Thiazide Diuretics Affect Osteocalcin Production in Human Osteoblasts at the Transcription Level Without Affecting Vitamin D3 Receptors

Authors

  • D. Lajeunesse,

    Corresponding author
    1. Unité de recherche en Arthrose, Centre Hospitalier de l'Université de Montréal, Campus Notre-Dame, Montréal, Québec, Canada.
    • Daniel Lajeunesse, Ph.D. Centre de Recherche L.C. Simard Centre Hospitalier de l'Université Montréal Campus Notre-Dame Pavillon J.A. DeSève, Y-2605 1560, rue Sherbrooke Est Montreal (Quebec) Canada H2L 4M1
    Search for more papers by this author
  • A. Delalandre,

    1. Unité de recherche en Arthrose, Centre Hospitalier de l'Université de Montréal, Campus Notre-Dame, Montréal, Québec, Canada.
    Search for more papers by this author
  • S. E. Guggino

    1. Gastroenterology Unit, Johns Hopkins Medical School, Baltimore, Maryland, U.S.A.
    Search for more papers by this author

Abstract

Besides their natriuretic and calciuretic effect, thiazide diuretics have been shown to decrease bone loss rate and improve bone mineral density. Clinical evidence suggests a specific role of thiazides on osteoblasts, because it reduces serum osteocalcin (OC), an osteoblast-specific protein, yet the mechanisms implicated are unknown. We therefore investigated the role of hydrochlorothiazide (HCTZ) on OC production by the human osteoblast-like cell line MG-63. HCTZ dose-dependently (1–100 μM) inhibited 1,25-dihydroxyvitamin D3 [1,25(OH)2D3]- induced OC release by these cells (maximal effect, −40–50% and p < 0.005 by analysis of variance [ANOVA]) as measured by ELISA. This effect of HCTZ on OC release was caused by a direct effect on OC gene expression because Northern blot analysis revealed that OC messenger RNA (mRNA) levels were reduced in the presence of increasing doses of the diuretic (–47.2 ± 4.0%; p < 0.0001 by paired ANOVA with 100 μM HCTZ). HCTZ (100 μM) also stimulated calcium (Ca2+) uptake (8.26 ± 1.78 pmol/mg protein/15 minutes vs. 13.6 ± 0.49 pmol/mg protein/15 minutes; p < 0.05) in MG-63 cells. Reducing extracellular Ca2+ concentration with 0.5 mM EDTA or 0.5 mM ethylene glycol-bis(β-amino ethyl ether)-N,N,N',N'-tetraacetic acid (EGTA) only partly prevented the inhibitory effect of the diuretic on OC secretion (maximal effect, −22.5 ± 6.9%), suggesting that thiazide-dependent Ca2+ influx is not sufficient to elicit the inhibition of OC secretion. Because OC production is strictly dependent on the presence of 1,25(OH)2D3 in human osteoblasts, we next evaluated the possible role of HCTZ on vitamin D3 receptors (VDR) at the mRNA and protein levels. Both Northern and Western blot analyses showed no effect of HCTZ (1–100 μM) on VDR levels. The presence of EGTA in the culture media reduced slightly the VDR mRNA levels under basal condition but this was not modified in the presence of increasing levels of HCTZ. The OC gene promoter also is under the control of transcription factors such as Yin Yang 1 (YY1) and cFOS. Western blot analysis revealed no changes in YY1 levels in response to HCTZ either in the presence or in the absence of 0.5 mM EGTA in the culture media. In contrast, HCTZ induced a dose-dependent increase in cFOS levels (p < 0.002 by ANOVA), a situation prevented by incubation with EGTA. These studies indicate that HCTZ inhibits OC mRNA expression independently of an effect on VDR, YY1, or extracellular Ca2+ levels but involves changes in cFOS levels. As OC retards bone formation/mineralization, the inhibition of OC production by HCTZ could explain its preventive role in bone loss rate. (J Bone Miner Res 2000;15:894–901)

