Structure of the 70S Ribosome from the Human Pathogen Acinetobacter baumannii in Complex with Clinically Relevant Antibiotics

T cell toxicity induced by tigecycline binding to the mitochondrial ribosome

Tetracyclines are essential bacterial protein synthesis inhibitors under continual development to combat antibiotic resistance yet suffer from unwanted side effects. Mitoribosomes - responsible for generating oxidative phosphorylation (OXPHOS) subunits - share structural similarities with bacterial machinery and may suffer from cross-reactivity. Since lymphocytes rely upon OXPHOS upregulation to establish immunity, we set out to assess the impact of ribosome-targeting antibiotics on human T cells. We find tigecycline, a third-generation tetracycline, to be the most cytotoxic compound tested. In vitro, 5-10 μM tigecycline inhibits mitochondrial but not cytosolic translation, mitochondrial complex I, III and IV expression, and curtails the activation and expansion of unique T cell subsets. By cryo-EM, we find tigecycline to occupy three sites on T cell mitoribosomes. In addition to the conserved A-site found in bacteria, tigecycline also attaches to the peptidyl transferase center of the large subunit. Furthermore, a third, distinct binding site on the large subunit, aligns with helices analogous to those in bacteria, albeit lacking methylation in humans. The data provide a mechanism to explain part of the anti-inflammatory effects of these drugs and inform antibiotic design.

Creation of a macrolide antibiotic against non-tuberculous Mycobacterium using late-stage boron-mediated aglycon delivery

Non-tuberculous mycobacteria (NTM) is gaining clinical recognition as a recently emerging pulmonary pathogen. Mycobacterium avium complex (MAC), the most common NTM, is the cause of pulmonary MAC disease. Currently, the macrolide azithromycin (AZM) is the standard first-line antibiotic for treatment of the disease. However, the rise of drug-resistant MAC necessitates the development of alternative therapeutics. Here, we present a late-stage boron-mediated aglycon delivery strategy for selective modification of AZM, generating a library of potential anti-MAC drugs designated KU01 to KU13. Screening of KU01 to KU13 revealed that KU13 exhibited enhanced antimicrobial activity against wild-type and macrolide-resistant MAC compared to AZM. Cryo-electron microscopy analysis indicated that the inserted tercyclic moiety of KU13 formed a robust anchor on the bacterial ribosome, creating a binding pocket with base flipping of U2847, potentially bypassing the standard mechanism of macrolide resistance. These results position KU13 as a promising lead for therapeutics against macrolide-resistant MAC.

The Janus Effect: The Biochemical Logic of Antibiotic Resistance

Antibiotics are essential medicines threatened by the emergence of resistance in all relevant bacterial pathogens. The engagement of the molecular targets of antibiotics offers multiple opportunities for resistance to emerge. Successful target engagement often requires passage of the antibiotic from outside into the cell interior through one or two distinct membrane barriers. Resistance can occur by preventing the accumulation of antibiotics in sufficient quantities outside the cell, decreasing the rates of entry into the cell, and modifying the antibiotic or the target once inside the cell. These competing equilibria and rates are the lens through which the balance of antibiotic efficacy or failure can be viewed. The two faces of antibiotics, cell growth inhibition or resistance, are reminiscent of Janus, the Roman god of doorways and beginnings and endings, and offer a framework through which antibiotic discovery, use, and the emergence of resistance can be rationally viewed.

Overcoming the preferred-orientation problem in cryo-EM with self-supervised deep learning

While advances in single-particle cryo-EM have enabled the structural determination of macromolecular complexes at atomic resolution, particle orientation bias (the 'preferred' orientation problem) remains a complication for most specimens. Existing solutions have relied on biochemical and physical strategies applied to the specimen and are often complex and challenging. Here, we develop spIsoNet, an end-to-end self-supervised deep learning-based software to address map anisotropy and particle misalignment caused by the preferred-orientation problem. Using preferred-orientation views to recover molecular information in under-sampled views, spIsoNet improves both angular isotropy and particle alignment accuracy during 3D reconstruction. We demonstrate spIsoNet's ability to generate near-isotropic reconstructions from representative biological systems with limited views, including ribosomes, β-galactosidases and a previously intractable hemagglutinin trimer dataset. spIsoNet can also be generalized to improve map isotropy and particle alignment of preferentially oriented molecules in subtomogram averaging. Therefore, without additional specimen-preparation procedures, spIsoNet provides a general computational solution to the preferred-orientation problem.

