Granulocyte-macrophage colony-stimulating factor enhancer activation requires cooperation between NFAT and AP-1 elements and is associated with extensive nucleosome reorganization

Abstract

The human granulocyte-macrophage colony-stimulating factor (GM-CSF) gene is activated by an NFAT-dependent enhancer forming an inducible DNase I hypersensitive (DH) site. The enhancer core comprising the DH site contains the GM330 and GM420 elements that bind NFAT and AP-1 cooperatively. Here we demonstrate that both elements are essential for enhancer activity and that Sp1 and AML1 sites in the enhancer become occupied in vivo only after activation. Chromatin structure analysis revealed that the GM-CSF enhancer core elements are divided between two adjacent nucleosomes that become destabilized and highly accessible after activation. Inducible chromatin reorganization was not restricted to the enhancer core but extended across a 3-kb domain of mobilized nucleosomes, within which the nucleosome repeat length was compressed from approximately 185 to 150 bp. The GM420 element is a high-affinity site that binds NFAT independently of AP-1 but depends on the linked AP-1 site for enhancer function. Nevertheless, just the NFAT motif from the GM420 element was sufficient to form a DH site within chromatin even in the absence of the AP-1 site. Hence, NFAT has the potential to cooperate with other transcription factors by promoting chromatin remodelling and increasing accessibility at inducible regulatory elements.

FIG. 1.

Map of the human GM-CSF enhancer. The locations of regulatory elements and the DH site in the GM-CSF enhancer are shown. Sites are labeled according to their positions in the 717-bp BglII fragment that defines the enhancer. DH sites (vertical arrows) and the probe used to map DH sites (box) are indicated. Composite NFAT/AP-1 sites are depicted below the map. HS, hypersensitive.

FIG. 2.

In vivo footprinting of the GM-CSF enhancer in human T cells. Cultured human peripheral blood T cells were either left unstimulated (−) or stimulated for 4 h with 20 ng of PMA per ml and 2 μM A23187 (+). Cells were then incubated with DMS, and sites of DMS-mediated DNA methylation for both DNA strands were identified by LM-PCR. DMS-methylated purified genomic DNA (G) was also included to indicate DMS reactivity towards protein-free DNA. The GM330 region of the enhancer was assayed using the 2R primer set for the upper strand and the 2F primer set for the lower strand. The GM420 region was assayed using the 2R primer set for the upper strand and the 3F primer set for the lower strand. Note that no data are displayed for the GC-rich region of the upper strand just upstream of position 298, because the LP25 linker primer was not able to amplify DNA modified within this section. Bases protected from methylation by DMS (open circles) and bases with enhanced DMS reactivity (closed circles) are indicated. Previously identified transcription factor binding sites are shown in bold type. A novel potential AML1 site is underlined. Bases are numbered relative to the upstream BglII site.

FIG. 3.

Fine structure of the DH site within the GM-CSF enhancer. Human T cells were either left unstimulated (−) or stimulated for 5 h with 20 ng of PMA per ml and 2 μM A23187 (+). Nuclei were isolated and briefly digested with 15 (−) or 12 (+) μg of DNase I per ml. These samples were selected for LM-PCR analysis from a DNase I titration series, because they had equivalent levels of overall DNase I digestion and were optimal for the detection of DH sites in other assays. Cleavage sites were identified by LM-PCR using the 1R primer set for the top strand and the 1F primer set for the lower strand. The positions of transcription factor binding were identified by aligning the DNase I pattern with the positions of G bases determined using DMS-treated DNA (G). To align the G reaction, we also included a DNA sample partially digested with BglII, BamHI, ApaI, DraI, and NcoI (R). We included a DNase I digestion product of purified DNA for a control (C). Note that the region immediately upstream of the Sp1 site on the upper strand represents a region not amplified by the LP25 primer, and therefore appears as a blank region. The positions of fragments digested by different restriction endonucleases are indicated to the left of the gels. Asterisks indicate background artifacts. Bases are numbered relative to the upstream BglII site.

FIG. 4.

