A small-molecule inhibitor of isoprenylcysteine carboxyl methyltransferase with antitumor activity in cancer cells

Ann M. Winter-Vann*, Rudi A. Baron*, Waihay Wong*, June dela Cruz*, John D. York*, David M. Gooden†, Martin O. Bergo‡, Stephen G. Young§, Eric J. Toone†, and Patrick J. Casey*¶

Departments of *Pharmacology and Cancer Biology and †Chemistry, Duke University Medical Center, Durham, NC 27710; ‡Department of Internal Medicine, Sahlgrenska University Hospital, S-413 45 Gothenburg, Sweden; and §Department of Medicine, University of California, Los Angeles, CA 90095

Edited by John A. Glomset, University of Washington, Seattle, WA, and approved February 7, 2005 (received for review November 1, 2004) Many key regulatory proteins, including members of the Ras family proteins have also been implicated in oncogenesis and tumor of GTPases, are modified at their C terminus by a process termed progression, and these proteins most likely require processing via prenylation. This processing is initiated by the addition of an the prenylation pathway for function (2, 15). isoprenoid lipid, and the proteins are further modified by a pro- Both the membrane targeting and the transforming abilities of teolytic event and methylation of the C-terminal prenylcysteine. Ras require processing through the prenylation pathway (16, 17). Although the biological consequences of prenylation have been For this reason, the protein prenyltransferases, most notably characterized extensively, the contributions of prenylcysteine FTase, have been targets of major drug discovery programs for methylation to the functions of the modified proteins are not well the last decade (18, 19). Presently, several FTase inhibitors are understood. This reaction is catalyzed by the enzyme isoprenyl- being evaluated in clinical trials (15, 19). These experimental cysteine carboxyl methyltransferase (Icmt). Recent genetic disrup- agents have shown significant activity in a number of clinical tion studies have provided strong evidence that blocking Icmt trials, but the overall response rates in patients have been less activity has profound consequences on oncogenic transformation. than initially hoped. One possible explanation for this lack of Here, we report the identification of a selective small-molecule efficacy is the process of alternate prenylation that allows some inhibitor of Icmt, 2-[5-(3-methylphenyl)-1-octyl-1H-indol-3-yl]acet- FTase substrates to be modified by geranylgeranyltransferase amide (cysmethynil). Cysmethynil treatment results in inhibition of type I when FTase activity is limiting (20–22). Recent studies cell growth in an Icmt-dependent fashion, demonstrating mecha- using genetic disruption of Icmt have demonstrated that Ras nism-based activity of the compound. Treatment of cancer cells proteins are significantly mislocalized and tumorigenesis is with cysmethynil results in mislocalization of Ras and impaired markedly impaired in cells that lack Icmt (23, 24). After this epidermal growth factor signaling. In a human colon cancer cell discovery, CaaX protein methylation has gained attention as a line, cysmethynil treatment blocks anchorage-independent target in oncogenesis (25). growth, and this effect is reversed by overexpression of Icmt. These With emerging evidence for the importance of Icmt-catalyzed findings provide a compelling rationale for development of Icmt CaaX protein methylation in oncogenesis, there is a clear need inhibitors as another approach to anticancer drug development. for specific pharmacological agents to target this process. How- ever, the only such agents available to date have been analogs of cell transformation ͉ protein isoprenylation ͉ protein methylation ͉ Ras the substrate prenylcysteine or the product S-adenosylhomocys- signaling teine; all of these analogs have been reported to have pleiotropic effects (26–28). Here, we report the discovery of an indole-based C-terminal CaaX motif, where C is cysteine, the a’s are small-molecule inhibitor of Icmt. Treatment of cancer cells with Aaliphatic amino acids, and X can be any of a number of this compound that we have named 2-[5-(3-methylphenyl)-1- amino acids, targets a variety of eukaryotic proteins to a series octyl-1H-indol-3-yl]acetamide (cysmethynil), results in a de- of posttranslational modifications important for their localiza- crease in Ras carboxylmethylation, mislocalization of Ras, and tion and function (1, 2). This processing is initiated by the impaired signaling through Ras pathways. Cysmethynil treat- covalent attachment of a 15-carbon farnesyl or a 20-carbon ment blocks anchorage-independent growth in a human colon geranylgeranyl lipid to the cysteine of the CaaX motif, a reaction cancer cell line, and this effect is reversed by overexpression of catalyzed by protein farnesyltransferase (FTase) or protein Icmt. These findings, together with the findings from genetic geranylgeranyltransferase type I (3). After prenylation, the disruption of this enzyme, suggest that Icmt inhibitors may have C-terminal three amino acids (i.e., the -aaX) are removed by a significant therapeutic potential. specific CaaX protease termed Rce1 (4, 5) and the now C- terminal prenylcysteine is methylated by isoprenylcysteine car- Materials and Methods boxyl methyltransferase (Icmt; refs. 6–8). As polytopic mem- Materials. Farnesyl pyrophosphate was from Biomol, the chem- brane proteins localized to the endoplasmic reticulum, both ical library was from PPD Discovery (Research Triangle Park, Rce1 and Icmt are unusual in their respective classes (9). NC), streptavidin-Sepharose beads were from Amersham Phar- Proteins that terminate in a –CaaX motif regulate a number macia, puromycin and S-adenosylmethionine (AdoMet) were of pathways important in oncogenesis. The best studied example from Sigma, and S-(5Ј-adenosyl)-L-homocysteine was from is the central role of the Ras family of proteins in growth factor activation of the MAP kinase signaling cascade (10, 11). Con-

