Nuclear proteins preferentially bound to the beta-globin gene: cellular specificity and variation during terminal differentiation of mouse erythroblasts

Restriction fragments of the mouse beta major globin gene and of the long terminal repeat (LTR) DNA fragment of the mouse mammary tumor provirus as a control, were used to analyze the specificity of DNA-protein interactions in nuclear extracts of mouse erythroleukemia (MEL) cells and of other differentiated mouse cultured cell lines. After gel electrophoresis and transfer to nitrocellulose, DNA-binding proteins with a preferential affinity for the cloned beta-globin genomic sequence were characterized and related to the level of globin gene expression during induction of differentiating mouse erythroblasts. Two proteins (110 K and 75 K) appear in differentiated MEL cells while another one (100 K), for which we have localized the binding site on the beta-globin gene, is present only in immature MEL cells.     

Structure and function of the murine beta-globin locus control region 5' HS-3.

We previously identified the murine homologue of the human beta-globin Locus Control Region (LCR) 5' HS-2. The lambda clone containing murine 5' HS-2 extends approximately 12 kb upstream from this site; here, we report the sequence of this entire upstream region. The murine homologue of 5' HS-3 is located approximately 16.0 kb upstream from the mouse epsilon y-globin gene, but no region homologous to human 5' HS-4 was present in our clone. Using a reporter system consisting of a human gamma-globin promoter driving the neomycin phosphotransferase gene (gamma-neo), we tested murine LCR fragments extending from -21 to -9 kb (with respect to the epsilon y-globin gene cap site) for activity in classical enhancer and integration site assays in K562 and MEL cells. 5' HS-2 behaved as a powerful enhancer and increased the number of productive integration events (as measured by a colony assay) in both K562 and MEL cells. 5' HS-3 had no activity in K562 cells or in transiently transfected MEL cells, but was nearly as active as 5' HS-2 in the MEL cell colony assay. Two additional tests confirmed the identification of murine 5' HS-3: first, a DNA fragment containing 5' HS-3 confers copy number-dependent, integration-site independent inducibility on a linked beta-globin gene in the MEL cell environment. Secondly, a strong DNAseI hypersensitive site maps to the location of the 5' HS-3 functional core in chromatin derived from MEL cells. Collectively, these data suggest that we have identified the murine homologue of human 5' HS-3, and that this site is functional when integrated into the chromatin of MEL cells but not K562 cells. 5' HS-3 may therefore contain information that contributes to the development-specific expression of the beta-like globin genes.     

Characterization of trans-acting factor(s) regulating beta-globin gene expression by in vivo competition

A beta-globin/TK fusion gene was microinjected into non-erythroid cells (Ltk- cells) and erythroid cells (murine erythroleukemia (MEL) cells), and the interactions of the regulatory cellular factors with the beta-globin sequences were investigated by the in vivo competition experiment. The fusion gene was expressed efficiently in Ltk- cells. This expression was inhibited by a co-injection with a three-fold molar excess of the 5'-flanking sequence of the beta-globin gene or with a nine-fold molar excess of the mammary tumor virus LTR, but not with the alpha-globin gene. The fusion gene was expressed very poorly in the uninduced MEL cells and highly in the induced MEL cells. The co-injection of the beta-globin gene did not affect expression in the MEL cells in either uninduced or induced conditions.     

CACCC and GATA-1 sequences make the constitutively expressed alpha-globin gene erythroid-responsive in mouse erythroleukemia cells.

Although the human alpha-globin and beta-globin genes are co-regulated in adult life, they achieve the same end by very different mechanisms. For example, a transfected beta-globin gene is expressed in an inducible manner in mouse erythroleukemia (MEL) cells while a transfected alpha-globin gene is constitutively expressed at a high level in induced and uninduced MEL cells. Interestingly, when the alpha-globin gene is transferred into MEL cells as part of human chromosome 16, it is appropriately expressed in an inducible manner. We explored the basis for the lack of erythroid-responsiveness of the proximal regulatory elements of the human alpha-globin gene. Since the alpha-globin gene is the only functional human globin gene that lacks CACCC and GATA-1 motifs, we asked whether their addition to the alpha-globin promoter would make the gene erythroid-responsive in MEL cells. The addition of each of these binding sites to the alpha-globin promoter separately did not result in inducibility in MEL cells. However, when both sites were added together, the alpha-globin gene became inducible in MEL cells. This suggests that erythroid non-responsiveness of the alpha-globin gene results from the lack of erythroid binding sites and is not necessarily a function of the constitutively active, GC rich promoter.     

