Anoctamin-6 Controls Bone Mineralization by Activating the Calcium Transporter NCX1*

Background: Deletion of Ano6 leads to reduced skeleton size and skeletal deformities. Results: Ca2+-activated anion currents are missing in primary osteoblasts from Ano6 null mice. Ano6 is required for the Na+/Ca2+ exchanger NCX1 to operate in osteoblasts. Conclusion: The Na+/Ca2+ exchanger NCX1 is essential for bone mineralization and requires Ano6 to fully operate. Significance: Ano6 supports NCX1, which is essential for bone mineralization. Anoctamin-6 (Ano6, TMEM16F) belongs to a family of putative Ca2+-activated Cl− channels and operates as membrane phospholipid scramblase. Deletion of Ano6 leads to reduced skeleton size, skeletal deformities, and mineralization defects in mice. However, it remains entirely unclear how a lack of Ano6 leads to a delay in bone mineralization by osteoblasts. The Na+/Ca2+ exchanger NCX1 was found to interact with Ano6 in a two-hybrid split-ubiquitin screen. Using human osteoblasts and osteoblasts from Ano6−/− and WT mice, we demonstrate that NCX1 requires Ano6 to efficiently translocate Ca2+ out of osteoblasts into the calcifying bone matrix. Ca2+-activated anion currents are missing in primary osteoblasts isolated from Ano6 null mice. Our findings demonstrate the importance of NCX1 for bone mineralization and explain why deletion of an ion channel leads to the observed mineralization defect: Ano6 Cl− currents are probably required to operate as a Cl− bypass channel, thereby compensating net Na+ charge movement by NCX1.

During vertebrate skeletal development, specialized cells, the osteoblasts, produce a mineralized bone matrix by deposition of hydroxyapatite crystals in the extracellular collagen matrix, where they become embedded and differentiate into osteocytes (1)(2)(3). This process requires the organized transport of high concentrations of Ca 2ϩ and phosphate ions into the extracellular matrix without inducing premature crystallization. Although of fundamental importance, the process of calcification is not well understood. Different concepts currently exist. (i) A cell-independent mechanism has been suggested, whereby charged non-collagenous proteins associate with the collagen matrix to form mineralization foci (4). (ii) The amorphous calcium phosphate theory predicts that initial amorphous calcium phosphate is deposited, which gradually converts to pure crystalline hydroxyapatite (5,6). (iii) A cell-controlled mechanism that requires budding of vesicles from the plasma membrane has also been proposed. These vesicles accumulate ions and organize the formation of calcium phosphate crystallization seeds containing hydroxyapatite particles. Subsequently, rupture of these vesicles spreads calcium phosphate onto the extracellular matrix (7,8). (iv) Recent work identified intracellular vesicles that contain mineral in a mostly amorphous phase form (9,10).
Whatever the precise mechanism is, mineralization requires Ca 2ϩ transport. The mechanisms of Ca 2ϩ transport by osteoblasts are not well understood at the molecular level. The Ca 2ϩpermeable transient receptor potential channels TRPV4 and TRPV6 are expressed in bone cells, but whether they affect bone mineralization has not yet been elucidated (11). Another group of Ca 2ϩ transporters, the plasma membrane Ca 2ϩ -ATPases, are strong candidates to transport Ca 2ϩ into the extracellular compartment. PMCA1, PMCA2, and PMCA4 are expressed in osteoblasts, but their role in mineralization in mammals in vivo has not been demonstrated. In zebrafish, the PMCA2 homolog Atp2b1a seems to be required for mineralization by odontoblasts and possibly the adjacent bone structures (11)(12)(13)(14). A third class of Ca 2ϩ transporters are the Na ϩ / Ca 2ϩ exchangers (NCX). Of these, NCX1 and NCX3 are expressed in mouse osteoblasts, and NCX3 has been demonstrated to regulate Ca 2ϩ transport in osteoblastic cell lines (14 -17).
