Many microbes rely on enzymes known as hydrogenases to catalyse the metabolic reactions that generate energy. These enzymes cleave hydrogen molecules to release electrons that go on to participate in further reactions. In order to fully understand how hydrogenases and other enzymes work it is necessary to work out their structure at the atomic level.

Last year a technique known as electron cryo-microscopy (cryo-EM) was used to show that Frh—a hydrogenase that is crucial for many different steps in the metabolic process of microbes that produce methane—had a tetrahedral structure. Cryo-EM involves freezing the molecule of interest in a layer of ice to preserve its structure as it is imaged with an electron beam. Unfortunately, the signal-to-noise ratio in each image is low, so researchers must combine many separate images in order to determine the structure of the molecule.

The use of a new type of electron detector can improve the performance of an electron cryo-microscope in several ways. Higher frame rates can be used, which makes it possible to correct for movement of the molecule caused by the electron beam. The new electron detectors are also more efficient, so samples can be exposed to lower doses of electrons, reducing damage to the sample.

Using the new direct electron detectors, Allegretti et al. were able to work out the structure of Frh in greater detail than before. The results confirm that the previously reported structure is correct. Furthermore, several new structural features were seen for the first time, including a previously unseen ion located between two protein subunits. Allegretti et al. also revealed that the structure of Frh is highly rigid, and so the process by which it catalyses reactions involving its substrate, the coenzyme F₄₂₀, does not involve changes in its shape. Instead, the reaction rate depends on the rate at which F₄₂₀ diffuses to the correct position in the enzyme, where the reaction occurs very rapidly.

It has long been recognized that electron cryo-microscopy (cryo-EM) has the potential to solve protein structures at near-atomic resolution (Henderson, 1995). Since the first sub-nanometer resolution cryo-EM structures of icosahedral viruses, which have the advantage of multi-MDa size and high symmetry (Böttcher et al., 1997; Conway et al., 1997), steady advances in instrumentation have yielded a number of virus structures at better than 4 Å resolution (Grigorieff and Harrison, 2011). For smaller and less symmetrical structures, progress has been slower, but several complexes have yielded structures at around 6 Å resolution where secondary structure is recognizable and X-ray structures of subcomplexes can be reliably fitted (Ludtke et al., 2004; Armache et al., 2010; Gipson et al., 2010). In a few cases, side chain densities have been resolved (Cong et al., 2010; Ludtke and Baker, 2008; Zhang et al., 2010, 2013; Mills et al., 2013). Radiation damage limits the electron dose that can be used in any individual EM image and the resulting low signal-to-noise ratio (SNR) has meant that until recently high-resolution maps have required the equivalent of 10⁶ images of asymmetric units.

The recent introduction of direct electron detection cameras with much better detective quantum efficiency (DQE) has meant that images having higher SNR can be obtained with the same, or lower, total electron exposure. In addition, these cameras have frame rates which make it possible to collect videos of the particles in the thin film of vitreous water during an exposure, and thus partially correct for the effects of beam-induced particle movement and specimen drift. Using a 70-MDa virus that can be accurately aligned even at short exposures, it was shown that electron irradiation causes random translations and rotations of the particles, which are typically largest in the first few frames (Brilot et al., 2012; Campbell et al., 2012). In two recent studies, image processing schemes were developed to correct for these beam-induced motions. Using the 700 kDa archaeal 20S proteasome, which is too small to reliably detect and align in images recorded with a short exposure, a protocol was developed to align entire frames or subareas of >2000 × 2000 pixels to each other (Li et al., 2013) (The motion correction software is available to download from http://www.nature.com/nmeth/journal/v10/n6/extref/nmeth.2472-S2.zip). For the much larger ribosome (∼4 MDa) a statistical video processing approach was developed, which acts on individual particles using a user-defined running average of video frames (Bai et al., 2013). Both studies produced maps of unprecedented resolution, 3.3 Å for the D7 proteasome from 126,000 particles and 4.5 Å from 35,000 asymmetric 80S ribosomes.

Recently we determined the structure of the 1.2 MDa Frh complex, the F₄₂₀-dependent hydrogenase from Methanothermobacter marburgensis, ab initio from cryo-EM data collected on photographic film (Mills et al., 2013). Frh is a key enzyme in the metabolism of methanogenic archaea, where the reduction of carbon dioxide to methane involves the oxidation of four molecules of H₂ by a number of different hydrogenases. The reduced form of the F₄₂₀ coenzyme, which is the electron donor in several of these steps, is regenerated by Frh (Thauer et al., 2010). The central role of Frh in the metabolism of methanogens is reflected in its abundance (∼2%) in the soluble cell protein (Fox et al., 1987). Frh is a heterotrimeric enzyme composed of the 43 kDa subunit FrhA that contains a [NiFe]-center, the 26 kDa subunit FrhG with three [4Fe4S] clusters, and the 31 kDa iron–sulphur flavoprotein FrhB, which contains the F₄₂₀-binding site and has one [4Fe4S] cluster and an FAD. Our cryo-EM study showed that the Frh complex is a dodecamer with tetrahedral symmetry. The map had ∼5 Å resolution which was sufficient to show secondary structure and density for many side chains as well as the cofactors forming the electron transfer chain, making it possible to trace the three protein chains, one of them without a template.

