Groupe Icare

Innovative in vitro 3D skin model for safety assessment : a case report for the skin sensitization

Filaire E. 1.2., Mejri H. 3., Fernandes P. 3., Simon M. 1., Homs Corbera A. 3., Poinsot C. 1., Gaudriault P.3., Fassini D.3.

  1. Icare Group, Biopôle Clermont-Limagne, 6 rue Emile Duclaux, 63360 Saint-Beauzire, France.
  2. UMR 1019 INRAE-University Clermont-Auvergne, Human Nutrition Unity, ECREINT Team, 63000 Clermont-Ferrand, France.
  3. Cherrybiotech, 14 rue de la Beaunne, 93100 Montreuil.

*Corresponding author e-mail :   00330648035198


Prevalence of allergic contact dermatitis is creasing, justifying the need to assess skin sensitization potential for any chemical products. Current assessements of potential sinsitization events of new compounds are performed according to three guidelines (OECD 442 C, D and E) and a battery of in vitro tests. Despite the advances of skin model equivalents (SME) validated by the governmental organizations, none of them consider the interplay between the epidermis:dermis cells and the resident immune cells. The aim of our study was to develop a SME that includes an endothelial barrier and some relevant resident immune cells involved in the skin sensitization cascade.

The model integrated adult Normal Human Epidermal Keratinocytes, Normal Human Dermal Fibroblasts, Human Umbilical Vein Endothelial cells, THP-1 (a human leukemia monocytic cell line), and THP-1 derived macrophages, Structure, cells and tissue-specific markers were revealed by immunocytochemistry techniques.

Our approach resulted in a multilayered, differentiate epidermis proliferating on top of the dermis as revealed by immunolabelling analyses, Presence of specific markers of human epidermis/dermis as well presence/activation of immune resident cell phenotype was verified using immune labelling techniques.

This 3D skin model should allow in future to obtain more translatable indications regarding possible inflammatory events induced by drugs and medical devices.


3D skin model ; immunity ; pseudo-vascularization ; skin sensitization ; cosmetic product safety.


Skin sensitization is a term used to refer to the regulatory hazard known as allergic contact dermatitis (ACD) in humans. This disease is a rather diffuse inflammatory disease originating from cellmediated immune response to skin sensitizers, knowing that up to 20% of the population is affected at different degrees [1]. From a cosmetology/vigilance perspective, it is interesting to note that ACD is one of the most frequently observed adverse reactions related to the use of cosmetics. Skin sensitization is a complex immunological process, originating from the contact of sensitizing products with the first layers of the epidermis and resulting in succession of physicochemical and biological events that lead to the recruitment and activation of allergenspecific T lymphocytes. Different steps have been described as more specific in the development of this disease. Those steps have therefore been the target of the current tests

given that information on skin sensitization is required in the safety assessment of EU Cosmetics Regulation [2]. Following the revision of Annex VII of the Registration, Evaluation and Authorization of Chemicals (REACH) regulation [3], animal models, such as the Guinea Pig based assays (GPMT or the Buchler test) [4] are no longer allowed to meet the information requirements for substances exclusively intended for use in cosmetic products. It is nevertheless important to note that an in vivo sensitization study, preferably Local Lymph Node Assay, can only be performed if the in chemico/in vitro methods are not applicable

New Approach Methods (NAMs), such as in chemico and in vitro methods, have been validated and incorporated since 2021 into the official test guidelines by the Organisation for Economic Cooperation and Development (OECD/ OECD TG 497: guidelines for chemical test), serving as viable replacements of animal modes [5]. These models focus on individual key events (KEin the adverse outcome pathway (AOP) for skin sensitization for which the molecular initiating event (MIE) is covalent binding to proteins [6]. In short, the chemical sensitizer penetrates the stratum corneum, the uppermost layer of the skin, and subsequently binds covalently to proteins (Key Event 1: KEI) to form haptenprotein conjugates, which can be immunogenic

