In an effort to elucidate the nature of the collagen-platelet interaction, the effects of collagen modification on platelet aggregation have been studied. We have shown that purified rat skin (salt) soluble collagen is effective at about 20 nM in mediating platelet aggregation in human platelet-rich plasma. This concentration is somewhat greater than that required of several skin insoluble collagens (ca. 10 nM). Both the alpha1(I) and alpha2 chains from rat skin soluble collagen produced platelet aggregation, but only at concentrations of about 13 muM and 55 muM, respectively. In contrast, heat-denatured collagen and chains (e.g., 65 muM alpha1(I) and 160 muM alpha2) failed to induce platelet aggregation and to inhibit platelet aggregation by native collagen. Glycopeptides were prepared from human skin insoluble collagen by extended digestion with bacterial collagenase and trypsin, and were purified by gel filtration into two classes. One class of higher molecular weight contained sialic acid, glucosamine, galactosamine, fucose, mannose, galactose, and glucose, and the other of lower molecular weight consisted primarily of a mixture of galactose and galactosyl-glucose units O-glycosidically linked to hydroxylysine-containing peptides. We found that, after the residual tryptic activity contaminating the higher molecular weight fraction was inhibited, neither of the glycopeptide classes produced nor inhibited native human skin insoluble collagen-mediated platelet aggregation at the highest concentration examined (ca. 1-2 mg glycopeptide per ml of platelet-rich plasma). Highly purified samples of the hydroxylysyl glycosides, hydroxylysylgalactose and hydroxylysylgalactosylglucose (Hyl-Gal and Hyl-Gal-Glc, respectively), were prepared from human urine and labeled at galactose using galactose oxidase followed by reduction with tritiated borohydride. Binding studies with platelet-rich plasma showed that, at concentrations greater than 50 nM, Hyl-Gal gives apparent binding to platelets, but there was no evidence of Hyl-Gal-Glc binding to platelets at concentrations up to 250 nM. At concentrations several hundredfold higher than the equivalents present in the minimum concentration of rat skin soluble collagen required for platelet aggregation, neither Hyl-Gal (at 29 muM) nor Hyl-Gal-Glc (at 18 muM) caused platelet aggregation or inhibited platelet aggregation by native collagen. Also, at a concentration of 85 muM (which represents a concentration about two thousandfold higher than the equivalents in the minimum concentration of soluble collagen required for platelet aggregation) the Gal-Glc-containing 36 residue rat skin soluble collagen alpha1(I)cyanogen bromide #5 peptide had no platelet aggregating or inhibiting activity. Modification of at least 90% of the rat skin soluble collagen carbohydrate by mild periodate oxidation had no effect on the platelet aggregating activity. Human skin insoluble collagen was reacted with periodate under the same conditions, and this had no demonstrable effect on its ability to induce platelet aggregation. This indicates that the normal carbohydrate side chains of these collagens are not required for the platelet interaction that produces the release of ADP and other metabolic constituents and leads to aggregation.Thus, collagen-platelet interactions appear to involve at least two distinct binding sites on the platelet plasma membrane. One is a protein binding site that activates platelet aggregation and has high specificity and affinity for the collagen triple-helical fold or perhaps even for a particular amino acid sequence in the triple helix.