Cell Biology and Toxicology. 2000; 16: 391^400. # 2001 Kluwer Academic Publishers. Printed in the Netherlands Development of a highly sensitive in vitro phototoxicity assay using the SkinEthicTM reconstructed human epidermis F-X. Bernard1, C. Barrault1, A. Deguercy1, B. De Wever2 and M. Rosdy2 1 BIOalternatives, Genc°ay; 2SkinEthic Laboratories, Nice, France Received 15 May 2000; accepted 29 December 2000 Keywords: phototoxicity, in vitro, ultraviolet A, reconstituted human epidermis, SkinEthicTM Abstract The reconstituted human epidermis model SkinEthicTM was used to evaluate the phototoxicity of topically applied chemicals. For comparison with published data, we ¢rst tested a library of 13 nonphototoxic (NPT) and phototoxic (PT) compounds, applied onto SkinEthicTM reconstituted human epidermal tissues, in a protocol as close as possible to the one described by Liebsch using another skin tissue model. The results showed that, under these nonoptimized conditions, the SkinEthicTM model was already able to fully discriminate between known NPT and PT compounds. Furthermore, these epidermal tissues being highly resistant to UVA irradiation, it was possible to increase irradiation by (at least) 3-fold without decrease in tissue viability. In such conditions, the phototoxicity assay is much more sensitive, so that the model is expected to be of great interest for the detection not only of strong but also of weak phototoxic compounds. Abbreviations: DMSO, dimethyl sulfoxide; 6-MC, 6-methylcoumarin; 5-MOP, 5-methoxypsoralen; 8-MOP, 8-methoxypsoralen; MTT, 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide; NPT, nonphototoxic; PABA, p-aminobenzoic acid; PT, phototoxic Introduction In recent years, considerable e¡ort has been dire cted toward the development and evaluation of alternative test methods for predicting chemical-induced skin irritancy and toxicity. It has been shown in a joint EU/ COLIPA validation project (Liebsch et al., 1995; Api, 1997), that the phototoxic potential of chemicals can be correctly predicted using a specially designed cytotoxicity assay, the mouse ¢broblast 3T3^neutral red uptake phototoxicity test (3T3-NRU assay). However, this cell culture monolayer assay has a major disadvantage since it cannot be used for the evaluation of nonhydrosoluble compounds or solid (e.g., creams) or liquid (e.g., perfumes) formulations. The use of reconstituted skin models was an attractive solution to this problem, and the Episkin1 (Saduc, France), Skin2 TM (Advanced Tissue Sciences, USA), and EpiDermTM (MatTek, USA) models have been shown to be able to predict photoirritancy (Edwards et al., 1994; Rouget et al., 1994; Liebsch et al., 1997). In the present study, we used reconstituted human epidermis tissues provided by SkinEthic Laboratories, Nice, France, for the evaluation of phototoxicity in vitro. Being developed in chemically de¢ned conditions, 392 this human epidermal model has been described as a true physiologically relevant tool for cutaneous in vitro absorption, metabolism, irritancy, and e¤cacy testing of skin care as well as of pharmaceutical products (Doucet et al., 1994, 1996; Botham et al., 1998; Rosdy et al., 1998; Coquette et al., 1999; De Brugerolle de Fraissinette et al., 1999; Gysler et al., 1999a,b; Laugier et al., 2000). Materials and methods Reconstituted human epidermis SkinEthicTM Surgical samples of adult normal human skin were incubated overnight at 48C in dispase grade II; then the epidermis was separated from the dermis using a ¢ne forceps. Epidermal cells (mainly