Cdc25 Inhibition and Cell Cycle Arrest by a Synthetic Thioalkyl Vitamin K Analogue¹

The vitamin K family of molecules comprises the natural forms vitamin K₁ (phylloquinone) and vitamin K₂ (menaquinones) and the synthetic form vitamin K₃ (menadione). These naphthoquinone-containing molecules inhibit tumor cell growth in culture, with vitamin K₃ being more potent than either vitamin K₁ or K₂ (1). Vitamin K₃ exhibits low toxicity to animals(2, 3) and can enhance the antiproliferative effects of other clinically used anticancer agents (4), although it is toxic to humans (5). The growth-inhibitory actions of vitamin K₃ have been ascribed to both sulfhydryl arylation and oxidative stress because of redox cycling (6, 7). We previously synthesized and characterized a thioalkyl K vitamin analogue, Cpd 5³(Fig. 1), with superior growth-inhibitory activity that also rapidly enhances cellular protein tyrosine phosphorylation and causes apoptosis(8). The antiproliferative and antiphosphatase activity of Cpd 5 is antagonized by exogenous thiols but not by nonthiol antioxidants, suggesting that unlike vitamin K₃,its inhibition is mediated by sulfhydryl arylation rather than oxidative stress (8). One proposed site for interaction is the catalytic cysteine(s) found in protein tyrosine phosphatases that regulate cell proliferation (2).

Protein tyrosine phosphatases that have an essential role in cell cycle progression include the Cdc25 phosphatases, which activate Cdks. In mammalian cells, Cdc25 phosphatases are encoded by a multigene family consisting of Cdc25A, Cdc25B, and Cdc25C (9, 10, 11). Each Cdc25 homologue controls distinct aspects of cell cycle progression. Cdc25C dephosphorylates and activates the mitotic kinase Cdc2/cyclin B,which is required for entry into mitosis (12). Cdc25A is important for entry into S-phase (13), whereas Cdc25B is essential for preinitiating G₂-M transition and S-phase progression (14). Cdc25A and Cdc25B have oncogenic properties in cells that have mutated Ha-ras or loss of Rb1, the Rb susceptibility gene (15). Cdc25A and Cdc25B are transcriptional targets of the c-myconcogene (16) and are overexpressed in several tumor types and may reflect poor prognosis (15, 17, 18, 19). Unfortunately, potent and selective inhibitors of Cdc25 phosphatases are currently unavailable but would be attractive candidates as potential anticancer agents.

In human hepatoma cells, vitamin K₃ induces hyperphosphorylation of p34^cdc2 (Cdc2) kinase and decreases the protein tyrosine phosphatase activity in cell lysates(20). Vitamin K₃ and other naphthoquinone analogues inhibit Cdc25A in vitro, and one of these analogues has been shown to cause G₁arrest (21). The mechanism by which the potent redox-deficient thioalkyl K vitamin analogue Cpd 5 inhibits cell growth is not known, although inhibition of Cdc25 has been hypothesized(2). Thus, we have examined the actions of Cpd 5 and two other vitamin K analogues on protein tyrosine phosphatases, including Cdc25A, Cdc25B, and Cdc25C, as well as their antiproliferative and cell cycle checkpoint activity.

tsFT210 cells were a generous gift from Dr. Chris Norbury (Oxford University, Oxford, United Kingdom) and were maintained for no longer than 30 passages as described elsewhere (22). The anti-Cdc2 (SC 54), anti-Cdk2 (SC 163G), anti-Cdk4 (SC 601G),anti-cyclin D1 (SC 6281), anti-cyclin E (SC 481), anti-p53 (SC 1312),anti-p21 (SC 3976), and anti-p16 (SC 1207) antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Agarose conjugate of each antibody was used for immunoprecipitation. Anti-cyclin A antibody was purchased from Oncogene Research Product(Cambridge, MA). Anti-phosphotyrosine antibody was purchased from Upstate Biotechnology (Lake Placid, NY). Phospho-Rb antibody and Rb antibody were purchased from New England Biolabs, Inc. (Beverly, MA),and anti-GAPDH antibody was purchased from Chemicon International, Inc.(Temecula, CA). Histone H1 was obtained from Boehringer Mannheim Co.(Indianapolis, IN). [γ-³²P]ATP (10 mCi/ml)was from Amersham Life Science, Inc. (Arlington Heights, IL).

