Abstract
Human skin cells (epidermal keratinocytes and dermal fibroblasts) in monolayer culture and human skin in organ culture were exposed to agents that are known to produce irritation (redness, dryness, edema and scaly crusts) when applied topically to skin. Among the agents used were three well accepted contact irritants (i.e., all-trans retinoic acid [RA], sodium lauryl sulfate [SLS] and benzalkonium chloride) as well as the corrosive organic mercury compound, aminophenyl mercuric acetate (APMA), and 5 contact sensitizers (oxazolone, nickel sulfate, eugenol, isoeugenol and ethylene glycol dimethacrylate [EGDM]). As a group, the contact irritants (including the corrosive mercuric compound) were cytotoxic for keratinocytes and fibroblasts and suppressed growth at lower concentrations than the contact sensitizers. The contact irritants also produced histological changes (hyperplasia, incomplete keratinization, loss of the granular layer, acantholysis and necrosis) in organ-cultured skin at dose levels at which the contact sensitizers appeared to be inert. Finally, the profile of secreted molecules from organ-cultured skin was different in the presence of contact irritants versus contact sensitizers. Taken together, these data suggest that the use of organ-cultured skin in conjunction with cells derived from the skin in monolayer culture may provide an initial approach to screening agents for deleterious changes in skin.
Introduction
Agents developed for formulation as cosmetics/cosmeceuticals need to be evaluated for detrimental effects on the skin. Topical pharmaceuticals also need to be evaluated for the same effects as does the large group of agents in the “household products” category. The goal of safety-assessment is to define the theoretical benefit of a compound compared to what damage, if any, the compound may cause. Safety assessment is currently based on clinical evidence of skin damage using animal models developed decades ago (Draize et al., 1944). However, several alternatives to animal testing have recently been or are currently being evaluated. In part, these efforts to eliminate or reduce animal use in testing are based on ethical considerations and/or marketing strategies and cost. In addition, however, it is recognized that results obtained in animal tests are not always predictive of the response of human skin (Nixon et al., 1975; York et al., 1996).
Among the alternatives being evaluated are three-dimensional organotypic cultures consisting of epidermal keratinocytes on a collagenous matrix (with or without imbedded dermal fibroblasts) and whole skin organ cultures from rodents, pigs or, in some cases, discarded human skin (Botham et al., 1998; Boelsma et al., 2000; Ponec et al., 2002; Jacobs et al., 2004; Welss et al., 2004). Multiple endpoints have been examined with the various culture systems in an effort to find functional and/or biochemical changes that are predictive of skin damage. To date, however, validated assays are only available to confirm corrosivity. No assay has proven capable of achieving all of the requirements needed to replace testing in animals (Botham, 2004). In the present report, we describe results using a combination of skin cells (human epidermal keratinocytes and human dermal fibroblasts) in monolayer culture and adult human skin in organ culture to identify and characterize changes in structure/function following exposure to known corrosive, irritant and sensitizing agents.
Materials and Methods
Chemical Agents
The organic mercury compound, aminophenyl mercuric acetate (APMA), was used as a corrosive agent. Irritants included the detergent, sodium lauryl sulfate (SLS), the biologically-active retinoid, all-trans retinoic acid (RA) and benzalkonium chloride. Five agents that are known to be human skin contact-sensitizers were also examined. These included oxazolone (classified as a strong sensitizer), nickel sulfate (strong), isoeugenol (strong), eugenol (moderate) and ethylene glycol dimethacrylate (EGDM; moderate) (Basketter et al., 2005).
Organ Culture
Replicate 2 mm punch biopsies were obtained from hip skin of volunteers less than 70 years of age. The participation of human subjects in this study was approved by the University of Michigan Institutional Review Board and all subjects provided written informed consent prior to their inclusion in the study. The punch biopsies were incubated in wells of a 24 well dish (1 tissue piece per 250 μl of culture medium). Culture medium consisted of Keratinocyte Basal Medium (KBM) (Cambrex Biologicals, Walkersville, MD). KBM is a modification of MCDB-153 medium. For our purposes, it was supplemented with calcium chloride to a final Ca2+ concentration of 1.4 mM. One or 2 tissue pieces were left as controls while the others were treated with the irritants or allergens over a range of concentrations. Fresh culture medium was provided at 2-day intervals.
