Structural polymorphism in bacterial EspA filaments revealed by cryo-EM and an improved approach to helical reconstruction

The traditional Fourier-Bessel approach to three-dimensional reconstruction from electron microscopic (EM) images of helical polymers involves averaging over filaments, assuming a homogeneous structure and symmetry. We have used a real-space reconstruction approach to study the EspA filaments formed by enteropathogenic E. coli. In negative stain, the symmetry of these filaments is ambiguous, and we suggest that such ambiguities may be more prevalent than realized. Using cryo-EM of frozen-hydrated filaments, we find that these filaments have a fixed twist with 5.6 subunits per turn but an axial rise per subunit that varies from about 3.6 A to 5.6 A. Reconstructions at approximately 15 A resolution show a switching between the more compressed and extended filaments in the packing of putative alpha helices around the hollow lumen. Outside of a crystal, where there is nothing to maintain long-range order, the structural polymorphism in helical polymers may be much greater than has been assumed.     

The iterative helical real space reconstruction method: surmounting the problems posed by real polymers

Many important biological macromolecules exist as helical polymers. Examples are actin, tubulin, myosin, RecA, Rad51, flagellin, pili, and filamentous bacteriophage. The first application of three-dimensional reconstruction from electron microscopic images was to a helical polymer, and a number of laboratories today are using helical tubes of integral membrane proteins for solving the structure of these proteins in the electron microscope at near atomic resolution. We have developed a method to analyze and reconstruct electron microscopic images of macromolecular helical polymers, the iterative helical real space reconstruction (IHRSR) algorithm. We can show that when there is disorder or heterogeneity, when the specimens diffract weakly, or when Bessel functions overlap, we can do far better with our method than can be done using traditional Fourier-Bessel approaches. In many cases, structures that were not even amenable to analysis can be solved at fairly high resolution using our method. The problems inherent in the traditional approach are discussed, and examples are presented illustrating how the IHRSR approach surmounts these problems.     

The structure of a filamentous bacteriophage

Many thin helical polymers, including bacterial pili and filamentous bacteriophage, have been seen as refractory to high-resolution studies by electron microscopy. Studies of the quaternary structure of such filaments have depended upon techniques such as modeling or X-ray fiber diffraction, given that direct visualization of the subunit organization has not been possible. We report the first image reconstruction of a filamentous virus, bacteriophage fd, by cryoelectron microscopy. Although these thin ( approximately 70 A in diameter) rather featureless filaments scatter weakly, we have been able to achieve a nominal resolution of approximately 8 A using an iterative helical reconstruction procedure. We show that two different conformations of the virus exist, and that in both states the subunits are packed differently than in conflicting models previously proposed on the basis of X-ray fiber diffraction or solid-state NMR studies. A significant fraction of the population of wild-type fd is either disordered or in multiple conformational states, while in the presence of the Y21M mutation, this heterogeneity is greatly reduced, consistent with previous observations. These results show that new computational approaches to helical reconstruction can greatly extend the ability to visualize heterogeneous protein polymers at a reasonably high resolution.     

An iterative Fourier-Bessel algorithm for reconstruction of helical structures with severe Bessel overlap

Classical Fourier-Bessel methodology fails when used to reconstruct helical structures with severe Bessel overlap on the layer lines. In the reconstruction of a peculiar type of double-layered helical tube of GDP-tubulin, we face the problem of Bessel overlap on all the layer lines due to the superposition of the Fourier components from the inner and outer layers of the tube. In order to decompose the Fourier terms of the inner and outer layers more than one image of the tubes must be combined and the orientations of their inner and outer layer helices must be determined. While there is no direct analytical method to determine these orientational parameters, we have devised an iterative Fourier-Bessel algorithm to calculate the correct orientations and thus allow us to obtain a reconstruction from multiple images of the double-layered tubes. The algorithm successfully works for the reconstruction of computer-modeled double-layered helical tubes as well as with real images obtained by cryo-electron microscopy. The algorithm has also been applied with very satisfactory results to the reconstruction of 13-protofilament microtubules, which is another helical structure that suffer Bessel overlap, suggesting its generality.     

