MET Kinase Inhibitor SGX523 Synergizes with Epidermal Growth Factor Receptor Inhibitor Erlotinib in a Hepatocyte Growth Factor–Dependent Fashion to Suppress Carcinoma Growth

MET receptor tyrosine kinase signaling, driven by its unique biological ligand hepatocyte growth factor (HGF), is a key signaling pathway that is not only essential for many normal developmental and homeostatic processes but is also responsible for the pathologic development and progression of many human cancers (http://www.vai.org/met/; refs. 1, 2). The complexity of the HGF-MET pathway arises from its ability to induce diverse biological activities, ranging from proliferation, motility, and invasion to survival and angiogenesis (1, 2), many of which are hallmark traits of cancer (3). Multiple mechanisms can lead to an aberrant activation of this pathway, including paracrine or autocrine signaling or activating mutations in the receptor, all contributing to malignancy (1). Amplification or overexpression of MET and/or HGF mediating paracrine or autocrine activation is observed in almost all types of human solid tumors, whereas MET amplification or mutation causing inappropriate activation has been identified in a few cancers of many different origins (1). Inappropriate MET signaling has been identified as one of the important pathways that cancer cells may use to bypass growth inhibition caused by drugs targeting other pathways such as epidermal growth factor receptor (EGFR; refs. 4–6).

Many successes have been achieved by targeting the receptor tyrosine kinases (RTK) with EGFR-, ErbB2-, or vascular endothelial growth factor (VEGF) receptor–targeting drugs to treat various human cancers (7). The compelling evidence for the importance of the HGF-MET pathway in many human cancers (1) suggests that it will also prove to be an effective target for cancer therapy. Various strategies have been used to interfere with the HGF-MET pathway (8), and a list of MET drugs having promising therapeutic potential has been developed (9–11). Among these are neutralizing antibodies against either HGF or MET, and small-molecule inhibitors that antagonize MET kinase activity (9–11). The development of drugs targeting HGF or MET demands relevant preclinical in vivo test systems in which tumorigenic activity due to MET can be properly assessed and drug efficacy evaluated in many cancer types.

Mouse HGF binds human MET with only low affinity and does not potently activate human MET signaling (12). Therefore, we previously generated a transgenic mouse model expressing human HGF on an immune-compromised severe combined immunodeficient (SCID) background (hHGF^tg-SCID). This background significantly enhances the growth of MET-expressing human tumors by providing a species-compatible ligand (13). Here, we show that the hHGF^tg-SCID mice can significantly promote subcutaneous xenograft growth of many MET-expressing carcinoma cell lines derived from a broad spectrum of organs including lung, breast, kidney, colon, stomach, and pancreas. Furthermore, we show that SGX523, an exquisitely selective, ATP-competitive small-molecule kinase inhibitor of MET (14), is able to synergize with erlotinib to inhibit cell proliferation in vitro and tumor xenograft growth in hHGF^tg-SCID mice. Our data show that the hHGF^tg-SCID mouse can serve as a unique and valuable animal model for preclinical testing of MET drugs against various human cancers and for exploring better therapeutic strategies through administration of MET drugs in combination with other cancer drugs.

The NCI-H596, NCI-H1373, Colo205, DLD1, HT-29, MKN-45, NCI-N87, and HCC1954 cell lines were cultured in RPMI Medium 1640 (Invitrogen) supplemented with 10% fetal bovine serum (FBS; HyClone). The SW480, MDA-MB-231, MDA-MB-435, and 786-0 cell lines were cultured in DMEM (Invitrogen) supplemented with 10% FBS. Capan-2 cells were cultured in McCoy's 5A (Invitrogen) supplemented with 10% FBS, and HPAF II cells were cultured in 50% DMEM plus 50% MEM (Invitrogen) supplemented with 10% FBS. The NCI-H596, NCI-H1373, Colo205, DLD-1, SW480, HCC1954, MDA-MB-435, NCI-87, Capan-2, and HPAF II cell lines were obtained from the American Type Culture Collection. The HT-29, MDA-MB-231, and 786-0 were from the NCI-60 cell lines. The MKN-45 was obtained from the Japanese Collection of Research Bioresources.

