Restraint validation of biomolecular structures determined by NMR in the Protein Data Bank.

Biomolecular structure analysis from experimental NMR studies generally relies on restraints derived from a combination of experimental and knowledge-based data. A challenge for the structural biology community has been a lack of standards for representing these restraints, preventing the establishment of uniform methods of model-vs-data structure validation against restraints and limiting interoperability between restraint-based structure modeling programs. The NEF and NMR-STAR formats provide a standardized approach for representing commonly used NMR restraints. Using these restraint formats, a standardized validation system for assessing structural models of biopolymers against restraints has been developed and implemented in the wwPDB OneDep data deposition-validation-biocuration system. The resulting wwPDB restraint violation report provides a model vs. data assessment of biomolecule structures determined using distance and dihedral restraints, with extensions to other restraint types currently being implemented. These tools are useful for assessing NMR models, as well as for assessing biomolecular structure predictions based on distance restraints.

Restraint Validation of Biomolecular Structures Determined by NMR in the Protein Data Bank.

Biomolecular structure analysis from experimental NMR studies generally relies on restraints derived from a combination of experimental and knowledge-based data. A challenge for the structural biology community has been a lack of standards for representing these restraints, preventing the establishment of uniform methods of model-vs-data structure validation against restraints and limiting interoperability between restraint-based structure modeling programs. The NMR exchange (NEF) and NMR-STAR formats provide a standardized approach for representing commonly used NMR restraints. Using these restraint formats, a standardized validation system for assessing structural models of biopolymers against restraints has been developed and implemented in the wwPDB OneDep data deposition-validation-biocuration system. The resulting wwPDB Restraint Violation Report provides a model vs. data assessment of biomolecule structures determined using distance and dihedral restraints, with extensions to other restraint types currently being implemented. These tools are useful for assessing NMR models, as well as for assessing biomolecular structure predictions based on distance restraints.

The 100-protein NMR spectra dataset: A resource for biomolecular NMR data analysis.

Multidimensional NMR spectra are the basis for studying proteins by NMR spectroscopy and crucial for the development and evaluation of methods for biomolecular NMR data analysis. Nevertheless, in contrast to derived data such as chemical shift assignments in the BMRB and protein structures in the PDB databases, this primary data is in general not publicly archived. To change this unsatisfactory situation, we present a standardized set of solution NMR data comprising 1329 2-4-dimensional NMR spectra and associated reference (chemical shift assignments, structures) and derived (peak lists, restraints for structure calculation, etc.) annotations. With the 100-protein NMR spectra dataset that was originally compiled for the development of the ARTINA deep learning-based spectra analysis method, 100 protein structures can be reproduced from their original experimental data. The 100-protein NMR spectra dataset is expected to help the development of computational methods for NMR spectroscopy, in particular machine learning approaches, and enable consistent and objective comparisons of these methods.

Representing structures of the multiple conformational states of proteins.

Biomolecules exhibit dynamic behavior that single-state models of their structures cannot fully capture. We review some recent advances for investigating multiple conformations of biomolecules, including experimental methods, molecular dynamics simulations, and machine learning. We also address the challenges associated with representing single- and multiple-state models in data archives, with a particular focus on NMR structures. Establishing standardized representations and annotations will facilitate effective communication and understanding of these complex models to the broader scientific community.

Cell-permeable chameleonic peptides: Exploiting conformational dynamics in de novo cyclic peptide design.

Membrane-traversing peptides offer opportunities for targeting intracellular proteins and oral delivery. Despite progress in understanding the mechanisms underlying membrane traversal in natural cell-permeable peptides, there are still several challenges to designing membrane-traversing peptides with diverse shapes and sizes. Conformational flexibility appears to be a key determinant of membrane permeability of large macrocycles. We review recent developments in the design and validation of chameleonic cyclic peptides, which can switch between alternative conformations to enable improved permeability through cell membranes, while still maintaining reasonable solubility and exposed polar functional groups for target protein binding. Finally, we discuss the principles, strategies, and practical considerations for rational design, discovery, and validation of permeable chameleonic peptides.

