Sorafenib Functions to Potently Suppress RET Tyrosine Kinase Activity by Direct Enzymatic Inhibition and Promoting RET Lysosomal Degradation Independent of Proteasomal Targeting*

Germ line missense mutations in the RET (rearranged during transfection) oncogene are the cause of multiple endocrine neoplasia, type 2 (MEN2), but at present surgery is the only treatment available for MEN2 patients. In this study, the ability of Sorafenib (BAY 43-9006) to act as a RET inhibitor was investigated. Sorafenib inhibited the activity of purified recombinant kinase domain of wild type RET and RETV804M with IC50 values of 5.9 and 7.9 nm, respectively. Interestingly, these values were 6–7-fold lower than the IC50 for the inhibition of B-RAFV600E. In cell-based assays, Sorafenib inhibited the kinase activity and signaling of wild type and oncogenic RET in MEN2 tumor and established cell lines at a concentration between 15 and 150 nm. In contrast, inhibition of oncogenic B-RAF- or epidermal growth factor-induced ERK1/2 phosphorylation required micromolar concentrations of Sorafenib demonstrating the high specificity of this drug in targeting RET. Moreover, prolonged exposure to Sorafenib resulted in inhibition of cell proliferation and RET protein degradation. Using lysosomal and proteasomal inhibitors, we demonstrate that Sorafenib induces RET lysosomal degradation independent of proteasomal targeting. Furthermore, we provide a structural model of the Sorafenib·RET complex in which Sorafenib binds to and induces the DFGout conformation of the RET kinase domain. These results strengthen the argument that Sorafenib may be effective in the treatment of MEN2 patients. In addition, because inhibition of RET is not impaired by mutation of the Val804 gatekeeper residue, MEN2 tumors may be less susceptible to acquired Sorafenib resistance.

Germ line missense mutations in the RET (rearranged during transfection) oncogene are the cause of multiple endocrine neoplasia, type 2 (MEN2), but at present surgery is the only treatment available for MEN2 patients. In this study, the ability of Sorafenib (BAY 43-9006) to act as a RET inhibitor was investigated. Sorafenib inhibited the activity of purified recombinant kinase domain of wild type RET and RET V804M with IC 50 values of 5.9 and 7.9 nM, respectively. Interestingly, these values were 6 -7-fold lower than the IC 50 for the inhibition of B-RAF V600E . In cell-based assays, Sorafenib inhibited the kinase activity and signaling of wild type and oncogenic RET in MEN2 tumor and established cell lines at a concentration between 15 and 150 nM. In contrast, inhibition of oncogenic B-RAF-or epidermal growth factor-induced ERK1/2 phosphorylation required micromolar concentrations of Sorafenib demonstrating the high specificity of this drug in targeting RET. Moreover, prolonged exposure to Sorafenib resulted in inhibition of cell proliferation and RET protein degradation. Using lysosomal and proteasomal inhibitors, we demonstrate that Sorafenib induces RET lysosomal degradation independent of proteasomal targeting. Furthermore, we provide a structural model of the Sorafenib⅐RET complex in which Sorafenib binds to and induces the DFG out conformation of the RET kinase domain. These results strengthen the argument that Sorafenib may be effective in the treatment of MEN2 patients. In addition, because inhibition of RET is not impaired by mutation of the Val 804 gatekeeper residue, MEN2 tumors may be less susceptible to acquired Sorafenib resistance.
