Noggin recruits mesoderm progenitors from the dorsal aorta to a skeletal myogenic fate.

Embryonic mesoangioblasts are the in vitro counterpart of vessel-associated progenitors, able to differentiate into different mesoderm cell types. To investigate signals recruiting these progenitors to a skeletal myogenic fate, we developed an in vitro assay, based upon co-culture of E11.5 dorsal aorta (from MLC3F-nLacZ transgenic embryos, expressing nuclear beta galactosidase only in striated muscle) with differentiating C2C12 or primary myoblasts. Under these conditions muscle differentiation from cells originating from the vessel can be quantified by counting the number of beta gal+nuclei. Results indicated that Noggin (but not Follistatin, Chordin or Gremlin) stimulates while BMP2/4 inhibits myogenesis from dorsal aorta progenitors; neutralizing antibodies and shRNA greatly reduce these effects. In contrast, TGF-β1, VEGF, Wnt7A, Wnt3A, bFGF, PDGF-BB and IGF1 have no effect. Sorting experiments indicated that the majority of these myogenic progenitors express the pericyte marker NG2. Moreover they are abundant in the thoracic segment at E10.5 and in the iliac bifurcation at E11.5 suggesting the occurrence of a cranio-caudal wave of competent cells along the aorta. BMP2 is expressed in the dorsal aorta and Noggin in newly formed muscle fibers suggesting that these two tissues compete to recruit mesoderm cells to a myogenic or to a perithelial fate in the developing fetal muscle.

Introduction

Classic transplantation experiments showed, and recent data confirmed that all skeletal myoblasts of the vertebrate body originate from somites (Buckingham et al., 2003; Christ and Ordahl, 1995), blocks of paraxial mesoderm that form and mature in a cranio-caudal sequence along the axial structures of the embryo, i.e. notochord and neural tube. Specifically, myogenic progenitors, identified by the expression of the Pair-ruled gene Pax3, are located in the dorsal dermo-myotome and are specified by Wnt and Shh signals emanating from the axial structures and from the dorsal ectoderm (Cossu and Borello, 1999; Reshef et al., 1998). The dorsal dermomyotome also contains progenitors for dermis, tendons, vascular endothelium and smooth muscle, some of which also express Pax3 and migrate towards the dorsal aorta (Brent and Tabin, 2002; Esner et al., 2006), while the anatomical location of skeletal myogenic progenitors is established, the exact lineage relationship among different cell types of the dorsal somite and their differentiated tissue progeny remains less defined. For example, retroviral lineage marking, unexpectedly but unequivocally demonstrated the existence of a common somitic progenitor for both endothelium and skeletal muscle (Kardon et al., 2002). Moreover, in ovo electroporation experiments have shown that BMP and Notch interfere with somitic cell fate diverting them from skeletal muscle and inducing endothelial and smooth muscle fate respectively (Ben-Yair and Kalcheim, 2008). Thus it appears that in mammalian mesoderm, cell fate is established in response to signaling molecules, locally produced by neighbor, differentiated cells. Interfering with the expression of one or more specific molecules thus results in altered proportion of competent cells undergoing a given differentiation pathway (Shin and O'Brien, 2009). While these reports focused on somites, much less is known on the subsequent phases of pre-natal skeletal muscle histogenesis. If multipotent progenitors exist in the somite and likely in other regions of the mesoderm, they should presumably undergo more than one differentiation pathways. In the last ten years a large number of progenitor cells have been clonally isolated and expanded in vitro from embryonic or adult mesoderm tissues, and shown to be multipotent (Asahara et al., 1997; Asakura and Rudnicki, 2002; De Bari et al., 2003; Minasi et al., 2002; Reyes and Verfaillie, 2001; Rodriguez et al., 2006; Tamaki et al., 2002; Toma et al., 2001; Torrente et al., 2004). With the possible exception of mesenchymal stem cells, little is known on the origin, lineage relationships and differentiation potency of these cells.

Mesoangioblasts were initially isolated from the embryonic dorsal aorta and partially characterized as cells expressing early endothelial and pericyte markers, and able to differentiate into different types of solid mesoderm, both in vitro and also when transplanted in chick embryos in ovo (Minasi et al., 2002) Embryonic mesoangioblasts undergo smooth muscle differentiation if exposed to TGF-β but do not spontaneously differentiate into skeletal muscle. However, if genetically labeled, mesoangioblasts, cultured together with unlabeled differentiating myoblasts undergo fusion and activate expression of muscle genes (Minasi et al., 2002). It is still currently unknown what are the signals released by differentiating muscle cells that activate myogenesis in mesoangioblasts. Here we show that muscle-derived Noggin – an antagonist of BMP-2/4 activity - recruits cells from the dorsal aorta to skeletal myogenesis and this activity is competed by endothelial-derived BMP that rather recruits these cells to a perithelial, smooth muscle fate.

