Computational modeling multiple conformational states of proteins with residual dipolar coupling data.

Solution nuclear magnetic resonance spectroscopy provides unique opportunities to study the structure and dynamics of biomolecules in aqueous environments. While spin relaxation methods are well recognized for their ability to probe timescales of motion, residual dipolar couplings (RDCs) provide access to amplitudes and directions of motion, characteristics that are important to the function of these molecules. Although observed in the 1960s, the acquisition and computational analysis of RDCs has gained significant momentum in recent years, and particularly applications to motion in proteins have become more numerous. This trend may well continue as RDCs can easily leverage structures produced by new computational methods (e.g., AlphaFold) to produce functional descriptions. In this report, we provide examples and a summary of the ways that RDCs have been used to confirm the existence of internal dynamics, characterize the type of dynamics, and recover atomic-scale structural ensembles that define the full range of conformational sampling.

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.

Glycan Conformation in the Heavily Glycosylated Protein, CEACAM1.

Glycans attached to glycoproteins can contribute to stability, mediate interactions with other proteins, and initiate signal transduction. Glycan conformation, which is critical to these processes, is highly variable and often depicted as sampling a multitude of conformers. These conformers can be generated by molecular dynamics simulations, and more inclusively by accelerated molecular dynamics, as well as other extended sampling methods. However, experimental assessments of the contribution that various conformers make to a native ensemble are rare. Here, we use long-range pseudo-contact shifts (PCSs) of NMR resonances from an isotopically labeled glycoprotein to identify preferred conformations of its glycans. The N-terminal domain from human Carcinoembryonic Antigen Cell Adhesion Molecule 1, hCEACAM1-Ig1, was used as the model glycoprotein in this study. It has been engineered to include a lanthanide-ion-binding loop that generates PCSs, as well as a homogeneous set of three 13C-labeled N-glycans. Analysis of the PCSs indicates that preferred glycan conformers have extensive contacts with the protein surface. Factors leading to this preference appear to include interactions between N-acetyl methyls of GlcNAc residues and hydrophobic surface pockets on the protein surface.

AssignSLP_GUI, a software tool exploiting AI for NMR resonance assignment of sparsely labeled proteins.

Not all proteins are amenable to uniform isotopic labeling with 13C and 15N, something needed for the widely used, and largely deductive, triple resonance assignment process. Among them are proteins expressed in mammalian cell culture where native glycosylation can be maintained, and proper formation of disulfide bonds facilitated. Uniform labeling in mammalian cells is prohibitively expensive, but sparse labeling with one or a few isotopically enriched amino acid types is an option for these proteins. However, assignment then relies on accessing the best match between a variety of measured NMR parameters and predictions based on 3D structure, often from X-ray crystallography. Finding this match is a challenging process that has benefitted from many computational tools, including trained neural nets for chemical shift prediction, genetic algorithms for searches through a myriad of assignment possibilities, and now AI-based prediction of high-quality structures for protein targets. AssignSLP_GUI, a new version of a software package for assignment of resonances from sparsely-labeled proteins, uses many of these tools. These tools and new additions to the package are highlighted in an application to a sparsely-labeled domain from a glycoprotein, CEACAM1.

Site-to-site cross-talk in OST-B glycosylation of hCEACAM1-IgV.

N-glycosylation is a common posttranslational modification of secreted proteins in eukaryotes. This modification targets asparagine residues within the consensus sequence, N-X-S/T. While this sequence is required for glycosylation, the initial transfer of a high-mannose glycan by oligosaccharyl transferases A or B (OST-A or OST-B) can lead to incomplete occupancy at a given site. Factors that determine the extent of transfer are not well understood, and understanding them may provide insight into the function of these important enzymes. Here, we use mass spectrometry (MS) to simultaneously measure relative occupancies for three N-glycosylation sites on the N-terminal IgV domain of the recombinant glycoprotein, hCEACAM1. We demonstrate that addition is primarily by the OST-B enzyme and propose a kinetic model of OST-B N-glycosylation. Fitting the kinetic model to the MS data yields distinct rates for glycan addition at most sites and suggests a largely stochastic initial order of glycan addition. The model also suggests that glycosylation at one site influences the efficiency of subsequent modifications at the other sites, and glycosylation at the central or N-terminal site leads to dead-end products that seldom lead to full glycosylation of all three sites. Only one path of progressive glycosylation, one initiated by glycosylation at the C-terminal site, can efficiently lead to full occupancy for all three sites. Thus, the hCEACAM1 domain provides an effective model system to study site-specific recognition of glycosylation sequons by OST-B and suggests that the order and efficiency of posttranslational glycosylation is influenced by steric cross-talk between adjoining acceptor sites.

NMR analysis suggests the terminal domains of Robo1 remain extended but are rigidified in the presence of heparan sulfate.

