|Home | About | Journals | Submit | Contact Us | Français|
Thiazide diuretics are used, worldwide, as the first-choice drug for patients with uncomplicated hypertension. In addition to their anti-hypertensive actions, they increase bone mineral density and reduce the prevalence of fractures, indicating that thiazides may have a role in the management of postmenopausal osteoporosis. Traditionally, the bone-protective effects of thiazides have been attributed to an increase in renal calcium reabsorption, secondary to the inhibition of the sodium chloride cotransporter, NCC, expressed in the kidney distal tubule. Whether thiazides exert a direct osteoanabolic effect independently of their renal action is controversial. Here we demonstrate that freshly frozen sections of human and rat bone express NCC, principally in bone-forming cells, the osteoblasts. In primary and established culture models of osteoblasts, fetal rat calvarial (FRC) and human MG63 cells, NCC protein is virtually absent in proliferating cells while its expression is dramatically increased during differentiation. Thiazides directly stimulate the production of osteoblast markers, runt-related transcription factor 2 (runx2) and osteopontin, in the absence of a proliferative effect. Using overexpression/knockdown studies in FRC cells, we show that thiazides, but not loop diuretics, increase mineralized nodule formation acting on NCC. Overall, our study demonstrates that thiazides stimulate osteoblast differentiation and bone mineral formation independently of their renal actions. In addition to their use as part of a therapeutic treatment plan for elderly, hypertensive individuals, our discovery opens up the possibility that bone-specific drug targeting by thiazides may be developed for the prevention and treatment of osteoporosis in the patient population as a whole.
It has been known for several decades that treatment of hypertension with thiazides has the beneficial side effect of strengthening bone,1-5. To date, this bone-protective effect has been attributed to thiazides acting at the distal nephron to inhibit the Na+-Cl- cotransporter (NCC),6. In favor of this hypothesis are two observations. Firstly, patients with Gitelman’s syndrome, and the equivalent murine model, where NCC is non-functional due to a mutation in the SLC12A3 gene, exhibit an increased bone mineral density,7,8. Secondly, patients with pseudohypoaldosteronism type II, who present with an increased NCC activity, exhibit reduced bone mineral density,9. The mechanism by which genetic (Gitelman’s syndrome) or pharmacological (thiazide treatment) inhibition of NCC results in enhanced bone mineral density has been hypothesized to be due to increased circulating serum calcium levels,10,11. This model proposes that NCC inhibition in the distal tubule evokes hyperpolarization, increased electrical driving force for calcium reabsorption and a subsequent decrease in urinary calcium loss,6. However, in Gitelman’s patients, as in patients undergoing thiazide therapy, the expected increase in circulating parathyroid hormone levels which should accompany the increase in free ionized plasma calcium concentration is not seen. In fact, plasma parathyroid hormone levels are either decreased or unchanged and serum calcium levels remain essentially normal,8,11. These observations suggest that the increase in bone mineral density is probably not directly related to the enhanced renal tubular calcium transport, but to a direct action of thiazides on bone. Potential mechanisms by which thiazides may exert their effect on bone are via an inhibition of osteoclast-mediated bone resorption and/or by an increase in osteoblastic bone formation. Although in vitro studies indicate that thiazides are capable of reducing osteoclastic activity independently of NCC,12,13 anabolic effects of thiazides have not been demonstrated. Since NCC mRNA has been previously reported in a bone-derived cell line,14 we hypothesized that thiazides increase bone mineral density by interacting directly with an osteoblast NCC protein. Given the potential therapeutic importance of chronic thiazide treatment in the prevention of age-related osteoporosis, it was the objective of the present study to investigate the effects of thiazides by establishing NCC expression in human bone, and to determine the effects of thiazides on osteoblast proliferation, differentiation and mineralized nodule formation in osteoblast models.
