HCO₃⁻/Cl⁻ anion exchanger SLC4A2 is required for proper osteoclast differentiation and function

Abstract

As the only cell capable of bone resorption, the osteoclast is a central mediator of skeletal homeostasis and disease. To efficiently degrade mineralized tissue, these multinucleated giant cells secrete acid into a resorption lacuna formed between their apical membrane and the bone surface. For each proton pumped into this extracellular compartment, one bicarbonate ion remains in the cytoplasm. To prevent alkalinization of the cytoplasm, a basolateral bicarbonate/chloride exchanger provides egress for intracellular bicarbonate. However, the identity of this exchanger is unknown. Here, we report that the bicarbonate/chloride exchanger, solute carrier family 4, anion exchanger, member 2 (SLC4A2), is up-regulated during osteoclast differentiation. Suppression of Slc4a2 expression by RNA interference inhibits the ability of RAW cells, a mouse macrophage cell line, to differentiate into osteoclasts and resorb mineralized matrix in vitro. Accordingly, Slc4a2-deficient mice fail to remodel the primary, cartilaginous skeletal anlagen. Abnormal multinucleated giant cells are present in the bone marrow of Slc4a2-deficient mice. Though these cells express the osteoclast markers CD68, cathepsin K, and NFATc1, compared with their wild-type (WT) counterparts they are larger, fail to express tartrate-resistant acid phosphatase (TRAP) activity, and display a propensity to undergo apoptosis. In vitro Slc4a2-deficient osteoclasts are unable to resorb mineralized tissue and cannot form an acidified, extracellular resorption compartment. These data highlight SLC4A2 as a critical mediator of osteoclast differentiation and function in vitro and in vivo.

Bone is a remarkable biomaterial composed of organic and inorganic molecules that remodels to preserve structural integrity and adapt to stress. Two cells execute this process: the osteoblast and osteoclast, which synthesize and catabolize bone respectively. An imbalance between osteoclast and osteoblast activity perturbs bone quality, leading to fractures or skeletal deformities (1, 2).

The osteoclast is a multinucleated giant cell that differentiates from myeloid precursors under the influence of the osteoblast-derived cytokines, macrophage-colony stimulating factor (MCSF) and receptor activator for nuclear factor-κB ligand (RANKL). This process is controlled by the transcription factor, nuclear factor of activated T cells c1 (NFATc1), which is induced by RANKL and governs the expression of genes necessary for osteoclast formation and function (3). After differentiation, the osteoclast polarizes, forming a resorption lacuna between its apical membrane and the mineralized bone surface. Within this space, the osteoclast secretes enzymes, as well as acid, via an apical, H⁺-vacuolar ATPase. The low pH of the resorption lacuna activates these proteolytic enzymes and promotes dissolution of crystalline calcium phosphate. To maintain electroneutrality, a parallel chloride/proton antiporter releases a chloride ion with each proton (4). In humans and mice, mutations in either the proton pump (TCIRG1) or the chloride/proton antiporter (CLCN7) lead to osteopetrosis, a condition characterized by high bone mass due to an inability of osteoclasts to remodel bone (5).

The protons released by the vacuolar proton pump are supplied by carbonic anhydrase II, which catalyzes the formation of carbonic acid from water and carbon dioxide. For each proton ejected by the pump, an equimolar amount of base in the form of bicarbonate (HCO₃⁻) is retained in the cytoplasm. A HCO₃⁻/Cl⁻ anion exchanger in the basolateral membrane prevents cytoplasmic alkalinization by providing egress for excess base. Accordingly, anion exchange inhibitors repress the bone resorbing activity of osteoclasts, and incubation of active osteoclasts in Cl⁻-free media results in an increase in cytoplasmic pH (4, 6–8). Despite these data, the genetic identity of the basolateral HCO₃⁻/Cl⁻ anion exchanger in osteoclasts, and its role in bone remodeling in vivo, has not been described. Here, we report that the HCO₃⁻/Cl⁻ anion exchanger SLC4A2 (also known as AE2) is up-regulated during osteoclastogenesis through an NFATc1-dependent pathway and is a critical mediator of osteoclast development and function in vivo and in vitro.