Simultaneously, keratinocytes are stimulated both to release proinflammatory signals (e.g., pro- inflammatory cytokines or ATP) and activate antioxidative response genes (Key Event 2: KE2)Dendritic cells (DCs), which are antigenpresenting cells that playing a major role in the primary immune response and can be activated by allergenic molecules, then develop a mature phenotype involving the induction of various costimulatory molecules and production of pro- inflammatory cytokines and chemokines with formation of haptenprotein conjugates on major histocompatibility complex molecules (Key Event 3: KE3). The activated DCs are mobilized and then migrate from the skin to the draining lymph nodes to present the allergen to T cells (Key Event 4). Currently, three technical Test Guidelines (OECD TG 442 C, D and E) [7, 89] describe a total of seven such methods. Those include: 1) the KEI based Direct Peptide Reactivity Assay (DPRA) and the Amino acid Derivative Reactivity Assay (ADRA); 2) the KE2 based assays KeratinoSens and LuSens; 3) the KE3 based assays hCLAT, USENS, and the IL8 Luc assay (see Tab. 1). As previously reported [10], none of the NAMs methods, except the KDPRA, is considered a sufficient standalone replacements of animal data to draw conclusions about the skin sensitization potential of chemicals or to provide information for potency subcategorization (subcategories 1A and 1B). Each of the tests listed above have some known technical limitations, and data generated can incorrect safety assessment (e.g.. falsenegative). In the Table 1 are reported the test included in the OECD 497 [5]. Novel state- oftheart scientific methods currently in the OECD Test Guideline Program, such as the SENS- IS, which is based on the toxicogenomic analysis on 3D reconstituted epidermidis model (Episkin®) to measure skin sensitization potency, are under evaluation by EURLECVAM

This test seems to address the three first key events of the Adverse Outcome Pathways: the KEL through the expression of genes under the control of «sensor» proteins, the KE2 through the expression of genes indicating keratinocytes activation and the KE3 through the expression of genes implicated in dendritic cells activation and maturation

Despite the scientific literature has demonstrated the possibility of obtaining reconstructed skin models (skin model equivalents, SME), those models have not yet specifically aimed at recreating in vitro the complex cellular interplay between epidermis/dermis cells and the resident immune cells with the goal of providing a fully biomimetic skin equivalent suitable to assess sensitization potential of a given compound. The aim of our study was to tailor a new skin equivalent model that includes more key actors of the sensitization cascade.

This new model included key immune resident cellular components as well as a functional endothelial barrier involved in the recruitment of immune cells; the aim is to investigate the possibility to have a model able to better predict the sensitization potency of single or multiple chemical compounds.

Table 1 List of Non-animal defined approaches (DAs) included in OECD 497 :

tableau recherche et innovation

DPRADirect Peptide Reactivity Assay (DPRA); ADRA Amino acid Derivative Reactivity Assay; KDPRA- kinetic Direct Peptide Reactivity Assay

Materials and Methods 

The model integrated Normal Human Epidermal Keratinocytes (NHEK), Normal Human Dermal Fibroblasts (NHDF), Human Umbilical Vein Endothelial cells (HUVECS), and THP- 1 (a human leukemia monocytic cell line). Both keratinocytes and fibroblasts were obtained from Promocell, (PromoCell GmbH, Heidelberg) and the cells expanded until passage according to the provider’s guidelines. Adult pooled keratinocytes (forehead skin of women aged 34-64 years) were selected to reduce the bias introduced by a single donor. HUVECs were obtained from Lonza Bioscience (Lonza, Basel) while THP1 (TIB202) were obtained from ATCC (LGS, Molsheim) and expanded according to the ATCC guidelines. THP1derived macrophages (MO) were obtained by treating THP1 cells with 150 nM phorbol12myristate- 13acetate (PMA) for 24 hours. The skin model was constructed (Figure 1) using an Sum porous Brandt insert 2in1 for 12 MW (BrandTech Scientific, Essex). The dermal compartment consisted of an extracellular matrix, obtained by mixing rat tail collagen type I (CorningTewksbury), fibrinogen (Sigma Aldrich, Saint Louis) and 1 UImg of fibrinogen of thrombin

(Sigma Aldrich, Saint Louis) together with fibroblasts and immune cells. 300 ul of hydrogel consisting of 9K THP1 or M0 cells and 45K fibroblasts prepared in a final concentration of 4mg ml collagen and 3mg.ml1 fibrinogen (Table 2) was added to each transwell and allowed polymerize for 20 minutes at 37°C in a 5% CO2 humidified incubator. HUVECs were seeded to on the bottom of the transwells at a density of and incubated for 1 hour in a 5CO2 humidified incubator to permit the cells to adhere to the bottom.