Materials and Methods

Mice

MLC3F-nlacZ transgenic mice express nuclear β-gal under the transcriptional control of the myosin light chain 1/3 F promoter/enhancer (Kelly et al., 1995). In Myf5nlacZ mice nuclear LacZ was targeted to the Myf5 locus (Tajbakhsh et al., 1996). EGFP mice have also been described (Hadjantonakis et al., 1998)

Co-culture of embryonic DA and C2C12 myoblasts

C2C12 myoblasts were plated at sub-confluence (104x ml) as a drop of 50 μl in a 0.5 cm area in the center of individual wells of a 24-well plate. After adhesion to the substrate, a single freshly isolated embryonic DA (dissected from the thoracic upper segment to the iliac bifurcation) from MLC3F-nlacZ embryo (Minasi et al., 2002) was added, and covered by a drop of Matrigel™ diluted 1:4. The co-culture was maintained in growth medium (DMEM + 10% FBS) for three days and then shifted to differentiation medium (DMEM + 5% horse serum). After three additional days the co-culture was fixed with paraformaldehyde 4% and then incubated with X-gal staining solution overnight at 37 °C. C2C12 myoblasts, 10 T1/2 fibroblasts, D16 mesoangioblasts and H5V endothelial cells were described before (Minasi et al., 2002). In some of these experiments, cells were labeled with BrdU (5 μM) in complete medium for 2 hours at different days of culture and in different experimental conditions.

DA-derived cells culture

Aorta-derived single cells were obtained by digestion of freshly isolated DA (E11.5) in PBS without calcium-magnesium containing 0.45 mg/ml of collagenase V (Sigma) and 0.15 mg/ml of dispase (Gibco) for 40 min at 37 °C. After recovery in 20% FBS containing growth medium, cells were either cultured for 12 hours and then stained with the antibodies described below, or separated by FACS sorting analysis (see below) and then co-cultured with C2C12 myoblasts as described above.

Growth Factors

Growth Factor Reduced Matrigel™ Matrix was purchased from BD Bioscience. Recombinant human Noggin/Fc chimera and neutralizing anti-human BMP-2/4 antibody from R&D Systems, Inc. BMP-2 and BMP-4 from Peprotech.

Cryosections

Mouse embryos were fixed in 4% PFA, washed in PBS, dehydrated by washing them in PBS containing increasing concentration of sucrose (10, 20 and 30%), included in O.C.T™ and frozen in liquid nitrogen-cooled isopentane. 7 μm-thick sections were cut with a Leica cryostat.

Immunofluorescence

Adherent cell cultures, co-cultures and cryosections were processed with the same standard protocol: after fixation with 4% PFA, samples were incubated with PBS 1% BSA 0.1% Triton X-100 (1 h at RT), blocked with donkey or goat serum 10% in PBS (30 min at RT), incubated with the primary antibody (1 h at RT), washed three times and incubated with the secondary antibody (Alexa Fluor® conjugated (Invitrogen), diluted 1:500, 1 h at RT). Then the samples were washed three times and incubated with Hoescht (1 mg/ml × 5 min at RT), washed and mounted. The primary antibodies used were: rabbit anti BMP-2 (1:250, Abcam), rat anti CD31 (1:2, Hybridoma Bank), mouse anti MyHC (1:2, MF20, Hybridoma Bank), goat anti Noggin (1:100, R&D Systems), mouse anti smooth muscle heavy chain (1:250, Abcam), rabbit anti NG2 (1:200, Chemicon), goat anti β-galactosidase (1:300, Biogenesis), rabbit anti Myf5 (1:150, Santa Cruz), mouse anti MyoD (1:200, Dako), anti-Phospho-Smad 1-5-8 (1:200 Cell Signaling Technology®), anti-BrdU (1:100 Amersham GE Healthcare).

In situ hybridization

Previously published probes were: Bmp4 (Furuta et al., 1997), and Noggin (Ybot-Gonzalez et al., 2007) In situ hybridization was carried out as described by (Henderson et al., 1999) with samples fixed for 2 h in 4% paraformaldehyde/phosphate buffered saline, then rinsed with PBS-Tween, dehydrated in 100% methanol and stored at − 20 °C until processed for WISH. Antisense riboprobes were previously in vitro labelled with modified nucleotides (digoxigenin, Roche).

Semiquantitative RT-PCR

Total RNA was isolated from cells using TRIzolr (Invitrogen) or RNeasyr Micro Kit (Quiagen). cDNA was synthesized from total RNA previously treated with DNAse I, primed with random primers, and then reversed transcribed with Moloney murine leukemia virus reverse transcriptase. PCR protocols for amplify transcripts for BMP-2, Noggin and BMPR1A were performed as described in (Goulley et al., 2007; Hager-Theodorides et al., 2002; Van der Horst et al., 2002).

Primers

Agarose beads

BMP-2 and BMP-4 were loaded in agarose beads (Affi-GelR Blue Gel, BioRad) incubating beads with growth factors at 3 μg/mL in PBS 0.1% BSA for 1 h at RT and then washed three times with PBS.

Transfection

C2C12 myoblasts were transfected with the shRNA for Noggin containing plasmid using Lipofectamine™ and PLUS™Reagent (Invitrogen) and further selectioned with puromycin (2 μg/ml) for three days.

