A Small Molecule Inhibitor of Isoprenylcysteine Carboxymethyltransferase Induces Autophagic Cell Death in PC3 Prostate Cancer Cells*♦

A number of proteins involved in cell growth control, including members of the Ras family of GTPases, are modified at their C terminus by a three-step posttranslational process termed prenylation. The enzyme isoprenylcysteine carboxylmethyl-transferase (Icmt) catalyzes the last step in this process, and genetic and pharmacological suppression of Icmt activity significantly impacts on cell growth and oncogenesis. Screening of a diverse chemical library led to the identification of a specific small molecule inhibitor of Icmt, cysmethynil, that inhibited growth factor signaling and tumorigenesis in an in vitro cancer cell model (Winter-Vann, A. M., Baron, R. A., Wong, W., dela Cruz, J., York, J. D., Gooden, D. M., Bergo, M. O., Young, S. G., Toone, E. J., and Casey, P. J. (2005) Proc. Natl. Acad. Sci. U. S. A. 102, 4336–4341). To further evaluate the mechanisms through which this Icmt inhibitor impacts on cancer cells, we developed both in vitro and in vivo models utilizing PC3 prostate cancer cells. Treatment of these cells with cysmethynil resulted in both an accumulation of cells in the G1 phase and cell death. Treatment of mice harboring PC3 cell-derived xenograft tumors with cysmethynil resulted in markedly reduced tumor size. Analysis of cell death pathways unexpectedly showed minimal impact of cysmethynil treatment on apoptosis; rather, drug treatment significantly enhanced autophagy and autophagic cell death. Cysmethynil-treated cells displayed reduced mammalian target of rapamycin (mTOR) signaling, providing a potential mechanism for the excessive autophagy as well as G1 cell cycle arrest observed. These results identify a novel mechanism for the antitumor activity of Icmt inhibition. Further, the dual effects of cell death and cell cycle arrest by cysmethynil treatment strengthen the rationale for targeting Icmt in cancer chemotherapy.

Posttranslational processing of so-called CaaX proteins has received much attention in the past two decades due to the important roles these proteins play in biological regulations and diseases (1,2). This processing is initiated by isoprenoid modification of the cysteine residue of the C-terminal CAAX motif of the protein, subsequent proteolytic removal of the three C-terminal amino acids, i.e. the ϪAAX residues, and the methylation of the newly exposed carboxyl group of the prenylated cysteine residue. The overall process, termed protein prenylation, has been shown to be important for the localization, stability, and ultimate functions of a broad array of CaaX proteins (3).
Most members of the Ras superfamily of GTPases are CaaX proteins, and Ras proteins themselves, which are farnesylated, have been extensively studied due to the high prevalence of dysregulated Ras signaling in human cancers (4). Inhibitors of protein farnesyltransferase (FTase) 2 have been under development as anticancer agents for over a decade, but their efficacy, especially in solid tumors, has been disappointing (5,6). The realization that some CaaX proteins, including forms of Ras in which mutations are prevalent in human tumors, are subject to alternative prenylation by protein geranylgeranyltransferase I when FTase is inhibited (7) spurred efforts to target the postprenylation processing steps of proteolysis and methylation since each of these steps is catalyzed by a single enzyme that acts on both farnesylated and geranylgeranylated proteins (8,9). In particular, targeting of CaaX protein methylation via inhibition of the enzyme responsible, isoprenylcysteine carboxylmethyltransferase (Icmt), through both genetic and pharmacological approaches, has been shown to dramatically impair oncogenesis in several tumor cell models (10,11).
The mechanism(s) through which inhibition of Icmt impacts on cell proliferation and oncogenesis are still far from clear. Interference with cell cycle progression, however, is a cornerstone of many chemotherapeutic agents, and both FTase inhibition and geranylgeranyltransferase I inhibition have been * This work was supported by the Agency for Science, Technology and demonstrated to arrest many types of tumor cells at G 1 phase of the cell cycle; FTase inhibitors also trigger a G 2 /M arrest in certain cell types (5,6,12). Another important property of cancer chemotherapeutic agents is the ability to induce cell death. The process of apoptosis in particular has been widely studied in this regard, and many current anticancer agents, including CaaX prenylation inhibitors, enhance apoptosis in cells (6,13). Quite recently, autophagic cell death has stepped into the spotlight as a type of programmed cell death with the potential to be enhanced by cancer therapeutics (14,15). As with many biological regulatory processes, autophagy seems to be a doubleedged sword in terms of an impact on cell functions. Most cell types display a baseline level of autophagy to clear damaged organelles and unwanted proteins, but dysregulation of the autophagy process can be detrimental to cell survival. Consequently, manipulation of autophagy is now considered to present therapeutic opportunities in several disease states, including cancer (16).
