Understanding the complex regulation of beta-globin genes is critically important to design therapeutic approaches to beta-thalassemia and sickle cell disease. We previously reported that a complex of erythroid proteins GATA-1, LMO2 and TAL1 as well as the more widely expressed nuclear factor LDB1 mediates beta-globin gene looping to the distant locus control region (LCR) enhancer, which is required for transcription activation. We have found that dimerization of LDB1 is required for looping and beta-globin transcription rescue in the background of LDB1 knock down (KD) erythroid cells. Deletion of a small conserved region within the dimerization domain (DD4/5) yielded an LDB1 mutant protein that can dimerize and rescue long-range LCR/beta-globin looping but is unable to rescue beta-globin gene expression. DD4/5 is required for the recruitment of the co-regulators FOG1 and the nucleosome remodeling and deacetylating (NuRD) complex. Lack of DD4/5 alters histone acetylation and RNA polymerase II recruitment and results in failure of the locus to migrate to the nuclear interior, as normally occurs during erythroid maturation. These results uncouple enhancer-promoter looping from nuclear migration and transcription activation. Using RNA-seq and DESeq analysis of LDB1 KD cells overexpressing full length LDB1 or LDB1 without DD4/5 we identified LDB1-dependent genes sensitive (4/5-dependent) and insensitive (4/5-independent) to loss of this region. ChIP confirmed that FOG1 occupancy relies on the DD4/5 region of LDB1 at the 4/5-dependent genes. These results confirm a broad role for LDB1 DD4/5 and FOG1 interaction in regulation of the erythroid gene transcriptome. In addition to LDB1/GATA-1/TAL1/LMO2 protein complex, histone-modifying enzymes play a significant role in regulation of globin gene expression. For example, data suggest that G9a methyltransferase, responsible for establishing H3K9me2 histone modification, may be involved in repressing fetal and activating adult globin gene expression in erythroid cells. Using ex vivo differentiation of primary CD34(+) adult human cells as a model system, we found that repression of G9a activity during the most proliferative phase of cell differentiation caused a highly significant increase in gamma-globin gene expression as well as reduction of beta-globin gene expression without substantial change in alpha-globin gene expression. ChIP confirmed G9a repression resulted in elimination of H3K9me2 histone modification from the beta-globin locus. Moreover, G9a repression was accompanied by a significant increase of LDB1 complex occupancy at the gamma-globin gene and decrease at delat- and beta-gene promoters as well as changes in LCR looping from interaction with the gamma- to the beta-globin gene. These results support a model whereby G9a establishes conditions preventing activation of gamma-globin genes by facilitating LCR looping with delat- and beta-gene promoters and subsequent strong activation of adult globin gene expression during differentiation of adult erythroid progenitor cells. In this view, G9a inhibition represents a promising approach for treatment of beta-hemoglobinopathies. Cell type specific regulatory elements, enhancers and promoter, control lineage specific transcriptomes. Our current understanding is that lineage-specific transcription factors and complexes that binds to these regulatory elements, can orchestrate these long range interactions, for example, by protein dimerization. In other cases, lineage-specific factors co-opt ubiquitous architectural looping proteins such as CTCF into cell type-specific enhancer long range interactions but in these cases molecular details are unclear. We identified a region upstream of the carbonic anhydrase gene 2 gene (Car2) that is strongly occupied by the LDB1 complex and that loops to Car2 when it is highly transcribed in erythroid cells. But the promoter of the gene is occupied by CTCF rather than LDB1 complex proteins. Both ectopic enhancer assays and genome editing with CRISPR-Cas9 identify the upstream LDB1-bound region as a Car2 enhancer and also showed that the promoter CTCF site is essential for Car2 activation. Knock down of either LDB1 or CTCF abrogates the loops and strongly reduces CAR2 transcription. And we found that LDB1 can directly interact with CTCF through the LCCD domain independent of GATA1 factor. Moreover, we identify a subset of genes with CTCF-bound promoters and long range interactions with LDB1-bound known or putative enhancers. CRISPR-Cas9 deletion of these enhancers compromises transcription of the genes, validating the enhancer function and generalizing the importance of LDB1-CTCF interactions in enhancer looping and establishment of the erythroid transcriptome. The principles underlying the architectural landscape of chromosomes in living cells remain largely unknown despite its potential to play a role in mammalian gene regulation. We investigated the 3-dimensional conformation of a 1 Mbp region of human chromosome 11 containing the beta-globin genes by integrating looping interactions of the insulator protein CTCF determined comprehensively by 3C into a polymer model of chromatin folding. Knock down of CTCF showed that regional CTCF contacts in mammalian nuclei functionally affect spatial distances between globin genes and their control elements and, hence, contribute to chromosomal reorganization required for transcription. We are using genome editing with TALENs to delete CTCF sites with strong 3C interaction frequencies such as HS5 and 3HS1 flanking the beta-globin locus and a CTCF binding site several hundred kb distant (C26)in erythroid cells and testing the effect on globin gene expression and interaction frequencies of the remaining CTCF sites. PCR amplification and sequencing was used to validate mono- and di-allelic deletions. Western blot showed that CTCF protein level remained the same in CTCF binding site deleted cells compared to normal cells. CHIP indicated that CTCF binding site deletion resulted in loss of CTCF binding at the targeted sites. Interestingly, deletion of one CTCF motif was capable of reducing CTCF occupancy at another CTCF site, possibly by affecting long range interactions. For example, 3HS1 CTCF deletion caused reduced CTCF binding in HS5 and C26 sites. Interaction frequencies of the remaining CTCF sites will be tested in the clones with edited CTCF sites. We will see the predictive nature of the polymer model with the new data.
Showing the most recent 10 out of 21 publications
