Deep mutational learning predicts ACE2 binding and antibody escape to combinatorial mutations in the SARS-CoV-2 receptor-binding domain

The biomedical applications of artificial intelligence: an overview of decades of research

A significant area of computer science called artificial intelligence (AI) is successfully applied to the analysis of intricate biological data and the extraction of substantial associations from datasets for a variety of biomedical uses. AI has attracted significant interest in biomedical research due to its features: (i) better patient care through early diagnosis and detection; (ii) enhanced workflow; (iii) lowering medical errors; (v) lowering medical costs; (vi) reducing morbidity and mortality; (vii) enhancing performance; (viii) enhancing precision; and (ix) time efficiency. Quantitative metrics are crucial for evaluating AI implementations, providing insights, enabling informed decisions, and measuring the impact of AI-driven initiatives, thereby enhancing transparency, accountability, and overall impact. The implementation of AI in biomedical fields faces challenges such as ethical and privacy concerns, lack of awareness, technology unreliability, and professional liability. A brief discussion is given of the AI techniques, which include Virtual screening (VS), DL, ML, Hidden Markov models (HMMs), Neural networks (NNs), Generative models (GMs), Molecular dynamics (MD), and Structure-activity relationship (SAR) models. The study explores the application of AI in biomedical fields, highlighting its enhanced predictive accuracy, treatment efficacy, diagnostic efficiency, faster decision-making, personalised treatment strategies, and precise medical interventions.

LL-37 Inhibits TMPRSS2-Mediated S2' Site Cleavage and SARS-CoV-2 Infection but Not Omicron Variants

Continual evolution of SARS-CoV-2 spike drives the emergence of Omicron variants that show increased spreading and immune evasion. Understanding how the variants orientate themselves towards host immune defence is crucial for controlling future pandemics. Herein, we demonstrate that human cathelicidin LL-37, a crucial component of innate immunity, predominantly binds to the S2 subunit of SARS-CoV-2 spike protein, occupying sites where TMPRSS2 typically binds. This binding impedes TMPRSS2-mediated priming at site S2' and subsequent membrane fusion processes. The mutation N764K within S2 subunit of Omicron variants reduces affinity for LL-37 significantly, thereby diminishing binding capacity and inhibitory effects on membrane fusion. Moreover, the early humoral immune response enhanced by LL-37 is observed in mice against SARS-CoV-2 spike but not Omicron BA.4/5 spike. These findings reveal the mechanism underlying interactions amongst LL-37, TMPRSS2 and SARS-CoV-2 and VOCs, and highlight the distinct mutation for Omicron variants to evade the fusion activity inhibition by host innate immunity.

Deep mutational learning for the selection of therapeutic antibodies resistant to the evolution of Omicron variants of SARS-CoV-2

Most antibodies for treating COVID-19 rely on binding the receptor-binding domain (RBD) of SARS-CoV-2 (severe acute respiratory syndrome coronavirus 2). However, Omicron and its sub-lineages, as well as other heavily mutated variants, have rendered many neutralizing antibodies ineffective. Here we show that antibodies with enhanced resistance to the evolution of SARS-CoV-2 can be identified via deep mutational learning. We constructed a library of full-length RBDs of Omicron BA.1 with high mutational distance and screened it for binding to the angiotensin-converting-enzyme-2 receptor and to neutralizing antibodies. After deep-sequencing the library, we used the data to train ensemble deep-learning models for the prediction of the binding and escape of a panel of eight therapeutic antibody candidates targeting a diverse range of RBD epitopes. By using in silico evolution to assess antibody breadth via the prediction of the binding and escape of the antibodies to millions of Omicron sequences, we found combinations of two antibodies with enhanced and complementary resistance to viral evolution. Deep learning may enable the development of therapeutic antibodies that remain effective against future SARS-CoV-2 variants.

Biophysics of SARS-CoV-2 spike protein's receptor-binding domain interaction with ACE2 and neutralizing antibodies: from computation to functional insights

The spike protein encoded by the SARS-CoV-2 has become one of the most studied macromolecules in recent years due to its central role in COVID-19 pathogenesis. The spike protein's receptor-binding domain (RBD) directly interacts with the host-encoded receptor protein, ACE2. This review critically examines computational insights into RBD's interaction with ACE2 and with therapeutic antibodies designed to interfere with this interaction. We begin by summarizing insights from early computational studies on pre-pandemic SARS-CoV-1 RBD interactions and how these early studies shaped the understanding of SARS-CoV-2. Next, we highlight key theoretical contributions that revealed the molecular mechanisms behind the binding affinity of SARS-CoV-2 RBD against ACE2, and the structural changes that have enhanced the infectivity of emerging variants. Special attention is given to the "RBD charge rule", a predictive framework for determining variant infectivity based on the electrostatic properties of the RBD. Towards applying the computational insights to therapy, we discuss a multiscale computational protocol for optimizing monoclonal antibodies to improve binding affinity across multiple spike protein variants, including representatives from the Omicron family. Finally, we explore how these insights can inform the development of future vaccines and therapeutic interventions for combating future coronavirus diseases.

Computational electrostatic engineering of nanobodies for enhanced SARS-CoV-2 receptor binding domain recognition

This study presents a novel computational approach for engineering nanobodies (Nbs) for improved interaction with receptor-binding domain (RBD) of the SARS-CoV-2 spike protein. Using Protein Structure Reliability reports, RBD (7VYR_R) was selected and refined for subsequent Nb-RBD interactions. By leveraging electrostatic complementarity (EC) analysis, we engineered and characterized five Electrostatically Complementary Nbs (ECSb1-ECSb5) based on the CeVICA library's SR6c3 Nb. Through targeted modifications in the complementarity-determining regions (CDR) and framework regions (FR), we optimized electrostatic interactions to improve binding affinity and specificity. The engineered Nbs (ECSb3, ECSb4, and ECSb5) demonstrated high binding specificity for AS3, CA1, and CA2 epitopes. Interestingly, ECSb1 and ECSb2 selectively engaged with AS3 and CA1 instead of AS1 and AS2, respectively, due to a preference for residues that conferred superior binding complementarities. Furthermore, ECSbs significantly outperformed SR6c3 Nb in MM/GBSA results, notably, ECSb4 and ECSb3 exhibited superior binding free energies of -182.58 kcal.mol-1 and -119.07 kcal.mol-1, respectively, compared to SR6c3 (-105.50 kcal.mol-1). ECSbs exhibited significantly higher thermostability (100.4-148.3 kcal·mol⁻1) compared to SR6c3 (62.6 kcal·mol⁻1). Similarly, enhanced electrostatic complementarity was also observed for ECSb4-RBD and ECSb3-RBD (0.305 and 0.390, respectively) relative to SR6c3-RBD (0.233). Surface analyses confirmed optimized electrostatic patches and reduced aggregation propensity in the engineered Nb. This integrated EC and structural engineering approach successfully developed engineered Nbs with enhanced binding specificity, increased thermostability, and reduced aggregation, laying the groundwork for novel therapeutic applications targeting the SARS-CoV-2 spike protein.

