Thiazolidine-Diones BIOCHEMICAL AND BIOLOGICAL ACTIVITY OF A NOVEL CLASS OF TYROSINE PROTEIN KINASE INHIBITORS*

Various derivatives of thiazolidine-diones have been identified as tyrosine protein kinase inhibitors. The epidermal growth factor (EGF) receptor kinase and c-src kinase were inhibited in vitro with IC50 values in the range of 1-7 microM. The v-abl tyrosine protein kinase was not inhibited by thiazolidine-diones. Inhibition was found to be specific for tyrosine protein kinases. Inhibition of serine/threonine protein kinases was not observed. The active derivatives were shown to inhibit EGF-induced receptor autophosphorylation, either in vitro or in intact cells, and were also found to inhibit growth of the EGF-dependent BALB/MK and A431 cell lines (IC50 1-3 microM). Growth of the interleukin-3-dependent myeloid cell line FDC-P1 was inhibited with equal efficiency. Thus, in these cell lines, members of the c-src kinase family are also potential targets for inhibition by the compounds.


Various
derivatives of thiazolidine-diones have been identified as tyrosine protein kinase inhibitors. The epidermal growth factor (EGF) receptor kinase and csrc kinase were inhibited in vitro with ICbO values in the range of l-7 MM. The v-abl tyrosine protein kinase was not inhibited by thiazolidine-diones. Inhibition was found to be specific for tyrosine protein kinases. Inhibition of serinelthreonine protein kinases was not observed.
The active derivatives were shown to inhibit EGF-induced receptor autophosphorylation, either in vitro or in intact cells, and were also found to inhibit growth of the EGF-dependent BALB/MK and A431 cell lines (I&,,. Growth of the interleukin-3dependent myeloid cell line FDC-Pl was inhibited with equal efficiency.
Thus, in these cell lines, members of the c-src kinase family are also potential targets for inhibition by the compounds.
Gene products involved in transducing growth stimuli in response to growth factors are frequently associated with malignant cellular proliferation. Such gene products thus represent attractive targets for chemotherapeutic intervention and interruption of signal transduction pathways. The EGF' receptor is a transmembrane glycoprotein which transduces mitogenic signals generated by binding of EGF or transforming growth factor o( (1). The receptor is composed of an extracellular hormone-binding domain and an intracellular tyrosine protein kinase domain which catalyzes receptor autophosphorylation and phosphorylation of cellular substrates upon stimulation . The growth factor-dependent tyrosine kinase activity of the cytoplasmic domain is regarded as the primary mechanism for the generation of intracellular signals that initiate the multiple cellular responses which culminate in DNA synthesis and cell proliferation. The dependence of this pathway on tyrosine kinase activity has been confirmed by site-directed mutagenesis. Mutations within the kinase domain which eliminate enzyme activity have been shown to abolish signal transduction via the EGF receptor (7,8).
Comparison of the primary amino acid sequence within the kinase domain of members of the tyrosine protein kinase gene family have revealed a high degree of homology due to conservation of residues essential for catalytic function, especially within the nucleotide-binding site (9). However, since tyrosine protein kinases show distinct substrate specificities (10, 11) the design of selective enzyme inhibitors should be possible. Several classes of compounds have been reported to inhibit tyrosine protein kinases. These compounds include halomethyl ketones (12, 13), amiloride (14), the microbial alkaloid staurosporine (15)(16)(17), and flavonoids such as genistein and quercetin (18)(19)(20) mean and standard deviations of two to three independent experiments.

