Drug-induced skin toxicity: gaps in preclinical testing cascade as opportunities for complex in vitro models and assays

Rhiannon N. Hardwick *a, Catherine J. Betts b, Jessica Whritenour c, Radhakrishna Sura d, Maike Thamsen e, Elad H. Kaufman f and Kristin Fabre g
aTranslational Safety Sciences, Theravance Biopharma, US, Inc., South San Francisco, CA, USA. E-mail: rhardwick@theravance.com
bPathology Sciences, Drug Safety and Metabolism, IMED Biotech Unit, AstraZeneca, Cambridge, UK
cPfizer, Inc., Drug Safety Research and Development, Eastern Point Rd, Groton, CT 06340, USA
dPreclinical Safety, Abbvie Inc., North Chicago, IL, USA
ePharmacology, Theravance Biopharma, US, Inc., South San Francisco, CA, USA
fBiology, Theravance Biopharma, US, Inc., South San Francisco, CA, USA
gMPS Center of Excellence, Drug Safety & Metabolism, IMED Biotech Unit, AstraZeneca, Waltham, MA, USA

Received 1st June 2019 , Accepted 17th September 2019

First published on 10th October 2019


Skin is the largest organ of the body and serves as the principle barrier to the environment. Composed of multiple cell types arranged in stratified layers with highly specialized appendages, it serves sensory and immune surveillance roles in addition to its primary mechanical function. Several complex in vitro models of skin (i.e. microphysiological systems (MPS) including but not limited to 3D tissues, organ-on-a-chip, organoids), have been developed and assays validated for regulatory purposes. As such, skin is arguably the most advanced organ with respect to model development and adoption across industries including chemical, cosmetic, and to a somewhat lesser extent, pharmaceutical. Early adoption of complex skin models and associated assays for assessment of irritation and corrosion spurred research into other areas such as sensitization, absorption, phototoxicity, and genotoxicity. Despite such considerable advancements, opportunities remain for immune capabilities, inclusion of appendages such as hair follicles, fluidics, and innervation, among others. Herein, we provide an overview of current complex skin model capabilities and limitations within the drug development scheme, and recommendations for future model development and assay qualification and/or validation with the intent to facilitate wider adoption of use within the pharmaceutical industry.


Introduction

The skin is the largest organ in the body; built to be mechanically flexible, but tough to protect against external mechanical, chemical, and biological challenges.1,2 The two main layers of the skin are the epidermis and dermis.3 The epidermis is made up of keratinocytes that primarily form the barrier of the skin, while the dermis contains the blood vessels, the majority of the extra-cellular matrices (which are responsible for the mechanical properties of the skin), fibroblasts and other mesenchymal cells, hair follicles, and sweat glands.3 Below the dermis is a subcutaneous layer that consists of loose connective tissue, elastin and cells such as fibroblasts, macrophages and adipocytes. Extensive innervation throughout allows the transmission of important touch, temperature, pain, and itch sensations. Given the critical role for skin, it is imperative to minimize adverse effects from therapeutics in order to maintain its physical barrier for protection, thermoregulation, sensation, homeostasis, and immunity. Adverse events for skin can primarily arise by 1) direct exposure to exogenous factors, such as topically applied therapeutics, or 2) indirect effects such as side effects from various systemically administered therapeutics (illustrated in Fig. 1). Given the potential frequency for such events to occur, which may be more or less dependent on the route of administration, there is a need for pharmaceutical companies to predict and/or potentially mitigate dermal adverse events.
image file: c9lc00519f-f1.tif
Fig. 1 Structure of human skin illustrating relationship between route of administration leading to direct versus indirect effects. Direct effects can result from topical application and may manifest in lower layers of the skin depending on permeability of the drug in the model. Indirect effects may result when a systemically administered drug (e.g., oral) is delivered to the skin via systemic circulation where accumulation may be possible depending on hydrophilicity/hydrophobicity.

Several informative reviews are available on microphysiological (MPS) skin models as well as their potential applications in drug development, and are encouraged reading.4–8 Herein, we aim to further such discussions by providing an overview of MPS skin model use within the pharmaceutical industry with particular attention to assessment of direct versus indirect adverse effects. Commentary on advantages with respect to models and assays, as well as opportunities for improvement also are included. Additionally, a perspective on future model development to fit the needs of pharmaceutical scientists is provided with recommendations for model assessment and endpoints of particular interest. It should be noted that for the purposes of the current discussion, the authors collectively define MPS as expanding beyond traditional 2D sandwich culture or monolayers and may include models with one or several of the following design aspects: having a multi-cellular environment within biopolymer or tissue-derived matrix, a 3D structure, mechanical factors such as stretch or perfusion (e.g. breathing, gut peristalsis, flow, etc.), incorporating primary or stem cell-derived cells, and/or inclusion of immune system components.

Direct toxic effects on skin

Significant efforts have been made to better predict skin toxicity when directly exposed to compounds or chemicals. Dermal toxicity assessment of compounds and subsequent hazard identification can generally be categorized as corrosion, irritation, and sensitization (Table 1). Many commercial models have been validated for specific uses by the European Centre for the Validation of Alternative Methods (ECVAM) and the Interagency Coordinating Committee on the Validation of Alternative Methods (ICCVAM). Subsequent guidelines have been accepted by the Organization for Economic Cooperation and Development (OECD), with a goal of standardizing methods for the testing and assessment of chemicals. As a result, methods have been codified for the assessment of skin corrosion and irritation following direct topical application (Fig. 1), as well as a battery of sensitization assays. While genotoxicity and phototoxicity assays remain in various stages of development.
Table 1 Dermal hazard categorization according to UN GHS11
Hazard Category/sub-category Hazard criteria
Corrosion 1 Corrosive—irreversible damage including necrosis through epidermis into dermis following ≤4 hours exposure
1A Corrosivity following exposure ≤3 minutes with ≤1 hour observation
1B Corrosivity following exposures >3 minutes and ≤1 hour with ≤1 hour observation
1C Corrosivity following >1 hour and ≤4 hours exposure with ≤14 days observation
Irritation 2 Irritant—reversible damage following ≤4 hours exposure with persistent inflammation, alopecia, hyperkeratosis, hyperplasia, or scaling with ≤14 days observation
Erythema/eschar mean score or edema mean score ≥2.3 < 4.0
Mild irritant—reversible damage following an exposure ≤4 hours
3 Erythema/eschar mean score ≥1.5 < 2.3
Sensitization 1A High frequency of sensitization occurrence in humans or high potency in animals with presumed potential to elicit effects in humans
1B Low/moderate frequency of sensitization occurrence in humans or low/moderate potency in animals with presumed potential to elicit effects in humans


The OECD test guideline 404 (OECD TG 404) has been used historically to evaluate the ability of test articles to produce skin toxicity following topical administration, and relies upon the scoring system originally established by Draize for assessment of erythema, eschar, and edema in rabbits.9,10 This system has been used as the base against which in vitro assays are compared for irritation and corrosion predictive capability following direct, topical application of a compound. A corrosive agent is the most severe classification and one that produces irreversible skin damage following application of a test article for up to 4 hours. It is characterized by ulceration, bleeding, scabbing, and necrosis through the epidermis reaching into the dermis.9,11 An irritant is classified as an agent that produces reversible damage following test article application for up to 4 hours with persistence through a typical observation period of 14 days, and may include inflammation, alopecia, hyperkeratosis, hyperplasia, or scaling.9,11 Skin sensitization is a term used to indicate the potential of a chemical to produce contact allergic dermatitis, a delayed T-cell-mediated response.12–14 The development of contact allergic dermatitis can primarily be described in two phases: induction and elicitation, and has been described in detail within a skin sensitization adverse outcome pathway (AOP) generated by the OECD, discussed below.15

Indirect toxic effects on skin

While the advancement of targeted and immune-based therapies has significantly improved patient outcomes, dermal adverse effects following systemic administration of a compound (Fig. 1) have become significant and can often lead to dose limitations. For example, incidence of skin reactions can be as frequent as 50–100% (varied based on type of skin toxicity) with compounds that target epidermal growth factor receptor (EGFR).16 Targeted therapies, such as EGFR, VEGF (vascular endothelial growth factor), BRAF, MEK (mitogen activated protein kinase), and mTOR (mechanistic target of rapamycin kinase) inhibitors exploit cell networks that contribute to cancer development and growth, but also impact the normal tissues that use these pathways for a healthy and constant state of proliferation and renewal, including skin.16–18 Common adverse events include papulopustular eruption, xerosis, pruritus, hand–foot syndrome, pigmentary changes, edema, foliculitis, and alopecia. Based on the Common Terminology Criteria for Adverse Events,19 severity ranges from mild to severe, with the latter having the potential to require an adjustment in treatment that could reduce the amount of drug needed to be fully effective in fighting the disease. Mitigative strategies can help, but there are numerous instances where the patient's quality of life and health are negatively impacted. Pain, risk of infection, and in some cases, life-threatening occurrences such as Stevens–Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) have been documented.20

Skin is one of the most frequently affected organs associated with hypersensitivity reactions for systemically administered small molecule drugs.21 Similar to other drug-induced indirect effects on skin, severity of these reactions can range from mild to severe. One or more mechanisms leading to the generation of an adaptive immune response against the drug or direct interaction of the drug with T-cells may be at play. Although all of the risk factors have not been defined, it is likely that there are both drug-specific characteristics as well as patient-specific factors that contribute to the risk of developing a drug hypersensitivity reaction.

Skin reactions can also be a frequent development in patients treated with checkpoint inhibitor immunotherapies such as cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4), programmed cell death protein (PD-1), and its ligand PD-L1, creating an immune-related adverse event.22 Not surprisingly, skin effects are similar to that of a hypersensitivity reaction because the therapy stimulates the patient's immune system to attack tumor cells, and this increase in immune activation can result in overt inflammation and attack on healthy tissues and organs, comparable to an autoimmune disorder. Similar to targeted therapies, rare, albeit severe, cases of SJS and TEN also have been observed.

