Calcium Input Potentiates the Transforming Growth Factor (TGF)-β1-dependent Signaling to Promote the Export of Inorganic Pyrophosphate by Articular Chondrocyte*

Transforming growth factor (TGF)-β1 stimulates extracellular PPi (ePPi) generation and promotes chondrocalcinosis, which also occurs secondary to hyperparathyroidism-induced hypercalcemia. We previously demonstrated that ANK was up-regulated by TGF-β1 activation of ERK1/2 and Ca2+-dependent protein kinase C (PKCα). Thus, we investigated mechanisms by which calcium could affect ePPi metabolism, especially its main regulating proteins ANK and PC-1 (plasma cell membrane glycoprotein-1). We stimulated articular chondrocytes with TGF-β1 under extracellular (eCa2+) or cytosolic Ca2+ (cCa2+) modulations. We studied ANK, PC-1 expression (quantitative RT-PCR, Western blotting), ePPi levels (radiometric assay), and cCa2+ input (fluorescent probe). Voltage-operated Ca2+-channels (VOC) and signaling pathways involved were investigated with selective inhibitors. Finally, Ank promoter activity was evaluated (gene reporter). TGF-β1 elevated cCa2+ and ePPi levels (by up-regulating Ank and PC-1 mRNA/proteins) in an eCa2+ dose-dependent manner. TGF-β1 effects were suppressed by cCa2+ chelation or L- and T-VOC blockade while being mostly reproduced by ionomycin. In the same experimental conditions, the activation of Ras, the phosphorylation of ERK1/2 and PKCα, and the stimulation of Ank promoter activity were affected similarly. Activation of SP1 (specific protein 1) and ELK-1 (Ets-like protein-1) transcription factors supported the regulatory role of Ca2+. SP1 or ELK-1 overexpression or blockade experiments demonstrated a major contribution of ELK-1, which acted synergistically with SP1 to activate Ank promoter in response to TGF-β1. TGF-β1 promotes input of eCa2+ through opening of L- and T-VOCs, to potentiate ERK1/2 and PKCα signaling cascades, resulting in an enhanced activation of Ank promoter and ePPi production in chondrocyte.

The balance between extracellular inorganic pyrophosphate (ePP i ) 3 and extracellular inorganic phosphate (eP i ) is critical for homeostasis of articular cartilage. Indeed, an increase in ePP i supports calcium pyrophosphate dihydrate crystal (CPPD) depositions responsible for articular chondrocalcinosis (ACC) (1). Production of ePP i by articular chondrocytes is dependent on PC-1 (plasma cell membrane glycoprotein-1; also known as ENPP1), an ectoenzyme that produces ePP i from nucleotide triphosphates (2), and on the transporter ANK, which exports PP i across the cell membrane (3). Although ePP i level can be controlled by tissue-nonspecific alkaline phosphatase, which hydrolyzes ePP i into eP i , no expression of this enzyme was detected in healthy articular cartilage compared with osteoarthritic cartilage (4). Therefore, CPPD formation only depends on the expression level of PC-1 and ANK in mature articular cartilage.
Transforming growth factor (TGF)-␤1 is a well known inducer of ePP i production by articular chondrocytes (1). We demonstrated previously that ANK was the major contributor of this TGF-␤1-induced ePP i production; meanwhile, PC-1 played a minor role (5). Furthermore, these effects of TGF-␤1 relied on an ERK1/2-calcium-dependent protein kinase C (PKC) signaling mechanism, which was independent from Smad activation. TGF-␤1 levels were also described to be gradually increased in a rat model of hyperparathyroidism consisting in a daily injection of parathyroid hormone (PTH) for 3 months (6). PTH levels are classically elevated in patients with hyperparathyroidism, and an increase in circulating PTH was also reported in patients with idiopathic CPPD depositions (7). Altogether, these observations support a potential role for TGF-␤1 in the formation of CPPD crystal formation in the context of primary hyperparathyroidism.
ACC linked to hyperparathyroidism reaches a prevalence of 21% (8). Besides development of ACC, patients with hyperparathyroidism have a chronic elevation of free calcium levels in their extracellular fluids. Although no direct link was reported between hypercalcemia and ACC (9), a very recent study demonstrated that patients with familial hypocalciuric hypercalcemia developed ACC (10). However, the modus operandi by which calcium may influence CPPD formation remains unclear and is therefore worthy of study.
The higher calcium levels found in patients with hyperparathyroidism could influence chondrocyte metabolism by stimulating calcium input through voltage-operated channels (VOCs). Indeed, several VOCs were described as being expressed in chondrocytes, such as N-VOCs (11) and L-and T-VOCs (12). Another possible mechanism could be related to the activation of the calcium-sensing receptor (CaSR), a G protein-coupled receptor (13). Furthermore, calcium was described as stimulating the activity of the transcription factor SP1 (specific protein 1), which increased the basal transcription of ␤1,3-glucuronosyltransferase I, a key enzyme responsible for the completion of cartilage proteoglycans (14). Interestingly, SP1 activity was also shown to be implicated both in the TGF-␤1-induced expression of tissue inhibitor of metalloproteinase-3 (15) and ␤1,3-glucuronosyltransferase I (16) in chondrocytes. These findings suggest a possible regulatory interplay between calcium and TGF-␤1 through the SP1 transcription factor, which could result possibly in CPPD crystal deposition in articular cartilage.
The present work investigated whether the variation in either extracellular or cytosolic calcium levels could influence the stimulating effects of TGF-␤1 on ePP i production by articular chondrocytes. Particular care was taken in deciphering the contribution of VOCs and the signaling events regulating the expression of the main ePP i -regulating genes PC-1 and especially Ank with a study of its promoter region.

EXPERIMENTAL PROCEDURES
Chondrocyte Isolation and Culture-Articular cartilage was obtained from 6-week-old healthy male Wistar rats (130 -150 g) killed under dissociative anesthesia (ketamine (Mérial, Lyon, France) and acepromazine (Sanofi-Aventis, Libourne, France)) in accordance with our institutional local ethics committee and the national animal care guidelines. Cartilage pieces were dissected aseptically from femoral head caps, and chondrocytes were obtained by sequential digestion with Pronase and collagenase B (Roche Applied Science), as described previously (17). Cells were washed twice in PBS and cultured to confluence in 75-cm 2 flasks at 37°C in a humidified atmosphere containing 5% CO 2 . Cells were maintained in calcium-free DMEM (Invitrogen), supplemented with 1.25 mM of CaCl 2 (Sigma, France), L-glutamine (2 mM), gentamicin (50 g/ml), amphotericin B (0.5 g/ml), and 10% heat-inactivated FCS (Invitrogen). Unless specified, we used first passage chondrocytes plated at 4 ϫ 10 5 cells/well in 6-well plates throughout the study. Except for the dose-ranging studies on eCa 2ϩ level, experiments were performed in calcium-free DMEM, supplemented with 1.25 mM Ca 2ϩ to mimic the physiological free circulating Ca 2ϩ levels.
