A novel neural network technique for analysis and classification of EM single-particle images

Machine Learning for Structure Determination in Single-Particle Cryo-Electron Microscopy: A Systematic Review

Recently, single-particle cryo-electron microscopy (cryo-EM) has become an indispensable method for determining macromolecular structures at high resolution to deeply explore the relevant molecular mechanism. Its recent breakthrough is mainly because of the rapid advances in hardware and image processing algorithms, especially machine learning. As an essential support of single-particle cryo-EM, machine learning has powered many aspects of structure determination and greatly promoted its development. In this article, we provide a systematic review of the applications of machine learning in this field. Our review begins with a brief introduction of single-particle cryo-EM, followed by the specific tasks and challenges of its image processing. Then, focusing on the workflow of structure determination, we describe relevant machine learning algorithms and applications at different steps, including particle picking, 2-D clustering, 3-D reconstruction, and other steps. As different tasks exhibit distinct characteristics, we introduce the evaluation metrics for each task and summarize their dynamics of technology development. Finally, we discuss the open issues and potential trends in this promising field.

Dynamic closed states of a ligand-gated ion channel captured by cryo-EM and simulations

Ligand-gated ion channels are critical mediators of electrochemical signal transduction across evolution. Biophysical and pharmacological characterization of these receptor proteins relies on high-quality structures in multiple, subtly distinct functional states. However, structural data in this family remain limited, particularly for resting and intermediate states on the activation pathway. Here, we report cryo-electron microscopy (cryo-EM) structures of the proton-activated Gloeobacter violaceus ligand-gated ion channel (GLIC) under three pH conditions. Decreased pH was associated with improved resolution and side chain rearrangements at the subunit/domain interface, particularly involving functionally important residues in the β1-β2 and M2-M3 loops. Molecular dynamics simulations substantiated flexibility in the closed-channel extracellular domains relative to the transmembrane ones and supported electrostatic remodeling around E35 and E243 in proton-induced gating. Exploration of secondary cryo-EM classes further indicated a low-pH population with an expanded pore. These results allow us to define distinct protonation and activation steps in pH-stimulated conformational cycling in GLIC, including interfacial rearrangements largely conserved in the pentameric channel family.

Likelihood-based structural analysis of electron microscopy images

Likelihood-based analysis of single-particle electron microscopy images has contributed much to the recent improvements in resolution. By treating particle orientations and classes probabilistically, uncertainties in the reconstruction process are explicitly accounted for, and the risk of bias towards the initial model is diminished. As a result, the quality and reliability of the reconstructions have greatly improved at manageable computational cost. Likelihood-based analysis of electron microscopy images also offers a route to direct coordinate refinement for dynamic systems, as an alternative to 3D density reconstruction. Here, we review recent developments in the algorithms used for reconstructions of high-resolution maps, and in the integrative framework of combining likelihood methods with simulations to address conformational variability in cryo-electron microscopy.

Massively parallel unsupervised single-particle cryo-EM data clustering via statistical manifold learning

Structural heterogeneity in single-particle cryo-electron microscopy (cryo-EM) data represents a major challenge for high-resolution structure determination. Unsupervised classification may serve as the first step in the assessment of structural heterogeneity. However, traditional algorithms for unsupervised classification, such as K-means clustering and maximum likelihood optimization, may classify images into wrong classes with decreasing signal-to-noise-ratio (SNR) in the image data, yet demand increased computational costs. Overcoming these limitations requires further development of clustering algorithms for high-performance cryo-EM data processing. Here we introduce an unsupervised single-particle clustering algorithm derived from a statistical manifold learning framework called generative topographic mapping (GTM). We show that unsupervised GTM clustering improves classification accuracy by about 40% in the absence of input references for data with lower SNRs. Applications to several experimental datasets suggest that our algorithm can detect subtle structural differences among classes via a hierarchical clustering strategy. After code optimization over a high-performance computing (HPC) environment, our software implementation was able to generate thousands of reference-free class averages within hours in a massively parallel fashion, which allows a significant improvement on ab initio 3D reconstruction and assists in the computational purification of homogeneous datasets for high-resolution visualization.

Structural Study of Heterogeneous Biological Samples by Cryoelectron Microscopy and Image Processing

In living organisms, biological macromolecules are intrinsically flexible and naturally exist in multiple conformations. Modern electron microscopy, especially at liquid nitrogen temperatures (cryo-EM), is able to visualise biocomplexes in nearly native conditions and in multiple conformational states. The advances made during the last decade in electronic technology and software development have led to the revelation of structural variations in complexes and also improved the resolution of EM structures. Nowadays, structural studies based on single particle analysis (SPA) suggests several approaches for the separation of different conformational states and therefore disclosure of the mechanisms for functioning of complexes. The task of resolving different states requires the examination of large datasets, sophisticated programs, and significant computing power. Some methods are based on analysis of two-dimensional images, while others are based on three-dimensional studies. In this review, we describe the basic principles implemented in the various techniques that are currently used in the analysis of structural conformations and provide some examples of successful applications of these methods in structural studies of biologically significant complexes.