Rapid restoration of potent neutralization activity against the latest Omicron variant JN.1 via AI rational design and antibody engineering

The rapid evolution of the viral genome has led to the continual generation of new variants of SARS-CoV-2. Developing antibody drugs with broad-spectrum and high efficiency is a long-term task. It is promising but challenging to develop therapeutic neutralizing antibodies (nAbs) through in vitro evolution based on antigen-antibody binding interactions. From an early B cell antibody repertoire, we isolated antibody 8G3 that retains its nonregressive neutralizing activity against Omicron BA.1 and various other strains in vitro. 8G3 protected ACE2 transgenic mice from BA.1 and WA1/2020 virus infection without adverse clinical manifestations and completely cleared viral load in the lungs. Similar to most IGHV3-53 antibodies, the binding sites of 8G3 and ACE2 largely overlap, enabling competition with ACE2 for binding to RBD. By comprehensively considering the binding free energy changes of the antigen-antibody complexes, the biological environment of their interactions, and the evolutionary direction of the antibodies, we were able to select 50 mutants. Among them, 11 were validated by experiments showing better neutralizing activities. Further, a combination of four mutations were identified in 8G3 that increased its neutralization potency against JN.1, the latest Omicron mutant, by approximately 1,500-fold, and one of the mutations led to an improvement in activity against multiple variants to a certain extent. Together, we established a procedure of rapid selection of neutralizing antibodies with potent SARS-CoV-2 neutralization activity. Our results provide a reference for engineering neutralizing antibodies against future SARS-CoV-2 variants and even other pandemic viruses.

Unveiling the ghost: machine learning's impact on the landscape of virology

The complexity and speed of evolution in viruses with RNA genomes makes predictive identification of variants with epidemic or pandemic potential challenging. In recent years, machine learning has become an increasingly capable technology for addressing this challenge, as advances in methods and computational power have dramatically improved the performance of models and led to their widespread adoption across industries and disciplines. Nascent applications of machine learning technology to virus research have now expanded, providing new tools for handling large-scale datasets and leading to a reshaping of existing workflows for phenotype prediction, phylogenetic analysis, drug discovery and more. This review explores how machine learning has been applied to and has impacted the study of viruses, before addressing the strengths and limitations of its techniques and finally highlighting the next steps that are needed for the technology to reach its full potential in this challenging and ever-relevant research area.

A predictive language model for SARS-CoV-2 evolution

Modeling and predicting mutations are critical for COVID-19 and similar pandemic preparedness. However, existing predictive models have yet to integrate the regularity and randomness of viral mutations with minimal data requirements. Here, we develop a non-demanding language model utilizing both regularity and randomness to predict candidate SARS-CoV-2 variants and mutations that might prevail. We constructed the "grammatical frameworks" of the available S1 sequences for dimension reduction and semantic representation to grasp the model's latent regularity. The mutational profile, defined as the frequency of mutations, was introduced into the model to incorporate randomness. With this model, we successfully identified and validated several variants with significantly enhanced viral infectivity and immune evasion by wet-lab experiments. By inputting the sequence data from three different time points, we detected circulating strains or vital mutations for XBB.1.16, EG.5, JN.1, and BA.2.86 strains before their emergence. In addition, our results also predicted the previously unknown variants that may cause future epidemics. With both the data validation and experiment evidence, our study represents a fast-responding, concise, and promising language model, potentially generalizable to other viral pathogens, to forecast viral evolution and detect crucial hot mutation spots, thus warning the emerging variants that might raise public health concern.

Predicting adaptive immune receptor specificities by machine learning is a data generation problem

Determining the specificity of adaptive immune receptors-B cell receptors (BCRs), their secreted form antibodies, and T cell receptors (TCRs)-is critical for understanding immune responses and advancing immunotherapy and drug discovery. Immune receptors exhibit extensive diversity in their variable domains, enabling them to interact with a plethora of antigens. Despite the significant progress made by AI tools such as AlphaFold in predicting protein structures, challenges remain in accurately modeling the structure and specificity of immune receptors, primarily due to the limited availability of high-quality crystal structures and the complexity of immune receptor-antigen interactions. In this perspective, we highlight recent advancements in sequence-based and structure-based data generation for immune receptors, which are crucial for training machine learning models that predict receptor specificity. We discuss the current bottlenecks and potential future directions in generating and utilizing high-dimensional datasets for predicting and designing the specificity of antibodies and TCRs.

A potent pan-sarbecovirus neutralizing antibody resilient to epitope diversification

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) evolution has resulted in viral escape from clinically authorized monoclonal antibodies (mAbs), creating a need for mAbs that are resilient to epitope diversification. Broadly neutralizing coronavirus mAbs that are sufficiently potent for clinical development and retain activity despite viral evolution remain elusive. We identified a human mAb, designated VIR-7229, which targets the viral receptor-binding motif (RBM) with unprecedented cross-reactivity to all sarbecovirus clades, including non-ACE2-utilizing bat sarbecoviruses, while potently neutralizing SARS-CoV-2 variants since 2019, including the recent EG.5, BA.2.86, and JN.1. VIR-7229 tolerates extraordinary epitope variability, partly attributed to its high binding affinity, receptor molecular mimicry, and interactions with RBM backbone atoms. Consequently, VIR-7229 features a high barrier for selection of escape mutants, which are rare and associated with reduced viral fitness, underscoring its potential to be resilient to future viral evolution. VIR-7229 is a strong candidate to become a next-generation medicine.

A deep learning approach predicting the activity of COVID-19 therapeutics and vaccines against emerging variants

Understanding which viral variants evade neutralization is crucial for improving antibody-based treatments, especially with rapidly evolving viruses like SARS-CoV-2. Yet, conventional assays are labor intensive and cannot capture the full spectrum of variants. We present a deep learning approach to predict changes in neutralizing antibody activity of COVID-19 therapeutics and vaccine-elicited sera/plasma against emerging viral variants. Our approach leverages data of 67,885 unique SARS-CoV-2 Spike sequences and 7,069 in vitro assays. The resulting model accurately predicted fold changes in neutralizing activity (R2 = 0.77) for a test set (N = 980) of data collected up to eight months after the training data. Next, the model was used to predict changes in activity of current therapeutic and vaccine-induced antibodies against emerging SARS-CoV-2 lineages. Consistent with other work, we found significantly reduced activity against newer XBB descendants, notably EG.5, FL.1.5.1, and XBB.1.16; primarily attributed to the F456L spike mutation.

Variation and evolution analysis of SARS-CoV-2 using self-game sequence optimization

Introduction

The evolution of SARS-CoV-2 has precipitated the emergence of new mutant strains, some exhibiting enhanced transmissibility and immune evasion capabilities, thus escalating the infection risk and diminishing vaccine efficacy. Given the continuous impact of SARS-CoV-2 mutations on global public health, the economy, and society, a profound comprehension of potential variations is crucial to effectively mitigate the impact of viral evolution. Yet, this task still faces considerable challenges.

Methods

This study introduces DARSEP, a method based on Deep learning Associates with Reinforcement learning for SARS-CoV-2 Evolution Prediction, combined with self-game sequence optimization and RetNet-based model.

Results

DARSEP accurately predicts evolutionary sequences and investigates the virus's evolutionary trajectory. It filters spike protein sequences with optimal fitness values from an extensive mutation space, selectively identifies those with a higher likelihood of evading immune detection, and devises a superior evolutionary analysis model for SARS-CoV-2 spike protein sequences. Comprehensive downstream task evaluations corroborate the model's efficacy in predicting potential mutation sites, elucidating SARS-CoV-2's evolutionary direction, and analyzing the development trends of Omicron variant strains through semantic changes.

Conclusion

Overall, DARSEP enriches our understanding of the dynamic evolution of SARS-CoV-2 and provides robust support for addressing present and future epidemic challenges.

On the Nature of the Interactions That Govern COV-2 Mutants Escape from Neutralizing Antibodies

The most fruitful prevention and treatment tools for the COVID-19 pandemic have proven to be vaccines and therapeutic antibodies, which have reduced the spread of the disease to manageable proportions. The search for the most effective antibodies against the widest set of COV-2 variants has required a long time and substantial resources. It would be desirable to have a tool that will enable us to understand the structural basis on which mutants escape at least some of the epitope-bound antibodies, a tool that may substantially reduce the time and resources invested in this effort. In this work, we applied a computational-based tool (employed previously by us to understand COV-2 spike binding to its cognate cell receptor) to the study of the effect of Delta and Omicron mutations on the escape tendencies. Our binding energy predictions agree extremely well with the experimentally observed escape tendencies. They have also allowed us to set forth a structural explanation for the results that could be used for the screening of antibodies. Lastly, our results explain the differences in molecular interactions that govern interaction of the spike variants with the receptor as opposed to those with antibodies.

