Ebf1 and c-Myb repress Rag transcription downstream of Stat5 during early B cell development

Abstract

The temporal control of recombination-activating gene (Rag) expression in developing lymphocytes prevents DNA breaks during periods of proliferation that could threaten genomic integrity. In developing B cells, the interleukin-7 receptor (IL-7R) and precursor B cell antigen receptor (pre-BCR) synergize to induce proliferation and the repression of Rag at the protein and mRNA levels for a brief period following successful immunoglobulin (Ig) heavy-chain gene rearrangement. While the mechanism of RAG2 protein downregulation is well-defined, little is known about the pathways and transcription factors that mediate transcriptional repression of Rag. Using Abelson Murine Leukemia Virus (AMuLV)-transformed B cells to model this stage of development, we identified Early B Cell Factor 1 (Ebf1) as a strong repressor of Rag transcription. shRNA-mediated knockdown of either Ebf1 or its downstream target c-Myb was sufficient to induce Rag transcription in these highly proliferative cells. Ebf1 and c-Myb antagonize Rag transcription by negatively regulating the binding of Foxo1 to the Rag locus. Ebf1 accomplishes this through both direct negative regulation of Foxo1 expression, and direct positive regulation of Gfi1b expression. Ebf1 expression is driven by the IL-7R downstream effector Stat5, providing a link between the negative regulation of Rag transcription by IL-7 and a novel repressive pathway involving Ebf1 and c-Myb.

Introduction

The generation of diverse B and T cell antigen (Ag) receptor repertoires is dependent on the expression of the recombination-activating genes Rag1 and Rag2 (collectively known as Rag) during early lymphocyte development (1). RAG1 and RAG2 proteins form a complex that binds to conserved recombination signal sequences (RSSs) flanking a pair of Ag receptor gene segments and synchronously generates double-stranded DNA breaks (DSBs) between each RSS and its corresponding gene segment (2). Rag expression and DSB generation are restricted to the G0–G1 phases of cell cycle such that repair of DNA coding ends in the RAG-stabilized post-cleavage complex is carried out by the non-homologous end-joining (NHEJ) pathway resulting in assembly of the variable domain exons of Ag receptor genes (3). RAG-induced DSBs produced during S phase have the potential to be repaired by homologous recombination, a process that can lead to chromosomal translocations and transformation (4, 5). As lymphocytes go through periods of clonal expansion during their development, the balance between proliferation and differentiation, along with the expression of Rag, must be tightly regulated to maintain genomic integrity and ensure the production of diverse pools of B and T cells to mediate adaptive immune responses.

Following productive immunoglobulin (Ig) heavy-chain gene rearrangement in developing pro-B cells in the bone marrow, expression of the precursor B cell antigen receptor (pre-BCR) activates signaling pathways that synergize with interleukin-7 receptor (IL-7R) signaling to direct two processes (6, 7). The first is clonal expansion of pre-BCR+ progenitors, a stage during which RAG protein and Rag mRNA levels are negatively regulated upon entry of these large, cycling pre-B cells into S phase (8, 9). The second process is differentiation to the small pre-B cell stage, which involves coordinated cell cycle exit, re-expression of Rag1 and Rag2, and increased chromatin accessibility at the Ig kappa light-chain locus (10) to allow light chain gene recombination and ultimately the assembly of a complete B cell receptor (BCR).

Phosphorylation and proteasome-dependent degradation of RAG2 controls recombinase protein levels during this proliferative burst (11). However, the mechanism by which Rag1 and Rag2 transcription is repressed by IL-7R and pre-BCR signaling is ill-defined. Activation of the PI(3)K-Akt pathway downstream of these receptors has been implicated in the inactivation of Rag transcription via antagonism of Foxo transcription factors (12–14). Gfi1b and Stat5 have been implicated as direct negative regulators of Rag transcription (15, 16). Stat5 is activated by IL-7R signaling (17), consistent with the ability of IL-7 to repress Rag transcription (6, 12).

Abelson Murine Leukemia Virus (AMuLV)-transformed B cell lines provide an in vitro model system to study the dynamics of Rag transcription during the developmental transition from the large to small pre-B cell stage. The constitutively active v-Abl kinase transforms B cell progenitors in a highly proliferative state where Rag expression is low, mimicking the large, cycling pre-B cell stage of development. This developmental block can be released by inhibition of v-Abl with the small molecule kinase inhibitor STI-571 (STI) (18). STI treatment induces cell cycle arrest, upregulation of Rag transcription, and differentiation to a small pre-B cell-like state where initiation of Ig kappa light-chain gene recombination occurs. In this study, we utilized the AMuLV system to identify novel pathways and factors responsible for the repression of Rag transcription. A gain-of-function screen identified unexpected roles for Early B Cell Factor 1 (Ebf1) and c-Myb in the repression of Rag transcription. The expression of these factors is driven by the IL-7R signaling effector Stat5, linking the negative regulation of Rag transcription by IL-7 to a novel repressive pathway involving Ebf1 and c-Myb.

Materials and Methods

Animal Use Statement

All experiments using C57/B6 mice were approved by the Animal Care and Use Committee at the University of California at Berkeley. The handling of animals was in accordance with protocol R253-0405.

Cell culture and chemicals

AMuLV-transformed B cells were cultured in RPMI 1640 (Gibco) supplemented with 5% vol/vol FCS (Gemini), 100 mg/mL penicillin and streptomycin (Gibco), and 55nM 2-mercaptoethanol (Gibco). Primary B cells isolated from C57/B6 mice were cultured in RPMI with 15% vol/vol FCS and supplemented with 2 ng/mL recombinant mouse IL-7 (R&D Systems). For IL-7 withdrawal experiments, IL-7 concentration was increased to 5 ng/mL for 24 h. Cells were then spun down and resuspended in media with 5 or 0.1 ng/mL IL-7 and cultured for an additional 24 h before harvest or analysis. GP2 retroviral packaging cells (a gift from G. Barton) were cultured in DMEM (Gibco) supplemented with 5% vol/vol FCS, 100 mg/mL penicillin and streptomycin, and 1mM sodium pyruvate (Gibco). All cells were grown at 37°C in 5% CO₂. STI-571 (Novartis) was used at a final concentration of 2.5 uM for 16 h for all experiments.

Expression plasmids

MSCV-based cDNA retroviral expression constructs were previously described (12). All cDNAs were PCR-amplified with Platinum Pfx DNA Polymerase (Invitrogen) and cloned into a multiple cloning site (MCS) upstream of an internal ribosome entry site (IRES). This IRES is followed by coding sequence for either human CD4 (hCD4) or Thy1.1 cell surface proteins to mark infected cells. All cDNA open reading frames (ORFs) were sequenced to confirm the absence of mutations.

