Date Published: October 01, 2017
Publisher: International Union of Crystallography
Author(s): Gergely Papp, Christopher Rossi, Robert Janocha, Clement Sorez, Marcos Lopez-Marrero, Anthony Astruc, Andrew McCarthy, Hassan Belrhali, Matthew W. Bowler, Florent Cipriani.
New compact and precise cryogenic sample holders for macromolecular crystallography are proposed as possible future European standards.
With the emergence of cryocrystallography (Teng, 1990 ▸) as a standard technique in macromolecular crystallography (MX), various sample holders for protein crystals were developed or adapted from existing supports for crystallographic cryogenic data collection (Garman & Owen, 2006 ▸). The ‘top-hat’ design, exemplified by the Hampton Research Magnetic Base, has proved to be highly successful and over many years has evolved into several similar designs that were subsequently standardized for the needs of robotic sample mounting (Cohen et al., 2002 ▸; Karain et al., 2002 ▸; Snell et al., 2004 ▸; Cipriani et al., 2006 ▸). Among them, the European SPINE standard was established in 2005 as an evolution of existing commercial cap-and-vial models. This standard played a key role in beamline automation in Europe and made it possible to collect data at different European beamlines with minimal compatibility problems (Beteva et al., 2006 ▸). Nevertheless, as with all existing sample-holder standards, the SPINE standard has become a limiting factor at high-throughput beamlines.
The first phase of the development process focused on the NewPin sample holder, which is a simple pin of 22 mm in length and 1.9 mm in diameter. Its design fully meets the requirements for high storage density (36 samples per puck and up to 64 for the model anticipated for fully automated robotic pin handling) with a repositioning precision of 10 µm (Papp et al., 2017 ▸). Nevertheless, after the initial evaluation it became clear that an intermediate version that was easier to handle manually and to integrate at beamlines would be required. Therefore, the miniSPINE sample holder was developed: a compact version of the SPINE sample holder that provides high storage density (36 samples per puck) and is easier to integrate on existing beamlines. The type of crystal mount (i.e. the loop or support that will hold the crystal) of the sample holders is not defined as they can receive any commercially available nylon loop/LithoLoop (Molecular Dimensions, Suffolk, England) or MicroMounts (MiTeGen, Ithaca, USA) or can be customized for specific requirements, such as for the CrystalDirect harvester (Cipriani et al., 2012 ▸; Zander et al., 2016 ▸). Attempts were made to design a vial for both new sample-holder types. Different vial-to-pin coupling methods were explored but insoluble handling problems, as well as the difficulties anticipated in manufacturing the vials at an affordable cost, led this option to be abandoned. Consequently, a closed robot gripper that acts as a cold buffer was developed to keep the crystals below 100 K during transfers in ambient air and to protect them from ice contamination. This gripper is compatible with both the NewPin and miniSPINE sample holders. Corresponding storage pucks and manual handling tools were also developed for the NewPin and miniSPINE sample holders. Both pucks can store 36 sample holders, leading to an increase in sample density in the widely used CX100 dry-shipping dewars by a factor of five versus the SC3-SPINE pucks (Cipriani et al., 2006 ▸) and of more than two versus uni-pucks. The uni-puck footprint standard (http://smb.slac.stanford.edu/robosync/Universal_Puck) was adopted to ensure backwards compatibility with uni-pucks and to facilitate the migration of sample changers already installed at European MX beamlines, such as CATS (Jacquamet et al. 2009 ▸), G-Rob (Ferrer et al., 2013 ▸), BART (Diamond Light Source, England) and ACTOR Rigaku systems (http://smb.slac.stanford.edu/robosync/). A specific dewar slot that is compatible with uni-pucks and that ensures precise positioning of the NewPin and miniSPINE pucks has also been developed. Furthermore, to offer an interoperable line of sample holders and pucks, we decided to slightly modify the SPINE standard, creating the SPINEplus sample holder and a corresponding sample-storage puck. The three new supports and associated pucks are shown in Fig. 1 ▸. The three sample-holder models are designed to be pre-oriented in specific storage racks to be further manipulated with known orientation (see §2.4).