Results

Slc4a2 Is Up-Regulated during Osteoclastogenesis in an NFATc1-Dependent Manner.

NFATc1 is induced by RANKL and orchestrates the osteoclast differentiation program (3). We recently generated a conditional knockout allele of Nfatc1 (Nfatc1^fl/fl), and deletion in the postnatal period using Mx1-Cre (Nfatc1^Δ/Δ) resulted in severe osteopetrosis with a marked attenuation of osteoclast differentiation (9). Because NFATc1 is required for expression of many genes important for osteoclast function, we reasoned that novel osteoclast regulators exist within the universe of NFATc1-dependent transcripts in these cells. A comparative DNA microarray analysis of FACS-sorted, bone marrow osteoclast precursors (BmOcPs) incubated with MCSF and RANKL was performed (9). This analysis revealed that the bicarbonate/chloride anion exchanger, Slc4a2, was expressed 5-fold higher in NFATc1-sufficient cells (Fig. 1A). The use of alternative promoters leads to multiple Slc4a2 isoforms designated Slc4a2a, Slc4a2b1, Slc4a2b2, Slc4a2c1, and Slc4a2c2 (10, 11). Though the Slc4a2a and Slc4a2b isoforms are ubiquitously expressed, the Slc4a2c1 isoform is stomach restricted (10). In response to RANKL, the Slc4a2a isoform was up-regulated almost 10-fold in BmOcPs (Fig. 1B), whereas expression of Slc4a2b1, Slc4a2b2, Slc4a2c1, and Slc4a2c2 was not discernable (Fig. 1B). Consistent with our microarray results (Fig. 1A), Slc4a2a was not induced in NFATc1-deficient BMOcPs treated with RANKL, and cyclosporine A (CsA), an inhibitor of the NFATc1-activating phosphatase calcineurin, attenuated Slc4a2a expression (Fig. 1 B and C). Moreover, SLC4A2 protein was expressed by osteoclast precursors at the basolateral membrane after treatment with RANKL (Fig. 1 D and E). Taken together, these data show that SLC4A2 is induced during osteoclastogenesis through an NFATc1-dependent pathway, suggesting this molecule plays an important role in bone resorption.

Fig. 1.

NFATc1-dependent up-regulation of Slc4a2 during osteoclastogenesis. (A) Microarray signal intensities for Slc4a2 from two independent mRNA samples isolated from BMOcPs treated for 3 days with MCSF and RANKL. (B) qRT-PCR analysis for the expression of Slc4a2 isoforms in BMOcPs cultured with MCSF or MCSF and RANKL. ex14 refers to qRT-PCR primers that produce an amplicon within exon 14 of Slc4a2, an exon present in all five isoforms. (C) qRT-PCR analysis for the expression of Slc4a2a in BMOcPs treated with MCSF or MCSF and RANKL in the presence of increasing concentrations of CsA (62.5, 125, 250 ng/ml). (D) SLC4A2 immunostaining of WT MCSF-primed BM osteoclast precursors cultured with MCSF or MCSF and RANKL. Two examples of SLC4A2-positive osteoclasts are shown. Green, SLC4A2; blue, Hoechst. (E) Confocal image along the z axis of the osteoclast shown in the rightmost panel of (D).

Knockdown of Slc4a2 Expression in RAW Cells Prevents Osteoclast Differentiation and Matrix Resorption.

