Geranylgeranylation

Geranylgeranylation is a form of prenylation, which is a post-translational modification of proteins that involves the attachment of one or two 20-carbon lipophilic geranylgeranyl isoprene units from geranylgeranyl diphosphate to one or two cysteine residue(s) at the C-terminus of specific proteins. Prenylation (including geranylgeranylation) is thought to function, at least in part, as a membrane anchor for proteins.

The process of geranylgeranylation can be catalyzed by either geranylgeranyl transferase I (GGTase I) or Rab GGTase (also GGTase II). GGTase I catalyzes the addition of one geranylgeranyl group onto the C-terminal consensus sequence CAAL (somewhat similar to farnesyltransferase reactions), where C=cysteine, A=any aliphatic amino acid, and L=leucine. Rab GGTase adds a total of two geranylgeranyl groups onto two cysteine residues at the C-terminal consensus sequence CXC or XXCC. The source of the geranylgeranyl group is geranylgeranyl diphosphate, which is synthesized from the isoprenoid biosynthetic pathway.

An example of this can be seen in the lipid anchoring of the Rho GTPase family of signaling molecules and the gamma subunit of heterotrimeric G proteins.

For more information see prenylation.

Bisphosphonates (formerly called diphosphonates) have become the drugs of choice for skeletal diseases involving excessive osteoclast-mediated bone resorption, such as Paget's disease, tumor-induced osteolysis, and postmenopausal osteoporosis. These compounds are synthetic analogues of pyrophosphate that contain nonhydrolyzable P―C―P bonds, have high affinity for bone mineral, and inhibit bone resorption by osteoclasts.(1) Different bisphosphonates can be generated by altering the structure of the two side-chains (R1 and R2) attached to the geminal carbon atom. The first generation of bisphosphonate drugs (such as dichloromethylene-1,1-bisphosphonate[CLO] and 1-hydroxyethylidene-1,1-bisphosphonate [ETI]) were introduced almost 30 years ago.(2,3) This was followed by the development of more potent bisphosphonates after the discovery that antiresorptive potency in vivo could be increased up to 30,000-fold by the insertion of a primary, secondary, or tertiary nitrogen function in the R2 side-chain (e.g., 3-amino-1-hydroxypropylidene-1,1-bisphosphonate[PAM], 4-amino-1-hydroxybutylidene-1,1-bishphosphonate [ALN], and 1-hydroxy-3(methylpentylamino)-propylidene-1,1-bisphosphonate [IBA], which have an alkyl R2 side-chain, or 2-(3-pyridinyl)-1-hydroxyethylidene-1,1-bisphosphonate [RIS] and 2-(imidazol-1-yl)-hydroxyethylidene-1,1-bisphosphonate [ZOL], which have a heterocyclic R2 side-chain).(4,5)

Bisphosphonates are internalized by osteoclasts(6,7) and appear to inhibit resorption by causing alterations to the osteoclast cytoskeleton,(8) by affecting protein or membrane trafficking(9) or by causing osteoclast apoptosis.(10,11) Bisphosphonates, at least in vitro, also inhibit the formation of mature, multinucleated osteoclasts from mononuclear, hematopoietic precursors(12) and appear to cause the production of factor(s) by osteoblasts that inhibit osteoclast formation.(13) Although all such effects may contribute to the ability of bisphosphonates to inhibit bone resorption in vivo,(14) the molecular mechanisms involved have remained largely unknown. We and others have recently shown that nitrogen-containing bisphosphonates can inhibit the biosynthetic mevalonate pathway.(15–18) This pathway is required for the production of sterols from mevalonate and involves the synthesis of isoprenyl diphosphate intermediates, such as farnesyl diphosphate (FPP) and geranylgeranyl diphosphate (GGPP). The latter compounds are substrates for prenyl:protein transferases (FTase and geranylgeranyl:protein transferase [GGTases] I and II),(19), which catalyze the post-translational modification of proteins that contain characteristic carboxy-terminal, cysteine-containing CaaX, xCxC, or xxCC prenylation motifs (in particular the small guanosine triphosphate [GTP]-binding proteins such as Ras, Rho, Rac, and Rab).(20,21) Prenylation of small GTP-binding proteins with farnesyl or geranylgeranyl groups is essential for the localization of these proteins to cell membranes and hence for their biological function as important components of intracellular signaling pathways.(22,23) Because bisphosphonates are potent inhibitors of protein prenylation in J774 macrophages, we have proposed that inhibition of protein prenylation also may account for their antiresorptive mechanism of action on osteoclasts.(17,18) In support of this hypothesis, addition of geranylgeraniol (GGOH) can overcome the inhibitory effect of the bisphosphonates ALN(24) and IBA(25) on osteoclast formation and bone resorption in vitro.

In this study, we have confirmed that nitrogen-containing bisphosphonates inhibit protein prenylation in osteoclasts. Furthermore, using specific, cell-permeable peptidomimetic inhibitors that prevent protein farnesylation (FTI-277) and protein geranylgeranylation (GGTI-298),(26,27) we have showed that geranylgeranylated proteins (to be specific, those prenylated by GGTase I) rather than farnesylated proteins regulate the morphology and cytoskeleton of osteoclasts and are required for osteoclast formation and bone resorptive activity and for suppression of osteoclast apoptosis.

