The Selective Tyrosine Kinase Inhibitor STI571 Inhibits Small Cell Lung Cancer Growth¹

Protein tyrosine kinases play a critical role in intracellular signal transduction pathways leading to diverse cellular responses,such as proliferation, apoptosis, and differentiation (1, 2). Consequently, these enzymes have become primary targets for the development of novel therapeutics designed to block cancer cell proliferation, metastasis, angiogenesis, and promote apoptosis (3, 4). The strategy that has progressed farthest in clinical development is the use of monoclonal antibodies to target growth factor receptor tyrosine kinases (5). The use of small molecule tyrosine kinase inhibitors, however, could have significant theoretical advantages over monoclonal antibodies. Small molecule inhibitors could have better tissue penetration, could have activity against intracellular targets, and could be designed to have oral bioavailability. Their interaction with the target enzyme can be more easily studied through the use of X-ray crystallography and computer modeling, and their structure could be more easily engineered to give the drug more specificity and more favorable pharmacokinetic properties. Several lead compounds have shown promising activity against such targets as the EGFR³(6, 7) and the vascular endothelial cell growth factor receptor (8).

The compound STI571 (formerly known as CGP 57148B) is a 2-phenylaminopyrimide derivative that was optimized for selective inhibition of the Bcr-Abl tyrosine kinase (9, 10), which is a fusion protein that results from the pathogenic 9:22 translocation that occurs in 95% of CML (11). The drug showed a cellular IC₅₀ for inhibition of Bcr-Abl kinase activity of ∼0.25 μm and inhibited growth of cell lines engineered to be dependent on Bcr-Abl kinase activity at 1μ m (10, 12). In addition, STI571 also selectively inhibited the growth of continuous cell lines derived from CML patients, as well as primary CML progenitors in clonogenic assays,at low micromolar concentrations (10, 13, 14). Recently,it has been demonstrated that the drug, given at pharmacokinetically appropriate doses, can completely block the outgrowth of CML cells in nude mice xenografts and cause complete regression of the majority of established tumors with oral dosing without detectable systemic toxicity (15). Although STI571 appears to be a potentially exciting new treatment for CML patients not curable with allogeneic bone marrow transplantation (16), it could also be potentially useful for the treatment of other malignancies. This is attributable to the fact that although STI571 is highly selective for Bcr-Abl and c-Abl versus most other tyrosine kinases, it does inhibit the PDGFR with similar efficacy to Bcr-Abl (12). Preliminary evidence also suggests that the drug can also inhibit the Kit receptor tyrosine kinase in vitro.⁴This cross-inhibition is likely secondary to a conserved structure of the ATP-binding pocket, the site of drug interaction, among these kinases. Thus, we hypothesized that STI571 might also be effective against malignancies in which Kit activation plays an important role.

The Kit growth factor receptor tyrosine kinase is in the same subclass as the receptors for platelet-derived growth factor and colony-stimulating factor-1 (17). Kit’s ligand is SCF(alternatively named mast cell growth factor or steelfactor), a hemopoietic growth factor that, in conjunction with other hemopoietic growth factors, supports the proliferation and differentiation of multiple hemopoietic cell lineages from early precursors (18). In addition to hemopoietic cells, Kit is also normally expressed in melanocytes and germ cells, as well as in a variety of hematological malignancies and solid tumors, including SCLC. In fact, >70% of SCLC cell lines and tumor specimens coexpress Kit and its ligand, SCF (19, 20, 21), leading to the hypothesis that this coexpression constitutes an autocrine growth loop. Evidence supporting a SCF/Kit autocrine loop includes the observation that growth and chemotaxis of selected SCLC cell lines is stimulated by exogenous SCF (22, 23). In addition, we demonstrated that reconstitution of SCF and Kit coexpression in a SCLC cell line led to enhanced growth factor independence, and inhibition of Kit activation by transfection of a dominant-negative c-kit gene resulted in a loss of growth factor independence (23, 24). Furthermore, the quinoxaline tyrphostin AG 1296, an inhibitor of Kit and the PDGFR, completely blocked SCF-dependent growth of the H526 cell line and promoted apoptosis within 48 h (23). Treatment of a representative panel of SCLC cell lines grown in FCS with an optimal concentration of AG 1296 led to growth inhibition averaging ∼50% in five of six responding cell lines. The use of AG 1296, however, is limited by is poor solubility in aqueous solution. The purpose of the present study was to determine whether the more soluble and p.o. bioavailable STI571 would inhibit SCF-dependent Kit activation and growth of SCLC cell lines.

