A TRPC1 Protein-dependent Pathway Regulates Osteoclast Formation and Function*

Background: Ca2+ signaling is essential for osteoclastogenesis. Results: I-mfa negatively regulates TRPC1-mediated Ca2+ signaling and osteoclastogenesis. Conclusion: TRPC1 and I-mfa fine-tune the dynamic range of store-operated Ca2+ entry channels during osteoclastogenesis. Significance: The TRPC1/I-mfa interaction is biologically relevant in osteoclastogenesis. Ca2+ signaling is essential for bone homeostasis and skeletal development. Here, we show that the transient receptor potential canonical 1 (TRPC1) channel and the inhibitor of MyoD family, I-mfa, function antagonistically in the regulation of osteoclastogenesis. I-mfa null mice have an osteopenic phenotype characterized by increased osteoclast numbers and surface, which are normalized in mice lacking both Trpc1 and I-mfa. In vitro differentiation of pre-osteoclasts derived from I-mfa-deficient mice leads to an increased number of mature osteoclasts and higher bone resorption per osteoclast. These parameters return to normal levels in osteoclasts derived from double mutant mice. Consistently, whole cell currents activated in response to the depletion of intracellular Ca2+ stores are larger in pre-osteoclasts derived from I-mfa knock-out mice compared with currents in wild type mice and normalized in cells derived from double mutant mice, suggesting a cell-autonomous effect of I-mfa on TRPC1 in these cells. A new splice variant of TRPC1 (TRPC1ϵ) was identified in early pre-osteoclasts. Heterologous expression of TRPC1ϵ in HEK293 cells revealed that it is unique among all known TRPC1 isoforms in its ability to amplify the activity of the Ca2+ release-activated Ca2+ (CRAC) channel, mediating store-operated currents. TRPC1ϵ physically interacts with Orai1, the pore-forming subunit of the CRAC channel, and I-mfa is recruited to the TRPC1ϵ-Orai1 complex through TRPC1ϵ suppressing CRAC channel activity. We propose that the positive and negative modulation of the CRAC channel by TRPC1ϵ and I-mfa, respectively, fine-tunes the dynamic range of the CRAC channel regulating osteoclastogenesis.

Ca 2؉ signaling is essential for bone homeostasis and skeletal development. Here, we show that the transient receptor potential canonical 1 (TRPC1) channel and the inhibitor of MyoD family, I-mfa, function antagonistically in the regulation of osteoclastogenesis. I-mfa null mice have an osteopenic phenotype characterized by increased osteoclast numbers and surface, which are normalized in mice lacking both Trpc1 and I-mfa. In vitro differentiation of pre-osteoclasts derived from I-mfa-deficient mice leads to an increased number of mature osteoclasts and higher bone resorption per osteoclast. These parameters return to normal levels in osteoclasts derived from double mutant mice. Consistently, whole cell currents activated in response to the depletion of intracellular Ca 2؉ stores are larger in pre-osteoclasts derived from I-mfa knock-out mice compared with currents in wild type mice and normalized in cells derived from double mutant mice, suggesting a cell-autonomous effect of I-mfa on TRPC1 in these cells. A new splice variant of TRPC1 (TRPC1⑀) was identified in early pre-osteoclasts. Heterologous expression of TRPC1⑀ in HEK293 cells revealed that it is unique among all known TRPC1 isoforms in its ability to amplify the activity of the Ca 2؉ release-activated Ca 2؉ (CRAC) channel, mediating store-operated currents. TRPC1⑀ physically interacts with Orai1, the pore-forming subunit of the CRAC channel, and I-mfa is recruited to the TRPC1⑀-Orai1 complex through TRPC1⑀ suppressing CRAC channel activity. We propose that the positive and negative modulation of the CRAC channel by TRPC1⑀ and I-mfa, respectively, fine-tunes the dynamic range of the CRAC channel regulating osteoclastogenesis.
Mature osteoclasts are derived from hematopoietic stem cells through a series of events initiated by the formation of myeloid precursors in response to macrophage-colony stimulating factor (M-CSF) 5 (1). Subsequently, these precursors differentiate into multinucleated osteoclasts in a multistep process dependent on M-CSF and receptor activator of nuclear factor-B ligand (RANKL) (2). Both of these factors act through Ca 2ϩ signaling to induce downstream regulators of osteoclastogenesis such as nuclear factor of activated T cells c1 (NFATc1), NF-B, c-fos, ␤-catenin, and others (3,4). However, the molecular identity of the Ca 2ϩ channels essential for osteoclastogenesis is only recently starting to emerge.
The molecular makeup and mode of activation of the channels mediating I SOC are less clear, but TRPC1 has been shown to produce I SOC in association with STIM1 and Orai1 (30 -38). However, TRPC1 alone or co-expressed with STIM1 and Orai1 has never resulted in the de novo generation or amplification of I CRAC , respectively. These results raise the question of whether the SOC and CRAC channel is the same channel modified by the presence or absence of TRPC1. Therefore, the exact role of TRPC1 in the regulation of store-operated Ca 2ϩ entry pathways has been unclear.
