Use of Raman spectroscopy and chemometrics for the quantification of metal ions attached to Lactobacillus kefir

Aims: To set‐up an experimental and analytical methodology to evaluate the feasibility of developing simple, accurate and quantitative models based on Raman spectroscopy and multivariate analysis for the quantification of metal ions adsorbed to the bacterial surface of Lactobacillus kefir. Methods and Results: One millilitre cultures from two strains of Lact. kefir in the stationary phase were harvested and washed twice with ultra pure water. The bacterial pellets were resuspended into 1 ml solutions of Pb+2, Cd+2 or Ni+2 ranging from 0 to 0·9 mmol l−1. The suspensions were further incubated for 1 h at 30°C at pH 5·5. After centrifugation, the pellets were kept to register the Raman spectra and the supernatants were used for the analytical determination of Pb+2, Cd+2 and Ni+2. Micro‐organisms nontreated with metal ions were used as controls. Principal component analysis (PCA) was performed over the preprocessed Raman spectra to evaluate whether the clusters obtained could be correlated with the concentration of metal ions attached to the bacterial biomass. After that, partial least squares (PLS) models were calibrated with the aim of quantifying the metal ions adsorbed to the bacterial surface. According to the analytical determinations, the maximum binding capacity of all the metals (qmax) attained values that are comparable with those observed for other lactic acid bacteria (ca. 0·200 mmol g−1). The spectral analysis revealed that the main functional groups involved in the bacteria/metal interaction are carboxylates, phosphates and polysaccharides. In PCA, the first two principal components explain more than 72% variance of the spectral data set contained in the data structure, allowing a clear discrimination among samples of different concentrations. Based on this information and using as reference the results obtained by analytical methods, PLS prediction models were successfully defined for the quantification of Pb+2, Cd+2 and Ni+2 attached to the bacterial surface. Conclusions: The calibration and validation of methods based on multivariate analysis allowed the definition of models for the quantification of Pb+2, Cd+2 and Ni+2 attached to bacterial surfaces. The high percentages of explained variances in PCA gave a strong support to calibrate the prediction models, depicting very good correlations with the reference method (correlations ∼0·90 in all cases). Significance and Impact of the study: Lactobacillus kefir CIDCA 8348 and JCM 5818 bind Pb+2, C