SATB1 family protein expressed during early erythroid differentiation modifies globin gene expression

The human β-globin gene cluster on chromosome 11 consists of 5 developmentally specific genes for embryonic (ϵ), fetal (^Gγ, ^Aγ), and adult (δ, β) globins. A strong enhancer, located in the far upstream region of the cluster called the locus control region (LCR), contains 5 DNase I hypersensitive (HS) sites and is able to enhance tissue-specific globin gene expression and provide a high level of transcription activity from human globin gene constructs in transgenic mice. Transcription factors such as erythroid Krüppel-like factor (EKLF), GATA-1, and NF-E2, that bind to the LCR and other regulatory elements, and promoters in the globin gene locus, have been reported to regulate chromatin histone acetylation by associating with histone acetyltransferases.^1-3  The LCR is required to increase the rate of transcription but may be dispensable for formation of an open chromatin domain of a downstream active globin gene in erythroid cells.^4,5  For globin gene expression, spatial organization of the β-globin cluster requires special interactions between distal transcriptional elements in the LCR and downstream active globin genes. Some developmental specificity between individual hypersensitive sites in the LCR and downstream globin genes is evident such as the interaction between HS2 and ϵ-globin for transcription activation.⁶ 

Complex packaging of eukaryotic chromosomes in nuclei creates chromatin loops and matrix/scaffold attachment regions (MARs/SARs; the term MARs is used here), originally identified as gDNA fragments that remain tightly associated with salt-extracted and DNase I-digested nuclei, have been postulated to be localized at the base of chromatin loops.⁷  MARs identified by such criteria often contain a base-unpairing region (BUR), the DNA bases of which become continuously unpaired when subjected to negative superhelical strain.^8,9  Many candidate MARs in the β-globin cluster appear to be in regions of mass binding sites for transcription factors, some of which are specific to the developmental stage.^10,11  MARs found flanking the ϵ- or γ-globin genes or within the β-globin second intervening sequence (IVS2) have been proposed as regulatory elements for specific globin gene expression or hemoglobin switching.^12-16 

Special AT-rich binding protein 1 (SATB1) binds to double-stranded BUR sequences, specifically recognizing a specialized DNA context (an ATC sequence context), characterized by a cluster of sequence stretches with well-mixed As and Ts but either Cs or Gs exclusively on one strand (designated as ATC sequences).¹⁷  SATB1 has roles in tissue-specific organization of DNA sequences, in regulation of gene expression by acting as a “landing platform” for chromatin-remodeling enzymes and in designation of the region-specific histone modification in vivo.^18,19  In vitro studies have indicated that SATB1 family protein can bind to some of the MARs in the β-globin gene cluster.^12,14,15  Here, we show that overexpression of SATB1 in K562 cells induces hemoglobin and globin gene expression concomitant with changes in chromatin structure and that SATB1 interacts directly with MARs in vivo in the β-globin cluster at LCR HS2 and at the ϵ-globin promoter region. These data suggest that during early erythroid development SATB1 may provide a previously unknown mechanism underlying differential globin gene regulation.

Human erythroleukemia K562 cells (American Type Culture Collection, Manassas, VA) were cultured in RPMI 1650/10% fetal bovine serum.²⁰  Human primary erythroid progenitor cells were purified using Ficoll-Hypaque (BioWhittaker, Walkersville, MD) from blood obtained from consenting healthy volunteers through the National Institutes of Health (NIH) Department of Transfusion Medicine and cultured as described.²¹  Approval was obtained from the NIH institutional review board for these studies. Informed consent was provided according to the Declaration of Helsinki.

An SATB1 expression vector was constructed by excising SATB1 cDNA (EcoR1) from pECHAT1146,¹⁷  ligating into pIRES2-EGFP (Clontech, Palo Alto, CA) to give pEGFP/SATB1; accuracy was confirmed by DNA sequencing. For stable cell lines, pEGFP/SATB1 or pIRES2-EGFP was transfected by electroporation into K562 cells.²²  Clones were selected using geneticin (500 μg/mL; Gibco, Grand Island, NY). For reporter gene assays, 5.0 × 10⁵ HeLa cells or 5.0 × 10⁶ K562 cells were transfected using Superfect reagent (Qiagen, Valencia, CA). After 72 hours, cells were harvested and assayed. PSV-β-galactosidase (Promega, Madison, WI) was cotransfected for normalization of transfection efficiency. CREB-binding protein (CBP; RSVCBP), E1A, and a mutant E1A (E1AΔ2-36), and SATB1 (pEGFP/SATB1) expression vectors were used for cotransfection with reporter gene constructs. pEGFP/SATB1 was transfected into primary erythroid progenitor cells by electroporation.

