A lineage-specific requirement for YY1 Polycomb Group protein function in early T cell development

T cell development originates from hematopoietic stem and progenitor cells in the bone marrow or fetal liver, which migrate to the thymus and obtain T cell identity (Scollay et al., 1986; Donskoy and Goldschneider, 1992; Schwarz and Bhandoola, 2004). Thymocytes are divided into multiple phenotypically distinct stages that are delineated by unique cell surface expression markers. T cell development is initiated from double negative (DN) cells that lack expression of both CD4 and CD8, which then become CD4⁺ CD8⁺ double positive (DP) and subsequently differentiate into mature CD4⁺ or CD8⁺ single positive cells (Ciofani and Zúñiga-Pflücker, 2007; Hayday and Pennington, 2007; Rothenberg et al., 2008; Shah and Zúñiga-Pflücker, 2014). Early thymic progenitors (ETPs) are identified as the earliest T cell precursors in the thymus. Development from ETP to DN3 stage is defined as pro-T-cell stage and requires a combination of receptor ligands and growth factors provided by the thymic epithelium to support pro-T cell differentiation, survival and proliferation (Rothenberg et al., 2008). Notch signaling is crucial throughout the pro-T cell stages (Radtke et al., 2004; Maillard et al., 2005). Although inactivation of Notch1 results in DN1T cell developmental blockage (Radtke et al., 1999), overexpression of constitutively active Notch1 in hematopoietic progenitors results in development of T lineage cells to the DP stage and abrogates B cell development (Pui et al., 1999).

T cell lineage commitment requires many regulatory events in addition to Notch signaling. At least four transcription factors, including PU.1 (also known as Spi1), Ikaros (Ikzf1), RUNX1 family factors in complex with CBFβ, and E2A (Tcf3), enable cell progenitors to develop in the T cell lineage before Notch signaling (Dakic et al., 2005; Yoshida et al., 2006; Talebian et al., 2007; Rothenberg et al., 2008). In addition, stem cell factor (SCF; Kitl)/c-Kit (Kit) is crucial for sustaining pro-T cell proliferation. Kit expression is essential from the ETP to DN2 stages and is downregulated through the DN3 stage (Tabrizifard et al., 2004; Massa et al., 2006). Previously, we demonstrated that YY1 promotes c-Kit cell surface expression and SCF/c-Kit signaling in hematopoietic stem cells (HSCs) (Lu et al., 2018). Thus, YY1 may also be a core T cell lineage regulatory factor required for progenitor cell commitment to T cell development. By conditional knockout of Yy1 in HSCs, in this study, we assessed the impact of YY1 deficiency on early T cell commitment, survival and receptor-ligand signaling.

YY1 is a ubiquitous multifunctional transcription factor with important roles in early embryonic development, cell cycle progression, apoptosis and hematopoiesis. We have demonstrated that YY1 promotes adult HSC self-renewal and maintains HSC quiescence (Lu et al., 2018). YY1 also has important roles in lymphocyte development, and is required for all stages of B lymphocyte development (Kleiman et al., 2016). Conditional knockout of YY1 in the B cell lineage generates an arrest at the pro-B cell stage with abnormal heavy chain rearrangement (Pan et al., 2013). Conditional knockout of YY1 by CD2-Cre and Lck-Cre in the T cell lineage leads to a T cell developmental blockage at the DN3 to DN4 stage and the DN-to-DP transitions, respectively (Chen et al., 2016). In addition to its conventional transcription factor attributes, YY1 is a Polycomb Group (PcG) protein and a founding member of a limited cohort of mammalian PcG proteins with sequence-specific DNA binding (Atchison et al., 2003; Srinivasan and Atchison, 2004; Srinivasan et al., 2005). Stable PcG-dependent repression involves the hierarchical recruitment of PcG complexes and chromatin modifications to establish a stable repressed state (Wang et al., 2004). By contrast, non-stable transcriptional repression involves direct competition for DNA binding by activators and repressors, recruitment of corepressors that deacetylate histones or direct interference with the transcriptional machinery. Although PcG complexes must function at specific DNA sites, the mechanisms responsible for recruiting mammalian PcG proteins to specific DNA regions are unclear, as the majority of PcG proteins do not exhibit sequence-specific DNA binding activity. YY1 can repress transcription in a PcG-dependent manner, recruit other PcG proteins to specific DNA sequences and increase H3K27me3 at the recruitment site (Atchison et al., 2003; Srinivasan and Atchison, 2004; Wilkinson et al., 2006). Our previous results have demonstrated that the YY1 PcG domain is required for Igκ chain rearrangement in early B cell development (Pan et al., 2013) but is not necessary for HSC self renewal and quiescence defects caused by YY1 deletion (Lu et al., 2018). This suggests that there may be a lineage-specific requirement for YY1 function in orchestrating the actions of other PcG proteins in adult hematopoiesis. In this study, we utilized a YY1 REPO domain mutant (YY1ΔREPO) to dissect the role of YY1 PcG function in different hematopoietic lineages. The small 25 amino acid REPO domain is necessary and sufficient for recruiting other PcG proteins to YY1-occupied chromatin sites in Drosophila. Although the REPO domain YY1 mutant (YY1ΔREPO) is competent for DNA binding, transcriptional activation, transient transcriptional repression and interaction with transcriptional coregulators such as histone deacetylases, YY1ΔREPO is defective in all YY1 PcG functions. Furthermore, it is unable to recruit other PcG proteins to DNA (Wilkinson et al., 2006). Thus, this mutant constitutes a powerful tool for dissecting mechanisms governing YY1 PcG-dependent versus -independent functions in diverse regulatory environments, e.g. in different hematopoietic lineages.

