The Volta phase plate (VPP) is a device which enables in-focus phase contrast in a transmission electron microscope (TEM) (Danev et al., 2014). It is the first phase-plate-type device to reach near-atomic resolutions in single particle cryo-EM (Chua et al., 2016; Danev and Baumeister, 2016; Khoshouei et al., 2016a, 2016b). Nowadays such resolutions are not uncommon, but there are still challenges related mainly to intrinsic properties of the target structures. Small, heterogeneous and/or flexible molecules are generally difficult to solve by the traditional defocus phase contrast (DPC) method (Cheng et al., 2015). The VPP improves image contrast and thus could help with solving the structures of such ‘difficult’ samples. Aside from the improved contrast, one of the main questions about the VPP in the cryo-EM community has been: does it in any way limit the achievable resolution as compared to DPC?

In our previous report (Danev and Baumeister, 2016) we presented an in-focus cryo-EM approach with the VPP. The in-focus method, although ideal from a theoretical point of view, has a few practical disadvantages and limitations. It requires very accurate focusing which complicates and slows down the data acquisition. Furthermore, the achievable resolution is limited by the spherical aberration of the objective lens to ~3 Å. At the time of our initial report the 3D reconstruction software did not yet support VPP phase shift (PS) in the contrast transfer function (CTF). In the meantime, such support was implemented which enabled the present work. Here, we present a hybrid approach which combines the VPP with a small amount of defocus. It simplifies the data acquisition to a level similar to DPC and solves the resolution limit issue of the in-focus approach by enabling CTF fitting and correction.

In our report on in-focus cryo-EM with the VPP we used Thermoplasma acidophilum 20S proteasome as a test sample (Danev and Baumeister, 2016). Here, we used the same sample for consistency and continuation. The defocused VPP dataset is illustrated in Figure 1. Acquiring images with defocus is easier and faster because the defocus and astigmatism can be fitted and corrected a posteriori. The VPP provides low frequency contrast which allows the use of defocus values smaller than those typically used in DPC. Because of the low defocus the images (Figure 1A) look similar to in-focus images (Figure 4A in Danev and Baumeister, 2016). The power spectrum in Figure 1B, however, clearly shows the effect of defocus with characteristic CTF rings (Thon rings). Fitting the CTFs of VPP data requires an additional PS parameter (Rohou and Grigorieff, 2015). Such fits provide a quantitative measure of the behavior of the VPP. Figure 1C shows a plot of the PS history throughout the dataset. The VPP was advanced to a new position every 1.5 hr (every ~40 images, total dose on the VPP ~50 nC). After each advance the PS drops abruptly to a low value (<0.2 π) and starts to gradually build up again. Figure 1D contains a histogram of the measured PS. The distribution has a maximum at ~0.6 π with a relatively small number of micrographs exhibiting low (<0.2 π) or high (>0.8 π) PS. The evolving PS of the VPP is an advantage for single particle analysis because it moves the positions of the CTF zeros thus mitigating the need to vary the defocus, which is necessary with DPC. With the VPP, datasets can be collected with a single, low defocus value.

Figure 1

Download asset Open asset

In this work (Figure 1C) as well as in the previous report (Figure 3 in Danev and Baumeister, 2016) the VPP exhibited more PS than in the original VPP paper (Figure 1C in Danev et al., 2014). In addition, there is some variation in the PS magnitude between different areas on the phase plate (Figure 1C). In practice, it is advantageous to have a VPP with slower PS development because this allows collection of more images at each position. The heater used in the original VPP paper (Danev et al., 2014) was a pre-production prototype and the phase plate was homemade thus the production versions used in this work seem to exhibit a faster PS development. The variation of the PS across VPP areas could indicate a local variation in the quality of the amorphous carbon film. Overall, those factors do not prevent the collection of high quality data, as illustrated by the results presented here, but there is definitely room for improvement and we hope that manufacturers will take those observations into account when they develop future versions of the hardware.

