Bone morphogenetic protein (BMP) receptor type 1A (BMPR1A) mutations are associated with facial dysmorphism, which is one of the main clinical signs in both juvenile polyposis and chromosome 10q23 deletion syndromes. Craniofacial development requires reciprocal epithelial/neural crest (NC)-derived mesenchymal interactions mediated by signaling factors, such as BMP, in both cell populations. To address the role of mesenchymal BMP signaling in craniofacial development, we generated a conditional knockdown mouse by expressing the dominant-negative Bmpr1a in NC-derived cells expressing the myelin protein zero(Mpz)-Cre transgene. At birth, 100% of the conditional mutant mice had wide-open anterior fontanelles, and 80% of them died because of cleft face and cleft palate soon after birth. The other 20% survived and developed short faces, hypertelorism and calvarial foramina. Analysis of the NC-derived craniofacial mesenchyme of mutant embryos revealed an activation of the P53 apoptosis pathway, downregulation of both c-Myc and Bcl-XL, a normal growth rate but an incomplete expansion of mesenchymal cells. These findings provide genetic evidence indicating that optimal Bmpr1a-mediated signaling is essential for NC-derived mesenchymal cell survival in both normal nasal and frontal bone development and suggest that our model is useful for studying some aspects of the molecular etiology of human craniofacial dysmorphism.
Bone morphogenetic proteins (BMPs) function via conserved type 1 and type 2 transmembrane receptors to regulate a range of biological processes, including cell proliferation, apoptosis, differentiation and cell shape, in a highly context-dependent manner (Massagué, 2000; Chen et al., 2004; Aubin et al., 2004; Kishigami and Mishina, 2005; Eblaghie et al., 2006).
Humans with germ line BMP receptor 1A (BMPR1A) mutations that produce truncated receptors suffer from juvenile polyposis syndrome and facial defects (OMIM ID: 174900) (Zhou et al., 2001). Truncated BMPR1A might act via dominant-negative mechanisms. Furthermore, chromosome 10q23 deletion syndrome (OMIM ID: 612242), which is associated with BMPR1A deletion, is marked by facial dysmorphism (Delnatte et al., 2006; Menko et al., 2008).
Vertebrate facial development starts with the emergence of five facial primordia: a frontonasal prominence and the paired maxillary and mandibular processes. These primordia mainly consist of neural crest (NC)-derived mesenchyme covered by epithelium (Chai and Maxson, 2006). Whereas the processes grow out in conjunction with regulated mesenchymal cell proliferation and apoptosis (Minkoff, 1980; Beverdam et al., 2001), the paired lateral and medial nasal processes bilaterally bulge at the frontonasal prominence. Then, two fusions occur: one in the midline between the right and left medial nasal processes and the other laterally between the maxillary and nasal processes. Cleft face is caused by the former fusion defect and cleft lip by the latter.
Bmpr1a is broadly expressed and its ligands, Bmp2, Bmp4 and Bmp7, are expressed at specific developing nasal regions (Danesh et al., 2009; Furuta and Hogan, 1998; Hu and Marcucio, 2009; Panchision et al., 2001). Previously, researchers generated three conditional Bmpr1a knockout mouse lines (Liu et al., 2005; Nomura-Kitabayashi et al., 2009; Stottmann et al., 2004). However, two mice died in the late embryonic stage and the other displayed no recombination in the mesenchymal cells of the nasal processes. To overcome these issues, we have established a new Bmpr1a-mediated signaling knockdown mouse line in NC cells and confirmed that the signal is involved in craniofacial developmental processes.
Expression and effect of dominant-negative Bmpr1a (Y176STOP) mutant in NC cells
The dominant-negative Bmpr1a protein (dnBmpr1a), which lacks the intracellular kinase domain, inhibits the Bmpr1a-mediated signaling pathway in vivo (Suzuki et al., 1994; Maéno et al., 1994). We generated a Tg(CAG-flox-dnBmpr1a-NLLacZ)1Nobs mouse line using the construct pCAG-XstopX-dnBmpr1a-IRES-NLLacZ (Fig. 1A). The Tg(CAG-flox-dnBmpr1a-NLLacZ)1Nobstg/tg mice were normal. To generate Tg(CAG-flox-dnBmpr1a-NLLacZ)1Nobstg/+;Tg(Mpz-cre)94Imegtg/+ mice (hereafter referred to as double-tg) expressing dnBmpr1a in NC cells, we crossed Tg(CAG-flox-dnBmpr1a-NLLacZ)1Nobstg/tg mice with Tg(Mpz-cre)94Imegtg/+ mice (Yamauchi et al., 1999). Controls throughout the study were single transgenic littermates without Cre transgene.