I-mfa is a cytosolic protein with a unique cysteine-rich domain, first identified as an interacting protein interacting with MyoD (39) and subsequently with components of the Wnt/␤-catenin pathway (40 -43). We have identified the inhibitor of MyoD family isoform "a" (I-mfa) as a binding partner for TRPC1 (44). Using an array of biochemical assays, we showed that TRPC1 associated directly with I-mfa in transfected cells, native tissues, and cell lines. Functional experiments in transfected and native A431 cells revealed that I-mfa suppressed I SOC through TRPC1. These gain-and loss-of-function experiments in combination with co-immunoprecipitation experiments in native tissues provided evidence for a physiological role of I-mfa in the regulation of endogenous TRPC1 activity. However, the biological role of the I-mfa-mediated inhibition of TRPC1 remained unknown. In this study, we identify a role of the TRPC1/I-mfa interaction in the regulation of osteoclastogenesis in vivo and in vitro through the modulation of the store-operated Ca 2ϩ entry channels.
Cell Culture-HEK293 cells were purchased from American Type Culture Collection (ATCC) and maintained in Dulbecco's modified Eagle's medium (DMEM, Mediatech) supplemented with 10% fetal bovine serum (FBS).
Expression of I-mfa and TRPC1␣/⑀ Isoforms in Preosteoclasts-Nonadherent bone marrow-derived cells from wild type mice were grown in ␣-minimal essential medium supplemented with 10% ES-FBS (Atlanta Biologicals), 1ϫ penicillin/ streptomycin/glutamine solution (Invitrogen), and in the presence of 10% CMG-conditioned media (containing M-CSF). For experiments without M-CSF, 10% CMG was omitted from culture media. After 2 days, cells in suspension were collected, and total RNA was isolated using TRIzol (Invitrogen). Five g of RNA was reverse-transcribed using SuperScript III (Invitrogen) and an equal mix of oligo(dT) and random hexamers (Roche Applied Science) as primers. I-mfa mRNA was detected using the following PCR primers: Fwd 5Ј-AGC CAC GAC CAC CTC TCA GAA CCG-3Ј and Rev 5Ј-CGC AGT CCA GGA GGA TGT TAC AGA-3Ј. TRPC1 product was amplified using a primer set spanning the exon 4 -5 junction, Fwd 5Ј-GTT GTC AGT CCG CAG ATG CAC TTT-3Ј and Rev 5Ј-TGT CCA AAC CAA ACC GTG TTC AGG-3Ј (694 bp). PCR conditions were as follows: initial denaturation for 2 min at 95°C, 35 cycles of 30 s at 95°C, 30 s at 57°C, 45 s at 72°C, and a final extension at 72°C for 7 min using Platinum Taq polymerase (Invitrogen). The product of this PCR was used as a template for a nested PCR using the internal primer set: Fwd 5Ј-ACG ATC ATC AAG ACC AAC CAT TG-3Ј and Rev 5Ј-AGT CCT CGT TTG TCA AGA GGC TCA-3Ј (495 bp). PCR conditions were as follows: initial denaturation for 4 min at 96°C, 30 cycles of 1 min at 96°C, 1 min at 55°C, 30s at 72°C, and a final extension at 72°C for 5 min using Vent polymerase (New England Biolabs) in a 200-l reaction volume. PCR products were phenol/chloroform-extracted and ethanol-precipitated. An equal amount of purified PCR products were digested with EcoRV (New England Biolabs) or left untreated, separated on a 2% agarose gel, and photographed. Digested PCR products were subjected to one more round of amplification, EcoRV digestion, and separation on an agarose gel to ensure complete digestion of the TRPC1␣ isoform. Final EcoRV-resistant 500-bp band was excised, purified (Qiagen), and sequenced.
Real Time Quantitative PCR-Using gene-specific primers, quantitative real time PCR was performed with RT2 Fast SYBR Green quantitative PCR master mix (SABioscience, Valencia, CA) and the CFX96 detection system (Bio-Rad). PCR conditions were as follows: initial denaturation at 95°C for 10 min, 40 cycles of 15 s at 95°C, and 1-min extension at 61°C for GAPDH and I-mfa, 55°C for TRPC1␣. Primer sets for I-mfa and GAPDH were same as above for RT-PCR. The TRPC1␣-specific primer set used for real time quantitative PCR was Fwd 5Ј-GGT TTC GTC TTG ATA TCT ATA G-3Ј and Rev 5Ј-TCG TTT GTC AAG AGG CTC ATC-3Ј.
Micro-computed Tomography (CT) Analysis-Twelveweek-old male mice were euthanized, and soft tissues were removed. After fixation in 70% ethanol, proximal tibiae were scanned by using the Scanco vivaCT 40 CT scanner (Scanco Medical, Bassersdorf, Switzerland) with a resolution size of 10 m. Three-dimensional reconstruction and quantification of structural parameters were calculated using the manufacturer's software. Scanning of the trabecular bone in the tibia was initiated proximal to the growth plate, and a total of 120 consecutive 10-m-thick sections were analyzed. Cortical bone was excluded from the analysis, and the segmentation values were set at 0.8/1/220 for all studies.
Bone Histology and Histomorphometry-Tibiae were cut in half, and the larger distal pieces were fixed in 4% paraformaldehyde for 24 h at room temperature and stored in 70% ethanol until sectioning. Longitudinal sections (5 m thick) were cut at the 50% plane from methyl methacrylate-embedded blocks using a Leica 2265 microtome. Sections were stained with Goldner's Trichrome. For histomorphometry, a region of interest was selected that was exactly 250 m distal to the growth plate and extended 1 mm downward (thereby avoiding the primary spongiosa) through the metaphysis of the tibia. Standard bone histomorphometry was performed by the methods of Parfitt et al. (47) using Bioquant Image Analysis software (R & M Biometrics, Nashville, TN). Four types of primary measurements were made: area, length (perimeter), distance, and number. Tissue volume, bone volume, bone surface, and osteoid surface were used to derive trabecular number and trabecular separation. Blind measurements were performed in all samples.
