MEG3 imprinted gene contribution in tumorigenesis



Maternally expressed gene 3 (MEG3) is a maternally expressed imprinted gene representing a large noncoding RNA in which microRNAs (miRNAs) and small nucleolar RNAs are also hosted. It is capable of interacting with cyclic AMP, p53, murine double minute 2 (MDM2) and growth differentiation factor 15 (GDF15) playing a role in cell proliferation control. MEG3 expression is under epigenetic control, and aberrant CpG methylation has been observed in several types of cancer. Moreover, gene copy number loss has been reported as additional mechanism associated with tumorigenesis. MEG3 deletion seems to upregulate the paternally expressed genes and on the other hand downregulate the expression of downstream maternally expressed genes and tumor suppressor miRNAs, although there are conflicting data on the topic. MEG3 could represent a tumor suppressor gene located in chromosome 14q32 and its association with tumorigenesis is growing every day.

Genomic imprinting refers to an epigenetic modification of a specific parental chromosome, imparted early during gametogenesis that results in the differential expression of the two alleles of a gene in the somatic cells of the offspring.1 Differential expression can occur in all cells, or in specific tissues or developmental stages. Several theories have been proposed to explain the adaptive advantage of imprinting evolution in placental mammals including the ovarian time bomb hypothesis, the complementation hypothesis, the host defense hypothesis and the parental conflict hypothesis. It has also been postulated that imprinting evolved as a defense against parthenogenesis and the associated developmental disorders that would result from the increased homozygosity of recessive mutations.2

Imprinted alleles distinguish their parent of origin by parent-specific epigenetic marks (mainly methylation of CpG nucleotides) often consisting of a single differentially marked cis element, termed imprint control region (ICR).3 Aberrant DNA methylation of imprinted genes results in loss of imprinting (LOI), that is, loss of the normal allele-specific gene expression and has been associated with cancer, imprinting-related diseases and psychiatric disorders.4, 5 Imprinted genes are not randomly distributed in the genome but they tend to occur in clusters suggesting that the primary control of imprinting is not at the single gene level but at the chromosome level.6 Large noncoding RNAs (ncRNAs) have been found at all imprinted clusters, are almost always expressed from the parental chromosome with the unmethylated ICR and have a role in silencing imprinted genes. In both the insulin growth factor 2 (IGF2)/H19 and the DLK1/MEG3 clusters, the ICR lies several kilobases upstream of the ncRNA promoter, and the ncRNA is expressed from the maternally derived chromosome.7 Other types of ncRNAs found at imprinted clusters include the small nucleolar RNAs (snoRNAs) and the microRNAs (miRNAs) which in the DLK1/MEG3 cluster are downstream maternally expressed genes and include various tumor suppressor miRNAs. SnoRNAs are approximately 80–300 nucleotide long, and miRNAs are around 22 nucleotide long. These, like the longer ncRNAs, are also expressed from the parental chromosome with the unmethylated ICR and may arise from processing of long ncRNAs.7 More than 80 genes are currently known to be imprinted, and this number continues to increase. A comprehensive list of mammalian imprinted ncRNAs is currently described in the ncRNAimprint database (