The ϵ-globin promoter with the 5′ flanking M_ϵ MAR (-445 to +17) or without (-365 to +17) was inserted between the KpnI and HindIII sites of the reporter vector pREP4/Luc (from Keji Zhao, NHLBI, National Institutes of Health, Bethesda, MD), an episomal vector containing the Epstein-Barr virus replication origin and encoding nuclear antigen EBNA-1 required for replication,²³  to create pREP4/ϵ or pREP4/Δϵ. The mutation of M_ϵ MAR was created using sense and antisense M_ϵ MAR mutant oligonucleotides (GACGGTACCGGGGTAGGGGGAGAGGGGCGCCGGTATCTAGAGGC) and ligation of the annealed oligonucleotides to the ϵ-promoter. The HS2 enhancer region with or without the HS2-M1 MAR-binding site (with MAR, 8244-8862; without MAR, 8461-8862) was inserted upstream of the ϵ-globin promoter in pREP4/ϵ between the Xba1 and BglII sites to create pREP4/HS2-ϵ and pREP4/ΔHS2-ϵ. Mutation of HS2-M1 MAR was created by synthesis of sense and antisense HS2-M1 MAR mutant oligonucleotides (CGCTCTAGACAGAGCACAGGAGAAGGAAGGGGGAGGGGGGAGGGGGGTACCTGG).

Total RNA was isolated and first-strand cDNA was synthesized using MuLV reverse transcriptase (RT) and oligo-d (T)₁₆ (Applied Biosystems, Foster City, CA). For quantitative real-time polymerase chain reaction (PCR) analysis, gene-specific primers and fluorescent labeled TaqMan probes (6-carboxy fluorescein [FAM] as the 5′ fluorescent reporter, tetramethylrhodamine [TAMRA] as 3′ end quencher) were used in a 7700 Sequence Detector (Applied Biosystems, Foster City, CA) as described.²⁴  All results were normalized to human β-actin.

Cell lysates were obtained by adding RIPA buffer (10 mM Tris [tris(hydroxymethyl)aminomethane] HCl, 1 mM EDTA [ethylenediaminetetraacetic acid], 0.1% sodium dodecyl sulfate [SDS], 0.1% Na₃VO₄, 1% Triton-X 100) and protease inhibitor (Roche Diagnostics, Mannheim, Germany) into the cell pellet, incubated on ice for 30 minutes and centrifuged at 17 000g for 10 minutes. The protein sample was run on NuPAGE 4% to 12% Bis-Tris Gel (Invitrogen, Carlsbad, CA) for 1 hour at 200 V. Protein was transferred to nitrocellulose by standard methods. The blot was blocked with 5% nonfat milk in Tween 20-Tris-buffered saline (TTBS) buffer for 1 hour at room temperature, probed with primary antibody for 1 hour at room temperature, washed in TTBS buffer, probed with horseradish peroxidase (HRP)-conjugated secondary antibody for 1 hour at room temperature, and rinsed in TTBS buffer for chemiluminescent detection.

K562 and K562/SATB1 nuclear extracts were prepared for electromobility shift assay (EMSA). Nuclei were extracted from washed cells using hypotonic buffer (10 mM HEPES [N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid], 1.5 mM MgCl₂, 10 mM KCl, 0.5 mM dithiothreitol [DTT], and 0.5 mM phenylmethylsulfonyl fluoride [PMSF]) and centrifuged. The cytoplasm containing supernatant was discarded, the pellet resuspended in 2 × volume of extraction buffer (20 mM HEPES, 25% glycerol, 0.42 M NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 0.5 mM DTT, 0.5 mM PMSF), placed on ice for 30 minutes, and centrifuged (10 000rpm; 10 minutes), and the supernatant containing the nuclear extract collected. For EMSA, DNA probes were labeled with γ-³²P-adenosine triphosphate (ATP) by reaction with T4 polynucleotide kinase (New England BioLabs, Beverly, MA). Probe binding to nuclear extract was carried out at room temperature for 30 minutes in binding buffer (20 mM Tris HCl, pH 8, 100 mM NaCl, 1 mM EDTA, 5% NP-40, 10 mM DTT, and 0.1 mg/μL bovine serum albumin). DNA-protein complexes were visualized using electrophoresis in a 4% nondenaturing polyacrylamide gel and autoradiography. For GATA-1 EMSA, reaction buffer contained 10 nM HEPES, 50 nM potassium glutamate, 5 mM MgCl₂, 1 mM EDTA, 2 mM DTT, 5% glycerol, and 1 μg poly(dI-dC).

