Lineage-specific activators affect β-globin locus chromatin in multipotent hematopoietic progenitors
2006; Springer Nature; Volume: 25; Issue: 15 Linguagem: Inglês
10.1038/sj.emboj.7601232
ISSN1460-2075
AutoresStefania Bottardi, Julie Ross, Natacha Pierre‐Charles, Volker Blank, Éric Milot,
Tópico(s)Acute Myeloid Leukemia Research
ResumoArticle13 July 2006free access Lineage-specific activators affect β-globin locus chromatin in multipotent hematopoietic progenitors Stefania Bottardi Stefania Bottardi Guy-Bernier Research Centre, Maisonneuve-Rosemont Hospital and Faculty of Medicine, University of Montreal, Montreal, Quebec, Canada Search for more papers by this author Julie Ross Julie Ross Guy-Bernier Research Centre, Maisonneuve-Rosemont Hospital and Faculty of Medicine, University of Montreal, Montreal, Quebec, Canada Search for more papers by this author Natacha Pierre-Charles Natacha Pierre-Charles Guy-Bernier Research Centre, Maisonneuve-Rosemont Hospital and Faculty of Medicine, University of Montreal, Montreal, Quebec, Canada Search for more papers by this author Volker Blank Volker Blank Lady Davis Institute for Medical Research, McGill University, Montreal, Quebec, Canada Department of Medicine, McGill University, Montreal, Quebec, Canada Search for more papers by this author Eric Milot Corresponding Author Eric Milot Guy-Bernier Research Centre, Maisonneuve-Rosemont Hospital and Faculty of Medicine, University of Montreal, Montreal, Quebec, Canada Institute for Research in Immunology and Cancer (IRIC), University of Montreal, CP Succursale Centre-ville, Montreal, Quebec, Canada Search for more papers by this author Stefania Bottardi Stefania Bottardi Guy-Bernier Research Centre, Maisonneuve-Rosemont Hospital and Faculty of Medicine, University of Montreal, Montreal, Quebec, Canada Search for more papers by this author Julie Ross Julie Ross Guy-Bernier Research Centre, Maisonneuve-Rosemont Hospital and Faculty of Medicine, University of Montreal, Montreal, Quebec, Canada Search for more papers by this author Natacha Pierre-Charles Natacha Pierre-Charles Guy-Bernier Research Centre, Maisonneuve-Rosemont Hospital and Faculty of Medicine, University of Montreal, Montreal, Quebec, Canada Search for more papers by this author Volker Blank Volker Blank Lady Davis Institute for Medical Research, McGill University, Montreal, Quebec, Canada Department of Medicine, McGill University, Montreal, Quebec, Canada Search for more papers by this author Eric Milot Corresponding Author Eric Milot Guy-Bernier Research Centre, Maisonneuve-Rosemont Hospital and Faculty of Medicine, University of Montreal, Montreal, Quebec, Canada Institute for Research in Immunology and Cancer (IRIC), University of Montreal, CP Succursale Centre-ville, Montreal, Quebec, Canada Search for more papers by this author Author Information Stefania Bottardi1,‡, Julie Ross1,‡, Natacha Pierre-Charles1, Volker Blank2,3 and Eric Milot 1,4 1Guy-Bernier Research Centre, Maisonneuve-Rosemont Hospital and Faculty of Medicine, University of Montreal, Montreal, Quebec, Canada 2Lady Davis Institute for Medical Research, McGill University, Montreal, Quebec, Canada 3Department of Medicine, McGill University, Montreal, Quebec, Canada 4Institute for Research in Immunology and Cancer (IRIC), University of Montreal, CP Succursale Centre-ville, Montreal, Quebec, Canada ‡These authors contributed equally to this work *Corresponding author. Guy-Bernier Research Centre, Maisonneuve-Rosemont Hospital and Faculty of Medicine, University of Montreal, 5415 boulevard l'Assomption, Montreal, Quebec, Canada H1T 2M4. Tel.: +1 514 252 3551; Fax: +1 514 252 3430; E-mail: [email protected] The EMBO Journal (2006)25:3586-3595https://doi.org/10.1038/sj.emboj.7601232 