Unveiling complexity and multipotentiality of early heart fields

Abstract

Rationale:

Extraembryonic tissues, including the yolk sac and placenta, and the heart within the embryo, work to provide crucial nutrients to the embryo. The association of congenital heart defects (CHDs) with extraembryonic tissue defects further supports the potential developmental relationship between the heart and extraembryonic tissues. Although the development of early cardiac lineages has been well-studied, the developmental relationship between cardiac lineages, including epicardium, and extraembryonic mesoderm remains to be defined.

Objective:

To explore the developmental relationships between cardiac and extraembryonic lineages.

Methods and Results:

Through high-resolution single cell and genetic lineage/clonal analyses, we show an unsuspected clonal relationship between extraembryonic mesoderm and cardiac lineages. Single-cell transcriptomics and trajectory analyses uncovered two mesodermal progenitor sources contributing to left ventricle cardiomyocytes, one embryonic and the other with an extraembryonic gene expression signature. Additional lineage-tracing studies revealed that the extraembryonic-related progenitors reside at the embryonic-extraembryonic interface in gastrulating embryos, and produce distinct cell types forming the pericardium, septum transversum, epicardium, dorsolateral regions of the left ventricle and atrioventricular canal myocardium, and extraembryonic mesoderm. Clonal analyses demonstrated that these progenitors are multipotent, giving rise to not only cardiomyocytes and serosal mesothelial cell types but also, remarkably, extraembryonic mesoderm.

Conclusions:

Overall, our results reveal the location of previously unknown multipotent cardiovascular progenitors at the embryonic-extraembryonic interface, and define the earliest embryonic origins of serosal mesothelial lineages, including the epicardium, which contributes fibroblasts and vascular support cells to the heart. The shared lineage relationship between embryonic cardiovascular lineages and extraembryonic mesoderm revealed by our studies underscores an underappreciated blurring of boundaries between embryonic and extraembryonic mesoderm. Our findings suggest unexpected underpinnings of the association between congenital heart disease and placental insufficiency anomalies, and the potential utility of extraembryonic cells for generating cardiovascular cell types for heart repair.

Introduction

A developmental field is a transient collection of cells with a shared potential to produce a restricted subset of embryonic structures, that is present only at specific developmental stages. Past studies of heart development have defined a first heart field (FHF) and second heart field (SHF), according to their potential to produce specific myocardial lineages within the developing heart.

The FHF and SHF were initially inferred by clonal analyses in the mouse embryo, which revealed two clonally distinct myocardial lineages, the first and second heart lineages^1–3. At E8.5, clones of the first heart lineage were observed to be excluded from the outflow tract and populate the left ventricle (LV) and left atrioventricular canal (AVC), whereas clones of the second heart lineage were found to be excluded from the LV¹. Of the heart fields predicted by this model, the SHF has been visualized and defined, as a population of cells medial to differentiating cardiomyocytes of the cardiac crescent that expresses the transcription factor Isl1 around E7.75⁴. SHF cells expressing Isl1 will also supply multiple other cell types that contribute to the heart, pharyngeal arches and head/neck including endothelial, endocardial and smooth/skeletal muscle cells^4–6. At E7.5, the first differentiating cells in the cardiac crescent are marked by Tbx5 and Hcn4. As Tbx5-CreERT2 and Hcn4-CreERT2 labeled cells in the crescent mainly contribute to cardiomyocyte lineages in the LV and parts of the atria, they are thought to represent more differentiated precursors of first heart lineage cardiomyocytes, and the FHF at crescent stages^{2, 7}. Since these original studies, lineage studies, clonal analyses and single cell studies have added to our knowledge of cardiac progenitors, and affirmed the presence of two heart fields^{2, 3, 8}. However, the location of distinct FHF and SHF progenitors in the early gastrulating embryo remains to be explored.

In addition to myocardial and endocardial lineages, the fully formed heart includes fibroblasts and vascular support cells derived from the epicardium. The proepicardium, a transient cluster of cells that forms at the base of the looping heart from the septum transversum (ST) during early heart development, produces cells which cover the heart to form the epicardium. Yet, the developmental origin of the proepicardium, and its relationship to previously described heart fields have been elusive.

To address the developmental origins, of cell lineages creating the heart, we performed single cell transcriptomic analyses on Mesp1-cre; Rosa26-tdTomato (Rosa26-tdT) mouse embryos across key developmental stages of cardiac development. Computational trajectory analyses of these data predicted a group of Hand1-expressing progenitors that may create a subset of first heart lineage cardiomyocytes. In situ and lineage tracing analyses utilizing Hand1-CreERT2 revealed a Hand1-expressing population at the extraembryonic/intraembryonic boundary of the gastrulating embryo that supplies first heart lineage cardiomyocytes residing largely within dorsolateral regions of the LV and AVC. Notably, the Hand1-CreERT2 lineage created a subset of (rather than all) LV cardiomyocytes. As the second heart lineage does not populate the LV¹, this finding implies an unexpected complexity of the FHF in which the FHF is not a single developmental heart field, but rather composed of at least two distinct developmental sources, one of which, identified here, is marked by Hand1. Utilizing Hand1-CreERT2 and Rosa26-Confetti clonal analyses, we furthermore discovered that this subset of FHF-derived cardiomyocytes derive from a common multipotent Hand1+ progenitor that also creates serosal mesothelial lineages (including proepicardium/epicardium and pericardium) and extra-embryonic mesoderm.

