I. Safety and Efficacy Testing of Chemicals, Pharmaceuticals and Biologicals

Computational System for Predicting Chemical Reactivity Towards Macromolecules and Subsequent Adverse Effects
G. Dimitrova ⁽¹⁾, C. Kuseva ⁽¹⁾, D. Duché ⁽²⁾, J. Eilstein ⁽²⁾, S. Ringeissen ⁽²⁾, and O. Mekenyan ⁽¹⁾⁽¹⁾ Laboratory of Mathematical Chemistry, "Prof. As. Zlatarov" University, Bourgas, Bulgaria
⁽²⁾ L’Oréal Research & Innovation, France

The ultimate goal of this work is the development of a computational tool for predicting chemical reactivity towards macromolecules (proteins, DNA) and subsequent adverse effects such as skin sensitization, genotoxicity, etc. Firstly, we focused on contact dermatitis which results from the interaction between a hapten (electrophile) and the side chain of nucleophilic aminoacids of proteins. Therefore the reactivity of chemicals towards peptides was studied. The domains of reactivity categories are defined according to the types of interaction mechanisms such as Michael addition, epoxide ring opening, Schiff-base formation, acylation etc. The chemicals acting by these mechanisms have specific structural functionalities with different hardness/softness which can be assessed by quantum-chemical parameters and used to predict chemical reactivity towards macromolecules. Experimental data of reactivity of chemicals on synthetic cysteine/lysine peptides (Direct Peptide Reactivity Assay) and/or Glutathione are used to calibrate the boundaries of defined reactivity categories as well as for defining new categories for protein binding potency. The ability of chemicals to interact by ionic or radical mechanisms were assessed by quantum-chemical parameters evaluating the stability of formed intermediates and hence, the energetically feasibility of the respective transformations. The classification of parent chemicals according to the reactivity categories will be used to provide a hazard assessment for subsequent adverse effects.

Towards the 21st Century, Advances and Refinements in the Prediction of Sensitization Potential Using the TIMES platform
G. Patlewicz ⁽¹⁾, G. Dimitrova ⁽²⁾, C. Kuseva ⁽²⁾, G. Ellis ⁽³⁾, R. Hunziker ⁽⁴⁾, P. Kern ⁽⁵⁾, L. Low ⁽⁶⁾, S. Ringeissen ⁽⁷⁾, G. Veith ⁽⁸⁾ and O. Mekenyan ⁽²⁾⁽¹⁾ DuPont Haskell Global Centers for Health & Environmental Sciences, Newark DE, USA
⁽²⁾ Laboratory of Mathematical Chemistry, Prof “As Zlatarov”, Bourgas, Bulgaria
⁽³⁾ Givaudan Swiss SA, Vernier, Switzerland
⁽⁴⁾ Dow Europe GmbH, Toxicology and Environmental Research and Consulting, Horgen, Switzerland
⁽⁵⁾ Procter & Gamble Eurocor, Strombeek-Bever, Belgium
⁽⁶⁾ ExxonMobil, Biomedical Sciences, Inc., Division of Toxicology and Environmental Sciences, Annandale NJ, USA
⁽⁷⁾ L’Oreal Research & Innovation, France
⁽⁸⁾ International QSAR Foundation (IQF), Two Harbors, MN, USA

The TImes MEtabolism Simulator platform for predicting Skin Sensitization (TIMES-SS) is a hybrid expert system that was developed at Bourgas University using funding and data from a consortium comprising experts from industry and regulatory agencies and coordinated by IQF. The model was developed with the aim of minimising animal testing, to be scientifically valid for regulatory purposes, and to be mechanistically transparent. TIMES-SS encodes 2D structure-toxicity and structure-skin metabolism relationships through a number of transformations, some of which are underpinned by mechanistic 3D QSARs. An external evaluation exercise was completed in 2007 where LLNA data were generated for 40 new chemicals and compared with predictions made by TIMES-SS. The results were promising with an initial concordance of 75%. An extensive evaluation followed to assess the results in light of reaction chemistry principles. The number of chemicals underpinning a given reaction chemistry alert was reviewed and four validation substances were subsequently tested. Recently, a 3 year research proposal was initiated as part of a new industry consortium. The skin (a)biotic metabolism simulators and the skin sensitisation model will be refined in light of new data and chemical insights. The applicability domain of the underlying experimental data will be evaluated to reflect current recognised inconsistencies between different in vivo assays. The feasibility of developing a respiratory sensitisation model will also be investigated. This presentation will provide an overview of the current status of TIMES-SS, highlight current refinement activities, and outline the strategy for deriving a respiratory sensitisation model based on preliminary investigations.

