Diverse contribution of amniogenic somatopleural cells to cardiovascular development: With special reference to thyroid vasculature

The somatopleure serves as the primordium of the amnion, an extraembryonic membrane surrounding the embryo. Recently, we have reported that amniogenic somatopleural cells (ASCs) not only form the amnion but also migrate into the embryo and differentiate into cardiomyocytes and vascular endothelial cells. However, detailed differentiation processes and final distributions of these intra‐embryonic ASCs (hereafter referred to as iASCs) remain largely unknown.


Key findings
Here, we show that amniogenic somatopleure cells (ASCs) differentiate into various cell types constituting the cardiovascular system, with some populations having a molecular background similar to that of hemangioblasts.Among them, those contributing to the thyroid vascular network were suggested to differentiate into vascular endothelial cells with fibroblast growth factor (FGF)-specification and vascular endothelial growth factor (VEGF)induced maturation.This study will provide a novel clue for understanding the cardiovascular development of amniotes from embryological and evolutionary perspectives.

| INTRODUCTION
We mammals, avians, and reptiles are classified as amniotes because of having the "amnion," a unique extraembryonic membrane that protects the developing embryo from the harsh land environment. 1,2The amnion provides embryos with an aquatic environment to adapt to terrestrial habitats.Therefore, the acquisition of an amniotic membrane has been considered to be one of the most important events of evolution. 1,32][13][14] However, the normal fate of these cells in the process of amnion development remains unclear.
The amniotic membrane consists of two layers of the ectoderm and the mesoderm origin, although its forming process differs depending on the species. 4,157][18] The coelomic cavity separates the lateral mesoderm into two components, somatic and splanchnic mesoderm.The somatic mesoderm forms the somatopleure with the ectoderm, whereas the splanchnic mesoderm forms the splanchnopleure with the endoderm.7][18] After the headfold stage, the subduction of the embryo causes the extraembryonic part of the somatopleure to cover from above, and finally, the membrane closest to the embryo becomes the amnion and the outside becomes the chorion. 3However, the behavior of this amniogenic somatopleural cell (ASC) population was unclear because the boundaries between the inside and outside of the embryo at this time were anatomically ambiguous.Therefore, it was not clear whether the ASCs constitute only the extraembryonic membranes (amnion and chorion) or are also involved in the embryonic composition.
In our previous study using chick and quail embryos, we found that ASCs close to the embryonic body not only contribute to the amnion but also migrate into the embryo and differentiate into endothelial cells and cardiomyocytes in the cardiac outflow tract. 16fibroblast growth factor (FGF) and BMP signals were suggested to be involved in the differentiation of ASCs into endothelial cells and cardiomyocytes, respectively. 16However, their intraembryonic distribution and fates have not been fully explored.It remains also unclear which cell population within ASCs is responsive to these signals and what kind of developmental processes take place under different cellular environments.
To address these issues, we first performed quailchick chimera experiments to trace ASCs entering the embryo (hereafter referred to as "intra-embryonic ASCs; iASCs") until late embryonic stages.Intra-embryonic ASCs were distributed to the cardiac outflow tract, where they differentiate into various cell types including cardiomyocytes, vascular endothelial and smooth muscle cells, and interstitial cells.They also distributed to the pharyngeal region, where they contribute to vasculature in the thyroid gland and infrahyoid muscles.Explant culture experiments indicated that an FGF-vascular endothelial growth factor (VEGF) signaling relay may drive the differentiation of iASCs into vascular endothelial cells.Single-cell (sc) RNA-seq analysis further revealed heterogeneity and the presence of a hemangioblast-like cell population within ASCs, with transcriptomic profiles consistent with a switch from FGF to VEGF signaling.Taken together with previous findings, the present study provides new perspectives on cardiopharyngeal development by revealing a novel cardiovascular cell origin and the embryological and evolutionary implications of amniogenesis.

