Proteins are biological macromolecules that have an increasing importance in food processing and in therapeutic products. Industrial crystallization for purification and separation of proteins has only been seriously considered during the past decade. It has many process parallels with small molecule crystallization, including a remarkable ability to purify proteins. Moreover, crystallization processes are distinctly cheaper and simpler to operate than traditional adsorption technologies, which are currently used to purify proteins. Overall, this thesis aimed to extend the engineering research knowledge of industrial bulk crystallization processes for proteins. The enzyme glucose isomerase (GI), which is used in the manufacture of high fructose corn syrups (HFCS), was studied in this work. The broad topics of study included the assessment and miniaturisation of experimental techniques, the solubility phase diagram, crystal growth and nucleation kinetics and graphoepitaxial crystallization.
To design a bulk crystallization process, kinetic data are required for the crystal growth and nucleation relationships with processing variables. The most important variables are commonly temperature and solute concentration. The crystal growth kinetics of GI were measured as a function of temperature (15, 20 and 25 °C) and concentration (2 -53 mg/mL) at pH 7.0 in 10% (w/v) ammonium sulphate solutions. This was performed using a miniaturised bulk crystallization technique that was developed in this work. Under equivalent conditions, the growth kinetics were also studied by the static technique. The static technique is more commonly used in fundamental studies of protein crystal growth. However, the growth kinetics of GI were found to be significantly different when measured by both techniques. In the agitated bulk crystallizer, GI demonstrated a typical second order growth dependence on supersaturation and a remarkably high dependence of growth rate on temperature. This was characterised as an activation energy of 179 ± 4 kJ/mol. In the static system, GI demonstrated a 3.1 growth order and an even higher activation energy of 280 ± 12 kJ/rnol. Moreover, static crystals had slower growth rates under equivalent conditions, with more than an order of magnitude difference at the lower supersatrations tested. Cold crystallization conditions (< 5 °C) are predicted as the conditions to maximise the growth rates of GI (up to about 5 μm per minute). However, the incorrect kinetics found by the static technique caused the predicted growth rates at these temperatures to be an order of magnitude greater than those predicted by bulk crystallization kinetics. Therefore industrially aimed research should measure growth kinetics by bulk crystallization techniques.
The miniaturised bulk crystallization technique that was developed only required about 1 gram of protein per experiment. This technique would enable kinetic studies in situations where a protein is scarce or costly. This was only possible by using image analysis to measure crystal size distributions. An improved means of data analysis was incorporated into the technique that enabled simultaneous measurement of crystal growth and nucleation kinetics from batch crystallizations. Simultaneous study of nucleation kinetics was necessary because the metastable zone of GI was very small (α= 2.5), resulting in a remarkably high rate of nucleation during batch crystallizations. The nucleation data was correlated with a new model for nucleation in batch crystallizations. This model pieces together the solute aggregation and collision sources of nuclei from a knowledge of the nucleation thresholds. The parameters suggested that collision nucleation was the key issue for GI and would become a greater problem as the crystal content increases.
A model for the approach to equilibrium solubility was developed. This found that the usual waiting period of a day or a few weeks for a solution to reach solubility equilibrium by crystallization is usually not long enough for proteins. The critical importance of accurate solubility data for the analysis of crystal growth kinetics was demonstrated. A solubility error can greatly affect the estimates of growth order and growth constant. This may explain some of the high growth orders (n > 3) found for several other proteins in the literature. To improve the accuracy and speed of solubility data generation, the approach to equilibrium model was adapted for curve fitting purposes in order to fit dynamic concentration data and thereby extrapolate it to the equilibrium solubility value.
Graphoepitaxial crystallization of GI was trailed using highly oriented friction deposited PTFE substrates. Oriented protein crystals could present new applications. However, no significant orientation of GI crystals was found on the PTFE substrates. Graphoepitaxial substrates with greater groove width and period than that for PTFE may induce macromolecular crystal orientation.
Observations on the GI crystal habit found that, unlike lysozyme, GI was insensitive to the variables of temperature and concentration. However, other species present in solution, such as the precipitant, appeared to control the crystal habit.
This work has extended the knowledge of protein crystallization for industrial purposes. The static technique was found to be a misleading way to study crystal growth kinetics of GI. In its place, a new miniaturised bulk technique was used that enabled simultaneous analysis of growth and nucleation kinetics from batch experiments. Correlation s for the solubility, growth and nucleation kinetics of GI were made. Further experimental work on GI is recommended to give more data for the parameters of the nucleation model. This would provide the data to support a process modelling and optimisation study.