T he subendothelial retention of LDLs through their interaction with proteoglycans has been proposed to be a key process in the pathogenesis of atherosclerosis. In vitro studies have identified eight clusters of basic amino acids in delipidated apo-B100, the protein moiety of LDL, that bind the negatively charged proteoglycans. To determine which of these sites is functional on the surface of LDL particles, we analyzed the proteoglycan-binding activity of recombinant human LDL isolated from transgenic mice. Substitution of neutral amino acids for the basic amino acids residues in site B (residues 3359-3369) abolished both the receptor-binding and the proteoglycan-binding activities of the recombinant LDL. Chemical modification of the remaining basic residues caused only a marginal further reduction in proteoglycan binding, indicating that site B is the primary proteoglycan-binding site of LDL. Although site B was essential for normal receptor-binding and proteoglycan-binding activities, these activities could be separated in recombinant LDL containing single-point mutation. Recombinant LDL with a K3363E mutation, in which a glutamic acid had been inserted into the basic cluster RKR in site B, had normal receptor binding but interacted defectively with proteoglycans; in contrast, another mutant LDL, R3500Q, displayed defective receptor binding but interacted normally with proteoglycans. LDL with normal receptor-binding activity but with severely impaired proteoglycan binding will be a unique resource for analyzing the importance of LDL- proteoglycan interaction in atherogenesis. If the subendothelial retention of LDL by proteoglycans is the initial event in early atherosclerosis, then LDL with defective proteoglycan binding may have little or no atherogenic potential.