Lentiviral vectors and cell transduction

shRNAs for mouse Noggin and mouse BMP receptor type 1A were purchased from Open Biosystems (Thermo Fisher Scientific Inc., USA). A third-generation lentiviral vector was produced as described in (Sampaolesi et al., 2003). The lentiviral particles were produced by transient transfection of the vector in association with the packing vectors (pREV, pD8.74 and pVSV-G) in HEK293T. After 30 h, the culture medium from transfected cells was filtered with a 0.22 μm filter and 250-times concentrated after centrifugation at 20.000 rpm for 2 h at 20 °C. The aorta-derived cells, plated 24 hours before, were infected with the lentivirus containing the shRNA for BMPR1A at a multiplicity of infection (MOI) of 400 for 12 h. Under these conditions, more that 90% of the cell population was efficiently transduced. As a control, a lentiviral vector expressing GFP expression was used (Sampaolesi et al., 2003).

Real time PCR

The RNA from infected cells was collected by RNeasy mini (or micro) kit (QIAGEN) converted into double-strand cDNA with the cDNA system kit ThermoScript RT-PCR (Invitrogen), according to the manufacturer's instructions. Real-time quantitative PCR analysis was carried out on cDNA from isolated cells by using an Mx3000P real-time PCR detection system (Stratagene, La Jolla, CA). Each cDNA sample was amplified in triplicate by using the SYBR Green Supermix (Invitrogen). Gene expression profiling was achieved using the Comparative CT method (DDCT) of relative quantification (Livak and Schmittgen, 2001).

FACS analysis

Aorta-derived single cells were labeled with PE-conjugated rat anti-mouse CD31 (BD Bioscience) and rabbit anti NG2 / secondary anti rabbit IgG FITC (Sigma) and then sorted in MoFlo System (DAKO).

Alkaline Phosphatase Staining

Fixed cells were incubated in alkaline buffer containing NBT/BCIP chromogenic solutions (Roche) for 30 min at RT.

Results

Embryonic dorsal aorta contains a cell population able to differentiate in skeletal muscle

To study the regulation of skeletal myogenic differentiation of aorta derived cells in vitro, co-cultures of freshly isolated embryonic dorsal aorta (DA) and C2C12 myoblasts were performed as described in Fig. 1A. Freshly isolated DA was dissected at E11.5 from MLC3F-nlacZ mouse embryos that express nuclear lacZ under the transcriptional control of the MLC1/3 F promoter of the myosin light chain gene that restricts expression to striated muscle (Kelly et al., 1995). Therefore, skeletal muscle differentiation of aorta-derived cells could be quantitatively analyzed by counting β-gal + nuclei, usually detected inside C2C12-derived myotubes. As shown in the Fig. 1B, after 6 days of co-culture, several aorta-derived cells (an average of 50 ± 15/dorsal aorta; n = 7) differentiate into skeletal muscle, likely by fusing with C2C12 myoblasts. This process was only observed in co-cultures of DA with myoblasts. When DA was cultured with 10 T1/2 fibroblasts (n = 3, Fig. 1C) or embryonic neural tube (n = 3, data not shown) no β-gal + nuclei were detected. This myogenic sub-population is not contaminated by somite-derived Myf5+ embryonic myoblasts because the dorsal aorta, freshly dissected from E 11.5 Myf5-nlacZ embryos does not contain any β − gal+ cell (Suppl. Fig. 1). At late stages of myogenic differentiation, all DA derived β-gal + nuclei contained in C2C12 myotubes expressed MyoD and/or Myf5, indicating that, as expected, skeletal myogenesis is related to Myogenic Regulatory Factor (MRF) expression (Suppl. Fig. 2).

Fig. 1

Co-culture assay to test skeletal myogenic potency of embryonic dorsal aorta. A. Scheme showing the cell assay used. B,C. X-gal staining of co-cultures of embryonic dorsal aorta (from MLC3F-nlacZ embryos at E11.5) with C2C12 myoblasts (B) or with 10 T1/2 fibroblasts (C). Myogenic differentiation of aorta-associated progenitors was scored by counting β-gal + nuclei (arrows). Scale bar: 100 μm.

Primary embryonic and fetal myoblasts are more potent than C2C12 myoblasts in activating skeletal myogenesis in cells from the DA

C2C12 are an established cell line and, although widely used for studying myogenesis, they may not entirely correspond to primary myoblasts, also because they were derived from adult satellite cells. To verify that the myogenic inductive potency was also present in myogenic cells freshly isolated from embryos or fetuses, we repeated the experiment by co-culturing DA from MLC3F-nlacZ mouse embryos with primary myoblasts purified from Myf5GFP embryos and fetuses at E11.5 and E15.5 respectively. The results indicated that primary embryonic myoblasts were approximately 8 times (Figs. 3A,D) and fetal myoblasts 4 times (Figs. 3C,D) times more effective than C2C12 in activating myogenesis of DA cells. Since adult primary satellite cells appeared equivalent to C2C12 myoblasts in their ability of activating myogenesis in DA cells (data not shown), it is likely that this potential is inversely related to developmental age. However, due to the difficulty of synchronizing mating as to obtain MLC3F-nlacZ and Myf5GFP mouse embryos of the right age on the same day, we carried out most of the experiments with C2C12 myoblasts and repeated critical ones with primary Myf5-GFP cells.

