Combining Rosetta Sequence Design with Protein Language Model Predictions Using Evolutionary Scale Modeling (ESM) as Restraint.

Computational protein sequence design has the ambitious goal of modifying existing or creating new proteins; however, designing stable and functional proteins is challenging without predictability of protein dynamics and allostery. Informing protein design methods with evolutionary information limits the mutational space to more native-like sequences and results in increased stability while maintaining functions. Recently, language models, trained on millions of protein sequences, have shown impressive performance in predicting the effects of mutations. Assessing Rosetta-designed sequences with a language model showed scores that were worse than those of their original sequence. To inform Rosetta design protocols with language model predictions, we added a new metric to restrain the energy function during design using the Evolutionary Scale Modeling (ESM) model. The resulting sequences have better language model scores and similar sequence recovery, with only a minor decrease in the fitness as assessed by Rosetta energy. In conclusion, our work combines the strength of recent machine learning approaches with the Rosetta protein design toolbox.

Backbone chemical shift and secondary structure assignments for mouse siderocalin.

The lipocalin protein family is a structurally conserved group of proteins with a variety of biological functions defined by their ability to bind small molecule ligands and interact with partner proteins. One member of this family is siderocalin, a protein found in mammals. Its role is discussed in inflammatory processes, iron trafficking, protection against bacterial infections and oxidative stress, cell migration, induction of apoptosis, and cancer. Though it seems to be involved in numerous essential pathways, the exact mechanisms are often not fully understood. The NMR backbone assignments for the human siderocalin and its rat ortholog have been published before. In this work we describe the backbone NMR assignments of siderocalin for another important model organism, the mouse - data that might become important for structure-based drug discovery. Secondary structure elements were predicted based on the assigned backbone chemical shifts using TALOS-N and CSI 3.0, revealing a high content of beta strands and one prominent alpha helical region. Our findings correlate well with the known crystal structure and the overall conserved fold of the lipocalin family.

Combining machine learning with structure-based protein design to predict and engineer post-translational modifications of proteins.

Post-translational modifications (PTMs) of proteins play a vital role in their function and stability. These modifications influence protein folding, signaling, protein-protein interactions, enzyme activity, binding affinity, aggregation, degradation, and much more. To date, over 400 types of PTMs have been described, representing chemical diversity well beyond the genetically encoded amino acids. Such modifications pose a challenge to the successful design of proteins, but also represent a major opportunity to diversify the protein engineering toolbox. To this end, we first trained artificial neural networks (ANNs) to predict eighteen of the most abundant PTMs, including protein glycosylation, phosphorylation, methylation, and deamidation. In a second step, these models were implemented inside the computational protein modeling suite Rosetta, which allows flexible combination with existing protocols to model the modified sites and understand their impact on protein stability as well as function. Lastly, we developed a new design protocol that either maximizes or minimizes the predicted probability of a particular site being modified. We find that this combination of ANN prediction and structure-based design can enable the modification of existing, as well as the introduction of novel, PTMs. The potential applications of our work include, but are not limited to, glycan masking of epitopes, strengthening protein-protein interactions through phosphorylation, as well as protecting proteins from deamidation liabilities. These applications are especially important for the design of new protein therapeutics where PTMs can drastically change the therapeutic properties of a protein. Our work adds novel tools to Rosetta's protein engineering toolbox that allow for the rational design of PTMs.

A tetravalent bispecific antibody outperforms the combination of its parental antibodies and neutralizes diverse SARS-CoV-2 variants.

The devastating impact of COVID-19 on global health shows the need to increase our pandemic preparedness. Recombinant therapeutic antibodies were successfully used to treat and protect at-risk patients from COVID-19. However, the currently circulating Omicron subvariants of SARS-CoV-2 are largely resistant to therapeutic antibodies, and novel approaches to generate broadly neutralizing antibodies are urgently needed. Here, we describe a tetravalent bispecific antibody, A7A9 TVB, which actively neutralized many SARS-CoV-2 variants of concern, including early Omicron subvariants. Interestingly, A7A9 TVB neutralized more variants at lower concentration as compared to the combination of its parental monoclonal antibodies, A7K and A9L. A7A9 also reduced the viral load of authentic Omicron BA.1 virus in infected pseudostratified primary human nasal epithelial cells. Overall, A7A9 displayed the characteristics of a potent broadly neutralizing antibody, which may be suitable for prophylactic and therapeutic applications in the clinics, thus highlighting the usefulness of an effective antibody-designing approach.