SCLC cell lines were grown in RPMI 1640 medium supplemented with 2 mm l-glutamine, with (complete medium) or without 10% fetal bovine serum (Life Technologies, Gaithersburg, MD);when grown in the absence of serum, 0.1% BSA (Sigma, St. Louis, MO)was added to the medium. Where indicated, serum-free medium was supplemented with recombinant SCF (Intergen, Purchase, NY) or IGF-1(R&D, Minneapolis, MN) at the indicated concentrations. The H146, H209,H432, H510, and H526 cell lines have been previously characterized by Carney et al. (25). The WBA cell line was grown from the involved bone marrow aspirate of an untreated patient, and it coexpresses SCF and Kit, as well as N-myc (20). Cells were stimulated with SCF and IGF-1 after preincubation in serum-free medium overnight. Cell growth was measured using the MTT(Sigma) colorimetric dye reduction method, an assay shown to correlate very well with viable SCLC cell number under the conditions used (26). Duplicate plates containing eight replicate wells per assay condition were seeded at a density of 1 ×10⁴ cells in 0.1 ml of medium, and data were expressed as the change in absorbance at 540 nmover 72 h, relative to initial values obtained 3 h after plating. STI571 was solubilized in DMSO; final concentration of DMSO in all cultures, including controls, was 0.1%. Stock solutions of carboplatinum (kindly provided by Bristol Myers Squib, Princeton, NJ)and etoposide (Calbiochem, San Diego, CA) were also made up at 1000-fold the desired final concentration in H₂O and DMSO, respectively.

DNA fragmentation was analyzed using the TUNEL technique (27), which was performed with the In Situ Cell Death Detection Kit (Boehringer-Mannheim, Indianapolis, IN) using the protocol supplied by the manufacturer. Fluorescence was quantitated using a Becton Dickinson FACScan flow cytometer with the FACScan research software package; nuclear labeling was confirmed by fluorescence microscopy. Caspase-3 activity was determined using the CaspACE assay system (Promega, Madison, WI). Briefly, after a 48-h incubation under the indicated conditions, cell pellets were resuspended in cell lysis buffer at a concentration of 10⁸ cells/ml. To achieve cell lysis, cells were frozen at −80°C overnight, thawed, and incubated on ice for 15 min. Lysates were cleared by centrifugation at 15,000 × gfor 20 min in the cold, and 500 μg of supernatant protein were used for the colorimetric assay exactly as described in the manufacturer’s protocol. Tritiated thymidine incorporation was determined after plating 2 × 10⁴ cells in eight replicate wells of a microplate dish containing 0.1 ml of medium to which 1 μCi of tritiated thymidine (DuPont NEN, Boston, MA) had been added. Cells were then transferred to glass fiber filter plates using an automated cell harvester (Packard Instruments, Meridian CT), extensively washed with H₂O, and counted using a liquid scintillation counter.

