Development of Protein Polycrystallography
Robert B. Von Dreele
IPNS/APS, Argonne National Laboratory, Argonne, IL 60439
The solution and refinement of structures with diffraction data from polycrystalline samples has undergone a revolution in the last two decades: once a phase identification tool used primarily in the industrial setting, this technique currently is a cornerstone for fundamental investigations of new and complex materials. In fact, it’s difficult to imagine progress in such diverse arenas from new classes of superconductors to small organic molecule structure solution without modern polycrystalline or “powder” diffraction science. This report describes the frontier of powder diffraction as it begins to make a significant impact on structural biology.
The reason that powder diffraction data from proteins and protein complexes have such a high potential is fourfold: 1) powders are easily prepared under a wide range of conditions and not at all subject to the vagaries and restricted circumstances required for growing suitable single crystals, 2) protein powders are inherently “perfect” for diffraction; they are of the right size (about 1 micron) and are almost completely strain free, 3) only very small samples sizes ( mg quantities) are required, and 4) data can be collected rapidly, making the observation of many dynamic processes straightforward.
The diffraction pattern that results from polycrystals consists of a series of rings (Figure 1) that are the superposition of the individual single-crystal diffraction patterns from the entire ensemble of a very large number (e.g. ~10 9/mm 3 for 1 mm crystallites) of randomly oriented crystallites. This pattern can display considerable sensitivity to subtle structural changes, typified by shifts in the diffraction peak positions and changes in intensity resulting in readily discernable changes in the powder diffraction profile.
Figure 1. Powder diffraction pattern of hen egg white lysozyme (HEWL) obtained at the 1-BM line at APS with 0.619A radiation in a 40s exposure. Crosses mark ring used to determine image center (+).
The development of a procedure, the Rietveld method, for curve fitting of the this powder diffraction profile, and thus extracting the maximum information, has revolutionized the use of powder diffraction for crystal structure analysis. Until recently, protein crystal structures were considered to be far too complex for any serious attempt at examining them by powder diffraction. However, by utilizing high-resolution (i.e. narrow instrumental diffraction line width) synchrotron radiation, we have quickly achieved three major milestones in the development of protein polycrystallography. The first of these was a successful refinement of the crystal structure of metmyoglobin (1). A second milestone was a demonstration that a protein structure could be solved from powder data by molecular replacement; the main features of the result on a new Zn-insulin phase were confirmed in an independent single crystal structure analysis (2). Thirdly, we showed that powder diffraction could be used to detect the formation of a protein-ligand complex and reveal the mode of ligand binding (3). In all cases diffraction patterns gave peak widths close to the inherent resolution of the diffractometer for a pattern that extended over the approximate d-spacing range of 40Å to 3Å. Detailed analysis of the diffraction line shapes revealed that the microcrystalline protein material was, in each case, virtually ideal material for powder diffraction, displaying little or no line broadening disorder and an approximate 1 mm crystallite size giving diffraction line widths of ~0.01 o 2 Q. Slight modification of the preparation conditions (pH, salt content, etc.) induced anisotropic changes in lattice parameters of 0.2-0.3%, far in excess of the measurement precision.
At line 1-BM at APS we developed a new method of data collection with the incident beam focused onto a MAR345 image plate thus making a low angle Guinier camera (cf. Fig. 1). In these experiments the focusing in the vertical plane is better than that in the horizontal plane. Thus, we perform integration of a "pie" section covering 60-90 o azimuthal angle of the image centered about the vertical and extending to a d-spacing of ~2.0Å. Considering the reduction in data acquisition time and sample size as well as improved counting statistics, the resulting powder diffraction data from a <1mg sample shows 10 4-10 6 improvement in data quality over that from a single point detector with 10mg of material.
This approach completely removes the problem of radiation damage in these materials and allows in situ tracking of processes on minutes-to-hours time scales; examples will be shown in the talk. Preliminary experiments using samples held in 50-200 mm nylon loops suggest that further reductions in sample size by an order of magnitude or more are possible. However, the 300 mm point spread on the image plate and ~200 mm sagittal focus image size gives diffraction peaks which are ~0.035 o 2 Q wide. Future developments in both optics and image plate technology can easily reduce this to ~0.01 o 2 Q matching that obtained from crystal-analyser detector systems.
The extreme sensitivity of a powder pattern to small changes in crystallization conditions provides an opportunity for possible near ab initio structure solution of proteins from powder diffraction data. An intensity extraction process that can use multiple and slightly different powder patterns should yield a suite of structure factors freer of the compromises in accuracy from overlap and may be sufficient for protein crystal determination. This approach may be of use for the very substantial fraction of proteins, especially membrane proteins, which have been very difficult to crystallize. It should be noted that these multiple powder patterns would then be used in a combined Rietveld and stereochemical restraint refinement of the protein structure.
References
(1) R.B. Von Dreele, Jour. Appl. Crystallogr. 32, 1084-1089 (1999).
(2) G.D. Smith, W.A. Pangborn and R.H. Blessing, Acta Crystallogr.D57, 1091-1100 (2001).
(3) R.B. Von Dreele, Acta Crystallogr.D57, 1836-1942 (2001).