Germ line missense activating point mutations in the RET oncogene cause the cancer syndrome multiple endocrine neoplasia type 2 (MEN2) (1,8). Three mutation-specific disease phenotypes can be recognized: MEN2A, MEN2B, and a familial form of medullary thyroid carcinoma (FMTC). Medullary thyroid carcinoma (MTC), the cancer arising from the calcitoninsecreting C cells of the thyroid, is the common clinical feature of the three clinical variants. MEN2A patients develop, in addition to MTC, pheochromocytoma (the tumor arising from the adrenal medulla cells) and hyperplasia of the parathyroid (HPT). MEN2B patients develop MTC and pheochromocytoma; however, instead of HPT, they develop a more complex and aggressive phenotype with neuromas in tongue, lips, and eyelids; intestinal ganglioneuromas; thickening of corneal nerves; and marfanoid habitus. In papillary thyroid carcinoma, the * This work was supported by the Breakthrough Breast Cancer, Ministry of Education and Science of Spain, the Medical Research Council, the Italian Association for Research on Cancer, and European Union Prokinase Network Grant 503467. 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. 1  tumor arises from the follicular cells of the thyroid and is caused by chromosomal inversions or translocations in which the 5Ј-end of various genes are fused in frame with the tyrosine kinase domain of the RET gene (9). The resulting chimera oncoproteins called RET-PTCs, of which more than 10 variant forms have been identified, display constitutive tyrosine kinase activity caused by formation of homodimers between the N-terminal domains. Surgical resection, radioiodine treatment, and hormone replacement are the standard treatment for papillary thyroid tumors. In the case of MEN2 patients, early diagnosis is crucial because the disease does not respond to standard chemotherapy or conventional radiotherapy. Surgery is at present the only curative treatment for MTC, consisting of a total thyroidectomy. However, more than 50% of patients have persistent or recurrent disease after surgery and are prone to develop distal metastasis, for which there is no systemic treatment (1). More recently attention has focused on the use of small molecule inhibitors to target RET in MEN2-related tumors (6). Indeed, various compounds have been reported to inhibit oncogenic RET including PP1 (10) and PP2 (11), RPI-1 (12), CEP-701, CEP-751 (13), ZD-6474 (11,14), Gleevec (15), SU5416 (16), and NVP-AST487 (17). Sorafenib (BAY 43-9006) was designed originally as a RAF inhibitor (18) but in addition to RAF-1 and B-RAF is also able to inhibit VEGFR-2, VEGFR-3, PDGFR␤, c-Kit, and Flt-3 (18,19). A recent study (20) reported that Sorafenib could inhibit the kinase activity and signaling of wild type and oncogenic RET. Here we have extended these studies to further investigate: (i) the effect of Sorafenib against the enzymatic activity of purified recombinant kinase domain of RET WT and RET V804M , (ii) the specificity of RET inhibition by Sorafenib, (iii) the mechanism of oncogenic RET degradation induced by Sorafenib, and (iv) the structural insights of the mechanism by which Sorafenib inhibits RET kinase activity.

MATERIALS AND METHODS
Cell Lines and Reagents-HEK293 (human embryonic kidney) cells were transfected with pCMV vector alone or pCMV vector containing wild type RET (RET WT ), the MEN2A mutant RET C634R , the MEN2B mutant RET M918T , or the FMTC mutant RET S891A . Transfected cells were selected in 500 g/ml of G418 (Sigma-Aldrich), screened by Western blotting and cultured in Dulbecco's modified Eagle's medium (Invitrogen) medium supplemented with 10% fetal bovine serum (Invitrogen) and 500 g/ml of G418. MTC-TT cells (human medullary thyroid carcinoma cell line harboring the MEN2A-RET C634W oncoprotein) were grown in RPMI medium (Invitrogen) containing 20% fetal bovine serum. MZ-CRC-1 (human medullary thyroid carcinoma cell line harboring the MEN2B-RET M918T oncoprotein) and TGW-1 (human neuroblastoma cell line expressing wild type RET and GFR␣1 co-receptor) were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. BAY 43-9006 was provided by Bayer HealthCare Pharmaceuticals (West Haven, CT) and made up as a 10 mM stock solution in Me 2 SO. STI571 (Novartis, Basel, Switzerland) and PP1 (Promega, Madison, WI) were made up as 10 and 3 mM stock solutions in Me 2 SO, respectively. Concanamycin A, lactacystin, and MG132 were purchase from Calbiochem and dis-solved in Me 2 SO to give stock solutions of 5 M, 10 mM, and 50 mM, respectively.
In Vitro Kinase Assay-Baculovirus expressed recombinant RET, ALK, and Abl kinases were purified and employed in an enzyme-linked immunosorbent assay-based kinase assay, as described previously (16,21). The data shown are from three independent experiments performed in triplicate.
Confocal Microscopy-The cells were plated on glass coverslips for 24 h, after which cells were fixed in 4% paraformaldehyde for 1 h at room temperature, followed by permeabilization in 0.5% Triton X-100 in phosphate-buffered saline for 10 min. Immunostaining with primary antibodies for 1 h was followed by incubation with Alexa 486 anti-rabbit Ig and Alexa 555 antimouse Ig for 45 min. Alexa 647-phalloidin conjugated was used to stain actin filaments, and 4Ј,6Ј-diamino-2-phenylindole was used to stain the nucleus.