To synthesize Cpd 5, we added 1,8-diaza-bicyclo[5.4.0]un-dec-7-ene (0.07 ml, 0.7 mmol) dropwise to a solution of menadione (5.154 g, 29.9 mmol) and 2-mercaptoethanol(2.10 ml, 29.9 mmol) in 150 ml ether at room temperature. Stirring was maintained at room temperature for 23 h, and then 20 ml of 3.6 m HCl were added. The organic and aqueous phases were separated, and the aqueous layer was extracted with ether. The combined organic layers were dried over magnesium sulfate, filtered, and concentrated to give 8.223 g of a dark brown viscous liquid. Purification by flash chromatography using 30% ethyl acetate/hexanes to elute the first two bands, followed by 50% ethyl acetate/hexanes,gave 2.698 g (36%) of an orange solid: melting point 78–80°C. ¹H NMR (CDCl₃) δ8.08–8.03 (m, 2H), 7.71–7.68 (m, 2H), 3.80 (t, J = 5.7, 2H), 3.34 (t, J = 5.8, 2H), 2.50 (s, 1H), 2.38 (s,3H); ¹³C NMR (CDCl₃)182.10, 181.42, 147.85, 145.87, 133.74, 133.41, 132.61, 131.86, 128.78,126.55, 62.02, 37.11, 15.40; IR (KBr) cm⁻¹ 3292(m), 1657 (s), 1585 (s), 1554 (s); UV (ethanol) λmax (log ε) 204(4.21), 260 (4.22), 408 (3.33); MS (m/z): 248 (2), 230 (63),221 (100), 197 (73); high resolution MS: calculated for C₁₃H₁₀O₂S:230.0412, found: 230.0405.

To synthesize Cpd 22, we used Fieser’s method. A solution of sodium carbonate (2.40 g, 22.6 mmol) and 30% hydrogen peroxide (10 ml, 97.9 mmol) in water (50 ml) was added together to a solution of menadione(10.168 g, 59.1 mmol) in warm ethanol (110–135 ml). The yellow quinone color disappeared, and the flask was then cooled in ice. Water was added (300 ml) to give 16.334 g of a white solid; melting point 87°C. 1.076 g of this solid was dissolved in concentrated sulfuric acid (6 ml), giving a deep red solution. The flask was swirled intermittently for 10 min, and then water (20 ml) was added slowly to give a yellow precipitate. The mixture was filtered, and the filtercake was washed with water until the filtrate was no longer acidic. This procedure afforded 0.724 g (67%) of Cpd 22; melting point 170–171°C. ¹H NMR (CDCl₃) δ8.15–8.07 (m, 2H), 7.79–7.66 (m, 2H), 7.3 (s, 1H), 2.12 (s,[³H]); ¹³C NMR(CDCl₃) 185.08, 181.23, 153.21, 134.89, 132.96,129.47, 126.78, 126.18, 120.60, 8.74; IR (KBr)cm⁻¹ 3312 (brs), 1643 (s), 1579 (m); UV(ethanol) λmax (log ε) 206 (4.54), 242 (4.48), 252 (4.47), 276(4.65); MS (m/z): 188 (100), 160 (30), 132 (42), 105 (33),77 (35); high resolution MS: calculated for C₁₁H₈O₃:188.0473442, found: 188.049347.