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info During development, the regulated expression of tissue-specific genes can be preceded by their potentiation, that is, by chromatin activation in progenitor cells. For example, the human β-like globin genes are potentiated in a gene- and developmental-specific manner in hematopoietic progenitors. Developmental regulation of human β-gene expression in erythroid cells is mostly determined by transcriptional activators; however, it is not clear how gene-specific potentiation is set in hematopoietic progenitors. Using human and transgenic multipotent hematopoietic progenitors, we demonstrate that human β-globin locus activation is characterized by TBP, NF-E2, CBP and BRG1 recruitment at both the Locus Control Region and human β-gene promoter. Our results further indicate that in hematopoietic progenitors, EKLF influences chromatin organization at the human β-globin locus and is instrumental for human β-gene potentiation. Thus, we show that lineage-specific transcriptional activators expressed at basal levels in progenitor cells can participate in gene potentiation. Introduction The expression of lineage-specific genes can be potentiated before transcriptional activation (Bonifer, 1999). Gene potentiation is characterized by an active chromatin organization in multipotent progenitor cells (Jimenez et al, 1992; Bottardi et al, 2003) and can be accompanied by partial occupancy of gene regulatory regions by transcriptional activators. Gene expression in differentiated cells requires synergy among activators, co-activators, general transcription factors, chromatin remodeling and histone modifying complexes (Struhl, 2005). In hematopoietic stem cells (Ye et al, 2003) and in multipotent hematopoietic progenitor cells (HPC) (Hu et al, 1997), the 'promiscuous' expression of lineage-specific factors such as PU.1, C/EBPα, PAX5, SCL, GATA-1, GATA-2 and GATA-3 does not alter the biological potential of these cells, but ostensibly provides a proper environment for stochastic lineage commitment mainly by repressing or activating groups of genes (Cantor and Orkin, 2002; Graf, 2002). However, it is not known if promiscuously expressed transcriptional activators influence gene potentiation in multipotent progenitor cells. In mammals, the human β-(huβ-) globin locus is a well-characterized multigenic locus, and therefore a good model to identify mechanisms that control epigenetic regulation of gene expression during hematopoiesis. The huβ-globin locus consists of five developmentally regulated genes (ε–Gγ–Aγ–δ–β), whose high-level expression in erythroid cells (EryC) depends upon the Locus Control Region (βLCR) comprised of four erythroid-specific DNaseI hypersensitive sites (HS). The βLCR enhances gene transcription through direct interaction with globin promoters (Carter et al, 2002; Tolhuis et al, 2002; Palstra et al, 2003). Even though DNaseI sensitivity at the endogenous murine locus is not lost when the βLCR is deleted (Bender et al, 2000), in EryC the human βLCR is a major determinant of chromatin organization at the huβ-globin locus (Grosveld et al, 1987). In progenitor cells, the β-globin locus is in an active chromatin organization (Jimenez et al, 1992) and potentiation of huβ-like genes, most likely mediated by as yet unidentified transcriptional activators, is developmental-specific (Bottardi et al, 2003). EKLF, GATA-1 and the p45 subunit of NF-E2 (p45) are among the best-characterized transcriptional activators involved in huβ-gene regulation in EryC. EKLF is essential for adult β-globin gene transcription and EKLF knockout (KO) mice suffer from severe anemia and die at approximately 14.5 d.p.c. (days