Development of New Insulin-like Growth Factor-1 Receptor Kinase Inhibitors Using Catechol Mimics*

Because insulin-like growth factor-1 (IGF-1) and its receptor play a pivotal role in many cancers, it is an attractive target for the design of inhibitors. Here we present a new family of bioisostere inhibitors, based on the structure of AG 538. AG 538 is a substrate-competitive inhibitor of the IGF-1 receptor (IGF-1R), with an IC50 = 61 nM in a cell-free kinase assay (Blum, G., Gazit, A., and Levitzki, A. (2000) Biochemistry 39, 15705–15712). AG 538 is a low molecular weight compound containing two catechol rings, which are sensitive to oxidation in cells. We have therefore prepared and examined catechol bioisosteres of AG 538. These AG 538 bioisosteres possess similar biological properties to AG 538; they inhibit IGF-1R by a substrate-competitive mechanism and are non-competitive vis à vis ATP. They inhibit IGF-1R kinase activity in the sub-micromolar concentration range in cell-free assays. IGF-1 induced IGF-1R autophosphorylation; IRS-1 phosphorylation and protein kinase B activation are inhibited at a low micromolar concentration range when applied to intact cells. These inhibitors also block the formation of colonies in soft agar by prostate and breast cancer cells. The ability to replace catechol groups with a moiety that is more stable in cells may aid in developing non-catechol-containing substrate-competitive inhibitors targeted toward IGF-1R and possibly against other protein-tyrosine kinases.

receptor activation and phosphorylation of downstream substrates (1).
The IGF-1R is essential for normal growth, development, and differentiation and mediates signals for the suppression of apoptosis, enhancement of mitogenesis, and anchorage-independent growth (2). The main mechanism by which IGF-1R protects cells from apoptosis is via the activation of the phosphatidylinositol 3-kinase (PI3K)/Akt/PKB pathway. An additional pathway was found in which IGF-1R protects cells from apoptosis by interacting with 14.3.3 proteins, causing the translocation of c-Raf to the mitochondria (3). In addition to these pathways, the IGF-1R promotes mitogenesis by activating the Erk/mitogen-activated protein kinase pathway (4).
Malignant transformation is often associated with increased expression and/or constitutive activation of the IGF-1R (5). For example, the highly metastatic H-59 Lewis lung carcinoma cells express high IGF-1R levels, and this IGF-1R overexpression was found to be critical for the ability of these cells to form metastases in mice livers. H-59 cells expressing antisense RNA of the IGF-1R became non-invasive and failed to form metastases in the mice livers (6). Also an IGF-1R-blocking antibody inhibits the growth of human breast cancer cells (7,8) and the formation of colonies in soft agar of these cells (9). These antibodies also inhibit the growth of Wilms tumor cells in culture and in nude mice (10).
Expression of dominant negative IGF-1R or expression of antisense RNA directed against the IGF-1R mRNA in various cancerous cell lines caused inhibition of the transformed phenotype as detected by the inhibition of colony formation in soft agar or the inhibition of tumor formation in nude mice. The cell lines inhibited include human cervical cancer (11), human prostate cancer (12), human gliomas cells (13), Rat-1 fibroblasts (14), and human rhabdomyosarcoma cells (15). Enhanced IGF-1R signaling has also been implicated in the development and progression of prostate cancer (16).
For all these reasons, we have been trying to generate IGF-1R kinase inhibitors as potential anti-neoplastic agents (17,18). We reported previously (18) on a substrate-competitive inhibitor of the IGF-1R, AG 538. AG 538 inhibits the IGF-1R with an IC 50 of 61 nM in a cell-free kinase assay, IGF-1 receptor autophosphorylation, as well as the activation of the downstream targets PKB and Erk2 in intact cells.
The catechol moiety present in AG 538 is unstable due to its vulnerability to oxidation (19). We therefore sought to generate substrate-competitive inhibitors in which the catechol moiety is replaced by bioisosteres (20). Here we report on the synthesis of a family of AG 538 analogs in which the catechol group was replaced with a benzoxazolone group on either side of the molecule. The benzoxazolone group can in principle function as a bioisostere of the catechol moiety, maintaining the bioactivity of the compound (21). The successful replacement of the catechol moieties in AG 538 with benzoxazolone groups are likely to be leads for the development of new class of IGF-1R kinase substrate-competitive inhibitors.