Ex Vivo Osteoclast Differentiation-Three 8 -12-week-old animals were used per experiment. Femurs, tibiae, and humeri were isolated, and soft tissue was removed. The bone marrow cavity was flushed with phosphate-buffered saline (PBS), and cells were grown in ␣-minimal essential medium supplemented with 10% embryonic stem cell-qualified (ES)-FBS (Atlanta Biologicals), 10% conditioned media from granulosa cells (CMG) (containing M-CSF), and 1ϫ penicillin/streptomycin/glutamine solution (Invitrogen). After 2 days, cells in suspension were seeded at 50,000 cells/well on a hydroxyapatite substrate (Corning Glass) or at 50,000 -200,000 cells/well on a 96-well plate, depending on the assay, and differentiated osteoclasts in medium were supplemented with 20 ng/ml recombinant mouse M-CSF and 50 ng/ml recombinant mouse RANKL (Shenandoah Biotechnology) for a defined period. To view resorption pits, osteoclasts were removed with 10% bleach, and the most representative areas of pits left by the osteoclasts were photographed and quantified using Metamorph (Molecular Devices) software. Pit area per osteoclast was determined only from nonoverlapping pits (100 pits/animal strain/experiment) using 50,000 cells plated per well onto osteologic plates (Corning Glass). Osteoclast resorption was confirmed by plating 50,000 pre-osteoclasts on dentin (Immunodiagnostic Systems Ltd.) for 10 days in the presence of 20 ng/ml M-CSF and 50 ng/ml RANKL. Cells were removed with a cotton swab and pits stained with Mayers hematoxylin (Sigma). Osteoclast multinucleation was determined by tartrate-resistant acid phosphatase staining of fixed cells. Fixed cells also were permeabilized with 0.1% Triton X-100 for 5 min, blocked with 1% BSA for 20 min at room temperature, and stained with phalloidin-Texas red (1:300, Molecular Probes) for 30 min at room temperature to visualize actin rings.
Transient Transfections-HEK293 cells were transfected in 35-mm dishes using Lipofectamine 2000 (Invitrogen) with the following plasmids: 1 g of Orai1, 1.6 g of STIM1, 1 g of TRPC1, 0.3 g of I-mfa or I-mfb, and 0.1 g of CD8␣. Cells were allowed to recover for 24 h, and then CD8␣ ϩ cells were identified by binding to magnetic beads coated with ␣-CD8␣ (Dynabeads, DYNAL) and processed for electrophysiology.
Electrophysiology-Whole cell patch clamp experiments were performed in voltage clamp tight-seal configuration at room temperature. Recordings were acquired using the Warner PC-505B amplifier (Warner Instrument Corp., Hamden, CT) and pClamp9.2 software (Axon Instrument, Foster City, CA). Pipettes were pulled from borosilicate glass capillaries (Warner Instruments, Corp.) and polished to a final resistance of 2-4 megohms. Voltage ramps of 100 ms duration spanning a range of Ϫ100 to ϩ100 mV were delivered from a holding potential of 0 mV at a rate of 0.5 Hz with an inter-ramp interval of 10 s. Currents were filtered at 2 kHz and digitized at 100-s intervals. Capacitive currents were determined and corrected before each voltage ramp. Traces recorded before I CRAC or I SOC current activation were used as templates for leak subtraction. Standard external solution (bath) was as follows (in mM): NaCl 120, KCl 2.8, CsCl 10, MgCl 2 2, CaCl 2 10, HEPES 10, and glucose 10 at pH 7.2 with 300 mOsm NaOH. In HEK293 cells, 10 mM tetraethylammonium was added to suppress delayed rectifier-mediated K ϩ currents (13). The standard internal solution (pipette) was as follow (in mM): cesium-methanesulfonate 120, NaCl 8, BAPTA 10, MgCl 2 3, HEPES 10 at pH 7.2 with 300 mOsm CsOH. Extracellular Na ϩ was replaced with an equimolar concentration of N-methyl-D-glucamine (Fig. 5F). Divalent replacement solution was based on the standard external solution, but 10 mM CaCl 2 was replaced by 10 mM BaCl 2 . Whole cell currents in myeloid precursors were measured as described above in transfected cells, except for the concentration of MgCl 2 in the pipette solution, which was raised from 3 to 8 mM, to block possible contamination from endogenous TRPM7 currents.

Trpc1 and I-mfa Function Antagonistically in the Regulation
of Osteoclastogenesis-Despite mild skeletal patterning defects manifested as rib fusions and bifurcations and a mild form of spina bifida, I-mfa Ϫ/Ϫ mice are fertile and live to adulthood (45). Trpc1 Ϫ/Ϫ mice also are fertile and live to adulthood (46). To test for a genetic interaction between the Trpc1 and I-mfa genes, we generated compound Trpc1 Ϫ/Ϫ ;I-mfa Ϫ/Ϫ mice and analyzed long bone histology and structure by histomorphometry and CT (Figs. 1 and 2).
