Physics-driven structural docking and protein language models accelerate antibody screening and design for broad-spectrum antiviral therapy.

Therapeutic antibodies have become one of the most influential therapeutics in modern medicine to fight against infectious pathogens, cancer, and many other diseases. However, experimental screening for highly efficacious targeting antibodies is labor-intensive and of high cost, which is exacerbated by evolving antigen targets under selective pressure such as fast-mutating viral variants. As a proof-of-concept, we developed a machine learning-assisted antibody generation pipeline that greatly accelerates the screening and re-design of immunoglobulins G (IgGs) against a broad spectrum of SARS-CoV-2 coronavirus variant strains. These viruses infect human host cells via the viral spike protein binding to the host cell receptor angiotensin-converting enzyme 2 (ACE2). Using over 1300 IgG sequences derived from convalescent patient B cells that bind with spike's receptor binding domain (RBD), we first established protein structural docking models in assessing the RBD-IgG-ACE2 interaction interfaces and predicting the virus-neutralizing activity of each IgG with a confidence score. Additionally, employing Gaussian process regression (also known as Kriging) in a latent space of an antibody language model, we predicted the landscape of IgGs' activity profiles against individual coronaviral variants of concern. With functional analyses and experimental validations, we efficiently prioritized IgG candidates for neutralizing a broad spectrum of viral variants (wildtype, Delta, and Omicron) to prevent the infection of host cells in vitro and hACE2 transgenic mice in vivo. Furthermore, the computational analyses enabled rational redesigns of selective IgG clones with single amino acid substitutions at the RBD-binding interface to improve the IgG blockade efficacy for one of the severe, therapy-resistant strains - Delta (B.1.617). Our work expedites applications of artificial intelligence in antibody screening and re-design even in low-data regimes combining protein language models and Kriging for antibody sequence analysis, activity prediction, and efficacy improvement, in synergy with physics-driven protein docking models for antibody-antigen interface structure analyses and functional optimization.

Soluble Angiotensin-Converting Enzyme 2 Protein Improves Survival and Lowers Viral Titers in Lethal Mouse Model of Severe Acute Respiratory Syndrome Coronavirus Type 2 Infection with the Delta Variant.

Severe acute respiratory syndrome coronavirus type 2 (SARS-CoV-2) utilizes angiotensin-converting enzyme 2 (ACE2) as its main receptor for cell entry. We bioengineered a soluble ACE2 protein termed ACE2 618-DDC-ABD that has increased binding to SARS-CoV-2 and prolonged duration of action. Here, we investigated the protective effect of this protein when administered intranasally to k18-hACE2 mice infected with the aggressive SARS-CoV-2 Delta variant. k18-hACE2 mice were infected with the SARS-CoV-2 Delta variant by inoculation of a lethal dose (2 × 104 PFU). ACE2 618-DDC-ABD (10 mg/kg) or PBS was administered intranasally six hours prior and 24 and 48 h post-viral inoculation. All animals in the PBS control group succumbed to the disease on day seven post-infection (0% survival), whereas, in contrast, there was only one casualty in the group that received ACE2 618-DDC-ABD (90% survival). Mice in the ACE2 618-DDC-ABD group had minimal disease as assessed using a clinical score and stable weight, and both brain and lung viral titers were markedly reduced. These findings demonstrate the efficacy of a bioengineered soluble ACE2 decoy with an extended duration of action in protecting against the aggressive Delta SARS-CoV-2 variant. Together with previous work, these findings underline the universal protective potential against current and future emerging SARS-CoV-2 variants.

Reply to Barton et al: signatures of natural selection during the Black Death.

Barton et al.1 raise several statistical concerns regarding our original analyses2 that highlight the challenge of inferring natural selection using ancient genomic data. We show here that these concerns have limited impact on our original conclusions. Specifically, we recover the same signature of enrichment for high FST values at the immune loci relative to putatively neutral sites after switching the allele frequency estimation method to a maximum likelihood approach, filtering to only consider known human variants, and down-sampling our data to the same mean coverage across sites. Furthermore, using permutations, we show that the rs2549794 variant near ERAP2 continues to emerge as the strongest candidate for selection (p = 1.2×10-5), falling below the Bonferroni-corrected significance threshold recommended by Barton et al. Importantly, the evidence for selection on ERAP2 is further supported by functional data demonstrating the impact of the ERAP2 genotype on the immune response to Y. pestis and by epidemiological data from an independent group showing that the putatively selected allele during the Black Death protects against severe respiratory infection in contemporary populations.

