Biomarker metabolites capturing the metabolite variance present in a rice plant developmental period

Abstract

Background: This study analyzes metabolomic data from a rice tillering (branching) developmental profile to define a set of biomarker metabolites that reliably captures the metabolite variance of this plant developmental event, and which has potential as a basis for rapid comparative screening of metabolite profiles in relation to change in development, environment, or genotype. Changes in metabolism, and in metabolite profile, occur as a part of, and in response to, developmental events. These changes are influenced by the developmental program, as well as external factors impinging on it. Many samples are needed, however, to characterize quantitative aspects of developmental variation. A biomarker metabolite set could benefit screening of quantitative plant developmental variation by providing some of the advantages of both comprehensive metabolomic studies and focused studies of particular metabolites or pathways.

Results: An appropriate set of biomarker metabolites to represent the plant developmental period including the initiation and early growth of rice tillering (branching) was obtained by: (1) determining principal components of the comprehensive metabolomic profile, then (2) identifying clusters of metabolites representing variation in loading on the first three principal components, and finally (3) selecting individual metabolites from these clusters that were known to be common among diverse organisms. The resultant set of 21 biomarker metabolites was reliable (P = 0.001) in capturing 83% of the metabolite variation in development. Furthermore, a subset of the biomarker metabolites was successful (P = 0.05) in correctly predicting metabolite change in response to environment as determined in another rice metabolomics study.

Conclusion: The ability to define a set of biomarker metabolites that reliably captures the metabolite variance of a plant developmental event was established. The biomarker metabolites are all commonly present in diverse organisms, so studies of their quantitative relationships can provide comparative information concerning metabolite profiles in relation to change in plant development, environment, or genotype.

Figure 1

Principal component scores during a rice plant developmental period bridging first tillering. The scores (categorized by value using a grey-scale as indicated in the legend) of Principal Components 1 to 5 (Panels 1 to 5, respectively) are plotted against the progression in sampling of days post-emergence (horizontal axis) and the height of the sampled tissue section (as height [mm] of mid-section – the vertical axis of each panel). Rice plants have a basal meristem, so an increase in mid-section height is also a progression in development. The principal components are of a standardized, centered metabolite space from an analysis of a comprehensive metabolomics dataset. Principal Components 1, 3 and 5 show a pattern of change relative to both developmental variables (axes), while Principal Component 4 and Principal Component 2 (if the influence of the 1-mm mid-section height samples are ignored) vary mainly with days post-emergence and are probably influenced by environment more than development. Each value is the mean of three replicates, each of which pooled sections representing 50 different plants.

Figure 2

Plot of correlations among samples based on biomarker metabolites vs. based on principal component scores. The biomarker metabolite set does a reasonable job of mimicking the pattern among the sampled tissues based on the top five principal components. The pattern among the tissues with respect to their metabolite composition is discerned by the set of all pairwise Pearson correlation values among the tissues. This "correlation measure" of the pattern among the tissues was applied using two different sets of markers. The set represented by the horizontal axis used the scores of the top five principal components from the analysis of the comprehensive metabolomics dataset. The set represented by the vertical axis used the biomarker metabolite concentrations. These metabolite concentrations are standardized and centered because our study was mainly interested in the magnitude and pattern of the variation in the metabolites during development. The correlation values plotted in the figure have been transformed to a Z-scale to bring out the accuracy (slope of a fitted line would be near 1 with an intercept near 0) in the ability of the biomarker metabolite set to mimic the pattern among the tissues, and with reasonable precision (r = 0.82).

Figure 3

Magnitude and pattern of variation in biomarker metabolite concentrations in samples ranging in development. Four of the tissue samples representing a developmental range are profiled with respect to biomarker metabolite variation. The samples progress in height at mid-section of the sampled tissue and in days post-emergence, thus they represent a cross-section of the larger set of tissue samples. The biomarker metabolites are listed along the horizontal axis, and each dot plot shows the Z-scores for the biomarker metabolite concentration. The metabolites vary a lot in absolute concentration, so the Z-score is used to equalize the overall variation in concentration of the metabolites during the study. Thus the figure shows the pattern and magnitude of the variation among the presented tissues, but also the amount of this variation relative to that of the metabolite concentration for the whole study. For example, oxalic acid and glutamate both have a fairly wide range of Z-scores for these samples, although the pattern of variation is opposite.