Gut-derived bacterial toxins impair memory CD4+ T cell mitochondrial function in HIV-1 infection

Abstract

People living with HIV (PLWH) who are immune nonresponders (INRs) are at greater risk of comorbidity and mortality than are immune responders (IRs) who restore their CD4+ T cell count after antiretroviral therapy (ART). INRs have low CD4+ T cell counts (<350 c/μL), heightened systemic inflammation, and increased CD4+ T cell cycling (Ki67+). Here, we report the findings that memory CD4+ T cells and plasma samples of INRs from several cohorts are enriched in gut-derived bacterial solutes p-cresol sulfate (PCS) and indoxyl sulfate (IS) that both negatively correlated with CD4+ T cell counts. In vitro PCS or IS blocked CD4+ T cell proliferation, induced apoptosis, and diminished the expression of mitochondrial proteins. Electron microscopy imaging revealed perturbations of mitochondrial networks similar to those found in INRs following incubation of healthy memory CD4+ T cells with PCS. Using bacterial 16S rDNA, INR stool samples were found enriched in proteolytic bacterial genera that metabolize tyrosine and phenylalanine to produce PCS. We propose that toxic solutes from the gut bacterial flora may impair CD4+ T cell recovery during ART and may contribute to CD4+ T cell lymphopenia characteristic of INRs.

Keywords: AIDS/HIV; Apoptosis; Infectious disease; Mitochondria; T cell development.

Figure 1. Enrichment of PCS and IS in INR samples.

Untargeted metabolic UPLC-MS/MS measurements showing levels of PCS (A) in 1 × 10⁶ to 2 × 10⁶ sorted memory CD4⁺ T cells of IRs (n = 15), INRs (n = 14), and HCs (n = 7). Concentration of PCS (B) and IS (C) in the plasma of the IR and INR groups in the CLIF (red dots) and SCOPE (green dots) cohorts. Median and interquartile range (solid lines) are displayed in the dot plots. P values were generated by Kruskal-Wallis with Dunn’s multiple-comparison test. Spearman’s correlation of plasma concentrations of PCS (D) and IS (E) with CD4⁺ T cell count. Blue dots are INR patient data.

Figure 2. Correlation of PCS and IS plasma concentration with markers of inflammation.

(A) Plasma concentrations of IFABP in IRs and INRs and Spearman’s correlation between PCS (middle panels)/IS (lower panels) and IL-6 (B), IP-10 (C), and hyaluronic acid (HA) (D). Median and interquartile range (in red) are shown.

Figure 3. PCS and IS block CD4⁺ T cell proliferation, increase apoptosis levels, and diminish mitochondrial protein expression in a dose-dependent manner.

(A) Representative CTV dye dilution proliferation blockade upon treating PBMCs of a healthy donor with gradient concentrations of PCS. CTV-labeled PBMCs were stimulated with anti-CD3/anti-CD28 in media with 2% FBS followed by treatment with increasing concentrations of PCS. CTV dye dilution was evaluated at 72 hours. Apoptosis levels were measured by annexin V binding. Cumulative data (n = 5) on the effect of PCS (B) and IS (C) on CD4⁺ T cell proliferation as monitored by CTV dye dilution. Cumulative data (n = 5) for the effect of PCS (D) and IS (E) on the levels of annexin V binding. (F) Effect of PCS on protein expression of mTFA and COXIV implicated in mitochondrial function. (G) The cumulative MFI (n = 5) for mTFA (green dots) and COXIV (red dots). Median and SD are displayed. *P < 0.05; **P < 0.01; ***P < 0.001 by Kruskal-Wallis with Dunn’s multiple-comparison test.

Figure 4. Treatment of memory CD4⁺ T cells with PCS reduces mitochondrial fitness.

(A) Representative electron microscopy imaging (EMI) from 1 EM section of sorted memory CD4⁺ T cells from HC, HC+PCS, IR, and INR samples. Scale bars: 200 nm (left image) and 500 nm (right 3 images). (B) Mitochondrial numbers per EMI section. (C) Numbers of mitochondria with dense folded cristae. (D) Numbers of mitochondria with vacuoles. Twenty EMI sections from 2 independent experiments were analyzed. P values generated by Kruskal-Wallis with Dunn’s multiple-comparison test. Median and interquartile range are shown.

Figure 5. Bacterial diversity is reduced in INR stool samples and enriched in genera able to produce PCS.

(A) Principal components distribution showing the segregation of HIV-1–infected subjects (n = 12) from HCs (n = 6). (B and C) Diversity of the gut microbiome as determined by Shannon’s index in HIV-1–infected and noninfected subjects and in HCs, INRs (n = 6), and IRs (n = 6). (D) Top 5 phyla detected in HC, INR, and IR stool samples showing significant enrichment of Firmicutes (P < 0.01) in INR samples. (E and F) Abundance of bacterial genera able to produce PCS in INRs and not producing PCS. *P < 0.05, **P < 0.01 by Kruskal-Wallis with Dunn’s multiple-comparison test.