Identification of antibody-resistant SARS-CoV-2 mutants via N4-Hydroxycytidine mutagenesis

Monoclonal antibodies targeting the Spike protein of SARS-CoV-2 are effective against COVID-19 and might mitigate future pandemics. However, their efficacy is challenged by the emergence of antibody-resistant virus variants. We developed a method to efficiently identify such resistant mutants based on selection from mutagenized virus pools. By inducing mutations with the active compound of Molnupiravir, N4-hydroxycytidine (NHC), and subsequently passaging the virus in the presence of antibodies, we identified specific Spike mutations linked to resistance. Validation of these mutations was conducted using pseudotypes and immunofluorescence analysis. From a Wuhan-like strain of SARS-CoV-2, we identified the following mutations conferring strong resistance towards the corresponding antibodies: Bamlanivimab - E484K, F490S and S494P; Sotrovimab - E340K; Cilgavimab - K444R/E and N450D. From the Omicron B.1.1.529 variant, the strongly selected mutations were: Bebtelovimab - V445A; Sotrovimab - E340K and K356M; Cilgavimab - K444R, V445A and N450D. We also identified escape mutations in the Wuhan-like Spike for the broadly neutralizing antibodies S2K146 - combined G485S and Q493R - and S2H97 - D428G, K462E and S514F. Structural analysis revealed that the selected mutations occurred at antibody-binding residues within the receptor-binding domains of the Spike protein. Most of the selected mutants largely maintained ACE2 binding and infectivity. Notably, many of the identified resistance-conferring mutations are prevalent in real-world SARS-CoV-2 variants, but some of them (G485S, D428G, and K462E) have not yet been observed in circulating strains. Our approach offers a strategy for predicting the therapeutic efficacy of antibodies against emerging virus variants.

The critical role of aromatic residues in the binding of the SARS-CoV-2 fusion peptide to phospholipid bilayer membranes

Based on the SARS-CoV-2 fusion peptide (FP) structure determined from the NMR experiment, we created six FP models under different environmental conditions to explore the effects of salt and cholesterol on FP-membrane binding. The all-atom molecular dynamics (MD) simulation results indicated that ionic environments notably impact the FP structure as well as the stability of the helical elements within the peptide. Our findings highlighted the unpredictable influence of ions on the secondary structures and dynamics of the FP, emphasizing the complexity and sensitivity of the peptide's conformations to ionic conditions. When exploring the peptide's interaction with a cholesterol-free phospholipid bilayer membrane, we found that the helical elements of the FP remain stable irrespective of the salt type (Na+ or Ca2+). This result emphasizes the crucial role of phospholipid bilayer membranes in supporting the secondary structures of the FP. The MD simulation results showed that Ca2+ ions facilitated deeper membrane penetration than Na+ ions, highlighting the critical role of calcium ions in the FP-membrane binding. Our study indicates the essential role of the aromatic residues (such as Phe833 and Tyr837) in the FP-membrane binding process. Finally, we investigated the FP-membrane binding patterns in the presence of cholesterol. The MD simulation results demonstrated that the coupling of Ca2+ ions and cholesterol would also benefit the FP-membrane binding. Furthermore, our findings reveal that while the type of ion and cholesterol content exert varied and unpredictable influences on FP-membrane binding patterns, aromatic residues like tyrosine (Tyr) and phenylalanine (Phe) play an essential role in FP-membrane binding. In particular, deep mutational scanning (DMS) experiments have confirmed that mutating phenylalanine in the FP significantly decreases viral mutational fitness, emphasizing the pivotal role of phenylalanine residues in membrane fusion. This knowledge can aid in developing more effective therapeutic strategies targeting the viral fusion peptide and its key amino acids, ultimately contributing to developing treatments and vaccines against the virus.

SARS-CoV-2 Evolution: Immune Dynamics, Omicron Specificity, and Predictive Modeling in Vaccinated Populations

Host immunity is central to the virus's spread dynamics, which is significantly influenced by vaccination and prior infection experiences. In this work, we analyzed the co-evolution of SARS-CoV-2 mutation, angiotensin-converting enzyme 2 (ACE2) receptor binding, and neutralizing antibody (NAb) responses across various variants in 822 human and mice vaccinated with different non-Omicron and Omicron vaccines is analyzed. The link between vaccine efficacy and vaccine type, dosing, and post-vaccination duration is revealed. The classification of immune protection against non-Omicron and Omicron variants is co-evolved with genetic mutations and vaccination. Additionally, a model, the Prevalence Score (P-Score) is introduced, which surpasses previous algorithm-based models in predicting the potential prevalence of new variants in vaccinated populations. The hybrid vaccination combining the wild-type (WT) inactivated vaccine with the Omicron BA.4/5 mRNA vaccine may provide broad protection against both non-Omicron variants and Omicron variants, albeit with EG.5.1 still posing a risk. In conclusion, these findings enhance understanding of population immunity variations and provide valuable insights for future vaccine development and public health strategies.

Forecasting dominance of SARS-CoV-2 lineages by anomaly detection using deep AutoEncoders

The coronavirus disease of 2019 (COVID-19) pandemic is characterized by sequential emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants, lineages, and sublineages, outcompeting previously circulating ones because of, among other factors, increased transmissibility and immune escape. We propose DeepAutoCoV, an unsupervised deep learning anomaly detection system to predict future dominant lineages (FDLs). We define FDLs as viral (sub)lineages that will constitute more than 10% of all the viral sequences added to the GISAID database on a given week. DeepAutoCoV is trained and validated by assembling global and country-specific data sets from over 16 million Spike protein sequences sampled over a period of about 4 years. DeepAutoCoV successfully flags FDLs at very low frequencies (0.01% - 3%), with median lead times of 4-17 weeks, and predicts FDLs ~5 and ~25 times better than a baseline approach For example, the B.1.617.2 vaccine reference strain was flagged as FDL when its frequency was only 0.01%, more than a year before it was considered for an updated COVID-19 vaccine. Furthermore, DeepAutoCoV outputs interpretable results by pinpointing specific mutations potentially linked to increased fitness, and may provide significant insights for the optimization of public health pre-emptive intervention strategies.

Forecasting dominance of SARS-CoV-2 lineages by anomaly detection using deep AutoEncoders

The COVID-19 pandemic is marked by the successive emergence of new SARS-CoV-2 variants, lineages, and sublineages that outcompete earlier strains, largely due to factors like increased transmissibility and immune escape. We propose DeepAutoCoV, an unsupervised deep learning anomaly detection system, to predict future dominant lineages (FDLs). We define FDLs as viral (sub)lineages that will constitute >10% of all the viral sequences added to the GISAID, a public database supporting viral genetic sequence sharing, in a given week. DeepAutoCoV is trained and validated by assembling global and country-specific data sets from over 16 million Spike protein sequences sampled over a period of ~4 years. DeepAutoCoV successfully flags FDLs at very low frequencies (0.01%-3%), with median lead times of 4-17 weeks, and predicts FDLs between ~5 and ~25 times better than a baseline approach. For example, the B.1.617.2 vaccine reference strain was flagged as FDL when its frequency was only 0.01%, more than a year before it was considered for an updated COVID-19 vaccine. Furthermore, DeepAutoCoV outputs interpretable results by pinpointing specific mutations potentially linked to increased fitness and may provide significant insights for the optimization of public health 'pre-emptive' intervention strategies.