Refining the structure of the Halobacterium salinarum flagellar filament using the iterative helical real space reconstruction method: insights into polymorphism

The eubacterial flagellar filament is an external, self-assembling, helical polymer approximately 220 A in diameter constructed from a highly conserved monomer, flagellin, which polymerizes externally at the distal end. The archaeal filament is only approximately 100 A in diameter, assembles at the proximal end and is constructed from different, glycosylated flagellins. Although the phenomenology of swimming is similar to that of eubacteria, the symmetry of the archebacterial filament is entirely different. Here, we extend our previous study on the flagellar coiled filament structure of strain R1M1 of Halobacterium salinarum. We use strain M175 of H.salinarum, which forms poly-flagellar bundles at high yield which, under conditions of relatively low ionic-strength (0.8 M versus 5 M) and low pH ( approximately 2.5 versus approximately 6.8), form straight filaments. We demonstrated previously that a single-particle approach to helical reconstruction has many advantages over conventional Fourier-Bessel methods when dealing with variable helical symmetry and heterogeneity. We show here that when this method is applied to the ordered helical structure of the archebacterial uncoiled flagellar filament, significant extensions in resolution can be obtained readily when compared to applying traditional helical techniques. The filament population can be separated into classes of different morphologies, which may represent polymorphic states. Using cryo-negatively stained images, a resolution of approximately 10-15 A has been achieved. Single alpha-helices can be fit into the reconstruction, supporting the proposed similarity of the structure to that of type IV bacterial pili.     

Do the utrophin tandem calponin homology domains bind F-actin in a compact or extended conformation?

Tandem calponin-homology (CH) domains play an important role in the actin-binding function of many spectrin superfamily proteins. Crystal structures from several of these proteins have suggested a flexibility between these domains, and the manner in which these domains bind to F-actin has been the subject of some controversy. A recent paper has used electron microscopy and three-dimensional reconstruction to examine the complex of the utrophin tandem CH domain with F-actin. In contrast to our previously published study, a closed conformation of the two calponin-homology domains was suggested in the new work. We show here that the new results can be explained by incomplete binding of utrophin to actin, heterogeneity in the mode of binding, and angular disorder in F-actin. We conclude that helical averaging applied to disordered filaments is responsible for their results, and that approaches designed to separate out homogeneous subsets within such filamentous complexes offer many advantages.     

A tale of two polymers: new insights into helical filaments

Many proteins function as helical polymers within the cell. Two intensively studied examples are eukaryotic actin and bacterial RecA, which belong to two different protein superfamilies. However, most other members of these superfamilies do not polymerize into helical filaments. General features of polymorphism, cooperativity and allostery that emerge from studies of eukaryotic actin and bacterial RecA raise more general issues about how conserved these filamentous structures have been during evolution.     

Application of the iterative helical real-space reconstruction method to large membranous tubular crystals of P-type ATPases

Since the development of three-dimensional helical reconstruction methods in the 1960's, advances in Fourier-Bessel methods have facilitated structure determination to near-atomic resolution. A recently developed iterative helical real-space reconstruction (IHRSR) method provides an alternative that uses single-particle analysis in conjunction with the imposition of helical symmetry. In this work, we have adapted the IHRSR algorithm to work with frozen-hydrated tubular crystals of P-type ATPases. In particular, we have implemented layer-line filtering to improve the signal-to-noise ratio, Wiener-filtering to compensate for the contrast transfer function, solvent flattening to improve reference reconstructions, out-of-plane tilt compensation to deal with flexibility in three dimensions, systematic calculation of Fourier shell correlations to track the progress of the refinement, and tools to control parameters as the refinement progresses. We have tested this procedure on datasets from Na(+)/K(+)-ATPase, rabbit skeletal Ca(2+)-ATPase and scallop Ca(2+)-ATPase in order to evaluate the potential for sub-nanometer resolution as well as the robustness in the presence of disorder. We found that Fourier-Bessel methods perform better for well-ordered samples of skeletal Ca(2+)-ATPase and Na(+)/K(+)-ATPase, although improvements to IHRSR are discussed that should reduce this disparity. On the other hand, IHRSR was very effective for scallop Ca(2+)-ATPase, which was too disordered to analyze by Fourier-Bessel methods.     

Image analysis reveals that Escherichia coli RecA protein consists of two domains.

