Indispensable epigenetic control of thymic epithelial cell development and function by polycomb repressive complex 2 - Nature
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
ARTICLE https://doi.org/10.1038/s41467-021-24158-w OPEN Indispensable epigenetic control of thymic epithelial cell development and function by polycomb repressive complex 2 Thomas Barthlott1,8, Adam E. Handel 2,8, Hong Ying Teh1,8, Rushika C. Wirasinha3, Katrin Hafen1, Saulius Žuklys1, Benoit Roch4, Stuart H. Orkin 5, Jean-Pierre de Villartay 4, Stephen R. Daley3,7 & Georg A. Holländer 1,6 ✉ 1234567890():,; Thymic T cell development and T cell receptor repertoire selection are dependent on essential molecular cues provided by thymic epithelial cells (TEC). TEC development and function are regulated by their epigenetic landscape, in which the repressive H3K27me3 epigenetic marks are catalyzed by polycomb repressive complex 2 (PRC2). Here we show that a TEC-targeted deficiency of PRC2 function results in a hypoplastic thymus with reduced ability to express antigens and select a normal repertoire of T cells. The absence of PRC2 activity reveals a transcriptomically distinct medullary TEC lineage that incompletely off-sets the shortage of canonically-derived medullary TEC whereas cortical TEC numbers remain unchanged. This alternative TEC development is associated with the generation of reduced TCR diversity. Hence, normal PRC2 activity and placement of H3K27me3 marks are required for TEC lineage differentiation and function and, in their absence, the thymus is unable to compensate for the loss of a normal TEC scaffold. 1 Department of Biomedicine and University Children’s Hospital of Basel, University of Basel, Basel, Switzerland. 2 Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, UK. 3 Infection and Immunity Program, Monash Biomedicine Discovery Institute and Department of Biochemistry and Molecular Biology, Monash University, Melbourne, VIC, Australia. 4 Genome Dynamics in the Immune System Laboratory, INSERM UMR 1163, Université de Paris, Imagine Institute, Paris, France. 5 Department of Pediatric Oncology, Dana-Farber Cancer Institute and Division of Hematology/ Oncology, Boston Children’s Hospital, Harvard Stem Cell Institute, Harvard Medical School, and Howard Hughes Medical Institute, Boston, MA, USA. 6 Department of Paediatrics and the Weatherall Institute of Molecular Medicine, University of Oxford, Oxford, UK. 7Present address: School of Health and Biomedical Sciences, Queensland University of Technology, Brisbane, QLD, Australia. 8These authors contributed equally: Thomas Barthlott, Adam E. Handel, Hong Ying Teh. ✉email: georg.hollander@paediatrics.ox.ac.uk NATURE COMMUNICATIONS | (2021)12:3933 | https://doi.org/10.1038/s41467-021-24158-w | www.nature.com/naturecommunications 1
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24158-w T he controlled balance between cell proliferation and dif- Cre recombinase is expressed under the control of Psmb11 reg- ferentiation is essential to maintain a normal thymic ulatory elements25,26. The resultant Eedfl/fl::β5tCre mice displayed microenvironment to promote the development and a small thymus with a reduced total cellularity, a normal corti- selection of naive T cells with a repertoire purged of vital “self” comedullary segregation and a number of small, cell-free cysts specificities but prepared to react to injurious “non-self”. Thymic surrounded by cytokeratin 8-positive epithelia (Fig. 1a–d). epithelial cells (TECs), which can be categorized based on specific However, the total number of TEC recovered from these mice molecular, structural and functional characteristics into separate was similar to that of wild-type and Eedfl/fl mice (Fig. 1e). EED cortical (cTECs) and medullary (mTECs) lineages, are essential deletion resulted in the expected depletion of H3K27me3 marks for this competence1. Both TEC lineages derive from a common in all cTECs and the majority of mTECs isolated from Eedfl/fl:: epithelial precursor of endodermal origin2,3 that expresses a range β5tCre mice, whereas H3 histone abundance remained unchan- of cTEC-specific markers including the proteasome component ged (Fig. 1f and see below; gating strategy in Supplementary β5t encoded by Psmb114–7. Fig. 1). cTECs induce the commitment of blood-borne precursor cells Eedfl/fl::β5tCre mice had a higher frequency and cellularity of to a T cell fate, foster their subsequent growth and initial cTECs but a lower rate and reduced cell number of mTECs maturation, and select immature thymocytes that express a T cell (Fig. 2a). The difference was present at birth and steadily antigen receptor (TCR) with sufficient affinity for self-peptides increased with age. Noticeably, cTEC cellularity did not diminish presented by major histocompatibility complexes (pMHC). In in adolescent Eedfl/fl::β5tCre mice contrary to controls. Moreover, contrast, mTECs promote the terminal differentiation of thy- the maturation of mTEC was compromised in Eed:fl/fl:β5tCre mocytes, which includes the establishment of immunological mice with fewer immature epithelia (MHCIIlo, AIRE−, designated tolerance to tissue-restricted antigens (TRAs) via two com- mTEClo) attaining a mature (MHChi; mTEChi) cell stage, either plementary mechanisms; a recessive activity of deleting thymo- positive or negative for the expression of AIRE (Fig. 2b). cytes expressing a TCR with high affinity for these antigens, and a Heterozygosity for a loss of Eed did not compromise thymus dominant mechanism of clonal diversion that generates thymic cellularity nor affect TEC numbers or subset composition regulatory T cells8. In addition, thymus immigrating dendritic (Fig. 2c). The proliferation rates of both cTEC and mTEC were cells (DCs) and B cells provide peripheral antigens for central significantly reduced in Eedfl/fl::β5tCre mice (Fig. 1h and tolerance induction9–11. Supplementary Fig. 1) rendering it an unlikely explanation for The ectopic representation of a significant number of TRAs by a the disparity in cTEC:mTEC ratios. Thus, the loss of PRC2 subpopulation of phenotypically mature mTECs is in part depen- activity (consequent to an absence of either EED or EZH1/2; dent on the expression of the transcriptional facilitator autoimmune Supplementary Fig. 2) resulted in an expansion of cTECs but a regulator (AIRE)12, which acts independently of lineage-specific reduction in mTECs. One possible explanation for this finding transcription factors13. Individual mTECs express AIRE-dependent could be that TEC precursors are preferentially differentiating TRAs to an extent significantly lower than that of corresponding into the cTEC lineage. peripheral tissues14,15 and in co-expression patterns that change with mTEC maturation16–19. AIRE limits the amplitude of TRA gene expression by repressing chromatin accessibility at distal cis- T cell differentiation in Eedfl/fl::β5tCre mice is affected at early regulatory sequences20. These regions, ostensibly placed within and late stages. We next evaluated the thymopoiesis of 3-week- facultative heterochromatin, are enriched in the repressive histone old Eedfl/fl and Eedfl/fl::β5tCre mice (Supplementary Fig. 3a, b). In mark, H3K27me3, and depleted for histone modifications permis- the absence of EED, the proportion of cells in the CD4−CD8− sive for conventional gene transcription8. double-negative (DN) thymoycte compartment increased 3–4- H3K27me3 is a