Here, a protocol describing the sample preparation and data collection steps required in cryo soft X-ray tomography (SXT) to image the ultrastructure of whole cryo-preserved cells at a resolution of 25 nm half pitch, is presented.
Imaging techniques are fundamental in order to understand cell organization and machinery in biological research and the related fields. Among these techniques, cryo soft X-ray tomography (SXT) allows imaging whole cryo-preserved cells in the water window X-ray energy range (284-543 eV), in which carbon structures have intrinsically higher absorption than water, allowing the 3D reconstruction of the linear absorption coefficient of the material contained in each voxel. Quantitative structural information at the level of whole cells up to 10 µm thick is then achievable this way, with high throughput and spatial resolution down to 25-30 nm half-pitch. Cryo-SXT has proven itself relevant to current biomedical research, providing 3D information on cellular infection processes (virus, bacteria, or parasites), morphological changes due to diseases (such as recessive genetic diseases) and helping us understand drug action at the cellular level, or locating specific structures in the 3D cellular environment. In addition, by taking advantage of the tunable wavelength at synchrotron facilities, spectro-microscopy or its 3D counterpart, spectro-tomography, can also be used to image and quantify specific elements in the cell, such as calcium in biomineralization processes. Cryo-SXT provides complementary information to other biological imaging techniques such as electron microscopy, X-ray fluorescence or visible light fluorescence, and is generally used as a partner method for 2D or 3D correlative imaging at cryogenic conditions in order to link function, location, and morphology.
Cryo-SXT can play a central role in biological imaging research as it provides 3D medium resolution (25-30 nm half pitch) volumes of hydrated whole cells1,2,3,4,5,6. In the water window energy range, between the carbon and the oxygen absorption K edges (4.4-2.3 nm), carbon-rich cellular structures absorb 10 times more than the oxygen-rich medium that permeates and surrounds them. In this energy range, vitrified cells up to 10 µm thickness can be imaged without the need for sectioning or staining, leading to quantitative high absorption contrast projections, which, combined with sample rotation capabilities, allow for the tomographic reconstruction of the cellular structure. Cryo-SXT fills a niche in terms of specimen dimensions and spatial resolution that is not easily accessible by any other imaging technique.
In brief, the absorption contrast of cryo-SXT is quantitative, as the attenuation of the photons through the specimen of thickness t obeys the Beer-Lambert law as follows: , where I0 represents the incident intensity and µl the linear absorption coefficient, which depends on the wavelength λ and the imaginary part β of the refractive index of the specimen (). The attenuation is a function of the biochemical composition and the thickness of the structures being imaged, with each biochemical component having a specific X-ray linear absorption coefficient µl (LAC). This means that each tomography voxel value depends on the chemical elements and their concentration in that voxel7. This allows for the natural discrimination of different organelles such as nuclei, nucleoli, lipid bodies or mitochondria, or different compaction states of chromatin solely based on their inherent LAC values reconstructed2,8,9.
In addition, cryo-SXT is a high throughput technique with tomograms being collected in few minutes. This specifically enables mesoscale imaging of cell populations that can be captured at key time points such as division, differentiation, and apoptosis, but also at different response states such as those induced by chemical exposure to specific drug therapies or to pathogenic infections. Data collected at those key points will deliver 3D description of the system with a faithful record of the spatial organization of the different cellular organelles at those specific moments.
Usually, cryo-SXT is used in combination with other techniques following correlative approaches that allow locating specific features, events, or macromolecules within the 3D cellular environment4,10,11,12,13,14,15,16, or hard X-ray fluorescence data17,18. Correlative approaches at cryogenic conditions are of paramount importance in order to obtain the most complete and valuable picture of the system of interest. A succinct summary of the typical workflow at the Mistral (Alba) and B24 (Diamond) cryo-SXT beamlines is sketched in Figure 1.
