IGF2 is parentally imprinted during human embryogenesis and in the Beckwith-Wiedemann syndrome.

The phenomenon of parental imprinting involves the preferential expression of one parental allele of a subset of chromosomal genes and has so far only been documented in the mouse. We show here, by exploiting sequence polymorphisms in exon nine of the human insulin-like growth factor 2 (IGF2) gene, that only the paternally-inherited allele is active in embryonic and extra-embryonic cells from first trimester pregnancies. In addition, only the paternal allele is expressed in tissues from a patient who suffered from Beckwith-Wiedemann syndrome. Thus the parental imprinting of IGF2 appears to be evolutionarily conserved from mouse to man and has implications for the generation of the Beckwith-Wiedemann syndrome.

Loss of imprinting in normal tissue of colorectal cancer patients with microsatellite instability.

Loss of imprinting (LOI) is an epigenetic alteration of some cancers involving loss of parental origin-specific expression of imprinted genes. We observed LOI of the insulin-like growth factor-II gene in twelve of twenty-seven informative colorectal cancer patients (44%), as well as in the matched normal colonic mucosa of the patients with LOI in their cancers, and in peripheral blood samples of four patients. Ten of eleven cancers (91%) with microsatellite instability showed LOI, compared with only two of sixteen tumors (12%) without microsatellite instability (P < 0.001). Control patients without cancer showed LOI in colonic mucosa of only two of sixteen cases (12%, P < 0.001) and two of fifteen blood samples (13%, P < 0.001). These data suggest that LOI in tumor and normal tissue identifies most colorectal cancer patients with microsatellite instability in their tumors, and that LO! may identify an important subset of the population with cancer or at risk of developing cancer.

Functional association of CTCF with the insulator upstream of the H19 gene is parent of origin-specific and methylation-sensitive.

In mammals, a subset of genes inherit gametic marks that establish parent of origin-dependent expression patterns in the soma ([1] and references therein). The currently most extensively studied examples of this phenomenon, termed genomic imprinting, are the physically linked Igf2 (insulin-like growth factor II) and H19 genes, which are expressed mono-allelically from opposite parental alleles [1] [2]. The repressed status of the maternal Igf2 allele is due to cis elements that prevent the H19 enhancers [3] from accessing the Igf2 promoters on the maternal chromosome [4] [5]. A differentially methylated domain (DMD) in the 5' flank of H19 is maintained paternally methylated and maternally unmethylated [6] [7]. We show here by gel-shift and chromatin immunopurification analyses that binding of the highly conserved multivalent factor CTCF ([8] [9] and references therein) to the H19 DMD is methylation-sensitive and parent of origin-dependent. Selectively mutating CTCF-contacting nucleotides, which were identified by methylation interference within the extended binding sites initially revealed by nuclease footprinting, abrogated the H19 DMD enhancer-blocking property. These observations suggest that molecular mechanisms of genomic imprinting may use an unusual ability of CTCF to interact with a diverse spectrum of variant target sites, some of which include CpGs that are responsible for methylation-sensitive CTCF binding in vitro and in vivo.

CpG methylation regulates the Igf2/H19 insulator.

The differentially methylated 5'-flank of the mouse H19 gene unidirectionally regulates the communication between enhancer elements and gene promoters and presumably represses maternal Igf2 expression in vivo [1-6]. The specific activation of the paternally inherited Igf2 allele has been proposed to involve methylation-mediated inactivation of the H19 insulator function during male germline development [1-4, 6]. Here, we addressed the role of methylation by inserting a methylated fragment of the H19-imprinting control region (ICR) into a nonmethylated episomal H19 minigene construct, followed by the transfection of ligation mixture into Hep3B cells. Individual clones were expanded and analyzed for genotype, methylation status, chromatin conformation, and insulator function. The results show that the methylated status of the H19 ICR could be propagated for several passages without spreading into the episomal vector. Moreover, the nuclease hypersensitive sites, which are typical for the maternally inherited H19 ICR allele [1], were absent on the methylated ICR, underscoring the suggestion that the methylation mark dictates parent of origin-specific chromatin conformations [1] that involve CTCF [2]. Finally, the insulator function was strongly attenuated in stably maintained episomes. Collectively, these results provide the first experimental support that the H19 insulator function is regulated by CpG methylation.