Chemicals-All chemical reagents were from Sigma, unless otherwise specified. All of the compounds used in this study were dissolved in DMSO (final concentration 0.1%), used as vehicle in control experiments. At the concentrations used, no cytotoxicity of either compound was detected in a 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide assay after 72 h of exposure (data not shown). Final concentrations were chosen after preliminary experiments demonstrating the selectivity of signal transduction inhibitors (immunoblotting of the phosphorylation level of the concerned signaling pathway; data not shown). TGF-␤1 (PeproTech, Neuilly-Sur-Seine, France) was used at a final concentration of 10 ng/ml throughout the study.
RNA Extraction and Reverse Transcription-Total RNA was isolated using the RNeasyplus minikit (Qiagen), which allows total removal of genomic DNA with an on-column DNA elimination step. Five hundred ng of total RNA were reverse-transcribed for 90 min at 37°C in a 20-l reaction mixture containing 2.5 mM dNTP, 5 M random hexamer primers, 1.5 mM MgCl 2 , and 200 units of Moloney murine leukemia virus reverse transcriptase (Invitrogen). cDNA production was performed in a Mastercycler gradient thermocycler (Eppendorf, France).
Real-time Quantitative PCR-To check our chondrocyte phenotype and measure mRNA levels of target genes, real-time quantitative PCR was performed using the Lightcycler (Roche Applied Science) technology. PCR was done with SYBR Green master mix system (Qiagen). The gene-specific primer pairs are provided in Table 1. A melting curve was performed to determine the melting temperature of the specific PCR products, and after amplification, the product size was checked on a 1% agarose gel stained with ethidium bromide (0.5 g/ml). Each run included positive and negative reaction controls. The mRNA level of the gene of interest and of S29, chosen as housekeeping gene, was determined in parallel for each sample. The S29 gene, which codes for a ribosomal protein, was shown previously to be invariable in TGF-␤1-challenged chondrocytes (5). Quantification was determined using a standard curve made from a purified PCR product for each gene tested, with concentrations ranging from 10 Ϫ3 to 10 Ϫ9 ng/l. Results were expressed as the ratio of the mRNA level of each gene of interest over the S29 gene.
Western Blot Analysis-Chondrocytes were harvested and lysed in 1ϫ Laemmli buffer. Samples were run on SDS-PAGE (10%) and transferred onto a polyvinylidene fluoride membrane (Immobilon TM , Waters, France). After 2 h in blocking buffer (TBS, 0.1% Tween (TBST) supplemented with 5% nonfat dry milk), membranes were washed three times with TBST and incubated overnight at 4°C with primary antibodies. The antibodies against ANK and PC-1 (Eurogentec, Angers, France) were used at a 1:500 dilution as we described previously (5,18). Antibodies directed against native and phosphorylated ERK1/2 and PKC␣ (Cell Signaling Technology) as well as antibodies against native (Santa Cruz Biotechnology) and phosphorylated ELK-1 (Ets-like protein-1) (Millipore) were used at 1:500. Antibodies against total (Millipore) and phosphorylated SP1 (Abcam) were used at 1:1000, and the anti-␤-actin (Sigma) was used at 1:4000. After three washings with TBST, each blot was incubated for 1 h at room temperature with anti-rabbit IgG conjugated with HRP (Cell Signaling Technology) at 1:2000 dilution in blocking buffer. After four washings in TBST, protein bands were detected by chemiluminescence with the Phototope Detection TM system according to manufacturer's recommendations (Cell Signaling Technology). Band intensities were quantified by densitometry with the Gnome TM computerized image processing system (Syngene, Cambridge, UK). ePP i Assays-ePP i levels were measured in culture supernatants, using the differential adsorption between UDP-[6-3 H]glucose (PerkinElmer Life Sciences) and its reaction product 6-phospho-[6-3 H]gluconate on activated charcoal, as described previously (19). The standards, ranging from 10 to 400 pmol of PP i , were included in each assay. After adsorption of the reaction mixture on charcoal and centrifugation at 16,000 ϫ g for 10 min, 100 l of the supernatant were counted for radioactivity in 5 ml of Bio-Safe II (Research Products International). Results were expressed as pmol of ePP i /g of total cell proteins (quantified by bicinchoninic acid assay).
Cytosolic Calcium Analyses-Calcium levels were measured in a fluorimeter (FLX-Xenius, SAFAS, Monaco) using the Fluo-4 NW TM calcium assay kit (Invitrogen) according to manufacturer's protocol (20). This assay is based on a dye solution containing the Fluo-4AM fluorescent probe and probenecid. This drug limits the extrusion of Fluo-4AM and reduces the base-line signal. Chondrocytes were plated in 96-well plates at 3 ϫ 10 4 cells/well and then cultured for 24 h until 90% confluence (ϳ4 ϫ 10 4 cells). Briefly, cells were washed with PBS and then incubated with 100 l/well of dye solution for 1 h at 37°C, in the presence or not of the different selective modulators of cytosolic Ca 2ϩ level and signaling inhibitors. Afterward, cells were placed back in culture medium containing TGF-␤1 for 1 min, which corresponded to the maximal cytosolic Ca 2ϩ level measured (time course over 30 min; data not shown). Fluorescence was read after PBS washing. Results were expressed as mean relative fluorescence units (RFU)/10 4 per 4 ϫ 10 4 cells.
Ras Activation Assay-The amount of Ras-GTP (active form) was evaluated using a Ras activation assay kit (Millipore) according to the manufacturer's protocol. This ELISA is based on a selective trapping of Ras-GTP using the Ras binding domain (RBD) of Raf-1 (a kinase downstream Ras) that fails to bind Ras-GDP (inactive form). The coupling of RBD to GST (RBD-GST) allows immobilization of the complex on microplate wells precoated with glutathione. The trapped Ras-GTP is further detected with a monoclonal antibody against Ras. Briefly, chondrocytes were lysed in the Mg 2ϩ lysis/wash buffer from the kit, supplemented with one tablet of complete mini protease inhibitor mixture (Roche Applied Science), before protein quantification by a Bradford assay. Cellular lysates containing 50 g (in 100 l) of proteins were incubated for 1 h in the RBD-GST-coated wells. After washing three times with TBST, wells were incubated for 1 h with the primary anti-Ras mouse antibody. After TBST washings, wells were incubated for 1 h with an HRP-conjugated goat anti-mouse secondary antibody. After four additional TBST washings, wells were finally rinsed three times with TBS, before incubation for 10 min with a chemiluminescent substrate. All incubation steps were performed at room temperature under gentle rocking. Luminescence was read with a CentroLB960 microplate luminometer (Berthold Technologies). Results were expressed as relative luminescence units (RLU)/10 6 .
siRNA Assays-RNA silencing was processed with either a control siRNA-A or a mix of three siRNA sequences against Sp1 from Santa Cruz Biotechnology. The final concentration used was 10 nM, and transfections were performed using INTERFERin TM (Polyplus Transfection). Briefly, siRNA were diluted in serum-free medium, and INTERFERin TM was then added to this mixture (12 l of INTERFERin TM /200-l mixture) for a short incubation at room temperature. During that time, cells were washed with PBS and then placed in serum-free medium. The siRNA-INTERFERin TM mix was then added to the culture for 24 h (200 l of mix/2 ml of medium).
