Identification and Characterization of Mechanism of Action of P61-E7, a Novel Phosphine Catalysis-Based Inhibitor of Geranylgeranyltransferase-I

Small molecule inhibitors of protein geranylgeranyltransferase-I (GGTase-I) provide a promising type of anticancer drugs. Here, we first report the identification of a novel tetrahydropyridine scaffold compound, P61-E7, and define effects of this compound on pancreatic cancer cells. P61-E7 was identified from a library of allenoate-derived compounds made through phosphine-catalyzed annulation reactions. P61-E7 inhibits protein geranylgeranylation and blocks membrane association of geranylgeranylated proteins. P61-E7 is effective at inhibiting both cell proliferation and cell cycle progression, and it induces high p21CIP1/WAF1 level in human cancer cells. P61-E7 also increases p27Kip1 protein level and inhibits phosphorylation of p27Kip1 on Thr187. We also report that P61-E7 treatment of Panc-1 cells causes cell rounding, disrupts actin cytoskeleton organization, abolishes focal adhesion assembly and inhibits anchorage independent growth. Because the cellular effects observed pointed to the involvement of RhoA, a geranylgeranylated small GTPase protein shown to influence a number of cellular processes including actin stress fiber organization, cell adhesion and cell proliferation, we have evaluated the significance of the inhibition of RhoA geranylgeranylation on the cellular effects of inhibitors of GGTase-I (GGTIs). Stable expression of farnesylated RhoA mutant (RhoA-F) results in partial resistance to the anti-proliferative effect of P61-E7 and prevents induction of p21CIP1/WAF1 and p27Kip1 by P61-E7 in Panc-1 cells. Moreover, stable expression of RhoA-F rescues Panc-1 cells from cell rounding and inhibition of focal adhesion formation caused by P61-E7. Taken together, these findings suggest that P61-E7 is a promising GGTI compound and that RhoA is an important target of P61-E7 in Panc-1 pancreatic cancer cells.


Introduction
Proteins such as the Rho family G-proteins are posttranslationally modified by the addition of a geranylgeranyl isoprenoid [1]. The isoprenoid modification is important for membrane association and functions of these proteins. Recent studies have highlighted the significance of protein geranylgeranylation in human cancers. First, it has been shown that a number of geranylgeranylated proteins play important roles in tumorigenesis and metastasis [2][3][4][5][6]. Second, characterization of GGTase-Ideficient cells showed that the inhibition of GGTase-I leads to proliferation inhibition and accumulation of p21 CIP1/WAF1 , pointing to the significance of GGTase-I in cell proliferation and cell cycle progression [7]. Furthermore, conditional knockout of the b-subunit of GGTase-I results in the inhibition of lung tumor growth and increased survival of mice expressing oncogenic K-ras [7]. Thus, inhibition of protein geranylgeranylation is a promising approach for developing anticancer drugs, and inhibitors of GGTase-I (GGTIs) are currently undergoing preclinical studies. The results obtained are consistent with the idea that GGTIs disrupt oncogenic and tumor survival pathways, inhibit prolifer-ation and anchorage-independent growth, and induce apoptosis [8][9][10][11][12].
A variety of approaches were taken to develop GGTI compounds. Peptidomimetic GGTI compounds were derived from the peptide bearing the C-terminal CAAL (cysteine followed by two aliphatic amino acids and the C-terminal residue is leucine or phenylalanine) motif, a sequence recognized by GGTase-I [11,[13][14][15][16]. The first nonpeptidomimetic inhibitor, GGTI-DU40, was identified via high-throughput screening of a compound library [17]. More recently, quantitative structure-activity relationship (QSAR) models for GGTIs have been developed and used to carry out virtual screen of more than 9 million commercially available compounds. This resulted in the identification of seven compounds with novel scaffolds [18]. Our approach [19,20] was to construct a library of allenoate-derived small molecules based on phosphine-catalyzed annulation reactions that produce diverse compounds such as dihydropyrroles [21,22], tetrahydropyridines [23][24][25], bicyclic succinimides, dioxanylidenes [26], a-pyrones [27], dihydropyrones [28,29], and cyclohexenes [30]. Initial GGTI compounds identified by screening a 171-compound pilot library were then used to synthesize derivatives with increased potency. Screening the 4288 compounds enabled us to identify two types of novel GGTI compounds: one group with a dihydropyrrole ring as its core scaffold and the other group with a tetrahydropyridine ring as its core scaffold [19,20]. In our previous study, we reported that derivatization of a carboxylic acid emanating from the dihydropyrrole ring of one of the GGTI compounds dramatically improves their cellular activity [20]. The improved GGTI, P61-A6, inhibits proliferation of a variety of human cancer cell lines, and causes G 1 cell cycle arrest and induction of p21 CIP1/WAF1 [20]. Our study with P61-A6 using a human pancreatic cancer xenograft model in mice showed that the GGTI exhibited prolonged blood circulation time and significant suppression of tumor growth [12].