An RNAi approach was used to explore the role of SLC4A2 in osteoclast differentiation and function using a highly osteoclastogenic subclone of the mouse macrophage cell line RAW 264.7 (RAW clone 6) (12). Similar to our findings with primary cells, RANKL induced SLC4A2 protein expression in RAW clone 6 (Fig. 2A). Of five different shRNA vectors targeting Slc4a2, hairpins 1 and 2 (Slc4a2–1 and Slc4a2–2) yielded the greatest reduction in Slc4a2 mRNA levels (Fig. 2B). Accordingly, Slc4a2–1 and Slc4a2–2 prevented the ability of RAW clone 6 cells to differentiate into osteoclasts in response to RANKL, as assessed by the liberation of TRAP into the tissue culture supernatants (Fig. 2C), the formation of TRAP-positive multinucleated cells (Fig. 2D) and the ability to resorb a synthetic calcified matrix (Fig. 2E). Hairpins 3 and 5 (Slc4a2–3 and Slc4a2–5), which knocked down Slc4a2 mRNA to intermediate levels (Fig. 2B), yielded a partial block in differentiation (Fig. 2 C and D). Lastly, shRNA clone 4 (Slc4a2–4) only modestly reduced Slc4a2 mRNA levels (Fig. 2B) and did not inhibit osteoclast differentiation (Fig. 2 C and D). These experiments identify SLC4A2 as an important mediator of osteoclast differentiation in RAW clone 6 cells and prompted an analysis of Slc4a2^−/− mice.

Fig. 2.

Knockdown of Slc4a2 expression in RAW clone 6 cells prevents osteoclast differentiation and matrix resorption. (A) Western blot analysis for SLC4A2 in RAW clone 6 cells cultured with or without RANKL. (B) qRT-PCR analysis for the expression of Slc4a2a in RAW clone 6 cells infected with lentiviruses expressing five different shRNA constructs targeting Slc4a2 and cultured with RANKL. The control shown is the average of RAW clone 6 cells infected with three different irrelevant lentiviral shRNA constructs and cultured with RANKL. (C) Supernatant TRAP assay or (D) TRAP stain of RAW clone 6 cells infected with the indicated lentiviral shRNA vectors and treated with or without ([C] only) RANKL. (E) Matrix resorption assay of RAW clone 6 cells infected with the indicated lentiviral shRNAs and cultured with RANKL.

Slc4a2-Deficient Mice Develop Osteopetrosis.

Mice deficient in all Slc4a2 isoforms are emaciated and achlorhyrdic, and exhibit early lethality (13). At 3 weeks of age, compared with WT littermates, Slc4a2^−/− mice showed growth retardation and clubbing of the long bones, most pronounced at the distal femur and proximal humerus (Fig. 3 A and B). A failure to form a proper bone marrow (BM) cavity, leading to an absence of discernable cortical bone, was evident in the long bones (Fig. 3B). Histology confirmed the lack of a BM cavity and cortical bone in Slc4a2^−/− mice (Fig. 3 C and D). The matrix filling the marrow cavity of Slc4a2^−/− mice stained positive with toluidine blue, indicating that these mice fail to resorb the primarily cartilaginous skeletal anlagen (Fig. 3E). These skeletal findings are consistent with congenital osteopetrosis.

Fig. 3.

Slc4a2-deficient mice fail to remodel the primary skeletal anlagen. (A and B) Digital radiographs of (A) mice and (B) femurs. (C and D) Hematoxylin and eosin stain of (C) vertebral bodies and (D) tibias (4× objective). (E) Toluidine blue stain of the tibial metaphysis (10× objective). The radiographs and histology are representative of at least three littermate-matched, 3-week-old mice analyzed per genotype.

Abnormal Osteoclast Differentiation and Increased Osteoclast Apoptosis in Slc4a2^−/− Mice.

