Dr Rogers receives grants from Novartis and Procter & Gamble. Dr Thompson has no conflict of interest.
The acute phase response is the major adverse effect of intravenously administered N-BPs. In this study we show that N-BPs cause γ,δ-T-cell activation and proliferation in vitro by an indirect mechanism through inhibition of FPP synthase, an effect that can be overcome by inhibiting HMG-CoA reductase with a statin. These studies clarify the probable initial cause of the acute phase response to N-BP drugs and suggest a possible way of preventing this phenomenon.
Introduction: The acute phase response is the major adverse effect of intravenously administered nitrogen-containing bisphosphonate drugs (N-BPs), used in the treatment of metabolic bone diseases. This effect has recently been attributed to their action as non-peptide antigens and direct stimulation of γ,δ-T-cells. However, because N-BPs are potent inhibitors of farnesyl diphosphate (FPP) synthase, they could cause indirect activation of γ,δ-T-cells owing to the accumulation of intermediates upstream of FPP synthase in the mevalonate pathway, such as isopentenyl diphosphate/dimethylallyl diphosphate, which are known γ,δ-T-cell agonists.
Materials and Methods: Peripheral blood mononuclear cells (PBMCs) were isolated from healthy volunteers and treated with N-BP, statin, or intermediates/inhibitors of the mevalonate pathway for 7 days in the presence of interleukin (IL)-2. Flow cytometric analysis of the T-cell-gated population was used to quantify the proportion of γ,δ-T-cells in the CD3+ population.
Results and Conclusions: The ability of N-BPs to stimulate proliferation of CD3+ γ,δ-T-cells in human PBMC cultures matched the ability to inhibit FPP synthase. γ,δ-T-cell proliferation and activation (interferon γ [IFNγ] and TNFα release) was prevented by mevastatin or lovastatin, which inhibit HMG-CoA reductase upstream of FPP synthase and prevent the synthesis of isopentenyl diphosphate/dimethylallyl diphosphate. Desoxolovastatin, an analog of lovastatin incapable of inhibiting HMG-CoA reductase, did not overcome the stimulatory effect of N-BP. Furthermore, statins did not prevent the activation of γ,δ-T-cells by a synthetic γ,δ-T-cell agonist or by anti-CD3 antibody. Together, these observations show that N-BPs indirectly stimulate the proliferation and activation of γ,δ-T-cells caused by inhibition of FPP synthase and intracellular accumulation of isopentenyl diphosphate/dimethylallyl diphosphate in PBMCs. Because activation of γ,δ-T-cells could be the initiating event in the acute phase response to bisphosphonate therapy, co-administration of a statin could be an effective approach to prevent this adverse effect.
BISPHOSPHONATES (BPs) are currently the most important class of drugs for inhibiting bone resorption in common metabolic bone disorders such as postmenopausal osteoporosis,(1–3) Paget's disease,(4, 5) tumor-associated osteolysis, and hypercalcemia.(6) BPs inhibit bone resorption by binding to bone mineral, from which they are internalized by bone-resorbing osteoclasts.(7) After cellular uptake, nitrogen-containing BPs (N-BPs) inhibit farnesyl diphosphate (FPP) synthase,(8-10) an enzyme in the mevalonate pathway that is required for the synthesis of cholesterol and the isoprenoid lipids FPP and geranylgeranyl diphosphate (GGPP). By inhibiting FPP synthase, N-BPs prevent the post-translational modification (isoprenylation) of small GTP-binding proteins that are necessary for osteoclast function.(11-13)
For Paget's disease and metastatic bone disease, the usual mode of administration of BPs is by intravenous infusion. For the treatment of postmenopausal osteoporosis (which currently involves oral administration of BPs), a once-yearly intravenous infusion may be a highly desirable therapeutic option in the future.(14) However, the major adverse effect of intravenous administration of N-BPs (such as pamidronate [PAM] and ibandronate [IBA]) is the development of an acute-phase response in approximately one-third of patients that receive the treatment for the first time.(15–17) The acute-phase response is characterized by a transient pyrexia and increased circulating levels of interleukin-6 (IL-6), TNFα, and interferon γ (IFNγ).(15, 17-19) The acute phase response is maximal within 28-36 h of intravenous administration and disappears 2-3 days later, despite continuing treatment.(16) The exact molecular basis for the acute-phase effect remains unclear. Several recent studies have suggested that N-BPs can directly activate and stimulate proliferation of γ,δ-T-cells(20, 21) by acting as non-peptide phosphoantigens (like isopentenyl diphosphate [IPP] or dimethylallyl diphosphate [DMAPP]) that bind to the γ,δ-T-cell receptor (TCR).(22) Engagement of the γ,δ-TCR by N-BPs could then directly activate and stimulate the proliferation of the major subset of γ,δ-T-cells in humans, the Vγ9Vδ2 subset(6) (also termed Vγ2Vδ2). However, because N-BPs can disrupt the mevalonate pathway by inhibiting FPP synthase,(8-10) it is also possible that stimulation of γ,δ-T-cells by N-BPs occurs through an indirect mechanism caused by altered intracellular synthesis of isoprenoid lipids or to inhibition of protein isoprenylation. If this were the case, modulation of the mevalonate pathway by other pharmacologic agents could be an approach to prevent the acute phase response to N-BPs.