In this study, we focused on P3-E5, a GGTI compound with a tetrahydropyridine ring [20], and modified it with an L-leucine methylester to replace the free acid. This led to the synthesis of GGTI P61-E7. Compound P61-E7 selectively inhibits protein geranylgeranlyation in cells with higher potency than P61-A6 and caused accumulation of both RhoA and RalA, which are known substrates of GGTase-I, in the cytosolic fractions. Furthermore, compared to P61-A6, P61-E7 exhibits improved potency to inhibit both cell proliferation and cell cycle progression and inducing p21 CIP1/WAF1 level. In addition, compound P61-E7 increases p27 Kip1 level and inhibits phosphorylation of p27 Kip1 (T187). P61-E7 also blocks anchorage-independent growth and disrupts cell morphology of Panc-1.
We also examined the mechanism of action of P61-E7 in pancreatic cancer cells. This is important because the cellular effects observed pointed to the involvement of RhoA, a geranylgeranylated small GTPase shown to mediate a number of cellular processes such as actin stress fiber organization, cell adhesion and cell proliferation, the significance of the inhibition of RhoA geranylgeranylation on GGTI-mediated cellular effects was evaluated. We showed that stable expression of farnesylated RhoA mutant (RhoA-F) prevents induction of p21 CIP1/WAF1 and p27 Kip1 by P61-E7 in Panc-1 and promotes partial resistance to the anti-proliferative effect of the GGTI compound. Furthermore, stable expression of RhoA-F rescues Panc-1 cells from cell rounding and inhibition of focal adhesion formation following P61-E7 treatment. Taken together, our findings show that P61-E7 provides a novel addition to the range of GGTI compounds that are in development as anticancer drugs and that RhoA is an important target of the GGTI compound in Panc-1 pancreatic cancer cells.

Cell Lines and Cell Cultures
MiaPaCa-2 (ATCC, Rockville, MD), NIH3T3 (ATCC, Rockville, MD) and Panc-1 (ATCC, Rockville, MD) cells were maintained in Dulbecco's modified Eagle's medium (DMEM; Cellgro, Herndon, VA). Jurkat (ATCC, Rockville, MD) cells were maintained in RPMI 1640 medium (Cellgro, Herndon, VA). Both media were supplemented with 10% (v/v) fetal bovine serum (FBS; HyClone, Logan, UT) and 1% penicillin/1% streptomycin stock solution (Invitrogen, Carlsbad, CA). All cells were cultured at 37uC in a humidified incubator at 5% CO 2 . Transfection Procedure and Creation of Stable Cell Lines Panc-1 cells were grown to 80% confluency on 6-well plates and transfected with Lipofectamine TM 2000 (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. Briefly, 10 ml of Lipofectamine TM 2000 reagent was diluted in 250 ml of OPTI-MEM medium (Invitrogen, Carlsbad, CA) and was allowed to equilibrate at room temperature for 5 min. 4.0 mg of plasmid DNA per 250 ml of medium was suspended and allowed to complex with the diluted Lipofectamine TM 2000 for 20 min at room temperature. Transfection complexes were added to cells in serum-free DMEM without antibiotics and incubated at 37uC for 4-6 h. Medium was then changed and replaced with DMEM containing 10% FBS, with penicillin-streptomycin, and the cells were further incubated at 37uC for 48 h. To generate stable cell lines, cells were passaged at 1:10, 1:20, and 1:50 dilutions into selective medium containing 800 mg/ml of GeneticinH (G418 sulfate; Invitrogen, Carlsbad, CA). Cell clones were expanded into mass culture, and expression of 3xHA-RhoA-GG and 3xHA-RhoA-F was analyzed by Western blotting using antibodies against the HA tag.

Subcellular Fractionation
NIH3T3 cells and Panc-1 cells stably expressing 3xHA-RhoA were treated with DMSO, P61-A6, or P61-E7 for 48 h. Cells were then washed and scraped into phosphate-buffered saline (PBS) and centrifuged at 2,500 rpm for 5 min. Pellets were resuspended (10 mM HEPES/KOH at pH 7.3, 10 mM KCl, 5 mM MgCl 2 , 0.5 mM DTT, and 16 protease inhibitor mixture), incubated on ice for 30 min and homogenized. Homogenates were centrifuged at 10006g for 10 min to collect the cytosolic fractions (supernatant). The remaining pellets were then resuspended in buffer containing 1% Triton X-100, 150 mM NaCl, 20 mM Tris-HCl at pH 7.5, 1 mM EDTA, and 16 protease inhibitor mixture, and centrifuged at 15, 000 rpm for 15 min to collect the membranecontaining fractions (supernatant). Na + /K + ATPase-a and RhoGDI were used as markers for the membrane-containing fractions and the cytosolic fractions, respectively.