Osteopetrosis is caused by either impaired osteoclast differentiation leading to reduced osteoclast numbers or a reduction in osteoclast activity, usually associated with normal or increased osteoclast numbers (2). Consistent with our RNAi findings (Fig. 2 C and D), a lack of TRAP staining in the BM cavities of long bones was observed in Slc4a2^−/− mice (Fig. 4A). However, multinucleated giant cells could still be identified in the sparse BM spaces of Slc4a2^−/− mice, some of which displayed features of programmed cell death previously described to characterize apoptotic osteoclasts, including nuclear condensation and fragmentation, and cytoplasmic shrinkage (Fig. 4B) (14). Immunohistochemistry for CD68 (Fig. 4C) and cathepsin K (Fig. 4D) confirmed that the multinucleated giant cells found in Slc4a2^−/− mice were of the macrophage/osteoclast lineage. Moreover, these cells expressed nuclear NFATc1, a master regulator of osteoclast development (Fig. 4E). Slc4a2^−/− osteoclasts tended to be larger and have more nuclei than their WT counterparts (Fig. 4 C and D). Terminal deoxynucleotidyl nick end labeling (TUNEL) confirmed that the multinucleated cells with features of programmed cell death (Fig. 4B) in Slc4a2^−/− mice were indeed apoptotic (Fig. 4F). Moreover, although similar numbers of osteoclasts were observed in Slc4a2^−/− mice compared with WT littermates, almost four times as many Slc4a2^−/− osteoclasts displayed cytologic features of apoptosis (Fig. 4 G and H). Our findings suggest that the dominant role of SLC4A2 in osteoclasts in vivo is to facilitate bone resorption and suppress apoptosis.

Fig. 4.

Abnormal osteoclast development and increased osteoclast apoptosis in Slc4a2-deficient mice. (A) TRAP stain of the tibial metaphysis (40× objective). (B) H&E stain showing multinucleated giant cells without (Left) and with (Right) apoptotic features in Slc4a2^−/− mice. (C–E) Immunohistochemistry for (C) CD68, (D) cathpesin K, and (E) NFATc1. (F) TUNEL staining at the femoral metaphysis of Slc4a2^−/− mice. Note brown staining in the condensed nuclei of apoptotic Slc4a2^−/− osteoclasts (arrows) compared with a nonapoptotic one (arrowhead). (G) Number of cathepsin K-positive, multinucleated cells per microscopic field (40× objective) in the distal femoral metaphysis (P > 0.5, Slc4a2^+/+ vs. Slc4a2^−/−) (H) Percentage of cathepsin K-positive, multinucleated cells with cytologic features of apoptosis in the distal femoral metaphysis (P < 0.01, Slc4a2^+/+ vs. Slc4a2^−/−). The images in (B)–(F) were cropped from pictures obtained in the femoral metaphysis with 40× or 60× ([F] only) objectives. Images are representative of at least three littermate-matched, 3-week-old mice analyzed per genotype. Data in (G) and (H) are the average plus SD of three mice per genotype.

SLC4A2 Is Required for Bone Resorption and Vacuolar Acidification ex Vivo.

To further explore the function of SLC4A2 in osteoclasts, in vitro experiments were performed with primary osteoclast precursors grown from the BM or spleens of WT and Slc4a2^−/− mice with MCSF. As expected, after culture with RANKL, WT osteoclast precursors expressed SLC4A2, whereas Slc4a2^−/− cells did not (Fig. 5A). Though Slc4a2^−/− osteoclast precursors formed TRAP-positive, multinucleated cells in vitro, these cells did not spread normally, released less TRAP into the culture supernatants, and expressed lower levels of genes associated with terminal osteoclast differentiation (Fig. 5 B–D). Moreover, Slc4a2^−/− osteoclasts were completely unable to resorb dentin (Fig. 5E). Because SLC4A2 promotes gastric acid secretion (13, 15) and osteoclasts secrete protons to resorb bone, the ability of Slc4a2^−/− cells to form an acidic, extracellular compartment was tested. Whereas WT osteoclasts formed large, acridine orange-positive discs, indicating acidic extracellular resorption pits, Slc4a2^−/− cells did not (Fig. 5F). Taken together, these data show that SLC4A2 facilitates optimal osteoclast differentiation and is required for acid secretion.

Fig. 5.

SLC4A2 is required for bone resorption and extracellular acidification by osteoclasts in vitro. (A) Immunostaining for SLC4A2 on MCSF-primed BM cells cultured with MCSF and RANKL. (B) TRAP stain and (C) supernatant TRAP assay of spleen cells cultured with MCSF ([B] only), or MCSF and RANKL. (D) qRT-PCR analysis of spleen cells cultured with MCSF, or MCSF and RANKL. (E) Lectin-TRITC stain of dentin slices cultured with spleen cells and MCSF and RANKL. (F) Acridine orange stain of spleen cells cultured on dentin slices with MCSF and RANKL.