Whole-cell lysates were extracted with radioimmunoprecipitation assay buffer [20 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl, 1% NP40, 0.5% deoxycholate, 0.1% SDS, 1 mmol/L EDTA, 50 mmol/L NaF, and 1 mmol/L sodium orthovanadate] supplemented with proteinase inhibitor cocktail (Roche), and they were quantified using a DC Protein Assay kit (Bio-Rad). The lysates were mixed with an equal volume of 2× sample buffer (Sigma-Aldrich), denatured in a boiling water bath for 5 minutes, and loaded onto a Tris-glycine gel (Invitrogen). After electrophoresis, the proteins were transferred onto a polyvinylidene difluoride membrane. The primary antibodies used for the Western blot analyses included anti-Met, anti-EGFR, anti-ErbB2, anti-ErbB3, and anti-ErbB4 from Santa Cruz; anti-phospho-Met (Tyr1234/1235), anti–phospho-ErbB3, anti–extracellular signal-regulated kinase (ERK) 1/2, anti–phospho-ERK1/2, anti-Akt, and anti–phospho-Akt from Cell Signaling Technology; anti–phospho-EGFR (Tyr1068) from Invitrogen; and anti-β actin from Sigma. Following secondary antibody reaction, the proteins were detected with ECL Western blotting Detection Reagents (Amersham) by exposure on a Kodak film.

For this assay, 1 × 10⁴ cells were seeded in each well of a 96-well dish. After 48 hours of serum starvation, the cells were treated with or without inhibitors of MET and/or EGFR, accompanied with or without the stimulation of HGF and/or EGF, for 20 hours. The cells were then treated with 1 μCi of ³H-thymidine (GE Lifesciences) per well and incubated at 37°C for 4 hours. Following ³H-thymidine incorporation, the cells were washed twice with ice-cold DPBS, treated with 200 μL of ice-cold 5% TCA on ice for 20 minutes, washed once with 95% ethanol, air-dried, and lysed in 75 μL of lysis solution (0.02 N NaOH, 0.1% SDS) for 15 minutes. The cell lysates were then transferred to a scintillation reader plate (Wallac) with 200 μl of scintillation fluid (Perkin-Elmer). The incorporated ³H radioactivity was measured as cpm in the MicroBeta TriLux 1450 LSC & Luminescence Counter (Perkin-Elmer). The assays were performed in triplicate for each treatment.

Cells were seeded in six-well dishes at a density of 5 × 10⁵ cells per well. After 24 hours of serum starvation, the cells were stimulated with or without growth factors for 48 hours. The cells were then harvested and stained with propidium iodine for flow cytometry analysis. For drug treatments, the cells were treated with or without drug for 1 hour before stimulation with growth factor.

The breeding and genotyping of the hHGF^tg-SCID mice has been previously described (13). Female SCID mice aged 4 to 8 weeks were used for the experiments. Age-matched C3H-SCID littermates from the same mating cages as the hHGF^tg-SCID mice were used as experimental controls. For subcutaneous tumor formation, subconfluent cells were harvested and resuspended in serum-free DMEM or RPMI at concentrations of 5 to 20 × 10⁶ cells/mL. Each animal received 100 μL of cell suspension injected s.c. into the right dorsal area. Following cell injection, mice were monitored for tumor formation twice weekly, and the dimensions of xenograft tumors were measured with manual calipers. The tumor volume was calculated by the formula volume (mm³) = length × width × depth. All animal studies were conducted with protocols approved by the Institutional Animal Care and Use Committee of the Van Andel Research Institute.

To test the efficacy of SGX523 and/or erlotinib on xenograft tumor growth, 40 tumor-bearing mice for each mouse strain (either hHGF^tg-SCID or C3H-SCID) were randomized into four groups (10 mice/group) for the following treatments: group 1, vehicle; group 2, erlotinib (150 mg/kg, once daily; refs. 15, 16); group 3, SGX523 (60 mg/kg, once daily); and group 4, erlotinib (150 mg/kg, once daily) plus SGX523 (60 mg/kg, once daily) combination. The erlotinib was dissolved in 0.5% (w/v) Methocel A4M Premium (DOW Chemicals), and the SGX523 was dissolved in 0.5% (w/v) Methocel A4M Premium plus 0.05% Tween 80 in distilled water. Treatments were started when the average tumor size reached ∼100 mm³, and drugs were administered for 14 days through oral gavage. The tumor sizes were monitored and measured twice weekly.