| Intra-embryonic ASCs are widely destined to cervical and thoracic tissues and organs
Here, we performed quail-chick chimera experiments to analyze in detail the distribution of iASCs to embryonic tissues and organs at the late stages, when the major organ primordia are almost complete and start to mature.The quail AS grafts at Hamburger and Hamilton stage (HH) 19 10 wild-type (WT) or mCherry transgenic embryo (referred to as chFP) 20 were orthotopically transplanted into WT chick AS regions at HH10 to generate quailchick chimeras (AS chimeras).After 8-day incubation, quail-derived iASCs were detected by quail cell marker antibody (QCPN) 21 or mCherry immunohistochemistry.We analyzed 22 chimeras at HH34-35 (embryonic days 8-9; E8-9), and 5 of them were confirmed intraembryonic contributions of ASCs.
Consistent with the previous study, 16 iASCs were observed in the outflow tract of the heart.In the aortic valve, QCPN-positive iASCs were distributed in the valvar interstitium, rather than in the innermost endothelial cell layer as shown in Figure 1B.Near the aortic valve, iASCs were also distributed in the myocardial sheath and the interventricular septum (Figure 1C,D), where some iASCs were co-stained with MF20, a cardiomyocyte marker but others were not (Figure 1E-H).Interestingly, iASC distribution was largely restricted to the right ventricle.Furthermore, some iASCs existed around the coronary septal branch and co-stained with a monoclonal antibody reactive with smooth muscle α-actin (α-SMA), indicating that they differentiated into coronary artery smooth muscle cells (Figure 1I-L).Thus, iASCs differentiated into not only vascular endothelial cells and cardiomyocytes but also more diverse cell types including vascular smooth muscle cells.
Previously, we have reported that some iASCs are localized to the pharyngeal region, where they are incorporated into vessel walls. 16To further examine the destination and fates of iASCs in the pharyngeal region, we explored the pharyngeal-arch derived cervical region of the AS chimeras at late developmental stages.As a result, we observed that iASCs were not evenly distributed within this region, but were rather restricted to specific tissues and organs.In particular, iASCs were concentrated in the infrahyoid muscles (Figure 1M,N) and the thyroid gland (Figure 1S,T).The infrahyoid muscles are external laryngeal and tracheal muscles, 22 which are particularly well developed in singing birds. 23In the AS chimeras, both QH1-positive and -negative iASCs existed in the infrahyoid muscles (Figure 1O-R).
5][26] In the AS chimeras, thyroid follicular cells were characterized by the expression of thyroid transcription factor-1 (TTF-1), whereas iASCs were distributed in the TTF1-negative interstitium and expressed QH1 (Figure 1U-X).From the cellular morphology of QH1-positive iASC, it appeared to be integrated into the thyroid vascular network.From the cellular morphology of QH1-positive iASCs, it appeared to be integrated into the thyroid vascular network.
In sum, iASCs differentiate into various cardiovascular cell types, including cardiomyocytes, vascular smooth muscle cells and vascular endothelial cells, depending on the tissue environment.

| Intra-embryonic ASCs contribute to intra-organic vascularization in the thyroid gland
The thyroid gland is a conserved organ in vertebrates and produces hormones that are important for maintaining homeostasis. 27However, the details of its developmental process have not been fully clarified.In avians, the thyroid gland is derived from the endodermal floor of the pharynx through a process consisting of five stages: specification, budding, descent, bilobation, and folliculogenesis. 28,29The vasculature of the thyroid gland develops at approximately HH27 (E5), just after the bilobation stage. 18o elucidate when iASCs are distributed into the thyroid, we performed immunohistochemistry of mCherry as well as QH1 in mCherry-positive chimeras at HH27, HH30 (E6.5), and HH32 (E7.5).At HH27, QH1-positive iASCs were observed to be incorporated into blood vessels around the thyroid anlage visualized by ink injection (Figure 2A-D, arrowheads).At HH30, blood vessels were found in the thyroid, indicating that vascular invasion starts between HH27 and HH30 (Figure 2E-H).At HH32, mCherry-positive iASCs were abundantly distributed in the thyroid, most of which were positive for QH1 (Figure 2I-L).These results indicate that iASCs are incorporated into perithyroid blood vessels through vasculogenesis before HH27 and migrate into the thyroid as angiogenic endothelial cells between HH25 and HH32.
F I G U R E 1 Legend on next page.

| FGF and signals from the thyroid gland are required for vascularization by ASCs
Then we focused on the selective incorporation of iASCs into the thyroid vasculature to clarify the underlying mechanisms.To investigate whether the incorporation of iASCs into the thyroid vasculature required signals from the thyroid gland, quail ASCs at HH10 were co-cultured with the chick thyroid anlage at HH30.For comparison, ASCs were also co-cultured with clumps including the thymus and parathyroid gland (Figure 3A).However, QH1-expressing cells did not appear under either condition (Figure 3A,B,E,F).
In the previous study, we revealed that FGF signaling from the pharyngeal mesenchyme is important for AS to endothelial differentiation. 16Therefore, the same experiments were performed in the presence of FGF2.As a result, many QH1-expressing cells appeared from ASCs around the chick thyroid anlage at HH30, forming a sheet-like structure (Figure 3B-D).By contrast, only a few QH1-expressing cells were observed in co-culture with the clumps including the thymus and parathyroid gland despite the presence of FGF2 (Figure 3B,G,H ).The It is known that the thyroid hormone thyroxine starts to accumulate from HH10 and increases logarithmically until HH20 (E3). 28To investigate whether thyroxine is involved in vascular endothelial differentiation, we cultured ASCs with thyroxine.However, no QH1 expression was observed without or with FGF2, indicating that the endothelial differentiation of ASCs was independent of the hormone-secreting function of the thyroid (data not shown).

