Quantitative Evaluation of Human Umbilical Vein and Induced Pluripotent Stem Cell-Derived Endothelial Cells as an Alternative Cell Source to Skin-Specific Endothelial Cells in Engineered Skin Grafts

Abstract

Objective: We compared the capability of human umbilical vein endothelial cells (HUVECs), induced pluripotent stem cell (iPSC)-derived endothelial cells (iECs), and human dermal blood endothelial cells (HDBECs) to effectively vascularize engineered human skin constructs (HSCs) in vitro and on immunodeficient mice.

Approach: We quantified the angiogenesis within HSCs both in vitro and in vivo through computational analyses of immunofluorescent (IF) staining. We assayed with real-time quantitative PCR (RT-qPCR) the expression of key endothelial, dermal, and epidermal genes in 2D culture and HSCs. Epidermal integrity and proliferation were also evaluated through haematoxylin and eosin staining, and IF staining.

Results: IF confirmed iEC commitment to endothelial phenotype. RT-qPCR showed HUVECs and iECs immaturity compared with HDBECs. In vitro, the vascular network extension was comparable for HDBECs and HUVECs despite differences in vascular diameter, whereas iECs formed unorganized rudimentary tubular structures. In vivo, all ECs produced discrete vascular networks of varying dimensions. HUVECs and HDBECs maintained a higher proliferation of basal keratinocytes. HDBECs had the best impact on extracellular matrix expression, and epidermal proliferation and differentiation.

Innovation: To our knowledge, this study represents the first direct and quantitative comparison of HDBECs, HUVECs, and iECs angiogenic performance in HSCs.

Conclusions: Our data indicate that HUVECs and iECs can be an alternative cell source to HDBEC to promote the short-term viability of prevascularized engineered grafts. Nevertheless, HDBECs maintain their capillary identity and outperform other EC types in promoting the maturation of the dermis and epidermis. These intrinsic characteristics of HDBECs may influence the long-term function of skin grafts.

Hasan Erbil Abaci, PhD

Introduction

Skin is a complex organ, articulated in epidermal, dermal, and hypodermal layers. It is rich in vasculature, innervation, appendages, and inhabited by more than 50 cell types. Over the past decades, many groups have focused on engineered human skin constructs (HSCs) to recreate a more accurate physiological environment for in vitro studies of skin diseases, drug therapies, and biomolecular signaling, and to address the medical needs of patients with genetic skin diseases, and victims of burns and injuries. The HSCs currently available on the market lack cell diversity, specialized structures such as hair and sweat glands, and vascularization, which typically lead to poor viability and poor integration of the grafts. Our previous work^1,2 and that of others^3–5 showed that prevascularization of HSCs is essential to promote neovascularization in the grafts by the host vessels and to improve graft survival by maintaining the regenerative capacity of HSCs. The presence of endothelial cells (ECs) in HSCs may also contribute to the microenvironment signaling and regulation of immune response in the skin, narrowing the gap between the construct and real human skin.

ECs have been demonstrated to be tissue-specific for different organs of the human body.^6–12 A recent global single-cell RNA-sequencing analysis indicated that capillaries have the most specialized and diverse ECs, compared with bigger vessels, and revealed the existence of highly specialized EC subclusters within the same organ.¹² According to these findings, human dermal blood endothelial cells (HDBECs) are possibly the most physiologically relevant EC subtype to create vasculature in HSCs. However, the vast majority of the vascularized HSCs in the literature were made by using human umbilical vein endothelial cells (HUVECs),^13–17 instead of HDBECs,¹⁸ due to their availability, good angiogenic potential, and perhaps especially because of their historical widespread use in other bioengineering applications. For these reasons, HUVECS represent an important candidate EC type to be used in HSCs. Finally, the recently acquired knowledge on cell reprograming and stem cell plasticity has led to the ability to differentiate induced pluripotent stem cells (iPSCs) into induced ECs (iECs).^19–28 Compared with the other cell types they are more demanding in terms of cost and time, but they can be produced from a tiny biopsy of patient tissue, minimizing the problem of histocompatibility and graft rejection. Thus, iECs represent a clinically promising EC type for HSC vascularization, for patient grafting applications. However, in light of tissue specificity and heterogeneity of ECs,^6–11,29 it is still unknown whether iECs and HUVECs can perform as effectively as HDBECs in HSCs, necessitating further study to assess whether either are clinically relevant.

In this article, we present a side-by-side comparison of three endothelial cell subtypes: HUVECs, a commonly studied and widely available EC type; iECs, a clinically promising immune-compatible EC type; and HDBECs, skin-specific ECs, in terms of their angiogenic potential in 3D HSCs and their global advantage for skin graft survival and integration in immunodeficient mice. Overall, our data highlight the potential of HUVECs and iECs to be used as alternatives to HDBECs in engineered skin grafts to promote the short-term viability of prevascularized engineered grafts, although HDBECs display a solid capillary identity and better promote dermal and epidermal maturation.

