A comparison of antibody-antigen complex sequence-to-structure prediction methods and their systematic biases.

The ability to accurately predict antibody-antigen complex structures from their sequences could greatly advance our understanding of the immune system and would aid in the development of novel antibody therapeutics. There have been considerable recent advancements in predicting protein-protein interactions (PPIs) fueled by progress in machine learning (ML). To understand the current state of the field, we compare six representative methods for predicting antibody-antigen complexes from sequence, including two deep learning approaches trained to predict PPIs in general (AlphaFold-Multimer and RoseTTAFold), two composite methods that initially predict antibody and antigen structures separately and dock them (using antibody-mode ClusPro), local refinement in Rosetta (SnugDock) of globally docked poses from ClusPro, and a pipeline combining homology modeling with rigid-body docking informed by ML-based epitope and paratope prediction (AbAdapt). We find that AlphaFold-Multimer outperformed other methods, although the absolute performance leaves considerable room for improvement. AlphaFold-Multimer models of lower quality display significant structural biases at the level of tertiary motifs (TERMs) toward having fewer structural matches in non-antibody-containing structures from the Protein Data Bank (PDB). Specifically, better models exhibit more common PDB-like TERMs at the antibody-antigen interface than worse ones. Importantly, the clear relationship between performance and the commonness of interfacial TERMs suggests that the scarcity of interfacial geometry data in the structural database may currently limit the application of ML to the prediction of antibody-antigen interactions.

Intranasal M2SR (M2-deficient Single Replication) Influenza Vaccine Induces Broadly Reactive Mucosal Antibody Production in Adults.

Intranasal M2SR (M2-deficient Single Replication influenza virus) vaccine induces robust immune responses in animal models and human subjects. A high-throughput multiplexed platform was used to analyze hemagglutinin-specific mucosal antibody responses in adults after a single dose of H3N2 M2SR. Nasal swab specimens were analyzed for total and hemagglutinin-specific IgA. Significant, dose-dependent increases in mucosal antibody responses to vaccine-matched and drifted H3N2 hemagglutinin were observed in M2SR vaccinated subjects regardless of baseline serum and mucosal immune status. These data suggest that M2SR induces broadly cross-reactive mucosal immune responses which may provide better protection against drifted and newly emerging influenza strains.

Leveraging permutation testing to assess confidence in positive-unlabeled learning applied to high-dimensional biological datasets.

BACKGROUND: Compared to traditional supervised machine learning approaches employing fully labeled samples, positive-unlabeled (PU) learning techniques aim to classify "unlabeled" samples based on a smaller proportion of known positive examples. This more challenging modeling goal reflects many real-world scenarios in which negative examples are not available-posing direct challenges to defining prediction accuracy and robustness. While several studies have evaluated predictions learned from only definitive positive examples, few have investigated whether correct classification of a high proportion of known positives (KP) samples from among unlabeled samples can act as a surrogate to indicate model quality. RESULTS: In this study, we report a novel methodology combining multiple established PU learning-based strategies with permutation testing to evaluate the potential of KP samples to accurately classify unlabeled samples without using "ground truth" positive and negative labels for validation. Multivariate synthetic and real-world high-dimensional benchmark datasets were employed to demonstrate the suitability of the proposed pipeline to provide evidence of model robustness across varied underlying ground truth class label compositions among the unlabeled set and with different proportions of KP examples. Comparisons between model performance with actual and permuted labels could be used to distinguish reliable from unreliable models. CONCLUSIONS: As in fully supervised machine learning, permutation testing offers a means to set a baseline "no-information rate" benchmark in the context of semi-supervised PU learning inference tasks-providing a standard against which model performance can be compared.

Characteristics and Functions of Infection-enhancing Antibodies to the N-terminal Domain of SARS-CoV-2.

