Intrinsic Viscosity of Proteins and Platonic Solids by Boundary Element Methods David K. Hahn and Sergio R. Aragon* Department of Chemistry and Biochemistry, San Francisco State UniVersity, 1600 Holloway AVe., San Francisco, California 94132 Received February 15, 2006 Abstract: The boundary element (BE) method is used to implement a very precise computation of the intrinsic viscosity for rigid molecules of arbitrary shape. The formulation, included in our program BEST, is tested against the analytical Simha formula for ellipsoids of revolution, and the results are essentially numerically exact. Previously unavailable, very precise results for a series of Platonic solids are also presented. The formulation includes the optional determination of the center of viscosity; however, for globular proteins, the difference compared to the computation based on the centroid is insignificant. The main application is to a series of 30 proteins ranging in molecular weight from 12 to 465 kD. The computation starts from the crystal structure as obtained from the Protein Data Bank, and a hydration thickness of 1.1 Å obtained in previous work with BEST was used. The results (extrapolated to an infinite number of triangular boundary elements) for the proteins are separated into two groups: monomeric and multimeric proteins. The agreement with experimental measurements of the intrinsic viscosity in the case of monomeric proteins is excellent and within experimental error of 5%, demonstrating that the solution and crystal structure are hydrodynamically equivalent. However, for some multimeric proteins, we observe strong systematic deviations around -20%, which we interpret as a systematic deviation of the solution structure from the crystal structure. A possible description of the structural change is deduced by using simple ellipsoid model parameters. A method to obtain intrinsic viscosity values for proteins to 1-2% accuracy (better than experimental error) on the basis of a single BE computation (avoiding the need for an extrapolation on the number of surface triangles) is also presented. I. Introduction The intrinsic viscosity, [η], is simply the initial fractional slope obtained when the solution viscosity, η, is plotted against the concentration, c: where η o is the viscosity of the pure solvent. Thus, [η] reflects the increase in viscosity brought about by the addition of an infinitesimal amount of solute to a pure solvent. Measurement of [η] provides a simple and inexpensive way of obtaining information about molecular shape in solution, allowing phenomena such as protein denaturation 1 and oligomerization 2 to be examined. Accurate computation of [η] for macromolecules requires microscopic detail in the representation of the molecular surface. To achieve microscopic detail, the surface may be modeled in one of two ways: by a collection of small spherical beads (the hydrodynamic bead model) or by an array of flat triangular platelets (the boundary element, BE, method). While the bead model is well-known and has been previously applied to the computation of [η] for proteins, 3 the problem can be formulated exactly as an integral equation that can be solved more accurately and precisely by the BE method. The BE treatment by Zhou 4 used an approximate expres- sion for [η] accurate to within 2% for ellipsoids with axial ratios between 1 / 4 and 4. Its application to globular proteins * Corresponding author fax: 415-338-2384; e-mail: aragons@sfsu.edu. [η] ) lim cf0 η - η o η o c (1) 1416 J. Chem. Theory Comput. 2006, 2, 1416-1428 10.1021/ct600062y CCC: $33.50 © 2006 American Chemical Society Published on Web 06/30/2006