LMO2 is essential to maintain the ability of progenitors to differentiate into T-cell lineage in mice

Notch signaling primarily determines T-cell fate. However, the molecular mechanisms underlying the maintenance of T-lineage potential in pre-thymic progenitors remain unclear. Here, we established two murine Ebf1-deficient pro-B cell lines, with and without T-lineage potential. The latter expressed lower levels of Lmo2; their potential was restored via ectopic expression of Lmo2. Conversely, the CRISPR/Cas9-mediated deletion of Lmo2 resulted in the loss of the T-lineage potential. Introduction of Bcl2 rescued massive cell death of Notch-stimulated pro-B cells without efficient LMO2-driven Bcl11a expression but was not sufficient to retain their T-lineage potential. Pro-B cells without T-lineage potential failed to activate Tcf7 due to DNA methylation; Tcf7 transduction restored this capacity. Moreover, direct binding of LMO2 to the Bcl11a and Tcf7 loci was observed. Altogether, our results highlight LMO2 as a crucial player in the survival and maintenance of T-lineage potential in T-cell progenitors via the regulation of the expression of Bcl11a and Tcf7.


Introduction
Hematopoietic stem cells (HSCs) have the ability to self-renew and differentiate into all blood cell types. HSCs begin to differentiate into various lineage cells, gradually losing their potential to become their descendants, hematopoietic progenitor cells (HPCs), and then differentiate into mature blood cells (Doulatov et al., 2012;Kosan and Godmann, 2016). This sequence of processes has long been imagined as a ball rolling down a valley track (Goldberg et al., 2007). Cell fate decisions in hematopoietic cells are controlled by continuous interactions between environmental influences and intrinsic cellular mechanisms, such as transcription factor networks and epigenetic regulation .
In the process of specification from HPCs, only the T-cell lineage requires a specialized environment of the thymus, where the immigrant cells receive Notch signaling induced by the interaction of Notch1 on the immigrant cells and a Notch ligand, delta-like 4 (Dll4), on the thymic epithelium, which determines their fate to the T-cell lineage (Radtke et al., 1999;Hozumi et al., 2008;Koch et al., 2008). In addition, several transcription factors contribute to the acquisition of T-cell identity, including TCF1 (encoded by Tcf7), basic helix-loop-helix (bHLH) factors, E2A and HEB, PU.1, GATA3, Bcl11b, and Runx family members, all of which are essential for T-cell development in the thymus Hosokawa et al., 2021a;Hosokawa et al., 2021b).
Among them, one of the earliest Notch-activated genes, Tcf7, plays a critical role in Notch-mediated initiation of the T-lineage program (Weber et al., 2011;Johnson et al., 2018). The coordinated action of these environmental and intrinsic factors completes the commitment to T-cell lineage in double-negative (DN; CD4 À CD8 À ) thymocytes at the transition from the DN2 to DN3 stages (Yui and Rothenberg, 2014;Hosokawa and Rothenberg, 2021). However, it remains unclear how the potential of the thymic immigrant cells is maintained to initiate their differentiation program toward the T-cell lineage.
LMO2 gene is located near the breakpoint of the chromosomal translocation t(11;14) (p13;q11) in human T-ALL (Boehm et al., 1991;Royer-Pokora et al., 1991), and LMO2 has been thought to function as a bridging factor in large transcriptional complexes with several DNA-binding and adaptor proteins (TAL1/SCL, E2A, GATA1, and Ldb1) (Wadman et al., 1997;Grütz et al., 1998;El Omari et al., 2013;Layer et al., 2016). Although deletion of Lmo2 in mice causes embryonic lethality due to embryonic erythropoiesis deficiency around 10 days post-fertilization (Yamada et al., 1998), conditional disruption of Lmo2 in T-lineage committed pro-T stages results in normal T-cell development in the thymus , indicating that LMO2 is dispensable for T-cell development after T-lineage commitment. However, it remains unclear whether LMO2 plays an important role before T-lineage commitment, including the maintenance of T-lineage potential in pre-thymic progenitors.