Modifying immune responses to adeno-associated virus vectors by capsid engineering.

De novo immune responses are considered major challenges in gene therapy. With the aim to lower innate immune responses directly in cells targeted by adeno-associated virus (AAV) vectors, we equipped the vector capsid with a peptide known to interfere with Toll-like receptor signaling. Specifically, we genetically inserted in each of the 60 AAV2 capsid subunits the myeloid differentiation primary response 88 (MyD88)-derived peptide RDVLPGT, known to block MyD88 dimerization. Inserting the peptide neither interfered with capsid assembly nor with vector production yield. The novel capsid variant, AAV2.MB453, showed superior transduction efficiency compared to AAV2 in human monocyte-derived dendritic cells and in primary human hepatocyte cultures. In line with our hypothesis, AAV2.MB453 and AAV2 differed regarding innate immune response activation in primary human cells, particularly for type I interferons. Furthermore, mice treated with AAV2.MB453 showed significantly reduced CD8+ T cell responses against the transgene product for different administration routes and against the capsid following intramuscular administration. Moreover, humoral responses against the capsid were mitigated as indicated by delayed IgG2a antibody formation and an increased NAb50. To conclude, insertion of the MyD88-derived peptide into the AAV2 capsid improved early steps of host-vector interaction and reduced innate and adaptive immune responses.

Interplay of thermodynamics and evolution within the ternary ligand-GPCR-G protein complex.

Recent studies on G protein-coupled receptors (GPCRs) dynamics report that GPCRs adopt a wide range of conformations that coexist in equilibrium, with the apo state of a GPCR having a high entropy. The formation of a ligand-GPCR-transducer complex comes with a reduction of conformational space and therefore with an entropic cost. We hypothesize that the availability of binding partners, their binding affinity and the rigidity of the respective binding sites are reflected in a distinct degree of sequence conservation to balance the energetic cost of intra- and extracellular binding events. Here, we outline the current findings in delineating the conformational space and include sequential conservation of many-to-many ligand-receptor systems to discuss the entropic cost that comes with GPCR signal transduction.

Computational modeling and prediction of deletion mutants.

In-frame deletion mutations can result in disease. The impact of these mutations on protein structure and subsequent functional changes remain understudied, partially due to the lack of comprehensive datasets including a structural readout. In addition, the recent breakthrough in structure prediction through deep learning demands an update of computational deletion mutation prediction. In this study, we deleted individually every residue of a small α-helical sterile alpha motif domain and investigated the structural and thermodynamic changes using 2D NMR spectroscopy and differential scanning fluorimetry. Then, we tested computational protocols to model and classify observed deletion mutants. We show a method using AlphaFold2 followed by RosettaRelax performs the best overall. In addition, a metric containing pLDDT values and Rosetta ΔΔG is most reliable in classifying tolerated deletion mutations. We further test this method on other datasets and show they hold for proteins known to harbor disease-causing deletion mutations.

Design, synthesis, and structure-activity relationships of diindolylmethane derivatives as cannabinoid CB2 receptor agonists.

3,3'-Diindolylmethane (DIM), a natural product-derived compound formed upon ingestion of cruciferous vegetables, was recently described to act as a partial agonist of the anti-inflammatory cannabinoid (CB) receptor subtype CB2 . In the present study, we synthesized and evaluated a series of DIM derivatives and determined their affinities for human CB receptor subtypes in radioligand binding studies. Potent compounds were additionally evaluated in functional cAMP accumulation and ß-arrestin recruitment assays. Small substituents in the 4-position of both indole rings of DIM were beneficial for high CB2 receptor affinity and efficacy. Di-(4-cyano-1H-indol-3-yl)methane (46, PSB-19837, EC50 : cAMP, 0.0144 µM, 95% efficacy compared to the full standard agonist CP55,940; ß-arrestin, 0.0149 µM, 67% efficacy) was the most potent CB2 receptor agonist of the present series. Di-(4-bromo-1H-indol-3-yl)methane (44, PSB-19571) showed higher potency in ß-arrestin (EC50 0.0450 µM, 61% efficacy) than in cAMP accumulation assays (EC50 0.509 µM, 85% efficacy) while 3-((1H-indol-3-yl)methyl)-4-methyl-1H-indole (149, PSB-18691) displayed a 19-fold bias for the G protein pathway (EC50 : cAMP, 0.0652 µM; ß-arrestin, 1.08 µM). DIM and its analogs act as allosteric CB2 receptor agonists. These potent CB2 receptor agonists have potential as novel drugs for the treatment of inflammatory diseases.

