Identification of receptors and signaling pathways for orphan bone morphogenetic protein/growth differentiation factor ligands based on genomic analyses.

There are more than 30 human transforming growth factor beta/bone morphogenetic protein/growth differentiation factor (TGFbeta/BMP/GDF)-related ligands known to be important during embryonic development, organogenesis, bone formation, reproduction, and other physiological processes. Although select TGFbeta/BMP/GDF proteins were found to interact with type II and type I serine/threonine receptors to activate downstream Smad and other proteins, the receptors and signaling pathways for one-third of these TGFbeta/BMP/GDF paralogs are still unclear. Based on a genomic analysis of the entire repertoire of TGFbeta/BMP/GDF ligands and serine/threonine kinase receptors, we tested the ability of three orphan BMP/GDF ligands to activate a limited number of phylogenetically related receptors. We characterized the dimeric nature of recombinant GDF6 (also known as BMP13), GDF7 (also known as BMP12), and BMP10. We demonstrated their bioactivities based on the activation of Smad1/5/8-, but not Smad2/3-, responsive promoter constructs in the MC3T3 cell line. Furthermore, we showed their ability to induce the phosphorylation of Smad1, but not Smad2, in these cells. In COS7 cells transfected with the seven known type I receptors, overexpression of ALK3 or ALK6 conferred ligand signaling by GDF6, GDF7, and BMP10. In contrast, transfection of MC3T3 cells with ALK3 small hairpin RNA suppressed Smad signaling induced by all three ligands. Based on the coevolution of ligands and receptors, we also tested the role of BMPRII and ActRIIA as the type II receptor candidates for the three orphan ligands. We found that transfection of small hairpin RNA for BMPRII and ActRIIA in MC3T3 cells suppressed the signaling of GDF6, GDF7, and BMP10. Thus, the present approach provides a genomic paradigm for matching paralogous polypeptide ligands with a limited number of evolutionarily related receptors capable of activating specific downstream Smad proteins.

Reagents and Hormones-Dulbecco's modified Eagle's medium, ␣-minimal essential medium, and Opti-MEM were obtained from Invitrogen. L-Glutamine, penicillin, and streptomycin were purchased from BioWhittaker (Walkersville, MD). Recombinant mouse GDF6 and GDF7 as well as human BMP2, BMP7, and TGF-␤ 1 were from R & D Systems (Minneapolis, MN). Phospho-Smad1 and phospho-Smad2 antibodies were obtained from Dr. C. H. Heldin (Ludwig Institute for Cancer Research, Uppsala, Sweden). Biotinylated anti-mouse GDF6 and GDF7 antibodies were purchased from R & D Systems. Anti-His monoclonal antibodies were from Amersham Biosciences. Peptide N-glycosidase F (PNGase) was obtained from New England Biolabs (Beverly, MA).
Generation of Recombinant Human BMP10-Human BMP10 cDNA was amplified from the human heart Marathon-Ready cDNA library (Clontech) using specific primers (forward, 5Ј-TTAGCGGCCGCAT-GGGCTCTCTGGTCCTGACACTGT-3Ј; Reverse, 5Ј-TTATCTAG-ACTATCTACAGCCACATTCGGAGACG-3Ј) and Turbo Pfu polymerase (Stratagene, La Jolla, CA). Fidelity of the PCR products was confirmed by sequencing after subcloning the cDNA into the expression vector pcDNA3.1/Zeo (Invitrogen). To facilitate purification, the BMP10 construct was tagged with a polyhistidine tag (His 6 ) at the N terminus. Recombinant BMP10 proteins were generated from 293T cells transfected with expression plasmids using Lipofectamine (Invitrogen) and maintained in Dulbecco's modified Eagle's medium/ F-12 supplemented with 10% FBS. Clonal cell lines stably expressing recombinant proteins were selected under 500 g/ml Zeocin (Invitrogen) and maintained in 200 g/ml Zeocin. After 3 days of serum-free culture, the conditioned media were harvested, adjusted to 20 mM imidazole, and incubated with metal-chelating Sepharose (Amersham Biosciences). The bound proteins were eluted with the wash buffer contain-ing 500 mM imidazole. Protein purity and biochemical characteristics were analyzed after electrophoresis using a 12% SDS-PAGE.
