Tpgen: a language model for stable protein design with a specific topology structure.

BACKGROUND: Natural proteins occupy a small portion of the protein sequence space, whereas artificial proteins can explore a wider range of possibilities within the sequence space. However, specific requirements may not be met when generating sequences blindly. Research indicates that small proteins have notable advantages, including high stability, accurate resolution prediction, and facile specificity modification. RESULTS: This study involves the construction of a neural network model named TopoProGenerator(TPGen) using a transformer decoder. The model is trained with sequences consisting of a maximum of 65 amino acids. The training process of TopoProGenerator incorporates reinforcement learning and adversarial learning, for fine-tuning. Additionally, it encompasses a stability predictive model trained with a dataset comprising over 200,000 sequences. The results demonstrate that TopoProGenerator is capable of designing stable small protein sequences with specified topology structures. CONCLUSION: TPGen has the ability to generate protein sequences that fold into the specified topology, and the pretraining and fine-tuning methods proposed in this study can serve as a framework for designing various types of proteins.

Structure and proposed DNA delivery mechanism of a marine roseophage.

Tailed bacteriophages (order, Caudovirales) account for the majority of all phages. However, the long flexible tail of siphophages hinders comprehensive investigation of the mechanism of viral gene delivery. Here, we report the atomic capsid and in-situ structures of the tail machine of the marine siphophage, vB_DshS-R4C (R4C), which infects Roseobacter. The R4C virion, comprising 12 distinct structural protein components, has a unique five-fold vertex of the icosahedral capsid that allows genome delivery. The specific position and interaction pattern of the tail tube proteins determine the atypical long rigid tail of R4C, and further provide negative charge distribution within the tail tube. A ratchet mechanism assists in DNA transmission, which is initiated by an absorption device that structurally resembles the phage-like particle, RcGTA. Overall, these results provide in-depth knowledge into the intact structure and underlining DNA delivery mechanism for the ecologically important siphophages.

Critical Residues Involved in the Coassembly of L1 and L2 Capsid Proteins of Human Papillomavirus 16.

Human papillomaviruses (HPV) are small DNA viruses associated with cervical cancer, warts, and other epithelial tumors. Structural studies have shown that the HPV capsid consists of 360 copies of the major capsid protein, L1, arranged as 72 pentamers in a T=7 icosahedral lattice, coassembling with substoichiometric amounts of the minor capsid protein, L2. However, the residues involved in the coassembly of L1 and L2 remain undefined due to the lack of structure information. Here, we investigated the solvent accessibility surfaces (SASs) of the central cavity residues of the HPV16 L1 pentamer in the crystal structure because those internal exposed residues might mediate the association with L2. Twenty residues in L1 protein were selected to be analyzed, with four residues in the lumen of the L1 pentamer identified as important: F256, R315, Q317, and T340. Mutations to these four residues reduced the PsV (pseudovirus) infection capacity in 293FT cells, and mutations to R315, Q317, and T340 substantially perturb L2 from coassembling into L1 capsid. Compared with wild-type (WT) PsVs, these mutant PsVs also have a reduced ability to become internalized into host cells. Finally, we identified a stretch of negatively charged residues on L2 (amino acids [aa] 337 to 340 [EEIE]), mutations to which completely abrogate L2 assembly into L1 capsid and subsequently impair the endocytosis and infectivity of HPV16 PsVs. These findings shed light on the elusive coassembly between HPV L1 and L2. IMPORTANCE Over 200 types of HPV have been isolated, with several high-risk types correlated with the occurrence of cervical cancer. The HPV major capsid protein, L1, assembles into a T=7 icosahedral viral shell, and associates with the minor capsid protein, L2, which plays a critical role in the HPV life cycle. Despite the important role of the L2 protein, its structure and coassembly with L1 remain elusive. In this study, we analyzed the amino acid residues at the proposed interface between L1 and L2. Certain mutations at these sites decreased the amount of L2 protein assembled into the capsid, which, in turn, led to a decrease in viral infectivity. Knowledge about these residues and the coassembly of L1 and L2 could help to expand our understanding of HPV biology and aid in the development of countermeasures against a wide range of HPV types by targeting the L2 protein.

