CryoEM screening with a blotting instrument: quantifying parameters affecting ice thickness using semi-automated image analysis
-
Sepideh Valimehr
,
Ellie Cho
,
Hamish G. Brown
Abstract
Cryo-electron microscopy (cryoEM) grid preparation is one of the bottlenecks in using cryoEM for macromolecular structure determination. Despite significant advancements in the development of blot-free instruments, their high cost limits their widespread usage. Understanding the effects of the different parameters of the blotting instruments, particularly the Vitrobot, can lead to producing high-quality grids while saving both time and resources. In this study, we focus specifically on low magnification cryoEM images as a rapid way to evaluate ice thickness and screen the effects of different grid preparation parameters. By employing a machine learning-based approach and semi-automated image analysis, we analysed large datasets of low-magnification atlas images to quantify ice thickness and distribution across grids prepared under varied conditions. Our results show that detectable changes in ice quality often require substantial adjustments in parameters, and even then, considerable grid-to-grid variability can persist. Notably, we observed that the presence of detergent improved consistency in ice thickness. While our approach does not assess protein distribution or particle behaviour at high magnification, it offers a scalable and efficient tool for early-stage grid screening and protocol optimization.
1 Introduction
Following the resolution revolution era in single particle cryo electron microscopy (cryoEM) [1], there has been a rapid and exponential advancement in obtaining high-resolution structures of macromolecular complexes using cryoEM, evidenced by the increased rate of PDB (Protein Data Bank) structure depositions [2]. Due to the continuous progress in both hardware and software in the field of cryo-EM, obtaining high resolution structures of macromolecules has become feasible for both soluble and membrane proteins [3], [4]. However, there remain certain challenges to address, particularly in the cryoEM grid preparation and screening steps [5]. The first step to obtain a high resolution cryoEM structure is preparing a high-quality sample via biochemical characterisation with additional validation by negative stain EM and mass photometry [6]. Grid preparation generally starts with chemical, or plasma treatment of the grid and its foil substrate followed by deposition of the sample and then vitrification in liquid ethane. The quality of cryoEM data critically depends on the preparation of the grids to obtain an ideal vitreous ice thickness, or sample thickness, where the protein particles in different orientations [7] are embedded in a thin layer of vitreous ice. The 3D reconstruction is then obtained from 2D projection images of the particle [8], [9]. While in the past it was sufficient to only have a few optimised areas, the advent of very fast detectors over the last few years has exacerbated the need for large area of grids to be collection ready and more grids to be produced.
There are different approaches to cryo-EM grid preparation [10]. The traditional method involves deposition of 3–5 µl of the specimen on the hydrophilic grid then removing the excess sample by manual blotting with a filter paper and gravity-driven plunge-freezing into a cryogen with a manually operated trigger [11], [12] with the apparatus often located in cold rooms to control temperature and humidity, preventing sample evaporation. The advent of the modern computer-controlled blotting devices with enclosed chamber for humidity and temperature control (e.g., the ThermoFisher Vitrobot Mark IV, the Gatan Cryoplunge 3, and the Leica EMGP2) moved toward more automated approach and improved the grid preparation process [13], [14]. Recently, new freezing instruments have been introduced that offer alternative methods for cryoEM grid preparation. More recent developments have incorporated glow discharge, pin-printing, and plunge-freezing for reproducible and contamination-free grid production [15], [16], [17]. Several of these devices have been specifically developed for time-resolved structural studies, enabling rapid mixing and vitrification within milliseconds. Early efforts in this direction, such as the spray-freezing and jet vitrification systems developed in the 1990s [18], established the foundation for capturing dynamic intermediates by minimizing sample exposure times before freezing. White et al. had demonstrated a second-generation time-resolved apparatus for capturing transient macromolecular reactions with millisecond precision [19]. Building on these concepts, Kontziampasis et al. described a microfluidic mixing-spraying apparatus that allows grid vitrification within ∼10 ms [20], [21]. Later innovations, including the Spotiton and Chameleon introduced precise sample dispensing and self-wicking grids to achieve more reproducible ice thickness and reduced particle denaturation at the air–water interface [15], [16]. A systematic study was performed to compare three vitrification methods, Vitrobot, chameleon, and a time-resolved cryoEM device (TED) to evaluate how time between sample application and vitrification affects macromolecular complex integrity [22]. The combination of these rapid-freezing strategies has already enabled visualization of dynamic molecular events, such as the swinging lever mechanism of myosin-V revealed by time-resolved cryo-EM [23]. However, the high cost of commercial prototypes limits their broader use in the scientific community. There are currently a few facilities around the world which can afford these types of instruments and it remains the case that the majority of high-quality 3D reconstructions deposited in the Electron Microscopy Databank (EMDB) are from grids prepared with conventional blotting devices [24].
While the traditional Vitrobot, a blotting instrument, has been widely used and is a relatively cost-effective solution for cryoEM grid preparation, there is no clear quantification of the effect of two main parameters of the Vitrobot, the displacement of the horizontal blot-pad position (blot force) and the time over which the specimen is absorbed into the filter paper (blot time) [25] on cryoEM grid preparation.
