The goal of this protocol is to demonstrate how to prepare serial crystallography samples for data collection on a high viscous injector, Lipidico, recently commissioned at the Australian synchrotron.
A facility for performing serial crystallography measurements has been developed at the Australian synchrotron. This facility incorporates a purpose built high viscous injector, Lipidico, as part of the macromolecular crystallography (MX2) beamline to measure large numbers of small crystals at room temperature. The goal of this technique is to enable crystals to be grown/transferred to glass syringes to be used directly in the injector for serial crystallography data collection. The advantages of this injector include the ability to respond rapidly to changes in the flow rate without interruption of the stream. Several limitations for this high viscosity injector (HVI) exist which include a restriction on the allowed sample viscosities to >10 Pa.s. Stream stability can also potentially be an issue depending on the specific properties of the sample. A detailed protocol for how to set up samples and operate the injector for serial crystallography measurements at the Australian synchrotron is presented here. The method demonstrates preparation of the sample, including the transfer of lysozyme crystals into a high viscous media (silicone grease), and the operation of the injector for data collection at MX2.
Serial crystallography (SX) is a technique that was developed initially in the context of X-ray Free Electron Lasers (XFELs)1,2,3,4. Though fixed target approaches can be used for SX5,6,7, typically, injector systems are employed to deliver crystals in a continuous stream to the X-ray beam. Because it combines data from a large number of crystals, SX avoids the need for any crystal alignment during the experiment, and enables data to be collected at room temperature8,9. With the aid of a suitable injector, the crystals are streamed one-by-one into the X-ray interaction area and the resulting diffraction data is collected on an area detector9,10. To date, SX has been successful in solving a number of protein structures1,11,12,13 including crystals too small to measure using conventional crystallography. It has also provided new insights into time-resolved molecular dynamics by exploiting the femtosecond pulse duration of the XFEL. By initiating pump-probe reactions with optical laser sources, in-depth studies have been carried out on photosystem II14,15, photoactive yellow protein16,17, cytochrome C oxidase18, as well as bacteriorhodopsin19,20,21. These studies have probed the electron transfer dynamics that occur following light activation demonstrating the significant potential of serial crystallography for understanding time-resolved biological processes.
Development of serial crystallography is also becoming increasingly prevalent at synchrotron sources9,12,20,22,23,24. Synchrotron based SX allows for large numbers of individual crystals to be measured efficiently at room temperature using an appropriate injector system. This approach is suitable for smaller crystals hence, in addition to requiring a fast frame-rate detector to collect the data, a micro-focused beam is also required. Compared to conventional crystallography, SX does not involve the mounting and alignment of individual crystals in the X-ray beam. Because data from a large number of individual crystals is merged, the radiation dose received by each crystal can be substantially reduced compared to conventional crystallography. Synchrotron SX can also be applied to the study of time-resolved reactions, even down to the millisecond regime, provided a detector with a sufficiently high frame rate is available (e.g., 100 Hz or more). Several serial crystallography experiments have been carried out at the synchrotron using injectors that were initially developed at XFEL sources20,22,23. The two most common types of injector are the Gas Dynamic Virtual Nozzle (GDVN)25 and High Viscous Injector (HVI)9,24,26,27,28. The GDVN is ideal for injecting low viscosity, liquid samples, but requires high flow rates to achieve stable streams, which in turn leads to high sample consumption rates. By contrast, HVI’s are suitable for high viscosity samples which allows generation of a stable stream at much lower flow rates, leading to much lower sample consumption. The HVI injector, therefore, favors delivery of samples where a viscous carrier is preferable (e.g., lipid-based for membrane proteins) and/or large quantities of sample are not available. SX injectors are generally challenging to use and require extensive training to operate. They also involve lengthy sample transfer protocols, as the sample needs to be loaded into a specialized reservoir, this generally has a high risk associated with it of sample being lost either in the ‘dead volume’ or via leakages in the connections. Therefore, it is desirable to optimize the injector design to mitigate any losses prior to the sample reaching the X-ray beam.
Recently, the first SX results were published using Lipidico23 with a lysozyme target, using an Eiger 16M detector. This injector design limits sample wastage by minimizing the number of steps involved in going from initial crystallization to the transfer of crystals into the injector followed by the delivery of sample to the X-ray beam. This manuscript describes and demonstrates the sample transfer procedure starting from sample preparation, moving on to the injection process, and finally data collection, using the same crystallization vessel. The operation of the injector is also described.
1. Preparation of crystals in a high viscous media using glass syringes
2. Injector mounting and control
3. Mounting the sample syringe
4. Running the injector
Lipidico is an HVI built as an alternative delivery system for use on MX2 (Figure 1). It is ideally suited for SX where crystals are either grown in lipidic cubic phase or transferred to a high viscous inert media.
To demonstrate the injector application silicone grease mixed with lysozyme crystals was used to collect SX data at the MX2 beamline at the Australian synchrotron. To mount the injector on the MX2 beamline the cryogenic nozzle is removed and replaced by the injector as shown in Figure 1. The syringe samples are mounted directly on the injector and fastened using O-rings (Figure 1B). Data collection can be initiated within minutes after a sample change. The injector is designed for the rapid sample control with a simple setup that is easy to operate for users who are unfamiliar with SX.