Modeling gastrointestinal anthrax disease.

Bacillus anthracis is a spore-forming microbe that persists in soil and causes anthrax disease. The most natural route of infection is ingestion by grazing animals. Gastrointestinal (GI) anthrax also occurs in their monogastric predators, including humans. Exposure of carcasses to oxygen triggers sporulation and contamination of the surrounding soil completing the unusual life cycle of this microbe. The pathogenesis of GI anthrax is poorly characterized. Here, we use B. anthracis carrying the virulence plasmids pXO1 and pXO2, to model gastrointestinal disease in Guinea pigs and mice. We find that spores germinate in the GI tract and precipitate disease in a dose-dependent manner. Inoculation of vegetative bacilli also results in GI anthrax. Virulence is impacted severely by the loss of capsule (pXO2-encoded) but only moderately in absence of toxins (pXO1-encoded). Nonetheless, the lack of toxins leads to reduced bacterial replication in infected hosts. B. cereus Elc4, a strain isolated from a fatal case of inhalational anthrax-like disease, was also found to cause GI anthrax. Because transmission to new hosts depends on the release of large numbers of spores in the environment, we propose that the acquisition of pXO1- and pXO2-like plasmids may promote the successful expansion of members of the Bacillus cereus sensu lato group able to cause anthrax-like disease.

Evolution of immune genes is associated with the Black Death.

Infectious diseases are among the strongest selective pressures driving human evolution1,2. This includes the single greatest mortality event in recorded history, the first outbreak of the second pandemic of plague, commonly called the Black Death, which was caused by the bacterium Yersinia pestis3. This pandemic devastated Afro-Eurasia, killing up to 30-50% of the population4. To identify loci that may have been under selection during the Black Death, we characterized genetic variation around immune-related genes from 206 ancient DNA extracts, stemming from two different European populations before, during and after the Black Death. Immune loci are strongly enriched for highly differentiated sites relative to a set of non-immune loci, suggesting positive selection. We identify 245 variants that are highly differentiated within the London dataset, four of which were replicated in an independent cohort from Denmark, and represent the strongest candidates for positive selection. The selected allele for one of these variants, rs2549794, is associated with the production of a full-length (versus truncated) ERAP2 transcript, variation in cytokine response to Y. pestis and increased ability to control intracellular Y. pestis in macrophages. Finally, we show that protective variants overlap with alleles that are today associated with increased susceptibility to autoimmune diseases, providing empirical evidence for the role played by past pandemics in shaping present-day susceptibility to disease.

Spherical nucleic acids as an infectious disease vaccine platform.

SignificanceUsing SARS-CoV-2 as a relevant case study for infectious disease, we investigate the structure-function relationships that dictate antiviral spherical nucleic acid (SNA) vaccine efficacy. We show that the SNA architecture can be rapidly employed to target COVID-19 through incorporation of the receptor-binding domain, and that the resulting vaccine potently activates human cells in vitro and mice in vivo. Furthermore, when challenged with a lethal viral infection, only mice treated with the SNA vaccine survived. Taken together, this work underscores the importance of rational vaccine design for infectious disease to yield vaccines that elicit more potent immune responses to effectively fight disease.

Glutathionylation of Yersinia pestis LcrV and Its Effects on Plague Pathogenesis.

Glutathionylation, the formation of reversible mixed disulfides between glutathione and protein cysteine residues, is a posttranslational modification previously observed for intracellular proteins of bacteria. Here we show that Yersinia pestis LcrV, a secreted protein capping the type III secretion machine, is glutathionylated at Cys273 and that this modification promotes association with host ribosomal protein S3 (RPS3), moderates Y. pestis type III effector transport and killing of macrophages, and enhances bubonic plague pathogenesis in mice and rats. Secreted LcrV was purified and analyzed by mass spectrometry to reveal glutathionylation, a modification that is abolished by the codon substitution Cys273Ala in lcrV Moreover, the lcrVC273A mutation enhanced the survival of animals in models of bubonic plague. Investigating the molecular mechanism responsible for these virulence attributes, we identified macrophage RPS3 as a ligand of LcrV, an association that is perturbed by the Cys273Ala substitution. Furthermore, macrophages infected by the lcrVC273A variant displayed accelerated apoptotic death and diminished proinflammatory cytokine release. Deletion of gshB, which encodes glutathione synthetase of Y. pestis, resulted in undetectable levels of intracellular glutathione, and we used a Y. pestis ΔgshB mutant to characterize the biochemical pathway of LcrV glutathionylation, establishing that LcrV is modified after its transport to the type III needle via disulfide bond formation with extracellular oxidized glutathione.IMPORTANCEYersinia pestis, the causative agent of plague, has killed large segments of the human population; however, the molecular bases for the extraordinary virulence attributes of this pathogen are not well understood. We show here that LcrV, the cap protein of bacterial type III secretion needles, is modified by host glutathione and that this modification contributes to the high virulence of Y. pestis in mouse and rat models for bubonic plague. These data suggest that Y. pestis exploits glutathione in host tissues to activate a virulence strategy, thereby accelerating plague pathogenesis.