MATERIALS AND METHODS Reagents CLO, ETI, chloro-4-phenylthiomethylene-1,1-bisphosphonate (TIL), ALN, RIS, PAM, and IBA were provided by Procter and Gamble Pharmaceuticals (Cincinnati, OH, U.S.A.). ZOL was provided by Novartis Pharma AG (Basle, Switzerland). The bisphosphonates were dissolved in phosphate-buffered saline (PBS) and the pH was adjusted to 7.4 with 1N NaOH and then filter-sterilized by using a 0.2-μm filter. Prenyl:protein transferase inhibitors (FTI-277 and GGTI-298) were prepared as described previously.(26,27) Mevastatin was purchased from Sigma Chemical Co. (Poole, U.K.) and was converted from the lactone as described by Luckman et al.(17) [14C]mevalonic acid lactone, Enhance and Hyperfilm-MP were from Amersham (Aylesbury, U.K.). [3H]all-trans farnesol (FOH) and [3H]all-trans geranylgeraniol (GGOH) were from American Radiochemicals, Ltd. (St. Louis, MO, U.S.A.). Solvent was removed from the radiochemicals by evaporating in nitrogen. Other reagents were from Sigma Chemical Co. unless stated otherwise.

Isolation and purification of rabbit osteoclasts Mature osteoclasts were isolated from 2-day-old New Zealand white rabbits using a modification of the method described by Tezuka et al.(28) The long bones from each rabbit were removed after they were killed and transferred into ice-cold PBS and then minced in 15 ml of α modified essential medium (α-MEM; Life Technologies, Paisley, U.K.) containing 100 U/ml penicillin and 100 μg/ml streptomycin, and vigorously vortexed three times. After allowing the bone pieces to settle for 1 minute, the supernatant was transferred to a fresh tube and made up to a final volume of 50 ml α-MEM containing 10% fetal calf serum. The cells were seeded into 6-well, 24-well, or 96-well plates (Costar, Cambridge, MA, U.S.A.; 3 ml, 0.5 ml, or 200 μl of cell suspension per well, respectively). The following day, contaminating adherent cells in 6-well and 24-well plates were removed by rinsing in PBS and then incubation for 2–3 minutes with 0.001% pronase/0.002% EDTA in PBS. The remaining cells (>95% tartrate-resistant acid phosphatase [TRAP]-positive osteoclasts) were rinsed twice in PBS and then cultured in fresh α-MEM containing 10% fetal calf serum, 100 U/ml penicillin, and 100 μg/ml streptomycin.

Isolation of avian osteoclasts Bone marrow cells and osteoclasts were isolated from the tibias and femora of 18-day-old embryonic chicks by a modification of the method described by van't Hof et al.(29) Chicks were killed by decapitation and then the long bones were removed and transferred into ice-cold PBS. Marrow was forced out of the bones into α-MEM containing 100 U/ml penicillin, 100 μg/ml streptomycin, 10% fetal calf serum, and 20 mM HEPES (1.5 ml per chick), using the end of a 5 ml syringe plunger. The cell suspension was disaggregated using a plastic pasteur pipette and bone fragments were allowed to settle out for 15 minutes and then the top 25% of the medium was discarded. The remaining cell suspension was then seeded into 24-well plates (0.5 ml/well). The medium was replaced 75 minutes later and then the cells (a mixed population of osteoclasts and bone marrow cells) were treated in fresh medium the following day.

Incorporation of [14C]mevalonate into prenylated proteins in osteoclasts Detection of prenylated proteins in rabbit osteoclasts was carried out as described previously for J774 macrophages.(30) Purified rabbit osteoclasts in 6-well plates were depleted of mevalonate by incubation in α-MEM containing 10% fetal calf serum and 5 μM mevastatin for 4 h. The medium was then replaced with 1.0 ml per well fresh α-MEM/10% fetal calf serum containing 5 μM mevastatin, 7.5 μCi/ml [14C]mevalonic acid lactone (specific activity 57 mCi/mmol) plus bisphosphonates (duplicate wells per treatment). After 40 h the cells were lysed in 0.2 ml per well 1% (vol/vol) NP-40, 0.1% (wt/vol) sodium dodecyl sulfate (SDS), 0.5% (wt/vol) sodium deoxycholate, and 10 μg/ml phenylmethylsulfonyl fluoride in PBS, pooling the lysate from duplicate wells. Lysates were concentrated to approximately 50 μl by centrifuging (10,000g) through 10-kDa cut-off microconcentrators (Whatman, Maidstone, U.K.). The protein concentration of the lysates was determined using the BCA protein assay (Pierce, Rockford, IL, U.S.A.) and then 50 μg of each osteoclast lysate was electrophoresed on 12% polyacrylamide-SDS gels under reducing conditions. After electrophoresis, the gels were fixed in 10% (vol/vol) acetic acid, 40% (vol/vol) methanol, and 50% (vol/vol) distilled water and then dried and labeled proteins were visualized on a BioRad Personal FX Imager (Hemel-Hempstead, UK) after exposure to a Kodak (Hemel-Hempstead, UK) phosphorimaging screen.

Inhibition of protein prenylation by FTI-277 and GGTI-298 in osteoclasts Osteoclasts were labeled metabolically with [3H]FOH or [3H]GGOH to determine the concentration of FTI-277 and GGTI-298, which inhibits protein farnesylation and geranylgeranylation, respectively. Purified rabbit osteoclasts in 6-well plates were depleted of mevalonate as described above and then incubated with 5 μM or 10 μM FTI-277 or GGTI-298 in the presence of 30 μCi/ml [3H]FOH or [3H]GGOH plus 5 μM mevastatin. Cell lysates were prepared 40 h later (or 16 h later in the case of cells treated with GGTI-298, which caused osteoclast apoptosis in longer incubations) and 60 μg protein from each sample were separated by electrophoresis on a 12% polyacrylamide gel as described above. Gels were then fixed, incubated in Enhance for 30 minutes, and dried. [3H]-labeled proteins were visualized by exposing the gel to preflashed Hyperfilm-MP for 6 days at −70°C.

