G. McMullan

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

Three-dimensional (3D) structure determination by single particle electron cryomicroscopy (cryoEM) involves the calculation of an initial 3D model, followed by extensive iterative improvement of the orientation determination of the individual particle images and the resulting 3D map. Because there is much more noise than signal at high resolution in the images, this creates the possibility of noise reinforcement in the 3D map, which can give a false impression of the resolution attained. The balance between signal and noise in the final map at its limiting resolution depends on the image processing procedure and is not easily predicted. There is a growing awareness in the cryoEM community of how to avoid such over-fitting and over-estimation of resolution. Equally, there has been a reluctance to use the two principal methods of avoidance because they give lower resolution estimates, which some people believe are too pessimistic. Here we describe a simple test that is compatible with any image processing protocol. The test allows measurement of the amount of signal and the amount of noise from overfitting that is present in the final 3D map. We have applied the method to two different sets of cryoEM images of the enzyme beta-galactosidase using several image processing packages. Our procedure involves substituting the Fourier components of the initial particle image stack beyond a chosen resolution by either the Fourier components from an adjacent area of background, or by simple randomisation of the phases of the particle structure factors. This substituted noise thus has the same spectral power distribution as the original data. Comparison of the Fourier Shell Correlation (FSC) plots from the 3D map obtained using the experimental data with that from the same data with high-resolution noise (HR-noise) substituted allows an unambiguous measurement of the amount of overfitting and an accompanying resolution assessment. A simple formula can be used to calculate an unbiased FSC from the two curves, even when a substantial amount of overfitting is present. The approach is software independent. The user is therefore completely free to use any established method or novel combination of methods, provided the HR-noise test is carried out in parallel. Applying this procedure to cryoEM images of beta-galactosidase shows how overfitting varies greatly depending on the procedure, but in the best case shows no overfitting and a resolution of ~6 Å. (382 words)

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

Although electron cryo-microscopy (cryo-EM) single-particle analysis has become an important tool for structural biology of large and flexible macro-molecular assemblies, the technique has not yet reached its full potential. Besides fundamental limits imposed by radiation damage, poor detectors and beam-induced sample movement have been shown to degrade attainable resolutions. A new generation of direct electron detectors may ameliorate both effects. Apart from exhibiting improved signal-to-noise performance, these cameras are also fast enough to follow particle movements during electron irradiation. Here, we assess the potentials of this technology for cryo-EM structure determination. Using a newly developed statistical movie processing approach to compensate for beam-induced movement, we show that ribosome reconstructions with unprecedented resolutions may be calculated from almost two orders of magnitude fewer particles than used previously. Therefore, this methodology may expand the scope of high-resolution cryo-EM to a broad range of biological specimens. DOI: http://dx.doi.org/10.7554/eLife.00461.001

How cryo-EM is revolutionizing structural biology

For many years, structure determination of biological macromolecules by cryo-electron microscopy (cryo-EM) was limited to large complexes or low-resolution models. With recent advances in electron detection and image processing, the resolution by cryo-EM is now beginning to rival X-ray crystallography. A new generation of electron detectors record images with unprecedented quality, while new image-processing tools correct for sample movements and classify images according to different structural states. Combined, these advances yield density maps with sufficient detail to deduce the atomic structure for a range of specimens. Here, we review the recent advances and illustrate the exciting new opportunities that they offer to structural biology research.

Detective quantum efficiency of electron area detectors in electron microscopy.

Recent progress in detector design has created the need for a careful side-by-side comparison of the modulation transfer function (MTF) and resolution-dependent detective quantum efficiency (DQE) of existing electron detectors with those of detectors based on new technology. We present MTF and DQE measurements for four types of detector: Kodak SO-163 film, TVIPS 224 charge coupled device (CCD) detector, the Medipix2 hybrid pixel detector, and an experimental direct electron monolithic active pixel sensor (MAPS) detector. Film and CCD performance was measured at 120 and 300 keV, while results are presented for the Medipix2 at 120 keV and for the MAPS detector at 300 keV. In the case of film, the effects of electron backscattering from both the holder and the plastic support have been investigated. We also show that part of the response of the emulsion in film comes from light generated in the plastic support. Computer simulations of film and the MAPS detector have been carried out and show good agreement with experiment. The agreement enables us to conclude that the DQE of a backthinned direct electron MAPS detector is likely to be equal to, or better than, that of film at 300 keV.

Comparison of optimal performance at 300 keV of three direct electron detectors for use in low dose electron microscopy

Low dose electron imaging applications such as electron cryo-microscopy are now benefitting from the improved performance and flexibility of recently introduced electron imaging detectors in which electrons are directly incident on backthinned CMOS sensors. There are currently three commercially available detectors of this type: the Direct Electron DE-20, the FEI Falcon II and the Gatan K2 Summit. These have different characteristics and so it is important to compare their imaging properties carefully with a view to optimise how each is used. Results at 300 keV for both the modulation transfer function (MTF) and the detective quantum efficiency (DQE) are presented. Of these, the DQE is the most important in the study of radiation sensitive samples where detector performance is crucial. We find that all three detectors have a better DQE than film. The K2 Summit has the best DQE at low spatial frequencies but with increasing spatial frequency its DQE falls below that of the Falcon II.