Background: Fcγ-receptor (FcγR)-independent enhancement of SARS-CoV-2 infection mediated by N-terminal domain (NTD)-binding monoclonal antibodies (mAbs) has been observed in vitro, but the functional significance of these antibodies in vivo is less clear. Methods: We characterized 1,213 SARS-CoV-2 spike (S)-binding mAbs derived from COVID-19 convalescent patients for binding specificity to the SARS-CoV-2 S protein, VH germ-line usage, and affinity maturation. Infection enhancement in a vesicular stomatitis virus (VSV)-SARS-CoV-2 S pseudovirus (PV) assay was characterized in respiratory and intestinal epithelial cell lines, and against SARS-CoV-2 variants of concern (VOC). Proteomic deconvolution of the serum antibody repertoire was used to determine functional attributes of secreted NTD-binding mAbs. Results: We identified 72/1213 (5.9%) mAbs that enhanced SARS-CoV-2 infection in a PV assay. The majority (68%) of these mAbs recognized the NTD, were identified in patients with mild and severe disease, and persisted for at least 5 months post-infection. Infection enhancement by NTD-binding mAbs was not observed in intestinal and respiratory epithelial cell lines and was diminished or lost against SARS-CoV-2 VOC. Proteomic deconvolution of the serum antibody repertoire from 2 of the convalescent patients identified, for the first time, NTD-binding, infection-enhancing mAbs among the circulating immunoglobulins directly isolated from serum. Functional analysis of these mAbs demonstrated robust activation of FcγRIIIa associated with antibody binding to recombinant S proteins. Conclusions: Functionally active NTD-specific mAbs arise frequently during natural infection and can last as major serum clonotypes during convalescence. These antibodies display functional attributes that include FcγR activation, and may be selected against by mutations in NTD associated with SARS-CoV-2 VOC.

HIV-related immune activation attenuates polyfunctional IgG and memory B-cell responses to Tdap immunization during pregnancy.

BACKGROUND: Maternal pertussis vaccination with Tdap vaccine is recommended to protect newborns from severe postnatal infection. HIV-exposed uninfected (HEU) infants have a higher incidence of pertussis infection and may particularly benefit from maternal immunization. The impact of HIV infection on the quality of IgG and memory B cell (MBC) responses to Tdap vaccination in pregnant women (PW) living with HIV (PWH) is unknown. METHODS: In this observational study, humoral immune responses to Tdap vaccination, including IgG levels, Fc-dependent effector functions, and MBC frequencies, were measured before and after vaccination in 40 PWH and 42 HIV-uninfected PW. Placental transfer of IgG and avidity were assessed in cord blood (CB). Soluble and cellular immune activation markers were quantified at baseline. FINDINGS: One month after vaccination, PWH had lower frequencies of MBC compared with HIV-uninfected PW. At delivery, PWH had attenuated pertussis-specific IgG levels and Fc-dependent effector functions. Reduced levels of maternal vaccine polyfunctional IgG and IgG avidity were transferred to HEU as compared to HIV-unexposed newborns. After adjustment with ethnicity, maternal antibody levels and gestational age at vaccination, HIV infection was independently associated with decreased levels of PT specific-IgG in CB. Both maternal and neonatal pertussis-specific IgG responses as well as PT-specific IgG avidity were inversely correlated with maternal sCD14 levels before vaccination among PWH. INTERPRETATION: Maternal HIV infection is associated with attenuated humoral immune responses to Tdap vaccination that correlate with sCD14. Suboptimal transfer of maternal immunity may further increase the risk of severe pertussis infection in HEU infants. FUNDING: This work was supported by IRIS Fund managed by the Foundation Roi Baudouin [2017J1820690206902], Association Vésale pour la Recherche Médicale and the Medical Council of CHU Saint-Pierre and has been funded in part with Federal funds from the National Institute of Allergy and Infectious Diseases, National Institutes of Health, US Department of Health and Human Services, under Award No. U19AI145825. N.D. is a clinical researcher and A.M. is Research Director at the Fonds de la Recherche Scientifique (F.R.S.-FNRS), Belgium. M.E.A. was partially supported by NIHNIAID1U19AI14825. This article is published with the support of the Fondation Universitaire of Belgium.

