A feedback amplifier circuit with Notch and E2A orchestrates T-cell fate and suppresses the innate lymphoid cell lineages during thymic ontogeny

In this study, Miyazaki et al. show that the E-protein–NOTCH1–ID2 axis serves as an amplifying feedback circuit that dictates T-cell lineage specification in the fetal and adult thymus. E2A and NOTCH1 synergistically regulate the chromatin accessibility at enhancers and transcriptional programs that specify T-cell identity while suppressing innate lymphoid cell fates, representing a developmental switch that underpins innate and adaptive immunity.

Abstract

External signals from the thymic microenvironment and the activities of lineage-specific transcription factors (TFs) instruct T-cell versus innate lymphoid cell (ILC) fates. However, mechanistic insights into how factors such as Notch1–Delta-like-4 (Dll4) signaling and E-protein TFs collaborate to establish T-cell identity remain rudimentary. Using multiple in vivo approaches and single-cell multiome analysis, we identified a feedback amplifier circuit that specifies fetal and adult T-cell fates. In early T progenitors (ETPs) in the fetal thymus, Notch signaling minimally lowered E-protein antagonist Id2 levels, and high Id2 abundance favored the differentiation of ETPs into ILCs. Conversely, in the adult thymus, Notch signaling markedly decreased Id2 abundance in ETPs, substantially elevating E-protein DNA binding and in turn promoting the activation of a T-cell lineage-specific gene expression program linked with V(D)J gene recombination and T-cell receptor signaling. Our findings indicate that, in the fetal versus the adult thymus, a simple feedback amplifier circuit dictated by Notch-mediated signals and Id2 abundance enforces T-cell identity and suppresses ILC development.

During cell differentiation, the interplay between a lineage-specific internal transcription factor (TF) and external stimuli derived from the tissue microenvironment controls cell lineage direction at developmental branchpoints, regulates signature gene expression, and enforces cell lineage commitment by suppressing the expression of other lineage-specific TFs (Heinz et al. 2015; Troutman et al. 2021). For T-cell lineage commitment, Notch1 signal constitutes an essential external signal, and E-protein represents an essential internal TF (Radtke et al. 2004; Shah and Zúñiga-Pflücker 2014; Murre 2019; Hosokawa and Rothenberg 2021). A deficiency in Notch1 or E-protein expression in lymphoid progenitors or the expression of Notch ligand Delta-like-4 (DLL4) in thymic epithelial cells (TECs) leads to the loss of T-cell potential and induction of other lymphoid lineages such as B cells, dendritic cells (DCs), and innate lymphoid cells (ILCs) in the thymus (Pui et al. 1999; Radtke et al. 1999; Hozumi et al. 2008; Feyerabend et al. 2009; Miyazaki et al. 2017; Wang et al. 2017; Koga et al. 2018; Hozumi 2020). Because E-protein directly regulates Rag gene enhancer activity to promote V(D)J recombination of the T-cell receptor (TCR) and immunoglobulin (Ig) genes, specification of adaptive lymphoid cells is controlled by this TF (Miyazaki et al. 2020). However, the interplay between Notch1 signaling and E-protein function in T-cell lineage commitment, especially how these factors establish the T-cell-specific enhancer landscape and regulate T-cell signature genes, has not yet been determined.

Innate and adaptive lymphoid cell development is orchestrated by the transcriptional balance between E-proteins and Id (inhibitor of DNA binding) proteins (E–Id axis) (Kee 2009; Lauritsen et al. 2009; Braunstein and Anderson 2011; Belle and Zhuang 2014). Lymphoid cells express four E-proteins: E2-2 (Tcf4), HEB (Tcf12), E12, and E47. E12 and E47 are encoded by the E2A (Tcf3) gene and are generated via differential splicing. E2A/E47 plays a key role in B-cell lineage commitment by regulating the enhancer landscape that orchestrates B-cell fate (Lin et al. 2010, 2012). The E-protein DNA binding activity is antagonized by Id proteins (Id1–4). In particular, Id2 is important for determining all ILC subsets and their functions (Verykokakis et al. 2014; Cherrier et al. 2018). Therefore, the E–Id axis plays a key role in the cell fate choice between adaptive and innate lymphoid lineages. However, the regulatory mechanism underlying Id2 expression, including Id2 suppression in T lineages, remains unknown.

T lymphopoiesis begins with the entry of lymphoid progenitors into the thymus, where the interaction of Notch1 on progenitors with DLL4 on TECs instructs T-cell lineage differentiation and gives rise to early T progenitors (ETPs) (Chen et al. 2019a; Hirano et al. 2022). These cells are defined by a CD4/CD8-double-negative (DN) lineage-negative (Lin⁻) CD44⁺CD25⁻Kit⁺ phenotype (ETPs) and further differentiate into DN2 thymocytes (CD44⁺CD25⁺; DN2a; Kit⁺, DN2b Kit⁻), cells in a developmental stage committed to the T lineage fate. Following T lineage commitment, TCRβ gene recombination occurs at the DN2–DN3 stages. After the β selection of the TCRβ gene, DN3 cells differentiate into DN4 cells and further into CD4⁺CD8⁺-double-positive (DP) cells. Loss of E-protein activity (E2A [Tcf3] and HEB [Tcf4]) in lymphoid progenitors results in lineage conversion of ETPs toward ILCs, including type 2 ILCs (ILC2s) and lymphoid tissue inducer (LTi)-like cells; however, these ILCs are rarely detected in wild-type adult mice (Miyazaki et al. 2017). In contrast to the adult thymus, recent single-cell RNA sequencing technology revealed the presence of ILCs in the thymus of wild-type fetuses, suggesting that ETPs favor the development of ILCs in the fetal thymus, whereas adult ETPs predominantly differentiate into cells in the αβT lineage (Kernfeld et al. 2018; Ferreira et al. 2021; Shin and McNagny 2021). This has raised the question of how the development of fetal versus adult thymocytes is mechanistically regulated.

In addition to displaying functions regulated by a common set of specific TFs (T-bet, Gata3, Rorgt, and Eomes), T cells and ILCs share TFs necessary for their development, including Gata3, Tcf1, and Bcl11b (Yagi et al. 2014; Yang et al. 2015; Hosokawa et al. 2018, 2020; Harly et al. 2019; Ren et al. 2022). However, how these common TFs are regulated at the developmental bifurcation of ETPs and how Notch signaling and the E-protein orchestrate these TFs for T-cell lineage commitment remain unknown.

