Adaptive patterning of vascular network during avian skin development: Mesenchymal plasticity and dermal vasculogenesis

Abstract

A vasculature network supplies blood to feather buds in the developing skin. Does the vasculature network during early skin development form by sequential sprouting from the central vasculature or does local vasculogenesis occur first that then connect with the central vascular tree? Using transgenic Japanese quail Tg(TIE1p.H2B-eYFP), we observe that vascular progenitor cells appear after feather primordia formation. The vasculature then radiates out from each bud and connects with primordial vessels from neighboring buds. Later they connect with the central vasculature. Epithelial-mesenchymal recombination shows local vasculature is patterned by the epithelium, which expresses FGF2 and VEGF. Perturbing noggin expression leads to abnormal vascularization. To study endothelial origin, we compare transcriptomes of TIE1p.H2B-eYFP⁺ cells collected from the skin and aorta. Endothelial cells from the skin more closely resemble skin dermal cells than those from the aorta. The results show developing chicken skin vasculature is assembled by (1) physiological vasculogenesis from the peripheral tissue, and (2) subsequently connects with the central vasculature. The work implies mesenchymal plasticity and convergent differentiation play significant roles in development, and such processes may be re-activated during adult regeneration.

Summary statement

We show the vasculature network in the chicken skin is assembled using existing feather buds as the template, and endothelia are derived from local bud dermis and central vasculature.

1. Introduction

Neovascularization is a critical step in development as all tissues require adequate nutrients, immune surveillance, and removal of metabolic wastes (Marcelo et al., 2013). Most research in this area has focused on heart formation in early embryos. How the vascular network is established in other areas remains largely unknown. Here we explore the vascularization process in avian skin. We previously proposed a multiscale model for avian skin development. First from the non-patterned morphogenic field, multiple feather primordia emerge via a “de novo periodic patterning” process (Jiang et al., 1999; Jung et al., 1998). Next, skin mesenchymal cells (e.g., adipocytes, endothelium, and muscle cells) are assembled using the existing feather patterns as reference points in an “adaptive patterning” process, as demonstrated by the self-assembly of chicken intradermal muscle networks (Wu et al., 2019). By the time feather muscles finish forming networks between feather follicles, an obvious vascular network is also established but no studies have clearly demonstrated the origins of those vasculatures. Here, we examine if the current adaptive patterning model can apply to avian skin neovascularization.

Neovascularization commences via two common mechanisms: vasculogenesis and angiogenesis (Venkatakrishnan and Parvathi, 2022). Vasculogenesis, defined as the de novo formation of vasculature from differentiating endothelial progenitor cells or angioblasts, gives rise to the first wave of blood vessels and blood cells (Metikala et al., 2022). Angiogenesis, on the other hand, is defined as the formation of vasculature from the extension of existing blood vessels (Metikala et al., 2022). In zebrafish, embryonic endothelial cells (ECs) share common progenitors with blood cells (Okuda and Hogan, 2020). While in chickens, some ECs are derived from somitic mesoderm and move into the dorsal aorta (Sahai-Hernandez et al., 2023). In addition to ECs, angiogenesis is dependent on the peripheral nervous system (Hao et al., 2015; Rasmussen et al., 2018). Although the existence of these two processes has been long known, it is still not clear whether skin vasculatures are formed by vasculogenesis or angiogenesis. Some investigators have suggested primary angiogenesis is a prominent feature of developing skin during embryogenesis (Detmar, 1996). At the same time, several studies suggested that the developing organ produces paracrine factors to induce blood vessel formation from the mesenchyme (Dight et al., 2022). This conflicting evidence prompted us to revisit this long-standing question using the latest techniques and animal models.

To study avian skin vascular patterning, researchers previously relied solely on immunostaining with a limited selection of molecular markers. Vascular endothelial growth factor-2 (VEGF-2) and its receptor, quail endothelial kinase (Quek1) also known as fetal liver kinase (Flk-1) (Sugishita et al., 2000), were only seen from day 7 onwards (Nimmagadda et al., 2004). Next, injecting virus expressing H2B-GFP to early embryonic blood islands enabled investigators to determine their contribution to embryonic blood vessels (LaRue et al., 2003). Later, to show EC lineage with more precision, they generated a transgenic quail line (TIE1p.H2B-eYFP) using a conserved endothelial marker, TIE1 (Sato et al., 1995), to express H2B-eYFP only in vascular cells. This transgenic line was used to analyze dynamic quail vascular morphogenesis from Hamburger and Hamilton stage 6 (HH6) up to HH12 at embryonic day 2 (E2) (Sato et al., 2010). Even in the mouse model, Tie1 is expressed in the developing heart, migrating aorta ECs differentiating angioblasts within the head mesenchyme, and yolk sac blood islands (Korhonen et al., 1994; Sato et al., 1993) and its expression persists in ECs throughout embryogenesis.

This led us to use a transgenic line that uses the Tie1 promoter to drive the expression of a fluorescent marker (TIE1p.H2B-eYFP quail and a chicken model to characterize vascular progenitor cell behavior in 3-dimensions from HH28-44. We show that H2B-eYFP+ cells are organized into a vasculature network pattern in a feather epithelium dependent manner. We also show that feather growth relies on normal vessel formation by over-expressing the neovascularization-inhibitory gene, noggin in ECs. Lastly, by comparing the transcriptome of H2B-eYFP+ cells in developing avian skin with that of the aorta, we propose that blood vessels within feathered skin are formed by vasculogenesis that may also involve other biological processes. Here, we identified endothelial progenitors that affect the next level of avian skin development. Understanding how vasculature is formed and identifying endothelial progenitor cues that influence developing skin patterning events may shed light on how to induce vascularization in regenerating skin after wounding, and in tissue-engineered organoids for future translational medical purposes.

2. Results

Formation of vascular network in the developing chicken skin is coupled with the formation of feather buds and follicles

We first characterized the morphogenic events of intra-feather vessels in a chicken model system from at embryonic day 10 (E10; HH36) to E18 (HH44) during their first regenerative feather cycle (n=10). Vascular development in quail embryos is shown schematically, in wholemount views and in longitudinal sections of feathers at E10 – E18 (Fig. 1A, B).

Figure 1. Developing skin vasculature from chicken and Tg(TIE1p.H2B-eYFP) quail.

(A). Schematic of blood vessel growth patterns in chicken feathers from E10 to E18. The vessels are indicated by red lines and the capillaries by red hatching marks.

(B). H&E stain of chicken skins from E10 to E18. The vessels in the skin and the feather are in red. The left column shows whole-mount samples; the right column shows skin longitudinal sections of developing feather follicles. Blood vessels are indicated by red arrows. Blood vessels within feather buds are indicated by white arrows. Sites of blood vessel primordia are indicated by white arrowheads. Left column, scale bar: 200 μm; right column, scale bar: 100 μm.