Moreover, taking advantage of the wavelength tuning capability at synchrotron facilities, spectroscopic information can be obtained in addition to the structural one using the specific differential absorption of particular elements contained in the sample. An example of this would be the location of calcium in the study of biomineralization processes in cells19,20,21. By taking 2D images at different photon energies (spectra) or tomograms below and at the x-ray absorption edge of interest, the pixels or voxels containing the selected element can be identified. Spectra also permits differentiating chemical states (i.e., the evolution of amorphous calcium to hydroxyapatite as in the previous biomineralization example20). Quantification of different elements is possible in 2D and 3D. Spectroscopic imaging of vitrified cells is typically done in the water window, but is also possible at other energy ranges if the water content is low enough or if other sample preparation protocols, including dehydration, are used22. A detailed spectroscopy step-by-step protocol is beyond the focus of the protocol herein.
In what follows, the protocol focuses on briefly summarizing the major sample preparation steps, although each system might need individual refinement, followed by a detailed step-by-step data collection procedure for cryo soft X-ray tomography.
1. Sample preparation
2. Loading into the Transmission X-Ray Microscope (TXM)
3. Imaging using the TXM software
NOTE: Grids at the sample stage are first imaged using an on-line visible light microscope (VLM) to map the grid either in brightfield and/or fluorescence mode before being imaged with X-rays. Use the joystick icon corresponding to the Motion Control tab on the top left panel to open the Motion Control.
4. Data analysis
NOTE: All data analysis is done with available open software and scripts developed with automated pipelines.
Preparing samples for cryo-SXT can be challenging. Even within the same sample grid, it is possible to have areas that are ideal, and areas that are non-ideal, as can be seen in Figure 5, which shows two squares from the same Au finder grid. The ideal sample should have single cells at the center of a square mesh, embedded in a thin layer of ice and surrounded by well-dispersed Au fiducial markers used for the alignment of tilt projections prior to tomography reconstruction. Figure 5A shows a fibroblast-like cell (NIH 3T3) that complies with many of these criteria. A single slice from a 3D reconstruction using ART27 of the area marked with the red box indicating the field of view (FoV) is shown in Figure 5B. Many different organelles such as mitochondria (M), endoplasmic reticulum (ER), vesicles (V) and the nucleus (N) can be distinguished thanks to the quantitative reconstruction of the LACs. In addition, the signal-to-noise ratio of the reconstruction is very high allowing to achieve high contrast of the cellular features. On the other hand, Figure 5C shows a square with higher cell density. Because of this, the blotting is usually less efficient, leading to a thicker ice layer, or even vitrification issues. In some cases, this can already be observed when screening the grid using epifluorescence mapping prior to the X-ray imaging, and those grids should be avoided at any cost. In Figure 5C, a crack within the grid and the vitrified ice can be observed going through the entire mesh square (marked by the red arrows). Any imaging near cracks should be avoided due to probable instability of the grid when exposed to the beam. In addition, cracks can be a sign of thick ice, as was the case in this area. A tilt series was recorded in the area marked with the red box. In Figure 5D, a single slice from the corresponding 3D reconstruction is shown. Even though some larger structures can be recognized, fine details are lost within the noise and grainy texture due to the poor vitrification quality of the thick ice, as can be seen specifically, for instance, on the upper mitochondria pointed by the arrow.
Figure 1: Workflow. Schematic workflow followed prior to cryo-SXT data collection. Please click here to view a larger version of this figure.
Figure 2: Growing cells on grids. (A) Cells growing in a P100 Petri dish with a confluence around 80%-90%. (B) P60 Petri dish with several grids after seeding the cells. (C) Cells growing on top of a grid after 24 h. Scale bars: 100 µm. Please click here to view a larger version of this figure.
Figure 3: Loading grids on the sample holders and into the transfer chamber. (A) Workstation filled with liquid nitrogen with the shuttle and the cryoboxes ready for loading the grids. (B) Sample holder inserted into the loader with the grid loaded. (C) Shuttle with the sample holder in position 3 without the cover. (D) Workstation with the transfer chamber attached. Please click here to view a larger version of this figure.
Figure 4: Loading samples into the TXM (A) Attaching the transfer chamber to the TXM. (B) Shuttle inside the TXM. (C) TXM robot arm inserting the sample holder into the sample stage. Please click here to view a larger version of this figure.
