Abstract
T cell development and selection are coordinated in the thymus by a specialized niche of diverse stromal populations1,2,3. Although much progress has been made over the years in identifying the functions of the different cell types of the thymic stromal compartment, there is no comprehensive characterization of their diversity and heterogeneity. Here we combined massively parallel single-cell RNA-sequencing4,5, spatial mapping, chromatin profiling and gene targeting to characterize de novo the entire stromal compartment of the mouse thymus. We identified dozens of cell states, with thymic epithelial cells (TECs) showing the highest degree of heterogeneity. Our analysis highlights four major medullary TEC (mTEC I–IV) populations, with distinct molecular functions, epigenetic landscapes and lineage regulators. Specifically, mTEC IV constitutes a new and highly divergent TEC lineage with molecular characteristics of the gut chemosensory epithelial tuft cells. Mice deficient in Pou2f3, a master regulator of tuft cells, have complete and specific depletion of mTEC IV cells, which results in increased levels of thymus-resident type-2 innate lymphoid cells. Overall, our study provides a comprehensive characterization of the thymic stroma and identifies a new tuft-like TEC population, which is critical for shaping the immune niche in the thymus.
This is a preview of subscription content, access via your institution
Access options
Access Nature and 54 other Nature Portfolio journals
Get Nature+, our best-value online-access subscription
$29.99 / 30 days
cancel any time
Subscribe to this journal
Receive 51 print issues and online access
$199.00 per year
only $3.90 per issue
Buy this article
- Purchase on Springer Link
- Instant access to full article PDF
Prices may be subject to local taxes which are calculated during checkout
Similar content being viewed by others
References
Klein, L., Kyewski, B., Allen, P. M. & Hogquist, K. A. Positive and negative selection of the T cell repertoire: what thymocytes see (and don’t see). Nat. Rev. Immunol. 14, 377–391 (2014).
Abramson, J. & Anderson, G. Thymic epithelial cells. Annu. Rev. Immunol. 35, 85–118 (2017).
Takahama, Y., Ohigashi, I., Baik, S. & Anderson, G. Generation of diversity in thymic epithelial cells. Nat. Rev. Immunol. 17, 295–305 (2017).
Jaitin, D. A. et al. Massively parallel single-cell RNA-seq for marker-free decomposition of tissues into cell types. Science 343, 776–779 (2014).
Paul, F. et al. Transcriptional heterogeneity and lineage commitment in myeloid progenitors. Cell 163, 1663–1677 (2015).
Takada, K. & Takahama, Y. Positive-selection-inducing self-peptides displayed by cortical thymic epithelial cells. Adv. Immunol. 125, 87–110 (2015).
Wong, K. et al. Multilineage potential and self-renewal define an epithelial progenitor cell population in the adult thymus. Cell Rep. 8, 1198–1209 (2014).
Galliano, M. F. et al. Characterization and expression analysis of the Spink5 gene, the mouse ortholog of the defective gene in Netherton syndrome. Genomics 85, 483–492 (2005).
Hale, L. P. & Markert, M. L. Corticosteroids regulate epithelial cell differentiation and Hassall body formation in the human thymus. J. Immunol. 172, 617–624 (2004).
Lara-Astiaso, D. et al. Chromatin state dynamics during blood formation. Science 345, 943–949 (2014).
Buenrostro, J. D., Giresi, P. G., Zaba, L. C., Chang, H. Y. & Greenleaf, W. J. Transposition of native chromatin for fast and sensitive epigenomic profiling of open chromatin, DNA-binding proteins and nucleosome position. Nat. Methods 10, 1213–1218 (2013).
Brennecke, P. et al. Single-cell transcriptome analysis reveals coordinated ectopic gene-expression patterns in medullary thymic epithelial cells. Nat. Immunol. 16, 933–941 (2015).
Meredith, M., Zemmour, D., Mathis, D. & Benoist, C. Aire controls gene expression in the thymic epithelium with ordered stochasticity. Nat. Immunol. 16, 942–949 (2015).
Gerbe, F. et al. Distinct ATOH1 and Neurog3 requirements define tuft cells as a new secretory cell type in the intestinal epithelium. J. Cell Biol. 192, 767–780 (2011).
Gerbe, F. et al. Intestinal epithelial tuft cells initiate type 2 mucosal immunity to helminth parasites. Nature 529, 226–230 (2016).