Murine c-Myb transcript variant 2 (NM_010848.3) ORF was PCR-amplified from a primary B cell cDNA library, while murine Foxo1, Pax5, and Ebf1 ORFs were PCR amplified from existing plasmids and cloned into CMSCV IRES Thy1.1 or CMSCV IRES hCD4 retroviral vectors. N-terminal 3XFLAG-tagged Ebf1 was created by amplifying the Ebf1 ORF lacking an ATG start codon and cloning it downstream of a 3XFLAG sequence inserted into the MCS of CMSCV IRES hCD4. CA-STAT5B cDNA (a gift from M. Clark and M. Mandal) was excised from MIGR1 and cloned into CMSCV IRES hCD4.

MSCV-based shRNA retroviral expression constructs contain a human CD2 (hCD2) cell surface marker cDNA followed by a miR-30 cassette as previously described (19). Seed sequences for desired shRNA targets were identified with siRNA Wizard (InvivoGen), and shRNA oligonucleotides containing these seed sequences were designed using RNAi Codex. shRNA oligonucleotides purchased from Elim Biopharm were PCR amplified using Vent Polymerase (NEB) with 5uL DMSO (Sigma-Aldrich) added to each reaction, and cloned into XhoI and EcoRI restriction sites in the miR-30 cassette. Seed sequences and PCR amplification primers are provided in the Supplemental Table 1.

Retrovirus production and infection

GP2 packaging cells were transfected with a 5:3 ratio of retroviral:VSVG plasmids using Lipofectamine 2000 (Invitrogen) and placed at 37°C. Cells were moved to 32°C at 24 h. At 48 h, retroviral supernatant was collected and concentrated (1.5 h, 16,800g, 4°C) using a SW 41 Ti rotor and L8-M ultracentrifuge (Beckman). Retroviral pellets were resuspended in media supplemented with 4 ug/mL polybrene (Sigma-Aldrich). Target cells were cultured in concentrated retrovirus at 32°C for 24 h, and then moved to 37°C and expanded. Analysis and sorting were performed 2–4 days post-infection.

Cell sorting, flow cytometry, and intracellular staining

Single-cell suspensions were prepared and cells labeled with fluorochrome-conjugated antibodies using standard techniques. A FC500 (Beckman Coulter) or LSRII (BD Biosciences) flow cytometer was used for analysis, while a MoFlo or Influx cell sorter (Dako-Cytomation) was used for sorting. Data were analyzed with FlowJo software (Tree Star).

Primary B cells were labeled with anti-IgM (II/41), anti-B220 (RA3-6B2), anti-CD43 (S7), and anti-CD19 (1D3) antibodies. Anti-hCD2 (RPA-2.10), anti-hCD4 (RPA-T4) and anti-Thy1.1 (OX-7) antibodies were used to label cells tranduced with retrovirus. Anti-B220, anti-CD43, anti-CD19, and anti-Thy1.1 antibodies were obtained from BD Pharmingen. All others were obtained from eBiosciences.

The Fix & Perm Cell Permeabilization Kit (Invitrogen) was used for intracellular staining. Primary antibody was rabbit anti-EBF-1 (Millipore AB10523) or rabbit IgG control (GenScript). Secondary antibody was Alexa Fluor 647 F(ab′)2 fragment goat anti-rabbit. Gating strategy was as follows: pro-B cells (IgM−, B220+, CD43+), large cycling pre-B cells (IgM−, B220+, CD43−, forward scatter high), small resting pre-B cells (IgM−, B220+, CD43−, forward scatter low).

Quantitative realtime PCR (qPCR)

Cells were collected by centrifugation and lysed in Trizol, or sorted directly into Trizol LS (both from Invitrogen). RNA was prepared and reverse transcription performed with MMLV-RT (Invitrogen) using random hexamer priming. Quantitative realtime PCR was carried out on an Applied Biosystems 7300 thermocycler using JumpStart Taq (Sigma) and EvaGreen (Biotium). PCR conditions were as follows: 50°C 2 min, 95°C 10 min, 40 cycles of 95°C 15s, 60°C 1 min. See Supplemental Table I for primer sequences. For gene expression analysis, transcript levels of all genes were normalized to Hprt. For ChIP qPCR, data is presented as a percentage of ChIP input. Error bars on all plots represent the standard deviation of triplicate qPCR assays.

Immunoblot

Cells were lysed in RIPA buffer supplemented with Complete Protease Inhibitor Cocktail (Roche). Cell debris was cleared by centrifugation and soluble protein quantified with Bradford Reagent (Bio-Rad). Laemmli SDS loading buffer was added to 20–40ug of protein per sample prior to separation on 8–10% SDS-PAGE gels. Following transfer to Immobilon-FL PVDF membranes (Millipore) and blocking with 5% vol/vol milk/PBS solution, blots were probed with primary and secondary antibodies and analyzed with the Odyssey Infrared Imaging System (LI-COR Biosciences). Primary antibodies used: anti-EBF-1 (R&D Systems AF5156 and Millipore AB10523), anti-Myb (Millipore 05-175), anti-Pax5 (Santa Cruz sc-1974), anti-Stat5 (Santa Cruz sc-835), anti-Foxo1 (Cell Signaling L27), anti-Lamin B1 (Abcam ab16048), anti-Tubulin (Calbiochem CP06), and anti-FLAG M2 (Sigma-Aldrich). Infrared dye–conjugated secondary antibodies were from Molecular Probes–Invitrogen. Quantification was performed using ImageJ.

Chromatin immunoprecipitation

Chromatin immunoprecipitation (ChIP) was performed as previously described (20). Briefly, 80 million cells per experimental condition were harvested, fixed, lysed, and sonicated using a Branson 450 Digital Sonifier. Following centrifugation to remove insoluble material, chromatin was quantified and equal amounts used in experimental and control immunoprecipitations. 5 μg of anti-FLAG M2 (Sigma-Aldrich), anti-Foxo1 (Abcam ab70382), anti-Pax5 (Santa Cruz sc-1974X), anti-Ebf1 (Sigma-Aldrich SAB2501166), or IgG control antibodies (GeneScript) were conjugated to Protein G Dynabeads (Invitrogen) and added to samples. Following overnight immunoprecipitation, beads were collected and washed three times with low salt buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris pH 8.1, 150 mM NaCl), once with high salt buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris pH 8.1, 500 mM NaCl), once with LiCl buffer (250 mM LiCl, 1% NP40, 1% sodium deoxycholate, 1 mM EDTA, 10 mM Tris pH 8.1) and twice with TE buffer. DNA-protein complexes were eluted from beads with TE elution buffer (TE w/1% SDS), and crosslinks reversed overnight at 65°C. Following RNAse A (Quiagen) and Proteinase K (Invitrogen) treatment, chromatin was isolated using QIAquick columns (Quiagen) and subjected to qPCR analysis. See Supplemental Table I for primer sequences.

