Ligand- and Stage-Dependent Divergent Functions of BMP Signaling in the Differentiation of Embryonic Skeletogenic Progenitors In Vitro



Bone morphogenetic proteins (BMPs) are key molecules in the differentiation of skeletal tissues. We have investigated whether differentiation of limb embryonic mesodermal progenitors into different connective tissue lineages depends on specific stimulation of distinct BMP ligands or on the differential response of target cells to a common BMP stimulus. We show that Bmp2,4,5,7 and Gdf5 exhibit differential expression domains during the formation of tendons, cartilages, and joint tissues in digit development, but their respective effects on digit progenitors cell cultures cannot sustain the divergent differentiation of these cells into tendons, joints, and cartilage. However, the influence of BMPs differs based on the culture length. Early cultures respond to any of the BMPs by inducing chondrogenic factors and inhibiting fibrogenic and osteogenic markers. Later, a second phase of the culture occurs when BMPs attenuate their prochondrogenic influence and promote the fibrogenic marker Scleraxis. At advanced culture stages, BMPs inhibit prochondrogenic and profibrogenic markers and promote osteogenic markers. The switch from the prochondrogenic to the profibrogenic response appears critically dependent on the basal expression of Noggin. Thus, the differential regulation of Scleraxis at these stages was abrogated by treatments with a BMP-analogous compound (AB204) that escapes NOGGIN antagonism. Gene regulation experiments in absence of protein synthesis during the first period of culture indicate that BMPs activate at the same time master chondrogenic and fibrogenic genes together with cofactors responsible for driving the signaling cascade toward chondrogenesis or fibrogenesis. Gene-silencing experiments indicate that Id2 is one of the factors limiting the profibrogenic influence of BMPs. We propose that connective tissues are dynamic structures composed of cartilage, fibrous tissue, and bone that form in successive steps from the differentiation of common progenitors. This sequential differentiation is regulated by BMPs through a process that is dependent on the basal expression of BMP cofactors or signaling modulators. © 2014 American Society for Bone and Mineral Research.


Unraveling the factors responsible for the differentiation and for the maintenance of the phenotype of connective tissues is a major challenge in regenerative medicine.[1] In the embryo, connective tissues, such as cartilage, tendons, or bone, are originated from mesoderm or neural crest mesenchyme through a differentiation process regulated by secreted growth factors.[2] In the adult, differentiated cells retain considerable plasticity and can lose differentiation when transferred to culture conditions,[3] or they can even undergo metaplastic transformations in a variety of pathological conditions.[4] Hence, a better understanding of connective tissue differentiation will be valuable in regenerative medicine for the development of new therapeutic treatments for orthopedic and sports medicine.

Bone morphogenetic proteins (BMPs) are major regulators of the differentiation of specialized connective tissues. BMPs were initially identified because of their potential to induce ectopic bone in soft tissues,[5] and exogenous BMPs induce bone regeneration in stillborn mice subjected to digit amputation.[6] The role of BMPs in the differentiation and phenotypic maintenance of cartilage is evident from the presence of cardiovascular and skeletal alterations caused by genetic mutations that affect the BMP signaling cascade in mice and humans[7-11] and from results obtained from in vitro studies using mesenchymal stem cell cultures.[12, 13] BMPs also have been implicated in tendon differentiation.[14-18] Clinical and preclinical attempts to employ BMPs for the treatment of bone, cartilage, and tendon injuries are numerous.[19-24] However, the mechanisms whereby distinct BMP ligands induce lineage-specific (bone, cartilage, or tendon) commitment of connective tissue progenitors are not well understood (see Berasi and colleagues[15]). Such information is critical to realizing the potential of individual BMPs as therapeutic agents and for designing appropriate clinical indications. Considering that canonical BMP signaling is mediated by phosphorylation and nuclear translocation of a common set of intracellular Smad proteins (Smad-1, -5, and -8 in conjunction with Smad-4), the differences in the response of BMP target cells are thought to reside on the intensity of the signal. In fact, BMP ligands, which signal through different receptors, exhibit similar osteoinductive influences on bone marrow mesenchymal stem cells.[25] However, other mechanisms, such as the presence of cofactors in the extracellular matrices[26, 27] or synergistic interactions with other signaling pathways,[2, 28-30] can also be important. The number of factors able to modify the intensity of the BMP signal in a target tissue is extraordinarily high,[31] and include at least the following: 1) the affinity of different ligands for BMP receptors; 2) the availability of BMP ligands in the extracellular space that is regulated by extracellular BMP-binding molecules; 3) the presence of ligand-sequestering pseudo-receptors in the target cells; and, 4) intracellular partner molecules, including different miRNAs, that regulate transcriptional function, degradation, or nuclear shuttling of Smad proteins. This complex regulation of BMP signaling hinders our ability to clarify the basis for the functional diversification and versatility of BMPs in the differentiation of connective tissues.[32] Hence, to evaluate the potential clinical benefits of different BMP ligands in the differentiation of stem cells for regenerative medicine, we need to understand the molecular mechanisms underlying their biological activity.

Here, we have analyzed the function of BMPs in the differentiation of mesodermal connective tissue progenitors obtained from the embryonic limb. For this study, we selected the BMP ligands that are expressed during digit development and correlated their transcriptional influence in vitro with their pattern of expression during normal development. In the developing limb, mesodermal progenitors originating from the lateral mesoderm differentiate into phalanges (cartilaginous differentiation occurs first, followed by bone), interphalangeal joints, and digit tendons. BMP genes, including Bmp2, Bmp4, Bmp5, Bmp7, and Gdf5, are highly expressed in the limb autopod (the region of the embryonic limb that forms the digits; see Zuzarte-Luis and colleagues[33] and Geetha-Loganathan and colleagues[34] and references therein), and their role in the formation of the digit tissues is supported by genetic approaches.[35-42] Some evidence suggests that different BMP ligands may exert distinct effects on the differentiation of limb mesodermal progenitors;[12] however, we do not yet know how they contribute to the establishment of divergent cell fates in the mesodermal progenitors.

Materials and Methods

In this work, we employed Rhode Island chicken embryos from day 4.5 to day 8 of incubation (id) equivalent to stages 24 to 32 HH.