Gerbe, F., Brulin, B., Makrini, L., Legraverend, C. & Jay, P. DCAMKL-1 expression identifies tuft cells rather than stem cells in the adult mouse intestinal epithelium. Gastroenterology 137, 2179–2180 (2009).
Tanimizu, N., Nishikawa, Y., Ichinohe, N., Akiyama, H. & Mitaka, T. Sry HMG box protein 9-positive (Sox9+) epithelial cell adhesion molecule-negative (EpCAM−) biphenotypic cells derived from hepatocytes are involved in mouse liver regeneration. J. Biol. Chem. 289, 7589–7598 (2014).
Gury-BenAri, M. et al. The spectrum and regulatory landscape of intestinal innate lymphoid cells are shaped by the microbiome. Cell 166, 1231–1246 (2016).
Kim, D., Langmead, B. & Salzberg, S. L. HISAT: a fast spliced aligner with low memory requirements. Nat. Methods 12, 357–360 (2015).
Scialdone, A. et al. Computational assignment of cell-cycle stage from single-cell transcriptome data. Methods 85, 54–61 (2015).
Sansom, S. N. et al. Population and single-cell genomics reveal the Aire dependency, relief from Polycomb silencing, and distribution of self-antigen expression in thymic epithelia. Genome Res. 24, 1918–1931 (2014).
Lyubimova, A. et al. Single-molecule mRNA detection and counting in mammalian tissue. Nat. Protoc. 8, 1743–1758 (2013).
Buenrostro, J. D., Wu, B., Chang, H. Y. & Greenleaf, W. J. ATAC-seq: a method for assaying chromatin accessibility genome-wide. Curr. Protoc. Mol. Biol. 109, 21.29.1–21.29.9 (2015).
Langmead, B., Trapnell, C., Pop, M. & Salzberg, S. L. Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol. 10, R25 (2009).
Heinz, S. et al. Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol. Cell 38, 576–589 (2010).
Li, Q., Brown, J. B., Huang, H. & Bickel, P. J. Measuring reproducibility of high-throughput experiments. Ann. Appl. Stat. 5, 1752–1779 (2011).
Acknowledgements
I.A. is supported by the Chan Zuckerberg Initiative, the HHMI International Scholar award, the European Research Council Consolidator Grant (ERC-COG)724471-HemTree2.0, the Israel Science Foundation (703/15), the Ernest and Bonnie Beutler Research Program of Excellence in Genomic Medicine, the Helen and Martin Kimmel award for innovative investigation, a Minerva Stiftung research grant, the Israeli Ministry of Science, Technology, and Space, the David and Fela Shapell Family Foundation, the NeuroMac DFG/Transregional Collaborative Research Center Grant, and the Abramson Family Center for Young Scientists. I.A. is the incumbent of the Alan and Laraine Fischer Career Development Chair. J.A. is supported by the ERC-2016-CoG-724821, Israel Science Foundation (1796/16 and 722/14), the Sy Syms Foundation; US–Israel Binational Foundation, Maurice and Vivienne Wohl Charitable Foundation; Goodman Family Charitable Lead Annuity Trust; Ruth and Samuel David Gameroff Family Foundation. J.A. is an incumbent of the Dr. Celia Zwillenberg-Fridman and Dr. Lutz Fridman Career Development Chair; A.G. is a recipient of the Clore fellowship; N.T., P.J. and V.Z. are supported by ANR-17-CE15-TUFTEFF, the Labex EpiGenMed and SIRIC Montpellier Cancer Grant INCa_Inserm_DGOS_12553; M.P. is supported by the Labex EpiGenMed and the FRM; F.G. and V.Z. are supported by CNRS; and N.T. and P.J. are supported by Inserm. We thank M. Boyer-Clavel and S. Gailhac of Montpellier Rio Imaging for cell sorting and the RAM animal facility of the IGMM.