Fig. 3

Primary embryonic and fetal myoblasts are more potent than C2C12 cells in promoting myogenic differentiation of aorta derived cells. A-B. X-gal staining of co-cultures of embryonic DA and E11.5 Myf5GFP purified embryonic myoblasts under control conditions (A) and after addition of BMP-2 (3 μM) (B). C. X-gal staining of co-cultures of embryonic DA and E16.5 Myf5GFP purified fetal myoblasts under control conditions. Scale bars: 200 μm. D. Quantification of differentiation index (number of β-gal + nuclei in myotubes or myocytes) in co-culture of DA cells with embryonic myoblasts (± BMP-2 or ± Noggin) or with fetal myoblasts. An average of four different co-cultures (each experimental condition in triplicate) is shown.

Neutralization of BMP-2/4 activities enhances myogenic differentiation of aorta-associated progenitors

To identify growth factors that may regulate skeletal myogenesis of DA cells we exposed the co-cultures to a large number of growth factors and cytokines such as Wnt, FGF, Noggin and BMP inhibitors that are known activate or enhance skeletal myogenesis and others such as TGF-β, VEGF, PDGF-BB, BMP that rather inhibit myogenesis and also influence recruitment and differentiation of mural cells during vessel development (Armulik et al., 2005). Addition of BMP-2 to the culture medium was not possible since the cytokine also inhibits C2C12 myogenesis. Therefore agarose beads loaded with BMP-2 were added close to the DA and were found to significantly inhibit skeletal muscle differentiation of DA cells (Figs. 2A,C, n = 3). Consistent with this observation, addition to the co-culture of a neutralizing antibody specific for BMP-2/4 (Fig. 2D, n = 3) caused a six fold enhancement of myogenesis. Furthermore, we took advantage of the fact that primary embryonic myoblasts are resistant to the inhibitory effect of BMP (Biressi et al., 2007) and thus cultured E11.5 Myf5GFP purified myoblasts with DA from MLC3F-nlacZ mouse embryos in the presence of BMP-2. As shown in Fig. 3B, many thin oligonucleated, typical embryonic myotubes formed in the presence of BMP-2 but the number of β-gal+, DA derived nuclei was dramatically reduced, indicating that, as expected, BMP-2 does not inhibit embryonic myogenesis but prevents the ability of embryonic myoblasts and myotubes to activate skeletal myogenesis in DA cells (compare Figs. 3A with B).

Fig. 2

Neutralization of BMP-2/4 activities enhances myogenic differentiation of aorta derived cells. A-C X-gal staining of co-cultures of embryonic DA and C2C12 myoblasts under control conditions (A), after addition of Noggin (0.5 μg/ml) (B) and in the presence of BMP-2 (3 μM) loaded beads (C). Scale bars: 200 μm. D. Quantification of effect of Noggin, anti BMP-2/4 neutralizing antibody and BMP-2 on differentiation index (number of β-gal + nuclei in C2C12 myotubes). The average of fifteen different co-cultures (each experimental condition in triplicate) is shown. Differences were found to be significant: noggin, P = 0.0036; anti BMP-2/4, P = 0.0126; BMP-2, P = 0.0096, using unpaired two-tailed t-Test.

On the other hand, the addition of Noggin to the co-cultures enhanced fourfold muscle differentiation of DA derived cells (Fig. 2B, n = 3). This was not due to enhanced myogenesis of C2C12 cells, since their differentiation was similar with or without Noggin (Suppl. Figs. 3A,B). To test whether Noggin induces skeletal myogenesis in a larger number of DA cells, or rather promote their proliferation, we administered BrdU to co-cultures, with or without Noggin, for 3 hours on day 2 and 4. The results showed that at day 6, almost all β-gal+, control or treated cells had incorporated BrdU (arrows show examples in Suppl. Fig. 4), thus suggesting that a major proliferative effect of Noggin is unlikely, also considering that at this embryonic stage cells proliferate actively in serum containing medium. Interestingly, addition of Noggin to co-cultures of primary Myf5GFP myoblasts and DA did not significantly enhance myogenesis (Fig. 3D), suggesting that no further enhancement of skeletal myogenesis was possible, at least under these experimental conditions. Assuming that a dissected dorsal aorta contains approximately 50,000 cells, we may indirectly conclude that approximately 1% of the total cell population would be competent to respond to myogenesis inducing signals such as Noggin.

On the other hand, other BMP-2/4 antagonists such as follistatin, chordin and gremlin had no effect (n = 3, data not shown). Similarly Wnt3A, Wnt7A, TGF-β1, VEGF and PDGF-BB had no effect (n = 3, data not shown).

