Premature skewing of T cell receptor clonality and delayed memory expansion in HIV-exposed infants

Abstract

While preventing vertical HIV transmission has been very successful, HIV-exposed uninfected infants (iHEU) experience an elevated risk to infections compared to HIV-unexposed and uninfected infants (iHUU). Here we present a longitudinal multimodal analysis of infant immune ontogeny that highlights the impact of HIV/ARV exposure. Using mass cytometry, we show alterations in T cell memory differentiation between iHEU and iHUU being significant from week 15 of life. The altered memory T cell differentiation in iHEU was preceded by lower TCR Vβ clonotypic diversity and linked to TCR clonal depletion within the naïve T cell compartment. Compared to iHUU, iHEU had elevated CD56^loCD16^loPerforin⁺CD38⁺CD45RA⁺FcεRIγ⁺ NK cells at 1 month postpartum and whose abundance pre-vaccination were predictive of vaccine-induced pertussis and rotavirus antibody responses post 3 months of life. Collectively, HIV/ARV exposure disrupted the trajectory of innate and adaptive immunity from birth which may underlie relative vulnerability to infections in iHEU.

Fig. 1. Immunophenotypic trajectory of infant NK cells and T cells in the first 9 months of life.

A−C Age related maturation trajectory of NK cells and CD4⁺ and CD8⁺ T cells depicted using multidimensional scaling (MDS) coordinates derived from median marker expressions for each infant samples collected in cord blood (CB, n = 5) and at birth (BTH, n = 44) and weeks (WK) 4 (n = 53), 15 (n = 52) and 36 (n = 53). D−F Boxplots summarising MDS coordinate associated with infant age for NK cells and CD4⁺ and CD8⁺ T cells. Boxplot central line depicts the median, the upper and lower lines represents the 75th and 25th percentile respectively, and the whiskers mark the 1.5 times the 75th and 25th percentile boundary, (p = Kruskal Wallis). Summary trendline shows local regression as determined by local estimated scatterplot smoothing (LOESS) with 95%CI error bands. G, I, K Uniform manifold approximation and proximation (UMAP) with density preservation showing dimensional reduction of FlowSOM clusters of NK cells and CD4⁺ and CD8⁺ T cells, respectively. H, J, L Age related phenotypic composition of NK cells and CD4⁺ and CD8⁺ T cells depicted as principal component (PC) coordinates of centred log-odd ratios of the FlowSOM clusters for each infant samples and arrows indicating contribution of each cell cluster in the scatter of PC components. M Spearman’s correlation of PC coordinates associated with immune trajectory for NK cells, and CD4⁺and CD8⁺ T cells. Summary trendline shows local regression as determined by LOESS with 95%CI error bands. N Circoplot showing correlations among immune cell clusters derived from NK cells, and CD4⁺ and CD8⁺ T cells.

Fig. 2. Divergent T cell memory differentiation in HIV-exposed uninfected infants (iHEU) compared to HIV-unexposed uninfected infants (iHUU).

A−C Boxplots comparing PC coordinates derived from centred log-odd ratios of FlowSOM NK cell, and CD4⁺ and CD8⁺ T clusters between iHEU and iHUU at birth (BTH), weeks (WK) 4, 15 and 36. Boxplot central line depicts the median, the upper and lower lines represents the 75th and 25th percentile respectively, and the whiskers mark the 1.5 times the 75th and 25th percentile boundary (*p < 0.05, **p < 0.01; Two-sided Wilcoxon). Summary trendline shows local regression as determined by local estimated scatterplot smoothing (LOESS) with 95%CI error bands. D, E Generalised linear mixed model (GLMM) comparing the abundances of CD4⁺ T cell clusters between iHEU and iHUU at weeks 15 and 36. F, G GLMM comparing the abundances of CD8⁺ T cell clusters between iHEU and iHUU at weeks 15 and 36. Positive log Fold change (FC) signified cell cluster frequencies were higher in iHEU compared to iHUU. False discovery rate (FDR) was used to correct for multiple comparison and adjusted p values are presented.

Fig. 3. Premature CD4+ and CD8 + T cell receptor (TCR) repertoire skewing in HIV-exposed uninfected infants (iHEU) relative to HIV-unexposed uninfected infants (iHUU).

A Boxplots comparing Inverse Simpson TCR diversity scores between iHUU and iHEU at birth (BTH) and weeks (WK) 4, 15 and 36. B Boxplots comparing Chao1 TCR clonotype richness between iHUU and iHEU at BTH and WK4, 15 and 36. All boxplots used the standard Tukey’s representation with the central line depicting the median, the upper and lower lines represent the 75th and 25th percentile respectively, and the whiskers mark the boundary at 1.5 times of the 75th and 25th percentile (*p < 0.05, **p < 0.01 and ***p < 0.001; Two-sided Wilcoxon). C Spearman’s rank correlation between naive CD4⁺ and CD8⁺ T cell Inverse Simpson scores and frequencies of CD4⁺ and CD8⁺ T cell clusters respectively. *adjusted p < 0.05, FDR used to correct for multiple comparisons. D GLIPH analysis showing antigen specificity that were significantly enriched in iHUU relative to iHEU and in naïve relative to memory CD4⁺ and CD8⁺ T cell subsets. Sample size for each group is shown in Fig. S1.

Fig. 4. Association of vaccine antibody responses to immune cell phenotypes.

A Comparing IgG levels against pertussis between HIV-exposed uninfected infants (iHEU) and HIV-unexposed uninfected infants (iHUU), grey shaded area indicate threshold IgG levels for protective pertussis vaccine response. B Boxplots comparing rotavirus specific IgA titres between iHEU and iHUU at week 36. All boxplots used the standard Tukey’s representation with the central line depicting the median, the upper and lower lines represent the 75th and 25th percentile respectively, and the whiskers mark the boundary 1.5 times of the 75th and 25th percentile (*p < 0.05, **p < 0.01 and ***p < 0.001; Two-sided Wilcoxon). C, D Spearman’s correlation of FlowSOM clusters abundances for NK cells, CD4⁺ and CD8⁺ T cells to anti-pertussis IgG and anti-rotavirus IgA respectively measured at week 15 and 36. *adjusted p < 0.05, FDR used to correct for multiple comparisons.

Fig. 5. Early-life immune cell compositions predictive of antibody responses post-vaccination.

A, C Summary of ROC analysis using the latent variable axis-1 derived from partial least square discriminate analysis (PLS-DA) of NK, CD4 and CD8 T cell clusters determined to be best predictors at birth and week 4 of pertussis antibody responses at weeks 15 and rotavirus antibody response at week 36. B, D Loading variables from PLS-DA showing pre-vaccine cell clusters associated with good or poor antibody responses against pertussis and rotavirus vaccine at week 15 and 36 respectively. Sample size for each group is shown in Fig. S1.