Alk1 acts in non-endothelial VE-cadherin+ perineurial cells to maintain nerve branching during hair homeostasis

Chovatiya, Gopal; Li, Kefei Nina; Li, Jonathan; Ghuwalewala, Sangeeta; Tumbar, Tudorita

doi:10.1038/s41467-023-40761-5

Download PDF

Article
Open access
Published: 12 September 2023

Alk1 acts in non-endothelial VE-cadherin⁺ perineurial cells to maintain nerve branching during hair homeostasis

Nature Communications volume 14, Article number: 5623 (2023) Cite this article

1953 Accesses
16 Altmetric
Metrics details

Subjects

Abstract

Vascular endothelial (VE)-cadherin is a well-recognized endothelial cell marker. One of its interacting partners, the TGF-β receptor Alk1, is essential in endothelial cells for adult skin vasculature remodeling during hair homeostasis. Using single-cell transcriptomics, lineage tracing and gene targeting in mice, we characterize the cellular and molecular dynamics of skin VE-cadherin⁺ cells during hair homeostasis. We describe dynamic changes of VE-cadherin⁺ endothelial cells specific to blood and lymphatic vessels and uncover an atypical VE-cadherin⁺ cell population. The latter is not a predicted adult endovascular progenitor, but rather a non-endothelial mesenchymal perineurial cell type, which forms nerve encapsulating tubular structures that undergo remodeling during hair homeostasis. Alk1 acts in the VE-cadherin⁺ perineurial cells to maintain proper homeostatic nerve branching by enforcing basement membrane and extracellular matrix molecular signatures. Our work implicates the VE-cadherin/Alk1 duo, classically known as endothelial-vascular specific, in perineurial-nerve homeostasis. This has broad implications in vascular and nerve disease.

Simultaneous single-cell three-dimensional genome and gene expression profiling uncovers dynamic enhancer connectivity underlying olfactory receptor choice

Article Open access 15 April 2024

CGRP sensory neurons promote tissue healing via neutrophils and macrophages

Article Open access 27 March 2024

Spatially organized cellular communities form the developing human heart

Article Open access 13 March 2024

Introduction

Vascular endothelial (VE) cadherin (encoded by Cdh5) has been traditionally known as an endothelial cell (EC) specific marker, expressed by both blood vessels (capillaries, veins, and arteries) and lymphatic vessels¹. The encoding gene and mRNA are referred to as Cdh5 and the protein product as VE-cadherin, a convention we follow throughout our paper. VE-cadherin is a cell-cell contact molecule, recognized for its essential role in the formation and maturation of vascular tubes^1,2. It also contributes to signaling within the ECs by interacting with receptors of the VEGF, Notch, FGF, and the TGF-β pathway, including Alk1, a TGF-β receptor expressed by ECs³. Its molecular interactions ensure vasculature resilience, flexibility, mobility, and permeability, thus endowing ECs with their specific characteristics^3,4,5,6. Nonendothelial tumor cells can also aberrantly activate VE-cadherin, forming tubular irrigation networks that assist tumor growth⁷. Single-cell (sc) RNA-seq analysis of ECs from multiple adult organs (but not skin) uncovered significant tissue-specific gene expressions^8,9,10, but generally identified the same type of mature ECs. Intriguingly, single-cell transcriptomics in the adult mouse dorsal aorta uniquely identified a distinct Cdh5⁺ cellular subset proposed to be an immature endothelial vascular progenitor (EVP)¹¹. A VE-cadherin⁺ EVP was also suggested to arise in skin wounds and tumors^12,13, but it has not yet been reported in normal skin. To date, single-cell transcriptomics have not yet characterized non-EC types within the normal VE-cadherin expressing lineage. In addition, there is little knowledge of the dynamic changes in VE-cadherin lineage heterogeneity and molecular makeup during adult tissue homeostasis, in the absence of injury or cancer.

Adult skin and hair follicles are an excellent model system to study the organization of adult VE-cadherin⁺ lineages during tissue homeostasis. Skin is a highly regenerative tissue that requires an ongoing supply of O₂ and nutrients. The hair follicles, made largely of epithelial cells, undergo periodic transitions from resting/quiescence (telogen) to growth (anagen) phase during the hair homeostatic cycle. These phases are synchronous in mice for the first cycle, providing an opportunity to access homeostatic processes that are generally more hidden in less synchronous adult tissues. At anagen, the skin hypodermis significantly expands in size and becomes filled with engorging adipocytes and blood vessels that wrap around the downward growing hair follicle bulbs, while the lymphatic vessels also remodel their caliber and change their drainage capacity^14,15 (Fig. 1a). Alk1, the known VE-cadherin interacting partner, acts in all skin ECs to regulate their migration and the vasculature remodeling during hair cycle¹⁶. Other skin structures, such as nerves and sub-cutaneous muscles, may also be remodeled during the hair cycle¹⁷. A thorough characterization of the cellular and functional heterogeneity of skin VE-cadherin⁺ lineage organization during adult skin homeostasis is currently lacking.

**Fig. 1: Endothelial and non-endothelial adult skin VE-cadherin-marked cells characterized by scRNA-seq.**

Here, we employed 10x Genomics scRNA-seq of sorted Cdh5-CreERT²;tdTomato marked cells isolated from adult mouse skin at two hair cycle stages: quiescence (telogen) and growth (anagen). This data combined with genetic lineage tracing and Alk1 gene targeting in specific skin cell types allows us to refine and redefine the cellular and functional heterogeneity of the VE-cadherin⁺ lineage during adult tissue homeostasis. We rule out a potential VE-cadherin⁺ endovascular progenitor (EVP) that may be responsible for vasculature growth during hair cycle. Moreover, we firmly expand the VE-cadherin lineage attributes and the function of Alk1 from vascular-endothelial into other cell types and functions, including perineurial-nerve remodeling. This work opens new avenues of investigation into vascular and nerve remodeling functions in adult regenerative tissues, with potential implications to nerve and vascular diseases.

Results

Labeling and isolation of adult skin VE-cadherin⁺ (Cdh5) cells documents turnover rates

Skin vasculature remodels during the hair cycle^18,19,20, as adult skin considerably expands in thickness from telogen to anagen²¹ (Fig. 1a). At anagen, the hair follicle grows downward into the hypodermis, which fills up with fat cells, while the blood vessels (BV) vasculature stretches out and nourishes tissue growth^16,22. A horizontal vascular plexus (CD31⁺/ VE-cadherin⁺) underneath the hair germ (HPuHG) made mostly of BV disperses out into the hypodermis at the telogen-anagen transition^16,18 (Fig. 1a). Whole mount skin imaging¹⁶ and 70 µm skin sections that were VE-cadherin stained or Cdh5-CreERT² x tdTomato (Cdh5-tdTomato⁺) labeled, followed by 3D confocal microscopy confirmed these findings (Fig. 1b and Supplementary Fig. 1a, b). Aside from BVs, the lymphatic capillaries also remodel by increasing in caliber and moving slightly away from the HFs^23,24.

We and others also reported an increase in Ki67⁺ or BrdU⁺ proliferative CD31⁺ EC at anagen by immunofluorescence of skin sections^18,19,20. To quantitatively evaluate the magnitude of VE-cadherin⁺ skin lineage proliferation in homeostasis, we performed BrdU labeling from postnatal (PD)17-25, encompassing the hair cycle transition from quiescence to growth. We used Cdh5-CreERT² x tdTomato (Cdh5-tdTomato⁺)^25,26 reporter mice induced with tamoxifen (TM) at PD17, followed by FACS isolation of tdTomato⁺ ECs at PD25 (Supplementary Fig. 1c, d). Confocal microscopy of skin sections confirmed both high BrdU labeling efficiency and the presence of BrdU⁺/VE-cadherin⁺ vascular cells in single optical images (Supplementary Fig. 1e). Analysis of FACS sorted tdTomato⁺ cells deposited on slides and stained with antibodies for BrdU and VE-cadherin (Cdh5) (Supplementary Fig. 1f) showed that ~80% of the sorted tdTomato⁺ were strongly VE-cadherin-positive by immunofluorescence (IF) staining, and of those ~15% incorporated BrdU during the 8-day labeling period (Supplementary Fig. 1g). This suggests that in the period that accompanies the transition from hair quiescence to growth ~ 1-2% ECs divide on average/day. Proliferation was reported to occur through early anagen and to peak at full anagen^18,19,20. Thus, we can estimate that at least one-third of skin VE-cadherin-expressing cells engage in proliferation during the first hair cycle in mice. This data quantitively documents the extent of cellular turnover in the VE-cadherin⁺ skin compartment during the hair cycle (Supplementary Fig. 1h).