Results

Ebf1 overexpression negatively regulates Rag transcription

To screen for transcriptional regulators of Rag, we used retroviruses to individually overexpress a panel of transcription factors important for B cell development in a previously described AMuLV-transformed Rag1-GFP reporter B cell line (12). This line was derived from a heterozygous Rag1-GFP knock-in mouse that has a GFP cDNA in place of the coding exon at one Rag1 allele (21), allowing changes in Rag transcription to be monitored by flow cytometry. We assessed Rag1-GFP levels following overexpression of the different factors in both untreated and STI-treated reporter cells.

To our surprise, Early B Cell Factor 1 (Ebf1), a transcription factor implicated as a positive regulator of Rag transcription during early B cell development (22), repressed baseline Rag1-GFP levels when overexpressed in the reporter cell line as compared to empty vector control cells (Fig. 1A, top row). Moreover, when the Ebf1-overexpressing cells were treated with STI, Rag1-GFP induction was severely blunted compared to STI-treated control cells (Fig. 1A, bottom row). DNA binding by Ebf1 was required for Rag1-GFP repression as mutation of H157 in the Ebf1 zinc-finger motif abolished activity(23) (Fig. 1A, right). The repression of both Rag1 and Rag2 by Ebf1 was confirmed by quantitative realtime PCR (qPCR) (Fig. 1B), and the repressive effect of Ebf1 overexpression on Rag transcription was validated in independently generated Abelson cell lines (unpublished observations). Interestingly, Ebf1 overexpression did not affect other aspects of STI-induced differentiation, such as the upregulation of Irf4 and Spi-B expression, or the induction of Ig kappa germline transcription (κGT)(Fig. 1C).

Figure 1. Ebf1 overexpression negatively regulates Rag transcription.

(A) Flow cytometry analysis of GFP expression in an AMuLV-transformed B cell line generated from heterozygous Rag1-GFP knock-in mice, where GFP reports transcription from the endogenous Rag locus. Cells were infected with empty retroviral vector, Ebf1 retrovirus, or Ebf1 H157A DNA-binding mutant retrovirus (23). Uninfected cells (shaded histogram) were distinguished from transduced cells (black line) by staining with anti-hCD4 (retroviral marker). Analysis was performed on untreated (UT, top row) or STI-treated (+STI, bottom row) cells. Numbers above gate indicate the percentage of GFP+ uninfected cells (top) or transduced cells (bottom). Data are representative of four independent experiments. (B) Quantitative realtime PCR (qPCR) measuring Rag1 and Rag2 transcript levels in sorted empty vector-transduced or Ebf1-overexpressing AMuLV-transformed B cells in normal culture conditions (left) or treated with STI (right). Data are representative of two independent experiments. (C) qPCR measuring Irf4, Spi-B and kappa germline (κGT) transcript levels in sorted empty vector-transduced or Ebf1-overexpressing AMuLV-transformed B cells in normal culture conditions (white bars) or treated with STI (gray bars). Data are representative of two independent experiments. (D) GFP expression in CD19+IgM− primary B cells from heterozygous Rag1-GFP knock-in mice transduced with empty retroviral vector (filled histogram) or Ebf1 retrovirus (black line) and cultured in high (left) or low (right) concentrations of IL-7. Numbers indicate the GFP mean fluorescence intensity (MFI) of the empty vector-transduced cells (top) or Ebf1-overexpressing cells (bottom). Data are representative of three independent experiments. (E) qPCR measuring Rag1 and Rag2 transcript levels in sorted CD19+IgM− primary B cells from wildtype C57/B6 mice transduced with empty retroviral vector or Ebf1 retrovirus. Data are representative of two independent experiments.

To test if Ebf1 overexpression represses Rag transcription in primary cells, we transduced cultured pro-B cells isolated from heterozygous Rag1-GFP mice with Ebf1 or empty vector control retroviruses. In cultures with a high concentration of IL-7, flow cytometry analysis revealed that the Ebf1-overexpressing cells had a lower GFP mean fluorescence intensity (MFI) than empty vector control cells (Fig. 1D, left). When the IL-7 concentration in these cultures was lowered to induce Rag expression and differentiation to the pre-B cell stage, an increase in the GFP MFI was seen in the control cells but not the Ebf1-overexpressing cells (Fig. 1D, right), confirming that Ebf1 represses Rag induction upon IL-7 withdrawal. The repression of both Rag1 and Rag2 by Ebf1 in primary IL-7-cultured pro-B cells from wildtype C57/B6 mice was confirmed by qPCR (Fig. 1E). Together these data show that Ebf1 overexpression in Abelson cells and committed primary pro- and pre-B cells represses Rag transcription.

Ebf1 levels are inversely correlated with Rag expression at the large-to-small pre-B cell transition

Ebf1 mRNA levels are low in lymphoid progenitors, increase during progression through the pro-, pre-, and immature B cell stages, and decrease in mature and peripheral B cell subsets (24, 25). Our Ebf1 overexpression results prompted us to assess Ebf1 mRNA and Ebf1 protein expression levels with respect to Rag expression in developing B cells. In agreement with microarray data that shows Ebf1 transcript levels increasing from Fraction C′ (large pre-B) to Fraction D (small pre-B)(24) both STI treatment of Abelson cells and IL-7 withdrawal in primary cell cultures resulted in an induction of Ebf1 mRNA (Fig. 2B, D). However, we observed a decrease in Ebf1 protein levels upon STI treatment and IL-7 withdrawal in lysates prepared from these same cells (Fig. 2A, C), indicating that Ebf1 protein levels are inversely correlated with Rag expression in both Abelson cells and IL-7-cultured primary cells.

Figure 2. Ebf1 protein levels are inversely correlated with Rag expression.

(A, B) Immunoblot measuring Ebf1 protein levels (A) and qPCR measuring Ebf1 transcript levels (B) in AMuLV-transformed B cells in normal culture conditions or treated with STI. Tubulin immunoblot serves as a protein loading control. Numbers below lanes indicate Ebf1/Tubulin ratio for each sample. Data are representative of three independent experiments. (C, D) Immunoblot measuring Ebf1 protein levels (C) and qPCR measuring Ebf1 transcript levels (D) in CD19+IgM− IL-7-cultured primary B cells before and after IL-7 withdrawal. Tubulin immunoblot serves as a protein loading control. Numbers below lanes indicate Ebf1/Tubulin ratio for each sample. Data are representative of two independent experiments. (E) Flow cytometry analysis of intracellular Ebf1 expression in pro-B cells (1, dotted line), large cycling pre-B cells (2, dashed), and small resting pre-B cells (3, solid) isolated from wildtype C57/B6 mouse bone marrow. Shaded histogram represents total pre-B cell isotype control staining. Data are representative of two independent experiments.

