Enhancing between-facility reproducibility of the SH test as an in vitro skin sensitization test by the improved test method

Abstract

There has been an increased demand to eliminate animal experiments and to replace the experiments with alternative tests for assessing the safety of cosmetics. The SH test is an in vitro skin sensitization test that evaluates the protein binding abilities of a test substance. Skin sensitization must be evaluated by multiple test methods. The SH test uses the same cell line and measuring instruments as the human Cell-Line Activation Test (h-CLAT), which is one of the test methods used to evaluate different key events and is listed in the OECD test guidelines. There are cost advantages to usher the SH test into facilities that are already running the h-CLAT. The SH test is conducted only at a facility that has developed the SH test because studies on the between-facility reproducibility and validity have not been performed. Therefore, to verify the transferability of the SH test and the between-facilities reproducibility, we evaluated the reproducibility of the SH test results at three facilities, including the development facility. After an initial round of testing, the protocol was refined as follows to improve reproducibility among the three facilities: i) determine the optimum pH range, ii) change the maximum applicable concentration of water-soluble substances, and iii) define the appropriate dispersion conditions for evaluating hydrophobic substances. These refinements markedly enhanced the between-facility reproducibility (from 76.0% to 96.0%) for the 25 substances evaluated in this study. This study confirmed that the SH test is an effective skin sensitization test method with high technical transferability and between-facility reproducibility.

INTRODUCTION

There has been an increased demand to reduce or eliminate animal experimentation in various fields including toxicology. The European Union (EU) has banned the use of animal experimentation to assess the safety of cosmetics and cosmetic ingredients. EU has also prohibited the sale of products containing ingredients which were evaluated by animal experimentation (Council of the European Union & European Parliament, 2003). Similar regulations have been established in other countries and regions; therefore, the global standards for evaluation of product safety are rapidly changing to exclude animal experimentation.

Under these circumstances, testing methods to replace animal experimentation have developed at a rapid pace. At present, many of the testing methods assessing toxicity and not relying on animal experimentation have been adopted in the test guidelines (TG) of the Organization for Economic Co-operation and Development (OECD). Substituting in vivo tests with a single in vitro or in chemico alternative testing method, however, is challenging. Integrated testing strategies are required to evaluate the systemic toxicity and complex biologic reactions of various compounds (Adler et al., 2011), especially for testing which requires repeated administration. The current mainstream approach is to use a framework that combines test methods for evaluating key events (KE) in the development of toxicity along the Adverse Outcome Pathway (AOP) in the context of the Integrated Approaches for Testing and Assessment (IATA) (OECD, 2016c).

The IATA, combining in chemico and in vitro testing methods that can evaluate KE based on the AOP, has been proposed to evaluate skin sensitization to chemicals (OECD, 2012a, 2012b). According to the OECD guidelines, there are four KE in the AOP for skin sensitization: KE1, covalent binding to skin proteins; KE2, epidermal keratinocyte inflammatory response; KE3, activation (maturation) and migration of Langerhans cell and dermal dendritic cells; and KE4, activation/proliferation of allergen specific T-cells. Of these, alternative test methods corresponding to KE1, KE2, and KE3 have been developed, and OECD has issued the test guidelines for each method.

Currently, two test methods are available to evaluate KE1: The Direct Peptide Reactivity Assay (DPRA) and the Amino acid Derivative Reactivity Assay (ADRA). These tests evaluate whether or not a test substance has the ability to bind to a peptide that is important for the expression of skin sensitization by assessing the reactivity of the test substance to a synthetic peptide (EURL-ECVAM, 2013; OECD, 2019b, 2019c).

Two other tests have also been developed to evaluate KE2. The KeratinoSens^TM test and the LuSens test were recently adopted by the OECD. Both test methods evaluate the inflammatory response of keratinocytes by assessing the induction of gene groups controlled by the antioxidant reactive region (ARE) during the induction of skin sensitization. Cells containing a reporter gene are used to evaluate whether an ARE gene group has been induced on the basis of the activation of the Keap1-Nrf2-ARE pathway by a skin sensitizer. (OECD, 2018a; Natsch, 2010; EURL-ECVAM, 2014; Ramirez et al., 2014; ESAC, 2016).

KE3 is currently evaluated using methods such as the human cell line activation test (h-CLAT), the U937 cell line activation test (U-SENS^TM), and the interleukin-8 reporter gene assay (IL-8 Luc assay). Dendritic cells activate during the induction phase of skin sensitization, migrate to lymph nodes, and present antigens to naïve T-cells. At this time, the expression of molecules that maintain adhesion to T cells such as CD54 and CD86 is enhanced on the surface of activated dendritic cells and the expression of inflammatory cytokines, such as IL-1β, and chemokines, such as IL-8, are induced (Aiba et al., 2003; dos Santos et al., 2009). These three test methods evaluate the activation of dendritic cells using the expression of cell surface markers or IL-8 as indicators (OECD, 2018b).

The current OECD test guidelines do not include any test methods to evaluate the biological response corresponding to KE4, which is the activation/proliferation of allergen-specific T cells.

Although many test methods have been developed, the non-animal-based test methods which are currently available only evaluate a part of the AOP; therefore they cannot completely replace animal-based test methods. It is necessary to combine multiple test methods along with the AOP to replace animal-based tests and various combinations have been proposed (OECD, 2016a). Even if an appropriate combination of test methods is identified, each test often requires expensive equipment and the use of different cells and reagents. The tests also require intensive labor and familiarity with a number of test methods. Thus, large financial and human labor costs are involved in preparing a location with the capabilities of performing all of the tests for KE1, KE2, and KE3.

