Fig 2: Canine CD5-depleted and sorted CD3-NKp46+ cell populations converge on identical gene expression in the activated state. Differential gene expression analysis of resting (Day 0) CD5-depleted and FACS-isolated CD3-NKp46+ versus activated (Day 14) NKp46+ and expanded CD5-depleted cells as candidate canine NK cell populations. (A) Individual gene counts of canonical NK markers by treatment group: CD5-depleted resting, CD3-NKp46+ resting, CD3-NKp46+ activated and unsorted expanded CD5-depleted activated cells. Sorting on CD5-depleted in the activated state is not possible as expression is lost after 14-days of feeder line co-culture. Resting CD3-NKp46+ cells have higher expression of canonical NK-associated genes than CD5-depleted cellsand therefore may represent a subset of circulating NK cells. (B) Flow cytometry results for CD5dim versus NKp46 staining of the starting population from healthy beagle PBMCs, demonstrating overlapping of the two phenotypes. (C, D, E) MA (ratio intensity) plot, mean of normalized counts by log fold change, compares differentially expressed genes (in red) between treatment conditions. In total, Figure 2B contains 2888 induced genes and 2109 repressed genes, Figure 2C contains 1893 induced genes and 1896 repressed genes, and Figure 2D contains just 1 induced gene and 7 repressed genes which are differentially expressed between the cell populations. Thectivated CD3-NKp46+ cells compared to activated unsorted cells in Figure 2D have very few differentially expressed genes and thus nearly identical transcriptional profiles. (F) Principal component analysis reveals resting CD5-depleted cells drive variance for PC1, while resting CD3-NKp46+ cells drive variance for PC2 (F). Almost no variance between activated CD3-(E). (G) The similarity matrix demonstrates Jaccard index (J(A,B) = |AnB| / |AUB|) approaching 1 (identical) between activated CD3-NKp46+ and CD5-depleted expanded cells based on pre-abundance count hash sketches. This signifies similarity based on fastq sequences prior to alignment to a transcriptome or normalization of counts. (Key: CD5dim_D0 = resting CD5dim cells, NKP46_D0 = resting NKp46+ cells, NKp46_D14 = coculture activated NKp46+ cells, US_D14 = unsorted coculture activated cells expanded from a CD5dim resting population.) (H) Dotplots of gene ontology analysis results (using KEGG) comparing gene pathways induced in resting CD5-depleted cells, resting CD3-NKp46+ cells and activated CD3-NKp46+ cells. Note that pathways induced in CD5-depleted cells include phagocytosis, endocytosis and lysosome functions which have been broadly implicated in innate immune cell pathways including NK cell trogocytosis. Also note resting CD3-NKp46+ cells express the NK cell-mediated cytotoxicity pathway, but also pathways associated with T-cell development. As before, activated CD3-NKp46+ cells express pathways associated with pathways associated with both high replication and cellular dysfunction.

Fig 3: CD5-depleted and CD5dim cells retain gene expression related to broad innate immune cell function when compared to NKp46+ cells in steady-state. DGE was performed between canine cells isolated using CD5-depletion via magnetic separation, cells isolated via FACS for CD5dim and cells isolated via FACS for NKp46+ in the steady-state (at rest). The purpose of this experiment was to compare the gene expression between three most common starting populations described most frequently in the canine NK cell literature. (A) Individual normalized gene counts of canonical NK markers by treatment group. Within these classically NK-associated genes, CD5-depleted and CD5dim expression is similar and much lower than expression in NKp46+ cells, with the exception of PRMD1/BLIMP1 and ZEB2 which is higher in the CD5-depleted and CD5dim populations. (B) A heatmap of the top 30 most differentially expressed genes demonstrates a subset of genes induced in CD5dim cells, compared to the other two populations, none of which are classically associated with NK cell function. CXCL8 is the gene for interleukin-8 and most commonly secreted by macrophages. The G0S2 gene governs the G0/G1 switch and governs cell cycle progression in blood mononuclear cells. FTH1 encodes ferritin heavy chain associated with iron binding. (C) Principal component analysis demonstrates the variance between CD5dim and NKp46+ cells drives PC1 (86%), whereas the difference between CD5-depleted and CD5dim cells drives the variance for PC2 (7%), which is much smaller. (D) Gene-ontology (GO) analysis between the cell populations is depicted as UpSet plots. UpSet plots the intersections of a set as a matrix. In this case, each set on the left represents a gene pathway. The quantity in the bar graph represents the normalized number of counts for that gene pathway. When genes contribute to multiple gene pathways (for example, cell cycle and oocyte meiosis) a bar connects the dots in the matrix. Thus, performing GO analysis the cell pathways on the left of each graph denote the pathways induced in a specific population when compared to a second. Gene overlap between pathways depicted by connected dots. In this case, CD5-depleted and CD5dim cells share induced pathways related to broad innate functions, to include phagosome, lysosome and endocytosis. In contrast, NKp46+ cells appear to have induced pathways involved in T cell function when compared to CD5-depleted and CD5dim cells. None of the populations appear to possess induced pathways related to NK cell function when compared to the others. It is important to note this does not denote the absence of NK cell function in the populations, but rather the absence of significant differences in any NK cell functions between the three populations.

