The dynamic relationship between protein folding/unfolding and self-assembly (aggregation) has important implications in multiple areas of research and product development, which span from biotherapeutics and neurodegenerative diseases to additives in foods. For example, even though Alzheimer’s afflicts about 10% of the population over age 70 [1
], the mechanism that triggers the misfolding of β-amyloid proteins and their subsequent aggregation into large plaques in brain cells is largely unknown. In the biotherapeutic world, where the development of stable protein formulations is constantly being sought, an understanding of the causes of protein aggregation would be advantageous to increase the drug’s bioavailability and thereby its efficacy [2
]. Additionally, the consumer industry has a need to manipulate the structure and self-assembly of proteins in foods because of their impact on the overall rheological properties which give rise to desirable texture and flavor for the consumer.
The necessity for understanding protein folding and self-assembly has led to numerous studies on a wide range of proteins [3
]. However developing a comprehensive understanding between colloidal and conformational dynamics in proteins has been challenging because the thermodynamic pathways that drive them, such as unfolding and aggregation, can occur simultaneously and, thereby, in a convoluted manner. Part of this challenge arises from the inability of analytical instruments to effectively tease apart and distinguish these convoluted unfolding and aggregation pathways.
To work toward successfully deconvolving these pathways, a combination of complementary techniques should be employed that can especially analyze several key properties of a single sample in tandem for more consistent results. Therefore, in this study a detailed concomitant dynamic light scattering (DLS), DLS microrheology, and Raman spectroscopy thermal aggregation study on a protein was carried out at a number of different pHs to explore the effects on both colloidal and conformational stability and associated rheological changes. Furthermore, in addition to Raman spectroscopy, differential scanning calorimetry (DSC) studies were carried out to explore the unfolding processes.
Colloidal stability plays a critical role in protein aggregation. The impact of weak protein–protein interactions on colloidal stability has been investigated for a number of protein systems [5
]. In this regard, the second osmotic virial coefficient B22
has been explored as a potential predictor for colloidal stability and as a link to the solvent-mediated interaction of a pair of proteins. This parameter is averaged over the separation distance and relative orientations. When the interaction between the pair is primarily repulsive in nature (e.g., in conditions where the proteins are highly charged and in a low ionic strength environment), B22
exhibits a positive value while under conditions where the interactions are primarily attractive (e.g., near the isoelectric point), B22
exhibits a negative value. The diffusion interaction parameter (kD
) has served as a surrogate for B22
because they exhibit a linear relationship with each other and the former can be directly obtained from DLS measurements with experimental ease and amenability to high-throughput measurements. The diffusion interaction parameter, kD
, has been shown to be a good colloidal stability predictor for therapeutic proteins, such as monoclonal antibodies [6
]. To the best of our knowledge, this understanding has, however, not been extended to food whey proteins, such as β-Lactoglobulin.
β-Lactoglobulin (βlg) is a milk protein with a molecular weight of 18.4 kDa and an isoelectric point around 5.13. It is known to exist as a dimer at neutral pH and dissociate into its constituent monomers below pH 3 [9
]. Like many proteins, the self-association properties of βlg are strongly influenced by electrostatic interactions. The pH and ionic strength of the solution therefore play a critical role in the self-association behavior [10
] which, for βlg, has been well investigated as a function of temperature [10
] and ionic strength [15
]. These studies have illustrated that a particular self-association route is dependent on both the pH and ionic strength, resulting in different onset temperatures for thermal aggregation, as well as different sizes and morphologies of the resulting aggregates. For example, the formation of rod-like aggregates, spherical aggregates, or worm-like aggregates have been reported based on the pH [20
] and ionic strength [21
] of the solution. Fibril formation has also been reported dependent upon formulation conditions [19
]. A detailed comprehensive investigation on the effect of pH on the colloidal stability, as revealed through kD
via DLS, and its corresponding impact on conformational stability via DSC and Raman spectroscopy, has not been reported for this system [30
]. Additionally the impact of the colloidal/conformational stability on the rheological response of the system has not been previously investigated systematically as a function of pH.
