Abrogation of TGF-beta signalling in TAGLN expressing cells recapitulates Pentalogy of Cantrell in the mouse

Pentalogy of Cantrell (PC) is a rare multi-organ congenital anomaly that impedes ventral body wall closure and results in diaphragmatic hernia, intra- and pericardial defects. The underlying cellular and molecular changes that lead to these severe developmental defects have remained unknown largely due to the lack of representative animal models. Here we provide in depth characterization of a mouse model with conditional ablation of TGFβRII in Transgelin (Tagln) expressing cells. We show that Tagln is transiently expressed in a variety of cells that participate in the embryonic development and patterning of ventral structures. Genetic ablation of TGFβRII in these cells leads to ventral midline closure defect, diaphragmatic hernia, dilated cardiac outflow tract and aberrant cardiac septation, providing a reliable model to study the morphological changes leading to PC. We show that myogenisis in the diaphragm is independent of TGFβ and the diaphragmatic hernia arises from fibroblast-specific migration defect. In the dorsal body wall Tagln expression is initiated after the closure process, revealing a remarkable difference between ventral and dorsal body walls development. Our study demonstrates the use of micro-CT scanning to obtain a 3-dimensional high-resolution overview of embryonic anomalies and provides the first mechanistic insight into the development of PC.

In the developing embryo transforming growth factor β (TGFβ) signalling plays a pivotal role in facilitating closure of the embryonic midline 11 . It is also essential for cell homeostasis in general and is particularly crucial for cardiac and vascular development 12,13 . Furthermore, defects in TGFΒ signalling pathways are known to associate with cardiac and midline closure defects [14][15][16][17][18][19] . However, they can also result in various congenital anomalies including dorsal midline closure defects, cleft palate, lung hypoplasia, craniofacial and limb malformations and urogenital defects 11,16,[20][21][22] . Although all TGFβ morphogens signal via common receptors (TGFβR1/2/3 complex) their expression varies between various cells and tissues, explaining the differences in phenotypes when knocked out in the mouse.
Transgelin (TAGLN, also known as SM22a) is an actin-binding cytoskeletal protein that is expressed in vascular smooth muscle cells 23 . At embryonic stages TAGLN is not a specific VSMC marker and is widely expressed by non-vascular mesenchymal tissues 24 . Moreover, Tagln was recently found to label a migratory myofibroblasts cell population that respond to TGFβ signalling 15 . TGFβ is also known to induce Transgelin (Tagln) in vitro and in vivo [25][26][27] through SMAD binding to the Tagln promoter 28 . Furthermore, TGFβ signalling inactivation in Tagln expressing cells leads to ventral body wall closure defect, cardiac and great vessels anomalies 15,17,29 .
We have recently demonstrated that selective removal of Tgfbr2 from TAGLN expressing cells results in exomphalos and ectopia cordis 15 . In addition, previous reports of the Tagln-Cre:Tgfbr2 flx/flx have demonstrated cardiac and major vessels defect that overlap with the spectrum of cardiac anomalies seen in PC 17,29 . We are now reporting an anterior diaphragmatic hernia in this model making the congenital defects in the Tagln-Cre:Tgfbr2 flx/flx mouse model of highly representation of the PC anomaly seen in humans.

Results
Tagln-Cre:Tgfbr2 knockout exhibits multiple congenital anomalies. The Tagln-Cre:Tgfbr2 flx/flx model exhibits three main categories of congenital anomalies. The first striking anomaly is the complete failure of ventral body wall closure as we have recently shown 15 . We used here micro-CT scanning to characterise the extent of the defect and to delineate its 3D topography. We found that the knockout embryos develop a large exomphalos and ectopia cordis. A thin sac covers the heart and embryonic intestine (Figs 1a,b and S1b,d,e and Movie 1) and the lateral body wall fails to develop beyond the ventral one-half (arrowheads in Fig. 1b and S1e). The wild type littermates show only a small physiological umbilical hernia and the ribs and intercostal muscles reach the midline (Figs 1c,d and S1a,c and Movie. 2). The mutant embryos also show gross cardiac and outflow tract anomalies. Large ventricular septal defect (VSD) is present (Fig. 1e) compared to complete ventricular septal closure in the WT littermates (Fig. 1g). Moreover, gross dilatation of the heart and outflow (OF) tract is evident (Figs 1f and S1f) when compared to the WT littermates (Fig. 1h). In addition, the central veins showed gross dilatation as well (Fig. S1g,h). Lastly, this model displays an anterior diaphragmatic hernia with the liver herniating into the chest (Fig. 1f), while in the WT littermates the diaphragm extend fully anteriorly and attaches to the sternum (arrow in Fig. 1h).
