Single-cell mass cytometry analysis of human tonsil T cell remodeling by varicella zoster virus

Abstract

Although pathogens must infect differentiated host cells that exhibit substantial diversity, documenting the consequences of infection against this heterogeneity is challenging. Single-cell mass cytometry permits deep profiling based on combinatorial expression of surface and intracellular proteins. We used this method to investigate varicella-zoster virus (VZV) infection of tonsil T cells, which mediate viral transport to skin. Our results indicate that VZV induces a continuum of changes regardless of basal phenotypic and functional T cell characteristics. Contrary to the premise that VZV selectively infects T cells with skin trafficking profiles, VZV infection altered T cell surface proteins to enhance or induce these properties. Zap70 and Akt signaling pathways that trigger such surface changes were activated in VZV-infected naive and memory cells by a T cell receptor (TCR)-independent process. Single-cell mass cytometry is likely to be broadly relevant for demonstrating how intracellular pathogens modulate differentiated cells to support pathogenesis in the natural host.

Fig. 1. VZV infection alters the phenotypic hierarchy of tonsil T cell populations

A) UI T cells were distributed by SPADE into 50 nodes based on four core markers CD4, CD8, CD45RO and CD45RA (representative experiment). Nodes are cells with a similar profile placed in proximity to nodes with closely related profiles, with size proportional to number of cells. Node colors in panels i and ii-v indicate cell count and median intensity of protein expression, respectively (blue: minimum; red: maximum. Arrows indicate the three major subpopulations. DP: double positive CD4+CD8+: DN: double negative CD4−CD8−. (B) 3D visualization of UI T cells by PCA displaying each cell (red dot) by the co-ordinates of its first three principal components based on expression of 17 surface proteins. The three distinct cell clouds correspond to the dominant subpopulations; M = memory, N = naïve (see Fig. S1B). (C) Subpopulations of UI and V+ T cells were characterized by proportions expressing the four core markers (yellow: minimum; red: maximum. (D) 3D visualization of V+ T cells by PCA (left); contribution of variance from each variable in each of the three principal components is shown in Fig. S1G. V+ T cells were distributed into two cell clouds (a and b). Hierarchical cluster heatmap of V+ T cells (right panel); each row denotes a cell and each column represents a protein; the color scale indicates protein expression levels. Dendrograms denote the Euclidean distance between clustered cells. (E) Hierarchical relationship of pooled UI, Bys, and V+ T cells based on surface phenotypes by PCA (left) and agglomerative clustering (right) of equal numbers of cells from each population. V+ T cells were iteratively compared with a randomly selected equal number of UI and Bys T cells; a representative iteration is shown. See also Fig.S1.

Fig. 2. Alterations in T cell surface proteins associated with activation and migration on VZV-infected T cells

(A) Boxplot of changes in cell surface protein and VZV gE protein expression (x-axis) on all V+ T cells (green) compared to UI (red) and Bys (blue) T cells (representative experiment). Each box depicts the 25^th and 75^th percentile and median (center line) expression values (y-axis); whiskers indicate 1.5x the interquartile range. Fold-change in the median expression of the proteins was determined between V+ vs UI, V+ vs Bys, and Bys vs UI T cells. Bonferroni correction adjusted p values from 1-sided, 2-sample Wilcoxon tests corresponding to the alternative hypotheses of more than 20% change in protein expressions (in arcsinh scale) in the V+ T cells were determined; p values (***p < 0.001) are shown where V+ T cells were significantly different from both UI and Bys T cells and UI and Bys T cells did not differ. (B-E) SPADE trees showing differential expression of cell surface proteins indicated on UI and V+ T cells across phenotypic hierarchies (Bys T cells shown in Fig. S2A). Median intensity values of protein expression are denoted by the node color (blue:minimum; red: maximum). Kinetics of PD-1 expression in UI and V+ T cells at 48 (left panel) and 72 hpi (right panel) across T cell phenotypes are shown in D. (F) Dotplots of CXCR5 and PD-1 dual expression on UI, Bys, and V+ T cells determined by flow cytometry. (G) Dotplots of CCR4 and CLA expression on all UI, Bys and V+ T cells (upper panels) and on the subpopulations of CXCR5+ T cells (lower panels). See also Fig.S2, S3.

Fig. 3. Analysis of combinatorial surface changes on VZV-infected T cells

(A) Boxplots showing variations in expression of the four core and 13 other T cell surface proteins on UI (red) and V+ (green) T cells in CD4+M, CD4+N and CD8+N subpopulations (representative experiment). Fold-changes in the median expression of the 13 non-core proteins were determined; Bonferroni correction adjusted p values from 1-sided, 2-sample Wilcoxon tests corresponding to the alternative hypotheses of more than 20% change in protein expressions (in arcsinh scale) in the V+ T cells were determined ***p < 0.001; ** 0.001 < p < 0.05; y-axis: the arcsinh value of protein expression, x-axis: proteins tested. See Fig. S4. (B) Schematic representation of single cell linkage distance estimation (SLIDE) analysis. Each V+ T cell (green) from any given subpopulation was matched with every UI T cell (red) to identify its closest neighbor, UI_nn (purple), which was most similar based on combinatorial expression of 13 non-core surface markers. The UI_nn T cell was subsequently matched to its nearest neighbor, UI_ui-nn (orange), in the UI population. Measured absolute distances between these nearest neighbor T cells were denoted as d1 and d2. (C) The d1:d2 ratio, calculated for each T cell in a given subpopulation (x-axis), was averaged and plotted (y-axis) with confidence intervals incorporating 95% of V+ T cells.