In this study, we aimed to demonstrate the novel mechanisms by which E-protein and Notch1 signaling cooperatively instruct the T lineage commitment and suppress ILC development in the thymus (the E2A–Notch1 amplifier circuit) using multiple in vivo and single-cell multiome approaches. We also found that this regulatory circuit instructs fetal versus adult thymocyte gene expression programs during thymic ontogeny.

Results

DLL4 on TECs and N1En in T-cell and thymic ILC development

To assess the suppression of aberrant thymic ILC development by Notch1–DLL4 signaling, we analyzed the thymus of DLL4^fl/flFoxn1^Cre mice in which Dll4 was deleted in TECs. We observed a considerable increase in the number of thymic ILC2s and LTi-like cells in these mutant mice (Fig. 1A,B; Supplemental Fig. S1A).

Figure 1.

Regulation of the Notch1 enhancer (N1En) by the E-protein is essential for Notch1 expression during hematopoiesis. (A,B) ILC2s and LTi-like cells in thymi derived from adult control and Dll4^fl/flFoxn1^Cre mice are shown. (C) Notch1 enhancer (N1En) location and T- and B-cell TF binding patterns in T and B progenitors (ChIP-seq). RNA-seq and ATAC-seq data of ETPs from control and E2A/HEB-double-deficient (EHdKO) mice are shown. The GEO accession numbers used in the figures are shown in Supplemental Table S1. (D) Accessibility of N1En in hematopoietic progenitor cells. (E–G) Deletion of N1En (N1En^d/d) led to partial impairment of thymocyte development and reduced Notch1 cell surface expression through lymphocyte development. (E) CD4 versus CD8 expression, KIT versus CD25 expression (CD4/CD8-double-negative [DN]; Lin⁻), and CD19 and B220 expression are shown. (ETPs) CD44⁺CD25⁻Kit⁺, (DN2a) CD44⁺CD25⁺Kit⁺. (F) Absolute numbers of the indicated cells and the ratio of DN2a to ETP cells. (G) Notch1 cell surface expression in the indicated cell populations and FMO (fluorescence minus one) of Notch1. Lineage markers in A and E were CD3e, TCRγδ, B220, CD19, CD11b, Gr1, Ter119, CD11c, and DX5. N1En^d/d and control mice were analyzed at 4 weeks old. The data represent the means ± SD. (*) P < 0.05, (**) P < 0.01, (***) P < 0.001 (Student's t-test). Two to three independent experiments produced similar results (E–G), as did one experiment (A); n = 3 biological replicates.

We identified a putative cis-regulatory element (CRE) for the expression of Notch1 (Notch1 enhancer (N1En]), which is located upstream of the Notch1 exon 1 in ETPs (Fig. 1C). Notch1 mRNA expression and N1En chromatin accessibility in ETPs were markedly decreased without affecting promoter accessibility in E2A/HEB gene-deficient ETPs (EHdKO). In addition to E2A, other TFs including Bcl11b, Ikaros, PU.1, and Ebf1 were recruited to N1En in T- or B-cell progenitors (pro-T or pro-B), respectively (Fig. 1C). Increased accessibility of the N1En sequence began in hematopoietic stem cells (HSCs), and chromatin remained open in multipotent progenitors (MPPs) and common lymphoid progenitors (CLPs) until reaching the DN3 and B-cell precursor (pre-B) stages during the T- or B-cell development, respectively (Fig. 1D; Supplemental Fig. S1B). This chromatin configuration was accompanied by hypomethylated CpG DNA (Supplemental Fig. S1B). To prove the enhancer activity of N1En in vivo, we generated an N1En deletion mouse line (N1En^d/d). In line with a recent report (Kashiwagi et al. 2022), N1En deletion resulted in mild to moderate impairment of thymocyte development in both fetal and adult mice, characterized by a defect in the transition of ETPs to DN2a cells (Fig. 1E,F; Supplemental Fig. S1C,D). N1En deletion led to an increase in thymic B cells but did not affect the presence of thymic γδT cells, NK cells, NKT cells, myeloid cells, and ILC2s (Fig. 1E, F; Supplemental Fig. S1E,F). Furthermore, we found that N1En controls Notch1 mRNA and cell surface Notch1 expression in hematopoietic progenitors and developing T and B cells (Fig. 1G; Supplemental Fig. S1H). These results suggest that N1En activity controls Notch1 expression not only in immature DN thymocytes but also in early lymphoid progenitors; however, this attenuated Notch1 signaling caused by N1En deletion was sufficient to induce T-cell lineage commitment and suppress thymic ILC development in the adult thymus.

Forced Notch signaling requires E-protein activity for the induction of T-cell lineage commitment and suppression of ILC development

The deletion of E2A (Tcf3) and HEB (Tcf12) reduced Notch1 expression, which is regulated by N1En (Fig. 1C). Therefore, we examined whether forced Notch signaling could restore T-cell development in E2A/HEB-deficient mice. For this purpose, we generated mice (designated Rosa26^NICDIl7r^Cre/+) in which the Notch intracellular domain (NICD) was constitutively expressed starting from the CLP stage (Murtaugh et al. 2003; Schlenner et al. 2010; Chea et al. 2016). NICD expression resulted in a loss of B-cell development and ectopic T-cell development in the bone marrow (BM), accompanied by the loss of ILC precursors (ILCps) and ILC2s in the BM (Fig. 2A). Moreover, induction of NICD expression by Mb1^Cre (Rosa26^NICDMb1^Cre) (Hobeika et al. 2006), which starts being expressed in early B-cell progenitor cells, suppressed B-cell development and aberrantly induced differentiation of DP cells in the BM (Fig. 2B; Supplemental Fig. S2A). However, DP cells were not observed in the BM of R26^NICDCd19^Cre mice (Supplemental Fig. S2B). To examine whether NICD-induced T-cell lineage differentiation requires E-protein activity, we crossed Rosa26^NICDIl7r^Cre/+ mice with E2A/HEB-deficient mice (Tcf3^fl/flTcf12^fl/flRosa26^NICDIl7r^Cre/+) (Jones and Zhuang 2007; Miyazaki et al. 2017). The loss of E2A and HEB genes resulted in the complete absence of DP cells in the BM despite enforced Notch1 signaling, indicating that NICD-mediated aberrant T-cell development depends on E-protein activity (Fig. 2C). However, Bcl11b deletion did not block NICD-mediated aberrant DP cell development in the BM (Fig. 2D; Supplemental Fig. S2C). Subsequently, we examined whether forced Notch signaling restores thymic T-cell development in Tcf3^fl/flTcf12^fl/flIl7r^Cre/+ (E2A/HEB dKO) mice that display abolished T-cell lineage and a drastic increase in thymic ILCs (Miyazaki et al. 2017). To conduct the experiment in the absence of systemic lymphopenia, we analyzed fetal thymi at 18.5 days postcoitum (dpc). Although transgenic NICD expression resulted in a higher frequency of DN2a cells in Tcf3^fl/flTcf12^fl/flRosa26^NICDIl7r^Cre/+ fetuses, their differentiation into DN2b and DN3 cells was blocked, resulting in the complete loss of DP cells (Fig. 2E,F). In addition, an increased number of thymic ILC2s was observed in 4 week old Tcf3^fl/flTcf12^fl/flRosa26^NICDIl7r^Cre/+ mice, similar to in Tcf3^fl/flTcf12^fl/flIl7r^Cre/+ mice (Fig. 2G,H). Because these results could be conflicting with previous reports that active Notch signaling caused the degradation of E2A protein (Ordentlich et al. 1998; Nie et al. 2003), we examined the E2A protein levels in NICD-expressing ETP/DN2/DN3 cells in vitro and also compared the E2A protein levels in pro-B and DN2 cells in vivo. We observed significant but subtle differences among them (Supplemental Fig. S2D,E), suggesting that Notch signaling did not aggressively induce E2A degradation to inhibit E2A activity. Thus, we concluded that both E-protein activity and Notch signaling are required for T lineage commitment and suppression of ILC development.

