Crystallization, followed by subsequent structure determination, is a major step in understanding the structure-function relationship of macromolecules. Understanding macromolecule structure has become a key part in the development of new pharmaceuticals, and is a major area of NIH research. Crystallization however is also the rate limiting step, despite technological efforts to automate the set-up and crystallization data acquisition processes. Macromolecule crystallization conditions are arrived at by screening experiments, where the target material is subjected to typically hundreds to even thousands of different chemical cocktails. In most cases screening experiments fail as they do not result in a crystal. We propose that the experiments contain useful information about the targets behavior in response to the imposed conditions, and that the results can be analyzed to extract the relevant parameters for guiding subsequent crystallization trials. No screen or group of screens can systematically cover the chemical space for protein crystallization, and we hypothesize that in the absence of clear positive hits scored results can be analyzed to determine these factors. As a test of this a preliminary screening results analysis program was written and tested using three hyperthermophile proteins, one each a very easy, moderate, and a very difficult crystallizer. Characterizations were on the basis of each proteins behavior in a single crystallization screen. The analysis is called the Associated Experimental Design (AED) approach. The analysis identified the most significant factors and a 96 condition screen based on those factors was prepared for each protein and set up. Crystals were obtained for all three proteins, and none of the crystallization conditions duplicated those in the original screen. Our Phase I goal is to further improve the preliminary AED software. The Phase I improvements are to include up to three or four screens in the analysis function, to treat salt anions and cations separately in the analysis, to output a 96 condition screen for each protein composed of the found significant factors for that protein, and to test a revised scoring scheme. For simplicity th same screens will be used throughout the Phase I effort. A pool of test proteins will be employed, 1/3 each being classified as easy, moderately difficult, and difficult crystallizers based upon their performance in screening trials. Each iteration of the software will be used with the test proteins original scoring data and optimization screens set up. The optimization results will be fed back to further guide the AED software development. Proteins classified as difficult are anticipated to require two or more optimization rounds. Based on the preliminary results the AED method shows considerable promise. A major advantage of this approach is that it will fit into existing practice, making use of the data already generated in crystallization screening. Success with this approach will increase the number of hits generated and greatly reduce the time and effort required for macromolecule crystallization.
Successful crystallization and X-ray data analysis provides important three-dimensional information on the macromolecules structure-function relationship, important to the design of pharmaceuticals. Screening experiments to identify crystallization conditions typically return non crystalline results. This proposal is to develop software for the analysis of screening data scores to identify likely significant crystallization factors, providingan analytical basis for subsequent experiments, and thereby increasing the chances of success.
| Dinc, Imren; Pusey, Marc L; Aygun, Ramazan S (2016) Optimizing Associative Experimental Design for Protein Crystallization Screening. IEEE Trans Nanobioscience 15:101-12 |