posttranslational modification catalysed by fold (2.5 ± 0.4% vs. 8.1 ± 0.9%; Fig. 3a). This relative increase was polycomb repressive complex 2 (PRC2) that depends for its mainly due to the presence of B cells although their absolute function on four core factors, including EED (embryonic ecto- numbers were not altered (4.8 ± 0.6 × 105 in Eedfl/fl::β5tCre and derm development), which is critical for physically binding to 3.2 ± 1.3 × 105 in Eedfl/fl). The relative number of early thymic H3K27me3, and both EZH1 and EZH2 (enhancer of zeste progenitors (ETP, defined as DN CD44hickithiSca-1posCD25−) homologue), which mediate the trimethylation of H3K2721,22. and DN2 (DN CD44hickithiSca-1pos CD25+) cells in relation to Compromised gene repression due to a loss of PRC2 activity23 all thymocytes of the αβ T cell lineage (i.e. excluding cells positive impairs tissue specification and maintenance, and consequently for TCRγδ, CD19, CD11b, CD11c, DX5, NK1.1, Gr1, F4/80, results in early embryonic lethality24. Ter119 and MHCII) was comparable in mutant and wild-type To establish a specific role for PRC2 in TEC development and mice (Fig. 3b and Supplementary Fig. 3a). We observed, however, function, we analyse mice with a loss of this complex targeted to a partial block at the DN3 (i.e. DN CD44−CD25+) β-selection TEC precursors and occurring after the formation of the thymus checkpoint. Compared to DN2 cells, DN3 thymocytes exhibited anlage. Absent PRC2 activity in TEC leads to an impairment in an almost 2-fold reduction in CD25 expression, which was par- canonical TEC development and function whilst in parallel an alleled by a significant reduction in the frequency of CD71+ pre- alternative mTEC differentiation pathway emerges. These double-positive (preDP) thymocytes in the ckitneg DN sub- abnormalities are associated with reduced TCR diversity in the population (27.5 ± 2.1% vs. 38.7 ± 2.8). The progression through CD4+ and CD8+ naive T cell pools, yet negative selection of self- subsequent maturation stages (i.e. CD71+CD8+ immature single- reactive TCR specificities remained intact. Taken together, we positive (ISP) and CD71+DP) was, however, not affected in conclude that PRC2 activity is essential for TEC lineage specifi- mutant mice (Fig. 3c). Hence, these results indicate a partial block cation and function. at the β-selection checkpoint caused by an EED deficiency in TEC. Whether the reduction in CD25 expression in DN3 is causally linked to the observed decrease in the generation of Results preDP thymocytes remains to be determined. Loss of EED in TEC differentially disrupts lineage develop- We next analysed positive selection into the single-positive ment. To generate mice with a loss of PRC2 activity in all thymic (SP) thymocytes. Although the frequencies of positively selected epithelia at a time after initiation of a thymus anlage, we interbred DPdull cells were unchanged, we detected in Eed:fl/fl:β5tCre mice animals harbouring a conditional Eed allele to mice in which the an accumulation of both CD8SP and CD4SP (Treg excluded) 2 NATURE COMMUNICATIONS | (2021)12:3933 | https://doi.org/10.1038/s41467-021-24158-w | www.nature.com/naturecommunications
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24158-w ARTICLE a b c Eedfl/fl Eedfl/fl::β5tCre *** Thymus cellularity (x108) *** Eedfl/fl 3.5 *** 3 Eedfl/fl:: 2.5 *** *** β5tCre 2 *** 1.5 Eedfl/fl:: 1 *** β5tCre 0.5 0 Eedfl/fl 0 1 4 2mm 200µm 200µm Time (weeks) d e Eedfl/fl Eedfl/fl::β5tCre *** *** TEC cellularity (x105) *** *** Eedfl/fl 4 8 *** T EC fr e q ue n c y (% ) ** CK14 CK8 Aire 3 *** 6 Eedfl/fl 0.15 ±0.05 2 4 Eedfl/fl:: *** β5tCre Eedfl/fl::β 5tCre 105 1 2 CD45 A700 104 10 3 0 0 0 1 4 0 1 4 102 0 2.81 Time (weeks) Time (weeks) ±0.63 CK14 CK8 Aire DAPI 0 102 103 104 105 EpCAM PerCp-Cy5.5 f cTEC mTEC CD45+ H3K27me3 % Max 0 103 104 105 Total H3 Eed fl/fl Eed fl/fl::β5tCre IgG control Fig. 1 Thymus phenotype of Eedfl/fl::β5tCre mice. a Absolute thymus cellularity of mutant (grey diamonds) and control mice (black circles) at the indicated postnatal ages (week 0 relates to newborn). b Macroscopic and c microscopic comparison of thymic lobes isolated from 4-week-old Eedfl/fl:: β5tCre and control Eedfl/fl mice. Scale bars represent 2 mm and 200 µm, respectively. Tissue sections were stained with haematoxylin and eosin. d Immunohistology of thymus sections prepared from 4-week-old Eedfl/fl and Eedfl/fl::β5tCre mice and stained for cytokeratin (CK) 8 (green, a cTEC marker), CK14 (red, an mTEC marker), AIRE (blue) and DAPI (grey; only in lower panels). Scale bars represent 50 µm. e Relative frequency and absolute cellularity of TEC isolated from Eedfl/fl::β5tCre (grey diamonds) and control Eedfl/fl mice (black circles) at indicated ages. Contour plots show representative flow cytometric analyses of thymic single-cell suspensions of 4-week-old mice, the gating for TEC and their relative frequencies. f Detection of H3K27me3 marks and histone H3 protein in mTEC, cTEC and CD45+ cells isolated from Eedfl/fl::β5tCre and Eedfl/fl mice at 4 weeks of age. Isotype control stains for Eedfl/fl mice are indicated as IgG control. Data in graphs (a, e) represent mean and standard deviations (SD) and are pooled from independent experiments: week 0: n = 3, week 1: n = 7, week 4: n = 6; *p < 0.05, **p < 0.01, ***p < 0.001 (two-tailed unpaired Student’s t test). Asterisks in red indicate comparisons between Eedfl/fl::β5tCre and Eedfl/fl mice, whereas black and grey asterisks compare results from identical mouse strains at different time points. Data in (b– d, f) are representative of two independent experiments with n = 3 mice each per strain investigated. n = biologically independent replicates per group. The gating for thymic epithelia is displayed in Supplementary Fig. 1. Source data including exact statistical test values are provided in the Source Data file. NATURE COMMUNICATIONS | (2021)12:3933 | https://doi.org/10.1038/s41467-021-24158-w | www.nature.com/naturecommunications 3
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24158-w a b cTEC Cellularity (x104) *** ** ** ** * ** 100 60 * mTEC 80 cTEC (%) * 15.3 ±1.0 100 *** 40 Subset of cells (%) TEC 60 *** 75 Eedfl/fl 10.7 40 * 33.6 ±2.9 *** 20 ±2.0 *** Eedfl/fl * 20 50.8 ±2.7 50 0 0 0 1 4 0 1 4 Eedfl/fl MHC II-APC/Cy7 Time (weeks) 105 8.0 ±0.8 86.2 Time (weeks) 25 Eedfl/fl::β 5tCre ±3.2 Eedfl/fl:: mTEC cellularity (x104) 104 *** Eedfl/fl::β 5tCre *** ** ** * 105 100 ** *** 50 Ly51 PE-Cy7 4 74.7 ±4.5 *** β5tCre 7.7 0 10 80 103 ±1.7 Eedfl/fl Eedfl/fl:: 40 mTEC (%) 103 *** 60 0 83.6 ±2.4 β5tCre 102 *** 30 0 mTEChiAire+ 40 0 10 3 10 4 10 5 13.8 ±4.3 20 *** mTEChiAire- ** Aire eF660 mTEClo 0 102 103 104 105 20 * 10 UEA1 Cy5 0 0 0 1 4 0 1 4 Time (weeks) Time (weeks) c 3.5 0.15 2.5 15 100 40 80 Thymus cellularity x 10 8 3 % CD40 hi in mTEC hi % mTEC hi in mTEC 2 80 % mTEC in TEC % cTEC in TEC 2.5 30 60 10 TEC x 10 5 0.1 % TEC 2 1.5 60 20 40 1.5 1 40 0.05 5 1 20 10 0.5 20 0.5 0 0 0 0 0 0 0 e e e e al e al e al al female Eed fl/wt al al female m m m fl/w t m m Eed ::β5tCre m fe fe male Eed fl/fl fe male d in TEC in mTEC 0.8 *** 100 *** *** 4 ** 0.7 80 Cellularity x105 0.6 3 % TEC 0.5 0.4 % 60 *** *** 2 *** 0.3 40 *** 0.2 1 0.1 20 0 0 0 cTEC mTEC mTEClo mTEChi TEC cTEC mTEC mTEClo mTEChi cTEC mTEClo mTEChi 25 Eed fl/fl ** Eed fl/fl :: β 5tCre 20 % BrdU+ 7.9 8.5 20.3 15 *** 10 ** # Cells 1.5 2.5 11.8 5 0 103 104 105 0 cTEC mTEClo mTEChi BrdU APC mature cells with a CD69−CD24− phenotype (CD8SP: 56.5 ± circulation from the periphery back to the thymus (Fig. 3e). 