Immunoprecipitation of p27 Kip1 Protein
Panc-1 cells were lysed in lysis buffer (1% Triton X-100, 150 mM NaCl, 20 mM Tris-HCl, pH 7.5, 1 mM EDTA, and 16 protease inhibitor mixture). Lysates were incubated with p27 Kip1 antibodies (rabbit, Santa Cruz Biotechnology, Inc.) in 50% Protein G/Sepharose beads slurry for 2 h at 4uC while rocking, then washed four times with an excess of lysis buffer. Samples were then boiled at 95uC for 10 min in 46 sodium dodecyl sulfate (SDS)sample buffer and analyzed by Western blotting as described above.

Cell Proliferation Assays and Cell Cycle Analyses
Effects of GGTIs on cell proliferation were examined using the CCK-8 cell counting kit (Dojindo Molecular Technologies, Kumamoto, Japan) as described previously [20]. Briefly, cells (3610 3 ) were seeded onto 96-well plates. The following day cells were treated with the appropriate inhibitor as indicated in the figure legends under low-serum conditions (0.5% FBS) for 72 hours. Cell Proliferation was calculated relative to the DMSO control. The cell cycle profiles were analyzed by flow cytometry as described previously [32].

Fluorescence Microscopy
To examine the effects of GGTI on actin cytoskeleton and focal adhesion formation, cells were seeded on 4-well or 8-well chamber slides. The following day, cells were treated as indicated in the figure legends. Then cells were fixed in 3.7% formaldehyde-PBS and permeabilized in 0.1% Trion X-100-PBS for actin fiber and vinculin detection. Actin fibers were detected by incubation with fluorescein isothiocyanate (FITC)-labeled phalloidin or tetramethyl rhodamine isothiocyanate (TRITC)-labeled phalloidin (Sigma-Aldrich, St. Louis, MO) as indicated in figure legends. Vinculin was detected with anti-vinculin antibodies (mouse, Sigma-Aldrich, St. Louis, MO) and fluorescently conjugated anti-mouse IgG antibody (Sigma-Aldrich, St. Louis, MO). Cell images were visualized using a Zeiss Microscope (406 magnification). Images were captured using AxioVision software.

Anchorage Independent Growth Assay
Cells were seeded at a cell density of 1000 cells/well in duplicate in 12-well culture dishes in 0.3% agar over a 0.5% bottom agar layer. Various concentrations of P61-E7 or DMSO were incubated in the top layer of cells. Cultures were re-fed and treated with the GGTI or DMSO once weekly (14 days of incubation in total). Colonies were stained with 1 mg/ml MTT (tetrazolium salt) for 1 hour and scanned.

Synthesis of GGTI P61-E7 by the Derivatization of the Carboxyl Group on the Dihydropyrrole Ring of GGTI P3-E5
We previously reported a novel small molecule GGTI compound P61-A6 ( Figure 1A) [20]. This compound was identified by screening a pilot library of allenoate-derived compounds made through phosphine catalysis for scaffolds that can be used for further derivatization. The dihydropyrrole scaffold identified was then derivatized to identify compounds with improved potency and bioavailability. A particularly useful observation was made concerning a modification of the carboxyl group off of the dihydropyrrole ring that resulted in a dramatic increase in cellular potency [19,20]. Along the way, we succeeded in establishing a novel chemical library of .4,000 compounds through phosphine-catalyzed annulations of resin-bound allenoates.
