Protein-DNA interactions in vivo of an erythroid-specific, human beta-globin locus enhancer.

The 5' DNase I-hypersensitive site 2 (5' HS-2) is an erythroid-specific enhancer located 11 kilobases (kb) upstream of the human beta-globin gene cluster. Presence in cis of 5' HS-2 confers a high level of erythroid cell-specific and developmentally regulated promoter activities of human globin genes in transfected cell cultures and in transgenic mice. Combining the use of the methylation protection assay and polymerase chain reaction, we have studied nuclear factor-DNA interactions of the 5' HS-2 enhancer in vitro and in vivo. The data from analyses of three different sequence motifs within 5' HS-2 represent three different modes of protein-DNA interaction with respect to cell-type specificities and in vivo vs. in vitro differences. First, a GATA-1 motif was found to bind nuclear factor(s), presumably the GATA-1 factor, present in K-562 cell extracts and in living K-562 cells. No such binding was seen in nonerythroid HeLa cells or extract. A second motif, NF-E2/AP1 (nuclear factor-erythroid 2/activator protein 1), consists of tandemly arranged dimers of AP1 binding consensus. The presence of either HeLa extract or K-562 extract protects the NF-E2/AP1 motif from methylation, but the footprints are different. This is most likely due to different protein-DNA contacts of the AP1-DNA complex formed in HeLa extract and the NF-E2-DNA complex in K-562 extract. In vivo methylation protection patterns of this motif parallel those observed in vitro, suggesting that it is also bound by NF-E2 in K-562 cells and by AP1 in HeLa cells. Finally, a GT-I motif binds apparently to one or more similar factors in both types of nuclear extracts, but the in vivo methylation protection patterns are not identical between living HeLa and K-562 cells. These data provide direct evidence that specific nuclear factor-DNA complexes form in vivo at functionally important sequence motifs of the 5' HS-2 enhancer in erythroid cells. The detection of conformationally different nuclear factor-DNA complexes at the same sequence motifs in HeLa and Raji cell lines also raises interesting questions regarding the origin and function of these complexes in nonerythroid cells.     

Development of a condensed locus control region cassette and testing in retrovirus vectors for A gamma-globin

Retrovirus vectors for A gamma-globin are being developed for the treatment of beta chain hemoglobinopathies. Toward the goal of achieving therapeutic expression levels, core elements of the beta-globin locus control region (LCR) hypersensitive sites (HS) were screened for enhancer activity in erythroid MEL and K562 cell lines using a drug-resistant colony assay. When used alone, core elements of HS1, HS3, and HS4 showed no activity and a fragment for HS2 showed only modest activity in the colony assay. However, a 1.1 kb combination of fragments for HS2, HS3, and HS4 (termed a nLCR) enhanced colony formation 17-fold in K562 cells and 94-fold in MEL cells. Addition of an HS1 fragment enhanced nLCR activity only modestly in MEL cells. When linked to a beta-globin gene, the 1.1 kb nLCR enhanced globin mRNA expression to 82% per copy of mouse alpha-globin in transfected MEL cells. Inclusion of a nLCR in retrovirus vectors containing a beta-globin promoter and various A gamma-globin gene expression cassettes resulted in extreme genetic instability and reduced titers. Specific deletions were abrogated by removing homologous sequences, but random recombinations were still observed at significant frequencies. In MEL cells containing intact provirus, A gamma-globin mRNA produced by an optimal vector containing the nLCR was only 2-fold higher (8.5% vs. 3.9% per copy of mouse alpha-globin) compared to the same vector without the nLCR. These data suggest that vector elements detract from the ability of the nLCR to enhance expression of the beta pr.A gamma cassettes.     

Synergistic enhancement of globin gene expression by activator protein-1-like proteins.

DNA sequences corresponding to the four major DNase I hypersensitive sites upstream of the beta-globin gene cluster are essential for the achievement of high levels of globin gene expression and development regulation. In this study, we focused on one of these sites, hypersensitive site 2, which behaves as a powerful enhancer in transient expression and transgenic mouse experiments. We identified a tandem repeat of the activator protein 1 (AP-1) consensus sequence that binds AP-1-like proteins from nuclear extracts of K562 and HeLa cells. These proteins have the same binding properties as HeLa AP-1 but differ in the electrophoretic mobility and in functional assays. Transient-expression experiments in K562 of various deletion and point mutation constructs derived from hypersensitive site 2 indicate that the enhancer activity and the inducibility of a linked gamma-globin promoter are dependent upon the synergistic action of proteins bound to the tandem AP-1 repeat.     