Effects of FTI-277 and GGTI-298 on osteoclast formation Osteoclasts were generated in vitro using the mouse bone marrow/osteoblast coculture previously described.(31) On the sixth day of the coculture, FTI-277 or GGTI-298 were added to a final concentration of 10 μM in quadruplicate wells. After a further 48 h, cultures were fixed for 10 minutes in buffered formalin and stained for TRAP by incubating with naphthol-AS-BI-phosphate, hexazotized pararosanilin, and 30 mM tartrate in acetate buffer (pH 5.5) at 37°C for 30 minutes.(29) The cultures were rinsed with PBS and the number of osteoclasts, defined as TRAP-positive, multinucleated cells (>2 nuclei/cell), were counted using a Zeiss Axiovert microscope (Welwyn Garden City, UK).

Effects of FTI-277 and GGTI-298 on osteoclast morphology and osteoclast number Representative photographic images of an individual live chick osteoclast treated with FTI-277 or GGTI-298 were taken after 20–28 h using a Zeiss Axiovert microscope equipped with an Optronics CCD camera. To assess osteoclast number, chick and rabbit osteoclasts in 24-well plates were treated in quadruplicate with FTI-277 or GGTI-298 for 24 h, the medium containing nonadherent cells was removed, and the remaining adherent cells were fixed for 5 minutes in 4% (vol/vol) formaldehyde in PBS. Cells were then stained with 1 μg/ml 4,6-diamidino-2-phenylindole (DAPI) in PBS for 10 minutes and the number of multinucleated (>2 nuclei) osteoclasts retaining normal nuclear morphology (i.e., without pyknotic or fragmenting nuclei) was counted. The results were then expressed as the mean number of nonapoptotic osteoclasts remaining per well. To visualize intracellular F-actin, fixed osteoclasts were permeabilized for 20 minutes with 0.2% Triton in PBS and then incubated for 30 minutes with 0.5 μg/ml tetramethylrhodamine isothiocyanate (TRITC)-phalloidin in PBS.

Effects of FTI-277 and GGTI-298 on osteoclasts cultured on ivory Rabbit osteoclasts were seeded on to 5-mm-diameter elephant tusk ivory slices in 96-well plates and allowed to adhere for 2 h. The medium was then removed and slices were rinsed in PBS and cultured with fresh α-MEM containing 5 μM and 10 μM FTI-277 or GGTI-298. forty-eight hours later, the medium was removed and the slices were fixed in 4% formaldehyde and then stained with DAPI and TRITC-phalloidin as described above. The number of actin rings, defined as a distinct and complete ring of podosomes(32), per slice was then counted. The slices were then stained for TRAP as above and the number of TRAP-positive, multinucleated cells per slice was counted. Resorptive activity of osteoclasts also was assessed as previously described.(29) Briefly, slices were immersed in 20% sodium hypochlorite and then pits were visualized by reflected light microscopy, and resorption was quantitated by measuring the area of the pits per slice using a Leitz Quantimet Q500MC image analysis system (Leitz, Milton Keynes, U.K.).

RESULTS Nitrogen-containing bisphosphonates inhibit protein prenylation in purified osteoclasts Electrophoretic analysis of lysates from purified rabbit osteoclasts that had been labeled metabolically with [14C]mevalonate showed major radiolabeled bands of molecular mass 17 kDa and approximately 21–26 kDa and minor bands of approximately 60–70 kDa (Fig. 1A). In accord with our previous studies(17,18) these represent farnesylated and geranylgeranylated small GTP-binding proteins (21-26 kDa)(33) and farnesylated lamin B and prelamin A (60-70 kDa).(34–36) In addition, the major radiolabeled band at the dye front represents mostly isoprene-containing intermediates of the mevalonate pathway, such as GGPP.(17) Treatment of osteoclasts for 40 h with 100 μM PAM, ALN, IBA, RIS, or ZOL markedly inhibited the incorporation of [14C]mevalonate into farnesylated and geranylgeranylated proteins and into the major band at the dye front. One hundred micromolar RIS, IBA, and ZOL appeared to be most effective, completely inhibiting prenylation, whereas 100 μM ALN and PAM were less effective, only partially inhibiting prenylation. For ZOL, a concentration as low as 10 μM inhibited protein prenylation, while 1 μM appeared to inhibit labeling of compounds at the dye front only (Fig. 1B). By contrast, treatment with 100 μM CLO, ETI, or TIL for 40 h did not affect protein prenylation in osteoclasts or the labeling of compounds at the dye front (Fig 1A).

Figure Figure 1. Inhibition of protein prenylation by bisphosphonates in purified rabbit osteoclasts. Cultures of purified osteoclasts were metabolically labeled for 40 h with [14C]mevalonate in the absence or presence of (A) 100 μM bisphosphonates or (B) 1–100 μM ZOL. Osteoclast lysates were analyzed by SDS-PAGE on 12% gels and radiolabeled, prenylated proteins were detected by phosphorimaging. Arrowheads indicate the position of molecular mass markers, arrows indicate the position of the labeled bands. To avoid possible experimental differences between gels in A, each gel contained lysates from osteoclasts treated with nitrogen-containing bisphosphonates (either ALN, RIS, PAM, IBA, or ZOL) and non-nitrogen-containing bisphosphonates (either CLO, ETI, or TIL). The data shown are representative of three independent experiments.

FTI-277 and GGTI-298 inhibit farnesylation and geranylgeranylation, respectively, in osteoclasts The ability of FTI-277 and GGTI-298 (specific peptidomimetic, cell-permeable inhibitors of FTase and GGTase I, respectively)(26,27) to inhibit farnesylation and geranylgeranylation in rabbit osteoclasts was investigated by metabolically labeling cells with [3H]FOH and [3H]GGOH. [3H]FOH was incorporated into proteins of molecular mass approximately 21 kDa, most likely Ras proteins(37) (data not shown) and a major band of approximately 65 kDa (nuclear lamin; Fig. 2A). Treatment of purified rabbit osteoclasts for 40 h with 5 μM or 10 μM FTI-277 completely inhibited the incorporation of [3H]FOH into the 65-kDa lamin protein. Because lamins are among the most resistant proteins to the inhibition of farnesylation, owing to the high affinity of FTase for lamin substrate,(20,38) the ability of 5 μM or more FTI-277 to inhibit lamin farnesylation indicates that these concentrations are sufficient to inhibit completely farnesylation in osteoclasts.