The Escherichia coli RecA protein catalyzes homologous genetic recombination by forming helical polymers around DNA molecules. These polymers have an ATPase activity, which is essential for the movement of strands between two DNA molecules. One obstacle to structural studies of the RecA filament has been that the ATPase results in a dynamical polymer containing a mixture of states with respect to the bound ATP and its hydrolytic products. We have formed filaments which are trapped in the ADP-Pi state by substituting AIF4- for the Pi, and have used these stable filaments to generate a three-dimensional reconstruction from electron micrographs. The resolution of the reconstruction is sufficient to resolve the 38-k RecA subunit into two nearly equal domains. This reconstruction provides the most detailed view yet of the RecA protein, and serves as a framework within which existing biochemical data on RecA can be understood.     

Helical crystallization on lipid nanotubes: streptavidin as a model protein

In this study, we use streptavidin (SA) as a model system to study helical protein array formation on lipid nanotubes, an alternative to 2D studies on lipid monolayers. We demonstrate that wild-type and a mutant form of SA form helical arrays on biotinylated lipid nanotubes. 3D maps from helical arrays of wild-type and mutant SA were reconstructed using two different approaches: Fourier-Bessel methods and an iterative single particle algorithm. The maps show that wild-type and mutant streptavidin molecules order differently. The molecular packing arrangements of SA on the surface of the lipid nanotubes differ from previously reported lattice packing of SA on biotinylated monolayers. Helical crystallization on lipid nanotubes presents an alternative platform to explore fundamentals of protein ordering, intermolecular protein interaction and phase behavior. We demonstrate that lipid nanotubes offer a robust and reproducible substrate for forming helical protein arrays which present a means for studying protein structure and structure-function relationships.     

Helical Filaments of Human Dmc1 Protein on Single-Stranded DNA: A Cautionary Tale

Proteins in the RecA/Rad51/RadA family form nucleoprotein filaments on DNA that catalyze a strand exchange reaction as part of homologous genetic recombination. Because of the centrality of this system to many aspects of DNA repair, the generation of genetic diversity, and cancer when this system fails or is not properly regulated, these filaments have been the object of many biochemical and biophysical studies. A recent paper has argued that the human Dmc1 protein, a meiotic homolog of bacterial RecA and human Rad51, forms filaments on single-stranded DNA with ∼ 9 subunits per turn in contrast to the filaments formed on double-stranded DNA with ∼ 6.4 subunits per turn and that the stoichiometry of DNA binding is different between these two filaments. We show using scanning transmission electron microscopy that the Dmc1 filament formed on single-stranded DNA has a mass per unit length expected from ∼ 6.5 subunits per turn. More generally, we show how ambiguities in helical symmetry determination can generate incorrect solutions and why one sometimes must use other techniques, such as biochemistry, metal shadowing, or scanning transmission electron microscopy, to resolve these ambiguities. While three-dimensional reconstruction of helical filaments from EM images is a powerful tool, the intrinsic ambiguities that may be present with limited resolution are not sufficiently appreciated.     

Conformational behavior of chemically reactive alanine-rich repetitive protein polymers

The synthesis of protein-based polymers with controlled conformational properties and functional group placement offers many opportunities for the design of advanced materials. In this work, protein engineering methods have been used to produce repetitive alanine-rich protein polymers with the sequence [(AAAQ)(5)(AAAE)(AAAQ)(5)](x) (x = 2 and 6); these macromolecules may mimic architectural features of certain alanine-rich helical sequences found in natural proteins. Various proteins from this family can be readily expressed and purified from Escherichia coli. Circular dichroic spectroscopy (CD) characterization demonstrates that the purified proteins are highly helical under a variety of conditions. Thermal analysis of [(AAAQ)(5)(AAAE)(AAAQ)(5)](2) via differential scanning calorimetry (DSC) and CD indicates that the protein undergoes a reversible helix-coil transition at approximately 45 degrees C and that the protein conformation can be manipulated at elevated temperatures depending on solution conditions. The demonstrated conformational properties of these artificial proteins suggest that they may be excellent candidates for elucidating structure-function relationships in biopolymers for nanotechnology and biological applications.     

Averaging data derived from images of helical structures with different symmetries.