Deep mutational scanning of SARS-CoV-2 Omicron BA.2.86 and epistatic emergence of the KP.3 variant

Deep mutational scanning experiments aid in the surveillance and forecasting of viral evolution by providing prospective measurements of mutational effects on viral traits, but epistatic shifts in the impacts of mutations can hinder viral forecasting when measurements were made in outdated strain backgrounds. Here, we report measurements of the impact of all single amino acid mutations on ACE2-binding affinity and protein folding and expression in the SARS-CoV-2 Omicron BA.2.86 spike receptor-binding domain. As with other SARS-CoV-2 variants, we find a plastic and evolvable basis for receptor binding, with many mutations at the ACE2 interface maintaining or even improving ACE2-binding affinity. Despite its large genetic divergence, mutational effects in BA.2.86 have not diverged greatly from those measured in its Omicron BA.2 ancestor. However, we do identify strong positive epistasis among subsequent mutations that have accrued in BA.2.86 descendants. Specifically, the Q493E mutation that decreased ACE2-binding affinity in all previous SARS-CoV-2 backgrounds is reversed in sign to enhance human ACE2-binding affinity when coupled with L455S and F456L in the currently emerging KP.3 variant. Our results point to a modest degree of epistatic drift in mutational effects during recent SARS-CoV-2 evolution but highlight how these small epistatic shifts can have important consequences for the emergence of new SARS-CoV-2 variants.

Zero-shot prediction of mutation effects with multimodal deep representation learning guides protein engineering

Mutations in amino acid sequences can provoke changes in protein function. Accurate and unsupervised prediction of mutation effects is critical in biotechnology and biomedicine, but remains a fundamental challenge. To resolve this challenge, here we present Protein Mutational Effect Predictor (ProMEP), a general and multiple sequence alignment-free method that enables zero-shot prediction of mutation effects. A multimodal deep representation learning model embedded in ProMEP was developed to comprehensively learn both sequence and structure contexts from ~160 million proteins. ProMEP achieves state-of-the-art performance in mutational effect prediction and accomplishes a tremendous improvement in speed, enabling efficient and intelligent protein engineering. Specifically, ProMEP accurately forecasts mutational consequences on the gene-editing enzymes TnpB and TadA, and successfully guides the development of high-performance gene-editing tools with their engineered variants. The gene-editing efficiency of a 5-site mutant of TnpB reaches up to 74.04% (vs 24.66% for the wild type); and the base editing tool developed on the basis of a TadA 15-site mutant (in addition to the A106V/D108N double mutation that renders deoxyadenosine deaminase activity to TadA) exhibits an A-to-G conversion frequency of up to 77.27% (vs 69.80% for ABE8e, a previous TadA-based adenine base editor) with significantly reduced bystander and off-target effects compared to ABE8e. ProMEP not only showcases superior performance in predicting mutational effects on proteins but also demonstrates a great capability to guide protein engineering. Therefore, ProMEP enables efficient exploration of the gigantic protein space and facilitates practical design of proteins, thereby advancing studies in biomedicine and synthetic biology.

Primate-specific BTN3A2 protects against SARS-CoV-2 infection by interacting with and reducing ACE2

BACKGROUND

Coronavirus disease 2019 (COVID-19) is an immune-related disorder caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). The complete pathogenesis of the virus remains to be determined. Unraveling the molecular mechanisms governing SARS-CoV-2 interactions with host cells is crucial for the formulation of effective prophylactic measures and the advancement of COVID-19 therapeutics.

METHODS

We analyzed human lung single-cell RNA sequencing dataset to discern the association of butyrophilin subfamily 3 member A2 (BTN3A2) expression with COVID-19. The BTN3A2 gene edited cell lines and transgenic mice were infected by live SARS-CoV-2 in a biosafety level 3 (BSL-3) laboratory. Immunoprecipitation, flow cytometry, biolayer interferometry and competition ELISA assays were performed in BTN3A2 gene edited cells. We performed quantitative real-time PCR, histological and/or immunohistochemical analyses for tissue samples from mice with or without SARS-CoV-2 infection.

FINDINGS

The BTN3A2 mRNA level was correlated with COVID-19 severity. BTN3A2 expression was predominantly identified in epithelial cells, elevated in pathological epithelial cells from COVID-19 patients and co-occurred with ACE2 expression in the same lung cell subtypes. BTN3A2 targeted the early stage of the viral life cycle by inhibiting SARS-CoV-2 attachment through interactions with the receptor-binding domain (RBD) of the Spike protein and ACE2. BTN3A2 inhibited ACE2-mediated SARS-CoV-2 infection by reducing ACE2 in vitro and in vivo.

INTERPRETATION

These results reveal a key role of BTN3A2 in the fight against COVID-19. Identifying potential monoclonal antibodies which mimic BTN3A2 may facilitate disruption of SARS-CoV-2 infection, providing a therapeutic avenue for COVID-19.

FUNDING

This study was supported by the National Natural Science Foundation of China (32070569, U1902215, and 32371017), the CAS "Light of West China" Program, and Yunnan Province (202305AH340006).

Leveraging artificial intelligence in vaccine development: A narrative review

Vaccine development stands as a cornerstone of public health efforts, pivotal in curbing infectious diseases and reducing global morbidity and mortality. However, traditional vaccine development methods are often time-consuming, costly, and inefficient. The advent of artificial intelligence (AI) has ushered in a new era in vaccine design, offering unprecedented opportunities to expedite the process. This narrative review explores the role of AI in vaccine development, focusing on antigen selection, epitope prediction, adjuvant identification, and optimization strategies. AI algorithms, including machine learning and deep learning, leverage genomic data, protein structures, and immune system interactions to predict antigenic epitopes, assess immunogenicity, and prioritize antigens for experimentation. Furthermore, AI-driven approaches facilitate the rational design of immunogens and the identification of novel adjuvant candidates with optimal safety and efficacy profiles. Challenges such as data heterogeneity, model interpretability, and regulatory considerations must be addressed to realize the full potential of AI in vaccine development. Integrating emerging technologies, such as single-cell omics and synthetic biology, promises to enhance vaccine design precision and scalability. This review underscores the transformative impact of AI on vaccine development and highlights the need for interdisciplinary collaborations and regulatory harmonization to accelerate the delivery of safe and effective vaccines against infectious diseases.

Complete combinatorial mutational enumeration of a protein functional site enables sequence-landscape mapping and identifies highly-mutated variants that retain activity

Understanding how proteins evolve under selective pressure is a longstanding challenge. The immensity of the search space has limited efforts to systematically evaluate the impact of multiple simultaneous mutations, so mutations have typically been assessed individually. However, epistasis, or the way in which mutations interact, prevents accurate prediction of combinatorial mutations based on measurements of individual mutations. Here, we use artificial intelligence to define the entire functional sequence landscape of a protein binding site in silico, and we call this approach Complete Combinatorial Mutational Enumeration (CCME). By leveraging CCME, we are able to construct a comprehensive map of the evolutionary connectivity within this functional sequence landscape. As a proof of concept, we applied CCME to the ACE2 binding site of the SARS-CoV-2 spike protein receptor binding domain. We selected representative variants from across the functional sequence landscape for testing in the laboratory. We identified variants that retained functionality to bind ACE2 despite changing over 40% of evaluated residue positions, and the variants now escape binding and neutralization by monoclonal antibodies. This work represents a crucial initial stride toward achieving precise predictions of pathogen evolution, opening avenues for proactive mitigation.