Fig 4: Activated canine NK cells expanded from a CD5-depleted starting population expresses canonical NK markers similar to activated CD3-CD56+ human NK cells. NK cells were isolated via magnetic separation for canine (CD5-depleted) or by RosetteSep NK enrichment for human NK cells (CD3-CD56+). Differential gene expression between resting canine and human NK cells compared to cells exposure to irradiated K562 human feeder cells transduced with 4-1BBL (CD137L) and membrane-bound rh-IL21 (K562C9IL21) supplemented with 100 IU/mL rh-IL2 for 14 days reveals similar profiles. (A) DGE using DESeq2 showed comparable expression levels of GZMA and IFNG in human and canine cells. Locus 490630 on the CanFam3.1 transcriptome was identified as GZMB homolog and its expression is similar in humans and canines. Resting humanCD3-CD56+ cells express more PRF1 mRNA than resting canine CD5dim/- cells. A panel of NK-associated genes demonstrated elevation in the activated state for canine CD5dim cells, with the exception of the canine low affinity IgG Fc Receptor (CD16). Using the CanFam3.1 transcriptome, Locus 486692 and Locus 478984 were identified for the first time as NKG2A/B and CD16 homologs, respectively. (B, C) MA (ratio intensity) plot, mean of normalized counts by log fold change, compares differentially expressed genes (blue = human, red = dog) between treatment conditions. In total, (B) contains 3305 induced genes and 2872 repressed genes and (C) contains 2573 induced genes and 2448 repressed genes. Each species has a similar number of differentially expressed genes. (D, E) Dotplots of gene ontology analysis results (using KEGG) comparing gene pathways induced in humans (D) and canines (E), respectively. Both populations express pathways associated with high cell turnover (eg: cell cycle, RNA transport, DNA replications) and dysfunction (eg: cellular senescence, disease-annotated). With KEGG, disease-annotated pathways denote dysfunction rather than actual pathologic state. (F) A more detailed representation of GO analysis as a cnetplot for activated canine CD5-depleted cells which includes specific gene names and overlap between interacting pathways. Predominant pathways include genes involved in cell cycle, p53 signaling, RNA transport, and ribosomes. Cell cycle and p53 pathways share the greatest number of induced genes.

Fig 5: Single-cell analysis of activated CD5dim and CD3-NKp46+ canine NK cells shows heterogeneity in time-to-response for NK cell activation. DGE following single-cell RNA sequencing of 4 populations: FACS-isolated resting CD5dim and FACS-isolatedCD3-NKp46+ cells from healthy beagle PBMCs, co-culture activated CD3-NKp46+ cells and unsorted activated NK cells expanded from healthy beagle PBMCs using CD5 depletion. All sorting was completed using canine-specific monoclonal antibodies. ScRNASeq was performed using the plate-based 3’ UPX Transcriptome kit with 24 cells/group given the comparative rarity of the NKp46+ group. (A) Resting CD5dim, resting CD3-NKp46+, and activated CD3-NKp46+ populations are depicted as seen by flow cytometry during cell sorting procedure. Sorted populations lie within the designated gate. Note the markedly lower percentage of resting NKp46+ cells compared to the resting CD5dim cells, highlighting that NKp46+ are a very small population of total PBMCs. (B) Dimensionality reduction using t-distributed stochastic neighbor embedding (t-SNE) plot of cells showing tighter clustering and high similarity among the activated CD3-NKp46+ and unsorted cell populations from a less dense distribution of the resting cell populations. Note, however, some activated, unsorted cells cluster with the resting CD5dim and CD3-NKp46+ groups, suggesting that while all “activated” cells were exposed to K562/IL-2 co-culture for a uniform 14 days, biologically there exists heterogeneity in the time to response of single cells within the selected populations. (C) Heatmap depicting genes driving the variance for PC1 and PC2, representing how the gene expression of the combined resting and combined activated populations differ from one another. Resting CD5dim and NKp46+ cells have higher expression of NKC gene KLRD1, and decreased expression of GZMA and CCL5 in comparison to the activated cell populations. (D) Panel of genes and corresponding expression in the single cell populations. Resting cells are in the top left portion of the graphs and activated cells are in the bottom right. Increasing gene expression is denoted by the color gradient. Note relatively constituitive expression of KLRB1, also known as NK1.1, which is a well-established phenotypic marker for murine NK cells and may represent a candidate in canines as well.