Molecular mechanisms leading to the phenotypic development in paternal and maternal uniparental disomy for chromosome 14.

Human chromosome 14q32.2 carries a cluster of imprinted genes. They include paternally expressed genes (PEGs) such as DLK1 and RTL1, and maternally expressed genes (MEGs) such as GTL2 (alias, MEG3), RTL1as (RTL1 antisense), and MEG8. Consistent with this, paternal and maternal uniparental disomies for chromosome 14 (upd(14)pat and upd(14)mat) cause distinct phenotypes. In this review, we summarize the current knowledge about the underlying factors for the development of upd(14)pat and upd(14)mat phenotypes. The data available suggest that the DLK1-GTL2 intergenic differentially methylated region (IG-DMR) plays an important role in the maternal to paternal epigenotypic switch, and that excessive RTL1 expression and decreased DLK1 and RTL1 expression play a major role in the development of upd(14)pat-like and upd(14)mat-like phenotypes, respectively.

Introduction

Human chromosome 14q32.2 carries a cluster of imprinted genes including paternally expressed genes (PEGs) such as DLK1 and RTL1, and maternally expressed genes (MEGs) such as GTL2 (alias, MEG3), RTL1as (RTL1 antisense), and MEG8 (Fig. 1) (1, 2). The parent-of-origin specific monoallelic expression patterns are tightly related to the methylation status of differentially methylated regions (DMRs) (3). For the 14q32.2 imprinted region, the DLK1-GTL2 intergenic DMR (IG-DMR) and the GTL2-DMR are extensively hypermethylated after paternal transmission and grossly hypomethylated after maternal transmission (4,5,6,7).

Fig. 1.

The human chromosome 14q32.2 imprinted region and the expression patterns of the imprinted genes. PEGs are shown in blue, MEGs in red, and biparentally expressed genes in black. The IG-DMR and the GTL2-DMR are depicted in green. In the normal status, all the PEGs and MEGs show monoallelic expression patterns, in association with the parental origin specific methylation patterns of the DMRs. In upd(14)pat, PEGs are expressed from both chromosomes, with severely increased RTL1 expression because of the absence of the functional RTL1as. In upd(14)mat, MEGs are expressed from both chromosomes.

Consistent with these findings, both paternal and maternal uniparental disomies for chromosome 14 (upd(14)pat and upd(14)mat) cause distinct phenotypes. Upd(14)pat results in a unique phenotype characterized by facial abnormality, small bell-shaped thorax, abdominal wall defects, and polyhydramnios (7). In particular, the bell-shaped thorax is pathognomonic and may be lethal or requires long-term mechanical ventilation, and polyhydramnios usually needs repeated amniocentesis before 30 wk of gestation. Upd(14)mat leads to clinically discernible features such as pre- and postnatal growth failure, hypotonia, mild facial abnormalities, small hands, and early onset of puberty (7). The phenotypic spectrum is wide and ranges from nearly normal phenotype to severe phenotype reminiscent to that of Prader-Willi syndrome (8, 9). Furthermore, in agreement with the pivotal role of imprinted genes in placental growth and development (10, 11), upd(14)pat is associated with placentomegaly (12), while placental size has not been examined in upd(14)mat.

Such phenotypic development is ascribed to perturbed expression of imprinted genes, i.e., increased expression of PEGs and absent expression of MEGs in upd(14)pat and increased expression of MEGs and absent expression of PEGs in upd(14)mat. In this regard, we have recently proposed the major factors for the development of upd(14)pat/mat phenotypes, on the basis of (epi)genotype-phenotypes correlations in a total of 12 patients with microdeletions and epimutations affecting the imprinted region as well as the mouse data available (7, 13). Here, we first summarize the current knowledge about the mouse homologous region and subsequently explain the underlying mechanisms leading to the development of human upd(14)pat/mat phenotypes. We also refer to the placental data in comparison with the body data.

Mouse Data

Uniparental disomy for chromosome 12

The human 14q32.2 imprinted region is highly conserved on the distal part of the mouse chromosome 12 (14). The imprinted genes on the distal chromosome 12 show monoallelic expression in both the normal embryos and placentas (10, 14). Thus, paternal uniparental disomy for chromosome 12 (PatDi(12)) results in distinct clinical phenotype such as prenatal lethality, cartilage defects, abdominal distension, and placentomegaly, whereas MatDi(12) leads to characteristic phenotype such as perinatal lethality, growth failure (~60%), and placental hypoplasia (15). Since these clinical features are grossly reminiscent of those of human upd(14)pat/mat, this suggests the involvement of similar (epi)genetic mechanisms in both human and mouse disomies for the conserved imprinted region.

