A human skin model to evaluate the protective effect of compounds against UVA damage.

Objective: The main function of skin is to protect the body from external aggressions. Over time, normal skin ageing is accelerated by external stresses such as smoking, pollution, chemical products and radiation. UV light, in particular UVA, causes DNA damage, apoptosis and morphological modifications, which are responsible for both premature ageing and cancer. The aim of this study was to establish a discriminatory and sufficiently reproducible cutaneous model for evaluating UVA damage, to enable testing for effectiveness of potentially protective compounds. Methods: The cutaneous model is based on Human skin explants irradiated with UVA. Deleterious effects on epidermis were observed and quantified by haematoxylin–eosin staining and by immunofluorescence of ɣ‐H2Ax, cytokeratin 5, involucrin and loricrin protein. Dermis deterioration was evaluated by transmission electronic microscopy and zymography in situ. Results: We were able to observe and quantify deleterious effects associated with UVA irradiation: epidermal and dermal disruption, appearance of Sunburn cells, increased DNA damage and induced apoptosis. The use of this model in the evaluation of protective compounds was first confirmed using sunscreens, then further validated with a panel of active ingredients which showed beneficial effects on epidermis morphology and DNA integrity after UVA exposure. Conclusion: We have developed a model and a standardized protocol, based on the use of human skin explants, which allows us to explore the protective effect of active ingredients to environmental stresses such as UVA.

Résumé: Objectifs: La fonction principale de la peau est de protéger le corps des agressions externes. Au cours du temps, le vieillissement naturel de la peau est accéléré par des stress externes comme la cigarette, la pollution, les produits chimiques et les radiations solaires. Le rayonnement ultraviolet, en particulier les UVA, cause des dommages de l'ADN, l'apoptose et des modifications morphologiques qui sont responsables du vieillissement prématuré et de cancers. Le but de cette étude est d'établir un modèle cutané reproductible et discriminatoire pour évaluer les dommages créé par les UVA et tester l'efficacité de potentiels produits protecteurs. Méthodes: Ce modèle cutané est basé sur un explant de peau humaine irradié aux UVA. Les effets délétères sur l'épiderme sont observés et quantifiés par coloration Hématoxyline‐éosine et par des immunofluorescence des protéines ɣ‐H2Ax, cytokératine 5, involucrine et loricrine. Le détérioration du derme est évaluée par microscopie électronique à transmission et par zymographie in situ. Résultats: Nous avons observé et quantifié des effets délétères associés aux irradiation UVA: détérioration de l'épiderme et du derme, apparition de cellules "coup de soleil", augmentation des dommages de l'ADN et l'induction de l'apoptose. L'utilisation du modèle pour tester des nouveaux composés a été premièrement validée avec l'utilisation d'un filtre solaire puis validée par le test d'un panel d'ingrédients actifs qui ont montré des effets bénéfiques sur la morphologie de l'épiderme et l'intégrité de l'ADN après exposition aux UVA. Conclusion: Nous avons développé un modèle et un protocole standardisé basé sur l'utilisation d'un explant de peau humaine qui permet d'explorer l'effet protecteur ingrédients actifs contre des stress environnementaux comme les UVA.

Keywords: UVA; skin explant; skin barrier; skin physiology/structure; safety testing

(a) Representative images of H&E staining acquired from sections of paraffin‐embedded human skin explants: control (foil protected), post UVA irradiation, or UVA irradiation with sunscreen (X40, scale bars 100μm, *stratum corneum, **epidermis, ***dermis). (b) Percentage of sunburn cells. ***P‐value < 0.001. Observation on 10 skin donors for control and irradiated skin, and 5 donors for irradiated with sunscreen skin.

The main function of skin is to protect the body against external aggressions. Its specific structure, organized in superposed layers, provides an effective outermost biological barrier [1]. Over time skin deteriorates in a process called intrinsic skin ageing. This phenomenon is accelerated by external stresses such as smoking, pollution, chemical products, and particularly by UV irradiation, which constitute extrinsic ageing.