Development of viral vectors for gene therapy of beta-chain hemoglobinopathies: optimization of a gamma-globin gene expression cassette

Progress toward gene therapy of beta-chain hemoglobinopathies has been limited in part by poor expression of globin genes in virus vectors. To derive an optimal expression cassette, we systematically analyzed the sequence requirements and relative strengths of the Agamma- and beta-globin promoters, the activities of various erythroid-specific enhancers, and the importance of flanking and intronic sequences. Expression was analyzed by RNase protection after stable plasmid transfection of the murine erythroleukemia cell line, MEL585. Promoter truncation studies showed that the Agamma-globin promoter could be deleted to -159 without affecting expression, while deleting the beta-globin promoter to -127 actually increased expression compared with longer fragments. Expression from the optimal beta-globin gene promoter was consistently higher than that from the optimal Agamma-globin promoter, regardless of the enhancer used. Enhancers tested included a 2.5-kb composite of the beta-globin locus control region (termed a muLCR), a combination of the HS2 and HS3 core elements of the LCR, and the HS-40 core element of the alpha-globin locus. All three enhancers increased expression from the beta-globin gene to roughly the same extent, while the HS-40 element was notably less effective with the Agamma-globin gene. However, the HS-40 element was able to efficiently enhance expression of a Agamma-globin gene linked to the beta-globin promoter. Inclusion of extended 3' sequences from either the beta-globin or the Agamma-globin genes had no significant effect on expression. A 714-bp internal deletion of Agamma-globin intron 2 unexpectedly increased expression more than twofold. With the combination of a -127 beta-globin promoter, an Agamma-globin gene with the internal deletion of intron 2, and a single copy of the HS-40 enhancer, gamma-globin expression averaged 166% of murine alpha-globin mRNA per copy in six pools and 105% in nine clones. When placed in a retrovirus vector, this cassette was also expressed at high levels in MEL585 cells (averaging 75% of murine alpha-globin mRNA per copy) without reducing virus titers. However, recombined provirus or aberrant splicing was observed in 5 of 12 clones, indicating a significant degree of genetic instability. Taken together, these data demonstrate the development of an optimal expression cassette for gamma-globin capable of efficient expression in a retrovirus vector and form the basis for further refinement of vectors containing this cassette.     

Increased protein binding to a -530 mutation of the human beta-globin gene associated with decreased beta-globin synthesis

Although some cases of the syndrome of hereditary persistence of fetal hemoglobin (HPFH) have been correlated with mutations causing a change in the binding of trans-acting factors to DNA sequences flanking the gamma-globin gene, this mechanism has not been described in beta-thalassemias upstream of the canonical promoter of the beta-globin gene. In this report we describe such a change in binding of a protein that may explain a silent carrier phenotype of beta-thalassemia. We have previously demonstrated the binding of a protein (BP1) derived from a nuclear extract of human K562 cells to DNA 5' to the human beta-globin gene in a region having a negative regulatory function. The binding of BP1 in this region can be detected by DNAse I footprinting and by gel mobility shift analysis. We have now compared binding of BP1 to the normal sequence and a mutated sequence (+ATA/-T at -530 bp from the cap site) from the silent carrier of beta-thalassemia. Using mobility shift assays we show that BP1 binds about nine times more strongly to the mutated sequence than the normal sequence. These results suggest the possibility that the decreased expression of the beta-globin gene exhibited by the carrier may be due, at least in part, to tighter binding of a protein which functions as a negative control element or repressor.     