We focused on the SH test; an in vitro skin sensitization test developed by Suzuki et al. (2009). This test is performed using the human monocytic leukemia cell line (THP-1) and flow cytometry, similar to the h-CLAT (Ashikaga et al., 2006; Sakaguchi et al., 2006). The SH test has advantages because a facility that is already performing the h-CLAT can perform the SH test by only purchasing the necessary detection reagents and learning a few simple procedures. The SH test is not listed in the OECD TG, but is considered a reliable testing method to evaluate KE1 in the IATA for skin sensitization (OECD, 2016b). Skin sensitizing KE1 is based on the binding of chemicals to proteins. After entering the body, a low-molecular-weight sensitizer (hapten) binds to a lysine or cysteine residue of the carrier protein to acquire immunogenicity recognizable by dendritic cells. The DPRA and the ADRA assess the protein binding ability of the test substance using an indicator that depends on the reactivity of the test substance with a synthetic peptide containing either lysine or cysteine (Gerberick et al., 2004, 2007; Fujita et al., 2014). The SH test is a method of detecting skin sensitizers on the basis of a decrease in the amount of cell surface thiols due to the binding of a test substance to the cell surface proteins (Hirota et al., 2009, 2013). In addition, the SH test can also detect certain skin sensitizers that increase the number of cell surface thiols via endoplasmic reticulum stress in THP-1 cells and activation of the mitogen-activated protein kinase cascade (Hirota et al., 2010). Furthermore, the use of the SH test for assessing KE1 in combination with an ARE assay for assessing KE2 (Natsch and Emter, 2008) and the h-CLAT for assessing KE3 produced good results in a risk assessment model of skin sensitization (Hirota et al., 2015). In their report, the SH test performance was equivalent to that of the DPRA test as an in vitro test method for KE1 (Hirota et al., 2015).

However, the SH test has not yet been validated. The SH test technical transferability to facilities other than the developing facility or its reproducibility between facilities have not been evaluated, either. Therefore, in this study, we verified the ability to establish the SH test at other facilities and evaluated the between-facility reproducibility.

First, two new facilities that adopted the SH test received technical guidance from the developer. Next, tests were conducted on common test substances at the three facilities and the reproducibility of the test results among the three facilities was evaluated. After the initial round of testing, it was revealed that the protocol needed improvement to enhance the reproducibility among the facilities. Therefore, the protocol was adjusted as follows; i) determine the optimum pH range, ii) change the maximum applicable concentration of water-soluble substances from 5000 µg/mL to 15000 µg/mL, and iii) define the appropriate dispersion conditions for evaluating poorly water-soluble substances. As a result of the adjustments, the between-facility reproducibility was markedly enhanced. The SH test became an in vitro skin sensitization test method with high technical transferability as well as between-facility reproducibility.

MATERIALS AND METHODS

Testing facilities

The testing was conducted at three facilities: the Shiseido Global Innovation Center (Facility A), the Frontier Research Center of POLA Chemical Industries, Inc. (Facility B), and the Safety and Analytical Research Laboratories of KOSÉ Corporation (Facility C). The SH test was developed at Facility A. In this report, we used old as well as new data reported by Facility A.

Test chemicals

A total of 25 chemicals were evaluated in this study, including 17 sensitizers and 8 non-sensitizers (Table 1). All of the chemicals were used in the studies of the Working Group for In vitro Skin Sensitization Evaluation of the Japan Cosmetic Industry Association (Hirota et al., 2015). The chemicals were previously evaluated and classified with a local lymph node assay (LLNA) (Urbisch et al., 2015). When selecting the test chemicals, we focused on the following: i) select a wide range of compounds in points of skin sensitization potency and n-octanol / water partition coefficient (log Kow) predicted by KOWWIN^TM contained in the Estimation Programs Interface (EPI) Suite provided by the United States Environmental Protection Agency (available at https://www.epa.gov/tsca-screening-tools/epi-suitetm-estimation-program-interface#models) and ii) increase the proportion of chemical substances that have similar characteristics as cosmetic ingredients (non or low skin sensitization potency and high log Kow). We purchased the test chemicals from MilliporeSigma Corporation (St. Louis, MO, USA) or FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan). The details of the test chemicals are provided in Table 1. The LLNA is known to sometimes obtain false positive results with irritants such as surfactants (OECD, 2010). One example is Sodium lauryl sulfate (SLS). During a peer review of the LLNA, it was confirmed that the positive reaction for SLS in the LLNA was not a skin sensitization but a false positive due to skin irritation (ICCVAM, 1999). In this study, SLS was treated as an LLNA negative substance.

Table 1. Summary of information on evaluated chemicals and results of the SH test by the three facilities.

Cells and cell culture

The human monocytic cell line, THP-1, was obtained from the ATCC (Manassas, VA, USA). These cells were cultured in RPMI 1640 medium with HEPES buffer (Gibco, Thermo Fisher Scientific, Waltham, MA, USA) or without HEPES buffer (Nissui Pharmaceutical Co., LTD, Tokyo, Japan), containing 1% antibiotic–antimycotic (Invitrogen, Thermo Fisher Scientific) and 10% fetal bovine serum (JRH Biosciences, Lenexa, KS, USA, or MilliporeSigma). Facility A and B used RPMI 1640 medium with HEPES buffer while Facility C used RPMI 1640 medium without HEPES buffer. The cells were maintained in an incubator at 37°C under an atmosphere of 5% CO₂ in air.

SH test procedure

The SH test was performed in accordance with the method of Hirota et al. (2013). THP-1 cells, which were maintained by subculture, were cultured at 37°C for 2 hr in a culture medium containing one of the test chemicals. Three concentrations of each test chemical were evaluated: one-third and one-ninth of the highest applied concentration, with the highest concentration set to the concentration that produced 50% cell survival (IC₅₀). Therefore, the cytotoxicity of each test chemical was evaluated in advance, and the IC₅₀ concentration was determined from a dose-survival curve. Cell viability was measured using an MTT assay. After being treated with the test chemical, the cells were washed with phosphate-buffered saline (PBS). Next, Alexa Fluor 488 C5 maleimide (Invitrogen, Thermo Fisher Scientific), a fluorescent dye that selectively binds to the thiols of proteins, was dissolve in PBS and added to the cells and the cells were further cultured for 30 min at 37°C. The cells were then washed again with PBS and analyzed by flow cytometry (BD FACSCanto^TM II or BD FACSCalibur^TM, Becton, Dickinson and Company, Franklin Lakes, NJ, USA). At the time of the analysis, dead cells were identified using 0.625 μg/mL propidium iodide. After the mean fluorescence intensity (MFI) at each test concentration was determined, the relative fluorescence intensity (RFI) was calculated by the following formula: RFI (% of control) = (MFI of chemical-treated cells / MFI of vehicle-treated control cells) × 100. The RFI was not calculated when cell viability was less than 50%. The SH test was performed independently at least three times, and the mean RFI value was calculated from the RFI value for each test. For each test concentration, a positive result required that the mean RFI value was ≤ 85% or ≥ 115%, and the difference from the control group was statistically significant as evaluated by a paired 2-tailed t-test with a p-value less 0.05. The SH test was conducted at three concentrations and was judged for each concentration based on the mean RFI value. The maximum amount of change of the mean RFI values in the SH test (MAC value) was calculated for each test substance and this is shown in the result as a guide of the amount of fluctuation in the RFI value of each substance. The MAC value is the maximum value of the fluctuation amount in the mean RFI value calculated for each test concentration by the following formula: The fluctuation amount in mean RFI = | 100 – (mean RFI value) |.