Structural basis for differential inhibition of eukaryotic ribosomes by tigecycline

Tigecycline is widely used for treating complicated bacterial infections for which there are no effective drugs. It inhibits bacterial protein translation by blocking the ribosomal A-site. However, even though it is also cytotoxic for human cells, the molecular mechanism of its inhibition remains unclear. Here, we present cryo-EM structures of tigecycline-bound human mitochondrial 55S, 39S, cytoplasmic 80S and yeast cytoplasmic 80S ribosomes. We find that at clinically relevant concentrations, tigecycline effectively targets human 55S mitoribosomes, potentially, by hindering A-site tRNA accommodation and by blocking the peptidyl transfer center. In contrast, tigecycline does not bind to human 80S ribosomes under physiological concentrations. However, at high tigecycline concentrations, in addition to blocking the A-site, both human and yeast 80S ribosomes bind tigecycline at another conserved binding site restricting the movement of the L1 stalk. In conclusion, the observed distinct binding properties of tigecycline may guide new pathways for drug design and therapy.

Overcoming the preferred orientation problem in cryoEM with self-supervised deep-learning

While advances in single-particle cryoEM have enabled the structural determination of macromolecular complexes at atomic resolution, particle orientation bias (the so-called "preferred" orientation problem) remains a complication for most specimens. Existing solutions have relied on biochemical and physical strategies applied to the specimen and are often complex and challenging. Here, we develop spIsoNet, an end-to-end self-supervised deep-learning-based software to address the preferred orientation problem. Using preferred-orientation views to recover molecular information in under-sampled views, spIsoNet improves both angular isotropy and particle alignment accuracy during 3D reconstruction. We demonstrate spIsoNet's capability of generating near-isotropic reconstructions from representative biological systems with limited views, including ribosomes, β-galactosidases, and a previously intractable hemagglutinin trimer dataset. spIsoNet can also be generalized to improve map isotropy and particle alignment of preferentially oriented molecules in subtomogram averaging. Therefore, without additional specimen-preparation procedures, spIsoNet provides a general computational solution to the preferred orientation problem.

Accurate cryo-EM protein particle picking by integrating the foundational AI image segmentation model and specialized U-Net

Picking protein particles in cryo-electron microscopy (cryo-EM) micrographs is a crucial step in the cryo-EM-based structure determination. However, existing methods trained on a limited amount of cryo-EM data still cannot accurately pick protein particles from noisy cryo-EM images. The general foundational artificial intelligence (AI)-based image segmentation model such as Meta's Segment Anything Model (SAM) cannot segment protein particles well because their training data do not include cryo-EM images. Here, we present a novel approach (CryoSegNet) of integrating an attention-gated U-shape network (U-Net) specially designed and trained for cryo-EM particle picking and the SAM. The U-Net is first trained on a large cryo-EM image dataset and then used to generate input from original cryo-EM images for SAM to make particle pickings. CryoSegNet shows both high precision and recall in segmenting protein particles from cryo-EM micrographs, irrespective of protein type, shape, and size. On several independent datasets of various protein types, CryoSegNet outperforms two top machine learning particle pickers crYOLO and Topaz as well as SAM itself. The average resolution of density maps reconstructed from the particles picked by CryoSegNet is 3.32 Å, 7% better than 3.57 Å of Topaz and 14% better than 3.85 Å of crYOLO.

CryoTransformer: A Transformer Model for Picking Protein Particles from Cryo-EM Micrographs

Cryo-electron microscopy (cryo-EM) is a powerful technique for determining the structures of large protein complexes. Picking single protein particles from cryo-EM micrographs (images) is a crucial step in reconstructing protein structures from them. However, the widely used template-based particle picking process requires some manual particle picking and is labor-intensive and time-consuming. Though machine learning and artificial intelligence (AI) can potentially automate particle picking, the current AI methods pick particles with low precision or low recall. The erroneously picked particles can severely reduce the quality of reconstructed protein structures, especially for the micrographs with low signal-to-noise (SNR) ratios. To address these shortcomings, we devised CryoTransformer based on transformers, residual networks, and image processing techniques to accurately pick protein particles from cryo-EM micrographs. CryoTransformer was trained and tested on the largest labelled cryo-EM protein particle dataset - CryoPPP. It outperforms the current state-of-the-art machine learning methods of particle picking in terms of the resolution of 3D density maps reconstructed from the picked particles as well as F1-score and is poised to facilitate the automation of the cryo-EM protein particle picking.

