Developmental features and unique characteristics of peptide-specific PLZF+ innate-like T cells

Innate-like T cells are considered a family of T lymphocytes capable of mounting quick responses, yet restricted to conserved antigens of non-peptide nature. Here, we performed single-cell transcriptomic analyses of peptide-specific PLZF+ innate-like T (PILT) cells and compared them to other members of the innate-like T cell family. Our data revealed that PILT cells share similar transcriptional profiles and overlapping developmental trajectories with invariant Natural Killer T (iNKT) cells as well as mucosal-associated invariant T (MAIT) cells. Yet, in contrast to iNKT and MAIT cells, PILT cells displayed a polyclonal T cell receptor (TCR) repertoire closely resembling the one of conventional CD8 T cells, inferring MHC I restriction and a broader range of antigen specificity. We further show that artificial thymic organoid cultures (ATOC) support selection and development of PILT cells in vitro exhibiting similar transcriptional profiles to their counterparts maturing in the thymus. Lastly, using an “on-time” T-cell receptor retrogenic ATOC system, we demonstrate that PILT1 derived TCR clonotypes favour acquisition of PILT1 innate-like phenotype during T cell development in vitro suggesting an instructive role of TCR specificity in PILT cell lineage commitment and functional differentiation. Overall design: Experiment 1. Single-cell suspensions of thymocytes were prepared from 3 WT mice and 3 MHC-T mice (10 weeks old). All samples were subjected to CD8? depletion, stained with appropriate antibody mix followed by FACS sorting. Sorted cells were mixed in a ratio of 1:4 (iNKT:PILT enriched population) and 20,000 cells were used for GEM1 generation. Experiment 2. Single-cell suspensions of thymocytes were prepared from 3 WT mice and 3 MHC-T mice (10 weeks old) and then split into two equal parts. Part 1 was depleted for CD8? (as in Experiment 1) and Part 2 was not. Both parts were stained with appropriate antibody mix followed by FACS sorting. Sorted cell fractions were then mixed in a ratio of 1:4 (iNKT:PILT-enriched population) and 40,000 cells were used for GEM2 (CD8? depleted) and GEM3 (non-deplete) generation. Experiment 3. Single-cell suspensions of thymocytes were prepared from 3 CD1d-/- mice (10 weeks old), stained with appropriate antibody mix followed by FACS sorting. Sorted cells were then mixed and 40,000 cells were used for GEM4 generation. Experiment 4. For samples week1:5, we collected and pooled three organoids per time point in organoid collection buffer. The sample was then split into two equal parts. Part 1 was stained with an appropriate antibody mix and subjected to FACS sorting for iNKT cells (Live/CD19-/TCRß+/CD1d-PBS57+). Part 2 was stained with an appropriate antibody mix and subjected to FACS sorting for PILT-enriched population (Live/CD19-/TCRb+/CD1d-PBS57-/TCRß+/CD44+ and/or PD-1+). All sorted iNKT cells were then pooled with 40,000 cells from the PILT-enriched population and used for GEM5 generation. Experiment 5. For week 0 sample, FACS sorted DN1-3 thymocytes from 3 WT mice (10 weeks old) were used as shown in (Supplementary Fig. 6B). Samples for weeks 1:5 were collected from the corresponding organoids in an organoid collection buffer and then stained using Zombie Green™ Fixable Viability Kit. FACS sorted live thymocytes from samples from weeks 0:5 were then mixed in equal numbers and 40,000-pooled cells were then used for GEM6 generation. For experiments 1, 2, and 3 the GEMs were generated by Chromium controller (10x Genomics) using Chromium Single Cell V(D)J Reagent Kits (v1.1). For experiments 4 and 5 the GEMs were generated by Chromium Xi (10x Genomics) using Chromium Single Cell 5' Reagent Kits (v2 - Dual Index). The library preparation was carried out as per manufacturer recommendations for each of the kits. Generated sequencing libraries were either sequenced by Illumina's NextSeq 550, NovaSeq 6000 or NovaSeqX with an average read per cell of 40,000 for gene expression libraries and an average of 10,000 reads per cell for VDJ and hashtag libraries. Schematic representation of the layout of all single-cell RNA sequencing experiments is depicted in (Supplementary Fig. 1B and C).