Figure 2.

Forced Notch signaling critically requires E2A (Tcf3) and HEB (Tcf12) to induce T-cell development. (A) Aberrant DPs and loss of ILC2s and PLZF-expressing ILC precursors (ILCps) in the BM from R26^NICDIL7R^Cre/+ mice. The data are representative of two independent experiments. (B) Mb1^Cre-dependent NICD expression results in aberrant DP cell development and a loss of B-cell development in the BM. The data are representative of three independent experiments. (C) Aberrant DPs in the BM are shown. (D) Aberrant T-cell development in the BM induced by NICD expression in the absence of the Bcl11b gene. The data are representative of two independent experiments. (E) T-cell development in fetal thymi derived from 18.5 dpc fetuses. (F) Numbers of total fetal thymocytes and the ratios of DN2a to ETP cells and DN3⁺DN2b to DN2a cells. (G) Thymic ILC2s from 4 week old mice. (H) The numbers of total thymocytes and thymic ILC2s. (A–D,G,H) All mice were analyzed at 4 weeks old. The data represent the means ± SD. (*) P < 0.05, (**) P < 0.01, (***) P < 0.001, (****) P < 0.0001 (Student's t-test). (E–H) The data are representative of more than two independent experiments.

An E2A–Notch1-amplified circuit determines T-cell lineage commitment and suppresses thymic ILC development

To investigate the mechanisms underlying the roles of E-proteins and Notch signaling in promoting T-cell lineage commitment, we examined gene expression and chromatin accessibility in ETPs isolated from the thymi of control, Tcf3^fl/flTcf12^fl/flIl7r^Cre/+ (E2A/HEB dKO [EHdKO]), and Tcf3^fl/flTcf12^fl/flRosa26^NICDIl7r^Cre/+ (EHdKO;NICD⁺) 18.5 dpc fetuses. We found that 1082 genes were differentially expressed genes (DEGs) in EHdKO or EHdKO;NICD⁺ ETPs versus control ETPs (Fig. 3A). Notably, among the DEGs, T-cell signature genes (Ptcra, Lef1, Rag1, and Rag2) were downregulated, whereas the genes characteristic of ILCs (Maf, Il4, Pdcd1, Stat1, and Gzmb) were upregulated (Fig. 3A). Gene ontology (GO) analysis of the upregulated genes in the EHdKO and EHdKO;NICD⁺ ETPs revealed effector cell functions, whereas those of the downregulated genes related to aspects of T-cell differentiation and consisted of a loss of T-cell potential and concurrent progression to an ILC lineage fate (Supplemental Fig. S3A,B). To directly relate DEGs to putative enhancer regions regulated by E2A and Notch signaling, we integrated chromatin immunoprecipitation with sequencing (ChIP-seq) data for E2A and RBPj (a downstream TF of Notch signaling) occupancies in developing T cells with DEGs. We identified putative enhancers co-occupied by both factors and related them to biological processes associated with DEGs (Fig. 3B; Supplemental Fig. S3B). Compared with the control ETPs, in both the EHdKO and EHdKO;NICD⁺ ETPs, Rag1, Ptcra, Notch3, and Tcf1 (Fig. 3B, marked in blue) were downregulated, whereas Id2, Rora, Socs1, and Runx3 (Fig. 3B, marked in red) were suppressed by transgenic NICD expression (Fig. 3B). These results suggest that E2A and Notch signaling cooperatively regulate the expression of T-cell signature genes and that Notch signaling alone is sufficient to suppress some genes characteristic of ILC identity.

Figure 3.

Notch signaling suppresses a set of ILC signature genes, such as Id2 and Rora, but fails to induce T-cell signature genes in the absence of E2A and HEB. (A) Heat map of differentially expressed genes identified via DEseq using ETPs in the fetal thymi from control, EHdKO (Tcf3^f/fTcf12^f/fIL7R^Cre/+), and EHdKO-NICD (Tcf3^f/fTcf12^f/fR26^NICDIL7R^Cre/+) mice. (B) The Venn diagram shows overlaps between E2A and RBPj binding sites (left), and the overlapping genes were compared with DEGs in EHdKO and EHdKO-NICD⁺ ETPs compared with wild-type control ETPs (right). (C) Volcano plots of the mRNA expression in ETPs from EHdKO and EHdKO-NICD⁺ mice. The red and blue dots represent upregulated and downregulated genes, respectively (more than twofold; P < 0.05). (D) Browser images of normalized RNA-seq and ATAC-seq reads in control, EHdKO, and EHdKO-NICD⁺ ETPs, and ChIP-seq data for E2A and RBPj and histone H3K27 acetylation in DN3 and ILC2s across the Id2, Rora, and Rag1/2 gene loci. (E) ATAC-seq read coverage in ETPs from control (blue line), EHdKO (orange line), or EHdKO-NICD⁺ (green line) mice, plotted as a function of genomic distance from E2A and RBPj cobinding sites (top); RBPj, but not E2A, binding sites (middle); and CTCF binding sites (bottom). (F) Model of the E2A–Notch1–Id2 amplifier circuit that induces T-cell lineage and simultaneously suppresses the ILC lineage.

