NF-E2 Disrupts Chromatin Structure at Human - Molecular and ...

MOLECULAR AND CELLULAR BIOLOGY, Oct. 1996, p. 5634–5644 0270-7306/96/$04.0010 Copyright q 1996, American Society for Microbiology

Vol. 16, No. 10

NF-E2 Disrupts Chromatin Structure at Human b-Globin Locus Control Region Hypersensitive Site 2 In Vitro JENNIFER A. ARMSTRONG

AND

BEVERLY M. EMERSON*

Regulatory Biology Laboratory, The Salk Institute for Biological Studies, La Jolla, California 92037 Received 1 May 1996/Returned for modification 7 June 1996/Accepted 19 July 1996

The human b-globin locus control region (LCR) is responsible for forming an active chromatin structure extending over the 100-kb locus, allowing expression of the b-globin gene family. The LCR consists of four erythroid-cell-specific DNase I hypersensitive sites (HS1 to -4). DNase I hypersensitive sites are thought to represent nucleosome-free regions of DNA which are bound by trans-acting factors. Of the four hypersensitive sites only HS2 acts as a transcriptional enhancer. In this study, we examine the binding of an erythroid protein to its site within HS2 in chromatin in vitro. NF-E2 is a transcriptional activator consisting of two subunits, the hematopoietic cell-specific p45 and the ubiquitous DNA-binding subunit, p18. NF-E2 binds two tandem AP1-like sites in HS2 which form the core of its enhancer activity. In this study, we show that when bound to in vitro-reconstituted chromatin, NF-E2 forms a DNase I hypersensitive site at HS2 similar to the site observed in vivo. Moreover, NF-E2 binding in vitro results in a disruption of nucleosome structure which can be detected 200 bp away. Although NF-E2 can disrupt nucleosomes when added to preformed chromatin, the disruption is more pronounced when NF-E2 is added to DNA prior to chromatin assembly. Interestingly, the hematopoietic cell-specific subunit, p45, is necessary for binding to chromatin but not to naked DNA. Interaction of NF-E2 with its site in chromatin-reconstituted HS2 allows a second erythroid factor, GATA-1, to bind its nearby sites. Lastly, nucleosome disruption by NF-E2 is an ATP-dependent process, suggesting the involvement of energydependent nucleosome remodeling factors. throid lineage is made (23). This suggests that alteration of chromatin structure poises the locus for the subsequent step of commitment to b-globin gene expression. However, the events involved in hypersensitive site formation in the b-globin LCR are poorly understood. It is unknown whether erythroid proteins which bind to the LCR at DNase I hypersensitive sites are capable of binding to their sites when assembled into chromatin and subsequently disrupting nucleosomal structure. The ability of a hematopoietic cell-specific trans-acting factor to disturb chromatin structure in the LCR would be consistent with the formation and/or maintenance of the DNase I hypersensitive sites and the open chromatin configuration which is necessary for transcription of the b-globin genes. Recently, many of the erythroid-cell-restricted transcription factors with conserved binding sites in the LCR have been elucidated. One surprising discovery was the small set of hematopoietic cell-specific factors which are involved. These include NF-E2 (2, 3, 37), GATA-1 (33, 48, 49), and EKLF (34). Such a limited number of trans-acting factors, in addition to extensive in vivo characterization, makes the next step to in vitro studies quite logical. In vitro approaches now enable investigation of the mechanisms of chromatin remodeling by these proteins within the LCR. Using an in vitro chromatin assembly system, we examined the ability of a key erythroid transcription factor, NF-E2, to alter chromatin structure in HS2 of the human b-globin LCR. NF-E2 is a basic leucine zipper heterodimer consisting of the hematopoietic cell-specific subunit, p45, and the small 18-kDa subunit of the maf proto-oncogene family (2, 3, 21), p18. Expression of NF-E2 is limited to erythroid, hematopoietic progenitor, megakaryocytic, and mast cells (2). Binding sites for NF-E2 are found throughout both the b- and a-globin LCRs, and NF-E2 is necessary for expression of the b-globin gene in mouse erythroleukemia (MEL) cells (25, 31) but not in transgenic mice (43). NF-E2 binds to two tandem AP1-like sites which form the core of the HS2 enhancer (32, 46). The 59 of the two NF-E2

Active eukaryotic genes are characterized by a general sensitivity to nucleases and early replication in S phase (18, 56). Nuclease sensitivity reflects an altered chromatin configuration with increased accessibility to the trans-acting factors which are necessary for DNA replication and gene expression. DNase I sensitivity throughout the 100-kb b-globin locus and hypersensitivity in the upstream locus control region (LCR) is well characterized in globin-expressing erythroid cells (20, 52). The human b-globin LCR consists of five DNase I hypersensitive sites, four of which are erythroid cell specific (HS1 to -4), and regulates expression of the five globin genes: epsilon, G gamma, A gamma, delta, and beta (52). The LCR also affects the chromatin structure and timing of DNA replication over the entire locus (16). All four erythroid-cell-specific sites in the LCR possess some level of LCR activity, which is defined as conferring high levels of integration site-independent and copy number-dependent erythroid-cell-specific expression to a linked gene (15, 17, 20, 42, 45). Thus, an LCR can form and maintain an active chromatin structure regardless of the state of surrounding chromatin, thereby overcoming position-effect variegation. Although the LCR has a clear role in determining chromatin structure, the mechanism by which it accomplishes this is unclear. The second hypersensitive site, HS2, located 10.8 kb upstream of the epsilon-globin gene, also acts as a powerful transcriptional enhancer, the core of which is formed by two tandem AP1-like sites (32, 36, 46, 47, 53). Hypersensitive sites are believed to represent nucleosomefree or -disrupted regions of chromatin (reviewed in reference 19). Previous studies have shown that DNase I hypersensitivity in the LCR is observed in multipotential hematopoietic progenitor cells before the decision to differentiate along the ery* Corresponding author. Mailing address: The Salk Institute, 10010 N. Torrey Pines Rd., La Jolla, CA 92037. Phone: (619) 453-6560. Fax: (619) 535-8194. Electronic mail address: [email protected] .edu. 5634