We recently reported the identification of a specific small molecule inhibitor of Icmt, cysmethynil, and demonstrated a mechanism-based impact on tumorigenesis in an in vitro cancer cell model (11). We now report that treatment of PC3 prostate cancer cells with cysmethynil impairs cell cycle progression and, unexpectedly, activates an autophagic process in the cells and promotes autophagy-dependent cell death. Further, treatment of mice harboring xenograft tumors with cysmethynil results in dramatic impairment of tumor growth. These results identify a novel mechanism for the antitumor effects of Icmt inhibition and strengthen the support for targeting Icmt in cancer chemotherapy.

EXPERIMENTAL PROCEDURES
Materials-Cysmethynil and biotin S-farnesylcysteine were synthesized by the Duke Small Molecule Synthesis Facility via established methods (11,17). Cysmethynil analog 1-octyl-mtolyl-1H-indole (J3) was synthesized via standard chemical procedures and characterized to confirm identity and purity (see the supplemental text and scheme). Stock solutions were prepared at 10 mM in DMSO and stored at Ϫ20°C. Antibodies recognizing cyclins D1 and B1, p27, poly(ADP-ribose) polymerase-2, caspase 3, eukaryotic initiation factor 4E-binding protein 1 (4EBP1), and GAPDH were all from Cell Signaling. The LC3 antibody was from Abgent.
Cell Culture and Proliferation Assays-The PC3 human prostate cell line was obtained from American Type Culture Collection (Rockville, MD). Cells were maintained at 37°C with 5% CO 2 in DMEM (Sigma) supplemented with 10% fetal bovine serum (Hyclone), 50 units/ml penicillin (Invitrogen), and 50 g/ml streptomycin (Invitrogen). For proliferation assays, cells were seeded at 15-20% confluency in DMEM containing 5% fetal bovine serum in 96-well plates for 24 h prior to treatment with specific agents (e.g. cysmethynil) or vehicle at the concentrations and length of time indicated in the legends for Figs. 1-2. The relative number of the live cells was determined using the CellTiter 96 AQueous One Solution cell proliferation assay (Promega). Each condition was performed in quadruplicate, and data presented that obtained from at least three separate experiments. For proliferation studies performed with 3-methyladenine (3-MA) in addition to cysmethynil, cells were seeded as above and incubated with 0.5 mM 3-MA or vehicle at 37°C overnight. Cells were then treated with the concentrations of cysmethynil as indicated in the legend for Fig. 1 in the continued presence of 0.5 mM of 3-MA for the indicated durations.
Flow Cytometry Analysis-PC3 cells (5 ϫ 10 5 ) were seeded in 100-mm dishes in DMEM containing 5% fetal bovine serum and incubated for 24 h, whereupon the cells were exposed to cysmethynil, the J3 analog, or vehicle at the concentration and for the time indicated in the appropriate figure legend. Cells were harvested by centrifugation at 300 ϫ g for 5 min, whereupon the cells were then fixed in ice cold 70% ethanol overnight before being resuspended in phosphate-buffered saline containing 40 g/ml propidium iodide and 0.1 mg/ml Ribonuclease A (both from Sigma) for 1 h at 37°C. Fluorescence was measured by flow cytometry analysis using an Excalibur Instrument (BD Biosciences).
Immunofluorescence of LC3-Cells subjected to the indicated treatments were fixed with 4% paraformaldehyde and permeabilized with 0.3% Triton using a standard protocol (18). Incubation with primary antibody to LC3 (MAP1LC3B, Abgent) was performed at 4°C overnight before incubation with Rhodamine Red-X secondary antibody (Jackson ImmunoResearch Laboratories). Analysis was performed using an Olympus fluorescent microscope fitted with the appropriate excitation and emission filters.