INTRODUCTION

Skeletal and joint ailments are becoming a major health and financial problem in our increasingly aging population. Among these health problems, osteoporosis and its associated fractures represent a major burden. Epidemiological studies indicate that thiazide diuretics may be beneficial for the treatment of osteoporosis and to reduce bone loss and bone fractures and may be indicated for women who cannot or do not want to take estrogens.(1–9) Indeed, increases in bone mineral density (measured by single-photon absorptiometry) in patients treated with thiazides alone or in combination with estrogens have been reported.(3,5) The long-term use of thiazides also resulted in a low prevalence of bone fractures in aged patients in two studies, whereas Adland-Davenport et al., using a small cohort of patients, failed to observe this trend.(6,7,10) However, a longitudinal study (5 years) with men comparing thiazide use to other antihypertensive medications showed a decreased bone-loss rate in thiazide users, whereas the other groups actually showed an increase in bone-loss rate, providing more evidence that thiazides reduce fracture risk by preserving bone mass.(8) LaCroix et al. also concluded that the use of thiazide diuretics reduced the risk of hip fracture by approximately one-third in both men and women.(9) The beneficial effect of thiazide may be linked to their ability to reduce bone resorption because in healthy men volunteers treated with thiazides and elevated 1,25-dihydroxyvitamin D3 [1,25(OH)2D3] doses serum indicators of bone resorption were reduced despite any significant changes in immunoreactive parathyroid hormone (PTH) levels.(11) A recent meta-analysis of clinical studies suggests that thiazides should be considered as part of an approach to osteoporotic fracture prevention, particularly in hypertensive subjects.(12)

The mechanism of action of thiazides on bone metabolism still remains uncertain. Long-term use of thiazides reduces serum PTH and osteocalcin (OC) levels in both aged men and aged women and in postmenopausal women.(13–15) It also reduces the synthesis of 1,25(OH)2D3 because of calcium retention induced by these diuretics.(15) Although 1,25(OH)2D3 stimulates OC synthesis in both rodent and human osteoblasts, which are the only cells producing OC, the small variations in circulating 1,25(OH)2D3 levels in thiazide users probably could not explain this reduction in OC levels.(16–19) OC is an osteoblast marker; however, serum OC levels generally are elevated in osteoporotic women and OC levels correlate inversely with bone mineral density.(16,20,21) Furthermore, OC knockout mice have better bone mass than wild-type mice, suggesting that OC retards bone mineralization.(22) We and others recently reported that thiazides can directly inhibit OC secretion by human osteoblasts.(23,24) However, the mechanisms involved in thiazide's action on OC secretion by osteoblasts remains unresolved at present.

Here, we investigated the mechanism by which thiazides inhibit OC production. We used the human osteosarcoma cell model MG-63, which reproduces some of the features of osteoblast-like cells in vitro. With this model cell line, we observed that thiazides act directly on OC messenger RNA (mRNA) levels and that lowering extracellular calcium levels does not modify significantly this effect. Thiazide's effect does not involve any changes in vitamin D3 receptor (VDR) nor in Yin Yang 1 (YY1) levels, yet it increases the production of cFOS in these cells.

MATERIALS AND METHODS

Cell culture

Human osteosarcoma MG-63 cells were obtained through the American Type Culture Collection (Rockville, MD, U.S.A.). They were grown in HAMF12/Dulbecco's modified Eagle medium (DMEM) media containing 10% charcoal-stripped fetal bovine serum (FBS) prepared as previously described, 1% penicillin-streptomycin (PS) mixture (Gibco-BRL, Burlington, Ontario), and 50 μg/ml ascorbic acid.(17) The cells were cultured in an incubator gassed with 5% CO2 at 37°C. Culture medium was changed twice a week. Cells were split once a week at a ratio of 1:9 with 0.025% trypsin-0.01% EDTA in phosphate-buffered saline (PBS), pH 7.4 (trypsin.EDTA). Confluent cells were treated with hydrochlorothiazide (HCTZ; 1–100 uM) for 48 h in a mixture of HAMF12/DMEM (1:1) containing 2% FBS and 50 μg/ml ascorbic acid and 50 nM 1,25(OH)2D3 to stimulate OC production. In some experiments, the incubation was performed in the presence or absence of 0.25 mM or 0.5 mM EDTA or 0.5 mM ethylene glycol-bis(β-amino ethyl ether)-N,N,N',N'-tetraacetic acid (EGTA) to reduce extracellular calcium concentration in the μM range. MG-63 cells were then treated as usual for the determination of alkaline phosphatase activity (ALPase)and OC secretion. In some experiments, cells in T25 flasks also were treated with vitamin D3 (50 nM) and increasing doses of HCTZ for 48 h. After 48 h of culture, cells were prepared for either Northern blot or Western blot analysis as described below.