We have now collected a new dataset of Frh as videos on a back-thinned Falcon II detector and reconstructed a cryo-EM using a ‘gold standard’ refinement approach. The resulting map from 26,000 particles has a resolution of 3.36 Å, which enabled us to refine the model obtained from film data. We can now trace the three proteins in the complex completely, and localize most side chains with confidence. We compare and discuss different processing schemes to deal with the video data.

Images of the Frh complex were collected using the back-thinned FEI Falcon-II in video mode with 18 frames per second. Using videos allows long exposures in which the individual particles and Thon rings are clearly visible, while retaining the radiation-sensitive higher resolution information only present during the initial exposure. The particles were clearly distinguishable below 1 μm defocus with a 1.5 s exposure (∼90 e/Ų) (Figure 1A). A dataset of 33,590 particles (403,080 asymmetric units) at a defocus range of 0.8–3.5 µm was collected from 235 images in two sessions on a single grid. Areas of very thin ice were selected that we estimate to be only slightly thicker than the particle diameter of ∼175 Å; often similar holes nearby had no particles in the center but a densely packed outer rim, so apparently the thin ice had squeezed the complexes out to the edge of the hole.

Figure 1

Download asset Open asset

The Frh dodecamer was refined from a low-pass filtered model to high resolution using the RELION gold standard refinement procedure. In this protocol, two half datasets are refined completely independently throughout, and after each cycle the new reference volumes are low-pass filtered to the resolution where the Fourier shell correlation (FSC) between the two volumes drops to 0.143, thus preventing overfitting of noise (Scheres and Chen, 2012). A first refinement using the sum of the first 20 video frames (discarding the last 4 frames because of likely radiation damage [Baker and Rubinstein, 2010]) yielded a map at 3.94 Å resolution (Figure 2A) that looked considerably better than our previous map from 84,000 particles on film. Also, the map was virtually noise-free due to the RELION refinement procedure (Scheres, 2012). Next we applied the frame alignment software (Li et al., 2013) and aligned all 24 frames to each other. In the majority of cases, the first frame showed a large movement in comparison with the others (Figure 3A). We refined the sum of the aligned frames with or without the first frame and found that discarding the first frame results in a higher-resolution map (Figure 3B). As found also by others (Li et al., 2013), the correction cannot deal with the fast motion of the particles during the first frame. On this basis we discarded the first frame of every video for all subsequent refinements. The refinement of the 20 aligned frames resulted in a 3.74 Å map (Figure 2A). In another refinement approach, we used the statistical alignment procedure (Bai et al., 2013) implemented in RELION (Scheres, 2012). This method follows the movement of the particles within the videos (particle-based method) using a running average of frames. This procedure also improved the resolution relative to the unaligned particles in a very similar way to the whole-frame alignment (Figure 2A); the best resolution was obtained when 5 frames were averaged. In yet another approach we combined the two alignment procedures (whole-frame and particle-based). The resolution curve of this refinement gave a better FSC than the others over all frequencies and the resulting reconstruction had a resolution of 3.69 Å (Figure 2A), indicating that for the intermediate-sized particle investigated here the combination of the two alignment procedures works best.

Figure 2

Download asset Open asset

Figure 3

Download asset Open asset

We had already discarded the last 4 frames of the 24-frame videos, because of likely radiation damage. We then continued to assess the effect of radiation damage on the quality of the map by successively reducing the number of frames. Every omitted frame improved the map resolution until only 6 frames were averaged. This gave the best resolution of 3.39 Å, compared to 3.74 Å for all 20 frames (Figure 2B). Reducing the number of averaged frames further to 4 or 5 made the resolution worse.

We attempted to improve the resolution further by adding more data. A third microscope session on a different grid yielded another 15,000 particles. Adding them, however, reduced the map resolution and these images were therefore not included in further steps. We assume the data quality was worse because of the thicker ice on this grid, which had made it necessary to use a higher defocus, never less than 1.2 µm, to distinguish the particles. We therefore decided to discard suboptimal images from the first dataset to improve the map. We removed the images with high defocus (>2.5 µm) and those that were of poor quality as judged by the visibility and symmetry of the Thon rings. This reduced the dataset by 20% from 33,590 to 26,635 particles (319,620 asymmetric units). Refinement of this smaller dataset under otherwise identical conditions resulted in maps with better resolution than the full dataset, and the best map, with 6 frames averaged, extended to 3.36 Å resolution (Figures 2C and 4). The accumulated dose for this data was ∼24 e⁻/Ų. We tested the local resolution of the map with the program ResMap (Kucukelbir et al., 2014) to identify possible flexible protein regions. The local resolution map was featureless (not shown), indicating that the dodecameric Frh complex is completely rigid.

Figure 4

Download asset Open asset

To assess the effects of radiation damage further, we also created maps of the last 6 and the middle 6 frames of the 20-frame videos of the reduced dataset, averaging particles that were in effect pre-irradiated with ∼24 and 49 e⁻/Ų. This resulted in maps of 3.94 and 4.16 Å resolution, respectively, which was clearly reflected in the poorer visibility of side chain densities (Figure 5).