Histomorphometry revealed that I-mfa Ϫ/Ϫ mice had a significant reduction (44%) in bone mass accompanied by an increase in the number of osteoclasts per bone surface and ero-TRPC1 and I-mfa Regulate Osteoclastogenesis AUGUST 2, 2013 • VOLUME 288 • NUMBER 31 sion (or osteoclast) surface per bone surface, by 72 and 38%, respectively, compared with wild type controls, indicating an osteopenic phenotype (Fig. 1, A-D). In contrast to I-mfa Ϫ/Ϫ mice, single Trpc1 Ϫ/Ϫ mice showed a substantial, but not significant, increase of 14% in bone mass and decreases of 27 and 28% in osteoclast numbers and eroded surface, respectively ( Fig. 1, A-D). The higher osteoclast numbers and larger erosion surface per bone surface in I-mfa null mice were normalized in the double knock-out mice (Fig. 1, A-D), suggesting that the increased osteoclastogenesis seen in I-mfa mutant mice was related to increased activity of TRPC1. CT analysis confirmed the histomorphometry results. I-mfa Ϫ/Ϫ mice had severely reduced bone mass (42% reduction, Fig. 1, A and F), trabecular thickness, numbers, and connectivity density and increased trabecular spacing (Fig. 2, B-E). In contrast, Trpc1-deficient mice had significantly increased bone mass (13%, Fig. 2A) and connectivity density (Fig. 2E), but all other parameters were similar to control mice ( Fig. 2, B-D). Double mutant mice had an intermediate phenotype with ϳ25% rescue of the I-mfa Ϫ/Ϫ -reduced bone phenotype in regard to bone mass and trabecular thickness (Fig. 2, A and C). All other parameters in these mice remained similar to I-mfa null mice (Fig. 2

, B, D, and E).
To determine whether the changes in bone mass were due in part to abnormalities in osteoblast numbers or function, we performed dynamic bone labeling prior to histomorphometry. Numbers and surface of osteoblasts and dynamic bone formation determined by calcein labeling were not different between wild type, I-mfa-knock-out mice, and DKO mice (Fig. 3) indicating a specific effect of I-mfa on osteoclasts and the lack of a significant genetic interaction of I-mfa and Trpc1 in osteoblasts. Interestingly, the numbers of osteoblasts (Fig. 3A), but not mineral apposition rate or bone formation rate (Fig. 3, D and E), were reduced in Trpc1-null mice suggesting that osteoblasts lacking Trpc1 may have increased function to compensate for the reduced numbers. In sum, both histomorphometric and CT studies showed that deletion of I-mfa caused an osteopenic phenotype that was partially rescued by the deletion of both genes. At the cellular level, I-mfa null mice had increased numbers of osteoclasts, which were completely restored by the additional deletion of Trpc1, suggesting a dominant effect of TRPC1 over I-mfa in osteoclastogenesis in vivo.
To determine whether the effect of I-mfa-mediated inhibition of TRPC1 could affect osteoclast function in a cell-autonomous fashion, we performed ex vivo experiments in which bone marrow pre-osteoclasts were differentiated into mature, multinucleated osteoclasts in the presence of M-CSF and RANKL (Fig. 4A). Osteoclast function was determined by the size of resorption pits formed by individual osteoclasts plated onto hydroxyapatite-coated plates (Fig. 4, B and C) or dentin discs (Fig. 4G). Deletion of Trpc1 did not significantly affect osteoclast formation (Fig. 4A) or resorption (Fig. 4, B and C), consistent with the idea that TRPC1 is blocked by I-mfa in wild type cells. However, deletion of I-mfa increased osteoclast numbers and resorption (Fig. 4, A-C). Inactivation of both Trpc1 and I-mfa normalized osteoclast numbers and resorption (Fig. 4, A-C), suggesting that the effect of I-mfa on osteoclast formation and function was primarily mediated through the inhibition of TRPC1. Deletion of I-mfa or both genes did not have a specific effect on small, medium, or large osteoclasts ex vivo or actin ring formation (Fig. 4, D-F). Overall, genetic experiments showed that disruption of I-mfa enhanced osteoclastogenesis in vivo and function in vitro and that both of these effects were suppressed by the additional disruption of Trpc1.
