Abstract
Disruption of cardiac neural crest cells (CNCCs) results in congenital heart disease, yet we do not understand the cell fate dynamics as these cells differentiate to vascular smooth muscle cells. Here we utilized single-cell RNA-sequencing of NCCs from the pharyngeal apparatus with heart in control mouse embryos and when Tbx1, the gene for 22q11.2 deletion syndrome, is inactivated. We uncovered three dynamic transitions of pharyngeal NCCs expressing Tbx2 and Tbx3 through differentiated CNCCs expressing cardiac transcription factors with smooth muscle genes, and that these transitions are altered non-autonomously by loss of Tbx1. Further, inactivation of Tbx2 and Tbx3 in early CNCCs resulted in aortic arch branching defects due to failed smooth muscle differentiation. Loss of Tbx1 interrupted mesoderm to CNCC cell-cell communication with upregulation of BMP signaling with reduced MAPK signaling and failed dynamic transitions of CNCCs leading to disruption of aortic arch artery formation and cardiac outflow tract septation.
Introduction
Neural crest cells (NCCs) are multipotent cells that migrate in three ordered streams from the rhombomeres in the neural tube to the pharyngeal apparatus where they differentiate to many cell types1. The pharyngeal apparatus is a dynamic embryonic structure consisting of individual pharyngeal arches (PA), forming in a rostral to caudal manner from mouse embryonic day (E) 8 to E10.5. A subset of pharyngeal NCCs migrate through the caudal pharyngeal arches, PA3-6, and surround the pharyngeal arch arteries (PAAs), while others continue to migrate to the cardiac outflow tract (OFT), both differentiating to vascular smooth muscle cells2. Ablation of NCCs from PA3-6 results in interruption of the aortic arch and arterial branching defects as well as persistent truncus arteriosus of the OFT3. These NCCs in PA3-6, are referred to as cardiac NCCs (CNCCs) based upon their position and known function in heart development as well as their differentiation to vascular smooth muscle. Understanding CNCC development is critical to determine the pathogenesis of human congenital heart defects such as those observed in 22q11.2 deletion syndrome (22q11.2DS) patients4, 5.
TBX1, encoding a T-box transcription factor, is the major gene for congenital heart disease in 22q11.2DS. Although 22q11.2DS is largely considered to be a neurocristopathy, Tbx1 is not significantly expressed in CNCCs6, but it is strongly expressed in adjacent cells in the pharyngeal apparatus including the mesoderm. Global inactivation of Tbx1 or conditional inactivation in the mesoderm using Mesp1Cre 7 in the mouse results in neonatal lethality with a persistent truncus arteriosus8–10, in part due to failed CNCC development6. Therefore, one of the main functions of Tbx1 in the pharyngeal mesoderm is to signal to CNCCs to promote their development. In order to understand how CNCCs are affected non-autonomously in Tbx1 mutant embryos, it is essential to define their transcriptional signatures and cardiac fate acquisition in the normal situation between E8.5 and E10.5, when Tbx1 is expressed in the pharyngeal apparatus and when inactivated, on a single cell level.
Previously, single cell RNA-sequencing (scRNA-seq) of NCCs from early stages in the chick embryo identified expression of Tgif1, Est1 and Sox8 being important for early CNCC identity and fate decisions11. However, these were early migrating mesenchymal NCCs that also have the potential to contribute to the craniofacial skeleton and other cell types. Another seminal scRNA-seq study demonstrated that NCC fate choices are made by a series of sequential binary decisions in mouse embryos at E8.5-10.512 but did not focus on detailed steps of cardiac fate acquisition or investigate Tbx1 function12.
To uncover genetic signatures and dynamic transitions of CNCCs in the normal situation and when Tbx1 is inactivated, we performed scRNA-seq of NCCs from control and Tbx1 null mutant mouse embryos. We found that smooth muscle cell fate acquisition is in part dependent on two other T-box genes, Tbx2 and Tbx3. When Tbx1 is inactivated, we found failure of dynamic progression of CNCC maturation due to disruption of cell-cell communication from mesodermal cells, resulting in down regulation of MAPK signaling and upregulation of the BMP pathway, as well as affecting other, known and novel, ligand receptor interactions.
Results
Single cell transcriptional profiling of NCCs in the pharyngeal apparatus
We performed scRNA-seq of the Wnt1-Cre, ROSA-EGFP genetic lineage13, 14 in the mouse pharyngeal apparatus at E8.5, E9.5 and E10.5 (Fig. 1A-F). These stages correspond to developmental time points when Tbx1 is highly expressed in cell types adjacent to NCCs. At E8.5, the anterior half of the embryo was dissected (Fig. 1A), while at E9.5 the pharyngeal apparatus with heart was microdissected (Fig. 1B). At E10.5 arches two to six and the heart were included in the dissection (Fig. 1C). EGFP positive NCCs were purified by FACS and the Chromium 10X platform was used to perform scRNA-seq and data from 36,721 NCCs were obtained (Supplementary Table 1). Unsupervised clustering was performed using Seurat software15 and individual clusters were identified (Fig. 1D-F).
Expression of Sox10 and Twist1 were used to identify early migratory and mesenchymal NCCs, respectively16. We used Hox and Dlx (Homeodomain) genes to provide spatial context to different arches (PA2-6; 17). At E8.5, Sox10 and Twist1 show overlap in expression, while at E9.5, expression became complementary, with a relative reduction of Sox10+ NCCs and increased Twist1+ NCCs in the expanded populations of mesenchymal NCCs (Fig. 1G-I). At E8.5 and E9.5, Hoxa2 was expressed in PA2 and PA3-6, while Hoxb3 was expressed only in PA3-6 containing NCCs that will invade the OFT and surround the PAAs (Fig. 1D,E,G,H). Using the Hox genes as a guide, at E9.5, early migrating Sox10+ NCCs of PA2 and PA3 were clustered together (cluster C4), suggesting that they have a similar transcriptional profile. At E10.5, the relative proportion of Twist1 expressing mesenchymal cells increased with respect to reduction of Sox10 expressing cells (Fig. 1F, 1I). Further, at E10.5, Hoxa2 expression was expanded within the mesenchymal cell populations, and Hoxb3 was expressed in NCCs of PA3-6 (cluster C3 at E9.5 is similar to C2 at E10.5; Fig. 1F, 1I). Additional marker genes are shown in Supplementary Data 1 (E8.5), 2 (E9.5) and 3 (E10.5). Spatial localization was confirmed for Sox10, Hoxa2 and Hoxb3 expression by wholemount RNAscope in situ hybridization (Fig. 1J-L). In addition, to anterior-posterior spatial localization of the cells, we identified their proximal-distal location in the PAs with Dlx2, Dlx5 and Dlx6 (Supplementary. Fig 1).