Herein, we demonstrate that a conditional knockout of Yy1 in the hematopoietic system results in an early T cell developmental blockage at the DN1 stage. YY1-deficient early T cells have reduced Notch1 expression and signaling. There is a lineage-specific requirement for YY1 PcG function/REPO domain, as the YY1 PcG function/REPO domain is important for T and B cell development, but not necessary for myeloid cell development. We present evidence for a multicomponent mechanism by which YY1 controls early T cell development. Although YY1 promotes early T cell survival through its PcG function, it mediates DN1-to-DN2 transition and promotes Notch1 expression and signaling in a PcG-independent manner.

To test how loss-of-function of YY1 impacts early T cell development, Yy1^f/f mice with loxP sites flanking the Yy1 promoter region and exon 1 (Liu et al., 2007) were crossed to the inducible Mx1-Cre. In Yy1^f/f Mx1-Cre mice, YY1 deletion was achieved after treatment with the IFN-α-stimulating polyinosinic:polycytidylic acid (pI-pC). Yy1^f/f Mx1-Cre and Mx1-Cre mice received four doses of pI-pC injections. Because Yy1^f/f Mx1-Cre mice were lethal at 7-10 days post pI-pC injection (Lu et al., 2018), Yy1^f/f Mx1-Cre mice were evaluated at 3-7 days post pI-pC injections. There was a greater than 80% reduction of YY1 transcript (Fig. 1A) and protein (Fig. 1B) levels in Yy1^f/f Mx1-Cre (Yy1^−/−) thymocytes in comparison with Mx1-Cre (Yy1^+/+) control. Yy1^−/− mice had smaller thymuses compared with Yy1^+/+ mice and there was a more than 80% reduction in thymus weight and 90% reduction in total thymocyte number (Fig. 1C). In Yy1^−/− mice, there was a statistically significant increase in percentage of DN2 (Lin⁻CD44⁺CD25⁺) and DN3 (Lin⁻CD44⁻CD25⁺), and a decrease in percentage of DN4 (Lin⁻CD44⁻CD25⁻) (Fig. 1D,E). Yy1^−/− mice had a significant reduction of DN1 (Lin⁻CD44⁺CD25⁻), c-Kit⁺ DN1 (Lin⁻c-Kit⁺CD44⁺CD25⁻), DN2 (Lin⁻CD44⁺CD25⁺), DN3 (Lin⁻CD44⁻CD25⁺) and DN4 (Lin⁻CD44⁻CD25⁻) absolute cell numbers compared with Yy1^+/+ mice (Fig. 1E; Fig. S1). In Yy1^−/− mice, there was a statistically significant decrease of the absolute number of DP (CD4⁺CD8⁺), CD4⁺ and CD8⁺ T cells, but with no significant change of the overall percentage compared with Yy1^+/+ mice (Fig. 1F,G). Our results support essential functions of YY1 in DN, DP and single-positive T cell development. YY1 deficiency leads to reduction of all early-stage T cell numbers and a blockage at the DN3-to-DN4 transition.

In the classical model of lymphopoiesis, ETPs are believed to derive from common lymphoid progenitors (CLPs), which are the obligated intermediate progenitors between lymphoid-primed multipotent progenitors (LMPPs) and mature lymphocytes (Kondo et al., 1997). On the contrary, many studies support that CLPs and ETPs are not obligatorily linked by a progenitor-successor relationship and T cells can also develop via a CLP-independent pathway (Allman et al., 2003; Schwarz and Bhandoola, 2004; Adolfsson et al., 2005; Maillard et al., 2006; Ghaedi et al., 2016). In a recent study, LMPPs were divided by CD127 (Il7r) cell surface expression and CD127⁺ LMPP cells were more efficient than the other progenitors including CLP and CD127⁻ LMPP at differentiating into T cells (Ghaedi et al., 2016). Thus, ETPs and therefore T cells may arise from a progenitor population upstream of CLPs and from a more heterogeneous pool of progenitors than previously thought. As YY1 deficiency causes decreased HSC number and function (Lu et al., 2018), we next assessed percentages and absolute numbers of CLP and LMPP in Yy1^+/+ and Yy1^−/− bone marrow cells. Yy1^−/− mice had normal percentages of CLP (Lin⁻Sca1^loc-Kit^loThy1.2⁻ST-2⁻CD135^hiCD127⁺), CD127⁻ LMPP (Lin⁻Sca1^hic-Kit^hiThy1.2⁻ST-2⁻CD135^hiCD127⁻) and CD127⁺ LMPP (Lin⁻Sca1^hic-Kit^hiThy1.2⁻ST-2⁻CD135^hiCD127⁺) cells compared with Yy1^+/+ mice; however, the absolute numbers were significantly reduced in CD127⁺ LMPP, CD127⁻ LMPP and CLP populations (Fig. 1H,I). Thus, YY1 is required for maintaining CLP and LMPP pools in adult bone marrow.