Small Hairpin RNA Plasmid Construction-The pSilencer 2.0-U6 plasmid (Ambion, Austin, TX) containing the human U6 RNA polymerase III promoter was used to construct the shRNA-encoding plasmids targeting BMPRII, ActRIIA, or ALK6 mRNAs. They were generated by inserting annealed oligonucleotides (TABLE ONE) into the pSilencer 2.0-U6 plasmid between the BamHI and HindIII restriction sites using the T4 DNA ligase (Roche Diagnostics). Each oligonucleotide contained the target sequences, the restriction sites at both ends, the linker sequences, and a stretch of T or A residues as the polymerase III terminator. The 21-nucleotide target sequences were defined from the mouse cDNA sequences by using the small interfering RNA Target Finder (Ambion). After synthesis, the sense and antisense oligonucleotides were annealed at 90°C for 3 min and 37°C for 1 h. In addition, the Sequences of shRNAs for different type I and type II receptors Sense and antisense oligonucleotides for each receptor gene were designed using the siRNA Target Finder (Ambion, Austin, TX). The GenBank TM accession numbers for each mouse gene are shown in parentheses. The sense (S) and antisense (AS) sequences for the shRNAs are shown on the right. Target RNA sequences are in boldface; lowercase letters represent restriction enzyme sites, underlined are the linker sequences, and a stretch of T and A residues serves as the polymerase III terminator. Control shRNA encodes random sequences with no homology to any known mouse genes. In contrast to other shRNAs, the shALK3 was purchased from OpenBiosystems (Huntsville, AL) and a single oligonucleotide, complementary to the target sequence (boldface), was amplified and cloned into the pSHAG-MAGIC2 vector. Cell Lines and Transfection-MC3T3 and P19 cells were cultured in ␣-minimal essential medium supplemented with 10% FBS, together with 100 units/ml penicillin, 100 g/ml streptomycin, and 2 mM L-glutamine (PSG). COS7 cells were cultured in Dulbecco's modified Eagle's medium/high glucose supplemented with 10% FBS and PSG. MC3T3 and COS7 cells were seeded at 90% confluency in 24-well plates and transiently transfected in Opti-MEM for 4 h with 1 g (COS7 cells) or 300 ng (MC3T3 cells) of plasmid DNA per well using Lipofectamine 2000 (Invitrogen). Fifty ng of the pCMV-␤-galactosidase vector was routinely cotransfected to monitor transfection efficiency. After transfection, cells were treated with the appropriate ligands. To monitor luciferase activities, lysis buffer (200 l) (Promega Corp., Madison, WI) was added to each well, and 30 l of the supernatant was used for luciferase determination using a luminometer (Bio-Rad). Aliquots of the cell lysate were also used to measure ␤-galactosidase activities. The reporter activity is expressed as the ratio of relative light unit/␤-galactosidase activity. Data are the mean Ϯ S.E. of triplicates from representative experiments.
Western Blotting Analysis of Smad Proteins-To investigate Smad activation by GDF6, GDF7, and BMP10, MC3T3 cells (180,000 cells/ well) were cultured overnight in 12-well plates in ␣-minimal essential medium containing 10% FBS and starved for 3 h in serum-free media to minimize basal Smad activity. MC3T3 cells were treated with different hormones for 30 min and washed once on ice with chilled phosphatebuffered saline before cell lysis in the loading buffer containing ␤-mercaptoethanol. Cells were gently sonicated on ice for 15 s with an MSE sonicator (Sanyo Corp., Osaka, Japan) and boiled for 3 min. Cellular proteins were separated on 10% SDS-polyacrylamide gels and electroblotted onto Hybond-electrochemiluminescence (for Smad2 experiments) and Hybond-P membranes (for Smad1 experiments) (Amersham Biosciences). For the detection of phosphorylated Smads, membranes were blocked for 1 h at room temperature in Tris-buffered saline containing 0.05% Tween and 5% fat-free dry milk. After blocking of nonspecific binding, membranes were incubated with antiphospho-Smad1 or antiphospho-Smad2 antibodies diluted at 1:5,000 at 4°C overnight. The secondary anti-rabbit antibodies were used following manufacturer's instructions (Amersham Biosciences). Immunoreactive proteins were detected using enhanced chemiluminescence (ECL kit, Amersham Biosciences).