(C). Plucked chicken or quail feathers from newborn chicks. Whole mount view of the base of a feather (left panel), Whole mount view of a feather shaft once the sheath is removed (2^nd panel from the left). A sectioned chicken feather stained with VEGFR2 (3^rd panel from the left, red) to identify blood vessel ECs. Whole mount feather from the TIE1p.H2B-eYFP (green) transgenic quail showing abundant ECs in the feather pulp (right panel). Scale bar: 500 μm.

(D). TIE1p.H2B-eYFP transgenic quail foot (E6, top panel) and body (E5, HH16; bottom panel). The EYFP is shown in green. Scale bar: 500 μm.

(E). Whole mount skins of E8 quails. a wildtype quail stained with quail endothelial antigen (red; left panel) and TIE1p.H2B-eYFP transgenic quail (green). Scale bar: 100 μm.

At E10 (HH36), the feather vessels are connected to central vasculatures and supplied with blood (Meyer and Baumgärtner, 1998), thus they are easy to visualize. Major vessels run through the interbud area with branches invaginating into each feather bud (Fig. 1A, B). The major vessel (red arrows) forms a coiled loop inside the growing feather bud. Small vessels can be seen connecting to the major vessel (white arrows) and between themselves (white arrowheads) within the forming feather pulp.

At E11 (HH37), the long bud invaginates into the dermis (Jiang et al., 2011). As short feather buds grow to become long buds at E11, the pulp blood vessels branch further and expand inside the feather bud to support feather development. The major vessels form linear connections from the bottom to the top of the feather.

At E14 (HH40), within the rapidly elongating buds (Meyer and Baumgärtner, 1998) the vasculature network narrows to accommodate the available space inside the keratinizing feathers. But the blood supply can still be observed from the bottom to the top of the feather.

At E18 (HH44), the growth of feather follicles and filaments are complete. As the distal feather becomes progressively keratinized, the vasculature begins to be absorbed at E16 (data not shown). At E18, no vasculature is seen inside the feather.

In summary, the avian skin vasculature develops in parallel with the feather buds and follicles during the first regenerative feather cycle. Vessels continuously grow and branch into feathers in anagen phase before HH42, while the pulp is resorbed in catagen phase after HH42.

2.1. Characterization of vascular network formation before blood flow

It is challenging to systematically characterize the prenatal vascular patterning in chicken embryos because E5.5 to E8 (HH26-HH35) embryos lack the blood flow needed for vessel visualization. Hence, to overcome this obstacle we switched to a transgenic Japanese quail (TIE1p.H2B-eYFP) line that expresses nuclear-localized eYFP in Tie1+ ECs. Tie1 is a receptor tyrosine kinase that is specifically expressed in the endothelial lineage (Partanen et al., 1992; Sato et al., 1995). The H2B-eYFP signal of this transgenic quail consistently colocalizes with TIE1, VE-CADHERIN mRNA, and QH1 immunostaining (a reference standard for quail vascular endothelial lineage) (Sato et al., 2010). Together these markers suggest H2B-eYFP+ cells in this quail model can faithfully represent the locations of vascular progenitor cells before blood is supplied.

The expression pattern of Tie1 H2B-eYFP from the pulp of newborn quail feathers resembles the pattern seen in chicken vasculature stained for VEGFR2 (Fig. 1C; n=3). This pattern also matches that of endogenous VEGFR2 mRNA in the lateral portion of somites and their immediate surroundings (Nimmagadda et al., 2004). The Tie1 H2B-eYFP expression pattern is found much earlier in HH16 (E4) quail indicating the first stages of vasculature cell lineage (Fig. 1D, bottom panel; n=3). Later, in the E6 embryonic limb, strong H2B-eYFP expression is seen at the boundary of the digits (Fig. 1D, top panel; n=3), consistent with that of endogenous VEGFR-2 mRNA. In E8 quail feather buds, the Tie1+ cells show identical patterns as that seen with anti-quail endothelial antigen staining (Fig 1E; n=3). With this evidence, we conclude that the TIE1p.H2B-eYFP transgenic quail is suitable for tracing the vascular cells even after HH12. Thus, we use this transgenic quail model to visualize vascular patterning in the feather primordia of developing avian skin prior to blood flow.

2.2. Vascularization takes place in feather bud dermis before connection to central vasculatures

To better understand initial stages of blood vessel formation we traced H2B-eYFP cells from E5.5 to E8 (HH26-HH35). During E5.5 – E6.5 (HH26-HH30) blood vessels rapidly expand within the skin. Then during E6.5-E8 the vasculature invaginates into the feather bud dermis. We discovered that vascular progenitors are organized into specific patterns in feather bud dermis before connecting to the central vasculature.

Our observations show that blood vessels first appear at E5.5-E6.5, then the eYFP+ cells radiate in every direction. As the feather primordia form, the eYFP+ cells aggregate under feather primordia. At this stage, the eYFP+ cells can form linear patterns that cannot be functional vessels because they are mostly 1- or 2-cells wide. A more complete description of EYFP+ cell patterning during various HH stages is detailed below.

At E5.5 (HH27-28), prior to obvious feather primordia formation, dense parallel bands of H2B-eYFP+ endothelial cell densities are visible from the head to the tail (Fig. 2A whole mount view; n=5). A magnified view of the skin is shown in Fig. 2B (n=5) for the region highlighted in Fig. 2C) In the dorsal skin cranial to the wing bud, near the midline the vasculature is forming with high H2B-eYFP+ endothelial cell densities. Low endothelial cell densities with few H2B-eYFP+ cells lie laterally adjacent to the midline. Toward the lateral regions higher H2B-eYFP+ endothelial cell density increases, but these primitive vasculatures do not connect with the central vasculature.

Figure 2. Neovascularization in the quail skin before the circulation is established.

(A). TIE1p.H2B-eYFP transgenic quail embryo at the age of E5.5. Brightfield (left) and EYFP imaging of the embryo. Scale bar: 1 mm.

(B). Whole-mount skins of TIE1p.H2B-eYFP quails from E5.5 to E8. The left column shows the brightfield skin images (with the midline in the center); the right column shows the EYFP (green) channel of the same region. Scale bar: 300 μm.

(C). Schematic drawing highlighting patterns of the Tie1+ cells and feather follicles (condensations, buds) in developing quail skin by age. The Tie1+ cells are drawn as green dots. Dotted vs solid green lines reflect continuity and solid lines indicate relative vessel thickness. Feather follicles (condensations, buds) are drawn as circles or ovals depending on their shape. The yellow circles indicate newly formed feather condensations; the orange ones are condensations that appeared in previous stages. Feather buds with keratinization are drawn with black lines. Boxes drawn with dotted lines depict the regions shown in panel B.