Our previous data show that YY1 stimulates c-Kit cell surface expression and SCF/c-Kit signaling in adult HSCs (Lu et al., 2018). To assess whether YY1 promotes early T cell development by regulating SCF/c-Kit signaling, we next compared c-Kit cell surface expression in Yy1^−/− ETP and DN2a with Yy1^+/+ ETP and DN2a cells respectively. Surprisingly, Yy1^−/− ETP and DN2a cells had similar c-Kit median fluorescent intensity (MFI) compared with Yy1^+/+ cells (Fig. S2B), whereas bone marrow Yy1^−/− Lin⁻Scal⁺Kit⁺ (LSK) cells had decreased c-Kit MFI compared with Yy1^+/+ LSKs (Fig. S2A). Thus, YY1 deficiency did not impact c-Kit cell surface expression in ETP and DN2a cells. As Notch signaling plays a pivotal role in early T cell development (Radtke et al., 2004; Maillard et al., 2005), and YY1 may regulate Notch1 expression or function by associating with the high molecular weight Notch complex in vitro (Yeh et al., 2003), we next assessed Notch1 expression and signaling in Yy1^−/− thymocytes. Yy1^−/− mice had a significant reduction in percentages of Notch1⁺ DN1 (Lin⁻CD44⁺CD25⁻), DN2 (Lin⁻CD44⁺CD25⁺) and DN4 (Lin⁻CD44⁻CD25⁻) cells compared with Yy1^+/+ mice (Fig. 2A), and the MFI of Notch1 was significantly reduced in YY1-deficient DN2 (Lin⁻CD44⁺CD25⁺), DN3 (Lin⁻CD44⁻CD25⁺) and DN4 (Lin⁻CD44⁻CD25⁻) cells (Fig. 2B). There was a more than 80% reduction of Notch1 mRNA and protein expression in Yy1^−/− thymocytes compared with Yy1^+/+ mice (Fig. 2C,D). In addition, Notch target gene Hes1 was downregulated in Yy1^−/− thymocytes (Fig. 2D). We analyzed YY1 occupancy at the Notch1 locus by evaluating previously published ChIP-seq data (Cuddapah et al., 2011; Barrett et al., 2013; Weintraub et al., 2017; Davis et al., 2018). We detected a strong occupancy of YY1 at the enhancer site (−62.8 from the transcription start site) of the Notch1 locus in human CD4⁺ T cells, patient-derived xenograft (PDX) cells from human T-cell acute lymphoid leukemia (T-ALL) patients and the T cell line Jurkat (Fig. 2E). We concluded that YY1-deficient T cells had reduced Notch1 expression and downregulation of a Notch target gene.

Inducible inactivation of Notch1 results in an early blockage in intrathymic T cell lineage differentiation at the DN1 stage (Radtke et al., 1999). Although Notch1 expression and target genes were significantly downregulated in Yy1^f/f Mx1-Cre mice (Fig. 2), YY1-deficient thymocytes had a T cell developmental blockage at the DN3 stage (Fig. 1D). The discrepancy may be caused by a long-term versus short-term deletion of YY1 impacting differently on T cell development. Our Yy1^f/f Mx1-Cre mice were assessed no longer than 7 days after the last dose of pI-pC injection due to the lethality beyond this time point, and thymocyte development may still be under the influence of pI-pC (Leonard et al., 1969). To address these questions, we crossed Yy1^f/f mice with Vav-Cre mice to generate heterozygous Yy1^f/+Vav-Cre mice. The Vav promoter drives Cre recombinase expression specifically in fetal liver hematopoietic cells starting at embryonic day (E) 11.5 (Georgiades et al., 2002). Yy1^f/+ Vav-Cre mice were then subsequently bred with Yy1^f/f mice to generate homozygous Yy1^f/f Vav-Cre mice. Although Yy1^f/f Vav-Cre mice died at birth, they survived at E14.5 of fetal development stage (Lu et al., 2018). This allowed us to use a powerful Lin⁻ fetal liver/OP9-DL1 co-culturing system to assess YY1 regulation of DN T cell development ex vivo. In the Lin⁻ fetal liver/OP9-DL1 co-culture system, over 90% of Lin⁻ fetal liver cells start to have DN1 (Lin⁻CD44⁺CD25⁻) cell phenotype after 3 days of co-culturing with OP9-DL1 feeder cells. After day 10 of co-culturing, over 80% of cells were in the DN3 stage, and at day 14 of co-culturing, Lin⁻ fetal liver cells had progressively differentiated into DN4 stage and DP T cells (Fig. S3). Yy1^f/f Vav-Cre (Yy1^−/−) and Yy1^f/f (Yy1^+/+) fetal liver cells were isolated at E14.5, and Lin⁻ cells were purified by magnetic beads and then plated on OP9-DL1 feeder cells. At day 10 of co-culturing, over 90% of cells from Yy1^−/− mice remained in the DN1 (Lin⁻CD44⁻CD25⁻) stage and there was a significant reduction in percentage of Yy1^−/− cells proceeding into the DN2 (Lin⁻CD44⁺CD25⁺) and DN3 (Lin⁻CD44⁻CD25⁺) stages compared with the wild-type cells (Fig. 3A,B). There was also a significant reduction of absolute numbers of DN2, DN3 and DN4 cells in the co-culture generated from Yy1^−/− mice compared with co-culture generated from Yy1^+/+ mice (Fig. 3C). Although there was no statistical difference between numbers of DN1 cells in Yy1^−/− versus Yy1^+/+ co-cultures, this was due to a significant increase of DN1 percentage caused by YY1 deficiency. Thus, YY1 deficiency causes DN1 blockage ex vivo and reduction of DN cell numbers.