To further confirm the inverse relationship between Ebf1 and Rag expression, we performed intracellular staining for Ebf1 in primary B cells isolated from wildtype C57/B6 mouse bone marrow. Ebf1 expression in pro-B cells (IgM−, B220+, CD43+) was lower than in large cycling pre-B cells (IgM−, B220+, CD43−, forward scatter high) (Fig. 2E). And in agreement with our western blot data, small resting pre-B cells (IgM−, B220+, CD43−, forward scatter low) expressed less Ebf1 than large pre-B cells (Fig. 2E). These data suggest that Ebf1 protein levels peak in large cycling pre-B cells in which Rag transcription is repressed, and that Ebf1 protein levels decrease upon differentiation to the small resting pre-B cell stage when Rag transcription is upregulated to initiate Ig light-chain gene recombination. This inverse relationship between Ebf1 and Rag expression strongly correlates with our observation that Ebf1 overexpression negatively influences Rag transcription in Abelson cells and IL-7-cultured primary cells.

Ebf1 knockdown is sufficient to induce Rag transcription in Abelson cells

Upon STI-induced differentiation of Abelson cell lines, a multitude of changes in gene expression takes place, several of which are important for the induction of Rag transcription. If a decrease in Ebf1 protein levels is one of the changes essential for this induction, we reasoned that knockdown of Ebf1 might be sufficient to upregulate Rag transcription even in the absence of other STI-mediated effects. This was indeed the case as shRNA knockdown of Ebf1 was sufficient to induce Rag transcription in the Rag1-GFP reporter cell line, while a control luciferase shRNA had no effect on Rag levels (Fig. 3A). Reduction of Ebf1 mRNA levels by the Ebf1 shRNA was confirmed by qPCR (Fig. 3B). Thus Ebf1 is a bona fide repressor of Rag transcription in Abelson cells.

Figure 3. Ebf1 knockdown is sufficient to induce Rag transcription.

(A) GFP expression in AMuLV-transformed Rag1-GFP reporter B cells transduced with luciferase (LUC, left) or Ebf1 (right) shRNA retrovirus. Uninfected cells (shaded histogram) were distinguished from shRNA-expressing cells (black line) by staining with anti-hCD2 (retroviral marker). Numbers above each gate indicate the percentage of GFP+ uninfected cells (top) or shRNA-expressing cells (bottom) in each culture. (B) qPCR measuring Ebf1 transcript levels in sorted AMuLV-transformed B cells transduced with luciferase or Ebf1 shRNA. Data in A and B are representative of three independent experiments.

The Ebf1 target gene c-Myb represses Rag transcription in Abelson cells

Conditional deletion in pro-B cells revealed that Ebf1 is critical for their proliferation and survival (26)In agreement with this, Ebf1 overexpression enhanced proliferation in our Rag1-GFP reporter cell line, while Ebf1 knockdown was deleterious to survival as Ebf1 shRNA-transduced cells could not be expanded in culture (Fig. S1). The same study showed that the Ebf1 target gene c-Myb was important for cell survival, and that c-Myb overexpression rescued cell death in Ebf1-deficient AMuLV-transformed B cells. Consistent with c-Myb being downstream of Ebf1, qPCR analysis revealed that c-Myb mRNA levels decreased significantly upon Ebf1 knockdown (Fig. 4A). Although c-Myb has been previously described as a positive regulator of Rag transcription (27, 28)we found that c-Myb overexpression in the Rag1-GFP reporter cell line both decreased baseline Rag levels and severely blunted STI-induced Rag transcription (Fig. 4B).

Figure 4. c-Myb negatively regulates Rag transcription downstream of Ebf1.

(A) qPCR measuring c-Myb transcript levels in sorted AMuLV-transformed B cells transduced with luciferase (LUC) or Ebf1 shRNA. Data is representative of three independent experiments. (B) GFP expression in AMuLV-transformed Rag1-GFP reporter B cells transduced with c-Myb retrovirus. Uninfected cells (shaded histogram) were distinguished from c-Myb-overexpressing cells (black line) by staining with anti-hCD4 (retroviral marker). Analysis was performed on untreated (UT) or STI-treated (+STI) cells. Numbers above gate indicate the percentage of GFP+ uninfected cells (top) or c-Myb-overexpressing cells (bottom). Data are representative of three independent experiments. (C) GFP expression in AMuLV-transformed Rag1-GFP reporter B cells transduced with luciferase (left) or c-Myb (right) shRNA. Uninfected cells (shaded histograms) were distinguished from shRNA-expressing cells (black line) by staining with anti-hCD2 (retroviral marker). Numbers above gate represent the percentage of GFP+ uninfected cells (top) or shRNA-expressing cells (bottom). Data are representative of three independent experiments. (D) qPCR measuring c-Myb (left) and Ebf1 (right) transcript levels in sorted AMuLV-transformed B cells transduced with luciferase or c-Myb shRNA. Data in C and D are representative of three independent experiments.

To test if c-Myb is a bona fide repressor of Rag transcription, we performed shRNA knockdown of c-Myb in the Rag1-GFP reporter line. As was the case for Ebf1 knockdown, c-Myb knockdown was sufficient to induce Rag transcription in the absence of STI treatment (Fig. 4C). Conditional deletion of c-Myb in lymphoid progenitors showed that this factor positively regulates Ebf1 expression during early B cell development (29). However in our cell line, c-Myb knockdown does not affect Ebf1 transcript levels (Fig. 4D), suggesting that the observed increase in Rag transcription is due to c-Myb knockdown alone and not a concurrent decrease in Ebf1 expression. Thus during later stages of B cell development represented by our Abelson cell line, c-Myb is downstream of Ebf1 in a pathway that represses Rag transcription.

