Dullard-mediated Smad1/5/8 inhibition controls mouse cardiac neural crest cells condensation and outflow tract septation

We thank the reviewers for their thoughtful and constructive comments. We have divided our response into three sections, the points highlighted by the reviewers and the editorial board. The answers to the specific points have been incorporated in these three points. We provide several new supplementary figures and the text has been modified accordingly.

A) Response to the general comments:

Point 1. The lack of detail on the temporal development and symmetry/asymmetry of NCC deployment into the OFT and how the phenotype evolves morphologically and in the context of related events such as EndMT.

Point 1.1 A phenotypic characterization of Dullard mutant hearts anchored in a larger developmental window.

We now provide substantial phenotypic analysis of Dullard mutants throughout a larger time window than presented in the previous version of the manuscript, which bring insights into the role and time of action of the phosphatase during heart morphogenesis (Figure 2C-F—figure supplement 2).

First, labelling the cardiac NCC, the myocardium and/or the endocardium of E10.5 embryos indicates that the morphology of the OFT is not impacted at this stage by the loss of Dullard in the NCC compartment (Figure 2—figure supplement 1A). Hence, the premature NCC condensation in Dullard mutant OFT is likely established after E10.5.

Second, we now show that Dullard deletion using either the Wnt1^cre or the Pax3^credrastically compromises the survival of embryos after E12.5 onwards (Figure 2C). Histological characterization of the rare embryos that can survive up to E14.5 or E18.5 revealed that their heart exhibits i) an opened interventricular septum, ii) a hypertrophy of the right ventricle, iii) an aorta overriding the two ventricles and finally a pulmonary stenosis (Figure 2D, E). Hence, as proposed by one of the reviewers, the phenotype of the heart developed in the presence of cardiac NCC missing Dullard is highly reminiscent to the heart of patients affected by the most cyanotic congenital heart disease, named tetralogy of Fallot (TOF) (Neeb et al., 2013). We now discuss how our work brings insight into the etiology of this condition, which is still poorly understood (cf. Snider and Conway, 2011).

Finally, we imaged the blood flow within the abdominal artery of E11.75 control and Dullard mutant embryos using doppler ultrasound (Figure 2F) (Nomura-Kitabayashi et al., 2009). It indicates that the flow is generally weaker in Dullard mutants compared to controls, as indicated by a lower systolic velocity peak in mutants than control. Hence the OFT defects triggered by the loss of the phosphatase in the NCC worsen the blood circulation of embryos and could in turn impair the survival rate of embryos.

Point 1.2 A better characterization of the effect of Dullard loss on the OFT cell types communicating with the NCC (experimental procedures detailed here will also address the questions raised in point 3).

We provide in our revised manuscript a better description of the differentiation and morphogenesis of the non-NCC derived OFT tissues. Overall these analyses sustain the idea that the NCC derived cardiac cushion is the tissue mostly affected by the loss of Dullard. No identity defect in the specification of the myocardium was found. Similarly, we had barely found any changes in the endocardium of Dullard mutants compared to controls. To illustrate this, we have:

– Performed bioinformatics analyses of our 44 gene base-transcriptomes centred on the RFP⁺;CD31⁺ endocardial cells and to the RFP⁺;CD31^- OFT cells following the pipeline used to describe GFP⁺ NCC transcriptomes (Figure 5A, Figure 5—figure supplement 1). As established for the GFP⁺ NCC, the endocardium cells and the RFP⁺;CD31^- OFT cells display transcriptomic variations and can be segregated into distinct populations (Figure 5Ai, ii). However, in contrast to the NCC cells, most of these populations all contain both control and mutant cells, with the exception of two populations of endocardial cells (Figure 5Ai’, ii’). The RFP⁺;CD31⁺ pop 1 is enriched for cells coming from control embryos, while the RFP⁺;CD31⁺ pop 6 contains more mutant cells than control cells. Again, in contrast to NCC case, very few of the genes that drive the segregation of the CD31⁺ cells into 6 populations display clear variations in the pop 6 or the pop 1 compared to the other pops (Figure 5—figure supplement 1B, C). Twist1 stands as one of the pop1 signature genes and appears downregulated in the absence of Dullard. Accordingly, immune detection of this transcription factor revealed few positive Twist1⁺ endocardial cells in E11.5 control embryos, which were barely detectable I n mutant endocardium (Figure 6B).

– Brought further evidence that the Wnt1^Cre and Pax3^Cre drivers activity are not detectable in the endocardium to the reviewers. The yellow staining detected on the sections stained for GFP and PECAM correspond to overlapping membranes at sites of cell-cell interaction captured with a large pinhole confocal acquisition. FACS plots further showed a clear segregation between the GFP⁺ and the CD31⁺ cells (Author response image 1). Finally, within the NCC transcriptomes, we did not detect a cohort of genes which define endocardial cells (e.g. Figure 5—figure supplement 2). We have not included the data of the Author response images in the current version of the manuscript, as many papers have already extensively described the activity of these Cre drivers in the developing heart (Brown et al., 2001; Jiang et al., 2000).