Other therapies, such as some dipeptidyl peptidase 4 (DPP-4) inhibitors have raised concerns due to rare skin reactions, which have been postulated to be associated with expression of DPP-4 (also known as CD26) on T-cells.23–29 Skin lesions such as flaking, ulceration, and scabbing with some, but not all DPP-4 inhibitors has been detected in nonclinical species.23,24 However, in post-approval surveillance multiple DPP-4 inhibitors have been shown to produce other skin lesions not detected in nonclinical species such as hypersensitivity reactions, pruritus, SJS, and even rare, debilitating bullous pemphigoid.23,25–27,29 Regardless of the underlying cause, dermal adverse events resulting from systemic therapeutics can severely impact quality of life, and have the potential to alter the course of treatment.

MPS: expanding opportunities in skin research

Recent advances in MPS development for several organ systems have significantly transformed the pharmaceutical scientist's approach to drug development. Many such models offer a human-derived system (availability of nonclinical species can be limited depending on organ system) in which histological endpoints may be evaluated in vitro in addition to traditional biochemical, transcriptomic, and proteomic approaches. However, their specific applications and the degree to which MPS are employed in the drug development pipeline vary across pharmaceutical companies. Currently, the most advanced organ system MPS with respect to validation and widespread use across industries (pharmaceutical, chemical, cosmetic, etc.) are reconstructed human skin models, specifically reconstructed human epidermis. Several academicians and commercial vendors have created 3D epidermal and/or full thickness (epidermis and dermis) skin models through differentiation of human fibroblasts and/or keratinocytes in an air–liquid-interface culture format. These skin MPS have been shown to exhibit drug metabolizing enzyme gene expression similar to that of human skin, thus providing an improved representation of the in vivo organ as compared to historical 2D cell culture.30,31 A significant advantage afforded by these skin MPS models has been the ability to test the direct effects of compound solutions applied topically to skin (Fig. 1), as well as assessment of complex mixtures such as formulations. Such models are often grown in a transwell-type culture dish, which is readily amenable to topical application, while systemic administration to study indirect effects (Fig. 1) must be modeled in a static manner via addition of the compound to the receiver well. Further, reconstructed skin models that maintain the air–liquid interface allow for application of compounds in evaporative solvents, such as acetone, and solids and semi-solids, such as ointments and creams. However, application of solids and semi-solids poses a challenge with respect to duration of exposure and maintenance of tissue differentiation at the air–liquid interface. Advancements in corrosion and irritation assays utilizing topical application, discussed below, have enabled the possibility of proactive formulation screening for irritation, while new assays are providing possibilities for follow-on genotoxicity and phototoxicity screening. There remains an opportunity for integration of skin sensitization assessment, evaluation of pruritic effects, evaluation of indirect effects following systemic administration, and many others as will be discussed.

Gaps in testing capabilities

In drug development, it is necessary to understand contributions from effects on individual cell types including keratinocytes, immune cells, and neurons, but also integrated effects that occur when these cell types signal in concert. Additionally, the skin condition of the target indication may need to be considered for aspects of increased/decreased barrier permeability, epidermal thickening or thinning, and ongoing inflammation and irritation as related to the disease of interest, particularly for topically applied drugs.32 Other skin condition-related concerns include increased sensory nerve branching due to immune cell and neuronal interactions.33–36 Demonstration of in vivo correlation of skin MPS with potential for assessment in tandem with other organ systems also is needed and may be critical for some systemically administered agents. Additionally, regulatory agencies provide lists of approved topical formulation ingredients and their recommended limits; however, this general guidance does not capture sensitivities specific to certain disease conditions or interactions of the active drug and its vehicle. Advances in model complexity and assay guidelines are necessary to enable efficient predictive screening of candidate topical formulations for local tolerance, irritation, and pain sensation prior to clinical evaluation.

Models and methods to test direct effects on skin

In vitro . In an effort to reduce the use of animals and harmonize the safety assessment of chemicals, OECD has published an Integrated Approach on Testing and Assessment (IATA) for skin corrosion and irritation that allows for a tiered approach to safety testing of directly applied (topical) chemicals.37,38 Specifically, the IATA for skin corrosion and irritation is divided into three parts, the first of which is an assessment of available information (including human and animal data, physico-chemical properties, quantitative structure activity relationships, etc.), followed by a weight of evidence (WoE) analysis, and lastly a roadmap for additional testing. While this approach is quite useful for chemicals with varying amounts of historical data available, a different approach must be considered for new chemical entities such as those faced by a pharmaceutical scientist. For this purpose, the IATA outlines a top-down or bottom-up approach that could be applied in a tiered manner within a proactive screening strategy. In short, if the WoE is inconclusive and the agent or formulation is presumed to have irritation or corrosive potential, an in vitro skin corrosion test followed by an in vitro skin irritation test (if the corrosion test is negative) may be employed in the top-down approach. Conversely if the agent or formulation is presumed not to be an irritant, a bottom-up approach may be employed in which the in vitro skin irritation test is performed first followed by the in vitro skin corrosion test if the irritation test is positive.

An important purpose of the tiered approach to skin corrosion and irritation assessment described in the IATA is to emphasize the use of animal testing as a last resort. Several in vitro assays have been established as acceptable by United States and European regulatory agencies; however, an integrated approach of safety assessment in which animal testing is fully replaced by in vitro methods has not yet been developed and accepted for pharmaceuticals.39 Of particular importance in the pharmaceutical industry is the need to establish the local, dermal toxicity of a compound following topical application, as well as identification of systemic effects following dermal absorption that may pose a risk to human health. Several chemicals have been implicated in producing systemic effects following dermal absorption including herbicides, pesticides, metals, polycyclic aromatic hydrocarbons, and some topical medications.13,40 Thus, it becomes necessary to accurately predict systemic exposure of a pharmaceutical following topical application in order to determine the potential risk to other organ systems. Currently available reconstructed epidermis models have demonstrated increased permeability compared to excised human skin as well as pig skin, which is often used as a surrogate dermal absorption model in drug development.41 An additional important point with respect to skin toxicity assessment is the manifestation of dermal effects following exposure to systemically administered drugs, which will be discussed separately below. While considerable efforts are being made for the assessment of local effects in new in vitro methods, integration with systemic safety assessment is still premature with considerable work to be done.

As mentioned previously, MPS have been successfully employed for the in vitro assessment of skin corrosion and irritation as occurs following topical application (Table 2). OECD Test Guidelines 431 and 439 have provided extensive details for the assessment of corrosion and irritation, respectively, with recommended positive and negative control compounds.42,43 The corrosion assay has been formally validated for use in the EpiSkin™ SM, EpiDerm™ EPI-200, SkinEthic™ RHE, and epiCS® models with EpiSkin™ SM, EpiDerm™ EPI-200, SkinEthic™ RHE, and LabCyte EPI-MODEL24 recommended in the irritation assay. Both protocols rely on reduction of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) as the primary endpoint.42,43 By employing either the top-down or bottom-up approach, the two assays enable the determination of corrosives as UN-GHS11 category 1A, category 1B/1C, and irritants as category 2. The assays are currently unable to differentiate between corrosive categories 1B and 1C, in part due to the minimal number of representative category 1C compounds available for assay validation.42 Additionally, OECD Test Guideline 439 provides a designation of “irritating” or “non-irritating”, but is unable to identify mild irritants (UN-GHS category 3) from non-irritating/no category compounds. Thus, there is currently no direct in vitro method capable of identifying mildly irritating compounds or formulations, which are of greater likelihood (versus strong irritants and corrosives) in pharmaceutical development.

Table 2 Summary of selected in vitro assays available or under evaluation to support skin research
Category Assay Available models OECD TG Primary endpoint
a Not available and/or not yet approved.
Corrosion Transcutaneous electrical resistance (TER) test method Excised skin 43044 TER
Membrane barrier test method Corrositex (artificial membrane) 43545 Color change in receiver fluid
Reconstructed human epidermis (RHE) test method EpiSkin™, EpiDerm™, SkinEthic™, epiCS® 43142 Viability via MTT reduction
Irritation Skin irritation epidermis method EpiSkin™, EpiDerm™, SkinEthic™, epiCS®, EPI-MODEL 43943 Viability via MTT reduction
Sensitization Direct peptide reactivity assay In chemico 442C4,46 HPLC detection of unbound peptides5,47
ARE-Nrf2 luciferase test method KeratinoSens™, (HaCaT human keratinocytes) 442D48,49 Luciferase activity8,50
NCTC 2544 IL-18 test NCTC 2544 cells (human keratinocyte cell line) N/Aa IL-18 levels via ELISA51,52
LuSens assay Immortalized keratinocytes N/Aa Luciferase activity53,54
SENS-IS® assay EpiSkin™ N/Aa qPCR analysis of selected biomarkers5,12,55
SenCeeTox® assay In chemico GSH reactivity N/Aa GSH depletion
Activation of Nrf2 pathway in HaCaT cell or EpiDerm™ qPCR analysis of selected Nrf2 target genes56
Human cell line activation test (h-CLAT) THP-1 cells (human monocytic leukemia cell line) 442E5,57 Upregulation of CD86 and CD5458
U-SENS™ U937 cells (human myeloid cell line) N/Aa Upregulation of CD86 and CD5459,60
GARDskin (genomic allergen rapid detection) MUTZ-3 cells (dendritic cell-like human myeloid cell line) N/Aa Array analysis of 196–200 genes4,5,33,61,62
VITOSENS™ Human CD34+ dendritic cells N/Aa Gene expression analysis of CCR2 and CREM63,64
Human T cell priming assay (hTCPA) Human monocyte-derived dendritic cell and naïve T cell co-culture N/Aa Flow cytometry analysis of T cell activation markers (e.g., IFN-γ and TNF-α)65,66
Genotoxicity Reconstructed skin micronucleus assay EpiDerm™ N/Aa % micronuclei (modification of TG 487)34,37,67,68
3D skin comet assay EpiSkin™, EpiDerm™ FT, Phenion® FT N/Aa Comet tail (based on TG 489)69–71
Phototoxicity 3T3 NRU assay Balb/c 3T3 fibroblasts 43272 Viability ± UVR
EpiDerm™ Phototoxicity test EpiDerm™ H3D-PT N/Aa Viability ± UVR (based on TG 432)72,73
Absorption Skin absorption: in vitro method Excised dermatomed or full thickness skin 42874 Concentration in receptor fluid, skin surface (wash), and skin layers


