Analysis of Four-Way Two-Dimensional Liquid Chromatography-Diode Array Data:  Application to Metabolomics

Abstract

Two-dimensional liquid chromatography (2D-LC) is rapidly gaining popularity for the analysis of very complex mixtures, including proteomic and metabolomic samples. It provides an effective strategy for separating such samples, because the resolving power of 2D-LC is far superior to that of traditional single-dimension separations. The present work focuses on the development of data analysis methods for the extremely large data sets, on the order of 10 million data points, generated by 2D-LC with diode-array detection (DAD). Specifically, we have applied and adapted chemometric methods to the analysis of maize seedling digests, focusing on compounds related to the biosynthetic pathways of indole-3-acetic acid, the primary growth regulator in plants. The chemometric techniques of window target testing factor analysis (WTTFA), along with parallel factor analysis - alternating least squares (PARAFAC-ALS) were used to analyze 2D-LC-DAD chromatograms of a sample composed of 26 indolic standards, 2 extracts of mutant orange pericarp maize seedlings, 2 extracts of wild-type maize seedlings, and a blank sample. The indolic compounds studied belonged to six spectrally unique groups, and WTTFA was able to specifically identify the presence or absence of any of the 26 indolic standards in the mutant and wild-type samples. A PARAFAC-ALS algorithm and an ALS algorithm with flexible constraints were successfully applied to resolve the spectrally rank deficient data and to demonstrate the quantitative potential of multivariate curve resolution methods. Using this procedure, 95 total peaks were resolved in the data set analyzed. Of those 95 peaks, 45 were found in both the mutant and wild-type maize samples, 16 peaks were unique to the mutant maize samples, 13 peaks were unique to the wild-type maize samples, and 15 peaks were unique to the standard chromatograms. Of the 26 standards included in the data set, several indole acetic acid conjugates were identified and quantified in the maize samples at levels of approximately 0.3-2 microg/g plant material.

Evaluation of the multivariate selectivity of multi‐way liquid chromatography methods

“…The 2D-LC system described by Stoll et al [7] uses reversed-phase gradient elution in both dimensions, where the first column is less retentive on average than the second column. The analysis of several extracts of maize seedlings and a set of indolic standards showed that the vast majority of the components in the mixtures eluted in the upper-left quadrant of the separation space and that only about 70% of the total possible separation space was utilized [19]. This limitation does not negate the potential of 2D-LC to separate complex samples with many components, and the practical peak capacity of the 2D separation is still far greater than for a 1D separation.…”

“…The spectra for the 500 components were randomly sampled from the 47 drug spectra described in Subsection 3.1. The peak widths in the 1st and 2nd dimension and the correlated peak function for both 1st dimension and 2nd dimension variation were designed to approximate the metabolomics data that were reported in Reference [19], and the recommendations of Wang et al [35], Wang X, Carr PW, Stoll DR, personal communication and Stoll et al [36] for optimal gradient conditions. All of the parameters used in the comprehensive 2D-LC simulations are summarized in Table I. A 2D-LC simulation was also carried out to determine the dependence of selectivity on the phase of the sampling rate.…”

“…Comprehensive two-dimensional separation techniques, including two-dimensional gas chromatography (2D-GC) and 2D-LC are becoming popular for the separation of complex mixtures [7,9,19]. Giddings defined comprehensive two-dimensional separations as those in which a sample is subjected to two separation techniques that are at right angles (or orthogonal) to one another.…”

Fast Determination of Phenanthrene in Soil by Gas Chromatography‐Mass Spectrometry Using Chemometric Resolution and Standard Addition Method

“…has been proved to be efficient approach for the analysis of complex system such as traditional Chinese herbal medicine and metabolomic samples. [21,22] Taking the advantage of 2-D bilinear signals, multivariate curve resolution-alternating least squares (MCR-ALS) [23] was developed and employed in the analysis of environmental and biological samples. [24,25] Furthermore, chemometric methods for resolving three-or even four-dimensional signals have been developed, such as alternating trilinear decomposition (ATLD), [26,27] parallel factor analysis (PARAFAC) [28,29] and four-way self-weighted alternating normalized residue fitting algorithm.…”

Efficient recovery of electrophoretic profiles of nucleoside metabolites from urine samples by multivariate curve resolution

“…These methods were initially developed to determine the number of compounds present in a multicomponent system; however, they have subsequently been extended by a least squares or iterative least-squares stage to resolve the data matrix into the individual concentration profiles [13]. Other multivariate curve resolution methods such as the generalized rank annihilation method, direct tri-linear decomposition, parallel factor analysis alternating least squares, and multivariate curve resolution alternating least squares (MCR-ALS) can simultaneously handle several data matrices originating from several HPLC-DAD runs (threeway data analysis) [16][17][18][19]. This makes them convenient tools in the analysis of numerous samples, e.g.…”