Figure 5

Download asset Open asset

We fitted the model of the FrhABG heterotrimer based on the model built into the ∼5 Å map obtained previously from film data (Mills et al., 2013). The overall features of the 3.36 Å map were similar, but the higher resolution of the new map and the low noise level made the interpretation of the electron densities unambiguous (Figure 6; Video 1). Alpha helices (Figure 7) and beta sheets (Figure 8) were easily recognizable and the large majority of side chains had good density (Figure 9; Video 2). Most of the protein could be fitted to the map using the ‘real space refine’ feature in Coot, which had not been possible for the lower resolution map from film images. The FSC between this map and a map calculated from the fitted model is 0.5 at 3.56 Å resolution, close to the resolution of the final map, confirming the correctness of the fit (Figure 10).

Figure 6

Download asset Open asset

Video 1

Download asset

Figure 7

Download asset Open asset

Figure 8

Download asset Open asset

Figure 9

Download asset Open asset

Video 2

Download asset

Figure 10

Download asset Open asset

All three subunits were traced completely except for a few N- or C-terminal residues, accounting for 893 out of 903 amino acids (Figure 11). We found that all three proteins had been traced essentially correctly in the lower-resolution film map (Video 3). The largest differences between the two models were found in surface loops that had been difficult to trace in the film map. The only unresolved part in the previous model involved a 12-residue stretch in FrhG (188–199), for which no density was found and which was thought to form a flexible surface loop. In the new map, two nearby density features, modelled as extended loops before, were clearly recognizable as alpha-helices (194–201 and 208–215) that accommodate the missing residues. The resulting shift of residues brings the cysteine residues C206 and C208, which are conserved in the FrhG family (Figure 11B), near a strong density on the dimer axis between two FrhG subunits. The cysteines of the dimer partner are just a few Å away, and the density in between the four cysteine side chains is strongly suggestive of a coordinated ion (Figure 12). Although the cysteine side chains are not completely resolved, their tetrahedral arrangement suggests that it is a zinc ion (Dokmanić et al., 2008; Harding et al., 2010; Rulíšek and Vondrášek, 1998; Zheng et al., 2008).

Figure 11

Download asset Open asset

Video 3

Download asset

Figure 12

Download asset Open asset

The FAD cofactor in FrhB has continuous density in the map, and its folded conformation with the adenine and isoalloxazine moieties in close proximity (Mills et al., 2013) is confirmed (Figure 13; Video 4). As noted before (Mills et al., 2013), the FAD is completely surrounded by conserved residues, including loops 23–30, 71–77, and 131–138 (which includes Cys134, a ligand of the nearby FeS cluster i.e., the electron donor to FAD) (Figure 11C). The binding pocket is well-defined in the new map, with the pyrophosphate moiety at the N-terminal end of helix 27–39 and the conserved 24QDGG as an extended chain around it (Figure 13A). There is no density for the substrate, F₄₂₀, but the FAD is accessible from the surface through a ∼5 Å gap between the main body of FrhB and a domain formed by residues 128–188 containing an alpha helix and a three-stranded beta sheet (Video 5). This putative substrate access channel is, like the FAD itself, completely lined by conserved residues (Video 6). The model for the FrhABG trimer and the dodecamer is shown in Figure 14.

Figure 13

Download asset Open asset

Video 4

Download asset

Video 5

Download asset

Video 6

Download asset

Figure 14

Download asset Open asset

Cryo-EM has long had the potential of reaching near-atomic resolution, but until very recently this has only been achieved for very large, highly symmetrical icosahedral viruses. The fundamental problem is the inevitable radiation damage to the biological sample caused by the electron beam. Reaching high resolution requires low exposures to maximize the SNR of rapidly destroyed high-resolution information. The resulting images are noisy, which makes it both difficult to detect particles and to determine the defocus parameters accurately. The high frame rate of direct electron detectors allows data collection in video mode, which removes these restrictions and allows beam-induced motion of particles to be identified and partially corrected for. Just as importantly, direct electron detectors offer increased detective quantum efficiency at high resolution and so while individual images are still noisy they contain more information. The combination of better images and improved data processing is now leading to better reconstructions which in turn allows more information to be extracted from the images due to better alignment of individual particles.

The advantages are illustrated in recently published studies of two very different specimens, the 4 MDa asymmetric Saccharomyces cerevisiae ribosome and the 700 kDa D7-symmetric proteasome. These studies used two different detectors, the FEI Falcon II and the Gatan K2, respectively, and employed different strategies to align the video frames (Bai et al., 2013; Li et al., 2013). We have now reconstructed the 1.2 MDa tetrahedral Frh complex from ∼26,000 particles collected on the Falcon II, and determined the structure of the three proteins in the complex at high resolution.

The Falcon II data quality is far superior to that obtained with photographic film. With film we obtained a 5 Å map from 84,000 Frh particles (Mills et al., 2013), but with Falcon II images of the same sample and the same electron microscope we were able to achieve a map with better than 4 Å resolution from 70% fewer particles. The higher DQE of the detector makes it possible to reliably detect particles at very low defocus. This is illustrated by Figure 1A, showing a micrograph taken at 900 nm defocus in which the particles are clearly visible, whereas with film we were not able to detect the particles at less than 1500 nm defocus.

