Phosphorylation of Runx protein controls helper CD4+ T cell versus cytotoxic CD8+ T cell lineage choice

Ogawa, Chihiro; Okuyama, Kazuki; Kojo, Satoshi; Nishino, Kohei; Machiyama, Hiroaki; Padhi, Aditya K.; Takahashi, Hirotaka; Ebihara, Takashi; Tenno, Mari; Zhizhen, Qin; Muroi, Sawako; Ito, Kosei; Sawasaki, Tatsuya; Zhang, Kam Y. J.; Xue, Hai-Hui; Yokosuka, Tadashi; Kosako, Hidetaka; Taniuchi, Ichiro

doi:10.1038/s41590-026-02441-6

Download PDF

Article
Open access
Published: 19 February 2026

Phosphorylation of Runx protein controls helper CD4⁺ T cell versus cytotoxic CD8⁺ T cell lineage choice

Nature Immunology (2026)Cite this article

Subjects

Abstract

MHC-I- and MHC-II-selected CD4⁺CD8⁺ precursor thymocytes differentiate into cytotoxic CD8⁺ and helper CD4⁺ lineage T cells, during which suppression of Cd4 and Thpok genes by Runx-dependent-silencers in those genes is crucial to segregate the two lineages. However, how TCR signals are linked to cytotoxic-lineage-specific Cd4-Thpok silencing remains unclear. Here we show that the terminal Y residue within the evolutionarily conserved C-terminal WRPY motif in Runx1, which is essential for interacting with TLE co-repressor proteins, was phosphorylated more in CD4⁻CD8⁺ thymocytes than in CD4⁺CD8⁻ thymocytes, inducing an interaction with TLE co-repressors for cytotoxic-lineage specific Cd4-Thpok silencing. Non-receptor tyrosine kinases Lck and Zap70 interacted with Runx in the cytoplasm more in MHC-I-signaled CD4⁻CD8⁺ thymocytes than in CD4⁺CD8⁻ thymocytes. Collectively, these findings reveal that differential phosphorylation states at the terminal tyrosine residue in Runx connect MHC restriction with the helper versus cytotoxic T cell lineage choice.

Main

During thymic T cell development, the developmental fate of αβT lymphocytes is controlled by a T cell receptor (TCR) that senses self-antigen presented on major histocompatibility complex (MHC). After a successful positive selection, a process that enriches useful T lymphocytes upon recognizing self-antigen on their MHC¹, CD4⁺CD8⁺ double-positive (DP) thymocyte precursors differentiate into CD4⁺CD8⁻ SP and CD4⁻CD8⁺ SP thymocytes, which are committed to the helper- and cytotoxic-lineage T cells via induction of the lineage-specifying transcription factor Thpok (encoded by Zbtb7b) and Runx3, respectively^2,3,4. The MHC specificity of TCRs expressed on the DP precursors correlates well with the CD4-helper versus CD8-cytotoxic choice. MHC-I selected precursors shut off both Zbtb7b and Cd4 genes via the activity of transcriptional silencers in these loci, whereas these silencers are nonfunctional in MHC-II-signaled thymocytes that become CD4⁺CD8⁻ SP thymocytes^5,6,7. Both Cd4 and Thpok silencers contain binding motifs for Runx transcription factor family proteins, and binding of Runx proteins is essential to activate these silencers^7,8. Runx proteins interact with transducin-like enhancer (TLE) co-repressor family proteins, which have a crucial role in suppressing Cd4 and Zbtb7b genes⁹. The evolutionarily conserved VWRPY penta-peptide motif at the C-terminal end of Runx family proteins and the WD40 motifs in TLE proteins function as a docking platform between Runx and TLEs^10,11. However, it remains elusive how MHC-I-engaged TCR signaling activates the Runx-TLE axis. There is a WRPW motif at the C terminus of basic helix loop helix (bHLH) transcription factor family proteins¹², such as Hes1. The WRPW motif also mediates interaction with TLEs with higher affinity than the WRPY peptide^11,13. Although the WRPW peptide interacts with the WD40 domain with an affinity high enough to be co-crystalized, the WRPY motif was unable to remain associated with the WD40 motif during crystallization¹¹. These observations have raised the possibility that having a tyrosine (Y) instead of tryptophan (W) at the C-terminal end endows Runx proteins with regulatory mechanisms that allow them to function as both activator and repressor in the same cell, that is, Cd4 and Cd8 genes are repressed and activated by Runx3 in CD8⁺ T cells, respectively^8,14. However, underlying molecular mechanisms for such bifunction of Runx proteins and how TCR signals initiated by distinct MHCs is linked with such regulatory mechanisms remain unclear.

Here we provided genetic evidence that replacement of the terminal Y residue within the WRPY motif in Runx proteins with W or phenylalanine (F) converted them to dominant repressors that induced redirection of MHC-II selected thymocytes into cytotoxic CD4⁻CD8⁺ T cells. Furthermore, the terminal Y residue was phosphorylated more in CD4⁻CD8⁺ than in CD4⁺CD8⁻ thymocytes, and induced an interaction with TLE co-repressors specifically in CD4⁻CD8⁺ thymocytes. Runx1 protein interacted with the non-receptor tyrosine kinases, Lck and Zap70, more frequently in MHC-I signaled CD4⁻CD8⁺ thymocytes than in CD4⁺CD8⁻ thymocytes. These findings provide insight into how TCR signals by distinct MHCs are coupled with transcriptional regulation governing the helper versus cytotoxic fate decision.

Results

A WRPW missense mutation transforms Runx proteins into repressors

To examine the impact of the Y to W single-amino-acid change within the WRPY motif on endogenous Runx proteins, hereafter referred to as the WRPW mutation, we generated F0 founder mice harboring the Runx3^WRPW(Rx3^W) allele by CRISPR/Cas9 genome editing on a C57BL/6N genetic background. However, all four F0 founders harboring the Rx3^W allele died before mating age. Given that Runx3-deficient (Rx3^Δ/Δ) mice on a C57BL/6 background die within 2 days after birth but survive longer on an Institute of Cancer Research (ICR) background¹⁵, we performed the same genome editing using ICR zygotes and established a Rx3^W mouse line (Extended Data Fig. 1a). Homozygous Rx3^W/W F2 founders on a C57BL/6 and ICR mixed background showed severe limb ataxia (Supplementary Video), as observed in Rx3^Δ/Δ mice^15,16, and died at around 4 weeks of age. The development of CD8⁺ SP thymocytes was nearly completely abrogated in the Rx3^W/W mice (Fig. 1a), whereas some mature CD8⁺ T cells developed in Rx3^Δ/Δ mice⁸. Although mature CD8⁺ single-positive (SP) thymocytes emerged in heterozygous Rx3^+/W mice, the frequency and absolute numbers of splenic CD8⁺ T cells were markedly reduced (Fig. 1a). Besides the CD8⁺ T cell lineage, development of natural killer (NK) cells, dendritic epidermal T cells and Langerhans cells was severely inhibited by expression of Runx3^WRPW protein in a gene-dosage-dependent manner (Extended Data Fig. 1b, c).

**Fig. 1: Runx3^WRPW inhibits the development of CD8 T cells.**

The maturation and survival of CD8⁺ SP thymocytes require stimulation through the common cytokine receptor γ chain by IL-7 or IL-15 (refs. ^17,18). The promoter region in the Il7ra locus contains a Runx motif¹⁹, and IL7Rα expression in CD4⁺ SP thymocytes is positively regulated by Runx1²⁰. We found that expression of IL7Rα was lost in Rx3^W/W CD8⁺ SP thymocytes (Fig. 1b), which showed a clear G0/G1 cell cycle arrest (Fig. 1c). A tendency for G0/G1 arrest was also observed in Rx3^+/W CD8⁺ SP thymocytes (Fig. 1c). RNA-seq analyses of the transcriptome of Rx3^+/+ and Rx3^+/W CD8⁺ SP thymocytes found several differentially expressed genes (Fig. 1d), and GO term analyses found that down regulation of cell-cycle-associated genes in Rx3^+/W CD8⁺ SP thymocytes (Fig. 1d), including Tnk2 (Fig. 1e). Low expression of the Tnk2 gene was confirmed by RT-qPCR (Fig. 1e). Although the Runx3 binding peak in the Tnk2 locus as assessed by chromatin immunoprecipitation sequencing (ChIP-seq) was comparable between Rx3^+/+ and Rx3^+/W CD8⁺ SP thymocytes, H3K27me3, known as a typical repressive histone post-translational modification, was spread from the Runx3 binding region specifically in Rx3^+/W cells (Fig. 1f). The emergence of such aberrant H3K27me3 peaks was observed at a genome-wide level in Rx3^+/W cells (Fig. 1g), whereas there were no appreciable changes in H3K4me3 peaks, a typical modification for active regions (Fig. 1f and Extended Data Fig. 2a). These observations suggest that Runx3^WRPW functions as a strong repressor and counteracts Runx3-mediated gene activation, and genes that should have been activated by Runx3 are conversely repressed by Runx3^WRPW in a dosage- and time-dependent manner.

Rx3^W/W mice also showed loss of peripheral lymph nodes (Fig. 2a and Extended Data Fig. 2b). In keeping with the loss of secondary lymphoid tissues, IL7Rα⁺α4β7⁺ lymphoid tissue inducer (LTi) cells, which are essential for the generation of secondary lymphoid organs²¹, were hardly detected in the gut of Rx3^W/W 17.5 dpc embryos (Fig. 2a). As lymphoid tissue formation is impaired by homozygous mutation in either the Runx1 or Runx3 gene^22,23, impaired LTi differentiation and lack of Peyer’s patch formation even in the Rx3^+/W mice suggested that the Runx3^WRPW protein might also interfere with Runx1 function, although heterodimerization between Runx proteins has not been recognized, unlike the well-established Runx and Cbfβ heterodimer²⁴. To address the interaction between endogenous Runx1 and Runx3 proteins, we inserted 3xTy1-tag and 1xFlag-Tag sequences into the distal (P1) promoter-derived Runx1 and Runx3 isoform by genome editing, respectively (Extended Data Fig. 2c, d). Using CD8⁺ T cells of Rx1^Ty1/Ty1:Rx3^Flag/Flag mice, TY1-Runx1 protein was co-immunoprecipitated by a Flag M2 antibody (Fig. 2b), indicating that TY1-Runx1 and Flag-Runx3 proteins form protein complexes. In addition, the luminescence-based interaction assay AlphaScreen²⁵ (Extended Data Fig. 2e), which uses an in vitro-synthesized Flag-tagged Runx and biotin-labeled Runx protein, showed clear and robust interaction when Runx1 and Runx3 were mixed (Fig. 2c), indicating a possible direct interaction between Runx1 and Runx3. Compared to the control DHFR protein, addition of the Cbfβ protein reduced AlphaScreen signals between Runx1 and Runx3, particularly in the presence of an oligonucleotide containing a Runx binding motif (Extended Data Fig. 2f), suggesting that the interaction of Runx1 and Runx3 was likely mediated through the Runt domain, which also mediates interaction with Cbfβ²⁶. An alternative possibility is that Runx3^WRPW-TLE complex might exhibit increased stability at occupying genomic regions, thereby sequestering Runx1 from those sites. Regardless of the mechanism, Runx3^WRPW likely interfered with Runx1-mediated gene activation during early LTi cell development. In line with previous reports showing Runx3-dependent development of group 1 and group 3 innate lymphoid cells (ILC1 and ILC3, respectively)²³, the frequency of ILC1 and ILC3 in the CD3⁻CD19⁻IL7Rα⁺ population in the large intestinal lamina propria was decreased, with a relative increase of ILC2, in a Runx3^WRPW-dosage-dependent manner (Fig. 2d). In the bone marrow, the frequency and number of Lin⁻IL7Rα⁺Flt3⁻CD25⁺α4β7^lo ILC2 progenitors (ILC2Ps) were increased in a Runx3^WRPW-dosage-dependent manner (Fig. 2e). In bone marrow chimeric mice that were transplanted with equal numbers of Rx3^+/+ and Rx3^+/W whole bone marrow cells, the increase of ILC2P occurred in a cell-intrinsic manner (Extended Data Fig. 2g), suggesting that Runx3^WRPW might bias ILCP differentiation toward ILC2Ps by an as-yet-uncharacterized mechanism.

**Fig. 2: Runx3^WRPWinhibits the development of innate lymphoid cells.**

Given the strong dominant-negative effect of Runx3^WRPW on the differentiation of several types of Runx3-dependent cells, we next tested whether Runx1^WRPW also severely inhibited the Runx1-dependent developmental program and recapitulated the embryonic lethal phenotype of Rx1^Δ/Δ embryos^27,28. Genotyping by PCR and sequencing of embryos generated by CRISPR/Cas9 genome editing detected seven, five and four embryos that had donor DNA incorporated at 11.5, 12.5 and 13.5 dpc, respectively (Extended Data Fig. 3a). Although all seven embryos harboring a knock-in Rx1^w allele were alive at E11.5 dpc, three out of five and three out of four embryos with the Rx1^w allele died at 12.5 and 13.5 dpc, respectively (Extended Data Fig. 3b). Three E12.5 embryos harboring the Rx1^w allele showed bleeding in the brain (Fig. 3a) and lacked Lin⁻ScaI⁺c-Kit⁺ hematopoietic stem-like cells in the fetal liver (Fig. 3b), as reported in Rx1^Δ/Δ embryos^27,28. Although the genotype of F0 founders generated by CRISPR/Cas9 genome editing was mosaic and thus hard to determine accurately, these observations indicated that heterozygous Rx1^+/W embryos likely phenocopied the Rx1^Δ/Δ embryos and lacked generation of hematopoietic stem cell, presumably due to inhibition of Runx1-dependent gene activation. Overall, the single-amino-acid change of the terminal Y to W on endogenous Runx1 and Runx3 proteins markedly impacted their function and inhibited the development of multiple Runx-dependent cells, presumably by altering Runx protein function into a constitutive and dominant transcriptional repressor.