This study aims to quantify and investigate the impact of these two crucial parameters in the blotting process on the quality of cryoEM grids using semi-automated analysis pipeline on low magnification micrographs that combines machine learning segmentation with automated image analysis. We varied these parameters while keeping all other conditions constant to evaluate their effects on grid quality.
By quantifying these effects, we found that both blot force and blot time have a mild positive correlation on the grid readable area coverage and a mild negative correlation on ice thickness. Interestingly, adding detergent to the solution perturbated this correlation in most tested groups. Notably, there was high variation of grid quality between grids prepared with identical parameters which sometimes exceed the variation observed between groups of grids prepared with different blot forces and times. Consequently, to achieve significant differences in ice thickness and coverage, substantial changes in blot time and force are usually required. This is also noted in Glaeser’s work that minor adjustments to Vitrobot blotting parameters, such as force or time, typically have little reproducible effect on ice thickness. Micro-scale interactions between the filter paper fibres and the sample dominate liquid removal, so only relatively large changes produce noticeable effects [26]. We therefore recommend avoiding small incremental changes to the Vitrobot preparation step and opt for larger changes to achieve desirable sample thickness.
2 Materials and methods
2.1 CryoEM grid preparation
The cytosolic AAA+ ATPase p97 protein was used for cryoEM grid preparation. The construct and the purification steps are described elsewhere [27]. Quantifoil R1.2/1.3, 300 mesh grids glow discharged at 15 mA for 30 s using air with GloQube (Quorum) was used for all experiments. The grids were prepared during two independent experiment using a Vitrobot IV (Thermofisher Scientific) (1) by varying the blot force to −10, −5, −1, 0, 1, 5 and 10 with 5 s blot time and (2) by changing the blot time to 3 s, 5 s, 10 s, 20 s and 30 s with a fixed −1 blot force at 22 °C and 100 % humidity. 4 µl of 800 μg/ml of the protein solution was applied to the grids. The same settings were used to prepare grids with detergent, except the protein concentration was increased to 4 mg/mL and 0.02 % NP40 (v/v) was added to the sample immediately before freezing. The grids were prepared in three replicates on three different days for each condition by the same experienced operator. To prevent variations due to environmental factors such as humidity and temperature [26], the filter papers were conditioned in the Vitrobot chamber at 22 °C and 100 % humidity for 15 min prior to use.
One blank cycle of plunge-freezing with tweezers was performed before starting each experiment.
2.2 Data acquisition
The data were collected with EPU, an automated acquisition software (Thermo Fisher Scientific), using Atlas magnification presets specific to each microscope and camera setup. On the Talos Arctica equipped with a K2 direct detector (Gatan), images were acquired at 110× magnification, corresponding to a pixel size of 0.125 µm/pixel. On the Titan Krios G4 equipped with a Falcon 4 direct detector (Thermo Fisher Scientific), images were acquired at 100× magnification, corresponding to a pixel size of 0.157 µm/pixel. On the Titan Krios G4 equipped with a K3 direct detector (Gatan), images were acquired at 135× magnification, corresponding to a pixel size of 0.067 µm/pixel. The Atlas images were used for quantification and image analysis. The high magnification images were acquired using Titan Krios G4 and K3 direct detector at 81,000× magnification corresponding to a pixel size of 1.069 Å/pixel with total dose of 50 e−/Å2 and defocus value of −2 μm.
2.3 Image analysis
2.3.1 Step 1 stitching
At Atlas magnification, a mosaic image was taken as a ‘step spiral’ pattern where the first image starts from the centre of the field (Figure S1A). Individual tiles of large mosaic scan were exported as .mrc file format and post-processed for stitching with the Bigstitcher [28] Plugin in FIJI [29].
To streamline the tile alignment, first the individual tile files were renamed to follow the Down and Right order, and the tile configuration file was generated based on its new sequence (Figure S1C, Script1). In some datasets, tiles were rotated slightly (0.5–2°) in a random manner, and this was corrected manually with the aid of auto importing and auto exporting of dataset (Figure S2, Script 2, Script2_1, Script2_2). Tiles corrected for random rotation were automatically imported to BigStitcher Plugin (Script3) and the individual tile positions were further corrected before export as the final stitched montage (Figure 1B).
2.3.2 Step 2 grid square measurement
In order to detect every grid square including ones rendered opaque by thick ice, a synthetic grid image was generated for each dataset type (Figure S3A, Script4) and registered to the stitched image. The stitched image was rotated to align the grid square to the horizontal and vertical axis. The rotation angle of grid was defined based on the visible squares using top-hat filter and measuring the dominant direction using OrientationJ [30] plugin (Figure S3B, Script5). For the sample that has very low number of visible squares, automated detection failed, and this was further corrected by user based on the user interactive prompt (Script5).