Solar UV rays are divided into UVC (100–290 nm) stopped by the ozone layer, UVB (290–320 nm, 5–10% of total UV) and UVA (320–400 nm, 90–95% of total UV), penetrating the ozone layer. UVB rays are directly absorbed by the epidermis and induce DNA damage in keratinocytes, such as formation of mutagenic DNA pyrimidine dimers and DNA photoproducts [2]. UVA rays, which can reach the deep dermis, are characterized by the generation of reactive oxygen species (ROS) which then lead to oxidative stress causing DNA damage similar to UVB and DNA strand breaks [3]. If the dose of UV exceeds a damage response threshold, keratinocytes activate apoptotic pathways identified by chromatin condensation resulting in pyknotic nuclei known as sunburn cells (SBC) [4]. UVA causes also the degradation of extracellular matrix of the dermis by activation of matrix metalloproteinases (MMP) and the inhibition of collagen synthesis. Both UVA and UVB are responsible for pigmentation, photoageing, photoimmunosuppression and photocarcinogenesis [5]. Since UVB rays are more energetic than UVA, their impact has been extensively studied and described [[6]]. But the penetrating properties of UVA, and their greater proportion, make them a harmful factor whose deleterious effects have long been underestimated [[10]].

In this context, it is necessary to develop in vitro skin models, for studying UVA damages in order to evaluate the effectiveness of active ingredients to protect the skin, thus helping to avoid testing on volunteers.

Some cutaneous models make it possible to study UVA effects and to test dermatological products. The simplest one is a two‐dimensional (2D) keratinocytes or fibroblasts cell culture, but which lacks the complexity of three‐dimensional (3D) skin structure and environment [13].

A more satisfactory model is reconstructed epidermis or reconstructed skin. These models are generated by culturing human skin cells to create a stratified epithelium. For reconstructed skin, the keratinocyte differentiating process and its environment are preserved thanks to the integration of a matrix representing dermis. This model is reproducible and can be used in toxicology screening [14]. Cells can be genetically modified before epidermis reconstruction to reproduce dermatologic diseases for studying these treatments [15]. It has already been used to study the effect of UVA, ROS, DNA damage, fibroblast apoptosis and lipid peroxidation. Links between observed UVA‐induced impact and clinical consequences of exposure were proven. These molecular events thus confirm the contribution of UVA to long‐term harmful effects of UV exposure and underscore the need for adequate UVA photoprotection [[16]]. However, this model is expensive and its 3D structure does, nonetheless, differ slightly from human skin [18].

In another current model, human skin explants are obtained following abdominoplasty surgery and maintained alive for several days under specific culture conditions [19]. Native 3D structure, appendages, cellular environment and metabolism are conserved. The limitation is that the nervous and vascular systems are not functional [20]. The skin barrier being complete, the skin explant allows one to study the efficacy of pharmaceutical or cosmetic products by topical application. But only few papers report studies of UVA effects on 3D skin explants.

The aim of our work was to elaborate, based on human skin explant, an evaluation model and protocol to test the effectiveness of protection afforded by new products against UVA, under conditions that attempt to mimic, via repeated UVA irradiation, the in vivo effects of chronical exposure to the sun. We demonstrate deleterious effects on the model similar to clinical observations [17] like epidermal disruption, appearance of sunburn cells, increase of DNA double‐strand breaks, apoptotic keratinocytes and alteration of dermis organization. We evaluated the relevance of this model by applying sunscreen on skin explants exposed to UVA and confirmed its use. Finally, our model protocol was validated with respect to reproducibility and discriminatory power to quantify the protective benefits of new active ingredients and thus avoiding the use of human volunteers.

Human skin explants (Nativeskin®) were obtained from GENOSKIN (Toulouse). Experiments in this study were conducted using explants from a population of 10 skin donors, all women, aged between 27 and 66 years, with a phototype 2 on Fitzpatrick scale and originating from abdominoplasty as indicated by the supplier. After the surgery, the skin is conserved 24 hours at 4°C and transformed into a living explant via punch biopsy for incorporation into a solid support and a nutrient matrix (NativeSkin®). The human skin explant is maintained in culture at 37°C, 5% CO2, in an incubator during 5 days with specific medium provided by the supplier, renewed daily.