For the immunological detection of NCC expression in skeletal tissue, we used freshly frozen rat femur, freshly frozen human mandible and EDTA-decalcified, wax-embedded rat femur. At low magnification, we observed NCC-immunoreactivity in human and rat cortical and trabecular bone, in both undecalcified, frozen (Figures 1a and 1b) and decalcified, paraffin wax-embedded human (not shown) and rat bone (Figure 1c). No non-specific immunoreactivity was detected when the primary antibodies were omitted (Figure 1d). NCC immunoreactivity was also observed in snap-frozen sections of rat femur (Figures 1e-i). Using two different antibodies raised against two different epitopes of either human or rat proteins, NCC-specific immunoreactivity was localized to cells of the osteoblastic lineage (particularly osteoblasts) in both human and rat bone (Figures 1a, 1g and upper panel of 1k). NCC protein was also present in some, but not all, osteocytes (Figures 1a, 1c and 1h), perhaps indicating that cells at different stages of differentiation display differential expression of this protein (see below). Occasionally, osteoclasts in the rat cryosections (Figure 1i), but not those in sections of decalcified bone (not shown), also expressed NCC; immunostaining in osteoclasts was absent in human freshly frozen and paraffin sections (not shown). Furthermore, no NCC immunoreactivity was detectable in either human or rat cartilage (not shown). Positive control experiments carried out on rat kidney cryosections (Figure 1j) confirmed the expression of NCC exclusively in the distal convoluted tubule,1 thus demonstrating the specificity of the antibody. To investigate the size of human NCC protein in osteoblasts, we performed western analysis on crude membrane-enriched fractions from human osteoblast-derived MG63 cells. Human kidney cortex was used as the control tissue. Results showed comparable immunoreactivities of the expected molecular weights for the NCC monomer (M) and dimer (D) (Figure 1k, upper panel). Figure 1k also shows that MG63 cells expressed the type 1 Na+-K+-2Cl- cotransporter (the ubiquitously expressed form, NKCC1, figure 1k, lower panel),1 while they lack the kidney-specific isoform of the type 2 Na+-K+-2Cl- cotransporter (NKCC2, data not shown),1. Western analysis was performed on homogenates of established osteoblast cellular models after they had reached confluence. These were rat osteosarcoma UMR-106 cell line, mouse osteoblast-derived 2T3 clonal cell line and fetal rat calvaria (FRC). To varying degrees, all of these osteoblast-derived lines expressed immunoreactivity of the expected size for NCC (Figure 2a). After enzymatic deglycosylation with N-glycanase, the molecular weight of the NCC monomeric protein in MG63 cells was reduced from ~160 kDa to ~120 kDa, a size close to the unglycosylated core protein. Interestingly, western analysis also showed that NCC protein expression levels in FRC cells (Figure 2c) and in MG63 cells (Figure 2d) were virtually undetectable in proliferating cells (i.e., up to 10 days in culture). In contrast, in post-confluent, differentiating FRC and MG63 cells, NCC protein expression levels increased up to 15 days post-confluency.
It is conceivable that the osteoanabolic effects of metolazone could be ascribed to stimulation of cell proliferation. In line with the absence of NCC expression in pre-confluent, proliferating osteoblasts, FRC cell proliferation was not affected by metolazone (1-100 μM) for up to two weeks in culture (Figure 3a, three independent experiments performed in triplicate). Type I collagen is the first marker of osteoblast differentiation,16. Our data show that neither metolazone, nor chlorothiazide had any effect on the production of Type I collagen (Figure 3b, three independent experiments performed in triplicate). In contrast, metolazone produced a concentration-dependent increase in the expression of later osteoblast differentiation markers, Runx2 (Figures 3c and 3d, five independent observations from three independent cell isolations) and osteopontin (Figures 3e and 3f, five independent observations from three independent cell isolations) in both FRC (Figures 3c and 3e) and MG63 (Figures 3d and 3f) cells. In addition, chronic metolazone treatment per se increased NCC expression in a dose-dependent fashion (not shown).
The effect of thiazides and of loop diuretics on mineralized nodule formation was tested in post-confluent FRC cells kept in culture up to 3 weeks and mineralization was visualized using von Kossa staining. Figure 4 shows that both metolazone and chlorothiazide treatment of FRC cells induced a dramatic, concentration-dependent increase in mineralized nodule formation. The figure also shows that, under the same experimental conditions, the loop diuretic bumetanide (BMT) did not increase mineralization of FRC cells. These results indicate that thiazides increase mineralization and strongly suggest that this effect is dependent upon NCC expression.