The 5' flank of mouse H19 in an unusual chromatin conformation unidirectionally blocks enhancer-promoter communication.

BACKGROUND: During mouse prenatal development, the neighbouring insulin-like growth factor II (Igf2) and H19 loci are expressed monoallelically from the paternal and maternal alleles, respectively. Identical spatiotemporal expression patterns and enhancer deletion experiments show that the Igf2 and H19 genes share a common set of enhancers. Deletion of a differentially methylated region in the 5' flank of the H19 gene partially relieves the repression of the maternal Igf2 and paternal H19 alleles in the soma. The mechanisms underlying the function of the 5' flank of the H19 gene are, however, unknown. RESULTS: Chromatin analysis showed that the 5' flank of the mouse H19 gene contains maternal-specific, multiple nuclease hypersensitive sites that map to linker regions between positioned nucleosomes. These features could be recapitulated in an episomal-based H19 minigene, which was propagated in human somatic cells. Although the 5' flank of the H19 promoter has no intrinsic silencer activity under these conditions, it unidirectionally extinguished promoter-enhancer communications in a position-dependent manner, without directly affecting the enhancer function. CONCLUSIONS: The unmethylated 5' flank of the H19 gene adopts an unusual and maternal-specific chromatin conformation in somatic cells and regulates enhancer-promoter communications, thereby providing an explanation for its role in manifesting the repressed state of the maternally inherited Igf2 allele.

Did genomic imprinting and X chromosome inactivation arise from stochastic expression?

Both X chromosome inactivation and autosomal genomic imprinting generate a functional hemizygosity. Here we consider models that explain the evolution of genomic imprinting and X chromosome inactivation from novel perspectives. Specifically, we suggest that random (in)activation events are common in genes and gene clusters with a low probability of transcription. These generate variability that natural selection has acted on to evolve stable monoallelic expression. Possible selection forces might include a need for dosage compensation and the prevention of biallelic silencing where a total switch off would be lethal. Two different mechanisms can accomplish regular monoallelic expression - genomic imprinting and gene counting.

CTCF is a uniquely versatile transcription regulator linked to epigenetics and disease.

CTCF is an evolutionarily conserved zinc finger (ZF) phosphoprotein that binds through combinatorial use of its 11 ZFs to approximately 50 bp target sites that have remarkable sequence variation. Formation of different CTCF-DNA complexes, some of which are methylation-sensitive, results in distinct functions, including gene activation, repression, silencing and chromatin insulation. Disrupting the spectrum of target specificities by ZF mutations or by abnormal selective methylation of targets is associated with cancer. CTCF emerges, therefore, as a central player in networks linking expression domains with epigenetics and cell growth regulation.

Successive activation of the platelet-derived growth factor beta receptor and platelet-derived growth factor B genes correlates with the genesis of human choriocarcinoma.

The hydatidiform mole is a benign disease of the placenta characterized by the absence of the maternal genome. Approximately 3% of the reported cases will develop into malignant choriocarcinoma. In situ hybridization analysis reveals that the paternal platelet-derived growth factor (PDGF) beta receptor gene is up to 2 orders of magnitude more active in cytotrophoblasts of the complete hydatidiform moles than in normal placentae. The transition between hyperplasia (complete hydatidiform mole) and neoplasia (choriocarcinoma) in these cells correlates with at least a 10- to 20-fold activation of the PDGF-B gene. Since the neoplastic cytotrophoblasts have maintained an abnormally high level of PDGF beta receptor expression, we propose that a deregulated PDGF autostimulatory loop is involved in the genesis of human choriocarcinoma from hydatidiform moles.

Paternally derived H19 is differentially expressed in malignant and nonmalignant trophoblast.