| Sequential FGF and VEGF signaling promotes vascular endothelial differentiation of ASCs
A previous study using mice reported that thyroid parenchymal cells express VEGF-A and the surrounding KDR/ VEGFR2-expressing cells differentiate into vascular endothelium. 30Indeed, the expression level of VEGF-A was higher in the thyroid gland than in the surrounding organs also in chickens (Figure 4A).These results led us to the idea that FGF signals from the pharyngeal mesenchyme and VEGF from the thyroid parenchyma may sequentially cooperate to promote endothelial differentiation of ASCs.Indeed, the expression of the major VEGF receptor KDR/VEGFR2 was found to be upregulated by FGF2 with a peak around 24 hours after treatment (Figure 4B).Furthermore, in vitro AS culture experiments showed that sequential treatment with FGF2 and VEGF-A for 24 hours each (FGF2 24 hours + VEGF 24 hours) significantly increased the number of QH1-positive cells in cultured quail ASCs (Figure 4C).QH1-positive cells typically constituted plexiform sheetlike structure expanding from AS explants, which were not induced to form by FGF or VEGF treatment alone, although sporadic QH1-positive cells were found in these conditions.This suggests that FGF2 may drive ASCs into a state similar to KDR/VEGFR2-expressing cells that differentiate various cell types including cardiovascular cells and hematopoietic cells. 31o further characterize the nature of QH1-positive cells, we investigated expression levels of vascular endothelium-related genes (undifferentiated endothelial and hematopoietic markers KDR/VEGFR2, TAL1, and LMO2, and mature vascular endothelial marker CDH5) [32][33][34] in the absence or presence of FGF2 and/or VEGF-A.As a result, the expression levels of these markers were significantly upregulated in the group of FGF2 24 hours + VEGF 24 hours compared to other groups (Figure 5A-D), suggesting that FGF signals stimulate KDR/VEGFR2 expression and subsequent VEGF signals promote vascular endothelial differentiation of ASCs.We also noticed that the expression levels of FGF receptors were significantly reduced in ASCs as endothelial differentiation progressed (Figure 5E-G).These results suggest that FGF works in the early stage of vascular endothelial differentiation.
To investigate whether vascular endothelial differentiation potentials of ASCs are dependent on developmental stages, the grafts of quail amniotic membrane at HH30 were cultured under specific conditions.In contrast to ASCs at HH10, amniotic grafts at HH30 did not yield QH1-positive cells even in the presence of both FGF and VEGF signaling (Figure 6).This result indicates that vascular endothelial differentiation potentials in response to FGF and VEGF signals are restricted to ASCs at early embryonic stages and disappear in mature amniotic cells, consistent with previous studies showing that mesodermal cells have differentiated into smooth muscle cells in the amniotic membrane by around HH27. [35][36][37]

| Single-cell transcriptome analysis identifies endothelial progenitor-like ASCs before intraembryonic translocation
To further investigate the heterogeneity of ASCs or iASCs in vivo, single-cell RNA sequencing (scRNAseq) was performed using the Fluidigm C1 system.First, we collected ASCs from HH10 chick which included cells of ectoderm and somatic mesoderm origins (Figure 7A).After quality filtering, a dataset comprising 89 ASCs was subjected to further analysis.To reveal characteristics of ASCs, scRNAseq data of ASCs were mapped on UMAP 38 and cells were divided into seven clusters (Figure 7B).Differentially expressed genes between each cluster shown in Figure 7C,D revealed that clusters 0 and 5 were likely ectodermal cells because they highly expressed the ectoderm or epithelium markers CDH1 and EPCAM. 39,40Therefore, other clusters are considered to be mesodermal, of which clusters 3 and 4 expressed mesodermal stemrelated genes such as CXCL12 and CDH11. 41,42NGPTL1, a growth factor largely specific for vascular endothelium, was also expressed in these clusters. 43ells in cluster 1 were characterized by the expression of some neuronal marker genes such as CDH2 and NCAM1.Of note, cells in cluster 6 had high expression of genes related to vascular endothelial or hematopoietic cells such as KDR/VEGFR2, CDH5, LMO2, PECAM1, and CD34.Cells in cluster 2 were regarded as of low quality because of the low number of detected genes (nFeature RNA) excluded from further analysis.Re-embedding with UMAP (Figure 8A) and velocity analysis 44 (Figure 8B) again showed ectodermal and mesodermal cluster groups, in the latter of which a flow towards cluster 6. Pseudotime analysis on the mesodermal clusters using Monocle 3 45 (Figure 8C,D) showed downregulation of FGF receptor gene expressions (Figure 8Ea-c) and upregulation of VEGF receptor gene expressions (Figure 8Ed-f) during the putative transition from mesodermal to endothelial cells.These results are consistent with the present in vitro experiments showing sequential requirement of FGF and VEGF signaling for ASC differentiation into vascular endothelial cells.