Acridine Orange Staining for Autophagy Detection-Acridine orange (Sigma) staining was performed according to a published protocol (19). Briefly, cells were washed twice with phosphate-buffered saline and then stained with 1 M/ml acridine orange for 15 min at 37°C. Analysis was performed via fluorescence microscopy using 490-nm band-pass blue excitation filters and a 515-nm long-pass barrier filter. In the study of the effect of bafilomycin co-treatment, cells were treated with 200 nM bafilomycin A1 for 40 min prior to the addition of acridine orange.
Atg5 Knockdown-Stealth siRNA duplex oligoribonucleotides targeting Atg5 (Invitrogen) were resuspended to make a 20 M solution following the manufacturer's instructions. Transfections were carried out using the Lipofectamine 2000 protocol provided by Invitrogen. Conditions were optimized with varying ratios of Lipofectamine and RNA, as well as different time intervals after the transfections as determined by immunoblot analysis of atg5 protein levels. Cell proliferation on both the atg5 siRNA-treated cells and the mock-transfected cells was then assessed using the assay described above.
Cysmethynil Treatment of Xenograft-harboring Mice-PC3 cells were grown in DMEM and 10% fetal calf serum until near confluence and then harvested with a standard method using trypsin. Cells (5 ϫ 106) were then mixed with Matrigel (BD Biosciences) to achieve 40% Matrigel in the final solution. The cell preparation was then injected subcutaneously into the flanks of 6 -8-week-old SCID mice. For treatment, cysmethynil was prepared in 4% DMSO, 1.4% Tween 80, and 1% sodium carboxymethyl cellulose (Sigma) normal saline solution; the vehicle control was with the same mixture lacking cysmethynil. In a preliminary acute toxicity study, mice were injected intra-peritoneally with vehicle control, 0.1, 0.2, 0.3, 0.6, and 1.0 mg/g of cysmethynil. The control, the 0.1 mg/g-, and the 0.2 mg/ginjected mice showed few signs of distress after the injection. The 0.3 mg/g and higher dosings resulted in sluggishness and diarrhea in the first 24 h after injection, although all recovered 2 days after injection. Based on the acute toxicity study, cysmethynil was dosed at 0.1 and 0.2 mg/g in two groups, together with the control group, at 48-h intervals. The animals were monitored for their general appearance and health, as well as body weight. Subcutaneous tumors were measured with the standard clipper ruler method. Final tumor weight at the end of the study was documented after animal sacrifice and dissection of tumor tissue.
Miscellaneous Procedures-Icmt activity was determined by the in vitro assay described previously (17,20). Briefly, the assay was carried out using recombinant Icmt produced in Sf9 cells, biotin S-farnesylcysteine as the prenylcysteine substrate, and [ 3 H]S-adenosylmethionine. Apoptosis was assessed by determination of activation of caspase 3 and poly(ADP-ribose) polymerase-2 through immunoblot analysis, DNA fragmentation assay, and microscopic observation of the abnormalities of nuclei stained with 4Ј,6-diamidino-2-phenylindole as described (21). To perform immunoblot analysis, cells subjected to the indicated treatment were harvested and lysed, and protein concentration was determined by Micro BCA protein assay (Pierce). Proteins were separated by 14% SDS-PAGE, and subsequent immunoblot procedures were performed using an enhanced chemiluminescence procedure (GE Healthcare) per the manufacturer's instructions.

Cysmethynil Exhibits Mechanism-based Antiproliferative
Activity toward Prostate Cancer Cells-To facilitate a rapid evaluation of the mechanism-based consequences of cysmethynil treatment on cells, we sought a structural analog of the compound that lacked activity toward the enzyme. Molecular modeling of available structure-activity data on the chemical series from which cysmethynil was identified (22) suggested that the amide portion of the indole ring might be important in this activity. An analog lacking only the amide portion was synthesized, termed J3 (Fig. 1A), and evaluated for in vitro activity toward Icmt. This analysis, shown in Fig. 1B, revealed that the J3 analog was essentially devoid of Icmt inhibitory activity despite its chemical similarity to cysmethynil.