scRNA-seq profiling of Cdh5-CreERT² marked skin cells uncovered VE-cadherin (Cdh5)⁺ endothelial and non-endothelial lineages

To understand the cellular heterogeneity and the dynamic molecular reorganization of VE-cadherin⁺ cells in the hair cycle, we performed 10x Genomics scRNA-seq analysis of Cdh5-CreERT² x tdTomato⁺; Krt14-H2BGFP^- cells purified from mouse dorsal skin at quiescence (telogen) and growth (anagen) stages (Fig. 1c–e and Supplementary Fig. 1a), using methods previously described²⁷. Two mouse replicates (S1 and S2) from each stage were subjected to 10x scRNA-seq library construction and sequencing (see Methods, GEO accession number GSE211381). After applying quality control parameters on individual samples (Supplementary Fig. 1i), we obtained a total of 7250 and 9262 high-quality cells at telogen and anagen, respectively. All samples were integrated using the Harmony package (see Methods), followed by Uniform Manifold Approximation and Projection (UMAP) analysis which produced 16 distinct cell populations from the tdTomato⁺/GFP^- cells, highly reproducible in all 4 mice analyzed (Fig. 1f and Supplementary Fig. 1j). Based on a panel of known lineage markers (Figs. 1g), 7 populations were ECs and showed robust Cdh5 expression (Fig. 1h). Previously characterized EC type-specific markers revealed distinct populations of artery, vein, capillaries, and lymphatics^28,29,30 (Fig. 1h). The capillaries (Cap) displayed two (Cxcl12⁺/CD36⁺) populations enriched in either Vegfr2 (Kdr) or Rgcc^9,28. The lymphatics were either collecting vessels (Lymph-Col, Prox1⁺) or capillaries (Lymph-Cap, Ccl21a, Pdpn and Lyve1⁺). The Ki67⁺ proliferative (Proli) EC population was ~1-2% of the total, which agrees with our BrdU data (Fig. 1f, g and Supplementary Fig. 1g).

Apart from these 7 endothelial populations, 6 populations were Cdh5 negative contaminating immune cells and fibroblasts, while 2 populations that expressed low levels of Cdh5 were vascular smooth muscle cells (vSMCs) or pericytes and schwann cells (Fig. 1h and Supplementary Data 1). Surprisingly, one population expressed medium Cdh5 levels and Krt19, previously reported as an exclusive HFSC marker³¹. This Cdh5⁺/Krt19⁺ population represented over 15% of all sorted cells (Fig. 1f), did not share expression of mature ECs and could not be classified as pericyte, schwann, or fibroblast cells. Together, these data provide high-quality transcriptomic profiles for adult VE-cadherin⁺ skin lineages at different hair cycle stages and identified known mature endothelial populations as well as a few not previously reported populations that were likely not endothelial.

VE-cadherin⁺ endothelial populations undergo dynamic redistribution and gene expression changes during the hair cycle

We first focused on the vascular EC clusters to probe in more depth the dynamics of cellular and molecular heterogeneity during the hair cycle. The 7 ECs populations from the VE-cadherin marked cells, represented 2543 (telogen) and 7498 (anagen) high-quality sequenced cells (Supplementary Fig. 2a) with high correlation between replicates (Supplementary Fig. 2b). Previous single-cell analysis suggested a tissue-specific marker expression of ECs⁹, thus we examined individual population enriched genes and identified skin-specific marker combination of ECs (Supplementary Fig. 2c, d and Supplementary Data 2).

UMAP analysis showed that all 7 EC populations that include the proliferative ECs are present at both telogen and anagen (Fig. 2a), with no emergence or loss of any EC type. Interestingly, the relative fraction of each population changed at anagen when compared to telogen, with a near doubling of blood EC populations at the expense of the lymphatic populations (Fig. 2b). To test for this change in the intact tissue, we co-stained skin sections at telogen and anagen with CD31 and either Endomucin (blood vessel - BV- marker) or Prox1 (lymphatic vessel - LV- marker). Using 8 µm thin sections to reduce cell overlap, we identified nuclei enclosed by CD31 signal and counted BV and LV EC density in both dermis and hypodermis (Supplementary Fig. 2e, f). The data showed a significant increase in BV EC density by ~45% at anagen, mainly in the hypodermis area (Fig. 2c, d). On the contrary, the LV EC counts showed decreased density at anagen mostly detected in the dermis (Fig. 2c, d). This in vivo validation by IF counts confirmed the population fractional change between the hair cycle rest and growth phases predicted by the scRNA-seq data.

**Fig. 2: Cellular and molecular dynamics in all endothelial populations at the telogen-to-anagen transition.**

The changes in BV/LV ECs density suggested either active conversions between these EC types or input of new ECs from an active progenitor. To gain more insight into the dynamics of the EC populations, we examined the gene expression changes during the two critical stages. Many genes changed expression (log2FC > 0.25) in each EC population at the telogen-anagen transition (Supplementary Fig. 3a). Differentially expressed genes (DEGs) (330 upregulated and 466 downregulated or 784 total unique genes >1.3x) extracted from the nonproliferative EC populations (Fig. 2e and Supplementary Data 3) revealed diverse functional categories via PANTHER³² analysis (Fig. 2f), while heatmap clustering revealed DEG changes from telogen to anagen (Supplementary Fig. 3b). Interestingly, the Cap-Kdr population clustered with the arteries and with the proliferative cells while the Cap-Rgcc population clustered with the veins (Supplementary Fig. 3b), suggesting potential sub-lineage relationships. Only a small fraction of 35 genes was upregulated in all EC populations at anagen, and those genes were involved in metabolism and oxidative phosphorylation, indicating increased cellular bioenergetics (Fig. 2g and Supplementary Data 3). On the other hand, Hallmark pathways³³ analysis of all detected genes show EC-population specific enrichment and changes between the stages (Fig. 2h, Supplementary Fig. 3c and Supplementary Data 4). Angiogenesis genes³⁴ were strongly enriched in arteries relative to the other clusters, along with related pathways, including Wnt, TGF-β, and TNF-α pathways (Fig. 2h and Supplementary Fig. 4), and these genes did not change at anagen. This observation, along with the BrdU incorporation at telogen/early anagen (Supplementary Fig. 1g), and the comparable percent of Ki67⁺ ECs at telogen and anagen (Fig. 2b) suggest that angiogenesis likely occurs in ECs at both stages, not just at anagen. Pseudotime analysis by Monocle³⁵ predicted the artery population as the most immature on the lineage trajectory, but a specific flow from one EC population into another was not clearly apparent (Supplementary Fig. 3d). EC populations other than arteries also showed changes in several signaling pathways previously implicated in vascular remodeling, such as TGF-β, NFK-β, PI3K-Akt-mTOR, Notch and MYC pathways (Fig. 2h). These were accompanied by changes in various gene ontology (GO) terms, which indicated overall vasculature growth and/or remodeling at anagen in all EC populations (Fig. 2i).

These data documented the EC heterogeneity in hair cycle and attested to the dynamic and intricate nature of cellular and molecular transformation that appear to implicate all EC populations during homeostasis (Fig. 2j). Despite significant shifts in the balance of EC types between stages, the analysis did not reveal a clear predicted flow of ECs from one population into another, nor did it uncover a common and highly active or changing immature vascular progenitor, as identified in skin injury and cancer¹². This prompted us to further characterize the remaining VE-cadherin⁺ lineages that did not clearly qualify as ECs, in search for a potential EC vascular progenitor.

VE-cadherin⁺ lineage contains a Krt19 ⁺ mesenchymal cell population outside the hair follicle

The remaining VE-cadherin (Cdh5) expressing populations that were not known ECs comprised of pericytes, schwann cells, and a previously unknown Cdh5⁺/Krt19⁺ population (Fig. 3a–c). VE-cadherin (Cdh5) levels were low when compared with the mature ECs (Fig. 3c), which may explain why previous analyses not based on the highly sensitive Cdh5-CreERT² x tdTomato labelling have missed this expression. Pericytes specifically expressed Des, Acta2, Rgs4 and Notch3, as expected, and displayed enrichment of metabolism, biosynthetic pathways, and growth-related features at anagen, similar with the EC populations (Supplementary Fig. 5a, b). The schwann cell population displayed the known markers expression pattern (Supplementary Fig. 5c), and unlike all other VE-cadherin populations, it remained unchanged in gene expression between the two hair cycle stages (Supplementary Data 5, adjusted p-val ns). The levels of VE-cadherin in this population were especially low and marked at most ~25% of cells in this population (Fig. 3e).