The high-resolution signal for the low-defocus images is very much better than that for images recorded at higher defocus, which was a critical factor in attaining the final map. This is illustrated by the fact that adding more particles recorded at higher defocus (from a grid with thicker ice) actually degraded the overall resolution, and that omitting particles with more than 2500 nm defocus, which did not contribute to the signal at high resolution, improved the map (Figure 2C).

A decisive advantage of direct electron detectors over film is the possibility of collecting video data. We found that two issues are important here: aligning the video frames to reduce the effects of beam-induced movement, and selecting the frame sequence used in the final map reconstruction to eliminate data affected by radiation damage. We used two approaches for video alignment. The first was the procedure developed by Li et al. (2013) for aligning whole frames (or large fractions of frames) independent of the visibility of the particles of interest in the individual frames. This was developed for the 700 kDa 20S proteasome and is most useful for relatively small protein complexes that are hard to detect on images recorded at low electron exposure. The second method is the statistical alignment procedure developed by Bai et al. (2013), which works on the particle level on user-defined running averages of substacks and requires the particles to be visible on the substacks. This was developed for ribosomes, which for a given dose are easier to detect due to their large size (several MDa) and high RNA content. The 1.2 MDa Frh complex, which contains 48 [4Fe4S] clusters and several other metal ions, is intermediate in density and size between the proteasome and the ribosome. For this complex we found that both video alignment methods gave similar improvements in final map resolution compared to unaligned frames (Figure 2A). However, the combination of the two methods, running the statistical particle alignment on pre-aligned stacks, yielded another improvement of similar magnitude. We used a running average of 5 frames for the particle-based alignment, so it is not unexpected that a pre-alignment of the frames was advantageous.

Collecting data in video mode also allows elimination of sub-optimal frames from the reconstruction. As was found in earlier studies (Brilot et al., 2012; Campbell et al., 2012; Li et al., 2013), beam-induced movement was much more noticeable in the first frame than in subsequent frames (Figure 3A). This large movement during the first fraction of the exposure would blur the first frame, and by omitting it, we obtained an improvement in resolution (Figure 3B). The later frames are increasingly affected by radiation damage. Reducing the number of frames used in the reconstruction to only 6 (out of the total 24) produced the best map (Figure 2B). The accumulated dose in these frames was ∼24 e/Ų, very similar to the accumulated dose (∼21 e/Ų) used for the 3.3 Å proteasome map (Li et al., 2013). A 4.16 Å reconstruction after a refinement of only the last 6 frames of the 20-frame videos still showed partial side chains and clear helices (Figure 5C). In case of smaller particles than the 1.2 MDa Frh complex, it would be advantageous to record videos over longer periods with higher cumulative doses, so that the particles are more visible and easier to align. The later, radiation-damaged frames can then be omitted from the reconstruction if necessary, as suggested before (Bai et al., 2013). Thus, direct electron detectors provide a way to optimize a posteriori the electron exposure of the data used for reconstruction.

Our 3.36 Å resolution map made it possible to determine the structure of the Frh complex unambiguously. All elements of secondary structure are obvious, and side chain density is clearly visible for most residues in the protein interior. Side chains on the surface are in general not seen, probably due to their flexibility in the aqueous solvent. The only conspicuous exceptions are the side chains of glutamates and aspartates, which are almost all absent (Figure 9), even in maps reconstructed from the earliest, least damaged frames (Figure 5A). An absence of aspartate and glutamate side chain densities can also be observed in other high-resolution cryo-EM maps, including the electron crystallography maps of LHC-II (Kühlbrandt et al., 1994) and bacteriorhodopsin (Kimura et al., 1997) and the recent 3.3-Å single particle map of the 20S proteasome (Li et al., 2013). In X-ray crystallography it was noted that carboxylate side chains have higher B-factors after extended exposure to intense synchrotron radiation, possibly due to decarboxylation (Weik et al., 2000). A quantification of synchrotron radiation damage on carboxylate side chains (Fioravanti et al., 2006) shows that this damage already occurs at radiation doses equivalent to less than 1 e/Ų (Henderson, 1990), a much lower dose than is feasible in cryo-EM. This implies that the decarboxylation would have occurred well within the first frame of our video data set and aspartate and glutamate side chains would in general not be visible in cryo-EM maps. However, not all are absent. Among the few visible carboxylate side chains in Frh, two are involved in ligand binding: FrhG Asp60 in the coordination of the proximal FeS cluster, and FrhA Glu44 in liganding a possible Fe or Mg ion (Mills et al., 2013; Figure 15). In X-ray crystallography it is also found that not all glutamate and aspartate side chains are equally susceptible to radiation damage, but the dependence on the structural and chemical environment is not well understood (Fioravanti et al., 2006). In the 3-Å bacteriorhodopsin EM map, the side chains of Asp85 and Asp212, which from spectroscopy are known to be deprotonated, were invisible, whereas the protonated side chains of Asp96 and Asp115 had good density (Kimura et al., 1997). Active sites are often found to contain the most radiation-sensitive residues in X-ray crystallography, but the protonation state was not found to be essential, and a relation to pKa was also not observed (Fioravanti et al., 2006). As more high-resolution cryo-EM maps become available, more data on the relative radiation sensitivity of side chains in different environments may give new insights in protein structure and dynamics.

