LMTK3 is essential for oncogenic KIT expression in KIT-mutant GIST and melanoma.

Certain cancers, including gastrointestinal stromal tumor (GIST) and subsets of melanoma, are caused by somatic KIT mutations that result in KIT receptor tyrosine kinase constitutive activity, which drives proliferation. The treatment of KIT-mutant GIST has been revolutionized with the advent of KIT-directed cancer therapies. KIT tyrosine kinase inhibitors (TKI) are superior to conventional chemotherapy in their ability to control advanced KIT-mutant disease. However, these therapies have a limited duration of activity due to drug-resistant secondary KIT mutations that arise (or that are selected for) during KIT TKI treatment. To overcome the problem of KIT TKI resistance, we sought to identify novel therapeutic targets in KIT-mutant GIST and melanoma cells using a human tyrosine kinome siRNA screen. From this screen, we identified lemur tyrosine kinase 3 (LMTK3) and herein describe its role as a novel KIT regulator in KIT-mutant GIST and melanoma cells. We find that LMTK3 regulated the translation rate of KIT, such that loss of LMTK3 reduced total KIT, and thus KIT downstream signaling in cancer cells. Silencing of LMTK3 decreased cell viability and increased cell death in KIT-dependent, but not KIT-independent GIST and melanoma cell lines. Notably, LMTK3 silencing reduced viability of all KIT-mutant cell lines tested, even those with drug-resistant KIT secondary mutations. Furthermore, targeting of LMTK3 with siRNA delayed KIT-dependent GIST growth in a xenograft model. Our data suggest the potential of LMTK3 as a target for treatment of patients with KIT-mutant cancer, particularly after failure of KIT TKIs.

INTRODUCTION

The class III receptor tyrosine kinase KIT and its cognate ligand, stem cell factor (SCF), play important roles during development, as well as in adult stem cell maintenance. Normal KIT activity is induced upon SCF binding and signals to numerous downstream pathways, including PI3K/AKT and MEK/ERK, to drive proliferation and survival of cells, . Gain-of-function mutations in KIT causing ligand-independent kinase activation are known to drive neoplastic growth in multiple tissues, , . KIT-mutant tumors include the majority of gastrointestinal stromal tumors (GIST) and mastocytosis, as well as subsets of melanoma, acute myeloid leukemia, and seminoma, , , , . Somatic activating KIT mutations are rare overall, but within certain tumor types or subtypes, the frequency can be quite high. Of the 5,000 new cases of GIST that are diagnosed each year in the U.S., over 70% of cases are caused by KIT mutations. In melanoma, KIT mutations make up the most common oncogenic driver mutations in acral and mucosal subtypes, as well as melanomas arising from chronically sun-damaged skin, . Both GIST and these melanoma subtypes have poor response to conventional cytotoxic therapies and radiation, . However, KIT TKIs, such as imatinib, have improved outcomes for these patients. The median overall survival of patients with advanced GIST is estimated to be 7–8 years, and a subset of patients live more than 10 years, , ; this is in contrast to an overall survival of 12–18 months with conventional chemotherapies. Although no KIT-targeted treatments are yet approved for KIT-mutant melanoma, early clinical trials have shown responses in some patients, –.

While KIT TKI treatments, can provide significant clinical benefit to patients, they are rarely curative. The majority of GIST patients will develop drug resistance over the course of KIT TKI treatment; the median time to tumor progression on first-line imatinib therapy is 20–24 months. GIST second- and third-line GIST therapies, sunitinib and regorafenib, respectively, are used to treat patients after failure on imatinib, but are also limited by resistance of their own, . Resistance to KIT TKIs in GIST is almost exclusively caused by secondary KIT mutations, most commonly affecting the ATP binding pocket (V654A, T670I) or the activation loop (codons 816, 820, 822, 823 or 829 with multiple amino acid substitutions reported for most of these codons), , , . Primary mutations that affect these domains can also confer drug resistance. Nonetheless, KIT TKI-resistant GIST remain dependent on KIT and therefore KIT is still a relevant target. Disease management is complicated in the advanced setting with the existence of inter- and intra-lesional heterogeneity of KIT mutations. Patients can have various secondary mutations between and within lesions, and each mutation can have different sensitivity profiles to individual KIT TKIs, .