Blind assessment of monomeric AlphaFold2 protein structure models with experimental NMR data.

Recent advances in molecular modeling of protein structures are changing the field of structural biology. AlphaFold-2 (AF2), an AI system developed by DeepMind, Inc., utilizes attention-based deep learning to predict models of protein structures with high accuracy relative to structures determined by X-ray crystallography and cryo-electron microscopy (cryoEM). Comparing AF2 models to structures determined using solution NMR data, both high similarities and distinct differences have been observed. Since AF2 was trained on X-ray crystal and cryoEM structures, we assessed how accurately AF2 can model small, monomeric, solution protein NMR structures which (i) were not used in the AF2 training data set, and (ii) did not have homologous structures in the Protein Data Bank at the time of AF2 training. We identified nine open-source protein NMR data sets for such "blind" targets, including chemical shift, raw NMR FID data, NOESY peak lists, and (for 1 case) 15N-1H residual dipolar coupling data. For these nine small (70-108 residues) monomeric proteins, we generated AF2 prediction models and assessed how well these models fit to these experimental NMR data, using several well-established NMR structure validation tools. In most of these cases, the AF2 models fit the NMR data nearly as well, or sometimes better than, the corresponding NMR structure models previously deposited in the Protein Data Bank. These results provide benchmark NMR data for assessing new NMR data analysis and protein structure prediction methods. They also document the potential for using AF2 as a guiding tool in protein NMR data analysis, and more generally for hypothesis generation in structural biology research.

Blind Assessment of Monomeric AlphaFold2 Protein Structure Models with Experimental NMR Data.

Recent advances in molecular modeling of protein structures are changing the field of structural biology. AlphaFold-2 (AF2), an AI system developed by DeepMind, Inc., utilizes attention-based deep learning to predict models of protein structures with high accuracy relative to structures determined by X-ray crystallography and cryo-electron microscopy (cryoEM). Comparing AF2 models to structures determined using solution NMR data, both high similarities and distinct differences have been observed. Since AF2 was trained on X-ray crystal and cryoEM structures, we assessed how accurately AF2 can model small, monomeric, solution protein NMR structures which (i) were not used in the AF2 training data set, and (ii) did not have homologous structures in the Protein Data Bank at the time of AF2 training. We identified nine open source protein NMR data sets for such "blind" targets, including chemical shift, raw NMR FID data, NOESY peak lists, and (for 1 case) 15 N- 1 H residual dipolar coupling data. For these nine small (70 - 108 residues) monomeric proteins, we generated AF2 prediction models and assessed how well these models fit to these experimental NMR data, using several well-established NMR structure validation tools. In most of these cases, the AF2 models fit the NMR data nearly as well, or sometimes better than, the corresponding NMR structure models previously deposited in the Protein Data Bank. These results provide benchmark NMR data for assessing new NMR data analysis and protein structure prediction methods. They also document the potential for using AF2 as a guiding tool in protein NMR data analysis, and more generally for hypothesis generation in structural biology research. Highlights: AF2 models assessed against NMR data for 9 monomeric proteins not used in training.AF2 models fit NMR data almost as well as the experimentally-determined structures. RPF-DP, PSVS , and PDBStat software provide structure quality and RDC assessment. RPF-DP analysis using AF2 models suggests multiple conformational states.

SpecDB: A relational database for archiving biomolecular NMR spectral data.