Author information
Authors and Affiliations
Contributions
C.B., S.N., A.G., N.K., J.A. and I.A. designed the project, planned the experiments and wrote the manuscript. C.B., S.N. and N.K. performed experiments, A.G. and E.D. analysed the data, M.P., F.G., A.M., N.T., P.J. and V.Z. performed intestinal tuft and Pou2f3 knockout mouse experiments. A.C. generated the Csnb-reporter mouse. B.T. and S.I. contributed to the single-molecule RNA fluorescence in situ hybridization experiment. O.G. provided human samples.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data figures and tables
Extended Data Fig. 1 Single-cell data quality controls.
a, Summary of all single cells analysed in this study, divided into experimental procedures. ‘nbatches’ indicates number of technical replicates; ‘ncells’ indicates number of cells after filtering (see Methods). b–e, Colour-coded tracks summarizing the number of Illumina reads per cell (b), transcripts (UMI) detected in each cell (c), fraction of analysed cells from each amplification batch (d) and estimation of technical noise for each amplification batch (e). Cells are coloured by experimental procedure. Technical noise is assessed by genomic UMIs in empty wells as previously described4 (see Methods).
Extended Data Fig. 2 Thymic stroma sorting and clustering.
a, Flow cytometry schematic of thymic cells showing isolation of stroma cells, as well as staining for known populations markers. Immune cells, CD45; fibroblast, CD34; endothelial, CD31; mTEC, UEA1; cTEC, LY51; mature mTEC, MHC-II. The red border marks stroma single-cell sorting gate. b, Index sort tracks showing the intensity of protein levels for MHC-II, UEA1 and LY51 in individual single cells shown in Fig. 1a–c. c, Cell–cell correlation of CD45− thymic stroma cells calculated over 132 differentially expressed genes. d, Pairwise distance distribution between cells within the three main stromal lineages. Distance is defined as 1 − Spearman. Box plots display median bar, first–third quantile box and 5th–95th percentile whiskers. n = 1,825 single cells. e, f, Gene expression profiles of 723 cells from fibroblast clusters of the CD45− stroma, marked by increased expression of Col1a1 (e, Supplementary Table 1) and 221 cells from the endothelial clusters of the CD45− stroma, marked by increased expression of Pecam1 (f, Supplementary Table 1).
Extended Data Fig. 3 Thymic epithelial cells are characterized by four subsets of mTEC and a single cTEC subset.
a, Heat map showing a metacell analysis of 2,341 thymic epithelial cells (CD45−EpCAM+), featuring the 73 most variable genes, from 15 biological replicates of 4–6-week-old mice. Colour bar represents separation of 36 metacells into five main populations. b, Two-dimensional graph representation of the metacell model in Fig. 1a (see Methods). Big circles represent metacells, and are colour-coded as shown in a. c, FACS index sorting measurement of LY51 and UEA1 in epithelial cells. Cells are coloured based on cluster association as determined in a. Dashed lines outline LY51+UEA1− and LY51−UEA1+ gates. d, Fraction of TEC subsets out of LY51+UEA1− and LY51−UEA1+ populations, assessed by gating single cells on index sorting protein measurements of UEA1 and LY51 in c. e, Controlling for batch effect as determined by the relative share of each batch in all metacells. Batches are ordered by biological replicate (marked by dashed lines) and sorting scheme (either CD45− or CD45−EpCAM+). f, g, Single molecule FISH assay on 5–8-μm-thick cryosections (see Methods) using fluorescent probes against the genes Epcam and Sbsn (f) or Avil (g). Blue, DAPI. The experiments were repeated independently four times with similar results. h, Immunofluorescence images of the protein markers: PIGR and DCLK1. Medulla (M) and cortex (C) are separated by dashed lines, distinguished by nuclei density. Blue, DAPI. f–h, Scale bars, 20 μm. The experiments were repeated independently twice with similar results. i, Projection of representative differentially expressed genes onto the two-dimensional graph of epithelial cells.
Extended Data Fig. 4 TEC dynamics during thymus development.
