Abstract
Proteins are being used as therapeutic active components by the pharmaceutical industry. However, protein drugs may suffer physical and chemical degradation as results of protein purification and formulation. The physical stability of proteins in solution may be affected by the interaction of proteins with surrounding molecules at the bulk phase (i.e. protein-protein, protein-water, protein-excipients), as well as protein adsorption to interfaces (i.e. solid-liquid, liquid-liquid and air-liquid). The objective of this thesis was to study the physical stability of proteins, at the bulk solution and oil-water interface, using bovine serum albumin (BSA). In particular, two hypotheses were stated, defining the specific objectives of the study: 1) spectroscopy and multivariate analysis (MVA) could be used to develop methods of protein characterization and quantification, able to detect small amount of unfolded proteins in the bulk solution, and 2) the adsorption of native protein to liquid-liquid interfaces is affected by the presence of unfolded proteins and excipients in the bulk solution.
BSA was dissolved in distilled water to prepare native BSA solutions at 1% w/w and 5% w/w. Samples were heated at 90°C for 30 minutes to prepare heat-denatured BSA solutions. Binary mixtures of native and heat-denatured BSA were prepared over the range of 100% to 50% w/w native BSA. These solutions were analysed using spectroscopy and chromatography methods. FTIR and fluorescence spectra were modified using diverse pre-processing methods like the second derivative (2ndD), area normalization (AN), baseline correction (BC), multiplicative scatter correction (MCS) and standard normal variate (SNV). These pre-processing techniques were used to investigate which of those techniques (or combination of techniques) help to visualise changes in the secondary and tertiary structure of proteins, as consequence of protein heat-denaturation. Additionally, these pre-processing methods were also used to study their capacity to reduce the spectral noise and increase the linearity between the spectral signal and protein concentration. These pre-processed spectra were then used to build partial least-square (PLS) regression models.
FTIR and fluorescence pre-processed spectra were used to quantify native protein concentration in the binary mixtures using PLS regression models. The quality of the PLS models was assessed by comparing the number of PLS factors, correlation coefficient (R2), root mean square error of calibration (RMSEC) and root mean square error of prediction (RMSEP) in the calibration and prediction sets, respectively. In the case of FTIR spectroscopy, spectra were pre-processed using 2ndD, BC and AN which explained the model using three PLS factors; RMSE= 0.91% and 1.64% and R2= 0.997 and 0.991 for the calibration and prediction sets, respectively. In the case of fluorescence spectroscopy, the best PLS model was obtained for spectra pre-processed using AN and BC. This model used one PLS factor to explain the 99% of the spectra variability, RMSE% = 1.38% and 1.32% and R2 of 0.993 and 0.994 for the calibration and prediction, respectively.
The physical stability of proteins is affected by its adsorption to interfaces. The adsorption of globular proteins like BSA to oil-water interfaces starts with protein diffusion from the bulk solution to the interface. If this interaction is favourable, proteins may undergo attachment, molecular relaxation and conformational rearrangement at the interface. At the equilibrium, the interface will be covered by a monolayer of proteins, which further evolves to a multilayer. Increases of G' and G'' moduli were observed from solutions of native BSA alone, as well as from solutions of heat-denatured BSA (0.15 mM) (90°C for 30 min). The resulting maximum G' value was lower in the presence of heat-denatured protein than for the native protein alone. Addition of heat-denatured BSA (0.07 mM) to the native protein solutions decreased the maximum elastic modulus reached at equilibrium (G' 42 ± 6 mN/m) but giving similar G'' (9 ± 1 mN/m) compared to native BSA alone. The delay in the protein adsorption to interfaces was attributed to differences between native and heat-denatured BSA proteins like: surface activity, protein flexibility and protein conformation (globular vs. extended), which may affect protein diffusion and adsorption to the oil-water interface.
The physical stability of proteins is also influenced by protein interaction with excipients presents in the bulk solution or at the interfaces. Excipients may prevent or increase protein adsorption to interfaces (i.e. NaCl, polymers: PBuA-PDMAEMA and PDMAEMA and phospholipids: DPPC, DSPG-Na and DSPC). NaCl at concentrations of 0.1 M and 0.5 M increased BSA adsorption to the oil-water interface giving values of G' and G'' moduli of 35 ± 20 mN/m and 90 ± 2 mN/m (0.1 M) and 25 ± 6 mN/m and 10 ± 0.2 mN/m (0.5 M), respectively. Native BSA in the presence of NaCl (1 M) showed reduced protein adsorption to the oil-water interface giving a lower value of G' (10 ± 1 mN/m) and G'' (7 ± 0.5 mN/m) moduli than for native BSA alone (27 ± 2 mN/m) and 10 ± 1 mN/m, respectively. In the case of native BSA interacting with phospholipids at the interface, results show an initial increase in the G' and G'' moduli followed by a progressive decrease in G' and G'' moduli from 1 x10-2 N/m to 1 x10-4 N/m, for the highest concentration of DPPC (1x10-3 % w/w). In the presence of polymers, the magnitude of G' and G'' moduli for native BSA alone (i.e. 27 ± 2 mN/m and 10 ± 1 mN/m, respectively) was reduced to 8 ± 3 mN/m (G' modulus) and 4 ± 1 mN/m (G'' modulus) for native BSA in the presence of PDMAEMA (8.75x10-3 % w/v) and, 7 ± 10 mN/m (G' modulus) and 3 ± 4 mN/m (G'' modulus) for native BSA in the presence of PBuA (8.75x10-3 % w/v). The decrease in the interfacial tension measurements (IFT) for native BSA in the presence of excipients was attributed to an increase in protein adsorption or excipient adsorption to interfaces, which was higher for native BSA in the presence of NaCl (0.5 M, 0.1 M and 1 M) than for BSA in the presence of phospholipids and polymers.
In conclusion, FTIR and fluorescence pre-processed spectra in combination with PLS regressions gave a suitable method to characterize and quantify native protein content in the bulk solution. Interfacial measurements confirmed that the kinetic and mechanism of native protein adsorption to interfaces is affected by the presence of heat-denatured BSA in the bulk solution, as well as due to the presence of excipients in the bulk phase (i.e. NaCl and polymers), or at the oil phase (i.e. phospholipids).