ALP

Confluent cells in 24-well/plates were washed twice with Hank's buffered salt solution (HBSS) containing CaCl2, 1.26 mM; KCl, 5.37 mM; KH2PO4, 0.44 mM; MgCl2, 0.49 mM; MgSO4, 0.41 mM; NaCl, 136.9 mM; NaHCO3, 25 mM; Na2HPO4, 0.336 mM; D-glucose, 5.55 mM; pH 7.4. Cells were used to evaluate final cell density, ALPase, or protein determination by the Bicinchoninic acid (BCA) method.(25) ALPase was determined as the release of p-nitrophenol hydrolyzed from p-nitrophenyl phosphate (12.5 mM) at 37°C for 30 minutes as previously described after cells were solubilized in ALPase buffer (100 mM glycine, 1 mM MgCl2, 1 mM ZnCl2, 1% Triton X-100, and pH 10.5) with agitation for 60 minutes at 5°C. All samples were then sonicated for 5 s with an ultrasonic sonifier cell disruptor at a setting of 5 just before enzymatic assay.(17,18)

OC secretion

At the end of the 48 h of incubation, the culture media was removed and frozen at 80°C. Intact OC levels were determined by ELISA (Biomedical Technologies, Inc., Stoughton, MA, U.S.A.) with antibodies raised against bovine OC, which cross-react with human OC. OC also was determined on aliquots of the media used for the incubations to correct for bovine OC contamination, which was very low with the procedure used (< 0.1 ng/ml).(17,18)

Northern blot analysis of OC and of VDRs

Confluent MG-63 cells in T25 flasks (1 flask per condition per experiment) were treated with or without increasing doses of HCTZ for 2 days in Ham's F12/DMEM containing 2% heat-inactivated fetal calf serum. At the end of the incubation, cells were washed twice with PBS and scraped into a Trisol lysing solution (Gibco-BRL, Burlington, Ontario). Total RNA was extracted by the method of Chomczynski and Sacchi.(26) RNA samples (10 μg/lane) were size-fractionated on 1.0% agarose gel containing 1.2 M formaldehyde and transferred to nylon membranes. The membranes were UV autocross-linked, prehybridized at 42°C for 20 h, and hybridized at 42°C overnight with digoxigenin (DIG)-labeled DNA probes. The DNA probe used for OC (phBGP/Sac complementary DNA [cDNA]) was generously provided by Dr. J.M. Wozney and T. Celeste (Genetics Institute, Inc., Cambridge, MA, U.S.A.). A 530-base pair (bp) cDNA probe for the VDR was generated via reverse-transcription polymerase chain reaction (RT-PCR)– based amplification from rat osteosarcoma cell ROS 17/2.8 using primers TCCAACACACTGCAGACGTACAT (sense) and ATCAGTCAGCAGCCACTTAGGCA (antisense). The 530-bp PCR product was put in PCR 2.1 plasmid, grown in competent Escherichia coli cells, and cut at EcoRI sites to recover the VDR probe. Both probes were labeled with DIG-uridine triphosphate (DIG-UTP) using a random-primed DNA labeling kit and following the protocol provided by the company (Boeringher-Mannheim, St-Laurent, Québec, Canada). Filters were washed once for 60 minutes in 2× SSC/1 × Denhardt's solution at 42°C, twice in 1 × SSC/0.1% sodium dodecyl sulfate (SDS) for 30 minutes each at 65°C, and 15 times in 0.1 × SSC/1% SDS at room temperature before exposure to Kodak X-Omat films for 1 h at room temperature. Hybridizing signals on the blots were analyzed quantitatively by densitometric scanning of autoradiograms. The filters were then reprobed with cDNA for glyceraldehyde phosphate dehydrogenase (GAPDH) to monitor loading between samples.