**Fig. 3: The *Cdh5*⁺*/Krt19*⁺ population identifies a non-hair mesenchymal VE-cadherin⁺ population.**

Finally, the Cdh5⁺/Krt19⁺ cluster expressed medium VE-cadherin (Cdh5) levels and captured a relatively large population that represents ~15% of all sorted cells (Fig. 3a). This population did not express CD31 (Pecam1) nor other mature EC markers (Fig. 1h), was highly distinct from all other Cdh5⁺ populations in UMAP analysis (Fig. 3b) and had its own defining set of marker expression (Fig. 3d, e). To probe the localization of this population in skin (and failing Keratin19 immunofluorescence staining), we employed the Krt19-CreERT; tdTomato mice induced with tamoxifen at PD17 and sacrificed at PD20 and PD49. As expected from previous studies³⁶, HFs showed prominent tdTomato⁺ labeling, but we also identified tdTomato⁺ signal in the dermis (Supplementary Fig. 5d). These non-HF structures stained clearly for VE-cadherin, though expression was weaker when compared with mature vasculature (Fig. 3f). GO analysis revealed a strong mesenchymal signature and enrichment in extracellular matrix (ECM) organization as well as structural assembly of an external encapsulating structure (Fig. 3g and Supplementary Fig. 5e). These features were common with some of the known Cdh5⁺ EC populations (Fig. 3h). In addition, many upregulated genes in the Cdh5⁺/Krt19⁺ population were uniquely associated with a mesenchymal identity that distinguishes these cells from known ECs (Supplementary Data 1) and are involved in adhesive functions such as collagen, heparin, and glycosaminoglycan binding (Fig. 3h). Together, these results confirmed the existence of a VE-cadherin (Cdh5)⁺/Krt19⁺ non-HF population in vivo, and revealed its strong and unique mesenchymal-type gene signature highly enriched in ECM secretion and binding.

Cdh5 ⁺/Krt19 ⁺ population forms tubular structures but does not contribute to mature vasculature

Having defined the molecular characteristics of the Cdh5⁺/Krt19⁺ population, we then interrogated its physiological function during skin homeostasis. We first wondered if these cells may be the reported immature endovascular progenitors (EVPs), with low VE-cadherin expression^11,12,13. Using a stringent set of genes specifically upregulated (FC > 4) in our Cdh5⁺/Krt19⁺ population relative to mature ECs, we found significant overlap with the reported aorta EVP signature (Fig. 4a) and markers (Fig. 4b and Supplementary Fig. 6a). Typical mature EC markers previously reported^11,12, such as CD31 and CD34 were downregulated in the Cdh5⁺/Krt19⁺ population (Fig. 4c and Supplementary Fig. 6b). Other observations consistent with a putative identity as EVPs of the Cdh5⁺/Krt19⁺ population were: an apparent fractional decrease from telogen to anagen in favor of the mature EC populations (Fig. 4d); differential gene expression changes in growth-associated genes (Fig. 4e); and suggestive Monocle lineage trajectory predictions placing the Cdh5⁺/Krt19⁺ population as the ground state (Fig. 4f and Supplementary Fig. 6c-d’). This prediction did not hold for a control fibroblast population (Supplementary Fig. 6e-f’).

**Fig. 4: *Cdh5*⁺/*Krt19*⁺ cells do not contribute into the mature vasculature in vivo.**

If the Cdh5⁺/Krt19⁺ cells were indeed skin EC vascular progenitors acting in the hair cycle, we reasoned they should eventually form mature vasculature. To test this, we employed lineage tracing using the Krt19-CreERT; tdTomato mice TM-induced at PD17 and stained the tissue sections at PD32 (anagen) and PD49 (telogen) with VE-cadherin, CD31 (Pecam1), CD34 and lymphatic capillary marker Lyve1 in n = 3 mice per stage. Excitingly, we did observe tdTomato⁺/VE-cadherin⁺ tubular structures resembling vasculature at both hair cycle stages analyzed (Fig. 4g, h). However, to our surprise these structures never stained positive for mature EC markers CD31, CD34, and Lyve1 (Fig. 4i–k) in >45 tdTomato⁺ tubular structures/staining/stage analyzed. Thus, our in vivo lineage tracing analysis did not support the computational prediction of Cdh5⁺/Krt19⁺ cells as a potential EVP subpopulation that may contribute to mature vasculature formation in adult skin during the hair cycle. This left unresolved the identity and function of the Cdh5⁺/Krt19⁺ population and of the tubular structures it formed.

Cdh5 ⁺/Krt19 ⁺ cells encapsulate neurofilament-positive nerve bundles

Recently, scRNA-seq of cells isolated from sciatic nerves were profiled and a Krt19⁺ population was reported as a putative perineurial population^37,38, although in vivo validation was lacking. Perineurial cells are a specialized non-EC type that encases thick nerve bundles, contributing to the nerve basement membrane and to nerve structural integrity³⁹.

Our Cdh5⁺/Krt19⁺ population molecular signature overlapped to some extent with the reported Krt19⁺ perineurial population (Fig. 5a–c), including some known perineurial genes such as Slc2a1, Perp and Itgb4⁴⁰ (Fig. 5b). VE-cadherin expression was not previously reported in these Krt19⁺ cells, but a decades-old histological study reported VE-cadherin staining around peripheral nerves⁴¹. We then stained skin sections from Krt19-CreERT; tdTomato mice with nerve-specific neurofilament (NF) and schwann cell-specific (S100β) antibodies. The data demonstrated exclusive localization of the tdTomato⁺ tubular structures with skin nerve bundles, which they appeared to encapsulate, as demonstrated by high-resolution confocal microscopy (Fig. 5d, e). Perineurial cells secrete basement membrane related proteins^42,43 that also matched with the strong ECM signature of the Cdh5⁺/Krt19⁺ population (Supplementary Fig. 6g). In fact, the Krt19-CreERT;tdTomato marked structures stained strongly for laminin, a basement membrane marker (Fig. 5f). We had already shown that the Krt19-CreERT;tdTomato structures expressed the VE-cadherin protein (Fig. 4g, h). In addition, Cdh5-CreERT²;tdTomato marking by TM injection at PD17 followed by NF staining also revealed abundant colocalization of tdTomato⁺ tubular structures around the nerves (Fig. 5g), confirming our findings. Collectively, these data demonstrate the identity of our Cdh5⁺/Krt19⁺ cell population as a previously un-recognized non-endothelial VE-cadherin⁺ perineural cell population that encapsulates the skin nerve bundles being embedded in basement membrane depositions (Fig. 5h).

**Fig. 5: *Cdh5*^+/*Krt19*⁺ cells are non-endothelial perineurial cells embedded in the nerve basement membrane.**

Dynamic remodeling of VE-cadherin⁺ perineurial cells and nerve bundles in hair cycle

Since VE-cadherin expression in ECs underlines highly dynamic tubular vascular structures that are strongly remodeled during hair cycle (Fig. 1a, b)^15,16,18, we wondered if our detection of VE-cadherin in the perineurial tubes may also underline dynamic structures. This may be especially the case, since skin nerves reportedly remodel during the hair cycle, when the hair follicle (HF) epithelium and the nerves likely crosstalk^15,44,45 (Fig. 6a).

**Fig. 6: Homeostatic remodeling of perineurial VE-cadherin⁺ lineage during the hair cycle.**

Nerve bundles arrive in skin through the subcutaneous muscle and form 3 horizontal plexuses, with the middle plexus at the hypodermal/dermal junction; medium-sized and thin nerve bundles derived from this plexus innervate the bulge, upper hair follicle and the epidermal site^15,44,45 (Fig. 6a). During anagen, the hypodermal nerves become stretched out and display new nerve terminations innervating the newly formed hair follicle (HF bulbs), while dermal nerves display increased density of thin filaments and more branching around the HF bulge^44,45. The early reports documenting nerve remodeling during hair cycle^44,45 were consistent with our images of thick (70 µm) NF/VE-cadherin-stained skin sections, followed by z-stack confocal imaging at telogen and anagen (Fig. 6b-d’). Dynamic remodeling of the nerves was also apparent in our quantification, where thick bundles were more frequently scored at telogen than at anagen (Fig. 6e).