To synthesize Cpd 16, we added a solution of Cpd 22 (0.713 g, 3.793 mmol) in dry THF (4 ml) via cannula to a suspension of potassium hydride (0.229 g, 5.711 mmol) in dry THF (10 ml) at 0°C. The resulting dark brown mixture was stirred for 5–10 min when a solution of 18-Crown-6 (1.542 g, 5.841 mmol) in dry THF (4 ml) was added. In some reactions, we used supplemental THF to aid in the stirring of the solution. The resulting burgundy mixture was stirred for 20 min, and then 1-iodooctane (0.68 ml, 3.768 mmol) was added. The mixture was refluxed for 21 h and then stirred at room temperature for 24 h. The reaction was quenched with saturated ammonium chloride and extracted with ether. The organic and aqueous phases were separated,and the aqueous layer was extracted with ether. The combined organics were dried over magnesium sulfate, filtered, and concentrated to give 1.704 g of a dark brown liquid. Purification by flash chromatography using 5% ethyl acetate/hexanes gave 1.11 g (98%) of a yellow solid: melting point, 38°C. ¹H NMR(CDCl₃) δ 7.85–7.80 (m, 2H), 7.53–7.46 (m,2H), 4.22 (t, J = 6.6, 2H), 1.94 (s, 3H), 1.68–1.59 (m,2H), 1.35–1.16 (m, 10H), 0.77 (t, J = 6, 3H). ¹³C NMR (CDCl₃) 185.26,180.90, 157.20, 133.33, 132.84, 131.70, 131.40, 131.22, 125.81, 73.51,31.67, 30.42, 29.18, 29.12, 25.69, 22.52, 13.97, 9.15; IR (KBr)cm⁻¹ 1668 (s), 1620 (s) 1593 (s); UV (ethanol)λmax (log ε) 208 (4.20), 248 (4.29), 276 (4.07), 334 (3.51); MS(m/z): 316 (15) 201 (30), 188 (100), 172 (45), 160 (30);high resolution MS: calculated for C₁₉H₂₄O₃:300.1725, found: 300.1745.

tsFT210 cells were plated at 2 × 10⁵ cells/ml and maintained at 32.0°C as described previously (22). Cell proliferation was blocked at G₂ phase by incubation at 39.4°C for 17 h. The synchronized cells were then released by reincubating at 32.0°C and treated immediately with Cpd 5, Cpd 16, Cpd 22, or SC-ααδ9, respectively, to probe for G₂-M arrest. Cells were treated 6 h after G₂-M release to determine G₁ arrest. For both G₂-M and G₁ blockage studies, treated cells were incubated at 32.0°C for an additional 6 h after each drug exposure and then harvested with PBS at 5 × 10⁵ cells/ml. The harvest cells were stained with a solution containing 50 μg/ml propidium iodide and 250 μg/ml RNase A. Flow cytometry analysis was conducted with a Becton Dickinson FACS Star (Franklin Lakes, NJ). Each compound was tested at least three independent times. A final concentration of 0.5% DMSO was used for all compounds and as a negative control. For positive controls, we used 100 μm SC-ααδ9 (for both G₂-M and G₁), 1μ m nocodazole (for G₂-M) ,or 50μ m roscovitine (for G₁).

The preparation of plasmid DNA and GST-fusion proteins has been described previously (23). The activities of the GST-fusion Cdc25A, Cdc25B₂, Cdc25C, and VHR, as well as human recombinant PTP1B, were measured as described previously(23) in a 96-well microtiter plate using the substrate OMFP (Molecular Probes, Inc., Eugene, OR), which is readily metabolized to the fluorescent o-methyl fluorescein. OMFP concentrations approximating the K_m were used:Cdc25A, Cdc25B₂ and Cdc25C, 40μ m; VHR, 10 μm; and PTP1B, 200 μm. Inhibitors were resuspended in DMSO, and all reactions including controls were performed at a final concentration of 7% DMSO. The final incubation mixture (150 μl) was optimized for enzyme activity and comprised 30 mmTris (pH 8.5 for Cdc25 phosphatases; pH 7.5 for VHR and PTP1B), 75 mm NaCl, 1 mm EDTA, 0.033%BSA, and 1 mm DTT. Reactions were initiated by adding 1 μg of Cdc25 phosphatases, 0.025 μg of VHR, or 0.25 μg of PTP1B phosphatase. Fluorescence emission from the product was measured over a 20–60 min reaction period at ambient temperature with a multiwell plate reader (PerSeptive Biosystems Cytofluor II; Framingham,MA; excitation filter, 485/20; emission filter, 530/30). For all enzymes, the reaction was linear over the time used in the experiments and was directly proportional to both the enzyme and substrate concentration. Best curve fit for Lineweaver-Burk plots and K_is was determined by using the curve-fitting programs Prism 3.0 (GraphPad Software, Inc., San Diego,CA) and EZ-Fit 5.03 (Perrella Scientific, Inc., Amherst, NH).

