Biochemical and Biophysical Research Communications 290, 869–873 (2002)

doi:10.1006/bbrc.2001.6289, available online at http://www.idealibrary.com on

Identification of a Bisphosphonate That InhibitsIsopentenyl Diphosphate Isomerase andFarnesyl Diphosphate Synthase

Keith Thompson,* James E. Dunford,* Frank H. Ebetino,† and Michael J. Rogers*,1

*Bone Research Group, Department of Medicine & Therapeutics, University of Aberdeen, Scotland, United Kingdom; and†Proctor & Gamble Pharmaceuticals, Health Care Research Center, Mason, Ohio 45040

Received December 6, 2001

tain a nitrogen in one of their two side-chains (N-BPs,

We and others have recently shown that the major

molecular target of nitrogen-containing bisphospho-nate drugs is farnesyl diphosphate synthase, an en-zyme in the mevalonate pathway. In an in vitro screen,we discovered a bisphosphonate, NE21650, that po-tently inhibited farnesyl diphosphate synthase but,unlike other N-BPs investigated, was also a weak in-hibitor of isopentenyl diphosphate isomerase. NE21650was a more potent inhibitor of protein prenylation inosteoclasts and macrophages, and a more potent in-hibitor of bone resorption in vitro, than alendronate,despite very similar IC50 values for inhibition of farne-syl diphosphate synthase. Our observations show thatminor changes to the structure of bisphosphonatesallow inhibition of more than one enzyme in the me-valonate pathway and suggest that loss of protein pre-nylation due to inhibition of more than one enzyme inthe mevalonate pathway may lead to an increase inantiresorptive potency compared to bisphosphonatesthat only inhibit farnesyl diphosphate synthase. © 2002

Elsevier Science

Key Words: bisphosphonate; farnesyl diphosphatesynthase; isopentenyl diphosphate isomerase; meval-onate; prenylation; osteoclast.

Bisphosphonates are commonly used drugs to inhibitexcessive osteoclast-mediated bone resorption in pa-tients with osteoporosis, metastatic bone disease andhypercalcemia (1, 2). Bisphosphonates inhibit osteo-clast formation in vitro, disrupt the osteoclast cyto-skeleton and ruffled border, and can cause osteoclastapoptosis, reviewed in Ref. (3). We and others haverecently demonstrated that bisphosphonates that con-

1 To whom correspondence and reprint requests should be ad-dressed at Bone Research Group, Department of Medicine & Ther-apeutics, University of Aberdeen, Polwarth Building, Forester-hill, Aberdeen AB25 2ZD, UK. Fax: 44 1224 699884. E-mail:m.j.rogers@abdn.ac.uk.

869

such as alendronate, risedronate, ibandronate, andpamidronate) act by inhibiting the mevalonate path-way and, therefore, indirectly prevent the posttransla-tional prenylation of small GTPases such as Rho, Rac,and cdc42 (4–8). By preventing the prenylation ofsmall GTPases in osteoclasts, N-BPs affect signaltransduction pathways regulating cytoskeletal ar-rangement, membrane ruffling, vesicular traffickingand apoptosis (9–11). Inhibition of protein prenylationtherefore accounts for the effects of N-BPs on osteoclastmorphology, function and survival. The enzyme of themevalonate pathway that is the target of N-BPs hasrecently been identified as farnesyl diphosphate (FPP)synthase (7, 12). N-BPs are potent inhibitors of FPPsynthase and there is a highly significant correlationbetween the rank order of potency for inhibition of FPPsynthase in vitro and inhibition of bone resorption invivo (13).

Although FPP synthase appears to be the majorpharmacological target of N-BPs, we have examinedwhether these compounds may also inhibit other en-zymes in the mevalonate pathway, such as isopentenyldiphosphate (IPP) isomerase, the enzyme proximal toFPP synthase in the mevalonate pathway (Fig. 1). Inthis study we describe a novel N-BP, NE21650, thatinhibits both FPP synthase and IPP isomerase, andNE10571, a pharmacologically inactive isomer ofNE21650.

MATERIALS AND METHODS

Reagents. The sodium salts of alendronate (ALN; 4-amino-butane-1,1-bisphosphonic acid), NE21650 (2-aminophenyl-2-ethane-1-hydroxy-1,1-bisphosphonic acid) and NE10571 (4-aminophenyl-2-ethane-1,1-bisphosphonic acid) were synthesized by Proctor & Gam-ble Pharmaceuticals (Cincinnati, OH). The N-BPs were dissolved inPBS, the pH adjusted to 7.4 with 1 N NaOH, and then filter-sterilized using a 0.2-mm filter. All other reagents were obtainedfrom Sigma Chemical Co., unless otherwise stated. [14C]Mevalonicacid lactone was purchased from NEN (Hounslow, UK).