NMR is a valuable experimental tool in the structural biologist's toolkit to elucidate the structures, functions, and motions of biomolecules. The progress of machine learning, particularly in structural biology, reveals the critical importance of large, diverse, and reliable datasets in developing new methods and understanding in structural biology and science more broadly. Biomolecular NMR research groups produce large amounts of data, and there is renewed interest in organizing these data to train new, sophisticated machine learning architectures and to improve biomolecular NMR analysis pipelines. The foundational data type in NMR is the free-induction decay (FID). There are opportunities to build sophisticated machine learning methods to tackle long-standing problems in NMR data processing, resonance assignment, dynamics analysis, and structure determination using NMR FIDs. Our goal in this study is to provide a lightweight, broadly available tool for archiving FID data as it is generated at the spectrometer, and grow a new resource of FID data and associated metadata. This study presents a relational schema for storing and organizing the metadata items that describe an NMR sample and FID data, which we call Spectral Database (SpecDB). SpecDB is implemented in SQLite and includes a Python software library providing a command-line application to create, organize, query, backup, share, and maintain the database. This set of software tools and database schema allow users to store, organize, share, and learn from NMR time domain data. SpecDB is freely available under an open source license at https://github.rpi.edu/RPIBioinformatics/SpecDB.

Accurate de novo design of membrane-traversing macrocycles.

We use computational design coupled with experimental characterization to systematically investigate the design principles for macrocycle membrane permeability and oral bioavailability. We designed 184 6-12 residue macrocycles with a wide range of predicted structures containing noncanonical backbone modifications and experimentally determined structures of 35; 29 are very close to the computational models. With such control, we show that membrane permeability can be systematically achieved by ensuring all amide (NH) groups are engaged in internal hydrogen bonding interactions. 84 designs over the 6-12 residue size range cross membranes with an apparent permeability greater than 1 × 10-6 cm/s. Designs with exposed NH groups can be made membrane permeable through the design of an alternative isoenergetic fully hydrogen-bonded state favored in the lipid membrane. The ability to robustly design membrane-permeable and orally bioavailable peptides with high structural accuracy should contribute to the next generation of designed macrocycle therapeutics.

AlphaFold Models of Small Proteins Rival the Accuracy of Solution NMR Structures.

Recent advances in molecular modeling using deep learning have the potential to revolutionize the field of structural biology. In particular, AlphaFold has been observed to provide models of protein structures with accuracies rivaling medium-resolution X-ray crystal structures, and with excellent atomic coordinate matches to experimental protein NMR and cryo-electron microscopy structures. Here we assess the hypothesis that AlphaFold models of small, relatively rigid proteins have accuracies (based on comparison against experimental data) similar to experimental solution NMR structures. We selected six representative small proteins with structures determined by both NMR and X-ray crystallography, and modeled each of them using AlphaFold. Using several structure validation tools integrated under the Protein Structure Validation Software suite (PSVS), we then assessed how well these models fit to experimental NMR data, including NOESY peak lists (RPF-DP scores), comparisons between predicted rigidity and chemical shift data (ANSURR scores), and 15N-1H residual dipolar coupling data (RDC Q factors) analyzed by software tools integrated in the PSVS suite. Remarkably, the fits to NMR data for the protein structure models predicted with AlphaFold are generally similar, or better, than for the corresponding experimental NMR or X-ray crystal structures. Similar conclusions were reached in comparing AlphaFold2 predictions and NMR structures for three targets from the Critical Assessment of Protein Structure Prediction (CASP). These results contradict the widely held misperception that AlphaFold cannot accurately model solution NMR structures. They also document the value of PSVS for model vs. data assessment of protein NMR structures, and the potential for using AlphaFold models for guiding analysis of experimental NMR data and more generally in structural biology.

De novo protein design by deep network hallucination.