To confirm that NCC is required for the thiazide-evoked mineralization, we genetically manipulated NCC expression in post-confluent FRC cells. First, to rule out non-specific effects, we transfected post-confluent FRC cells with an empty vector (pcDNA3.1) and measured mineralization in the presence or absence of concentrations of metolazone known to evoke statistically significant effects in non-transfected cells (i.e., 10 μM, see Figure 4). Figure 5a shows that the effects of metolazone were maintained even after plasmid delivery. Indeed, 10 μM metolazone increased the number of mineralized nodules per well from 285.9±16.3 to 470.2±18.6 (n=9 observations from 3 independent cell isolations, p<0.05). In the absence of metolazone treatment, neither overexpression nor antisense knockdown evoked significant changes in mineralized nodule formation (Figure 5b). In contrast, overexpression of NCC increased metolazone-dependent mineralized nodule formation by ~50% (+49.3±13.0%, n=12 replicates from 4 independent cell isolations), an effect which was completely prevented by NCC knockdown (-13±8.7%,=12 replicates from 4 independent cell isolations, p<0.05).
Chronic thiazide treatment is associated with a reduction in the risk of hip and wrist fractures in postmenopausal women and elderly men,1-5. This bone-sparing effect is thought to occur through blockage of the renal sodium chloride cotransporter, NCC and subsequent reduction in urinary calcium excretion. Whether thiazides directly effect new bone formation independently of their renal action has never been demonstrated. Here, we show that NCC is expressed in freshly frozen and decalcified sections of human and rat bone, in cells of the osteoblast lineage, particularly osteoblasts and, to a lesser extent, in the osteocytes, in both rat and human bone.
Immunoblotting performed on crude membrane extracts of freshly isolated (FRC), osteosarcoma-derived (UMR-106) and virally transformed (2T3) cells confirmed that NCC in osteoblasts is of an equivalent molecular mass as its renal counterpart. The renal NCC contains two N-linked glycosylation sites which are important for sensitivity to thiazides,22. Enzymatic deglycosylation of NCC in MG63 cells demonstrated the presence of carbohydrate residues in the osteoblast protein. Taken together, these observations show that NCC is expressed in bone, and suggest that the osteoblast NCC, like the kidney NCC, may also be a target for thiazide diuretics. Thus, the bone-protective effects of thiazides may be due to their direct interaction with this protein in the osteoblasts.
Bone formation is characterized by a distinctive sequence of events beginning with the commitment of mesenchymal cells to osteoblast lineage, followed by osteoblastic proliferation and differentiation. This sequence of events culminates in the formation of mineralized extracellular matrix by terminally differentiated osteoblasts,16. Previous studies have ruled out an effect of hydrochlorothiazide on human bone marrow stromal cells, suggesting that the thiazide-dependent enhanced bone mineral density is not due to an increase in osteoblast progenitors,23.
However, the effects of thiazides on osteoblast proliferation are controversial, with either an increase or no change in proliferation having been reported,13,15 depending on species. Therefore, we investigated the effects of thiazides on osteoblast proliferation, differentiation and mineralization, in order to ascribe a potential role of NCC in each of these events. To exclude the possibility that the effects of thiazides could be indirect (i.e., a consequence of their renal actions), we carried out these studies in vitro, using both primary and established models of osteoblasts, of either rat or human derivation, namely FRC and MG63 cells, respectively. First, we tested the effects of metolazone and of chlorothiazide on FRC cells, and found that neither of these compounds affected proliferation rates for up to ten days in culture. Consistent with a lack of effect of thiazides on osteoblast proliferation, NCC protein expression levels in FRC cells and in MG63 cells were negligible during the proliferative phase (i.e., up to 10 days in culture). At this point, FRC cells had stopped proliferating and begun their differentiation process, evident from increased expression of alkaline phosphatase (not shown) and aggregation of cells into nodular areas. In post-confluent, differentiating osteoblasts, NCC protein expression levels in both FRC and MG63 cells gradually rose, and peaked at approximately two weeks post-confluency, indicating that NCC acts as a novel potential osteoblast differentiation marker. Having ascertained the presence of NCC in post-confluent, differentiating osteoblasts, we then tested the effects of thiazides on the expression levels of known osteoblast markers. Type I collagen is the major structural component of the organic matrix of bone and one of the earliest marker of osteoblast differentiation16,17. Our data show that, at a stage when NCC protein expression is absent (i.e., two days post-confluency), neither metolazone, nor chlorothiazide had any effect on the production of Type I collagen. In contrast, metolazone produced a concentration-dependent increase in the expression of Runx2, a master osteoblast-specific transcription regulator, in both FRC and MG63 cells. The ability of thiazides to regulate the osteoblast differentiation process was tested by measuring the levels of osteopontin, a soluble, secreted phosphoprotein that is a component of the bone mineralized extracellular matrix (also called bone sialoprotein I, secreted phosphoprotein, 2ar and bp69). In both FRC and MG63 cells, metolazone significantly increased osteopontin production. These data, taken together with our observations that metolazone was ineffective during stages when NCC expression was undetectable and up regulated NCC expression in MG63 cells, suggest that thiazides act directly on differentiating osteoblasts through NCC.