Predicting Sensitizing Potential of Cosmetic Ingredients: Enlargement of the Applicability Domain of the MUSST Assay by Using Complementary U937-based Assays
S. Teissier, C. Piroird, D. Verda, A. Staropoli, J.M. Ovigne, J.R. Meunier L’Oréal Research & Innovation, France

Allergic contact dermatitis resulting from industrial, environmental or domestic exposure to sensitizers is the most common manifestation of immunotoxicity in humans. Skin sensitization risk assessment and more precisely the sensitizing potential of ingredients used in the cosmetic and pharmaceutical industries so far essentially relies on available animal test methods, such as the mouse local lymph node assay. In the context of the 7th amendment to the Cosmetic Directive as well as the recent EU-legislation on chemicals (REACH), the cosmetic industry is particularly concerned by the challenge of finding in vitro alternatives to assess the sensitizing potential of chemicals.

Contact sensitizers induce several phenotypic and functional changes on dendritic cells (DC) in vivo and in vitro. One of these changes, the induction of CD86, is the most frequently analyzed endpoint for the in vitro prediction of contact sensitizers using different cellular models based on DC or human myeloid cell lines. We developed the Myeloid U937 Skin Sensitization Test (MUSST) based on the induction of CD86 on U937 cells. Year’s in-house experience with this assay led us to identify its limits, and to develop further methods and further models (including 3D-models) to overcome these limits. We will describe here how an adequate use of CD86-mRNA test, U937-apoptosis assay and the Episkin-U937 co-culture assay can complement the MUSST assay to enlarge its applicability domain and thus to cover a larger physicochemical diversity encountered in cosmetic ingredients.

Contact Sensitizers Modulate the Arachidonic Acid Metabolism of PMA-differentiated U-937 Monocytic Cells Activated by LPS
A. Del Bufalo ⁽¹⁾, J. Bernad ⁽²⁾, D. Verda ⁽¹⁾, J.R. Meunier ⁽¹⁾, F. Rousset ⁽¹⁾, S. Teissier ⁽¹⁾ and B. Pipy ⁽²⁾⁽¹⁾ L’Oréal Research & Innovation, France
⁽²⁾ UMD3, INSERM-IFR150, Université de Toulouse, Université Paul Sabatier, France

Contact sensitizers are defined as reactive molecules which have the ability to modify skin proteins to form an antigen. In addition to the haptenation mechanism, inflammatory signals, leading to the activation of dendritic cells, are described to be crucial for the effective induction of an antigen-specific T cell immune response. However, the sensitization phase is often a silent process, without obvious clinical manifestations of inflammation/irritation. Even more, anti-inflammatory properties of some molecules don’t prevent them to induce skin sensitization. The aim of this study was to better understand how sensitizers modulate an inflammatory response. To address this purpose, we used the human monocytic-like U-937 cell line differentiated by phorbol myristate acetate (PMA) and investigated the effect of 6 contact sensitizers (DNCB, PPD, hydroquinone, propyl gallate, cinnamaldehyde and eugenol) and 3 non sensitizers (lactic acid, glycerol and tween 20) on the production of pro-inflammatory cytokines (IL-1β and TNF-α) and on the arachidonic acid metabolic profile after bacterial lipopolysaccharide (LPS) stimulation. Our results showed that among the tested molecules, all sensitizers specifically prevent the production of PMA/LPS-induced COX-2 metabolites (PGE2, TxB2 and PGD2). We further demonstrated that there is no unique PGE2 inhibition mechanism: while the release of arachidonic acid (AA) from membrane phospholipids does not appear do be a target of modulation, COX-2 expression and/or COX-2 enzymatic activity are the major steps of prostaglandin synthesis that are inhibited by sensitizers. Altogether these results add a new insight into the multiple biochemical effects described for sensitizers.