To determine an optimal UVA dose, we used morphological modification and apparition of DNA damage as indicators and adapted to technical constraints. Thus, the protocol turned out to require treating the skins with the test compounds 48 h before (pre‐treatment) and after irradiation (post‐treatment) and to distribute repeated irradiations during the 7 days of explant life. Consequently, the maximum number of irradiation steps was limited to 3 days. To define the dose, we tested the range from 20 to 80 J cm−2 and determined that a dose of 60 J cm−2 induced a minimum measurable deleterious effect (data not shown).

Thus, the protocol was established as an UVA irradiation at 60 J cm−2 with a Biolink® UVA lamp (main peak at 365 nm), at a distance of 16 cm from the skin, for a duration of 5h30, and a temperature of 36°C, applied to skin samples once a day for three days. The total dose is 180 J cm−2 of UVA. This total dose of UVA corresponds to one day of UVA irradiation (at 320–400 nm) in June in southwest France [21]. Samples were collected 24 h after the last irradiation. The negative control sample was a human skin explant that was not irradiated (maintained in the irradiator but protected with aluminium).

The sunscreen (homemade formulation, SPF30) was applied on the surface of the explant (2 mg cm−2) to protect irradiated explants. The sunscreen was applied during exposure and was changed every day. Before and after each application of the product, the skin is cleaned. The sunscreen is composed of 24.5% solar filter (octocrylen, butyl methoxydibenzoylmethane, ethylhexylsalicylate, homomenthylsalicylate, ethylhexylmethoxycinnamate), sunflower oil and Aerosil® 200 (Sigma‐aldrich, Missouri, USA). The absorbance spectrum of sunscreen is complete from 200 to 400 nm (data not shown). The control and irradiated skin were treated just with sunflower oil and Aerosil® 200. The sunscreen was tested on five donors.

Test compounds, provided by Syntivia S.A.S, were composed of vitamin B6 (Pyridoxamine) and various acid synthons (Table):

Composition of active ingredients

Test compounds were dissolved in DMSO (Dimethylsulfoxide, Sigma) at a concentration of 0.9 M. These solutions were then diluted to 0.9 mM with a Carbopol gel at 0.5% (ULTREZ 10®, Lubrizol, Ohio, USA). The solutions (10 µL) were applied before (24 h), during (3 days) and after (24 h) exposure and were changed every day (before and after irradiation). Before and after each application of the product, the skin is rinsed. The control skins were treated with Carbopol gel (ULTREZ 10®) at 0.5% and 0.1% DMSO. Compounds were tested on 1 donor.

Human skin biopsies were fixed in 4% neutral‐buffered formalin (Sigma‐aldrich) at room temperature (RT) 24 h after the last irradiation. Skin explants were dehydrated, paraffin‐embedded and cross sections of 5 µm thick were prepared using a Microtome. Haematoxylin and eosin (HE) staining was performed to identify skin components. Transmitted‐light images of the staining were acquired with an optical microscope (Leica, Germany, DMi1) and a Leica MC170HD camera, using Leica Application Suite (LAS®) for image capture. SBC were counted and compared to the total number of nuclei in each image. The percentage of positive SBC was determined from five images for each of the 10 donors for control and irradiated skin (n = 50), and from five donors for sunscreen‐protected irradiated skin (n = 25).

Paraffined skin sections were deparaffinised and hydrated. The antigen retrieval was performed in citrate pH6 at 90°C for 40 minutes. Sections were blocked with PBS (phosphate‐buffered saline) containing 2% bovine serum albumin, 30 minutes and incubated with primary antibodies anti‐ɣ‐H2Ax (Cell signalling, Leiden, Netherlands), or anti‐loricrin (Abcam, Cambridge, UK), or anti‐involucrin (Abcam) and anti‐cytokeratin 5 (Abcam) overnight at 4°C. A secondary antibody, Alexa Fluor 594 anti‐rabbit or anti‐mouse (Invitrogen Life Technologies, Carlsbad, USA), was then added for detection. Nuclei were stained with 4,6‐diamidino‐2‐Phenylindole (DAPI, Sigma‐Aldrich), and slides were observed with a fluorescence microscope (Leica, DM5000B). The percentage of positive cells was determined by image analysis (Image J) from 10 images for each of the 10 donors for controls and irradiated skin (n = 100) and from five donors for sunscreen‐protected irradiated skin (n = 50).