Figure Figure 2. Inhibition of protein farnesylation and geranylgeranylation in osteoclasts by FTI-277 and GGTI-298. Cultures of purified rabbit osteoclasts were metabolically labeled (A) for 40 h with [3H]FOH or (B) for 16 h with [3H]GGOH in the absence or presence of 5 μM or 10 μM FTI-277 or GGTI-298. Osteoclast lysates were separated by SDS-PAGE on 12% gels and [3H]-labeled, prenylated proteins were detected by fluorography. Arrowheads indicate the position of major labeled proteins; (A) approximately 65 kDa (nuclear lamin); (B) approximately 21 kDa (small GTP-binding proteins). The data shown are representative of three independent experiments.

[3H]GGOH was incorporated into proteins of approximately 21–26 kDa (small GTP-binding proteins) in osteoclasts (Fig. 2B). Five micromolar and 10 μM GGTI-298 effectively inhibited the incorporation of [3H]GGOH into a major band of these geranylgeranylated proteins on SDS-polyacrylamide gel electrophoresis (PAGE) gels, but a further band of small GTPase protein of slightly higher molecular mass was less affected. The latter band may include Rab proteins, which are geranylgeranylated by GGTase II (an enzyme that is not inhibited by GGTI-298(27)), or proteins more resistant to inhibition by GGTI-298 owing to greater affinity of GGTase I for these proteins. Nevertheless, ≥5 μM GGTI-298 was sufficient to inhibit geranylgeranylation of at least some proteins that are prenylated by GGTase I in osteoclasts.

Inhibition of protein geranylgeranylation prevents osteoclast formation To determine whether loss of farnesylated or geranylgeranylated proteins (or both) may account for the ability of bisphosphonates to inhibit osteoclast formation, we compared the effects of 10 μM FTI-277 and GGTI-298 on osteoclast formation in cocultures of bone marrow cells and calvaria-derived osteoblasts. FTI-277 and GGTI-298 were added on day 6 of the coculture (when fusion of mononuclear osteoclast precursors begins to occur) and cultures were incubated for a further 48 h. Although 10 μM FTI-277 had no effect on the number of multinucleated TRAP-positive osteoclasts formed compared with control, 10 μM GGTI-298 inhibited formation of murine osteoclasts by more than 70% (Fig. 3). Higher concentrations of GGTI-298 completely inhibited osteoclast formation (data not shown).

Figure Figure 3. Inhibition of osteoclast formation by GGTI-298 but not FTI-277. Murine calvarial osteoblasts and bone marrow cells were cocultured for 6 days and then treated with 10 μM GGTI-298 or FTI-277 for a further 48 h. Cultures were fixed and stained for TRAP and then the number of multinucleated, TRAP-positive cells per well was counted. The results are expressed as the mean ± SEM (n = 3).

Inhibition of protein geranylgeranylation causes cytoskeletal disruption and osteoclast apoptosis We next examined the effects of FTI-277 and GGTI-298 on isolated rabbit and chick osteoclasts in vitro. Chick osteoclasts treated with 10 μM GGTI-298 rapidly retracted and lost adherence, retraction typically beginning after about 20 h of culture (Fig. 4A). In contrast, the stromal cells also present in these cultures retained normal morphology for 48 h or longer. The ultimate fate of osteoclasts treated with GGTI-298 appeared to be apoptosis, shown by characteristic nuclear condensation and fragmentation observed after staining nuclei of rabbit osteoclasts with DAPI (Fig. 4B). Osteoclast apoptosis was assessed indirectly by counting the number of adherent, multinucleated cells that retained normal nuclear morphology after treatment. Twenty micromolar GGTI-298 reduced the number of adherent rabbit and chick osteoclasts by 75% and 66%, respectively, after 24 h of treatment (Fig. 4C), whereas 5 μM GGTI-298 reduced osteoclast number after 48 h but not 24 h (not shown). By contrast, concentrations up to 20 μM FTI-277 had little effect on the number of adherent rabbit or chick osteoclasts after 24 h (Fig. 4C) or 48 h.

Figure Figure 4. GGTI-298 causes cell retraction, loss of adherence, and apoptosis in osteoclasts. (A) An individual chick osteoclast (arrow) observed by phase contrast microscopy at the indicated time after commencing treatment with 10 μM GGTI-298, showing retraction, cytoplasmic condensation, and loss of adherence. (B) Nuclear morphology of DAPI-stained rabbit osteoclasts after treatment for 24 h with or without 20 μM GGTI-298. A single osteoclast is shown in each photograph. (C) Rabbit (open bars) and chick (hatched bars) osteoclasts in tissue culture plates were treated with 5 μM or 20 μM FTI-277 or GGTI-298 for 24 h and then remaining adherent cells were fixed and nuclei were stained with DAPI. Multinucleated cells (more than two nuclei) retaining normal nuclear morphology were then counted in each well and results were expressed as the mean ± SEM (n = 4).

Podosomes were observed around the periphery of both chick and rabbit osteoclasts on tissue culture plates after staining with TRITC-phalloidin. The formation of podosomes in chick osteoclasts was not affected by 20 μM FTI-277 but was disrupted after treatment with 20 μM GGTI-298 for 24 h, giving rise to diffuse staining of actin throughout the cytoplasm (Fig. 5A). Rabbit osteoclasts cultured on ivory slices also formed actin rings characteristic of polarized osteoclasts. As with the podosomes in osteoclasts cultured on plastic, these actin rings were disrupted completely by treatment with 10 μM GGTI-298, whereas 10 μM FTI-277 had no effect (Fig. 5B). The 5-μM and 10-μM GGTI-298 caused a dose-dependent reduction in the number of actin rings, of 39% and 85%, respectively, compared with control cultures. By contrast, neither 5 μM nor 10 μM FTI-277 affected the number of actin rings compared with control cultures (Fig. 6A).

