The E2A gene encodes two E protein/class I basic helix-loop-helix transcription factors, E12 and E47, that are essential for B lymphopoiesis. In addition to the DNA-binding and protein dimerization domain, the E proteins share two highly conserved transcription activation domains. In this study, we show that both activation domains are required for optimal E2A-dependent transcription. Surprisingly, however, neither activation domain is required for E2A to rescue B lymphopoiesis from E2A−/− hemopoietic progenitors, although the N terminus of E2A, which harbors some transcription capacity, is required. Therefore, the E protein activation domains function redundantly in promoting B cell development. In contrast, the N-terminal activation domain, AD1, is required for a newly described ability of E2A to suppress macrophage development in vitro. Our findings demonstrate distinct functionalities for the E protein activation domains in B lymphocytes and macrophages.
Blymphocytes develop from hemopoietic stem cells in the embryonic liver and adult bone marrow (1). This developmental program is marked by the progressive loss of myeloid differentiation potential followed by activation of lymphoid-associated genes and commitment to the B lymphocyte lineage (2). The mechanisms underlying the loss of myeloid potential and induction of the B cell gene program are not fully understood but appear to involve the activation of transcription regulatory proteins that promote or repress gene expression. Activation of the B cell gene program is dependent on numerous transcription factors including Bcl11a, E2A, early B cell factor (EBF1),3 Pax5, Pu.1, and Sox4 (2, 3, 4). E2A, EBF1, and Pax5 are activated in a transcriptional hierarchy with E2A promoting expression of Ebf1 and EBF1 regulating expression of Pax5, although recent studies revealed multiple levels of feedback control within this simple network (2, 5). Nonetheless, the essential function of E2A in this differentiation program is induction of Ebf1 as demonstrated by the ability of ectopic EBF1 to rescue B cell lineage specification and commitment from E2A−/− progenitors (6). E2A cooperates with multiple factors including Pu.1, Ets1, Stat5, and Pax5 to promote Ebf1 transcription (5, 7, 8, 9). Although E2A DNA-binding sequences have been identified in the Ebf1 promoter, precisely how E2A activates Ebf1 transcription remains to be determined (5, 10).
The E2A proteins, E12 and E47, are basic helix-loop-helix (bHLH) transcription factors that differ only in the use of an alternatively spliced exon encoding the bHLH domain. Their bHLH domains share >80% amino acid identity and bind to the same DNA sequence (an E box) and dimerize with the same proteins (11). E2A proteins are related to two additional E proteins in mammals, E2-2 and HEB, and all four proteins have overlapping patterns of expression within the hemopoietic system (12). The E proteins function redundantly and expression of HEB from the E2A gene is sufficient to rescue B lymphopoiesis in E2A−/− mice (13). Although HEB and E2-2 are not essential for B lymphocyte development, they are expressed in these cells and compound E protein-deficient mice (i.e., E2A+/−HEB+/− and E2A+/−E2-2+/−) reveal some function for all E proteins in B cell development (14, 15). Indeed, the ability of EBF1 to rescue B cell development from E2A−/− fetal liver multipotent progenitors (FL MPPs) is dependent on the activity of these other E proteins because complete inhibition of E protein activity in pro-B cells results in growth arrest and apoptosis (6, 16). Therefore, E2A specifically is required for induction of Ebf1 but activation of B cell genes in collaboration with EBF1 may occur with the lower levels of E protein activity conferred by HEB and E2-2 in the absence of E2A.
E2A, HEB, and E2-2 share homology (>40%) within two well-defined transcription activation domains (ADs), referred to as AD1 and AD2, or the loop-helix (LH) AD (17, 18, 19, 20). When fused to a Gal4 DNA-binding domain AD1 and AD2 are able to activate transcription from a reporter construct containing 5 Gal4 DNA-binding sites upstream of a minimal promoter (17, 19). However, the requirement for these ADs has not been investigated in the context of a full-length E protein. In this regard, it is interesting that the Drosophila homolog of E2A, daughterless, has a sequence similar to AD2 but does not contain an AD1-related domain (17). The HEB gene also produces an alternatively spliced form (HEBAlt) that lacks the AD1 domain and instead encodes a novel sequence that confers unique properties on HEB in T lymphocytes (21). In addition, isoforms of E2A that lack AD1 have been described (22). Interestingly, AD1- and AD2-Gal4 fusion proteins were shown to have cell type- specific activity in zebrafish suggesting that they may interact with distinct coactivator complexes (20). These findings raise the possibility that the ADs of E2A may play distinct roles in gene activation in different cell types.