Ank Promoter Cloning in pGL3-Human Ank promoter sequence (GenBank TM accession number AC016575) was cloned from a mix of genomic DNA from different organs (brain, heart, liver, and kidney) (Clinisciences) in the pGL3 luciferase reporter vector (Promega), using the KpnI and XhoI sites. Primers used for the amplification are described in Table  1. We cloned a Ϫ720 bp fragment containing two putative Sp1responsive elements (Ϫ125/15 bp and Ϫ532/22 bp) and a Ϫ2715 bp fragment containing the Sp1-responsive elements and one putative Elk-1-responsive element (Ϫ1860/44 bp). These responsive elements were identified using the TFSearch 1.3 algorithm (21).
Plasmid Electroporation-Cells were electroporated as described previously (5), using the Chondrocyte Nucleofector kit (Lonza) according to the manufacturer's protocol, with the Nucleofector (Lonza) program U-28. In one set of experiments, we transfected either the Ϫ720 bp or the Ϫ2715 bp Ank promoter fragment (1 g/3 ϫ 10 5 cells) in chondrocytes exposed or not to TGF-␤1. In another set of experiments, cells were co-transfected with either one of these constructs, along with 1 g/3 ϫ 10 5 cells of either pCMV-Sp1 (overexpressing wild-type human SP1 protein) or pCMV-Elk-1 (overexpressing wild-type human ELK-1 protein) or both vectors. These constructs were a generous gift from Dr. M. Ouzzine (UMR7561 CNRS-Nancy University (14)). In each experiment, chondrocytes were transfected with pTK-RL expressing Renilla luciferase (500 ng/3.10 5 cells) to normalize the results. Plasmid pmaxGFP TM (Lonza), encoding a GFP, was used to assess transfection efficiency.
Reporter Gene Assay-Luciferase activities were measured using the Dual-Luciferase TM reporter assay kit (Promega). Briefly, chondrocytes electroporated with reporter gene constructs were stimulated or not for 48 h with TGF-␤1. After two washings in PBS, cells were lysed in 1ϫ passive lysis buffer for 15 min at room temperature. Cell lysate was placed in LAR II (firefly luciferase substrate), and luminescence was read with a Sirius luminometer (Berthold Technologies). Afterward, Stop & Glo reagent (Renilla luciferase substrate) was added to the sample, and luminescence was acquired in the same way. Results were expressed as the mean ratio of firefly/Renilla luciferase activity, in -fold induction over control.
Statistical Analysis-Results are expressed as the mean Ϯ S.D. of at least three independent assays. Comparisons were made by analysis of variance, followed by Fisher's t post hoc test, using the Statview TM 5.0 software (SAS Institute Inc.). A value of p Ͻ 0.05 was considered significant.

Effects of eCa 2ϩ Levels on TGF-␤1-stimulated Metabolism of PP i and Calcium Mobilization in
Chondrocytes-First, we confirmed that the articular phenotype of chondrocyte was maintained throughout the study. Indeed, a strong expression of type II collagen was observed, whereas collagens type I and X, Runx2, and tissue-nonspecific alkaline phosphatase remained undetectable (data not shown). This confirmed our previous report (5) and underlined that no dedifferentiation or switch toward hypertrophy occurred in our culture conditions.
To determine the influence of eCa 2ϩ level on the TGF-␤1stimulated metabolism of ePP i , chondrocytes were incubated in calcium-free DMEM containing 1% FCS, which was supplemented or not with 1.25, 2.5, or 5 mM CaCl 2 for 24 h. Then cells were exposed to TGF-␤1 for either 24 h (mRNA expression), 48 h (protein level), or 72 h (ePP i production). Of note, all results were analyzed in comparison with the "physiological" concentration of 1.25 mM eCa 2ϩ . Basal ePP i level was reduced from 65 to 42 pmol/g protein (Ϫ35%) in the calcium-free medium while being increased to 89 pmol/g protein (1.4-fold) at 5 mM eCa 2ϩ (Fig. 1A). TGF-␤1 increased the ePP i production from 65 to 274 pmol/g protein (4.2-fold) at 1.25 mM eCa 2ϩ . This effect was limited to 94 pmol/g protein (2.3-fold) in the absence of eCa 2ϩ , while increasing gradually with eCa 2ϩ to peak at 389 pmol/g protein (4.4-fold) at 5 mM eCa 2ϩ . Compared with the 1.25 mM eCa 2ϩ condition, basal Ank mRNA level (evaluated by real-time quantitative PCR) was reduced from 3.8 to 0.8 (Ϫ79%) in the absence of eCa 2ϩ , whereas PC-1 expression was decreased from 2.5 to 1.2 (Ϫ52%) (Fig. 1B). Conversely, supplementation of culture medium with 5 mM eCa 2ϩ up-regulated the expression of Ank from 3.8 to 5.4 (1.4-fold) and of PC-1 from 2.5 to 5 (2-fold) (Fig. 1B). TGF-␤1 increased Ank and PC-1 mRNA expression to 19.2 (5-fold) and 9.6 (3.8fold), respectively, at 1.25 mM eCa 2ϩ . In the absence of eCa 2ϩ , TGF-␤1 elevated the expression of Ank only from 0.8 to 2.1 (2.6-fold) and of PC-1 from 1.2 to 2.5 (2.1-fold). However, the effect of TGF-␤1 reached 32 (8.4-fold) for Ank and 22.5 (9-fold) for PC-1 at 5 mM eCa 2ϩ . Interestingly, the articular phenotype of chondrocytes was affected neither by changes in eCa 2ϩ levels nor by TGF-␤1 challenge because type I and type X collagen remained undetectable, whereas a high mRNA level of type II collagen was maintained (data not shown). Western blot analysis ( Fig. 1C) confirmed the influence of eCa 2ϩ on the inducing effect of TGF-␤1 on Ank and PC-1 mRNA levels because the progressive increase in eCa 2ϩ boosted the TGF-␤1 up-regulation of ANK and PC-1 protein levels. Finally, we checked whether TGF-␤1 and/or eCa 2ϩ could initiate calcium mobilization in chondrocytes. For that goal, cytosolic calcium (cCa 2ϩ ) level was evaluated in cells exposed for 24 h to the different eCa 2ϩ concentrations before being challenged or not with TGF-␤1. Per se, eCa 2ϩ variation was unable to modify cCa 2ϩ level after 1 min of exposure (Fig. 1D). However, after 5 min of exposure, the cCa 2ϩ level was increased from 0.5 to 0.8 RFU (1.6-fold) and from 0.5 to 1 RFU (2-fold) in cells incubated with either 2.5 or 5 mM eCa 2ϩ , respectively (Fig. 1D). After 1 min, TGF-␤1 was ineffective in the absence of eCa 2ϩ , whereas it increased the cCa 2ϩ level from 0.5 to 2.7 RFU (5.4-fold) at 1.25 mM eCa 2ϩ (Fig. 1D). At higher eCa 2ϩ concentrations, this increase in cCa 2ϩ level was amplified to 3.6 RFU (7.2-fold, at 2.5 mM) and 5 RFU (10-fold, at 5 mM). Interestingly, after 5 min of exposure to the same culture conditions, TGF-␤1 remained ineffective in the absence of eCa 2ϩ while producing lower effects than at 1 min for the other eCa 2ϩ concentrations (Fig.  1D). This was consistent with the time course analysis showing that TGF-␤1 influence was maximal after 1 min (data not shown). Taken together, these data demonstrate that the stimulating effect of TGF-␤1 on Ank and PC-1 was enhanced by eCa 2ϩ in a dose-related fashion, which suggests a key regulatory role for calcium on the modulation of ePP i metabolism by TGF-␤1 in chondrocyte.