Opportunities and challenges in design and optimization of protein function

The field of protein design has made remarkable progress over the past decade. Historically, the low reliability of purely structure-based design methods limited their application, but recent strategies that combine structure-based and sequence-based calculations, as well as machine learning tools, have dramatically improved protein engineering and design. In this Review, we discuss how these methods have enabled the design of increasingly complex structures and therapeutically relevant activities. Additionally, protein optimization methods have improved the stability and activity of complex eukaryotic proteins. Thanks to their increased reliability, computational design methods have been applied to improve therapeutics and enzymes for green chemistry and have generated vaccine antigens, antivirals and drug-delivery nano-vehicles. Moreover, the high success of design methods reflects an increased understanding of basic rules that govern the relationships among protein sequence, structure and function. However, de novo design is still limited mostly to α-helix bundles, restricting its potential to generate sophisticated enzymes and diverse protein and small-molecule binders. Designing complex protein structures is a challenging but necessary next step if we are to realize our objective of generating new-to-nature activities.

Deep mutational scanning of SARS-CoV-2 Omicron BA.2.86 and epistatic emergence of the KP.3 variant

Deep mutational scanning experiments aid in the surveillance and forecasting of viral evolution by providing prospective measurements of mutational effects on viral traits, but epistatic shifts in the impacts of mutations can hinder viral forecasting when measurements were made in outdated strain backgrounds. Here, we report measurements of the impact of all single amino acid mutations on ACE2-binding affinity and protein folding and expression in the SARS-CoV-2 Omicron BA.2.86 spike receptor-binding domain (RBD). As with other SARS-CoV-2 variants, we find a plastic and evolvable basis for receptor binding, with many mutations at the ACE2 interface maintaining or even improving ACE2-binding affinity. Despite its large genetic divergence, mutational effects in BA.2.86 have not diverged greatly from those measured in its Omicron BA.2 ancestor. However, we do identify strong positive epistasis among subsequent mutations that have accrued in BA.2.86 descendants. Specifically, the Q493E mutation that decreased ACE2-binding affinity in all previous SARS-CoV-2 backgrounds is reversed in sign to enhance human ACE2-binding affinity when coupled with L455S and F456L in the currently emerging KP.3 variant. Our results point to a modest degree of epistatic drift in mutational effects during recent SARS-CoV-2 evolution but highlight how these small epistatic shifts can have important consequences for the emergence of new SARS-CoV-2 variants.

Predicting Natural Evolution in the RBD Region of the Spike Glycoprotein of SARS-CoV-2 by Machine Learning

Machine learning (ML) is a key focus in predicting protein mutations and aiding directed evolution. Research on potential virus variants is crucial for vaccine development. In this study, the machine learning software PyPEF was employed to conduct mutation analysis within the receptor-binding domain (RBD) of the Spike glycoprotein of SARS-CoV-2. Over 48,960,000 variants were predicted. Eight prospective variants that could surface in the future underwent modeling and molecular dynamics simulations. The study forecasts that the latest variant, ISOY2P5O1, may potentially emerge around 17 November 2023, with an approximate window of uncertainty of ±22 days. The ISOY8P5O2 variant displayed an increased binding capacity in the dry assay, with a total predicted binding energy of -110.306 kcal/mol. This represents an 8.25% enhancement in total binding energy compared to the original SARS-CoV-2 strain discovered in Wuhan (-101.892 kcal/mol). Reverse research confirmed the structural significance of mutation sites using ML models, particularly in the context of protein folding. The study validated regression methods (SVR, RF, and PLS) with different data structures. This study investigates the effectiveness of the "ML-Guided Design Correctly Predicts Combinatorial Effects Strategy" compared to the "ML-Guided Design Correctly Predicts Natural Evolution Prediction Strategy". To enhance machine learning, we created a timestamping algorithm and two auxiliary programs using advanced techniques to rapidly process extensive data, surpassing batch sequencing capabilities. This study not only advances machine learning in guiding protein evolution but also holds potential for forecasting future viruses and vaccine development.

High-throughput screening of genetic and cellular drivers of syncytium formation induced by the spike protein of SARS-CoV-2

Mapping mutations and discovering cellular determinants that cause the spike protein of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) to induce infected cells to form syncytia would facilitate the development of strategies for blocking the formation of such cell-cell fusion. Here we describe high-throughput screening methods based on droplet microfluidics and the size-exclusion selection of syncytia, coupled with large-scale mutagenesis and genome-wide knockout screening via clustered regularly interspaced short palindromic repeats (CRISPR), for the large-scale identification of determinants of cell-cell fusion. We used the methods to perform deep mutational scans in spike-presenting cells to pinpoint mutable syncytium-enhancing substitutions in two regions of the spike protein (the fusion peptide proximal region and the furin-cleavage site). We also used a genome-wide CRISPR screen in cells expressing the receptor angiotensin-converting enzyme 2 to identify inhibitors of clathrin-mediated endocytosis that impede syncytium formation, which we validated in hamsters infected with SARS-CoV-2. Finding genetic and cellular determinants of the formation of syncytia may reveal insights into the physiological and pathological consequences of cell-cell fusion.

Fast-Track Discovery of SARS-CoV-2-Neutralizing Antibodies from Human B Cells by Direct Functional Screening

As the COVID-19 pandemic revealed, rapid development of vaccines and therapeutic antibodies are crucial to guarantee a quick return to the status quo of society. In early 2020, we deployed our droplet microfluidic single-cell-based platform DROPZYLLA® for the generation of cognate antibody repertoires of convalescent COVID-19 donors. Discovery of SARS-CoV-2-specific antibodies was performed upon display of antibodies on the surface of HEK293T cells by antigen-specific sorting using binding to the SARS-CoV-2 spike and absence of binding to huACE2 as the sort criteria. This efficiently yielded antibodies within 3-6 weeks, of which up to 100% were neutralizing. One of these, MTX-COVAB, displaying low picomolar neutralization IC50 of SARS-CoV-2 and with a neutralization potency on par with the Regeneron antibodies, was selected for GMP manufacturing and clinical development in June 2020. MTX-COVAB showed strong efficacy in vivo and neutralized all identified clinically relevant variants of SARS-CoV-2 at the time of its selection. MTX-COVAB completed GMP manufacturing by the end of 2020, but clinical development was stopped when the Omicron variant emerged, a variant that proved to be detrimental to all monoclonal antibodies already approved. The present study describes the capabilities of the DROPZYLLA® platform to identify antibodies of high virus-neutralizing capacity rapidly and directly.

Identifying LasR Quorum Sensors with Improved Signal Specificity by Mapping the Sequence-Function Landscape

Programmable intercellular signaling using components of naturally occurring quorum sensing can allow for coordinated functions to be engineered in microbial consortia. LuxR-type transcriptional regulators are widely used for this purpose and are activated by homoserine lactone (HSL) signals. However, they often suffer from imperfect molecular discrimination of structurally similar HSLs, causing misregulation within engineered consortia containing multiple HSL signals. Here, we studied one such example, the regulator LasR from Pseudomonas aeruginosa. We elucidated its sequence-function relationship for ligand specificity using targeted protein engineering and multiplexed high-throughput biosensor screening. A pooled combinatorial saturation mutagenesis library (9,486 LasR DNA sequences) was created by mutating six residues in LasR's β5 sheet with single, double, or triple amino acid substitutions. Sort-seq assays were performed in parallel using cognate and noncognate HSLs to quantify each corresponding sensor's response to each HSL signal, which identified hundreds of highly specific variants. Sensor variants identified were individually assayed and exhibited up to 60.6-fold (p = 0.0013) improved relative activation by the cognate signal compared to the wildtype. Interestingly, we uncovered prevalent mutational epistasis and previously unidentified residues contributing to signal specificity. The resulting sensors with negligible signal crosstalk could be broadly applied to engineer bacteria consortia.

