There are considerable variations in serum concentrations of the high density lipoprotein (HDL)-associated enzyme, paraoxonase (PON), which is an important determinant of the antioxidant capacity of HDL. The present study examined the hypothesis that differences in the stability of isoforms arising from the coding region L54M polymorphism could contribute to such variations. A model system was developed using transfected Chinese hamster ovary cells to secrete recombinant PON corresponding to human L or M isoforms. The recombinant peptides exhibited the molecular properties of human serum PON. They formed complexes with lipoproteins in culture medium, notably binding to apolipoprotein A-I-containing particles. The enzymatic properties of the recombinant isoforms were comparable to those of human serum PON. The recombinant M isoform lost activity more rapidly and to a greater extent than the recombinant L isoform [26.0 +/- 3.0% vs. 14.0 +/- 1.0% (phenylacetate substrate) and 36.1 +/- 2.0% vs. 19.3 +/- 2.0% (paraoxon substrate) over 96 h (P < 0.01)] in medium containing fetal calf serum or PON-free human serum. Addition of a protease inhibitor resulted in retention of activity by both isoforms. Parallel results were obtained in incubation studies of human serum from donors homozygous LL or MM for the L54M polymorphism. Enzyme activity was lost more rapidly and to a greater extent from MM than LL sera (P < 0.01). A parallel loss of PON peptide mass was also observed, with a significantly greater loss from MM homozygotes (P < 0.001). It corresponded to the appearance of a smaller molecular mass band on SDS-PAGE analysis. Direct analysis of the proteolytic effect using HDL isolated from homozygotes and incubated with purified kallikrein confirmed the greater loss of activity from MM homozygotes and the protective effect of proteolysis inhibitor. The results provide evidence for lesser stability of the M54 isoform of PON, apparently involving greater susceptibility to proteolysis. It provides one mechanism to explain variations in serum levels of PON and has implications for the antioxidant capacity of HDL.