Organ culture-conditioned medium collected on day 2 was saved for assessment of growth factors, cytokines and type I procollagen levels as described below. At the end of the incubation period (day 8), organ-cultured tissue was fixed in 10% buffered formalin and embedded in paraffin. Five-μm-thick sections were cut and stained with hematoxylin and eosin. Representative sections of each biopsy were selected for histological evaluation and examined. In certain experiments, tissue was fixed at day 2 and stained with methyl green pyronine (Jacobs et al., 2002). Human skin in organ culture has been extensively used in the past. The protocol used here was virtually identical to that described in our past reports (Varani et al., 1993, 1994).
Cells in Monolayer Culture
Human epidermal keratinocytes and human dermal fibroblasts were used in these studies. Keratinocytes were isolated from neonatal foreskin tissue obtained at circumcision (Varani et al., 1994). They were maintained in Keratinocyte Growth Medium (KGM). KGM consistes of the same basal medium as KBM but is supplemented with epidermal growth factor (EGF), insulin and pituitary extract. Cell growth was at 37°C in a humidified atmosphere containing 5% CO2 and 95% air. Keratinocytes were subcultured using trypsin/EDTA as required, and used at passage 3 to 5. An immortalized keratinocyte cell line (Hacat) (Boukamp et al., 1988) was also utilized as the epithelial cell component in some of the studies. These cells were handled exactly as low-passage keratinocytes and used interchangeably with those cells.
Fibroblasts were isolated from the same tissue as used for the isolation of keratinocytes (Varani et al., 1994). Fibroblasts were maintained in Dulbecco’s MEM supplemented with non-essential amino acids and 10% fetal bovine serum (DMEM-FBS). Cell growth was at 37°C in a humidified atmosphere containing 5% CO2 and 95% air. Fibroblasts were subcultured using trypsin/EDTA as required, and used at passage 3 to 5.
Cytotoxicity and Growth Inhibition Assays
Keratinocytes and fibroblasts were added to wells of a 24-well dish at 8 × 104 cells per well using their respective growth media (KGM for keratinocytes and DMEM-FBS for fibroblasts). One day later, after the cells had attached and spread, the culture medium was removed and the cells washed 1 time in serum-free medium. At this point, 0.5 ml of KGM was added to keratinocyte cultures. The same medium was used for fibroblasts with the exception that it was supplemented with extracellular Ca2+ to bring the final concentration to 1.4 mM (Varani et al., 1994). The agents to be examined were added to the wells of both cell types over a wide range of concentrations. Cells were incubated for 4 hours and then harvested with trypsin/EDTA and counted. Following this, the cells were replated in wells of a 24-well dish using DMEM-FBS for both cell types. The number of cells that reattached and spread was determined four hours later. Cells capable of successfully “replating” were assumed to be viable and functional (Varani et al., 1992).
To assess growth inhibition, keratinocytes and fibroblasts were added to wells of a 24-well dish and treated exactly as above. Cells were incubated for 48 hours and then harvested with trypsin/EDTA and enumerated.
Assessment of Secreted Molecules
Organ culture fluids collected at day 2 were analyzed for levels of amphiregulin by enzyme-linked immunosorbant assay (ELISA) using a commercially available kit (R&D Systems; Minneapolis, MN) as described previously (Rittie et al., 2006). The same organ culture fluids were assayed for type I procollagen levels using a commercially available ELISA kit (Pan Vera Corp., Madison, WS) (Varani et al., 2004). Levels of interleukin-6 (IL-6) and interleukin-8 (IL-8) were quantified as part of a multiplex assay (Luminex antibody bead kit; BioSource International, Inc.; Camarillo, CA). Results from the multiplex assay were confirmed using individual ELISA kits for the 2 cytokines (Upstate Biologicals; Charlottesville, VA and Biosource International, Inc.).
Statistical Analysis
Differences between groups in experiments with multiple groups were analyzed for statistical significance by ANOVA followed by paired-group comparisons. Differences from control values at the p < 0.05 level were considered significant. Correlation coefficients were determined in experiments designed to assess relationships between levels of different metabolites in a given sample.