tsFT210 cells were harvested and sonicated in the lysis buffer using the same procedure for cell synchronizing and drug exposure as described above for the G₁ flow cytometric analysis. For the phospho-Rb study, we harvested the cells at each time point: -6 h (releasing point from the G₂-M synchronizing); and -3, 0, 1.5, 3, and 6 h after treatment with 20μ m Cpd 5. The protein lysates were analyzed by Western Blot for phospho-Rb, Rb, GAPDH, p53, p21, and p16. Immunoprecipitation assays were performed essentially as described previously(24), except we replaced 0.1% Tween 20 for 1% Triton X-100 in the lysis buffer. We incubated 2 mg of protein lysate on a rocker platform with 10 μg of anti-Cdk2 or anti-Cdk4 agarose conjugate for 4 h at 4°C. The immunocomplexes were washed four times with the same lysis buffer. After the final wash, the immunocomplexes were suspended with SDS-electrophoresis loading buffer and analyzed by Western blotting for Cdk2, tyrosine phosphorylated Cdk2, Cdk4, tyrosine phosphorylated Cdk4, cyclin A, cyclin E, and cyclin D1 as described above. To quantify the phosphorylation level of Cdk2 or Cdk4, we scanned X-ray films and analyzed band intensity on a Molecular Dynamics personal SI densitometer and analyzed them using the Image Quant software package (Ver. 4.1; Molecular Dynamics, Sunnyvale,CA). The phosphorylation level (pCdk2/Cdk2) was calculated by using the formula; pCdk2/Cdk2 = (a)/(b),where a was the intensity of the phosphorylated Cdk2 band and b was the intensity of the Cdk2 band, respectively. Statistical significance was analyzed using Student’s unpaired t test.

tsFT210 cells were synchronized, exposed to drugs, and harvested as described above for the G₂-M flow cytometric analysis. The protein lysates were analyzed by Western blot for Cdc2 as described previously (25). Cdc2 kinase activity assay was performed as described previously (26). Briefly, the Cdc2 immunoprecipitates were incubated in 20 μl of kinase reaction buffer(26) for 30 min at 37°C, with 3 μg of histone H1, 20 mm Tris-HCl, 10 mm MgCl₂,5 μm cold ATP, and 10 μCi of[γ-³²P]ATP. The proteins were separated by SDS-PAGE and analyzed with a Molecular Dynamics STORM 860 PhosphorImager (Sunnyvale, CA).

We initially determined whether Cpd 5 blocked cell cycle progression through checkpoints using murine tsFT210 cells, because they can be readily synchronized with exogenous compounds because of a ts Cdc2(22). When tsFT210 cells were incubated at the permissive temperature of 32.0°C, they had a normal cell cycle distribution(Figs. 2,A and 3,A); when cells were incubated at the nonpermissive temperature of 39.4°C, they arrested at G₂-M, because of Cdc2 inactivation (Ref.22; Figs. 2,B and 3,B). When G₂-M arrested cells were cultured at the permissive temperature for 6 h with DMSO vehicle alone, we saw clear evidence of entry into G₁ (Fig. 2,C). In contrast, 1 μm nocodazole blocked cell passage through G₂-M (Fig. 2,D). To determine the effect of Cpd 5 on G₂-M cell cycle transition, we treated cells with either 10 or 20 μm Cpd 5 for 6 h after releasing cells at 32.0°C. As indicated in Fig. 2, E and F, both concentrations of Cpd 5 significantly arrested cells at G₂-M phase. This G₂-M inhibition was selective to Cpd 5 and not seen with two structural analogues, i.e., Cpd 16 and Cpd 22 (Fig. 1, G and H). The G₂-M inhibition was similar to that seen with another inhibitor of the Cdc25 family of phosphatases,SC-ααδ9, which is structurally unrelated (Refs. 23and 27; Fig. 2 I).