In this study, we demonstrate that both AD1 and AD2 are required for optimal E2A-dependent transcription of an E box reporter construct. Nonetheless, E2A lacking both AD1 and AD2 shows some transcription capacity that is abrogated by deletion of the entire N terminus of E2A. For the most part, stable expression of E2A AD mutants in the 70Z/3m cell line fails to promote strong induction of B cell traits, in contrast to full-length E2A (23). Surprisingly, however, neither AD is required for E2A to rescue Ebf1 transcription and B cell development from E2A−/− FL MPPs, although selection for higher expression of E2A AD mutants with reduced transcription capacity is observed. Therefore, while E2A-dependent transcription is likely essential for B cell lineage specification, the conserved ADs of E2A are not absolutely required. In contrast to pro-B cells, there is a strong selection against ectopic E2A in CD11b+ macrophages developing from E2A−/− FL MPPs and this selection is dependent on AD1 but not AD2. Therefore, while the E2A ADs function redundantly in promoting B cell development, AD1 is specifically required for suppression of CD11b+ cells. Our study is the first to show a differential role for the E2A ADs in B lymphocytes and macrophages.
Materials and Methods
Mice were housed at the University of Chicago in accordance with the guidelines of the University of Chicago Animal Care and Use Committee. E2A−/− mice with a deletion of the E47-specific exon have been described previously and were on a C57BL/6 background (24).
In vitro culture conditions
Livers from mouse embryos at days 12.5–14.5 of gestation were dispersed in PBS/2% FBS and depleted of CD11b+, Gr1+, and Ter119+ cells by incubation with specific biotinylated Abs followed by streptavidin magnetic beads and passage over a magnetic column (LS columns; Miltenyi Biotec). The cells were transduced (1 × 105/ml) with the indicated retrovirus by spinning at 2,500 rpm for 2 h in flat-bottom 24-well plates as described previously (25). The cells were then plated with 30,000 irradiated (1,000 rad) S17 stromal cells/24-well, in OPTI-MEM containing 10% FBS, 100 μg/ml penicillin, 100 μg/ml streptomycin, a 1/100 dilution of IL-7 supernatant (from J558-IL-7 cells), and a 1/500 dilution of c-kit ligand (from CHO-MGF). The cells were transferred to new wells when confluent, generally every 4–5 days.
The pHβAPneoE12, pHβAPneobHLH, and S003-E12 plasmids have been described previously (23, 25). The pHβAPneoΔAD2 was made by PCR amplification of pHβAPneoE12 using the primers 5′ LH: 5′-CGATCTACTCCCCGGATAGATCTG-3′ and 3′ LH: 5′-CCTGAGGCCAGCGCTGATCTAT-3′ followed by digestion with BglII and ligation. The construct was sequenced to ensure correct deletion of the AD2 domain. The ΔAD1 and ΔTA constructs in were created in pSS-Flag by digestion of E12 or ΔAD2 with NarI, made blunt with Klenow, and subsequently digested with XbaI. The fragment was cloned into pSS-Flag digested with BamHI, made blunt with Klenow and then digested with XbaI such that the Flag sequences are in-frame with the E12-coding region. For cloning into pHβAPneo, pSS-ΔAD1 and pSS-ΔTA were digested with XbaI, made blunt with Klenow, and subsequently digested with HindIII before cloning into pHβAPneo digested with BamHI, made blunt with Klenow, and digested with HindIII resulting in pHβAPneoΔAD1 or pHβAPneoΔTA. S003 retrovirus vectors were made by PCR amplification of pSS-Flag plasmids using the primers Flag: 5′-GCTTCAGCATGGACTACAAGGACG-3′ and HLH reverse: 5′-CCATCTAGAGCTGAAAGCACCATCTG-3′ and blunt end ligation into EcoRV digested pBSK. The resulting pBSK construct was digested with XbaI, made blunt with Klenow, and then digested with SalI and the relevant E12 AD mutant was cloned into S003 digested with NotI, made blunt with Klenow, followed by digestion with XhoI. The resulting retrovirus vectors encode the E12 AD mutant followed by and internal ribosomal entry site (IRES) and the coding sequence for GFP.
Approximately 1 × 106 cells were stained in a 200-μl volume of PBS/2% FBS/0.02% NaN3. The Abs used include anti-CD19 (clone 1D3), anti-B220 (clone RA3 6B2), anti-CD11b (clone M1/70), anti-Igμ (clone II/41; from eBioscience or BD Pharmingen). Intracellular stain for E12 was preformed as described previously (25). Data were acquired on a LSRII (BD Biosciences) using FACS Diva software and analyzed using FlowJo (Tree Star).