stitutive activation of this pathway is transforming in a wide This paper was submitted directly (Track II) to the PNAS office. variety of cell types, and activating mutations in Ras have been Abbreviations: Icmt, isoprenylcysteine carboxyl methyltransferase; cysmethynil, 2-[5-(3- found in almost 30% of all cancers, including 50% of colon methylphenyl)-1-octyl-1H-indol-3-yl]acetamide; Rce1, CaaX protease; AdoMet, S-adeno- cancers and up to 90% of pancreatic cancers (12). In addition, sylmethionine; BFC, biotin-S-farnesyl-L-cysteine; MAPK, mitogen-activated protein kinase; many cancers contain alterations upstream of Ras, and the MDCK, Madin–Darby canine kidney; FTase, farnesyltransferase. resultant hyperactivation of Ras is thought to contribute to ¶To whom correspondence should be addressed. E-mail: [email protected]. tumorigenesis in these cancers as well (13, 14). Many other CaaX © 2005 by The National Academy of Sciences of the USA

4336–4341 ͉ PNAS ͉ March 22, 2005 ͉ vol. 102 ͉ no. 12 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0408107102 Downloaded by guest on September 30, 2021 Fluka. Epidermal growth factor was from EMD Bioscience. Background absorbance from blank wells containing only media [methyl-3H]methionine and [methyl-3H]AdoMet were from with compound or vehicle were subtracted from each test well. PerkinElmer. CellTiter 96 Aqueous One solution cell prolifer- ation assay was from Promega. pEGFP and pLPCX were from Localization of GFP Proteins in MDCK Cells. MDCK cells stably Clontech. Sf9 membranes containing recombinant Rce1 and expressing GFP-H-Ras, GFP-K-Ras, or GFP-N-Ras were grown Icmt, termed Rce1 membranes and Icmt membranes, respec- on 35-mm coverslips treated with poly-D-lysine. MDCK cells tively, were made in our laboratory as described (5). Farnesy- expressing Yes-GFP were prepared by transient transfection of lated K-Ras was made by in vitro modification of bacterially the Yes-GFP construct (30), using Superfect reagent (Qiagen, expressed K-Ras with purified FTase as described (5). Biotin- Valencia, CA) following the manufacturer’s instructions. Cells S-farnesyl-L-cysteine (BFC) was synthesized as detailed else- were grown in media containing 10% FBS to Ϸ25% confluence where (R.A.B. and P.J.C., unpublished work). Mouse embryonic and then treated with 1% DMSO or cysmethynil at the indicated fibroblast cell lines were established and maintained as described concentrations. Cells were imaged 72 h after drug treatment (24 (29). Madin–Darby canine kidney (MDCK) cells stably express- h for transiently transfected cells) on an Olympus inverted ing GFP-H-Ras, GFP-K-Ras and GFP-N-Ras were a generous microscope with an UltraView spinning-disk confocal gift of M. Philips (New York University Medical School, New (PerkinElmer LAS) and a krypton͞argon laser with a 488-nm York). line, attached to a cooled charge-coupled device camera (Hamamatsu). Initial image acquisition and manipulation was Icmt Assay: Screening. The small-molecule screen was performed performed with METAMORPH software (Universal Imaging, in 96-well multiscreen filtration plates (Millipore) by using a Downingtown, PA). Beckman Biomek FX robot. Briefly, Sf9 membranes containing Rce1 and Icmt were suspended in 100 mM Hepes, pH 7.4͞5mM Phosphoprotein Analysis. Cells were grown for three days in media MgCl2 (Buffer A) such that 40 ␮l of suspension had 1 ␮g of Rce1 containing 1% FBS with cysmethynil or vehicle as indicated. membrane protein and 0.2 ␮g of Icmt membrane protein; Half of the wells were then treated with EGF (10 ng͞ml) and half protein values are of total membrane protein. This membrane with vehicle for 10 min, whereupon cells were rinsed with PBS suspension was dispensed into wells, to which was added the and harvested. Total