Human roundabout 1 (hRobo1) is an extracellular receptor glycoprotein that plays important roles in angiogenesis, organ development, and tumor progression. Interaction between hRobo1 and heparan sulfate (HS) has been shown to be essential for its biological activity. To better understand the effect of HS binding we engineered a lanthanide-binding peptide sequence (Loop) into the Ig2 domain of hRobo1. Native mass spectrometry was used to verify that loop introduction did not inhibit HS binding or conformational changes previously suggested by gas phase ion mobility measurements. NMR experiments measuring long-range pseudocontact shifts were then performed on 13C-methyl labeled hRobo1-Ig1-2-Loop in HS-bound and unbound forms. The magnitude of most PCSs for methyl groups in the Ig1 domain increase in the bound state confirming a change in the distribution of interdomain geometries. A grid search over Ig1 orientations to optimize the fit of data to a single conformer for both forms produced two similar structures, both of which differ from existing X-ray crystal structures and structures inferred from gas-phase ion mobility measurements. The structures and degree of fit suggest that the hRobo1-Ig1-2 structure changes slightly and becomes more rigid on HS binding. This may have implications for Robo-Slit signaling.

O-fucosylation stabilizes the TSR3 motif in thrombospondin-1 by interacting with nearby amino acids and protecting a disulfide bond.

Thrombospondin type-1 repeats (TSRs) are small protein motifs containing six conserved cysteines forming three disulfide bonds that can be modified with an O-linked fucose. Protein O-fucosyltransferase 2 (POFUT2) catalyzes the addition of O-fucose to TSRs containing the appropriate consensus sequence, and the O-fucose modification can be elongated to a Glucose-Fucose disaccharide with the addition of glucose by ß3-glucosyltransferase (B3GLCT). Elimination of Pofut2 in mice results in embryonic lethality in mice, highlighting the biological significance of O-fucose modification on TSRs. Knockout of POFUT2 in HEK293T cells has been shown to cause complete or partial loss of secretion of many proteins containing O-fucosylated TSRs. In addition, POFUT2 is localized to the endoplasmic reticulum (ER) and only modifies folded TSRs, stabilizing their structures. These observations suggest that POFUT2 is involved in an ER quality control mechanism for TSR folding and that B3GLCT also participates in quality control by providing additional stabilization to TSRs. However, the mechanisms by which addition of these sugars result in stabilization are poorly understood. Here, we conducted molecular dynamics (MD) simulations and provide crystallographic and NMR evidence that the Glucose-Fucose disaccharide interacts with specific amino acids in the TSR3 domain in thrombospondin-1 that are within proximity to the O-fucosylation modification site resulting in protection of a nearby disulfide bond. We also show that mutation of these amino acids reduces the stabilizing effect of the sugars in vitro. These data provide mechanistic details regarding the importance of O-fucosylation and how it participates in quality control mechanisms inside the ER.

Validated determination of NRG1 Ig-like domain structure by mass spectrometry coupled with computational modeling.

High resolution hydroxyl radical protein footprinting (HR-HRPF) is a mass spectrometry-based method that measures the solvent exposure of multiple amino acids in a single experiment, offering constraints for experimentally informed computational modeling. HR-HRPF-based modeling has previously been used to accurately model the structure of proteins of known structure, but the technique has never been used to determine the structure of a protein of unknown structure. Here, we present the use of HR-HRPF-based modeling to determine the structure of the Ig-like domain of NRG1, a protein with no close homolog of known structure. Independent determination of the protein structure by both HR-HRPF-based modeling and heteronuclear NMR was carried out, with results compared only after both processes were complete. The HR-HRPF-based model was highly similar to the lowest energy NMR model, with a backbone RMSD of 1.6 Å. To our knowledge, this is the first use of HR-HRPF-based modeling to determine a previously uncharacterized protein structure.

Human CEACAM1 N-domain dimerization is independent from glycan modifications.

Carcinoembryonic cellular adhesion molecules (CEACAMs) serve diverse roles in cell signaling, proliferation, and survival and are made up of one or several immunoglobulin (Ig)-like ectodomains glycosylated in vivo. The physiological oligomeric state and how it contributes to protein function are central to understanding CEACAMs. Two putative dimer conformations involving different CEACAM1 N-terminal Ig-like domain (CCM1) protein faces (ABED and GFCC'C″) were identified from crystal structures. GFCC'C″ was identified as the dominant CCM1 solution dimer, but ambiguity regarding the effect of glycosylation on dimer formation calls its physiological relevance into question. We present the first crystal structure of minimally glycosylated CCM1 in the GFCC'C″ dimer conformation and characterization in solution by continuous-wave and double electron-electron resonance electron paramagnetic resonance spectroscopy. Our results suggest the GFCC'C″ dimer is dominant in solution with different levels of glycosylation, and structural conservation and co-evolved residues support that the GFCC'C″ dimer is conserved across CEACAMs.