0006-291X/02 $35.00© 2002 Elsevier ScienceAll rights reserved.

Expression of recombinant human IPP isomerase and FPP syn-

Vol. 290, No. 2, 2002 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

thase. The human IPP isomerase clone pFMH12 in the bacterialexpression vector pARC306N (14) was kindly provided by C. D.Poulter, University of Utah and expressed as previously described(13). Human FPP synthase (Clone KIA1293, a kind gift from theKazusa DNA Research Institute, Kisarazu, Chiba, Japan) was ex-pressed and partially purified as described previously (13).

In vitro assay of IPP isomerase and FPP synthase activity. IPPisomerase activity was assayed by the acid-lability method (15) using[14C]IPP as a substrate, as previously described (13). The hydrolyzedreaction products were solvent extracted into ligroin and the radio-activity in the ligroin fraction (dimethylallyl diphosphate, the prod-uct of the isomerase reaction) was determined using a PackardTricarb 1900CA scintillation counter. N-BPs were pre-incubatedwith the enzyme for 10 min prior to initiation of the reaction.

FPP synthase activity was assayed using [14C]IPP and geranyldiphosphate as substrates, as described previously (13). Reactionproducts were extracted with water-saturated butyl alcohol and theamount of radioactivity in the upper phase was determined by scin-tillation counting, using a Packard Tricarb 1900CA scintillationcounter. To determine the effects of N-BPs on FPP synthase activity,N-BPs were pre-incubated with the enzyme preparation for 10 min,prior to initiation of the reaction.

Effect of N-BPs on Rap1A prenylation in osteoclasts and macro-phages. Osteoclasts were isolated from the long bones of 2-day-oldNew Zealand white rabbits, seeded into 6-well culture plates thenpurified according to the method of Coxon et al. (10). J774 cells werecultured in 12-well plates (1 3 105 cells/well). After treatment with50 mM N-BPs for 24 h, J774 cells and osteoclasts were lysed in 0.1 ml1% (v/v) NP-40, 0.1% (w/v) sodium dodecyl sulfate, 0.5% (w/v) sodiumdeoxycholate, 1% (v/v) protease inhibitor cocktail (Sigma). Twentymicrograms of J774 cell lysate or 50 mg of rabbit osteoclast lysate wasthen electrophoresed on 12% polyacrylamide–SDS gels under reduc-ing conditions. Following electrophoresis, the proteins were trans-ferred to polyvinyldifluoride (PVDF) membrane and then hybridizedwith 0.2 mg/ml goat polyclonal anti-Rap1A antibody (Santa CruzBiotechnology Inc.), which recognizes only the unprenylated form ofRap1A (16, 17) followed by 1 mg/ml anti-goat IgG–horseradish per-oxidase (HRP) conjugate (Sigma). Chemiluminescent bands werevisualized using Supersignal reagent (Pierce) and a Bio-Rad Fluor-SMax MultiImager.

Incorporation of [14C]mevalonate into prenylated proteins in mac-rophages. Protein prenylation in J774 macrophages was studied bymetabolically labeling cells with [14C]mevalonic acid lactone for 20 hin the presence of 50 or 100 mM N-BPs, as described previously (4).Cell lysates were then examined for the presence of radiolabeled,prenylated proteins. Fifty micrograms of protein from each cell ly-sate was separated on 12% polyacrylamide–SDS gels under reducingconditions, then prenylated proteins were visualized by phosphor-imaging, using a Kodak phosphorscreen and Bio-Rad Personal FXimager.

Effects of N-BPs on osteoclast-mediated bone resorption. Rabbitosteoclasts were seeded on to 5-mm-diameter elephant tusk ivoryslices in 96-well plates and allowed to adhere for 2 h. Following 48 hexposure with 0.1–100 mM N-BPs, the slices were fixed with 4%formaldehyde and resorptive activity of osteoclasts was assessed byreflected light microscopy using a Leitz Quantimet Q500MC imageanalysis system, as described by van’t Hof et al. (18).