Methods

Detailed Methods are provided in Online Data Supplement. Please see the Major Resources Table in the Supplemental Materials. Raw data used in this study are available upon request. The raw scRNA-seq data can be found in GEO (GSE176306). Visualization of gene expression of the scRNA-seq is available on the UCSC cell browser at https://chi-10x-mouse-cardiomyocytes.cells.ucsc.edu.

Results

scRNA-seq analysis of Mesp1 lineage-traced cells reveals developmental cell types participating in mesoderm-related organogenesis.

To track cell lineages contributing to the development of mesoderm-derived organs including the heart, we employed a mouse Mesp1-Cre⁹; Rosa26-tdTomato (R26R-tdT)¹⁰ genetic fate mapping system (Figure 1A). Utilizing single-cell RNA-sequencing (scRNA-seq), we interrogated the transcriptomes of Mesp1-Cre; Rosa26-tdT genetically-labeled cells at E7.25 (no bud stage), E7.5 (early bud stage), E7.75 (late head fold stage) and E8.25 (somite stage) (Figure 1A, Online Figure I). Unsupervised k-means clustering¹¹ of these single cell data revealed a broad array of cell types, which were identified based on gene expression (Figure 1B–F, Online Figure I and II, Data Set I). As mesodermal progenitors differentiated into organ-specific cell types during mouse embryogenesis, we observed that the number of identified cell-types increased with developmental age. For example, nascent, early-extraembryonic and hemangiogenic mesoderm (NM, EEM, Hem) cell types were detected at E7.25 as described¹²; however, many more intermediate and differentiated organ-specific cell types were identified by E7.75 and E8.25, including two cardiomyocyte clusters – developing cardiomyocytes (DC) and cardiomyocytes (CM) (Figure 1B–E), which appear to represent developing and more established cardiomyocytes based on their differential expression of sarcomeric (Tnnt2, Ttn and Myl3) and cardiac progenitor genes (Tbx5, Sfrp1/5 and Meis1) (Figure 1F).

Figure 1. Mesp1-Cre single-cell maps reveal diverse cell types participating in early mouse mesoderm development.

(A), Mesp1-Cre scRNA-seq experimental design. Mesp1-Cre; Rosa26-tdT embryos were harvested for scRNA-seq at E7.25 (no bud stage); E7.5 (early bud stage); E7.75 (early head fold stage); and E8.25 (somite stage) as shown in representative bright-field and Mesp1-Cre; tdT+ (Mesp1 lineage) micrographs. Illustration below these micrographs shows tissues genetically labeled by Mesp1-Cre in embryos, and workflow for capturing these labeled single cells for RNA sequencing. Scale bars, 150 μm. (B), scRNA-seq data is displayed by tSNE plots at each developmental stage. Cells are colored according to their cell identities in D, E, F. (C, D), tSNE plot of scRNA-seq data across all examined stages displays individual cells (single dots) by (C) developmental stages or (D) cell types. (E), Dot plot shows distribution of each cell type across different embryonic stages. (F), Dot plot of key marker genes identifies each cell cluster. Al, Allantois; Bl, Blood; MM, Mixed Mesoderm; CrPh, Cranial-pharyngeal mesoderm; CM, Cardiomyocytes; DC, Developing Cardiomyocytes; En, Endothelium; Ep, Epithelium; EEM, Early Extraembryonic Mesoderm; Hem, Hemangiogenic mesoderm; HT, Heart tube; LEM, Late Extraembryonic Mesoderm; LPM, Lateral plate mesoderm; NM, Nascent Mesoderm; PSM, Pre-somitic mesoderm; PS, Primitive streak, SM, Somite mesoderm.

Trajectory analysis elucidates developmental pathways during mesoderm organogenesis.

To illuminate the developmental origins and cell fate decisions of organ-specific cell types arising from mesodermal progenitor cells, we organized cells from our single cell data along developmental trajectories using the lineage inference analysis, URD¹³. These reconstructed developmental trajectories not only ordered cells by a pseudotime which correlate with the developmental age of analyzed cells but also revealed new and known developmental cell fate decisions (Figure 2A, Online Figure IIIA). In particular, we observed developmental trajectories that identified previously described mechanisms of development for some cell types including endothelial, blood, muscle and cranial pharyngeal cells^{6, 14, 15} (Figure 2A). On the other hand, examination of the cardiomyocyte developmental trajectory uncovered two potential developmental sources that may contribute to developing cardiomyocytes: a known intraembryonic progenitor from the lateral plate mesoderm (LPM) and a previously undescribed cardiac progenitor from the late extraembryonic mesoderm (LEM) (Figure 2A, box). A three-dimensional force-directed URD representation revealed how these two progenitor sources originate and converge to independently contribute to developing cardiomyocytes (Figure 2B, Online Figure IIIB). Consistent with these findings, a URD lineage inference analysis of previously published mouse embryonic scRNA-seq data¹² showed that analogous LPM and LEM cells could be identified forming similar developmental trajectories contributing to developing cardiomyocytes (Online Figure IIIC–E). Thus, our bioinformatic analyses support that cells with an extraembryonic signature (LEM) may contribute to the heart that is separate from that of the embryonic (LPM) lineage.

Figure 2. Mesp1-Cre scRNA-seq trajectory analysis reconstructs developmental cell lineage trees during mesoderm/heart organogenesis.