Identification of cardiac NCC gene signatures
Differentiated NCCs of the OFT and PAAs express smooth muscle genes such as smooth muscle actin, Acta218, 19. Acta2 is a representative marker gene of smooth muscle cells that include expression of Tagln, Myl9, Myh9, and Cnn1. We identified a cluster of cells expressing Acta2 at E9.5 and 10.5 (cluster C14, Fig. 2A; C10, Fig.3A), but not at E8.5. To delineate molecular signatures of cardiac NCCs (CNCCs), we evaluated genes that are co-expressed with Acta2 and identified known genes for cardiac development including Tbx2, Tbx3, Msx2, Isl1, Gata3 and Hand2, at E9.5 and E10.5 (Fig. 2A; Fig. 3A). These genes are not only expressed in Acta2+ cells but also in NCCs in the distal PA1-3 expressing Dlx5 at E9.5 (C1, C3; Fig. 2A) and in PA2-6 at E10.5 (C2, C3, C4; Fig. 3A). At E9.5, we validated co-expression of ISL1 in CNCCs in the OFT of which some expressed ACTA2 (Fig. 2B). Further, RNAscope in situ analysis confirmed the expression of Gata3, Isl1 and Msx2 in CNCCs within the OFT at E9.5 (Fig. 2C-E). At E10.5, Isl1 and Gata3 were expressed in CNCCs within the cardiac cushions of the distal OFT and in the mesenchyme of the dorsal aortic sac wall and aortic sac protrusion (Fig. 3B). Gata3 was expressed in a larger domain of the OFT than Isl1. When taken together, we now identify the genetic signatures of CNCCs of the forming OFT.
Tbx2 and Tbx3 were widely expressed in pharyngeal NCCs at E9.5 (Fig. 2A), but their expression was restricted to cell clusters comprising PA3-6 at E10.5 (Fig. 3A). Tbx2 and Tbx3 were expressed immediately lateral and dorsal to Isl1 and Gata3 expressing CNCCs in embryos at E10.5 by RNAscope analysis (Fig. 3C-E). In addition, Tbx2 and Tbx3, but not Isl1 and Gata3, were expressed in NCCs surrounding the PAAs that are differentiating to smooth muscle at E10.5 (Fig. 3D,E,G). Both ISL1 and TBX2 proteins were expressed in smooth muscle cells of the OFT and PAAs, respectively (ACTA2 or TAGLN; Fig. 3F,G). Expression of Tbx2, Tbx3, Isl1, Gata3 and Acta2 in NCCs at E10.5 is illustrated in Fig. 3H. A subset of pharyngeal NCCs will form the CNCCs, defined as NCCs expressing markers specific to the cardiac or smooth muscle lineages. We refer to the CNCCs in the pharyngeal arches as P-CNCCs (Fig. 3H). Therefore, CNCCs can be subdivided into four populations based upon position and expression of cardiac or smooth muscle genes, referred to as P-CNCCs, PAA-CNCCs of PAAs expressing Acta2, OFT-CNCCs of the OFT expressing Isl1 and Gata3 and SM-CNCCs of the OFT that express Acta2 (Fig. 3H).
We noted earlier that some CNCCs were located in clusters from PA1 and PA2 (C1; Fig. 2A), that are not typically considered to harbor CNCCs. Consistent with this, at E9.5 (20 somites) we found that the OFT was connected to PA2 and CNCCs from PA2 are entering the OFT (Fig. 2F). At late E9.5 (24 somites), the OFT was located between PA2 and PA3 and the first CNCCs from PA3 were entering the OFT (Fig. 2G). These data are consistent with evidence from a previous report20,that NCCs from anterior arches also contribute to the developing heart.
Cardiac NCC fate dynamics that drive differentiation to smooth muscle cells
To uncover CNCC cell fate dynamics at E9.5, we used CellRank software21 (Fig. 2H, I). We discovered genes that were progressively activated during the transition from pharyngeal NCCs to SM-CNCCs, which are candidate cardiac lineage driver genes (Fig. 2J, K; Supplementary data 4 for the full list of genes). Our analysis indicates that CNCCs progressively activate Tbx2, Tbx3, Msx2, Hand2, Gata3 and Isl1 expression during their commitment towards Acta2+ smooth muscle cells at E9.5 (Fig. 2J,K).
To understand how CNCCs progress at E10.5, when there are more smooth muscle cells in the pharyngeal arches, we used CellRank software and generated PAGA (partition-based graph abstract) plots (Fig. 3I-K). The cell fate probability map from CellRank identified cells with a high potential to differentiate to smooth muscle fates (from cluster C2 and C3 to C10; Fig. 3J). The PAGA plots further indicated that some pharyngeal NCCs (cluster C2) are P-CNCCs and they transition to OFT-CNCCs (cluster C3) that then transition to SM-CNCCs (cluster C10; blue color fraction in pie chart; Fig. 3K). This data also indicates that a small fraction of P-CNCCs may directly differentiate to smooth muscle cells (blue fraction in the pie chart in C2), in agreement with Tbx2 and Tbx3 expression in PAA-CNCCs at E10.5 (Fig. 3D, E, G). We identified genes whose expression correlates with SM-CNCC fate acquisition (Fig. 3L, M; Supplementary data 5 for full list of genes at E10.5). Representative genes were ordered according to their expression peak in pseudotime and included Tbx2, Tbx3, Foxf1, Isl2, Msx2, Isl1, Hand1, Hand2, Mef2c, Rgs5, Gata3, Acta2, Gata4 and Gata6.
