My 40+ year research career has focused primarily on the role of biotransformation and disposition as a determinant of chemical carcinogenicity, although my graduate work in the late 1970s was focused on membrane transporters—there were only three known hepatic xenobiotic transporters then: organic anion, organic cation, and “neutral” (ouabain) transporters. A search of “liver transporter” in the Human Protein Atlas now lists 349 different genes!

I also was fascinated by the role of the then recently discovered cytochrome P450 enzymes—there were only two known at the time: cytochrome P450 and cytochrome P448. We know now of 58 different human CYP genes. But much of my early career research focused on the glutathione S-transferase (GST) family of genes. I was fascinated by the discovery that the potent rat liver toxin and carcinogen, aflatoxin B1 (AFB), was not carcinogenic at all in adult mice but was one of the most potent known rat carcinogens.

After several years of work in my new lab, we cloned a GST gene in the mouse that conferred nearly complete resistance to the genotoxic effects of AFB (mGstA3). We subsequently showed that there was an orthologous gene in the rat, rGSTA5, that had high activity toward the genotoxic AFB,8,9-oxide (AFBO) but was not constitutively expressed in the rat liver. But we could turn the rGSTA5 gene on with sulforaphane—present in broccoli and other cruciferous vegetables. We then showed that feeding rats broccoli protected them from AFB genotoxicity/carcinogenicity. We also found that none of the human alpha-class GSTs had any measurable activity toward AFBO, suggesting that humans—relative to rodents—might be highly susceptible to AFB hepatocarcinogenesis, but we found that the human Mu class GSTM1, which is highly polymorphic in the human population, had small but measurable activity toward AFBO. We then showed that human-isolated hepatocytes that were GSTM1-null had about three times more AFB-DNA adducts compared to human hepatocytes that were GSTM1-positive.

Although several studies at the time found that human CYP3A4 had the highest activity of any P450 in human liver toward activating AFB to AFBO, we showed that human CYP1A2 was actually more important—it had a much lower Vmax than CYP3A4 but had a much higher affinity (lower K_M). Thus, at the low concentrations encountered in the human diet, virtually all activation of AFB to AFBO occurred by CYP1A2, not CYP3A4.

More recently, I’ve been interested in how adverse outcome pathway approaches to risk assessment can do a much better job of estimating human health risks from potential carcinogens and provide a more valid approach to incorporating known species differences between rats, mice, and humans in estimating potential health risks. I will explore two such examples—the importance of using human-based toxic equivalence factors for estimating risk of dioxins and dioxin-like compounds and the importance of species differences in DNA repair in estimating human health risks of the potent rat liver carcinogen, N-nitrosodimethylamine (NDMA), recently found as a contaminant in numerous pharmaceuticals.