Clinical Problem Addressed

Skin replacement therapy has classically consisted of using skin flaps or grafts taken from donor sites of the body, resulting in a loss of tissue from other areas. Although generally efficacious, this therapy often causes skin mismatch, and additional surgical wounds and scars. Furthermore, when the surface to cover is too extensive or the patient has a genetic disease affecting the skin, this approach is not practicable. For these reasons, in recent years many new protocols have been developed, and are successfully employed in clinics.^30–33 In this context, the presence of pre-established blood vessels within the graft is crucial to support viability and cell proliferation, allowing for faster integration with the surrounding tissue. Also, it paves the way for a full-thickness restoration of the skin. Taking into consideration the rapidly growing knowledge on tissue specificity of ECs, our goal was to understand by direct comparison which kinds of endothelial cells may suit HSCs. Our study represents a critical evaluation of alternative strategies for EC sourcing to establish an improved marketable vascularized model and provide patients with a full-thickness skin replacement in the near future.

Materials and Methods

Cell culture

Human fibroblasts (FBs) and keratinocytes (KCs) were isolated from newborn foreskin samples, and they were cultured (<passage 3) in FB medium (DMEM; Gibco #10566–016 with 10% fetal bovine serum, Gibco #16000–044) and EpiLife S7 (Gibco; #MEPI500CA) respectively. HUVECs and HDBECs (PromoCell; #C-12203 and #C-12211) were cultured (<passage 3) in Microvascular Endothelial Cell Growth Medium (PromoCell; #CC-3202).

iEC differentiation and culture

We differentiated an in-house line of dermal FB-derived iPSCs into ECs following the protocol of Patsch et al.¹⁹ that we slightly modified. Briefly, cells were plated on T75 flasks coated with Geltrex (Gibco; #2084839) at a density of 45,000 cells/cm² in mTeSR1 supplemented with ROCK inhibitor (EMD Millipore; #SCM075) [10 μM] and incubated overnight. The next day, the medium was switched to 36 mL of DMEM/F12 (Gibco; #10565018) and Neurobasal medium (Thermo Fisher Scientific; #21103049) in a 1/1 ratio, supplemented with N2 (Gibco; #17502001), B27 (Gibco; #12587010), CP21R7 [1 μM] (Sigma-Aldrich; #ADV638391855), and BMP4 [25 ng/mL] (R&D Systems; #314-BP); it was incubated for 3 days to obtain the mesoderm transition. On the 4th day, medium was switched to 18 mL StemPro-34 Medium supplemented with Glutamax-I (Gibco; #35050–061), Forskolin [2 μM] (Sigma Life Science #F3917), and vascular endothelial growth factor A (VEGFA) [200 ng/mL] (Peprotech; # 100–20) and it was replaced every day for 2 more days. No antibiotics were used during the differentiation. On day 7, cells were MACS sorted with anti-CD144 magnetic beads (Miltenyi; #130-097-857) and replated at 25,000 cells/cm² on T75 flasks coated with vitronectin. ECs were cultured in StemPro-34 Medium supplemented with Glutamax-I, and VEGFA [50 ng/mL], and they were passaged twice at 80% confluency with trypsin (Fisher Scientific; #25300062) before being seeded on the constructs.

Generation of vascularized HSCs

Three-dimensional HSCs were generated by placing on a deepwell insert (Corning; # 353092) 1 mL of uncrosslinked rat tail collagen type I [3 mg/mL] (EMD Millipore; # 08–115) as an acellular layer, followed by 3 mL of collagen solution mixed with 500,000 human FBs at passage 3 to recreate the dermis, following the protocol of Gangatirkar et al.³⁴ After culturing the dermis in FB medium for 7 days, human KCs at passage 3 were seeded on the top of the construct and cultured in Epidermalization medium for 5 days, and Cornification medium for 7 days.³⁴ Once the epidermis cornified, constructs were flipped and ECs were seeded on the top (150,000 cells/80 μL) and incubated for 30 min. The constructs were then flipped again and cultured in Microvascular Endothelial Cell Growth Medium 2 (Lonza; # cc-3202) supplemented with 25 ng/mL of VEGFA (Peprotech; # 100–20). The in vitro HSCs were kept in culture for 10 days with medium changes every other day.

HSC engraftment onto mice

All experimental animal protocols were approved by the Institutional Animal Care and Use Committee at Columbia University Medical Center. The day after ECs seeding, using the pinch cutting technique an area of mouse skin equivalent to the size of our HSCs (∼0.8 cm²) was excised from the dorsal antero-posterior midline surface of 10 week-old male immunodeficient nude mice (athymic nude, Crl:NU(NCr)-Foxn1^nu; Charles River, Wilmington, MA). The HSCs were grafted on the surgical site with three interrupted suture stiches, and they were covered with devitalized mouse skin on the top. During the surgery, isoflurane (1–5%) and Carprofen (5 mg/kg) were used for anesthesia and analgesia, respectively. The surgical wound was then wrapped with a band-aid (an OpSite Flexifix Transparent Film) to assure extra protection and keep moisturization. The mice were kept for 14 days before being euthanized.