Adeno-associated virus serotype 2 capsid variants for improved liver-directed gene therapy.

BACKGROUND AND AIMS: Current liver-directed gene therapies look for adeno-associated virus (AAV) vectors with improved efficacy. With this background, capsid engineering is explored. Whereas shuffled capsid library screenings have resulted in potent liver targeting variants with one first vector in human clinical trials, modifying natural serotypes by peptide insertion has so far been less successful. Here, we now report on two capsid variants, MLIV.K and MLIV.A, both derived from a high-throughput in vivo AAV peptide display selection screen in mice. APPROACH AND RESULTS: The variants transduce primary murine and human hepatocytes at comparable efficiencies, a valuable feature in clinical development, and show significantly improved liver transduction efficacy, thereby allowing a dose reduction, and outperform parental AAV2 and AAV8 in targeting human hepatocytes in humanized mice. The natural heparan sulfate proteoglycan binding ability is markedly reduced, a feature that correlates with improved hepatocyte transduction. A further property that might contribute to the improved transduction efficacy is the lower capsid melting temperature. Peptide insertion also caused a moderate change in sensitivity to human sera containing anti-AAV2 neutralizing antibodies, revealing the impact of epitopes located at the basis of the AAV capsid protrusions. CONCLUSIONS: In conclusion, MLIV.K and MLIV.A are AAV peptide display variants selected in immunocompetent mice with improved hepatocyte tropism and transduction efficiency. Because these features are maintained across species, MLIV variants provide remarkable potential for translation of therapeutic approaches from mice to men.

Computational Structure Prediction for Antibody-Antigen Complexes From Hydrogen-Deuterium Exchange Mass Spectrometry: Challenges and Outlook.

Although computational structure prediction has had great successes in recent years, it regularly fails to predict the interactions of large protein complexes with residue-level accuracy, or even the correct orientation of the protein partners. The performance of computational docking can be notably enhanced by incorporating experimental data from structural biology techniques. A rapid method to probe protein-protein interactions is hydrogen-deuterium exchange mass spectrometry (HDX-MS). HDX-MS has been increasingly used for epitope-mapping of antibodies (Abs) to their respective antigens (Ags) in the past few years. In this paper, we review the current state of HDX-MS in studying protein interactions, specifically Ab-Ag interactions, and how it has been used to inform computational structure prediction calculations. Particularly, we address the limitations of HDX-MS in epitope mapping and techniques and protocols applied to overcome these barriers. Furthermore, we explore computational methods that leverage HDX-MS to aid structure prediction, including the computational simulation of HDX-MS data and the combination of HDX-MS and protein docking. We point out challenges in interpreting and incorporating HDX-MS data into Ab-Ag complex docking and highlight the opportunities they provide to build towards a more optimized hybrid method, allowing for more reliable, high throughput epitope identification.

Epitope-focused immunogen design based on the ebolavirus glycoprotein HR2-MPER region.

The three human pathogenic ebolaviruses: Zaire (EBOV), Bundibugyo (BDBV), and Sudan (SUDV) virus, cause severe disease with high fatality rates. Epitopes of ebolavirus glycoprotein (GP) recognized by antibodies with binding breadth for all three ebolaviruses are of major interest for rational vaccine design. In particular, the heptad repeat 2 -membrane-proximal external region (HR2-MPER) epitope is relatively conserved between EBOV, BDBV, and SUDV GP and targeted by human broadly-neutralizing antibodies. To study whether this epitope can serve as an immunogen for the elicitation of broadly-reactive antibody responses, protein design in Rosetta was employed to transplant the HR2-MPER epitope identified from a co-crystal structure with the known broadly-reactive monoclonal antibody (mAb) BDBV223 onto smaller scaffold proteins. From computational analysis, selected immunogen designs were produced as recombinant proteins and functionally validated, leading to the identification of a sterile alpha motif (SAM) domain displaying the BDBV-HR2-MPER epitope near its C terminus as a promising candidate. The immunogen was fused to one component of a self-assembling, two-component nanoparticle and tested for immunogenicity in rabbits. Robust titers of cross-reactive serum antibodies to BDBV and EBOV GPs and moderate titers to SUDV GP were induced following immunization. To confirm the structural composition of the immunogens, solution NMR studies were conducted and revealed structural flexibility in the C-terminal residues of the epitope. Overall, our study represents the first report on an epitope-focused immunogen design based on the structurally challenging BDBV-HR2-MPER epitope.