All solvents for HPLC use were from J. T. Baker Inc.; reagents for chemical synthesis were from Frutarom Ltd.
R ϩ and R Ϫ cells were a generous gift from Dr. R. Baserga. R Ϫ cells are mouse embryo fibroblasts, which are 3T3-like cells devoid of IGF-IR (23). R ϩ cells were generated by stable transfection of IGF-1R to R Ϫ cells, and these cells express ϳ1 ϫ 10 6 copies of IGF-1R (23). Cells were cultured in DMEM supplemented with 10% FCS, 100 units/ml penicillin, 100 g/ml streptomycin, 50 g/ml G418, and 50 g/ml hygromycin B in a humidified atmosphere of 94% air and 6% CO 2 at 37°C.
MDA MB-468 is a breast cancer cell line, a generous gift from Prof. Axel Ullrich from the Max Planck Institute (Martinsreid, Germany). These cells were cultured in DMEM supplemented with 10% FCS, 100 units/ml penicillin, 100 g/ml streptomycin in a humidified atmosphere of 94% air and 6% CO 2 at 37°C. PC-3, obtained from the ATCC, and LNCaP, obtained from CaPCURE Israel, both are prostate adenocarcinoma cell lines. MCF-7 is a mammary carcinoma cell line obtained from the ATCC. PC-3 cells were cultured in DMEM; LNCaP cells were cultured in RPMI 1640 with 5 g/ml insulin and 1 nM testosterone. MCF-7 cells were cultured in RPMI 1640. Media of all cell lines were supplemented with 10% FCS, 100 units/ml penicillin, 100 g/ml streptomycin in a humidified atmosphere of 94% air and 6% CO 2 at 37°C.
Inhibition of IGF-1R or IR-catalyzed Substrate Phosphorylation-The general protein-tyrosine kinase substrate, poly(Glu,Tyr) 4:1 (pGT), was coated onto a 96-well Maxisorp plates (Nunc) by adding 125 l of 0.1 mg/ml pGT in PBS to each well. Plates were sealed and incubated for 16 h at 37°C, washed once with TBST (10 mM Tris-HCl, pH 7.5, 50 mM NaCl, and 0.1% Triton X-100), dried for 2-3 h, and stored at 4°C. Semi-purified IGF-1R from R ϩ or IR from WTIR cells were prepared as described (18). The receptor was incubated (10 ng/well) in 20 M ATP, 10 mM MgCl 2 , 5 mM MnAc 2 , and 20 mM Tris-HCl, pH 7.4, with or without inhibitors, for 20 min at 30°C. The plate was then washed with TBS with 0.2% Tween 20 (TBST) and blocked with 5% low fat milk (1%). Rabbit polyclonal anti-phosphotyrosine serum (1:3000) was added to the plate, with incubation for 45 min at room temperature. The plate was washed repeatedly with TBST and then treated with anti-rabbit peroxidase conjugate antibody for 30 min. Detection was carried out with a color reagent, 2,2Ј-azido-bis-3-ethylbenzithiazoline-6-sulfonic acid, in citrate-phosphate buffer, pH 4.0, with 0.004% H 2 O 2 for 10 min and monitored at 405 nm, all at room temperature. IC 50 values of inhibitors were calculated using the REGRESSION program.
The assay was optimized with respect to the amount of IGF-1R or IR, reaction time, and ATP concentration. The signal was linear for 30 min in the range of IGF-1R/IR protein concentrations up to 35 ng/well (18).
Inhibition of IGF-1R Substrate Phosphorylation by Radioactive Methods-200 ng/sample of semi-purified IGF-1R from R ϩ cells was SCHEME 2 SCHEME 1 added to a solution containing inhibitor and pGT at various concentrations. The reaction was initiated by the addition of reaction buffer (30 mM HEPES, 12 mM MgCl 2 , 0.04 mM NaVO 3 , 5 mM Mn(Ac) 2 , 125 M ATP, and 1.5 Ci/sample [␥-32 P]ATP, final concentration) at 30°C for 10 min. The reaction was stopped by the addition of EDTA, pH 8, 0.1 M final concentration. Reaction samples were absorbed onto 3MM Whatman paper squares. The paper squares were then washed in 10% trichloroacetic acid, 1% sodium pyrophosphate at room temperature and dried in ethanol. Cerenkov radiation was measured by a 1600CA TRI-CARB Packard liquid scintillation counter. The data were analyzed by a Lineweaver-Burk plot using the Microsoft Excel program (18).