Evaluation of SensCeeTox® an Integrative Model for Identifying Chemical Sensitizers
C. Gomes ⁽¹⁾, A. Del Bufalo ⁽¹⁾, R. Note ⁽¹⁾, S. Teissier ⁽¹⁾, C. Piroird ⁽¹⁾, J.M. Ovigne ⁽¹⁾, J.M. McKim Jr ⁽²⁾ and J.R. Meunier ⁽¹⁾ L’Oréal Research & Innovation, France
⁽²⁾ Ceetox, Inc., United States

Allergic contact dermatitis is the result of an adaptive immune response of the skin to direct exposure to an allergen. Because it is the most common manifestation of immunotoxicity in humans, a strict screening of all ingredients in consumer products is required. Current methods rely on animal testing (eg. local lymph node assay) for determining chemical sensitization. Ethical concerns and regulatory changes in the EU have stimulated the development of alternative tests for the assessment of potential sensitizers. There is a common view that a strategy for the integration of the currently available methods will be required. Ceetox Inc. has developed an integrative approach, namely SensCeeTox® allowing the EC3 value to be estimated and differentiating the degree of response from non-sensitizer (NS), weak (W), moderate (M), and strong (S), up to extreme (E). The purpose of this study was to evaluate the predictive capacity of this approach in a blinded manner. L’Oréal provided a set of 40 compounds (20 positive and 20 negative), consisting of 24 proprietary and 16 public domain chemicals that were assessed by CeeTox. All 40 compounds could be classified; those placed into E, S, and M were considered positive, while compounds classified as W or N were considered negative. Results obtained for the prediction of positive and negative compounds were promising. However, the model failed to accurately predict each sensitization category. Refinements and automation of the algorithm and the incorporation of additional assays should improve the models’ ability to predict potency.

Evaluation of SENS-IS®, an Episkin® based model for Identifying Chemical Sensitizers
S. Teissier ⁽¹⁾, F. Tourneix ⁽¹⁾, A. Del Bufalo ⁽¹⁾, C. Gomes ⁽¹⁾, H. Groux ⁽²⁾ and J.R. Meunier ⁽¹⁾⁽¹⁾ L’Oréal Research & Innovation, France
⁽²⁾ ImmunoSearch, Grasse, France

In the context of the 2013 ban given by EU Cosmetics Directive, the ability to identify and classify the skin sensitization potential of chemicals without animals is of high importance for the cosmetic industry. A range of different in vitro chemistry-based (DPRA, GSH reactivity) and cell-based methods (MUSST, hCLAT, Keratinosens) have been developed and we are currently evaluating some of them for their applicability to cosmetic ingredients and its physicochemical diversity. Although these assays appear to be promising for hazard identification, potency assessment is still limited. Possible limitations may be linked to the metabolism that may differ between the models and native skin, to bioavailability which is not considered in monolayer cultures, and to the danger signal that may be different in monolayers as compared to a natural tridimensional microenvironment.
Immunosearch developed SENS-IS, a new method, based on the quantitative analysis of specific biomarkers expressed in 3D reconstructed epidermis (Episkin®), thus providing a possible way to encompass these limitations and come closer to potency assessement. With the aim to evaluate the predictive capacity of this approach on a cosmetic ingredient constituted set, L’Oréal provided in a blinded manner a set of 40 proprietary as well as public domain chemicals that were assessed by Immunosearch. We present here the results of this study and will analyze the genomic signatures among chemical and ingredient classes.

The Myeloid U937 Skin Sensitization Test (MUSST) for the Prediction Of Skin Sensitization Potential
J.M. Ovigne, C. Piroird, D. Verda, F. Rousset, S. Teissier and J.R. Meunier L’Oréal Research & Innovation, France

Skin sensitization is a delayed type allergy consisting of a cellular immune reaction to small molecular weight chemicals, so far predicted using animal test methods such as the local lymph node assay (LLNA). In line with the 3Rs, in vitro alternatives are being developed based on early events of the skin sensitization process. One of these is the capacity of dendritic cells to recognize a chemical as a danger. The Myeloid U937 Skin Sensitization Test (MUSST) models this by measuring the up-regulation of CD86 expression on U937 cells. A chemical is classified as sensitizer if it induces a dose-dependent up-regulation of CD86 expression at non toxic doses in two concordant experiments.

The MUSST prediction is exemplified with 2 sensitizers (Phenyl benzoate, Methylchloroisothiazolinone), 2 pre/prohaptens (ethylene diamine, eugenol) and 2 non sensitizers (lactic acid, benzaldehyde) correctly classified by the assay.

The predictive performances of the MUSST are evaluated with a panel of 50 reference chemicals (31 sensitizers and 19 non sensitizers) against the LLNA data to support its submission to and its acceptation by ECVAM for a pre-validation. With the 40 classified chemicals (10 are inconclusive), the MUSST displays a concordance of 83% with 81% sensitivity and 84% specificity.

The performances, further evaluated with an extended set of 83 classified reference chemicals, show concordance, sensitivity and specificity above 75%.