Skin sensitization is a term used to indicate the potential of a chemical to produce allergic contact dermatitis, a delayed-type T-cell-mediated hypersensitivity response.13,14 As mentioned previously, the development of allergic contact dermatitis can primarily be described in two phases: induction and elicitation, which have been described in detail within the skin sensitization AOP generated by the OECD.15 During induction, the agent must gain access to the epidermis where it can form a hapten, a small molecule–protein complex.14 Haptens may form through spontaneous conjugation with endogenous proteins, such as that observed with highly reactive molecules, or be processed via attachment to a carrier protein by antigen presenting cells (APC) such as Langerhans cells, the resident dendritic cells of the skin. In vivo, the Langerhans cells migrate to local draining lymph nodes where the antigenic complex is then presented to T lymphocytes, leading to activation and proliferation, and subsequent generation of T memory cells.14 The elicitation phase consists of re-exposure to the chemical and recognition by memory T-cells, followed by release of inflammatory cytokines leading to tissue damage.14 The concentration of chemical needed for immune induction is typically higher than that required in the elicitation phase, the response then occurs over the next 24 hours, so producing the delayed-type response characteristic of contact sensitization. Additionally, multiple exposures may be needed in the elicitation phase to manifest a response, and may be an additional factor contributing to the delayed nature of allergic contact dermatitis.

It has generally been accepted that a battery of in vitro assays is necessary to determine the skin sensitization potential of a molecule with no one model completely replacing the in vivo local lymph node assay (LLNA), described below. Each of the currently available assays for skin sensitization testing aim to address a key event in the AOP: reactivity, keratinocyte activation resulting in inflammation and oxidative stress pathway activation, dendritic cell activation and maturation, and T-cell proliferation, with many highlighted in Table 2.15 The direct peptide reactivity assay (OECD TG 442C) is an in chemico test aimed at determining the reactivity of a molecule.46 The ARE-Nrf2 Luciferase Test (OECD TG 442D), NCTC 2544 IL-18 Test, LuSens Assay, SENS-IS® Assay, and SenCeeTox® Assay all address the inflammation and/or oxidative stress component of the AOP.12,48,50,52,54,56 Dendritic cell activation and/or maturation may be addressed with the h-CLAT (OECD TG 442E), U-SENS™, GARDskin, and VITOSENS™ assays.57–64 The final component of the AOP, T-cell proliferation, may be evaluated through the human T-cell priming assay, a complex process involving separation of cell populations from fresh human peripheral blood and carefully timed differentiation and exposure procedures.66 Of the battery of skin sensitization assays, SenCeeTox® and SENS-IS® alone employ the use of skin MPS. Integration of all four key events in the AOP into a single assay, including an MPS model, has not yet been achieved.

Photosensitization generally refers to effects on tissues as a result of treatment with a chemical and exposure to sunlight, and is particularly important for topically administered drugs; however, photosensitivity may also manifest as an indirect effect with systemically administered agents. Photosensitization reactions may be classified as phototoxic or photoallergic. Phototoxicity, also referred to as photoirritation, is an acute light-induced response to a chemical that is photoreactive; whereas a photoallergy is an immunological response to a light-induced antigenic molecule such as a chemical-protein adduct formed by a photochemical reaction.14,75,76 In order to have potential as a photosensitiser, a compound typically must absorb light within the range of sunlight (290–700 nm), generate a reactive species in response to UV visible light absorption, and exhibit sufficient exposure in light exposed tissues.76 To aid in the evaluation of new pharmaceuticals, the International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH), working with an expert group, developed a guideline for assessing photosafety (ICH S10).76 As with irritation/corrosion testing, a tiered approach also is suggested for photosafety assessment. The first step is to analyse the UV-visible spectrum of the molecule to obtain a molar extinction coefficient (MEC) value. Any molecules with an MEC > 1000 L mol−1 cm−1 require additional evaluation, such as a reactive oxygen species assay (ROS) or the 3T3 Neutral Red Uptake (NRU) phototoxicity test.72,76,77 The 3T3 NRU test has been accepted by OECD (OECD TG 432), and evaluates the cytotoxicity of a molecule in the presence and absence of ultraviolet radiation (UVR) exposure (5 J cm−2) via uptake of the neutral red dye. A determination of phototoxicity is made by evaluation of the photo-irritation factor (ratio of cytotoxic IC50 ± UVR) and the mean photo effect (evaluation of concentration response curves).72 Compounds are then rated as phototoxic or not phototoxic, with any equivocal results requiring follow-up in vivo. The 3T3 NRU test does have potential to generate false positives, necessitating additional studies to determine in vivo risk.

A phototoxicity test in the EpiDerm™ model (EpiDerm™ H3D-PT) has been developed for use as an accompaniment to the 3T3 NRU test, or as a method to verify suspected false positives (Table 2).73 The assay, which has previously undergone early testing, is based on cytotoxicity measurement via MTT in the presence and absence of UV exposure (6 J cm−2), and has the advantage of the ability to test compounds in solution or in topical formulation for assessment of direct effects.73 Compounds that directly interact with MTT and thus perturb integrity of the assay pose a challenge.73 Based on the early assessment studies in collaboration with ECVAM, the assay appears to have reasonable ability to detect strong photosensitizers, specifically topically administered compounds, whereas weak photosensitizers remain a challenge. Also, the assay may under-predict the indirect phototoxic potential of systemically administered compounds, leaving opportunity for development and validation of a fluidic-based skin MPS for assessment of photosafety. The ability of the assay to predict photoallergens is unknown, and likely will require incorporation of an immune component. Nonetheless, the EpiDerm™ H3D-PT assay provides another asset in the toolbox for industrial toxicologists though full regulatory acceptance is still needed.

Genotoxicity assessment by the pharmaceutical scientist involves a tiered approach as is recommended for skin irritation and corrosion whereby in vitro assays to screen for mutagenicity and chromosomal damage are employed prior to assessments in vivo.78 Mutagenicity is determined by the bacterial Ames test, while chromosomal damage can be assessed in mammalian cell lines (e.g., in vitro metaphase chromosome aberration, human peripheral blood micronucleus). The assays generally assess the ability of a molecule to cause genetic damage on a global scale with specific target organs often evaluated in vivo. Indeed, should a positive result occur in an in vitro assay, follow-up in vivo tests are then required with evaluation of specific target organs, such as skin, depending upon the needed endpoint with respect to route of administration. Chromosomal damage may be evaluated in vivo in a number of organ systems, and capabilities are heavily dependent upon historical data generated in said organ by the investigational laboratory.

Two assays aimed at providing a means of in vitro dermal genotoxicity assessment are the Reconstructed Skin Micronucleus Test (RSMNT) and the 3D Skin Comet test (Table 2). The RSMNT has undergone extensive interlaboratory and intralaboratory validation with the EpiDerm™ model, and is conducted similarly to the in vitro Mammalian Cell Micronucleus Test (OECD TG 487).37,68,69 The EpiDerm™ model is amenable to cytokinesis block and exhibits a low background frequency of micronuclei. Furthermore, the assay has shown utility in identifying both direct-acting and indirect-acting (metabolic activation required) genotoxic agents with the ability to determine genotoxic mechanism (i.e., aneugen versus clastogen).34,37,79 The in vivo comet assay (OECD TG 489), which is not dependent on cell proliferation, has been adapted for use in the EpiSkin™, Phenion® FT (full thickness) and EpiDerm™ FT models.69–71,80 Each model has been investigated for frequency of background DNA damage with differences identified between models.69 The development of the RSMNT and 3D Skin Comet test has expanded the capabilities of follow-up for early genotoxicity screening assays. Acceptance of the assays by OECD may further expand regulatory decision-making abilities.

In vivo . In the interest of practicing sound science and keeping in mind the principles of 3Rs (replacement, reduction and refinement), a sound rationale needs to be provided before initiating in vivo dermal studies. Data which needs to be evaluated include available in vitro, in silico, and physico-chemical data (such as pH of the material especially if strong acid or base). Only if the WoE is inconclusive, is there a need for in vivo testing for dermal toxicities.

In the in vivo skin corrosion/irritation models, a single dose of the test material (solid/liquid) is applied uniformly to the skin of albino rabbits, and the untreated skin areas of the test animals serve as control. The degree of irritation/corrosion is read and scored at specific intervals to evaluate the reversibility or irreversibility of the observed effects. If no corrosion is observed, there is a follow-up confirmatory test for the irritant or negative response in additional animals. Similarly, equivocal responses also warrant additional animal testing. The grading of skin responses is subjective, hence is always evaluated in conjunction with the nature and severity of lesions and their reversibility or lack of reversibility (OECD TG 404).9

The in vivo test has some advantages such that it gives an indication of the reversibility of an observed effect as well as some information into the potential mode of action. However, given it is a subjective test, there is potential for variability in results interpretation between study personnel. Additionally, the test is conducted in rabbits, and could result in species differences with respect to test article reactivity (i.e., human versus rabbit). Depending on the impurity profile of the test article, results may also vary. Since the test article is applied uniformly to the test area, this can pose a challenge for solid and semi-solid materials.