The impact of treatment, by cysmethynil and the inactive J3 analog, on the prostate cancer-derived cell line PC3 was evaluated using a cell viability assay. Cysmethynil treatment resulted in a dose-and time-dependent reduction in the number of viable PC3 cells, whereas the J3 analog at the highest corresponding dose was ineffective (Fig. 1C). These data, along with a previous study using a colon cancer cell line in which the antiproliferative activity of cysmethynil was markedly diminished by overexpression of Icmt in the cells (11), provide compelling evidence that the impact of cysmethynil treatment on cells is directly due to inhibition of Icmt. An additional finding from this study is that at moderate dosage, cysmethynil appears to exhibit primarily cytostatic activity, whereas at higher con-centrations, both cytostatic and cytotoxic activity are observed (Fig. 1C).
Cysmethynil Is Efficacious in Controlling Tumor Growth in a Xenograft Mouse Model of Prostate Cancer-To investigate the ability of cysmethynil to inhibit tumor growth in vivo, we first conducted a dose escalation toxicity trial in Balb/C mice and found that an intraperitoneal dose of less than 0.3 mg/g was well tolerated. We then established a xenograft model of PC3 cells in SCID mice. Tumor cells were implanted subcutaneously in the flank. When tumors started to increase in size (usually 100 -200 mm 3 ), confirming the successful grafting of the tumor cells, mice were randomly assigned to control (vehicle) and cysmethynil treatment groups. Mice thus bearing established PC3 tumors were given intraperitoneal injections of vehicle only, 0.1 or 0.2 mg/g of cysmethynil every 48 h for 28 days. The duration of the experiments were dictated by the tumor burden of the control mice. As shown in Fig. 2A, both doses of cysmethynil significantly suppressed the growth of PC3 tumors when com- pared with the vehicle control group. Neither dose was accompanied by any appreciable toxicity as assessed by body weight determinations of the treated mice when compared with the control group (Fig. 2B). These data indicate that pharmacological inhibition of Icmt in vivo significantly impacts on tumor growth and further reinforces the notion of Icmt as an anticancer drug target.
Cysmethynil Treatment Induces Cell Cycle Arrest in PC3 cells-As noted in the Introduction, there is considerable evidence that inhibition of CaaX protein processing by either of the protein prenyltransferases can impact on cell cycle progression. This information, coupled with the finding that moderate doses of cysmethynil have apparently predominant cytostatic activity on PC3 cells (Fig. 1C), prompted us to examine more closely cell cycle parameters in cells treated with this Icmt inhibitor. Flow cytometry analysis of PC3 cells treated 48 h with 20 M cysmethynil showed a significantly increased population of cells in G 1 (Fig. 3A). Further examination of molecular markers associated with G 1 arrest showed remarkable changes, including increased p27 levels, reduced cyclin D1 levels, and showed almost complete loss of phospho-Rb (Fig. 3B). These data are all consistent with the G 1 arrest observed in the flow cytometry analysis.
Cysmethynil Treatment Induces Autophagic Cell Death-Although the ability of cysmethynil treatment to trigger a G 1 arrest in PC3 cells could account for the cytostatic capacity of this compound, the reduction in cell count following longer term and higher dose treatments in vitro suggested that an increase in cell death was also being triggered by Icmt inhibition. Our initial set of experiments to examine this potential outcome of cysmethynil treatment was focused on apoptotic pathways since as noted above, inhibition of CaaX protein processing had been linked to elevated apoptosis in many cells types (6). However, no consistent impact of cysmethynil treatment at the concentration of cysmethynil that decreased the number of viable cells was observed on apoptotic markers such as caspase 3 and poly(ADP-ribose) polymerase-2 cleavage in the treated PC3 cells (data not shown). Additionally, neither DNA fragmentation nor abnormal nuclear morphology was observed in PC3 cells following cysmethynil treatment at the concentration that resulted in cell death (not shown), arguing against a significant contribution of apoptosis to diminished cell viability.
The inability to link apoptosis to the cell death observed in the PC3 cells following cysmethynil treatment prompted us to consider whether the drug induced a non-apoptotic form of cell death. Specifically, we examined autophagy, a process that involves the degradation of cellular components through an autophagosome-lysosome pathway, as this process has recently become appreciated as important in cell growth and survival    14). Indeed, cysmethynil treatment of PC3 cells resulted in a dramatic elevation of the LC3-II protein (Fig. 4A), which is characteristic for the activation of autophagic pathway (23). Further cell-based analysis employing anti-LC3 immunofluorescence revealed that the LC3 protein in drug-treated cells both increased in total abundance and also aggregated into granular/vesicular structures (Fig. 4B), consistent with the expected changes that signal autophagosome formation (23).