Recently, anoctamin-6 (Ano6, TMEM16F) has been reported as another protein required for proper bone mineralization (1). Ano6 belongs to a family of putative Ca 2ϩ -activated Cl Ϫ channels (18 -21). Several reports demonstrated that Ano6 generates an outwardly rectifying Cl Ϫ current when activated by a large increase in local intracellular Ca 2ϩ concentrations (22)(23)(24). Moreover, Ano6 operates as a Ca 2ϩ -activated phospholipid scramblase that leads to movement of phosphatidylserine from the inner to the outer leaflet of the plasma membrane bilayer (25). Impaired Ca 2ϩ -dependent phospholipid scrambling of phospholipids in platelets leads to Scott syndrome, a rare inherited defect in primary hemostasis due to mutations in Ano6 (25,26). The defect in Ca 2ϩ -dependent phospholipid scrambling in cells from patients with Scott disease is paralleled by lack of Ca 2ϩ -dependent activation of Cl Ϫ currents, which is necessary for full platelet phosphatidylserine exposure and procoagulant activity (26,27). Recently, reconstitution of a purified Ano6 homolog from Aspergillus fumigatus confirmed that Ano6 has both ion channel and scramblase function and that transport of both ions and lipids is controlled by a single Ca 2ϩbinding site (28).
A role of Ano6 in activating phosphatidylserine scrambling in osteoblasts has been demonstrated previously (1). However, its function as a Ca 2ϩ -activated Cl Ϫ channel has not been investigated in osteoblasts. In this study, we therefore examined the role of Ano6 Cl Ϫ currents in the mineralization process. Notably, in a yeast two-hybrid split-ubiquitin screen using Ano6 as bait, we found the Na ϩ /Ca 2ϩ exchanger NCX1 to be a potential binding partner, suggesting a molecular and functional link between Ano6 and NCX1.

EXPERIMENTAL PROCEDURES
Ano6-deficient Mice, RT-PCR, and Two-hybrid Analysis-Generation of Ano6 (TMEM16F)-deficient mice has been described (1). Total RNA was isolated from human and mouse osteoblasts using NucleoSpin RNA II columns (Macherey-Nagel GmbH & Co. KG, Düren, Germany). Total RNA (1 g/50-l reaction) was reverse-transcribed using random primers (Promega, Mannheim, Germany) and Moloney murine leukemia virus reverse transcriptase (RNase H Minus; Promega). Each RT-PCR contained sense and antisense primers for anoctamins (0.5 M; see supplemental Tables 1 and 2) or for GAPDH (0.5 M), 0.5 l of cDNA, and GoTaq polymerase (Promega). After 2 min at 95°C, cDNA was amplified in 30 cycles for 30 s at 95°C, 30 s at 57°C, and 1 min at 72°C. PCR products were visualized by loading on ethidium bromide-containing agarose gels and analyzed using MetaMorph Version 6.2 (Molecular Devices). The two-hybrid split-ubiquitin screenings were performed using a normalized human universal NubG-x library with human Ano6 as a bait (Dualsystems Biotech AG, Zurich, Switzerland).
Primary Osteoblasts, Mineralization Assay, and HEK293 Cells-Frontal and parietal parts of mouse calvariae were dissected from day 18.5 embryos, and primary osteoblasts were isolated by two 30-min digestion steps in 0.1% collagenase IV (Sigma). Cells from individual animals were cultured separately, seeded at a density of 10 4 cells/cm 2 , and cultured in DMEM supplemented with 10% fetal bovine serum and penicillin/streptomycin. Upon confluence, cells were reseeded at a density of 10 4 cells/cm 2 and cultured in the presence of 10 mM glycerophosphate and 5 mM ascorbate phosphate to induce osteogenic differentiation and mineralization (day 0). After establishing a nontoxic effective concentration, cells were treated with 5 M tannic acid or KB-R7943 (29). The potential toxicity of these inhibitors was excluded by negative trypan blue staining and by assessing the effect of these inhibitors on cell morphology (data not shown). Moreover, previous in vivo and in vitro studies demonstrated the nontoxicity of these inhibitors (30 -32).
During differentiation, triplicate wells of each culture were fixed at days 3, 10, 17, and 20 in methanol and washed in deionized water. Staining intensity was quantified as described by Gregory et al. (33). Human primary osteoblasts were purchased from PromoCell (Heidelberg, Germany). Cells were grown in either osteoblast growth medium or osteoblast mineralization medium (PromoCell) according to the manufacturer's protocol. Mineralization was visualized by Alizarin Red staining, which was subsequently quantified (33). Briefly, Alizarin Red S was extracted from the cell layer in 10% acetic acid, and cell debris was removed by centrifugation and quantified at 405 nm in a microplate reader (NOVOstar, BMG LABTECH, Offenburg, Germany). HEK293 human embryonic kidney cells were grown in DMEM supplemented with 10% FBS. Cells were incubated in 5% CO 2 at 37°C. pcDNA3.1 encoding human Ano6 and human NCX1 was transfected into HEK293 cells using standard methods (Lipofectamine, Invitrogen). Cells were examined 48 or 72 h after transfection.