On the stitched image, the visible grid area was determined with ‘Default’ automatic threshold [31]. Then the synthetic grid was imposed and aligned to achieve the maximum overlap with the visible grid area (Figure S3C). The threshold area of each grid was measured and presented as a percentage of full-size grid (theoretical value). The squares near the edge of the image were not measured to avoid partial detection of grid (Figure S3D, Script6).
2.3.3 Step 3 hole measurement
To measure the thickness of vitreous ice, the grey intensity of individual holes within the grid was measured and converted to equivalent thickness (see Ice thickness measurement section below). To detect individual holes, a similar registration approach was used. First, the individual grid square area identified from the previous step was extracted and combined as a tiff stack per dataset for easier handling (Figure S4A, Script6). A synthetic hole image for each dataset type was generated (Figure S4B, Script7). Visible hole detection was performed with machine learning based pixel classification using Labkit [32] in FIJI (Figure S4C, Script8) followed by measuring the rotation angle (script9). During the pixel classification, an additional class was set to detect empty (vacuum) areas.
On the hole segmented image, the synthetic hole image was aligned to identify total possible holes. From the possible holes, readable holes were defined if they overlap with visible hole (detected using Labkit) (Figure S4D, script10). The intensity was measured only on the readable holes and further normalised by the average intensity of the vacuum (empty) area from each dataset.
2.4 Ice thickness measurement
To measure the ice thickness, images were collected using K3 camera at atlas magnification both with and without energy filter. The slit width for the energy filter was set at 20 eV. The equation d = Λln I/Izlp [33] was used to generate the standard curve to convert intensity to the ice thickness (Figure S5), where d is the ice thickness in nanometres, Λ is the apparent mean free path (435 ± 30 nm) for the 300 KeV calculated by Rice et al. [33], I is total intensity without Gatan Imaging Filter (GIF), and Izlp is total intensity with GIF. To create a standard curve for Falcon 4 camera, images were taken from the same holes previously imaged with the K3 to measure the intensity and the ice thickness. A separate equation, based on the slope of the Falcon 4 standard curve, was then used to convert Falcon 4 intensities into ice thickness.
2.5 Statistical analysis
For the group comparison (different blot time or blot force) within each treatment (i.e. non-detergent or detergent), One-way ANOVA with Tukey’s post hoc test was performed. For the treatment comparison (i.e. non-detergent vs. detergent), Two-way ANOVA with Tukey’s post hoc test was performed. For measuring correlation, Pearson’s correlation coefficient was measured. For measuring data variability between groups, Coefficient of Variation was compared between groups using Unpaired T-test. All statistics were measured using GraphPad Prism (v10.4.1). P values under 0.05 were reported as significant.
2.6 Scanning electron microscopy
Standard Vitrobot filter paper (grade 595, Whatman) with an outer diameter of 55 mm and an inner diameter of 20 mm was cut into small pieces (∼3 mm), approximately matching the size of the cryo-EM grids. Because the filter paper rotates during the blotting process, pieces were taken from three different regions of the paper. The filter paper pieces were then sputter-coated with a 10 nm layer of gold using a Safematic CCU-010 coater. Subsequently, the coated pieces were mounted on scanning electron microscopy (SEM) stubs, and micrographs were acquired using a Teneo SEM (Thermo Fisher Scientific) operated at an accelerating voltage of 10 kV and a beam current of 50 pA at 50×, 250× and 800× magnification.
3 Results
In this study, we quantified the impact of grid preparation parameters on the distribution and thickness of vitreous ice at low magnification. To obtain a comprehensive overview of the entire grid, which contains more than 300 grid squares and 200,000 individual holes, an automated detection and measurement workflow was essential. We established a workflow that encompasses the entire image analysis process, from stitching to automated detection and measurement, using ImageJ plugins and macro scripts. Example dataset and scripts are available at https://tinyurl.com/yc5c324k. The first step of this workflow involved stitching individual tiles to create a mosaic image, ensuring high-accuracy alignment of the grid squares. This mosaic image was then used for the automatic detection of grid squares, regardless of their visibility (Figure 1). Since the Atlas image acquisition was not designed for a precisely aligned mosaic image, simple alignment based on positions extracted from metadata was unsuccessful (Figure 1A). This issue was primarily due to the rotation of specimen movement axes with respect to image axes (Figure S1B), which was not calibrated in the metadata, and the random rotation of the sample (0.5–2°) in some acquisitions (Figure S2). Therefore, a semi-automated approach was employed, where the random component of the rotation was manually corrected before automatic stitching.
Image analysis workflow. A: Raw images aligned in BigStitcher based on the stage position extracted from metadata. B: Final stitched image after rotation is corrected. C: Registration with synthetic grid image (yellow) allowing detection of both visible and non-visible grid square. D: Registration with synthetic hole image for complete hole detection.
