Nanometre-sized objects with highly symmetrical, cage-like polyhedral shapes, often with icosahedral symmetry, have recently been assembled from DNA, RNA or proteins for applications in biology and medicine. These achievements relied on advances in the development of programmable self-assembling biological materials, and on rapidly developing techniques for generating three-dimensional (3D) reconstructions from cryo-electron microscopy images of single particles, which provide high-resolution structural characterization of biological complexes. Such single-particle 3D reconstruction approaches have not yet been successfully applied to the identification of synthetic inorganic nanomaterials with highly symmetrical cage-like shapes. Here, however, using a combination of cryo-electron microscopy and single-particle 3D reconstruction, we suggest the existence of isolated ultrasmall (less than 10 nm) silica cages ('silicages') with dodecahedral structure. We propose that such highly symmetrical, self-assembled cages form through the arrangement of primary silica clusters in aqueous solutions on the surface of oppositely charged surfactant micelles. This discovery paves the way for nanoscale cages made from silica and other inorganic materials to be used as building blocks for a wide range of advanced functional-materials applications.

No abstract

The direct electron detector has revolutionized electron cryo-microscopy (CryoEM). Icosahedral virus structures are routinely produced at 4Å resolution or better and the approach has largely displaced virus crystallography, as it requires less material, less purity and often produces a structure more rapidly. Largely ignored in this new era of CryoEM is the dynamic information in the data sets that was not available in X-ray structures. Here we review an approach that captures the dynamic character of viruses displayed in the CryoEM ensemble of particles at the moment of freezing. We illustrate the approach with a simple model, briefly describe the details and provide a practical application to virus particle maturation.

Proposed: ConceptGAN Baseline CycleGAN colorbags edgebags colorbags edgebags reconstructions reconstructions no training data edgeshoes no training data edgeshoes colorshoes colorshoes Figure 1: We propose ConceptGAN, a framework that can jointly learn, transfer and compose concepts to generate semantically meaningful images, even in subdomains with no training data (highlighted) while the state-of-the-art methods such as CycleGAN [49] fail to do so. AbstractCompositionality of semantic concepts in image synthesis and analysis is appealing as it can help in decomposing known and generatively recomposing unknown data. For instance, we may learn concepts of changing illumination, geometry or albedo of a scene, and try to recombine them to generate physically meaningful, but unseen data for training and testing. In practice however we often do not have samples from the joint concept space available: We may have data on illumination change in one data set and on geometric change in another one without complete overlap. We pose the following question: How can we learn two or more concepts jointly from different data sets with mutual consistency where we do not have samples from the full joint space? We present a novel answer in this paper based on cyclic consistency over multiple concepts, represented individually by generative adversarial networks (GANs). Our method, ConceptGAN, can be understood as a drop in for data augmentation to improve resilience for real world applications. Qualitative and quantitative evaluations demonstrate its efficacy in generating semantically meaningful images, as well as one shot face verification as an example application.

Cryo EM structures of maturation-intermediate Prohead I of bacteriophage HK97 with (PhI(Pro+)) and without (PhI(Pro-)) the viral protease packaged have been reported (Veesler et al., 2014). In spite of PhI(Pro+) containing an additional ∼ 100 × 24 kD of protein, the two structures appeared identical although the two particles have substantially different biochemical properties, e.g., PhI(Pro-) is less stable to disassembly conditions such as urea. Here the same cryo EM images are used to characterize the spatial heterogeneity of the particles at 17Å resolution by variance analysis and show that PhI(Pro-) has roughly twice the standard deviation of PhI(Pro+). Furthermore, the greatest differences in standard deviation are present in the region where the δ-domain, not seen in X-ray crystallographic structures or fully seen in cryo EM, is expected to be located. Thus presence of the protease appears to stabilize the δ-domain which the protease will eventually digest.