Cross-validation of cyanogen bromide-peptide ratios to measure the proportion of type II collagen in pepsin digests of equine articular cartilage, meniscus, and cartilage repair tissue.

Collagen type I and type II were purified from equine flexor tendon and articular cartilage, respectively. Equal amounts of these collagens were cleaved with cyanogen bromide, and 11 mixtures containing increasing proportions of type II collagen were separated in seven identical sodium dodecyl sulfate-polyacrylamide gels. The density of bands was measured in wet gels and the peak areas were used to form six ratios of peptide bands that had polynomial relationships with the known proportions of type I and type II collagen in the mixtures. Calibration curves for determining the proportion of type II collagen in the mixtures were constructed using ratios and combinations of ratios of peak areas. Cross-validation was used to identify calibration curves with the smallest squared prediction error or squared average prediction error for all combinations of ratios. Ratios of peak areas of each one of the seven gels were treated, in turn, as the "unknown," and a prediction was carried out using these unknowns and the ratios from the other six gels. Two ratios had the smallest squared average prediction error and calibration curves were computed for these ratios with all seven gels. These curves were used to estimate the proportion of type II collagen in the pepsin-soluble and the pepsin-resistant fractions of articular cartilage inner and outer meniscus, and cartilage repair tissue. Cross-validation enabled selection of the cyanogen bromide-peptide ratios for calibration curves that resulted in the most accurate estimation of the proportion of type II collagen in pepsin digests of tissues.