Figure 1: Images of Lipidico, a viscous injector used at the Australian synchrotron for SX experiments. (A) Shows a photo of Lipidico incorporated into MX2. The red arrow indicates the direction of the X-ray beam. (B) A closer view of the sample holder region on Lipidico. The syringe is held in place by two O-rings and the sample is collected in the catcher. (D) A close-up view of the sample stream showing the injector needle and the sample waste catcher which can be altered to incorporate a polystyrene/venturi suction catcher seen under the needle in (A). The figure has been adapted from Berntsen et. al. 201923 with the permission of AIP Publishing. Please click here to view a larger version of this figure.
The injector is compatible with various types of highly viscous carriers. The optimal sample running speed for the injector is crucial for the stream stability, too slow will cause a curling effect/stream expansion and too fast will result in the stream breakup. The optimal speed will vary depending on the carrier system used. Figure 2 and Figure 3 show silicone grease tested at different injector speeds and demonstrates how the stream behavior can vary. A slow flow rate (Figure 2A, 2.3 nL/s) results in expansion of the silicone grease extruded from the injector whilst a faster flow rate (Figure 2C, 6.6 nL/s) produces a thinner stream. Curling of the viscous media stream has regularly been observed (Figure 3A). To overcome this issue two innovative solutions were tested: a polystyrene catcher and a venturi suction funnel below the needle. The polystyrene catcher introduces a weak electrostatic force and works best on highly charged samples, as in the case of silicone grease (Figure 3B), while the venturi suction funnel aids in guiding the stream vertically downwards to the catcher, independent of sample charging. Depending on the characteristics of the sample either option can be used successfully.
Figure 2: The effect of a high viscous carrier stream using different injector speeds. 100% silicone grease was tested at different sample velocities to assess its flow characteristics. (A). 2.3 nL/s (30 rpm), (B) 3.5 nl/s (90 rpm) and (C) 6.6 n/s (170 rpm). Please click here to view a larger version of this figure.
Figure 3: Stream behavior of 100% silicone grease demonstrating curling of the viscous media. (A) Curling effect at 2.3 nl/s (30 rpm) (B) The effect of adding the polystyrene catcher to the sample waste holder with the injector speed operating at 2.3 nL/s (30 rpm). Please click here to view a larger version of this figure.
A glass syringe containing 26 µL of lysozyme crystals (crystal size 10 µm x 10 µm x 20 µm) suspended in silicone grease was used (Figure 4A). Lipidico generated a constant sample stream, using the 108 µm needle I.D., with an average flow rate of 1.14 nL/s (with the motor set to 30 rpm). A peak finding algorithm36 was used to assess the hit rate and a sufficient amount of data (total of 224,200 images) was collected within 38 min of data collection time, enabling structure retrieval (PDB code 6MQV). Table 1 provides a summary of the data collection statistics and Figure 4B shows a representative image of the electron density map surrounding one of the di-sulfide bonds in the lysozyme structure20.
Figure 4: Lysozyme images. (A) Optical images of the lysozyme crystals in silicone grease. A cross polarized (40x magnification) image of lysozyme crystals mixed with a high viscous media, silicone grease, imaged using a visible light microscope. The size of the crystals range between 15 and 20 µm and (B) A representative image of the electron density map (2Fo-Fc, 1σ) surrounding the disulfide bonds of lysozyme. The figure has been adapted from Berntsen et. al. 201923 with the permission of AIP Publishing. Please click here to view a larger version of this figure.
Supplementary Information. Please click here to download this file.
Table 1: Summary table showing SX data statistics from Lipidico. Lysozyme crystals were mixed with a high viscous media, silicone grease, and streamed through the X-ray interaction region of the MX2 beamline. This is a summary of the complete table of results published in Berntsen et. al.23. The table is adapted here with the permission of AIP Publishing.
Data statistics | Lipidico, MX2 beamline |
Detector | Dectris EIGER X 16M, |
frame rate | 100Hz |
Sample-detector distance (mm) | 300 |
X-ray energy (KeV) | 13 |
Beam size (WxH) (µm) | 12×22 |
No. of frames | 224200 |
Hit rate (%) | 2.95 |
No. of indexed frames | 4852 |
Resolution (Å) | 17.74-1.83 |
Completeness | 99.44 |
Space group | P43212 |
Unit cell | |
a, b, c (Å) | 78.68, 78.68, 34.48 |
α, β, γ (°) | 90, 90, 90 |
I/σ(I) | 5.08 |
Redundancy | 73.97 |
CC1/2 | 0.96 |
Rwork/Rfree | 0.18/0.26 |
An alternative HVI has been developed, ideal for carrying out SX experiments at synchrotron sources. It has two key advantages over existing HVIs. First, it is easy to install on the beamline allowing rapid switching between conventional crystallography and SX, just ~30 minutes is required for installation and alignment on MX2. Second, the sample syringes used to grow crystals can be directly used as the reservoirs for injection, limiting wastage during sample transfer. The protocol for changing samples has been described and demonstrated. The design eliminates the need for a complicated gas stream to control the jet compared to other HVIs. A simple process for changing the flow rate of the injector was demonstrated which can be adjusted without a delayed response.