(A, B), URD inferred lineage tree, as displayed by (A) dendogram or (B) force-directed layout, reveals the developmental history of Mesp1 mesoderm-derived organs. Red dashed box in A, B outlines cardiomyocyte branch, which is further magnified in B. The magnified cardiomyocyte branch shows that cardiomyocytes may derive from both late extraembryonic mesoderm (LEM) and lateral plate mesoderm (LPM) progenitor cells. (C), tSNE layout of cells from only the cardiomyocyte branch (boxed area in A, B) reveals seven cardiac subclusters composing the cardiomyocyte branch including three distinct cardiomyocyte populations (CM1–3) and four specific cardiac progenitor cell-types (CP4–7). (D), Heatmap of differentially expressed marker genes identifies each cardiac subcluster.

Multiple developmental pathways create distinct cardiomyocyte populations.

Previous studies have reported distinct populations of cardiomyocytes during heart development which arise from specific heart fields^1–3. Thus, we investigated whether these cardiomyocyte populations could be detected as subclusters within our identified developing cardiomyocyte (DC)/cardiomyocyte (CM) clusters (Figure 1D), and how LPM and LEM cells in our cardiomyocyte trajectories may contribute to these subclusters (Figure 2B - magnification). Subclustering analysis of cells comprising the initial DC/CM branches (Figure 2A, B –boxed area, magnification: LEM, LPM, DC and CM) uncovered seven distinct sub-populations (Figure 2C, D, Online Figure IVA, B, Data Set II). Three of these sub-clusters exhibited increased expression of cardiomyocyte sarcomeric genes including Ttn, Tnnt2, and Myl7 (Figure 2C, D, Cardiomyocyte/CM1–3, Online Figure IVA, C, E), and correlated with the CM cluster (Figure 2C - boxed area), whereas the other four sub-clusters displayed relatively low expression of these sarcomeric genes but high expression of cardiac progenitor (CP) markers including Isl1, Sfrp5, Tbx5 (Figure 2C, D, Cardiac Progenitor/CP4–7, Online Figure IVA, D, E), and associated closely with the DC cluster and specific portions of LPM and LEM clusters (Figure 2C - unboxed area). Differential gene marker analyses of the CM subclusters revealed that CM1, CM2 and CM3 cells displayed a combinatorial enrichment of Irx4/Tbx5, Tdgf1/Isl1 and Mab21l2/Tbx5, respectively, and that CM1 cells exhibited increased expression of mature cardiomyocyte gene markers including Actc1, Actn2, Myh6, Myh7 and Myl1 (Online Figure IVA, F). These findings indicate that CM1 and CM2 subclusters may represent cardiomyocytes arising from the FHF and SHF^{2–4, 7, 16, 17}, whereas the developmental source of the CM3 subcluster remains to be identified. Additional gene marker analyses of CP subclusters support that cell types from these subclusters represent cardiac progenitors for cardiomyocytes and potentially other differentiated cardiac cell types (Figure 2D, Online Figure IVA, D, E). For instance, CP6 and CP7 expressed genes overlapping with those in CM3 and proepicardial cells (Upk3b, Ccbe1, Sfrp5, Mab21l2, Tbx18^18–23) (Online Figure IVA), suggesting that CP6/CP7 subclusters contain progenitors for CM3 and proepicardial cells. Additionally, CP5 and a small region of CP4 are enriched for genes expressed within the anterior and posterior SHF including Isl1, Fgf10, Hoxb1, Hoxa1 and Osr1 (Online Figure IVA, D, E, V).

To confirm the identity of subcluster cell types and investigate their relationship during embryogenesis, we spatially mapped these cell types in E8.25 embryos using RNAscope in situ hybridization (ISH) analysis of markers that are specific to these subclusters (Online Figure VI). Irx4, Tdgf1 and Mab21l2, markers of CM1, CM2 and CM3 subclusters, were expressed in distinct regions of the heart tube as labeled by Myl7 and Nkx2–5: the middle segment (Primitive left ventricle/LV), arterial pole (Primitive outflow tract/OFT and right ventricle/RV), and venous pole of the heart tube (Online Figure VIA, B), indicating that CM1 and CM2 cells correspond to cardiomyocytes derived from the FHF and SHF ^{2–4, 7, 16, 17}, whereas the source of progenitors giving rise to CM3 cardiomyocytes remains to be determined. Using differentially expressed genes in the CP subclusters, we investigated the location of CP subcluster cell types during embryogenesis (Figure 2D, Online Figure IVA, D, E and VIC–K), and discovered that CP4 markers Sfrp5, Nr2f2 and Arg1 were co-expressed in regions posterior to the venous pole and contiguous with CM1 (Online Figure VIC, E, J).

The CP5 marker Isl1 was enriched in regions anterior and dorsal to the arterial pole and contiguous with CM2 (Online Figure VID, F). The combined CP6 markers Smoc2 and Mab21l2 were expressed at the interface between the forming heart and extraembryonic tissues, near the ventral venous pole and contiguous with CM3 (Online Figure VIH), and the CP7 markers Twist1, Sfrp5 and Mab21l2 were located in regions connected to the ventral side of the venous pole and contiguous with CP6 (Online Figure VIE, G, I). The adjacent locations of CP6, CP7, and CM3 and overlapping genes between them (Cpa2, Mab21l2, Bmp4, Hand1) (Online Figure IVA, D, VI), suggest that these cell types may be developmentally related.