postcoitus) (Nuez et al, 1995; Perkins et al, 1995). EKLF binds to βLCR and globin promoters (Bieker, 2001) and it is required for βLCR-β major (βmaj) gene direct interaction (Drissen et al, 2004). EKLF recruits the erythroid-specific SWI/SNF chromatin remodeling complex 1 (E-RC1) to the β-globin locus (Armstrong et al, 1998) and the absence of EKLF leads to reduced DNaseI HS formation at the βmaj and huβ-promoters (Wijgerde et al, 1996). GATA-1 binds to βLCR HSs and globin promoters and is critical for primitive and definitive erythroid cell differentiation (Pevny et al, 1991; Fujiwara et al, 1996). It interacts with CBP, SWI/SNF, ACF/WCRF and NuRD complexes (Blobel et al, 1998; Kadam et al, 2000; Hong et al, 2005; Rodriguez et al, 2005). GATA-1 is also required for βLCR-βmaj direct interaction (Vakoc et al, 2005). NF-E2 is comprised of a ubiquitously expressed subunit, MafK, and the hematopoietic-specific activator p45 (Andrews et al, 1993), and it is recruited to mouse and human βLCR and promoter regions (Daftari et al, 1999; Forsberg et al, 2000; Sawado et al, 2001; Leach et al, 2003). NF-E2 is necessary for globin gene expression in differentiated MEL (mouse erythroleukemia) cells but p45 KO mice exhibit only mild effects on erythropoiesis and no significant influence on globin gene expression (Shivdasani and Orkin, 1995). Interestingly EKLF, GATA-1 and p45 are also promiscuously expressed in hematopoietic progenitors (Hu et al, 1997). p45 could be involved in priming/potentiation of the mouse α-globin locus (Anguita et al, 2004), whereas PU.1–GATA-1 interaction influences lineage commitment (Cantor and Orkin, 2002; Graf, 2002). However, it has never been investigated whether these factors play roles in developmental-specific potentiation of the huβ-gene and globin locus activation in multipotent HPC. To understand the molecular events leading to huβ-like globin gene potentiation, we studied the influence of promiscuously expressed transcriptional activators on recruitment of general transcription factors as well as chromatin remodeling and histone modifying activities at βLCR HS2, huγ- and huβ-promoter in primary multipotent HPC. Our results, obtained with human cells and transgenic mice expressing the huβ-globin locus, suggest that EKLF is instrumental for globin gene potentiation in HPC, facilitating p45, CBP, BRG1 and TBP recruitment at the huβ-promoter. Finally, we provide insight into the cooperative role of EKLF and p45 for promoting appropriate chromatin activation in HPC, as well as for Pol II recruitment and subsequent transcriptional elongation in EryC. Based on these results, we suggest a model whereby EKLF is a key regulator of huβ-gene potentiation in HPC. Results HPC purification and characterization Mouse HPC were isolated from 13.5 d.p.c. fetal livers of line 2 mice (ln2 HPC), which are transgenic for the entire huβ-globin locus and express the huβ-globin genes normally (Strouboulis et al, 1992). Mouse HPC corresponds to a Ter119−, Gr-1−, B220−, CD31high population, which represents 1–2% of the total fetal liver and does not contain mature or late committed cells (see below). The sorted population was always ⩾98% pure. The hematopoietic potential of these cells was ascertained by in vitro clonal assays in methylcellulose (Table I, ln2 HPC). On average, out of 100 colonies, 85 originated from progenitors with multilineage potential (CFU-GEMM: colony forming unit-granulocyte, erythrocyte, megakaryocyte, macrophage; CFU-GM: colony forming unit-granulocyte, megakaryocyte), whereas 15 showed unilineage commitment (BFU-E: Burst forming unit-erythrocyte). No CFU-E (colony forming unit-erythrocyte) was detected. Table 1. Clonal assays in methylcellulose: 150 HPC