a, Flow cytometry scheme of thymic epithelial cells from different development time points. Numbers indicate fraction of CD45−EpCAM+ cells. b, Summary of the K (K = 50) nearest neighbours of embryonic cells, grouped into metacells. Neighbours of adult origin are coloured by TEC subsets, and of embryonic origin are coloured by developmental time point. Metacells with more than 20% adult neighbours were assigned to a TEC subset. c, Normalized mean expression of differential genes across TEC mature populations (4 weeks old) and unassigned cells from developmental time points (grey). n = 4 (E14.5), 3 (E18.5) and 2 (6 days) independent animals. Data are median (bars) and individual animals (dots). d, Differential gene expression between early cTEC (e-cTEC) from three developmental time points and the mature cTEC. Axes represent UMI count per 1,000 UMI, normalized to cell numbers. e, Distribution of cell cycle gene expression across cells from developing TEC is bimodal. The red line indicates the empirical proliferation threshold. f, Frequency of proliferating cells in the epithelial population at each developmental time point. Colour code as in b. g, Gene pairwise Spearman correlation over 2,319 e-cTEC single cells reveals three gene modules jointly expressed across embryonic early TEC and mature cTEC populations. h, GO annotations enrichment analysis of the three cTEC gene modules. For cell cycle, n = 114; antigen presentation, n = 33; and antigen processing, n = 56.
Extended Data Fig. 5 Genetic and epigenetic characterization of TEC subsets.
a, In silico gating of mTEC I–IV populations by index sorting measurements of surface markers. The same gating schemes were used to purify these populations by FACS (Fig. 2a). Cells are colour-coded as shown in Fig. 1. Blue, cTEC; light blue, mTEC I; red, mTEC II; yellow, mTEC III; green, mTEC IV. b, Relative enrichment (log2 fold change compared to total CD45−EpCAM+ epithelial cells) of the individual mTEC I–IV subsets gated according to a. c, Heat map showing pairwise Spearman correlation of 29,472 H3K4me2 peaks (top) or ATAC-seq peaks (bottom) from mTEC I–IV sorted populations. Biological replicates for each population are shown. d, Scatter plots of mTEC I–IV H3K4me2 ChIP–seq peaks in biological duplicates. e, Summary of the most significant motifs enriched in each cluster of mTEC I–IV H3K4me2 differential peaks (Fig. 2e). P values are derived from binomial tests after FDR correction for multiple hypotheses. n = 2,302 differential peaks.
Extended Data Fig. 6 Characterization of AIRE-dependent mTEC subsets.
a, Variance of genes plotted against their mean value (genes with >50 total UMI are shown). Orange dots indicate variable genes. b, Pairwise Pearson gene correlations in AIRE-dependent (left) and AIRE-independent (right) TRA gene lists across 2,341 TEC single cells. Levels of differential expression (highest change of expression in cluster compared to median across all clusters) are indicated as bars; bar colours indicate cluster association to TEC population. c, Comparison of stochastic gene expression between Aire knockout and wild-type cells in mTEC III and IV populations. Marginal distribution is shown as histogram. Axes represent UMI count per 1,000 UMI, normalized to cell numbers. d, Flow cytometry scheme of thymic Aire knockout cells showing the percentage of each TEC population compared to wild-type percentage (shown in brackets). e, Representative immunofluorescence images of two independent experiments, for tdTomato across different organs. In the thymus, the medulla (M) and cortex (C) are separated by dashed lines, distinguished by nuclei density. Blue, DAPI. Scale bars, 100 μm. f, qPCR analysis of Csnb (left) and cre (right) genes in Csnbcre+Rosa26tdTomato (cre+) and Csnbcre−Rosa26tdTomato (cre−) across thymic populations. Dot plots display mean and error bars indicate s.e.m. n = 2 (wild type) or n = 3 (Csnbcre+) biologically independent animals. g, Flow cytometric analysis of tdTomato expression in mTEC subsets (colours as in Fig. 1) isolated from thymi of Csnbcre+Rosa26tdTomato- or Csnbcre−Rosa26tdTomato-reporter mice.
Extended Data Fig. 7 Comparing intestinal tuft cells and the mTEC IV population.
a, Flow cytometry scheme of mTEC IV cells (Sox9-eGFP+L1CAM+) sorting. b, Flow cytometry scheme of small intestine Hpgds-tdTomato+ tuft cell sorting. c, Heat map showing gene expression profiles across 1,903 intestinal tuft (Hpgds-tdTomato+) single cells, grouped into 68 metacells. d, Comparison of gene expression between tuft cells isolated from small intestine (Hpgds-tdTomato+; x axis) and mTEC IV cells isolated from thymus (CD45−EpCAM+Sox9-eGFP+L1CAM+; y axis). Axes represent UMI count per 1,000 UMI, normalized to cell numbers. e, Normalized mean expression of differential genes across TEC populations, sorted mTEC IV cells (L1CAM+ Sox9-eGFP+) and intestinal Tuft (Hpgds-tdTomato+) cells. f, GO annotations enrichment in differential genes (fold change >2) between intestinal tufts (n = 634) and mTEC IV cells (L1CAM+Sox9-eGFP+) (n = 1,308).
