Sequencing of this newly discovered Bmp5 mutation revealed a correctly spliced Bmp5 transcript with a G-to-A substitution at base 932 of the Bmp5 coding region (Fig. 1a, b). This change destroys a Taq1 site in the second exon (Fig. 1c), providing a simple assay for following the mutation in genetic crosses.

Most TGF-β superfamily proteins are synthesized as larger precursors that are cleaved at an RXXR consensus site by proprotein convertase endoproteases to generate an N-terminal pro domain and a C-terminal signaling domain 4344. The G-to-A mutation changes the first arginine in the RXXR processing site of BMP5 to a glutamine (Fig. 1a). To test whether the Arg311Gln mutation disrupts BMP5 protein processing, we expressed the wild-type or mutant forms in COS-7 cells and analyzed conditioned media by Western blot analysis with antibodies raised to the pro and mature regions of BMP5. Cells transfected with the wild-type construct produced BMP5 protein bands of ~40 kDa and ~20 kDa, consistent with the expected sizes of the cleaved BMP5 pro and mature domains (Fig. 1d). Both antibodies also detected a minor protein band of ~55 kDa (Fig. 1d), suggesting that some secreted BMP5 protein was unprocessed. Cells expressing mutant BMP5 produced only the unprocessed ~55 kDa protein form (Fig. 1d), confirming that the mutation in the RXXR site blocks normal proteolytic cleavage. We termed this mutation Bmp5^clv to denote the lesion at the cleavage site.

Mutations in the RXXR site of other TGF-β family members have inhibited the processing and activity of the corresponding protein 45464748. Such mutations also have acted as dominant-negative mutations that block the function of other coexpressed TGF-β members, presumably by sequestering them into inactive heterodimer complexes with the unprocessed mutant subunits 4547. Mice heterozygous for the Bmp5^clv mutation showed mild skeletal defects not seen in wild-type or +/Bmp5^null mice, including reduction of the spinous process at the second thoracic vertebrae (data not shown). Such defects are consistent with a mild dominant-negative effect observed when the Bmp5^clv allele is present in single copy, consistent with the mode of action observed for similar mutations in other TGF-β members 4547.

We expect homozygosity for the mutation to further decrease the activity of BMP5 and increase the production of the non-processed BMP5 molecules that may inactivate other coexpressed BMPs. To determine the effect of this mutation on homozygotes carrying this allele, we crossed heterozygous carriers of the Bmp5^clv mutation and generated viable Bmp5^clv/Bmp5^clv homozygotes, but at rates ~10-times lower than Mendelian predictions (4/156 progeny, P < 0.001). Despite the increased prenatal lethality, some Bmp5^clv homozygotes survived with normal life spans and fertility. Among these surviving homozygous Bmp5^clv mice, we noted more severe defects than those seen with age-controlled Bmp5^null homozygotes, including shorter external ears (wild-type: 6.4 ± 0.4 mm, null: 4.8 ± 0.2 mm, Bmp5^clv: 2.9 ± 0.2 mm; P < 0.01), loss of lesser horns of the hyoid (Fig. 2a), more misshapen xiphisternum and missing ribs (Fig. 2f), less calcification of thyroid cartilage (Fig. 2a), abnormal bony fusion (fused posterior sternum; Fig. 2f), and reduced or absent processes on the sixth cervical (Fig. 2b), second thoracic (Fig. 2d), and lumbar vertebrae (Fig. 2e). The spectrum of phenotypes, and the consistent reduction rather than overgrowth of skeletal tissue, both suggest that the Bmp5^clvmutation leads to loss rather than gain of BMP5 activity.

An interesting new phenotype in Bmp5^clv mice is marked reduction or complete elimination of the deltoid crest on the humerus bone (Fig. 2c). The deltoid crest is a prominent bony ridge that is the insertion site for the deltoid muscle, and it normally forms in response to mechanical interaction between muscle and bone. In paralyzed animals or those exhibiting genetically defective muscle formation, the deltoid crest does not form 4950. In contrast, mutant animals showing abnormal increases in muscle mass develop bigger deltoid tuberosities 51.

