Longitudinal analysis of microbiome composition in Ghanaians living with HIV-1

Abstract

Human immunodeficiency virus (HIV) 1 infection is known to cause gut microbiota dysbiosis. Among the causes is the direct infection of HIV-1 in gut-resident CD4⁺ T cells, causing a cascade of phenomena resulting in the instability of the gut mucosa. The effect of HIV infection on gut microbiome dysbiosis remains unresolved despite antiretroviral therapy. Here, we show the results of a longitudinal study of microbiome analysis of people living with HIV (PLWH). We contrasted the diversity and composition of the microbiome of patients with HIV at the first and second time points (baseline_case and six months later follow-up_case, respectively) with those of healthy individuals (baseline_control). We found that despite low diversity indices in the follow-up_case, the abundance of some genera was recovered but not completely, similar to baseline_control. Some genera were consistently in high abundance in PLWH. Furthermore, we found that the CD4⁺ T-cell count and soluble CD14 level were significantly related to high and low diversity indices, respectively. We also found that the abundance of some genera was highly correlated with clinical features, especially with antiretroviral duration. This includes genera known to be correlated with worse HIV-1 progression (Achromobacter and Stenotrophomonas) and a genus associated with gut protection (Akkermansia). The fact that a protector of the gut and genera linked to a worse progression of HIV-1 are both enriched may signify that despite the improvement of clinical features, the gut mucosa remains compromised.

Keywords: Ghana; HIV; PLWH; gut microbiome dysbiois; longitudinal analysis.

Figure 1

Alpha and beta diversity of the cohort. Principal component analysis shows that the microbial abundance of each of the three condition-timepoints can be clustered together. Each dot represents each sample’s taxon abundance in two dimensions. Beta diversity is measured using Curtis-Bray, Unweighted UniFrac, and Weighted UniFrac distances (A). (B) Shows the diversity indices between condition-timepoint. The relative abundance of the 19 highest expressed genera in each condition-timepoint is shown in (C). Statistical significance is calculated by the Wilcoxon test. **p < = 0.01, ***p < = 0.001, ****p < = 0.0001.

Figure 2

The result of linear discriminant analysis (LDA) effect size (LEfSe) analyses. LEfSe analysis between baseline_case and baseline_control and baseline_case and follow-up_case is shown in (A,B), respectively. The difference in genera abundance between baseline_case and baseline_control and baseline_case and follow-up_case is shown in (C,D), respectively. The difference in genera abundance between HIV negative and positive is shown in (E). LDA threshold is 2.5.

Figure 3

The relative abundance of select genera in each condition-timepoint. The genera Achromobacter, Dorea, Stenotrophomonas, Blautia, and Subdoligranulum shows low abundance in the baseline _control, high abundance in the baseline_case, and low abundance in the follow-up_case, albeit a non-significance in the genus Subdoligranulum. In contrast, the genus Faecalibacterium has a high abundance in the baseline_control, a reduction in the baseline_case, and a rebound in the follow-up_case. The genera Staphylococcus and Streptococcus, however, have high abundance in positive cases regardless of time point. Statistical significance is calculated by the Wilcoxon test. **p < = 0.01, ***p < = 0.001, ****p < = 0.0001.

Figure 4

The correlation of microbiome abundance with clinical features. The diversity indices between low and high CD4⁺ cell count are shown in (A). The diversity indices between low and high soluble CD14 level are shown in (B). The cutoff value between high and low is the median value. CD4⁺ T-cell count is in cells/μL of blood and soluble CD4 is in pg/mL of blood. The correlation between genera and clinical features is observable in (C). Statistical significance is calculated by the Wilcoxon test. **p < = 0.01, ***p < = 0.001, ****p < = 0.0001. ART: antiretroviral, sCD14: soluble CD14, IFABP: intestinal fatty acid binding protein, LBP: lipopolysaccharide-binding protein.

Figure 5

Predicted pathways upregulated/downregulated in our cohort. PICRUSt2 predicts that baseline_case samples have pathways that are significantly enriched compared to baseline_control and follow-up_case.

Figure 6

Diversity indices and clinical features in regard to treatment duration. Values of CD4⁺ T-cell count, viral load, soluble CD14, IFABP, and LBP in regard to treatment duration is shown in (A–E), respectively. Antiretroviral (ART) duration is in months, CD4⁺ T-cell count is in cells/μL of blood, viral load is in copies/μL of blood, soluble CD14 is in μg/mL of blood, intestinal fatty acid binding protein (IFABP) is in ng/mL of blood, and lipopolysaccharide-binding protein (LBP) is in μg/mL of blood. Each white dot is the value of a sample while the black dots show the median value. Diversity indices in regard to duration treatment is shown in (F,G). Statistical significance is calculated by the Wilcoxon test. **p < = 0.01, ***p < = 0.001, ****p < = 0.0001.

Figure 7

Linear discriminant analysis (LDA) effect size (LEfSe) in regard to treatment duration. LEfSe analysis between baseline_case and baseline_control and baseline_case and follow-up_case in regard to treatment duration is shown here. There is no follow-up result in naive due to its effect size is lower than the threshold 2.5.