Western blot analysis of VDRs, YY1, and cFOS

Confluent MG-63 cells in T25 flasks (1 flask per condition per experiment) were washed twice with PBS and scraped into the same buffer as used for ALP determinations. The cell suspension was then diluted in 2 × electrophoresis buffer made of 8% SDS, 24% glycerol, 100 mM Trizma base, pH 6.8, 4% β-mercaptoethanol, and 0.02% bromophenol blue. SDS-polyacrylamide gel electrophoresis (PAGE) was performed (25 μg protein per lane) in 10% acrylamide gel for 150 minutes at 100 V. After SDS-PAGE, proteins were electrotransferred for 50 minutes at 150 mA (constant current) in a semidry electroblot apparatus on nitrocellulose. Western blots were then performed in the presence of polyclonal rabbit anti-VDR, YY1, or cFOS antibodies (Geneka, Montréal, Québec, Canada) at a final dilution of 1:1000. These antibodies were then detected by horseradish peroxydase–coupled sheep anti-rabbit immunoglobulin Gs (IgGs) (dilution 1:5000), and the protein bands were visualized using the enhanced chemiluminescence (ECL)-plus chemiluminescence kit with Lumigen-PS3 (Amersham Pharmacia Biotech, Baie d'Urfé, Québec, Canada). Specific bands were then detected using Kodak X–Omat films, and scanning densitometry was performed directly on these films.

Table Table 1.. Dose-Response Effectof Thiazideson ALPase and OC Productionby MG-63 Cells
HCTZ (μM)Osteocalcin (ng/mg protein per 48 hours)Alkaline phosphatase (nmol/mg protein per 30 minutes)
  1. Confluent MG-63 cells were incubated for 48 h in the presence of 50 ng/ml 1,25(OH)2D3 and increasing doses (1–100 mM) of HCTZ. At the end of the incubation, the supernatant was recuperated for OC determination by ELISA, and cells were solubilized in ALPase buffer (see Materials and Methods section) before determining ALPase. Results are the mean ± SEM of four individual experiments. Similar letters for individual treatments indicate significant differences (p < 0.05) between these treatments using the Bonferroni/Dunn multiple comparison test.

0120 ± 5a305 ± 16.5a,b
1100 ± 10b,c357 ± 13.4a,b
1070 ± 9a,b379 ± 13.6a,b
10060 ± 6a,b,c354 ± 14.1a
 p < 0.005 by ANOVAp < 0.005 by ANOVA

Statistics

Statistical differences were evaluated by either analysis of variance (ANOVA) or Kruskal-Wallis Nonparametric ANOVA in dose-response experiments. In each case, the statistical test used is indicated and the number of experiments is stated individually in the legend of each figure.

RESULTS

ALPase

As previously shown,(18) HCTZ slightly stimulated 1,25(OH)2D3-induced ALPase in MG-63 cells (Table 1). The increase in ALPase was gradual, and a significant effect of the diuretic was noted at doses as low as 1 μM (p < 0.005 by ANOVA; Table 1).

OC secretion

In contrast to the increase in ALPase induced by HCTZ in MG-63 cells, OC secretion was reduced by the diuretic (Table 1). Our results indicated that OC release in the absence of 1,25(OH)2D3, which is very low in MG-63 cells, was not affected by HCTZ. However, HCTZ dose-dependently (1–100 μM) inhibited 1,25(OH)2D3-induced OC secretion by these cells (Table 1; −40–50%; p < 0.005 by ANOVA). Because HCTZ can increase calcium uptake by MG-63 cells, we next evaluated if reducing the extracellular calcium concentration would affect OC release.(27) Reducing the extracellular calcium concentration in the culture media with either 0.25 mM or 0.5 mM EDTA while incubating cells in the presence of 1,25(OH)2D3 (50 nM) reduced slightly the control values (not significant), yet it did not prevent the inhibitory effect of 100 μM HCTZ on OC secretion (Fig. 1), which was slightly increased under these conditions. In contrast, using 0.5 mM EGTA reduced 1,25(OH)2D3-induced OC release alone, but this did not prevent the HCTZ dose-dependent inhibition of OC release in MG-63 cells (–22.5 ± 6.9%; p < 0.003 by Kruskal-Wallis nonparametric ANOVA test) compared with assays without EGTA (–26.8 ± 2.4%; p < 0.0001 by Kruskal-Wallis nonparametric ANOVA test; Fig. 2).

Figure Fig. 1..

Effect of reducing the extracellular calcium concentration with EDTA on OC production by osteoblasts in response to thiazides. Confluent MG-63 cells were treated with 1,25(OH)2D3 (50 nM) and HCTZ (100 μM), either in the presence or absence of 0, 0.25, or 0.5 mM EDTA in the culture media for the last 48 h of culture. At the end of the incubation, aliquots of the supernatants were used to determine OC by ELISA. The results are the mean ± SEM of four experiments. Statistical differences are given compared with control values.

Figure Fig. 2..