The architecture of the Schizosaccharomyces pombe CCR4-NOT complex

CCR4-NOT is a large protein complex present both in cytoplasm and the nucleus of eukaryotic cells. Although it is involved in a variety of distinct processes related to expression of genetic information such as poly(A) tail shortening, transcription regulation, nuclear export and protein degradation, there is only fragmentary information available on some of its nine subunits. Here we show a comprehensive structural characterization of the native CCR4-NOT complex from Schizosaccharomyces pombe. Our cryo-EM 3D reconstruction of the complex, combined with techniques such as immunomicroscopy, RNA-nanogold labelling, docking of the available high-resolution structures and models of different subunits and domains, allow us to propose its full molecular architecture. We locate all functionally defined domains endowed with deadenylating and ubiquitinating activities, the nucleus-specific RNA-interacting subunit Mmi1, as well as surfaces responsible for protein–protein interactions. This information provides insight into cooperation of the different CCR4-NOT complex functions.

CCR4-NOT is a protein complex involved in a variety of important genetic processes. Here, the authors report the mid-resolution structure of this complex, and model the positions and contacts between the subunits, providing structural support for the previously reported functions of the complex.

Bipartite Topology of Treponema pallidum Repeat Proteins C/D and I: OUTER MEMBRANE INSERTION, TRIMERIZATION, AND PORIN FUNCTION REQUIRE A C-TERMINAL β-BARREL DOMAIN

We previously identified Treponema pallidum repeat proteins TprC/D, TprF, and TprI as candidate outer membrane proteins (OMPs) and subsequently demonstrated that TprC is not only a rare OMP but also forms trimers and has porin activity. We also reported that TprC contains N- and C-terminal domains (TprC(N) and TprC(C)) orthologous to regions in the major outer sheath protein (MOSP(N) and MOSP(C)) of Treponema denticola and that TprC(C) is solely responsible for β-barrel formation, trimerization, and porin function by the full-length protein. Herein, we show that TprI also possesses bipartite architecture, trimeric structure, and porin function and that the MOSP(C)-like domains of native TprC and TprI are surface-exposed in T. pallidum, whereas their MOSP(N)-like domains are tethered within the periplasm. TprF, which does not contain a MOSP(C)-like domain, lacks amphiphilicity and porin activity, adopts an extended inflexible structure, and, in T. pallidum, is tightly bound to the protoplasmic cylinder. By thermal denaturation, the MOSP(N) and MOSP(C)-like domains of TprC and TprI are highly thermostable, endowing the full-length proteins with impressive conformational stability. When expressed in Escherichia coli with PelB signal sequences, TprC and TprI localize to the outer membrane, adopting bipartite topologies, whereas TprF is periplasmic. We propose that the MOSP(N)-like domains enhance the structural integrity of the cell envelope by anchoring the β-barrels within the periplasm. In addition to being bona fide T. pallidum rare outer membrane proteins, TprC/D and TprI represent a new class of dual function, bipartite bacterial OMP.

Three distinct ribosome assemblies modulated by translation are the building blocks of polysomes

Translation is increasingly recognized as a central control layer of gene expression in eukaryotic cells. The overall organization of mRNA and ribosomes within polysomes, as well as the possible role of this organization in translation are poorly understood. Here we show that polysomes are primarily formed by three distinct classes of ribosome assemblies. We observe that these assemblies can be connected by naked RNA regions of the transcript. We show that the relative proportions of the three classes of ribosome assemblies reflect, and probably dictate, the level of translational activity. These results reveal the existence of recurrent supra-ribosomal building blocks forming polysomes and suggest the presence of unexplored translational controls embedded in the polysome structure.

Structural characterization of the substrate transfer mechanism in Hsp70/Hsp90 folding machinery mediated by Hop

In eukarya, chaperones Hsp70 and Hsp90 act coordinately in the folding and maturation of a range of key proteins with the help of several co-chaperones, especially Hop. Although biochemical data define the Hop-mediated Hsp70-Hsp90 substrate transfer mechanism, the intrinsic flexibility of these proteins and the dynamic nature of their complexes have limited the structural studies of this mechanism. Here we generate several complexes in the Hsp70/Hsp90 folding pathway (Hsp90:Hop, Hsp90:Hop:Hsp70 and Hsp90:Hop:Hsp70 with a fragment of the client protein glucocorticoid receptor (GR-LBD)), and determine their 3D structure using electron microscopy techniques. Our results show that one Hop molecule binds to one side of the Hsp90 dimer in both extended and compact conformations, through Hop domain rearrangement that take place when Hsp70 or Hsp70:GR-LBD bind to Hsp90:Hop. The compact conformation of the Hsp90:Hop:Hsp70:GR-LBD complex shows that GR-LBD binds to the side of the Hsp90 dimer opposite the Hop attachment site.

