Inhibition of the 3-Hydroxy-3-methylglutaryl-coenzyme A Reductase Pathway Induces p53-independent Transcriptional Regulation of p21 WAF1/CIP1 in Human Prostate Carcinoma Cells*

Progression through the cell cycle is controlled by the induction of cyclins and the activation of cognate cyclin-dependent kinases. The 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase inhibitor lovastatin induces growth arrest and cell death in certain cancer cell types. We have pursued the mechanism of growth arrest in PC-3-M cells, a p53-null human prostate carcinoma cell line. Lovastatin treatment increased protein and mRNA levels of the cyclin-dependent kinase inhibitor p21 WAF1/CIP1 , increased binding of p21 with Cdk2, markedly inhibited cyclin E- and Cdk2-associated phosphorylation of histone H1 or GST-retinoblastoma protein, enhanced binding of the retinoblastoma protein to the transcription factor E2F-1 in vivo, and induced the activation of a p21 promoter reporter construct. By using p21 promoter deletion constructs, the lovastatin-responsive element was mapped to a region between −93 and −64 relative to the transcription start site. Promoter mutation analysis indicated that the lovastatin-responsive site coincided with the previously identified transforming growth factor-β-responsive element. These data indicate that in human prostate carcinoma cells an inhibitor of the HMG-CoA reductase pathway can circumvent the loss of wild-type p53 function and induce critical downstream regulatory events leading to transcriptional activation of p21.

hibiting Cdk catalytic activity and by inhibiting proliferating cell nuclear antigen, which is required for DNA synthesis (10). By inhibiting the activity of G 1 cyclin-Cdk complexes, p21 blocks phosphorylation of the retinoblastoma protein (pRB). This leaves pRB in its active, hypophosphorylated form, in which it is able to induce G 1 arrest through negative regulation of the heterodimeric transcription factor E2F/DP (11)(12)(13).
We have found that human prostate carcinoma cells are very sensitive to the induction of growth arrest and cell death by the HMG-CoA reductase inhibitor lovastatin (14). 2 Lovastatin blocks the rate-limiting step in the cholesterol-synthesis pathway, the formation of mevalonic acid from HMG-CoA (15)(16)(17). The study presented here investigates the mechanism by which lovastatin induces cell cycle arrest in the androgen-independent prostate carcinoma line PC-3-M, which is lacking both p53 alleles (18). The data show that lovastatin treatment caused a marked shift of pRB to its underphosphorylated form and enhanced the in vivo binding of pRB and E2F-1. In addition, lovastatin inhibited Cdk activity, and this inhibition was associated with elevated protein and RNA expression of the Cdk inhibitor p21, similar to the induction of p21 by lovastatin in human breast cells (19). We investigated the mechanism of lovastatin-induced p21 expression using p21 promoter-reporter assays. Lovastatin treatment induced transcriptional activation of the p21 promoter as indicated by increased activity of a p21 promoter-luciferase reporter construct, and this transcriptional response did not require the p53 response element. All effects of lovastatin were completely reversed by addition of mevalonic acid. These data demonstrate a link between the mevalonate pathway and activation of the promoter of the critical cell cycle-regulatory protein p21. To begin to understand the mechanism of lovastatin-induced transcription, promoter deletion and mutation studies were performed. This analysis demonstrated that lovastatin responsiveness is mediated by a promoter-proximal region that coincides with the recently identified transforming growth factor-␤ (TGF-␤)-responsive element.

Cells-
The human prostate carcinoma cell line PC-3-M was a gift of Dr. James Kozlowski, Department of Urology, Northwestern University Medical School (20). Cells were cultured as described previously (21).
Antibodies-Polyclonal antibodies to cyclin E, Cdk2, Cdk4, and monoclonal anti-E2F-1 (clone 20) were obtained from Santa Cruz Biotechnology. Monoclonal anti-pRB antibody (clone G3-245) and monoclonal anti-p21 antibody were purchased from PharMingen. * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Cell Cycle Analysis-Cells were fixed in 50% ethanol and incubated with RNase A and the DNA intercalating dye propidium iodide, and cell cycle phase analysis was performed by flow cytometry using a Becton Dickinson FACStar flow cytometer and Becton Dickinson Cell Fit software.