Safety and pharmacokinetics of VRC07-523LS administered via different routes and doses (HVTN 127/HPTN 087): A Phase I randomized clinical trial.

BACKGROUND: Broadly neutralizing antibodies (bnAbs) are a promising approach for HIV-1 prevention. In the Antibody Mediated Prevention (AMP) trials, a CD4-binding site targeting bnAb, VRC01, administered intravenously (IV), demonstrated 75% prevention efficacy against highly neutralization-sensitive viruses but was ineffective against less sensitive viruses. VRC07-523LS is a next-generation bnAb targeting the CD4-binding site and was engineered for increased neutralization breadth and half-life. We conducted a multicenter, randomized, partially blinded Phase I clinical trial to evaluate the safety and serum concentrations of VRC07-523LS, administered in multiple doses and routes to healthy adults without HIV. METHODS AND FINDINGS: Participants were recruited between 2 February 2018 and 9 October 2018. A total of 124 participants were randomized to receive 5 VRC07-523LS administrations via IV (T1: 2.5 mg/kg, T2: 5 mg/kg, T3: 20 mg/kg), subcutaneous (SC) (T4: 2.5 mg/kg, T5: 5 mg/kg), or intramuscular (IM) (T6: 2.5 mg/kg or P6: placebo) routes at 4-month intervals. Participants and site staff were blinded to VRC07-523LS versus placebo for the IM group, while all other doses and routes were open-label. Safety data were collected for 144 weeks following the first administration. VRC07-523LS serum concentrations were measured by ELISA through Day 112 in all participants and by binding antibody multiplex assay (BAMA) thereafter in 60 participants (10 per treatment group) through Day 784. Compartmental population pharmacokinetic (PK) analyses were conducted to evaluate the VRC07-523LS serum PK. Neutralization activity was measured in a TZM-bl assay and antidrug antibodies (ADAs) were assayed using a tiered bridging assay testing strategy. Injections and infusions were well tolerated, with mild pain or tenderness reported commonly in the SC and IM groups, and mild to moderate erythema or induration reported commonly in the SC groups. Infusion reactions were reported in 3 of 20 participants in the 20 mg/kg IV group. Peak geometric mean (GM) concentrations (95% confidence intervals [95% CIs]) following the first administration were 29.0 µg/mL (25.2, 33.4), 58.5 µg/mL (49.4, 69.3), and 257.2 µg/mL (127.5, 518.9) in T1-T3 with IV dosing; 10.8 µg/mL (8.8, 13.3) and 22.8 µg/mL (20.1, 25.9) in T4-T5 with SC dosing; and 16.4 µg/mL (14.7, 18.2) in T6 with IM dosing. Trough GM (95% CIs) concentrations immediately prior to the second administration were 3.4 µg/mL (2.5, 4.6), 6.5 µg/mL (5.6, 7.5), and 27.2 µg/mL (23.9, 31.0) with IV dosing; 0.97 µg/mL (0.65, 1.4) and 3.1 µg/mL (2.2, 4.3) with SC dosing, and 2.6 µg/mL (2.05, 3.31) with IM dosing. Peak VRC07-523LS serum concentrations increased linearly with the administered dose. At a given dose, peak and trough concentrations, as well as serum neutralization titers, were highest in the IV groups, reflecting the lower bioavailability following SC and IM administration. A single participant was found to have low titer ADA at a lone time point. VRC07-523LS has an estimated mean half-life of 42 days across all doses and routes (95% CI: 40.5, 43.5), over twice as long as VRC01 (15 days). CONCLUSIONS: VRC07-523LS was safe and well tolerated across a range of doses and routes and is a promising long-acting bnAb for inclusion in HIV-1 prevention regimens. TRIAL REGISTRATION: ClinicalTrials.gov/ NCT03387150 (posted on 21 December 2017).

Nasal and pharyngeal mucosal immunity to poliovirus in children following routine immunization with inactivated polio vaccine (IPV) in the United States of America.

