TGFβ−induced embryonic cell senescence at the origin of the Cornelia de Lange syndrome

Cornelia de Lange Syndrome (CdLS) largely caused by mutation of the cohesin loader NIPBL is a rare developmental disorder affecting the formation of many organs. Besides a short body size and neurological defects, more than half of CdLS children feature various cardiac malformations. To mimic the disease and test a therapeutic strategy, we generated a C57/Bl6 Nipbl+/- mouse model of the disease. These mice featured a severe delay in both embryonic and postnatal growth. The Nipbl-deficient embryonic and neonatal hearts developed ventricular hypertrophy, aortic and valve defects associated with a persistent truncus arteriosus and a ventricular septal defect. Muscles derived from the second heart field were deficient in the Nipbl haplo-insufficient mouse embryos. The adult hearts then featured a severe aortic senescence phenotype and a stenosis resulting in an increase in aortic flux velocity and persistent left ventricular hypertrophy. Using proteomics and RNA-sequencing in embryos, we identified a dysregulated TGFβ pathway in the outflow tract of embryonic hearts as well as the presence of senescent cells as early as in E13.5 Nipbl+/- embryonic hearts, limb primordium cartilage as well as in different post-natal tissues including muscle and brain cortex. Treatment of pregnant Nipbl+/- mice with a TGFβR (ALK5) inhibitor from E9.5 to E13.5 prevented cell -senescence and rescued the cardiac phenotype as well as the body size of mice at birth. Altogether our data revealed that an exacerbated TGFβ pathway associated with cell senescence is at the origin of many defects in a CdL mouse model. This druggable pathway opens the path toward a potential preventive and/or therapeutic strategy for post-natal CdLS patients.


Introduction
Cornelia de Lange syndrome (CdLS) is a rare genetic and developmental disorder affecting about 1:10,000/1:30,000 children.The syndrome is sometimes diagnosed at prenatal stages and most often at birth because of distinct facial features of babies due to cranio-facial malformations.A majority of CdLS children also presents with a short stature, mild to profound neuro-cognitive disabilities, microcephaly, upper limb defects, as well as gastro-esophageal reflux 1,2 .
Cardiac defects are also observed in more than 50 % of patients 3,4 .Cardiac malformations arise in principle from defects in differentiation, and/or migration of cardiac progenitors and of cells emerging from the embryonic second heart field 5 .
CdLS children hearts feature septal defects and outflow tract defects including hypoplastic aorta, stenosis, or coartation of great arteries as well as Tetralogy of Fallot.
Since its discovery, CdLS has been described as a clinically highly variable disease, which suggested a multigenic origin or a causative gene playing a multifunctional role.
The first and main NIPBL (Nipped-B-like protein) gene whose mutations are responsible for the syndrome has been uncovered in 2004 6 .NIPBL main function is to load the cohesin complex onto DNA and to ensure a stability of the genome 7 although NIPBL may also work like a transcription factor binding gene regulatory regions independently from cohesin 8 .
Although other genes encoding proteins of the cohesin complex or associated proteins (RAD21, SMC1A, SMC3, HDAC8, BRD4) have also been found mutated in CdLS patients or in patients with associated syndromes, NIPBL mutations and in turn gene haploinsufficiency explain a large spectrum of CdLS patients 9 .While the syndrome has been described in 1933 by Pr Cornelia de Lange 10 , no therapeutic strategy has been proposed and patients have to be managed by clinicians on a symptomatic and social basis.Both the clinical heterogeneity of patients and the likely multiple functions of a large multi-domain protein such as NIPBL make both the understanding of the disease and the research of therapeutic targets challenging.
However, most of these models do not faithfully recapitulate the human syndrome at least in a reproducible manner.They however allowed to identify dysregulation of specific genes and of pleiotropic signaling pathways such as the Wnt pathway 11,13 as well as DNA repair and cell senescence pathways [19][20][21] in nipbl haploinsufficient cells but no therapeutic strategy could emerge from these studies.Interestingly, a crosstalk was recently described between cohesinopathies and TGF-related disorders 22 .For example, CdLS patients feature some defects in great arteries that are common to those observed in Marfan or Loeys-Dietz syndrome 23 .
More specifically, TGF-dependent pathway was found to be constitutively activated in a recently described cohesinopathy 24 .Interestingly, SMC protein (SMC3), a cohesin complex component also named chondroitin sulfate proteoglycan 6 is an extracellular protein expressed in smooth muscle cells and activated by TGF 25 .
Interestingly, SMC3 mutated CdLS patients feature a high incidence of cardiac defects 26 , Being aware of mouse models of genetic diseases that may or not recapitulate human diseases according to their genetic background and overall level of expression of the gene of interest 27,28 , we first characterized the cardiac phenotype of a novel C57Bl/6J mouse model of Nipbl haplo-insufficiency and combined it with human CdLS patient-specific iPS cells in order to uncover a potential therapeutic target of CdLS.
We found that Nipbl+/-mice featured a significant decrease in Nipbl mRNAs as well as a moderate decrease in the protein in the heart.These mice also present a severe delay in embryonic and postnatal growth.The heart at birth featured ventricular hypertrophy associated with a persistent truncus arteriosus (PTA).The adult hearts then feature a severe aortic phenotype with an enlargement of the intima including senescent cells and a stenosis resulting in an increase in aortic flux velocity and persistent left ventricular hypertrophy.Using proteomics and RNA-sequencing, we identified a dysregulated TGF pathway in the outflow tract of embryonic hearts and the presence of senescent cells as early as in E13.5 Nipbl+/-embryonic hearts, as well as in post-natal muscle and neonatal brain cortex.
Treatment of pregnant mice with a TGFR (ALK5) inhibitor from E9.5 to E13.5 prevented cell senescence and rescued the cardiac phenotype as well as the body size of mice at birth.