Targeted deletion of the IG-DMR (ΔIG-DMR)

This experiment has shown that the germ-line derived IG-DMR functions as a cis-acting regulator for the imprinted region of maternal origin in the bodies (5, 16). Namely, ΔIG-DMR causes paternalization of a maternally derived imprinted region and a unique phenotype comparable to that of PatDi(12) in embryos, with ~4.5 times of Rtl1 expression and ~2 times of Dlk1 and Dio3 expression as well as nearly absent Megs expression (5, 15, 16). The marked Rtl1 expression is ascribed to a synergic effect between activation of the usually silent maternally derived Rtl1 and loss of functional Rtl1as as a repressor for Rtl1 (5, 17). The doubled Dlk1 and Dio3 expression is simply due to the activation of Pegs of maternal origin (5). The absent Megs expression is associated with hypermethylation of the Gtl2-DMR (5), consistent with the notion that methylation pattern of the Gtl2-DMR is established after fertilization depending on the methylation pattern of the IG-DMR (18). By contrast, the ΔIG-DMR has no imprinting or clinical effect after paternal transmission (5).

Knockout mouse experiments

Knockout mouse experiments have successfully been performed for Dlk1, Rtl1/Rtl1as, and Dio3. The Dlk1 mutation causes several upd(14)mat-like features such as pre- and postnatal growth deficiency (~80%), obesity, and facial abnormalities only after paternal transmission (19). The paternally inherited Rtl1 deletion results in mild growth deficiency (~80%) and perinatal lethality (20), and the maternally derived Rtl1as deletion leads to placentomegaly and dilated fetal capillaries in the labyrinth zone in association with 2.5–3.0 times of Rtl1 expression (20). Dio3 knockout mice show reduced enzyme activities and some phenotypic effects after paternal transmission (21). Furthermore, Gtl2 mice with dysregulated imprinting status caused by a transgene insertion have a normal phenotype with at least 60–80% reduction of all the Megs (22).

Placental analysis

Placental analyses have revealed different expression patterns and phenotypes between the PatDi(12) mice and the ΔIG-DMR mice. While mice with PatDi(12) have placentomegaly (15), those with maternally derived ΔIG-DMR have normal placentas with mildly increased Pegs expression and considerably preserved Megs expression (16). In addition, while the IG-DMR methylation pattern is comparable between the normal mouse embryo and placenta, the Gtl2-DMR in the embryo does not behave as a DMR in the placenta, with the ratio of methylated to unmethylated clones being ~50%:50% after paternal transmission and ~25%:75% after maternal transmission (16). These findings suggest the differential imprinting control and resultant phenotypic consequences between the embryos and placentas with ΔIG-DMR.

Human Data

Identification of the IG-DMR and the GTL2-DMR

We and other investigators have identified the human IG-DMR (5, 7). In particular, we found two regions with the property of the IG-DMR, and designated CG4 and CG6. We have also identified the human GTL2-DMR, and designated CG7; here, although the GTL2-DMR has been reported (6), CG7 is the first region that was confirmed to be the GTL2-DMR by bisulfite sequencing (7).

Proposed hypothesis

We have proposed that the IG-DMR plays an important role in the maternal to paternal epigenotypic switch, and that excessive RTL1 expression and decreased DLK1 and RTL1 expression play a major role in the development of upd(14)pat-like and upd(14)mat-like phenotypes, respectively. This hypothesis assumes that the functions of the IG-DMR and the imprinted genes within this domain are primarily similar between the human and the mouse (5, 7). We present here how this hypothesis can explain the development of upd(14)pat/mat-like phenotypes in non-disomic patients. We do not refer to DIO3 and MEGs other than RTL1as; although the relevance of DIO3 and the total absence of MEGs still remains tenable at this time, upd(14)pat/mat patients are apparently free from thyroid dysfunction (7), despite the primary function of DIO3 being thyroid hormone metabolism (21), and biological functions remains totally unknown for MEGs other than RTL1as.