The acquired atlas images were analysed at both grid square and individual hole level. Given that detergent is commonly included in membrane protein sample buffers [34] and it is also a way to mitigate preferred orientation in cryo-EM samples [35], the experiments were performed with and without adding detergent to the sample prior to the grid freezing. At square level, the area size of individual square was measured from the thresholded image and the percentage area to its possible maximum square size was calculated. Total 376 ± 3.2 squares were measured per stitched image and the average value of percentage area per dataset was used for statistical analysis (n = 6–9 per dataset). For samples without detergent, percentage area measurements revealed positive correlation to the blot force applied (Figure 2A, Pearson r = 0.51, p < 0.001). As blot force increases from −10 to +10, the grid square area percentage gradually increases and become significantly different at blot force of +1, +5 and +10 compared to −10 (ANOVA p < 0.01). Such correlation was not observed for grids with detergent (p = 0.12) and there were no significant differences in the grid square area percentage between blot force groups (Figure 2B, p = 0.15). There was a positive correlation with increasing blot time from 3 to 30 s (i.e. increasing blot time appears to increase the grid square area percentage) on the grids without adding detergent (Figure 2C, Pearson r = 0.31, p < 0.05) although no significant difference was detected between individual groups (ANOVA p = 0.08). On the grids with added detergent, neither positive correlation (p = 0.96) nor a statistically significant difference between groups was observed (Figure 2D, p = 0.97). Interestingly, the grid square area percentage exhibited much higher variability with detergent (Figure 2D, p < 0.01). For example, with the blot time 30 s, square grid area percentage ranged from 0.16 % to 99.9 %. These results indicate that to have adequate square area for data collection, higher blot force and longer blot time is required. In this study, blot force greater than 1 and blot time longer than 10 s resulted in an increase of the area percentage of grid squares by more than 50 %. In addition, it shows that the usage of detergent disrupts the correlation between blot force and ice coverage and generates more variable outcomes.
The effect of blot force and blot time at grid squares level. (A, B) Box plots showing the relationship between blot force and grid square area percentage. (A) Data for samples without detergent (red). (B) Data for samples with detergent (blue). Asterisks indicate statistically significant differences between groups. (C, D) Box plots illustrating the effect of blot time on grid square area percentage. (C) Without detergent (red). (D) With detergent (blue). Data points represent individual measurements.
To better understand the impact of blot force and blot time, we examined the quality of individual grid squares. We measured how many holes within the square were readable and examined their intensity to estimate ice thickness. The number of readable holes followed a similar trend to the grid square area percentage. Without adding detergent, increasing blot force led to higher readable hole counts (Figure 3A, r = 0.55, p < 0.0001) resulting significantly higher numbers of readable holes at positive force groups (+1, +5, +10) compared to the negative blot force groups (−10 and −5) (Figure 3A, p < 0.001). By adding detergent, the positive correlation was no longer evident (p = 0.05) and no significant differences between blot force groups was observed (Figure 3B). With varying the blot time in the absence of detergent, readable hole count showed weak positive correlation although it was not significant (r = 0.28, p = 0.07) and not different between groups (p = 0.12) (Figure 3C). Interestingly, the hole counts increased significantly with blot time in the presence of detergent (r = 0.50, p < 0.001) with longer blot times (20 and 30 s) producing a significantly higher readable hole count compared to shorter blot times (3 and 5 s) (Figure 3D, p < 0.05).
The effect of blot force and blot time on readable hole count. (A, B) Box plots showing how different blot forces impact the percentage of readable hole counts. (A) Results for samples without detergent (red). (B) Results for samples with detergent (blue). Asterisks indicate statistically significant differences between groups. (C, D) Box plots illustrating the effect of blot time on the percentage of readable hole counts. (C) Without detergent (red). (D) With detergent (blue). Each dot represents an individual measurement.
To evaluate the influence of blot force and blot time on sample thickness, the grayscale intensity of each readable hole in the grid was measured and converted to its corresponding thickness. Ice thickness was relatively similar at negative blot forces and then significantly decreased at +1 and +10 blot forces in absence of detergent (Figure 4A, Pearson r = −0.29, p < 0.05, ANOVA p < 0.01). Upon adding detergent, samples were also thinner with higher blot force (r = −0.36, p < 0.01). As expected by lowering the surface tension on addition of detergent [36] the vitreous ice layer was thinner compared to the sample without detergent (p < 0.0001), while it remained relatively stable across all blot forces that there were no significant decreases and increases. (Figure 4B Pearson r = −0.29, p = 0.05, ANOVA p = 0.08). In contrast to the blot force, vitreous ice thickness decreased in longer blot time but without any significant correlation (r = −0.27, p = 0.07). The thickness dropped only once from 60–80 nm at 3 s to 40–50 nm at 5 s, but no further decrease was observed with longer blot time (Figure 4C). Upon adding detergent, vitreous ice thickness also decreased significantly compared to the sample without detergent (p < 0.0001). Similar to the Blot Force dataset, ice thickness was not significantly affected by different blot times (Pearson r = −0.27, p = 0.07, ANOVA p = 0.50) (Figure 4D). This indicates that detergent plays an important role in reducing ice thickness dependent on blot force or blot time. High magnification images confirm the presence and absence of the protein at different ice thicknesses (Figure S7).
