine (100 nM, 16 hr) treated A549 cells were fixed and stained with Alexa 488 labeled anti-alpha-tubulin antibody, Alexa 568 conjugated phalloidin (F-actin) or Hoechst33342 (nucleus). White scale bars indicate 20 µm. B, Tivantinib inhibits tubulin polymerization similar to vincristine in vitro. A tubulin polymerization assay was performed with or without indicated concentration of tivantinib, vincristine, paclitaxel, crizotinib or PHA-665752. Experiments were carried out in triplicate. The average values and SEMs are shown.

To explore whether tivantinib directly affects microtubule formation in vitro, we performed a tubulin polymerization assay. Highly purified α- and β-tubulin and GTP were mixed with tivantinib, vincristine, crizotinib or paclitaxel, and incubated. Compare to control, paclitaxel enhanced tubulin polymerization, whereas vincristine inhibited tubulin polymerization completely as has been well-described previously (Fig. 5B) (31, 33). Similar to vincristine, tivantinib inhibited tubulin polymerization in a dose-dependent manner (Fig. 5B and Supplementary Fig. S10). In contrast, c-MET inhibitors crizotinib and PHA-665752 did not affect tubulin polymerization (Fig. 5B). These results suggest that tivantinib disrupts microtubules in cells by abrogating microtubule assembly.

Discussion

The HGF/c-MET pathway is required for the normal development in mammals (34). However, evidence to date suggests that the activity of this pathway is normally low in adults except for some specific physiological responses including embryogenesis and wound healing (35, 36). However, the HGF/c-MET pathway is frequently dysregulated in many types of human cancers, and the c-MET expression is correlated with the patient poor prognosis (35, 37). c-MET transmits signals for cell proliferation, survival, invasion, and metastasis. In addition, activation of c-MET can lead to resistance to other tyrosine kinase inhibitors, and the HGF/c-MET pathway is believed to be an attractive target for EGFR-resistant cancer therapy.

Tivantinib (ARQ 197) was originally reported as a selective, orally administrated, small molecule, non-ATP competitive inhibitor of c-MET (12). The sensitivity to tivantinib was reported to be dependent on the c-MET status (12). Tivantinib is now entered into clinical trials (13–18) and has shown preliminary activity in combination with erlotinib in KRAS mutant NSCLCs (16, 38).

We compared the tivantinib sensitivity between c-MET-addicted and c-MET-independent cancer cell lines (Supplementary Fig. S1). To our surprise, we found that the IC50 values of tivantinib in c-MET-independent A549, H460 and HCC827 were similar to those in c-MET-addicted EBC1, MKN45 and SNU638 cells (Fig. 1), suggesting that the toxicity of this compound is unrelated to inhibition of c-MET. In addition, our previously established c-MET inhibitor PHA-665752 resistant SNU638 clones were c-MET-independent were as sensitive to tivantinib were almost the same as that in parental SNU638 cells (Fig. 2). Further, tivantinib did not downregulate phospho-c-MET levels after 6 hr treatment (Fig. 3A). When cells were treated with doses of tivantinib exceeding the cytotoxic doses for 24 hr, a modest decrease in phospho-c-MET levels was observed in EBC1 and MKN45 cells (Supplementary Fig. S3). However, this did not lead to marked suppression of downstream PI3K/AKT and MEK/ERK pathways, suggesting that this is not the cause of its cytotoxicity. In total, these experiments provide compelling evidence that the cytotoxic activity of this compound is not due to its effects on c-MET.

Tivantinib was reported to stabilize c-MET in inactive form (12, 19). MKN45 cells have MET gene amplification (7) and constitutive c-MET activation. However, when MKN45 cells were treated with 100 nmol/L of crizotinib, phospho-c-MET levels were potently suppressed within 6 hr. Crizotinib binds MET in the inactive conformation (39). Even though MET was in the inactive conformation, tivantinib was unable to block the re-activation of MET (Fig. 3B). In A549 cells, we did observe that tivantinib incubation mitigated activation of c-MET by HGF (Supplementary. Fig. S4). Thus, it remains possible that tivantinib has an effect on c-MET activation by ligand. However, the data in this manuscript demonstrate that the indiscriminatory toxicity of this compound is not due to its modest effects on c-MET signaling.