We generated lineage driver gene sets by dividing the genes in the fate probabilities heat map from Bmp4 to Gata6, least to most differentiated to SM-CNCCs, to four groups of equal size (Fig. 3L; Supplementary data 5). Next, we performed Gene Ontology (GO) enrichment analysis using ToppGene Suite22 to understand the function of the genes in each group (Fig. 3N; Supplementary data 6). Our analysis indicates that the initially activated genes of pharyngeal NCCs that include some P-CNCCs, are associated with general pharyngeal arch development processes (e.g. Hox, Dlx, Six2 genes). Then cell division (cell cycle) genes are highly expressed, consistent with the expansion of pharyngeal NCCs during development23, together with cardiac development genes. Finally, genes important in cardiac development, cell adhesion and actin-filament processes (Hand1, Gata3/4/5/6, Isl1, Acta2) become strongly expressed (Fig. 3N). In addition, our functional enrichment analysis identified genes associated with congenital heart disease such as tetralogy of Fallot, double outlet right ventricle and ventricular septal defects (Supplementary data 6), which supports the importance of the genetic program of CNCCs in OFT formation and disease.
Thus, here we identified a specific CNCC transcriptomic signature at E9.5-10.5 and revealed that cell fate acquisition to smooth muscle cells requires a multistep specification process. We additionally identified new genes such as Dkk1, Gata3, Foxf1, Isl2, Tbx2, Tbx3, Rgs5, among others, which have not yet been considered as CNCC markers (Supplementary data 4 and 5).
Tbx2 and Tbx3 are required in cardiac NCCs for aortic arch branching
Tbx2 and Tbx3 are expressed in multiple tissue types within the pharyngeal apparatus and global inactivation of both genes leads to early embryonic lethality with severe cardiac defects24–27. To understand the requirement of Tbx2 and Tbx3 in NCCs, we generated Wnt1-Cre/+;Tbx2f/f;Tbx3f/fdouble conditional mutant embryos (Tbx2/3 cKO). We performed intracardiac ink injection and histological analysis at E15.5 (Fig. 4A-I) and found that 38.5% of Tbx2/3 cKO embryos had an aberrant retro-esophageal right subclavian artery (ARSA) but no intracardiac defects (Fig 4A-I and M). No defects were identified in Wnt1-Cre/+;Tbx2f/f;Tbx3f/+ nor in Wnt1-Cre/+;Tbx2f/+;Tbx3f/f embryos. The right subclavian artery is formed from the right 4th PAA. By immunostaining on coronal sections of Wnt1-Cre/+;Tbx2f/f;Tbx3f/f;ROSA-EGFPf/+embryos at E11.5 using GFP and ACTA2 antibodies, we found that NCCs contributed to the right 4th PAAs but failed to differentiate into smooth muscle cells (Fig. 4J-L). Bmp4 and Foxf1 have been identified as regulators of smooth muscle cell differentiation in other organs28. We found that Bmp4 and Foxf1 expression is activated temporally after Tbx2 and Tbx3 expression during cardiac fate acquisition (Fig. 3L).
Disruption of cardiac NCCs by loss of Tbx1
In Tbx1 null mutant embryos, the caudal pharyngeal apparatus is hypoplastic and unsegmented at E9.5 and E10.5 due in part to failed deployment of NCCs6 (Fig. 5A, Fig 7A). Further, CNCCs fail to enter the shortened cardiac OFT, leading to a persistent truncus arteriosus later in development6. We found that Tbx1 was not noticeably expressed in NCCs (Supplementary Fig. 2) and its conditional deletion in NCCs using Wnt1-Cre did not lead to cardiac defects (Supplementary Fig. 3).
To understand how the absence of Tbx1 affects development of CNCCs, we performed scRNA-seq of NCCs isolated from the microdissected pharyngeal region plus heart of Tbx1 null mutant embryos at E9.5. We obtained sequencing data from 11,301 NCCs (Fig 5A; Supplementary Table 1) and integrated scRNA-seq data from control and Tbx1 null embryos using RISC (Robust Integration of scRNA-seq) software29. Even though there were visibly fewer NCCs in the pharyngeal apparatus (Fig. 5A), there were no missing cell clusters in Tbx1 null embryos (Fig. 5B-C). We then compared the proportion of cells in each cluster among the total number of NCCs in each dataset. As expected, there was a reduction in the relative proportion of NCCs from Tbx1 null embryos as compared to controls, as shown in Fig. 5D, in PA3 (C4,1.5 fold), in proximal PA2 (C8, 1.4 fold), in cluster C9 corresponding early migrating NCCs in PA2 and PA3 (1.2 fold), and in C5 corresponding to early migrating NCCs in PA1 (1.4 fold). In addition, there was an increase of the relative proportion of NCCs in cluster C3 (1.6 fold) that corresponds to the distal part of PA1 and PA2 (Fig. 5D) and it is known that cells from PA2 abnormally migrate to PA1 in Tbx1 null embryos6.
Altered BMP and MAPK signaling pathways in the absence of Tbx1
We examined the data to identify differentially expressed genes (DEGs) in mutant versus control embryos at E9.5. Surprisingly, we found few DEGs per cluster at this stage (Supplementary data 7). However, in Tbx1 null embryos there was a clear increase in the expression of genes that act downstream of BMP signaling in proximal PA2 and PA3 (clusters C4, C8; Fig. 5E). This includes increased expression of Msx2, Bambi, Gata3, Dkk1, Smad6, Id2 and Id3. Our analysis also revealed a downregulation of the expression of genes in the MAP kinase (mitogen-activated protein kinase) signaling pathway including Spry2, Spry4, Myc, Foxo1, Lyn, and Dusp3 (Fig 5E). Signaling by BMP30 and growth factors activating the MAPK pathway31 are two signaling pathways known to be critical for NCC development and migration during embryogenesis. Gene enrichment analysis of DEGs by cluster profiler R software32 confirmed an increase in expression of genes in the BMP signaling pathway (Fig. 5F, Supplementary data 8) and a decrease in expression of genes related to negative regulation of MAPK cascade and activity (Fig. 5G, Supplementary data 9). There was an increase in expression of Msx2 and Gata3 (downstream in the BMP pathway) as well as reduced expression of Spry4 (MAPK pathway) in scRNA-seq data of Tbx1 null embryos (Fig. 5H). Expression of BMP downstream genes, Msx2 and Bambi, were expanded dorsally in Tbx1 null mutant embryos by wholemount RNAscope in situ and 3D reconstruction (Fig. 5I). These results were confirmed by RNAscope assays on traverse sections of control and Tbx1 null embryos (Fig. 5J). In addition, there was an increase and ectopic expression of P-SMAD1/5/9, marking an increase in BMP signaling, in NCCs towards the dorsal part of the pharyngeal apparatus in Tbx1 null embryos (PA2, PA3; Fig. 5K). These data suggest that altered BMP and MAPK signaling might affect NCC development in Tbx1 null embryos. A schematic representation of expanded BMP signaling and reduced NCCs migrating to the shortened OFT in Tbx1 null embryos at E9.5 is shown in Fig. 5L.