Structural Comparative Modeling of Multi-Domain F508del CFTR.

Cystic fibrosis (CF) is a rare genetic disease caused by mutations in the cystic fibrosis transmembrane conductance regulator (CFTR), an epithelial anion channel expressed in several vital organs. Absence of functional CFTR results in imbalanced osmotic equilibrium and subsequent mucus build up in the lungs-which increases the risk of infection and eventually causes death. CFTR is an ATP-binding cassette (ABC) transporter family protein composed of two transmembrane domains (TMDs), two nucleotide binding domains (NBDs), and an unstructured regulatory domain. The most prevalent patient mutation is the deletion of F508 (F508del), making F508del CFTR the primary target for current FDA approved CF therapies. However, no experimental multi-domain F508del CFTR structure has been determined and few studies have modeled F508del using multi-domain WT CFTR structures. Here, we used cryo-EM density data and Rosetta comparative modeling (RosettaCM) to compare a F508del model with published experimental data on CFTR NBD1 thermodynamics. We then apply this modeling method to generate multi-domain WT and F508del CFTR structural models. These models demonstrate the destabilizing effects of F508del on NBD1 and the NBD1/TMD interface in both the inactive and active conformation of CFTR. Furthermore, we modeled F508del/R1070W and F508del bound to the CFTR corrector VX-809. Our models reveal the stabilizing effects of VX-809 on multi-domain models of F508del CFTR and pave the way for rational design of additional drugs that target F508del CFTR for treatment of CF.

Therapeutic alphavirus cross-reactive E1 human antibodies inhibit viral egress.

Alphaviruses cause severe arthritogenic or encephalitic disease. The E1 structural glycoprotein is highly conserved in these viruses and mediates viral fusion with host cells. However, the role of antibody responses to the E1 protein in immunity is poorly understood. We isolated E1-specific human monoclonal antibodies (mAbs) with diverse patterns of recognition for alphaviruses (ranging from Eastern equine encephalitis virus [EEEV]-specific to alphavirus cross-reactive) from survivors of natural EEEV infection. Antibody binding patterns and epitope mapping experiments identified differences in E1 reactivity based on exposure of epitopes on the glycoprotein through pH-dependent mechanisms or presentation on the cell surface prior to virus egress. Therapeutic efficacy in vivo of these mAbs corresponded with potency of virus egress inhibition in vitro and did not require Fc-mediated effector functions for treatment against subcutaneous EEEV challenge. These studies reveal the molecular basis for broad and protective antibody responses to alphavirus E1 proteins.

Fintiamin: A diketopiperazine from the marine sponge-derived fungus Eurotium sp.

The fungus Eurotium sp., derived from the marine sponge Ircinia variabilis, was found to produce a diketopiperazine-indole alkaloid that we named fintiamin (1). Structural elucidation of 1 was achieved by extensive spectroscopic analysis including nuclear magnetic resonance spectroscopy and mass spectrometry. Compound 1 is a lipophilic terpenoid-dipeptide hybrid molecule that shows affinity for the cannabinoid CB1 receptor at low micromolar concentrations. Docking studies based on previous X-ray structures provide a plausible binding pose for compound 1 in the orthosteric binding site of the CB1 receptor.

Ligand-binding and -scavenging of the chemerin receptor GPR1.

Tight regulation of cytokines is essential for the initiation and resolution of inflammation. Chemerin, a mediator of innate immunity, mainly acts on chemokine-like receptor 1 (CMKLR1) to induce the migration of macrophages and dendritic cells. The role of the second chemerin receptor, G protein-coupled receptor 1 (GPR1), is still unclear. Here we demonstrate that GPR1 shows ligand-induced arrestin3 recruitment and internalization. The chemerin C-terminus triggers this activation by folding into a loop structure, binding to aromatic residues in the extracellular loops of GPR1. While this overall binding mode is shared between GPR1 and CMKLR1, differences in their respective extracellular loop 2 allowed for the design of the first GPR1-selective peptide. However, our results suggest that ligand-induced arrestin recruitment is not the only mode of action of GPR1. This receptor also displays constitutive internalization, which allows GPR1 to internalize inactive peptides efficiently by an activation-independent pathway. Our results demonstrate that GPR1 takes a dual role in regulating chemerin activity: as a signaling receptor for arrestin-based signaling on one hand, and as a scavenging receptor with broader ligand specificity on the other.