Two cell lines, Bosc23 and 70Z/3 macrophage, were used for luciferase assays. Bosc23 cells are a derivative of 293T that were created as a retroviral packaging cell line and have a very high transfection efficiency (26). 70Z/3m cells were derived from 70Z/3 pre-B cells by selection for plastic adherence and lack E protein activity but can be converted into pre-B cells by expression of E12 (23, 27). A total of 1.1 × 105 Bosc23 cells/well were grown overnight in 1 ml of complete DMEM medium in a 24-well plate. Cells were transiently transfected using calcium phosphate. Just before transfection, the medium was replaced with 300 μl of fresh medium. A total of 100 ng/ml CMV-Renilla luciferase reporter gene (Promega), used as an internal control, was cotransfected with 0.5 μg/ml EE6 luciferase and 0.5 μg/ml E12 or E12 AD mutants in pHβAPneo. Cell were harvested 48 h after transfection. Protein extracts were prepared by adding 200 μl of passive lysis buffer (Promega). The EE6-luciferase plasmid contains six copies of tandem μE2-μE5 sequences in pGL3 (Promega). Luciferase assays for all transfections were performed using the Dual-Luciferase Reporter Assay System following the manufacturers directions (Promega).
For transfection of 70Z/3m cells, ∼1 × 107 cells were incubated for 20 min in 2.5 ml TS buffer (140 mM NaCl, 5 mM KCl, 25 mM Tris (pH 7.4), 0.6 mM Na2HPO4, 0.7 mM CaCl2, and 0.5 mM MgCl2) containing 75 μl of DEAE-dextran (10 mg/ml), 100 ng of CMV-Renilla luciferase reporter gene, used as an internal control, 5 μg of μE2-E5 reporter construct, and 5 μg of E12 or E12 AD mutants. After transfection the cells were washed twice and cultured for 48 h. Protein extracts were prepared by adding 500 μl of passive lysis buffer to pelleted cells.
Nuclear extracts were made by the method of Dignam et al. (28). A total of 20 μg of nuclear extract was electrophoresed through an 8% SDS-polyacrylamide gel and transferred to Immobulon membranes. The membranes were stained with amido black for 2 min and subsequently washed with 10% methanol and 10% acetic acid to determine that equivalent amounts of nuclear protein had been transferred in each lane. The membranes were blocked for 1 h at room temperature in TBST (50 mM Tris (pH 7.5), 150 mM NaCl, and 0.5% Tween 20) containing 5% milk powder. The blots were then incubated for 2 h in TBST containing a 1/500 dilution of anti-E2A (Ab G98-271; BD Pharmingen). The blots were washed three times for 15 min in TBST before addition of HRP-conjugated goat anti-mouse IgG at a dilution of 1/15,000 (Invitrogen Life Technologies) for 1 h. The blots were washed three times for 15 min in TBST and developed using the ECL detection system (GE Healthcare).
RNA was isolated by TRIzol reagent (Invitrogen Life Technologies) and 10 μg was separated through a 0.75% agarose gel buffered with 18 mM Na2HPO4, 2 mM NaH3PO4, and 6% formaldehyde after incubation for 5 min at 65°C in the same buffer containing 50% formamide. The gel was washed extensively with ddH2O and the RNA was transferred to nytran membrane by capillary action using 10× SSC. The blot was hybridized with random primed, 32P-labeled gene-specific probes in 50% formamide, 5× SSC, 0.15% SDS, 1× Denhardt’s, and 10% dextran sulfate containing 100 μg/ml heat-denatured herring sperm DNA at 42°C overnight. The blots were washed two times for 30 min at 42°C in 2× SSC/0.2% SDS and 2 × 15 min at 60°C in 0.2× SSC/0.2% SDS before exposure to autoradiographic film at −70°C for 12–24 h.
Quantitative real-time PCR (QPCR)
RNA was reverse-transcribed to cDNA using standard techniques and was amplified in an iCycler in a 25-μl reaction containing iQ SYBR Green Supermix from Bio-Rad. Expression was calculated for each gene relative to Hprt using the ΔCT method. The primers were designed to span introns of >1 kb where possible and include: Ebf1, Pax5, and Hprt. The D-JH PCR was performed exactly as described previously (29).
Protein extracts and EMSA
Nuclear extracts were prepared as described (28). Protein concentrations were determined using Bio-Rad’s protein assay reagent. dsDNA probes were end labeled using T4 polynucleotide kinase and purified over a G25-Sepharose column. EMSAs using the μE5 oligos (sequence) were performed as described previously (30). The DNA-protein complexes were resolved by electrophoresis through a 5% polyacrylamide gel containing 45 mM Tris, 45 mM boric acid, and 1 mM EDTA, dried, and then exposed to autoradiographic film at −70°C.