Cell-cell communication from the mesoderm to NCCs is disrupted in the absence of Tbx1
During formation of the heart, NCCs receive critical signaling from adjacent mesodermal cells (Fig. 6A). We investigated cell-cell communication and how it is disrupted in the absence of Tbx1 at single cell resolution using CellChat software33.
Inactivation of Tbx1 in the mesoderm results in similar pharyngeal hypoplasia and altered NCC distribution as in global null embryos, implicating the pharyngeal mesoderm as being critical to signal to NCCs34. To identify Tbx1-dependent signals from the mesoderm to NCCs, we investigated existing scRNA-seq data from Mesp1Cre control and Tbx1 conditional null embryos at E9.535. We focused on mesodermal subpopulations, expressing Tbx1, that are adjacent to the NCCs including the anterior and posterior second heart field (aSHF; pSHF)36, 37. We also included a critical Tbx1-dependent multilineage progenitor population (MLP) in the pharyngeal mesoderm required for cell fate progression to the aSHF and pSHF35. We examined signaling to NCCs in clusters corresponding to migrating NCCs of the future PA2-6 (C9), distal part of PA1-2 (C3), mesenchyme of PA2 (C8) and PA3-6 (C4), and CNCCs of the OFT (C12) in integrated scRNA-seq data from control and Tbx1 null embryos at E9.5 (Fig. 5; Fig. 6B). Representative results of ligand-receptor pairs altered when Tbx1 is inactivated are shown in Fig. 6B, and the complete set of pairs are in Supplementary Fig. 4.
Affected ligands in the mesoderm include Wnt5a, Wnt2, Sema3c, Pdgfa Nrg1, Fgf8, Fgf10, Bmp4 and Edn3 and others. To validate relationships, we analyzed integrated Mesp1Cre data (Fig. 6C; MLPs-C8, aSHF-C10 and pSHF-C1+C12). Isl1, is a critical gene required for OFT development38, and it is expressed in the MLPs, aSHF and pSHF (Fig. 6D). We examined expression changes of Wnt5a, Wnt2, Sema3c, Pdgfa, Nrg1, Fgf10 and Edn3 (Fig. 6E-K). These genes were altered in expression in the cell types specified and, in the direction of altered signaling (decreased or increased in the mutant embryos), as indicated in Fig. 6B.
Reduced expression of Wnt5a39, 40, Fgf835, 41, 42, Fgf1043, 44, Sema3c45 and Nrg135 ligands and increase of Wnt246 are consistent with previous in vivo studies of Tbx1 mutant embryos, however these were not known with respect to cell-cell communication to NCCs. With this data, we show on a single cell level, that this signaling to NCCs is altered in Tbx1 mutant embryos. Two additional ligand genes, Edn3 and Pdgfa were not investigated regarding Tbx1 (Fig. 6H, 6K). Edn3 encodes an endothelin ligand important in cell migration but not well known with respect to Tbx1. Pdgfa encodes a growth factor regulating cell survival, proliferation and migration and PDGF signaling is required in NCC development47.
We investigated CellChat results for the BMP and MAPK pathways that were altered in NCCs when Tbx1 was inactivated. An abnormal increase of the BMP signaling from the mesoderm to NCCs in Tbx1 mutant embryos was found through Bmp4/5/7 ligands (Fig. 6B). Fgf8 and Fgf10 are ligands in FGF signaling that act through the MAPK pathway and it is well known that they are reduced in expression in Tbx1 mutant embryos35, 41–44 (Fig. 6J). Similarly, we found that Fgf8 and Fgf10 were reduced in expression in the mesoderm and signaling to NCCs was altered (Fig. 6B and Supplementary Fig. 4). It is known that FGF and BMP pathways can act antagonistically48, 49. Therefore, it is possible that reduction of FGF and changes in other ligands in the adjacent mesoderm, could result in ectopic BMP and reduced MAPK signaling in NCCs leading to their failed progression.
Failed cardiac cell fate progression of NCCs in the absence of Tbx1 at E10.5
To further understand how contribution of NCCs to the OFT is altered in the absence of Tbx1, we performed scRNA-seq of NCCs in Tbx1 null embryos at E10.5, when the caudal pharyngeal apparatus is extremely hypoplastic (Fig. 7A). We integrated scRNA-seq data from two control replicates (21,561 cells) and two Tbx1 null replicates (17,840 cells) using RISC software (Fig. 7B). The integrated datasets clearly show a strong reduction in the number of NCCs in most clusters in Tbx1 null embryos including OFT-CNCCs (Isl1/Gata3+; C10) and SM-CNCCs (Acta2+; C13) as shown in Fig. 7C, except that the relative proportion of pharyngeal NCCs that contain P-CNCCs (Tbx2/Tbx3+; C4) is not changed, after considering the total cells being sequenced (Fig. 7D). Our analysis showed that the cell fate probabilities point from pharyngeal NCCs containing P-CNCCs (C4) and OFT-CNCCs (C10) toward SM-CNCCs (C13) as shown in Fig. 7E. We generated a list of DEGs in clusters C4 and C10, between control and mutant embryos at E10.5 (Supplementary data 10). Consistent with data from E9.5, Msx2, Bambi, Gata3 and Dkk1 were increased and ectopically expressed in pharyngeal NCCs in Tbx1 null embryos (Fig. 7F, Supplementary data 10), suggesting an abnormal upregulation of BMP signaling in the absence of Tbx1, consist with data in Figure 6. By GO analysis, we found downregulation in expression of genes involved in embryonic organ development and mesenchyme development in pharyngeal NCCs that contain P-CNCCs (Fig. 7G; Supplementary data 11), suggesting dysregulation of NCC development. Interestingly, there was an upregulation of genes that inhibit cell cycle progression of OFT-CNCCs (Fig. 7H; Supplementary data 12). By immunostaining, we confirmed the overall reduction in the number of CNCCs within the OFT and NCCs in the pharyngeal region of Tbx1 null embryos (Fig. 7I,J). We also confirmed a reduced number of ISL1+ NCCs in dorsal aortic sac wall mesenchyme and distal OFT and absence of the aortic sac protrusion in Tbx1 null embryos (Fig. 7I). Supporting the scRNA-seq data, immunostaining experiments indicated that TBX2 (and likely TBX3) expression is maintained in NCCs located in the lateral part of the pharyngeal apparatus. We also noticed normal differentiation of the CNCCs within the OFT of Tbx1 null embryos despite that there are fewer cells (Fig. 7J). Together this suggests a failure of cardiac fate progression between pharyngeal NCCs and OFT-CNCCs states in the absence of Tbx1.