keratinocytes) were detached by a 10 min treatment with 0.05% trypsin^0.02% EDTA at 378C. Primary cultures were initiated at a density of 3000 cells/cm2 in cell culture £asks (Falcon, France) in modi¢ed MCDB 153 medium (prepared at SkinEthic Laboratories) containing 0.15 mmol/L CaCl2, 0.1 ng/ml human recombinant epidermal growth factor, 5 mg/ ml human recombinant insulin, and 0.4 mg/ml hydrocortisone. Bovine pituitary extract (BPE) 75 mg/ml was added for the ¢rst 2 days. All chemicals used in the medium were from Sigma Aldrich, France; dispase, trypsinEDTA, soybean trypsin inhibitor, and growth factors were purchased from Boehringer Mannheim, France. Before con£uence was reached, cells were harvested and subcultured in the same medium without BPE (i.e., chemically de¢ned medium). Second-passage keratinocytes were inoculated at a density of 56105 cells/cm2 on air-lifted Millipore polycarbonate culture inserts. Reconstituted epidermis was obtained by growing the keratinocytes for 16^17 days at the surface of high-calcium de¢ned medium, as described previously (Rosdy and Clauss, 1990). Phototoxicity experiments were performed with tissues grown for 17 days, in 24-well plates, on 300 ml maintenance medium per well, at 378C under a 95% air^5% CO2 atmosphere. The same medium was used during UV exposure. Test substances and administration protocols l -Histidine (Sigma H8000), sodium lauryl sulfate (Sigma L4390), penicillin G (Sigma PENNA), rose bengal sodium salt (Sigma R3877), and neutral red (Sigma N4638) were dissolved in ultrapure sterile water. Chlorpromazine hydrochloride (Sigma C8138), promethazine hydrochloride (Sigma P4651) and 8-methoxypsoralen (8-MOP, Sigma M3501) were dissolved in dimethylsulfoxide (DMSO, Sigma D8418) and subsequently diluted in water; the maximal ¢nal DMSO concentration was 0.1%. Benzophenone-3 (Sigma H2758), p-aminobenzoic acid (PABA, Sigma A9878), tetracycline base (Sigma T3258), 5-methoxyp s o ral e n ( 5 -MOP, Sig m a M 1 6 4 8 ), 6 methylcoumarin (6-MC, Aldrich M3,620-3) and bithionol (Sigma T9881) were dissolved or suspended directly in sesame oil (Cooper A28326/4). Chemicals dissolved in water were directly applied by gentle pipetting of 50 ml on the stratum corneum side of the tissues (without using a pad). Chemicals in oil were applied using a pad (8 mm diameter disks, Promedica). Each pad was soaked by pipetting 20 ml of test chemical and each pad was then put onto the corresponding tissue. Irradiation source and irradiation conditions The irradiation source was a UV-sun simulator lamp SOL 500 equipped with a H1 UVA ¢lter (Dr. Ho«nle, Martinsried). All irradiations, for both calibrations and ¢nal experiments, were performed through the lid of a 24-well plate; irradiance was measured with a VLX 3W UVA 393 radiometer (Cofralab). The lamp was adjusted to achieve an equally distributed irradiance of 1.7 mW/cm2 (resulting dose of 1 J/cm2 per 10 min exposure time) in the area of exposure. The temperature was controlled and maintained at 37(+1)8C by continuous ventilation during the irradiation step. Experimental procedure Upon receipt, epidermal tissues were incubated at 378C, under 5% CO2, on de¢ned maintenance medium (SkinEthic Laboratories). The next day, compounds were applied to the epidermal surface, and the tissues were incubated overnight (18 h) on 300 ml fresh medium (in 24-well plates). The treatments were performed in duplicate (two tissues per experimental condition), except for the UVA sensitivity range determinations, which were performed in triplicate. After the incubation