cell lysates (30 ␮g protein) were resolved library compounds in DMSO (1.3 ␮l) to yield a final concen- on 4–20% Tris-glycine gels (Invitrogen). Proteins were trans- tration of 15–30 ␮M. After 15 min at room temperature, 10 ␮l ferred to nitrocellulose and probed with a mix of anti-phospho- of buffer A containing 13 ␮M[3H]AdoMet (5 Ci͞mmol; 1 Ci ϭ Akt and anti-phospho-p42͞44 mitogen-activated protein kinase 37 GBq) and 5 ␮M farnesylated K-Ras were added. After (MAPK) antibodies, or with anti-tubulin or anti-phospho NF␬B another 30 min, the reaction was quenched with 100 ␮lof6% antibody (Cell Signaling Technology, Beverly, MA) as indicated. SDS͞45% trichloroacetic acid (TCA). Precipitated proteins Visualization was performed with alkaline phosphatase (Pro- were recovered on the membranes of the plate wells by vacuum mega) as per the manufacturer’s instructions. filtration of the entire plate, and the wells were washed three times with 200 ␮l of 6% TCA. Scintillation fluid [Microscint 20 Generation of Stable Cell Lines Expressing GFP or GFP-Icmt. Full- (Packard), 50 ␮l per plate per well] was added, the plates were length human ICMT was cloned into pEGFP after restriction sealed as per manufacturer’s instructions, and radioactivity in endonuclease digestion with BamHI and XhoI. Retroviral con- each well was determined by using a Packard TopCount NXT structs were generated by cloning EGFP or GFP-ICMT into the microplate scintillation counter. pLPCX retroviral vector. These constructs, along with the helper plasmid 467, were transfected into human embryonic kidney 293 Icmt Assay: Secondary Analysis. Secondary assays and kinetic cells. Virus was harvested 48 h after transfection and used to analyses were by adding recombinant Icmt (0.5 ␮gofSf9 infect DKOB8 cells. Cells were treated with virus for 24 h, membrane protein) to an assay mixture containing 4 ␮M BFC, allowed to recover for another 24 h, selected in 0.5 ␮g͞ml 5 ␮M[3H]AdoMet (1.2 Ci͞mmol), and either inhibitor or puromycin for Ϸ3 weeks, and then sorted for expression of GFP DMSO in a total volume of 45 ␮l of Buffer A. Reactions were on a Becton Dickinson FACSVantageSE cell sorter. GFP- incubated for 20 min at 37°C, terminated with 5 ␮l of 10% Tween positive cells selected in this manner were cultured as per normal 20, and then 10 ␮l of streptavidin beads in 500 ␮lof20mM DKOB8 cells (31). NaH2PO4,pH7.4͞150 mM NaCl (Buffer B) were added. The interaction between biotin and streptavidin was allowed to Soft Agar Growth Assay. Soft agar culture media was prepared proceed overnight at 4°C under gentle agitation, whereupon the with 10% FBS in 1ϫ minimum essential medium ␣. Bottom agar beads were harvested by centrifugation, washed three times with (0.5 ml 2.4% noble agar in 1.5 ml of soft agar culture media) was 0.5 ml of buffer B, and resuspended in 100 ␮l of the same buffer plated in each cell culture dish. Cells were harvested by for radioactivity determination. In the experiments to measure trypsinization at Ϸ80% confluence, mixed into top agar (10,000 potential time-dependent inhibition of Icmt, the same assay was cells per plate, 0.3% noble agar in soft agar culture media) and used except that the Icmt membrane suspension was first mixed poured onto prepared plates. Cysmethynil or DMSO was in- with inhibitor in buffer A; this solution was incubated for 30 min cluded at the indicated concentrations in both the top and at 37°C, whereupon the remaining components of the reaction bottom agar layers. For each condition, triplicate samples were mixture were added and the reactions were incubated an addi- prepared. Plates were fed twice a week with 300 ␮lof1ϫ MEM tional 20 min at 37°C before product isolation and radioactivity containing 10% FBS and the appropriate concentrations of