Calcium Entrance Stimulates TGF-␤1-induced PP i Export
Effect of Selective Ca 2ϩ -VOC Blockers and CaSR Agonist on TGF-␤1-activated Metabolism of PP i and Calcium Mobilization in Chondrocytes-We next investigated the mechanisms involved in the cCa 2ϩ mobilization and stimulation of ePP i metabolism by TGF-␤1. With that goal, we examined the consequence of a 1-h preincubation of chondrocytes with either a selective CaSR agonist, GdCl 3 (100 M), or the following selective VOCs blockers (Merck, Darmstadt, Germany): -agatoxin (1 M, for P/Q-type), -conotoxin (1 M, for N-type), lercanidipin (10 M, for L-type), and NiCl 2 (50 M, for a selective blockade of T-VOCs (22)). Their effects were then assessed after either 24 h (mRNA expression), 48 h (protein level), or 72 h (ePP i production) of stimulation with TGF-␤1 in the presence of 1.25 mM eCa 2ϩ .
The basal ePP i production was unaffected by any of the compounds listed above ( Fig. 2A). Moreover, neither activation of CaSR nor blockade of N-VOCs affected the TGF-␤1-stimulated ePP i production, whereas -agatoxin slightly decreased it from 274 to 235 pmol/g protein (Ϫ14%) ( Fig. 2A). In contrast, lercanidipin and NiCl 2 markedly diminished the TGF-␤1 effect to 162 pmol/g protein (Ϫ41%) and 93 pmol/g protein (Ϫ66%) respectively ( Fig. 2A). In accordance with these findings, none of these compounds modified the basal mRNA expression of Ank and PC-1 (Fig. 2B). Once again, neither GdCl 3 nor -conotoxin modulated the stimulating effect of TGF-␤1 on Ank and PC-1 mRNA levels, whereas P/Q-VOC blockade by -agatoxin diminished it from 19.2 to 15.3 (Ϫ20%) for Ank and from 9.6 to 7.9 (Ϫ18%) for PC-1 (Fig. 2B). More importantly, blockade of L-or T-VOCs by lercanidipin or NiCl 2 reduced the impact of TGF-␤1 on Ank mRNA level to 9.9 (Ϫ49%) and 6.6 (Ϫ66%), respectively (Fig. 2B). Similarly, PC-1 expression was reduced to 6.1 (Ϫ36%) and 5 (Ϫ48%), respectively. Western blot analysis (Fig. 2C) corroborated the assets obtained at the mRNA level because only lercanidipin and NiCl 2 significantly reduced the increase in ANK and PC-1 protein level induced by TGF-␤1 (Fig. 2C), whereas GdCl 3 and -conotoxin remained ineffective (data not shown). Finally, to evaluate the possible involvement of these VOCs in the TGF-␤1-induced calcium mobilization, we measured the cCa 2ϩ level in the same experimental conditions as above. None of the compounds affected the basal cCa 2ϩ level (Fig. 2D). Interestingly, only the blockade of L-and T-VOCs lowered the TGF-␤1stimulated calcium mobilization, from 2.7 RFU to 1.7 RFU (Ϫ37%) and 1.1 RFU (Ϫ60%), respectively (Fig. 2D). Although -agatoxin displayed a moderate inhibitory effect on TGF-␤1stimulated metabolism of ePP i , it remained ineffective on the cCa 2ϩ level. These results demonstrate a major role for calcium entry through L-and T-type VOCs in the modulation of the TGF-␤1-stimulating effect.

Effect of Calcium Chelation or Release from Internal Stores on TGF-␤1-induced Metabolism of PP i and Ca 2ϩ Mobilization in
Chondrocytes-To evaluate the respective contribution of cCa 2ϩ level and calcium located in internal stores on the stimulating effect of TGF-␤1 on ePP i production, we examined the consequence of a 1-h preincubation of cells with the three following compounds (Merck): a calcium chelator, BAPTA-AM (1 M); a calcium ionophore, ionomycin (1 M); and an inhibitor of phospholipase C (PLC)-dependent calcium release from  None of the compounds listed above modified the basal ePP i production (Fig. 3A). Moreover, PLC inhibition did not affect the TGF-␤1-stimulated production of ePP i (Fig. 3A). Interestingly, although BAPTA significantly decreased the TGF-␤1stimulated ePP i level from 274 to 145 pmol/g protein (Ϫ49%), ionomycin raised it to 442 pmol/g protein (1.6-fold). None of the compounds significantly modified the basal mRNA expression of Ank and PC-1 (Fig. 3B). Chelation of cCa 2ϩ by BAPTA reduced the effect of TGF-␤1 from 19.2 to 7.9 (Ϫ59%) for Ank and from 9.6 to 4.9 (Ϫ49%) for PC-1 mRNA levels. In contrast, cCa 2ϩ elevation induced by ionomycin boosted the effect of TGF-␤1 from 19.2 to 36.4 (1.9-fold) for Ank and from 9.6 to 17.8 (1.9-fold) for PC-1. Once again, U73122 remained ineffective (Fig. 3B). Western blot analysis (Fig. 3C) confirmed the opposite influence of BAPTA and ionomycin on the inducing effect of TGF-␤1 on ANK and PC-1 protein levels (Fig. 3C). Interestingly, although ionomycin alone failed to modify Ank mRNA level, it strongly induced ANK expression. This apparent discrepancy may come from the ability of ionomycin to stabilize mRNA, as demonstrated previously for the interleukin-3 transcript (23), therefore allowing an increase in the translation step. Finally, neither the basal nor the TGF-␤1-induced increase in cCa 2ϩ level was affected by PLC inhibition (Fig. 3D). In contrast, although BAPTA did not modify the basal cCa 2ϩ level, it lowered the TGF-␤1-induced influx from 2.7 RFU to 1 RFU (Ϫ63%). In an opposite fashion, ionomycin increased the basal cCa 2ϩ level to 3.3 RFU (6.6-fold) and potentiated the stimulating effect of TGF-␤1 to 5.5 RFU (1.7-fold) (Fig. 3D). Taken together, these results demonstrate that an increase in cCa 2ϩ , independent of PLC-dependent calcium release from the endoplasmic reticulum, is critical for the stimulating effect of TGF-␤1 on ePP i metabolism.