The 87-kD A gamma-globin enhancer-binding protein is a product of the HOXB2(HOX2H) locus

Developmental regulation of globin gene expression may be controlled by developmental stage-specific nuclear proteins that influence interactions between the locus control region and local regulatory sequences near individual globin genes. We previously isolated an 87-kD nuclear protein from K562 cells that bound to DNA sequences in the beta- globin locus control region, gamma-globin promoter, and A gamma-globin enhancer. The presence of this protein in fetal globin-expressing cells and its absence in adult globin-expressing cells suggested that it may be a developmental stage-specific factor. A lambda gt11 K562 cDNA clone encoding a portion of the HOXB2 (formerly HOX2H) homeobox gene was isolated on the basis of the ability of its beta-galactosidase fusion protein to bind to the same DNA sequences as the 87-kD K562 protein. Because no other relationship had been established between the 87-kD K562 protein and the HOXB2 protein other than their ability to bind ot the same DNA sequences, we have investigated whether the two proteins are related antigenically. Our data show that antisera produced against the HOXB2-beta-gal fusion protein and a synthetic HOXB2 decapeptide react specifically with an 87-kD protein from K562 nuclear extract, showing that the 87-kD K562 nuclear protein is a product of the HOXB2 locus, and is the first demonstration of cellular HOXB2 protein.     

Determinants of specific RNA interference-mediated silencing of human beta-globin alleles differing by a single nucleotide polymorphism

A single nucleotide polymorphism (SNP) in the sickle beta-globin gene (beta(S)) leads to sickle cell anemia. Sickling increases sharply with deoxy sickle Hb concentration and decreases with increasing fetal gamma-globin concentration. Measures that decrease sickle Hb concentration should have an antisickling effect. RNA interference (RNAi) uses small interfering (si)RNAs for sequence-specific gene silencing. A beta(S) siRNA with position 10 of the guide strand designed to align with the targeted beta(S) SNP specifically silences beta(S) gene expression without affecting the expression of the gamma-globin or normal beta-globin (beta(A)) genes. Silencing is increased by altering the 5' end of the siRNA antisense (guide) strand to enhance its binding to the RNA-induced silencing complex (RISC). Specific beta(S) silencing was demonstrated by using a luciferase reporter and full-length beta(S) cDNA transfected into HeLa cells and mouse erythroleukemia cells, where it was expressed in the context of the endogenous beta-globin gene promoter and the locus control region enhancers. When this strategy was used to target beta(E), silencing was not limited to the mutant gene but also targeted the normal beta(A) gene. siRNAs, mismatched with their target at position 10, guided mRNA cleavage in all cases except when two bulky purines were aligned. The specific silencing of the beta(S)-globin gene, as compared with beta(E), as well as studies of silencing SNP mutants in other diseases, indicates that siRNAs developed to target a disease-causing SNP will be specific if the mutant residue is a pyrimidine and the normal residue is a purine.     

Functional properties of the beta-globin locus control region in K562 erythroleukemia cells.

In this report, we compare the function of the human beta-globin locus control region (LCR) in three K562 erythroleukemia cell assays, including (1) a transient transfection assay for "classical" enhancer activity, (2) a colony assay that detects "productive integration events," and (3) an assay that detects the ability of LCR fragments to confer hemin inducibility on linked, stably integrated gamma-globin promoters. Various LCR fragments were inserted into an expression vector consisting of an A gamma-globin promoter driving the neomycin phosphotransferase gene (gamma-neo). Using these vectors, we determined that a 2.5-kb DNA fragment containing LCR sites I through IV (previously named mu locus activation region [mu LAR]) had activity in all three assays; of the individual LCR sites, only site II was highly active in all three assays. One region within site II, consisting of tandem AP-1/NF-E2 consensus elements, had approximately 10% as much colony assay activity as the entire mu LAR. However, this region did not have detectable activity in a transient enhancer assay in uninduced K562 cells, nor was it capable of conferring hemin inducibility on linked gamma-globin promoters in stably transfected cells. Finally, we tested the ability of the mu LAR to activate promoters (beta-globin and cathepsin G) that are not normally expressed in K562 cells. beta-neo was minimally activated by the mu LAR in transient transfection experiments. The mu LAR increased the number of stably transfected colonies produced by beta-neo, but the absolute number of beta-neo colonies, with or without the mu LAR, was approximately 10% to 20% that of gamma-neo. In contrast, a minimal cathepsin G promoter was activated by the mu LAR in K562 cells. Our results suggest that LCR functions are dependent in part on the environments and the promoters with which the LCR is tested.     

