L-3-hydroxyacyl-coenzyme A dehydrogenase type II (HADH) was described as an endoplasmic reticulum amyloid beta-peptide-binding protein (ERAB), which enhances Abeta toxicity, and accumulates in neurons in Alzheimer's disease (AD). Hence, HADH/ERAB was suggested to mediate the amyloid-induced neurodegeneration. We estimated the in vivo interactions of HADH and Abeta in an immunocytochemical study of ten Alzheimer's disease and seven normal brains using five monoclonal HADH-specific antibodies. We found no HADH in amyloid plaques or vascular amyloid. The neuronal expression of HADH was not correlated with the severity of amyloid load in neuropil. HADH was expressed in vascular smooth muscle cells in young and old controls and in amyloid-free blood vessels in AD cases, but little or no HADH was in smooth muscle cells in arteries with amyloid deposits. The putative intracellular interaction between HADH and Abeta in amyloid-producing cells was further studied in vascular smooth muscle cells isolated from brain blood vessels with amyloid-beta angiopathy - the cells that were shown previously to accumulate Abeta intracellularly ['Research advances in Alzheimer's disease and related disorders' (1995) 747; Brain Res. 676 (1995) 225; Neurosci. Lett. 183 (1995) 120]. HADH had a mitochondrial localization and did not co-localize with an endoplasmic reticulum marker. Cells that accumulated Abeta were those with low expression of HADH and the proteins did not co-localize. Explanation of the association between low levels of HADH and deposition of Abeta by brain smooth muscle cells requires further studies.