Since different cell types exhibit varying degrees of autophagocytosis at baseline (14), we quantified the autophagosomes in the cysmethynil-and vehicle-treated cells by determination of the fraction of cells exhibiting elevated level of LC3 aggregation; this quantitation showed that cysmethynil treatment significantly elevated cellular abundance of autophagosomes (35% versus 2%, Fig. 4B).
The process of autophagy starts with the autophagosome formation and then progresses to autophagolysosomes through the fusion of acidic lysosomes with autophagosomes (15). Therefore, acridine orange staining of the live cells was also employed to visualize acidic autophagolysosomes in control and cysmethynil-treated PC3 cells. As shown in Fig. 4C, cysmethynil treatment markedly elevated the amount of autophagolysosomes in the cells, providing further evidence that the autophagic process was being activated by drug treatment and that the autophagosome formation is not the result of the inhibition of lysosomal fusion.
Bafilomycin A1, an inhibitor of vacuolar Hϩ ATPase, prevents the transition of autophagosome to autophagolysosomes by disrupting the fusion of autophagosomes to lysosomes (24). Hence, bafilomycin A1 provides a useful way of studying the autophagy process. The treatment of the PC3 cells with bafilomycin A1 markedly reduced the quantity of acridine orangepositive vesicles (Fig. 4C), confirming that the autophagosomes produced by the treatment of cysmethynil undergo the same maturation process with the fusion with lysosomes. Indeed, in the bafilomycin A1-treated cells, LC3-II levels remain elevated despite a marked reduction of acidic vesicles (not shown), indicating that the earlier stages of autophagy prior to lysosomal fusion were not affected by this lysosome fusion inhibitor.
Inhibition of Autophagy Protects PC3 Cells from Cell Death Elicited by Cysmethynil Treatment-To assess whether the induction of autophagy observed in cysmethynil-treated PC3 cells actually contributed to the cell death elicited by treatment with the compound, we first assessed whether 3-MA, a known inhibitor of autophagy that acts through inhibition of type 3 PI3 kinase (25), could alleviate cysmethynil-induced cell death. PC3 cells were treated with vehicle alone, cysmethynil, 3-MA alone, or 3-MA plus cysmethynil, and viability of the cells was assessed 72 h later. As seen in Fig. 5A, 3-MA treatment alone had little impact on cell viability; the viability of cysmethynil-treated cells was markedly increased if 3-MA was present during the course of the treatment. Immunoblot analysis showed that the cells treated with both 3-MA and cysmethynil had much lower levels of the autophagy marker LC3-II when compared with the cells treated with cysmethynil alone, suggesting that the autophagic process stimulated by cysmethynil is subjected to the regulation by type 3 PI 3-kinase (Fig. 5A).
We also employed a knockdown strategy to impair autophagy to further assess its impact on cysmethynil-induced cell death. RNA interference-mediated knockdown of atg5, a crucial component of the autophagy cascade (14), markedly reduced cell death elicited by cysmethynil treatment (Fig. 5B). This has provided additional evidence supporting a crucial role for autophagy-dependent cell death in the efficacy of cysmethynil. Knockdown of atg5 also resulted in the reduction of LC3-II production in cysmethynil-treated cells (Fig. 5B), confirm- ing a direct impact of the knockdown on the autophagic process in the cells. Taken together, the survival benefits achieved with both 3-MA treatment and atg5 RNA knockdown in the cysmethynil-treated PC3 cells provide compel-ling evidence that cysmethynil not only induces autophagy but that the accompanying autophagy-dependent cell death contributes significantly to the efficacy of cysmethynil in inducing cancer cell death.