Familial microdeletions (Family A)

(Fig. 2)

Fig. 2.

Family A. a. The pedigree. Cases III-1 (deceased) and III-3 have typical upd(14)pat phenotype, and cases I-3 and II-2 exhibit upd(14)mat-like phenotype. b. Methylation analysis of the DMRs. The IG-DMR (CG4 and CG6) and the GTL2-DMR (CG7) are severely hypermethylated in case III-3 and grossly hypomethylated in cases I-3 and II-2. For the IG-DMR, each lane indicates a single clone, and each circle denotes a CpG island; filled and open circles represent methylated and unmethylated cytosines, respectively. For the GTL2-DMR (CG7), methylated allele-specific primers (M) and unmethylated allele-specific primers (U) have been utilized. c. FISH analysis for the IG-DMR. Heterozygous microdeletion is identified in cases III-3, II-2, and I-3. d. Direct sequencing for a long and accurate (LA)-PCR product, demonstrating a ~109 kb deletion in cases III-1, II-2, and I-3. The predicted gene dosages are indicated on the right part.

This unique three-generation family contains two sibs (III-1 and III-3) with typical upd(14)pat phenotype and the mother (II-2) and the maternal grandfather (I-3) with upd(14)mat-like phenotype including mild short stature (–2.2 SD in the mother and –2.9 SD in the grandfather) (7). Methylation analysis showed hypermethylated DMRs in case III-3 and hypomethylated DMRs in cases II-2 and I-3. Deletion analysis revealed ~109 kb deletion involving DLK1, the IG-DMR, the GTL2-DMR, and GTL2 in cases with upd(14)pat/mat-like phenotypes. Thus, the deletion has caused typical upd(14)pat phenotype after maternal transmission and upd(14)mat-like phenotype after paternal transmission.

The results are well explained by the above notion. In the two sibs (III-1 and III-3) with typical upd(14)pat phenotype, since the loss of IG-DMR is derived from the mother, this would have caused paternalization of the imprinted domain, resulting in the expression of PEGs from both chromosomes. However, since DLK1 is deleted from the maternally inherited chromosome, DLK1 should be present in a single active copy, as in normal individuals. By contrast, since RTL1 is present in two copies in the absence of functional RTL1as, the expression dosage of RTL1 should be increased markedly (4–5 times), as in upd(14)pat patients. Thus, it is likely that severely increased RTL1 dosage has played a critical role in the development of typical upd(14)pat phenotype.

In the mother (II-2) and the maternal grandfather (I-3) with upd(14)mat-like phenotype, since the loss of IG-DMR is of paternal origin, this would have no effect on the imprinting status. Thus, the upd(14)mat-like phenotype would simply be ascribed to the loss of DLK1 from the paternally derived chromosome.

Familial microdeletions (family B)

(Fig. 3)

Fig. 3.

Family B. e. The pedigree. Cases III-1 show upd(14)pat-like phenotype, and cases I-3 and II-2 exhibit upd(14)mat-like phenotype. f. Methylation analysis of the DMRs. The IG-DMR (CG4 and CG6) and the GTL2-DMR (CG7) are severely hypermethylated in case III-3 and grossly hypomethylated in case II-2. g. FISH analysis for the IG-DMR. Heterozygous microdeletion is identified in cases III-1 and II-2. h. Direct sequencing for an LA-PCR product, demonstrating a ~411 kb deletion in cases III-1 and II-2. The predicted gene dosages are indicated on the right part.

This two-generation family contains the daughter (III-1) with relatively mild upd(14)pat-like phenotype in terms of the duration of respiratory duration, abdominal defects, and the degree of polyhydramnios, and the mother (II-2) with upd(14)mat-like phenotype including severe short stature (–4.4 SD) (7). Methylation analysis showed hypermethylated DMRs in case III-1 and hypomethylated DMRs in case II-2. Deletion analysis revealed a ~411 kb deletion involving WDR25, BEGAIN, DLK1, the IG-DMR, the GTL2-DMR, GTL2, RTL1, RTL1as, and MEG8 in cases with upd(14)pat/mat-like phenotypes. Thus, the deletion has caused relatively mild upd(14)pat phenotype after maternal transmission and upd(14)mat-like phenotype with severe short stature after paternal transmission.