We further investigated the DEGs in EHdKO:NICD⁺ ETPs and found 849 DEGs (upregulated, 317; downregulated, 532) compared with the EHdKO ETPs (Fig. 3C; Supplemental Fig. S3C). Notably, the downregulated genes included several ILC signature genes such as Id2, Rora, Maf, Rorc, Ikzf3, Cxcr6, Il2rb, Icos, and Il18r (Fig. 3B,C). Here, we focused on the Id2, Rora, and Rag1/2 gene loci, which are representative ILC and T-cell signature gene loci regulated by E2A and Notch signaling (Wong et al. 2012; Miyazaki et al. 2020; Ferreira et al. 2021). R-TEn (Rag–T-cell enhancer), a T lineage-specific enhancer region for Rag1/2 expression, was co-occupied by E2A and RBPj, and NICD expression did not fully restore Rag1/2 expression or chromatin accessibility in the R-TEn region (Fig. 3D; Supplemental Fig. S3E). Chromatin accessibility (assay for transposase-accessible chromatin with sequencing [ATAC-seq]) of the E2A- and RBPj-co-occupied regions upstream of Id2 and Rora genes was markedly reduced in EHdKO/EHdKO:NICD⁺ ETPs, and transgenic NICD expression did not alter chromatin accessibility in these regions despite the suppression of Id2 and Rora expression (Fig. 3D; Supplemental Fig. S3D,E). ATAC-seq signals in the RBPj- and E2A-co-occupied regions were weaker in both EHdKO (Fig. 3E, top panel, orange line) and EHdKO;NICD⁺ (Fig. 3E, top panel, green line) ETPs than in control ETPs (Fig. 3E, top panel, blue line). In contrast, the RBPj binding sites without E2A occupancy or CTCF binding sites used as control loci remained unchanged (Fig. 3E, middle and bottom panels). Collectively, these results suggest that E-protein principally maintains chromatin accessibility at CREs bound by E2A and Notch–RBPj and that Notch-mediated signals suppress the expression of a set of ILC genes without affecting enhancer accessibility. These findings suggest that E2A/HEB regulate Notch1 expression through the Notch1 enhancer, and E-protein and Notch signaling cooperatively activate a gene expression program of T-cell lineage and, in parallel, suppress expression of a set of ILC signature genes. In addition, Notch1-controlled inhibition of Id2 expression increased E2A DNA binding activity, which further upregulated T-cell gene expression while simultaneously suppressing ILC genes (E2A–Notch1 amplifier circuit) (Fig. 3F).

Notch signaling and E2A/E47 synergistically activate a set of T-cell-specific transcription signatures and suppress ILC development by modulating the enhancer landscape

Although N1En deletion led to reduced Notch1 expression, the signals generated by this reduced receptor were sufficient to induce T-cell development (Fig. 1E). Similarly, mice deficient in E47 (a splicing isoform of the E2A gene) showed normal thymocyte development accompanied by reduced thymocyte cellularity (one-half to one-third of control thymocytes) owing to the compensatory expression of HEB and E12 (another isoform of the E2A gene) (Bain et al. 1999; Beck et al. 2009). To demonstrate the significance of the E2A–Notch1 circuit, we attempted to attenuate the function of this circuit by generating mice that were double-deficient in N1En and E47 (designated N1En^d/dE47^−/−) (Fig. 4A). Notably, these mice demonstrated a developmental block at the DN1 stage, as indicated by the absence of thymocytes at all maturation stages (Fig. 4A). Additionally, they demonstrated a substantial increase in thymic ILC2s, which was comparable with that in mice that were either double-deficient in E2A and HEB or lacked DLL4 expression (Fig. 4B,C). We further analyzed thymi isolated from 18.5 dpc control, N1En^d/d, and N1En^d/dE47^−/− fetuses. Although lymphoid progenitor cells from control and N1En^d/d fetuses differentiated into the DP stage, N1En^d/dE47^−/− fetuses displayed developmental arrest during the transition from DN2a to DN2b cells (Fig. 4D; Supplemental Fig. S4A). Increased numbers of thymic ILC2s were noted in the thymi of N1En^d/d, N1En^d/dE47^+/−, and N1En^d/dE47^−/− fetuses (Fig. 4E; Supplemental Fig. S4B). These results clearly demonstrated not only the functional redundancy between Notch1 signaling and E-protein activity but also the importance of the Notch1–E2A circuit in T lineage commitment and the suppression of ILC development.

Figure 4.

Notch signaling and E2A/E47 synergistically activate a set of T-cell signature genes and suppress ILC development by orchestrating the enhancer landscape. (A,B) CD4 versus CD8 expression, CD44 versus CD25, and ST2 versus Sca1 are shown (4 week old control, N1En^d/d, or N1En^d/dE47^−/− mice). (C) Numbers of total thymocytes and thymic ILC2s from 4 week old mice. The data are representative of three independent experiments. (D,E) Flow cytometric analysis of fetal thymi from 18.5 dpc fetuses. The absolute numbers of the indicated cell populations (D) and thymic ILC2s (E) and the ratios of DN2a to ETP and DN2b to DN2a cells (D) are shown. The data are representative of four independent experiments. Control (B6) and mutant fetuses (18.5 dpc) were analyzed on the same day in one experiment. (F) Circos plots showing overlapping genes among N1En^d/d, E47^−/−, and N1En^d/dE47^−/− ETPs. Upregulated and downregulated gene lists compared with genes in control ETPs. All the ETPs were derived from 16.5 dpc fetal thymi. (G) Gene ontology (GO) analysis of DEGs in ETPs from N1En^d/d, E47^−/−, and N1En^d/dE47^−/− fetuses. (H) Heat map showing the correlation of ATAC-seq reads in the distal peaks for the indicated cell types, including C57BL/6 (B6), N1En^d/d, E47^−/−, and N1En^d/dE47^−/− ETPs. Ctrl:ETP and EHdKO (E2A/HEB dKO). (I) Venn diagram showing overlaps among E2A and RBPj occupancies and downregulated ATAC-seq peaks near downregulated genes according to RNA-seq in N1En^d/dE47^−/− ETPs. GO analysis and representative overlapping genes are shown. The data represent the means ± SD. (*) P < 0.05, (**) P < 0.01, (***) P < 0.001, (****) P < 0.0001 (Student's t-test).

