Maternal but not infant anti-HIV-1 neutralizing antibody response associates with enhanced transmission and infant morbidity

A significant number of infants acquire HIV-1 through their infected mother’s breast milk, primarily due to limited access to antiretrovirals. Passive immunization with neutralizing antibodies (nAbs) may prevent this transmission. Previous studies, however, have generated conflicting results about the ability of nAbs to halt mother-to-child transmission (MTCT) and their impact on infant outcomes. This study compared plasma neutralizing activity in exposed infants and the infected mothers (n = 63) against heterologous HIV-1 variants and the quasispecies present in the mother. HIV-exposed uninfected infants (HEU) (n = 42), compared to those that eventually acquired infection (n = 21), did not possess higher nAb responses against heterologous envelopes (P = 0.46) or their mothers’ variants (P = 0.45). Transmitting compared to nontransmitting mothers, however, had significantly higher plasma neutralizing activity against heterologous envelopes (P = 0.03), although these two groups did not have significant differences in their ability to neutralize autologous strains (P = 0.39). Furthermore, infants born to mothers with greater neutralizing breadth and potency were significantly more likely to have a serious adverse event (P = 0.03). These results imply that preexisting anti-HIV-1 neutralizing activity does not prevent breast milk transmission. Additionally, high maternal neutralizing breadth and potency may adversely influence both the frequency of breast milk transmission and subsequent infant morbidity. IMPORTANCE Passive immunization trials are under way to understand if preexisting antibodies can decrease mother-to-child HIV-1 transmission and improve infant outcomes. We examined the influence of preexisting maternal and infant neutralizing activity on transmission and infant morbidity in a breastfeeding mother-infant cohort. Neutralization was examined against both the exposure strains circulating in the infected mothers and a standardized reference panel previously used to estimate breadth. HIV-exposed uninfected infants did not possess a broader and more potent response against both the exposure and heterologous strains compared to infants that acquired infection. Transmitting, compared to nontransmitting, mothers had significantly higher neutralization breadth and potency but similar responses against autologous variants. Infants born to mothers with higher neutralization responses were more likely to have a serious adverse event. Our results suggest that preexisting antibodies do not protect against breast milk HIV-1 acquisition and may have negative consequences for the baby.

The BAN Study was approved by the Malawi National Health Science Research Committee, the institutional review boards at the University of the North Carolina, the U.S. Centers for Disease Control and Prevention, and Boston University. All women provided written informed consent for themselves as well as on behalf of their infants. Buffy coats from anonymized uninfected donor volunteers were obtained from the Kraft Family Blood Donor Center at the Dana Farber Cancer Center after written informed consent. Mother and infant plasma samples were acquired from the control group of the BAN Study. The BAN Study compared a maternal combination antiretroviral therapy (cART) regimen, infant nevirapine, or no therapy after the peripartum period in HIV-1 breast milk transmission frequency (ClinicalTrials.gov no. {“type”:”clinical-trial”,”attrs”:{“text”:”NCT00164736″,”term_id”:”NCT00164736″}}NCT00164736). All mother-infant pairs examined in this study were treated peripartum with single-dose oral nevirapine followed by zidovudine-lamivudine (ZDV-3TC) for 7 days postpartum. This brief peripartum treatment in the control arm was deemed ineffective to prevent subsequent breast milk HIV-1 transmission, and it was deemed likely to have an insignificant impact on the maternal Env quasispecies or antibody repertoire. An infant negative-DNA PCR at birth and 14 days postpartum was required for enrollment to rule out intrauterine and intrapartum transmission. Mothers were instructed to exclusively breastfeed for 24 weeks, and mother-infant pairs were followed and tested with sample collection at regular intervals for 48 weeks (42). An infant was deemed as having acquired infection through breast milk when HIV-1 RNA was detected in a follow-up plasma sample and the previously collected sample was negative. All infant samples examined in this study were obtained prior to the documented HIV-1 acquisition. Each infant that eventually acquired infection