BMP-2 is expressed by endothelial cells in embryonic dorsal aorta

To determine if BMP-2 is expressed in the embryonic DA, immunofluorescence, in situ hybridization and semiquantitative RT-PCR assays were performed. Fig. 4A shows that isolated dorsal aorta and associated inter-somitic arteries are brightly stained by an anti-CD31 antibody (green) and also by an antibody (red) recognizing BMP-2/4 (Fig. 4B). This is also evident in a transverse section showing a merged image (Fig. 4C). Since immunostaining of tissue section with antibodies against secreted molecules exposes to the risk of background staining (despite very low signal detected with secondary antibody alone) we confirmed this observation by a number of methods: semiquantitative RT-PCR assay shows that the transcripts for BMP-2 are present in the embryonic DA and H5V murine endothelial cell line but not in differentiating or differentiated C2C12 cells (Fig. 4D). In situ hybridization with a probe recognizing BMP2 strongly stained isolated dorsal aorta (appearing violet in Fig. 4E, blood appears red inside) indicating that the message is transcribed as expected. Finally, almost all freshly isolated cells, obtained by enzymatic digestion of freshly isolated DA that were stained by an anti-CD31 antibody (green) were also stained by the anti BMP antibody (red) as shown in Figs. 4F-H. These results were confirmed in a functional assay. Suppl. Fig. 5 shows that the addition of conditioned medium from ten DA, cultured as explants, to C2C12 myoblasts induced inhibition of myogenesis and expression of the pericyte marker alkaline phosphatase, similar to the addition of BMP-2 to the culture medium. This effect was reversed by the simultaneous addition of neutralizing anti BMP2/4 antibody, thus confirming that cells of the aorta explants secrete BMP2.

Fig. 4

BMP-2 is expressed in embryonic dorsal aorta. Immunofluorescence for CD31 (A) and BMP-2 (B) of a dissected dorsal aorta from a mouse embryo at E11.5; merged image of a transverse cryosection of a similar embryo is show in C. BMP is localized in and around the endothelium of dorsal aorta. Scale bar: 80 μm. D. RT-PCR analysis for BMP-2 expression in different cells lines and embryonic DA. MW: molecular markers; H5V: mouse endothelial cell line; D16: mouse mesoangioblast clone; C2mb: C2C12 myoblasts; C2mt: C2C12 myotubes; DA: embryonic Dorsal Aorta. E: In situ hybridization with a BMP2 probe of a dissected DA, with adjacent mesoderm. Note that signal (violet) is in the internal region surrounding the residual blood (red). F-H. Immunofluorescence for CD31 and BMP-2 in embryonic DA-derived cells after 12 hours in culture. Note co-expression of CD31 + and BMP-2 (F). Scale bar: 15 μm.

Noggin is expressed during myoblast differentiation.

Noggin is expressed during myoblast differentiation

Noggin is a secreted protein that binds with high affinity and antagonizes the biological activity of several BMP proteins such as BMP-2, BMP-4 and BMP-7 (Zimmerman et al., 1996). Since Noggin did enhance skeletal myogenesis of DA derived cells in vitro, we performed immunofluorescence and semiquantative RT-PCR analysis for Noggin in differentiating myoblasts and myotubes during in vitro differentiation and in vivo mouse limb development. Figs. 5A-B show that, Noggin is immune-detected predominantly in MyHC + differentiated myotubes in vitro (A, B) and in MyHC + newly formed myofibers in vivo at E 14.5 (C-D). Likewise, transcripts for Noggin markedly increase during differentiation of C2C12 myoblasts (lanes 1–3, Fig. 5E) but are absent in 10 T1/2 fibroblasts (lane 5). These results demonstrate that this BMP antagonist is expressed by differentiating skeletal muscle cells in vitro and in vivo.

Fig. 5

A,B. Immunofluorescence for Noggin (B) in C2C12 myotubes at day 3 of differentiation. Reactivity for Noggin co-localizes with myosin heavy chain (MyHC) in multinucleated myotubes (A). Merged image is shown in the inset. Scale bar: 50 μm. C. Immunofluorescence for Noggin in a cryosection of a limb of a mouse embryo (E14.5). The arrow shows one MyHC+ (C) muscle fiber showing reactivity for Noggin (D). Merged image is shown in the inset. E. RT-PCR analysis for Noggin expression in C2C12 myoblasts at day 1, 3 and 5 of differentiation (lanes 1–3). Noggin is not expressed in 10 T1/2 fibroblasts (lane 5). Lane 4 is the negative control of the PCR reaction.

BMP-4 induces aortic mural cells to differentiate in smooth muscle

To evaluate whether BMP-2/4 inhibits differentiation of DA derived in skeletal muscle because it drives differentiation towards the default smooth muscle phenotype, we analyzed co-cultures of DA with C2C12 in which BMP-4 was loaded onto beads as described in Fig. 2. Figs. 6A,B shows that many cells, adjacent to BMP-4-loaded beads, differentiated into smooth muscle as revealed by immune detection of smooth muscle myosin heavy chain (SM-MyHC). To distinguish between DA-derived and C2C12 cells, in a subsequent experiment the DA was dissected from EGFP/MLC3F-nlacZ embryos. Figs. 6C-E shows that the expression of smooth muscle myosin heavy chain (SmMyHC) is associated to GFP + aorta-derived cells, even though not all the GFP + cells expressed SmMyHC. To rule out the possibility that BMP may activate smooth muscle genes in C2C12 cells, we treated these cells alone with BMP-2 and after 5 days we analyzed by RT-PCT the expression of the smooth muscle marker Calponin. As shown in Fig. 6F, BMP-2 did not induce expression of this gene in C2C12 cells, that readily express SMA as the untreated cells. These results indicate that BMP-2 stimulates the differentiation of DA-derived cells into smooth muscle.