H526 cells were lysed in a buffer containing 50 mm HEPES(pH 7.5), 150 mm NaCl, 1% NP40, 1.5 mmMgCl₂, 1 mm EGTA, 10% glycerol, 0.2 mm NaVO₄, 100 μg/ml phenylmethylsulfonyl fluoride, 1 μg/ml aprotinin, and 10 μg/ml leupeptin using a Dounce homogenizer with a tight-fitting pestle;protein concentrations were determined by bicinchoninic acid assay(Pierce, Rockford, IL). The lysate, containing 1–1.5 mg of protein,was centrifuged for 10 min at 10,000 × g to obtain a soluble postnuclear supernatant. IP was initiated by the addition of 10μg of monoclonal anti-Kit antibody (K45; NeoMarkers, Fremont, CA),followed by incubation for 2 h at 4°C and by an additional 2 h in the presence of Protein A+G agarose. The IP was washed four times in lysis buffer and then once in PBS. The pellet was aspirated to dryness, and 30 μl of kinase buffer (20 mm1,4-piperazinediethanesulfonic acid, 10 mmMnCl₂) containing 0.1 μm[γ-P³²]ATP (3000 Ci/mmol; DuPont NEN) was added. The kinase reaction was carried out for 10 min at room temperature and terminated by the addition of an equal volume of 2×SDS sample loading buffer; the reaction was then resolved on a 10%polyacrylamide gel and analyzed by autoradiography. Western blotting was performed using standard procedures, with detection using the enhanced chemiluminescence system (Amersham, Arlington Heights,IL). Staining was accomplished using the following antibodies:anti-Kit, 3D6 monoclonal (Boehringer-Mannheim); antiphosphotyrosine,PY20 monoclonal (Transduction Labs, Lexington, KY); antiactive MAP kinase polyclonal (Promega); anti-phospho-Akt Thr308, Ser473, and pan-Akt polyclonals (New England BioLabs, Beverly, MA).

As is the case for other receptors in its class, ligand binding induces receptor dimerization, which in turn leads to autophosphorylation and an enhancement of the specific activity of the Kit kinase domain (18). The H526 SCLC cell line expresses high levels of Kit, relatively low levels of SCF, and can grow in serum-free medium containing recombinant SCF (as the sole growth factor), which rapidly induces tyrosine phosphorylation and an increase in the specific enzymatic activity of Kit (23). To determine the effect of STI571 on receptor activation, H526 cells made quiescent by overnight incubation in serum-free medium were pretreated with increasing concentrations of drug or DMSO vehicle for 30 min followed by the addition of SCF. Five min after the addition of SCF, cells were divided into two equal portions, and the degree of Kit activation was determined by assessing the extent of in vivoautophosphorylation by immunoblotting in one portion and in the other portion by directly measuring kinase activity in an IP-kinase assay. As illustrated in Fig. 1, both the kinase assay and the assessment of in vivotyrosine phosphorylation revealed an IC₅₀ of 0.2 and 0.1 μm, respectively, in good agreement with previous studies on the highly related PDGFR, as well as the originally intended target, Bcr-Abl (9, 12). The residual kinase activity seen at high drug concentrations may be attributable to a small degree of binding reversibility during the IP and kinase reaction, both performed in the absence of additional drug.

To determine the effect of STI571 on SCF-mediated growth, H526 cells were incubated in serum-free medium containing SCF, and increasing concentrations of drug and growth were measured at the end of 3 days using the MTT dye reduction assay. IGF-1-mediated growth was also assayed in the same fashion to determine the specificity of the drug. As illustrated in Fig. 2, SCF-mediated growth was inhibited in a dose-dependent fashion with an IC₅₀ of ∼0.3 μm, with complete inhibition obtained at 1 μm. The average IC₅₀ for three independent determinations was 0.8 ± 0.6 μm. Growth inhibition by STI571 clearly paralleled its inhibition of Kit kinase activity (Fig. 1). IGF-1-mediated growth was only modestly inhibited at 1μ m, but significant growth inhibition was seen at 10μ m. At this concentration, however, a large proportion of cells in the SCF-treated cultures underwent cell death, presumably by apoptosis as previously demonstrated for cells treated with AG 1296 (23).