2.5% vs. 41.8 ± 2.6%; CD4SP: 45.4 ± 4.3% vs. 28.5 ± 0.5%, Fig. 3d, Within that CD24− subset, the frequency of re-circulated e). Both lineages displayed a normal progression from DPdull to peripheral T cells with a CD44+ phenotype (excluded from the CD69+CD24+ immature SP stages, suggesting that the relative above analysis) was significantly increased in Eedfl/fl::β5tCre mice, increase of mature CD24− SP was caused by retention in the especially among CD8SP cells (CD24−CD8:+ 16.4 ± 5.3% vs. 3.5 thymus and not by an altered developmental transition. Although ± 0.8%; CD24−CD4+: 2.7 ± 0.5% vs. 1.2 ± 0.2%, Fig. 3f and the overall frequency of Foxp3+ Treg cells within CD4 SP Supplementary Fig. 3b). Hence, the absence of PRC2 function in thymocytes was only marginally, yet significantly elevated in TEC caused an accumulation of specific SP thymocytes. mutant mice, their phenotype was severely skewed towards a Next, we characterized the negative selection of CD4 lineage CD24−CCR7lo expression pattern (67.4 ± 5.0% vs. 28.9 ± 3.2%) thymocytes. The first wave of negative selection within the cortex indicative of extended residence of cells in the medulla and/or re- (as identified by the co-expression of Helios and PD1 on Foxp3− 4 NATURE COMMUNICATIONS | (2021)12:3933 | https://doi.org/10.1038/s41467-021-24158-w | www.nature.com/naturecommunications
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24158-w ARTICLE Fig. 2 TEC phenotype of Eedfl/fl::β5tCre mice. a Quantification of cortical (CD45−EpCAM+Ly51+UEA1−) and medullary (CD45−EpCAM+Ly51−UEA1+) TEC subpopulations isolated from mice with the indicated genotype. Representative contour plots are from 4-week-old mice, frequencies and absolute cell numbers of cTEC and mTEC are from mice with the indicated genotype (Eedfl/fl::β5tCre: grey diamonds, Eedfl/fl mice: black circles) at ages shown (week 0: n = 3, week 1: n = 7, week 4: n = 6). b Flow cytometric analysis of mTEC from 4-week-old mice for MHCII and AIRE expression. Representative contour plots and bar graphs displaying the frequencies of individual subpopulations (mTEC:lo white bars black circles, mTEChiAIRE:− light grey bars black diamonds, mTEChi AIRE:+ grey bars black triangles) for each of the indicated genotypes (n = 5). c Quantification of total and TEC cellularity and determination of TEC subset distributions in Eedfl/wt::β5tCre and control mice. Data are pooled from two independent experiments with female (n = 4 Eed: fl/wt:β5tCre: light grey bars with grey diamonds, n = 4 Eed:fl/wt black bars with grey circles) and male mice (n = 6 Eedfl/wt::β5tCre: light grey bars with white diamonds, n = 2 Eed:fl/fl black bars with white circles) at 4–6 weeks of age. Cellularities are displayed for each sex separately, TEC subset frequencies are pooled. d TEC subset frequencies, cellularities and proliferation in 4-week-old Eedfl/fl::β5tCre (n = 3, white bars with grey diamonds) and Eedfl/fl mice (n = 3, grey bars with black circles). Representative histograms showing BrdU incorporation and frequencies of BrdU-positive cTEC and mTEC subsets. Data in graphs represent mean and standard deviations (SD) and are pooled from independent experiments (a, c) or display one experiment representative of two independent experiments (b, d). Contour plots (a, b) and histograms (d) are representative of data in bar graphs. The gating for thymic epithelia is displayed in Supplementary Fig. 1. n = biologically independent replicates per group. *p < 0.05, **p < 0.01, ***p < 0.001 (two-tailed unpaired Student’s t test). Asterisks in red indicate comparisons between Eedfl/fl::β5tCre and Eedfl/fl mice, whereas black and grey asterisks compare results from identical mouse strains at different time points. Source data including exact statistical test values are provided in the Source Data file. and CCR7− TCR+ DP or CD4SP thymocytes) was undisturbed of TCRα clonotypes bearing these amino acid motifs varied by the absence of EED in TEC (Fig. 4a). The second wave, which between T cell lineages33–35 but did not differ significantly takes place in the medulla and is characterized by Helios expression between mutant and control samples (Fig. 5b, c). Except for a on Foxp3− CD4SP thymocytes27, was reduced (CD24+: 4.1 ± 0.4% distal shift in Traj segment usage by DP thymocytes in mutant vs. 6.6 ± 0.6%; CD24−: 2.5 ± 0.3% vs. 3.4 ± 0.4%, Fig. 4b). This mice, Trav and Traj gene segment usage and CDR3 length dis- reduction correlated with both lower CD5 expression on SP tributions were comparable to controls (Supplementary Fig. 5). thymocytes (Fig. 4c) and significantly fewer apoptotic cells present This shift in Traj may reflect a longer sojourn in the cortex due to in the Helios-positive subsets (Fig. 4d) suggestive of having received the presence of a larger cTEC scaffold relative to the thymocyte TCR-mediated signals of too low avidity to commit to programmed pool in EED-mutant mice. Whereas the diversity of the pre- cell death28. In contrast, we observed no differences in thymocyte selection and wave 1 TCRα repertoires was normal, the CD4SP selection and maturation in Eed:fl/wt:β5tCre mice indicating that the and CD8SP TCRα repertoires selected by EED-deficient TEC presence of one Eed allele in TEC is sufficient to instruct regular T showed a 2-fold lower diversity than controls (Fig. 5d, e). These cell development (Fig. 4e). results indicate that EED deficiency did significantly narrow the Under physiological conditions, apoptotic cells are rapidly TCRα repertoires of CD4SP and CD8SP thymocytes. engulfed and cleared by thymic macrophages (Mø)29 and conventional Type 1 DCs (cDC1)30. Thymus immigrating DCs and B cells can also provide peripheral antigens for central Increased frequency of activated Treg cells in Eedfl/fl::β5tCre tolerance induction9–11. We therefore determined the frequencies mice. The peripheral T cell compartment of Eedfl/fl::β5tCre mice of thymic Mø, B cell and DC subsets in control and Eedfl/fl:: demonstrated a T cell lymphopenia, equally affecting CD4 and β5tCre mice and found that their relative frequency was CD8 T cells, (Fig. 6a, b). This resulted in a reduced frequency of significantly increased in mutant animals (Supplementary Fig. 4). naive and an expansion of central and effector memory T cells Increased Mø and cDC1 frequencies also indicated effective (Fig. 6c). Naive T cells from Eedfl/fl::β5tCre mice neither showed a removal of apoptotic cells11,30,31 thus providing a cellular deficiency in proliferation nor IL-2 production in response to mechanism for the reduced detection of cells having received CD3 mAb and co-stimulatory signals (Fig. 6d). negative selection signals. In addition, the relative increase in B We next determined the frequency and activation state of Treg cells and thymus immigrating DCs could compensate for defects cells. In comparison to controls, mutant mice showed a in selection mechanisms by PRC-deficient mTECs. Thus, the significant increase in the frequency of Treg among CD4 cells frequency of detectable negatively selected thymocytes was (Fig. 6b) with a larger proportion of these cells displaying an significantly reduced in the medulla, but not the cortex, of activated phenotype, as revealed by a higher frequency of cells Eedfl/fl::β5tCre mice. negative for CD62L (Fig. 6c) and positive for CD103 and/or