UVA exposure additionally degrades the dermal matrix through decreases in procollagen synthesis and increases in MMP-1, -3, and – 9 expression. Photoaged skin has also displayed reduced dermal vasculature and dermal connective tissue breakdown and disorganization in human explant cultures. Human Skin Equivalents are in vitro tissue models that have been previously used for studies on photoaging, wound healing, skin development, alopecia, disease, stem cell renewal, and toxicology screening research.Traditionally-used animal models such as rabbits, pigs, mice, and rats have different anatomies than humans, and so do not accurately model human healing/recovery rates. In contrast, the self-assembly development of HSEs has similar development pace to normal human skin. Although HSE research has been well developed to recreate the dermal and epidermal layers using fibroblasts and keratinocytes, more complex co-culture systems are needed to recapitulate human anatomy more closely and mimic trophic factor exchange of different cell populations in vivo. Building on our previously published protocol generating vascularized human skin equivalents, here we demonstrate inclusion of a hypodermis,square pot which we term adipose and vascular human skin equivalent , and demonstrate suitability for UVA photoaging studies.

Multi-cellular skin models similar to this AVHSE have been previously explored but with fewer cell types, much shorter culture lengths, and little to no volumetric characterization. UV photoaging has been previously investigated with in vitro skin models of the epidermis, keratinocytes in 2D, dermal fibroblasts in 2D, adipose components in 2D. This work combines photoaging studies with comprehensive in vitro skin models and allows for volumetric quantification of epidermal, dermal, and hypodermal components through volumetric imaging . Further, the effects of photoaging on adipokine and inflammatory cytokines have been quantified using ELISA.AVHSE cultures were created using N/TERT1 human keratinocytes , HMEC1 human microvascular endothelial cells primary adult human dermal fibroblasts , and ASC52telo adipose derived mesenchymal stem cells. All cell lines were routinely cultured at 37 °C and 5% CO2; all media blends given in supplemental Table 4.1. N/TERT1 cells have been shown to maintain normal epidermal behavior in previous organotypic skin cultures. N/TERT1 cells were grown up in a modified K-SFM media blend including K-SFM base, 0.2 ng/mL endocrine growth factor , 25 µg/mL bovine pituitary extract, 0.3 mM CaCl2, and 1% penicillin/streptomycin . N/TERT1 were routinely passaged once 30% confluence was met to prevent undesired differentiation in 2D cultures. HMEC1 cells were grown up in MCDB1 base media with 10 mM L-glutamine, 1 µg/mL hydrocortisone, 10 ng/mL EGF, 10% FBS, and 1% PCN/STREP. HMEC1 cells at passages 9 and 11 were used. HDFa were originally expanded in fibroblast basal media supplemented with fibroblast growth kit as outlined by manufacturer. For short term expansion upon pull up for AVHSE cultures, HDFa cells were grown up in DMEM supplemented with 5% FBS and 1% FBS.

ASC52telo was used to generate the adipose component of the skin construct. Cells were originally expanded in mesenchymal stem cell basal media with added supplements from a mesenchymal stem cell growth kit and G418 at 0.2 mg/mL; this was used as the 2D culture media until adipogenesis induction. Adipogenesis media was administered once ASC52telo plates were ~90% confluent . Thickness of epidermal layers were automatically detected from confocal images via thresholding differences using a custom analysis algorithm designed in MATLAB . For each sample, five confocal sub-volumes in the center of the AVHSE were used to detect thickness . An average thickness was found for each XY position to obtain a volumetric thickness indication rather than from a single cross-sectional position or from max projection. Briefly, epidermis was localized using DRAQ7, cytokeratin 10, and involucrin stains. Noise was removed using median filters applied to each XY-plate and intensities were scaled by linear image adjustment. Background auto-fluorescence was removed using rolling ball filters on each XY plane and the epidermis was segmented using hysteresis thresholding. Gaps in the epidermal binary volume were removed via morphological closing and opening with a disk structuring element. The resulting binary volume created a computational plane from which the top and bottom difference could be calculated and metrically scaled by appropriate voxel size. Intensity comparison of the suprabasal markers, Cytokeratin 10 and Involucrin was completed across all samples using confocal images. A maximum projection image of ten positions per sample was generated and average intensity values were calculated. For all epidermal quantification, three biological replicates with one/two technical replicates each were used for analysis, n = 5 for each groupAdipose thickness, volume fraction , and integrated intensity quantification was completed from 10 confocal sub-volumes per each sample . VF is an estimate of the volume that adipose takes up within the hypodermis and dermal space together. Volumetric thickness was calculated using localization of the BODIPY mature adipose marker, as described for epidermal thickness quantification. Integrated intensity of BODIPY was quantified via custom algorithms. Briefly, image sub-volumes were segmented and the resulting binary masks were used to isolate BODIPY stain from background noise and autofluorescence. The sum of raw intensity along the z-axis was calculated for each sub-volume within its binary map, then all sub-volume values were averaged as a metric of the whole sample volume. These data were gathered from images taken in the 3rd imaging phase .