Characterization of Nipbl+/-mice
Nipbl+/-haplo-insufficient mice were generated by deleting the exon 2 that includes the ATG of Nipbl in one allele using an ubiquitous CAG cre mouse and the Nipbl floxed mouse 29 .The Nipbl+/-mice were then backcrossed for ten generations in the C57Bl/6J genetic background.
C57Bl/6J Nipbl+/-mice featured a severe growth delay as shown by the small size of E13.5 embryos, neonates as well as 2 months old adult mice (Fig. 1a).In order to better evaluate this growth delay, we first measured the length of the tibia, the thickness of ribs as well as the length of fingers in the front leg of neonatal mice in skeleton stained with Alizarin Red and Alcian Blue. Figure 1b and 1c revealed a significant decrease in all measured parameters in Nipbl+/-mice compared to wild type (wt) mice.We then more specifically investigated the radius composition in neonatal mice at the cellular level.Both the radius and its hypertrophic zone were significantly shorter in Nipbl+/-mice when compared to wild type (Fig. 1d).The reduced hypertrophic zone was associated with a reduced expression domain of Collagen type 2 (Col2) and 10a1 (Col10a1) in Nipbl+/-mice when compared to wild type littermates (Fig. S1).
We next monitored expression of Nipbl mRNA in E9.5 and neonatal whole Nipbl+/and wild type hearts by Q-PCR. Figure 1e shows a significant decrease (down to 20 %) in Nipbl transcripts in Nipbl+/-when compared to wild type.We also looked at the protein level more specifically in cardiomyocytes isolated and purified from neonatal hearts.Western blot (direct or after immunoprecipitation) also showed a decrease of about 30 % confirmed by blotting after enrichment by immunoprecipitation of the NIPBL protein.Interestingly, Mau2 was also decreased in Nipbl+/-haploinsufficient myocytes (Fig. 1f) while cohesin complex subunit SMC1A was unaffected.