Fig. 6

BMP-4 induce aortic mural cells to differentiate in smooth muscle. A. X-gal staining of a co-culture in which BMP-4-loaded agarose beads (arrows) were added close to the dissected DA. One myotube containing few Β-gal + nuclei is shown by the arrowhead. Scale bar: 100 μm. B. immunofluorescence for smooth muscle myosin heavy chain (SM-MyHC) of the same co-culture (the area shown in the dot line square in A). Scale bar: 60 μm. C-E. Immunofluorescence for smooth muscle myosin heavy chain in a co-culture of a HGFP; MLC3F-nlacZ dorsal aorta and C2C12 myoblasts. GFP + cells (and DAPI) are shown in C, SM-MyHC + cells (red) in D; the double positive cells, shown in the merged image in E originate from the aorta. Scale bar: 50 μm. F. RT-PCR of Control and BMP-2 treated C2C12 cells for the expression of Calponin (Calp), Smooth Alpha Actin (SMA) and Gapdh. RNA extracted from E11.5 embryo (E) was used as a positive control.

Myoblast-derived noggin is required for in vitro myogenic differentiation of aorta-derived cells that is also enhanced by silencing of BMPR1A

To determinate whether Noggin synthesized by muscle cells is required for differentiation of DA derived cells we knocked-down Noggin in C2C12 myoblasts by transfection of a specific siRNA-containing plasmid. Fig. 7A shows that under these conditions, Noggin transcripts level decreased to approximately 55% of control myotubes (evaluated by band quantification of RT-PCR products and Real Time PCR: Fig. 7B). Fig. 7C shows that the differentiation of DA derived cells is strongly inhibited (< 40% of control) when the embryonic DA are plated with C2C12 transfected with the siRNA-contained plasmid. The process of differentiation of C2C12 myoblasts transfected with the shRNA containing plasmid was not compromised since they formed multinucleated myotubes as control cultures (Suppl. Figs. 3A,C). On the other hand, silencing BMP signaling should enhance myogenesis of DA derived cells. BMP-dependent signaling is mediated by two membrane receptors BMPRI and BMPRII, which dimerize after ligand binding and activate a Smad1,5,8-dependent cascade (ten Dijke et al., 2003). Fig. 7A shows that after transduction of the Dorsal Aorta with a lentiviral vector expressing a specific shRNA for the BMP receptor type 1A, the levels of this transducer receptor for BMP-2 was reduced to 40% of control myotubes (evaluated by band quantification of RT-PCR products and Real Time PCR: Fig. 7B) and the differentiation of DA derived cells into skeletal muscle was enhanced twofold (Fig. 7C). Altogether these results suggest that interference with the BMP pathway, either by modulation of Noggin levels or by downregulation of the BMP membrane receptor, modulates the skeletal myogenic potency of DA derived cells.

Fig. 7

Silencing of Noggin in myoblasts and of BMPRIA in dorsal aorta. A. Semi-quantitative RT-PCR for Noggin of total RNA from C2C12 myoblasts transfected with a plasmid expressing either GFP or shRNA for Noggin. B. Semi-quantitative RT-PCR for of BMP receptor 1A of total RNA from explanted Dorsal Aorta transduced with a lentivector expressing either GFP or SiRNA for BMPR1A. Data show one representative of four different experiments. Differences were found to be significant (P = 0.007 and P < 0.001, respectively) using unpaired two-tailed t-Test. C. Real-time quantitative RT-PCR for Noggin and BMPR1A of total RNA of C2C12 and aorta-derived cells transduced with silencing vectors as described in A,B. Data are the average (± SD) of four different experiments. Differences were found to be significant (P < 0.002 and P < 0.001, respectively) using unpaired two-tailed t-Test. D. Quantification of the effect of Noggin or BMPR1A silencing on myogenic differentiation of aorta-derived cells co-cultured with C2C12 myoblasts. Data plotted correspond to four different paired co-cultures after four different pairs of infections. Difference was found to be significant (P = 0.015 and P = 0.024) using unpaired two-tailed t-Test .

Aorta derived cells endowed with myogenic potency express the NG2 proteoglycan

As shown in Fig. 8A, mouse embryonic DA at E11.5 is composed basically by CD31 + endothelial and by NG2 + perivascular cells. NG2 is a chondroitin sulfate proteoglycan described as a cell marker for embryonic and adult pericytes (Ozerdem and Stallcup, 2004). To determine which cell type of the embryonic DA express a skeletal myogenic potency, a fluorescence activated cell sorting (FACS) analysis of DA-derived cells was performed. FACS analysis of single cell suspension obtained by mild enzymatic digestion of freshly isolated DA showed that approximately 20% of the total population expressed NG2, 6% CD31 and 4% both antigens (Fig. 8B). After cell sorting, the same number of cells (5 × 103) from each sub-population was co-cultured with C2C12 myoblasts and the number of β-gal + nuclei in C2C12 myotubes was evaluated. These experiments showed that the NG2 + cell fraction contains the large majority of the myogenic progenitors (Fig. 8C). Many freshly isolated NG2 + cells were found to express phospho-SMAD 1-5-8 (Fig. 8D) suggesting that, within the aorta wall, these cells respond to BMP, and are therefore likely primed to a smooth muscle fate, to which they may escape if they migrate away from the endothelium.