**Fig. 3: Runx1^WRPW causes embryonic lethality and loss of hematopoietic stem cells.**

Runx^WRPW impairs MHC-II-specific T cell development

During thymocyte differentiation, CD4⁺CD8⁺ DP precursors mainly express Runx1 protein among Runx family proteins²⁰. To assess the impact of the endogenous expression of Runx1^WRPW in CD4⁺CD8⁺ DPs on CD4 helper versus CD8 cytotoxic choice, which could not be analyzed in Rx1^+/W mice due to embryonic lethality, we utilized an inducible transgene expression system at the Rosa26 locus, in which transgenic protein expression from an inserted cDNA was induced upon removal of a loxP-Stop-loxP (lsl) sequence (Extended Data Fig. 4a), to establish R26^lsl-R1, R26^lsl-R1-WRPW, R26^lsl-R3 and R26^lsl-R3-WRPW transgenic strains that were crossed with Cd4^Cre for expression of Flag-Runx1, Flag-Runx1^WRPW, Flag-Runx3, Flag-Runx3^WRPW protein starting at the CD4⁺CD8⁺ DP thymocyte stage (Extended Data Fig. 4b), hereafter referred to as R26^DP-R1, R26^DP-R1-WRPW, R26^DP-R3 and R26^DP-R3-WRPW mice, whereas non-transgenic wild-type (WT) mice were used as control. CD24⁺ thymocytes were divided into CD69⁻TCRβ⁻ pre-selection and CD69⁺TCRβ⁺ post-selection thymocytes and mature thymocytes were defined as CD24⁻TCRβ⁺ cells (Extended Data Fig. 4c). Cells expressing transgenic Runx proteins from the Rosa26 locus were identified by eGFP expression driven by ires-GFP sequences located downstream of Runx1 cDNA (Extended Data Fig. 4a). Although the CD4 and CD8α expression profile was comparable in pre-selection CD69⁻TCRβ⁻ thymocytes among non-transgenic WT, R26^DP-R3 and R26^DP-R3-WRPW mice, the frequency of CD4⁺ cell population was reduced in post-selection GFP⁺CD69⁺TCRβ⁺ thymocytes in R26^DP-R3-WRPW mice compared to non-transgenic WT and R26^DP-R3 mice (Fig. 4a). Mature thymocyte populations (CD4⁺CD8⁻ SP and CD4⁻CD8⁺ SP subsets) were segregated with a similar frequency in non-transgenic control and R26^DP-R3 mice (Fig. 4a), whereas R26^DP-R3-WRPW mice had almost no CD4⁺CD8⁻ SP subset and had an increase in CD4⁻CD8⁻ double-negative (DN) cells (Fig. 4a). Similarly, transgenic expression of Flag-Runx1^WRPW in CD4⁺CD8⁺ DPs resulted in loss of CD4 expression in post-selection GFP⁺CD69⁺TCRβ⁺ thymocytes and loss of the CD4⁺CD8⁻ SP subset in the mature thymocyte population (Fig. 4b). The percentage of CD4⁻CD8⁻ DN cells in the mature CD24⁻TCRβ⁺ thymocyte population was lower, with a relative increase of CD4⁻CD8⁺ cells in R26^DP-R1-WRPW compared to R26^DP-R3-WRPW mice (Fig. 4a, b), suggesting a functional difference between Runx1^WRPW and Rux3^WRPW, presumably due to difference in interacting proteins. Zbtb7b expression was not detected in mature CD4⁻CD8⁻CD24⁻TCRβ⁺ thymocytes in both R26^DP-R1-WRPW and R26^DP-R3-WRPW mice (Fig. 4a, b). The frequency and number of post-selection and mature CD24⁻TCRβ⁺ thymocytes was lower in both R26^DP-R1-WRPW and R26^DP-R3-WRPW mice compared to non-transgenic WT mice (Extended Data Fig. 4c), indicating that Runx^WRPW proteins inhibited the efficiency of positive selection, consistent with the reduction of mature thymocytes in Cd4^CreRunx1^fl/flRunx3^fl/fl mice²⁰. Because CD4⁺CD8⁻ thymocytes in WT mice are MHC-II restricted, we examined the differentiation of MHC-II-restricted cells in b2m^−/−R26^DP-R1-WRPW mice, which are MHC-I deficient due to lack of β2-microglobulin, and found that most MHC-II-restricted cells were redirected into the CD4⁻CD8⁺ cell subset (Fig. 4c). In addition, thymocytes expressing the OT-II transgenic TCR, which recognizes an OVA-derived-peptide on MHC-II and are guided to differentiate into CD4⁺CD8⁻ cells, differentiated into CD4⁻CD8⁺ SP in the R26^DP-R1-WRPW mice (Fig. 4c). Thus, transgenic expression of Runx1^WRPW in CD4⁺CD8⁺ DPs redirected MHC-II-restricted cells into the CD4⁻CD8⁺ lineage and inhibited CD4 expression after positive selection, but not in CD4⁺CD8⁺ DP cells.

**Fig. 4: Runx^WRPW redirects MHC-II-restricted thymocytes to the CD8⁺ T cell lineage.**

To examine whether the silencers in the Cd4 and Zbtb7b loci were involved in the redirection of MHC-II restricted cells in R26^DP-R1-WRPW mice, we generated Cd4^ΔS/+R26^DP-R1-WRPW and Zbtb7b^ΔS/+R26^DP-R1-WRPW mice, which lacked the Cd4 silencer and Thpok silencer in the Cd4 and Zbtb7b locus, respectively. Lack of the Thpok silencer in the Zbtb7b locus induces expression of Thpok even in MHC-I selected cells, guiding them into the CD4⁺CD8⁻ lineage⁷. Differentiation of CD4⁺CD8⁻ SP thymocytes was recovered approximately to half by the loss of the Thpok silencer in Zbtb7b^ΔS/+R26^DP-R1-WRPW compared to R26^DP-R1-WRPW mice (Fig. 4d), whereas removal of the Cd4 silencer from the Cd4 locus in Cd4^ΔS/+R26^DP-R1-WRPW mice failed to restore the differentiation of CD4⁺CD8⁻ SPs, although CD8⁺ cells de-repressed CD4 (Extended Data Fig. 4d). The activity of these silencers requires TLE proteins⁹. Development of CD4⁺CD8⁻ SP thymocytes was partially restored by the lack of TLE proteins in R26^DP-R1-WRPW Tle1/3/4^TKO mice (Fig. 4d), whereas loss of TLEs resulted in a nearly complete lack of CD4⁻CD8⁺ SPs in Tle1/3/4^TKO mice through redirection of MHC-I selected cells into CD4⁺CD8⁻ SP cells⁹. These observations indicated that Runx1^WRPW likely altered the lineage specificity of those silencers. Although the Cd4 silencer and Thpok silencer are non-functional in MHC-II restricted cells in WT mice, they were activated in MHC-II signaled post-selection thymocytes by Runx1^WRPW in a TLE-dependent manner (Fig. 4d). When expression of Runx1^WRPW was induced in MHC-II-restricted post-selection thymocytes through the generation of Thpok-^CreR26^4SP-R1-WRPW mice²⁹, CD4 expression was slightly lower in CD69⁺TCRβ⁺ post-selected thymocytes, and the frequency of CD4⁺CD8⁻ SP thymocytes declined in the CD24⁻TCRβ⁺ mature thymocyte population compared to R26^DP-R1 mice (Fig. 4b), indicating that Runx1^WRPW still exhibited a dominant function at later stages of thymocyte differentiation in MHC-II restricted cells.

To addressed if changes of the Y residue to other amino acids, such as F or glutamic acid (E), influenced Runx function, we generated R26^DP-R1-WRPE and R26^DP-R1-WRPF mice by CRISPR/Cas9 genome editing onto the R26^DP-R1 allele and gene targeting, respectively (Extended Data Fig. 4e). There was no decrease of CD4⁺ cells in CD69⁺TCRβ⁺ post-selection and CD24⁻TCRβ⁺ mature thymocyte populations, whereas CD24⁻TCRβ⁺mature CD8⁺ thymocytes de-repressed CD4 in the R26^DP-R1-WRPE mice (Fig. 5a). On the contrary, the frequency of CD4⁺CD8⁻ cells was decreased, whereas that of CD4⁻CD8⁻ cells was increased in CD69⁺TCRβ⁺ post-selection and CD24⁻TCRβ⁺ mature thymocytes, together with a reduction in the frequency and numbers of CD24⁻TCRβ⁺ mature thymocytes in R26^DP-R1-WRPF mice compared to R26^DP-R1mice (Fig. 5a), although the loss of CD4⁺CD8⁻ cells was milder in the R26^DP-R1-WRPF mice compared to R26^DP-R1-WRPW mice (Fig. 5a). To test the effect of Y to E or Y to F amino acid changes on endogenous Runx3, we generated Rx3^E and Rx3^F alleles by CRISPR/Cas9 genome editing (Extended Data Fig. 4f). CD8⁺ T lineage cells in Rx3^E/E mice, which expressed the Runx3^WRPE protein, de-repressed CD4 (Fig. 5b), as reported in Runx3^ΔWRPY mice³⁰. Although TLE3 was co-immunoprecipitated by a Runx3 antibody in spleen Rx3^+/+ CD8⁺ T cells, TLE3 association with Runx3^WRPE protein was hardly detected in spleen Rx3^E/E CD8⁺ T cells (Fig. 5c). Rx3^F/F mice survived longer and had a mature CD4⁻CD8⁺ SP thymocyte population compared to Rx3^W/W mice, whereas the CD8⁺ T cells were almost absent in spleen (Fig. 5b) and formation of peripheral lymph nodes was impaired in a gene-dosage-dependent manner, although the phenotype was milder than that caused by the Rx3^W mutation (Fig. 5d). Thus, Y was needed in the terminal residue in Runx proteins for their appropriate function.

**Fig. 5: Mutation of the Runx terminal Y residue influences its function.**

Other tyrosine residues (Y375, Y378, Y379 and Y386) are present within the inhibitory domain of mouse Runx1 protein, and they were shown to be phosphorylated in L8057 megakaryoblastic cells³¹. To investigate whether F substitutions at these Y sites affected T cell development, we generated a mutant Runx1 allele, designated Runx1^Y3F, in which Y375, Y378 and Y379 were substituted with F residues (Extended Data Fig. 5a). In contrast to the severe reduction of CD4 SP thymocytes observed in R26^DP-R1-WRPF mice expressing the transgenic Runx1^WRPF form, Runx1^Y3F/Y3F homozygous mice did not show a statistically significant decrease in CD4SP thymocyte number (Extended Data Fig. 5b). Thus, the substitution of the terminal Y residue had a distinct and pronounced effect on CD4⁺ versus CD8⁺ T cell lineage commitment.