Although tivantinib was developed as a c-MET inhibitor and exhibits potent antitumor activity, the data suggested that tivantinib exhibits its antitumor activity by affecting unknown targets in addition to c-MET. Indeed, this was underscored by the effect of tivantinib on the cell cycle, that was quite distinct from other c-MET inhibitors, crizotinib and PHA-665752 (Figure 4A and Supplementary Fig. S6). To identify a potential target of tivantinib, we utilized the COMPARE analysis using JFCR39 cell line panel. The fingerprint of tivantinib was most similar to microtubule inhibitors. In contrast, the fingerprints of crizotinib and PHA-665752 were not similar to tivantinib (Fig. 4B and C). Furthermore, tivantinib treatment disrupted microtubules both in c-MET addicted EBC1 and c-MET non-addicted A549 cells (Fig. 5A, Supplementary Fig. S8 and S9). These results indicate that tivantinib has an ability to inhibit microtubule function.

Both microtubule polymerization and depolymerization are required for mitosis and normal homeostasis of the cells (40). The in vitro tubulin polymerization assay suggested that tivantinib inhibits tubulin polymerization similar to vincristine (Fig. 5B). Thus, our results suggest that growth inhibition/cytotoxicity of tivantinib may be mainly due to inhibition of tubulin polymerization. In the in vitro tubulin polymerization experiments, we found that 3 µM of tivantinib was required to inhibit polymerization. In this assay, the concentration of purified tubulin is approximately 18 µM (2 mg/mL). Since the concentration of drug required to inhibit tubulin polymerization is likely impacted by the concentration of tubulin, higher inhibitor concentration might be required to inhibit tubulin polymerization in these assays. However, it is important to note 10 µM doses of c-MET inhibitors, crizotinib and PHA-665752, do not inhibit tubulin polymerization (Figure 5A). Further studies are needed to clarify how tivantinib inhibits tubulin polymerization. Since the in vitro tubulin polymerization assay does not contain microtubule associate proteins, we hypothesize that tivantinib directly binds tubulin.

Currently, many clinical trials are based on the fact that tivantinib is a c-MET inhibitor. The phase II trial of erlotinib with tivantinib or placebo did not reveal a significant difference between 2 groups. Although the study did not meet its primary end point, subset analysis suggest that tivantinib showed increased efficacy among patients with KRAS mutation (16). Since our study suggests that tivantinib exhibits antitumor effect via microtubule inhibition, it remains possible that tivantinib may also be effective to other solid cancers. Indeed, the combination with erlotinib may not be the most active since this combination was developed with the notion that tivantinib is a potent c-MET inhibitor. As the mode of action of tivantinib seems to be different from classical anti-mitotic drugs, further studies are necessary to understand the detailed mechanism how tivantinib inhibits tubulin polymerization and how to use this agent most effectively to treat cancer.

Supplementary Material

1
2
3

Acknowledgements

We thank Dr. Shingo Dan for valuable discussions. We also thank to Ms. Sumie Koike for data analysis.

Grant Support

The study was supported in part by R01CA137008, R01CA140594, National Cancer Institute (NCI) Lung SPORE (to J.A. Engelman), and JSPS KAKENHI Grant Number, 24300344 and 22112008 (to N. Fujita).

Footnotes

Disclosure of Potential Conflicts of Interest

Jeffrey Engelman is co-inventor on IP combining EGFR and MET inhibitors

Author’s Contributions

Conception and design: J.A. Engelman, N. Fujita

Development of Methodology: R. Katayama, A. Aoyama, T. Yamori, J. Qi, J.A. Engelman, N. Fujita

Acquisition of Data: R. Katayama, A. Aoyama, T. Yamori, J. Qi, T. Oh-hara, Y. Song, J.A. Engelman, N. Fujita

Analysis and interpretation of data: R. Katayama, T. Yamori, J.A. Engelman, N. Fujita

Writing, review, and/or revision of the manuscript: R. Katayama, J.A. Engelman, N. Fujita

Administrative, technical, or material support: R. Katayama, A. Aoyama, J. Qi, T. Yamori, J.A. Engelman, N. Fujita