To compare amounts of proteins, the membranes were probed with rabbit anti-␤-actin antibody (diluted 1:1000 in 1% NFM/ TBST; Sigma). The intensity of bands was quantified by densitometric scanning and normalized to ␤-actin.
Cell lysates in Nonidet P-40 lysis buffer were precleaned with protein G-agarose at 4°C for 1 h. Cell lysates were incubated with 5 g of primary antibody (goat anti-GFP or rabbit anti-NCX1 polyclonal antibody) overnight at 4°C on a rotator. Cell lysates (70 l containing 50 g of protein) were incubated with protein G-agarose at 4°C for 3 h, followed by centrifugation three times at 1000 ϫ g for 1 min. Pellets were collected and resuspended in 1ϫ loading buffer. The supernatant was collected, separated by 10% SDS-PAGE, and transferred to PVDF membrane. The membrane was blocked with 5% NFM/TBST for 1 h at room temperature and incubated overnight at 4°C with goat anti-GFP polyclonal antibody (diluted 1:3000 in 1% NFM/TBST) or rabbit anti-NCX1 polyclonal antibody (diluted 1:1000 in 5% NFM/TBST). Subsequently, the membrane was incubated with HRP-conjugated donkey anti-goat IgG (diluted 1:2000 in 5% NFM/TBST) at room temperature for 1 h or with HRP-conjugated goat anti-rabbit IgG (diluted 1:10,000 in 1% NFM/TBST) at room temperature for 2 h. Immunoreactive signals were visualized using the SuperSignal chemiluminescence substrate detection kit.
Patch Clamping-Cells grown on coverslips were mounted in a perfused bath on the stage of a Zeiss IM35 inverted microscope and kept at 37°C. The bath was perfused continuously with Ringer's solution (145 mM NaCl, 0.4 mM KH 2 PO 4 , 1.6 mM K 2 HPO 4 , 6 mM D-glucose, 1 mM MgCl 2 , and 1.3 mM calcium gluconate (pH 7.4)) at ϳ10 ml/min. Patch clamp experiments were performed in the fast whole cell configuration. Patch pipettes had an input resistance of 4 -6 megohms when filled with an intracellular-like solution containing 30 mM KCl, mM 95 potassium gluconate, 1.2 mM NaH 2 PO 4 , 4.8 mM Na 2 HPO 4 , 1 mM EGTA, 0.758 mM calcium gluconate, 1.034 mM MgCl 2 , 5 mM D-glucose, and 3 mM ATP. The pH was 7.2, and the Ca 2ϩ activity was 0.1 M or 1 mM. The access conductance was measured continuously and was 90 -140 nanosiemens (EPC-9 amplifier, List Medical Electronics, Darmstadt, Germany). In regular intervals, membrane voltages were clamped in steps of 20 mV from Ϫ100 to ϩ100 mV from a holding potential of Ϫ60 mV.
Intracellular Ca 2ϩ Concentrations in HEK293 Cells Coexpressing Ano6 and NCX1-Cells were seeded on glass coverslips and loaded with 2 M Fura-2/AM and 0.02% Pluronic F-127 (Invitrogen) in Ringer's solution (145 mmol/liter NaCl, 0.4 mmol/liter KH 2 PO 4 , 1.6 mmol/liter K 2 HPO 4 , 5 mmol/liter glucose, 1 mmol/liter MgCl 2 , and 1.3 mmol/liter calcium gluconate) for 1 h at room temperature. Fluorescence was detected in cells perfused with Ringer's solution at 37°C using an inverted microscope (Axiovert S100, Zeiss) and a high speed polychromator system (VisiChrome, Puchheim, Germany). Fura-2 was excited at 340/380 nm, and emission was recorded between 470 and 550 nm using a CoolSNAP HQ camera (Visitron Systems). [Ca 2ϩ ] i was calculated from the 340/380 nm fluorescence ratio after background subtraction. The formula used to calculate [Ca 2ϩ ] i was as follows; where R is the observed fluorescence ratio. The values R max and R min (maximum and minimum ratios) and the constant S f2 /S b2 (fluorescence of free and Ca 2ϩ -bound Fura-2 at 380 nm) were calculated using 1 M ionomycin (Calbiochem), 5 M nigericin, 10 M monensin (Sigma), and 5 mM EGTA to equilibrate intracellular and extracellular Ca 2ϩ in intact Fura-2-loaded cells. The dissociation constant for the Fura-2⅐Ca 2ϩ complex was taken as 224 nmol/ Removal of Cl Ϫ ions inhibited whole cell currents, indicating activation of Cl Ϫ currents by ionomycin. b, current/voltage relationships before (control (con)) and after stimulation with ionomycin in the presence of high (145 mM; left) or low (5 mM; right) extracellular Cl Ϫ . V c , voltage clamp. c, whole cell currents were not enhanced by ionomycin in knock-out osteoblasts (Ano6 Ϫ/Ϫ ). d, corresponding current/voltage relationships before and after stimulation with ionomycin. Values represent mean Ϯ S.E. *, significant increase by ionomycin (paired t test). The number of cells is indicated in parentheses.