We next performed RNA-seq and ATAC-seq using fetal ETPs (Fig. 4F; Supplemental Fig. S4C). The GO analysis of the upregulated genes showed that N1En^d/dE47^−/− ETPs tended to differentiate into effector cells. In contrast, the genes downregulated in N1En^d/dE47^−/− ETPs were associated with thymic T-cell differentiation (Fig. 4G; Supplemental Fig. S4D). The profiles of ATAC-seq reads differed from those of the transcriptome. Specifically, the accessible distal enhancer landscape in N1En^d/d ETPs was similar to that in control ETPs but distinct from that in E47^−/− and N1En^d/dE47^−/− ETPs (Supplemental Fig. S4E,F).

To further define the roles of E2A and Notch1 signaling in the establishment of enhancer landscapes and to identify similarities between each ETP and hematopoietic cell, we compared ATAC-seq reads from each ETP with public ATAC-seq data derived from murine immune cell subsets. We found that the enhancer repertoires detected in control and N1En^d/d ETPs were closely related to those detected in DN1/DN2a cells and hematopoietic progenitors, whereas enhancer repertoires identified in the E47^−/− and N1En^d/dE47^−/− ETPs were most closely related to those in EHdKO ETPs, precursors of NK/ILC2s, and mature ILC2s (Fig. 4H). These results suggest that E2A/E47 maintains an enhancer landscape relevant to the T lineage, whereas Notch1 signaling largely affects gene expression but not enhancer accessibility.

To identify genes directly controlled by E2A activity and Notch signaling, we integrated the transcriptome and ATAC-seq data of N1En^d/dE47^−/− ETPs with the E2A and RBPj ChIP-seq data. This analysis revealed a twofold decrease in ATAC-seq peaks near genes whose expression was downregulated in N1En^d/dE47^−/− ETPs. Next, we complemented this information with E2A- and RBPj-co-occupied regions. We identified 31 putative enhancer regions for T-cell signature genes closely related to the Rag1/2–Ku70/80 complex and Notch signaling pathway by GO analysis (Fig. 4I). Indeed, these putative regulatory regions were regulated by E2A and Notch signaling, and the chromatin accessibility of these regions depended on E47 activity (Supplemental Fig. S4G). Together, these results suggest that the E2A–Notch1 circuit regulates the enhancer activity of T-cell signature genes that related to TCR recombination.

Next, we attempted to identify T-cell core enhancer regions by analyzing core TF occupancies. Because Bcl11b and Gata3 are shared between T cells and ILCs, we integrated the E2A- and RBPj-co-occupied regions with the Bcl11b and Gata3 binding sites identified in DN3 cells. We identified 416 T-cell core TF binding sites (Supplemental Fig. S4H). Integration of ChIP-seq data for Rora, Bcl11b, Runx1, and Gata3 in ILC2s detected 990 co-occupied putative enhancer regions as ILC2 core TF binding sites (Supplemental Fig. S4I). Importantly, only four peaks were shared between T-cell and ILC2 core TF binding sites despite the presence of 46 shared target genes (Supplemental Fig. S4J), suggesting that there are distinct regulatory mechanisms involved in T-cell versus ILC2 development and that E2A activity and Notch signaling recruit Gata3 and Bcl11b in a cell type-specific manner.

Single-cell multiome (scMultiome) analysis reveals developmental trajectories that underpin fetal thymocyte development

To elucidate the involvement of E2A and Notch signaling in the developmental trajectories of T-cell lineage and ILCs at a single-cell level, we used a single-cell multiome (scMultiome) to simultaneously resolve the single-cell transcriptome (scRNA-seq) and regulome (scATAC-seq). We used weighted nearest neighbor (WNN) analysis for the simultaneous comparison of scRAN-seq and scATAC-seq data to characterize cell states and allow for better visualization of cell developmental trajectories (Supplemental Fig. S5A; Buenrostro et al. 2018; Hao et al. 2021). We identified clusters of thymocytes, endothelial cells, mesenchymal cells, and TECs by evaluating the expression of Ptprc, Cd3e, Egfl7, Pdgfra, Epcam, and Icos, respectively (Supplemental Fig. S5A). Furthermore, to analyze the developmental bifurcation of T-cell and ILC lineages, we extracted Ptprc^hiCd3e⁺ and Ptprc^hiIcos⁺ clusters (clusters 0 and 1) (Supplemental Fig. S5B). Next, we identified 11 distinct clusters (T0–T11) according to their representative gene expression of T-cell signature genes, cytokine genes, surface marker genes, and TFs; these included ETPs (T11/T4), αβT lineages (DN2–3; T2, T1, T7, and T0), γδT (T5/T6), ILC precursor (ILCp; T9), ILC1/NK (T3), ILC2 (T10), and unknown T cells (T8) (Fig. 5A–C; Supplemental Fig. S5C,D). Consistent with the flow cytometric data, we observed fewer αβT cells and an increased frequency of ILC1/NK, ILC2, and ILCp cells in the N1En^d/dE47^−/− thymi, and the frequency of cells expressing T-cell signature genes was markedly reduced, whereas the frequencies of Il4-, Il13-, and Ifng-expressing cells (compatible with ILC2s and ILC1/NK cells) were enhanced (Fig. 5B,C).

Figure 5.

Single-cell multiome analysis reveals the developmental trajectories toward T-cell and ILC lineages. (A) Uniform manifold approximation and projection (UMAP) of scMultiome data of Ptprc^hiCd3e⁺ (cluster 0) and Ptprc^hiIcos⁺ (cluster 1) thymocyte-gated cells derived from 15.5 dpc control (C57BL/6) and N1En^d/dE47^−/− fetuses (Supplemental Fig. S5B). Cells are colored according to the results of clustering analysis using thymocyte-gated cells. (B) The ratio of each cluster is shown. (C) UMAP plot showing the expression of the indicated genes. (D) Graphs showing the accessibility of the T-cell core and ILC2 core enhancer regions (as described in Supplemental Fig. 4I,J) in single cells in each cluster. (E) Arch plots showing the peak correlation of open chromatin regions across the Rag1/2 and Id2 gene loci. (F,G) Pseudotime analyses showing developmental trajectories toward αβT and ILC lineages from the earliest Flt3-expressing ETPs. (F) Flt3 expression is shown in UMAP plots. (G) Pseudotime trajectories are shown in the UMAP space for the combined Ctrl and N1En^d/dE47^−/− thymocyte scMultiome data sets (left) and Ctrl (middle) and N1En^d/dE47^−/− (right) thymocytes.