The MUSST is an efficient assay for the skin sensitization hazard characterization, promising as a tool to be integrated within a battery of assays to perform a skin sensitization risk assessment.

Methods Development Through Recognition in 3Rs : L’Oréal Commitment
J.R. Meunier, N. Alépée, J. Cotovio, D. Duché, G. Ouédraogo-Arras, L. Marrot, R. Note, J.M. Ovigne, S. Ringeissen, S. Teissier and I. Marey-Semper L’Oréal Research & Innovation, France

Dissemination of advances on alternative methods represents a step to promote alternatives to animal testing in line with the EU Cosmetics Directive. L’Oréal has, based on these principles, developed test methods to screen and test potential effects on chemicals.

We have focused initially on approaches for skin irritation. A peer review on various aspects of alternative techniques was performed at all stages of the R&D with a focus on in vitro methods improvement of chemicals selection (screening) as well as quality testing. To ensure quality and objectivity, experts from international committees oversee the content of EpiSkin and SkinEthic RHE protocols. Details of the approach will be presented for both skin corrosion and irritation with a set of 50 reference chemicals. Using computational approaches, an automated workflow algorithm was developed to predict a molecule’s potential for skin irritancy based on in vivo Draize data. The practical approaches developed by L’Oréal in the areas of eye irritation (SkinEthic HCE defined with 90 chemicals), skin sensitization (MUSST assay optimized with 50 chemicals), phototoxicity and genotoxicity will be also described. For chronic and systemic toxicity testing, a realistic approach relied on the combination of data generated for multiple endpoints. Preliminary studies indicated that the method had good sensitivity and specificity (91% and 78%) while defining a LD50 threshold at 2000 mg/kg.

Combination of in silico, read across and in vitro strategies assure realistic scientific approaches suitable for the safety assessment process within industry.

Acute Dermal Toxicity Using the OECD TG 404 Integrated Testing Strategy Combining the Use of the EpiSkin Test Methods.
N. Alépée ⁽¹⁾, M.H. Grandidier ⁽¹⁾, N. Seyler ⁽²⁾, F. Soler ⁽²⁾, J. Cotovio ⁽¹⁾ and J.R. Meunier ⁽¹⁾⁽¹⁾ L’Oréal Research & Innovation, France
⁽²⁾ Episkin SNC, France

Identification of corrosive and skin irritant chemicals was based, for regulatory purposes, on their ability to produce irreversible or reversible alterations of the skin at the site of contact. During the skin corrosion validation study of the EpiSkin test method, some in vivo corrosives were identified as non corrosives in vitro. Since under classification of chemicals may be due to non specific reduction of MTT in solution, MTT interference corrections were performed on 5 chemicals detected as direct MTT reducers, indicating the need to adapt the EpiSkin skin corrosion test method by including specific controls for MTT reducers.

A stepwise testing strategy for the prediction of skin irritation and then skin corrosion was developed using the validated EpiSkin test methods to support the ongoing revision of the OECD test guidelines TG404 and TG431. When applying the testing strategy on about 50 reference substances (from the ECVAM validation studies), 20 in vivo irritant chemicals were identified in vitro as non corrosive but correctly classified as irritants. In addition 17 in vivo non irritants and non corrosives were correctly predictive in vitro using both skin corrosion and irritation test methods. Finally 12 corrosive chemicals identified by NICEATM/ICCVAM as incorrectly predicted in vitro were evaluated. The results showed that corrosive chemicals misclassified in the in vitro corrosion test were identified in vitro as irritants.

This analysis of those new data should decrease the need for testing for both dermal skin corrosivity and skin irritation of substances for which sufficient evidence already exists.

In vitro eye Irritation Assessment Using the SkinEthic HCE Test Method Applied to Ingredients Used in Cosmetics
J. Cotovio, N. Alépée, M.H. Grandidier, D. Lelièvre, S. Blond and J.R. Meunier L’Oréal Research & Innovation, France

