A novel bisphosphonate inhibitor of squalene synthase combined with a statin or a nitrogenous bisphosphonate in vitro.

Statins and nitrogenous bisphosphonates (NBP) inhibit 3-hydroxy-3-methylglutaryl-coenzyme-A reductase (HMGCR) and farnesyl diphosphate synthase (FDPS), respectively, leading to depletion of farnesyl diphosphate (FPP) and disruption of protein prenylation. Squalene synthase (SQS) utilizes FPP in the first committed step from the mevalonate pathway toward cholesterol biosynthesis. Herein, we have identified novel bisphosphonates as potent and specific inhibitors of SQS, including the tetrasodium salt of 9-biphenyl-4,8-dimethyl-nona-3,7-dienyl-1,1-bisphosphonic acid (compound 5). Compound 5 reduced cholesterol biosynthesis and lead to a substantial intracellular accumulation of FPP without reducing cell viability in HepG2 cells. At high concentrations, lovastatin and zoledronate impaired protein prenylation and decreased cell viability, which limits their potential use for cholesterol depletion. When combined with lovastatin, compound 5 prevented lovastatin-induced FPP depletion and impairment of protein farnesylation. Compound 5 in combination with the NBP zoledronate completely prevented zoledronate-induced impairment of both protein farnesylation and geranylgeranylation. Cotreatment of cells with compound 5 and either lovastatin or zoledronate was able to significantly prevent the reduction of cell viability caused by lovastatin or zoledronate alone. The combination of an SQS inhibitor with an HMGCR or FDPS inhibitor provides a rational approach for reducing cholesterol synthesis while preventing nonsterol isoprenoid depletion.

inhibitors have shown the potential for added benefi ts due to decreased triglyceride biosynthesis ( 17 ), likely resulting from a farnesol-induced mechanism ( 18 ).
Nitrogenous bisphosphonates (NBP; e.g., zoledronate and alendronate) are a second class of clinical drugs targeting the mevalonate pathway, and they are used for treatment of bone-related disorders such as osteoporosis. NBPs function by inhibition of FDPS, thus depleting cellular levels of FPP and other downstream isoprenoids ( 19 ). Bisphosphonates may be regarded as analogs of diphosphates, in which the central bridging oxygen atom (P-O-P) has been replaced with a carbon (P-C-P). This results in increased metabolic stability and allows chemical functionalization of the bisphosphonate core. Furthermore, the P-C-P linkage combined with an ␣ -hydroxy group facilitates bone targeting ( 20 ). Although these compounds have a high affi nity for bone, there are also reports of decreased cholesterol levels with patients treated with nitrogenous bisphosphonates ( 21 ). To our knowledge, the combination of an FDPS inhibitor with an SQS inhibitor has not been evaluated.
While various SQS inhibitors exist, relatively few are based on a bisphosphonate structure ( 22 ), and their specifi city for SQS relative to the prenylation arm of the mevalonate pathway has not been reported. Herein, we describe the synthesis and identifi cation of novel bisphosphonates as potent inhibitors of SQS. A structure-activity relationship is evaluated in the context of potency and specifi city for these novel compounds. Emphasis is placed on the evaluation of a lead compound 5 ( Fig. 2 ) in combination with lovastatin or zoledronate in HepG2 cells. We hypothesized that these combinations of inhibitors would decrease cholesterol biosynthesis while preventing the depletion of nonsterol isoprenoid levels, resulting in reduced adverse cellular effects compared with inhibition of HMGCR or FDPS alone.