VOL. 16, 1996

NUCLEOSOME DISRUPTION AT HS2 BY NF-E2

sites is necessary for enhancer function, while the 39 site contributes to overall activity (46). In this study, we show that DNase I HS2 of the LCR can be formed in vitro by the interaction of recombinant NF-E2 which disrupts local chromatin structure in an energy-dependent manner. Nucleosome disruption requires the p45 subunit and allows recruitment of GATA-1 to HS2. Since active genes may be depleted of histone H1 (7), NF-E2 binding to chromatin is examined in the presence and absence of H1. Chromatin binding and nucleosome disruption by a hematopoietic cell-specific transcription factor provide a potential mechanism for the establishment and maintenance of a chromatin configuration poised for expression of the b-globin genes. MATERIALS AND METHODS Plasmid constructions. HS2bCAT was constructed as previously described (22). Expression vectors for p45 and p18 were generated in pET-15b (Novagen). p18/pET-15b was constructed by inserting a SmaI-BamHI fragment from p18pBluescript into the NdeI (blunt)-BamHI sites of pET-15b. p45/pET-15b was constructed by first cloning the p45 cDNA into pBluescript; this was followed by a three-way ligation inserting a 120-bp PCR product from the start codon of p45 (with an NdeI site included in the primer) into an internal PstI site and a PstI-BamHI fragment into the NdeI-BamHI sites of pET-15b. The integrity of the PCR-generated portion of p45 was verified by sequencing. Expression and purification of recombinant proteins. All proteins were expressed in bacterial BL21(DE3) cells. Both p18 and p45 were purified under denaturing conditions on Ni21-nitrilotriacetic acid (NTA) resin (Qiagen) as described in the Qiagen QIAexpressionist booklet. p18 was renatured by dialysis into storage buffer {20 mM HEPES [N-2-hydroxyethylpiperazine-N9-2-ethanesulfonic acid, pH 7.9], 20% glycerol, 0.2 mM EDTA, 0.2 mM EGTA [ethylene glycol-bis(b-aminoethyl ether)-N,N,N9,N9-tetraacetic acid], 5 mM dithiothreitol [DTT]} with 0.5 M KCl for 6 h at 48C. p45 was renatured by stepwise dialysis from 8 M urea to 0 M urea over a period of 9 h at 48C into storage buffer with 0.1 M KCl. Both p45 and p18 were quite pure; when analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and silver stained, single bands were observed. NF-E2 was reconstituted by combining equal molar ratios of the two proteins at high salt and dialyzing to low salt (from 1 M to 70 mM KCl in 20 mM HEPES [pH 7.9]–10% glycerol–5 mM MgCl2–2 mM DTT). A form of NF-E2 with higher activity was generated by renaturing equal molar ratios of the two subunits together by stepwise dialysis as described above for p45. Both types of reconstituted NF-E2 protein were used in this study, with identical results. The amount of NF-E2 protein used in these experiments was always comparable to that needed to visualize a footprint on plasmid DNA. mGATA-1 was expressed in bacteria from the pET-mGATA-1 vector, bound to double-stranded DNA-cellulose (Sigma) in buffer G (20 mM HEPES, pH 7.9, 5 mM MgCl2, 10% glycerol, 0.1% Brij 35, 0.1 mM EDTA) with 25 mM ammonium sulfate, washed with buffer G with 90 mM ammonium sulfate and eluted in buffer G with 250 mM ammonium sulfate. GATA-1 was dialyzed into dialysis buffer (20 mM HEPES, pH 7.9 50 mM KCl, 0.2 mM EDTA, 10% glycerol, 1 mM DTT) with 5 mM MgCl2. MEL extract preparation. MEL cells (30) were grown in a minimal essential medium with 10% calf serum and 13 penicillin–streptomycin–L-glutamine and induced with N,N9-hexamethylene bisacetamide (HMBA) at 5 mM 36 to 48 h prior to extract preparation. All centrifugation steps were carried out at 48C. Typically, 10 to 20 liters of cells were harvested and washed twice with cold phosphate-buffered saline, and cells were resuspended in hypotonic buffer (20 mM HEPES, pH 7.9, 10 mM NaCl, 1.5 mM MgCl2, 2 mM DTT), swelled 10 min on ice, and spun at 4,000 rpm for 10 min in a Sorvall SS-34 rotor. Cells were resuspended in modified hypotonic buffer (hypotonic buffer with 15% glycerol), proteinase inhibitors were added (10-mg/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, 10-mg/ml pepstatin, 10-mg/ml E-64, 1-mg/ml aprotinin, 1 mM benzamidine), and the cells were homogenized on ice by using a Dounce homogenizer for 20 strokes with pestle B. NaCl was added dropwise to approximately 0.3 M, and extract was stirred for 30 min at 48C. The extract was spun at 45,000 rpm for 1 h in a 50.2 Ti rotor, the lipid pellicle was aspirated off, and the supernatant was dialyzed against 2 liters of dialysis buffer (20 mM HEPES, pH 7.9, 50 mM KCl, 0.2 mM EDTA, 10% glycerol, 1 mM DTT) for 6 h at 48C with one change of dialysis buffer and spun at 15,000 rpm for 20 minutes in a Sorvall SS-34 rotor. MEL transcription extract was aliquoted, frozen in liquid nitrogen, and stored at 21008C. Chromatin assembly and transcription reactions. Drosophila S-190 chromatin extracts were prepared from 6-h embryos as previously described (8, 24). As these extracts contain low levels of histones, they must be supplemented with core histones (see reference 28 for a description of the preparation of core histones). Histone H1 was prepared as described previously (13). Plasmid DNA (0.5 to 1.0 mg) was assembled in 50- to 100-ml reaction volumes. For a typical assembly reaction, 1 mg of plasmid DNA was preincubated with or without MEL extract or purified recombinant NF-E2 protein for 20 min in a room temperature

5635

water bath in a volume of 50 ml in 20 mM HEPES (pH 7.9)–50 mM NaCl–0.1 mM EDTA (pH 8.0)–10% glycerol–5 mM DTT. To this sample, 40 ml of Drosophila assembly extract (preincubated with 0.8 mg of core histones for 30 min on ice) and 10 ml of 103 ATP mix (300 mM phosphocreatine, 30 mM ATP, 41 mM MgCl2, and 10-mg/ml creatine kinase) was added, and assembly was allowed to proceed for 41⁄2 h at 278C. Mock assembly was achieved by incubation of the DNA with S-190 buffer (10 mM HEPES, pH 7.5, 10 mM KCl, 1.5 mM MgCl2, 0.5 mM EGTA, 10% glycerol) instead of S-190 extract. For transcription assays, chromatin was incubated with MEL extract (total MEL extract was 80 ml per reaction) on ice for 10 min, 103 transcription mix was added (200 mM HEPES, pH 7.9, 500 mM KCl, 55 mM MgCl2, 2-mg/ml bovine serum albumin [BSA], 6 mM [each] CTP, UTP, and GTP, 7 mM DTT, 5% glycerol), and transcription was allowed to proceed at 308C for 30 min. Transcription reactions were divided into three equal aliquots and stopped with 130 ml of transcription stop buffer (1% SDS, 20 mM EDTA). RNA purification by proteinase K digestion (0.2 mg/ml at 378C for 15 min) was followed by phenol-chloroform extraction, chloroform extraction, and sodium acetate (pH 5.2)-ethanol precipitation with 10 mg of glycogen as carrier. The RNA was analyzed by primer extension analysis. MNase and DNase I digestion. Immediately following chromatin assembly, 50 ml of chromatin was digested with 15 U of micrococcal nuclease (MNase) (Boehringer Mannheim). MNase ladders were generated by removing 15-ml aliquots at 1-, 3-, and 9-min time points. Chromatin which was to be analyzed by plasmid footprinting was digested with MNase for 10 min. Reactions were stopped by the addition of 53 stop buffer (2.5% Sarkosyl, 10 mM EDTA); this was followed by RNase A treatment (0.2 mg/ml at 378C for 10 min). SDS was added to 0.2%, proteinase K was added to 0.2 mg/ml, and the reaction was incubated at 558C for 15 min. The DNA was purified by phenol-chloroform extraction and ammonium acetate-ethanol precipitation with 10 mg of glycogen as carrier. For ladders, MNase-digested DNA was electrophoresed on a 1.5% Tris-glycine agarose gel (44 mM Tris base, 384 mM glycine) at 10 V/cm for 3 h at 48C, Southern blotted onto a Gene Screen Plus membrane which had been prehybridized for 1 h at 378C in hybridization buffer (63 SSC [13 SSC is 0.15 M NaCl plus 0.015 sodium citrate], 13 Denhardt’s solution, 0.1-mg/ml salmon sperm DNA, 0.5% SDS), and hybridized with 5 pmol of oligonucleotide in the same buffer for 5 h at 378C. Washes were carried out in wash buffer (63 SSC, 0.5% SDS) for 5 min at room temperature, for 10 min at 378C (twice), and for 20 min at 428C. The oligonucleotide probes that were used are as follows: CAT, GCCATTGGGATATATCAACGGTGG; A, TCCCTTCCAGCATCCTCATC TCTG; B, TCAGTGCCCCACCCCCGCCTTCTG; C, GATCATGCTGAGTC ATGATGAGTCATG; D, GCATCCTGCTGGGGACCCAGATAG; and E, TGTCACATTCTGTCTCAGGCATCC. DNase I digestions were carried out for 1 min at 308C with the amounts of DNase I indicated in the figure legends. DNase I was diluted when necessary into dilution buffer (25 mM HEPES, pH 7.9, 80 mM NaCl, 5 mM MgCl2, 0.1-mg/ml BSA, 5 mM CaCl2). For hypersensitive site analysis, the DNA was purified as described above, digested with DraI, run on a 1% TBE agarose gel, Southern blotted, and probed with a DraI-HindIII fragment from HS2bCAT. Plasmid footprinting. Immediately following assembly, 12 ml of chromatin was digested with either MNase or DNase I (amounts as indicated in figure legends), and the DNA was purified as described above, denatured by incubation in 0.2 N NaOH–2 mM EDTA for 10 min at room temperature, and precipitated with sodium acetate (pH 4.5)-ethanol. End-labeled primer (0.25 pmol) was added to the DNA in Vent buffer (40 mM NaCl, 10 mM Tris [pH 8.9], 5 mM MgSO4) containing 0.5 mM deoxynucleoside triphosphate and annealed at 608C for 30 min. Vent DNA polymerase (0.4 U) was added, and the reaction was incubated at 768C for 10 min to allow extension. The products were precipitated with sodium acetate (pH 5.2)-ethanol and analyzed on a 6% sequencing gel. Sequencing reactions were carried out by dideoxy sequencing with the same primer. Reactions analyzed by plasmid footprinting were also analyzed by MNase ladders to verify the presence of chromatin.