In situ hybridization

Expression of Bmp2, Bmp4, Bmp5, Bmp7, Gdf5, and Scleraxis (Sclx) during digit development was analyzed by in situ hybridization in 100 µm vibratome sections of the embryonic limb autopod. Samples were treated with 10 µg/mL of proteinase K for 20 minutes at 20°C. Hybridization with digoxigenin-labeled antisense RNA probes was performed at 68°C. Alkaline phosphatase-conjugated anti-digoxigenin antibody (dilution 1:2000) was used (Roche Applied Bioscience, Indianapolis, IN, USA). Reactions were developed with BM Purple AP Substrate (Roche).

Micromass mesodermal cultures

Progenitor mesodermal cells of the digit tissues were obtained from the “progress zone” region located under the apical ectodermal ridge of chick leg buds of embryos at 4.5 id (25 HH). Cells were dissociated and suspended in medium DMEM with 10% fetal bovine serum (FBS), 100 units/mL penicillin and 100 µg/mL streptomycin. In long-term cultures, 50 µgr/mL of ascorbic acid were added to the medium from day 7 of culture. Cultures were made by pipetting 10 µL drops of cell suspension at a density of 2.0 × 107 cells/mL into each well of a 24-well plate. The cells were allowed to attach for 2 hours and then 200 µL of medium was added. We employed these cultures for analyzing the effects of adding BMPs on gene regulation.

The panel of selected genes analyzed in the study included well-known markers (Supplemental Table S1) of cartilage differentiation (Sox9, Col2α1); fibrogenic markers (Scleraxis; Col1α1); bone differentiation markers (Alkaline Phosphatase, liver/bone/kidney; ALPL, and Runt-related transcription factor 2; Runx2); secreted growth factors implicated in connective tissue differentiation (Tgfβ2; Gdf5; Activin βα); type I BMP receptors (Alk1/Acvr-l1; Alk2/Acvr1; Alk3/Bmpr1a; Alk6/Bmpr1b); and extracellular BMP modulators (Noggin; Chordin (Chd); Chordin-like 1 (Chdl-1/Ventroptin); Chordin-like 2 (Chdl-2); Twisted Gastrulation (Tsg; Twsg 1); Dan (differential screening-selected gene aberrative in neuroblastoma); Bmper (BMP binding endothelial regulator; crossveinless 2); Sost (Sclerostin); SostDc1 (Sclerostin domain containing-1; Uterine sensitization associated gene-1; Wise; Ectoidin); Follistatin (Fst); Follistatin-like 1(Fst-l1); Tolloid like 1 (Tll1; Colloid).


The effects of BMP2, BMP4, BMP5, BMP7, and GDF5 were analyzed by adding recombinant protein to the micromass culture medium. Treatments were maintained for a 1-hour (short-length) or 6-hour (medium-length) period. After testing different protein concentrations, we selected doses of 200 ng/mL of each BMP, which caused a similar upregulation of Sox9. We employed human recombinant BMP2 (Peprotech, Lodon, UK); human recombinant BMP4 (R&D Systems, Minneapolis, MN, USA); human recombinant BMP5 (R&D Systems); human recombinant BMP7 (R&D Systems); and human recombinant GDF5 (R&D Systems).

To explore the influence of endogenous NOGGIN on the transcriptional effects of BMPs, we employed the BMP ligand analogous AB204 (200 ngr/mL). This compound was engineered by RASCH (Random Assembly of Segmental Chimera and Heteromer) and has the functional properties of BMP2 but escapes from the antagonism of NOGGIN.[43]

Histological analysis of tissue differentiation

Morphological analysis of the developing digits was performed in samples fixed in trichloroacetic acid and whole-mount stained with alcian blue. The specimens were next embedded in paraffin wax, sectioned at 10 µm, and stained with hematoxylin and eosin.

The sequence of differentiation of the micromass cultures was established by histological analysis. For this purpose, the cultures were fixed in 4% paraformaldehyde (PFA) in 0.1 M phosphate-buffered saline (PBS) for 10 minutes at room temperature, washed in PBS, dehydrated in acetone, and embedded in araldite. Semithin sections (1 µm) were stained with toluidine blue and examined in the microscope.

Cellular and extracellular matrix modifications in the treated cultures were monitored by immunolabeling with SOX9 (Calbiochem, San DIego, USA) and Tenascin (M1B4, Developmental Studies Hybridoma Bank, University of Iowa, Des Moines, IA, USA) antibodies, as positive markers of skeletal progenitors and tendon matrix, respectively.[44]

Cell nucleofection

The function of Id2 in differential transcriptional effects of BMP signaling was analyzed through a loss-of-function approach. For this purpose, limb mesodermal cells were transfected with a short hairpin RNAi for Id2 (ShId2) cloned into the pcU6-1-shRNA (a generous gift of Dr Tim J Doran) as described by Wise and colleagues.[45] The limb mesodermal cells were electroporated using the Eppendorf Multiporator system (Eppendorf, Manborg, Germany) following the manufacturer's instructions. The level of gene silencing was evaluated by Q-PCR. Control transfections using a chU6-1v irrelevant control plasmid were performed in all experiments.

Western blot

Analysis of the intensity of BMP signaling was evaluated by western blot. For this purpose, total protein extracts were obtained from control and BMP-treated mesodermal cultures. Cell lysis was performed with RIPA buffer (150 mM NaCl, 1.5 mM MgCl2, 10 mM NaF, 10% glycerol, 4 mM EDTA, 1% Triton X-100, 0.1% SDS, 1% deoxicolate, 50 mM Hepes, pH 7.4) supplemented with the protease inhibitors phenylmethylsulfonyl fluoride (PMSF, 1 mM), leupetin (10 µg/mL), and aprotinin (10 µg/mL) for 15 minutes on ice. The cell lysates were clarified of cellular debris by centrifugation (13,200g) for 10 minutes at 4°C. Proteins were separated by 10% polyacrilamide gel electrophoresis containing 0.1% SDS and transferred to polyvinylidene fluoride (PVDF) membrane (Bio-Rad, Hercules, CA, USA). The membranes were incubated for 1 hour at room temperature in bovine serum albumin (BSA) and incubated overnight with the following antibodies: Rabbit polyclonal antibody against phospho-SMAD1/SMAD5/SMAD8 (Ser463/465; Cell Signaling, Danvers, MA, USA); Rabbit monoclonal antibody against Phospho-p38 MAPK (Thr180/Tyr182; Cell Signaling); and mouse polyclonal antibody anti-tubulin (T5168, Sigma, St. Louis, MO, USA). Protein bands were detected with an ODISSEY infrared-imaging system (Li-Cor Bioscience, Lincoln, NE, USA) according to ODISSEY western blot protocol. Immunoblots were developed with anti-mouse IRDye 800DX or anti-rabbit IRDye 680DX (Rockland Immunochemicals USA, Gilbertsville, PA, USA). The public domain image analysis program ImageJ, was employed to perform quantitative analysis.