The ultimate osteoblast differentiation marker is the formation of new bone and FRC cells in culture form calcified nodules which can be visualized by von Kossa staining. Metolazone treatment of post-confluent FRC cells induced a dramatic, concentration-dependent increase in mineralized nodule formation. This effect was specific to thiazide diuretics as it could be mimicked by chlorothiazide but could not be emulated by the loop diuretic, bumetanide, even though our results show that osteoblast models express NKCC1, the molecular target for loop diuretics.
As proof of concept, we used plasmid delivery of NCC antisense and sense cDNAs in FRC cells and assessed the effects of metolazone on mineralized nodule formation. Overexpression of NCC resulted in a significant increase in metolazone-induced mineralized nodule formation. This increase was completely prevented by NCC knockdown with the antisense construct. This effect is even more striking if one considers that the efficiency of transfection of the plasmid in primary cells is only ~10% (estimated with co-transfection with a fluorescent reporter, not shown). The evidence that the increase in mineralization is only observed in the presence of thiazides suggests the possibility that NCC might act as receptor for thiazides, rather than a cotransporter although such an interpretation would not explain why Gitelman’s patients exhibit an increased bone mineral density.
Finally, it has been suggested that thiazides prevent bone loss because they reduce acid production by inhibiting carbonic anhydrase activity in osteoclasts,24. Interestingly, we demonstrate NCC immunostaining in some osteoclasts of cryoprepared rat femora, but not human bone. Given that osteoclast staining was observed in five different preparations, with the appropriate positive and negative controls, we believe that NCC immunofluorescence in a sub-population of osteoclasts is real. This observation opens the possibility that thiazides might affect osteoclastic function through NCC in addition to creating alkalinization of the resorption milieu. The dual action of thiazide drugs on both osteoblast and osteoclast function could account for the observed reduced remodeling in patients taking such drugs in the absence of changes in plasma circulating PTH levels.
The main finding of this study is the demonstration that thiazides directly stimulate osteoblast differentiation and mineral production independently of their renal action. This effect of thiazides is concentration-dependent, is not mimicked by loop diuretics, is not due to increased osteoblast proliferation and is enhanced by NCC overexpression. Together with the observations that thiazide treatment and inactivating mutations of NCC are associated with an increased bone mineral density in humans and in knockout murine models, our findings support a pivotal role for the osteoblast NCC in mediating thiazide-induced bone formation. Thiazide diuretics are inexpensive and exhibit a good safety profile. Our findings suggest that it might be possible to develop osteoblast-specific thiazides as part of osteoporosis prevention and therapeutic programs.
Sprague-Dawley rats (Charles River Laboratories, Wilmington, Kent, UK) were sacrificed by cervical dislocation and used in accordance to the UK Animals Scientific Procedures Act of 1986.
The human osteoblast cell line, MG-63 cells, was cultured as previously described,15. Fetal rat calvarial (FRC) cells were isolated as previously described,16 FRC cells, rat UMR-106 and mouse osteoblast-derived 2T3 cells were cultured as described elsewhere,17. Metolazone (MET, Sigma-Aldrich, Poole, Dorset, U.K.), was dissolved at 37°C for 2 h in the culture medium before being added to the cells. This procedure was repeated every 3 days. From confluency onwards, the media were supplemented with ascorbic acid (284 μM for MG-63, UMR-106 and 2T3 and 568 μM for FRC cells; Sigma-Aldrich) and β-glycerophosphate (3 mM; Sigma-Aldrich).