BACKGROUND: Although polioviruses (PVs) replicate in lymphoid tissue of both the pharynx and ileum, research on polio vaccine-induced mucosal immunity has predominantly focused on intestinal neutralizing and binding antibody levels measured in stool. METHODS: To investigate the extent to which routine immunization with intramuscularly injected inactivated polio vaccine (IPV) may induce nasal and pharyngeal mucosal immunity, we measured PV type-specific neutralization and immunoglobulin (Ig) G, IgA, and IgM levels in nasal secretions, adenoid cell supernatants, and sera collected from 12 children, aged 2 to 5 years, undergoing planned adenoidectomies. All participants were routinely immunized with IPV and had no known contact with live PVs. RESULTS: PV-specific mucosal neutralization was detected in nasal and adenoid samples, mostly from children who had previously received four IPV doses. Across the three PV serotypes, both nasal (Spearman's rho ≥ 0.87, p≤0.0003 for all) and adenoid (Spearman's rho ≥0.57, p≤0.05 for all) neutralization titers correlated with serum neutralization titers. In this small study sample, there was insufficient evidence to determine which Ig isotype(s) was correlated with neutralization. CONCLUSIONS: Our findings provide policy-relevant evidence that routine immunization with IPV may induce nasal and pharyngeal mucosal immunity. The observed correlations of nasal and pharyngeal mucosal neutralization with serum neutralization contrast with previous observations of distinct intestinal and serum responses to PV vaccines. Further research is warranted to determine which antibody isotype(s) correlate with polio vaccine-induced nasal and pharyngeal mucosal neutralizing activity and to understand the differences from intestinal mucosal immunity.

Hierarchal single-cell lineage tracing reveals differential fate commitment of CD8 T-cell clones in response to acute infection.

Generating balanced populations of CD8 effector and memory T cells is necessary for immediate and durable immunity to infections and cancer. Yet, a definitive understanding of CD8 differentiation remains unclear. We used CARLIN, a processive lineage recording mouse model with single-cell RNA-seq and TCR-seq to track endogenous antigen-specific CD8 T cells during acute viral infection. We identified a diverse repertoire of expanded T-cell clones represented by seven transcriptional states. TCR enrichment analysis revealed differential memory- or effector-fate biases within clonal populations. Shared Vb segments and amino acid motifs were found within biased categories despite high TCR diversity. Using single-cell CARLIN barcode-seq we tracked multi-generational clones and found that unlike unbiased or memory-biased clones, which stably retain their fate profiles, effector-biased clones could adopt memory- or effector-bias within subclones. Collectively, our study demonstrates that a heterogenous T-cell repertoire specific for a shared antigen is composed of clones with distinct TCR-intrinsic fate-biases.

Selection of positive controls and their impact on anti-drug antibody assay performance.

Development of assays to reliably identify and characterize anti-drug antibodies (ADAs) depends on positive control anti-idiotype (anti-id) reagents, which are used to demonstrate that the standards recommended by regulatory authorities are met. This work employs a set of therapeutic antibodies under clinical development and their corresponding anti-ids to investigate how different positive control reagent properties impact ADA assay development. Positive controls exhibited different response profiles and apparent assay analytical sensitivity values depending on assay format. Neither anti-id affinity for drug, nor sensitivity in direct immunoassays related to sensitivity in ADA assays. Anti-ids were differentially able to detect damage to drug conjugates used in bridging assays and were differentially drug tolerant. These parameters also failed to relate to assay sensitivity, further complicating selection of anti-ids for use in ADA assay development based on functional characteristics. Given this variability among anti-ids, alternative controls that could be employed across multiple antibody drugs were investigated as a more uniform means to define ADA detection sensitivity across drug products and assay protocols, which could help better relate assay results to clinical risks of ADA responses. Overall, this study highlights the importance of positive control selection to reliable detection and clinical interpretation of the presence and magnitude of ADA responses.

Hybrid Immunity Overcomes Defective Immune Response to COVID-19 Vaccination in Kidney Transplant Recipients.