period, the pads were removed and the tissues were irradiated at 1.7 mW/cm2 (most of the experiments used irradiation doses of 6 J/cm2, so the irradiation times were generally 1 h). The nonirradiated tissues were placed in the dark, at room temperature. After washing of the stratum corneum with 46500 ml isotonic phosphate bu¡er, the tissues were incubated for 18 h on 300 ml assay medium (in 24-well plates). On the last day of the experiment, a 24-well plate with 300 ml MTT medium (0.5 mg/ml solution) per well was prepared and the inserts were incubated for 3 h, at 378C, 5% CO2. After aspiration of MTT medium and washing, the tissues were cut out of the inserts and transferred into individual microtubes, and formazan was extracted in 1.25 ml isopropanol^HCl, for 3 h under agitation in the dark. Three 200 ml aliquots of each tube were distributed in a 96-well plate and the viability of tissues was estimated by recording the absorbance at 540 nm with a ThermoMax microplate reader controlled by SoftMax software (Molecular Devices). Calculations and prediction model For each concentration of test compound, the mean optical density (OD) of the two tissues treated was expressed as relative percentage viability of the untreated controls tissues. Identical calculations were performed for the irradiated and nonirradiated samples. Concentrations of test compounds that decreased tissue viability by 50% (IC50) were estimated by linear regression and, when possible, the ratio IC50(^UV)/IC50(+UV) was calculated and considered as a possible parameter for a prediction model. Alternatively, the viabilities of irradiated and nonirradiated conditions were compared for each concentration of a test compound: a 30% decrease in the viability of tissues treated with a test chemical concentration followed by irradiation as compared to treated but nonirradiated control tissues was considered as a possible criterion for predicting phototoxicity. Results UVA sensitivity of tissues The sensitivity of the SkinEthicTM tissues to UVA was ¢rst evaluated by irradiation with increasing doses (up to 21 J/cm2 UVA), using the Ho«nle lamp, in two di¡erent experiments (Figure 1). No signi¢cant decrease in tissue viability was observed, even at the highest UVA dose applied. Chlorpromazine: interexperimental variability In three consecutive experiments with the protocol using 6 J/cm2, the phototoxic (PT) reference compound chlorpromazine gave highly reproducible phototoxicity results with IC50s(^UV) ranging between 0.21 and 0.26 mg/ml and IC 50s (+UV) ranging between 0.026 and 0.027 mg/ml (Table 1). 394 are listed in Table 4. The IC 5 0 (^UV)/ IC 50(+UV) ratios were only calculable for penicillin G, lauryl sulfate, and PABA; the ratios were below 1.5 in all cases. Furthermore, for the ¢ve in vivo NPT compounds assayed, no test concentration in the +UVA part of the experiment revealed a decrease in viability exceeding 30% when compared to identical concentrations of the ^UVA part of the experiment (Table 4). A series of in vivo phototoxic (PT) chemicals was also tested in the 6 J/cm2 protocol (Tables 3 and 4). The 6 J/cm 2 UVA irradiation drastically enhanced the toxicity of promethazine, 5-MOP, and tetracycline (Table 3) with IC50(^UV)/IC50(+UV) ratios for these compounds above 25 (Table 4). Four concentrations of promethazine and tetracyclin, and (all the) ¢ve concentrations of 5MOP reduced viability by more than 30% in irradiation conditions when compared with UV-protected conditions (Table 4). Owing to its high coloration, the dye neutral red gave