Menu
Menu
Menu
Menu
Background: Access to antiretroviral therapy (ART) during pregnancy and breastfeeding for mothers with HIV has resulted in fewer children acquiring HIV peri- and postnatally, resulting in an increase in the number of children who are exposed to the virus but are not infected (HEU). HEU infants have an increased likelihood of childhood infections and adverse growth outcomes, as well as increased mortality compared to their HIV-unexposed (HUU) peers. We explored potential differences in the gut microbiota in a cohort of 272 Nigerian infants born to HIV-positive and negative mothers in this study during the first 18 months of life. Results: The taxonomic composition of the maternal vaginal and gut microbiota showed no significant differences based on HIV status, and the composition of the infant gut microbiota at birth was similar between HUU and HEU. Longitudinal taxonomic composition of the infant gut microbiota and weight-for-age z-scores (WAZ) differed depending on access to breast milk. HEU infants displayed overall lower WAZ than HUU infants at all time points. We observed a significantly lower relative abundance of Bifidobacterium in HEU infants at 6 months postpartum. Breast milk composition also differed by time point and HIV infection status. The antiretroviral therapy drugs, lamivudine and nevirapine, as well as kynurenine, were significantly more abundant in the breast milk of mothers with HIV. Levels of tiglyl carnitine (C5) were significantly lower in the breast milk of mothers without HIV. ART drugs in the breast milk of mothers with HIV were associated with a lower relative abundance of Bifidobacterium longum. Conclusions: Maternal HIV infection was associated with adverse growth outcomes of HEU infants in this study, and these differences persist from birth through at least 18 months, which is a critical window for the development of the immune and central nervous systems. We observed that the relative abundance of Bifidobacterium spp. was significantly lower in the gut microbiota of all HEU infants over the first 6 months postpartum, even if HEU infants were receiving breast milk. Breastfeeding was of benefit in our HEU infant cohort in the first weeks postpartum; however, ART drug metabolites in breast milk were associated with a lower abundance of Bifidobacterium. [MediaObject not available: see fulltext.]
This was a prospective cohort study of mother-infant pairs conducted at the University of Benin Teaching Hospital Nigeria (UBTH) between 2015 and 2018. The study was approved by the UBTH research ethics committee and the University of Maryland Baltimore Institutional Review Board. Pregnant women with and without HIV infection (~ 150 each) were recruited from the University of Benin Teaching Hospital located in Edo State, Southern Nigeria. Participating women were required to be aged between 18 and 45 years, have documented evidence of HIV status, and willing to comply with follow-up assessment schedule. Babies born to these women were also enrolled at birth. Recruited mother-infant pairs were assessed at birth and followed up for 18 months with scheduled assessment visits at 1, 6, 9, 15, and 18 months. Demographic, clinical, feeding, anthropometric and microbiome data were collected at each visit. Informed consent was also obtained from all mothers. University of Maryland Baltimore and UBTH Institutional Review Boards approved all study procedures. HIV DNA PCR test was done for all HEU babies at 6 weeks postpartum and at 4 months for non-breastfed infants or 2 months after breastfeeding cessation. About 70% of the mothers with HIV were already on highly active antiretroviral treatment (HAART) prior to the index pregnancy, and their triple regimens were continued. Others were initiated on antiretroviral drugs in line with Nigerian guidelines, which recommend HAART for women requiring treatment for their own disease or option B prophylaxis with triple regimen until 1 week after breastfeeding ceases, as well as nevirapine to the baby from birth to 6 weeks. Standardized questionnaires were utilized at each study visit to document general medical and obstetric information, including medication and comorbidity history, general physical examination findings, and anthropometric assessment. Information on feeding practices was collected using structured feeding questionnaires. This included type, pattern, and duration of breastfeeding as well as complementary and alternate feeding practices. Weight was measured to the nearest 0.1 Kg using “Salter Baby Scale (Model 180)” at birth and “Seca Digital Scale (Model 872)” subsequently. For the latter, baby’s weight was determined from the combined mother-baby weight measurement. Recumbent length was measured using an infantometer (“Seca 416”). A flexible non-elastic tape (“Seca 