RESULTS

Confirmation of transferability and reproducibility among three facilities

Confirmation of transferability

After Facility A provided the hands-on training for conducting the SH test to Facility B and C, we evaluated potential differences in the detection sensitivity among the three facilities by having each facility independently perform the SH test using dinitrochlorobenzene (DNCB), which is used as a positive control substance for the SH test. Similar results were obtained for the dose response of DNCB among the three facilities (Fig. 1). The lowest dose of DNCB (3 μg/mL) was negative in all three facilities, and doses ≥ 4 µg/mL of DNCB were clearly positive in all three facilities. Thus, the detection limit for DNCB was 4 μg/mL in all three facilities. These results indicate that there were no major issues in the technical transferability of the SH test.

Fig. 1

Dose response curve of RFI value with DNCB treatment in the three facilities. The RFI value is the mean of three independent test results ± S.D. (n = 3). The test facilities are (▲: Facility A, ●: Facility B, ■: Facility C). The dotted line shows the 85% criterion for a positive result. *: The RFI values of all three facilities were 85% or less and significant (p < 0.05) by t-test, indicating that a positive test result was obtained.

Confirmation of reproducibility among facilities

Next, we verified the reproducibility among the facilities using the 25 test substances (Table 1 and 2). Two of the weak sensitizers in the LLNA, 1-bromohexane and pyridine, were detected as positive at Facility A, but negative at Facility B and C, which means that they were not detected as sensitizers by the new facilities. Methyl salicylate, a non-sensitizer, was negative at Facility A, but positive at Facility B and C. Another non-sensitizing substance, lactic acid, was negative at Facility A and B, and positive at Facility C. In addition, chlorobenzene, another non-sensitizing substance, was positive at Facility A and C but negative at Facility B. 1,2,4-benzenetricarboxylic anhydride, which was previously reported as a moderate skin sensitizer, was not detected as a sensitizer in any of the facilities, but the same results were obtained in all facilities. Even though the results were different from the LLNA data, the reproducibility between facilities was confirmed. In comparison with the results reported by Facility A, the test method developer, the concordance rates (i.e., the accuracy) of the results of Facility B and C were 84.0% and 80.0%, respectively, and the concordance rate of the results among all three facilities was 76.0%. This value is below 80% and meets the criteria of Between Laboratory Reproducibility (BLR) provided in the performance standards of some OECD Guidelines (OECD, 2015a, 2015b, 2019a).

Table 2. Consistency of test results for the three facilities.

For 5 of the 6 substances that did not produce reproducible results among the facilities (1-bromohexane, pyridine, 1-iodohexane, lactic acid, and methyl salicylate), the results of Facility A were in agreement with the results of the LLNA, whereas the results of Facility C did not agree with the LLNA results. For the remaining substance (chlorobenzene), the SH test results at Facility A and C differed from the LLNA results, but the results at Facility B were consistent with the LLNA results. Based on Table 1, Facility A differed from the LLNA for three substances (1,2,4-benzenetricarboxylic anhydride, 2-hydroxypropyl methacrylate, and chlorobenzene). Facility B differed from the LLNA for five substances (1,2,4-benzenetricarboxylic anhydride, 1-bromohexane, pyridine, 2-hydroxypropyl methacrylate, and methyl salicylate). Facility C differed from the LLNA for eight substances (1,2,4-benzenetricarboxylic anhydride, 1-bromohexane, pyridine, 1-iodohexane, 2-hydroxypropyl methacrylate, chlorobenzene, lactic acid, and methyl salicylate). Regarding the predictive performance for LLNA results at each test facility (Table 3), the probability of correctly predicting a positive substance as positive (i.e., the sensitivity) was 94.1% for Facility A and 82.4% for both Facility B and C. The probability of predicting a negative substance as negative (i.e., the specificity) was 75.0% for Facility A and B and remarkably lower (37.5%) for Facility C. The accuracies were 88.0%, 80.0%, and 68.0% for Facility A, B, and C, respectively. The between-facility reproducibility was only slightly below the OECD BLR criterion of 80%, but there was a large difference in the prediction performance of each facility. Therefore, to improve the test accuracy and further improve reproducibility to meet the BLR criteria, the causes of the differences in reproducibility and prediction performance as well as potential improvements of the original standard operating procedure (SOP) were studied as described below.

Table 3. Prediction performance against LLNA results from each facilities.

SH test protocol refinements

Optimization of the pH range during cell treatment

For lactic acid, a non-skin sensitizer, Facility A and B produced negative results while Facility C produced a positive result. In addition, the MAC value was small (4.4) in Facility A, but large in both Facility B (26.4), which had a negative test result, and Facility C (20.1), which had a positive test result. Although the RFI value in Facility B for lactic acid changed in relation to the control treatment, the test was considered to be negative because the change was not significantly different from the control treatment. Because the two facilities with negative lactic acid results (Facility A and B) used HEPES buffer in the cell medium while the facility with positive results (Facility C) used medium without HEPES buffer, we hypothesized that differences in the pH of the medium affected the SH test results. Therefore, we evaluated the effect of the pH of the treatment medium on the SH test (Fig. 2). Lactic acid (3500 μg/mL, the IC₅₀ value) was added to the medium (pH 3.8) and then NaOH was added to prepare a treatment medium with a pH ranging from 4.0 to 10.0 in 1-pH value increments. The RFI value clearly decreased in the unadjusted and the pH 4.0 medium compared to the higher pH adjusted medium (Fig. 2). Although the SH test result could be correctly evaluated as negative in medium with a pH ranging from 5.0 to 10.0, there was a tendency for the tests of the pH 10 medium to become positive due to the mean RFI value. Thus, we determined that it was necessary to maintain the pH of the treatment medium near neutral to obtain good predictability and reproducibility of the SH test. Adjusting the pH changed the lactic acid results at Facility C from a false positive to negative. Furthermore, the amount of fluctuation in the RFI values was reduced and stable negative results were obtained (refer to lactic acid’s data in Table 4).

