Dimitry Tegunov

Cryo-electron tomography (cryo-ET) visualizes cellular environments in a near-native state at macromolecular resolution. Accurate alignment of the tilted projection images is essential for data interpretation, yet existing reference-free algorithms often fail due to limited information overlap between images and inaccurate assumptions about the sample. Human experts, however, can easily recognize the misalignment left by these tools. We introduce MissAlignment, a machine learning approach that trains similar intuition to improve tilt-series alignment. A convolutional neural network learns to score alignment accuracy using a contrastive loss that does not require well-aligned ground truth, and gradient back-propagation from this score optimizes individual image alignment parameters. MissAlignment significantly outperforms reference-free techniques, rivals reference-based alignment, and improves all downstream analyses, making cryo-ET applicable to a broader range of biological samples.

Fourier-space projection operations are central to electron microscopy single-particle analysis and electron tomography algorithms. Machine learning methods require differentiable implementations for end-to-end model training, but PyTorch’s built-in operations are too slow for practical use. This paper introduces torch-projectors : a high-performance library for differentiable Fourier-space projections in PyTorch. The library provides 2D and 3D forward and backward projection operators with linear and cubic interpolation, supporting gradient calculation for all inputs. Optimized for CPU, Apple Silicon (MPS), and CUDA devices, torch-projectors outperforms torch-fourier-slice by 1–2 orders of magnitude.

BamA is the central component of the essential β-barrel assembly machine (BAM), a conserved multi-subunit complex that dynamically inserts and folds β-barrel proteins into the outer membrane of Gram-negative bacteria. Despite recent advances in our mechanistic and structural understanding of BamA, there are few potent and selective tool molecules that can bind to and modulate BamA activity. Here, we explored in vitro selection methods and different BamA/BAM protein formulations to discover peptide macrocycles that kill Escherichia coli by targeting extreme conformational states of BamA. Our studies show that Peptide Targeting BamA-1 (PTB1) targets an extracellular divalent cation-dependent binding site and locks BamA into a closed lateral gate conformation. By contrast, PTB2 targets a luminal binding site and traps BamA into an open lateral gate conformation. Our results will inform future antibiotic discovery efforts targeting BamA and provide a template to prospectively discover modulators of other dynamic integral membrane proteins.

High-resolution structures of proteins are critical to understanding molecular mechanisms of biological processes and in the discovery of therapeutic molecules. Cryo-EM has revolutionized structure determination of large proteins and their complexes ¹ , but a vast majority of proteins that underlie human diseases are small (< 50 kDa) and usually beyond its reach due to low signal-to-noise images and difficulties in particle alignment ² . Current strategies to overcome this problem increase the overall size of small protein targets using scaffold proteins that bind to the target, but are limited by inherent flexibility and not being bound to their targets in a rigid manner, resulting in the target being poorly resolved compared to the scaffolds ^3–11 . Here we present an iteratively engineered molecular design for transforming Fabs (antibody fragments), into conformationally rigid scaffolds (Rigid-Fabs) that, when bound to small proteins (∼20 kDa), can enable high-resolution structure determination using cryo-EM. This design introduces multiple disulfide bonds at strategic locations, generates a well-folded Fab constrained into a rigid conformation and can be applied to Fabs from various species, isotypes and chimeric Fabs. We present examples of the Rigid Fab design enabling high-resolution (2.3–2.5 Å) structures of small proteins, Ang2 (26 kDa) and KRAS (21 kDa) by cryo-EM. The strategies for designing disulfide constrained Rigid Fabs in our work thus establish a general approach to overcome the target size limitation of single particle cryo-EM.

Translation is the fundamental process of protein synthesis and is catalysed by the ribosome in all living cells. Here, we use cryo-electron tomography and sub-tomogram analysis to visualize the dynamics of translation inside the prokaryote Mycoplasma pneumoniae . We first obtain an in-cell atomic model for the M. pneumoniae ribosome that reveals distinct extensions of ribosomal proteins. Classification then resolves thirteen ribosome states that differ in conformation and composition and reflect intermediates during translation. Based on these states, we animate translation elongation and demonstrate how antibiotics reshape the translation landscape inside cells. During translation elongation, ribosomes often arrange in a defined manner to form polysomes. By mapping the intracellular three-dimensional organization of translating ribosomes, we show that their association into polysomes exerts a local coordination mechanism that is mediated by the ribosomal protein L9. Our work demonstrates the feasibility of visualizing molecular processes at atomic detail inside cells.