ICOS (Fig. 6e). In comparison to control mice, peripheral Treg cells from Eedfl/fl::β5tCre mice exerted a higher potency suppressing The TCR repertoire in single-positive thymocytes of Eedfl/fl:: in vitro the proliferation of mitogenically activated CD4 effector β5tCre mice is narrowed. We extended the analyses of thymo- cells (Fig. 6f). Taken together, these findings indicate that the cyte selection by characterizing the TCR repertoire selected in the increased frequency and higher suppressive capacity of Treg cells presence of EED-deficient TEC. Eedfl/fl::β5tCre and β5tCre in Eedfl/fl::β5tCre mice likely control a dysregulated immune (CD45.2) mice were transplanted with T cell-depleted bone response as a result of either by lymphopenia or self-reactivity. marrow cells from mice transgenic for the YAe62 TCR β chain (TCRVb8.2)32. Donor-derived (CD45.1) thymocytes at distinct developmental stages (see Supplementary Fig. 5a) were isolated Eedfl/fl::ZsGreen::β5tCre mice contain mTEC with PRC2 4 weeks after transplantation and their TCRα chain repertoire activity. TEC isolated from Eed:fl/fl:β5tCre mice contained, as was sequenced. Each T cell lineage showed a high degree of TCRα noted above (Fig.1f), a subpopulation that had escaped gene repertoire overlap between mutant and control samples, and the deletion and thus maintained its H3K27me3 marks. These “EED T cell lineages were equally distinct from each other regardless of escapees” were predominantly mTEC, a finding that was also the PRC2-deficiency (Fig. 5a and Supplementary Fig. 5). Certain observed in Ezh1ko::Ezh2fl/fl::β5tCre mice (Supplementary Fig. 2). amino acid doublets and cysteine promote T cell self-reactivity This result was unexpected as Cre expression under the Psmb11 when present at the apex of the complementarity-determining transcriptional control targets >99.5 % of all mTECs26. To region 3 (CDR3) of the TCRα or TCRβ chain33–35. The frequency exclude a lack of Cre activity as an explanation for the detection NATURE COMMUNICATIONS | (2021)12:3933 | https://doi.org/10.1038/s41467-021-24158-w | www.nature.com/naturecommunications 5
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24158-w a b c Eed fl/fl fl/fl Eed fl/fl Eed fl/fl ::β5tCre Eed fl/fl Eed fl/fl ::β5tCre Eed ::β5tCre live cells DN T cell precursors CD8 CD4 - + CD4 PE-eFluor610 105 105 105 79.8 84.3 9.6 9.5 1.35 1.55 CD24 FITC ckit APC 104 104 104 2.7 103 22.1 19 8.1 103 103 0 102 2.6 0 3.3 98.7 98.4 0 0 103 104 105 0 103 104 105 0 103 104 105 CD8 A700 CD25 BV605 TCRβ PE DN ckit + TCR - CD19 APC-Cy7 CD71 PE-Cy7 105 Sca-1 BV510 105 105 CD8_ISP 28.2 23.3 104 4.5 57.3 104 104 103 93 41.4 103 103 98.2 95.3 102 71.8 ETP 76.7 102 0 0 DN2 0 0 103 104 105 0 103 104 105 0 50 100 150 200 250 TCRβ PE CD25 BV605 FSC-A (K) ckit - DP B cells CD71 PE-Cy7 CD71 PE-Cy7 105 105 DN in DN preDP 14.4 13.4 CD71+ 104 39.4 26.9 104 10 *** 70 *** DN3 103 103 60 8 102 102 50 0 7.7 52.6 9.2 62.1 0 % 6 % 40 0 103 104 105 0 103 104 105 30 CD25 BV605 TCRβ PE 4 20 ETP in maturation progression 2 T cell lineage in ckit + in ckit - 10 after β selection 0.01 100 70 ** 1.4 ** 0 0 100 60 1.2 ratio CD25 gMFI 0.008 80 80 50 ** 1 Eed fl/fl Eed fl/fl:: % 0.006 % 60 % 40 0.8 60 ** β5tCre 30 0.6 % 0.004 40 40 20 0.4 0.002 20 10 0.2 20 0 0 0 0 ETP DN2 DN3 pre CD71- DN3 0 DP CD25- p in P n DP 71 + in P& 1 + P DN2 CD reD i 3& eDP IS D7 &IS eD SP P CD D pr 8_I P DN pr C d Eed fl/fl Eed fl/fl ::β5tCre e Eed fl/fl Eed fl/fl ::β5tCre f re-circulated Foxp3 + in CD4 in CD24- SP DP CD4 CD8 + - 10 25 *** CD69 PE-Cy7 5 5 10 10 4.43 4.9 4.9 8 Foxp3 PE 3.63 10 4 10 4 20 3 * 10 6 % 15 3 10 % 0 0 4 10 3 4 5 94.4 94.1 0 10 10 10 0 103 104 105 TCR Bio/SA-BV650 2 5 *** CD25 BV605 CD69 +TCR + 0 0 CCR7 BV421 5 10 CD4 CD8 4 CD8 +CD4 lo Foxp3 + CD4 10 13.7 11 CD24 - in Foxp3 + 10 3 105 CD24 FITC dull 70.1 29.8 80 *** 2 DP 104 10 0 86.2 88.7 103 3 4 0 10 10 10 5 60 TCR Bio/SA-BV650 102 0 28.8 69 % concatenated 0102 103 104 105 40 CD8 +CD4 lo & CD8 +CD4 - CCR7 BV421 5 20 CD24 FITC 10 CD24 + 10 4 55.3 Foxp3 - CD4 39.6 10 3 10 5 0 CD24 FITC 70 10 2 CD24- 10 4 48.7 0 44.4 60.2 3 10 0 103 104 105 CD69 PE-Cy7 10 2 29 50.6 maturation progression 0 CD24- Foxp3 - CD4 in Foxp3 - CD4 maturation 0102 103 104 105 *** DP dull 70 70 progression CD69 PE-Cy7 in DP 60 *** 60 in CD8 *** CD24 + 50 50 10 60 5 *** *** % 40 CCR7 BV421 10 % 40 50 4 CCR7 + 30 8 10 47.9 52 30 40 10 3 - 20 20 % 6 % 30 2 CCR7 10 10 0 51.2 10 47.2 4 0 0 20 0 103 104 105 R7 CD + 4- CR in R7 CR 7 - D2 in CC C CR in + - - 7 &- C 7 + &C 4 - 24 R7 &ll CR7 - R7 2 10 TCR Bio/SA-BV650 + 2 CD CC CC du C DP C 0 0 24 CD + 4- CC D2 in D2 in 4 &ll C24 + &C 4 - + 2 du D DP C CD of mTEC with regular H3K27me3 marks, we next generated Eedfl/ loss of ZsGreen expression coincided with the detection of fl::ZsGreen::β5tCre mice where ZsGreen expression depends on H3K27me3 marks in 93.2 ± 2% of all TEC, whereas the vast the Cre-mediated removal of a floxed stop-cassette 5ʹ of the majority of ZsGreen+ TEC lacked H3K27me3 (97.4 ± 0.6%; transgenic reporter. In these mice, 98.6 ± 0.3% of all cTEC, yet Fig. 7b) although their total histone H3 abundance was com- only 80.3 ± 5.6% of all mTEC expressed ZsGreen (Fig. 7a). The parable (Fig. 7c). Furthermore, H3K27me2 median fluorescence 6 NATURE COMMUNICATIONS | (2021)12:3933 | https://doi.org/10.1038/s41467-021-24158-w | www.nature.com/naturecommunications
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24158-w ARTICLE Fig. 3 EED-deficient TEC affect thymocyte development at early and late stages. a Distribution of B cells and T lineage precursors within DN cells. b Quantification of ETP, DN2, DN3 and preDP subsets within DN T cell precursors. c Analysis of maturational progression from CD8 ISP to CD71pos DP thymocyte stages. Characterization of the developmental progression within CD8 (d) and CD4 lineage (e) selected thymocytes. f Quantification of re- circulated peripheral T cells within mature CD24neg SP subsets. Representative flow cytometry contour plots are shown, bar graphs display mean values with SD together with individual data points from n = 4 Eedfl/fl and n = 3 Eedfl/fl::β5tCre mice (a–c) or n = 6 Eedfl/fl and n = 9 Eedfl/fl::β5tCre mice (d–f) at 3 weeks of age. *p < .05, **p < .01, ***p < .001 (two-tailed unpaired Student’s t test). Data in bar graphs are from one experiment representative of three independent experiments. Eedfl/fl::β5tCre: white bars with grey diamonds, Eedfl/fl grey bars with black circles. n = biologically independent replicates per group. *p < 0.05, **p < 0.01, ***p < 0.001 (two-tailed unpaired Student’s t test). The gating for thymocyte subsets is displayed in Supplementary Fig. 3. Source data including exact statistical test values are provided in the Source Data file. was drastically reduced in ZsGreen+ but not ZsGreen− mTEC For example, genes encoding proteins involved in the immuno- subsets (mTEClo: ZsG+ 8.5 ± 3.0%, ZsG− 30.1 ± 23.5%, control proteasome were expressed at lower copy numbers in the absence 38.6 ± 16.0%; mTEChi: ZsG+ 13.7 ± 5.2%, ZsG− 115.6 ± 28.6%, of EED (ZsGreen-positive vs. Eedfl/fl::ZsGreen: 1.58-fold, p < control 100 ± 19.6%; Fig. 7d, left panels). In addition, the lack of 0.002; ZsGreen-positive vs. ZsGreen-negative: 2.66-fold, p < EED correlated with a 2-fold reduction in EZH2 fluorescence 0.002; Fig. 8b, c). Endopeptidase activity was particularly marked (cTEC: ZsG+ 6.8 ± 1.5%, control 14.2 ± 5.2%; mTEClo: ZsG+ in the