The history of the measured defocus is plotted in Figure 1E. Approximately halfway through the dataset acquisition we changed the target defocus from 500 nm to 300 nm to evaluate the performance at different defocus values. The measured defocus has periodic oscillations, with ~16 image period, probably caused by local variations in the slant of the support film (waviness), which can introduce a defocus difference between the focusing and acquisition positions. Histograms of the measured defocus values are shown in Figure 1F. The distribution of the 300 nm target defocus data is wider than the 500 nm one but this seems to be caused by the systematic defocus oscillations and not by random focusing errors (Figure 1E). The 300 nm target defocus had a practical disadvantage in that fitting the CTFs of micrographs with defocus <300 nm was difficult because the CTF has fewer rings and their period is similar to power spectrum features, such as the amorphous ice ring at ~3.7 Å. In practice, the 500 nm defocus was more robust and easier to process.

High PS (>0.8 π) is undesirable because it causes CTF artifacts and thus reduces the quality of the data. Figure 2A shows examples of images with different amounts of PS. The image on the left has a low PS (0.1 π) and consequently lower contrast. The middle image is close to the optimal PS of 0.5 π and has good contrast and the best appearance. The image on the right was acquired with a high PS of 0.9 π and looks ‘blurry’. The blurriness is caused by an extra CTF maximum at very low spatial frequencies illustrated in Figure 2B. In practice, it is difficult to predict the exact shape or size of the central spot of the VPP. Factors such as specimen charging, specimen thickness and beam drift could cause the central beam to change its shape or move on the phase plate and thus modify the central spot. We approximated the central spot of the VPP with a Gaussian function which has only one parameter – the size of the spot. This is a simple model which, although not accurate in every case, provides a satisfactory simulation of a rather unpredictable function. The bottom plot in Figure 2B shows the approximated relative PS profile of the VPP. The top plot shows CTFs for various saturation PS values. The CTFs were calculated using the formula:

Figure 2

Download asset Open asset

where φ is the phase shift of the VPP in radians, k is the spatial frequency, s is the radius of the VPP spot in reciprocal space (0.05 1/nm), △z is the defocus (500 nm), λ is the electron wavelength (0.002 nm) and C_S is the spherical aberration coefficient (2.6*10⁶ nm). In the case of an ideal phase plate the VPP spot is infinitely small, i.e. s = 0. Up to a PS of 0.5 π (dotted lines) the CTFs behave well in a sense that they rise gradually in accordance with the PS profile. For high PSs (red line) the CTF rises quickly and passes through a maximum at the point where the PS crosses 0.5 π on its way to the saturation value. This peak at very low spatial frequencies and the shift of the first CTF zero towards the low frequencies are the reasons for the blurry appearance of the high PS image in Figure 2A. The CTF model used in data processing is that of an ideal phase plate with a delta-function-like central spot (black dashed line). The large deviation between the practical and the theoretical CTFs (pink area in Figure 2B) means that high PS images are not handled optimally during processing.

Figure 2C shows PS histograms of the particles before and after 3D classification. Particles with low (<0.2 π) and high (>0.8 π) PSs were predominantly rejected indicating that such particles had low correspondence with the reference. There was no correlation between defocus and particle retention. The first few images after moving the VPP to a new position have a low PS which quickly develops (Figure 3 in Danev and Baumeister, 2016). In those images there may be noticeable PS evolution throughout the multiframe movie. A CTF fit taking into account such evolution is possible, but in general, throwing away such images has little effect on the data acquisition throughput. For high PS images, the CTF model can, in principle, be modified to include the central VPP spot which should improve their handling. However, it is best to avoid both low and high PS images which could be achieved through an improved data acquisition strategy. An optimal solution would be to incorporate PS monitoring in the automated data acquisition software through on-the-fly CTF fitting of the acquired images. Once the PS reaches a preset value, e.g. 0.8 π, the VPP is advanced to the next position and pre-irradiated for a given time to build up a desired initial PS, e.g. 0.2 π, before continuing the acquisition. The dataset presented here was acquired with less frequent changes of the VPP position and without throwing away the first few images in order to explore the effect of PS on the quality of the data.