There has been considerable recent progress in protein structure prediction using deep neural networks to predict inter-residue distances from amino acid sequences1-3. Here we investigate whether the information captured by such networks is sufficiently rich to generate new folded proteins with sequences unrelated to those of the naturally occurring proteins used in training the models. We generate random amino acid sequences, and input them into the trRosetta structure prediction network to predict starting residue-residue distance maps, which, as expected, are quite featureless. We then carry out Monte Carlo sampling in amino acid sequence space, optimizing the contrast (Kullback-Leibler divergence) between the inter-residue distance distributions predicted by the network and background distributions averaged over all proteins. Optimization from different random starting points resulted in novel proteins spanning a wide range of sequences and predicted structures. We obtained synthetic genes encoding 129 of the network-'hallucinated' sequences, and expressed and purified the proteins in Escherichia coli; 27 of the proteins yielded monodisperse species with circular dichroism spectra consistent with the hallucinated structures. We determined the three-dimensional structures of three of the hallucinated proteins, two by X-ray crystallography and one by NMR, and these closely matched the hallucinated models. Thus, deep networks trained to predict native protein structures from their sequences can be inverted to design new proteins, and such networks and methods should contribute alongside traditional physics-based models to the de novo design of proteins with new functions.

Hepatitis C virus drugs that inhibit SARS-CoV-2 papain-like protease synergize with remdesivir to suppress viral replication in cell culture.

Effective control of COVID-19 requires antivirals directed against SARS-CoV-2. We assessed 10 hepatitis C virus (HCV) protease-inhibitor drugs as potential SARS-CoV-2 antivirals. There is a striking structural similarity of the substrate binding clefts of SARS-CoV-2 main protease (Mpro) and HCV NS3/4A protease. Virtual docking experiments show that these HCV drugs can potentially bind into the Mpro substrate-binding cleft. We show that seven HCV drugs inhibit both SARS-CoV-2 Mpro protease activity and SARS-CoV-2 virus replication in Vero and/or human cells. However, their Mpro inhibiting activities did not correlate with their antiviral activities. This conundrum is resolved by demonstrating that four HCV protease inhibitor drugs, simeprevir, vaniprevir, paritaprevir, and grazoprevir inhibit the SARS CoV-2 papain-like protease (PLpro). HCV drugs that inhibit PLpro synergize with the viral polymerase inhibitor remdesivir to inhibit virus replication, increasing remdesivir's antiviral activity as much as 10-fold, while those that only inhibit Mpro do not synergize with remdesivir.

Proanthocyanidin Structural Details Revealed by Ultrahigh Resolution FT-ICR MALDI-Mass Spectrometry, 1H-13C HSQC NMR, and Thiolysis-HPLC-DAD.

Proanthocyanidins (condensed tannins) are important in food chemistry, agriculture, and health, driving demand for improvements in structure determination. We used ultrahigh resolution Fourier transform-ion cyclotron resonance mass spectrometry (FT-ICR MS) methods to determine the exact composition of individual species in heterogeneous mixtures of proanthocyanidin polymers from Sorghum bicolor grain and Neptunia lutea leaves. Fragmentation patterns obtained with FT-ICR ESI MS-MS (electrospray ionization) confirmed structural details from thiolysis-high-performance liquid chromatography (HPLC)-diode array detection (DAD) and 1H-13C heteronuclear single quantum coherence (HSQC) NMR. We found that A-type linkages were characteristic of shorter polymers in predominantly B-linked proanthocyanidin. We suggest that supramolecular complex formation between proanthocyanidins and matrix components such as 2,5-dihydroxybenzoic acid was responsible for anomalous 152 dalton peaks, incorrectly assigned as 3-O-galloylation, when using FT-ICR matrix-assisted laser desorption ionization (MALDI-MS). Our data illustrate the power of the ultrahigh resolution FT-ICR methods but include the caveat that MALDI-MS must be paired with complementary analytical tools to avoid artifacts.

Mapping protein-polymer conformations in bioconjugates with atomic precision.