RESULTS AND DISCUSSION

In this study we screened N-BPs for the ability toinhibit FPP synthase and IPP isomerase. NE21650, abisphosphonate containing an amino group in theortho-position of an aromatic ring (Fig. 1), was a potent

870

inhibitor of recombinant human FPP synthase, compa-rable in potency to ALN (IC50 58 and 53 nM, respec-tively; Fig. 2). However, unlike ALN (Fig. 2) and otherN-BPs studied previously, such as ibandronate, rise-dronate and incadronate (13), NE21650 also inhibitedrecombinant human IPP isomerase in vitro (IC50 ;70mM; Fig. 2). NE10571 (an isomer of NE21650 contain-ing a para-amino group rather than an ortho-aminogroup), had no inhibitory effect on IPP isomerase atconcentrations up to 300 mM, and was almost 700-foldless potent than NE21650 at inhibiting recombinanthuman FPP synthase. This indicates that the positionof the amino group in an aromatic ring is an importantdeterminant for inhibition of IPP isomerase and FPPsynthase by bisphosphonates. This is in accord withour recent observation that the three-dimensional po-sition of a nitrogen within a heterocyclic ring is criticalfor effective inhibition of FPP synthase by bisphospho-nates (13).

We next investigated whether, by inhibiting bothFPP synthase and IPP isomerase, NE21650 was amore effective inhibitor of protein prenylation in vitrothan ALN (which inhibits FPP synthase only, with avery similar IC50 to NE21650). Western blot analysis,using an antibody that selectively recognizes the un-prenylated form of Rap1A (16, 17), showed thatNE21650 was a more effective inhibitor of protein pre-

FIG. 1. Schematic representation of the mevalonate pathway,and the structures of nitrogen-containing bisphosphonates.

Vol. 290, No. 2, 2002 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

nylation than ALN in both J774 macrophages and pu-rified rabbit osteoclasts, since unprenylated Rap1A ac-cumulated to a greater extent in cells treated withNE21650 compared to cells treated with ALN (Fig. 3).This suggests that inhibition of both IPP isomeraseand FPP synthase prevents protein prenylation moreeffectively in intact cells than inhibition of FPP syn-thase alone. NE10571 did not cause accumulation ofunprenylated Rap1A in J774 cells (Fig. 3), consistentwith the lack of effect of this compound on IPP isomer-ase or FPP synthase.

The greater effectiveness of NE21650 for inhibitingprotein prenylation was confirmed by labeling J774macrophages with [14C]mevalonate. At a concentrationof 50 mM, NE21650 clearly prevented the synthesis ofisoprenoid lipids, that migrate at the dye front of SDS–PAGE gels (4, 10), but only slightly prevented the

FIG. 2. (A) Inhibition of recombinant human IPP isomerase byNE21650, but not NE10571 or ALN. N-BPs were incubated withrecombinant human IPP isomerase for 10 min prior to assayingactivity. [14C]DMAPP formed in the reaction was extracted intoligroin and the activity determined by liquid scintillation counting.Data are the mean 6 SEM of at least three independent experi-ments, expressed as a percentage of IPP isomerase activity in theabsence of N-BP. (B) Inhibition of recombinant human FPP synthaseby NE21650 and ALN. N-BPs were incubated with recombinanthuman FPP synthase for 10 min prior to assaying for FPP synthaseactivity. 14C-labeled FPP and geranyl diphosphate formed in thereaction were extracted into water-saturated butanol and activitydetermined by liquid scintillation counting. Data are means 6 SEMof at least three independent experiments, expressed as a percentageof FPP synthase activity in the absence of N-BP.

871

prenylation of 21- to 26-kDa small GTPases (Fig. 4).ALN (50 mM) had little effect on the synthesis of iso-prenoid lipids or on protein prenylation. At a concen-tration of 100 mM, NE21650 completely prevented thesynthesis of isoprenoid lipids and very markedly pre-vented prenylation of small GTPases, whereas ALNwas less effective. As expected, NE10571 had no effecton the synthesis of isoprenoid lipids or on protein pre-nylation (Fig. 4).

Finally, given the apparent increase in the effective-ness of NE21650 for inhibiting protein prenylation, weexamined whether this would confer an increasein antiresorptive potency compared to ALN. In an invitro bone resorption assay using rabbit osteoclasts,NE21650 was slightly (but significantly) more effectivethan ALN at inhibiting resorption at concentrations$10 mM (Fig. 5). Hence, NE21650, by inhibiting both

FIG. 3. Inhibition of prenylation of Rap1A by ALN andNE21650, but not NE10571, in J774 macrophages (A) and purifiedrabbit osteoclasts (B). Cells were treated with 50 mM N-BPs for 24 hprior to separation of proteins in cell lysates by SDS–PAGE, thenWestern blotting using an antibody that recognizes the 21-kDa,unprenylated form of Rap1A (arrowheads). Blots were also probedwith anti-actin antibody to ensure equal loading per lane (notshown). Data are representative of three independent experiments.