Analysis of dnBmpr1a expression by LacZ staining (Fig. 1B,C) confirmed that the expression pattern of dnBmpr1a was identical to that of LacZ in Tg(CAG-cat-lacZ)11Miyatg/+;Tg(Mpz-cre)94Imegtg/+ mice (Fig. 1D). We then investigated the putative changes in the downstream effector pathways by using reverse transcriptase-polymerase chain reaction (RT-PCR) and western blot analysis of the developing nasal region at 10.5 days post coitum (d.p.c.) and 15.5 d.p.c., respectively. In the double-tg embryos, expression levels of Msx1 and Msx2 were significantly reduced to 32.7% (P=0.0026) and 50.6% (P=0.048) of control levels, respectively (Fig. 1E,F), and the expression of phosphorylated Smad1, Smad5 and Smad8 proteins significantly decreased to 64.5% (P=0.03) of control levels (Fig. 1G,H). However, levels of phosphorylated Smad2, phosphorylated Smad3 and Bmpr1a proteins in the nasal processes of the double-tg embryos were normal (Fig. 1G), indicating that expression of dnBmpr1a in the double-tg embryos had no effect on TGF-β pathways and endogenous Bmpr1a expression level.
Perinatal death and facial defects in the double-tg mice
At 3 weeks after birth, the ratio of double-tg was 17.9% (n=21/117). Double-tg did not show embryonic lethality but perinatal death (supplementary material Table S1). Among the newborns, 41.9% were double-tg (n=18/43). Double-tg newborns without cleft palate were indistinguishable from control newborns in appearance (Fig. 2A,B,D,E). Among the double-tg newborns, 83.3% (n=15/18) had midline fusion defects, cleft face (Fig. 2C,F) and cleft palate (Fig. 2G,H). These defects affected suckling but not breathing. The double-tg mice without cleft survived and had a characteristic short face (Fig. 2I–L). To quantify the facial morphology, we measured facial length (FL) and the distance between the eyes (DE) at 3 weeks. The FL and DE of double-tg were significantly shorter (0.87-fold; P<0.001; Fig. 2M) and greater (1.1-fold; P=0.024; Fig. 2N), respectively, than that of the controls.
Dependency of frontonasal mesenchymal cell survival on Bmpr1a-mediated signaling
At 9.5 d.p.c., which marks the end of cranial NC cell migration (Jiang et al., 2002), the double-tg embryos were histomorphologically indistinguishable from the controls (Fig. 3A,B). Furthermore, the expression pattern of Tfap2a, one of the marker genes of NC cells, of double-tg mice was indistinguishable from that of the controls (Fig. 3C,D). These results suggest that in our mutants NC cell migration was not affected by the reduction in Bmpr1a-mediated signaling.