Immunoprecipitation and Western Blot Analysis-Immunoprecipitations were performed as described (22). Cell lysates were incubated with anti-Cdk2 polyclonal antibody, anti-E2F-1, or normal rabbit serum. The immunocomplex was collected on protein A-Sepharose (Sigma) and analyzed by SDS-polyacrylamide gel electrophoresis using enhanced chemiluminescence detection (Amersham Pharmacia Biotech). Western blot analysis was performed as described (23).
Northern Blot Analysis-Total RNA was isolated from cells lysed in guanidinium thiocyanate solution, and Northern blot analysis was carried out as described previously (24) using 32 P-labeled (ICN) cDNA probes and 30 g of total RNA per lane. The human p21 DNA probe was a gift of Dr. David Givol.
Immune Complex Kinase Assay-Immune complex kinase assays were performed as described (22). Briefly, cell lysates were incubated with primary antibody, and immune complexes were collected on protein A-Sepharose beads and resuspended in kinase assay mixture containing [␥-32 P]ATP (ICN) and GST-RB or histone H1 (Life Technologies, Inc.) as substrate. The GST-RB fusion protein containing amino acids 379 -928 of human pRB (25) was a gift of Dr. Frederic Kaye, NCI, National Institutes of Health. The proteins were separated on 10% SDS-polyacrylamide gels, and bands were detected by autoradiography.
p21 Promoter-Luciferase Constructs-Drs. Bert Vogelstein and Todd Waldman, The Johns Hopkins Oncology Center, provided the fulllength 2.4-kb human p21 promoter-chloramphenicol acetyltransferase constructs as well as p21 promoter deletion constructs, in which 300 base pairs were progressively removed from the 3Ј terminus. Three constructs were tested as follows: a 2.4-kb p21 promoter fragment containing the p21 transcription initiation start site at its 3Ј end and a p53 response element at the 5Ј end, a 2.1-kb fragment consisting of the 2.4-kb fragment from which 300 base pairs, including the p53 response element, had been deleted from the 5Ј end, and a 0.3-kb fragment consisting of the 300 base pairs most proximal to the transcription start site (9). The original chloramphenicol acetyltransferase constructs were digested with HindIII and XhoI and subcloned into the HindIII-XhoI sites of the luciferase reporter vector pGL2-basic (Promega Corp.), to create p21 promoter-luciferase constructs. The deletion constructs p21P 93-S (0 to Ϫ93), p21P Sma ⌬1 (0 to Ϫ63) p21P Sma ⌬2 (full-length p21 promoter from which the region Ϫ64 to Ϫ110 was deleted) and p21P (0 to Ϫ100), and the p21P 93-S mutants were created as described previously (26).
Luciferase Assay-PC-3-M cells were co-transfected with the p21 promoter-luciferase reporter constructs and with the ␤-galactosidase reporter vector pCMV␤ (Promega) using LipofectAMINE (Life Technologies, Inc.), as recommended by the manufacturer. Following transfection the cells were incubated for 6 h; the medium was exchanged, and the cells were incubated for an additional 36 h in the presence or absence of 10 M lovastatin. The cells were then lysed, and luciferase activity in the lysates was assayed using a Dynatech ML1000 luminometer (Dynatech Laboratories). Luciferase activity was normalized to ␤-galactosidase activity which was assayed using the ␤-galactosidase Enzyme Assay System (Promega).

Lovastatin Induces G 1 Arrest and Cell Death-24 h after 10
M lovastatin addition there was an increase in the percent of PC-3-M cells in G 1 . By 48 h following lovastatin addition greater than 90% of the PC-3-M cells were arrested in G 1 (Table I). Cell viability was consistently greater than 75% in cells treated for 48 h, after which there was a marked loss of viability, culminating in cell death (data not shown).
Loss of Phosphorylated pRB in Cells Exposed to Lovastatin-In cells treated with lovastatin pRB appeared as a single band on Western blot analysis, migrating in the position characteristic of unphosphorylated or hypophosphorylated pRB (27) (Fig. 1A). The position of pRB after lovastatin coincided with the position of pRB detected with an antibody specific for the un-or hypophosphorylated form (Pharmingen clone G99 -549, data not shown).
