Submitted to: Journal of Agricultural and Food Chemistry
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: March 14, 2008
Publication Date: November 12, 2008
Citation: Luthria, D.L., Lin, L., Robbins, R.J., Finley, J.W., Banuelos, G.S., Harnly, J.M. 2008. Discriminating between cultivars and treatments of broccoli using mass spectral fingerprinting and analysis of variance-principal component analysis. Journal of Agricultural and Food Chemistry. 56(21):8130-8140. Interpretive Summary: Genetics and a variety of environmental factors (such as rainfall, pests, soil, irrigation levels, and fertilization) can lead to chemical differences in the same plant materials. A mass spectrometric (MS) method is described that allows the overall chemical composition of plant materials to be compared. This method uses an extraction of a powdered plant material that is analyzed using direct injection MS (no prior chromatographic separation). The spectra of different plant materials are then analyzed using a pattern recognition program called analysis of variance-principal components analysis (ANOVA-PCA). With this method, differences in materials show up as a horizontal separation on the PCA score plots. Further calculations allow the influence (identified as a percent of variance differences in the plots) of the experimental factors (cultivar, treatment, and analytical uncertainty) to be quantified. This method also allows the identification of specific compounds in the plant materials that are most responsible for their chemical differences. This method will prove useful to analysts in comparing and classifying food materials.
Technical Abstract: Metabolite fingerprints, obtained with direct injection mass spectrometry (MS) with both positive and negative ionization, were used with analysis of variance-principal components analysis (ANOVA-PCA) to discriminate between cultivars and growing treatments of broccoli. The sample set consisted of two cultivars of broccoli, Majestic and Legacy, the first grown with 4 different levels of Se and the second grown organically and conventionally with two rates of irrigation. Chemical composition differences in the 2 cultivars and 7 treatments produced patterns that were visually and statistically distinguishable using ANOVA-PCA. PCA loadings allowed identification of the molecular and fragment ions that provided the most significant chemical differences. A standardized profiling method for phenolic compounds showed that important discriminating ions were not phenolic compounds. The elution time of the discriminating ions and previous reports suggest that they were common sugars and organic acids. The mean of the ANOVA calculations of the positive and negative ionization MS fingerprints showed that for UV, 33% of the variance came from the cultivar, 59% from the growing treatment, and 8% from analytical uncertainty. Although the positive and negative ionization fingerprints differed significantly, there was no difference in the distribution of variance. High variance of individual masses with cultivars or growing treatment was correlated with high PCA loadings. The ANOVA data suggests that only variables with high variance for analytical uncertainty should be deleted. All other variables represent discriminating masses that allow separation of the samples with respect to cultivar and treatment.