There are many examples of macromolecules that form helical tubes or crystals, which are useful for structure determination by electron microscopy and image processing. Helical crystals can be thought of as two-dimensional crystals that have been rolled into a cylinder such that two lattice points are superimposed. In many real cases, helical crystals of a particular macromolecule derive from an identical two-dimensional lattice but have different lattice points superimposed, thus producing different helical symmetries which cannot be simply averaged in Fourier-space. When confronted with this situation, one can select images corresponding to one of the observed symmetries at the expense of reducing the number of images that can be used for data collection and averaging, or one can calculate separate density maps from each symmetry, then align and average them together in real-space. Here, we present a third alternative, which is based on averaging of the Fourier-Bessel coefficients, gn,l(r), and which allows the inclusion of data from all symmetry groups derived from a common two-dimensional lattice. The method is straightforward and simple in practice and is shown, through a specific example with real data, to give results comparable to real-space averaging.     

Ruby-Helix: an implementation of helical image processing based on object-oriented scripting language

Helical image analysis in combination with electron microscopy has been used to study three-dimensional structures of various biological filaments or tubes, such as microtubules, actin filaments, and bacterial flagella. A number of packages have been developed to carry out helical image analysis. Some biological specimens, however, have a symmetry break (seam) in their three-dimensional structure, even though their subunits are mostly arranged in a helical manner. We refer to these objects as "asymmetric helices". All the existing packages are designed for helically symmetric specimens, and do not allow analysis of asymmetric helical objects, such as microtubules with seams. Here, we describe Ruby-Helix, a new set of programs for the analysis of "helical" objects with or without a seam. Ruby-Helix is built on top of the Ruby programming language and is the first implementation of asymmetric helical reconstruction for practical image analysis. It also allows easier and semi-automated analysis, performing iterative unbending and accurate determination of the repeat length. As a result, Ruby-Helix enables us to analyze motor-microtubule complexes with higher throughput to higher resolution.     

The utrophin actin-binding domain binds F-actin in two different modes: implications for the spectrin superfamily of proteins

Utrophin, like its homologue dystrophin, forms a link between the actin cytoskeleton and the extracellular matrix. We have used a new method of image analysis to reconstruct actin filaments decorated with the actin-binding domain of utrophin, which contains two calponin homology domains. We find two different modes of binding, with either one or two calponin-homology (CH) domains bound per actin subunit, and these modes are also distinguishable by their very different effects on F-actin rigidity. Both modes involve an extended conformation of the CH domains, as predicted by a previous crystal structure. The separation of these two modes has been largely dependent upon the use of our new approach to reconstruction of helical filaments. When existing information about tropomyosin, myosin, actin-depolymerizing factor, and nebulin is considered, these results suggest that many actin-binding proteins may have multiple binding sites on F-actin. The cell may use the modular CH domains found in the spectrin superfamily of actin-binding proteins to bind actin in manifold ways, allowing for complexity to arise from the interactions of a relatively few simple modules with actin.     

The mammalian cardiac muscle thick filament: crossbridge arrangement

Although skeletal muscle thick filaments have been extensively studied, information on the structure of cardiac thick filaments is limited. Since cardiac muscle differs in many physiological properties from skeletal muscle it is important to elucidate the structure of the cardiac thick filament. The structure of isolated and negatively stained rabbit cardiac thick filaments has been analyzed from computed Fourier transforms and image analysis. The transforms are detailed, showing a strong set of layer lines corresponding to a 42.9 nm quasi-helical repeat. The presence of relatively strong "forbidden" meridional reflections not expected from ideal helical symmetry on the second, fourth, fifth, seventh, eighth, and tenth layer lines suggest that the crossbridge array is perturbed from ideal helical symmetry. Analysis of the phase differences for the primary reflections on the first layer line of transforms from 15 filaments showed an average difference of 170 degrees, close to the value of 180 degrees expected for an odd-stranded structure. Computer-filtered images of the isolated thick filaments unequivocally demonstrate a three-stranded arrangement of the crossbridges on the filaments and provide evidence that the crossbridge arrangement is axially perturbed from ideal helical symmetry.     