Chemical synthesis
Preparation of compounds 1 ( 23 ) and 2 ( 24,25 ) has been described, while compounds 3-5 were prepared as follows. In short, geranyl acetate was oxidized with SeO 2 under reported literature procedures ( 26 ) ( Fig. 3 ). The newly formed hydroxyl group was protected by reaction with 3,4-dihydro-2H-pyran, and the acetate group was removed ( 27 ). The resulting product then was split and carried forward in three divergent directions. The 2-tetrahydro pyranyl (THP) protecting group of alcohol 10 was subjected to cop per-mediated Grignard displacement ( 24,25,28 ) with the corresponding biphenyl Grignard reagents to provide independently the ortho, meta, and para analogs 11, 12, and 13, respectively. Standard conversion of the free hydroxyl group to the allylic bromide was performed using PBr 3 . After workup, the bromides were used without further purifi cation in the subsequent alkylation of tetraethyl methylenebisphosphonate ( 29 ). The resulting phosphonate esters 17, 18, and 19 were hydrolyzed to their corresponding salts (3, 4, and 5, respectively) under standard McKenna procedures ( 30 ). All bisphosphonate salts were dissolved in water prior to use for biological studies. Detailed experimental and characterization data is provided in the supplementary data.
(LDL) receptor (LDLR) in the liver and clearance of cholesterol-containing LDL particles from the bloodstream. The use of statin drugs is prevalent because elevated total cholesterol and LDL levels are major risk factors for coronary heart disease ( 4 ). Although the statins are used abundantly and effectively, there are various reasons for developing novel inhibitors of cholesterol biosynthesis. There can be side effects associated with statin use, such as myopathy and hepatotoxicity ( 5 ), which are commonly speculated to be due to the depletion of nonsterol components of the mevalonate pathway ( 6 ). Furthermore, statin use does not always reduce LDL to desired levels ( 7 ), which is particularly important as lower LDL target levels are suggested for some patients (8)(9)(10).
Inhibition of SQS has attracted much interest as a pharmacological target, and various compounds have been identifi ed as inhibitors ( 11,12 ). Lapaquistat (TAK-475, Takeda) progressed to phase III clinical trials, but studies were discontinued after the US Food and Drug Administration recommended suspension of studies with high-dose (100 mg/kg) monotherapy due to hepatotoxicity manifested as elevated levels of liver transaminases ( 12 ). It is currently unknown whether this was due to an enzyme inhibitory class effect or whether it was specifi c to the drug. Inhibition of SQS can result in the accumulation of both FPP and FPP metabolites, such as farnesol-derived dicarboxylic acids ( 13 ), which could be responsible for the possible hepatotoxicity with the high-dose monotherapy of lapaquistat. Farnesol itself can be proapoptotic at high concentrations ( 14 ). Other reported results appeared promising with lapaquistat, with cholesterol levels decreasing in monotherapy treatment. Moreover, the combination therapy of lapaquistat with statins showed additional LDL reduction compared with statins alone ( 12 ). Also of interest, lapaquistat's active metabolite T-91485 was capable of preventing statin-induced myotoxicity in a human skeletal muscle cell model ( 15 ). Similarly, lapaquistat was able to prevent statin-induced myotoxicity in a guinea pig model ( 16 ). In addition to the expected cholesterol depletion, other SQS

Western blot analysis
Protein concentrations were determined by the bicinchoninic acid (BCA) method. Proteins were resolved on 12 or 15% gels and transferred to polyvinylidene difl uoride membranes via electrophoresis. Blocking was performed in 5% nonfat dry milk for 45 min, after which primary and secondary antibodies were added sequentially for 1 h each at 37°C. Proteins were visualized using enhanced chemiluminescence detection. Rap1a and ␣ -tubulin antibodies were acquired from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA), and anti-pan-Ras was acquired from InterBiotechnology (Tokyo, Japan).

Cholesterol biosynthesis assay
Cells were plated in 12-well plates and grown to near confl uency. Compounds were added for 1 h followed by the addition of 1 µCi of 1-14 C-acetate (Sigma) for 4 h. Cells were harvested using trypsin, and lipids were extracted using the Bligh and Dyer method ( 31 ). Chloroform extracts were dried, resuspended in a 30 µl of