The impact of blot force and blot time on thickness of the ice inside the holes. (A, B) Box plots showing how different blot forces impact the ice thickness. (A) Results for samples without detergent (red). (B) Results for samples with detergent (blue). Asterisks indicate statistically significant differences between groups. (C, D) Box plots illustrating the effect of blot time on the ice thickness. (C) Without detergent (red). (D) With detergent (blue). Each dot represents an individual measurement.
4 Discussion
During our experiments we have attempted to keep everything constant between the replicates. Recognizing that several factors such as filter paper variability, instrument condition, and operator technique can influence cryo-EM grid preparation, we took deliberate steps to minimize their impact. For example, the same operator performed all grid preparations using a single Vitrobot, with the same environmental settings. Nonetheless, the inherent microscale irregularities in filter paper remain a likely contributor to subtle grid-to-grid variability, even when all other parameters are held constant. Our own scanning electron microscopy experiments (Figure S6) and two separate studies by Zi Tan et al. [37] and Penzien et al. [38], revealed high irregularities of the fibres in the standard blotting filter papers for the Vitrobot. Despite this fact, different laboratories claim to have developed their own protocols for using the Vitrobot which work reproducibly [25]. However, parameters that work well in one lab may not produce the same results in another due to variability between different instruments, operators and laboratory environments. We have found that consistency can be achieved through protocol optimization and operator experience. Researchers typically start with an optimized protocol suitable for most samples in their lab and then fine-tune the parameters to match their specific needs. Different Vitrobot parameters ranging from 0 blot force to 20 in combination with different blot times are routinely reported in the literature [39], [40], [41]. Our research suggests that large changes in blot force and time are required to induce significant changes in overall grid ice thickness and that small changes of 1–2 s of blot time or 1–2 increments in blot force won’t make much difference.
Armstrong et al. [26]. looked at the microscale fluid behaviour during grid blotting in the absence of added detergent and suggested that, based on measured pressures on the grid during blotting, the range of blot force available in the Vitrobot does not have a large effect on the outcome. They also suggest that most of the wicking occurs during the first 1–2 s and increases beyond that should only have a minor impact. Here, we demonstrated that changing blot force and blot time affect the freezing outcome in the absence of the detergent. However, in the presence of detergent modifying the Vitrobot parameters have very little effect on sample thickness and that concentration of detergent more than blotting parameters is the main parameters guiding sample thickness [36].
We also evaluated the effect of a non-ionic surfactant [42], which was included to explore its potential to reduce preferred orientation and improve ice consistency. While surfactants are commonly used for this purpose in cryo-EM, their effects are highly protein-dependent. Prior studies have shown that surfactants such as CHAPS or fluorinated detergents can enhance particle distribution in some cases but may also induce aggregation or loss of particles in others [43]. Our aim here was not to optimize conditions for a particular protein but to assess general trends at the grid level. As such, we used a single surfactant and matched particle density across conditions to isolate its impact on ice. Future work could expand on this by correlating low-magnification ice behaviour with high-magnification particle quality across a range of surfactant conditions.
Numerous environmental factors, such as temperature and humidity, as well as preparation parameters like the glow discharge, can also significantly influence grid quality [44]. Our extensive experience in preparing grids within the core facility has shown that lowering the temperature reduces the incidence of empty holes on the grid. However, this observation has been based on empirical experience rather than a rigorously quantified scientific method. The semi-automated analysis approach developed in this study facilitates the systematic quantification of these effects, thereby enhancing quality control in future research endeavours.
5 Conclusion
Our study successfully quantified the impact of grid preparation parameters on the distribution and thickness of ice. By implementing a semi-automated detection and measurement workflow using ImageJ plugins and macro scripts, we achieved a comprehensive analysis of the entire grid vitreous ice thickness.
Our results indicate that adding a detergent reduces the effects of blot force and blot time. While the use of detergent introduces some variability in the readable area (as indicated by the grid square area results), it is beneficial in achieving stable ice thickness under varying force and time conditions. Conversely, in the absence of detergent, applying large increases in blot force and blot time results in a more readable area and thinner ice.
List of abbreviations
- cryoEM
-
cryo-electron microscopy
- PDB
-
protein data bank
- EMDB
-
electron microscopy databank
- GIF
-
gatan imaging filter
Acknowledgments
We acknowledge Professor Isabelle Rouiller (Department of Biochemistry and Pharmacology, University of Melbourne) for providing the construct and materials for protein purification, and the Melbourne Protein Characterisation Facility (Bio21 Molecular Science and Biotechnology Institute) for access to protein purification instruments. We acknowledge the Biological Optical Microscopy Platform for providing the computer workstations used to perform image analysis, and the Ian Holmes Imaging Centre for access to the microscopes and cryoEM data collection. We thank Phil Francis for training and assistance with scanning electron microscopy at the Ian Holmes Imaging Centre.
-
Research ethics: Not applicable.
-
Informed consent: Not applicable.