determination. cysmethynil or DMSO. After 3 weeks, plates were stained by CELL BIOLOGY addition of 300 ␮lof10mg͞ml 3-(4,5-dimethylthiazol-2-yl)-2,5- Cell Growth Determination. Cell growth assays were performed in diphenyl tetrazolium bromide (Sigma-Aldrich) in PBS followed a 96-well plate format. Briefly, Ϸ1,000 cells were plated into each by incubation at 37°C in an incubator with 5% CO2 for 3 h. Plates well of the plate. After 24 h, the media were replaced with media were then treated with 0.4 M HCl in 300 ␮l of isopropyl alcohol containing either compound or vehicle. Media and drug were and incubated overnight before imaging. replaced every 24 h. Cell determinations were made at the times indicated by adding 19 ␮l of CellTiter 96 Aqueous One solution Results to each well followed by incubating the plates in the dark at 37°C Identification of an Indole-Based Selective Inhibitor of Icmt. To for 2 h, after which the absorbance at 490 nm was read. identify small-molecule inhibitors of Icmt, we screened a diverse

Winter-Vann et al. PNAS ͉ March 22, 2005 ͉ vol. 102 ͉ no. 12 ͉ 4337 Downloaded by guest on September 30, 2021 Fig. 1. Cysmethynil, a small-molecule inhibitor of Icmt. (A) Structure of the indole-based compound, cysmethynil. (B) Time-dependent inhibition of Icmt by cysmethynil. Icmt activity was measured as the incorporation of [3H] from [3H]AdoMet into the Icmt substrate BFC. The assay was performed either with (E) or without (F) preincubating Icmt with cysmethynil as described in Mate- rials and Methods.

chemical library of Ϸ10,000 compounds. The library contained 70ϩ subfamilies derived from unique scaffolds. We used an in vitro screen in which Icmt activity was measured as the incor- poration of a [3H]methyl group into a farnesylated, Rce1- proteolyzed, K-Ras substrate (see Materials and Methods). Com- pounds that showed Ͼ50% inhibition at 50 ␮M were subjected Fig. 2. Icmt-dependent growth inhibition of mouse embryonic fibroblasts by to a secondary screen by using a small-molecule substrate of using cysmethynil. Mouse embryonic fibroblasts grown in media containing Icmt, BFC. From this screen, we identified a group of com- 8% serum were treated with DMSO (■) or cysmethynil at concentrations of 15 pounds with an indole core structure that had significant activity (‚), 20 (), or 30 (F) ␮M. Media and drug were replaced daily, and cell growth was monitored for 6 days as described in Materials and Methods. Data against Icmt. The most potent of these compounds was what we ϩ/ϩ Ϫ/Ϫ term cysmethynil (Fig. 1A). This compound was independently represent the mean and SD of four replicate wells. (A) Icmt cells. (B) Icmt cells. (C) IcmtϪ/Ϫ cells stably transfected with a gene expressing human ICMT synthesized and characterized to confirm identity and purity (see Ϫ Ϫ (Icmt / ͞ICMT cells). Supporting Text and Scheme 1, which are published as supporting information on the PNAS web site), and all studies described below were performed by using the independently synthesized from the gene-disruption studies (32). Reasoning that cells that compound. had adapted to grow in the absence of Icmt activity should be In the initial in vitro assay using BFC as the prenylcysteine resistant to the effects of the inhibitor, we treated IcmtϪ/Ϫ mouse substrate, the IC50 for Icmt inhibition by cysmethynil was determined to be 2.4 ␮M (Fig. 1B). In this assay, the substrates embryonic fibroblasts and matched wild-type cells with increas- and the inhibitor were premixed, and the reaction was initiated ing concentrations of the compound and monitored cell growth by the addition of enzyme. However, when the enzyme was for 6 days (Fig. 2). Treatment with cysmethynil resulted in a premixed with inhibitor and AdoMet for 15 min before initiation dose-dependent inhibition of growth wild-type cells (Fig. 2A), of the reaction with BFC, a dramatic increase in inhibitor but IcmtϪ/Ϫ cells were largely unaffected (Fig. 2B). Further- potency was observed with a measured IC50 of Ͻ200 nM (Fig. more, when the human ICMT gene was stably expressed in 1B). These data suggest that cysmethynil is a time-dependent IcmtϪ/Ϫ cells, the reconstituted cell line regained sensitivity to inhibitor of Icmt. Importantly, even at concentrations up to 50 cysmethynil (Fig. 2C). These results provide strong evidence for ␮ M, cysmethynil did not inhibit the other enzymes in the an antiproliferative activity of cysmethynil that is mechanism- prenylation pathway (FTase, geranylgeranyltransferase type I, based, i.e., directly due to an impact on Icmt activity. and Rce1), nor did it inhibit an AdoMet-dependent DNA methyltransferase or an unrelated protein methyltransferase (the SssI DNA methyltransferase and PCMT1 protein methyl- Cysmethynil Treatment of Cells Results in Mislocalization of Ras and transferase, respectively) (data not shown). Impairment of Growth Factor Signaling. Carboxylmethylation is important for proper plasma membrane localization of Ras (24). Cysmethynil Treatment Impacts Cell Growth in an Icmt-Dependent Based on this observation, we predicted that treatment of cells Fashion. To evaluate the potential cellular activity of cysmethynil, with an Icmt inhibitor would lead to a loss of Ras from the we took advantage of a cell model of Icmt deficiency developed plasma membrane. To test this hypothesis, MDCK cells stably

