|Home | About | Journals | Submit | Contact Us | Français|
In this issue of Developmental Cell, Kernohan et al. link the chromatin regulatory proteins ATRX, MeCP2, CTCF, and cohesin with silencing of H19 and other imprinted genes during critical stages of postnatal brain development, perhaps suggesting a common etiology for several human diseases that exhibit defects in brain development and function.
Chromatin architecture and genome organization play essential roles in establishing the gene expression patterns that are critical for normal development and function. A number of human disorders have been identified whose primary genetic defect is in loci whose gene products regulate chromatin and chromosome architecture. Examples include Rett syndrome, which is associated with loss of function of the methylated-DNA binding protein, MeCP2; Cornelia de Lange syndrome, which is associated with mutations in the cohesin-related factors SMC1A, SMC3, and NIPBL; and Alpha-Thalassemia mental Retardation, X-linked (ATR-X), which is associated with mutations in ATRX, an ATP-dependent chromatin remodeling protein. All three of these diseases present broad and partially overlapping phenotypes that include deficits in brain development and function. While the primary regulatory defect for each of these syndromes has been determined, a major challenge still exists to identify the downstream targets of these genes with the goal of eventually developing appropriate therapies. In this issue, Kernohan et al. (Kernohan et al., 2010)analyze the roles of ATRX, cohesin, and MeCP2 at the imprinted H19 gene and demonstrate these factors interact together with the transcription factor/insulator protein CTCF to regulate expression of an entire network of imprinted genes. This leads to the intriguing hypothesis that changes in expression of imprinted genes may contribute to the brain defects associated with these diseases.
The Igf2/H19 locus is a typical imprinted gene cluster(Wan and Bartolomei, 2008). It includes paternal specific genes, Insulin2 and Insulin-like growth factor 2 (Igf2), and one maternal specific gene, H19. The maternal and paternal chromosomes differ not only in gene activity but also in chromosome architecture. They display distinct patterns of DNA methylation, histone modification, and DNA loop formation. All of the differences across the >100 kb locus depend entirely upon a 2.4 kb element, the H19 Imprinting Control Region (H19ICR) that lies just adjacent to the H19 promoter and about 100 kb upstream of Igf2 and Insulin2. Deletion of the ICR renders the two chromosomes indistinguishable, while insertion of this element ectopically will result in artificial imprinting of the targeted locus.
A basic understanding of the H19ICR's abilities has been obtained through molecular and genetic analyses. Maternally inherited ICR DNA is not methylated and binds the CTCF transcription factor in a cohesin-dependent manner, resulting in expression of H19 in the embryo and early neonatal stages. Moreover, CTCF-bound maternal ICR interacts through DNA looping with the Igf2 promoter region, preventing Igf2 promoters from physically interacting/looping with downstream enhancers. In contrast, paternal ICR DNA is methylated and therefore cannot bind CTCF. Since CTCF-dependent loops are not formed between the ICR and the Igf2 gene on the paternal chromosome, the Igf2 promoters are instead free to interact with the downstream enhancers and activate paternal expression. In addition, the methylated ICR induces chromatin changes at the adjacent H19 promoter that prevent its activation.
Besides these intra-chromosomal interactions, recent studies have demonstrated CTCF dependent inter-chromosomal interactions between the maternal H19ICR and other loci, with a particular enrichment for imprinted genes(Ling et al., 2006; Sandhu et al., 2009; Zhao et al., 2006). Thus there is quite a lot going on at the H19ICR, and Kernohan et al. demonstrate that the element is even busier than heretofore thought and in ways that link critical disease loci to the H19ICR.
Based on initial findings that H19 and Igf2 were significantly upregulated in ATRX-null mice, Kernohan et al. characterize protein complexes at the H19ICR. They demonstrate that in addition to cohesin and CTCF, ATRX and MeCP2 also bind specifically to the maternal (i.e. non-methylated) ICR. Furthermore, cohesin and CTCF binding are ATRX-dependent. The MeCP2 binding to the maternal ICR is surprising, since it was always assumed that this CpG methyl-binding protein interacted specifically with the methylated paternal ICR. Kernohan's results thus support newer hypotheses that MeCP2's primary role is in forming DNA loops that stimulate gene activity and is not limited to mediating DNA-methylation induced repression(LaSalle, 2007).
The increase in H19 expression in ATRX-deficient mice is maternal in origin, i.e. there is no loss of imprinting. This implicates ATRX in postnatal repression of H19 (and possibly Igf2). After birth, expression of these two genes is repressed hundreds of fold. The biological necessity for this repression is absolute, but the underlying mechanisms have been elusive until now. The authors do not determine the parental origin of the extra Igf2, but their results predict that it is likely maternal in origin.
Kernohan et al. also demonstrate interactions of cohesin, MeCP2, and ATRX proteins at the Glt2 ICR, although in this case their analyses did not allow them to determine the parent-of-origin specificity of the interactions. The Dlk1/Glt2 locus is somewhat analogous to the Igf2/H19 locus in gene organization and regulation(Wan and Bartolomei, 2008). An ICR adjacent to the Gtl2 promoter regulates Gtl2 and the far upstream Dlk1. Kernohan et al. results support the idea that interactions of cohesin, MeCP2, and ATRX proteins are of general importance and not restricted to H19/Igf2. However, several differences between the two loci suggest that the nature of these interactions may not be straightforward. At Dlk1/Glt2, the proteins each bind to distinct parts of the ICR and not to a single region like they do at the H19 locus. Furthermore, MeCP2 binding to the ICR is ATRX-dependent at Glt2 but ATRX-independent at H19.
By several criteria, Igf2, H19, Dlk1, and Glt2 are part of a network of at least 10 imprinted genes(Varrault et al., 2006). These genes all share developmental and tissue-specific patterns of expression and respond similarly to mutations at the Zac1 locus. Curiously this network shows almost no overlap with imprinted genes involved in interchromosomal interactions with the H19ICR. Kernohan et al. provides good evidence that ATRX is required for the down regulation of expression of each of these genes in late embryonic or in postnatal development. The key question remains whether the down regulation of any of these imprinted genes, including H19, is important in the ATRX syndrome.
Evolutionary theory and analysis of many knockout mouse strains both support the idea that imprinted genes are likely to play their major role in regulating fetal and early neonatal growth. However, some experiments have suggested a role for imprinted genes in brain development and function(Wilkinson et al., 2007). Chimeric animals generated by mixtures of wild type and gynogenetic (maternal chromosomes only) cells or by mixtures of wild type and androgenetic (paternal chromosomes only) cells, show divergent phenotypes with gynogenetic and androgenetic cells each contributing to distinct brain structures(Keverne et al., 1996). These experiments are hard to interpret on a molecular level. Gynogenetic cells not only lack any paternal specific transcripts but also have two-fold over-expression of all maternal specific RNAs. Nonetheless the results are intriguing. Mammalian cells go to great effort to carefully regulate the doses of imprinted genes. Whether mis-expression of H19 or of any of the imprinted genes plays a clinically important role in brain development and function is an important and difficult one that remains to be addressed.
This work was supported by the Intramural Research Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development.