Sciatic nerve from streptozotocin-induced diabetic rats has previously been shown to incorporate more 32P into phosphatidylinositol-4,5-bisphosphate (PIP2) and the principal myelin proteins than normal nerve. In the present study, labeling of ATP and PIP2 was compared. Using nerve segments, [gamma-32P]ATP specific activity reached a plateau after incubation for 4 h with [32P]orthophosphate, whereas the specific activity of [32P]PIP2 rose much more slowly and was still increasing after 8 h. The rate of disappearance of radioactivity from prelabeled ATP was biphasic, with 75% being lost within 30 min and the remainder declining much more slowly for several hours thereafter. In contrast, no decrease in prelabeled PIP2 radioactivity could be detected for up to 4 h. The kinetics of ATP metabolism were not appreciably different for normal and diabetic nerve. However, after incubation with [32P]orthophosphate for 2 h, the specific activity of PIP2 was 50-120% higher in diabetic nerve. This phenomenon, therefore, cannot be ascribed to altered specific activity of the ATP precursor pool. Greater labeling of PIP2 in 32P-labeled diabetic nerve was present in purified myelin isolated using a simple discontinuous sucrose density gradient, but not in a "nonmyelin" fraction. When nerve homogenate was fractionated on a more complex gradient, three myelin-enriched subfractions were obtained which were heterogeneous as judged by morphological appearance, protein profile, and lipid metabolic activity. The proportion of total lipid radioactivity accounted for by PIP2 was elevated in all the subfractions relative to the homogenate. As compared to myelin subfractions from normal nerve, an increased percentage of 32P in PIP2 was obtained only in the major myelin subfraction from diabetic nerve. The phosphorylation of P0 relative to the other myelin proteins was also enhanced in this subfraction in nerve from diabetic animals.