Omics-guided bacterial engineering of Escherichia coli ER2566 for recombinant protein expression.

The goal of bacterial engineering is to rewire metabolic pathways to generate high-value molecules for various applications. However, the production of recombinant proteins is constrained by the complexity of the connections between cellular physiology and recombinant protein synthesis. Here, we used a rational and highly efficient approach to improve bacterial engineering. Based on the complete genome and annotation information of the Escherichia coli ER2566 strain, we compared the transcriptomic profiles of the strain under leaky expression and low temperature-induced stress. Combining the gene ontology (GO) enrichment terms and differentially expressed genes (DEGs) with higher expression, we selected and knocked out 36 genes to determine the potential impact of these genes on protein production. Deletion of bluF, cydA, mngR, and udp led to a significant decrease in soluble recombinant protein production. Moreover, at low-temperature induction, 4 DEGs (gntK, flgH, flgK, flgL) were associated with enhanced expression of the recombinant protein. Knocking out several motility-related DEGs (ER2666-ΔflgH-ΔflgL-ΔflgK) simultaneously improved the protein yield by 1.5-fold at 24 °C induction, and the recombinant strain had the potential to be applied in the expression studies of different exogenous proteins, aiming to improve the yields of soluble form to varying degrees in comparison to the ER2566 strain. Totally, this study focused on the anabolic and stress-responsive hub genes of the adaptation of E. coli to recombinant protein overexpression on the transcriptome level and constructs a series of engineering strains increasing the soluble protein yield of recombinant proteins which lays a solid foundation for the engineering of bacterial strains for recombinant technological advances. KEY POINTS: • Comparative transcriptome analysis shows host responses with altered induction stress. • Deletion of bluF, cydA, mngR, and udp genes was identified to significantly decrease the soluble recombinant protein productions. • Synchronal knockout of flagellar genes in E. coli can enhance recombinant protein yield up to ~ 1.5-fold at 24 °C induction. • Non-model bacterial strains can be re-engineered for recombinant protein expression.

Structural basis for the synergistic neutralization of coxsackievirus B1 by a triple-antibody cocktail.

Coxsackievirus B1 (CVB1) is an emerging pathogen associated with severe neonatal diseases including aseptic meningitis, myocarditis, and pancreatitis and also with the development of type 1 diabetes. We characterize the binding and therapeutic efficacies of three CVB1-specific neutralizing antibodies (nAbs) identified for their ability to inhibit host receptor engagement. High-resolution cryo-EM structures showed that these antibodies recognize different epitopes but with an overlapping region in the capsid VP2 protein and specifically the highly variable EF loop. Moreover, they perturb capsid-receptor interactions by binding various viral particle forms. Antibody combinations achieve synergetic neutralization via a stepwise capsid transition and virion disruption, indicating dynamic changes in the virion in response to multiple nAbs targeting the receptor-binding site. Furthermore, this three-antibody cocktail protects against lethal challenge in neonatal mice and limits pancreatitis and viral replication in a non-obese diabetic mouse model. These results illustrate the utility of nAbs for rational design of therapeutics against picornaviruses such as CVB.

A Bacterially Expressed SARS-CoV-2 Receptor Binding Domain Fused With Cross-Reacting Material 197 A-Domain Elicits High Level of Neutralizing Antibodies in Mice.