Phosphorylation of Runx1 regulates association with TLE3

Next, we examined whether post-translational modifications occurred at the terminal Y residue of Runx proteins. The PhosphoSitePlus database (https://www.phosphosite.org/homeAction) shows phosphorylation on the terminal Y residue on human RUNX1 in the Jurkat T cell line. In Jurkat cells transduced with a retroviral vector encoding Ty1-tagged murine Runx1 or Runx1^WRPF, and treated or not with the tyrosine phosphatase inhibitor pervanadate, immunoprecipitation with a Ty1 antibody and mass spectrometry after trypsin digestion detected a phosphorylated LEEAVWRPpY peptide in pervanadate-treated Jurkat cells (Fig. 6a). Using J.CaM1.6 cells, a Jurkat variant cell line lacking Lck activity³², we tested whether Runx1-Lck interactions were required for the phosphorylation of the terminal Y residue in Runx1. LEEAVWRPpY peptide was hardly detected in J.CaM1.6 cells transduced with Ty1-Runx1, even after pervanadate treatment, compared to Jurkat cells (Fig. 6a), suggested that Lck was involved in the phosphorylation of Runx1 Y terminal residue. In primary CD4 SP or CD8 SP thymocytes isolated from Rx1^Ty1/Ty1 mice expressing OT-II or OT-I transgenic TCRs, respectively, the abundance of the LEEAVWRPpY peptide was higher in CD8 SP thymocytes than in CD4 SP thymocytes, whereas the non-phosphorylated LEEAVWRPY peptide was more abundant in CD4 SP thymocytes (Fig. 6b). Consequently, the ratio of LEEAVWRPpY to LEEAVWRPY was approximately 30-fold higher in CD8SP than in CD4 SP thymocytes (Fig. 6b). Immunoprecipitation and mass spectrometry analysis indicated that the interaction of Ty1-Runx1 with TLE3 was enhanced in pervanadate-treated compared to non-treated Jurkat cells transduced with Ty1-Runx1(Fig. 6c), suggesting that phosphorylation was involved in the regulation of the Runx1-TLE3 interaction. In cell lysate from Jurkat cells expressing Ty1-Runx1 or Ty1-Runx1^WRPF, Ty1-Runx1-TLE3 interaction was abrogated, whereas the Ty1-Runx1^WRPF-TLE3 interaction was retained by treatment with alkaline phosphatase (AP) before co-immunoprecipitation, and the TCF1-TLE3 interaction, which is mediated through the Q domain in TLE proteins³³, was increased by AP treatment (Fig. 6d), suggesting the Y to F replacement bypassed the phosphorylation-mediated interaction between Runx1 and TLE3. The increased interaction between Runx1^WRPF-TLE3 upon pervanadate treatment suggested that additional phosphorylation events, at other tyrosine or serine/threonine residues in either or both Runx1 and TLE3 may also contribute to the Runx1-TLE3 interaction. More endogenous Ty1-Runx1 was co-immunoprecipitated by TLE3 antibody in CD8 SP than in CD4 SP thymocytes from Rx1^Ty1/Ty1 mice (Fig. 6e), although the amount of Ty1-Runx1 protein was higher in CD4 SP thymocytes than CD8 SP thymocytes (Fig. 6e). In line with these findings and previous work⁹, ChIP-qPCR detected TLE3 binding to the Cd4 silencer in CD8 SP, but not in CD4 SP thymocytes from WT mice (Fig. 6f). Runx-Cbfβ binding to the Cd4 silencer, which was not detected in spleen CD4⁺ T cells⁹, was detected in CD4 SP thymocytes, albeit at lower level compared to that in CD8 SP thymocytes from WT mice (Extended Data Fig. 6a). To investigate the effects of phosphorylation on the terminal Y residue of Runx1, we conducted in silico modeling to simulate the conformational changes, leveraging structural data on the interaction between the WD40 domain in humanTLE3 (hTLE3) and the WRPW peptide¹¹. Molecular simulations revealed an increase in TLE3 association with WRPF peptide (Fig. 6g, Extended Data Fig. 6b–e and Extended Data Table 1a, b). Although F was initially expected to mimic unphosphorylated Y, binding free-energy calculations showed that WRPF bound mTLE3 in a manner comparable to phosphorylated WRPpY (Extended Data Table 1). Principal-component analysis (PCA) and free energy landscape (FEL) analyses further suggested WRPF maintains a stable conformation, like WRPpY (Extended Data Fig. 6d, e). Although structural simulations did not reveal major perturbations in the overall conformation of the mTLE3-mRunx1 complex, they did show that Y phosphorylation facilitated key stabilizing interactions, including van der Waals, electrostatic and hydrophobic forces (Extended Data Fig. 6b–e, Extended Data Table 1). Collectively, these observations indicated that a distinct phosphorylation status at the terminal Y residue of Runx1 between CD4 SP and CD8 SP thymocytes regulated recruitment of TLE to the Cd4 silencer.

**Fig. 6: Phosphorylation of Runx1 terminal Y residue regulates interaction with TLE3.**

To address what kinase(s) phosphorylated the terminal Y residue in Runx1 protein, we used the AlphaScreen system with a customized library covering 497 possible kinases to search for Runx interacting proteins in vitro. Analyses focusing on tyrosine kinases listed the top 30 candidates and ranked Fyn kinase fourth among 497 kinases (Extended Data Fig. 7). Src family kinases have been reported to phosphorylate Y on Runx3 in HEK293 cells³⁴. Mass spectrometry analysis of Ty1-Runx1-interacting proteins in Jurkat cells showed that LCK, FYN and ZAP70 co-immunoprecipitated with Ty1-Runx1 and Ty1-Runx1^WRPF (Fig. 6h). Only the Runx1-ZAP70 interaction was enhanced by pervanadate treatment (Fig. 6h), suggesting that the Runx1-Zap70 may be regulated by phosphorylation at other tyrosine or serine/threonine residues. Thus, non-receptor tyrosine kinases, such as Lck, are associated with Runx1 protein.

Runx1 interacts with Lck and Zap70 more in CD8 SP thymocytes

Lck and Runx1 are located preferentially in the cytoplasm and nucleus, respectively. To assess where in the cell they encountered each other, we used in situ proximity ligation assay (PLA), which can visualize interactions between two molecules in a proximity as well as the intracellular location of such interaction³⁵. PLA staining with WRPY³⁶ and Ty1 antibodies in Jurkat and J.CaM1.6 cells transduced with Ty1-Runx1 detected Ty1-Runx1 intramolecular interactions mainly in the nucleus of both Jurkat and J.Cam1.6 cells (Fig. 7a), whereas staining of Jurkat cells, but not J.CaM1.6 cells, with Ty1 and LCK antibodies visualized interactions between Ty1-Runx1 and endogenous LCK in the cytoplasm and beneath the plasma membrane (Fig. 7a). Although Ty1-Runx1 intramolecular staining was equivalently detected in CD4⁺CD8⁺ DP, CD4 SP and CD8 SP thymocytes isolated from Rx1^Ty1/Ty1 mice, Runx1-Lck interactions were statistically more abundant in CD8 SP thymocytes compared to DP and CD4 SP thymocytes (Fig. 7b). Runx1-Lck PLA signals were also detected more abundant in spleen CD8⁺ T cells than in spleen CD4⁺ T cells from Rx1^Ty1/Ty1 mice (Extended Data Fig. 8a). High resolution microscopy and measurement of the distance of PLA signals from the center of nucleus indicated that Runx1-Lck PLA signals were dominantly developed in the cytoplasmic region (Fig. 7c). PLA staining with Ty1 and Cbfβ antibodies visualized Runx1-Cbfβ interactions mainly in the nucleus, whereas the Cbfβ-Lck interaction was detected in the cytoplasmic region (Extended Data Fig. 8b). PLA also detected Runx1-Bcl11b interaction in the nucleus (Extended Data Fig. 8c), which was in line with co-immunoprecipitation of Runx1 by Bcl11b antibody³⁷, whereas Bcl11b-Lck interactions were not detected by PLA (Extended Data Fig. 8c). Runx1-Zap70 interactions were visualized by PLA in the cytoplasmic region of primary total thymocytes (Fig. 7d). PLA staining of CD8 SP thymocytes from Rx3^Flag/Flag mice with Flag and Lck antibodies detected Runx3-Lck interaction in the cytoplasm (Fig. 7d and Extended Data Fig. 8d), indicating that Runx proteins were associated with Lck and Zap70 in the cytoplasm, and Runx1-Lck interactions were formed more in CD8 SPs than in CD4 SPs. To address whether TCR signals engaged by MHC-I or differentiation of CD8 SP thymocytes were important for more abundant Runx1-Lck interactions, we used Zbtb7b^GFP/GFP mice, in which MHC-II restricted thymocytes are marked by GFP expression as well as redirected to CD8⁺ T cell lineage due to lack of Thpok protein³⁸. We used PLA to detect Runx1-Lck interactions in mature CD8 SP thymocytes from Rx1^Ty1/Ty1Zbtb7b^GFP/GFP mice sorted into MHC-II-signaled GFP⁺ and MHC-I-signaled GFP⁻ cells. More PLA signals were detected in MHC-I-signaled GFP⁻ cells than in MHC-II-signaled GFP⁺ cells or DP thymocytes (Fig. 7e). Although these results did not formally exclude the possibility that differences in Lck-Runx1 association were a consequence of CD4⁺ T cell and CD8⁺ T cell differentiation, they supported the notion that increased Lck-Runx1 association was correlated with, or regulated by, MHC-I-engaged TCR signaling, rather than being merely a consequence of CD8 SP differentiation.

**Fig. 7: CD8 SP thymocytes dominate the Runx1-Lck interaction in the cytoplasmic region.**

Discussion

Here we provide genetic evidence showing that Y is needed as the terminal amino acid residue in mouse Runx1 and Runx3 proteins and show that the distinct phosphorylation status at this terminal Y residue served as a crucial regulatory mechanism to dissect the helper- versus cytotoxic T cell lineage through phosphorylation-dependent association of Runx proteins with the TLE family of co-repressor proteins. This finding provides insights into the role of post-translational modifications into the function of the transcription factor Runx but also unravels a mechanism that links MHC-I-engaged TCR signals with cytotoxic CD8⁺ T cell-lineage-specific activation of transcriptional silencers in Cd4 and Zbtb7b loci.

An important next question to address is how the phosphorylation at the terminal Y residue of Runx is regulated. Besides the interaction of Runx1 with the non-receptor tyrosine kinases Lck and Zap70 in the cytoplasmic region, Lck activity was involved in phosphorylation of Runx1 in Jurkat cells. Because these observations do not necessarily mean that Lck or Zap70 directly phosphorylated the terminal Y residue in Runx proteins, the identification of the kinase responsible for this modification is needed.

Despite many past studies attempting to identify the differential signaling cues received by MHC-I-selected versus MHC-II-selected thymocytes, there is no clear evidence showing qualitative or quantitative differences in TCR signals initiated by either MHC-I or MHC-II engagement. The kinetic signaling model, which is based on temporal down regulation of CD8, regardless of MHC specificity of TCR on CD4⁺CD8⁺ DP precursors, proposes that the duration of TCR signaling is a key regulator for helper CD4⁺ versus cytotoxic CD8⁺ T cell lineage choice³. Thus, MHC-I selected precursors that receive shorter TCR signals due to CD8 downregulation are guided to become CD8 SP thymocytes. We found that phosphorylation at the terminal Y residue of Runx1 was more abundant in CD8 SP thymocytes than in CD4 SP thymocytes. We also observed that MHC-I-specific TCR signals, rather than CD8 SP thymocyte-related characteristics, enhanced Runx1-Lck interactions. For now, it remains unclear whether shorter TCR signals are involved in inducing Runx1 phosphorylation in CD8 SP thymocyte or whether there are uncharacterized qualitative or quantitative differences in TCR signaling by MHC-I or MHC-II engagement to control phosphorylation of Runx1 or Runx1-Lck interaction. Unraveling the mechanisms that create a distinct phosphorylation status of Runx protein and/or stabilize Runx1-Lck interaction is essential to understanding how MHC-I- and MHC-II-mediated TCR signals are translated into distinct transcriptional programs governing thymocyte fate. To address this point, it would be informative to acquire the Lck or Zap70-interactomes³⁹ between CD4 SP and CD8 SP thymocytes.

Replacement of the terminal Y residue in the Runx protein with F likely increased the binding affinity to TLE proteins, bypassing the phosphorylation-mediated Runx-TLE association. Such constitutive association of Runx with TLE disrupted the lineage specificity of Runx-dependent silencers and redirected MHC-II-restricted cells to cytotoxic CD8⁺ T cell lineages. This indicates that the ratio of phosphorylated Runx bound by TLE to non-phosphorylated Runx unbound by TLE must be tightly regulated. Indeed, our mass spectrometry assay estimated that a very limited fraction (<1%) of Runx1 protein was phosphorylated at the terminal Y residue, suggesting that there are uncharacterized mechanisms that efficiently and selectively recruit TLE-bound Runx protein to silencers and not enhancers. It is important to understand how repressive Runx complexes, including TLE, are guided to appropriate genome regions. Y to F replacement has been used in many studies to mimic phosphorylation-dead residues to address the physiological meaning of Y phosphorylation. In the case of the terminal Y residue in Runx proteins and F or W residue might be structurally closer to a phosphorylated Y and might functionally induce TLE binding. Structural analyses of the WD domain with the LEEAVWRPpY peptide will provide detailed insight into this point.

Evolutionarily, the WRPY motif was likely to have been acquired at some point toward Eumetazoa⁴⁰. The Caenorhabditis elegans Runx protein, CeRUN, has an WRPF sequence⁴¹, whereas the fruit fly (Drosophila melanogaster) DmRunt has the WRPY sequence. In C. elegans, the Groucho/TLE-like co-repressor UNC-37 specifies the choice in motor neuron circuit through its repressive activity⁴². It remains elusive whether CeRUN and UNC-37 are co-expressed in cells such as motor neurons, and if so, if they stably associate with each other. Our results showed that phosphorylation at a single Y reside endowed diversity in Runx function, enabling Runx proteins to function as activator and repressor within a single cell. This finding sheds light on the role of Runx phosphorylation in linking environmental cues sensed by the plasma membrane TCR receptor with transcriptional regulation of a developmental program in the nucleus. It would be interesting to address how this regulation was acquired during evolution and whether this Runx phosphorylation axis is utilized to control developmental programs of a variety of cell types other than T lymphocytes in metazoan species.