Manual handling of the SPINEplus, miniSPINE and NewPin sample holders as well as of the related manipulation tools has been tested during the development phase. Once selected for the first implementation of a future sample-holder standard, the miniSPINE system was tested at 12 different partner sites using evaluation kits (Supplementary Fig. S5). Its design was then upgraded following the feedback received. In parallel, all of the models have been tested under real conditions (crystal harvesting, automated sample mounting and data collection) at the ESRF–EMBL–India BM14 beamline using a FlexED8 sample changer (Papp et al., 2017 ▸).
Here, we only report on the manual usability of the sample holders. The results related to robotic sample transfers are published in the accompanying article (Papp et al., 2017 ▸) as they depend on the robot grippers, the transfer times and the precision of the robotics used. The SPINEplus sample holders were found to be easy to manipulate as they are identical in size to the sample holders commonly used at beamlines. The miniSPINE harvesting tool was judged to be convenient. However, owing to the angle between the pin and the tool, some users found it more difficult to use than the usual straight tools that are used to handle the SPINE sample holders where the orientation of the loop can be easily selected. A major problem was encountered with NewPin and miniSPINE pucks when inserting the pins with a crystal mounted on them into a puck under liquid nitrogen. Owing to the small size of the wells, it can be difficult to know whether a position is occupied or free. An additional difficulty when using NewPin was to properly orient the pins to plug them correctly into the puck slots. This made it clear that NewPin should be reserved for fully robotic systems, including the crystal-harvesting step. For miniSPINE, the puck-loading assistant (Fig. 4 ▸c) greatly improved the manual harvesting process (Supplementary Video S1).
Two new sample holders for macromolecular cryocrystallography have been developed together with corresponding pucks and robotic and manual handling tools to enhance crystal-processing times at beamlines and to reduce the handling effort and transportation costs. miniSPINE allows the storage of 36 samples in a puck. NewPin offers a storage density of up to 64 samples per puck combined with highly accurate positioning on goniometers. The two models can be handled with the same robot gripper and use pucks with the same footprint. They fit in a specific uni-puck-compliant dewar slot and in canisters compatible with CX100 shipping dewars. A goniometer mount compatible with SPINE and miniSPINE was also developed to facilitate the integration of miniSPINE at beamlines. The NewPin model requires a specific goniometer mount and is more difficult to handle manually. It is adapted to fully automated pipelines covering all of the steps from crystal harvesting to processing at beamlines. The maximum storage density of the NewPin pins could potentially exceed 64 pins per puck as it depends on the space allocated around the pins for handling. This is related to the outer dimensions of the gripper and to the precision of the handling robotics. Finally, a modified SPINE sample holder called SPINEplus, which is backwards-compatible with SPINE, was developed together with a miniSPINE/NewPin dewar slot compliant puck. This interoperable line of sample holders and pucks should facilitate the transition from the current SPINE sample-holder standard to miniSPINE and further to NewPin, in particular on beamlines that are equipped with flexible sample changers based on six-axis industrial robots and tool changers. Initially, sample tracking was identified as highly important when moving to the densities proposed here; therefore, both pins and pucks are ready to receive RFID tags. Work is continuing to improve the reliability of the RFID tags used to identify the pins so that they can resist the extreme thermal cycling experienced during their lifetime. The identification of the pucks also includes Datamatrix and human-readable labels. Defined as an important feature at the beginning of the project, the identification of the pins was over time judged to be less important than the identification of pucks. In practice, even when available, SPINE-standard pin identification is rarely used, probably because of the additional effort needed during crystal harvesting. Currently, sample tracking almost exclusively relies on the position of the pins in human-identifiable pucks. Furthermore, as automation becomes more extensive, the potential for human error decreases. In the near future, projects where individual sample tracking is important should benefit from automated harvesting and storing, making pin identification unnecessary.