Ebf1 and c-Myb independently repress Rag transcription

The observation that c-Myb is downstream of Ebf1 (Fig. 4A) suggested that upon Ebf1 knockdown, the increase in Rag transcription may be due to the subsequent decrease in c-Myb levels. To test if a decrease in Ebf1 levels that occurs independent of a decrease in c-Myb levels would be sufficient to induce Rag transcription, we transduced cells overexpressing a c-Myb cDNA with our Ebf1 shRNA or control luciferase shRNA. qPCR confirmed that the expression level of c-Myb mRNA was stable upon Ebf1 knockdown due to rescue by the exogenous c-Myb cDNA (Fig. 5A). However, Ebf1 knockdown still induced higher levels of Rag transcription in the absence of STI as compared to luciferase shRNA control cells (Fig. 5B, left). Moreover, while STI treatment was unable to fully upregulate Rag transcription in the luciferase shRNA control cells due to the repressive effects of c-Myb overexpression, STI-induced Rag transcription in the Ebf1 knockdown cells is robust (Fig. 5B, right). This result, along with the c-Myb knockdown result, shows that an independent decrease in either Ebf1 or c-Myb levels in our Rag1-GFP reporter Abelson cell line is sufficient to activate Rag transcription.

Figure 5. Ebf1 negatively regulates Rag transcription independent of c-Myb.

(A) qPCR measuring c-Myb transcript levels in sorted c-Myb-overexpressing AMuLV-transformed Rag1-GFP reporter B cells transduced with luciferase (LUC) or Ebf1 shRNA. (B) GFP expression in cells from A under normal culture conditions (UT) or treated with STI (+STI). Data in A and B are representative of three independent experiments.

Ebf1 and c-Myb antagonize Foxo1 binding to the Rag locus upon STI-induced differentiation of Abelson cells

We hypothesized that Ebf1 and c-Myb might repress Rag transcription by antagonizing the activity of positive regulators of Rag. A recent study using Irf4^−/−Irf8^−/− pre-B cells implicated Foxo1 and Pax5 as critical factors for the upregulation of Rag transcription during the large to small pre-B cell transition (14). Individual overexpression of Foxo1 or Pax5 is sufficient to increase Rag1-GFP levels in the reporter Abelson cell line (Fig. 6A, Fig. S2) (12) and knockdown of either factor severely blunts STI-induced Rag transcription (Fig. 6B, Fig. S2)(12). Genome-wide ChIP-seq analysis in the Irf4^−/−Irf8^−/− pre-B cells identified Foxo1 and Pax5 binding sites in the Rag locus that are inducibly bound by these factors upon IL-7 withdrawal and pre-B cell differentiation (14). These sites included previously identified elements such as the Erag enhancer(30) and Rag2 promoter Pax5 binding site(28), along with novel promoter (P) and intergenic (I) binding sites (see Rag locus schematic in Fig. 6C). We performed Foxo1 and Pax5 ChIP in untreated and STI-treated Abelson cells and observed increased occupancy of these factors at these Rag locus sites following STI treatment (Fig. 6D, left, and Fig. S2). We also observed increased Foxo1 occupancy at sites in the Blnk and Syk loci, two other genes identified as direct Foxo1 targets during Irf4^−/−Irf8^−/− pre-B cell differentiation(14)(Fig. 6D, right). Together, these experiments implicate Foxo1 and Pax5 as direct transcriptional activators of Rag and other genes involved in pre-B cell differentiation in the AMuLV system following STI treatment.

Figure 6. Ebf1 and c-Myb antagonize Foxo1 binding to the Rag locus during pre-B cell differentiation.

(A) GFP expression in AMuLV-transformed Rag1-GFP reporter B cells transduced with Foxo1 cDNA retrovirus. Uninfected cells (shaded histogram) were distinguished cDNA-overexpressing cells (black line) by staining with anti-Thy1.1 (retroviral marker). Numbers above gate indicate the percentage of GFP+ uninfected cells (top) or cDNA-overexpressing cells (bottom). Data are representative of two independent experiments. (B) GFP expression in AMuLV-transformed Rag1-GFP reporter B cells transduced with Foxo1 shRNA and treated with STI. Uninfected cells (shaded histogram) were distinguished from shRNA-expressing cells (black line) by staining with anti-hCD2 (retroviral marker). Numbers above gate indicate the percentage of GFP+ uninfected cells (top) or shRNA-expressing cells (bottom). Data are representative of two independent experiments. (C) Schematic of the Rag locus showing approximate locations of sites bound by Foxo1 and Pax5 upon pre-B cell differentiation (14). “P” and “I” denote promoter and intergenic binding sites, respectively. (D) ChIP in AMuLV-transformed B cells with anti-Foxo1 antibody. Cells were cultured under normal conditions (white bars) or treated with STI (gray bars) prior to harvest. qPCR was performed with recovered chromatin to assess Foxo1 occupancy Rag locus sites depicted in C, or at binding sites near other Foxo1 target genes (14). “Rag2Pax5BS” serves as negative control as Foxo1 does not bind this site. Data are representative of two independent experiments. (E) ChIP in AMuLV-transformed B cells with anti-Foxo1 or IgG control antibody. Empty vector-transduced, Ebf1-overexpressing, and c-Myb-overexpressing cells were treated with STI and harvested. qPCR was performed with recovered chromatin to assess Foxo1 occupancy at the binding sites studied in D. Data are representative of two independent experiments.

We then tested if Ebf1 and c-Myb overexpression affected the binding of Foxo1 or Pax5 to the Rag locus sites upon STI treatment. While coimmunoprecipitation experiments revealed that both Ebf1 and c-Myb physically interact with Pax5 (unpublished observations), Pax5 ChIP experiments showed that binding of this factor to the Rag locus upon STI treatment was not affected by Ebf1 or c-Myb overexpression (Fig. S2). However, Foxo1 ChIP experiments revealed that overexpression of either Ebf1 or c-Myb clearly reduced the amount of Foxo1 bound to the Rag locus upon STI treatment (Fig. 6E, left). Ebf1 or c-Myb overexpression did not affect Foxo1 binding to the Blnk and Syk loci (Fig. 6E, right), suggesting these factors repress Rag induction but not other aspects of pre-B cell differentiation. Thus Ebf1 and c-Myb negatively regulate Rag transcription by antagonizing Foxo1 binding to the Rag locus.

Ebf1 and c-Myb negatively regulate Foxo1 expression

ImmGen microarray data shows that Foxo1 mRNA levels are high in Fractions B and C, decrease in Fraction C′, and increase again in Fraction D (24). In agreement with this, we found that Foxo1 transcript levels increase upon differentiation of primary B cell cultures and AMuLV cells following IL-7 withdrawal and STI treatment, respectively (Fig. 7A). Together these data suggest that factors actively repress Foxo1 transcription in large cycling pre-B cells. Reduced Foxo1 binding to the Rag locus in STI-treated Ebf1- and c-Myb-overexpressing cell lines prompted us to assess Foxo1 expression levels in these cells. Western blotting revealed lower levels of Foxo1 protein in cells overexpressing Ebf1 and c-Myb compared to control empty vector-transduced cells (Fig. 7B), suggesting that these factors act as negative regulators of Foxo1 expression. To further assess the relationship between Ebf1, c-Myb, and Foxo1, we assayed Foxo1 mRNA levels in cells transduced with shRNAs targeting Ebf1 and c-Myb. Indeed both Ebf1 and c-Myb knockdown resulted in an increase in Foxo1 transcript levels compared to cells transduced with a control luciferase shRNA (Fig. 7C), supporting the notion that Ebf1 and c-Myb are negative regulators of Foxo1 transcription in Abelson cells.