Small molecule inhibitor of lipoteichoic acid synthesis is an antibiotic for Gram-positive bacteria.

The current epidemic of infections caused by antibiotic-resistant gram-positive bacteria requires the discovery of new drug targets and the development of new therapeutics. Lipoteichoic acid (LTA), a cell wall polymer of gram-positive bacteria, consists of 1,3-polyglycerol-phosphate linked to glycolipid. LTA synthase (LtaS) polymerizes polyglycerol-phosphate from phosphatidylglycerol, a reaction that is essential for the growth of gram-positive bacteria. We screened small molecule libraries for compounds inhibiting growth of Staphylococcus aureus but not of gram-negative bacteria. Compound 1771 [2-oxo-2-(5-phenyl-1,3,4-oxadiazol-2-ylamino)ethyl 2-naphtho[2,1-b]furan-1-ylacetate] blocked phosphatidylglycerol binding to LtaS and inhibited LTA synthesis in S. aureus and in Escherichia coli expressing ltaS. Compound 1771 inhibited the growth of antibiotic-resistant gram-positive bacteria and prolonged the survival of mice with lethal S. aureus challenge, validating LtaS as a target for the development of antibiotics.

Vaccine protection against Bacillus cereus-mediated respiratory anthrax-like disease in mice.

Bacillus cereus strains harboring a pXO1-like virulence plasmid cause respiratory anthrax-like disease in humans, particularly in welders. We developed mouse models for intraperitoneal as well as aerosol challenge with spores of B. cereus G9241, harboring pBCXO1 and pBC218 virulence plasmids. Compared to wild-type B. cereus G9241, spores with a deletion of the pBCXO1-carried protective antigen gene (pagA1) were severely attenuated, whereas spores with a deletion of the pBC218-carried protective antigen homologue (pagA2) were not. Anthrax vaccine adsorbed (AVA) immunization raised antibodies that bound and neutralized the pagA1-encoded protective antigen (PA1) but not the PA2 orthologue encoded by pagA2. AVA immunization protected mice against a lethal challenge with spores from B. cereus G9241 or B. cereus Elc4, a strain that had been isolated from a fatal case of anthrax-like disease. As the pathogenesis of B. cereus anthrax-like disease in mice is dependent on pagA1 and PA-neutralizing antibodies provide protection, AVA immunization may also protect humans from respiratory anthrax-like death.

Hereditary hemochromatosis restores the virulence of plague vaccine strains.

Nonpigmented Yersinia pestis (pgm) strains are defective in scavenging host iron and have been used in live-attenuated vaccines to combat plague epidemics. Recently, a Y. pestis pgm strain was isolated from a researcher with hereditary hemochromatosis who died from laboratory-acquired plague. We used hemojuvelin-knockout (Hjv(-/-)) mice to examine whether iron-storage disease restores the virulence defects of nonpigmented Y. pestis. Unlike wild-type mice, Hjv(-/-) mice developed lethal plague when challenged with Y. pestis pgm strains. Immunization of Hjv(-/-) mice with a subunit vaccine that blocks Y. pestis type III secretion generated protection against plague. Thus, individuals with hereditary hemochromatosis may be protected with subunit vaccines but should not be exposed to live-attenuated plague vaccines.

Polymorphisms in the lcrV gene of Yersinia enterocolitica and their effect on plague protective immunity.