RNA Extraction, cDNA Synthesis, and Quantitative Real Time PCR-RNA extraction was performed using the RNeasy kit (Qiagen) under the manufacturer's instructions. For cDNA synthesis, 200 ng of mRNA were used with the Omniscript RT kit (Qiagen) following the manufacturer's instructions using oligo(dT) 20 primer (Qiagen). 1 l of cDNA was used per qPCR. Each analysis reaction was performed in triplicate. Glyceraldehyde-3-phosphate dehydrogenase was used as an endogenous control throughout all experimental analysis. Gene expression analysis was performed using TaqMan gene expression assays on an ABI Prism 7900HT sequence detection system (Applied Biosystems, Foster City, CA). Analysis was performed using the ⌬ Ϫ ⌬C t method, which determines fold changes in gene expression relative to a comparator sample (control). qPCR primers were purchased from Applied Biosystems. The Assayson-Demand references for RET and glyceraldehyde-3-phosphate dehydrogenase are Hs00240887_m1 and 4310884E, respectively.
Structural Model Analysis-To generate a model of RET in complex with Sorafenib, we assumed that Sorafenib would bind to the DFG out conformation of the kinase. A model of the RET⅐Sorafenib complex was therefore derived from the unphosphorylated RET kinase structure (Protein Data Bank code 2IVS) (22) with its activation segment substituted (DFG to APE motif, Asp 892 -Ala 919 ) with that of the c-Kit⅐Gleevec (STI-571) complex (Protein Data Bank code 1T46) (23), with mutations to convert the c-Kit activation segment sequence to that of RET. Superimposing this model onto the B-RAF⅐Sorafenib complex (Protein Data Bank code 1UWH) (24) allowed a fit of Sorafenib to the RET protein model. Further changes to the model included a small shift of the C-helix and of residues of the Gly loop to conformations observed in the c-Kit-STI-571 crystal structure. The resultant model provides a stereochemically plausible structure of the RET⅐Sorafenib complex. All modeling work was performed using COOT (CCP4), and the figures were generated with PyMOL (DeLano Scientific; www.pymol.org).
Statistical Analysis-Statistical analysis was performed using Prism software. The cell number assays were compared using one-way analysis of variance for repeating measures and the Dunnet test. Differences were considered statistically significant when p Ͻ 0.05.

Sorafenib Inhibits RET Enzymatic Activity at Low Nanomolar Concentration in Vitro-
To determine the inhibition of RET by Sorafenib, the His-tagged tyrosine kinase domain (amino acids 700 -1020) from wild type RET was expressed in Sf9 cells and purified using a two-step chromatographic proto-col ( Fig. 1A, inset). In an in vitro kinase assay, Sorafenib blocked enzymatic activity with an IC 50 ϭ 5.9 Ϯ 1.2 nM (Fig. 1A). To compare this potent inhibitory capacity of Sorafenib with other known RET tyrosine kinase inhibitors, the in vitro kinase assay was repeated in the presence of Gleevec (15). Gleevec blocked the kinase activity of wild type RET with an IC 50 ϭ 12 Ϯ 2 M compared with an IC 50 ϭ 0.3 Ϯ 0.06 M for the well characterized Gleevec target, Abl (Fig. 1B). Next, the specificity of Sorafenib was investigated by in vitro kinase assay with recombinant RET, Abl, and ALK kinase domains (Fig. 1C). Sorafenib blocked the enzymatic activity of RET, Abl, or ALK with IC 50 ϭ 0.006, 25, and Ͼ500 M, respectively. Finally, to assess the activity of the Sorafenib used in these experiments, an in vitro kinase assay with recombinant B-RAF V600E was performed (Fig.  1D). Sorafenib inhibited B-RAF V600E with an IC 50 ϭ 45 nM, which is essentially identical to that previously reported (25). The K i values for Sorafenib inhibition of wild type RET and B-RAF V600E were 0.45 and 5.6 nM, respectively. The lower K i value for RET indicates a high binding affinity of the drug to the protein. Together with the lower IC 50 values, this indicates that in vitro Sorafenib is a more potent inhibitor of RET tyrosine kinase rather than oncogenic B-RAF.