Human skin sections were deparaffinised and rehydrated with PBS. Tissue sections were permeabilized with proteinase K (20 µg ml−1) for 30 minutes at 37°C, washed with PBS and incubated with TdT buffer (Invitrogen) for 10 minutes at RT. They were then incubated with dUTP 16 biotin (Roche, Bâle, Swiss) and TdT recombinant enzyme (Invitrogen) for 1 h at 37°C. Finally, staining was revealed by Streptavidin TRITC (Invitrogen), and nuclei were localized by DAPI staining. The percentage of TUNEL‐positive cells was calculated for 10 images in all experimental conditions for three donors (n = 30).

We observed by transmission electron microscopy (TEM) a control skin and an irradiated skin, on one donor. Skins were fixed with a 2% glutaraldehyde solution and 0.1 M of phosphate buffer overnight at 4°C. Samples were rinsed with phosphate buffer and post‐fixed with a solution of 250 mM of sucrose and 0.05 M of phosphate buffer for 1 h in the dark at RT. Skins were dehydrated by increasing concentration of ethanol. The tissue was then included in an EMbeb 812 resin® (Electron microscopy sciences, Hatfield, UK). Finally, the sample was cut into a 70 nm thick section and mounted on a Collodion 100 coated copper grid prior to staining with 3% uranyl acetate in 50% ethanol and Reynold's lead citrate. Samples were imaged with a Hitachi HT7700 transmission electron microscope at an 80‐kV acceleration voltage.

The explants of three donors were frozen and cross sections of 5 µm thick were prepared using a cryostat (Leica CM1950). The sections were rehydrated in PBS, and the probe Protease Enzcheck® (Thermofisher, Waltham, Massachusetts, USA) at 200 µg mL−1 was added to each sample and incubated overnight at 37°C. Sections were rinsed with PBS, and the nuclei were stained with DAPI. The sections were observed with a fluorescence microscope (Leica DM5000B).

Data represent the mean ± standard deviation. Significant differences between averages were analysed with Student's t‐test at a P‐value of 5% (*P‐value < 0.05, **P‐value < 0.01, ***P‐value < 0.001, non‐significant). The comparisons were performed in relation to respective donors between control and treated skin.

We first examined the effect of UVA exposure on the integrity and organization of skin explants under the conditions of the described protocol. Following three days of UVA irradiation at 60 J cm−2 (Fig. a), major changes were observed only in the epidermis compared to control skins. A disorganization of the layer was observed, characterized by a modification of cell cohesion and superposition. Moreover, the thickness of the epidermis decreased. Chromatin condensation within the nuclei of the keratinocytes in the superior stratum is characterized by small dark nuclei and known as sunburn cells (SBC). Degradation of nuclei after UVA irradiation showed a significant increase of SBC (36%) when compared to controls (2.9%) or with sunscreen (7%) (Fig. b). When skin was protected by sunscreen, the phenotype is similar to controls, demonstrating the effectiveness of the former.

It is known that UVA causes DNA strand breaks [16] and histone H2AX becomes phosphorylated on serine 139 (γ‐H2AX) in response to the formation of DNA double‐strand breaks. This histone is commonly used as a sensitive marker of DNA double‐strand breaks. In order to detect DNA breaks [[22]], ɣ‐H2Ax immunostaining was performed on paraffin skin sections (Fig. a). The percentage of stained positive nuclei was determined (Fig. b). On control skin sections, only 1.6% of nuclei presented staining, a result similar to skin protected by sunscreen (3.8%). On the irradiated skin sections, we detected a significant increase in positive ɣ‐H2Ax stained nuclei (32.1%) in all layers of the epidermis (Fig. b). The results showed a significantly higher percentage of positive cells in irradiated skin compared to controls.

Moreover, it is known that the appearance of sunburn cells implies the beginning of the apoptosis process. To confirm the occurrence of this biological mechanism, a TUNEL assay was performed (Fig. a), and the percentage of stained positive nuclei determined (Fig. b). On irradiated skin, many nuclei (19%) in the intermediary epidermis layers were positive with TUNEL staining, revealing apoptotic cells. For protected skin, the percentage was similar to controls (2.2% and 1.4%, respectively).