3. Experimental Section
A Zetasizer Helix (patent pending) from Malvern Instruments (Malvern, UK) has been used to obtain DLS (nanoscale particle size) and Raman (structural information) data sequentially on a single sample. The Helix system uses a proprietary non-invasive backscatter (NIBS) detector (Malvern Instruments, Malvern, UK) with dynamic (DLS), static (SLS) and electrophoretic (ELS) light scattering to measure the hydrodynamic radius of particles from 0.15 nm to 5 µm. Raman spectra are collected using 785 nm excitation (~280 mW) from 150 to 1925 cm−1 at 4 cm−1 resolution.
β-Lactoglobulin (βlg) (obtained in 90% purity from Sigma-Aldrich Corp. (St. Louis, MO, USA)) was dissolved at 40 mg/mL in a solution of one of the following three pHs: (1) 20 mM sodium phosphate buffer (pH 7.0); (2) 20 mM sum of sodium phosphate and sodium citrate buffer pH 5.5; and (3) 20 mM Trizma base buffer (pH 8.5). After preparation, samples were filtered through 200 nm syringe filters and final concentration of βlg was confirmed using UV spectroscopy (Thermo Fisher Scientific, Waltham, MA, USA).
DLS, microrheology measurements and Raman spectroscopy were performed on the Helix instrument (Malvern Instruments, Malvern, UK) for the three aforementioned samples as well as their associated buffers for reference. Each sample was subjected to a temperature range from 50 to 90 °C at 1 °C intervals with a 30 min equilibration window at each temperature. After equilibration, data from each sample was collected over 15 min in triplicate for DLS, over 15 min for microrheology, and over a 30 s exposure time with 10 scans for Raman spectroscopy at each temperature. Sample aliquots (~20 µL) for Raman and DLS work were placed into a proprietary titanium cuvette with 120 µm thick quartz windows, and positioned in a temperature-stabilized sample holder. For microrheology experiments, 500 µm tracer particles that contain a silica core, and is surface-functionalized with polyethylene glycol polymers (Micromod, GmbH, Rostock, Germany), was added to the protein samples to final concentration of 0.15% w
before each measurement. The percentage of secondary structure components were deconvolved from a Raman protein library of 18 model proteins with a PLS model. The accuracy of the library has been validated and confirmed within 5% deviation from commonly applied CD library or corresponding secondary structure percentage from Worldwide Protein Data Bank for the corresponding protein [41
Differential scanning calorimetry (DSC) was performed using a MicroCal VP-Capillary DSC (Malvern Instruments, Northampton, MA, USA). Thermograms for each protein and buffer sample (350 μL at 40 mg/mL) were obtained from 25 to 100 °C using a scan rate of 60 °C·h−1. Thermograms of the buffer alone were subtracted from each protein prior to analysis using Origin 7.0 (OriginLab, Northampton, MA, USA) graphing software equipped with the MicroCal VP-Capillary DSC (Malvern Instruments, Malvern, UK) analysis software add-on. Fitting was done using a Levenberg–Marquadt non-linear least-squares method that specifically involved a non-2-state model.
This study has illustrated the strong effect of pH on both colloidal and conformational stability and associated rheological changes for βlg. At pH 5.5, the βlg undergoes a rapid decrease in the diffusion coefficient and more profound secondary structural changes with increasing temperature. Corresponding increases in the elastic modulus (G′) indicates that this system is undergoing gelation as a result of protein aggregation. At higher pH, equivalent changes in the diffusion coefficient and secondary structure are not seen. Significant changes in the elastic modulus G′ are also not observed. The melting transition for the pH 5.5 solution occurs at higher temperatures as seen in the DSC measurements and we propose that this is due to a convolution of unfolding, aggregation and gelation. An attempt to deconvolve these effects has been undertaken through resolving the DSC melting transition into two events and comparing it with the low-frequency Raman data, which we assign to the development of an intermolecular hydrogen bonding network. These results further indicate that, at pH 5.5, βlg undergoes further structural changes after aggregation and gelation. Although this study provides a preliminary view of the differences in structural changes and their associated effect on aggregation and rheological changes for a gelling system versus a non-gelling system, a broader comprehensive understanding of the effect of pH requires an extension to both higher and lower pH values than investigated in this study. This is currently ongoing and will be the subject of a future publication. Overall, this study has shown that the combination of DLS, Raman spectroscopy, microrheology, and DSC allows a more comprehensive understanding of the unfolding mechanisms and self-assembly mechanisms in protein systems like βlg.