FIG. 4. Inhibition of protein prenylation by ALN and NE21650,but not NE10571, in J774 macrophages. Cells were metabolicallylabeled for 20 h with [14C]mevalonate in the absence or presence of 50or 100 mM ALN, NE10571, or NE21650. Cell lysates were analyzedby SDS–PAGE on 12% polyacrylamide gels and radiolabeled, preny-lated proteins were detected by phosphorimaging. The positionof isoprenoid intermediates and 21- to 26-kDa, prenylated smallGTPases are indicated.

ACKNOWLEDGMENTS

Vol. 290, No. 2, 2002 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

IPP isomerase and FPP synthase, is more effectivethan ALN at inhibiting protein prenylation in osteo-clasts and, consequently, has slightly increased anti-resorptive potency. NE10571 did not inhibit bone re-sorption at all concentrations studied (up to 100 mM).

This study emphasizes the critical importance of thethree-dimensional position of the nitrogen group in thebisphosphonate side chain (for example, ortho- vs para-position in an aromatic ring) for determining the abil-ity to inhibit FPP synthase and hence for determiningthe potency for inhibiting bone resorption. Further-more, our findings illustrate that certain side-chainstructures, such as NE21650, also allow the inhibitionof other enzymes in the mevalonate pathway, such asIPP isomerase. The N-BPs ibandronate and incadr-onate, as well as inhibiting FPP synthase, are alsoknown to inhibit squalene synthase (19, 20), therebypreventing cholesterol biosynthesis (Fig. 1). However,inhibition of squalene synthase does not prevent pro-tein prenylation, and replenishing osteoclasts withcholesterol does not overcome the effects of N-BPs (5).Furthermore, the antiresorptive potency of ibandro-nate and incadronate correlates well with their abilityto inhibit FPP synthase alone (13). Hence, inhibition ofsqualene synthase does not appear to directly enhancethe antiresorptive effect of ibandronate or incadronate.By contrast, inhibition of IPP isomerase synthase byNE21650 appears to enhance the ability to preventprotein prenylation in intact cells and hence leads toa slight increase in antiresorptive potency comparedto bisphosphonates that prevent protein prenylationby inhibiting only FPP synthase. The exact three-dimensional structure of NE21650 that allows inhibi-tion of IPP isomerase and FPP synthase remains to bedetermined.

FIG. 5. Inhibition of osteoclastic bone resorption by N-BPs inrabbit osteoclasts. Cells were treated with 0.1–100 mM N-BPs for48 h prior to quantification of the area of resorbed mineral by re-flected light microscopy. Data are means 6 SEM of at least threeindependent experiments, expressed as a percentage of the area ofmineral resorbed in untreated cultures.

872

We are grateful to C. D. Poulter and F. M. Hahn for providing thecDNA clone for IPP isomerase.

REFERENCES

1. Russell, G., Mueller, G., Shipman, C., and Croucher, P. (2001)Clinical disorders of bone resorption. Novartis Found. Symp.232, 251–267.

2. Watts, N. B. (2001) Treatment of osteoporosis with bisphospho-nates. Rheum. Dis. Clin. North Am. 27, 197–214.

3. Rogers, M. J., Gordon, S., Benford, H. L., Coxon, F. P., Luckman,S. P., Monkkonen, J., and Frith, J. C. (2000) Cellular and mo-lecular mechanisms of action of bisphosphonates. Cancer 88,2961–2978.

4. Luckman, S. P., Hughes, D. E., Coxon, F. P., Russell, R. G. G.,and Rogers, M. J. (1998) Nitrogen-containing bisphosphonatesinhibit the mevalonate pathway and prevent post-translationalprenylation of GTP-binding proteins, including Ras. J. BoneMiner. Res. 13, 581–589.

5. Fisher, J., Rogers, M. J., Halasy, J. M., Luckman, S. P., Hughes,D. E., Masarachia, P. J., Wesolowski, G., Russell, R. G. G.,Rodan, G. A., and Reszka, A. A. (1999) Alendronate mechanismof action: Geranylgeraniol, an intermediate in the mevalonatepathway, prevents inhibition of osteoclast formation, bone re-sorption and kinase activation in vitro. Proc. Natl. Acad. Sci.USA 96, 133–138.

6. van Beek, E., Lowik, C., Van der Pluijm, G., and Papapoulos, S.(1999) The role of geranylgeranylation in bone resorption and itssuppression by bisphosphonates in fetal bone explants in vitro: Aclue to the mechanism of action of nitrogen-containing bisphos-phonates. J. Bone Miner. Res. 14, 722–729.