Methods

Mice

Runx1^+/WRPW embryos were generated by incorporation single-strand donor DNA into the Runx1 locus by the CRISPR/Cas9 genome editing technology. Zygotes generated from C57BL/6NJcl strain purchased from CLEA Japan were injected with mRNA encoding humanized S. pyogenes Cas9 that was invitro transcribed from pX330 plasmid (#42230, Addgene), single guide RNA (sgRNA) and single strand donor DNA, both of which were synthesized at Integrated DNA Technology. Similarly, Runx3^WRPW, Runx3^WRPF, Runx3^WRPE mouse strains were generated by the CRISPR/Cas9 genome editing technology with appropriate mRNA for Cas9, sgRNA and donor DNA. Zygotes generated from C57BL/6NJcl for Runx3^WRPW, Runx3^WRPF and Runx3^WRPE strains and ICR for Runx3^WRPW strain purchased from CLEA Japan were injected with Cas9 mRNA, single guide RNA (sgRNA) and single-strand donor DNA. All these Runx3 mutant strains were established by crossing F0 founders with C57BL/6NJcl strain. In order to construct the target vector for R26^lsl-R1, R26^lsl-R1-WRPW, R26^lsl-R1-WRPF, R26^lsl-R3 and R26^lsl-R3-WRPW mice, cDNA encoding WT and mutant Runx1 and Runx3 proteins were amplified by PCR to add AscI sites at the both ends, and these DNA fragments was ligated into an AscI-cleaved pCTV vector (#15912, Addgene). 30 μg of the target vector was linearized by AsiSI enzyme and transfected into the M1 ES cell line by electroporation as previously described⁵. After G418 selection, G418 resistant ES clones were screened for homologous recombination event by PCR. Appropriate ES clones were aggregated with blastocysts to generate chimera mice through which these mutant Rosa26 alleles were germline transmitted to the offspring, establishing the Rosa26 mouse lines. Runx1^TY1 mice were generated by insertion of three copies of the TY1 (EVHTNQDPLD) tag followed by the hinge (GGG) region at the N-terminal end of P1-Runx1 protein by the CRISPR/Cas9 genome editing technology. Runx3^Flag mice were generated by insertion of one copy of the Flag (DYKDDDDKLD) tag followed by the hinge (GG) region at the N-terminal end of P1-Runx3 protein by the CRISPR/Cas9 genome editing technology. R26^DP-R3-WRPE allele was generated by CRISPR /Cas9 genome editing onto the R26^lsl-R3 allele using in vitro fertilized zygotes between R26^{lsl-R3/lsl-R3} sperm and C57B6/N eggs. Runx1^Y3F allele was also generated by CRISPR/Cas9-mediated genome editing. Sequences for sgRNA and donor DNA used in genome editing are listed in Extended Data Table 2. Cd4-Cre⁴³, OT-I⁴⁴, OT-II⁴⁵, Thpok-Cre²⁹, Zbtb7b^ΔS ⁷, Zbtb7b^GFP ³⁸ and TLE1/3/4^TKO ⁹ mice have been previously described. Β2m-deficient mice (stock 002070) were from Jackson Laboratory. All mouse strains were bred and maintained in the animal facility at the RIKEN IMS. All animal procedures were in accordance with protocol AEY2022-019(2) approved by the institutional Animal Care and Use Committee of RIKEN Yokohama Branch. All the mice were euthanized with CO₂ overdose following anesthesia using isoflurane. Data from both genders were combined for analysis, and unless otherwise specified, mice aged 4 to 14 weeks were analyzed.

Cells

Jurkat cells (RCB0806) and J.Cam1.6 cells were a gift from T. Saito at RIKEN IMS. Both Jurkat and J.Cam1.6 cells were maintained in RPMI-1640 medium supplemented with 10% FBS and antibiotics. pMYs retroviral vectors encoding TY1-tagged murine Runx1, Runx1WRPF or MigR1 retroviral vector encoding 3xFlag-TCF1 were transfected into GP2-293 cells (Clontech) with pVSV-G using FuGENE HD Transfection Reagent (Promega) according to the manufacturer’s protocol. Two days after transfection, supernatant was collected, passed through a 0.45 μm syringe filter and supplemented with 8 μg ml⁻¹ polybrene (Sigma-Aldrich). Jurkat cells were suspended in the virus-containing medium at a density of 2.5 × 10⁵ cells ml⁻¹, and spin-infection was performed at 1,800 x g for 90 min at 32 °C.

Tissue processing

Thymus, spleen, lymph nodes and other organs were harvested from mice at 4 to 14 weeks of age and then mashed and passed through 100 μm pore cell strainer to make single-cell suspensions. After the hemolysis with ACK Lysing Buffer (Thermo Fisher Scientific), cells were washed with ice-cold staining buffer (D-PBS (-), 2 mM EDTA, 0.05% NaN₃ and 2% FBS). Lamina propria lymphocytes from the small intestine were isolated as previously described²³. Small intestine was cut into 5 mm pieces and incubated in 20 ml RPMI containing 2% FBS and 5 mM EDTA, with shaking at 200 rpm and 37 °C for 20 min. After incubation, tissue was washed with PBS twice by vortexing for 20 s to remove epithelial cells. The remainder of tissue was filtered and digested with RPMI with 2% FBS, 0.5 mg ml⁻¹ collagenase IV (Sigma-Aldrich, C-5138) and 50 μg ml⁻¹ DNase (043-26773; FUJIFILM) at 200 rpm and 37 °C for 30 min. Digested tissue was passed through a 100 µm strainer and subjected to Percoll gradient centrifugation at 1,800 rpm for 20 min using 40% and 80% Percoll in RPMI with 2% FBS. Cells located between the 40% and 80% Percoll layers were collected as LPLs.

Flow cytometry

Surface molecules were stained with specific antibodies by incubating for 30 min on ice. Following antibodies for surface molecules were purchased from BD Biosciences, BioLegend or Thermo Fisher Scientific; CD3ε (clone: 145-2C11), CD4 (clone: RM4-5), CD8α (clone: 53-6.7), CD19 (clone:6D5), CD24 (clone: M1/69), CD25 (clone: PC61.5), CD45 (clone: 30-F11), CD45.1 (clone: A20), CD45.2 (clone: 104), CD69 (clone: H1.2F3), CD117 (clone: 2B8), CD127 (clone: A7R34), CD135 (clone: A2F10), CD161 (clone: PK136), EpCAM (clone: G8.8), Ly-6G (clone: RB6-8C5), Integrin α4β7 (clone: DATK32), MHC-II (clone: M5/114.15.2), NKp46 (clone: 29A1.4), ScaI (clone: E13-161.7) TCRβ (clone: H57-597), and γδTCR (clone: GL3) and Vα2TCR (B20.1). In order to investigate intracellular molecules, cells were fixed and permeabilized by use of Transcription Factor Buffer Set (BD Biosciences) following surface staining. Following antibodies for intracellular molecules were purchased from BD Biosciences or Thermo Fisher Scientific; Zbtb7b (clone: 2POK), Runx3 (clone: R3-5G4), Rorγt (clone: B2D), PLZF (clone: 9E12) and Gata3 (clone: TWAJ). Dead cells were distinguished by use of 7-AAD (BD Biosciences) and LIVE/DEAD Fixable Dead Cell Stain (Thermo Fisher Scientific) for general and intracellular staining, respectively. Multi-color flow cytometric analysis was performed using BD FACSCanto II (BD Biosciences), and data was processed with FlowJo software (BD Biosciences). Cell subsets were sorted using a BD FACSAria III (BD Biosciences).

Cell cycle analyses

After the cell surface staining of freshly prepared whole thymocytes, cells were fixed with ice-cold 70% ethanol for 30 min on ice. After the meticulous cell wash using PBS, RNA was removed by the treatment of 0.1% Triton X-100 PBS containing 100 μg ml⁻¹ Ribonuclease A (Thermo Fischer) for 30 min at room temperature. Intracellular DNA was stained using PBS (2% FCS) containing SYTOX Green (Thermo Fischer), and DNA content distribution was assessed by flow cytometry.

RNA sequencing

RNA was extracted from sorted CD8⁺ mature thymocytes of Wt and Rx^+/W mice using the RNeasy Mini kit (Qiagen). Sequencing libraries were prepared using a NEBNext RNA Library Prep Kit for Illumina (NEB) according to the manufacturer’s protocol. Single-end 50 bp reads were obtained by Illumina HiSeq 2500. The reads were mapped to mm10 and analyzed with HISAT2 (v2.2.1) and edgeR (v3.36.0), respectively.

Bone marrow chimera

Bone marrow cells were collected from tibia and femur by flashing with RPMI. To generate bone marrow chimeras, recipient CD45.1⁺ ICR mice were irradiated at 950 rad. Donor cells were obtained from CD45.1⁺ ICR mice and CD45.1⁺/CD45.2⁺ F1 littermate mice, which were generated by crossing CD45.2⁺ C57BL/6 mice with CD45.1⁺ Rx3^+/W mice in ICR background. The irradiated CD45.1⁺ ICR mice were reconstituted with a 50:50 mixture of CD45.1⁺ Rx3^+/+ bone marrow cells and CD45.1⁺/CD45.2⁺ either Rx3^+/+ or Rx3^+/W bone marrow cells. Eight weeks after transplantation, chimeric mice were analyzed by flowcytometry to assess chimerism.

Cell sorting and enrichment

CD4 SP and CD8 SP thymocytes were enriched using EasySep Cell Separation with EasySep Mouse Streptavidin RapidSpheres Isolation Kit (STEMCELL Technologies). Thymi were harvested from C57BL/6NJcl mice at 3 to 5 weeks of age and mashed on a 100 μm pore cell strainer to obtain single-cell suspension. After the hemolysis with ACK Lysing Buffer (Thermo Fisher Scientific), cells were washed twice with D-PBS (-), and 10⁸ cells were resuspended in 1 ml EasySep buffer (D-PBS (-), 2% FBS, 1 mM EDTA) at room temperature. The cell suspension was mixed with 50 μl normal rat serum and 10 μl biotin-conjugated anti-CD24 antibody (clone M1/69, BD Biosciences) with either 10 μl biotin-conjugated anti-CD4 (clone RM4-5, BioLegend) or CD8a (clone 53-6.7, BioLegend) antibody and incubated for 10 min at room temperature. The cell suspension was then mixed with 50 μl EasySep Mouse Streptavidin RapidSpheres and incubated for 5 min. After the reaction with RapidSpheres, the mixture was placed on the EasySep Magnet for 2.5 min, and the unlabeled cells were harvested. The purity the CD4 SP or CD8 SP thymocytes obtained was over 85%.

RNA isolation and quantitative PCR

DNaseI-treated total RNA was prepared from sorted CD8⁺ T cells using RNeasy mini kit (QIAGEN), and cDNA was synthesized by SuperScript IV reverse transcriptase (Thermo Fisher Scientific). Quantitative RT-PCR was performed using QuantStudio 3 Real-Time PCR system (Applied Biosystems) with Universal ProbeLibrary (Roche). The following primer sets and probes were used for Tnk2 and Hprt mRNA quantification: Tnk2-F, 5′-GGCCCTGCTCATCACAAA-3′, Tnk2-R, 5′-CCGTGATAGCTGTGCTCTGA-3′ and UPL #108. Hprt-F, 5′-TCCTCCTCAGACCGCTTTT-3′, Hprt-R, 5′-CCTGGTTCATCATCGCTAATC-3′ and UPL #95.

ChIP-seq and ChIP-qPCR

For ChIP-seq, MACS-enriched and FACS-sorted 5 × 10⁶ Splenic CD8⁺ T cells were washed once with PBS supplemented with 2% FCS and cross-linked by incubation in a 1% formaldehyde solution for 10 min with gentle rotation at 22 to 26 °C. The reaction was quenched in 0.15 M glycine. Cells were then washed with ice-cold PBS containing 2% FCS for 10 min with gentle rotation at 4 °C and were lysed in Lysis Buffer 1 (50 mM HEPES pH 7.5, 140 mM NaCl, 1 mM EDTA, 10% Glycerol, 0.5% NP-40, 0.25% Triton X-100) supplemented with cOmplete Protease Inhibitor Cocktail (Roche) for 10 min at 4 °C with gentle rotation. Nuclei were pelleted and were washed by Lysis Buffer 2 (10 mM Tris-HCl pH 8.0, 200 mM NaCl, 1 mM EDTA and 0.5 mM EGTA) supplemented with protease inhibitor cocktail. Pelleted nuclei were resuspended in 300 μl Lysis Buffer 3 (10 mM Tris-HCl pH 8.0, 100 mM NaCl, 1 mM EDTA, 0.5 mM EGTA, 0.1% sodium deoxycholate and 0.5% N-laurylsarcosine sodium salt) supplemented with protease inhibitor cocktail and were sonicated 10 times using a model XL2000 ultrasonic cell disruptor (MICROSON) at output level 6 for 15 s. After removing debris by centrifugation, 30 μl 10% Triton X-100 was added to 270 μl supernatant, and sonicated chromatin was incubated overnight at 4 °C with anti-Histone H3K27me3 (clone C36B11, Cell Signaling Technology), anti-Histone H3K4me3 (clone C42D8, Cell Signaling Technology), or anti-RUNX3 (clone D6E2, Cell Signaling Technology) monoclonal antibodies, that were pre-conjugated with 50 μl Dynabeads M-280 Sheep anti-Rabbit IgG (Thermo Fisher Scientific). Magnetically collected beads were washed with ChIP-RIPA (50 mM HEPES (pH 7.6), 500 mM LiCl, 1 mM EDTA, 1% NP-40, 0.7% sodium deoxycholate) and TE buffer supplemented with 50 mM NaCl. Immunoprecipitates were eluted from beads into 100 μl of elution buffer (50 mM Tris-HCl pH 8.0, 10 mM EDTA, 1% SDS) by incubation for 15 min at 65 °C with vigorous shaking. Eluted immunoprecipitates were then incubated at 65 °C overnight for reverse-crosslinking. ChIP DNA samples were treated with RNase A (Thermo Fisher Scientific) at 37 °C for 1 h, followed by incubation with Proteinase K (Thermo Fisher Scientific) at 55 °C in the presence of 6 mM CaCl2 for 1 h and purified by phenol/chloroform extraction and ethanol precipitation. Purified DNA samples were subjected to re-sonication with a Covaris S220 to produce DNA fragments with an average size of 200 bp, and were used for library construction with NEBNext ChIP-seq Library Prep Master Mix set for Illumina Kit (NEB). Sequencing was performed by the RIKEN IMS sequence facility with Illumina HiSeq 1500. Sequencing data was analyzed on Galaxy platform (https://usegalaxy.org/). Adaptor-trimmed fastq files were mapped onto reference mouse genome (mm10) using ‘Bowtie2’ (v.2.5.3) with default setting. Generated bam datasets were processed with ‘Filter’ (v.2.5.2) and ‘MarkDuplicates’ (v.3.1.1.0) to remove reads with low mapping quality (<30), reads mapped on mitochondrial DNA, and PCR duplicates. Peaks were identified by use of ‘MACS2 callpeak’ (v.2.2.9.1). Bigwig files were generated from BedGraph using ‘Convert BedGraph to BigWig’ (v.1.0.1). ‘computeMatrix’ (v.3.5.4) and ‘plotHeatmap’ (v.3.5.4) were used to make heatmaps, and bigwig tracks were generated using ‘pyGenomeTraks’ (v.3.8).