Enhanced production of Chikungunya virus-like particles using a high-pH adapted spodoptera frugiperda insect cell line

Chikungunya virus-like particles (VLPs) have potential to be used as a prophylactic vaccine based on testing in multiple animal models and are currently being evaluated for human use in a Phase I clinical trial. The current method for producing these enveloped alphavirus VLPs by transient gene expression in mammalian cells presents challenges for scalable and robust industrial manufacturing, so the insect cell baculovirus expression vector system was evaluated as an alternative expression technology. Subsequent to recombinant baculovirus infection of Sf21 cells in standard culture media (pH 6.2–6.4), properly processed Chikungunya structural proteins were detected and assembled capsids were observed. However, an increase in culture pH to 6.6–6.8 was necessary to produce detectable concentrations of assembled VLPs. Since this elevated production pH exceeds the optimum for growth medium stability and Sf21 culture, medium modifications were made and a novel insect cell variant (SfBasic) was derived by exposure of Sf21 to elevated culture pH for a prolonged period of time. The high-pH adapted SfBasic insect cell line described herein is capable of maintaining normal cell growth into the typical mammalian cell culture pH range of 7.0–7.2 and produces 11-fold higher Chikungunya VLP yields relative to the parental Sf21 cell line. After scale-up into stirred tank bioreactors, SfBasic derived VLPs were chromatographically purified and shown to be similar in size and structure to a VLP standard derived from transient gene expression in HEK293 cells. Total serum anti-Chikungunya IgG and neutralizing titers from guinea pigs vaccinated with SfBasic derived VLPs or HEK293 derived VLPs were not significantly different with respect to production method, suggesting that this adapted insect cell line and production process could be useful for manufacturing Chikungunya VLPs for use as a vaccine. The adaptation of Sf21 to produce high levels of recombinant protein and VLPs in an elevated pH range may also have applications for other pH-sensitive protein or VLP targets.

The structure of Sinorhizobium meliloti phage ΦM12, which has a novel T=19l triangulation number and is the founder of a new group of T4-superfamily phages

ΦM12 is the first example of a T=19l geometry capsid, encapsulating the recently sequenced genome. Here, we present structures determined by cryo-EM of full and empty capsids. The structure reveals the pattern for assembly of 1140 HK97-like capsid proteins, pointing to interactions at the pseudo 3-fold symmetry axes that hold together the asymmetric unit. The particular smooth surface of the capsid, along with a lack of accessory coat proteins encoded by the genome, suggest that this interface is the primary mechanism for capsid assembly. Two-dimensional averages of the tail, including the neck and baseplate, reveal that ΦM12 has a relatively narrow neck that attaches the tail to the capsid, as well as a three-layer baseplate. When free from DNA, the icosahedral edges expand by about 5nm, while the vertices stay at the same position, forming a similarly smooth, but bowed, T=19l icosahedral capsid.

The architecture of yeast DNA polymerase ζ

DNA polymerase ζ (Polζ) is specialized for the extension step of translesion DNA synthesis (TLS). Despite its central role in maintaining genome integrity, little is known about its overall architecture. Initially identified as a heterodimer of the catalytic subunit Rev3 and the accessory subunit Rev7, yeast Polζ has recently been shown to form a stable four-subunit enzyme (Polζ-d) upon the incorporation of Pol31 and Pol32, the accessory subunits of yeast Polδ. To understand the 3D architecture and assembly of Polζ and Polζ-d, we employed electron microscopy. We show here how the catalytic and accessory subunits of Polζ and Polζ-d are organized relative to each other. In particular, we show that Polζ-d has a bilobal architecture resembling the replicative polymerases and that Pol32 lies in proximity to Rev7. Collectively, our study provides views of Polζ and Polζ-d and a structural framework for understanding their roles in DNA damage bypass.

Maskiton: Interactive, web-based classification of single-particle electron microscopy images

Electron microscopy (EM) is an important tool for determining the composition, arrangement and structure of biological macromolecules. When studying structurally heterogeneous samples using EM, classification is a critical step toward achieving higher resolution and identifying biologically significant conformations. We have developed an interactive, web-based tool, called Maskiton, for creating custom masks and performing 2D classifications on aligned single-particle EM images. The Maskiton interface makes it considerably easier and faster to explore the significance of heterogeneity in single-particle datasets. Maskiton features include: resumable uploads to facilitate transfer of large datasets to the server, custom mask creation in the browser, continual progress updates, and interactive viewing of classification results. To demonstrate the value of this tool, we provide examples of its use on several experimental datasets and include analyses of the independent terminus mobility within the Ltn1 E3 ubiquitin ligase, the in vitro assembly of 30S ribosomal subunits, and classification complexity reduction within Immunoglobulin M. This work also serves as a proof-of-concept for the development of future cross-platform, interactive user interfaces for electron microscopy data processing.

Semiautomatic, high-throughput, high-resolution protocol for three-dimensional reconstruction of single particles in electron microscopy

In this chapter we describe the steps needed for reconstructing the three-dimensional structure of a macromolecular complex starting from its projections collected in electron micrographs. The concepts are shown through the use of Xmipp 3.0, a software suite specifically designed for the image processing of biological structures imaged with electron or X-ray microscopy. We illustrate the image processing workflow by applying it to the images of Bovine Papilloma virus published in Wolf et al. (Proc Natl Acad Sci USA 107:6298-6303, 2010). We show that in the case of high-quality, homogeneous datasets with a priori knowledge about the initial volume, we can have a high-resolution 3D reconstruction in less than 1 day using a computer cluster with only 32 processors.