Structural conservation of antibiotic interaction with ribosomes

The ribosome is a major target for clinically used antibiotics, but multidrug resistant pathogenic bacteria are making our current arsenal of antimicrobials obsolete. Here we present cryo-electron-microscopy structures of 17 distinct compounds from six different antibiotic classes bound to the bacterial ribosome at resolutions ranging from 1.6 to 2.2 Å. The improved resolution enables a precise description of antibiotic-ribosome interactions, encompassing solvent networks that mediate multiple additional interactions between the drugs and their target. Our results reveal a high structural conservation in the binding mode between antibiotics with the same scaffold, including ordered water molecules. Water molecules are visualized within the antibiotic binding sites that are preordered, become ordered in the presence of the drug and that are physically displaced on drug binding. Insight into RNA-ligand interactions will facilitate development of new antimicrobial agents, as well as other RNA-targeting therapies.

Molecular basis of the pleiotropic effects by the antibiotic amikacin on the ribosome

Aminoglycosides are a class of antibiotics that bind to ribosomal RNA and exert pleiotropic effects on ribosome function. Amikacin, the semisynthetic derivative of kanamycin, is commonly used for treating severe infections with multidrug-resistant, aerobic Gram-negative bacteria. Amikacin carries the 4-amino-2-hydroxy butyrate (AHB) moiety at the N1 amino group of the central 2-deoxystreptamine (2-DOS) ring, which may confer amikacin a unique ribosome inhibition profile. Here we use in vitro fast kinetics combined with X-ray crystallography and cryo-EM to dissect the mechanisms of ribosome inhibition by amikacin and the parent compound, kanamycin. Amikacin interferes with tRNA translocation, release factor-mediated peptidyl-tRNA hydrolysis, and ribosome recycling, traits attributed to the additional interactions amikacin makes with the decoding center. The binding site in the large ribosomal subunit proximal to the 3'-end of tRNA in the peptidyl (P) site lays the groundwork for rational design of amikacin derivatives with improved antibacterial properties.

A large expert-curated cryo-EM image dataset for machine learning protein particle picking

Cryo-electron microscopy (cryo-EM) is a powerful technique for determining the structures of biological macromolecular complexes. Picking single-protein particles from cryo-EM micrographs is a crucial step in reconstructing protein structures. However, the widely used template-based particle picking process is labor-intensive and time-consuming. Though machine learning and artificial intelligence (AI) based particle picking can potentially automate the process, its development is hindered by lack of large, high-quality labelled training data. To address this bottleneck, we present CryoPPP, a large, diverse, expert-curated cryo-EM image dataset for protein particle picking and analysis. It consists of labelled cryo-EM micrographs (images) of 34 representative protein datasets selected from the Electron Microscopy Public Image Archive (EMPIAR). The dataset is 2.6 terabytes and includes 9,893 high-resolution micrographs with labelled protein particle coordinates. The labelling process was rigorously validated through 2D particle class validation and 3D density map validation with the gold standard. The dataset is expected to greatly facilitate the development of both AI and classical methods for automated cryo-EM protein particle picking.

Ribosome-targeting antibiotics and resistance via ribosomal RNA methylation

The rise of multidrug-resistant bacterial infections is a cause of global concern. There is an urgent need to both revitalize antibacterial agents that are ineffective due to resistance while concurrently developing new antibiotics with novel targets and mechanisms of action. Pathogen associated resistance-conferring ribosomal RNA (rRNA) methyltransferases are a growing threat that, as a group, collectively render a total of seven clinically-relevant ribosome-targeting antibiotic classes ineffective. Increasing frequency of identification and their growing prevalence relative to other resistance mechanisms suggests that these resistance determinants are rapidly spreading among human pathogens and could contribute significantly to the increased likelihood of a post-antibiotic era. Herein, with a view toward stimulating future studies to counter the effects of these rRNA methyltransferases, we summarize their prevalence, the fitness cost(s) to bacteria of their acquisition and expression, and current efforts toward targeting clinically relevant enzymes of this class.