At E6 (HH29), the first row of feather primordia forms symmetrically and bifurcates around the trunk region only to rejoin along the caudal skin midline (Fig. 2B, C). At this stage, H2B-eYFP+ cell densities consolidate in the midline and begin to enter the newly formed feather primordia. H2B-eYFP+ cells in the mid-dorsal region become enriched within the thicker tissue near the midline and connect to the central vascular network. In the caudal skin these cells are connected to the lateral trunk vasculature.

At E6.5 (HH30), feather primordia begin to grow out from the skin surface to become feather buds; the number of feather buds in a row is 1-2 (cranial), 5 (middle), 2-3 (caudal) plus one feather bud along the midline (Fig. 2B, C). H2B-eYFP+ cells develop inside the feather buds accordingly and inter-bud connections between each feather bud begin to be observed.

The second stage starts from HH31 to HH35 (E6.5 to E8). At this stage, the eYFP+ cells within the feather buds aggregate and become more obvious. As each feather bud matures and elongates, the eYFP+ cells gradually form a tubular pattern and connect with each other. A more detailed description for pattern of the EYFP+ cell during various HH stages is available below.

At E7 (HH31-32), feather buds elongate and become narrower; 1-2 newly formed feather buds are added to the lateral side of each feather row (Fig. 2B, C). H2B-eYFP+ cells continue to reside within the feather bud.

At E7.5 (HH33), pigmentation begins to appear. The innermost two rows of feather buds have black and yellow pigmentation, and a new row of feather primordia begin to be seen along the midline (Fig. 2B, C). H2B-eYFP+ cells become more organized and form inter-connected linear patterns within the feather bud and interbud regions.

At E8 (HH34-35), the feather buds elongate further, and the distal tip becomes pointed (Fig. 2B, C). H2B-eYFP+ cells appear to grow from a narrow base, enter the feather bud core and spread out in several linear tracks, then grow together with the feather buds. At this stage chick embryos were reported to establish a good capillary supply which penetrates into the young feather papilla of the protruding feather buds (Meyer and Baumgärtner, 1998).

Together, these data show that vascularization takes place in feather bud dermis with specific pattens at each different stage before blood flow can be seen. This observation indicates the neovascularization in early skin is likely contributed by in situ vasculogenesis.

2.3. Dynamic imaging of the avian skin vasculature

Observing the highly synergistic development of feathers and vasculatures, we wondered whether the feather bud cells directly interact with the vessels. The model of adaptive patterning suggests the feather bud cells serve as the primary reference points for ancillary structures which differentiate from the skin mesenchyme (Wu et al., 2019). To study this question, we turned to an ex vivo skin explant model (Fig. S1A) because it can be readily observed and manipulated. Previously, chicken skin explants were shown to be a powerful model to study feather morphogenesis and skin myogenesis (Chuong, 2000; Wu et al., 2019) (Chuong et al., 2000; Wu et al., 2019). Such a model can be cultured in a dish for more than 10 days and recapitulate the major developmental events of skin including invagination as well as feather follicle and feather muscle formation (Jiang et al., 2011, 2023). The skin explant culture method was also used to study quail pigment formation (Inaba et al., 2019). As the skin neovascularization takes place before pigmentation, we first confirm if the quail skin explant model before HH35 can be used to study neovascularization.

We verified the skin explant model by comparing samples grown in culture for 24 hours (24hr) with skins grown in ovo (Fig. S1A-C; n=3). For E5.5 Tg(TIE1p.H2B-eYFP) quail skin grown in culture 24hr, feather primordia form and the H2B-eYFP⁺ cells aggregate to the newly formed feather primordia (Fig. 3C) paralleling their behavior in vivo (Fig. 2B). However, H2B-eYFP⁺ cells weren’t as bright if the skins were cultured for more than 72hr (Fig. S1B). We also characterized the distribution of H2B-eYFP+ cells in E6.5 and E7.5 cultures grown for 24hr (Fig. S1C). Comparing these data with E7 and E8 skins, we found that the feather buds continued to grow and H2B-eYFP+ cells moved towards the developing feather buds in both the in vivo and ex vivo skin models.

Figure 3. Time-lapse images of H2B-eYFP+ cells in developing avian skin.

(A-D). Frames from the dynamic imaging of the explant cultures. Skin specimens sampled from TIE1p.H2B-eYFP quails of different ages are shown (A, E5.5; B, E6; C, E6.5 and D, E7). In panels A-C, skin explants are cultured for about 15 hours and were imaged by focusing on the epithelia. Single frames at 3-hour intervals are shown. In panel D, the E7 skin explant development is recorded for 15 hours after culture for 0 (D0), 1 (D1) and 2 (D2) days. The images presented were focused on the feather bud level. The timing of the images is listed to the left of each picture. In each panel, a schematic drawing shows the approximate position of the field of view (FOV) on the embryonic skin. The 4 movies from which these still images are derived are presented as movies 1-4. Scale bar: 200 μm.

In addition, we used live imaging to further characterize the dynamics of transgenic quail H2B-eYFP endothelial cell patterning during early skin morphogenesis. We were able to capture these movies for about 15 hours before the field lost focus due to continued skin explant growth. The data confirmed that eYFP+ cell movement is consistent with our earlier histological evidence. In the E5 +15hr movie, disseminated H2B-eYFP⁺ cells in the lateral trunk migrate both medially and laterally (Fig. 3A; n=3; Movie 1). Cells moving towards the midline move in lines, which might be the path of future vasculature. In the E5.5 +15hr movie, the first row of feather primordia form. The H2B-eYFP⁺ cells inside the feather primordia further proliferate and move towards other primordia forming vascular connections (Fig. 3B; n=3; Movie 2). In the E6 +15hr movie, H2B-eYFP⁺ cells in feather primordia are organized into a polygonal pattern. The vasculature in primordia that developed earlier are more compact than those within newly formed primordia (Fig. 3C; n=3; Movie 3); ECs in each feather primordia build connections with other feather buds. E7 skin was cultured for 0, 1 or 2 days prior to observing them after culture for an additional 15 hours (Fig. 3D; n=3; Movie 4). In the E7 +2D movie (D2+ 0hr, D2+ 15hr) feather buds elevate from the skin surface, H2B-eYFP⁺ cells in feather buds develop in accord with feather bud shape changes and melanocytes are seen at the feather epithelium.