Inappropriate expression of MET and/or its ligand HGF is widely observed in almost all types of human cancers (http://www.vai.org/met/; ref. 1). We have previously shown that the growth of MET-expressing human xenografts, such as SK-LMS-1 leiomyosarcoma cell tumors, is significantly accelerated in hHGF^tg-SCID mice (13), indicating that human HGF in the host mice can activate MET in the human tumor cell lines through a paracrine mechanism.

To extend our understanding of how paracrine HGF-MET activation contributes to tumor growth, we tested various MET-expressing human cancer cell lines derived from a broad spectrum of organ/tissue origins in the female hHGF^tg-SCID mouse model. These cell lines represented lung, colon, breast, kidney, gastric, and pancreatic cancers (Table 1). Varied levels of the MET receptor were detected in these carcinoma cell lines; the levels of EGFR/ErbB receptor tyrosine kinase family members, especially EGFR and ErbB3, varied as well (Fig. 1A). We chose female mice for our preclinical testing because the concentration of circulating human HGF in female hHGF^tg-SCID mice was four times higher than in their male littermates, which allowed better growth of MET-expressing tumors (Supplementary Figs. S1 and S2).

Despite the variability in MET protein expression in the carcinoma cell lines tested in Fig. 1, they all displayed a significant growth advantage when s.c. injected into hHGF^tg-SCID mice versus into control C3H-SCID animals (Fig. 1B). The NCI-H1373, Colo205, DLD1, HT-29, SW480, MDA-MB-231, MDA-MB-435, MKN-45, NCI-N87, 786-0, and Capan-2 lines showed a 1.5- to 2-fold xenograft tumor growth advantage in the human HGF host environment (Table 1). By comparison, almost complete growth dependence was observed with the human lung cancer cell line NCI-H596, the breast cancer line HCC1954, and the pancreatic cancer line HPAF II in hHGF^tg-SCID mice, suggesting that paracrine activation of the HGF-MET pathway is critical for the tumorigenicity of these lines and in a sense reveals MET addiction in this mouse model. This was striking, given that MET protein levels in the NCI-H596 and HPAF II cells were not particularly high relative to levels in other lines (Fig. 1A). Moreover, not all cancer cell lines expressing MET showed a growth advantage in hHGF^tg-SCID mice (Supplementary Fig. S3). The discordance between the level of MET protein and the growth advantage in hHGF^tg-SCID mice might be explained in part by high MET expression, which can drive ligand-independent activation in addition to the ligand-dependent paracrine activation.

The NCI-H596, NCI-H1373, HCC1954, and HPAF II cells all express a high level of EGFR in addition to MET (Fig. 1A). With the exception of NCI-H596, the lines also express a moderate to high level of ErbB3 (Fig. 1A). To assess the status of MET, EGFR, and their downstream signaling in these cells, we treated them with or without SGX523 and/or erlotinib before HGF or EGF stimulation, and then performed Western blot analyses.

SGX523 inhibited MET phosphorylation and the downstream ERK and AKT activation in a dose-dependent manner, and at 1 μmol/L almost completely eliminated the MET activation induced by HGF (Fig. 2). Although the inhibition of MET phosphorylation by SGX523 did not seem to synergize with the addition of erlotinib, the suppression of downstream ERK activation was clearly enhanced by the combination of these two drugs (SGX523 at 200 nmol/L to 1 μmol/L with 5 μmol/L erlotinib) in NCI-H1373, HCC1954, and HPAF II cells (Fig. 2B–D), but not in the NCI-H596 cells (Fig. 2A). In NCI-H1373 cells, which express high levels of MET and display constitutive MET phosphorylation, the combination of SGX523 and erlotinib seemed to more strongly inhibit tyrosine phosphorylation of ErbB3 and EGFR induced by EGF (Fig. 2B), suggesting possible cross-talk between the MET and EGFR-ErbB3 pathways. However, EGFR and/or ErbB3 phosphorylation was not affected by SGX523 in the HCC1954, HPAF II, and NCI-H596 cells, which express low to moderate levels of MET and show no ligand-independent MET phosphorylation (Fig. 2A, C, and D).

Proliferation of NCI-H1373, HCC1954, HPAF II, and Capan-2 cells was induced 24 hours after HGF stimulation as measured by a ³H-thymidine incorporation assay (Fig. 3A and B). However, NCI-H596 displayed enhanced ³H-thymidine incorporation only at 48 hours and not at 24 hours after HGF stimulation (data not shown; Fig. 3C), indicating a relatively slow cell doubling property. To determine if blocking MET activation impaired HGF-induced cell proliferation, we treated the cells with SGX523 together with HGF stimulation. SGX523 strongly suppressed the HGF-induced proliferation of NCI-H596 and Capan-2 cells (Fig. 3B and C), whereas only moderate inhibition was observed for the NCI-H1373, HCC1954, and HPAF II cells (Fig. 3A and B).