Of note, the mutant mouse did not express any dorsal closure defect (Fig. 1i,k) and the closure of the palate was comparable between the knockout and the WT littermates (Figs 1j,l and S1i-l).
The expression of Tagln in the developing embryonic organs. The 3D analysis of the knockout model enables to distinguish embryonic morphogenetic processes that depend on TGFβ signalling in Tagln expressing cells. Tagln is primarily recognized as a marker for smooth muscle cells and fibroblasts, yet a complete overview of its expression at the time of dorsal and ventral body wall closure was missing. We have recently demonstrated the Tagln-Cre transgene specificity and shown complete overlap between the transgene and the native TAGLN protein up until E13.5 15 . We hence used Tagln-Cre:Rosa26tdTom mouse model to study TAGLN expression and performed serial sectioning and IHC staining at different embryonic stages. At E11.5 Tagln expression is seen in the heart, outflow tract and the myotome (Fig. 2a). It also labels the aorta and the developing vasculature in diverse organs and the intestine inside and outside of the umbilical hernia ( Fig. 2b-d). Remarkably, whereas Tagln is highly expressed in the ventral body wall, no such expression is evident in the dorsal body wall and it is absent also from the neural tube ( Fig. 2a-d). At E13.5 tdTomato labelled cells are seen in many organs. In the thorax the tdTomato protein labels the heart, lungs and the major vessels (Fig. 2e). The diaphragm crura and the septum transversum are also tdTomato+ (Fig. 2e,f). Similarly, the perivascular cells in the liver and intestinal smooth muscle cells are labelled by the transgene at this stage (Fig. 2f,g). Moreover, the intercostal and abdominal muscles also express tdTomato. At E13.5 tdTomato is strongly expressed in the dorsal body wall muscles and in the interneurons of the neural tube ( Fig. 2e-g). By E15.5 the expression pattern of Tagln is already established, the tdTomato protein expression is indistinguishable from that at E13.5 and is mainly evident in the heart, diaphragm, body wall muscles and smooth muscles of the lungs and intestine. In addition, tdTomato expression is more widely spread in the neural tube ( Fig. 2h-i). These data collectively demonstrate that Tagln expression marks the vasculature, skeletal muscles and ventral body wall at the time of its closure, while it is initiated in dorsal structures only after the dorsal closure period.

Diaphragmatic myogenic and myofibroblast cells express Tagln.
Considering the developmental defects in the diaphragm we next focused more specifically on Tagln expression dynamics in the developing diaphragm. We used Tagln-Cre crossed to the Rosa26-NGZ reporter strain and performed whole-mount β-galactosidase staining. The expression pattern of Tagln from early time points of diaphragm development was remarkable in the pleuro-peritoneal folds (PPF) and the septum transversum (ST) (Fig. 3a). Later on during diaphragm development the muscular components, also nLacz+, can be seen displacing the septum transversum into an anterior and central position (Fig. 3b). The diaphragm continues to develop in a dorso-ventral and lateral to central fashion and the septum transversum is limited to a narrow triangular area at E14.5 (dotted area in Fig. 3c). In the developed diaphragm the Tagln-Cre derived nLacz signal was evident in the muscle part of the diaphragm, and to a lesser degree in the central tendon (Fig. 3d). We next studied the cells that initiate Tagln expression in the developing diaphragm and constructed the cellular dynamics of the congenital defect. We analysed the developing diaphragm cells in serial sectioning and IHC staining of Tagln-Cre:Rosa26tdTom mouse model. At early time points of diaphragm development (E11.5) the pleuro-peritoneal folds express TAGLN, but do not express the myogenic markers paired box protien7 (PAX7)  or MyoD (Fig. 3e). These early pleuro-peritoneal folds' cells are likely fibroblasts as they co-express transcription factor 4 (Tcf4) (Fig. 3f) and TAGLN. In the PPF of E12.5 embryo tdTomato+ cells are seen in a more lateral position to the myogenic cells of the diaphragm labelled by Pax7 (Fig. 3g). These pioneering (Tagln+) cells at the leading edge of the developing diaphragm at E12.5 co-express fibroblast marker Tcf4, but do not express myogenic markers (Fig. 3g). This becomes more evident at E13.5 where TAGLN expression becomes increasingly restricted to the cell population at the most rostral part of the advancing diaphragm (Fig. 3h). Furthermore, tdTomato+ cells at the advancing edge of the developing diaphragm at E13.5 markedly express TGFβ receptor-2 (TGFβR2) (Fig. 3i). Starting from E13.5 some of the more dorsally-located cells of the developed part of the diaphragm start to down regulate TAGLN (Fig. 3j) while maintaining lineage tracing marker tdTomato. The TAGLN+ cells at the leading edge of the diaphragm do not appear to undergo apoptosis at time points of anterior hemi diaphragms closure and fusion (E13.5 till E15.5). We did not detect nuclear accumulation of activated Caspase 3 protein in the anterior diaphragmatic TAGLN + cells at these time points (Fig. S2a-c). The TAGLN + cells are maintained in the adult diaphragm of Tagln-Cre:Rosa26tdTomato as myogenic cells, pleural cells and fibroblasts of the central tendon (Figs 3k and S2d), while TAGLN expression in the adult diaphragm is only present in perivascular cells (Fig. 3l). This data demonstrates that TAGLN expressing fibroblasts arise early in PPF development and migrate ahead of the diaphragm muscle cells during diaphragm morphogenesis. It also suggests that Tagln is essential during diaphragm embryogenesis, but probably not for adult diaphragm maintenance.
