Merestinib (LY2801653) inhibits neurotrophic receptor kinase (NTRK) and suppresses growth of NTRK fusion bearing tumors

Merestinib is an oral multi-kinase inhibitor targeting a limited number of oncokinases including MET, AXL, RON and MKNK1/2. Here, we report that merestinib inhibits neurotrophic receptor tyrosine kinases NTRK1/2/3 which are oncogenic drivers in tumors bearing NTRK fusion resulting from chromosomal rearrangements. Merestinib is shown to be a type II NTRK1 kinase inhibitor as determined by x-ray crystallography. In KM-12 cells harboring TPM3-NTRK1 fusion, merestinib exhibits potent p-NTRK1 inhibition in vitro by western blot and elicits an anti-proliferative response in two- and three-dimensional growth. Merestinib treatment demonstrated profound tumor growth inhibition in in vivo cancer models harboring either a TPM3-NTRK1 or an ETV6-NTRK3 gene fusion. To recapitulate resistance observed from type I NTRK kinase inhibitors entrectinib and larotrectinib, we generated NIH-3T3 cells exogenously expressing TPM3-NTRK1 wild-type, or acquired mutations G595R and G667C in vitro and in vivo. Merestinib blocks tumor growth of both wild-type and mutant G667C TPM3-NTRK1 expressing NIH-3T3 cell-derived tumors. These preclinical data support the clinical evaluation of merestinib, a type II NTRK kinase inhibitor (NCT02920996), both in treatment naïve patients and in patients progressed on type I NTRK kinase inhibitors with acquired secondary G667C mutation in NTRK fusion bearing tumors.


Anchorage dependent and independent cell proliferation
KM-12 cells were acquired from the NCI-Frederick Cancer DCTC Repository and cultured in RPMI media supplemented with 10% fetal bovine serum (FBS) and were plated in 96 well plates at 4000 cells/well. The following day, culture media was replaced with 2% FBS containing media. Cells were treated with merestinib, M1, or M2 metabolite at a dose concentration from 0.0003 μM to 10 μM for 72 hours. CellTiter-Glo ® was added according to manufacturer's protocol (Promega, Madison WI). Cell viability was determined by measuring relative luminescence unit (RLU) using a Wallac Victor 2 1420 multilable counter (Perkin Elmer, Waltham MA). DMSO treated cells served as the reference comparator (0% inhibition) while 10 μM staurosporin was used for normalization defined as 0% control. Values recorded as % control were subsequently plotted using GraphPad Prism 7 with IC 50 calculations based on 4-parameter curve-fit analysis.
Assessment of anchorage independent growth of KM-12 cells treated with merestinib, M1, M2 and crizotinib was performed using the S+ Chip Analyzer (Samsung Electro-Mechanics Co. Ltd., South Korea) as described in detail previously [1]. Briefly, 80 cells were spotted in the presence of RPMI-1640 media containing 2% FBS (final) onto microwells with alginate hydrogel added to a final concentration of 0.5% with inhibitor concentration ranging from 0.17 nM to 10,000 nM. DMSO served as the untreated control. Cells were incubated with compounds for 3 days, calcein stained and subsequently scanned. IC 50 calculations using variable slope (4-parameter) of log inhibitor concentration versus response were performed in GraphPad Prism.

In vivo mouse studies
For KM-12 cell-derived xenograft studies, 5 million cells were implanted in the rear flank of athymic nu/nu mice (Envigo, Indianapolis, IN). Dosing began once average tumor volume reached 200 mm 3 . The colorectal PDX model EL1989 was originally established from a pT3a tumor of the colon excised from an 80 year old male Caucasian patient. Presence of the TPM3-NTRK1 genomic rearrangement in this tumor was identified through exome and RNA sequencing analysis. This was independently confirmed through RT-PCR analysis of the tumor. EL1989 tumor was propagated in athymic nude mice, harvested, sectioned into small fragments and subsequently implanted in nude mice. Merestinib and crizotinib treatment began once average tumor volume reached 150 mm 3 . Merestinib was formulated in 10% PEG 400 (Fisher Chemical, Pittsburgh PA)/90% (20% Captisol) (Cydex Pharmaceuticals, San Diego CA), administered orally at 24 mg/kg once daily. Crizotinib formulated in 10% Acacia and 0.05% antifoam (Dow Chemical, Bristol PA) was administered orally at 25 mg/kg twice daily. Entrectinib (MedKoo Biosciences, Inc., Morrisville NC) was formulated in 0.5% methylcellulose, 1% Tween 80 in water and administered orally at 30 mg/kg twice daily. Animal body weights were recorded twice weekly. All in vivo experimental protocols were approved by the Eli Lilly and Company Animal Care and Use Committee. Eli Lilly and Company is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International. The in vivo efficacy study involving the CTG-0798 HNSCC PDX model harboring an ETV6-NTRK3 fusion (Supplementary Figure 7) was performed at Champions Oncology (Hackensack, NJ). CTG-0798 tumors were propagated in immunocompromised nude mice and tumor fragments were implanted in athymic nude mice for the study. Compound treatment began once average tumor volume reached 150-300 mm 3 . Statistical analysis was performed as described previously [2].