4338 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0408107102 Winter-Vann et al. Downloaded by guest on September 30, 2021 GFP-N-Ras in these cells and for all three Ras isoforms ex- pressed in mouse embryonic fibroblasts (data not shown). As a control, MDCK cells expressing a fusion of GFP to the N terminus of Yes, a protein kinase localized to the plasma membrane by N-terminal myristoylation and palmitoylation (30), were treated under the same conditions. Cysmethynil treatment did not affect the plasma membrane localization of Yes-GFP (Fig. 3 e–h), indicating that the compound does not globally disrupt trafficking to the plasma membrane. Growth factor signaling to MAPK involves CaaX proteins, again most notably Ras, and inhibition of CaaX protein meth- ylation has been reported to impair EGF-mediated phosphory- lation and activation of MAPK (29, 33). To evaluate the effect of cysmethynil treatment on EGF-mediated activation of MAPK and other signaling proteins, DKOB8 colon cancer cells were grown under low serum (1%) conditions in the presence of either vehicle or 1 ␮M cysmethynil. After 3 days of treatment, the cells were treated with either EGF or additional vehicle. Cell lysates were separated on SDS͞PAGE and immunoblotted with a mixture of phospho-specific antibodies for Akt and p42͞44 MAPK, or with an antibody for ␤-tubulin. As shown in Fig. 3 Lower, the level of activated Akt increased Ϸ3-fold, and that of activated p42͞44 MAPK nearly 10-fold in EGF-treated cells. The EGF-induced increase in MAPK was almost completely blocked by cysmethynil treatment, whereas the increase in Akt phosphorylation was partially but not completely attenuated. This observation supports the hypothesis that cysmethynil treat- ment impacts on signaling through Ras-dependent pathways, because the activation of MAPK by EGF occurs primarily via the Ras pathway; corresponding activation of Akt involves both Ras-independent and Ras-dependent processes (13). Also, it is interesting to note that under low (1%) serum conditions, cysmethynil affects cellular processes at 5- to 10-fold lower concentrations than when the cells are grown in higher (8–10%) serum, suggesting that this compound, like many pharmacolog- ical agents, is buffered by serum.