In Vivo Inhibition of RET Phosphorylation, Downstream Signaling, and Cell Proliferation by Sorafenib-It has recently been demonstrated that Sorafenib can inhibit RET activation and cell proliferation in tumor cells harboring a MEN2A-RET C634W mutation (MTC-TT cell line) or a RET/PTC1 oncogene (TPC1 cell line) (20). Here we have extended these studies to also examine the effects of Sorafenib on tumor cells expressing a MEN2B-RET M918T mutation (MZ-CRC-1 cell line) or wild type RET together with the GFR␣1 co-receptor (TGW-1 cell line). The cells were treated with Sorafenib for 90 min, and the levels of RET phosphorylation were assessed by Western blotting using an antibody that recognizes RET only when phosphorylated at Tyr 1062 . For the TGW-1 cells expressing wild type RET, GDNF (10 ng/ml) was added 30 min after drug treatment. In all of the tumor cell lines examined, a reduction in RET Tyr 1062 phosphorylation was observed at a concentration of Sorafenib between 15 and 150 nM ( Fig. 2A). Moreover, this decrease in RET phosphorylation was accompanied by a decrease in ERK1/2 phosphorylation. In cell growth assays it was demonstrated that Sorafenib treatment at concentrations as low as 0.1 M resulted in a significant reduction in cell number (Fig. 2B). The in vitro kinase assays demonstrated that RET is more sensitive than oncogenic B-RAF V600E to Sorafenib inhibition (Fig. 1).
The tumor cell lines expressing the different oncogenic forms of RET were derived from different patients and will therefore have very different genetic backgrounds. Consequently, to assess whether the observed inhibitor effects of Sorafenib were specific to RET, the assays were repeated using established HEK293 cells expressing wild type or different oncogenic RET variants (RET C634R , RET M918T , and RET S891A ). Again an inhibition of RET Tyr 1062 phosphorylation was detected at concentrations of Sorafenib between 15 and 150 nM (Fig. 3, A and B). As expected HEK293 cells transfected with vector alone showed no expression of RET and a very low level of ERK1/2 phosphorylation (Fig. 3B). Similarly, treatment of cells expressing oncogenic RET variants with 0.1 or 1 M Sorafenib over an 8-day period resulted in a significant decrease in cell number, whereas no effect of Sorafenib was observed on HEK293 cells transfected with vector alone (Fig. 3C).
Sorafenib Inhibits Oncogenic B-RAF-and EGF-induced ERK1/2 Phosphorylation at Micromolar Concentrations: Specificity of Sorafenib Inhibition-Sorafenib was first designed as a RAF inhibitor. As a consequence inhibition of RAF by Sorafenib will inhibit downstream ERK1/2 signaling. To demonstrate that the inhibition of RET-induced ERK1/2 phosphorylation by Sorafenib is not caused by the dual specificity of the compound for RET and RAF, two experiments were performed. First, HEK293 cells transfected with Myc-tagged oncogenic B-RAF V600E were treated with increasing amounts of Sorafenib for 90 min (Fig.  4A). Western blotting analysis revealed that Sorafenib was able to inhibit oncogenic B-RAF-induced ERK1/2 phosphorylation only at concentrations of 4.5 M (Fig. 4A). Second, TGW cells, which do not express oncogenic RET, were stimulated with EGF and treated with increasing amounts of Sorafenib (Fig. 4B). Activation of the EGF receptor results in activation of RAF and phosphorylation of ERK1/2. Again this ERK1/2 phosphorylation was only inhibited by high concentrations (3 M) of Sorafenib. Together these experiments demonstrate that inhibition of RET-induced ERK1/2 phosphorylation at submicromolar concentrations of Sorafenib results from direct inhibition of RET rather than inhibition of other kinase targets.
Sorafenib Induces Oncogenic RET Lysosomal Degradation-To further investigate the mechanisms by which Sorafenib impairs proliferation in RET-expressing cells, HEK293 cells expressing RET C634R or RET M918T were treated with Sorafenib for 48 h, and the level of RET protein was monitored by Western blotting. In both cell lines, treatment with Sorafenib resulted in a significant reduction in RET protein levels, and it was noted that this effect was stronger for RET M918T compared with RET C634R (Fig. 5A). Loss of the RET-PTC1 fusion protein following Sorafenib treatment was also observed in the TPC1 cell line (data not shown). Real time quantitative PCR analysis revealed no change in RET C634R or RET M918T mRNA levels following Sorafenib treatment (Fig. 5B), suggesting that the decrease in RET protein levels does not result from transcriptional regulation. To uncover the mechanism of oncogenic RET degradation, the cells were treated with Sorafenib for 48 h, and then lysosome or proteasome inhibitors were added for 8 h  prior to cell lysis. In HEK293 cells expressing RET C634R , Sorafenib-induced RET degradation was rescued by the lysosome inhibitor concanamycin A but not by the proteasome inhibitors lactacystin and MG132. Similarly, Sorafenib-induced degradation of RET M918T was unaffected by treatment with either MG132 or lactacystin, whereas degradation was rescued by concanamycin A, albeit to a lesser extent than observed with RET C634R (Fig. 5C, left-hand panels). Treatment of these established cell lines with the lysosome and proteasome inhibitors in the absence of Sorafenib revealed no effect of MG132 or lactacystin but a significant increase in RET protein levels followed treatment with the lysosomal inhibitor concanamycin A (Fig.  5C, right-hand panels). Again, a greater effect on RET C634R compared with RET M918T was observed.