To assess whether YY1 is crucial for DN1 T cell survival, we performed the annexin V/DAPI apoptosis assay on DN1 cells at day 10 of Lin⁻ fetal liver/OP9-DL1 co-culturing (Fig. 3D,E). There was ∼10 fold increase in percentage of early apoptotic DN1 cells (annexin V⁺DAPI⁺) in the Yy1^−/− group compared with the Yy1^+/+ group, and there was a significant reduction in percentage of non-apoptotic cells (annexin V⁻DAPI⁻) in Yy1^−/− compared with Yy1^+/+ (Fig. 3D,E). Consistently, at day 3 of co-culturing, there was an increase in the percentage of non-viable cells in Yy1^−/− DN1 compared with Yy1^+/+ DN1 (Fig. S4). Thus, we concluded that YY1 plays a crucial role in DN1 transition and DN1 cell survival.

Our previous studies have shown that YY1 PcG function/REPO domain is crucial for early B cell development (Pan et al., 2013). Therefore, we next assessed whether YY1 functions in early T cell development are dependent on its PcG function/REPO domain by using a unique YY1 mutant, YY1ΔREPO, which contains all YY1 functions except its PcG function (Wilkinson et al., 2006). Bone marrow cells from Yy1^f/f Mx1-Cre mice were transduced retrovirally with MigR1-FlagYY1, MigR1-FlagYY1ΔREPO or MigR1 vector and transplanted into lethally irradiated CD45.1⁺ mice. In addition, Mx1-Cre bone marrow cells infected with MigR1 vector were used as the wild-type control and were transplanted into CD45.1⁺ recipient mice (Fig. 4A). Before bone marrow transplantation (BMT), viral transduction efficiency was assessed 48 h post-infection (Fig. 4B). MigR1-FlagYY1, MigR1-FlagYY1ΔREPO, or MigR1 had a similar infection rate (% GFP⁺) and a similar GFP MFI (Fig. 4B). Upon pI-pC injection 4 weeks post-transplantation, endogenous YY1 was deleted, and YY1 function beyond this stage was dependent upon the exogenous YY1 constructs transduced into the cells. Western blot analysis revealed that over 90% of endogenous YY1 was deleted in GFP⁺ peripheral blood lymphocytes. Exogenous YY1 and YY1ΔREPO were expressed at similar levels 7 days post-pI-pC injection (Fig. 4C). Peripheral blood or bone marrow cells were analyzed by gating CD45.2⁺ donor-derived cells and further gating for GFP⁺ cells within T cell (Thy1.2⁺CD19⁻), B cell (Thy1.2⁻CD19⁺), monocyte (Mac1⁺Gr1⁻), and neutrophil (Mac1⁺Gr1⁺) populations (Fig. S5). Before the endogenous YY1 deletion 4 weeks post-transplantation, peripheral blood from all cohorts ­ Yy1^+/++MigR1, Yy1^−/−+MigR1, Yy1^−/−+MigR1-Flag YY1 and Yy1^−/−+MigR1-Flag YY1ΔREPO – had an equivalent percentage of CD45.2⁺ GFP⁺ cell within T cell, B cell, neutrophil and monocyte populations. After YY1 endogenous deletion, at 20-weeks post-transplantation, Yy1^−/−+YY1ΔREPO mice had a significantly higher percentage of donor-derived GFP⁺ cells within neutrophil and monocyte populations compared with Yy1^−/−+MigR1, and was at a similar level compared with the wild-type YY1 rescued group (Yy1^−/−+YY1) and wild-type control (Yy1^+/++MigR1) (Fig. 4D). However, Yy1^−/−+YY1ΔREPO mice had a significantly reduced percentage of donor-derived GFP⁺ cells in T and B lymphocytes compared with Yy1^+/++MigR1 and Yy1^−/−+YY1 (Fig. 4D). Similar results were also observed in lineage evaluation of bone marrow cells that YY1ΔREPO was able to rescue monocytes and neutrophils, but not T and B cell lineages (Fig. 4E). Our results support that YY1 PcG function/REPO domain is selectively required for B and T cell development and is unnecessary for myeloid lineage.

To assess YY1 PcG function/REPO domain in early T cell development, thymocytes from Yy1^+/++MigR1, Yy1^−/−+MigR1, Yy1^−/−+MigR1-Flag YY1 and Yy1^−/−+MigR1 Flag-YY1ΔREPO mice were assessed at 20 weeks post-BMT (Fig. 4A). Although the percentage of CD45.2⁺GFP⁺ cells within total thymocytes is similar among Yy1^+/++MigR1 and Yy1^−/−+YY1ΔREPO, the absolute number of donor-derived GFP⁺ thymocytes was significantly reduced in Yy1^−/−+YY1ΔREPO mice compared with Yy1^+/++MigR1 or Yy1^−/−+YY1 (Fig. 5A). Similarly, Yy1^−/−+YY1ΔREPO mice had a significantly decreased number of DN1 (Lin⁻CD44⁺CD25⁻), c-Kit⁺ DN1 (Lin⁻c-Kit⁺CD44⁺CD25⁻), DN2 (Lin⁻CD44⁺CD25⁺), DN3 (Lin⁻CD44⁻CD25⁺) and DN4 (Lin⁻CD44⁻CD25⁻) cells compared with Yy1^+/++MigR1 or Yy1^−/−+YY1, whereas percentages of DN cells remained at a similar level compared with the wild-type control (Fig. 5B,C). Consistent with our findings in the fetal liver co-culturing system (Fig. 3), there was a significant increase in percentage of DN1 and c-Kit⁺ DN1 in donor-derived GFP⁺ cells of the Yy1^−/−+MigR1 group, indicating that YY1 is required for DN1 transition. In contrast, there was no difference in percentages of DN1 or c-Kit⁺ DN1 in the donor-derived GFP⁺ cell population among Yy1^+/++MigR1, Yy1^−/−+YY1 and Yy1^−/−+YY1ΔREPO groups (Fig. 5C), indicating that YY1 PcG function/REPO domain was not required for YY1 function in promoting DN1 T cell transition. Consistently, MFI of Notch1 was similar among Yy1^+/++MigR1, Yy1^−/−+YY1 and Yy1^−/−+YY1ΔREPO groups (Fig. 5D), thus Yy1^+/++MigR1, Yy1^−/−+YY1 and Yy1^−/−+YY1ΔREPO groups had similar Notch1 cell surface expressions in DN1, DN2 and DN3 cells. We concluded that YY1 PcG function/REPO domain is required for maintaining proper T cell numbers during development; however, it is not required for DN1-to-DN2 transition or Notch1 expression.