The in vivo assessment of skin sensitization potential has largely relied upon the LLNA (OECD TG 429), which is regarded as the gold standard for safety assessment against which new in vitro assays are benchmarked, as mentioned previously.14,81 The LLNA specifically evaluates the induction of skin sensitization, and is based on the dose-dependent proliferation of lymphocytes in draining lymph nodes. In an IATA for skin sensitization, the OECD has mapped the flow of events involved in contact allergic dermatitis based on the aforementioned skin sensitization AOP.82 The IATA also outlines several models and assays available to investigate the sensitizing potential of a test article within the context of each key event, as described above. Once the sensitization potential of a test article has been identified, it may then be categorized. The UN GHS offers two sub-categories for skin sensitization, namely 1A and 1B (Table 1).11 Category 1A sensitizers are those with a high frequency of occurrence in humans (typically determined by a patch test) and/or a high potency of effects in animals with a LLNA EC3 ≤2%, where an EC3 refers to the concentration required to produce a 3-fold increase in lymphocyte proliferation above the vehicle control.83 Category 1B sensitizers generally exhibit low/moderate frequency of occurrence in humans and/or a low/moderate potency of effects in animals with a LLNA EC3 >2%.

Models and methods to test indirect effects on skin

In vitro . As discussed earlier, immune responses to systemically administered drugs can lead to hypersensitivity reactions which manifest as skin toxicity, ranging in severity from a mild maculopapular rash to a life-threatening response such as TEN. There are no in vitro tests for specifically predicting the potential of systemically administered drugs to indirectly induce skin hypersensitivity reactions; however, in vitro assays that have been developed for predicting the direct sensitization potential of topically applied chemicals may be of value. A recent review by Corsini, et al., summarizes in vitro (and in vivo) tools that may have utility for testing the potential of systemically administered therapeutics to induce indirect effects on skin.84 Some of the in vitro assays described in Table 2 may prove useful for predicting drugs with the potential to indirectly induce skin toxicity via a hypersensitivity reaction, particularly those assays which assess chemical reactivity, dendritic cell activation, and dendritic cell/T cell co-culture endpoints. Other potential tools include an in vitro assay using THP-1 cells (a human monocytic cell line), which showed promising results for several drugs associated with hypersensitivity reactions using IL-8 production and CD86 upregulation as endpoints.85 In addition, a human skin explant model has also been evaluated for its ability to predict the sensitization potential of chemicals,86 and more recently, of systemically administered drugs.87 While these and other assays may show promise, until a greater number of drugs are tested, the true potential of these tools for predicting immune responses to systemically administered agents remains unknown. It should also be noted that unlike the case for sensitization following topical administration of chemicals where protein reactivity (haptenation) is the first key event of the AOP, other mechanisms may be involved in driving immune responses to systemically administered compounds. For example, the p–i (pharmacological interaction with immune receptors) concept, a model that proposes that non-covalent interactions of a drug with a T-cell receptor and HLA (human leukocyte antigen) molecule can lead to the activation of drug-specific T-cells and hypersensitivity reactions,88 suggests that haptenation is not a prerequisite for inducing immune responses to a systemically administered drug. In addition, HLA genotype also has been shown to play a role in susceptibility to skin hypersensitivity reactions for some drugs,89 a potential risk factor not addressed with the current in vitro tools for assessing skin sensitization.

Less attention has been directed to improved prediction of skin hazards following systemic administration in the validation of in vitro assays. In addition to immune reactivity, reliable prediction of irritation and phototoxicity in currently available assays, as mentioned briefly above, is limited. While the gap has been acknowledged, specific methods to overcome these limitations are still needed and of importance in pharmaceutical development. As an example, underprediction of systemically administered phototoxicants hypothetically may be alleviated with application of a fluidic component to a skin MPS; however, this remains an unknown. Further model and assay development with respect to assessment of indirect effects is strongly encouraged.

In vivo . Similar to the in vitro situation, there are no well characterized or validated in vivo models for predicting the potential of a systemically administered drug to induce hypersensitivity reactions. Within the past two decades, in vivo rodent models have been evaluated, and to some extent, have shown promise.84 Largely, these models are variations of the LLNA that incorporate subcutaneous or oral administration of drugs as opposed to topical application. However, because of the limitations of these in vivo models, including the lack of a human relevant MHC component, poor understanding of the mechanism(s) by which the drugs induce lymphocyte activation/proliferation, and the unknown translatability of the findings to humans, these models are not routinely used during drug development.

Opportunities for development and recommendations

Despite the many advancements and capabilities with skin models made thus far, several gaps in capabilities necessary for proactive screening of systemically administered pharmaceutical agents remain, while further assay and model development would expand investigative capabilities for topically applied agents. One key gap is the availability of skin models and validated assays with functional immune cells resulting in an immunocompetent system. This is critical for the skin, a barrier organ where detection and reaction to pathogens is a key role, and for which the immune system is critical in mediating many of the adverse toxicity reactions discussed previously. An in vitro skin MPS comprising immune cells with proven assay guidelines would permit understanding of innate and initial immune induction responses to a pharmaceutical agent. This would enable improved translation of in vitro toxicity assessment for issues such as irritation and sensitization. In addition, coupling of a tissue system containing relevant immune cells to a fluidic system would enable improved pharmacokinetic/pharmacodynamic screening abilities. Furthermore, incorporation of a separate lymphoid-like reserve of circulating immune cells (lymphocytes, monocytes and phagocytes) would potentially enable investigation of adverse immune reactions that are initiated systemically but manifest in the skin and perhaps, investigation of immune-mediated tissue recovery.

Additional gaps crucial to pharmaceutical development include improved barrier function to enable more accurate predication of skin permeability and subsequent systemic exposure with respect to topically applied pharmaceuticals. However, it is appreciated that many factors may contribute to establishing a barrier model that is truly representative of human skin. For example, full thickness models representing epidermis and dermis as opposed to epidermis only, as well as addition of adipose, would be expected to have a more representative permeability profile. Additionally, application of fluidics as well as inclusion of an adipose compartment with or without endothelial cell vascularization and addition of hair follicles are expected to have an influence on the overall permeability profile of a model.

Thus, while several desired model features will be discussed in isolation in the remaining sections and are presented separately in Table 3, it is expected that many biological and anatomical improvements may work in concert to achieve one or more desired functional outcomes as depicted in Fig. 2. Irritation, corrosion, and sensitization stimuli and endpoints have been detailed by ECVAM and OECD.15,90,91 Additional endpoints for inflammation, barrier integrity, pruritus, and sensory neuron activity are suggested in Table 4.

Table 3 Desired features and recommendations for future skin model research
Desired feature Measurement and/or desired outcome Recommended test compounds(s) and/or stimuli
Extended tissue longevity ≥2 weeks viability and stability of drug metabolizing enzyme gene expression/function No specific recommendation
Improved ability to model, investigate, and modulate disease mechanisms
Improved barrier function TER Testosterone, caffeine hydrocortisone, tape stripping, Th2 cytokine mix
TEWL
Tissue permeability reflective of normal and/or diseased human skin
Full thickness tissue assay validation Irritation/corrosion predictivity Eskes, et al., 200790 and Casati, et al., 200991 ECVAM/OECD irritation and sensitization reference compounds4
Sensitization predictivity
Immune component Minimum of Langerhans cell inclusion with demonstrated langerin expression Eskes, et al., 200790 and Casati, et al., 200991 ECVAM/OECD irritation and sensitization reference compounds4
Sensitization predictivity
Adipose/vascularized compartment Improved permeability prediction, tissue health/longevity No specific recommendation
Leptin and glycerol production
Fluidics Systemic exposure estimation Testosterone, caffeine, hydrocortisone
Circulating immune cell component to enable modelling of cell recruitment to damaged/diseased tissue
Validated methods for assessment of repeat-dosing effects
Pigmentation Melanocyte inclusion A-MSH, basic fibroblast growth factor (FGF-2)
Responsive to α-melanocyte stimulating hormone (α-MSH)
Potential to investigate pigmentation disorders (e.g., vitiligo)
Hair follicle Mimic permeability of hydrophilic molecules Valproate, isoniazid, testosterone, minoxidil
Ability to investigate alopecia and hirsutism
Innervation Ability to measure and investigate pruritus ACE inhibitors, allopurinol, amiodarone, opioids, simvastatin, capsaicin, dust mite allergen, histamine, cowhage, trypsin, chloroquine
Histopathological analysis Scoring system with in vivo correlates No specific recommendation
Genotoxicity Micronucleus and comet validation and guideline acceptance in epidermis and/or full thickness models N-Ethyl-N-nitrosourea (ENU), 7,12-dimethylbenzanthracene (DMBA), mitomycin C, cyclophosphamide, vinblastine79
Photosensitivity Phototoxicity validation and guideline acceptance in epidermis and/or full thickness models Tetracycline, chlorpromazine, vemurafenib, nalidixic acid, voriconazole, anthracene, promethazine, chloroquine, amiodarone, norfloxacin
Development of photoallergenicity assay



image file: c9lc00519f-f2.tif
Fig. 2 Desired physical features in future skin models. Several physical features desired in future skin models are shown including fluidics and drug administration considerations. It is expected that many features may work in concert to achieve an improvement such as that of improved barrier function.
Table 4 Recommended markers
Feature Markers
Skin inflammation IL-17, IL-2, IL-4, IL-6, IL-12, IL-8, IL-10, TNF-α, IgE, IFN-γ, TGF-β, MIG
Skin barrier & lipid synthesis Filaggrin, involucrin, loricrin, S100A7, keratins, elafin, cystatin A, desmosomal proteins (desmocollins & desmogleins), sphingomyelin, phosphodiesterase, kallikrein-related peptidases, ceramide, synthases, stearoyl CoA desaturase, ELOVL1, ELOVL6
Pruritus IL-31, TSLP, IL-4, IL-13
Skin sensory neuronal activity TRPV1, TRPA1, S1PR3, TRPM8, P2RX3, MrgprX1, HRH1, F2RL1 (PAR2), F2RL3 (PAR4), PIEZO1, PIEZO2


Inclusion of appendages and additional compartments

Skin is a complex organ with three distinct layers namely: epidermis, dermis and subcutaneous adipose tissue. The hair follicle is the major appendage that traverses the three layers of the skin and acts as a gatekeeper as well as portal of entry for some hydrophilic materials coming in contact with skin. It also serves a role in thermoregulation and external appearance. Other skin appendages include sebaceous gland, sweat gland, arrector pili muscle, and nerves, all of which enable skin to maintain its integrity and homeostasis.