In the face of heterogeneous KIT mutations in these tumors, KIT TKIs have limited ability to control KIT-mutant disease once resistance develops, leaving patients with few treatment options. Because of this, we sought to identify novel targets in KIT-mutant cancer cells using a previously described siRNA screen methodology called RNAi Assisted Protein Identification (RAPID), which knocks down potential targets, including all predicted members of the human tyrosine kinome, . Using this approach, we identified the protein kinase lemur tyrosine kinase 3 (LMTK3) as an essential gene in KIT-mutant GIST and melanoma cells. Despite its name, LMTK3 has been found to be a serine/threonine kinase with only a few identified kinase substrates or functions, . LMTK3 has been implicated in promoting cancer growth, and some roles of LMTK3 have been described in breast cancer, including driver protein and transcriptional regulation, , , . However, the roles for LMTK3 are not fully understood in other tumor types.

This study describes our findings of the role for LMTK3 in promoting the viability of all KIT-mutant GIST and melanoma cells studied to date, including those with mutations conferring KIT TKI resistance. Our data show that LMTK3 is both necessary and sufficient for KIT protein expression. Specifically, we show that loss of LMTK3 reduces KIT translation in GIST cells and thereby reduces KIT expression and downstream signaling. These data support that LMTK3 is a novel regulator of KIT expression, and because of its essential role in KIT-mutant cells regardless of KIT mutation, LMTK3 represents a potential therapeutic target in TKI-resistant KIT-mutant cancers.

RESULTS

LMTK3 identified as essential for viability of mutant KIT-dependent cells

To identify novel targets in KIT-mutant cancers, we performed a siRNA screen using cell viability as a read-out. The siRNA library encompassed all known and predicted human tyrosine kinases, as well as NRAS and KRAS (93 genes total), , . We measured viability 96 hours after transfecting cells with siRNA pools against each target in three KIT-mutant cell lines: two GIST and one melanoma (GIST430 [exon 11], GIST-T1, and MaMel) (Supplemental Table 1).

Candidate genes that negatively affected viability were determined for each cell line by calculating the median and standard deviation across each entire screen experiment, as has been described previously, ; candidates for further comparison were those that reduced cell viability greater than one standard deviation from the median. All candidates within this cut off were found to have a statistically significant effect on cell viability (Supplemental Figure 1). These genes were then compared across the three cell lines to identify common candidates. There were two candidates besides KIT (a positive control) that were shared by all three cell lines: PTK2 and LMTK3 (Figure 1A). Protein tyrosine kinase 2 (PTK2), or focal adhesion kinase (FAK) has been described to have a role in GIST viability and imatinib resistance, , . LMTK3, however, is a novel candidate in KIT-mutant cancers.

Figure 1:

Silencing of the protein kinase LMTK3 specifically reduces viability of mutant KIT-dependent GIST and melanoma cells.

A. Venn diagram of hits from RAPID tyrosine kinase siRNA screens performed in KIT-mutant GIST430 (ex11), GIST-T1, and MaMel cell lines. B. Viability 96 hours post-transfection with non-targeting (NT), LMTK3, and KIT siRNA. C. Viability of KIT-mutant GIST cell lines was measured 96 hours post-transfection with indicated siRNAs. D. Viability of KIT-independent GIST and melanoma cells measured 96 hours post-transfection with indicated siRNA. E-F. Viability of GIST430 (ex11) and GIST430-LMTK3myc cells 96 hours post-transfection with shown siRNA. The p values of one-way ANOVA for each cell line to NT siRNA are indicated by asterisks: *, p<0.05; **, p<0.005; ***, p<0.001; ****, p<0.0001. (N>3).