Discussion
In this report, we identified the signatures and cell fate dynamics of CNCCs. We focused on pharyngeal NCCs in the mouse at developmental stages when Tbx1, the gene for 22q11.2DS is highly expressed and functions. We determined the mechanisms by which Tbx1 non-autonomously regulates CNCC maturation at a single cell level and found altered BMP and MAPK signaling may contribute to cardiovascular malformations when Tbx1 is inactivated. We also uncovered novel genes and ligand-receptor pairs with respect to cell-cell communication from mesoderm to NCCs that are new to our understanding of CNCC fate progression.
NCCs are multipotent and differentiate to many cell types including smooth muscle cells. Through examination of transitional dynamics along with embryonic localization by in situ analysis, we uncovered three main transition states from pharyngeal to smooth muscle expressing CNCC derivatives, termed P-CNCCs, OFT-CNCCs and SM-CNCCs as shown in the model in Fig. 8A. The P-CNCCs express Hox and Dlx genes, as well as those implicated in cardiac development including Tbx2, Tbx3, Six2, Shox2, Bmp4, Prdm1, Daam2, Scube1, Angptl1, Tfap2b and Mef2c. Our data also suggests that some Tbx2/3+ cells differentiate directly into Acta2+ smooth muscle in the PAAs (PAA-CNCCs; Fig. 8A). A role of Tbx2 and Tbx3 in CNCCs in PAA development have not been previously described and we found that inactivation results in abnormal arterial branching at reduced penetrance, indicating besides function as markers of P-CNCCs, they have a role in smooth muscle differentiation. The second state of OFT-CNCCs express cardiac transcription factors such as Hand1/2, Msx1/2, Mef2c and Gata3 as well as Isl1 and Isl2 (Fig. 8A). These cells are required for OFT development, as determined by conditional inactivation studies in NCCs of Hand250, Msx1 and Msx251. Expression of Isl1 in CNCCs contributing to the OFT is consistent with a dual lineage tracing study52. We also identified genes not previously connected to CNCCs that include Dkk1, Foxf1, Rgs5, Isl2 and Gata3. Finally, the third state is SM-CNCCs within the OFT that express Acta2 smooth muscle genes together with Gata4, Gata5 and Gata6. Supporting their requirement, conditional deletion of Gata6 in NCCs results in septation defects of the OFT53. We additionally found novel genes not yet connected to these cells, including Meox1, Bambi, Smad6 and Smad7. We found that progression of CNCC fate toward smooth muscle of the OFT is associated with progressive downregulation of genes involved in pharyngeal embryonic development and progressive increase in cardiac specification genes, consistent with maturation of the unipotent cellular state to smooth muscle cells (Fig. 8A). Development of NCCs in the pharyngeal apparatus is regulated by cell-cell signaling, in particular from the pharyngeal mesoderm as uncovered by studies of Tbx1 mutant embryos34, 54. In global null or Mesp1Cre mediated Tbx1 conditional null embryos, there is altered deployment of CNCCs and reduced contribution to the OFT leading to a persistent truncus arteriosus34, 54. Our results suggest that NCCs are produced normally in the neural tube in the absence of Tbx1, but they fail to migrate to the caudal pharyngeal arches and OFT and we suggest this is due to disrupted signaling from adjacent mesodermal cells to the NCCs. Here, we identified several receptor-ligand interactions that are disrupted by comparing scRNA-seq data from NCCs and the
Mesp1Cre lineages
Using CellChat software33, we confirmed known interactions and their disruption in Tbx1 mutant embryos, now for the first time to NCCs at a single cell level. For example, we identified a reduction of FGF signaling from mesodermal cells to NCCs. Reduced expression of Fgf ligands including Fgf8 and Fgf10 in the mesoderm of Tbx1 null embryos have been reported previously44, 55. FGF and BMP signaling, both important for NCCs development, can act in an antagonistic manner48. In addition, FGF signaling activates the MAPK signaling pathway that is critical for NCC development and migration31. Here we propose a model in which FGF paracrine signaling from the mesoderm is required to restrict BMP signaling and activate MAPK signaling in pharyngeal CNCCs necessary for their development and progression to the heart (Fig. 8B). In the absence of Tbx1, FGF signaling from the mesoderm is reduced leading to ectopic and overactivation of the BMP pathway and abnormal down-regulation of MAPK pathway in adjacent pharyngeal CNCCs that fail to develop correctly (Fig. 8B). This is consistent with our in vivo observation that BMP signaling is abnormally expanded in NCCs in the pharyngeal region of Tbx1 null mutant embryos and with our in silico study showing reduction of downstream effector genes in the MAPK pathway. Reduced phospho-ERK1/2 has previously been reported in NCCs of Tbx1 null mutant embryos56. Our investigation also indicates that BMP4-Bmpr1a+Bmpr2 signaling from mesoderm progenitor populations to NCCs in PA2-6 is abnormally upregulated in Tbx1 mutant embryos. BMP4 can activate BMP signaling through p-SMAD1/5/957, 58 and our analysis indicates no change in expression of Bmpr1a and Bmpr2 genes in NCCs of Tbx1 null embryos at E9.5, raising the possibility of a direct upregulation of BMP signaling in CNCCs by mesoderm cells.
