362 JOURNAL OF COSMETIC SCIENCE ing methods, such as measuring the inhibition of L-DOPA auto-oxidation (12) and inhibition of tyrosinase and melanin production in melanocytes (13,14) are available however, these in vitro methods are not suitable to measure the activity of whitening ingredients in a finished product. Finished products are cytotoxic to the monolayer cell culture and interfere in L-DOPA auto-oxidation assays. These methods are also limited to compounds that are soluble in water or in culture media. In contrast, a living skin equivalent allows the topical application of a finished product or a water-insoluble compound as it is applied to human skin. Living skin equivalent models are not complete in vitro systems, as they lack dermis and other components of human skin such as hair follicles, sebaceous glands, blood vessels, sweat ducts, and sensory nerves. How- ever, these three-dimensional skin models are very useful for testing cosmetics and pharmaceutical products and ingredients before testing on humans. This article reviews a new method to screen tyrosinase-inhibiting agents and subsequently develop whiten- ing products. We utilized a three-dimensional human skin model, Melanoderm, to screen whitening agents before testing the product on humans. Melanoderm is a living skin equivalent in vitro model of the human epidermis, consisting of well-differentiated human keratinocytes and melanocytes. The biochemical, histological, and ultrastructural properties of Melanoderm are similar to those of human epidermis. A cross section of Melanoderm shows the presence of stratum corneum, keratinocytes, and dendritic me- lanocytes localized in the basal cell layer (15). Melanocytes stain positive when exposed to L-dihydroxyphenylalanin (L-DOPA), a precursor of melanin. The relative activity of whitening agents such as kojic and lactic acids and magnesium ascorbyl phosphate (MAP) was measured on Melanoderm. A new method was developed to measure total kojic acid in a finished product using HPLC. Finally, a complete formula was tested on human subjects to determine the correlation with in vitro test methods. MATERIALS AND METHODS TESTING ON MELANODERM Melanoderm was purchased from the Mat-Tek Co, Ashland, MA. Kojic acid, lactic acid, and MAP in either an aqueous or anhydrous base or the bases alone were applied to Melanoderm and incubated for two to three days. Tissues were refed with the fresh medium daily. At the end of incubation, the cream was removed. The Melanoderm was first rinsed in phosphate-buffered saline (PBS), treated with 10% buffered formalin for ten minutes at room temperature, and incubated in 0.1% L-DOPA for one hour at room temperature. After one hour Melanoderm was placed in 0.1% fresh L-DOPA and in- cubated 4 to 16 hours. Melanin from the Melanoderm was extracted and measured at 490 nm using a microplate reader. HPLC The kojic acid assay was performed using a Hewlett Packard 1090 model HPLC in- strument comprised of an automatic injector, ternary pump, photodiode array detector, and a Chemstation to control the instrument and manage the data. The chromatographic conditions consisted of a C s, ODS-5, 150 x 4.5 mm column (Metachem #0297) and a

SKIN-WHITENING PRODUCTS 363 Figure 1. Effect of whitening agents on Melanoderm. Tissues were treated with vehicle control (A), 1% magnesium ascorbyl phosphate (B), 3% lactic acid (C), 1% kojic acid (D), 1% Ascorbic acid (E), and 1% plant extract (F). Tyrosinase activity was detected by L-DOPA as described in Materials and Methods. mobile phase composed of water:acetonitrile (20:80 v/v). The wavelength of detection was 254 nm. Samples were diluted with water:acetonitrile (50:50 v/v). CLINICAL TESTING Ten healthy adult subjects, who gave informed consent, participated in a 12-week study.