Reduction of early T cell numbers can be caused by decreased T cell survival or by interference with cell proliferation. Because apoptotic thymocytes are often undetected in vivo due to clearance by phagocytosis (Schlegel et al., 2000), we next used the Lin⁻ fetal liver/OP9-DL1 co-culturing system to assess whether YY1 PcG function/REPO domain was needed for T cell survival ex vivo. Yy1^f/f Vav-Cre (Yy1^−/−) fetal liver cells were isolated at E14.5 and Lin⁻ liver cells were then purified using magnetic beads. Yy1^−/− Lin⁻ fetal liver cells were retrovirally transduced with MigR1, MigR1-FlagYY1 and MigR1-FlagYY1ΔREPO. And as a wild-type control, Yy1^f/f (Yy1^+/+) Lin⁻ fetal liver cells were transduced with the MigR1 vector (Fig. 6A). At day 0 of co-culturing, the same number of Yy1^−/−+MigR1, Yy1^−/−+YY1, Yy1^−/−+YY1ΔREPO and Yy1^+/++MigR1 Lin⁻ fetal liver cells were plated to the OP9-DL1 covered plates (Fig. 6C). At day 3 of co-culturing, a similar percentage (∼80-90%) of GFP⁺ Lin⁻ fetal liver cells start to have DN1 (Lin⁻CD44⁺CD25⁻) cell phenotype in Yy1^−/−+MigR1, Yy1^−/−+YY1, Yy1^−/−+YY1ΔREPO and Yy1^+/++MigR1 groups (Fig. 6B,D). However, Yy1^−/−+MigR1 and Yy1^−/−+YY1ΔREPO samples had a significant decrease of absolute number of GFP⁺ DN1 cells compared with Yy1^−/−+YY1 and Yy1^+/++MigR1 (Fig. 6E). To assess the impact of YY1 PcG function/REPO domain on DN1 cell survival, we evaluated early and late apoptosis using annexin V/DAPI staining of DN1 cells at day 3 of fetal liver/OP9-DL1 co-culturing (Fig. 6F). There was a 5- to 15-fold increase in percentage of late apoptotic DN1 (Lin⁻CD44⁺CD25⁻) cells (annexin V⁺DAPI⁺) in the Yy1^−/−+YY1ΔREPO group compared with the Yy1^+/++MigR1 or Yy1^−/−+YY1 group. And there was a significant reduction in the percentage of non-apoptotic cells (annexin V⁻DAPI⁻) in Yy1^−/−+MigR1 and Yy1^−/−+YY1ΔREPO groups compared with Yy1^+/++MigR1 and Yy1^−/−+YY1 (Fig. 6G). Thus, we concluded that YY1 PcG function/REPO domain plays a crucial role in DN1 T cell survival.

In addition to interfering with cell survival, reduction of DN1 cell number could also be caused by inhibition of cell proliferation. Next, we assessed how YY1 PcG function/REPO domain impacted cell cycle progression. There was a significant reduction of percentage of G0 (Ki67⁻DAPI⁻) and G1 (Ki67⁺DAPI⁻) cells in Yy1^−/−+MigR1 and Yy1^−/−+YY1ΔREPO groups compared with Yy1^+/++MigR1 and Yy1^−/−+YY1 groups. There was a significant increase in percentage of S-G2-M phase cells (Ki67⁺ DAPI⁺) in Yy1^−/−+MigR1 and Yy1^−/−+YY1ΔREPO groups compared with Yy1^+/++MigR1 and Yy1^−/−+YY1 groups (Fig. 6H,I). Thus, the reduction of DN1 cell numbers in the Yy1^−/−+MigR1 and Yy1^−/−+YY1ΔREPO groups was not caused by inhibition of cell proliferation. Instead, DN1 cells in Yy1^−/−+MigR1 and Yy1^−/−+YY1ΔREPO group had increased proliferation, which was probably caused by a compensatory mechanism due to the severe lymphopenia.