SQS enzyme assays
Enzyme assays were performed in 20 µl reactions containing 50 mM phosphate buffer (pH 7.4, 5 mM MgCl 2 , 4 mM CHAPS, 10 mM DTT), 400 ng recombinant enzyme, 0.25 µM [1-3 H]FPP (20 Ci/mmol; American Radiolabeled Chemicals, St. Louis, MO) and 2 mM NADPH. Inhibitors were added with enzyme and incubated for 10 min at 37°C. Substrate was then added and reactions were incubated for 10 min at 37°C. Reactions were stopped by addition of 300 µl 1 mM EDTA, and then 1 ml ice-cold petroleum ether was added. After freezing the lower aqueous phase, the upper phase containing the products was transferred to a scintillation vial containing liquid scintillation fl uid, and radioactivity was quantitated using a Beckman liquid scintillation counter. Data was analyzed using Prism Graphpad software.  pattern resulted in IC 50 values of 5.7, 13.4, and 7.1 nM, respectively. One-way ANOVA was used to test for statistical differences in IC 50 values, and the means were significantly different across the samples ( P < 0.05). Tukey post hoc analysis indicated that compounds 2-5 were each statistically different from compound 1, but they were not statistically different from each other.

Effects of compounds 1-5 on protein prenylation in HepG2 cells
Substrate-like inhibitors targeted against SQS have potential for off-target effects due to inhibition of other FPP utilizing enzymes. Our laboratories previously identifi ed geranyl bisphosphonate ( 33 ) as an inhibitor of GGDPS ( 34 ). With this in mind, we set out to determine whether the compounds active against SQS impaired protein farnesylation or geranylgeranylation, outputs that can identify inhibitors of FDPS, GGDPS, or prenyltransferases ( 35,36 ). The prenylation of select individual proteins was assessed by Western blot for use as markers of the cellular status of protein farnesylation and geranylgeranylation. The Ras antibody utilized in these experiments recognizes both the modifi ed (farnesylated) and unmodifi ed (nonfarnesylated) forms of the protein. Impairment of farnesylation on the Ras Western blot panel was noted by the presence of a more slowly migrating upper unmodifi ed band. The impairment of Rap1a geranylgeranylation was noted by the appearance of a band on the Western blot (the antibody used only detects the unmodifi ed form of Rap1a). HepG2 cells were treated with 25 µM lovastatin for 24 h as a positive control, as this concentration was required to impair both farnesylation and geranylgeranylation ( Fig. 4A ). Lovastatin depletes mevalonate and the downstream products (e.g., FPP and GGPP) and thus impairs both protein farnesylation and geranylgeranylation. HepG2 cells were treated with 50 µM of compounds 1-5 for 24 h. Monogeranyl bisphosphonate (compound 1) impaired protein geranylgeranylation ( Fig.  4A ). Compound 2 also impaired geranylgeranylation, whereas compound 3 displayed a slight impairment. No detectable impairment was noted with compounds 4 or 5. Compound 5 was utilized in subsequent studies as the lead inhibitor due to its potency and specifi city for SQS.

Effect of compound 5 on lovastatin-or zoledronateinduced impairment of protein prenylation
Treatment of HepG2 cells with 25 µM lovastatin for 24 h resulted in impairment of both farnesylation of Ras and geranylgeranylation of Rap1a. Lovastatin-induced impairment of Ras farnesylation was prevented by cotreatment chloroform, and loaded on S-60 silica TLC plates. TLC was performed using an eluting solvent system of toluene and isopropyl ether (1:1) as the mobile phase. Plates were stained with iodide to determine the location of a cholesterol standard. Regions corresponding to cholesterol were excised from the plate, and radioactivity was quantifi ed using a liquid scintillation counter.

Measurement of FPP and GGPP levels
Both FPP and GGPP levels were determined as reported ( 32 ). Briefl y, FPP and GGPP were extracted from cells and incorporated into fl uorescently-labeled CAAX peptides by FTase and GGTase, which were then quantifi ed by fl uorescent detection on an HPLC. Levels were normalized to total protein as measured by BCA assay.