To comply with the EU Cosmetic Directive, predictive alternative methods are required to evaluate eye damage potential of chemicals. Linked to the complexity of both eye irritation mechanisms and the diversity of chemicals, the in vitro assessment of eye irritation potential is a complicated issue. The corneal epithelium is crucial and represents the first line of defense against injury. 3-D in vitro tissues sustained by adapted technologies allow the testing of substances in conditions similar to in vivo exposure. In this study, we have used the standardized SkinEthic reconstructed human corneal epithelial (HCE) model to evaluate in vitro eye irritancy. A specific protocol was developed aiming to match chemicals properties with adapted exposure steps (1h exposure, 16 hrs post exposure incubation) in particular for cosmetic ingredient families. Analyses were performed according to the "Globally Harmonized System (GHS) classification. The Prediction Model using a 50% viability cut-off allowed the drawing up of 2 categories: Irritants (grouping Cat1 and 2) and No Category. Applied to a broad set of 435 cosmetic substances, the SkinEthic HCE test method showed good and balanced prediction performances (81% sensitivity; 82% specificity). Furthermore, by using appropriate controls, the applicability domain of the method can be extended to the MTT reducers and/or dyes substances by using additional controls. Severe or irreversible irritant chemicals were not specifically differentiated from reversible and mildly irritants. This test method is part of the ongoing ECVAM eye irritation validation. It could be part of a specific tiered test strategy for hazard assessment of test substances in regulatory schemes.

Characterization of Alcohol- and Aldehyde- Deshydrogenase Activities in Normal Human Skin Compared with Reconstructed Human Skin Models
J. Eilstein, G. Léreaux, J.R. Meunier, J. Leclaire and D. Duché L’Oréal Research & Innovation, France

Skin is considered as the body envelope and a physical barrier to its environment. However, it contains numerous metabolizing enzymes which give to it a potential role in terms of metabolism and detoxification. The 7th European amendment to the cosmetic directive bans the use of animal testing to evaluate the efficacy and safety of new cosmetic ingredients. This policy has forced the cosmetic industry to develop reconstructed human skin models (skin models) as tools for alternative methods to the animal experimentation. For these reasons, the models require to be characterized and compared with a normal human skin (NHS) in terms of metabolic capabilities. In this work, we characterized alcohol deshydrogenase (ADH) and aldehyde deshydrogenase (ALDH) activities. Previous studies showed that NHS and reconstructed human epidermis such as EpiskinTM, SkinEthic-RHETM and the full thickness model of EpiskinTM expressed several ADH and ALDH isoforms. Their global catalytic activities were quantified in NHS and skin models using cinnamyl alcohol and cinnamic aldehyde as substrates, respectively. Apparent Vmax, Km and ratio Vmax/Km (estimating metabolic clearances) were calculated for each tissue from metabolites measurement of dose effect studies. Results showed that in NHS and in skin models, ADH and ALDH enzymes are functional and that ALDH activity is more important than ADH activity (ratio Vmax/Km ALDH > ratio Vmax/Km ADH). To conclude, the skin models can be easy used to study the detoxication process of primary alcohols or aldehydes, considered as potential sentitizers, and define their levels of cytotoxicity at the skin level.

Comparison of Xenobiotic Metabolizing Enzyme Activities in Normal Human Skin and Reconstructed Human Skin Models from SkinEthic Laboratories
J. Eilstein, G. Léreaux, J.R. Meunier, J. Leclaire and D. Duché L’Oréal Research & Innovation, France

Skin represents the major protective barrier of the body to its environment. Also, skin is an organ involved in the metabolism of xenobiotics and its ability to metabolize xenobiotics can become consequent when considering its total surface area (2 square meters). Consequently, research on skin metabolism would need a real scientific effort to characterize skin metabolizing enzymes and their activities. In addition, the 7th European amendment to the cosmetic directive forbids the use of animal testing to assess the efficacy and safety of new cosmetic ingredients. This policy has forced the cosmetic industry to develop in vitro tools such as reconstructed human skin models (skin model) as alternative methods to animal experiments. For these reasons, these skin models require to be characterized and compared with normal human skin (NHS) samples in terms of metabolic capabilities. This work presents the mRNA expression of several enzymes (CYP450, Esterase, ADH, ALDH, NAT, GST, UGT, SULT…) and their apparent catalytic parameters (apparent Km, Vmax and the ratio Vmax/Km) in skin models compared with NHS. Results showed that all these enzymes involved in the metabolism of xenobiotics are expressed and effective in the NHS and skin models. Also, apparent ratio Vmax/Km (estimating the metabolic clearance) and then the metabolic abilities were often comparable between skin models and NHS. These results indicate that the skin models can substitute to NHS to select cosmetic ingredients on the basis of their metabolism, efficacy or/and safety.