As Notch1 is an essential survival factor for DN T cells (Ciofani and Zúñiga-Pflücker, 2005), we next assessed whether YY1-mediated early T cell survival was dependent on its regulation of Notch1 expression. Yy1^−/− Lin⁻ fetal liver cells were isolated and retrovirally transduced with retroviral vector expressing Notch1 intracellular domain (N1-ICD) (MigR1-N1-ICD) (Pear et al., 1996) and vector control (MigR1) (Fig. S7B). At day 0 of co-culturing, the same number of Yy1^−/−+MigR1 and Yy1^−/−+N1-ICD were plated to the OP9 covered plates (Fig. S7A). By expressing the MigR1-N1-ICD in the Lin⁻ fetal liver/OP9 co-culture, ∼28% of wild-type Lin⁻ fetal liver cells differentiated to DN3 T cells at day 3 of co-culturing (Fig. S7C). Although at day 3 of co-culturing there was a small but significant increase of DN2 percentage in Yy1^−/−+N1-ICD co-culture compared with Yy1^−/−+MigR1, there were no DN3 cells detected (Fig. S7D,E). Surprisingly, there was a significant increase in the percentage of early apoptotic DN1 (Lin⁻CD44⁺CD25⁻) cells in Yy1^−/−+N1-ICD co-culture compared with Yy1^−/−+MigR1 (Fig. S7F,G). Our results indicate that YY1 promotes DN1 transition by a multi-component mechanism that is not solely dependent on its regulation of Notch1 expression. It is unlikely that YY1 promotes early T cell survival by maintaining the Notch1 expression level. Physiologically, the expression level of N1-ICD is tightly controlled. Overexpressed N1-ICD can interact with different transcriptional cofactors from multiple signaling pathways such as SMADs, NFκB and HIF1α (Kopan and Ilagan, 2009). However, under physiological conditions, it is unlikely that there will be many free N1-ICD available to associate with other factors in the nucleus. The survival defect we observed here could also merely be the result of excessive N1-ICD expression (Fig. S7B).

Multiple transcription factors and epigenetic regulators function as intrinsic regulators of T cell development. YY1, a multifunctional transcription factor and PcG protein, plays crucial roles in lymphocyte development (Liu et al., 2007; Zaprazna and Atchison, 2012; Chen et al., 2016; Kleiman et al., 2016). When YY1 was conditionally deleted using CD2-Cre and Lck-Cre mice, there was a DN3-to-DN4 developmental blockage with a skewed TCRα repertoire or a blockage at the DP stage. In addition, YY1 deficiency results in decreased survival of thymocytes by downregulating p53 expression (Chen et al., 2016). Clearly, YY1 functions to control T cell development and survival via multiple mechanisms and would not be expected to be solely involved in the regulation of p53. The CD2 transgene starts expressing between the DN1 and DN3 stages and the expression is universal in all thymocytes by the DN4 stage (de Boer et al., 2003). In Yy1^f/f Lck-Cre mice, YY1 protein is not substantially depleted until the DP stage (Chen et al., 2016). Thus, neither system is suitable for assessing the impact of YY1 on T cell development before the DN1 stage. Our previous results have shown that YY1-deficient mice have decreased HSC number and function (Lu et al., 2018). Thus, YY1 regulation of T cell development may start at a level upstream of the DN1 stage, and YY1-deficient ETPs may fail to differentiate into the downstream T cell lineages.

In this study, we used Mx1-Cre and Vav-Cre to delete YY1 before the DN1 stage. Our results demonstrate that YY1 is required for DN1-to-DN2 transition and promotes Notch1 expression and signaling in thymocytes (Figs 1 and 2). By introducing a YY1 mutant (YY1ΔREPO) into a YY1-null genetic background, we were able to determine YY1 PcG function/REPO domain in different hematopoietic compartments. We found that YY1 PcG function/REPO domain is required for T and B cell development. However, YY1 control of the myeloid lineage is insensitive to YY1 PcG domain deletion (Fig. 4). Although YY1 PcG function/REPO domain is required for early T cell survival (Fig. 6), YY1 regulation of DN1-to-DN2 transition and Notch signaling are independent of its PcG function/REPO domain (Figs 5 and 7).

YY1 is a ubiquitously expressed transcription factor with a high expression level in HSCs, multipotent progenitors, myeloid progenitors, CLPs and lymphocytes (https://gexc.riken.jp/models/3/genes/Yy1), and mechanisms underlying YY1 functions in adult hematopoietic compartments appear to be distinct. In HSCs, YY1 overlaps with other hematopoietic regulators, including GATA-2 and SCL (Tal1) at the Kit locus and regulates c-Kit cell surface expression and signaling. However, our results demonstrate that YY1-deficient ETP had normal c-Kit cell surface expression, and YY1 function in early T cell precursors is independent of its regulation of c-Kit expression (Fig. S2). Instead, YY1 may promote early T cell development by promoting Notch1 expression and its target genes (Fig. 2). Previous studies have shown that endogenous YY1 associates with the Notch1 receptor intracellular domain in Jurkat cells, and YY1 may promote Notch signaling by associating with a high molecular weight Notch complex (Yeh et al., 2003). By evaluating published ChIP-seq data, we were able to detect a strong YY1 binding site at the Notch1 enhancer in T cell lines and primary T cell samples (Fig. 2E). YY1 may transcriptionally regulate Notch1 expression by binding at the Notch1 locus to promote its expression and signaling. Although conditional inactivation of Notch1 was reported to cause increased numbers of B lineage cells (Radtke et al., 1999), we did not observe this phenotype in YY1 conditional knockout mice, instead YY1 deficiency caused a significant reduction of mature B cells in both bone marrow and peripheral blood (Fig. 4D,E). This discrepancy was due to the requirement of YY1 for Ig locus contraction needed for distal variable gene rearrangement of Ig heavy chain (Liu et al., 2007) and for all stages of B cell development (Kleiman et al., 2016). Due to the heterogeneity of DN1 (Lin⁻CD44⁺CD25⁻) cells, ETPs (cKit⁺Lin⁻CD44⁺CD25⁻) account for less than 10% of the total DN1 population. Thus, further evaluation of YY1 regulation of Notch1 expression and signaling in lymphoid progenitors such as ETPs and LMPPs will be interesting for future studies.