Lovastatin Enhances Association of pRB and E2F-1-Association of pRB with the transcription factor E2F-1 was almost undetectable by co-immunoprecipitation analysis of the untreated log phase cells (Fig. 1B). However, 36 h after addition of lovastatin there was a strong increase in the association of pRB and E2F-1.
Levels of Cdk2, Cdk4, and Cyclin E Protein Are Unchanged following Lovastatin Treatment-The G 1 /S transition is regulated by D-type cyclins which bind to and activate Cdk4 and Cdk6 and cyclin E, which activates Cdk2. Because lovastatin arrests PC-3-M cells in G 1 , we tested whether lovastatin decreased protein levels of Cdk4, Cdk2, or cyclin E, and as demonstrated in Fig. 2A, there was little detectectable change in the level of these proteins, even after 48 h of exposure to lovastatin. It should be noted, however, that the actual effects of lovastatin on these proteins may be larger than is apparent in the Western blot because of the nonlinearity of the chemiluminescence detection system.
Lovastatin Inhibits Cdk2 and Cyclin E-associated Kinase Activity-Cdk2-and cyclin E-associated kinase activity were reduced to undetectable levels after 36 h treatment with lovastatin (Fig. 2B). These results indicated that the inhibition of Cdk2-and cyclin E-associated kinase activity might involve binding of a Cdk inhibitor protein to the Cdk2-cyclin E complex.
Lovastatin Induces p21 Protein and Association of p21 with Cdk2-Western blot analysis showed that lovastatin induced a marked increase in the level of the Cdk inhibitor p21 (Fig. 3A). We next asked whether the p21 protein induced by lovastatin treatment was associated with Cdk2 in vivo. As shown in Fig.  3B, p21 was detected in anti-Cdk2 immunoprecipitates from lovastatin-treated cells and not in immunoprecipitates from  untreated cells. Thus lovastatin induced the association of p21 with Cdk2, suggesting that lovastatin-induced p21 blocks Cdk2 kinase activity by direct binding to the cyclin-Cdk2 complex.
Lovastatin Induces p21 mRNA-To examine the mechanism by which lovastatin induces p21 we performed Northern blot analysis of the effect of lovastatin on the level of p21 mRNA. The 2.4-kb p21 transcript was increased in PC-3-M cells as early as 12 h after lovastatin treatment (Fig. 4). The kinetics of induction were consistent with the time course of inhibition of Cdk activity. The increase in p21 mRNA induced by lovastatin was blocked by actinomycin D (data not shown), suggesting that lovastatin may be regulating p21 mRNA levels through an effect on p21 transcription. p53 acts at the transcriptional level to increase p21 expression in response to DNA damage (9). In PC-3-M cells the p53 gene has been reported to be truncated and p53 protein absent (18). As a control we performed Western and Northern blot analysis for p53 and found no evidence of p53 expression in control or lovastatin-treated cells (data not shown). These data demonstrate that lovastatin is regulating the level of p21 mRNA through a p53-independent mechanism.
Transcriptional Activation of the p21 Promoter by Lovastatin-To test for transcriptional regulation of p21 by lovastatin p21 promoter-luciferase reporter constructs were transfected into PC-3-M cells, and luciferase activity was measured in the presence and absence of lovastatin. The full-length 2.4-kb promoter construct containing a p53 response element was activated 3-4-fold by lovastatin (Fig. 5A). Activation was not seen with a control vector consisting of the pGL2-basic luciferase reporter vector alone, and activation of the p21 promoter by lovastatin was completely blocked by incubation of the cells with 5 mM mevalonic acid, demonstrating that the ability of lovastatin to activate the p21 promoter is mediated by a component of the HMG-CoA reductase pathway (data not shown).