Modification of globin gene expression by RNA targeting strategies

OBJECTIVE: Sickle cell anemia is a genetic blood disease resulting from production of mutant beta-globin (beta(S)) and has severe clinical consequences. It is known that a higher cellular gamma-globin level, e.g., higher ratio of cellular gamma-globin to beta(S)-globin (gamma/beta(S) ratio), inhibits sickle hemoglobin (HbS) polymerization tendency. Hence, therapeutic treatment of sickle cell anemia has been focused on introducing gamma-globin gene into red blood cells to increase the cellular gamma/beta(S) ratio. Here, we have introduced ribozymes and small interfering RNAs (siRNAs) against beta(S)-globin mRNA into blood cells as a means to increase the gamma/beta(S) ratio. MATERIALS AND METHODS: Single and multiribozymes against beta(S)-globin mRNA have been tested in vitro and in human erythroleukemia K562beta(S) cells that stably express exogenous beta(S)-globin gene. Primary human hematopoietic progenitor cells were also transfected with multiribozyme and the gamma/(gamma + beta) ratio determined and compared with cells transfected with long hairpin beta-globin cDNA and synthetic siRNA genes. RESULTS: We have found that the multiribozyme zb21A containing two ribozyme units effectively reduces beta(S)-globin mRNA both in vitro and in K562beta(S) cells. The gamma-globin mRNA to beta(S)-globin mRNA ratio in the multiribozyme transfected cells is about a factor of 2 more than that in the control cells. We have also found that the gamma/(gamma + beta) ratio in the transfected hematopoietic progenitor cells is increased by more than twofold in cells treated with multiribozyme zb21A or siRNA ib5. CONCLUSION: Our results suggest that introducing multiribozymes or siRNAs into red blood cells is comparable in their effectiveness to increase the ratio of cellular gamma-globin mRNA to beta- or beta(S)-globin mRNA, providing possible strategies to increase the effectiveness of gamma-globin gene transfer as gene therapy for treatment of patients with sickle cell anemia.     

Regulated high level expression of a human gamma-globin gene introduced into erythroid cells by an adeno-associated virus vector.

Gene therapy of severe hemoglobinopathies will require high-level expression of a transferred globin gene in erythroid cells. Distant regulatory elements flanking the beta-globin gene cluster, the locus control region, are needed for appropriate expression. We have explored the use of a human parvovirus, the adeno-associated virus (AAV), for globin gene transfer. The human A gamma-globin gene, linked to hypersensitivity site 2 from the locus control region of the beta-globin gene cluster, was subcloned into a plasmid (psub201) containing the AAV inverted terminal repeats. This construct was cotransfected with a helper plasmid containing trans-acting AAV genes into human 293 cells that had been infected with adenovirus. The recombinant AAV vector containing hypersensitivity site 2 stably introduced on average one or two unrearranged proviral copies into human K562 erythroleukemia cells. The transferred globin gene exhibited normal regulation upon hemin induction of erythroid maturation and was expressed at a level equivalent to a native chromosomal A gamma-globin gene.     

Developmental- and differentiation-specific patterns of human gamma- and beta-globin promoter DNA methylation

The mechanisms underlying the human fetal-to-adult beta-globin gene switch remain to be determined. While there is substantial experimental evidence to suggest that promoter DNA methylation is involved in this process, most data come from studies in nonhuman systems. We have evaluated human gamma- and beta-globin promoter methylation in primary human fetal liver (FL) and adult bone marrow (ABM) erythroid cells. Our results show that, in general, promoter methylation and gene expression are inversely related. However, CpGs at -162 of the gamma promoter and -126 of the beta promoter are hypomethylated in ABM and FL, respectively. We also studied gamma-globin promoter methylation during in vitro differentiation of erythroid cells. The gamma promoters are initially hypermethylated in CD34(+) cells. The upstream gamma promoter CpGs become hypomethylated during the preerythroid phase of differentiation and are then remethylated later, during erythropoiesis. The period of promoter hypomethylation correlates with transient gamma-globin gene expression and may explain the previously observed fetal hemoglobin production that occurs during early adult erythropoiesis. These results provide the first comprehensive survey of developmental changes in human gamma- and beta-globin promoter methylation and support the hypothesis that promoter methylation plays a role in human beta-globin locus gene switching.