To see if the VE-cadherin⁺ perineurial structures also undergo remodeling during hair cycle, we used Krt19-CreERT;tdTomato mice TM induced at PD17 and analyzed skin at telogen (PD20) and anagen (PD32) for changes in non-HF tdTomato⁺ structure morphology (Fig. 6f–i). At telogen, the tdTomato⁺ structures outside the HF appeared compact and only rarely extended into long tubular structures, which appeared more prominent and frequent by anagen (Fig. 6g, h). The tdTomato⁺ signal enclosed variable structures that contained 1–12 total DAPI⁺ nuclei within the nerve bundle. Quantification of the number of nuclei belonging to schwann cells that were enclosed by tdTomato signal/structure demonstrates an average increase and a distribution shift towards larger structures at anagen (Fig. 6i). This data may suggest that perineurial cells spread or stretch out around and over the nerves and schwann cells. Ki67 staining of skin sections at the different stages did not catch co-localization with the perineurial cells, suggesting little if any proliferation of these cells.

To document how VE-cadherin⁺ perineurial cells associate with different nerve bundles during the hair cycle, we performed co-immunostaining with NF and VE-cadherin antibodies and divided the NF⁺ bundles into thick (>5 fibers), medium (3–5 fibers) and thin (1–2 fibers) (Fig. 6j). Quantification of this data at 3 hair cycle stages (n = 3 mice/stage and N > 45 bundles/stage) showed ~100% VE-cadherin⁺ perineurial signal associated with the thick bundles, ~40–80% association with the medium bundles depending on hair cycle stage, and only ~10% association with thin bundles or individual axons (Fig. 6k).

Taken together, these data demonstrate a dynamic homeostatic re-organization of VE-cadherin⁺ non-endothelial perineurial cells along with the skin nerve bundles and the hair follicles during hair cycle and a preference of the perineurial cells for thick nerve bundles (Fig. 6l).

Alk1 regulates expression of ECM-associated genes in VE-cadherin (Cdh5)⁺ lineages

To understand the function and regulation of the VE-cadherin⁺ non-endothelial perineurial population during hair cycle, we turned to known VE-cadherin interacting partners from ECs^3,46. Of these, the TGF-β pathway surfaced most strongly in our skin EC populations during hair cycle (Fig. 2h). In fact, we have reported one TGF-β receptor - activin receptor-like kinase 1 (Alk1) known as an important regulator of EC function^47,48- as playing an essential role in skin vasculature remodeling during hair cycle^16,18. Here, we found that Cdh5⁺/Krt19⁺ cells indeed expressed detectable levels of several TGF-β members, including Alk1 (Fig. 7a). To ask if Alk1 may play a role in the Cdh5⁺ perineurial cells, we first examined our previous scRNA-seq dataset generated from tdTomato⁺ sorted cells from Cdh5-CreERT² x Alk1^fl/fl mice TM induced at PD17 and sacrificed at 1^st telogen¹⁶, which we referred to from here on as KO (Fig. 7b–e). When compared with the telogen dataset that served here as wild-type control (CT), hundreds of genes changed expression in Cdh5⁺ perineurial cells due to Alk1 loss, and of these, 53 changed by >3x, which is highly significant in single-cell analysis (Supplementary Table 1). Among the 33 downregulated genes, we noted signaling processes, cytoskeletal, and ECM and basement membrane formation (Supplementary Table 1 and Supplementary Data 6). Notably, expression of encapsulating structure signatures, ECM constituents, and collagen-associated ECM, which were the top functional categories enriched in the Cdh5⁺ perineurial cells, were also significantly reduced in KO perineurial cells (Fig. 7f, g). To compare the Alk1 effects in ECs with those in the Cdh5⁺ perineurial cells, we used GO terms analysis using DEGs (FC > 1.3). Interestingly, several top down-regulated processes in Cdh5⁺ perineurial cells showed similarities in blood vessels (but not in lymphatic vessels) EC populations (Fig. 7g), which includes collagen-containing ECM, external encapsulating structure signatures, and oxidative metabolism. To further investigate if Alk1 has direct roles in regulating expression of these ECM constituents, we FACS sorted tdTomato⁺ cells from the skin tissue of Cdh5-CreERT² x Alk1^fl/fl x tdTomato mice only after 3 days of tamoxifen injection (Supplementary Fig. 7a). We then qPCR profiled mRNA levels in CT and KO cells that confirmed downregulation of ECM factors Col1a1, Lypd2, Col3a1, Serpinh1, Pcolce, Col1a2, Sbspon, Vtn and Fn1, which were the top genes downregulated by Alk1 loss in perineurial cells (Fig. 7h).

**Fig. 7: Alk1 regulates expression of encapsulation and basement membrane components in vivo.**

Taken together, these findings demonstrate that Alk1, a known VE-cadherin partner in ECs function^3,46,47,48, is also utilized by the Cdh5⁺ non-endothelial perineurial cells to regulate gene expression signatures, defining both common and unique characteristics of cells within the skin VE-cadherin lineage. In perineurial cells, we find that Alk1 specifically promotes expression of ECM molecules along with the encapsulating structure identity. This was likely associated with the known function of this cell type in basement membrane deposition around the nerves^39,42.

Alk1 acts in perineurial cells to maintain proper nerve branching during the hair-cycle

To understand how Alk1 loss in perineurial cells may affect nerve organization and remodeling, we first examined skin from Cdh5-CreERT² x Alk1^fl/fl; tdTomato mice, TM induced at PD17 and sacrificed at PD25, by NF and laminin staining. Little if any changes were noted relative to control mice at this early stage post Alk1 loss. Since these mice die shortly after PD25 due to defects in the vasculature, we crossed the Krt19-CreERT x Alk1^fl/fl; tdTomato mice and examined tdTomato⁺ clones outside the HFs at PD32 and PD49 after TM induction at PD17. These mice survived and appeared normal 2-3 weeks post TM induction with no obvious phenotypes, indicating that loss of Alk1 in Krt19-CreERT targeted cells does not induce major physiological effects in these adult mice. This may not be surprising as the patches of non-HF tdTomato⁺ cells are relatively small and rare with only 1-3 structure present in each 30–60 µm thick and ~1.5 cm long skin sections analyzed. Second, since Krt19-CreERT also targets Alk1 in the hair follicles, where it is lowly expressed (Supplementary Fig. 7b), we inspected the tissue for any defects in hair morphology or hair cycle progression. However, these appeared normal at both PD32 (anagen) and PD49 (telogen) stages analyzed in all n = 3 mice/stage analyzed (Supplementary Fig. 7c). The tdTomato⁺ non-HF cells and clones were present at their normal relatively rare frequency of 1–3 nerve structures/skin section in both CT and KO skin. Interestingly, IF staining for laminin – a marker that provides a collective assessment of the basement membrane’s architecture and stability in nerves and other structures - was strongly reduced in Krt19-CreERT targeted tdTomato⁺ clones in dermis of KO when compared with CT mice (Fig. 8a-c).

**Fig. 8: Alk1 loss in perineurial cells results in increased nerve branching during hair-cycle.**

Next, we examined nerve bundle morphology in the tdTomato⁺ non-HF clones using IF-staining for S100β, a schwann cell marker that shows a tightly colocalized staining pattern with NF (Supplementary Fig. 7d). Strikingly, nerves marked as tdTomato⁺ clones appeared disorganized, highly branched and with fewer fibers/bundle in Krt19-CreERT driven Alk1 KO skin dermis as compared to the CT dermis at both hair cycle stages analyzed (Fig. 8d, e). This was largely confirmed by our quantification, especially in the number of branches per tdTomato⁺ nerve structure in CT and KO mice in more than 60 structures for each genotype/stage (Fig. 8f, g). Together, these results identified Alk1-driven molecular control of perineurial function and established one role of these perineurial cells in maintaining the physiological nerve branching patterns during the hair cycle via Alk1 activity (Fig. 8h).

Discussion

Our work re-defines the cellular and molecular heterogeneity of the essential VE-cadherin⁺ lineage and its dynamics during homeostasis, firmly expanding its functions beyond classical endothelial vasculature remodeling^1,49 into perineurial cells and skin nerve maintenance. Most importantly, we uncovered a previously unrecognized VE-cadherin⁺ non-endothelial perineurial cell population encapsulating the skin nerves, contributing to the nerve basement membrane, and remodeling into tubular-like structures along with the nerves during the hair cycle. We implicate Alk1, a known VE-cadherin interacting partner from the TGF-β pathway, in the function of these perineurial cells. We show that Alk1 in the perineurium promotes production of ECM factors, deposition of nerve basement membrane, and maintenance of proper branching and organization of nerve bundles during the hair cycle.