artifactual responses at concentrations up to 0.11 mg/ml. However, the phototoxic potential of neutral red was undoubtly established in the assay since a dramatic UVA- In vivo PT versus non-PT compounds in the 6 J/cm2 protocol Five in vivo nonphototoxic (NPT) compounds were tested with the same protocol. For all concentrations of these NPT compounds, the decrease in viability following the 6 J/cm2 exposure did not exceed 20% except for penicillin G at 100 mg/ml, for which UVA exposure decreased the viability by 22% (Table 2). The IC50 calculated for these compounds Figure 1. UVA sensitivity of SkinEthic tissues (two individual experiments, n = 3 tissues). Viability was evaluated by the MTT test. Table 1. Chlorpromazine as reference PT compound Concentration (mg/ml) Viability (%) öööööööööööööööööööööööö Exp. 1: Exp. 2: Exp. 3: batch a batch b batch c Non-irradiated 1 0.33 0.11 0.037 0.012 2 38 75 80 82 1 7 82 85 94 0 5 102 109 111 UVA-irradiated (6 J/cm2) 1 0.33 0.11 0.037 0.012 2 2 8 28 82 2 1 5 26 88 1 1 1 21 86 Assay in three di¡erent experiments (Exp. 1^3) with three di¡erent batches of tissues (batch a^c). Viability was evaluated by the MTT test 395 Table 2. Test of in vivo NPT chemicals (irradiation 6 J/cm2) Compound/ solvent Penicillin G/H2O Histidine/H2O Viability (%) ööööööööö Concentration Non(mg/ml) irradiated Irradiated 100 33.3 11.1 3.7 1.2 38 94 95 102 92 16 75 93 100 95 33 11 3.7 1.2 0.4 85 93 98 95 104 Table 3. Test of in vivo PT chemicals (irradiation 6 J/cm2) Compound/ solvent Viability (%) ööööööööö Concentration Non(mg/ml) irradiated Irradiated Promethazine/H2O 1 0.33 0.11 0.037 0.012 39 89 114 113 117 2 2 11 31 109 90 84 92 92 86 Neutral red/H2O 1 0.33 0.11 0.037 0.012 130 b 195 b 125 b 111 110 79 b 21 b 6b 2 2 5-MOP/oil 3 1 0.33 0.11 0.037 101 114 94 99 103 24 23 33 34 68 Lauryl sulfate/H2O 5 1.5 0.5 0.16 0.05 0 0 62 87 97 0 0 65 86 93 Benzophenone-3/oil 10 3.3 1.1 0.37 0.12 114 109 104 104 106 107 106 99 100 95 Tetracycline/oil 3 1 0.33 0.11 0.037 101 99 105 101 101 33 49 33 51 87 PABA/oil 10 3.3 1.1 0.37 0.12 5 100 100 106 107 5 96 93 97 100 Rose bengala/H2O 3 1 0.33 0.11 0.037 98 101 101 102 100 2 66 84 91 91 Viability was evaluated by theMTT test. a The in vivo phototoxicity of RB is controversial; it seems not to cross the stratum corneum freely. b Overestimated value due to absorbance of test compound. Viability was evaluated by the MTT test. dependent reduction of cell viability was already observed in the concentration range 0.012^0.037 mg/ml. Furthermore, the decrease in cell viability following irradiation can clearly be uncoupled from artifactual response at the three higher doses tested. In this assay, the IC50 ratio for neutral red was estimated to be 440 and all the concentrations tested gave more than 30% decrease in viability due to UVA (Table 4). In the absence of UVA irradiation, rose bengal was not cytotoxic at the highest concentration tested (3 mg/ml). The 6 J/cm2 UVA irradiation led to partial or complete cell mortality at 1 mg/ml or 3 mg/ml, respectively. The IC50 ratio for rose bengal was 42 and the two highest concentrations assayed (1 mg/ml and 3 mg/ml) induced more than 30% decrease in viability due to UVA (Table 4). 