Study of the Cross-talk between the eCa 2ϩ Level, cCa 2ϩ Level, and the TGF-␤1-activated Signaling Pathways and Influence on ANK and PC-1 Expression-To evaluate the influence of calcium on the signaling pathways activated by TGF-␤1 and implicated in the regulation of ePP i metabolism, we assessed, by immunoblotting, the activation of ERK1/2 and Ca 2ϩ -dependent PKC. For that purpose, chondrocytes were challenged with TGF-␤1 for 5 min (PKC) or 15 min (ERK), as described elsewhere (5). During these experiments, we modulated either the eCa 2ϩ level or modified the eCa 2ϩ entrance (with the only efficient VOC blockers, lercanidipin and NiCl 2 ) as well as the cCa 2ϩ level (using BAPTA and ionomycin). We completed this study by evaluating the consequence of ERK1/2 or PKC␣ inhibition on the increase in ANK and PC-1 protein level induced by high eCa 2ϩ levels. With that goal, cells were preincubated for 1 h with the following selective inhibitors (Merck): PD98059 (10 M; a MEK 1 inhibitor that prevents ERK1/2 activation) or Gö6976 (5 M; a selective calcium-dependent PKC inhibitor). Then ANK and PC-1 protein levels were determined after an additional 48-h incubation in culture medium containing either 1.25 or 5 mM Ca 2ϩ . As shown in Fig. 4A, eCa 2ϩ alone was sufficient to stimulate the phosphorylation cascades, from 1.25 mM for ERK1/2 and 2.5 mM for PKC␣. Interestingly, although TGF-␤1 was able to induce ERK1/2 phosphorylation in the absence of calcium, this activation increased gradually with eCa 2ϩ level. Likewise, the phosphorylation of PKC␣ by TGF-␤1 was also related to eCa 2ϩ level (Fig. 4A). When chondrocytes cultured at 1.25 mM Ca 2ϩ were preincubated with L-VOC blocker lercanidipin, the TGF-␤1-induced ERK1/2 phosphorylation was weakened, whereas the activation of PKC␣ was totally suppressed (Fig. 4B). The same results were observed for T-VOC blockade by NiCl 2 but with a more marked reduction of ERK1/2 phosphorylation. As shown in Fig. 4C, BAPTA totally prevented ERK1/2 and PKC␣ phosphorylation by TGF-␤1. In contrast, ionomycin was sufficient per se to enhance the basal phosphorylation of PKC␣ and ERK1/2. However, in response to TGF-␤1, only the phosphorylation of PKC␣ was enhanced by the cCa 2ϩ burst, suggesting that activation of the ERK1/2 pathway was still maximal under ionomycin challenge. As demonstrated in Fig. 4D, the increase in ANK and PC-1 protein levels by 5 mM Ca 2ϩ was almost totally suppressed by both PD98059 and Gö6976. Finally, to assess how calcium input could regulate TGF-␤1 signaling, we checked the activation status of Ras, which was previously shown to be activated secondary to an increase in cCa 2ϩ level (24). With that goal, chondrocytes at 90% confluence were incubated in calcium-free DMEM containing 1% FCS, which was supplemented or not with 1.25 mM eCa 2ϩ for 24 h. Then cells cultured without eCa 2ϩ were preincubated for 1 h with vehicle, whereas cells exposed to 1.25 mM eCa 2ϩ were preincubated with either vehicle or calcium modulators (i.e. lercanidipin, NiCl 2 , or BAPTA). Afterward, chondrocytes were exposed to TGF-␤1 for 5 min, and the amount of Ras-GTP was estimated in total cell lysates. As shown in Fig. 4E, no increase in Ras-GTP level was detected in cells stimulated with TGF-␤1 in the absence of eCa 2ϩ . In contrast, in the presence of 1.25 mM eCa 2ϩ , TGF-␤1 induced a 2.8-fold increase in Ras-GTP level (from 1.1 to 3.1 RLU). Interestingly, the blockade of VOCs, of L-type by lercanidipin and T-type by NiCl 2 , reduced this effect, from 3.1 to 2.2 RLU (Ϫ43%) and 1.4 RLU (Ϫ85%), respectively (Fig. 4E). In these experimental conditions, BAPTA reduced the TGF-␤1-induced increase in Ras-GTP level to those observed in the absence of eCa 2ϩ .
These results demonstrate that the stimulatory effect of TGF-␤1 on Ras, ERK1/2, and PKC␣ activation is critically regulated by eCa 2ϩ entry. ERK1/2 and PKC␣ directly contribute to the increase in ANK and PC-1 protein level provoked by high calcium levels.
Contribution of SP1 to the Regulatory Role of Calcium in the Chondrocyte Response to TGF-␤1-Because SP1 was reported to relay the effect of TGF-␤1, we assessed its possible role in the modulation of TGF-␤1-stimulated metabolism of ePP i in regard to calcium. With that goal, we investigated the influence of eCa 2ϩ level on SP1 protein expression by immunoblotting. Thus, chondrocytes were challenged with TGF-␤1 for 24 h while modulating either the eCa 2ϩ level, the eCa 2ϩ entrance (preincubation of 1 h with the efficient VOC blockers lercanidipin and NiCl 2 ), or the cCa 2ϩ level (1 h of pre-exposure with BAPTA or ionomycin). We completed this study using chondrocytes either pretreated for 1 h with WP631 (Merck; 1 M; inhibitor of the DNA-binding ability of SP1) or transfected with a control siRNA or a mix of siRNA targeting Sp1 (Santa Cruz Biotechnology; 10 nM). We then exposed cells to TGF-␤1 for either 24 h (mRNA expression), 48 h (protein level), or 72 h (ePP i production) in the presence of 1.25 mM eCa 2ϩ . As shown in Fig. 5A, a physiological eCa 2ϩ level alone was sufficient to elevate SP1 protein level compared with the no calcium condition. Interestingly, although TGF-␤1 was able to induce SP1 expression by itself, this effect increased gradually with eCa 2ϩ concentration. In chondrocytes pretreated with lercanidipin or NiCl 2 , the TGF-␤1-induced increase of SP1 was totally suppressed, as was also the case for BAPTA (Fig. 5A). Ionomycin increased the total expression of SP1 by itself but blunted the stimulatory effect of TGF-␤1 on this protein (Fig. 5A). When we assessed the phosphorylation of SP1 after 30 min of stimulation, we showed that TGF-␤1 induced SP1 phosphorylation (Fig. 5A). This activation of SP1 was totally suppressed by BAPTA. In contrast, ionomycin enhanced both the basal and the TGF-␤1-induced increase in SP1 phosphorylation (Fig. 5A). Neither WP631 nor Sp1 silencing modulated the basal ePP i production (Fig. 5B). In contrast, WP631 significantly decreased the TGF-␤1-stimulated ePP i production from 274 to 99 pmol/g protein (Ϫ64%). Similarly, Sp1 silencing reduced the TGF-␤1-induced ePP i production to 87 pmol/g protein (Ϫ68%) (Fig. 5B). Once again, neither WP631 nor Sp1 silencing modified the basal mRNA expression of Ank and PC-1 (Fig.  5C). However, WP631 reduced the TGF-␤1 induction from 19.2 to 7.5 (Ϫ61%) for Ank and from 9.6 to 5 (Ϫ48%) for PC-1 mRNA expression. Similarly, Sp1 silencing decreased the TGF-␤1-induced Ank expression to 5.7 (Ϫ70%) as well as decreasing the PC-1 level to 4.8 (Ϫ50%). Western blot analysis (Fig. 5D) validated the efficiency of Sp1 silencing and confirmed its major impact on the stimulatory effects of TGF-␤1 on ANK and PC-1 protein level. These results show that SP1 contributes to the modulatory influence of calcium on the TGF-␤1-induced production of ePP i .