Inhibition of Autophosphorylation of IGF-1R and IGF-1 Signaling in Intact Cells-Tyrosine autophosphorylation of the ␤-subunit of IGF-1R and the IGF-1R signaling was assayed as described (18). Briefly, subconfluent NWTc43 or PC-3 or MDA MB-468 cells in 6-well polypropylene plates were incubated for 17 h with inhibitor at various concentrations in DMEM supplemented with 10% FCS, 100 units/ml penicillin, 100 g/ml, 0.1% Me 2 SO, and 0.1% ethanol (this final concentration of Me 2 SO/ethanol was kept constant in all samples), in a humidified atmosphere of 94% air and 6% CO 2 at 37°C. Cells were then starved by incubation at 37°C for 1 h in DMEM, containing inhibitors at the same concentration as before, in 0.1% Me 2 SO and 0.1% ethanol. Cells were then stimulated with 50 ng/ml IGF-1 for 5 min. After IGF-1 treatment, cells were washed twice with ice-cold PBS and lysed by the addition of boiling sample buffer (10% glycerol, 50 mM Tris-HCl, pH 6.8, 3% SDS, and 5% ␤-mercaptoethanol). Lysates were boiled for 15 min, and clarified by a 10-s 18,000 ϫ g centrifugation at room temperature. Equal amounts of protein per lane were separated by 8% SDS-PAGE and transferred to a nitrocellulose membrane (Sartorius AG). Phosphorylated proteins were immunoblotted with monoclonal anti-phosphotyrosine, anti-phospho-IRS-1, anti-phospho-Akt (Thr-308), and anti-phospho-Erk antibodies. Detection was performed with horseradish peroxidase-conjugated secondary antibody using the ECL system. Blots were then stripped of antibodies by incubating in 2% SDS, 10 mM ␤-mercaptoethanol, 62.5 mM Tris-HCl, pH 6.8, at 55°C for 20 min, washed with TBS with 0.2% Tween 20 (TBST), and then blocked with TBST with 5% low fat milk and re-probed. Blots that had been probed with anti-phosphotyrosine were re-probed with anti-IGF-1R␤. Other blots were re-probed with antibodies against the corresponding nonphosphorylated protein anti-IRS-1, anti-Akt1/2, and anti-Erk2 antibodies (e.g. the blots with anti-phospho-IRS-1 was re-probed with anti-IRS-1). Band intensities were quantified using the NIH image program. To determine IC 50 values of inhibition of PKB activation, the PKB levels in these samples divided the amount of phosphorylated PKB. This value was normalized to the maximal activity (designated as 100%) detected in the samples treated with IGF-1 without inhibitor. IC 50 values were calculated using the REGRESSION program.
Activation of PKB Phosphorylation by PDGF in Intact Cells-The activation of PKB by PDGF was performed as described above for activation by IGF-1, with the following changes. Sub-confluent NIH-3T3 cells in 6-well polypropylene plates were incubated for 20 h with inhibitor at various concentrations in DMEM supplemented with 10% FCS, 100 units/ml penicillin, 100 g/ml, 0.1% Me 2 SO (This final concentration of Me 2 SO was kept constant in all samples), in a humidified atmosphere of 94% air and 6% CO 2 at 37°C. Cells were then starved by incubation at 37°C for 1 h in DMEM, containing inhibitors at the same concentration as before, in 0.1% Me 2 SO. Cells were then stimulated with 50 ng/ml PDGF for 5 min. After PDGF treatment, cells were washed twice with ice-cold PBS and lysed by the addition of boiling sample buffer. Lysates were boiled for 15 min and clarified by short centrifugation at room temperature. 50 g of protein per lane were separated by 8% SDS-PAGE and transferred to a nitrocellulose membrane.
Phosphorylated PKB was immunoblotted with anti-phospho-Akt (Thr-308). Detection was performed with horseradish peroxidase-conjugated secondary antibody using an ECL system. Blots were then stripped of antibodies, washed with TBST, blocked with TBST with 5% milk, and re-probed with anti-Akt1/2 and detected as above.