Study supervision: J.A. Engelman, N. Fujita

References

  • 1.Cooper CS, Park M, Blair DG, Tainsky MA, Huebner K, Croce CM, Vande Woude GF. Molecular cloning of a new transforming gene from a chemically transformed human cell line. Nature. 1984;311:29–33. doi: 10.1038/311029a0. [DOI] [PubMed] [Google Scholar]
  • 2.Danilkovitch-Miagkova A, Zbar B. Dysregulation of Met receptor tyrosine kinase activity in invasive tumors. J Clin Invest. 2002;109:863–867. doi: 10.1172/JCI15418. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Takayama H, LaRochelle WJ, Sharp R, Otsuka T, Kriebel P, Anver M, Aaronson SA, Merlino G. Diverse tumorigenesis associated with aberrant development in mice overexpressing hepatocyte growth factor/scatter factor. Proc Natl Acad Sci U S A. 1997;94:701–706. doi: 10.1073/pnas.94.2.701. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Rong S, Segal S, Anver M, Resau JH, Vande Woude GF. Invasiveness and metastasis of NIH 3T3 cells induced by Met-hepatocyte growth factor/scatter factor autocrine stimulation. Proc Natl Acad Sci U S A. 1994;91:4731–4735. doi: 10.1073/pnas.91.11.4731. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Lennerz JK, Kwak EL, Ackerman A, Michael M, Fox SB, Bergethon K, Lauwers GY, Christensen JG, Wilner KD, Haber DA, Salgia R, Bang YJ, Clark JW, Solomon BJ, Iafrate AJ. MET amplification identifies a small and aggressive subgroup of esophagogastric adenocarcinoma with evidence of responsiveness to crizotinib. J Clin Oncol. 2011;29:4803–4810. doi: 10.1200/JCO.2011.35.4928. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Ou SH, Kwak EL, Siwak-Tapp C, Dy J, Bergethon K, Clark JW, Camidge DR, Solomon BJ, Maki RG, Bang YJ, Kim DW, Christensen J, Tan W, Wilner KD, Salgia R, Iafrate AJ. Activity of crizotinib (PF02341066), a dual mesenchymal-epithelial transition (MET) and anaplastic lymphoma kinase (ALK) inhibitor, in a non-small cell lung cancer patient with de novo MET amplification. J Thorac Oncol. 2011;6:942–946. doi: 10.1097/JTO.0b013e31821528d3. [DOI] [PubMed] [Google Scholar]
  • 7.Smolen GA, Sordella R, Muir B, Mohapatra G, Barmettler A, Archibald H, Kim WJ, Okimoto RA, Bell DW, Sgroi DC, Christensen JG, Settleman J, Haber DA. Amplification of MET may identify a subset of cancers with extreme sensitivity to the selective tyrosine kinase inhibitor PHA-665752. Proc Natl Acad Sci U S A. 2006;103:2316–2321. doi: 10.1073/pnas.0508776103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Bean J, Brennan C, Shih JY, Riely G, Viale A, Wang L, Chitale D, Motoi N, Szoke J, Broderick S, Balak M, Chang WC, Yu CJ, Gazdar A, Pass H, Rusch V, Gerald W, Huang SF, Yang PC, Miller V, Ladanyi M, Yang CH, Pao W. MET amplification occurs with or without T790M mutations in EGFR mutant lung tumors with acquired resistance to gefitinib or erlotinib. Proc Natl Acad Sci U S A. 2007;104:20932–20937. doi: 10.1073/pnas.0710370104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Engelman JA, Zejnullahu K, Mitsudomi T, Song Y, Hyland C, Park JO, Lindeman N, Gale CM, Zhao X, Christensen J, Kosaka T, Holmes AJ, Rogers AM, Cappuzzo F, Mok T, Lee C, Johnson BE, Cantley LC, Janne PA. MET amplification leads to gefitinib resistance in lung cancer by activating ERBB3 signaling. Science. 2007;316:1039–1043. doi: 10.1126/science.1141478. [DOI] [PubMed] [Google Scholar]