Expression of Trpc1 and I-mfa mRNAs in Early Osteoclast Progenitors and Identification of a New TRPC1 Isoform (TRPC1⑀)-To begin investigating whether TRPC1 and I-mfa mediate their effects on osteoclastogenesis at an early essential step in this process, we examined expression of I-mfa and Trpc1 mRNAs in hematopoietic progenitors (no M-CSF), myeloid precursors (ϩM-CSF, no RANKL), and early pre-osteoclasts (ϩM-CSF, ϩRANKL) (Fig. 5, A-D). I-mfa mRNA was induced by more than 20-fold during an early stage of differentiation in the presence M-CSF and then down-regulated in the later stage in the presence of M-CSF and RANKL. This regulation was consistent with its role as an inhibitor of Ca 2ϩ signaling at an early stage in osteoclast differentiation (Fig. 5, A and B). Alternative splicing of the Trpc1 gene results in several isoforms (␣-␦) (48), with TRPC1␣ being the longest known isoform. RT-PCR showed expression of TRPC1␣ in hematopoietic progenitors and myeloid precursors that persisted in pre-osteoclasts   Cells were classified as small (Ͻ50,000 pixels), medium (50,001-100,000 pixels), or large (Ͼ100,000 pixels) based on pixels per cell. Data represent mean Ϯ S.E. G, in vitro derived osteoclasts effectively resorb dentin. 50,000 bone marrowderived osteoclast precursors were plated on dentin discs in the presence of 50 ng/ml RANKL and 20 ng/ml MCSF for 10 days. Media were refreshed every 3 days. Cells were removed with a cotton swab and discs stained with Mayers hematoxylin to reveal the resorption pits (ϫ20 magnification). EcoRV-resistant PCR fragment at day 2 was gel-purified and directly sequenced. Asterisks indicate nonspecific products. E, nucleotide and corresponding amino acid sequence of the junction between exons 4 and 5 of mouse TRPC1␣ and TRPC1⑀ isoforms. Deleted sequence in TRPC1⑀ isoform is boxed. Unique EcoRV site in TRPC1␣ is shown in red. (Fig. 5C, lanes 2, 4, 6, 8, and 10). Real time quantitative PCR however using 40 cycles, revealed down-regulation of TRPC1␣ in response to M-CSF by ϳ4 -5-fold. However, due to very low expression level of TRPC1 in these cells and/or its expression in a subpopulation of these cells, quantitative changes in TRPC1 mRNA should be interpreted with caution. Interestingly, a new isoform (TRPC1⑀) was induced in myeloid precursors in response to M-CSF and its expression persisted for 4 days following RANKL stimulation (Fig. 5C, EcoRV-resistant fragment  in lanes 2, 4, 6, 8, and 10). TRPC1⑀ differed from TRPC1␣ by the deletion of 21 nucleotides containing a unique EcoRV site that allowed us to efficiently distinguish the mRNAs of the two isoforms. The 21-nucleotide deletion, which was caused by an alternative use of a splice acceptor site residing 21 nucleotides downstream of the normal acceptor site at the beginning of exon 5 (Fig. 5E), resulted in an in-frame deletion of seven amino acids within the N-terminal cytosolic region (Fig. 5E).

Store-operated Whole Cell Currents in Myeloid Precursors-
The expression of I-mfa and TRPC1 in myeloid precursors prompted us to examine store-operated currents in these cells. Cells derived from Trpc1 ϩ/ϩ ;I-mfa ϩ/ϩ (WT) mice showed typical I CRAC induced by 10 mM BAPTA in the pipette solution (Fig. 6, A and B, black). Deletion of Trpc1 (C1 Ϫ/Ϫ cells) did not affect I CRAC (Fig. 6, A and B, blue), consistent with the idea that TRPC1 was blocked by I-mfa in these cells. However, deletion of I-mfa (I Ϫ/Ϫ cells) resulted in almost linear store-operated current(s) displaying a larger inward component than that of I CRAC but also an outward component (Fig. 6, A and B, red). Currents in these cells were the largest among all groups (Fig. 6,  C and D). Deletion of both genes (DKO cells) restored I CRAC that was indistinguishable from I CRAC seen in wild type or Trpc1 null cells (Fig. 6, A and B, green), indicating that Trpc1 was responsible for the formation of large linear currents(s). These data showed that endogenous TRPC1 augments storeoperated currents and I-mfa suppresses the contribution of TRPC1 to these currents. Moreover, the effects of these two proteins on store-operated currents correlated well with their ability to modulate osteoclast formation and function, suggesting that modulation of store-operated currents by TRPC1 and I-mfa may underlie the mechanism of action of TRPC1 and I-mfa on osteoclastogenesis.
Translation of Trpc1 mRNA Is Initiated from a Non-AUG Codon-To obtain a mechanistic insight of how TRPC1 modulated these currents, we proceeded with a heterologous system whereby the two TRPC1 isoforms were functionally evaluated in HEK293 cells. However, we noticed that the 5Ј-untranslated region of mouse or human TRPC1␣ or TRPC1⑀ mRNA in exon 1 was extended far beyond the first methionine without an upstream in-frame STOP codon ( Fig. 7A and supplemental Fig. S1). In fact, five putative non-AUG translation initiation sites, as predicted by Ivanov et al. (49), were identified upstream of the first methionine in 13 mammalian TRPC1 species (Fig. 7A and supplemental Fig. S1). To identify the most upstream functional non-AUG translational start site in TRPC1, we deleted or mutated sites 1-3 and tested for their effects on TRPC1 mobility in SDS-PAGE. Deletion of site 1 did not cause a significant change in TRPC1 size (Fig. 7, B and C, lane 3), suggesting that site 1 either was not utilized or it was utilized, but upon its deletion, translation was initiated at a nearby non-AUG site, possibly site 2. When site 2 was deleted along with site 1 (Fig. 7C, lanes 4 and 5) or singly mutated (Fig. 7C, lane 6), translation was initiated from a downstream site causing a reduction in TRPC1 size. This analysis suggested that site 2 functions as the most upstream non-AUG translational start site of mouse TRPC1 expressed in HEK293 cells (Fig. 7, B and C). To test whether translation of endogenous TRPC1 also is initiated upstream of the predicted AUG site, endogenous TRPC1 was immunoprecipitated from HEK293T cell lysates and detected with a monoclonal TRPC1-specific antibody (1F1) (Fig. 7D, lane 5). A TRPC1-specific band with a molecular size similar to TRPC1␣/⑀ was detected (Fig. 7D, lane 5), arguing that endogenous TRPC1 contains a species with the N-terminal extension, as seen with transfected TRPC1 (Fig. 7D, lane 2). Overall, these data identify a new splice variant of TRPC1 induced in an early osteoclast precursor population by M-CSF and reveal that translation of transfected mouse or endogenous human TRPC1 is initiated at a CUG codon resulting in an N-terminal extension by 78 amino acids.