396 Table 4. Application of two di¡erent parameters for phototoxicity prediction (irradiation 6 J/cm2): the IC50 ratio and the `30% rule' IC50 (mg/ml) öööööööööööööö `30% rule' for phototoxicity prediction ööööööööööööööööö Phototoxic Decrease in concentration viability due (mg/ml) to UVA (%) IC50(^UVA) IC50(+UVA) Ratio (+)/(^) 86 433 0.69 4100 67 61 433 0.73 4100 65 1.4 ^ 1.1 ^ 1 No No No No No ^ ^ ^ ^ ^ 0.85 0.03 28 0.037 0.11 0.33 1 73 90 98 95 0.5^1 (estimation) 50.012 441 0.012 0.037 0.11 0.33 1 98 98 95 89 39 5-MOP 43 0.07 441 0.037 0.11 0.33 1 34 66 65 80 Tetracycline 430 1.1 427 0.11 0.33 1 3 49 69 50 67 Rose bengal 43 1.5 42 1 3 35 98 Compound Penicillin G Histidine Lauryl sulfate Benzophenone-3 PABA Promethazine Neutral red Viability was evaluated by the MTT test. Comparison between 6 J/cm2 and 18 J/cm2 UVA irradiation As UVA irradiation of SkinEthicTM epidermis could be increased up to 21 J/cm2 without loss of tissue viability, the phototoxic properties of four di¡erent compounds were assayed in parallel, at 0, 6, and 18 J/cm2. The strong PT (reference) compounds chlorpromazine and 8MOP showed a dramatic increased phototoxicity when irradiated at 18 J/cm2 (Tables 5 and 6) compared to the 6 J/cm2 irradiation. The two weak PT compounds 6-MC and bithionol (dissolved in oil) exhibited no phototoxicity when irradiated with 6 mJ/cm2 UVA (Tables 5 and 6), but both were found to be phototoxic when irradiated with 18 J/cm2. For 6-MC, one concentration (10 mg/ml) was clearly phototoxic with the 18 J/cm 2 irradiation, leading to a ratio of 5.5 with 18 J/ cm 2 . Bithionol was phototoxic in two concentrations (0.3 and 0.1 mg/ml) following 18 J/cm2 irradiation, whereas it was not phototoxic after 6 J/cm2 irradiation. 397 Table 5. Cytotoxic e¡ects of chlorpromazine, 8-MOP and the two weakly PT compounds 6-methycoumarin (6-MC) and bithionol after irradiation by 0, 6 and 18 mJ/cm2 UVA Compound Concentration (mg/ml) Viability (%) öööööööööööööööööööööööööö Irradiated Irradiated 18 J/cm2 Nonirradiated 6 J/cm2 Chlorpromazine 0.33 0.11 0.037 0.012 0.004 1 91 96 98 100 2 1 18 76 92 0 2 1 31 40 6-MC 100 33 11 3.7 1 21 102 104 0 9 89 86 0 0 38 64 1.2 0.01 0.0033 0.0011 0.0003 101 89 83 88 97 98 50 80 97 96 83 29 26 38 97 0.33 0.11 0.037 0.011 0.004 62 114 111 108 104 66 105 103 102 99 0 57 84 85 98 8-MOP Bithionol Viability was evaluated by MTT test. Table 6. E¡ects of increasing UVA irradiation from 6 J/cm2 to 18 J/cm2 on the phototoxic response of chlorpromazine, 8-MOP and the two weak phototoxic compounds 6-methycoumarin (6-MC) and bithionol IC50(+UVA) Ratio (+)/(^) `30% rule' for phototoxicity prediction öööööööööö Phototoxic concentration (mg/ml) 0.02 50.04 10.5 452 0.11; 0.037 0.11; 0.037; 0.012; 0.004 0.01 0.001 41 410 0.01 0.01; 0.0033; 0.0011 22 8.2 1.2 3.2 No 10 40.3 0.11 ? 42.7 No 0.3; 0.1 IC50 (mg/ml) ööööööööööö Compound/solvent Chlorpromazine/water 8-MOP/water 6-MC/oil Bithionol/oil Irradiation IC50(^UVA) Nonirradiated 6 J/cm2 18 J/cm2 0.21 Nonirradiated 6 J/cm2 18 J/cm2 40.01 Nonirradiated 6 J/cm2 18 J/cm2 26 Nonirradiated 6 J/cm2 18 J/cm2 40.3 Viability was evaluated by the MTT test. 398 Discussion Development of in vitro methods as an alternative to animal testing encounters numerous di¤culties owing to the variety of compounds that have to be evaluated for their toxicity. Phototoxicity evaluation was thought to be important since many pharmaceutical or cosmetic ingredients have been shown to be activated by UV irradiation. The ¢rst in vitro methods employed to assess the phototoxic potential of compounds used conventional monolayer cell cultures. Because these methods were not adapted to test solid and liposoluble materials, reconstituted tissues (epidermis or skin equivalents) have been employed for the evaluation of the phototoxic potential of nonhydrosoluble components (Roguet et al., 1994; Liebsch et al., 1997). Topical application of the tested