We further investigated the developmental relationship of the CM1–3 subpopulations and how LPM and LEM progenitors contribute to them. Using the three CM subcluster populations CM1–3 (Figure 2C) as end points for URD trajectory analysis¹³, we reconstructed our developmental trajectories (Figure 3A, B, Online Figure VII), which created three new cardiomyocyte trajectory branches (Figure 3A, B, box). The CM1 and CM2 trajectory branches, whose cells expressed genes associated with FHF (Tbx5) and SHF (Isl1/Tdgf1), respectively (Figure 2D, Online Figure IVA, C, E), shared a common intraembryonic cellular origin associated with LPM and NM cells, whereas the CM3 lineage branch was distinct from the CM1 and CM2 branches and shared an origin with early and late extraembryonic mesoderm (EEM and LEM) cells (Figure 3A, B, Online Figure VII). The CM2 and cranial-pharyngeal (CrPh) branches expressed the SHF marker, Isl1, and appeared along a developmentally related trajectory consistent with previous studies of SHF development (Figure 3A–C) ^4–6.

Figure 3. Distinct cardiomyocyte lineages derive from intra- and extra-embryonic related developmental origins.

(A, B), Reconstructed URD developmental cell lineage trees using the three distinct subclustered cardiomyocyte populations predict that CM1/CM2 and CM3 cardiomyocytes derive respectively from intra- and extra-embryonic related progenitor sources, as displayed by (A) cell type and (B) developmental stages. The cardiomyocyte-related branches of the URD developmental tree are outlined with box. (C), Marker genes differentially expressed among the lineages for each cardiomyocyte subcluster are plotted on the URD cardiomyocyte-related branches. Hand1 and Mab21l2 mark early and late regions of the CM3 lineage, respectively. Mesp1, Tbx5, Isl1, Irx4 and Tdgf1 label different regions of the CM1 and CM2 lineage branches. (D, E), RNAscope in situ hybridization (ISH) of Mesp1 and Hand1 was performed in (D) E7.25 and (E) E7.5 Mesp1-Cre; Rosa26-tdT embryos. The diagram illustrates both the gene expression pattern of Hand1 and Mesp1 and Mesp1-Cre lineage-traced cells in these embryos. (F, G), RNAscope ISH of Hand1, Tbx5, and Isl1 was performed in E7.75 embryos. The diagram illustrates the expression pattern of Hand1, Tbx5, and Isl1 in these embryos. n = 3 per panel. Scale bars, 100 μm. EXE, Extraembryonic Ectoderm; YS, Yolk Sac.

Interrogating transcriptional profiles of CM1–3 lineage branches uncovers distinct cell fate programs for each cardiomyocyte population.

To identify gene programs that regulate the cell fate decisions creating these distinct cardiomyocyte lineages, we interrogated transcriptional profiles of cells along each of the cardiomyocyte developmental trajectories. We analyzed differentially expressed genes, including transcription factors, at each branch point, and found that Hand1 appeared important for the branch point decision between embryonic NM and EEM (branch point 1) (Figure 3C, Online Figure VIIIA–C), which coincides with Hand1’s role in extraembryonic mesoderm development^{24, 25}. Supporting previous cardiac developmental studies^{2, 4, 26, 27}, the transcription factors Tbx5, Isl1, Hand2 and Tbx1 exhibited differential expression in FHF-related CM1, SHF-related CM2 and SHF-related CrPh cell-types at branch point 3 (Figure 3C, Online Figure VIIIA, B, E, Supplemental Data Set III). In the extraembryonic branch, Cdx2/Cdx4 and Tsc22d1 were expressed at branch point 2 where allantois (Al) and LEM cells arise from EEM cells (Online Figure VIIIA, B, D, Supplemental Data Set III). Mef2c, Id2 and Cited2 in CM3 cells^28–30, and Hoxb6 and Hand1 in LEM cells, were predicted to regulate cell fate decisions at branch point 4 (Online Figure VIIIA, B, F, Supplemental Data Set III).

To illuminate the dynamics of cell fate choices in these cardiomyocyte lineages, we examined genes differentially expressed in each cardiomyocyte lineage trajectory along a pseudotime, which revealed three major differentiation states for each trajectory: an early, intermediate and late state (Online Figure IX). Consistent with our branch point analyses (Online Figure VIII), CM1 and CM2 early state genes were similar to each other but distinct from CM3 early state genes; however, genes across these pseudotime analyses converged as each intermediate state cardiac progenitor differentiated into its corresponding late state cardiomyocyte population (Online Figure IX). Specifically, Mesp1 was expressed in the CM1 and CM2 early states but Tbx5 and Isl1 were activated in these lineages at intermediate states (Figure 3C, Online Figure IXA, B, D, E, G, H), suggesting that CM1 and CM2 may derive from a common developmental trajectory but Tbx5 and Isl1 may direct their specification in more distinct cardiomyocyte populations. In contrast, Hand1 and BMP signaling-related genes Bmp4 and Msx2 were expressed in CM3 early states, and Mab21l2 and Cpa2 were activated in CM3 intermediate states (Figure 3C, Online Figure IXC, F, I). Finally, Mef2c and sarcomeric genes were expressed at CM1–3 late stages; however, some genes appeared specific for each CM population including Irx4 (CM1) and Tdgf1 (CM2) (Online Figure IXD–I). Confirming these analyses, RNAscope ISH revealed that Hand1 was expressed at the embryonic and extraembryonic boundary, whereas Mesp1 was expressed in the intraembryonic migrating mesoderm at E7.25 and E7.5 (Figure 3D, E). Furthermore, Hand1, Tbx5 and Isl1 marked different locations in the crescent region at E7.75 where Hand1 labeled regions anterolateral to the cardiac crescent, which was marked by Tbx5², while Isl1 labeled regions posteromedial to the cardiac crescent (Figure 3F, G, Online Figure X). These bioinformatic and spatial gene expression analyses reveal a potentially unexplored developmental source of cardiomyocytes along the proximal extraembryonic-embryonic boundary that appears distinct from FHF and SHF progenitors.