Impact of Cysmethynil Treatment on the Transformed Phenotype of Colon Cancer Cells. The data detailed above all point to a potential impact of Icmt inhibition on blocking the transformed pheno- type of cancer cells. One of the classic methods to assess transformation of cells is by measuring their ability to grow in soft agar (34), and genetic disruption of Icmt in cells has been shown to block mouse anchorage-independent growth triggered by activated Ras (23). To directly assess the role of Icmt in any ability of cysmethynil to block the transformed phenotype of cancer cells, we first set out to engineer the DKOB8 colon cancer cells to stably overexpress Icmt; conferring resistance to a pharmacological agent by overexpression of the target is a classic Fig. 3. Impact of cysmethynil treatment on Ras localization and signaling. means to confirm the mechanism of action of the agent. Through (Upper) Mislocalization of GFP Ras in cysmethynil-treated cells. MDCK cells expressing GFP-tagged K-Ras (a–d) or Yes-GFP (e–h) were treated with 1% this approach, we were able to create a line of DKOB8 cells DMSO (a and e)or10(b and f), 20 (c and g), or 30 (d and h) ␮M cysmethynil. expressing GFP-Icmt in which the level of Icmt activity was Live cells were imaged on a confocal microscope as described in Materials and elevated 4-fold compared with a parallel line in which GFP alone Methods.(Lower) Impact of cysmethynil treatment on EGF-stimulated protein was expressed (Fig. 4A). Although modest, this level of Icmt phosphorylation. Wild-type mouse embryonic fibroblasts were grown for 3 overexpression was sufficient to protect the cells from cysmethy- days in media containing 1% serum in the presence of DMSO or 1 ␮m nil blockade of EGF-stimulated MAPK activation (Fig. 4B), cysmethynil as described in Materials and Methods. Where indicated, cells providing further confirmation that this effect of the inhibitor is were treated with EGF for the final 10 min before harvesting. Cell lysates because of its ability to impact on Icmt activity in the cells. containing equal amounts of protein were resolved on a 13% SDS- Having established that overexpression of Icmt conferred CELL BIOLOGY polyacrylamide gel and probed with antiphospho-Akt, antiphospho-p42͞44 MAPK, or anti-␤-tubulin antibody as indicated. resistance to cysmethynil, we then evaluated both cell lines for the effect of cysmethynil on anchorage-independent growth. DKOB8͞GFP and DKOB8͞GFP-Icmt cells growing in soft agar were treated with either vehicle or increasing concentrations of expressing GFP-tagged K-Ras were treated with increasing cysmethynil. After 3 weeks, the plates were stained with 3-(4,5- concentrations of cysmethynil for 72 h before imaging by con- dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide to iden- focal fluorescence microscopy. As shown in Fig. 3 a–d, cys- tify viable cells and imaged. As shown in Fig. 4C, treatment with methynil treatment led to a dose-dependent mislocalization of cysmethynil significantly impaired the ability of the DKOB8͞ GFP-K-Ras. Similar effects were noted for GFP-H-Ras and GFP cells to grow in soft agar, with a concentration of 20 ␮M

Winter-Vann et al. PNAS ͉ March 22, 2005 ͉ vol. 102 ͉ no. 12 ͉ 4339 Downloaded by guest on September 30, 2021 Fig. 4. Overexpression of Icmt rescues EGF-stimulated MAPK activation and anchorage-independent growth in cysmethynil-treated cells. (A) Creation of cell lines stably overexpressing Icmt. DKOB8 cells were engineered to stably express either GFP alone or a GFP–Icmt fusion protein as described in Materials and Methods. Membrane fractions from GFP- and GFP-ICMT-expressing lines were assayed for Icmt activity by using the BFC assay. (B) Overexpression of Icmt restores EGF-stimulated MAPK activation in cysmethynil-treated cells. DKOB8 cells stably expressing GFP or GFP-ICMT were either left untreated or treated with 5 ␮M cysmethynil for 3 days in reduced-serum media as described in Materials and Methods. Where indicated, cells were treated with EGF for the final 10 min before harvesting. Cell lysates containing equal amounts of protein were resolved on a 13% SDS-polyacrylamide gel and probed with antiphospho-p42͞44 MAPK, or as a control, antiphospho NF␬B antibody as indicated. (C) Impact of cysmethynil on anchorage-independent growth of DKOB8 colon cancer cells. DKOB8 cells stably expressing GFP (first row) or GFP-ICMT (second row) were suspended in 0.3% noble agar and plated on a base of 0.6% noble agar; both the top and bottom layers contained either 1% DMSO or 10 or 20 ␮M cysmethynil as indicated. After 3 weeks of growth, colonies were stained with 3-(4,5-dimethylthiazol-2-yl)- 2,5-diphenyl tetrazolium bromide and imaged.