Current commercially available skin MPS are derived from keratinocytes and fibroblasts but do not contain endothelial cells, a full complement of immune cells, adipose tissues or skin appendages, hence their barrier properties are remarkably lower than the human skin, and also fail to mimic the complexity of human intact skin. Addition of the different cell types will improve the utility of the models and potentially reduce model/assay variability (Fig. 2, Table 3).

iPSCs have been explored for the development of skin MPS due to their ability to differentiate into a wide variety of cells, including those that are more difficult to isolate. An ideal in vitro model should have all the three layers of the skin and iPSC have an opportunity to fill void of missing cell types in current models.

Sensory afferent neurons in the skin comprised predominantly of unmyelinated C-fibers and thinly myelinated Aδ-fibers contribute to sensations of pain and itch,92 but may also contribute local signals of neurogenic inflammation. Some of these sensations are adaptive in nature and necessary to protect from injurious insult. Others are physiological responses related to the substance to which the skin is exposed. High doses of capsaicin result in neurogenic inflammation, but repeated low dose administration of capsaicin leads to decreased pain sensation due to neuronal desensitization,93 an effect that could be viewed as beneficial or detrimental depending on context and condition. Models that can recapitulate neurogenic inflammation, pain, or itch sensation potential could contribute to understanding and de-risking these factors early in drug development (Fig. 2, Table 3).

The condition of the skin related to the disease being treated should also be considered for interrogating neuronal effects. Decreased barrier function in some dermal conditions such as atopic dermatitis can result in increased exposure at the sensory nerve endings.94 Changes in sensory neuronal branching and innervation related to inflammation could also result in increased sensation of pain or itch. Association of eosinophils with human sensory nerves in vitro and in atopic human and mouse skin in vivo results in increased sensory nerve branching and itch sensation.35,36

Skin MPS with a subcutaneous layer containing adipocytes and vascular networks are not commercially available. To fully take advantage of fluidics in an in vitro skin model, endothelial cells are regarded as a necessary feature and likely to significantly contribute not only to the overall health of the tissue, but also analysis of pharmacokinetic parameters (Fig. 2). Furthermore, a vascularized skin MPS would be ideal to couple with circulating immune cells (or immune tissue compartment) and lymphatic system, and enable study of immune cell recruitment to damaged and/or diseased tissue (Fig. 2).

Improved barrier function and fluidics

In vitro assessment of skin absorption has largely relied on methods outlined in OECD TG 428, and utilize excised human skin mounted in a static or flow-through diffusion cell.74 Measurement of drug concentrations within the skin, receptor fluid, and wash of skin surface can enable determination of the percentage of drug absorbed, and in some cases a permeability constant. However, the variability of donor skin can influence the analysis and prove problematic when assessing compounds over time as is typical in drug development. Additionally, due to limited availability of donor skin, comparisons of pharmacokinetic parameters in healthy versus diseased skin proves particularly challenging.

Reconstructed skin models such as Episkin™, Episkin™ full thickness (FT) model, and EpiDerm™ have been investigated for expression of drug metabolizing enzymes. Luu-The, et al., compared expression of phase 1 and phase 2 enzymes in the Episkin™ and Episkin™ FT model to human epidermis, dermis, and whole (FT) skin, and shown similarity in expression of 61 enzymes.30 In particular, phase 1 enzymes such as cytochrome P450 (CYP) isozymes exhibit low expression, with CYP4B1 and CYP26B1 being the highest expressed. In contrast to the low expression of phase 1 enzymes, phase 2 enzymes including catechol O-methyltransferase, glutathione transferase P1, sulfotransferase 2B1b, and N-acetyl transferase 5 are highly expressed.30 Hu, et al., similarly evaluated drug metabolizing enzyme expression in the EpiDerm™ model with comparison to whole skin noting considerable congruence, and also demonstrated capability of CYP induction with 3-methylcholanthrene.31 Thus, the metabolic capabilities of reconstructed human skin models has been clearly demonstrated; however, there remains a challenge in assessing drug absorption and systemic availability in these models.

Reconstructed epidermis models have been investigated for utility in predicting skin permeability, and validation activities commenced.41,95,96 Testosterone and caffeine have been employed as tool compounds to compare permeability of Episkin™, EpiDerm™, and SkinEthic™ with excised human and pig skin. Reconstructed epidermis models generally exhibit much higher absorption compared to human epidermis across compounds with a range of lipophilicity.97 Although the reconstructed epidermis models have demonstrated ability to rank compounds by permeability similarly to human skin, the observed higher absorption and variability with respect to lipophilicity/hydrophilicity makes for tentative pharmacokinetic prediction, particularly with respect to systemic exposure. It is likely that improvement in barrier function will impart more accurate predictivity with respect to drug absorption and pharmacokinetic analysis of systemic exposure following topical application. Factors such as inclusion of a dermal compartment (FT), adipose compartment, vascularization, and modulation of lipid synthesis are expected to influence barrier capabilities (Fig. 2). However, regulatory acceptance of a drug absorption assay in skin MPS is lacking with significant opportunity for further development. The ability to establish drug absorption parameters with systemic exposure estimates is highly desired in skin MPS to enable clinical exposure prediction and risk assessment earlier in the drug development process. Lastly, the effect of fluidics on pharmacokinetic analyses in skin MPS remains unclear with much work to be done.

Pigmentation

Melanocytes have been successfully incorporated into the EpiDerm™ epidermis model (MelanoDerm), and appear responsive to modulation of pigmentation with melanogenic agents such as α-melanocyte stimulating hormone (α-MSH) and basic fibroblast growth factor (FGF-2). Others have demonstrated an effect of fibroblasts on basic melanocyte biology and response to stimulatory agents, suggesting the importance of a full thickness model in recapitulating and investigating pigmentation disorders.98 While pigmentation is not a critical feature in standard skin safety assessment during drug development, availability of a full thickness, pigmented model would be useful for pharmacological research in disorders such as vitiligo (Table 3).

Tissue longevity

Current skin MPS used in the assessment of irritation and corrosion require use of the models within a fixed, short period of time from the point of tissue maturation. The ability of complex skin models to sustain viability and function with respect to drug metabolizing enzymes for greater than 2 weeks is highly desired. For some skin toxicities the manifestation of the adverse event is fairly rapid, or an early signal predictive of eventual outcome is known. However, some reactions in the skin such as scabbing or immune reactions as can be observed in vivo can take time to manifest or require repeated dose administration, thus necessitating extended culture duration. With respect to disease modelling, culture longevity becomes even more important depending upon the time needed for disease induction. For example, if it takes 7 days to induce a disease phenotype and an additional 7–14 days to probe the pharmacological effects of a test article, tissues ultimately must be viable for a minimum 21 days from achievement of maturation.

Immune competence

In vitro skin models to date have comprised engineered skin constructs or ex vivo punch biopsies, both of which may lack immune cells completely or certainly lack access to a recruitable immune component, and have limited culture duration. This limits investigation of any adverse reactions to the very initiation of a response – if immune cells or antigen-presenting cells are present in the static system. For simpler skin toxicity questions based upon corrosivity and irritation, skin MPS lacking immune cells have proven quite useful. However, to mimic inflammatory skin conditions and adverse immune reactions – such as contact hypersensitivity and even skin reactions mediated by systemic exposure, a competent immune component is essential (Fig. 2, Table 3). This has been a key target for development of MPS for some time, with some limited examples of success.

Of course, whilst these models have moved some way to inclusion of immune cell populations with a static immune component they are limited to investigation of innate and early events only, which might be of use for some adverse issues but does not help with more complex mechanistic understanding. Key considerations for existing in vitro assessments have been that the use of primary cells from donors results in high variability, and the requirement to match cells with respect to HLA. Thus, many of the models developed to date instead use cell lines to mitigate variability. However, whilst this might be appropriate for more simple models, use of cell lines in skin MPS may not be the optimal representation of the human organ. The matching of cell types in skin MPS could be severely restrictive, thus going forward the utility of iPSCs (induced pluripotent stem cells) could become a key feature of these more complex models as this allows donor matching across the cell types used.

Another driver for the inclusion and validation of an immune component in skin MPS is the advantage of enabling investigation of inflammatory skin diseases. There is strong pharmaceutical interest in developing novel anti-inflammatory drugs for skin diseases like atopic dermatitis and psoriasis. Atopic dermatitis, the most common inflammatory skin disease, is a chronic condition characterized by eczematous skin lesions and intense itch (pruritus). The immune reaction is mediated by Th2 cells inducing the inflammatory response through the key Th2 cytokines IL-4, IL-5, and IL-13.99 Th2 cytokines have been shown to lead to downregulation of epidermal barrier proteins like filaggrin, involucrin and loricrin.100–102 In addition, a defect in terminal differentiation of keratinocytes, changes in the skin microbiome, and altered skin lipid composition have been associated with skin barrier dysfunction in atopic dermatitis.100 The defective epidermal barrier leads to the induction of Langerhans cells, antigen-presenting dendritic cells of the skin, residing in the epidermis.103 In atopic dermatitis, Langerhans cells take up antigen from the exposed epidermis and then migrate to the draining lymph nodes where they initiate T-cell-mediated immune responses.104

Like atopic dermatitis, psoriasis is an inflammatory skin disease that presents as scaling plaques caused by epidermal hyperplasia, infiltration of large numbers of T-cells and dendritic cells, altered keratinocyte differentiation, and skin barrier dysfunction.105,106 While atopic dermatitis is mediated by a Th2 response, in psoriasis Th1 cytokines, like IFN-γ and TNF-α, are prevalent.