The paternal allele of the H19 gene has been shown to be transcriptionally inactive in the developing human embryo. Using reverse transcription PCR and RNase protection assays, we demonstrate that expression of H19 is predominantly, but not exclusively, from the maternal allele in the human placenta. In situ hybridization analysis shows strong expression of the H19 gene in eight complete hydatidiform moles, hyperplastic tissues consisting of trophoblasts which contain only paternally derived genetic material, indicating that H19 is not functionally imprinted in this tissue. H19, a putative growth suppressor, is oppositely imprinted to the neighboring insulin-like growth factor II (IGF2) gene and an up-regulation of IGF2 expression has been linked previously to a down-regulation of H19 expression in the progression to Wilms' tumor. Two cases of complete hydatidiform mole which progressed to choriocarcinoma show high levels of expression of both H19 and IGF2. The choriocarcinomas which developed from these complete hydatidiform moles showed similar expression of IGF2 but a decreased number of H19-positive cells, which may reflect selection for cells expressing IGF2 and against those expressing H19 in this tissue.

Inactivation of H19, an imprinted and putative tumor repressor gene, is a preneoplastic event during Wilms' tumorigenesis.

Genetic evidence shows that the parent of origin-dependent expression patterns of the Igf2 and H19 genes is coordinated in mouse, such that H19 controls the activity of Igf2 in cis. Equally compelling evidence for a similar situation in humans is absent, although the frequently observed activation of the maternal IGF2 allele (ie., loss of imprinting) in Wilms' tumors has been attributed to the silencing of the maternal H19 locus. We show here that loss of H19 activity is generally a preneoplastic event, which may be linked with an overgrowth lesion that has been proposed to be permissive for tumor formation. Although our results document one instance in which a postneoplastic loss of H19 activity correlates with loss of IGF2 imprinting at the cellular level, it appears that inactivation of H19 is more generally independent of loss of imprinting of IGF2, at least in our specimens. Our results imply that inactivation of H19 correlates with blastema overgrowth and can be independent of a regulatory role with respect to IGF2 imprinting status in cis.

Mosaic allelic insulin-like growth factor 2 expression patterns reveal a link between Wilms' tumorigenesis and epigenetic heterogeneity.

Numerous observations link the loss of imprinting of insulin-like growth factor 2 (IGF2) and an overdosage of this growth factor gene with cancer, in general, and with Wilms' tumorigenesis, in particular. It is not known, however, if loss of imprinting correlates with specific stages of neoplasia or if allelic expression patterns vary within the tumor. By applying an allele-specific in situ hybridization technique to formalin-fixed thin sections, we show that the parental IGF2 alleles can be differentially expressed, not only in Wilms' tumors, but also in nephrogenic rests (which represent premalignant lesions) of Wilms' tumor patients. Moreover, a subpopulation of mesenchymal cells, which surrounds tumor nodules, expresses IGF2 biallelically irrespective of the imprinted state of IGF2 within the tumor. These data show that Wilms' tumorigenesis involves epigenetic heterogeneity as visualized by variable allelic IGF2 expression patterns.

Potential multiple functions of the v-myc oncogene within a single cell clone of OK10 retrovirus-transformed quail fibroblasts.

Propagation of in vitro transformed OK10 QDP 9c cells, cloned in soft agar, results in selective amplification of the OK10 pro-viral genome to yield a 12-fold increase in v-myc oncogene expression. In addition, increased v-myc oncogene dosage correlates with an increase of the proliferative potential of already transformed cells and relieves dependence on high serum concentrations for optimal cell growth. The increased rate of cell proliferation is reflected by a much more rapid progression through all parts of the cell cycle. These results suggest that the attainment of transformed status on one hand and progressive increase in growth factor independence for optimal cell proliferation on the other hand, may be initiated by different myc oncogene dosages within the OK10 QDP 9c cell clone.

Hypervariable allelic expression patterns of the imprinted IGF2 gene in tumor cells.

