Identification of Novel Small Molecule Inhibitors of Oncogenic RET Kinase

Oncogenic mutation of the RET receptor tyrosine kinase is observed in several human malignancies. Here, we describe three novel type II RET tyrosine kinase inhibitors (TKI), ALW-II-41-27, XMD15-44 and HG-6-63-01, that inhibit the cellular activity of oncogenic RET mutants at two digit nanomolar concentration. These three compounds shared a 3-trifluoromethyl-4-methylpiperazinephenyl pharmacophore that stabilizes the ‘DFG-out’ inactive conformation of RET activation loop. They blocked RET-mediated signaling and proliferation with an IC50 in the nM range in fibroblasts transformed by the RET/C634R and RET/M918T oncogenes. They also inhibited autophosphorylation of several additional oncogenic RET-derived point mutants and chimeric oncogenes. At a concentration of 10 nM, ALW-II-41-27, XMD15-44 and HG-6-63-01 inhibited RET kinase and signaling in human thyroid cancer cell lines carrying oncogenic RET alleles; they also inhibited proliferation of cancer, but not non-tumoral Nthy-ori-3-1, thyroid cells, with an IC50 in the nM range. The three compounds were capable of inhibiting the ‘gatekeeper’ V804M mutant which confers substantial resistance to established RET inhibitors. In conclusion, we have identified a type II TKI scaffold, shared by ALW-II-41-27, XMD15-44 and HG-6-63-01, that may be used as novel lead for the development of novel agents for the treatment of cancers harboring oncogenic activation of RET.


Introduction
The REarranged during Transfection (RET) gene codes for a single pass transmembrane tyrosine kinase (TK) receptor that is mutated in several human cancers [1]. In approximately 20% of human papillary thyroid carcinoma (PTC), RET exons encoding the tyrosine kinase domain are fused to the promoter region and the 5'-ter exons of heterologous genes, generating chimeric oncogenes, such as CCDC6-RET (RET/PTC1) or NCOA4-RET (RET/PTC3) [1,2]. Missense germline and somatic point mutations of RET are associated to familial (95%) and sporadic (50%) cases of medullary thyroid carcinoma (MTC), respectively. MTC associated RET mutations commonly target cysteine residues in the extracellular domain or the intracellular tyrosine kinase domain [1][2][3].
PTC-, NSCLC-and MPD-associated RET rearrangements and MTC-associated RET point mutations induce an oncogenic conversion of RET gene product by promoting ligandindependent kinase activation [1,11]. Unscheduled RET TK activation results in its constitutive autophosphorylation on specific tyrosine residues, such as Y905 and Y1062, in the intracellular domain. This, in turn, switches-on several signalling pathways, like the SHC/ RAS/MAPK pathway, that support cell transformation [1,11].
Based on this knowledge, RET targeting in cancer has been exploited via the identification of small molecule RET tyrosine kinase inhibitors (TKI) [12,13]. Two of them, vandetanib (ZD6474) and cabozantinib (XL184), have been approved for locally advanced or metastatic MTC [14,15]. Vandetanib binds to the active conformation of RET kinase (DFG-in) in the ATP-binding pocket and it is therefore a type I kinase inhibitor [16,17]. Though in vivo molecular mechanisms of acquired resistance are still unknown, RET mutations V804M/L or Y806C are able to cause a 50-(V804M/L) and 10-fold (Y806C) increase of in vitro IC 50 dose of vandetanib for RET [18,19] and V804M causes resistance to cabozantinib, as well [20]. It is still unknown whether such mutations might be involved in building resistance in patients.
Here we describe the identification and characterization of ALW-II-41-27 [21,22], XMD15-44 [23] and HG-6-63-01 as novel potent inhibitors of RET kinase. These compounds were type II inhibitors, designed to bind to the 'DFG-out' inactive kinase conformation, and all contain a 3-trifluoromethyl-4-methylpiperazinephenyl pharmacophore which occupies the hydrophobic pocket created by the rearrangement of the activation loop [23]. Thus, the common structure shared by the three compounds may represent a novel scaffold to generate potent and selective type II TKIs for cancers that exhibit constitutively active RET signaling.

Compounds
Compounds were synthesized in the Gray's laboratory according to published procedures [21,23], dissolved in dimethyl sulfoxide (DMSO) at 10 mM concentration and stored at -80°C. The synthetic procedure and characterization for HG-6-63-01 is provided in the Supplemental informations (S1 Methods). Final dosing solution was prepared on the day of use by dilution of the stock solution in cell growth media.