Chromatin extracts were prepared as described.²⁵  In brief, 2 × 10⁷ cells were fixed with formaldehyde and incubated at 37°C for 3 minutes, washed in phosphate-buffered saline (PBS), resuspended in 15 mL Triton buffer (0.1 M Tris HCl, 0.05 M EDTA, 0.01 M EGTA [ethylene glycol tetraacetic acid], 0.25% Triton X-100) and incubated for 15 minutes. Triton-washed cells were centrifuged (1000g for 5 minutes), resuspended in 15 mL NaCl buffer (0.1 M Tris HCl, 0.01 M EGTA, 0.05 M EDTA, 5 M NaCl), and incubated for an additional 15 minutes. The samples were centrifuged, resuspended in 1 mL sonication buffer (0.1 M Tris HCl, 0.05 M EDTA, 0.01 M EGTA, 1% SDS), and sonicated for 10 bursts of 10 seconds. Cell debris was removed by centrifugation (17 000g for 5 minutes) and supernatant stored at -80°C as chromatin extracts. For chromatin immunoprecipitation (ChIP) analysis, specific antibodies against SATB1, acetylated or methylated isoforms of histone 3 and 4 (Upstate Biotechnology, Lake Placid, NY), or preimmune serum and 40 μL protein A-Sepharose suspension were added to the chromatin extract, incubated at 4°C overnight, and washed. Bound and input chromatin samples were placed in 0.5% (wt/vol) SDS and incubated overnight at 65°C to reverse the formaldehyde cross-linking. DNA was further purified by phenol-chloroform extraction and ethanol precipitated using glycogen (10 μg) as a carrier.

For DNA sequence-specific quantification by real-time PCR, primers and fluorescent-labeled TaqMan probes were used. DNA (2 ng) from the ChIP selected (IP) fraction and 2 ng gDNA as a reference control (Ref) were used as templates. Using sequence-specific primers and TaqMan probes for quantitative real-time PCR, at low amplification the threshold cycle number (Ct) is directly proportional to the amount of corresponding specific DNA in the sample and is in the linear range. Each cycle represents a 2 × amplification of the amount of product. Enrichment of the specific sequences in IP was compared with Ref and calculated from the difference of the threshold cycle number (Ct) for the respective DNA pools. Specifically, IP/Ref = 2^{(Ct (Ref) - Ct (IP))} is used to determine the fold difference.²⁶ 

K562 nuclear extract (100 μg) was incubated with SATB1 or CBP antibodies, or preimmune serum and protein A-Agarose (Santa Cruz, Biotechnology, Santa Cruz, CA) in 1 mL binding buffer (0.5% NP-40, 10 mM Tris HCl, 150 mM NaCl, 2 mM EDTA, 10% glycerol, protease inhibitor) for 3 hours at 5°C. The reaction mixture was briefly centrifuged, the pellet washed in 1 mL binding buffer 3 times at 5°C for 5 minutes, and the immunoprecipitated SATB1/CBP or CBP/SATB1 complexes were separated on a polyacrylamide gel. Anti-SATB1 or anti-CBP antibody was used for Western blot analysis of bound proteins.

SATB1 sense and antisense oligonucleotides flanking the translation start site (5′-GCCTCGTTCAAATGATCCATACTCAGTC-3′) were synthesized with a phosphorothioate backbone and purified by high-performance liquid chromatography (HPLC; Synthegen, Houston, TX). Fresh antisense or sense (control) oligonucleotide was added to the primary erythroid progenitor cells at day 1 and day 3 of phase 2 culture and the cells were harvested at day 5.

Statistical analysis was carried out by standard methods. Error bars used throughout indicate SD of the mean.

To investigate the influence of SATB1 on globin gene regulation, a SATB1 expression vector was stably transfected into K562 cells that endogenously express a low level (compared with T lymphocytes) of SATB1 (or its isoform),^14,27  to generate K562/SATB1. Western blotting confirmed the increase in SATB1 expression (Figure 1A). Surprisingly, a red cell pellet clearly indicated a marked elevation in hemoglobin production in the K562/SATB1 cells without hemin (Figure 1A). Benzidine staining shows hemoglobinization increasing from 4% (control K562 population) to 48% in the K562/SATB1 cells (Figure 1B). Additional stable K562/SATB1 clones were isolated and analyzed. SATB1 levels, determined by Western blotting, correlated with hemoglobin production measured spectrophotometrically (Figure 1C). Because globin protein expression is primarily transcriptionally regulated, quantification of globin mRNA reflects the amount of globin produced. Globin gene expression was determined using gene-specific primers and TaqMan probes (Table 1). K562 cells express predominantly γ-globin and ϵ-globin and low levels of β-globin mRNA (Figure 1D). Quantitative RT-PCR analysis indicated that overexpression of SATB1 increased ϵ-globin transcripts to 250% compared to stable transfection with the vector control (mock) and reduced γ-globin transcripts while leaving β-globin unchanged (Figure 1D). Induction of corresponding hemoglobin was confirmed by HPLC. SATB1 increased ϵ-globin gene expression in a dose-dependent manner (Figure 1E).

