Zhenget al.,Science 374 , eabe6474 (2021) 17 December 2021 4 of 11
Fig. 2. Heterogeneity and dynamics of CD8+exhausted T cells.(A) Dot plot
showing the expression of signature genes of the four CD8+exhausted T cells.
Both color and size indicate the effect size. (B) RNA velocities overlaid on UMAP
showing two major state transition paths from naïve to exhaustion. Arrows on a
grid show the RNA velocity field, and dots are colored by metaclusters. (C) The
pair-wise transition index (pTrans) of terminal Texcells. Three metaclusters, with
which terminal Texcells are highly connected (pTrans > 0.1), are highlighted.
(D) Heatmap showing the pTrans between terminal Texcells and other
metaclusters, stratified by cancer types. Color represents thez-score scaled
pTrans value by row. Metaclusters belonging to the two major paths to
exhaustion are highlighted with a dashed line box. pTrans are whited out if the
value < 0.1, representing low transition potential. (E) UMAP visualization showing
the expression of selected marker genes in terminal Texcells of ovarian cancer
(OV). Color representsz-score–scaled gene expression values. (F) Violin and
boxplot showing the expression ofIL26across metaclusters in pan-cancer
(PanC) and esophageal cancer (ESCA). The expression pattern in pan-cancer
serves as a“reference,”comparing with the“ectopic”expression of genesIL26in
the terminal Texcells of ESCA. (G) Boxplot showing the frequency ofIL26+cells in
terminal Texcells among different cancer types. ThePvalue is calculated with
the Kruskal-Wallis test. (H) Bar plot showing the effect size of the top 10
universal TFs of terminal Texcells. (I) Scatter plot showing the specificity scores
of regulons of terminal Texcells. The top 10 regulons are highlighted.RESEARCH | RESEARCH ARTICLE