Fig. 4. Activation of the Zap70 pathway in VZV-infected T cells

Fold-change in expression of pZap70 (A), pSLP76 (B) and pErk1/2 (C) in V+ and Bys T cells at 48 and 72 hpi relative to UI T cells by SPADE (see Fig.S4A). Relevant subpopulations on the SPADE trees are marked for reference in (A), left panel. (D) Maximum likelihood estimate (MLE) showing the proportion of pZap70+ T cells that also expressed the indicated downstream signaling proteins in V+, Bys, and UI T cells. (E-G) Comparison of the fold-change in expression of the indicated phosphoproteins in V+, Bys and UI T cells at 48 hpi (blue: minimum; red: maximum). (H) MLE showing the proportion of pMAPKAPK2+ T cells that also expressed pCREB and pS6 in the V+, Bys, and UI T cells. MLEs in (D) and (H) were determined using data from five independent experiments; p values (***p < 0.05; Mann-Whitney-Wilcoxon test). See Fig.S4; Table S2-3.

Fig. 5. Activation of the Akt pathway in VZV-infected T cells

Fold-change in expression levels of pAkt (A), p4EBP1 (B), IκBα (C), and pFAK (D) in V+ and Bys T cells at 48 and 72 hpi relative to UI T cells by SPADE (see Fig. S5A). (blue: minimum; red: maximum). (E) Maximum likelihood estimate (MLE) showing the proportion of pAkt+ T cells also expressing the indicated downstream signaling proteins in the V+, Bys, and UI T cells determined from five independent experiments; p values (***p < 0.05; Mann-Whitney-Wilcoxon test). See Fig.S5; Table S4.

Fig. 6. Analysis of cell signaling pathway activation in VZV-infected T cells

A) Schematic of T cell signaling, indicating the hierarchical locations of the phosphoproteins that were tested (bold font). VZV particles and green arrows indicate the point at which activation of the signaling pathway was detected in VZV-infected T cells or where it was downregulated. The activation index (AI) was calculated for each phosphoprotein in V+ and UI T cells and represented as circles on the pathway diagram. Basal activation in UI T cells is shown as red circles on the left branch, with circle sizes corresponding to the AI for each protein. The right branches show the fold changes in AI for each protein in V+ T cells relative to UI T cells, green circles indicate an increase and yellow circles indicate a decrease in AI. ‘—’ denote activation and ‘- -’ denote inhibition. Proteins noted in small font are intermediate pathway components, not tested. See Fig. S6A. (B) Comparison of phosphorylation determined by calculating the AIs for each phosphoprotein in V+, UI_nn, and total UI T cells for two representative CD4+ T cell subpopulations, CD4+RO+ and CD4+RO+RA+. The mean AI (+SD) (y-axis) for each protein in V+, UI_nn, and total UI T cells was calculated from five independent experiments. P values were determined by one-sided Student t test (p < 0.05) for each protein compared between V+ and UInn T cells and for V+ and total UI T cells. See Fig.S6B; Table S5–6.

Fig. 7. Verification of the remodeling capacity of VZV infection using naïve T cells

A) Activation index (AI) of each phosphoprotein in the UI, Bys, and V+ T cells in the two subpopulations expressing CD45RA, either with or without CD45RO. p values were determined by one-sided Student t test (p < 0.05). Changes were significant in V+ T cells when compared to UI and/or Bys cells in the CD45RA+RO- subpopulation (lower panel) except for those marked with black dots. See Table S7. (B) Boxplots showing expression intensity (y-axis) of CD69 and CD279. Fold-changes in the median expression of the two proteins were determined; p values: Wilcoxon signed-rank non-parametric test ***p < 0.001. (C) PCA scatterplot of pooled UI (red), Bys (blue), and V+ (green) purified naïve CD4+ T cells and the agglomerative clustering heatmap, as determined using equal number of cells from each population. The arrow indicates V+ T cells that form a cluster distinct from UI/Bys T cells on the PCA plot and also form a hierarchically distinct group on the heatmap. V+ T cells were iteratively compared with a randomly selected equal number of UI and Bys T cells; a representative iteration is shown. (D) Fold-changes in expression of indicated surface markers on V+ T cells, relative to UI T cells. The white arrow indicates the SPADE node displaying changes in several surface markers compared to the UI and Bys T cells. See Fig.S7.