EGFR is strongly coexpressed with MET in the NCI-H596, NCI-H1373, HCC1954, HPAF II, and Capan-2 cells (see Fig. 1A). To determine if cross-talk between MET and EGFR influences proliferation, the cells were also treated with erlotinib alone or with a combination of SGX523 and erlotinib. Erlotinib alone, like SGX523, also significantly inhibited HGF-induced proliferation of these cells (Fig. 3A–C), indicating that such proliferation might be in part mediated by EGFR activation. However, the inhibition was much greater, and the proliferation was far below the basal level when the cells were treated with the combination of SGX523 and erlotinib in the presence of HGF (Fig. 3A–C). This might be explained in part by the fact that SGX523 and erlotinib combination almost completely eliminated the basal and HGF-induced ERK and/or AKT phosphorylation in the presence of HGF (see Fig. 2). In contrast, the proliferation of these cells in the absence of HGF (serum free or induced by EGF alone) was completely suppressed by erlotinib alone to the level seen when they were treated with the SGX523/erlotinib combination in the presence of HGF (Supplementary Fig. S4; Fig. 3A–C). The synergy of SGX523 and erlotinib in inhibiting cell proliferation in the presence of HGF stimulation implies cross-talk between the MET and EGFR pathways in these cells.

Both HGF and EGF can induce cell cycle progression in NCI-H596 cells, as manifested by increasing cell counts in the S phase (Fig. 3D). SGX523 strongly inhibited HGF-induced cell cycle progression by reducing the number of cells in the S phase from 32.72% to 9.53% (Fig. 3D). Erlotinib alone also provided significant inhibition of HGF-induced cell cycle progression from G₁ to S phase, but inhibition was maximized by the combination of SGX523 and erlotinib, as the percentage of the cells in the S phase was down to 5.49% (Fig. 3D). Interestingly, the EGF-induced cell cycle progression was inhibited only by erlotinib but not by SGX523. This result is consistent with the observation that erlotinib (but not SGX523) inhibited the ³H-thymidine–incorporating activity of NCI-H596 cells in the presence of EGF stimulation (Fig. 3C).

To determine the in vivo effect of SGX523 and/or erlotinib on tumor growth, we evaluated the efficacy of these two drugs (alone or in combination) on various human carcinoma xenografts raised in hHGF^tg-SCID mice and in control nontransgenic SCID mice. For established NCI-H596 lung cancer xenografts in the hHGF^tg-SCID mice, the individual drugs alone only provided partial growth inhibition, 19% by erlotinib and 30% by SGX523 (Table 2; Fig. 4A). However, the combination of SGX523 and erlotinib exhibited a synergistic effect and strongly suppressed NCI-H596 tumor xenograft growth by 75% (Table 2; Fig. 4A). Strikingly, SGX523 displayed a much stronger growth inhibition when treatment began immediately after the inoculation of NCI-H596 cells in mice: Tumor growth was decreased by 83% relative to controls (Fig. 4B).

The SGX523/erlotinib combination also provided greater inhibition of the growth of xenografts derived from HCC1954 breast cancer cells and HPAF II pancreatic cancer cells in hHGF^tg-SCID mice relative to single-agent treatment (Table 2; Fig. 4C and D). As summarized in Table 2, HCC1954 tumor growth was suppressed by 84% with the combination, compared with 69% by erlotinib (P = 0.0487) and 51% by SGX523 (P = 0.0041) alone. HPAF II tumor growth was down by 68% when treated with the combination, compared with 60% with erlotinib (P = 0.0265) and 32% with SGX523 (P = 0.00001) alone. However, in nontransgenic C3H-SCID mice, there was no significant difference in tumor growth inhibition between erlotinib alone and the combination of SGX523/erlotinib (P = 0.425 for HCC1954 and P = 0.492 for HPAF II; Table 2; Fig. 4C and D). The efficacy of the two drugs on NCI-H596 xenografts in the nontransgenic SCID mice was not evaluated due to poor tumorigenicity (Table 1).