Fig. 8

Aorta-derived cells endowed with skeletal myogenic potency express NG2. A. Immunofluorescence for CD31 and NG2 expression in a transversal cryosection of a dorsal aorta (mouse embryo at E11.5). Scale bar: 50 μm. B. A representative FACS profile of freshly isolated DA-derived cells, labeled with antibodies against CD31 or NG2. C. Histogram showing the myogenic potential of each sorted fraction from three different experiments, evaluated as the number of β-gal + nuclei contained in C2C12 myotubes, normalized by the number of sorted cells. Values plotted correspond to averages ± SD. Cells generating β-gal + clones were significantly enriched in the NG2 + fraction (vs all other positive fractions: ***P < 0.001) by the ANOVA Test. DN: Double Negative; DP: Double Positive. D. Double IF of NG2 + cells, co-stained with an antibody recognizing phospho-SMAD 1-5-8, showing co-expression in the majority of the cells.

Aorta-associated progenitors with myogenic potency are distributed along a cranio-caudal, temporal gradient

We investigated whether DA cells able to activate skeletal myogenesis are equally distributed along the cranio-caudal axis of the embryonic dorsal aorta. To this aim the DA were dissected from MLC3F-nlacZ embryos at different stages of development and further subdivided in a thoracic, an abdominal and a pelvic segment as indicated in the upper scheme of Fig. 9. Each segment taken from the same DA (n = 3), was cultured with C2C12 myoblasts and, after differentiation, the number of β-gal + cells was recorded. Results indicated that at E10.5 the large majority of myogenic cells is present in the most cranial segment, with very few myogenic cells present in the other segments (Fig. 9 upper lane). In contrast, when the same experiment was repeated at E12.5 the results were completely different, with a very large predominance of myogenic cells in the iliac segment (Fig. 9, lower lane). Interestingly, the abdominal segment was always almost devoid of myogenic cells, even at E11.5, when one would predict that the wave of myogenic cells should be present (Fig. 9 middle lane). These results suggest that proliferating competent cells able to respond to skeletal myogenic induction are unequally distributed along the cranio-caudal axis and this distribution changes with time, roughly corresponding to the cranio-caudal wave of skeletal myogenic differentiation of somite-derived myoblasts in the embryonic trunk (Tajbakhsh and Buckingham, 1994). The lower amount of β-gal + nuclei associated to abdominal region of DA suggests the presence of some inhibitory factor of muscle differentiation of DA derived cells, likely the high levels of BMP-2 (or BMP-secreting cells) in the AGM region.

Fig. 9

Distribution of myogenic aorta-associated progenitors follows a cranio-caudal wave during embryonic development. A. Scheme depicting the cranio-caudal level of dissection of the dorsal aorta. B-J. X-gal staining of co-cultures of C2C12 myoblasts with the different regions of DA (A, anterior B,E,H; M, medial, C,F,I; P, posterior, D,G,J) as shown in the upper drawing. Co-cultures of corresponding segments of DA dissected from embryos at E10.5 (B-D), E11.5 (E-G) and E12.5 (h-J) are shown. Scale bar: 80 μm. K. Quantification of β-gal + nuclei in co-cultures with C2C12 myotubes at the three different stages of development. Number in the ordinate show the n° of β-gal + nuclei/ explanted segment of the dorsal aorta. Each dot represents the result of a separate experiment, each performed in triplicate.

Discussion

The identification of signaling molecules that instruct competent cells to adopt one or another cell fate remains a central issue in developmental biology.

In the case of skeletal muscle the main signaling molecules that activate myogenesis in the somites and the responding genes were identified and characterized in the last twenty years (reviewed in Buckingham et al., 2003). It is currently assumed that commitment occurs in somites for all myogenic progenitors and thus it remains to be understood why certain progenitors differentiate immediately to form the myotomes while others proliferate and differentiate later during pre- and post-natal development. For example Pax3+/MRF- myoblasts at E11.5, once explanted in vitro differentiate immediately into thin myotubes, thus suggesting that they do not require (and likely had already received) instructive signaling molecules needed to initiate myogenesis. The situation is different for a number of ill-defined mesoderm cells that do not spontaneously undergo myogenesis but do so if co-cultured with genetically distinguishable bona-fide myogenic cells. These “occasional” myogenic progenitors (Cossu and Biressi, 2005) are clearly competent to be recruited to myogenesis but are not committed to a myogenic fate. They may represent reserve cells to be utilized during development or tissue repair but the precise role and the extent of their contribution are still largely unknown.

