Abstract:
Proteins and their interactions with each other are of interest for fundamental
research as well as numerous applications in biology, biochemistry or medical and
pharmaceutic industry. In recent years, research in our group has shown that
trivalent metal salts can cause a rich phase behavior in aqueous protein solutions,
including reentrant condensation, metastable liquid–liquid phase separation and
crystallization. The topic of this thesis is on the one hand the influence of control
parameters in these systems on protein aggregation and especially crystallization.
On the other hand, we focus on the discrimination between classical and non-
classical pathways of crystal nucleation and growth. We study mainly the model
protein β-lactoglobulin (BLG) in the presence of the divalent salts zinc chloride
and cadmium chloride and the trivalent yttrium chloride.
It has been observed that trivalent salts can cause a reentrant condensation in
protein solutions: below a certain salt concentration c* and above a second one,
c**, samples are clear, indicating stable solutions. At intermediate salt concentra-
tions, samples become turbid due to aggregation. The divalent salts CdCl2 and
ZnCl2 cause a first boundary in BLG solutions and a partial re-clearing at a salt
concentration we denote as pseudo-c**. Using SAXS, we show that the effective
interactions become less repulsive approaching c* and finally attractive. Although
the attractive interaction becomes weaker when pseudo-c** is crossed, repulsion is not reached again. The weak attraction close to the boundaries provides optimal conditions for crystallization. Depending on salt concentration and temperature,
different pathways are found. We discuss possible scenarios based on the phase
behavior of colloidal systems with short-range attractions. Structure determina-
tion by X-ray diffraction shows that the cations are an integral part of the lattice
and mediate new contacts by the formation of ion bridges.
While it was not possible to distinguish if the observed intermediate phases
are a direct precursor for crystals or merely act as an agent for heterogeneous,
classical nucleation in the system with ZnCl2 , real-time SAXS measurements of
BLG in the presence of CdCl2 provide evidence for a multistep process around
pseudo-c**: in the SAXS data, a broad peak forms first, indicating an intermedi-
ate structure, followed by Bragg peaks. To quantify the relationship between the
crystals and their possible precursor, the area under the Bragg peaks and under
the additional broad peak were evaluated and plotted as a function of time. It
can be observed that the maximum amount of the intermediate coincides with the
steepest increase of the crystalline state. We propose a first step in which a non-
crystalline, ordered structure is formed within metastable aggregates, followed by
crystal nucleation in this precursor phase. Our observations further suggest that the number of crystals increases strongly in the beginning, but their growth speed
is low inside the aggregates that have a high viscosity which hinders the diffusion
of protein molecules. After enough aggregate material has been consumed by crystal nucleation and growth, the crystals are in contact with the dilute solution and grow faster. The number of crystals, however, stays almost constant in this stage.
Based on these assumptions, a rate equation model was applied which reproduces the experimentally observed kinetics well. Increasing the salt concentration in the
vicinity of pseudo-c** leads to a reduction of aggregates and to fewer crystals that grow larger due the larger amount of available proteins per crystal.
The nonclassical crystallization pathway described above suggests that the struc-
ture of the intermediate phase may be crucial for the subsequent crystal nucleation.
We thus performed a study on the structure of potential precursor phases in the
system of BLG in the presence of YCl3 on different length scales by a combination
of SAXS, SANS and VSANS. The monomer–monomer correlation within larger
compounds, the scattering of dimers as well as the correlation between clusters
and their spacial arrangement can be identified. When the system is cooled below
a certain transition temperature, changes in the internal structure of the inter-
mediate phase can be observed. Moreover, time-dependent SAXS measurements
show clear isosbestic points, allowing to describe the kinetics of the structural
evolution of the intermediate phase by a two-state model.
Experiments on other protein–salt systems suggest that our work can be, to
some extent, generalized. For some of the systems, structure determination by X-ray diffraction has been performed successfully. In all cases, crystal contacts are mediated by ion bridging. Tuning the protein–protein interactions by the addition of multivalent metal salts provides a method for the growth of potentially high-quality crystals.