Novel in vitro genotoxicity Assays Using Reconstructed Human Tissues
G. Ouédraogo ⁽¹⁾, F. Nesslany ⁽²⁾, S. Simar ⁽²⁾, S. Talahari ⁽²⁾, D. Lagache ⁽²⁾, E. Vercauteren ⁽²⁾, M.C. Polyn ⁽²⁾, D. Marzin ⁽²⁾, N. Flamand ⁽¹⁾ and J.R. Meunier ⁽¹⁾⁽¹⁾ L’Oréal Research & Innovation, France
⁽²⁾ Institut Pasteur de Lille, Laboratoire de Toxicologie, Lille, France

The poor specificity (tendency to yield a high number of positive results whereas in vivo genotoxicty results are negative) of in vitro genotoxicity assays (especially the mammalian cell based ones) is of concern in the context of the European legislation (7th amendment to the European Cosmetic directive, REACH).

An in vitro micronucleus assay using human reconstructed skin tissues and target cells grown beneath the skin was developed. The purpose was to improve the relevance of exposure conditions in in vitro genotoxicity assays for dermally applied compounds. Previous results have shown that this method was reproducible and could be transferred to other laboratories. In addition, clastogens as well as aneugens could be detected.

The system has now evolved to combine both the comet assay and the micronucleus assay. A set of 12 chemicals have been tested with this system. The results obtained will be shown. So far, they look promising. Most of the “irrelevant positives” yielded negative in vitro results using this system.

II. Replacement and Reduction in Basic Research

Adaptation of the Validated SkinEthic RHE Skin Corrosion Test Method to 0.5 cm2 Tissue Sample
C. Tornier ⁽¹⁾, N. Alépée ⁽²⁾, J. Cotovio ⁽²⁾, S. Martins ⁽¹⁾ and J.R. Meunier ⁽²⁾⁽¹⁾ SkinEthic Laboratories, France
⁽²⁾ L’Oréal Research & Innovation, France

In vitro human reconstructed epidermal models have been used to develop protocols able to discriminate corrosives from non corrosives. The SkinEthic RHE test method, using 0.63 cm2 inserts was validated in 2006. Due to manufacturing constraints, the model is now produced on 0.5 cm2 instead of 0.63 cm2.

The study demonstrates that the RHE skin corrosion protocol could be adapted from 0.63 cm2 to 0.5 cm2 RHE samples. For such purpose, the protocol size adaptation was performed using 25 chemicals including the 12 OECD TG431 reference chemicals. To test the robustness and relevance of the test method, particular attention was given to choose chemicals correctly classified (non corrosive, corrosive) but also misclassified chemicals (false positives / negatives). Results obtained with the 0.5 cm2 skin corrosion test method showed that all corrosives were correctly classified, and 11 out of 13 chemicals were identified as non corrosives. The overall accuracy over the 25 chemicals was 92%. The specificity and the sensitivity of the OECD chemicals were 100%.

In addition, the quality of RHE tissues was not only maintained but also improved. The quality control, performed on 136 (0.63 cm2) and 262 (0.5 cm2) RHE batches, showed a mean of 0.99 and 1.17, respectively. This similarity over years demonstrates the high quality production of the tissues using both viability and morphology parameters.

In conclusion, the quality of 0.5 cm2 SkinEthic RHE tissues was thoroughly maintained over 9 years and the performance of the skin corrosion test method fully meet OECD and ECVAM requirements.

Predicting Dermal Toxicity Using the OECD TG 404 Integrated Testing Strategy: An Evaluation of the SkinEthic RHE Test Methods
N. Alépée ⁽¹⁾, C. Tornier ⁽²⁾, S. Martins ⁽²⁾, J. Cotovio ⁽¹⁾ and J.R. Meunier ⁽¹⁾⁽¹⁾ L’Oréal Research & Innovation, France
⁽²⁾ SkinEthic Laboratories, France

The use of testing strategies based on alternatives methods was adopted by the OECD test guidelines TG404 for predicting dermal toxicity. To date the test methods using the SkinEthic reconstructed human epidermis (RHE) were independently validated for dermal skin corrosion and skin irritation.

The aim of the present study was to develop a testing strategy combining both SkinEthic RHE skin corrosion and skin irritation test methods to support the ongoing revision of the OECD TG404 and TG431.

For such purpose, about 50 chemicals (from the ECVAM validation studies) were evaluated in both skin corrosion and skin irritation. The results showed that 20 in vivo irritant chemicals were identified in vitro as non corrosives but correctly classified as irritants using the SkinEthic RHE skin irritation test method. In addition 15 in vivo non irritants and non corrosives were correctly predictive in vitro using both skin corrosion and irritation test methods. Finally 12 corrosive chemicals identified by NICEATM/ICCVAM as incorrectly predicted in vitro were evaluated. The results showed that corrosive chemicals misclassified in the in vitro corrosion test were identified in vitro as irritants. So, when applying the OECD TG404 testing strategy on the substances identified by NICEATM/ICCVAM as potentially false negative corrosives, all these substances were correctly identified.