Tagln-Cre:Tgfbr2 knockout shows anterior diaphragmatic hernia. We have demonstrated that TAGLN is widely expressed in the embryonic diaphragm and TGFβR2 is abundant in tdTomato+ cells at the anterior part of the developing diaphragm. In line with these observations, we found an extensive anterior diaphragmatic hernia in Tagln-Cre:Tgfbr2 mutants. The anterior part of the diaphragm fails to develop completely and the liver herniates to the thoracic cavity through the anterior part of the diaphragm (Fig. 4a). We have observed the earliest evidence of this anterior diaphragmatic hernia at E13.5 when the liver can be seen herniating to the thoracic cavity displacing the heart and lungs (Fig. 4b). In the WT littermate, the abdominal and thoracic cavities are completely separate at E14.5. The anterior diaphragm reaches the sternum anteriorly and is made of differentiated skeletal muscle cells (Fig. 4c,d). In the mutant, by E14.5 the lateral and posterior elements of the diaphragm have developed normally and multi-layered differentiated muscle cells labelled by myosin heavy chain are seen extending between the posterio-lateral body wall and the central tendon medially (Fig. 4e,e' and S3a [arrows]). On the other hand, the only septum between the thoracic and abdominal cavities anteriorly is a thin, Laminin positive, sac (Figs 4f,f ' and S3a [arrowheads]). Here, the most rostral portion of the developing diaphragm is made of condensated fibroblasts (Tcf4+) and lack the presence of myogenic progenitor cells (Fig. 4g).
In contrast to what is seen in the WT at this stage where the anterior diaphragm is made of muscle cells expressing sarcomeric myosin (Fig. 4d).
The effects of liver herniation into the thorax are readily observed in the developing lungs. By using organ segmentation and volume rendering of micro-CT scans, it becomes evident that in the mutant the herniating liver pressures the lower lung lobes (Fig. 4h) compared to the WT littermate (Fig. 4j). The anterior parts of the lower lobes of the lungs reveal indentation from the herniating liver (Figs 4i and S3b) and are less developed than the WT littermates (Figs 4k and S3c). In addition, the upper lobes appear to be hypoplastic (Figs 4i and S3b) when compared to the WT (Figs 4k and S3c). This is probably due to direct pressure from the dilated right and left superior vena cava (Figs 4a and S3d). By using organ segmentation methods, the dilated SVCs are seen encirculating and compressing the upper lung lobes (Fig. S3e-g). The lung architecture in the mutant is otherwise maintained. Lung branching and lung lobar structure in the mutant and WT littermate are similar, the three right lung lobes (superior, middle and inferior), the left lobe and posterior caval lobes are all developed (Figs 4I,k and S3b,c) and the bronchial spaces are patent (Figs 4b,e and S3a). Lastly, the dilated outflow tract displaces the trachea to the left, however the tracheal lumen remains patent as in the WT littermates (Fig. S3h,i).

TGF-β signalling in myogenic progenitors is not essential for diaphragm development.