At 10.5 d.p.c., the nasal processes of the double-tg mice were apparently smaller than those of the controls (Fig. 3E,F). Calculation of cell density of the frontonasal mesenchyme revealed that the cell density of the mutant embryos at 9.5 d.p.c. [(2.09±0.22)′104 cells/mm2; n=3] was the same as that of the controls [(1.86±0.19)′104 cells/mm2; n=3], but that the cell density of the mutant embryos at 10.5 d.p.c. was significantly lower than that of the controls (0.79-fold; P=0.002; Fig. 3G), suggesting that a decrease in proliferation and/or an increase in apoptosis occurred in the frontonasal mesenchymal cells between 9.5 and 10.5 d.p.c. Therefore, 10.5 d.p.c. nasal processes were assessed for apoptosis and proliferation. Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) and P53 double staining assay of frontonasal mesenchyme showed 0% of TUNEL single-positive cells, 2.46% of TUNEL/P53 double-positive cells and 0.80% of P53 single-positive cells in the control, and 0% of TUNEL single-positive cells, 18.24% of TUNEL/P53 double-positive cells and 4.21% of P53 single-positive cells in double-tg mice (Fig. 3H–J). No TUNEL single-positive cells were detected in either double-tg or control frontonasal mesenchyme and there was a 7.4-fold enhancement in the ratio of the double-positive cells in double-tg frontonasal mesenchyme. These results indicated that the p53 apoptotic pathway was activated by a reduction in Bmpr1a-mediated signaling in double-tg frontonasal mesenchyme. RT-PCR revealed that the levels of the anti-apoptotic Bcl-XL and the apoptosis-related c-Myc gene transcripts in the double-tg nasal processes were significantly reduced to 33.7% (P=0.018) and 49.2% (P=0.006) of the control levels, respectively (Fig. 3K,L). Proliferating cell nuclear antigen (PCNA) immunostaining assay showed no change in proliferation (Fig. 3M–O). At 11.5 d.p.c., when medial nasal processes normally fused in the midline (Fig. 3P), 77.8% of the double-tg embryos (n=7/9) had an abnormally unfused nasal process (Fig. 3Q). Furthermore, at both 13.5 d.p.c. and 15.5 d.p.c., after the completion of upper lip formation, 75% of double-tg embryos exhibited facial cleft (13.5 d.p.c., n=3/4; 15.5 d.p.c., n=6/8), which was consistent with the incidence of both unfused nasal processes at 11.5 d.p.c. and cleft face at birth. These observations strongly suggest that a decrement in BMP signaling induces an unusual P53-mediated apoptosis without affecting cell proliferation in the frontonasal mesenchyme around 10.5 d.p.c. and that the apoptosis eventually results in the facial cleft.
Dependency of frontal bone primordial mesenchymal cell survival on Bmpr1a-mediated signaling
Micro-computer tomographic (micro-CT) scanning at 3 weeks clearly showed calvarial ossification defects between the frontal bones of double-tg mice (n=5/5; Fig. 4A–D). No obvious defects were detected in the other parts of the double-tg skulls. Analysis of newborn skulls using Alizarin Red and Alcian Blue staining (skeletal staining) revealed a reduction in the size of the frontal bones of double-tg mice in the presence (n=3) and absence (n=1) of cleft face and cleft palate (Fig. 4E–J). The double-tg mice with cleft face and cleft palate exhibited more severe defects in craniofacial bone formation (e.g. nasal bones dysplasia and septal cartilage separation) (Fig. 4G,J). The other craniofacial bones and cartilage of the double-tg mice were indistinguishable from those of the controls. We hypothesized that the etiology of calvarial foramina is similar to that of facial dysmorphism in double-tg mice.
PCNA immunostaining and the TUNEL assay revealed an unchanged proliferation ratio and higher apoptotic ratio (2.7-fold; P=0.048), respectively, in the developing frontal bone primordial mesenchymal cells of double-tg embryos at 10.5 d.p.c. (Fig. 4K–N,Q,R). Furthermore, apoptotic p53-positive cells were detected only in the frontal bone primordia of double-tg embryos (Fig. 4O,P,S). At 12.5 d.p.c., we observed a marked reduction in the number of frontal bone primordial cells but no changes in apoptotic or proliferating cell ratios (Fig. 4T,U and data not shown). At 15.5 d.p.c., hematoxylin-eosin (HE) and skeletal staining revealed that the frontal bone primordial cells of the double-tg embryos formed the frontal bone matrix normally and differentiated into the frontal bones, but their sizes were smaller than those of the controls (Fig. 4V–Y). These observations suggest that a decrement in BMP signaling in the developing frontal bone primordial mesenchyme contributed to unusual apoptosis, which resulted in cell expansion failure around 10.5 d.p.c. This then caused a reduction in both the frontal bone primordial cell number at 12.5 d.p.c. and the frontal bone size after 15.5 d.p.c. Thus, in our model, etiology of the calvarial foramina appears to be similar to that of facial dysmorphism.
Heart septum defects and pigmentary abnormality
Because patients with juvenile polyposis syndrome and chromosome 10q23 deletion syndrome exhibit heart defects (Delnatte et al., 2006; Menko et al., 2008; Zhou et al., 2001), we assessed heart morphology in double-tg at birth by micro-CT and histomorphological analyses. Three of the six mutant mice with facial cleft (50%) exhibited a defect in the ventricular septum of the heart (supplementary material Fig. S1A–D and Movies 1, 2). Furthermore, 28.6% of the surviving double-tg mice without cleft face or cleft palate (n=6/21) exhibited a pigmentary abnormality, a belly spot (supplementary material Fig. S1E).