It is well known that the recent gene therapy trials using retrovirus-mediated introduction of the common cytokine receptor gamma chain (gc; CD132) in X-linked severe combined immunodeficiency (X-SCID) patients resulted in the development of T-cell leukemia due to retroviral insertion at the LMO2 locus (Hacein-Bey-Abina et al., 2003;Hacein-Bey-Abina et al., 2008;Howe et al., 2008). Moreover, transgenic overexpression of Lmo2 in various murine tissues only results in T-cell leukemia, indicating its involvement exclusively in T-cell malignancies when abnormally expressed (Neale et al., 1995). On the other hand, Lmo2 is more highly expressed in HSCs and HPCs than in mature blood cells (Yoshida et al., 2019) and is known to induce reprogramming of HSCs from differentiated blood cells (Riddell et al., 2014) and fibroblasts (Batta et al., 2014;Vereide et al., 2014), suggesting that LMO2 actively contributes to the maintenance of the undifferentiated state of HSCs. Consistently, aberrant induction of Lmo2 in thymocytes generates self-renewing cells that retain the capacity for T-cell differentiation Cleveland et al., 2013). Thus, it is not clear why Lmo2 overexpression causes only T-cell malignancies and how LMO2 contributes to the maintenance of an undifferentiated state.
In attempts to establish HPC lines with T-cell differentiation potential, it has been found that pro-B cells without transcription factors essential for early B-cell development, including Pax5, Ebf1, and E2A (encoded by Tcf3), can be maintained in vitro with an HPC-like phenotype, retaining T-cell potential on OP9 stromal cells in the presence of IL-7 Rolink et al., 1999;Ikawa et al., 2004;Pongubala et al., 2008). In this study, based on their protocol, we established pro-B cell lines, called pro-B(+) cells, that retain T-cell potential on OP9 stromal cells. In contrast, proliferating pro-B cells, named pro-B(À) cells, were also obtained in our originally established thymic stromal cells, TD7, which die abruptly immediately following activation of Dll4-mediated Notch signaling in vitro. Thereafter, we identified Lmo2 as a gene whose expression is downregulated in pro-B(À) cells compared to pro-B(+) cells and found that the forced expression of Lmo2 was sufficient for the pro-B(À) cells to reacquire the ability to differentiate into T-cell lineage by Notch signaling. Furthermore, LMO2 ensures survival of pro-B cells through the activation of the Bcl11a/Bcl2 pathway and contributes to maintaining the accessibility of the Tcf7 locus, which is one of the earliest Notch downstream targets in thymic immigrant cells. These epigenetic alterations could be mediated by direct binding of the transcriptional complex, including LMO2, to the target loci. Together with loss-of-function experiments, we demonstrate that LMO2 is significant for maintaining T-cell progenitors for the progression of the T-cell differentiation program initiated by Notch signaling.

Results
LMO2 has a crucial role in the maintenance of T-cell differentiation capacity in Ebf1-deficient pro-B cells Using different stromal cells, OP9 and TD7, with fetal liver progenitor cells from Ebf1-deficient mice, we established two types of pro-B cell lines, with and without the ability to differentiate into T-cell lineages designated pro-B(+) and pro-B(À) cells, respectively ( Figure 1-figure supplement 1A,B). Both the Ebf1-deficient pro-B lines grew robustly on OP9 cells with IL-7, Flt3L, and SCF ( Figure 1A, upper panels) and expressed intermediate levels of B220 (Figure 1-figure supplement 1C). Pro-B (+) cells were able to initiate differentiation into the T-cell lineage, DN2 (CD25 + CD44 + ), DN3 (CD25 + CD44 lo ), and double-positive (DP) stages on Dll4-expressing OP9 (OP9-Dll4) cells in vitro ( Figure 1A, lower right, Figure 1-figure supplement 1D), and mature into CD4 and CD8 T cells in vivo ( Figure 1-figure supplement 1E), as shown previously (Pongubala et al., 2008). In contrast, pro-B(À) cells remained in the pro-B cell phenotype (CD25 À CD44 + B220 int ) and died with Dll4-mediated Notch stimulation, although the expression levels of the Notch receptors were comparable ( Figure 1A, Figure 1-figure supplement 1F). To explore their different characteristics at the molecular level, we carried out comparative microarray analysis and identified the differentially expressed genes between pro-B(+) and pro-B(À) cells ( Figure 1B, Supplementary file 1). We identified approximately 400 differentially expressed genes (FC > 10) between the pro-B(+) and pro-B(À) cells, and these genes were enriched for genes related to the 'Hematopoietic cell lineage' pathway ( Figure 1-figure supplement 1G). Among them, the functional importance of Meis1, Hmga2, and Bcl11a has been reported in undifferentiated HPCs (Wong et al., 2007;Ariki et al., 2014;Nishino et al., 2008;Copley et al., 2013;Yu et al., 2012). However, the introduction of Meis1 and Hmga2 failed to overcome the defective T-cell differentiation in the presence of Notch signaling, and that of Bcl11a was highly toxic in pro-B(À) cells (data not shown). Lmo2 has also been shown to be linked to the reprogramming of lineage-committed blood cells or mesenchymal cells to the induced HSCs (Riddell et al., 2014;Batta et al., 2014;Vereide et al., 2014). We found that the expression levels of Lmo2 mRNA and protein were approximately threefold higher in pro-B(+) than pro-B(À) cells ( Figure 1B,C), and enforced expression of Lmo2 markedly provided their capacity to differentiate into T-cell lineage following Notch stimulation in vitro ( Figure 1D). Therefore, these results suggest that LMO2 plays a pivotal role in the maintenance of the capacity to differentiate into T-cell lineage driven by Notch signaling in Ebf1-deficient pro-B cells.