Figure Figure 5. GGTI-298, but not FTI-277, disrupts actin distribution in osteoclasts. (A) Chick osteoclasts cultured on plastic plates and (B) rabbit osteoclasts cultured on ivory slices were treated with FTI-277 or GGTI-298 for 24 h and then fixed and stained with TRITC-phalloidin to visualize F-actin. Photographic images of representative osteoclasts are shown.

Figure Figure 6. GGTI-298 inhibits osteoclast activity and bone resorption and reduces osteoclast number on ivory slices. (A) Rabbit osteoclasts were seeded on to ivory slices and then treated for 48 h with 5 μM or 10 μM FTI-277 or with 5 μM or 10 μM GGTI-298. Slices were fixed and stained with TRITC-phalloidin and then the number of actin rings per ivory slice was counted (values are the mean ± SEM; n = 4). (B) Osteoclasts were then stained for TRAP and the number of TRAP-positive multinucleated osteoclasts per slice was counted (values are the mean ± SEM; n = 4). (C) Resorption pits were visualized by reflected light microscopy and the total resorbed area per slice was quantitated. Results are expressed as the mean resorbed area per slice ± SEM (n = 4).

After staining for TRAP, the total number of osteoclasts per slice was determined. On untreated slices, there were approximately four times as many osteoclasts as actin rings (Fig. 6B). Treatment with 10 μM GGTI-298 for 48 h reduced osteoclast number on ivory by 57%, whereas 5 μM had no effect. Neither 5 μM nor 10 μM FTI-277 affected osteoclast number (Fig. 6B).

Inhibition of protein geranylgeranylation inhibits bone resorption The importance of geranylgeranylated and farnesylated proteins in osteoclastic bone resorption was assessed by measuring total resorption pit area after incubation of rabbit osteoclasts on ivory slices. GGTI-298 dose dependently inhibited resorption, with 5 μM and 10 μM reducing pit area by 58% and 92%, respectively, compared with control (Fig. 6C). By contrast, neither 5 μM nor 10 μM FTI-277 had any effect on resorption pit area.

DISCUSSION The molecular mechanisms by which bisphosphonate drugs inhibit bone resorption have been a contentious issue for many years, largely because of the difficulty in isolating and culturing sufficient numbers of pure, authentic osteoclasts. Using the macrophage cell line J774 as a surrogate osteoclast model with which to study the mechanism of bisphosphonate-induced apoptosis, we recently found that nitrogen-containing bisphosphonates such as RIS, ALN, and IBA prevented protein prenylation in intact cells metabolically labeled with [14C]mevalonate.(17) In addition, the induction of J774 apoptosis could be partially overcome by addition of FPP or GGPP, suggesting that apoptosis could be the result of inhibition of protein farnesylation and/or geranylgeranylation.(17) Furthermore, we found a clear correlation between the ability of bisphosphonates to inhibit protein prenylation in macrophages and the ability to inhibit bone resorption, suggesting that nitrogen-containing bisphosphonates also could inhibit bone resorption by this mechanism.(18) In later studies by Fisher et al. (24) and van Beek et al.,(25) the inhibitory effect of the nitrogen-containing bisphosphonates ALN and IBA on osteoclast formation and bone resorption in vitro could be prevented by replenishing cells with GGOH but not FOH, indicating that ALN and IBA affect osteoclasts and osteoclast precursors by inhibiting enzymes of the mevalonate pathway, and suggesting that loss of protein geranylgeranylation in osteoclasts may be of greater consequence than loss of protein farnesylation.

In the present study, we have confirmed that nitrogen-containing bisphosphonates inhibit protein prenylation in osteoclasts. One hundred micromolar RIS, ZOL, IBA, ALN, and PAM (a concentration that causes osteoclast apoptosis(11) and caspase activation in vitro (H.L. Benford and M.J. Rogers, unpublished data, 1999) and could be achieved in the vicinity of osteoclasts in vivo (8)) inhibited the incorporation of [14C]mevalonate into both farnesylated and geranylgeranylated proteins in cultures of purified rabbit osteoclasts. At this concentration, ZOL, IBA, and RIS appeared to be more effective than PAM and ALN, in accord with their relative antiresorptive potencies. ZOL inhibited protein prenylation in osteoclasts even at a concentration of 10 μM, in agreement with our previous demonstration that RIS inhibits protein prenylation in macrophages at concentrations ≥10 μM.(17) In contrast to the nitrogen-containing bisphosphonates, 100 μM CLO, ETI, and TIL had little effect on protein prenylation in osteoclasts, even though this concentration causes osteoclast apoptosis (H.L. Benford and M.J. Rogers, unpublished data, 1999).(11) The lack of effect of CLO and ETI on protein prenylation is consistent with the evidence that CLO and ETI do not inhibit FPP synthase or protein prenylation in J774 macrophages(17,39,40) and have a different molecular mechanism of action to the nitrogen-containing bisphosphonates.(41,42) Electrophoretic analysis of osteoclast lysates suggested that RIS, ZOL, IBA, ALN, and PAM also prevented the incorporation of [14C]mevalonate into isoprene-containing intermediates of the mevalonate pathway and therefore prevent prenylation by preventing the synthesis of FPP and GGPP rather than by inhibiting FTase or GGTase I or II. Recent evidence that nitrogen-containing bisphosphonates (including PAM, IBA, and RIS) can inhibit FPP synthase(39,40,43) supports this conclusion.