RNA expression analysis

The RNA expression of HDBECs, HUVECs, and iECs at P0 (HUVECs and HDBECs were purchased from PromoCell, and they underwent three doublings from isolation to commercialization; we considered these cells at P0 on receipt) was assayed through quantitative RT-qPCR with QuantStudio 7 Flex Real-Time PCR System following the manufacturer's instructions. We used as master mix TaqMan Universal MasterMix II, no UNG (Applied Biosystem # 4427788), and Taqman probes for GAPDH (Hs99999905_m1), ICAM1 (Hs00164932_m1), PECAM1 (Hs00169777_m1), VECAD (Hs00901465_m1), vWF (Hs01109446_m1), eNOS (Hs01574659_m1), KLF2 (Hs00360439_g1), VEGFA (Hs00900055_m1), TGF1B (Hs00998133_m1), COL1A2 (Hs01028956_m1), COL3A1 (Hs00943809_m1), MMP2 (Hs01548727_m1), MMP9 (Hs00957562_m1), bFGF (Hs00266645_m1), SQLE (Hs01123768_m1), HMGCS1 (Hs00940429_m1), COL4A1 (Hs00266237_m1), COL7A1 (Hs00164310_m1), ACTA2 (Hs00426835_g1), LAMA3 (Hs00165042_m1), FN1 (Hs01549976_m1), KRT14 (Hs00265033_m1), EGR3 (Hs00231780_m1), KRT10 (Hs00166289_m1), KRT16 (Hs00373910_g1), IVL (Hs00846307_s1), and FLG (Hs00856927_g1). HUVEC and iEC gene expression is compared with that of HDBECs, and the variation is calculated through ΔΔCT formula.

Immunofluorescent staining of samples

Coverslips

HUVECs, HDBECs, and iECs were cultured for 4 days on coverslips coated with vitronectin (Gibco; #A14700) for iECs and Quick Coating Solution for HUVECs and HDBECs (Angio-Proteomie; #cAP-01) in Endothelial Cell Growth Medium 2 (PromoCell; #C-22111) until they reached confluency. The cell monolayer was then fixed in PFA 4%, and it was stained with anti-VECAD (Abcam; #ab232880), anti-ICAM-1 (Abcam; #ab179707), anti-vWF (Sigma Life Science; #F3250), and anti-KLF-2 (R&D Systems; #MAB5466) primary antibodies, AlexaFluor 594 anti-Rabbit (Life Technologies; #A11012) and AlexaFluor 594 anti-Mouse (Life Technologies; #A11005) as secondary antibodies, and DAPI (Life Technologies; #D21490).

Sections

After fixation in 4% PFA, in vivo samples were cut into halves: One half of each sample was cryopreserved in OCT (Fisher Health Care; #4585) and sectioned for further staining with hematoxylin and eosin (H&E) and IF. To evaluate epidermal integrity and proliferation, samples were incubated with primary antibodies anti-Loricrin (Abcam; #ab85679), anti-K14 (Biolegend; #906001), and anti-ki67 (Abcam; #ab15580), then with secondary antibodies AlexaFluor 594 anti-Rabbit, AlexaFluor 680 anti-Rabbit (Invitrogen; #A27042), and AlexaFluor 594 anti-Chicken (Invitrogen; #A11042), and DAPI. To assess angiogenesis, we used primary anti-CD31 (Abcam; #Ab28364), secondary AlexaFluor 594 anti-Rabbit, and DAPI.

Wholemounts

For the wholemount staining of in vitro samples, we utilized primary anti-CD31 and anti-Vimentin (Scbt; #sc-6260), secondary AlexaFluor 594 anti-Rabbit and Alexa fluor 680 anti-Mouse (Invitrogen; #A21058), and DAPI.

Live staining

The mouse tail veins were first injected with fluorescein-labeled ulex europaeus agglutinin I (100 μg/100 μL) followed after 10 min by Rhodamine-labeled Dextran (1 mg/100 μL). Twenty minutes from the last injection, the animals were euthanized and the grafts were excised with surgical scissors. Wholemount specimens were then imaged with a confocal microscope.

Quantification of in vivo tissue constructs

In vivo quantification was largely automated by using the software CellProfiler (Broad Institute, Cambridge, MA).³⁵ To quantify the percentage of Ki67-positive cells in the tissue construct, a pipeline was created that identified the total number of objects stained with DAPI and the total number of objects stained with Ki67 that were within the expected size threshold of nuclei in the image. Ki67⁺ nuclei were counted if a DAPI-stained object colocalized with a Ki67-stained object. The percentage of Ki67-positive cells in the epidermis was determined similarly by using preprocessed images containing only the epidermis. Vasculature area was determined by using a different CellProfiler pipeline that identified CD31-stained objects, merged touching objects, and finally filtered these objects by size to obtain a representative image of vasculature in each image while minimizing noise from nonspecific staining. The software then measured the total occupied area of these objects compared with the total area of the image.