LMTK3 was validated in independent siRNA experiments using silencing of the essential cell cycle gene PLK1 served as a positive control as indication of efficiency of transfection. KIT siRNA served as an additional positive control in mutant KIT-dependent cell lines and showed significant negative effect on cell viability in GIST-T1, GIST430 (ex11), and MaMel, in most cases comparable to PLK1 silencing; the silencing of LMTK3 decreased viability to similar levels in all three cell lines (Figure 1B). Moreover, to corroborate these data, we found that multiple individual siRNAs against LMTK3 decreased viability in KIT-mutant GIST and melanoma cell lines, as well as knocked down LMTK3 at the protein level (Supplemental Figure 2).

LMTK3 is specifically essential for cell viability of KIT-dependent GIST and melanoma

To determine the breadth of the effect of LMTK3 silencing in KIT-mutant cells, we expanded experiments to include a library of GIST and melanoma cell lines. These included GIST cell lines derived from those used in our initial screens (GIST430 [ex 11] and GIST-T1) and others that have secondary KIT mutations conferring resistance to KIT TKIs (Supplemental Table 2). Similar to KIT or PLK1 silencing, LMTK3 silencing in all mutant KIT-dependent cell lines, including those with KIT TKI-resistance mutations, decreased cell viability relative to non-targeting (NT) control siRNA (Figure 1C). In contrast, KIT-independent fibrosarcoma (HT1080), GIST (GIST54), and melanoma (SKMEL2) cell lines showed no significant change in cell viability after LMTK3 silencing when compared to the NT siRNA (Figure 1D).

To further determine the specificity of the effects of LMTK3 silencing on KIT-mutant cells, we created a stable GIST430 (ex 11) cell line expressing a c-myc epitope-tagged LMTK3 by lentiviral transduction (GIST430-LMTK3myc). This construct contained the coding DNA sequence (CDS) of LMTK3 but lacked 5’ and 3’ untranslated regions (UTRs). Experiments were then performed in these, as well as control GIST430 (ex 11) cells using siRNAs targeting the LMTK3 CDS (siLMTK3_CDS), which knocks down both endogenous and exogenous versions, or the LMTK3 3’UTR (siLMTK3_3’UTR), which only knocks down the endogenous version. LMTK3 knockdown with either the CDS-targeting or 3’UTR-targeting siRNAs significantly decreased cell viability in GIST430 (ex 11) cells, which only express endogenous LMTK3 (Figure 1E). However, only siRNA targeting the LMTK3 CDS, but not the 3’UTR, decreased cell viability in the GIST430-LMTK3myc cells (Figure 1F), suggesting LMTK3myc is sufficient to maintain cell viability and the impact of LMTK3 silencing is due to on-target effects on endogenous LMTK3.

Silencing LMTK3 reduces proliferation in vitro and in vivo in KIT-dependent cells

To understand the role of LMTK3 on the proliferation of KIT-dependent cells, we measured total cell number over time after LMTK3 silencing. The proliferation of GIST430 (exon 11, Figure 2A), GIST-T1, and MaMel cells in vitro (Supplemental Figure 3) was significantly impaired by 96 hours post-transfection with LMTK3 siRNA. To understand the role of LMTK3 on the in vivo growth of KIT-mutant GIST cells, we injected siRNA-treated GIST430 (ex 11) cells subcutaneously into NOD.Cg-Rag1 Il2rg (NRG) mice. GIST430 (ex 11) cells were transfected with NT or LMTK3 siRNA 24 hours prior to injection. LMTK3 or NT siRNA-treated cells were implanted separately into the right or left flank, respectively. Non-targeted tumors were palpable within 3 weeks and animals were euthanized 6 weeks post-implantation. Non-targeting tumors grew at a rapid rate after becoming palpable, reaching an average volume of 1 cm3 at the time of euthanasia. In contrast, tumors in which LMTK3 had been knocked down were not palpable until at least 4 weeks. These tumors grew much slower and were significantly smaller at the time of sacrifice (0.1 cm3 average volume, p <0.005, Figure 2B). Taken together these data suggest that GIST cells in which LMTK3 has been transiently silenced have reduced cell viability and proliferation in both in vitro and in vivo experiments.