The Coronavirus disease 2019 (COVID-19) pandemic presents an unprecedented public health crisis worldwide. Although several vaccines are available, the global supply of vaccines, particularly within developing countries, is inadequate, and this necessitates a need for the development of less expensive, accessible vaccine options. To this end, here, we used the Escherichia coli expression system to produce a recombinant fusion protein comprising the receptor binding domain (RBD) of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2; residues 319-541) and the fragment A domain of Cross-Reacting Material 197 (CRM197); hereafter, CRMA-RBD. We show that this CRMA-RBD fusion protein has excellent physicochemical properties and strong reactivity with COVID-19 convalescent sera and representative neutralizing antibodies (nAbs). Furthermore, compared with the use of a traditional aluminum adjuvant, we find that combining the CRMA-RBD protein with a nitrogen bisphosphonate-modified zinc-aluminum hybrid adjuvant (FH-002C-Ac) leads to stronger humoral immune responses in mice, with 4-log neutralizing antibody titers. Overall, our study highlights the value of this E. coli-expressed fusion protein as an alternative vaccine candidate strategy against COVID-19.

Structures of pseudorabies virus capsids.

Pseudorabies virus (PRV) is a major etiological agent of swine infectious diseases and is responsible for significant economic losses in the swine industry. Recent data points to human viral encephalitis caused by PRV infection, suggesting that PRV may be able to overcome the species barrier to infect humans. To date, there is no available therapeutic for PRV infection. Here, we report the near-atomic structures of the PRV A-capsid and C-capsid, and illustrate the interaction that occurs between these subunits. We show that the C-capsid portal complex is decorated with capsid-associated tegument complexes. The PRV capsid structure is highly reminiscent of other α-herpesviruses, with some additional structural features of ß- and γ-herpesviruses. These results illustrate the structure of the PRV capsid and elucidate the underlying assembly mechanism at the molecular level. This knowledge may be useful for the development of oncolytic agents or specific therapeutics against this arm of the herpesvirus family.

A stepwise docking molecular dynamics approach for simulating antibody recognition with substantial conformational changes.

Conformational changes or rearrangements are common events during inter-biomolecular recognition. Tracking these changes are essential for exploring the allosteric mechanism and it is usually achieved by molecular dynamics simulation in silico. We previously identified a broad-neutralizing antibody against H5 influenza virus, 13D4, and solved the crystal structures of the free 13D4 Fab and its complex with hemagglutinin (HA). Structural comparison of the unbound and bound 13D4 Fabs showed that the heavy chain complementarity-determining region 3 (HCDR3) undergoes a substantial conformational rearrangement when it recognizes the receptor-binding site (RBS). Here, we used molecular dynamics (MD) to simulate the conformational changes that occur during antibody recognition. We showed that neither conventional MD nor steered MD could recapitulate the loop fitting of the RBS structure contour. Consequently, to simulate these challenging conformational changes, we engaged a stepwise docking MD method that allowed for the gradual docking of the ligand to receptor. This new method recapitulates the bound shape of the HCDR3 and provides the best approximation of the shape rendered by the co-crystal structure, with an RMSD of 0.926 Å. This strategy affords a flexible MD approach for simulating complicated conformational changes that occur during molecular recognition, and helps to provide an understanding of the involved allosteric mechanism.

Engineering for an HPV 9-valent vaccine candidate using genomic constitutive over-expression and low lipopolysaccharide levels in Escherichia coli cells.