RNA from rat hepatoma cells can activate phenylalanine hydroxylase gene of mouse erythroleukemia cells

Mouse erythroleukemia (MEL) cells do not synthesize any detectable level of phenylalanine hydroxylase and thus do not grow in Tyr- medium. Rat hepatoma cells that constitutively express phenylalanine hydroxylase were treated prior to fusion with MEL cells with biochemical inhibitors to inactivate different macromolecular components of the cells, and the fusion products were selected in Tyr- medium. Continuously growing populations of cells resembling the parental MEL cells and expressing mouse phenylalanine hydroxylase were obtained only when rat hepatoma cells treated with mitomycin or iodoacetamide, which inactivate DNA and SH proteins, respectively, were fused with MEL cells. Fusion of MEL cells with UV-treated rat hepatoma cells did not result in the activation of the mouse phenylalanine hydroxylase gene. UV treatment damages both DNA and RNA. These data suggested that RNA was involved in the regulation of phenylalanine hydroxylase gene. Additional evidence for the role of RNA in the phenylalanine hydroxylase gene regulation was obtained from RNA transfection studies. RNA only from cells which express phenylalanine hydroxylase, such as rat hepatoma cells and MEL cybrids, when introduced into MEL cells by the CaPO4 coprecipitation method, resulted in the permanent activation of the mouse phenylalanine hydroxylase gene.     

LCR/MEL: a versatile system for high-level expression of heterologous proteins in erythroid cells.

We have used the human globin locus control region (LCR) to assemble an expression system capable of high-level, integration position-independent expression of heterologous genes and cDNAs in murine erythroleukaemia (MEL) cells. The cDNAs are inserted between the human beta-globin promoter and the second intron of the human beta-globin gene, and this expression cassette is then placed downstream of the LCR and transfected into MEL cells. The cDNAs are expressed at levels similar to those of the murine beta-globin in the induced MEL cells. Heterologous genomic sequences can also be expressed at similar levels when linked to to the LCR and beta-globin promoter. In addition we demonstrate that, after induction of differentiation, MEL cells are capable of secreting heterologous proteins over a prolonged time period, making this system suitable for use in continuous production systems such as hollow fibre bioreactors. The utility of the LCR/MEL cell system is demonstrated by the expression of growth hormone at high levels (greater than 100 mg/l) 7 days after induction. Since the expression levels seen do not depend upon gene amplification and are independent of the integration position of the expression cassette, it is possible to obtain clones with stable high-level expression within 3-4 weeks after transfection.     

Affinity isolation of active murine erythroleukemia cell chromatin: uniform distribution of ubiquitinated histone H2A between active and inactive fractions.

This laboratory recently reported the development of a biotin-cellulose/streptavidin affinity chromatography method based on the DNase I sensitivity of active chromatin to isolate a DNA fraction from murine erythroleukemia (MEL) cells that is more than 15-fold enriched in active genes (Dawson et al.: Journal of Biological Chemistry 264:12830-12837, 1989). We now report the extension of this technique to isolate and characterize chromatin that is enriched in active genes. In this approach, DNA in nuclei isolated from MEL cells was nicked with DNase I at a concentration that does not digest the active beta-globin gene, followed by repair of the nicks with a cleavable biotinylated nucleotide analog, 5-[(N-biotin-amido)hexanoamido-ethyl-1,3'-dithiopropionyl-3- aminoallyl]-2'- deoxyuridine 5'-triphosphate (Bio-19-SS-dUTP), during a nick-translation reaction. After shearing and sonication of the nuclei to solubilize chromatin, chromatin fragments containing biotin were separated from non-biotinylated fragments by sequential binding to streptavidin and biotin cellulose. The bound complex contained approximately 10% of the bulk DNA. Reduction of the disulfide bond in the biotinylated nucleotide eluted approximately one-half of the affinity isolated chromatin. Hybridization analysis of DNA revealed that whereas inactive albumin sequences were equally distributed among the chromatin fractions, virtually all of the active beta-globin sequences were associated with chromatin fragments which had bound to the affinity complex. Western blot assessment for ubiquitinate histones revealed that ubiquitinated histone H2A (uH2A) was uniformly distributed among active (bound) and inactive (unbound) chromatin fractions.     

Expression of chimeric human beta- and delta-globin genes during erythroid differentiation

To determine whether sequences contained within the small intervening sequence (IVS 1) or large intervening sequence (IVS 2) are involved in the regulated expression of the human beta-globin gene, chimeric genes containing portions of the human beta- and delta-globin genes were stably transfected into mouse erythroleukemia (MEL) cells. Since MEL cells can be induced to differentiate in culture, the expression of the chimeric genes was compared to the expression of beta and delta both before and after the induction of erythroid differentiation. The expression of beta delta 1, a beta-globin gene containing delta IVS 1 in place of beta IVS 1, was comparable to the expression of a beta-globin gene both before and after erythroid differentiation. However, the base-line expression of human beta-globin genes containing delta IVS 2 in place of beta IVS 2 was dramatically decreased. Furthermore, the substitution of delta IVS 2 for beta IVS 2 prevented the regulated increase in expression of the beta-globin gene upon induction. The results also indicate that sequences present in beta IVS 2 are not sufficient for this induced increase in expression since the substitution of beta IVS 2 for delta IVS 2 in a delta gene does not increase the regulated expression of delta during differentiation. These experiments suggest that either the presence of delta IVS 2 in a beta gene interrupts sequences required for the induced expression of beta-globin or that sequences in beta IVS 2 act in concert with other beta globin sequences not present in the delta-globin gene to permit optimal expression.