Fig. 2

Effect of pH on RFI value with lactic acid treatment in the SH test. RFI values are mean ± S.D. (n = 3) of three independent tests. The dotted line shows the cut-off line for a positive result with an RFI value of ≤ 85%. Lactic acid, which is a non-sensitizing substance, was added at 3500 µg/mL, and pH of the medium was adjusted to each pH value with NaOH.

Table 4. Results of the SH test with improved methods at the tree facilities.

Maximum applicable concentration

We then investigated why pyridine, which is a weak sensitizer, produced false negative results at Facility B and C. Both of the new facilities had low MAC values and little fluctuation of the RFI values. This suggests that the exposure level to the test substance was insufficient. Therefore, we examined the maximum applicable concentration of the test substance. In the conventional SOP, the maximum applicable concentration of water-soluble substances for the SH test is 5000 μg/mL. By applying 15000 μg/mL, we succeeded in detecting pyridine as positive (Fig. 3-1). When 10000 μg/mL was applied, the mean RFI value exceeded 115% of the cut-off line, but the change was not statistically significant by t-test and was therefore not determined as a positive result.

Fig. 3-1

Dose response when evaluating pyridine in the SH test. The maximum applicable concentration of the original standard operating procedure of the SH test is 5000 µg/mL. * indicates that the mean RFI value was in the positive region of the SH test and was significant (p < 0.05) by t-test. The gray area shows the negative region of the mean RFI value in the SH test (85 < RFI < 115). The mean RFI values (bars) and cell viability (lines) indicate the mean ± S.D. of three independent test results.

On the other hand, Fig. 3-2 shows the results of the evaluation of 4 non-sensitizing substances (2-propanol, glycerol, ethanol, and propylene glycol) that are commonly used as solvents. Increasing the maximum applicable concentration of each of these substances to 15000 µg/mL did not produce a false positive. Therefore, changing the maximum applicable concentration of the water-soluble test substances for the SH test to 15000 μg/mL improved the detection sensitivity without increasing the false positive rate.

Fig. 3-2

The SH test results for each solvent when changing the maximum applicable concentration. The maximum applicable concentration of the original standard operating procedure of the SH test is 5000 µg/mL. * indicates that the mean RFI value was in the positive region of the SH test and was significant (p < 0.05) by t-test. The gray area shows the negative region of the mean RFI value in the SH test (85 < RFI < 115). Each bar represents the mean RFI value ± S.D.. In the test, three tests were independently repeated 7 times.

As a result of changing the maximum applicable concentration of pyridine to 15000 μg/mL, the result of Facility B turned from false negative to positive, but Facility C did not (refer to pyridine’s data in Table 4). Compared to the conventional SOP, the MAC values for both Facility B and C were clearly increased, suggesting that the cells are reacting to the sensitizer. However, in Facility C the amount of change was not statistically significant, and the result was not determined as positive.

Of the 25 substances examined in this study, the water-soluble substances that required re-testing after changing the maximum concentration were pyridine and propylene glycol. For the other water-soluble substances, it was not necessary to change the maximum concentration due to cytotoxicity and solubility.

Dispersibility of poorly water-soluble substances

Some of the substances with low between-facility reproducibility had high log Kow and low water solubility. The predicted log Kow, determined using KOWWIN™ (https://www.epa.gov/tsca-screening-tools/epi-suitetm-estimation-program-interface), was 3.63 for 1-bromohexane, 4.05 for 1-iodohexane, and 2.6 for both chlorobenzene and methyl salicylate. The SH test incubation time is 2 hr shorter than general testing methods using cells. The evaluation can be performed within the time when the dispersed state can be maintained even if a substance with low water solubility is not uniformly dissolved. As shown in Table 1, substances, such as hexyl salicylate, hexyl cinnamic aldehyde, abietic acid, benzyl cinnamate, and N,N-dibutylaniline, with a predicted log Kow of 4 to 6 were able to be evaluated. The difference in the dispersion state was considered to increase the variability of the test results. Therefore, in the SH tests for substances with poor water solubility, the dispersion criterion of test substance in the medium was set to create a dispersion state in which the test substance would not quickly coalesce during the 2 hr incubation period. In Facility B and C where the false-negative result had been obtained to 1-bromohexane, the improved dispersion state reduced variability of the result and made a statistically significant difference compared to the control group, and the result turned to the positive. On the other hand, for non-sensitizing substances, there were some cases where the result changed to a false positive due to the improvement of dispersibility. In Facility A, the methyl salicylate result was negative and the MAC value was 15 on the borderline. This result was changed to a false positive because the MAC value became larger and the variability was reduced due to the improved dispersion state and associated improved exposure condition in this retest. The results of chlorobenzene at Facility B were also affected by changes in exposure conditions associated with improved dispersibility. Initially, although the MAC value was 5.9, which was almost unchanged, the improved dispersion state increased the MAC value to 69.0 at a lower concentration than before. In Facility C, 1-iodohexane results changed from false positives to negatives because variability in results could not be completely ruled out and no significant difference was obtained (Table 4).

Effect of improved test conditions on between-facility reproducibility

After adjusting the pH, increasing the maximum applicable concentration of water-soluble substances, and improving the dispersion of the poorly water-soluble substances, we again tested the various test substances in the three facilities (Table 4).

Adjusting the pH of lactic acid changed the positive result in Facility C to a negative result, matching those results of facility A and B. Increasing the maximum concentration of pyridine changed the test result from negative to positive in Facility B, similar to Facility A. By increasing the dispersion, 1-bromohexane, chlorobenzene, and methyl salicylate were positive at all facilities, and 1-iodohexane was negative at all facilities. Comparisons of the SH test results between Facility A and Facility B and C are summarized in Table 5. Facility B gave the same results as Facility A for all substances, and Facility C gave the same results as Facility A, except for pyridine. Reproducibility at the three facilities was 96.0%, and the reproducibility between the facilities was greatly increased by improving the test conditions.