Effect of a Selective Inhibition of TGF-␤1-induced Signaling Pathways on cCa 2ϩ Level and Transcription Factor Activation-
To evaluate further the respective roles of calcium mobilization and TGF-␤1 in the activation of signaling pathways, we assessed the consequence of a selective inhibition of ERK1/2 and PKC␣. First, cells were preincubated for 1 h in the presence of 1.25 mM eCa 2ϩ with one of the following selective inhibitors: PD98059, Gö6976, and WP631. Then cCa 2ϩ levels cells were determined after 1 min of stimulation with TGF-␤1. As shown in Fig. 6A, neither the basal nor the TGF-␤1-induced increase in cCa 2ϩ level was affected by the inhibition of either signaling pathway. Second, we assessed whether inhibition of ERK1/2 and PKC␣ activation could alter SP1 expression and phosphorylation status because cross-talks between ERK1/2 and PKC␣ (25) and between ERK2 and SP1 (26) pathways have been reported. For that purpose, in the same preincubation conditions as above, chondrocytes were stimulated with TGF-␤1 for 30 min (SP1 phosphorylation) or 24 h (total SP1 expression level) in the presence of 1.25 mM eCa 2ϩ . As demonstrated in Fig. 6B, the TGF-␤1-induced phosphorylation of SP1 was totally suppressed by MEK1 inhibition (PD98059 treatment), whereas Gö6976-induced blockade of PKC␣ activation had no effect. Third, we assessed whether the blockade of ERK1/2 and PKC␣ activation could influence the phosphorylation of the transcription factor ELK-1, described as a downstream effector of ERK1/2 activation (27). With that goal, in the same preincubation conditions as above, ELK-1 phosphorylation was evaluated after 30 min of stimulation of chondrocytes with TGF-␤1 in the presence of 1.25 mM eCa 2ϩ . As seen in Fig. 6C, the TGF-␤1-induced phosphorylation of ELK-1 was totally suppressed by PD98059 treatment, whereas Gö6976-induced blockade of PKC␣ activation remained ineffective.
These results demonstrated that TGF-␤1-induced mobilization of cCa 2ϩ is an early event independent of the activation of ERK1/2, PKC␣, and SP1. Moreover, TGF-␤1 is responsible for a ERK1/2-dependent activation of SP1 and ELK-1, independently of PKC␣ activation.
Regulation of Ank Promoter by cCa 2ϩ Level, Sp1, and Elk-1 in Chondrocytes-Because we demonstrated previously that Ank contributed much more than PC-1 to the TGF-␤1-induced production of ePP i by chondrocytes (5), we therefore focused our investigation on the Ank promoter sequence. For that purpose, we challenged chondrocytes, electroporated with either the Ϫ720 bp or the Ϫ2715 bp Ank promoter construct (Fig. 7A), with TGF-␤1 for 48 h in the presence of 1.25 mM eCa 2ϩ while modulating the eCa 2ϩ entrance (pretreatment of 1 h with lercanidipin or NiCl 2 ) or the cCa 2ϩ level (1 h of pre-exposure with BAPTA or ionomycin).
As depicted in Fig. 7B, neither the VOCs blockade nor the dramatic changes in cCa 2ϩ modified the basal firefly/Renilla FIGURE 5. Contribution of SP1 to the effect of eCa 2؉ level and selected Ca 2؉ modulators in the chondrocyte response to TGF-␤1. A, effect of eCa 2ϩ level, selected VOC blockers, and cCa 2ϩ modulators on SP1 expression and SP1 phosphorylation status. Phosphorylated SP1 relative abundance was normalized to both total SP1 and ␤-actin. Blots are representative of three independent experiments. B, influence of Sp1 blocker WP631 and Sp1 silencing on ePP i production. ePP i level was determined in culture supernatants and normalized to total cell proteins. Data (n ϭ 6) are expressed as mean Ϯ S.D. (error bars) in pmol/g protein. C, influence of WP631 and Sp1 silencing on Ank and PC-1 mRNA expression. mRNA levels were normalized to S29 (reference gene) (n ϭ 3). Results are presented as mean Ϯ S.D. over the S29 value. D, influence of Sp1 silencing by a pool of three siRNA sequences on SP1, ANK, and PC-1 protein level. Their relative abundance was normalized to ␤-actin. Images are representative of three independent experiments. Except for A, cells were maintained for 24 h in medium containing 1.25 mM eCa 2ϩ before starting these experiments. Statistically significant differences versus vehicle are indicated as p Ͻ 0.05 (*), and significant differences versus TGF-␤1 condition are shown as p Ͻ 0.05 (#).

Calcium Entrance Stimulates TGF-␤1-induced PP i Export
ratio in chondrocytes transfected with the Ϫ720 bp Ank construct. In contrast, TGF-␤1 increased the luciferase activity by 1.3-fold. This weak induction was totally suppressed by the blockade of VOCs as well as by cCa 2ϩ chelation. In contrast, ionomycin increased the effect of TGF-␤1 from 1.3-to 1.6-fold. In chondrocytes electroporated with the Ϫ2715 bp fragment, the basal firefly/Renilla ratio was increased by 2.7-fold by TGF-␤1 (Fig. 7B). When L-and T-VOCs were blocked, the stimulatory effect of TGF-␤1 was reduced from 2.7-fold to 2and 1.3-fold, respectively. Similarly, BAPTA decreased the TGF-␤1 effect from 2.7-to 1.3-fold. In contrast, ionomycin alone was sufficient to induce a 1.4-fold activation of Ank promoter but also boosted the TGF-␤1-stimulated activation from 2.7-to 4.4-fold (Fig. 7B). Besides highlighting the regulatory role of cCa 2ϩ level, the highest response to TGF␤-1 obtained with the long promoter construct suggested a critical and major contribution of ELK-1. Because we showed in the present study that TGF-␤1 increased SP1 and ELK-1 phosphorylation, we focused on the possible regulatory role of both SP1 and ELK-1 on the promoter constructs. In the presence of 1.25 mM eCa 2ϩ , we analyzed the effects of a 48-h overexpression of either Sp1 or Elk-1 or both in chondrocytes, electroporated with either the Ϫ720 bp or the Ϫ2715 bp fragment. As depicted in Fig. 7C, in chondrocytes transfected with the Ϫ720 bp fragment, overexpression of Elk-1 remained ineffective. In contrast, overexpression of Sp1 alone and in combination with Elk-1 induced the basal firefly/Renilla ratio by 1.2-fold. Moreover, in chondrocytes electroporated with the Ϫ2715 bp construct, overexpression of Sp1 raised the basal luciferase ratio by 1.3-fold, whereas overexpression of Elk-1 elevated it by 1.8-fold (Fig. 7C). Likewise, simultaneous overexpression of Sp1 and Elk-1 increased the basal firefly/Renilla ratio until 2.6-fold. Finally, to further evaluate the respective contribution of SP1 and ERK1/2, we assessed the activity of Ank promoter constructs in chondrocytes challenged or not with TGF-␤1 for 48 h after a 1-h pretreatment with WP631 or PD98059. As depicted in Fig. 7D, the basal firefly/Renilla ratios were unaffected by either inhibitor. In chondrocytes transfected with the Ϫ720 bp construct, the stimulating effect of TGF-␤1 on the Ank promoter activity was reduced from 1.3-to 1.2-fold in the presence of WP631 or PD98059 (Fig. 7D). This result confirmed that activation of SP1 was a downstream event of ERK1/2 activation, consistent with the data provided in Fig. 6B. In chondrocytes transfected with the Ϫ2715 bp construct, the stimulating effect of TGF-␤1 on the reporter activity was reduced from 2.7-fold to 2-and 1.55-fold in the presence of WP631 and PD98059, respectively (Fig. 7D). The greater inhibitory potency of PD98059 supported the dual contribution of SP1 and ELK-1 to the control of the Ank promoter by TGF-␤1, consistent with the data obtained with overexpressing plasmids (Fig. 7C).