Importantly, VE-cadherin expression in a non-endothelial (perineurial) population and the implication of its endothelial partner Alk1^3,46,47,48, in homeostatic remodeling of nerve structures, changes the existing dogma in the field that considers these interacting factors as specific to the endothelium. Our result is supported by a decades-old, and since largely ignored VE-cadherin staining near skin nerves⁴¹. VE-cadherin is a cell-cell contact molecule with specialized function in tubular structures that have been so-far thought of as exclusively vascular, until now. VE-cadherin is known to confer tubular EC structures flexibility, resilience, and permeability to gases, nutrients, and blood/immune cells^50,51,52, and it’ll be interesting in the future to examine its role in the perineurium. Given its known functions in endothelial cells, the identification of VE-cadherin in the non-endothelial dynamic perineurial tubes is intriguing. The perineurium is already known to be endowed with characteristics that VE-cadherin promotes such as being flexible, resistant to mechanical damage, and playing important roles in permeability through the blood-nerve barrier⁵³. In fact, acute and chronic pain is related to nerve inflammation and the infiltration of immune cells to the nerve terminations⁵⁴, which may be mediated in the perineurium via VE-cadherin, a tantalizing possibility for future investigation.

Our data clearly implicated Alk1 in promoting ECM gene expression in both endothelial and perineurial cells, which both form dynamic VE-cadherin⁺ tubular structures. On the other hand, Alk1 represses ECM gene expression in other non-endothelial mesenchymal cell types including structural but non-tubule forming cells, such as fibroblasts⁵⁵ and chondrocytes⁵⁶. This suggest that although perineurial cells have a strong mesenchymal identity, Alk1’s effect on ECM genes in perineurial cells resembles more the endothelial, than the mesenchymal cell types. Interestingly, Alk1 loss in endothelial cells results in increased permeability and leakiness of the vasculature, which loses its integrity⁵⁷. Similarly, we show that nerve bundles branch more when Alk1 is lost from their surrounding perineurium, possibly due to low laminin content in the basement membrane. In contrast, the known Alk1 role in EC migration¹⁶ may not apply in perineurial cells, since the latter seem to remodel correctly into branched out tubular structures at anagen. This remodeling may involve perineurial cell stretching, migrating and encapsulating schwann cells and nerve terminations, and this seems to occur even in the absence of Alk1. Interestingly, gene expression changes upon Alk1 loss are more similar in perineurial and blood vessel endothelial cells, when compared to lymphatic vessel ECs, perhaps attesting to the less dynamic nature of the latter.

That the VE-cadherin/Alk1 signaling axis, previously implicated exclusively in the endothelium^3,46,47,48, is present in the perineurial cells as well, may have important implications in understanding the known coordinated nerve/vasculature remodeling that has intrigued scientists for decades⁵⁸. Nerve and vasculature structures are created by branching morphogenesis, and both structures share and respond to growth factor derived from the other compartment⁵⁸. Moreover, they are both known to respond to signals from hair follicle epithelial cells during the hair cycle¹⁵. Perhaps a VE-cadherin⁺/Alk1 axis may be part of a general tissue network that drives the coordinated nerve/vasculature remodelling during homeostasis, with implications in vasculature and nerve disease.

In addition to shifting the dogma for the VE-cadherin⁺ population by expanding its known heterogeneity, our data provides the first high resolution cellular and molecular characterization of vasculature dynamics in the homeostasis of a highly regenerative tissue. While many adult vascular systems are quiescent during homeostasis⁵⁹, a few adult tissues besides skin¹⁵ also employ dynamic vasculature changes in the absence of injury. The latter include menstruation in female endometrial homeostasis⁶⁰, exercise in the skeletal muscle⁶¹, and the ongoing regeneration of lacteals in the intestine⁶². The synchronicity of hair follicle homeostatic growth and regression in skin facilitates the characterization of this system, illuminating the way for other adult regenerative tissues. Interestingly, we detected an anagen-specific increase in blood ECs at the expense of the lymphatic ECs, suggesting the two lineages are in opposite flux during skin homeostasis. While blood vessel EC number increase is not surprising, and aligns with proposed high energy needs to supply O₂ and nutrients for the production of new hair shafts^15,20 and to expand the skin hypodermis, the concomitant decrease in lymphatics ECs is intriguing. A reported reduction in the dermal lymphatic capillaries area at anagen²⁴, and the reduced drainage observed at anagen²³ may corroborate our finding. The lymphatic vasculature is generally static in adult tissues^63,64, but homeostatic regeneration also occurs in the intestinal ‘lacteal’ lymphatic capillaries and the ovary during folliculogenesis^62,65. Our study in skin adds important insight into active vascular homeostasis of another highly regenerative adult tissue.

Importantly, we exclude here the existence of a hair-cycle active skin resident VE-cadherin⁺ endovascular progenitor (EVP) that was previously documented in the aorta, skin wounding and skin tumors^11,12,13. Our ultra-sensitive labeling method (Cdh5-CreERT² x tdTomato) uncovered a mesenchymal perineurial cell type and not an immature EC type. Thus, it appears that skin ECs change their overall numbers and balance between blood and lymphatic ECs during hair cycle, without cellular input from a skin-resident VE-cadherin⁺ immature cell type or EVP.

Furthermore, our data demonstrate that in homeostasis, skin EC populations coordinately change their molecular makeup to reflect increased metabolic and biosynthetic activities. This corroborates with previous findings of microvasculature remodeling through capillary pruning and duplication⁶⁶ and through exercise-induced angiogenic stimulus⁶¹. Several major signaling pathways surfaced from our transcriptomic analysis as differentially regulated in the specific EC populations, including TGF-β, Notch and Wnt signaling. These data should aid in the emerging cross-communication of ECs with hair follicle stem cells¹⁵, which is poorly understood. More importantly, this work may aid in efforts to engineer proper organotypic skin cultures for chronic and diabetic wounds⁶⁷ and for understanding skin vascular disease such as Human hereditary hemorrhagic telangiectasia (HHT)⁶⁸.

Finally, our study has some technical strengths and advances worth mentioning. First, using FACS on the highly sensitive Cdh5-CreERT²;tdTomato labeling system permitted access to rare and low-expressing VE-cadherin cell types, which baring one decades-old study⁴¹ were generally missed in antibody staining. Second, despite our sensitive methods, we did not find a VE-cadherin⁺ endothelial vascular progenitor (EVP), as suggested in the aorta, in cancer or in injury^11,12,13. Significantly, our computational lineage trajectory analysis and specific gene expression signatures as done in aorta¹¹ supported the EVP hypothesis but this was refuted by our genetic lineage tracing in skin, cautioning against the use of single-cell transcriptomic computational methods alone to predict cell identity and functions. Third, our single-cell transcriptomic data revealed strong expression of Krt19 mRNA in perineurial cells, which prompted us to employ the Krt19-CreERT driver that also marked the skin perineurium. The Krt19-CreERT was previously considered specific to epithelial hair follicle stem cells³⁶, but evidently, this assumption should be revised given our data. The expression of a keratin in a perineurial cell type with a strong mesenchymal signature is intriguing and requires further investigation and confirmation at the protein level.

In conclusion, our systematic analysis of the VE-cadherin⁺ population in adult skin revealed previously un-recognized non-endothelial cell types and identified dynamic reorganization that highlights the coordinated nerve and vasculature remodeling during homeostasis. This data un-earths unexpected cellular and functional heterogeneity in the essential VE-cadherin⁺ population using skin as a highly regenerative adult tissue and will have relevance to other adult tissues and to vascular and nerve disease.

Methods

Mice and treatments

All the mouse work was performed in accordance with the Cornell University Institutional Animal Care and Use Committee (IACUC) guidelines (protocol no. 2007-0125). The tdTomato²⁶ (Stock no 007905) and Krt19-CreERT³⁶ (Stock no 026925) mice were obtained from Jackson Laboratory. The Alk1^fl/fl and Cdh5-CreERT² ²⁵ mice were imported from Dr. Anne Eichmann at Yale University and used as previously described¹⁸. Mouse sex was not considered in the study design. The mice were intraperitoneally (IP) injected with tamoxifen at a dose of 200 µg/g body weight to induce Cre recombination. For BrdU-pulse experiments, mice were fed with 0.8 mg/ml BrdU in daily water supply.