purified from 13.5 d.p.c. ln2 or ln2 EKLF−/− fetal livers were plated onto methylcellulose plates Clonogenic ability of Ter119−, Gr-1−, B220−, CD31high cells CFU-GEMM (%) CFU-GM (%) BFU-E (%) Colonies/103 cells ln2 HPC 38.9 44.8 16.3 210±40 ln2 EKLF−/− HPC 34 48.2 17.8 240±35 To further characterize this population, ln2 HPC cDNA was used to study the expression of marker genes. SCL, GATA-1, p45, GATA-2, EKLF, C/EBPα and globin transcripts are all expressed, whereas transcripts corresponding to GATA-3 are not detected (Figure 1A; ln2 HPC). This expression pattern suggests that the Ter119−, Gr-1−, B220−, CD31high cell population is mainly composed of multipotent HPC (Akashi et al, 2000), namely common myeloid and megakaryocyte/erythrocyte lineage-restricted progenitors, a finding supported by our in vitro clonal assays (see above). As evaluated by quantitative real-time reverse transcriptase (RT)–PCR, the levels of p45, EKLF and huβ-globin mRNA are, respectively, ∼7-, ∼6- and ∼100-fold higher in EryC than in HPC (Supplementary Figure 1). Accordingly, Western blot analyses revealed that GATA-1, p45 and EKLF protein levels are approximately, 12-, 6- and 11-fold higher in EryC than in HPC (Figure 1B). Figure 1.Expression of marker genes in ln2 HPC and ln2 EKLF−/− HPC. (A) Semiquantitative RT–PCR performed on equal amounts of RNA purified from ln2 HPC or ln2 EKLF−/− HPC. PCR samples were resolved onto agarose gel. γ-Globin: fetal human (γ) and mouse embryonic (βH1) transcripts; β-globin: adult human (β and δ) and mouse (βmin and βmaj) globin transcripts; NPM: ubiquitously expressed nucleophosmin transcript, used as internal control; Neg: negative control; Pos: positive control; (B) Western blot analysis of ln2 HPC, ln2 EKLF−/− HPC and ln2 EryC; 4 and 8 μg of whole-cell protein extract were loaded in each lane of a 10% SDS–PAGE. Anti-GATA-1 and -p45 antibodies were purchased from SantaCruz; anti-actin antibodies were purchased from LabVision; anti-EKLF serum is a generous gift of S Philipsen. Protein levels in ln2 HPC versus ln2 EKLF−/− HPC or ln2 EryC versus ln2 HPC were calculated using actin as internal control and they are shown on the right side of each panel together with their standard error of means; (C–E) representative examples of quantitative real-time RT–PCR; the relative level of GATA-1 or p45 gene expression in ln2 EKLF−/− HPC versus ln2 EKLF HPC were calculated according to Pfaffl (2001), using mouse actin as internal control and expressed as KO/WT ratio; x axis: cycle number; y axis: derivative of SYBR Green fluorescence. Blue dots: ln2 HPC; green triangles: ln2 EKLF−/− HPC. Download figure Download PowerPoint Since an EKLF KO background is extensively used in this study, we investigated whether the absence of EKLF causes a general failure of cellular differentiation during hematopoiesis because EKLF regulates other genes crucial for final EryC differentiation (Drissen et al, 2005). HPC purified from 13.5 d.p.c. ln2 EKLF KO fetal livers (ln2 EKLF−/− HPC) represent 1–2% of total fetal liver (as observed in ln2 HPC) and, as expected (Nuez et al, 1995; Perkins et al, 1995), give rise to erythroid colonies in methylcellulose (Table I). Furthermore, the gated populations and marker gene expression analyses (with the exception of globin genes) do not reveal any apparent difference between ln2 HPC and ln2 EKLF−/− HPC (Figure 1A). In particular, GATA-1 and p45 expression levels, as evaluated by quantitative real-time RT–PCR and Western blot analysis, do not change significantly among the two populations (Figure 1B–E). Thus, the absence of EKLF does not preclude normal hematopoietic differentiation of HPC. Transcriptional activators recruitment at the huβ-globin locus in HPC