BACKGROUND: The various advantages associated with the growth properties of Escherichia coli have justified their use in the production of genetically engineered vaccines. However, endotoxin contamination, plasmid vector instability, and the requirement for antibiotic supplementation are frequent bottlenecks in the successful production of recombinant proteins that are safe for industrial-scaled applications. To overcome these drawbacks, we focused on interrupting the expression of several key genes involved in the synthesis of lipopolysaccharide (LPS), an endotoxin frequently responsible for toxicity in recombinant proteins, to eliminate endotoxin contamination and produce better recombinant proteins with E. coli. RESULTS: Of 8 potential target genes associated with LPS synthesis, we successfully constructed 7 LPS biosynthesis-defective recombinant strains to reduce the production of LPS. The endotoxin residue in the protein products from these modified E. coli strains were about two orders of magnitude lower than that produced by the wild-type strain. Further, we found that 6 loci-lpxM, lpxP, lpxL, eptA, gutQ and kdsD-were suitable for chromosomal integrated expression of HPV L1 protein. We found that a single copy of the expression cassette conferred stable expression during long-term antibiotic-free cultivation as compared with the more variable protein production from plasmid-based expression. In large-scale fermentation, we found that recombinant strains bearing 3 to 5 copies of the expression cassette had 1.5- to 2-fold higher overall expression along with lower endotoxin levels as compared with the parental ER2566 strain. Finally, we engineered and constructed 9 recombinant E. coli strains for the later production of an HPV 9-valent capsid protein with desirable purity, VLP morphology, and antigenicity. CONCLUSIONS: Reengineering the LPS synthesis loci in the E. coli ER2566 strain through chromosomal integration of expression cassettes has potential uses for the production of a 9-valent HPV vaccine candidate, with markedly reduced residual endotoxin levels. Our results offer a new strategy for recombinant E. coli strain construction, engineering, and the development of suitable recombinant protein drugs.

In situ analysis of biomolecular interactions on nano/microparticles in formulations.

Nano/microparticle-based formulations are attracting more and more attention in the drug and vaccine formulation field. Analytical methodologies are needed for analyzing the function of the protein molecules before and after their adsorption onto particles. Here, we report a fluorescence imaging-based high content analysis (HCA) for analyzing bimolecular interactions on the particle surface. Subtle conformational alteration of antigen epitope overlapping were shown to be feasible using this microplate-based HCA method. This HCA is highly amenable for automation, making it a method of choice for early formulation development as well as for stability test of products in late stage development.

A SCID mouse-human lung xenograft model of SARS-CoV-2 infection.

SARS-CoV-2 infection, which is responsible for the current COVID-19 pandemic, can cause life-threatening pneumonia, respiratory failure and even death. Characterizing SARS-CoV-2 pathogenesis in primary human target cells and tissues is crucial for developing vaccines and therapeutics. However, given the limited access to clinical samples from COVID-19 patients, there is a pressing need for in vitro/in vivo models to investigate authentic SARS-CoV-2 infection in primary human lung cells or tissues with mature structures. The present study was designed to evaluate a humanized mouse model carrying human lung xenografts for SARS-CoV-2 infection in vivo. Methods: Human fetal lung tissue surgically grafted under the dorsal skin of SCID mice were assessed for growth and development after 8 weeks. Following SARS-CoV-2 inoculation into the differentiated lung xenografts, viral replication, cell-type tropism and histopathology of SARS-CoV-2 infection, and local cytokine/chemokine expression were determined over a 6-day period. The effect of IFN-α treatment against SARS-CoV-2 infection was tested in the lung xenografts. Results: Human lung xenografts expanded and developed mature structures closely resembling normal human lung. SARS-CoV-2 replicated and spread efficiently in the lung xenografts with the epithelial cells as the main target, caused severe lung damage, and induced a robust pro-inflammatory response. IFN-α treatment effectively inhibited SARS-CoV-2 replication in the lung xenografts. Conclusions: These data support the human lung xenograft mouse model as a useful and biological relevant tool that should facilitate studies on the pathogenesis of SARS-CoV-2 lung infection and the evaluation of potential antiviral therapies.

Rational design of a multi-valent human papillomavirus vaccine by capsomere-hybrid co-assembly of virus-like particles.

The capsid of human papillomavirus (HPV) spontaneously arranges into a T = 7 icosahedral particle with 72 L1 pentameric capsomeres associating via disulfide bonds between Cys175 and Cys428. Here, we design a capsomere-hybrid virus-like particle (chVLP) to accommodate multiple types of L1 pentamers by the reciprocal assembly of single C175A and C428A L1 mutants, either of which alone encumbers L1 pentamer particle self-assembly. We show that co-assembly between any pair of C175A and C428A mutants across at least nine HPV genotypes occurs at a preferred equal molar stoichiometry, irrespective of the type or number of L1 sequences. A nine-valent chVLP vaccine-formed through the structural clustering of HPV epitopes-confers neutralization titers that are comparable with that of Gardasil 9 and elicits minor cross-neutralizing antibodies against some heterologous HPV types. These findings may pave the way for a new vaccine design that targets multiple pathogenic variants or cancer cells bearing diverse neoantigens.