TRPC1 and I-mfa Regulate Osteoclastogenesis
Remarkably, co-expression of TRPC1⑀ with STIM1 and Orai1 did not produce I SOC but instead amplified Orai1-mediated I CRAC (Fig. 8D), as judged by activation by passive store depletion induced by BAPTA and formation of an inwardly rectifying current. Therefore, the 7-amino acid deletion generated by alternative splicing in TRPC1⑀ dramatically changed the permeability properties of TRPC1⑀ and/or Orai1/TRPC1⑀ complexes by completely eliminating the outward component in Orai1/TRPC1␣-transfected cells, while enhancing inward currents in Orai1/TRPC1⑀-transfected cells compared with Orai1-transfected cells.
Deletion of the N-terminal extension from TRPC1⑀ (TRPC1⑀-⌬NTx) resulted in linear currents but with an E rev closer to the E rev of I SOC in STIM1 ϩ Orai1 ϩ TRPC1␣-transfected cells and some inward rectification (Fig. 8C). These data indicated that both the N-terminal extension and the 7-amino acid deletion were required for the amplification of Orai1-mediated currents by TRPC1⑀, demonstrating for the first time that a transient receptor potential channel can amplify I CRAC or a current closely resembling I CRAC . Changes in STIM1 or Orai1 expression levels in cells transfected with or without TRPC1⑀ could not account for such an effect (Fig. 8E). Interestingly, overexpression of STIM1 and Orai1 in transiently transfected cells induced the expression of an endogenous TRPC1 species with a molecular size similar to TRPC1 constructs containing the N-terminal extension (Fig. 8E, lanes 2, 4, and 6), supporting previous findings (Fig. 7D) that native human TRPC1 mRNA utilizes an upstream non-AUG site as seen in transfected mouse TRPC1. These data further reveal a possible regulation of TRPC1 protein by Orai1 and STIM1.
Next, we examined Ca 2ϩ and Ba 2ϩ permeability of Orai1 in the presence of TRPC1⑀. Substitution of extracellular Na ϩ with an equimolar concentration of the nonpermeable N-methyl-Dglucamine did not affect the size of store-operated inward currents in cells transfected with STIM1 ϩ Orai1 or STIM1 ϩ Orai1 ϩTRPC1⑀ suggesting that TRPC1⑀ amplified Orai1-mediated Ca 2ϩ currents (Fig. 8F). However, TRPC1⑀ did increase the permeability of Orai1 to Ba 2ϩ . Co-expression of TRPC1⑀ with STIM1 and Orai1 resulted in larger Ba 2ϩ currents compared with STIM1 ϩ Orai1 (Fig. 8G), strongly suggesting the formation of a heteromultimeric channel of Orai1 and TRPC1⑀. As shown in Fig. 8E, expression levels of STIM1 and Orai1 were not affected by the presence or absence of TRPC1⑀, which could potentially affect Ba 2ϩ permeability in triple-transfected cells (50). Substitution of extracellular Ca 2ϩ with a divalent-free solution proportionally augmented currents mediated by STIM1-activated Orai1 or Orai1/TRPC1⑀, indicating that TRPC1⑀ did not affect monovalent permeability of STIM1-activated Orai1 channel (data not shown). Consistent with the qualitative rather than quantitative effects of TRPC1␣ and TRPC1⑀ on I CRAC , both isoforms were expressed at comparable levels in the plasma membrane (Fig. 9, middle panel, lanes 2 and 3).

Suppression of I SOC and I CRAC by I-mfa through TRPC1-
Next, we tested the effect of I-mfa on I SOC and I CRAC in the presence or absence of TRPC1␣ or TRPC1⑀, respectively. Fig.  10A shows that I-mfa suppressed I CRAC in cells transfected with STIM1, Orai1, and TRPC1⑀ but not in cells transfected with STIM1 and Orai1. Expression levels of STIM1, Orai1, or TRPC1 did not change by co-transfection with I-mfa or I-mfb (Fig. 10C). I-mfb, which is a splice variant of I-mf lacking the TRPC1-binding site did not suppress I CRAC (Fig. 10, A and D). Interestingly, I-mfa suppressed I CRAC to a lower magnitude than the magnitude of I CRAC mediated by Orai1 in the absence of TRPC1⑀, suggesting that TRPC1⑀ "sensitized" Orai1 to I-mfa-mediated inhibition. I-mfa had a similar effect on I SOC in cells transfected with STIM1, Orai1, and TRPC1␣ (Fig. 10, B  and D). Because I-mfa does not physically interact with STIM1 or Orai1 (data not shown), does not disrupt the Orai1/TRPC1 interaction (Fig. 10E), and suppresses I SOC /I CRAC only in the presence of TRPC1␣/⑀ (Fig. 10, A and B), we conclude that I-mfa suppresses these currents by being recruited to the Orai1-TRPC1 complex through an interaction with TRPC1␣/⑀. In sum, these results show that TRPC1 has a dual effect on Orai1-mediated I CRAC . In the absence of I-mfa, TRPC1⑀ amplifies Orai1-mediated I CRAC , whereas in the presence of I-mfa, it mediates I-mfa-induced inhibition of Orai1-mediated I CRAC . The positive and negative modulation of Orai1-mediated current by TRPC1 and I-mfa, respectively, suggests that the dynamic range of the CRAC channel can be enhanced by TRPC1 and I-mfa.

