![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Previous Article | Next Article 
Molecular and Cellular Biology, April 2007, p. 2870-2885, Vol. 27, No. 8
0270-7306/07/$08.00+0 doi:10.1128/MCB.02323-06
Copyright © 2007, American Society for Microbiology. All Rights Reserved.
Brett V. Johnson,2,,
Euan W. Baxter,2 and
Peter N. Cockerill2*
Hanson Institute, Institute of Medical and Veterinary Science, Adelaide 5000, Australia,1 Experimental Haematology, Leeds Institute of Molecular Medicine, St. James's University Hospital, Leeds LS9 7TF, United Kingdom2
Received 12 December 2006/ Returned for modification 24 January 2007/ Accepted 27 January 2007
 |
ABSTRACT |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
There is also considerable evidence that the context of transcription factor binding sites is crucial to their function, with arrays of specifically positioned elements assembling multiprotein complexes termed enhanceosomes (5, 8, 28, 29). Enhanceosomes are likely to assemble at sites of protein-protein interactions on DNA, such as those that join NFAT and AP-1 at composite elements (8), and can assemble using coactivators such as CBP and P300 as scaffolding proteins (5, 7, 29). We have, therefore, also investigated whether different specific clusters of transcription factors function within enhancers as discrete enhanceosome-like complexes.
We have employed the human granulocyte-macrophage colony-stimulating factor (GM-CSF) locus as a model to unravel the mechanisms utilized by diverse cell types to activate the same gene in a highly tissue-specific pattern (9). The GM-CSF gene can be activated by a variety of signaling pathways and is expressed inducibly in a diverse range of cell types that includes T cells, monocytes, mast cells, myeloid precursor cells, endothelial cells, fibroblasts, and epithelial cells. The GM-CSF gene is regulated by an inducible enhancer located 3 kb upstream of the GM-CSF gene that functions in most GM-CSF-expressing cell types (11). The enhancer is activated via protein kinase and calcium signaling pathways, which in T cells are coupled to the T-cell receptor.
The GM-CSF enhancer encompasses three composite NFAT/AP-1 elements, termed the GM330, GM420, and GM550 elements, which bind NFAT and AP-1 in a highly cooperative manner, in addition to the GM170 NFAT and AP-1 sites that do not function cooperatively (10) (as depicted in Fig. 2A below). The enhancer core also encompasses a Runx1 (AML1) site (GM450) and an Sp1-like element (GM300) (9). In recent studies, we demonstrated that, in T cells, enhancer function requires just the two centrally located NFAT/AP-1 elements (GM330 and GM420) that exist on adjacent nucleosomes (termed N1 and N2) (21) that span the conserved core region (9). Enhancer activation is accompanied by the rapid induction of a DNase I-hypersensitive (DH) site spanning these two specifically positioned nucleosomes, which become destabilized upon recruitment of NFAT/AP-1 complexes to the GM330 and GM420 elements. Enhancer function in T cells is absolutely dependent on the integrity of both the NFAT and AP-1 motifs within each of the GM330 and GM420 elements. An Sp1-like element (GM300) located just upstream of the GM330 element is also required for maximal activity.
|
The GM-CSF enhancer encompasses additional conserved sequences that extend immediately upstream of the essential core required for function in T cells (9) (see Fig. 2 below). One of these conserved regions includes two GATA-like elements, the GM170 AP-1 site, and an Ets-like element, which are all redundant for enhancer function in Jurkat T-cell transfection assays and lie outside of the DH site in T cells (21). However, we speculated that this region might direct an alternate mode of GATA-dependent transcription in erythroid/megakaryocytic lineage cells expressing GATA-1 and in mast cells expressing GATA-1 and GATA-2. In support of this model, we found that the GATA elements define the 3' boundary of an upstream nucleosome (N0) that undergoes constitutive chromatin remodeling to form a DH site in myeloid cells that express GATA factors but not NFAT. In erythroleukemic and mast cell lines, the GATA elements were indeed required for enhancer function. In bone marrow-derived mast cells, we observed that GATA-2 binding to the enhancer was accompanied by repositioning of nucleosome N0 to a location 140 bp upstream from the position that it occupies in T cells and repositioning of the adjacent nucleosomes. Hence, this study reveals that the GM-CSF enhancer can function in a modular fashion, utilizing different sets of transcription factors in different cell types to establish alternate specific chromatin architectures and thus drive different expression patterns. These findings also imply that the transcription factors themselves dictate nucleosome positioning. We further demonstrate that enhancer activation was accompanied by a profound change in chromatin architecture over the flanking regions in both T cells and mast cells, where extensive nucleosome mobilization was observed.
|
MATERIALS AND METHODS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Human GM-CSF transgenic mouse mast cells were grown from adult femur and tibia bone marrow by culture in Iscove's modified Dulbecco's medium supplemented with 10% fetal calf serum (HyClone), 150 µM monothioglycerol, 500 U/ml penicillin and streptomycin, 10 ng/ml recombinant mouse IL-3 (Biosource), and 10 ng/ml recombinant mouse stem cell factor (SCF; R and D Systems). After 1 week, nonadherent cells were moved into a new flask and cultured at a density of 0.5 x 106 to 1 x 106 for a further 3 weeks with several changes of flask to remove any emerging adherent macrophages from the culture.
Mast cell analyses. Mast cells were stained using toluidine blue in 1% NaCl (pH 2.3) and were seen to have purple granules in the cytoplasm and round nuclei. Mast cell identity was further confirmed by fluorescence-activated cell sorting and cells stained positive using antibodies against c-Kit, CD34, and Sca-1, all of which are known to be expressed on mouse mast cells (14).
mRNA analyses. Human GM-CSF mRNA expression relative to mouse glyceraldehyde phosphate dehydrogenase (GAPDH) mRNA expression was measured by real-time PCR analysis with SYBR green as previously described (21). Primers used for the mouse GAPDH, Oct1A, Runx1, NFATc1, PU.1, GATA-1, and GATA-2 genes were designed using PCR Express and Oligo 5 software to anneal at 60°C. Standard curves for mRNA quantitation were prepared using serial dilutions of stimulated mast cell cDNA. The sequences used as primers were as follows: mOct-1A, GGACCAGCAGCTCACCTATCAA and CCG TACTGCTTGTCACAAGGCT; mRunx1, AGCGGTAGAGGCAAGAGCTTC and CGGATTTGTAAAGACGGTGATG; mNFATc1, TGGAAGCAAAGACTGACCGG and CTGGTTGCGGAAAGGTGGTAT; mPU.1, CCATAGCGATCACTACTGGGATTT and TGTGAAGTGGTTCTCAGGGAAGT; mGATA-1, GTCAGAACCGGCCTCTCATC and GTGGTCGTTTGACAGTTAGTGCAT; mGATA-2, TAAGCAGAGAAGCAAGGCTCGC and ACAGGCATTGCACAGGTAGTGG; and m c-fms, CTTTGGTCTGGGCAAGAAGAT and CAGGGCCTCCTTCTCATCAG.