Quantification of in vitro tissue constructs

Angiogenesis formation data were quantified from in vitro samples by using image software Cellprofiler and ImageJ. A z-stack image of the base of each in vitro wholemount-stained construct was taken. Cellprofiler was used to automatically identify CD31-stained vasculature, as described earlier, and to create a binary image of this vasculature for each slice of the z-stack. An average projection of all the binary images in the z-stack was then created. The vasculature projection was then analyzed by using the Angiogenesis Analyzer plug-in on ImageJ, and these data were then normalized to the number of endothelial cells in each image projection.³⁶ Mean vessel diameter was quantified by using ImageJ.

Statistical analyses

All assays were repeated at least in triplicate, and the data are presented as the mean ± standard deviation (SD). Results were analyzed through the software Prism 8 by using multiple unpaired t test, not assuming a consistent standard deviation for all rows. Deviations were considered statistically significant when p < 0.05.

Results

Molecular differences in early passage HDMECs, HUVECs, and iECs

To understand the molecular differences in HDBECs, HUVECs, and iECs, we performed RT-qPCR for 2D cultures of cells at P0 (we considered HUVECs and HDBECs at P0 on receipt; according to the manufacturer, they underwent three duplications from their isolation). We compared HUVECs and iECs gene expression to that of HDBECs by using the ΔΔCT formula. The overall expression of EC marker genes (Fig. 1a) was higher in HDBECs, reflecting their established EC identity. All markers were detected in HUVECs at lower levels except for eNOS. iECs highly expressed CD31 and vascular endothelial caderin (VECAD), but vWF was downregulated; they were also lacking eNOS and KLF2, which might indicate an ongoing transitional status. VEGFA and TGFB1 (Fig. 1b), genes involved in neovascularization, again were higher in HDBECs. The panel of extracellular matrix (ECM) genes (Fig. 1c) revealed variability in the expression of specific ECM genes by different EC types. HUVECs showed a slightly but statistically significant upregulation in the levels of COL1A2 transcript. COL3A1 was upregulated in both iECs and HUVECs by almost 200-fold, whereas COL4A1 was downregulated in HUVECs and upregulated in iECs around nine-fold. As expected, none of the ECs expressed COL7A1, a type of collagen typically expressed by basal KCs. Both laminin and fibronectin (FN) were downregulated or undetected in HUVECs and iECs. The ECM remodeling panel (Fig. 1d) shows downregulation of MMP2 for both HUVECs and iECs, whereas MMP9 exhibits strong upregulation in iECs. ACTA2 was upregulated by two-fold in iECs; conversely, HUVECs showed a two-fold downregulation. Finally, we tested for a set of metabolic genes along with bFGF (Fig. 1e), all of which were previously shown to be upregulated in skin ECs compared with ECs from other tissues,¹¹ and we found that the gene expression of HMGCS1 and SQLE was slightly, yet significantly higher in iECs and HUVECs, whereas bFGF was downregulated in both cell types.

Figure 1.

Differential gene expression of HDMECs, HUVECs, and iECs in 2D culture. Transcript expression of the different EC types in P.0 (3 doublings from isolation, according to the manufacturer, for HDBECs and HUVECs) assayed through RT-qPCR. The level of transcript is compared with that of HDBECs, using the ΔΔCT formula and expressed as a fold change. (a) General EC phenotype marker genes: ICAM1, CD31, VECAD, vWF, eNOS, and KLF2. (b) Genes involved in angiogenesis: VEGFA, TGFB1. (c) Genes encoding extracellular matrix proteins: COL1A2, COL3A1, COL4A1, COL7A1, ACTA2, LAMA3, and FN1. (d) Genes encoding proteins responsible for extracellular matrix remodeling MMP2, MMP9, ACTA2. (e) Genes reported by Chi et al. as. skin-specific EC genes involved in cellular metabolism: bFGF, SQLE, and HMGCS1. bFGF, fibroblast growth factor 2; EC, endothelial cell; HDBEC, human dermal blood endothelial cell; HUVEC, human umbilical vein endothelial cell; SQLE, squalene epoxidase; TGFB1, transforming growth factor beta 1; VECAD, vascular endothelial caderin; VEGFA, vascular endothelial growth factor A; vWF, von Willebrand factor. Color images are available online.

The IF staining on 2D coverslip-cultured cells (Fig. 2) confirmed some results of the RT-qPCR and iECs differentiation (Supplementary Figs. S1). VECAD (Fig. 2a) was roughly two-fold more intense in HDBECs compared with HUVECS and iECs. ICAM1 (Fig. 2b) and vWF (Fig. 2c) similarly showed the highest intensity in HDBECs, about 60% in iECs and 30% in HUVECs. This data further support the iECs' commitment to the endothelial lineage and highlight the differences in these EC subtypes in 2D cultures at early passages before their integration into HSC for vascularization.

Figure 2.