The above experiments demonstrate that STI571 selectively blocked growth of a SCLC cell line dependent on SCF as the only exogenous growth factor. However, to be clinically relevant, growth inhibition should extend to growth in the presence of a complex mix of growth factors, such as found in serum. To determine whether the drug could inhibit growth under these circumstances, a panel of six representative cell lines was incubated in medium containing 10% FCS, and increasing concentrations of STI571 and growth were measured over 72 h by MTT assay. All cell lines used coexpress SCF and Kit to some degree, with the exception of H146, which expresses high levels of SCF but not Kit (20). Three of the cell lines (H146, H432, and H510) were derived from patients previously treated with chemotherapy, and three were derived from previously untreated patients (20, 25). One cell line, H510, showed no response to concentrations of STI571 up to 10 μm (data not shown); this is consistent with the lack of response this cell line displayed to another Kit inhibitor, AG 1296 (23). The other five cell lines showed a dose-dependent decline in growth as illustrated in Fig. 3. For these cell lines, growth inhibition at 10 μm STI571 was between 50 and 90%; the average IC₅₀ for the five responsive cell lines was ∼5 μm. Of note, H146,the only one of the six cell lines that does not express Kit, had an intermediate sensitivity to the drug, suggesting that, in addition to inhibiting Kit, the drug must block other targets necessary for SCLC growth. This point is further illustrated by the response of H526,which despite its sensitivity to low concentration of the drug when grown in SCF (Fig. 2), only showed 20–30% growth inhibition in serum-containing medium until a concentration of 10 μmwas reached. It is likely that at this concentration, Kit is not the only affected target, as suggested by the inhibition of IGF-1-mediated growth at 10 μm STI571 (Fig. 2).

Surprisingly, despite the decreased growth of cells treated with STI571 noted by MTT assay, no increase in the appearance of morphologically apoptotic cells over controls was observed in any of the treated cell lines, suggesting a cytostatic mechanism of action. This was opposed to the cell death observed when cells were treated while growing in medium containing SCF as the sole growth factor (Fig. 2). Because it is important to understand whether the drug mediates its effect through cytotoxic or cytostatic mechanisms, we directly assayed for the induction of apoptosis. First, we sought to confirm that the cell death we observed in serum-free medium containing SCF was attributable to apoptosis. H526 cells were exposed to STI571 or DMSO vehicle for 48 h in SCF medium, and apoptosis was assayed using a flow cytometric TUNEL assay. Fig. 4,A illustrates the marked increase in terminal deoxynucleotidyl transferase labeling of cells treated with STI571,indicative of DNA strand breakage and apoptosis. H526 cells exposed to STI571 in serum-containing medium were also analyzed using the flow cytometric TUNEL assay. No difference in the curves generated from vehicle and STI571-treated cells were noted. However, because of enhanced cell aggregation when grown in serum, the curves were complex and harder to interpret (data not shown). We therefore switched to an alternative assay that measured the activation of caspase-3, a critical mediator of apoptosis. As illustrated in Fig. 4 B, etoposide efficiently activated caspase-3 in all three cell lines tested but STI571 did not produce a statistically significant increase in caspase-3 in any of the cell lines tested. These observations are consistent with the lack of morphological evidence for apoptosis in STI571-treated cells grown in serum.

On the basis of the above evidence, it appeared that STI571 mediated its growth inhibition in serum-containing medium through cytostatic mechanisms. To confirm this impression, the thymidine incorporation of cells treated with 10 μm STI571 for 24 h in serum-containing medium was determined; Fig. 5 illustrates representative results for the WBA and H146 cell lines. Tritiated thymidine incorporation decreased by ∼50%, correlating well with results obtained in the MTT assay (Fig. 3) and therefore indicating that STI571 acts predominantly by slowing proliferation in the presence of serum.

To confirm that STI571 blocked downstream signal transduction from Kit and to develop some insight into the reasons for its cytotoxic effect on cells grown in SCF versus its cytostatic effect on cells grown in serum, we studied the effect of the drug on activation of MAP kinase and Akt, which lies downstream of phosphatidylinositol-3 kinase. Activation of these signal transduction proteins was studied by Western blotting using phospho-specific antibodies. As illustrated in Fig. 6, the addition of SCF to quiescent H526 cells resulted in a potent activation of both MAP kinase and Akt that, as expected, was completely inhibited by pretreatment with 10 μm STI571. IGF-1 produced a weak activation of MAP kinase and a potent activation of Akt, which were unaffected by the presence of STI571. The addition of 10% FCS to quiescent cells produced an intermediate activation of MAP kinase, which was unaffected by the presence of the drug; no FCS-induced activation of Akt was seen in three replicate experiments. These experiments suggest that persistent activation of MAP kinase could be partially responsible for the lack of apoptosis of STI571-treated H526 cells when grown in FCS.

