Although we did not directly compare skin equivalents without adipose to AVHSEs here or directly compare culture time points, we have not observed any obvious changes in epidermal coverage compared to our previous work in vascularized human skin equivalents that do not contain a subcutaneous adipose compartment. While the model is customizable to study the effects of intrinsic and extrinsic aging factors, as a test case we have demonstrated suitability for studies in UVA photo aging due to the strong literature base of both in vitro and in vivo studies available for comparison. Finally, we demonstrated the accessibility of the model for both molecular and morphological studies . A key aspect of any HSE model is a differentiated and stratified epidermis. Here, N/TERT-1 keratinocytes were used to generate skin epidermis as previously completed. Importantly, N/TERTs are a suitable and robust substitute to primary keratinocytes which have disadvantages including limited supply, limited in vitro passage capabilities, and donor variability. HSEs generated with N/TERT keratinocytes demonstrate comparable tissue morphology, appropriate epidermal protein expression, and similar stratum corneum permeability when compared to HSEs generated with primary keratinocytes. Similar to prior models, we demonstrate AVHSEs appropriately model the skin epidermis with correct localization of involucrin , and cytokeratin , and nuclei localized in the lower stratified layers . Further, volumetric imaging and automated analysis allows for epidermal thickness to be robustly calculated. AVHSE present with median epidermal thicknesses within 90-100 µm, similar to values in both prior in vitro studies 100-200 µm and in vivo optical coherence tomography imaging of adult skin 59±6.4 to 77.5±10 µm 194. Consistent with prior in vitro and in vivo results showing UVA wavelengths predominantly impact dermal rather that epidermal layers, UVA photoaging resulted in no observable changes in epidermal thickness or expression of differentiation markers in AVHSE . In the dermis and hypodermis,square plastic plant pot skin is highly vascularized with cutaneous microcirculation playing important roles in thermal regulation and immune function.

Many prior HSE models have not included a vascular component4; however, there is increasing recognition of its importance. In the present work, we used collagen IV as a marker of the vascular basement membrane, enabling the automated segmentation and mapping of a vascular network within AVHSEs. The vascular VF of AVHSEs is lower than in vivo dermis , but prior work has shown this is tunable by using different cell seeding conditions . Optimizing the VF may be more involved in the AVHSE, since the ratio of adipose and vascular cells has been shown to be important in regulating tissue morphology. Thus, ratio of adipose and vascular cells would need to be optimized again for new cell and collagen densities. Adipose tissue is densely vascularized and the ability of adipocytes to generate lipid droplets and adipokines in the presence of endothelial cells is important to replicate the in vivo environment. Previous work has shown that in co-culture of endothelial cells and mature adipocytes can lead to dedifferentiation of mature adipocytes, but in homeostatic cultures ECs and adipocyte crosstalk is important. Through soluble factor release, ECs regulate lipolysis and lipogenesis and adipocytes regulate vasodilation and contraction. Secretion of adipokines by adipocytes aids vascular formation and adipose tissue stability. In prior work, Hammel & Bellas demonstrated that 1:1 is the optimal ratio for vessel network within 3D adipose, and we matched the 1:1 cell ratio in the present work. Quantification of vessel diameter in the Hammel & Bellas study shows that a 1:1 ratio of adipocytes to endothelial cells gives an average vessel diameter of ~10 µm235, our work supports this finding with a median inner vessel diameter of ~6 µm. Importantly, these data are within the range of human cutaneous microvascular of the papillary dermis . We did not observe morphological changes of VF and diameter within the vasculature due to photoaging. This is not entirely unexpected, as UVA exposure and its effects on vasculature are still poorly understood. While it is established that chronic UVA exposure can contribute to vascular breakdown, the duration of our studies may be too short to see this effect in diameter and VF . However, photoaging did induce an increase in diffusion length . Rk is a measure of the 90th percentile of distance from the vascular network and so a higher value corresponds to less coverage; values presented here match previous studies of vascularized collagen.

