Deep generative priors for biomolecular 3D heterogeneous reconstruction from cryo-EM projections.

Cryo-electron microscopy has become a powerful tool to determine three-dimensional (3D) structures of rigid biological macromolecules from noisy micrographs with single-particle reconstruction. Recently, deep neural networks, e.g., CryoDRGN, have demonstrated conformational and compositional heterogeneity of complexes. However, the lack of ground-truth conformations poses a challenge to assess the performance of heterogeneity analysis methods. In this work, variational autoencoders (VAE) with three types of deep generative priors were learned for latent variable inference and heterogeneous 3D reconstruction via Bayesian inference. More specifically, VAEs with "Variational Mixture of Posteriors" priors (VampPrior-SPR), non-parametric exemplar-based priors (ExemplarPrior-SPR) and priors from latent score-based generative models (LSGM-SPR) were quantitatively compared with CryoDRGN. We built four simulated datasets composed of hypothetical continuous conformation or discrete states of the hERG K + channel. Empirical and quantitative comparisons of inferred latent representations were performed with affine-transformation-based metrics. These models with more informative priors gave better regularized, interpretable factorized latent representations with better conserved pairwise distances, less deformed latent distributions and lower within-cluster variances. They were also tested on experimental datasets to resolve compositional and conformational heterogeneity (50S ribosome assembly, cowpea chlorotic mottle virus, and pre-catalytic spliceosome) with comparable high resolution. Codes and data are available: https://github.com/benjamin3344/DGP-SPR.

Structure of a zosuquidar and UIC2-bound human-mouse chimeric ABCB1.

The multidrug transporter ABCB1 (P-glycoprotein) is an ATP-binding cassette transporter that has a key role in protecting tissues from toxic insult and contributes to multidrug extrusion from cancer cells. Here, we report the near-atomic resolution cryo-EM structure of nucleotide-free ABCB1 trapped by an engineered disulfide cross-link between the nucleotide-binding domains (NBDs) and bound to the antigen-binding fragment of the human-specific inhibitory antibody UIC2 and to the third-generation ABCB1 inhibitor zosuquidar. Our structure reveals the transporter in an occluded conformation with a central, enclosed, inhibitor-binding pocket lined by residues from all transmembrane (TM) helices of ABCB1. The pocket spans almost the entire width of the lipid membrane and is occupied exclusively by two closely interacting zosuquidar molecules. The external, conformational epitope facilitating UIC2 binding is also visualized, providing a basis for its inhibition of substrate efflux. Additional cryo-EM structures suggest concerted movement of TM helices from both halves of the transporters associated with closing the NBD gap, as well as zosuquidar binding. Our results define distinct recognition interfaces of ABCB1 inhibitory agents, which may be exploited for therapeutic purposes.

MRCZ - A file format for cryo-TEM data with fast compression.

The introduction of fast CMOS detectors is moving the field of transmission electron microscopy into the computer science field of big data. Automated data pipelines control the instrument and initial processing steps which imposes more onerous data transfer and archiving requirements. Here we conduct a technical demonstration whereby storage and read/write times are improved 10× at a dose rate of 1 e-/pix/frame for data from a Gatan K2 direct-detection device by combination of integer decimation and lossless compression. The example project is hosted at github.com/em-MRCZ and released under the BSD license.

Robust image alignment for cryogenic transmission electron microscopy.

Cryo-electron microscopy recently experienced great improvements in structure resolution due to direct electron detectors with improved contrast and fast read-out leading to single electron counting. High frames rates enabled dose fractionation, where a long exposure is broken into a movie, permitting specimen drift to be registered and corrected. The typical approach for image registration, with high shot noise and low contrast, is multi-reference (MR) cross-correlation. Here we present the software package Zorro, which provides robust drift correction for dose fractionation by use of an intensity-normalized cross-correlation and logistic noise model to weight each cross-correlation in the MR model and filter each cross-correlation optimally. Frames are reliably registered by Zorro with low dose and defocus. Methods to evaluate performance are presented, by use of independently-evaluated even- and odd-frame stacks by trajectory comparison and Fourier ring correlation. Alignment of tiled sub-frames is also introduced, and demonstrated on an example dataset. Zorro source code is available at github.com/CINA/zorro.

Blotting-free and lossless cryo-electron microscopy grid preparation from nanoliter-sized protein samples and single-cell extracts.

We present a sample preparation method for cryo-electron microscopy (cryo-EM) that requires only 3-20nL of sample to prepare a cryo-EM grid, depending on the protocol used. The sample is applied and spread on the grid by a microcapillary. The procedure does not involve any blotting steps, and real-time monitoring allows the water film thickness to be assessed and decreased to an optimum value prior to vitrification. We demonstrate that the method is suitable for high-resolution cryo-EM and will enable alternative electron microscopy approaches, such as single-cell visual proteomics.

Practical implementation of high-resolution electron ptychography and comparison with off-axis electron holography.