It has been demonstrated that treatment of tumors expressing epidermal growth factor family receptors with blocking antibodies sensitizes the tumor cells to the effects of cis-platinum both in vitro and in vivo (28, 29, 30). This observation could indicate a specific property of epidermal growth factor family receptors, of the antibodies used, or a general phenomenon relating to the interaction between the chemotherapeutic agent and blockade of necessary receptor tyrosine kinases. Because platinum compounds are an integral part of present chemotherapeutic regimens for SCLC (31, 32), we decided to determine whether STI571 would sensitize SCLC cell lines to the cytotoxic effects of carboplatinum, which was chosen because of its better aqueous solubility compared to cis-platinum. Cells were incubated in the continuous presence of increasing concentrations of carboplatinum in the presence of 5 μm STI571. The 5-μm concentration was chosen because it was the average IC₅₀, as well as a potentially achievable serum concentration based on preliminary results of the ongoing Phase I trial (33). The degree of growth inhibition relative to control cells incubated in the presence of carboplatinum alone was determined after 72 h by MTT assay. Three cell lines were chosen for study based on their varying sensitivity to carboplatinum, with IC₅₀s ranging from <0.1μg/ml to 5 μg/ml, and their varying sensitivity to STI571. As illustrated in Fig. 7, STI571 failed to synergize with the cytotoxic effects of carboplatinum in any of these three cell lines; a fourth cell line (WBA) also showed no synergy (data not shown). On the contrary, in experiments using H526, there appeared to be a mild but reproducible protective effect in the presence of the kinase inhibitor, especially at high carboplatinum concentrations. In addition to carboplatinum, etoposide, the other integral component of SCLC chemotherapy regimens, was also tested in combination with STI571; no enhanced cytotoxicity over that produced by etoposide (1–10 μm) alone could be demonstrated for the combination using H526, WBA, or H432 cell lines(data not shown).

We have demonstrated that STI571 efficiently inhibits SCF-mediated Kit activation (in serum-free medium) at concentrations similar to those that inhibit both Bcr-Abl and the PDGFR in cellular assays, with an IC₅₀ in the 0.1–0.5 μm range (9, 12). This observation likely reflects the related nature of the ATP-binding sites of these three kinases because the drug is competitive for ATP binding (9). The inhibition of SCF-mediated SCLC growth closely parallels inhibition of Kit kinase activity, with complete inhibition of H526 growth occurring at 1μ m. The drug was clearly selective because the same degree of growth inhibition was not obtained in IGF-1-containing medium until a 10-fold higher concentration was attained.

The selectivity seen in serum-free medium supplemented with single growth factors was apparently reduced in serum-containing medium. The IC₅₀ for H146, which does not express Kit, was 5μ m, which was approximately the average for all of the sensitive cell lines combined. This observation suggests two conclusions. The first is that the contribution of Kit activation to growth of SCLC cell lines in medium containing 10% FCS is likely to be relatively limited because the IC₅₀ in serum is far higher than the drug concentration needed to completely block growth mediated by saturating concentrations of SCF. A potentially confounding factor affecting this conclusion could be binding or inactivation of the drug by some component of serum. We feel that this is not likely to be an important factor because the growth inhibition of H526 in serum relatively closely paralleled the growth inhibition seen in serum-free medium containing IGF-1, suggesting that drug inactivation by serum does not significantly contribute to the higher IC₅₀. Response to the drug was fairly consistent,despite distinct biological properties of the cell lines used (25) and despite variation in the sensitivity to carboplatinum (Fig. 7) and etoposide. It is worth pointing out,however, that there appears to be some correlation with the degree of endogenous SCF expression and the sensitivity to STI571. Both the H432 and H209 cell lines express high levels of both SCF and Kit (20) and are the most sensitive to drug in serum, whereas H526, which expresses very high levels of Kit and little SCF, is the least sensitive. However, expression levels of SCF cannot explain the growth inhibition of H146 because it expresses no Kit. Therefore, it is likely that there is an additional drug target that is partially responsible for SCLC growth inhibition by STI571, especially at drug concentrations that exceed 2.5 μm. The identity of this hypothetical target is not known at this time, but given the in vitro selectivity of this compound for a conserved ATP-binding pocket, it is possible that the additional target is a tyrosine kinase related to the PDGFR family or c-ABL. The fact that the relative sensitivity of a panel of cell lines to STI571 parallels their sensitivity to the quinoxaline tyrphostin AG1296 (23), an inhibitor of PDGFR and Kit but not Abl (34), suggests that unidentified drug target may be more likely related to the PDGFR family. In addition, the similar pattern of growth inhibition observed using two tyrosine kinase inhibitors having dissimilar chemical structures suggests that the inhibition is attributable to their ability to inhibit Kit and related kinases. One important question that arises is whether the existence of additional targets of STI571 affects its potential as a therapeutic agent for SCLC. Because the toxicity of the drug at concentrations up to 10 μm for nontransformed cells in culture was nonexistent (10, 12)or limited (9, 13) and limited animal studies produced no detectable toxicity (15), it is possible that the additional drug target is selectively activated in SCLC. The existence of an additional drug target may be an advantage because it provides the potential for inhibition of the 25–30% of SCLCs, represented by the H146 cell line, which do not express Kit. If further in vivo toxicity studies confirm little toxicity at concentrations that inhibit SCLC growth, there could be an exploitable therapeutic index even if the effect of STI571 cannot entirely be attributable to inhibition of Kit.