The facial abnormalities in our double-tg mice mimic the hypertelorism and flat nasal bridge observed in patients with juvenile polyposis syndrome and chromosome 10q23 deletion syndrome. Furthermore, these patients and our double-tg mice both suffer from similar heart septal defects (Delnatte et al., 2006; Jacoby et al., 1997; Menko et al., 2008; Zhou et al., 2001). Thus, facial abnormalities and heart septal defects in human patients might be partly caused by the knockdown of BMPR1A-mediated signaling in NC-derived cells.
Compared with our double-tg mice, two previous mutant mice lacking Bmpr1a in NC cells exhibited more severe heart septal defects (Nomura-Kitabayashi et al., 2009; Stottmann et al., 2004) that correlated with the level of Bmpr1a-mediated signaling. However, their facial development was normal. Bmp2, Bmp4 and Bmp7 ligands bind to Bmpr1a-Bmpr1b and Bmpr2 serine/threonine kinase receptor complexes and transmit signals via common downstream proteins Smad1, Smad5 and Smad8 (R-Smads). Two of any activated R-Smad proteins form a heterotrimer with a common partner, Smad4. The trimer translocates to the nucleus and regulates the transcription of target genes (Miyazawa et al., 2002). Panchision and co-workers showed that the expression levels of Bmpr1b in the developing heart mesenchyme of wild-type animals were much lower than those in the developing nasal mesenchyme (Panchision et al., 2001). In conditional knockout animals, the free BMPs generated by an ablation of Bmpr1a-Bmpr2 receptor complexes can bind the Bmpr1b-Bmpr2 complexes instead and transmit compensatory signals to R-Smads-Smad4 in facial development but not in heart development. In our double-tg, no free BMPs were generated because BMPs could form stable but inactive ternary complexes with dnBmpr1a-Bmpr2 receptor complexes, resulting in a failure to transmit sufficient signals for both face and heart development.
The c-Myc gene, a member of the helix-loop-helix family of transcription factors, is one of apoptotic regulators (Pelengaris and Khan, 2003). Both downregulation and upregulation of c-Myc expression are associated with an activation of apoptosis (Amendola et al., 2009; Askew et al., 1993). In our double-tg embryos, an activation of apoptosis of the frontonasal mesenchymal cells is associated with c-Myc downregulation. Our results showed that an optimal Bmpr1a-mediated signal was involved in maintenance of c-Myc expression levels in nasal process developing.
The NC cell-derived mesenchyme is the source for populating the frontal bone primordium in mice (Jiang et al., 2002). Bmp2, Bmp4, Bmp7 and their receptors Bmpr1a and Bmpr1b are expressed in primordial tissues (Dewulf et al., 1995; Kim et al., 1998; Zouvelou et al., 2009). Mutations in Msx1 and/or Msx2 transcription factors, which are downstream in the BMP signaling pathway, result in frontal ossification defects (Han et al., 2007). These results imply that downregulation of BMP signaling via Bmpr1a in NC cells can lead to frontal bone malformation. However, early embryonic lethality of the previous Bmpr1a mutant mice prevented the testing of this hypothesis (Mishina et al., 1995; Nomura-Kitabayashi et al., 2009; Stottmann et al., 2004). Because our double-tg mice overcome embryonic lethality and displayed ossification defects in frontal bones, we clearly demonstrated, for the first time, a significant role of Bmpr1a-mediated signaling in normal frontal bone development.
Thus, we have developed a new useful model system for addressing the pathological mechanism of diseases caused by the downregulation of Bmpr1a-mediated signaling.
We inserted the 0.6 kb dominant-negative mouse Bmpr1a cDNA (ΔmTFR11) (Suzuki et al., 1994), 0.5 kb internal ribosomal entry site and 3.5 kb LacZ gene containing the nuclear localization signal into the PmeI site of pCAG-XstopX-polyA (Saito et al., 2005). Founders were made by pronuclear injection into zygotes obtained by crossing Pcp2tm1(cre)Nobs/+ males (129X1/SVJ*C57BL/6J) (Saito et al., 2005) and C57BL/6J females and selected by monitoring the LacZ expression in retina cells (data not shown). Mice heterozygous for the transgene were backcrossed to C57BL/6J for at least ten generations. All the animals were cared for in accordance with the ethical guidelines established by the Institutional Animal Care and Use Committee at Mie University.