Receptors Expression Profiles Based on RT-PCR and Real-time PCR-Messenger RNAs were extracted from MC3T3, P19, and COS cell lines using the RNeasy kit (Qiagen, Valencia, CA). cDNA was synthesized by using reverse transcriptase (Clontech). Traditional PCR was carried out using the Advantage cDNA polymerase kit (Clontech), and real-time PCR was performed using SmartCycler (Cepheid Inc., Sunnyvale, CA) according to the manufacturer's protocol. Primers used for individual genes are listed in TABLE TWO. To determine the copy number of target transcripts, ALK3, ALK6, BMPRII, and ActRIIA cDNAs were used to generate calibration curves by plotting the threshold cycle (Ct) versus the known copy number for each plasmid template. The copy numbers for target samples were determined according to the calibration curve.

Phylogenetic Relationship of TGF␤ Family Ligands and Their
Corresponding Signaling Pathways-The TGF␤ superfamily includes 33 related ligands, but there are only 5 type II and 7 type I serine/threonine kinase receptors in the human genome (1,19). Although approximately one-third of these ligands are still orphans without known receptors and downstream signaling pathways, cognate receptors and downstream Smads for the remaining ligands have been characterized (Fig. 1). After interacting with type I and type II serine/threonine kinase receptors, the ligands activate two major pathways characterized by the activation of two different groups of intracellular Smad proteins. Smad1, Smad5, and Smad8 are phosphorylated by the type I receptors ALK1-3 or -6, whereas Smad2 and Smad3 are activated by the receptors ALK4, -5, or -7 (6 -9). It is apparent that combinatorial uses of a limited number of type I and type II receptors led to differential activation by the large number of ligands. Because many of the BMP and GDF proteins are still orphan ligands, we hypothesized that these ligands are likely to interact with the limited number of receptors, and phylogenetically related Primers and probes for RT-PCR analyses of type I and type II receptor transcripts from different species FAM means 6-carboxyfluorescein; TAMRA means 6-carboxytetramethylrhodamine.

Genes
GenBank TM accession no.

Primers forward (5-3) Primers reverse (5-3) Probes (5FAM-TAMRA3)
ligands are likely to interact with the same receptors. We selected GDF6, GDF7, and BMP10 as prototypic orphan ligands, and we investigated their signaling pathways. Based on the phylogenetic analysis ( Fig. 1), the three orphan ligands are closely related to BMP2 and GDF5. Thus, they could activate the Smad1/5/8 pathway through interactions with the type I receptors ALK3 and ALK6 and the type II receptors BMPRII and ActRIIA.
Characterization of Recombinant GDF6, GDF7, and BMP10-Members of the TGF␤ superfamily are synthesized as large precursor molecules that are cleaved proteolytically at an RXXR site to release a C-terminal mature peptide of 110 -140 amino acids in length (2,20). This region shares conserved cysteine residues involved in the cystine knot conformation (21). In many cases, an extra cysteine residue is engaged in intermolecular disulfide bond formation necessary for the assembly of biologically active dimers. We analyzed the biochemical properties of recombinant GDF6 and GDF7 under both reducing and nonreducing conditions using anti-GDF6 ( Fig. 2A, lanes 1-2) and anti-GDF7 ( Fig.  2A, lanes 3-4) antibodies. Under reducing conditions, a band of 13 kDa likely represented the GDF6 monomer ( Fig. 2A, lane 2). Under nonreducing conditions (Fig. 2A, lane 1), an ϳ30-kDa band and a 13-kDa band likely represented GDF6 dimers and monomers, respectively. For GDF7, distinct dimer and monomer bands of 32 ( Fig. 2A, lane 3) and 16 kDa (Fig. 2A, lane 4) were found under nonreduced and reduced conditions, respectively. These data suggested that both GDF6 and GDF7 formed disulfide-linked homodimers.