A perspective on the PDB's impact on the field of glycobiology.

Structures deposited in the Protein Data Bank (PDB) facilitate our understanding of many biological processes including those that fall under the general category of glycobiology. However, structure-based studies of how glycans affect protein structure, how they are synthesized, and how they regulate other biological processes remain challenging. Despite the abundant presence of glycans on proteins and the dense layers of glycans that surround most of our cells, structures containing glycans are underrepresented in the PDB. There are sound reasons for this, including difficulties in producing proteins with well-defined glycosylation and the tendency of mobile and heterogeneous glycans to inhibit crystallization. Nevertheless, the structures we do find in the PDB, even some of the earliest deposited structures, have had an impact on our understanding of function. I highlight a few examples in this review and point to some promises for the future. Promises include new structures from methodologies, such as cryo-EM, that are less affected by the presence of glycans and experiment-aided computational methods that build on existing structures to provide insight into the many ways glycans affect biological function.

Using molecular dynamics trajectories to predict nuclear spin relaxation behaviour in large spin systems.

Molecular dynamics (MD) trajectories provide useful insights into molecular structure and dynamics. However, questions persist about the quantitative accuracy of those insights. Experimental NMR spin relaxation rates can be used as tests, but only if relaxation superoperators can be efficiently computed from MD trajectories - no mean feat for the quantum Liouville space formalism where matrix dimensions quadruple with each added spin 1/2. Here we report a module for the Spinach software framework that computes Bloch-Redfield-Wangsness relaxation superoperators (including non-secular terms and cross-correlations) from MD trajectories. Predicted initial slopes of nuclear Overhauser effects for sucrose trajectories using advanced water models and a force field optimised for glycans are within 25% of experimental values.

Sparse isotope labeling for nuclear magnetic resonance (NMR) of glycoproteins using 13C-glucose.

Preparation of samples for nuclear magnetic resonance (NMR) characterization of larger proteins requires enrichment with less abundant, NMR-active, isotopes such as 13C and 15N. This is routine for proteins that can be expressed in bacterial culture where low-cost isotopically enriched metabolic substrates can be used. However, it can be expensive for glycosylated proteins expressed in mammalian culture where more costly isotopically enriched amino acids are usually used. We describe a simple, relatively inexpensive procedure in which standard commercial media is supplemented with 13C-enriched glucose to achieve labeling of all glycans plus all alanines of the N-terminal domain of the highly glycosylated protein, CEACAM1. We demonstrate an ability to detect partially occupied N-glycan sites, sites less susceptible to processing by an endoglycosidase, and some unexpected truncation of the amino acid sequence. The labeling of both the protein (through alanines) and the glycans in a single culture requiring no additional technical expertise past standard mammalian expression requirements is anticipated to have several applications, including structural and functional screening of the many glycosylated proteins important to human health.

Skp1 Dimerization Conceals Its F-Box Protein Binding Site.

Skp1 is an adapter that links F-box proteins to cullin-1 in the Skp1/cullin-1/F-box (SCF) protein family of E3 ubiquitin ligases that targets specific proteins for polyubiquitination and subsequent protein degradation. Skp1 from the amoebozoan Dictyostelium forms a stable homodimer in vitro with a Kd of 2.5 µM as determined by sedimentation velocity studies yet is monomeric in crystal complexes with F-box proteins. To investigate the molecular basis for the difference, we determined the solution NMR structure of a doubly truncated Skp1 homodimer (Skp1ΔΔ). The solution structure of the Skp1ΔΔ dimer reveals a 2-fold symmetry with an interface that buries ∼750 Å2 of predominantly hydrophobic surface. The dimer interface overlaps with subsite 1 of the F-box interaction area, explaining why only the Skp1 monomer binds F-box proteins (FBPs). To confirm the model, Rosetta was used to predict amino acid substitutions that might disrupt the dimer interface, and the F97E substitution was chosen to potentially minimize interference with F-box interactions. A nearly full-length version of Skp1 with this substitution (Skp1ΔF97E) behaved as a stable monomer at concentrations of ≤500 µM and actively bound a model FBP, mammalian Fbs1, which suggests that the dimeric state is not required for Skp1 to carry out a basic biochemical function. Finally, Skp1ΔF97E is expected to serve as a monomer model for high-resolution NMR studies previously hindered by dimerization.

Measurement of residual dipolar couplings in methyl groups via carbon detection.