Immunofluorescent analyses of HDMECs, HUVECs, and iECs in 2D culture. IF staining of three main EC markers, differentially expressed by HDBECs, HUVECs, and iECs. (a) anti-VECAD (red) and DAPI (b) anti-vWF (red) and DAPI (c) anti-ICAM1 (red) and DAPI. Cells were cultured on coverslips for 4 days until they reached confluency. Scale bars represent 20 μm. (d) Quantification of fluorescence intensity normalized by cell nuclei. HSC, human skin construct; IF, immunofluorescence. Color images are available online.

Performance of HDBECs, HUVECs, and iECs to vascularize engineered HSCs

After confirming the molecular differences in 2D cultures of three EC subtypes in vitro, we next tested their ability to self-organize to form vasculature within the HSCs. After generating the HSCs, we seeded the ECs (HDBECs and HUVECs at P3, iECs at P2) on the bottom and cultured them in EGM2-MV. As expected, on day 10 of culture, HDBECs and HUVECs formed an intricate vascular network (Fig. 3a), covering the lower layer of the dermis. iECs instead formed a few short vessel-like structures, lacking the extensive branching and connectivity observed for the other ECs. Vessels stained with anti-CD31 appear surrounded in close proximity by FBs stained with anti-vimentin (Fig. 3b). Computational analysis revealed that the total length was highest for HUVECs with an average of 11.7 mm and it was significantly low for iECs with an average of 6.9 mm, whereas there was no significant difference in the vessel length normalized by the number of cells between three cell types (Fig. 3c). The total number of nodes, which indicates the degree of branching, was higher in HUVECs than HDBECs and iECs. However, the mean number of nodes normalized by the number of cells was very similar for HUVECs, iECs, and HDBECs with an average of 1.28, 1.28, and 1.31, respectively. Notably, each endothelial cell type formed vessels of different caliber. The highest mean diameter was seen in HUVECs, 10–14 μm, followed by HDBECs, 8–12 μm, and iECs, 8.3–8.5 μm.

Figure 3.

Vascularization of 3D-engineered skin with HDBECs, HUVECs, and iECs. In vitro, HSCs vascularized with different EC types were cultured on transwells in deep-well plates for 10 days, before fixation and staining. (a) IF anti-CD31 (red) wholemount staining. (b) IF anti-CD31 (red), anti-Vimentin (cyan) and DAPI wholemount staining. (c) Quantification of total vascular length–vascular length/endothelial nuclei, number of nodes-nodes/nuclei, and average vessel diameter. Scale bars represent 200 μm (a) and 100 μm (b). Color images are available online.

Influence of the EC types on dermal and epidermal maturation in vitro

We performed RT-qPCR for the dermis and epidermis separately to assess the influence of different EC types on the dermal and epidermal development and maturation. The expression of all genes encoding for ECM proteins in the dermis (Fig. 4a) was significantly enhanced by the presence of HDBECs. The other conditions showed lower ECM expression except for COL1A2 and FN1 in non-vascularized HSC (NV-HSC). Laminin, in particular, was reduced by about 20-fold in NV-HSC, 35-fold in HUVEC-HSC, and 50-fold in iEC-HSC. The epidermal expression of ECM proteins followed a similar trend (Fig. 4b), although in this case iEC-HSC showed an increase in LAMA3 and FN1 by six- and eight-fold, respectively. KRT 14, a marker for basal KC, showed no significant difference, whereas KRT16, a marker for KC proliferation, was downregulated especially in NV-HSC (Fig. 4c). KRT10, EGR3, IVL, and FLG, markers for epidermal differentiation and integrity, were generally downregulated in all conditions compared with HDBEC-HSC. The only significant exception was IVL in iEC-HSC, which was similar to HDBEC-HSC.

Figure 4.

Differential dermal and epidermal gene expression of NV-HSC, HDBEC-HSC, HUVEC-HSC, and iEC-HSC in vitro. Dermis and epidermis were physically separated, and mRNA level of ECM proteins and epidermal proliferation and differentiation markers were assessed through RT-qPCR analysis. The level of transcript is compared with that of HDBECs by the ΔΔCT formula and expressed as a fold change. (a) Dermal ECM genes expression: COL1A2, COL4A1, COL7A1, ACTA2, LAMA3, and FN1. (b) Epidermal ECM gene expression: COL1A2, COL4A1, COL7A1, LAMA3, and FN1. (c) Epidermal markers of proliferation and differentiation: KRT14, KRT16, KRT10, EGR3, IVL, and FLG. ECM, extracellular matrix. Color images are available online.