We next examined whether Cpd 5 caused G₁ arrest in tsFT210 cells. To investigate the mechanism of G₁ cell cycle block by Cpd 5, we arrested tsFT210 cells at G₂-M by shifting to the nonpermissive temperature, then released them into G₁ by shifting to the permissive temperature, and subsequently added either Cpd 5 or DMSO vehicle 6 h later. Cells that were treated with the DMSO vehicle passed through G₁ phase as expected and produced the broad S-phase peak (Fig. 3,D), whereas cells exposed continuously to 50μ m roscovitine were blocked and did not pass through G₁ (Fig. 3,E). As illustrated in Fig. 3, F–I, cells treated with 5 or 10μ m Cpd 5 were delayed, whereas cells treated with 15 or 20 μm Cpd 5 were fully blocked at G₁. In contrast, neither Cpd 16 nor Cpd 22 at 20μ m blocked G₁ transition(data not shown). As expected from our previous studies (Fig. 2), Cpd 5 not only caused a G₁ block but also prevented cells that were in the G₂ phase from progressing through G₂-M, which resulted in two prominent cell cycle peaks (Fig. 3, H and I). This dual G₁ and G₂-M inhibition was similar to that seen with a much higher concentration of the structurally unrelated and less potent Cdc25 inhibitor, SC-ααδ9(Fig. 3 J).

Because of the dual G₁ and G₂-M blockage with Cpd 5 and previous speculation concerning possible phosphatase inhibitory activity (8),we examined the inhibitory activity of Cpd 5, Cpd 16, and Cpd 22 (Fig. 1) against the dual specificity phosphatases Cdc25B₂ and VHR and the tyrosine phosphatase PTP1B. At 30 μm, Cpd 5 caused >75% inhibition of recombinant human Cdc25B₂ activity with only a small effect on VHR and no inhibition of PTP1B (Fig. 4,A). In contrast, identical concentrations of the close structural analogues, Cpd 16 and Cpd 22, did not inhibit any of these protein phosphatases, indicating the essential nature of theβ-mercaptoethanol moiety for enzyme inhibition. A more extensive study revealed that the Cdc25B₂IC₅₀ for Cpd 5 was 3.8 ± 0.6μ m compared with >150μ m for the close analogues Cpd 16 and Cpd 22(Fig. 4,B). Thus, Cpd 5 was 40-fold more active than Cpd 16 or Cpd 22. The selectivity of Cpd 5 is illustrated in Fig. 4 C; the IC₅₀s for VHR and PTP1B were 45 and 3200 μm, respectively. We also found that 40 and 80 μm Cpd 5 lacked any significant inhibitory activity against recombinant human mitogen-activated protein kinase (data not shown).

The inhibition of Cdc25B₂ was dependent on the length of enzyme exposure to Cpd 5; a 30-min preincubation with 2μ m Cpd 5 caused almost 50% more inhibition in enzyme activity than in samples that were exposed to Cpd 5 at the time of substrate addition (Fig. 5,A). Preincubation longer than 30 min did not produce greater inhibition, possibly because Cpd 5 became inactivated. No reduction in enzyme activity was seen when Cdc25B₂ was preincubated with 0.5% DMSO for 90 min or less. The time-dependent inhibition was irreversible; a 90-min incubation in Cpd 5-free buffer did not restore the lost enzyme activity (Fig. 5,B). Similar results were seen with both Cdc25A and VHR (data not shown). A kinetic analysis of the inhibition indicated a partial competitive inhibition for full-length human Cdc25A, Cdc25B₂, and Cdc25C(Fig. 6). The K_is for Cdc25A,Cdc25B₂, and Cdc25C were 15, 1.7, and 1.3μ m, respectively.