-
Author contributions: Conceptualization, E.H., S.V. and E.C.; methodology, E.H., S.V., E.C. and H.B.; data acquisition, S.V.; software, E.C.; formal analysis, S.V. and E.C.; writing, original draft preparation, S.V. and E.C.; writing, review and editing, E.H., S.V., E.C., H.B. and P.M.; funding acquisition, E.H. and P.M.
-
Use of Large Language Models, AI and Machine Learning Tools: None declared.
-
Conflict of interest: The authors declare no competing interests.
-
Research funding: None declared.
-
Data availability: Example dataset and scripts are available at https://tinyurl.com/yc5c324k.
References
[1] W. Kühlbrandt, “The resolution revolution,” Science, vol. 343, no. 6178, pp. 1443–4, 2014, https://doi.org/10.1126/science.1251652.Suche in Google Scholar PubMed
[2] S. K. Burley, H. M. Berman, G. J. Kleywegt, J. L. Markley, H. Nakamura, and S. Velankar, “Protein data bank (PDB): The single global macromolecular structure archive,” in Protein Crystallography: Methods and Protocols, New York, NY, USA, Humana Press, 2017, pp. 627–41.10.1007/978-1-4939-7000-1_26Suche in Google Scholar PubMed PubMed Central
[3] K. M. Yip, N. Fischer, E. Paknia, A. Chari, and H. Stark, “Atomic-resolution protein structure determination by cryo-EM,” Nature, vol. 587, no. 7832, pp. 157–61, 2020, https://doi.org/10.1038/s41586-020-2833-4.Suche in Google Scholar PubMed
[4] M. A. Herzik, M. Wu, and G. C. Lander, “High-resolution structure determination of sub-100 kDa complexes using conventional cryo-EM,” Nat. Commun., vol. 10, no. 1, pp. 1–9, 2019, https://doi.org/10.1038/s41467-019-08991-8.Suche in Google Scholar PubMed PubMed Central
[5] G. Weissenberger, R. J. Henderikx, and P. J. Peters, “Understanding the invisible hands of sample preparation for cryo-EM,” Nat. Methods, vol. 18, no. 5, pp. 463–71, 2021, https://doi.org/10.1038/s41592-021-01130-6.Suche in Google Scholar PubMed
[6] M. Joppe et al.., “The resolution revolution in cryoEM requires high-quality sample preparation: A rapid pipeline to a high-resolution map of yeast fatty acid synthase,” IUCrJ, vol. 7, no. 2, pp. 220–7, 2020, https://doi.org/10.1107/s2052252519017366.Suche in Google Scholar
[7] I. J. Hirst, W. J. Thomas, R. A. Davies, and S. P. Muench, “CryoEM grid preparation: A closer look at advancements and impact of preparation mode and new approaches,” Biochem. Soc. Trans., vol. 52, no. 3, pp. 1529, 2024.10.1042/BST20231553Suche in Google Scholar PubMed PubMed Central
[8] E. Nogales and S. H. Scheres, “Cryo-EM: A unique tool for the visualization of macromolecular complexity,” Mol. Cell, vol. 58, no. 4, pp. 677–89, 2015, https://doi.org/10.1016/j.molcel.2015.02.019.Suche in Google Scholar PubMed PubMed Central
[9] J. Dubochet et al.., “Cryo-electron microscopy of vitrified specimens,” Q. Rev. Biophys., vol. 21, no. 2, pp. 129–228, 1988, https://doi.org/10.1017/s0033583500004297.Suche in Google Scholar PubMed
[10] R. F. Thompson, M. Walker, C. A. Siebert, S. P. Muench, and N. A. Ranson, “An introduction to sample preparation and imaging by cryo-electron microscopy for structural biology,” Methods, vol. 100, no. 100, pp. 3–15, 2016. https://doi.org/10.1016/j.ymeth.2016.02.017.Suche in Google Scholar PubMed PubMed Central
[11] R. A. Grassucci, D. J. Taylor, and J. Frank, “Preparation of macromolecular complexes for cryo-electron microscopy,” Nat. Protoc., vol. 2, no. 12, pp. 3239–46, 2007, https://doi.org/10.1038/nprot.2007.452.Suche in Google Scholar PubMed PubMed Central
[12] J. S. Depelteau, G. Koning, W. Yang, and A. Briegel, “An economical, portable manual cryogenic plunge freezer for the preparation of vitrified biological samples for cryogenic electron microscopy,” Microsc. Microanal., vol. 26, no. 3, pp. 413–8, 2020, https://doi.org/10.1017/s1431927620001385.Suche in Google Scholar PubMed
[13] P. M. Frederik and M. M. Storms, “Automated, robotic preparation of vitrified samples for 2D and 3D cryo electron microscopy,” Microsc. Today, vol. 13, no. 6, pp. 32–9, 2005, https://doi.org/10.1017/s1551929500053967.Suche in Google Scholar