We previously showed that increasing GATA-1 in K562 cells decreases ϵ-globin expression, whereas increasing GATA-2 increases ϵ- and γ-globin expression.²⁸  SATB1 did not affect GATA-1 or GATA-2 production. Western blot analysis of K562, K562/SATB1, and K562 mock cells showed comparable amounts of GATA-1 and GATA-2 (Figure 1F). EMSA determined GATA-1 binding in nuclear extract from K562 cells and several clones of K562/SATB1 cells and indicated that increasing SATB1 only modestly affected GATA-1 binding to DNA in K562/SATB1 clones (Figure 1G, lanes 8-11), compared to the reported 10-fold or more increase by hemin.^29-31  As controls for GATA-1 binding, cold probe (G) and GATA-1 antibody (αG) displaced GATA-1 binding, whereas a mutant GATA-1 competitor (ΔG) had no effect (Figure 1G, lanes 1-7).

Hemin induction of K562 cell erythroid differentiation increases hemoglobin production and γ-globin gene expression.³²  Using hemin-induced K562 nuclear extract for EMSA, protein binding to the SATB1 probe (S), Wt (25)₇, consisting of the core sequence of the BUR from the immunoglobulin heavy-chain enhancer (5′-(TCTTTAATTTCTAATATATTTAGAA)₇-3′)¹⁷  markedly decreased after a 72-hour exposure to hemin (Figure 2A). Western blotting confirmed a reduction in protein detected using the anti-SATB1 antibody with hemin treatment (Figure 2B). Although the predicted molecular mass of SATB1 is 85.9 kDa, SATB1 from thymus (T lymphocytes) migrates at 103 kDa on SDS-polyacrylamide gels,³³  and the detected protein in K562 cells migrates faster at 96 kDa, suggesting a modified or isoform of SATB1 protein in erythroid cells.²⁷  Quantification of globin transcripts in these K562 cells revealed the expected increase in γ-globin mRNA reaching 9-fold and a decrease in ϵ-globin mRNA by 8-fold, after 96 hours of hemin induction (Figure 2C). Hence, during erythroid differentiation of K562 cells by hemin, induction of γ-globin transcription and hemoglobin production proceeds with a decrease in anti-SATB1 immunoreactive protein and ϵ-globin gene expression. The decrease in ϵ-globin expression following hemin induction is consistent with the late passage of the K562 cells used in this study, and is in contrast with the induction of ϵ-globin expression observed in earlier passages of K562 cells.^32,34 

To identify new SATB1-binding sites in the β-globin locus in addition to those previously reported in this region, we examined SATB1 binding to potential MAR sequences from HS2 and the 5′ flanking region of the ϵ-globin gene (Figure 3A).¹³  Using the potential MAR at the 5′ region of HS2 (HS2-M1, CATTATAATTAACTGTTATTTTTTA, located 158 bp upstream of HindIII in the HS2 core) as a DNA probe for EMSA, we observed a slow migrating band from the K562 nuclear extract (Figure 3B, lanes 1-4). This band was competed by the SATB1-binding sequence Wt (25)₇) (S),¹⁷  and by SATB1 antibody (αS), but not by a mutated SATB1-binding competitor (ΔS), mut (24)₈ (5′-(TCTTTAATTTCTACTGCTTTAGAA)₈-3′).¹⁷  We also observed SATB1 binding to M_ϵ, the 5′ ϵ-globin (TTCCTATTTTGAGATTTGCTCCTTT) located 392 bp 5′ flanking the ϵ-globin proximal promoter (Figure 3B, lanes 5-8). The slow migrating SATB1 band was competed by the SATB1 competitor and the SATB1 antibody, but not by the mutant competitor.

