Date Published: January 01, 2016
Publisher: International Union of Crystallography
Author(s): Elizabeth L. Baxter, Laura Aguila, Roberto Alonso-Mori, Christopher O. Barnes, Christopher A. Bonagura, Winnie Brehmer, Axel T. Brunger, Guillermo Calero, Tom T. Caradoc-Davies, Ruchira Chatterjee, William F. Degrado, James S. Fraser, Mohamed Ibrahim, Jan Kern, Brian K. Kobilka, Andrew C. Kruse, Karl M. Larsson, Heinrik T. Lemke, Artem Y. Lyubimov, Aashish Manglik, Scott E. McPhillips, Erik Norgren, Siew S. Pang, S. M. Soltis, Jinhu Song, Jessica Thomaston, Yingssu Tsai, William I. Weis, Rahel A. Woldeyes, Vittal Yachandra, Junko Yano, Athina Zouni, Aina E. Cohen.
A high-density sample mount has been developed for efficient goniometer-based sample delivery at synchrotron and XFEL sources.
As structural biologists tackle ever more challenging systems, the development of efficient methods to deliver large quantities of crystals for X-ray diffraction studies is increasingly important. Proteins that are difficult to crystallize will often produce only small crystals that yield only a few degrees of diffraction data before succumbing to the damaging effects of radiation exposure. For many systems, obtaining a complete data set to high resolution using very small crystals is possible through the use of microfocus synchrotron beams and the collection and combination of data from multiple crystals. For example, when solving the structure of the β2 adrenergic receptor–Gs protein complex, hundreds of microcrystals were screened and 20 microcrystals were used in the final data set (Rasmussen et al., 2011 ▸). The structural information accessible from small or radiation-sensitive crystals may be extended through the application of femtosecond crystallography (FX), an emerging method that capitalizes on the extremely bright, short time-scale X-ray pulses produced by X-ray free-electron lasers (XFELs). This approach exploits a ‘diffraction before destruction’ phenomenon (Neutze et al., 2000 ▸) where a still diffraction pattern is produced by a single X-ray pulse before significant radiation-induced electronic and atomic rearrangements occur within the crystal (Barty et al., 2011 ▸; Lomb et al., 2011 ▸). Since the area of the crystal exposed to the X-ray pulse is completely destroyed after each shot, multiple crystals are required for these experiments (Chapman et al., 2011 ▸). In addition, FX provides a means to determine catalytically accurate structures of radiation-sensitive metalloenzymes (Kern et al., 2013 ▸) which may undergo structural rearrangement upon photo-reduction of the metal center at a synchrotron (Peters et al., 1998 ▸; Yano et al., 2005 ▸; Corbett et al., 2007 ▸; Meharenna et al., 2010 ▸; Davis et al., 2013 ▸). In most cases, these experiments also require a large quantity of crystals as each area of the crystal can only be exposed once.
The grid scaffold consists of a piece of 100 or 200 µm thick polycarbonate plastic with laser-cut rows of holes (or ports). This polycarbonate scaffold is affixed to a standard magnetic base to produce the ‘grid assembly’ (Fig. 1 ▸a). A specialized bonding jig is used to hold the polycarbonate scaffold inside the magnetic base as the epoxy sets (Fig. 1 ▸b). Grid ports may hold either large crystals or groups of smaller crystals in known locations. The current grid layout has 75 ports of 400, 200 and 125 µm in diameter (Fig. 1 ▸d); however, the size and arrangement of ports may be altered to better suit different experimental setups. Since crystals are held in known locations, rapid and precise automated crystal positioning into the X-ray beam path is possible.
Grids may be manually filled with crystals. To view crystals during this process, it is helpful to position the grid assembly underneath a microscope using a magnetic holder (Supplementary Fig. S2). Grid ports should be prefilled with cryoprotectant oil such as Paratone-N or paraffin to prevent crystal dehydration. A fine needle may be used to apply oil to each grid port. A cryo-loop may be used as a tool to remove a crystal from the crystallization tray, coat the crystal with a thin layer of oil and then transfer it to an appropriately sized grid port. It is helpful to match the size of the cryo-loop tool to a port size in the grid. Filling all ports in a grid may be impractical because crystals may degrade over time in the cryoprotectant oil. Testing is necessary to determine the maximum timeframe for filling grids with a particular sample and oil. This may be accomplished by filling a grid with crystals and recording the loading time for each port. Diffraction data may then be collected and compared for crystals with known exposure times to the oil. Detailed instructions for grid usage may be found at http://smb.slac.stanford.edu/hardware/sample_mounting_grids/.