In this study, we used a wide range of N-BPs and inhibitors of the mevalonate pathway (Fig. 1) to examine the relationship between inhibition of FPP synthase and stimulation of γ,δ-T-cell proliferation in 7-day cultures of human peripheral blood mononuclear cells (PBMCs) in vitro.
MATERIALS AND METHODS
Zoledronic acid (ZOL; the hydrated disodium salt) was from Novartis Pharma AG. NE-10790;(23) alendronate (ALN), clodronate (CLO), ibandronate (IBA), pamidronate (PAM), NE-21650, NE-10571, NE-11808, and NE-11809 were kindly provided by Procter & Gamble Pharmaceuticals. Isopentenyl diphosphate (IPP), dimethylallyl diphosphate (DMAPP), mevalonic acid lactone (MVA), mevastatin (MEV), lovastatin (LOV), squalene, zaragozic acid, 4-aminobutyl phosphonate, and recombinant human IL-2 were from Sigma. Mevastatin was converted from the lactone form to the free acid as previously described.(11) All-trans geranylgeraniol (GGOH) was from American Radiolabeled Chemicals. Desoxolovastatin and FTI-277/GGTI-298 were kind gifts of Dr G Weitz-Schmidt (Novartis Pharma AG, Basle, Switzerland) and Dr S Sebti (University of South Florida, Tampa, FL, USA), respectively. Bromohydrin pyrophosphate (BrHPP)(24) was kindly provided by Dr J-J Fournié and Dr C Belmant (Innate Pharma, Marseille, France). Stock solutions of mevalonic acid lactone, squalene, GGOH, IPP, and DMAPP were prepared in ethanol, and BPs were prepared as described previously.(11) Stock solutions of lovastatin (lactone form) and desoxolovastatin were prepared in DMSO.
PBMC isolation and culture conditions
Informed consent was obtained for the collection of peripheral blood from healthy volunteers, and PBMCs were isolated by Lymphoprep (Nycomed) density gradient separation. PBMCs were cultured in RPMI 1640 media supplemented with 100U/ml penicillin, 100 μg/ml streptomycin, 1 mM glutamine, 10% fetal calf serum (FCS), and 10 U/ml rhIL-2. Cells were routinely cultured at a concentration of 1 × 106 cells/ml in 24-well plates. Unless stated otherwise, cells were treated with 1 μM N-BP, 1 μM IPP, 1 μM DMAPP, 1 μM zaragozic acid, 5 μM FTI-277, 5 μM GGTI-298, 1 mM NE-10790, 1 μM mevastatin, lovastatin, or desoxolovastatin, or an anti-pan-CD3-coated magnetic beads (1:100 dilution; 2.5 × 106 beads/well; Dynal) for a period of 7 days, in the presence of 10U/ml rhIL-2, before harvesting and subsequent flow cytometric analysis. On day 4 of the culture period, one-half of the medium was replaced with fresh medium containing IL-2 with or without treatments. Total cell number was calculated using an improved Neubauer hemocytometer.