RosettaCM for antibodies with very long HCDR3s and low template availability.

Antibody-antigen co-crystal structures are a valuable resource for the fundamental understanding of antibody-mediated immunity. Determination of structures with antibodies in complex with their antigens, however, is a laborious task without guarantee of success. Therefore, homology modeling of antibodies and docking to their respective antigens has become a very important technique to drive antibody and vaccine design. The quality of the antibody modeling process is critical for the success of these endeavors. Here, we compare different computational protocols for predicting antibody structure from sequence in the biomolecular modeling software Rosetta-all of which use multiple existing antibody structures to guide modeling. Specifically, we compare protocols developed solely to predict antibody structure (RosettaAntibody, AbPredict) with a universal homology modeling protocol (RosettaCM). Following recent advances in homology modeling with multiple templates simultaneously, we propose that the use of multiple templates over the same antibody regions may improve modeling performance. To evaluate whether multi-template comparative modeling with RosettaCM can improve the modeling accuracy of antibodies over existing methods, this study compares the performance of the three modeling algorithms when modeling human antibodies taken from antibody-antigen co-crystal structures. In these benchmarking experiments, RosettaCM outperformed other methods when modeling antibodies with long HCDR3s and few available templates.

Modeling Immunity with Rosetta: Methods for Antibody and Antigen Design.

Structure-based antibody and antigen design has advanced greatly in recent years, due not only to the increasing availability of experimentally determined structures but also to improved computational methods for both prediction and design. Constant improvements in performance within the Rosetta software suite for biomolecular modeling have given rise to a greater breadth of structure prediction, including docking and design application cases for antibody and antigen modeling. Here, we present an overview of current protocols for antibody and antigen modeling using Rosetta and exemplify those by detailed tutorials originally developed for a Rosetta workshop at Vanderbilt University. These tutorials cover antibody structure prediction, docking, and design and antigen design strategies, including the addition of glycans in Rosetta. We expect that these materials will allow novice users to apply Rosetta in their own projects for modeling antibodies and antigens.

Cyclic Analogues of the Chemerin C-Terminus Mimic a Loop Conformation Essential for Activating the Chemokine-like Receptor 1.

The chemokine-like receptor 1 (CMKLR1) is a promising target for treating autoinflammatory diseases, cancer, and reproductive disorders. However, the interaction between CMKLR1 and its protein-ligand chemerin remains uncharacterized, and no drugs targeting this interaction have passed clinical trials. Here, we identify the binding mode of chemerin-9, the C-terminus of chemerin, at the receptor by combining complementary mutagenesis with structure-based modeling. Incorporating our experimental data, we present a detailed model of this binding site, including experimentally confirmed pairwise interactions for the most critical ligand residues: Chemerin-9 residue F8 binds to a hydrophobic pocket in CMKLR1 formed by the extracellular loop (ECL) 2, while F6 interacts with Y2.68, suggesting a turn-like structure. On the basis of this model, we created the first cyclic peptide with nanomolar activity, confirming the overall binding conformation. This constrained agonist mimics the loop conformation adopted by the natural ligand and can serve as a lead compound for future drug design.

Predicting susceptibility to SARS-CoV-2 infection based on structural differences in ACE2 across species.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the cause of the global pandemic of coronavirus disease-2019 (COVID-19). SARS-CoV-2 is a zoonotic disease, but little is known about variations in species susceptibility that could identify potential reservoir species, animal models, and the risk to pets, wildlife, and livestock. Certain species, such as domestic cats and tigers, are susceptible to SARS-CoV-2 infection, while other species such as mice and chickens are not. Most animal species, including those in close contact with humans, have unknown susceptibility. Hence, methods to predict the infection risk of animal species are urgently needed. SARS-CoV-2 spike protein binding to angiotensin-converting enzyme 2 (ACE2) is critical for viral cell entry and infection. Here we integrate species differences in susceptibility with multiple in-depth structural analyses to identify key ACE2 amino acid positions including 30, 83, 90, 322, and 354 that distinguish susceptible from resistant species. Using differences in these residues across species, we developed a susceptibility score that predicts an elevated risk of SARS-CoV-2 infection for multiple species including horses and camels. We also demonstrate that SARS-CoV-2 is nearly optimal for binding ACE2 of humans compared to other animals, which may underlie the highly contagious transmissibility of this virus among humans. Taken together, our findings define potential ACE2 and SARS-CoV-2 residues for therapeutic targeting and identification of animal species on which to focus research and protection measures for environmental and public health.