The addition of a thin polymer film to one face of the grid creates a scaffold for crystallization experiments. To fill grids with crystallization solutions for this purpose, an adaptor was developed that holds a grid in the destination-plate position of liquid-handling robots (Fig. 2 ▸a). A neoprene-lined torsion clip grips the magnetic base of the grid assembly and holds it in place. Grids are indexed against two metal surfaces to ensure accurate reproducible drop placement (Fig. 2 ▸b). The adaptor has been successfully tested with the Labcyte Echo 550 (Fig. 2 ▸c), the Art Robbins Gryphon (Supplementary Video S1) and the TTP Labtech Mosquito (Supplementary Video S2). After the sample has been deposited on grids, the grids may be incubated in specialized crystal-growth containers which support hanging-drop or sitting-drop experiments and lipidic cubic phase (LCP) crystallization experiments. Liquid-handling robots may also be used to load crystal suspensions onto grids.
A grid vapor-diffusion chamber was developed to hold a grid in a controlled environment for incubating sitting-drop or hanging-drop crystallization experiments (Fig. 3 ▸a). The chambers are capped with transparent X-seal crystallization supports so that crystal growth may be monitored with a microscope. Silicone O-rings with a thin film of vacuum grease are used to ensure a tight seal around the grid scaffold (Fig. 3 ▸b). A well in the base of the chamber holds up to 350 µl of desiccant below the grid.
LCP crystallization experiments may also be performed on grids with the use of a specialized tray, eliminating the need to cut open a glass sandwich plate and manually harvest crystals for data collection. The grid LCP tray assembly consists of two siliconized glass slides, 1 mm thick double-sided tape and a tray with a support for a magnetic base and glass slides. The grid is sandwiched between the two sheets of glass and is surrounded by precipitating agent (Fig. 4 ▸a). To aid in removal, a thin polycarbonate sheet may be laid on top of the grid and under the top glass plate.
Grids have been used in combination with goniometer-based instrumentation installed at the LCLS-XPP instrument (Cohen et al., 2014 ▸) for XFEL diffraction experiments. During these experiments, the Stanford Automated Mounter (SAM) robot (Cohen et al., 2002 ▸) was used to mount crystals inside grids onto the goniometer. To control these experiments, automated routines were added to the Blu-Ice/DCSS experimental control software (McPhillips et al., 2002 ▸; Cohen et al., 2014 ▸). To define the position of all grid ports in relation to the X-ray beam position, a semi-automated alignment procedure takes advantage of the predefined spatial arrangement of the laser-cut grid ports. This process begins by first defining the position of the edge of the grid by rotating it until it is edge-on in the software video display and then clicking on the video image of the grid to move the edge of the grid into the X-ray beam position. If the grid is tilted in this view, two positions may be identified to define the translation path. Next, the grid is rotated by 90° to put the face-on view of the grid in the video image. Four ports on the outer corners of the grid are then clicked in a specified order, which act as fiducial markers to define the port locations and the grid rotation (Supplementary Fig. S3; Fig. 1 ▸d). This procedure calibrates all of the grid ports to the coordinate system of the goniometer and the beam-interaction region.
Data collection for grids is also supported at the SSRL macromolecular crystallography beamlines, and a tab for grid data collection has been added to the Blu-Ice/DCSS experimental control software (McPhillips et al., 2002 ▸). Similar to data collection at LCLS, a semi-automated alignment procedure is performed for each grid, after which data collection may proceed. Similar options for automated data collection supported at LCLS also work at SSRL (Supplementary Fig. S3), with the addition of options to collect oscillation data at each crystal position and during X-ray rastering. This automation may be further incorporated into automated workflows (Tsai et al., 2013 ▸) where low-dose X-ray rastering may be used to locate crystals within grid ports for automated multi-crystal data-collection strategies.