Although nitrogen-containing bisphosphonates inhibit both farnesylation and geranylgeranylation in osteoclasts, it appears that inhibition of osteoclast formation is mainly a consequence of loss of geranylgeranylated proteins. Concentrations of 5–10 μM GGTI-298, a cell-permeable, peptidomimetic inhibitor of GGTase I,(27) inhibited the geranylgeranylation of some small GTP-binding proteins in purified rabbit osteoclasts (although other proteins, probably Rab GTPases, which are prenylated by GGTase II, were less affected). These concentrations of GGTI-298 inhibited the formation of multinucleated, osteoclast-like cells in vitro (although we did not determine whether this was because of a decreased fusion of precursors or an increased apoptosis of newly formed osteoclasts). By contrast, 10 μM FTI-277 (an inhibitor of FTase), which completely inhibited protein farnesylation in rabbit osteoclasts, had no effect on osteoclast formation. Other inhibitors of the mevalonate pathway, statins and ketoconazole, also prevent osteoclast formation in vitro.(17,24,44,45) Although Sato et al. concluded that simvastatin prevented multinucleated osteoclast formation by decreasing the level of cholesterol required for membrane fusion of precursors,(44) Fisher et al. recently showed that lovastatin-mediated inhibition of osteoclast formation in murine bone marrow cocultures could be overcome by the addition of geranylgeraniol.(24) The latter observation is consistent with the findings described here that geranylgeranylated proteins are required for osteoclast formation.

As well as preventing osteoclast formation, GGTI-298 reduced the number of viable chick and rabbit osteoclasts in culture by causing apoptosis. These observations are consistent with other studies in which inhibition of protein geranylgeranylation by GGTI-298 caused apoptosis.(46–48) GGTI-298 also caused disruption of podosomes in chick and rabbit osteoclasts cultured on tissue culture plastic and reduced the number of osteoclasts with actin rings, a characteristic cytoskeletal feature of polarized osteoclasts,(32,49) when cultured on a resorbable substrate (ivory). Approximately 25% of rabbit osteoclasts seeded on ivory formed actin rings, which is similar to the proportion of rat osteoclasts found to form actin rings on ivory.(50) Because other studies have shown that bisphosphonates disrupt actin ring formation,(8) the finding that GGTI-298 also disrupts actin rings reaffirms our hypothesis that nitrogen-containing bisphosphonates can indirectly affect the osteoclast cytoskeleton by preventing protein geranylgeranylation.(17) GGTI-298 also has been reported to disrupt stress fibers in NIH3T3 cells,(51) presumably because proteins modified by geranylgeranylation, such as Rac and Rho, are known to play important roles in the regulation of the actin cytoskeleton.(52,53)

GGTI-298 dose dependently inhibited bone resorption by rabbit osteoclasts in vitro. This effect correlated with the reduction in the number of osteoclast actin rings, consistent with the requirement of the actin ring for bone resorption.(49) Inhibition of bone resorption by 20 μM GGTI-298 also was associated with a decrease in osteoclast number. However, a lower concentration of GGTI-298 (5 μM) inhibited bone resorption and disrupted actin ring formation without affecting osteoclast number, suggesting that disruption of the cytoskeleton by GGTI-298 is sufficient to inhibit osteoclast activity without causing apoptosis.

Because, in contrast to the effects of GGTI-298, FTI-277 had little effect on the morphology, actin organization, resorptive activity, or apoptosis of osteoclasts at concentrations that inhibited protein farnesylation, it appears that geranylgeranylated proteins (specifically, those prenylated by GGTase I) rather than farnesylated proteins are essential for osteoclast function. The role of geranylgeranylated proteins in osteoclasts has not been studied extensively, although Zhang et al. have shown that specific inhibition of Rho activity using C3 exoenzyme from Clostridium botulinum caused disruption of the osteoclast cytoskeleton and inhibited bone resorption by murine osteoclasts.(54) Together with our observations, this suggests that the loss of geranylgeranylation of Rho proteins could be responsible for the cytoskeletal effects of nitrogen-containing bisphosphonates and GGTI-298 in osteoclasts.

In conclusion, it is clear from our studies that geranylgeranylated proteins, rather than farnesylated proteins, are fundamental to osteoclast function. Furthermore, because we have shown conclusively that nitrogen-containing bisphosphonates inhibit protein prenylation in osteoclasts, the potent, antiresorptive property of these compounds is likely to be caused by the loss of protein geranylgeranylation. Further studies are required to elucidate the molecular events that occur after loss of protein geranylgeranylation in osteoclasts.

Acknowledgements This work was supported by project and equipment grants from the Arthritis Research Campaign (CO583, RO568, RO552, and HO535) and the National Cancer Institute, NIH RO1 grant CA-67771 (S.M.S.).