Although YY1 PcG function/REPO domain is required for proper Ig rearrangement and early B cell development (Pan et al., 2013), YY1 function in HSC self-renewal and quiescence is independent of its PcG function. Our study demonstrated that the requirement for YY1 PcG function/REPO domain is lineage-specific. YY1 PcG function/REPO domain is required for B and T lymphocyte, but not for myeloid cell development (Fig. 4D,E). The ability of YY1ΔREPO to functionally rescue monocyte and neutrophil development provides evidence that YY1ΔREPO was producing appropriate amounts of functional protein. Furthermore, its failure to rescue B and T cells was not due to a low protein expression or protein misfolding. It is intriguing that YY1 regulates early T cell development by both PcG development and PcG-independent pathways. Although YY1 PcG function/REPO domain was required for DN cell survival, YY1 mediated DN1-to-DN2 transition and Notch1 expression were independent of its PcG function/REPO domain (Fig. 7). As one of a few mammalian PcG proteins with intrinsic sequence-specific DNA binding activity (Atchison et al., 2003; Srinivasan and Atchison, 2004), YY1 recruits other PcG proteins, including EZH2, to specific DNA sequences, leading to catalysis of H3K27me3 at the recruitment site (Srinivasan and Atchison, 2004; Wilkinson et al., 2006). In Drosophila, Polycomb Repressive Complex (PRC) recruitment depends on Pleiohomeotic Repressive Complex (PhoRC) (Levine et al., 2002), but no central recruitment mechanism or specific DNA elements have been identified in mammalian cells (Bauer et al., 2016). Many studies support an important role for YY1 in PRC recruitment (Wilkinson et al., 2006; Ku et al., 2008; Woo et al., 2010; Basu et al., 2014). However, YY1 and Polycomb target genes do not have a high degree of overlap in certain cell types (Xi et al., 2007) and YY1 also has many PcG-independent functions in the cell (Lu et al., 2018). Other factors such as Gata1, Hic1, Rest and Runx1, and long non-coding RNA, have been reported to recruit the PRC complex to specific DNA sites (Kaneko et al., 2010; Boulay et al., 2012; Dietrich et al., 2012; Ross et al., 2012; Yu et al., 2012). Thus, YY1-mediated PRC recruitment may not be a general mechanism in all mammalian cells. Distinct Polycomb sub-complexes might use different recruiting mechanisms, thus binding different collections of genes in the lineage-dependent and context-dependent manner; the non-canonical PRC complex, may use YY1-mediated PRC recruitment in pre-B and early T cells, but not in myeloid cells. Different hematopoietic lineages may use a distinct mechanism of PRC recruitment depending on interrelated DNA binding context, histone recognition and protein-protein interactions. Our study provides an important model/tool for further dissecting the lineage-dependent and context-dependent requirement of PRC recruitment in the adult hematopoietic system.

Yy1^f/f mice (Affar et al., 2006; Liu et al., 2007) in which the Yy1 promoter region and exon 1 are flanked by loxP sites, were crossed to Mx1-Cre mice to generate heterozygous Yy1^f/+ Mx1-Cre mice. Yy1^f/+ Mx1-Cre mice were then subsequently bred with Yy1^f/f mice to generate homozygous Yy1^f/f Mx1-Cre mice. Yy1^f/f mice were bred with Vav-Cre mice to generate Yy1^f/+Vav-Cre mice. Then, Yy1^f/+Vav-Cre mice were crossed with Yy1^f/f mice, and Yy1^f/f Vav-Cre fetuses were harvested at E14.5. To induce the expression of Cre recombinase, 6- to 8-week-old Yy1^f/f Mx1-Cre mice were injected with 100 μg of pI-pC (GE Healthcare Life Science) every other day for four doses. Approximately equal portions of mice of both genders were used, and aggregated data are presented because gender-specific differences were not found. All experiments described in this manuscript were performed 7 days after the last injection of pI-pC unless stated otherwise. All experiments involving mice were approved by the Institutional Laboratory Animal Care and Use Committee of the University of Wisconsin-Madison and conform to the appropriate regulatory standards.

Total RNA was isolated using the RNeasy Mini Kit (Qiagen) according to the manufacturer's protocol. Complementary DNA (cDNA) was synthesized with the SuperScript III First-Strand Synthesis System Kit (Invitrogen™) with oligo (dt) primers and was subjected to the reverse transcriptase-polymerase chain reaction (RT-PCR) assay. Yy1, Notch1 and Hes1 transcripts were detected by Roche Lightcycler96 with the cycle setting at 95°C (10 min), 95°C (10 s), 60°C (10 s) and 72°C (20 s) for a total of 40 cycles. 18s was used as an internal control for normalization, and quantification was determined by the delta-delta cycle threshold (ΔΔCt) method.

The primers used were: Yy1 forward TCAGACCCTAAGCAACTGGCAGAA, reverse TTGAGCTCTCAACGAACGCTTTGC; Notch1 forward AGATCGACAACCGGCAATGT, reverse CCCACAGCCCACAAAGAAC (Rodríguez-Caparrós et al., 2019); Hes1 forward CGGCATTCCAAGCTAGAGAAGG, reverse GGTAGGTCATGGCGTTGATCTG (Espinosa et al., 2010); 18s forward CGCCGCTAGAGGTGAAATTCT, reverse CGAACCTCCGACTTTCGTTCT.