Anchorage-dependent Growth Inhibition-For growth curves, R ϩ cells (450 cells/well), MDA MB-468 cells (2000 cells/well), and MCF-7 cells (2000 cells/well) were plated on 96-well plates in the appropriate media supplemented with 10% FCS, 100 units/ml penicillin, 100 g/ml, in a humidified atmosphere of 94% air and 6% CO 2 at 37°C. One day after seeding, cultures were treated with compounds 4 or 10 or AG 538 at various concentrations in growth medium. The final concentration of 0.1% Me 2 SO was kept constant in all samples. Triplicate samples were utilized for each concentration. Medium with and without inhibitors was replaced every day. Cell growth was determined by the micro culture methylene blue assay (25). Cells were fixed in glutaraldehyde, 0.05% final concentration, for 10 min at room temperature. After washing, the microplates were stained with 1% methylene blue in 0.1 M borate buffer, pH 8.5, for 60 min at room temperature. The plates were then washed extensively and rigorously (8 times) to remove excess dye and dried. The dye taken up by cells was eluted in 0.1 M HCl for 60 min at 37°C, and absorbance was monitored at 620 nm.
Anchorage-independent Cell Growth Assay-Colony formation in soft agar was performed mainly as described previously (26). A suspension of separated MDA MB-468, PC-3, LNCaP, or MCF-7 cells was plated in agar at a density of 5000 cells/well in a 96-well plate in growth medium containing 0.3% agar (50 l/well), on top of a layer of growth medium containing 1% agar (100 l/well). One day later, 50 l of growth medium supplemented with inhibitors at 4 times the final desired concentration was added on top of each cell line. The final concentration of 0.1% Me 2 SO was kept constant in all samples. 7-12 days after plating, cells were stained with MTT, photographed at 100 times magnification, and counted. The assays were performed twice in triplicate.
Chemistry-The aldehyde benzoxazolones were prepared from the acetal protected nitro hydroxy aldehydes. The nitro group was reduced and benzoxazolone formed with phenyl chloroformate instead of urea (Scheme 1) (27). In addition to the higher yields and less drastic conditions, this has the advantage of yielding the isomeric benzoxazolone aldehydes 3 and 9. Friedel-Crafts acetylation of benzoxazolone yields only the 6-acetyl isomer (28). The bromoacetyl benzoxazolone 13 was prepared by this route. The isomeric bromoacetyl 19 required the preparation of the 3-nitro-4-hydroxyacetophenone 16, which led to the 5-acetylbenzoxazolone isomer (Scheme 2). An attempt to obtain the bromoacetyl benzoxazolone 13 with N-bromosuccinimide or Br 2 failed. Only ring bromination on the benzoxazolone occurred. The bromoacetyl benzoxazolone 13 was prepared by Friedel-Crafts reaction of benzoxazolone with bromoacetyl bromide. We failed to obtain compound 14 by condensation of cyanoacetic acid to benzoxazolone with PPA.
The yields in the reactions of the bromoacetyl 13 and 19 with KCN were low or failed (11 and 0%, respectively). We therefore tried to improve the yield with LiCN (29) and NaCN. The higher yields with LiCN (19 and 22% with 13 and 19) do not justify the preparation of LiCN and the anhydrous conditions, and the commercial NaCN gave compound 20 at 29% yield, so we chose this latter route.
Reaction on a larger scale for 6 h gave 86% yield and for 12 h gave a 75% yield. For compound 8A, compound 7 (1.47 g, 6.7 mM) was hydrogenated with 10% Pd/C in ethanol for 6 h. Filtration and evaporation gave 0.82 g, light yellow solid: mp 125°C; 55% yield; NMR (CDCl 3 ) For compound 8B, 100 mg of Ra-Ni suspension was added to compound 7 (1.5 g, 6 mM) plus 1 ml of hydrazine hydrate in 30 ml of ethanol and 10 ml of water. The reaction was refluxed 40 min, decanted, and worked up to give 0.75 g, 58% yield, white solid, identical (TLC, NMR) to compound 8A.