Figure 2:

LMTK3 silencing inhibits proliferation of KIT-dependent cells in vitro and in vivo.

A. Total cell number over time of GIST430 (exon 11) after LMTK3 silencing, (N=3). B. Subcutaneous tumor volume after implantation of 1 × 106 GIST430 (ex11) cells treated with the indicated siRNAs into each flank of an NRG mouse, (N=8). The p values for t tests between NT and LMTK3 siRNA on each day are indicated by asterisks: **, p<0.005.

LMTK3 silencing induces cell death in KIT-dependent cells

To clarify the mechanism by which LMTK3 silencing decreased viability in KIT-dependent cells, we investigated the role of apoptosis. We measured the activity of caspases 3 and 7 in KIT-dependent cell lines 96 hours post-siRNA transfection. We observed a significant increase in caspase 3/7 activity after LMTK3 silencing (Figure 3A), as well as a concordant increase in the cleavage of PARP, a downstream target of caspases, by immunoblotting (Figure 3B). Further, a greater percentage KIT-dependent cells were permeable to propidium iodide (PI) after exposure to LMTK3 siRNA (Supplemental Figure 4), indicating they were undergoing cell death.

Figure 3:

LMTK3 silencing induces apoptosis.

A. Activity of caspases 3 and 7 96 hours post-transfection with NT or LMTK3 siRNA in KIT-mutant cells. (N=5) B. Immunoblot showing cleavage (lower arrowhead, 90kDa) of full-length PARP (upper arrowhead, 110kDa), 72 hours post-siRNA transfection. C. Activity of caspases 3 and 7 96 hours post-transfection with NT, LMTK3 CDS or 3’UTR siRNA in GIST430 (ex11) or GIST430-LMTK3myc cells. (N=3) The p values of t test for each cell line are indicated by asterisks: **, p<0.005; ***, p<0.001; ****, p<0.0001.

Apoptotic markers were also measured in GIST430 (ex11) vs. GIST430-LMTK3myc cells treated with LMTK3 CDS siRNA or LMTK3 3’UTR siRNA to determine the specificity of LMTK3 knockdown for this phenotype. We observed that LMTK3myc expression was sufficient to prevent the induction of caspase activity seen when endogenous LMTK3 is knocked down, further indicating that LMTK3 silencing specifically induces apoptosis to reduce viability in mutant KIT-dependent cells (Figure 3C).

LMTK3 positively regulates KIT protein expression

Since KIT-mutant cells specifically depend on KIT expression and activity for proliferation and survival, as evidenced by sensitivity to KIT knockdown, we hypothesized that LMTK3 silencing was affecting KIT. To test this possibility, we examined KIT protein after LMTK3 siRNA transfection by immunoblotting. LMTK3 silencing significantly reduced the total amount of KIT protein in the imatinib-sensitive GIST and melanoma cell lines used in our screen (Figure 4A, C) and in imatinib-resistant KIT-mutant GIST cells (Figure 4B, C). The reduction in total KIT protein upon LMTK3 silencing was reflected in a proportional reduction in KIT phosphorylation (Y721, Y703, and pan- phospho-tyrosine, Figure 4A-B, D, Supplemental Figure 5). Loss of KIT phosphorylation translated downstream to a reduction in the activity of pro-survival and proliferative signaling effectors, AKT (phospho-S473) and ERK1/2 (phospho-T202/pY204) in KIT-dependent cells (Figure 4A-B).

Figure 4:

Silencing of LMTK3 in KIT-mutant GIST and melanoma cells reduces KIT protein expression.