Antibodies detecting YY1 (EPR4652; Abcam; 1:5000), cleaved Notch1 (D3B8; Cell Signaling Technology; 1:1000), Notch1 (D6F11; Cell Signaling Technology; 1:1000) and β-actin (A1978; Sigma-Aldrich; 1:5000) proteins were used according to the manufacturer's recommendation. The intensity of specific protein bands was quantified using Adobe Photoshop software.

Fetal liver cells were isolated from Yy1^f/f and Yy1^f/f Vav-Cre fetuses at E14.5 of fetal development. Lin⁻ fetal liver cells were then isolated using the EasySep™ magnet system (Stemcell Technologies) with antibodies against CD3 (145-2C11), CD4 (RM4-5), CD5 (53-7.3), CD8 (53-6.7), B220 (RA3-6B2), CD19 (eBio1D3), Ter 119 (TER-119) and Gr1 (RB6-8C5) (eBioscience). All antibodies were diluted at 1:200 ratio. At 24 h after Lin⁻ fetal liver cells were retrovirally infected, cells were plated on OP9-DL1 (Fu et al., 2017) or OP9 (Choi et al., 2011) feeder cells with MEM, fetal bovine serum, penicillin/streptomycin, L-glutamine, Flt-3L (PeproTech) and IL7 (PeproTech). Cells were transferred to new OP9-DL1 or OP9 feeder cells every 72 h. At day 3 and day 10 after Lin⁻ fetal liver cell plating, flow analysis was conducted. DN cells were defined by lineage markers CD44 and CD25. Lineage markers included CD4, CD8, B220, Ter 119, Gr1, Mac1 and NK1.1.

For retroviral bone marrow transplantation, bone marrow cells were harvested from 6- to 8-week-old C57BL/6 (CD45.2⁺) mice 4 days after receiving 5 mg 5-fluorouracil IV (5-FU, APP Pharmaceuticals) and transduced as previously described (Pan et al., 2012, 2013).

Directly conjugated or biotin-conjugated antibodies specific for the following surface antigens were purchased from eBioscience: CD3 (145-2C11), CD4 (RM4-5), CD8 (53-6.7), B220 (RA3-6B2), Ter 119 (TER-119), Gr1 (RB6-8C5), IgM (eB121-15F9), CD19 (eBio1D3), IL7Ra (A7R34), CD45.2 (104), CD45.1 (A20), Sca1 (D7), c-Kit (2B8), Mac1 (M1/70), Thy1.2 (53-2.1), ST-2 (RMST2-2), CD25 (PC61.5) and CD44 (1M7). Nk1.1 (PK136), CD127 (A7R34), CD135 (A2F10) and Notch1 (HMN1-12) were purchased from BioLegend. All antibodies were diluted at 1:200 ratio, except Thy 1.2 (1:800), CD127 (1:50) and CD45.1 (1:400). Ghostdye™ Violet510 (Tonbo Bioscience) or DAPI (Thermo Fisher Scientific) were diluted at 1:1000 ratio and were used to exclude non-viable cells. Acquisitions were performed on an LSR II or LSR II Fortessa (BD Biosciences). Data were analyzed using BDFlowJo v10.0.7 software.

Thymocytes from the OP9-DL1 or OP9 co-culturing system were harvested and were stained for lineage markers, CD25 and CD44 expression. Then, cells were resuspended in the binding buffer solution with PE anti-annexin V (559763; BD Biosciences; 1:40). Next, cells were resuspended in DAPI solution (Thermo Fisher Scientific, 1:1000). Samples were analyzed using the BD LSRII Fortessa and results were analyzed using FlowJo v10.0.7.

Cells from day 10 of the OP9-DL1 co-culture system were fixed with 4% paraformaldehyde, permeabilized with 0.1% saponin in PBS, and stained with PE-conjugated Ki67 (BD Biosciences, 1:40) and DAPI solution (Thermo Fisher Scientific, 1:250) in addition to DN1, DN2, DN3 and DN4 markers as described above.

ChIP-seq data in Jurkat cell line (GSE99521), PDX T-ALL cell (GSE145549), and human CD4⁺ T cell (GSE25674) were obtained from the Gene Expression Omnibus portal. LiftOver was used to convert the aligned reads from hg18 to hg19.

All statistical analyses were conducted using GraphPad Prism v7.04. The two-tailed unpaired Student's t-test was used to determine statistical significance between Mx1-Cre and Yy1^f/fMx1-Cre groups. Differences between the in vitro and in vivo rescue system groups were determined using a two-tailed unpaired Student's t-test when comparing two groups, or a one-way ANOVA followed by Tukey's post-hoc test when comparing three or more groups. Two-way ANOVA followed by Tukey's post hoc test was used to compare the changes in the proportion of cells undergoing apoptosis, as measured by annexin V and DAPI flow cytometric analysis, and to compare cells in different phases of the cell cycle, as measured by Ki67 and DAPI flow cytometry analysis. P-values ≤0.05 were considered statistically significant.

We thank the University of Wisconsin Carbone Comprehensive Cancer Center (UWCCC) for use of its Shared Services (Flow Cytometry Core and Cancer Informatics Shared Resources), which are supported by UWCCC grant P30 CA014520, to complete this research.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/148/7/dev197319/