Unraveling the Dynamics of SARS-CoV-2 Mutations: Insights from Surface Plasmon Resonance Biosensor Kinetics

Surface Plasmon Resonance (SPR) technology is known to be a powerful tool for studying biomolecular interactions because it offers real-time and label-free multiparameter analysis with high sensitivity. This article summarizes the results that have been obtained from the use of SPR technology in studying the dynamics of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) mutations. This paper will begin by introducing the working principle of SPR and the kinetic parameters of the sensorgram, which include the association rate constant (ka), dissociation rate constant (kd), equilibrium association constant (KA), and equilibrium dissociation constant (KD). At the end of the paper, we will summarize the kinetic data on the interaction between angiotensin-converting enzyme 2 (ACE2) and SARS-CoV-2 obtained from the results of SPR signal analysis. ACE2 is a material that mediates virus entry. Therefore, understanding the kinetic changes between ACE2 and SARS-CoV-2 caused by the mutation will provide beneficial information for drug discovery, vaccine development, and other therapeutic purposes.

Predicting the immune escape of SARS-CoV-2 neutralizing antibodies upon mutation

COVID-19 has resulted in millions of deaths and severe impact on economies worldwide. Moreover, the emergence of SARS-CoV-2 variants presented significant challenges in controlling the pandemic, particularly their potential to avoid the immune system and evade vaccine immunity. This has led to a growing need for research to predict how mutations in SARS-CoV-2 reduces the ability of antibodies to neutralize the virus. In this study, we assembled a set of 1813 mutations from the interface of SARS-CoV-2 spike protein's receptor binding domain (RBD) and neutralizing antibody complexes and developed a machine learning model to classify high or low escape mutations using interaction energy, inter-residue contacts and predicted binding free energy change. Our approach achieved an Area under the Receiver Operating Characteristics (ROC) Curve (AUC) of 0.91 using the Random Forest classifier on the test dataset with 217 mutations. The model was further utilized to predict the escape mutations on a dataset of 29,165 mutations located at the interface of 83 RBD-neutralizing antibody complexes. A small subset of this dataset was also validated based on available experimental data. We found that top 10 % high escape mutations were dominated by charged to nonpolar mutations whereas low escape mutations were dominated by polar to nonpolar mutations. We believe that the present method will allow prioritization of high/low escape mutations in the context of neutralizing antibodies targeting SARS-CoV-2 RBD region and assist antibody design for current and emerging variants.

Meta learning addresses noisy and under-labeled data in machine learning-guided antibody engineering

Machine learning-guided protein engineering is rapidly progressing; however, collecting high-quality, large datasets remains a bottleneck. Directed evolution and protein engineering studies often require extensive experimental processes to eliminate noise and label protein sequence-function data. Meta learning has proven effective in other fields in learning from noisy data via bi-level optimization given the availability of a small dataset with trusted labels. Here, we leverage meta learning approaches to overcome noisy and under-labeled data and expedite workflows in antibody engineering. We generate yeast display antibody mutagenesis libraries and screen them for target antigen binding followed by deep sequencing. We then create representative learning tasks, including learning from noisy training data, positive and unlabeled learning, and learning out of distribution properties. We demonstrate that meta learning has the potential to reduce experimental screening time and improve the robustness of machine learning models by training with noisy and under-labeled training data.

Advancing Antibody Engineering through Synthetic Evolution and Machine Learning

Abs are versatile molecules with the potential to achieve exceptional binding to target Ags, while also possessing biophysical properties suitable for therapeutic drug development. Protein display and directed evolution systems have transformed synthetic Ab discovery, engineering, and optimization, vastly expanding the number of Ab clones able to be experimentally screened for binding. Moreover, the burgeoning integration of high-throughput screening, deep sequencing, and machine learning has further augmented in vitro Ab optimization, promising to accelerate the design process and massively expand the Ab sequence space interrogated. In this Brief Review, we discuss the experimental and computational tools employed in synthetic Ab engineering and optimization. We also explore the therapeutic challenges posed by developing Abs for infectious diseases, and the prospects for leveraging machine learning-guided protein engineering to prospectively design Abs resistant to viral escape.

PandoGen: Generating complete instances of future SARS-CoV-2 sequences using Deep Learning

One of the challenges in a viral pandemic is the emergence of novel variants with different phenotypical characteristics. An ability to forecast future viral individuals at the sequence level enables advance preparation by characterizing the sequences and closing vulnerabilities in current preventative and therapeutic methods. In this article, we explore, in the context of a viral pandemic, the problem of generating complete instances of undiscovered viral protein sequences, which have a high likelihood of being discovered in the future using protein language models. Current approaches to training these models fit model parameters to a known sequence set, which does not suit pandemic forecasting as future sequences differ from known sequences in some respects. To address this, we develop a novel method, called PandoGen, to train protein language models towards the pandemic protein forecasting task. PandoGen combines techniques such as synthetic data generation, conditional sequence generation, and reward-based learning, enabling the model to forecast future sequences, with a high propensity to spread. Applying our method to modeling the SARS-CoV-2 Spike protein sequence, we find empirically that our model forecasts twice as many novel sequences with five times the case counts compared to a model that is 30× larger. Our method forecasts unseen lineages months in advance, whereas models 4× and 30× larger forecast almost no new lineages. When trained on data available up to a month before the onset of important Variants of Concern, our method consistently forecasts sequences belonging to those variants within tight sequence budgets.

Protein Engineering and High-Throughput Screening by Yeast Surface Display: Survey of Current Methods

Yeast surface display (YSD) is a powerful tool in biotechnology that links genotype to phenotype. In this review, the latest advancements in protein engineering and high-throughput screening based on YSD are covered. The focus is on innovative methods for overcoming challenges in YSD in the context of biotherapeutic drug discovery and diagnostics. Topics ranging from titrating avidity in YSD using transcriptional control to the development of serological diagnostic assays relying on serum biopanning and mitigation of unspecific binding are covered. Screening techniques against nontraditional cellular antigens, such as cell lysates, membrane proteins, and extracellular matrices are summarized and techniques are further delved into for expansion of the chemical repertoire, considering protein-small molecule hybrids and noncanonical amino acid incorporation. Additionally, in vivo gene diversification and continuous evolution in yeast is discussed. Collectively, these techniques enhance the diversity and functionality of engineered proteins isolated via YSD, broadening the scope of applications that can be addressed. The review concludes with future perspectives and potential impact of these advancements on protein engineering. The goal is to provide a focused summary of recent progress in the field.

Deep mutational scans of XBB.1.5 and BQ.1.1 reveal ongoing epistatic drift during SARS-CoV-2 evolution

Substitutions that fix between SARS-CoV-2 variants can transform the mutational landscape of future evolution via epistasis. For example, large epistatic shifts in mutational effects caused by N501Y underlied the original emergence of Omicron, but whether such epistatic saltations continue to define ongoing SARS-CoV-2 evolution remains unclear. We conducted deep mutational scans to measure the impacts of all single amino acid mutations and single-codon deletions in the spike receptor-binding domain (RBD) on ACE2-binding affinity and protein expression in the recent Omicron BQ.1.1 and XBB.1.5 variants, and we compared mutational patterns to earlier viral strains that we have previously profiled. As with previous deep mutational scans, we find many mutations that are tolerated or even enhance binding to ACE2 receptor. The tolerance of sites to single-codon deletion largely conforms with tolerance to amino acid mutation. Though deletions in the RBD have not yet been seen in dominant lineages, we observe tolerated deletions including at positions that exhibit indel variation across broader sarbecovirus evolution and in emerging SARS-CoV-2 variants of interest, most notably the well-tolerated Δ483 deletion in BA.2.86. The substitutions that distinguish recent viral variants have not induced as dramatic of epistatic perturbations as N501Y, but we identify ongoing epistatic drift in SARS-CoV-2 variants, including interaction between R493Q reversions and mutations at positions 453, 455, and 456, including F456L that defines the XBB.1.5-derived EG.5 lineage. Our results highlight ongoing drift in the effects of mutations due to epistasis, which may continue to direct SARS-CoV-2 evolution into new regions of sequence space.

