A growing field of interest in immunology and is the study of T cell subtype heterogeneity and T-cell receptor (TCR) diversity. This is of particular interest in immuno-oncology, where a deep understanding of T-cell and TCR diversity is essential to understanding the tumor microenvironment and, importantly, for developing new cancer therapies.
Recent advances in 5’ single-cell sequencing, such as 10X Genomics Chromium Immune Profiling, enables simultaneous profiling of gene expression and TCR V(D)J sequences.
Here, we reanalyze a mutliomic scRNA+scTCR dataset from Luoma et al.. In this manuscript, they investigate immune cell behavior in the colon during CTLA-4 and PD-1/PD-L1 checkpoint inhibitor (CPI) treatment. This is of particular importance because 60% of patients treated with these checkpoint blockades experience severe treatment-limiting toxicities – particularly in the gastrointestinal mucosa. 10-20% of patients treated with CTLA-4 checkpoint inhibitors experience severe, and potentially life-threatening, colon inflammation (a.k.a. colitis).
This dataset contains colon biopsies from n = 8 melanoma patients with colitis, n = 6 melanoma patients without colitis, and n = 8 healthy adults who underwent screening colonoscopies. Of the 6 melanoma patients without colitis, 3 had enteritis (inflammation of the small intestine) and 3 did not.
Following biopsy, the colon tissue was enzymatically digested, FACS sorted into a CD3+ T-cell population, and then sequenced with 10X Genomics 5’ single-cell immune profiling. For 16 of the patients, a CD45+ mononuclear cell population was also sorted and sequenced.
The UMAPs below give some basic information regarding the dataset.
While looking at only the CD45+ separated cells, we see that CPI+ colitis (C) patients have a higher proportion of T-cells than both control (CT) and CPI+ non-colitis (NC) patients.
Additionally, we can see that there are large differences between treatment groups within different celltypes.
Next, we investigate the T-cells in greater detail. We see that the T-cells generally separate into CD8 and CD4 subtypes. However, it isn’t perfect – there are some clusters (e.g. cluster 10) which contain both CD4 & CD8 cells, and there are some clusters which contain neither CD4 or CD8 cells (e.g. cluster 18), or contain only very low expression of a marker gene (e.g. cluster 13).
To remedy this, we instead use Azimuth to annotate the cell types. We found that this worked very well for the initial labeling of CD4/CD8/other T-cell subtypes, but did a poor job of capturing heterogeneity in sub-subtypes. So, we next separated the CD4 and CD8 T-cells based on the Azimuth level 1 annotation, and studied them separately.
We can see that the proportion of CD4+ to CD8+ is relatively consistent between the treatment groups.
First, we analyze the CD8+ Tcells. We see distinct clusters based on known celltype markers, and we see a large difference in cell-type proportions of CPI colitis patients compared to both controls and CPI without colitis. In particular, we see a large increase in the cycling and cytotoxic effector cells in CPI colitis, and a loss of Trm IEL and Trm LP1 cells.
Next, we analyze the CD4+ T-cells. We see distinct clusters based on known celltype markers, and we again see a large difference in cell-type proportions of CPI colitis patients compared to both controls and CPI without colitis. In particular, we see a notable decrease in the proportion of Tissue Resident Memory cells (Trm) and a large increase in the proportion of Th1 effector cells, cycling cells, and regulatory T-cells.
We next investigated the expression of immune checkpoint genes and inhibitory receptors in T-cell subclusters. We see that there is an increased expression of checkpoint-related genes in both CD8 and CD4 T-cells.
We can see from density plots that colitis-enriched celltypes are sources of expression for these genes.
However, in a pseudobulk comparison of colitis vs non-colitis + control samples, we see that CTLA4, HAVCR2, LAG3, and CD38 are significantly upregulated in colitis CD8+ T-cells, but only HAVCR2 and CD38 are upregulated in colitis CD4+ T-cells.
Our pseudobulk analysis also reveals that the colitis results in similar changes in gene expression between CD4+ and CD8+ T-cells.
Now we will turn our attention to the TCR data. Initial analysis of the TCR data reveals that CD4+ cells have a higher number of unique clonotypes, but have a much lower ratio of expanded clonotypes than CD8+ T-cells.
When looking at the number of expanded clonotypes per subcluster, we can see that colitis-enriched celltypes have the highest humber of expanded clonotypes. This is especially prominent in the CD8+ cytotoxic effector cells.
We next wanted to investigate the origin of this colitis-associated CD8+ cytotoxic effector celltype. To do this, we investigated overlap of clonotypes between CD8+ clusters. We see the strongest overlaps between this cluster and the Trm_LP1 & Trm_LP2 clusters, suggesting these clusters are the likely origin for the cytotoxic effector cells.
Finally, we confirm this by visualizing the location of the top 20 most expanded clonotype amino acids in the cytotoxic effector cells.
For further analysis of this dataset, the next step would be to cluster the other (non-Tcell) celltypes, and then look at the interactions between different celltypes. This was done to an extent with myeloid cells in the original manuscript, but the authors did not dive into the large B-cell and Plasma cell clusters. Additionally, utilizing a more-advanced cell-cell communication method like cellchat could provide additional insights and potential therapeutic targets.