The beta-globin dominant control region: hypersensitive site 2.

The Dominant Control Region (DCR) of the human beta-globin gene locus consists of four strong hypersensitive sites (HSS) upstream of the epsilon-globin gene. Addition of these sites confers copy number dependent expression on the human beta-globin gene in murine erythroleukaemia cells and transgenic mice, at levels comparable with the endogenous mouse globin genes. We have shown previously that a 1.9 kb fragment comprising HSS 2 accounts for 40-50% of the full effect of the DCR. In this paper we describe a deletional analysis of HSS 2. We show that a 225 bp fragment is sufficient to direct high levels of expression of the human beta-globin gene which is copy number dependent and integration site independent. This 225 bp fragment overlaps the major region that is hypersensitive 'in vivo'. DNase I footprinting shows the presence of four binding sites for the erythroid specific protein NF-E1; the three other footprinted regions display a remarkable redundancy of the sequence GGTGG and bind a number of proteins including Sp1 and the CACC box protein. The significance of these results for the regulation of globin gene expression is discussed.     

Structural analysis and mapping of DNase I hypersensitivity of HS5 of the beta-globin locus control region.

The beta-globin locus control region (LCR) is a cis regulatory element that is located in the 5' part of the locus and confers high-level erythroid lineage-specific and position-independent expression of the globin genes. The LCR is composed of five DNase I hypersensitive sites (HSs), four of which are formed in erythroid cells. The function of the 5'-most site, HS5, remains unknown. To gain insights into its function, mouse HS5 was cloned and sequenced. Comparison of the HS5 sequences of mouse, human, and galago revealed two extensively conserved regions, designated HS5A and HS5B. DNase I hypersensitivity mapping revealed that two hypersensitive sites are located within the HS5A region (designated HS5A(major) and HS5A(minor)), and two are located within the HS5B region (HS5B(major), HS5B(minor)). The positions of each of these HSs colocalize with either GATA-1 or Ap1/NF-E2 motifs, suggesting that these protein binding sites are implicated in the formation of HS5. Gel retardation assays indicated that the Ap1/NF-E2 motifs identified in murine HS5A and HS5B interact with NF-E2 or similar proteins. Studies of primary murine cells showed that HS5 is formed in all hemopoietic tissues tested (fetal liver, adult thymus, and spleen), indicating that this HS is not erythroid lineage specific. HS5 was detected in murine brain but not in murine kidney or adult liver, suggesting that this site is not ubiquitous. The presence of GATA-1 and NF-E2 motifs (which are common features of the DNase I hypersensitive sites of the LCR) suggests that the HS5 is organized in a manner similar to that of the other HSs. Taken together, our results suggest that HS5 is an inherent component of the beta-globin locus control region.     

Inducibility of the HS II enhancer depends on binding of an erythroid specific nuclear protein.

An erythroid specific, inducible enhancer associated with hypersensitive site II (HS II) plays a central role in the function of the human beta globin dominant control region. The HS II enhancer consists of tandem AP-1 binding sites and has been shown to bind members of the ubiquitous jun and fos families of proteins. The same sites are now shown to bind the erythroid specific protein, NF-E2. Inducibility of the HS II enhancer depends on NF-E2 binding, even in the presence of another hypersensitive site. Further, increased activity of the enhancer in induced K562 cells correlates with the presence of NF-E2, which appears to be present in a modified form. NF-E2 is distinct from some enhancer binding proteins in K562 nuclear extracts, in that it does not contain Fos or Fra-1 protein. Thus, binding by NF-E2 may be the mechanism, whereby tandem AP-1 binding sites confer erythroid specificity on the HS II enhancer.     