To generate recombinant BMP10, we first performed RT-PCR using a heart cDNA library to obtain the full-length human BMP10 cDNA. Human 293T cells were transfected with expression vectors containing the BMP10 cDNA tagged at its N terminus with a polyhistidine epitope His 6 . Following selection of permanent clones, cell lines were cultured under serum-free conditions to obtain conditioned media. Recombinant BMP10 was purified using metal chelate chromatography based on the presence of the His 6 tag. Following Coomassie Blue staining, three distinct bands were found under nonreducing conditions. The 25-kDa band represented the dimer form, and the 64-and 120-kDa bands likely represented the unprocessed monomers and dimers, respectively (Fig.  2B, lane 1). These unprocessed BMP10 forms contained the uncleaved large prodomain in addition to the C-terminal mature region. Under reducing conditions, BMP10 migrated as a 12-kDa band representing the monomer form and a 60-kDa band corresponding to the unproc- BMP10 with an N-terminal His 6 tag was purified from conditioned media of 293T cells using metal chelate chromatography. BMP10 was detected by Coomassie Blue staining (lanes 1-3) and immunoblotting using antibodies against polyhistidine His 6 (lanes 4 and 5) under nonreduced or reduced conditions. Some samples were treated with peptide N-glycosidase F (PNGase) to remove N-linked carbohydrate side chains. essed monomer (Fig. 2B, lane 2). A third weak band of 48 kDa could represent a shorter unprocessed form of BMP10 generated after cleavage of the N-terminal region of the prodomain (Fig. 2B, lane 2). In addition, under reducing conditions, the anti-His 6 antibody revealed the same three bands representing the monomer (12 kDa) and the unprocessed monomer forms of 60 and 48 kDa (Fig. 2B, lane 4). After treatment with PNGase F to remove N-linked carbohydrate chains, the 60-and 48-kDa unprocessed forms of BMP10 migrated faster (Fig. 2B,  lanes 3 and 5), suggesting that the N-glycosylation sites are located in the prodomain. These data suggested that the recombinant BMP10 was secreted as an N-glycosylated homodimeric proform, and after processing, the mature protein formed nonglycosylated homodimers.
Treatment with GDF6, GDF7, or BMP10 Activates the Smad1/5/8 Pathway in MC3T3 Cells-We hypothesized that GDF6, GDF7, and BMP10 shares downstream signaling molecules with other TGF␤ ligands. We took advantage of the availability of different promoterluciferase constructs for these hormones to elucidate their signaling pathways. The BRE promoter is known to be activated by several BMPs mediated by Smad1/5/8, whereas the CAGA promoter is activated by TGF␤/activin mediated by Smad2 and Smad3 (15,16,22). We tested the responses of the two reporter constructs in several cell lines transfected with these plasmids following treatment with the three orphan ligands. Although most cell lines tested did not respond to the ligand stimulation (data not shown), treatment of the MC3T3 cells with GDF6, GDF7, or BMP10 stimulated the luciferase reporter driven by the BRE promoter (Fig. 3A). Consistent with earlier findings (15,22), the BRE promoter was activated by BMP2 and BMP7. Although TGF-␤ was ineffective in activating the BRE promoter (Fig. 3A), treatment with TGF-␤ stimulated the reporter driven by the CAGA promoter (Fig. 3B) (16). In contrast, treatment with GDF6, GDF7, or BMP10 did not activate the CAGA promoter (Fig. 3B). Similarly, treatment with BMP2 or BMP7 did not stimulate the CAGA promoter (Fig. 3B). Furthermore, the stimulatory effects of GDF6, GDF7, or BMP10 on the BRE promoter were found to be dose-dependent (Fig. 3C). The ability of GDF6, GDF7, and BMP10 to activate the Smad 1/5/8 pathway was further confirmed by immunoblotting analysis using specific antibodies against phospho-Smad1 or phospho-Smad2. As shown in Fig. 3D, treatment with GDF6, GDF7, or BMP10, similar to BMP2, induced the phosphorylation of Smad1 in MC3T3 cells. In contrast, they were ineffective in inducing phospho-Smad2 in the same cells. These findings suggest that MC3T3, a mouse pre-osteoblast cell line derived from neonatal mouse calvaria (23), likely expresses receptors necessary for GDF6, GDF7, and BMP10 signaling.