  • 10.Turke AB, Zejnullahu K, Wu YL, Song Y, Dias-Santagata D, Lifshits E, Toschi L, Rogers A, Mok T, Sequist L, Lindeman NI, Murphy C, Akhavanfard S, Yeap BY, Xiao Y, Capelletti M, Iafrate AJ, Lee C, Christensen JG, Engelman JA, Janne PA. Preexistence and clonal selection of MET amplification in EGFR mutant NSCLC. Cancer Cell. 2010;17:77–88. doi: 10.1016/j.ccr.2009.11.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Yano S, Wang W, Li Q, Matsumoto K, Sakurama H, Nakamura T, Ogino H, Kakiuchi S, Hanibuchi M, Nishioka Y, Uehara H, Mitsudomi T, Yatabe Y, Nakamura T, Sone S. Hepatocyte growth factor induces gefitinib resistance of lung adenocarcinoma with epidermal growth factor receptor-activating mutations. Cancer Res. 2008;68:9479–9487. doi: 10.1158/0008-5472.CAN-08-1643. [DOI] [PubMed] [Google Scholar]
  • 12.Munshi N, Jeay S, Li Y, Chen CR, France DS, Ashwell MA, Hill J, Moussa MM, Leggett DS, Li CJ. ARQ 197, a novel and selective inhibitor of the human c-Met receptor tyrosine kinase with antitumor activity. Mol Cancer Ther. 2010;9:1544–1553. doi: 10.1158/1535-7163.MCT-09-1173. [DOI] [PubMed] [Google Scholar]
  • 13.Adjei AA, Schwartz B, Garmey E. Early clinical development of ARQ 197, a selective, non-ATP-competitive inhibitor targeting MET tyrosine kinase for the treatment of advanced cancers. Oncologist. 2011;16:788–799. doi: 10.1634/theoncologist.2010-0380. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Goldman JW, Laux I, Chai F, Savage RE, Ferrari D, Garmey EG, Just RG, Rosen LS. Phase 1 dose-escalation trial evaluating the combination of the selective MET (mesenchymal-epithelial transition factor) inhibitor tivantinib (ARQ 197) plus erlotinib. Cancer. 2012 doi: 10.1002/cncr.27575. [DOI] [PubMed] [Google Scholar]
  • 15.Rosen LS, Senzer N, Mekhail T, Ganapathi R, Chai F, Savage RE, Waghorne C, Abbadessa G, Schwartz B, Dreicer R. A phase I dose-escalation study of Tivantinib (ARQ 197) in adult patients with metastatic solid tumors. Clin Cancer Res. 2011;17:7754–7764. doi: 10.1158/1078-0432.CCR-11-1002. [DOI] [PubMed] [Google Scholar]
  • 16.Sequist LV, von Pawel J, Garmey EG, Akerley WL, Brugger W, Ferrari D, Chen Y, Costa DB, Gerber DE, Orlov S, Ramlau R, Arthur S, Gorbachevsky I, Schwartz B, Schiller JH. Randomized phase II study of erlotinib plus tivantinib versus erlotinib plus placebo in previously treated non-small-cell lung cancer. J Clin Oncol. 2011;29:3307–3315. doi: 10.1200/JCO.2010.34.0570. [DOI] [PubMed] [Google Scholar]
  • 17.Wagner AJ, Goldberg JM, Dubois SG, Choy E, Rosen L, Pappo A, Geller J, Judson I, Hogg D, Senzer N, Davis IJ, Chai F, Waghorne C, Schwartz B, Demetri GD. Tivantinib (ARQ 197), a selective inhibitor of MET, in patients with microphthalmia transcription factor-associated tumors: Results of a multicenter phase 2 trial. Cancer. 2012 doi: 10.1002/cncr.27582. [DOI] [PubMed] [Google Scholar]
  • 18.Yap TA, Olmos D, Brunetto AT, Tunariu N, Barriuso J, Riisnaes R, Pope L, Clark J, Futreal A, Germuska M, Collins D, deSouza NM, Leach MO, Savage RE, Waghorne C, Chai F, Garmey E, Schwartz B, Kaye SB, de Bono JS. Phase I trial of a selective c-MET inhibitor ARQ 197 incorporating proof of mechanism pharmacodynamic studies. J Clin Oncol. 2011;29:1271–1279. doi: 10.1200/JCO.2010.31.0367. [DOI] [PubMed] [Google Scholar]