Both AD1 and AD2 contribute to E2A-dependent transcription activation
E2A has two ADs that, when fused to a Gal4 DNA-binding domain, activate transcription from a luciferase reporter containing Gal4 DNA-binding sites (Fig. 1,A) (17, 18, 31). However, the necessity for these domains in the context of the full-length E2A protein has not been investigated. Therefore, we cloned cDNA-encoding mutant versions of E12 that lack either the AD1 (ΔAD1) or AD2 (ΔAD2) domain, or both domains (ΔTA) into the pHβAPneo vector containing the β-actin promoter, which drives expression in all mammalian cells (Fig. 1,A). Wild-type (WT) and E12 AD mutant expression vectors were transfected into Bosc23 cells along with an E box reporter that promotes expression of firefly luciferase (EE6 luciferase). We used Bosc23 cells for this assay because of their high transfection efficiency (>90%) and low level of spontaneous activation of the EE6-luciferase reporter. As an internal standard for transfection efficiency, we cotransfected the cells with a plasmid in which the CMV promoter drives expression of Renilla luciferase (pCMV-RL). We found that deletion of either AD reduces E box-dependent transcription by ∼4- to 5-fold compared with WT E12, and loss of both ADs results in an additional 2-fold decrease in luciferase activity (Fig. 1,B). Deletion of the remaining N-terminal sequences upstream of the bHLH domain (bHLH) led to a further 4-fold decrease in transcription, which is indistinguishable from the control pHβAPneo vector producing neomycin but no E12-related protein (Fig. 1, A and B). Each of the E12 proteins was produced efficiently in Bosc23 cells, as determined by western blot analysis (Fig. 1 C). Therefore, in the context of E12 both the AD1 and AD2 domains contribute substantially to transcription activation and they function cooperatively because loss of one domain has a similar consequence as loss of both domains. In addition, our data suggest that a minor transcription activation capacity remains in the N-terminal portion of E12 outside of the conserved ADs because ΔTA is able to induce transcription significantly above that of the bHLH domain.
The ADs of E2A, when fused to Gal4, have cell type-specific activity in zebrafish (20). Therefore, we tested the requirement for AD1 and AD2 in another cell line, 70Z/3m, which lacks functional E proteins and can be converted to the B cell lineage by expression of E12 (23). The pattern of transcription activation in 70Z/3m cells was similar to that observed in Bosc23 cells, with deletion of either AD resulting in an approximate 4-fold decrease in luciferase activity when compared with WT E12 (Fig. 1 D). In addition, ΔTA showed only a further 1.5-fold decrease whereas the bHLH domain did not induce luciferase activity above the level of the control vector. Therefore, both of the conserved ADs are functional in Bosc23 and 70Z/3m cells.
Characterization of 70Z/3m cells stably expressing E12 AD mutants
We have shown previously that stable expression of E12 in 70Z/3m cells is sufficient to induce B cell genes and suppress some myeloid-associated gene expression (23). To examine the role of AD1 and AD2 in promoting this macrophage to B cell lineage transition, we created stable 70Z/3m cells expressing each of the E12 AD mutants. We generated neomycin-resistant clones after transfection of each of the E12 AD mutants into 70Z/3m cells and three to six clones of each mutant were isolated for further expansion and analysis. We screened each of the clones for expression of E12 by a combination of intracellular flow cytometry, Western blot, and EMSA and selected cells with the highest expression of the E12 protein for our subsequent analysis (Fig. 2, and data not shown). We were able to obtain E12-expressing 70Z/3m lines (m/E12) with a range of E12 protein with m/E12 2C1 expressing the highest and m/E12 D3 having the lowest concentration of protein, as described previously (Fig. 2,A) (23). Transfection of these m/E12 lines with the EE6-luciferase reporter confirmed that these cells had increased E protein activity compared with m/neo clones (stably expressing neomycin but no E12 protein) and the level of reporter activation was consistent with the amount of E12 expressed in each clone (Fig. 2,B). However, we note that the amount of E12 protein produced on a per cell basis in these stable lines is much lower than that produced per cell after transient transfection of pHβAPneo-E12. Therefore, the level of luciferase activity produced by EE6 luciferase after transfection into the stable clones is much lower than in assays where both E12 and EE6 luciferase were added transiently (compare Fig. 1,D and Fig. 2 B). Interestingly, however, we find that the amount of E12 is not directly proportional to the induction of B cell traits because m/E12 D3 has higher expression of Pax5 mRNA than m/E12 2C6 but m/E12 2C6 expresses more Ebf1 (23). This observation suggests that Pax5 mRNA is not directly dependent on the amount of E12 in these cells.
To assure ourselves that we were examining E12 AD-mutant clones that expressed the E12 AD-mutant protein at levels comparable to E12 clones that could undergo lineage conversion, we examined E box-binding activity by EMSA. Our previous experiments revealed that m/E12 clones express Id2, a characteristic of 70Z/3m, which could antagonize E box binding by E12 AD mutants in these cells (23). Therefore, we selected clones that showed E box binding that was comparable or higher than that in m/E12 D3, a level of E12 that is able to induce B cell lineage conversion (Fig. 2,C). To determine whether binding to the μE5 E box sequence was specific, the protein extracts were preincubated with excess cold μE5 probe, and additional experiments incorporated an Ab against human E2A that shifted each of the E12 containing complexes (see Fig. 4 C, and data not shown). Therefore, we conclude that the level of E box-binding activity in E12 AD-mutant clones is comparable to or higher than m/E12 D3 cells.