dramatically reducing the ability of cells to form colonies. Even with these limitations, there are exciting hints about the However, elevated Icmt activity was sufficient to rescue the involvement of Icmt in a number of biological systems. Increas- ability of the cells to form colonies in soft agar in the presence ing evidence suggests that Icmt-catalyzed methylation impacts of 20 ␮M cysmethynil (Fig. 4C, second row of wells). These signaling through Ras, and more importantly, that a lack of Icmt results provide compelling evidence that cysmethynil blockade can slow or even stop cellular transformation (23, 29, 33). In of Icmt is responsible for the ability of the compound to impact addition, several studies have linked Icmt inhibition to significant on both growth factor signaling pathways and on anchorage- effects on endothelial cells, including increased permeability and independent growth of these cancer cells. apoptosis (35, 36, 40). Inhibitors of Icmt might therefore have significant utility as anti-cancer agents. In fact, there is evidence Discussion that one existing anti-cancer drug, methotrexate, targets Icmt The functional consequence of carboxylmethylation of CaaX through an elevation of its product inhibitor S-(5Ј-adenosyl)-L- proteins has been a matter of speculation since the prenylation homocysteine (29). pathway was first identified (6, 7). Icmt-catalyzed methylation is Although much of the work on Icmt has centered on the clearly essential for some biologies, as evidenced by the embry- consequences of carboxylmethylation of Ras proteins, some onic lethal phenotype when the gene encoding this enzyme is intriguing findings have been reported for other CaaX proteins disrupted in mice (32). However, the contribution of carboxyl- processed by Icmt. Methylation of RhoA plays a major role in methylation to specific biological processes has been difficult to stability of the protein (23, 41), and the effects of Icmt inhibition address. Although much work has been performed by using on endothelial cells noted above have been suggested to be due prenylcysteine analogs or agents that elevate S-(5Ј-adenosyl)-L- to impact on carboxylmethylation of RhoA in these cells (35, 36). homocysteine to inhibit Icmt activity in cells (29, 35–38), the Outside the family of GTPases, methylation of lamin B clearly nonspecific nature of these compounds has made it difficult to influences its interaction with the nuclear envelope (39). The attribute specific outcomes to an inhibition of Icmt (27, 28). The identification of cysmethynil as an inhibitor of Icmt provides a establishment of cell lines lacking Icmt has greatly helped the selective pharmacological tool to probe the potential functional field (23, 39), but researchers are restricted to these few cell lines. consequences of CaaX protein methylation in cellular systems

4340 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0408107102 Winter-Vann et al. Downloaded by guest on September 30, 2021 and also the involvement of Icmt in biologies that are important Grants GM46372 (to P.J.C.) and AR050200 and HL076839 (to in both normal and pathological cellular processes. S.G.Y.), a Howard Hughes Medical Institute Predoctoral Fellowship (to A.M.W.-V.), and a fellowship from l’Association Pour la Recherche Contre le Cancer (to R.A.B.). M.O.B. was supported by We thank James Otto (Duke University) for recombinant human Rce1 grants from the Swedish Cancer Society and the Swedish Research and Icmt. This work was supported by National Institutes of Health Council.

