Application of Rotated PCA Models to Facilitate Interpretation of Metabolite Profiles: Commercial Preparations of St. John's Wort Author Anders Juul Lawaetz 1 , Bonnie Schmidt 1 , Dan Staerk 2 , Jerzy W. Jaroszewski 3 , Rasmus Bro 1 Affiliation 1 Department of Food Science, Faculty of Life Sciences, University of Copenhagen, Copenhagen, Denmark 2 Department of Basic Sciences and Environment, Faculty of Life Sciences, University of Copenhagen, Copenhagen, Denmark 3 Department of Medicinal Chemistry, Faculty of Pharmaceutical Sciences, University of Copenhagen, Copenhagen, Denmark Key words ● " Hypericum perforatum L. ● " St. John's wort ● " Clusiaceae ● " principal component analysis (PCA) ● " orthogonal rotation ● " metabolite profiling received June 4, 2008 revised September 26, 2008 accepted October 27, 2008 Bibliography DOI 10.1055/s-0028-1112194 Planta Med 2009; 75: 271–279 © Georg Thieme Verlag KG Stuttgart · New York Published online December 18, 2008 ISSN 0032-0943 Correspondence Anders Juul Lawaetz Department of Food Science Faculty of Life Sciences University of Copenhagen Rolighedsvej 30 1958 Frederiksberg C Denmark Tel.: +45-3533-3254 Fax: +45-3533-3245 ajla@life.ku.dk Original Paper 271 Introduction ! 1 H-NMR spectroscopy is an attractive analytical technique for assessment of samples of biological origin, i. e., biofluids and plant extracts. The tech- nique is non-destructive, applicable to intact bio- material and information-rich with regard to molecular structure elucidation. Thus, the tech- nique has been widely used as the analytical plat- form to generate information-dense data in me- tabonomic, metabolomic, and metabolite profil- ing studies. However, 1 H-NMR spectra of biologi- cal samples can be extremely complex as they may contain thousands of distinctive resonances. Therefore, visual inspection of a series of such spectra may only release a small percentage of the total information available. Computer-based methods are often used to re- duce the complexity of data to a suitable level. In 1 H-NMR-based metabolite profiling studies, principal component analysis (PCA) is often used [1]. Graphical outputs from PCA enable research- ers across disciplines to discuss detailed facets of conceivably complex mathematical models. A PCA model uses orthogonal and intrinsically ab- stract latent variables. This means that interpre- tation of the model in terms of finding the con- nection between loadings and the variables used in the analysis can be difficult. Even though a PCA bi-plot of scores and loadings provides insight into the structure of the data, it can still be diffi- cult to interpret the many correlations occurring in NMR-based metabonomic studies. In this study we explore a route to simplify the interpretation of complex PCA models with re- spect to the influence of individual compounds on the observed clustering of samples. 1 H-NMR spectra and HPLC-PDA profiles of extracts of 24 commercially available preparations of St. John's wort, a popular herbal medicine, are used as model data sets. Metabolite profiles based on 1 H-NMR spectroscopy have previously proven useful for assessment of herbal medi- cines or plant extracts using different two-way chemometric methods [2], [3], [4], [5], [6], [7], [8], [9]. Abstract ! This paper describes the application of orthogo- nal rotation of models based on principal compo- nent analysis (PCA) of 1 H nuclear magnetic reso- nance (NMR) spectra and high-performance liq- uid chromatography-photo diode array detection (HPLC-PDA) profiles of natural product mixtures using extracts of antidepressive pharmaceutical preparations of St. John's wort as an example. 1 H-NMR spectroscopy of complex mixtures is of- ten used in metabolomic, metabonomic and me- tabolite profiling studies for assessment of sam- ple composition. Interpretation of the derived chemometric models may be complicated be- cause several sample properties often contribute to each principal component and because the in- fluence of individual metabolites may be shared by several principal components. Furthermore, extensive signal overlap in 1 H-NMR spectra poses additional challenges to the interpretation of PCA models derived from such data. Orthogonal rota- tion of PCA models derived from 1 H-NMR spectra and HPLC-PDA profiles of the extracts of St. John's wort preparations facilitate interpretation of the model. Using the varimax criterion, rotation of loadings provides simpler conditions for under- standing the influence of individual metabolites on the observed clustering. Alternatively, rota- tion of scores simplifies the understanding of the influence of whole metabolite profiles on the clustering of individual samples. Lawaetz AJ et al. Application of Rotated … Planta Med 2009; 75: 271 – 279