The hexameric structure of a conjugative VirB4 protein ATPase provides new insights for a functional and phylogenetic relationship with DNA translocases

VirB4 proteins are ATPases essential for pilus biogenesis and protein transport in type IV secretion systems. These proteins contain a motor domain that shares structural similarities with the motor domains of DNA translocases, such as the VirD4/TrwB conjugative coupling proteins and the chromosome segregation pump FtsK. Here, we report the three-dimensional structure of full-length TrwK, the VirB4 homologue in the conjugative plasmid R388, determined by single-particle electron microscopy. The structure consists of a hexameric double ring with a barrel-shaped structure. The C-terminal half of VirB4 proteins shares a striking structural similarity with the DNA translocase TrwB. Docking the atomic coordinates of the crystal structures of TrwB and FtsK into the EM map revealed a better fit for FtsK. Interestingly, we have found that like TrwB, TrwK is able to bind DNA with a higher affinity for G4 quadruplex structures than for single-stranded DNA. Furthermore, TrwK exerts a dominant negative effect on the ATPase activity of TrwB, which reflects an interaction between the two proteins. Our studies provide new insights into the structure-function relationship and the evolution of these DNA and protein translocases.

A clarification of the terms used in comparing semi-automated particle selection algorithms in cryo-EM

Many cyro-EM datasets are heterogeneous stemming from molecules undergoing conformational changes. The need to characterize each of the substrates with sufficient resolution entails a large increase in the data flow and motivates the development of more effective automated particle selection algorithms. Concepts and procedures from the machine-learning field are increasingly employed toward this end. However, a review of recent literature has revealed a discrepancy in terminology of the performance scores used to compare particle selection algorithms, and this has subsequently led to ambiguities in the meaning of claimed performance. In an attempt to curtail the perpetuation of this confusion and to disentangle past mistakes, we review the performance of published particle selection efforts with a set of explicitly defined performance scores using the terminology established and accepted within the field of machine learning.

Classification of electron sub-tomograms with neural networks and its application to template-matching

Classification of electron sub-tomograms is a challenging task, due the missing-wedge and the low signal-to-noise ratio of the data. Classification algorithms tend to classify data according to their orientation to the missing-wedge, rather than to the underlying signal. Here we use a neural network approach, called the Kernel Density Estimator Self-Organizing Map (KerDenSOM3D), which we have implemented in three-dimensions (3D), also having compensated for the missing-wedge, and we comprehensively compare it to other classification methods. For this purpose, we use various simulated macromolecules, as well as tomographically reconstructed in vitro GroEL and GroEL/GroES molecules. We show that the performance of this classification method is superior to previously used algorithms. Furthermore, we show how this algorithm can be used to provide an initial cross-validation of template-matching approaches. For the example of sub-tomogram classification extracted from cellular tomograms of Mycoplasma pneumonia and Spiroplasma melliferum cells, we show the bias of template-matching, and by using differing search and classification areas, we demonstrate how the bias can be significantly reduced.

Structural and functional characterization of an influenza virus RNA polymerase-genomic RNA complex

The replication and transcription of influenza A virus are carried out by ribonucleoproteins (RNPs) containing each genomic RNA segment associated with nucleoprotein monomers and the heterotrimeric polymerase complex. These RNPs are responsible for virus transcription and replication in the infected cell nucleus. Here we have expressed, purified, and analyzed, structurally and functionally, for the first time, polymerase-RNA template complexes obtained after replication in vivo. These complexes were generated by the cotransfection of plasmids expressing the polymerase subunits and a genomic plasmid expressing a minimal template of positive or negative polarity. Their generation in vivo was strictly dependent on the polymerase activity; they contained mainly negative-polarity viral RNA (vRNA) and could transcribe and replicate in vitro. The three-dimensional structure of the monomeric polymerase-vRNA complexes was similar to that of the RNP-associated polymerase and distinct from that of the polymerase devoid of template. These results suggest that the interaction with the template is sufficient to induce a significant conformation switch in the polymerase complex.

The multidrug resistance half-transporter ABCG2 is purified as a tetramer upon selective extraction from membranes

ABCG2 is a human membrane ATP-binding cassette half-transporter that hydrolyzes ATP to efflux a large number of chemotherapeutic agents. Several oligomeric states of ABCG2 from homodimers to dodecamers have been reported depending on the overexpression systems and/or the protocols used for purification. Here, we compared the oligomeric state of His(6)-ABCG2 expressed in Sf9 insect cells and in human Flp-In-293/ABCG2 cells after solubilization in mild detergents. His(6)-ABCG2 was purified through a new approach involving its specific recognition onto a functionalized lipid layer containing a Ni-NTA lipid. This approach allowed the purification of His-ABCG2 in presence of all solubilized membrane components that might be involved in the stabilisation of native oligomers and without requiring any additional washing or concentration passages. ABCG2 purified onto the NiNTA lipid surfaces were directly analyzed by electron microscopy and by biochemical assays. Altogether, our data are consistent with a tetrameric organization of ABCG2 when expressed in either heterologous Sf9 insect cells or in human homologous cells.