Mutant allele genotyping
The genotypes were determined by PCR analysis. The primers for LacZ were as follows: LacZ forward, 5′-CTACACCAACCTGACCTATC-3′ and LacZ reverse, 5′-CGCTCATCGATAATTTCACC-3′. The primers for Cre were: Cre forward, 5′-GTCGATGCAACGAGTGATGA-3′ and Cre reverse, 5′-CAGCGTTTTCGTTCTGCCAA-3′. The PCR conditions were 94°C for 1 minute; 30 cycles of 94°C for 1 minute, 60°C for 1 minute, and 72°C for 1 minute; and 72°C for 5 minutes.
Data were analysed using Student’s t-test to compare two groups. The results are presented as mean ± s.e.m.
Western blot analysis
Total protein was extracted as described previously (Tsumura et al., 2006). Rabbit antibody against phosphorylated Smad1/5/8 (Cell Signaling Technology, Beverly, MA), goat antibody against phosphorylated Smad2/3 (Santa Cruz Biotechnology), goat anti-Bmpr1a antibody (Santa Cruz Biotechnology) and rabbit anti-actin antibody (Santa Cruz Biotechnology) were used at 1:1000, 1:200, 1:200 and 1:1000 dilution, respectively. Secondary goat anti-rabbit IgG (Santa Cruz Biotechnology) and donkey anti-goat IgG (Santa Cruz Biotechnology) antibodies were both used at a dilution of 1:2000. Blotted proteins were visualized as described previously (Tsumura et al., 2006). Expression levels of proteins were normalized to that of β-actin.
Mutations in the gene encoding bone morphogenetic protein receptor type 1A (BMPR1A) are associated with facial dysmorphism and heart defects, the main clinical features of juvenile polyposis syndrome and chromosome 10q23 deletion syndrome. BMPR1A is broadly expressed and its ligands (BMP2, BMP4 and BMP7) are expressed in specific nasal regions during development. Mouse models have been used as a means to better understand the role of BMPR1A in craniofacial development. Of three previously generated conditional BMPR1A knockout mouse lines, two died in the late embryonic stage and the other did not display recombination in the mesenchymal cells of the nasal processes. In this paper, the authors developed a new mouse model that overcomes embryonic lethality, and demonstrate that features in individuals with juvenile polyposis or chromosome 10q23 deletion syndromes might be caused in part by the knockdown of BMPR1A-mediated signaling in neural crest (NC)-derived cells.
The authors generated a BMPR1A knockdown mouse by conditionally expressing a dominant-negative (dn) form of BMPR1A in NC-derived cells. 100% of the conditional mutant mice had wide-open anterior fontanelles, and 80% died of cleft face and cleft palate soon after birth. The 20% that survived developed short faces, hypertelorism and calvarial foramina. 50% of mice with facial cleft at birth also showed abnormal heart morphology. Analysis of the NC-derived craniofacial mesenchyme of mutant embryos revealed a normal growth rate but an incomplete expansion of mesenchymal cells. At the molecular level, it was found that knockdown of BMPR1A-mediated signaling triggered accumulation of p53, downregulation of Bcl-XL and c-Myc, and extensive apoptosis.
Implications and future directions
These findings provide genetic evidence that optimal BMRP1A-mediated signaling is essential for NC-derived mesenchymal cell survival in normal nasal and frontal bone development. This work indicates that inhibition of BMPR1A-mediated signaling in NC-derived cells leads to facial dysmorphism and heart defects in mice that are similar to those seen in humans with juvenile polyposis syndrome and chromosome 10q23 deletion syndrome. By crossing the floxed-dnBMRP1A transgenic mice developed here with other Cre strains, it will be possible to probe the role of BMRP1A in other cell types. For example, crossing them to a colon-specific Cre mouse or a brain-specific Cre mouse will enable studies of colorectal polyposis and mental retardations, respectively, which are additional features of juvenile polyposis syndrome and chromosome 10q23 deletion syndrome.