To investigate further this difference in efficiency between RET C634R and RET M918T degradation, the subcellular localization of RET was examined by confocal microscopy. RET C634R was detected at the cell surface and showed substantial co-localization intracellularly with the lysosomal marker LAMP1 (lysosome-associated membrane protein 1). RET M918T was found to localize with intracellular organelles, although the overlap with LAMP1 was less pronounced (Fig. 5D). When cells were treated with the Sorafenib for 48 h, as expected, a reduction in levels of the RET protein was observed by confocal analysis (data not shown). Similarly, treatment of cells with concanamycin A in the absence of Sorafenib resulted in an accumulation of intracellular RET (data not shown). To confirm that these data were not cell type-dependent, Western blotting analyses were repeated with the tumor cell lines MTC-TT and MZ-CRC-1, which express MEN2A-RET C634W and MEN2B-RET M918T , respectively. As with the transfected cell lines, Sorafenib treatment resulted in RET degradation. In addition, this degradation was reverted with the lysosomal inhibitor concanamycin A (Fig. 6A). Finally, in HEK293-RET WT cells, Sorafenib treatment decreased the level of wild type RET protein, but this was not as dramatic as observed with oncogenic RET and failed to reach statistically significance. This partial decrease was rescued by concanamycin A treatment, and as with oncogenic RET, concanamycin A treatment alone resulted in a significant increase in the levels of RET protein (Fig. 6B).
Sorafenib Inhibits RET V804M Kinase Activity with Similar Efficacy as Wild Type RET-Mutations affecting the active domain of tyrosine kinase receptors can confer resistance to small molecule inhibitors (26). In the case of RET, mutations at the gatekeeper residue, Val 804 , have been reported to confer resistance to PP1, PP2, and ZD6474 (11). Similarly a novel MEN2B tandem mutation involving Val 804 (V804M/E805K) also confers resistance to the PP1 inhibitor (27). However, in in vitro kinase assays using immunoprecipitated RET, the IC 50 for Sorafenib inhibition of RET C634R , which contains a wild type kinase domain, was only two-3-fold lower then for RET V804L or RET V804M (20). Here we have performed in vitro kinase assays using recombinant wild type RET and RET V804M kinase domains. Both kinase domains showed a similar IC 50 of Sorafenib inhibition (Fig. 7A). As a control, kinase activity was assessed in the presence of PP1 inhibitor. Consistent with previous reports (11), wild type RET kinase activity, but not the activity of the RET V804M mutant, was inhibited by PP1 (Fig. 7B).
Structural Model of RET in Complex with Sorafenib-A model for RET in complex with Sorafenib was generated guided by x-ray crystal structures of the B-RAF⅐Sorafenib complex (24), the unphosphorylated state of RET (22), and the structure of c-Kit in complex with Gleevec (STI-571) (23). In the B-RAF⅐Sorafenib crystal structure, Sorafenib spans the interface between the N-and C-lobes of the kinase domain. The proximal trifluoromethyl phenyl ring (ring A) of the inhibitor engages a pocket between the C and E ␣-helices, whereas its urea moiety forms two hydrogen bonds with the protein: one via its amide nitrogen atom to the carboxylate side chain of the catalytic Glu 501 residue and the second via its carbonyl moiety to the main chain nitrogen of Asp 594 of the DFG motif (24). Together, these interactions promote the inactive conformation of the kinase with the DFG motif adopting the DFG out conformation, and the N-terminal region of the activation segment swung out toward the Gly loop of the kinase. This structure of the B-RAF⅐Sorafenib complex is reminiscent of the inactive conformation of the c-Abl tyrosine kinase domain in complex with Gleevec (28,29) and with the mitogen-activated protein kinase⅐BIRB 796 complex (30). Specifically, Gleevec and Sorafenib induce the DFG out conformation of their cognate kinases via similar mechanisms: displacement of the phenyl ring of the DFG motif Phe residue via an aromatic moiety augmented by hydrogen bonds between the urea group of the inhibitor and the DFG motif.
