Adaptation of the KeratinoSens TM Skin Sensitization Test to Animal-Product-Free Cell Culture

560 Received January 31, 2017; Accepted March 16, 2017; Epub March 16, 2017; doi:10.14573/altex.1701311 The KeratinoSensTM method addresses the second key event of skin sensitization, i.e., the activation of keratinocytes. KeratinoSensTM cells are derived from the human keratinocyte cell line HaCat, containing a luciferase gene that is under the control of a constitutive promoter fused with the antioxidant response element (ARE) from a gene that is known to be up-regulated by contact sensitizers. The majority of skin sensitizers induce this pathway and, therefore, the luciferase signal reflects the activation by sensitizers of endogenous Nrf2 dependent genes. This method was fully validated and gained regulatory acceptance in 2015 (Natsch et al., 2010, 2011, 2013). The KeratinoSensTM method represents a great advancement in the replacement of animal testing. However, the published protocol currently includes the use of animal-derived cell culture components (bovine serum and porcine trypsin). XCellR8’s mission is the full replacement of animal testing for cosmetics and their ingredients globally, motivated by an industry requirement to maximize the human relevance of in vitro models, along with


Introduction
Skin sensitization is a multi-step process that has been described as an Adverse Outcome Pathway (AOP).Following skin penetration, chemicals may bind covalently to skin proteins (key event 1), followed by the activation of keratinocytes (key event 2) and activation of dendritic cells (key event 3).Dendritic cells migrate to the lymph nodes and induce the proliferation of T-lymphocytes (key event 4).
Under pressure from legislation, including the European Cosmetics Regulation 1223/2009 and REACH (Registration, Evaluation, Authorization and restriction of CHemicals), many groups have worked in recent years on identifying relevant alternatives to traditional animal methods to determine the skin sensitization potential of chemicals (Reisinger et al., 2015).Alternatives for key events 1-3 have recently been adopted as OECD guidelines: DPRA (OECD TG 442c), KeratinoSens™ (OECD TG 442d) and h-CLAT (OECD TG 442e).
Aldrich UK) and Geneticin (500 µg/ml, Life Technologies) at 37°C, 5% CO 2 to a maximum passage number of 22.They were grown to a maximum of 80-90% confluence.Sub-culture was performed using TrypZean ® (Sigma-Aldrich), a recombinant, animal-free equivalent of porcine trypsin.

KeratinoSens™ assay
The KeratinoSens™ assay was performed as previously described (Emter et al., 2010;DB-ALM, 2013) with the following adaptations and optimization under animal-product-free conditions.
Cells were seeded in 3 parallel replicate 96-well plates for luciferase activity testing and in one 96-well plate for MTT viability testing in medium containing 10% human serum.Plates were incubated at 37°C, 5% CO 2 for 24 h prior to treatment.At the dosing time point, culture medium was removed and replaced with fresh culture medium containing 1% human serum and 1% DMSO.Cells were then exposed to the reference chemicals and controls at 12 different concentrations (max.2000 µM) and incubated for 48 h at 37°C, 5% CO 2 .After the incubation period, luciferase activity was evaluated by luminescence measurement (Luciferase Assay System, Promega, UK) and cell viability was evaluated using the MTT viability assay (Sigma-Aldrich).The assay was repeated 3 times.For each chemical in each repetition and at each concentration, the luciferase gene induction was compared to the negative control (1% DMSO).A maximum of 3 outliers out of 9 values were removed to obtain a CV value below 20%.As per the VRM (Validated Reference Method), the following parameters were calculated: the I max value (maximal average fold induction of luciferase activity observed at any concentration of the tested chemical and positive control); the EC 1.5 value, representing the lowest concentration for which induction of luciferase activity is above the 1.5 fold threshold (i.e., 50% enhanced luciferase activity); and the IC 50 and IC 30 concentration values for 50% and 30% reduction of cellular viability, respectively.
The positive control, cinnamic aldehyde, was used at a concentration range of 8 to 128 µM and required to meet the following acceptance criteria: average induction (I max ) in the three replicates for cinnamic aldehyde at 32 µM between 1.6 and 3; EC 1.5 value between 6 µM and 39 µM.
Test items were considered as positive for skin sensitization if the following conditions were met in 2 of 3 repetitions: The I max was higher than 1.5 fold and statistically significantly different as compared to the solvent (negative) control (as determined by a two-tailed, unpaired Student's t-test); the cellular viability was higher than 70% at the lowest concentration with induction of luciferase activity above 1.5 fold (i.e., at the EC 1.5 determining concentration); the EC 1.5 value was less than 1000 µM; there was an apparent overall dose-response for luciferase induction.