material showed the phototoxicity of most of the known highly phototoxic materials, but failed to detect the phototoxicity of weak phototoxics because the high dose of UVA needed to activate their phototoxic properties was itself toxic for the tissues used. The results described in this paper clearly show that the SkinEthicTM reconstituted human epidermis model can be irradiated with high doses of UVA without detectable loss in tissue viability. This di¡erence in sensitivity can be explained by the presence of a fully di¡erentiated stratum corneum in the SkinEthicTM reconstituted human epidermal tissues featuring a normal and functional permeability barrier both for xenobiotics and light irradiation (Rosdy and Clauss, 1990; Rosdy et al., 1996; Fartasch and Rosdy, 1996; Liebsch et al., 1997; Gysler et al., 1999a). We therefore compared data obtained with SkinEthicTM tissues with data reported using another reconstructed human epidermal model (EpiDermTM, MatTek), using a protocol as close as possible to the basic protocol described by Liebsch et al. (1997). Two criteria for phototoxicity prediction were used, the IC50(^UV)/IC50(+UV) ratio (used in the 3T3NRU protocol) and the prediction model previously described (Liebsch et al., 1997), where a chemical is predicted to have a phototoxic potential if one or more tested concentrations of the (+UVA) part of the experiment reveal a decrease in viability exceeding 30% when compared with identical concentrations tested without irradiation. The excellent reproducibility of the results obtained in our model, with this protocol (6 J/ cm2), was established through six experiments with the PT reference compound chlorpromazine. The discriminant power between NPT and PT compounds was demonstrated by the correct classi¢cation of three hydrosoluble (penicillin G, histidine, lauryl sulfate) and two nonhydrosoluble (benzophenone-3, PABA) compounds that are nonphototoxic in vivo, compounds, and three hydrosoluble (promethazine, neutral red, and rose bengal) and two nonhydrosoluble (5-MOP and tetracycline prepared in oil) compounds that are phototoxic in vivo. Tetracycline antibiotic derivatives are known to be phototoxic in vivo, but they were often misclassi¢ed as nonphototoxic in in vitro assays (Spielmann et al., 1994). The high amplitude of the phototoxic response obtained in the present study with tetracycline clearly points out the sensitivity of the SkinEthicTM epidermal model for phototoxicity prediction. Moreover, the high phototoxicity observed here with neutral red i n d i c ate s th at dye s o r oth e r c o lo re d compounds or formulations are evaluable with this test model. We classi¢ed the compound rose bengal as phototoxic despite its limited phototoxicity in humans. Rose bengal is clearly phototoxic in the in vitro 3T3-NRU assay, but its phototoxicity is limited in vivo by its relative inability to penetrate the stratum corneum (Kaidbey and Kligman, 1978). The assay predicted rose bengal to be potentially phototoxic since it 399 showed a IC50(^UV)/IC50(+UV) value 42, and because two concentrations gave a UVAdependent reduction of viability 430%. However, the phototoxic potential of rose bengal seems to be low compared to that of the other PT compounds listed in Table 3 (the decrease in viability due to UVA was only of 35% at the concentration of 1 mg/ml). The SkinEthicTM epidermis was found to be highly resistant to UVA irradiation; important di¡erences in barrier properties and molecular organization of the stratum corneum in SkinEthic, Episkin 1 , and EpiDermTM