Figures and data
A novel bioinformatics approach for the identification of predicted inhibitory receptors.
A. Schematic overview of the bioinformatics pipeline with the number of unique genes and corresponding proteins remaining at every step. All amino acid sequences corresponding to a protein-coding transcript were retrieved from Ensembl. B. Intracellular domains of identified proteins were permutated 10,000 times, and the number of ITIM or ITSM occurrences were compared to the number of ITIMs in the original sequence to determine the likelihood of a specific intracellular domain containing an ITIM or ITSM. Threshold was determined based on known inhibitory receptors, and set at 0.25+ε, with ε being a random number between 0 and 0.01 to better predict the borderline predictions. C. Three-dimensional structure for all proteins was predicted using AlphaFold, and the average model prediction score (pLDDT) was determined for individual each individual ITIM or ITSM in the protein. Proteins with all ITIMs above 80 pLDDT were excluded. For B and C, one protein is plotted for every unique gene symbol for clarity.
Genes encoding previously described ITIM-bearing inhibitory receptors
Number of samples for different resting and activated immune cell subsets
Known and predicted single-spanning inhibitory receptors are expressed in different cell types in the resting state and after activation.
A. Heatmap with normalized expression data for known and predicted single-spanning inhibitory receptor genes in different cell types, in the resting state or after activation. Receptors were considered not expressed (black) when expression was below median overall gene expression in the sample. Data for neutrophils was retrieved from a different source than the other cell types. B. Novel and known receptors were classified into different functional categories based on changes, or lack thereof, in expression after stimulation. Threshold receptors were expressed at resting state, and did not change after activation (i.e., change in expression < 0.5 log2 fold change). Threshold-negative feedback and threshold-disinhibition receptors were defined by > 0.5 log2 fold change up or downregulation, respectively, in expression after activation. Negative feedback receptors were absent in the resting state, but were expressed after activation. C. Upset plot showing the number of single-spanning receptors that are expressed uniquely by individual immune cell subsets, or shared between subsets as indicated by connected circles. Sixty-four genes are expressed in all cell types (not depicted).
Number of single-spanning receptors in different functional categories for each immune cell subset
Single-spanning predicted inhibitory receptors are expressed across a wide variety of tumour infiltrating T cell subsets.
A. Heatmap with row-normalized expression data for known and predicted single-spanning inhibitory receptor genes in different tumour infiltrating T cell subsets. Receptors were considered not expressed in a T cell subset (black) when expression was below median across all subsets. B. Number of inhibitory receptor genes expressed by different CD4+ and CD8+ T cell subsets.
A novel bioinformatics approach for the identification of predicted inhibitory receptors.
A. The overall likelihood of finding an ITIM in the intracellular domain based on the permutation method was plotted as function of total intracellular domain length for known and predicted single-spanner and multi-spanner proteins.
B. The average size of the intracellular domain of proteins that were excluded by the likelihood filter was significantly higher than of those that were included. C. Examples of three-dimensional structure predictions by AlphaFold. Colouring indicates high (red) and low (blue) confidence residues in the protein, ITIMs for PDCD1 (left) and BMP1RB (right) are depicted as purple spheres.
Known and predicted multi-spanning inhibitory receptors are expressed in different cell types in the resting state and after activation.
A. Heatmap with normalized expression data for known and predicted multi-spanning inhibitory receptor genes in different cell types, in the resting state or after activation. Receptors were considered not expressed (grey) when expression was below median overall gene expression in the sample. Data for neutrophils was retrieved from a different source than the other cell types. B. Known and predicted multi-spanning receptors were classified into different functional categories based on changes, or lack thereof, in expression after stimulation. Threshold receptors were expressed at resting state, and did not change after activation (i.e., change in expression < 0.5 log2 fold change). Threshold-negative feedback and threshold-disinhibition receptors were defined by > 0.5 log2 fold change up or downregulation, respectively, in expression after activation. Negative feedback receptors were absent in the resting state, but expressed after activation. C. Upset plot showing the number of multi-spanning receptors that are expressed uniquely by individual immune cell subsets, or shared between subsets as indicated by connected circles. Seventy-five genes are expressed in all cell types (not depicted). Cell subsets without uniquely expressed putative inhibitory receptors, i.e., B cells and T cells, were excluded from the panel for clarity.
Multi-spanning predicted inhibitory receptors are expressed across a wide variety of tumour infiltrating T cell subsets.
A. Heatmap with row-normalized expression data for known and predicted multi-spanning inhibitory receptor genes in different tumour infiltrating T cell subsets. Receptors were considered not expressed in a T cell subset (black) when expression was below median across all subsets. B. Number of inhibitory receptor genes expressed by different CD4+ and CD8+ T cell subsets.
