This paper describes proteoglycan catabolism by adult bovine articular cartilage treated with retinoic acid as a means of stimulating the loss of this macromolecule from the extracellular matrix of cartilage. Addition of retinoic acid (10(-12)-10(-6) M) to adult bovine articular cartilage which had been labeled with [35S]sulfate for 6 h after 5 days in culture, resulted in a dose-dependent increase in the rate of loss of 35S-labeled proteoglycans from the matrix of the tissue. Concomitant with this loss was a decrease in the proteoglycan content of the tissue. Incubation of cultures treated with 1 microM retinoic acid, at 4 degrees C, or with 0.5 mM cycloheximide, resulted in a significant decrease in the rate of retinoic acid-induced loss of proteoglycans and demonstrated cellular involvement in this process. Analysis of the 35S-labeled proteoglycans remaining in the matrix showed that the percentage of radioactivity associated with the small proteoglycan species extracted from the matrix of articular cartilage explants labeled with [35S]sulfate after 5 days in culture was 15% and this increased to 22% in tissue maintained in medium alone. In tissue treated with 1 microM retinoic acid for 6 days, the percentage of radioactivity associated with the small proteoglycan was 58%. Approximately 93% of the 35S-labeled proteoglycans released into the medium of control and retinoic acid-treated cultures was recovered in high density fractions after CsCl gradient centrifugation and eluted on Sepharose CL-2B as a broad peak with a Kav of 0.30-0.37. Less than 17% of these proteoglycans was capable of aggregating with hyaluronate. These results indicate that in both control and retinoic acid-treated cultures the larger proteoglycan species is lost to the medium at a greater rate than the small proteoglycan species. The effect of retinoic acid on proteoglycan turnover was shown to be reversible. Cartilage cultures maintained with retinoic acid for 1 day then switched to medium with 20% (v/v) fetal calf serum for the remainder of the culture period exhibited decreased rates of loss of 35S-labeled proteoglycans from the matrix and increased tissue hexuronate contents to levels near those observed in tissue maintained in medium with 20% (v/v) fetal calf serum throughout. Furthermore, following switching to 20% (v/v) fetal calf serum, the relative proportions of the 35S-labeled proteoglycan species remaining in the matrix of these cultures were similar to those of control cultures.