MTT assay
The MTT assay measures the activity of enzymes that reduce the MTT substrate within metabolically active cells. It is commonly used as a measure of cell viability. Cells were allowed to adhere in 24-well plates and grown until approximately 50% confl uent. Cells were treated with indicated compounds and incubated for 45 h, followed by addition of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; EMD Chemicals; La Jolla, CA) and incubation for an additional 3 h. MTT stop solution (HCl, triton X-100, and isopropyl alcohol) was then added, and plates were gently agitated at 37°C overnight. Absorbance was measured at 540 nm with a reference wavelength at 650 nm.

Real-time quantitative PCR
HepG2 cells were allowed to adhere in 6-well plates and grown until approximately 50% confl uent. Cells were then washed with PBS, and the cells were equilibrated in media containing 10% lipoprotein defi cient serum (LPDS) for 24 h. Cells were then treated for 24 h in media containing 10% LPDS with indicated compounds. Total RNA was isolated using Qiashredders and RNase easy mini kit (Qiagen), with inclusion of a DNase step as recommended. cDNA was made from 1 µg of RNA by reverse-transcription using the iScript DNA Synthesis Kit (Bio-Rad). Primers were from Integrated DNA Technologies (Coralville, IA). The primers used were: LDLR forward, TCAACACACAACAGCAGATGGCAC; LDLR reverse, AAGGCTAACCTGGCTGTCTAGCAA; GAPDH forward, TCGA-CAGTCAGCCGCATCTTCTTT; GAPDH reverse, ACCAAATCCGT-TGACTCCGACCTT. Real-time PCR was performed using Sybr Green Master Mix (Applied Biosystems) using an Applied Biosystems Model 7000 real-time thermalcycler. The protocol for real-time was: 2 min at 50°C, 10 min at 95°C, followed by 40 cycles of 95°C for 15s and 60°C for 1 min.

In vitro inhibition of recombinant SQS
A screen of in-house compounds was performed to identify in vitro inhibitors of SQS. A small panel of compounds (1, 2, 3, 4, and 5) was selected for further study of their inhibitory activity ( Fig. 2 ). These compounds were synthesized as described in "Experimental Procedures." Doseresponse curves were used to determine the concentration of compound required to inhibit 50% of SQS activity (IC 50 value, Table 1 ). Geranyl bisphosphonate (compound 1) had an IC 50 value of 1,361 nM in this assay. Addition of a phenyl ring at the C-9 position of the geranyl chain (compound 2) enhanced potency to an IC 50 of 26.5 nM. The addition of a biphenyl group in an ortho-(compound 3), meta-(compound 4), or para-(compound 5) substituted Three independent dose-response curves of compounds 1-5 were used to generate IC 50 values. Values are expressed as the mean ± SE, n = 3.

Effect of compound 5 with lovastatin or zoledronate on cholesterol biosynthesis
Lovastatin at 50 nM or compound 5 at 50 µM significantly inhibited de novo cholesterol biosynthesis in HepG2 cells compared with untreated HepG2 cells ( Fig. 6 ). The combination of these concentrations of lovastatin and compound 5 showed a trend toward enhanced inhibition of cholesterol synthesis compared with single treatments, but it was not statistically signifi cant. Zoledronate at 10 µM also reduced cholesterol biosynthesis compared with control, and the combination of 25 µM compound 5 with 10 µM zoledronate did not signifi cantly enhance cholesterol depletion compared with the respective single treatments.

Effect of compound 5 with lovastatin or zoledronate on LDLR mRNA levels
Lovastatin at 10 µM signifi cantly increased LDLR mRNA levels 2.3-fold compared with control HepG2 cells after 24 h ( Fig. 7 ). Compound 5 increased LDLR mRNA levels to 1.4-fold of control, but this was not statistically signifi cant ( P > 0.05). The combination of lovastatin and compound 5 resulted in increased LDLR mRNA levels compared with control, but levels were not elevated further than only lovastatintreated cells. Zoledronate at 10 µM also signifi cantly increased LDLR mRNA compared with control. The combination of zoledronate with compound 5 did not further increase LDLR mRNA levels compared with only zoledronate-treated cells, but the levels remained signifi cantly enhanced compared with control.