Myosin filament structure in vertebrate smooth muscle

The in vivo structure of the myosin filaments in vertebrate smooth muscle is unknown. Evidence from purified smooth muscle myosin and from some studies of intact smooth muscle suggests that they may have a nonhelical, side-polar arrangement of crossbridges. However, the bipolar, helical structure characteristic of myosin filaments in striated muscle has not been disproved for smooth muscle. We have used EM to investigate this question in a functionally diverse group of smooth muscles (from the vascular, gastrointestinal, reproductive, and visual systems) from mammalian, amphibian, and avian species. Intact muscle under physiological conditions, rapidly frozen and then freeze substituted, shows many myosin filaments with a square backbone in transverse profile. Transverse sections of fixed, chemically skinned muscles also show square backbones and, in addition, reveal projections (crossbridges) on only two opposite sides of the square. Filaments gently isolated from skinned smooth muscles and observed by negative staining show crossbridges with a 14.5-nm repeat projecting in opposite directions on opposite sides of the filament. Such filaments subjected to low ionic strength conditions show bare filament ends and an antiparallel arrangement of myosin tails along the length of the filament. All of these observations are consistent with a side-polar structure and argue against a bipolar, helical crossbridge arrangement. We conclude that myosin filaments in all smooth muscles, regardless of function, are likely to be side-polar. Such a structure could be an important factor in the ability of smooth muscles to contract by large amounts.     

Flagellar filaments of the deep-sea bacteria Idiomarina loihiensis belong to a family different from those of Salmonella typhimurium

The helical filaments of the bacterial flagella so far studied seem to be universal in the bacterial kingdom. Despite the variation in flagellin molecular masses, which range from 24 kDa to 62 kDa in different species, there are only two forms: either the so-called Normal (left-handed) or the Curly (right-handed). The Normal and Curly helical forms are asymmetric; the two characteristic helical parameters, which are the pitch and diameter, of Normal filaments are twice those of Curly filaments. Both the universality of these two helical forms and their asymmetry are biological puzzles. We found that the marine bacteria Idiomarina loihiensis have flagella with left-handed Curly-like filaments. Analysis of the polymorphic forms under different pH conditions showed that the Curly-like filaments are actually Normal filaments having a smaller pitch and diameter than those of Salmonella typhimurium. A minor modification of Calladine's model for a filament lattice can explain the variant helical forms. Pseudomonas aeruginosa filaments also belong to the family of I.loihiensis filaments. Thus, there are at least two families of flagella filaments.     

A helical processing pipeline for EM structure determination of membrane proteins

Electron crystallography plays a key role in the structural biology of integral membrane proteins (IMPs) by offering one of the most direct means of providing insight into the functional state of these molecular machines in their lipid-associated forms, and also has the potential to facilitate examination of physiologically relevant transitional states and complexes. Helical or tubular crystals, which are the natural product of proteins crystallizing on the surface of a cylindrical vesicle, offer some unique advantages, such as three-dimensional (3D) information from a single view, compared to other crystalline forms. While a number of software packages are available for processing images of helical crystals to produce 3D electron density maps, widespread exploitation of helical image reconstruction is limited by a lack of standardized approaches and the initial effort and specialized expertise required. Our goal is to develop an integrated pipeline to enable structure determination by transmission electron microscopy (TEM) of IMPs in the form of tubular crystals. We describe here the integration of standard Fourier-Bessel helical analysis techniques into Appion, an integrated, database-driven pipeline.     

AN ultrastructural study of cross-bridge arrangement in the frog thigh muscle thick filament.

We have developed thick filament isolation methods that preserve the relaxed cross-bridge order of frog thick filaments such that the filaments can be analyzed by the convergent techniques of electron microscopy, optical diffraction, and computer image analysis. Images of the filaments shadowed by using either unidirectional shadowing or rotary shadowing show a series of subunits arranged along a series of right-handed near-helical strands that occur every 43 nm axially along the filament arms. Optical filtrations of images of these shadowed filaments show 4-5 subunits per half-turn of the strands, consistent with a three-stranded arrangement of the cross-bridges, thus supporting our earlier results from negative staining and computer-image analysis. The optical diffraction patterns of the shadowed filaments show a departure from the pattern expected for helical symmetry consistent with the presence of cylindrical symmetry and a departure of the cross-bridges from helical symmetry. We also describe a modified negative staining procedure that gives improved delineation of the cross-bridge arrangement. From analysis of micrographs of these negatively stained filament tilted about their long axes, we have computed a preliminary three-dimensional reconstruction of the filament that clearly confirms the three-stranded arrangement of the myosin heads.