ELISAs. Human and mouse GM-CSF protein concentrations were measured by enzyme-linked immunosorbent assay (ELISA) of the culture media, after stimulation of splenic cells or peritoneal cells, as previously described (11). Relative levels of GM-CSF transgene expression per gene copy were calculated as the ratio of human GM-CSF to mouse GM-CSF to transgene copy number, multiplied by 2 to correct for the mouse genome.
EMSAs. Electrophoretic mobility shift assays (EMSAs) were performed as described previously (21), using K562 nuclear extracts and the following probe and competitor sequences: GM138 GATA, CCCCTTGTGATAATGTCTCTCGT; GM153 GATA, TCTCTCGTGATAAGGATCCTGGA; GM-CSF promoter –114 GATA, GTGTGGCTGATAAGGGCCAGGAG; GATA consensus, CAGGGACATGATAAGGGAGCCAA; GM300 Sp1 site, TCGTGGTCCCCGCCTCTGCCCTGCC; GM-CSF promoter Sp1 site, AGTTCCCCCGCCTCCCTGGCATTTT; and Sp1 consensus, GCTCGCCCCGCCCCGATCGAATG.
Analysis of nuclease cleavage sites within nuclei. DNase I and micrococcal nuclease (MNase) digestions and analysis of chromatin fragments were performed essentially as previously described (21). DNase I (6 U/µg) and MNase were obtained from Worthington. The DH site analyses of human cell nuclei presented in Fig. 5A employed previously characterized DNase I-digested samples that represent optimally digested samples taken from series of DNase I titrations where we have previously demonstrated that we can efficiently detect constitutive and inducible DH sites (11). In each case, cells had been either not stimulated or stimulated for 4 to 6 h with 20 ng/ml phorbol myristate acetate (PMA) and 2 µM calcium ionophore A23187.
|
30 to 100 U/ml) DNase I for 3 min. Analyses of mono- and oligonucleosome fragments by Southern blot hybridization were performed using 2% agarose gels and alkaline transfer (in 0.5 M NaOH plus 1.5 M NaCl) to GeneScreen Plus (Perkin-Elmer) nylon membranes as described previously (21), using 1.5 µg of DNA samples obtained from chromatin digested within the range of 50 to 500 U/ml MNase for 3 to 15 min at 22°C. Specific nucleosomal regions were probed using the N2 probe described previously (21), and PCR generated fragments using the N0 primers TCTCAGGTCCCCAGAGATAACAAG and GGGCTGATCTGATCCTTCTTTG and the primers located 1.3 kb upstream of the gene (–1.3-kb primers) GACCACAGTGCTAAAGCCATGA and GGTTGGACACAAGCGAGAACA. The 3' end of the locus (+4.0 to +4.8 kb) was probed using a 0.85-kb BglII-HindIII fragment.
All imaging and densitometry of hybridization patterns were performed using a Bio-Rad FX phosphorimager and Quantity One software to collect images and traces of digestion patterns. Excel software (Microsoft) was used to digitally calculate ratios between each point on the traces derived from chromatin digestions and DNA control digestions and to transform migration distances within lanes to sizes in base pairs via an equation derived from mobilities of the restriction enzyme fragments.
ChIP. Chromatin immunoprecipitation (ChIP) assays were performed essentially as published in reference 2, with the following minor modifications. Cells were fixed in 0.33 M (1%) formaldehyde for 5 min, before adding 5 volumes of phosphate-buffered saline containing 0.125 M glycine, to provide a twofold molar excess of glycine over formaldehyde. DNA was purified after proteinase K digestion using QIAGEN PCR purification columns. ChIPs employed 5 µg of antibodies recognizing either mouse and human GATA-2 (sc-9008; Santa Cruz Biotech), acetyl histone H3 K9 (07-352; Upstate Biotechnology), pan-acetyl istone H4 (06-598; Upstate Biotechnology), or unmodified histone H3 (1791-100; Abcam). Control ChIPs employed purified normal rabbit immunoglobulin G (IgG) (12-370; Upstate Biotechnology). Real-time PCR was used to determine the amount of each gene-specific amplicon present. All values were determined from standard curves using input DNA (DNA purified from sonicated chromatin before ChIP). All ChIP data were normalized by expressing the amount of each specific DNA precipitated as a ratio with the values obtained using a region of mouse chromosome 1. This control was selected on the basis that it is located within an extensive gene-free region, is free of repetitive or conserved sequences, and is at least 200 kb from the nearest known transcribed sequence. As a positive control, we selected the intronic enhancer from the IL-4 gene which is known to bind GATA-2 in mast cells in vivo (24). For negative controls, we employed (i) a region 1.5 kb upstream of the c-fms promoter, which we confirmed was inactive in mast cells, and (ii) a region 17 kb upstream of the mouse PU.1 gene.
The sequences of the primers were as follows: m IL-4 intron 2, AACCACCCCACAGGTCTTTGTT and TGAAAGGCCCCAAAGTCTTGA; m c-fms –1.5 kb, CACGCCGGCTGAGTGTCT and TCCACGTAGATGGTGTCAGCAT; m chromosome 1 control sequence, TGCTCCACAGTGTCCATGTACA and AGCAATTTCATGGGTGAGAGAAG; h GM-CSF enhancer, GGAGCCCCTGAGTCAGCAT and CATGACACAGGCAGGCATTC; h GM-CSF promoter, TGTCGGTTCTTGGAAAGGTTCA and TGTGGAATCTCCTGGCCCTTA; and m PU.1 –17 kb, CCTCCCAAGTGCTGGGATTAA and TGAGCACACGAACACACTTAAGTGTA.