Results
Keratinocyte and Fibroblast Cytotoxicity and Growth Inhibition: Comparison of Chemicals with Corrosive, Irritant and Allergenic Potential
In the first series of experiments, a panel of 3 contact irritants and 5 contact sensitizers was examined for cytotoxicity with human epidermal keratinocytes and human dermal fibroblasts in monolayer culture. A corrosive organic mercury compound (APMA) was examined in the same assay. The 3 irritants demonstrated minimal (or no) cytotoxic activity at concentrations up to 1 μM while the organic mercury compound was cytotoxic to virtually 100% of the keratinocytes and fibroblasts at 1 μM and killing of both cell types (30–40% killed) was seen at concentrations as low as 0.1 μM (Figure 1). In contrast, there was no toxicity with any of the 5 contact sensitizers at concentration below 100 μM (Figure 1).
Following these initial studies, the 3 irritants and 5 allergens were examined for effects on keratinocyte and fibroblast proliferation in a 2-day growth assay (Figure 2 and Table 1). All of the agents caused dose-dependent inhibition of growth with both cell types. The major difference between the 2 groups of agents was in the concentrations needed for growth inhibition. Growth inhibition with the 3 contact irritants occurred over the range of 0.5–10 μM while significantly higher concentrations (100–500 μM) with the contact sensitizers were required to produce growth inhibition. Even at the highest concentration, growth inhibition was minimal with 3 of the sensitizers.
Comparison of Contact Irritants and Contact Sensitizers for Effects on Histological Features of Human Skin in Organ Culture
Biopsies of sun-protected skin were obtained from healthy volunteers and maintained in organ culture for 8 days in the presence or absence of the same irritants and sensitizers used in the monolayer culture studies. With RA there was a significant hyperplasia observed at 1 μM. At 2 μM, the epidermis remained viable and was still thickened as compared to the control. However, abnormalities in the upper epidermis including incomplete keratinization and loss of the granular layer were observed. The abnormalities in differentiation were more pronounced at 10 μM, where, in addition to the other change, overt acantholysis was observed in virtually every tissue. In some areas, the upper epidermis separated from the viable layers below (Figure 3).
Histological changes were also observed after treatment for 8 days with SLS. At 1 μM SLS, the tissue was indistinguishable from control. Epidermal hyperplasia in some of the cultures was observed at 5 μM, but the degree of epidermal thickening was not as substantial as seen with RA. At 10 μM, the lower portion of the epidermis remained viable and intact, but at this concentration, areas of necrosis in the upper epidermis were evident. The histological abnormalities seen at high SLS concentrations were distinctly different from those seen in RA-treated skin. There was no acantholysis or separation of the upper nonviable layers from the cells below. Rather, upper epidermal necrosis was the primary finding (Figure 3).
Figure 3 also demonstrates histological changes in organ-cultured skin treated with benzalkonium chloride. There were no consistent changes in tissue exposed to 1 μM benzalkonium chloride. At 5–10 μM, epidermal necrosis was observed. With benzalkonium chloride treatment, necrosis was observed throughout the entire tissue, including the basal layer. This was in contrast to what was seen with RA and SLS, where the major effects were on the upper (differentiating) layers of the epidermis. Finally, organ-cultured skin was exposed to 1 μM APMA for 8 days. Complete necrosis throughout the epidermis and dermis was observed in every specimen (not shown). Histological changes induced by APMA and benzalkonium chloride were indistinguishable, except for the rapidity with which they developed and the dosage. Skin organ cultures were also treated with increasing concentrations of the five contact sensitizers. Concentrations as high as 100–250 μM did not cause necrosis with any of the agents. In some tissues, epidermal thinning was observed at the highest concentration (500 μM). However, even at 500 μM, most of the tissue samples were indistinguishable from controls (not shown).
In an effort to detect cytotoxic changes in the tissue before overt necrosis was apparent, we stained sections of irritant-treated, organ-cultured skin with methyl green-pyronine after incubation for 2 days. With irritant concentrations that eventually produced overt cytotoxicity (i.e., 10 μM RA, SLS and benzalkonium chloride), there was a loss of RNA (decreased methyl green-pyronine staining) evident at day 2. However, decreased staining (relative to freshly biopsied skin) was also seen in sections of control tissue. The loss of RNA in the control tissue was never as great as seen following exposure to the irritants, but differences between treated and control skin were subtle (data not shown).