Disruption of Lmo2 in pro-B(+) cells leads to differentiation arrest following stimulation by Notch signaling
To confirm the necessity of LMO2 for the maintenance of the differentiation capacity into T-cell lineage, we tested the effect of Lmo2 disruption on the developmental potential of T-cell lineage in bone marrow (BM)-derived progenitor cells and pro-B(+) cells. BM progenitors from Cas9;Bcl2 Tg mice were infected with a bicistronic retroviral vector carrying an sgRNA against Lmo2 with the human nerve growth factor receptor (hNGFR) or control sgRNA against luciferase. Two days after sgRNA introduction, the cells were transferred onto an OP9-Dll1 monolayer to initiate T-lineage differentiation ( Figure 2-figure supplement 1A). The ability sgLMO2-treated progenitors to differentiate into the T-lineage was comparable to that of control ( Figure 2-figure supplement 1B). Next, Cas9-introduced pro-B(+) cells were transduced with sgRNA against LMO2, which caused specific loss of LMO2 protein 4 days after retroviral infection as detected by immunoblotting (Figure 2-figure supplement 1C). Five or ten days after sgLMO2 transduction, the cells were transferred onto a OP9-Dll4 monolayer to induce Notch-dependent T-lineage differentiation ( Figure 2A). Lmo2-deficient pro-B(+) cells progressed to the DN2 stage (CD25 + CD44 + ) as well as control cells on day 5, however, developmental arrest was observed ten days after sgLMO2 transduction ( Figure 2B). These results indicate that LMO2 is necessary for pro-B(+) cells to maintain their ability to differentiate into T-cell lineage, and loss of their differentiation potential took several days (~10 days) after Lmo2 disruption. Thus, the process may require loss of LMO2-mediated slower time-scale transcriptional changes, including histone modifications, chromatin remodeling, and DNA methylation. Source data 1. Raw data used to generate the graph in Figure 1A. Source data 2. Raw data used to generate the graph in Figure 1B. Source data 3. Raw data used to generate the graph in Figure 1C. Source data 4. Raw data used to generate the graph in Figure 1D. pro-B(À) cells, whereas Meis1 further downregulated Bcl11a expression. The expression of other potential target genes, including Meis1 and Hmga2, were also changed following Lmo2-and Meis1transduction, which failed to restore T-lineage differentiation capacity in pro-B(À) cells ( Figure 3A). Bcl11a is known to regulate the survival of lymphoid progenitors via induction of the anti-apoptotic gene, Bcl2 (Yu et al., 2012). Therefore, we examined the roles of the Bcl11a/Bcl2 pathway in the survival of pro-B(À) cells and their potential to differentiate into T-lineage cells. Transduction of Bcl2 significantly protected pro-B(À) cells from cell death, before and after Notch stimulation ( Figure 3B, Figure 3-figure supplement 1). Conversly, overexpression of the oncogene, Bcl11a, was highly toxic in pro-B cells. Importantly, Bcl2-introduced pro-B(À) cells (pro-B(À)/Bcl2) still failed to progress into the CD25 + DN2 stage ( Figure 3C). Thus, these results suggest that LMO2 would regulate survival of pro-B(À) cells via the Bcl11a/Bcl2 pathway, and other mechanisms contribute to the maintenance of the potential to differentiate into T-cell lineage.