SDS-PAGE immunoblotting of MG-63, UMR-106, 2T3 and FRC cells (whole cell lysates), human and rat kidney was performed as previously described,18,19 using the following primary antibodies: affinity-purified rabbit anti-human NCC polyclonal antibodies (1:1,000)19, affinity-purified rabbit anti-rat NCC polyclonal antibodies (1:5,000)20, affinity-purified rabbit anti-rat NKCC1 (a gift of Dr. R. James Turner, NIDCR) (1:1,000), affinity-purified rabbit anti-rat NKCC2 polyclonal antibodies (1:1,000),21 mouse anti-human runx2 monoclonal antibody (a gift of Dr. Andree von Wijnen, University of Massachusetts, USA) (1:4,000), mouse anti-rat osteopontin monoclonal antibody (Iowa Hybridoma Bank) (1:4,000) and mouse anti-β actin monoclonal antibody (Abcam, Cambridge, Cambridgeshire, U.K.) (1:10,000). For SDS PAGE, samples were heated to 60°C for 10-15 min in a 5x Laemmli sample buffer, in the presence of DTT (30 mg/ml) or β-mercaptoethanol (143 mM). Proteins were resolved by SDS-PAGE electrophoresis and transferred onto nitrocellulose membranes prior to blocking (30 min.; phosphate-buffered saline containing 5% semi-skimmed milk powder or Odyssey blocking buffer; Li-Cor, Lincoln, Nebraska, U.S.A.) and antibody incubations (1-12 h). Membranes were washed in Tween-Tris-buffered saline (15 mM Tris, pH 8, 150 mM NaCl, 0.1% (v/v) Tween 20). Antibody binding was visualized by an enhanced chemiluminescence system (Amersham Pharmacia Biotech, Little Chalfont, Buckinghamshire, U.K.). Human kidney samples for immunoblotting were obtained from the unaffected portion of a kidney which had been resected because of a renal tumor (approved as exempt from review by Office of Human Subjects Research).
Undecalcified, snap-frozen rat femora (2 months) and human mandible resections and EDTA-decalcified rat femora were prepared and used for immunofluorescence and immunoperoxidase experiments as previously described17. Anti-human and anti-rat NCC polyclonal antibodies (see above) were used at 1:10 and 1:100 dilutions, respectively. The human mandible tissue was from neck resections from patients with squamous cell carcinoma invading bone (Ethical approval and signed informed patient consent were obtained).
Cells were plated in 12-well plates at a density of 5,000 cells/cm2 and were treated with metolazone (1-100 μM) for up to two weeks. At each time point, triplicate wells were trypsinized and the cells were counted with a Coulter counter (Beckman Coulter, High Wycombe, Buckinghamshire, U.K.).
The effects of metolazone on Coll1A content were quantified as measurements of the hydroxyproline content in FRC cells after 48 h treatment. Following HCl digestion for 24 h at 110°C, the samples were freeze-dried to remove the acid, diluted in distilled water, oxidized with chloramine followed by coupling with dimethylamino benzaldehyde at 70 °C for 10-20 min. The colored product was measured at 550 nm. Standards of 1-10 μgml-1 hydroxyproline were used to calculate the standard curve.
FRC and MG63 cells were cultured in 35 mm dishes (20,000 cells/cm2) and treated with metolazone or chlorothiazide (CTZ), for 2 days post-confluency for early differentiation experiments, or for 7 to 14 days post-confluency for mineralization experiments. For immunoblotting, the cells were washed in phosphate-buffered saline and lysed in RIPA buffer as described previously18. Semi-quantitative changes in Runx2 and osteopontin immunoreactivities were normalized for the levels of β-actin (mouse monoclonal, Abcam, Cambridge, Cambridgeshire, U.K.).
FRC cells were treated from confluency for up to three weeks. Mineralized nodules were visualized by von Kossa staining, as described previously,17,18. The images of mineralized nodules were captured on a flatbed scanner and image analysis software (Scion Image, Frederick, MD, USA) was used to count the number of mineralized nodules.
The statistical significance was assessed by one way ANOVA with the Tukey’s post-hoc test or with the Student’s unpaired t-test, as appropriate. Observations were considered to be statistically significant different for p values ≤ 0.05.
This work was funded by the Arthritis Research Campaign (R0626, to DR and DHC) and The Wellcome Trust (CRIG 070159, to DR and GG). The authors thank Prof. V. Duance and Dr. S. Gilbert, Cardiff University, for the help with the Collagen I assay, Dr. A. von Wijnen, University of Massachusetts, USA, for the gift of the Runx2 antibody, Dr. A. Mee, University of Manchester, UK, for the gift of MG-63 cells, Mrs. N Vasquez, National University of Mexico, for technical help and the Iowa Hybridoma Bank for the osteopontin antibody.