| DISCUSSION
Here, we describe that intra-embryonic amniogenic somatopleure cells (iASCs) differentiate into various cell types constituting the cardiovascular system including the thyroid vascular network.
Recent studies have identified three of heart fields (eg, the first, second, and tertiary heart fields) in intraembryonic splanchnic mesoderm. 46,477][48] In contrast, iASCs, which arise from the extra-embryonic somatopleure, migrate into the primary heart tube from the outflow tract during amnion formation. 16reviously, we have proposed a model in which FGF and BMP signaling from pharyngeal and outflow regions may direct iASC differentiation into endothelial cells and cardiomyocytes, respectively, at the outflow tract and right ventricle. 16The present study revealed that the distribution of iASCs in the interventricular septum at HH34-35 was enriched in the right ventricular side (Figure 1A-D).This distribution pattern of iASCs is similar to that of the cardiogenic precursor cells derived from the second heart field (SHF). 46,49It is an interesting possibility that iASCs may be a previously unrecognized extra-embryonic subpopulation of SHF.However, the expression of ISL1 (a SHF marker 49,50 ) was not detected in the AS mesoderm at HH10, 16 implying that iASCs are not incorporated in SHF at this stage.Further studies, including lineage analysis of iASCs with SHF markers (such as Isl1 and Fgf10 51,52 ) from earlier stages, will clarify the relationship between iASCs and SHF.Although it remains to be determined whether these cell populations represent distinct cell lineages or not, they likely share common behavioral characteristics to contribute to the formation of the outflow tract.
Stimulation of FGF significantly increased KDR/ VEGFR2 expression in ASCs, whereas the expression of FGFRs tended to decrease when the differentiation into the vascular endothelium progressed.This suggests that initial stimulation by FGF signal may act as a trigger for the early stages of vascular endothelial differentiation, and then thyroid-derived signals including VEGF may promote vascular endothelial cell maturation.Vascular invasion into the thyroid had already started at HH27 (Figure 2E-H), and transplanted quail ASCs (QCPNpositive cells) entered the thyroid and differentiated into QH1-positive vascular endothelial cells (Figure 1S-X) before HH35, by which differentiation steps involving sequential FGF and VEGF-A signaling are already accomplished.Since VEGF-A is also secreted from the thyroid gland at HH35, differentiation of QH1-positive endothelial cells in the co-culture with quail ASCs seems to recapitulate these processes (Figure 3C).6][37] Based on these findings, we propose a model for the differentiation process of iASCs as shown in Figure 9.
According to scRNAseq, some ASCs express markers of hematopoietic progenitors.Interestingly, it has been reported that hematopoietic foci like blood islands are found in the mesenchyme close to the branchial artery in HH15 chick embryos. 53,54][57] Moreover, Shigeta et al demonstrated macrophage differentiation from the hemogenic endocardium and its essential role in valvular remodeling, 58 suggesting that the temporary hematopoiesis may be important for tissue remodeling, as well as for supplying circulating blood to the whole body.Although we were unable to show directly that ASCs differentiate into blood cells, it is quite possible that iASC may contribute to hematopoiesis through endothelial differentiation in the cardiac outflow tract and pharyngeal arch regions or directly forming hematopoietic foci.
The process of amnion formation varies greatly among species.Since reptiles and birds diverged from the same ancestor and developed amnion, 59 the properties of the amnion of these two species may be similar.Rather, mammals have acquired novel traits such as retraction of the chorioallantoic membrane and formation of the placenta.Despite these differences, mouse studies have reported that Flk1 (KDR/VEGFR2 herein)-expressing cells are present in the amnion, 31 and that there are a few cells destined for blood islands and PGCs in the presumptive The number of feature RNA (nFeature RNA) and expression levels of some genes of interest on the UMAP.Each plot is colored according to the Z score of the expression value amnion. 60Moreover, Zhang et al recently reported that a part of the extraembryonic mesoderm gives rise to cardiomyocytes and serosal mesothelial cell types. 61These cells may be considered comparable with iASCs in avians, although the route of entrance into the developing heart appears to be different between arterial and venous poles possibly due to topological differences in the position of the extraembryonic/intraembryonic boundary.
From an evolutionary viewpoint, it is intriguing to pursue whether iASCs contribute to the anatomical and functional diversity among species.The thyroid gland is considered to be evolutionarily derived from the endostyle, an exocrine organ, and the branching structure is a remnant of that. 62In particular, the morphology and anatomy of the thyroid and associated vessels changes between species. 27,63Therefore, it is interesting to investigate if the different usage of iASC might be involved in thyroid morphogenesis.
The next challenge is to explore how the AS was acquired in the process of evolution.One fascinating possibility is that some migratory pluripotent cells with a potential to diverse cell types might led to the acquisition of the amniotic membrane, as neural crest cells, pluripotent stem cells with migratory capacity arising in the neural plate border, generated the "new head" as an evolutionary novelty in vertebrates. 64For this purpose, it is necessary to explore whether there is a homologue of the AS in non-amniotes such as fish and amphibians.