Rational design of protein-polymer bioconjugates is hindered by limited experimental data and mechanistic understanding on interactions between the two. In this communication, nuclear magnetic resonance (NMR) paramagnetic relaxation enhancement (PRE) reports on distances between paramagnetic spin labels and NMR active nuclei, informing on the conformation of conjugated polymers. 1H/15N-heteronuclear single quantum coherence (HSQC) NMR spectra were collected for ubiquitin (Ub) modified with block copolymers incorporating spin labels at different positions along their backbone. The resultant PRE data show that the conjugated polymers have conformations biased towards the nonpolar ß-sheet face of Ub, rather than behaving as if in solution. The bioconjugates are stabilized against denaturation by guanidine-hydrochloride, as measured by circular dichroism (CD), and this stabilization is attributed to the interaction between the protein and conjugated polymer.

Solution NMR structure of CGL2373, a polyketide cyclase-like protein from Corynebacterium glutamicum.

Protein CGL2373 from Corynebacterium glutamicum was previously proposed to be a member of the polyketide_cyc2 family, based on amino-acid sequence and secondary structure features derived from NMR chemical shift assignments. We report here the solution NMR structure of CGL2373, which contains three α-helices and one antiparallel ß-sheet and adopts a helix-grip fold. This structure shows moderate similarities to the representative polyketide cyclases, TcmN, WhiE, and ZhuI. Nevertheless, unlike the structures of these homologs, CGL2373 structure looks like a half-open shell with a much larger pocket, and key residues in the representative polyketide cyclases for binding substrate and catalyzing aromatic ring formation are replaced with different residues in CGL2373. Also, the gene cluster where the CGL2373-encoding gene is located in C. glutamicum contains additional genes encoding nucleoside diphosphate kinase, folylpolyglutamate synthase, and valine-tRNA ligase, different from the typical gene cluster encoding polyketide cyclase in Streptomyces. Thus, although CGL2373 is structurally a polyketide cyclase-like protein, the function of CGL2373 may differ from the known polyketide cyclases and needs to be further investigated. The solution structure of CGL2373 lays a foundation for in silico ligand screening and binding site identifying in future functional study.

Structural insight into the length-dependent binding of ssDNA by SP_0782 from Streptococcus pneumoniae, reveals a divergence in the DNA-binding interface of PC4-like proteins.

SP_0782 from Streptococcus pneumoniae is a dimeric protein that potentially binds with single-stranded DNA (ssDNA) in a manner similar to human PC4, the prototype of PC4-like proteins, which plays roles in transcription and maintenance of genome stability. In a previous NMR study, SP_0782 exhibited an ssDNA-binding property different from YdbC, a prokaryotic PC4-like protein from Lactococcus lactis, but the underlying mechanism remains unclear. Here, we show that although SP_0782 adopts an overall fold similar to those of PC4 and YdbC, the ssDNA length occupied by SP_0782 is shorter than those occupied by PC4 and YdbC. SP_0782 exhibits varied binding patterns for different lengths of ssDNA, and tends to form large complexes with ssDNA in a potential high-density binding manner. The structures of SP_0782 complexed with different ssDNAs reveal that the varied binding patterns are associated with distinct capture of nucleotides in two major DNA-binding regions of SP_0782. Moreover, a comparison of known structures of PC4-like proteins complexed with ssDNA reveals a divergence in the binding interface between prokaryotic and eukaryotic PC4-like proteins. This study provides insights into the ssDNA-binding mechanism of PC4-like proteins, and benefits further study regarding the biological function of SP_0782, probably in DNA protection and natural transformation.

In vivo production of psilocybin in E. coli.

Psilocybin, the prodrug of the psychoactive molecule psilocin, has demonstrated promising results in clinical trials for the treatment of addiction, depression, and post-traumatic stress disorder. The development of a psilocybin production platform in a highly engineerable microbe could lead to rapid advances towards the bioproduction of psilocybin for use in ongoing clinical trials. Here, we present the development of a modular biosynthetic production platform in the model microbe, Escherichia coli. Efforts to optimize and improve pathway performance using multiple genetic optimization techniques were evaluated, resulting in a 32-fold improvement in psilocybin titer. Further enhancements to this genetically superior strain were achieved through fermentation optimization, ultimately resulting in a fed-batch fermentation study, with a production titer of 1.16 g/L of psilocybin. This is the highest psilocybin titer achieved to date from a recombinant organism and a significant step towards demonstrating the feasibility of industrial production of biologically-derived psilocybin.