The deltoid muscle remained present in Bmp5^clv/Bmp5^clv mutants, suggesting that the deltoid crest defect was likely due to changes in the humerus bone. In situ hybridization showed Bmp5 expression in the developing deltoid crest of the humerus (Fig. 3b), and coexpression of the Bmp2 and Bmp6 genes in similar regions (Fig. 3c, d). The coexpression of multiple Bmps may explain why the deltoid crest is only mildly reduced in Bmp5^nullmice but completely eliminated in Bmp5^clv mice. The expression of multiple Bmp genes at the deltoid crest and the defect of this structure in Bmp5^clv mutants suggest that Bmp signaling is important at this mechanosensitive site.

The loss of a prominent muscle-induced skeletal feature in Bmp5^clv mice suggests that Bmp signaling plays a key role in bone cells' response to mechanical activity. To test this hypothesis, we isolated osteoblasts from the deltoid tuberosity of 10-month-old wild-type and Bmp5^clv mice and subjected the cells to 24 hours of cyclic uniform radial strain in culture. The stretch regimen we applied (10-second maximum 15% elongation, then 10-second relaxation, frequency 0.05 Hz or 3 cycles/minute) is similar to the mechanical stimulation known to induce cellular responses in cultured osteoblasts 1819. After 24-hour cyclic strain, control osteoblasts became spindle-shaped, showed elongation of cellular processes, and were largely oriented perpendicular to the radial strain field (Fig. 4 and Table 1). In contrast, osteoblasts from the deltoid tuberosity of the Bmp5^clv mice displayed no significant changes in morphology or orientation after mechanical strain (Fig. 4 and Table 1), suggesting that the defect in BMP signaling blocked normal response of bone cells to mechanical stimulation.

Osteoblasts cultured from an independent bone-muscle interaction site that does not show morphologic defects in Bmp5^clv mice responded normally to mechanical strain in vitro (femur trochanter osteoblasts; Fig. 4). The anatomic site-specificity of osteoblast response to mechanical stimuli in Bmp5^clv mice is consistent with the specificity of the skeletal defects seen at the tissue level of these mutants. Furthermore, muscle fibroblasts isolated from the deltoid of both wild-type and Bmp5^clv mice showed normal changes in morphology and orientation after mechanical stimulation in vitro (Fig. 4 and Table 1), suggesting that the mutation primarily affects bone cells and not muscle cells at these sites.

To further characterize the relationship between mechanical stimulation and BMP signaling, we studied the effect of mechanical strain on cellular translocation of SMAD proteins, key transcription factors that translocate from the cytoplasm to the nucleus upon activation of BMP receptors 52. Non-strained wild-type deltoid tuberosity osteoblasts exhibited SMAD1/5 immunoreactivity predominantly in the cytoplasm (Fig. 5 and Table 1). Within 30 minutes of applying cyclic strain to these cells, we detected a significant increase in nuclear localization of SMAD1/5 (Fig. 5 and Table 1). Peak nuclear localization occurred 1 hour after mechanical stimulation, SMAD1/5 was then predominantly cytoplasmic again by 24 hours (Fig. 5 and Table 1), at which point cells also had reoriented perpendicularly to the strain axis. In contrast, SMAD1/5 in Bmp5^clv deltoid tuberosity cells remained mostly cytoplasmic before and after cyclic stretch (Fig. 5 and Table 1), demonstrating that perturbation of BMP signaling by the Bmp5^clv mutation disrupts bone cells' rapid response to mechanical stretch.

While mutant osteoblasts from deltoid crest displayed altered response to mechanical stimuli, it was unclear whether this was due to an ongoing requirement for BMP-mediated signaling or abnormal cell development at the deltoid crest of Bmp5^clv mice. To distinguish between these possibilities, we tested the effect of transiently inhibiting BMP signaling in wild-type osteoblasts cultured with increasing concentrations of noggin, a secreted protein that binds BMP and inhibits its activity 53. Incubation of adult wild-type osteoblasts with noggin produced dose-dependent decreases in their reorientation response to mechanical stimulation (Table 1), confirming that BMP signaling is important in maintaining the normal mechanical responses of osteoblasts at postnatal stages.