Transient transfection assays. Transfection assays and luciferase assays were performed using the luciferase reporter gene plasmid pXPG plus either just the bp –627 to +28 segment of the human GM-CSF promoter (pXPG-GM627) or the promoter plus the 717-bp fragment of the human GM-CSF enhancer (pXPG-GM627-B717), with and without point mutations, as described previously (21). The plasmid derivatives BM575, BH355, BA260, AM315, and HM231 contain the enhancer fragments depicted in Fig. 4 in place of the BglII fragment.
|
|
RESULTS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The GM-CSF enhancer is essential for GM-CSF gene expression in myeloid cells. We previously determined that the GM-CSF enhancer is required for efficient GM-CSF gene expression in spleen cells derived from human GM-CSF transgenic mice (11). While T cells represent the main source of GM-CSF expression in the spleen, in this study we aimed to identify an additional role for the GM-CSF enhancer in the myeloid compartment and to determine whether it functioned by the same mechanisms in these different cell types.
We first investigated GM-CSF enhancer function in both spleen cells and a mixed population of myeloid cells recovered from the peritoneal cavity of GM-CSF transgenic mice. The peritoneal population primarily includes GM-CSF-expressing cell types such as monocytes and macrophages. Cells were obtained from both the peritonea and spleens of seven lines of transgenic mice containing the intact 10.5-kb GM-CSF locus and six lines of GM-CSF transgenic mice lacking the enhancer (
E GM-CSF) (Fig. 1A). Human and mouse GM-CSF expression levels were assayed by ELISA after stimulation with phorbol ester and calcium and calcium ionophore (PMA/I) (Fig. 1B). These agents act predominantly to activate kinase pathways that induce AP-1 and NF-
B and calcium pathways that induce NFAT, although there is also a lot of cross talk between these pathways in the activation of these factors.
In the spleen, the activity per gene copy of the intact human GM-CSF locus was equal to or greater than that of each copy of the endogenous mouse GM-CSF gene (Fig. 1B). Transgenes lacking the enhancer were on average 14% as active as the intact transgenes. In the peritoneal cells, the intact human GM-CSF gene was several times more active than the mouse GM-CSF gene. In these myeloid cells, deletion of the enhancer almost abolished gene activity, reducing average expression levels to just 2.6% of that of the wild-type transgene. Similar results were obtained with peritoneal cells enriched for macrophages by adherence to culture dishes (data not shown).
In myeloid cells, cytokine gene activation can also be triggered by bacterial lipopolysaccharide (LPS) via Ca2+-independent pathways that induce AP-1 and NF-B. To determine whether the GM-CSF enhancer could also be activated via these pathways, we assayed transgene expression in peritoneal cells stimulated with LPS. The intact transgene was induced by LPS to a level similar to that of the endogenous GM-CSF gene, and loss of the enhancer reduced expression levels to just 1% of wild type on average (Fig. 1). Hence, the enhancer appears to be essential for GM-CSF expression induced in myeloid cells by both NFAT-dependent and -independent pathways.
We next purified CD11b-positive myeloid cells from the spleens of several of our lines of GM-CSF transgenic mice. These cells represent primarily monocyte/macrophage lineage cells and account for about 5% of the cells in the spleen. Cells were stimulated for 4 h with PMA/I, and the levels of GM-CSF transgene mRNA expression were determined relative to mouse Gapdh mRNA expression. Mirroring the above observations, the intact GM-CSF locus was induced to similar levels of expression in two independent lines of mice, whereas transgene expression was almost eliminated in mouse lines lacking the GM-CSF enhancer (Fig. 1C). However, the level of GM-CSF mRNA expression in splenic myeloid cells was only about 0.1 to 0.2% of that of Gapdh. This is in marked contrast to activated T cells, which express approximately the same levels of GM-CSF and Gapdh mRNA (data not shown).
The GM-CSF enhancer and promoter encompass GATA and Sp1 elements. Most myeloid cell types (other than macrophages) employ GATA family transcription factors to drive differentiation and specific gene expression. These include myeloid precursor cells, mast cells, erythroid cells, megakaryocytes, and eosinophils, which express GATA-1 and/or GATA-2 to various degrees, with GATA-1 most prominent in erythroid cells and GATA-2 most prominent in mast cells (31). It is, therefore, highly significant that the GM-CSF enhancer also encompasses two stretches of conserved sequence which lie just upstream of the core elements that function in T cells and include two conserved ideal consensus GATA elements (Fig. 2). These two closely coupled GATA sequences (TGATAAT and TGATAAG) reside between a conserved Ets-1-like consensus sequence and the downstream GM170 AP-1 site (Fig. 2B). We defined these GATA-like sequences as the GM138 and GM153 elements, according to their location within the 717-bp BglII enhancer fragment, in line with the convention used for all elements within the GM-CSF enhancer. To determine whether these two elements function as genuine GATA sites, we performed EMSAs with nuclear extracts prepared from the erythroleukemic cell line K562, which expresses high levels of GATA-1. The GM138 and GM153 probes each functioned as highly efficient GATA-1-binding elements that could both compete for GATA-1 binding and be blocked by a GATA consensus sequence (Fig. 2D).
We also performed EMSAs on a conserved GATA-like element (TGATAAG) that is located in the GM-CSF promoter 114 bp upstream of the transcription start site. This sequence also functioned as an efficient GATA-1-binding element (Fig. 2D). We performed additional GATA EMSAs to define cell lines representing useful models that express high or low levels of GATA family factors (Fig. 2E). The HMC-1 mast cell line expressed significant levels of GATA-like factors, which comigrated with the K562 GATA-like complexes, whereas Jurkat and CEM T cells both expressed much lower levels. Because GATA elements frequently cooperate with Sp1 sites, we performed additional EMSAs with Sp1-like elements located in the GM-CSF enhancer (GM300) and promoter (9). These assays confirmed that both sites functioned as efficient Sp1 sites when assayed in parallel with an ideal consensus Sp1 site (Fig. 2F).