| Animals
Fertilized Boris Brown (Gallus gallus, wild type, WT) eggs and Japanese quail (Coturnix japonica, WT) eggs were obtained from Inoue Egg Farm (Kanagawa, Japan) and Motoki Hatchery (Saitama, Japan), independently.PGK:H2B-chFP-TG (chFP) quail eggs 20 were provided by the Avian Bioscience Research Center at Nagoya University as part of the National Bioresources Program.All eggs were incubated at 37 C in a humid incubator at 37 C to appropriate embryonic stages.All experiments using animals were reviewed and approved by the University of Tokyo Animal Care and Use Committee and were performed in accordance with the institutional guidelines for care and use of laboratory animals.Staging of chick/quail embryos was according to Hamburger and Hamilton. 19

| Quail-chick chimera
To reveal the distributions of specific cells of interest, quail-chick chimera analysis was performed.Quail-chick chimeras of the AS ("AS chimera") have been described in the previous study. 16When making AS chimera, HH10 WT, or chFP quail AS grafts were transplanted into both the right and left regions of the AS of HH10 host chick embryos.

| Histological analysis
For histological analyses, AS chimeras and embryos were collected, some of which injected ink (Kiwa-Guro, Sailor, Japan) into the extra-embryonic vitelline vein or into the heart using a glass micropipette to visualize whole blood vessels.Embryos were then fixed with 4% paraformaldehyde (PFA), stored in PBS, embedded in paraffin (Kanto Chemical, Tokyo, Japan).
Images of the immunofluorescence staining were captured using a Nikon C2 confocal microscope.DAB staining was observed under OLYMPUS AX80 motorized microscope.All images were processed using ImageJ and Nikon NIS Elements software.

| Explant cultures
The AS grafts which included both ectodermal and mesodermal layers were excised from HH10 WT or chFP quail embryos and placed in a gelatin-coated 3.5-cm dish (IWAKI) containing 1%FBS DMEM (Wako) medium at 37 C in 5% CO 2 to appropriate time.In organ culture, the thyroid, parathyroid, and thymus from HH30 or HH35 chick embryos were collected, and coculture with the quail AS graft.The explants were pasted on the bottom dish with a glass needle or fine forceps before incubation.When the cut out HH10 quail AS tissue which contained both ectodermal and mesodermal layer was cultured on the gelatin coated dish, the ectodermal cells became shrunk and mesodermal cells expanded radially.In some experiments, L- thyroxine (Nacalai, 100 nmol/L), rm-bFGF (FGF2) (233-FB, R&D, 20 ng/mL) and VEGF (293-VE, R&D, 50 ng/mL) were added in the culture medium (1% FBS in DMEM).The concentration of each additive was determined after preliminary experiments based on previous reports.For control, an appropriate amount of vehicle was added in the culture medium.
For immunocytochemistry, cells were fixed with 4% PFA for 10 min.After washing with PBS, cells were immunoreacted with first antibodies for overnight and secondary antibodies for 2 hours and stained with DAPI solution as described in histological analysis.Images of the immunofluorescence staining were captured using a Keyence BZ-X700 microscope.
The explant cell areas and QH1 positive areas were quantified after binarizing the images with ImageJ software.All ectodermal regions or the part where the grafts overlapped to form a thick cell layer were excluded from analysis.

