Fluorescence spectroscopy as a tool for monitoring solubility and aggregation behavior of β‐lactoglobulin after heat treatment

Abstract

Denaturation and aggregation of whey proteins are of interest to the food and pharmaceutical industry due to the importance of final structure in functionality, impact on food texture, and the chemical stability of the final product. In this study, we demonstrate the potential of fluorescence spectrometry combined with multivariate chemometric methods for quantifying solubility and aggregation behavior of beta-lactoglobulin (beta-LG); a major whey protein and a frequent food ingredient. Heat-induced aggregation of beta-LG was studied under different conditions including pH, temperature and heating durations. Results showed very good agreement between the fluorescence-based predictions and measurements obtained by HPLC and gravimetric analysis regardless of the conditions. Standard normal variate (SNV), a signal preprocessing and filtering tool, was found to enhance the predictive accuracy and robustness of the fluorescence-based model.

Fluorescence‐based soft‐sensor for monitoring β‐lactoglobulin and α‐lactalbumin solubility during thermal aggregation

“…PLS is able to handle problems associated with noise and collinearity in multivariate data sets with large numbers of highly correlated variables (Geladi and Kowalski, 1986;Qin and McAvoy, 1992). Collinearity is very high among the different fluorescence readings obtained at different combinations of emission and excitation wavelengths (Elshereef et al, 2006). Hence, the original input variables x in the matrix X are replaced by a smaller set of underlying new variables that are uncorrelated, mutually independent (orthogonal) and are mathematically represented by linear combinations of the original descriptors.…”

“…HPLC measurements of b-LG and a-LA concentrations, were arranged in an output matrix to be referred to heretofore as matrix Y whereas the fluorescence measurements were arranged into an input data (X-matrix). The rows in the X matrix correspond to different samples while the columns correspond to fluorescence intensities at different excitation-emission wavelength pairs (Elshereef et al, 2006). The input data is considered to be multivariate in nature since it consists of many measurements (1700 fluorescence intensity data points) for many samples.…”

“…Therefore, heat-induced aggregation and precipitation is an important treatment process in the manufacture of many dairy products, and is used to modify functional properties with the goal of ensuring food safety of the product. Functional, physical and chemical properties of milk such as texture, heat stability, foaming properties and rheological behavior are all affected by the heat treatment (Elshereef et al, 2006;Kessler and Beyer, 1991;Morr, 1985;Newstead et al, 1975;Lucey et al, 1998;Zhu and Damodaran, 1994).…”

“…The soft-sensor proposed was designed by combining fluorescence spectroscopic data acquired for b-LG/a-LA solutions subjected to different heat conditions and Partial Least Squares (PLS) modelling for monitoring solubility of a-LA and b-LG in their mixtures and their aggregation behavior during heat treatment under different treatment conditions. Partial Least Squares regression is a well-known chemometric tool for developing a calibration model which correlates the set of known measurements represented by multi-wavelength fluorescence data to the desired property to be predicted (Becker et al, 2003;Elshereef et al, 2006;Herbert et al, 2000;Hagedorn et al, 2004). This is because the PLS method is capable of handling data sets with large numbers of highlycorrelated variables such as the fluorescence spectral measurements and with few observations (Elshereef et al, 2006).…”

“…Partial Least Squares regression is a well-known chemometric tool for developing a calibration model which correlates the set of known measurements represented by multi-wavelength fluorescence data to the desired property to be predicted (Becker et al, 2003;Elshereef et al, 2006;Herbert et al, 2000;Hagedorn et al, 2004). This is because the PLS method is capable of handling data sets with large numbers of highlycorrelated variables such as the fluorescence spectral measurements and with few observations (Elshereef et al, 2006).…”