– Performed several immunostaining or ISH for key markers of the myocardium, showing in agreement with transcriptomic data that the differentiation tissue and its pattern along thee bis not majorly affected by Dullard loss. Isl1 or myosin heavy chain (MF20) in E11.5 control and Dullard hearts displayed similar profile and expression levels in control and Dullard mutant embryos (Figure 2—figure supplement 1). Similarly, Tbx1 and Pitx2 expression patterns at the base of the pulmonary trunk and its expression level were comparable in control and mutant embryos (Author response image 2). Finally, more data on Sema3c mRNA distribution in Dullard mutants also indicate that Sema3c expression in the myocardial cuff is not altered by Dullard’s deletion in the NCC (Figure 1—figure supplement 1D).

N.B. We have tested several antibodies against the Cdh5, none of those were efficient enough to mark whether the elevated Cdh5 mRNA levels in Dullard mutant could also be detected at the protein level.

Author response image 1

Download asset Open asset

Author response image 2

Download asset Open asset

Point 2. The specificity of the phenotype with respect to BMP signalling.

We now provide further support for the lack of Dullard leading to an elevation of BMP signalling.

– A new statistical test (two way-Anova) indicate that the variations in PSmads levels both between control and mutant cells at a given proximo-distal axis level and between the distal and proximal regions of the OFT in a given genetic background are significant (significance for the differences between Wnt1^cre,Dullard^Flox/Flox and Wnt1^cre;Dullard^Flox/+; p-value<0.002).

– Our single cells transcriptomic analysis reveals that out of the genes differentially expressed between control and Dullard mutant cardiac NCC stand well known pan BMP signalling direct target genes, namely Id1, Id2, Msx1 and Msx2 (Figure 1F). Accordingly, Id2 and Msx2 detected using RNAscope probes exhibit elevated levels in the mutant NCC compared to control NCC or to adjacent positive tissues (Figure 1F). Hence, not only the phosphorylation of Smad1/5/8 is increased by the loss of Dullard, but their transcriptional activity is likely to be upregulated as well.

– We provide also data showing that the loss of Dullard using the Pax3^cre driver not only lead to elevation of BMP signalling in cardiac NCC, but also in other lineages marked by this Cre driver and exposed to BMP ligands in their environment (Figure supplement 1D and data not shown). It includes for instance several muscle masses in the trunk and limb bud regions, as well as the dorsal neural tube. Furthermore, the muscle masses of Dullard mutants also display elevated levels of Sema3c (Figure 1—figure supplement 1D). These results further support the idea that Dullard stands as a pan and generic inhibitor of BMP signalling in several embryonic tissues.

Furthermore, we have attempted to rescue Dullard mutants by performing intraperitoneal injections to pregnant females with the Alk2 and Alk3 BMP receptors inhibitor LDN193189 at two different doses (Author response image 3). Unfortunately, we haven’t been able to downregulate the levels of PSmads neither in controls nor in Dullard mutants, and Dullard mutants treated with LDN still displayed OFT septation defect. This could stem from the instability in the LDN, issues from its absorption and diffusion within the mother and embryonic tissues.

Author response image 3

Download asset Open asset

We have also checked the state of TGFβ signalling in Dullard mutants, as this pathway has been shown to be increased upon deletion of the phosphatase in the limb bud mesenchyme (Hayata et al., 2015). Immunostaining for the phosphorylated form of one of the downstream effectors of this signalling pathway (Smad2) indicates that the levels of activation of this TF is not impaired in absence of Dullard (Figure 1—figure supplement 1E). Hence, in NCC Dullard does not appear to be required for dampening TGFβ signalling levels.

Finally, we have clarified in the Discussion our statement on whether the ingression of NCC may contribute to the generation of a BMP signalling gradient. From our data, it is clear that Dullard is not implicated in the establishment of the gradient of BMP signalling, but in the regulation of its amplitude. We have proposed that this is regulated by other negative regulators of the pathway and could be linked to temporal dynamics in this signalling pathway. Over time the signal decreases, hence cells which have been submitted earlier to BMP and which have ingressed further display lower levels of signalling than cells which have been in the presence of BMP ligands for a shorter period of time (Bier and Robertis, 2015).

Point 3. The possible role of Sema3C and other differentially expressed genes in the development of the phenotype.

As proposed by one of the reviewers, we have taken the opportunity to further develop our discussion in the revised version of the paper on how the differentially expressed genes may contribute to the establishment of the Dullard mutant phenotype. Notably, we believe that the elevation of Sema3c expression is central to the establishment of the OFT defects in Dullard mutants. Based on a handful of papers that have demonstrated in different cell lines Sema3c requirement for cohesion between cells, we propose that the elevated levels could promote the early compaction of the NCC in Dullard mutants.

References:

Bier E and Robertis EMD (2015) BMP gradients: A paradigm for morphogen-mediated developmental patterning. Science 348: aaa5838

Brown CB, Feiner L, Lu MM, Li J, Ma X, Webber AL, Jia L, Raper JA and Epstein JA (2001) PlexinA2 and semaphoring signaling during cardiac neural crest development. Dev. Camb. Engl. 128: 3071–3080

Hayata T, Yoichi, Ezura, Asashima M, Nishinakamura R and Noda M (2015) Dullard/Ctdnep1 Regulates Endochondral Ossification via Suppression of TGF-β Signaling. J. Bone Miner. Res. 30: 318–329

Jiang X, Rowitch DH, Soriano P, McMahon AP and Sucov HM (2000) Fate of the mammalian cardiac neural crest. Development 127: 1607–1616