LMO2 is required for the activation of Tcf7 after Notch signaling
Our comparative microarray analysis of pro-B(À) and pro-B(+) cells showed that expression of Tcf7 (encoding TCF1), one of the most important and the earliest Notch target genes in thymic immigrant cells, was weakly detected in pro-B(+) cells; however, its level was approximately 100-fold lower in pro-B(À) cells. Therefore, we next examined the expression kinetics of Tcf7 in pro-B(+)/Bcl2 and pro-B(À)/Bcl2 cells that survive after Notch stimulation but fail to differentiate into T-cell lineages. The upregulation of Tcf7 expression observed in pro-B(+)/Bcl2 cells after Notch stimulation was significantly abrogated in pro-B(À)/Bcl2 cells ( Figure 4A). In contrast, another Notch-regulated gene, Gata3, was upregulated by Notch stimulation but modestly lower in pro-B(À)/Bcl2 cells compared to that in pro-B(+)/Bcl2 cells ( Figure 4A). These results were further confirmed at the protein level using pro-B(À)/Bcl2 and pro-B(+)/Bcl2 cells at 7 days post-Notch stimulation ( Figure 4B). In addition, there was a decrease in Tcf7 expression in Lmo2-deficient pro-B(+) cells ( Figure 2) after 10 days but Source data 1. Raw data used to generate the graph in Figure 2B.   Figure 1D. After culturing, the live cells were analyzed for the expression of CD44 and CD25 (left). The data represent the mean values of percentages of CD25 + cells in three independent biological replicates with SD (right). ***p<0.001 by two-sided Student's t-test. The online version of this article includes the following source data and figure supplement(s) for figure 3: Source data 1. Raw data used to generate the graph in Figure 3A. Source data 2. Raw data used to generate the graph in Figure 3B. Source data 3. Raw data used to generate the graph in Figure 3C.  progenitors plays a crucial role in the activation of Tcf7 once the progenitors migrate into the thymus and are stimulated by Notch signaling.
DNA methylation status of the Tcf7 locus is maintained by LMO2 Next, we examined the epigenetic status of the Tcf7 locus in pro-B cell lines. We observed that in pro-B(+) cells, the transcriptional start site (TSS) of the Tcf7 locus was highly enriched for the active histone mark H3K4-3Me. However, Lmo2 disruption caused a significant reduction in the H3K4-3Me marks 10 days post-sgRNA introduction (~30%), and a modest reduction in the H3K4-3Me levels after 5 days of Lmo2 deletion (~70%) ( Figure 5-figure supplement 1). Moreover, we found a CpG island at the TSSs of Tcf7 ( Figure 5A). Thus, we examined the DNA methylation levels of the CpG Source data 1. Raw data used to generate the graph in Figure 4A. Source data 2. Raw data used to generate the graph in Figure 4C. island at the TSSs of the Tcf7 locus in the pro-B cell lines. In pro-B(À) cells, the CpG island at the Tcf7 locus was highly methylated, in contrast to pro-B(+) cells that had a highly demethylated TSSs (77.6% vs. 29.8%) ( Figure 5B). Moreover, enforced expression of Lmo2 in pro-B(À) cells induced demethylation of the CpG island at the TSSs ( Figure 5B, pro-B(À)/LMO2). These results demonstrate that the epigenetic status of the Tcf7 locus in the progenitor cells is maintained by LMO2 in a transcriptionally poised chromatin state for quick responsiveness following Notch stimulation.
LMO2 directly binds to the Bcl11a and Tcf7 loci Finally, we performed chromatin immunoprecipitation followed by deep-sequencing (ChIP-seq) analysis to identify genome-wide LMO2 occupancy sites in the pro-B(+) cells. We identified more than 1500 reproducible LMO2 binding peaks, and enrichment of Runx and ETS motifs that are frequently found in open chromatin regions of hematopoietic progenitors and pre-commitment pro-T cells (Yoshida et al., 2019;Ungerbäck et al., 2018; Figure 6A). The motif for a previously reported LMO2 interacting partner, the bHLH factors, was also coenriched ( Figure 6A). Many of the bHLHregulated genes, including Lyl1, Erg, and Hhex, were bound by LMO2 ( Figure 6B). In addition, these motifs are highly relevant to the binding sites for major transcription factors, such as LMO2, in HPCs (Wilson et al., 2010). Moreover, two LMO2 binding peaks were detected in the downstream regions of the Bcl11a locus, one of the LMO2-sensitive genes ( Figures 3A and 6B). Most importantly, a clear LMO2 peak was observed at the À35 kb upstream region of the Tcf7 locus. This binding site overlapped with RBPJ, a DNA-binding subunit of the Notch intracellular domain (NICD) transcriptional complex, binding sites at the Tcf7 locus, and co-occupied with Runx1 in pre-commitment DN1 cells Tcf7 loci and regulates their expression to maintain the ability of progenitors to differentiate into T-cell lineage.