Immunoblotting of imatinib-sensitive GIST and melanoma cell lines (A) or imatinib-resistant GIST cell lines (B) 72 hrs post-transfection with NT or LMTK3 siRNA. C-D. Quantification of phospho-KIT (Y721) and total KIT protein from immunoblots, normalized to β-tubulin. (N=3) E. Immunoblot and quantification of GIST430-LMTK3myc stable cells 72 hrs post-transfection with NT, LMTK3 CDS, or LMTK3 3’UTR siRNA, (N=4). Bars show average protein relative to NT siRNA. The p values of t tests for each cell line compared to NT indicated by asterisks: **, p<0.005; ***, p<0.001; ****, p<0.0001.

Because loss of KIT can result in cell death in mutant KIT-dependent cells, we tested the ability of exogenous LMTK3myc to affect this phenotype by targeting GIST430-LMTK3myc cells with LMTK3 CDS siRNA or LMTK3 3’UTR siRNA. Maintenance of LMTK3 expression partially restored the loss of KIT protein in GIST430-LMTK3myc cells (Figure 4E). These results are in agreement with our above data showing the restoration of cell viability under the same conditions.

Upon observing the effect of LMTK3 knockdown on KIT abundance, we investigated how KIT expression was affected by LMTK3 overexpression. We measured KIT in GIST430-LMTK3myc clones with variable levels LMTK3myc expression. We found that LMTK3myc protein abundance was highly correlated with KIT protein abundance (Supplemental Figures 6, R2= 0.9459). Collectively, these data suggest that LMTK3 positively regulates KIT protein expression to support proliferative signaling and viability in mutant KIT-dependent cells, regardless of driver KIT mutation.

LMTK3 regulates KIT translation in KIT-dependent GIST cells

There are multiple points of KIT regulation at which LMTK3 may play a role to control total KIT protein abundance, including transcription, translation, or protein stability. Because LMTK3 has been previously implicated to have roles in controlling transcription of oncogenes and cancer promoting genes, , , , we began by investigating whether LMTK3 is involved in regulating KIT transcription. We measured KIT mRNA abundance by qRT-PCR 72 hours post-transfection with NT or LMTK3 siRNA in GIST and melanoma cells. We found that LMTK3 siRNA did not significantly change KIT transcript abundance, despite a significant decrease in LMTK3 mRNA (Figure 5A, Supplemental Figure 7A, C).

Figure 5:

LMTK3 regulates KIT translation in KIT-dependent GIST.

A. LMTK3 and KIT transcript abundance relative to NT siRNA at 72 hours post-transfection with LMTK3 3’UTR siRNA in GIST430 (ex 11). B. KIT protein abundance after inhibition of translation with cycloheximide in GIST430 (ex 11) 48 hrs post-siRNA transfection. Protein half-life calculated by one-phase decay. C. Gel showing immunoprecipitated S35-KIT and quantification relative to NT (N=4). GIST430 (ex 11) cells labeled 48 hrs post- siRNA transfection. The p values of one-way ANOVA indicated by asterisks: *, p<0.05; ***, p<0.001.

LMTK3 has also been implicated in regulating protein stability of ERα in breast cancer models. To determine if LMTK3 may be playing a similar role in KIT-mutant GIST or melanoma cells, we calculated the half-life of KIT protein. We transfected cells with NT or LMTK3 siRNA then treated with the translation inhibitor cycloheximide (CHX) at least 48 hours post-transfection. We measured total KIT protein in whole cell lysates over a time course. Knockdown of LMTK3 did not shorten KIT protein half-life in GIST430 (ex 11) cell lines (Figure 5B, Supplemental Figure 7E), nor was KIT protein half-life significantly different in MaMel and GIST-T1 cell lines transfected with NT vs. LMTK3 siRNA (Supplemental Figure 7B, D, F-G). These results suggest that LMTK3 has a role that is independent of controlling KIT transcription or KIT protein stability in KIT-mutant cells.