The coronavirus SARS-CoV-2 uses an RNA-dependent RNA polymerase (RdRp) to replicate and transcribe its genome. Structures of the RdRp revealed a monomeric enzyme composed of the catalytic subunit nsp12, two copies of subunit nsp8, and one copy of subunit nsp7. Here we report an alternative, dimeric form of the coronavirus polymerase and resolve its structure at 4.8 Å resolution. In this structure, the two RdRps contain only one copy of nsp8 and dimerize via their nsp7 subunits to adopt an antiparallel arrangement. We speculate that the RdRp dimer facilitates template switching during production of sub-genomic RNAs for transcription.

Cryo-electron microscopy (cryo-EM) enables macromolecular structure determination in vitro and inside cells. In addition to aligning individual particles, accurate registration of sample motion and three-dimensional deformation during exposures are crucial for achieving high-resolution reconstructions. Here we describe M, a software tool that establishes a reference-based, multi-particle refinement framework for cryo-EM data and couples a comprehensive spatial deformation model to in silico correction of electron-optical aberrations. M provides a unified optimization framework for both frame-series and tomographic tilt-series data. We show that tilt-series data can provide the same resolution as frame-series data on a purified protein specimen, indicating that the alignment step no longer limits the resolution obtainable from tomographic data. In combination with Warp and RELION, M resolves to residue level a 70S ribosome bound to an antibiotic inside intact bacterial cells. Our work provides a computational tool that facilitates structural biology in cells.

Remdesivir is the only FDA-approved drug for the treatment of COVID-19 patients. The active form of remdesivir acts as a nucleoside analog and inhibits the RNA-dependent RNA polymerase (RdRp) of coronaviruses including SARS-CoV-2. Remdesivir is incorporated by the RdRp into the growing RNA product and allows for addition of three more nucleotides before RNA synthesis stalls. Here we use synthetic RNA chemistry, biochemistry and cryo-electron microscopy to establish the molecular mechanism of remdesivir-induced RdRp stalling. We show that addition of the fourth nucleotide following remdesivir incorporation into the RNA product is impaired by a barrier to further RNA translocation. This translocation barrier causes retention of the RNA 3ʹ-nucleotide in the substrate-binding site of the RdRp and interferes with entry of the next nucleoside triphosphate, thereby stalling RdRp. In the structure of the remdesivir-stalled state, the 3ʹ-nucleotide of the RNA product is matched and located with the template base in the active center, and this may impair proofreading by the viral 3ʹ-exonuclease. These mechanistic insights should facilitate the quest for improved antivirals that target coronavirus replication.

Integrative in-cell structural biology: In bacteria, RNA polymerases can associate with ribosomes to form transcription-translation units called expressomes. Multiple models based on structural data of in vitro reconstitution assays have been proposed for how the two machineries interface with one another. Understanding this bacteria-specific coupling mechanism offers insight regarding the central dogma of molecular biology and might be leveraged for antibiotic development. O'Reilly et al. found that the NusA protein interfaces between the two complexes. The authors combined cryo–electron tomography and cross-linking mass spectrometry to produce an integrative model of the transcribing, translating expressome of Mycoplasma pneumoniae obtained entirely from in-cell data. This approach contributes to the development of in-cell structural biology. Science , this issue p. 554

Cryo-electron microscopy (cryo-EM) enables macromolecular structure determination in vitro and in situ . In addition to aligning individual particles, accurate registration of sample motion and 3D deformation during exposures is crucial for achieving high resolution. Here we describe M , a software tool that establishes a reference-based, multi-particle refinement framework for cryo-EM data and improves the results of structure determination. M provides a unified optimization framework for both in vitro frame series and in situ tomographic tilt series data. We show that tilt series data can provide the same resolution as frame series, indicating that the alignment step no longer limits the resolution obtainable from tomographic data. In combination with Warp and RELION, M improves upon previous methods, and resolves a 70S ribosome bound to an antibiotic inside bacterial cells at a nominal resolution of 3.7 Å. Thus, computational tools are now available to resolve structures from tomographic in situ cryo-EM data at residue level.