Figure 3

Download asset Open asset

We first processed the data through the standard Relion workflow (Scheres, 2012, 2014). The results are presented in Figure 3. The reconstruction reached a resolution of 2.4 Å according to the gold-standard Fourier shell correlation (FSC) 0.143 criterion (Figure 3C). The visibility of primary structure features, such as the holes in the tyrosine and phenylalanine rings (Figure 3B), is in accordance with that estimate. The local resolution varies between 2.2 Å and 2.6 Å (Figure 3A). The significant improvement in resolution compared to our previous in-focus dataset, at 3.2 Å (Danev and Baumeister, 2016), confirms the effectiveness of the CTF correction. Figure 3E shows a plot of the resolution versus number of particles calculated using random particle subsets. The resolution improvement is consistent throughout the particle number range with a significant extension towards low particle numbers. The improvement is also a result of the fact that the defocused dataset was collected with a smaller pixel size (1.054 Å vs 1.35 Å previously) and in super-resolution mode of the K2 camera. Figure 3F contains a plot of the logarithm of the number of particles vs the squared reciprocal resolution, with a linear fit. The B-factor calculated as twice the slope of the fit (Rosenthal and Henderson, 2003) was 103 Ų. In this work we did not collect a DPC dataset because our main goal was to explore the capabilities of the VPP. We expect that with an ‘easy’ specimen, such as the 20S proteasome, DPC would have produced very similar results to those of the VPP.

We also tried a simplified data processing workflow based on the MotionCor2 software (Zheng et al., 2016). It performs movie frame alignment with local motion tracking and dose-weighting. Such preprocessing provides similar performance gains to particle polishing in Relion (Scheres, 2014) but is more efficient in terms of computation and storage space. The results was, for all practical purposes, virtually identical to that from particle polishing both in terms of resolution (2.42 ŠMotionCor2 vs 2.36 Špolishing) and B-factor (85 Ų MotionCor2 vs 74 Ų polishing). Because of its simplicity, speed and comparable performance, nowadays we use the MotionCor2 workflow for most of our projects.

In summary, the use of VPP with defocus simplifies the data acquisition and improves the acquisition speed. The evolving PS of the VPP allows the use of a constant, low target defocus. In our experience, 500 nm is a good practical value because it ensures reliable CTF fits with some margin for random or systematic focusing errors. Future modifications of the data acquisition software, to include on-the-fly CTF fitting, PS monitoring and VPP pre-irradiation, would improve the performance by preventing the collection of low and high PS images. Here, we demonstrated that the VPP is not limited in resolution and matches the state of the art of DPC for ‘easy’ samples, such as the 20S proteasome. The real performance advantage of the VPP needs further exploration using ‘difficult’ samples, such as small, flexible and/or heterogeneous samples.

1. 1. Danev R
   2. Tegunov D
   3. Baumeister W

   (2016) Volta phase plate with defocus cryo-EM dataset of Thermoplasma acidophilum 20S proteasome

   Publicly available at the Electron Microscopy Pilot Image Archive (accession no: EMPIAR-10078).

   https://www.ebi.ac.uk/pdbe/emdb/empiar/entry/10078/
2. 1. Danev R
   2. Tegunov D
   3. Baumeister W

   (2016) Thermoplasma acidophilum 20S proteasome map reconstructed with Relion 2.0 particle polishing from defocused Volta phase plate cryo-EM data

   Publicly available at the EMDB Protein Data Bank in Europe (accession no: EMD-3455).

   http://www.ebi.ac.uk/pdbe/entry/emdb/EMD-3455
3. 1. Danev R
   2. Tegunov D
   3. Baumeister W