Zmmunoblot
Analysis-A431 cells were seeded in T-25 tissue culture flasks at lo5 cells/flask in DMEM with 10% FCS, grown to 80% confluence and then starved for 24 h in DMEM containing 0.5% FCS. Compounds were added and incubated with cells for the indicated pretreatment times. EGF (400 rig/ml) was added to cultures which were then incubated for another 10 min. Cells were scraped from the dish surface into the culture medium, centrifuged at 1000 x g for 5 min and washed with 10 ml of phosphate-buffered saline. Cells were lysed in SDS-PAGE sample buffer and boiled for 5 min. Following SDS-PAGE (8% gels), proteins were transferred onto Immobilon membranes (Millipore, Bedford, MA) by semidry blotting. Membranes were blocked overnight at 4 "C! with 3% bovine serum albumin/ phosphate-buffered saline and then incubated with primary antibody (1 pg/ml in 3% bovine serum albumin/phosphate-buffered saline) for 2 h. Biotinylated secondary antibody and Avidin/Biotin complex with alkaline phosphatase were used for development according to the manufacturer (Vector). I& values for phosphotyrosine blots were estimated by laser densitometry.

Protein
Determination-Protein concentrations were determined according to the method of Bradford (39), using bovine serum albumin as a standard.

RESULTS
In Vitro Activity of Compounds as Inhibitors of Protein Kinases-A series of thiazolidine-diones of the general formula I (Fig. 1) were synthesized as described (Fig. 2) and tested in vitro for inhibition of the EGF receptor tyrosine protein kinase (Table I). Compounds l-3, which have either a =CH2 or -CH*N(CH& group at Rz, were found to be potent inhibitors of the EGF receptor kinase with I& values of 6, 1, and 2 PM, respectively, when assayed in vitro using an exogenous peptide substrate (angiotensin II). The compounds were further tested for selectivity within the tyrosine protein kinase enzyme family using recombinant c-src and v-abl kinases. The inhibition profiles of compounds l-10 with respect to c-src kinase were essentially the same as for the EGF receptor kinase (i.e. compounds l-3 were inhibitors of the c-src kinase with ICsO values of 4, 3, and 7 KM, respectively), while compounds 4-10 were inactive. In marked contrast, the v-abl kinase was not inhibited by compounds l-10 (I&, > 100 PM). Only compound 2 was found to have marginal inhibitory  Details of the chemical synthesis of the compounds are described under "Experimental Procedures."   (Fig. 3A). The IC,() value is similar to that obtained using the non-physiological exogenous peptide substrate angiotensin II (Table I). The effect of compound 2 on v-abl kinase autophosphorylation was tested (Fig. 3B). Only at the highest concentration of drug (100 PM) was inhibition of autophosphorylation observed (Fig. 3B, lane 2)  A, inhibition of EGF receptor autophosphorylation. A431 membranes were incubated for 10 min with EGF (lanes 2-7) before addition of the indicated concentrations of compound 2 (lanes 3-7). Autophosphorylation reactions were started by addition of radiolabeled ATP and terminated by the addition of SDS-PAGE sample buffer. An autoradiogram of the dried SDS gel (8%) after 2 h of exposure is shown. B, effect of compound 2 on v-abl autophosphorylation.
Recombinant v-abl kinase was assayed using 2 j.~cM radiolabeled ATP. Compound 2 (lanes 2-7) and compound 6 (lane I) were added at the indicated concentrations. After stopping reactions with sample buffer, the mixture was subjected to SDS-PAGE (10% gel). An autoradiogram of the dried gel is shown. c-src kinase was assayed as described under "Experimental Procedures" using 0.125 mg/ml Glu,Tyr (4:l) as exogenous substrate. Concentrations of compound 2 were 0 (W), 3.1 pM (V), and 6.2 I.~M (0). Points represent the mean of quadruplicate determinations. Curves were calculated using the Enzfitter program (40) with simple weighting. The Lineweaver-Burk representation is shown as an inset. performed to determine the kinetic nature of inhibition by thiazolidine-diones relative to ATP. The c-src kinase was used as the target enzyme in kinetic studies due to the higher quality of results obtained with pure enzyme. Using the random polymer Glu,Tyr (4:l) at 0.125 mg/ml as substrate, the K,, for ATP was 17.6 pM (S.D. = 3.9), and the V,,,, was 150 (SD. = 8) nmol/min/mg. Competition experiments between the inhibitor and ATP indicated that compound 2 was not a competitive inhibitor with respect to the nucleotide, but exhibited mixed type inhibition (Fig. 