Gene Array Analysis of Bone Morphogenetic Protein Type I Receptor-Induced Osteoblast Differentiation

Authors


  • The authors have no conflict of interest

Abstract

The genomic response to BMP was investigated by ectopic expression of activated BMP type I receptors in C2C12 myoblast using cDNA microarrays. Novel BMP receptor target genes with possible roles in inhibition of myoblast differentiation and stimulation of osteoblast differentiation were identified.

Introduction: Bone morphogenetic proteins (BMPs) have an important role in controlling mesenchymal cell fate and mediate these effects by regulating gene expression. BMPs signal through three distinct specific BMP type I receptors (also termed activin receptor-like kinases) and their downstream nuclear effectors, termed Smads. The critical target genes by which activated BMP receptors mediate change cell fate are poorly characterized.

Materials and Methods: We performed transcriptional profiling of C2C12 myoblasts differentiation into osteoblast-like cells by ectopic expression of three distinct constitutively active (ca)BMP type I receptors using adenoviral gene transfer. Cells were harvested 48 h after infection, which allowed detection of both early and late response genes. Expression analysis was performed using the mouse GEM1 microarray, which is comprised of approximately 8700 unique sequences. Hybridizations were performed in duplicate with a reverse fluor labeling. Genes were considered to be significantly regulated if the p value for differential expression was less than 0.01 and inverted expression ratios per duplicate successful reciprocal hybridizations differed by less than 25%.

Results and Conclusions: Each of the three caBMP type I receptors stimulated equal levels of R-Smad phosphorylation and alkaline phosphatase activity, an early marker for osteoblast differentiation. Interestingly, all three type I receptors induced identical transcriptional profiles; 97 genes were significantly upregulated and 103 genes were downregulated. Many extracellular matrix genes were upregulated, muscle-related genes downregulated, and transcription factors/signaling components modulated. In addition to 41 expressed sequence tags without known function and a number of known BMP target genes, including PPAR-γ and fibromodulin, a large number of novel BMP target genes with an annotated function were identified, including transcription factors HesR1, ITF-2, and ICSBP, apoptosis mediators DRP-1 death kinase and ZIP kinase, IκBα, Edg-2, ZO-1, and E3 ligase Dactylin. These target genes, some of them unexpected, offer new insights into how BMPs elicit biological effects, in particular into the mechanism of inhibition of myoblast differentiation and stimulation of osteoblast differentiation.

INTRODUCTION

Bone morphogenetic proteins (BMPs) were originally identified by their ability to induce ectopic cartilage and bone formation(1–3) and are members of the multifunctional transforming growth factor-β (TGF-β) superfamily, which also includes TGF-βs and activins. BMPs have important roles in directing cell fate choices of mesenchymal cells; they stimulate osteoblast and adipocyte differentiation while inhibit myogenic differentiation.(4–8)

BMPs elicit their cellular effects by inducing heteromeric complexes of type I and type II serine/threonine kinase receptors.(9) For BMPs, three distinct type II receptors, that is, BMP type II receptor (BMPR-II) and activin type II and type IIB (ActR-II and ActR-IIB), and three type I receptors, also termed activin receptor-like kinases (ALKs), that is, BMPR-IA/ALK3, BMPR-IB/ALK-6, and ALK-2, have been identified. On ligand-induced heteromeric complex formation, the constitutively active type II receptor kinase phosphorylates the type I receptor.(10) Thus, the type I receptors act downstream of type II receptors and confer signaling specificity to the heteromeric receptor complex.(11,12)

Activated BMP type I receptors initiate intracellular signaling through phosphorylation of specific receptor-regulated (R-)Smad proteins, that is, Smad1, Smad5, and Smad8, at their extreme carboxy terminal SSVS motif.(13,14) Activated R-Smads form heteromeric complexes with common partner (Co-)Smad, that is, Smad4,(15,16) which efficiently accumulate in the nucleus. Within the nucleus, these Smad complexes regulate the transcription of target genes.(15,17)

Transcriptional activation of R- and Co-Smads is mediated through their conserved carboxy terminal regions, known as MH2 domains, which can recruit transcriptional co-activators such as CBP/p300. Smads can bind directly to specific DNA sequences through their conserved amino-terminal regions, known as MH1 domains.(15,17)

