Analytica Chimica Acta 424 (2000) 105–114 Multivariate extension of the continuous variation and mole-ratio methods for the study of the interaction of intercalators with polynucleotides M. Vives, R. Gargallo, R. Tauler ∗ Dept. Qu´ ımica Anal´ ıtica, Universitat de Barcelona, Diagonal 647, E-08028 Barcelona, Spain Received 14 January 2000; received in revised form 22 May 2000; accepted 13 June 2000 Abstract A recently proposed approach for the study of the intercalation equilibria of drugs and polynucleotides is applied to ethidium bromide and poly(inosinic)–poly(cytidylic) acid. The procedure consists of the extension of the continuous variation and mole-ratio methods by recording all the UV–VIS absorption, fluorescence and circular dichroism spectra for a set of solutions containing a range of polynucleotide:dye concentration ratios. The whole set of spectroscopic data matrices was simultaneously analyzed by a multivariate curve resolution method based on alternating least squares. This procedure allowed the detection of the intercalation complex, the recovery of the concentration profiles and pure spectra for each species and the calculation of the polynucleotide:dye ratio in the complex and the apparent equilibrium constant. The intercalation sites occur every three base pairs and the value for the logarithm of the apparent equilibrium constant at 37 ◦ C in 0.26 M NaCl was 4.6±0.1 M −1 . © 2000 Elsevier Science B.V. All rights reserved. Keywords: Multivariate curve resolution; Factor analysis; Self-modeling; Intercalation; Polynucleotides; Ethidium bromide 1. Introduction The intercalation of drugs or dyes with polynu- cleotides in solution is studied by using spectroscopic techniques. Hence, UV–VIS absorption, molecular fluorescence or circular dichroism spectra are usually recorded for the free polynucleotide, for the free dye and for the mixture at different experimental condi- tions of pH, temperature or ionic strength. From this qualitative analysis of the whole spectra, the presence of an intercalation complex can be deduced. How- ever, in most such studies only the changes at a single wavelength are usually measured for the calculation ∗ Corresponding author. Fax: +34-3-402-12-33 E-mail address: roma@quimio.qui.ub.es (R. Tauler). of equilibrium constants (K eq ) or for the determination of the polynucleotide:dye ratio. Hence, quantitative studies using the continuous variation or Job’s method [1,2] or the mole-ratio method [3] usually monitor the changes at a single wavelength because there is no way to analyze the changes in the whole spectrum [4–6]. This implies that this wavelength must be as se- lective as possible, i.e. only the complex must absorb at it. This disadvantage can be solved by recording the full spectrum and by analyzing the whole data set with an appropriate mathematical method. Several approaches have been proposed for the analysis of multivariate data, i.e. data recorded at more than one wavelength, in biophysical and ana- lytical studies. Most of these approaches are based on the postulation of a chemical model [7,8], i.e. 0003-2670/00/$ – see front matter © 2000 Elsevier Science B.V. All rights reserved. PII:S0003-2670(00)01029-1