2.4. Avian skin vascular network pattern forms in a feather epithelium-dependent manner

To search for clues for vascular assembly, we used a skin explant model which enables us to evaluate epithelial-mesenchymal interactions. For this purpose, we carried out 2 sets of epithelium-mesenchyme recombination experiments. In the first experiment, we separated the epithelium from the mesenchyme of the skin by gently peeling the epithelium away at HH29-30 (Fig. 4A). If the feather bud cells are essential for forming a feather-vessel network, we expect that loss of the feather bud epithelium would disrupt the formation of a normal vascular network in the dermis. In agreement with this concept, the H2B-eYFP⁺ cells in the control group, in which the epithelium was separated and then replaced, would form a polygonal feather pattern after 24hr of culture; whereas in the experimental group without the epithelium, the polygonal pattern of H2B-eYFP⁺ cells in feather primordia at E6.5 gradually dispersed, suggesting epidermis is essential for this vascular formation (Fig. 4B; n=5).

Figure 4. H2B-eYFP+ cells pattern is oriented by the epithelium.

(A) (C). Schematics demonstrating the manipulations performed on the skin explants. A, anterior; P, posterior.

(A) The epithelium in the skin is indicated in white, while the dermis is in grey. For the “no epithelium” condition, the epithelium is removed from the dissected skin and the remaining dermis is cultured on the insert. (C) For the “recombination” condition, the epithelium is first separated from dermis but then rotated 90 degrees and placed back onto the dermis. The recombinant skin is then cultured on the insert.

(B). Skin explants before and after 24-hr culture. The control sample is an intact piece of skin. The “no epithelium” sample is a dermis-only skin. The first rows are brightfield images and the second rows are EYFP images from the same FOV. Scale bar: 300 μm.

(D). The recombinant skin explant cultured for 24 hours. eYFP is shown in green. An enlarged view of the boxed region of the explant is shown on the right. The original orientation of the epithelium is turned 90 degrees counterclockwise relative to the dermis. Scale bar: 300 μm.

(E). ISH staining of embryonic quail sections for FGF2, VEGF and Noggin at E6-E8. Scale bar: 500 μm.

(F). Schematic diagram of gene expression patterns for growth factors FGF/VEGF and Noggin.

In the second experiment, we replaced the skin epithelium after its separation from mesenchyme. However, in this scenario, the epithelium was rotated 90° relative to the dermis (Fig. 4C; n=5). If the feather bud cells are indeed guiding the vascular progenitor cells, epithelial-mesenchymal interactions after this rotation would form the vascular network but might be oriented along the anterior-posterior axis of the epithelium. Indeed, the H2B-eYFP⁺ cells were redirected to the orientation of the epithelium and not the mesenchyme (Fig. 4D; n=5). Together, these results support that early skin development also fits into our model of adaptive patterning, and the development is epithelium-dependent.

2.5. Expression of pro-angiogenic factors in developing feather bud epidermis

Since feather bud cells interact with and may guide the organization of Tie+ ECs, we decided to map molecules that might mediate these cellular interactions using in situ hybridization (Fig. 4E, F; n=3). Here, based on the literature we chose to reveal the expression patterns of FGF2, VEGF and Noggin in E6-E8 chicken embryo skin to avoid having quail pigmentation obscure our view.

FGF2 was shown to promote endothelial cell survival while VEGF promoted both endothelial cell survival and the formation of primitive vasculogenesis (Kazemi et al., 2002). Furthermore, an inhibitor of FGFR blocked the recruitment of endothelial precursor cells during vasculogenesis (Fons et al., 2015). In the avian skin, FGF2 was expressed throughout the epithelium at E6 and became enriched in feather primordia epithelium at E7. By E8 it was enriched in the bud epithelium and posterior dermis.

Vascular endothelial growth factor (VEGF) and its receptor (VEGFR) play important roles in vasculogenesis (Apte et al., 2019; Wang et al., 2020). Drugs targeting VEGF or VEGFR (ie., Bevacizumab, Ramucirumab) have shown success as anti-cancer therapeutic agents (Ribatti et al., 2021). VEGF was expressed uniformly in the bud epithelium at E6. It was enriched in the feather primordia epithelium at E7. It then was expressed in the anterior bud epithelium and distal dermis at E8.

While BMP increases the number of blood vessels, Noggin reduces their numbers by inhibiting VEGFR-2 (Nimmagadda et al., 2005). At E6 Noggin was expressed in the epithelium. By E7, it was enriched in the primordia epithelium. At E8 Noggin was expressed throughout the epithelium and was also expressed in the posterior bud dermis.

2.6. Perturbation of vascular formation locally in feather buds disrupt normal bud morphogenesis

Because both VEGF and FGF2 are strong pro-angiogenic factors in development (Laddha and Kulkarni, 2019), we tested whether adding VEGF and FGF2 proteins can promote local H2B-eYFP⁺ cell aggregation. For this test this, we used agarose beads to deliver the angiogenic factors to the skin explant at E7. Beads coated with VEGF and FGF2 or with BSA as a control were placed on top of the explant and co-cultured for 1 day (Fig. 5A). Our data shows that both VEGF and FGF2 coated beads caused H2B-eYFP⁺ cells to aggregate under the beads (Fig. 5B). The effect is mostly local as the cells going toward neighboring feather primordia were attracted to the beads instead (Fig. 5B, the second row; n=5), while feather primordia distant from the beads still retained H2B-eYFP⁺ cells. Together, these results suggest that the aggregation of H2B-eYFP⁺ cells to feather primordia is caused by the local release of VEGF and FGF2 from the dermal cells, leading to the future formation of inter- and intra-feather vessel networks.

Figure 5. VEGF, FGF2, and noggin functional studies.

(A). Schematic diagram of the placement of control (BSA) or FGF/VGF soaked beads at day 0 and of their effects on feather growth after one day in culture.

(B). Skin explants 24 hours after growth factor-soaked agarose beads are laid on top of skins. The skins are sampled from E7 TIE1p.H2B-eYFP (green) quails. The agarose beads appear as dark circles in the EYFP channel (indicated by red arrows). The lower row in each column shows a magnified view of the highlighted region in the top row. Scale bar: 300 μm.

(C). Whole-mount and sectioned skins with Noggin overexpression. The RCAS virus directing the expression of Noggin was injected into E2 embryos, and the skins were sampled at E14. Scale bar: 200 μm.

(D). Immunostaining for the presence of Fibronectin and Flk1 in control and Noggin expressing samples. Noggin causes blood vessel formation resembling a plexus and increased ECM expression. This suggests an imbalance of epithelial to mesenchymal components leads to an expansion of mesenchymal cells within the feather pulp.

We next perturbed these genes in vivo for functional verification and characterization. For this purpose, we used the RCAS retrovirus system to deliver Noggin into early-stage chicken embryos. When the chicken skin developed to E12-E14, follicle formation and the beginning of feather keratinization were observed in non-viral affected areas. RCAS-noggin infected regions showed abnormal vasculogenesis. The affected feathers were enlarged and balloon shaped (Fig. 5C; n=5). These results reveal that the vasculature-feather interactions are mediated by VEGF/FGF/Noggin related pathways and can affect both vasculogenesis and feather morphogenesis.