Dynamics of water-mediated interaction effects on the stability and transmission of Omicron

SARS-Cov-2 Omicron variant and its highly transmissible sublineages amidst news of emerging hybrid variants strengthen the evidence of its ability to rapidly spread and evolve giving rise to unprecedented future waves. Owing to the presence of isolated RBD, monomeric and trimeric Cryo-EM structures of spike protein in complex with ACE2 receptor, comparative analysis of Alpha, Beta, Gamma, Delta, and Omicron assist in a rational assessment of their probability to evolve as new or hybrid variants in future. This study proposes the role of hydration forces in mediating Omicron function and dynamics based on a stronger interplay between protein and solvent with each Covid wave. Mutations of multiple hydrophobic residues into hydrophilic residues underwent concerted interactions with water leading to variations in charge distribution in Delta and Omicron during molecular dynamics simulations. Moreover, comparative analysis of interacting moieties characterized a large number of mutations lying at RBD into constrained, homologous and low-affinity groups referred to as mutational drivers inferring that the probability of future mutations relies on their function. Furthermore, the computational findings reveal a significant difference in angular distances among variants of concern due 3 amino acid insertion (EPE) in Omicron variant that not only facilitates tight domain organization but also seems requisite for characterization of mutational processes. The outcome of this work signifies the possible relation between hydration forces, their impact on conformation and binding affinities, and viral fitness that will significantly aid in understanding dynamics of drug targets for Covid-19 countermeasures. The emerging scenario is that hydration forces and hydrophobic interactions are crucial variables to probe in mutational analysis to explore conformational landscape of macromolecules and reveal the molecular origins of protein behaviors.

Insights into next generation sequencing guided antibody selection strategies

Therapeutic antibody discovery often relies on in-vitro display methods to identify lead candidates. Assessing selected output diversity traditionally involves random colony picking and Sanger sequencing, which has limitations. Next-generation sequencing (NGS) offers a cost-effective solution with increased read depth, allowing a comprehensive understanding of diversity. Our study establishes NGS guidelines for antibody drug discovery, demonstrating its advantages in expanding the number of unique HCDR3 clusters, broadening the number of high affinity antibodies, expanding the total number of antibodies recognizing different epitopes, and improving lead prioritization. Surprisingly, our investigation into the correlation between NGS-derived frequencies of CDRs and affinity revealed a lack of association, although this limitation could be moderately mitigated by leveraging NGS clustering, enrichment and/or relative abundance across different regions to enhance lead prioritization. This study highlights NGS benefits, offering insights, recommendations, and the most effective approach to leverage NGS in therapeutic antibody discovery.

An AsCas12f-based compact genome-editing tool derived by deep mutational scanning and structural analysis

SpCas9 and AsCas12a are widely utilized as genome-editing tools in human cells. However, their relatively large size poses a limitation for delivery by cargo-size-limited adeno-associated virus (AAV) vectors. The type V-F Cas12f from Acidibacillus sulfuroxidans is exceptionally compact (422 amino acids) and has been harnessed as a compact genome-editing tool. Here, we developed an approach, combining deep mutational scanning and structure-informed design, to successfully generate two AsCas12f activity-enhanced (enAsCas12f) variants. Remarkably, the enAsCas12f variants exhibited genome-editing activities in human cells comparable with those of SpCas9 and AsCas12a. The cryoelectron microscopy (cryo-EM) structures revealed that the mutations stabilize the dimer formation and reinforce interactions with nucleic acids to enhance their DNA cleavage activities. Moreover, enAsCas12f packaged with partner genes in an all-in-one AAV vector exhibited efficient knock-in/knock-out activities and transcriptional activation in mice. Taken together, enAsCas12f variants could offer a minimal genome-editing platform for in vivo gene therapy.

Binding of synthetic nanobodies to the SARS-CoV-2 receptor-binding domain: the importance of salt bridges

In this study, five different SARS-CoV-2 receptor-binding domain (RBD) models were created based on the crystal structures of RBD complexes with two synthetic nanobodies (Sb16 and Sb45). Microsecond all-atom MD simulations revealed that Sb16 and Sb45 substantially stabilized the flexible RBD loop (residues GLU471-SER494) due to the salt bridges and hydrogen bonding interactions between RBD and the synthetic nanobodies. However, the calculation of binding free energy displayed that Sb45 had a higher binding affinity to RBD than Sb16, in agreement with the experimental result. This is because Sb45 has stronger electrostatic attraction to RBD as compared to Sb16. In particular, the salt bridge GLU484-ARG33 in Sb45-RBD is stronger than the GLU484-LYS32 in Sb16-RBD. Furthermore, by comparing the binding affinity of Sb16 for two RBD mutants (E484K and K417N), we found that E484K mutation substantially reduced the binding affinity to Sb16, and K417N mutation had no significant effect, qualitatively in agreement with experimental studies. According to the binding free energy calculation, the strong electrostatic repulsion between LYS32 and LYS484 caused by E484K mutation destroys the salt bridge between LYS32 and GLU484 in the RBD wild type (WT). In contrast, the binding of the K417N mutant to Sb16 effectively maintains the salt bridge between LYS32 and GLU484. Therefore, our research suggests that the salt bridges between RBD and synthetic nanobodies are crucial for binding synthetic nanobodies to RBD, and a SARS-CoV-2 variant can escape neutralization from nanobodies by creating electrostatic repulsion between them.

Deep mutational scans of XBB.1.5 and BQ.1.1 reveal ongoing epistatic drift during SARS-CoV-2 evolution

Substitutions that fix between SARS-CoV-2 variants can transform the mutational landscape of future evolution via epistasis. For example, large epistatic shifts in mutational effects caused by N501Y underlied the original emergence of Omicron variants, but whether such large epistatic saltations continue to define ongoing SARS-CoV-2 evolution remains unclear. We conducted deep mutational scans to measure the impacts of all single amino acid mutations and single-codon deletions in the spike receptor-binding domain (RBD) on ACE2-binding affinity and protein expression in the recent Omicron BQ.1.1 and XBB.1.5 variants, and we compared mutational patterns to earlier viral strains that we have previously profiled. As with previous RBD deep mutational scans, we find many mutations that are tolerated or even enhance binding to ACE2 receptor. The tolerance of sites to single-codon deletion largely conforms with tolerance to amino acid mutation. Though deletions in the RBD have not yet been seen in dominant lineages, we observe many tolerated deletions including at positions that exhibit indel variation across broader sarbecovirus evolution and in emerging SARS-CoV-2 variants of interest, most notably the well-tolerated Δ483 deletion in BA.2.86. The substitutions that distinguish recent viral variants have not induced as dramatic of epistatic perturbations as N501Y, but we identify ongoing epistatic drift in SARS-CoV-2 variants, including interaction between R493Q reversions and mutations at positions 453, 455, and 456, including mutations like F456L that define the newly emerging EG.5 lineage. Our results highlight ongoing drift in the effects of mutations due to epistasis, which may continue to direct SARS-CoV-2 evolution into new regions of sequence space.