4). Inhibition of EGF-induced Cell Proliferation-BALBIMK mouse epidermal keratinocytes have a strong dependence on EGF for proliferation (34; Fig. 5, panel B). Thiazolidine-dione analogues were tested for inhibition of EGF-stimulated cellular proliferation of BALB/MK cells. Compounds 1-3, which were potent inhibitors of the EGF receptor kinase in vitro, showed potent antiproliferative activity with I&) values of -1 pM. (Fig. 5, panel A). Enzymatically inactive analogues were without antiproliferative activity, Table II). In addition, compounds were tested for antiproliferative activity using the A431 cell line, which is known to express high levels of the EGF receptor (41). The thiazolidine-diones were found to have an essentially identical antiproliferative activity profile as compared to that obtained using the BALB/MK cell line (active compounds having IC,o values in the range l-2 ELM (Table II). Compounds were then tested for antiproliferative activity using the FDC-Pl cell line which lacks EGF receptor expression. FDC-Pl is a myeloid, bone marrow-derived cell line (35), which requires either IL-3 or granulocyte macrophage colony stimulating factor for growth. Growth of this cell line was also inhibited by compounds l-3 (Table II)  dependent cellular tyrosine phosphorylation and EGF receptor expression was monitored by immunoblotting with phosphotyrosine and EGF receptor-specific antibodies.
Compounds 1-3, which inhibited the EGF receptor kinase in vitro, Only data on inhibition of the epidermal growth factor-R (EGF-R), insulin receptor (IR), v-src, c-src, v-abl, PKA (CAMP-dependent protein kinase) and PKC (protein kinase C) protein kinases have been included. Inhibitors have been limited to the best characterized members of each structural class. To enable comparison with thiazolidine-diones (Table I) showed clear inhibition of EGF receptor autophosphorylation in A431 cells (Fig. 6A, lanes 3-5). In contrast, the negative control compound 6, which was inactive as an inhibitor in u&o, had no effect on cellular EGF receptor autophosphorylation (kme 6). Drug treatment was found to have no significant effect on EGF receptor expression when compared with untreated control cells (Fig. 6B). The inhibitory effect of compound 2 on EGF receptor autophosphorylation was dependent on the drug concentration and treatment time. Following a 2-h preincubation period with compound 2, an ICsO value in the range of 12.5-25 ~.LM was observed (Fig. 7). Similar results were obtained for compounds 1 and 3 (results not shown). The inhibitory effect of the compounds increased with increasing preincubation time from 2 to 10 h, while the level of EGF receptor expression was essentially constant (data not shown). Following treatment with high concentrations of compounds 1-3, A431 cells detached from the culture dish surface in extended sheets (50% detachment with 100 FM of compounds l-3 after 1 h of treatment). This effect was not observed with the enzymatically inactive analogues (compounds d-10).

DISCUSSION
A number of tyrosine protein kinase inhibitors of different structural classes have previously been reported. For comparative purposes a selection of the best characterized examples are shown in Table III. The thiazolidine-diones are a new class of protein kinase inhibitors with specificity for tyrosine protein kinases. With the serinelthreonine protein kinases tested, no inhibition was observed. Within the family of tyrosine protein kinases, which are highly homologous in their kinase domains, inhibitors of the thiazolidine-dione class exhibit partially selective enzyme inhibition.
While both the EGF receptor kinase and the c-src kinase were inhibited with equal efficiency, the same compounds were inactive (I& > 100 pM) on the v-&l kinase. These results were surprising as the c-src and v-&l kinases are viewed as more closely related to each other than to the EGF receptor kinase when compared at the level of sequence homology within their kinase domains (9). However, in terms of substrate preference in u&o, c-src kinase appears to be more similar to the EGF receptor kinase. Both kinases accept acidic random amino acid polymers, which are poor substrates for the v-abl kinase (1O).3 From their kinetic behavior, thiazolidine-diones were not competitive with ATP and thus do not behave as ATP analogues. However, as mixed type inhibitors they may discriminate between members of the tyrosine protein kinase family with different substrate preferences.