Total RNA was extracted using Trizol reagent (Invitrogen, Carlsbad, CA). cDNA was synthesized from 1 μg of total RNA by using the reverse transcription system (Promega, Madison, WI). The primer sets were as follows: Msx1, 5′-ATGACTTCTTTGCCACTCGG-3′ and 5′-GTCAGGTGGTACATGCTGTA-3′; Msx2, 5′-ATGGCTTCTCCGACTAAAGG-3′ and 5′-GTCCATCTGGTCTTCCTTAG-3′; dnBmpr1a, 5′-CTGCTAACCATGTTCATGCC-3′ and 5′-GCTGCAAATACTGGTTGCAC-3′; Bax, 5′-ACAGATCATGAAGACAGGGG-3′ and 5′-CAAAGTAGAAGAGGGCAACC-3′; Bcl-XL, 5′-TGGTCGACTTTCTCTCCTAC-3′ and 5′-TGAGTGGACGGTCAGTGTCT-3′; c-Myc, 5′-CGCGCCCAGTGAGGATATC-3′ and 5′-CCACATAC-AGTCCTGGATGAT-3′; and Hprt, 5′-CGTGATTAGCGATGATGAAC-3′ and 5′-GGCTTTGTATTTGGCTTTTC-3′.
Each experiment was performed using cDNA from the frontonasal processes of the control and double-tg mice at 10.5 d.p.c. (n=4 each). The PCR conditions were 94°C for 1 minute; 35 cycles of 94°C for 1 minute, 60°C for 1 minute, and 72°C for 1 minute; and 72°C for 5 minutes. Expression levels were normalized to that of Hprt.
Histological and TUNEL analysis
Histological analysis was performed as previously described (Hogan et al., 1994). For cell density analysis, we counted cells in 30,000μm2 area of the nasal processes in a HE-stained section of each embryo (three samples for each genotype). Primers for the 1.3-kb Tfap2a cDNA were: AP2 forward, 5′-ATGCTTTGGAAACTGACGGA-3′ and AP2 reverse, 5′-TTTGTCACTGCTTTTGGCGC-3′. The TUNEL assay was performed using an in situ apoptosis detection kit (Takara, Shiga, Japan). Peroxidase-conjugated anti-PCNA antibody (Dako Cytomation, Glostrup, Denmark) and rabbit anti-p53 antibody (Santa Cruz Biotechnology) were used at 1:2 and 1:1000 dilutions, respectively. Secondary goat anti-rabbit IgG antibody (Santa Cruz Biotechnology) was used at 1:2000 dilution. Signals of TUNEL, P53 and PCNA were visualized with HistoMark Black (KPL, Gaithersburg, MD), HistoMark Orange (KPL) and di-amino-benzidine (Dojindo, Kumamoto, Japan), respectively. For the statistical analysis, we counted TUNEL-, PCNA- and p53-positive cells among Kernechtrot-positive cells in the nasal processes in a section of each embryo (three or four samples for each genotype).
Three-dimensional images of the skulls and hearts were obtained by CT analysis using the R_mCT system (Rigaku, Tokyo, Japan). For heart analysis, mice were injected with 0.2 ml Omnipaque 350 (Daiichi-sankyo, Tokyo, Japan) via the intraorbital vein under anesthesia, and then sacrificed. The lamp voltage and current were 90 kV and 100 mA, respectively. The scanning times were 17 seconds and 2 minutes for skulls and hearts, respectively. Volume rendering and measurements were performed using the software i-View (Morita, Kyoto, Japan).
We thank Naoto Ueno (National Institute for Basic Biology) for the dominant-negative mouse Bmpr1a cDNA (ΔmTFR11). We also thank Drs Hidetoshi Yamazaki (Mie University), Kyoko Imanaka-Yoshida (Mie University), and Yuji Nakajima (Osaka City University) for helpful discussions.
The authors declare that they do not have any competing or financial interests.
H.S.: data collection and analysis, and manuscript preparation. K.Y.: Tg(Mpz-cre)94Imegtg/+ mouse preparation. N.S.: data analysis, manuscript preparation and study design. All authors edited the manuscript prior to submission.
This work was supported in part by Grant-in-Aid for Scientific Research, and by the National Cancer Center Research and Development Fund (23-A-2).
Supplementary material for this article is available at http://dmm.biologists.org/lookup/suppl/doi:10.1242/dmm.009274/-/DC1
- Received November 29, 2011.
- Accepted May 23, 2012.
- © 2012. Published by The Company of Biologists Ltd
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial Share Alike License (http://creativecommons.org/licenses/by-nc-sa/3.0), which permits unrestricted non-commercial use, distribution and reproduction in any medium provided that the original work is properly cited and all further distributions of the work or adaptation are subject to the same Creative Commons License terms