Dynamic effects of cofactors and DNA on the oligomeric state of human mitochondrial DNA helicase

We examined the effects of cofactors and DNA on the stability, oligomeric state and conformation of the human mitochondrial DNA helicase. We demonstrate that low salt conditions result in protein aggregation that may cause dissociation of oligomeric structure. The low salt sensitivity of the mitochondrial DNA helicase is mitigated by the presence of magnesium, nucleotide, and increased temperature. Electron microscopic and glutaraldehyde cross-linking analyses provide the first evidence of a heptameric oligomer and its interconversion from a hexameric form. Limited proteolysis by trypsin shows that binding of nucleoside triphosphate produces a conformational change that is distinct from the conformation observed in the presence of nucleoside diphosphate. We find that single-stranded DNA binding occurs in the absence of cofactors and renders the mitochondrial DNA helicase more susceptible to proteolytic digestion. Our studies indicate that the human mitochondrial DNA helicase shares basic properties with the SF4 replicative helicases, but also identify common features with helicases outside the superfamily, including dynamic conformations similar to other AAA(+) ATPases.

Conformational rearrangements of SV40 large T antigen during early replication events

The Simian virus 40 (SV40) large tumor antigen (LTag) functions as the replicative helicase and initiator for viral DNA replication. For SV40 replication, the first essential step is the assembly of an LTag double hexamer at the origin DNA that will subsequently melt the origin DNA to initiate fork unwinding. In this study, we used three-dimensional cryo-electron microscopy to visualize early events in the activation of DNA replication in the SV40 model system. We obtained structures of wild-type double-hexamer complexes of LTag bound to SV40 origin DNA, to which atomic structures have been fitted. Wild-type LTag was observed in two distinct conformations: In one conformation, the central module containing the J-domains and the origin binding domains of both hexamers is a compact closed ring. In the other conformation, the central module is an open ring with a gap formed by rearrangement of the N-terminal regions of the two hexamers, potentially allowing for the passage of single-stranded DNA generated from the melted origin DNA. Double-hexamer complexes containing mutant LTag that lacks the N-terminal J-domain show the central module predominantly in the closed-ring state. Analyses of the LTag C-terminal regions reveal that the LTag hexamers bound to the A/T-rich tract origin of replication and early palindrome origin of replication elements are structurally distinct. Lastly, visualization of DNA density protruding from the LTag C-terminal domains suggests that oligomerization of the LTag complex takes place on double-stranded DNA.

An introduction to maximum-likelihood methods in cryo-EM

The maximum-likelihood method provides a powerful approach to many problems in cryo-electron microscopy (cryo-EM) image processing. This contribution aims to provide an accessible introduction to the underlying theory and reviews existing applications in the field. In addition, current developments to reduce computational costs and to improve the statistical description of cryo-EM images are discussed. Combined with the increasing power of modern computers and yet unexplored possibilities provided by theory, these developments are expected to turn the statistical approach into an essential image-processing tool for the electron microscopist.

Classification of structural heterogeneity by maximum-likelihood methods

With the advent of computationally feasible approaches to maximum-likelihood (ML) image processing for cryo-electron microscopy, these methods have proven particularly useful in the classification of structurally heterogeneous single-particle data. A growing number of experimental studies have applied these algorithms to study macromolecular complexes with a wide range of structural variability, including nonstoichiometric complex formation, large conformational changes, and combinations of both. This chapter aims to share the practical experience that has been gained from the application of these novel approaches. Current insights on how to prepare the data and how to perform two- or three-dimensional classifications are discussed together with the aspects related to high-performance computing. Thereby, this chapter will hopefully be of practical use for those microscopists wishing to apply ML methods in their own investigations.

Automation in single-particle electron microscopy connecting the pieces

Throughout the history of single-particle electron microscopy (EM), automated technologies have seen varying degrees of emphasis and development, usually depending upon the contemporary demands of the field. We are currently faced with increasingly sophisticated devices for specimen preparation, vast increases in the size of collected data sets, comprehensive algorithms for image processing, sophisticated tools for quality assessment, and an influx of interested scientists from outside the field who might lack the skills of experienced microscopists. This situation places automated techniques in high demand. In this chapter, we provide a generic definition of and discuss some of the most important advances in automated approaches to specimen preparation, grid handling, robotic screening, microscope calibrations, data acquisition, image processing, and computational infrastructure. Each section describes the general problem and then provides examples of how that problem has been addressed through automation, highlighting available processing packages, and sometimes describing the particular approach at the National Resource for Automated Molecular Microscopy (NRAMM). We contrast the more familiar manual procedures with automated approaches, emphasizing breakthroughs as well as current limitations. Finally, we speculate on future directions and improvements in automated technologies. Our overall goal is to present automation as more than simply a tool to save time. Rather, we aim to illustrate that automation is a comprehensive and versatile strategy that can deliver biological information on an unprecedented scale beyond the scope available with classical manual approaches.

A toolbox for ab initio 3-D reconstructions in single-particle electron microscopy

Structure determination of a novel macromolecular complex via single-particle electron microscopy depends upon overcoming the challenge of establishing a reliable 3-D reconstruction using only 2-D images. There are a variety of strategies that deal with this issue, but not all of them are readily accessible and straightforward to use. We have developed a "toolbox" of ab initio reconstruction techniques that provide several options for calculating 3-D volumes in an easily managed and tightly controlled work-flow that adheres to standard conventions and formats. This toolbox is designed to streamline the reconstruction process by removing the necessity for bookkeeping, while facilitating transparent data transfer between different software packages. It currently includes procedures for calculating ab initio reconstructions via random or orthogonal tilt geometry, tomograms, and common lines, all of which have been tested using the 50S ribosomal subunit. Our goal is that the accessibility of multiple independent reconstruction algorithms via this toolbox will improve the ease with which models can be generated, and provide a means of evaluating the confidence and reliability of the final reconstructed map.