Genome re-sequencing and reannotation of the Escherichia coli ER2566 strain and transcriptome sequencing under overexpression conditions.

BACKGROUND: The Escherichia coli ER2566 strain (NC_CP014268.2) was developed as a BL21 (DE3) derivative strain and had been widely used in recombinant protein expression. However, like many other current RefSeq annotations, the annotation of the ER2566 strain was incomplete, with missing gene names and miscellaneous RNAs, as well as uncorrected annotations of some pseudogenes. Here, we performed a systematic reannotation of the ER2566 genome by combining multiple annotation tools with manual revision to provide a comprehensive understanding of the E. coli ER2566 strain, and used high-throughput sequencing to explore how the strain adapted under external pressure. RESULTS: The reannotation included noteworthy corrections to all protein-coding genes, led to the exclusion of 190 hypothetical genes or pseudogenes, and resulted in the addition of 237 coding sequences and 230 miscellaneous noncoding RNAs and 2 tRNAs. In addition, we further manually examined all 194 pseudogenes in the Ref-seq annotation and directly identified 123 (63%) as coding genes. We then used whole-genome sequencing and high-throughput RNA sequencing to assess mutational adaptations under consecutive subculture or overexpression burden. Whereas no mutations were detected in response to consecutive subculture, overexpression of the human papillomavirus 16 type capsid led to the identification of a mutation (position 1,094,824 within the 3' non-coding region) positioned 19-bp away from the lacI gene in the transcribed RNA, which was not detected at the genomic level by Sanger sequencing. CONCLUSION: The ER2566 strain was used by both the general scientific community and the biotechnology industry. Reannotation of the E. coli ER2566 strain not only improved the RefSeq data but uncovered a key site that might be involved in the transcription and translation of genes encoding the lactose operon repressor. We proposed that our pipeline might offer a universal method for the reannotation of other bacterial genomes with high speed and accuracy. This study might facilitate a better understanding of gene function for the ER2566 strain under external burden and provided more clues to engineer bacteria for biotechnological applications.

Neutralization sites of human papillomavirus-6 relate to virus attachment and entry phase in viral infection.

Human papillomavirus type 6 (HPV6) is the major etiologic agent of genital warts and recurrent respiratory papillomatosis. Although the commercial HPV vaccines cover HPV6, the neutralization sites and mode for HPV6 are poorly understood. Here, we identify the HPV6 neutralization sites and discriminate the inhibition of virus attachment and entry by three potent neutralizing antibodies (nAbs), 5D3, 17D5, and 15F7. Mutagenesis assays showed that these nAbs predominantly target surface loops BC, DE, and FG of HPV6 L1. Cryo-EM structures of the HPV6 pseudovirus (PsV) and its immune complexes revealed three distinct binding modalities - full-occupation-bound to capsid, top-center-bound-, and top-rim-bound to pentamers - and illustrated a structural atlas for three classes of antibody-bound footprints that are located at center-distal ring, center, and center-proximal ring of pentamer surface for 5D3, 17D5, and 15F7, respectively. Two modes of neutralization were identified: mAb 5D3 and 17D5 block HPV PsV from attaching to the extracellular matrix (ECM) and the cell surface, whereas 15F7 allows PsV attachment but prohibits PsV from entering the cell. These findings highlight three neutralization sites of HPV6 L1 and outline two antibody-mediated neutralization mechanisms against HPV6, which will be relevant for HPV virology and antiviral inhibitor design. HighlightsMajor neutralization sites of HPV6 were mapped on the pseudovirus cryo-EM structuremAb 15F7 binds HPV6 capsid with a novel top-rim binding modality and confers a post-attachment neutralizationmAb 17D5 binds capsid in top-centre manner but unexpectedly prevents virus from attachment to cell surface.