Efficacy of engineered skin grafts vascularized with HDBECs, HUVECs, and iECs

To evaluate the efficacy of vascularization in vivo, we seeded the ECs on the bottom of HSCs as we did for the in vitro experiment and engrafted the HSCs on athymic nude mice the next day by using our previously established grafting strategy.^1,2

After graft excision on day 14 (Fig. 5a corner panel and Supplementary Fig. S2), newly formed blood vessels were clearly visible to the naked eye. The 3D rendering of wholemount samples (Fig. 5a main panel and Supplementary Fig. S3) shows a higher number of vessels from the host invading the vHSCs for all EC types compared with control HSCs without ECs, indicating the successful integration of the prevascularized grafts. We immunostained the sections of explanted HSCs for CD31 to assess neovascularization (Fig. 5b). Overall, HDBECs exhibit a capillary pattern, HUVECs larger vessels, and iECs an intermediate one. vHSCs with all EC types were significantly vascularized, whereas the vascularization in control was poor. Quantification of the total area covered by vasculature (Fig. 5c) places HUVECs as first with 9.75%, followed by HDBECs with 4.63%, and iECs with 3.67%, whereas the control was as low as 1.7%, highlighting the importance of prevascularization of HSCs for grafting and the dependence of degree of neovascularization on the EC type.

Figure 5.

Engraftment of vascularized HSCs with HDBECs, HUVECs, and iECs onto immunodeficient mice. vHSC, grafted on the dorsal antero-posterior midline surface of athymic nude mice (Crl:NU(NCr)-Foxn1nu; N = 4 per condition), were explanted 14 days after grafting. (a) Comparison of vascular bed in explanted grafts and IF wholemount displaying blood vessels stained with antiVECAD (green) invading the HSCs. The dash line demarcates the interface between the grafts and the mouse skin (b) IF staining of frozen sections with anti-CD31 (red). (c) Quantification of total area covered by vasculature. Scale bars represent 100 μm (b). Color images are available online.

Epidermis quality and integrity were not equal for HSCs with different ECs, as demonstrated through H&E staining (Fig. 6a), where the epidermis appeared thick in HDBECs and iECs, thinner in HUVECs, and very thin with a disrupted terminal differentiation in control. K14 and loricrin staining are in line with what we observed in H&E (Fig. 6b, c). The percentage of Ki67-positive nuclei in cells in the upper dermis and epidermis was highest for HDBECs (Fig. 7b) with 22.9% positive nuclei, followed by HUVECs with 21.5%, iECs with 18.3% compared with control with 2.4% of positive nuclei, substantially lower than vHSCs. In the epidermis alone (Fig. 7c), HUVECs showed the highest epidermal proliferation with 32.5% Ki67⁺ cells; HDBECs followed closely with 29.5%; and iECs had a lower percentage of 22.9%, whereas the control had almost no proliferative KCs (∼0.3%).

Figure 6.

Epidermal integrity, proliferation, and differentiation of the graft. Frozen sections of explanted grafts were stained to evaluate the integrity of the epidermis, the proliferation of basal keratinocytes, and the terminal differentiation in the stratum corneum. (a) H&E. (b) IF with anti-K14 (red) and DAPI. (c) IF with anti-Loricrin (cyan) and DAPI. Scale bars represent 100 μm for (b) and (c). Color images are available online.

Figure 7.

Epidermal Regeneration Capacity. Frozen sections of explanted grafts were stained for Ki67, and proliferative cell fraction was quantified. (a) IF with anti-Ki67 (red), scalebar 100 μm. (b) Total fraction of Ki67-positive cells in upper dermis and epidermis of control HSC, and vHSCs with HDBECs, HUVECs, and iECs. (c) Quantification of %Ki67-positive cells in the epidermis. Scale bars represent 100 μm (a). Color images are available online.

Discussion

The ever-growing literature describing organotypic vasculature, and the known heterogeneity within the same organ,¹² makes it necessary to screen for the efficacy of different EC types in the generation of organ models. Halaidych et al. previously compared side by side³⁷ the properties of HUVECs, HDBECs, and iECs in 2D culture and within Matrigel plugs grafted onto mice, concluding that iECs can generate functional blood vessels in vivo, but transplantation conditions and stromal cell source need to be improved to equal primary cell performance. Similarly, another group compared iECs with HUVECs in fibrinogen grafts and highlighted the need for increased iPSC-EC maturation for a realistic clinical translation.³⁸ Here, we compared for the first time the spontaneous vasculature formation of HDBECs, HUVECs, and iECs within HSCs in the context of skin replacement therapy. The in vivo performance of all three cell types in HSCs was in agreement with those found in previous studies.^37,38 However, the angiogenesis capability of iECs in vitro was lower in our HSCs, due in part to monolayer sprouting assays used in those studies instead of 3D angiogenesis in ECM.

The iECs differentiation protocols currently available^19–28 entail several challenges. Besides being potentially cost-prohibitive, the EC yield and maturity at the end of differentiation varies with different protocols. Primary ECs, such as HUVECs and HDBECs, benefit from tissue-specific microenvironmental cues and physical factors (e.g., shear stress) during differentiation. iECs, originating from FBs through iPSCs, lack such commitment, and may need the support of a biological tissue microenvironment in both the differentiation and vascularization processes. Our RT-qPCR analyses revealed the absence of eNOS and KLF2 in iECs, which could be due to the variable outcome of iEC differentiation protocols. The reduction in VEGFA and TGFB1 expression in both iECs and HUVECs can possibly reflect the immaturity of those cell types compared with HDBECs. The EC markers, instead, were generally expressed in the three lines.