and the corresponding transmitting mother were matched to two mother-infant pairs with no documented transmission based on maternal plasma virus level, maternal CD4+ T-cell counts, and days postpartum to sample collection. The following MAbs were obtained from the National Institutes of Health (NIH) AIDS Reagent Program: VRC01, NIH-45-46, 10E8, 3BNC117, PG9, PG16, PGT121, PGT128, 10-1074, 4E10, 2F5, 2G12, B12, and A32. For maternal and infant plasma samples collected within the first 7 days after birth (maternal, 99, 1844, and 2315; infant, 99 and 2315), IgG was isolated using the Melon gel IgG spin purification kit (Thermo Scientific, Pierce Biotechnology) according to the manufacturer’s instructions. The concentration of collected IgG was measured on the NanoDrop 2000 spectrophotometer (Thermo Scientific) and stored at 4°C for downstream use. An equivalent amount of isolated IgG to that present in a 1:50 plasma dilution was used in the subsequent neutralization assays. Human epithelial kidney HEK293T cells and TZM-bl cells were acquired from the NIH AIDS Reagent Program. Cells were maintained in Dulbecco’s modified Eagle medium (DMEM) with 10% fetal bovine serum (FBS) (Thermo Fisher Scientific), 2 mM l-glutamine, 100 U of penicillin per ml, and 100 μg of streptomycin per ml. Peripheral blood mononuclear cells (PBMCs) were purified using the Ficoll-Hypaque method from at least 3 separate HIV-seronegative donors and propagated in RPMI 1640 containing 10% FBS, 100 μg/ml penicillin-streptomycin, 5 μg/ml of phytohemagglutinin (Sigma), and interleukin-2 for 4 days prior to infection. Expression plasmids for the 12 rev-env cassettes selected as global reference strains were acquired from the NIH AIDS Reagent Program. These 12 Envs were previously deemed to best represent the spectrum of neutralizing activity of a larger Env panel (43). Env amplicons were generated using PCR with primers IR delta ecto (5′-AAGCCTCCTACTATCATTAT) and envb3 out (5′-TTGCTACTTGTGATTGCTCCATGT) under previously described conditions (85). The QIAamp viral RNA isolation kit (Qiagen) was used to isolate RNA from maternal pretransmission plasma that was also used for the neutralization assessments. Envs were amplified using single-genome amplification (SGA) or bulk PCR under previously described conditions (86). All SGA-amplified amplicons or minimum of 3 independent bulk PCRs were pooled to generate a library of maternal Envs (Table S3). Maternal Env pools and reference Envs were inserted into a subtype C T/F plasmid, pZM247Fv2 (87), and NL4-3 ({“type”:”entrez-nucleotide”,”attrs”:{“text”:”AF324493″,”term_id”:”296556482″,”term_text”:”AF324493″}}AF324493) plasmid, respectively, using yeast gap repair homologous recombination with minor modifications from previously described methods (88, 89). Virus stocks were generated from HEK293T transfections as described previously (88). Briefly, HEK293T cells were cotransfected with a cytomegalovirus–NL4-3–long terminal repeat→Gag4 (CMV-NL4-3-LTR→Gag4) plasmid and a recombinant NL4-3 plasmid incorporating a reference Env or CMV-Q-23-LTR→Gag4 plasmid and plasmid pools incorporating maternal Envs. The supernatant was harvested 48 h posttransfection, filtered through a 0.45-μm-pore filter, and stored at −80°C. The 293T virus was passaged on PBMCs for a maximum of 7 days. At harvest, supernatants were centrifuged, filtered through a 0.45-μm-pore filter to remove cellular debris, and stored at −80°C. Titers of virus stocks were determined on TZM-bl cells in the presence of 10 μg/ml DEAE-dextran. A replication-competent virus could not be generated for reference panel subtype C Env CE0217, and thus the heterologous Env panel used in this study consisted of 11 variants. All maternal and infant plasma samples were heat inactivated for 1 h at 56°C. All neutralization assays were performed in duplicate or triplicate a minimum of 2 independent times using TZM-bl cells as described previously (33). Briefly, neutralization of viruses incorporating maternal Envs was tested against 1:50 maternal or infant plasma and 2-fold serial dilutions. Neutralization of viruses incorporating a reference Env from the global panel was tested against either a 1:50 plasma dilution, an equivalent amount of isolated IgG to that present in a 1:50 plasma dilution, or 50 μg/ml MAb. A NL4-3 Env-deleted vesicular stomatitis virus G protein Env pseudotype was also used as a negative control in panel neutralizations. Virus was incubated with antibody, heat-inactivated plasma, or growth medium alone in a total volume of 50 μl for 1 h at 37°C, and approximately 1E5 TZM-bl cells with 10 μg/ml DEAE-dextran was added to each well after this incubation. After 48 