The IGF2 gene, which encodes a growth factor, is subject to genomic imprinting. The frequently observed loss of IGF2 imprinting in a variety of tumors has been suggested to contribute to neoplasia. Since these reports have not documented the imprinting status of IGF2 at the cellular level, it cannot be excluded that the imprinting status might vary within the tumor. The possibility that loss of IGF2 imprinting in neoplastic cells reflects random imprinting patterns, was therefore addressed. We show here that individual cell populations of the JEG-3 choriocarcinoma cell line display heterogenous imprinting patterns of both IGF2 and H19. In addition, a lack of correlation between IGF2 and H19 imprinting status suggests that any regional parental imprint has been functionally lost. This notion is reinforced by the observation that JEG-3 cell subclones display a range of promoter-specific IGF2 allele usage. Moreover, we observed that the imprinting status of H19 and IGF2 were differentially modulated in JEG-3-derived tumors generated in nude mice. The results suggest that allele-specific expression of IGF2 operates in the absence of a parental imprint. Finally, our observations urge caution with respect to the general interpretation of biallelic expression as 'loss of imprinting'.

An Inr-containing sequence flanking the TATA box of the human c-sis (PDGF-B) proto-oncogene promoter functions in cis as a co-activator for its intronic enhancer.

High-level activity of the human PDGF-B promoter in choriocarcinoma cell lines depends upon an atypical, intronic enhancer-like element which does not function with heterologous promoters tested. An extensive series of mutant PDGF-B promoter-driven constructs identified a sequence flanking the TATA box which is required specifically for enhancer-mediated transcription in human choriocarcinoma cell lines. This element, which we here term an enhancer-dependent cis co-activator (EDC) contains an Inr (initiator) consensus sequence upstream of the TATA box which is required, but not sufficient for its function. Requirement for the EDC is cell type-specific, since it was dispensable for enhancer-mediated transcription in a human breast cancer cell line. Although it lies within the region defined, the TATA box itself is not required for EDC function, or for basal promoter function which may derive from a second Inr-like sequence situated at the transcriptional start site. These observations indicate that interactions between some promoter and enhancer elements may be more complex than that generally described for 'classical' enhancer systems and may suggest an additional function for the initiator motif.

Expression, promoter usage and parental imprinting status of insulin-like growth factor II (IGF2) in human hepatoblastoma: uncoupling of IGF2 and H19 imprinting.

We have studied the promoter utilization and parental imprinting status of human IGF2 in three genetically informative hepatoblastomas from patients ranging in age from 9 months to 3 years. In all three cases, there is a downregulation of promoter P1 in the tumor tissues while the P2 and P3 promoters are upregulated compared to the normal liver. One of three patients displayed loss of imprinting (LOI) of IGF2 in the tumor tissue. We also investigated the expression of the H19 gene in all three cases and the methylation pattern in H19 from the patient with LOI of IGF2. The expression of H19 was greatly reduced in all tumors. Monoallelic H19 expression however, was retained even in the case which showed LOI of IGF2. Unlike the situation in Wilms' tumor, no differences in the methylation pattern between the normal liver and tumor tissues were observed in the H19 promoter or 3' region, using HpII analysis. We show here, that in contrast to the situation in Wilms' tumor, H19 expression is not a prerequisite for maintaining a monoallelic IGF2 expression.

H19 is imprinted in the choroid plexus and leptomeninges of the mouse foetus.

It has been proposed that either the Igf-2 gene or the H19 gene--but not both--can be expressed from a given chromosome. Igf-2 is known to be biallelically expressed in the choroid plexus and leptomeninges, however, raising the question of whether H19 is down-regulated or absent there. We found by in situ hybridization that H19 is indeed expressed in the choroid plexus and leptomeninges of the developing mouse foetus. Comparison with the expression pattern of Igf-2 showed that the genes are coexpressed in all areas, with the exception of the choroid plexus epithelium. To evaluate whether H19 is also biallelically expressed in these tissues, we microdissected embryos from interspecific crosses and performed RNAse protection analysis on the isolated RNA. This revealed that H19 maintains its imprint in the choroid plexus/leptomeninges, being transcribed from the maternal allele at a level comparable to that in normal liver. We discuss the significance of these results for current models of Igf-2 and H19 imprinting.

The non-viability of uniparental mouse conceptuses correlates with the loss of the products of imprinted genes.