The accumulating evidence suggests that fibroblast migration drives the morphogenesis of the diaphragm 30,31 . Our lineage tracing data using Tagln-Cre:Rosa26tdTomato model indicated that in addition to fibroblasts myogenic cells in the diaphragm undergo a phase of Tagln expression (Fig. 3a-c,h). This leaves open the possibility that in addition to fibroblasts myogenic cells may be affected in the Tagln-Cre:Tgfbr2 flx/flx mutant. We next analysed whether Tgfbr2 elimination from myogenic cells may underlie the diaphragmatic development defect that we observed in Tagln-Cre:Tgfbr2 flx/flx model. We found that Tagln is expressed in MYOD + embryonic myotubes in the PPF from E12.5 (Fig. 5a). To characterize specifically the importance of TGFβ signalling in myogenic cells we crossed theTgfbr2 flx/flx strain to the MyoD-Cre mouse line 32 and analysed MyoD-Cre:Tgfbr2 flx/flx embryos and postnatal mice. Embryonic diaphragm development in MyoD-Cre:Tgfbr2 flx/flx was normal, fully muscularised diaphragmatic crura, dome and lateral diaphragm is present at E14.5 (Fig. 5b-d) similar to the WT (Figs 3c and 4d). These mice are born in good condition, do not show any signs of respiratory distress at birth and have intact and fully developed diaphragm (Fig. 5e,f). Thereby, we can conclude that TGFβ signalling during diaphragm development acts preferentially on fibroblasts and is not necessary for the patterning or development of diaphragm muscle cells.  scanning. These congenital anomalies are most evident at E14.5 (Fig. 6a), both the heart and the outflow tract show gross dilatation in the mutant compared to the WT littermates (Fig. 6a' ,b') and the aneurysmal outflow tract occupies the majority of the superior thoracic and inferior neck spaces (Figs 6a,c,d and S4a) compared to the WT littermate (Fig. 6b,f). In the mutant there is a single outflow tract originating from the right ventricle and overriding the large ventricular septal defect (VSD) (Figs 6e and S4b), while the WT littermates at this stage show a left sided aortic arch and a separate pulmonary trunk (Fig. 6f,g). Nevertheless, in the mutant the atria and ventricles contract independently and fill the dilated outflow tract (Movie. 3). This dilated outflow tract is thick walled and displaces the right main bronchus to the left (Fig. 6h). All systemic and pulmonary branches in the mutant originate from the single outflow tract. The left and right subclavian arteries originate first, either individually directly from the outflow tract (Fig. S4b,c) or through a common arterial stem (Fig. S4d). While, the left and right common carotid arteries originate from the dome of the aneurysmal outflow tract (Fig. S4b,c,e). The pulmonary arteries originate from the posterior wall of the outflow tract origin individually and not through a common pulmonary trunk (Fig. S4c,g,h). The pulmonary veins join before draining into the left atrium (Fig. S4i,j). Of note, the aneurysmal anomaly of the arterial system is limited to the outflow tract and none of the systemic or pulmonary arteries show dilatation (Fig. S4b-h).
Structural cardiac anomalies are also found in the Tagln-Cre:Tgfbr2 knockout. A large VSD is evident and the ventricular muscle wall (expressing MyHC and smooth muscle α actin (αSMA)) is thinner and less compact compared to the WT ventricular wall (Fig. 6i-k). These results confirms previous reports and indicates a direct

Discussion
We provide here a comprehensive description of a mouse model for Pentalogy of Cantrell. We have used whole embryo micro-CT scanning in combination with immunohistochemistry to generate a holistic overview of the pathological changes that occur when TGFβ signalling is eliminated from Tagln expressing cells. This analysis provides the first mechanistic insight into the development of PC, an otherwise poorly characterised anomaly.
The phenotype expressed by the Tagln-Cre:Tgfbr2 knockout resembles the phenotype seen in Pentalogy of Cantrell anomaly in humans 5 . The pentad of anomalies expressed in PC are readily observed in the mutant mouse; supra-umbilical exomphalos, ectopia cordis, anterior diaphragmatic hernia, failed formation of the diaphragmatic pericardium and intra-cardiac defects are all present. The single outflow tract overriding a VSD is normally referred to in humans as truncus arteriosus 33 . This anomaly is generally lethal and the outcome is poor 34 . This may explain the lethality seen in the Tagln-Cre:Tgfbr2 flx/flx model. In addition, our knockout model shows gross aneurysmal dilatation of the outflow tract that has not been described in newborn infants with PC. It is likely that the lethality of this anomaly induces abortion and stillbirth and hence does not present in viable neonates. In addition, the complete loss of function nature in mouse knockout model may be the reason behind the severe anomaly spectrum in comparison to some milder anomalies in some cases of PC. Nevertheless, the resemblance between Tagln-Cre:Tgfbr2 flx/flx and PC phenotype is quite striking.
