Research Article A multiway approach for classification and characterization of rabbit liver apothioneins by CE-ESI-MS We applied a multiway approach to extract information from the analysis of protein isoforms by CE-ESI-MS. Metallothioneins (MT) are low-molecular-weight proteins (6–7 kDa) with a strong affinity for heavy-metal ions. Rabbit liver MT-I and MT-II fractions are purified from MT samples. At low pH, the bound metal ions were released from the amino acid structures, giving rise to apothioneins. MT-I, MT-II and MT apothioneins, which are complex mixtures of protein isoforms, were analyzed by CE-ESI- MS. After data pre-processing, parallel factor analysis (PARAFAC) and multivariate curve resolution-alternating least squares (MCR-ALS) were applied to the data sets. In both cases, the models enabled classification of the protein samples and identification of their characteristic sub-isoforms using a set of three components. MCR-ALS required an initial estimate of the pure mass spectra of the three components. Thus, PARAFAC loadings were used to initialize the MCR-ALS optimization. The classifications obtained with MCR-ALS were slightly better than those obtained with PARAFAC, probably because MCR-ALS was less affected by the small migration time shifts of the pre- processed electropherograms. However, no differences were found between the pure mass spectra of the three components in either model. Finally, MCR-ALS allowed us to obtain an individual electrophoretic profile of each of the three components for each of the samples analyzed, which proved valuable for characterization and quantification purposes. Keywords: CE-MS / Isoforms / MCR-ALS / Metallothioneins / PARAFAC DOI 10.1002/elps.200800212 1 Introduction CE is one of the techniques of choice for separation of the protein isoforms that are a result of microheterogeneity arising in the biosynthesis of a large number of proteins [1–4]. In CE, protein isoforms (e.g. apothioneins of metallo- forms, metalloforms or glycoforms) are primarily separated according to their charge-to-mass ratios, and despite its limited selectivity, UV absorbance detection is used extensively for detection [5–7]. In recent years, CE coupled online with ICP-MS has proved to be useful for quantitative speciation of several metals in metalloproteins [8]. However, CE-ICP-MS is unsuitable for obtaining molecular mass (Mm) information from the different apothionein metallo- forms. For this purpose, CE-ESI-MS is preferred [1–4, 6–7, 9]. Several CE-ESI-MS have been described for the selective separation and characterization of protein isoforms [1–4, 6–7, 9]. However, the performance of CE-ESI-MS is limited and resolution problems could arise when handling complex mixtures of protein isoforms, such as human erythropoietin, which is a mixture of around 100 glycoforms [9]. In such cases, the methods traditionally used for the analysis of MS data may be excessively time consuming. Chemometrics-assisted multiway data analysis is an excellent alternative for handling these complex data sets [10–14]. Multiway data analysis methods in combination with CE-UV have been used for sample classification and char- acterization, peak purity analysis, peak resolution and quantification [10]. In general, UV absorbance at a single wavelength as a function of migration time (first-order data for a single sample or two-way data for a set of samples) or UV spectra, acquired with a DAD, as a function of migration time (second-order data for a single sample or three-way data for a set of samples) have been employed for the Fernando Benavente 1 Balbina Ando ´n 1 Estela Gime ´ nez 1 Alejandro C. Olivieri 2 Jose ´ Barbosa 1 Victoria Sanz-Nebot 1 1 Departamento de Quı ´mica Analı ´tica, Universidad de Barcelona, Barcelona, Espan ˜a 2 Departamento de Quı ´mica Analı ´tica, Facultad de Ciencias Bioquı ´micas y Farmace ´ uticas, Universidad Nacional de Rosario, Rosario, Argentina Received March 31, 2008 Revised June 19, 2008 Accepted June 20, 2008 Abbreviations: ALS, alternating least squares; MCR, multivariate curve resolution; Mm, molecular mass; MT, metallothioneins; PARAFAC, parallel factor analysis; TIE, total iron electropherogram Correspondence: Dr. Fernando Benavente, Departamento de Quı´mica Analı´tica, Universidad de Barcelona, Diagonal 647, E-08028 Barcelona, Espan ˜a E-mail: fbenavente@ub.edu Fax: 134-934021233 & 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com Electrophoresis 2008, 29, 4355–4367 4355