Results
The KeratinoSens™ protocol adapted to animal-free conditions was performed to predict the skin sensitization potential of 21 reference chemicals, including those listed in the Performance Standards (OECD, 2015) and OECD TG 442d.animal welfare considerations.This strategy includes eradication of the significant number of animal-derived components that are still widely used in cell culture methods (including serum, tissue extracts and antibodies).In this project, minor adaptations to an existing method enable the complete replacement of animals for testing key event 2 in the skin sensitization AOP.They enable the method described in OECD TG 442d to be conducted without the need for a number of animal components by replacing them with human equivalents.Key animal components replaced are fetal calf serum and porcine trypsin.The significant animal welfare concerns around the production of these components has been well documented (Jochems et al., 2002;van der Valk et al., 2017).
Here, we describe the internal validation of the modified method using 21 reference chemicals, including those described in the Performance Standards (OECD, 2015) for skin sensitization and in OECD TG 442d.The reference chemicals are comprised of a panel of different sensitizer categories (minimal to severe and non-sensitizers), as defined by the animal-based Local Lymph Node Assay (LLNA).

Cell culture
The KeratinoSens™ cell line contains a stable insertion of a luciferase gene under the transcriptional control of a constitutive promoter fused with the ARE (Antioxidant Response Element) from the AKR1C2 gene (Lou et al., 2006).KeratinoSens™ cells were developed by Emter et al. (2010) and purchased from Givaudan.Upon first thawing from cryopreservation, cells were adapted to cell culture medium containing human serum.Cells were routinely maintained in Dulbecco's Modified Eagle Medium (DMEM) containing Glutamax (Life Technologies), 10% human serum (pooled human male AB plasma, 60-70 donors, Sigma-For each chemical, the I max value and the EC 1.5 concentration were calculated (Tab. 1) (I max maximal average fold induction of luciferase activity observed at any concentration of the tested chemical and positive control); the EC 1.5 value, representing the lowest concentration for which induction of luciferase activity is above the 1.5 fold threshold (i.e., 50% enhanced luciferase activity).
Among the 8 reference compounds classified as non-sensitizers in vivo (based on published LLNA results), one was misclassified (4-methoxy-acetophenone), in common with the KeratinoSens™ VRM.
Regarding the 13 chemicals classified as sensitizers in vivo (based on published LLNA results), 2 were misclassified (phenyl benzoate and eugenol), in common with the KeratinoSens™ VRM.
Results show that the prediction for the 21 reference chemicals was identical to the VRM (Emter et al., 2010;Natsch and Emter, 2008).
Both assays correctly predicted 18 chemicals of the total 21 chemicals and incorrectly predicted the same 3 chemicals, when compared with LLNA results: 4-methoxy-acetophenone was predicted as a skin sensitizer ("false positive") while phenyl benzoate and eugenol were predicted to be non-sensitizers ("false negatives").In compliance with the acceptance criteria of the Performance Standards, no strong or extreme sensitizer was under-predicted with the adapted method.
Furthermore, in the case of eugenol, the result obtained in our laboratory was a "borderline" classification, as the 2 last concentrations were determined as positive but with a cell viability below the 70% threshold.
Tab. 1: Results obtained for the 21 reference chemicals (mean of 3 independent repetitions in triplicate) These results are compared to VRM KeratinoSens™ assay results and to in vivo classification of the chemicals, based on EC3 values from literature results of the LLNA test method: > 10 weak, < 10 and > 1 moderate, < 1 and > 0.1 strong, < 0.1 extreme (OECD, 2015).The 3 underlined chemicals were misclassified as compared to in vivo results.S, sensitizer; NS, non-sensitizer; I max , maximal fold gene induction; EC 1.5 , concentration in µM for 1.5-fold induction; n.i., no induction.