Collagen type III is a homotrimer that is prevalent in hollow organs, such as large blood vessels, bowel, and uterus, and together with collagen type I is a principal constituent of ECM in the human body.³⁹ The strong upregulation of COL3A1 in HUVECs and iECs compared with HDBECs is consistent with the capillary identity of HDBECs. The higher expression of laminin and fibronectin, some of the principal constituents of ECM in the skin, in HDBECs is likely due to the dermal origin of these ECs. In iECs the increase of MMP9, which is involved in proinflammatory angiogenic responses,⁴⁰ may be due in part to the activation of the WNT signaling pathway⁴¹ through GSK3 inhibition from CP21R7 during the differentiation protocol.

Despite the lack of significant data in literature on the skin-specific EC phenotype, Chi et al. previously described bFGF, SQLE, and HMGCS1 as metabolic skin-specific endothelial genes¹¹ compared with ECs from other organs. Interestingly, our data showed a higher expression of SQLE (involved in sterol synthesis), and HMGCS1 (also implicated in and sterol synthesis, and catabolism of valine and leucine) in HUVECs and iECs compared with HDBECs. A possible explanation of this finding is that those cells are juvenile phenotypes, not committed to any specific tissue, and with a faster metabolism than HDBECs. However, HDBECs outperformed HUVECs and iECs in terms of promoting the expression of dermal and epidermal ECM and maturation markers in HSCs, suggesting that HMGCS1 and SQLE expression alone may not be predictive of skin specificity or angiogenic performance in the skin and calling for identification of other skin-specific EC genes for humans.

The transcriptional analysis of ECM proteins in the dermis and epidermis of in vitro cultured HSCs highlights the influence of HDBECs within the skin microenvironment, and their role in enhancing the expression of skin-relevant ECM proteins, and the epidermal proliferation and differentiation. Considering the endothelial contribution to specific function of many organs,¹² this is not surprising, although the precise mechanisms through which this is achieved are yet to be discovered. It is possible that a more abundant and physiological ECM secretion contributed to basal KC maintenance and proliferation through the interaction with ECM receptors, such as integrins.⁴² A paracrine crosstalk between HDBECs and KCs, connected to either the skin specificity of these ECs or their capillary identity, might also play a role in skin homeostasis and needs additional investigation.

We further observed that the vascular development in HSCs in vitro was similar for HUVECs and HDBECs, whereas iECs were not able to self-organize into a cohesive network. Current iEC protocols in literature,^20–28 including the one used in our study,¹⁹ typically confirm the angiogenic function of iECs on 2D Matrigel, instead of physiologically relevant ECM components, such as collagen. Therefore, our data underline the importance of testing the functionality of iECs in the context of a relevant ECM or tissue microenvironment. Our in vitro HSCs resemble the real skin only partially, since they do not include many cell types and ECM proteins present in human skin, the multitude of hormones, growth factors, and other signaling molecules circulating throughout the human blood, nor the shear stress exerted by blood flow. We previously showed that the use of iECs in HSCs with a perfusable vasculature¹ was comparable to HUVECs. In that model, iECs were able to attach on a premade lumen surface and form an endothelial barrier, overcoming the need for iECs to spontaneously organize to generate a vascular network. Considering our previous and current results, having a premade lumen may be the best strategy to utilize the iECs generated through the currently available protocols. Especially since iECs can express adherent junction proteins, such as VECAD and CD31, they are able to form the endothelial barrier function once they attach on this premade lumen surface.

In our engraftment experiments in mice, all skin grafts proved to be viable 15 days after grafting, where the vascularized HSCs showed a much thicker and moisturized epidermis. HSCs with HDBECs appear the best for keeping the epidermis viable and proliferative, followed by HUVECs and iECs, respectively. Interestingly, despite the fact that vHSCs with HDBECs had a vascular area less than half of vHSCs with HUVECs, they developed a thicker and more proliferative epidermis, suggesting that the degree of vascularization may not be the primary predictor of the epidermal regenerative capacity. Despite these differences, all ECs significantly supported basal epidermal proliferation, elevating the fraction of Ki67-positive nuclei in the epidermis from 0.3% of the control up to 20–30%. Further analysis of the anatomy, lumen formation, and maturation of these vessels to demonstrate the capacity of the different ECs to recruit pericytes and their physical interaction and to assess the regenerative potency of these ECs would help to better characterize these engineered vessels.