Introduction: Comorbidities and immunosuppressive therapies are associated with reduced immune responses to primary COVID-19 mRNA vaccination in kidney transplant recipients (KTRs). In healthy individuals, prior SARS-COV-2 infection is associated with increased vaccine responses, a phenotype called hybrid immunity. In this study, we explored the potential influence of immune suppression on hybrid immunity in KTRs. Methods: Eighty-two KTRs, including 59 SARS-CoV-2-naïve (naïve KTRs [N-KTRs]) and 23 SARS-CoV-2-experienced (experienced KTRs [E-KTRs]) patients, were prospectively studied and compared to 106 healthy controls (HCs), including 40 SARS-CoV-2-naïve (N-HCs) and 66 SARS-CoV-2-experienced (E-HCs) subjects. Polyfunctional antibody and T cell responses were measured following 2 doses of BNT162b2 mRNA vaccine. Associations between vaccine responses and clinical characteristics were studied by univariate and multivariate analyses. Results: In naïve KTRs, vaccine responses were markedly lower than in HCs and were correlated with older age, more recent transplantation, kidney retransplantation after graft failure, arterial hypertension, and treatment with mycophenolate mofetil (MMF). In contrast, vaccine responses of E-KTRs were similar to those of HCs and were associated with time between transplantation and vaccination, but not with the other risk factors associated with low vaccine responses in naïve KTRs. Conclusion: In conclusion, hybrid immunity overcomes immune suppression and provides potent humoral and cellular immunity to SARS-CoV-2 in KTRs.

Sex- and species-associated differences in complement-mediated immunity in humans and rhesus macaques.

The complement system can be viewed as a "moderator" of innate immunity, "instructor" of humoral immunity, and "regulator" of adaptive immunity. While sex is known to affect humoral and cellular immune systems, its impact on complement in humans and rhesus macaques, a commonly used non-human primate model system, has not been well studied. To address this knowledge gap, we analyzed serum samples from 90 humans and 72 rhesus macaques for the abundance and activity of the complement system components. While sequences of cascade proteins were highly conserved, dramatically different levels were observed between species. Whereas the low levels detected in rhesus samples raised questions about the suitability of the test for use with macaque samples, differences in levels of complement proteins were observed in male and female humans. Levels of total and antibody-dependent deposition of C1q and C3b on a glycosylated antigen differed between humans and rhesus, suggesting differential recognition of glycans and balance between classical and alternative activation pathways. Functional differences in complement-mediated lysis of antibody-sensitized cells were observed in multiple assays and showed that human females frequently exhibited higher lytic activity than human males or rhesus macaques, which typically did not exhibit such sex-associated differences. Other differences between species and sexes were observed in more narrow contexts-for only certain antibodies, antigens, or assays. Collectively, these results expand knowledge of sex-associated differences in the complement system in humans, identifying differences absent from rhesus macaques.IMPORTANCEThe complement system is a critical part of host defense to many bacterial, fungal, and viral infections. In parallel, rich epidemiological, clinical, and biomedical research evidence demonstrates that sex is an important biological variable in immunity, and many sex-specific differences in immune system are intimately tied with disease outcomes. This study focuses on the intersection of these two factors to define the impact of sex on complement pathway components and activities. This work expands our knowledge of sex-associated differences in the complement system in humans and also identifies the differences that appear to be absent in rhesus macaques, a popular non-human primate model. Whereas differences between species suggest potential limitations in the ability of macaque model to recapitulate human biology, knowledge of sex-based differences in humans has the potential to inform clinical research and practice.

Effector functions are required for broad and potent protection of neonatal mice with antibodies targeting HSV glycoprotein D.

Multiple failed herpes simplex virus (HSV) vaccine candidates induce robust neutralizing antibody (Ab) responses in clinical trials, raising the hypothesis that Fc-domain-dependent effector functions may be critical for protection. While neonatal HSV (nHSV) infection results in mortality and lifelong neurological morbidity in humans, it is uncommon among neonates with a seropositive birthing parent, supporting the hypothesis that Ab-based therapeutics could protect neonates from HSV. We therefore investigated the mechanisms of monoclonal Ab (mAb)-mediated protection in a mouse model of nHSV infection. For a panel of glycoprotein D (gD)-specific mAbs, neutralization and effector functions contributed to nHSV-1 protection. In contrast, effector functions alone were sufficient to protect against nHSV-2, exposing a functional dichotomy between virus types consistent with vaccine trial results. Effector functions are therefore crucial for protection by these gD-specific mAbs, informing effective Ab and vaccine design and demonstrating the potential of polyfunctional Abs as therapeutics for nHSV infections.