Flow cytometric analysis
PBMCs were harvested as above and washed twice with PBS containing 2% (wt/vol) bovine serum albumin (BSA), and 0.5 × 106 cells were dual-stained with 20 μl anti-CD3-FITC antibody and 10 μl anti-pan-γ,δ-TCR-PC5 antibody, or the respective isotype-matched controls (all from Coulter-Immunotech), in a final volume of 100 μl PBS/BSA. After another wash in PBS, cells were resuspended in PBS and fixed with formaldehyde (final concentration 1% vol/vol). Cells were analyzed using a FACSCalibur (BD Biosciences), using the FL1 detector for FITC and the FL3 detector for PC5. CellQuest (BD Biosciences) software was used for quantitative analysis of the T-cell-gated population.
Quantification of TNFα and IFNγ release
PBMCs were cultured in RPMI 1640 media supplemented with 100U/ml penicillin, 100 μg/ml streptomycin, 1 mM glutamine, 10% FCS, and 10U/ml rhIL-2. Cells were cultured at a density of 1 × 106 cells/ml in 24-well plates and treated with 1 μM ZOL ± 1 μM MEV, or 100 nM BrHPP ± 1 μM MEV, for 24 or 48 h, in the presence of 10U/ml rhIL-2. The medium was removed and centrifuged at 400g and then 13,000g to remove cells and particulate matter, and levels of TNFα and IFNγ in the supernatant were quantified with Quantikine Human TNFα and IFNγ Immunoassays (R&D Systems), according to the manufacturer's instructions.
Results are expressed as mean ± SE from experiments with PBMCs from four independent blood donors. To determine statistical significance, a one-way ANOVA was performed followed by a Bonferroni posthoc test. Unless indicated otherwise, asterisks show values significantly different from centrol. Values of p < 0.05 were considered significant.
N-BPs stimulate proliferation of γ,δ-T-cells in vitro
In cultures of human PBMCs, the proportion of γ,δ-T-cells was typically 3–8% of the CD3+ cell population (n = 4 different donors). Treatment with 1 μM of the N-BPs ZOL, IBA, ALN, and PAM significantly increased the proportion of γ,δ-T-cells to 30-74% of the CD3+ population in the presence of rhIL-2 (Fig. 2). The increase in the proportion of γ,δ-T-cells was comparable with the increase in total cell number in the cultures at the end of the experiments (data not shown) and therefore was not caused simply by a decrease in the number of α,β-T-cells.
The order of effectiveness of 1 μM N-BPs for eliciting proliferation of γ,δ-T-cells was similar to the order of potency for inhibiting recombinant human FPP synthase (ZOL > ALN ∼ IBA > PAM).(10) CLO, which lacks a nitrogen moiety and does not inhibit FPP synthase,(10) had no stimulatory effect on γ,δ-T-cells at a concentration of 1 (Fig. 2), 50 (Fig. 9), or 150 μM (data not shown). When T-cells were purified from PBMCs using a negative selection kit (Dynal), 1 μM N-BP did not stimulate the proliferation of γ,δ-T-cells (data not shown).
Only potent inhibitors of FPP synthase stimulate proliferation of γ,δ-T-cells
We next investigated whether minor changes in N-BP structure that affect the potency for inhibiting FPP synthase also affected the ability to stimulate γ,δ-T-cell proliferation; 1 μM NE-11808 (IC50 40 nM for rhFPP synthase)(10) significantly stimulated γ,δ-T-cell proliferation, whereas the closely related analog NE-11809 (IC50 2900 nM for rhFPP synthase)(10) had no effect (Fig. 3). Similarly, 1 μM NE-21650 (IC50 58 nM for rhFPP synthase)(25) significantly stimulated γ,δ-T-cell proliferation, whereas the closely related analog NE-10571 (IC50 20,000 nM for rhFPP synthase)(25) did not affect proliferation (Fig. 3).
Whereas 1 μM ALN stimulated proliferation of γ,δ-T-cells (Fig. 2), 4-aminobutyl phosphonate (a monophosphonate analog of ALN) had no effect on γ,δ-T-cells at concentrations up to 10 μM (data not shown).