One of the putative, endogenous, cellular substrates for both Cdc25B and Cdc25C is the mitotic inhibitor Cdc2, which must be dephosphorylated to allow entry into mitosis (14, 22, 28). Thus, we reasoned that an effective Cdc25 inhibitor would not only cause a G₂-M cell cycle block but would also prevent Cdc2 dephosphorylation. We, therefore, performed Western blotting on tsFT210 cell extracts to determine the Cdc2 phosphorylation levels in the presence or absence of Cpd 5. Protein lysates of tsFT210 cells arrested at the G₂-M boundary were harvested and analyzed by SDS-PAGE. Approximately 50% of Cdc2 was in the mitotic-inactive, hyperphosphorylated form, as reflected by a slower migrating Cdc2 (Fig. 7,A). The phosphorylation of Cdc2 decreased gradually after cells were released from G₂-M block, and most of the Cdc2 was dephosphorylated 6 h after G₂-M release, even in the presence of the DMSO vehicle (Fig. 7,A). When we incubated cells with 1 μm nocodazole,which caused a G₂-M arrest, no hyperphosphorylation of Cdc2 was seen (Fig. 7,B), consistent with its proposed inhibitory activity after Cdc2 activation. In contrast, Cdc2 dephosphorylation was partially blocked (∼70%) with 10 μm Cpd 5 and completely blocked (94%) with 20 μm (Fig. 7,C). The Cdc2 phosphorylation status after 6 h with DMSO alone was similar in Fig. 7, A and C. SC-ααδ9 at 50μ m also caused hyperphosphorylation of Cdc2(Fig. 7,B). Because the phosphorylation status of Cdc2 determines its enzymatic activity (29), we examined the kinase activity of immunoprecipitated Cdc2 by measuring histone H1 phosphorylation in vitro. We found that the Cdc2 kinase activity in cells treated with 1 μm nocodazole was significantly increased, which was consistent with a previous study using tsFT210 cells (26). The Cdc2 kinase activity in cells treated with 10–20 μm Cpd 5 was markedly reduced (Fig. 7,D). The congeners, Cpd 16 and Cpd 22,however, did not block this kinase activity as expected by their lack of effect on Cdc25 activity (Fig. 4,A). Similar amounts of Cdc2 were immunoprecipitated in cells treated with Cpd 5 (Fig. 7 E).

Cdk4 plays a central role in regulating the G₁transition by its association with cyclin D1 (30). This complex remains inactive until Cdc25A dephosphorylates it. Cdk2 is also involved in regulating the G₁-S transition by its association with cyclin E or cyclin A. The Cdk2/cyclin E complex has been shown to be dephosphorylated at Thr-14 and Tyr-15 and, thereby,activated by Cdc25A treatment in vitro (31). To clarify the mechanism of G₁ cell cycle block by Cpd 5, we treated tsFT210 cells with ≤20 μmCpd 5 for 6 h, immunoprecipitated Cdk2 or Cdk4 from the cell lysates, and then determined tyrosine phosphorylation by Western blotting, using anti-phosphotyrosine monoclonal antibody. As illustrated in Fig. 8, the phosphorylation of both Cdk2 and Cdk4 increased after Cpd 5 treatment. We confirmed that there was equivalent loading of Cdk2 or Cdk4 with anti-Cdk2 or anti-Cdk4 antibody, respectively (Fig. 8). To quantify the phosphorylation level of Cdk2 or Cdk4, we determined the intensity of the bands by densitometer and calculated a phosphorylation level as described in “Materials and Methods.” Both Cdk2 and Cdk4 tyrosine phosphorylation increased in a concentration-dependent manner after Cpd 5 treatment, with a >5-fold increase being seen after exposure to 20 μm Cpd 5.

Cdk requires noncovalent interactions with cyclins to be functional. To exclude the possibility that Cpd 5 simply blocked such an intracellular interaction, we treated tsFT210 cells with Cpd 5 for 6 h,immunoprecipitated Cdk2 from cell lysates with an anti-Cdk2 antibody,and then examined the immunoprecipitate for cyclin A and E content by Western blotting. We also immunoprecipitated Cdk4 from cell lysates with anti-Cdk4 and determined cyclin D1 protein levels. Cyclin A or E association with Cdk2 was unchanged after Cpd 5 treatment (Fig. 9,A). Similarly, Cdk4 association with cyclin D1 was unaffected by the Cpd 5 treatment (Fig. 9,B). For both analyses, we loaded equivalent amounts of Cdk2 or Cdk4 as detected with anti-Cdk2 or anti-Cdk4, respectively (Fig. 9).