Figure 5: Example of cryo soft X-ray tomograms. Upper row: ideal sample, (A) 2D mosaic view of a grid square showing an isolated cell at the center. (B) One slice from the reconstructed 3D volume showing the marked area with the red box (A). Compared to (D) the image is much smoother and more details are visible. Lower row: non-ideal samples, (C) 2D mosaic view of a grid square showing too high cell confluency and cracks in the ice and grid foil (red arrows). (D) One slice from the reconstructed 3D volume showing the area marked with the red box in (C). The poor or suboptimal vitrification can be identified by the grainy texture of the image. N: Nucleus; M: Mitochondria; ER: Endoplasmic Reticulum; MV: Multivesicular bodies; V: Vacuole; Scale bars: A & C 20 µm; B & D 2 µm.VPlease click here to view a larger version of this figure.
Sample preparation is a critical step to obtain high quality soft X-ray tomograms, as their quality directly depends on the quality of the sample vitrification and the ice layer thickness in which the cell is embedded. Projections with high signal-to-noise ratio will be collected in regions with thin ice layer, allowing to minimize the radiation dose required to achieve the highest possible resolution. In addition, the cell confluency will also affect the final tomogram quality, since one should avoid having neighboring cells entering the FoV upon rotation. Finally, the right dispersion of Au fiducial markers will determine the accuracy of the projection alignment and then ultimately determine the quality of the final 3D reconstructed volume. Note that a proper spread of Au fiducials on the grid enables automatization of the projection alignment step, without which a high expertise is needed for such a critical step.
The protocol herein only depicts one possible sample preparation strategy, which has similarities with the ones used in cryo electron tomography (cryo-ET). In both cases, protocols improving the demanding sample preparation for better reproducibility will be fundamental for the success of these techniques, and efforts are being made toward this goal29. It is worth mentioning that in addition to imaging isolated cells, sections of tissue can also be visualized provided the transmission signal through the section will be enough at high tilt angles. Typically, this will imply sections of few microns (below 10 µm).
To image a specific structure or event inside a cell, one needs to make sure this particular feature is inside the FoV of the tilt series. As the FoV in cryo-SXT is limited to 10 x 10 µm2 to 15 x 15 µm2 depending on the lens and accounting for a pixel oversampling of the resolution to at least a factor of 2, it is often smaller than the full cell extension (see the red squares indicated in Figure 5). Therefore, the ROI must be found and properly labeled. This is usually done by means of fluorescent tags and visible light correlative approaches. 2D strategies combining epifluorescence are straightforward as the soft X-ray transmission microscope has an integrated on-line visible light fluorescence microscope, but other approaches for high resolution 2D or 3D fluorescence signal are also available4,12,13,15,16. In those cases, the grid needs to be imaged first in specific instruments such as super resolution microscopes. Note that the most efficient correlative approaches are those involving data collection at cryogenic conditions. This is because the time lapse between room temperature (RT), visible light fluorescence imaging, and sample vitrification, for instance, will hinder catching the right cellular event on time; in addition, the vitrification procedure might detach the cell of interest that has been imaged at RT from the grid. Even if most correlative imaging approaches might imply that the sample grids have to be manipulated and transported from one instrument to the other, and despite the increased risk of grid contamination or damage this poses, the reward is clear: to be able to pinpoint specific events or molecules within the cellular landscape.
When whole cell imaging is required, stitching different tomograms is possible provided the total dose applied does not exceed the radiation damage limit. Usually, the deposited dose for collecting few tomograms on the same cell is well below the limit at the achievable resolution (109 Gy) and, therefore, no specific strategy is required to lower the dose, although this is sample- and experiment-dependent. In the case of intensive data collection such as spectro-tomography, minimizing the dose would indeed be required and convenient data collection and specific processing strategies would need to be applied.
