Signal peptide mimicry primes Sec61 for client-selective inhibition.

Preventing the biogenesis of disease-relevant proteins is an attractive therapeutic strategy, but attempts to target essential protein biogenesis factors have been hampered by excessive toxicity. Here we describe KZR-8445, a cyclic depsipeptide that targets the Sec61 translocon and selectively disrupts secretory and membrane protein biogenesis in a signal peptide-dependent manner. KZR-8445 potently inhibits the secretion of pro-inflammatory cytokines in primary immune cells and is highly efficacious in a mouse model of rheumatoid arthritis. A cryogenic electron microscopy structure reveals that KZR-8445 occupies the fully opened Se61 lateral gate and blocks access to the lumenal plug domain. KZR-8445 binding stabilizes the lateral gate helices in a manner that traps select signal peptides in the Sec61 channel and prevents their movement into the lipid bilayer. Our results establish a framework for the structure-guided discovery of novel therapeutics that selectively modulate Sec61-mediated protein biogenesis.

Cryo-EM structure of ssDNA bacteriophage ΦCjT23 provides insight into early virus evolution.

The origin of viruses remains an open question. While lack of detectable sequence similarity hampers the analysis of distantly related viruses, structural biology investigations of conserved capsid protein structures facilitate the study of distant evolutionary relationships. Here we characterize the lipid-containing ssDNA temperate bacteriophage ΦCjT23, which infects Flavobacterium sp. (Bacteroidetes). We report ΦCjT23-like sequences in the genome of strains belonging to several Flavobacterium species. The virion structure determined by cryogenic electron microscopy reveals similarities to members of the viral kingdom Bamfordvirae that currently consists solely of dsDNA viruses with a major capsid protein composed of two upright ß-sandwiches. The minimalistic structure of ΦCjT23 suggests that this phage serves as a model for the last common ancestor between ssDNA and dsDNA viruses in the Bamfordvirae. Both ΦCjT23 and the related phage FLiP infect Flavobacterium species found in several environments, suggesting that these types of viruses have a global distribution and a shared evolutionary origin. Detailed comparisons to related, more complex viruses not only expand our knowledge about this group of viruses but also provide a rare glimpse into early virus evolution.

Structure of a Cell Entry Defective Human Adenovirus Provides Insights into Precursor Proteins and Capsid Maturation.

Maturation of adenoviruses is distinguished by proteolytic processing of several interior minor capsid proteins and core proteins by the adenoviral protease and subsequent reorganization of adenovirus core. We report the results derived from the icosahedrally averaged cryo-EM structure of a cell entry defective form of adenovirus, designated ts1, at a resolution of 3.7 Å as well as of the localized reconstructions of unique hexons and penton base. The virion structure revealed the structures and organization of precursors of minor capsid proteins, pIIIa, pVI and pVIII, which are closely associated with the hexons on the capsid interior. In addition to a well-ordered helical domain (a.a. 310-397) of pIIIa, highlights of the structure include the precursors of VIII display significantly different structures near the cleavage sites. Moreover, we traced residues 4-96 of the membrane lytic protein (pVI) that includes an amphipathic helix occluded deep in the hexon cavity suggesting the possibility of co-assembly of hexons with the precursors of VI. In addition, we observe a second copy of pVI ordered up to residue L40 in the peripentonal hexons and a few fragments of density corresponding to 2nd and 3rd copies of pVI in other hexons. However, we see no evidence of precursors of VII binding in the hexon cavity. These findings suggest the possibility that differently bound pVI molecules undergo processing at the N-terminal cleavage sites at varying efficiencies, subsequently creating competition between the cleaved and uncleaved forms of VI, followed by reorganization, processing, and release of VI molecules from the hexon cavities.

Localized reconstruction in Scipion expedites the analysis of symmetry mismatches in cryo-EM data.