A number of target genes for BMPs in mesenchymal cells and osteoblasts have been identified.(18) BMPs have been reported to induce the expression of various extracellular matrix proteins, including collagens, bone sialoprotein, decorin, and osteocalcin. BMP-induced expression of osteocalcin and alkaline phosphatase (ALP) are often used as indicators of BMP-induced osteoblast differentiation. However, BMP-induced increases in osteocalcin and ALP expression require protein synthesis, and therefore, both osteocalcin and ALP are indirect BMP target genes.(18)

BMPs affect cell fate by regulating the expression of transcription factors, among which is a member of the core binding factor 1α(CBF) family of transcription factors (Cbfa1), also termed Runt-related gene 2 (RUNX2), that has a pivotal effector role in BMP-induced osteoblast differentiation. Other transcription factors that were identified to be induced by BMPs, including Id1 and JunB, may indirectly support osteoblast differentiation of mesenchymal precursor cells by preventing their differentiation to alternate pathways.(19–21) To fully understand the molecular mechanisms that govern BMP-induced mesenchymal cell differentiation, additional BMP target genes with critical roles in this process remain to be identified.(18)

Different BMP type I receptors have been reported to mediate distinct responses; in bone marrow stromal 2T3 cells, ALK3 was reported to function as an inducer of both adipogenesis and osteogenesis, whereas ALK6 induced only osteogenesis while blocking adipogenic differentiation.(7) During limb bud morphogenesis in the chick, ALK3 was found to mediate osteogenesis, whereas ALK6 induced preferentially chondrocyte differentiation.(22) In neural precursor cells, caALK3 was found to induce expression of ALK6 and to promote proliferation, whereas caALK6 caused mitotic arrest resulting in apoptosis or terminal differentiation in early or mid-gestation stages mouse embryos, respectively.(23) However, ALK2, ALK3, and ALK6 were all found to stimulate the differentiation of C2C12 myoblasts into osteoblasts.(24,25) Available data do not allow understanding of which target genes can be specific for different BMPR-Is and responsible for different biological effects.

In this study, we have investigated the effect of ectopic expression of three distinct BMP type I receptors on gene expression in C2C12 cells. Novel target genes for BMP type I receptors were identified, and their potential significance in BMP-induced differentiation of mesenchymal cells is discussed.

MATERIALS AND METHODS

cDNA probes and ligands

Id1, Id2, and Id3 cDNAs were provided by the Dr H Weintraub Laboratory, HesR1 cDNA by Dr CCW Hughes, PPARγ cDNA by Dr S Harris, RUNX2/Osf2 by Dr G Karsenty, slug cDNA by Dr P Savagner, osteocalcin cDNA by Dr M Noda, Edg1 and Edg2 cDNAs by Dr W Moolenaar, ZO-1 cDNA by Dr A Fanning, IκBα by Dr S Makarov, Drp1 cDNA by Dr A Kimchi, Dlk construct by Dr D Kögel, and mouse CD44s cDNA by Dr R van der Neut. Recombinant BMPs were a gift from Dr K Sampath.

Cell culture and ligands

C2C12 cells were cultured in DMEM (Invitrogen, Paisley, UK) containing 10% fetal calf serum (FCS; Sigma, Irvine, UK). Cells were grown in 5% CO2-containing atmosphere at 37°C.