ChIP assay was used to assess SATB1 binding in living cells. SATB1-bound chromatin complexes from K562 cells were isolated using the anti-SATB1 antibody. We examined binding in vivo to ATC sequence-rich regions previously reported to bind to SATB1 in vitro or to nuclear matrix proteins as well as selected ATC sequence-rich regions from γ-globin 5′ (39081-39310), β-globin 5′ (60661-60870), HS4 (853-1525), HS3 (4239-4909), and HS2 (M1 (8127-8327), M2 (8346-8546), M3 (8525-8725), M4 (8579-8797), and M5 (8832-9032) as shown in Figure 3A. Specific primer pairs and corresponding probes (Table 1) were used for quantitative real-time PCR analyses of chromatin fragments immunoprecipitated with the anti-SATB1 antibody (SATB1-ChIP DNA). All primer pairs yielded PCR products with control gDNA (Figure 3C,E-F). With SATB1-ChIP DNA, only HS2-M1 and M_ϵ produced PCR products indicating SATB1 binding in vivo at these sites (Figure 3C, lanes 7, 12). No PCR products were produced using preimmune serum (Figure 3C, lanes 13-18). Increasing amounts of SATB1-ChIP DNA produced corresponding increases in PCR products for HS2-M1 and M_ϵ (19149-19346) in the ϵ-globin 5′ but not HS2-M2 primer pairs (Figure 3C, lanes 19-28). Quantitative real-time PCR demonstrated an increased binding of M1 and M_ϵ to SATB1 in the K562/SATB1 cells compared with K562 cells (Figure 3D). No binding of SATB1 was observed for MARs in the 3′ ^Aγ-globin enhancer region (γ1 [41334-41441] and γ2 [41549-41654]) or in the β-globin IVS2 (63006-63114; Figure 3E), showing that SATB1 binding in vitro to the 3′ ^Aγ-globin enhancer or β-globin IVS2 does not correlate with binding in living cells. In addition, no binding of SATB1 was observed for other ATC sequence-rich regions localized at HS3, HS4, ^Aγ-globin 5′ (-386 bp), and β-globin 5′ (-1526 bp; Figure 3F).

SATB1 activation of ϵ-globin transcription is mediated in part by its direct effect on the ϵ-globin proximal promoter. The ϵ-globin promoter region with flanking M_ϵ-MAR was cloned into the pREP4/Luc episomal reporter vector to produce pREP4/ϵ (Figure 4A). pREP4/Luc does not integrate stably into the genome, but rather propagates as an episome and displays appropriate nucleosomal chromatin structure.³⁵  Analysis of these pREP4/Luc-derived constructs in K562 cells revealed a 3-fold increase in luciferase activity by cotransfection of the SATB1 expression construct (pEGFP/SATB1) with pREP4/ϵ (Figure 4B). Deletion of M_ϵ-MAR (pREP4/Δϵ) reduced luciferase activity by about 2-fold, and cotransfection of pEGFP/SATB1 had no effect on pREP4/Δϵ activity. HS2 containing the 5′ SATB1-binding site M1 was inserted into pREP4/ϵ to create pREP4/HS2-ϵ. HS2-ϵ increased ϵ-globin promoter activity, which was further enhanced by cotransfection with pEGFP/SATB1. With increased SATB1 expression, deletion of M1 MAR in HS-2 (ΔHS2-ϵ) reduced transcription activity to 0.7 the level of HS2-ϵ. The effect of SATB1 did not appear to depend on the presence of erythroid-specific transcription factors such as GATA-1 or NF-E2, as indicated by analysis of these pREP4/Luc-derived constructs in HeLa cells that do not express endogenous SATB1 (Figure 4B). Cotransfection of pEGFP/SATB1 and pREP4/ϵ in HeLa cells led to increased reporter activity comparable to the increase observed in K562 cells. pEGFP/SATB1 cotransfection led to the 2-fold increase in reporter activity of pREP4/HS2-ϵ in HeLa cells with a two-thirds decrease in reporter activity for the M1 MAR-deleted pREP4/ΔHS2-ϵ compared with pREP4/HS2-ϵ. These data provide evidence for a direct effect of SATB1 on ϵ-globin promoter activity, mediated by the SATB1 binding M_ϵ and on the enhancing effect of HS2, mediated by M1 MAR in the 5′ region of HS2.

These constructs were further analyzed in K562/SATB1 cells with elevated SATB1 expression (Figure 1B). Additional constructs were examined including direct mutation of M_ϵ in pREP4/ϵ to give pREP4/mut-ϵ, and mutation of M1 in pREP4/HS2-ϵ to give pREP4/mut-HS2-ϵ. Transfection of pREP4/ϵ into K562/SATB1 cells resulted in robust transcription activity. This activity decreased by mutation of M_ϵ in pREP4/mut-ϵ and was comparable to the reduction obtained with pREP4/Δϵ. Addition of HS2 (pREP4/HS2-ϵ) increased transcription activity by 3.4-fold compared with pREP4/ϵ. The effect of mutating HS2-M1 was comparable to deletion of HS2-M1, and the transcription activity of pREP4/mut-HS2-ϵ and pREP4/ΔHS2-ϵ was reduced to about 0.5 that of pREP4/HS2-ϵ. These results in the K562/SATB1 cells are comparable to those obtained for cotransfection of the reporter gene with the SATB1 expression vector in the control K562 cells.