Taken together, our data suggest that SGX523 administered in conjunction with erlotinib can provide a strong suppression of tumor growth of human carcinomas and that the hHGF^tg-SCID mouse model can serve as a valuable tool for preclinical drug evaluation of MET drugs alone or in combination. In addition, early treatment of tumors with MET drugs may provide better drug efficacy against tumor growth (Fig. 4B).

The MET receptor tyrosine kinase is the sole receptor recognized by HGF and, upon activation, promotes diverse biological activities such as cell growth, motility, and morphogenesis in different cell contexts (1). Important roles of the HGF-MET pathway in cancer have been unveiled by many studies that have included the identification of mutations in the MET gene, frequent observations of MET and/or HGF overexpression in many types of human cancers, and the in vitro and in vivo cancer-related functional activities of this ligand-receptor pair, as well as tumorigenic and metastatic studies in animal models (1, 2, 17). These studies have drawn a great deal of interest and effort into developing drugs that block the HGF-MET axis for targeted cancer intervention (8). Various MET drugs targeting this pathway have emerged, and several are currently in phase I/II clinical trials (9). Some are selective drugs specifically targeting the HGF-MET, such as Amgen's AMG102 anti-human HGF-neutralizing antibody (18), Genentech's MetMAb anti-MET monovalent (one armed) antibody (19), and ArQule's ARQ197 small-molecule kinase inhibitor of MET (9), whereas the others are multikinase inhibitors that simultaneously abolish the kinase activities of MET and other RTKs (9–11).

Cancer drug development includes several critical steps: identifying a potential therapeutic target, engineering a drugable reagent, preclinical evaluation in animal models, and, ultimately, clinical trials. Preclinical evaluation is critical not only for determining the efficacy and safety of a drug, but also in helping to identify the patient groups most likely to benefit from the drug, alone or in combination. The common preclinical approach is to test a drug against tumor xenografts raised in immune-compromised athymic nude or SCID mice. However, because mouse HGF has a very low affinity for and activity on the human MET receptor (12), these models do not mimic the full range of MET activation in human cancers. The hHGF^tg-SCID mouse, a genetically engineered strain that produces human HGF/SF (13), overcomes this limitation. The hHGF^tg-SCID mice strongly enhance the xenograft growth of MET-expressing tumors such as the SK-LMS-1 leiomyosarcoma model (13), which is a valuable system for evaluating MET drugs such as AMG102 (20).

Because aberrant MET activation is observed in a wide spectrum of human carcinomas and sarcomas (http://www.vai.org/met/), MET drugs could have therapeutic value in most types of human cancer. In this report, we explored various human carcinoma cells implanted in hHGF^tg-SCID mice as in vivo carcinoma xenograft systems suitable for human MET drug evaluation. We showed that the hHGF^tg-SCID mice significantly enhanced xenograft growth of many human MET-expressing carcinomas derived from lung, breast, gastric, pancreatic, colon, and kidney (Fig. 1B). These data provided straightforward in vivo evidence that paracrine activationof the HGF-MET pathway is critical for the tumorigenicity of many human carcinomas. Among the tested cancer lines, NCI-H596, HCC1954, and HPAF II displayed particularly strong HGF-dependent xenograft growth. The presence of human HGF in the SCID host environment dramatically shortened the times required for these cell lines to develop tumors, enhanced tumor takes, and/or greatly advanced the growth rate of their xenografts (Table 1).

Nonetheless, not all MET-expressing cancer cell lines displayed a significant growth advantage in the hHGF^tg-SCID mice (Supplementary Fig. S3), and HGF-dependent xenograft growth did not necessarily parallel the level of MET protein expressed. It remains to be determined if ligand-independent or autocrine MET activation is accountable for the uncoupling of MET protein levels detected in the cells and the HGF-dependent xenograft growth in the hHGF^tg-SCID mice. In addition, neither the scale of ERK and/or AKT activation downstream of MET (Fig. 2), nor the levels of in vitro cell proliferative activity induced by HGF (Fig. 3) predicted how cell growth might respond in the hHGF^tg-SCID mice (Fig. 1). These observations may not be surprising in light of the fact that the HGF-MET axis contributes to many key tumorigenic events including cell proliferation, motility, invasion, survival, and angiogenesis. Cell lines of different origins may depend to different extents on MET signaling for these processes and may vary in their requirements for the level of MET expression and HGF stimulation of the pathway.