Effect of reducing the extracellular calcium concentration with EGTA on OC production by osteoblasts in response to thiazides. Confluent MG-63 cells were treated with 1,25(OH)2D3 (50 nM) and increasing doses of HCTZ (1 to 100 μM), either in the presence or in the absence of 0.5 mM EGTA in the culture media for the last 48 h of culture. At the end of the incubation, aliquots of the supernatants were used to determine OC by ELISA. The results are the mean ± SEM of four experiments.

Northern blot analysis of OC: effect of extracellular calcium

To assess at which level HCTZ affected OC release by MG-63 cells, we evaluated the possible role of this diuretic on mRNA synthesis. HCTZ dose-dependently inhibited the OC mRNA synthesis by MG-63 cells, and densitometric analysis of bands relative to GAPDH loading revealed that OC mRNA abundance was reduced by 9.0 ± 6.5%, 26.7 ± 3.8%, and 47.2 ± 4.0% in the presence of 1, 10, and 100 μM HCTZ, respectively (p < 0.0001 by paired ANOVA; n = 4; Fig. 3). The maximal inhibition observed was within the same range as for OC protein secretion (Table 1). Interestingly, reducing the extracellular calcium concentration to the micromolar range with 0.5 mM EGTA only partly modified the dose-dependent effect of HCTZ on OC mRNA levels (maximal inhibition, −33.2 ± 6.2%, p < 0.005 by paired ANOVA, and n = 3; Fig. 3), reflecting the results obtained for OC release (see Fig. 2).

Northern and Western blot analysis of VDRs

Because OC secretion is tightly controlled by 1,25(OH)2D3 via specific receptors in osteoblasts, the decrease in OC mRNA synthesis may be caused by a reduction in VDRs in these cells. However, Western blot analysis (Fig. 4, n = 5 for control and n = 3 for HCTZ) and Northern blot analysis (Fig. 5, n = 4 experiments) of VDRs failed to show any effect of HCTZ on either protein or mRNA levels. However, reducing extracellular calcium concentration with 0.5 mM EGTA did alter slightly basal VDR mRNA level in response to increasing doses of HCTZ (Fig. 5) yet it failed to show an effect of HCTZ on this parameter under similar culture conditions.

Figure Fig. 3..

Representative Northern blot analysis of OC expression by MG-63 cells: combined effect of extracellular calcium removal and HCTZ treatment. Confluent MG-63 cells were incubated with 1,25(OH)2D3 (50 ng/ml) for the last 48 h of culture in the presence of 1–100 μM HCTZ, either in the presence or in the absence of 0.5 mM EGTA to reduce the extracellular calcium concentration. After solubilizing cells with Trizol, 10 μg total RNA was electrophoresed and OC mRNA was detected by a specific probe labeled with DIG (see Material and Methods section). Membranes were then reprobed with a cDNA probe for GAPDH to assess similar loadings between samples. The densitometric analysis of the OC mRNA bands against GADPH are given as percent of control (without HCTZ) in the text.

Figure Fig. 4..

Representative Western blot analysis of VDR abundance of MG-63 cells in response to HCTZ treatment. Confluent MG-63 cells were incubated with 1,25(OH)2D3 (50 ng/ml) for the last 48 h of culture in the presence or absence of 100 μM HCTZ. Aliquots of 25-mg proteins were electrophoresed, electroblotted onto nylon membrane, and detected with a polyclonal antibody against the nuclear VDR. Five control preparations and three thiazide preparations are represented on this blot.

Figure Fig. 5..

Representative Northern blot analysis of VDR expression by MG-63 cells: combined effect of extracellular calcium removal and HCTZ treatment. Confluent MG-63 cells were incubated with 1,25(OH)2D3 (50 ng/ml) for the last 48 h of culture in the presence of 0–100 μM HCTZ, either in the presence or in the absence of 0.5 mM EGTA to reduce the extracellular calcium concentration. Total RNA was extracted with Trizol, and 10 μg RNA per lane was electrophoresed. VDR mRNA was detected using a cDNA probe labeled with DIG (see Material and Methods section). Membranes were then reprobed with a cDNA probe for GAPDH to assess similar loadings between samples.

Figure Fig. 6..