For ChIP-qPCR, an equal number of CD4 SP and CD8 SP thymocytes were washed with D-PBS (-) twice, and the crosslink was performed by incubating in 50 mg ml⁻¹ disuccinimidyl glutarate (Thermo Fisher Scientific) for 30 min at room temperature with gentle rotating followed by reaction with 1% paraformaldehyde (Sigma-Aldrich) for 10 min. The crosslink was quenched by adding 0.125 M glycine, and cells were rinsed with ice-cold D-PBS (-) containing 2% FBS. Cells were pelleted by centrifugation, and stored at −80 °C until the day of use. Pellet was defrosted on ice, suspended to Nuclear Isolation Buffer (50 mM Tris-HCl [pH 7.4], 60 mM KCl, 0.5% NP-40) supplemented with cOmplete, Mini, EDTA-free Protease Inhibitor Cocktail (PI) (Sigma-Aldrich) and incubated for 10 min at 4 °C with gentle rotation to isolate nuclei. Nuclei were resuspended in 100 μl PI-supplied Lysis Buffer (50 mM Tris-HCl (pH 7.4), 10 mM EDTA, 0.5 mM EGTA and 0.5% SDS), and sonicated using Picoruptor 2 sonication device (Diagnode) with 16 cycles of output at high for 30 s followed by 30 s rest. Cell debris were removed by centrifugation for 10 min at maximum speed. Supernatant was transferred to clean tube, and 500 μl PI-supplied Dilution Buffer (20 mM Tris-HCl (pH 7.4), 150 mM NaCl, 2 mM EDTA and 1% Triton X-100) was added. The lysate was incubated with rabbit anti-TLE3 antibody (Proteintech, 11372-1-AP) or anti-Cbfb antibody pre-bound to Dynabeads Protein G beads (Thermo Fisher Scientific) overnight at 4 °C with gentle rotating. Chromatin-beads complex was washed with ChIP-RIPA Buffer (50 mM HEPES [pH 7.6], 500 mM LiCl, 1 mM EDTA, 1% NP-40, 0.7% Sodium Deoxycholate) > 5 times, and with TE buffer supplemented with 50 mM NaCl once. Chromatin-beads complex was resuspended to 100 ml of Elution Buffer (50 mM Tris-HCl [pH 8.0], 10 mM EDTA, 1% SDS) and incubated overnight at 65 °C with vigorous shaking. Samples were treated with RNaseA (Thermo Fisher Scientific) at 37 °C for 2 h followed by incubation with Proteinase K (Thermo Fisher Scientific) at 55 °C in the presence of 6 mM CaCl₂ for 1 h. Precipitated chromatin DNA was purified using ChIP DNA Clean & Concentrator (ZYMO Research). Purified DNA was subjected to the quantification of E4, S4 region with PowerUp SYBR Green Master Mix (Thermo Fisher Scientific), QuantStudio 3 Real-Time PCR System (Thermo Fisher Scientific) and the following specific primers (Eurofins Genomics); E4p-F: 5′-TCTCCAAAGGGTAACAGGTGTCAG-3′, E4p-R: 5′-TGTGACTTACAAAGGTGCCTCCA-3′, S4-F: 5′-CCTTGTGTGGTCCCTCTCTTTG-3′, S4-R: 5′-GCAACAACCACCCTTCACAGG-3′, Alb-F: 5′-ACCTGCGTTACAGCATCCAC-3′, and Alb-R: 5′-TGCTGACAGAGCAGGAGACA-3′.

AlphaScreen

Recombinant proteins of mouse Runx1 and Runx3 were synthesized as N-terminal Flag-tagged and N-terminal single biotinylated form using a wheat cell-free protein synthesis system according to the procedures described previously²⁵. To increase the yield and solubility of Runx proteins, self-annealed oligonucleotides encoding Runx binding sequence (sense, 5′-AGATGTGTGGTTAACCACAAAC-3′ and antisense, 5′-GTTTGTGGTTAACCACACATCT-3′) were added to the cell-free synthesis reaction⁴⁶. In some reaction, the nucleotides in which mutations were subjected to one or both of the two Runx binding sequences were added. The sequences were as follows: M1: sense, 5′-AGATGTGTCCTTAACCACAAAC-3′ and antisense, 5′-GTTTGTGGTTAAGGACACATCT-3′, M2: sense, 5′-AGATGTGTGGTTAAGGACAAAC-3′ and antisense, 5′-GTTTGTCCTTAACCACACATCT-3′, and M3: sense, 5′-AGATGTGTCCTTAAGGACAAAC-3′ and antisense, 5′-GTTTGTCCTTAAGGACACATCT-3′ (where underline represents sequences that were mutated). These oligonucleotides were purchased from Thermos Fisher Scientific. A total of 497 human protein kinases containing 96 tyrosine kinases were synthesized as N-terminal FLAG-GST-tagged form using the same procedure used for Runx synthesis. In vitro binding assay using AlphaScreen was performed as previously described²⁵ with slight modifications. Briefly, the binding reaction was performed in 15 µl binding buffer (50 mM Tris-HCl, pH 7.5, 1 mg ml⁻¹ bovine serum albumin, 0.01% Tween20) containing 1 µl crude translation mixture of Flag-tagged and biotinylated proteins in a 384-well Opti-plate (PerkinElmer). The binding between two proteins was detected by anti-DYKDDDDK antibody (1E6, Wako) using AlphaScreen IgG (protein A) detection kit (Perkin Elmer). After 1 h incubation at room temperature, 10 µl detection mixture containing 7.5 ng anti-DYKDDDDK antibody (1E6, Wako), 0.08 μl streptavidin-coated donor beads and 0.08 μl protein A-conjugated acceptor beads (Perkin Elmer) were added to the binding rection in the Opti-plate. Binding between two proteins was detected with an EnVision multi-plate reader.

Mass spectrometry

Jurkat cells and J.CaM1.6 cells transduced with Ty1-tagged murine Runx1 or Runx1^WRPF were treated with or without pervanadate (500 μM pervanadate, 3 mM hydrogen peroxide) for 15 min at 37 °C. Murine primary CD4 SP and CD8 SP thymocytes were prepared from Runx1^Ty1/Ty1 mice expressing OT-II or OT-I transgenic TCR, respectively. The murine primary cells (10⁷ cells) or the cultured cells (10⁸ cells) were lysed with ice-cold RIPA buffer (20 mM HEPES-NaOH pH7.5, 1 mM EGTA, 1 mM MgCl₂, 150 mM NaCl, 0.25% Na-deoxycholate, 0.05% SDS, 1% NP-40, Benzonase (Merck), PhosSTOP phosphatase inhibitor (Roche) and cOmplete protease inhibitor cocktail (Roche)). After the lysates were centrifuged at 20,000 × g for 15 min at 4 °C, the supernatants were incubated for 2 h at 4 °C with a 2.5 µl slurry of Sera-Mag SpeedBeads Protein A/G (Cytiva) pre-incubated with 1 µg anti-Ty1-tag mouse monoclonal IgG (Thermo Fisher Scientific, MA5-23513). The beads were washed four times with RIPA buffer and then twice with 50 mM ammonium bicarbonate. Proteins on the beads were digested by adding 200 ng trypsin/Lys-C mix (Promega) at 37 °C overnight. The resulting digests were reduced, alkylated, acidified, and desalted using GL-Tip SDB (GL Sciences). The eluates were evaporated and dissolved in 0.1% trifluoroacetic acid and 3% acetonitrile (ACN). LC-MS/MS analysis of the resultant peptides was performed on an EASY-nLC 1200 UHPLC connected to a Q Exactive Plus mass spectrometer through a nanoelectrospray ion source (Thermo Fisher Scientific). The peptides were separated on a C18 reversed-phase column (75 µm (inner diameter) × 150 mm; Nikkyo Technos) with a linear 4–32% ACN gradient for 0–100 min, followed by an increase to 80% ACN for 10 min and final hold at 80% ACN for 10 min. The mass spectrometer was operated in data-dependent acquisition mode with a top 10 MS/MS method. MS1 spectra were measured with a resolution of 70,000, an automatic gain control (AGC) target of 1e6 and a mass range of 350 to 1,500 m/z. HCD MS/MS spectra were acquired at a resolution of 17,500, an AGC target of 5e4, an isolation window of 2.0 m/z, a maximum injection time of 60 ms and a normalized collision energy of 27. Dynamic exclusion was set to 20 s. Raw data were directly analyzed against the SwissProt database restricted to Homo sapiens supplemented with Ty1-tagged murine Runx1 sequence using Proteome Discoverer v.2.4 (Thermo Fisher Scientific) with Sequest HT search engine. The search parameters were as follows: (i) trypsin as an enzyme with up to two missed cleavages, (ii) precursor mass tolerance of 10 ppm, (iii) fragment mass tolerance of 0.02 Da, (iv) carbamidomethylation of cysteine as a fixed modification and (v) acetylation of protein N terminus, oxidation of methionine and phosphorylation of serine, threonine and Y as variable modifications. Peptides were filtered at a false discovery rate of 1% using the Percolator node. Several selected peptides of Runx1, TLE3, LCK, FYN and ZAP70 were measured by parallel reaction monitoring (PRM)⁴⁷, an MS/MS-based targeted quantification method using high-resolution MS. For quantitative analyses of phosphorylated versus unphosphorylated LEEVWRPY peptide of Runx1 protein, synthetic LEEAVWRPY and LEEAVWRPpY peptides (GL Biochem Japan) were measured to generate a standard curve. Targeted MS/MS scans were acquired by a time-scheduled inclusion list at a resolution of 70,000, an AGC target of 2e5, an isolation window of 2.0 m/z, a maximum injection time of 1 s and a normalized collision energy of 27. Time alignment and relative quantification of the transitions were performed using Skyline software.

Co-immunoprecipitation and immunoblotting

Peripheral CD8⁺ T cells were isolated form the spleens of Rx1^Ty1/Ty1:Rx3^Flag/Flag mice using EasySep Mouse CD8⁺ T Cell Isolation Kit (STEMCELL Technologies). Murine CD4 SP or CD 8SP thymocytes were obtained by beads cell sorting. Murine CD4 SP or CD 8SP thymocytes (10⁶ cells), peripheral CD8⁺ T cells (10⁷ cells) or Jurkat cells (10⁷ cells) were lysed in RIPA buffer (20 mM HEPES-NaOH pH 7.5, 1 mM EGTA, 1 mM MgCl₂, 150 mM NaCl, 0.25% sodium deoxycholate, 1% NP-40, Benzonase (Merck), PhosSTOP phosphatase inhibitor (Roche) and cOmplete protease inhibitor cocktail (Roche)). After centrifuging the lysates at 20,000 × g for 30 min at 4 °C, the supernatants were incubated for 4 h at 4 °C with 50 μl Protein A beads (Dynabeads) pre-incubated with 2.5 μg of anti-TLE3 antibody (Proteintech). The beads were washed four times with RIPA buffer, boiled in SDS-sample buffer and subjected to SDS-PAGE. Immunoblotting was performed using anti-Ty1 (Invitrogen, MA5-23513) to detect Ty1-Runx1, anti-Flag (Sigma-Aldrich: A8592) to detect 3xFlag-TCF1, and anti-Gapdh (Santa Cruz Biotechnology, sc-32233). Antibody binding was detected using ECL Prime Western Blotting Detection Reagent (Cytiva), and images were acquired on an Amersham Imager 680 (Cytiva). For AP treatment, 20 U alkaline phosphatase (Roche, 11097075001) was added to the Jurkat cell lysate and incubated at room temperature for 15 min.

Model building

To understand the dynamics and molecular interactions between TLE3 and the C-terminal WRPY-peptide of Runx1, the homology models of mouse TLE3 (mTLE3) and WRPY-peptide of mouse Runx1 (mRunx1) were first constructed using the crystal structure of human TLE-WD domain (hTLE) and the AlphaFold2⁴⁸ predicted model of human Runx1 (hRunx1), respectively. The crystal structure of the C-terminal SMWRPW peptide from the human Hes1 bound to the TLE-WD domain (PDB ID: 2CE9)¹¹ was used as a template for the modeling of mRunx1 WRPY-peptide. For executing AlphaFold2, the run_docker.py python script was used to point to the FASTA file of the sequence, and max_template_date was set to a very recent date so that all the latest and up-to-date templates can be searched and used for model building. Structural superposition between mTLE3-hTLE1 and mRunx1-hRunx1 models was performed to evaluate the degree of similarity.