Our previous results suggested that in bone marrow-derived ln2 HPC and human CD34+ cells, chromatin at the huβ-globin LCR is in an active state, and the huβ-promoter is potentiated for transcriptional activation (Bottardi et al, 2003). We now investigate how histone covalent modification and chromatin remodeling activities are recruited at HS2, huβ- and huγ-promoters during hematopoiesis. Chromatin immunoprecipitation (ChIP) assays were performed on 13.5 d.p.c. fetal liver-derived hematopoietic cells. In 13.5 d.p.c. fetal livers, the huβ-gene is expressed and huγ-genes are mostly silent as judged by the fact that huγ-transcripts represent ∼10% of total huβ-globin level (Strouboulis et al, 1992). HS2 was chosen because it is important for high-level huβ-gene expression in EryC (Morley et al, 1992), and moreover HS2 deletion can abrogate epigenetic regulation of globin genes in EryC (Milot et al, 1996) as well as locus activation in HPC (Bottardi et al, 2005). Histone acetylation of the huβ-globin locus in fetal liver ln2 HPC (Figure 2B; Supplementary Figure 2A) is similar to the pattern observed in bone marrow HPC (Bottardi et al, 2003), suggesting that βLCR and huβ-promoter are in an active/potent chromatin organization also in 13.5 d.p.c. fetal liver HPC. As shown in Figure 2C, GATA-1 is not significantly crosslinked at HS2, huβ- or huγ-promoters while, as reported previously (Anguita et al, 2004), HS-12 of the mouse α-globin locus is slightly enriched (Supplementary Figure 3A). p45 is detected at HS2 and huβ-promoter in ln2 HPC and in CD34+ cells (Figure 2D; Supplementary Figure 3B). p45 is also detected at huγ-promoters in ln2 HPC (Figure 2D). Thus, p45 is recruited to the huβ-globin locus at early stages of hematopoiesis. Figure 2.Histone H3 acetylation, GATA-1 and p45 recruitment at the huβ-globin locus in HPC. (A) A map of the huβ-globin locus; genes are shown as black boxes and the location of βLCR HSs is indicated by arrows. Amplified regions used for ChIP analyses are indicated by grey boxes. (B–G) ChIP assays: immunoprecipitated and unbound (input) chromatin samples were subjected to duplex radioactive PCR with one primer set specific for huβ-globin regions and another primer set specific for mouse zfp37 (ZFP, zinc-finger protein 37) or human pax6 (pax6, paired box protein 6) regulatory regions ('Ctl', indicated by arrows), two genes that are not expressed in hematopoietic cells. All PCR reactions were performed in parallel under conditions of linear amplification. Products were quantified by PhosphorImager. The level of enrichment of globin regions relative to the control and input samples is represented by bars, with their corresponding standard deviations. A value of 1 indicates no enrichment. Mouse THP/ZFP or human THP/pax6 controls are included (THP, kidney-specific Tamm-Horsfall gene) to confirm that no enrichment is detected at regulatory regions of non-hematopoietic genes. To obviate for weak signals, THP/ZFP PCR reactions were run longer but always in conditions of linear amplification. Representative gel images are shown below each panel; AcH3: anti-acetylated histone H3 antibodies; dark blue bars: ln2 HPC; dark blue hatched bars: human CD34+ cells; light blue bars: ln2 EKLF−/− HPC. Download figure Download PowerPoint Since EKLF is expressed at basal levels in HPC, we investigated whether it can influence huβ-gene potentiation in HPC. Commercially available antibodies against EKLF are not suitable for ChIP analysis (Feng and Kan, 2005). Thus, we used a genetic approach in comparing ln2 HPC with ln2 EKLF−/− HPC. As shown in Figure 2E, histone H3 acetylation at the huβ-globin locus decreases significantly in ln2 EKLF−/− HPC relative to ln2 HPC, whereas no