The diverse pathologies present in PC are likely caused by morphogenetic defects mainly in the somatic mesoderm. Tagln is expressed in a variety of developing tissues during embryogenesis 24,35 . Beside the developing heart and outflow tract Tagln is seen in the myotome from as early as E9.5 24 . Moreover, Tagln labels migratory myofibroblasts in the developing abdominal wall 15 . We have demonstrated here that Tagln is abundantly expressed in the embryonic diaphragm and that TAGLN + diaphragm cells express TGFβR2. The role of TGFβ signalling in cardiac development and midline closure is widely established [14][15][16]36 . We have recently demonstrated a TGFβ gradient initiating from the epithelium of the primary ventral body wall (VBW) 15 . This gradient regulates the patterning of the skeletal and muscular components of the closing VBW. The peak of this TGFβ gradient is at E13.5 and the anterior diaphragmatic hernia seen in the Tagln-Cre:Tgfbr2 knockout is probably a representation of a halt in diaphragmatic development at this stage. The importance of this narrow developmental window for the full closure of the diaphragm is further supported by the presence of normal posterior and lateral diaphragmatic elements in the mutant embryos. Thus, diaphragm formation is largely independent of TGFβ, whereas its morphogenetic movement depends on it. We show that both TAGLN and TGFβR2 are highly enriched in fibroblasts that accumulate at the leading edge of the closing diaphragm, making them the primary candidate to respond to the TGFβ gradient. We propose that the anterior diaphragmatic hernia seen in the Tagln-Cre:Tgfbr2 flx/flx is probably due to dysfunction in the migration of the TGFβR2+TAGLN+cells at the leading edge of the developing diaphragm similar to what is seen in the developing body wall 15 . Importantly, we show that elimination of Tgfbr2 from myogenic cells does not lead to any diaphragmatic defects. This data highlights the importance of TGFβ as a morphogen for controlling fibroblast-dependent tissue organization in the embryo and suggests that this may represent the key mechanism in the development of PC.
An epithelial-mesenchymal signalling is essential for the patterning and closure of the ventral body structures. Mutations in several genes involved in epithelial and mesenchymal growth result in ventral body wall closure defects. The elimination of Wnt signalling from mesenchymal cells (Wntless knockout in Dermo1Cre mouse model; Wls f/f ;Dermo1 Cre/+ ) leads to ectopia cordis and failure of the thoracic rib cage to close 37 . In addition, elimination of Wnt signalling in ectodermal cells (in murine msh homeobox 2 Msx2-cre; Wls c/c knockout mouse model) disrupt the Wnt-Pitx 2 (Paired-like homeodomain transcription factor 2) axis, impairs ventral musculature formation and leads to ventral body wall closure defects 38 . Similarly, disruption of epithelial transcription factor AP-2 alpha (AP2α), aortic caroboxy-peptidase like protein (ACLP) and other components of the Wnt pathway lead to different ventral body wall closure phenotypes. For example, Wntlss and β-catenin knockout manifesting in Prune belly syndrome and Wnt/β-catenin, Gsk-3b, Lrp5 and Lrp6 knockout causing ectopia cordis [39][40][41][42] . We have demonstrated that the epithelium of the primary ventral body wall regulates a temporal TGFβ gradient that recruits and directs the migration of TAGLN+ fibroblasts at the leading edges of the ventral body wall and anterior diaphragm 15 . The morphological changes that shape and close the ventral body wall are dynamically controlled by multiple signalling pathways and tight cross-talk between the developing epithelial and mesenchymal components.
Micro-CT scanning has allowed for the first time complete characterisation of a multi-organ mutant mouse model. It has facilitated displaying the full picture and enabled visualization of smaller defects and anatomical variations that are otherwise difficult to establish by simple histological analysis of sectioned specimens. The micro-CT scanning method we have used allowed us to visualise sections at 6 µm intervals without the risk of losing any section that is commonly encountered in standard histological analyses. In addition, the use of contrast staining methods allows staining soft tissues differentially and hence acquiring images with high inter-tissue resolution. Furthermore, the ability to generate different angle views and off-axis sections provides 3-dimensional overview of the development of the pathology.
We have demonstrated in this study that TGFβ signalling in Tagln expressing cells displays a significant role in the development of mesodermal tissues and the loss of this role manifests in embryonic defects that are highly similar to the Pentalogy of Cantrell.

