High Throughput Precise Protein Stability Measurements Facilitated By Robotics

Marshall Edgell ^{‡ ¬}, Dorothy Sims^‡, Gary Pielak^§¬, Fang Yi^¬

‡Department of Microbiology and Immunology, ^§Department of Chemistry, ^¬Department of Biochemistry and Biophysics, University of North Carolina at Chapel Hill, 27599

Abstract:

Protein stability (DG_{N-D, H2O}) derived from fluorescence monitored solvent denaturation is a standard metric to study the relationship between protein sequence and stability. The complexity of the interactions for a single residue in a protein implies an anecdotal nature for testing models about protein stability using a few mutants. To overcome this, we chose to employ statistical modeling as a formalism for model testing, which requires stability data from a large number (hundreds) of protein mutants. Simulation studies indicate that the variance introduced into model fitting by DG_{N-D, H2O} increases in proportion to the square of measurement error. A commonly reported error range for DG_{N-D, H2O} derived from solvent denaturation (0.1- 0.6 kcal/mol) accounts for 2% to 50% of the variation in stability behavior of the mutants, an observation that poses a need to obtain high precision stability data. This is a challenging task especially with the added stipulation of high throughput. Excel based simulations were used to dissect the measurement error from each step. The precision of DG_{N-D, H2O} data depends most on the precision of the titrating fluorometer, followed by protein purity and then the capacity to precisely prepare the solutions used for solvent denaturation. To achieve the desired precision, we have developed a robotics facilitated approach which also supports high throughput. A laboratory automation workstation (Beckman Biomek 2000) was used to purify his-tagged mutant proteins in 96-well format, and to prepare mutant protein solutions for solvent denaturation carried out on a semi-automated titrating fluorometer (Protein Solutions). The variation in protein concentrations of solutions prepared by the robot was determined by weighing. For eglin c we can attain a throughput of stability measurements of about 20 mutants per day. The use of robotics for protein purification and solution preparation gives highly precise DG_{N-D, H2O} values in which the standard deviation of values from multiple protein preparations (± 0.05 kcal/mol) differs very little from multiple measurements from a single preparation (± 0.04 kcal/mol). These are the most precise stability measurements that have ever been reported in the literature. This approach can be applied to characterize combinatorial mutant libraries built around any protein. The high precision stability data from these measurements serve not only to parameterize models concerning protein stability determinants, also they can be used as valuable test data for computational biophysicists to design programs to predict protein stability and structure based on protein sequence information.