Cardiac phenotype of nipbl+/-mice
We first looked at the cardiac phenotype of Nipbl+/-mice at birth.High Resolution Episcopic Microscopy (HREM) revealed that Nipbl+/-mice featured a severe ventricular hypertrophy (Fig. 2a).The wall of left ventricle was twice as thick in Nipbl+/-mice when compared to wt (Fig. 2b).This hypertrophy was associated with a septation defect of the great vessels and in turn, both a persistent truncus arteriosus and a stenosis of the distal artery in Nipbl+/-hearts.The aortic valve leaflets were also thickened in Nipbl+/-hearts when compared to wild type hearts (Fig 2a ) as further revealed by Mowat staining (Fig. 2c).The right ventricle, which usually appears as crescent shaped in wt hearts, largely lost this shape in Nipbl+/-hearts (Fig. S2).The latter also revealed a ventricular septum defect (i.e., an interventricular communication) at the apical region (Fig. S2).
Cardiac hypertrophy was also observed in two months old adult Nipbl+/-mice first revealed by an increase in contractility of left ventricle as monitored in echocardiography (Fig. 3a).Doppler Echocardiography also showed an increase in aortic flux velocity (Fig. 3b) associated with a decrease in the diameter of the aorta (Fig. 3c) of Nipbl+/-mice when compared with wild type mice.The phenotype was fully penetrant and observed in 92% of mice (n=24).
Immunostaining of hearts and more specifically of the aorta with an anti-smooth muscle actin (SMA) antibody showed an increase in thickness of the aortic wall of Nipbl+/-mice (Fig. 3d-e) when compared to wild type mice.Interestingly the media of Nipbl+/-mice aorta was filled with large cells which did not express SMA (Fig. 3d yellow inset).Of important note, these cells were positive for H2AX suggesting a senescent phenotype.
Conditional deletion of Nipbl specifically in the smooth muscle lineage using the SMA CreERT2 mouse with the recombinase activated by tamoxifen at E11.5 fully recapitulated the functional aortic phenotype observed in Nipbl+/-mice.As a consequence, adult mice lacking Nipbl in smooth muscle cells featured an increased aortic flux at 4, 7 and 11 weeks (Fig. S3).

Skeletal muscle phenotype of nipbl+/-mice
We next looked at the skeletal muscles derived from the second heart field 30,31,32 in both Nipbl+/-mice and after specifically deleting Nipbl in the second heart field using the Mef2cAHF cre mouse.In Mef2cAHF cre /Nipbl fl/fl homozygous E13.5 embryos, the trapezius muscles, the extra-ocular muscles, the oesophagal muscle, the pervertebral muscles of the neck and the masseters were atrophied or missing when compared to heterozygous Mef2cAHF cre /Nipbl fl/+ embryos (Fig. S4).Similar observations were recorded in Nipbl+/-embryonic mice (data not shown).
As many CdLS patients present gastro-oesophagal reflux, we further examined the skeletal oesophagal muscle in 2 months old adult mice and found that this muscle was markedly less developed in Nipbl+/-mice (thickness 19±3 m, n=3) than in wild type mice (47±5 m, n=3) (Fig. S5a).

Senescence at the origin of tissue and organ defects in Nipbl+/-mouse
In order to characterize the origin of PTA we investigated back the embryonic heart.
Nipbl+/-haplo-insufficient embryonic E13.5 hearts (n=9 from 4 separate litters) also featured the septation defect of the outflow tract (Fig. 4).Interestingly we found that many p21+ senescent cells accumulated at the base of the common outflow vessel.
We found that each 10 m section of wt muscle cell featured one or none p21+ cells while the ones of Nipbl+/-mice revealed 8±2 p21+ cells (n=3) (Fig. S5b).We also observed in the cortex of neonatal mouse brains an abnormal presence of p21+ neuronal cells in Nipbl+/-mice compared to wt (Fig S5d).Altogether, these observations strongly point to senescence as a major consequence of nipbl haploinsufficiency.
We thus searched for the origin of cell senescence in Nipbl+/-mice.More specifically, to better understand the outflow tract phenotype of Nipbl+/-mice, we performed a proteomic analysis of the outflow tract in E16.5 embryos.6 wt and 6 Nipbl+/-embryos were collected, the heart explanted and the outflow tract entirely dissected out.Each OFT was individually subjected to mass spectrometry analysis.
As shown in Figure 5, Nipbl+/-OFT featured a loss of proteins present in the extracellular matrix (Supplementary data 1).We specifically highlighted four proteins emelin1, fibrillin 1 and 2 and actbl2 that are dramatically under-expressed in Nipbl+/-OFT as compared to that of wild-type.Remarkably, all these proteins are involved in the TGF signaling pathway including fibrillins, which sequester TGF 33 or in related aortic diseases 23 .
RNA-sequencing of the same cardiac regions (7 OFT from E16.5 wild type and Nipbl+/-embryonic hearts) revealed an increase in expression of 1199 genes (Supplementary data 2) including genes involved in OFT formation such as slit2, Robo1, six2 and Hoxb, Hoxc and Hoxd as well as TGF2 genes in Nipbl+/compared to wt OFT.TGF2 upregulation was further confirmed by RT-Q-PCR (Fig. 5b, inset).Anterior Hoxa genes were for most of them downregulated in OFT of Nipbl+/-vs wild type mice as well as pbx1, all genes whose haploinsufficiency are associated with PTA 34,35 .