Figure 15

Download asset Open asset

The structure of Frh was traced ab initio in our previous cryo-EM map from film data, which had a nominal overall resolution of 5.5 Å as determined by gold-standard FSC, but a significantly higher resolution of ∼4.5 Å in well-defined regions of the map (Mills et al., 2013). There were many indications that our original Frh structure was essentially correct, for example well-resolved hydrophobic side chains in the protein interior and the observation of secondary structure in regions where it was expected from prediction programs. It was nevertheless gratifying to see how well the model fits the new, higher-resolution map. Note that it would hardly be possible to trace an electron density map obtained by X-ray crystallography at a nominal resolution of 5.5 Å. This clear difference reflects the quality of the phase information, which is conserved in the phase contrast electron micrographs and determined directly by image processing, whereas in X-ray crystallography it is determined indirectly by isomorphous replacement or anomalous scattering.

Although the overall chain trace is the same, in detail the model based on the 3.36 Å map is of course much improved. The function of Frh is the hydrogenation of F₄₂₀ by molecular hydrogen. Electrons are extracted from hydrogen by the [NiFe] cluster in FrhA, and transferred via the three [4Fe4S] clusters in FrhG and another one in FrhB to FAD and from there to the substrate. The electron transfer chain is clearly recognizable. The [NiFe] cluster has a high density (Figure 15B) and is coordinated by conserved residues, as in the large subunits of other [NiFe] hydrogenases (Figure 11A). Nearby is another ion, also conserved, coordinated among others by the C-terminal histidine of FrhA, His386 (Figure 15B). The four FeS clusters were easily localized in the film map because of their high density; moreover, they are arranged in a chain with distances of ∼10 Å (Mills et al., 2013). In the Falcon II map, the clusters are not just featureless blobs, but their tetrahedral coordination (mostly by cysteine residues, or, in one case, an aspartate) can be clearly seen (Figure 6, Figure 15A). Thus their orientation can be modelled correctly, with the irons facing the sulphurs of the cysteine side chains. One unexpected feature found in the new map was the density for an ion on the dimer interface between two FrhG subunits (Figure 12). This density feature was actually visible in the film map, but since the flanking conserved cysteines from the two FrhG subunits were not recognized, its interpretation as an ion was not obvious. The function of this ion is most likely to stabilize the dodecameric complex, but its location between the two medial FeS clusters at a distance of only 9 Å, which is less than the distance of ∼10 Å between adjacent FeS clusters in the electron transfer chains, means that a role in electron transfer cannot be ruled out. The final part of the electron transfer chain, the FAD in FrhB, which is close to the last FeS cluster, has unambiguous, continuous density in the Falcon II map (Figure 13; Video 4). FAD acts as a one-electron/two-electron redox switch between the FeS clusters and the F₄₂₀ substrate (Fox et al., 1987; Alex et al., 1990; Thauer et al., 2010). This means that the substrate needs to approach the FAD to within van der Waals distance to enable hydride transfer (Ceh et al., 2009). In our previous study, we collected data from Frh with and without the substrate. In the map obtained in the presence of substrate we had identified a density near the FAD isoalloxazine ring that we interpreted as part of F₄₂₀ at low occupancy (Mills et al., 2013). For the new Falcon-II map, we used an Frh sample with a large excess of F₄₂₀, but the F₄₂₀ binding pocket is clearly empty. As the same sample was used in both studies, we conclude that the density in the film map must have been due to noise. An access channel for the substrate can easily be recognized (Video 5) and its significance is confirmed by the high conservation of the amino acid residues lining the channel (Video 6). There is no indication of even a low occupancy of substrate or of flexibility in this map region. A map of local resolution (Kucukelbir et al., 2014) confirmed that the protein region around this pocket is rigid. This suggests that conformational changes do not play a role for the access of F₄₂₀ to the hydride donor, consistent with the finding that F₄₂₀ reduction by Frh is very rapid and most likely diffusion-limited (Livingston et al., 1987).

The advent of direct electron detection cameras opens new horizons for cryo-EM. It is now possible to obtain higher-resolution maps with many fewer images. Using the video data collection mode, the optimal data in terms of SNR or radiation damage can be extracted a posteriori, and frame alignment schemes can be fine-tuned to the object under investigation.

1. 1. Allegretti M
   2. Mills DJ
   3. McMullan G
   4. Kühlbrandt W
   5. Vonck J

   (2013) Electron cryo-microscopy of F420-reducing [NiFe] hydrogenase Frh

   ID EMD-2513. In the public domain at the Electron Microscopy Data Bank (EMDB) at PDBE: http://www.ebi.ac.uk/pdbe/emdb/.

   http://www.ebi.ac.uk/pdbe-srv/emsearch/atlas/2513_summary.html
2. 1. Allegretti M
   2. Mills DJ
   3. McMullan G
   4. Kühlbrandt W
   5. Vonck J

   (2013) Atomic model of the F420-reducing [NiFe] hydrogenase by electron cryo-microscopy using a direct electron detector

   ID 4CI0. Publicly available at the RCSB Protein Data Bank (http://www.rcsb.org/pdb/).