Adenoviral infection, RNA isolation, and Northern blot analysis

caALK2, caALK3, caALK6, and LacZ adenoviral constructs were provided by Dr K Miyazono.(25) C2C12 mouse mesenchymal cells were seeded at density of 2.5 × 104/cm2, and 12 h later were infected with caALK2, caALK3, caALK6, or LacZ (β-galactosidase-expressing control) adenoviral constructs. Cells were incubated in DMEM supplemented with 10% FCS, and before collecting cells, were extensively washed and starved for 12 h in serum-free medium supplemented with 0.2% bovine serum albumin (BSA; Sigma). This protocol was adopted because we found that serum in culture medium is needed to get efficient adenoviral infection, but that serum induces some basal Smad1/5/8 phosphorylation. Before harvesting, cells were washed twice with ice-cold PBS. Isolation of total RNA was performed with using TriZol reagent (Invitrogen) according to manufacturers recommendations. For microarray experiments, polyA+ RNA material was selected with Oligotex polyA+ selection mini kit (Qiagen, Valencia, CA, USA). Concentration of polyA+ RNA was measured spectrofluorometrically with RibogreenTM RNA quantitation assay (Molecular Probes, Leiden, The Netherlands). For Northern blot analysis, 20 μg total RNA per lane was loaded on denaturng (6% formaldehyde) 1% agarose gels. Blotting and hybridization using probe labeled with α-32P-dCTP (Amersham Biosciences, Buckinghamshire, UK) was performed as previously described.(26)

Western blot analysis

Western blotting was performed as described previously.(26) Polyclonal rabbit PS1 antibody (used at a dilution of 1:1000) that specifically recognizes phosphorylated Smad1, Smad5, and/or Smad8 has been previously described.(27) Antibody against HA-tag (used at a dilution of 1:1000) was from Roche (Mannheim, Germany). Secondary horseradish peroxidase-conjugated goat anti-rabbit IgG antibody (Amersham Pharmacia) was used in a 104-fold dilution. Detection was performed by enhanced chemoluminescence (ECL; Amersham Biosciences, Buckinghamshire, UK).

ALP assays

Histochemical examination of ALP activity produced by C2C12 cells was performed using naphtol AS-MX phosphate (Sigma, St Louis, MO, USA) and fast blue RR salt (Sigma) as previously described.(6)

Microarray analysis

Mirroarray hybridizations were performed with mouse GEM1 array (Incyte Genomics (Mountain View, CA, USA) containing 8700 cDNA spots. Both labeling and microarray hybridizations were performed by Incyte Genomics in accordance to previously described protocols.(28) Briefly, 200 ng of PolyA+ RNA from every experimental variant or reference sample (prepared from nontreated cells) was labeled using reverse transcriptase reaction in the presence of dNTPs and either a Cy3 or Cy5 random 9mer. Hybridizations were performed in duplicate using a reciprocal Cy3/Cy5 fluor labeling. Hybridization results on individual spots were selected for further analysis when the intensity signal to background ratio was greater than 2.5 and the area occupancy under the grid greater than 40%. Following signal quantification, a signal correction algorithm was used to correct for systematic differences between the Cy3 and Cy5 labels. This algorithm applied a 2nd order polynomal regression model to the data by fitting a parabola through log-transformed Cy3 versus Cy5 intensities. The residuals of the regression model were taken as the new gene expression ratios. Genes that were significantly differentially expressed in the experimental versus the reference sample were selected by a statistical method derived from boxplots, which are widely used to visualize the overall shape of a data set.(29) In our experiments that compare a single perturbation to an isotypic reference, the expression of the majority of genes will not differ between the experimental sample and the reference sample and their ratios will be in the center of the distribution of expression ratios. Therefore, we computed the first (Q1) and third (Q3) quartiles of the distribution of the residuals of the regression model and the interquartile range (IQR) of the distribution as a measure for the variation in the expression ratios of nondifferentially expressed genes. Then, in analogy with box plots, an inner fence was set at Q1 − (1.41 × IQR) and Q3 + (1.41 × IQR). Using these criteria, genes within the inner fence have a probability of p = 0.99 to be nondifferentially expressed, whereas the outlier group will harbor the differentially expressed genes. Genes were selected for further study if they fell outside the inner fence in both of the duplicate experiments and displayed an inverted ratio in the reciprocal labeling experiment. Selected genes were categorized into functional groups using information from the Gene Ontology Consortium (http://www.geneontology.org). Data were visualized using the Spotfire Decision Site 7.0 analysis package (Spotfire Inc., Somerville, MA, USA).

