Comparative analysis reveals a role for TGF-β in shaping the residency-related transcriptional signature in tissue-resident memory CD8+ T cells

Abstract

Tissue-resident CD8+ memory T (TRM) cells are immune cells that permanently reside at tissue sites where they play an important role in providing rapid protection against reinfection. They are not only phenotypically and functionally distinct from their circulating memory counterparts, but also exhibit a unique transcriptional profile. To date, the local tissue signals required for their development and long-term residency are not well understood. So far, the best-characterised tissue-derived signal is transforming growth factor-β (TGF-β), which has been shown to promote the development of these cells within tissues. In this study, we aimed to determine to what extent the transcriptional signatures of TRM cells from multiple tissues reflects TGF-β imprinting. We activated murine CD8+ T cells, stimulated them in vitro by TGF-β, and profiled their transcriptomes using RNA-seq. Upon comparison, we identified a TGF-β-induced signature of differentially expressed genes between TGF-β-stimulated and -unstimulated cells. Next, we linked this in vitro TGF-β-induced signature to a previously identified in vivo TRM-specific gene set and found considerable (>50%) overlap between the two gene sets, thus showing that a substantial part of the TRM signature can be attributed to TGF-β signalling. Finally, gene set enrichment analysis further revealed that the altered gene signature following TGF-β exposure reflected transcriptional signatures found in TRM cells from both epithelial and non-epithelial tissues. In summary, these findings show that TGF-β has a broad footprint in establishing the residency-specific transcriptional profile of TRM cells, which is detectable in TRM cells from diverse tissues. They further suggest that constitutive TGF-β signaling might be involved for their long-term persistence at tissue sites.

Fig 1. Schematic overview of the experimental design.

(A) On day 0, wild type C57BL/6 splenocytes were coated with gB peptide and cocultured with gBT-I splenocytes in order to specifically activate gB-specific CD8⁺ T cells. Addition of IL-2 on days 2, 3, and 4 maintains T cell proliferation. At day 5, a highly pure population of activated CD8⁺ gBT-I T cells was subjected to four conditions: unstimulated control, TGF-β, IL-2, and TGF-β/IL-2. After 40h of TGF-β stimulation, cells were subjected to RNA extraction and RNA sequencing; N = 3 independent experiments. (B) Flow cytometric analysis of activated gBT-I cells on day 4. After pre-gating on live singlets, gBT-I cells are distinguished as CD45.1⁺, vα2 T cell receptor-positive (TCR⁺), and CD8⁺ cells. The representative dot plot (left) of CD45.1 and vα2 TCR staining, indicates the enrichment of gBT-I cells (red quadrant). The representative histogram (right) shows the expression of CD8 on gBT-I cells. The dot plot and histogram are representative of over 3 independent experiments.

Fig 2. Principal component analysis and hierarchical clustering.

(A) PCA plot (PC1 vs. PC2) of all 12 RNA samples. The numbers in parenthesis beside the PC labels denote the percentage of variance in the data explained by the respective PC. Clusters of TGF-β-treated (TGF-β and IL-2/TGF-Β) and TGF-β-untreated (Untreated and IL-2) groups, which separated along PC2, are circled. (B) Dendrogram from hierarchical cluster analysis of all the samples based on their expression profiles. Clustering was done using the Ward’s method with the “maximum” distances measure provided as the dissimilarity matrix. Dendrogram branches are coloured by TGF-β treatment: TGF-β-treated groups (blue) and TGF-β-untreated groups (orange). The dots represent biological samples coloured according to TGF-β treatment they received. Biological replicates (Rep.1-3) for each treatment have the same colour.

Fig 3. GO terms enriched among genes differentially expressed in the TGF-β-treated groups compared to their untreated counterparts.

Top representative GO (biological processes) terms based on REVIGO output, enriched among (A) all the DE genes, (B) up-regulated genes, and (C) down-regulated genes in the TGF-β-treated groups. The GO terms (y-axes) were ranked according to their enrichment P-values (x-axes). The numbers on top of the bar plots in parentheses denote the total number of all DE, up-, and down-regulated genes with GO term annotations. The numbers at the end of each bar represent the actual number of all DE genes; up- or down-regulated genes that were classified to a particular biological process. All GO terms listed were significant at FDR < 0.05. All of the 254 common up-regulated genes in the TGF-β-treated groups were associated with GO terms. 160 out of 162 common down-regulated genes in the TGF-β-treated groups were annotated with GO terms.

Fig 4. TGF-β induced transcriptional profiles are enriched for the TRM-related signature genes identified in murine TRM cells.

(A-C) The bar plots show the log2 fold change in expression (x-axes) of the 35 TRM-related signature genes in each of the pairwise comparison of the TGF-β-treated groups vs. TGF-β-untreated groups. The bars are colour-coded based on their expression in both TRM cells and TGF-β-treated groups. Consistently up-regulated genes are depicted in red, and down-regulated genes are shown in blue. Grey represents genes that were either not significantly DE in TGF-β-treated groups vs. TGF-β-untreated groups or whose direction of DE was not concordant between TRM cells and TGF-β-treated groups. The numbers in the pie chart denote the percentage overlap between the genes in the TRM-related signature and genes DE (FDR < 0.05) in the TGF-β-treated groups. The P-values next to the pie charts denote the enrichment P-values calculated via bootstrapping (Methods).

Fig 5. Enrichment plot for tissue-specific TRM-associated gene sets in the TGF-β-treated group.

GSEA enrichment plots for the (A) five TRM-related up-regulated and (B) five TRM-related down-regulated gene sets in the TGF-β-treated group. All 10 gene sets shown were significantly enriched at P-value < 0.05 and FDR q-value < 0.25. Each plot shows the running enrichment scores (y-axes) and the position (denoted by dots) of the members of each gene set in the ranked list of genes DE between TGF-β-treated group and TGF-β-untreated group (x-axes). The genes in the ranked list are ordered along the x-axes based on fold change, where the most up-regulated genes in the TGF-β-treated group are on the far left and the most down-regulated genes on the far right. The dotted vertical grey line represents fold change of zero. The curved lines, coloured by tissue type, show the cumulative enrichment score.