To further explore the effects of Noggin, we immunostained the RCAS-perturbed skin with antibodies to fibronectin to analyze changes in extracellular matrix and FLK1, the VEGF receptor, to assess its involvement in increasing blood vessel formation (Fig. 5D; n=5). In control skin, fibronectin was seen in the epithelium and dermis outside of the feather follicle. Fibronectin was highly up-regulated within the expanded feather bud regions of ectopic Noggin expression. FLK1 was only expressed around the central blood vessel of the control feather but was significantly increased within the epithelium and pulp of the RCAS-Noggin skin.

2.7. Vasculogenesis or angiogenesis? Transcriptome analyses reveals mesenchymal cell fate plasticity in the developing avian skin

We next sought to determine whether the vasculature in the dorsal skin originated from angiogenesis whereby new vessels are produced by branching from existing blood vessels or if they form from interactions with the local dermis. Interestingly, we observed that eYFP-Tie1+ cells light up simultaneously in most parts of the dermis and later form a vascular network. This phenomenon suggests that the intra-feather vessels result from vasculogenesis, and interfeather vessels may be the result of angiogenesis. To explore this further, we used RNA-Sequencing (RNA-seq) to track vascular cell origins. Vascular cells in the aorta arise during the 1^st neovascularization wave at HH11 (E1.5-2); whereas those in skin dermis develop between E5.5 and E7 (Sato et al., 2010). If the H2B-eYFP⁺ cells arise from dermal stem cells, their transcriptome profile should resemble those of other skin dermal cells; however, if they are derived from the extension of existing blood vessels, their profile should resemble H2B-eYFP⁺ cells in the existing vasculature, such as the aorta.

Aorta and skin specimens were collected from Tg(TIE1p.H2B-eYFP) quails at E6 and E7 and analyzed by bulk RNA-seq (n=2 biological samples per time point). We identified 4243 annotated differentially expressed genes (DEGs) between dermis and aorta H2B-eYFP+ cells (adjusted p-value (padj) <0.05, total n=4243). We then used FACS sorting to separate eYFP+ from eYFP− cell (Fig. 6A; S2A; n=4). Both principal component analysis (PCA) and Euclidian distance (heatmap of the sample-to-sample distances) of RNA-seq data showed that dermal H2B-eYFP⁺ cells are closer to dermal H2B-eYFP⁻ cells than aorta H2B-eYFP⁺ cells (Fig. 6B-D; S2B). In addition, the E6 and E7 aortic H2B-eYFP+ cells show greater similarity in transcriptomic profile compared to their H2B-eYFP− counterparts, suggesting that the aortic blood vessel cells have reached a relatively stable state with minimal undergoing differentiation, which is consistent with our presumption. In contrast, the dermal H2B-eYFP+ cells show greater similarity to their dermal H2B-eYFP− counterparts, suggesting that the H2B-eYFP+ cells were derived from the local H2B-eYFP− cells and the dermal cells are still actively differentiating. Therefore, the similarity between dermal H2B-eYFP⁺ and H2B-eYFP⁻ cells is not due to faulty FACS sorting (Fig. S2A). The Tie1 gene expression level (Fig. 6B, C) recovered from RNA-seq data further confirmed that our FACS process was valid. Together, these data suggest that the transcription profile of dermal vascular cells are more similar to the local dermal non-vascular cells, compared to well-differentiated vascular cells. This finding suggests they are not migrating or budding from existing vasculatures; instead, they more likely are induced and differentiated locally. In other words, the Tie1+ dermal cells should be the result of vasculogenesis rather than angiogenesis.

Figure 6. RNA-seq analysis of skin and aorta neovascularization at E6 and E7.

(A). Workflow of the RNA-Seq experiment. First, skin (dermis) and aorta are separately dissected from E6 or E7 TIE1p.H2B-eYFP quails. Then, cells are dissociated from the tissues and sorted into eYFP-positive and −negative groups by FACS. mRNA from eYFP+ and eYFP− cells are extracted and sequenced separately. Thus, 8 sample types were generated based on their age, tissue type and Tie1 expression.

(B). An example of FACS sorting result of E6 and E7 quail cells. The scale bar to the right shows Tie1 gene expression levels in the eYFP+ and eFYP− samples. In the labels to the right, T means Tie1+ samples while the remainder are Tie1− samples.

(C) Venn diagram showing differentially expressed genes in the transcriptomes of E6 and E7 Tie+ samples (ECs) versus aorta cells (E6T_DvsA and E7T_DvsA) and Tie1− dermis (dermal cells) versus aorta samples (E6_DvsA and E7_DvsA) . The gene count in each sector of the Venn diagram is color coded (see the legend to the right of the diagram).

(D). Cluster analysis shows differentially expressed genes by comparing E6 and E7 Tie+ and Tie− dermis vs aorta samples.

3. Discussion

Avian skin is an excellent model in which to study tissue patterning (Chang et al., 2019). We propose that the integument pattern is built stepwise. First, feather primordia form from the non-patterned morphogenic field by de novo patterning following Turing principles (Inaba et al., 2019; Jung et al., 1998; Turing, 1952). Second, the feather dermal muscle network is patterned using feather primordia as reference points for further mechanical force-dependent assembly in the process we named “adaptive patterning” (Wu et al., 2019). To investigate intricate skin vascular network assembly, we systematically characterized how blood vessels are initiated at an early developmental stage within feathers since such data hasn’t been previously reported. The Tg(TIE1p.H2B-eYFP) quail provides a powerful approach to study the early vascular formation process (Sato et al., 2010). Using this model, we demonstrate the spatiotemporal vascular pattern formation in avian skin both before and after circulation development. We observe a transition of skin vasculature from irregularly disseminated H2B-eYFP+ cells to an interconnected and organized tubular structure.

While observing skin morphogenesis and cell-cell interactions with dynamic imaging using the explant model may not be ideal, it is the most pragmatic approach. Using timelapse imaging of the developing avian skin explants, we systematically characterize skin neovascularization from HH26 – HH35 using both the quail and chicken models. We show that the explant model can represent in vivo conditions, although it does show a slightly slower developmental process. By capturing cell behavior using movies over several 24-hour intervals, we observe many important features of dermal neovascularization. For instance, the aggregation of H2B-eYFP⁺ quail ECs in feather primordia, their attempt to form connections between feather buds around E6, and the transformation of a previous capillary network to a branching structure around E7 (Honda and Yoshizato, 1997). Our study using this model demonstrates a transition of avian skin vasculature from an irregularly disseminated pattern to an organized tubular structure from HH26 to HH35 (Fig. 7A). This knowledge can serve as a framework for future studies and help us gain insights to build and vitalize skin grafts with blood flow in translational settings.