Current efforts to develop plague vaccines focus on LcrV, a polypeptide that resides at the tip of type III secretion needles. LcrV-specific antibodies block Yersinia pestis type III injection of Yop effectors into host immune cells, thereby enabling phagocytes to kill the invading pathogen. Earlier work reported that antibodies against Y. pestis LcrV cannot block type III injection by Yersinia enterocolitica strains and suggested that lcrV polymorphisms may provide for escape from LcrV-mediated plague immunity. We show here that polyclonal or monoclonal antibodies raised against Y. pestis KIM D27 LcrV (LcrV(D27)) bind LcrV from Y. enterocolitica O:9 strain W22703 (LcrV(W22703)) or O:8 strain WA-314 (LcrV(WA-314)) but are otherwise unable to block type III injection by Y. enterocolitica strains. Replacing the lcrV gene on the pCD1 virulence plasmid of Y. pestis KIM D27 with either lcrV(W22703) or lcrV(WA-314) does not affect the ability of plague bacteria to secrete proteins via the type III pathway, to inject Yops into macrophages, or to cause lethal plague infections in mice. LcrV(D27)-specific antibodies blocked type III injection by Y. pestis expressing lcrV(W22703) or lcrV(WA-314) and protected mice against intravenous lethal plague challenge with these strains. Thus, although antibodies raised against LcrV(D27) are unable to block the type III injection of Y. enterocolitica strains, expression of lcrV(W22703) or lcrV(WA-314) in Y. pestis did not allow these strains to escape LcrV-mediated plague protective immunity in the intravenous challenge model.

Prevention of pneumonic plague in mice, rats, guinea pigs and non-human primates with clinical grade rV10, rV10-2 or F1-V vaccines.

Yersinia pestis causes plague, a disease with high mortality in humans that can be transmitted by fleabite or aerosol. A US Food and Drug Administration (FDA)-licensed plague vaccine is currently not available. Vaccine developers have focused on two subunits of Y. pestis: LcrV, a protein at the tip of type III secretion needles, and F1, the fraction 1 pilus antigen. F1-V, a hybrid generated via translational fusion of both antigens, is being developed for licensure as a plague vaccine. The rV10 vaccine is a non-toxigenic variant of LcrV lacking residues 271-300. Here we developed Current Good Manufacturing Practice (cGMP) protocols for rV10. Comparison of clinical grade rV10 with F1-V did not reveal significant differences in plague protection in mice, guinea pigs or cynomolgus macaques. We also developed cGMP protocols for rV10-2, a variant of rV10 with an altered affinity tag. Immunization with rV10-2 adsorbed to aluminum hydroxide elicited antibodies against LcrV and conferred pneumonic plague protection in mice, rats, guinea pigs, cynomolgus macaques and African Green monkeys. The data support further development of rV10-2 for FDA Investigational New Drug (IND) authorization review and clinical testing.

Plague in Guinea pigs and its prevention by subunit vaccines.

Human pneumonic plague is a devastating and transmissible disease for which a Food and Drug Administration-approved vaccine is not available. Suitable animal models may be adopted as a surrogate for human plague to fulfill regulatory requirements for vaccine efficacy testing. To develop an alternative to pneumonic plague in nonhuman primates, we explored guinea pigs as a model system. On intranasal instillation of a fully virulent strain, Yersinia pestis CO92, guinea pigs developed lethal lung infections with hemorrhagic necrosis, massive bacterial replication in the respiratory system, and blood-borne dissemination to other organ systems. Expression of the Y. pestis F1 capsule was not required for the development of pulmonary infection; however, the capsule seemed to be important for the establishment of bubonic plague. The mean lethal dose (MLD) for pneumonic plague in guinea pigs was estimated to be 1000 colony-forming units. Immunization of guinea pigs with the recombinant forms of LcrV, a protein that resides at the tip of Yersinia type III secretion needles, or F1 capsule generated robust humoral immune responses. Whereas LcrV immunization resulted in partial protection against pneumonic plague challenge with 250 MLD Y. pestis CO92, immunization with recombinant F1 did not. rV10, a vaccine variant lacking LcrV residues 271-300, elicited protection against pneumonic plague, which seemed to be based on conformational antibodies directed against LcrV.

Amino acid residues 196-225 of LcrV represent a plague protective epitope.

LcrV, a protein that resides at the tip of the type III secretion needles of Yersinia pestis, is the single most important plague protective antigen. Earlier work reported monoclonal antibody MAb 7.3, which binds a conformational epitope of LcrV and protects experimental animals against lethal plague challenge. By screening monoclonal antibodies directed against LcrV for their ability to protect immunized mice against bubonic plague challenge, we examined here the possibility of additional protective epitopes. MAb BA5 protected animals against plague, neutralized the Y. pestis type III secretion pathway and promoted opsonophagocytic clearance of bacteria in blood. LcrV residues 196-225 were necessary and sufficient for MAb BA5 binding. Compared to full-length LcrV, a variant lacking its residues 196-225 retained the ability of eliciting plague protection. These results identify LcrV residues 196-225 as a linear epitope that is recognized by the murine immune system to confer plague protection.

Yersinia pestis IS1541 transposition provides for escape from plague immunity.