tissues (Fartasch et al., 1996; Roguet et al., 1996; Gysler et al., 1999b) are likely to be responsible for the di¡erences in their response to UVA. By increasing UVA irradiation from 6 J/cm2 to 18 J/cm2, we showed that phototoxicity of chlorpromazine and 8-MOP was drastically enhanced. The IC50 obtained with the high UVA irradiation was decreased more than 5fold when compared to the IC50 obtained with the 6 J/cm2 protocol. The psoralen compound 8-MOP is highly phototoxic in cell monolayer assays (i.e., 3T3-NRU assay); however, its phototoxicity is heterogeneous and di¤cult to demonstrate clearly in in vivo studies (Guillot et al., 1985). Using another epidermal model, Liebsch et al. (1997) reported phototoxic activity of 8-MOP; however, this compound tested in the range 0.01^100 mg/ml, did not induce a complete loss of cell viability under UVA, and showed no dose-dependent phototoxic e¡ects (the viability after UVA was between 30% and 50% irrespective of the 8MOP concentration tested). Using the nonoptimal 6 J/cm2 irradiation protocol, we found 8-MOP to be phototoxic only at the maximal tested concentration of 10 mg/ml (IC50 = 10 mg/ml, with the 6 J/cm2 UVA dose). As before for chlorpromazine, increasing the irradiation from 6 J/cm2 to 18 J/cm2 multiplied by ten the phototoxicity of 8-MOP (IC50 = 1 mg/ml, with the 18 J/cm 2 UVA dose). After 18 J/cm2 irradiation, 8-MOP exhibited three phototoxic concentrations leading to 60^70% decrease in cell viability. Encouraged by these results, we tested two weak PT compounds, 6-MC and bithionol, using the two irradiation protocols. Both compounds were found to be nonphototoxic in the 6 J/cm2 protocol, as found with other epidermal tissue (Epiderm) or cell culture (3T3-NRU) phototoxicity test models. Interestingly, increasing the irradiation dose to 18 J/cm2 unmasked a phototoxic potential of both 6-MC and bithionol. Thus, the increase in UV doses maximizes the sensitivity of the assay and allows the clear detection of compounds with low or borderline phototoxicity. Additional screening of a large number of such compounds from various chemical families is necessary to evaluate the potential of the present assay for the detection of weakly phototoxic compounds. In conclusion, the SkinEthicTM reconstituted epidermal model predicted well the phototoxic potential of chemicals under nonoptimized conditions (6 J/cm2), but it has a much greater potential for in vitro phototoxicity prediction for all type of compounds, including low-phototoxicity compounds, when a higher dose of UV is applied (18 J/cm2). References Api AM. In vitro assessment of phototoxicity. In Vitro Toxicol. 1997;10:339^50. Botham P, Earl L, Fentem J, Roguet R, Van de Sandt J. Alternative methods for skin irritation testing: the current status. ECVAM Skin Irritation Task Force Report 1. ATLA. 1998;26:195^211. Coquette A, Berna N, Vandenbosch A, Rosdy M, Poumay Y. Di¡erential expression and release of cytokines by an in vitro reconstituted human epidermis model following skin irritant and sensitizing compounds. Toxicol in Vitro. 1999; 13:867^77. De Brugerolle de Fraissinette A, Picarles V, Chibout S, et al. Predictivity of an in vitro model for acute and chronic skin irritation (SkinEthicTM) applied to the testing of topical vehicles. Cell Biol Toxicol. 1999;15:121^35. Doucet O, Garcia N, Zastrow L. Skin culture model: a possible alternative to the use of excised human skin for assessing in 400 vitro percutaneous absorption. Toxicol in Vitro. 