Interestingly, our analysis also reveals that Neuregulin (Nrg1)-ERBB3 signaling from the MLP to the pharyngeal NCCs of PA2 to PA6 at E9.5 is downregulated in Tbx1 mutant embryos. Neuregulin is important for migration of NCCs acting as a chemoattractant and chemokinetic molecule59 and it is involved in heart development. It has been shown recently that Nrg1 is a direct transcriptional target gene of Tbx1 in the multilineage progenitors (MLPs) in the mesoderm35. Therefore, alteration of Neuregulin signaling in Tbx1 mutant embryo could contribute to explain failed cardiac contribution of the NCCs. Together, our analyses indicate that a combination of important signaling from the pharyngeal mesoderm to NCCs are affected in Tbx1 mutant embryos and could contribute to failure of fate progression of CNCCs.
The pharyngeal endoderm is also an important source of signaling during development, including FGF ligands60, that could potentially affect CNCCs development. It will be interesting to evaluate how the exchange of signaling between the pharyngeal endoderm and NCCs are affected in the absence of Tbx1.
In conclusion, in this report we identified the transcriptional signature that defines the CNCCs and identified the gene expression dynamics that regulates CNCC fate progression into smooth muscle of the OFT and PAAs. In addition, we highlight direct alteration of FGF signaling from the mesoderm to CNCCs resulting in an abnormal increase in the BMP pathway and failed cardiac contributions in the absence of Tbx1 at a single cell level. Together our results allow a better understanding of the normal development of CNCCs and provide new insights into the origin of congenital heart defects associated with defective NCCs and 22q11.2DS.
Methods
Mouse lines
The following transgenic mouse lines were used: Wnt1-Cre 13, ROSA26R-GFPf/+ 14 that we refer to as ROSA-EGFP, Tbx1+/-61, Tbx1f/+ 61. Mice were maintained on a mixed Swiss Webster genetic background. Tbx2f/+ and Tbx3f/+ mutant mouse lines were generated in Dr. Chenleng Cai’s laboratory by inserting a two LoxP sites into the intron sequences flanking exon 2 of the Tbx2 gene and exons 2-4 of Tbx3 gene, by gene targeting using homologous recombination. When the genomic sequences between the LoxP sites are floxed out, the reading frame and T-Box domain of Tbx2 and Tbx3 are both disrupted. Tbx2f/+ and Tbx3f/+ mice were maintained on a mixed Swiss Webster and C557BL/6 genetic background. Mice and embryos resulting from the different crosses were genotyped by PCR using standard protocols from DNA extracted from toes tips or yolk sac. Animal experiments were carried out in agreement with the National Institutes of Health and the Institute for Animal Studies, Albert Einstein College of Medicine (https://www.einsteinmed.org/administration/animal-studies/) regulations. The IACUC number protocol is #00001034. Embryos were collected at different embryonic days dated from the day of the vaginal plug (E0.5). For each experiment, a representative result is presented from at least three analyzed embryos.
Immunofluorescence staining on paraffin sections
Embryos were collected in cold 1x PBS (Phosphate Buffered Saline) and fixed in 4% PFA (Paraformaldehyde) for 1 hour at 4°C under constant agitation. Embryos were progressively dehydrated in ethanol then xylene and embedded in paraffin (Paraplast X-tra, Sigma P3808). The embryos were sectioned to 10μm thickness and sections were deparaffinized in xylene and progressively rehydrated in an ethanol series. Sections were incubated and boiled in antigen unmasking solution (Vector laboratories, H-3300) for 15 minutes. After cooling at room temperature, sections were washed in PBS containing 0.05% Tween (PBST) and blocked for 1 hour in TNB buffer (0.1M Tris-HCl, pH 7.5, 0.15M NaCl, 0.5% Blocking reagent [PerkinElmer FP1020]) at room temperature. Then sections were incubated with primary antibodies diluted in TNB overnight at 4°C. Sections were washed in PBST and incubated with secondary antibodies diluted in TNB for 1 hour at room temperature. After washes in PBST, nuclei were stained with DAPI (1/1,000; Thermo Scientific, 62248) and slides were mounted in Fluoromount (Southern Biotech). Embryonic sections were imaged using a Zeiss Axio Imager M2 microscope with ApoTome.2 module. The following primary antibodies were used: goat anti-GFP (1/200, Abcam ab6673), mouse anti-alpha smooth muscle actin ACTA2) (1/200, Abcam ab7817), rabbit anti-αSMA (ACTA2; 1/200, Abcam ab5694), mouse anti-Isl1/2 (1/100, DSHB 40.2D6 and DSHB 39.4D5), mouse anti-TBX2 (1/100, Santa Cruz sc514291), rabbit anti-TAGLN (1/200, Abcam ab14106), rabbit anti-pSMAD1/5/9 (1/100, Cell Signaling D5B10) and rabbit anti-TBX1 (1/100, Lifescience LS-C31179). Donkey secondary antibodies from Invitrogen (Thermo Fisher Scientific) were used (anti-goat, anti-mouse and anti-rabbit).
RNAscope
RNAscope on wholemount embryos
Embryos were collected and dissected in 1x PBS at 4°C and fixed in 4% PFA overnight. Then embryos were dehydrated in progressive methanol washes and stored in 100% methanol at −20°C. Wholemount RNAscope was performed using RNAscope Multiplex Fluorescent Detection Reagents v2 kit (Advanced Cell Diagnostics, ref 323110). Embryos were progressively rehydrated in 1x PBS containing 0.01% Tween (PBST) and were permeabilized using Protease III (Advanced Cell Diagnostics, ref 323110) for 20 minutes at room temperature followed by washes in PBST. Embryos were then incubated with 100μl of pre-warmed mixed C1, C2 and C3 (ratio 50:1:1, respectively) RNAscope probes at 40°C, overnight. After 3 washes in 0.2x SSC (Saline Sodium Citrate), 0.01%Tween, embryos were fixed in 4% PFA for 10 minutes at room temperature. Embryos were then incubated in Amp1 for 30 minutes at 40°C, Amp2 for 30 minutes at 40°C and Amp3 for 15 minutes at 40°C, with washes in 0.2x SSC, 0.01%Tween between each step. Tyramide Signal Amplification (TSA) solutions were prepared as follows: 1/500 for TSA-Fluorescein (Akoya Biosciences, NEL741001KT), 1/2000 for TSA-CY3 (Akoya Biosciences, NEL744001KT) and 1/1000 for TSA-CY5 (Akoya Biosciences, NEL745001KT). To reveal C1 probes, embryos were incubated in HRP1-C1 for 15 minutes at 40°C then washed and incubated with the chosen TSA solution, 30 minutes at 40°C. The amplification reaction was blocked using HRP-Blocker during 15 minutes at 40°C. C2 and C3 probes were revealed following the previous steps and using HRP-C2 for C2 probes and HRP-C3 for C3 probes. Nuclei were stained overnight with DAPI. Wholemount embryos were imaged as Z-stacks using a Leica SP5 confocal microscope or a Nikon CSU-W1 spinning disk confocal microscope. 3D image reconstruction and analyses were performed using Fiji and ImarisViewer 9.8.0 software.