  • 19.Eathiraj S, Palma R, Volckova E, Hirschi M, France DS, Ashwell MA, Chan TC. Discovery of a novel mode of protein kinase inhibition characterized by the mechanism of inhibition of human mesenchymal-epithelial transition factor (c-Met) protein autophosphorylation by ARQ 197. J Biol Chem. 2011;286:20666–20676. doi: 10.1074/jbc.M110.213801. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Monks A, Scudiero D, Skehan P, Shoemaker R, Paull K, Vistica D, Hose C, Langley J, Cronise P, Vaigro-Wolff A, et al. Feasibility of a high-flux anticancer drug screen using a diverse panel of cultured human tumor cell lines. J Natl Cancer Inst. 1991;83:757–766. doi: 10.1093/jnci/83.11.757. [DOI] [PubMed] [Google Scholar]
  • 21.Paull KD, Shoemaker RH, Hodes L, Monks A, Scudiero DA, Rubinstein L, Plowman J, Boyd MR. Display and analysis of patterns of differential activity of drugs against human tumor cell lines: development of mean graph and COMPARE algorithm. J Natl Cancer Inst. 1989;81:1088–1092. doi: 10.1093/jnci/81.14.1088. [DOI] [PubMed] [Google Scholar]
  • 22.Shoemaker RH. The NCI60 human tumour cell line anticancer drug screen. Nat Rev Cancer. 2006;6:813–823. doi: 10.1038/nrc1951. [DOI] [PubMed] [Google Scholar]
  • 23.Yamori T. Panel of human cancer cell lines provides valuable database for drug discovery and bioinformatics. Cancer Chemother Pharmacol. 2003;52(Suppl 1):S74–S79. doi: 10.1007/s00280-003-0649-1. [DOI] [PubMed] [Google Scholar]
  • 24.Yamori T, Matsunaga A, Sato S, Yamazaki K, Komi A, Ishizu K, Mita I, Edatsugi H, Matsuba Y, Takezawa K, Nakanishi O, Kohno H, Nakajima Y, Komatsu H, Andoh T, Tsuruo T. Potent antitumor activity of MS-247, a novel DNA minor groove binder, evaluated by an in vitro and in vivo human cancer cell line panel. Cancer Res. 1999;59:4042–4049. [PubMed] [Google Scholar]
  • 25.Naasani I, Seimiya H, Yamori T, Tsuruo T. FJ5002: a potent telomerase inhibitor identified by exploiting the disease-oriented screening program with COMPARE analysis. Cancer Res. 1999;59:4004–4011. [PubMed] [Google Scholar]
  • 26.Ohashi Y, Iijima H, Yamaotsu N, Yamazaki K, Sato S, Okamura M, Sugimoto K, Dan S, Hirono S, Yamori T. AMF-26, a novel inhibitor of the Golgi system, targeting ADP-ribosylation factor 1 (Arf1) with potential for cancer therapy. J Biol Chem. 2012;287:3885–3897. doi: 10.1074/jbc.M111.316125. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Yaguchi S, Fukui Y, Koshimizu I, Yoshimi H, Matsuno T, Gouda H, Hirono S, Yamazaki K, Yamori T. Antitumor activity of ZSTK474, a new phosphatidylinositol 3-kinase inhibitor. J Natl Cancer Inst. 2006;98:545–556. doi: 10.1093/jnci/djj133. [DOI] [PubMed] [Google Scholar]
  • 28.Qi J, McTigue MA, Rogers A, Lifshits E, Christensen JG, Janne PA, Engelman JA. Multiple mutations and bypass mechanisms can contribute to development of acquired resistance to MET inhibitors. Cancer Res. 2011;71:1081–1091. doi: 10.1158/0008-5472.CAN-10-1623. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Katayama R, Shaw AT, Khan TM, Mino-Kenudson M, Solomon BJ, Halmos B, Jessop NA, Wain JC, Yeo AT, Benes C, Drew L, Saeh JC, Crosby K, Sequist LV, Iafrate AJ, Engelman JA. Mechanisms of acquired crizotinib resistance in ALK-rearranged lung Cancers. Sci Transl Med. 2012;4:120ra117. doi: 10.1126/scitranslmed.3003316. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Dan S, Tsunoda T, Kitahara O, Yanagawa R, Zembutsu H, Katagiri T, Yamazaki K, Nakamura Y, Yamori T. An integrated database of chemosensitivity to 55 anticancer drugs and gene expression profiles of 39 human cancer cell lines. Cancer Res. 2002;62:1139–1147. [PubMed] [Google Scholar]