System preparation and all-atom MD simulations

The modeled mTLE3-mRunx1 WRPY-peptide complex structures were used for molecular dynamics (MD) simulations. To assess the impact of mutations, four mutated complexes were prepared and an additional system comprising only WRP-peptide of Runx1 was also prepared (Extended Data Table 1). The four mutated complexes harbored (i) WRPpY, (ii) WRPF, (iii) WRPE and (iv) WRPW peptides at the C terminus of Runx1 (where pY denotes phosphorylated tyrosine). For mutating the tyrosine residue and generating the mutant complexes (except WRPpY), the ‘swapaa’ tool of UCSF Chimera was used. For modeling the WRPpY structure, the UCSF Chimera package with SwissSidechain module was used to incorporate the nnAA-library, where the ‘swapnaa’ tool was used to select the chain ID, residue index and automatically assign the most favorable rotamer.

A total of six systems in addition to the WT complex (Extended Data Table 1) were used for MD simulations. The simulations of the complexes were carried out using GROMACS 5.1.2 software package with the CHARMM27 force field⁴⁹. Each system was subjected to the addition of hydrogen atoms and subsequently solvated in a dodecahedron box of TIP3P water in the center at least 1.0 nm from the box boundary. The individual solvated system was then electrostatically neutralized by the addition of counterions, and subjected to 50,000 steps of energy minimization and an equilibration step consisting of a heating step from 0 to 300 K in 200 ps and a constant temperature phase at 300 K for 1 ns. The Parrinello-Rahman barostat pressure coupling was used to avoid the impact of velocity. Then, production runs were carried out for 200 ns for each system (total 1.2 µs) with periodic boundary conditions in an NPT ensemble with modified Berendsen temperature coupling and the constant pressure of 1 atm⁵⁰. In this step, the LINCS algorithm was used to constrain the bond lengths, the Particle-mesh Ewald method was employed to calculate the electrostatic forces, the Fourier grid spacing and Coulomb radius were set at 0.16 and 1.4 nm respectively, and the van der Waals interactions were limited to 1.4 nm. The snapshots from the production MD were saved every 10 picoseconds for structural and dynamic analyses.

Analysis from MD simulations

The MD simulation trajectories of mTLE3-mRunx1 WRPY-peptide complex structures were analyzed using gmx rms and gmx gyrate GROMACS utilities to obtain the root mean square deviation and radius of gyration of each system, respectively. Various types of intermolecular interactions between mTLE3 and mRunx1 C-terminal peptides for all the molecular systems were computed using the Arpeggio web server⁵¹. The intermolecular hydrogen bond interactions formed between mTLE3 and mRunx1 C-terminal peptide for all the molecular systems were analyzed from the MD simulation trajectories using GetContacts (https://getcontacts.github.io/). The contacts were shown in a clustergram to make the interpretation clear for visualization.

Essential dynamics of mTLE3-mRunx1 C-terminal peptide complexes

The essential dynamics, which represent the principal motion directions by a set of eigenvectors, were performed on the MD trajectories of mTLE3-mRunx1 WRPY (WT), mTLE3-mRunx1 WRPpY, mTLE3-mRunx1 WRPF, mTLE3-mRunx1 WRPE, mTLE3-mRunx1 WRPW, and mTLE3-mRunx1 WRP complexes. In this analysis, a variance/covariance matrix was constructed by calculating the eigenvectors and eigenvalues, and their projection along the first two principal components was monitored by the PCA. The eigenvalues associated with each of the eigenvectors of mTLE3-mRunx1 WRPY (WT), mTLE3-mRunx1 WRPpY, mTLE3-mRunx1 WRPF, mTLE3-mRunx1 WRPE, mTLE3-mRunx1 WRPW, and mTLE3-mRunx1 WRP complexes were used to calculate the percentage of variability. The conformational changes associated with the FEL for the complexes were computed by gmx sham package⁵².

Binding free energy of mTLE3-mRunx1 C-terminal peptide complexes

The binding free energies between mTLE3 and mRunx1 C-terminal peptides for the mTLE3-mRunx1 WRPY (WT), mTLE3-mRunx1 WRPpY, mTLE3-mRunx1 WRPF, mTLE3-mRunx1 WRPE, mTLE3-mRunx1 WRPW, and mTLE3-mRunx1 WRP complexes were calculated using the molecular mechanics/Poisson Boltzmann surface area (MM/PBSA) methodology employed in the g_mmpbsa tool⁵³ of GROMACS. The binding free energies of the complexes were computed as per the methodology described previously⁵⁴. Finally, the binding free energy of complexes was calculated from 200 snapshots over the last 20 ns of the simulation trajectories as all the systems were stable during this time period.

In situ PLA

Total thymocytes and splenocytes were freshly isolated from mice, and specific populations were purified by sorting with BD FACSAria III as needed. Sample cells except derivates of cell lines and primary cells with GFP expression were stained with Alexa Fluor 488-conjugated anti-CD45 (Clone: 30-F11, Dilution: 1/200, BioLegend, 103122) antibody for 30 min on ice. Cells were washed with D-PBS (-) twice, resuspended to D-PBS (-) supplemented with 2% FBS at a density of 2 ×105 cells/ml, and 100 μl of the suspension was mounted on a glass slide by centrifugation with Cytospin 4 (Thermo Fisher Scientific) at 800 rpm for 5 min. Cells were fixed for 10 min with 4% paraformaldehyde in D-PBS (-) followed by the permeabilization for 10 min with 0.5% Triton-X-100. Glass slides were rinsed in D-PBS (-), and subjected to in situ PLA. After the blocking for an hour with supplied blocking buffer (Sigma-Aldrich), slides were incubated overnight with primary antibodies at 4 °C. Following antibodies were used; rabbit anti-TY1-tag polyclonal IgG (dilution: 1/100; GenScript, A01004), and mouse anti-FLAG-tag monoclonal IgG (Dilution: 1/100, Sigma-Aldrich, E3165), mouse anti-Lck monoclonal IgG (Dilution: 1/40, Santa Cruz Biotechnology, sc-433), rabbit anti-Bcl11b polyclonal IgG (dilution: 1/100; Bethyl Laboratories, A300-383A), rabbit anti-Cbfb polyclonal IgG (Dilution: 1/100, homemade), and mouse anti-Zap70 monoclonal IgG (dilution: 1/40; Santa Cruz Biotechnology, sc-32760). To generate mouse anti-VWRPY monoclonal IgG (dilution: 1/100, clone: Rp-6C2), 6xHis-tagged full-length human RUNX3 protein, expressed baculovirally in Sf9 cells and purified using a Ni-NTA resin (Superflow; QIAGEN), was used as an antigen to immunize mice⁵⁵. Of the hybridoma clones producing pan-reactive antibodies to RUNX1, RUNX2 and RUNX3, the Rp-6C2 clones were found to be VWRPY specific, as revealed by western blotting³⁶. Slides were washed with Wash Buffer A (Sigma-Aldrich) twice, and reacted with Duolink In Situ PLA Probe Anti-Mouse PLUS and Anti-Rabbit MINUS (Sigma-Aldrich) for 1 h at 37 °C followed by washing steps with Wash Buffer A. Ligation and amplification procedures were performed with Duolink In Situ Detection Reagents Orange (Sigma-Aldrich) according to manufacturer’s protocol. After the amplification step, samples were wash with 1x Wash Buffer B (Sigma-Aldrich) twice and once with 0.1x Wash Buffer B, and encapsulated with cover glass and Duolink In Situ Mounting Medium with DAPI (Sigma-Aldrich). Specimens were examined using BZ-X810 (Keyence) or ECLIPSE Ti-TIRF (Nikon) equipped with an sCMOS camera (ORCA flash 4.0, Hamamatsu Photonics), and densitometry was done with Image J software.

Data and statistical analyses

Data distribution was assumed to be normal, but this was not formally tested. Statistical analysis was performed by F-test and unpaired Students’ t-test with or without Welch’s correction using GraphPad Prism (v8.4.3) (GraphPad Software) or Excel (v2408) (Microsoft). The sequence information of the oligonucleotides is provided in Extended Data Table 2. The number of mice per group, the number of replicates per experiment, summary statistics and measures of dispersion are indicated in the legend of each figure. Mouse phenotyping was conducted under conditions that were largely randomized, and PLA experiments were fully randomized. Data collection and analysis were conducted without blinding to experimental groups because the mice had been genotyped before experimentation. No data were excluded from the analyses. Sample sizes were consistent with those commonly reported in the literature.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

ChIP-seq and RNA-seq data that support the findings of this study have been deposited with accession GSE280658 and GSE307041, respectively. Source data are provided with this paper.