RESULTS

Ectopic expression of constitutively active BMP type I receptors

BMP stimulates osteoblast differentiation and inhibits myogenic differentiation of C2C12 cells. ALK2 and ALK3 are expressed by C2C12 cells.(30) Each BMP can bind, although with different affinity, to multiple BMP type I receptors.(31) Thus, to investigate signaling responses downstream of each specific BMP type I receptor, we cannot stimulate C2C12 cells with different BMP isoforms. Therefore, we have chosen to ectopically express in C2C12 cells constitutively active (ca) mutants of BMP type I receptors(12,22) that signal in the absence of ligand. A disadvantage of this approach is that normal kinetics of signaling is eliminated. caBMP type I receptors were overexpressed by adenoviral infection. First we analyzed the expression of HA-epitope tagged receptors in time and at different multiplicity of infection (moi). Using adenoviral vectors allowed us to get 100% of transduced cells already at moi of 1500 (evaluated histochemically by staining for specific galactosidase activity after infection with control LacZ adenovirus; data not shown). BMP type I receptors were found to be expressed after 10 h, increased in expression until 24–36 h postinfection, and remained expressed at that level until 4 days, after which expression decreased (data not shown). In a subsequent experiment, the expression pattern (between 24 and 48 h) of three caALKs were compared at different moi. Expression at moi 500 was found to elicit a similar expression level as moi 2000 (Fig. 1). We subsequently analyzed Smad1, Smad5, and/or Smad8 phosphorylation by the three adenovirally expressed BMP type I receptors. Western blotting of cell lysates of adenoviral infected C2C12 cells with phospho-Smad1 antibody (which cross-reacts equally efficiently with phosphorylated Smad5 and phosphorylated Smad8) showed that all three BMP type I receptors induced similar levels of phosphorylated BMP R-Smads (total pull of Smad1, Smad5, and/or Smad8 together (Fig. 1A). Thus, differential effects on gene expression by caALKs caused by different Smad activation levels were ruled out. All three type I receptors also activated to similar levels of ALP activity, an early marker for osteoblast differentiation (Fig. 1B) and did not influence significantly viability of cells (data not shown).

Figure FIG. 1..

Ectopic expression of caBMP receptors in C2C12 cells. (A) Overexpression of HA-epitope tagged caALKs in C2C12 cells by adenoviral gene transfer. Receptor expression was measured by blotting cell lysates with HA antibody. Smad phosphorylation was detected with PS1 antibody. BMP-7 stimulated cells were taken along as positive control for BMP-induced Smad1 phosphorylation. Multiplicity of infection (moi) is indicated. (B) Effect of ectopic caALK expression (moi = 2000) on alkaline phosphatase activity. (C and D) Northern blot analysis of mRNA expression 29, 36, or 48 h after caALK infection. Effect of caALK expression (moi = 2000) on (C) Id1, Smad6, and osteocalcin mRNA expression. (D) Quality control for RNA samples for microarray analysis. Effect of caALK adenoviral transduction expression 48 h after infection (moi = 2000) on Id2 and Runx2 expression is shown. In C and D, GAPDH hybridization pattern is shown to control for equal loading. Positions of 18S and 28S are indicated.

Constitutively active receptors are continuously signaling and therefore early and late response genes will be activated when receptors are expressed for some time. We therefore analyzed in adenoviral infected C2C12 cells the expression of Id1, a direct target for BMP, of which the expression peaks around 1–2 h, and decreases thereafter,(19) and Smad6, a direct target of which expression remains high with time,(32,33) as well as osteocalcin, a late target for BMP, of which expression is induced only after 1–2 days.(34) We found that Id1 expression peaked 24 h after infection and decreased in expression thereafter. Smad6 was found to be expressed at similar levels in the 24–48 h postinfection period. Osteocalcin was found to be expressed only 48 h after infection after caALK adenovirus transduction (Fig. 1C). To investigate both early and late target genes and thereby to maximize our chances in finding differences in gene expression between the three BMP type I receptors, we have chosen to analyze gene expression by cDNA microarrays 48 h after caALK infection with moi of 2000.