  • 31.Himes RH, Kersey RN, Heller-Bettinger I, Samson FE. Action of the vinca alkaloids vincristine, vinblastine, and desacetyl vinblastine amide on microtubules in vitro. Cancer Res. 1976;36:3798–3802. [PubMed] [Google Scholar]
  • 32.Yoshimatsu K, Yamaguchi A, Yoshino H, Koyanagi N, Kitoh K. Mechanism of action of E7010, an orally active sulfonamide antitumor agent: inhibition of mitosis by binding to the colchicine site of tubulin. Cancer Res. 1997;57:3208–3213. [PubMed] [Google Scholar]
  • 33.De Brabander M, Geuens G, Nuydens R, Willebrords R, De Mey J. Taxol induces the assembly of free microtubules in living cells and blocks the organizing capacity of the centrosomes and kinetochores. Proc Natl Acad Sci U S A. 1981;78:5608–5612. doi: 10.1073/pnas.78.9.5608. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Birchmeier C, Gherardi E. Developmental roles of HGF/SF and its receptor, the c-Met tyrosine kinase. Trends Cell Biol. 1998;8:404–410. doi: 10.1016/s0962-8924(98)01359-2. [DOI] [PubMed] [Google Scholar]
  • 35.Boccaccio C, Comoglio PM. Invasive growth: a MET-driven genetic programme for cancer and stem cells. Nat Rev Cancer. 2006;6:637–645. doi: 10.1038/nrc1912. [DOI] [PubMed] [Google Scholar]
  • 36.Chmielowiec J, Borowiak M, Morkel M, Stradal T, Munz B, Werner S, Wehland J, Birchmeier C, Birchmeier W. c-Met is essential for wound healing in the skin. J Cell Biol. 2007;177:151–162. doi: 10.1083/jcb.200701086. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Birchmeier C, Birchmeier W, Gherardi E, Vande Woude GF. Met, metastasis, motility and more. Nat Rev Mol Cell Biol. 2003;4:915–925. doi: 10.1038/nrm1261. [DOI] [PubMed] [Google Scholar]
  • 38.Scagliotti GV, Novello S, Schiller JH, Hirsh V, Sequist LV, Soria JC, von Pawel J, Schwartz B, Von Roemeling R, Sandler AB. Rationale and Design of MARQUEE: A Phase III, Randomized, Double-Blind Study of Tivantinib Plus Erlotinib Versus Placebo Plus Erlotinib in Previously Treated Patients With Locally Advanced or Metastatic, Nonsquamous, Non-Small-Cell Lung Cancer. Clin Lung Cancer. 2012 doi: 10.1016/j.cllc.2012.01.003. [DOI] [PubMed] [Google Scholar]
  • 39.Timofeevski SL, McTigue MA, Ryan K, Cui J, Zou HY, Zhu JX, Chau F, Alton G, Karlicek S, Christensen JG, Murray BW. Enzymatic characterization of c-Met receptor tyrosine kinase oncogenic mutants and kinetic studies with aminopyridine and triazolopyrazine inhibitors. Biochemistry. 2009;48:5339–5349. doi: 10.1021/bi900438w. [DOI] [PubMed] [Google Scholar]
  • 40.Jordan MA, Wilson L. Microtubules and actin filaments: dynamic targets for cancer chemotherapy. Curr Opin Cell Biol. 1998;10:123–130. doi: 10.1016/s0955-0674(98)80095-1. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS456298-supplement-1.pdf (101KB, pdf)
2
NIHMS456298-supplement-2.pdf (896KB, pdf)
3
NIHMS456298-supplement-3.pdf (150.6KB, pdf)

RESOURCES