Table 1 Cells concentration used to reconstruct the skin model equivalent

tableau recherche et innovation

The transwells were then transferred to a 6 multi well (MW) with 3ml of EBM (Lonza, Basel)300K Keratinoctyes suspended in 400 ul of Keratinocyte Medium 2 (Promocells, Heidelbergwere finally added on top of the dermal layer of each transwell (Figure 1). The tranwells were then transferred to the incubator. Two days post seeding, the keratinocytes were exposed to air and the medium inside the MW was lowered to 1.5 ml to avoid potential medium infiltration in the transwell due to hydrostatic pressure

organoïdes sur puce recherche et innovation

Figure 1. Graphical representation of the protocol steps used to obtain the reconstructed skin model equivalent.

The full model was then cultured in the incubator and then perfused with the Microphysiological System (MPS) to stimulate endothelial cell growth and differentiation with an adequate amount of shear stress (Figure 2). The MPS allows the control of gas concentration in the medium; we used 20% O2, 4% CO2 and 76% N2 with the MPS, to compare results with static incubator. MPS was used in a closed loop configuration, the medium was enriched for 10 minutes at 10sccm, and a flow of 150ul.min was obtained with a peristaltic pump. For the whole duration of the perfusion, the medium flow rate was kept constant at 150 ul.min« . To keep the level of O2 and CO2 stable in the medium a continuous flow of 0.15 sccm was injected into the reservoir, and the temperature kept stable at 37 C by using a MPS heating plate (Figure 2)

organoïdes sur puce recherche et innovation

Figure 2. Schematic representation of the MPS configuration used to perfuse the model and condition the cell medium. Medium were recirculated using a peristaltic pump. The MPS was providing constant microvolumes of gas to keep the partial pressure of gasses constant among the duration of the experiment; at the same time the control unit guarantee a constant temperature inside the standard 6MW. The transwell is placed in a patented polycarbonate adaptor to seal the cells from environment and provide a laminar flow to stimulate the endothelial cells

Viability of cells was assessed at the end of the maturation stage using a live dead assay (MerckBurlington). To assess the expression of skin tissuespecific proteins (ie., involucrin) and the presence and localization of the different cell populations sample were fixed, and paraffin embedded or directly subjected to standard immune labelling and ICC and observed with Leica Thunder microscope (Leica Microsystem, Wetzlar)

CD37 Immunofluorescence labelling was performed on slides and whole model. The slides were deparaffinised using a standard protocol. For whole mount labelling, the transwells were washed twice with phosphate buffer saline (PBS) and then fixed with cold 4paraformaldehyde (PFA) in PBS at 4°C, overnight. Following 2 washes with PBS, both the slides/transwells were blocked with 5% bovin serum albumin (BSA) in PBS for 1-2 hrs at room temperature. The primary antibodies to involucrin (Abcam #ab68, 1:200)(Thermofischer #MA5_18135, 1:100), CD54 (Abcam #ab214860, 1:100) and CD206 (Thermofischer #MA5_32498, 1:200) were diluted in 1% BSA/PBS containing 0.1% tween 20and incubated with the slides/transwells overnight at 4°C. After 3 washes with PBS, the slidestranswells were incubated with the secondary antibodies to involucrin (Goat AntiMouse IgG H&L Alexa Fluor® 647, Abcam #ab150115, 1:500) and CD206 (Goat antiRabbit IgG (H+LHighly CrossAdsorbed Secondary Antibody, Alexa FluorTM Plus 488, Invitrogen# A327311:500) diluted in 1% BSA/PBS containing 0.1% tween 20, for 2 hours at room temperatureFollowing 3 washes, the slides/transwells were labelled with Actin (ActinRed 555 Ready Probes reagent, Thermofisher # R37112) and DAPI (NucBlue Fixed Cell stain ready probesreagent, Thermofischer # R37606) according to the manufacturer’s instructions, for 30 minutesat room temperature. The slides/ transwells were then mounted in PBS before imaging using fluorescence microscope


The obtained reconstructed human skin model showed good viability of THP1 whereas fibroblasts easily tent to proliferate too much when the endothelial cells monolayer is not present (Figure 3). Images showed no mortality of cells at the end of the experiment.

organoïdes sur puce recherche et innovation

Figure 3. IF Live/dead micrograph of the dermis layers cultured in the incubator (A) and under MPS normoxic perfusion (B). In both the modelsendothelial cells were not seeded on the bottom of the membrane; an excess of fibroblasts was observed. Green = calcein AM; orange = propidium iodine; orange arrows = THP1; blue arrow = fibroblasts; red arrow = HUVECsScale bar = 50 um