Analysis of B cell and macrophage traits in 70Z/3m cells expressing E12 AD mutants
We next examined the m/E12 AD-mutant clones by flow cytometry for surface Igμ, a characteristic of 70Z/3 pre-B cells. The 70Z/3m, m/ΔTA and m/bHLH cells lacked detectable surface Igμ whereas Igμ was highly expressed on m/E12 2C1 (Fig. 3,A). Notably, m/ΔAD1 expressed surface Igμ although the intensity of staining was lower than that on m/E12 2C1 cells, and m/ΔAD2 had even lower, but detectable, amounts of surface Igμ (Fig. 3,A). Therefore, loss of either AD impairs surface Igμ expression but deletion of both ADs results in a failure to express detectable surface Igμ. QPCR analysis revealed that the m/ΔAD1 and m/ΔAD2 clones have reduced amounts of Ebf1 mRNA compared with m/E12 2C1 or m/E12 D3 (Fig. 3,C). Interestingly, while m/bHLH cells did not express Ebf1 mRNA above the level in 70Z/3m the m/ΔTA cells consistently showed a low level of Ebf1 mRNA expression (Fig. 3,C). This observation suggests that the transcriptional activity remaining in the N-terminal portion of the ΔTA protein is sufficient for low level activation of Ebf1 in the 70Z/3m line and this activity may be partially masked by the conserved ADs. In contrast, while Pax5 mRNA is detectable in 70Z/3m the amount of Pax5 mRNA also remained low, but detectable, in all of the AD-mutant clones (Fig. 3 D). Therefore, the E12 AD mutants are poor activators of B cell genes in 70Z/3m.
To further examine the lineage-related changes in 70Z/3m cells expressing E12 AD mutants, we performed FACS analysis for CD11b, a marker of 70Z/3m that is down-regulated after expression of E12 (Fig. 3,B) (23). Interestingly, the m/ΔTA and m/bHLH cells expressed CD11b indicating that at least one of the E12 ADs is required for down-regulation of CD11b in 70Z/3m cells (Fig. 3,B). In contrast, both m/ΔAD1 and m/ΔAD2 showed reduced expression of CD11b similar to m/E12 2C1 (Fig. 3 B). Therefore, AD1 and AD2 function redundantly to promote the down-regulation of CD11b in 70Z/3m cells. In contrast, we did not find decreased expression of any other macrophage genes in E12 AD-mutant clones including mRNA encoding lysozyme or c-fms, as was detected in m/E12 clones (23). Therefore, suppression of most macrophage genes may require increased E12 transcription capacity or, more likely, is a consequence of E12-induced lineage conversion. Taken together, our results indicate that loss of either E12 AD domain impairs the ability of E12 to promote B cell lineage conversion of 70Z/3m cells.
Ectopic E12 promotes B cell development and suppresses development of macrophages from E2A−/− FL MPPs
Our studies in 70Z/3m cells are limited by the fact that this is a transformed cell line, and transition from the macrophage to the B cell fate may have different requirements in these cells than B cell and macrophage development from a multipotent progenitor. Therefore, we examined the activity of the E12 AD mutants in an established assay of B cell and macrophage development from FL MPPs (32). We have shown previously that transduction of E2A−/− FL MPPs with a retrovirus producing E12 rescues development of pro-B cells in vitro in S17 stromal cell plus cytokine-supplemented cultures (6). Indeed, B220+ CD19+ pro-B cells can be detected in cultures of E2A−/− Lin- FL MPPs by days 9 and 10 of culture (Fig. 4,A) (6). In contrast, the bHLH domain of E12 does not promote development of pro-B cells from E2A−/− FL MPPs (Fig. 4,A). The retroviral vectors used in these experiments produce E12- or bHLH- from an mRNA that also codes for GFP downstream of an IRES. Therefore, virus-transduced cells can be detected by expression of GFP. Remarkably, only 14.5% (2.1/12.4 + 2.1 × 100) of CD19−GFP+ cells from E12 virus-transduced cultures were CD11b+ whereas 63% (53.8/53.8 + 31.6 × 100) of CD19−GFP− cells were CD11b+ in the same cultures indicating that CD11b+ cells are specifically depleted from the E12 virus-transduced population. This observation suggests that E12 expression is incompatible with macrophage development (Fig. 4,B). In contrast, neither the bHLH-producing virus nor the control retrovirus (producing GFP only) affected development of CD11b+ cells (Fig. 4,B). The decline in the percent of cells with high forward scatter (FSC) by side scatter (SSC) properties and the increased percent of cells with low FSC by SSC properties in the CD19−GFP+ population of E12- vs bHLH-transduced cultures further suggests that E12 prevents development of macrophages as opposed to simply suppressing CD11b expression because macrophages have a high FSC/SSC profile (Fig. 4 C). Our data indicate that ectopic expression of E12 promotes B cell and suppresses macrophage development from E2A−/− FL MPPs and that E12 function is critically dependent on sequences N-terminal to the bHLH domain.