Functional and binding studies of HS3.2 of the beta-globin locus control region

The distal locus control region (LCR) is required for high-level expression of the complex of genes (HBBC) encoding the beta-like globins of mammals in erythroid cells. Several major DNase hypersensitive sites (HSs 1-5) mark the LCR. Sequence conservation and direct experimental evidence have implicated sequences within and between the HS cores in function of the LCR. In this report we confirm the mapping of a minor HS between HS3 and HS4, called HS3.2, and show that sequences including it increase the number of random integration sites at which a drug resistance gene is expressed. We also show that nuclear proteins including GATA1 and Oct1 bind specifically to sequences within HS3.2. However, the protein Pbx1, whose binding site is the best match to one highly conserved sequence, does not bind strongly. GATA1 and Oct1 also bind in the HS cores of the LCR and to promoters in HBBC. Their binding to this minor HS suggests that they may be used in assembly of a large complex containing multiple regulatory sequences.     

Characterization of a DNA binding activity in DNAse I hypersensitive site 4 of the human globin locus control region.

A portion of the beta-globin Locus Control Region (LCR), which included DNAse I hypersensitive site 4 (HS4), was analyzed for its interactions with nuclear extracts and its contribution to LCR activity in a functional assay. In gel retardation assays, a short fragment from HS4 formed complexes with nuclear extracts from both erythroid and nonerythroid cells, and a core protected sequence 5'GACTGGC3' was revealed by DNAse I protection and methylation interference studies. This sequence resembles the binding sites of CCAAT-family members. Purified CP-2 but not CP-1 was shown to bind this HS4 sequence in a gel shift reaction, suggesting that the HS4 binding activity shares some sequence specificity with the CCAAT-factor family. Utilizing a transient expression assay in murine erythroleukemia cells, steady-state RNA levels were measured from pairs of LCR constructs linked to distinguishable beta-globin reporter genes. A short DNA fragment from HS4 which included the binding site for this novel binding activity accounted for most of the contribution to high level expression made by the entire HS4 region.     

Chromatin structure at the flanking regions of the human beta-globin locus control region DNase I hypersensitive site-2: proposed nucleosome positioning by DNA-binding proteins including GATA-1

The human beta-globin locus control region DNase I hypersensitive site-2 (LCR HS-2) is erythroid-specific and is located 10.9 kb upstream of the epsilon-globin gene. Most studies have only examined the core region of HS-2. However, previous studies in this laboratory indicate that positioned nucleosomes are present at the 5'- and 3'-flanking regions of HS-2. In addition, footprints were observed that indicated the involvement of DNA-binding proteins in positioning the nucleosome cores. A consensus GATA-1 site exists in the region of the 3'-footprint. In this study, using an electrophoretic mobility shift assay (EMSA) and DNase I footprinting, we confirmed that GATA-1 binds in vitro at the 3'-end of HS-2. An additional GATA-1 site was found to bind GATA-1 in vitro at a site positioned 40 bp upstream. At the 5'-end of HS-2, DNase I footprinting revealed a series of footprints showing a marked correlation with the in vivo footprints. EMSA indicated the presence of several erythroid-specific complexes in this region including GATA-1 binding. Sequence alignment for 12 mammalian species in HS-2 confirmed that the highest conservation to be in the HS-2 core. However, a second level of conservation extends from the core to the sites of the proposed positioning proteins at the HS-2 flanking regions, before declining rapidly. This indicates the importance of the HS-2 flanking regions and supports the proposal of nucleosome positioning proteins in these regions.     

In vivo formation of a human beta-globin locus control region core element requires binding sites for multiple factors including GATA-1, NF-E2, erythroid Kruppel-like factor, and Sp1