RNAscope on cryosections
Embryos were collected and dissected in 1x PBS at 4°C then fixed in PFA 4% overnight and incubated in successive 10%, 20% and 30% sucrose (Sigma-Aldrich S8501) solutions and then embedded in OCT (Optimal Cutting Temperature compound). Embryos were stored at −80°C until they are used. RNAscope was performed on 10μm sections mounted on SuperFrost Plus slides (FisherScientific, 12-550-15) following the RNAscope Multiplex Fluorescent Reagent Kit v2 assay protocol from Advanced Cell Diagnostics.
Probes used for RNAscope
Egfp (400281, 400281-C3), Tbx2 (448991), Hoxb3 (515851), Dlx2 (555951), Bambi (523071), Hoxa2 (451261), Gata3 (4033321-C2), Isl1 (451931-C2), Tbx3 (832891-C2), Sox10 (435931-C2), Msx2 (421851-C2), Meox1 (530641-C2), Dlx5 (478151-C3).
Histology and staining with Hematoxylin & Eosin
Fetuses were collected and dissected in 1x PBS and fixed overnight in 4% PFA. They were progressively dehydrated in ethanol and incubated in xylene prior to embedding in paraffin. Tissue sections of 12μm thickness were deparaffinized in xylene and progressively rehydrated in ethanol washes and incubated for 10 minutes in Hematoxylin (Sigma-Aldrich, HHS16) then rinsed in water and dehydrated in 70% ethanol. Sections were then incubated in alcoholic Eosin (70%) (Sigma-Aldrich, HT110116) solution and progressively dehydrated in ethanol and xylene washes prior to mounting in Permount mounting medium (Fisher Chemical SP15100). Sections were imaged using a Zeiss Axioskop 2 plus microscope.
Intracardiac India ink injection
Fetuses at E15.5 were dissected in 1x PBS and the chest was carefully opened to avoid damaging the cardiovascular system. A solution containing 50% India ink and 50% 1x PBS was injected into the left ventricle of the heart by blowing gently into an aspirator tube assembly connected to a microcapillary. Immediately after filling the left ventricle and arterial branches, the heart and aortic arch with arterial branches were imaged using a Leica MZ125 stereomicroscope.
scRNA-seq data generation
Embryos were collected and microdissected in 1x PBS at 4°C. Dissected tissues of interest were maintained in DMEM (Dulbecco’s Modified Eagle Medium, GIBCO 11885084) at 4°C. For E8.5 embryos, the rostral half of the embryos including the heart was collected. At E9.5 and E10.5 the pharyngeal region plus heart were collected. Pharyngeal arch 1 was removed at E10.5 as shown in Fig. 1. Then, tissues were incubated in 1ml of 0.25% Trypsin-EDTA (GIBCO, 25200056) containing 50U/ml of DNase I (Milipore, 260913-10MU), at room temperature for 7 minutes. Next, heat inactivated FBS (Fetal Bovine Serum, ATCC, 30-2021) was added to stop the reaction at a final concentration of 10%, at 4°C. Dissociated cells were centrifugated for 5 minutes at 300 x g at 4°C and the supernatant was removed. Cells were then resuspended in 1x PBS without Ca2+ and Mg2+ (Coming, 21-031-cv) containing 10% FBS at 4°C and filtered with a 100μm cell strainer. A total of 1μl DAPI (1mM) (Thermo Fisher Scientific, D3571) was added before FACS using a BD FACSAria II system. EGFP positive, DAPI negative cells were then centrifuged at 4°C, 5 minutes at 300 x g and resuspended in 50μl of 1xPBS without Ca2+ and Mg2+ with 10% FBS. Cells were then loaded in a 10x Chromium instrument (10x Genomics) using the Chromium Single Cell 3’ Library & Reagent kit v2 or single index Chromium Next GEM Single Cell 3’ GEM, Library & Gel Bead kit v3 or v3.1.
Sequencing
Sequencing of the DNA libraries was performed using an Illumina Hiseq4000 system (Genewiz Company, South Plainfield, NY, USA) with paired-end, 150 bp read length.
scRNA-seq data analysis
CellRanger (v6.0.1, from 10x Genomics) was used to align scRNA-seq reads to the mouse reference genome (assembly and annotation, mm10-2020-A) to generate gene-by-cell count matrices. All the samples passed quality control measures for Cell Ranger version 6.0.1 (Supplementary Table 1).
Seurat analysis for filtering and clustering
Individual scRNA-seq sample data were analyzed using Seurat V4.0.5 15, with parameters as recommended by Seurat software.
Integrated scRNA-seq analysis
After individual samples were analyzed by Seurat for clustering, the data were integrated by the RISC software (v1.5) using the Reference Principal Component Integration (RPCI) algorithm for removing batch effects and aligning gene expression values between the control and Tbx1 null samples at E9.5 and E10.5 29. The integrated data were re-clustered by RISC, using parameters adjusted to match the cell type clusters in the Seurat results. Gene expression differences between control and Tbx1 null embryos was determined by RISC software for each of the clusters at an adjusted p-value < 0.05 and log2(fold change) > 0.25. The GO enrichment for the differentially expressed genes were identified by clusterProfiler (v4.0.5) 32. Cell compositions were computed from the integrated cell clusters and used for two-proportion Z-test as implemented in the R prop.test() function to evaluate the statistical significance in changes between control and null embryos.
Cell trajectory analysis
CellRank (v1.5.1) 21 was used to infer differentiation trajectory, focusing on determining the probability of cells to adopt the smooth muscle, Acta2+ cell fate. The analysis used RNA velocity from Velocyto (v0.17.17) 62 and scVelo (v0.2.4) 62, and cell-cell similarity to infer trajectories and cell differentiation potential. The analysis was performed for all cells in either the E9.5 control sample or the two E10.5 control samples and then for the cells in the selected clusters that were predicted to have connections to the smooth muscle Acta2+ cluster.