The E12 ADs are not required for B cell development from E2A−/− FL MPPs
We tested the ability of each of the E12 AD mutants to rescue B cell development from E2A−/− FL MPPs in vitro. Two days after transduction, GFP+ virus-transduced cells could be detected in all cultures (Fig. 5,A). To our surprise, by day 9 of culture, B220+CD19+ cells could be detected in cultures of E2A−/− FL MPPs transduced with the ΔAD1, ΔAD2, or ΔTA producing virus indicating that neither of the E2A ADs is essential for E12 to promote B cell development (Fig. 5,B). The B220+CD19+ cells developing in these cultures expand well in response to IL-7 and expressed Ebf1, Pax5, Igll1, and Rag1 mRNA indicating that they are committed pro-B cells (Fig. 5,C, and data not shown). In addition, D-JH rearrangement was detected in each of these cultures further demonstrating that the cells had differentiated to the pro-B cell stage (Fig. 5 D). Therefore, E12 mutants lacking the AD1 and/or AD2 ADs are able to promote B cell development from E2A−/− FL MPPs.
The retroviral vector that we are using is capable of producing a wide range of protein levels as observed by the heterogeneous expression of GFP early after infection (Fig. 5,A). Because E12 and GFP are produced from the same mRNA the level of GFP is an indirect measure of the amount of E12. We questioned whether the B220+CD19+ cells that are rescued by E12 AD-mutant proteins would be selected for higher concentrations of E12 protein, potentially compensating for decreased transcription capacity. Indeed, we found that the level of GFP, in CD19+ cells at day 10 of culture was highest in ΔTA-expressing E2A−/− cells (mean fluorescence intensity (MFI) = 126) and lowest in E12-expressing CD19+ E2A−/− cells (MFI = 54.1), with ΔAD1- and ΔAD2-expressing cells showing an intermediate level of GFP (MFI = 95.7 and 64.4, respectively) (Fig. 6 A). Therefore, there appears to be a selection for cells with higher expression of the E12 AD mutant-IRES-GFP sequence when the E12 AD mutant has reduced transcription activation capacity. Taken together, our data indicate that E12 proteins lacking both AD1 and AD2 are able to promote pro-B cell development from E2A−/− FL MPPs but higher concentrations of protein may be required to support B lymphopoiesis.
Interestingly, we also observed that a majority of E12-expressing E2A−/− (E2A−/−E12) pro-B cells lose expression of GFP after attaining CD19 expression (Fig. 6,B and data not shown). In E2A−/−ΔTA virus-transduced cells ∼30% of CD19+ cells were GFP− on day 12 of culture whereas nearly 60% of E2A−/−E12 CD19+ cells were GFP− (Fig. 6,B). By day 19 of culture ∼90% of E2A−/−E12 CD19+ cells were GFP− whereas <50% of E2A−/−ΔTA CD19+ cells had reduced GFP expression (data not shown). The E2A−/−ΔAD2 CD19+ cells also showed a lower percent of GFP−CD19+ cells, ∼34% on day 12 of culture (Fig. 6,B). However, the E2A−/−ΔAD1 CD19+ cells showed a slightly higher proportion of cells that down-regulate GFP, in this experiment 74%, although in different experiments this varied between 40 and 74% (Fig. 6 B and data not shown). These observations suggest that there may be selection for cells with lower amounts of E12 protein after cells attain the CD19+ pro-B cell stage.
To determine whether this decrease in GFP correlated with a decrease in E12 protein we examined E12 by intracellular staining with an anti-human E2A Ab. We can identify cells with two different levels of E12 protein in rescued E2A−/− CD19+ cells (Fig. 6,C). Nonetheless, the lowest amount of E12 in the CD19+ population from E12, ΔAD1, or ΔAD2 virus-transduced E2A−/− cells is above the level observed in control virus-transduced cells indicating that some E12, or E12 AD-mutant protein is produced in all CD19+ cells (Fig. 6 C). Moreover, the percent of cells with the highest amount of E12 correlates well with the percent of GFP-positive cells (data not shown). The observation that E12 protein concentration decreases after cells attain the pro-B cell stage is consistent with our previous observation that initiation of B cell development requires high concentrations of E protein but lower concentrations are sufficient to promote survival and pro-B cell gene expression in synergy with EBF1 after B cell lineage commitment (6). Our observations suggest that there may also be an upper bound on the amount of E12 that is compatible with survival or expansion of pro-B cells.