RESULTS HS2 is formed in vitro with erythroid-cell-specific factors. The in vitro Drosophila chromatin assembly system used in this study allows transcription of the reconstituted HS2bCAT template and generates the chromatin structure seen in erythroid cells in vivo. The 6.15-kb plasmid construct used in this analysis is shown in Fig. 1A. This construct has been extensively characterized in transfection studies and reflects in vivo patterns of gene regulation (22). When assembled into chromatin in vitro, the HS2bCAT template was competent to allow transcription from the b-globin promoter (Fig. 1B). Similar to most RNA Pol II genes, the human b-globin promoter was transcriptionally repressed by chromatin assembly (lanes 5 to 7). This repression was overcome by including an erythroid protein extract during chromatin assembly (lanes 8 to 13); this was followed by transcription initiated by the addition of nucleo-

5636

ARMSTRONG AND EMERSON

FIG. 1. b-globin chromatin template assembled in the presence of erythroid factors is transcriptionally active and exhibits DNase I hypersensitivity at HS2. (A) Schematic diagram of human b-globin locus and HS2bCAT construct. The 1.45-kb HS2 element extends from KpnI to BglII sites, and the b-globin promoter extends from 2378 to 118. Drawing is not to scale. (B) Transcription of in vitro-assembled chromatin template. HS2bCAT plasmid DNA was assembled into chromatin with a Drosophila S-190 extract in the presence of MEL extract in the amounts as indicated. After assembly the chromatin was transcribed with additional MEL extract. Following transcription, each reaction was split equally into three tubes for RNA purification and primer extension analysis. Naked DNA was mock assembled by incubation with S-190 buffer instead of S-190 extract. The primer extension product is 99 nucleotides and is indicated by an arrow. M indicates HindIII-digested pBR322 as marker. (C) Hypersensitive site formation at HS2 in in vitro-reconstituted chromatin. A DNase I hypersensitive site is observed at HS2 when assembled into chromatin in the presence of either an MEL extract or NF-E2. HS2bCAT was prebound with MEL extract (lanes 9 to 12), or recombinant NF-E2 (lanes 13 to 16). Lanes 1 to 4 contain mockassembled naked DNA. Following assembly, reactions were split into four aliquots (100 ng of DNA each) and digested with DNase I as follows: 0, 0.0005, 0.0015, and 0.0045 U (lanes 1 to 4); 0, 0.1, 0.3, and 0.9 U (lanes 5 to 8 and 13 to 16); and 0, 0.3, 0.9, and 2.7 U (lanes 9 to 12). M indicates marker lane in which HS2bCAT was digested with DraI; this was followed by digestion with MscI, BspHI, XbaI, or ApaI as a size marker and a hybridization control. Arrows point to hypersensitive sites or regions. P, promoter; HS2, hypersensitive site 2. Schematic diagram of construct is shown alongside for reference.

MOL. CELL. BIOL.

side triphosphates and either additional MEL extract (Fig. 1B) or a nonerythroid HeLa transcription extract (data not shown). Presumably, the erythroid proteins present during chromatin assembly were able to bind their sites in the b-globin promoter, thus preventing nucleosome formation over the transcription start site and allowing transcription to occur. This experiment illustrates that our in vitro-reconstituted chromatin does not nonspecifically interfere with normal functions such as transcription. Transcription under these conditions is not dependent on HS2 (data not shown), as the promoter contains multiple sites for transcriptional activators and is very active in vitro. The transcription start site observed in vitro is the same as that utilized in vivo (data not shown). When the DNase I sensitivity of the template was examined, the promoter region showed hypersensitivity not seen in naked DNA or in chromatin reconstituted in the absence of erythroid factors (Fig. 1C, compare lanes 3 and 8 to lane 12). DNase I hypersensitivity in the b-globin promoter is observed in vivo only in erythroid cells expressing the gene (52). The HS2 region also revealed DNase I hypersensitivity in a pattern very similar to that seen in vivo (47), centering around the tandem NF-E2 sites and including an upstream CACC site and downstream GATA-1 sites. In order to determine whether a single erythroid factor was able to generate a DNase I hypersensitive site, purified recombinant NF-E2 was preincubated with the plasmid and chromatin was assembled. As shown in Fig. 1C, a DNase I hypersensitive site is generated by NF-E2 alone (lane 16) and corresponds to a portion of the hypersensitive region produced by preincubation with an MEL extract. The hypersensitive site comigrates with the marker fragment generated by digestion with DraI and BspHI. The BspHI site is located between the two tandem NF-E2 sites in HS2. Interestingly, the DNase I pattern generated with NF-E2 looks remarkably similar to the hypersensitive site at HS2 seen in induced MEL cells which carry multiple copies of the human b-globin minilocus (47). Thus, our in vitro chromatin assembly system correctly recapitulates the chromatin structure observed in erythroid cells in vivo, allowing us to address specific questions regarding the mechanism of interaction between a key erythroid factor and chromatin. Binding of NF-E2 results in nucleosome disruption. A DNase I hypersensitive site in chromatin is believed to represent a nucleosome-free or -disrupted region of DNA. Since NF-E2 alone was competent to form a hypersensitive site, we wished to examine its effect on nucleosomes. Although the binding of many transcription factors to chromatin has been studied (reviewed in references 6, 57, and 58), the analyses have never been extended to erythroid-cell-specific factors. We incubated HS2bCAT DNA with NF-E2, reconstituted chromatin, and digested the chromatin with MNase; this was followed by agarose gel electrophoresis and Southern blotting. The blots were then probed with oligonucleotides corresponding to the NF-E2 binding site. MNase digests chromatin in the linker region between nucleosomes. If nucleosomes are evenly spaced throughout the DNA, a ladder will be generated. Any perturbation of this ladder indicates that the nucleosomes are no longer present in a regular array at that site, an effect we shall refer to as nucleosome disruption. As shown in Fig. 2A, increasing the amount of NF-E2 results in a disruption of the MNase ladder at the NF-E2 site. This disruption is characterized by the smearing of the ladder and by the generation of pieces of DNA smaller than a mononucleosome. These small DNA fragments may correspond to naked DNA protected from MNase by NF-E2 itself or to DNA bound to a disturbed histone octamer. The level of nucleosome disruption reaches a saturation of 40% at 75 nM NF-E2. This nucleosome disrup-