Complete Combinatorial Mutational Enumeration of a protein functional site enables sequence-landscape mapping and identifies highly-mutated variants that retain activity

Understanding how proteins evolve under selective pressure is a longstanding challenge. The immensity of the search space has limited efforts to systematically evaluate the impact of multiple simultaneous mutations, so mutations have typically been assessed individually. However, epistasis, or the way in which mutations interact, prevents accurate prediction of combinatorial mutations based on measurements of individual mutations. Here, we use artificial intelligence to define the entire functional sequence landscape of a protein binding site in silico, and we call this approach Complete Combinatorial Mutational Enumeration (CCME). By leveraging CCME, we are able to construct a comprehensive map of the evolutionary connectivity within this functional sequence landscape. As a proof of concept, we applied CCME to the ACE2 binding site of the SARS-CoV-2 spike protein receptor binding domain. We selected representative variants from across the functional sequence landscape for testing in the laboratory. We identified variants that retained functionality to bind ACE2 despite changing over 40% of evaluated residue positions, and the variants now escape binding and neutralization by monoclonal antibodies. This work represents a crucial initial stride towards achieving precise predictions of pathogen evolution, opening avenues for proactive mitigation.

Learning protein fitness landscapes with deep mutational scanning data from multiple sources

One of the key points of machine learning-assisted directed evolution (MLDE) is the accurate learning of the fitness landscape, a conceptual mapping from sequence variants to the desired function. Here, we describe a multi-protein training scheme that leverages the existing deep mutational scanning data from diverse proteins to aid in understanding the fitness landscape of a new protein. Proof-of-concept trials are designed to validate this training scheme in three aspects: random and positional extrapolation for single-variant effects, zero-shot fitness predictions for new proteins, and extrapolation for higher-order variant effects from single-variant effects. Moreover, our study identified previously overlooked strong baselines, and their unexpectedly good performance brings our attention to the pitfalls of MLDE. Overall, these results may improve our understanding of the association between different protein fitness profiles and shed light on developing better machine learning-assisted approaches to the directed evolution of proteins. A record of this paper's transparent peer review process is included in the supplemental information.

Challenges and Progress in Designing Broad-Spectrum Vaccines Against Rapidly Mutating Viruses

Viruses evolve to evade prior immunity, causing significant disease burden. Vaccine effectiveness deteriorates as pathogens mutate, requiring redesign. This is a problem that has grown worse due to population increase, global travel, and farming practices. Thus, there is significant interest in developing broad-spectrum vaccines that mitigate disease severity and ideally inhibit disease transmission without requiring frequent updates. Even in cases where vaccines against rapidly mutating pathogens have been somewhat effective, such as seasonal influenza and SARS-CoV-2 (severe acute respiratory syndrome coronavirus 2), designing vaccines that provide broad-spectrum immunity against routinely observed viral variation remains a desirable but not yet achieved goal. This review highlights the key theoretical advances in understanding the interplay between polymorphism and vaccine efficacy, challenges in designing broad-spectrum vaccines, and technology advances and possible avenues forward. We also discuss data-driven approaches for monitoring vaccine efficacy and predicting viral escape from vaccine-induced protection. In each case, we consider illustrative examples in vaccine development from influenza, SARS-CoV-2, and HIV (human immunodeficiency virus)-three examples of highly prevalent rapidly mutating viruses with distinct phylogenetics and unique histories of vaccine technology development.

Deep-learning-enabled protein-protein interaction analysis for prediction of SARS-CoV-2 infectivity and variant evolution

Host-pathogen interactions and pathogen evolution are underpinned by protein-protein interactions between viral and host proteins. An understanding of how viral variants affect protein-protein binding is important for predicting viral-host interactions, such as the emergence of new pathogenic SARS-CoV-2 variants. Here we propose an artificial intelligence-based framework called UniBind, in which proteins are represented as a graph at the residue and atom levels. UniBind integrates protein three-dimensional structure and binding affinity and is capable of multi-task learning for heterogeneous biological data integration. In systematic tests on benchmark datasets and further experimental validation, UniBind effectively and scalably predicted the effects of SARS-CoV-2 spike protein variants on their binding affinities to the human ACE2 receptor, as well as to SARS-CoV-2 neutralizing monoclonal antibodies. Furthermore, in a cross-species analysis, UniBind could be applied to predict host susceptibility to SARS-CoV-2 variants and to predict future viral variant evolutionary trends. This in silico approach has the potential to serve as an early warning system for problematic emerging SARS-CoV-2 variants, as well as to facilitate research on protein-protein interactions in general.

Evaluation of Mutual Information and Feature Selection for SARS-CoV-2 Respiratory Infection

This study aims to develop a predictive model for SARS-CoV-2 using machine-learning techniques and to explore various feature selection methods to enhance the accuracy of predictions. A precise forecast of the SARS-CoV-2 respiratory infections spread can help with efficient planning and resource allocation. The proposed model utilizes stochastic regression to capture the virus transmission's stochastic nature, considering data uncertainties. Feature selection techniques are employed to identify the most relevant and informative features contributing to prediction accuracy. Furthermore, the study explores the use of neighbor embedding and Sammon mapping algorithms to visualize high-dimensional SARS-CoV-2 respiratory infection data in a lower-dimensional space, enabling better interpretation and understanding of the underlying patterns. The application of machine-learning techniques for predicting SARS-CoV-2 respiratory infections, the use of statistical measures in healthcare, including confirmed cases, deaths, and recoveries, and an analysis of country-wise dynamics of the pandemic using machine-learning models are used. Our analysis involves the performance of various algorithms, including neural networks (NN), decision trees (DT), random forests (RF), the Adam optimizer (AD), hyperparameters (HP), stochastic regression (SR), neighbor embedding (NE), and Sammon mapping (SM). A pre-processed and feature-extracted SARS-CoV-2 respiratory infection dataset is combined with ADHPSRNESM to form a new orchestration in the proposed model for a perfect prediction to increase the precision of accuracy. The findings of this research can contribute to public health efforts by enabling policymakers and healthcare professionals to make informed decisions based on accurate predictions, ultimately aiding in managing and controlling the SARS-CoV-2 pandemic.

Multiplexed on-yeast serological assay for immune escape screening of SARS-CoV-2 variants

The emergence of the SARS-CoV-2 Omicron variant altered patient risk profiles and shifted the trajectory of the COVID-19 pandemic. Therefore, sensitive serological tests capable of analyzing patient IgG responses to multiple variants in parallel are highly desirable. Here, we present an adaptable serological test based on yeast surface display and serum biopanning that characterizes immune profiles against SARS-CoV-2 Wuhan (B lineage), Delta (B.1.617.2 lineage), and Omicron (B.1.1.529 lineage) receptor-binding domain (RBD) variants. We examined IgG titers from 30 serum samples from COVID-19-convalescent and vaccinated cohorts in Switzerland, and assessed the relative affinity of polyclonal serum IgG for RBD domains. We demonstrate that serum IgGs from patients recovered from severe COVID-19 between March-June 2021 bound tightly to both original Wuhan and Delta RBD variants, but failed to recognize Omicron RBDs, representing an affinity loss of >10- to 20-fold. Our yeast immunoassay is easily tailored, expandable and parallelized with newly emerging RBD variants.

Genotype-phenotype landscapes for immune-pathogen coevolution

Our immune systems constantly coevolve with the pathogens that challenge them, as pathogens adapt to evade our defense responses, with our immune repertoires shifting in turn. These coevolutionary dynamics take place across a vast and high-dimensional landscape of potential pathogen and immune receptor sequence variants. Mapping the relationship between these genotypes and the phenotypes that determine immune-pathogen interactions is crucial for understanding, predicting, and controlling disease. Here, we review recent developments applying high-throughput methods to create large libraries of immune receptor and pathogen protein sequence variants and measure relevant phenotypes. We describe several approaches that probe different regions of the high-dimensional sequence space and comment on how combinations of these methods may offer novel insight into immune-pathogen coevolution.