   (2016) Thermoplasma acidophilum 20S proteasome map reconstructed with MotionCor2 and Relion 2.0 from defocused Volta phase plate cryo-EM data

   Publicly available at the EMDB Protein Data Bank in Europe (accession no: EMD-3456).

   http://www.ebi.ac.uk/pdbe/entry/emdb/EMD-3456

1. 1. Campbell MG
   2. Cheng A
   3. Brilot AF
   4. Moeller A
   5. Lyumkis D
   6. Veesler D
   7. Pan J
   8. Harrison SC
   9. Potter CS
   10. Carragher B
   11. Grigorieff N

   (2012) Movies of ice-embedded particles enhance resolution in electron cryo-microscopy

   Structure 20:1823–1828.

   https://doi.org/10.1016/j.str.2012.08.026

   • PubMed
   • Google Scholar
2. 1. Campbell MG
   2. Veesler D
   3. Cheng A
   4. Potter CS
   5. Carragher B

   (2015) 2.8 å resolution reconstruction of the Thermoplasma acidophilum 20S proteasome using cryo-electron microscopy

   eLife 4:e06380.

   https://doi.org/10.7554/eLife.06380

   • Google Scholar
3. 1. Cheng Y
   2. Grigorieff N
   3. Penczek PA
   4. Walz T

   (2015) A primer to single-particle cryo-electron microscopy

   Cell 161:438–449.

   https://doi.org/10.1016/j.cell.2015.03.050

   • PubMed
   • Google Scholar
4. 1. Chua EY
   2. Vogirala VK
   3. Inian O
   4. Wong AS
   5. Nordenskiöld L
   6. Plitzko JM
   7. Danev R
   8. Sandin S

   (2016) 3.9 å structure of the nucleosome core particle determined by phase-plate cryo-EM

   Nucleic Acids Research 44:8013–8019.

   https://doi.org/10.1093/nar/gkw708

   • PubMed
   • Google Scholar
5. 1. Danev R
   2. Baumeister W

   (2016) Cryo-EM single particle analysis with the Volta phase plate

   eLife 5:e13046.

   https://doi.org/10.7554/eLife.13046

   • PubMed
   • Google Scholar
6. 1. Danev R
   2. Buijsse B
   3. Khoshouei M
   4. Plitzko JM
   5. Baumeister W

   (2014) Volta potential phase plate for in-focus phase contrast transmission electron microscopy

   PNAS 111:15635–15640.

   https://doi.org/10.1073/pnas.1418377111

   • PubMed
   • Google Scholar
7. 1. Grant T
   2. Grigorieff N

   (2015) Automatic estimation and correction of anisotropic magnification distortion in electron microscopes

   Journal of Structural Biology 192:204–208.

   https://doi.org/10.1016/j.jsb.2015.08.006

   • PubMed
   • Google Scholar
8. 1. Heymann JB
   2. Belnap DM

   (2007) Bsoft: image processing and molecular modeling for electron microscopy

   Journal of Structural Biology 157:3–18.

   https://doi.org/10.1016/j.jsb.2006.06.006

   • PubMed
   • Google Scholar
9. Preprint

   1. Khoshouei M
   2. Radjainia M
   3. Baumeister W
   4. Danev R

   (2016a) Cryo-EM structure of haemoglobin at 3.2 A determined with the volta phase plate

   bioRxiv.

   https://doi.org/10.1101/087841

   • Google Scholar
10. 1. Khoshouei M
    2. Radjainia M
    3. Phillips AJ
    4. Gerrard JA
    5. Mitra AK
    6. Plitzko JM
    7. Baumeister W
    8. Danev R

    (2016b) Volta phase plate cryo-EM of the small protein complex Prx3

    Nature Communications 7:10534.