The above initial study identified another scaffold that consists of a tetrahydropyridine ring. P3-E5, a representative tetrahydropyridine carboxylic acid compound, inhibits GGTase-I with IC 50 value of 0.31 mM [20]. The inhibition was specific to GGTase-I, as little inhibition was observed with FTase and RabGGTase ( Figure  S1). In this study, we derivatized the carboxyl group of P3-E5 to generate compounds that exhibit increased cellular activity. From a series of modified compounds, we identified P61-E7 that was synthesized by coupling carboxylic acid of P3-E5 with L-leucine methyl ester to yield an amido ester P61-E7 ( Figure 1B). The reaction carried out under the influence of O-benzotriazole-N,N,N9,N9-tetramethyluronium hexafluorophosphate (HBTU) and N,N-diisopropylethylamine (DIPEA) in chloromethane [33] was efficient with 95% yield. P61-E7 demonstrated better cellular GGTI activity than P3-E5; GI 50 values with Jurkat cells were 3.5 mM and 20 mM for P61-E7 and P3-E5, respectively. GGTI P61-E7 Exhibits Improved Ability to Inhibit Geranylgeranylation in Cultured Mammalian Cells When Compared to GGTI P61-A6 P61-E7 exhibits increased potency to inhibit protein geranylgeranylation in cells (Figure 2A and 2B). In the first assay, we assessed inhibition of protein geranylgeranylation by using an antibody that specifically detects unprenylated form of Rap1 [34]. Rap1 is a Raslike GTPase that plays a role in cadherin-based cell adhesion [34], and is a known substrate of GGTase-I. Panc-1 ( Figure 2A) and NIH3T3 cells ( Figure 2B) were treated with DMSO, P61-A6, or P61-E7 for 48 hours. Treatment with either P61-A6 or P61-E7 led to the accumulation of unprenylated Rap1 in a dose-dependent manner in both Panc-1 ( Figure 2A) and NIH3T3 ( Figure 2B) cells. P61-E7 treatment resulted in the inhibition of protein geranylgeranylation at much lower concentrations when compared to P61-A6 in both cell lines tested (Figure 2A and 2B). For example, at 5 mM, P61-A6 increased the level of unprenylated Rap1 in Panc-1 cells by 2.6-fold when compared to DMSO, while 5 mM of P61-E7 caused the level of unprenylated Rap1 in Panc-1 cells to increase by 12.5fold ( Figure 2A). Likewise, at 5 mM, P61-A6 treatment did not significantly change the level of unprenyalted Rap1 in NIH3T3 cells compared to DMSO treatment, while 5 mM of P61-E7 resulted in 21-fold increase in unprenylated Rap1 level ( Figure 2B). All these findings reflected a significant improvement in the potency of P61-E7 in inhibiting protein geranylgeranylation in cells.
Next, the specificity of P61-E7 was measured by determining its ability to inhibit farnesylation of HDJ2 which is a cochaperone of the heat-shock cognate protein 70 (hsc70) involved in protein folding and traffic [35] and a known substrate of farnesyltransferase (FTase). NIH3T3 cells were treated with various concentrations of P61-E7 (5, 10, and 20 mM) or BMS-225975, an FTase inhibitor (FTI). While treatment with BMS-225975 slowed the mobility of HDJ2 on SDS-polyacrylamide gel, no such mobility shift was observed with P61-E7 treatment even when cells were treated with 20 mM of P61-E7 ( Figure 2C). These findings indicated that P61-E7 selectively inhibits protein geranylgeranylation but not farnesylation in cells.

Treatment with P61-E7 results in increase of cytosolic localization of geranylgeranylated proteins
Because prenylation is required for geranylgeranylated proteins to be associated with cellular membranes, we assessed whether P61-E7 treatment inhibits membrane association of proteins such as RhoA and RalA, known substrates of GGTase-I. NIH3T3 cells ( Figure 3A and 3B) and Panc-1 cells stably expressing 3xHA-RhoA ( Figure 3C and 3D) were treated with DMSO, P61-A6 or P61-E7, and membrane and cytosolic fractions were prepared and processed for SDS-PAGE followed by Western blotting. The Na + /K + ATPase-a membrane marker and the RhoGDI cytosolic marker were used to confirm the separation of membranecontaining and cytosolic fractions, respectively. As shown in Figure 3, treatment with P61-E7 resulted in increases of RalA ( Figure 3A and 3B) and 3xHA-RhoA ( Figure 3C and 3D) in the cytosolic fractions of NIH3T3 cells and Panc-1cells stably expressing 3xHA-RhoA, respectively, while their association with the membrane fractions was decreased. Furthermore, at both concentrations tested (5 and 10 mM), P61-E7 is more effective at disrupting association of RalA ( Figure 3A and 3B) and 3xHA-RhoA (Figure 3Cand 3D) with the membrane fractions when compared to P61-A6.
P61-E7 is More Effective at Inhibiting Cell Cycle Progression, Inducing p21 CIP1/WAF1 Protein Level and Blocking Cell Proliferation When Compared to P61-A6 As shown in Figure 4, treatment of Panc-1 ( Figure 4A) and MiaPaCa-2 ( Figure 4B) cells with either P61-A6 or P61-E7 for 48 h caused dose-dependent enrichment of G 1 phase cells, whereas the percentage of S phase cells decreased. Consistent with findings discussed above, P61-E7 is more effective at inhibiting cell cycle progression in both cell lines tested ( Figure 4A and 4B). For example, following treatment with P61-A6 (5 mM), the proportion of Panc-1 cells in the G 1 phase increased from 45.1%63.1% (DMSO) to 48.4%61.4%, whereas the proportion of Panc-1 cells in the G 1 phase increased to 62.7%62.2% following treatment with P61-E7 (5 mM) ( Figure 4A). Likewise, at 5 mM, P61-A6 treatment caused an increase in proportion of MiaPaCa-2 cells in the G 1 phase by 2.6%61.3% in comparison to DMSO treatment, while treatment of MiaPaCa-2 cells with 5 mM of P61-E7 increased the proportion of cells in the G 1 phase by 9.7%60.67% when compared to DMSO treatment ( Figure 4B).