ZsGreen-negative mTECs, with the highest expression of 10.2 ± 2.9%, control 18.6 ± 2.0%; mTEChi: 46.1 ± 4.2%, control 12 components of the proteasome seen in these cells. Conversely, 100 ± 4.5%; Fig. 7d, right panels). However, a minor fraction of ZsGreen-positive mTECs had elevated expression of genes within ZsGreen+ and H3K27me3+ TEC were detected (1.9 ± 0.8%; pathways regulating transcriptional activity related to cellular Fig. 7b) that were predominantly mTEC but displayed lower stress, such as Fosb and Jun (Fig. 8c). Moreover, genes that were amounts of ZsGreen (Fig. 7b). This phenotype is compatible with differentially expressed between ZsGreen-positive and either promiscuous β5tCre expression in “EED mTEC escapees”. Hence, ZsGreen-negative or Eedfl/fl mTECs were also enriched within most H3K27me3+ “EED escapees” must be the progeny of rare molecular pathways associated with antigen processing and pre- TEC progenitors that do not express Cre. sentation (Supplementary Fig. 7). These results suggested a defi- Eedfl/fl::ZsGreen::β5tCre mice provided an opportunity to ciency in normal antigen processing in the absence of EED compare in the same thymus ZsGreen-positive EED-deficient expression along with upregulation of pathways indicative of and ZsGreen-negative EED-proficient TEC (aka “EED escapees”). cellular stress. ZsGreen-negative cTEC had a reduced Ly51 and MHC class II expression, thus providing a phenotype distinct from that of Differential gene expression in EED-deficient and EED- ZsGreen-positive cTEC (Fig. 7f). mTEC and the contingent of proficient mTEC of Eedfl/fl::ZsGreen::β5tCre mice. mTEC sub- mature mTEChi demonstrated markedly reduced frequencies in populations provide an almost complete molecular representation the absence PRC2 activity in contrast to corresponding cells of TRAs with some of them being expressed at low frequency and among “EED escapees” (Fig. 7e). The frequency of CD40+ EED- in low copy number8. This capacity is aided by the transcriptional deficient mTEChi remained unchanged but that of AIRE+ cells facilitator AIRE, which promiscuously regulates the expression of was reduced when compared to controls (Fig. 7g). The genes marked at their transcription start site (TSS) by maturational progression of EED-deficient mTEC thus progres- H3K27me38. We therefore determined at single-cell resolution sively declined, whereas that of “EED escapees” was comparable whether EED-deficient and -proficient mTECs from Eedfl/fl:: to control mTEC. Both ZsGreen-positive and -negative mTEC ZsGreen::β5tCre mice maintained their normal transcriptional were uniformly distributed throughout the medulla of Eedfl/fl:: programmes. Single-cell analysis of cTECs showed a clear ZsGreen::β5tCre mice, yet the latter cell type occasionally formed separation between EED-deficient and Eedfl/fl cells based on small cell clusters (Fig. 7h). Taken together, these results are changes in transcripts associated with antigen processing and compatible with the notion that EED-proficient mTEC had most presentation (including Psmb11 and Ctsl; Supplementary Fig. 8). likely arisen via a non-canonical differentiation path unrelated to Significant differences were observed in the clustering centroids ineffective Cre recombination. of each of the three separate mTEChi populations analysed (all p < 0.05), further supporting the notion that EED-proficient EED-competent mTEC in Eedfl/fl::β5tCre mice represent a mTEC in Eedfl/fl::ZsGreen::β5tCre mice represent the emergence novel, non-canonical mTEC differentiation pathway. In wild- of a non-canonical mTEC lineage (Supplementary Fig. 9). type mice, mTECs are derived from progenitors that express Integration with existing single-cell mTEC transcriptomic Psmb1126. To test the hypothesis that “EED escapees” constitute a datasets enabled us to further delineate mTEC differentiation in separate lineage, we next compared the transcriptome of EED-deficient and -proficient mTEChi (Fig. 9a, b)36,37. EED- ZsGreen-positive EED-deficient and ZsGreen-negative EED- deficient mTEChi from Eedfl/fl::ZsGreen::β5tCre mice were proficient mTEChi. A principal component analysis of the gene diverted away from mature mTEC into an immature, intertypical expression profiles identified three distinct clusters with “EED TEC phenotype (Fig. 9c, left and middle panel). Conversely, EED- escapees” mTEChi grouping separately from EED-deficient and proficient mTEChi from Eedfl/fl::ZsGreen::β5tCre mice were able Eedfl/fl::ZsGreen mTEChi, suggesting that ZsGreen-negative EED- to differentiate into mature mTEC similarly to Eedfl/fl mice but proficient cells either represent a distinct, non-canonical TEC with a significant expansion of intertypical TEC (Fig. 9c, right lineage or a distinct differentiation pathway ending in a similar, panel). We used gene-regulatory network inference to identify but nonetheless transcriptomically distinct, medullary compart- transcription factor networks driving differences in mTEC ment (Fig. 8a). Both ZsGreen-positive and -negative mTECs differentiation and found Ascl1 and Irf7 network expression showed marked transcriptomic differences to Eedfl/fl::ZsGreen was significantly increased in EED-deficient mTEChi (Fig. 9d). mTECs (518 and 305 genes, respectively, significant at FDR < Jun and Fos network expression was specifically elevated in EED- 0.05). There were also transcriptomic differences between proficient mTEChi from Eedfl/fl::ZsGreen::β5tCre mice relative to ZsGreen-positive and -negative mTECs (664 genes, significant at Eedfl/fl mice (Fig. 9e). This confirmed that EED-deficiency FDR < 0.05; Supplementary Fig. 6 and Supplementary Data 1–3). diverted mTEC differentiation towards a more immature NATURE COMMUNICATIONS | (2021)12:3933 | https://doi.org/10.1038/s41467-021-24158-w | www.nature.com/naturecommunications 7
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24158-w a CCR7 - Foxp3 - TCR + b CCR7 + Foxp3 - CD4 SP excluded CD4SP CD24 + CD24 - 0.14 1 10 fl/fl fl/fl 0.09 0.54 0.12 3.93 % Helios + PD1 + % Helios + PD1 + Eed 0.8 7.06 *** Eed 8 0.1 % Helios + 0.08 0.6 6 ** 0.06 ::β5tCre 0.4 4 ::β5tCre 5 5 10 10 Helios APC Helios APC 4 0.12 0.51 0.04 4 10 0.2 10 4.52 2.3 2 0.02 103 103 fl/fl 0 0 0 fl/fl CCR7 - CCR7 - + 102 102 Eed 24 Foxp3 - 24 0 Eed 0 Foxp3 - D D TCR + CD4SP C 3 4 5 2 3 4 5 C 0 10 10 10 010 10 10 10 PD1 BV786 Eed fl/fl:: CCR7 BV421 fl/fl Eed fl/fl:: Eed fl/fl Eed β5tCre β5tCre - c 35 * 30 TCR gMFI (K) 25 20 15 10 ** 5 0 20 *** *** *** * * *** CD5 gMFI (K) 16 *** 12 *** 8 4 fl/fl Eed fl/fl:: Eed 0 β5tCre CD24- CD8 Foxp3+ CD4 CD24+ CD8 Helios+ PD1+ DPdull TCRpos DP pre-selection CCR7- CD4 CCR7+ CD4 CD24+ CD24- Helios+ PD1+ Helios- PD1- Helios+ Helios+ Helios- Helios- d CCR7 - Foxp3 - TCR + CCR7 + Foxp3 - CD4 Foxp3 + CD4 SP excluded CD4SP CD24 + CD24 - 100 30 * * *** *** Heliosneg Heliospos % Lactadherinpos 25 pos fl/fl 80 % Lactadherin Eed 20 60 70.7 79.8 4.2 20.1 18.1 15 40 10 Eed fl/fl:: β5tCre 20 5 33.4 29.8 4.3 7.71 5.7 0 0 0 103 104 105 CCR7 - Foxp3 - TCR + CCR7 + Foxp3 - CD4 Foxp3 + CD4 Lactadherin FITC CD24 + CD24 - P ed 4S cl SP Eed fl/fl:: ud D fl/fl Eed β5tCre C ex e 100 female Eed fl/wt female fl/wt male Eed fl/fl male Eed ::β5tCre 80 60 0.5 10 % in thymus 10 2.5 60 60 10 % Foxp3 + in CD4SP % Helios +PD1 + 8 2 50 50 8 0.4 8 % in preDP % Helios + 40 40 0.3 6 % in SP 6 1.5 6 30 30 4 1 4 0.2 4 20 20 2 0.5 2 0.1 2 10 10 0 0 0 0 0 0 0 CCR7 - Foxp3 - CCR7 + Foxp3 - DN1 DN2 DN3 DN4 CD8 ISP preDP DP CD4SP CD8SP Foxp3 - CD4SP CD8SP TCR + TCR + CD24 + CCR7 - CCR7 + CD24 - CD24 + CD24 - excluded CD4SP CD24 - SP medullary phenotype, whereas the majority of “EED escapees”, transcripts depend on AIRE expression, are enhanced by AIRE despite not originating from the canonical β5t+ TEC progenitor or are generated independently of AIRE expression8. There were pool, were able to adopt a mature mTEC identity, although with small changes in the overall expression of AIRE-controlled genes several differences in transcription factor network expression. between EED-deficient and -proficient mTEC (Supplementary We next assessed the consequences of the loss of EED on Fig. 10a). For example, AIRE-dependent TSPAN8 expression promiscuous expression of genes encoding TRAs whose remained unchanged since this TRA was detected at comparable 8 NATURE COMMUNICATIONS | (2021)12:3933 | https://doi.org/10.1038/s41467-021-24158-w | www.nature.com/naturecommunications