The phosphorylation of Rb, which is a critical regulator of the G₁ checkpoint, is controlled in part by Cdk2. Thus, we examined the phosphorylation status of Rb in synchronous tsFT210 cells at various times after addition of 20 μmCpd 5 (Fig. 10). As expected, the Rb phosphorylation increased with passage into G₁ phase (Fig. 10). Within 1.5 h after exposure of cells to Cpd 5, however, there was a marked inhibition of Rb phosphorylation with no alteration of Rb protein levels. Equivalent loading was confirmed by measuring GAPDH (Fig. 10). Thus, our results support the hypothesis that Cpd 5 blocked cell cycle progression through the G₁ checkpoint by disruption of functional Cdk activity through inhibition of Cdc25A activity.

To ensure that the inhibition of Cdk4 kinase activity and cell cycle arrest were not secondary to p53 induction or increased Cdk inhibitors, we measured p53, p21, and p16 levels in tsFT210 cells after Cpd 5 treatment (Fig. 11). tsFT210 cells, which had been treated with an equitoxic etoposide concentration, displayed elevated p53 levels, whereas Cpd 5 produced no increase (Fig. 11,A). We also saw no increase in p21 or p16 with Cpd 5 (Fig. 11, B and C), suggesting that the dual cell cycle phase arrest was not due simply to nonspecific cell stress or DNA damage.

The Cdc25 dual-specific phosphatases have an essential role in controlling cell proliferation by regulating the activities of Cdks(14, 31). In higher eukaryotes, Cdc25A is responsible for governing G₁ transition into S phase, Cdc25B probably initiates cell cycle movement through the G2 phase, and Cdc25C is required for entry into mitosis, because of its ability to dephosphorylate and activate Cdc2. Because Cdc25A and Cdc25B have also been reported to be oncogenic (15) and to be overexpressed in several tumor types (17, 18), Cdc25 is an attractive therapeutic target. Although the Cdc25 family members appear to have distinct biological functions and possibly substrates, the amino acids comprising their active site HC(X₅)R region are identical, suggesting that inhibitors with specificity to all three Cdc25s are feasible. Moreover, significant structural differences exist among the other protein tyrosine phosphatases and Cdc25(32). Thus, it may be possible to identify selective inhibitors of this family of enzymes.

Except for the widely used broad-spectrum protein phosphatase inhibitor vanadate (33), few dual-specificity protein phosphatase inhibitors have been reported (34, 35). Moreover, these analogues are generally in limited supply, and the effects of these compounds on cell cycle transition or other enzymes are not known. We have previously synthesized and evaluated a small molecule,SC-ααδ9, that was among the most potent of the known synthetic inhibitors of the Cdc25 dual-specificity phosphatases(23). As noted in our current studies and elsewhere(27), this competitive inhibitor of Cdc25 caused both G₂-M and G₁ inhibition.

Previously, we reported that the thioether vitamin K analogue Cpd 5 was a more potent inhibitor of hepatoma cell proliferation than other K vitamins (8). Hepatoma cells normally only arrest in G₁. Moreover, we found that growth-inhibitory concentrations of Cpd 5 caused a rapid increase in protein tyrosine phosphorylation that could be blocked by elevating intracellular stores of thiols, such as cysteine (36, 39). Although we proposed sulfhydryl arylation of protein phosphatases as a potential mechanism for enhanced phosphorylation and growth inhibition, no experimental examination of the effects of Cpd 5 on specific phosphatases was performed previously. We now report that Cpd 5 inhibited Cdc25 in a partially competitive manner that was time dependent and ultimately irreversible. The Cdc25 enzymes share a conserved COOH-terminal catalytic domain containing the Cys-(X)₅-Arg motif. In Cdc25A and presumably other Cdc25 enzymes, Cys-430 forms a disulfide bond with Cys-384 that may be self-inhibiting and redox sensitive (32). In contrast to other K vitamins, however,Cpd 5 lacks significant redox activity (8). Thus, we hypothesize that arylation, possibly of Cys-430 or Cys-384, is responsible for the enzyme inhibition. This will require additional experimental studies to establish.