Immunofluorescence staining, microscopy and image processing

Immunofluorescence (IF) staining on the non-prefixed tissue sections was performed by following the standard protocol, as described previously¹⁸. The tdTomato⁺ skin tissues were prefixed in 4% paraformaldehyde for 2 h at 4 °C followed by passing them through a sucrose gradient before embedding in OCT compound (Tissue Tek, Sakura). The non-prefixed samples were first fixed in 4% paraformaldehyde (PFA) for 10 minutes at room temperature (RT), followed by washing with PBS and 20 mM glycine in PBS. The sections were blocked with 5% normal serum for 1 h at RT before incubating them with primary antibodies at 4 °C overnight. Next day, sections were 3x washed with PBST and then incubated with secondary antibodies (1:500). The sections were washed 3x with PBST and stained with DAPI. The slides were mounted with antifade solution. For thick sections (70 µm), primary antibodies incubation was performed for 48 hours at 4 °C and secondary antibodies overnight at 4 °C. For BrdU staining, tissue sections or cells were first stained with non-BrdU primary antibodies, followed by fixation using 4% PFA. DNA was denatured by incubating the slides in 1 M HCl solution for 55 minutes at 37 °C. The slides were washed with PBS and re-blocked using 0.5% Tween-20 and 1% BSA before incubating them with anti-BrdU antibody overnight at 4 °C. Then slides were washed in PBST and incubated with FITC-conjugated anti-rat antibody. Primary antibodies used in this study include CD31 (1:100, BD Biosciences, 550274); VE-cadherin (1:500, R&D Systems, AF1002); LYVE1 (1:400, Thermo Scientific, 14-0443-82); Prox1 (1:300, Abcam, ab199359); BrdU (1:300, Abcam, ab6326); Endomucin (1:300, Santa Cruz Biotechnology, sc-65495); CD34 (1:100, BD Biosciences, 553731); Laminin (α1) (1:500, Sigma-Aldrich, L9393); S100β (1:400, Proteintech, 15146-1-AP) and Neurofilament Heavy chain (1:500, Millipore, AB5539). The IF images were captured using either the Leica DMI6000B microscope with Leica K5 camera or the Zeiss LSM inverted 880 confocal microscopes. The z-stacks (1 µm stacks) images of thick sections were deconvoluted using AutoQuant X software and were further processed for brightness, contrast and stitching using Fiji⁶⁹. Multiple images were combined using stitching package⁷⁰ in Fiji⁶⁹ for Fig. 1. The quantification graphs were generated using GraphPad Prism 9.

Fluorescence-activated cell sorting (FACS) of VE-cadherin⁺ cells

FACS isolation of VE-cadherin⁺ cells and its purity check was performed as previously described²⁷. Briefly, tdTomato (Jax Stock #007905), Cdh5-CreERT² ²⁵ and Krt14-H2BGFP⁷¹ mice were used for VE-cadherin⁺ cells and control keratinocytes isolation. The VE-cadherin⁺ cells were labeled with tdTomato by injecting tamoxifen (200 μg/g) at postnatal day (PD)17, and dorsal skin was collected at PD20 or PD32. The dorsal skin was digested in collagenase and Dispase mixture as previously described²⁷. The dead cells were removed by LIVE/DEAD™ Fixable Aqua Dead Cell Stain Kit (ThermoFisher). FACS Aria (BD Biosciences) was used for cell sorting in Cornell Flow Cytometry facility. FACS data were analyzed with the FlowJo (FlowJo™ Software, v10.5.0, BD Biosciences).

Single cell library preparation and data analysis

The barcoded single-cell 3′ cDNA libraries were generated using Chromium Single Cell 3′ gel bead and library Kit v3 (10x Genomics) and sequenced using an Illumina NextSeq-500. The raw data were aligned to the mouse reference genome (mm10-2020-A) using the 10X Genomics Cell Ranger pipeline (v6.0.1). The data analysis was performed in R using the Seurat package version 4.0^72,73. Cells that had between 200 and 5000 genes expressed and had under 10% of the UMIs mapped to mitochondrial genes were retained. We obtained a total of 3957 and 8627 high-quality cells for telogen and anagen samples, respectively. For further analysis, all samples were merged, the transcript counts were log-normalized, and the expression of each gene was scaled so that the variance in gene expression across cells was one, followed by Principal Component Analysis (PCA). The data integration was performed by PCA embeddings using Harmony⁷⁴. The corrplot R function was used to measure the correlation between the two biological replicates. Differentially expressed genes (DEGs) were identified by ‘FindAllMarkers’ function using the Wilcox Rank Sum test.

Cell lineage trajectory, gene scoring and gene ontology analysis

Trajectory analysis was performed using Monocle³⁵. We used the top 2000 genes to order cells in pseudotime trajectory by Monocle and the cells having lower than 200 transcripts were removed from further analysis. The DDRTree package was employed to construct the trajectories. To determine the collective enrichment of a set of genes in different populations, the AddModuleScore function in the Seurat R package was used. For pathway analysis, DEGs were identified by Seurat’s ‘FindMarkers’ function using the Wilcox Rank Sum test (FC > 1.3). Further enrichment analysis and plotting were performed as previously described²⁷.

Heatmap of differentially expressed genes (DEGs) using raw counts

First, DEGs were identified using FindMarkers function of Seurat. Next, to obtain the expression value for each gene in all populations, the raw counts for the DEGs were extracted using the FetchData function. Of 205 genes (FC > 1.5) 6 showed 0 values in at least one population, whereas the corresponding population in the other stage showed a very low expression level relative to other populations. Thus ‘0’ was replaced with the low ‘background’ value from the corresponding population at the other stage. This way, these genes with very low expression levels in some populations appear ‘unchanged’ from telogen to anagen, thus eliminating the possibility that small fluctuations in background expression are assigned a ‘changed’ value. The Xist gene showed 0 values in more than one population but did not display an obvious background value in the corresponding population at the other stage, and thus they were not analyzed further. The same method was applied to the gene set of 784 genes (FC > 1.3), where 18 showed 0 values in at least one population.

RNA isolation and real-time quantitative RT-PCR

FACS sorting of VE-cadherin⁺ cells was performed as described above. Sorted cells were directly collected in TRIzol LS (ThermoFisher) for total RNA isolation, and c-DNA synthesis was performed using Superscript III (Invitrogen) following manufactures instructions. The primer sequences are provided in Supplementary Data 7.

Statistics and reproducibility

Data comparisons between two groups was performed using the two-tailed Student’s t-test. For multiple comparisons, we used two-way ANOVA followed by Tukey’s test. We used GraphPad Prism (Prism 9) for statistical calculations and analyses. Statistical significance was indicated in all experiments by p value. No statistical method was used to predetermine the sample size. The experiments were not randomized. The investigators were not blinded to allocation during experiments and outcome assessment.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The scRNA-seq data generated in this study have been deposited in the NCBI Gene Expression Omnibus (GEO) database under the accession number GSE211381. Source data are provided with this paper.

Code availability

R script code for Seurat analysis is deposited in a GitHub repository and is freely accessible at https://github.com/GChovatiya/Chovatiya_et_al_Nat_Comm_SourceCode.