Sarecycline inhibits protein translation in Cutibacterium acnes 70S ribosome using a two-site mechanism

Acne vulgaris is a chronic disfiguring skin disease affecting ∼1 billion people worldwide, often having persistent negative effects on physical and mental health. The Gram-positive anaerobe, Cutibacterium acnes is implicated in acne pathogenesis and is, therefore, a main target for antibiotic-based acne therapy. We determined a 2.8-Å resolution structure of the 70S ribosome of Cutibacterium acnes by cryogenic electron microscopy and discovered that sarecycline, a narrow-spectrum antibiotic against Cutibacterium acnes, may inhibit two active sites of this bacterium's ribosome in contrast to the one site detected previously on the model ribosome of Thermus thermophilus. Apart from the canonical binding site at the mRNA decoding center, the second binding site for sarecycline exists at the nascent peptide exit tunnel, reminiscent of the macrolides class of antibiotics. The structure also revealed Cutibacterium acnes-specific features of the ribosomal RNA and proteins. Unlike the ribosome of the Gram-negative bacterium Escherichia coli, Cutibacterium acnes ribosome has two additional proteins, bS22 and bL37, which are also present in the ribosomes of Mycobacterium smegmatis and Mycobacterium tuberculosis. We show that bS22 and bL37 have antimicrobial properties and may be involved in maintaining the healthy homeostasis of the human skin microbiome.

The small mycobacterial ribosomal protein, bS22, modulates aminoglycoside accessibility to its 16S rRNA helix-44 binding site

Treatment of tuberculosis continues to be challenging due to the widespread latent form of the disease and the emergence of antibiotic-resistant strains of the pathogen, Mycobacterium tuberculosis. Bacterial ribosomes are a common and effective target for antibiotics. Several second line anti-tuberculosis drugs, e.g. kanamycin, amikacin, and capreomycin, target ribosomal RNA to inhibit protein synthesis. However, M. tuberculosis can acquire resistance to these drugs, emphasizing the need to identify new drug targets. Previous cryo-EM structures of the M. tuberculosis and M. smegmatis ribosomes identified two novel ribosomal proteins, bS22 and bL37, in the vicinity of two crucial drug-binding sites: the mRNA-decoding center on the small (30S), and the peptidyl-transferase center on the large (50S) ribosomal subunits, respectively. The functional significance of these two small proteins is unknown. In this study, we observe that an M. smegmatis strain lacking the bs22 gene shows enhanced susceptibility to kanamycin compared to the wild-type strain. Cryo-EM structures of the ribosomes lacking bS22 in the presence and absence of kanamycin suggest a direct role of bS22 in modulating the 16S rRNA kanamycin-binding site. Our structures suggest that amino-acid residue Lys-16 of bS22 interacts directly with the phosphate backbone of helix 44 of 16S rRNA to influence the micro-configuration of the kanamycin-binding pocket. Our analysis shows that similar interactions occur between eukaryotic homologues of bS22, and their corresponding rRNAs, pointing to a common mechanism of aminoglycoside resistance in higher organisms.

CryoPPP: A Large Expert-Labelled Cryo-EM Image Dataset for Machine Learning Protein Particle Picking

Cryo-electron microscopy (cryo-EM) is currently the most powerful technique for determining the structures of large protein complexes and assemblies. Picking single-protein particles from cryo-EM micrographs (images) is a key step in reconstructing protein structures. However, the widely used template-based particle picking process is labor-intensive and time-consuming. Though the emerging machine learning-based particle picking can potentially automate the process, its development is severely hindered by lack of large, high-quality, manually labelled training data. Here, we present CryoPPP, a large, diverse, expert-curated cryo-EM image dataset for single protein particle picking and analysis to address this bottleneck. It consists of manually labelled cryo-EM micrographs of 32 non-redundant, representative protein datasets selected from the Electron Microscopy Public Image Archive (EMPIAR). It includes 9,089 diverse, high-resolution micrographs (∼300 cryo-EM images per EMPIAR dataset) in which the coordinates of protein particles were labelled by human experts. The protein particle labelling process was rigorously validated by both 2D particle class validation and 3D density map validation with the gold standard. The dataset is expected to greatly facilitate the development of machine learning and artificial intelligence methods for automated cryo-EM protein particle picking. The dataset and data processing scripts are available at https://github.com/BioinfoMachineLearning/cryoppp.