Real-time quantitative PCR (Q-PCR) for gene expression analysis

In each experiment, total RNA was extracted and cleaned from specimens using the RNA Mini Kit (Bioline, Taunton, MA, USA). RNA samples were quantified using a spectrophotometer (Nanodrop Technologies ND-1000, Wilmingston, DE, USA). First-strand cDNA was synthesized by RT-PCR using random hexamers and M-MulV reverse transcriptase (Fermentas, GmbH, St. Leon-Rot, Germany). The cDNA concentration was measured in a spectrophotometer (Nanodrop Technologies ND-1000) and adjusted to 0.5 µg/µL. Q-PCR was performed using the Mx3005P system (Agilent, Santa Clara, CA, USA) with automation attachment. In this work, we have used SYBRGreen (Applied Biosystems, Carlsbad, CA, USA) based Q-PCR. Gapdh had no significant variation in expression across the sample set and therefore was chosen as the normalizer in our experiments. Mean values for fold changes were calculated for each gene. Each value in this work represents the mean ± SEM of at least three independent samples obtained under the same conditions. Samples consisted of four micromass cultures. Primers for Q-PCR are included in Supplemental Table S2.


Data were analyzed using MANOVAs (gene × treatment) followed by Bonferroni tests for post hoc comparisons or Student's t test for gene expression levels. Statistical significance was set at p < 0.05. All the analyses were done using SPSS for Windows version 18.0 (SPSS Inc., Chicago, IL, USA).


Bmp gene expression during the specification of connective tissues in the embryonic digits

Digits develop in the autopod between 5 and 8.5 id under the influence of BMP signaling.[34] Each digit is composed of a rod of differentiating cartilage with extensor and flexor tendon primordia differentiating along their dorsal and ventral surface, respectively (Fig. 1A–D and A′–D′). The cartilage exhibits local discontinuities at the zones of joint formation (arrows in Fig. 1A), which are considered to be regions in which cartilage differentiation is reverted to form the joint components.

Figure 1.

Histology and in situ hybridizations in longitudinal and transverse sections of the chick developing autopod in representative stages of digit formation (id 6.5, A–N; id 8, A′–N′). (A) Whole-mount staining with alcian blue to show the discontinuity of cartilage staining in the metatarso-phalangeal joints (arrows). (B) Transverse section of the autopod showing the cartilaginous digit skeleton (blue) encompassed by the perichondrium (red) and the associated extensor and flexor tendon condensations (T). (C, D) In situ hybridizations in sections of the autopod showing the expression of Scleraxis (Sclx) in the developing tendons. (E, F) Bmp5 expression is specifically distributed in the region subjacent to the ectoderm. (G–J) transcripts of Bmp7 (G, H) and Bmp2 (I, J) are specifically distributed in the core of the interdigits. Note also the positive expression of Bmp2 in the developing joints (arrows in I). (K, L) Expression of Bmp4 in longitudinal (K) and transverse (L) sections of the autopod; the transcripts are concentrated in the tendon blastemas (arrows) and interdigits. Note that interdigital expression occupies the border area between the Bmp5 and Bmp7/2 expressions. (M, N) Expression of Gdf5 in prominent joint domains that extend into the neighboring tissue surrounding the joints. (A′–N′) Longitudinal and transverse sections of the autopod and digits at id 8 show changes in the pattern of Bmp gene expression at advanced stages of digit formation. Note that Bmp5 (F′), Bmp7 (H′), Bmp2 (J′), and Bmp4 (L′) are all expressed, at variable levels, in the contour of the differentiating tendons. Note also the intense domains of Bmp4 at both sides of the differentiating phalanges (L′) and the presence of perichondrial domains of Bmp5 in the differentiating phalanges (arrow in E′).

As shown in Fig. 1E–N, during the early stages of digit formation, the BMP ligands in the autopod exhibit a striking pattern of expression suggestive of differential functions. Bmp5 (Fig. 1E, F) is expressed in the subectodermal mesoderm, which is the region where the tendon blastemas develop. Bmp7 (Fig. 1G, H) and Bmp2 (Fig. 1I, J) are expressed in the core of the interdigits. Bmp4 (Fig. 1K, L) is expressed in the interface between the interdigital domains of Bmp2/7 and Bmp5. Additionally, Bmp2 is also expressed in joint domains, and Bmp4 is expressed in the tendon blastemas (arrows in Fig. 1I–L). Gdf5 (Fig. 1M, N) is expressed in the developing joints, forming intense domains that extend to the mesoderm located around the digit where ligaments and tendon enthesis develop.

Once the digits are established (Fig. 1A′–N′), the interdigital tissue undergoes regression and the patterns of expression of each BMP ligand become more similar to each other, except for specific domains of Gdf5 (Fig. 1M′) and Bmp2 (Fig. 1I′) in the developing joints, and Bmp5 domains in the perichondrium of the diaphysis of the differentiating phalanges (Fig. 1E′). As illustrated in Fig. 1K′, L′, Bmp4 shows prominent domains along the margins of the digits in zones of differentiating connective tissue through which run digital blood vessels. In addition, Bmp2, Bmp4, Bmp5, and Bmp7 show tenuous expression domains around the differentiating tendons with a predominant distribution of Bmp2 and Bmp7 in the extensor tendons.

In conclusion, these results demonstrate that various BMPs are expressed in the connective tissues of the digits, implying that they play a role in differentiation of connective tissue progenitors.