Discussion
Here, we demonstrate that LMO2 contributes to the differentiation capacity of T-cell lineage in Ebf1-deficient pro-B cells via activation of the Bcl11a/Bcl2 pathway, which is critical for cell survival, and maintenance of DNA methylation status of the Tcf7 locus. LMO2 directly binds to its target loci and enables the Tcf7 locus to achieve a state that is responsive to T-cell induction following Notch signaling. We established our original thymic stromal cell line, TD7, in which HSCs differentiate into CD19 + B-lineage cells in the presence of IL-7 (Figure 1-figure supplement 1A) or into Thy1 + CD25 + T-lineage cells upon receiving Notch signaling. To establish pro-B cell lines possessing pluripotency, we cultured HSCs from the fetal liver of Ebf1-deficient mice on TD7 or OP9 stromal cells, which have been reported to establish pro-B cell lines (Pongubala et al., 2008). The pro-B cells derived from cultures with TD7 grew robustly in the presence of IL-7, but abruptly died when Notch signaling is provided, in contrast to the cells from cultures with OP9, which differentiated into DN2/3 stages with Notch signaling. The characteristics observed in the former pro-B cells were found to be due to decreased expression of Lmo2. These results suggest that TD7 does not support the pluripotency of Ebf1-deficient pro-B cells, leading to loss of LMO2. Although the molecular machinery underlying  LMO2 expression in HSCs has not been fully described, it was shown that several cis-regulatory elements around the Lmo2 locus cooperatively function and ensure the full expression pattern of Lmo2 (Landry et al., 2005;Landry et al., 2009), which appears to be driven in part by PU.1, TAL1, GATA factors, and LMO2 itself. The culture conditions provided by TD7 may be unable to sustain sufficient expression of these transcription factors. Alternatively, it is possible that TD7 could not support pro-B cells that retained their pluripotency because M-CSF, which is not produced by OP9 (Nakano et al., 1994), promoted the differentiation of HSCs into myeloid cells. In fact, the establishment of pro-B cells on TD7 was accompanied by a prolonged emergence of myeloid marker-positive cells compared to that on OP9, and the remaining pro-B cells lost their pluripotency (Figure 1-figure supplement 1B). Therefore, HSCs may be depleted by differentiation into the myeloid lineage, resulting in IL-7-dependent proliferation of pro-B cells without pluripotency that express reduced Lmo2. In either case, substantial expression of Lmo2 is critical for the maintenance of pluripotency in Ebf1-deficient pro-B cells, which is consistent with the finding that LMO2 is an important factor for the reprogramming of committed blood or mesenchymal cells to the induced HSCs (Riddell et al., 2014;Batta et al., 2014;Vereide et al., 2014).
A remarkable feature of LMO2 is that it only generates T-cell malignancies by abnormal expression (Neale et al., 1995;Curtis and McCormack, 2010;Matthews et al., 2013). A recent study indicated that LMO2 overexpression induces T-cell malignancies not only at the hematopoietic undifferentiated stages, but also after differentiation to B-lineage cells (García-Ramírez et al., 2018). However, the strong association between LMO2 and T-cell malignancy is not well understood at the molecular level. We showed here that LMO2 plays an important role in maintaining the chromatin structure of the Tcf7 locus in an accessible state necessary for Notch signaling-mediated activation of Tcf7. The expression of Tcf7 is highly specific to the T-cell lineage among hematopoietic cells, and TCF1 is important for initiating the epigenetic identity of the T-cell lineage (Johnson et al., 2018). Therefore, it can be inferred that, as the Tcf7 locus remains in an accessible state in HPCs or lineage-committed cells with ectopic LMO2 expression, they efficiently differentiate into T-lineage cells and selectively develop T-cell malignancy. In addition, TCF1 forms a complex with b-catenin, which contributes to the development of T-cell malignancy (Bigas et al., 2020). Although the contribution of the b-catenin/TCF1 complex in normal T-cell development remains unclear, the complex, together with RAG molecules, induces various genetic instabilities including those in the Myc gene in immature T cells and causes Myc gene-targeted tumorigenesis (Dose et al., 2014;Gekas et al., 2016). As TCF1 is also involved in other tumorigenesis (Liu et al., 2019), aberrant expression of LMO2 may promote T-cell malignancy via TCF1.