The results are similarly explained by the above notion. In case III-1, loss of IG-DMR from the maternally derived chromosome would have caused paternalization of the imprinted domain. However, since DLK1 is deleted from the maternally inherited chromosome, DLK1 should be present in a single active copy, as in normal individuals. By contrast, while RTL1 is also present in a single copy, the expression dosage of RTL1 should be 2.5–3.0 times higher than the normal individuals because of the absence of functional RTL1as. Thus, it appears that moderately increased RTL1 dosage is essential for the development of relatively mild upd(14)pat phenotype.

In the mother (II-2), there should be no alteration of the imprinting status because of the loss of IG-DMR from the paternally derived chromosome. Thus, the upd(14)mat-like phenotype with severe short stature would simply be ascribed to the loss of DLK1 and RTL1 from the paternally derived chromosome.

Sporadic microdeletions

We have also identified two deletions of maternal origin, i.e., a ~475 kb deletion involving the IG-DMR, the GTL2-DMR, GTL2, RTL1, RTL1as, and MEG8 and a ~6.5 Mb deletion involving the whole imprinted region in patients with relatively mild upd(14)pat-like phenotype (7). This would also be explained by the moderately increased RTL1 dosage, as in the case III-1 of family B.

Epimutations (hypermethylated DMRs)

We have identified three patients with typical upd(14)pat phenotype and hypermethylated DMRs (epimutations) (7). In particular, genotyping analysis for a SNP within the IG-DMR (CG4) confirmed hypermethylation of the usually hypomethylated DMRs of maternal origin. The results are explained by assuming that the hypermethylation of the maternally inherited IG-DMR has caused paternalization of the imprinted region, as in the loss of IG-DMR of maternal origin. In this case, since the expression pattern of the imprinted domain would be comparable to those in upd(14)pat, with 4–5 times of RTL1 expression dosage, this explains the development of typical upd(14)pat phenotype in these patients.

Epimutations (hypomethylated DMRs)

We have also found a patient with fairly typical upd(14)mat-like phenotype and hypomethylated DMRs (epimutations) (13). The results suggest that the hypomethylation of the paternally inherited IG-DMR has resulted in maternalization of the imprinted region, leading to the development of upd(14)mat-like phenotype.

Since virtually all the imprinted genes studied to date are expressed in the placenta (10, 14), we examined placental samples obtained from case III-3 in family A with the microdeletion and from one case with epimutation (hypermethylation), as well as from a upd(14)pat patient and a nearly gestational age-matched control subject (7). We could also obtain three sets of samples consisting of cDNA and genomic DNA of normal fresh placenta and leukocyte genomic DNA of the mother.

Consequently, we have shown the following: [1] monoallelic paternal DLK1 expression and maternal GTL2 expression in the placentas (the genotyping results were not informative for other imprinted genes); [2] paternalization of the maternally inherited imprinted region with markedly elevated RTL1 expression dosage in case III-3 of family A and one case with epimutation; [3] parental origin dependent differential methylation pattern of the IG-DMR and grossly hypomethylated GTL2-DMR (the results are consistent with the IG-DMR being the germline derived DMR and the GTL2-DMR being the secondary DMR, because the germline derived DMRs are delineated as DMRs in the placentas as well as in the bodies, whereas the secondary DMRs, though they behave as DMRs in the bodies, are rather hypomethylated irrespective of the parental origin in the placentas) (7, 23,24,25); and [4] characteristic histological findings such as proliferation of dilated and congested chorionic villi. These findings imply that the phenotypic development is closely associated with altered expression dosage of the imprinted genes in both the body and the placenta, and that the epigenetic control is different between the bodies and the placentas in the human as well as in the mouse and between the human and the mouse placentas with maternally derived deletion of the IG-DMR.

Perspectives

Despite the above progress, many matters remain to be clarified. They include: the precise mechanisms involved in the imprinting regulation, the clinical and molecular consequences caused by the GTL2-DMR deletion alone, the repressor function of RTL1as for RTL1 in the human, the biological functions of most MEGs/Megs, the presence or absence of multiple microRNAs in the human imprinted region, the imprinting status of human DIO3, the mechanisms leading to epimutations, and the mechanisms involved in the placental imprinting regulation. These matters await further investigations.