Separation of peptides by capillary electrophoresis : Application of chemometrics for evaluation of separation performance in micellar electrokinetic chromatography

Abstract: The separation of enkephalin-related peptides and protein kinase A peptide substrates werestudied in micellar electrokinetic chromatography and compared with capillary zoneelectrophoresis. Special interest was focused on systems where the electroosmosis wasreversed in combination with neutral micellar agents. Such systems involve the unique combination of minimizing surface interactions by electrostatic effects and increasing themigration time window. This was accomplished either by addition of cationic monomericamines to the background electrolyte or by using fused silica capillaries modified byimmobilizing an amine to the surface.The selectivities were enhanced by adding anionic taurodeoxycholate micelles to the BGE.The efficiencies obtained were highly dependent on the extent of distribution of the peptides tothe micellar phase. The effect of experimental factors that cause band broadening duringseparation were evaluated by fractional factorial design and response surface modelling. Partialleast square (PLS) regression analysis revealed that very high efficiencies were obtained forpeptides with low distribution to the micelles, while the efficiencies drastically decreased forpeptides strongly associated to the micelles, probably due to slow sorption-desorption kinetics.The separation of enkephalin-related peptides in a system with anionic sodium dodecylmicelles (SDS) was optimized by utilizing experimental design and response surfacemodelling. Such strategies have the advantages of greatly reducing the number of experimentsand allowing identification of interaction effects between the variables. The effect ofacetonitrile was studied, in four different concentration domains, together with SDSconcentration, temperature and the ionic strength of the buffer via central composite design,and related to resolution, migration factor and migration time window utilizing PLS-regression. The results revealed a complex system with very different influences of theexperimental variables in respective domain. The effect of acetonitrile is highly non-linear insystems of this kind and dependent on the temperature used. A change in the thermodynamicsof the distribution behaviour was observed at increasing acetonitrile concentrations. Further,the relationships between different parameters that influence the resolution were investigatedby principal component analysis.