These results demonstrate that calcium is a key regulator of activation of the Ank promoter by TGF-␤1 in chondrocytes. This activation requires the phosphorylation of ERK1/2, which in turn activates the SP1 and ELK-1 transcription factors, where SP1 seems to behave as a transcriptional co-activator for ELK-1.

DISCUSSION
Although the clinical association between ACC and primary hyperparathyroidism was established 40 years ago (28), very few studies have addressed the relationship between CPPD deposition and calcium levels in extracellular fluids. Recently, Volpe et al. (10) demonstrated the high frequency of ACC in patients with familial hypocalciuric hypercalcemia and ACC. However, the molecular mechanisms linking high Ca 2ϩ levels to ACC still needed to be elucidated.
Because an increase in PTH-mid-region fragments was noticed in patients with CPPD crystal deposition disease (7), the possible influence of PTH on Ank and PC-1 expression was studied previously in our laboratory. 4 At that time, we failed to  . B, influence of ERK1/2 and PKC␣ activation blockers on TGF-␤1mediated activation of SP1. Phospho-SP1 relative abundance was normalized to both total SP-1 and ␤-actin. Images are representative of three independent experiments. C, influence of ERK1/2 and PKC␣ activation blockers on TGF-␤1-mediated activation of ELK-1. Phospho-ELK-1 relative abundance was normalized to non-phospho-ELK-1. Their relative abundance was normalized to ␤-actin. Blots are representative of four independent experiments. Cells were maintained for 24 h in medium containing 1.25 mM eCa 2ϩ before starting these experiments. observe any difference between controls and PTH-stimulated chondrocytes. This finding is consistent with a previous study that described the ineffectiveness of PTH on PC-1 enzymatic activity in human osteoblast-like cells (29). Therefore, we focused our work on the regulation of the ePP i /P i balance by TGF-␤1, whose elevated synovial fluid levels were reported to be associated with CPPD formation in OA patients (30).
In the present study, we paid first special attention to the characterization of the articular phenotype of the chondrocytes. We confirmed the absence of tissue-nonspecific alkaline phosphatase mRNA, thus indicating that our evaluation of the ePP i levels throughout the study was not biased by its possible hydrolysis into eP i . As a consequence, no eP i accumulation could occur in the culture medium. This feature is critical because eP i was also demonstrated to increase the TGF-␤1induced expression of Ank in chondrogenic cells (31), a finding that we confirmed recently in articular chondrocytes supplemented with eP i (32). Moreover, eP i was previously shown to enhance the expression of matrix ␥-carboxyglutamic acid (Gla) protein in growth plate chondrocytes, thus inhibiting the mineralization process (33). This suggests that eP i alone could orient the chondrocyte phenotype toward a hypertrophic state, whereas we demonstrated previously that ePP i contributed to the maintenance of the differentiated chondrocyte phenotype (18).
The present work demonstrates clearly that increasing eCa 2ϩ concentrations potentiated the TGF-␤1-induced production of ePP i in a concentration-related fashion by increasing Ank and PC-1 expression. These results are consistent with the very recent work from Oca et al. (31), who showed that the up-regulation of Ank by TGF-␤1 was increased in the presence of a higher eCa 2ϩ level during chondrogenesis of the ATDC5 cell line. However, neither the dose dependence on eCa 2ϩ concentration nor the deciphering of the underlying signaling pathway implicated was determined. A possible regulation by calcium could occur by CaSR activation, which, however, failed to modify the TGF-␤1-mediated ePP i production in our experimental conditions. This lack of efficiency of GdCl 3 was consistent with the differentiated phenotype of chondrocyte because a strong expression of CaSR was reported in growth plate or hypertrophic chondrocytes rather than in articular chondrocytes (34). Altogether, these findings demonstrated that the eCa 2ϩ pool modulated the control of the ePP i /eP i balance by TGF-␤1, providing a possible mechanistic link between hypercalcemia and secondary CPPD deposition.
In our experimental conditions, TGF-␤1 provoked an entrance of eCa 2ϩ in chondrocytes, which depended on L-and T-VOCs. The eCa 2ϩ input was consistent with previous works that have demonstrated a similar influx in other cell types, like  4). Statistically significant differences versus vehicle are indicated as p Ͻ 0.05 (*), and significant differences versus TGF-␤1 condition are shown as p Ͻ 0.05 (#). C, influence of Sp1, Elk-1, and Sp1 ϩ Elk-1 overexpression on the basal activity of Ank promoter. Chondrocytes were transfected either with the Ϫ720 or Ϫ2715 bp construct and co-transfected with either pCMV-Sp1 or pCMV-Elk-1 or with both. Normalization was performed using the TK-Renilla reporter gene. Data are expressed as the mean -fold induction of firefly/Renilla luciferase activity ratio Ϯ S.D. (n ϭ 4). Statistically significant differences versus empty vectortransfected cells are indicated as p Ͻ 0.05 (*), and significant differences versus single construct-transfected cells are shown as p Ͻ 0.05 (#). D, influence of SP1 (WP631) or ERK1/2 (PD98059) blockade on the basal and calciuminduced activity of Ank promoter. Chondrocytes were transfected with either the Ϫ720 or Ϫ2715 bp construct. Normalization was performed using the TK-Renilla reporter gene. Data are expressed as the mean -fold induction of firefly/Renilla luciferase activity ratio Ϯ S.D. (n ϭ 4). Statistically significant differences versus cells without TGF-␤1 at 1.25 mM eCa 2ϩ are indicated as p Ͻ 0.05 (*), and significant differences versus TGF-␤1-stimulated cells are shown as p Ͻ 0.05 (#). Cells were maintained for 24 h in medium containing 1.25 mM eCa 2ϩ before starting these experiments. fibroblasts (35) or vascular smooth muscle cells (36). In the present study, the contribution of VOCs was evaluated indirectly by their blockade with selective inhibitors, but there was a good correlation between the changes observed in cCa 2ϩ levels and ePP i levels in our study. The contribution of L-VOCs to TGF-␤1-induced calcium entrance was shown recently in ATDC5 cells, although no contribution of T-VOCs was reported (31). This was quite unexpected because T-VOCs were shown to be expressed not only in primary chondrocytes but also in the ATDC5 cell line (12). Our data on cCa 2ϩ level support a major role for T-VOCs compared with L-VOCs. In contrast, -agatoxin was unable to block the calcium influx in our experimental system, although a modest, probably Ca 2ϩ -independent effect was observed on ePP i levels and expression of Ank and PC-1. This result is in line with the current lack of a demonstration that P/Q-VOCs are expressed in articular chondrocytes or in ATDC5 cells.