Tissue differentiation in micromass cultures of mesenchymal digit precursors

The transcriptional effect of the different BMPs was analyzed in high-density micromass cultures of digit mesodermal progenitors. The micromass culture provides a reliable procedure to analyze the differentiation of limb mesoderm into cartilage and associated connective tissues.[46, 47] After histological examination of the micromasses, we selected three differentiation stages within the duration of culture for BMP treatment (Fig. 2). In the first period (covering the first 2 days of culture; Fig. 2A, A′), the tissue appears as undifferentiated mesenchyme with occasional nodular regions of increased cell condensation. In the second stage (days 3 to 7 of culture; Fig. 2B, B′), an increasing number of cartilage nodules are identified (arrows in Fig. 2B) and a fibroblastic-like tissue occupies the space between nodules. In the third stage (days 11 and 12 of culture; Fig. 2C, C′), the cartilage appears hypertrophic. Remarkably, the hypertrophic cartilage is covered by a layer of flattened cells that have a perichondrial-like morphology (arrowheads in Fig. 2C′). Fibroblastic tissue at this stage appears in zones of discontinuity of the cartilage plate and in the periphery of the culture.

Figure 2.

(A–C) Stages of differentiation of micromass cultures of digit progenitors. Low (A–C) and high (A′–C′) magnification images of semithin sections at days 2 (A, A′), 7 (B, B′), and 12 (C, C′) of culture stained with toluidine blue. Arrows indicate the presence of chondrogenic nodules in B. Note the hypertrophyc appearance of the cartilage in C and the presence of a perichondrial-like superficial layer of flattened cells (arrowheads) and internodular connective tissue (ct). (D) SOX9 expression and Tenascin-positive matrix in control and BMP-treated micromass cultures at days 2 and 4 of culture. Note increased immunolabeling for SOX9 and decreased distribution of Tenascin in micromasses treated at day 2 in contrast with the absence of significant modifications in treatments performed at day 4.

These observations confirm that the micromass culture assay replicates the differentiation events of mesodermal progenitors occurring in vivo.

Comparative analysis of connective tissue lineage determination by addition of distinct BMP ligands

The regulation of specialized connective tissue gene markers (cartilage, tendon, joint, and bone tissues; Supplemental Table S1) was studied in 2-day cultures subjected to 6-hour treatment with BMP2, BMP4, BMP5, BMP7, or GDF5. Because the effect of BMP ligands is dose-dependent, we selected the doses for each BMP based on the dose that regulated a target gene (Sox9) to a set level. As shown in Table 1, markers of chondrogenic differentiation (Sox9, Col2a1, Bmpr1b, Fz4, and Activin βα) were upregulated by all BMP ligands. The strong upregulation of Activin βα (between 10-fold and 16-fold) is remarkable; Activin βα is the earliest marker of the digit cartilage primordial.[48] In contrast with these effects, fibrous tissue (Scleraxis), bone (Runx2 and ALPL), and joint (Gdf5 and Wnt14) markers were all downregulated.

Table 1. Expression of Connective Tissue Lineage Markers 6 Hours After the Addition of BMP2, BMP4, BMP5, BMP7, and GDF5 to 2-Day Micromass Cultures of Digit Progenitors in the Second Day of Culture
  • *p < 0.05;
  • **p < 0.01;
  • ***p < 0.001.
Sox95.65 ± 0.65**5.47 ± 1.11*5.68 ± 1.00**3.83 ± 0.29***4.77 ± 0.47**
Bmpr1B3.83 ± 0.48*5.40 ± 0.37**4.78 ± 0.65*3.56 ± 0.43***2.98 ± 0.59*
Col2α14.04 ± 0.58*9.62 ± 1.85*6.58 ± 1.29**2.15 ± 0.12***2.35 ± 0.41*
Fzd42.75 ± 0.33*4.05 ± 0.58*3.17 ± 0.41*2.25 ± 0.23*2.52 ± 0.48*
Scleraxis0.14 ± 0.01*0.22 ± 0.02***0.44 ± 0.08**0.20 ± 0.02***0.53 ± 0.10**
Col1α10.72 ± 0.030.89 ± 0.111.23 ± 0.150.99 ± 0.030.85 ± 0.07
Wnt140.34 ± 0.05**0.61 ± 0.02***0.95 ± 0.100.57 ± 0.08**0.41 ± 0.05***
Tgfβ20.67 ± 0.121.26 ± 0.120.71 ± 0.070.81 ± 0.070.46 ± 0.02*
Gdf50.02 ± 0.00*0.03 ± 0.00***0.09 ± 0.02**0.06 ± 0.02***0.05 ± 0.01***
Activin βα11.13 ± 1.17**15.77 ± 5.51*14.84 ± 3.18**12.53 ± 2.61*11.29 ± 2.09**
Acvr11.01 ± 0.071.17 ± 0.221.54 ± 0.23*1.58 ± 0.12*1.05 ± 0.28
Bmpr1A1.02 ± 0.041.32 ± 0.141.66 ± 0.30*1.81 ± 0.34*2.33 ± 0.52*
Runx20.31 ± 0.04**0.54 ± 0.17*1.00 ± 0.240.75 ± 0.021.16 ± 0.35
ALPL0.51 ± 0.09 *0.51 ± 0.02*0.58 ± 0.08**1.04 ± 0.030.46 ± 0.07**

Differential regulatory effects between the studied ligands were moderate (Table 1). In general, such differences were shared by members of the same BMP structural subgroup (see Discussion). The following differences were the most remarkable: GDF5 at difference of other BMP ligands had a negative influence on the expression of Tgf β2. BMP5 and BMP7 at difference of the other BMPs induced a mild upregulation of Acvr1 and Bmpr1a. The latter was also upregulated by GDF5. In addition, BMP7 was the only BMP that did not downregulate ALPL (Alkaline phosphatase L). BMP2 and BMP4 at difference of the other BMPs downregulated the expression of Runx2.

Comparative analysis of the intensity of gene regulation revealed some differences among the tested BMPs (Supplemental Fig. S1). The most remarkable difference was an increased positive influence of BMP4 on Type 2 Collagen gene expression and a reduced negative influence of BMP5 and GDF5 on the expression of Scleraxis.

The occurrence of cell-context differences in the response to BMPs has often been recognized. However, the precise nature of the “cell-context” in the case of the prochondrogenic versus the profibrogenic effects of BMPs remains largely undetermined. To assess the extent of variability in the response of the target mesodermal progenitors to BMPs that is attributable to variations in the stage of differentiation, we compared the transcriptional response induced in micromass cultures at different growing stage by 6-hour exposure to one of the BMPs expressed in the developing digits. We selected BMP2 and BMP4 at random for these experiments.