Figure 7.

A. A summary schematic for neovascularization and feather development. B. Contributions of the roles of VEGF, FGF2 and Noggin in vasculogenesis.

Besides characterizing vasculature formation, we note an interesting observation in the developing skin that may be worth future investigation. The enrichment of H2B-eYFP⁺ skin cells within this tissue made us wonder if the H2B-eYFP⁺ skin cells are locally induced or are merely migrating through this layer. In an image of the E5.5 Tg(TIE1p.H2B-eYFP) quail lateral view, the H2B-eYFP⁺ cells show a linear pattern between vertebrae but the skin has not yet developed at this early stage. It is also interesting to ask if the intervertebral H2B-eYFP⁺ cells under the forming skin directly contribute to the skin H2B-eYFP⁺ cells given their obvious spatial proximity. To the best of our knowledge, this specialized tissue layer has never been reported in the literature and its identity and function(s) are a total mystery. We speculate that it is the avian analog to the interstitium, which is a novel reticular patterned tissue in humans that exists in the submucosae of the gastrointestinal tract, urinary bladder, the dermis, the peri-bronchial and peri-arterial soft tissues, and fascia (Benias et al., 2018). This tissue layer first appears in the cranial and caudal parts of the body and later appears in the middle part of the back skin. It is interesting to ask whether forming the middle section is the result of migration and fusion of the cranial and caudal sections or is the result of late onset in situ development.

ECs form new vasculature structures that interact with their local environments to guide the differentiation of specific organ types (Witjas et al., 2019). Epithelial-dermal tissue recombination studies show that the separation of the epithelium and dermis diminished the vascular pattern. Also, rotating the epithelium resulted in changing the H2B-eYFP⁺ cell growth axis. Moreover, vascular cells were found to play an important role in pattern formation during organogenesis. Hepatic growth would fail in the absence of ECs (Matsumoto et al., 2001). In the pancreas, ectopic vascularization can lead to pathological conditions such as islet hyperplasia and ectopic insulin expression (Lammert et al., 2001). In mice, signaling from the vasculature is found to regulate hair follicle stem cell activation to begin the next hair cycle (Li et al., 2023).

We map the expression of proangiogenic factors, FGF2 and VEGF and related molecules which have long been shown to be highly expressed in endoderm during gastrulation (Riese et al., 1995). These factors, their receptors and modulators are critical for growth and morphogenesis of angioblasts into the initial vascular pattern (Poole et al., 2001), and are sufficient to induce ectopic vessel formation just like the endoderm (Cox and Poole, 2000). We find that FGF2 and VEGF are initially highly expressed in the epithelium, then in localize to feather placodes within developing feather primordia. Release of FGF2 or VEGF from beads is able to induce or promote the migration and patterning of H2B-eYFP⁺ cells in the skin. Ectopic noggin expression from RCAS (replication competent avian sarcoma virus) vectors dramatically interfere with the neovascularization process via the VEGF and FGF signaling pathways and show that feather morphogenesis is impaired in conjunction with higher vascularity in infected regions. The loss of VEGF in quail embryos was previously shown to result in EC aggregation and lack of polygonal structures for the primary vascular plexus (Drake et al., 2000). Such phenotypes show great resemblance with what we observe when overexpressing RCAS-noggin. Our findings agree with previous literature supporting the essential role of FGF2 and VEGF for proper patterning vascular cells in avian skin and attracting ECs to feather placodes.

Our transcriptomic analyses favor vasculogenesis when considering the developmental process that forms skin vasculatures. Many studies on different tissues have shown that vascular endothelium differentiates in a tissue specific manner, as capillary ECs could express an array of surface antigens selectivity for brain, ovary, lung, steroidogenic glands, ovary, testis, adrenal and placenta (Witjas et al., 2019)(Auerbach et al., 1985; LeCouter et al., 2001). Previous studies showed that skin lymphatic ECs (LECs) could derive from the cardinal veins and intersomitic veins (Srinivasan et al., 2007; Yang et al., 2012), or blood capillary plexus (Pichol-Thievend et al., 2018). Our PCA analysis showing a higher similarity between skin H2B-eYFP⁺ cells and H2B-eYFP⁻ cells than with aorta H2B-eYFP⁺ cells (Fig. S2C) also favors the involvement of vasculogenesis in skin neovascularization. Indeed, it is more plausible that vasculogenesis is the dominant mechanism of vascular morphogenesis during pre-circulation stage (Poole et al., 2001). In the aorta, the high expression of EC lineage markers in Tie1⁺ cells suggest a more committed cell fate. Whereas the dermal Tie1⁺ cells express TFs involved in multiple biological processes, suggesting cells with both differentiating and flexible cell fates may be contributing. The formation of vasculature from local cells also has been observed in tumors. For instance, in vascular mimicry, melanoma cells can transdifferentiate into ECs and together with other ECs form vasculature. After blood flow is restored, these cells can spread further in the blood stream (Ge and Luo, 2018). Exogeneous transcription factors could be added to reprogram fibroblast into endothelium (Cui et al., 2022) and fibroblasts could undergo reprogramming to transdifferentiate into ECs in response to wound healing (Cooke and Lai, 2023). Additionally, human dermal fibroblasts could be engineered to become ECs by the ectopic expression of ETV2, FLI1 and FOXC2 (Rincon-Benavides et al., 2023). However, here the fibroblast- endothelium transition we show is under a physiological process.

In summary, this study shows the skin vascular network is promoted by VEGF and FGF2. Blood vessels are assembled from ECs of two different lineages produced by neovasculogenesis from cells derived from the local dermis followed by angiogenesis to expand within individual feather buds, representing an example of convergent differentiation (Wolock et al., 2019). These findings are revealed only by the avian model here because of the unique possibility to examine developing skins from early embryos. On the other hand, the avian model is limited compared to available transgenic mouse lines and direct lineage tracing evidence is not available now. Progress is being made in transgenic bird technology and we hope those developments will help in the future. The findings in this study highlight the plasticity of mesenchymal cells and further characterizing the mechanism of mesenchymal cell fate determination may shed light for future applications toward skin regeneration after severe wounding.

4.0. Materials and methods

4.1. Experimental model

Fertilized pathogen-free (SPAFAS) chicken eggs were from and staged according to the method described by Hamburger and Hamilton (Hamburger and Hamilton, 1951). Fertilized wild-type Japanese quail eggs were from (Westminster, CA); fertilized Tg(TIE1p.H2B-eYFP) and Tg(TIE1p.H2B-eYFP) X Tg(PGKp.H2B-mcherry) quail eggs were from USC Quail Aviary and are staged according to Ainsworth et al. (Ainsworth et al., 2010). Eggs were incubated at a temperature of 38°C with 60-65% humidity with turning every 2 hours.