| RNA extraction and quantitative real-time polymerase chain reaction
For RNA extraction from cultured quail ASCs, the cells were washed with PBS to remove medium and lysed with ISOGEN2 (500 μL, Wako).After lysing cells, total RNA was obtained following the manufacturing procedure.The RNA concentration was determined at 260 nm by SimpliNano.The total RNA was reverse transcribed following the ReverTra Ace qRT-PCR Kit (Toyobo) procedure.The primers used for quantitative reverse transcription-polymerase chain reaction (qRT-PCR) are listed in Table S1.For primers of quail GAPDH, VEGFR2, and TAL1 were referred to Giles et al 65 and qual CDH5 was referred to Yvernogeau et al. 66 Other primers were designed by Primer-Blast (https://www.ncbi.nlm.nih.gov/tools/primer-blast/) or Primer3Plus (https://www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi).PCR was performed on GeneAmp PCR SYSTEM 9700 (Applied Biosystems).Quantitative RT-PCR was performed using the LightCycler 480 (Roche) real-time system according to the manufacturer's instructions.To reduce errors when applying to 96-well plates, duplicate samples were created.Absolute quantification was performed using the second derivative max method with the number of cycles of the maximum value when the amplified signal was differentiated twice as the cross point (Cp).Then, the concentration was calculated from the standard curve and the value divided by the GAPDH value was compared.The value of concentration of duplicated samples was averaged.All experiments were performed in parallel and repeated three or more independently.

| Cell preparation
For ASCs single-cell preparation, AS grafts including ectodermal and mesodermal layers were collected from more than 15 chick embryos at HH10 (referred to as st10 or E1.5 in the dataset).The grafts were treated with 0.25 w/v% trypsin/EDTA (0.2 g/L) at 37 C, 15 min to dissociate into single cells.Enzymes were neutralized with an equal volume of DMEM with 10% fetal bovine serum (FBS).Cell suspensions were filtered through a 35-μm nylon mesh cell strainer (FALCON, 352235) to exclude debris and cell clumps.

| mRNA sequencing
Prepared cells were captured on the Fluidigm C1 Single-Cell Auto Prep Array integrated fluidic circuit (IFC) for small cells (5-10 μm in diameter).Cells were loaded on the IFC chips at concentration of 500 cells per μL.All captured cells were photographed to check if they were truly dissociated for single cells by using KEYENCE BZ-X710.cDNAs of captured single cells were prepared by using SMARTer v4 Ultra Input Low RNA kit for the Fluidigm C1 System (Clontech).Then cDNAs were quality checked by using the Agilent 2200 TapeStation system and quantified by Qubit (Thermo Fisher).High-quality cDNAs were further constructed sequencing libraries by using Nextera XT DNA Sample Preparation Kit (Illumina).Each single-cell cDNAs were sequenced 50 pair-end on the Illumina Hiseq 2500.

| Data analysis
Data analysis was performed in R version 3.6.0and Python 3.7.All graphical figures were created by the ggplot2 R package.Sequence output fastq files were aligned to chick ensemble genome reference (Gallus_gallus.GRCg6a.96)with extra GFP and spike DNA sequences by using HISAT2 software version 2.1.0.Gene expression counts were calculated with the feature-Counts function from the Rsubread version 1.34.7 package using R.All scRNA-seq data have 123 ASCs.
The gene name was converted from chick Ensembl ID to gene symbol, and duplicated genes integrated as Ensembl ID to use for further analysis.Each count data F I G U R E 9 Presumptive model of the process by which amniogenic somatopleural cells (ASCs) or Intra-embryonic ASCs (iASCs) differentiate into the thyroid vascular endothelial cells.scRNA-seq data reveal the existence of VEGFR2-positive cells in the AS before entering the embryo.After migrating into the embryo, fibroblast growth factor (FGF) signaling from the surrounding mesenchymal cells upregulates VEGFR2 expression in iASCs to be incorporated into pharyngeal vasculature through vasculogenesis.Then, vascular endothelial growth factor (VEGF) signaling from the thyroid further accelerates the maturation and iASCs migrate into the thyroid by angiogenesis were converted into a Seurat object by using Seurat version 4.0.4R package. 67H10 ASCs were normalized using 89 cells of number of feature RNA > 2000.For cell clustering, the twodimensional UMAP space was used. 38Each differentially expressed genes (DEGs) corresponding to each cluster were calculated by using FindAllMarkers function.The statistical significance was performed with a Wilcoxon Rank Sum test and the threshold was limited by log2 fold change was more than 0.25.For Figure 7C, the heatmap was created by using DoHeatmap function and the color indicates z-scored expression values in each cell.
RNA velocity analysis 44 was performed by using velocyto, including velocyto.pyversion 0.17.17Python package and velocyto.R version 0.6.The expression counts of each spliced and unspliced genes were calculated through mapping to chick genome reference (Gallus_gallus.GRCg6a.96)by using run_smartseq2 function from velocyto.py.Then, RunVelocity and show.velocity.on.embedding.corfunction in SeuratWrappers version 0.1.0package in R were used for the visualization of RNA velocity on the UMAP space.From the direction of RNA velocity, each sub trajectory of pseudotime was calculated by using monocle3 version 0.1.1R package.This was used for the visualization of transition of gene expression patterns along with pseudotime.
The dataset is available at the NCBI Gene Expression Omnibus, under accession GSE201285.