   http://www.rcsb.org/pdb/search/structidSearch.do?structureId=4ci0

1. 1. Alex LA
   2. Reeve JN
   3. Orme-Johnson WH
   4. Walsh CT

   (1990) Cloning, sequence determination, and expression of the genes encoding the subunits of the nickel-containing 8-hydroxy-5-deazaflavin reducing hydrogenase from Methanobacterium thermoautotrophicum deltaH

   Biochemistry 29:7237–7244.

   https://doi.org/10.1021/bi00483a011

   • Google Scholar
2. 1. Armache J-P
   2. Jarasch A
   3. Anger AM
   4. Villa E
   5. Becker T
   6. Bhushan S
   7. Jossinet F
   8. Habeck M
   9. Dindar GL
   10. Franckenberg S
   11. Marquez V
   12. Mielke T
   13. Thomm M
   14. Berninghausen O
   15. Beatrix B
   16. Söding J
   17. Westhof E
   18. Wilson DN
   19. Beckmann R

   (2010) Cryo-EM structure and rRNA model of a translating eukaryotic 80S ribosome at 5.5-Å resolution

   Proceedings of the National Academy of Sciences of the United States of America 107:19748–19753.

   https://doi.org/10.1073/pnas.1009999107

   • Google Scholar
3. 1. Bai X-c
   2. Fernandez IS
   3. McMullan G
   4. Scheres SHW

   (2013) Ribosome structures to near-atomic resolution from thirty thousand cryo-EM particles

   eLife 2:00461.

   https://doi.org/10.7554/eLife.00461

   • Google Scholar
4. 1. Baker LA
   2. Rubinstein JL

   (2010) Radiation damage in electron cryomicroscopy

   Methods Enzymology 481:371–388.

   https://doi.org/10.1016/S0076-6879(10)81015-8

   • Google Scholar
5. 1. Böttcher B
   2. Wynne SA
   3. Crowther RA

   (1997) Determination of the fold of the core protein of hepatitis B virus by electron cryomicroscopy

   Nature 386:88–91.

   https://doi.org/10.1038/386088a0

   • Google Scholar
6. 1. Brilot AF
   2. Chen JZ
   3. Cheng A
   4. Pan J
   5. Harrison SC
   6. Potter CS
   7. Carragher B
   8. Henderson R
   9. Grigorieff N

   (2012) Beam-induced motion of vitrified specimen on holey carbon film

   Journal of Structural Biology 177:630–637.

   https://doi.org/10.1016/j.jsb.2012.02.003

   • Google Scholar
7. 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

   • Google Scholar
8. 1. Ceh K
   2. Demmer U
   3. Warkentin E
   4. Moll J
   5. Thauer RK
   6. Shima S
   7. Ermler U

   (2009) Structural basis of the hydride transfer mechanism in F₄₂₀-dependent methylenetetrahydromethanopterin dehydrogenase

   Biochemistry 48:10098–10105.

   https://doi.org/10.1021/bi901104d

   • Google Scholar
9. 1. Chen S
   2. McMullan G
   3. Faruqi AR
   4. Murshudov GN
   5. Short JM
   6. Scheres SHW
   7. Henderson R

   (2013) High-resolution noise substitution to measure overfitting and validate resolution in 3D structure determination by single particle electron cryomicroscopy

   Ultramicroscopy 135:24–35.

   https://doi.org/10.1016/j.ultramic.2013.06.004

   • Google Scholar
10. 1. Cong Y
    2. Baker ML
    3. Jakana J
    4. Woolford D
    5. Miller EJ
    6. Reissmann S
    7. Kumar RN
    8. Redding-Johanson AM
    9. Batth TS
    10. Mukhopadhyay A
    11. Ludtke SL
    12. Frydman J
    13. Chiu W

    (2010) 4.0-Å resolution cryo-EM structure of the mammalian chaperonin TRiC/CCT reveals its unique subunit arrangement

    Proceedings of the National Academy of Sciences of the United States of America 107:4967–4972.

    https://doi.org/10.1073/pnas.0913774107

    • Google Scholar
11. 1. Conway JF
    2. Cheng N
    3. Zlotnick A
    4. Wingfield PT
    5. Stahl SJ
    6. Steven AC

    (1997) Visualization of a 4-helix bundle in the hepatitis B virus capsid by cryo-electron microscopy

    Nature 386:91–94.

    https://doi.org/10.1038/386091a0

    • Google Scholar
12. 1. Dokmanić I
    2. Šikić M
    3. Tomić S

    (2008) Metals in proteins: correlation between the metal-ion type, coordination number and the amino-acid residues involved in the coordination

    Acta Crystallographica Section D, Biological Crystallography 64:257–263.

    https://doi.org/10.1107/S090744490706595X

    • Google Scholar
13. 1. Emsley P
    2. Cowtan K

    (2004) Coot: model-building tools for molecular graphics

    Acta Crystallographica Section D, Biological Crystallography 60:2126–2132.

    https://doi.org/10.1107/S0907444904019158

    • Google Scholar
14. 1. Fioravanti E
    2. Vellieux FMD
    3. Amara P
    4. Madern D
    5. Weik M

    (2006) Specific radiation damage to acidic residues and its relation to their chemical and structural environment

    Journal of Synchrotron Radiation 14:84–91.