Figure 7. Ebf1 and c-Myb negatively regulate Foxo1 expression.

(A) qPCR measuring Rag1 and Foxo1 transcript levels upon IL-7 withdrawal-induced differentiation of primary B cells from C57/B6 mouse bone marrow (top), and STI-induced differentiation of AMuLV-transformed B cells (bottom). Data are representative of two independent experiments. (B) Immunoblot measuring Foxo1 protein levels in sorted AMuLV-transformed B cells transduced with empty retroviral vector, Ebf1 retrovirus (top) or c-Myb retrovirus (bottom). Tubulin and Lamin serve as loading controls. Numbers below lanes indicate Ebf1/loading control ratio for each sample. Data are representative of two independent experiments. (C) qPCR measuring Foxo1 transcript levels in sorted AMuLV-transformed B cells transduced with luciferase shRNA (LUC), and either Ebf1 (top) or c-Myb shRNA (bottom). Data are representative of three independent experiments. (D) Published Ebf1 ChIP-seq data (26, 31) visualized in the UCSC Genome Browser at the Foxo1 locus. “P” and “I” denote promoter and intergenic Ebf1 peaks, respectively. Ebf1 binding sites (5′-CCCNNGGG-3′) identified within each peak are shown below, along with their location relative to the Foxo1 TSS. (E) ChIP in AMuLV-transformed B cells overexpressing a 3XFLAG-tagged Ebf1 protein with anti-FLAG or IgG control antibody. qPCR was performed with recovered chromatin to assess 3XFLAG-Ebf1 occupancy at sites in the Foxo1 locus described in D compared to a negative control site in the Rag1 promoter. Data are representative of three independent experiments. (F) ChIP in AMuLV-transformed B cells with anti-Ebf1 antibody. Cells were cultured in normal conditions (white bars) or treated with STI (gray bars) prior to harvest. qPCR was performed on recovered chromatin to assess Ebf1 occupancy at sites in the Foxo1 locus described in D. Data are representative of two independent experiments.

Genome-wide ChIP-seq analysis in pro-B and splenic B cells revealed Ebf1 binding sites in both the Foxo1 promoter and downstream intergenic sites (26, 31)(Fig. 7D). Conventional ChIP in cells overexpressing a 3XFLAG-tagged Ebf1 protein revealed strong Ebf1 binding to these sites, but not to a control site in the Rag1 promoter (Fig. 7E), or to other sites in Rag locus including the Erag enhancer (Fig. S3). Furthermore, ChIP with an anti-Ebf1 antibody revealed that endogenous Ebf1 is bound to these sites in resting Abelson cells, and that Ebf1 occupancy decreases upon STI-induced differentiation (Fig. 7F). These experiments suggest that Ebf1 represses Foxo1 transcription directly and that this repression is relieved upon STI treatment to allowing induction of Foxo1 mRNA and Rag transcription.

Ebf1 and c-Myb positively regulate Gfi1b expression

In spite of their negative influence on Foxo1 expression, Foxo1 binding to non-Rag locus targets upon STI-induced differentiation was unaffected by Ebf1 or c-Myb overexpression (Fig. 6E, right), suggesting Ebf1 and c-Myb influence chromatin accessibility at the Rag locus but not other Foxo1 target genes. Gfi1b is both a direct negative regulator of Rag transcription, and an indirect regulator of the Rag genes via its antagonism of Foxo1 expression (15). Gfi1b directly represses Rag transcription by biding near the Irag2 and Erag elements also bound by Foxo1 (Fig. 6) and recruiting chromatin modifiers that deposit H3K9me2, a chromatin mark associated with transcriptional repression (15). Given the effects of Ebf1 and c-Myb on both Foxo1 expression and Foxo1 binding to the Rag locus, we hypothesized that these factors may also regulate Gfi1b expression. Indeed, knockdown of both Ebf1 and c-Myb resulted in a strong decrease of Gfi1b mRNA levels (Fig. 8A), suggesting that Ebf1 and c-Myb repress Rag and Foxo1 in part by driving Gfi1b expression. Gfi1b has been previously identified as a direct target of Ebf1 in both pro-B and splenic B cells, with ChIP-seq revealing Ebf1 binding sites the Gfi1b promoter (26, 31)(Fig. 8B). Conventional ChIP in cells overexpressing a 3XFLAG-tagged Ebf1 protein revealed strong binding of Ebf1 to this site (Fig. 8C), suggesting that Ebf1’s positive influence on Gfi1b expression is direct.

Figure 8. Ebf1 and c-Myb positively regulate Gfi1b expression.

(A) qPCR measuring Gfi1b transcript levels in sorted AMuLV-transformed B cells transduced with luciferase shRNA (LUC), and either Ebf1 (left) or c-Myb shRNA (right). Data are representative of three independent experiments. (B) Published Ebf1 ChIP-seq data (26, 31) visualized in the UCSC Genome Browser at the Gfi1b locus. An Ebf1 binding site (5′-CCCNNGGG-3′) identified within the Gfi1b promoter peak is shown below, along with its location relative to the Gfi1b TSS. (C) ChIP in AMuLV-transformed B cells overexpressing a 3XFLAG-tagged Ebf1 protein with anti-FLAG or IgG control antibody. qPCR was performed on recovered chromatin to assess Ebf1 occupancy at the Gfi1b promoter site described in B compared to a negative control site in the Rag1 promoter. Data are representative of four independent experiments.