The accumulation of micro‐injuries induced by UVA at the level of DNA and proteins can modify the proliferative capacity and the differentiation of keratinocytes and thus explain the morphological changes [24]. To observe this effect on our model, we examined the localization of three proteins involved in each layer of epidermis differentiation; cytokeratin 5 (KRT5) in the basal layer; involucrin (INV) in the granular layer and loricrin (LOR) in the stratum corneum. Their localization was visualized by immunostaining performed on paraffin skin sections (10 donors or five donors) (Fig.). Immunolabeling of the KRT5 is localized in the cytoplasm of the keratinocytes of the basal layer. The fluorescence, reflecting its expression within the tissue at a given time, is equivalent for all three conditions (control, irradiated, treated). Therefore, UVA does not seem to cause a modification of KRT5 under the conditions of this protocol.

We analysed the staining of involucrin, which is involved in the terminal and intermediate differentiation of the epidermis. It is localized in the cytoplasm of the keratinocytes from spinous to granular layers in the control and protected skin. The intensity of the labelling forms a gradient with increased fluorescence in the upper layers. The overall intensity of involucrin staining is decreased and heterogeneously present in some areas in the samples exposed to UVA. With sunscreen treatment, the location of involucrin is comparable to control skin.

To observe the effect of UVA on terminal differentiation, we analysed the location of loricrin in the stratum corneum. The staining is discontinuous in all samples examined.

In conclusion, we could observe that our conditions of UVA irradiation seem to alter the proteins of intermediate layers of the epidermis (involucrin) but not proteins in the upper or deeper layers.

At the molecular level, photoageing is characterized by an accumulation of disorganized elastin called solar elastosis and an alteration of the extracellular matrix with the degradation of collagen I by the activation of MMPs [25]. We examined whether our skin model exposed to UVA was able to reproduce this effect on the integrity of the matrix fibres. To this aim, we used TEM to observe structures at the ultrastructure level. A control explant and an irradiated explant from a single donor were fixed and processed to be imaged by TEM (Fig. a). Dermal analysis of the control sample shows a high concentration of collagen fibres. The fibres are contiguous and form a compact homogeneous structure.

In UVA irradiated skin, at ×500 magnification, papillary dermal fibres appear fragmented and small at the level of dermo‐epidermal junction. The density of the fibres is lower than in the control explant. In the reticular dermis, the fibres are dispersed and non‐cohesive. At ×1500 magnification, the transversal section of the fibres (white arrowhead) shows many small filaments that are peeling off the main fibre (feathering). This TEM analysis shows that the ultrastructural aspect and organization of the papillary dermis fibres are modified after the standardized UVA irradiation of our protocol. No quantification of the effects was attempted at this stage.

Many studies have shown that UV induces the activation of MMPs, dermal protease, by the AP‐1 pathway. UVA rays penetrating to the dermis are involved in this process.

We wanted to know if these enzymes are activated in this model. On the other hand, the use of a skin explant limits the techniques for analysing enzymatic activities compared to cell culture. Its 3D conformation reduces the accessibility of the substrate within the tissue that must be alive to maintain enzymatic activity. We therefore used the in situ zymography method on cryosection. The disadvantage of this technique is that it does not specifically target MMPs but global skin proteases (elastases, collagenases, etc.).

The explants were collected 24 h after exposure to UVA and frozen. Cryosections were made and the probe, revealing the enzymatic activity, was disposed on the sections. The probe becomes fluorescent if it is cleaved by proteases. The stronger the enzymatic activity, the more intense the fluorescence detected. The study was carried out on three donors. The explants protected by sunscreen protection were used as a reference (Fig. b).

In the explant control, fluorescence is detected in the epidermis and dermis. In the dermis, fluorescence is localized more particularly at the level of the matrix fibres. In irradiated skin, an increase of the fluorescence is visible in the epidermis and dermis. The explant with sun protection has the same phenotype as the control explant. This experiment shows us that during repeated irradiation with UVA the general activity of the proteases of the dermis is increased.

To validate the use of this model and our standardized protocol, we evaluated the protective effect of molecules of the pyridoxamine family from our chemical library. Previously, these compounds have shown results for genes encoding antioxidant enzymes. These compounds are amides composed of pyridoxamine reacted with different acids. These compounds do not absorb in UVA (data not shown).