h, infection levels were determined using Galacto-Light Plus (Applied Biosystems, Foster City, CA). Differences between relative light units (RLU) in the presence of antibody or plasma and growth medium alone were calculated as the percentage of neutralization. Background RLU in the TZM-bl cells alone were subtracted from all wells. Neutralization responses against the global reference Env panel for each plasma sample and MAb were summarized by two different but related estimates. The first estimate, termed “breadth-potency” (BP), incorporated both the amount of neutralization against an Env at either a 1:50 plasma dilution or 50 µg/ml MAb concentration and responses against the entire 11 Env panel. BP was estimated by averaging the percentage of neutralization across all 11 Envs, and in this calculation, an Env-plasma/MAb combination that yielded a negative percentage of neutralization was assigned a value of 0. This mean was log2 transformed so the score ranged from 0 to 1, with 0 representing no neutralization and 1 being 100% neutralization against all Envs, as shown in equation 1: The second estimate, termed “breadth,” was defined as the percentage of Env variants neutralized at >50% at the highest tested plasma/MAb concentration. As opposed to BP, breadth did not incorporate the observed degree of neutralization against an Env. For autologous variants, an IC50 was calculated as the dilution that gave 50% inhibition. Cases were assigned an IC50 of 25 (half of the highest tested plasma dilution) when 50% inhibition was not observed. AUC was also estimated because IC50 could not be estimated for a large number of samples (47). Differences in frequencies among two groups were examined using a two-sample test of proportions. Differences among groups were analyzed using the Wilcoxon matched-pair test, and for these comparisons, average values were used for the 2 controls matched to each case. Linear regression models were also fit with generalized estimating equations (GEEs) because averaging estimates from the 2 matched controls for each case is not always deemed ideal. All measures were arcsine transformed for better fit in the GEE models. We considered nutrition supplementation, infant birth weight, maternal age, infant death, and serious adverse events (SAEs) in the infant as potential confounders of the relationship between AI and HEU infants and between TMs and NTMs and each of the scores of interest (BP, breadth, and AUC). Results were similar between the Wilcoxon matched-pair test and linear regression models; therefore, only P values from the latter are reported. Spearman’s rank tests were used to examine correlation among all continuous variables. BP and breadth scores were also calculated for MAbs using published IC50s against the global reference Env panel variants. The IC50s of an antibody against a specific reference Env was obtained from the Los Alamos database (http://hiv.lanl.gov/catnap) (90). This MAb BP-IC50 score was generated using equation 2: In cases, where an IC50 could not be estimated because 50% neutralization could not be achieved, IC50 was set as the highest tested antibody concentration. The BP-IC50 approaches 1 for the potent antibodies that have low IC50 against the majority of virus variants. The BP-IC50 is 0 for MAbs that demonstrate no neutralization capacity. Breadth was defined as the percentage of reference Envs neutralized with an IC50 of <25 μg/ml. Clinical adverse events for these infants were graded by the BAN Study investigators prior to our sample evaluations and according to toxicity tables from the Division of AIDS at the National Institute of Allergy and Infectious Diseases (NIAID) (42). We used Cox proportional hazard models to study the risk of grade 4 SAE or death as a function of BP score and adjusted for the matching in the data by using a robust estimate of the standard errors. For this analysis, BP scores were dichotomized as high (BP score ≥ cohort median) versus low (BP score < cohort median). Nutrition, maternal age, and infant birth weight were considered potential confounders in this model. We also considered an analysis stratified by HIV status of the infant. The BH correction was used to control for possible inflation of the type I error rate due to multiple testing. The adjusted P value with BH correction is presented for instances of multiple comparisons where the unadjusted P value is <0.05. Heat maps were generated using the Los Alamos HIV sequence database heat map tool (https://www.hiv.lanl.gov/). All heat maps used hierarchical clustering with the Euclidean distance method. Bootstraps were generated using standard procedure in the Los Alamos tool.

N/A

• Dima Health