Here we report that a sub-population of mesoderm cells, physically associated to the embryonic dorsal aorta, can undergo myogenic differentiation if co-cultured with differentiating myogenic cells. Contact is necessary, as conditioned medium or co-culture with physical separation (trans-well) are not sufficient to induce myogenesis (Minasi et al., 2002) and unpublished results), and fusion is the most likely underlying mechanism as most β-gal + nuclei are found inside multinucleated myotubes. Even when DA cells are co-cultured with primary embryonic myoblasts, that form thin oligonucleated myotubes, more than one DA-derived nucleus are often found in the same cytoplasm, even though, mononucleated, Β-gal + cells can be detected. Interestingly, when DA cells are co-cultured with embryonic myoblasts, approximately 8 times more cells are recruited to myogenesis, in comparison with C2C12 cells, whereas this value is decreased to fourfold when the co-culture is performed with fetal myoblasts, suggesting that the ability of newly forming myotubes to recruit DA cells is inversely proportional to developmental age; indeed primary adult satellite cells recruit to myogenesis approximately the same number of cells than C2C12 which were derived from mouse adult satellite cells.

In this work we show that this process is positively regulated by Noggin (but not by other BMP inhibitors) and negatively by BMP2 that rather directs DA cells to a perithelial/smooth muscle fate. It is somewhat surprising that other BMP inhibitors such as Chordin, Follistatin and Gremlin did not show any activity much as other unrelated signaling molecules. It will be interesting to investigate the mechanism through which Noggin activates skeletal myogenesis in DA cells: our current data only suggest that Noggin may not promote proliferation of responsive DA and thus possibly activates MRF or upstream genes in these cells and/or inhibits expression of genes that drive smooth muscle differentiation such as Myocardin or Necdin (Brunelli et al., 2004; Hu et al., 2003) beside inhibiting BMP activity, as known.

Several reports indicate that BMP-2 and BMP-4, in addition to their canonical role as osteogenic factors, can also induce smooth muscle markers (Lagha et al., 2009) and it is well known that BMP signaling inhibits skeletal myogenesis (Duprez et al., 1996; Yamaguchi et al., 1991). BMP is expressed in the endothelium at the proper time as previously reported and confirmed in this study (Durand et al., 2007) and thus may play this role in vivo by promoting smooth at the expense of skeletal myogenesis. On the other hand we show here that Noggin is expressed in developing muscle fibers in vivo, consistent with a possible role in recruiting aorta cells to skeletal myogenesis. Moreover, gain and loss of function experiments by both neutralizing antibodies and silencing RNA support this conclusion. Unfortunately, the relative inaccessibility of the mammalian embryo in this specific period of development prevents in utero experiments that have been rather performed in ovo with the chick embryo.

Sorting experiments identified the DA cells endowed with this myogenic potency as pericytes. These data confirm previous transplantation experiments in the chick embryo (Minasi et al., 2002). However it should be considered that pericytes are still poorly characterized and likely heterogenous cells, part of which may derive from endothelial cells. In vivo lineage tracing experiments are in progress to define the fate of early endothelial cells (Azzoni et al. under revision) but similar experiments are not feasible for pericytes since none of the genes expressed in these cells (NG2, alkaline phosphatase, smooth muscle actin, desmin, PDGF receptor β, etc.) is unique to pericytes, at least pre-natally.

Finally, we report that DA cells are not homogeneously distributed along the cranio-caudal axis of the dorsal aorta but appear to follow a cranio-caudal wave that roughly corresponds to the cranio-caudal progress of skeletal myogenesis along the main body axis (Tajbakhsh and Buckingham, 1994). This would suggest that developing skeletal muscle recruits mesoderm progenitors to a myogenic fate at the time when primary multinucleated fibers are forming by fusion of myoblasts. Right after formation of skeletal muscle, new vessels grow in between newly formed fibers and, as known, endothelial cells recruit mesoderm progenitors to a pericyte fate by secretion of PDGF-BB. Thus it appears that a competition exists in the developing skeletal muscle between endothelium and differentiating muscle to recruit mesoderm progenitors to a perithelial or to a myogenic fate and that the local relative concentrations of BMP and Noggin, that likely reflect physical contiguity, dictate this choice (Fig. 10).

Fig. 10

Proposed model for cell fate regulation of vessels-associated progenitors. During embryonic development, cell fate of vessel-associated progenitors depends upon a competition between BMP-2, synthesized by the endothelium (inhibiting skeletal myogenesis and promoting smooth myogenesis) and its antagonist Noggin, secreted by myofibers and recruiting progenitors to skeletal myogenesis.

Conclusions

These data suggest that during tissue histogenesis, endothelial cells and newly formed skeletal muscle fibers compete to recruit mesoderm progenitors to different fates (pericytes vs skeletal muscle) through a balance between BMP2 and Noggin.

BMP-2 Fw	TGTGACCAGACTATTGGACACC	(454 bp)
BMP-2 Rv	AGTTCAGGTGGTCAGCAAGG
GAPDH Fw	TTCACCACCATGGAGAAGGC	(238 bp)
GAPDH Rv	GGCATGGACTGTGGTCATGA
Noggin Fw	TGAGCAAGAAGCTGAGGAGG	(334 bp)
Noggin Rv	GGATCGATCAAGTGTCTGGG
BMPR1A Fw	GAAGTTGCTGTATTGCTGA	(216 bp)
BMPR1A Rv	GTAATACAACGACGAGCC