Mutational analysis of the GM-CSF enhancer in transient-transfection assays. Luciferase reporter gene plasmids containing the luciferase vector pXPG and a segment of the human GM-CSF promoter covering positions −627 bp to +28 bp relative to the transcription start site and an upstream 717-bp BglII segment of the human GM-CSF enhancer were transiently transfected into Jurkat T cells and assayed for luciferase activity after stimulation for 9 h with 20 ng of PMA per ml and 1 μM calcium ionophore A23187. The assays included plasmids containing the promoter lacking the enhancer (pXPG-GM627) or containing the intact enhancer (pXPG-GM627-B717) or the enhancer carrying the specified mutations to destroy the indicated binding sites. At least two independent preparations of each plasmid were assayed, and each plasmid was assayed an average of 34 times to maximize reliability of the data. The luciferase activity of each plasmid is expressed relative to the plasmid containing the intact enhancer. Note that as reported previously (13), in the absence of stimulation the luciferase activity of each plasmid was approximately 1% of the activity obtained after stimulation, and these data are not presented here. Error bars represent the standard errors of the means. HS, hypersensitive. White bars indicate the activity of the promster alone, black bars indicate the additional activity supported by the enhancer.

FIG. 5.

Kinetics of GM-CSF gene activation. (A) GM-CSF mRNA expression in T cells stimulated for the indicated times with 20 ng of PMA per ml and 2 μM A23187 (PMA/I). RNA expression was assayed by real-time PCR and expressed relative to GAPDH expression where the value for unstimulated cells was set at 1. (B) Assay of DH sites upstream of the GM-CSF gene in T cells from the same cultures assayed in panel A. For each preparation of T-cell nuclei, aliquots of nuclei were digested with DNase I in the range of 4.5 to 8 μg of DNase I per ml and only the optimally digested samples are displayed. DH sites were mapped by indirect end labeling of EcoRI-digested DNA using an upstream 1.4-kb BamHI fragment of DNA as a probe as indicated in Fig. 1. The positions of the promoter (P) and GM-CSF enhancer (E) are indicated to the right of the gel. (C) EMSA with oligonucleotides carrying binding sites for the indicated transcription factors (see Materials and Methods) using nuclear extracts prepared from the same cultures shown in panels A and B.

FIG. 6.

Cooperative transcription factor binding and function of the GM420 NFAT/AP-1 site. (A) EMSA with recombinant NFAT, AP-1, and GM420 duplex oligonucleotides with or without the indicated mutations. (B) Transient-transfection assays of pXPG-GM55 luciferase reporter gene plasmids containing the minimal GM-CSF promoter and tandem arrays of three copies of the GM420 elements with and without the indicated mutations. Transfected Jurkat T cells were stimulated for 9 h with 20 ng of PMA per ml and 1 μM calcium ionophore A23187. Luciferase activities are shown as the means ± standard errors of the means (error bars) for 12 transfection assays, employing two separate preparations of each plasmid construct, and are expressed relative to the luciferase activity of pXPG-GM55, which was set at 1.

FIG. 7.

Identification of DH sites encompassing NFAT elements in stably transfected cells. DH sites were assayed in single clones of Jurkat T cells transfected with pHGM0.6 GM-CSF promoter plus chloramphenicol acetyltransferase (CAT) reporter gene plasmids (14) containing either the intact GM-CSF enhancer or three copies of just the NFAT-binding region of the GM420 NFAT site. Cells were either left unstimulated (−) or stimulated for 6 h with 20 ng of PMA per ml and 2 μM A23187 (PMA/I) (+). DH sites were identified by indirect end labeling using a 0.7-kb SacI-ScaI fragment of pHGM0.6 as a probe.

FIG. 8.

Chromatin architecture of the GM-CSF enhancer. (A to E) Indirect end-labeling assays of MNase and DNase I sites within the GM-CSF enhancer in nuclei prepared from cells that were either left unstimulated (nil) or stimulated for 4 h with 20 ng of PMA per ml and 2 μM A23187 (PMA/I). Analyses were performed by Southern blot hybridization using either 10 μg of human T-cell DNA per lane (A to C) or 1 μg of Gm-CSF transgenic mouse T-cell DNA per lane (D and E). The positions of nucleosomes N1 and N2 and the DH site (DHS) are indicated by brackets. Band positions are calculated relative to the upstream BglII site at position 1 in the 717-bp BglII enhancer sequence. Asterisks indicate the positions of new inducible MNase sites at positions 260 and 470. Panel A depicts the ethidium bromide staining pattern for the analyses performed in panels B and C and the positions of bands containing two to six nucleosomes (indicated to the left of the gel). The restriction enzyme marker in panels B and C consists of DNA partially digested with BglII, PstI, and ApaI, with the positions of these sites depicted on the right. (F) Map of the major nuclease cleavage sites within the 1,660-bp BglI fragment encompassing the GM-CSF enhancer, showing the predicted positions of nucleosomes, the DH site, and the indirect end-labeling probes used to map cleavage sites. HS, hypersensitive. (G) LM-PCR mapping of nucleosome boundaries in nucleosome length DNA fragments purified from human T cells that were either left nonstimulated (nil) or stimulated for 4 h with 20 ng of PMA per ml and 2 μM A23187 (PMA/I). The leftmost panel depicts agarose gel electrophoresis of 146-bp nucleosome-length DNA (N) and 167-bp chromatosome length DNA (C) purified from MNase digestion products of T-cell nuclei. Twenty-five nanograms of purified nucleosomal DNA was assayed by LM-PCR using the 2R2 and 4F primer sets. DMS-treated DNA samples to align the sequence (G) and a control MNase digest of purified DNA (M) were also included. The map shows the relative positions of the primer sets used to map nucleosome boundaries.