The accumulation of G 1 phase cells was associated with an induction of proteins involved in the regulation of cell cycle. Both P61-A6 and P61-E7 induced p21 CIP1/WAF1 expression level in Panc-1 ( Figure 4C) and NIH3T3 (data not shown) cells in a dosedependent manner. Similar to observations discussed above, P61-E7 treatment resulted in a higher induction of p21 CIP1/WAF1 level when compared to treatment with P61-A6 in both cell lines tested. For example, at 20 mM, P61-A6 induced p21 CIP1/WAF1 level by 25.9-fold in comparison to DMSO treatment, while treatment with 20 mM of P61-E7 induced p21 CIP1/WAF1 level by 61.2-fold ( Figure 4C). Besides inducing p21 CIP1/WAF1 expression, P61-E7 increased p27 Kip1 protein level in Panc-1 cells ( Figure S2A). Phosphorylation of p27 Kip1 on Thr187 by cyclin E/CDK2 results in the association of p27 Kip1 with SCF ubiquitin ligase, targeting it for degradation. It has been shown that Cdk2 phosphorylates p27 Kip1 on Thr187 and promotes its nuclear degradation, and that the inhibition of the phosphorylation results in nuclear accumulation of p27 Kip1 [36]. Therefore, we investigated whether P61-E7 inhibits phosphorylation of p27 Kip1 on Thr187 in Panc-1 cells. P61-E7 increased p27 Kip1 protein level in Panc-1 cells ( Figure S2B and S2C), while inhibiting the phosphorylation of p27 Kip1 on Thr187 ( Figure S2B).
The abilities of P61-A6 and P61-E7 to inhibit proliferation of Panc-1 cells under low-serum conditions (0.5%) were examined and compared. Consistent with findings discussed above, P61-E7 is more effective at inhibiting cell proliferation ( Figure 4D). For example, at 1.25 mM, P61-A6 treatment inhibited proliferation by 9.72%611.5% in comparison to DMSO treatment, while treatment with 1.25 mM of P61-E7 inhibited proliferation by 48.6%64.49%. The improved potency of P61-E7 to inhibit cell proliferation when compared to P61-A6 reflects its increased ability to inhibit protein geranylgeranylation inside the cells (Figure 2).

P61-E7 Causes Actin Cytoskeleton Disorganization, Abolishes Formation of Focal Adhesions, and Inhibits Anchorage-Independent Growth of Panc-1 Cells
To further investigate cellular effects of P61-E7, we carried out experiments using Panc-1 cells. First, we examined the effects of P61-E7 on the cell morphology of Panc-1 cell. Previous studies have shown that GGTase-I deficiency results in disrupted actin cytoskeleton in fibroblasts [7], and that GGTI treatment causes disorganization of actin cytoskeleton and inhibits formation of vinculin-containing focal adhesions [37]. Panc-1 cells were treated with DMSO or P61-E7 and the cells were stained with phalloidin to observe actin cytoskeleton. As shown in Figure 5A, P61-E7 treatment caused disorganization of actin cytoskeleton and significant cell rounding. We next examined whether P61-E7 treatment had effects on focal adhesion. This was examined by using antibody against vinculin, one of the components of focal adhesion. As seen in Figure 5A, the treatment with P61-E7 led to the disappearance of vinculin punctate immunostaining at the periphery of Panc-1 cells. The staining became uniform throughout the cell. The GGTI treatment had no effect on vinculin expression level (data not shown). These results suggest that focal adhesion assembly was inhibited by P61-E7.
In addition, a soft agar clonogenicity assay was used to measure the effect of P61-E7 on the anchorage-independent growth of Panc-1 cells. Treatment with increasing concentrations of P61-E7 impaired the anchorage-independent growth of Panc-1 cells in a dose-dependent manner ( Figure 5B and 5C). For examples, at 5 mM, P61-E7 inhibited colony formation by 42.9%64.1%, while it inhibited colony formation by 71.2%65.2% at 10 mM ( Figure 5B).