Grid ports may be covered with a thin polymer film to prevent evaporation for room-temperature data collection. This approach was used for room-temperature screening of light-sensitive photosystem II (PSII) crystals at LCLS-XPP. Polycarbonate backing was glued to one face of a grid with epoxy. A suspension of PSII crystals was pipetted over the open grid ports, and a loop was used to drag crystals into grid ports (Fig. 5 ▸a). A second sheet of polycarbonate was placed over the open grid ports and held in place by capillary action (Fig. 5 ▸b). Grids were then hand-mounted on the goniometer for data collection at XPP. To avoid exposure of the light-sensitive PSII crystals to ambient light, data collection was performed in the dark. The crystals were substantially smaller (∼20–30 µm) than the grid ports and were difficult to visualize under low light conditions. Therefore, data were collected by rastering the individual 400 µm ports. In a first round of experiments a total of 500 images were collected from three grids. After crystal suspension conditions and data-collection parameters had been optimized, 280 images were collected from a single grid in 18 min and minimal evaporation was observed. Out of this set 87 images contained diffraction data, and the best diffraction was observed to better than 2.5 Å resolution (Fig. 5 ▸c).
Liquid-handling robots may also be used to dispense suspensions of microcrystals into grids. A suspension of mouse perforin crystals was loaded into grids for screening at LCLS-XPP with the use of an Echo 550 liquid-handling robot. Perforin crystals were prepared as described previously (Law et al., 2010 ▸). The slurry was centrifuged, the supernatant was poured off and the pellet was resuspended in cryprotectant. An Echo 550 liquid-handling robot was used to transfer the suspension onto grids with polycarbonate backing. 30 nl droplets of the crystal suspension were deposited on the surface of the polycarbonate backing inline with each grid port (Fig. 6 ▸a). The grids were immediately flash-cooled and used to screen perforin crystals at LCLS-XPP in December 2014 (Fig. 6 ▸b).
To test grids as scaffolds for crystal growth, we set up sitting-drop vapor-diffusion experiments on grids with lysozyme. An Echo 550 liquid handler was used to dispense drops of protein and precipitant solution onto a grid with a thin polycarbonate backing. Grids were then incubated in a vapor-diffusion chamber drop-side up with desiccant for 5 d. Lysozyme crystals grew on the grids (Fig. 3 ▸c), and the Echo 550 liquid handler was used to dispense drops of cryoprotectant on top of the crystallization drops.
Proof of principle LCP crystallization experiments were performed on grids using an adaptation of a protocol for the growth of lysozyme in LCP (Aherne et al., 2012 ▸). An Art Robbins Gryphon was used to dispense cubic phase into grid ports (Supplementary Video S1), and the grids were incubated with precipitating agent in a glass sandwich (Fig. 4 ▸b). Lysozyme crystals up to 50 µm wide were observed in grid ports after 16 h of incubation.
High-density sample-mounting devices dramatically reduce the amount of time needed for multi-crystal data collection. Data collection of crystals held in loops requires that each crystal be individually mounted on the goniometer, centered in the X-ray beam and then dismounted. Automated alignment of crystal-containing loops can take between 15 and 30 s each, depending on the beamline hardware and software (Sharff, 2003 ▸; Snell et al., 2004 ▸; Pothineni et al., 2006 ▸; Jain & Stojanoff, 2007 ▸; Song et al., 2007 ▸). The time required to mount and dismount a sample varies; however, even if sample exchange requires 25 s, sample exchange, alignment and exposure for 1000 crystals in loops would consume at least 12 h of beam time. Furthermore, 1000 samples mounted individually in loops would require 11 SSRL cassettes or 62 uni-pucks for storage. The sample-mounting grid enables up to 75 conventionally sized crystals (∼100–300 µm in diameter) to be mounted on the goniometer at once, or many thousands of microcrystals, circumventing much of the time involved in sample exchange. Time is further saved because alignment, which takes about 30 s to 1 min depending on the operator, is performed once for the entire grid, after which the position of each sample location is automatically calculated by the Blu-Ice/DCSS control software (Cohen et al., 2014 ▸). When conventionally sized crystals are used, the use of grids to screen 1000 crystals would reduce the time spent on sample exchange and alignment from about 12 h at best to under 1 h, and a single uni-puck would be sufficient to hold 1000 samples. Additionally, a single grid port may also be filled with multiple microcrystals, in which case thousands of crystals may be mounted on the goniometer at once. In practice, the efficiency of data collection with grids will depend on a number of factors, including the detector readout speed, the number of crystals in each grid, the type of data collection performed and the skill of the user.
The use of high-density containers such as grids that hold crystals in known locations enables efficient highly automated data-collection strategies. Adaptors and specialized grid holders allow crystallization experiments to be performed inside grids, bypassing the tedious step of harvesting and mounting crystals. This approach is particularly attractive for very small and delicate crystals.
The following references are cited in the Supporting Information for this article: Caffrey & Cherezov (2009 ▸), Hellmich et al. (2014 ▸) and Pullara et al. (2013 ▸).