Residual dipolar couplings (RDCs) provide both structural and dynamical information useful in the characterization of biological macromolecules. While most data come from the interaction of simple pairs of directly bonded spin-1/2 nuclei (1H-15N, 1H-13C, 1H-1H), it is possible to acquire data from interactions among the multiple spins of 13C-labeled methyl groups (1H3-13C). This is especially important because of the advantages that observation of 13C-labeled methyl groups offers in working with very large molecules. Here we consider some of the options for measurement of methyl RDCs in large and often fully protonated proteins and arrive at a pulse sequence that exploits both J-modulation and direct detection of 13C. Its utility is illustrated by application to a fully protonated two domain fragment from the mammalian glycoprotein, Robo1, 13C-methyl-labeled in all valines.

NMR Resonance Assignment Methodology: Characterizing Large Sparsely Labeled Glycoproteins.

Characterization of proteins using NMR methods begins with assignment of resonances to specific residues. This is usually accomplished using sequential connectivities between nuclear pairs in proteins uniformly labeled with NMR active isotopes. This becomes impractical for larger proteins, and especially for proteins that are best expressed in mammalian cells, including glycoproteins. Here an alternate protocol for the assignment of NMR resonances of sparsely labeled proteins, namely, the ones labeled with a single amino acid type, or a limited subset of types, isotopically enriched with 15N or 13C, is described. The protocol is based on comparison of data collected using extensions of simple two-dimensional NMR experiments (correlated chemical shifts, nuclear Overhauser effects, residual dipolar couplings) to predictions from molecular dynamics trajectories that begin with known protein structures. Optimal pairing of predicted and experimental values is facilitated by a software package that employs a genetic algorithm, ASSIGN_SLP_MD. The approach is applied to the 36-kDa luminal domain of the sialyltransferase, rST6Gal1, in which all phenylalanines are labeled with 15N, and the results are validated by elimination of resonances via single-point mutations of selected phenylalanines to tyrosines. Assignment allows the use of previously published paramagnetic relaxation enhancements to evaluate placement of a substrate analog in the active site of this protein. The protocol will open the way to structural characterization of the many glycosylated and other proteins that are best expressed in mammalian cells.

Paramagnetic Tag for Glycosylation Sites in Glycoproteins: Structural Constraints on Heparan Sulfate Binding to Robo1.

An enzyme- and click chemistry-mediated methodology for the site-specific nitroxide spin labeling of glycoproteins has been developed and applied. The procedure relies on the presence of single N-glycosylation sites that are present natively in proteins or that can be engineered into glycoproteins by mutational elimination of all but one glycosylation site. Recombinantly expressing glycoproteins in HEK293S (GnT1-) cells results in N-glycans with high-mannose structures that can be processed to leave a single GlcNAc residue. This can in turn be modified by enzymatic addition of a GalNAz residue that is subject to reaction with an alkyne-carrying TEMPO moiety using copper(I)-catalyzed click chemistry. To illustrate the procedure, we have made an application to a two-domain construct of Robo1, a protein that carries a single N-glycosylation site in its N-terminal domains. The construct has also been labeled with 15N at amide nitrogens of lysine residues to provide a set of sites that are used to derive an effective location of the paramagnetic nitroxide moiety of the TEMPO group. This, in turn, allowed measurements of paramagnetic perturbations to the spectra of a new high affinity heparan sulfate ligand. Calculation of distance constraints from these data facilitated determination of an atomic level model for the docked complex.

Structural Characterization of a Heparan Sulfate Pentamer Interacting with LAR-Ig1-2.

Leukocyte common antigen-related (LAR) protein is one of the type IIa receptor protein tyrosine phosphatases (RPTPs) that are important for signal transduction in biological processes, including axon growth and regeneration. Glycosaminoglycan chains, including heparan sulfate (HS) and chondroitin sulfate (CS), act as ligands that regulate LAR signaling. Here, we report the structural characterization of the first two immunoglobulin domains (Ig1-2) of LAR interacting with an HS pentasaccharide (GlcNS6S-GlcA-GlcNS3,6S-IdoA2S-GlcNS6S-OME, fondaparinux) using multiple solution-based NMR methods. In the course of the study, we extended an assignment strategy useful for sparsely labeled proteins expressed in mammalian cell culture supplemented with a single type of isotopically enriched amino acid ([15N]-Lys in this case) by including paramagnetic perturbations to NMR resonances. The folded two-domain structure for LAR-Ig1-2 seen in previous crystal structures has been validated in solution using residual dipolar coupling data, and a combination of chemical shift perturbation on titration of LAR-Ig1-2 with fondaparinux, saturation transfer difference (STD) spectra, and transferred nuclear Overhauser effects (trNOEs) have been employed in the docking program HADDOCK to generate models for the LAR-fondaparinux complex. These models are further analyzed by postprocessing energetic analysis to identify key binding interactions. In addition to providing insight into the ligand interaction mechanisms of type IIa RPTPs and the origin of opposing effects of CS and HS ligands, these results may assist in future design of therapeutic compounds for nervous system repair.