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

The introduction of direct electron detectors with higher detective quantum efficiency and fast read-out marks the beginning of a new era in electron cryo-microscopy. Using the FEI Falcon II direct electron detector in video mode, we have reconstructed a map at 3.36 Å resolution of the 1.2 MDa F420-reducing hydrogenase (Frh) from methanogenic archaea from only 320,000 asymmetric units. Videos frames were aligned by a combination of image and particle alignment procedures to overcome the effects of beam-induced motion. The reconstructed density map shows all secondary structure as well as clear side chain densities for most residues. The full coordination of all cofactors in the electron transfer chain (a [NiFe] center, four [4Fe4S] clusters and an FAD) is clearly visible along with a well-defined substrate access channel. From the rigidity of the complex we conclude that catalysis is diffusion-limited and does not depend on protein flexibility or conformational changes. DOI: http://dx.doi.org/10.7554/eLife.01963.001

Experimental observation of the improvement in MTF from backthinning a CMOS direct electron detector

The advantages of backthinning monolithic active pixel sensors (MAPS) based on complementary metal oxide semiconductor (CMOS) direct electron detectors for electron microscopy have been discussed previously; they include better spatial resolution (modulation transfer function or MTF) and efficiency at all spatial frequencies (detective quantum efficiency or DQE). It was suggested that a ‘thin’ CMOS detector would have the most outstanding properties [1–3] because of a reduction in the proportion of backscattered electrons. In this paper we show, theoretically (using Monte Carlo simulations of electron trajectories) and experimentally that this is indeed the case. The modulation transfer functions of prototype backthinned CMOS direct electron detectors have been measured at 300 keV. At zero spatial frequency, in non-backthinned 700-μm-thick detectors, the backscattered component makes up over 40% of the total signal but, by backthinning to 100, 50 or 35 μm, this can be reduced to 25%, 15% and 10%, respectively. For the 35 μm backthinned detector, this reduction in backscatter increases the MTF by 40% for spatial frequencies between 0.1 and 1.0 Nyquist. As discussed in the main text, reducing backscattering in backthinned detectors should also improve DQE.

Enhanced imaging in low dose electron microscopy using electron counting

We compare the direct electron imaging performance at 120 keV of a monolithic active pixel sensor (MAPS) operated in a conventional integrating mode with the performance obtained when operated in a single event counting mode. For the combination of sensor and incident electron energy used here, we propose a heuristic approach with which to process the single event images in which each event is renormalised to have an integrated weight of unity. Using this approach we find enhancements in the Nyquist frequency modulation transfer function (MTF) and detective quantum efficiency (DQE) over the corresponding integrating mode values by factors of 8 and 3, respectively.

Electronic detectors for electron microscopy.

Electron microscopy (EM) is an important tool for high-resolution structure determination in applications ranging from condensed matter to biology. Electronic detectors are now used in most applications in EM as they offer convenience and immediate feedback that is not possible with film or image plates. The earliest forms of electronic detector used routinely in transmission electron microscopy (TEM) were charge coupled devices (CCDs) and for many applications these remain perfectly adequate. There are however applications, such as the study of radiation-sensitive biological samples, where film is still used and improved detectors would be of great value. The emphasis in this review is therefore on detectors for use in such applications. Two of the most promising candidates for improved detection are: monolithic active pixel sensors (MAPS) and hybrid pixel detectors (of which Medipix2 was chosen for this study). From the studies described in this review, a back-thinned MAPS detector appears well suited to replace film in for the study of radiation-sensitive samples at 300 keV, while Medipix2 is suited to use at lower energies and especially in situations with very low count rates. The performance of a detector depends on the energy of electrons to be recorded, which in turn is dependent on the application it is being used for; results are described for a wide range of electron energies ranging from 40 to 300 keV. The basic properties of detectors are discussed in terms of their modulation transfer function (MTF) and detective quantum efficiency (DQE) as a function of spatial frequency.

Images of paraffin monolayer crystals with perfect contrast: Minimization of beam-induced specimen motion

Quantitative analysis of electron microscope images of organic and biological two-dimensional crystals has previously shown that the absolute contrast reached only a fraction of that expected theoretically from the electron diffraction amplitudes. The accepted explanation for this is that irradiation of the specimen causes beam-induced charging or movement, which in turn causes blurring of the image due to image or specimen movement. In this paper, we used three different approaches to try to overcome this image-blurring problem in monolayer crystals of paraffin. Our first approach was to use an extreme form of spotscan imaging, in which a single image was assembled on film by the successive illumination of up to 50,000 spots, each of a diameter of around 7 nm. The second approach was to use the Medipix II detector with its zero-noise readout to assemble a time-sliced series of images of the same area in which each frame from a movie with up to 400 frames had an exposure of only 500 electrons. In the third approach, we simply used a much thicker carbon support film to increase the physical strength and conductivity of the support. Surprisingly, the first two methods involving dose fractionation in space or time produced only partial improvements in contrast whereas the third approach produced many virtually perfect images, where the absolute contrast predicted from the electron diffraction amplitudes was observed in the images. We conclude that it is possible to obtain consistently almost perfect images of beam-sensitive specimens if they are attached to an appropriately strong and conductive support; however great care is needed in practice and the problem remains of how to best image ice-embedded biological structures in the absence of a strong, conductive support film.