Hand1 lineage tracing reveals an unexpected heart field that contributes to specific subsets of the first heart lineage and serosal mesothelial lineages.

To examine this predicted extraembryonic-related heart field, we employed an inducible Cre-recombinase fate mapping strategy to lineage trace cells from this potential heart field. Because Hand1 was expressed in early extraembryonic-related CM3 progenitors but not CM1 and CM2 progenitors, we examined whether Hand1 could be used to genetically label progenitors from the CM3 heart field (Figure 3C). To explore this possibility, we performed RNAscope ISH analyses to examine Hand1 expression in developing embryos. At E6.75, Hand1+ cells co-expressed Mesp1 at the extraembryonic/embryonic boundary (Online Figure XIA) but downregulated Mesp1 after E6.75 (Figure 3D, E). Hand1 was expressed in the extraembryonic mesoderm at E7.75, E8.25 and pericardium at E8.25 but not in Hcn4+ or Myl7+ cardiomyocytes (Online Figure XIB–E). However, at E8.5 and E9.0, Hand1 was expressed in a portion of cardiomyocytes in the LV and AVC as well as the pericardium and septum transversum (ST) (Online Figure XIF, G).

Based on these Hand1 expression studies, which reveals a developmental time window (E6.25 - E8.25) in which the contributions of early gastrulating Hand1+ CM3 progenitors to the heart can be investigated, we generated a Hand1-CreERT2 mouse that was confirmed by RNAscope ISH of CreERT2 and analysis of Hand1-CreERT2; Rosa26-tdT embryos with and without tamoxifen induction (Figure 4A, B, Online Figure XII). Because the half-life of tamoxifen in mice is ~12 hours and persists over a ~24–36 hour time period³¹ (Online Figure XIII), we studied Hand1-CreERT2; Rosa26-tdT embryos from pregnant mice given tamoxifen at E5.75 (Figure 4B) to avoid CreERT2 activation in differentiated cardiomyocytes expressing Hand1 at E8.5. Consistent with our CM trajectory branches (Figure 3C), examining these embryos at E7.75, E8.25, E8.5, E9.5 and E12.5 revealed that Hand1 lineage-traced cells contributed to the heart and extraembryonic tissues (Figure 4C–F, Online Figure XIV). Within extraembryonic tissues including the yolk sac (Figure 4C–E, Online Figure XIVB, E, F), Hand1 lineage-traced tdT+ cells produced Pecam+ endothelial cells, α-SMA+ smooth muscle cells and Pdgfrβ+ mesothelial cells (Online Figure XIVE, F). In the developing embryo, Hand1 lineage-traced tdT+ cells contributed in a more spatiotemporal restricted manner (Figure 4C–F). Specifically, Hand1 lineage-traced tdT+ cells supplied Hcn4+ cardiomyocytes in the cardiac crescent at E7.75 and then cardiomyocytes (Myl7+) on the ventral side of the venous pole and medial regions of the heart tube at E8.25 (Figure 4C, D). Between E8.5 - E12.5, tdT+ cardiomyocytes were restricted spatially to the primitive AVC region and LV at E8.5 and then further to the AVC/sinus venosus (SV), dorsolateral LV and atrial regions of E12.5 hearts (Figure 4E, F, Online Figure XIVB, D). Furthermore, tdT+ cells appeared in non-myocardial heart tissue including the pericardium, proepicardium/ST, epicardium, and occasionally endocardium from E8.25 - E12.5 (Figure 4D–F, Online Figure XIVA–D). Supporting these lineage studies, the CM3 URD trajectory, including CP6 and CP7 subclusters, comprised cells that express AVC (Msx1/2, Twist1, Tbx2^{23, 32}), and proepicardial/pericardial markers (Upk3b, Ccbe1, Sfrp5, Mab21l2, Tbx18¹⁸⁻²³) (Online Figure IVA), while the lateral location of E8.25 tdT+ cells coincides with Mab21l2 expression (Online Figure VIH) and the previously reported location of Tbx18+ venous pole progenitors³³. To confirm these findings, we examined Hand1-CreERT2; Rosa26-tdT embryos from pregnant mice given tamoxifen at E6.25 (Online Figure XV). In addition to recapitulating results in E5.75 tamoxifen-induced Hand1-CreERT2; Rosa26-tdT embryos (Online Figure XVA–F compared to Figure 4C–F), Hand1 tdT+ cells genetically-labeled at E6.25 also contributed cardiomyocytes and epicardial-derived cell-types including fibroblasts and vascular support cells to E17.5 embryos (Online Figure XVG). Altogether, these data suggest that during early gastrulation, Hand1 marks a progenitor population that produces not only cardiomyocytes within the AVC and LV prior to when Hand1 is expressed in differentiated CMs, but also pericardial, epicardial and extraembryonic-derived mesoderm cell types. As these Hand1+ cardiomyocyte progenitors contribute to cardiomyocytes within the developing AVC and dorsolateral LV, they likely represent a distinct subset of the reported first heart lineage cardiomyocytes¹, suggesting that FHF is not a single heart field, but is rather composed of at least two distinct sources, including a Hand1-derived source.

Figure 4. Lineage tracing studies reveal that early gastrulating Hand1+ cells contribute to not only a distinct subpopulation of first heart lineage cardiomyocytes but also serosal mesothelial lineages (pericardial, epicardial cells) in the heart.

