It was postulated that thalidomide causes birth defects by being metabolized to a toxic electrophilic intermediate. This hypothesis was tested by using an in vitro assay in which drug toxicity to human lymphocytes was assessed in the presence of a hepatic microsomal drug metabolizing system. Maternal hepatic microsomes from pregnant rabbits mediated the production of a metabolite that was toxic to lymphocytes. Toxicity was enhanced by inhibitors of epoxide hydrolase (EC 3.3.2.3) and abolished by adding the purified enzyme to the incubation medium. The metabolite thus appears to be an arene oxide, consistent with the previously reported isolation of phenolic metabolites of thalidomide from the urine of treated animals. Two teratogenic analogs of thalidomide (phthalimidophthalimide and phthalimidinoglutarimide) were also toxic in the system; two nonteratogenic analogs (phthalimide and hexahydrothalidomide) were not toxic, even in the presence of epoxide hydrolase inhibitors. The toxic metabolite of thalidomide was not produced by rat liver microsomes (the rat is not sensitive to thalidomide teratogenesis) but was produced by hepatic preparations from maternal rabbits, and rabbit, monkey, and human (all sensitive species) fetuses. A toxic arene oxide therefore may be involved in the teratogenicity of thalidomide.Thalidomide was identified as a human teratogen 20 years ago (1-3). Compared to other teratogens, thalidomide's selective toxicity in the embryo, particularly for the developing limbs, and its relative lack oftoxicity in the adult is striking (4). Despite intensive investigation, however, the mechanism of the fetal toxicity of the drug remains unknown.An interesting early observation was that rats were resistant to the teratogenic effects of thalidomide but rabbits and monkeys were sensitive (5-7). Differences in species susceptibility could result from differences in biotransformation of the compound. It was noted that rabbit liver homogenates enhanced the rate of disappearance of the drug from incubation mixtures whereas rat liver homogenates did not (8). Similarly, in vivo, more thalidomide metabolites were bound to liver macromolecules in the rabbit than in the rat, suggesting that a metabolite might interact covalently with macromolecules important in morphogenesis. Furthermore, after thalidomide treatment, 4-and 5-hydroxylated metabolites of thalidomide were recovered from the urine of rabbits but not from rats (9).The presence ofphenolic derivatives ofthalidomide suggests that the drug might undergo oxidative metabolism via an arene oxide intermediate. Arene oxides have been implicated as mutagens, cytotoxins, and teratogens (10-12). Thus, birth defects caused by the anticonvulsant phenytoin may result from an arene oxide metabolite which covalently binds to critical structures in the developing fetus (12). Therefore, we have attempted to look for a possible toxic arene oxide metabolite ofthalidomide. We have used an in vitro assay system in which human lymphocytes are the target of metabol...