The AD1 domain of E12 is required for suppression of CD11b+ cells
In addition to promoting B cell development, ectopic expression of E12 in FL MPPs suppresses the outgrowth of CD11b+ cells (Fig. 4,A). On day 10 of culture essentially no CD19−GFP+ cells from E12 virus-transduced cultures express CD11b whereas 48% of the CD19−GFP− cells in the same cultures were CD11b+ (Fig. 7, A and C). In contrast, in ΔTA transduced cultures there appeared to be little selection against GFP+CD11b+ cells because CD19−GFP+ and CD19−GFP− cells express CD11b at equivalent frequencies (Fig. 7, A and C). Similarly, in ΔAD1 virus-transduced cells there appears to be little selection against CD11b+ cells (Fig. 7, A and C). In contrast, we consistently observed a decrease in CD11b+ cells in the CD19−GFP+ population of ΔAD2 virus-transduced cultures. Only 14% of CD19−GFP+ cells were CD11b+ whereas 46% of CD19−GFP− cells were CD11b+. In addition, ΔAD2- but not ΔAD1- or ΔTA-expressing cells showed a decreased percent of cells with a high FSC/SSC profile and an increased percent of cells with a low FSC/SSC profile (Fig. 7,B). Interestingly, however, more ΔAD2 cells showed high FSC/SSC than E12-expressing cells, although both were lower than ΔAD1- or ΔTA-expressing cells (Fig. 7 B). Therefore, we conclude that the AD1 domain of E12 is necessary and sufficient to effectively repress development of macrophages in E2A−/− FL MPP cultures. This data contrasts with the requirements for E12 to rescue pro-B cell development, which is not dependent on either of the conserved ADs.
We show for the first time that both AD1 and AD2 are required for optimal E box-dependent transcription by an E protein in at least two different cell types. However, a mutant version of E12 that lacks both AD1 and AD2 retains some transcription capacity that is terminated by removal of sequences N-terminal to the bHLH domain. Although AD1 and AD2 function cooperatively to promote E box transcription, neither AD is required for E12-dependent rescue of B cell development from E2A−/− FL MPPs in vitro. Nonetheless, transcription activation by E12 likely plays a role in E12’s ability to promote B cell development because the bHLH domain alone is not sufficient and there is a selection for increased concentrations of E12 AD mutants in rescued GFP+ pro-B cells. These observations indicate that the E12 ADs provide cooperative but redundant functions to promote B cell development. In contrast, the AD1 domain is required for E12-dependent suppression of CD11b+ macrophages in vitro. Therefore, our results indicate that the E2A ADs provide distinct functions in pro-B cells and macrophages. These findings highlight the importance of cell context for the functions of, and requirements for, the E2A ADs.
In the 70Z/3m cell line, E12 induces a transition from a macrophage to a B cell gene program, although some macrophage traits remain in these converted pre-B cells (23). Low levels of Ebf1 mRNA were detected in the m/ΔAD1 and m/ΔTA clone and both m/ΔAD1 and m/ΔAD2 showed a low level of surface Igμ and repressed CD11b surface expression. However, none of the E12 AD mutants were able to efficiently induce the macrophage to pre-B cell transition. Indeed, all of the E12 AD mutants remain tightly adherent to plastic and failed to induce Pax5 or repress most myeloid genes. These findings suggest that in the context of the 70Z/3m cell line both ADs are necessary for E12-dependent lineage conversion.
In contrast to 70Z/3m cells, primary E2A−/− FL MPPs express the E proteins HEB and E2-2, which can dimerize with E12 (11). Therefore, while our data clearly indicate that neither AD1 nor AD2 are required for E12 to promote B lymphopoiesis, we cannot conclude that E protein-dependent transcription activation is not essential. One possibility is that dimerization of the E12 AD-mutant proteins with full-length HEB or E2-2 is sufficient to produce transcription competent E protein complexes. However, the bHLH domain alone does not support B cell development indicating that DNA-binding competent E protein complexes containing the HEB or E2-2 ADs alone are not sufficient. Moreover, the observed selection for higher amounts of E12 AD mutants with lower transcription capacity suggests that E12 functions as a homodimer to rescue B lymphopoiesis and that a minimal level of transcription activation is essential. We favor this second model and suggest that 70Z/3m cells fail to express E12 AD-mutant proteins at sufficiently high concentrations to allow lineage conversion whereas the range of protein concentrations permitted by retroviral-transduction of primary cells allows for selection of high amounts of E12 AD mutants to support B cell development. Regardless of which scenario is correct, our data suggest that AD1 and AD2 provide redundant functions for initiation of B lymphopoiesis.