We determined the acetylation and methylation states of core histones in HS2 and ϵ-globin, γ-globin, and β-globin promoters in the K562/SATB1 cells, where the SATB1 levels are elevated. Antibodies against acetylated isoforms of histone 3 and 4 were used for ChIP analysis. Primers and TaqMan probes, specific for the HS2 enhancer core sequence, ϵ-globin, γ-globin, and β-globin promoters (Figure 5A), were used for quantitative real-time PCR (Table 1). In K562/SATB1 cells, histone H3 acetylation in the ϵ-globin promoter increased by 2-fold compared with control K562 cells with a small decrease or no change observed for γ-globin and β-globin promoters (Figure 5B). Histone H4 acetylation in HS2 and the ϵ-globin promoter increased about 3-fold compared with the control K562 cells and more modest (2-fold or less) or no changes were observed for the γ-globin and β-globin promoters. Methylation of histone H3 at lysine 9 (H3-MeK9) decreased 2-fold in HS2 and the ϵ-globin promoter, and increased 2-fold in the γ-globin promoter (Figure 5B). Histone acetylation associates with actively transcribed genes and the hyperacetylation in HS2 and the ϵ-globin promoter may be indicative of a preferential shift of HS2 and ϵ-globin to a more transcriptionally active state in the K562/SATB1 cells. H3-MeK9 is linked to a less active transcriptional state, and the changes in H3-MeK9 provide further evidence for a shift of HS-2 and the ϵ-globin promoter to a more transcriptionally active state. SATB1 reduces γ-globin expression and may be indicative of the increase in H3-MeK9 associated with the γ-globin promoter offsetting the modest increase in histone H4 acetylation to reduce transcription activation of γ-globin relative to ϵ-globin gene expression.

Hemin induction decreases SATB1 immunoreactive protein with concomitant decreasing ϵ-globin expression and increasing γ-globin expression. ChIP analysis showed that hemin induction decreased histones H3 and H4 acetylation in chromatin associated with HS2 by 2-fold, and with the ϵ-globin promoter by 3- to 4-fold following 48 hours of hemin induction (Figure 5C). These changes, in addition to increases in associated H3-MeK9, are indicative of a shift of HS2 and the ϵ-globin promoter to a less transcriptionally active state, consistent with the reduced expression. Conversely, hemin increases H3 and H4 acetylation and decreases H3-MeK9 in chromatin associated with the γ-globin promoter, as expected for transcription activation and the marked increase in γ-globin expression (Figure 5C).

CBP/p300 is known to interact with a variety of DNA-binding transcription factors and to possess intrinsic histone acetyltransferase activity.^36,37  CBP cooperates with GATA-1 and is required for erythroid differentiation.³⁸  Using coimmunoprecipitation analysis and antibodies specific for SATB1 and CBP, we found that a complex containing CBP and SATB1 in K562 cells could be immunoprecipitated by both CBP- and SATB1-specific antibodies but not by preimmune serum (Figure 6A, lanes 1-6). In contrast, no SATB1 and p300 protein complex in K562 cells was detected (Figure 6A, lane 8). E1A is known to be able to repress CBP transcriptional activity and has been used to test the requirement of CBP.³⁹  In K562 cells, cotransfection of pEGFP/SATB1 and an E1A expression vector with pREP4/ϵ abrogated the increase in reporter gene activity observed with cotransfection of pEGFP/SATB1 alone with pREP4/ϵ (Figure 6B). There was no inhibition of luciferase activity when using the mutant E1AΔ2-36. In the presence of SATB1, CBP exhibited a dose-dependent increase in pREP4/ϵ reporter gene activity (Figure 6C), whereas cotransfection of increasing amounts of the CBP expression construct with the pREP4/ϵ reporter gene in the absence of pEGFP/SATB1 showed little change in transcription activity (Figure 6D). These results provide evidence that SATB1 and the transcription coactivator, CBP, are in the same protein complex that is functionally important for ϵ-globin expression and that CBP enhances SATB1-mediated transcriptional activity. Reporter gene assays in HeLa cells exhibited comparable inhibition by E1A of SATB1 activity on the ϵ-globin promoter with no effect on the mutant E1AΔ2-36 (Figure 6E). The dose-dependent enhancement by CBP of SATB1 activation of the ϵ-globin promoter was also observed in HeLa cells (Figure 6F) providing additional evidence for participation of SATB1 and CBP in a protein complex that contributes to ϵ-globin promoter activity. Further support for cooperation between CBP and SATB1 for ϵ-globin activation is given by the ability of E1A but not mutant E1AΔ2-36 to abrogate the increase in endogenous ϵ-globin expression by SATB1 (Figure 6G). Interestingly, E1A also inhibits the SATB1 decrease of endogenous γ-globin expression.