major differences are detected at HS-29 of the mouse α-globin locus (Supplementary Figure 3C). (Unless specified, 'significant' refers to a statistically significant difference between wild type and KO cells, based on unpaired student's t-test P-values <0.05). As in ln2 HPC, GATA-1 is not detected at HS2 or globin promoters in ln2 EKLF−/− HPC (Figure 2F). p45 binding to HS2 and huβ-promoter is significantly affected in ln2 EKLF−/− HPC (Figure 2G), suggesting that EKLF can influence p45 recruitment to these regions and that p45 as well as EKLF participate in locus organization in HPC. Recruitment of histone modifying and chromatin remodeling activities at the huβ-globin locus in HPC Based on the above results, we investigated whether the role of p45 and EKLF in globin potentiation includes their capacity to recruit histone modifying and chromatin remodeling activities to the globin locus. CBP is detected at HS2 and huβ-promoter in ln2 HPC and CD34+ cells (Figure 3A; Supplementary Figure 3D). However, CBP binding is significantly reduced at huβ-promoter in ln2 EKLF−/− HPC (Figure 3B). p45 (Hung et al, 2001) and EKLF (Zhang and Bieker, 1998) interact with and are acetylated by CBP, but unlike p45 (Johnson et al, 2001), the ability of EKLF to recruit CBP to any region of the globin locus has never been demonstrated. The fact that in ln2 EKLF−/− HPC H3 acetylation and p45 recruitment are affected at both HS2 and huβ-promoter (Figure 2E and G), whereas CBP binding is reduced only at huβ-promoter, suggests that the absence of EKLF could either: (i) directly impede CBP recruitment at the huβ-promoter; or (ii) preclude p45 and CBP recruitment at the huβ-promoter with consequent hypoacetylation. Therefore, the role of p45 in locus acetylation was verified in HPC purified from 13.5 d.p.c. ln2 p45 KO fetal livers (ln2 p45−/− HPC). The absence of p45 does not significantly change histone H3 acetylation level at HS2 (data not shown), whereas H3 acetylation and CBP recruitment at the huβ-promoter significantly decrease in ln2 p45−/− HPC (Supplementary Figure 4). Thus, it appears that p45-mediated recruitment of CBP, which is facilitated by EKLF, contributes to histone acetylation of the huβ-promoter in HPC. The detection of histone H3 acetylation at HS2 in ln2 p45−/− HPC suggests that other activator/s capable of interacting with histone acetyltransferases can bind HS2 in HPC. Figure 3.CBP, BRG1, TBP and Pol II recruitment at the huβ-globin locus in HPC. Ln2 HPC, human CD34+ cells, or ln2 EKLF−/− HPC were subjected to ChIP analyses with anti-CPB (A, B), -BRG1 (C, D), -TBP (E, F), or -Pol II (G, H) antibodies. Analysis and quantification of immunoprecipitated samples were performed as described in Figure 2. Dark blue bars: ln2 HPC; dark blue hatched bars: human CD34+ cells; light blue bars: ln2 EKLF−/− HPC. Download figure Download PowerPoint The remodeling complex E-RC1 contains, among other proteins, the yeast homologue BRG1 and EKLF (Armstrong et al, 1998). We therefore wondered whether EKLF might be part of a remodeling complex in HPC. Since the EKLF-BRG1 interaction is sufficient to remodel a chromatin-assembled β-globin promoter (Kadam et al, 2000), and BRG1 is the first among the SWI/SNF components to be recruited to promoter regions, we verified whether EKLF could influence huβ-gene potentiation through BRG1 recruitment and hence chromatin remodeling. The comparison between ln2 HPC or CD34+ cells and ln2 EKLF−/− HPC (Figure 3C and D) reveals a significant reduction of BRG1 binding at huβ-promoter in ln2 EKLF−/− HPC. As expected, BRG1 binding at the c-myc promoter (Nagl et al, 2006) is similar in ln2 HPC versus ln2 EKLF−/− HPC (Supplementary Figure 3E). Based on these results, we