Table 5. Consistency of test results with the improved protocol at the three facilities.

The predictability to the LLNA results are summarized in Table 6. The sensitivity was 94.1%, 94.1%, and 88.2% for Facility A, B, and C, respectively. The specificity was 62.5% at all facilities, and the accuracy was 84.0% for Facility A and B, and 80.0% for Facility C. Compared with results obtained before improving the test conditions, the accuracy of Facility A and the specificity of Facility A and B slightly decreased, but the accuracy and sensitivity of Facility B and C improved, and the specificity of Facility C was remarkably improved.

Table 6. Prediction performance with improved protocol against LLNA for each facilities.

DISCUSSION

To avoid animal testing, a number of alternative skin sensitization test methods have been developed. The SH test examined in this study is a method that evaluates the protein binding of sensitizers, which is an important KE for establishing skin sensitization (Hirota et al., 2010; OECD, 2016a). Therefore, we examined the BLR of the SH test, which has not yet been evaluated. The developer, Facility A, provided hands-on training to Facility B and C, and we first compared the detection sensitivity to the positive control substance DNCB among the three facilities. In all three facilities, the limit of detection of DNCB was 4 μg/mL, and the dose-effect curves were similar among the facilities. These results confirmed that the newly introduced technologies had sufficient detection sensitivity in the two new facilities and that it was possible to easily transfer the technique between laboratories.

In the SH test development at Facility A, the positive control was set at 4 μg/mL DNCB as one of the acceptance criteria. It should be noted that the positive control is close to the detection limit, but this was considered to be an appropriate index for confirming the sensitivity of the test method. In addition, confirming the detection sensitivity for DNCB, including the dose-responsiveness, can guide new facilities in the technical acquisition of the SH test. Although the principle of the test method is different between the SH test and the h-CLAT, they both use the same cells and DNCB as a positive control. As described in OECD TG442E (OECD, 2018b), one of the criteria for the technical mastering of the h-CLAT is that the range of the lower limit concentration that can detect DNCB as positive is 0.5 to 10 μg/mL. In other words, if DNCB of 10 μg/mL or less can be detected, it is determined that the h-CLAT has reached a certain technical level. Although the SH test and the h-CLAT use the same cells, the treatment conditions and principles of the tests differ, making direct comparison difficult. However, the DNCB detection sensitivity of the SH test (its detection limit was 4 μg/mL) was not so different from the h-CLAT. Therefore, it was considered that the detection sensitivity of the SH test was sufficient for a skin sensitization test.

To confirm the reproducibility between facilities, the SH test was performed on 25 chemical substances and the results were compared with the data previously reported by Facility A. The concordance rate of the results of all three facilities was 76.0%, which is close to the 80% criteria for BLR in the in vitro test method performance standard of the OECD test guidelines (OECD, 2015a, 2015b, 2019a). Relative to the LLNA data, however, the sensitivity of Facility A was very high (94.1%), while that of Facility B and C was only 82.4%. In addition, there were many false positives in Facility C, with an accuracy of 68.0% and a specificity of less than 40%. The accuracy and specificity are significantly lower than those of Facility A, which were both 75.0%. Even though the SH test tends to have higher sensitivity and lower specificity than other test methods (Hirota et al., 2015), a specificity less than 40% is too low.

Therefore, we investigated the cause of the decrease in specificity and improved the protocol. Lactic acid is a typical non-sensitizing substance and is used as a negative control in the h-CLAT. Although the reported data from Facility A for lactic acid was negative and the MAC value was small, the result was negative with a large MAC value in Facility B and positive with a large MAC value in Facility C. Adjusting the pH of the medium during cell treatment to near neutral decreased the MAC value and stable negative results were obtained among the three facilities.

The two facilities (Facility A and B) that were initially negative for lactic acid used a HEPES buffer-containing medium, while Facility C, which was positive, used a HEPES buffer-free medium. Thus, we hypothesized that the SH test results might be influenced by pH. We determined that a low pH gave false positives for lactic acid; therefore, it was necessary to adjust the pH of the culture solution during cell treatment to be within the range of 5.0 to 9.0 (7.0 ± 2.0). Regarding the use of HEPES buffer-containing medium, the MAC value was large, even in Facility B, and the results from Facility B showed the same changes as in Facility C due to the pH adjustment. In addition, adjusting the pH was more important and the use of HEPES buffer-containing medium was set as a recommended condition rather than an indispensable condition for the SH test.

After the SH test was reported by Suzuki et al. (2009), the maximum applicable concentration of poorly water-soluble components was increased from 1000 µg/mL to 2500 µg/mL by Hirota et al. (2013). To correctly evaluate pyridine, we further increased the maximum applicable concentration of water-soluble components from 5000 µg/mL to 15000 µg/mL. In Facility B and C, the MAC value was small when evaluating 5000 µg/mL pyridine and there was almost no change in the cell surface thiols. Because the same phenomenon was observed at the two facilities, it was not considered to be due to the technical or experimental environment of the facilities. When the maximum applicable concentration was increased to 15000 µg/mL pyridine, the MAC value greatly increased. Although a positive result was obtained in Facility B, a positive result was not obtained in Facility C because the change was not statistically significant. The RFI values of each independent testing fluctuated sufficiently, but the fluctuation was not significant and the test result was not determined as a positive (data not shown). For the current SH test to produce a positive result, the mean RFI value obtained by each independent test must be ≤ 85% or ≥ 115% and the fluctuation must be statistically significant. The variability among each individual test must be sufficiently small to produce a significant difference relative to the solvent control group. With this method, even if all the test results are positive, it is difficult to make a positive judgment when the variability of the result is large across each test.

In the study of dispersibility, the concordance rate of the results was improved by making the dispersal state between facilities similar. We considered specifying the dispersion method, such as the mixing device, ultrasonication conditions, and processing time. When considering the possibility that the SH test will be introduced to many facilities in the future, however, the processing conditions will depend on the equipment used in each test facility and it may not be appropriate to attempt to set specific dispersion conditions. In addition, the appropriate mixing conditions will also depend on the substance being evaluated. Thus, we did not define these conditions in the SOP.