7. Bergstrom, J. D., Bostedor, R. G., Masarachia, P. J., Reszka,A. A., and Rodan, G. (2000) Alendronate is a specific, nanomolarinhibitor of farnesyl diphosphate synthase. Arch. Biochem. Bio-phys. 373, 231–241.

8. Benford, H. L., Frith, J. C., Auriola, S., Monkkonen, J., andRogers, M. J. (1999) Farnesol and geranylgeraniol prevent acti-vation of caspases by aminobisphosphonates: Biochemical evi-dence for two distinct pharmacological classes of bisphosphonatedrugs. Mol. Pharmacol. 56, 131–140.

9. Reszka, A. A., Halasy-Nagy, J. M., Masarachia, P. J., and Rodan,G. A. (1999) Bisphosphonates act directly on the osteoclast toinduce caspase cleavage of mst1 kinase during apoptosis. A linkbetween inhibition of the mevalonate pathway and regulation ofan apoptosis-promoting kinase. J. Biol. Chem. 274, 34967–34973.

10. Coxon, F. P., Helfrich, M. H., van ‘t Hof, R. J., Sebti, S. M.,Ralston, S. H., Hamilton, A. D., and Rogers, M. J. (2000) Proteingeranylgeranylation is required for osteoclast formation, func-tion, and survival: Inhibition by bisphosphonates and GGTI-298.J. Bone Miner. Res. 15, 1467–1476.

11. Benford, H. L., McGowan, N. W., Helfrich, M. H., Nuttall, M. E.,and Rogers, M. J. (2001) Visualization of bisphosphonate-induced caspase-3 activity in apoptotic osteoclasts in vitro. Bone28, 465–473.

12. van Beek, E., Pieterman, E., Cohen, L., Lowik, C., and Papapou-los, S. (1999) Farnesyl pyrophosphate synthase is the moleculartarget of nitrogen-containing bisphosphonates. Biochem. Bio-phys. Res Commun. 264, 108–111.

13. Dunford, J. E., Thompson, K., Coxon, F. P., Luckman, S. P.,Hahn, F. M., Poulter, C. D., Ebetino, F. H., and Rogers, M. J.(2001) Structure–activity relationships for inhibition of farnesyldiphosphate synthase in vitro and inhibition of bone resorption

in vivo by nitrogen-containing bisphosphonates. J. Pharmacol. dence for the formation in vivo of a metabolite that inhibits bone

Vol. 290, No. 2, 2002 BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS

Exp. Ther., 235–242.14. Hahn, F. M., Xuan, J. W., Chambers, A. F., and Poulter, C. D.

(1996) Human isopentenyl diphosphate:dimethylallyl diphos-phate isomerase: Overproduction, purification, and characteriza-tion. Arch. Biochem. Biophys. 332, 30–34.

15. Satterwhite, D. M. (1985) Isopentenyldiphosphate delta-isomerase. Methods Enzymol. 110, 92–99.

16. Reszka, A. A., Halasy-Nagy, J., and Rodan, G. A. (2001)Nitrogen-bisphosphonates block retinoblastoma phosphoryla-tion and cell growth by inhibiting the cholesterol biosyntheticpathway in a keratinocyte model for esophageal irritation. Mol.Pharmacol. 59, 193–202.

17. Frith, J. C., Monkkonen, J., Auriola, S., Monkkonen, H., andRogers, M. J. (2001) The molecular mechanism of action of theanti-resorptive and anti-inflammatory drug clodronate: Evi-

873

resorption and causes osteoclast and macrophage apoptosis. Ar-thritis Rheum. 44, 2201–2210.

18. Van’t Hof, R. J., Tuinenburg-Bol, R. A., and Nijweide, P. J.(1995) Induction of osteoclast characteristics in cultured avianblood monocytes: Modulation by osteoblasts and 1,25-(OH)2 vi-tamin D3. Int. J. Exp. Pathol. 76, 205–214.

19. Amin, D., Cornell, S. A., Gustafson, S. K., Needle, S. J., Ullrich,J. W., Bilder, G. E., and Perrone, M. H. (1992) Bisphosphonatesused for the treatment of bone disorders inhibit squalenesynthase and cholesterol biosynthesis. J. Lipid Res. 33, 1657–1663.

20. Amin, D., Cornell, S. A., Perrone, M. H., and Bilder, G. E. (1996)1-Hydroxy-3-(methylpentylamino)-propylidene-1,1-bisphosphonicacid as a potent inhibitor of squalene synthase. Arzneimittel-Forschung. 46, 759–762.