To determine expression of SATB1 during erythroid differentiation of human adult erythroid progenitor cells, human primary hematopoietic progenitor cells were isolated from peripheral blood and stimulated for erythroid differentiation. Western blotting with anti-SATB1 antibody detected protein early during erythroid differentiation at day 5 of erythropoietin stimulation (Figure 7A). By day 8 with erythropoietin, this protein band was markedly decreased to low levels. We have previously shown ϵ-globin gene activation during early adult erythropoiesis.²⁰  Analysis of β-like globins in these cultures revealed that the peak ϵ-globin expression at day 5 following erythropoietin stimulation coincided with the peak in reactivity to anti-SATB1 antibody and markedly decreased by day 8 of erythropoietin stimulation, whereas γ-globin expression continued to increase (Figure 7B). The induction of β-globin gene expression followed and then surpassed γ-globin gene expression late in erythroid differentiation (Figure 7B). These data show a possible correlation between ϵ-globin gene expression and SATB1 family protein expression.

To investigate the effect of SATB1 on globin gene transcription, a SATB1 expression vector was transfected into human primary adult erythroid progenitor cells, and the cells were harvested at day 8. Maintaining SATB1 expression at a high level beyond day 5 resulted in an increase in ϵ-globin expression (3-fold) with a reduction in γ-globin expression (3-fold; Figure 7C). These changes in ϵ-globin and γ-globin expression are concomitant with increases in histone H3 and H4 acetylation in the ϵ-globin promoter and with decreases in histone H3 and H4 acetylation in the γ-globin promoter (Figure 7D). These data suggest that, as observed in K562 cells, manipulation of the SATB1 level during differentiation of adult early erythroid progenitor cells can alter chromatin associated with ϵ-globin and γ-globin and change the balance of globin gene expression, particularly between ϵ-globin and γ-globin. To down-regulate SATB1 family protein expression, an SATB1 antisense oligonucleotide was synthesized. Transfection into cells down-regulated anti-SATB1 immunoreactive protein when compared with the sense control (Figure 7E). As observed with hemin induction in K562 cells, the antisense oligonucleotide resulted in a decrease in SATB1, a decrease in ϵ-globin, and an increase in γ-globin expression compared with the sense control (Figure 7E).

SATB1 is required for coordinating gene expression during T-cell development⁴⁰  and can target multiple chromatin-remodeling complexes to specific genomic sites to regulate chromatin structure.¹⁸  We found that SATB1 family protein, which is expressed in K562 cells,²⁷  is down-regulated by hemin induction, which is concomitant with increased γ-globin expression and decreased ϵ-globin expression. Conversely, increased SATB1 expression in transfected K562 cells results in activation of ϵ-globin and a decrease in γ-globin expression, with a marked induction of total hemoglobin production in the absence of hemin.

Transcription factors, such as GATA-1 and NF-E2, and EKLF for β-globin expression, can interact specifically within the β-globin cluster at the LCR and downstream globin genes, and can associate with histone acetylases to modify chromatin structure to activate globin gene transcription. In vivo SATB1-binding genomic sites are tightly associated with the base of chromatin loops.⁴¹  Linkage between SATB1 and globin gene expression was initially suggested by the discovery of SATB1-binding sites localized to important regulatory regions within the β-globin cluster. These sites include MARs or ATC sequence-rich regions in HS2 (but not HS3 or HS4) and in the downstream ϵ-, γ-, and β-globin genes.^12-15,42  It has been suggested that SATB1 may participate in a dynamic process to mediate looping of the β-globin locus to achieve transcriptional control, advancing the notion that SATB1 binding to MARs may facilitate the remodeling of local chromatin structure and may bring distal regulatory elements in close proximity to promoters.¹⁶  We screened these potential MARs or SATB1-binding regions^12-15,42  and found that SATB1 bound in vivo to HS2 (M1) and the ϵ-globin promoter M_ϵ, but did not bind to previously reported in vitro binding sites in the 3′ γ-globin enhancer or β-IVS2 or to ATC sequence-rich regions in HS3, HS4, or the 5′ region of the γ- or β-globin genes. HS2 was the first hypersensitive site identified in the LCR with erythroid-specific developmental stage-independent enhancer activity⁴³  and HS2, but not HS3 or HS4, activates transcription of the ϵ-globin gene in K562 cells in studies using chromatinized episomes.^44,45  The data presented here suggest a role for SATB1 in globin gene regulation by selective activation of ϵ-globin transcription. This activation is mediated in part via the formation of a complex containing SATB1 and CBP to increase ϵ-globin expression.