| Statistical analysis
For Figures 3D and 4D, P-values were determined by ANOVA with Tukey-Kramer's post hoc test.Pvalues < 0.05 were considered statistically significant.For Figure 5A-D, the Kruskal-Wallis test as a nonparametric method was used to compare mean values, followed by the Mann-Whitney U-test.R environment (Rstudio, Inc., Boston, USA) was used.P-values < 0125 were considered statistically significant.For Figures 4A and Figure 5E-H, Welch's t-test for unequal variances was performed after the normality of the data was confirmed using the Shapiro-Wilk normality test (α = 0.05).P-values <. 05 were considered statistically significant.Data were obtained from three or more independent experiments.

| CONCLUSION
Here, we showed that amniogenic somatopleure cells (ASCs) differentiate into various cell types constituting the cardiovascular system, with some populations having a molecular background similar to that of hemangioblasts.Among them, those contributing to the thyroid vascular network were suggested to differentiate into vascular endothelial cells with FGF-specification and VEGFinduced maturation.This study will provide a novel clue for understanding the cardiovascular development of amniotes from embryological and evolutionary perspectives.

K
E Y W O R D S amniogenic somatopleure, angiogenesis, cardiovascular development, FGF, quail-chick chimera, thyroid development, VEGF

F I G U R E 1
Intra-embryonic amniogenic somatopleural cells (iASCs) in the heart, infrahyoid muscles, and thyroid gland of HH34-35 AS chimeras.(A-D) Images of quail cell marker antibody (QCPN) immunostaining in transverse sections of the heart.QCPN-positive cells (brown) are considered iASCs of quail origin.(B) and (D) are higher magnifications of boxes in (A) and (C), respectively.(E-H) Immunostained images of a section adjacent to the box indicated in (D).mCherry signals indicate iASCs.MF20 and 4 0 ,6-diamidino-2-phenylindole (DAPI) indicate cardiomyocytes and nuclei, respectively.Arrowheads and arrows indicate MF20 positive and negative iASCs, respectively.(I-L) Immunostained coronary septal branch in a section adjacent to the box indicated in (D).Arrowheads indicate smooth muscle α-actin (αSMA) and mCherry double positive iASCs.(M) QCPN immunostaining in transverse sections of the cervical tissue.(N) Magnified image of the box in (M).(O-R) Immunostained images of a section adjacent to (N).mCherry and QH1 indicate iASCs and quail-derived angioblasts/vascular endothelial cells, respectively.Thyroid transcription factor-1 (TTF1) stains the epithelium of trachea.Arrowheads and arrows indicate QH-positive and -negative iASCs, respectively.(S) QCPN immunostaining in a transverse section of the thyroid gland.(T) Magnified image of the box in (S).(U-X) Immunostained images of a section adjacent to the box in (T).TTF1 stains the thyroid parenchyma.Scale bars: 1 mm for (A, C, M, and S), 50 μm for other images.av, aortic valve; ca, carotid artery; es, esophagus; ifh, infrahyoid muscle; jv, jugular vein; LV, left ventricle; ng, ganglion; pa, pulmonary artery; pt, parathyroid; RA, right ventricle; RV, right ventricle; sb, coronary septal branch; thy, thyroid gland; tr, trachea F I G U R E 2 Intra-embryonic amniogenic somatopleural cells (iASCs) integrated into the peri-and intra-thyroid vasculature.Immunostained or angiographic images of AS chimeras with QH1 for quail-derived angioblasts/vascular endothelial cells and thyroid transcription factor-1 (TTF1) for the thyroid parenchyma at HH27 (A-D), HH30 (E-H), and HH32 (I-L).Arrowheads in (A) and (E) indicate QH1-positive iASCs, which line vessel lumens visualized by ink (D, H).Arrowheads in (I) indicate QH1 and mCherry double positive iASCs in the thyroid.Note that QH1 antibody reacted with hematopoietic cells as a background, as shown in (J).Scale bars = 100 μm.thy, thyroid; TW, thoracic wall