The most crucial steps for successful SX data collection are optimizing the crystal concentration, obtaining a homogeneous mixture of high viscous media, and producing a stable stream for data collection. The optimal crystal concentration and homogeneity can be achieved by centrifuging the crystals down to a pellet and adding a higher concentration of crystals to the carrier media, ensuring that the sample is thoroughly mixed in the coupler system. Once the crystal concentration is optimized, the injector flow rate can easily be adjusted during beamtime to obtain a stable stream. Curling of the viscous stream is commonly observed with high viscous samples. Two solutions are presented: use of a polystyrene sample catcher for charged samples or the addition of a venturi suction funnel acting as a catcher. This has been successful in controlling the stream in the initial tests, but under different sample conditions curling of the stream may result in sticking of the sample to the needle point. It is possible that this could be overcome by changing the surface charge properties of the needle by adding special coatings. Needles coated with different chemicals (i.e. silicone) have previously been successfully adapted for use with liquid handling robots and can be custom ordered from manufactures to suit the syringe37.
The injector is not limited to silicone grease and can be used with other highly viscous liquids. Several different alternatives have been demonstrated23 which can be successfully used for crystal injection and are described in a number of other publications22,30,33. The upper limit on sample viscosity using this HVI is still under investigation, however sample viscosities of up to 25 Pa.s (using 100 % silicone grease) have resulted in a stable stream. However, when the sample viscosity was reduced to <10 Pa.s (using 70 % silicone grease sample) a reliable stream could not be produced. Hence, 10 Pa.s represents the current lower limit on the sample viscosity for injection using this HVI. The needle I.D. is also an important consideration. Although 51 µm I.D. was successfully tested whilst injecting various carriers without any crystals, the introduction of crystals disrupts the sample flow. Hence, depending on their size, with crystals introduced into the matrix, there is a much higher likelihood of blockage when using the 51 µm I.D. needle compared to the larger needle size. When a blockage occurs a critical buildup of pressure can occur, in other injectors this could result in syringe breakage. However, the design of this HVI includes a safety mechanism where the drive mechanics will pull away from the syringe plunger if the pressure becomes too high until a deactivation switch is enabled. This results in the drive being stopped and prevents the glass syringe from rupturing.
Several limitations for this injector exist. The injector is specifically designed for highly viscous samples, therefore, liquid like samples cannot be used directly for sample delivery. To overcome this limitation, crystals grown in buffer systems can be mixed with a suitable inert media as described in step 1, however, there is a possibility the crystal may become unstable. Therefore, screening of a variety of inert media26,27,28,29 should be investigated first to ensure sample stability is maintained. Secondly, the size of the crystal and quality of crystal packing will affect the diffraction quality. The MX2 beamline operates at an optimal beam size of 22 x 12 µm (H x W) and flux ~1012 photon/s. The optimal crystal size to carry out SX at MX2 is ~10 µm and has been shown to yield a high signal-to-noise ratio. However, it is possible to collect data on smaller sized crystals if they are well ordered and diffract to high-resolution. There is an option to slit down the beam size at this beam line to ~7.5 µm, however, this comes at a cost since it reduces the incident flux which needs to be considered during the experiment.
The development of this injector makes SX easily accessible to mainstream crystallographers. The successful operation of Lipidico, opens the door to a fast and easy method for SX data collection at a wide variety of synchrotron sources. It enables room temperature data collection on crystals which are 10 µm or less at MX2, limiting radiation damage effects for individual crystals. It also provides a new opportunity in Australia to carry out millisecond time resolved SX which is the current state-of-the-art for crystallographers. Future applications of this injector system extend to X-ray characterization of highly viscous materials using Small Angle X-ray Scattering (SAXS) allowing the injector to be readily adapted to other beamlines at the Australian Synchrotron.
The authors have nothing to disclose.
This work was supported by the Australian Research Council Centre of Excellence in Advanced Molecular Imaging (CE140100011) (http://www.imagingcoe.org/). This research was undertaken in part using the MX2 beamline at the Australian Synchrotron, part of ANSTO, and made use of the Australian Cancer Research Foundation (ACRF) detector.
Hen eggwhite lysozyme | Sigma-Aldrich | L6876 | Used to grow crystals for testing the injector and the crystals are transferred into silicon grease. https://www.sigmaaldrich.com/ |
High vacuum silicon grease | Dow Corning | Z273554-1EA | Used for testing of injector. https://www.sigmaaldrich.com/ |
Injector needle (108 µm ID) | Hamilton | part No: 7803-05 | www.hamiltoncompany.com |
Glass gas-tight syringes, 100 µl | Hamilton | part no: 7656-01 | Syringes used for sample injection. www.hamiltoncompany.com |
LCP syringe coupler | Formulatrix | 209526 | Syringe coupler to mix the samples |
Lipidico injector | La Trobe Univerity/ANSTO | This is a specific piece of equipment that can be accessed through La Trobe University / ANSTO Australian Synchrotron Facility |