Stable Expression of Farnesylated RhoA in Panc-1 Cells Suppresses GGTI-induced Phenotypes
RhoA, a geranylgeranylated small GTPase, has been shown to influence a number of cellular processes including actin stress fiber organization, cell adhesion and cell proliferation. Interestingly, the cellular effects of P61-E7 observed pointed to the possible involvement of RhoA in GGTI-mediated cellular effects. First, the above results on the effects of GGTI to induce p21 CIP1/WAF1 are consistent with previous studies that have shown that small GTPase protein RhoA (RhoA-GG) acts as a negative regulator of p21 CIP1/WAF1 expression [38,39]. Effects of GGTI on actin cytoskeleton also support the possibility that P61-E7 induces cellular effects by blocking geranylgeranylation of RhoA, thereby disrupting its subcellular localization and inhibiting its functions. Indeed, results from our RhoA-GTP pull-down assay indicated that P61-E7 inhibited activation of RhoA in a dose-dependent manner ( Figure S3). To assess the implication of RhoA in P61-E7 induced phenotypes, we constructed Panc-1 stable cell line expressing farnesylated RhoA (RhoA-F). Using site-directed mutagenesis, a RhoA-F mutant was generated by altering the CAAX from CLVL to CLVS sequence to render RhoA a substrate of farnesyltransferase. Previous studies showed that farnesylated RhoA has the same subcellular localization as geranylgeranylated RhoA, and that both RhoA prenyl isoforms have similar effects on cell morphology, actin organization, vinculin distribution, cell proliferation and p21 CIP1/WAF1 expression [40]. However, unlike wild-type RhoA (RhoA-GG), RhoA-F is expected to be resistant to GGTI treatment. Therefore, if P61-E7 does indeed increase p21 CIP1/WAF1 expression by blocking geranylgeranylation of RhoA, stable expression of RhoA-F is expected to overcome P61-E7's effect on p21 CIP1/WAF1 expression.
We established populations of Panc-1 cells stably expressing either 3xHA-tagged wild-type RhoA (3xHA-RhoA-GG) or 3xHA-RhoA mutant (3xHA-RhoA-F) as described in ''Materials and Methods,'' and confirmed the prenylation of 3xHA-RhoA-GG and 3xHA-RhoA-F in these stable cell lines by treating them with DMSO, FTI BMS-225975, or GGTI-298 which is a commercially available GGTI. Treatment with GGTI-298 inhibited protein geranylgeranylation as indicated by the appearance of the unprenylated Rap1 band ( Figure 6A, left and right panels). Moreover, as expected, GGTI-298 treatment slowed the mobility of 3xHA-RhoA-GG on SDS-PAGE, indicating geranylgeranylation of 3xHA-RhoA-GG was inhibited ( Figure 6A, right panel). However, GGTI-298 did not change the mobility of 3xHA-RhoA-F protein ( Figure 6A, left panel). Consistent with observations discussed above ( Figure 2C), FTI BMS-225975 inhibited farnesylation of HDJ2, a known substrate of FTase, as shown by the appearance of the slower migrating form of HDJ2 on SDS-PAGE ( Figure 6A, left and right panels). Furthermore, treatment with the FTI inhibited farnesylation of 3xHA-RhoA-F ( Figure 6A, left panel) while it had no effect on the mobility of 3xHA-RhoA-GG ( Figure 6A, right panel). We also checked the expression levels of RhoA-GG and RhoA-F in the Panc-1 stable cell lines used in these experiments. As indicated in Figure 6B, the levels of RhoA-GG and Rho-F were similar in these stable cell lines.
Next, we investigated whether stable expression of farnesylated RhoA confers resistance to the antiproliferative effect of P61-E7. Following treatment with DMSO or 5 mM of P61-E7, cell proliferation of Panc-1 cells, Panc-1 cells stably expressing 3xHA-RhoA-GG, and 3 different clones of Panc-1 cells stably expressing 3xHA-RhoA-F with different expression levels ( Figure 7B) was measured as described previously [20]. As shown in Figure 7A, stable expression of 3xHA-RhoA-F promotes partial resistance to the anti-proliferative effect of P61-E7 in a manner that depends on its expression level in Panc-1 cells. With the highest expression level of RhoA-F, Panc-1 (3xHA-RhoA-F) Clone No. 8 showed the highest resistance to the anti-proliferative effect of P61-E7 when compared with Panc-1 (3xHA-RhoA-F) Clone No. 1 (lowest expression level) and Clone No. 4 (intermediate expression level). These findings suggested that P61-E7 inhibition of cell proliferation in Panc-1 cells is due, in part, to inhibition of RhoA geranylgeranylation and hence its function.
