BIOLOGY OF REPRODUCTION 54, 273-278 (1996) Genomic Imprinting in Mice: Its Function and Mechanism' Philip A. Leighton, Jennifer R. Saam, Robert S. Ingram, and Shirley M. Tilghman 2 Howard Hughes Medical Institute and Departmentof Molecular Biology Princeton University, Princeton, New Jersey 08544 ABSTRACT Genomic imprinting is an epigenetic phenomenon by which the two parental alleles of a gene are differentially expressed. Although Downloaded from https://academic.oup.com/biolreprod/article/54/2/273/2761811 by guest on 30 September 2021 the function of genomic imprinting is not clear, it has been proposed that it evolved in mammals to regulate intrauterine growth. This proposal is consistent with experiments that were designed to reveal the mechanism and impact of genomic imprinting in a region of mouse chromosome 7 that contains four imprinted genes: Mash-2 (a transcription factor) and H19 (a noncoding RNA) are maternally expressed, whereas Insulin-2(lns-2) and Insulin-like growth factor2(lgf-2) are paternally expressed. Two targeted disruptions at the locus were generated in mice; these support the hypothesis that the function of the H19 gene is to set up the imprinting of both Igf-2 and Ins- 2.H19transcription on the maternal chromosome precludes transcription of the other two genes by a mechanism that involves competition for a common set of enhancers. On the paternal chromosome the H19 gene is silenced by DNA methylation, thus permitting the use of enhancers by the other genes. INTRODUCTION allelic inactivation by imprinting. The function presumably Genomic imprinting is an epigenetic phenomenon by must confer selective advantage to the organism, as it must which the two parental alleles of a gene are differentially counterbalance the obvious attendant risk of hemizygosity. expressed [1, 2]. The existence of genomic imprinting in Several explanations have been proposed for the acquisi- mice was first deduced from nuclear transplantation exper- tion of genomic imprinting in eutherian mammals, includ- iments in which diploid zygotes were generated with ex- ing protection against parthenogenesis and ovarian tumors clusive genetic contribution from either mothers or fathers as well as chromosome loss [11-14]. To date the most com- [3, 4]. Neither kind of uniparental embryo was able to sus- pelling model has been provided by Haig and his col- tain normal development, and the conclusion was drawn leagues [15], who suggest that imprinting evolved in mam- that the lethality was due to the absence or overexpression mals because of the conflicting interests of maternal and of one or more gene products as the result of the unequal paternal genes within a litter. In mammals, which are pri- expression of genes in the maternal and paternal genomes. marily nonmonogamous, the mother provides significant As of 1995 there are twelve genes known to be subject maternal resources to the offspring both during intrauterine to genomic imprinting in mammals, three that are mater- life and during suckling after birth. Successful passage of nally expressed and nine that are paternally expressed (Ta- paternal genes into the next generation is best ensured by ble 1). However, on the basis of the mapping of imprinted having the embryos consume maternal resources, even if regions in the mouse genome using classical genetics [5], it by so doing the fitness of her future litters is compromised. is evident that Table 1 contains only a subset of the im- The mother's interests are best served by distributing her printed genes in the genome. For most genes that have been resources more equitably among litters. tested, there is conservation of imprinting between mice Haig's model predicts that paternally expressed genes and humans, although there is at least one exception, the will promote embryonic growth whereas maternal genes Insulin-like growth factor 2 receptor (Igf-2r), which is ma- will act to restrain the use of maternal resources. In so do- ternally expressed in mice but biallelic in humans [6-8]. ing, this model is in remarkably good agreement with the Some of the imprinted genes are organized in clusters, and function of at least three imprinted genes: Insulin-like two of the clusters, on mouse chromosomes 7 and 17, con- growthfactor2 (Igf-2), H19, and Igf-2r. Mouse embryos are tain both maternally and paternally expressed genes [9, 10]. exquisitely sensitive to the levels of the paternally ex- pressed growth factor insulin-like growth factor II (IGFII). A complete loss of function of the gene encoding the THE FUNCTION OF GENOMIC IMPRINTING growth factor, Igf-2, leads to a 40% reduction in birth weight With a growing list of imprinted genes in hand, the chal- [16]. Igf-2r, which is maternally expressed, encodes a bi- lenge is now to determine the function and mechanism of functional protein that acts as a sink for the growth factor. Fetuses that lack the receptor are approximately 30% larger 1 Supported by a grant from the National Institutes of Health (GM51460). 