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24158-w ARTICLE Fig. 4 Negative selection and TCR avidity are reduced in a EED-deficient thymic microenvironment. Analysis of negative selection of CD4 lineage thymocytes in the cortex (a) and the medulla (b). CD5 and TCR geometric mean fluorescence intensity on indicated thymocyte subsets (c). Quantification of apoptotic cells in Helios-positive thymocyte subsets (d). Analysis of thymocyte development in Eed:fl/wt:β5tCre and control mice. Frequencies of preDP, DP and SP subsets, developmental stages within preDP thymocytes, maturation within SP cells, frequencies of Treg in CD4SP and frequencies of negatively selected cells (e). Representative flow cytometry contour and histogram plots are shown. Bar graphs display mean values with SD together with individual data points from n = 6 Eedfl/fl and n = 9 Eedfl/fl::β5tCre mice at 3 weeks of age (a–c) or n = 3 Eedfl/fl and n = 3 Eedfl/fl::β5tCre mice at 4 weeks of age (d) representative of at least two independent experiments. Eedfl/fl::β5tCre: white bars with grey diamonds, Eedfl/fl grey bars with black circles. Data in (e) are pooled from two independent experiments with female (n = 4 Eedfl/wt::β5tCre: light grey bars with grey diamonds, n = 4 Eedfl/wt black bars with grey circles) and male mice (n = 6 Eedfl/wt::β5tCre: light grey bars with white diamonds, n = 2 Eedfl/fl black bars with white circles) at 4–6 weeks of age. n = biologically independent replicates per group. *p < 0.05, **p < 0.01, ***p < 0.001 (two-tailed unpaired Student’s t test). The gating for thymocyte subsets is displayed in Supplementary Fig. 3. Source data including exact statistical test values are provided in the Source Data file. a b c Eed 0.016 cysteine within 2 positions 0.3 that promote self-reactivity Proportion of clones with *** β5tCre Proportion of clones with *** β5tCre CDR3 apical doublets Morisita Horn *** Pre-sel’n fl/fl *** of the CDR3 apex 0.31 0.012 Index 0.2 Eedfl/fl:: Eedfl/fl:: +/+ 0.16 0.18 ≤0.15 β5tCre β5tCre Wave 1 0.008 fl/fl 0.15 0.16 0.49 >0.15 0.1 0.004 +/+ 0.16 0.19 0.13 0.10 >0.30 CD4SP 0 0 fl/fl 0.11 0.12 0.09 0.08 0.58 >0.45 C P P W ’n C 1 C P P W ’n 1 4S 8S l 4S 8S e l e se se av av D D D D e- +/+ e- 0.09 0.10 0.11 0.10 0.13 0.10 C Pr Pr CD8SP fl/fl 0.07 0.08 0.07 0.07 0.10 0.10 0.38 +/+ fl/fl +/+ fl/fl +/+ fl/fl +/+ Eed d e Pre-sel’n Wave 1 CD4SP CD8SP 40 Pre-sel’n 7 β5tCre clonotypes (10–3) Wave 1 TCRα clonotype relative diversity Unique TCRα 30 Eedfl/fl:: CD4SP 5 β5tCre 20 CD8SP 3 Eedfl/fl:: β5tCre β5tCre 10 0 1 0 0.2 0.4 0.6 1 P l’n P 8S 4S e se av Coverage D D e- W C C Pr Fig. 5 TCRα repertoire analysis of YAe62 TCRβ-transgenic thymocytes in β5tCre or Eedfl/fl::β5tCre bone marrow chimeras. Unless stated otherwise, the analyses compared TCR catalogues, which were formed by aggregating samples of a given T cell subset/genotype combination. The β5tCre and Eedfl/fl:: β5tCre TCR catalogues contained the following numbers of samples: 9 and 14 mice, respectively (pre-selection, Wave 1 and CD8SP); 8 and 14 mice, respectively (CD4SP). a Heatmap shows the Morisita-Horn index for each pair of TCR catalogues. b, c Amino acid composition of the CDR3 apex. Considering the amino acid doublets at CDR3 positions 6 and 7, (b) shows the proportion of TCRα clones with doublets that promote self-reactivity33. Each symbol represents an individual sample. Samples with fewer than 100 clones were excluded, (c) depicts the proportion of TCRα clones with cysteine anywhere from CDR3 position −2 to +2 as determined by apex-out CDR3 alignment35. ***p = 4.3 × 10−5 (pre-selection vs. Wave 1) and p = 1.7 × 10 −6 (pre-selection vs. CD4SP) using Student’s t test (unpaired, two-sided, Bonferroni-corrected) in (b) or p = 1.5 × 10−6 (pre-selection vs. Wave 1) and p = 2.8 × 10−15 using Fisher’s exact test (2 × 2 table, two-sided, Bonferroni-corrected) in (c). d, e TCRα clonotype diversities. d Cumulative number of unique TCRα clonotypes plotted as a function of “coverage”, defined as the proportion of all clones (encountered and unencountered) in a given T cell subset that belong to clonotypes encountered in the catalogue. Symbols show the observed values and curves outline the 95% confidence intervals of the rarefaction and extrapolation curves determined using 10 bootstrap replications. e Relative diversity of TCR catalogues. Symbols indicate ratios of the numbers of unique TCRα clonotypes per TCR catalogue at the level of coverage indicated by the vertical dashed line in (d). Error bars in (e) show the 95% confidence intervals of the relative diversities of TCR catalogues based on 10 bootstrap replications. level in ZsGreen-positive and -negative mTEChi of experimental subtle reduction in the breadth of promiscuous gene expression in mice and their controls (Fig. 10a). By randomly sampling AIRE mTEChi proficient or deficient in PRC2 activity within the same expressing single cells, we estimated for each of the different thymic environment. However, multivariate analysis showed that mTEC populations the number of detectable genes encoding the presence of H3K27me3 marks near the TSS predicted whether TRAs (Fig. 10b). The breadth of promiscuous TRA expression or not a gene was upregulated in EED-deficient mTECs relative to was reduced in EED-deficient mTECs, affecting particularly those that were EED-proficient independently of whether or not AIRE-controlled loci (AIRE-dependent: 0.86-fold, p < 0.0001; that gene was regulated by AIRE (Supplementary Fig. 10b). This AIRE-enhanced: 0.96-fold, p < 0.0001; AIRE-independent TRA: finding suggests that at least a portion of the mild deficiency in 0.98-fold, p < 0.0001; all other genes: 0.99-fold, p < 0.0001). A promiscuous gene expression was more likely to be driven by similar change in the pattern of TRA expression was observed in thymic microenvironmental changes than a direct effect of EED- ZsGreen-negative mTEC isolated from Eedfl/fl::ZsGreen::β5tCre deficiency. Eedfl/fl::β5tCre mice did not develop auto-antibodies or mice (AIRE-dependent: 0.84-fold, p < 0.0001; AIRE-enhanced: overt organ infiltration nor did they reveal signs of autoimmunity 0.94-fold, p < 0.0001; AIRE-independent TRA: 0.97-fold, p < after antibody-mediated T cell depletion and subsequent 0.0001; all other genes: 0.99-fold, p < 0.0001). Hence, there was a endogenous T cell reconstitution (Supplementary Fig. 11). NATURE COMMUNICATIONS | (2021)12:3933 | https://doi.org/10.1038/s41467-021-24158-w | www.nature.com/naturecommunications 9