Rk of the vascular network for both control and photoaged samples was within the range of 51 – 128 µm which is importantly below the 200 µm diffusion limit. Upon photoaging, AVHSEs did demonstrate a significant increase in Rk compared to controls.In vascularized tissue, a high VF and low Rk is preferable and the Rk increase demonstrated indicates a loss in vascular coverage in photoaged AVHSEs. These findings conflict with studies of acute UV exposure in skin, which show stimulation of angiogenesis. It has been proposed that UV light exposure may improve psoriasis by normalizing disrupted capillary loops through upregulation of VEGF by keratinocytes. The AVHSE model could be used to more thoroughly test the effects of UV light and other molecular mechanisms it induces in future studies. The vascular networks extend from adipose to the epidermal-dermal junction , consistent with previous literature and to normal human skin histology/stereography. Further, we observed vasculature colocalized with the lipid droplet BODIPY staining , indicating recruitment of the vascular cells to the hypodermis. Importantly, the vascular networks in prior studies and the present AVHSE are self-assembled. While there are advantages to self-assembly, especially the simplicity of the method, it is important to note the limitations. Cutaneous microcirculation in vivo has a particular anatomical arrangement with two horizontal plexus planes, one deep into the tissue in the subcutaneous fat region and one just under the dermal-epidermal junction. Between these two planes are connecting vessels running along the apicobasal axis that both supply dermal tissues with nutrients and are an important part of thermoregulation. Although the AVHSEs presented here are fully vascularized up to the epidermal junction they do not recapitulate this organization. While not covered in this work, future studies could incorporate layers of patterned or semi-patterned vasculature to more closely match the dermal organization, depending on the needs of the researcher. In contrast to the epidermal and some vascular components, photoaging impacted the hypodermis. Volumetric imaging of BODIPY, which stains lipid droplets, was used to identify the adipose. While small reductions in the morphological parameters were observed, they were not significant, suggesting there was not large-scale necrosis or loss of fat mass. However, there was a significant decrease in the intensity of BODIPY staining, indicating decreased lipid levels. This is consistent with photoaging of excised human skin showing that UV exposure decreases lipid synthesis in subcutaneous fat tissue228. We further collected culture supernatant and tested for the presence of adiponectin, IL6, and MMP-1. The data collected through ELISA show that this AVHSE model secretes both adiponectin and IL6, which are also present innative skin and both considered important adipokines. Elevated serum adiponectin levels are linked to anti-inflammatory effects in humans and centenarians have elevated levels of adiponectin.Decreased adiponectin has previously been associated with photoaging in both excised human skin that was sun-exposed compared to protected skin and in protected skin that was exposed to acute UV irradiation. Conversely,25 liter square pot IL6 is a key factor in acute inflammation in skin, and has been shown to regulate subcutaneous fat function. In prior studies of photoaging, IL6 has demonstrated an increase after UVA irradiation in monolayer fibroblast cultures and excised human skin. IL-6 is released after UV irradiation and has been linked to decreased expression of adipokine receptors and mRNA associated with lipid synthesis, decreases in lipid droplet accumulation, and enhanced biosynthesis of MMP1. However, after one week of photoaging we did not observe an increase in IL-6 or MMP-1 via ELISA. The absence of changes in IL-6 and MMP-1 expression but decreases in lipid accumulation and adiponectin are not expected results but they could be due to methodology differences in UVA exposure. We determined our UVA dose and exposure based on literature values. The dose used here was 0.45 ± 0.15 mW/cm2with exposure for 2 hours daily for 7 d which roughly converts to 3.24 J/cm2 per day and a total of 22.68 J/cm2.

Many studies do not report exposure time and/or present ambiguous time points. This, compounded with the practice of using doses based on sample pigmentation threshold and broad definition of UVA wavelengths is likely contributing to the discrepancy in IL-6 and MMP-1 expressions. Previous work has shown that neutralizing anti-IL-6 antibody prevents UV induced decrease of important fat associated mRNA and that IL-6 secreted from keratinocytes and fibroblasts following UV irradiation inhibits lipid synthesis. From previous work, it is clear that IL-6 secretion is upregulated by UVA and presence contributes to negative adipose function but more investigation is necessary to understand what UVA doses and exposures induce IL-6 and further at what time points after photoaging are these expressions quantifiable. In this model, it is possible that there were increases in IL-6 that contributed to adiponectin decreases in photoaged samples, these trends may have been caught with different media collection time points. Alternatively, other analysis of inflammatory responses and adipokines may show generalized inflammatory responses identified in literature and further, changes in dose/exposure or continued photo aging may mimic the previously shown effects. There are notable limitations of the AVHSE model presented. Although we have presented a skin model that is closer to both anatomy and biology of human skin in comparison to past HSEs, we have not modeled skin fully through inclusion of other features of in vivo skin such as immune and nerve components. Including a functional immune system is important in understanding autoimmune diseases, cancer, wound healing, and decline of immune function in aged skin. Additionally, neuronal cell inclusion will allow for modeling of sensory processes necessary for grafting and modeling of skin disorders associated with nerve dysregulation. Further, while the cell lines used in this study were chosen for their low cost and accessibility, primary cells or populations differentiated from induced pluripotent stem cells would more closely match the physiology in vivo. While changing cell populations would likely require some adjustment to the culture system, we have previously demonstrated that cell types can be replaced with minimal changes. We model epidermis, dermis, and hypodermis here, but we do not model the depth that is present in thick skin tissue; to mimic thicker skin the model would need to be taller. As nutrient and waste diffusion in tissues is limited to ~200 µm, thick tissues will likely require perfusion to maintain throughout culture. Vasculature in thicker skin has higher diameters, especially in the lower dermis and hypodermis, these can be up to 50 µm. Finally, for ease of use, initial collagen density in the AVHSE model is 3 mg/mL, much lower than in vivo densities. Decline of collagen density is an important aspect of skin aging, correlating with skin elasticity and wound healing. Varying collagen density influences vascular self-assembly, but higher collagen densities are possible through a variety of techniques, including dense collagen extractions146, and compression of the collagen culture. By incorporating these tools, AVHSE could be modified to more closely represent the in vivo dermal matrix. Further, the AVHSE method was demonstrated with low serum requirements; but serum was used for initial growth and the cultures are maintained for weeks without serum. Serum replacements during the growth phase could potentially provide a chemically defined xeno-free culture condition in beginning culture stages for greater reproducibility and bio-compatibility. The presented AVHSE model provides unique capabilities compared to cell culture, ex vivo, and animal models. Excised human skin appropriately models penetration of dermatological products but there is limited supply and high donor variability; replacing excised human skin with animal models or commercially available skin equivalents is not the best course of action because of the differences such as varying penetration rates, lipid composition, lipid content, morphological appearance, healing rates, and costs; and limitations of customization. AVHSE can be cultured using routinely available cell populations, are cost effective, and are customizable for specific research questions. Further, the model is accessible for live imaging, volumetric imaging, and molecular studies, enabling a wide range of quantitative studies.