In conclusion, when the determination of corrosivity or irritation cannot be made using a weight-of-the-evidence analysis, a preferred sequential testing strategy (skin irritation / corrosion), which includes the performance of accepted in vitro SkinEthic RHE tests should be considered.

A Pertinent Screening Tool to Measure Permeability Coefficient : Episkin® Reconstructed Human Skin Model
A. Garrigues-Mazert, S. Grégoire, N. Zeman and J.R. Meunier L’Oréal Research & Innovation, France

According to their similarities to native human tissue in terms of morphology, lipid composition and biochemical markers, reconstructed human epidermis (RhE) have been identified as useful tools for the in vitro testing of phototoxicity, corrosivity and irritancy. These last years, some papers claim that RhE are appropriate alternatives to human skin for the assessment of skin permeation and penetration in vitro.

Both RHE and Episkin®, commercially available wtith SkinEthic, are particularly adapted for testing. Indeed, their design allow to measure penetration directly in the insert without mounting the tissue in a diffusion cell. These results lead to the development of reliable protocols for the upstream ranking assessment of the skin penetration of cosmetic ingredients under their conditions of cosmetic use.

Permeability coefficient (Kp) measurement requires sampling as a function of time. It could be done using flow through diffusion cell (as PermeGear cell) or using directly insert with totally or partially replacement of a given volume of receptor fluid at given time gap. Both approaches have been tested with caffeine as reference compound. With sink condition and infinite dose, flux as a function of time should reach a constant value corresponding to the steady state. Results show that steady state is not reached with PermeGear cell contrary to Insert.

Comparison with human skin data reinforce previous studies conclusion on RhE model as relevant alternative to human skin for in vitro penetration study.

Alternative Approach to Maximum Flux for TTC Applied to Safety Evaluationof Cosmetic Ingredients
S. Grégoire, D. Duché, A. Garrigues-Mazert and J.R. Meunier L’ORÉAL Research & Innovation, France

REACH legislation would increase drastically the toxicology evaluation. To meet this challenge, relevant and accurate prediction of dermal uptake/exposure of topically applied chemicals is essential for risk assessment. It could be obtained without recourse to an experimental measurement and avoids any problems with ethical issues, recruiting volunteers or housing animals. Typically, QSAR model predicting permeability coefficients (i.e. kp) are used. Many models were developed; all of them lead to the same conclusion: small lipophilic chemicals have greatest skin permeability. This analysis often rises to confusion. Dataset used to build up this relation concerns percutaneous transport from aqueous solution. Whereas, kp increases with log P, aqueous solubility decreases with lipophily. Resulting flux, and effective absorbed amount of chemical, is then balanced between two competitive factors (permeability and solubility).

Concept of maximum flux means that a chemical cannot cross the skin higher than flux measured at steady state with a chemical applied on the surface in saturated solution (or in neat chemical form). It allows assessing the maximum absorbed dose. This concept was recently used in TTC approach for cosmetic ingredient. A classification of potential of cutaneous chemical absorption was proposed on the basis of their physico-chemical properties. Unfortunately, the proposed classification does not cover all range of molecular weight and log P. Moreover, some physico-chemical properties known to affect cutaneous absorption (i.e. ionisation,; volatility) are not considered. At least, the default proposed values greatly overestimate the absorption experimentally obtained.

An alternative approach was developed by L’Oréal to overcome these limitations.

Development of an Integrative Approach For The Prediction of Systemic Toxicity: Combination of Cell Toxicity, Pharmacological and Physical Chemical Properties
R. Note ⁽¹⁾, H. Nocairi ⁽¹⁾, M. Thomas ⁽¹⁾, L. Bourouf ⁽¹⁾, J.M. McKim Jr. ⁽²⁾, G. Ouédraogo ⁽¹⁾ and J.R. Meunier ⁽¹⁾⁽¹⁾ L’Oréal Research & Innovation, France
⁽²⁾ Ceetox, United States