The Bmp5^clv mutation was originally designated "se-^4J" when discovered as a spontaneous mutation in a backcross between C57Bl/6J and B6.Cg-Otop1^tlt (for further strain details, see JAX stock number 001496). The Bmp5^clv stock was maintained in the laboratory of DMK by intercrossing Bmp5^clv/+ heterozygotes. Mutants were identified by the short ear phenotype, and their genotype was confirmed by molecular typing as described in this paper. The classical se mutation was also maintained on the C57Bl/6J background (Jackson Laboratory stock number 000056). Skeletons from age- and sex-matched mice from different genotypes were fixed in 95% ethanol, processed using a potassium hydroxide/alizarin red (Sigma) staining procedure 58, disarticulated, and analyzed.

The Bmp5 open reading frame from total lung RNA of a 1-month-old C57Bl/6J male mouse and an age- and sex-matched Bmp5^clv/Bmp5^clv mutant mouse was amplified. Reverse transcription was carried out with Superscript RT (Gibco) using 1.5 μg of RNA and 2.5 μL of 20-μM reverse primer (5'-CGC GGA TCC CTA GTG GCA GCC ACA CGA-3') in a 20-μL reaction at 37°C for 1 h. Polymerase chain reaction (PCR) was performed with Amplitaq DNA polymerase (Perkin-Elmer) in a 50-μL reaction containing 1 μL each of 20-μM forward (5'-CGC GGA TCC ACC ATG CAT TGG ACT GTA TTT TTA C-3') and reverse (as above) primers and 1 μL of the RT reaction product from above. The PCR products were purified from a 1% regular agarose gel with the Gene-clean kit (Bio 101), digested with BamHI, and cloned into pBluescript II SK(+) (Stratagene). The Bmp5 cDNA insert on both strands from multiple clones of each genotype was sequenced with the Sequenase kit (United States Biochemicals) and primers that span the Bmp5 open reading frame (primer sequences available upon request).

The genomic region affected by the Bmp5^clv mutation was amplified with primers 1 (5'-AAA TCT GCT GGT CTT GTG GG-3') and 2 (5'-GGG TCC TGA TGA GAG TTG GA-3') in a 50-μL PCR reaction containing 2.5 μL each of 20-μM primers 1 and 2 and 1.5 μL of genomic DNA. The amplified products were digested by TaqI and separated by electrophoresis on a 3% low melt agarose gel.

COS-7 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS, Hyclone), 2-mM GlutaMax, 0.1-mM non-essential amino acids, penicillin G (100 U/mL), and streptomycin (100 μg/mL) at 37°C in 5% CO₂/95% air and were passaged at confluency and maintained according to standard tissue culture practices. All tissue culture reagents were from Gibco except as indicated otherwise.

The Bmp5 open reading frame was amplified from wild-type or Bmp5^clv homozygous mice as described above, subcloned into the mammalian expression vector pCDNA3 (Invitrogen), and sequenced to ensure that no mutations were introduced in the cloning steps. In transfection experiments, 2 × 10⁶ COS-7 cells were plated on a 100-mm tissue culture dish (Corning) and transfected with 3.3 μg of DNA and 10 μL of lipofectamine (Gibco) per dish under reduced serum condition for 7 h (as instructed by the manufacturer). The transfected cells were allowed to recover in full medium for 5 h, at which point the normal medium was replaced with reduced serum medium (growth medium with 1% FBS, 3 mL/plate). After ~60 h, the conditioned media from 2 identically treated plates were collected and pooled, clarified by centrifugation, made 2% SDS by addition of a 10% SDS stock, boiled for 10 min, and aliquoted and stored at -20°C.