Representative Westerm blot analysis of YY1 (top) and cFOS (bottom) in cell extracts of MG-63 cells. Confluent MG-63 cells were incubated with 1,25(OH)2D3 (50 ng/ml) for the last 48 h of culture in the presence of 0–100 μM HCTZ, either in the presence or in the absence of 0.5 mM EGTA to reduce the extracellular calcium concentration. YY1 and cFOS were detected using selective antibodies followed by peroxydase-coupled second antibody and ECL.

Western blot analysis of YY1 and cFOS

The OC gene promoter region responds to a number of transactivating factors either inhibitory or stimulatory.(28–30) Among these, YY1 and cFOS are important regulators of OC expression, which have been identified recently.(29,30) HCTZ failed to alter YY1 levels in MG-63 cells either in the presence or in the absence of 0.5 mM EGTA during the last 48 h of culture (Fig. 6, top). In contrast, cFOS levels dose-dependently increased (p < 0.002 by ANOVA) in response to HCTZ (Fig. 6, bottom), a situation that was prevented by EGTA treatment. Table 2 resumes densitometric evaluations of YY1 and cFOS levels obtained by Western blot analysis from three separate experiments.

DISCUSSION

It is increasingly recognized that thiazide diuretics have positive effects on bone mineral density in patients using this medication instead of other diuretics or antihypertensive agents. Interestingly, alendronate, a bisphosphonate that reduces bone resorption and stone formation during immobilization or bed rest, also reduces OC release by osteoblasts in vitro.(30,31) The present study indicates a direct action of thiazides on OC gene expression by human osteoblasts. Our previous studies as well as those of other investigators indicated that thiazides modulate OC secretion and macrophage colony–stimulating factor release.(23,24,27) Barry et al. recently showed thiazide receptors in rat osteoblast-like UMR-106 cells and proposed a role for thiazides on the regulation of intracellular calcium concentration in these cells.(32) Hence, these studies are consistent with a selective role of thiazides in osteoblasts as we previously proposed.(23) In contrast, one study suggested a possible direct action on osteoclasts(33) via a role of thiazides on carbonic anhydrase in these cells. However, a similar mechanism in osteoblasts is unlikely because we previously showed that acetazolamide, a specific stimulator of carbonic anhydrase, failed to modify OC secretion in MG-63 cells while other sulfonylureas were also inactive (diazoxide) whereas thiazide derivatives (chlorothiazide, cyclothiazide, and HCTZ) were all active.(23)

Table Table 2.. Densitometric Evaluationof YY1 and cFOS Levels Obtainedby Western Blot Analysis
 Densitometric units
 YY1cFOS
 no EGTA+ EGTAno EGTA+ EGTA
  1. Confluent MG-63 cells were incubated for 48 h in the presence of 50 ng/ml 1,25(OH)2D3 and increasing doses (1–100 μM) of HCTZ, in the presence or absence of 0.5 mM EGTA. At the end of the incubation, cells were solubilized in ALPase buffer and protein determined as usual. Aliquots were recuperated and solubilized in 2× electrophoresis buffer before separation. Results are the mean of three experiments ± SD and are given in percent densitometric unit against the control assays without HCTZ and EGTA in each experiment. Statistical analysis was performed by ANOVA and is indicated for each condition.

Control100.0 ± 12.1105.3 ± 21.2100.0 ± 5.9111.0 ± 16.4
1 μM HCTZ101.1 ± 5115.3 ± 21.2105.3 ± 4.4112.3 ± 12.5
10 μM HCTZ99.5 ± 15.3111.1 ± 10.5115.5 ± 5.3111.7 ± 11.6
100 μM HCTZ103.0 ± 17.3106.4 ± 15.7122.3 ± 2.2112.3 ± 12.0
 NSNSp < 0.002NS

We previously showed that HCTZ neither affected DNA synthesis nor cell growth by these cells in vitro.(23,27) Hall and Schaueblin obtained similar results with rat osteosarcoma UMR cells, whereas Song and Wergedal concluded that 1 μM HCTZ increased [3H]thymidine incorporation in human primary osteoblasts after 7 days of exposition to the diuretic.(24,34) This discrepancy between the results of Song and Wergedal and other groups may be related to the length of exposure of cells to HCTZ (7 days as opposed to 2 days) or to the use of primary human osteoblasts in the former study. The slight increase in ALPase measured in MG-63 cells in vitro in response to HCTZ may reflect an increased differentiation of these cells with thiazides (23 and this study).