Mast cells as a model for GATA-dependent GM-CSF gene regulation. Mast cells are known to utilize both GATA and NFAT family transcription factors in the regulation of other cytokine genes located near the GM-CSF locus (24, 30), and GATA-2 is a major factor directing mast cell differentiation (35). To prepare an enriched population of mast cells, we cultured nonadherent bone marrow cells in the presence of IL-3 and SCF for approximately 4 weeks. The resulting cells resembled mature differentiated mast cells in that they were uniformly packed with toluidine blue staining granules and were essentially 100% Sca-1, CD34, and c-Kit positive (data not shown) (14). We confirmed that these cells represented an appropriate model for studying mechanisms of GM-CSF enhancer regulation by demonstrating (i) that they expressed high levels of the transcription factors GATA-1, GATA-2, Runx1, and NFATc1 relative to the ubiquitous factor Oct-1 (Fig. 3A) and (ii) that they expressed very high levels of GM-CSF mRNA expression relative to Gapdh mRNA after stimulation with PMA/I (Fig. 3B). The relative level of GM-CSF gene expression in mast cells was approximately 104-fold higher than that detected in splenic CD11b-positive myeloid cells (Fig. 1C) and severalfold higher than that detected in transgenic T cells (data not shown). This suggests that mast cells and T cells represent more meaningful sources of GM-CSF than monocyte-macrophage lineage cells. Significantly, we showed that the GM-CSF enhancer was essential for GM-CSF gene expression in mast cells because transgenes lacking the enhancer expressed almost undetectable levels of GM-CSF mRNA (Fig. 3B). We also demonstrated that the enhancer existed as a highly inducible DH site in mast cells, but in contrast to T cells (21), it also existed as a weak DH site prior to stimulation (Fig. 3C), suggesting a different pattern of regulation. For DNase I control purposes, we performed a parallel analysis of a cluster of DH sites located just downstream of the IL-3 gene (Fig. 3C) that includes three ubiquitous constitutive DH sites and one inducible site (9). This analysis confirmed that the two samples were digested to similar extents by DNase I and demonstrated that the IL-3 kb +4.5 DH site is also inducible in mast cells. Note that the GM-CSF promoter appears to be decrease in hypersensitivity after stimulation (Fig. 3C) as a result of the increased cleavage at the enhancer which eliminates the longer fragments. In contrast, the analyses described below in Fig. 6 demonstrate that hypersensitivity at the promoter actually increases after stimulation.
|
|
The GM-CSF enhancer is modular and is differentially regulated in T cells and myeloid cells. The transcription factor binding sites in the GM-CSF enhancer are organized as discrete clusters of sites that potentially function as differentially regulated enhanceosome-like complexes. To assess the degree of tissue specificity and the cooperativity of the different regions of the enhancer, we performed transient transfection assays using GM-CSF promoter/enhancer luciferase reporter gene plasmids (Fig. 4) in a panel of myeloid and T-cell lines that express GATA and/or NFAT family proteins. This panel included Jurkat and CEM T cells that each express NFAT but little GATA protein, K562 and HEL pro-erythroid/megakaryocytic cells that express GATA-1 but little NFAT, and HMC-1 mast cells and KG1a CD34-positive primitive myeloid cells that express both families of factors (Fig. 2E) (3; data not shown).
The first set of luciferase reporter gene constructs tested each contained the bp –627 to +28 fragment of the human GM-CSF gene promoter (GM627), with or without different combinations of the three major clusters of regulatory elements that exist within the enhancer (Fig. 4A). The 575-bp BglII/MscI segment (BM575) encompasses all of the known binding sites and nuclease-sensitive regions and functioned as an inducible enhancer to essentially the same extent as the previously defined 717-bp BglII fragment (GME) in all of the cell lines tested. These constructs only supported near-background levels of activity in the absence of stimulation (data not shown). In these studies, it was also significant that the relative activity of the enhancer compared to that of the promoter alone was five- to eightfold greater in the GATA-1-expressing cell lines K562 and HEL than it was in Jurkat cells.
A truncated 315-bp ApaI/Msc I segment (AM315) that encompasses the GM330 and GM420 elements retained full enhancer activity in Jurkat T cells. However, this segment lacks the GATA/GM170 region and had little activity in the HEL, HMC-1, and KG1a myeloid cell lines and reduced activity in CEM and K562 cells. Conversely, the BglII/HincII segment (BH355), which includes the GATA, GM170, and GM330 elements but not the GM420 element, had no activity in the Jurkat and CEM T-cell lines but retained full or substantial activity in all of the myeloid cell lines. The BA260 and HM231 constructs lacking the central GM330 region were essentially inactive in most cases, except for HMC-1 cells, where the BA260 GATA region was sufficient for significant enhancer function.
To identify roles for individual elements in the enhancer, we performed site-directed mutagenesis in the context of the 717-bp BglII fragment inserted in pGM627 (Fig. 4B). In line with the above results, mutation of the GATA elements had no effect in Jurkat cells but decreased enhancer activity in all of the other cell lines tested. The combined mutation of all four defined NFAT sites abolished activity in Jurkat and CEM T cells, but had no effect in HMC-1 cells and intermediate effects in the other cell lines. Mutation of just the GM420 NFAT site also abolished activity in Jurkat and CEM T cells, but had little or no impact in KG1a or HMC-1 cells (data not shown). Mutations in individual AP-1 sites had very specific effects depending upon their context. The GM170 AP-1 site adjacent to the GATA elements was required for function in GATA-expressing K562, HEL, and HMC-1 cells, but not in Jurkat cells. Conversely, the GM330 and GM420 AP-1 sites, which are both linked to NFAT sites, were essential for function in Jurkat T cells but played no role in HMC-1 cells. This indicates that the different AP-1 sites must cooperate in a specific differentially regulated manner with neighboring sites. The Sp1 site contributed to function in every cell type, but was not essential. There was, however, no effect from mutations in the GM170 or GM550 NFAT sites, the GM450 Runx1 site, or the Ets-like element in any of the cell lines (data not shown).
The GATA region of the GM-CSF enhancer forms a distinct DH site in myeloid cells. Active regions within enhancers and promoters are commonly associated with discrete domains of chromatin remodeling that are detected as DH sites. In order to obtain evidence whether specific enhancer modules in the endogenous GM-CSF gene are active in different cell types, we proceeded to map DH sites in the cells used in the above studies. In addition, we included peripheral blood T cells, which express GM-CSF and NFAT; HUVECs, which express GM-CSF, NFAT and GATA factors; and HeLa carcinoma cells, which express GM-CSF but not GATA or NFAT. DH sites were mapped at high resolution across the GM-CSF enhancer from a PshA1 site located 1.5 kb upstream of the BglII site that defines the 5' boundary of the enhancer (Fig. 5A and B). In all T lineage cells, the DH site was only present in stimulated cells, and was confined to nucleosomes N1 and N2, which encompass the GM330 and GM420 elements. This region is roughly defined by the ApaI and PstI sites and includes all of the sites known to be required for enhancer function in T cells (21). In myeloid cells, the DNase I mapping revealed distinctly different patterns of DH sites. Significantly, the GATA-1- and/or GATA-2-expressing K562, HEL, and HMC-1 cells each had a novel myeloid-specific DH site spanning the GATA elements and nucleosome N0 that reside between the ApaI and 5' BglII sites, and this site was detected in both stimulated and unstimulated cells. (In parallel experiments, this site was maintained in the same position in activated HEL cells [data not shown].) This lower myeloid-specific DH site was in some cases accompanied by an upper DH site equivalent to the inducible site seen in T cells.