Our data are consistent with the hypothesis that all of the ADs in E12 (AD1, AD2, and the unknown AD in ΔTA) function similarly to promote B lymphopoiesis. The AD1 domain has been shown to interact with the coactivator proteins p300/CBP and the SAGA histone acetyltransferase complex through an LDFS motif (18, 33, 34). A recent study demonstrated that the AD2 domain of E2-2, which is highly conserved with the E2A AD2 domain, also interacts with p300/CBP (35). Moreover, this study showed that haploinsufficiency of p300 exacerbates the effects of E2A haploinsufficiency in B lymphopoiesis in vivo, indicating that p300 is an essential coactivator of E2A dependent genes in B lymphocytes. Therefore, we predict that AD1 and AD2 function redundantly to stabilize p300/CBP at target promoters. It will be of interest to determine whether p300/CBP interacts with AD1 and AD2 through distinct domains, because cooperative binding of p300/CBP to E2A through multiple interactions could influence the stability of coactivator recruitment at a target promoter. The KIX domain of p300/CBP binds to AD1 but additional studies will be needed to determine which domains mediate p300/CBP-AD2 interaction (33). We propose that ΔTA may also be able to interact with p300/CBP, possibly with low affinity, explaining its low transcription activation potential. Moreover, in vivo multiple transcription factors that recruit p300/CBP may interact with the promoters of essential B-lineage genes; therefore loss of an E2A AD may have only a minor impact on the stability of p300/CBP at the promoter. Indeed, selection for increased amounts of E12 AD mutants may partially overcome decreased coactivator affinity by increasing occupancy of the promoter and thereby helping to stabilize p300/CBP in cooperation with other transcription factors. Notably, the initial induction of the Ebf1 gene, which is a critical target of E2A in B lymphopoiesis, is regulated by multiple transcription factors including at least E2A, Stat5, and Ebf1 itself (5).
Interestingly, we found that there was selection against high concentrations of ectopic E12 in CD19+ pro-B lymphocytes cultured in vitro. This observation suggests that at high concentrations E12 may be detrimental for growth or survival of pro-B lymphocytes. Consistent with this hypothesis we have also found that E12 protein transduced directly into WT pro-B lymphocytes cultured in vitro is rapidly selected against (data not shown). These observations are consistent with the finding that ectopic expression of E2A in E2A−/− T cell lymphomas leads to growth arrest and apoptosis (25). Moreover, loss of E2A leads to T lymphocyte progenitor cell transformation, and has recently been implicated in B cell lineage acute lymphoblastic leukemia, indicating that E2A may function to repress lymphocyte growth and/or survival (36, 37, 38, 39). Combined with our previous observation that inhibition of E protein activity in pro-B cells results in growth arrest and apoptosis, we conclude that a narrow range of E protein activity is compatible with pro-B cell proliferation and survival (16, 40).
In contrast to B cell development, the AD1 domain appears to be critical for E2A to repress development of CD11b+ cells. Interestingly, the transcription repressor ETO, as well as the oncogene AML-ETO, interacts with the AD1 domain in myeloid cells (41). ETO interacts with AD1 though a sequence that overlaps with the LDFS motif and therefore competes with p300/CBP for binding to E2A. Although this interaction would be predicted to reduce E2A function, it may not abolish E protein activity because the AD2 domain could still interact with p300/CBP. However, it is likely that ETO directly represses E protein-driven gene expression because it associates with histone deacetylases that can modify chromatin accessibility (42). Therefore, the failure of ΔAD1 to repress development of CD11b+ cells may be related to an inability to inhibit specific genes required for macrophage differentiation. Alternatively, E2A may recruit novel coactivators specifically through AD1 in macrophages, possibly as a dimer with class II bHLH proteins expressed in these cells, and thereby activate genes which repress macrophage differentiation (11). The observation that lymphoid lines derived from E2A−/− mice retain the ability to differentiate into myeloid cells is consistent with a need for E2A to repress macrophage differentiation during specification and commitment to the lymphoid lineages (43, 44).
Taken together, our data reveal redundant functions for the E2A ADs in promoting B cell development and further demonstrate that neither AD is absolutely essential provided that sufficient E2A protein is produced. In contrast, AD1 is required for E2A to suppress macrophage development demonstrating distinct functions for these ADs in B cells and macrophages.
We thank Amelia Santos and Laura Glasgow for technical assistance; Ryan Duggan and David LeClerc in the Immunology Applications Core Facility for assistance with flow cytometry; and members of the Kee laboratory for discussions and comments on the manuscript.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by the National Institutes of Health (R01 CA099978, to B.L.K.).
Abbreviations used in this paper: EBF, early B cell factor; bHLH, basic helix-loop-helix; FL MPP, fetal liver multipotent progenitor; AD, activation domain; LH, loop helix; IRES, internal ribosomal entry site; QPCR, quantitative real-time PCR; WT, wild type; FSC, forward scatter; SSC, side scatter; MFI, mean fluorescence intensity.