Convolutional networks for supervised mining of molecular patterns within cellular context.

Cryo-electron tomograms capture a wealth of structural information on the molecular constituents of cells and tissues. We present DeePiCt (deep picker in context), an open-source deep-learning framework for supervised segmentation and macromolecular complex localization in cryo-electron tomography. To train and benchmark DeePiCt on experimental data, we comprehensively annotated 20 tomograms of Schizosaccharomyces pombe for ribosomes, fatty acid synthases, membranes, nuclear pore complexes, organelles, and cytosol. By comparing DeePiCt to state-of-the-art approaches on this dataset, we show its unique ability to identify low-abundance and low-density complexes. We use DeePiCt to study compositionally distinct subpopulations of cellular ribosomes, with emphasis on their contextual association with mitochondria and the endoplasmic reticulum. Finally, applying pre-trained networks to a HeLa cell tomogram demonstrates that DeePiCt achieves high-quality predictions in unseen datasets from different biological species in a matter of minutes. The comprehensively annotated experimental data and pre-trained networks are provided for immediate use by the community.

Intracellular nanoscale architecture as a master regulator of calcium carbonate crystallization in marine microalgae.

Unicellular marine microalgae are responsible for one of the largest carbon sinks on Earth. This is in part due to intracellular formation of calcium carbonate scales termed coccoliths. Traditionally, the influence of changing environmental conditions on this process has been estimated using poorly constrained analogies to crystallization mechanisms in bulk solution, yielding ambiguous predictions. Here, we elucidated the intracellular nanoscale environment of coccolith formation in the model species Pleurochrysis carterae using cryoelectron tomography. By visualizing cells at various stages of the crystallization process, we reconstructed a timeline of coccolith development. The three-dimensional data portray the native-state structural details of coccolith formation, uncovering the crystallization mechanism, and how it is spatially and temporally controlled. Most strikingly, the developing crystals are only tens of nanometers away from delimiting membranes, resulting in a highly confined volume for crystal growth. We calculate that the number of soluble ions that can be found in such a minute volume at any given time point is less than the number needed to allow the growth of a single atomic layer of the crystal and that the uptake of single protons can markedly affect nominal pH values. In such extreme confinement, the crystallization process is expected to depend primarily on the regulation of ion fluxes by the living cell, and nominal ion concentrations, such as pH, become the result, rather than a driver, of the crystallization process. These findings call for a new perspective on coccolith formation that does not rely exclusively on solution chemistry.

Native molecular architectures of centrosomes in C. elegans embryos.

Centrosomes organize microtubules that are essential for mitotic divisions in animal cells. They consist of centrioles surrounded by Pericentriolar Material (PCM). Questions related to mechanisms of centriole assembly, PCM organization, and microtubule formation remain unanswered, in part due to limited availability of molecular-resolution structural analyses in situ. Here, we use cryo-electron tomography to visualize centrosomes across the cell cycle in cells isolated from C. elegans embryos. We describe a pseudo-timeline of centriole assembly and identify distinct structural features including a cartwheel in daughter centrioles, and incomplete microtubule doublets surrounded by a star-shaped density in mother centrioles. We find that centriole and PCM microtubules differ in protofilament number (13 versus 11) indicating distinct nucleation mechanisms. This difference could be explained by atypical γ-tubulin ring complexes with 11-fold symmetry identified at the minus ends of short PCM microtubules. We further characterize a porous and disordered network that forms the interconnected PCM. Thus, our work builds a three-dimensional structural atlas that helps explain how centrosomes assemble, grow, and achieve function.

Cryo-Electron Tomography of Reconstituted Biomolecular Condensates.

The assembly of membraneless compartments by phase separation has recently been recognized as a mechanism for spatial and temporal organization of biomolecules within the cell. The functions of such mesoscale assemblies, termed biomolecular condensates, depend on networks of multivalent interactions between proteins, their structured and disordered domains, and commonly also include nucleic acids. Cryo-electron tomography is an ideal tool to investigate the three-dimensional architecture of such pleomorphic interaction networks at nanometer resolution and thus form inferences about function. However, preparation of suitable cryo-electron microscopy samples of condensates may be prone to protein denaturation, low retention of material on the sample carrier, and contamination associated with cryo-sample preparation and transfers. Here, we describe a series of protocols designed to obtain high-quality cryo-electron tomography data of biomolecular condensates reconstituted in vitro. These include critical screening by light microscopy, cryo-fixation by plunge freezing, sample loading into an electron microscope operated at liquid nitrogen temperature, data collection, processing of the data into three-dimensional tomograms, and their interpretation.