212”) was used to measure head and arm circumference. Low birth weight was defined as birth weight < 2.5 kg [103]. World Health Organization (WHO) child growth standards were used to generate z scores for weight for age (WAZ). WAZ ≤ 2 z-scores were defined as underweight [104, 105]. About 0.5 g of meconium and stool samples were collected at birth and at each follow-up study visit respectively (Table S1). Similarly, 0.5 g of stool sample was collected from the mothers at enrollment and following delivery (Table S1). Breast milk was collected by trained research nurses at 6 weeks and 6 months postpartum (Table S1). After washing hands with soap and water and cleaning the nipples and areolar area with cotton wool soaked in normal saline, 10 ml of breast milk was manually expressed and collected into a falcon tube. This was aliquoted into cryogenic vials and immediately stored at − 20 °C and later in − 80 °C freezers. Vaginal swab was collected from the mothers at enrollment and again following delivery (Table S1). Specimen was collected using “Isohelix Sk-2” swab (Geneflow, Ltd, UK) following aseptic procedures. The swab was then inserted back into its container tube, the cap closed, and tube placed in a ziploc with ice pack, and this was subsequently stored at − 70 °C freezers (Table S1). DNA was extracted from each fecal, meconium, and vaginal specimen. Both positive and negative controls (Zymo, Irvine, CA) were included in the DNA extraction process and the 16S rRNA gene sequence amplification process as previously described [106]. Samples were thawed at 4°C and, in aliquots of 0.15 g per tube, resuspended in 1 ml of 1 × phosphate-buffered saline. Cell lysis was initiated with two enzymatic incubations: 1. using 5 μl of lysozyme (10 mg/ml; Amresco, Solon, OH), 13 μl of mutanolysin (11.7 U/μl; Sigma-Aldrich), and 3 μl of lysostaphin (4.5 U/μl; Sigma-Aldrich) for an incubation of 30 min at 37 °C and, 2. using 10 μl of proteinase K (20 mg/ml; Research Products International, Mt. Prospect, IL), 50 μl of 10% SDS, and 2 μl of RNase (10 mg/ml) for an incubation of 45 min at 56 °C. After the enzyme treatments, cells were disrupted by bead beating in tubes with lysing matrix B (0.1-mm silica spheres; MP Biomedicals, Solon, OH), at 6 m/s at room temperature in a FastPrep-24 (MP Biomedicals). The resulting crude lysate was processed using the ZR fecal DNA miniprep kit (Zymo, Irvine, CA) according to the manufacturer’s recommendations. The samples were eluted with 100 μl of ultrapure water into separate tubes. DNA concentrations in the samples was determined with the Bioanalyzer 2100 DNA 1000 chip (Agilent, Santa Clara, CA). Hypervariable regions V3 and V4 of the bacterial 16S rRNA gene were amplified with primers 319F and 806R as previously described by [107, 108]. High-quality amplicon sequences were obtained on an Illumina HiSeq 2500 modified to generate 300 bp paired-end reads [108]. A total of 139 million reads were retained following chimera removal and 45,556 amplicon sequence variants (ASVs) were generated by DADA2 and taxonomically classified using the RDP Naïve Bayesian Classifier [109] trained with the SILVA v128 16S rRNA gene database [110]. ASVs of major stool taxa were assigned species-level taxonomy using speciateIT (http://ravel-lab.org/speciateit). Negative controls generated a negligible amount of sequencing reads, whereas the positive controls generated the expected mock community [106]. Taxa present at a relative abundance of less than 10-5 across all samples was removed from the dataset. The phyloseq R package [111] was used for analysis of the microbial community data. A selection of 17 breast milk samples from mothers with HIV and 17 breast milk samples from mothers without HIV at visits 6 weeks and 6 months were shipped to Metabolon, Inc. (Durham, NC, USA) for metabolomics. The samples were chosen haphazardly, making sure the baseline characteristics of the mothers were similar and their infants were breastfed for at least 6 months postpartum (Table (Table2).2). Untargeted metabolite profiling was carried out by Metabolon Inc. (Durham, NC, USA) using ultra-high-performance liquid chromatography/mass spectrometry/mass spectrometry (UHPLC/MS/MS). Breast milk was mixed with methanol to recover chemically diverse metabolites after precipitating proteins. The methanol extract was divided into five fractions: two for analysis by two separate reverse-phase (RP) UPLC/MS/MS methods with positive ion mode electrospray ionization (ESI), one for analysis by RP/UPLC/MS/MS with negative ion mode ESI, one for analysis by hydrophilic interaction (HILIC) UPLC/MS/MS with negative ion mode ESI, and one sample was reserved for backup. The mass spectrometry (MS) analysis alternated between MS and data-dependent MS scans using dynamic exclusion. A pooled sample was