Cell-cell communication analysis
CellChat (v1.1.3) 33 was used to identify the ligand-receptor interactions between Mesp1Cre mesoderm lineage and NCC lineages and then compare the change between control and Tbx1 null mutant data at p < 0.05. Data included ligands in the Mesp1Cre lineage and receptors in CNCC lineage.
Data Availability
All scRNA-seq datasets generated in this study have been submitted to GEO (Gene Expression Omnibus) repository July 13, 2022, and are awaiting approval.
Author contributions
C.D. and B.E.M. designed the study and experiments. C.D. performed all wet laboratory experiments. C.D., Y.L., A.F., A.V. performed computational analysis of single cell RNA-sequencing data. Y.L., A.F. and D.Z. provided bioinformatics expertise and guidance. C.D. and B.E.M wrote the manuscript. All authors read, intellectually contributed, edited, and approved the manuscript.
Competing interests
The authors declare no competing interests.
Supplementary table 1: Summary of scRNA-seq experiments. This is summary of the scRNA-seq experiments shown in Figures 1, 2, 3, 5 and 7.
Supplementary Figure 1: Dlx and Hox genes provide proximal-distal and anterior-posterior identities of NCCs in the pharyngeal arches, respectively. A) UMAP plots showing expression levels of Hoxb3, Hoxa2, Sox10, Dlx2, Dlx5 and Dlx6 genes in cell specific clusters in scRNA-seq data of NCCs at E9.5. B) Wholemount RNAscope in situ hybridization of Wnt1-Cre;ROSA-EGFP embryos at E9.5 with probes for Egfp, Dlx2 and Dlx5. PA, pharyngeal arch. Scale bar: 200 μm. This figure is related to Figure 1.
Supplementary Figure 2: TBX1 expression is not detected in NCCs at E8.5, E9.5 and E10.5. Immunostaining for EGFP (green) and TBX1 (red) on sagittal sections of Wnt1-Cre;ROSA26-EGFP embryos E8.5 (A) (n=3) and E9.5 (B) (n=6) and on transverse sections of Wnt1-Cre;ROSA26-EGFP embryos at E10.5 (C) (n=5). Note that TBX1 is not noticeably expressed in NCCs, but it is expressed in adjacent mesodermal cells. PA, pharyngeal arch; NT, neural tube; end, endoderm; OFT, outflow tract. Scale bars: 100 μm.
Supplementary Figure 3: Conditional deletion of Tbx1 in NCCs does not affect heart development. Hematoxylin and eosin staining on Wnt1-Cre;Tbx1f/+ (n=3) and Wnt1-Cre;Tbx1f/f (n=3) embryos at E14.5 showing normal aorta and pulmonary trunk septation (A,B) and normal interventricular septation in Wnt1-Cre;Tbx1f/f embryos (C,D). Ao, aorta; PT, pulmonary trunk; RA, right atrium; LA, left atrium; RV, right ventricle; LV, left ventricle. Scale bars: 500 μm.
Supplementary Figure 4: Bubble plots for all ligand-receptor pairs showing significant cell-cell signaling changes from mesodermal cells to NCCs in control and Tbx1 mutant embryos at E9.5. This version is the complete version compared to the image shown in Fig. 6B.
Supplementary data 1: Marker genes and statistics of cell clusters from scRNA-seq of NCCs at E8.5. These data are related to Fig. 1.
Supplementary data 2: Marker genes and statistics of cell clusters from scRNA-seq of NCCs at E9.5. These data are related to Fig. 1 and 2.
Supplementary data 3: Marker genes and statistics of cell clusters from scRNA-seq of NCCs at E10.5. These data are related to Fig. 1 and 3.
Supplementary data 4: Heatmap of gene expression that correlates with cardiac fate probabilities with cells ordered by fate probabilities at E9.5. This heatmap is the complete version of the selected genes shown in Fig. 2.
Supplementary data 5: Heatmap of gene expression that correlates with cardiac fate probabilities with cells ordered by fate probabilities at E10.5. This heatmap is the complete version of the selected genes shown in Fig. 3.
Supplementary data 6: Gene ontology biological processes and disease processes of lineage driver genes at E10.5. These are the complete lists of gene ontology biological processes and disease processes in each group of genes after dividing the ordered gene list in Supplementary data 5 into four groups of equal number of genes from Bmp4 to Gata6.
Supplementary data 7: List of differentially expressed genes and statistics for each NCC cluster of integrated data from control and Tbx1 null embryos at E9.5. This list is associated with the data shown in Fig. 5B and C.
Supplementary data 8: Gene ontology biological processes of upregulated genes in proximal PA2 (C8 in Figure 5) of control and Tbx1 null embryos at E9.5. This is a complete version of the data shown in Fig. 5F.
Supplementary data 9: Gene ontology biological processes of downregulated genes in proximal PA2 (C8 in Figure 5) of control and Tbx1 null embryos at E9.5. This is a complete version of the data shown in Fig. 5G.
Supplementary data 10: List of differentially expressed genes and statistics for each cell cluster of integrated data from control and Tbx1 null embryos at E10.5. This list is related to data shown in Fig. 7B and C.
Supplementary data 11: Gene ontology biological processes of downregulated genes in pharyngeal NCCs (C4 in Figure 7) from control and Tbx1 null embryos at E10.5. This is a complete version of the data shown in Fig. 7G.
Supplementary data 12: Gene ontology biological processes of upregulated genes in OFT-CNCCs (C10 in Figure 7) from control and Tbx1 null embryos at E10.5. This is a complete version of the data shown in Fig. 7H
Acknowledgements
We thank members of the Genomics core, Flow Cytometry and Analytical Imaging facilities at Albert Einstein College of Medicine. We are grateful to Professor Chenleng Cai for Tbx2f/+ and Tbx3f/+ mice. We thank Professor Robert G. Kelly for insightful comments on the manuscript. This work was supported by grants from the National Institutes of Health (P01HD070454, R01HL138470, R01HL153920, R01HL163667) and by grant from the Fondation Leducq (Transatlantic Network of Excellence 15CVD01). C.D thanks the Fondation Bettencourt-Schueller and the Philippe Foundation for their financial support.