A Transfer Learning-Based Classification Model for Particle Pruning in Cryo-Electron Microscopy

The cryo-electron microscopy (cryo-EM) single-particle analysis requires tens of thousands of particle projections to reveal structural information of macromolecular complexes. However, due to the low signal-to-noise ratio and the presence of high contrast artifacts and contaminants in the micrographs, the semiautomatic and fully automatic particle picking algorithms tend to suffer from high false-positive rates, which degrades the confidence of structure determination. In this study, we introduce PickerOptimizer (PO), a transfer learning-based classification neural network for particle pruning in cryo-EM, as an additional strategy to complement the current automated particle picking algorithms. To achieve high classification performance with minimal human intervention, we adopted two key strategies: (1) utilizing the transfer learning techniques to train the convolutional neural network, where the knowledge gained from public classification datasets is applied to the field of cryo-EM. (2) Designing a multiloss strategy, a combination of multiple loss functions, to guide the optimization of the network parameters. To reduce the domain shift between cryo-EM images and natural images for pretraining, we build the first image classification dataset for cryo-EM, which contains positive and negative samples collected from EMPIAR entries. The PO is tested on 14 public experimental datasets, achieving accuracy and F1 scores above 95% in most cases. Furthermore, three case studies are provided to verify the model performance by applying PO on problematic particle selections, showing that our algorithm achieved better or comparable performance compared with other particle pruning strategies.

Crystal Structure of the Disease-Specific Protein of Rice Stripe Virus

The rice stripe virus (RSV) is responsible for devastating effects in East Asian rice-producing areas. The disease-specific protein (SP) level in rice plants determines the severity of RSV symptoms. Isothermal titration calorimetry (ITC) and bimolecular fluorescence complementation (BiFC) assays confirmed the interaction between an R3H domain-containing host factor, OsR3H3, and RSV SP in vitro and in vivo. This study determined the crystal structure of SP at 1.71 Å. It is a monomer with a clear shallow groove to accommodate host factors. Docking OsR3H3 into the groove generates an SP/OsR3H3 complex, which provides insights into the protein-binding mechanism of SP. Furthermore, SP's protein-binding properties and model-defined recognition residues were assessed using mutagenesis, ITC, and BiFC assays. This study revealed the structure and preliminary protein interaction mechanisms of RSV SP, shedding light on the molecular mechanism underlying the development of RSV infection symptoms.

Mechanistic insights into translation inhibition by aminoglycoside antibiotic arbekacin

How aminoglycoside antibiotics limit bacterial growth and viability is not clearly understood. Here we employ fast kinetics to reveal the molecular mechanism of action of a clinically used, new-generation, semisynthetic aminoglycoside Arbekacin (ABK), which is designed to avoid enzyme-mediated deactivation common to other aminoglycosides. Our results portray complete picture of ABK inhibition of bacterial translation with precise quantitative characterizations. We find that ABK inhibits different steps of translation in nanomolar to micromolar concentrations by imparting pleotropic effects. ABK binding stalls elongating ribosomes to a state, which is unfavorable for EF-G binding. This prolongs individual translocation step from ∼50 ms to at least 2 s; the mean time of translocation increases inversely with EF-G concentration. ABK also inhibits translation termination by obstructing RF1/RF2 binding to the ribosome. Furthermore, ABK decreases accuracy of mRNA decoding (UUC vs. CUC) by ∼80 000 fold, causing aberrant protein production. Importantly, translocation and termination events cannot be completely stopped even with high ABK concentration. Extrapolating our kinetic model of ABK action, we postulate that aminoglycosides impose bacteriostatic effect mainly by inhibiting translocation, while they become bactericidal in combination with decoding errors.

Cryo-EM Determination of Eravacycline-Bound Structures of the Ribosome and the Multidrug Efflux Pump AdeJ of Acinetobacter baumannii

Antibiotic-resistant strains of the Gram-negative pathogen Acinetobacter baumannii have emerged as a significant global health threat. One successful therapeutic option to treat bacterial infections has been to target the bacterial ribosome. However, in many cases, multidrug efflux pumps within the bacterium recognize and extrude these clinically important antibiotics designed to inhibit the protein synthesis function of the bacterial ribosome. Thus, multidrug efflux within A. baumannii and other highly drug-resistant strains is a major cause of failure of drug-based treatments of infectious diseases. We here report the first structures of the Acinetobacterdrug efflux (Ade)J pump in the presence of the antibiotic eravacycline, using single-particle cryo-electron microscopy (cryo-EM). We also describe cryo-EM structures of the eravacycline-bound forms of the A. baumannii ribosome, including the 70S, 50S, and 30S forms. Our data indicate that the AdeJ pump primarily uses hydrophobic interactions to bind eravacycline, while the 70S ribosome utilizes electrostatic interactions to bind this drug. Our work here highlights how an antibiotic can bind multiple bacterial targets through different mechanisms and potentially enables drug optimization by taking advantage of these different modes of ligand binding.