Recombinant protein containing murine BMP5 pro region (amino acids 1–310) fused with thioredoxin was expressed under the IPTG-inducible pET32 system in BL21 E. Coli bacterial cells as instructed by the manufacturer (Novagen). Briefly, 8 L transformed bacterial culture was grown to a density of OD600 ~0.5 and induced with 1-mM IPTG for 4 h at 37°C. Cells were collected by centrifugation and resuspended in Buffer A (6-M Guanidine HCl, 0.1-M NaH2PO4, 0.01-M Tris pH 8.0, 5 mL/g cell pellet weight), and the clarified supernatant was incubated with Ni-NTA nickel resin beads (Qiagen) for 1 h at room temperature (RT). The beads were washed with 10 volumes of Buffer A, 10 volumes of Buffer B (8-M urea, 0.1-M NaH2PO4, 0.01-M Tris pH 8.0), 10 volumes of Buffer C (8-M urea, 0.1-M NaH2PO4, 0.01-M Tris pH 6.3) with 5-mM imidazole, 10 volumes of Super Buffer C (Buffer C plus 250-mM NaCl, 0.5% Tween-20, 10-mM β-mercaptoethanol, 100-mM imidazole), and 2 washes of 10 volumes of Buffer C with 20-mM imidazole. Bound protein was eluted by boiling the beads in 2.5× Laemmli buffer. Polyclonal antibodies against the BMP5 pro domain (anti-PRO) were raised by immunizing rabbits with recombinant BMP5 pro region protein at the Berkeley Antibody Company. Monoclonal antibodies against the human BMP5 mature domain (anti-MAT) were provided by Genetics Institute, Inc., and were shown to cross-recognize the highly conserved murine BMP5 mature region.

For western blot analysis, 60 μL of the processed SDS samples were boiled for 5 min in reducing Laemmli buffer, fractionated on a 15% SDS-polyacrylamide gel, and transferred onto nitrocellulose. The membrane was incubated at RT for 3 h in blocking solution (5% non-fat dried milk in 150-mM NaCl, 10-mM Tris pH 8.0, and 0.05% Tween-20), probed at RT for 12 h with anti-BMP5 primary antibodies (anti-PRO 1:667 or anti-MAT 1:172) diluted in blocking solution, and incubated for 1 h at RT in blocking solution containing donkey anti-rabbit-IgG coupled to horseradish peroxidase diluted 1:5000 (anti-PRO) or sheep anti-mouse-IgG coupled to horseradish peroxidase diluted 1:1000 (anti-MAT). Signals on the blot were detected by the ECL system (Amersham).

Forelimbs were dissected from E13.5 CD1 embryos, frozen immediately in O.C.T. embedding medium (VWR), and stored at -80°C. Twelve-micrometer cross sections of the humerus at the level of the deltoid tuberosity were collected with a cryostat microtome and were processed, prehybridized, and hybridized with digoxygenin-labeled cRNA probes, as described in previously published protocols 40. Bmp probes were generated by in vitro transcription of linearized constructs containing 200- to 500-bp DNA inserts from the pro region or untranslated region of Bmp genes as described previously 59. Descriptions of the washing conditions and the methods used to detect the labeled signal have been published previously 40.