It is noteworthy that the NCI-H596 cells displayed almost absolute dependence on the HGF-MET pathway for xenograft growth (Fig. 1B). NCI-H596 is a non–small cell lung cancer (NSCLC) line carrying an intronic mutation in the MET gene (19). The mutation causes an abnormal splicing of the exon 14 and results in a shorter form of the MET receptor lacking its juxtamembrane domain, which contains a Cbl-binding site (19). Consequently, the MET mutant can no longer be regulated by Cbl-mediated ubiquitination and degradation, causing a more sustained MET activation upon HGF stimulation of these cells (19). Contrary to this extreme growth dependence, established NCI-H596 xenograft growth was only partially inhibited by SGX523 (Fig. 4A), which suppressed MET phosphorylation and downstream ERK and AKT activation (Fig. 2A), and also impaired proliferation and cell cycle progression of NCI-H596 cells induced by HGF in vitro (Fig. 3C and D). However, SGX523 had a much stronger inhibitory activity on residual tumor growth when administered concurrent with inoculation (Fig. 4B), indicating that early treatment of tumors with MET drugs may achieve a better therapeutic outcome.

Similar to SGX523, another MET drug, MetMAb, also partially inhibited established NCI-H596 xenograft growth in hHGF^tg-SCID mice (21), suggesting that the cells might rely on other signaling pathways to evade HGF-MET pathway inhibition. This seems to be true because a much greater suppression of NCI-H596 xenograft growth in hHGF^tg-SCID mice was achieved by the combination of SGX523 with erlotinib (Table 2; Fig. 4A) or of MetMAb with erlotinib (21), relative to individual drug treatment alone. Consistent with the in vivo efficacy (Fig. 4A), the SGX523/erlotinib combination also showed synergy in vitro, inhibiting cell proliferation (Fig. 3C) and causing an enhanced impairment of NCI-H596 cell cycle progression from G₁ to S phase in the presence of HGF (Fig. 3D). These data indicated that simultaneous targeting of the MET and EGFR pathways can provide a synergistic inhibitory effect on the growth of NCI-H596 lung cancer. Strikingly, it was recently reported that a combination of erlotinib and the MET inhibitor ARQ197 significantly improved median progression-free survival by 66% in patients with advanced NSCLC cancer in a phase II clinical trial (22). This promising clinical data, in conjunction with our results, supports that the hHGF^tg-SCID xenograft model is a sound preclinical system for exploring optimal cancer therapeutic strategies involving MET drugs, and drug evaluation in this system can be used to help predict clinical therapeutic outcome.

Likewise, a combination of SGX523 and erlotinib may also provide stronger suppression of the growth of HCC1954 and HPAF II xenografts in hHGF^tg-SCID mice perhaps due to additive effects of the two drugs (Table 2; Fig. 4C and D). Neither displayed as robust combinatorial suppression as with NCI-H596 (Fig. 4A). More importantly, the enhanced suppression of tumor xenograft growth by the SGX523/erlotinib combination was observed only in hHGF^tg-SCID mice (Table 2), indicating that the hHGF^tg-SCID mouse model is an excellent in vivo system for preclinically evaluating not only MET drugs but also their combinations with other cancer drugs. This is further exemplified by the observation that a combination of MetMAb and anti-VEGF neutralizing antibody also displayed synergistic inhibition on NCI-H596 xenograft growth (23).

Taken together, we showed here that hHGF^tg-SCID mice can enhance the xenograft growth of many human carcinoma cell lines through paracrine MET expression, and we have provided evidence that these xenograft models can serve as valuable tools for preclinical evaluation of MET drugs alone or in combination with other cancer drugs. We are confident that the hHGF^tg-SCID mouse system can provide unique strength for facilitating the development of MET drugs against various human cancers.

Sean Buchanan is an employee of the Lilly Research Laboratories. The other authors disclosed no potential conflicts of interest.

We thank the Vivarium staff, Kyle Furge, Kellie Leali, YiMi Wu, Kay Koo, and Amy Nelson at Van Andel Research Institute (VARI) for their help; David Nadziejka for editing the manuscript; Carrie Graveel at VARI and Kenneth Iwata at OSI Pharmaceuticals for critical reading of the manuscript; SGX and Lilly Pharmaceuticals for providing SGX523; OSI Pharmaceuticals, Inc. for providing the erlotinib; and Van Andel Foundation for funding.

Grant Support: Van Andel Foundation.

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.