We next investigated a role for LMTK3 in KIT translational control. At 48 hours post- siRNA transfection, we labeled nascent proteins in GIST430 (ex 11) cells with radioactive (S35) methionine and isolated KIT protein by immunoprecipitation and SDS-PAGE to quantify S35-KIT. We observed a significant reduction in the abundance of S35-labeled KIT with two individual LMTK3 targeting siRNAs (CDS and 3’UTR) compared to NT control (Figure 5C). These data provide evidence that LMTK3 plays an important role in regulating translation of KIT in KIT-dependent GIST cells to promote proliferative signaling that maintains these cancer cells.

DISCUSSION

Despite the therapeutic success of KIT TKIs for treating KIT-mutant cancers, drug resistance presents a significant clinical barrier. With this in mind, we used a siRNA screen to identify novel therapeutic candidates in KIT-mutant solid tumor cells. In this report, we describe a novel target, LMTK3, which was identified using three individual mutant KIT-dependent cell lines, two GIST and one melanoma. We found that silencing LMTK3 specifically killed KIT-dependent cells bearing various KIT mutations, including those that confer imatinib-resistance, and severely slowed GIST growth in vivo. We show that LMTK3 supports the viability of KIT-dependent GIST and melanoma cells by maintaining KIT expression.

The positive gene-specific regulation of KIT translation in KIT-dependent cells represents a novel function of LMTK3. LMTK3 has been shown to promote the expression of other oncogenes, including ERα in breast cancer, but via regulation of transcription and protein stability. LMTK3 has not yet been directly shown to play roles in translation regulation in breast cancer models, but LMTK3 silencing was shown to alter the abundance of many proteins in breast cancer, thus, LMTK3 may have translational regulation in this cancer as well, but this remains to be investigated.

Our data suggest that LMTK3 plays a role as a KIT gene-specific translational regulator in KIT-mutant GIST and melanoma cells, not as a regulator of global translation. Upon LMTK3 knockdown in mutant KIT-dependent cells we do not see global reduction in protein levels; for example, ERK1 and ERK2 protein remain unchanged. Moreover, LMTK3 silencing is not toxic to all cells, as would be the case if LMTK3 controlled total protein synthesis. KIT-dependent cells are uniquely sensitive to loss of LMTK3, supporting that LMTK3 can affect translation of KIT specifically. Prior to this study, it was not known that mutant KIT was regulated by gene-specific translation, although, WT KIT-specific translation has been reported in hematopoietic cells. Gene-specific translational regulation is not uncommon, especially in cancer cells. Oncogenes can be specifically translated in a cap-dependent manner, often bypassing conventional translation regulation. It is not yet clear if LMTK3 controls the translation of mutant KIT in this way, or if LMTK3 controls WT KIT in other cell types. This will be important to understand for assessing the therapeutic utility of LMTK3 as a target because some normal cells, such as hematopoietic stem cells depend on WT KIT. Interestingly, LMTK3 seems to have a specific role in KIT-mutant GIST and melanoma that is distinct from other KIT-mutant disease; this RAPID screen has been run on many cell lines and primary leukemia samples, including those with activating KIT mutations. LMTK3 silencing does not affect the cell viability of the human KIT-mutant mastocytosis cell line, HMC-1.1, nor do most KIT-mutant primary leukemia, suggesting it regulates KIT specifically in solid KIT-mutant tumor cells.

Most significantly, both KIT TKI-sensitive and -resistant cell lines showed decreased viability and decreased KIT expression after LMTK3 silencing. Secondary KIT mutations, T670I, V654A, or D820A, confer resistance to imatinib and have variable sensitivity to second and third line KIT TKI (sunitinib and regorafenib) in GIST, . Moreover, in advanced GIST, these mutations often co-occur to create complex intra- and inter-lesional heterogeneity, which cannot be controlled by any single clinically available KIT TKI. Resistant KIT-mutant GIST cells still depend on KIT for proliferation and survival, therefore KIT remains an important target. Silencing LMTK3 can affect KIT T670I, V654A, or D820A, because LMTK3 regulates KIT at an early step of KIT expression, synthesis, suggesting it could be useful as a treatment strategy in advanced patients with various resistance mutations.