Amphiregulin and Type I Procollagen Production in Organ-Cultured Skin Exposed to Contact Irritants and Contact Sensitizers
Organ culture fluids obtained 2 days after treatment with each of the 3 irritants and 5 sensitizers were analyzed for amphiregulin. The amphiregulin level was significantly higher in culture fluid from RA-treated skin than control skin (Figure 4). Amphiregulin in organ culture fluid from SLS-treated tissue was also elevated, but the increase with SLS, although statistically significant, was not as dramatic as seen with RA. In contrast, neither benzalkonium chloride nor any of the 5 sensitizers had a significant effect on amphiregulin production (Figure 4). Thus, the increase in amphiregulin correlated with epidermal hyperplasia rather than with necrosis, per se.
The same organ culture fluids were assessed for type I procollagen. As can be seen from Figure 5, the level of type I procollagen was elevated in culture fluid from RA-treated and benzalkonium chloride-treated organ cultures. There was also a slight increase in type I procollagen in the organ culture fluid from SLS-treated skin. None of the contact sensitizers had a measurable effect on levels of this protein.
Elaboration of IL-6 and IL-8 in Organ-Cultured Skin Exposed to Contact Irritants and Contact Sensitizers
Levels of 2 pro-inflammatory cytokines, IL-6 and IL-8, were assessed in organ culture fluids collected on day 2 from tissue exposed to each of the 3 contact irritants and to 2 of the contact sensitizers (nickel sulfate and oxazolone). Tissue from 8 individuals was assessed with each agent. As seen in Table 2, SLS failed to raise the level of either IL-6 or IL-8 over control values. RA, benzalkonium chloride, oxazolone and nickel sulfate, on the other hand, induced a measurable increase in the levels of both cytokines. Variability increased along with mean values for the various treatments. This reflects the fact that with tissue from some individuals, levels of the 2 cytokines increased greatly in response to treatment while with tissue from other individuals, there was little or no increase. For example, 7 of 8 individuals’ skin samples demonstrated a response to RA with increased production of the 2 cytokines. With benzalkonium chloride and nickel sulfate, there was a response in tissue from 5 of 8 and 4 of 8 individuals, respectively. With oxazolone, only 3 of 8 individuals demonstrated a response and only IL-6 was significantly elevated.
Additional experiments were carried out to further explore the variability observed in the cytokine measurements. For this we obtained replicate (n = 12) biopsies from each of 4 subjects and assessed IL-6 and IL-8 in culture fluid from each biopsy after incubation for 2 days under control conditions. With both cytokines, the variability among replicate biopsies from a given subject as well as the average variability among the different subjects was much lower than the variability observed with the various treatments. For example, the standard deviations as a percentage of the mean values for IL-6 was 47%, 42%, 46% and 51% for the 4 subjects, respectively (n = 12 biopsies per subject). For IL-8, the standard deviations were 50%, 49%, 33% and 51%, respectively. The inter-subject variability for the 4 subjects (standard deviations as a percent of the mean values) was 47% for IL-6 and 49% for IL-8. Additionally, when levels of IL-6 and IL-8 among replicate biopsies in each subject were examined for degree of correlation, a highly significant correlation was achieved in all cases (not shown).
Discussion
Redness, dryness, edema, and scaly crusts are common manifestations of chemically induced skin irritation. Such changes can be brought about through multiple mechanisms (Goh and Soh, 1984; Corsini and Galli, 1998; Welss et al., 2004). At one extreme, widespread killing of the major cellular elements of the skin may occur upon a single exposure to a highly toxic substance. This is referred to as acute skin irritation or as a corrosive effect. More common is irritation resulting from a single or repetitive exposure to mild irritants. These reactions may have a component of cytotoxic cell injury, but the pathophysiology may also reflect non-lethal changes in cell physiology. Agents that cause epidermal hyperplasia and hyperkeratosis or other manifestations of altered differentiation often have an associated irritant dermatitis (Welss et al., 2004). In addition to these reactions that result from direct effects on keratinocytes and fibroblasts are reactions that engage the adaptive immune system as part of the pathophysiology (Goh and Soh, 1984; York et al., 1996; Coquette et al., 2003). Given this multiplicity of mechanisms, and the possibility that certain agents may be allergenic as well as directly irritating (Basketter et al., 2004), it is not surprising that no single test is capable of providing a reliable indicator of risk.