Single-cell-resolved ATAC-seq reads mapped around T-cell core enhancers (Supplemental Fig. S4H) were more numerous in the control ETPs than in the N1En^d/dE47^−/− ETPs and subsequent αβT lineage clusters (Fig. 5D). In the N1En^d/dE47^−/− αβT-cell clusters, this accessibility pattern remained unchanged, whereas that of the ILC2 core enhancers (Supplemental Fig. S4I) increased (Fig. 5D). In contrast to wild-type cells, N1En^d/dE47^−/− thymocytes displayed fewer predicted interactions across the Rag1/2 gene locus and more interactions across the Id2 locus (representing typical T-cell and ILC loci, respectively) (Fig. 5E). To characterize the separate developmental paths of ETPs toward αβT cells and ILCs, we arranged incremental thymocyte developmental stages along a pseudotime trajectory using Monocle (Qiu et al. 2017). Two separate trajectories emanated from the earliest T progenitors, as defined by their high Flt3 expression (Fig. 5F; Allman et al. 2003; Sambandam et al. 2005). They were directed toward the αβT lineage (T2) and ILCps (T9), respectively (Fig. 5G). Among N1En^d/dE47^−/− thymocytes, there was also a trajectory from αβT lineage cells (T2) to ILCps (T9) (Fig. 5G, conversion path). These results suggest that Notch1 signaling and E-protein activity cooperatively drive T lineage specification in ETPs and enforce T lineage fate determination to prevent conversion to ILCs, which is achieved by regulating T-cell core enhancers and suppressing ILC core enhancers.

Single-cell multiome analysis reveals an E2A–Notch1 circuit that amplifies a T-cell-specific gene expression program

Next, we analyzed the regulome to explore the mechanism by which Notch1 signaling and E-protein activity enforce the T-cell lineage determination. The expression of TFs and access to their cognate binding motifs strongly diverged with cell differentiation (Fig. 6A). In control cells, although E2A and Notch1 were expressed at all analyzed thymocyte stages, their binding motifs were mostly accessible after T-cell lineage (Fig. 6A). Gata3 was also expressed in both lineages, but accessibility to its motif was transient in both lineages (early T cells [T2] and ILCps [T9]) and γδT cells (T5/T8) (Fig. 6A). Expression of Spi1, which is characteristic of the progenitors, and of Rora, which is typical of ILCs, was consistent with the accessibility of their respective motifs. In contrast, there was limited access to the E2A and RBPj motifs in N1En^d/dE47^−/− T lineage cells (Fig. 6A). Higher Rora and Gata3 expression levels and corresponding access to specific binding motifs were observed in N1En^d/dE47^−/− thymocytes (Fig. 6A). To investigate the mechanism by which N1En^d/dE47^−/− T2 cells of the αβT lineage were converted to ILCps (T9), we analyzed TF expression and motifs in this trajectory. The binding motifs of Rora, Runx3, Tcf1, and Gata3, but not Nfil3, were significantly enriched in the accessible regions of converting cells (Fig. 6B; Supplemental Fig. S6A). We further examined the variation in TF expression and access to binding motifs, accompanied by T and ILC differentiation. Although the E2A mRNA level was not altered during T-cell differentiation, accessibility to its specific binding motif increased toward DN2a thymocytes and was further enhanced at DN2b and DN3 (Fig. 6C). The expression levels of Notch1, Tcf12/HEB, and Lef1 and access to their motifs were synchronized with increased accessibility to the E2A motif (Fig. 6C; Supplemental Fig. S6B). In contrast, Gata3 was stably expressed in the T and ILC lineages, yet accessibility to its binding motif transiently increased toward either DN2a thymocytes or ILCps but then decreased with further differentiation into either cell lineage (Fig. 6C). Rora and Gata3 expression and accessibility to their motif were largely enhanced in N1En^d/dE47^−/− T cells and ILCs, suggesting that increased transcriptional activity of Rora and Gata3 may induce conversion of cells from the T lineage to the ILC lineage (Fig. 6C). We also assessed the time required for cells to branch into separate T and ILC lineages by using the single-cell trajectory reconstruction, exploration, and mapping (STREAM) Python pipeline (Chen et al. 2019b). By inferring a developmental hierarchy from ETPs to each lineage, we found that the differentiation time for the αβT lineage was substantially longer than that for the ILC2 and ILC1/NK cell lineages and that the E2A binding motif was mostly accessible after T lineage commitment (Supplemental Fig. S6C). We next classified the expression of representative TFs of the T and ILC lineages into six groups (Supplemental Fig. S6D). We assessed the differences in the expression of these genes in N1En^d/dE47^−/− thymocytes (Supplemental Fig. S6E). These results suggested that, during T-cell specification, E2A activity and Notch1 signaling induce Tcf12, Lef1, Gfi1, and Bcl11b and suppress Maf, Rora, Gata3, Runx3, Zbtb16, and Id2 transcription, thereby enabling a gene expression program that enforces T lineage commitment and cell fate (Supplemental Fig. S6F).

Figure 6.

The E2A–Notch1 circuit orchestrates appropriate TF recruitment and T-cell signature gene expression for T-cell lineage commitment. (A) UMAP plots showing the expression and accessible binding motifs of the indicated transcription factors. (B) Cells converting from T2 to T9 have higher frequencies of access to the Rora, Runx3, Tcf7, and Gata3 binding motifs. Dot plots revealing access to their motifs in each cell population. (C) Curve graphs showing the mRNA expression (top) and accessible binding motif (middle) of the indicated TF during T-cell (solid line) and ILC (dotted line) development. (Blue line) Control, (orange line) N1En^d/dE47^−/−. The bottom panel shows the densities of clusters at each pseudotime point analyzed by Monocle 3. The line colors correspond to the clusters defined in Figure 5A. (D) Curve graphs showing the mRNA expression of the indicated genes (solid line) and accessible binding motifs of E2A (green dotted line) and RBPj (blue dotted line) at the indicated gene locus during αβT-cell and ILC development, as shown in Supplemental Figure S6G. (E) Model depicting how the E2A–Notch1 circuit functions along with other TFs in the specification (top) and commitment (bottom) stages, respectively (see the Discussion).

Because accessibility to the binding motifs of E2A and Notch1/RBPj was markedly increased, especially after T lineage commitment (DN2b/DN3 stages), we hypothesized that the dynamic expression of T-cell signature genes was controlled by the E2A–Notch1 amplifier circuit. This contention was confirmed by our observation that, in parallel with the increased accessibility to the E2A and RBPj binding motifs at T-cell signature gene loci (Cd3e/d, Bcl11b, Lef1, Ptcra, Xrcc6, and Rag1), expression of these genes was increased in control T cells but not in ILCs (Fig. 6D; Supplemental Fig. S6G). Taken together, the results of the scMultiome analysis revealed precise regulation of T-cell signature genes by the E2A–Notch1 amplifier circuit during priming, specification, and commitment to the T-cell lineage (Fig. 6E).