Transcriptional profiling of caBMP type I receptors

To analyze the effect of caALK2, caALK3, or caALK6 on gene expression we compared caBMPR-Is adenovirus infected RNA samples with LacZ adenovirus infected RNA. RNA from LacZ adenovirus infected C2C12 cells was also compared with noninfected cells to analyze the effect of LacZ adenovirus infection on gene expression. We first validated the RNA samples to be used for cDNA microarrays by Northern blotting using BMP target genes Id2(19) and Runx2(35) cDNA probes. Both genes were found to be potently induced on infection by caALKs, but not LacZ or noninfected and non-BMP-treated cells (Fig. 1D). No difference in intensity of induction for either genes was observed between the three different ALKs. We used the mouse GEM1 array, which consists of approximately 8700 annotated genes and ESTs (Incyte Genomics). Each experiment was performed twice with reversal of the Cy3 and Cy5 dyes. Genes were considered to be significantly regulated by caALK, if they exhibited inverse expression ratios per duplicate hybridization and if the p value was less than 0.01. The microarray data were found to be very reliable as the reproducibility of two independent labeling experiments was very high (Fig. 2). Furthermore, genes that are represented by multiple clones on the array (e.g., Sox11, DRP-1, extracellular matrix protein 1) showed identical results from the different clones.

Figure FIG. 2..

Heat map of the transcriptional profiles induced by the different BMP receptors. The figure shows log-transformed expression ratios from paired duplicates on a pseudocolor scale indicated at the bottom of the figure; black color shows no changes. Genes were grouped using the biological process annotation from the Gene Ontology Consortium. The right column of the heat map shows the subcellular localization of protein products of each particular gene (color code indicated at the bottom of the figure).

Expression analysis revealed 200 genes to be at least 1.7-fold differentially regulated by caALK expression, among which 97 genes were upregulated and 103 genes were downregulated. The data reveal a global pattern of downregulation of muscle-related genes, a concurrent upregulation of extracellular matrix associated genes, and a broad pattern of modulation of genes involved in signal transduction (Fig. 2). Forty-one expressed sequence tags without known function were identified. The control experiment with an adenovirally transduced LacZ gene revealed that the adenoviral infection itself had very little effect on gene expression (Fig. 2). Surprisingly, no significant differences in gene expression were observed when comparing the caALK2-, caALK3-, and caALK6-infected RNA samples (Fig. 2). Several genes that previously have been reported to be regulated by BMPs were identified as caALKs-regulated genes, including the upregulation of peroxisome proliferator-activator receptor γ (PPAR-γ)(7) and fibromodulin,(36) and the downregulation of insulin-like growth factor binding protein-5,(37,38) and different muscle-specific genes, such as myosin heavy and light chains.(39) Furthermore, we observed differential expression of the genes encoding the extracellular matrix proteins collagen IIIα1 and lysyl oxidase(40) and the genes encoding insulin-like growth factor-1,(40) integrin α7,(40) and CD44,(41) which all have been implicated in osteoblast differentiation previously.

Among the newly identified BMP receptor targets that were found upregulated are the transcription factors HesR1 and interferon consensus sequence binding protein (ICSBP), the NFκβ inhibitor IκBα, and the LPA receptor Edg2. Among the genes that were observed to be downregulated are the transcription factors Sox11 and ITF-2, tight junction protein Zonula occludens 1 (ZO-1), pro-apoptotic kinases DRP-1 and Zip kinase, and E3 ligase Dactylin. Based on their function, the BMP targets can be divided into different groups (Fig. 2).

Northern blot analysis of selected BMP receptor target genes

A number of BMP target genes were selected for further analysis by Northern blotting to verify their differential expression. The upregulation of PPAR-γ, HesR-1, Edg-2, and the downregulation of DRP-1 and ZO-1 was confirmed (Fig. 3). Edg-1, closely related to Edg-2, was also found upregulated. In most cases the fold induction obtained by Northern blot analysis was comparable with that obtained in microarray studies (data not shown). Thus, none of the genes as identified to be regulated by BMP receptors in microarray experiment showed discrepancy with results obtained from Northern blot analysis.

Figure FIG. 3..