1. Zhang, F. L. & Casey, P. J. (1996) Annu. Rev. Biochem. 65, 241–269. 23. Bergo, M. O., Gavino, B. J., Hong, C., Beigneux, A. P., McMahon, M., Casey, 2. Kloog, Y. & Cox, A. D. (2004) Semin. Cancer Biol. 14, 253–261. P. J. & Young, S. G. (2004) J. Clin. Invest. 113, 539–550. 3. Casey, P. J. & Seabra, M. C. (1996) J. Biol. Chem. 271, 5289–5292. 24. Bergo, M. O., Leung, G. K., Ambroziak, P., Otto, J. C., Casey, P. J. & Young, 4. Boyartchuk, V. L., Ashby, M. N. & Rine, J. (1997) Science 275, 1796–1800. S. G. (2000) J. Biol. Chem. 275, 17605–17610. 5. Otto, J. C., Kim, E., Young, S. G. & Casey, P. J. (1999) J. Biol. Chem. 274, 25. Clarke, S. & Tamanoi, F. (2004) J. Clin. Invest. 113, 513–515. 8379–8382. 26. Chiang, P. K., Gordon, R. K., Tal, J., Zeng, G. C., Doctor, B. P., Pardhasaradhi, 6. Clarke, S., Vogel, J. P., Deschenes, R. J. & Stock, J. (1988) Proc. Natl. Acad. K. & McCann, P. P. (1996) FASEB J. 10, 471–480. Sci. USA 85, 4643–4637. 27. Ma, Y. T., Shi, Y. Q., Lim, Y. H., McGrail, S. H., Ware, J. A. & Rando, R. R. 7. Hrycyna, C. A., Sapperstein, S. K., Clarke, S. & Michaelis, S. (1991) EMBO J. (1994) Biochemistry 33, 5414–5420. 10, 1699–1709. 28. Scheer, A. & Gierschik, P. (1993) FEBS Lett. 319, 110–114. 8. Dai, Q., Choy, E., Chiu, V., Romano, J., Slivka, S. R., Steitz, S. A., Michaelis, 29. Winter-Vann, A. M., Kamen, B. A., Bergo, M. O., Young, S. G., Melnyk, S., S. & Philips, M. R. (1998) J. Biol. Chem. 273, 15030–15034. James, S. J. & Casey, P. J. (2003) Proc. Natl. Acad. Sci. USA 100, 6529–6534. 9. Young, S. G., Ambroziak, P., Kim, E. & Clarke, S. (2000) in The Enzymes, 30. McCabe, J. B. & Berthiaume, L. G. (1999) Mol. Biol. Cell 10, 3771–3786. eds. Tamanoi, F. & Sigman, D.G. (Academic, San Diego), Vol. 21, pp. 31. Habets, G. G., Knepper, M., Sumortin, J., Choi, Y. J., Sasazuki, T., Shirasawa, 156–213. S. & Bollag, G. (2001) Methods Enzymol. 332, 245–260. 10. Malumbres, M. & Barbacid, M. (2003) Nat. Rev. Cancer 3, 459–465. 32. Bergo, M. O., Leung, G. K., Ambroziak, P., Otto, J. C., Casey, P. J., Gomes, 11. Shields, J. M., Pruitt, K., McFall, A., Shaub, A. & Der, C. J. (2000) Trends Cell A. Q., Seabra, M. C. & Young, S. G. (2001) J. Biol. Chem. 276, 5841–5845. Biol. 10, 147–154. 33. Chiu, V. K., Silletti, J., Dinsell, V., Wiener, H., Loukeris, K., Ou, G., Philips, 12. Bos, J. L. (1989) Cancer Res. 49, 4682–4689. M. R. & Pillinger, M. H. (2003) J. Biol. Chem. 279, 7346–7352. 13. Gschwind, A., Fischer, O. M. & Ullrich, A. (2004) Nat. Rev. Cancer 4, 361–370. 34. Clark, G. J., Cox, A. D., Graham, S. M. & Der, C. J. (1995) Methods Enzymol. 14. Schlessinger, J. (2000) Cell 103, 211–225. 255, 395–412. 15. Doll, R. J., Kirschmeier, P. & Bishop, W. R. (2004) Curr. Opin. Drug Discov. 35. Kramer, K., Harrington, E. O., Lu, Q., Bellas, R., Newton, J., Sheahan, K. L. Devel. 7, 478–486. & Rounds, S. (2003) Mol. Biol. Cell 14, 848–857. 16. Hancock, J. F., Magee, A. I., Childs, J. E. & Marshall, C. J. (1989) Cell 57, 36. Lu, Q., Harrington, E. O., Hai, C. M., Newton, J., Garber, M., Hirase, T. & 1167–1177. Rounds, S. (2004) Circ. Res. 94, 306–315. 17. Kato, K., Cox, A. D., Hisaka, M. M., Graham, S. M., Buss, J. E. & Der, C. J. 37. Roullet, J. B., Xue, H., Chapman, J., McDougal, P., Roullet, C. M. & (1992) Proc. Natl. Acad. Sci. USA 89, 6403–6407. McCarron, D. A. (1996) J. Clin. Invest. 97, 2384–2390. 18. Gibbs, J. B., Oliff, A. & Kohl, N. E. (1994) Cell 77, 175–178. 38. Kowluru, A., Seavey, S. E., Li, G., Sorenson, R. L., Weinhaus, A. J., Nesher, 19. Karp, J. E., Kaufmann, S. H., Adjei, A. A., Lancet, J. E., Wright, J. J. & End, R., Rabaglia, M. E., Vadakekalam, J. & Metz, S. A. (1996) J. Clin. Invest. 98, D. W. (2001) Curr. Opin. Oncol. 13, 470–476. 540–555. 20. James, G. L., Goldstein, J. L. & Brown, M. S. (1995) J. Biol. Chem. 270, 39. Maske, C. P., Hollinshead, M. S., Higbee, N. C., Bergo, M. O., Young, S. G. 6221–6226. & Vaux, D. J. (2003) J. Cell Biol. 162, 1223–1232. 21. Whyte, D. B., Kirschmeier, P., Hockenberry, T. N., Nunez-Oliva, I., James, L., 40. Wang, H., Yoshizumi, M., Lai, K., Tsai, J. C., Perrella, M. A., Haber, E. & Lee, Catino, J. J., Bishop, W. R. & Pai, J. K. (1997) J. Biol. Chem. 272, 14459–14464. M. E. (1997) J. Biol. Chem. 272, 25380–25385. 22. Sebti, S. M. & Der, C. J. (2003) Nat. Rev. Cancer 3, 945–951. 41. Backlund, P. S., Jr. (1997) J. Biol. Chem. 272, 33175–33180. CELL BIOLOGY

Winter-Vann et al. PNAS ͉ March 22, 2005 ͉ vol. 102 ͉ no. 12 ͉ 4341 Downloaded by guest on September 30, 2021