Reference substances
In (cinnamic aldehyde) concentration range optimized to 8-128 µM for the adapted test conditions.The KeratinoSens™ cell line adapted well to routine culture in human serum (in place of bovine serum), showing healthy morphology and comparable growth rates.This enabled the rapid creation of an internal cell bank for long term use in the test.A maximum passage number of 22 was set to reflect the optimum passage window for assay performance using the cells cultured in human serum.
Since there have been no changes to the cell lines or key steps in the protocols, the changes constitute an adaptation to OECD TG442d, and not a "me-too" method.
Subsequently, the adapted method was validated in-house using the 10 proficiency chemicals listed in OECD TG 442d, and the 11 additional chemicals listed in the associated Performance Standards (of the 20 chemicals listed in the Performance Standards, 9 were already present in the Proficiency Chemicals listed in OECD TG 442d).These chemicals spanned the full range of levels of skin sensitization observed in the LLNA, from extreme to non-sensitizers.The adapted method showed full concordance of all 21 chemicals with the VRM, i.e., the KeratinoSens™ published protocol.In common with the VRM, the adapted method identified two "false negatives" (phenyl benzoate and eugenol) and one "false positive" (4-methoxyacetophenone) when compared with the LLNA.It is worth noting that the classification of chemicals as "false positive" or "false negative" by in vitro tests makes an assumption that the classification generated by the animal in vivo test (in this case the LLNA) is predictive of the human response.This is a common practice in the development of in vitro methods that is worthy of review.
The Performance Standards stipulate that "The accuracy, sensitivity and specificity of the proposed similar or modified test method should be comparable or better to that of the VRM.The accuracy, sensitivity and specificity obtained with the 20 reference substances should all be equal or higher (≥) than 80.0% (actual for KeratinoSens™ based on the 20 reference substances and using a weighted calculation: 87.0% accuracy, 86.7% sensitivity and 87.5% specificity)." All acceptance criteria of the Performance Standards were met as follows: Results obtained showed that the accuracy of the adapted method was 85.7%, the sensitivity was 84.6%, and the specificity was 87.5%.The "false positive" and "false negative" results were the same as those obtained with the VRM and no strong or extreme sensitizers were under-predicted.The cell line was the one used in the VRM and the endpoint was the same (luciferase activity measurement).
Our animal-product-free adaptation of the KeratinoSens™ test has proven to be robust and reproducible in our Good Laboratory Practice (GLP) accredited laboratory for over two years, and is performed in full compliance with GLP requirements.
The adapted method provides added value from both scientific and ethical perspectives.From a scientific perspective, the eradication of animal-derived components from in vitro test systems, and their replacement with human equivalents, directly increases the relevance of the test system to humans and would be expected to at least equal, if not enhance, the predictivity of the tests across a wide range of chemical categories.The adaptations also provide

Discussion
Many advances have been made in terms of achieving full regulatory acceptance for scientifically advanced in vitro methods to replace traditional animal-based safety testing.Examples include: OECD TG431, OECD TG439 and OECD TG492 using reconstructed human tissue models to assess skin corrosion, skin irritation and eye irritation, respectively.Recently, three methods for the assessment of key events in the skin sensitization AOP have gained full acceptance and have been incorporated into the REACH Regulation in Europe as the methods of choice for this endpoint, i.e., OECD TG442c, OECD TG442d and OECD TG442e.The traditional animal-based method for the assessment of skin sensitization, OECD TG429, the Local Lymph Node Assay (LLNA), may now only be used as a last resort, representing a significant shift towards the adoption of in vitro technologies in the context of regulatory safety testing.
In spite of this substantial progress, it is important to recognize that many in vitro methods currently use animal-derived components such as bovine serum, mouse antibodies, rat tissue extracts, and a variety of animal-derived cell lines.As such, these methods still ultimately require the sacrifice of animals and cannot be considered completely animal-free.Such an approach presents scientific limitations by unnecessarily compromising the direct relevance of the test system to humans.The significant animal welfare concerns around the production of these components has also been well documented (Jochems et al., 2002;Even et al., 2006;van der Valk et al., 2017).Importantly, these scientific and ethical drawbacks give rise to concerns within the cosmetics and personal care industry.The European Cosmetics Regulation 1223/2009 includes an animal testing ban for cosmetic products and ingredients.
Globally, animal testing is a highly emotive issue in many countries, and there is increasing demand from consumers for "cruelty-free" cosmetic products, in which neither the formulation nor individual ingredients have been tested on animals.Many ethical cosmetic companies seek to go beyond the "bare minimum" expectation, and to satisfy consumer demand by avoiding the use of animal-derived components altogether.In response to this demand, new methods developed at XCellR8 are animal-product-free from the outset, and the organization openly encourages other laboratories to adopt the same approach, avoiding the need for adaptation after the formal process of validation and regulatory acceptance has been completed.In addition, we have been asked by leading cosmetic companies to adapt existing regulatory methods to animal-product-free conditions, and to seek regulatory acceptance for the updated protocols.
In this project, the widely used KeratinoSens™ test method for skin sensitization (OECD TG 442d) was adapted to animal-product-free conditions and further optimized by making the following minor adaptations to the established protocol: 1.The use of human serum in cell culture (OECD TG 442d states only "serum" but the Validated Reference Method (VRM) protocol states "foetal calf serum"); 2. non-animal recombinant trypsin (TrypZean ® ) used to harvest cells; 3. MTT method adapted to the widely-used protocol consisting of 3 h incubation, solubilization in isopropanol and absorbance measurement at 570 nm; 4. positive control