To replicate an inflammatory skin disease in skin MPS, several approaches are in the realm of possibility. The most straightforward approach is the cultivation of cells from patients with an inflammatory skin disease. The psoriasis model produced by MatTek Corporation uses normal human epidermal keratinocytes in combination with fibroblasts harvested from psoriatic lesions. This model maintains an inflamed phenotype as demonstrated by increased basal cell proliferation, expression of psoriasis-specific markers and inflammatory cytokines, and mimics the morphology of psoriatic skin. These characteristics allow its use in pre-clinical efficacy testing. No commercial skin MPS utilizing cells from atopic dermatitis patients is currently available; however, it would be a worthwhile addition to the skin MPS market. Other approaches to induce inflammatory conditions in skin MPS are the addition of pro-inflammatory cytokines in the media and co-culture of inflammatory cells (e.g., dendritic cells). For instance, culture of skin MPS in the presence of Th2 cytokines affects keratinocyte differentiation and skin barrier dysfunction comparable to lesional atopic dermatitis skin (MatTek, SID 2016 Annual Meeting). The development and validation for context of use of a skin MPS featuring the inclusion of immune cells, like Langerhans cells, is highly desirable. As well, the interaction of keratinocytes with the immune system is essential in the development of inflammatory skin diseases. A model incorporating this immune aspect of the disease would improve the ability to study therapeutic candidates targeting inflammatory cells in vitro.

The ultimate goal with respect to immune competence in any model is the ability to recruit the wider immune system, and thus model the factors released into the systemic circulation and potentially model a more adaptive-type response. Here, the advantages of a fluidics system come into play (Fig. 2). Adding flow into and out of a skin MPS would not only permit dosing that might mimic a systemic delivery, alongside pharmacokinetic investigations into uptake and metabolism, but also introduce investigation into the soluble factors secreted by the skin tissue cells and released into the circulation. The ability to add a recruitable cellular component, in a physiologically-relevant medium, to the fluidics system would further enhance the potential of skin MPS (Fig. 2). For example, if a white blood cell mixture (i.e., lymphocytes, monocytes, granulocytes, etc.) of proportion akin to human circulation were added to culture fluidics, it is possible that migration of cells to sites of damage, as observed in vivo, could be replicated. The homing and activation of these cell types within the skin MPS could be investigated to understand if the system can replicate a typical inflammatory reaction, for example. Further enhancements such as coupling the skin MPS to a lymph node or spleen MPS via a fluidics system, would enable modelling of the multisite nature of immune activation (Fig. 2). This might be of particular importance for investigating potential hypersensitivity reactions (from sensitization to systemic activation of the immune system) where antigen presenting cells detecting and engulfing foreign proteins or haptens in the skin can potentially exit the tissue and migrate to lymphoid tissues with subsequent priming of T-cells. The release of proliferating T-cells into the system could also be monitored, and potentially tracked back to the site of insult.

This type of development would require extremely complex systems. It could be envisaged that in fact two fluidics systems would be required: a substitute for lymph collection and transport (lymphatics) would be required to deliver cells to the lymphoid tissue system, and then a vascular component would be needed to deliver cells back to the target tissue. In such a design, the potential to truly mimic the immune responses in not only the skin, but any target organ, becomes attainable.

Histopathological analytical methods

Validation of a system comprising human and potentially animal cells organized in a tissue-like structure will naturally need comparative investigations with the tissues they are designed to model. It is envisaged that new skin MPS might be compared structurally to human punch biopsies for comparative physiology and arrangements of cells within the tissue. Histopathology should also be compared for known chemical effects resulting in corrosivity, irritation, and sensitization to validate the changes in cell health and state that occur in vivo. With models where fluidics and cell migration can be incorporated, histopathologic investigation will be essential to understand the properties of the model and whether cells entering or exiting the tissues locate in a physiologically relevant way. This could entail standard H&E with codified histopathologic scoring, but also comprise multiplex IHC and transcript detailing by in situ technologies to identify cell types and activation states. Furthermore, spatial transcriptomics could help understand cell activities in a morphological state.107

Biologics in skin toxicity

Of course, complex models developed for assessing skin toxicity with small molecules and traditional pharmaceuticals will be of similar relevance for understanding peptide and large molecule therapeutics if the target is applicable. The irritancy and corrosivity testing of protein therapeutics is of less concern, given these therapeutics are usually dosed systemically and are not of a corrosive nature. Contact sensitization is also of minimal concern given the parenteral dose route; however, skin models may be of interest for assessing inherent hypersensitivity potential even if dose routes differ. These models might be of particular relevance for understanding subcutaneous or intradermal dosing of therapeutics encased in novel delivery systems such as lipid nanoparticles, dendrimers, or polymers. Advanced skin MPS may be able to predict adverse dose site reactions and pro-inflammatory potential, and model mitigation measures to enable smoother transition to a clinical setting. For considerations of biologic testing in MPS the reader is referred to Application of 3D Cultures and Microphysiologic Systems in Biopharmaceuticals Research and Development (Peterson, et al.) in the current series.

Conclusions

Our goal has been to provide deeper insight to the use of skin MPS in the drug development setting, and highlight key areas of interest for further development from the perspective of the pharmaceutical scientist. While skin MPS are the most advanced models with respect to regulatory acceptance for specific contexts of use, there are many more opportunities of which to take advantage. Though current skin MPS regulatory guidelines are specifically geared toward safety assessment, such guidelines are not necessary for biology and pharmacology research. However, it is generally accepted that advancements in skin MPS capabilities in the context of safety assessment will naturally spill over to the benefit of other areas in the drug development process. Model advancements continue to be made at a rapid pace, and the compilation of accepted guidelines expands with each year. Therefore, we anticipate this progress will translate to wider acceptance across the pharmaceutical industry, yet encourage continued investigation to further advance the possibilities with skin MPS.

Conflicts of interest

There are no conflicts of interest to declare. All authors are/were current employees of their respective institutions with substantive contributions to the manuscript in the form of conceptualization, analysis, and writing (original drafting, reviewing, and editing).

Acknowledgements

This manuscript is part of a manuscript series led by the IQ MPS Affiliate. The IQ MPS Affiliate launched in June 2018 within the International Consortium for Innovation and Quality in Pharmaceutical Development (also known as the IQ Consortium). The IQ Consortium is a not-for-profit organization of pharmaceutical and biotechnology companies with a mission of advancing science and technology to augment the capability of member companies to develop transformational solutions that benefit patients, regulators, and the broader research and development community (https://iqconsortium.org). Of those 39 companies, 23 are members of the IQ MPS Affiliate, an initiative of the IQ Consortium established to provide a venue for appropriate cross-pharma collaboration and data sharing to facilitate industry implementation and qualification of MPS models. The purpose of the IQ MPS Affiliate-led manuscript series is to specifically outline the contemporary pharmaceutical industry perspectives and considerations for developing, evaluating, and characterizing MPS models to support drug discovery and development. The authors acknowledge and thank members of the IQ Consortium and the IQ MPS Affiliate for review of, and recommendations to the manuscript.