Effect of compound 5 on lovastatin-or zoledronateinduced reduction of MTT activity
Treatment of HepG2 cells for 48 h with 25 µM lovastatin signifi cantly reduced MTT activity compared with control with 25 µM exogenous FPP, whereas lovastatin-induced impairment of Rap1a geranylgeranylation was prevented by cotreatment with 25 µM GGPP ( Fig. 4B ). Cotreatment of 25 µM lovastatin with 25 µM compound 5 prevented lovastatininduced impairment of Ras farnesylation, but it did not completely restore Rap1a geranylgeranylation.

Effect of compound 5 and lovastatin on FPP and GGPP levels
We next measured the FPP and GGPP levels from HepG2 cells in response to treatment with either compound 5 or lovastatin alone and in combination for 24 h. In HepG2 cells treated with 25 µM compound 5, FPP levels were increased approximately 16-fold and GGPP levels were approximately 1.6-fold compared with control ( Fig. 5 ). As expected, lovastatin (25 µM) reduced both FPP and GGPP levels compared with control. The combination of 25 µM lovastatin with 25 µM compound 5 resulted in increased FPP levels compared with lovastatin-treated cells; however, GGPP levels remained diminished. This data correlates with the results showing prevention of lovastatin-induced impairment of farnesylation, but not geranylgeranylation, by cotreatment with compound 5.