Stimulation of the BRE Promoter by GDF6, GDF7, or BMP10 Is Blocked by the Inhibitory Smad6 and Smad7-The transcriptional activities of all R-Smad proteins (Smad1-3, -5, and -8) are blocked by the inhibitory Smad7 (24 -27), whereas those of Smad1, -5, and -8, but not Smad2 and -3, are blocked by the inhibitory Smad6 (27)(28)(29). We tested the inhibitory activities of these two inhibitory Smad proteins in MC3T3 cells treated with GDF6, GDF7, or BMP10. As shown in Fig. 4, cotransfection with plasmids encoding either Smad6 or Smad7 led to a dose-dependent suppression of the stimulation of the BRE promoter activity by GDF6, GDF7, or BMP10. In addition, a minor inhibition of the basal promoter activity was also found. Consistent with the reported role of Smad1, -5, and -8 in BMP2 signaling (30), overexpression of either Smad6 or Smad7 blocked the BMP2 stimulation of the BRE promoter in the same cells (Fig. 4). These results suggest that ligand signaling by GDF6, GDF7, or BMP10 does not involve the Smad2 and Smad3 pathway but is mediated by the BMP-responsive pathway through Smad1, -5, and -8.
Overexpression of ALK3 or ALK6 Confers GDF6 and GDF7 Responsiveness in the COS7 Cells-In preliminary tests, we found that treatment with GDF6 or GDF7 was minimally effective in activating the BRE promoter in the COS7 cells (Fig. 5A). We hypothesized that the type I receptor levels were too low in this cell line to allow signaling by the two orphan ligands. We performed overexpression tests in an attempt to gain ligand responsiveness and to identify the type I receptors involved in GDF6 and GDF7 signaling. As shown in Fig. 5A, transfection of these cells with increasing amounts of the ALK3 or the ALK6 expression plasmid led to a dose-dependent stimulation of the BRE promoter activity by GDF6 or GDF7. In contrast, overexpression of ALK1, -2, -4, -5, or -7 did not confer GDF6 or GDF7 activation of the BRE promoter.
Unlike GDF6 and GDF7, COS7 cells were already responsive to BMP10 treatment, and transfection of increasing amounts of the ALK3 plasmid, but not the ALK6 plasmid, showed a small increase in the stimulation of the BRE promoter activity (Fig. 5A). In addition, overexpression of ALK1, -2, -4, -5, or -7 did not confer BMP10 activation of the BRE promoter. The increased expression of all seven ALK receptors in transfected cells was confirmed using RT-PCR (Fig. 5B).
Expression of the Type I Receptor Transcripts in MC3T3 Cells-To confirm the importance of ALK3 and ALK6 in MC3T3 cells that were responsive to GDF6, GDF7, or BMP10, we performed RT-PCR using total RNA extracted from these cells to define the expression pattern of these receptors. As shown in Fig. 6, transcripts for ALK3, but not ALK6, were expressed in MC3T3. In addition, P19 cells were used as a positive control to verify the amplification of ALK6 transcript by RT-PCR (Fig.  6). The transcript levels for ALK3 and ALK6 in MC3T3 cells were confirmed by quantitative real-time PCR analysis (ALK3, 798,190 Ϯ 134,79 copies/80,000 cells; ALK6, Ͻ2,000 copies/80,000 cells).
The Roles of Endogenous ALK3, BMPRII, and ActRIIA in Ligand Signaling by GDF6, GDF7, and BMP10-Because MC3T3 cells are responsive to GDF6, GDF7, and BMP10 and express transcripts for the candidate type I receptor ALK3, we further tested the involvement of endogenous ALK3 in GDF6, GDF7, and BMP10 signaling. MC3T3 cells were transfected with plasmids encoding small hairpin (sh)RNA before testing the ability of the ligands to activate the BRE promoter. As shown in Fig. 7, transfection with increasing amounts of the ALK3 shRNA, but not the control shRNA, led to dose-dependent decreases in the BRE promoter activity stimulated by GDF6, GDF7, or BMP10. Although transfection with the ALK3 shRNA also suppressed BRE promoter activity stimulated by BMP2 in a dose-dependent manner, no suppression of the CAGA promoter activity by TGF␤ was observed (Fig. 7). These findings demonstrated the role of endogenous ALK3 in GDF6, GDF7, and BMP10 signaling in MC3T3 cells.