Recently, a structure of Gleevec in complex with c-Kit revealed a mode of inhibition similar to that observed for c-Abl. The kinase domains of RET and c-Kit are structurally very similar, sharing 41% sequence identity overall, increasing to 57% for residues of the activation segment. Such high sequence similarity would suggest that the activation segments of both kinases would adopt similar DFG out conformations in their inactive states. We based our model of a RET⅐Sorafenib complex on the assumption that the inhibitor would promote the DFG out conformation of the activation segment, similar to that of B-RAF in complex with Sorafenib, and c-Abl and c-Kit in complex with Gleevec. In all three kinases, B-RAF, c-Kit, and RET, the relative interdomain arrangements are very similar. A model of the RET⅐Sorafenib complex was therefore derived using the RET kinase struc-ture with its activation segment substituted with that of the c-Kit⅐Gleevec complex. RET residues were then substituted for c-Kit residues within the activation segment. Superimposing this model onto the B-RAF⅐Sorafenib complex allowed a fit of Sorafenib to the RET protein model (Fig.  8A). This provides a stereochemically plausible model of the RET⅐Sorafenib complex. Overall, nine residues that contact Sorafenib (defined as a contact distances less than 4 Å) differ between the two kinases (Fig. 8D). Residues of RET that interact with ring A and the urea group of the inhibitor are highly conserved with those in the B-RAF⅐Sorafenib complex, with only Val 787 and Leu 854 for Ile substitutions in RET. There are a total of six differences in the residues between B-RAF and RET that contact the central (B) and pyridyl (C) rings between the two kinases. These mainly comprise changes in van der Waals' contacts but also include a difference in the gatekeeper residue (Thr 529 and Val 804 in B-RAF and RET, respectively). In the B-RAF⅐Sorafenib complex, the hydroxyl group of the Thr 529 side chain is buried by the central ring of the inhibitor. Substitution of Val 804 for Thr in RET would potentially remove an unfavorable unsatisfied hydrogen bond present in the B-RAF⅐Sorafenib complex.

DISCUSSION
Tyrosine kinase receptors play a major role in cancer development. They control many biological processes in the cell, but when mutated, they become constitutively active or display altered signaling properties; hence they become oncoproteins. Understanding the molecular bases of cancer is necessary to gain a better insight into the genotype-phenotype correlations but also is crucial for the discovery of new therapeutic approaches. Indeed, tyrosine kinase receptor-based therapies have reached widespread clinical use in, for example, breast cancer (inhibition of HER2 by Herceptin), gastrointestinal stromal tumors (inhibi-  -hand panels). In parallel experiments, the cells were treated with concanamycin A or lactacystin in the absence of Sorafenib (right-hand panels). The cell lysates were analyzed by Western blotting using the indicated antibodies. The data represent the means in percentage of RET protein levels versus control (␣-tubulin) Ϯ S.E. of four independent experiments. *, p Ͻ 0.05; **, p Ͻ 0.01. D, confocal microscopy images of established HEK293 cells expressing RET C634R and RET M918T inmunostained with RET and LAMP1 antibodies, followed by Alexa 488 anti-rabbit Ig (green) and Alexa 555 anti-mouse Ig (red). Actin filaments were stained with Alexa 647-phalloidin (blue), and nuclei were counterstained with 4Ј,6Ј-diamino-2phenylindole (gray). Scale bar, 15 m. tion of c-Kit by Gleevec), and non-small cell lung cancer (inhibition of EGF receptor by Gefitinib) (31).