VOL. 16, 1996


5637

FIG. 2. Chromatin assembly in the presence of NF-E2 results in nucleosome disruption at the NF-E2 site in HS2. MNase digestion of chromatin-reconstituted HS2bCAT plasmid. DNA (1 mg) was preincubated with the indicated amounts of recombinant NF-E2. (A) The membrane was hybridized with oligonucleotide C, which corresponds to the NF-E2 site. The percent nucleosome disruption was calculated by phosphoimaging analysis of the membrane blot as follows: the level of small DNA pieces was determined for the highest degree of MNase digestion for each reaction and the amount of submononucleosomal DNA was divided by total mononucleosomal and submononucleosomal DNA and multiplied by 100. Arrows indicate species as mentioned in the figure, and the bracket indicates a smear seen between the mono- and the dinucleosome upon binding of NF-E2. (B) The membrane was stripped and rehybridized with an oligonucleotide probe within the CAT reporter gene, 1 kb downstream of the NF-E2 sites.

tion is specific to the NF-E2 site, since an oligonucleotide probe 1 kb away in the chloramphenicol acetyltransferase (CAT) gene shows no disturbance of the nucleosome ladder (Fig. 2B). These experiments clearly show that NF-E2 can bind chromatin and disrupt local nucleosomal organization. However, the erythroid factor was preincubated with naked plasmid DNA and was present throughout chromatin assembly. NF-E2 may simply be binding to naked DNA and passively excluding nucleosomes during chromatin formation. To examine whether NF-E2 could disrupt preformed chromatin, the protein was added to the reaction following chromatin assembly. As shown in the middle of Fig. 3A, NF-E2 can indeed disrupt nucleosomal organization at its binding site when added postchromatin assembly, as seen by the smearing of the MNase ladder and by the generation of submononucleosomal-sized fragments. However, nucleosome disruption is more pronounced if NF-E2 is added before chromatin assembly. Another interesting question involves the extent of the chromatin disruption. By stripping and reprobing the same blot multiple times with oligonucleotides which spanned the core HS2 enhancer, we found that smearing of the MNase ladder between the mono- and dinucleosomes was detectable up to 200 bp away from the NF-E2 site in both directions (Fig. 3A). Oligonucleotides A and B are located 208 and 75 bp upstream of the NF-E2 site, respectively. Oligonucleotides D and E are located 90 and 193 bp, respectively, downstream of the NF-E2 site. Note that small pieces of DNA were not generated when distal oligonucleotide probes were used. This is consistent with the idea that the submononucleosomal-sized fragments result from protection from MNase digestion by NF-E2. At a distance of 200 bp away from the NF-E2 site it is the neighboring nucleosome that is affected, hence the smearing between the mono- and dinucleosomes. Again, the CAT oligonucleotide probe in Fig. 3B indicates that this disruption is specific and does not extend to 1 kb away from the NF-E2 sites. We conclude that NF-E2 can disrupt nucleosomes when added both before and after chromatin assembly, and this disruption is detectable at least 200 bp away.

Binding of NF-E2 is not altered by histone H1 and shows strand preference when NF-E2 is added after chromatin assembly. Our chromatin up to this point had been assembled in the absence of histone H1, which binds linker DNA on both sides of the nucleosome in decondensed chromatin (reviewed in reference 14). Plasmid footprinting of MNase-digested chromatin was performed to assess whether the binding of NF-E2 to chromatin could position nucleosomes immediately upstream or downstream and whether incorporation of H1 into this chromatin affected NF-E2 binding. Note that this technique cannot address whether NF-E2 is bound in a ternary complex with a nucleosome since MNase only digests in the linker region between nucleosomes. Figure 4 shows that upon binding of NF-E2, no obvious change in the MNase digestion pattern of the chromatin flanking the NF-E2 sites was seen (compare lanes 9 and 10 to 11 and 12; also compare lanes 21 and 22 to 23 and 24). NF-E2 was capable of interacting with its site in chromatin even when reconstituted with histone H1. Thus the binding of NF-E2, while disrupting the local chromatin, does not result in specifically repositioned nucleosomes. Histone H1 does not affect this process. MNase ladders generated from the same chromatin reactions showed an increase in nucleosome repeat length (from 174 to 193 bp), indicating that histone H1 was properly incorporated into the chromatin (data not shown). Comparison of NF-E2 binding when added pre- and postchromatin assembly was also examined by plasmid footprinting of MNase-digested chromatin. This chromatin was reconstituted in the absence of histone H1. Low amounts of NF-E2 were used to visualize differences in binding. As detected by extension with the 39 oligonucleotide, NF-E2 binding to the top strand was complete whether added pre- or postchromatin assembly (Fig. 5, compare lanes 12 through 15 to lanes 17 through 20). However, protection is incomplete on the bottom strand, as detected with the 59 oligonucleotide, when NF-E2 is added after chromatin assembly (compare lanes 2 through 5 to lanes 7 through 10). Thus, NF-E2 preferentially binds the top strand when added to preformed chromatin.

5638


MOL. CELL. BIOL.

FIG. 3. Nucleosome disruption can be detected up to 200 bp away and is more pronounced if NF-E2 is added prior to chromatin assembly. MNase nuclease digestion of chromatin-reconstituted HS2bCAT plasmid. DNA (1 mg) was reconstituted into chromatin as described. A total of 43.2 pmol of NF-E2 was included either prior to assembly and incubated 30 min at room temperature (pre-) or added after assembly was complete and allowed to bind for 30 min at 278C (post-). (A) The membrane was repeatedly stripped and rehybridized with oligonucleotides A to E throughout HS2 as indicated. The diagram is to scale and the size bar represents 40 bp. NF-E2 is represented by the ovals, the arrow indicates submononucleosomal-sized DNA fragments, and the bracket indicates smear between mono- and dinucleosome. (B) The membrane was stripped and rehybridized with the CAT oligonucleotide. The order of oligonucleotide hybridization and membrane stripping was unimportant to the outcome of the experiment.