Northern blot analysis of selected BMP receptor target genes. Differential gene expression in response to caALKs were validated for HesR1, Slug, PPAR-γ, Edg-1, Edg-2, Drp-1 death kinase, and ZO-1. C2C12 cells were nontreated (control) or infected with LacZ, caALK2, caALK3, or caALK6 (moi = 2000) for 48 h and were analyzed by Northern blot analysis. Blots were hybridized with indicated cDNA probes. Equal loading of RNA samples in left panel is shown by ethidium bromide stain of gel before Northern blotting and for right panel by GAPDH hybridization.

The zinc finger transcriptional repressor Slug was reported to be induced by TGF-β2.(42) We were therefore interested whether Slug is induced by caALKs; we found Slug to be induced by all three of caALKs (Fig. 3).

DISCUSSION

BMPs have an important function in controlling cell fate, and they mediate these effects, at least in part, by regulating gene expression. We performed microarray analysis on C2C12 cells in which the three distinct caBMP receptors were ectopically expressed to obtain more insight into the mechanisms that govern BMP receptor-induced differentiation of mesenchymal cells. We used C2C12 cells as an experimental system, which on BMP stimulation, are inhibited in myoblast differentiation and promoted to differentiate into osteoblasts.

Ectopic expression of caALK2, caALK3, and caALK6 induced equally efficient levels of total BMP R-Smad pull (Smad1, Smad5, and/or Smad8) phosphorylation and ALP activity in C2C12 cells. In addition, myoblast differentiation was equally well inhibited by caALK2, caALK3, and caALK6 (data not shown). These data are consistent with those obtained by Fujii et al.(25) Analysis of GEM1 array containing 8700 cDNA spots revealed that 200 genes were significantly differentially regulated by caALK expression, among which 97 genes were upregulated and 103 genes were downregulated. For 41 genes, no functional annotation was described in various databases. Several of genes we found induced on caALKs adenoviral transduction were already described before after induction of osteoblastic differentiation in C2C12 cells with different BMPs.

Different BMP receptors have been reported to mediate different biological responses. For example, 2T3 cells were found to differentiate into adipocytes by caALK3, but not ALK6, and consistent with this finding, PPAR-γ, which has been reported to play an important role in the differentiation pathway to adipocytes, was found to be induced by caALK3, but not caALK6 in these cells.(7) To our surprise, this study showed no significant differences in gene expression on ectopic expression of caALK2-, caALK3-, and caALK in C2C12 cells. One reason may be the high expression that caALK levels achieved by adenoviral infection; another reason may be that these effects are cell-type dependent. Moreover, we cannot distinguish between phosphorylated Smad1, Smad5, or Smad8 using our phospho-Smad antibody, and thus cannot exclude the possibility that the three type I receptors activate Smad1, Smad5, and/or Smad8 with unequal efficiency in different cell types. Different BMPs have been shown to activate Smad1, Smad5, and Smad8 differentially,(43) and all three Smad isoforms may not be functionally equivalent (our unpublished data). To obtain further insight into why different type I receptors may induce distinct responses, we are currently investigating the transcriptional profile of caALK2, caALK3, and caALK6 in other mesenchymal cells and determining which Smad isoforms are activated in each cell type. Preliminary data indicate that different responses can be achieved by caALKs in other mesenchymal cells, despite reaching similar overexpression levels as in C2C12 cells.

PPAR-γ was found to be equally induced by caALK2, ALK3, and ALK6 in C2C12 cells. Stimulation of C2C12 cells with activators of PPAR family members has been shown to induce their differentiation in adipocytes.(44) However, in our studies infection with caALK-2, -3, and -6 in C2C12 cells did not induce adipogenesis even in the presence of PPAR-γ agonist indomethacin and vitamin C (data not shown), suggesting that PPARγ alone is not sufficient for adipocyte differentiation, and/or that other factors that are induced by caALK2, -3 and -6 override the adipogenic promoting signals.

Among the targets, we found genes previously shown to be induced by BMP, but also many targets not previously implicated in BMP signaling. Below we discuss the possible implications of newly identified BMP receptor targets for inhibition of myoblast differentiation and stimulation of osteoblast differentiation.