2 than normal, have elevated circulating levels of IGFII, and Correspondence: Shirley M. Tilghman, Department of Molecular Biology, Princeton University, Lewis Thomas Laboratory, Room 127, Washington Road, Princeton, NJ 08544. die around birth [17, 18]. It has been argued that the lethality FAX: (609) 258-3345; e-mail: [email protected] is the consequence of a gain of function of the effective 273 274 LEIGHTON ET AL. TABLE 1. Chromosomal location if imprinted Gene Expression In mouse In human References Igf-2 paternal distal 7 11p15.5 [16, 47 481 H19 maternal distal 7 11p15.5 [28, 49, 50] Insulin-2 paternal distal 7 not tested [241 Mash-2 maternal distal 7 not tested [22] Snrpn paternal central 7 15q13-14 [51, 521 Znf-127 paternal central 7 15q13-14 R. Nicholls, personal comm. IPW paternal not tested 15q13-14 [451 Igf-2r maternal proximal 17 not imprinted 16, 7, 81 Mas paternal proximal 17 not tested 191 Downloaded from https://academic.oup.com/biolreprod/article/54/2/273/2761811 by guest on 30 September 2021 Xist paternal x X [53, 54] U2afbp-rs paternal proximal 11 not tested [551 Peg-I/Mest paternal proximal 6 not tested [561 . ____________ ._ _n- ~11tr\~n the~ distnl~t"I1 ~ nnrtinn Cu1 1 nf1 mnlm11 o rchrnmnnmo1 1 11 1 1 7 thath t Itnntains tI. Mash-2 Ins-2 IXf2 H19 four imprinted genes (Fig. 1). The Mash-2 gene, one of a family of helix-loop-helix proteins, is expressed from the Mat Pat Pat Mat maternal chromosome in spongiotrophoblasts of the mouse placenta [21-23]. The Insulin-2gene (Ins-2), which lies ap- 100kb ---" -- - I 1. -- ... - II I I...1. "1 1 - I "~ .. I ~ [ l. alllla.ly AJtJ- JVJ llThI 11 IJIUJIfLJ 1 Jt'l CA- I ly FIG. 1. An imprinted region on the distal end of nnouse chromosome 7. The po- pressed in the yolk sac of the mouse embryo but biallelically sitions of the imprinted genes Mash-2, Ins-2, Igf- (Igf2 on figure), and H19 on expressed in the pancreas [24]. The paternally expressed mouse chromosome 7 [10, 221 are indicated. Mat, maternally expressed; Pat, pa- Igf-2 gene lies immediately downstream from Ins-2 [16]; and ternally expressed. at the end of the cluster lies the H19gene, an unusual gene concentration of circulating IGFII, orn the basis of the ob- that codes for a nontranslatable RNA [25-27]. Its expression servation that the lethality can be abr<ogated by paternal in- is exclusively maternal [28]. heritance of an Igf-2 null mutation []18, 19]. As will be dis- The clustering of four imprinted genes raised the possibility cussed below, the maternally exprn2ssed H19 gene also that imprinting, like X chromosome inactivation, is controlled functions to lower the concentration of IGFII by suppress- by a signal or signals that can act over large distances and ing its transcription on the maternal chromosome [20]. The multiple genes. Although the Igf-2 and H19 genes are ex- implication of these experiments is that imprinting of all pressed from different parental chromosomes, it seemed three genes is functioning to finely reregulate the expression likely that their imprinting was mechanistically linked, as evi- of the Igf-2 gene. denced by the observation that their temporal and tissue-spe- cific patterns of expression are essentially identical [29, 30]. A MODEL FOR THE MECHANIS ;M OF GENOMIC We proposed that the mechanism relied on a competition be- IMPRINTING tween their promoters for a set of shared regulatory elements such as enhancers (Fig. 2) [31, 32]. To begin to dissect the molecular rmechanism of imprint- The original model proposed that the competition was ing, we have focused on a region of approximately 300 kb biased differently on the two parental chromosomes by epi- genetic mark(s) placed there during gametogenesis. A search for candidates for the epigenetic mark focused from -T__ 9 the outset on parent-specific DNA methylation, as it was possible to envisage a scenario in which the sperm and egg Ins-2 Igf2 H19 chromosomes could be transmitted with different methyl- ation patterns and that those patterns could be propagated through many somatic cell divisions through the activity of ) DNA methyltransferase, a hemi-methylase that acts at rep- lication forks to reassemble fully methylated DNA. FIG. 2. The enhancer competition model to explain the opposite imprinting of H19 The best candidate for the epigenetic mark at the H19/ and Igf-2(1gf2 on figure)/Ins-2. The genes are indicalted by the boxes, with the hor- izontal arrows indicating the transcribed alleles. The two H193' endoderm enhanc- Igf-2/Ins-2 locus is a region of extensive paternal-specific ers are designated by filled circles. The positions of allele-specific methylation of DNA methylation of the H19 gene and its 5' flank [31, 33].