Ptychography is a coherent diffractive imaging technique that can determine how an electron wave is transmitted through an object by probing it in many small overlapping regions and processing the diffraction data obtained at each point. The resulting electron transmission model describes both phase and amplitude changes to the electron wave. Ptychography has been adopted in transmission electron microscopy in recent years following advances in high-speed direct electron detectors and computer algorithms which now make the technique suitable for practical applications. Its ability to retrieve quantitative phase information at high spatial resolution makes it a plausible alternative or complement to electron holography. Furthermore, unlike off-axis electron holography, it can provide phase information without an electron bi-prism assembly or the requirement of a minimally structured region adjacent to the region of interest in the object. However, it does require a well-calibrated scanning transmission electron microscope and a well-managed workflow to manage the calibration, data acquisition and reconstruction process to yield a practical technique. Here we detail this workflow and highlight how this is greatly assisted by acquisition management software. Through experimental data and modelling we also explore the similarities and differences between high-resolution ptychography and electron holography. Both techniques show a dependence of the recovered phase on the crystalline orientation of the material which is attributable to dynamical scattering. However, the exact nature of the variation differs reflecting fundamental expectations, but nonetheless equally useful information is obtained from electron holography and the ptychographically determined object transmission function.

An algorithm for 3-D refractive index measurement in holographic confocal microscopy.

The principle of holographic confocal microscopy is introduced in this paper. An algorithm is presented to retrieve the phase information from the holograms and calculate the refractive index of the specimen from the phase. It is shown that this algorithm can be applied to a refractive index distribution in which the variation in x and z directions can be separated. The two examples used for the simulations are a linear gradient distribution and a circular distribution, which can be found in practical situations.

Solution structure of discoidal high-density lipoprotein particles with a shortened apolipoprotein A-I.

High-density lipoprotein (HDL) particles are cholesterol and lipid transport containers. Mature HDL particles destined for the liver develop through the formation of intermediate discoidal HDL particles, which are the primary acceptors for cholesterol. Here we present the three-dimensional structure of reconstituted discoidal HDL (rdHDL) particles, using a shortened construct of human apolipoprotein A-I, determined from a combination of nuclear magnetic resonance (NMR), electron paramagnetic resonance (EPR) and transmission electron microscopy (TEM) data. The rdHDL particles feature a protein double belt surrounding a lipid bilayer patch in an antiparallel fashion. The integrity of this structure is maintained by up to 28 salt bridges and a zipper-like pattern of cation-π interactions between helices 4 and 6. To accommodate a hydrophobic interior, a gross 'right-to-right' rotation of the helices after lipidation is necessary. The structure reflects the complexity required for a shuttling container to hold a fluid lipid or cholesterol interior at a protein:lipid ratio of 1:50.

Determination of localized visibility in off-axis electron holography.

Off-axis electron holography is a wavefront-split interference method for the transmission electron microscope that allows the phase shift and amplitude of the electron wavefront to be separated and quantitatively measured. An additional, third component of the holographic signal is the coherence of the electron wavefront. Historically, wavefront coherence has been evaluated by measurement of the holographic fringe visibility (or contrast) based on the minimum and maximum intensity values. We present a method based on statistical moments is presented that allows allow the visibility to be measured in a deterministic and reproducible fashion suitable for quantitative analysis. We also present an algorithm, based on the Fourier-ratio method, which allows the visibility to be resolved in two-dimensions, which we term the local visibility. The local visibility may be used to evaluate the loss of coherence due to electron scattering within a specimen, or as an aid in image analysis and segmentation. The relationship between amplitude and visibility may be used to evaluate the composition and mass thickness of a specimen by means of a 2-D histogram. Results for a selection of elements (C, Al, Si, Ti, Cr, Cu, Ge, and Au) are provided. All presented visibility metrics are biased at low-dose conditions by the presence of shot-noise, for which we provide methods for empirical normalization to achieve linear response.

Phase measurement error in summation of electron holography series.

Off-axis electron holography is a method for the transmission electron microscope (TEM) that measures the electric and magnetic properties of a specimen. The electrostatic and magnetic potentials modulate the electron wavefront phase. The error in measurement of the phase therefore determines the smallest observable changes in electric and magnetic properties. Here we explore the summation of a hologram series to reduce the phase error and thereby improve the sensitivity of electron holography. Summation of hologram series requires independent registration and correction of image drift and phase wavefront drift, the consequences of which are discussed. Optimization of the electro-optical configuration of the TEM for the double biprism configuration is examined. An analytical model of image and phase drift, composed of a combination of linear drift and Brownian random-walk, is derived and experimentally verified. The accuracy of image registration via cross-correlation and phase registration is characterized by simulated hologram series. The model of series summation errors allows the optimization of phase error as a function of exposure time and fringe carrier frequency for a target spatial resolution. An experimental example of hologram series summation is provided on WS2 fullerenes. A metric is provided to measure the object phase error from experimental results and compared to analytical predictions. The ultimate experimental object root-mean-square phase error is 0.006 rad (2π/1050) at a spatial resolution less than 0.615 nm and a total exposure time of 900 s. The ultimate phase error in vacuum adjacent to the specimen is 0.0037 rad (2π/1700). The analytical prediction of phase error differs with the experimental metrics by +7% inside the object and -5% in the vacuum, indicating that the model can provide reliable quantitative predictions.

Characterization of detector modulation-transfer function with noise, edge, and holographic methods.

We developed a new method for characterization of detector performance used in the transmission electron microscope (TEM) based on the measured contrast of holographic fringes. The new method changes spatial frequency of the measured holographic fringes, generated by an electrostatic biprism and Schottky or cold field-emission gun, to sample the modulation-transfer function (MTF) of the detector. The MTF of a Gatan Ultrascan™ 1000 charged-coupled detector (CCD) is evaluated using the new method and the results are compared to the established noise and slanted-edge method results. Requirements for accuracy of the edge and noise MTF methods are discussed. We consider issues surrounding incomplete read-out and how it affects the gain reference normalization of the detector. We evaluate how the MTF affects optimization of experimental parameters in the TEM.