Autophosphorylation of the EGF receptor is considered to be the first enzymatic event following ligand binding (l-6, 42). The physiological function of this event is incompletely understood. However, it is generally considered as a regulatory mechanism which results in the relief of an inhibitory constraint (2) From the immunoblot analysis of EGF receptor autophosphorylation in A431 cells, it was found that compounds l-3 were able to inhibit EGF receptor autophosphorylation within intact cells without affecting the level of EGF receptor expression. From the time dependence of the inhibitory effect, it can be concluded that permeation or equilibration of the compounds within the cell is a relatively slow process occurring over a period of hours. Following a 2-h preincubation period, the IC& for compound 2 was in the range of 12.5-25 PM.
Since compounds were shown to exert their inhibitory effect on EGF receptor in vitro as well as in intact cells, they were tested for growth inhibition of the EGF-dependent cell line BALB/MK, and cells with overexpressed levels of the EGF receptor (A431 cells). Inhibition was observed at I& values of l-2 ELM, i.e. -lo-fold lower concentrations than inhibition of autophosphorylation in intact cells. However, in antiproliferative assays, cells were incubated with drugs over a period of several days. This long term treatment might explain the low drug concentrations required for inhibition of cellular growth by compounds l-3. It is also possible that the threshold drug concentration for inhibition of proliferation is less than the concentration required for complete inhibition of receptor autophosphorylation.
Additionally, phosphorylation of other cellular substrates of the EGF receptor kinase might be more sensitive to inhibition than those substrates tested. The possibility cannot be excluded that effects unrelated to inhibition of EGF receptor kinase activity were responsible for inhibition of cell proliferation. However, an effect completely unrelated to kinase inhibition appears unlikely, since inhibition of EGF dependent growth was only observed with thiazolidine-dione analogues which were active as inhibitors in vitro.
The proliferation of the EGF-independent FDC-Pl cell line was equally well inhibited by compounds l-3 as were the EGF-dependent BALB/MK and A431 cell lines. The finding, that the c-src kinase was inhibited in vitro by thiazolidinediones with the same efficiency as the EGF receptor kinase " J. F. Geissler, N. B. Lydon, unpublished observations. might explain the broad spectrum of antiproliferative effects of these compounds. In the case of FDC-Pl, possible targets are the c-src kinase or other members of the c-src enzyme family (9). Evidence is accumulating, that products of the csrc gene family may play an important role in the mitogenic response of hematopoietic cells to proliferation signals. In FDC-Pl cells, either of the mitogenic factors granulocyte macrophage colony stimulating factor or IL-3 have been shown to activate c-src kinase (46). Tyrosine phosphorylation also occurs in response to IL-3 treatment in other myeloid cells (47)(48)(49). Thus, it is likely that the inhibition of IL-3dependent FDC-Pl cell proliferation is due to inhibition of a tyrosine kinase activity required for growth factor-dependent proliferation.
Recently, it has been suggested that phosphorylation of c-src on mitosis-specific sites may influence mitotic events, presumably by regulating c-src kinase activity (50, 51). Compounds that inhibit c-src kinase might thus be general inhibitors of cell growth due to interference with events in the M-phase of the cell cycle. In this respect the detachment of A431 cells from the culture dish surface upon treatment with active thiazolidine-diones was of interest. It has been suggested that phosphorylation of mitosis-specific sites of csrc might influence the cytoskeleton, bringing about shape changes during mitosis (50, 51). Such effects of c-src kinase on cell morphology and cell to cell communication are well documented (52). However, due to other possible targets, including other members of the protein kinase family, the antiproliferative effects of thiazolidine-diones cannot at present be unambiguously linked to inhibition of a single target enzyme. Y., Yaish, P., Chorev, M., Gilon, C., Braun, S., and Levitzki,