FIG. 9.

Nucleosome destabilization and mobilization in the GM-CSF locus. Southern blot hybridization analysis of DNA purified from MNase-digested nuclei isolated from GM-CSF transgenic mouse T cells that were either left unstimulated (nil) or stimulated for 4 h with 20 ng of PMA per ml and 2 μM A23187 (PMA/I). The three sets of MNase digestion conditions employed were, from left to right, 250 U of MNase/ml for 3 min, 150 U of MNase/ml for 15 min, and 500 U of MNase/ml for 15 min. The latter was chosen for performing densitometry of the hybridization patterns, as depicted below each panel, with nonstimulated T-cell samples in grey and stimulated T-cell samples in black, and the average nucleosome repeat length was calculated for nonstimulated (−) and stimulated (+) cells. DNA was analyzed on a total of four replica 2% agarose gels, using 3 μg of DNA per lane. HaeIII-digested φX174 DNA was employed as a size marker in the rightmost lanes, and as can still be seen in panel H, the positions of the marker and the major 146-bp nucleosome bands were marked on the gel and filter with a needle and Indian ink to enable accurate size estimations. The leftmost lanes contain 3 μg of DNA digested with BamHI, HincII, and PstI. Panels B, C, and D were reprobed, and the results are shown in panels E, G, and H, respectively. The map depicts the 10.5-kb XhoI-HindIII fragment used as a transgene and the positions of the GM-CSF probes used in panels B to G. Probe B extended from positions 219 to 390, and probe C extended from positions 386 to 559 within the GM-CSF enhancer. Pr., promoter. Panel H was probed with a 17-kb EcoRI DNA fragment spanning the Cɛ region of the mouse heavy-chain immunoglobulin (IgH) gene, which is inactive in T cells.

FIG. 10.

Restriction enzyme accessibility assays of the GM-CSF enhancer. Assays were performed on nuclei isolated from cultured human T cells that were either left unstimulated (−) or stimulated for 4 h (A and B) or 6 h (E) with 20 ng of PMA per ml and 2 μM A23187 (PMA/I) (+). (A and B) Nuclei were digested with the restriction enzymes indicated above each lane. Purified DNA was then redigested with either BglII (the ApaI, HincII, NcoI, BamHI, and PstI lanes and the first G lane) or BglI (the BglII lanes and the rightmost G lane) prior to electrophoresis on a 2% agarose gel and Southern blot hybridization analysis using either the 717-bp BglII fragment of the enhancer (A) or a BglII-BamHI fragment downstream of the enhancer (B) as probes. ApaI, HincII, PstI, and BamHI each cut the enhancer just once. Because NcoI and BglII cut the enhancer region two or three times, the migration positions of complete and partial digestion products are shown on the righthand side of panel A and on each side of panel B, and a map of these digestion products is provided in panel D. The size markers to the left of the gel in panel A indicate the migration positions of selected bands of HaeIII-digested φX174 DNA. (C) Locations of nuclease cleavage sites in the GM-CSF enhancer, showing the total percentage of DNA cut at each site before (nil) and after stimulation with 20 ng of PMA per ml and 2 μM A23187 (PMA/I). (D) Summary of the products detected by probes A and B after digestion with either BglII followed by BglI or NcoI followed by BglII. The positions of sites cleaved by the enzymes used in the accessibility assays are indicated by asterisks. (E) Nuclei from nonstimulated (0 lanes) and stimulated cells (P/I lanes) were digested with HaeIII, and cleavage at the indicated sites or positions was assayed by LM-PCR. The fold increase in cleavage products detected is shown below each panel. The map below shows the locations of HaeIII sites in the enhancer and the positions of LM-PCR primers used to detect cleavage at these sites.