The endogenous antigen-specific CD8+ T cell repertoire is composed of unbiased and biased clonotypes with differential fate commitments

Abstract

Generating balanced populations of CD8⁺ effector and memory T cells is necessary for immediate and durable immunity to infections and cancer. Yet, a definitive understanding of how a diverse CD8⁺ T cell repertoire differentiates remains unclear. We identified several hundred T cell receptor (TCR) clonotypes that constitute the polyclonal response against a single antigen and found that a majority of TCR clonotypes were highly biased toward memory or effector fates. TCR-intrinsic biases were not stochastic and were dominant over environmental cues. Differential gene expression analysis of memory- or effector-biased TCR clonotypes showed bifurcation of differential fates at the early effector stage. Additionally, phylogenetic analysis revealed that memory-biased clonotypes retain their fate preferences in subclonal populations but effector-biased subclones can switch to a memory fate. Our study highlights that the polyclonal CD8⁺ T cell response is a composite of unbiased and biased clonotypes with varying capacity to incorporate environmental cues in their cell fate decisions.

Keywords: T cell fate bias; endogenous CD8(+) T cell response; hierarchal lineage tracing.

Figure 1.. Endogenous OVA-specific CD8⁺ T cells exhibit a spectrum of cell states at the peak of the immune response to acute VSV-OVA infection.

A. Schematic of the CARLIN mouse line expressing reverse tetracycline-controlled transactivator (rtTA) from the Rosa26 (R26) locus. Doxycycline (Dox)-inducible Cas9 expression and ten U6 promoter-controlled gRNAs and CAG promoter-controlled CARLIN-barcode array within the 3’ UTR of GFP are expressed from the Col1a1 locus. Cas9-mediated CARLIN-barcode edits include deletions or insertions (blue blocks). B. Experimental schematic. In one independent experiment dual OVA-K_b Tetramer+splenic CD8⁺ T cells were sorted from three VSV-OVA infected mice (LA1, LA2, LA3) on D7 post-infection and analyzed by scRNA-seq, scCARLIN-barcode-seq and LR-scTCR-seq. Dox was given on days 1 and 4. C. UMAP of OVA-K_b-specific CD8⁺ T cells from all mice. Cell identities were assigned based on DEGs: Naïve-like (N), Memory precursors (MP), Early effectors (EE), Cycling effectors (CE), Intermediate effectors (IE), Interferon responsive (IF) and Terminal effector/SLECs (TE). Feature plots of select top DEGs. D. Dot plot of select top DEGs. E. Dot plot summarizing the GSEA of gene clusters identified in Best et al. The color represents normalized enrichment score (NES). Dot size indicates −log₁₀(adjusted p-value). F. RNA velocity field projected onto the UMAP. Direction of arrows marks the direction of differentiation. Arrow length indicates the rate of differentiation (left). Scatter plot of Zeb2 spliced and unspliced mRNA counts per cell where the colors represent cluster identity (right). The dotted line marks the steady-state and gene velocity per cell is a measure of deviation from the steady state where positive velocity means gene induction and negative velocity means repression. ECM=extracellular matrix. See also Figure S1 and Table S1.

Figure 2.. Long-read TCR sequencing reveals multiple VSV-OVA specific T cell clonotypes with differential fate biases.

A. Schematic of Nanopore-based LR-scTCR-seq pipeline. B. Pie charts depict the percentage recovery of CDR3a and CDR3b sequences in the mouse replicates. n: number of T cells in each replicate. C. Overlap of CDR3a chains, CDR3b chains and paired CDR3a and CDR3b clonotypes across replicates. D. Enrichment score of the biased or unbiased TCR clonotypes in the Seurat clusters (Top). n: total number of TCR clonotypes in each category. The dotted line represents ES=1 (no enrichment). Data are individual TCR clonotypes (circles) and mean (bar) +SD. Statistical analysis was performed using one-way ANOVA and Tukey’s multiple comparisons test. **** P < 0.0001. Significance values for major comparisons are shown for clarity. Other significant comparisons: IE vs TE (P <0.001), IE vs N (P <0.0001), EE vs TE (P <0.001), EE vs N (P <0.0001), CE vs N (P <0.01) for memory-biased clonotypes, IF vs CE (P <0.05), IE vs CE (P <0.05), EE vs N (P <0.01) and CE vs N (P <0.0001) for effector-biased clonotypes and N vs all other clusters (P <0.0001) for unbiased clonotypes. All the cells present in each biased category are displayed on the UMAP (bottom). E. and F. The distribution of the biased and unbiased clonotypes E. across MP and TE clusters F. across replicates. G. Volcano plot depicting top DEGs across EE cells from effector-biased TCR clonotypes versus memory-biased TCR clonotypes. H. The DEGs across EE cells were converted into gene modules and the UMAP represents module score per cell. See also Figure S2, S3, S4 and Table S2.

Figure 3.. Biased TCR clones maintain fate preference under different inflammatory cues.