1994;12: 423^30. Doucet O, Robert C, Zastrow L. Use of a serum-free reconstituted epidermis as a skin pharmacological model. Toxicol in Vitro. 1996;10:305^13. Edwards SM, Donally TA, Sayre RM, Rheins LA, Spielmann H, Liebsch M. Quantitative in vitro assessment of phototoxicity using a human skin model; Skin. Photodermatol Photoimmunol Photomed. 1994;10:111^7. Fartasch M, Rosdy M. Maturation of the epidermal barrier in air-exposed keratinocyte cultures: a time course study. J Invest Dermatol. 1996;107:518. Fartasch M, Rosdy M, Ponec M. Development of a structurally competent epidermal barrier in air-exposed keratinocyte cultures: a time course study. J Invest Dermatol. 1996;107: 656. Guillot J-P, Gonnet J-F, Loquerie J-F, Martini M-C, Convert P, Cotte J. A new method for the assessment of phototoxic and photoallergic potentials by topical applications in the albino guinea pig. J Toxicol Cutan Ocul Toxicol. 1985;4:117^33. Gysler A, Koenigsmann U, Schafer-Korting M. Tridimentional skin models recording percutaneous absorption. ALTEX. 1999a;16:63^76. Gysler A, Kleuser B, Sippl W, et al. Skin penetration and metabolism of topical glucocorticoids in reconstructed epidermis and excised human skin. Pharm Res. 1999b;16: 1386^91. Kaidbey KH, Kligman AM. Identi¢cation of topical photosensitizing agents in humans. J Invest Dermatol. 1978;70: 148^51. Laugier JP, Schuster S, Rosdy M, Csoka AB, Stern R, Maibach HI. Topical hyaluronidase decreases hyaluronic acid and CD44 in human skin and in reconstituted human epidermis: evidence that hyaluronidase can permeate the stratum corneum. Br J Dermatol. 2000;142:226^33. Liebsch M, Do«ring B, Donelly TA, Logemann P, Rheins LA, Spielmann H. Application of the human dermal model Skin2 ZK 1350 to phototoxicity and skin corrosivity testing. Toxicol in Vitro. 1995;9:557^62. Liebsch M, Barrabas C, Traue D, Spielmann H. Entwicklung eines neuen in vitro tests auf dermale phototoxizita«t mit einem modell menschlicher epidermis (EpiDermTM). ALTEX. 1997;14:165^74 Mosmann T. Rapid colorimetric assay for cellular growth and survival: application to proliferation and cytotoxicity assays. J Immunol Methods. 1983;65:55^63. Roguet R, Cohen C, Leclaire J, Rougier A. Utilisation d'un ëpiderme reconstruit dans l'ëvaluation des e¡ets cytotoxiques des rayonnements UV. Nouv Dermatol. 1996;15: 384^9. Roguet R, Cohen C, Rougier A. A reconstituted human epidermis to assess cutaneous irritation, photoirritation and photoprotection in vitro. In Rougier A, Goldberg AM, Maibach, HI, eds. In vitro skin toxicology ^ irritation, phototoxicity, sensitization. New York: Mary Ann Liebert; 1994:141^9. (Alternative Methods in Toxicology, vol. 10). Rosdy M and Clauss LC. Terminal epidermal di¡erentiation of human keratinocytes grown in chemically de¢ned medium on inert ¢lter substrates at the air^liquid interface. J Invest Dermatol. 1990;95:409^14. Rosdy M, Fartasch M, Ponec M. Structurally and biochemically normal permeability barrier of human epidermis reconstituted in chemically de¢ned medium. J Invest Dermatol. 1996;107:664. Rosdy M, Bertino B, Butet V, Gibbs S, Darmon M, Ponec M. Retinoic acid inhibits epidermal di¡erentiation when applied topically on the stratum corneum of epidermis formed in vitro by human keratinocytes grown on de¢ned medium. In Vitro Toxicol. 1998;10:39^47. Spielmann H, Lovell WW, Ho«lzle E, et al. In vitro phototoxicity testing. The report and recommendations of ECVAM workshop. ATLA 1994;22:314^48. Address for correspondence: Dr Franc° ois-Xavier Bernard, BIOalternatives, 1 bis, rue des Plantes 86160 Genc°ay, France. E-mail: Bioalternatives@compuserve.com