In this study, we found that thiazides act directly on OC gene expression. Our data would preclude that the inhibitory effect of HCTZ on OC secretion is related exclusively to an effect on Ca influx in MG-63 cells. Indeed, reducing the extracellular Ca concentration with either EDTA or EGTA did not prevent the effect of thiazides on OC production both at the protein and at the mRNA level. However, this does not rule out a role for HCTZ on the intracellular Ca2+ concentration via a role on Ca store release of these cells as proposed by Barry et al.(32) Moreover, whether 0.25 mM and 0.5 mM EDTA or 0.5 mM EGTA was enough to remove fully extracellular Ca, and thereby reduce calcium influx to such low levels that this inhibited secretion/gene expression, is possible. However, we could not increase the concentration of EDTA and/or EGTA further during the 2 days of conditioning because this lead to cell loss (not shown). Further studies will be needed to determine the exact role of intracellular calcium concentration on OC secretion and the role played by thiazides on this mechanism because we previously showed that variations of extracellular calcium in the culture media of MG-63 cells between 0.25 and 1 mM did not modify OC secretion by these cells whereas only at higher calcium concentrations (>1.5 mM) did it increase OC release.(18) Further, our results on the modulation of OC mRNA and OC protein levels in response to HCTZ treatment in the presence/absence of EGTA suggest that thiazides cannot exert exclusively their effect via a modulation of the expression of the OC gene but that posttranscription mechanisms also can be involved, a situation that corroborates the observations of Shalhoub et al.(35) However, this cannot be related to an effect on VDR mRNA levels under these conditions. Indeed, although EGTA treatments slightly reduced VDR mRNA levels and OC release, the combination with HCTZ did not induce any further reduction of VDR mRNA levels whereas it inhibited OC release, albeit slightly less than without EGTA treatments. This would then indicate that extracellular calcium levels modulate somewhat VDR levels but not the full response to thiazides.

Vitamin D3 and cyclic adenosine monophosphate (cAMP) responsive elements, AP1 and AP2 sites, glucocorticoids and estrogens responsive elements, etc. are present in the promoter region of the OC gene.(16,28,36) All of these regions may be affected by HCTZ possibly via its role on intracellular Ca concentration and/or via other as yet undefined effectors. Our data indicated no effect of thiazides on YY1 levels, either in the presence or in the absence of EGTA. In contrast, HCTZ dose-dependently enhanced cFOS production in MG-63 cells in the absence of EGTA, whereas in the presence of EGTA cFOS levels remained elevated. This increase in cFOS production in the absence of EGTA was paralled by a decrease in OC release by these cells. A positive effect of an intact AP-1 site (response element for Fos-Jun interaction) distal to the vitamin D responsive element in the human OC gene has been proposed by Ozono et al.; however cFOS and the VDR can form a complex that blocks the interaction of the VDR with the vitamin D responsive element on the promoter region of OC.(30,37,38) Hence, the increase in cFOS levels without any changes in VDR could explain the observed inhibition of OC secretion in response to HCTZ. The increase in cFOS levels in the presence of EGTA could then explain the partial reduction in OC secretion noted here because the Jun-Fos complex directly inhibits OC gene expression.(30) However, this could not explain the HCTZ-dependent inhibition of OC mRNA and protein production still observed in the presence of EGTA because cFOS levels remained elevated under these conditions, implying that other mechanisms besides cFOS, VDR, or YY1 are targeted by thiazides. Indeed, thiazides may be interfering with other pathways such as YY1-mediated VDR/VDR response element protein–DNA interaction and the VDR/transcription factor TFIIB protein–protein interaction, which is essential for VDR-driven OC gene expression.(39) However, direct regulation of cAMP responsive elements via the modulation of the protein kinase A (PKA) pathway is unlikely, because Barry et al. recently showed that thiazides do not influence basal and PTH-stimulated cAMP synthesis in the rat osteoblast-like cell UMR 106, and we also failed to show any effect of HCTZ on cAMP synthesis in MG-63 cells (not shown).(32) In conclusion, HCTZ directly affects the regulation of OC production by MG-63 cells both at the mRNA and at the protein level, by a process that does not involve calcium uptake, but may involve changes in cFOS levels.

Acknowledgements

D. Lajeunesse is a senior scholar from the “Fonds de la Recherche en Santé du Québec.” This work was supported by a grant from the Kidney Foundation of Canad a (D.L.) and grant NIH DK 43423 (S.E.G.).

Ancillary