The effects of peroxisome proliferators on protein abundances in mouse liver.

We have investigated the effects of five peroxisome proliferators (PPs : clofibric acid, DEHP, WY14,643, nafenopin, and LY171883) on the abundances of a large number of proteins in the livers of treated mice at 5- and 35-day time points. LY171883 was investigated at a range of doses, and one of its close structural analogs that is not a peroxisome proliferator (LY163443) was included as a negative control compound. Liver samples were analyzed by quantitative 2-D electrophoresis. Data for a selected set of 107 liver protein spots that respond strongly to at least one of the test compounds was subjected to principal component analysis to search for global protein pattern changes. The first component (PC1) accounted for 51% of the total data variance and was identified as a global measure of peroxisome proliferation by its correlation with enzymatic peroxisomal beta-oxidation. Component PC2 (7%) separated 5- and 35-day exposures, and PC3 (5%) separated groups treated with LY163443 from the rest. We used PC1 as a surrogate for equivalent dose in order to examine the effects of diverse compounds, with widely differing potencies, on a common scale. Analyzed in this way, the data indicate that all the peroxisome proliferators tested produce effects over wide time and dose ranges that fall on or near a single curve. Examination of specific protein responses showed that many proteins individually show a unified response curve, but that curves for different proteins were different. In particular, it appears that some constitutive proteins showing modest inductions with a high dose plateau (such as cytosolic epoxide hydrolase) are inducible at lower doses than some proteins showing very strong, nonplateaued inductions (such as the 80-kDa peroxisomal bifunctional enzyme). The results provide support for a unified receptor-based mechanism controlling the main PP response, but demonstrate that individual responsive genes can show quite different dose-response curves.