propose that in HPC EKLF facilitates chromatin remodeling of the huβ-promoter through BRG1 recruitment. Transcription preinitiation complex assembly and globin gene potentiation in HPC The preinitiation complex (PIC) includes Pol II, general transcription factors, for example TBP, and additional cofactors. Basal levels of huβ-gene expression in HPC (Figure 1A) prompted us to ask whether gene potentiation in these cells requires PIC assembly. TBP was detected at huβ-promoter in ln2 HPC and CD34+ cells (Figure 3E). Even though bound to the huγ-promoter in CD34+ cells (see Discussion), TBP is not found at huγ- (Figure 3E) or huε- (data not shown) promoters in ln2 HPC. Thus, globin potentiation in ln2 HPC is developmental-specific and is associated with TBP recruitment at the huβ-promoter before the onset of high-level transcriptional activity. This is supported by our results obtained in EKLF KO cells. In EKLF KO mice, the switch from γ-to-β-globin expression is delayed; indeed, around 13.5 d.p.c. γ-globins are expressed in a greater number of fetal liver EryC derived from these mice relative to the situation for wild-type counterparts (Perkins et al, 1996; Wijgerde et al, 1996). Consequently, in ln2 EKLF−/− HPC the TBP binding significantly decreases at huβ-promoter, whereas it increases at huγ-promoters (Figure 3F). It has been proposed that NF-E2−TAFII130 interaction facilitates PIC assembly at globin promoters (Amrolia et al, 1997). Accordingly, in ln2 p45−/− HPC, TBP recruitment at the huβ-promoter is significantly lower than in wild-type cells (Supplementary Figure 4). Altogether, these results suggest that EKLF, either directly or through p45, contributes to huβ-gene potentiation also by modulating the developmental-specific recruitment of TBP at the huβ-promoter in HPC. Ln2 HPC were then subjected to ChIP analyses with an antibody that binds Pol II in a phosphorylation-independent manner. Pol II is detected at HS2, but not at the huβ-, huγ- (Figure 3G and H) or βmaj- (Supplementary Figure 5) promoters, suggesting that HS2-bound Pol II is not efficiently transferred to globin promoters in ln2 HPC or ln2 EKLF−/− HPC. However, Pol II is crosslinked at huγ-promoters in CD34+ cells (see Discussion). Thus, the presence of TBP but not Pol II suggests that the PIC is partially assembled at the huβ-promoter and contributes to globin gene potentiation in HPC. Basal levels of globin gene expression in HPC Globins are expressed at basal levels in multipotent progenitors (Figure 1A) (Hu et al, 1997; Bottardi et al, 2003). EKLF is essential for huβ-gene expression in EryC, but it is not known whether EKLF is also important for basal levels of globin gene expression in HPC. By quantitative real-time RT–PCR (Figure 4), we show that huβ-gene expression is 10-fold lower and huγ-gene expression is 4-fold higher in ln2 EKLF−/− HPC than in ln2 HPC. Thus, fetal liver-derived HPC express huβ- and huγ-globin genes and the absence of EKLF favors huγ- to the detriment of huβ-gene basal levels of expression. Figure 4.Quantitative real-time RT–PCR of ln2 HPC and ln2 EKLF−/− HPC. Representative examples of huβ- (A) and huγ-gene (B) expression were evaluated by real-time RT–PCR. The relative levels of globin gene expression in ln2 EKLF−/− HPC versus ln2 HPC were calculated according to Pfaffl (2001), using mouse actin as internal control (as described in Figure 1) and expressed as KO/WT ratios. Blue dots: ln2 HPC; green triangles: ln2 EKLF−/− HPC. Download figure Download PowerPoint Chromatin organization at the huβ-globin locus in EryC EKLF is required for enucleation and final erythrocyte formation during in vitro culture of definitive erythrocytes, whereas EKLF is dispensable for the first steps of