Sorafenib was initially identified as a potent RAF kinase inhibitor, but in addition to RAF-1 and B-RAF, Sorafenib has subsequently been shown to also inhibit VEGFR-2, VEGFR-3, PDGFR␤, c-Kit, and Flt-3 (18,19,26). Recently, Carlomagno et al. (20) have demonstrated that Sorafenib can inhibit the kinase activity of full-length RET immunoprecipitated from cultured cells with a an IC 50 of 50 nM. Using recombinant soluble kinase domain in in vitro kinase assays, we have confirmed that Sorafenib is a potent RET kinase inhibitor. However, the IC 50 of inhibition (5.9 nM) was ϳ10-fold lower than previously reported. The reasons for this difference may reflect differences in the assays employed; however, it is also possible that the kinase activity of RET following immunoprecipitation may be altered by co-immunoprecipitation of RET associated kinases. A novel finding in our studies was that the IC 50 values of Sorafenib inhibition obtained with wild type RET were 6 -7-fold lower than those obtained with B-RAF V600E . These findings were also supported by comparing their K i values: 0.45 and 5.6 nM for wild type RET and B-RAF V600E , respectively, indicating a higher affinity and efficacy of Sorafenib inhibition of wild type RET than for B-RAF V600E . Moreover, in cell-based assays, Sorafenib inhibited oncogenic B-RAF-or EGF-induced ERK1/2 activation at micromolar concentrations, giving evidence of the high specificity of RET inhibition by Sorafenib.
Our structural model of a RET⅐Sorafenib complex predicts that the inhibitor will promote the DFG out conformation of the activation segment in the kinase domain. Consequently, the overall affinity of the inhibitor for RET will be determined by the overall energy difference between the DFG out and DFG in conformations and by the specific interactions between the protein and inhibitor. The inability of the Src kinase to adopt the DFG out conformation explains its insensitivity to Gleevec, despite sharing with c-Abl all the residues that contact the inhibitor (32). Our modeling indicates that the increased affinity of Sorafenib for RET could be due to a combination of more favorable interactions between the kinase and inhibitor and an equilibrium that favors the transition to the DFG out conformation. One of the nine Sorafenib contact residues that differ between RET and B-RAF is Ser 891 . Interestingly, a mutation at this residue to an alanine (RET S891A ) is found in FMTC patients and, as illustrated in Fig. 3C, displays constitutive kinase activity as monitored by RET phosphorylation and downstream phosphorylation of ERK1/2. In cell-based assays, substitution of alanine for serine at residue 891 did not reduce sensitivity to Sorafenib inhibition (Fig. 3B). Similarly, substitution of Val 804 for Met in RET (RET V804M ) did not alter the kinetics of Sorafenib inhibition in vitro (Fig. 7A). Together, these data indicate that the difference in sensitivity to Sorafenib observed between RET and B-RAF does not reside in a single amino acid interaction between the kinase domain and the drug rather than a combination of different amino acids along the interface between kinase and inhibitor. The potent inhibition of RET by Sorafenib was confirmed in tumor cell lines and established HEK293 cells expressing wild type and oncogenic RET. In these cell lines, Sorafenib was found to inhibit RET tyrosine kinase activity and downstream signaling and to specifically inhibit RET-dependent cell proliferation. This confirms and extends the work of Carlomagno et al. (20). To further investigate the mecha-  nism of RET inhibition in these cells, the effects of Sorafenib on RET degradation were investigated. Exposure to the drug for 48 h had no effect on RET mRNA levels but resulted in a significant decrease in RET protein, indicating that Sorafenib has a 2-fold mechanism of receptor inhibition by both directly inhibiting RET kinase activity and promoting its degradation. The mechanism of Sorafenib-induced RET degradation was addressed by using proteasomally and lysosomally specific inhibitors in Western blotting experiments. Treatment with the lysosomal inhibitor concanamycin A reversed the Sorafenib-induced RET C634R degradation and provided a lesser, but still significant, rescue of RET M918T degradation. In contrast, no reversal was observed in the presence of two proteasome inhibitors, lactacystin and MG132, demonstrating that Sorafenib induces the lysosomal degradation of oncogenic RET independent of proteasome targeting. Similar data were obtained with tumor cell lines with endogenous expression of these oncogenic RET receptors. In these degradation experiments it should be noted that concanamycin A treatment does not rescue the level of RET protein in Sorafenib-treated cells to that found after treatment with concanamycin A treatment alone. The reason for this is most likely that cells were treated with Sorafenib for 48 h and that lysosomal or proteosomal inhibitors were only added for the final 8 h of the experiment because of their cytotoxic effects in longer term culture.