    https://doi.org/10.1038/ncomms10534

    • PubMed
    • Google Scholar
11. Preprint

    1. Kimanius D
    2. Forsberg BO
    3. Scheres S
    4. Lindahl E

    (2016) Accelerated cryo-EM structure determination with parallelisation using GPUs in RELION-2

    bioRxiv.

    https://doi.org/10.1101/059717

    • Google Scholar
12. 1. Mastronarde DN

    (2005) Automated electron microscope tomography using robust prediction of specimen movements

    Journal of Structural Biology 152:36–51.

    https://doi.org/10.1016/j.jsb.2005.07.007

    • PubMed
    • Google Scholar
13. 1. Mastronarde DN
    2. Held SR

    (2016) Automated tilt series alignment and tomographic reconstruction in IMOD

    Journal of Structural Biology.

    https://doi.org/10.1016/j.jsb.2016.07.011

    • PubMed
    • Google Scholar
14. 1. McMullan G
    2. Vinothkumar KR
    3. Henderson R

    (2015) Thon rings from amorphous ice and implications of beam-induced brownian motion in single particle electron cryo-microscopy

    Ultramicroscopy 158:26–32.

    https://doi.org/10.1016/j.ultramic.2015.05.017

    • PubMed
    • Google Scholar
15. 1. Rohou A
    2. Grigorieff N

    (2015) CTFFIND4: Fast and accurate defocus estimation from electron micrographs

    Journal of Structural Biology 192:216–221.

    https://doi.org/10.1016/j.jsb.2015.08.008

    • PubMed
    • Google Scholar
16. 1. Rosenthal PB
    2. Henderson R

    (2003) Optimal determination of particle orientation, absolute hand, and contrast loss in single-particle electron cryomicroscopy

    Journal of Molecular Biology 333:721–745.

    https://doi.org/10.1016/j.jmb.2003.07.013

    • PubMed
    • Google Scholar
17. 1. Scheres SH

    (2012) RELION: implementation of a Bayesian approach to cryo-EM structure determination

    Journal of Structural Biology 180:519–530.

    https://doi.org/10.1016/j.jsb.2012.09.006

    • PubMed
    • Google Scholar
18. 1. Scheres SH

    (2014) Beam-induced motion correction for sub-megadalton cryo-EM particles

    eLife 3:e03665.

    https://doi.org/10.7554/eLife.03665

    • PubMed
    • Google Scholar
19. Preprint

    1. Zheng S
    2. Palovcak E
    3. Armache J-P
    4. Cheng Y
    5. Agard D

    (2016) Anisotropic correction of Beam-induced motion for improved Single-particle electron Cryo-microscopy

    bioRxiv.

    https://doi.org/10.1101/061960

    • Google Scholar

Author details

1. Radostin Danev

   Department of Molecular Structural Biology, Max Planck Institute of Biochemistry, Martinsried, Germany

   Contribution

   RD, Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing—original draft, Project administration, Writing—review and editing

   For correspondence

   danev@biochem.mpg.de

   Competing interests

   RD: RD is a co-inventor in US patent US9129774 B2 "Method of using a phase plate in a transmission electron microscope"

   "This ORCID iD identifies the author of this article:" 0000-0001-6406-8993

2. Dimitry Tegunov

   Department of Molecular Biology, Max Planck Institute for Biophysical Chemistry, Göttingen, Germany

   Contribution

   DT, Software, Formal analysis, Validation, Visualization, Writing—original draft, Writing—review and editing

   Competing interests

   No competing interests declared.

   "This ORCID iD identifies the author of this article:" 0000-0001-7019-3221

3. Wolfgang Baumeister

   Department of Molecular Structural Biology, Max Planck Institute of Biochemistry, Martinsried, Germany

   Contribution

   WB, Conceptualization, Supervision, Funding acquisition, Investigation, Writing—original draft, Project administration, Writing—review and editing

   Competing interests

   WB: WB is on the Scientific Advisory Board of FEI Company

• 13,156

  views

• 1,766

  downloads

• 114

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.