Stimulated endothelial cells (HUVECs), which express both GATA and NFAT family proteins, also had two DH sites spanning the GATA and GM330/GM420 regions. KG1a cells, however, presented a uniquely different pattern. KG1a cells had no constitutive DH site, but had a broad inducible DH site spanning the ApaI site and including the novel NF-B and NFAT-like elements at positions 220 and 240 (that were identified as conserved elements in Fig. 2C), but not extending into the nucleosome N0 region. Activated HeLa cells had a single sharply defined DH site centered between nucleosomes N1 and N2, midway between the ApaI and PstI sites.
The pattern of chromatin remodeling largely reflected the pattern of DNA element utilization in each cell type and provided further evidence for distinct mechanisms of gene regulation being employed in different lineages. The myeloid-specific DH site spans several conserved known and potential transcription factor binding sites (Fig. 2). Although this study has concentrated on the role of the GATA elements, it is noteworthy that this region also encompasses the GM170 NFAT and AP-1 elements, the highly conserved Ets-like sequence (of unknown function) just upstream of the GATA sites, and the highly conserved NF-B and NFAT-like elements at positions 220 and 240.
Differential regulation of nucleosome positioning and chromatin remodeling across the GM-CSF enhancer in mast cells and T cells. To investigate the impact of distinct tissue-specific mechanisms of transcription factor recruitment on chromatin remodeling in normal cells, we performed very-high-resolution DNase I and MNase mapping of the chromatin structure of the GM-CSF enhancer in GM-CSF transgenic T cells and mast cells. These analyses revealed that the in vivo association of GATA-2 with the enhancer in mast cells has profound consequences for nucleosomal organization within the enhancer and identified different inducible patterns of nucleosome reorganization in mast cells and T cells. DH sites and MNase cleavage sites were mapped in stimulated and unstimulated cells from the BglI site located 660 bp downstream of the enhancer (Fig. 5B and C), and band sizes were calculated from the mobilities of the restriction enzyme fragments. The positions of prominent MNase bands, numbered relative to the 5' BglII site of the enhancer (which is located at –3289 relative to the transcription start site), are shown to the right (Fig. 5C).
MNase and DNase I digestion of mast cell nuclei revealed complex specific cleavage patterns with nuclease-hypersensitive sites spanning nucleosome N0 and the N0/N1 linker region which were essentially absent in T cells. Additional strong nuclease-hypersensitive sites associated with the nucleosomes N1 and N2 were present in both the activated T cells and the activated mast cells. Hence, chromatin remodeling in T cells appears to be restricted to just the GM330 and GM420 regions, whereas all known regulated elements within the enhancer become remodeled in activated mast cells and form a 600-bp hypersensitive region.
As the first stage of a detailed analysis of the chromatin structure data presented here, we remapped the positions of nucleosomes in both mast cells and T cells. Because MNase exhibits pronounced sequence specificity, we further refined these analyses by employing densitometry and digitally calculated the relative sensitivity of MNase cleavage within each lane in Fig. 5C. We calculated the ratios of MNase digestion of chromatin samples compared to a genomic DNA digest and then replotted both the relative MNase sensitivity and the absolute DNase I sensitivity as a function of size. The ratios determined from the BglI analyses of MNase digests of stimulated and unstimulated mast cells and T cells (Fig. 5C) are presented as the lower four graphs in Fig. 5D. In unstimulated T cells, the analyses of relative MNase cleavage revealed a strikingly regular specific positioning and 200-bp periodicity of nucleosomes across the enhancer and extending into the flanking sequences (Fig. 5D). The predicted nucleosome positions are displayed underneath the traces in Fig. 5D and differ only marginally from our previous less-refined analysis of nucleosome positions (21). From these maps (and from the additional analyses presented below in Fig. 6), we can define approximate positions (shown in parentheses) in T cells for nucleosomes N-1 (–230 to –70), N0 (–40 to 120), N1 (180 to 340), N2 (380 to 540), N3 (600 to 760), and N4 (800 to 960). These data confirm that the GM330 and GM420 NFAT/AP-1 elements lie within nucleosomes N1 and N2, respectively, and that the GATA elements must lie within a broad linker region separating nucleosomes N0 and N1.
In mast cells, the nucleosome N0 region and its 3' linker adopted a structure distinctly different from that seen in T cells. Nucleosome N0 was essentially excluded from its previously determined location and was either evicted or shifted to a new position approximately 140 bp upstream (approximately –180 to –20) and replaced by an extensive MNase-hypersensitive (MH) region spanning the GATA elements, with a short protected region at position 160 where the GATA sites reside (Fig. 5C and D). Downstream of the GATA region, nucleosomes also existed in slightly different positions from those seen in T cells, being located 30 bp further downstream (N1, 210 to 370; N2, 410 to 570; N3, 630 to 790; and N4, 820 to 980). These findings were confirmed by subsequent analyses (to be presented in full below in Fig. 6) mapping enhancer nucleosome positions from an upstream PshAI site, which are summarized in Fig. 5E. These studies demonstrate that the same nucleosome positions are predicted regardless of whether positions were mapped from upstream or downstream (Fig. 5D and E). Significantly, these observations reveal that transcription factors can play a major role in defining nucleosome positions within an array of nucleosomes, even before an enhancer has become activated.
In both mast cells and T cells, enhancer activation was associated with the appearance of MH sites across the GM330 and GM420 regions and extensive disruption of the organization of the regular nucleosome array in the sequences flanking the enhancer (Fig. 5D). In addition, prominent inducible MH sites flanking the GATA region were seen in activated mast cells but not in T cells. Densitometric traces of DNase I cleavage within the DH sites in activated mast cells and T cells are also displayed above the MNase traces in Fig. 5D. It is striking that the pattern of DH sites closely mirrors the pattern of MH sites, and the overall pattern suggests that nucleosomes are either absent or extensively disrupted in all of the regions where factors are predicted to bind. Hence, for both MNase and DNase I digests, distinct protected regions are seen that correspond to the GM330, GM420, and GM550 NFAT/AP-1 elements in activated mast cells and T cells and at the conserved GATA region specifically in activated mast cells. Also just visible in the traces of the activated mast cell MNase and DNase I analyses is a weakly protected region at position 230, which may represent the predicted NF-B or NFAT sites present within conserved sequence C (Fig. 2C). The calculated positions of the protected regions are shown at the very top of Fig. 5D, and these precisely match the positions of all of the previously defined and predicted clusters of transcription factor binding sites within the enhancer. Furthermore, the spacing between nuclease-hypersensitive sites within the enhancer N0/N1/N2 region in activated mast cells is only about 100 bp throughout this segment, which is incompatible with the coexistence of conventional phased nucleosomes within this 600-bp region. These observations suggest that the enhancer functions as distinct modules of mini-enhanceosomes, which essentially replace the nucleosomes within this region.