Notes and references

  1. E. Fuchs, Curr. Top. Dev. Biol., 2016, 116, 357–374 CrossRef PubMed.
  2. A. Nystrom and L. Bruckner-Tuderman, Semin. Cell Dev. Biol., 2019, 89, 136–146 CrossRef.
  3. C. K. Hsu, H. H. Lin, H. I. Harn, M. W. Hughes, M. J. Tang and C. C. Yang, J. Dermatol. Sci., 2018, 90, 232–240 CrossRef PubMed.
  4. D. M. Danilenko, G. D. Phillips and D. Diaz, Toxicol. Pathol., 2016, 44, 555–563 CrossRef CAS PubMed.
  5. E. Abd, S. A. Yousef, M. N. Pastore, K. Telaprolu, Y. H. Mohammed, S. Namjoshi, J. E. Grice and M. S. Roberts, J. Clin. Pharmacol., 2016, 8, 163–176 CAS.
  6. S. H. Mathes, H. Ruffner and U. Graf-Hausner, Adv. Drug Delivery Rev., 2014, 69-70, 81–102 CrossRef CAS PubMed.
  7. V. Planz, C. M. Lehr and M. Windbergs, J. Controlled Release, 2016, 242, 89–104 CrossRef CAS.
  8. N. E. Vrana, P. Lavalle, M. R. Dokmeci, F. Dehghani, A. M. Ghaemmaghami and A. Khademhosseini, Tissue Eng., Part B, 2013, 19, 529–543 CrossRef CAS.
  9. OECD, Test No. 404: Acute Dermal Irritation/Corrosion, 2015.
  10. OECD, Explanatory background document to the OECD draft Test Guideline on in vitro skin irritation testing, 2010 Search PubMed.
  11. UN, Globally Harmonized System of Classification and Labelling of Chemicals (GHS), 2019.
  12. F. Cottrez, E. Boitel, C. Auriault and H. Groux, in Alternatives for Dermal Toxicity Testing, ed. C. Eskes, E. Van Vliet and H. I. Maibach, Springer Cham, 2017, pp. 361–375 Search PubMed.
  13. R. H. Rice and T. M. Mauro, in Casarett and Doull's Toxicology: The Basic Science of Poisons, ed. C. D. Klaassen, McGraw Hill Education, 8th edn, 2013, ch. 19, pp. 839–859 Search PubMed.
  14. B. B. Hayes, E. Patrick and H. Maibach, in Hayes' Principles and Methods of Toxicology, ed. A. W. Hayes and C. L. Kruger, CRC Press, 6th edn, 2014, ch. 27, pp. 1345–1384 Search PubMed.
  15. OECD, The Adverse Outcome Pathway for Skin Sensitisation Initiated by Covalent Binding to Proteins, 2014.
  16. L. Peuvrel and B. Dreno, Am. J. Clin. Dermatol., 2014, 15, 425–444 CrossRef.
  17. D. I. G. Cubero, B. M. Z. Abdalla, J. Schoueri, F. I. Lopes, K. C. Turke, J. Guzman, A. Del Giglio, C. Filho, V. Salzano and D. G. Fabra, Drugs Context, 2018, 7, 212516 Search PubMed.
  18. Y. Balagula, S. T. Rosen and M. E. Lacouture, J. Am. Acad. Dermatol., 2011, 65, 624–635 CrossRef.
  19. A. P. Chen, A. Setser, M. J. Anadkat, J. Cotliar, E. A. Olsen, B. C. Garden and M. E. Lacouture, J. Am. Acad. Dermatol., 2012, 67, 1025–1039 CrossRef PubMed.
  20. A. C. Rosen, Y. Balagula, D. W. Raisch, V. Garg, B. Nardone, N. Larsen, J. Sorrell, D. P. West, M. J. Anadkat and M. E. Lacouture, Anti-Cancer Drugs, 2014, 25, 225–234 CrossRef CAS.
  21. D. J. Naisbitt, M. Pirmohamed and B. K. Park, Expert Opin. Drug Saf., 2007, 6, 109–124 CrossRef CAS.
  22. J. M. Michot, C. Bigenwald, S. Champiat, M. Collins, F. Carbonnel, S. Postel-Vinay, A. Berdelou, A. Varga, R. Bahleda, A. Hollebecque, C. Massard, A. Fuerea, V. Ribrag, A. Gazzah, J. P. Armand, N. Amellal, E. Angevin, N. Noel, C. Boutros, C. Mateus, C. Robert, J. C. Soria, A. Marabelle and O. Lambotte, Eur. J. Cancer, 2016, 54, 139–148 CrossRef CAS.
  23. S. H. Tella and M. S. Rendell, Expert Opin. Drug Saf., 2015, 14, 127–140 CrossRef CAS.
  24. P. Hoffmann, P. Bentley, P. Sahota, H. Schoenfeld, L. Martin, L. Longo, R. Spaet, P. Moulin, S. Pantano, V. Dubost, D. Lapadula, B. Burkey, V. Kaushik, W. Zhou, M. Hayes, N. Flavahan, S. D. Chibout and S. Busch, Toxicol. Pathol., 2014, 42, 684–695 CrossRef CAS.
  25. I. Garcia-Diez, M. Ivars-Lleo, D. Lopez-Aventin, N. Ishii, T. Hashimoto, P. Iranzo, R. M. Pujol, A. Espana and J. E. Herrero-Gonzalez, Int. J. Dermatol., 2018, 57, 810–816 CrossRef CAS.
  26. C. Chijiwa, S. Takeoka, M. Kamata, M. Tateishi, S. Fukaya, K. Hayashi, A. Fukuyasu, T. Tanaka, T. Ishikawa, T. Ohnishi, S. Watanabe and Y. Tada, J. Dermatol., 2018, 45, 596–599 CrossRef CAS.
  27. X. W. Chen, Z. X. He, Z. W. Zhou, T. Yang, X. Zhang, Y. X. Yang, W. Duan and S. F. Zhou, Clin. Exp. Pharmacol. Physiol., 2015, 42, 999–1024 CrossRef CAS.
  28. X. W. Chen, Z. X. He, Z. W. Zhou, T. Yang, X. Zhang, Y. X. Yang, W. Duan and S. F. Zhou, Clin. Exp. Pharmacol. Physiol., 2015, 42, 1225–1238 CrossRef CAS PubMed.
  29. R. Andukuri, A. Drincic and M. Rendell, Diabetes, Metab. Syndr. Obes.: Targets Ther., 2009, 2, 117–126 CrossRef CAS.
  30. V. Luu-The, D. Duche, C. Ferraris, J. R. Meunier, J. Leclaire and F. Labrie, J. Steroid Biochem. Mol. Biol., 2009, 116, 178–186 CrossRef CAS.
  31. T. Hu, Z. S. Khambatta, P. J. Hayden, J. Bolmarcich, R. L. Binder, M. K. Robinson, G. J. Carr, J. P. Tiesman, B. B. Jarrold, R. Osborne, T. D. Reichling, S. T. Nemeth and M. J. Aardema, Toxicol. In Vitro, 2010, 24, 1450–1463 CrossRef CAS.
  32. S. Gattu and H. I. Maibach, Skin Pharmacol. Physiol., 2010, 23, 171–176 CrossRef CAS PubMed.
  33. D. W. Roberts, Regul. Toxicol. Pharmacol., 2018, 98, 155–160 CrossRef CAS.
  34. M. J. Aardema, B. B. Barnett, G. C. Mun, E. L. Dahl, R. D. Curren, N. J. Hewitt and S. Pfuhler, Mutat. Res., 2013, 750, 40–49 CAS.
  35. J. J. Lee, C. A. Protheroe, H. Luo, S. I. Ochkur, G. D. Scott, K. R. Zellner, R. J. Raish, M. V. Dahl, M. L. Vega, O. Conley, R. M. Condjella, J. A. Kloeber, J. L. Neely, Y. S. Patel, P. Maizer, A. Mazzolini, A. D. Fryer, N. W. Jacoby, D. B. Jacoby and N. A. Lee, J. Allergy Clin. Immunol., 2015, 135, 477–487 CrossRef CAS.
  36. E. L. Foster, E. L. Simpson, L. J. Fredrikson, J. J. Lee, N. A. Lee, A. D. Fryer and D. B. Jacoby, PLoS One, 2011, 6, e22029 CrossRef CAS.
  37. E. L. Dahl, R. Curren, B. C. Barnett, Z. Khambatta, K. Reisinger, G. Ouedraogo, B. Faquet, A. C. Ginestet, G. Mun, N. J. Hewitt, G. Carr, S. Pfuhler and M. J. Aardema, Mutat. Res., 2011, 720, 42–52 CAS.
  38. OECD, Guidance Document on an Integrated Approach on Testing and Assessment (IATA) for Skin Corrosion and Irritation, 2017.
  39. NTP, Alternative Methods Accepted by US Agencies, https://ntp.niehs.nih.gov/pubhealth/evalatm/accept-methods/index.html, (accessed 08/31/2019, 2019).
  40. F. S. Alikhan and H. Maibach, Cutaneous Ocul. Toxicol., 2011, 30, 175–186 CrossRef.
  41. M. Schafer-Korting, U. Bock, W. Diembeck, H. J. Dusing, A. Gamer, E. Haltner-Ukomadu, C. Hoffmann, M. Kaca, H. Kamp, S. Kersen, M. Kietzmann, H. C. Korting, H. U. Krachter, C. M. Lehr, M. Liebsch, A. Mehling, C. Muller-Goymann, F. Netzlaff, F. Niedorf, M. K. Rubbelke, U. Schafer, E. Schmidt, S. Schreiber, H. Spielmann, A. Vuia and M. Weimer, ATLA, Altern. Lab. Anim., 2008, 36, 161–187 CrossRef PubMed.
  42. OECD, Test No. 431: In vitro skin corrosion: reconstructed human epidermis (RHE) test method, 2019.
  43. OECD, Test No. 439: In Vitro Skin Irritation: Reconstructed Human Epidermis Test Method, 2019.
  44. OECD, Test No. 430: In Vitro Skin Corrosion: Transcutaneous Electrical Resistance Test Method (TER), 2015.
  45. OECD, Test No. 435: In Vitro Membrane Barrier Test Method for Skin Corrosion, 2006.
  46. OECD, Test No. 442C: In Chemico Skin Sensitisation, 2019.
  47. G. F. Gerberick, J. A. Troutman, L. M. Foertsch and P. S. Kern, in Alternatives for Dermal Toxicity Testing, ed. C. Eskes, E. Van Vliet and H. I. Maibach, Springer Cham, 2017, pp. 225–233 Search PubMed.
  48. OECD, Test No. 442D: In Vitro Skin Sensitisation, 2018.
  49. L. J. van den Broek, L. Bergers, C. M. A. Reijnders and S. Gibbs, Stem Cell Rev. Rep., 2017, 13, 418–429 CrossRef CAS PubMed.
  50. A. Natsh, in Alternatives for Dermal Toxicity Testing, ed. C. Eskes, E. Van Vliet and H. I. Maibach, Springer Cham, 2017, pp. 235–248 Search PubMed.
  51. E. Corsini, V. Galbiati, M. Mitjans, C. L. Galli and M. Marinovich, Toxicol. In Vitro, 2013, 27, 1127–1134 CrossRef CAS PubMed.
  52. E. Corsini and V. Galbiati, in Alternatives for Dermal Toxicity Testing, ed. C. Eskes, E. Van Vliet and H. I. Maibach, Springer Cham, 2017, pp. 263–272 Search PubMed.
  53. T. Ramirez, A. Mehling, S. N. Kolle, C. J. Wruck, W. Teubner, T. Eltze, A. Aumann, D. Urbisch, B. van Ravenzwaay and R. Landsiedel, Toxicol. In Vitro, 2014, 28, 1482–1497 CrossRef CAS PubMed.