F I G U R E 3
Fibroblast growth factors (FGF) and signals from the thyroid gland are required for endothelial differentiation of amniogenic somatopleural cells (ASCs).(A) Schematic illustration of co-culture experiment.(B, C) QH1-and 4 0 ,6-diamidino-2-phenylindole (DAPI)-stained images.(a-d) Quail AS grafts are co-cultured with HH30 or HH35 chick thyroid explants in DMEM with 1% FBS in the absence (a and b; n = 12 for HH30, n = 3 for HH35), or the presence (c and d; n = 13 for HH30, n = 4 for HH35) of FGF2 (20 ng/mL).Arrowheads indicate QH1-positive endothelial cells.(e-h) Quail AS grafts are co-cultured with HH30 or HH35 chick parathyroid and thymus explants in DMEM with 1%FBS in the absence (e and f; n = 3 for HH30, n = 3 for HH35), or the presence (g and h; n = 4 for HH30, n = 3 for HH35) of FGF2 (20 ng/mL).Asterisks indicate remnant organ grafts.Scale bars = 200 μm.(D) Quantitative comparison of QH1-positive areas."para" indicates parathyroid and thymus.Box plot shows the median and quartiles 1 and 3 of the ratio of QH1 positive-to DAPI positive-areas of each group.*P < 0.05, **P < 0.0001 same results were obtained when quail ASCs were cocultured with HH35 (E9) chick organs (Figure 3C,D).These results indicate that both FGF and thyroid-derived signals are required for endothelial differentiation of ASCs.

F I G U R E 5
Effects of fibroblast growth factors (FGF) and/or vascular endothelial growth factors (VEGF) on the expression levels of endothelial differentiation-related genes.(A-D) quantitative real-time polymerase chain reaction (qRT-PCR) analysis of 48 h-cultured quail amniogenic somatopleural cells (ASCs) under conditions as in Figure 4C for mRNA levels of VEGFR2 (A), TAL1 (B), LMO2 (C), and CDH5 (D).Data were obtained from five independent experiments.Error bars indicate SE. *P < 0.0125.(E-H) The expression levels of KDR/ VEGFR2 and FGF receptors before (culture 0 h) and after (FGF24h + VEGF24h) endothelial differentiation of ASCs.Error bars indicate SE. *P < 0.05, **P < 0.01

F
I G U R E 6 Mature amnion cells do not respond to fibroblast growth factors (FGF) and vascular endothelial growth factors (VEGF) signaling.(A) HH30 quail embryo which is completely wrapped in amniotic membrane (arrow).(B-I) QH1-and 4 0 ,6-diamidino-2-phenylindole (DAPI)-stained images of HH30 amnion grafts are cultured under conditions as follows: Control 48 h, n = 7 (B and C); control 24 h + 50 ng/mL VEGF-A 24 h, n = 5 (D and E); 20 ng/mL FGF2 48 h, n = 5 (F and G); FGF2 24 h + VEGF-A 24 h, n = 7 (H and I).Asterisks indicate ectodermal tissues.Scale bars = 200 μm F I G U R E 7 Single-cell RNA sequence (scRNA-seq) of HH10 chick amniogenic somatopleural cells (ASCs) reveals the existence of vascular endothelial progenitor-like cells in the AS even before intraembryonic translocation.(A) Schematic illustration of single-cell preparation.Both ectodermal (ect) and somatic mesoderm (Som-mes) cells were collected from HH10 chick somatopleural tissues.(B) scRNA-seq data of all ASCs (89 cells) are displayed as plots on UMAP with different colors representing different clusters.(C) The heatmap shows top 50 differentially expressed genes (DEGs) sorted by log2 fold change between clusters detected by Seurat FindAllMarkers function.The heatmap is colored according to the Z scores of the average expression values in each cluster.Yellow and purple represent high and low expression levels, respectively.(D)

F I G U R E 8
Fibroblast growth factors (FGF) and vascular endothelial growth factor (VEGF) receptors expression dynamics along pseudotime.(A) UMAP with a dataset of 75 ASCs from which low-quality ASCs are excluded (cells in cluster 2 shown in Figure 7B).The cluster numbers correspond to those in Figure 7B.The clusters are clearly divided into two major populations: the ectoderm (clusters 0 and 9) and mesoderm (clusters 1, 3, 4, and 6) components.(B) RNA velocity based on UMAP.Arrows indicate local average velocity on a regular grid vector field.(C) The mesoderm component in (A) is extracted.(D) Pseudotime values were calculated and plotted.The starting point of the pseudotime is set arbitrarily by referring to the result of velocity shown in (B).(E) Trajectory mapping of the mesoderm component shows some stratification between cell states along the pseudotime, with high expression of FGF receptors at the early stage (a-c) and vascular endothelial growth factor (VEGF) receptors at the late stage (d-f).The color of each plot corresponds to the color shown in (C) discussion and encouragement.This work was supported in part by JSPS KAKENHI Grant Numbers 20 J11793 (Yuka Haneda), 21 K19519, 19H01048, and 22H04991 (Hiroki Kurihara), 19 K08308 and 22 K07877 (Sachiko Miyagawa-Tomita) and a stipend from the World-leading Innovative Graduate Study Program for Life Science and Technology (WINGS-LST) (Yuka Haneda).