Stat5 represses Rag through upregulation of Ebf1

The IL-7R signaling pathway has been implicated in both the control of Ebf1 expression (32–35) and the repression of Rag transcription (12, 16) during B cell development. Signaling through the IL-7R and transformation by v-Abl both result in the phosphorylation and activation of Stat5 (17, 36). Our IL-7 withdrawal and STI-treatment experiments showed that both IL-7R and v-Abl signaling positively regulate Ebf1 protein expression (Fig. 2). Thus, we hypothesized that Stat5 is the downstream effector in these signaling pathways that upregulates Ebf1 to repress Rag transcription. As with Ebf1 and c-Myb, overexpression of a constitutively active form of Stat5 (CA-STAT5B)(37) in the Rag1-GFP reporter cell line repressed baseline Rag levels and severely blunted STI-induced Rag transcription (Fig. 9A). When we compared Ebf1 protein levels in sorted empty vector control and CA-STAT5B-transduced cells, Stat5 overexpression clearly upregulated Ebf1 (Fig. 9B, left lanes). Upon inhibition of v-Abl with STI, levels of phosphorylated endogenous Stat5a decrease (Fig. 9B, top 100kDa band), and this correlates with a decrease in Ebf1 levels in empty vector control cells as observed in Fig. 2A. However in cells overexpressing CA-STAT5B, Ebf1 expression is rescued and remains high despite v-Abl inhibition and loss of active endogenous Stat5a (Fig. 9B, right lanes). This result is consistent with a model in which activation of Stat5 downstream of v-Abl (and likely the IL-7 receptor) drives high levels of Ebf1 expression that mediate the transcriptional repression of Rag1 and Rag2.

Figure 9. Stat5 represses Rag transcription through upregulation of Ebf1.

(A) GFP expression in AMuLV-transformed Rag1-GFP reporter B cells transduced with constitutively active STAT5B (CA-STAT5B) retrovirus. Uninfected cells (shaded histogram) were distinguished from CA-STAT5B-overexpressing cells (black line) by staining with anti-hCD4 (retroviral marker). Analysis was performed on cells in normal culture conditions (UT) or treated with STI (+STI). Numbers above gate indicate the percentage of GFP+ uninfected cells (top) or CA-STAT5-overexpressing cells (bottom). Data are representative of two independent experiments. (B) Immunoblot measuring Stat5 and Ebf1 protein levels in sorted AMuLV-transformed B cells transduced with empty vector or CA-STAT5B retrovirus. Cells were cultured under normal conditions (UT) or treated with STI (+STI) prior to harvest. Lamin serves as protein loading control. Endogenous Stat5a is distinguished from exogenous CA-STAT5B based on molecular weight. Slower-migrating Stat5a species is the phosphorylated form of the protein (denoted “p-Stat5a”). Numbers below lanes indicate Ebf1/Lamin ratio for each sample. Non-specific band below Ebf1 arises from anti-Stat5 probing. Data are representative of two independent experiments. (C) Model for stage-specific repression of Rag transcription by Ebf1 during B cell development.

Based on our data we propose the following model for the role Ebf1 in regulating Rag transcription during B cell development (Fig. 9C). Initial IL-7R signaling events minimally activate the Ebf1 locus during the early stages of B cell development. Low levels of Ebf1 drive B lineage specification through activation of genes including Foxo1 and Rag and the initiation of Ig heavy-chain gene rearrangement. Enhanced IL-7R signaling in late pro-B cells undergoing transient proliferative bursts upon exposure to high local concentrations of IL-7, or in pre-BCR(+) large pre-B cells undergoing clonal expansion, strongly upregulates Ebf1 expression. Increased Ebf1 dosage drives proliferation through activation target genes involved in cell cycle progression such as cyclin D3 (Ccnd3)(26, 38) while simultaneously repressing Rag transcription via antagonism of Foxo1 and activation of Gfi1b. Thus the promotion of proliferation by increased Ebf1 dosage is tightly linked with repression of the recombination machinery, ensuring genomic integrity is maintained. Upon the attenuation of IL-7R signaling and differentiation to the small pre-B cell stage, Ebf1 protein levels are reduced and Rag expression is upregulated to mediate Ig light-chain gene rearrangement.

Discussion

Restricting expression of the recombinase machinery in developing lymphocytes to non-dividing cells prevents DSBs during S-phase that could be aberrantly repaired and lead to leukemic transformation. During B cell development, signals through the IL-7R and pre-BCR drive proliferation and the repression of Rag expression via targeted degradation of RAG2 protein and transcriptional repression of the Rag genes. We have identified a novel pathway that adds an additional layer of complexity to how this transcriptional repression is accomplished. While this pathway involves upstream components in IL-7 and Stat5 that have been previously implicated in Rag repression, the downstream effectors identified (Ebf1 and c-Myb) paradoxically have been previously described as activators of the Rag locus during early B cell development.

Ebf1 was identified as a B lineage-specific factor that bound and activated the mb-1 promoter (39, 40), and Ebf1-deficient mice revealed its essential role in early B cell development (41). Ebf1 promotes specification to the B lineage in multipotent progenitors via simultaneous activation of B lineage genes and repression of genes involved in the development of other lineages (42, 43). Predicted Ebf1 targets during the earliest stages of development include Rag1, Rag2, and Foxo1, as Ebf1-deficient common lymphoid progenitors (CLPs) display reduced expression of these genes (22). This raises the question of how Ebf1 is converted to a repressor of Foxo1 transcription during later stages of B cell development. Our experiments suggest that Ebf1 dosage may play a role, as unexpectedly we found increased levels of Ebf1 in both AMuLV-transformed cells and IL-7-cultured primary B cells result in decreased Rag expression (Fig. 1). While Ebf1 transcript levels peak in small pre-B cells (24), our data show that Ebf1 protein levels are highest in proliferating pro-B and early pre-B cells where Rag expression is repressed (Fig. 2). IL-7R expression levels vary within the pro-B cell compartment (16), and pre-BCR expression enhances IL-7R responsiveness (44, 45). We hypothesize that brief periods of strong IL-7R signaling in proliferating pro-B and large cycling pre-B cells could cause a spike in Ebf1 dosage that contributes to Rag transcriptional repression. Ebf1 dosage has been shown to be important for transcriptional repression of natural killer (NK) cell-specific genes by Ebf1 (46). Perhaps high levels of Ebf1 protein induced by strong IL-7R signaling allow for the formation of repressive complexes at binding sites in the Foxo1 locus that are not formed under conditions of low Ebf1 expression during earlier stages of development. Stat5 has been shown to simultaneously activate and repress genes in B cells by assembling into distinct dimeric and tetrameric complexes (47). It would be interesting to test biochemically if Ebf1 can form dose-dependent higher-order complexes on Foxo1 locus binding sites, and if such complexes are competent for co-repressor recruitment.