DISCUSSION
Our study provides several lines of evidence supporting the hypothesis that Trpc1 and I-mfa genetically and functionally interact to regulate osteoclastogenesis through store-operated Ca 2ϩ entry channels. First, I-mfa Ϫ/Ϫ mice show increased osteoclast formation in vivo, which is suppressed in mice lacking both genes. Second, pre-osteoclasts derived from I-mfa null mice have an increased number of mature osteoclasts and higher resorptive activity per osteoclast, which are normalized in cells derived from double mutant mice. Third, store-operated Ca 2ϩ currents are enhanced in I-mfa-null myeloid precursors and suppressed in double mutant cells. Fourth, TRPC1 requires core components of the CRAC channel, STIM1 and Orai1 for function, and I-mfa suppresses store-operated currents only in the presence of TRPC1. The data lead us to propose that TRPC1 and I-mfa increase the dynamic range of the CRAC channel, which can account for the observed effects on osteoclastogenesis.
I-mfa null mice have a significant osteopenic phenotype with increased osteoclastogenesis. Because the effect of I-mfa deletion on osteoclastogenesis was completely rescued by the additional deletion of Trpc1, we suggest that I-mfa promoted osteoclastogenesis by a mechanism related to TRPC1 and unrelated to its role as an inhibitor of the MyoD and other basic helixloop-helix transcription factors and/or through the canonical Wnt/␤-catenin pathway. However, involvement of these pathways is likely to be important in regulating bone mass independently of osteoclastogenesis in I-mfa null mice. This is supported by the incomplete normalization of bone mass in compound mice compared with mice lacking I-mfa. In contrast to the osteoporosis of I-mfa null mice, TRPC1 null mice showed a mild increase in bone mass. The effect of the Trpc1 deletion on osteoclastogenesis is revealed only in mice lacking I-mfa. These observations lead us to suggest that under normal physiological conditions, where I-mfa is quickly down-regulated by RANKL while TRPC1⑀ is up-regulated, TRPC1 can promote osteoclastogenesis. In regard to the genetic interaction of Trpc1 and I-mfa in osteoclastogenesis, we propose that maximal and/or persistent activation of I CRAC /I SOC through TRPC1 in cells lacking I-mfa leads to excessive osteoclastogenesis and reduced bone mass. This suggestion is supported by our experiments in myeloid precursors and in studies in Orai1 null mice showing similar but more severely defective osteoclastogenesis (51) and in vitro studies using Orai1-depleted osteoclasts (52,53). The more severe and nonspecific effect of the deletion of Orai1 in numerous cell types, including osteoclasts compared with TRPC1, is in agreement with Orai1 being a core component of the CRAC channel and TRPC1 being a regulatory protein whose function is dispensable for Orai1.
The TRPC1/I-mfa interaction is likely to affect osteoclastogenesis at both an early stage, possibly at a step sensitive to M-CSF-induced Ca 2ϩ signaling and at a later stage affecting bone resorption. The idea that an early step is affected by TRPC1 and I-mfa is supported by the M-CSF-induced expression of I-mfa and Trpc1⑀ mRNAs in myeloid precursors, promoting not only the formation of a highly Ca 2ϩ -selective CRAC channel complex but also its negative regulation by I-mfa. Our electrophysiological experiments in these cells clearly demonstrate a role of these two proteins in Ca 2ϩ signaling at this stage of osteoclastogenesis. However, we cannot pinpoint which TRPC1 isoform is responsible for the observed effects on Ca 2ϩ signaling and phenotypes associated with osteoclastogenesis. The development of a linear current in cells lacking I-mfa clearly argues for the functional expression of TRPC1␣ in these cells. However, a role of TRPC1⑀ cannot be ruled out, as the contribution of TRPC1␣, TRPC1⑀, or other TRPC1 isoforms in the inward component of store operated currents in these cells is unknown.
In regard to the molecular mechanism by which Trpc1 and I-mfa affect early osteoclastogenesis, we speculate that M-CSF "primes" myeloid precursors for RANKL-mediated signaling, not only through the well known up-regulation of RANK (54) but also through the up-regulation of both TRPC1⑀ and I-mfa (Fig. 11). However, Ca 2ϩ signaling and downstream activation of NFATc1 is suppressed at this stage as I-mfa suppresses CRAC channel activity through TRPC1⑀. Upon stimulation with RANKL, cells become competent for Ca 2ϩ signaling by down-regulating I-mfa releasing the block on CRAC channel. This idea is consistent with the lack of an effect of I-mfa on a specific class of osteoclasts (small, medium, or large sized), as all groups were up-regulated proportionally, arguing against a possible effect of I-mfa within the differentiation process. One possibility is that I-mfa suppresses the survival and/or proliferation of early osteoclast progenitors in response to M-CSF. Interestingly, M-CSF signaling is essential for the proliferation and survival of these progenitors through the up-regulation of ␤-catenin mediated by the action of the Ca 2ϩ -sensitive Pyk2 tyrosine kinase (3). Therefore, we could envision a positive feedback loop whereby accumulated ␤-catenin could compete with TRPC1 for binding to I-mfa, relieving the I-mfa-mediated suppression of TRPC1 activity, thus allowing for Ca 2ϩ influx. A recent study showed that M-CSF instructs hematopoietic stem cells toward the myeloid lineage in addition to previously known effects on survival and/or proliferation of committed hematopoietic progenitors (55). In light of these data, our study   11. Hypothetical model for the modulation of I SOC and I CRAC by TRPC1 and I-mfa in early stages of osteoclastogenesis. M-CSF primes myeloid precursors for Ca 2ϩ signaling by up-regulating not only TRPC1⑀ but also its negative regulator, I-mfa. As a result, Ca 2ϩ signaling is maintained at a low level in this stage. In response to RANKL, I-mfa is down-regulated, although TRPC1⑀ expression persists, setting up myeloid precursors/early pre-osteoclasts highly competent for Ca 2ϩ signaling, which is crucial for the downstream activation of NFATc1 and other regulators of osteoclastogenesis.