We also examined whether cell morphology and actin cytoskeleton changes induced by P61-E7 can be rescued by the stable expression of 3xHA-RhoA-F in Panc-1 cells. As shown in Figure 8, Panc-1 cells expressing 3xHA-RhoA-GG responded to P61-E7, and morphological changes and actin cytoskeleton disorganization were observed. In addition, vinculin staining showed that focal adhesion formation was inhibited by the GGTI treatment. In contrast, P61-E7 failed to induce morphological changes and actin cytoskeleton changes in Panc-1 cells expressing HA-RhoA-F. Furthermore, focal adhesions were still detected even after P61-E7 treatment in this cell line. These results suggest that stable expression of RhoA-F rescues Panc-1 cells from cell rounding and inhibition of focal adhesion formation caused by P61-E7.

Discussion
In this paper, we report successful identification of a novel GGTI compound P61-E7 that has a tetrahydropyridine scaffold. P61-E7 was derived by converting the free carboxylic acid moiety on the tetrahydropyridine ring of a parent compound P3-E5 to Lleucine amido methyl ester. P61-E7 exhibits increased potency to inhibit protein geranylgeranylation compared to our previous preclinical compound P61-A6 [12,20]. Inhibition of geranylgeranylation was first shown by detecting the appearance of unprenylated Rap1. In addition, P61-E7 was shown to be more effective at inducing cytoplasmic accumulation of RalA and RhoA in comparison to P61-A6.
The availability of a compound with increased potency enabled us to define cellular consequences of inhibiting protein geranylgeranylation. By using two compounds with different levels of potency to inhibit geranylgeranylation (P61-E7 and P61-A6), we could look for cellular effects that correlate with the inhibition of geranylgeranylation. We examined the effects of P61-E7 on pancreatic cancer cells Panc-1. We have shown that P61-E7 causes more pronounced cell cycle block at the G 1 phase compared with P61- A6. In addition, P61-E7 showed higher level of p21 CIP1/WAF1 induction in comparison to P61-A6. Furthermore, P61-E7 treatment resulted in a more pronounced inhibition of cellular proliferation under low serum condition when compared to P61-A6. The excellent correlation we observed between cellular effects and increased potency to inhibit protein geranylgeranylation supports the idea that the consequences of geranylgeranylation inhibition include inhibition of cell cycle progression, induction of p21 CIP1/WAF1 and inhibition of proliferation.
We also examined the mechanism of action of P61-E7 in this paper, and our work suggests that RhoA is a major player in GGTI-mediated effects on pancreatic cancer cells. We have shown that P61-E7 induces p27 Kip1 level, suggesting that the treatment with GGTI results in the increases of both p21 CIP1/WAF1 and p27 Kip1 in Panc-1 cells. The mechanism of induction, however, appears to differ between the two Cdk inhibitors. In the case of p21 CIP1/WAF1 , it appears that the induction is regulated at the transcriptional level, as we have shown that GGTI induces expression of p21-reporter construct [20]. On the other hand, GGTI effect on p27 Kip1 is associated with the inhibition of phosphorylation of Thr187. Phosphorylation of p27 Kip1 (T187) results in the association of p27 Kip1 with SCF ubiquitin ligase, targeting p27 Kip1 for degradation [36]. Because P61-E7 inhibits phosphorylation of p27 Kip1 (T187), it is possible that P61-E7 prevents degradation of p27 Kip1 , causing accumulation of p27 Kip1 in the nucleus in Panc-1 cells.
The above results on the effects of GGTI on Cdk inhibitors raise the possibility that the GGTI effects are mediated by RhoA. First, it is known that RhoA acts as a negative regulator of p21 CIP1/WAF1 expression [38,39,41]. Second, RhoA is shown to be essential for the degradation of p27 Kip1 [42][43][44] and that RhoA stimulates p27 Kip1 degradation through its regulation of cyclin E/CDK2 activity [43]. Further support for the idea that many of the GGTI effects we observed with Panc-1 cells are due to the inhibition of RhoA function comes from additional effects of P61-E7 we observed. We found that P61-E7 treatment results in cell rounding, actin cytoskeleton disorganization and inhibition of focal adhesion formation.