liter. Experiments, imaging acquisition, and data analysis were controlled with the software packages MetaFluor (Molecular Devices) and Origin (OriginLab Corp.).
Materials and Statistics-All compounds used were the highest grade from Sigma (ionomycin, cyclopiazonic acid, tannic acid, KB-R7943, and CaCCinh-A01) or Merck. Student's t test (for paired or unpaired samples as appropriate) was used for statistical analysis. A p value Ͻ0.05 was accepted as significant.

Ca 2ϩ -dependent Ion Currents and Mineralization by Mouse
Primary Osteoblasts Require Ano6 and NCX-In two-hybrid split-ubiquitin screenings of a normalized human universal NubG-x library with human Ano6 as bait, we identified NCX1 as a potential Ano6-interacting protein. Although two-hybrid screening identified a number of potential Ano6 targets, NCX1 appeared most interesting (Table 1). We further analyzed the potential interaction partner NCX1 because Ano6-deficient mice have been shown to suffer from a bone mineralization defect, whereas NCX1 has been reported previously to control bone mineralization (1,34). Moreover, Ano6 has been shown to form a Ca 2ϩ channel (35) or may alternatively control Ca 2ϩ transport through NCX1. Further evidence for a possible direct coupling of NCX1 and Ano6 was provided by co-immunoprecipitation of both proteins (see Fig. 2c). We hypothesized that both Ano6 and NCX1 may be required for Ca 2ϩ transport by osteoblasts. To test if Ano6 and NCX1 are coexpressed in primary osteoblasts, cells were isolated from WT and Ano6 Ϫ/Ϫ mice and grown in the presence of ␤-glycerophosphate and

Ano6 Activates NCX1
MARCH 6, 2015 • VOLUME 290 • NUMBER 10 ascorbate phosphate to induce osteogenic differentiation. By RT-PCR, we observed expression of Ano6, Ano8, and Ano10 in WT osteoblasts. Expression of Ano6 and Ano10 was upregulated during differentiation (see Fig. 2a). No Ano6 mRNA was detected in osteoblasts from Ano6 Ϫ/Ϫ animals (see Fig. 2a) (1). We found that an increase in [Ca 2ϩ ] i by the Ca 2ϩ ionophore ionomycin activated Cl Ϫ currents only in WT (but not Ano6 Ϫ/Ϫ ) osteoblasts (Fig. 1). Apart from Ano6, also Na ϩ /Ca 2ϩ exchangers (NCX1 and NCX3) were expressed in primary osteoblasts (Fig. 2b). Using Alizarin Red staining, we detected prominent mineralization at days 17 and 20 of mouse primary calvarial cultures (Fig. 2d). Mineralization was delayed after treatment of these osteoblasts with the NCX inhibitor KB-R7943, supporting a role of Na ϩ / Ca 2ϩ exchange in the mineralization process (36). Mineralization was also significantly inhibited by the Ano6 inhibitor tannic acid (30), demonstrating the role of Ano6 Cl Ϫ currents in mineralization. Ano6 Drives the Forward Mode of the Na ϩ /Ca 2ϩ Exchanger-We further examined the functional relationship between Ano6 and NCX1 in HEK293 cells (14). An increase in [Ca 2ϩ ] i by ionomycin activated whole cell Cl Ϫ currents in HEK293 cells overexpressing Ano6 (Fig. 3a) (22). We observed a small negative (Na ϩ inward) current that was inhibited when extracellular Na ϩ was replaced with impermeable N-methyl-Dglucamine (NMDG ϩ ). 2 We suggested that the NMDG ϩ -sensitive inward current is due to electrogenic transport by the Na ϩ / Ca 2ϩ exchanger NCX1 (Fig. 3, a and b). Endogenous expression of NCX1 in HEK293 cells was effectively knocked down by siRNA (Fig. 3, c and d). Inhibition of NCX1 expression attenuated NMDG ϩ -sensitive Na ϩ inward currents and augmented Cl Ϫ outward currents. Thus, ionomycin-induced increase in [Ca 2ϩ ] i appears to be larger in the absence of Na ϩ or after reduced expression of NCX1. Measurements of intracellular Ca 2ϩ demonstrated that inhibition of NCX1 by NMDG ϩ augmented [Ca 2ϩ ] i (Fig. 3h, upper panel). In the presence of extracellular NMDG ϩ , NCX1 operated in the reverse mode, thus transporting Ca 2ϩ into the cell. Blocking NCX1 with KB-R7943 inhibited Ca 2ϩ outward transport; therefore, the Ca 2ϩ increase was less pronounced. Notably, NMDG ϩ had no effects on [Ca 2ϩ ] i in the presence of the NCX1 blocker KB-R7943 (Fig. 3h,  lower panel).