Phylogenetic analysis of the entire repertoire of BMP/GDF ligands suggested that GDF6, GDF7, and BMP10 are closely related to BMP2, BMP7, and GDF5 (Fig. 1) and are likely to share the same type II receptors, BMPRII and ActRIIA. We first performed RT-PCR analysis to confirm the expression of BMPRII and ActRIIA in MC3T3 cells (Fig. 6). In addition, quantitative RT-PCR indicated that the BMPRII expression level was higher (836,360 Ϯ 166,050 copies/80,000 cells) than the ActRIIA expression level (43,890 Ϯ 7,790 copies/80,000 cells).
To test the role of endogenous BMPRII and ActRIIA in GDF6, GDF7, and BMP10 signaling, plasmids encoding BMPRII and ActRIIA shRNA were transfected into MC3T3 before testing the ability of the three ligands to activate the BRE promoter. As shown in Fig. 8, transfection of MC3T3 cells with BMPRII shRNA strongly suppressed the BRE promoter activity induced by GDF6, GDF7, or BMP10 in a dose-dependent manner, reaching a complete inhibition with 3 ng of transfected plasmid. Although ActRIIA shRNA was transfected at higher doses, GDF6 and GDF7 signaling was decreased by only 25%, and the BMP10 signaling was decreased by 50%. In addition, treatment with BMPRII shRNA or ActRIIA shRNA also suppressed BRE promoter activity stimulated by BMP2 in a dose-dependent manner (Fig. 8). However, no suppression of CAGA promoter activity induced by TGF␤ was observed following the same treatments (Fig. 8, bottom panel). These data demonstrated the major role of endogenous ALK3 and BMPRII in GDF6, GDF7, and BMP10 signaling in MC3T3 cells.

DISCUSSION
Based on a genomic analysis of the entire repertoire of TGF␤/BMP/ GDF ligands and their corresponding serine/threonine kinase receptors, we tested the ability of three orphan GDF/BMP ligands to activate a limited number of phylogenetically related receptors. We characterized dimeric recombinant GDF6, GDF7, and BMP10 proteins, followed by the identification of the MC3T3 cell line responsive to these ligands based on the activation of the BMP-responsive promoter construct and the phosphorylation of the intracellular Smad1. Overexpression studies in a low responsive COS7 cell line and RNA interference analyses in MC3T3 cells further demonstrated the role of the type I receptors ALK3 and ALK6 in the mediation of GDF6, GDF7, or BMP10 signaling. We also found that overexpression of shRNA for the type II receptors BMP- RII and ActRIIA suppressed the signaling of GDF6, GDF7, and BMP10 in MC3T3 cells. Thus, the present approach provides a genomic paradigm for matching paralogous polypeptide ligands with a limited number of evolutionarily related receptors capable of activating downstream Smad proteins.
The three orphan ligands studied here play important roles in diverse physiological processes, including bone morphogenesis, nervous system development, and heart formation. GDF6 and GDF7 are expressed in developing cartilage, tendons, and ligaments, as well as in the nervous system (11,13,(33)(34)(35)(36). GDF6 mutant mice showed joint defects with fusion between bones in the wrists and ankles (35). In Xenopus and Zebrafish embryos, GDF6 is expressed in the ectoderm and acts as an epidermal inducer and a neural inhibitor (37,38). In contrast, GDF7 mutant mice were characterized by the absence of a subclass of commissural interneurons in the spinal cord and defects in the development of the seminal vesicle (11,12,39). A recent study also demonstrated the role of GDF7 in restricting the development of neural crest cells to sensory neurons (13). Furthermore, BMP10 expression is restricted to the developing and postnatal heart in the mouse and chicken (40 -42). BMP10-deficient mice are embryonic lethal at E9.5-10 because of cardiac dysgenesis (14).