Inhibition of protein isoprenylation does not stimulate γ,δ-T-cell proliferation
To determine whether the stimulatory effect of N-BPs on γ,δ-T-cell proliferation could be caused by inhibition of FPP synthase and loss of isoprenylated proteins, we compared the effects of N-BPs with specific protein:prenyl transferase inhibitors (5 μM FTI-277, 5 μM GGTI-298, and 1 mM NE-10790; concentrations that inhibit farnesyl transferase, geranylgeranyl transferase I, and Rab geranylgeranyltransferase, respectively, in intact cells).(13, 23) None of the protein:prenyl transferase inhibitors stimulated γ,δ-T-cell proliferation (Fig. 4). Furthermore, 1 μM mevastatin, which inhibits HMG-CoA reductase, thereby indirectly preventing protein isoprenylation (Fig. 1), did not stimulate γ,δ-T-cell proliferation (see below and Fig. 7). The lack of effect of these inhibitors of protein prenylation on γ,δ-T-cell proliferation was not caused by cytotoxicity, because they did not affect the total cell number at the end of the culture (data not shown).
Products of the mevalonate pathway downstream of FPP synthase do not prevent N-BP-induced γ,δ-T-cell proliferation
To determine whether the stimulatory effect of N-BPs on γ,δ-T-cell proliferation could be caused by inhibition of FPP synthase and altered flux through the mevalonate pathway, we examined whether replenishment of cells with products of the mevalonate pathway downstream of FPP synthase could abrogate the stimulatory effect. Neither 1 mM squalene, an intermediate in the biosynthesis of cholesterol, nor 10 μM GGOH (a cell-permeable form of GGPP, necessary for protein isoprenylation; Fig. 1) affected N-BP-induced proliferation of γ,δ-T-cells (Fig. 5).
Mevalonate stimulates γ,δ-T-cell proliferation
Although low concentrations (1–100 μM) of mevalonic acid lactone (a cell-permeable form of mevalonate) did not stimulate proliferation of γ,δ-T-cells, a higher concentration (1 mM) caused a consistent increase in the proportion of γ,δ-TCR+ cells in the CD3+ population (from ∼7% in control cultures to ∼17%; Fig. 6A). The well-characterized isoprenoid lipid phosphoantigens IPP and DMAPP, metabolites upstream of FPP synthase in the mevalonate pathway (Fig. 1), also stimulated γ,δ-T-cell proliferation (Fig. 6B). Consistent with a previous study(26) 1 μM IPP was significantly more effective at stimulating γ,δ-T-cell proliferation than 1 μM DMAPP.
Mevastatin abolishes the stimulatory effect of N-BP on γ,δ-T-cell proliferation
Although replenishment of cells with metabolites downstream of FPP synthase had no preventative effect, the stimulatory effect of three N-BPs (ALN, IBA, and ZOL) on γ,δ-T-cell proliferation was consistently abrogated by simultaneous treatment with 1 μM mevastatin (Figs. 7A and 8A), which inhibits the mevalonate pathway upstream of FPP synthase (Fig. 1). The ability of mevastatin to prevent the stimulatory effect of 1 μM N-BP was concentration-dependent and occurred at concentrations as low as 10 nM (Fig. 7B). This inhibitory effect of mevastatin was not caused by general cytotoxicity, because treatment with 1 μM mevastatin did not significantly inhibit the increase in proliferation of CD3+ T-cells in cultures of PBMCs in response to BrHPP (Fig. 7D), a synthetic agonist that is a more potent stimulator of γ,δ-T-cell proliferation than IPP.(24) Furthermore, 1 μM mevastatin did not significantly reduce the stimulation of γ,δ-T-cell proliferation induced by anti-CD3 antibody (Fig. 7C) or by treatment with 1 μM IPP or 1 μM DMAPP (data not shown).
The inhibitory effect of 1 μM mevastatin on ZOL-induced γ,δ-T-cell proliferation could be overcome by replenishing cells with 100 μM mevalonic acid lactone (Fig. 8A), a cell-permeable form of mevalonate, showing that mevastatin inhibits the proliferative response to ZOL caused by inhibition of HMG-CoA reductase, and consequently, decreased synthesis of mevalonate and its downstream metabolites. In addition, the attenuation of ZOL-induced γ,δ-T-cell proliferation by 1 μM mevastatin could be reversed by the addition of 1 μM IPP, indicating that the attenuative effect of mevastatin was not simply caused by enhanced cytotoxicity when in combination with ZOL (Fig. 8A).