Thymic epithelial Dll4 expression regulates Id2 expression in ETPs, which modulates the development of ILC2s during thymic ontogeny

Our findings suggest a differential yet central biological role of the E2A–Notch1 circuit for the differentiation of ILCs and T cells from ETPs. Thymic ILC2 cellularity was the highest in the fetal thymus and started to decline 2 weeks after birth (Fig. 7A). To investigate changes in thymic ILC2 development, we used mice with either an E2A-GFP fusion protein reporter or an Id2-YFP reporter (Zhuang et al. 2004; Yang et al. 2011). E2A protein levels were highest in ETPs isolated from 3 day old mice (Supplemental Fig. S7A). In line with E2A, Notch1 expression was also changed (Supplemental Fig. S7B). Notably, the frequency of Id2-YFP-positive cells among the ETPs was ∼14% in 13.5 dpc fetal thymi and rapidly decreased to barely detectable levels after birth (Fig. 7B; Supplemental Fig. S7C). Furthermore, Id2-YFP-positive ETPs from 13.5 dpc fetal thymi preferentially gave rise to ILC1/NK cells when cocultured with OP9/DLL1 stromal cells (Fig. 7C).

Figure 7.

The E2A–Notch1/DLL4–Id2 axis modulates age-controlled thymic ILC2 development during ontogeny. (A) Absolute number of thymic ILC2s in C57BL/6 mice. (FT18) Fetal thymus at 18.5 dpc and 2, 4, 8, and 12 weeks after birth. (B) Id2-YFP expression in ETPs from Id2-YFP reporter (Id2^yfp/+) mice with aging. To avoid dispersion among the experiments, all the samples were analyzed on the same day. (C) Sorted YFP-expressing ETPs from 13.5 dpc Id2^yfp/+ fetuses were cocultured with OP9/DLL1 stromal cells. Representative flow cytometric analysis of CD4 and CD8 expression and NK1.1 and Id2-YFP expression are shown. The data are representative of two independent experiments. (D) Violin plot of Dll4 and Il7 expression in each cluster (E0–E14) of TECs from adults (pink) and fetuses (green) (related to Supplemental Fig. S7C). (E) Representative Id2-YFP expressions in ETPs from 14.5 dpc fetuses are shown. The data are representative of two independent experiments. (F) Model depicting the changes in DLL4 expression and the E2A–Notch1–Id2 axis between fetal and adult thymi. The data represent the means ± SD. (*) P < 0.05, (**) P < 0.01, (***) P < 0.001, (****) P < 0.0001 (Student's t-test).

Given the change in frequency of Id2-positive ETPs with age and because we demonstrated suppression of Id2 expression with Notch signaling activation (Fig. 3), we postulated that the change in Id2 expression was caused by age-controlled alterations in Notch1–DLL4 signaling. To test this notion, we examined Dll4 expression in TECs isolated from fetal and adult thymi (Supplemental Fig. S7D). Dll4 was highly expressed in TEC clusters (epithelial [E]) E0, E2, E4, and E11 in adult TECs; in contrast, clusters of E0 and E2 mTECs were not abundant, and E4 and E11 cTECs exhibited reduced Dll4 expression in the fetal thymus compared with Il7 expression in these clusters (Fig. 7D; Supplemental Fig. S7D). The fetal Id2 expression pattern was confirmed in 14.5 dpc fetuses with TEC targeted disruption of DLL4 (Hozumi et al. 2008), as the frequency of Id2-YFP-positive ETPs was significantly increased in DLL4^fl/flFoxn1^Cre fetuses (Fig. 7E). Thus, Dll4 expression levels in fetal and adult thymi may control Id2 expression in ETPs, which determines T versus ILC lineage choice during thymic ontogeny by modulating the E2A–Notch1 circuit (Fig. 7F).

Discussion

T-cell development in the thymus undergoes changes during thymic ontogeny: There are the developmental waves of tissue-resident γδT cells, regulatory T cells, and invariant NK T cells. Similarly, ETPs in the fetal thymus favor the development of ILC2 (Ferreira et al. 2021), whereas adult ETPs differentiate into αβT lineage. However, the age-related modulation of T-cell versus ILC development and the mechanisms underlying this distinction remained poorly understood. Here, we showed that a simple amplifier circuit—involving Notch signaling and Id2 expression—can induce differences in E-protein activity to orchestrate a fetal versus adult thymocyte gene expression program. Consistent with the dynamic changes in Dll4 expression on TECs during the early thymus anlages (Tsukamoto et al. 2005), Id2-expressing ETPs are drastically declined even in the fetal thymus (Fig. 7). This finding also suggests that increasing Dll4 expression on TECs at early thymus anlages induces a developmental shift from innate to adaptive lymphocytes.

E2A (Tcf3) is critically required for not only B-cell but also T-cell lineage commitment, whereas Id2 is essential for ILC development, reflecting the role of the E2A–Id2 axis in determining the lineage fates of adaptive versus innate lymphocytes (Hidaka et al. 2022). Because the E2A/E-protein regulates Rag1/2 gene enhancers, E2A specifies adaptive lymphocyte characteristics; that is, V(D)J recombination of Ig and TCR genes (Miyazaki et al. 2020). In the same way that E2A orchestrates B-cell fate together with Ebf1 and Foxo1 (Lin et al. 2010), E2A plus Notch signaling instructs T lineage commitment; however, their synergistic function remains unclear because expression of the Notch1 gene is already compromised in the absence of E-protein genes (Ikawa et al. 2006; Yashiro-Ohtani et al. 2009; Braunstein and Anderson 2011; Miyazaki et al. 2011). Here, we evidently show that Notch signaling requires E-protein activity for T-cell lineage commitment. E-protein orchestrates enhancer landscapes characteristic of the T-cell lineage by priming and maintaining enhancer accessibilities of common target genes, and Notch1 signaling enhances the expression of these genes via either a positive (T-cell genes) or negative (ILC genes) mechanism. Notch/RBPj can bind to either chromatin-accessible or -inaccessible sites during neuronal development, and Notch/RBPj-bound accessible sites correlate with active expression (van den Ameele et al. 2022). Strong Notch signaling induces transcriptional bursts of target genes, and lineage-specific TF-mediated enhancer priming is indispensable for this robust gene expression (Falo-Sanjuan et al. 2019). In concert, E-protein binding to T-cell lineage-specific enhancers enables RBPj binding and activation of a T-cell lineage-specific transcription signature upon commitment.

