We describe a new protein characterization technique called self‐interaction chromatography (SIC), which exploits the specificity of protein–protein interactions that is common to protein aggregates and enables the rapid screening of protein formulation additives as physical stabilizers against aggregation. This technique also enables the identification of specific interaction sites and the determination of their relative importance for self‐association. Mannitol, glycine, and dextran 40 were tested for their stabilizing effect toward the model protein lysozyme. Dextran 40 exhibited a poor stabilizing effect. While mannitol stabilized both the native and acid‐denatured forms of lysozyme, glycine stabilized the native form with respect to the denatured species. These results are in good agreement with findings in the formulation literature. The SIC shows tremendous potential as a rapid formulation development tool. We also screened two putative interaction sites for involvement in the self‐association of lysozyme and estimated the associated binding energies using a binding isotherm model that we developed. The sites screened consisted of residues 41–48 and 125–128 and were selected based on their apparent importance in forming crystal contacts in several different crystal forms of lysozyme. Of the two sites, only residues 125–128 were found to influence self‐association under the conditions we employed. Because the success of this technique depends on the exploitation of self‐interactions between native species, several important applications are also suggested such as separating native from misfolded or variant species and probing site utilization in aggregation versus crystallization phenomena. © 1996 John Wiley & Sons, Inc.

No abstract

No abstract

We simulated the structure of reversible protein aggregates as a function of protein surface characteristics, protein-protein interaction energies, and the entropic penalty accompanying the immobilization of protein in a solid phase. These simulations represent an extension of our previous work on kinetically irreversible protein aggregate structure and are based on an explicit accounting of the specific protein-protein interactions that occur within reversible aggregates and crystals. We considered protein monomers with a mixture of hydrophobic and hydrophilic surface regions suspended in a polar solvent; the energetic driving force for aggregation is provided by the burial of solvent-exposed hydrophobic surface area. We analyzed the physical properties of the generated aggregates, including density, protein-protein contact distributions, solvent accessible surface area, porosity, and order, and compared our results with the protein crystallization literature as well as with the kinetically irreversible case. The physical properties of reversible aggregates were consonant with those observed for the irreversible aggregates, although in general, reversible aggregates were more stable energetically and were more crystal-like in their order content than their irreversible counterparts. The reversible aggregates were less dense than the irreversible aggregates, indicating that the increased energetic stability is derived primarily from the optimality rather than the density of the packing in the solid phase. The extent of hydrophobic protein-protein contacts and solvent-exposed surface area within the aggregate phase depended on the aggregation pathway: reversible aggregates tended to have a greater proportion of hydrophobic-hydrophobic contacts and a smaller fraction of hydrophobic solvent-exposed surface area. Furthermore, the arrangement of hydrophobic patches on the protein surface played a major role in the distribution of protein contacts and solvent content. This was readily reflected in the order of the aggregates: the greater the contiguity of the hydrophobic patches on the monomer surface, the less ordered the aggregates became, despite the opportunities for rearrangement offered by a reversible pathway. These simulations have enhanced our understanding of the impact of protein structural motifs on aggregate properties and on the demarcation between aggregation and crystallization.

We have simulated the structure of kinetically irreversible protein aggregates in two-dimensional space using a lattice-based Monte-Carlo routine. Our model specifically accounts for the intermolecular interactions between hydrophobic and hydrophilic protein surfaces and a polar solvent. The simulations provide information about the aggregate density, the types of inter-monomer contacts and solvent content within the aggregates, the type and extent of solvent exposed perimeter, and the short- and long-range order all as a function of (i) the extent of monomer hydrophobic surface area and its distribution on the model protein surface and (ii) the magnitude of the hydrophobic-hydrophobic contact energy. An increase in the extent of monomer hydrophobic surface area resulted in increased aggregate densities with concomitant decreased system free energies. These effects are accompanied by increases in the number of hydrophobic-hydrophobic contacts and decreases in the solvent-exposed hydrophobic surface area of the aggregates. Grouping monomer hydrophobic surfaces in a single contiguous stretch resulted in lower aggregate densities and lower short range order. More favorable hydrophobic-hydrophobic contact energies produced structures with higher densities but the number of unfavorable protein-protein contacts was also observed to increase; greater configurational entropy produced the opposite effect. Properties predicted by our model are in good qualitative agreement with available experimental observations.