The active elements of the beta-globin locus control region (LCR) are located within domains of unique chromatin structure. These nuclease hypersensitive sites (HSs) are characterized by high DNase I sensitivity, erythroid specificity, similar nucleosomal structure, and evolutionarily conserved clusters of cis-acting elements that are required for the formation and function of the core elements. To determine the requirements for HS core formation in the setting of nuclear chromatin, we constructed a series of artificial HS cores containing binding sites for GATA-1, NF-E2, and Sp1. In contrast to the results of previous in vitro experiments, we found that when constructs were stably integrated in mouse erythroleukemia cells the binding sites for NF-E2, GATA-1, or Sp1 alone or in any combination were unable to form core HS structures. We subsequently identified two new cis-acting elements from the LCR HS4 core that, when combined with the NF-E2, Sp1, and tandem inverted GATA elements, result in core structure formation. Both new cis-acting elements bind Sp1, and one binds erythroid Kruppel-like factor (EKLF). We conclude that in vivo beta-globin LCR HS core formation is more complex than previously thought and that several factors are required for this process to occur.     

Structure and function of the enhancer 3'' to the human A gamma globin gene.

An enhancer is located immediately 3' to the A gamma globin gene. We have used DNase I footprinting to map the sites of interaction of nuclear proteins with the DNA sequences of this enhancer. Eight footprints were discovered, distributed over 600 base pairs of DNA. Three of these contain a consensus binding site for the erythroid specific factor GATA-I. Each of these GATA-1 sites had an enhancer activity when inserted into a reporter plasmid and tested in human erythroleukemia cells. Other footprints within the enhancer contained consensus binding sequences for the ubiquitous, positive regulatory proteins AP2 and CBP-1. An Sp1-like recognition sequence was also identified. Synthetic oligonucleotides encompassing two of the footprints generated a slowly migrating complex in gel mobility shift assays. The same complex forms on a fragment of the human gamma globin gene promoter extending from -260 to -200. The DNaseI footprint of this protein complex with the enhancer overlapped a sequence, AGGAGGA, found within the binding site for a protein that interacts with the chicken beta globin promoter and enhancer, termed the stage selector element. We propose that this complex of proteins may be involved in the human gamma globin promoter-enhancer interaction.     

Nuclear factors regulating erythroid progenitor cell differentiation and gene expression

Nuclear factors are effector molecules that directly regulate proliferation and differentiation of progenitor cells. Gene targeting experiments in mice have demonstrated that several nuclear factors are essential at different stages of erythroid development including cell cycling factors (Rb), and both ubiquitous (c-myb) and erythroid-specific (GA TA-1) DNA binding factors. In addition, DNA binding factors are required to establish the DNase I hypersensitive sites in the human β-globin locus control region (LCR). By 'opening' chromatin in the β-globin locus early in development, the LCR permits correct temporal and spatial expression of the globin genes in the maturing cells of the erythroid lineage.     

Quantification of DNaseI-sensitivity by real-time PCR: quantitative analysis of DNaseI-hypersensitivity of the mouse beta-globin LCR

We employ real-time PCR to allow us to quantify the sensitivity of chromatin to digestion by DNaseI. This approach has three clear advantages over the more conventional use of the Southern hybridization assay: the accuracy of quantification is improved; the resolution of the assay is enhanced, by designing primers to amplify small amplicons it is possible to analyze sequences both co-incident and proximal to sites of DNaseI-hypersensitivity; less material is needed, as little as 5 ng of treated genomic DNA. We applied this method in an analysis of the chromatin structure of the previously described mouse beta-globin locus control region (LCR) using fetal liver cells. The four hypersensitive sites of the canonical mouse LCR, HS1 to HS4, are shown to have kinetics of digestion consistent with these sequences being nucleosome-free in vivo. A different pattern was seen for HS6, a recently described "weak" hypersensitive site. The site was also rapidly lost but more of the sites proved resistant, we interpreted this to show that this hypersensitive was only forming in a portion of the erythroid cells. This finding implies that in vivo the LCR is structurally heterogeneous. Sequences proximal to the hypersensitive sites show a third pattern of intermediate sensitivity, consistent with the chromatin being unfolded but the sites still bound by a continual nucleosomal array. Our results demonstrate that this method has the potential to achieve accurate and detailed mapping of chromatin structure from small amounts of tissue samples.