T cells and ILCs commonly share TFs for their development, and our scMultiome analysis demonstrated differences in the kinetics of their expression and function between T-cell and ILC development. We also identified distinct T-cell or ILC2 core enhancer regions, and E2A and Notch1 signaling regulate T-cell core enhancers during αβT-cell differentiation. These results indicate that E-protein and Notch signaling appropriately recruit Gata3 and Bcl11b to T-cell core enhancers in a cell type-specific manner. N1En^d/dE47^−/− T cells and ILCs showed higher levels of Gata3 expression/activity, and cells that converted from the αβT lineage to ILCp showed increased Gata3 activity, suggesting that excessive Gata3 activity facilitates ILC development at the expense of differentiation into the T-cell lineage. Consistent with this, a reduction in Gata3 expression in E2A^−/− thymocytes facilitates thymocyte development, and forced Gata3 expression inhibits T-cell potential development (Taghon et al. 2007; Rothenberg and Scripture-Adams 2008; Xu et al. 2013). In summary, E2A and Notch1 signaling modulate Gata3 activity to enforce the αβT lineage by inhibiting lineage conversion to alternative ILCs (Supplemental Fig. S6F).

Although Bcl11b plays a role in the development of both T cells and ILCs (Hosokawa et al. 2020; Holmes et al. 2021), we found that its expression level was even greater in αβT cells than in thymic ILCs, which was induced by Notch1 and E2A. The ThymoD locus, which regulates the high level of Bcl11b expression in the T lineage, is defined as a superenhancer (SE) accompanied by a high level of histone H3K27 acetylation in T-cell progenitors/precursors (Supplemental Fig. S6I; Whyte et al. 2013). However, we found that ILC2s exhibited decreased H3K27 acetylation at the ThymoD locus but increased H3K27 acetylation (SE) near the Bcl11b promoter. Additionally, we identified potential RBPj and E2A binding sites at the ThymoD locus (Supplemental Fig. S6I). These results suggested that Notch1 signaling and E2A amplify Bcl11b expression, possibly through activation of ThymoD expression, during T-cell lineage specification. Because Tcf1, E-proteins, and Bcl11b are involved in SE formation (Hu et al. 2018; Johnson et al. 2018; Miyazaki et al. 2020), we suggest that the E2A–Notch1 circuit induces and amplifies HEB, Lef1, and Bcl11b expression and that these TFs cooperate with each other to amplify T-cell signature gene expression by orchestrating SE formation, which causes T cells to diverge from ILCs (Fig. 6E).

In conclusion, our findings indicate that a distinct feedback amplifier circuit involving E-proteins and Notch signaling specifies fetal and adult T-cell fates. Furthermore, in the fetal versus adult thymus, a simple feedback amplifier circuit, dictated by Notch-mediated signals and Id2 abundance, differentially orchestrates αβT versus ILC development. These results indicate the possibility of engineering de novo regulatory circuits that convert fetal-specific into adult-specific gene programs.

Materials and methods

Sorting of total thymic cells for single-cell multiome (scMultiome) analysis and library preparation

Thymi were obtained from three control (B6) and four N1En^d/dE47^−/− fetuses (15.5 dpc) and pooled for each genotype. Pooled thymi were digested in 0.05 U/mL liberase (Roche) and 0.01% (w/v) DNase I (Sigma-Aldrich) in RPMI1640 medium (Wako) by incubating three times for 12 min at 37°C. Digestion reactions were stopped by 2× volume of phosphate-buffered saline (PBS; Wako) plus 2% fetal bovine serum (Cosmo Bio) and 1 mM EDTA. Cells were spun down and stained with 7-amino actinomycin-D (7AAD; Calbiochem) diluted 500-fold with PBS plus 2% FBS to exclude dead cells. 7AAD-negative cells were isolated into RPMI1640 medium plus 5% FBS using a FACS Aria III (BD). After washing with PBS containing 0.04% BSA, sorted cells were suspended in lysis buffer containing 10 mM Tris-HCl (pH 7.4), 10 mM NaCl, 3 mM MgCl₂, 0.1% Tween-20, 0.1% NP-40, 0.01% digitonin, 1% BSA, 1 mM DTT (Sigma-Aldrich), and RNase inhibitor (Sigma-Aldrich) for 3 min on ice. Lysed cells were washed with wash buffer containing 10 mM Tris-HCl (pH 7.4), 10 mM NaCl, 3 mM MgCl₂, 0.1% Tween- 20, and 1% BSA. Nucleus concentration was adjusted to 3200–3300 nuclei/μL to target up to 10,000 nuclei with diluted nucleus buffer containing nucleus buffer (10x Genomics), 1 mM DTT (Sigma-Aldrich), and RNase inhibitor (Sigma-Aldrich). After centrifuging the solution, a nuclear pellet was obtained by removing the supernatant, and the pellet was resuspended in wash buffer.

scATAC and snRNA sequencing

Nuclei were captured via 10x Chromium single-cell Epi Multiome ATAC + Gene Expression Chemistry v1 (10x Genomics) to target up to 10,000 nuclei per channel, and scATAC-seq and snRNA-seq libraries were generated according to the manufacturer's instructions. The resultant libraries were then sequenced using an MGI DNBSEQ G400RS platform (version 1.1.0.108; MGI Tech). Briefly, for G400RS runs, libraries were subjected to adapter conversion PCR using the MGIEasy universal library conversion kit (MGI Tech 1000004155), circularized from the dsDNA, and sequenced using the DNBSEQ-G400RS high-throughput sequencing set (FCL PE100; MGI Tech 1000016950) according to the manufacturer's protocol (scATAC-seq: 89 cycles for read1, 89 cycles for read2, 24 cycles for i5, and eight cycles for i7; snRNA-seq: 28 cycles for read1, 150 cycles for read2, 10 cycles for i5, and 10 cycles for i7).

Quantification and statistical analysis

P-values were calculated with the two-tailed Student's test for two-group comparison and one-way ANOVA as applicable with Microsoft Excel or Prism. The statistical significance level was 0.05.

Data and material availability

Data for RNA-seq, ATAC-seq, and scMultiome have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus (RNA-seq and ATAC-seq: GSE245401; scMultiome: GSE254290). All (other) data needed to evaluate the conclusions in this study are presented here or in the Supplemental Material and Supplemental Table S1. All of the mice were either commercially available or available under a material transfer agreement (MTA). The codes and scripts used in this study are available at GitHub (https://github.com/tken18/2025GenDev_Tcell).

Competing interest statement

The authors declare no competing interests.