Collectively, these data presented here support the potential of LMTK3 as a therapeutic target for mutant KIT-driven GIST and melanoma. However, strategies for targeting LMTK3 have not been developed. Moreover, it is not yet clear what specific function of LMTK3 is responsible for regulating KIT translation in KIT-mutant cells. LMTK3 has indeed been shown to have serine/threonine kinase activity, but, it has also been shown to play roles in normal cells that are independent from its kinase activity, instead acting as a protein scaffold to bring together multi-protein regulatory complexes. Either kinase activity or protein complex scaffolding could play important roles in translation but targeting enzymatic activity of kinases has had the most success in the clinic historically. Future studies will be necessary to determine the crucial function of LMTK3 in order to design therapeutic strategies targeting LMTK3 in KIT-mutant solid tumors.

MATERIALS AND METHODS

Cell lines:

A summary of the KIT-mutant cell lines used in this study are shown in Supplemental Table 2. KIT-mutant cell lines were authenticated by KIT sequencing and KIT TKI sensitivity experiments. Upon receipt, GIST and MaMel cell lines were tested for mycoplasma contamination by Venor™ GeM Mycoplasma Detection Kit (Sigma, St. Louis, MO) and were found to be negative. HEK293 (ATCC® CRL-1573™), HT1080 (ATCC® CCL-121™) and SKMEL2 (ATCC® HTB-68™) were purchased from ATCC (Manassas, VA) and experiments were performed within 6 months of purchase. The stable LMTK3myc expressing GIST430 (ex11) cell line was created by viral transduction and selection in G418 (800ng/mL, Sigma). Cell lines were maintained in culture at 37°C with 5% CO2 for the minimum necessary time to perform experiments, generally not exceeding one month from thaw.

RAPID siRNA screen and siRNA transfection:

Human tyrosine kinase siRNA screen (RAPID) was performed as previously described, , for each cell line (N=3). Cells were plated in 96-well plates and transfected with siRNA using oligofectamine (Invitrogen, Carlsbad, CA) following manufacturer’s instructions. After 96-hour incubation at 37°C, cell viability was assayed using MTS reagent. Data were corrected for row or column plating bias and triplicate experiments were averaged. The median effect and standard deviation of the data from each screen experiment were calculated as described previously. Genes exceeding one standard deviation from the median of each screen were considered for comparison between the three KIT-mutant cell lines to identify common hits. For other siRNA experiments (N>3), siGENOME siRNA pools were purchased from Dharmacon: LMTK3: M-005338-03-0005, KIT: M-003150-02-0005, NT: D-001206-13-05, PLK1: M-003290-01-0005, custom LMTK3 3’UTR: 5’CAGAAGAGGGGUUGAGAAUUU-3’.

Cell viability and caspase activity assays:

Cells were plated, transfected, and incubated in opaque 96-well plates (Corning, Corning, NY). Cell viability was measured using the Cell Titer Glo reagent (Promega, Madison, WI) according to manufacturer’s instructions. Activity of caspases 3 and 7 was measured using Caspase 3/7 Glo (Promega). Both assays were measured on the GloMax luminometer (Promega). (N>3)

Flow Cytometry:

Total cell number and percent of cells with inclusion of PI were determined by flow cytometry on the GUAVA EasyCyte5 (Millipore, Burlington, MA) by staining with ViaCount reagent (Millipore) (N=3).

GIST xenograft:

One million viable GIST430 (ex 11) cells suspended in 50% matrigel (Corning) were implanted subcutaneously in the flank of four week old male NOD.Cg-Rag1 Il2rg (NRG) mice (Jackson Laboratories, Bar Harbor, ME) 24 hours post-transfection with non-targeting or LMTK3-targeting siRNA using oligofectamine reagent in vitro, as described above. Each mouse bore one tumor with each siRNA on opposing flanks (N=8). Once a palpable mass was detected, tumors were measured using calipers every three days until sacrifice. Animals were humanely euthanized before total tumor burden reached 2 cm3, as dictated by our IACUC protocol. Neither randomization nor blinding were necessary for these experiments. Sample size was determined by performing power calculations based on preliminary experiments.