Alternatively, late pro-B and early pre-B cells may express an Ebf1 co-repressor that is not present at appreciable levels during earlier stages of B cell development to allow for stage-specific Foxo1 repression. Interestingly, expression levels of the H3K27 methyltransferase Ezh2, which is recruited to the Ig kappa locus by tetrameric Stat5 complexes to repress premature recombination in cycling pre-B cells (47), increases throughout early B cell development and peaks in Fraction C′ when Rag is repressed (24). As Ebf1 has been shown to target the repressive H3K27me3 chromatin mark to non-B lineage genes it actively represses in pro-B cells (48, 49), it would be interesting to test if a stage-specific interaction between Ebf1 and Ezh2 is responsible for Foxo1 and Rag repression in late pro-B and early pre-B cells.

Our experiments also revealed that Ebf1 protein levels decrease upon differentiation to the small pre-B stage and re-expression of Rag1 and Rag2 (Fig. 2). This occurs despite an increase in Ebf1 transcript levels upon differentiation, suggesting Ebf1 protein is actively targeted for degradation at this developmental transition. Interestingly, analysis of GFP expression in mice expressing an Ebf1(1-148aa)-GFP fusion protein revealed high GFP levels in pro-B cells, but greatly reduced GFP expression in pre-B cells (25). While the authors attribute this to non-specific destabilization of the Ebf1-GFP fusion protein, it is consistent with the decrease in Ebf1 protein levels we observed during pre-B cell differentiation that appears necessary for full re-expression of the Rag genes. Identifying the factors and pathways responsible for this post-translational control of Ebf1 will be of considerable interest as disruption of this process could prevent recombinase re-expression and impair Ig light-chain gene rearrangement in resting pre-B cells.

Previous reports by our group and others implicated c-Myb as a positive regulator of Rag promoter activity (27, 28). While these studies relied heavily on reporter constructs containing cloned Rag locus elements transfected into both B and non-B lineage cells, our analysis of c-Myb’s effects on transcription from the endogenous Rag locus in an AMuLV-transformed B cell line clearly shows this factor represses Rag expression in this context. Analysis of mice in which c-Myb is conditionally deleted in early B cell progenitors, and of mice expressing a hypomorphic c-Myb allele, revealed defects in expression of IL-7R and of other early B lymphoid genes (29, 50). While Rag expression was not affected in the B lineage compartments of these mice, it appeared that similar to Ebf1, c-Myb plays a role in B lineage specification via the activation of a large set of genes important for early B cell development. Thus, how c-Myb is co-opted along with Ebf1 as a repressor of the Rag genes during a later stage of B cell development will be an important question to answer. Unlike Ebf1, c-Myb protein levels are unchanged following STI treatment of AMuLV cells (unpublished observations), suggesting that post-translational modifications or the availability of novel co-repressors control c-Myb activity at this developmental transition. Identifying the factors controlling c-Myb activity, along with the location of functional c-Myb binding sites in the Rag, Foxo1, and Gfi1b loci in B cells, will be essential for understanding the mechanism by which c-Myb represses Rag transcription.

Foxo1 activity is necessary for the two waves of Rag expression that mediate Ig heavy-chain and light-chain gene rearrangement in pro-B and pre-B cells respectively (51). Studies suggest that post-translational mechanisms negatively regulate Foxo1 activity in large pre-B cells, and that upon differentiation this regulation is relieved to allow Foxo1 to activate Rag transcription in small pre-B cells (12, 14). However, microarray analysis with sorted primary cells shows a clear decrease in Foxo1 transcript levels as pro-B cells (Fractions B/C) differentiate to large cycling pre-B cells (Fraction C′), followed by an increase in Foxo1 mRNA in small pre-B cells (Fraction D) (24). When we quantified Foxo1 transcripts in both Abelson-transformed and primary cultured pro-B cells, we observed a significant increase in Foxo1 mRNA levels upon STI- and IL-7 withdrawal-induced differentiation (Fig. 7A), indicating that Foxo1 is also regulated at the transcription level during this developmental transition. Our experiments suggest that the repression of Foxo1 transcription by Ebf1 and c-Myb is an important additional level of control of Foxo1 activity (and thus Rag expression) in proliferating late pro-B and early pre-B cells.

IL-7R signaling and Stat5 have been implicated in the ordered rearrangement of the Ig loci via activation of chromatin in the heavy-chain locus (52, 53) and repression of chromatin in the light-chain locus (47, 54) in pro-B cells. Recent studies have shown that in addition to controlling recombinase accessibility, IL-7R signaling and Stat5 regulate recombination at the level of transcription of the Rag genes (6, 12, 14, 16). Our findings provide additional mechanistic insight into how this negative regulation by Stat5 is accomplished. While potentially acting as a direct negative regulator of Rag transcription (16), we showed that Stat5 also activates an Ebf1-dependent pathway that leads to enhanced proliferation and the transcriptional repression of Foxo1. Thus, in addition to being controlled post-translationally by the PI(3)K-Akt pathway, Foxo1 is also regulated at the transcriptional level by the IL-7R/Stat5 pathway to prevent Rag expression during periods of cellular proliferation.

The transcriptional repressor Gfi1b is an additional downstream factor in the repressive Ebf1-dependent pathway (Fig. 8). While its ability to negatively regulate Rag transcription both directly and through repression of Foxo1 has been described (15), the pathways controlling Gfi1b expression and activity during early B cell development are ill-defined. Gfi1b mRNA levels are inversely correlated with Rag transcription and peak in large cycling pre-B cells (15). This is consistent with a model where enhanced IL-7R signaling and Stat5 activity at this stage upregulates Ebf1 and c-Myb, which in turn drive increased levels of Gfi1b transcription that lead to Rag repression. But while our experiments shed light on how Gfi1b expression might be regulated during this specific stage of B cell development, it has been reported that Ebf1 is a negative regulator of Gfi1b expression in CLPs (43). So again, the paradox of how Ebf1 switches between transcriptional activator and repressor must be solved to better understand how this factor directs gene expression during distinct developmental stages.

Stat5 is constitutively activated in AMuLV-transformed cells (36) and in human hematopoietic malignancies including BCR-ABL(+) acute lymphoblastic leukemia (ALL) (55). Stat5 is essential for ALL as Stat5-deficient hematopoietic progenitor cells cannot be transformed by retroviral expression of v-Abl or BCR-ABL oncogenes (56). While Stat5 itself directly activates genes involved in proliferation and cell survival, our experiments revealed that Ebf1 and c-Myb act downstream of Stat5 to repress differentiation (as evident by their repression of Rag transcription) and promote proliferation and survival of transformed cells. Consistent with other studies (26, 57, 58), reducing Ebf1 and c-Myb levels using shRNAs dramatically reduced proliferation and survival of AMuLV-transformed cells (Fig. S1 and unpublished observations). Thus Ebf1 and c-Myb represent attractive therapeutic targets, and defining the mechanisms regulating their expression and activity during early B cell development will likely contribute to a better understanding of the transformation process.