Main content area

A Quantitative Structure-Property Relationship Study of the Release of Some Esters and Alcohols from Barley and Oat β-Glucan Matrices

Christensen, Niels Johan, Leitao, Susana Murtinheira da Trindade, Petersen, Mikael Agerlin, Jespersen, Birthe Møller, Engelsen, Søren Balling
Journal of agricultural and food chemistry 2009 v.57 no.11 pp. 4924-4930
barley, oats, beta-glucans, protein structure, flavor compounds, esters, alcohols, strawberries, molecular weight, chemical concentration, flavor, multivariate analysis, data analysis, simulation models, model validation, structure-activity relationships
This study investigates the release of selected strawberry flavor compounds from aqueous solutions of two barley and oat β-glucan products at concentrations of 5, 10, and 15% (w/w). The flavor release of 12 esters and 3 alcohols was measured by dynamic headspace GC-MS. For each compound the ratio of the flavor release from the β-glucan solution to the release from aqueous solution, Arel, was recorded. In general, esters were retained in the β-glucan matrices in a mass-dependent manner where heavier molecules were retained more. Arel for alcohols was found to be significantly larger than for the esters. Whereas Arel values for esters were always below unity, this parameter was above unity for alcohols in some preparations of β-glucan. This implies that relative to esters, alcohols were rejected from some matrices. An increase in the concentration of the β-glucan products was associated with an increased retention of alcohols and esters. For solutions of oat and barley β-glucan products at the same concentration, the oat product retained the flavor compounds more strongly. This difference was more pronounced at low concentrations of the β-glucan products. To investigate the potential of a multivariate approach for the analysis of the flavor release from β-glucan products, partial least-squares regression was employed on a large selection of calculated molecular descriptors, yielding simple QSPR models capable of explaining the variation in Arel. The robustness of the QSPR models was verified by cross-validation and permutation tests. The results indicate that the multivariate modeling approach might provide a useful tool for the investigation of flavor release systems similar to those studied here.