Materials and Methods
Animals. Mice  5 minutes) and then blocked in an incubation buffer (10% normal donkey serum, 1% bovine serum albumin (BSA) and 0.2% tween 20 in PBS) for 4 hours. Primary antibody was added and incubated overnight at 4 °C. On day 2 the slides were washed with 0.2% PBS-Tween 20 and blocked with a second incubation buffer (1% BSA and 0.2% tween 20 in PBS) for 1 hour at room temperature. Secondary antibodies specific to the primary antibody host species was added and incubated for 1 hour at room temperature. Slides were washed and mounted in Vectashield mounting media with DAPI (Vectalabs). Slides were imaged using Zeiss Axio Imager M2. Zeiss Zen software was used for image analysis.
Whole mount IHC. Whole mount IHC were described elsewhere 30 . Briefly, embryos were fixed in 4% PFA for 24 hours, washed with PBS, bleached for 24 h with Dent's bleach, rinsed with 100% methanol and then fixed in Dent's fix. Specimens were placed in Dent's fixed for at least one week prior to incubation with primary antibody. Specimens were then washed with PBS, blocked in 5% donkey serum in PBS and incubated with primary antibody(s) for 72 hours. Specimens were washed with PBS thoroughly and incubated with secondary antibody(s) for 72 hours. Specimens were then rinsed in methanol, methanol-BABB and cleared in 100% BABB solution. Specimens were then placed in a silicone container, covered in BABB and imaged using two photon laser confocal microscope (Leica SP8 Upright). Three-dimensional data sets were analysed with Leica confocal and Bitplane/ Imaris software. Solutions: Dent's fixative: 1 part DMSO: 4 parts Methanol Dent's Bleach: 1 part H2O2: 2 parts Dent's fix BABB: 1 part benzyl alcohol: 2 parts benzyl benzoate Blocking Serum: 5% donkey serum (also from Jackson Labs), 75% 1xPBS, 20% DMSO

Antibodies.
A full list of primary antibodies used in this work can be found in supplementary table (1).
Diverse secondary antibodies were used in this study (lifetech).
Whole mount β -gal staining in diaphragm explants. Diaphragm explants were generated from fixed embryos (as above). The Torsos was dissected to include the lower thorax with the upper abdomen regions. The heart and lungs were removed carefully not to damage the diaphragm beneath and the liver was left en-bloc with the diaphragm explant. Diaphragm explants were permeabilised in (1% Triton X and 0.4% NP40 in PBS) solution for 4 hours and incubated overnight for β-galactosidase activity at 37 °C as described 46 . Embryos were imaged using Zeiss Axio Zoom microscope and Zeiss Zen software was used for image analysis.
Micro-CT scans. Tissue treatment and staining. Embryos of desired gestation were collected fresh and fixed for 24 hours in 4% PFA solution. They were rinsed thoroughly in PBS and placed in PBS 10% sucrose for 4-6 hours to avoid shrinkage when adding staining (contrast) solution. The staining method was adapted from 47,48 . The staining solution is made of 1 part Lugol's solution and 3 parts iodine free water with 10% sucrose. Samples were incubated in staining solution for 48-72 hours at 4 °c on a rocking surface and protected from light. Solutions: Lugol's solution (Make fresh): 10 g KI (Potassium iodine) in 100 mls H2O when dissociates add 5 g I2 (elemental iodine) [protect from direct light] Staining solution (25% Lugol's in 7% sucrose): 1 part 100% Lugol's to 3 parts iodine free water 10% sucrose Micro-CT acquisition. Micro-CT data acquisitions were performed using a Nikon XTH 225 kV instrument with a tungsten target. Acquisition parameters were chosen to optimise absorption contrast and were 80 kV and 180 µA, without using a filter in the beam path. Around 5000 projections with an exposure time of 500 ms were taken per scan. Each specimen was scanned individually to achieve a voxel size of 6 µm x 6 µm x 6 µm. The resulting data were reconstructed using Nikon 3D Pro reconstruction software, before exporting for segmentation and visualisation in FEI VSG Avizo software. All CT scan images comparing mutant and WT embryos were performed on littermate embryos (from the same conception) after confirmation with genotyping.
Ethical approval. All animal work in this study was conducted according to the Home Office regulations and was approved by the Home Office under license number 70/7435 according to the UK Animals (Scientific Procedures) Act (1986). The University of Manchester Research Ethics Review board has approved this study.