Introducing robustness to maximum-likelihood refinement of electron-microscopy data

An expectation-maximization algorithm for maximum-likelihood refinement of electron-microscopy images is presented that is based on fitting mixtures of multivariate t-distributions. The novel algorithm has intrinsic characteristics for providing robustness against atypical observations in the data, which is illustrated using an experimental test set with artificially generated outliers. Tests on experimental data revealed only minor differences in two-dimensional classifications, while three-dimensional classification with the new algorithm gave stronger elongation factor G density in the corresponding class of a structurally heterogeneous ribosome data set than the conventional algorithm for Gaussian mixtures.

Maximum likelihood refinement of electron microscopy data with normalization errors

Commonly employed data models for maximum likelihood refinement of electron microscopy images behave poorly in the presence of normalization errors. Small variations in background mean or signal brightness are relatively common in cryo-electron microscopy data, and varying signal-to-noise ratios or artifacts in the images interfere with standard normalization procedures. In this paper, a statistical data model that accounts for normalization errors is presented, and a corresponding algorithm for maximum likelihood classification of structurally heterogeneous projection data is derived. The extended data model has general relevance, since similar algorithms may be derived for other maximum likelihood approaches in the field. The potentials of this approach are illustrated for two structurally heterogeneous data sets: 70S E.coli ribosomes and human RNA polymerase II complexes. In both cases, maximum likelihood classification based on the conventional data model failed, whereas the new approach was capable of revealing previously unobserved conformations.

Image processing for electron microscopy single-particle analysis using XMIPP

We describe a collection of standardized image processing protocols for electron microscopy single-particle analysis using the XMIPP software package. These protocols allow performing the entire processing workflow starting from digitized micrographs up to the final refinement and evaluation of 3D models. A particular emphasis has been placed on the treatment of structurally heterogeneous data through maximum-likelihood refinements and self-organizing maps as well as the generation of initial 3D models for such data sets through random conical tilt reconstruction methods. All protocols presented have been implemented as stand-alone, executable python scripts, for which a dedicated graphical user interface has been developed. Thereby, they may provide novice users with a convenient tool to quickly obtain useful results with minimum efforts in learning about the details of this comprehensive package. Examples of applications are presented for a negative stain random conical tilt data set on the hexameric helicase G40P and for a structurally heterogeneous data set on 70S Escherichia coli ribosomes embedded in vitrified ice.

Modeling experimental image formation for likelihood-based classification of electron microscopy data

The coexistence of multiple distinct structural states often obstructs the application of three-dimensional cryo-electron microscopy to large macromolecular complexes. Maximum likelihood approaches are emerging as robust tools for solving the image classification problems that are posed by such samples. Here, we propose a statistical data model that allows for a description of the experimental image formation within the formulation of 2D and 3D maximum-likelihood refinement. The proposed approach comprises a formulation of the probability calculations in Fourier space, including a spatial frequency-dependent noise model and a description of defocus-dependent imaging effects. The Expectation-Maximization-like algorithms presented are generally applicable to the alignment and classification of structurally heterogeneous projection data. Their effectiveness is demonstrated with various examples, including 2D classification of top views of the archaeal helicase MCM and 3D classification of 70S E. coli ribosome and Simian Virus 40 large T-antigen projections.

The ATPase activity of the DNA transporter TrwB is modulated by protein TrwA: implications for a common assembly mechanism of DNA translocating motors

Conjugative systems contain an essential integral membrane protein involved in DNA transport called the Type IV coupling protein (T4CP). The T4CP of conjugative plasmid R388 is TrwB, a DNA-dependent ATPase. Biochemical and structural data suggest that TrwB uses energy released from ATP hydrolysis to pump DNA through its central channel by a mechanism similar to that used by F1-ATPase or ring helicases. For DNA transport, TrwB couples the relaxosome (a DNA-protein complex) to the secretion channel. In this work we show that TrwA, a tetrameric oriT DNA-binding protein and a component of the R388 relaxosome, stimulates TrwBDeltaN70 ATPase activity, revealing a specific interaction between the two proteins. This interaction occurs via the TrwA C-terminal domain. A 68-kDa complex between TrwBDeltaN70 and TrwA C-terminal domain was observed by gel filtration chromatography, consistent with a 1:1 stoichiometry. Additionally, electron microscopy revealed the formation of oligomeric TrwB complexes in the presence, but not in the absence, of TrwA protein. TrwBDeltaN70 ATPase activity in the presence of TrwA was further enhanced by DNA. Interestingly, maximal ATPase rates were achieved with TrwA and different types of dsDNA substrates. This is consistent with a role of TrwA in facilitating the interaction between TrwB and DNA. Our findings provide a new insight into the mechanism by which TrwB recruits the relaxosome for DNA transport. The process resembles the mechanism used by other DNA-dependent molecular motors, such as the RuvA/RuvB system, to be targeted to the DNA followed by hexamer assembly.