Yersinia pestis is perhaps the most feared infectious agent due to its ability to cause epidemic outbreaks of plague disease in animals and humans with high mortality. Plague infections elicit strong humoral immune responses against the capsular antigen (fraction 1 [F1]) of Y. pestis, and F1-specific antibodies provide protective immunity. Here we asked whether Y. pestis generates mutations that enable bacterial escape from protective immunity and isolated a variant with an IS1541 insertion in caf1A encoding the F1 outer membrane usher. The caf1A::IS1541 insertion prevented assembly of F1 pili and provided escape from plague immunity via F1-specific antibodies without a reduction in virulence in mouse models of bubonic or pneumonic plague. F1-specific antibodies interfere with Y. pestis type III transport of effector proteins into host cells, an inhibitory effect that was overcome by the caf1A::IS1541 insertion. These findings suggest a model in which IS1541 insertion into caf1A provides for reversible changes in envelope structure, enabling Y. pestis to escape from adaptive immune responses and plague immunity.

Pneumonic plague pathogenesis and immunity in Brown Norway rats.

The Brown Norway rat was recently described as a bubonic plague model that closely mimics human disease. We therefore evaluated the Brown Norway rat as an alternative small animal model for pneumonic plague and characterized both the efficacy and potency of vaccine candidates. When infected by intranasal instillation, these rats rapidly developed fatal pneumonic plague within 2 to 4 days of infection. Plague disease was characterized by severe alveolar edema and vascular hemorrhage in the lung in addition to fulminant necrotizing pneumonia caused by massive bacterial replication and inflammation. Twenty-four hours before death, animals developed systemic disease with an apparent delayed inflammatory response. We evaluated the ability of the protective antigen, LcrV, and a mutant derivative, V10, to protect these rats from pneumonic plague. Both were highly effective vaccines because complete protection was observed at challenge doses of 7500 LD(50). Antibody analyses suggested stronger potency of V10 immune sera compared with LcrV in the passive transfer of immunity to bubonic plague, with multiple neutralizing epitopes in LcrV. Taken together, these data demonstrate the effectiveness of inhibiting type III secretion in the prevention of pneumonic plague in rats and reveal critical contributions from both the cellular and humoral immune systems. Thus, the Brown Norway rat is an appealing alternative small animal model for the study of pneumonic plague pathogenesis and immunity.

Immunization with recombinant V10 protects cynomolgus macaques from lethal pneumonic plague.

Vaccine and therapeutic strategies that prevent infections with Yersinia pestis have been sought for over a century. Immunization with live attenuated (nonpigmented) strains and immunization with subunit vaccines containing recombinant low-calcium-response V antigen (rLcrV) and recombinant F1 (rF1) antigens are considered effective in animal models. Current antiplague subunit vaccines in development for utilization in humans contain both antigens, either as equal concentrations of the two components (rF1 plus rLcrV) or as a fusion protein (rF1-rLcrV). Here, we show that immunization with either purified rLcrV (a protein at the tip of type III needles) or a variant of this protein, recombinant V10 (rV10) (lacking amino acid residues 271 to 300), alone or in combination with rF1, prevented pneumonic lesions and disease pathogenesis. In addition, passive immunization studies showed that specific antibodies of macaques immunized with rLcrV, rV10, or rF1, either alone or in combination, conferred protection against bubonic plague challenge in mice. Finally, we found that when we compared the reactivities of anti-rLcrV and anti-rV10 immune sera from cynomolgus macaques, BALB/c mice, and brown Norway rats with LcrV-derived peptides, rV10, but not rLcrV immune sera, lacked antibodies recognizing linear LcrV oligopeptides.

Yersinia pestis caf1 variants and the limits of plague vaccine protection.

Yersinia pestis, the highly virulent agent of plague, is a biological weapon. Strategies that prevent plague have been sought for centuries, and immunization with live, attenuated (nonpigmented) strains or subunit vaccines with F1 (Caf1) antigen is considered effective. We show here that immunization with live, attenuated strains generates plague-protective immunity and humoral immune responses against F1 pilus antigen and LcrV. Y. pestis variants lacking caf1 (F1 pili) are not only fully virulent in animal models of bubonic and pneumonic plague but also break through immune responses generated with live, attenuated strains or F1 subunit vaccines. In contrast, immunization with purified LcrV, a protein at the tip of type III needles, generates protective immunity against the wild-type and the fully virulent caf1 mutant strain, in agreement with the notion that LcrV can elicit vaccine protection against both types of virulent plague strains.