For compound 9, to compound 8A (0.67 g, 3 mM), NaHCO 3 (0.3 g) in 25 ml of water and 25 ml of ethanol and phenyl chloroformate (0.5 g, 3.2 mM) was added. After 20 min NaOH (0.15 g) in 20 ml of water was added. After 0.5 h HCl was added until acidic pH and the reaction worked up to give after trituration from 284 mg of hexane: mp 163°C; For compound 14, compound 13 (2 g, 7.87 mM) and KCN (1.02 g, 15.7 mM) were dissolved in 10% water/ethanol (400 ml). The mixture was stirred and heated to 45°C for 1.5 h; 200 ml of water were then added, and the mixture was titrated to pH 6 with HCl. The mixture was stirred 0.5 h, then brought to pH 7 with KOH. The ethanol was evaporated, and the product was extracted with 3ϫ 200 ml of ethyl acetate and evaporated, and chromatography was performed on silica gel (35-70 mesh 1. Kinetics of inhibition of the IGF-1R by compound 4. pGT was phosphorylated by semi-purified IGF-1R, and IGF-1R activity was determined by the radioactive assay described under "Experimental Procedures." IGF-1R was assayed at concentrations of pGT ranging from 0 to 24 g/ml and compound 4 ranging from 0 to 6 M. The data were analyzed by the Lineweaver-Burk plot using the Excel program. For compound 16, 4-hydroxyacetophenone (27 g, 0.2 M) was added to 100 ml of glacial acetic acid and 25 ml of 70% nitric acid, with ice cooling and stirring for 6 h. The reaction mixture was kept overnight at 4°C while crystallization started. 100 ml of water was added, and the mixture was cooled in crushed ice for 0.5 h, and the precipitate was filtered, washed, and dried. The precipitate was recrystallized from ethanol to give 19.6 g, light yellow solid: mp 118°C; 54% yield; NMR (Me 2 SO-d 6  For compound 18, to compound 7 (7.6 g, 50 mM), NaHCO 3 (4.2 g) and phenyl chloroformate (7.8 g) in 150 ml of water and 350 ml of ethanol were stirred at room temperature for 35 min. NaOH (3.75 g) in 80 ml of water was added, and the reaction was stirred an additional 30 min. HCl was added slowly until pH 4, and the reaction was stirred 10 min and brought back to pH 7 with 1 M NaOH. The product was extracted 3ϫ with 150 ml of ethyl acetate and then evaporated to dryness.   3. Inhibition of IGF-1 induced PKB activation. A-C, IC 50 values were calculated using data obtained by a similar procedure to that described in Fig. 2C (i-iii). IC 50 curves and values are shown on the right. D, activation of PKB with PDGF in NIH-3T3 cells, in the presence of inhibitors as indicated, was performed as described in Fig. 2. For compound 20, compound 19 (1.024 g, 4 mM) and NaCN (0.98 g, 20 mM) in 100 ml of ethanol and 20 ml of water were heated to 55°C for 25 min. Ethanol was evaporated; 150 ml of water was added, and the mixture was brought to pH 7 with 1 M HCl (about 5 ml). The product was extracted 3ϫ with 100 ml of ethyl acetate, washed with water, and purified by HPLC using preparative RP18 column. The product was eluted by 25% acetonitrile in water and lyophilized to give 235 mg of white powder: mp 230°C; 29% yield; NMR (acetone-d 6  For compound 27, compound 26 (0.71 g, 2.9 mM) and NaCN (0.71 g, 14.5 mM) in 100 ml of ethanol and 20 ml of water were heated to 55°C for 30 min. Ethanol was evaporated, and 100 ml water of was added, and the mixture was brought to pH 7 with 1 M HCl (about 10 ml). The product was extracted 3ϫ with 100 ml of ethyl acetate, washed with water, and purified by HPLC by using preparative RP18 column. The product was eluted by 28% acetonitrile in water and lyophilized to give 112 mg, white powder: mp 150°C; 20.

RESULTS
Inhibition of IGF-1R and of Other Kinases in Cell-free Assays-AG 538 bioisosteres and other analogs were prepared by the Knoevenagel condensation of an aldehyde with a derivative of cyanoacetophenone. The products were purified by HPLC using a preparative RP18 column, then lyophilized, and ana- The efficacy of the compounds as kinase inhibitors was examined by enzyme-linked immunosorbent assays. Phosphorylation of pGT by the IGF-1R, IR, or Src was performed in the presence or absence of inhibitor. Tyrosine-phosphorylated pGT was detected by binding to anti-phosphotyrosine antibody and then horseradish peroxidase-secondary antibody followed by a color reaction as described under "Experimental Procedures" (Src enzyme-linked immunosorbent assay is described in Ref. 32). PKB inhibition was examined by a radioactive assay, in which the RPRTSSF peptide was phosphorylated by PKB in the presence or absence of inhibitor, as described (33).