Ion transport was also activated in NCX1/Ano6-expressing cells with high Ca 2ϩ concentrations (1 mmol/liter) in the patch pipette. Notably, NMDG ϩ -inhibitable Na ϩ currents (NCX1) were detected only in the presence of Ano6 (Fig. 3, e and f). Finally, when Ano6 and NCX1 were coexpressed, ionomycinactivated inward currents were larger, whereas outward currents were smaller compared with cells expressing only Ano6 (Fig. 3g). In contrast, inward currents decreased and outward currents increased after inhibition of NCX1 by KB-R7943 (Fig. 4). Both inward and outward currents were abolished by further removal of extracellular Cl Ϫ (5Cl Ϫ ). These results indicate that Cl Ϫ transport through Ano6 is essential for NCX1 to operate in the forward mode, i.e. to transport Ca 2ϩ out of the cell.
Ano6 Supports Outward Transport of Ca 2ϩ through the Na ϩ / Ca 2ϩ Exchanger-We further demonstrated regulation of NCX1 by Ano6 in [Ca 2ϩ ] i measurements using Fura-2. Endoplasmic reticulum Ca 2ϩ stores were emptied by cyclopiazonic acid in Ca 2ϩ -free bath solution (Fig. 5a, upper right scheme). Re-addition of extracellular Ca 2ϩ increased [Ca 2ϩ ] i only transiently through store-operated Ca 2ϩ influx because Ca 2ϩ was transported out of the cell by NCX1 [Ca 2ϩ ] i . NCX1 was inhibited by KB-R7943, Na ϩ replacement with NMDG ϩ , or inhibition of Ano6 currents by removal of extracellular Cl Ϫ (5Cl Ϫ ) ( Fig. 5, a (lower right scheme), b, and c). Moreover, siRNA knockdown of Ano6 also inhibited outward transport of Ca 2ϩ (Fig. 5, d and e). The results demonstrate the essential role of NCX1/Ano6 in controlling [Ca 2ϩ ] i in mouse osteoblasts and HEK293 cells.
Ano6 Supports Ca 2ϩ Transport and Mineralization in Primary Human Osteoblasts-Mouse primary osteoblasts up-regulated Ano6 upon differentiation by ␤-glycerophosphate and ascorbate phosphate (Fig. 2a). Similar results were found in human primary osteoblasts. Human primary osteoblasts expressed predominately Ano6 and Ano10, with Ano6 being strongly up-regulated during differentiation (Fig. 6, a-c). An increase in intracellular Ca 2ϩ by ionomycin activated whole cell currents, which were larger in differentiated cells. Similar, CaCCinh-A01-sensitive Cl Ϫ currents were enhanced in differ-entiated cells (Fig. 6, d-f). These results suggest that similar to recombinant NCX1, also in mouse and human osteoblasts, NCX1 might be regulated by Ano6. We performed a mineralization assay in differentiated primary human osteoblasts, which demonstrated extracellular calcium phosphate deposition (Fig. 7a). Extracellular precipitation made patch clamp experiments often difficult because it limited the access of the patch pipette to the cell membrane. Similar to mouse osteoblasts, also in human osteoblasts, mineralization was strongly inhibited by inhibition of NCX1 with KB-R7943 or blocking of Ano6 with tannic acid (Fig. 7). The fact that blocking NCX1 reduced mineralization less than inhibition of Ano6 suggests additional Ca 2ϩ export mechanisms unrelated to NCX1. Nevertheless, the data indicate that functional coupling of Ano6 and NCX1 is essential for bone min-  eralization, which explains the skeletal defect observed in Ano6 null mice (Fig. 8).