The exposure time of the test substance to the cells in the SH test was 2 hr, which, is relatively short for a test method using cultured cells. In addition, the test system was not greatly affected even if the test substance was not completely dissolved in the medium. Therefore, we confirmed that complete dissolution of the test substance was not an essential condition during the test, and that the test can be performed on substances that are not in a uniformly dispersed state. Thus, if the dispersion state can be maintained only for 2 hr, which is the exposure time required for the SH test, reproducible results can be obtained. For 1-bromohexane, 1-iodohexane, chlorobenzene, and methyl salicylate, which had a large log Kow and low consistency at the three facilities, a stirring method that could maintain a good dispersion state for two hours was set, and the test was conducted again. By improving the dispersion conditions, the reproducibility between facilities for these substances was increased. In the case of chlorobenzene, while cytotoxicity was observed more than when the MAC value of 5.9 was obtained, the result was considered to be the effect of improved exposure of the test substance on the cells. Similar levels of cytotoxicity were observed at the other two facilities, suggesting that the test conditions were homogenized. Although it was possible to turn false negatives to positives by improving the dispersion state, there were also some cases where negatives were turned to false positives such as the case of methyl salicylate and chlorobenzene. However, in most cases the variability in test results was reduced and the reproducibility between facilities could be enhanced by improving and homogenizing the dispersion state. This greatly contributed to the purpose behind this study. As shown in the remarks column of Table 4, among all the test conditions examined in this study, the improvement of dispersion state contributed most to the enhance of reproducibility of the 25 substances. The improvement of dispersion state also contributed to the predictability of the LLNA results. It is ideal to maintain a uniformly dispersed state but our results indicated that even if the test substance slightly precipitates in the medium, an appropriate evaluation can be made as long as it can be readily redispersed by gently shaking the test tube.

By paying attention to the dispersion state, the reproducibility rates between the facilities were improved, but the variation in the RFI value between tests was relatively large because the substances were not homogeneously dissolved. In such conditions, it is difficult to make a positive judgment and the results might be negative. When evaluating poorly water-soluble substances, it is necessary to pay attention to the variability in the RFI of each test and to judge whether the test result is appropriate or not, but the problem may be solved by improving the procedures and conditions. We will consider this issue in a future study.

By improving the testing conditions, reproducibility rates among the three facilities was increased from 76.0% to 96.0%. For the predictability to the LLNA results, Facility C showed a marked improvement in the specificity from 37.5% to 62.5%, and sensitivity and accuracy also improved from 82.4% to 88.2% and 68.0% to 80.0%, respectively. Facility B showed an improvement in the sensitivity from 82.4% to 94.1% and the accuracy from 80.0% to 84.0%, but the specificity was slightly decreased from 75.0% to 62.5%. Facility A showed slight decrease in the accuracy from 88.0% to 84.0% and specificity from 75.0% to 62.5%. This is because, as a result of improving and homogenizing the dispersion state, exposure to the test substance was also improved and homogenized, reproducibility between facilities was improved, and false negative results turned positive, but some negative results turned also false positive. According to the report by Hirota et al. (2015), the predictability of the SH test for the LLNA results of 73 substances is 96.2% for sensitivity, 52.4% for specificity, and 83.6% for accuracy. Compared with these earlier reports, the predicted performance of the refined SH test was almost the same as that of the conventional SH test, and it is considered that the new test conditions resulted in better between-facility reproducibility. In addition, the performance standard for other in vitro skin sensitization test methods requires a criterion of 80% to be cleared for sensitivity, specificity, and accuracy (OECD, 2015a, 2015b, 2019a). Although the improved SH test did not meet the standard for specificity, it did meet the standard for sensitivity and accuracy.

Of the seven LLNA-negative substances used in this study, three substances (2-hydroxypropyl methacrylate, chlorobenzene, and methyl salicylate) were determined to be positive in the new SH test. These substances, however, also give false positives in other test methods. Specifically, 2-hydroxypropyl methacrylate was determined to be positive in the DPRA and the Keratinosens™ test methods, and chlorobenzene was determined to be positive in the h-CLAT (Hirota et al., 2013).

Regarding the false-negative findings, only 1,2,4-benzenetricarboxylic anhydride in this chemical set was not detected as positive in all three facilities. According to previous reports, this substance was detected as positive in the h-CLAT, but produced a false negative in the Keratinosens™ test. In the DPRA, Hirota et al. (2013) reported a positive result while Urbisch et al. (2015) reported a false negative result. These test methods, like the SH test, are performed in an aqueous solution, and it is reportedly difficult to evaluate skin sensitizing properties of acid anhydrides under these conditions (Narita et al., 2017, 2018). In addition, since 1,2,4-benzenetricarboxylic anhydride binds to the lysine group instead of the cysteine group during protein binding (Urbisch et al., 2015), it is reasonable that it cannot be detected by the SH test.

This study revealed that the SH test still has some issues with the operational efficiency of the test and protocol clarity. These issues are currently being evaluated and will be reported in the near future.

In conclusion, based on the results of this study, the SOP of the SH test was improved as follows: i) the maximum applicable concentration of water-soluble substances was changed from the conventional level of 5000 µg/mL to 15000 µg/mL, ii) the pH was set to pH 7.0 ± 2 during the treatment of the test substance, and iii) the dispersion state was maintained for 2 hr in order to evaluate a poorly water-soluble substance. This study suggested that the SH test with the improved protocols is a reliable skin sensitization test with high technical transferability and between-facility reproducibility.

Conflict of interest

The authors declare that there is no conflict of interest.