REFERENCES 1 Rogers MJ, Frith JC, Luckman SP, Coxon FP, Benford HL, Monkkonen J, Auriola S, Chilton KM, Russell RG 1999 Molecular mechanisms of action of bisphosphonates. Bone 24:73. S-79S 2 Smith R, Russell RGG, Bishop M 1971 Diphosphonates and Paget's disease of bone. Lancet 1:945–947. 3 Fleisch H, Russell RGG, Francis MD 1969 Diphosphonates inhibit hydroxyapatite dissolution in vitro and bone resorption in tissue culture and in vivo. Science 165:1262–1264. 4 Shinoda H, Adamek G, Felix R, Fleisch H, Schenk R, Hagan P 1983 Structure-activity relationships of various bisphosphonates. Calcif Tiss Int 35:87–99. 5 Geddes AD, D'Souza SM, Ebetino FH, Ibbotson KJ 1994 Bisphosphonates: Structure-activity relationships and therapeutic implications In: HeerscheJNM, KanisJA, (eds.) Bone and Mineral Research. Elsevier Science BV Amsterdam, pp. 265–306. 6 Flanagan AM, Chambers TJ 1991 Inhibition of bone resorption by bisphosphonates: Interactions between bisphosphonates, osteoclasts, and bone. Calcif Tiss Int 49:407–415. 7 Murakami H, Takahashi N, Sasaki T, Udagawa N, Tanaka S, Nakamura I, Zhang D, Barbier A, Suda T 1995 A possible mechanism of the specific action of bisphosphonates on osteoclasts: tiludronate preferentially affects polarized osteoclasts having ruffled borders. Bone 17:137–144. 8 Sato M, Grasser W, Endo N, Akins R, Simmons H, Thompson DD, Golub E, Rodan GA 1991 Bisphosphonate action. Alendronate localization in rat bone and effects on osteoclast ultrastructure. J Clin Invest 88:2095–2105. 9 Zimolo Z, Wesolowski G, Rodan GA 1995 Acid extrusion is induced by osteoclast attachment to bone. Inhibition by alendronate and calcitonin. J Clin Invest 96:2277–2283. 10 Selander KS, Monkkonen J, Karhukorpi EK, Harkonen P, Hannuniemi R, Vaananen HK 1996 Characteristics of clodronate-induced apoptosis in osteoclasts and macrophages. Mol Pharmacol 50:1127–1138. 11 Hughes DE, Wright KR, Uy HL, Sasaki A, Yoneda T, Roodman GD, Mundy GR, Boyce BF 1995 Bisphosphonates promote apoptosis in murine osteoclasts in vitro and in vivo. J Bone Miner Res 10:1478–1487. 12 Boonekamp PM, van der Wee-Pals LJ, van Wijk-van Lennep MM, Thesing CW, Bijvoet OLM 1986 Two modes of action of bisphosphonates on osteoclastic resorption of mineralized matrix. Bone Miner 1:27–39. 13 Sahni M, Guenther HL, Fleisch H, Collin P, Martin TJ 1993 Bisphosphonates act on rat bone resorption through the mediation of osteoblasts. J Clin Invest 91:2004–2011. 14 Ebetino FH, Francis MD, Rogers MJ, Russell RGG 1998 Mechanisms of action of etidronate and other bisphosphonates. Rev Contemp Pharmacother 9:233–243. 15 Amin D, Cornell SA, Gustafson SK, Needle SJ, Ullrich JW, Bilder GE, Perrone MH 1992 Bisphosphonates used for the treatment of bone disorders inhibit squalene synthase and cholesterol biosynthesis. J Lipid Res 33:1657–1663. 16 Amin D, Cornell SA, Perrone MH, Bilder GE 1996 1-Hydroxy-3-(methylpentylamino)-propylidene-1,1-bisphos-phonic acid as a potent inhibitor of squalene synthase. Arzneimittelforschung 46:759–762. 17 Luckman SP, Hughes DE, Coxon FP, Russell RGG, Rogers MJ 1998 Nitrogen-containing bisphosphonates inhibit the mevalonate pathway and prevent post-translational prenylation of GTP-binding proteins, including Ras. J Bone Miner Res 13:581–589. 18 Luckman SP, Coxon FP, Ebetino FH, Russell RGG, Rogers MJ 1998 Heterocycle-containing bisphosphonates cause apoptosis and inhibit bone resorption by preventing protein prenylation: evidence from structure-activity relationships in J774 macrophages. J Bone Miner Res 13:1668–1678. 19 Clarke S 1992 Protein isoprenylation and methylation at carboxyl-terminal cysteine residues. Annu Rev Biochem 61:355. −86:355−386 20 Gibbs JB, Oliff A 1997 The potential of farnesyltransferase inhibitors as cancer chemotherapeutics. Annu Rev Pharmacol Toxicol 37:143–166. 21 Zhang FL, Casey PJ 1996 Protein prenylation: Molecular mechanisms and functional consequences. Annu Rev Biochem 65:241–269. 22 Casey PJ 1995 Protein lipidation in cell signaling. Science 268:221–225. 23 Seabra MC 1998 Membrane association and targeting of prenylated Ras-like GTPases. Cell Signal 10:167–172. 24 Fisher JE, Rogers MJ, Halasy JM, Luckman SP, Hughes DE, Masarachia PJ, Wesolowski G, Russell RGG, Rodan GA, Reszka AA 1999 Alendronate mechanism of action: Geranylgeraniol, an intermediate in the mevalonate pathway, prevents inhibition of osteoclast formation, bone resorption, and kinase activation in vitro. Proc Natl Acad Sci U S A 96:133–138. 25 van Beek E, Lowik C, Van der Pluijm G, Papapoulos S 1999 The role of geranylgeranylation in bone resorption and its suppression by bisphosphonates in fetal bone explants In vitro: A clue to the mechanism of action of nitrogen-containing bisphosphonates. J Bone Miner Res 14:722–729. 26 Lerner EC, Qian Y, Blaskovich MA, Fossum RD, Vogt A, Sun J, Cox AD, Der CJ, Hamilton AD, Sebti SM 1995 Ras CAAX peptidomimetic FTI-277 selectively blocks oncogenic Ras signaling by inducing cytoplasmic accumulation of inactive Ras-Raf complexes. J Biol Chem 270:26802–26806. 27 McGuire TF, Qian Y, Vogt A, Hamilton AD, Sebti SM 1996 Platelet-derived growth factor receptor tyrosine phosphorylation requires protein geranylgeranylation but not farnesylation. J Biol Chem 271:27402–27407. 