Studies using mice with defective type I and type II receptors only provided limited information on the potential receptors involved for GDF6, GDF7, and BMP10 because the phenotypes of different receptor null mice were distinct from those of mice with GDF6, GDF7, or BMP10 mutations. BMPRII and ALK3 are expressed in diverse embryonic and adult tissues, and null mice for each receptor showed early embryonic lethality (1, 3, 43-46). Of interest, targeted deletion of ALK3 in the heart leads to the thinning of the myocardial wall and expansion of the trabe-culated myocardium, similar to the phenotypes of BMP10 null mice (14,47). In addition, the ActRIIA null mice showed reproductive defects, whereas the ALK6 null mice exhibited a similar phenotype to that of the GDF5 null mice, both exhibiting joint defects (1,45,48,49). The present findings indicate that BMPRII and ALK3 are receptors for GDF6, GDF7, and BMP10. These data provide the basis to generate mice with tissuespecific deletion of these receptor genes in specific target tissues for future investigation.
In the present study, we found a pre-osteoblast MC3T3 cell line responsive to the three orphan ligands GDF6, GDF7, and BMP10. These ligands activated the BRE promoter through the phosphorylation of Smad1/5/8 (15,22). As shown in Fig. 1, the three orphan ligands are closely related to BMP2, BMP7, and BMP9, reported previously to stimulate the BRE promoter (2,15,17,22,50). The essential role of the Smad1/5/8 pathway in GDF6, GDF7, and BMP10 signaling is further confirmed by the blocking effect of the inhibitory Smad6. Smad6 specifically competes with R-Smad1 for complex formation with Smad4 (28), thus preferentially inhibiting the Smad1/5/8 pathway. We also confirmed the blocking effect of Smad7, which stably interacts with all activated type I receptors to prevent R-Smad activation and down-  stream transcriptional modulation (24,25). Similarly, Nakahara et al. (51) found that Smad6 and Smad7 also suppressed GDF5-mediated signaling in mouse B lineage cells.
Although GDF6 and GDF7 were expressed in the joint (34 -36), previous data suggested that GDF6 and GDF7 are not bone-inducing factors because, unlike BMP2, they did not induce the expression of the osteogenic marker alkaline phosphatase in the C2C12 cell lines (52,53). However, by using the BRE promoter activation assay directly associated with the activation of the Smad pathway, our results demonstrated that GDF6 and GDF7 were able to activate the Smad1/5/8 signaling in the pre-osteblast MC3T3 cell line. These data suggest that GDF6 and GDF7 may play a role on bone cells, likely independent of the stimulation of alkaline phosphatase expression. In contrast, osteoblasts are not likely to be the BMP10 target in vivo because of the restricted expression of BMP10 in the heart. Nevertheless, our results showed its potent ability to activate the BRE promoter in the pre-osteoblast cell line MC3T3. Our findings are consistent with earlier data showing the induction of the osteogenic marker alkaline phosphatase by BMP10 in the C2C12 cell line (53). It is clear that the specificity of the paracrine actions of various BMP family members is not only dependent on the available repertoire of receptors expressed in target cells but also on the tissue-specific expression of ligands.
