Previous studies have suggested that proline endopeptidase (PE) could participate to the catabolism of the beta-amyloid peptide (Abeta) or to the physiopathological maturation of the beta-amyloid protein precursor (betaAPP). We have examined the putative ability of human purified PE to catabolize Abeta40 and Abeta42 and the possible contribution of this enzyme to the generation of Abeta40 and Abeta42 in human HEK293 cells. We show first that purified human PE does not degrade synthetic Abeta40 and Abeta42, in vitro. We establish that HEK293 cell homogenates exhibit a Z-Gly-Pro-7AMC-cleaving enzyme, the activity of which is inhibited by Z-Pro-Prolinal and S17092 and S19825, two novel PE inhibitors, with affinities similar to those displayed on the purified human PE. These inhibitors also penetrate cells and achieve a full inhibition of endogenous proline endopeptidase in human cells. By means of selective antibodies directed towards the C-terminal of Abeta40 and Abeta42, we assessed the effect of PE inhibitors on the recovery of both Abeta species. This was examined in HEK293 cells stably overexpressing the wild-type and the familial Alzheimer's disease-related Swedish mutated beta-APP. We establish that none of these inhibitors affected Abeta40 or Abeta42 production in these transfected cells. Overall, our study indicates that human PE does not degrade Abeta40 and Abeta42. Furthermore, PE does not contribute to Abeta40 and Abeta42 formation in HEK293 cells. Therefore, PE does not appear to contribute to the Abeta-related aetiology of Alzheimer's disease.