The IC 50 values of these compounds were compared with the IC 50 of AG 538 as shown in Table I. Substitution of the catechol ring in compounds 23 and 29 with a methoxy catechol group reduced the potency of the inhibitor 25-and 55-fold, respectively, toward IGF-1R. Acetylating all 4 hydroxyls in compound 25 totally destroyed potency, whereas replacing one of the catechol rings with a benzoxazolone ring (compounds 4 and 10) reduced the inhibition only by 6 -7-fold. We found minor differences between the two isomers 4 and 10. In compound 21, substitution of the catechol ring with a benzoxazolone on the other side of the molecule reduced the inhibition by 10-fold as compared with AG 538. Replacing both catechol rings with a benzoxazolone reduced markedly the activity of the compounds (5, 6, 11, and 12). Compounds 4 and 10 inhibited other kinases than IGF-1R, with reduced efficacy, similarly to AG 538.
Mechanism of Inhibition of IGF-1R by Compounds 10 and 4-The mechanism of inhibition of IGF-1R was tested for the two most active molecules: 10 and 4. Table II shows that the IC 50 for inhibition of IGF-1R substrate phosphorylation by compound 10 is independent of ATP over a wide concentration range, suggesting that compound 10 is non-competitive vis à vis ATP. The use of the radioactive assay (described under "Experimental Procedures") allowed us to perform kinetic analysis of IGF-1R inhibition by compounds 4 and 10. Analyzing the data with a Lineweaver-Burk plot demonstrated that compound 4 inhibited the IGF-1R in a substrate-competitive manner (Fig. 1). Compound 10 was also found to inhibit the IGF-1R in the same manner (data not shown).
Inhibition of IGF-1R Autophosphorylation and Downstream Signaling in Intact Cells-To assay inhibition of IGF-1R autophosphorylation and of the downstream elements, IRS-1, PKB, and Erk-2 in intact cells, AG 538, compounds 4 and 10 were incubated with NWTc43 cells for 17 h as described under "Experimental Procedures." Gels were blotted and probed with anti-phosphotyrosine 4G10 antibody ( Fig. 2A), anti-phospho-IRS-1 antibody (Fig. 2B), anti-phospho-Akt antibody (Fig. 2C), and anti-phospho-Erk antibody (Fig. 2D). IGF-1R autophosphorylation and activation of IRS-1, Erk, and PKB/Akt were all inhibited by AG 538, compounds 10 and 4, in a dose-dependent manner over a similar concentration range (Fig. 2). Inhibition of IGF-1R downstream signaling, i.e. the phosphorylation of IRS-1, PKB, and Erk-2, was also detected after a short incubation of 40 min with the inhibitors with a similar dose dependence (data not shown).
Inhibition of IGF-1-Induced PKB Activation in Intact Cells-AG 538 and compounds 4 and 10 caused pronounced inhibition of IGF-1 induced PKB/Akt phosphorylation. We determined the IC 50 values of these inhibitors after overnight incubation of NWTC43 cells with the inhibitors, as described under "Experimental Procedures" (Fig. 3). Gels were blotted and probed with anti-phospho-Akt antibody, then stripped, and re-probed with anti-AKT1/2. Compounds 4 and 10 inhibited PKB activation more efficiently, at lower concentrations than AG 538. The IC 50 of compound 4 is 12 M, that of compound 10 is 6 M, and that of AG 538 is 19 M (Fig. 3).
PKB was activated by PDGF as described under "Experimental Procedures." The effect of the inhibitors AG 538 and compounds 4 and 10 on PDGF-induced activation was assayed  Growth Inhibition of R ϩ , MDA MB-468, and MCF-7 Cells-MDA MB-468 and MCF-7 are breast cancer-derived cell lines. R ϩ cells (described above) and MDA MB-468 and MCF-7 cells were sparsely seeded, and 1 day after seeding the cells were incubated with AG 538 or compounds 10 or 4 at various concentrations in growth media for 72 h. The medium was replaced every 24 h, and fresh inhibitor was added to the appropriate samples. Fig. 4 shows the growth inhibition of R ϩ cells by compound 10. Compounds 10 and 4 inhibited the growth of R ϩ , MDA MB-468, and MCF-7 cells, at a similar concentration range as AG 538, as shown in Table III.
