Single cell RNA-sequencing of feline peripheral immune cells with V(D)J repertoire and cross species analysis of T lymphocytes

Abstract

Introduction: The domestic cat (Felis catus) is a valued companion animal and a model for virally induced cancers and immunodeficiencies. However, species-specific limitations such as a scarcity of immune cell markers constrain our ability to resolve immune cell subsets at sufficient detail. The goal of this study was to characterize circulating feline T cells and other leukocytes based on their transcriptomic landscape and T-cell receptor repertoire using single cell RNA-sequencing.

Methods: Peripheral blood from 4 healthy cats was enriched for T cells by flow cytometry cell sorting using a mouse anti-feline CD5 monoclonal antibody. Libraries for whole transcriptome, alpha/beta T cell receptor transcripts and gamma/delta T cell receptor transcripts were constructed using the 10x Genomics Chromium Next GEM Single Cell 5' reagent kit and the Chromium Single Cell V(D)J Enrichment Kit with custom reverse primers for the feline orthologs.

Results: Unsupervised clustering of whole transcriptome data revealed 7 major cell populations - T cells, neutrophils, monocytic cells, B cells, plasmacytoid dendritic cells, mast cells and platelets. Sub cluster analysis of T cells resolved naive (CD4+ and CD8+), CD4+ effector T cells, CD8+ cytotoxic T cells and gamma/delta T cells. Cross species analysis revealed a high conservation of T cell subsets along an effector gradient with equitable representation of veterinary species (horse, dog, pig) and humans with the cat. Our V(D)J repertoire analysis demonstrated a skewed T-cell receptor alpha gene usage and a restricted T-cell receptor gamma junctional length in CD8+ cytotoxic T cells compared to other alpha/beta T cell subsets. Among myeloid cells, we resolved three clusters of classical monocytes with polarization into pro- and anti-inflammatory phenotypes in addition to a cluster of conventional dendritic cells. Lastly, our neutrophil sub clustering revealed a larger mature neutrophil cluster and a smaller exhausted/activated cluster.

Discussion: Our study is the first to characterize subsets of circulating T cells utilizing an integrative approach of single cell RNA-sequencing, V(D)J repertoire analysis and cross species analysis. In addition, we characterize the transcriptome of several myeloid cell subsets and demonstrate immune cell relatedness across different species.

Keywords: T cells; T-cell receptor repertoire; V(D)J; cross species analysis; feline; myeloid cells; single cell RNA-sequencing (scRNA-seq).

Figure 1.

scRNA-seq atlas of CD5⁺ enriched circulating feline immune cells revealed 7 major types across 21 clusters. (A) Table of cell counts from each age group. (B) UMAP plot demonstrating the unsupervised clustering results for feline circulating immune cells from 4 healthy cats of different age groups. (C) UMAP plot of global clustering colored by age. (D) Table of cell type counts across clusters by age. (E) Dot plot demonstrating cell type specific marker expression of the 7 major cell types. (F-O) Feature UMAP plots demonstrating expression profile of key markers for each of the 7 different cell types.

Figure 2.

CD5⁺ enriched T cells segregate into naïve T cell subtypes and effectors. Unsupervised clustering of T cells reveals 7 subtypes. (A) UMAP of the scRNA-seq atlas of T cells. (B) UMAP of T cells colored by Pseudotime. (C) Dot plot of marker genes expressed by T cell type. (D) Table of cell type frequencies across ages by cluster. (E) Feature plots of expression and co-expression of TRG and TRDC. (F-Q) UMAP of T cells colored by classical T marker genes defining CD4/CD8 status, naive (SELL, CCR7), effectorness (ANXA2, LGALS3, GZMK, PRF1), terminal differentiation (CCL5) and T exhaustion (TIGIT, PDCD1).

Figure 3.