One micromolar lovastatin (the lactone form) was also found to attenuate ZOL-induced proliferation in γ,δ-T-cells (Fig. 8B). As with mevastatin, this inhibitory effect could be reversed by replenishing cells with 100 μM mevalonic acid lactone (data not shown). However, 1 μM desoxolovastatin, which lacks the ability to inhibit HMG-CoA reductase,(27) did not have any inhibitory effect on ZOL-induced proliferation of γ,δ-T-cells (Fig. 8B).
CLO reduces the stimulatory effect of N-BP on γ,δ-T-cell proliferation
Although CLO alone did not stimulate γ,δ-T-cell proliferation (Figs. 2 and 9), co-treatment of PBMC cultures with 1 μM IBA + 50 μM CLO significantly reduced proliferation of γ,δ-T-cells compared with treatment with IBA alone (Fig. 9). The attenuative effect of CLO was not caused by cytotoxicity, because 50 μM CLO did not affect total cell number (data not shown).
Mevastatin abolishes TNFα and IFNγ release in response to ZOL but not BrHPP
Treatment with either 1 μM ZOL or 100 nM BrHPP caused a large increase in TNFα release in PBMC cultures after 24 h, and particularly after 48 h (Figs. 10A and 10B). Co-treatment with 1 μM mevastatin completely abolished TNFα release in ZOL-stimulated cultures, whereas TNFα release was not affected in BrHPP-stimulated cultures.
Treatment with 1 μM ZOL or 100 nM BrHPP also resulted in a large increase in IFNγ release in PBMC cultures after 24 h, and particularly after 48 h (Figs. 10C and 10D). As observed with TNFα release, co-treatment with 1 μM mevastatin completely abolished ZOL-induced IFNγ production but did not affect BrHPP-induced IFNγ production.
An acute phase response is the major adverse effect of intravenous administration of N-BP drugs in patients with tumor-associated osteolysis, hypercalcemia, or Paget's disease.(16, 17, 19–21) BPs that lack a nitrogen-containing moiety, such as CLO, do not seem to cause this response.(16, 20, 21) Recently, it was proposed that N-BPs may act as non-peptide phosphoantigens, capable of binding to γ,δ-TCR.(20, 21, 28) This would lead to activation and proliferation of γ,δ-T-cells, with subsequent release of pro-inflammatory cytokines such TNFα and IFNγ, the systemic levels of which are known to increase after N-BP treatment.(17-19) A variety of non-peptide antigens have been shown to activate γ,δ-T-cells, including alkylamines(29) and isoprenyl pyrophosphate monoesters such as IPP and DMAPP.(22, 29) We and others have shown recently that N-BPs probably resemble the structure of isoprenyl pyrophosphate monoesters such as IPP, DMAPP, or GPP because N-BPs are potent inhibitors of FPP synthase,(8-10) an enzyme of the mevalonate pathway that uses IPP, DMAPP, and GPP as substrates for conversion to FPP. Hence, N-BPs could bind to the γ,δ-TCR by mimicking isoprenyl pyrophosphate antigens such as IPP. However, this paradigm does not fully explain the ability of N-BPs to activate γ,δ-T-cells. The observation that only N-BPs (i.e., only the BPs that are nanomolar inhibitors of FPP synthase) stimulate γ,δ-T-cells(20, 21, 28) raises an alternative explanation for the previously reported in vitro effects on γ,δ-T-cells. Internalization of N-BPs by T-cells or other mononuclear PBMCs would rapidly lead to inhibition of FPP synthase, causing intracellular accumulation of isoprenyl pyrophosphate metabolites upstream of FPP synthase in the mevalonate pathway.(8) These metabolites include IPP and/or DMAPP, which are known to be potent agonists of the Vγ9Vδ2 TCR. Several studies have recently confirmed that N-BPs cause accumulation of isoprenoid lipids, such as IPP and DMAPP.(8, 9, 30, 31) Release or presentation of these metabolites could therefore account for the selective activation and proliferation of Vγ9Vδ2 T-cells. We addressed this possibility by examining in more detail the relationship between the ability to inhibit FPP synthase and the ability to stimulate proliferation of γ,δ-T-cells.