Quantitative RT-PCR:

Total RNA was extracted (RNeasy, Qiagen, Hilden, Germany), cDNA was synthesized using 1ug of total RNA (MultiScribe RT, Applied Biosystems, Foster City, CA), and quantitative RT-PCR (qRT-PCR) was performed on a LightCycler 480 (Roche, Basel, Switzerland) using Probes Master Mix (Roche). FAM Taqman primers for LMTK3 (Hs01090726_g1) and KIT (Hs00174029_m1) were purchased from Life Technologies (Carlsbad, CA). Custom primers and hydrolysis probe (IDT, San Jose, CA) were used to detect a 66-bp GAPDH amplicon: GAPDH forward CACTAGGCGCTCACTGTTCT, GAPDH reverse GCGAACTCCCCGTTG, GAPDH probe 5′TexRd-XN/TGGGGAAGGTGAAGGTCGGA/3′IAbRQSp.

Protein harvest and Immunoblotting:

Cells were scraped from flasks for lysis with lysis buffer (50mM HEPES, 150mM NaCl, 1mM EDTA, 1.5mM MgCl2, 1% Triton X-100, 10% Glycerol) with 1X protease and phosphatase inhibitor cocktail (Cell signaling, Danvers, MA). Immunoblotting was performed by standard SDS-PAGE protocol using the Criterion electrophoresis system (Biorad, Hercules, CA), Transblot Turbo transfer system (Biorad), and imaged and quantitated using the Chemidoc imaging system (Biorad).

Antibodies:

Antibodies used for immunoblotting are shown in Supplemental Table 3. Total KIT antibody (0.5ug, Abcam) bound to Protein A/G PLUS beads (sc-2003, Santa Cruz, Dallas, TX) was used for KIT immunoprecipitations.

Reagents:

Imatinib mesylate (STI571, Novartis, Basel, Switzerland), cycloheximide (Sigma), and G418 (Sigma) were dissolved in sterile water or PBS. The expression plasmid pCMV6-LMTK3myc-DDK (cat # RC223140) was obtained from Origene (Rockville, MD) and used for construction of pLENTI-LMTK3myc-DDK (using pLENTI-C-myc-DDK-IRES-Neo, PS100081).

Cycloheximide time course:

Cells were treated with cycloheximide (20ug/mL, Sigma), for the indicated times 48–72 hours post-siRNA transfection. KIT protein was quantified from whole cell lysates using PathScan Total c-KIT ELISA kit (Cell Signaling, #7197C), following manufacturer’s instructions. Experiments were performed in triplicate and KIT half-life was estimated by one-phase decay.

(S35) Methionine Labeling:

At 48 hrs post-siRNA transfection, GIST430 (ex 11) cells were starved in DMEM lacking methionine (and cysteine, Invitrogen Cat# 21013), supplemented with 5% dialyzed FBS and 2mM glutamine (N=4). Cells were then labeled with 100μCi/mL S35-methionine (EXPRE35S35S Protein Labeling Mix, PerkinElmer, Waltham, MA) for 30 minutes, washed twice with PBS, and lysed in RIPA lysis buffer. The supernatant was collected and immunoprecipitation was performed with KIT antibody (Abcam), or control rabbit IgG (Santa Cruz). Immunoprecipitated proteins were resolved by SDS-PAGE. Newly synthesized S35-KIT was visualized and quantitated on a Bio-Rad FX Molecular Imager using Quantity One Software (Bio-Rad).

Statistics:

All quantitative experimental data represent mean± SEM (unless otherwise indicated), performed with a minimum of three independent biological replicates. Statistical tests performed as appropriate, including one-way ANOVA (with multiple comparisons), unpaired, two-tailed t tests, or one phase-decay (to calculate protein half-life), using PRISM (Graph Pad, La Jolla, CA) software. P values indicated by asterisks: *, p<0.05; **, p<0.005; ***, p<0.001; ****, p<0.0001.