Three biological replicates with two technical replicates each were used for analysis, n = 6 for each group. Vascular quantification parameters of diameter, VF, and diffusion length were determined from the average of 6 confocal sub-volumes per each sample . Using the Collagen IV marker from cleared AVHSE structures , vessels were located through segmentation and ultimately edge detection via an enhanced hessian based frangi filter written for vessel detection. VF was determined using the resulting volume segmentation. After locating vessel location, segmentation data was skeletonized through a fast marching algorithm and the center of each vessel was found which allows for calculations of diameter and Rk . Briefly,square plastic planter diameter was quantified by calculating the difference in pixel location of the vascular segmentation orthogonal to the vessel skeleton and Rk was obtained by determining the Euclidean distance between each point in the collagen volume and the nearest point on the network . The vascular processing algorithm is given in supplementary appendix chapter 4. Two biological replicates with two technical replicates each were used for analysis, n = 4 for each group. Biological and technical replicates are noted for each quantification. Pairwise comparisons of control v. photoaged samples were completed through two-tailed t-test. ANOVA followed by Tukey’s HSD post-hoc test was used to test for statistically significant differences when applicable. Un-normalized data points are shown for comparison to tissue scale morphology. For statistical comparison, data were normalized to each biological control for epidermal, vascular, and adipose quantification. Significant differences of normalized data are plotted with p < 0.05 represented by a single asterisk; p < 0.01 represented by a double asterisk. Differences between ELISA values were run on raw data due to uneven biological/technical replicates; the same p-value representation is indicated. AVHSEs and the analysis techniques presented here enable study of skin volumetrically and at the tissue scale. We have shown that AVHSEs recapitulate epidermal, dermal, and hypodermal morphologies through co-localization of epidermal markers, vascular markers throughout the dermis, and co-localization of vascular and adipose markers in the hypodermis . The automated image analysis of the three skin compartments was completed on biologically large volumetric areas with minimum volumes of 1.85 x 0.37 x 0.25 mm to analyze the epidermis and up to 3.6 x 0.37 x 0.35 mm to analyze the hypodermis. Importantly, tissue analysis at this scale provides a higher degree of precision over histological analysis in understanding skin as a whole. Prior studies have demonstrated decreased adipokine production during photoaging, and adipokines are mediators of the dermal photoaging mechanism. To test if the AVHSE cultures were similarly responsive to UVA, we measured production of adiponectin using ELISA. AVHSE cultures were prepared and maintained through ALI as described in the methods. After 7 weeks of ALI, AVHSE were exposed to 7 days of UVA , or left as controls. Media supernatant was collected from both photoaged and control samples after UVA exposure. Adiponectin expression was significantly reduced, in agreement with prior in vivo studies. This was not accompanied by a general inflammatory response or increased matrix metalloproteinase-1 presence, as indicated by stable IL-6 and MMP-1 expression .Photoaging by UVA largely acts on the dermal and hypodermal portions of the skin rather than the epidermis, in contrast to UVB which shows epidermal toxicity. To assess any changes in epidermal morphology, we stained suprabasal markers along with the nuclear stain DRAQ7 to assess epidermal thickness. No statistically significant differences were observed in the staining intensity of involucrin and cytokeratin 10 , or in the overall thickness of the epidermis , when comparing the control and the photoaged AVHSEs. These data are consistent with the minimal in vivo effects of UVA on the epidermis. Prior studies have shown dermal vascular damage is associated with chronic UVA exposure, as determined from sun-exposed skin biopsies from young v. aged individuals.

As a proxy for vascular damage, we quantified overall morphology in the AVHSE. Vascular structures were identified through localization of collagen IV . Formation of well-developed vascular networks was observed throughout the dermal layer and hypodermis as shown in both the en face and orthogonal projections. Imaging for vascular quantification was performed after tissue clearing, to minimize the loss of signal deeper in the confocal volume. The 3D rendering shown is representative of the vascular network segmentation and skeletonization that was made possible with cleared tissues . Importantly, these techniques are possible with uncleared images as well48 but clearing increases the volume of tissue that can be quantified. Vascular network diameters are representative of the inner vessel diameter and were quantified as 6.45±0.14 μm for control and 6.34±0.12 μm for photoaged . Volume fraction ofvasculature had median values of 0.037±0.01 and 0.032±0.007 . No statistical difference was determined in comparison of diameter or vascular VF. Diffusion length was calculated with median values of 73.16±23.75 and 83±29.36 microns . A significant increase in diffusion length of photoaged AVHSEs was detected which corresponds to a slight non-significant decrease in VF of photoaged samples. Increased VF and decreased Rk is preferable in metabolic tissues and data here reflects that preference in control samples v. photoaged. Hypodermal adiposity is reduced with photoaging Prior in vivo studies have shown reduced lipid synthesis and lower amounts of fat in hypodermal adipose associated with UVA photoaging . To test if this was mimicked in the AVHSE model, we used confocal imaging of the lipid stain BODIPY in both controls and photoaged AVHSE. Representative images shown in Figure 4.6 A show decreased staining intensity and representative volume renderings are shown in Figure 4.6 B. To quantify adiposity, we utilized two morphological measures as well as the integrated intensity. Both morphological measures exhibit subtle declines, but the results are non-significant . However, the overall stain intensity was significantly decreased , indicating an overall loss of lipid content in the photo-aged AVHSE. Skin provides critical barrier, insulation, and homeostatic functions in human physiology; these are known to be disrupted in aging. Despite the importance, research is limited by the accessibility of physiologically relevant models, with conventional culture methods lacking the structure and organization of the overall tissue11 and conventional animal models presenting key differences from human aging physiology. To address this, human skin equivalents have been previously established as valuable models in the study of skin and aging; however limitations remain. Of special relevance, loss and dysregulation of hypodermal adipose is implicated in physiological aging and aging-associated diseases including lipoatrophy associated with insulinresistant diabetes mellitus. This dysregulation is poorly captured in current HSE. To address this, we have developed a robust and reproducible HSE that includes adipose and vascular components . This methodology builds off of previous studies and provides a model to study crosstalk between adipose, vascular, stromal, and epithelial components of skin in the context of aging. Further, this model is tissue-scale, stable for long culture durations, and suitable for aging studies. Other researchers have reported that when skin models are cultured with adipose tissue, after 2 weeks of culture, there was epidermal disintegration and that 7 days is enough time at ALI to produce a fully functional skin equivalent.