DISCUSSION
In the present work, we synthesized and identifi ed novel inhibitors of SQS. A lead inhibitor (compound 5) was identifi ed through in vitro enzyme assays and did not impair protein prenylation in HepG2 cells, suggesting specifi city for SQS over the prenylation branch of the mevalonate pathway. Other bisphosphonates have been shown to inhibit SQS ( 22 ); however, their effects on protein prenylation were not reported. Enzyme assays established that the addition of a phenyl group at the C-9 position (compound 2) of geranyl bisphosphonate greatly increased potency for SQS in vitro relative to monogeranyl bisphosphonate (compound 1). The addition of a second phenyl ring in an ortho-(compound 3), meta-(compound 4), or para-(compound 5) substitution pattern enhanced potency against SQS as measured by IC 50 values. This suggests that the hydrophobic portion of the molecule is important for potency. It would be expected that the bisphosphonate portion of compound 5 likely interacts with magnesium ( Fig. 8A ). Cotreatment with 25 µM FPP, GGPP, or compound 5 signifi cantly diminished lovastatin-mediated reduction of MTT activity. Compound 5 (25 µM) had no signifi cant effect on MTT levels as a single treatment. Zoledronate (10 µM) decreased MTT activity to ‫ف‬ 50% of control cells, and it was not reduced by FPP cotreatment ( Fig. 8B ). Cotreatment of HepG2 cells with 10 µM zoledronate and 25 µM GGPP or 25 µM compound 5 signifi cantly diminished zoledronate-mediated reduction of MTT activity.   in FPP available for other necessary cellular processes but not in such extreme excess to produce adverse effects. It remains to be determined if a reduction in GGPP levels resulting from this combination of drugs would be desirable or undesirable.
With FDPS inhibition by nitrogenous bisphosphonates, there is decreased product formation and accumulation of the upstream intermediates DMAPP and IPP ( 45 ). Notably, nitrogenous bisphosphonate-induced IPP accumulation can facilitate the formation of the possible proapoptotic isoprene ATP analog known as ApppI ( 46 ). With the combination of FDPS and SQS inhibition, the carbon fl ux will be fi rst inhibited at FDPS, resulting in accumulation of IPP. Albeit limited, the remaining carbon fl ux through the FDPS inhibition will yield FPP and encounter a second inhibition at SQS, preventing the formation of squalene and resulting in some relative accumulation of FPP. This SQS-induced FPP accumulation when combined with FDPS-induced IPP accumulation yields the necessary substrates for GGPP synthesis. The FPP and GPPP available could then be utilized in other nonsterol branches of the mevalonate pathway, such as protein farnesylation and geranylgeranylation. Indeed, in HepG2 cells, the combination of compound 5 and zoledronate prevents zoledronate-induced impairment of farnesylation and geranylgeranylation and diminishes zoledronate-mediated reduction in MTT activity.
In conclusion, our results suggest that dual inhibition of HMGCR and SQS or FDPS and SQS could yield a means of cholesterol reduction with the potential for minimal offtarget effects due to decreased depletion of nonsterol isoprenoids compared with HMGCR or FDPS inhibition alone. Treatment using multiple drugs (SQS inhibitors combined with statins being a possible approach in the near term) or the development of novel compounds capable of dual inhibition could be feasible. In particular, design and synthesis of a single agent capable of potent dual FDPS and SQS inhibition is likely to be an achievable goal, and future studies could evaluate this approach. The results presented herein are from cell culture (in vitro), and future studies would be necessary to validate these hypotheses in vivo because it is diffi cult to predict the effects of multienzyme targeting in vivo. atoms bound within a DDXXD motif. Other bisphosphonates have been shown to interact in this manner with other DDXXD-containing proteins ( 37 ). The hydrophobic portion of the molecule may extend into a hydrophobic fl ap of the protein, consisting of residues 50-54, that has been found to display conformational changes with different inhibitors, including an inhibitor containing a biphenyl group ( 38 ). Compounds 1, 2, and the orthosubstituted compound 3 display impairment of protein geranylgeranylation, whereas no impairment was detectable with the meta-(compound 4) or para-(compound 5) compounds. Due in part to the lack of success in clinical trials by SQS inhibitors as monotreatments, emphasis was placed on characterization of compound 5 in combination with HMGCR inhibition (lovastatin) or FDPS inhibition (zoledronate) in HepG2 cells. Simultaneous inhibition of SQS and HMGCR or FDPS provides possible mechanisms for decreasing cholesterol synthesis with less disruption of nonsterol isoprenoid levels compared with single enzyme inhibition, potentially alleviating toxicity due to excessive isoprenoid accumulation or depletion.
Treatment of HepG2 cells with compound 5 results in an inhibition of cholesterol biosynthesis and a substantial accumulation of FPP, which could result in the formation of potentially toxic levels of farnesol ( 14 ) or farnesolderived dicarboxylic acids ( 13 ). Bisphosphonates are notorious for poor cellular entry due to their high charge-to-mass ratio, likely partially explaining the relatively high concentrations required in cell culture assays. Future studies could evaluate prodrug approaches that might confer liver targeting and mask the bisphosphonate negative charge, thus signifi cantly enhancing cellular entry (39)(40)(41).
With statin treatment, there is depletion of mevalonate and all downstream components of the mevalonate pathway ( 2 ). The combination of lovastatin with compound 5 is able to inhibit cholesterol biosynthesis and upregulate LDLR mRNA. Furthermore, compound 5 prevents lovastatinmediated reduction in FPP levels and impairment of protein farnesylation, although GGPP levels and protein geranylgeranylation are not restored. Many of the pleiotropic effects of statins are thought to be mediated by GGPP depletion and impairment of protein geranylgeranylation ( 42 ). For example, statins can upregulate eNOS and have antioxidant effects ( 43 ), which are prevented by addition of GGPP but not FPP ( 44 ). Due to the diminished GGPP levels and impairment of protein geranylgeranylation with the combination of compound 5 and lovastatin, some statin-mediated pleiotropic effects may be retained. In addition, the combination of compound 5 with lovastatin can partially prevent statin-mediated reduction in cell viability. These results may explain why other SQS inhibitors can reverse models of statin-induced myopathy ( 15,16 ). While adverse effects of the SQS inhibitor lapaquistat as a monotherapy were discouraging, our results suggest a mechanism by which the dual administration of SQS inhibitors with statins may alleviate potential problems due to treatment with either agent alone. The FPP accumulation due to SQS inhibition and FPP depletion due to HMGCR inhibition may offset each other, resulting