A statistically harmonized alignment-classification in image space enables accurate and robust alignment of noisy images in single particle analysis

In determining the three-dimensional (3D) structure of macromolecular assemblies in single particle analysis, a large representative dataset of two-dimensional (2D) average images from huge number of raw images is a key for high resolution. Because alignments prior to averaging are computationally intensive, currently available multireference alignment (MRA) software does not survey every possible alignment. This leads to misaligned images, creating blurred averages and reducing the quality of the final 3D reconstruction. We present a new method, in which multireference alignment is harmonized with classification (multireference multiple alignment: MRMA). This method enables a statistical comparison of multiple alignment peaks, reflecting the similarities between each raw image and a set of reference images. Among the selected alignment candidates for each raw image, misaligned images are statistically excluded, based on the principle that aligned raw images of similar projections have a dense distribution around the correctly aligned coordinates in image space. This newly developed method was examined for accuracy and speed using model image sets with various signal-to-noise ratios, and with electron microscope images of the Transient Receptor Potential C3 and the sodium channel. In every data set, the newly developed method outperformed conventional methods in robustness against noise and in speed, creating 2D average images of higher quality. This statistically harmonized alignment-classification combination should greatly improve the quality of single particle analysis.

Unsupervised classification of single particles by cluster tracking in multi-dimensional space

In cryo-electron microscopy (cryo-EM) single-particle reconstruction, the heterogeneity of two-dimensional projection image data resulting from the co-existence of different conformational or ligand binding states of a macromolecular complex remains a major obstacle as it impairs the validity of reconstructed density maps and limits the progress toward higher resolution. Classification of cryo-EM data according to the different conformations is difficult because of the coexistence of multiple orientations in a single dataset. Here, we present an unsupervised classification method, termed cluster tracking, which utilizes the continuity in multi-dimensional space induced by angular adjacency of projections in large datasets. In a proof of concept, the testing of cluster tracking on simulated projection data, which were generated from multiple conformations and orientations of an existing volume, produced clusters that are consistent with the conformational identity of the data. The application of the method to experimental cryo-EM projection data is found to result in a partition similar to the one generated by supervised classification.

Quaternary polymorphism of replicative helicase G40P: structural mapping and domain rearrangement

Quaternary polymorphism is a distinctive structural feature of the DnaB family of replicative DNA hexameric helicases. The Bacillus subtilis bacteriophage SPP1 gene 40 product (G40P) belongs to this family. Three different quaternary states have been described for G40P homohexamers, two of them with C(3) symmetry, and the other with C(6) symmetry. We present three-dimensional reconstructions of the different architectures of G40P hexamers and a variant lacking the N-terminal domain. Comparison of the G40P and the deletion mutant structures sheds new light on the functional roles of the N and C-terminal domains, at the same time that it allows the direct structural mapping of these domains. Based on this new information, hybrid EM/X-ray models are presented for all the different symmetries. These results suggest that quaternary polymorphism of hexameric helicases may be implicated in the translocation along the DNA.

Structural basis for the cooperative assembly of large T antigen on the origin of replication

Large T antigen (LTag) from simian virus 40 (SV40) is an ATP-driven DNA helicase that specifically recognizes the core of the viral origin of replication (ori), where it oligomerizes as a double hexamer. During this process, binding of the first hexamer stimulates the assembly of a second one. Using electron microscopy, we show that the N-terminal part of LTag that includes the origin-binding domain does not present a stable quaternary structure in single hexamers. This disordered region, however, is well arranged within the LTag double hexamer after specific ori recognition, where it mediates the interactions between hexamers and constructs a separated structural module at their junction. We conclude that full assembly of LTag hexamers occurs only within the dodecamer, and requires the specific hexamer-hexamer interactions established upon binding to the origin of replication. This mechanism provides the structural basis for the cooperative assembly of LTag double hexamer on the cognate viral ori.

Polymorphism and double hexamer structure in the archaeal minichromosome maintenance (MCM) helicase from Methanobacterium thermoautotrophicum

Methanobacterium thermoautotrophicum minichromosome maintenance complex (mtMCM), a cellular replicative helicase, is a useful model for the more complex eukaryotic MCMs. Biochemical and crystallographic evidence indicates that mtMCM assembles as a double hexamer (dHex), but previous electron microscopy studies reported only the presence of single heptamers or single hexamers (Pape, T., Meka, H., Chen, S., Vicentini, G., Van Heel, M., and Onesti, S. (2003) EMBO Rep. 4, 1079-1083; Yu, X., VanLoock, M. S., Poplawski, A., Kelman, Z., Xiang, T., Tye, B. K., and Egelman, E. H. (2002) EMBO Rep. 3, 792-797). Here we present the first three-dimensional electron microscopy reconstruction of the full-length mtMCM dHex in which two hexamers contact each other via the structurally well defined N-terminal domains. The dHex has obvious side openings that resemble the side channels of LTag (large T antigen). 6-fold and 7-fold rings were observed in the same mtMCM preparation, but we determined that assembly as a double ring favors 6-fold structures. Additionally, open rings were also detected, which suggests a direct mtMCM loading mechanism onto DNA.