The link between IP 3 and calcium is well described in chondrocytes because other studies have suggested (37) or demonstrated (38) an IP 3 -dependent calcium release from the internal stores, using hypotonic shock and fluid flow experiments, respectively. It is noteworthy that, in mesangial cells, TGF-␤1 induced a calcium mobilization that depended on both IP 3 release and the disruption of the actin cytoskeleton organization (39). In our study, we used U73122, an inhibitor of PLC, which cleaves phosphatidylinositol 2-phosphate in IP 3 and diacylglycerol. We demonstrated that U73122 was ineffective on the TGF-␤1-mediated calcium influx, indicating that this phenomenon was independent of the IP 3 pathway in articular chondrocytes. Similar results were reported in chondrocytes in which a calcium influx was generated by an electrical stimulation (40). Thus, the variable implication of IP 3 could depend on the nature of the stimulation used (i.e. localized voltage modifications versus cell membrane stretches). TGF-␤1 was also shown to modify the cell membrane voltage by modulating K ϩ currents in microglia cells (41). Interestingly, rat articular chondrocytes express such K ϩ channels (42), suggesting that electrophysiological changes may have contributed to the observed IP 3 -independent mechanism. Altogether, these findings support the hypothesis of a quick TGF-␤1-mediated modification of the K ϩ currents, resulting in the opening of L-and T-VOCs, and secondary calcium influx into chondrocytes. This was already demonstrated in pituitary cells, where a variation in K ϩ currents induced an L-VOC-dependent Ca 2ϩ influx (43).
Although ionomycin alone induced an increase in cytosolic calcium level in chondrocytes (calcium released from internal stores), the addition of TGF-␤1 further increased the Ca 2ϩ mobilization in our system. Therefore, the TGF-␤1-induced production of ePP i is thought to depend on the availability of the eCa 2ϩ pool rather than on intracellular stores. Indeed, we showed here that no significant calcium mobilization was detectable in the absence of eCa 2ϩ in the culture medium and that cCa 2ϩ levels measured were higher in the presence of elevated eCa 2ϩ concentrations. Likewise, another work (40) reported an exclusive dependence of the increase in aggrecan and type II collagen mRNA to the eCa 2ϩ reservoir in electrically stimulated chondrocytes.
We deciphered previously the signaling pathways activated by TGF-␤1 responsible for Ank induction and thus ePP i production by articular chondrocytes (5). These were Smad-independent but relied on ERK1/2 and Ca 2ϩ -dependent PKCs. Oca et al. (31) confirmed recently that the induction of Ank expression depended on activation of Ca 2ϩ -dependent PKCs. We demonstrate here that ePP i production mostly depended upon the synergistic action of TGF-␤1 and eCa 2ϩ input, which both controlled the protein level of the transcription factor SP1. We also showed that TGF-␤1 induced the phosphorylation of SP1 in a ERK1/2-dependent mechanism, as demonstrated by others for the regulation of the gastrin promoter (26). TGF-␤1 is often described as a biological agent promoting SP1 activity. For example, TGF-␤1 enhanced the DNA binding capacity of SP1 in trabecular bone-derived cells (44), while promoting the SP1-dependent transcription of type I collagen in fibroblasts (45). Calcium was also demonstrated to be critical for the ability of SP1 to stimulate the basal (14) and TGF-␤1-induced expression of ␤1,3-glucuronosyltransferase I (16).
Our experiments underline that the calcium-increased effects of TGF-␤1 on ePP i metabolism were supported by an enhanced activation of the ERK1/2 and PKC␣ pathways. This is consistent with another study (46) that has shown the calcium dependence of ERK1/2 phosphorylation during the induction of the phosphate transporters Glvr-1 and -2 by P i in odontoblasts. It is noteworthy that we demonstrated that the ability of eCa 2ϩ to potentiate the phosphorylation of ERK1/2 by TGF-␤1 was probably supported by the activation of Ras because the blockade of VOCs inhibited both Ras and ERK1/2 activation. These data are consistent with the contribution of the Ras/ Raf-1 pathway to the stimulatory effect of TGF-␤1 on Ank expression at a "physiological" (1.25 mM) free eCa 2ϩ level (5) and the previous demonstration that eCa 2ϩ influxes could activate Ras, which is subsequently able to activate MEK1 and MAPK (47). Such a close relationship was further supported by the inability of TGF-␤1 to activate Ras in the absence of eCa 2ϩ or after Ca 2ϩ chelation. Additionally, PKCs are well described to directly activate ERK1/2 (48 -50), and Ca 2ϩ -dependent PKCs are key regulators of Ank expression in chondrocytes. Taken together, these data indicate that a critical cytosolic Ca 2ϩ level must be reached to activate Ras, ERK1/2, and PKC␣. Such a level is achieved at a physiological free eCa 2ϩ level by a TGF-␤1-dependent calcium influx through L-and T-VOCs.
Finally, using two constructs of the Ank promoter, we demonstrated that the presence of the Elk-1-responsive element was indisputable for the inducing effect of TGF-␤1. This finding corroborated the ERK1/2-dependent activation of ELK-1 by TGF-␤1. A comparable signaling cascade was reported for the regulation of type I collagen expression in fibroblasts (51) and after the stimulation of the human chondrosarcoma-derived chondrocytic cell line HCS-2/8 by connective tissue growth factor stimulation (52).
It is noteworthy that a synergistic action of ELK-1 and SP1 was detectable in our experimental conditions. One can suggest the formation of a transcription enhancer complex containing these two transcription factors, as observed during the lipopo-lysaccharide-induced expression of tumor necrosis factor-␣ in macrophages (53). Therefore, the cooperative effect between SP1 and ELK-1 could depend upon a direct protein-protein interaction, although Tsai et al. (53) are rather in favor of an indirect interaction through a protein scaffold depending on CBP/p300.
To summarize, we show here that high extracellular calcium level strongly increases the TGF-␤1-induced production of ePP i . This effect is due to an enhancement of Ank and PC-1 mRNA and protein expression. The underlying molecular mechanisms implicated include a TGF-␤1-mediated calcium influx through L-and T-VOCs. This results in an elevation of SP1 level, the activation of Ras, and the phosphorylation of ERK1/2 and PKC␣, to enhance finally the transcriptional activity of ELK-1 and SP1 acting synergistically on Ank expression.
Our results provide for the first time a possible mechanistic explanation for the onset of ACC secondary to primary hyperparathyroidism. This could support the research of new therapeutic options, such as the use of calcimimetics. For example, cinacalcet has proven to reduce biochemical parameters such as PTH and hypercalcemia in patients with primary or secondary hyperparathyroidism (54). However, clinical outcomes in ACC still need to be demonstrated. Calcimimetics target the CaSR, which remains very marginally expressed in articular cartilage (34). However, we show here that articular chondrocytes are very responsive to high eCa 2ϩ level by producing ePP i . Therefore, modulation of calcium levels and/or entrance could be an interesting avenue of clinical research aiming at the prevention of secondary ACC.