We first determined the effective dose of product to be used in the assay by evaluating the effect of increasing concentration of the SV27 molecule, a promising agent validated by other experimental approaches. We found that a concentration of 0.9mM of SV27 protected both the morphology and DNA damages of keratinocytes, thus presenting itself as our benchmark.

We next compared SV27 with a panel of other compounds (SV31, SV37, SV229, SV313) at the concentration of 0.9mM on one donor.

On irradiated skin treated with SV31, epidermis morphology was protected from deleterious UVA effects. With SV27 and SV37, protection was partial since a few pyknotic nuclei still appeared (20.83% and 13.39%, respectively) compared to irradiated skin (42.3%). When skin samples were treated with SV229 or SV313, perforations appear and lots of pyknotic nuclei (47.6% and 58.8%, respectively) in the epidermis compared to control (1.4%) (Fig. a).

To study protection from DNA damage, we observed and analysed the percentage of ɣ‐H2Ax positive cells (Fig. b). The number of positive nuclei between controls (2.7%) and SV31 treated irradiated skin (4.4%) was comparable, but significantly different compared to irradiated skin (20.7%). A slight increase of positive cells was observed with SV27 (8.6%) or SV37 (6.5%) treatments. But for SV229 and SV313, there was no significant protective effect.

Protecting skin from solar irradiation has long been a challenge, given its responsibility in skin cancer and extrinsic ageing which accelerates chronological ageing. UVA irradiation, rather than UVB, because of its greater penetration depth into the skin, and more importantly because they are the dominant proportion of sunlight, make them a dangerous factor whose deleterious effects have long been underestimated until recently. But today the deleterious effects of UVA are known and demonstrated [26]. To understand the biology of skin, the effects of UVA and to test pharmaceutical/cosmetic products, different models have been developed.

Since 2013, the European Union has banned animal testing [13]. In the same spirit, seeking to limit human volunteer testing has made it more than ever necessary to develop non‐harmful cutaneous models. Recent advances in tissue engineering have resulted in such skin models and skin disease models to test bacterial infection, psoriasis, irritant and allergic contact dermatitis and melanoma [15].

In our research to provide a solution for photodamage, particularly of UVA origin, human skin explants, rather than volunteers or even the best skin equivalents, offer study conditions as close as possible to in vivo situations, preserving skin structure and metabolism. The explant provides access to information from the stratum corneum to the hypodermis. On the other hand, the 3D complexity and tissue density of the model limits biological analysis techniques and the duration of observation and possible treatment is limited to 7 days. This limited lifetime makes it only possible to monitor the evolution of the model after damage and to study natural repair mechanisms in the short term.

The other characteristics of human skin explant are the variability of donor tissue, that may impact the reproducibility of results. Moreover, because this model mimics real‐life situations, it is influenced by extrinsic stress, age and diseases of the donors and hence by unknown parameters. Despite these limitations, skin explants seem to be a good representative model of physiological human skin and final consumers of cosmetic products [27].

In this study, we propose a standardized model protocol which consists of controlled UVA treatment conditions for human skin explants specifically chosen to originate from abdominoplasty and with a phototype 2. We chose specific parameters for analysis (DNA damage, epidermal and dermal morphology, SBC counts), which we can be quantified or imaged and compared with benchmarks.

As explants come from abdominoplasty, they are thus less exposed to extrinsic ageing (UV, pollution, etc.) linked to the donor lifestyle. We postulated that the effects observed with our model protocol are therefore caused only by our standardized in vitro irradiation treatment.

The first effect visualized is DNA damages, DNA condensation, appearing as SBC and pyknotic nuclei, corresponding to the first step in the apoptotic process [28], already correlated with UV damage [5]. In this model, UVA damage is confirmed by the appearance of these characteristics. Data are available in the literature concerning SBC, skin morphology and DNA damage in skin explants, but data are often difficult to compare because of the wide variations in irradiation protocols. These results suggest that the model reproduces similar DNA damage to that described in vivo[[22]].