To evaluate the possibility that GGTI targets RhoA functions, we have established stable cell lines of Panc-1 expressing a mutated form of RhoA (RhoA-F). RhoA-F has the CAAL (CLVL) box changed to CAAX (CLVS) so that the mutant protein is farnesylated instead of being geranylgeranylated. Characterization of these stable cell lines showed that the expression of RhoA-F reversed the ability of GGTI to induce p21 CIP1/WAF1 and p27 Kip1 . Stable expression of RhoA-F rescued cells from cell rounding and focal adhesion inhibition caused by P61-E7. Finally, we observed partial rescue of GGTI-induced proliferation inhibition and this rescue correlated with the expression levels of RhoA-F. These results suggest that many of the GGTI effects are due to the inhibition of RhoA functions. Our results on the rescue of proliferation inhibition by the expression of RhoA-F is consistent with the previous report that the expression of farnesylatable RhoA and Cdc42 restores proliferation of GGTase-I deficient fibroblasts expressing oncogenic K-Ras [7]. There are other proteins implicated in the GGTI effects. One type of proteins is Ral suggested to be involved in the GGTI effects of MiaPaca-2 pancreatic cancer cells [4]. It is possible that the Ral proteins are involved in other aspects of GGTI effects such as tumor growth. In addition, it has been suggested that Rac1and Rac3 proteins are involved in the GGTI effects on fibroblast cells [45]. Further work is needed to express mutant forms of these proteins that can bypass geranylgeranylation to gain complete understanding of the mechanism of GGTI effects. Figure S1 Effects of P3-E5 on the enzymatic activities of GGTase-I, RabGGTase, and FTase. GGTase-I and FTase activities were determined by following the incorporation of radiolabeled isoprenoid [  Elmer Life Sciences) and 2 mM maltose-binding protein-tagged substrates (K-Ras4B for FTase and RhoA for GGTase-I) in 20 ml of buffer containing 50 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 5 mM ZnCl 2 , 0.01% Triton X-100, and 1 mM dithiothreitol. Inhibitors were added at the indicated concentrations. The final DMSO concentration was 2.5% for all samples. Reactions were carried out for 10 min at 30uC. The reaction mixture was spotted onto a filter paper and treated with 10% trichloroacetic acid, followed by ethanol and acetone washing. The filter was counted using a scintillation counter. For RabGGTase assays, the reaction contained the following components in 20 ml: 0.625 ml of [ 3 H]GGPP (0.7 mM), 25 nM RabGGTase, 0.6 mM REP-1, 0.6 mM purified Rab7 or Ypt1 protein, 40 mM HEPES (pH 7.5), 150 mM NaCl, 5 mM dithiothreitol, 3 mM MgCl 2 , and 0.3% CHAPS. Reactions were carried out for 20 min at 37uC, and the products were analyzed as described above for the GGTase-I reaction. Data represent the mean 6 S.D. of two measurements from two independent experiments. (TIF) Figure S2 Effects of P61-E7 on p27 Kip1 protein levels.

Supporting Information
(A) Panc-1 cells were treated with DMSO or various concentrations of P61-E7, and whole cell lysates were collected and resolved on SDS-PAGE for immunoblotting using antibodies against p27 Kip1 . Results shown are representative of three independent experiments. (B) and (C) Panc-1 cells were treated with DMSO or P61-E7 (5 mM) for 48 h. Whole cell lysates were collected and were immunoprecipitated (B) with anti-p27 Kip1 antibodies in 50% Protein G/Sepharose beads slurry (negative control for immunoprecipitation: whole cell lysates from untreated Panc-1 cells mixed with slurry, without antibodies). (B) Immunoprecipitates were then resolved on SDS-PAGE for immunoblot analysis using phospho-27 Kip1 (T187) antibodies (B, top panel) or p27 Kip1 antibodies (B, bottom panel). Lanes: 1, DMSO-treated; 2, P61-E7-treated; 3, No antibody control. (C) Remaining lysates (10 mg) from each sample were resolved on SDS-PAGE for immunoblotting using antibodies against p27 Kip1 (C, top panel) or actin (C, bottom panel) to determine total p27 Kip1 level in each input used for immunoprecipitation. Results shown are representative of three independent experiments. The RhoGDI, Actin, phospho-p27 Kip1 and total p27 Kip1 bands were quantified using ImageJ, and the results are given above the images as fold change compared to the DMSO control. (TIF) Figure S3 P61-E7 inhibits RhoA activation in Panc-1 cells. Panc-1 cells were transfected with 3xHA-RhoA pcDNA expression vector. Twenty-four hours after transfection, cells were serum-starved in the presence of DMSO or P61-E7 for 24 h. Then, cells were stimulated with 10% FBS in DMEM in the presence of DMSO or P61-E7 for 30 min. Whole cell lysates were collected using Mg 2+ -containing lysis buffer, and GTP-RhoA was pulled down using GST-tagged Rhotekin-RBD protein beads (Cytoskeleton) following the manufacturer's instructions. Whole cell lysates (inputs) and pull-down were resolved on SDS-PAGE for immunoblotting analysis using HA.11 antibodies to detect total 3xHA-RhoA (bottom panel) and GTP-bound 3xHA-RhoA (top panel), respectively. (TIF)