In this study, we confirmed that N-BPs that are potent, nanomolar inhibitors of FPP synthase (ZOL, ALN, IBA, and PAM) stimulate the proliferation of γ,δ-T-cells in cultures of human PBMCs in vitro, in accord with previous reports.(20, 21) The stimulation of γ,δ-T-cells was caused by a clinically relevant concentration of 1 μM N-BP (the plasma Cmax is approximately 1 μM for ZOL or IBA and 10 μM for PAM after standard intravenous administration).(32–35) The effectiveness of ZOL, ALN, IBA, and PAM for eliciting a response by γ,δ-T-cells matched the order of potency for inhibiting recombinant human FPP synthase. Furthermore, using two pairs of closely related N-BPs that differ markedly in the potency for inhibiting FPP synthase, NE-11808/NE-11809 and NE-21650/NE-10571,(10, 25) we found that only the analog capable of potently inhibiting FPP synthase could effectively stimulate proliferation of γ,δ-T-cells. CLO (even at concentrations up to 150 μM) and a monophosphonate analog of ALN, neither of which inhibit FPP synthase or the mevalonate pathway,(10, 11) had no stimulatory effect. The lack of effect of CLO on γ,δ-T-cells is consistent with the majority of previous studies,(16, 20, 21) although a brief report by Schilbach et al.(36) suggested that a concentration of approximately 150 μM stimulates positively selected γ,δ-T-cells. However, this concentration is likely higher than the plasma concentration achieved in patients in vivo (a 300-mg infusion of CLO has a Cmax of approximately 50 μM after 2 h).(37)
Disruption of the mevalonate pathway through inhibition of FPP synthase would be expected to have two major intracellular effects: first, loss of GGPP and FPP required for protein isoprenylation and cholesterol biosynthesis; and second, the intracellular accumulation of the isoprenoid pyrophosphate metabolites immediately proximal to FPP synthase in the mevalonate pathway (i.e., IPP and/or DMAPP). However, bypassing inhibition of FPP synthase by replenishment of cultures of PBMCs with GGOH, a substrate for protein isoprenylation, did not diminish the ZOL-stimulated proliferation of γ,δ-T-cells. Furthermore, FTI-277, GGTI-298, and NE-10790 (specific inhibitors of protein isoprenylation) or mevastatin (a nonspecific, indirect inhibitor of protein isoprenylation; Fig. 1) did not mimic the ability of ZOL to stimulate γ,δ-T-cells. Hence, stimulation of γ,δ-T-cells does not seem to be a consequence of inhibition of protein isoprenylation. Also, replenishment of cultures of PBMCs with squalene (a precursor of cholesterol) did not diminish the ZOL-stimulated proliferation of γ,δ-T-cells, and zaragozic acid (which prevents cholesterol synthesis by inhibiting squalene synthase; Fig. 1) did not affect γ,δ-T-cell proliferation. Hence, N-BPs do not stimulate γ,δ-T-cell proliferation by decreasing cholesterol biosynthesis.
To examine whether accumulation of IPP and/or DMAPP could account for the effects of N-BPs, we treated cultures of PBMCs with ZOL in the presence of 1 μM mevastatin. The latter is a specific inhibitor of HMG-CoA reductase, the rate-limiting enzyme in the mevalonate pathway. Inhibition of HMG-CoA reductase would be expected to prevent the ZOL-induced accumulation of IPP and DMAPP caused by decreased synthesis of mevalonate, a precursor of IPP/DMAPP (Fig. 1). Mevastatin consistently attenuated the ZOL-induced proliferation of γ,δ-T-cells. This attenuation could be overcome by addition of mevalonate, showing that the effect of mevastatin was caused by inhibition of HMG-CoA reductase.