SATB1 orchestrates gene expression by recruiting chromatin modifiers and regulating region-specific histone modification.^18,19  In the K562/SATB1 cells, ϵ-globin expression increases concomitant with the formation of a permissive chromatin state in HS-2 and the ϵ-promoter, characterized by increased histone H3 and H4 acetylation and decreased H3-MeK9 in these regions. A decrease in γ-globin expression in the K562/SATB1 cells is accompanied by an increase in H3 MeK9, but little or no change in histones H3 and H4 acetylation. With hemin induction in K562 cells resulting in a decrease in SATB1 family protein expression, the converse was observed; decreased expression of ϵ-globin was accompanied by a change in histone modification toward that representing silent chromatin, whereas increased expression of γ-globin was associated with a shift toward a more active chromatin state. Interestingly, these data suggest a reciprocal relationship between ϵ- and γ-globin gene expression during the early stage of erythroid differentiation.

Increased ϵ-globin expression in K562/SATB1 cells is mediated in part via ϵ-promoter activation dependent on the presence of an intact M_ϵ motif, demonstrated by the episomal pREP4/ϵ reporter gene assay. Further enhancement of ϵ-promoter activity by HS2 with an intact HS2-M1 motif in the presence of SATB1 identifies this motif as another functionally important SATB1-binding site. The contribution of SATB1 to ϵ-globin activation, via interaction with HS2 and the ϵ-globin promoter, may relate to previously described developmental stage-specific epigenetic changes that occur prior to recruitment of NF-E2 binding to Mafk-recognition elements (MAREs) in HS2, and to high-level transcription activation of downstream globin genes.^46-49  Reporter gene analysis in HeLa cells indicates that SATB1 activation of ϵ-globin promoter activity does not depend on other erythroid transcription factors such as GATA-1 or NF-E2. Together, these underscore the importance of chromatin modification in gene expression and the potential role for SATB1 during erythroid differentiation.

CBP/p300 is a cofactor for many erythroid-specific transcription factors, such as GATA-1, NF-E2, and EKLF, and can bridge activators with TBP or pol II or other components of the basal transcription machinery.^1,37,38,50  We found that SATB1 and CBP are part of a protein complex that is important for ϵ-globin activation but does not appear to require other erythroid transcription factors. Whereas CBP alone has minimal effect on ϵ-globin promoter activity, CBP augments SATB1 activation of the ϵ-globin promoter. Conversely, E1A blocks SATB1 induction of ϵ-globin gene expression. Adenoviral E1A is a viral oncoprotein that can physically interact with cellular proteins, including CBP/p300, YY1, and SWI/SNF complexes, to affect cellular proliferation and differentiation, and can inhibit erythroid differentiation through sequestering CBP/p300.³⁹  Similar augmentation by CBP and suppression by E1A of SATB1 activation of the ϵ-promoter were also observed in HeLa cells, indicating that other erythroid-specific factors are dispensable for this activity. Although in T cells, SATB1 appears to interact with p300 but not CBP,⁵¹  a complex containing SATB1 and p300 was not detected in K562 cells.

A variety of processes regulate globin gene expression, including activator binding to regulatory regions, recruitment of basal transcription factors to the promoter, and chromatin remodeling. Therefore, the study of SATB1 interacting with the β-globin cluster through specific binding offers an additional mechanism to be considered for modification of globin gene expression and for possible chromatin remodeling in erythroid cells. Manipulation of the SATB1 protein level affects both ϵ- and γ-globin gene expression in K562 cells and in primary human adult hematopoietic progenitor cell cultures, providing further evidence for a reciprocal relationship between ϵ- and γ-globin expression during early erythroid differentiation. The data presented here provide evidence that SATB1 family protein has a role in activation of the β-globin cluster, especially the ϵ-globin gene, mediated by interaction with HS2 and the ϵ-globin promoter. The potential for SATB1 manipulation to affect fetal hemoglobin production in adult erythroid progenitor cells awaits further study.

We thank Dr Keji Zhao for pREP4/Luc vector and Dr Emory Bresnick and Dr Alan Schechter for helpful discussion.