Cloning and cell transfection
Wild-type TPM3-NTRK1 as well as the NTRK1 kinase domain G595R and G667C mutants were cloned, including a 3' 3X FLAG-Tag inserted into a pcDNA3.1 vector backbone. NIH-3T3 cells were transfected using Lipofectamine™ LTX (ThermoFisher Scientific, Grand Island, NY). Stable clonal cell pools were generated after 350 μg/ml hygromycin selection for 14 days. Each stable pool was DNA sequenced confirming the appropriate wild-type or mutated NTRK1. Control vector containing eGFP served as a negative control.

Western blotting
After treatment, KM-12 cells were washed once with dPBS and harvested with 200 μL cell extraction buffer (ThermoScientific, CA) in the presence of HALT protease and phosphatase inhibitors (Pierce, Rockford, IL). 20 μg cell lysates were electrophoretically separated, transferred to PVDF membrane, and blocked in StartingBlock T20 (ThermoScientific, Rockford, IL) for 30 minutes.

Histological assessment of PDX tumors
Subcutaneous tumors were collected in 10% neutral buffered formalin, fixed, processed and embedded in paraffin. Tumor samples were trimmed to provide the maximum tissue area in the histologic section. Sections were made at 4 microns and were stained with hematoxylin and eosin. Slides were evaluated qualitatively by a board-certified pathologist (KMC).

Immunofluorescence and iCys imaging and quantitation
EL1989 PDX tumors were processed to paraffin blocks and sections were made as described above. Slides were baked at 60°F for 1 hour and then deparaffinized in xylene (4 × 10 minutes); rehydrated with ethanol/water immersions with final washes in TBST; blocked with Protein Block (Dako, Santa Clara CA) for 30 minutes; stained with a combination of Hoechst 33324 and Ki67 (Thermo Scientific RM9106, clone SP6)/anti-rabbit Alexa Fluor-647 (Invitrogen, Carlsbad CA) then imaged using an iCys Laser Scanning Cytometer (CompuCyte, Newton NJ) and a Marianas Digital Imaging Workstation configured with a Zeiss Axiovert 200M inverted fluorescence microscope (Intelligent Imaging Innovations, Denver CO). Percent Ki67+ area was calculated as the percentage of total tissue area (Hoechst positive) that is also Ki67+ positive. Quantitative data comparisons of treatment groups were done using the Dunnett's analysis in JMP statistics software (SAS).