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24158-w a b TCR+ CD4+ 52.9 59.5 100 10.9 fl/fl Cellularity (x106) 80 Eed 60 60 *** 3 20 *** % Foxp3+ in CD4+ 39.8 87.9 40 CD4/CD8 ratio 15 ::β5tCre % TCR+ *** 105 26.7 40 105 60 2 105 CD5 APC-Cy7 CD4 PE-Cy7 20 16.7 Foxp3 APC 104 104 104 10 0 B T 103 20 103 1 103 fl/fl 5 Eed 102 Spleen 0 0 0 38.9 81.6 Eed fl/fl 0 0 0 3 4 5 3 4 5 3 4 5 Eed fl/fl::β5tCre 0 10 10 10 0 10 10 10 0 10 10 10 TCR BV650 CD8 A700 CD25 BV605 Eed fl/fl Eed fl/fl::β5tCre c CD4+ Foxp3+ 86.9 CD4+ Foxp3- 3.5 67.1 CD8+ 28.1 74.5 fl/fl 100 Eed *** % in T cell subsets 80 *** *** 23.3 7.8 1.3 60 *** CD62L PerCP-Cy5.5 105 42.4 72.1 5.8 33.7 59.1 ::β5tCre 40 104 20 *** 103 Eed fl/fl ** fl/fl Eed fl/fl::β5tCre 0 Eed 0 54.5 18.5 4 CD62L: - + - + + - + 2 3 4 0 10 10 10 10 5 CD44: + - + + - + + CD4+ CD4+Foxp3- CD8+ CD44 BV510 Foxp3+ d e CD4+ CD25+ f 4.18 5.95 (CD4 without TReg=100%) 100 Eed fl/fl fl/fl 100 fl/fl % of max. proliferation Eed fl/fl Eed fl/fl::β5tCre Eed % in CD4+ CD25+ *** Eed fl/fl::β5tCre Eed 80 80 ** 60 60 Eed fl/fl::β5tCre 73.3 16.6 *** * 40 40 ::β5tCre 105 8.8 20.8 *** ICOS PE-Cy7 20 20 % Max 104 ** 0 103 0 1:4 1:2 1:1 fl/fl 2 3 4 5 ICOS: - - + + Eed 0 10 10 10 10 102 TReg Dilution CFSE 0 36.2 34.2 CD103: - + + - 5 100 * 0 102 103 104 105 * CD103 FITC 80 Expansion Index 4 IL2 (ng/ml) 3 60 2 40 1 20 0 0 Eed fl/fl Eed fl/fl::β5tCre Fig. 6 Peripheral T cells and their function in Eedfl/fl::β5tCre and Eedfl/fl mice. a Absolute numbers of splenic B and T cells in Eedfl/fl (n = 3) and Eedfl/fl:: β5tCre mice (n = 3) at 4 weeks of age. b Representative contour plots and frequencies of lymph-node-derived T cell subsets. Frequency of TCRβ+ T cells, the CD4/CD8 ratio within TCRβ+ T cells and percentage of Foxp3+ TReg in CD4 T cells. c Frequencies of activated cells (CD44+CD62L−) in Foxp3+ TReg and frequencies of naive (CD44−CD62L+), central memory (CD44+CD62L+) and effector memory (CD44+CD62L−) T cells within the populations of Foxp3− CD4 and CD8 lymph node cells. b, c Eedfl/fl (n = 3) and Eed:fl/fl:β5tCre mice (n = 4) at 4 weeks of age. d Proliferative response and IL-2 secretion of naive lymph node CD4 T cells (CD44loCD25−) from Eedfl/fl (n = 3) and Eedfl/fl::β5tCre mice (n = 3) at 4 weeks of age. Representative histograms of CFSE dilution (including model components used for expansion index calculations; grey dotted lines) following stimulation for three days with CD3 mAb (1 µg/ ml) and irradiated Rag2−/− splenocytes. Bar graphs display the mean expansion index (left) and mean IL2 concentrations in the supernatant (right) with SD. e Representative contour plots and frequencies of lymph-node-derived TReg subsets defined by their differential expression of ICOS and CD103 from Eedfl/fl (n = 3) and Eedfl/fl::β5tCre mice (n = 4) at 4 weeks of age. f T cell suppression assay: CD3 mAb-induced proliferation of sorted naive WT CD4 T cells (CD44loCD25−) in the presence of sorted lymph node Treg (CD4+CD25+) from Eedfl/fl (n = 3) and from pools of two Eedfl/fl::β5tCre mice each (n = 3) of mixed sex at 7–9 weeks of age. *p < 0.05, **p < 0.01, ***p < 0.001 (unpaired Student’s t test). Data in graphs show mean ± SD and are representative of two independent experiments. Contour plots (b, c, e) and histograms (d) are representative of data in bar graphs. Eedfl/fl::β5tCre: white bars with grey diamonds, Eedfl/fl grey bars with black circles. n = biologically independent replicates per group. Source data including exact statistical test values are provided in the Source Data file. 10 NATURE COMMUNICATIONS | (2021)12:3933 | https://doi.org/10.1038/s41467-021-24158-w | www.nature.com/naturecommunications
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24158-w ARTICLE a CD45neg b CD45neg Eedfl/fl::ZsGreen::β 5tCre Eedfl/fl::ZsGreen::β 5tCre EpCAMposMHCIIpos EpCAMposMHCIIpos 5 100 ** *** 5 60.7 10 ZsGpos 1.7 87.6 Ly51 PE-Cy7 10 ZsGreen 4 80 10 4 % mTEC 10 3 94.8 10 100 60 3 10 % H3K27me3pos 0 0 4.8 80 40 97.8 33.1 ** 60 H3K27me3 A647 0 10 3 4 10 10 5 0 50 100 150 200 250 10 5 91.2 31.2 20 100 FSC-A (K) UEA1 BV786 4 4 0 10 80 ZsG: + + - % ZsGpos mTEC ZsGneg 10 3 2 H3K27me3: + - + 80 60 0 0 mTEC Zs neg os 40 Gp H3K27me3+ G 0 50 100 150 200 250 5 Zs 98.9 10 ZsG+ 91.7 Ly51 PE-Cy7 cTEC 20 FSC-A (K) 4 10 0 3 4 10 10 10 5 0 H3K27me3- EC EC 3 ZsG+ 10 ZsGreen cT mT 0 H3K27me3+ ZsG- 3 4 5 0 103 104 105 0 10 10 10 UEA1 BV786 ZsGreen c CD45neg d cTEC mTEC lo mTEC hi cTEC mTEC lo mTEC hi Eedfl/fl::ZsGreen::β 5tCre EpCAMposMHCIIpos ZsG+ ZsG- 3 3 4 5 3 3 4 5 -10 0 10 10 10 -10 0 10 10 10 3 4 5 0 10 10 10 H3K27me2 A647 EZH2 A647 Total H3 A647 *** *** *** Eed fl/fl::ZsGreen *** (% of β5tCre neg mTEC hi ) (% of β5tCre neg mTEC hi ) 160 Eed fl/fl 140 median fluorescence median fluorescence *** ZsGneg 140 ::ZsGreen 120 * *** ZsGpos 120 *** ::β5tCre H3K27me2 100 100 * EZH2 80 80 60 60 40 40 ** *** 20 20 0 0 cTEC mTEC lo mTEC hi cTEC mTEC lo mTEC hi e CD45neg f cTEC g EpCAMposMHCIIpos *** mTEChi 100 % CD40pos in mTEChi 100 *** % Ly51hi in cTEC ZsG pos 74.4 22.1 mTEC mTEChi 70.5 80.8 96.8 80 ** 80 Eedfl/fl::ZsGreen::β 5tCre 4.8 60 60 cTEC 51.6 81.4 40 3 40 85.8 20 20 76.1 10.9 86.8 0 10 3 10 4 10 5 0 103 104 105 0 0 ZsGneg 36 Ly51 Bio/SA BV786 CD40 eF450 MHCII median fluorescence Ly51pos cTEC 100 40K mTEChi % Airepos in mTEChi ** 80 30K 65.8 Ly51 Bio/SA BV786 60 MHCII APC-Cy7 Eedfl/fl::ZsGreen 5 5 9.2 5 10 10 89.4 10 20K ZsGreen 4 4 4 40 10 10 10 42.9 71.2 10 3 3 3 10K 20 10 10 10 2 100 0 0 71.6 0 0 0 10 3 10 4 10 5 0 0 103 104 105 0 103 104 105 0 103 104 105 0 103 104 105 Ly51hi cTEC MHCII APC-Cy7 Aire eF660 EpCAM PerCP-Cy5.5 UEA1 Cy5 CD80 BV605 fl/fl Eed fl/fl::ZsGreen ZsGneg Eed ::ZsGreen * ZsGpos ::β5tCre 100 60 % mTEChi in mTEC 80 50 h % in TEC 40 Eedfl/fl::ZsGreen::β5tCre 60 Eed fl/fl::ZsGreen 30 40 20 Eed fl/fl * ZsGneg 20 10 ::ZsGreen pos ZsG ::β5tCre 0 0 cTEC mTEC 15μm 50μm 50μm ZsGreen ZsGreen CK14 ZsGreen CK14 Given the near-complete loss of H3K27me3 marks in ZsGreen- expressed in ZsGreen-positive mTECs than in either of the two positive EED-deficient mTEC, we assessed the distribution of other TEC populations (H3K27me3 ChIP/input signal within 4 H3K27me3 ChIP-seq signal in wild-type mTEChi around those kb of the TSS: ZsGreen-positive vs. ZsGreen-negative: median genes that were differentially expressed between separate mTEC 2.36-fold, FDR < 0.0001; ZsGreen-positive vs. Eedfl/fl median populations. H3K27me3 ChIP-seq signals were significantly 2.14-fold, FDR < 0.0001; Fig. 10c). Differences in the correspond- increased around the TSS of genes that were most highly ing H3K27me3 ChIP-seq signals were, however, not observed NATURE COMMUNICATIONS | (2021)12:3933 | https://doi.org/10.1038/s41467-021-24158-w | www.nature.com/naturecommunications 11