As summarized in Fig. 3, our results reveal dramatic differences in the pattern of gene regulation by BMPs in cultures of 4-day compared with 2-day cultures. These changes are characterized by reduced upregulation of Sox9 and Noggin, the absence of positive regulation of Col2α1, Bmpr1b, and Chdl-1, and strong downregulation of Activin βα. All of these genes are chondrogenic markers that are upregulated by BMP treatment at day 2 of culture. These changes were accompanied by a positive regulation of fibrogenic (Scleraxis) and joint (Chdl-2) markers, which were both negatively regulated at day 2 of culture. Other transcriptional modifications (see below) induced at day 2 of culture were maintained (Gdf5, Chd, and Dan) or attenuated (Fst, Bmper, Sost, Tll1, and Tsg).

Figure 3.

Transcriptional regulations in cultures of 2, 4, and 12 days induced after a 6-hour exposure to BMP4 (table on the left and heat map on the right). The intense upregulation of chondrogenic markers (Sox9, Bmpr1b, Col2α1, Noggin, Chdl-1) at day 2 of culture is nearly undetected in treatments at day 4. Note that Scleraxis is downregulated in 2-day cultures and upregulated in 4-day cultures. Note also that bone markers (ALPL and Runx2) upregulated at day 12 of culture are downregulated after the same treatment at day 2.

Structural differences in the cultures after 6-hour exposure to BMPs were appreciable by immunolabeling of SOX9 and Tenascin in micromasses treated at day 2. As shown in Fig. 2D, at this period, almost every cell of the culture became positive for SOX9, whereas immunolabeling for Tenascin-positive matrix was considerably reduced. At day 4 of culture, positivity for SOX9 was intense in the cartilage nodules, but no significant differences were appreciable between treated and control cultures. In the same fashion, Tenascin-positive matrix was abundant both around and within the cartilage nodules, but its distribution was not significantly modified after 6-hour BMP treatments.

Transcriptional regulation in day 1 cultures was similar to that in day 2 cultures (not shown). Similarly, gene regulation at day 7 of culture was similar to that at day 4 (not shown). In contrast, as shown in Fig. 3, changes in gene regulation at day 12 of culture were dramatic. Gdf-5, Chd, Chdl-2, and Scleraxis and the chondrogenic markers Sox9, Type 2 Collagen, Noggin, Chdl-1, and Bmpr1b were no longer regulated. Tgfβ2, Sost, Dan, Tsg, and Bmper appeared upregulated. The bone markers Runx2 and ALPL that were downregulated in treatments performed at earlier stages of culture became upregulated.

All the treatments and transcriptional analysis of the cultures described above reveal the existence of switches in the early response to BMPs of the connective tissue precursors, varying from prochondrogenic to profibrogenic to pro-osteogenic commitment, depending on the culture length preceding the exposure.

Regulation of BMP antagonists by BMP ligands

During the stages of digit formation, the developing limb exhibits a complex expression pattern of extracellular BMP modulators that create spatial domains with differential intensities of BMP signaling.[49] Therefore, we analyzed in the “2 day of culture/6 hour of treatment” assay possible differences between the BMP ligands in the regulation of extracellular BMP modulators that are associated with digit development (Table 2). All of the tested BMPs strongly induced the expression of Noggin and Fst genes and moderately induced Chdl-1 expression. In addition, all of the tested BMPs repressed Chd, Dan, Sost, Bmper, and Tll 1 expression. Remarkably, unlike other BMPs, BMP2 and BMP4 appeared to be negative regulators of Fstl-1. In a comparable fashion, Tsg was downregulated by BMP2, BMP4, and GDF5 but not by BMP5 and BMP7. In addition, BMP7, unlike other BMPs, downregulated the expression of Chdl-2, which is a specific marker of joint formation.[49]

Table 2. Transcriptional Regulation of BMP Antagonists (Including the Tolloid Metaloprotease Tll1) 6 Hours After the Addition of BMP2, BMP4, BMP5, BMP7, and GDF5 to 2-Day Micromass Cultures of Digit Progenitors
  • *p < 0.05;
  • **p < 0.01;
  • ***p < 0.001.
Noggin33.87 ± 2.53**19.65 ± 5.86*12.65 ± 1.82**19.97 ± 4.43**15.64 ± 1.75**
Chd0.17 ± 0.01***0.17 ± 0.03*0.31 ± 0.01**0.22 ± 0.08**0.26 ± 0.01***
Chdl-13.75 ± 0.18***2.86 ± 0.50**3.76 ± 0.72*2.20 ± 0.23**2.98 ± 0.76*
Chdl-20.79 ± 0.191.08 ± 0.141.30 ± 0.290.55 ± 0.11**0.79 ± 0.11
Tsg0.52 ± 0.01***0.66 ± 0.14*0.87 ± 0.150.97 ± 0.080.66 ± 0.06**
Dan0.14 ± 0.01***0.28 ± 0.06 **0.35 ± 0.08***0.31 ± 0.07***0.29 ± 0.08***
Bmper0.33 ± 0.09**0.46 ± 0.07**0.57 ± 0.09**0.52 ± 0.02***0.39 ± 0.07***
Sost0.38 ± 0.08*0.39 ± 0.13**0.51 ± 0.15*0.72 ± 0.10*0.68 ± 0.08**
Sostdc-10.50 ± 0.05***0.52 ± 0.11*0.77 ± 0.180.43 ± 0.02***0.42 ± 0.09***
Fst7.47 ± 0.82**10.00 ± 3.41*7.20 ± 2.69*4.01 ± 0.63**6.97 ± 0.58**
Fstl-10.45 ± 0.09*0.60 ± 0.11*0.83 ± 0.131.09 ± 0.070.93 ± 0.16
Tll10.21 ± 0.04***0.22 ± 0.02*0.54 ± 0.05*0.60 ± 0.10**0.44 ± 0.16*

Stage-dependent differences in signaling intensity are elicited by BMPs

Based on the above-described stage-dependent BMP responses of mesodermal cultures, we selected the changes between days 2 and 4 of culture for further analysis. The differences in gene regulation prompted us to first analyze the intensity of signaling elicited by the same BMP doses at 2 and 4 days of culture. We observed that phosphorylation of Smad 1/5/8 was threefold more intense in 2-day cultures than in 4-day cultures (Fig. 4). Similar findings were observed for the phosphorylation of p38 MAPK (Fig. 4).