Diploid parthenogenetic or androgenetic mouse conceptuses produce characteristic and opposite mutant phenotypes and are non-viable, presumably due to different contributions from the maternal and paternal genomes. This is likely to be the result of the preferential expression of only one parent's copy of certain genes in the offspring. So far, four such endogenous imprinted genes are known: the paternal alleles of Igf2 and Snrpn and the maternal alleles of Igf2r and H19 are active, while their opposite parental alleles are inactive. Here we demonstrate that the expression patterns of the Igf2 and Igf2r genes in androgenetic and parthenogenetic conceptuses correlate with which parental alleles normally express them, implying that the imprint can be maintained in the absence of the other parent's genome for these genes. This also indicates that both types of uniparental conceptuses are lacking developmentally important gene products. We did find, however, that the H19 gene was highly expressed not only in the parthenogenetic conceptus, but also in giant trophoblasts and secondary giant cells in the androgenetic placenta, in spite of the imprinting of the H19 gene in normal mouse extra embryonic tissues. We discuss these observations with respect to the non-viability of uniparental conceptuses and the reciprocal imprinting patterns of the Igf2 and H19 genes.

Expression of a novel member of estrogen response element-binding nuclear receptors is restricted to the early stages of chorion formation during mouse embryogenesis.

Members of the nuclear hormone receptor gene family of transcription factors have been shown to be expressed in characteristic patterns during mouse organogenesis and postnatal development. Using an RT-PCR based screening assay, we have identified nuclear receptors expressed in embryonal carcinoma stem cells. One of the cDNAs characterized, mERR-2, was found to be expressed exclusively during a narrow developmental window in trophoblast progenitor cells between days 6.5 and 7.5 post coitum (p.c.). From 8.5 days p.c. and onwards, the mERR-2 gene activity evaded detection as analysed by in situ hybridization. We also show that the mERR-2 gene product and the estrogen receptor share a common target DNA-sequence recognition specificity unique among members of the gene family. Furthermore, efficient homodimerization and DNA-binding of the orphan receptor mERR-2 was found to be dependent on interaction with the heat shock protein 90, a molecular chaperone hitherto recognized to interact only with the steroid hormone receptor subgroup of nuclear receptors. Based on our results we suggest that the mouse orphan receptor mERR-2 has the potential to regulate overlapping gene networks with the estrogen receptor and may participate in signal transduction pathways during a short developmental period coinciding with the formation of the chorion.

The genotype and epigenotype synergize to diversify the spatial pattern of expression of the imprinted H19 gene.

Little is known of how the genetic background effects the phenomenon of genomic imprinting. The H19 gene belongs to a cluster of imprinted genes on human chromosome 11. Here we show that the alternative splicing of a human H19 transcript is genotype-specific. Moreover, this variant transcript, which lacks exon 4, is either not found at all, is widely expressed or is confined to extra-villous cytotrophoblasts in first trimester placenta, depending on a combination of the genotype and the sex of the transmitting parent.

Normal development and neoplasia: the imprinting connection.

The observation that a number of autosomal genes are expressed in a parent of origin-dependent monoallelic manner has fuelled a frantic research effort into the underlying mechanisms and biological functions of this phenomenon, termed genomic or parental imprinting. The level of intrigue associated with this subject has been heightened by the discovery that the "transcriptional phenotype" of some imprinted genes shows developmental and tissue-specific variation, and that some imprinted genes are expressed biallelically in tumors. Here we describe some further examples of variation in the allele-specific transcription of an imprinted gene, human IGF2. Analysis of different sub-clones of an established tumor cell line (Jeg-3) revealed examples of both a switch from monoallelic to biallelic expression, as well as monoallelic expression from the opposite parental allele. Examination of IGF2 expression in adult human liver clearly demonstrated that the functional imprinting is manifested in a promoter-specific manner. The P1 promoter produced biallelically derived transcripts, whereas the remaining three promoters were utilized in a complex pattern of mono- and biallelic expression which varied from sample to sample. These observations emphasize the need to re-examine the imprinting phenomenon and its plasticity in terms of the cis elements and trans-acting factors involved in the transcriptional regulation of these genes both in the normal and pathological contexts.