Selecting particles from digital micrographs is an essential step in single-particle electron cryomicroscopy (cryo-EM). As manual selection of complete datasets—typically comprising thousands of particles—is a tedious and time-consuming process, numerous automatic particle pickers have been developed. However, non-ideal datasets pose a challenge to particle picking. Here we present the particle picking software crYOLO which is based on the deep-learning object detection system You Only Look Once (YOLO). After training the network with 200–2500 particles per dataset it automatically recognizes particles with high recall and precision while reaching a speed of up to five micrographs per second. Further, we present a general crYOLO network able to pick from previously unseen datasets, allowing for completely automated on-the-fly cryo-EM data preprocessing during data acquisition. crYOLO is available as a standalone program under http://sphire.mpg.de/ and is distributed as part of the image processing workflow in SPHIRE.

Selecting particles from digital micrographs is an essential step in single particle electron cryomicroscopy (cryo-EM). Since manual selection of complete datasets typically comprising many thousands of particles is a tedious and time-consuming process, many automatic particle pickers have been developed in the past few decades.However, non-ideal datasets pose a challenge to particle picking. Here, we present a novel automated particle picking software called crYOLO, which is based on the deep learning object detection system "You Only Look Once" (YOLO). After training the network with 500 -2,500 particles per dataset, it automatically recognizes particles with high recall and precision reaching a speed of up to five micrographs per second.Importantly, we demonstrate a powerful general network trained on more than 40 datasets to select previously unseen datasets, thus paving the way for completely automated "on-the-fly" cryo-EM data pre-processing during data acquisition. CrYOLO is available as a standalone program under http://sphire.mpg.de/ and will be part of the image processing workflow in SPHIRE.

The CCR4-NOT deadenylase complex is a master regulator of translation and mRNA stability. Its NOT module orchestrates recruitment of the catalytic subunits to target mRNAs. We report the crystal structure of the human NOT module formed by the CNOT1, CNOT2 and CNOT3 C-terminal (-C) regions. CNOT1-C provides a rigid scaffold consisting of two perpendicular stacks of HEAT-like repeats. CNOT2-C and CNOT3-C heterodimerize through their SH3-like NOT-box domains. The heterodimer is stabilized and tightly anchored to the surface of CNOT1 through an unexpected intertwined arrangement of peptide regions lacking defined secondary structure. These assembly peptides mold onto their respective binding surfaces and form extensive interfaces. Mutagenesis of individual interfaces and perturbation of endogenous protein ratios cause defects in complex assembly and mRNA decay. Our studies provide a structural framework for understanding the recruitment of the CCR4-NOT complex to mRNA targets.

The RNA-binding proteins of the Nanos family play an essential role in germ cell development and survival in a wide range of metazoan species. They function by suppressing the expression of target mRNAs through the recruitment of effector complexes, which include the CCR4-NOT deadenylase complex. Here, we show that the three human Nanos paralogs (Nanos1-3) interact with the CNOT1 C-terminal domain and determine the structural basis for the specific molecular recognition. Nanos1-3 bind CNOT1 through a short CNOT1-interacting motif (NIM) that is conserved in all vertebrates and some invertebrate species. The crystal structure of the human Nanos1 NIM peptide bound to CNOT1 reveals that the peptide opens a conserved hydrophobic pocket on the CNOT1 surface by inserting conserved aromatic residues. The substitutions of these aromatic residues in the Nanos1-3 NIMs abolish binding to CNOT1 and abrogate the ability of the proteins to repress translation. Our findings provide the structural basis for the recruitment of the CCR4-NOT complex by vertebrate Nanos, indicate that the NIMs are the major determinants of the translational repression mediated by Nanos, and identify the CCR4-NOT complex as the main effector complex for Nanos function.

Human (Hs) Roquin1 and Roquin2 are RNA-binding proteins that promote mRNA target degradation through the recruitment of the CCR4-NOT deadenylase complex and are implicated in the prevention of autoimmunity. Roquin1 recruits CCR4-NOT via a C-terminal region that is not conserved in Roquin2 or in invertebrate Roquin. Here we show that Roquin2 and Drosophila melanogaster (Dm) Roquin also interact with the CCR4-NOT complex through their C-terminal regions. The C-terminal region of Dm Roquin contains multiple motifs that mediate CCR4-NOT binding. One motif binds to the CAF40 subunit of the CCR4-NOT complex. The crystal structure of the Dm Roquin CAF40-binding motif (CBM) bound to CAF40 reveals that the CBM adopts an a-helical conformation upon binding to a conserved surface of CAF40. Thus, despite the lack of sequence conservation, the C-terminal regions of Roquin proteins act as an effector domain that represses the expression of mRNA targets via recruitment of the CCR4-NOT complex.

CCR4-NOT is a conserved multiprotein complex which regulates eukaryotic gene expression principally via shortening of poly(A) tails of messenger RNA or deadenylation. Here, we reconstitute a complete, recombinant human CCR4-NOT complex. Our reconstitution strategy permits strict compositional control to test mechanistic hypotheses with purified component variants. CCR4-NOT is more active and selective for poly(A) than the isolated exonucleases, CCR4a and CAF1, which have distinct deadenylation profiles in vitro. The exonucleases require at least two out of three conserved non-enzymatic modules (CAF40, NOT10:NOT11 or NOT) for full activity in CCR4-NOT. CAF40 and the NOT10:NOT11 module both bind RNA directly and stimulate deadenylation in a partially redundant manner. Linear motifs from different RNA-binding factors that recruit CCR4-NOT to specific mRNAs via protein-protein interactions with CAF40 can inhibit bulk deadenylation. We reveal an additional layer of regulatory complexity to the human deadenylation machinery, which may prime it either for general or target-specific degradation.

In metazoans, a ≈1 megadalton (MDa) multiprotein complex comprising the dynein–dynactin adaptor Spindly and the ROD–Zwilch–ZW10 (RZZ) complex is the building block of a fibrous biopolymer, the kinetochore fibrous corona. The corona assembles on mitotic kinetochores to promote microtubule capture and spindle assembly checkpoint (SAC) signaling. We report here a high‐resolution cryo‐EM structure that captures the essential features of the RZZ complex, including a farnesyl‐binding site required for Spindly binding. Using a highly predictive in vitro assay, we demonstrate that the SAC kinase MPS1 is necessary and sufficient for corona assembly at supercritical concentrations of the RZZ–Spindly (RZZS) complex, and describe the molecular mechanism of phosphorylation‐dependent filament nucleation. We identify several structural requirements for RZZS polymerization in rings and sheets. Finally, we identify determinants of kinetochore localization and corona assembly of Spindly. Our results describe a framework for the long‐sought‐for molecular basis of corona assembly on metazoan kinetochores.

No abstract