Inhibition of Colony Formation in Soft Agar-MDA MB-468, PC-3, LNCaP, and MCF-7 cells were seeded in agar; 1 day later inhibitors were added at various concentrations. After 7-12 days, colonies were stained with MTT, counted, and photographed. Fig. 5 shows that treatment with 25 and 30 M inhibitors blocked colony formation in soft agar almost completely in MDA MB-468 and PC-3 cells. Table IV summarizes the IC 50 values for the inhibition of colony formation by the inhibitors on various cell lines.
Inhibition of IGF-1R Signaling in Breast Cancer and Prostate Cancer Cell Lines-The IGF-1R blockers AG 538 and compounds 4 and 10 inhibit colony formation in soft agar of two breast and two prostate cancer cells. The inhibitors were examined for their ability to inhibit the IGF-1R signaling in MDA MB-468 (breast cancer cells) and PC-3 (prostate cell lines). The IGF-1-induced tyrosine phosphorylation of IRS-1 was brought to a lesser degree than maximal activity at 10 M inhibitor, and at 50 M the extent of IRS-1 was below basal as shown in Fig.  6. The signal of PKB was below detection levels. DISCUSSION The catechol moiety mimics the phenol group of tyrosine, and therefore many tyrosine kinase inhibitors from the tyrphostin family contain a catechol moiety. It is well known that the catechol moiety is oxidized fairly rapidly, rendering catechol compounds unstable (19). In order to improve the stability of catechol compounds, bioisosteres were developed (compounds 4 -6, 10 -12, 15, 21, and 28).
Attempts to alter the catechol moieties of AG 538 caused reduction in potency. Substitution of the catechol ring in compounds 23 and 29 with a methoxy group reduced the potency of the inhibitor 25-and 55-fold, respectively, and acetylation of the hydroxyl group (compound 25) totally abolished inhibition (Table I). In this report we describe the synthesis of eight new AG 538 analogs that possess a benzoxazolone group in place of the catechol moiety on one or both sides of AG 538. Compounds that possess one benzoxazolone ring, namely 4, 10, 15, and 21, inhibit the IGF-1R selectively and do not inhibit Src and PKB ( Table I). Two of the more potent compounds, 4 and 10, were chosen for further investigation. These inhibitors are 6 -7-fold less active as compared with AG 538. AG 538 as well as its bioisostere analogs are selective for IGF-1R, as compared with their inhibitory activity toward Src and PKB in cell free assays (Table I) and toward the PDGFR signaling pathway in intact cells (Fig. 3D).
Both compounds 4 and 10 are substrate-competitive inhibitors of the IGF-1R ( Fig. 1 and data not shown). Such inhibitors may have an advantage over ATP-mimicking inhibitors that have to compete with high intracellular ATP concentrations (for review see Ref. 34).
The IGF-1R is known to activate two major pathways, the anti-apoptotic pathway mediated by PI3K-PKB and the Shc-Ras-Erk proliferative pathway. The IRS proteins are direct substrates of the IGF-1R and mediate the PI3K-PKB pathway as well as others. When applied to cells, compounds 4 and 10 inhibit IGF-1R autophosphorylation as well as IGF-1R-induced phosphorylation of IRS-1, PKB, and Erk2 in a dose-responsive manner, similarly to AG 538 (Fig. 2). A shorter (40 min) incubation of these inhibitors in intact cells also inhibits IGF-1R downstream signaling, i.e. phosphorylation of IRS-1, PKB, and Erk-2, in a dose-responsive manner (data not shown). For the compounds tested, the concentrations needed to inhibit the autophosphorylation of IGF-1R were higher than the concentrations needed to inhibit PKB and IRS-1. This difference may result because autophosphorylation represents intermolecular trans-autophosphorylation within the IGF-1R dimer. Therefore, the inhibitor must compete against a local substrate concentration that is much higher than the concentrations of exogenous substrates like IRS-1 or PI3K.