Basic helix-loop-helix protein hairy enhancer-of-split-related 1 (HesR1) was found to be one of most intensively upregulated genes by caALK2, -3, and -6 in our microarray experiments. HesR1 is a basic helix-loop-helix factor and a direct target of Notch in C2C12.(45)HesR1 has been reported to function as a transcriptional repressor of myogenesis by forming an inactive complex with MyoD.(46) Thus, the inhibition of myogenesis by BMP of C2C12 cells may be mediated through upregulation of HesR1. This adds HesR1 to the list of previously identified negative regulators of myogenesis that are induced BMPs in C2C12 cells, that is, Id proteins(47) and JunB.(21)

The zinc finger transcriptional repressor Slug was also induced by caALK2, -3, and -6. Slug has been shown to cause repression of E-cadherin(48) and desmosome dissociation and cell spreading.(49) The function of Slug as a downstream target in osteoblast differentiation remains to be investigated.

Inhibitory kappa B-α (IκB-α), a component of the NF-κB pathway, was found to be upregulated by caALK2, -3, and -6. Its main function is keeping p65/p50 complex inactive in the cytoplasm.(50) Thus, upregulation of IκB-α by caALKs may inhibit NF-κB pathway, which was shown to counteract Smad signaling.(51)

E3 ligase Dactylin was found to be downregulated by all three BMP type I receptors. Dactylin is important during limb development. Alterations in dactylin expression cause dactylaplasia, and adult homozygotes lacking its expression also lack hands and feet except malformed single digits.(52) Which proteins are specifically targeted to proteasomal degradation by dactylin and the functional importance of its downregulation by BMPs during osteogenesis remain to be investigated.

The two proapoptotic serine/threonine kinases, DRP-1(53) and ZIP kinase (also called DLK),(54,55) were downregulated in response to ectopic expression of BMP type I receptors. Downregulation of Drp-1 was found only after 6 h of BMP-6 stimulation (data not shown). Both DRP-1 and ZIP kinase are related to death-associated protein (DAP) kinase. ZIP kinase induces apoptosis and has also been shown to be involved in myosin phosphorylation and contraction of smooth muscle cells.(56) Interestingly, DAP kinase was recently shown to be induced by TGF-β and mediate TGF-β-induced apoptosis in Hep3B cells.(57) Thus, BMPR-mediated downregulation of DRP-1 and ZIP kinase may be important for survival of cells during osteoblastic differentiation.

Edg1 and Edg2 are sphingosine 1-phosphate (S1P) and lysophosphatidic acid (LPA) receptors, respectively, which are upregulated by BMP stimulation. Interestingly, BMP has also been shown to upregulate autotaxin, a lysophospholipase D that generates LPA from lysophosphatidyl choline, in mesenchymal progenitor cells.(58) Both LPA and S1P have been reported to be osteoblast mitogens, and the induction of their receptors (and possibly LPA) on BMP receptor activation may thus provide a proliferative signal. Interestingly, Edg2 expression is quickly (already after 2 h) induced by BMPs and reaches peak expression after 1 day and remains at this high level for at least 6 days. Edg1 expression was found to be significantly elevated only after 1 day (data not shown). Interestingly, mice deficient in Edg2 show developmental abnormalities including craniofacial dysmorphism.(59)Zonula occludens 1 (ZO-1), also termed tight junction protein 1, was found to be downregulated by caALKs. ZO-1 has been shown to interact with connexins in osteoblasts,(60) which modulate gap junctional communication and production of bone matrix proteins in osteoblasts.

In conclusion, using a microarray approach, we have identified BMP receptor target genes already known to have an important function in mesenchymal differentiation, genes not previously associated with the process, and some sequences without known function. In addition, we found that the three distinct BMP type I receptors induce an identical genomic response when ectopically expressed in C2C12 cells. It might be highly interesting to examine whether the identified targets are direct or indirect, how their promoters are activated, and what role the encoded gene products play in mesenchymal differentiation and BMP-induced responses in other cell types.

Acknowledgements

We thank Marie-Jose Goumans, Fumiko Itoh, and Susanne Bauerschmidt for fruitful discussion and Christophe Lefèvre and Tim Hulsen for bioinformatics support. This research was supported by the Netherlands Institute for Earth and Life Sciences (ALW 809.67.024) and TMR EC network grant (ERB FMRX-CT98–0216).

Ancillary