Imprinting regulation of the murine Meg1/Grb10 and human GRB10 genes; roles of brain-specific promoters and mouse-specific CTCF-binding sites

Abstract

The imprinted mouse gene Meg1/Grb10 is expres sed from maternal alleles in almost all tissues and organs, except in the brain, where it is expressed biallelically, and the paternal allele is expressed preferentially in adulthood. In contrast, the human GRB10 gene shows equal biallelic expression in almost all tissues and organs, while it is almost always expressed paternally in the fetal brain. To elucidate the molecular mechanisms of the complex imprinting patterns among the different tissues and organs of humans and mice, we analyzed in detail both the genomic structures and tissue-specific expression profiles of these species. Experiments using 5'-RACE and RT-PCR demonstrated the existence in both humans and mice of novel brain- specific promoters, in which only the paternal allele was active. The promoters were located in the primary differentially methylated regions. Interest ingly, CTCF-binding sites were found only in the mouse promoter region where CTCF showed DNA methylation-sensitive binding activity. Thus, the insulator function of CTCF might cause reciprocal maternal expression of the Meg1/Grb10 gene from another upstream promoter in the mouse, whereas the human upstream promoter is active in both parental alleles due to the lack of the corresponding insulator sequence in this region.

Figure 1

Two types of transcript from mouse Meg1/Grb10 and human GRB10. (A) Brain-specific transcripts (Brain type) are identified by 5′-RACE from exon 3 of the mouse Meg1/Grb10 gene and exon 2 of the human GRB10 gene, as well as the major types in both humans and mice, as described previously. Black and white arrowheads indicate the positions and directions of the 5′-RACE and RT–PCR primers [see (B)], respectively. (B) Exon-specific RT–PCR experiments show that the transcripts in the brain are transcribed mainly from exon 1b in the mouse and exon un2 in the human, while those in other tissues are transcribed exclusively from mouse exon 1a and human exon un1. The primers used are shown in Materials and Methods. No other spliced forms of transcripts were detected in the mouse, whereas two types of transcripts with or without the un4 exon were detected in the human. The results of the RT–PCR between exons 2 and 3 in both humans and mice, which are common to the two types of transcripts, indicate that the relative expression levels in the brains are almost identical to those in other tissues and organs of both humans and mice. β-actin was measured as the control for the RT–PCR. (C) Different parental origins of the two types of transcript in the mouse. In contrast to the maternal expression of mouse Meg1/Grb10 in embryos, adult kidneys and liver, as previously published, Meg1/Grb10 was expressed mainly from the paternal allele in adult brains. BJ, (B6 mother × JF1 father) F₁; JB, (JF1 father × B6 mother) F₁.

Figure 2

DNA methylation status of two promoter regions in the mouse Meg1/Grb10 gene. Eight to ten clones from the paternal and maternal alleles were sequenced. DNA methylation was absent in both the paternal and maternal alleles of the Meg promoter region. In contrast, differential methylation (i.e. full methylation in paternal alleles and non-methylation in maternal alleles) was detected in the Peg promoter region. Differential methylation patterns were already established in the sperm and eggs. Parental alleles are distinguished by DNA polymorphisms between JF1 and B6. The black circles indicate methylated CpGs and the white circles indicate non-methylated CpGs.

Figure 3

Comparison of the human and mouse promoter regions. (A) The two promoter regions that are located in the CpG islands show high sequence homologies between humans and mice, with the exception of the 600 bp mouse-specific tandem repeat in the Peg promoter region. Putative translation start sites are indicated by ATG. DNA homologies between these regions were calculated using the program VISTA, which is based on moving a user-specified window (100 bp in this work) over the entire alignment and calculating the percent identity over the window at each base pair. (B) Twelve repeats of the 10 bp motif GGCGCGTG(C/T)T are observed in the mouse-specific region. The 40 bp sequence (boxed) was used as the probe in the gel shift assays (see Fig. 4, Meg1 repeat)

Figure 4

CTCF binds specifically to the Meg1 repeat. (A) Specific competition for CTCF binding to the Meg1 repeat was seen for the canonical CTCF- binding sequence of the chicken β-globin FII (as shown in F), but not for the transcription factor recognition sequences of SP1 and AP1 (as shown in S and A) and vice versa. Mouse recombinant CTCF containing a full-length coding sequence was used. There were no DNA-binding factors included in the in vitro reticulocyte transcription/translation reaction mixture without CTCF-containing vector (left most lane). (B) The methylated Meg1 repeat was a less effective competitor than the non-methylated sequence.

Figure 5

Working models of tissue-specific expression regulation of the mouse Meg1/Grb10 and human GRB10 genes. The mouse model with an insulator sequence (left) and the human model without an insulator sequence (right) are shown. White lollipops indicate the unmethylated CpG motifs and black lollipops indicate the methylated CpG motifs. The small ovals indicate putative downstream enhancers and the large ovals indicate CTCF. Big and small arrows correspond to the expression levels measured by RT–PCT, as shown in Figure 1B and C. In both the mouse and human models, we presuppose that paternal expression from Peg promoters is regulated by both DNA methylation and putative brain-specific activators (not drawn in these figures); therefore, paternal alleles in other tissues are not active due to the lack of activator and maternal alleles are inactivated by DNA methylation. In the non-methylated paternal allele in the mouse, the upstream Meg promoter is repressed because the insulator blocks the function of the downstream enhancer, whereas in methylated maternal alleles the major promoter is active, since there is no interference with the enhancer blocking effect of the insulator. However, maternal expression in the brain is low compared with expression in other tissues (shown with dashed lines). It is possible that the downstream enhancer acts in a somewhat tissue-specific manner and is weak in the brain. Therefore, it shows biallelic, but strong paternally biased, expression in the brain, and maternal expression in other tissues. In humans (right), both of the parental major promoters are active, since the insulator sequence is absent. Therefore, it shows biallelic, but strong paternally biased, expression in the fetal brain, and biallelic expression in other tissues.