28 Tezuka K, Sato T, Kamioka H, Nijweide PJ, Tanaka K, Matsuo T, Ohta M, Kurihara N, Hakeda Y, Kumegawa, M 1992 Identification of osteopontin in isolated rabbit osteoclasts. Biochem Biophys Res Commun 186:911–917. 29 van't Hof RJ, Tuinenburg-Bol Raap AC, Nijweide PJ 1995 Induction of osteoclast characteristics in cultured avian blood monocytes; modulation by osteoblasts and 1,25-(OH)2 vitamin D3. Int J Exp Pathol 76:205–214. 30 Coxon FP, Benford HL, Russell RGG, Rogers MJ 1998 Protein synthesis is required for caspase activation and induction of apoptosis by bisphosphonate drugs. Mol Pharmacol 54:631–638. 31 van't Hof RJ, Ralston SH 1997 Cytokine induced nitric oxide inhibits bone resorption by inducing apoptosis of osteoclast progenitors and suppressing osteoclast activity. J Bone Miner Res 12:1797–1804. 32 Turksen K, Kanehisa J, Opas M, Heersche JN, Aubin JE 1988 Adhesion patterns and cytoskeleton of rabbit osteoclasts on bone slices and glass. J Bone Miner Res 3:389–400. 33 Maltese WA, Sheridan KM, Repko EM, Erdman RA 1990 Post-translational modification of low molecular mass GTP-binding proteins by isoprenoid. J Biol Chem 265:2148–2155. 34 Wolda SL, Glomset JA 1988 Evidence for modification of lamin B by a product of mevalonic acid. J Biol Chem 263:5997–6000. 35 Farnsworth CC, Wolda SL, Gelb MH, Glomset JA 1989 Human lamin B contains a farnesylated cysteine residue. J Biol Chem 264:20422–20429. 36 Beck LA, Hosick TJ, Sinensky M 1988 Incorporation of a product of mevalonic acid metabolism into proteins of Chinese hamster ovary cell nuclei. J Cell Biol 107:1307–1316. 37 Casey PJ, Solski PA, Der CJ, Buss JE 1989 p21ras is modified by a farnesyl isoprenoid. Proc Natl Acad Sci U S A 86:8323–8327. 38 Reiss Y, Goldstein JL, Seabra MC, Casey PJ, Brown MS 1990 Inhibition of purified p21ras farnesyl:protein transferase by Cys-AAX tetrapeptides. Cell 62:81–88. 39 Bergstrom JD, Bostedor RG, Masarachia PJ, Reszka AA, Rodan G 2000 alendronate is a specific, nanomolar inhibitor of farnesyl diphosphate synthase Arch Biochem Biophys 373:231–241. 40 van Beek E, Pieterman E, Cohen Lowik C, Papapoulos S 1999 Farnesyl pyrophosphate synthase is the molecular target of nitrogen-containing bisphosphonates. Biochem Biophys Res Commun 264:108–111. 41 Frith JC, Monkkonen J, Blackburn GM, Russell RGG, Rogers MJ 1997 Clodronate and liposome-encapsulated clodronate are metabolized to a toxic ATP analog, adenosine 5′-(β,γ-dichloromethylene) triphosphate, by mammalian cells in vitro. J Bone Miner Res 12:1358–1367. 42 Benford HL, Frith JC, Auriola S, Monkkonen J, Rogers MJ 1999 Farnesol and geranylgeraniol prevent activation of caspases by aminobisphosphonates: Biochemical evidence for two distinct pharmacological classes of bisphosphonate drugs. Mol Pharmacol 56:131–40. 43 Martin MB, Arnold W, Heath HT third, Urbina JA, Oldfield E 1999 Nitrogen-containing bisphosphonates as carbocation transition state analogs for isoprenoid biosynthesis. Biochem Biophys Res Commun 263:754–8. 44 Sato T, Morita I, Murota S 1998 Involvement of cholesterol in osteoclast-like cell formation via cellular fusion. Bone 23:135–140. 45 Fan X, Biskobing DM, Bain S, Rubin J 1996 Ketoconazole and phorbol myristate acetate regulate osteoclast precursor fusion in primary murine marrow culture. J Bone Miner Res 11:1274–1280. 46 Stark WW, Jr., Blaskovich MA, Johnson BA, Qian Y, Vasudevan A, Pitt B, Hamilton AD, Sebti SM, Davies P 1998 Inhibiting geranylgeranylation blocks growth and promotes apoptosis in pulmonary vascular smooth muscle cells. Am J Physiol 275:L55–L63. 47 Miquel K, Pradines A, Sun J, Qian Y, Hamilton AD, Sebti SM, Favre G 1997 GGTI-298 induces G0-G1 block and apoptosis whereas FTI-277 causes G2-M enrichment in A549 cells. Cancer Res 57:1846–1850. 48 Vitale M, Di Matola T, Rossi G, Laezza C, Fenzi G, Bifulco M 1999 Prenyltransferase inhibitors induce apoptosis in proliferating thyroid cells through a p53-independent CrmA-sensitive, and caspase-3-like protease-dependent mechanism. Endocrinology 140:698–704. 49 Lakkakorpi P, Tuukkanen J, Hentunen T, Jarvelin K, Vaananen K 1989 Organization of osteoclast microfilaments during the attachment to bone surface in vitro. J Bone Miner Res 4:817–825. 50 Burgess TL, Qian Y, Kaufman S, Ring BD, Van G, Capparelli C, Kelley M, Hsu H, Boyle WJ, Dunstan CR, Hu S, Lacey DL 1999 The ligand for osteoprotegerin (OPGL) directly activates mature osteoclasts. J Cell Biol 145:527–538. 51 Vogt A, Qian Y, McGuire TF, Hamilton AD, Sebti SM 1996 Protein geranylgeranylation, not farnesylation, is required for the G1 to S phase transition in mouse fibroblasts. Oncogene 13:1991–1999. 52 Hall A 1998 Rho GTPases and the actin cytoskeleton. Science 279:509–514. 53 Ridley AJ 1999 Rho family proteins and regulation of the actin cytoskeleton. Prog Mol Subcell Biol 22:1–22. 54 Zhang D, Udagawa N, Nakamura I, Murakami H, Saito S, Yamasaki K, Shibasaki Y, Morii N, Narumiya S, Takahashi N, Suda T. 1995 The small GTP-binding protein, rho p21, is involved in bone resorption by regulating cytoskeletal organization in osteoclasts. J Cell Sci 108:2285–2292.