CdLS iPS cells and smooth muscle derivative feature a cell senescence phenotype
To get more insight into the molecular mechanisms underlying senescence in Nipbl+/-haplo-insufficient mice, we switched to a human model using iPS cells derived from CdLS patient cells.3 iPS cell lines were used from patients harboring the NIPBL mutations c.6242.g>C in exon 35, or c.6860T>C in exon 40 or c.6516-6517 in exon38.All patients featured cardiac malformations.Both the human embryonic stem cell line H9 and an iPS cell line from a healthy volunteer were used as controls.CdLS IPS cells showed a decrease in NIPBL and an exclusion of the protein from the nucleolus.We also observed fragmentations of nucleoli in all CdLS IPS cells (Fig. 6 a,b) in contrast to wt cells which showed only two nucleoli, a feature of pluripotent stem cells.Cells were then differentiated in smooth muscle cells.While wild type cells still proliferated after one week of differentiation, CdLS cells stopped dividing and expressed for most of them p21 (Fig. 6c).Thus, the analysis of human cells confirmed that CdLS cells are programmed early towards a cell senescence phenotype.

Decreased motility of mesenchymal cells of proximal OFT
As senescence is known to affect cell motility, and in order to confirm senescence phenotype, we investigated the motility of cells undergoing epithelial-to-mesenchymal transition (EMT) in the proximal outflow tract at E10.5 in both wt and nipbl+/embryonic hearts.To such an aim we cultured explants dissected out from proximal OFTs on collagen gel for 48 hrs.We then monitored the distance of migration of cells undergoing EMT from the explant.Wild type cells migrated significantly farther than Nipbl+/-cells (Fig. S6) consistent with a senescence state conferred by Nipbl haplo-insufficiency.

