Search

On BL32XU, a microfocus beamline at SPring-8, oscillation data are collected with typical horizontal beam size of 1 μm. Hence it requires very accurate crystal centering, which is difficult especially for invisible crystals e.g. LCP crystals. Therefore, we perform raster diffraction scan to find crystal positions based on their diffractivity using low-dose exposure. It had been time consuming process due to two reasons; i) slow readout time of CCD, ii) manual inspection of hundreds of diffraction images. To tackle this problem, we installed new fast-readout CCD detector, MX225HS (Rayonix, L.L.C.), and developed support tool for raster scan based crystal centering. The tool visually shows possible crystal position on 2D map based on spot populations, and therefore it is named SHIKA (Spot-wo Hirotte Ichiwo Kimeru Application; a Japanese abbreviation which means the application for crystal positioning by picking up spots). SHIKA automatically detects new images when raster scan started and finishes just after raster scan ends. On GUI, user can find and pass the crystal position information to KUMA (a tool suggesting helical data collection strategy with predicted radiation damage) to start data collection immediately. User can also see picked spots on diffraction images with GUI. SHIKA has been developed based on DISTL [1] and modified to be faster and more accurate, especially for LCP crystal which is an important target on BL32XU. SHIKA picks up spots after subtracting smoothed pseudo-background which is a key for better separation of spots and ring-like diffuse background of lipids. Smoothing is time-consuming, but SHIKA now uses GPUs for almost all process including high-speed median filter [2] so that it can be done within ~100 msec. Further development is under way for faster processing. Now SHIKA can be also used on BL41XU, a high flux beamline at SPring-8 with some adjustment for PILATUS3 (Detectris Ltd.) detector.

Protein micro-crystallography is one of the most advanced technologies for protein structure analysis. In order to realize this, an undulator beamline, named BL32XU, was constructed at SPring-8. The beamline can provide beam with size of 0.9 x 0.9 µm and photon flux of 6E10 photons/s. The beam size can be easily changed by users from 1 to 10 µm square with the same flux density. Through three years user operation, we have established several key systems for efficient protein micro-crystallography. One of them is the software for precise positioning of micro-crystals in `raster scan'. SHIKA is a program with GUI which searches diffraction spots in a plenty of low dose diffraction images obtained in raster scan. Finally, it generates 2D map of crystal positions based on the number of spots or spot intensities. Parameters and thresholds in peak search have been empirically optimized for LCP crystals and it provides robust results. Another system is for the data collection strategy. Almost all successful data collections were conducted via `helical data collection' on BL32XU using the line-focused beam. The GUI software, named KUMA, enables estimation of an accumulated dose and suggests suitable experimental conditions for helical data collection. The system is proven to be useful for experimental phasing using tiny LCP crystals of membrane proteins[1-3]. Based on them, the rapid and automatic data collection system using protein micro-crystals is under development. The new CCD detector, Rayonix MX225HS, was installed for faster data acquisition in 10 Hz with the pixel size of 78 µm square. The new SHIKA using GPUs is under development for faster and more accurate crystal alignment. Following this step, KUMA system can suggest experimental conditions for each crystal found on the loop. We also report about the effects of higher dose rate in protein crystallography up to the order of 100 MGy/s. This work was supported by Platform for Drug Discovery, Informatics, and Structural Life Science from the Ministry of Education, Culture, Sports, Science and Technology, Japan.

The absorption of X-rays which pass through the protein crystal is possibly the largest source of systematic errors in macromolecular crystallography. Therefore we are developing protein crystal processing system using Pulsed UV Laser Soft Ablation (PULSA) technique [1] to reduce the systematic error as well as background scattering from cryoprotectant agents. For high-quality diffraction data collection from organic material, crystals are usually processed to spherical shape in order to keep X-ray path length in crystal constant. This dramatically reduces systematic errors caused by `absorption of X-rays'. Although shaping crystal was thought to be effective for protein crystallography, there was no usual technique to achieve this because protein crystals are exceedingly fragile against mechanical stress. We are developing protein crystal processing system using PULSA technique. In this system, short pulsed UV-laser (maximum power: 1.0 μJ/pulse, wavelength: 193.4 nm, duration: less than 1.3 nsec) is raised by NSL-193L (Nikon Corporation) and focused on 4 μmφ (FWHM). The focused laser is controlled by galvanomic mirror system and irradiates a sample. Combining this mirror system with four-axis goniometer enables to process crystal to arbitrary shape that is easily defined on GUI. Several protein crystals have been successfully processed into spherical, column and square pole shape, etc. In the case of crystal processed into column shape (diameter is 50 μm), in addition to reducing absorption effects, signal-noise ratio of diffraction data can be increased by removing cryoprotectant agent surrounding the crystal. This work was supported by "Platform for Drug Discovery, Informatics, and Structural Life Science" from MEXT, Japan.