(A) RNAscope ISH in Hand1-CreERT2 embryos shows that expression of CreERT2 precisely recapitulates the expression of Hand1. (B), Lineage tracing studies using Hand1-CreERT2 and Rosa26-tdT mice (shown in left) map the fate of early gastrulating Hand1+ cells. Schematic outlines the experimental strategy for Hand1-CreERT2 genetic fate mapping studies shown in right. Tamoxifen was given at E5.75, and embryos were examined for Hand1-CreERT2 genetically-labeled tdT+ cells at E7.75, E8.25, E9.5 and E12.5. (C-F), RNAscope ISH and immunohistochemistry of whole mount and cross sections of these embryos reveal the contribution of Hand1-CreERT2 genetically-labeled tdT+ cells at (C) E7.75, (D) E8.25, (E) E9.5 and (F) E12.5. (C’, D’, E’, E’’, E’’’), Insets show transverse sections of C, D, E at indicated dashed lines, respectively. (F’), Inset shows coronal section of F. (C’’, F’’), Insets are magnification of C’, F’ boxed area. Arrowheads point to tdT+ cells expressing (C, C’’) Hcn4, (D, D’) Myl7, (E’’) Erg1, (E’’’) Wt1 and (F’’) α-Actinin. (C’’’, D’’’, E’’’’, F’’’), Diagrams summarize the anatomical location of Hand1-CreERT2 genetically-labeled tdT+ cells at the embryonic stages analyzed. n = 3 embryos for each stage. Scale bars, 100 μm. AM, Amnion; AVC, Atrioventricular Canal; BA, Base of the Atrium; CC, Cardiac Crescent; Epi, Epicardium; HT, Heart tube; LA, Left Atrium; LV, Left Ventricle; OFT, Outflow Tract; Peri, Pericardium; Pro, Proepicardium; RA, Right Atrium; RV, Right Ventricle; SV, Sinus Venosus; ST, Septum transversum; YS, Yolk Sac.

Genetic clonal analysis reveals the multipotentiality of Hand1+ cardiac precursor cells.

To examine the lineage potential of single Hand1-expressing cells during early gastrulation, we crossed Hand1-CreERT2 mice with Rosa26-Confetti multicolor reporter mice³⁴ to genetically fate map early Hand1+ individual clones expressing a specific fluorescent protein (Figure 5A). Tamoxifen doses were initially titrated to ensure low-levels of Hand1+ labeling (Figure 5B). The observed frequency of each confetti color upon addition of tamoxifen was consistent with those previously reported³ (Online Figure XVIB). At the lowest dose of tamoxifen (0.0025 mg/g) given at E6.75, only 8.6% of exposed embryos (n = 73/849) were labeled, which was less than the expected Hand1-CreERT2; Rosa26-confetti genotype positivity rate (50%) (Figure 5B). At this dose, uni-color embryos occurred in the highest proportion followed by bi-color and tri-color embryos (Online Figure XVID). To reduce the probability of observing a multi-recombinant clone, only the uni-color embryos were further analyzed. At E9.5, the majority of the clones (n = 23) from the uni-color embryos were present in only the extraembryonic tissue. However, twelve clones did fluorescently-label both extraembryonic and cardiac tissues including the proepicardium/ST, pericardium, LV and AVC (Figure 5C, D), which was consistent with the distribution of genetically-labeled Hand1-CreERT2; Rosa26-tdT cells at a similar stage, thus supporting the multi-potentiality of these clones (Figure 4E, compared to Figure 5C, D). To confirm this observation on a larger sample size, clones induced with a higher dose of tamoxifen (0.005mg/g) given at E6.25 or E7.25 were analyzed (Online Figure XVIA, E). We used a rigorous statistical analysis to reduce the chance of analyzing clones generated from multi-recombinant events at this higher dose. The number of cells in each clone (an individual color) which contributes to cardiac tissues or to both extraembryonic and cardiac tissues was used to create a model with a mixture of two Gaussian distributions: one for cell counts that would be expected for a single recombination event, and the other for cell counts that would be expected for two or more recombination events (Online Figure XVIP, R). Based on this model, we excluded 35 (out of 88) clones, which likely derive from > 1 progenitor (Online Figure XVIQ, S). The majority of these remaining 53 clones analyzed contributed to two or three distinct lineages, including combinations of extraembryonic mesoderm, pericardium, proepicardium/ST, and AVC or LV myocardium (Online Figure XVIQ, S), thus supporting the multipotentiality of Hand1+ progenitors. Immunofluorescence staining show that these Hand1+ clones contributed to not only α-Actinin+ cardiomyocytes in the AVC and LV but also Wt1+ proepicardial and pericardial cells; however, Hand1+ clones were not observed in the endocardium (Online Figure XVIH–K). Finally, Hand1+ clones supplied α-SMA+ smooth muscle cells and Pdgfrβ+ mesothelial cells to the extraembryonic mesoderm (Online Figure XVIL, M).

Figure 5. Clonal analysis reveals multipotentiality in early Hand1+ progenitors.