To identify the type I receptors for GDF6, GDF7, and BMP10, we used overexpression and RNA interference approaches. In COS7 cells, we overexpressed individually the seven type I receptors, and we identified ALK3 and ALK6 as candidate type I receptors for GDF6 and GDF7 based on the stimulation of the BRE promoter. It is interesting to note that COS7 cells were already responsive to BMP10 treatment, and only overexpression of ALK3 slightly increased BMP10 actions. These results suggested that the endogenous levels of type I and type II receptors in COS7 cells are sufficient for BMP10 signaling but inadequate for Smad1/5/8 activation by GDF6 or GDF7, raising the possibility of differences in receptor binding properties between BMP10 and GDF6/ GDF7. We tested BMP10 action in diverse cell lines (P19, 293, CHO, and COS7) and found COS7 cells to be least responsive. Despite the observed responsiveness in COS7 cells, a potential role of ALK3 could still be demonstrated. Regarding the endogenous ALK proteins, our RT-PCR analyses indicated that ALK3, but not ALK6, was expressed in MC3T3 cells, suggesting that ALK3 is the type I receptor mediating GDF6, GDF7, and BMP10 signaling in this cell line. Indeed, transfection with the ALK3 shRNA suppressed GDF6, GDF7, and BMP10 stimulation of the BRE promoter. As expected, overexpression of the ALK3 shRNA also suppressed BMP2 signaling (43,54) but not the TGF␤ activation of the CAGA promoter mediated by ALK5 (16,55,56). By using the same approach, we induced the gene silencing of the two type II receptors, BMPRII and ActRIIA, expressed by MC3T3 cells, and we demonstrated the role of these proteins in BMP2, BMP7, and GDF5 signaling (43,54,57). Because of the pronounced inhibitory effects of BMPRII shRNA on GDF6, GDF7, and BMP10 signaling, BMPRII is likely the preferential type II receptor for the three ligands. These data showed that GDF6, GDF7, and BMP10 are signaling through the receptor complex BMPRII/ALK3 in MC3T3 cells. However, one cannot rule out the possibility that other cells use a different combination of the type I ALK3/ALK6 receptors and the type II BMPRII/ActRIIA receptors for signal transduction.
ALK3, ALK6, and ALK2 have been shown to be the three type I receptors mediating responses by at least 10 BMP-related ligands (Fig.  1). The effects seen with various BMP family members are highly dependent on the tissue/cell type being investigated and likely reflect the available repertoire and expression levels of individual receptors and downstream effectors. Thus, a specific response to a given ligand may be determined by both the stoichiometry of ALKs in the cell and their ability to form complexes with particular type II receptors. Moreover, the specificity of hormonal responses may also depend on the levels of specific ligands and the presence of accessory receptors, thus leading to different degrees of Smad activation (58). Formation of heterodimers between ligands may also modify the affinity for a given receptor complex and induce different physiological responses (59 -62). Indeed, GDF7 has been shown to enhance the axon-orienting activity of BMP7 (62) likely through the formation of GDF7/BMP7 heterodimers. Because our results demonstrated that BMP7 and GDF7 homodimers share the same receptors, heterodimers formed by BMP7 and GDF7 could bind the receptor complex with different affinities to initiate differential downstream signaling. Furthermore, various receptor-interacting proteins could regulate signaling by TGF␤/BMP/GDF ligands through different mechanisms (63). For example, GDF5 signaling through ALK6 is modulated by the tyrosine kinase receptor Ror2, which interacts with ALK6 and inhibits Smad1/5/8 signaling (64). Also, recep-FIGURE 8. Overexpression of BMPRII shRNA or ActRIIA shRNA blocked the stimulation of the BRE promoter activity by GDF6, GDF7, and BMP10. MC3T3 cells were transfected with 150 ng of the BRE reporter plasmid with or without increasing amounts of the BMPRII, ActRIIA, or control shRNA. As a negative control, cells were transfected with the BMPRII shRNA or ActRIIA shRNA before treatment with TGF␤. The relative luciferase activity was normalized based on the ␤-galactosidase activity to correct for variations in transfection efficiency. Results are presented as the mean Ϯ S.E. tor-interacting proteins, such as the TGF␤-activated kinase TAK1, may initiate complementary signals through the mitogen-activated protein kinase pathway (63,65,66).
Genomic analysis of the entire repertoire of TGF␤/BMP/GDF ligands and serine/threonine kinase receptors demonstrates that diverse related ligands utilize a limited set of receptors and Smad proteins to exert distinct and overlapping effects during embryogenesis and in adult tissues. We showed that three orphan ligands known to be important for joint and cartilage formation (GDF6) (10), interneuron, sensory neurons, and seminal vesicle formation (GDF7) (11)(12)(13), and heart development (BMP10) (14) used the type I receptors ALK3 or ALK6 and the type II receptors BMPRII or ActRIIA to activate the Smad1/5/8 proteins. Because many ligands of the TGF␤/activin/BMP family are still orphans, the present approach provides a genomic paradigm for matching paralogous polypeptide ligands with a limited number of evolutionarily related receptors capable of activating downstream Smad proteins.