Double hexamer disruption and biochemical activities of Methanobacterium thermoautotrophicum MCM

Methanobacterium thermoautotrophicum MCM (mtMCM) is a helicase required for DNA replication. Previous electron microscopy studies have shown mtMCM in several oligomeric forms. However, biochemical studies suggest that mtMCM is a dodecamer, likely a double hexamer (dHex). The crystal structure of the N-terminal fragment of mtMCM reveals a stable dHex architecture. To further confirm that the dHex is not an artifact of crystal packing of two hexamers, we investigated the relevance of the dHex by disrupting the hexamer-hexamer interactions seen in the crystal structure via site-directed mutagenesis and examining various biochemical activities of the mutants in vitro. Using a combination of biochemical and structural assays, we demonstrated that changing arginine to alanine at amino acid position 161 or the insertion of a six-aminoacid peptide at the hexamer-hexamer interface disrupted dHex formation and produced stable single hexamers (sHex). Furthermore, we showed that the sHex mutants retained wild-type level of ATPase and DNA binding activities but had decreased helicase activity when compared with the wild type dHex protein. These biochemical properties of mtMCM are reminiscent of those of SV40 large T antigen, suggesting that the dHex form of mtMCM may be the active helicase for DNA unwinding during the bidirectional DNA replication.

Maximum-likelihood multi-reference refinement for electron microscopy images

A maximum-likelihood approach to multi-reference image refinement is presented. In contrast to conventional cross-correlation refinement, the new approach includes a formal description of the noise, implying that it is especially suited to cases with low signal-to-noise ratios. Application of this approach to a cryo-electron microscopy dataset revealed two major classes for projections of simian virus 40 large T-antigen in complex with an asymmetric DNA-probe, containing the origin of simian virus 40 replication. Strongly bent projections of dodecamers showed density that may be attributed to the complexed double-stranded DNA, while almost straight projections revealed a twist in the relative orientation of the hexameric subunits. This new level of detail for large T-antigen projections was not detected using conventional techniques. For a negative stain dataset, maximum-likelihood refinement yielded results that were practically identical to those obtained using conventional multi-reference refinement. Results obtained using simulated data suggest that the efficiency of the maximum-likelihood approach may be further enhanced by explicitly incorporating the microscope contrast transfer function in the image formation model.

XMIPP: a new generation of an open-source image processing package for electron microscopy

X-windows based microscopy image processing package (Xmipp) is a specialized suit of image processing programs, primarily aimed at obtaining the 3D reconstruction of biological specimens from large sets of projection images acquired by transmission electron microscopy. This public-domain software package was introduced to the electron microscopy field eight years ago, and since then it has changed drastically. New methodologies for the analysis of single-particle projection images have been added to classification, contrast transfer function correction, angular assignment, 3D reconstruction, reconstruction of crystals, etc. In addition, the package has been extended with functionalities for 2D crystal and electron tomography data. Furthermore, its current implementation in C++, with a highly modular design of well-documented data structures and functions, offers a convenient environment for the development of novel algorithms. In this paper, we present a general overview of a new generation of Xmipp that has been re-engineered to maximize flexibility and modularity, potentially facilitating its integration in future standardization efforts in the field. Moreover, by focusing on those developments that distinguish Xmipp from other packages available, we illustrate its added value to the electron microscopy community.

Electron tomography of biological samples

Electron tomography allows computing three-dimensional (3D) reconstructions of objects from their projections recorded at several angles. Combined with transmission electron microscopy, electron tomography has contributed greatly to the understanding of subcellular structures and organelles. Performed on frozen-hydrated samples, electron tomography has yielded useful information about complex biological structures. Combined with energy filtered transmission electron microscopy (EFTEM) it can be used to analyze the spatial distribution of chemical elements in biological or material sciences samples. In the present review, we present an overview of the requirements, applications, and perspectives of electron tomography in structural biology.

Normalizing projection images: a study of image normalizing procedures for single particle three-dimensional electron microscopy

In the process of three-dimensional reconstruction of single particle biological macromolecules several hundreds, or thousands, of projection images are taken from tens or hundreds of independently digitized micrographs. These different micrographs show differences in the background grey level and particle contrast and, therefore, have to be normalized by scaling their pixel values before entering the reconstruction process. In this work several normalization procedures are studied using a statistical comparison framework. We finally show that the use of the different normalization methods affects the reconstruction quality, providing guidance on the choice of normalization procedures.

Some trends in microscope image processing

The present review tries to identify some trends among the multitude of ways followed by image processing developments in the field of microscopy. Nine topics were selected. They cover the fields of: signal processing, statistical analysis, artificial intelligence, three-dimensional microscopy, multidimensional microscopy, multimodality microscopy, theory, simulation and multidisciplinarity. A specific topic is dedicated to a trend towards semi-automation instead of full automation in image processing.

A refined circular template matching method for classification of human cytomegalovirus capsids in TEM images

An automatic image analysis method for describing, segmenting, and classifying human cytomegalovirus capsids in transmission electron micrograph (TEM) images of host cell nuclei has been developed. Three stages of the capsid assembly process in the host cell nucleus have been investigated. Each class is described by a radial density profile, which is the average grey-level at each radial distance from the center. A template, constructed from the profile, is used to find possible capsid locations by correlation based matching. The matching results are further refined by size and distortion analysis of each possible capsid, resulting in a final segmentation and classification.