created by taking a small aliquot from each of the samples, which served as technical replicates in the assay, whereas pure water samples served as a process blank, and a cocktail of quality control (QC) standards (Metabolon) was spiked into every standard sample to identify the instrument variability. The instrument variability determined by calculating the median relative standard deviation for the internal standards was 3%. The samples were randomized across the platforms, and internal standards and process blanks were added to each sample prior to injection into the mass spectrometers. The raw data extraction, peak identification, and QC process were performed using Metabolon's proprietary hardware and software. The metabolites were identified using a proprietary in-house library based on standards that contained the retention time/index, mass to charge ratio, and chromatographic data (including MS/MS spectral data) on molecules present in the library. Additional mass spectral entries were created for structurally unnamed biochemicals, which were identified by their recurrent nature (both chromatographic and mass spectral). Peaks were quantified using the area under the curve. The biochemical data were normalized for the volume of breast milk used. Raw data was extracted, peak-identified and QC processed using Metabolon’s hardware and software. Compounds were identified by comparison to library entries of purified standards or recurrent unknown entities. Metabolon maintains a library based on authenticated standards that contains the retention time/index, mass to charge ratio, and chromatographic data (including MS/MS spectral data) on all molecules present in the library. More than 3300 commercially available purified standard compounds have been acquired and registered into Metabolon’s system for analysis on all platforms for determination of their analytical characteristics. Additional mass spectral entries have been created for structurally unnamed biochemicals, which have been identified by virtue of their recurrent nature (both chromatographic and mass spectral). A data normalization step was performed to correct variation resulting from instrument inter-day tuning differences. Essentially, each compound was corrected in run-day blocks by registering the medians to equal one (1.00) and normalizing each data point proportionately (termed the “block correction”). Statistical analyses were performed using R (version 3.6.0). Demographic and clinical characteristics were compared between HEU and HUU children and their mothers using Wilcoxon, Fisher’s exact and t tests. Longitudinal comparisons of alpha diversity were performed using univariable and multivariable linear regression. Pairwise comparisons were performed with post hoc Tukey HSD test with FDR P value adjustment set at level 0.05. Principal coordinates analysis (PCoA) using Bray-Curtis dissimilarity was performed to assess the beta diversity. Permutational multivariate analysis of variance (PERMANOVA) was conducted to test whether the bacterial communities sequenced have different centroids based on HIV-status (mothers) or HIV-exposure (infants). Significance of the results was confirmed with a test of heterogeneity (ensure homogenous dispersion). In addition, multivariate association with linear models (MaAsLin2) [112], an additive general linear model with boosting that can capture the effects of a parameter of interest while deconfounding the effects of other metadata, was used to efficiently determine multivariable association between clinical metadata, 16S rRNA gene sequence data, and breast milk metabolomic data. MaAsLin2 analysis for the infants was adjusted for delivery type, prematurity, timepoint, antibiotic use, and breastfeeding at the time of visit. Additionally, MaAsLin2 parameters for taxa analysis were set as follows: P value control for Benjamini-Hochberg FDR was set at level 0.05, the minimum abundance for each taxon was set to 1% and the minimum percent of samples for which a taxon is detected at 1% was set to 10%. The parameters for the metabolomic analysis were as follows: P value control for Benjamini-Hochberg FDR was set at level 0.05, the minimum abundance for each metabolite was set to 0.001, and the minimum percent of samples for which a metabolite is detected at 0.001 was set to 10%. The heatmap was created using the statistical package in MetaboAnalyst 5.0 (http://www.metaboanalyst.ca/MetaboAnalyst/). ANOVA and post hoc test were performed by MetaboAnalyst 5.0. The P value was obtained by running the Fishers’ LSD after the ANOVA test and adjusted by multiple test corrections using the Benjamin-Hochberg procedure (FDR was set at level 0.05). Pearson's correlations were run using Benjamini-Hochberg multiple comparison adjustment (FDR P value adjustment set at 0.05).