Lovastatin (the lactone form, which is converted intracellularly to the acid form of lovastatin, capable of inhibiting HMG-CoA reductase) also attenuated the ZOL-induced stimulation of γ,δ-T-cell proliferation. However, desoxolovastatin (an analog of lovastatin that cannot be converted to the acid form and hence does not inhibit HMG-CoA reductase)(27) did not affect ZOL-stimulated γ,δ-T-cell proliferation. Statins have recently been reported to affect lymphocyte adhesion and co-stimulation through binding to the β2 integrin leukocyte function-associated antigen-1 (LFA-1) site,(27) subsequently disrupting ICAM-1 binding. However, the lack of any attenuative effect of desoxolovastatin (which still binds to the LFA-1 site) on ZOL-induced γ,δ-T-cell proliferation suggests that, at the concentrations used in this study, a disruption of LFA-1:ICAM-1 binding is not responsible for the abrogative effect of statins on γ,δ-T-cell proliferation. We therefore conclude that inhibition of HMG-CoA reductase, and the subsequent decrease in the synthesis and accumulation of IPP/DMAPP, is likely to be the mechanism whereby statins prevent the stimulatory effect of N-BP on γ,δ-T-cell proliferation. Although concentrations of statins higher than that used in this study have also been reported to decrease the proliferative response to IL-2 or phytohemagglutinin,(38) in our study, the low concentration of mevastatin that attenuated ZOL-induced γ,δ-T-cell proliferation (1 μM) did not significantly affect γ,δ-T-cell proliferation stimulated by BrHPP, anti-CD3 antibody, or IPP/DMAPP. Furthermore, addition of IPP could still stimulate γ,δ-T-cell proliferation in the presence of mevastatin + ZOL, indicating that the proliferative capacity of T-lymphocytes was not affected by treatment with mevastatin + ZOL. As well as preventing the stimulatory effect of N-BPs on γ,δ-T-cell proliferation, mevastatin also completely abolished N-BP-induced IFNγ and TNFα release in cultures of PBMCs. However, mevastatin treatment did not affect BrHPP-induced IFNγ and TNFα release, suggesting that the selective abrogative effect of mevastatin on N-BP-induced γ,δ-T-cell proliferation and activation is indeed caused by HMG-CoA reductase inhibition and the subsequent decreased synthesis of IPP/DMAPP.
Together, these observations show the likely cause of the acute phase response observed in patients after intravenous administration of N-BP drugs. This effect is dependent on the ability of N-BPs to inhibit the intracellular enzyme FPP synthase, either in γ,δ-T-cells or in other PBMCs. Inhibition of FPP synthase causes the accumulation of isoprenoid lipids (e.g., IPP and DMAPP, upstream of FPP in the mevalonate pathway) that then stimulate the proliferation and activation of γ,δ-T-cells. Because negatively selected T-cells failed to respond to either N-BP or IPP at concentrations up to 10 μM (data not shown), consistent with other studies,(39, 40) this suggests that either non-T-cells are the major source of IPP or that accessory cells are required for co-activation of γ,δ-T-cells by IPP.
Activation of γ,δ-T-cells by N-BPs may be the primary, initiating event that then triggers a cascade of other effects in vivo. In this study and in others,(17–19, 21, 31, 39, 41) N-BPs caused a substantial increase in the release of TNFα and IFNγ by PBMCs in vitro within 24 h, one of the main features of the acute phase reaction to N-BPs in vivo. Pyrexia and increased levels of CRP and IL-6 observed in vivo(16-19) may be a consequence of the increase in TNFα and IFNγ release. Consistent with this, we could not detect an increase in IL-6 release after N-BP treatment of PBMCs in vitro (data not shown). Because activation of lymphocyte subpopulations is also known to increase extravasation of these cells into peripheral tissues,(42) the decreased number of circulating lymphocytes after N-BP treatment that has been reported in some studies(16-18) may also be a consequence of γ,δ-T-cell activation and increased cytokine release.