REFERENCES

•  Adler,  S.,  Basketter,  D.,  Creton,  S.,  Pelkonen,  O.,  van Benthem,  J.,  Zuang,  V.,  Andersen,  K. E.,  Angers-Loustau,  A.,  Aptula,  A.,  Bal-Price,  A.,  Benfenati,  E.,  Bernauer,  U.,  Bessems,  J.,  Bois,  F.Y.,  Boobis,  A.,  Brandon,  E.,  Bremer,  S.,  Broschard,  T.,  Casati,  S.,  Coecke,  S.,  Corvi,  R.,  Cronin,  M.,  Daston,  G.,  Dekant,  W.,  Felter,  S.,  Grignard,  E.,  Gundert-Remy,  U.,  Heinonen,  T.,  Kimber,  I.,  Kleinjans,  J.,  Komulainen,  H.,  Kreiling,  R.,  Kreysa,  J., Batista, Leite, S., Loizou, G., Maxwell, G., Mazzatorta, P., Munn, S., Pfuhler, S., Phrakonkham, P., Piersma, A., Poth, A., Prieto, P., Repetto, G., Rogiers, V., Schoeters, G., Schwarz, M., Serafimova, R., Tähti, H., Testai, E., van Delft, J., van Loveren, H., Vinken, M., Worth, A. and Zaldivar, J.M. (2011): Alternative (non-animal) methods for cosmetics testing: current status and future prospects-2010. Arch. Toxicol., 85, 367-485.
•  Aiba,  S.,  Manome,  H.,  Nakagawa,  S.,  Mollah,  Z.U.,  Mizuashi,  M.,  Ohtani,  T.,  Yoshino,  Y. and  Tagami,  H. (2003): p38 Mitogen-activated protein kinase and extracellular signal-regulated kinases play distinct roles in the activation of dendritic cells by two representative haptens, NiCl₂ and 2,4-dinitrochlorobenzene. J. Invest. Dermatol., 120, 390-399.
•  Ashikaga,  T.,  Yoshida,  Y.,  Hirota,  M.,  Yoneyama,  K.,  Itagaki,  H.,  Sakaguchi,  H.,  Miyazawa,  M.,  Ito,  Y.,  Suzuki,  H. and  Toyoda,  H. (2006): Development of an in vitro skin sensitization test using human cell lines: the human Cell Line Activation Test (h-CLAT). I. Optimization of the h-CLAT protocol. Toxicol. In Vitro, 20, 767-773.
• Council of the European Union and European Parliament. (2003): Directive 2003/15/EC of the European Parliament and of the council of 27 February 2003 amending Council Directive 76/768/EEC on the approximation of the laws of the Member States relating to cosmetic products. Official Journal, L66, 26-35.
•  dos Santos,  G.G.,  Reinders,  J.,  Ouwehand,  K.,  Rustemeyer,  T.,  Scheper,  R.J. and  Gibbs,  S. (2009): Progress on the development of human in vitro dendritic cell based assays for assessment of the sensitizing potential of a compound. Toxicol. Appl. Pharmacol., 236, 372-382.
• ESAC. (2016): ESAC opinion on the BASF-coordinated performance ptandards-based validation of the LuSens test method for skin sensitisation testing, Publications Office of the European Union.
• EURL-ECVAM. (2013): EC EURL-ECVAM recommendation on the Direct Peptide Reactivity Assay (DPRA) for the skin sensitisation testing. Publications Office of the European Union.
• EURL-ECVAM. (2014): EURL-ECVAM recommendation on the KeratinoSens^TM assay for skin sensitisation testing. Publications Office of the European Union.
•  Fujita,  M.,  Yamamoto,  Y.,  Tahara,  H.,  Kasahara,  T.,  Jimbo,  Y. and  Hioki,  T. (2014): Development of a prediction method for skin sensitization using novel cysteine and lysine derivatives. J. Pharmacol. Toxicol. Methods, 70, 94-105.
•  Gerberick,  G.F.,  Vassallo,  J.D.,  Bailey,  R.E.,  Chaney,  J.G.,  Morrall,  S.W. and  Lepoittevin,  J.P. (2004): Development of a peptide reactivity assay for screening contact allergens. Toxicol. Sci., 81, 332-343.
•  Gerberick,  G.F.,  Vassallo,  J.D.,  Foertsch,  L.M.,  Price,  B.B.,  Chaney,  J.G. and  Lepoittevin,  J.P. (2007): Quantification of chemical peptide reactivity for screening contact allergens: a classification tree model approach. Toxicol. Sci., 97, 417-427.
•  Hirota,  M.,  Fukui,  S.,  Okamoto,  K.,  Kurotani,  S.,  Imai,  N.,  Fujishiro,  M.,  Kyotani,  D.,  Kato,  Y.,  Kasahara,  T.,  Fujita,  M.,  Toyoda,  A.,  Sekiya,  D.,  Watanabe,  S.,  Seto,  H.,  Takenouchi,  O.,  Ashikaga,  T. and  Miyazawa,  M. (2015): Evaluation of combinations of in vitro sensitization test descriptors for the artificial neural network-based risk assessment model of skin sensitization. J. Appl. Toxicol., 35, 1333-1347.
•  Hirota,  M.,  Kouzuki,  H.,  Ashikaga,  T.,  Sono,  S.,  Tsujita,  K.,  Sasa,  H. and  Aiba,  S. (2013): Artificial neural network analysis of data from multiple in vitro assays for prediction of skin sensitization potency of chemicals. Toxicol. In Vitro, 27, 1233-1246.
•  Hirota,  M.,  Motoyama,  A.,  Suzuki,  M.,  Yanagi,  M.,  Kitagaki,  M.,  Kouzuki,  H.,  Hagino,  S.,  Itagaki,  H.,  Sasa,  H.,  Kagatani,  S. and  Aiba,  S. (2010): Changes of cell-surface thiols and intracellular signaling in human monocytic cell line THP-1 treated with diphenylcyclopropenone. J. Toxicol. Sci., 35, 871-879.
•  Hirota,  M.,  Suzuki,  M.,  Hagino,  S.,  Kagatani,  S.,  Sasaki,  Y.,  Aiba,  S. and  Itagaki,  H. (2009): Modification of cell-surface thiols elicits activation of human monocytic cell line THP-1: possible involvement in effect of haptens 2,4-dinitrochlorobenzene and nickel sulfate. J. Toxicol. Sci., 34, 139-150.
• ICCVAM. (1999): The murine local lymph lode assay: A test method for assessing the allergic contact dermatitis potential of chemicals/compounds, The results of an independent peer review evaluation coordinated by the interagency coordinating committee on the validation of alternative methods (ICCVAM) and the national toxicology program interagency center for the evaluation of alternative toxicological methods (NICEATM), NIH Publication No. 99-4494, Research Triangle Park, NC: National Institute of Environmental Health Sciences.
•  Narita,  K.,  Ishii,  Y.,  Vo,  P.T.,  Nakagawa,  F.,  Ogata,  S.,  Yamashita,  K.,  Kojima,  H. and  Itagaki,  H. (2018): Improvement of human cell line activation test (h-CLAT) using short-time exposure methods for prevention of false-negative results. J. Toxicol. Sci., 43, 229-240.