Figure 4.

Western blot showing the phosphorylation of SMAD 1/5/8 and p-38 MAPK at 6 hours after the addition of BMP2 (+BMP2) to the medium compared with untreated cultures (Cont) at days 2 and 4 of culture.

To the light of those results, we next analyzed changes in the expression profile of type I BMP receptors and extracellular BMP antagonists during these 2 days of culture. As shown in Fig. 5A, the rate of expression of type I BMP receptors was unchanged or moderately upregulated at day 4 versus day 2 of culture; however, most extracellular BMP antagonists, including Noggin, Chordins, and Follistatin and the protease Tolloid-like 1, were upregulated up to 18-fold (Fig. 5B).

Figure 5.

Charts showing changes in the expression of type 1 Bmp receptor genes (A) and Bmp antagonists (B) in cultures of 4 days (gray column) compared with 2-day (white column). Note that Noggin is upregulated approximately 18-fold.

To evaluate whether decrease in signaling intensity caused by the increased expression of BMP antagonists at day 4 of culture explains the divergent transcriptional responses between 4-day and 2-day cultures, we analyzed the transcriptional effects of AB204. AB204 is an engineered compound analogous to BMP2 that is not blocked by NOGGIN.[43] Treatments for 6 hours in 4-day cultures restored the positive regulation of chondrogenic markers (Sox9, Col2α1, and Noggin) and the negative regulation of fibrogenic markers (Scleraxis), similar to the results obtained with BMP treatment at day 2 of culture (Fig. 6). In contrast, the characteristic negative regulation of Activin βα induced by BMPs in 4-day micromasses was conserved.

Figure 6.

Charts showing differences in the regulation of a selected group of gene markers after 6-hour treatments of 4-day cultures with BMP2 (white bars) or AB204 (gray bars). The levels of basal expression in untreated cultures are represented by the dotted line. Major chondrogenic markers including Sox9, type 2 Collagen, and Noggin (A) are increased to higher levels by AB204. Remarkably, Scleraxis (B) is negatively regulated by AB204, contrasting with the positive influence of BMP2 at this stage of culture. Asterisks indicate the statistically significant differences compared with the basal expression of untreated cultures. # indicates the level of significance of differences between AB204 and BMP2 treatments. #/*p < 0.05; ##/**p < 0.01; ###/***p <0.001.

In summary, these findings demonstrate that the differential transcriptional regulation of chondrogenic and fibrogenic genes induced by BMPs in 2- and 4-day cultures is associated with a NOGGIN-dependent diminution in the signaling intensity evoked by BMP treatments in 4-day cultures.

The influence of BMP signaling on fibrogenesis is modulated by transcriptional repressors

To gain insight into the mechanism responsible for the stage-dependent regulation of connective tissue differentiation markers, we further analyzed the regulation of Scleraxis. We compared the pattern of regulation of Scleraxis expression induced in 2-day cultures after 6 hours of treatment with the pattern after short 1-hour treatments. For this purpose, we employed BMP2 and AB204. Unlike the transcriptional downregulation observed after 6 hours of treatment, Scleraxis was upregulated after a short (1-hour) treatment (Fig. 7). This opposite regulation was particularly evident in treatments using AB204. This finding suggests that both Scleraxis and a repressor of its transcription are induced by the initial activation of BMP signaling at day 2 of culture. To further investigate this hypothesis, we analyzed the effect of BMP2 on 2-day micromasses preincubated with cycloheximide to inhibit de novo protein synthesis. After 4 hours of culture, this combined treatment caused a dramatic upregulation of Scleraxis (Fig. 8). Positive regulation of Scleraxis, although at more moderated levels, was also observed by the addition of cycloheximide alone, suggesting that at this stage endogenous BMPs are active in the control micromasses.

Figure 7.

Chart showing the positive regulation of Sox9 and Scleraxis in 2-day cultures treated for 1 hour with either BMP2 (white bars) or AB204 (gray bars). Compare these regulations with those shown in Table 1 to note the opposite regulation of Scleraxis but not Sox 9 in short (1-hour) treatments (this chart) compared with medium-length (6-hour) treatments (Table 1).

Figure 8.

Chart showing the strong upregulation of Scleraxis by BMP2 in 2-day cultures treated with BMP2 after inhibition of protein synthesis with cycloheximide (CHX). Note the negative regulation of Scleraxis by BMP2 (second bar) in untreated cultures compared with cultures subjected to inhibition of protein synthesis (third bar). Cycloheximide alone (fourth bar) induces a reduced upregulation of Scleraxis. The first bar shows the basal expression of Scleraxis in untreated control cultures. Asterisks indicate significant differences from the basal expression in untreated cultures. # indicates significant differences in gene regulation under conditions of inhibited protein synthesis with respect of that induced by BMP2 under normal protein synthesis conditions. + indicates the significant differences in the regulation of Scleraxis between combined treatments of BMP2 and cycloheximide and cycloheximide alone. *p < 0.05; +++/ ###/***p < 0.001.

BMP signaling requires Inhibitor of differentiation 2 (Id2) function to repress the expression of Scleraxis

To identify factors implicated in the negative influence of BMPs on the expression of Scleraxis, we analyzed the regulation of a panel of genes able to modulate BMP signaling in micromasses treated with BMPs. We selected Tgif1 and SnoN because they participate in the profibrogenic influence of Tgfβs on limb mesodermal progenitors[50] and Id2 because of its functional association with the helix-loop-helix (bHLH) transcription factors, to which Scleraxis belongs. We observed that expression of Id2 was upregulated up to 10-fold in 2-day micromasses exposed to BMP2 for 6 hours; Tgif1 was not regulated by BMP treatments; and SnoN was upregulated at much lower levels than Id2 (Fig. 9A). According with these findings, we selected Id2 for further studies. Id2 is a direct target of BMPs, with a demonstrated function in chondrogenesis.[50-52] As shown in Fig. 9B, transcripts of Id2 are abundant in micromass cultures of digit progenitors and, in vivo, Id2 shows well-defined expression domains in the growing tip of the digits, where progenitors diverge to differentiate into cartilage instead of forming tendon (Fig. 9C). Furthermore, the level of expression of Id2 related to that of Scleraxis drops by half in cultures of 4 days in relation to 2-day cultures (Fig. 9D). Hence, we analyzed the effect of BMP treatments in cultures subjected to Id2 silencing by transfection of sh-Id2. As shown in Fig. 9E, F, Scleraxis, unlike Sox9, was upregulated and addition of BMP2 did not induce transcriptional changes under these conditions.