References

Ashby, K. M. & Hogquist, K. A. A guide to thymic selection of T cells. Nat. Rev. 24, 103–117 (2024).
CAS Google Scholar
Taniuchi, I. CD4 helper and CD8 cytotoxic T cell differentiation. Annu. Rev. Immunol. 36, 579–601 (2018).
Article CAS PubMed Google Scholar
Singer, A., Adoro, S. & Park, J. H. Lineage fate and intense debate: myths, models and mechanisms of CD4- versus CD8-lineage choice. Nat. Rev. Immunol. 8, 788–801 (2008).
Article CAS PubMed PubMed Central Google Scholar
Kappes, D. J., He, X. & He, X. Role of the transcription factor Th-POK in CD4:CD8 lineage commitment. Immunol. Rev. 209, 237–252 (2006).
Article CAS PubMed Google Scholar
Sawada, S., Scarborough, J. D., Killeen, N. & Littman, D. R. A lineage-specific transcriptional silencer regulates CD4 gene expression during T lymphocyte development. Cell 77, 917–929 (1994).
Article CAS PubMed Google Scholar
He, X. et al. CD4-CD8 lineage commitment is regulated by a silencer element at the ThPOK transcription-factor locus. Immunity 28, 346–358 (2008).
Article CAS PubMed Google Scholar
Setoguchi, R. et al. Repression of the transcription factor Th-POK by Runx complexes in cytotoxic T cell development. Science 319, 822–825 (2008).
Article CAS PubMed Google Scholar
Taniuchi, I. et al. Differential requirements for Runx proteins in CD4 repression and epigenetic silencing during T lymphocyte development. Cell 111, 621–633 (2002).
Article CAS PubMed Google Scholar
Xing, S. et al. Tle corepressors are differentially partitioned to instruct CD8⁺ T cell lineage choice and identity. J. Exp. Med. 215, 2211–2226 (2018).
Article CAS PubMed PubMed Central Google Scholar
Levanon, D. et al. Transcriptional repression by AML1 and LEF-1 is mediated by the TLE/Groucho corepressors. Proc. Natl Acad. Sci. USA 95, 11590–11595 (1998).
Article CAS PubMed PubMed Central Google Scholar
Jennings, B. H. et al. Molecular recognition of transcriptional repressor motifs by the WD domain of the Groucho/TLE corepressor. Mol. Cell 22, 645–655 (2006).
Article CAS PubMed Google Scholar
Paroush, Z. et al. Groucho is required for Drosophila neurogenesis, segmentation, and sex determination and interacts directly with hairy-related bHLH proteins. Cell 79, 805–815 (1994).
Article CAS PubMed Google Scholar
Fisher, A. L., Ohsako, S. & Caudy, M. The WRPW motif of the hairy-related basic helix-loop-helix repressor proteins acts as a 4-amino-acid transcription repression and protein-protein interaction domain. Mol. Cell. Biol. 16, 2670–2677 (1996).
Article CAS PubMed PubMed Central Google Scholar
Hassan, H. et al. Cd8 enhancer E8I and Runx factors regulate CD8α expression in activated CD8⁺ T cells. Proc. Natl Acad. Sci. USA 108, 18330–18335 (2011).
Article CAS PubMed PubMed Central Google Scholar
Levanon, D. et al. The Runx3 transcription factor regulates development and survival of TrkC dorsal root ganglia neurons. EMBO J. 21, 3454–3463 (2002).
Article CAS PubMed PubMed Central Google Scholar
Inoue, K. et al. Runx3 controls the axonal projection of proprioceptive dorsal root ganglion neurons. Nat. Neurosci. 5, 946–954 (2002).
Article CAS PubMed Google Scholar
Etzensperger, R. et al. Identification of lineage-specifying cytokines that signal all CD8⁺-cytotoxic-lineage-fate ‘decisions’ in the thymus. Nat. Immunol. 18, 1218–1227 (2017).
Article CAS PubMed PubMed Central Google Scholar
Park, J. H. et al. Signaling by intrathymic cytokines, not T cell antigen receptors, specifies CD8 lineage choice and promotes the differentiation of cytotoxic-lineage T cells. Nat. Immunol. 11, 257–264 (2010).
Article CAS PubMed PubMed Central Google Scholar
Lee, H. C., Shibata, H., Ogawa, S., Maki, K. & Ikuta, K. Transcriptional regulation of the mouse IL-7 receptor alpha promoter by glucocorticoid receptor. J. Immunol. 174, 7800–7806 (2005).
Article CAS PubMed Google Scholar
Egawa, T., Tillman, R. E., Naoe, Y., Taniuchi, I. & Littman, D. R. The role of the Runx transcription factors in thymocyte differentiation and in homeostasis of naive T cells. J. Exp. Med. 204, 1945–1957 (2007).
Article CAS PubMed PubMed Central Google Scholar
Mebius, R. E. Organogenesis of lymphoid tissues. Nat. Rev. Immunol. 3, 292–303 (2003).
Article CAS PubMed Google Scholar
Tachibana, M. et al. Runx1/Cbfβ2 complexes are required for lymphoid tissue inducer cell differentiation at two developmental stages. J. Immunol. 186, 1450–1457 (2011).
Article CAS PubMed Google Scholar
Ebihara, T. et al. Runx3 specifies lineage commitment of innate lymphoid cells. Nat. Immunol. 16, 1124–1133 (2015).
Article CAS PubMed PubMed Central Google Scholar
Ogawa, E. et al. Molecular cloning and characterization of PEBP2 beta, the heterodimeric partner of a novel Drosophila runt-related DNA binding protein PEBP2 alpha. Virology 194, 314–331 (1993).
Article CAS PubMed Google Scholar
Takahashi, H. et al. Establishment of a wheat cell-free synthesized protein array containing 250 human and mouse E3 ubiquitin ligases to identify novel interaction between E3 ligases and substrate proteins. PLoS ONE 11, e0156718 (2016).
Article PubMed PubMed Central Google Scholar
Huang, X., Peng, J. W., Speck, N. A. & Bushweller, J. H. Solution structure of core binding factor beta and map of the CBF alpha binding site. Nat. Struct. Biol. 6, 624–627 (1999).
Article CAS PubMed Google Scholar
Wang, Q. et al. Disruption of the Cbfa2 gene causes necrosis and hemorrhaging in the central nervous system and blocks definitive hematopoiesis. Proc. Natl Acad. Sci. USA 93, 3444–3449 (1996).
Article CAS PubMed PubMed Central Google Scholar
Okuda, T., van Deursen, J., Hiebert, S. W., Grosveld, G. & Downing, J. R. AML1, the target of multiple chromosomal translocations in human leukemia, is essential for normal fetal liver hematopoiesis. Cell 84, 321–330 (1996).
Article CAS PubMed Google Scholar
Mucida, D. et al. Transcriptional reprogramming of mature CD4⁺ helper T cells generates distinct MHC class II-restricted cytotoxic T lymphocytes. Nat. Immunol. 14, 281–289 (2013).
Article CAS PubMed PubMed Central Google Scholar
Seo, W. et al. Roles of VWRPY motif-mediated gene repression by Runx proteins during T-cell development. Immunol. Cell Biol. 90, 827–830 (2012).
Article CAS PubMed Google Scholar
Huang, H. et al. A Src family kinase-Shp2 axis controls RUNX1 activity in megakaryocyte and T-lymphocyte differentiation. Genes Dev. 26, 1587–1601 (2012).
Article CAS PubMed PubMed Central Google Scholar
Goldsmith, M. A. & Weiss, A. Isolation and characterization of a T-lymphocyte somatic mutant with altered signal transduction by the antigen receptor. Proc. Natl Acad. Sci. USA 84, 6879–6883 (1987).
Article CAS PubMed PubMed Central Google Scholar
Chodaparambil, J. V. et al. Molecular functions of the TLE tetramerization domain in Wnt target gene repression. EMBO J. 33, 719–731 (2014).
Article CAS PubMed PubMed Central Google Scholar
Goh, Y. M. et al. Src kinase phosphorylates RUNX3 at tyrosine residues and localizes the protein in the cytoplasm. J. Biol. Chem. 285, 10122–10129 (2010).
Article CAS PubMed PubMed Central Google Scholar
Fredriksson, S. et al. Protein detection using proximity-dependent DNA ligation assays. Nat. Biotechnol. 20, 473–477 (2002).
Article CAS PubMed Google Scholar
Otani, S. et al. Runx3 is required for oncogenic Myc upregulation in p53-deficient osteosarcoma. Oncogene 41, 683–691 (2022).
Article CAS PubMed Google Scholar
Kojo, S. et al. Priming of lineage-specifying genes by Bcl11b is required for lineage choice in post-selection thymocytes. Nat. Commun. 8, 702 (2017).
Article PubMed PubMed Central Google Scholar
Muroi, S. et al. Cascading suppression of transcriptional silencers by ThPOK seals helper T cell fate. Nat. Immunol. 9, 1113–1121 (2008).
Article CAS PubMed Google Scholar
Voisinne, G. et al. Quantitative interactomics in primary T cells unveils TCR signal diversification extent and dynamics. Nat. Immunol. 20, 1530–1541 (2019).
Article CAS PubMed PubMed Central Google Scholar
Robertson, A. J., Larroux, C., Degnan, B. M. & Coffman, J. A. The evolution of Runx genes II. The C-terminal Groucho recruitment motif is present in both eumetazoans and homoscleromorphs but absent in a haplosclerid demosponge. BMC Res. Notes 2, 59 (2009).
Article PubMed PubMed Central Google Scholar
Bae, S. C. & Lee, J. cDNA cloning of run, a Caenorhabditis elegans Runt domain encoding gene. Gene 241, 255–258 (2000).
Article CAS PubMed Google Scholar
Winnier, A. R. et al. UNC-4/UNC-37-dependent repression of motor neuron-specific genes controls synaptic choice in Caenorhabditis elegans. Genes Dev. 13, 2774–2786 (1999).
Article CAS PubMed PubMed Central Google Scholar
Lee, P. P. et al. A critical role for Dnmt1 and DNA methylation in T cell development, function, and survival. Immunity 15, 763–774 (2001).
Article CAS PubMed Google Scholar
Hogquist, K. A. et al. T cell receptor antagonist peptides induce positive selection. Cell 76, 17–27 (1994).
Article CAS PubMed Google Scholar
Barnden, M. J., Allison, J., Heath, W. R. & Carbone, F. R. Defective TCR expression in transgenic mice constructed using cDNA-based alpha- and beta-chain genes under the control of heterologous regulatory elements. Immunol. Cell Biol. 76, 34–40 (1998).
Article CAS PubMed Google Scholar
Yonezawa, T. et al. The ubiquitin ligase STUB1 regulates stability and activity of RUNX1 and RUNX1-RUNX1T1. J. Biol. Chem. 292, 12528–12541 (2017).
Article CAS PubMed PubMed Central Google Scholar
Peterson, A. C., Russell, J. D., Bailey, D. J., Westphall, M. S. & Coon, J. J. Parallel reaction monitoring for high resolution and high mass accuracy quantitative, targeted proteomics. Mol. Cell Proteom. 11, 1475–1488 (2012).
Article Google Scholar
Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).
Article CAS PubMed PubMed Central Google Scholar
Pronk, S. et al. GROMACS 4.5: a high-throughput and highly parallel open source molecular simulation toolkit. Bioinformatics 29, 845–854 (2013).
Article CAS PubMed PubMed Central Google Scholar
Bussi, G., Donadio, D. & Parrinello, M. Canonical sampling through velocity rescaling. J. Chem. Phys. 126, 014101 (2007).
Article PubMed Google Scholar
Jubb, H. C. et al. Arpeggio: a Web server for calculating and visualising interatomic interactions in protein structures. J. Mol. Biol. 429, 365–371 (2017).
Article CAS PubMed Google Scholar
Brooks, C. L. III, Onuchic, J. N. & Wales, D. J. Statistical thermodynamics. Taking a walk on a landscape. Science 293, 612–613 (2001).
Article CAS PubMed Google Scholar
Homeyer, N. & Gohlke, H. Free energy calculations by the molecular mechanics Poisson-Boltzmann surface area method. Mol. Inform. 31, 114–122 (2012).
Article CAS PubMed Google Scholar
Padhi, A. K. et al. An integrated computational pipeline for designing high-affinity nanobodies with expanded genetic codes. Brief. Bioinform. 22, bbab338 (2021).
Article PubMed Google Scholar
Ito, K. et al. RUNX3, a novel tumor suppressor, is frequently inactivated in gastric cancer by protein mislocalization. Cancer Res. 65, 7743–7750 (2005).
Article CAS PubMed Google Scholar

Download references

Acknowledgements

We thank Y. Taniguchi and C. Miyamoto for the maintenance and genotyping of mice, N. Yoza for the cell sorting, Y. Iizuka and H. Tatsumi for genome editing and in vitro fertilization (IVF). We thank S. Yuki, C. Furukawa for preparation of recombinant human tyrosine kinases and C. Takahashi for performing AlphaScreen assay of Runx proteins. This work was supported by Grant-in-Aid for Scientific Research (C) (23K07823 to K.O.), Grant-in-Aid for Scientific Research (B) (17H04090 to I.T.), and Grant-in-Aid for Scientific Research on Innovative Area (19H05747 to I.T.) from the Japanese Society for the Promotion of Science, and Joint Usage and Joint Research Programs of the Institute of Advanced Medical Sciences of Tokushima University (to C.O. and I.T.).

Author information

Satoshi Kojo
Present address: Department of Immunology and Stem Cell Biology, Faculty of Medicine, Institute of Medical, Pharmaceutical and Health Sciences, Kanazawa University, Kanazawa, Ishikawa, Japan
Aditya K. Padhi
Present address: Laboratory for Computational Biology and Biomolecular Design, School of Biochemical Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh, India
Takashi Ebihara
Present address: Department of Medical Biology, Akita University Graduate School of Medicine, Akita, Japan
Mari Tenno
Present address: Research Institute for Biomedical Sciences, Tokyo University of Science, Noda, Chiba, Japan
Kam Y. J. Zhang
Present address: Laboratory for Computational Molecular Design, Center for Biosystems Dynamics Research, RIKEN, Yokohama, Kanagawa, Japan
These authors contributed equally: Chihiro Ogawa, Kazuki Okuyama, Satoshi Kojo.

Authors and Affiliations

Laboratory for Transcriptional Regulation, RIKEN Center for Integrative Medical Sciences (IMS), Yokohama, Kanagawa, Japan
Chihiro Ogawa, Kazuki Okuyama, Satoshi Kojo, Takashi Ebihara, Mari Tenno, Qin Zhizhen, Sawako Muroi & Ichiro Taniuchi
Division of Cell Signaling, Fujii Memorial Institute of Medical Sciences, Institute of Advanced Medical Sciences, Tokushima University, Tokushima, Japan
Kohei Nishino & Hidetaka Kosako
Department of Immunology, Tokyo Medical University, Tokyo, Japan
Hiroaki Machiyama & Tadashi Yokosuka
Laboratory for Structural Bioinformatics, RIKEN Center for Biosystems Dynamics Research, Yokohama, Kanagawa, Japan
Aditya K. Padhi & Kam Y. J. Zhang
Division of Cell-Free Science, Proteo-Science Center, Ehime University, Matsuyama, Ehime, Japan
Hirotaka Takahashi & Tatsuya Sawasaki
Department of Molecular Tumor Biology, Graduate School of Biomedical Sciences, Nagasaki University, Nagasaki, Japan
Kosei Ito
Center for Discovery and Innovation, Hackensack University Medical Center, Nutley, NJ, USA
Hai-Hui Xue

Authors

Chihiro Ogawa
View author publications
Search author on:PubMed Google Scholar
Kazuki Okuyama
View author publications
Search author on:PubMed Google Scholar
Satoshi Kojo
View author publications
Search author on:PubMed Google Scholar
Kohei Nishino
View author publications
Search author on:PubMed Google Scholar
Hiroaki Machiyama
View author publications
Search author on:PubMed Google Scholar
Aditya K. Padhi
View author publications
Search author on:PubMed Google Scholar
Hirotaka Takahashi
View author publications
Search author on:PubMed Google Scholar
Takashi Ebihara
View author publications
Search author on:PubMed Google Scholar
Mari Tenno
View author publications
Search author on:PubMed Google Scholar
Qin Zhizhen
View author publications
Search author on:PubMed Google Scholar
Sawako Muroi
View author publications
Search author on:PubMed Google Scholar
Kosei Ito
View author publications
Search author on:PubMed Google Scholar
Tatsuya Sawasaki
View author publications
Search author on:PubMed Google Scholar
Kam Y. J. Zhang
View author publications
Search author on:PubMed Google Scholar
Hai-Hui Xue
View author publications
Search author on:PubMed Google Scholar
Tadashi Yokosuka
View author publications
Search author on:PubMed Google Scholar
Hidetaka Kosako
View author publications
Search author on:PubMed Google Scholar
Ichiro Taniuchi
View author publications
Search author on:PubMed Google Scholar

Contributions

C.O., K.O., S.K., M.T., T.E. and Q.Z. analyzed the phenotype of gene-modified mice and performed molecular biological experiments. S.M. and I.T. generated gene-modified mice. K.N. and H.K. performed mass spectrometry analysis. H.T. and T.S. performed AlphaScreen. A.K.P. and K.Z. performed modeling. K.I. and H-H. X provided essential regents. H.M. and T.Y. analyzed the microscopic data. I.T. designed this study and wrote the paper.

Corresponding author

Correspondence to Ichiro Taniuchi.

Ethics declarations

Competing interests

Authors declare no competing interests.

Peer review

Peer review information

Nature Immunology thanks the anonymous reviewers for their contribution to the peer review of this work. Peer reviewer reports are available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Generation and characterization of Runx3WRPW mouse strain.

(a) Image of an agarose electrophoresis gel of Bcl1-digested PCR products of nine one-week-old F0 founders on a B6 background generated by CRISPR/Cas9 genome editing to generate the Rx3^W allele. Predicted nucleotide sequences for WT(WRPY) and mutant WRPW alleles and the Bcl1 enzyme site are shown at the bottom. The DNA sequence of the PCR product of an F0 founder on an ICR background, through which Rx3^W strain was established, is shown at the right. PCR products were cloned into the plasmid and sequenced by the Sanger method. A single experiment was performed. (b) Pseudo-color plots showing CD3ε and NK1.1 expression by spleen cells. The graph at the bottom shows a statistical summary of the frequency of NK1.1⁺ cells in CD45⁺ spleen cells. Rx3^+/+ (n = 5), Rx3^+/W (n = 5) and Rx3^W/W (n = 5). Mean ± SD. *: p < 0.05, ***: p < 0.005, ****: p < 0.0001 (one-way ANOVA with Tukey’s multiple comparison). (c) Pseudo-color plots showing CD45 expression by epidermal cells, MHC-II and CD3ε expression by the CD45⁺ subset, and EpCAM and CD24 expression in the CD45⁺ MHC-II⁺ subset. The graph at the bottom shows a statistical summary of the frequency of the indicated cells in the indicated population. Rx3^+/+ (n = 3), Rx3^+/W (n = 3) and Rx3^W/W (n = 3). Mean ± SD. *: p < 0.05, **: p < 0.01, ***: p < 0.005, ****: p < 0.0001 (one-way ANOVA with Tukey’s multiple comparison).