When endothelial cells were seeded on the bottom of the membrane, the presence of continuous and cohesive monolayer of endothelial cells can be observed. Images obtained on the whole model using ActinRed as well as CD31 immunolabelling on slices showed the presence of a well tight and aligned layer of cells on the bottom side of the transwell (Figure 4). A specific CD31 signal in some putative fibroblasts inside the dermis layer as well as the presence of the peculiar large monocyte nuclei were also observed (Figure 4

organoïdes sur puce recherche et innovation

Figure 4. Micrograph showing the presence of a continuous and aligned monolayer of endothelial cells on the bottom side of the transwell. (A) = lateral view obtained from a thin slice labelled with antiCD31 (green) and NucBlue (Nuclei, cyan). (B) = lateral view obtained from a thin slice labelled with antiCD31 (green) and NucBlue (Nuclei, cyan) and phase contrast (greay). (C) bottomup view of aligned endothelial cells cytoskeleton revealed by ActinRed ICC after 5 days of perfusions using the MPS; pores of membrane are visible in the background. Bar = 50 microns

Images of whole mount sections revealed the distribution and localization of cells included in the models as well as presence of involucrine, a standard marker of keratinocytes differentiation (Figures 5, 6)

microscope science recherche innovation icare

Figure 5. Cross section IF micrograph of a of a whole mount skin model. The discshaped dermis/epidermis layers were removed from the transwell, cut with a razor blade and mounted on a support to arrange the tissue orthogonal to the microscope objective. Red channel ActinRed; cyan channel = NucBlue; green channel = CD206; yellow channel involucrinecyan circles = putative THP1; green stars = putative macrophages; red stars = fibroblastsyellow star = keratinocytes. Scale bar = 100 um

Our data suggest a not homogeneous distribution of THP1 and macrophages (Figure 5, 6). The presence of THP1 was found more localized between the endothelial barrier and the first half of the dermis whereas the macrophages seemed more abundant in the dermis layer closer to the keratinocytes.

microscope science recherche innovation icare

Figure 6. Bottom up IF optical sections obtained in the middle of the dermis (A) and close to the epidermis (B) of a whole mount skin model. Red channel = ActinRed; cyan channel NucBlue; green channel CD206; cyan circles putative THP1; green stars = putative macrophages; red stars = fibroblasts


European legislation has paved the way of more ethical test of chemicals envisaging cosmetic applications [11] banning test on animal models. Nevertheless, the chemical industry is constantly developing new and more performing/ecological chemicals. Each new chemicals need to undergo a rigorous risk assessment independently from the potential field of application

To date most of those assessment rely on the use of animal models that have the drawback of raising more and more ethical concerns as well as potential translatability of the results. To overcome those drawback new alternative methods are/have been developed. Those models to be accepted by the legislations must be reliable, reproducible, and relevant. With respect to skin and skin reactions to chemical compounds, SME based on human cell lines are currently the most promising technology to develop alternative guidelines aimed at replacing the animal models. The increasing number of literatures on SME has demonstrated the superior capacity of those models to mimic native skin’s cellular organization and function when compared with traditional 2D cells culture [12]. Different SME are currently commercialized and used as alternative methods to assess specific hazards of new compounds (e.g. EpiSkin®, SkinEthic® and TSkin® (L’Oreal, France), EpiDerm® (MatTek Corporation, USA), epiCS® (CellSystems, Germany), Phenion (Henkel, Germany), NeoDerm® (Tego science, Korea)LabCyte® (JTEC, Japan)). These models have the advantage of being reproducible and relatively simple as most consist of mono or bilayered constructs with few cell types

To further enlarge the availability of alternative animal model we decided to develop a model dedicated to skin sensitization assessment give that, to the best of our knowledge, none of the existing models have the needed cellular complexity to recapitulate the main sensitization- specific events. The presence of immune resident cells and of a functional vasculature are expected to increase the increase the relevancy of a new model dedicated to skin sensitization assessments

SME all relies on some support (typically a transwell) to expose the epidermis to air and allow the maturation/differentiation of keratinocytes into corneocytes and formation of the stratus corneum. Whereas most of the models are cultured in static conditions inside incubators there have been more and more effort to integrate SME with microfluidic technologies. These combinations of technologies are referred as skinonachip models. Their main advantages are to enable physiologically relevant transport of nutrients and exogenous substances to the skin tissue, better control over physical and chemical factors in the cell microenvironment [13], and more reliable evaluation of drug candidates in terms of toxicity, efficacy, and delivery compared to static conditions. A vascularized SME could represent an invaluable tool for basic research (e.g, study vascular remodeling during wound healing, aging,) but has the potential to become a new tool to assess topical and/or systemic skin reactions to specific chemical compounds and eventually test therapies to mitigate or treat such adverse reactions.