has implications in M-CSF-induced differentiation of hematopoietic stem cells. Future studies are needed to investigate which M-CSF-dependent process is affected by TRPC1/I-mfamediated Ca 2ϩ signaling.
The enhanced ability of I-mfa-deficient osteoclasts to resorb bone is consistent with a role of TRPC1 in the regulation of secretion per se through SOCE channels. In fact, TRPC1 knockout mice used in our study show severely reduced salivary gland fluid secretion (56), and secretion and exocytosis are known functions of the CRAC channel (11). Therefore, it is tempting to speculate that increased resorptive activity in I-mfa-deficient osteoclasts is due to enhanced acid secretion secondary to upregulated SOCE.
We present several lines of evidence supporting the hypothesis that Orai1 forms a complex with TRPC1. First, the TRPC1␣ and TRPC1⑀ isoforms require Orai1 for functional expression. Second, transfected TRPC1 co-immunoprecipitates with endogenous or transfected Orai1 in HEK293T cells. Third, TRPC1␣ and TRPC1⑀ each modulate the current mediated by Orai1, by generating I SOC or by forming a Ba 2ϩ -permeable channel complex, respectively. Fourth, I-mfa suppresses Orai1mediated current only in the presence of TRPC1⑀. These data lead us to propose a model whereby STIM1 and Orai1 form the core module of the CRAC channel, whereas TRPC1 and I-mfa form a regulatory module that enhances the dynamic range of this channel. However, we do not completely understand how this channel complex is formed. One possibility is that TRPC1␣/⑀ and Orai1 form a heteromultimeric complex with a "chimeric" pore region. A second possibility is that they form different assemblies, but TRPC1 can regulate the activity of Orai1 by physical interactions through cytosolic fragments, possibly interfering with the gating of Orai1 by STIM1. We favor the second possibility, for several reasons. First, it is difficult to envision how TRPC1␣/Orai1 and TRPC1⑀/Orai1 could have different chimeric pores, because TRPC1␣ and TRPC1⑀ only differ by seven amino acids, which are located in N-terminal cytosolic region of TRPC1␣. Second, the crystal structure of Orai1 revealed that purified Orai1 could form a functional pore without the need for additional subunits (57). Third, STIM1 not only gates but also determines the cation selectivity of Orai1 (50,58). Therefore, it is conceivable that physical interactions through the N-terminal cytosolic region of TRPC1 and Orai1 could indirectly affect the cation permeability of Orai1 by interfering with binding to STIM1. We also favor this model because it does not require TRPC1 and Orai1 to be present in the same membrane. For example, TRPC1 can be in the endoplasmic reticulum or the plasma membrane.
The formation of I SOC or I CRAC by different TRPC1 isoforms implies that the magnitude of I CRAC and I SOC in different cell types can greatly vary depending on the expression levels of TRPC1 and I-mfa, even if levels of STIM1 and Orai1 are similar. Furthermore, our model implies that a cell would express I SOC or I CRAC depending on the TRPC1 isoform present in the cell. If there is no TRPC1 expression or I-mfa is in excess of TRPC1, cells would have Orai(1-3)-mediated I CRAC . TRPC1 isoform switching in response to an extracellular stimulus (i.e. M-CSF in osteoclast precursors) along with the profound functional differences between isoforms adds an additional layer of complexity in the regulation of SOCE channels.
Our studies suggest that modulating the dynamic range of the CRAC channel can control osteoclastogenesis. Therefore, inhibition of TRPC1 through small molecules or pore-blocking antibodies, suppression of its expression, or up-regulation of I-mfa could constitute new ways to combat conditions associated with abnormally enhanced osteoclastogenesis. Many disease states, including chronic periodontitis, osteoporosis, rheumatoid arthritis, Paget disease, and cancer metastases develop when osteoclasts are excessively recruited or inappropriately activated. Targeting TRPC1 in these conditions can be considered more advantageous than targeting STIM and/or Orai molecules, which can have more severe and widespread side effects. Alternatively, small molecule inhibitors targeting the interaction between TRPC1 and I-mfa might be beneficial for high bone mass-related diseases such as osteopetrosis, where the balance is shifted toward reduced osteoclastogenesis. Although it is premature to speculate on targeting strategies at this time, our work offers new approaches to therapeutic interventions for a wide variety of bone diseases.