PCR and DNA sequence verification of NTRK fusions
Material and method for EL1989 and KM-12 TPM3-NTRK1 fusion confirmation RNA was extracted from frozen EL1989 tumor section and KM-12 cell line using Direct-zol™ RNA MiniPrep cat# R2051 following manufacturer's protocol. cDNA was prepared from 1 ug RNA in 20 uL volume using SuperScript VILO MasterMix (Invitrogen, Carlsbad CA). mRNA sequences of TPM3 (NM_153649.3) and NTRK1 (NM_001012331.1) were obtained from NCBI for primer design. A forward primer upstream of the TPM3 fusion junction (as previously identified from RNAseq data) and a reverse primer downstream of the NTRK1 fusion junction were designed with online program Primer3 for PCR and Sanger sequencing primers. The primer sequences are TPM3_Ex3B_F5: AGGCAGATAGGAAGTATGAAGAGG, NTRK1_ Ex12_R5: AGGCATCACTGAAGTATTGTGG. PCR was carried out with Advantage ® 2 Polymerase Mix (Clontech, Mountainview CA) with Advantage 2 PCR buffer and cycled at 95°C for 2 minutes; 35 cycles of 95°C for 30 seconds; 65°C for 30 seconds, 72°C for 30 seconds, and a final extension of 72°C for 10 minutes. PCR was also performed for GAPDH house-keeping gene as internal control (GAPDH Ex3F: ATCCCATCACCATCTTCCAG, GAPDH Ex7R: CCATCACGCCACAGTTTCC). PCR products were purified with Wizard ® SV Gel and PCR Clean-Up System (Promega, Madison WI). Sequencing PCR was carried out using ABI BigDye Terminator v3.1 cycle sequencing kit (Life Technologies, Carlsbad CA). The resulting products were run on an ABI 3730xl DNA analyzer. All sequences were visually analyzed with Sequencher (Gene Codes Corp., Ann Arbor MI) and sequencing results confirmed TPM3 exon 7 (NM_153649.3) fusion to NTRK1 exon 9 (NM_001012331.1) in EL1989 PDX model and KM-12 cell line.

Material and method for CTG-0798 ETV6-NTRK3 fusion confirmation
CTG-0798 tumor frozen section was received from Champions. RNA was extracted using Direct-zol™ RNA MiniPrep following manufacturer's protocol (Zymo Research, Irvine CA). cDNA was prepared from 1 μg RNA in 20 μl volume using SuperScript VILO MasterMix (Invitrogen). mRNA sequences of ETV6 (NM_001987) and NTRK3 (NM_002530) were obtained from NCBI for primer design. A forward primer upstream of the ETV6 fusion junction (as previously identified from RNAseq data) and a reverse primer downstream of the NTRK3 fusion junction were designed with online program Primer3 for PCR and Sanger sequencing primers. The primer sequences are ETV6_Ex4_F5: CCTGGAAACTCTATACACACACAGC, NTRK3_ Ex16D_R5: GGAACTTATTCAGGTCTCCATGC. PCR was carried out with Advantage ® 2 Polymerase Mix (Clontech) with Advantage 2 PCR buffer and cycled at 95°C for 2 minutes; 35 cycles of 95°C for 30 seconds; 65°C for 30 seconds, 72°C for 30 seconds, and a final extension of 72°C for 10 minutes. PCR was also performed for GAPDH house-keeping gene as internal control (GAPDH Ex3F: ATCCCATCACCATCTTCCAG, GAPDH Ex7R: CCATCACGCCACAGTTTCC). PCR products were purified with Wizard ® SV Gel and PCR Clean-Up System (Promega #A9281). Sequencing PCR was carried out using ABI BigDye Terminator v3.1 cycle sequencing kit (Life Technologies). The resulting products were run on an ABI 3730xl DNA analyzer. All sequences were visually analyzed with Sequencher (Gene Codes Corp.) and sequencing results confirmed ETV6 exon 5 (NM_001987) fusion to NTRK3 exon 15 (NM_002530) in CTG-0798 PDX model.

Co-crystallization of merestinib with NTRK1 kinase and x-ray diffraction analysis
The kinase domain of TrkA containing residues 500-787 was cloned, expressed, and purified as previously described [3]. Purified protein was treated with 1.5 mM compound and crystallized in the space group P64 (a = b = 75.36 Å, c = 113.1 Å) at room temperature by hanging drop vapor diffusion against 2.2 M Sodium Chloride and 100 mM Bis-Tris Propane pH 7.0. Crystals were harvested with 25% glycerol as the cryoprotectant and flash frozen in liquid nitrogen. Diffraction data were collected at the LRL-CAT 31-ID beamline of the Advanced Photon Source, Argonne. The structure was solved by molecular replacement using the program PHASER [4] and Protein Data Bank entry 4PMM.pdb as a starting model [3]. The structure was subject to iterative rounds of manual rebuilding with the program COOT [5] and restrained refinement using the progam autoBUSTER (Global Phasing Ltd, Cambridge, U.K.). The final structure has crystallographic R/Rfree values of 0.173/0.207 at a resolution of 1.94 Å, and merestinib binding is unambiguously defined by the electron density maps.