SCLC is generally thought of as a highly chemotherapy-responsive disease, with a variety of multidrug combinations producing response rates of >80%, with a third or more being complete responses (31, 32). Despite this high response rate, virtually all extensive stage patients and 75–80% of limited stage patients(treated with combined modality therapy) succumb to the disease within∼2 years. Increasing the dose of standard chemotherapeutic agents has produced increased toxicity without increasing response or survival (32, 35). Despite the relative paucity of long-term survivors, chemotherapy has made a major impact on the treatment of this disease, with present regimens improving stage-specific survival at least 3–4 fold, in addition to concomitant improvements in quality of life (31). Thus, novel therapeutic approaches that prolong overall survival with minimal toxicity, even if they do not increase the percentage of long-term survivors, could be of significant benefit in this disease. STI571 could fall into this category based on its ability to be p.o.-dosed and its limited toxicity for nontransformed cells seen in previous studies at concentrations of drug we have shown to significantly inhibit SCLC growth. We had hoped that the effects of STI571 would synergize with the effects of standard chemotherapeutic agents, such as carboplatinum based on previous studies demonstrating enhanced platinum sensitivity of EGFR family-expressing tumors treated with blocking monoclonal antibodies (5) or a small molecule antagonist (36). This turned out not to be the case, at least for a schedule using continuous coadministration of both drugs. The different effects of the receptor tyrosine kinase inhibitors on platinum sensitivity could be related to differences between the biological effects of receptor tyrosine kinase subclasses, between SCLC and the predominantly breast and non-SCLC cell lines used in the EGFR studies, or between the specific agents themselves.

STI571 appears to be cytostatic for SCLC cells grown in FCS based on its ability to inhibit growth without an increase in apoptosis. This is in contrast to the cytotoxicity secondary to apoptosis seen when SCF was the only exogenous growth factor provided (Fig. 2; Fig. 4 A). However, it is unlikely that FCS accurately mimics all relevant in vivo situations. For example, for SCLC to recur after a chemotherapy-induced complete remission, the tumor cells must repopulate from relatively few cells originally located in a mostly necrotic tumor mass or alternatively from micrometastases in sanctuary sites such as the central nervous system. In these locations it could be hypothesized that autocrine growth stimulation, such as provided by coexpression of SCF and Kit, could be relatively more important than in environments rich in growth factors. Therefore, inhibition of autocrine growth with agents such as STI571 could potentially block recurrence after a chemotherapy-induced remission; however, even if the time to a clinically symptomatic recurrence was delayed by slowing tumor growth,this could have significant impact on survival and quality of life. The ability to p.o. dose the drug, along with the minimal toxicity seen in preliminary studies, suggests that it could be useful in such a setting. Therefore, based on these hypotheses and the demonstration that STI571 significantly inhibits SCLC growth, we believe further studies of STI571 in nude mouse SCLC xenograft models are warranted. We also believe that it would be reasonable to include recurrent SCLC patients in expansions of present ongoing Phase I clinical trials of STI571 (33).