In summary, although NF-E2 can clearly bind to its site when reconstituted into chromatin, it does not appear to position nearby nucleosomes. Binding of NF-E2 seems to be unaffected by the presence of histone H1. When bound to preassembled chromatin, NF-E2 preferentially interacts with the top strand of DNA, indicating that it may be restricted from binding the bottom strand by the presence of nucleosomes. Both subunits of NF-E2 are necessary for chromatin binding, and NF-E2 is able to recruit GATA-1 to its site in chromatin. NF-E2 is composed of two subunits, the hematopoietic cell-specific component, p45, and the DNA-binding Maf protein, p18 (3, 21). As shown in the plasmid footprint in Fig. 6A, both NF-E2 and p18 were competent to bind the HS2 NF-E2 site in naked DNA (lanes 6 to 9). The NF-E2 binding site is similar to the 14-bp palindromic T-MARE site bound by proteins of the maf gene family (21). The DNase I protection patterns generated by p18 and by NF-E2 look remarkably similar. However, when the HS2bCAT DNA was reconstituted into chromatin following prebinding with the individual subunits or NF-E2, only NF-E2 was able to remain bound in a chromatin context (Fig. 6B, compare lanes 6 and 7 to 10 and 11). This suggests that p45 is involved in stabilizing the binding of the p18 subunit to chromatin. Note that the chromatin was digested with DNase I in this plasmid footprint analysis. Therefore all factor binding, whether in the linker or on the nucleosome, is detectable. We next wanted to address whether interaction of NF-E2 with chromatin affected the binding of nearby erythroid factors. Immediately downstream of the tandem NF-E2 sites are two head-to-head GATA-1 motifs. As shown in Fig. 6B, GATA-1 was only able to bind chromatin in the presence of NF-E2 (lanes 6 and 7), suggesting that nucleosome disruption by NF-E2 facilitates the binding of neighboring factors. In the

presence of the individual p45 or p18 subunits, no binding by GATA-1 was observed. Thus, not all factors are competent to bind to their sites in HS2 when the DNA is reconstituted into chromatin. The homodimer p18 cannot bind to a reconstituted plasmid, and GATA-1 only binds in the presence of NF-E2. Nucleosome disruption by NF-E2 is dependent on ATP. Since NF-E2 is able to bind chromatin when added after assembly is complete, this suggests more than just a passive process. The assembly of chromatin in vitro requires ATP and an ATP-regenerating system, and nucleosomal reorganization by many transcription factors is ATP dependent (12, 26, 50, 55). The ATP-hydrolyzing enzyme apyrase was added after chromatin assembly and prior to incubation with NF-E2 to determine whether the specific nucleosome disruption by NF-E2 was ATP dependent. As shown in Fig. 7A, the generation of submononucleosomal-sized DNA fragments, which indicates free DNA or a disrupted nucleosome, was not seen in the presence of apyrase. Note that chromatin is more resistant to digestion by MNase in the presence of apyrase. We speculate that in the absence of functional energy-dependent factors which allow nucleosome reorganization, the chromatin may become less dynamic and more difficult to digest. Furthermore, digestion of naked DNA by MNase is slightly inhibited by the presence of apyrase (data not shown). NF-E2 was able to bind DNA in the presence of apyrase in gel mobility shift analyses (data not shown). Thus it appears that nucleosome disruption by NF-E2 is dependent on the function of an ATP-hydrolyzing enzyme. We then examined whether NF-E2 is able to bind chromatin in the presence of apyrase. As shown in Fig. 7B, a plasmid footprint of MNase-digested chromatin clearly indicates that NF-E2 is able to bind to its site in HS2 in chromatin in the absence of ATP (compare lanes 1 with 3 and 2 with 4). These same reactions were also analyzed by MNase ladders and then

VOL. 16, 1996


5639

FIG. 4. Binding of NF-E2 is unaffected by incorporation of histone H1 into the chromatin. Plasmid footprinting of chromatin-reconstituted HS2bCAT plasmid. DNA (0.5 mg) was preincubated with 24.6 pmol of NF-E2 (lanes 11, 12, 23, and 24) and chromatin was assembled in the presence (lanes 10, 12, 22, and 24) or the absence (lanes 9, 11, 21, and 23) of 0.12 mg of histone H1. Histone H1 was included with the core histones and incubated with the S-190 extract for 30 min on ice prior to addition to the DNA. Following chromatin assembly and digestion with MNase, each reaction was split into two equal aliquots, and plasmid footprinting was carried out with either the 59 oligonucleotide, oligo B (lanes 5 to 12), or the 39 oligonucleotide, oligo D (lanes 17 to 24). Naked DNA was digested with 0.0075, 0.075, 0.75, and 7.5 U of MNase (lanes 5 to 8 and 17 to 20), whereas chromatin was digested with 7.5 U of MNase. Protection by NF-E2 is indicated by brackets. The autoradiogram was scanned with a Hewlett Packard ScanJet IIcx/T, and the image was processed in Canvas 3.5.1.

by Southern blotting to ensure that the apyrase was functioning, and as before, the generation of submononucleosomesized DNA fragments was not seen in the presence of apyrase (data not shown). Thus, it appears that nucleosome disruption by NF-E2 is separable into two steps: factor binding and subsequently the energy-dependent nucleosome disturbance. DISCUSSION This study is the first to demonstrate chromatin disruption in the human b-globin LCR by a specific erythroid factor in vitro. We have shown that HS2 can be reconstructed in in vitro-

assembled chromatin by preincubation of the DNA with either an erythroid extract or with recombinant NF-E2. NF-E2 can specifically disrupt nucleosomes, even when added to preformed chromatin. NF-E2 binding to chromatin does not position nearby nucleosomes and is unaffected by the presence of histone H1. When added after chromatin assembly, NF-E2 preferentially binds the top strand of DNA. As naked DNA, the HS2 NF-E2 site can be bound by the non-erythroid-cellspecific subunit, p18. However, the hematopoietic cell-specific subunit, p45, is necessary for chromatin binding. Binding of NF-E2 to chromatin increases binding of GATA-1 to its sites 60 bp downstream. Lastly, nucleosome disruption by NF-E2 is

5640


MOL. CELL. BIOL.

FIG. 5. When added after chromatin assembly, NF-E2 binds DNA with strand preference. Plasmid footprinting of chromatin-reconstituted HS2bCAT plasmid is shown. Either DNA (0.5 mg) was preincubated with NF-E2, or NF-E2 was added after chromatin assembly was complete. Either 2.6, 5.1, 10.2, or 20.4 pmol of NF-E2 was included pre- (lanes 2 to 5 and 12 to 15) or postassembly (lanes 7 to 10 and 17 to 20). Following digestion with MNase and DNA purification, the samples were split in two equal aliquots and analyzed by plasmid footprinting with either the 59 oligonucleotide, oligo B (lanes 1 to 10), or the 39 oligonucleotide, oligo D (lanes 11 to 20). Protection by NF-E2 is indicated by brackets. The autoradiogram was scanned with a Hewlett Packard ScanJet IIcx/T, and the image was processed in Canvas 3.5.1.

dependent upon the presence of ATP while binding of NF-E2 is not. The ability of NF-E2 to disrupt chromatin structure is consistent with LCR function as an element which forms, maintains, and insulates an active chromatin configuration (11, 29, 38, 39). Interestingly, there is a strong correlation between DNase I HS2 and the presence of NF-E2 protein in vivo. Both HS2 (23) and NF-E2 (27, 40) are present in hematopoietic progenitor

cells prior to differentiation along the erythroid pathway. If the cells then differentiate along a nonerythroid pathway, HS2 is lost and the p45 subunit of NF-E2 is no longer detected (23, 27). Multiple proteins bind in HS2, including NF-E2, GATA-1, and a CACC-binding protein. These factors and others likely contribute to the generation of the hypersensitive site in vivo. Indeed, in MEL cells lacking the p45 subunit of NF-E2, a hypersensitive site is seen at mouse HS2 (25). This could result