Extended Data Fig. 8 The transcription factor Pou2f3 is a master regulator of mTEC IV.
a, Flow cytometry scheme for sorting of EpCAM+ cells from Pou2f3 knockout thymi. b, Projection of representative TEC subtype-specific markers onto the two-dimensional mapping of Pou2f3 wild-type and knockout cells to the epithelial model of Fig. 1 (see Methods). c, Bar plot showing log2 fold change between TEC subpopulation abundances in Pou2f3 knockout (n = 451 single cells) and wild-type (n = 1121) mice. Error bars represent 95% confidence intervals. d, Pooled expression of mTEC IV genes across cells from Pou2f3 knockout and wild-type (WT) mice. The x axis represents UMI count per 1,000 UMI; y axis represents fraction of expression from total UMI count of the cells. Green cells indicate cells classified as mTEC IV (two-sided binomial test; FDR-adjusted P < 10−10). n = 1,572 single cells. e, Differential gene expression between mTEC I–III cells isolated from control (wild-type) and Pou2f3 knockout mice. Axes represent UMI count per 1,000 UMI, normalized to cell numbers.
Extended Data Fig. 9 mTEC IV shape the thymus immune niche.
a, Heat map showing a metacell analysis of 3,500 CD45+IL-25R+ cells across five clusters. b, Differential gene expression in ILC2 cells from two biological replicates compared to other CD45+IL-25R+ cells (rest) from each replicate. Axes represent log2 fold change. c, Percentages of CD45+ cells in Pou2f3 knockout and wild-type thymi, determined by flow cytometry. Circles and diamonds indicate independent mice, centre line indicates the mean value. d, Flow cytometry analysis of cells expressing CD4, CD8, CD25 and CD44 in Pou2f3 knockout and wild-type thymi. The experiment was repeated independently four times with similar results to confirm reproducibility. e, Flow cytometry sorting scheme of thymic CD45+IL-25R+ cells from Pou2f3 knockout and wild-type thymi. f, Percentage of CD45+IL-25R+ cells in Pou2f3 knockout and wild-type thymi. Circles and diamonds indicate independent mice, centre line indicates the mean value. g, Percentages (left) and numbers (right) of ILC2 (Lin−TCR−CD127+GATA3+Rorγt−) cells within the single-cell gate in Pou2f3 knockout and wild-type mice. Circles and diamonds indicate independent mice, centre line indicates the mean value. A one-tailed Student’s t-test was used for the comparison, *P < 0.05.
Supplementary information
Supplementary Information
This file contains Supplementary Note 1: MetaCell - Correcting and clustering single cell RNA-seq data using k-nn graph covering.
Supplementary Table 1
Average gene expression across CD45− thymic fibroblasts and endothelial cells.
Supplementary Table 2
Average gene expression across CD45− thymic epithelial cells.
Supplementary Table 3
Average gene expression across the development of thymic epithelial cells (E14.5, E18.5 & 6 days PN).
Supplementary Table 4
Definition of stochastic genes and TRA.
Rights and permissions
About this article
Cite this article
Bornstein, C., Nevo, S., Giladi, A. et al. Single-cell mapping of the thymic stroma identifies IL-25-producing tuft epithelial cells. Nature 559, 622–626 (2018). https://doi.org/10.1038/s41586-018-0346-1
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1038/s41586-018-0346-1
This article is cited by
-
Mitochondrial protein C15ORF48 is a stress-independent inducer of autophagy that regulates oxidative stress and autoimmunity
Nature Communications (2024)
-
The Proteostasis of Thymic Stromal Cells in Health and Diseases
The Protein Journal (2024)
-
Expression of FOXI1 and POU2F3 varies among different salivary gland neoplasms and is higher in Warthin tumor
Discover Oncology (2024)
-
New insights into the function of Interleukin-25 in disease pathogenesis
Biomarker Research (2023)
-
An exploratory study for tuft cells in the breast and their relevance in triple-negative breast cancer: the possible relationship of SOX9
BMC Cancer (2023)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.