    https://doi.org/10.1107/S0909049506038623

    • Google Scholar
15. 1. Fox JA
    2. Livingston DJ
    3. Orme-Johnson WH
    4. Walsh CT

    (1987) 8-Hydroxy-5-deazaflavin-reducing hydrogenase from Methanobacterium thermoautotrophicum: 1. Purification and characterization

    Biochemistry 26:4219–4227.

    https://doi.org/10.1021/bi00388a007

    • Google Scholar
16. 1. Gipson P
    2. Mills DJ
    3. Wouts R
    4. Grininger M
    5. Vonck J
    6. Kühlbrandt W

    (2010) Direct structural insight into the substrate-shuttling mechanism of yeast fatty acid synthase by electron cryo-microscopy

    Proceedings of the National Academy of Sciences of the United States of America 107:9164–9169.

    https://doi.org/10.1073/pnas.0913547107

    • Google Scholar
17. 1. Grigorieff N
    2. Harrison SC

    (2011) Near-atomic resolution reconstructions of icosahedral viruses from electron cryo-microscopy

    Current Opinion in Structural Biology 21:265–273.

    https://doi.org/10.1016/j.sbi.2011.01.008

    • Google Scholar
18. 1. Harding MM
    2. Nowicki MW
    3. Walkinshaw MD

    (2010) Metals in protein structures: a review of their principal features

    Crystallography Reviews 16:247–302.

    https://doi.org/10.1080/0889311X.2010.485616

    • Google Scholar
19. 1. Henderson R

    (1990) Cryo-protection of protein crystals against radiation damage in electron and X-ray diffraction

    Proceedings of the Royal Society of London B 241:6–8.

    https://doi.org/10.1098/rspd.1990.0057

    • Google Scholar
20. 1. Henderson R

    (1995) The potential and limitations of neutrons, electrons and X-rays for atomic resolution microscopy of unstained biological molecules

    Quarterly Reviews of Biophysics 28:171–193.

    https://doi.org/10.1017/S003358350000305X

    • Google Scholar
21. 1. Kimura Y
    2. Vassylyev DG
    3. Miyazawa A
    4. Kidera A
    5. Matsushima M
    6. Mitsuoka K
    7. Murata K
    8. Hirai T
    9. Fujiyoshi Y

    (1997) Surface of bacteriorhodopsin revealed by high-resolution electron crystallography

    Nature 389:206–211.

    https://doi.org/10.1038/38323

    • Google Scholar
22. 1. Kucukelbir A
    2. Sigworth FJ
    3. Tagare HD

    (2014) Quantifying the local resolution of cryo-EM density maps

    Nature Methods 11:63–65.

    https://doi.org/10.1038/nmeth.2727

    • Google Scholar
23. 1. Kühlbrandt W
    2. Wang DN
    3. Fujiyoshi Y

    (1994) Atomic model of plant light-harvesting complex by electron crystallography

    Nature 367:614–621.

    https://doi.org/10.1038/367614a0

    • Google Scholar
24. 1. Li X
    2. Mooney P
    3. Zheng S
    4. Booth CR
    5. Braunfeld MB
    6. Gubbens S
    7. Agard DA
    8. Cheng Y

    (2013) Electron counting and beam-induced motion correction enable near-atomic-resolution single-particle cryo-EM

    Nature Methods 10:584–590.

    https://doi.org/10.1038/nmeth.2472

    • Google Scholar
25. 1. Livingston DJ
    2. Fox JA
    3. Orme-Johnson W
    4. Walsh CT

    (1987) 8-hydroxy-5-deazaflavin-reducing hydrogenase from methanobacterium thermoautotrophicum: 2. Kinetic and hydrogen-transfer studies. Derivation of a steady-state rate equation for deazaflavin-reducing hydrogenase

    Biochemistry 26:4228–4237.

    https://doi.org/10.1021/bi00388a008

    • Google Scholar
26. 1. Ludtke SJ
    2. Baker ML
    3. chen D-H
    4. Song J-L
    5. Chuang DT
    6. Chiu W

    (2008) De novo backbone trace of GroEL from single particle electron cryomicrosopy

    Structure 16:441–448.

    https://doi.org/10.1016/j.str.2008.02.007

    • Google Scholar
27. 1. Ludtke SJ
    2. Baldwin PR
    3. Chiu W

    (1999) EMAN: semiautomated software for high-resolution single-particle reconstructions

    Journal of Structural Biology 128:82–97.

    https://doi.org/10.1006/jsbi.1999.4174

    • Google Scholar
28. 1. Ludtke SJ
    2. Chen D-H
    3. Song J-L
    4. Chuang DT
    5. Chiu W

    (2004) Seeing GroEL at 6 A resolution by single particle electron cryomicroscopy

    Structure 12:1129–1136.

    https://doi.org/10.1016/j.str.2004.05.006

    • Google Scholar
29. 1. Mills DJ
    2. Vitt S
    3. Strauss M
    4. Shima S
    5. Vonck J

    (2013) De novo modeling of the F₄₂₀-reducing [NiFe]-hydrogenase from a methanogenic archaeon by cryo-electron microscopy

    eLife 2:e00218.

    https://doi.org/10.7554/elife.00218

    • Google Scholar
30. 1. Mindell JA
    2. Grigorieff N

    (2003) Accurate determination of local defocus and specimen tilt in electron microscopy