VOL. 16, 1996


5641

FIG. 6. The erythroid-cell-specific subunit of NF-E2, p45, is required for binding to chromatin-reconstituted HS2, and only the complete NF-E2 is able to recruit GATA-1. (A) Both p18 and NF-E2 (p18 and p45) can bind to the NF-E2 site in HS2 as naked DNA. HS2bCAT DNA was bound by 4.9 or 9.8 pmol of NF-E2 or p18 for 30 min at 308C. The DNA was digested with 0.002 U of DNase I per reaction. Following DNA purification, plasmid footprinting analysis was carried out with oligonucleotide B. The bracket indicates factor protection. (B) Upon assembly of the DNA into chromatin, NF-E2 remains bound and allows recruitment of GATA-1, while p18 is not able to bind chromatin or recruit GATA-1. HS2bCAT plasmid was prebound with 6.1 or 12.3 pmol of either NF-E2 or the individual subunits p45 or p18. GATA-1 protein (6.1 pmol) was also included where indicated. Chromatin was digested with 1.5 U of DNase I per reaction, and plasmid footprinting analysis was carried out for the top strand with oligonucleotide B. Brackets show protection by GATA-1 or NF-E2. Open circles indicate protected bands and the closed circle refers to a hypersensitive band generated by NF-E2 binding.

from the binding of neighboring factors in HS2 and/or occupancy of the NF-E2 site by an alternate factor such as AP1, Nfr1/LCR-F1 (9, 10), or Nrf2 (35), all of which are capable of binding the NF-E2 site in vitro. As NF-E2 can form a DNase I hypersensitive site, and hypersensitive sites result from altered chromatin structures, it was exciting to discover that NF-E2 can bind chromatin and disrupt nucleosomes. This perturbation of chromatin structure is clearly seen by the smearing of MNase ladders and the generation of submononucleosomal-sized DNA fragments at the NF-E2 site upon factor binding. Plasmid footprinting of MNase-digested chromatin shows binding of NF-E2 but does not suggest that the translational positions of nearby nucleosomes are altered. We believe that the apparent discrepancy between the two assays is due to a combination of three points: the different levels of resolution of the assays, the distinct nature of the assays, and the lack of nucleosome positioning at

HS2 in our in vitro system. MNase ladders, despite their low resolution, have an advantage in that they can detect nucleosome disruption on what may be a small percentage of total DNA. Footprinting analysis, which has high resolution, requires that a majority of the chromatin-reconstituted plasmids be in the same conformation in order to be detected. NF-E2 disrupts nucleosomes, but they do not appear to be repositioned in fixed sites. There is sufficient variability in the positions which they now occupy to generate what appears to be a random pattern in footprinting assays. However, the same nucleosomal disruption results in a smear in a MNase ladder. The erythroid-cell-restricted subunit, p45, was found to be essential for chromatin interaction. Since p18 is able to bind HS2 only as naked DNA, this finding suggests that factor binding in a chromatin context is far more stringent. Perhaps the DNA-p18 homodimer interaction at HS2 is too labile to remain bound during chromatin assembly; the p45 subunit may

5642


MOL. CELL. BIOL.

FIG. 7. Disruption of nucleosomes by NF-E2 is ATP dependent. (A) Nucleosome disruption cannot occur in the presence of the ATP-hydrolyzing enzyme apyrase. HS2bCAT DNA (1 mg) was assembled into chromatin and incubated in the presence (1) or absence (2) of 0.2 U of apyrase (Sigma, grade VI) for 15 min at 278C; this was followed by incubation with 37 pmol of NF-E2. Chromatin was digested with MNase and the purified DNA was electrophoresed on an agarose gel as described in Materials and Methods. The membrane was hybridized with oligonucleotide C to the NF-E2 site. Arrow indicates submononucleosomal-sized fragments generated upon binding of NF-E2. (B) NF-E2 is able to bind to chromatin in the presence of apyrase. HS2bCAT DNA (1 mg) was assembled into chromatin and incubated in the presence (1) or absence (2) of 0.2 U of apyrase (Sigma, grade VI) for 15 min at 278C; this was followed by incubation with 7.3 pmol of NF-E2. Chromatin was digested with 30 U (lanes 1 and 3) or 37.5 U (lanes 2 and 4) of MNase for 10 min at room temperature. The DNA was purified, and plasmid footprinting analysis was carried out with oligonucleotide D.

act to stabilize this interaction. Interestingly, p18 is able to bind to HS2 in transient transfections and acts as a repressor in the absence of p45 (21). Nucleosome disruption by NF-E2 also occurred when the factor was added after chromatin assembly was complete, suggesting an active process. This was confirmed by the loss of nucleosome disruption in the presence of apyrase, the ATPhydrolyzing enzyme. Thus, NF-E2 must be dependent on a second activity, which utilizes energy, to disrupt nucleosomes. Multiple ATP-dependent factors capable of remodeling chromatin in vitro have recently been reported, such as SWI/SNF

from yeasts (12) and humans (26), NURF from Drosophila melanogaster (51), and another yet unidentified factor which allows general accessibility in chromatin (54). It is unknown whether nucleosome disruption by NF-E2 occurs by the same mechanism when the factor is added pre- or postchromatin assembly. When prebound to DNA before chromatin assembly, NF-E2 could be excluding nucleosomes from forming over HS2. Regarding the disruption seen when NF-E2 is added after assembly, two mechanisms are discussed as follows. First, it is possible that the chromatin is always in a state of dynamic flux (6). NF-E2 could take advantage of this constant change in

VOL. 16, 1996


nucleosome arrangement and, when the opportunity arises, bind to its site when the site becomes more available. Second, NF-E2 may target a nucleosome-reorganizing factor present in the S-190 extract to HS2. There the energy-dependent factor could disrupt chromatin structure and perhaps further promote NF-E2 binding. It will be informative to learn whether NF-E2 can interact with any of the identified chromatin-remodeling factors. It was very interesting to find that NF-E2 could bind to its site in chromatin in the absence of ATP. The separation of nucleosome disruption into two steps, the energy-independent step of factor binding and the energy-dependent factor-mediated chromatin reconfiguration, has been reported for GAL4-VP16 (41) and for both Drosophila heat shock factor and GAF/GAGA (55), and it is seen here for NF-E2. Nucleosome disruption by NF-E2 presents an interesting question: once the local area of chromatin is disturbed, are nearby sites for additional factors now accessible? Two inverted GATA-1 motifs are found 60 bp downstream of the NF-E2 site. Our results indicate that GATA-1 binding was increased in the presence of NF-E2. The individual subunits p45 or p18 did not bind to the NF-E2 site in chromatin and were not capable of recruiting GATA-1. The increased GATA-1 binding seen in chromatin appears to be a direct result of nucleosome disruption by NF-E2. This cooperative binding in a chromatin context could serve to nucleate transcriptional activators in the HS2 enhancer. The distance between the NF-E2 and the inverted GATA-1 sites in HS2 is conserved in a variety of b-globin elements, including human HS1, HS3, and HS4; the chicken b-globin enhancer; mouse HS2, HS3, and HS4; and goat HS2 and HS3 (44), suggesting a conserved mechanism of interaction between the two factors. Indeed, both NF-E2 and GATA-1 motifs are necessary for human b-globin HS4 formation (44). A classic example of nucleosome disruption by one factor leading to the binding of a second is the case of the glucocorticoid receptor binding to chromatin upon activation and recruiting the transcriptional activator NF1 (4, 5). It has also been shown by mononucleosomal binding studies that the binding of a variety of transcriptional activators to nucleosomes is inherently cooperative (1). The hypersensitive sites of the LCR appear to open the locus in preparation for globin gene expression. When HS2 to -4 are lost, as seen in the naturally occurring Hispanic (gdb)0-thalassemia deletion, the entire b-globin locus becomes late replicating and DNase I resistant, and no globin RNA is detected (16). Our experiments are a first step towards a more complete understanding at the molecular level of the early events in hypersensitive site formation and chromatin domain activation. Furthermore, as this in vitro chromatin assembly system reconstitutes a transcriptionally active human b-globin promoter in the presence of erythroid factors, it is ideal for investigation of the relationship between promoter activity and chromatin structure at both proximal and distal elements. In conclusion, we have characterized the interaction of NF-E2 with its site in chromatin-reconstituted HS2. Using this in vitro system, it will be interesting to determine if other erythroid transcription factors that bind to the LCR also possess the ability to disrupt nucleosomes and to examine how these proteins interact with each other in a chromatin context. ACKNOWLEDGMENTS We are grateful to Stephen Jane for the HS2bCAT plasmid, Stuart Orkin for p45 and p18 cDNAs, Mike Pazin for histone H1, Bin-Ru She for her NF-E2 preparation, Russell Kaufman and Scott Langdon for the mGATA-1 expression vector, and Phil Sheridan and members of the Kadonaga lab, especially Suman Paranjape and Mike Bulger, for

