HIV Infection is associated with compositional and functional shifts in the rectal mucosal microbiota

Abstract

Background: Regardless of infection route, the intestine is the primary site for HIV-1 infection establishment and results in significant mucosal CD4+ T lymphocyte depletion, induces an inflammatory state that propagates viral dissemination, facilitates microbial translocation, and fosters establishment of one of the largest HIV reservoirs. Here we test the prediction that HIV infection modifies the composition and function of the mucosal commensal microbiota.

Results: Rectal mucosal microbiota were collected from human subjects using a sponge-based sampling methodology. Samples were collected from 20 HIV-positive men not receiving combination anti-retroviral therapy (cART), 20 HIV-positive men on cART and 20 healthy, HIV-negative men. Microbial composition of samples was analyzed using barcoded 16S Illumina deep sequencing (85,900 reads per sample after processing). Microbial metagenomic information for the samples was imputed using the bioinformatic tools PICRUST and HUMAnN. Microbial composition and imputed function in HIV-positive individuals not receiving cART was significantly different from HIV-negative individuals. Genera including Roseburia, Coprococcus, Ruminococcus, Eubacterium, Alistipes and Lachnospira were depleted in HIV-infected subjects not receiving cART, while Fusobacteria, Anaerococcus, Peptostreptococcus and Porphyromonas were significantly enriched. HIV-positive subjects receiving cART exhibited similar depletion and enrichment for these genera, but were of intermediate magnitude and did not achieve statistical significance. Imputed metagenomic functions, including amino acid metabolism, vitamin biosynthesis, and siderophore biosynthesis differed significantly between healthy controls and HIV-infected subjects not receiving cART.

Conclusions: HIV infection was associated with rectal mucosal changes in microbiota composition and imputed function that cART failed to completely reverse. HIV infection was associated with depletion of some commensal species and enrichment of a few opportunistic pathogens. Many imputed metagenomic functions differed between samples from HIV-negative and HIV-positive subjects not receiving cART, possibly reflecting mucosal metabolic changes associated with HIV infection. Such functional pathways may represent novel interventional targets for HIV therapy if normalizing the microbial composition or functional activity of the microbiota proves therapeutically useful.

Figure 1

Schematic of sampling and bioinformatic methodology. Rectal mucosa secretions were collected from 60 human subjects as shown and subjected to high-throughput deep V4 16S sequencing. Alpha and beta diversity analyses were then performed, which suggested microbial composition was altered in HIV-infected subjects who were not receiving combination anti-retroviral therapy (cART). Microbial differences in these subjects were identified and compared with HIV-infected subjects receiving cART. In addition, the different classes of cART were analyzed to determine whether any class was significantly associated with differences in microbial composition. Imputed metagenomic differences between HIV-infected subjects not receiving cART and healthy control subjects were then identified and compared between the three patient cohorts. PICRUSt, phylotypic investigation of communities by reconstruction of unobserved states; HUMAnN, HMP unified metabolic analysis network.

Figure 2

Alpha and beta diversity. (A) Chao1 alpha diversity indicated subjects on combination anti-retroviral therapy (cART(-)) had reduced species richness than healthy controls (HC) and subjects on (cART(+)), though only the comparison with HC was significant (t-test, P ≤0.05 at all sampling depths >17,188). Abundance curves for HC and cART(+) subjects were nearly indistinguishable. (B) Beta diversity was analyzed by unweighted unifrac analysis using the first and fourth principal components. These principal components were selected because principal coordinate 1 (PC1) was significantly different between cART(-) and cART(+) subjects (Kruskal-Wallis, P = 0.02) and both PC1 and PC4 were significantly different between cART(-) and HC subjects (Kruskal-Wallis, both P <0.05). HC subjects (blue) clustered relatively tightly with cART(+) subjects (yellow), whereas cART(-) subjects (red) were more diffusely scattered along PC1 and PC4 (Adonis, for cART(+) vs cART(-) P = 0.06, and for cART(-) versus HC P = 0.02). (C) Phylum level composition of each subject was sorted based on HIV and cART status. The abundance of Firmicutes was significantly reduced in cART(-) subjects compared with healthy subjects (analysis of variance (ANOVA), q = 0.06) while Fusobacteria were significantly enriched (ANOVA, q = 0.11).

Figure 3

Phylogenetic differences in subjects not on combination anti-retroviral therapy (cART(-)). A phylogenetic tree populated with taxa that were significantly associated with HIV infection. Blue boxes indicate taxa that were significantly depleted in cART(-) subjects; Red boxes indicate clades or taxa that were enriched in cART(-) subjects; and grey boxes indicate clades or taxa that were neither enriched nor depleted in HIV, and are only included to contextualize taxa that were differentially abundant. All associations were determined by analysis of variance and were significant at q <0.15 except Porphyromonas (q = 0.155).

Figure 4

Combination anti-retroviral therapy (Cart) partially normalizes microbial composition. Genera that were significantly depleted or enriched in subjects not on cART (cART(-)) were analyzed in those on cART (cART(+)) and healthy controls (HC) to reveal the extent to which cART use reversed microbial changes. While no genera were significantly different between cART(+) and HC subjects, cART(+) subjects displayed intermediate levels of enrichment and depletion in all tested genera. The largest vertical black bars in each bean plot represent group averages. Thin vertical black bars represent at least one sample. The x-axis represents the percent abundance of genera for each sample; *q <0.15 relative to HC.

Figure 5

Imputed metagenomic differences between subjects not on combination anti-retroviral therapy (cART(-)) and healthy control (HC) subjects. The relative abundance of metabolic pathways encoded in each imputed sample metagenome was analyzed by HIV infection status of each subject using box plots. From these box plots, clear differences are observed between the relative abundance of several imputed metagenomic functions between cART(-) subjects and HC. Significance of each comparison was determined using Kruskal-Wallis one way analysis of variance. Box plots of subjects on cART (cART (+)) are included to provide context for each comparison. Vertical black bars represent group averages. The x-axis represents the percent abundance of pathways for each imputed sample metagenome. Whiskers represent the interquartile ranges multiplied by 1.5; *q <0.25 relative to HC.