A. UMAP plots showing the members of effector- or memory-biased TCR clonotypes with paired CDR3a _CD3R3b sequences displayed. B. Activation of naive OT1, E1, M1 and M2 TCR-Rg T cells by serial dilutions of OVA(SIINFEKL)-K_b Tetramer-PE. The expression of TCR activation markers was determined by flow cytometry and plotted as mean fluorescent intensity (MFI). Data are from 3 technical replicates. C. Schematic of experimental layout. 5,000 naïve E1, M1, or M2 TCR-Rg T cells (CD45.1^+/+) were co-transferred with OT1 TCR-Tg T cells (CD45.1^+/−) at 1:1 ratio in naïve B6 mice. On the next day, mice were infected intravenously with 10⁶ plaque-forming units (PFU) of VSV-OVA or retro-orbitally with 10⁶ colony-forming units (CFU) of ΔActA Lm-OVA. Six to 8 days later, splenic CD8⁺ T cells were enriched and analyzed for KLRG1 and CD127 expression. D. Relative proportion of CD8⁺CD44^highCD45.1^+/+GFP⁺ TCR-Rg T cells compared to CD8⁺CD44^highCD45.1^+/− OT1 TCR-Tg T cells per mouse. Data are individual mice (circles) with mean (bar) ±SD. Statistical analysis was conducted using one-way ANOVA and Tukey’s multiple comparisons test. E. and F. Representative flow cytometric plots of co-transferred OT1 TCR-Tg (gated on CD44^highCD45.1^+/−) and TCR-Rg T cells (gated on CD44^high GFP⁺CD45.1^+/+) and quantification of MPECs (CD127⁺KLRG1⁻) to SLECs (CD127⁻KLRG1⁺) ratio post-infection with E. VSV-OVA F. ΔActA Lm-OVA. G. Naïve OT1 TCR-Tg T cells were co-transferred with M1 TCR-Rg T cells at 1:10 ratio, and some recipient mice were administered two doses (2hr and 24hr post VSV-OVA infection) of 0.5 μg rIL-12 intraperitoneally. MPEC to SLEC proportions were analyzed 7 days post-infection. For D., E. and F. data are pooled from 2 independent experiments with 3-5 mice per group. For E., F. and G. the circles and lines represent the co-transferred T cells from the same mouse. Statistical analysis was performed by paired T-test. In one experimental replicate in D., E. and F., 1,400 M1 TCR-Rg CD8⁺ T cells were co-transferred at 1:1 ratio with OT1 TCR-Tg T cells instead of 5,000 T cells. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.001. See also Figures S5.

Figure 4.. TCR fate bias correlates with the size of the memory pool.

A. Schematic of experimental layout. Five hundred or 1,000 naïve OT1 TCR-Tg T cells (CD45.1^+/−) were co-transferred with 5,000 naïve E1, M1, or M2 TCR-Rg T cells (CD45.1^+/+) into naïve B6 mice. On the next day, mice were infected intravenously with 3.3 × 10⁵ or 10⁶ PFU of VSV-OVA. 40+ days later, CD8⁺ T cells were enriched from the spleens and salivary glands (SG) of the infected mice and analyzed by flow cytometry. B. The percentage of TCM (CD44^highCD62L⁺CD69⁻), TEM (CD44^highCD62L⁻CD69⁻CD127⁺KLRG1⁻), LLEC (CD44^highCD62L⁻CD69⁻CD127^+/−KLRG1⁺) and TRM (CD44^highCD62L⁻CD69⁺CD103⁺) in the OT1 TCR-Tg and E1, M1, or M2 TCR-Rg T cells was determined from the Spleen and SG of the infected mice. Representative flow plots and quantified data are shown from 2 independent experiments (except E1 data is from one experiment) with 3-5 mice in each group. C. Number of cells in the indicated memory subsets per T cell clone. For B. and C. data are individual samples (circles) and mean (bar) ±SD. Statistical analysis was performed using one-way ANOVA and Tukey’s multiple comparisons test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. See also Figures S6.

Figure 5.. Unified Lineage Trees of OVA-K_b-specific TCR clonotypes.

Lineage trees depicting all OVA-K_b-specific TCR clonotypes for which CARLIN barcode information was captured are combined into a single mega tree per mouse: LA2 (left) and LA3 (right). The roots of individual TCR clonotypes are connected to a central node by green branches. Each node contains a pie chart showing the cell-type proportion of downstream cells. Terminal nodes are colored by their cell identity. Edited CARLIN barcode deletions (red) and insertions (blue) are represented within white bars present on the outer edge of the trees. A sample TCR clonotype (TCR-Clone-X, marked by black arrow) is highlighted in the box and shows how CARLIN edits can be informative of lineage relationships.

Figure 6.. CARLIN barcode edits allow tracking of multiple generations and subclones to reveal fate biases within each TCR clonotype.

Individual lineage trees and UMAP plots of representative TCR clonotypes in A. memory-biased and B. effector-biased categories from LA2 and LA3 replicates. Each node in the tree contains a pie chart showing the cell-type proportion of its daughter cells. The center larger pie chart represents the overall cluster distribution of the clonotype. The outer rectangles represent the members of the TCR clonotype colored by their cell identity. Cdr3a_Cdr3b sequences are displayed for each TCR clone. The UMAP plots show all members of the individual TCR clonotypes.