Figure 9.

(A) Chart showing the regulation of Tgif1, SnoN, and Id2 in micromass cultures after 6-hour treatment with BMP2 (gray bars). The expression in untreated cultures is shown in white bars. **p < 0.01. (B) In situ hybridization showing expression of Id2 in a 2-day-old micromass culture. Note that transcripts mark nodular prechondrogenic aggregates. (C) Expression of Id2 in the autopod at day 6 of incubation. Note the abundance of transcripts in the growing tip of the digits in a crescentlike pattern. (D) Chart showing the expression of Id2 (black bars) compared with that of Scleraxis (white bars; expression level was set to 1 in each day for comparison), at days 2 and 4 of culture. Note that the relative expression of Id2 drops by half at day 4 of culture. *p < 0.05. (E, F) Charts showing the regulation of Sox9 (E) and Scleraxis (F) by 6-hour BMP2 treatments in 2-day control cultures (white bars) versus cultures subjected to Id2 gene silencing (gray bars). Note that the expression of Sox9 is not modified in cultures subjected to Id2 gene silencing. In contrast, basal expression of Scleraxis is increased after Id2 knockdown, and cultures do not respond to the addition of BMP2. Asterisks indicate statistically significant differences from the basal expression in untreated control cultures. + indicates the significant differences between BMP2 treatments in control versus shId2-transfected cultures. +/*p < 0.05; **p < 0.01.


BMPs are multifunctional cellular regulators with promising potential in regenerative medicine to direct the differentiation of mesenchymal stem cells into connective tissue cell lineages for repairing cartilage, tendon, or bone.[27, 53] However, understanding the molecular basis of this biological property of BMPs requires further investigation to define the biological activity of each molecule individually (see review by Matthews and colleagues[54]). In this study, we have addressed two questions with clinical relevance. First, do individual BMP family members specifically induce the differentiation of connective tissue progenitors into distinct cell lineages (chondrocytes, tenocytes, and osteocytes)? Second, to what extent does the initial differentiation of mesenchymal progenitors condition their response to BMP-driven connective tissue differentiation? For our purposes, we have chosen a developmental approach employing primary cultures of embryonic limb connective tissue progenitors to test the effects of the BMP ligands that are physiologically expressed throughout digit formation.

Comparative analyses of the effects of BMP ligands on the diversification of the connective tissues are scarce in the literature[12, 25, 55] and have focused mainly on osteogenesis.[30, 56-59] Our observations indicate that during the initial stages of digit morphogenesis, the formation of tendons, cartilage, and joint tissues is accompanied by a regionalization of the autopod mesoderm into domains of distinct BMP ligand expression. The formation of digit cartilage primordia occurs within the mesenchymal core of the autopod expressing Bmp2, Bmp7, and Bmp4. Tendon blastemas appear next in development in the mesodermal tissue located between the cartilaginous digit primordia and the dorsal or ventral ectoderm.[44] These regions express high levels of Bmp5 and Bmp4. Joints are formed at discrete regions of the digit cartilages expressing a thin linear domain of Bmp2 together with a broad domain of Gdf5[60] that extends to the mesenchyme surrounding the digit ray where ligaments and tendon entheses develop. Differences in the affinity of these five BMP ligands for the different BMP type I receptors have been reported.[31] BMP-2 and BMP-4 belong to the same structural group of BMP ligands and preferentially bind to ALK-3 (BMPR-1A) and ALK-6 (BMPR-1B). BMP5 and BMP7 are members of the OP-1 group and bind to ALK-2 (ACVR1) and ALK-6/BMPR1B. GDF-5, which also names the other structural group of BMPs, binds to ALK-6/BMPR1B but not efficiently to other receptors. Our study showed that the 6-hour administration of the five analyzed BMPs to micromass cultures of digit progenitors elicited common transcriptional effects on the expression of cartilage markers (Sox9, type 2 Collagen, Fz4, Bmpr1b, and Activin βα), joint markers (Gdf5 and Wnt14), and BMP antagonists (Noggin, Chordin, Chordin-like1, Dan, Sost, Bmper, Follistatin, and Tolloid-like 1). This result can be explained by the common binding of all of the BMPs to BMPR1B/ALK6, which is the receptor responsible for the prochondrogenic effects of BMP signaling.[7, 38] In addition to this common effect, we detected differential transcriptional regulation of a number of markers. Remarkably, GDF5 exhibited a significantly negative influence on the expression of Tgfβ2, which is a key regulator of tendon and ligament differentiation.[61, 62] The precise significance of this transcriptional effect requires further investigation, but genetic deficiencies in Gdf5 generate not only cartilage and joint anomalies (brachypodism/brachydactyly[63-65]) but also structural tendon anomalies.[39, 66] The potential functional importance of other differences in regulation observed in this study (the negative influence of BMP7 on the expression of Chdl-2; the negative influence of BMP2 and BMP4 on the expression of Runx2, Tsg, and Fstl-1; and the positive influence of BMP5 and BMP7 on the expression of Acvr1 and Bmpr1a) is difficult to establish. However, considering the overlapping expression of BMP ligands in different digit tissues (Bmp5 and Bmp4 first and Bmp2, Bmp4, Bmp5, and Bmp7 next in tendon blastemas; Bmp2, Bmp4, and Bmp7 in zones of chondrogenesis; and Gdf5 and Bmp2 in joints), these transcriptional differences are difficult to assign specifically to BMP 2, 4, 5, or 7 in the lineage diversification of mesodermal progenitors. In accordance with this interpretation, apart from a specific function of Bmp2 in bone maturation and fracture healing,[67] genetic analyses in mice and humans indicate strong functional redundancy between Bmp2, Bmp4, Bmp5, and Bmp7 in limb skeletogenesis.[40, 68, 69]