Source data

Extended Data Fig. 2 Runx1 and Runx3 heterodimerization and cell intrinsic increase of ILC2 progenitor.

(a) Venn diagram showing the H3K4me3 regions in the whole genome between Rx3^+/+ and Rx3^+/W mice. (b) Graphs showing numbers of lymph nodes (LNs) at the indicated location and Peyer’s patches (PPs) in Rx3^+/+(n = 8 for LNs and n = 6 for PPs), Rx3^+/W(n = 8 for LNs and n = 6 for PPs), and Rx3^W/W (n = 8 for LNs and n = 6 for PPs) mice. Mean ± SD. ****: p < 0.0001 (ANOVA with Tukey’s multiple comparisons test for LNs and unpaired student t-test,two-sided for PPs). (c, d) Schemas showing the structure of the Rx1^TY1 allele with amino acid sequences of theTY1 tag (c) and the Rx3^Flag allele with amino acid sequences of the Flag tag (d) generated by CRISPR/Cas9 genome editing. Images of agarose gel electrophoresis of one-week-old F0 founders are shown at the bottom. A single experiment was performed. (e) Scheme showing the principle of AlphaScreen to examine the interaction of biotin-tagged Runx with Flag-tagged Runx. (f) Effect of Cbfβ and different oligonucleotides with or without the canonical Runx recognition motif (5′-TGTGGT-3′and 5′-ACCACA-3′) on the interaction between biotin-tagged Runx1 and Flag-tagged Runx3 in AlphaScreen. (g) Scheme showing the experimental flow of the bone marrow chimera experiment. The right graph shows a summary of the frequencies of CD45.1⁺ and CD45.1⁺CD45.2⁺ ILC1, ILC2P, ILC2, and ILC3 in the bone marrow of three recipient mice (n = 3). Mean ± SEM. *: p < 0.05, ****: p < 0.0001 (unpaired student t-test, two-sided for ILC2 and Welch’s t-test, two-sided for ILC2P).

Source data

Extended Data Fig. 3 Generation of Runx1WRPW mouse strain.

(a) Image of an agarose electrophoresis gel of Bcl1 digested PCR products of twelve 12.5 dpc F0 embryos modified by CRISPR/Cas9 genome editing to generate the Rx1^W allele. The predicted nucleotide sequences for the WT(WRPY) and mutant WRPW alleles and the Bcl1 enzyme site are shown at the bottom. The DNA sequence of the PCR product of the #6 embryo is shown at the right. PCR products were cloned into the plasmid and then sequenced. A single experiment was performed. (b) Table showing numbers of Runx1(Rx1)^+/+ and knock-in (KI), which are presumably Rx1^+/W, embryos, at the indicated dpc generated by CRISPR/Cas9 genome editing. Figures indicated numbers of live embryos and those in parentheses indicate total embryos.

Source data

Extended Data Fig. 4 Generation and characterization of Runx transgenic mouse strains.

(a) A scheme showing the structure of the Rosa26 allele with DNA fragments inserted for the inducible transgenic expression of Runx proteins. (b) Immunoblots showing expression of Flag-tagged transgenic Runx proteins in total thymocyte of non-transgenic (NTG), R26^DP-R3, R26^DP-R3-WRPW, R26^DP-R1, R26^DP-R1-WRPE, R26^DP-R1-WRPF, and R26^DP-R1-WRPW mice. Antibodies against Runx1 and Runx3 were used to detect both endogenous and transgenic Runx1 and Runx3 proteins. Gapdh was used for a loading control. Two independent experiments were performed. (c) Dot plots showing CD24 and TCRβ expression in total thymocytes and contour plots showing CD69 and TCRβ expression in CD24⁺ thymocytes. Graphs at the right show absolute numbers of CD24⁺CD69⁺ and CD24⁻TCRβ⁺ cells of mice with the indicated genotype. NTG (n = 14), R26^DP-R1-WRPW (n = 6) and R26^DP-R3-WRPW (n = 5). Mean ± SD. *: p < 0.05, **: p < 0.01 (one-way ANOVA with Tukey’s multiple comparison) (d) Dot Plots showing CD4 and CD8α expression in mature thymocytes of Cd4^ΔS/+ and R26^DP-R1-WRPW: Cd4^ΔS/+ mice. The right graph shows the frequency of CD4SP and DP in mature thymocytes and the numbers of mature CD4SP and DP thymocytes. Cont (n = 5) and R26^DP-R1-WRPW (n = 5). Mean ± SD. **: p < 0.01 (unpaired Student’s t-test, two-sided) (e) DNA sequences of PCR products of the R26^lsl-R1-WRPE allele generated by CRISPR/Cas9 genome editing on the R26^lsl-R3 allele and the R26^lsl-R1-WRPF allele generated by gene targeting in ES cells. (f) DNA sequences of PCR products of ICR background Rx3^E and Rx3^F F0 founder generated by CRISPR/Cas9 genome editing.

Source data

Extended Data Fig. 5 Normal CD4 SP/CD8SP ratio in Runx1^Y3F/Y3F mutant mice.

(a) Scheme shows structure of Runx1 protein and position of three tyrosine resides, Y375, Y378 and Y379, within inhibitory domain (ID). The nucleotide sequences around the region encoding Y375, Y378 and Y379 and designed mutation for inducing phenylalanine substitution on these tyrosine residues are shown at the middle. One representative result of Sanger sequences of Runx1^+/Y3F mice is shown at the bottom. (b) Pseudocolor plots showing CD4 and CD8α expression in total and mature thymocytes of Runx1^+/+ and Runx1^Y3F/Y3F mice. The right graph shows the frequency of CD4SP and DP in total and mature thymocytes population. Runx1^+/+ (n = 3) and Runx1^Y3F/Y3F (n = 3). Mean ± SD. n.s: not statistically significant (unpaired Student’s t-test,two-sided).

Source data

Extended Data Fig. 6 An in silico modeling to simulate the conformational changes.

(a) ChIP-qPCR showing binding of Cbfβ to the Cd4 enhancer (E4p) and Cd4 silencer (S4) in the Cd4 locus in CD4 SP and CD8 SP thymocytes. The Alb region was used for a negative control. Three independent experiments were conducted. Mean ± SD. *: p < 0.05, **: p < 0.01, ***: p < 0.005 (unpaired Student’s t-test, two-sided) (b) Structural superimposition of the mTLE3 WD-domain bound to the mRunx1 WRPY-peptide complex with the hTLE1 WD-domain bound to the hRunx1 WRPY-peptide complex. The resulting superimposition yielded an RMSD of 0.18 Å. (c) Physicochemical properties of mTLE3-mRunx1 C-terminal peptide complexes from all-atom MD simulations. Backbone RMSDs and Rg profiles of the mTLE3-mRunx1 C-terminal peptide complexes obtained from all-atom MD simulations of 200 ns each. The colored lines denoting each complex are highlighted and presented as an inset inside the figures. (d) The hydrogen bond interactions obtained between the key residues of mTLE3 and the C-terminal peptide of mRunx1 are shown for mTLE3-mRunx1 WRPY (WT), mTLE3-mRunx1 WRPE, mTLE3-mRunx1 WRPF, mTLE3-mRunx1 WRPpY, and mTLE3-mRunx1 WRPW complexes. The gradient of green shows the frequency of interactions over the simulation trajectories. (e) Free-energy landscapes (FELs) generated by projecting the principal components, PC1 and PC2, of mTLE3-mRunx1 C-terminal peptide complexes from MD simulations. FELs for mTLE3-mRunx1 WRPY (WT), mTLE3-mRunx1 WRPE, mTLE3-mRunx1 WRPF, mTLE3-mRunx1 WRPpY, mTLE3-mRunx1 WRPW, and mTLE3-mRunx1 WRP complexes are shown.

Source data

Extended Data Fig. 7 Results of AlphaScreen for Runx1 interacting protein kinases.

(a) Scheme shows the principle of AlphaScreen to examine the interaction of biotin-tagged Runx1 with Flag-tagged Protein kinases (PKs). (b) Summary table showing the results of the AlphaScreen of a customized PKs library covering 497 possible kinases. The table was modified to focus on the group of tyrosine kinases (TK). Numbers in the Rank column indicate ranking among the 497 kinases.

Extended Data Fig. 8 Results of in situ proximity ligation assay (PLA).

(a) Representative in situ PLA images showing the interaction of Runx1 with Lck in splenic CD4⁺ and CD8⁺ T cells of Rx1^Ty1/Ty1 mice. Ty1/NTC indicates a negative control for staining. The right graph shows a summary of the average PLA signal intensity (randomly selected 8 cells). Error bars indicate 95% C.I. values. *: p < 0.05, **: p < 0.01, ****: p < 0.0001 (two-tailed Student’s t-test, two-sided). (b) Representative PLA images showing interactions of Runx1 with Lck, Cbfβ with Lck, and Cbfβ and Runx1 in Runx1^Ty1/Ty1 CD8 SP thymocytes. The right graph shows a summary of the average PLA signal intensity (randomly selected 8 cells). Error bars indicate 95% C.I. values. *: p < 0.05, **: p < 0.01, ****: p < 0.0001 (one-way ANOVA with Tukey’s multiple comparison). (c) Representative PLA images addressing interaction of Bcl11b with Lck and Bcl11b with Runx1 in total thymocytes of Rx1^Ty1/Ty1 mice. The right graph shows a summary of the average PLA signal intensity (randomly selected 8 cells). Error bars indicate 95% C.I. values. (d) A representative higher resolution image of PLA signals in CD8 SP thymocytes of Rx3^Flag/Flag mice with anti-Flag for Runx3 and anti-Lck antibodies obtained by N-SIM TIRF microscopy. Two independent experiments were performed.

Source data

Extended Data Table 1 Comparison of the binding free energy and various energy components between mTLE3and the mRunx1 C-terminal peptide from respective MD simulation trajectories

Full size table

Extended Data Table 2 The sequence information of the oligonucleotides used in the study

Full size table

Supplementary information

Supplementary Information

Caption for Supplementary Video.

Reporting Summary

Peer Review File

Supplementary Video

Video showing two Runx3^+/+ and two Runx3^WRPW/WRPW homozygous mice, at two-weeks old, where Runx3^WRPW/WRPW mice were smaller and suffered from severe limb ataxia. Arrows indicate Runx3^WRPW/WRPW mice.

Source data

Source Data Fig. 1

Statistical source data.

Source Data Fig. 2

Unprocessed western blots, statistical source data.

Source Data Fig. 3

Statistical source data.

Source Data Fig. 4

Statistical source data.

Source Data Fig. 5

Unprocessed western blots, statistical source data.

Source Data Fig. 6

Unprocessed western blots, statistical source data.

Source Data Fig. 7

Statistical source data.

Source Data Extended Data Fig. 1

Unprocessed gel, statistical source data.

Source Data Extended Data Fig. 2

Unprocessed gels, statistical source data.

Source Data Extended Data Fig. 3

Unprocessed gel.

Source Data Extended Data Fig. 4

Unprocessed western blots, statistical source data.

Source Data Extended Data Fig. 5

Statistical source data.

Source Data Extended Data Fig. 6

Statistical source data.

Source Data Extended Data Fig. 8

Statistical source data.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Ogawa, C., Okuyama, K., Kojo, S. et al. Phosphorylation of Runx protein controls helper CD4⁺ T cell versus cytotoxic CD8⁺ T cell lineage choice. Nat Immunol (2026). https://doi.org/10.1038/s41590-026-02441-6

Download citation

Received: 02 February 2025
Accepted: 21 January 2026
Published: 19 February 2026
Version of record: 19 February 2026
DOI: https://doi.org/10.1038/s41590-026-02441-6

Subjects

Abstract

Main

Results

A WRPW missense mutation transforms Runx proteins into repressors

RunxWRPW impairs MHC-II-specific T cell development

Phosphorylation of Runx1 regulates association with TLE3

Runx1 interacts with Lck and Zap70 more in CD8 SP thymocytes

Discussion

Methods

Mice

Cells

Tissue processing

Flow cytometry

Cell cycle analyses

RNA sequencing

Bone marrow chimera

Cell sorting and enrichment

RNA isolation and quantitative PCR

ChIP-seq and ChIP-qPCR

AlphaScreen

Mass spectrometry

Co-immunoprecipitation and immunoblotting

Model building

System preparation and all-atom MD simulations

Analysis from MD simulations

Essential dynamics of mTLE3-mRunx1 C-terminal peptide complexes

Binding free energy of mTLE3-mRunx1 C-terminal peptide complexes

In situ PLA

Data and statistical analyses

Reporting summary

Data availability

References

Acknowledgements

Author information

Authors and Affiliations

Contributions

Corresponding author

Ethics declarations

Competing interests

Peer review

Peer review information

Additional information

Extended data

Supplementary information

Source data

Rights and permissions

About this article

Cite this article

Share this article

Search

Quick links

Runx^WRPW impairs MHC-II-specific T cell development