Technological advances in transmission electron microscopes and detectors have turned cryogenic electron microscopy (cryo-EM) into an essential tool for structural biology. A commonly used cryo-EM data analysis method, single particle analysis, averages hundreds of thousands of low-dose images of individual macromolecular complexes to determine a density map of the complex. The presence of symmetry in the complex is beneficial since each projection image can be assigned to multiple views of the complex. However, data processing that applies symmetry can average out asymmetric features and consequently data analysis methods are required to resolve asymmetric structural features. Scipion is a cryo-EM image processing framework that integrates functions from different image processing packages as plugins. To extend its functionality for handling symmetry mismatches, we present here a Scipion plugin termed LocalRec implementing the localized reconstruction method. When tested on an adenovirus data set, the plugin enables resolving the symmetry-mismatched trimeric fibre bound to the five-fold vertices of the capsid. Furthermore, it improves the structure determination of the icosahedral capsid by dealing with the defocus gradient across the particle. LocalRec is expected to be widely applicable in a range of cryo-EM investigations of flexible and symmetry mismatched complexes.

Tox_(R)CNN: Deep learning-based nuclei profiling tool for drug toxicity screening.

Toxicity is an important factor in failed drug development, and its efficient identification and prediction is a major challenge in drug discovery. We have explored the potential of microscopy images of fluorescently labeled nuclei for the prediction of toxicity based on nucleus pattern recognition. Deep learning algorithms obtain abstract representations of images through an automated process, allowing them to efficiently classify complex patterns, and have become the state-of-the art in machine learning for computer vision. Here, deep convolutional neural networks (CNN) were trained to predict toxicity from images of DAPI-stained cells pre-treated with a set of drugs with differing toxicity mechanisms. Different cropping strategies were used for training CNN models, the nuclei-cropping-based Tox_CNN model outperformed other models classifying cells according to health status. Tox_CNN allowed automated extraction of feature maps that clustered compounds according to mechanism of action. Moreover, fully automated region-based CNNs (RCNN) were implemented to detect and classify nuclei, providing per-cell toxicity prediction from raw screening images. We validated both Tox_(R)CNN models for detection of pre-lethal toxicity from nuclei images, which proved to be more sensitive and have broader specificity than established toxicity readouts. These models predicted toxicity of drugs with mechanisms of action other than those they had been trained for and were successfully transferred to other cell assays. The Tox_(R)CNN models thus provide robust, sensitive, and cost-effective tools for in vitro screening of drug-induced toxicity. These models can be adopted for compound prioritization in drug screening campaigns, and could thereby increase the efficiency of drug discovery.

Structures of Human Peroxiredoxin 3 Suggest Self-Chaperoning Assembly that Maintains Catalytic State.

Peroxiredoxins are antioxidant proteins primarily responsible for detoxification of hydroperoxides in cells. On exposure to various cellular stresses, peroxiredoxins can acquire chaperone activity, manifested as quaternary reorganization into a high molecular weight (HMW) form. Acidification, for example, causes dodecameric rings of human peroxiredoxin 3 (HsPrx3) to stack into long helical filaments. In this work, a 4.1-Å resolution structure of low-pH-instigated helical filaments was elucidated, showing a locally unfolded active site and partially folded C terminus. A 2.8-Å crystal structure of HsPrx3 was determined at pH 8.5 under reducing conditions, wherein dodecameric rings are arranged as a short stack, with symmetry similar to low-pH filaments. In contrast to previous observations, the crystal structure displays both a fully folded active site and ordered C terminus, suggesting that the HsPrx3 HMW form maintains catalytic activity. We propose a new role for the HMW form as a self-chaperoning assembly maintaining HsPrx3 function under stress.

Alignment of direct detection device micrographs using a robust Optical Flow approach.

The introduction of direct detection devices in cryo-EM has shown that specimens present beam-induced motion (BIM). Consequently, in this work, we develop a BIM correction method at the image level, resulting in an integrated image in which the in-plane BIM blurring is compensated prior to particle picking. The methodology is based on a robust Optical Flow (OF) approach that can efficiently correct for local movements in a rapid manner. The OF works particularly well if the BIM pattern presents a substantial degree of local movements, which occurs in our data sets for Falcon II data. However, for those cases in which the BIM pattern corresponds to global movements, we have found it advantageous to first run a global motion correction approach and to subsequently apply OF. Additionally, spatial analysis of the Optical Flow allows for quantitative analysis of the BIM pattern. The software that incorporates the new approach is available in XMIPP (http://xmipp.cnb.csic.es).