Molecule modeling
Though currently there are seven available X-ray structures of RET kinase in the public domain, all of them exhibit the 'DFG-in' active conformation of the activation loop and would not accommodate type II inhibitors. Therefore, here we first built the DFG-out model of RET kinase using the homology modelling method based on the RET sequence and the high-homology structure (PDB ID: 3DZQ) as the template with Swiss-model web server [24][25][26][27]. Then we used the autodock4.0 software to dock each ligand into the modeled DFG-out conformation of RET. The ligands were constructed by the online-tool: CORINA (http://www.molecular-networks. com). Lamarckian genetic algorithm with the default parameters was performed to get the candidate compounds. Then the docked compounds were clustered and sorted based on the binding free energy. The compound with the lowest binding free energy was shown as the binding mode.
TT human cell line was obtained in 2002 from ATCC and authenticated by RET genotyping; it was derived from a MTC and harbors a cysteine 634 to tryptophan (C634W) RET mutation [30]. TT cells were grown in RPMI 1640 supplemented with 20% fetal calf serum (GIBCO). MZ-CRC-1 human cells were kindly provided in 2009 by Robert F. Gagel (MD Anderson, Houston, TX) and authenticated by RET genotyping. MZ-CRC-1 cells were derived from a malignant pleural effusion from a patient with metastatic MTC and were found to bear a heterozygous (ATG to ACG) transition in RET resulting in the MEN2B-associated substitution of threonine 918 for methionine (M918T) [31]. MZ-CRC-1 cells were grown in DMEM supplemented with 10% fetal calf serum (GIBCO). Nthy-ori 3-1, a human thyroid follicular epithelial cell line immortalized by the SV40 large T gene, was obtained from European Collection of Cell Cultures (ECACC) (Wiltshire, UK) in 2010. Nthy-ori 3-1 was grown in DMEM supplemented with 10% fetal calf serum (GIBCO). TPC-1 cells were obtained in 1990 from M. Nagao (National Cancer Center Research Institute, Tokyo, Japan) and authenticated by RET/PTC genotyping [32]; they were grown in DMEM supplemented with 10% fetal calf serum (GIBCO).
Anti-SHC (#H-108) was from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-RET is a polyclonal antibody raised against the tyrosine kinase protein fragment of human RET [28]. Anti-phospho905 is a phospho-specific polyclonal antibody recognizing RET proteins phosphorylated at Y905 and anti-phospho1062 is a phospho-specific polyclonal antibody recognizing RET proteins phosphorylated at Y1062 [13]. Secondary antibodies coupled to horseradish peroxidase were from Santa Cruz Biotechnology.

Statistical analysis
To compare cell growth, we performed unpaired Student's t test using the Instat software program (Graphpad Software Inc). P values were two-sided, and differences were considered statistically significant at P <.02. IC 50 doses were calculated through a curve fitting analysis from last day values using the PRISM software program (Graphpad Software Inc).
Though more rarely than C634 and M918 mutations, other mutations targeting RET residues in the tyrosine kinase domain can cause familial or sporadic MTC [1,2]. We tested the activity of ALW-II-41-27, XMD15-44 and HG-6-63-01 towards various kinase domain-mutated RET oncoproteins. RET/L790F, RET/V804M and RET/S891A showed a sensitivity to ALW-II-41-27, XMD15-44 and HG-6-63-01 comparable to that of RET/C634R and RET/M918T proteins (S1 Fig). Instead, E768D, A883F and V804L RET mutants were less efficiently inhibited by the three compounds (S1 Fig). Of note, RET A883 residue is located in the VI Hanks domain adjacent to RET activation loop and it corresponds to G372 residue in ABL kinase whose mutation has been isolated in imatinib-resistant patients [33]. V804 corresponds to RET gatekeeper residue (T315 in ABL) in the ATP-binding pocket and was shown to mediate RET resistance to several kinase inhibitors when mutated to Methionine or Leucine [18,20]. Characterization of the three compounds in biochemical kinase assays using the Invitrogen Selectscreen indicated that they are equally potent against both wild type RET and RET/V804M. However, RET/ V804L mutants, likely because bulky nature of the leucine residue, featured 4-10 fold increased IC 50 compared to wt kinase (Fig 2C).
ALW-II-41-27 has been previously reported as a potent EPH family kinase inhibitor [21,22]; XMD15-44 has been characterized as a potent ABL inhibitor and HG-6-63-01 was developed as a general type II kinase inhibitor [23,24]. Kinome wide selectivity profiling indicates that the three compounds have multiple targets in addition to RET (S2 Fig and S1 Table).

Discussion
RET oncogenic conversion is a hallmark of several human cancers, including papillary and medullary thyroid carcinoma, lung adenocarcinoma and chronic myelomonocytic leukemia. In this light, RET kinase appears an attractive molecular target for anti-cancer therapy. Several anti-RET TKIs have been identified and vandetanib and cabozantinib have been recently approved for locally advanced or metastatic medullary thyroid carcinoma treatment [14,15]. Both compounds are multitarget kinase inhibitors able to inhibit kinases other than RET, including VEGFRII (vandetanib and cabozantinib), EGFR (vandetanib) and MET (cabozantinib) [11,12,14,15].
Here, we applied a structure-guided screen in order to identify novel RET TKIs. This screening resulted in the identification of ALW-II-41-27, XMD15-44, and HG-6-63-01 as novel and potent RET TKIs. Molecular modelling suggests that ALW-II-41-27, XMD15-44, and HG-6-63-01 recognize the 'DFG-out' conformation, consistent with being designed as type II kinase inhibitors. The three drugs impaired phosphorylation and signalling of various RET oncogenic mutants at nanomolar concentrations. They blunted proliferation of RET/ C634R and RET/M918T-transformed fibroblasts and of RET mutant thyroid cancer cells. Of note, the three compounds bound to the RET kinase bearing V804M mutation that instead are refractory to vandetanib and cabozantinib [19,20], whilst V804L mutation caused a 5-10 fold increase of the IC 50 dose of the three drugs. This could be explained by the bulky nature of Leucine that may interfere with drug binding. On the other hand, differently from vandetanib, whose activity was not affected by mutations in RET kinase activation loop, RET inhibitory effect of ALW-II-41-27, XMD15-44, and HG-6-63-01 was impaired secondary to A883F mutation targeting the RET activation loop, which is consistent with their type II binding mode.
In conclusion, ALW-II-41-27, HG-6-63-01 and XMD15-44 represent novel lead compounds able to efficiently inhibit RET. However, given their broad kinase selectivity, it will be important their further optimization to develop clinically-relevant agents against RET.