A TGF(ALK5) receptor inhibitor rescues the phenotype of nipbl+/-mice
As our data pointed to a hyperactive TGFpathway, we reasoned that its inhibition could prevent the senescence and the associated phenotypes observed in Nipbl haplo-insufficiency mouse model.Thus, we tested the effect of the ALK5 inhibitor Galunisertib on Nipbl+/-mice.Pregnant mice were treated from E9.5 up to E13.5 with 30 mg/kg/day of galunisertib.Embryos were first collected at E13.5.Remarkably, and as illustrated in Figure 7, embryonic Nipbl+/-hearts did not feature anymore PTA when compared to wild type hearts from the same litter.Distinct pulmonary trunk originating from the right ventricle and aorta from the left ventricle were indeed observed (Fig. 7a). in addition, very few p21+ cells were observed at the base of the outflow region of both the wild type and Nipbl+/-embryonic hearts (inset in Fig. 7a).
Expression of both Col2 and Col10 was similarly fully rescued (Fig. S8).Such a rescue was already observed at E18.5 stage of embryonic development (data not shown).
Furthermore, the thickness (32± 5m, n=3) of second heart field derived skeletal oesophagal muscle was fully restored in adult offspring from a galunisertib-treated mother and no or very few p21+ cells were detected in the muscle (Fig. S8a).
Similarly, the number of p21+cells that were scored in the cortex of neonatal Nipbl+/mice born from galunisertib-treated others was not significantly different from the one scored in wt neonatal mice (Fig. S8 b,c) The aorta of nipbl+/-adult mice originally born from galunisertib-treated mothers did not feature stenosis anymore (Fig S9 ) Finally, in order to assess the relevance of these findings in human cells, we treated CdLS patient-specific iPS cells-derived smooth muscle cells with 10 M galunisertib during their differentiation together with TGF1 and PDGFfor 6 days (Fig. S10).In The smooth muscle cells at the base of the outflow trunk derive from progenitors of the second heart field 37 and possibly from a myocardial to smooth muscle cell transdifferentiation 38 .This explains that deletion of Nipbl in neural crest cell lineage did not affect heart and more specifically OFT formation 29 .
A good candidate to account for the senescence of smooth muscle cells is TGF2.
Indeed TGF family has been involved in senescence of many cell types 39 .More specifically TGF2 is an inducer of smooth muscle cell senescence 40 , consistent with the protection of cells against senescence conferred by TGF inhibition we report in this work.
Smooth muscle cells at the base of the outflow trunk deficient in Nipbl+/-mice are regulated by both TGF2 and retinoic acid pathways, two pathways dysregulated in Nipbl+/-mice.Overactivation of TGF2 associated with a decrease in RA signaling 36 , a frequent association of events 41,42 , lead to defect in septation of OFT 37 .
TGF2 availability in OFT is likely increased in Nipbl+/-mice by a decrease in expression of extracellular matrix proteins including fibrillin, known to sequester the growth factor 33 .
The OFT phenotype could have been further worsened by a collagenolytic activity of senescent smooth muscle cells, a phenomenon mediated by the Senescence Associated Secretory Phenotype (SASP), and regulated by p38 MAPK 43 .More generally senescence is associated with a dysregulation of the extracellular matrix (ECM) 40 .This may account for the downregulation of many ECM proteins we observed in the proteomics profile of the E16.5 embryonic OFT of Nipbl+/-mice while they remained unaffected at the RNA level.The fact that many genes were upregulated while many proteins were downregulated in Nipbl+/-mice also point to a block in translation likely due to defect in nucleoli (Fig 6) and in turn in ribosomal activity.
We also observed senescence of both undifferentiated and differentiated smooth muscle cells from CdLS patients (Fig 6).Finally, cell senescence was also present in smooth muscle cells of adult Nipbl+/-aorta (Fig 3) as well as in the second heart field derived skeletal oesophagal muscle, in hypertrophic chondrocytes and in neonatal brains (Fig. S5).
The regulation of embryonic senescence of the growth plate cartilage by the TGF superfamily was reported as a determinant of the length of the bones 441.An exacerbated TGFpathway induced cell senescence in primordium cartilage (Fig 4c ) in Nipbl+/-mice likely explains their delayed growth.Along the same line, impairment in neuron migration was reported in Nipbl+/-neonatal brain 45 , which may also be due to a senescence of neurons just as the one we observed in neonatal brains.
Using a TGFRI, ALK5 inhibitor, used in oncology (Galunisertib) in pregnant mice at mid-gestation in a time window between E9.5 and E13.5 rescued the size of the neonate (Fig. 7, S7), restored the septation of the OFT in Nipbl+/-mice (Fig. 7) and prevented cell senescence in skeletal oesophagal muscle, aorta, and brain cortex (Fig. S8,S9).The prevention of cell senescence in brain cortex might be clinically relevant as post-natal brain cortex features changes in chromatin 3D structure 46 and still performs synaptogenesis 47 .This opens a postnatal therapeutic window.
The treatment of mothers also significantly decreased the number of senescent cells surrounding or in the OFT (Fig. 7).NIPBL+/-IPS cell-derived smooth muscle cells treated with Galunisertib also prevented their senescence (Fig. S10).This further suggests that TGF2 mediates its deleterious function through cell senescence of many cell types.Thus, collectively our data revealed that an exacerbated TFGpathway and associated cell senescence are at the origin of many defects in a CdLS mouse model.As this pathway is druggable, our research opens the path toward potential preventive, or even curative (at least for neuronal defects) therapeutic strategies for postnatal CdLS patients,
Nipbl flox/flox mice were obtained from Dr Heiko Peters, Institute of Genetic medicine, Newcastle University, UK.Mice were kept under standardized conditions (20-25°C temperature; 50%±20% humidity) on a 12 h light/12 h dark cycle in an enriched environment (kraft paper enrichment).Food and tap water were provided ad libitum.
Nipbl+/-mice were generated by deleting exon 2 of the gene using Nipbl flox/flox mouse, bred with a CAG CreERT2+/-(JAX laboratory) mouse and gavaged with tamoxifen.The mice were then backcrossed for 10 generations in C57Bl6j genetic background.
Nipbl+/-mice were maintained as heterozygous mice by breeding them with C57Bl/6J mice.Pregnant mice were separated from the male as soon as a vaginal plug was observed and daily observed.Two to three days before the female gives birth, a wild type C57Bl/6J mouse was added to the cage to prevent the mother from killing the small Nipbl+/-neonate.Nipbl+/-neonates were weaned after 4 weeks.
Mice were genotyped by PCR of tail biopsies lysed in proteinase K for 3 hours using the primers of Table 1.

High Resolution Episcopic Microscopy.
Neonatal hearts of wt or Nipbl+/-were fixed for 24h in 4% PFA and washed in PBS before being embedded in plastic resin for HREM according to the manufacturer.Images were processed using Imaris software.

M-Mode and Doppler Ultrasound Transthoracic Echocardiography.
Echocardiography was performed using an Affiniti 50 (Philips) and a 15 MHz probe.