While this manuscript was in preparation, Gober et al.(31) also reported that the N-BP-induced anti-tumor effect of γ,δ-T-cell clones in vitro is caused by the accumulation and recognition of endogenous mevalonate metabolites such as IPP in N-BP-treated tumor cell lines. Consistent with our study, the authors reported that mevastatin decreased TNFα release and γ,δ-T-cell activation by preventing the accumulation of IPP in tumor cells. Although very complementary, our study was performed with authentic human PBMCs from healthy donors rather than with T-cell clones and used concentrations of N-BPs and statins (1 μM) that are clinically relevant and known to be achieved in patients in vivo.(32–35, 43)
Although our study and that of Gober et al.(31) strongly suggest that N-BPs activate γ,δ-T-cells through accumulation of IPP, another report (using a short pulse of 1000 μM PAM on monocytic cells) suggests that N-BPs stimulate γ,δ-T-cells through extracellular presentation.(39) However, in the latter study, a high concentration of PAM would result in a significant amount being internalized by monocytic cells, even in 10 minutes. In confirmation of this, we have recently found that a fluorescently labeled N-BP can be rapidly internalized by monocytic cells in vitro (Coxon FP and Rogers MJ, unpublished data, 2003). PAM is a nanomolar inhibitor of FPP synthase(10); therefore, intracellular uptake of the N-BP would lead to immediate inhibition of FPP synthase, rapid accumulation of IPP, and subsequent activation of γ,δ-T-cells.
In our study with human, primary PBMC cultures, we showed for the first time that a high concentration (1 mM) of mevalonic acid lactone, a cell-permeable source of mevalonate, can also stimulate proliferation of γ,δ-T-cells in vitro. Because mevalonate does not closely resemble any other known γ,δ-TCR agonist (alkylamines or phosphomonoesters), it is likely that mevalonic acid lactone stimulates γ,δ-T-cell proliferation in vitro by bypassing the rate-limiting step in the mevalonate pathway and increasing the rate of synthesis of IPP/DMAPP in PBMCs.
Importantly, our studies also identify possible routes for preventing the acute phase response to N-BPs. First, stimulation of γ,δ-T-cell proliferation by IBA could be reduced in the presence of a 50-fold molar excess of CLO, possibly by competing for cellular uptake through a membrane transport mechanism and hence decreasing the intracellular concentration of IBA available for inhibiting FPP synthase.(44) However, co-administration of N-BP with CLO may not be desirable because our recent studies suggest that CLO can antagonize the effects of N-BPs on osteoclasts.(44) An alternative explanation for the antagonistic effect of CLO is that CLO antagonizes the agonistic effect of IPP/DMAPP on γ,δ-T-cells, because methylenebisphosphonate derivatives that are resistant to phosphate hydrolysis are known to antagonize IPP-mediated γ,δ-T-cell activation.(45) Second, proliferation and activation of γ,δ-T-cells in response to N-BP treatment in vitro could be effectively abolished by statins. The concentrations of statins that inhibited γ,δ-T-cell proliferation and activation in vitro (10 nM-1 μM) are similar to the peak plasma concentrations that could be achieved with statins currently licensed for the treatment of hypercholesterolemia.(43) Co-administration of a statin could, therefore, potentially prevent the acute phase response after N-BP infusion, thereby preventing the major side effect of this important class of drugs in patients with Paget's disease and tumor-associated bone disease. The ability to overcome the acute phase response is of even greater significance for the development of future treatments for postmenopausal osteoporosis, a disease affecting one in three elderly women, which is likely to involve intravenous infusion of N-BP.(14)
Finally, because activated γ,δ-T-cells possess potent antibacterial, antiparasitic, and anti-tumor activities,(46–51) there is considerable interest in the development of compounds such as BrHPP and N-BPs as immunostimulatory agents capable of initiating a γ,δ-T-cell-mediated response. However, because statins are among the most widely prescribed drugs worldwide, our demonstration that N-BP-induced activation of γ,δ-T-cells, unlike with true agonists such as BrHPP, occurs through an indirect mechanism and can be suppressed by statins indicates that N-BPs may not be ideal for development as vaccines for immunotherapy.
We thank Dr Sharon Gordon and Linda Duncan (University of Aberdeen, UK) for help with FACS analysis and Dr Martin Wilhelm (Klinikum Nuernberg, Germany), Dr Jean-Jacques Fournié (INSERM, Toulouse, France), and Dr Jonathan Green (Novartis Pharma, Switzerland) for critical comments on the manuscript. This work was funded by a studentship to KT from Novartis Pharma.