Cryo-SXT has several limitations, which should be mentioned here. The first one is the well-known missing wedge, which is intrinsic to using flat sample supports. Capillary sample supports allowing 180-degree rotation have been used in the past and are still used at some facilities, but they also present drawbacks such as an impoverished contrast due to the glass absorption and the restriction of using cells in suspension. A way to diminish the effect of the missing wedge is by performing dual tilt tomography. This is indeed possible at the Mistral beamline nowadays. The second limitation is set by the Fresnel zone plate lens used in such microscopes. This lens sets the ultimate resolution achievable and the depth of field (DoF), both being tightly related. This implies that increasing the resolution will diminish the DoF while the thickness of the cell will often be larger. For example, a 40 nm lens will have in theory a DoF of 3 µm and a resolution of 24.4 nm half pitch. The compromise between resolution and DoF is therefore strategic and the choice of the lens will depend on the type of the cell30,31. Finally, operational TXMs worldwide are far from being ideal microscopes and efforts are being made to improve the optical systems to reach the theoretical expectations. Finally, the visualization and segmentation of the reconstructed volumes can be carried out with specific software tools25,32,33,34.
In summary, cryo-SXT allows imaging cells quantitatively at medium resolution (25-30 nm half pitch) and in statistical numbers (few tens of tomograms per day). This allows obtaining the organization, distribution, and dimension of organelles at specific conditions, for instance, during pathogen infection or diseases, at precise time points or after particular treatments. It is, therefore, a useful complementary biological imaging technique to the more common electron and visible light microscopies, each of them tackling a specific range of sample dimensions and resolution. Cryo-SXT is frequently used in correlative approaches involving visible light fluorescence, but other cryo correlative strategies are also possible.
The authors have nothing to disclose.
This project has received funding from the European Commission Horizon 2020 iNEXT-Discovery project and the European Union's Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement No 75439.
Amira | Thermo Fisher | Software for segmentation | |
Au Holey Carbon Films finder grids | (Quantifoil Micro Tools Gmb | R 2/2 Au G200F1 | Au Holey Carbon Films finder grids |
Au nanoparticles | BBI Group, Cardiff, UK | Au nanoparticles 100nm | 100 nm Au nanoparticles (NPs) at Mistral (Alba) |
Au nanoparticles | BBI Group, Cardiff, UK | Au nanoparticles 250nm | 250 nm Au nanoparticles (NPs) at B24 (Diamond) |
Axio Scope A1 | Zeiss | 430035 9060 | Fluorescence microscope |
Blotting No.1 filter paper | Whatman | WHA10010155 | Blotting filter |
Bsoft | Software for projection alignment, reconstruction and visualization (Heymann et al., 2008) | ||
Chimera | Software for segmentation (Pettersen et al. 2004) | ||
Cryo-EM Glow Discharge Set | PELCO easiGlow | 91000S | Glow Discharge Cleaning System |
Cu Holey Carbon Films finder grids | (Quantifoil Micro Tools Gmb | R 2/2Cu G200F1 | Cu Holey Carbon Films finder grids |
Fetal calf serum | Sigma | F9665 | Heat Inactivated, sterile-filtered, suitable for cell culture |
ImageJ | Software for image processing and analysis in Java (NIH & LOCI University of Wisconsin) | ||
IMOD | Software for projection alignment, reconstruction and visualization (Kremer et al., 1996) | ||
Leica EM GP Grid Plunger | Leica | 16706401 | Automatic Plunge Freezer EM |
LINKAM cryo-stage | Linkam Scientific Instruments | CMS 196 | Cryo-Correlative Microscopy Stage |
MIB | Software for segmentation (Belevich et al. 2016) | ||
P100 Petri dish | Sigma | P6106 | Treated for cell culture and sterile |
P60 Petri dish | Sigma | D8054 | Treated for cell culture and sterile |
Polylysin | Sigma | P4707 | Poly-L-lysine solution 0.01%, sterile-filtered |
Soft X-Ray microscope 0.25-1.2keV | Xradia | NCT-SB | Transmission soft X-Ray microscope |
SURVOS | Software for segmentation (Luengo et al. 2017) | ||
Tomo3d | Software for reconstruction (SIRT, WBP) (Agulleiro et al. 2011) | ||
TomoJ | Software for reconstruction (ART) (Messaoudi et al., 2007) | ||
XM Data Explorer | Zeiss | TXM software |