We used the well-studied tsFT210 cell system, because the cells can be synchronized without any exogenous agents or drugs. Both Cpd 16 and Cpd 22, which are close congeners of Cpd 5, failed to block cell cycle progression. Neither Cpd 16 nor Cpd 22 had inhibitory activity against Cdc25 phosphatases in vitro. By contrast, Cpd 5 inhibited both cell cycle progression and Cdc25 phosphatase activity in vitro. These data revealed a close correlation between Cdc25 inhibition in vitro and disruption of cell cycle regulation. The observed elevated Cdc2 phosphorylation and the loss of Cdc2 kinase activity provided additional biochemical evidence for intracellular Cdc25 inhibition. Cpd 5, but not the other closely related but biochemically inactive analogues, decreased Cdc2 kinase activity in the intact cells. These concentration-response studies showing that Cpd 5 induced Cdc2 phosphorylation and inhibited its kinase activity suggest that Cpd 5 had an inhibitory effect on Cdc25B and Cdc25C within the cell and provide a mechanistic basis for the blockage at G₂-M.

We hypothesized that inhibition of Cdc25A might mediate the G₁ block caused by Cpd 5, because Cdc25A seems to be important for entry into S phase (13, 31). The tyrosine phosphorylation status of both Cdk2 and Cdk4 was markedly increased by the actions of Cpd 5. Both of these Cdks have a central role in regulating the G₁-S transition (30). Cdc25A dephosphorylates Cdk2 at Thr-14 and Tyr-15 and activates the functional Cdk2/cyclin E complex required for progression through the S phase of the cell cycle (31). Cdc25A also controls the tyrosine phosphorylation status of Cdk4, which regulates G₁ arrest by agents such as UV irradiation(37). Furthermore, the activity of Cdc25A determines the phosphorylation status of Rb through its effects on Cdk4 kinase. Our data show that Cpd 5 increased Cdk4 tyrosine phosphorylation, thereby decreasing kinase activity against Rb. Thus, the dephosphorylated Rb might cause the G₁ block in tsFT210.

We cannot, however, formally exclude that Cpd 5 acts on other cell cycle control mechanisms or on other protein phosphatases. Indeed, in hepatoma cells, Cpd 5 transiently enhanced the phosphorylation of a number of proteins (36, 39). Nonetheless, we found that differences exist in the in vitro sensitivity of several classes of protein phosphatases and that Cpd 5 induced persistent inhibition of one class of protein phosphatases, i.e.,Cdc25. Furthermore, we have established that cell cycle arrest resulted from the cellular effects of Cpd 5 consistent with intracellular inhibition of the catalytic activity of Cdc25. It is well established that DNA damage, such as that induced by ionizing radiation, produces a p53 induction and blocks the cell cycle at both G₁ and G₂-M(38). Our results indicate, however, that exposure of tsFT210 cells to Cpd 5 for 6 h did not produce p53, p21, or p16 induction (Fig. 11). These data suggest that the main pathway causing the dual cell cycle arrest by Cpd 5 is different from p53 (p21) or p16 induction pathways.

In summary, we demonstrated that the potent K vitamin analogue, Cpd 5,inhibited an important class of growth-regulatory, dual-specificity phosphatases and arrested cells in both G₁ and G₂-M phases. We suggest that small molecule inhibitors derived from the Cpd 5 pharmacophore will be useful for furthering our understanding of the role of Cdc25 in regulating G₁ and G₂ transition and may contribute to a further development of novel anticancer agents.

We thank Andreas Vogt, Alexander P. Ducruet, Angela Wang, and the other members of the Lazo Laboratory for their comments and scientific support. We also thank Professor Peter Wipf and members of his laboratory for synthesizing SC-ααδ9 and Professor Michio Yamakido at Hiroshima University for his assistance. This report is dedicated to the memory of our respected colleague and friend, the late Professor Paul Dowd, who first synthesized the enzyme inhibitor naphthoquinone.