5643

assistance in Drosophila extract and histone preparation. We also thank Sara Anderberg, Lori Westin, Bin-Ru She, Rajesh Bagga, and Kathryn Calame for helpful discussions and/or critical comments on the manuscript. Much appreciation goes to Joseph Schulz for helpful discussions and critical comments on the manuscript. This work was supported by grants from the National Institutes of Health (GM38760) and the Mathers Foundation to B.M.E. REFERENCES 1. Adams, C. C., and J. L. Workman. 1995. Binding of disparate transcriptional activators to nucleosomal DNA is inherently cooperative. Mol. Cell. Biol. 15:1405–1421. 2. Andrews, N. C., H. Erdjument-Bromage, M. B. Davidson, P. Tempst, and S. H. Orkin. 1993. Erythroid transcription factor NF-E2 is a haematopoieticspecific basic-leucine zipper protein. Nature (London) 362:722–728. 3. Andrews, N. C., K. J. Kotkow, P. A. Ney, H. Erdjument-Bromage, P. Tempst, and S. H. Orkin. 1993. The ubiquitous subunit of erythroid transcription factor NF-E2 is a small basic-leucine zipper protein related to the v-maf oncogene. Proc. Natl. Acad. Sci. USA 90:11488–11492. 4. Archer, T. K., M. G. Cordingley, R. G. Wolford, and G. L. Hager. 1991. Transcription factor access is mediated by accurately positioned nucleosomes on the mouse mammary tumor virus promoter. Mol. Cell. Biol. 11: 688–698. 5. Archer, T. K., P. Lefebvre, R. G. Wolford, and G. L. Hager. 1992. Transcriptional factor loading on the MMTV promoter: a bimodal mechanism for promoter activation. Science 255:1573–1576. 6. Becker, P. B. 1994. The establishment of active promoters in chromatin. Bioessays 16:541–547. 7. Bresnick, E. H., M. Bustin, V. Marsaud, H. R. Foy, and G. L. Hager. 1992. The transcriptionally active MMTV promoter is depleted of histone H1. Nucleic Acids Res. 20:273–278. 8. Bulger, M., and J. T. Kadonaga. 1994. Biochemical reconstitution of chromatin with physiological nucleosome spacing. Methods Mol. Genet. 5:241– 262. 9. Caterina, J. J., D. Donze, C.-W. Sun, D. J. Ciavatta, and T. M. Townes. 1994. Cloning and functional characterization of LCR-F1: a bZIP transcription factor that activates erythroid-specific, human globin gene expression. Nucleic Acids Res. 22:2383–2391. 10. Chan, J. Y., X.-L. Han, and Y. W. Kan. 1993. Cloning of Nrf1, an NF-E2related transcription factor, by genetic selection in yeast. Proc. Natl. Acad. Sci. USA 90:11371–11375. 11. Chung, J. H., M. Whiteley, and G. Felsenfeld. 1993. A 59 element of the chicken b-globin domain serves as an insulator in human erythroid cells and protects against position effect in Drosophila. Cell 74:505–514. 12. Coˆte´, J., J. Quinn, J. L. Workman, and C. L. Peterson. 1994. Stimulation of GAL4 derivative binding to nucleosomal DNA by the yeast SWI/SNF complex. Science 265:53–60. 13. Croston, G., L. Lira, and J. Kadonaga. 1991. A general method for purification of H1 histones that are active for repression of basal RNA polymerase II transcription. Protein Expr. Purif. 2:162–169. 14. Elgin, S. C. R. 1995. Chromatin structure and gene expression. Oxford University Press, New York. 15. Enver, T., N. Raich, A. J. Ebens, T. Papayannopoulou, F. Constantini, and G. Stamatoyannopoulos. 1990. Developmental regulation of human fetal-toadult globin gene switching in transgenic mice. Nature (London) 344:309– 313. 16. Forrester, W. C. F., E. Epner, M. C. Driscoll, T. Enver, M. Brice, T. Papayannopoulou, and M. Groudine. 1990. A deletion of the human b-globin locus activation region causes a major alteration in chromatin structure and replication across the entire b-globin locus. Genes Dev. 4:1637–1649. 17. Fraser, P., S. Pruzina, M. Antoniou, and F. Grosveld. 1993. Each hypersensitive site of the human b-globin locus control region confers a different developmental pattern of expression of the globin genes. Genes Dev. 7:106– 113. 18. Goldman, M. A., G. P. Holmquist, M. C. Gray, L. A. Caston, and A. Nag. 1984. Replication timing of genes and middle repetitive sequences. Science 224:686–692. 19. Gross, D. S., and W. T. Garrard. 1988. Nuclease hypersensitive sites in chromatin. Annu. Rev. Biochem. 57:159–197. 20. Grosveld, F., G. Blom van Assendelft, D. R. Greaves, and G. Kollias. 1987. Position-independent, high-level expression of the human b-globin gene in transgenic mice. Cell 51:975–985. 21. Igarashi, K., K. Kataoka, K. Itoh, N. Hayashi, M. Nishizawa, and M. Yamamoto. 1994. Regulation of transcription by dimerization of erythroid factor NF-E2 p45 with small Maf proteins. Nature (London) 367:568–572. 22. Jane, S. M., P. A. Ney, E. F. Vanin, D. L. Gumucio, and A. W. Nienhuis. 1992. Identification of a stage selector element in the human gamma-globin gene promoter that fosters preferential interaction with the 59 HS2 enhancer when in competition with the b-promoter. EMBO J. 11:2961–2969. 23. Jime´nez, G., S. D. Griffiths, A. M. Ford, M. F. Greaves, and T. Enver. 1992. Activation of the b-globin locus control region precedes commitment to the