[14] L. Melanson, “Cryoplunge™ 3 and solarus® 950: A perfect duet for high quality frozen hydrated specimen preparations for CryoEM,” Microsc. Microanal., vol. 15, no. S2, pp. 2–3, 2009, https://doi.org/10.1017/s1431927609093647.Suche in Google Scholar
[15] T. Jain, P. Sheehan, J. Crum, B. Carragher, and C. S. Potter, “Spotiton: A prototype for an integrated inkjet dispense and vitrification system for cryo-TEM,” J. Struct. Biol., vol. 179, no. 1, pp. 68–75, 2012, https://doi.org/10.1016/j.jsb.2012.04.020.Suche in Google Scholar PubMed PubMed Central
[16] M. C. Darrow, J. P. Moore, R. J. Walker, K. Doering, and R. S. King, “Chameleon: Next generation sample preparation for CryoEM based on spotiton,” Microsc. Microanal., vol. 25, no. S2, pp. 994–5, 2019, https://doi.org/10.1017/s1431927619005701.Suche in Google Scholar
[17] G. Weissenberger, F. Nijpels, R. Henderikx, B. Beulen, R. Ravelli, and P. Peters, “VitroJet: Bridging the gap in sample prep,” Microsc. Microanal., vol. 26, no. S2, pp. 328–, 2020, https://doi.org/10.1017/s1431927620014257.Suche in Google Scholar
[18] J. Berriman and N. Unwin, “Analysis of transient structures by cryo-microscopy combined with rapid mixing of spray droplets,” Ultramicroscopy, vol. 56, no. 4, pp. 241–52, 1994, https://doi.org/10.1016/0304-3991(94)90012-4.Suche in Google Scholar PubMed
[19] H. White, K. Thirumurugan, M. Walker, and J. Trinick, “A second generation apparatus for time-resolved electron cryo-microscopy using stepper motors and electrospray,” J. Struct. Biol., vol. 144, nos. 1–2, pp. 246–52, 2003, https://doi.org/10.1016/j.jsb.2003.09.027.Suche in Google Scholar PubMed
[20] D. Kontziampasis et al.., “A cryo-EM grid preparation device for time-resolved structural studies,” IUCrJ, vol. 6, no. 6, pp. 1024–31, 2019, https://doi.org/10.1107/s2052252519011345.Suche in Google Scholar
[21] D. P. Klebl et al.., “Sample deposition onto cryo-EM grids: From sprays to jets and back,” Biol. Crystallogr., vol. 76, no. 4, pp. 340–9, 2020, https://doi.org/10.1107/s2059798320002958.Suche in Google Scholar
[22] D. P. Klebl et al.., “Need for speed: Examining protein behavior during CryoEM grid preparation at different timescales,” Structure, vol. 28, no. 11, pp. 1238–48. e4, 2020, https://doi.org/10.1016/j.str.2020.07.018.Suche in Google Scholar PubMed PubMed Central
[23] D. P. Klebl et al.., “Swinging lever mechanism of myosin directly shown by time-resolved cryo-EM,” Nature, pp. 1–8, 2025, https://doi.org/10.1038/s41586-025-08876-5.Suche in Google Scholar PubMed PubMed Central
[24] J. Turner et al.., “EMDB—The electron microscopy data bank,” Nucl. Acids Res., vol. 52, no. D1, pp. D456–65, 2024, https://doi.org/10.1093/nar/gkad1019.Suche in Google Scholar PubMed PubMed Central
[25] K. Sader, R. Matadeen, P. Castro Hartmann, T. Halsan, and C. Schlichten, “Industrial cryo-EM facility setup and management,” Acta Crystallogr., Sect. D:Struct. Biol., vol. 76, no. 4, pp. 313–25, 2020, https://doi.org/10.1107/s2059798320002223.Suche in Google Scholar PubMed PubMed Central
[26] M. Armstrong, B.-G. Han, S. Gomez, J. Turner, D. A. Fletcher, and R. M. Glaeser, “Microscale fluid behavior during cryo-EM sample blotting,” Biophys. J., vol. 118, no. 3, pp. 708–19, 2020, https://doi.org/10.1016/j.bpj.2019.12.017.Suche in Google Scholar PubMed PubMed Central
[27] D. Halawani, A. C. LeBlanc, I. Rouiller, S. W. Michnick, M. J. Servant, and M. Latterich, “Hereditary inclusion body myopathy-linked p97/VCP mutations in the NH2 domain and the D1 ring modulate p97/VCP ATPase activity and D2 ring conformation,” Mol. Cell. Biol., vol. 29, no. 16, pp. 4484–94, 2009, https://doi.org/10.1128/MCB.00252-09.Suche in Google Scholar PubMed PubMed Central
[28] D. Hörl et al.., “BigStitcher: Reconstructing high-resolution image datasets of cleared and expanded samples,” Nat. Methods, vol. 16, no. 9, pp. 870–4, 2019, https://doi.org/10.1038/s41592-019-0501-0.Suche in Google Scholar PubMed
[29] J. Schindelin et al.., “Fiji: An open-source platform for biological-image analysis,” Nat. Methods, vol. 9, no. 7, pp. 676–82, 2012, https://doi.org/10.1038/nmeth.2019.Suche in Google Scholar PubMed PubMed Central