4.2. Specimen harvesting and processing

Chicken embryos staged HH36~44 and quail embryos staged HH26~35, were collected for sectioning, we removed the head, four limbs, internal organs, and feathers, and only the back area, from the spine and dorsal ribs, is preserved. The samples are fixed with 4% formaldehyde (Sigma) overnight at 4°C. For cryosection, we changed samples to phosphate buffered saline (PBS) and then 15% sucrose in PBS the next day. After samples sank to the bottom of the vial, samples were changed from 15% sucrose in PBS to 30% sucrose in PBS. And then the samples were embedded in mounting media (Tissue-Tek^R O.C.T^™ Sakura) and snap frozen on dry ice and stored at −80°C. For wholemount purpose, quail embryos staging from HH13~25 and skin samples were peeled off from the back area of HH26~35 quail embryos including dorsal-pelvic, apteric, and femoral tracts. And then fixed with 4% formaldehyde (Sigma) overnight.

4.3. Imaging

Imaging of wholemount embryos and skin samples was performed using a Nikon SMZ1500 or Olympus MVX10 or Leica TCS SP8 confocal microscopy. Sectioned samples were imaged with a Leica TCS SP8 confocal microscope or Keyence BZ-X700.

4.4. Explant Imaging Condition

Skin samples from quail embryos staging from HH28~35 were peeled off for dynamic analysis. We placed the embryonic skin on a culture insert in 6-well culture plate (Falcon) containing Dulbecco’ modified Eagle’s medium (DMEM) supplemented with 2% fetal bovine serum (FBS, Gemini #100-106), 10% chicken serum (CS, Sigma C5405) and Penicillin-Streptomycin (Gibco) (1:1000). The explants were incubated at 38 °C at an atmosphere of 5% carbon dioxide and 95% air (Jiang et al., 2023). Dynamic imaging was performed using an Olympus MVX10 or Leica TCS SP8 confocal microscopy or Keyence BZ-X700.

4.5. In situ hybridization

Gene-specific fragments were amplified (SuperScript III First-Strand Synthesis system, Invitrogen) from RNA extracted from the dorsal skin of quail and chicken embryos by using Trizol reagent (Invitrogen), and subsequently cloned into pDrive cloning vector system (Qiagen). PCR primers for the cDNA amplifications are listed below. Both sense and antisense RNA probes were made by in vitro transcription according to manufacturer’s instructions (Roche). Wholemount in situ hybridization was done according to the procedure described in (Chuong et al., 1996). Briefly, after fixation in 4% formaldehyde (Sigma) overnight at 4°C, samples were sequentially dehydrated with methanol. For section in situ hybridization, after dehydration with methanol, samples were embedded in paraffin wax (McCormick) and sections were prepared as 7um thickness. Hybridizations with probes were carried out overnight at 65 °C in the hybridization buffer, containing a cocktail of the digoxigenin-labeled RNA probes. After wash and blocking, color reaction was done with NBT/BCIP substrate (Promega).

Primers	Forward	Reverse	Product
VEGF	AGCGGAAGCCCAATGAAGTT	TCTTTGGTCTGCAGTCACATT	300bp
FGF2	AGCATCACCACGCTGC	GATTCCAAGCGCTCAAAAA	314bp

^*Other probe: noggin (Tzahor et al., 2003)

4.6. Retrovirus production and mis-expression

For RCAS production, retroviruses were cultured and harvested as described (Morgan and Fekete, 1996). RCAS-noggin was a gift from Dr. R. Johnson (Johnson and Tabin, 1997). The virus was injected into E3 chicken embryos. The presence of virus was detected by in situ hybridization using a probe to the viral polymerase gene (Crowe et al., 1998).

4.7. Epidermis removal and epidermis-dermis recombination

The separation of epidermis and dermis was performed according to the procedure described in (Chuong, 1998). Briefly, dissected quail skin samples were incubated in 2X calcium-magnesium free saline with 0.25% EDTA on ice for 10-15 minutes. Epidermis and dermis were separated by forceps carefully. For recombination, epidermis was rotated 90° and recombined on top of dermis. The recombinants were incubated in the culture insert.

4.8. Isolation of H2B-eYFP^+/− cells, immunofluorescence staining and FACS

After removal of epidermis as described in previous paragraph, dermis was broken up into small pieces with forceps. After dissected out from the circulation system, aorta was broken up into small pieces, too. Samples were incubated in 0.1% trypsin and 0.1% collagenase in calcium-magnesium free Hanks’ Balanced Salt Solution in 37°C water bath for 5 minutes. The cell suspension was agitated and inactivated with 100% FBS. The cell suspension was then diluted with DMEM and filtered through a 0.7um cell strainer. The filter cells were spinned down at low-speed centrifugation (300g, 1221rpm) at 4°C for 15 minutes. After removal of the supernatant, cells were resuspended and filtered through a 0.4um cell strainer into a FACS tube.

4.9. Immunofluorescence staining with DAPI and Draq5

The filtered cells were resuspended with DAPI (1:100,000) and Draq5 (1:2500) in staining buffer (2% FBS in PBS).

4.10. FACS

FACS was done using ARIA I and program. The excitation was available from 405nm (for measurement of DAPI), 510nm (for measurement of eYFP) and 633nm (for measurement of Draq5). FSC (Forward-Scattered) versus SSC (Side-Scattered) plot was run first to adjust the voltage on each detector. And then DAPI and Draq5 was run to ensure the viability of each cell sample. The compensation was adjusted between channels to eliminate overlap in the florescence signal. Finally, the eYFP gate was defined and H2B-eYFP⁺ and H2B-eYFP⁻ cells were collected into microfuge tubes containing lysis buffer RA1 and reducing agent TCEP (NucleoSpin RNA XS, Takara Clontech #740902.50) for RNA extraction. The samples were kept at −80 °C preparing for RNA extraction.

4.11. Total RNA isolation

After thawing, the samples were homogenized and total RNA extracted using NucleoSpin RNA XS kit (Takara Clontech #740902.50). The 15min rDNase treatment was done at room temperature as described in the manual to remove the DNA thoroughly.

Sample	Sequencing ID (biological repeats)
E7 dermis Tie+	695, 699
E7 dermis Tie−	696, 700
E6 dermis Tie+	675, 737
E6 dermis Tie−	676, 738
E7 aorta Tie+	697, 701
E7 aorta Tie−	698, 702
E6 aorta Tie+	677, 739
E6 aorta Tie−	678, 740

4.12. Stranded RNA sequencing

Total RNA concentrations from sixteen samples (eight samples with two biological repeats) were measured by NanoDrop Lite Spectrophotometer (Thermo Fisher Scientific), and quality was assessed by BioAnalyzer 2100 RNA Pico kit (Agilent). The samples had RNA integrity number (RIN) values ranging from 8.5 to 10. Ten nanogram of total RNA from each sample was used for library construction using the SMARTer Stranded Total RNA-Seq Kit v2 - Pico Input Kit (Takara Bio), following the manufacturer’s instructions. Deep sequencing of single-end 75 nt was carried out on NextSeq500 (Illumina). The Illumina sequencing was conducted by USC Molecular Genomics Core.