Feline Effector T cells (T_EM) segregate into transcriptionally similar clusters and reveal the presence of helper T phenotypes. Unsupervised clustering of T_EM reveals 13 subtypes. (A) UMAP of scRNA-seq atlas of T_EM. (B) Dot plot demonstrating up to 3 top differentially expressed genes for each cluster determined via Wilcoxon rank sum testing (Adj P <0.05). (C,E,G,I) UMAP of T_EM colored by helper T subtype gene modules. (D,F,H,J) Expression UMAP of a representative gene from each helper T gene module presented in parallel.

Figure 4.

Cross-species integrative analysis of T cells reveals missing cytotoxic effectors in the cat. Unsupervised clustering revealed 12 clusters across 5 species. (A) UMAP of scRNA-seq atlas of T cells. (B-F) UMAP split by species- dog, horse, cat, human and pig. (G) UMAP of T cells colored by T cell phenotype. (H) Dot plot of marker gene sets for T cell subtypes. (I) Percentage bar chart of clusters stacked by species. (J) UMAP colored by CD5 expression. (K) Bar chart of percentage CD5⁺ cells per cluster. (L) Bar chart of average CD5 expression across cells in each cluster. (M) Scatter plot of percentage CD5⁺ cells per cluster versus number of cat cells within the corresponding cluster.

Figure 5.

(A) Percentage of T cells expressing a certain combination of productive TR chain transcripts. TRA/TRB transcripts dominate in all ab T cell subsets except for CD8⁺ cytotoxic T cells, which primarily express TRA/TRB/TRG transcripts. (B) TRA & TRB V gene usage of the 4 major ab T cell subsets. CD8⁺ cytotoxic T cells show differential usage compared to the other T cell subset but are also substantiated by fewer cells. (C) Junctional length of TRA rearrangements of the 4 major ab T cell subsets. Junctional regions with dominant TRAV genes in CD8⁺ cytotoxic T cells are 13 and 14 amino acids long. (D) TRG transcripts in CD8⁺ cytotoxic T cells have a skewed junctional length with 16 amino acid length (top) and a conserved motif (bottom) (E) The majority of TRG rearrangements in CD8⁺ cytotoxic T cells involved cassette 2 (TRGV2–2/TRGJ2–2). Genes are displayed in order of genomic location and colored by cassette.

Figure 6.

(A) Shared clonotypes across 4 cats. (B) Characterization of clusters based on mean centrality and cluster density. (C) Representative examples of TRA clusters with high centrality and density.

Figure 7.

Feline circulating monocytes include classical monocytes with more differentiated clusters with traditional proinflammatory and unconventional anti-inflammatory phenotypes. Unsupervised clustering revealed 5 clusters. (A) UMAP of scRNA-seq atlas of monocytes. (B-F) UMAP of monocytes colored by expression levels of canonical markers. (G) Dot plot of genes of monocytic phenotypes across the 4 monocytic clusters. (H-J) Dot plots of top GO biological process terms called based on sets of positive differentially expressed genes identified via Seurat FindMarker function (Wilcoxon rank sum, Adj P <0.05). Classical monocyte, CM; Conventional dendritic cell, cDC; Pro-inf, proinflammatory; Anti-inf, anti-inflammatory.

Figure 8.

Feline circulating neutrophils separate into 2 clusters based on activation state. (A) UMAP of scRNA-seq atlas of neutrophils. (B-G) UMAP of neutrophils colored by expression levels of canonical and functional neutrophil markers. (H) Dot plots of top GO biological process terms in cluster 1 called based on sets of positive differentially expressed genes identified via Seurat FindMarker function (Wilcoxon rank sum, Adj P <0.05). (I-K) UMAP of neutrophils colored by expression levels of select differentially upregulated genes in cluster 1. (L,M) Violin plots of RNA counts and feature (unique RNA) counts per cell by cluster; (****) indicates P value less than 0.001 by T-test. (N) UMAP of neutrophils colored by interferon (IFN) gene composite score per cell; genes included are named in figure.