References

Vestweber, D. VE-cadherin: the major endothelial adhesion molecule controlling cellular junctions and blood vessel formation. Arterioscler Thromb. Vasc. Biol. 28, 223–232 (2008).
Article CAS PubMed Google Scholar
Giannotta, M., Trani, M. & Dejana, E. VE-Cadherin and Endothelial Adherens Junctions: Active Guardians of Vascular Integrity. Dev. Cell 26, 441–454 (2013).
Article CAS PubMed Google Scholar
Lagendijk, A. K. & Hogan, B. M. Chapter Ten - VE-cadherin in Vascular Development: A Coordinator of Cell Signaling and Tissue Morphogenesis. In: Current Topics in Developmental Biology (ed Yap AS). Academic Press (2015).
Carmeliet, P. et al. Targeted Deficiency or Cytosolic Truncation of the VE-cadherin Gene in Mice Impairs VEGF-Mediated Endothelial Survival and Angiogenesis. Cell 98, 147–157 (1999).
Article CAS PubMed Google Scholar
Corada, M. et al. Vascular endothelial-cadherin is an important determinant of microvascular integrity in vivo. Proc. Natl Acad. Sci. USA 96, 9815–9820 (1999).
Article CAS PubMed PubMed Central ADS Google Scholar
Matsuyoshi, N. et al. In vivo evidence of the critical role of cadherin-5 in murine vascular integrity. Proc. Assoc. Am. Physicians 109, 362–371 (1997).
CAS PubMed Google Scholar
Delgado-Bellido, D., Serrano-Saenz, S., Fernández-Cortés, M. & Oliver, F. J. Vasculogenic mimicry signaling revisited: focus on non-vascular VE-cadherin. Mol. Cancer 16, 65 (2017).
Article PubMed PubMed Central Google Scholar
Jambusaria, A. et al. Endothelial heterogeneity across distinct vascular beds during homeostasis and inflammation. eLife 9, e51413 (2020).
Article CAS PubMed PubMed Central Google Scholar
Kalucka, J. et al. Single-Cell Transcriptome Atlas of Murine Endothelial Cells. Cell 180, 764–779.e720 (2020).
Article CAS PubMed Google Scholar
Marcu, R. et al. Human Organ-Specific Endothelial Cell Heterogeneity. iScience 4, 20–35 (2018).
Article CAS PubMed PubMed Central ADS Google Scholar
Lukowski, S. W. et al. Single-Cell Transcriptional Profiling of Aortic Endothelium Identifies a Hierarchy from Endovascular Progenitors to Differentiated Cells. Cell Rep. 27, 2748–2758.e2743 (2019).
Article CAS PubMed Google Scholar
Donovan, P. et al. Endovascular progenitors infiltrate melanomas and differentiate towards a variety of vascular beds promoting tumor metastasis. Nat. Commun. 10, 18 (2019).
Article CAS PubMed PubMed Central ADS Google Scholar
Patel, J. et al. Functional Definition of Progenitors Versus Mature Endothelial Cells Reveals Key SoxF-Dependent Differentiation Process. Circulation 135, 786–805 (2017).
Article CAS PubMed Google Scholar
Hsu, Y. C. & Fuchs, E. Building and Maintaining the Skin. Cold Spring Harb Perspect Biol. 14, a040840 (2022).
Li, K. N. & Tumbar, T. Hair follicle stem cells as a skin-organizing signaling center during adult homeostasis. Embo j. 40, e107135 (2021).
Article CAS PubMed PubMed Central Google Scholar
Li, K. N., Chovatiya, G., Ko, D. Y., Sureshbabu, S. & Tumbar, T. Blood endothelial ALK1-BMP4 signaling axis regulates adult hair follicle stem cell activation. EMBO J. 42, e112196 (2023).
Peng, J., Chen, H. & Zhang, B. Nerve-stem cell crosstalk in skin regeneration and diseases. Trends Mol. Med. 28, 583–595 (2022).
Article CAS PubMed Google Scholar
Li, K. N. et al. Skin vasculature and hair follicle cross-talking associated with stem cell activation and tissue homeostasis. Elife. 8, e45977 (2019).
Mecklenburg, L. et al. Active Hair Growth (Anagen) is Associated with Angiogenesis. J. Invest. Dermatol. 114, 909–916 (2000).
Article CAS PubMed Google Scholar
Yano, K., Brown, L. F. & Detmar, M. Control of hair growth and follicle size by VEGF-mediated angiogenesis. J. Clin. Invest 107, 409–417 (2001).
Article CAS PubMed PubMed Central Google Scholar
Müller-Röver, S. et al. A Comprehensive Guide for the Accurate Classification of Murine Hair Follicles in Distinct Hair Cycle Stages. J. Invest. Dermatol. 117, 3–15 (2001).
Article PubMed Google Scholar
Goldstein, J. & Horsley, V. Home sweet home: skin stem cell niches. Cell. Mol. life Sci.: CMLS 69, 2573–2582 (2012).
Article CAS PubMed Google Scholar
Gur-Cohen, S. et al. Stem cell-driven lymphatic remodeling coordinates tissue regeneration. Science 366, 1218–1225 (2019).
Article CAS PubMed PubMed Central ADS Google Scholar
Peña-Jimenez, D. et al. Lymphatic vessels interact dynamically with the hair follicle stem cell niche during skin regeneration in vivo. Embo j. 38, e101688 (2019).
Article PubMed PubMed Central Google Scholar
Sörensen, I., Adams, R. H. & Gossler, A. DLL1-mediated Notch activation regulates endothelial identity in mouse fetal arteries. Blood 113, 5680–5688 (2009).
Article PubMed Google Scholar
Madisen, L. et al. A robust and high-throughput Cre reporting and characterization system for the whole mouse brain. Nat. Neurosci. 13, 133–140 (2010).
Article CAS PubMed Google Scholar
Chovatiya, G., Ghuwalewala, S., Walter, L. D., Cosgrove, B. D. & Tumbar, T. High-resolution single-cell transcriptomics reveals heterogeneity of self-renewing hair follicle stem cells. Exp. Dermatol 30, 457–471 (2021).
Article CAS PubMed PubMed Central Google Scholar
Chavkin, N. W. & Hirschi, K. K. Single Cell Analysis in Vascular Biology. Front Cardiovasc Med. 7, 42 (2020).
Rocha, S. F. & Adams, R. H. Molecular differentiation and specialization of vascular beds. Angiogenesis 12, 139–147 (2009).
Article CAS PubMed Google Scholar
Goveia, J. et al. An Integrated Gene Expression Landscape Profiling Approach to Identify Lung Tumor Endothelial Cell Heterogeneity and Angiogenic Candidates. Cancer Cell 37, 21–36.e13 (2020).
Article CAS PubMed Google Scholar
Michel, M. et al. Keratin 19 as a biochemical marker of skin stem cells in vivo and in vitro: keratin 19 expressing cells are differentially localized in function of anatomic sites, and their number varies with donor age and culture stage. J. Cell Sci. 109, 1017–1028 (1996).
Article CAS PubMed Google Scholar
Mi, H. et al. PANTHER version 16: a revised family classification, tree-based classification tool, enhancer regions and extensive API. Nucleic Acids Res. 49, D394–D403 (2020).
Article PubMed Central Google Scholar
Subramanian, A. et al. Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl Acad. Sci. 102, 15545–15550 (2005).
Article CAS PubMed PubMed Central ADS Google Scholar
Brulois, K. et al. A molecular map of murine lymph node blood vascular endothelium at single cell resolution. Nat. Commun. 11, 3798 (2020).
Article CAS PubMed PubMed Central ADS Google Scholar
Trapnell, C. et al. The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells. Nat. Biotechnol. 32, 381–386 (2014).
Article CAS PubMed PubMed Central Google Scholar
Means, A. L., Xu, Y., Zhao, A., Ray, K. C. & Gu, G. A CK19CreERT knockin mouse line allows for conditional DNA recombination in epithelial cells in multiple endodermal organs. genesis 46, 318–323 (2008).
Article CAS PubMed PubMed Central Google Scholar
Gerber, D. et al. Transcriptional profiling of mouse peripheral nerves to the single-cell level to build a sciatic nerve ATlas (SNAT). eLife 10, e58591 (2021).
Article CAS PubMed PubMed Central Google Scholar
Chen, B., Banton, M. C., Singh, L., Parkinson, D. B. & Dun, X.-P. Single Cell Transcriptome Data Analysis Defines the Heterogeneity of Peripheral Nerve Cells in Homeostasis and Regeneration. Front Cell Neurosci. 15, 624826 (2021).
Petrova, E. S. & Kolos, E. A. Current Views on Perineurial Cells: Unique Origin, Structure, Functions. J. Evolut. Biochem. Physiol. 58, 1–23 (2022).
Article Google Scholar
Carr, M. J. et al. Mesenchymal Precursor Cells in Adult Nerves Contribute to Mammalian Tissue Repair and Regeneration. Cell Stem Cell 24, 240–256.e249 (2019).
Article CAS PubMed Google Scholar
Smith, M. E., Jones, T. A. & Hilton, D. Vascular endothelial cadherin is expressed by perineurial cells of peripheral nerve. Histopathology 32, 411–413 (1998).
Article CAS PubMed Google Scholar
Jaakkola, S., Peltonen, J. & Uitto, J. J. Perineurial cells coexpress genes encoding interstitial collagens and basement membrane zone components. J. Cell Biol. 108, 1157–1163 (1989).
Article CAS PubMed Google Scholar
Piña-Oviedo, S. & Ortiz-Hidalgo, C. The Normal and Neoplastic Perineurium: A Review. Adv. Anat. Pathol. 15, 147–164 (2008).
Article PubMed Google Scholar
Botchkarev, V. A., Eichmüller, S., Johansson, O. & Paus, R. Hair cycle-dependent plasticity of skin and hair follicle innervation in normal murine skin. J. Comp. Neurol. 386, 379–395 (1997).
Article CAS PubMed Google Scholar
Botchkarev, V. A., Peters, E. M., Botchkareva, N. V., Maurer, M. & Paus, R. Hair cycle-dependent changes in adrenergic skin innervation, and hair growth modulation by adrenergic drugs. J. Invest Dermatol 113, 878–887 (1999).
Article CAS PubMed Google Scholar
Rudini, N. et al. VE-cadherin is a critical endothelial regulator of TGF-beta signalling. Embo j. 27, 993–1004 (2008).
Article CAS PubMed PubMed Central Google Scholar
Seki, T., Yun, J. & Oh, S. P. Arterial Endothelium-Specific Activin Receptor-Like Kinase 1 Expression Suggests Its Role in Arterialization and Vascular Remodeling. Circulation Res. 93, 682–689 (2003).
Article CAS PubMed Google Scholar
Oh, S. P. et al. Activin receptor-like kinase 1 modulates transforming growth factor-β1 signaling in the regulation of angiogenesis. Proc. Natl Acad. Sci. 97, 2626 (2000).
Article CAS PubMed PubMed Central ADS Google Scholar
Harris, E. S. & Nelson, W. J. VE-cadherin: at the front, center, and sides of endothelial cell organization and function. Curr. Opin. Cell Biol. 22, 651–658 (2010).
Article CAS PubMed PubMed Central Google Scholar
Duong, C. N. & Vestweber, D. Mechanisms Ensuring Endothelial Junction Integrity Beyond VE-Cadherin. Front Physiol. 11, 519 (2020).
Article PubMed PubMed Central Google Scholar
Gavard, J. Endothelial permeability and VE-cadherin: a wacky comradeship. Cell Adh Migr. 7, 455–461 (2013).
Article PubMed PubMed Central Google Scholar
Orsenigo, F. et al. Phosphorylation of VE-cadherin is modulated by haemodynamic forces and contributes to the regulation of vascular permeability in vivo. Nat. Commun. 3, 1208 (2012).
Article PubMed ADS Google Scholar
Liu, Q., Wang, X. & Yi, S. Pathophysiological Changes of Physical Barriers of Peripheral Nerves After Injury. Front Neurosci. 12, 597 (2018).
Article CAS PubMed PubMed Central Google Scholar
Liu, J. A., Yu, J., Cheung, C. W. Immune Actions on the Peripheral Nervous System in Pain. Int J Mol Sci. 22, 1448 (2021).
Muñoz-Félix, J. M., Perretta-Tejedor, N., Eleno, N., López-Novoa, J. M. & Martínez-Salgado, C. ALK1 heterozygosity increases extracellular matrix protein expression, proliferation and migration in fibroblasts. Biochim Biophys. Acta 1843, 1111–1122 (2014).
Article PubMed Google Scholar
Finnson, K. W., Parker, W. L., ten Dijke, P., Thorikay, M. & Philip, A. ALK1 Opposes ALK5/Smad3 Signaling and Expression of Extracellular Matrix Components in Human Chondrocytes. J. Bone Miner. Res. 23, 896–906 (2008).
Article CAS PubMed Google Scholar
Akla, N. et al. BMP9 (Bone Morphogenetic Protein-9)/Alk1 (Activin-Like Kinase Receptor Type I) Signaling Prevents Hyperglycemia-Induced Vascular Permeability. Arterioscler Thromb. Vasc. Biol. 38, 1821–1836 (2018).
Article CAS PubMed Google Scholar
Ribatti, D. & Guidolin, D. Morphogenesis of vascular and neuronal networks and the relationships between their remodeling processes. Brain Res Bull. 186, 62–69 (2022).
Article CAS PubMed Google Scholar
Carmeliet, P. Angiogenesis in health and disease. Nat. Med 9, 653–660 (2003).
Article CAS PubMed Google Scholar
Rogers, P. A. & Gargett, C. E. Endometrial angiogenesis. Angiogenesis 2, 287–294 (1998).
Article CAS PubMed Google Scholar
Prior, B. M., Lloyd, P. G., Yang, H. T. & Terjung, R. L. Exercise-Induced Vascular Remodeling. Exerc. Sport Sci. Rev. 31, 26–33 (2003).
Article PubMed Google Scholar
Bernier-Latmani, J. et al. DLL4 promotes continuous adult intestinal lacteal regeneration and dietary fat transport. J. Clin. Invest 125, 4572–4586 (2015).
Article PubMed PubMed Central Google Scholar
Stritt, S., Koltowska, K. & Mäkinen, T. Homeostatic maintenance of the lymphatic vasculature. Trends Mol. Med. 27, 955–970 (2021).
Article CAS PubMed Google Scholar
Connor, A. L., Kelley, P. M. & Tempero, R. M. Invariant asymmetry renews the lymphatic vasculature during homeostasis. J. Transl. Med. 14, 209 (2016).
Article PubMed PubMed Central Google Scholar
Breslin, J. W. et al. Lymphatic Vessel Network Structure and Physiology. Compr. Physiol. 9, 207–299 (2018).
Article PubMed PubMed Central Google Scholar
Santamaría, R., González-Álvarez, M., Delgado, R., Esteban, S. & Arroyo, A. G. Remodeling of the Microvasculature: May the Blood Flow Be With You. Front Physiol. 11, 586852 (2020).
Randall, M. J., Jungel, A., Rimann, M. & Wuertz-Kozak, K. Advances in the Biofabrication of 3D Skin in vitro: Healthy and Pathological Models. Front Bioeng. Biotechnol. 6, 154 (2018).
Article PubMed PubMed Central Google Scholar
Ruiz-Llorente, L. et al. Endoglin and alk1 as therapeutic targets for hereditary hemorrhagic telangiectasia. Expert Opin. Therapeutic Targets 21, 933–947 (2017).
Article CAS Google Scholar
Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nat. Methods 9, 676–682 (2012).
Article CAS PubMed Google Scholar
Preibisch, S., Saalfeld, S. & Tomancak, P. Globally optimal stitching of tiled 3D microscopic image acquisitions. Bioinformatics 25, 1463–1465 (2009).
Article CAS PubMed PubMed Central Google Scholar
Tumbar, T. et al. Defining the epithelial stem cell niche in skin. Science 303, 359–363 (2004).
Article CAS PubMed ADS Google Scholar
Hao, Y. et al. Integrated analysis of multimodal single-cell data. Cell 184, 3573–3587.e3529 (2021).
Article CAS PubMed PubMed Central Google Scholar
Satija, R., Farrell, J. A., Gennert, D., Schier, A. F. & Regev, A. Spatial reconstruction of single-cell gene expression data. Nat. Biotechnol. 33, 495–502 (2015).
Article CAS PubMed PubMed Central Google Scholar
Korsunsky, I. et al. Fast, sensitive and accurate integration of single-cell data with Harmony. Nat. Methods 16, 1289–1296 (2019).
Article CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We thank Drs. Lauren D. Walter and Benjamin D. Cosgrove for their help with scRNA-seq analysis and helpful suggestions. We also thank the Center for Animal Resources and Education (CARE) for mouse care and the Cornell University Biotechnology Resource Center (BRC) for their help in scRNA-seq experiments. Confocal microscopy work on Zeiss upright 880 was supported by the Cornell BRC facility (NYSTEM C029155 and NIH S10OD018516). This work was supported by NIH/NIAMS Grants R01AR070157, R56AR081021, R01AR081021 and R01AR073806 to TT and by the Empire State Stem Cell Fund through New York State Department of Health Contract # C30293GG training grant to GC. Opinions expressed here are solely those of the author and do not necessarily reflect those of the Empire State Stem Cell Board, the New York State Department of Health, or the State of New York.