DISCUSSION
Our results indicate that decreased mineral deposition in the skeletons of Ano6 null mice (1) is at least in part due to defective regulation of plasma membrane Na ϩ /Ca 2ϩ exchange. Cl Ϫ transport by Ano6 drives Ca 2ϩ transport by NCX1, as indicated by inhibitors of Ano6 (CaCCinh-A01 and tannic acid), or removal of extracellular Cl Ϫ . All inhibited deposition of hydroxyapatite by osteoblasts. We propose a model in which the Cl Ϫ channel function of Ano6 serves to activate NCX1 and possibly NCX3, thereby increasing extracellular Ca 2ϩ deposition, which is required for formation of hydroxyapatite in the extracellular matrix. As NCX1-or NCX2-deficient mice have not yet been reported, it is difficult to estimate if other Ca 2ϩ transporters are also regulated by Ano6. Likewise, other anion channels may be linked to Ca 2ϩ secretion. It is noteworthy that the cystic fibrosis transmembrane conductance regulator (CFTR) Cl Ϫ channel is expressed in osteoblasts and in ameloblasts/odontoblasts (37). Cystic fibrosis patients carrying the common F508del mutation in the CFTR gene and mice carrying an inactivated CFTR allele suffer from bone mineralization defects leading to osteoporosis and osteopenia (37)(38)(39)(40). The mechanisms of how CFTR operates in osteoblasts are only poorly understood. Nevertheless, the reduced bone mineral density underlines the importance of Cl Ϫ transport in osteoblasts for proper bone formation.
Ano6-dependent Na ϩ /Ca 2ϩ transport may also be related to the nonselective cation conductance that has been reported for Ano6 (35). We propose that instead of directly transporting Ca 2ϩ ions, Ano6 may support the activity of the Na ϩ /Ca 2ϩ exchanger.
The function of Ano6 in osteoblasts is reminiscent of the function of Ano1 in proximal tubular epithelial cells of the kidney (41). In the proximal tubule, protons are secreted by the vacuolar proton pump (V-ATPase). Translocation of H ϩ to the tubular lumen by the Ca 2ϩ -activated V-ATPase is an electrogenic process, and accumulation of charge may limit the transport activity of the pump. Parallel activation of Ano1 leads to Cl Ϫ efflux that counterbalances charge movement, thereby facilitating H ϩ secretion (41).
The mineralization defect in Ano6-deficient mice has previously been discussed as being linked to the scramblase function of Ano6, which is significantly impaired in osteoblasts (1). Here, we have shown that Ano6 facilitates matrix mineralization by supporting the function of the Na ϩ /Ca 2ϩ exchanger (Fig. 8). However, the present data do not elucidate whether Na ϩ /Ca 2ϩ exchangers operate to directly transport Ca 2ϩ into the extracellular collagen matrix, where it gradually converts to crystalline hydroxyapatite (5), or whether Na ϩ /Ca 2ϩ exchange takes place in budding vesicles or in intracellular calcium phosphate-containing vesicles (8 -10). In this respect, it is of interest that exposure of phosphatidylserine has been proposed to be required for the formation of matrix vesicles and/or the localization of these vesicles to the collagen network (2). Ano6 might thus couple two essential functions: Ca 2ϩ secretion and formation of matrix vesicles (Fig. 8).
Acknowledgments-We thank Daniel Auerbach and Nicolas Lentze (Dualsystems Biotech) for performing the two-hybrid split-ubiquitin analysis.  Shown is a schematic model of the roles of Ano6 and NCX1 in bone mineralization. Ca 2ϩ increases in osteoblasts activate outwardly rectifying Ano6 Cl Ϫ currents and the outward transport of Ca 2ϩ by the Na ϩ /Ca 2ϩ exchanger NCX1. The inward transport of Cl Ϫ compensates the cytosolic positive transport potential by the electrogenic Na ϩ /Ca 2ϩ exchanger, thereby maintaining Ca 2ϩ outward transport at a high rate. NCX1 may directly interact with Ano6 (yellow arrow). Scrambling of phosphatidylserine (PS; green) by Ano6 supports the mineralization process.