Cysmethynil Treatment Impacts on mTOR Signaling in PC3 Cells-The data presented above indicate that cysmethynil treatment induces both G 1 cell cycle arrest and autophagy; a common link of these two processes is that they can be modulated by mTOR signaling. Two types of CaaX proteins are known to be important in regulating mTOR signaling, Ras GTPases and the Rheb GTPase. Ras activates PI 3-kinase, leading to the activation of Akt, which in turn activates Rheb by inhibiting tuberous sclerosis complex, a negative regulator of Rheb. Rheb positively regulates mTOR kinase, with a resultant positive impact on cell cycle progression and negative impact on autophagy (26). Inhibition of Ras methylation has been shown to impair Ras activity (11); Rheb could also be affected by Icmt inhibition, and this could further impact mTOR signaling. Indeed, cysmethynil treatment of PC3 cells resulted in a marked reduction of phosphorylation of Akt (Fig. 6A), as would be expected from inhibition of Ras. In addition, phosphorylation of the mTOR downstream effector ribosomal protein S6 was markedly reduced in cysmethynil-treated cells, and another effector, 4EBP1, showed the characteristic shift in phosphorylation pattern consistent with inhibition of the mTOR kinase (Fig. 6A). Furthermore, Rheb levels in PC3 cells were markedly reduced following 48 h of cysmethynil treatment, providing evidence that the general availability and activity of Rheb GTPase were reduced. Based on these data, we propose a model whereby inhibition of Icmt leads to a reduction of Ras and Rheb activity and consequent inhibition of Akt and mTOR signaling, contributing to the effects on cell cycle progression and autophagy (Fig. 6B).

DISCUSSION
Autophagy as a means of induction of cancer cell death has attracted increasing attention recently, especially in circumstances of apoptosis-resistant cancer cells. For example, the  mTOR inhibitor rapamycin has been reported to induce cell death through autophagy in glioblastoma cell lines that are resistant to many therapies that induce apoptosis (27). In this regard as well, a recent study showed that temozolomide treatment alone of glioblastoma cell lines causes autophagic cell death (28). Although the realization that autophagic cell death may be an important component of the action of certain cancer drugs is relatively new, the connection between the PI3K/Akt/ mTOR inhibition and autophagy induction has been well established (14).
The PI3K/Akt/mTOR signaling pathway activation enhances cell proliferation, survival, and growth and inhibits autophagy in many cells, and aberrant up-regulation of the PI3K/Akt/ mTOR axis has been frequently linked to oncogenesis in many cancers (29). In many cases, up-regulation of the pathway has been linked to loss of the phosphatase and tensin homolog (PTEN) phosphatase. Although these aberrancies do contribute to resistance to therapies many cancers, they might also render the cancers more sensitive to mTOR inhibition, as the sustained activation would render the cells more dependent on this pathway for proliferation (21,30). On this note, the PC3 prostate cancer cell used in the current studies is PTEN-null and exhibits elevated intrinsic PI3K/Akt/mTOR signaling, which likely contributes to the resistance of this cell line to a variety of therapeutic interventions. The efficient induction of autophagic cell death by cysmethynil suggests that the inhibition of the CaaX protein processing in this group of more resistant cancers could provide an effective therapeutic alternative, or enhancement of, existing chemotherapies.
Targeting mTOR signaling by cysmethynil makes use of two fundamental mechanisms of treating cancer, induction of cell death (through autophagy) and inhibition of cell cycle progression. Inhibiting mTOR blocks the phosphorylation of two key downstream effectors, p70S6 kinase and 4EBP1. Both proteins play important roles in translational regulation; in particular, inhibition of expression of the G 1 cell cycle regulatory protein cyclin D1 leads to G 1 arrest in cells in which mTOR is inhibited (30). The ability of cysmethynil to induce G 1 arrest as well as autophagy could make it broadly applicable as an anticancer agent.
The identification of cysmethynil as a specific and bioavailable Icmt inhibitor provides a selective tool to probe the involvement of Icmt in both normal and pathological cellular processes. The current study not only strengthens the rationale for targeting Icmt as an anticancer strategy from a mechanistic standpoint but also demonstrates significant in vivo efficacy of cysmethynil through its administration to mice bearing xenograft prostate cancer-derived tumors. Our data clearly indicate that an induction of autophagy by cysmethynil is a major contributor to the cell death that accompanies pharmacological inhibition of Icmt. Although the specific CaaX protein(s) underlying this phenomenon have not yet been unam-biguously identified, the Ras and Rheb GTPases are potential players due to their abilities to control PI 3-kinase and mTOR signaling.