A 2.1-kb construct lacking the p53 response element was activated by lovastatin to the same extent as the full-length promoter (Fig. 5A). Constructs consisting of 300 base pairs and 100 base pairs proximal to the transcription start site were fully activated by lovastatin when compared with the 2.4-kb promoter. Thus, the lovastatin-responsive site is localized to the promoter proximal region between 0 and Ϫ100 base pairs. Three deletion constructs were employed to further map this region of the p21 promoter. A construct consisting of 0 to Ϫ93 base pairs was fully responsive to lovastatin (Fig. 5B). In contrast, a construct consisting of the region from 0 to Ϫ63 was non-responsive to lovastatin. A deletion construct consisting of the full-length promoter minus the region Ϫ64 to Ϫ110 was also non-responsive to lovastatin. These data localize the region responsive to lovastatin to the sequences between Ϫ64 and Ϫ93 base pairs.
The region between Ϫ64 and Ϫ93 contains two Sp1 binding sites. The second of these sites, between Ϫ78 and Ϫ84, has been identified as both the TGF-␤-responsive element (T␤RE) and the butyrate-responsive element (28). To test whether the lovastatin-responsive element coincides with this Sp1 binding site, we tested two constructs of the 0 to Ϫ93 region in which the Sp1 site was mutated. Mutant 1, containing 10 consecutive mutated base pairs, did not respond to lovastatin (Fig. 6). Mutation of bases Ϫ78 and Ϫ79 also resulted in loss of lovastatin responsiveness. These data localize the lovastatin-responsive region and suggest that Sp1 may play a critical role in the activation of p21 transcription in response to lovastatin.  4. Lovastatin induces expression of p21 mRNA. Cells were incubated with 10 M lovastatin for the times indicated. Total RNA was prepared, and p21 mRNA was detected by Northern blot analysis. The same blot was probed with a cDNA to glyceraldehyde-3Ј-phosphate dehydrogenase (GAPDH) to demonstrate comparable levels of mRNA per lane. The fold increase in p21 mRNA level was determined by integrating the signal using NIH Image version 1.6 software.

DISCUSSION
The hormone refractory prostate cancer cell line studied here represents the most advanced and least treatable form of adult carcinoma (29,30). Progressive malignant tranformation in epithelial tumors has been associated with a variety of genetic changes (31). Recent data suggest that mutations in cell cycle genes are the most common genetic change in tumor cells (32). Although inactivation or deletion of the RB gene itself is limited to a subset of human tumors (33), it has been suggested that continued molecular characterization may reveal that virtually all tumors harbor a mutation in one of the genes responsible for regulating G 1 progression (32) and that the final common target of the genetic changes in G 1 regulatory proteins is the functional inactivation of pRB (13).
Passage from G 1 to S phase is regulated by D-type cyclins and their cognate kinases Cdk4 and Cdk6, which act by phosphorylating and inactivating pRB prior to the restriction point at which cells commit to DNA synthesis (34), and by cyclin E-Cdk2, which acts later in G 1 (35). Genetic changes associated with G 1 cyclins and their Cdks in human cancers include overexpression of cyclin D1, overexpression or mutation of Cdk4, loss of expression of the cyclin D-directed Cdk inhibitor p16, mutation or deletion of p53 and RB, and overexpression of cyclin E (36 -39). The data presented here demonstrate that in the human androgen-independent prostate carcinoma cell line PC-3-M, the mevalonate pathway inhibitor lovastatin induces G 1 arrest, blocks pRB phosphorylation, and increases in vivo binding of pRB to E2F-1, consistent with restoration of the growth-regulatory activity of pRB. Furthermore, lovastatin blocks cyclin E-Cdk2 kinase activity, in association with enhanced expression of the Cdk inhibitor p21.
Lovastatin has been shown previously to increase the protein level of the Cdk inhibitor p27 KIP1 in HeLa cells, HL60 cells, and normal fibroblasts (40,41). There was no effect on p27 KIP1 in the PC-3-M cells (data not shown), and thus G 1 arrest appears to be independent of p27 KIP1 in the prostate cancer cells studied.