A Randomised Clinical Trial of the Safety and Pharmacokinetics of VRC07-523LS Administered via Different Routes and Doses (HVTN 127/HPTN 087).

Background: Broadly neutralizing antibodies (bnAbs) are a promising approach for HIV-1 prevention. In the only bnAb HIV prevention efficacy studies to date, the Antibody Mediated Prevention (AMP) trials, a CD4-binding site targeting bnAb, VRC01, administered intravenously (IV), demonstrated 75% prevention efficacy against highly neutralization-sensitive viruses but was ineffective against less sensitive viruses. Greater efficacy is required before passively administered bnAbs become a viable option for HIV prevention; furthermore subcutaneous (SC) or intramuscular (IM) administration may be preferred. VRC07-523LS is a next-generation bnAb targeting the CD4-binding site and was engineered for increased neutralization breadth and half-life. Methods: Participants were recruited between 02 February 2018 and 09 October 2018. 124 healthy participants without HIV were randomized to receive five VRC07-523LS administrations via IV (T1: 2.5 mg/kg, T2: 5 mg/kg, T3: 20 mg/kg), SC (T4: 2.5 mg/kg, T5: 5 mg/kg) or IM (T6: 2.5 mg/kg or P6: placebo) routes at four-month intervals. Safety data were collected for 144 weeks following the first administration. VRC07-523LS serum concentrations were measured by ELISA after the first dose through Day 112 in all participants and by binding antibody multiplex assay (BAMA) thereafter in 60 participants (10 per treatment group) through Day 784. Compartmental population pharmacokinetic (PK) analyses were conducted to evaluate the VRC07-523LS serum pharmacokinetics. Neutralization activity was measured in a TZM-bl assay and anti-drug antibodies (ADA) were assayed using a tiered bridging assay testing strategy. Results: Injections were well-tolerated, with mild pain or tenderness reported commonly in the SC and IM groups, and mild to moderate erythema or induration reported commonly in the SC groups. Infusions were generally well-tolerated, with infusion reactions reported in 3 of 20 participants in the 20 mg/kg IV group. Peak geometric mean (GM) concentrations (95% confidence intervals) following the first administration were 29.0 µg/mL (25.2, 33.4), 58.5 µg/mL (49.4, 69.3), and 257.2 µg/mL (127.5, 518.9) in T1-T3 with IV dosing; 10.8 µg/mL (8.8, 13.3) and 22.8 µg/mL (20.1, 25.9) in T4-T5 with SC dosing; and 16.4 µg/mL (14.7, 18.2) in T6 with IM dosing. Trough GM concentrations immediately prior to the second administration were 3.4 µg/mL (2.5, 4.6), 6.5 µg/mL (5.6, 7.5), and 27.2 µg/mL (23.9, 31.0) with IV dosing; 0.97 µg/mL (0.65, 1.4) and 3.1 µg/mL (2.2, 4.3) with SC dosing, and 2.6 µg/mL (2.05, 3.31) with IM dosing. Peak VRC07-523LS serum concentrations increased linearly with the administered dose. At a given dose, peak and trough concentrations, as well as serum neutralization titres, were highest in the IV groups, reflecting the lower bioavailability following SC and IM administration. A single participant was found to have low titre ADA at a lone timepoint. VRC07-523LS has an estimated mean half-life of 42 days (95% CI: 40.5, 43.5), approximately twice as long as VRC01. Conclusions: VRC07-523LS was safe and well-tolerated across a range of doses and routes and is a promising long-acting bnAb for inclusion in HIV-1 prevention regimens.