Nucleosomes are specifically positioned across at least 5 kb of the GM-CSF locus and are reorganized in activated cells.
To confirm the above observations, DNase I and MNase cleavage sites were mapped in the same samples used in Fig. 5C from a PshA I site located 1.5 kb upstream of the enhancer (Fig. 6A and C). To further extend this analysis, the filter containing these samples was subsequently reprobed from a PshA I site 1.8 kb downstream of the transcription start site so as to examine sequences upstream of the enhancer and spanning the promoter (Fig. 6B and C). The analyses of relative MNase cleavage using the upstream probe revealed that nucleosomes were strictly phased and regularly positioned at 200-bp intervals both within the enhancer and for at least 1 kb upstream of the enhancer in both T cells and mast cells (Fig. 6A and C). When combined with the analyses using the downstream probe, these data suggested that this regular 200-bp repeating pattern continued for at least 4 kb downstream of the enhancer in mast cells, indicating that at least 5 to 6 kb of the GM-CSF locus was occupied by a regular array of sequence-positioned nucleosomes. However, in T cells, this regular pattern of specific nucleosome positioning was less well defined than in mast cells for the regions just downstream of the enhancer and within the GM-CSF gene. There was also an indication that two alternate phases of nucleosome positions may overlap across a short transition region from approximately +600 to +900, where an additional series of MH bands (marked by asterisks in Fig. 6C) was present. Note that a 500-bp region midway between the two probes (–2000 to –1500) cannot be accurately mapped here due to the compression of bands within the upper part of each panel.
A strikingly different pattern of nucleosome organization was observed in activated T cells and mast cells. Here, the regular phasing of nucleosomes was largely randomized for both the sequences directly flanking the enhancer (Fig. 5D and 6C) and for much of the region between the enhancer and the promoter, but not for the far upstream sequences (Fig. 6C). Interestingly, nucleosomes flanking the promoter were maintained in strict positions after activation, and a region spanning the transcription start became strongly DNase I and MNase hypersensitive (Fig. 6C). The predicted nucleosomal linker region at positions 200 to 250 was also hypersensitive in activated mast cells, with nucleosome-size protected regions remaining on either side. This suggests that two specifically positioned nucleosomes may remain between positions 0 and 400, but with the linkers more accessible in activated mast cells. Previous reports indicate that the nucleosome occupying the promoter is excluded in activated T cells (19, 21), and the presence of the broad peak of nuclease-sensitive peaks seen across the promoter here (Fig. 6C) is consistent with these findings.
The analyses of DNase I and MNase cleavage sites within the enhancer mapped from the upstream PshA I site (Fig. 6B and C) confirmed the observations presented in Fig. 5C. These studies similarly (i) revealed that nucleosomes occupy different positions within the enhancer in nonactivated T cells and mast cells (also summarized in Fig. 5E) and (ii) revealed the presence of DH and MH sites at 100-bp intervals in activated cells, suggesting that nucleosomes are replaced by enhanceosome-like complexes. Once again, the DNase I and relative MNase cleavage patterns in activated cells were essentially identical and encompassed protected regions at the GM330 and GM420 elements in T cells and at both these sites and the GATA elements in mast cells (Fig. 6C). The analyses of nonactivated mast cells revealed the loss of nucleosome N0 and the clear presence of a short protected region corresponding to the GATA elements (marked by an arrow labeled G) (Fig. 5E and 6A and C). In this analysis, it is also evident that nucleosomes N1, N2, and N3 are maintained as specifically positioned nucleosomes within the enhancer in nonactivated mast cells but the linker regions flanking these nucleosomes are all hypersensitive to MNase relative to the flanking nucleosomes (Fig. 6C). These analyses also reveal a weak DH site present at –4 kb, corresponding to a myeloid-specific DH site (11).
Enhancer-dependent nucleosome mobilization in the GM-CSF locus. The above analyses suggested that nucleosomes were disrupted within the GM-CSF enhancer and disorganized across the sequences between the enhancer and promoter in both activated mast cells and T cells. To study nucleosome destabilization and mobilization more directly, we have employed internal nucleosome-length probes to determine whether nucleosomes are still present at specific locations and whether the surrounding nucleosome array is reorganized upon activation. Analyses of MNase-digested activated and nonactivated mast cells and T cells are shown in Fig. 7. Two different MNase concentrations were used for each cell type, and samples were assayed (i) using three nucleosome-length probes corresponding to the T-cell positions for nucleosome N0, nucleosome N2, and a region midway between the enhancer and promoter (–1.3 kb); and (ii) with a 0.85-kb probe at the far 3' end of the transgene (+4.0 to +4.8). The average nucleosome spacing in each lane was determined from the average band intervals by densitometry, using the DNA size marker as a reference, and is shown at the bottom of each lane (Fig. 7A to E). The average of each pair of lanes is presented in the summary in Fig. 7F. However, in these analyses, it is probably most useful to compare the most heavily digested mast cell samples with the equivalent least digested T-cell samples, because these show similar levels of cleavage of bulk DNA.
|
190 to 200 bp. In unstimulated mast cells, the locus also existed as a regular array but the repeat length was reduced to about 166 to 168 bp in the vicinity of the enhancer and was 10 bp shorter than in T cells in other regions. There was also significant loss of the enhancer N0 mononucleosome band, but not N2, in unstimulated mast cells but not in T cells (compare Fig. 7B and C). Nucleosomes N0 and N2 were significantly disrupted in both activated T cells and mast cells. Furthermore, in activated T cells, nucleosomes were mobilized and randomized both across the enhancer and within the –1.3-kb region between the enhancer and the GM-CSF promoter. With the N0, N2 and –1.3-kb probes, the sharp banding pattern was lost in activated T cells and the faint remaining oligonucleosome bands had an average repeat length of just 150 to 165 bp, which is inconsistent with predicted models of a conventional 30-nm chromatin fiber (Fig. 7B to D).
