Relationships between storage protein composition, protein content, growing season and flour quaility of bread wheat

The storage protein composition from the Glu-1, Glu-3 and Gli-1 loci encoding high and low molecular weight glutenin subunits (HMW-GS and LMW-GS) and gliadins, respectively, was determined on 30 wheat (T aestivum L) genotypes from three growing seasons. The gliadins and the LMW-GS were identified as gliadin/LMW-GS pairs.

All samples were analysed by two one-dimensional electrophoretic techniques, and selected samples were also subjected to two-dimensional electrophoretic separation. Different statistical/data-analytical techniques were evaluated in the study of how the presence or absence of the protein alleles, the protein content and the growing seasons are related to flour quality.

The year of growth had a large impact on mixograph peak time. When predicting mixograph peak time from the presence or absence of significant proteins and the year of growth, 70% of the variability in mixograph peak time could be explained, whereas only 49% of the variability could be explained when the year of growth was deleted from the model.

Protein had no effect on mixograph peak time as expected, and the well-known positive effect of HMW-GS 5 + 10, and the negative effects of 2 + 12 and 6 + 8 was observed. Furthermore, some of the gliadin/LMW-GS combinations influenced mixograph peak time significantly. The gliadin/LMW-GS at the combined Gli-Al, Glu-A3 loci b;f was positively related to mixograph peak time, whereas f;f and a;a was negatively related.

Although the LMW-GS component f of the alleles b;f and f;f alleles appear similar on one-dimensional gels, two-dimensional separation of selected samples may suggest that the f components in these alleles are different proteins. Cross-validated partial least squares regression combined with empirical uncertainty estimates (jack-knifing) of the parameters estimated in the model, gave similar results to ANOVA in identifying quality related protein alleles.

The applicability of the multivariate approach in proteomics is, however, much wider.