Book of Abstracts: Albany 2011

category image Albany 2011
Conversation 17
June 14-18 2011
©Adenine Press (2010)

Largescale Protein Properties: Crystallization, Amyloid Formation, and the Stability of The Proteome

We are interested in largescale protein behaviors. By largescale, we mean processes such as amyloid aggregation or protein crystallization that involve multiple proteins associating into complexes, or collective properties of the thousands of proteins in whole proteomes. There are many experimentally measurable properties for which some basic insights do not require large computer simulations of atomically detailed models.

Protein crystallization. Why do proteins crystallize? Under typical crystallization conditions, two protein molecules have charges of the same sign, so they should repel. So, the usual explanation for protein crystallization is that proteins also have a neutral sticking energy that overcomes the charge-charge repulsion. However, more is needed in order to account for the strong observed dependence of crystal stability on salt, which is of practical importance for crystallizing proteins. We treat the crystallization of proteins, such as lysozyme, as an association of charged spheres in the presence of salts (1). The model is in good agreement with lysozyme crystal solubility data as functions of temperature, pH, and salt. A key conclusion is that because the crystal is macroscopic, it must be electroneutral, so: (a) increasing the protein’s charge has little effect on its crystal stability because counterions are sequestered in proportion to the charge, and (b) the reason that crystals melt upon heating is because it melts the counterion `glue’, not because it melts the protein-protein interactions.

Amyloid aggregation. What are the forces of amyloid aggregation? We consider 3 states in equilibrium: monomeric amyloid peptide molecules, oligomers (micelle-like loose clusters of a few peptide chains), and fibrils (ordered fibers of many chains). We suppose the chains are driven by hydrophobic interactions into the oligomeric state and by additional interactions due to steric zipping (packing plus hydrogen bonding) into the fibrillar state (2). The model predicts two transitions, monomer to oligomer, and oligomer to fibril, the latter of which, interestingly, is predicted to be essentially independent of peptide concentration. The model predicts that if the oligomers are the toxic species, then the fibrils are `good guys’, because they soak up oligomeric chains and buffer their concentration. The model resolves an experimental puzzle from two research groups regarding denaturant disruption. Slightly different peptide concentrations lead to either stable oligomeric intermediates, or none. Model predictions are in good agreement with dependences on salt and pH.

Proteome stability. We compute the stabilities of all the proteins in various proteomes, including Ecoli, worm and yeast based on a simple thermodynamic parameterization of the database of known protein stabilities (3). We find that proteomes are marginally stable. That is, even though thousands of proteins have a stability averaging around 6 kcal/mol, about 650 of Ecoli’s proteins are less stable than 4 kcal/mol. Stability distributions such as these are useful for computing the temperature at which the whole proteome denatures, which coincides closely with the experimental temperatures of cell death.


  1. J. D. Schmit and K. A. Dill. The Stabilities of Protein Crystals. J. Phys. Chem. B 114, 4020--4027 (2010).
  2. J. Schmit, K. Ghosh and K. A. Dill. What drives amyloid molecules to assemble into oligomers and fibrils? Biophysical Journal 100, 450--458 (2011).
  3. K. Ghosh and K. A. Dill. Cellular proteomes have broad distributions of protein stability. Biophysical Journal 99, 3996--4002 (2010).

Ken A Dill1
Kings Ghosh2
Jeremy Schmit3

1Stony Brook University
2Denver University
3University of California at San Francisco

Ph: (631) 632-4132
Fx: (631) 632-4125