  54. T. Ramirez, A. Mehling and R. Landsiedel, in Alternatives for Dermal Toxicity Testing, ed. C. Eskes, E. Van Vliet and H. I. Maibach, Springer Cham, 2017, pp. 249–262 Search PubMed.
  55. F. Cottrez, E. Boitel, J. C. Ourlin, J. L. Peiffer, I. Fabre, I. S. Henaoui, B. Mari, A. Vallauri, A. Paquet, P. Barbry, C. Auriault, P. Aeby and H. Groux, Toxicol. In Vitro, 2016, 32, 248–260 CrossRef CAS PubMed.
  56. D. Keller, C. Bauch and P. Patel, in Alternatives for Dermal Toxicity Testing, ed. C. Eskes, E. Van Vliet and H. I. Maibach, Spinger Cham, 2017, pp. 377–391 Search PubMed.
  57. OECD, Test No. 442E: In Vitro Skin Sensitisation, 2018.
  58. H. Sakaguchi and T. Ashikaga, in Alternatives for Dermal Toxicity Testing, ed. C. Eskes, E. Van Vliet and H. I. Maibach, Springer Cham, 2017, pp. 289–309 Search PubMed.
  59. C. Piroird, J. M. Ovigne, F. Rousset, S. Martinozzi-Teissier, C. Gomes, J. Cotovio and N. Alepee, Toxicol. In Vitro, 2015, 29, 901–916 CrossRef CAS PubMed.
  60. N. Alepee, C. Piroird and L. Nardelli, in Alternatives for Dermal Toxicity Testing, ed. C. Eskes, E. Van Vliet and H. I. Maibach, Spinger Cham, 2017, pp. 311–330 Search PubMed.
  61. H. Johansson, F. Rydnert, J. Kuhnl, A. Schepky, C. Borrebaeck and M. Lindstedt, Toxicol. Sci., 2014, 139, 362–370 CrossRef CAS PubMed.
  62. M. Lindstedt, K. S. Zeller, H. Johansson and C. Borrebaeck, in Alternatives for Dermal Toxicity Testing, ed. C. Eskes, E. Van Vliet and H. I. Maibach, Springer Cham, 2017, pp. 393–403 Search PubMed.
  63. N. Lambrechts, I. Nelissen, V. Van Tendeloo, H. Witters, R. Van Den Heuvel, J. Hooyberghs and G. Schoeters, Toxicol. Lett., 2011, 203, 106–110 CrossRef CAS PubMed.
  64. N. Lambrechts, G. Schoeters, R. Van Den Heuvel, H. Witters, I. Nelissen and J. Hooyberghs, in Alternatives for Dermal Toxicity Testing, ed. C. Eskes, E. Van Vliet and H. I. Maibach, Springer Cham, 2017, pp. 347–359 Search PubMed.
  65. A. Richter, S. S. Schmucker, P. R. Esser, V. Traska, V. Weber, L. Dietz, H. J. Thierse, D. Pennino, A. Cavani and S. F. Martin, Toxicol. In Vitro, 2013, 27, 1180–1185 CrossRef CAS PubMed.
  66. P. R. Esser and S. F. Martin, in Alternatives for Dermal Toxicity Testing, ed. C. Eskes, E. Van Vliet and H. I. Maibach, Springer Cham, 2017, pp. 449–454 Search PubMed.
  67. S. Pfuhler and K. Reisinger, in Alternatives for Dermal Toxicity Testing, ed. C. Eskes, E. Van Vliet and H. I. Maibach, Springer Cham, 2017, pp. 513–525 Search PubMed.
  68. OECD, Test No. 487: In Vitro Mammalian Cell Micronucleus Test, 2016.
  69. K. Reisinger and S. Pfuhler, in Alternatives for Dermal Toxicity Testing, ed. C. Eskes, E. Van Vliet and H. I. Maibach, Springer Cham, 2017, pp. 527–539 Search PubMed.
  70. A. A. Reus, K. Reisinger, T. R. Downs, G. J. Carr, A. Zeller, R. Corvi, C. A. Krul and S. Pfuhler, Mutagenesis, 2013, 28, 709–720 CrossRef CAS PubMed.
  71. OECD, Test No. 489: In Vivo Mammalian Alkaline Comet Assay, 2016.
  72. OECD, Test No. 432: In Vitro 3T3 NRU Phototoxicity Test, 2019.
  73. H. Kandarova and M. Liebsch, in Alternatives for Dermal Toxicity Testing, ed. C. Eskes, E. van Vliet and H. I. Maibach, Springer Cham, 2017, pp. 483–503 Search PubMed.
  74. OECD, Test No. 428: Skin Absorption: In Vitro Method, 2004.
  75. A. F. Monteiro, M. Rato and C. Martins, Clin. Dermatol., 2016, 34, 571–581 CrossRef PubMed.
  76. ICH, S10 Guideline: Photosafety Evaluation of Pharmaceuticals, ICH, 2013.
  77. H. Kojima, K. Hosoi and S. Onoue, in Alternatives for Dermal Toxicity Testing, ed. C. Eskes, E. Van Vliet and H. I. Maibach, Springer Cham, 2017, pp. 477–482 Search PubMed.
  78. ICH, S2(R1) Guideline: Guidance on Genotoxicity Testing and Data Interpretation for Pharmaceuticals Intended for Human Use, ICH, 2011.
  79. S. Roy, R. Kulkarni, N. J. Hewitt and M. J. Aardema, Mutat. Res., Genet. Toxicol. Environ. Mutagen., 2016, 805, 25–37 CrossRef CAS PubMed.
  80. N. Flamand, L. Marrot, J. P. Belaidi, L. Bourouf, E. Dourille, M. Feltes and J. R. Meunier, Mutat. Res., 2006, 606, 39–51 CAS.
  81. OECD, Test No. 429: Skin Sensitisation, 2010.
  82. OECD, Guidance Document on the Reporting of Defined Approaches to be Used Within Integrated Approaches to Testing and Assessment, 2017 Search PubMed.
  83. I. Kimber, R. J. Dearman, D. A. Basketter, C. A. Ryan and G. F. Gerberick, Contact Dermatitis, 2002, 47, 315–328 CrossRef CAS.
  84. E. Corsini, A. B. Engin, M. Neagu, V. Galbiati, D. Nikitovic, G. Tzanakakis and A. M. Tsatsakis, Arch. Toxicol., 2018, 92, 3031–3050 CrossRef CAS.
  85. D. Corti, V. Galbiati, N. Gatti, M. Marinovich, C. L. Galli and E. Corsini, Toxicol. In Vitro, 2015, 29, 1339–1349 CrossRef CAS PubMed.
  86. S. S. Ahmed, X. N. Wang, M. Fielding, A. Kerry, I. Dickinson, R. Munuswamy, I. Kimber and A. M. Dickinson, J. Appl. Toxicol., 2016, 36, 669–684 CrossRef CAS PubMed.
  87. S. S. Ahmed, J. Whritenour, M. M. Ahmed, L. Bibby, L. Darby, X. N. Wang, J. Watson and A. M. Dickinson, Toxicol. Appl. Pharmacol., 2019, 369, 39–48 CrossRef CAS PubMed.
  88. W. J. Pichler, Toxicology, 2005, 209, 95–100 CrossRef CAS PubMed.
  89. P. T. Illing, J. P. Vivian, N. L. Dudek, L. Kostenko, Z. Chen, M. Bharadwaj, J. J. Miles, L. Kjer-Nielsen, S. Gras, N. A. Williamson, S. R. Burrows, A. W. Purcell, J. Rossjohn and J. McCluskey, Nature, 2012, 486, 554–558 CrossRef CAS PubMed.
  90. C. Eskes, T. Cole, S. Hoffmann, A. Worth, A. Cockshott, I. Gerner and V. Zuang, ATLA, Altern. Lab. Anim., 2007, 35, 603–619 CrossRef CAS PubMed.
  91. S. Casati, P. Aeby, I. Kimber, G. Maxwell, J. M. Ovigne, E. Roggen, C. Rovida, L. Tosti and D. Basketter, ATLA, Altern. Lab. Anim., 2009, 37, 305–312 CrossRef CAS PubMed.
  92. X. Kodji, A. A. Aubdool and S. D. Brain, Curr. Res. Transl. Med., 2016, 64, 195–201 CrossRef CAS PubMed.
  93. G. Jancso, E. Kiraly and A. Jancso-Gabor, Nature, 1977, 270, 741–743 CrossRef CAS.
  94. T. Mori, K. Ishida, S. Mukumoto, Y. Yamada, G. Imokawa, K. Kabashima, M. Kobayashi, T. Bito, M. Nakamura, K. Ogasawara and Y. Tokura, Br. J. Dermatol., 2010, 162, 83–90 CrossRef CAS PubMed.
  95. S. Schreiber, A. Mahmoud, A. Vuia, M. K. Rubbelke, E. Schmidt, M. Schaller, H. Kandarova, A. Haberland, U. F. Schafer, U. Bock, H. C. Korting, M. Liebsch and M. Schafer-Korting, Toxicol. In Vitro, 2005, 19, 813–822 CrossRef CAS PubMed.
  96. M. Schafer-Korting and S. Schreiber, in Dermal Absorption and Toxicity Assessment, ed. M. S. Roberts and K. A. Walters, CRC Press, 2nd edn, 2007 DOI:10.3109/9780849375927, pp. 141–160.
  97. S. Kuchler, K. Struver and W. Friess, Expert Opin. Drug Metab. Toxicol., 2013, 9, 1255–1263 CrossRef PubMed.
  98. C. Duval, C. Cohen, C. Chagnoleau, V. Flouret, E. Bourreau and F. Bernerd, PLoS One, 2014, 9, e114182 CrossRef PubMed.
  99. H. A. Sampson, J. R. Soc. Med., 1997, 90(Suppl 30), 2–8 CrossRef CAS PubMed.
  100. B. E. Kim and D. Y. M. Leung, Allergy, Asthma Immunol. Res., 2018, 10, 207–215 CrossRef CAS.
  101. M. D. Howell, B. E. Kim, P. Gao, A. V. Grant, M. Boguniewicz, A. Debenedetto, L. Schneider, L. A. Beck, K. C. Barnes and D. Y. Leung, J. Allergy Clin. Immunol., 2007, 120, 150–155 CrossRef CAS.
  102. B. E. Kim, D. Y. Leung, M. Boguniewicz and M. D. Howell, Clin. Immunol., 2008, 126, 332–337 CrossRef CAS PubMed.
  103. J. Deckers, H. Hammad and E. Hoste, Front. Immunol., 2018, 9, 93 CrossRef.
  104. R. E. Callard and J. I. Harper, Trends Immunol., 2007, 28, 294–298 CrossRef CAS.
  105. C. E. Griffiths, P. van de Kerkhof and M. Czarnecka-Operacz, Dermatol. Ther., 2017, 7, 31–41 CrossRef.
  106. S. Coimbra and A. Santos-Silva, Ann. Transl. Med., 2015, 3, 76 Search PubMed.
  107. P. L. Stahl, F. Salmen, S. Vickovic, A. Lundmark, J. F. Navarro, J. Magnusson, S. Giacomello, M. Asp, J. O. Westholm, M. Huss, A. Mollbrink, S. Linnarsson, S. Codeluppi, A. Borg, F. Ponten, P. I. Costea, P. Sahlen, J. Mulder, O. Bergmann, J. Lundeberg and J. Frisen, Science, 2016, 353, 78–82 CrossRef CAS PubMed.

Footnote

Current address: Translational Research Institute for Space Health, Baylor College of Medicine, Houston, TX, USA.

This journal is © The Royal Society of Chemistry 2020