Discussion

Bone resorption begins with the differentiation of macrophage precursors into osteoclasts, which adhere to the bone surface and polarize to form an extracellular hemivacuole into which acid is secreted. The secretion of protons into the resorption lacuna generates an intracellular acid deficit. A basolateral HCO₃⁻/Cl⁻ anion exchanger regulates intracellular pH through the efflux of excess bicarbonate (4, 8). Until this study, the genetic identity of this exchanger was unknown. Here, we define Slc4a2 as an NFATc1-regulated transcript in osteoclasts, whose mutation blocks skeletal remodeling.

The SLC4 family of anion transporters share a common cytoplasmic N-terminal domain, followed by a polytopic transmembrane domain responsible for anion transport (16). The use of alternative, tissue-specific promoters leads to the transcription of variant 5′ mRNAs that encode truncated N-terminal isoforms. Osteoclasts up-regulate the Slc4a2a isoform, but not the Slc4a2b and Slc4a2c isoforms (Fig. 1B). Whether NFATc1 directly regulates Slc4a2a expression in osteoclasts is currently under investigation. Two different strains of Slc4a2 knockout mice have been described. The first (Slc4a2a,b^−/−) targeted the Slc4a2a,b isoforms (15, 17), leaving expression of Slc4a2c intact, a stomach-specific isoform. The second strain (Slc4a2^−/−), used here, deleted all five isoforms (13). Though both strains display perinatal lethality, Slc4a2^−/− mice are growth retarded and edentulous, whereas Slc4a2a,b^−/− mice appear grossly normal (13, 15, 17). Here, we show that Slc4a2^−/− mice fail to remodel the primary cartilaginous anlagen leading to the absence of cortical bone and a discernable BM cavity (Fig. 3 A–E). Although the skeletal phenotype of Slc4a2a,b^−/− has not been reported, the lack of growth retardation in this strain, a common feature of severe osteopetrosis, suggests that SLC4A2c may compensate for SLC4A2a in osteoclasts. Further studies are needed to resolve this important issue.

To explore the function of SLC4A2 in osteoclasts, an RNAi approach was used using RAW clone 6 cells (Fig. 2B). Two shRNAs against Slc4a2 blocked osteoclast differentiation (Fig. 2 C and D). In contrast, multinucleated giant cells expressing osteoclast lineage markers could be identified in the BM of Slc4a2^−/− mice (Fig. 4 B–E). Moreover, primary Slc4a2-deficient macrophages formed multinucleated giant cells in response to RANKL (Fig. 5B). However, similar to the knockdown cells, Slc4a2-deficient osteoclasts expressed less TRAP activity (Figs. 4A and 5C). In addition, these cells expressed lower levels of osteoclast differentiation markers (Fig. 5D). However, though the block in differentiation appears to be partial, the osteoclasts that form in Slc4a2^−/− cultures were completely defective in the resorption of dentin (Fig. 5E) and vacuole acidification (Fig. 5F). Our histologic data indirectly support a role for SLC4A2 in osteoclast development. In other models, where genes are mutated that purely affect osteoclast function, like Src^−/− and Itgb3^−/− mice, increased osteoclast numbers are observed (18, 19). In contrast, we found similar numbers of cathepsin K-positive multinucleated cells in Slc4a2^+/+ and Slc4a2^−/− mice (Fig. 4G). Certainly, increased numbers of osteoclasts in Slc4a2^−/− mice could be offset by increased apoptosis (Fig. 4H), in addition to a defect in differentiation.