Contractility of the left ventricle was acquired in a long-axis configuration in M-Mode.
Aortic flow was monitored using PW Doppler-mode, by positioning the Doppler sample volume parallel to flow direction, assisted by Color Doppler-mode.
Echocardiography recordings were analysed blinded by a cardiologist who had not performed the recordings.A minimum of 6 mice in groups were monitored.
Cell imaging.Images were acquired using a confocal LSM 800 Zeiss microscope equipped with an airryscan of 32 detectors.Light was provided by a laser module 405/488,561 and 640 nm wavelengths, respectively.
All images were acquired using the ZEN ZEISS software.Then some images were deconvoluted using Autoquant and reconstructed in 3D using Imaris software (IMARIS).Episcopic images were also reconstructed in 3D using Imaris software.Some large images (whole embryonic section, brain sections) were acquired using an Axioscan Z1 (Zeiss).All samples were mounted in Fluoromount™ (Cliniscience, France).
Alzarin Red and Alcian Blue staining of neonatal skeleton.Skin and viscera were removed from sacrificed neonatal mice.The skeletons were then place for 5 days in Ethanol 95% and 2 days in acetone at room temperature.They were the stained with Alizarin Red/Alcian Blue for 5 days at 37°C.After a quick rinse in distilled water, they were placed in 1% KOH for 4 days and then in 20% glycerol/1% KOH for 5 more days.The skeletons were imaged using an Axiozoom V.16 and a 0.5x/0.125FWD 114mm PlanAPo Z objective (Zeiss).

Eosin and hematoxylin and MOWAT staining of hearts. Cryosections of hearts
were stained with eosin and hematoxylin according to standard procedures or with MOWAT pentachrome stain kit (Biosite MPS2) according to manufacturer instruction.