To date, in vivo assays have been used to assess the potential hazard of skin-contact agents. Patch-testing in humans has the potential to be the most reliable since the multiple read-outs (visual observation, color change, itchiness, transepidermal water loss, etc.) that are used as indicators constitute actual skin irritation. Most agents that come into contact with human skin, including the irritants and sensitizers examined here, have been examined in this manner. There appears to be a good relationship between concentrations that give positive results in human patch test assays with concentrations that elicited a response in the culture models examined here. For example, RA and benzalkonium chloride are positive in most individuals tested in the 0.02–0.2% range while SLS gives a positive response in most individuals tested at 1–2% (Wilhelm et al., 1989; Berardesca et al., 1990; Fisher et al., 1991; Geier et al., 2003; Basketter et al., 2004; Jacobs et al., 2004; Molander et al., 2004; Marrakchi and Maibach 2006).
With some of the contact sensitizers, concentrations of up to 5% are routinely used (Hinson et al., 1991; Anderson et al., 2001). Patch testing has also been used in an effort to distinguish irritant and allergic responses. In general, irritant responses arise more rapidly and often subside whereas allergic responses tend to arise more slowly and persist (Jacobs et al., 2004; Welss et al., 2004). Another difference is that irritant responses tend to be dose-dependent. In contrast, once an allergic response has developed, lower concentrations often elicit the same response. Overall, the patch test provides reliable information on irritancy but, for obvious reasons, it is not useful for early-stage testing. Further, it does not lend itself to mechanistic studies.
In vivo testing for skin irritation in animals has been used for years as a way to assess contact hazards (Draize et al., 1944). While this approach provides reliable information at an early stage in a compound’s development, issues include a lack of complete correlation between findings in animals and those in humans, overall expense and ethical considerations. Also, it is difficult to distinguish contact irritant from contact allergenic responses. On the other hand, the Draize approach, like the patch test in humans, relies heavily on clinical features of irritation. Other in vivo assays including the local lymph node assay (LLNA) in its various forms, the guinea pig maximization test and rodent “ear-swelling” assays have proven useful for detecting irritation and, in some cases at least, for predicting potential for contact sensitization (Basketter and Scholes 1992; Van Och et al., 2001). Even these assays are not completely specific, however, since all of the irritants used here have been shown to give positive responses in one or more of these predictive assays (Basketter et al., 2004, 2006). Furthermore, as with other in vivo assays, expense and ethical considerations must be taken into account.
Given the limitation of in vivo testing (especially for hazard assessment at an early stage of a product’s development), it is not surprising that a number of in vitro approaches have been or are being developed as replacements. Epidermal-equivalent and skin-equivalent (epidermis and dermis) cultures have been utilized in conjunction with a variety of biomarkers as a way to assess risk (Ponec and Kempenaar, 1995; Botham et al., 1998; Boelsma et al., 2000; Ponec et al., 2002). While some success has been achieved to date, development of alternative in vitro approaches is still ongoing. The studies described here suggest that a combination of human keratinocytes and fibroblasts in monolayer culture and intact human skin in organ culture may provide information that can be used as an initial screen for skin irritation potential.
Using a series of agents that act directly on the major cellular elements of the skin (i.e., contact irritants) and a panel of contact sensitizers, we observed a variety of responses that appear to distinguish the 2 groups. As a group, the irritants inhibited keratinocyte and fibroblast growth (and were overtly cytotoxic) at concentrations far lower than those needed for growth inhibition with the contact sensitizers. Concomitantly, the same contact irritants produced changes in organ-cultured skin (epidermal hyperplasia/epidermal thickening, abnormal differentiation and necrosis) while the contact sensitizers were virtually without effect. Likewise, we observed changes in the profile of secreted products with the irritants that were not observed with the contact sensitizers.