Significantly, we also obtained evidence that the process of long-range nucleosome mobilization was enhancer dependent because no change in the chromatin structure of the –1.3-kb region was seen in activated T cells prepared from transgenic line F140, which lacks the GM-CSF enhancer (Fig. 7G). This indicates that the nucleosome mobilization seen upstream of the GM-CSF gene is a process that spreads from the enhancer, not the promoter.
In activated mast cells, disruption of N0 and N2 was also accompanied by randomization of the nucleosome array in this region and a decrease in the apparent nucleosome repeat length from 167 to 147 bp (Fig. 7B and C). Nucleosome mobilization across the GM-CSF locus was more extensive in mast cells than in T cells, with reduced nucleosome repeat lengths being detected at both the –1.3-kb region, and at the far 3' end of the transgene (Fig. 7D and E). However, there was less randomization of nucleosome positions across the –1.3-kb region in activated mast cells than in T cells (Fig. 7D). In this instance, the banding patterns remained more sharply defined in mast cells, even though the repeat length in this region had decreased from 188 to 156 bp upon stimulation. This suggests that nucleosomes are uniformly closely packed at 156-bp intervals in this region, but are not positioned at specific sequences as they are in nonstimulated mast cells (Fig. 6C). Taken together, these observations suggest that nucleosomes are excluded from the enhancer in activated mast cells and become compacted and randomized in the flanking sequences.
GM-CSF enhancer nucleosomes are acetylated in mast cells. ChIP assays revealed that histones H3 and H4 were significantly acetylated within the enhancer relative to an inactive locus in both unstimulated and stimulated mast cells (Fig. 8A). The unmodified histone H3 assays also confirmed that nucleosomes were lost from the enhancer in stimulated mast cells (Fig. 8A), meaning that the relative level of histone acetylation was significantly elevated in stimulated mast cells (Fig. 8B). We found that these elevated levels of H3 and H4 acetylation extended from at least 7 kb upstream to at least 4.5 kb downstream of the GM-CSF gene, thereby encompassing the entire region of the locus where we observed nucleosome mobilization in mast cells (data not shown).
|
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
As summarized in the model in Fig. 9A, two modules of the GM-CSF enhancer encompassing the GM330 and GM420 NFAT/AP-1 elements normally exist as nucleosomes N1 and N2. These modules can be defined as the GM330 element plus the adjacent Sp1 site on nucleosome N1 and the GM420 element plus the adjacent Runx1 site on nucleosome N2. In T cells, these regulatory elements are the only sites required for function and they also represent the only sites where we have detected in vivo footprints in activated T cells (21). In the previous study, NFAT was implicated as a factor that promoted nucleosome destabilization. Significantly, the earlier study found that an array of NFAT sites was sufficient to direct formation of an inducible DH site in Jurkat cells but was insufficient to activate transcription in the absence of AP-1 elements. Hence, at the GM-CSF enhancer, cooperation between NFAT and AP-1 appears to be obligatory in T cells, with NFAT directing a chromatin remodeling activity and AP-1 being recruited by NFAT and required for transcriptional activation. This model also accounts for the lack of in vivo footprints for the constitutively expressed factors Sp1 and Runx1 in unstimulated T cells, because these elements also reside within nucleosomes and thus depend on NFAT-dependent mechanisms for accessibility within chromatin.
|
2,500 versus 108 kDa), and the length of DNA that is released by the eviction of two nucleosomes is very broad in the context of the folded nucleosomes. The NFAT/AP-1 complexes alone constitute 240 kDa. When viewed in this way, it begins to become apparent why DNase I HS sites are so much more accessible than folded chromatin fibers.
Regulation of the GM-CSF enhancer by GATA elements.
In contrast to the pattern seen in T cells, the GATA region of the enhancer is governed very differently at the level of chromatin structure in myeloid cells. The presence of nuclear GATA-1 and GATA-2 in unstimulated mast cells appears to be sufficient to direct a pattern of nucleosome positioning very different from that seen in T cells. Unlike other elements in the enhancer, it is unlikely that GATA factors require significant chromatin remodeling to access the GATA elements because they lie within a broad 60- to 75-bp nucleosome linker region. Furthermore, our previous restriction accessibility assays in T cells determined that this region is accessible to BamHI even in nonstimulated cells, whereas the ApaI and HincII sites just downstream are only accessible in activated cells (21). Our ChIP assays confirmed that GATA-2 binds to the enhancer in mast cells prior to activation. This event is likely to be responsible for promoting the acetylation of histones and the creation of an extensive nucleosome-free region that spans the normal position of nucleosome N0 and the entire conserved region stretching from the Ets-like element to the GM170 AP-1 element. This mode of regulation by GATA factors would also facilitate enhancer activation in myeloid cells lacking NFAT (such as K562 and HEL cells) where the enhancer functions efficiently in the absence of NFAT. As modeled in Fig. 9, the regulation of the enhancer in K562 and HEL cells by GATA-1 and AP-1 is associated with the formation of a discrete DH site spanning the GATA sites and the GM170 and GM330 AP-1 sites. In this context, activation is likely to lead to recruitment of AP-1 to the high-affinity GM170 and GM330 AP-1 elements and Sp1 to the GM300 element. We also have evidence that the GATA sites and the closely linked GM170 AP-1 site function as a discrete module because the GATA and GM170 sites are both needed for function in HMC-1 cells, whereas the GM330 and GM420 AP-1 sites are not (Fig. 3). It is also significant that GATA-2 and AP-1 have the capacity to directly associate with each other and are known to cooperate in the activation of the IL-13 promoter in mast cells (27). These observations, and the close proximity of the GATA and GM170 AP-1 sites, make it even more likely that this region of the GM-CSF enhancer functions as a discrete enhanceosome. GATA elements have also long been recognized as sites of tissue-specific chromatin remodeling, which is thought to be directed via SNF2H/ACF1 remodeling complexes recruited by GATA factors (32). This type of complex could also account for the repositioned nucleosomes that we observe in the GM-CSF enhancer because SNF2H is known to mediate nucleosome sliding (15).
Mechanisms of tissue-specific and high-level cytokine gene regulation.