erythroid cell differentiation (Table I; Drissen et al, 2005). Wright–Giemsa staining of ln2 EKLF wild type or ln2 EKLF KO 13.5 d.p.c. fetal livers revealed that, as expected (Perkins et al, 1995), late erythroblasts and enucleated red cells are less represented in 13.5 d.p.c. EKLF KO than wild-type fetal liver. However, only minor variations in the ratio of erythroid versus nonerythroid cells and in the percentage of primitive erythroid cells are observed among the two backgrounds (Supplementary Figure 6). In view of the above, ln2 EKLF wild type (ln2 EryC) as well as ln2 EKLF KO (ln2 EKLF−/− EryC) 13.5 d.p.c. fetal livers were used for ChIP analyses. First, compared with ln2 EryC, H3 lysine 4 (K4) dimethylation and H3 acetylation levels in ln2 EKLF−/− EryC (Figure 5A and B), though still considerable (except for K4 dimethylation at huβ-promoter), are lower at HS2, huγ- and huβ-promoters. GATA-1 and p45 are crosslinked to HS2 and huβ-promoter in ln2 EryC (Figure 5C and D). However, in ln2 EKLF−/− EryC, GATA-1 recruitment at the huβ-promoter (Figure 5C) and p45 recruitment at HS2 and huβ-promoter (Figure 5D) are reduced. Likewise, in ln2 EKLF−/− EryC CBP, BRG1, TBP and Pol II recruitment are affected at HS2 and huβ-promoter (Figure 5E–H). Pol II binding is also reduced at the βmaj promoter in ln2 EKLF−/− EryC (Supplementary Figure 5). Interestingly, in ln2 EKLF−/− EryC, TBP (Figure 5G) and Pol II (Figure 5H) binding to huγ-promoters is greater than in ln2 EryC. These results are in agreement with previous observations indicating that γ-genes are expressed at higher levels in EKLF KO than EKLF wild-type 13.5 d.p.c. fetal livers (Perkins et al, 1996; Wijgerde et al, 1996). To ascertain the significance (P<0.05) of the variations in histone acetylation as well as p45, CBP, and BRG1 recruitment in ln2 EryC versus ln2 EKLF−/− EryC, other regions were analyzed (Supplementary Figure 7A–D). Since no significant differences are observed at any of the additional regions analyzed, it is likely that the EKLF effect at the β-globin locus is not related to a global defect in transcriptional activation. Thus, the defect in huβ-gene expression documented in ln2 EKLF−/− EryC is most likely due to abnormal recruitment of chromatin proteins and histone modifying/chromatin remodeling activities, transcriptional activators, TBP, and Pol II at the βLCR, and, in particular, at the huβ-globin promoter. Figure 5.Chromatin organization at the huβ-globin locus in EryC. (A–I) ChIP analyses of ln2 EryC (dark orange bars) or ln2 EKLF−/− EryC (light orange bars). Analysis and quantification of immunoprecipitated samples, as well as antibodies used for ChIP assays are as described in Figures 2 and 3; MeK4: anti-lysine 4 dimethylated histone H3 antibodies; PCTD: anti-phospho-Ser2 Pol II CTD antibodies. Download figure Download PowerPoint Pol II phosphorylation and transcriptional elongation In the absence of EKLF, transgenic mice carrying the huβ-globin locus do not express mouse or human adult globin genes in EryC (Perkins et al, 1996; Wijgerde et al, 1996). Nonetheless, we detect Pol II at the huβ-promoter in ln2 EKLF−/− EryC (Figure 5H). To elucidate this conflicting result, ln2 EryC or ln2 EKLF−/− EryC were subjected to ChIP analyses with an antibody that recognizes phospho-Ser2 at the C-terminal domain (CTD) of Pol II (PCTD). It is known that the overall phosphorylation level of Pol II increases during transcriptional elongation. In fact, during promoter clearance and elongation, Pol II CTD is phosphorylated and this post-translational modification renders Pol II a transcription- and elongation-competent enzyme. As shown in Figure 5I, PCTD detection at the 3′end of the huβ-gene (Huβ3) is
Referência(s)