(A), Schematic outlines experimental strategy for Hand1-CreERT2; Rosa26-Confetti clonal analyses. (B), Bar graph reveals the percentage of E9.5 embryos that displayed fluorescence at titrated doses of tamoxifen. The numerator is the number of fluorescence-positive embryos, and the denominator is the number of total embryos examined. (C), Clonal analyses of uni-color E9.5 embryos reveal that individual Hand1-CreERT2; Rosa26-Confetti clones labeled at E6.75 (0.0025 mg/g Tamoxifen) have the capacity to generate multiple cell types that can contribute to the yolk sac and/or heart. (D), Diagram summarizes the contribution of Hand1-CreERT2; Rosa26-Confetti genetically-labeled clones in the heart and yolk sac at E9.5. (E), Clonal analyses of uni-color E12.5 hearts reveal that individual Hand1-CreERT2; Rosa26-Confetti clones labeled at E6.75 or E7.25 (0.005 mg/g Tamoxifen) have the capacity to generate multiple cell types within the heart. (F), Diagram summarizes the contribution of Hand1-CreERT2; Rosa26-Confetti genetically-labeled clones in the heart at E12.5. (G), Representative E12.5 uni-color hearts show individual Hand1-CreERT2; Rosa26-Confetti genetically-labeled clones contributing to different combinations of tissues and cell types: AVC/SV and RA (clone # 237-RFP); LV and Epi (clone # 329-RFP); the AVC/SV and Epi (clone # 319-YFP). Scale bars, 200 μm. (H), Bar graph displays the number of uni-color E12.5 hearts with clones contributing to cardiomyocytes only, epicardial cells and cardiomyocytes, or only epicardial cells. (I), Model summarizes the multipotentiality of Hand1+ cardiac progenitor cells (CPC) between E6.75 - E7.25 in relation to the contribution of reported FHF/SHF progenitors. AVC, Atrioventricular Canal; BA, Base of the Atrium; CM, cardiomyocytes; Epi, Epicardium; LA, Left Atrium; LV, Left Ventricle; OFT, Outflow Tract; Peri, Pericardium; Pro, Proepicardium; RA, Right Atrium; RV, Right Ventricle; SV, Sinus Venosus; ST, Septum Transversum; YS, Yolk Sac.

To investigate the clonal relationships among Hand1+ progenitor-derived cardiac cell types and their location in later stage hearts, we also examined E6.75 or E7.25 tamoxifen-induced Hand1-CreERT2; Rosa26-Confetti clones at E12.5 when most cardiac structures and cell types have been determined (Online Figure XVIF). In uni-color embryos, we identified clones marking the epicardium and cardiomyocytes in the AVC or LV (Figure 5E–H, Online Figure XVIN, O), supporting that multipotential Hand1+ cardiac progenitor cells can give rise to both cardiomyocytes and non-cardiomyocytes cardiac cells. Altogether, these results reveal multipotential Hand1 cardiac progenitors in the early ingressing mesoderm that can create extraembryonic mesoderm, mesothelial cardiac lineages (epicardium and pericardium) and LV and AVC myocardium (Figure 5I).

Discussion

Our single-cell transcriptomic, genetic fate-mapping and clonal analyses, reveal not only a developmental cardiac-extraembryonic tissue connection but also delineate the progenitors creating these lineages and their specific contributions to the developing heart and extraembryonic structures (Figure 5I). Highlighting the complexity of organogenesis, we show how similar cell types, such as cardiomyocytes, can derive from multiple developmental origins/progenitors that have potential to produce not only other cell types but also multiple organs/tissue structures. Single-cell subcluster analyses of isolated cardiomyocytes identified at least three distinct myocardial lineages including a heart lineage whose progenitor shares a gene signature with extraembryonic mesodermal progenitors including Hand1. Trajectory analyses predicted that two of these heart lineages derive from a common embryonic source prior to E7.25, with marker expression at E8.25 suggesting their correspondence to first and second heart lineages, whereas the Hand1+ extraembryonic-related heart lineage originates from a distinct developmental source. Expression analyses revealed that at early gastrula stages, Hand1+ progenitors reside at the intra-/extra-embryonic boundary, with genetic fate mapping demonstrating that Hand1+ progenitors create cardiomyocytes localized to the dorsal regions of the LV and AVC at E12.5. As LV cardiomyocytes have been shown to derive from the FHF¹, our results support that this Hand1+ cardiac progenitor field represents a distinct subset of the FHF, suggesting that the FHF consists of at least two distinct progenitor sources. This finding is supported by the complementary nature of the Hand1-lineage to previous studies examining the contribution of the lineages labeled by Tbx5-CreERT2, Mef2c-AHF-Cre, SMA-CreERT2 ^{2, 35, 36}. Notably, these Hand1+ FHF subpopulation findings are also consistent with a mathematically-inferred retrospective myocardial clonal analyses¹ that predicted two separate FHF lineages.

One limitation for understanding the lineage potential of the FHF or SHF from previous retrospective clonal studies¹, is that only myocardial clones were studied. However, when Isl1 was identified as a SHF marker, studies with Isl1-Cre or inducible Isl1-CreERT2 revealed that SHF produces both myocardial lineages as well as multiple other cardiac lineages^{4, 37}. Utilizing Hand1-CreERT2 with a confetti clonal indicator³⁴, we uncovered an unsuspected multipotentiality of Hand1 mesodermal cells, which produce not only a specific subset of FHF-derived LV cardiomyocytes, but also extraembryonic mesoderm, septum transversum/epicardial, and pericardial cells. Although previous findings suggest that Mesp1+ clones may contribute to LV and epicardial cell types³, our results reveal a closer lineage relationship between cardiomyocytes of the AVC and dorsolateral LV, and extraembryonic mesodermal and serosal mesothelial cells than previously expected, and furthermore expand upon previous studies³⁸ investigating the elusive origins of the proepicardium/epicardium.