Although the contact sensitizers were much less stimulatory than the irritants in a variety of assays and required higher concentrations for suppression of keratinocyte and fibroblast proliferation, organ culture fluids from tissue treated with either contact irritants or contact sensitizers elaborated two pro-inflammatory cytokines (IL-6 and IL-8). The 2 cytokines could be detected in control skin organ cultures, and levels of both were increased in the presence of the two contact-sensitizers examined. These data are consistent with previous findings from epidermal-equivalent or skin-equivalent cultures (Ponec and Kempenaar, 1995; Boxman et al., 1996; Corsini and Galli, 1998; Bernhofer et al., 1999a, 1999b; Coquette et al., 2003). While in our small sample of test reagents we found no definitive differences between the contact irritants and contact sensitizers (as groups), the finding that the contact sensitizers did, in fact, raise the levels of the 2 cytokines is significant in that it shows that these agents are not inert in the ex vivo assays utilized here.
Increased cytokine elaboration in organ culture is of interest in that it suggests local production. While it is impossible to have a complete inflammatory response in an isolated tissue piece, these data suggest that at least some of the initiating events following exposure to irritants of one type or another originate in the skin itself. Elaboration of pro-inflammatory cytokines in the epidermis could promote local inflammation (and skin irritation) through the numerous events occurring downstream of the cytokines (Mantovani, 1999). While it is not likely that assessing only two cytokines will be sufficient to establish irritancy or to completely define mechanisms, assessing a panel of cytokines may provide a clear indication of a chemical’s ability to cause irritation, at least in a subset of individuals.
As with all predictive assays, there are issues that need to be addressed with the human skin approach described here. Adult human skin may not be readily available. While we used skin from adult volunteers, others have used discarded skin following surgeries (Lehe et al., 2003). Furthermore, we have demonstrated that neonatal human foreskin can be maintained in organ culture (Varani et al., 1994). Tissue obtained at circumcision would, presumably, be more readily available. Another possibility is rodent skin. While skin from rats and mice can be maintained in organ culture (Varani et al., 2002), the structure of rodent skin is very different from that of humans, which could make interpretation of results difficult. Skin from other animals is also a possibility. Pig skin is very similar to human skin and pig skin has been successfully maintained in organ culture (Ghazizadeh et al., 1998). While the use of either rodent skin or skin from another species does not fit the definition of “animal-free testing,” it would greatly reduce the number of animals used.
Another issue is variability. With regard to the cytokine measurements in particular, but with all of the endpoints to some extent, the variability among tissue samples was higher in the presence of the various irritants and sensitizers than it was in controls (either intra-subject or between subjects). The high variability, thus, appears to reflect a differential response to the eliciting agent among individuals rather than differences from biopsy to biopsy within a given donor, or inherent (basal) differences among donors. The high variability, while making statistical analysis difficult (and larger sample size necessary), may be important in its own right. Similar variability is observed in patch-testing (Wilhelm et al., 1989; Fisher et al., 1991; Marrakchi and Maibach, 2006) and in other in vivo risk-assessment assays. It may, ultimately, be possible to use responses seen in organ culture to estimate the percentage of individuals who are likely to have an adverse event to a given agent or as a predictor of who will or will not develop a response to a given agent.
In summary, a wide variety of chemical agents produce irritation when applied topically to the skin. A similar cluster of clinical features can be produced through multiple diverse pathways. Given this multiplicity of mechanisms, it is not surprising that no single profile of cellular, histological or biochemical parameters characterizes all contact irritants or contact sensitizers. Nevertheless, the data presented here suggest that a combination of keratinocytes and fibroblasts in monolayer culture and intact skin in organ culture may provide information that can be used as an initial screen for irritation. The data presented here demonstrate that agents that act directly on the major cellular elements of the skin produce a variety of effects that are not seen with contact sensitizers. While the contact sensitizers used in the present studies were inert in a variety of bioassays, these agents were capable of stimulating production of pro-inflammatory cytokines. While these data need to be verified with a larger panel of agents, our findings suggest that it will, ultimately, be possible to use a combination of cell culture and organ culture as an initial screen for irritant potential, and to help distinguish among potential mechanisms.