Discovery and structural characterization of a therapeutic antibody against coxsackievirus A10.

Coxsackievirus A10 (CVA10) recently emerged as a major pathogen of hand, foot, and mouth disease and herpangina in children worldwide, and lack of a vaccine or a cure against CVA10 infections has made therapeutic antibody identification a public health priority. By targeting a local isolate, CVA10-FJ-01, we obtained a potent antibody, 2G8, against all three capsid forms of CVA10. We show that 2G8 exhibited both 100% preventive and 100% therapeutic efficacy against CVA10 infection in mice. Comparisons of the near-atomic cryo-electron microscopy structures of the three forms of CVA10 capsid and their complexes with 2G8 Fab reveal that a single Fab binds a border region across the three capsid proteins (VP1 to VP3) and explain 2G8's remarkable cross-reactivities against all three capsid forms. The atomic structures of this first neutralizing antibody of CVA10 should inform strategies for designing vaccines and therapeutics against CVA10 infections.

Identification of Strategic Residues at the Interface of Antigen-Antibody Interactions by In Silico Mutagenesis.

Structural information pertaining to antigen-antibody interactions is fundamental in immunology, and benefits structure-based vaccine design. Modeling of antigen-antibody immune complexes from co-crystal structures or molecular docking simulations provides an extensive profile of the epitope at the interface; however, the key amino acids involved in the interaction must be further clarified, often through the use of experimental mutagenesis and subsequent binding assays. Here, we describe an in silico mutagenesis method to identify key sites at antigen-antibody interfaces, using significant increase in pH-dependency energy among saturated point mutations. Through a comprehensive analysis of the crystal structures of three antigen-antibody immune complexes, we show that a cutoff value of 1 kcal/mol of increased interaction energy provides good congruency with the experimental non-binding mutations conducted in vitro. This in silico mutagenesis strategy, in association with energy calculations, may provide an efficient tool for antibody-antigen interface analyses, epitope optimization, and/or conformation prediction in structure-based vaccine design.

Identification of Broad-Genotype HPV L2 Neutralization Site for Pan-HPV Vaccine Development by a Cross-Neutralizing Antibody.

Human Papillomavirus (HPV), a non-enveloped, double-stranded DNA virus, is responsible for 5% of human cancers. The HPV capsid consists of major and minor structural proteins, L1 and L2. L1 proteins form an icosahedral shell with building blocks of the pentameric capsomere, and one L2 molecule extends outward from the central hole of the capsid. Thus, L2 is concealed within L1 and only becomes exposed when the capsid interacts with host cells. The low antigenic variation of L2 means that this protein could offer a target for the development of a pan-HPV vaccine. Toward this goal, here we describe an anti-L2 monoclonal antibody, 14H6, which broadly neutralizes at least 11 types of HPV, covering types 6, 11, 16, 18, 31, 33, 35, 45, 52, 58 and 59, in pseudovirion--based cell neutralization assay. The mAb 14H6 recognizes a minimal linear epitope located on amino acids 21 to 30 of the L2 protein. Alanine scanning mutagenesis and sequence alignment identified several conserved residues (Cys22, Lys23, Thr27, Cys28 and Pro29) that are involved in the 14H6 binding with L2. The epitope was grafted to several scaffolding proteins, including HPV16 L1 virus-like particles, HBV 149 core antigen and CRM197. The resultant chimeric constructs were expressed in Escherichia coli and purified with high efficiency. Immunization with these pan-HPV vaccine candidates elicited high titers of the L2-specific antibody in mice and conferred robust (3-log) titers of cross-genotype neutralization, including against HPV11, 16, 18, 45, 52, 58 and 59. These findings will help in the development of an L2-based, pan-HPV vaccine.