We studied also the impact of irradiation on the organization of the epidermis, by detecting proteins representative of epidermal differentiation, and the dermis. The alteration of proteins in the epidermis after UVA irradiation is poorly described in the literature. Whereas UVB irradiation studies appear to be contradictory with respect to involucrin and keratin modification [[8], [29]], we estimate that as well as our observations on loricrin, involucrin and KRT5, further studies on other differentiated keratinocyte proteins (such as keratin or filaggrin) might be of interest in our model.

Dermal degradation, by activation of the AP‐1 pathway which activates specific collagenase, MMP1, and inhibits collagen synthesis, is one of the characteristics of UVA's impact [30]. In order to confirm this characteristic signature in our UVA model, we carried out TME imaging of the organization of dermal ultrastructure and in situ monitoring of increased protease activity. We observed a fragmentation and a decrease of the extracellular matrix fibres. The dermal degradation observed with our model is consistent with in vivo results as well as the increase of the proteases activity in the dermis [31].

Strategies to fight photoageing are based on prevention, sun avoidance, protective clothing and sunscreens. Consequently, it is interesting to develop models to define the protective effects of sunscreens or other skincare products against UVA radiation. Our model of UVA photodamage was tested and validated by applying a protective‐sunscreen treatment during irradiation. Our results proved its effectiveness, along with the pertinence of the model in evaluating UVA protection.

The second way to fight photoageing is the use of specific preventive anti‐ageing products. After UVA irradiation, biological defects observed in human skin explants show some characteristics of this extrinsic ageing phenomenon [[25], [32]]. Consequently, we wanted to explore whether our model of UVA irradiation of human skin explants would also be appropriate to test potential preventive effects of anti‐ageing cosmetic treatments.

We evaluated five compounds in the pyridoxamine family which do not absorb in the UVA wavelength. It is known that pyridoxamine (vitamin B6) plays a crucial role as an essential cofactor in several metabolic transformations of amino acids. It is described as a scavenger candidate against glycation of proteins involved in skin ageing [[33]]. All five compounds were coupled with an acid to reinforce the beneficial effect of pyridoxamine.

The results of our studies demonstrate good sensitivity and discriminatory power of our model protocol, able to test the effects of different molecular structures in an exclusively ex vivo test. The observed differences in UVA protective power between these five substances are probably due to compounds penetration, varying ROS scavenging/antioxidant and chelating properties which will require more specific tests to substantiate in 3D skin. For further, it will be interesting to test the reproducibility of the results on several donors. Moreover, the study is incomplete because we have studied only 3 parameters (morphology, SBC and DNA damage) to compounds discrimination. Today, a study of the dermis morphology is necessary to visualize the effect in depth of the active ingredients. It would also be convincing to continue the study by observing parameters: epidermis and dermis protein expression, cells oxidation and skin inflammation.

Skin explants represent an increasingly valuable model, thanks to their 3D conformation, by making it possible to study genetic and molecular consequences resulting from the influence of exogenously added factors such as external damage and topical treatments [35]. They also replicate multiple aspects of in vivo skin biology due to the presence of all cell types as well as the natural microenvironment and the preservation of signalling pathways.

For the study of dermatological processes and their treatments, implementing 3D skin models into routine applications for drug development, developing effective standardized methods for their analyses and obtaining reliable readouts are crucial.

This statement is borne out by the relatively small inter‐individual variations observed in the data from the ten donor samples on morphology, DNA and protein damage, which were found within a narrow range under the conditions of treatment with the described protocol. While the study protocol described in the present paper does not claim or aspire to replace official sun protection standards, it is proposed as a tool to screen and discriminate novel candidates for UVA protection, allowing comparison between different classes of substances, studying mechanisms and developing further refinements.

The authors would like to thank Bernard Ducommun for this interest to this work. The authors thank Karl Lintner, Marine Norlund, Hilary Brooks and Gail Taillefer for their critical reading and Celine Damez for her help in producing compounds. We thank the centre of electron microscopy applied to biology of Toulouse Rangueil and ITAV imaging of the TRI‐Genotoul facilities. Sophie Abadie was supported by an ANRT CIFRE fellowship. The present study was financially supported by the CNRS and the University of Toulouse and by grants from the Occitanie region.

The authors declare that they have no conflict of interest.