    Journal of Structural Biology 142:334–347.

    https://doi.org/10.1016/S1047-8477(03)00069-8

    • Google Scholar
31. 1. Pettersen EF
    2. Goddard TD
    3. Huang CC
    4. Couch GS
    5. Greenblatt DM
    6. Meng EC
    7. Ferrin TE

    (2004) UCSF Chimera: a visualisation system for exploratory research and analysis

    Journal of Computational Chemistry 25:1605–1612.

    https://doi.org/10.1002/jcc.20084

    • Google Scholar
32. 1. Rosenthal P
    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

    • Google Scholar
33. 1. Rulíšek L
    2. Vondrášek J

    (1998) Coordination geometries of selected transition metal ions (Co²⁺, Ni²⁺, Cu²⁺, Zn²⁺, Cd²⁺, and Hg²⁺) in metalloproteins

    Journal of Inorganic Biochemistry 71:115–127.

    https://doi.org/10.1016/S0162-0134(98)10042-9

    • Google Scholar
34. 1. Scheres SHW

    (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

    • Google Scholar
35. 1. Scheres SHW
    2. Chen S

    (2012) Prevention of overfitting in cryo-EM structure determination

    Nature Methods 12:853–854.

    https://doi.org/10.1038/nmeth.2115

    • Google Scholar
36. 1. Tang G
    2. Peng L
    3. Baldwin PR
    4. Mann DS
    5. Jiang W
    6. Rees I
    7. Ludtke SJ

    (2007) EMAN2: an extensible image processing suite for electron microscopy

    Journal of Structural Biology 157:38–46.

    https://doi.org/10.1016/j.jsb.2006.05.009

    • Google Scholar
37. 1. Thauer RK
    2. Kaster A-K
    3. Goenrich M
    4. Schick M
    5. Hiromoto T
    6. Shima S

    (2010) Hydrogenases from methanogenic archaea, nickel, a novel cofactor, and H₂ storage

    Annual Review of Biochemistry 79:507–536.

    https://doi.org/10.1146/annurev.biochem.030508.152103

    • Google Scholar
38. 1. Weik M
    2. Ravelli RBG
    3. Kryger G
    4. McSweeney S
    5. Raves ML
    6. Harel M
    7. Gros P
    8. Silman I
    9. Kroon J
    10. Sussman JL

    (2000) Specific chemical and structural damage to proteins produced by synchrotron radiation

    Proceedings of the National Academy of Sciences of the United States of America 97:623–628.

    https://doi.org/10.1073/pnas.97.2.623

    • Google Scholar
39. 1. Zhang J
    2. Baker ML
    3. Schröder GF
    4. Douglas NR
    5. Reissmann S
    6. Jakana J
    7. Dougherty M
    8. Fu CJ
    9. Levitt M
    10. Ludtke SJ
    11. Frydman J
    12. Chiu W

    (2010) Mechanism of folding chamber closure in a group II chaperonin

    Nature 463:379–383.

    https://doi.org/10.1038/nature08701

    • Google Scholar
40. 1. Zhang Q
    2. Dai X
    3. Cong Y
    4. Zhang J
    5. Chen D-H
    6. Dougherty MT
    7. Wang J
    8. Ludtke SJ
    9. Schmid MF
    10. Chiu W

    (2013) Cryo-EM structure of a molluscan hemocyanin suggests its allosteric mechanism

    Structure 21:604–613.

    https://doi.org/10.1016/j.str.2013.02.018

    • Google Scholar
41. 1. Zheng H
    2. Chruszcz M
    3. Lasota P
    4. Lebioda L
    5. Minor W

    (2008) Data mining of metal ion environments present in protein structures

    Journal of Inorganic Biochemistry 102:1765–1776.

    https://doi.org/10.1016/j.jinorgbio.2008.05.006

    • Google Scholar

Author details

1. Matteo Allegretti

   Department of Structural Biology, Max Planck Institute of Biophysics, Frankfurt, Germany

   Contribution

   MA, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   No competing interests declared.

2. Deryck J Mills

   Department of Structural Biology, Max Planck Institute of Biophysics, Frankfurt, Germany

   Contribution

   DJM, Acquisition of data, Drafting or revising the article

   Competing interests

   No competing interests declared.

3. Greg McMullan

   Medical Research Council Laboratory of Molecular Biology, Cambridge, United Kingdom

   Contribution

   GM, Acquisition of data, Drafting or revising the article, Contributed unpublished essential data or reagents

   Competing interests

   No competing interests declared.

4. Werner Kühlbrandt

   Department of Structural Biology, Max Planck Institute of Biophysics, Frankfurt, Germany

   Contribution

   WK, Conception and design, Drafting or revising the article

   Competing interests

   WK: Reviewing editor, eLife.

5. Janet Vonck

   Department of Structural Biology, Max Planck Institute of Biophysics, Frankfurt, Germany

   Contribution

   JV, Conception and design, Analysis and interpretation of data, Drafting or revising the article

   For correspondence

   janet.vonck@biophys.mpg.de

   Competing interests

   No competing interests declared.

• 2,637

  views

• 355

  downloads

• 128

  citations

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