Our study also adds a new twist to the accepted models of the differential regulation of tissue-specific cytokine genes. In differentiated T cells, many cytokine genes are expressed in a mutually exclusive fashion by either Th2 or Th1 T-cell-specific mechanisms, whereas GM-CSF is expressed more widely. Th2-type cytokine genes, which include the IL-4, IL-5, and IL-13 genes, are all activated via NFAT and GATA-3 in Th2 T cells (13), and these genes are also coexpressed in mast cells. This makes it likely that the GATA and NFAT elements of the GM-CSF enhancer will also cooperate in Th2 T cells. This means that the GM-CSF enhancer may be activated by Th2-type mechanisms in Th2 cells, mast cells, and eosinophils, but by quite distinct mechanisms in other cells such as Th1 T cells and nonhemopoietic cells. It is also apparent that the recruitment of GATA factors leads to very high levels of GM-CSF locus activation. Mast cells which express GATA, NFAT, and AP-1 factors expressed 104-fold-higher levels of GM-CSF than CD11b-positive splenocytes (Fig. 1C) and fetal liver-derived macrophages (E. W. Baxter and P. N. Cockerill, unpublished data), which both express NFAT and AP-1 but not GATA factors. The enhancer was also considerably more active in transfection assays of myeloid cells expressing GATA factors than in Jurkat T cells, which do not (Fig. 4). Furthermore, it is likely that the IL-4-dependent activation of GATA-3 gene expression will synergize with T-cell receptor signaling pathways to drive higher levels of GM-CSF expression in Th2 T cells than in undifferentiated T cells.
Enhancer-dependent nucleosome mobilization in the GM-CSF locus. Surprisingly few previous studies have looked at long-range nucleosome mobilization or found alterations in nucleosome repeat length at active gene loci in mammalian cells. We have made the highly significant observation that activation of an enhancer can lead to dramatic changes in the structure of the whole chromatin fiber across several kilobases of DNA. In both T cells and mast cells, activation of the enhancer led to a reduction in local nucleosome repeat length from a normal value of about 190 bp in nonactivated T cells to about 153 bp in activated T cells and 146 bp in activated mast cells. This reduced value is essentially the same as the size of the 146-bp nucleosome core particle and is significantly shorter than the 167 bp required to assemble a chromatosome comprising the core plus histone H1. With essentially no significant nucleosomal linkers, it is difficult to see how this structure could form any kind of folded chromatin fiber that might resemble the 30-nm fiber.
Chromatin reorganization in the active GM-CSF locus was not restricted to the immediate vicinity of the enhancer, but continued more than a kilobase into the –1.3-kb region, where the nucleosome repeat lengths were 165 in activated T cells and 156 in activated mast cells. In activated mast cells, inducible histone acetylation and nucleosome reorganization also extended at least 2 kb past the 3' end of the GM-CSF gene where nucleosome repeat lengths were still about 20 bp less than those in nonactivated cells. Significantly, we also demonstrated that nucleosome mobilization upstream of the GM-CSF promoter was dependent upon the enhancer, suggesting that this may represent a significant component of the mechanism of enhancer function.
Nucleosome mobilization leads to randomization. When attempting to interpret our data, it is important to acknowledge that the predominant pattern of nucleosome mobilization in the GM-CSF locus (Fig. 6 and 7) is actually randomization of nucleosome positioning rather than simply shortening of nucleosome repeat lengths. In activated T cells, the discrete nucleosomal fragments seen after MNase digestion of mobilized regions (Fig. 7) do in fact only represent a small proportion of the total products, which mostly appear as a continuous spectrum of sizes. The small proportion of discrete products emanating from the activated GM-CSF locus most likely arise by a stochastic process whereby a subset of nucleosomes will inevitably lie in closely packed positions within an essentially random organization. In the case of the –1.3-kb region in activated mast cells, it would appear that a higher proportion of nucleosomes become more tightly packed and generate a more uniform nucleosome ladder (Fig. 7D), without adopting new specific positions (Fig. 6C). We have attempted to represent models of the regular array and the mobilized array in Fig. 9B.
Other examples of nucleosome mobilization at active genes. There are currently very few known examples of any other gene loci where gene activation is accompanied by a reduction in apparent nucleosome repeat length. One of the only mammalian examples is the active mouse kappa Ig gene locus, which has a calculated repeat length of 155 bp in B cells compared to 175 bp for an inactive globin gene (36). Other nonmammalian examples are the Chlamydomonas HSP70A and RBCS2 loci, with repeat lengths of 160 and 156 bp compared to 178 bp in bulk chromatin (26), and an inducible yeast GAL1 promoter construct which undergoes a repeat length decrease from 180 to 160 bp upon activation (6). The simplest explanation for the observed compaction of nucleosomes in the active GM-CSF locus is the exclusion of nucleosomes from the enhancer, accompanied by the loss of histone H1 from the adjacent sequences. Histone H1 depletion in mice is known to decrease the global nucleosome repeat length (16). Furthermore, there is a precedent for gene locus activation leading to a reduction of H1 levels, not only at cis-regulatory elements, but also within flanking regions (25).
There are also other similar previous examples of active gene loci undergoing extensive randomization of nucleosome organization (6, 12, 22, 23, 36), although in one such study of the ß-globin locus (22), there was no parallel decrease in nucleosome repeat length. In one of the yeast studies mentioned above, nucleosome mobilization was dependent upon SWI/SNF (23), and we and others have previously suggested that a Brg1-dependent SWI/SNF-like mechanism may underlie nucleosome disruption and mobilization at the GM-CSF locus (9, 19, 21). Another mechanism likely to be involved in nucleosome mobilization is the acetylation of histone H4 at lysine 16, which is known to be sufficient to induce unfolding of the 30-nm chromatin fiber (33). It is therefore also significant that we detect acetylation of histones H3 and H4 at the GM-CSF enhancer. We anticipate that long-range nucleosome reorganization occurs more commonly in active genes than is currently recognized and represents a fundamental aspect of gene activation that is likely to involve a variety of chromatin modifications acting together to create an active chromatin structure.
|
ACKNOWLEDGMENTS |
|---|
This work was supported by Yorkshire Cancer Research, the AICR, and the NH&MRC.
|
FOOTNOTES |
|---|
X174 marker used to determine sizes of oligonucleosome fragments. (B to E) Hybridization patterns obtained with the indicated probes. The left-hand lanes in each panel represent genomic DNA digested with PstI, ApaI, and BglII to produce fragments used to confirm appropriate hybridization specificity and validate quantification of size estimations. (F) Map of the GM-CSF transgene showing the locations of hybridization probes. Summarized below each probe are the average nucleosome repeat lengths determined from densitometry of the images in panels A to F. (G) MNase analyses of M268 and F140 transgenic T cells performed as described above, using 150 U/ml MNase for 15 min, and the –1.3-kb probe. F140 contains a transgene lacking the GM-CSF enhancer, as in Fig. 