Libraries	Raw reads	Processed reads	Mapping rate
Chuong675	29,681,072	28,220,335	73.83%
Chuong676	29,477,793	27,983,967	73.49%
Chuong677	27,607,414	26,215,226	72.31%
Chuong678	27,570,230	26,163,957	72.62%
Chuong695	30,754,020	30,072,354	78.98%
Chuong696	24,591,999	24,079,882	78.15%
Chuong697	27,504,374	26,935,623	79.73%
Chuong698	30,666,419	29,985,709	78.50%
Chuong699	26,800,723	26,234,859	78.63%
Chuong700	27,317,784	26,736,640	78.36%
Chuong701	31,765,782	31,071,406	76.86%
Chuong702	30,559,056	29,870,444	74.86%
Chuong737	29,087,674	28,190,085	74.77%
Chuong738	27,767,609	26,911,764	75.95%
Chuong739	22,158,973	21,462,867	73.36%
Chuong740	36,018,724	34,812,579	73.68%

4.13. RNA-Seq data analysis

Low-quality bases and reads were removed by using Trimomatic (Bolger et al., 2014) according to the following procedure: 1) Remove adaptors; 2) remove leading low quality bases below Q score 15; 3) remove trailing low quality bases below Q score 15; 4) scan the read with a 4-base wide sliding window, cutting when the average quality per base drops below 15; and 5) drop trimmed reads below 36 bases long. The latest Japanese quail genome assembly (version Coturnix_japonica_2.0) and its annotation file were downloaded from Ensembl FTP. The processed reads were mapped to the genome using HISAT2 (Kim et al., 2015) with the parameter: --score-min L,0,−0.7. The statistics of processed and mapped reads were listed in the table below. The read counts for each gene were obtained from the mapping files using StringTie (Pertea et al., 2016, 2015) with the default parameter and the genome annotation file.

4.14. Identification of differentially expressed genes (DEGs)

We identified the DEGs through several sets of comparisons. Set 1: Gene expressions between dermis and aorta were compared. Set 2: Gene expressions between dermis and aorta in Tie1+ samples were compared. Set 3: Gene expressions between Tie1+ and Tie1− in dermis samples were compared. Set 4: Gene expressions in E6 Tie1+ versus those in E6 and E7 Tie1− dermis samples were compared. Set 5: Gene expressions between E6 and E7 in Tie1+ dermis samples were compared. Set 6: Gene expressions between Tie1+ and Tie1− in E6 dermis samples were compared. Set 7: Gene expressions between Tie1+ and Tie1− in E7 dermis samples were compared. Set 8: Gene expressions between E6 and E7 in Tie1+ dermis samples were compared. Set 9: Gene expressions between E6 and E7 in Tie1− dermis samples were compared. Set 10: Gene expressions between E6 and E7 in Tie1+ aorta samples were compared. Set 11: Gene expressions between E6 and E7 in Tie1− aorta samples were compared. Set 12: Gene expressions between Tie1+ and aorta dermis samples were compared. Set 13: Gene expressions between Tie1− and aorta dermis samples were compared. Set 14: Gene expressions between Tie1+ and Tie1− in aorta samples were compared. These two replicates were compared with each other. The DEGs from the comparisons were computed by DESeq2 (Love et al., 2014). Only the genes with q value < 0.05; fold change > 1 were defined as DEGs.

Supplementary Material

1

Supplementary Figure 1. The explant culture system can be used for dynamic imaging.

(A). Schematics of the explant culture system for dynamic imaging. In brief, whole-mount skin is dissected, flattened, and cultured on a porous insert, at the air:culture media interface.

(B). E5 skin explant from TIE1p.H2B-eYFP quails cultured over time. The white dashed lines indicate the midline of the skin. Scale bar: 500 μm.

(C). Skin explants from TIE1p.H2B-eYFP quails before and after 24 hours in culture. The highlighted regions are enlarged in the panels below. Both brightfield and EYFP images are shown. Skins are sampled from E5.5, 6.5 or 7.5 quail embryos. Top row, scale bar: 2000 μm; second row, scale bar: 200 μm.

2

Supplementary Figure 2.

(A). Results of Flow cytometry analyses. Population P4 (orange) were collected as EYFP+ cells while population P5 (pink) were EYFP−. Cells in population 3 (blue) were not clearly distinguished from the surrounding populations and were not collected. The number of EYFP+ and EYFP− cells are indicated in the right panel.

(B). PCA analysis shows the similarity of overall RNA expression between replicate samples. Samples from skin dermis are shown as triangles; samples from aorta are shown as circle dots. Tie1+ samples are in red; Tie1− samples are in blue. The dash lines in the graph indicates the possible grouping scheme of the data points.

3

(A). Movie 1. An E5 TIE1p.H2B-eYFP quail skin explant shows early stages of vascularization. ECs move toward feather primordia that are forming on either side of the midline of the dorsal skin. The vasculature then begins to extend along the Anterior-Posterior axis. This video was recorded for 15 hours with a 10-minute interval between frames.

4

(B) Movie 2. An E5.5 TIE1p.H2B-eYFP quail skin explant shows the ECs forming around the newly forming feather primordia. This can be seen along the initial 2 central rows plus the initiation of the adjacent feather bud primordia. This video was recorded for 15 hours with a 10-minute interval between frames.

5

(C) Movie 3. An E6 TIE1p.H2B-eYFP quail skin explant more clearly shows expansion of the vasculature within the forming feather buds. Note the seven spacing between adjacent buds. The vasculature begins to extend in all directions from the buds. This video was recorded for 15 hours with a 10-minute interval between frames.

6

(D) Movie 4. An E7 TIE1p.H2B-eYFP quail skin explant shows expansion of the vasculature within the growing feather buds. Sparse vascular cells form between the buds but are not shown in this video. This video was recorded for 2 days with a 5-minute interval between frames.

Highlights.

• Skin vasculature forms from peripheral feather bud endothelia and the central vessel

• Skin vascular network patterning is epithelium-dependent, mediated by FGF and VEGF

• Noggin misexpression disrupts normal skin vasculature formation

• RNA-seq analysis shows skin endothelia are derived from dermal cells, not central endothelia

Data Availability

Upon publication, the datasets generated for this study will be deposited in the Gene Expression Omnibus (FEO) database.