•  Narita,  K.,  Vo,  P.T.,  Yamamoto,  K.,  Kojima,  H. and  Itagaki,  H. (2017): Preventing false-negatives in the in vitro skin sensitization testing of acid anhydrides using interleukin-8 release assays. Toxicol. In Vitro, 42, 69-75.
•  Natsch,  A. (2010): The Nrf2-Keap1-ARE toxicity pathway as a cellular sensor for skin sensitizers--functional relevance and a hypothesis on innate reactions to skin sensitizers. Toxicol. Sci., 113, 284-292.
•  Natsch,  A. and  Emter,  R. (2008): Skin sensitizers induce antioxidant response element dependent genes: application to the in vitro testing of the sensitization potential of chemicals. Toxicol. Sci., 102, 110-119.
• OECD. (2010): Test Guideline No. 429. Skin sensitisation: local lymph node assay. In: OECD Guidelines for the Testing of Chemicals), Organisation for Economic Cooperation and Development, Paris.
• OECD. (2012a): The adverse outcome pathway for skin sensitisation initiated by covalent binding to proteins. Part 1: Scientific evidence. In: Series on Testing & Assessment), Organisation for Economic Cooperation and Development, Paris.
• OECD. (2012b): The adverse outcome pathway for skin sensitisation initiated by covalent binding to proteins. Part 2: Use of the AOP to develop chemical categories and integrated assessment and testing approaches. In: Series on Testing & Assessment), Organisation for Economic Cooperation and Development, Paris.
• OECD. (2015a): Performance standards for assessment of proposed similar or modified in vitro skin sensitisation ARE-NRF2 luciferase test methods. In: OECD Environment, Health and Safety Publications: Series on Testing & Assessment), Organisation for Economic Cooperation and Development, Paris.
• OECD. (2015b): Performance standards for the assessment of proposed similar or modified in vitro reconstructed human epidermis (RhE) test methods for skin irritation testing as described in TG 439. In: OECD Environment, Health and Safety Publications: Series on Testing & Assessment), Organisation for Economic Cooperation and Development, Paris.
• OECD. (2016a): Annex I: Case studies to the guidance document on the reporting of defined approaches and individual information sources to be used within integrated approaches to testing and assessment (IATA) for skin sensitisation. In: OECD Environment, Health and Safety Publications: Series on Testing & Assessment), Organisation for Economic Cooperation and Development, Paris.
• OECD. (2016b): Guidance document for the reporting of defined approaches and individual information sources to be used within integrated approaches to testing and assessment (IATA) for skin sensitisation. In: OECD Environment, Health and Safety Publications: Series on Testing & Assessment), Organisation for Economic Cooperation and Development, Paris.
• OECD. (2016c): Guidance document for the use of adverse outcome pathways in developing integrated approaches to testing and assessment (IATA). In: OECD Environment, Health and Safety Publications: Series on Testing & Assessment), Organisation for Economic Cooperation and Development, Paris.
• OECD. (2018a): Test Guideline No. 442D. In vitro skin sensitisation assays addressing the key event on keratinocyte activation. In: OECD Guidelines for the Testing of Chemicals), Organisation for Economic Cooperation and Development, Paris.
• OECD. (2018b): Test Guideline No. 442E. In vitro skin sensitisation assays addressing the key event on activation of dendritic cells on the adverse outcome pathway for skin sensitisation. In: OECD Guidelines for the Testing of Chemicals), Organisation for Economic Cooperation and Development, Paris.
• OECD. (2019a): Performance standards for the assessment of proposed similar or modified in vitro skin sensitisation DPRA and ADRA test methods. In: OECD Environment, Health and Safety Publications: Series on Testing & Assessment), Organisation for Economic Cooperation and Development, Paris.
• OECD. (2019b): Report of the peer review panel on the validation of the amino acid derivative reactivity assay (ADRA). In: Series on Testing & Assessment), Organisation for Economic Cooperation and Development, Paris.
• OECD. (2019c): Test Guideline No. 442C. Key-event-based test guideline for In chemico skin sensitisation assays addressing the adverse outcome pathway key event on covalent binding to proteins. In: OECD Guidelines for the Testing of Chemicals), Organisation for Economic Cooperation and Development, Paris.
•  Ramirez,  T.,  Mehling,  A.,  Kolle,  S.N.,  Wruck,  C.J.,  Teubner,  W.,  Eltze,  T.,  Aumann,  A.,  Urbisch,  D.,  van Ravenzwaay,  B. and  Landsiedel,  R. (2014): LuSens: a keratinocyte based ARE reporter gene assay for use in integrated testing strategies for skin sensitization hazard identification. Toxicol. In Vitro, 28, 1482-1497.
•  Sakaguchi,  H.,  Ashikaga,  T.,  Miyazawa,  M.,  Yoshida,  Y.,  Ito,  Y.,  Yoneyama,  K.,  Hirota,  M.,  Itagaki,  H.,  Toyoda,  H. and  Suzuki,  H. (2006): Development of an in vitro skin sensitization test using human cell lines; human Cell Line Activation Test (h-CLAT). II. An inter-laboratory study of the h-CLAT. Toxicol. In Vitro, 20, 774-784.
•  Suzuki,  M.,  Hirota,  M.,  Hagino,  S.,  Itagaki,  H. and  Aiba,  S. (2009): Evaluation of changes of cell-surface thiols as a new biomarker for in vitro sensitization test. Toxicol. In Vitro, 23, 687-696.
•  Urbisch,  D.,  Mehling,  A.,  Guth,  K.,  Ramirez,  T.,  Honarvar,  N.,  Kolle,  S.,  Landsiedel,  R.,  Jaworska,  J.,  Kern,  P.S.,  Gerberick,  F.,  Natsch,  A.,  Emter,  R.,  Ashikaga,  T.,  Miyazawa,  M. and  Sakaguchi,  H. (2015): Assessing skin sensitization hazard in mice and men using non-animal test methods. Regul. Toxicol. Pharmacol., 71, 337-351.