Our data suggest that HUVECs can be a practicable alternative to HDBECs for vascularization, whereas iECs are promising but need more rigorous characterization and development before being an equally successful EC source for engineered skin constructs and grafts. It would be interesting to assess whether iPSCs derived from alternative sources, such as bone marrow, would generate iECs with different characteristics; although, due to clinical relevance, we kept our primary focus on the dermal FB-derived iPSCs. Previous work of Lanner et al.⁴³ suggested that it is possible to direct EC fate toward arterial or venous phenotypes through modulation of VEGF concentration. Developing a similar approach to establish a capillary identity may improve the tissue vascularization of iECs in the future. In addition, the culture or derivation of iECs in relevant tissue microenvironments, such as engineered organ models, may be an effective strategy to generate tissue-specific iECs.

Overall, our data highlighted the importance of prevascularizing engineered skin grafts to improve their viability and regenerative potential, and the use of HUVECs and iECs as an alternative cell source for skin grafting. However, their significant differences observed in ECM expression induction, vascular network formation, and vessel diameter, compared with skin-specific HDBECs, should be taken into account when considering the potential long-term effects on skin regeneration and homeostasis, and they should be further investigated before any realistic use in skin replacement therapy.

Innovation

To our knowledge, this study represents the first direct and quantitative comparison of HDBECs, HUVECs, and iECs in the context of their angiogenic performance in HSCs cultured in vitro and engrafted onto mice. We demonstrated the differences between three EC types in vascularization of HSCs both in vitro and in mice, highlighting the importance of EC sourcing for engineered skin constructs and grafts.

Key Findings

• HDBECs, HUVECs, and iECs exhibit differential gene expression at low passages in in vitro culture.

• HDBECS and HUVECs can vascularize HSCs in vitro similarly, although there are differences in the mean vascular diameter, whereas iECs resulted in less organized network formation.

• The transcriptional analysis of genetic markers for dermal and epidermal maturation and differentiation revealed that HDBECs promote the expression of these genes compared with the other EC types and the non-vascularized control.

• HDBECs, HUVECs, and iECs can sufficiently promote the neovascularization in engineered skin grafts to maintain graft viability.

• Although HUVECs and iEC performed similarly to HDBECs in vivo, HDBECs still appear to be the EC type that provides the best epidermal integrity and regeneration.

Abbreviations and Acronyms

ACTA2

actin alpha 2 (smooth muscle)

bFGF

fibroblast growth factor 2

CD31

platelet and endothelial cell adhesion molecule 1

COL1A2

collagen type I alpha 2 chain

COL3A1

collagen type III alpha 1 chain

COL4A1

collagen type IV alpha 1 chain

COL7A1

collagen type VII alpha 1 chain

EC

endothelial cell

ECM

extracellular matrix

EGR3

early growth response 3

eNOS

nitric oxide synthase 3

FB

fibroblast

FLG

filaggrin

FN1

fibronectin 1

H&E

hematoxylin and eosin

HDBEC

human dermal blood endothelial cell

HMGCS1

3-hydroxy-3-methylglutaryl-CoA synthase 1

HSC

human skin construct

HUVEC

human umbilical vein endothelial cell

ICAM1

intercellular adhesion molecule 1

iEC

induced pluripotent stem cell derived endothelial cell

IF

immunofluorescence

iPSC

induced pluripotent stem cell

IVL

involucrin

K14

keratin 14

KC

keratinocyte

KLF2

kruppel-like factor 2

KRT10

keratin 10

KRT14

keratin 14

KRT16

keratin16

LAMA3

laminin subunit alpha 3

RT-qPCR

real-time quantitative PCR

SQLE

squalene epoxidase

TGFB1

transforming growth factor beta 1

UEA1

ulex europaeus agglutinin I

VECAD

vascular endothelial caderin (cadherin 5)

VEGFA

vascular endothelial growth factor A

vHSC

vascularized human skin construct

vWF

von Willebrand factor

Acknowledgments and Funding Sources

The authors thank Ming Zhang, Emily Chang, and Wan Huang for their technical assistance in histology and mouse engraftment studies. This project is partially funded by the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS: 5K01AR072131–02), and Ines Mandl Foundation Connective Tissue Grant.

Author Disclosure and Ghostwriting

No competing financial interests exist. The content of this article was expressly written by the author(s) listed. No ghostwriters were used to write this article.

About the Authors

Alberto Pappalardo is an MD and works as a postdoctoral researcher at the Department of Dermatology, at Columbia University Medical Center. He is interested in tissue engineering in dermatology, stem cells, and immunology.

Lauren Herron, BS, is a research technician in the Department of Dermatology at Columbia University Medical Center. Her research interests are tissue engineering and immunology.

David E. Alvarez Cespedes is an MD/MS candidate in Biomedical Engineering at Columbia University continuing his research in Dr. Abaci's lab. He is interested in using tissue engineering for clinical applications in dermatology.

Hasan Erbil Abaci, PhD, is an Assistant Professor at the Department of Dermatology at Columbia University Medical Center. Dr. Abaci's work is essentially focused on the recapitulation of skin and tissue-specific vascular microenvironments using the state-of-the-art microfabrication tools, such as microfluidics and 3D-printing.

Supplementary Material

Supplementary Figure S1

Supplementary Figure S2

Supplementary Figure S3