5644


erythroid lineage. Proc. Natl. Acad. Sci. USA 89:10618–10622. 24. Kamakaka, R. T., M. T. Bulger, and J. T. Kadonaga. 1993. Potentiation of RNA polymerase II transcription by Gal4-VP16 during but not after DNA replication and chromatin assembly. Genes Dev. 7:1779–1795. 25. Kotkow, K. J., and S. H. Orkin. 1995. Dependence of globin gene expression in mouse erythroleukemia cells on the NF-E2 heterodimer. Mol. Cell. Biol. 15:4640–4647. 26. Kwon, H., A. N. Imbalzano, P. A. Khavari, R. E. Kingston, and M. R. Green. 1994. Nucleosome disruption and enhancement of activator binding by a human SW1/SNF complex. Nature (London) 370:477–481. 27. Labbaye, C., M. Valtieri, T. Barberi, E. Meccia, B. Masella, E. Pelosi, G. L. Condorelli, U. Testa, and C. Peschle. 1995. Differential expression and functional role of GATA-2, NF-E2, and GATA-1 in normal adult hematopoiesis. J. Clin. Invest. 95:2346–2358. 28. Laybourn, P., and J. Kadonaga. 1991. Role of nucleosomal cores and histone H1 in regulation of transcription by RNA polymerase II. Science 254:238– 245. 29. Li, Q., and G. Stamatoyannopoulos. 1994. Hypersensitive site 5 of the human beta locus control region functions as a chromatin insulator. Blood 84:1399–1401. 30. Lozzio, C. B., and B. B. Lozzio. 1975. Human chronic myelogenous leukemia cell-line with positive Philadelphia chromosome. Blood 45:321–334. 31. Lu, S.-J., S. Rowan, M. R. Bani, and Y. Ben-David. 1994. Retroviral integration within the Fli-2 locus results in inactivation of the erythroid transcription factor NF-E2 in Friend erythroleukemias: evidence that NF-E2 is essential for globin expression. Proc. Natl. Acad. Sci. USA 91:8398–8402. 32. Lui, D., J. C. Chang, P. Moi, W. Liu, Y. W. Kan, and P. T. Curtin. 1992. Dissection of the enhancer activity of b-globin 59 DNase I-hypersensitive site 2 in transgenic mice. Proc. Natl. Acad. Sci. USA 89:3899–3903. 33. Martin, D. I. K., and S. H. Orkin. 1990. Transcriptional activation and DNA binding by the erythroid factor GF-1/NF-E1/Erf1. Genes Dev. 4:1886–1898. 34. Miller, I. J., and J. J. Bieker. 1993. A novel, erythroid cell-specific murine transcription factor that binds to the CACCC element and is related to the Kruppel family of nuclear proteins. Mol. Cell. Biol. 13:2776–2786. 35. Moi, P., K. Chan, I. Asunis, A. Cao, and Y. W. Kan. 1994. Isolation of NF-E2-related factor 2 (Nrf2), a NF-E2-like basic leucine zipper transcriptional activator that binds to the tandem NF-E2/AP1 repeat of the b-globin locus control region. Proc. Natl. Acad. Sci. USA 91:9926–9930. 36. Ney, P., B. P. Sorrentino, K. T. McDonagh, and A. W. Nienhuis. 1990. Tandem AP-1-binding sites within the human b-globin dominant control region function as an inducible enhancer in erythroid cells. Genes Dev. 4:993–1006. 37. Ney, P. A., N. C. Andrews, S. M. Jane, B. Safer, M. E. Purucker, S. Weremowicz, C. C. Morton, S. C. Goff, S. H. Orkin, and A. W. Nienhuis. 1993. Purification of the human NF-E2 complex: cDNA cloning of the hematopoietic cell-specific subunit and evidence for an associated partner. Mol. Cell. Biol. 13:5604–5612. 38. Orkin, S. H. 1990. Globin gene regulation and switching: circa 1990. Cell 63:665–672. 39. Orkin, S. H. 1995. Regulation of globin gene expression in erythroid cells. Eur. J. Biochem. 231:271–281. 40. Orlic, D., S. Anderson, L. G. Biesecker, B. P. Sorrentino, and D. M. Bodine. 1995. Pluripotent hematopoietic stem cells contain high levels of mRNA for c-kit, GATA-2, p45 NF-E2, and c-myb and low levels or no mRNA for c-fms and the receptors for granulocyte colony-stimulating factor and interleukins 5 and 7. Proc. Natl. Acad. Sci. USA 92:4601–4605.

MOL. CELL. BIOL. 41. Pazin, M. J., R. T. Kamakaka, and J. T. Kadonaga. 1994. ATP-dependent nucleosome reconfiguration and transcriptional activation from preassembled chromatin templates. Science 266:2007–2011. 42. Ryan, T. M., R. R. Behringer, N. C. Martin, T. M. Townes, R. D. Palmiter, and R. L. Brinster. 1989. A single erythroid-specific DNase I super-hypersensitive site activates high levels of human b-globin gene expression in transgenic mice. Genes Dev. 3:314–323. 43. Shivdasani, R. A., and S. H. Orkin. 1995. Erythropoiesis and globin gene expression in mice lacking the transcription factor NF-E2. Proc. Natl. Acad. Sci. USA 92:8690–8694. 44. Stamatoyannopoulos, J. A., A. Goodwin, T. Joyce, and C. H. Lowrey. 1995. NF-E2 and GATA binding motifs are required for the formation of DNase I hypersensitive site 4 of the human b-globin locus control region. EMBO J. 14:106–116. 45. Talbot, D., P. Collis, M. Antoniou, M. Vidal, F. Grosveld, and D. R. Greaves. 1989. A dominant control region from the human b-globin locus conferring integration site-independent gene expression. Nature (London) 338:352– 355. 46. Talbot, D., and F. Grosveld. 1991. The 59 HS2 of the globin locus control region enhances transcription through the interaction of a multimeric complex binding at two functionally distinct NF-E2 binding sites. EMBO J. 10:1391–1398. 47. Talbot, D., S. Philipsen, P. Fraser, and F. Grosveld. 1990. Detailed analysis of the site 3 region of the human b-globin dominant control region. EMBO J. 9:2169–2178. 48. Trainor, C. D., T. Evans, G. Felsenfeld, and M. S. Boguski. 1990. Structure and evolution of a human erythroid transcription factor. Nature (London) 343:92–96. 49. Tsai, S. F., D. I. K. Martin, L. I. Zon, A. D. D’Andrea, G. G. Wong, and S. H. Orkin. 1989. Cloning of cDNA for the major DNA-binding protein of the erythroid lineage through expression in mammalian cells. Nature (London) 339:446–451. 50. Tsukiyama, T., P. B. Becker, and C. Wu. 1994. ATP-dependent nucleosome disruption at a heat-shock promoter mediated by binding of GAGA transcription factor. Nature (London) 367:525–532. 51. Tsukiyama, T., and C. Wu. 1995. Purification and properties of an ATPdependent nucleosome remodeling factor. Cell 83:1011–1020. 52. Tuan, D., W. Solomon, Q. Li, and I. M. London. 1985. The “b-like-globin” gene domain in human erythroid cells. Proc. Natl. Acad. Sci. USA 82:6384– 6388. 53. Tuan, D. Y. H., W. B. Solomon, I. M. London, and D. P. Lee. 1989. An erythroid-specific, developmental-stage-independent enhancer far upstream of the human “b-like globin” genes. Proc. Natl. Acad. Sci. USA 86:2554– 2558. 54. Varga-Weisz, P. D., T. A. Blank, and P. B. Becker. 1995. Energy-dependent chromatin accessibility and nucleosome mobility in a cell-free system. EMBO J. 14:2209–2216. 55. Wall, G., P. D. Varga-Weisz, R. Sandaltzopoulos, and P. B. Becker. 1995. Chromatin remodeling by GAGA factor and heat shock factor at the hypersensitive Drosophila hsp 26 promoter in vitro. EMBO J. 14:1727–1736. 56. Weintraub, H., and M. Groudine. 1976. Chromosomal subunits in active genes have an altered conformation. Science 193:848–856. 57. Wolffe, A. P. 1994. Transcription: in tune with the histones. Cell 77:13–16. 58. Workman, J. L., and A. R. Buchman. 1993. Multiple functions of nucleosomes and regulatory factors in transcription. Trends Biochem. Sci. 18:90– 95.