Previous reports suggest that LMO2 forms complexes with the bHLH transcription factors, including E2A, Lyl1, and TAL1/SCL, and binds to their target genomic sites (Wadman et al., 1997;Grütz et al., 1998;El Omari et al., 2013;Layer et al., 2016). Consistently, in this study, we found that the binding motif for bHLH factors was enriched in LMO2 binding regions in pro-B(+) cells. Thus, LMO2 seems to function along with these bHLH factors in the lymphoid progenitors. In addition, we found that LMO2 directly binds to the Tcf7 locus. Initiation of Tcf7 expression requires Notch signaling and Runx factors in T-cell progenitors (Weber et al., 2011;Guo et al., 2008;Shin et al., 2021). Two NICD-RBPJ complex and Runx1 binding sites were found at the À31 and À35 kb upstream regions of the Tcf7 locus ( Figure 6-figure supplement 1; Weber et al., 2011;Romero-Wolf et al., 2020;Hosokawa et al., 2018). Among these, the functional importance of the À31 kb region has been previously reported as a Notch-dependent enhancer of Tcf7 (Weber et al., 2011;Harly et al., 2020). The LMO2 binding site, identified in this study, overlapped with the other RBPJ binding site, at the À35 kb upstream region. While the physiological role of the À35 kb RBPJ binding site has not been clarified, our data suggest that direct binding of LMO2 to this region in progenitor cells plays an important role in maintaining the accessible chromatin configuration of the Tcf7 locus. In fact, Lmo2 expression levels were found to be associated with DNA methylation status around the Tcf7 locus and subsequent Notch-mediated activation of Tcf7 expression. Taken together, in the T-cell progenitor stage, LMO2 acts as a gatekeeper for maintaining the transcriptionally poised chromatin state of the Tcf7 locus via direct binding to the À35 kb region, and guarantees T-cell differentiation potential.

Isolation of hematopoietic progenitors and establishment of pro-B(+) and pro-B(-) cell lines
For the establishment of HPC lines, Ebf1-deficient progenitor cells were isolated from Ebf1-deficient E14.5 fetal liver (both male and female). FL cells were incubated with biotin-conjugated mAbs against lineage markers (TER119 and Gr-1, Lin). The cells were then washed and incubated with antibiotin Microbeads (Miltenyi Biotec), and Lin cells were enriched using autoMACS (Miltenyi Biotec). Lin À cells were cultured on OP9 or TD7 cells in the presence of mSCF, hFlt3L and mIL7. We established two pro-B cell lines, one that was maintained on OP9 cells was called as pro-B(+) cells, and the other that was maintained on TD7 cells was called as pro-B(-) cells.

Viral vector transduction of pro-B cell lines
Retroviral vector GCDN (mock vector containing IRES-hNGFR) (Hirano et al., 2015), encoding hBCL2, mMeis1, and mHmga2 were generated by transient transfection into PLAT-E packaging cells. Lentiviral vector pLVS-EF-IR2 (mock vector containing IRES-rat CD2), encoding mLmo2 and mTcf7 were generated by transient transfection into 293T as described previously (Hirano et al., 2020). Pro-B cells were transduced with viral supernatants by spin infection as described previously (Hozumi et al., 2003). After the infection, cells were plated and maintained on stromal cells.

RNA preparation and RT-qPCR
Total RNA was isolated from 4510 5 cells using Trizol reagent (Thermo Fisher Scientific) according to the manufacturer's instructions. cDNA was synthesized with High Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). Quantitative PCR (qPCR) was performed with Fast SYBR Green Master Mix (Thermo Fisher Scientific) and LightCycler 480 System II (Roche). The following primer sets were used for qPCR are listed below.