Our data are the first encouraging step toward this direction. Our data demonstrated the capability to obtain a partially immunocompetent models that recapitulate the main structure and functional elements of human skin. The presence of a micro/milli fluidic circuit allowed us to obtain continuous and aligned endothelial cells interface. The model is highly customizable in terms of cells population as well as vasculature to tissue ratio: this is expected to bring benefits when traying to recreate in vivo not only healthy and young skin models but also models with a pre- inflammatory condition or skin fragility.


Other groups created several SME models for the prediction of skin sensitization potential, but to our knowledge, none of them had integrated dermis, epidermis, THP1 and HUVECS. For the first time, this study demonstrates the feasibility of such approach. The next steps will be to demonstrate the capability of our skin model to recapitulate a sensitization episode. This will be assessed by evaluating the shift in the macrophage/THP1 population using: 1- reference chemicals representative of raw materials in cosmetics but also in medical devices; 2chemicals representative of a balanced range of skin sensitizer potency (weak, moderate, and strong).


This project was support by « PIA3 régionalisé » and AuvergneRhôneAlpes Region

Conflict of interest statement 

Any conflict of interest 


  1. SedóMejía G, SotoRodriguez A, PinoGarcía C, SanabriaCastro A, MongeOrtega OP (2020) Contact dermatitis: Clinical practice findings from a single tertiary referral hospitala 4Year retrospective study. World Allergy Organ J 13: 100440
  1. European Parliament (2009) Regulation (EC) no 1223/2009 of the European Parliament and of the council of 30 November 2009 on cosmetic products. OffJ Eur Union. 2009; L396:1- 1355 Available from: 13.55.
  2. Commission Regulation (EU) (2017) 2017/706 of 19 April 2017 amending Annex VII to Regulation (EC) No 1907/2006 of the European Parliament and of the Council on the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) as regards skin sensitisation and repealing Commission Regulation (EU) 2016/1688 (Text with EEA relevance.) C/2017/2369. Off. J. Eur. Union, 8-11
  1. OECD 2022 Guidelines for the testing of chemicals. Section 4. Test N°406: Skin Sensitization, health effects, OECD: Paris, France
  1. OECD 2021 Defined Approaches on Skin Sensitization. Test N° 497. In OECD Guidelines for the Testing of Chemicals, Section 4; OECD Publishing: Paris, France
  1. Silva JR, Tamburic S (2022) A stateoftheart review on the alternatives to animal testing for the safety assessment of cosmetics. Cosmetics, 9: 20
  1. OECD 2021 In Chemico Skin Sensitization: Assays addressing the Adverse Outcome Pathway key event on covalent binding to proteins. Test N°442C. In OECD Guidelines for the Testing of Chemicals, Section 4; OECD Publishing: Paris, France
  1. OECD 2018 In Vitro Skin Sensitization: ARENrf2 Luciferase Test Method Test N° 442D. In OECD Guidelines for the Testing of Chemicals, Section 4; OECD Publishing: ParisFrance
  1. OECD 2018 In Vitro Skin Sensitization assays addressing the Key Event on activation of dendritic cells on the Adverse Outcome Pathway for Skin Sensitisation. Test No 442E: In OECD Guidelines for the Testing of Chemicals, Section 4; OECD Publishing: Paris, France
  1. Risueno I, Valencia L, Jorcano JL, Velasco D 2021 Skinona- chip models: general overview and future perspectives. APL Bioengineering 5: 03090
  1. Almeida A, Sarmento B, Rodrigues F (2017) Insights on in vitro models for safety and toxicity assessment of cosmetic ingredients. Int J Pharm 519:178
  1. Sriram G. Alberti M, Dancik Y, Wu B, Wu R, Feng Z, Ramasamy S. Bigliardi PLBigliardiQi M, Wang Z (2017) Fullthickness human skinonchip with enhanced epidermal morphogenesis and barrier function. Materials Today 21, 4: 326-339
  1. Risueno I, Valencia L, Jorcano JL, Velasco D (2021) Skinona- chip models: general overview and future perspectives. APL Bioengineering.5:03090.
Retour en haut