It is possible that Slc4a2-deficient osteoclasts do not resorb mineralized matrix because they fail to express molecules necessary to efficiently resorb bone, in addition to an inability to form an acidified vacuole (Fig. 5F). For example, β3 integrin is necessary for attachment and spreading of osteoclasts (18), and this molecule is not optimally expressed in Slc4a2-deficient cells (Fig. 5D). Consistent with this, Slc4a2-deficient osteoclasts do not spread normally (Fig. 5B). How an anion exchanger could affect differentiation and gene expression is not apparent. Certainly, perturbations to cytoplasmic pH, generated even before activation of the proton pump, may affect signaling pathways. Furthermore, SLC4A2 may have functions in the osteoclast independent of its anion exchange activity. For example, in red blood cells, the N-terminal domain of SLC4A1, a homolog of SLC4A2, anchors the membrane to the cytoskeleton (16). Mutations in SLC4A1 cause hereditary spherocytosis, a genetic disease characterized by anemia due to erythrocyte fragility. Clearly, further experimentation is needed to resolve the independent contributions of SLC4A2 to osteoclast differentiation and function.

It is currently unclear why Slc4a2^−/− osteoclasts do not display histochemical TRAP activity in vivo, despite the expression of other osteoclast markers (Fig. 4 A and C–E). Osteoclast TRAP is translated as a monomer with low phosphatase activity. Proteolytic cleavage excises an internal, repressive peptide resulting in a more active heterodimer (20). The lack of histochemical TRAP activity in Slc4a2^−/− mice may be related to inefficient processing of the TRAP monomer due to improper vacuole acidification. Reports showing that cathepsins process monomeric TRAP (20), and that cathepsins are more stable and active at low pH (21), support this hypothesis. However, osteoclasts from Atp6i^−/− and Clcn7^−/− mice, which also display acidification defects, express TRAP activity (22, 23). Given the importance of TRAP to osteoclast function (24), further investigation is needed to explain this observation.

The most intriguing finding in this study was the dramatic increase in osteoclast apoptosis observed in Slc4a2^−/− mice (Fig. 4 F and H). Compared with the large body of knowledge about osteoclast differentiation, relatively little is known about osteoclast apoptosis in vivo. The reversal phase of bone resorption is associated with cell death of osteoclasts, most dramatically observed at the termination of lactation (25). Two drugs, estrogen and bisphosphonates, currently used to treat osteoporosis, induce osteoclast apoptosis (26, 27). Although the mechanism by which Slc4a2 deficiency results in osteoclast apoptosis is unresolved, it may be a direct consequence of cytoplasmic alkalinization in the absence of HCO₃⁻/Cl⁻ transport activity. In multiple systems a rise in intracellular pH triggers apoptosis (28). Alternatively, increased cell death may not be intrinsic to Slc4a2^−/− osteoclasts, but rather a consequence of proapoptotic signals generated in these mice. For example, sex hormones, glucocorticoids, and TGFβ all promote osteoclast apoptosis (29) and could be deregulated in Slc4a2^−/− mice. Slc4a2^−/− osteoclasts may also die more readily because of decreased matrix or stromal cell-derived prosurvival signals.

Taken together, our data suggest therapeutics targeting SLC4A2 may block bone degradation by at least three mechanisms: inhibiting acidification of the resorption lacuna, reducing TRAP activity, and initiating osteoclast apoptosis. The induction of cell death is a particularly attractive mechanism of action because intermittent inhibition of SLC4A2 could have long-lasting effects on osteoclast numbers and thus obviate the need for daily therapy.

Materials and Methods

Mice.

The generation of the Nfatc1 conditional knockout strain, and the use of Mx1-cre to delete Nfatc1 in osteoclast precursors, is described elsewhere (9). Slc4a2^−/− mice were obtained from G.E. Shull (University of Cincinnati, Cincinnati, OH) (13). Experimental protocols were approved by the Standing Committee on Animals at the Harvard Medical School and were designed with institutional and National Institutes of Health guidelines for the humane use of animals.

Cell Culture.