Inhibition of the mevalonate pathway has many ramifications including loss of the protein isoprenylation modifications farnesylation and geranylgeranylation, loss of sterol synthesis, and potential perturbation of N-linked glycosylation through inhibition of dolichol synthesis (42). To begin to map the target of lovastatin action within the mevalonate pathway, we incubated PC-3-M cells with the squalene synthase inhibitor zaragozic acid (43). We found that zaragozic acid treatment had no effect on p21 levels or cell cycle distribution (data not shown). Because lovastatin effects on the cell cycle were completely reversible by mevalonate and not mimicked by zaragozic acid, the critical point in the mevalonate pathway appears to be between HMG-CoA reductase and squalene synthase. One possible mechanism of lovastatin action is through Ras inactivation due to inhibition of Ras isoprenylation. Evidence suggesting that Ras may not be the mediator of the lovastatin response was presented by De Clue et al. (44) who showed that lovastatin arrests the growth of Ras-transformed cells even when the transforming Ras protein utilizes a myristate-dependent rather than farnesyl-dependent membrane-targeting moiety, and thus is not dependent on the mevalonate pathway. These data do not, however, eliminate Ras inactivation as the mechanism of lovastatin-induced G 1 arrest, and further experimental approaches, including studies with specific farnesyltransferase inhibitors, will be required.
Although lovastatin itself may not be the optimal HMG-CoA reductase inhibitor for cancer treatment, it has been shown to induce growth arrest and cell death in tumor cells in vitro and in athymic mice in vivo (44 -47). A phase I study of lovastatin was completed recently at the National Cancer Institute, in which it was demonstrated that doses as high as 25 times the standard anti-hypercholesterolemic dose were well tolerated (48). More precise identification of the target of lovastatin action would aid in developing new anticancer drugs that act through perturbation of the HMG-CoA reductase pathway.
In addition to p53, recent studies have demonstrated activation of the p21 promoter by TGF-␤ (49, 50), vitamin D 3 (51), phorbol ester, okadaic acid (52-54), STAT1 (55), and butyrate (28). Activation by vitamin D 3 and STAT1 occurs more than 500 base pairs upstream of the lovastatin-responsive region (51,55). In the U937 myeloid leukemia cell line, it has been demonstrated that activation of p21 transcription by phorbol esters and okadaic acid is mediated by the promoter region between Ϫ154 and ϩ16 and that Sp1 plays a critical role in this response. In K562, SW480, and HepG2 cells the activation of p21 transcription by phorbol ester was mapped to the region between Ϫ95 and Ϫ121, and AP2 was shown to act as a positive regulator of the phorbol ester response. The T␤RE is located between Ϫ71 and Ϫ86 relative to the p21 transcription start site (49,50). We found that the lovastatin response region maps between Ϫ64 and Ϫ93. Furthermore, mutation of the T␤RE abrogated lovastatin-induced p21 promoter activation. Thus the lovastatin response site corresponds with the T␤RE. One possible mechanism for the induction of p21 by lovastatin in PC-3-M cells would be induction of TGF-␤ by lovastatin and activation of p21 transcription by TGF-␤. It is likely, however, that lovastatin effects on p21 are unrelated to TGF-␤ because PC-3-M cells have an attenuated response to TGF-␤, and direct addition of TGF-␤1 to PC-3-M cells did not activate the p21 promoter reporter construct (data not shown). Recently it was reported that butyrate activates the p21 promoter and that this response maps to the T␤RE (28). Sp1 is required for p21 promoter activation in response to both TGF-␤ and butyrate (28,50). We have shown that mutation of an Sp1 binding site also blocks lovastatin activation of the p21 promoter. Several studies have shown that Sp1 is a critical factor in regulating transcriptional responses to pRB. Sp1 binds to the pRB-responsive element, and Sp1 binding is required and sufficient for promoter activation in response to pRB (56 -58). Although the overall level of pRB protein decreases in PC-3-M cells in response to lovastatin, that decrease is predominantly or entirely in the hyperphosphorylated form ( Fig. 1A and data not shown), and there is a substantial increase in active pRB as assessed by the markedly enhanced E2F-pRB binding in the in vivo coimmunoprecipitation experiment (Fig. 1B). These data suggest that both pRB and Sp1 may play a role in lovastatin-induced activation of the p21 promoter.
The data presented here demonstrate that lovastatin induces p21 transcription in the absence of wild-type p53. The most frequently mutated tumor suppressor in human cancer is p53. Therefore, p53-independent induction of p21 presents an attractive target for anticancer drug development (59). The data presented here suggest that perturbation of the mevalonate pathway may offer a new approach to transcriptional control of this critical negative growth-regulatory gene.