[30] Z. Püspöki, M. Storath, D. Sage, and M. Unser, “Transforms and operators for directional bioimage analysis: A survey,” in Focus on Bio-Image Informatics, Cham, Switzerland, Springer International Publishing, 2016, pp. 69–93.10.1007/978-3-319-28549-8_3Suche in Google Scholar PubMed
[31] M. D. Abràmoff, P. J. Magalhães, and S. J. Ram, “Image processing with ImageJ,” Biophoton. Int., vol. 11, no. 7, pp. 36–42, 2004.Suche in Google Scholar
[32] M. Arzt et al.., “LABKIT: Labeling and segmentation toolkit for big image data,” Front. Comput. Sci., vol. 4, no. 4, p. 777728, 2022. https://doi.org/10.3389/fcomp.2022.777728.Suche in Google Scholar
[33] W. J. Rice et al.., “Routine determination of ice thickness for cryo-EM grids,” J. Struct. Biol., vol. 204, no. 1, pp. 38–44, 2018, https://doi.org/10.1016/j.jsb.2018.06.007.Suche in Google Scholar PubMed PubMed Central
[34] S. Li, “Detergents and alternatives in cryo-EM studies of membrane proteins,” Acta Biochim. Biophys. Sin., vol. 54, no. 8, p. 1049, 2022, https://doi.org/10.3724/abbs.2022088.Suche in Google Scholar PubMed PubMed Central
[35] B. Li, D. Zhu, H. Shi, and X. Zhang, “Effect of charge on protein preferred orientation at the air–water interface in cryo-electron microscopy,” J. Struct. Biol., vol. 213, no. 4, p. 107783, 2021, https://doi.org/10.1016/j.jsb.2021.107783.Suche in Google Scholar PubMed
[36] B. Michon et al.., “Role of surfactants in electron cryo-microscopy film preparation,” Biophys. J., vol. 122, no. 10, pp. 1846–57, 2023, https://doi.org/10.1016/j.bpj.2023.04.016.Suche in Google Scholar PubMed PubMed Central
[37] Y. Z. Tan and J. L. Rubinstein, “Through-grid wicking enables high-speed cryoEM specimen preparation,” Acta Crystallogr., Sect. D: Struct. Biol., vol. 76, no. 11, pp. 1092–103, 2020, https://doi.org/10.1107/s0108767320099043.Suche in Google Scholar
[38] S. Pentzien, A. Conradi, R. Koter, and J. Krüger, “Cleaning of artificially soiled paper using nanosecond, picosecond and femtosecond laser pulses,” Appl. Phys. A, vol. 101, no. 2, pp. 441–6, 2010. https://doi.org/10.1007/s00339-010-5809-7.Suche in Google Scholar
[39] D. Hee Chung et al.., “Recombinant antibodies inhibit enzymatic activity of the E3 ubiquitin ligase CHIP via multiple mechanisms,” J. Biol. Chem., vol. 301, no. 3, p. 108220, 2025, https://doi.org/10.1016/j.jbc.2025.108220.Suche in Google Scholar PubMed PubMed Central
[40] S. Callegari et al.., “Structure of human PINK1 at a mitochondrial TOM-VDAC array,” Science, vol. 388, no. 6744, p. 303, 2025.10.1126/science.adu6445Suche in Google Scholar PubMed
[41] Y. Lee et al.., “Structural basis for malate-driven, pore lipid-regulated activation of the arabidopsis vacuolar anion channel ALMT9,” Nat. Commun., vol. 16, no. 1, p. 1817, 2025, https://doi.org/10.1038/s41467-025-56940-5.Suche in Google Scholar PubMed PubMed Central
[42] S. Valimehr, R. Vuillemot, M. Kazemi, S. Jonic, and I. Rouiller, “Analysis of the conformational landscape of the N-domains of the AAA ATPase p97: Disentangling the continuous conformational variability in partially symmetrical complexes,” Int. J. Mol. Sci., vol. 25, no. 6, p. 3371, 2024, https://doi.org/10.3390/ijms25063371.Suche in Google Scholar PubMed PubMed Central
[43] J. Chen, A. J. Noble, J. Y. Kang, and S. A. Darst, “Eliminating effects of particle adsorption to the air/water interface in single-particle cryo-electron microscopy: Bacterial RNA polymerase and CHAPSO,” J. Struct. Biol. X, vol. 1, p. 100005, 2019, https://doi.org/10.1016/j.yjsbx.2019.100005.Suche in Google Scholar PubMed PubMed Central
[44] I. Drulyte et al.., “Approaches to altering particle distributions in cryo-electron microscopy sample preparation,” Acta Crystallogr., Sect. D: Struct. Biol., vol. 74, no. 6, pp. 560–71, 2018, https://doi.org/10.1107/s2059798318006496.Suche in Google Scholar
Supplementary Material
This article contains supplementary material (https://doi.org/10.1515/mim-2025-0028).
© 2025 the author(s), published by De Gruyter on behalf of Thoss Media
This work is licensed under the Creative Commons Attribution 4.0 International License.