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24158-w Fig. 7 Eedfl/fl::ZsGreen::β5tCre mice contain a subpopulation of Eed-proficient mTEC. a ZsGreen expression in cTEC and mTEC isolated from 4-week-old Eedfl/fl::ZsGreen::β5tCre mice. b Quantification of H3K27me3 abundance in ZsGreen-positive and -negative TEC and their phenotypic characterization. c Representative histograms of histone H3 abundance in ZsGreen-positive and -negative TEC. d Representative histograms and median fluorescence quantification of H3K27me2 and EZH2. Fluorescence of control mTEChi was set as 100%. Data are pooled from two independent experiments identified by circles and diamonds, respectively. e Representative FACS plots and quantitative analysis of cTEC, mTEC and mTEChi frequencies within ZsGreen-positive and -negative TEC. f, g Representative histograms and quantitative analysis of ZsGreen-positive and -negative cTEC (f) and mTEChi (g) phenotypes. h Spatial distribution of ZsGreen-positive and -negative TEC in thymus tissue sections of 4-week-old Eedfl/fl::ZsGreen::β5tCre mice. Immunofluorescence analysis of mTEC expressing ZsGreen and cytokeratin 14 (CK14) in the medulla (marked by dashed blue line) identifying single cells and small aggregates of ZsGreen-negative cells (white dashed lines). Data in bar graphs show mean ± SD and are pooled from two independent experiments (a, b, d) or are representative of two independent experiments (e–g). Contour plots and histograms are representative for data displayed as frequencies in bar graphs (a, b, d, e–g). Immunofluorescence data are representative of two independent experiments with 3 sections each. a, b Eedfl/fl::ZsGreen::β5tCre (n = 3), in (b) dark green represents ZsGreen-positive H3K27me3-positive, light green shows ZsGreen-positive H3K27me3-negative, and grey depicts ZsGreen-negative H3K27me3-positive, (d) Eedfl/fl::ZsGreen (n = 5, grey histograms and bars with black circles and diamonds) Eedfl/fl::ZsGreen::β5tCre (n = 7, ZsGreen- negative: white histograms and bars with grey circles and diamonds, ZsGreen-positive: black histograms and bars with grey circles and diamonds), (e–g) Eedfl/fl::ZsGreen (n = 5, grey histograms and bars with black circles), Eedfl/fl::ZsGreen::β5tCre (n = 5, ZsGreen-negative: white histograms and bars with grey diamonds, ZsGreen-positive: black histograms and bars with grey diamonds); n = biologically independent replicates per group, mice were 3–4 weeks of age. *p < 0.05, **p < 0.01, ***p < 0.001 (two-tailed unpaired Student’s t test). Source data including exact statistical test values are provided in the Source Data file. between ZsGreen-negative and Eedfl/fl mTEC (FDR > 0.05). and function cannot be interrogated in detail42. Thus, it remains Taken together, these data demonstrated that the loss of EED unknown whether mTEC development bypassing β5t-positive bi- early in TEC development resulted in an mTEC compartment potent TEC precursors can also occur in Eedfl/fl::Foxn1-Cre mice. with a relatively modestly reduced capacity for promiscuous gene We therefore conclude that the loss of EED expression early expression. This finding suggests that the expression of TRAs is during TEC differentiation (i.e. before the emergence of bi-potent not critically dependent on repressive chromatin modifications. β5t-positive TEC precursors) and the consequent loss of cano- nical gene silencing by PRC2 precludes not only the normal Discussion progression along the canonical TEC differentiation pathway but TEC development and function are dependent on EED, an also excludes the emergence of the alternative mTEC pathway essential component for assembly and activity of the PRC2 described here. complex. The loss of EED in β5t-expressing TEC precursors and In addition to intrathymic B cells, Eedfl/fl::β5tCre mice also their progeny resulted in an absence of PRC2 activity in practi- show increased frequencies of thymic macrophages and DCs, cally all cTEC (with the notable exception of few non-recombined although it is possible that this could simply be due to the cTEC displaying an atypical phenotype) and the majority of observed reduction of thymocytes. The impaired antigen pre- mTEC. However, up to 15% of all mTEC present in Eedfl/fl:: sentation by TEC may be compensated by immigrating DCs and β5tCre mice arise from an alternative mTEC developmental B cells thus contributing to establishing central T cell pathway originating from cells other than bi-potent, β5t-positive tolerance9–11. Antigen transfer from mTEC to DCs may fur- precursors and characterized by a distinct transcriptome, repre- thermore offset defects in mTEC-mediated antigen presentation43 senting either a distinct TEC lineage or an alternate differentia- and selecting a TCR repertoire with sufficient self-tolerance. tion pathway into canonical mTEC cell types within an abnormal Signs of autoimmunity are not detected in Eedfl/fl::β5tCre mice thymic microenvironment26,38. EED-proficient mTEC exhibits in despite an impaired expression of antigen processing pathways, a Eedfl/fl::β5tCre mice a competitive growth/differentiation advan- limited reduction in the representation of TRA in mTEC, a tage over EED-deficient mTEC. narrowed breadth of the TCR repertoire and peripheral T cell Despite the severely reduced cellularity of ETP and a partial lymophopenia. Although reliant on mTEC and DCs for their block in β-selection, cTECs of Eedfl/fl::β5tCre mice support the selection44–46, Treg generation is not offset by the reduced mTEC differentiation to DP thymocytes and their positive selection cellularity, indicating a likely compensation by the increased despite differences in the expression of many genes required for frequency of intrathymic DCs. The heightened frequency and antigen processing and presentation. Any deficiencies in antigen function of peripheral Treg—resulting from lymphopenia-induced presentation may likely be compensated by the decreased proliferation—contribute further to the observed lack of auto- thymocytes-to-cTEC ratio, thus lessening the competition between immunity. Because peripheral Treg cannot compensate for AIRE thymocytes and the normally limiting numbers of cTEC. This will deficiency, the absence of autoimmunity corroborated our finding result in an increased cross-talk between these two cell types and a that EED deficiency results only in a minor alteration of TRA longer sojourn of thymocytes in the cortex39. Indeed, the subtle representation47. differences in the choice of Traj segments within the TCRα Gene expression profiling of EED-deficient mTEChi demon- repertoire of Eedfl/fl::β5tCre mice support this explanation as strated a significant upregulation of genes that are normally asso- reflected by a more frequent use of distally positioned Traj family ciated with H3K27me marks in EED-proficient mTEChi, both members40. Furthermore, the transition of thymocytes across the wild-type and “EED escapees”, a finding consistent with the corticomedullary zone may be secondary to the decrease in Psmb11 removal of canonical gene silencing by PRC2. Changes in TEC expression, as observed in β5t-deficient mice41. development and composition are thus explained by gene de- The exact timing of the loss of EED expression in TEC repression in pathways involved in epithelial differentiation as determines thymus organogenesis. In a recently described mouse highlighted in other organs, such as lung, including Foxp1 and Nfib strain (Eedfl/fl::Foxn1-Cre), an identical ablation of the Eed locus (Supplementary Fig. 7). This mechanism of de-repression was targeted to a time point early during TEC fate specification but previously suspected as known PRC2 targets are characterized by prior to Psmb11 expression hinders TEC maturation resulting in active enhancer marks in immature but not mature mTEC48. The a severely dysplastic thymus in which PRC2’s role in development current study demonstrates that these chromatin changes are not 12 NATURE COMMUNICATIONS | (2021)12:3933 | https://doi.org/10.1038/s41467-021-24158-w | www.nature.com/naturecommunications
You can also read