Unlike the relative functional similarity observed for all five selected BMP ligands, the transcriptional response of skeletogenic mesodermal progenitors to BMPs differed with the successive steps of cell differentiation. The initial response of the undifferentiated mesoderm (days 1 and 2 of culture) consisted of strong activation of prochondrogenic genes. This early response was followed by activation of profibrogenic genes accompanied by attenuation of the prochondrogenic effect (days 4 to 7 of culture). Remarkably, genes strongly upregulated in the first stage, such as Activinβα, became strongly downregulated in the second stage. In the third period studied here, chondrogenic and fibrogenic markers were no longer upregulated, whereas bone gene markers, such as ALPL and Runx2, that were downregulated in previous stages became upregulated. Changes in the response of progenitor cells to BMPs at different stages of differentiation are poorly characterized. A sequential variation in the biological effects of BMPs has been reported in the chondrogenic cell line ATDC5,[70] and cultured chondrocytes obtained from different layers of the joint surfaces, which are at different stages of differentiation, respond differently to both BMPs and TGFβs.[71] Consistent with these observations, our findings indicate that the basal gene expression profile of progenitors might be a critical condition for the subsequent BMP-driven lineage differentiation. Furthermore, our findings suggest that connective tissues differentiate dynamically under the influence of BMPs. Cartilage, fibrous tissue (tendon, joint capsules, perichondrium, and periosteum), and bone appear sequentially in the differentiation of common progenitors. According to this interpretation, BMPs could promote the successive steps in connective tissue differentiation through a process dependent on the basal expression of BMP cofactors or signaling modulators. This feature should be considered in the elaboration of protocols to differentiate stem cells for regenerative medicine, as the differentiated cell type may be different after short-term exposures to BMPs, long-term exposures to BMPs, or viral-based overexpression of BMPs.[72]

Our study provides some cues for understanding the divergent functions of BMP signaling in the differentiation of connective tissues. We show that the expression level of Noggin (together with other cartilage-expressed BMP antagonists) is a landmark in the transition between the chondrogenic and fibrogenic response to BMPs. This finding is consistent with an analysis of expression over the duration of joints and tendon enthesis formation in the developing limb.[73] Furthermore, we observed that AB204, which is a compound analogous to BMP that escapes NOGGIN antagonism, maintains the chondrogenic rather than the profibrogenic transcriptional influence of BMPs on 4-day micromass cultures. These findings support previous studies proposing that NOGGIN is a key profibrogenic factor in the embryonic limb.[16, 74] However, according to the present findings, the effect of NOGGIN is to attenuate rather than block BMP signaling.

The stage-dependent transcriptional regulation of Scleraxis by BMPs is an additional finding relevant to understanding BMP's function in connective tissue differentiation (summarized in Fig. 10). Scleraxis is a helix-loop-helix transcriptional factor that is considered to be the master regulator for the differentiation of mesodermal progenitors into fibrous connective tissue.[16] In the undifferentiated mesoderm, Scleraxis is upregulated after very short BMP treatments (1 hour), but the gene is downregulated after a longer exposure to BMPs (6 hours). Remarkably, we have observed that the inhibition of protein synthesis by cycloheximide blocks Scleraxis downregulation and potentiates Scleraxis upregulation. These findings indicate that in the undifferentiated mesoderm BMPs induce both Scleraxis and one or more factors that repress its expression. In contrast, at day 4 of culture when cartilage nodules are formed, 6-hour BMP treatments upregulate Scleraxis, indicating that the putative repressor is no longer induced at appropriate levels. Because the addition of AB204 to 4-day micromasses still downregulates the expression of Scleraxis, the differential regulation by BMPs at 2 days versus 4 days of culture is associated with the reduced level of signaling at this stage, which is mediated by the increased expression of BMP antagonists. In our attempt to identify factors responsible for the differential regulation of Scleraxis at days 2 and 4 of culture, we observed that Id2 is required for BMPs to inhibit Scleraxis expression at day 2. Id2 is a direct target of BMPs that is expressed in the developing digits[50] and that potentiates the chondrogenic effect of BMPs.[52] The effect of Id2 includes the regulation of basic helix-loop-helix (bHLH) transcription factors and the inhibition of the anti-BMP influence of Smad 7.[52] However, the precise developmental significance of Id2 in this regulatory process requires further clarification. Mice deficient for Id2 lack anatomical skeletal phenotype at birth, although skull differentiation is impaired during the postnatal period.[52] It is likely that other regulatory factors cooperate with Id2 to direct the differentiation of mesodermal progenitors into distinct connective tissues. The functional repressor of Id2, FHL2 gene,[75] might be a good candidate for this function because it is upregulated in micromass cultures treated with NOGGIN (data not shown).

Figure 10.

Schematic diagram illustrating the major findings of the study. Note that an early response of BMPs positively modulated both Sox9 and Scleraxis. However, they also promote early BMP response modulators (as Id2), which may initially repress the expression of the fibrogenic marker Scleraxis.

In summary, this study indicates that the divergent response of the skeletogenic mesoderm in the developing digits to BMP stimulation depends largely on the gene expression profile found in the target cells at the moment of treatment and not on the action of distinct members of the BMP family. Whether the differential expression of BMP ligands during digit formation reflects differences in the physiological regulation of BMP gene expression rather than differences in their skeletogenic function is a tempting hypothesis that requires further investigation.


All authors state that they have no conflicts of interest.


Thanks are owed to Montse Fernandez-Calderon, Sonia Perez-Mantecón, and Susana Dawalibi for excellent technical assistance. This work was supported by a grant from the Spanish Science and Innovation Ministry to JMH (BFU2011-24169).

Authors' roles: All authors meet the requirements for authorship as specified in the JBMR Author Guidelines. CILD, JAM, JAGP, and JMH conceived and designed the experiments. CILD and JAM performed the experiments. CILD, JAM, JAGP, and JMH analyzed the data and interpreted the results. SC codesigned experiments, discussed analyses, and provided interpretation. CILD, JAM, SC, JAGP, and JMH contributed reagents/materials/analysis tools. CILD, JAM, and JMH wrote the paper. All authors approved the final version of the submitted manuscript.