All cells were maintained in Minimum Essential Medium-Alpha (Cellgro) containing 10% FBS and 1% penicillin/streptomycin. RANKL was a gift from Y. Choi (University of Pennsylvania, Philadelphia). Osteoclast differentiation in RAW clone 6 cells (12), a gift of K.P. McHugh (Harvard Medical School, Boston), was induced by adding 200 ng/ml RANKL for 4 days. The isolation of CD11b^low/−CD3ε⁻B220⁻c-kit⁺c-fms⁺ BMOcPs, used for the experiments in Fig. 1 A–C, is described elsewhere (9, 30). For immunofluoresence studies, MCSF-primed BM cells were used as osteoclast precursors (18). For mouse spleen cell cultures, a single cell suspension of splenocytes was cultured for 3 days with 30 ng/ml MCSF (R&D Systems) and differentiation was induced by adding 500 ng/ml RANKL for 5 days. Biochemical assays for TRAP were performed as described (31, 32) or with a commercial kit (Sigma).

RNA Extraction and Quantitative Real-Time PCR.

Total RNA was extracted using an RNeasy kit (Qiagen). cDNA was synthesized using AffinityScript cDNA Synthesis Kit (Stratagene). qRT-PCR reactions were performed using the primers listed in supporting information (SI) Table S1 and the Brilliant II SYBR Green QPCR Master Mix (Stratagene) in an Mx3005P QPCR System (Stratagene). Relative amounts of mRNA were calculated by the ΔΔCt method using Hprt1 as an internal control.

SLC4A2 Western Blot and Immunofluorescence.

An affinity purified rabbit antibody to mouse SLC4A2 was provided by S.L. Alper (Harvard Medical School, Boston) (33). For immunofluorescence, osteoclast precursors were cultured with MCSF and RANKL for 5 days, fixed, permeabilized, and stained with anti-SLC4A2 and goat anti-rabbit Alexa488 (Invitrogen). Hoechst (Sigma) was used as a nuclear counterstain.

shRNA-Mediated Knockdown of Slc4a2 mRNA.

The RNAi Consortium at the Broad Institute (34) provided lentivirus-based shRNA constructs targeting Slc4a2 (Table S2), or control sequences, which were used infect RAW clone 6 cells. Assays were performed on a stable pool of transduced cells.

Histology and Immunohistochemistry.

Tissues were fixed in 4% paraformaldehyde, decalcified, and embedded in paraffin. TRAP staining was performed as described (31). The following monoclonal antibodies were used for immunohistochemistry: anti-CD68 (FA-11; Abcam), anti-cathepsin K (182–12G5; Calbiochem), and anti-NFATc1 (7A6; PharMingen). The appropriate HRP conjugated secondary antibodies and 3,3′-diaminobenzidine revealed binding of primary antibodies to their respective antigens. TUNEL was performed using the In Situ Cell Death Detection Kit (Roche).

Quantification of Osteoclasts and Apoptosis in Vivo.

To quantify osteoclasts and apoptotic osteoclasts, three nonoverlapping microscopic fields, obtained with a 40× objective, were analyzed per mouse from femur sections stained for cathepsin K. Osteoclasts were defined as cathepsin K-positive cells with two or more nuclei. Cells with chromatin condensation or nuclear fragmentation were considered apoptotic (29).

Osteologic Matrix Resorption and Dentin Resorption Assay.

RAW clone 6 cells were plated on Osteologic slices (BD Biosciences), treated with RANKL for 6 days, and analyzed following manufacturer's instructions. For dentin assays, spleen cells were plated on dentin slices (provided by P. Hauschka, Harvard Medical School, Boston) and differentiated with MCSF and RANKL for 6 days. Resorption pits were revealed using wheat germ lectin-TRITC (Sigma) after sonication in 0.5 M NH₄OH.

Evaluation of Extracellular Acidification Using Acridine Orange.

Acridine orange staining was performed as described (35). Briefly, spleen cells were plated on dentin slices, differentiated with MCSF and RANKL for 6 days, and incubated with 5 μg/ml acridine orange (Sigma) for 15 min. The cells were observed under a fluorescence microscope with a 490 nm excitation filter and a 525 nm arrest filter.

Statistical Analysis.

Upaired, two-tailed Student's t tests were used for statistical comparisons. A P value of <0.05 was considered significant.