The existence of a Hand11+ progenitor that supplies cells to extraembryonic and intraembryonic tissues provides further examples of blurred boundaries between extra/intraembryonic tissues³⁹. These findings may also account for previous observations that epicardial progenitors can adopt cardiomyocyte fates⁴⁰, and that loss of Scl can result in transdifferentiation of yolk sac hematopoietic cells to beating cardiomyocytes⁴¹. However, whether extraembryonic mesoderm cells can differentiate into cardiomyocytes in vivo as suggested by our scRNA-seq studies remains to be further examined, but the close developmental relationship between mesothelial lineages of both extraembryonic and intraembryonic tissues, and Hand1+ cardiac lineages, coupled to the high plasticity of mesothelial cells⁴², suggests the possibility of transforming extraembryonic and serosal mesothelial tissues into cardiomyocytes to treat heart failure.

The role of Hand1 in early specification of LV cardiomyocytes and FHF progenitors will be of interest to examine in future studies. Global Hand1 knockout embryos exhibit placental, yolk sac and heart defects^{24, 25} resulting in embryonic lethality at E8.5. As placenta and yolk sac defects can secondarily impact the heart, direct requirements for Hand1 in early heart progenitors remains unclear. Although experiments, including cardiac-specific conditional knockout and tetraploid rescue studies^{24, 25}, confirmed heart defects in Hand1 mutant embryos, these studies could not rule out requirements for Hand1 in differentiated cardiomyocytes, rather than undifferentiated progenitors. Our findings suggest that these heart defects may be due to abnormalities in undifferentiated progenitors which can produce cardiomyocytes. Thus, future studies utilizing scRNAseq of Hand1 knock-out embryos will be important to investigate whether Hand1 is important for the early stages of CM3 differentiation and the branch-point decisions between embryonic and extra-embryonic mesoderm. Additionally, due to the limited number of cells examined at the earliest stages in our study, interrogating these early decisions will also require more robust datasets.

Overall, our studies reveal a multipotential Hand1+ progenitor that produces not only cardiovascular lineages contributing to distinct regions of the left ventricle and pericardium but also extraembryonic cell types within the yolk sac. Consistent with our findings, Tyser et al. also recently reported Mab21l2 expressing progenitors present at cardiac crescent stages that contain precursors of heart and epicardium⁴³. Based on comparison between our studies (Online Figure XVII), their population likely represents a similar but later subset of the Hand1 progenitors that we describe at the embryonic/extraembryonic boundary at E6.25-E7.5. Although clonal relationships were not performed by Tyser et al., our results suggest that these progenitors are likely multipotent and contribute to distinct regions of the heart. Because these Hand1+ progenitors can also contribute to extraembryonic tissue, our studies also raise the possibility that congenital heart diseases thought to be caused by placental anomalies⁴⁴ may be due to perturbations of complex interplays between genetic pathways shared between extraembryonic and cardiac lineages.

Novelty and Significance.

What Is Known?

Two developmental heart fields – the first heart field (FHF) and the second heart field (SHF), have been defined as having the potential to produce specific myocardial lineages within the developing heart; however, whether all myocardial progenitors within these lineages differentiate in a similar manner remains unknown.

The SHF has been shown to supply other cardiac cell types to the heart, pharyngeal arches and head/neck including endothelial, endocardial, and smooth/skeletal muscle cells, but whether FHF progenitors are similarly multipotent is unknown.

The proepicardium produces epicardial cells which cover the heart but the origins of the cells remain to be defined.

What New Information Does This Article Contribute?

Hand1-expressing progenitors at the extraembryonic/intraembryonic boundary of the gastrulating embryo create a subset of FHF cardiomyocytes that contribute to the dorsolateral regions of the left ventricle and atrioventricular canal.

These Hand1-expressing progenitors are multipotent and give rise to not only cardiomyocytes but also serosal mesothelial lineages (including proepicardium/epicardium and pericardium) and extra-embryonic mesoderm.

Mammalian organs, including the heart have evolved a broad spectrum of specialized cell types that organize into intricate structures critical for their function. How early multipotent germ layer progenitors diversify to create these cell lineages, especially those deriving from mesoderm such as the heart remains to be fully illuminated. In the following study exploring the developmental cellular constituents during the ontogeny of mammalian mesoderm, we reveal an unexpected complexity of the contribution and multi-potentiality of mesodermal progenitors to cardiac lineages creating specific cell types such as myocardial, proepicardial/epicardial and pericardial cells in distinct regions of the heart.

Non-standard Abbreviations and Acronyms

Al

Allantois

AM

Amnion

AVC

Atrioventricular Canal

Bl

Blood

BA

Base of the Atrium

CC

Cardiac Crescent

CHDs

Congenital Heart Defects

CPC

Cardiac Progenitor Cells

CM

Cardiomyocytes

MM

Mixed Mesoderm

CrPh

Cranial-pharyngeal Mesoderm

DC

Developing Cardiomyocytes

Epi

Epicardium

EXE

Extraembryonic Ectoderm

En

Endothelium

Ep

Epithelium

EEM

Early Extraembryonic Mesoderm

FHF

First Heart Field

Hem

Hemangiogenic mesoderm

HT

Heart Tube

LA

Left Atrium

LEM

Late Extraembryonic Mesoderm

LPM

Lateral Plate Mesoderm

LV

Left Ventricle

NM

Nascent Mesoderm

OFT

Outflow Tract

Peri

Pericardium

Pro

Proepicardium

PSM

Pre-somitic Mesoderm

PS

Primitive Streak

Pro

Proepicardium

RA

Right Atrium

RV

Right Ventricle

SHF

Second Heart Field

SM

Somite Mesoderm

SV

Sinus Venosus

ST

Septum Transversum

YS

Yolk Sac

t-SNE

t-distributed Stochastic Neighbor Embedding

scRNA-seq

single-cell RNA-sequencing