Further, confocal microscopy analysis revealed that in untreated cells, a significant proportion of RET C634R co-localizes with LAMP1, suggesting that this active RET oncoprotein is constitutively internalized from the plasma membrane and targeted for lysosomal degradation. In support of this, treatment of cells with concanamycin A in the absence of Sorafenib resulted in a significant increase in level of RET C634R protein. It was notable in these experiments that concanamycin A was less effective at blocking Sorafenib-induced RET M918T degradation. In this respect, it is of interest that in untreated cells a lower proportion of RET M918T was found co-localized with LAMP1, indicating that the rate of constitutive RET M918T lysosomal degradation is lower than that for RET C634R . Again, this is supported by data showing that concanamycin A treatment in the absence of Sorafenib resulted in a statistically significant increase in RET M918T levels but that this increase was lower than that observed with RET C634R . The inhibitor studies and confocal analysis presented here support a model in which oncogenic RET, and in particular the MEN2A-RET C634R mutant, is constitutively internalized from the plasma membrane and that a proportion of this internalized receptor is targeted for degradation in the lysosomes. Treatment with concanamycin A prevents this lysosomal degradation and results in an accumulation of intracellular RET. Because treatment with Sorafenib accelerates the rate of degradation, it is likely that inhibition of the kinase activity shifts the equilibrium toward lysosomal degradation. Interestingly, in these experiments, concanamycin A treatment alone also increased the levels of wild type RET, indicating that in the absence of ligand, this receptor can be internalized and targeted to the lysosomes. However, it was notable that the effects of Sorafenib on receptor degradation were less striking that for oncogenic RET, suggesting that the majority of wild type RET in the absence of GDNF stimulation is kinase inactive and therefore shows little response to Sorafenib. This demonstration of a lysosomal degradation mechanism for RET is in keeping with data from other receptor tyrosine kinases that have been shown to be internalized via clathrin-coated pits and then trafficked through early and late endosome for either recycling back to the plasma membrane or targeting to the lysosome (33). However, to date there is conflicting data on the mechanism of wild type RET degradation following binding to its GNDF ligand. In cell lines it has been demonstrated that GDNF treatment results in clathrin-mediated internalization of RET into endosomes, suggesting that subsequent degradation will occur in the lysosome (34). In primary sympathetic neuron and transfected COS cells, activation with GDNF is rapidly followed by wild type RET ubiquitination and degradation, which can be blocked by proteasome inhibitors (35,36). In contrast to the data reported here, Carniti et al. (10) have reported that treatment of cells expressing either RET C634R or RET M918T with the kinase inhibitor PP1 results in relatively rapid degradation of the oncogenic RET through proteasomal targeting. The reasons for the differences obtained with these two kinase inhibitors is not known, but it has been suggested that PP1 may induce structural perturbations in RET that result in the recruitment of the stress-inducible machinery and subsequent proteasomal degradation (10). If binding of Sorafenib does not induce such perturbations, then this would account for the observed differences in mechanism and time course of receptor degradation.
The effect of Sorafenib on the kinase activity of mutant RET V804M was also analyzed. This mutant is associated with the FMTC phenotype and affects the gatekeeper amino acid of the kinase domain. Importantly, mutations at this site have been shown to confer resistance to previously tested drugs (11,20). Here we demonstrate that mutation of the gatekeeper residue (Val 804 ) to Met desensitizes the kinase to inhibition by PP1 but has essentially no effect on the inhibition by Sorafenib (Fig. 5). Analysis of the RET-PP1 structure (Protein Data Bank code 2IVV) (22) indicates that the aliphatic side chain of Val 804 packs against the benzyl moiety of PP1; a bulky Met residue could only be accommodated by a shift of position of the PP1 inhibitor, and presumably this weakens dramatically its affinity for the kinase. Intriguingly, the model of the RET⅐Sorafenib complex indicates a similar close contact between the Val 804 side chain and the central ring of the inhibitor (Fig. 6). To accommodate a Met residue at position 804, the Sorafenib molecule would also be required to shift its position. Because inhibitory potency is maintained, we assume that compensatory new interactions are formed with the protein, possibly facilitated by conformational changes in RET. The different mode of binding of Sorafenib and PP1 to the DFG out and DFG in conformations of RET, respectively, and the more extensive contacts formed between Sorafenib and RET compared with PP1 and RET could account for these differences. In this respect, it is interesting that mutation of the gatekeeper residue Thr 674 in the FIP1L1-PDGFR␣ oncogene conferred resistance to Gleevec but not to Sorafenib (26).
Taken together, the data presented here indicate that Sorafenib is indeed a potent RET kinase inhibitor and points toward its promising therapeutic potential for patients affected with MEN2-related tumors. In particular, the demonstration that Sorafenib is an effective inhibitor of oncogenic RET containing mutations at Val 804 indicates that tumors may not be able to acquire resistance to this drug by subsequent somatic mutation in this gatekeeper residue.