(Steve Almo, Director)
The Structure Core is responsible for high-throughput X-ray crystal structure determination of EFI targets. Through collaboration with all members of the EFI, the Structure Core implements strategies to maximize the number of “dockable entities” (i.e., experimental and homology-based models suitable for computational docking) available for each EFI target. The individual stages of the Structure Core pipeline are modeled after the Protein Structure Initiative (PSI) in order to take advantage of the PSI’s extensive experience maximizing throughput and efficiency. While the infrastructure and organization were inspired by the PSI, the EFI is explicitly concerned with the discovery of function, so extensive additional effort is often made to implement rescue strategies for targets that are the most informative in terms of function.
A typical crystallization experiment for an EFI target begins with a crystallization robot (PHOENIX or GRIFFEN) and uses ~200 µl of sample per protein “state” (e.g., apo vs. liganded, His-tag vs. cleaved) with 384 crystallization conditions. Initial hits with promising morphology are passed on to crystallographers, while hits that are too small or poorly formed are subjected to optimization screens that systematically vary components of the initial crystallization condition (e.g., pH, salt, precipitant). Inclusion of additives is also employed to improve crystal size, morphology, and diffraction resolution. In order to minimize chemical use and increase the number and accuracy of crystal optimizations that can be performed per year, the Structure Core uses a Formulator liquid handling robot (FORMULATRIX) to dispense individual chemicals from chemical stocks in an ordered array into the reservoirs of 96 well sitting drop crystallization plates. Crystallization plates are monitored with a Rock Imager (FORMULATRIX) which holds up to 1000 microplates in a climate controlled incubator and images drops on a user defined schedule. Since a major goal of the Structure Core is to produce structures that are amenable to computational ligand docking more than a single structure “snapshot” may be required to capture an active site in the correct conformation for catalysis. Many proteins will crystallize in the same crystal form under unique crystallization environment (pH range, salt vs. PEG, +/- divalent ions, +/- small molecule additives. Others will crystallize in more than one crystal form; therefore, the Structure Core vigorously pursues all initial crystallization hits in order to identify dockable structures. Hits of sufficient size are tested for diffraction on either an in-house or synchrotron source as appropriate, and promising crystals are subjected to synchrotron data collection either at the NSLS X29 undulator beamline or via the mail-in program of APS LRL-CAT (31-ID).
Given adequate diffraction, structure determination follows different trajectories contingent on if a structure of sufficient similarity has already been solved or if the structure has to be solved de novo. Molecular replacement (MR) is routinely attempted on targets with >30% sequence identity to a known PDB entry, using programs and models best appropriate for a given target. Potential MR solutions are subjected to numerous statistical and crystallographic tests to confirm their veracity, including favorable crystallographic R-factors, crystallographic contacts in three dimensions, strong difference density in electron density maps, etc. De novo structure determination poses a different set of challenges, but the EFI Structure Core has extensive experience in a wide range of approaches. These include classical single or multiple isomorpous replacement (SIR/MIR) as well as phasing based on endogenous metals and sulfur single-wavelength anomalous diffraction (SAD), radiation-induced damage, and halide soaks. Refined structures are then immediately published to the PDB and passed to the respective EFI Bridging Projects and Scientific Cores for docking and mechanistic analysis. Extensive communication between the EFI Bridging Projects and the EFI Structure Core assists in the iterative process of obtaining a useful structure and eventually a liganded structure to validate functional predictions.