In situ hybridization and Safranin Weigert staining of forelimbs
Forelimbs were fixed in 4% formalin overnight.E18.5 and forelimbs were decalcified in 25% EDTA overnight prior to embedding in paraffin.5 µm serial sections were stained with Safranin Weigert and used for in situ hybridization using Digoxigenin-11-UTP labelled antisense probes against Col2, Col10, Ihh and p21 as described previously (Schroeder et al. 2019).Brightfield images were taken on a Zeiss Axioplan2 microscope with a SPOT 14.2 camera and measures using SPOT advanced software.Fluorescent images were taken on a Zeiss Axio Observer7 microscope with an AxioCam 506 mono CCD camera and Zen2.3 software.
Expression domains of all genes were measured using ImageJ.For each measurement, 3 to 6 sections were analyzed per animal.Length differences were compared between littermates.iPS cells culture and differentiation : IPS cells were derived from skin fibroblasts using Cytotune Sendai virus kit (Thermofisher, France) and cultured on MEF in KO-DMEM supplemented with KO-SR (Thermofisher, France), 10 -7 M mercaptoethanol, Non-Essential Amino Acids and 10 ng/ml FGF2 as previously described 48 .Cells were screened for any chromosomal defects using the iCS-digital TM PSC 24-probes kit ( Stemgenomics, Montpellier, France) Cells were transferred on geltrex-coated plate in StemFit4 medium (NipponGenetics, Europe, Germany) for two passages and were differentiated into smooth muscle cells in RPMI-B27 medium supplemented with CHIR9901 8 M for 24hrs, CHIR9901 4 M and BMP2 10 ng/ml for 24hrs, BMP2 10 ng/ml and IWR1 5M for 24 hrs and for 6 days with TGF1 10 ng and PDGF20 ng/ml.Proximal Outflow tract explants.Outflow tracts were dissected out from E10.5 wt and nipbl+/-embryos.The proximal part was cut and open along the long axis to be transferred on a collagen gel 493 , the endocardial side being on the gel.The explants were cultured for 48 hrs before being fixed with PFA 4%, permeabilised with Triton X100 0.1% and stained with DAPI.The explants were imaged in confocal microscopy using a 10X long distance objective.A z-stack of images was acquired.Distance of cell migration was scored using Image-J.
RNA extraction, real time PCR and RNA sequencing.Total RNA was extracted from E16.5 OFT hearts (Nipbl+/-n=12 and controls n=12 from 6 separate litters) using Zymo Research Corp kit ZR RNA Miniprep following the manufacturer's protocol.For the real time PCR, one µg of total RNA was used to synthetize cDNAs using oligo(dT) primers and affinity script reverse transcriptase (Agilent technologies France).Real-time quantitative PCR analyses were performed using the Light Cycler LC 1.5 (Roche, France).For each condition, expression was quantified in duplicate, and GAPDH was used as the housekeeping gene or normalizing RNA in the comparative cycle threshold (CT) method 44  For the RNA sequencing, total RNA was isolated from E16.5 OFT hearts (Nipbl+/-n=12 and controls n=12 from 6 separate litters) and pooled either for wt or Nipbl+/groups.The two pooled samples were used for the RNA-seq library preparation, using the kit TruSeq Stranded mRNA by Illumina.
Transcripts discovery was performed using Cufflinks (v2.2.1) with the "library-type" argument set to fr-firstrand, and a GTF file obtained from GENCODE ("Comprehensive gene annotation", vM1) provided as the genomic annotation.The GTF files produced for each sample by Cufflinks were combined using Cuffmerge.
The "class code" assigned to each transcript by Cuffmerge was used to defined unknown transcripts (class code"u").Only de novo transcripts with counts greater than 0 in at least one RNA-seq sample were kept for subsequent analyses.These de contrast to non-treated cells, galunisertib treated smooth muscle cells did not express p21 (Fig S10), indicating that in humans as in mouse model, attenuation of TGFsignaling pathway protects cells from senescence.Altogether, these results demonstrate that pharmacological inhibition of TGFsignaling pathway, which is found hyperactive in our CdLS mouse model, in pregnant mice fully restores cardiac function and body size in newborn mice and also prevents senescence in newborn mice and in CdLS iPSCs derived smooth muscle cells.Discussion Nipbl+/-haplo-insufficient mice in the C57Bl/6J genetic background recapitulate many features of the CdLS patients.More specifically, cardiac hypertrophy, PTA, ventricular septal defect and aortic stenosis were observed in a large majority of mice.The mice were shorter and featured deficient skeletal muscle derived from the second heart field including the oesophagal muscle, the masseters, all muscles involved in food intake.This muscle phenotype could explain the gastro-oesophagal reflux observed in CdLS children, at least partly.The ventricular hypertrophy was likely a consequence of the persistent truncus arteriosus affecting the hemodynamics in the outflow trunk following both a thickening of the vessel wall and stenosis 36 .Several observations point to senescence of several cell types including smooth and skeletal muscle cells in nipbl+/-mice which limited both their number and their motility.Indeed, E13.5 nipbl+/-embryos featured many senescent cells at the base of the outflow trunk (Fig 4).OFT explant experiments further highlighted a decrease in motility of these cells at E10.5, i.e., at the onset of EMT of endocardial cells in the outflow trunk (Fig S7) This observation likely accounts for a septation defect of the OFT.

Fig. 1 Fig. 2
Fig.1 Characterisation of Nipbl+/-mouse.a, from left to right: E13.5 embryos, neonates and 2 months adult mice.The images are representative of at least 15 embryos or mice at each stage of development.b, Alizarin red and Alcian Blue stained neonatal skeletons.c, Graph of the size of tibia, the thickness of ribs and the length of fingers (measurements made as shown in the right inset) (n=5, ** p<0.01).d, Total radius length and the hypertrophic zone (black line) were measured in Safranin-Weigert stained forelimbs sections of E16.5 Nipbl+/-and wt mice.Dots show individual measurements of serial sections from 2 Nipbl+/-, or 2 wt embryos,* p<0.05, scale bar: 100 µm).e, Q-PCR of nipbl from E9.5 embryonic hearts and neonatal hearts (n=3 from 3 separate litters;** p<0.01), f, Western blot of NIPBL using an antibody directed against the N-terminal domain of NIPBL (Bethyl) and SMC1A in whole lysate of neonatal heart or after immunoprecipitation using anti-whole NIPBL antibody (Abcam)

Fig. 3 Fig. 4 Fig. 5 :Fig. 6 Fig. 7 .
Fig. 3 Adult cardiac phenotype of nipbl+/-mice.a,b,c echocardiography of 2 months old mice.a, shortening fraction b, maximal aortic flux velocity c, diameter of the aorta (n≥ 7 mice) d, Anti-SMA staining of aorta.Yellow insets show high magnification of cells within the aortas.Green inset:the aorta of WT and NIPBl+/-aorta were stained by an anti H2AX antibody.e, graph of aortic wall thickness (n=5) t-test **p<0.001.