Ethical, scientific and economical constraints have motivated the scientific community to develop alternatives to animal testing. Developing alternatives for acute/chronic systemic toxicity testing represents a challenge because of the complex biological processes implied. A realistic approach could rely on the combination of data generated for multiple endpoints. The Ctox panel® which is a multiparameter cell-based in vitro system for predicting rat acute systemic toxicity is a typical example. Preliminary studies conducted in a blinded manner showed a good sensitivity and specificity (91% and 78%) while defining a LD50 threshold at 2000 mg/kg. However, the model failed to accurately predict very toxic chemicals displaying (LD50 below 300 mg/kg). Further to an in-depth analysis of the misclassified chemicals, we concluded that both pharmacological data (for the reduction of false negatives) and physical-chemical properties (for the reduction of false positives) had to be considered. The modified approach was applied to 76 non-proprietary compounds previously tested with the standard method. A significant improvement in the prediction of the GHS categories was observed. Indeed, 75% of the chemicals pertaining to GHS 1, 2 and 3 were correctly classified, compared to 50% with the standard model. In addition, at an arbitrarily defined LD50 threshold of 500 mg/kg, the sensitivity and specificity were 85% and 89% with the new model against 71% and 83% with the standard model. Future directions will consist of challenging the newly built model with a new set of chemicals and foreseeing the application of such a strategy for repeated dose-toxicity.

Development of an Integrative Approach For The Prediction of Systemic Toxicity : Combination of Cell Toxicity and Metabolism Data
M. Thomas, R. Note, J. Eilstein, L. Bourouf, D. Duché, G. Ouédraogo and J.R. Meunier L’Oréal Research & Innovation, France

Increasing societal concerns for animal welfare and current legislation constraints have made the industry enter a new phase in its innovation and R&D processes. For acute and chronic toxicity testing, which consumes a large number of laboratory animals, no alternative is available, one of the reason being the complexity of the biological processes involved. As such, many research programs have been launched with the aim of developing integrative approaches that would accurately predict such an endpoint. The purpose of the study presented herein was to develop an integrated testing strategy on the basis of the data previously generated with The Ctox panel®, a multiparameter, cell-based in vitro system and the Solidus® DataChip/MetaChip platform, developed to assess metabolism-mediated toxicity. The set used consisted of 63 proprietary chemicals, categorized as Toxic (25 compounds) and Non Toxic (38 compounds) on the basis of a LD50 threshold of 500 mg/kg in rat following an oral administration. A statistical analysis of the data led to the construction of several integrative models. The most predictive model (discriminant analysis) required the consideration of a total of 22 parameters. On the basis of a LD50 threshold of 500 mg/kg, the sensitivity and specificity of the prediction was 92% and 87%. The next steps will consist of challenging the model with a set of diverse chemical classes. At last, it is noteworthy that the number of parameters considered for the model correlates well with the complexity of the endpoint mentioned above.

Optimizing Fluorescent Protein Choice for Transgenic Embryonic Medaka Models
A. Tindall ⁽¹⁾, N. Loire ⁽¹⁾, G. Lemkine ⁽¹⁾, E. Barbeau ⁽²⁾, and Marc Léonard ⁽²⁾⁽¹⁾ WatchFrog, Evry France
⁽²⁾ L'Oréal, Research & Innovation France

Transgenic medaka (Oryzias latipes) have proved to be pertinent models for a variety of ecotoxicological and research applications. The medaka embryo model has much promise as an in vitro model to investigate the toxicity potential of chemicals on a developing vertebrate.

However, a number of technical issues need to be solved such as limiting the impact of auto-fluorescence at different embryonic stages, which can be problematic for the quantification of biomarkers in vivo. To overcome this problem, we have determined the spectrum of emission wavelengths of different developmental stages of medaka submitted to a broad range of excitation wavelengths.

For each developmental stage tested, ten medaka embryos were individualized in a 384 well plate. Each well was subjected to excitation wavelengths from 350-670nm with 5nm increments using a SpectraMax M2 (Molecular Devices). For each excitation wavelength, emitted light was quantified from 20nm above the excitation wavelength to 700nm in 5nm increments. All measures were repeated 6 times.

The results show a fairly homogenous level of autofluorescence across the spectrum prior to hatching with almost undetectable levels at 6 days post fertilization. After hatching, higher levels of autofluorescence were observed in specific regions of the spectrum from blue to red (375-700nm) emission with excitation wavelengths close to the emission wavelength. This information will allow selection of optimal reporter genetic constructs, providing high signal to noise ratio for the quantification of fluorescence. Furthermore, this will pave the way to combining multiple biomarkers with different fluorescent proteins to detect various signals within the same organism.