Primary osteoblast and muscle fibroblast isolates from sex-matched 10-month-old C57Bl/6J and Bmp5^clv mice were established. Under microscopy, the humerus was exposed and the surrounding musculature was dissected off the bone. A ~3-mm segment of the humerus containing the deltoid tuberosity (wild type) or the deltoid muscle attachment site (Bmp5^clv) was removed. Similarly, a ~5-mm segment of the femur shaft containing the femoral trochanter was dissected from each mouse. The marrow cavity of the bone segments was flushed with ice cold phosphate-buffered saline, and the bone was rinsed in phosphate-buffered saline containing penicillin G (200 U/mL), streptomycin (200 μg/mL), and amphotericin B (0.5 μg/mL). The rinsed bones were minced into small pieces with a bone cutter in 2 mL dissociation medium (21.3-mM Tris pH 7.4, 111.2-mM NaCl, 5.4-mM KCl, 1.3-mM MgCl2, 0.5-mM ZnCl2, and 13-mM glucose supplemented with type II collagenase [5 U/mL], elastase [6.25 U/mL], D-sorbitol [18.22 mg/mL], and chondroitin sulfate [6 mg/mL]) and digested at 37°C for up to 180 min. At 10, 20, and 30 min and every 30 min thereafter, the dissociated cells were removed with the medium and added to 1 volume of FBS (Invitrogen), and a fresh 2 mL of dissociation medium was added to the bone fragments. Cells from the 10–20 min, 30–90 min, and 120–180 min time points were pooled and recovered by centrifugation, resuspended in osteoblast growth medium (αMEM with 10% FBS, 10% horse serum [Invitrogen], 2-mM GlutaMax, and 0.1-mM nonessential amino acids) supplemented with penicillin G (100 U/mL), streptomycin (100 μg/mL), and amphotericin B (0.25 μg/mL), and plated in a 175-cm² tissue culture flask. The plated cells were allowed to attach for 72 h at 37°C in 5% CO₂/95% air, and non-adherent cells were rinsed off. Fresh growth medium with antibiotics and antimycotics was replaced every 4 days until the osteoblasts reached confluency (typically ~14 days after initial plating), when they were passaged and maintained similarly afterwards. An aliquot of cells from each flask was plated separately at first passage and assayed for alkaline phosphatase activity, an indicator of osteogenic differentiation, using the AP assay kit (Sigma). Cultures containing fewer than 50% alkaline phosphatase-positive cells were discarded. Antibiotics and antimycotics were omitted from the growth medium after the second passage. Experiments were typically performed on cells from the second and third passages.

A previously described fibroblast isolation protocol 60 was used to harvest primary fibroblasts from the distal third of the deltoid muscle by enzymatic dissociation. The fibroblasts were maintained in DMEM with 15% FBS, 2-mM GlutaMax, and 0.1-mM non-essential amino acids supplemented with penicillin G (100 U/mL), streptomycin (100 μg/mL), and amphotericin B (0.25 μg/mL).

A total of 5 × 10⁴ osteoblasts or 1 × 10⁴ muscle fibroblasts per well were plated on BioFlex 6-well plates with collagen 1-coated flexible membrane bottoms (Flexcell) and allowed to attach overnight. For noggin treatment, fresh medium containing no (control) or recombinant mouse noggin protein (R&D, 0.1–10 μg/mL) was replaced 3 h before mechanical strain. The plates were mounted onto the base plate of a Flexercell FX-4000T system (Flexcell) and left undisturbed for 1 h before initiation of mechanical stimulation to minimize the extraneous strain and fluid stress secondary to handling of the plates.

Cyclic uniform radial strain was applied with -70 kPa negative pressure to produce a 15% maximum elongation of the membranes in square wave cycles of 10 sec strain and 10 sec relaxation at a frequency of 0.05 Hz or 3 cycles per minute.

We visualized the cells using a previously described indirect immunofluorescence protocol 60 with the following modifications. The cells were fixed in ice-cold 50% methanol/50% acetone at -20°C for 20 min; blocked in 10% serum with 0.5% Triton-X (Biorad); and probed with primary antibodies against type I collagen (Chemicon), vimentin (Sigma), or SMAD1/5 (Chemicon) and secondary antibodies coupled to FITC or Cy3 (Sigma). Nuclei were counterstained with propidium iodide (Molecular Probes). The round flexible membrane was cut into equal quadrants and each quadrant was mounted onto a slide. This allowed the angles between the cell axis and the strain field to be measured accurately. Angles were measured from 80–150 cells per condition over 3 independent experiments, and SMAD subcellular localization was determined from 80–150 cells per condition over 3 independent experiments. Images were captured with a Retiga 1300 digital camera (QImaging) attached to a Leica DM IRB microscope and were processed with Northern Eclipse 6.0 software (MVIA). P values and statistical significance between the means of each group were determined by the modified Bonferroni's t test.