Author information

Authors and Affiliations

Department of Molecular Biology and Genetics, Cornell University, Ithaca, NY, USA
Gopal Chovatiya, Kefei Nina Li, Jonathan Li, Sangeeta Ghuwalewala & Tudorita Tumbar

Authors

Gopal Chovatiya
View author publications
You can also search for this author in PubMed Google Scholar
Kefei Nina Li
View author publications
You can also search for this author in PubMed Google Scholar
Jonathan Li
View author publications
You can also search for this author in PubMed Google Scholar
Sangeeta Ghuwalewala
View author publications
You can also search for this author in PubMed Google Scholar
Tudorita Tumbar
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

G.C., N.L. and T.T. designed the experiments. G.C., N.L. and J.L. performed the experiments. G.C. did the computational analysis and in vivo validations. N.L. did the clonal analysis in Fig. 6. J.L. performed laminin and S100β staining and their quantifications in Fig. 8. S.G. contributed to pathway analysis in Fig. 2. G.C. and T.T. analyzed and interpreted all data. G.C., N.L. and T.T. prepared the figures. G.C. and T.T. wrote the manuscript.

Corresponding author

Correspondence to Tudorita Tumbar.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Maksim Plikus and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Peer Review File

Description of Additional Supplementary Files

Supplementary Data 1

Supplementary Data 2

Supplementary Data 3

Supplementary Data 4

Supplementary Data 5

Supplementary Data 6

Supplementary Data 7

Reporting Summary

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Chovatiya, G., Li, K.N., Li, J. et al. Alk1 acts in non-endothelial VE-cadherin⁺ perineurial cells to maintain nerve branching during hair homeostasis. Nat Commun 14, 5623 (2023). https://doi.org/10.1038/s41467-023-40761-5

Download citation

Received: 02 August 2022
Accepted: 09 August 2023
Published: 12 September 2023
DOI: https://doi.org/10.1038/s41467-023-40761-5

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.