The effect of the inhibitors on IGF-1-induced Erk2 phosphorylation in cells is less pronounced than the effect on PKB and IRS-1 phosphorylation (Fig. 2). Erk2 activation is mediated by Shc protein, a separate signaling pathway from IRS-1/PI-3 kinase/PKB (35). Since AG 538 and compounds 4 and 10 are IGF-1R substrate-competitive, the affinity of Shc proteins to IGF-1R and their concentrations in the cells should determine the efficacy of the inhibitors toward the Shc protein pathway (Erk2), whereas the affinity of IRS-1 to IGF-1R and the concentration of IRS-1 should determine the efficacy toward the IRS-1/PKB pathway. The different effects of the inhibitors on IGF-1-induced Erk2 and IRS-1/PKB phosphorylation may derive from differences in the affinities and/or concentrations of the relevant IGF-1R substrates.
Compounds 4 and 10 and AG 538 inhibit the activation of PKB by the IGF-1R in intact cells, with IC 50 values of 12, 6, and 19 M, respectively. These inhibitors do not affect PKB activation by PDGF-R in intact cells as shown in Fig. 3, confirming their selectivity.
Although the cell-free potency of compounds 4 and 10 is 6 -7-fold lower than that of AG 538 (IGF-1R inhibition, Table I), their potency in cells is similar to that of AG 538, when examined in long term assays of inhibition of growth and of colony formation in soft agar as shown in Tables III and IV. This most likely reflects the increased stability of the bezoxazolone-containing compounds in cells.
Inhibitors of the IGF-1R and its signaling pathways could be extremely useful as anti-neoplastic agents. Indeed the compounds described in this paper inhibit anchorage-dependent growth of R ϩ and MDA MB-468 breast cancer cell lines ( Fig. 4 and Table III). These compounds also led to morphological reversion of transformed R ϩ cells (IGF-1R overexpressors), leading to a morphology similar to that of the non-transformed R Ϫ (IGF-1R deleted) cells (data not shown).
The formation of colonies in soft agar by cancer cells is an indication of their invasiveness, and the IGF-1R is known to be a positive regulator of this phenotype (36,37). Compounds 4 and 10 and AG 538 were found to inhibit the formation of colonies in soft agar in two breast cancer cell lines and two prostate cancer cell lines (Fig. 5 and data not shown). The IC 50 values of these inhibitors are presented in Table IV. It can be seen that compounds 4 and 10 and AG 538 inhibit IRS-1 phosphorylation/activation by IGF-1 in representative breast and prostate cancer cell lines (Fig. 6). Compounds 4 and 10 and AG 538 inhibit anchorage-independent cell growth of the MDA MB-468 and MCF-7 breast cancer lines at lower concentrations than the inhibition of anchoragedependent cell growth. For example, comparison of unanchored to anchored growth inhibition in MDA MB-468 cells: IC 50 of 8, 10, or 6 M versus IC 50 of 51, 38, or 25 M, respectively. This finding supports the view that IGF-1R may be more important in maintaining the transformed phenotype than in promoting monolayer cell growth as suggested previously (38). This study describes the action of antisense IGF-1R RNA on human melanoma cells incubated subcutaneously or grown in soft agar or in monolayer. The antisense induced massive apoptosis of the subcutaneous incubated cells, whereas the same cells grown in monolayer treated with the same antisense were only moderately inhibited, and the effect on cells growing in soft agar was somewhere in-between. It therefore seems that when cells grow without anchorage they depend for their survival on anti-apoptotic pathways such as the IGF-1R-mediated pathways. Therefore, inhibition of the IGF-1R in unanchored cells causes massive apoptosis (39).
The ability of the IGF-1R inhibitors to inhibit anchorageindependent cell growth better than anchorage-dependent growth has important implications for the toxicity of an IGF-1R inhibitor as a potential therapeutic agent. An agent that can selectively inhibit transformed, unanchored cells without inhibiting normal, anchored cells would be a highly valuable anti-tumor drug, since it is likely to possess less toxicity.
In summary, in this study, we present a new class of substrate-competitive IGF-1R inhibitors, which possess a novel chemical moiety, resistant to oxidation. The ability to replace the catechol groups in substrate-competitive tyrphostins with a moiety that is more stable in cells should aid in developing non-catechol tyrosine kinase inhibitor drugs for clinical use. Furthermore, development of bioisostere tyrphostins will be necessary in order to generate potent and selective IGF-1R kinase inhibitors with anti-tumor activity.