Solution NMR structure and ligand identification of human Gas7 SH3 domain reveal a typical SH3 fold but a non-canonical ligand-binding mode.

Growth arrest specific 7 (Gas7) protein is a cytoskeleton regulator playing a crucial role in neural cell development and function, and has been implicated in Alzheimer disease, schizophrenia and cancers. In human, three Gas7 isoforms can be expressed from a single Gas7 gene, while only the longest isoform, hGas7c, possesses an SH3 domain at the N-terminus. To date, the structure and function of hGas7 SH3 domain are still unclear. Here, we reported the solution NMR structure of hGas7 SH3 domain (hGas7-SH3), which displays a typical SH3 ß-barrel fold comprising five ß-strands and one 310-helix. Structural and sequence comparison showed that hGas7-SH3 shares high similarity with Abl SH3 domain, which binds to a high-affinity proline-rich peptide P41 in a canonical SH3-ligand binding mode through two hydrophobic pockets and a specificity site in the RT-loop. However, unlike Abl-SH3, only six residues in the RT-loop and two residues adjacent to but not in the two hydrophobic pockets of hGas7-SH3 showed significant chemical shift perturbations in NMR titrations, suggesting a low affinity and a non-canonical binding mode of hGas7-SH3 for P41. Furthermore, four peptides selected from phage-displayed libraries also bound weakly to hGas7-SH3, and the binding region of hGas7-SH3 was mainly located in the RT-loop as well. The ligand identifications through structural similarity searching and peptide library screening in this study imply that although hGas7-SH3 adopts a typical SH3 fold, it probably possesses distinctive ligand-binding specificity.

1H, 13C, and 15N backbone resonance assignments for KPC-2, a class A serine-ß-lactamase.

The ever-increasing occurrence of antibiotic resistance presents a major threat to public health. Specifically, resistance conferred by ß-lactamases places the efficacy of currently available antibiotics at risk. Klebsiella pneumoniae carbapenemase-2 (KPC-2) is a ß-lactamase that enables carbapenem resistance and represents a clear and present danger to global public health. In order to combat bacterial infections harboring KPC-2 expression, inhibitors with improved potency need to be developed. Although the structure of KPC-2 has been solved by X-ray crystallography, NMR provides the unique opportunity to study the structure and dynamics of flexible loop regions in solution. Here we report the 1H, 13C, and 15N backbone chemical shift assignments for KPC-2 in the apo state as the first step towards the study of KPC-2 dynamics in the presence and absence of ligands to enable the rational design of optimized inhibitors.

Solution NMR structure of CHU_1110 from Cytophaga hutchinsonii, an AHSA1 protein potentially involved in metal ion stress response.

We report the solution nuclear magnetic resonance (NMR) structure of CHU_1110 from Cytophaga hutchinsonii. CHU_1110 contains three α-helices and one antiparallel ß-sheet, forming a large cavity in the center of the protein, which are consistent with the structural characteristics of AHSA1 protein family. This protein shows high structural similarities to the prokaryotic proteins RHE_CH02687 from Rhizobium etli and YndB from Bacillus subtilis, which can bind with flavinoids. Unlike these two homologs, CHU_1110 shows no obvious interaction with flavonoids in NMR titration experiments. In addition, no direct interaction has been observed between CHU_1110 and ATP, although many homologous sequences of CHU_1110 have been annotated as ATPase. Combining the analysis of structural similarity of CHU_1110 and genomic context of its encoding gene, we speculate that CHU_1110 may be involved in the stress response of bacteria to heavy metal ions, even though its specific biological functions that need to be further investigated.