Microarray analysis
Total RNA was extracted from Ebf1-deficient pro-B(-) and pro-B(+) cells, and each 100 ng RNA was used for microarray analysis. Microarray experiments were performed with Whole Mouse Genome 4x44K ver. 2 Microarray Kit (Agilent) using the Agilent One-Color Microarray-Based Gene Expression Analysis protocol. Data from samples that passed the QC parameters were subjected to 75th percentile normalization and analyzed using Genespring GX (version 12, Agilent Technologies).
CRISPR/Cas9-mediated deletion of LMO2 in Pro-B cell lines sgRNA expression vector (E42-dTet) and Cas9-GFP expression vector were described previously . 19-mer sgRNAs were designed using the Benchiling web tool (https://www.benchling.com) and inserted into the empty sgRNA-expression vector by PCR-based insertion. Three sgRNA-expression vectors were generated for one gene, and pooled retroviral plasmids were used to make retroviral supernatant. Sequences of sgRNAs used in this study are listed below. Control (Luciferase); ggcatttcgcagcctaccg LMO2 #1; tcgatggccgaggacattg LMO2 #2; aatgtcctcggccatcgaa LMO2 #3; gaaagccatcgaccagtac Pro-B cell lines were transduced with retroviral vectors encoding Cas9-GFP, and 3 days after infection, GFP + retrovirus-infected cells were sorted. Then, they were expanded for a week and subjected second retrovirus transduction with sgLMO2-hNGFR. They were transferred onto OP9-Dll4 on day5 or day10 after snd infection, then CD25 and CD44 profiles on GFP + hNGFR + retrovirusinfected cells were analyzed, 3 days later.

Immunoblotting
Cytoplasmic and nuclear extracts, used to the detection of Tubulina and LMO2, respectively, were prepared by using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Pierce). Lysates were run on 12.5% polyacrylamide gel, followed by immunoblotting. The antibodies used for the immunoblot analysis were anti-Tubulina (Sigma, T6199) and anti-LMO2 (Novus, NB110-78626).

DNA methylation analysis
Genomic DNA was isolated from pro-B(+), pro-B(-) and LMO2/pro-B(À) cells. The DNA was purified using NucleoSpin Tissue (TaKaRa Bio) and was treated with bisulfite using MethylEasy Xceed (TaKaRa Bio) according to the manufacturer's instructions. For the evaluation of DNA methylation status of CpG island in Tcf7 promoter region, which contains 25 potential CpG methylation sites, bisulfite-converted DNA was amplified by PCR using TaKaRa EpiTaq HS (for bisulfite-treated DNA, TaKaRa Bio) and the following primer set: 5'-ttaagtttttattggtgaatgagtt-3' and 5'-aaaaaactccaaaaataaaacccac-3'. Amplified PCR products were cloned into pMD20-T vector using Mighty TA-cloning kit (TaKaRa Bio) and then sequenced.
Chromatin immunoprecipitation (ChIP) and ChIP-sequencing 1 Â 10 7 cells were fixed with 1 mg/ml DSG (Thermo Scientific) in PBS for 30 min at RT followed by an additional 10 min with addition of formaldehyde up to 1%. The reaction was quenched by addition of 1/10 vol of 0.125 M glycine and the cells were washed with HBSS (Gibco). Pelleted nuclei were dissolved in lysis buffer (0.5% SDS, 10 mM EDTA, 0.5 mM EGTA, 50 mM TrisHCl [pH 8] and PIC) and sonicated on SFX150 (Branson) for scycles of 20 s sonication followed by 1 min rest, with 30% amplitude. Six mof anti-LMO2 Abs (a mixture of 2 mg of NB110-78626 [Novus], 2 mg of ab91652 [Abcam] and 2 mg of AF2726 [R and D systems]) or anti-H3K4-3Me (07-043, Millipore) were pre-bound to Dynabeads Protein A/G or Dynabeads anti-Rabbit Ig (Invitrogen) and then added to the diluted chromatin complexes. The samples were incubated overnight at 4˚C, then washed and eluted for 6 hr at 65˚C in ChIP elution buffer (20 mM Tris-HCl, pH 7.5, 5 mM EDTA, 50 mM NaCl, 1% SDS, and 50 mg/ml proteinase K). Precipitated chromatin fragments were cleaned up using PCR purification Kit (Qiagen).