Maternally expressed gene-MEG3 (also known as gene trap locus 2 (GTL2)) is reciprocally imprinted with the paternally expressed gene DLK1 constituting an imprinting domain on human chromosome 14q32 and on mouse chromosome 12 (Meg3). MEG3 represents a large ncRNA as its transcript lacks a significant open reading frame, is positioned ∼100 kb from the protein encoding DLK1, its promoter is differentially methylated and is transcribed in the same orientation with DLK1.9, 10 This specific imprinted domain contains two differentially methylated regions (DMRs), which is the germline-derived primary DLK1-MEG3 intergenic DMR (IG-DMR) and the postfertilization-derived secondary MEG3-DMR, respectively.11 It also seems that the methylation pattern of the MEG3-DMR depends on that of the IG-DMR suggesting a hierarchical interaction and distinct properties of both DMRs in imprinting control.11 MEG3 is expressed during development with higher levels in the paraxial mesoderm, the developing central nervous system, and the epithelia of salivary glands, pancreas and kidney. It is also expressed in adult mouse adrenal and pituitary glands and brain.9 The parent-of-origin monoallelic expression pattern is related to the methylation profile of the ICR, and subjects presenting uniparental disomies (upd) result in distinct phenotypes ranging from almost a normal phenotype to facial abnormalities, distorted fingers, small thorax, scoliosis, abdominal wall defects, esophageal atresias and polyhydramnios or lethal abnormalities in the case of upd(14)pat, while upd(14)mat leads to prenatal and postnatal growth failure, present increased susceptibility to sporadic cancers and to cancer-prone diseases.12–14 Single nucleotide polymorphisms lying within intron 6 of MEG3 increases the susceptibility to type 1 diabetes by direct expression modulation of the paternally inherited DLK1 gene.15 Aberrant methylation pattern of the MEG3 gene and consequent altered expression profile of the imprinted gene may also result in spontaneous pregnancy loss.16

The functional properties of MEG3 and its involvement in physiological and pathologic processes of cell biology, is under continuous investigation. MEG3 has tumor suppressor properties and exerts its antiproliferative function by suppressing MDM2 expression with consequent increase of the p53 protein levels which in turn stimulates GDF15 expression in human cancer cells. MEG3 transcription and not MEG3 translation is required for p53 activation, and its activation requires full length MEG3 transcripts while is not achieved by accumulation of random MEG3 RNA transcripts. However, MEG3 is capable of inhibition of cell proliferation even in the absence of p53.17 Another possible mechanism through which MEG3 exerts its function is represented by the cyclic AMP (cAMP) response element (CRE), which seems to be an important positive regulatory element for MEG3 expression and resides in its proximal promoter between −69 and −49 sequences.18 It is probable that promoter hypermethylation blocks the binding of Cyclic-AMP response element binding protein (CREB) family to CRE and consequently blocks MEG3 transcription initiation. It has also been reported that cAMP stimulates MEG3 expression through the CRE binding site and plays a role in the cAMP-mediated physiological function in endocrine tissues and also in the mitogen-activated protein kinase (MAPK) signaling pathway inhibiting cell proliferation and growth.18 MEG3 is also capable of suppressing cell proliferation by direct regulation of Rb phosphorylation or indirectly by activation of Rb positive regulator p16INK4A.19 Nevertheless, it should be clear that MEG3 functional properties is a working model in progress and the above-described mechanisms, such as the p53 target GDF15 or the MEG3 upstream activator cAMP, probably represent some of the mechanisms existing. Notably, in a Meg3 knockout mice model, the Notch, Wnt, GnRH, vascular endothelial growth factor (VEGF), adherens junctions and calcium signaling pathways were enriched in the brain of Meg3-null mice embryos defining the specific gene's role in various important pathways in cell physiology. Five genes involved in the VEGF pathway also presented increased expression in Meg3-null embryos and were associated with increased brain blood vessel development highlighting a potential tumor suppressive role of Meg3 in controlling neoangiogenesis which characterizes tumorigenesis.20 Changes in the Notch signaling pathway were also observed suggesting that Meg3 is involved in different biological processes that are under the Notch signaling control.20 The known interactions between MEG3, its upstream regulators and the downstream involved signaling pathways are schematically represented in Figure 1. Recently, 12 MEG3 cDNA isoforms have been identified variably distributed in human tissues. MEG3 isoforms are capable of stimulating p53 expression and suppressing cell growth in HCT116 cells at different rates among various isoforms. It seems that the folding structure of MEG3 RNA is more important than its primary sequence for its function.21 As there is growing evidence of MEG3 involvement in cell biology pathways and in cancer, we sought to summarize the current knowledge on MEG3 contribution in tumorigenesis.

Figure 1.

Cyclic AMP (cAMP) induces MEG3 expression which in turn activates both the p53 and Rb pathways. The p53 pathway is activated directly or indirectly (via MDM2 suppression) and consequently activation of its downstream target GDF15 with known tumor suppressive function. The Rb pathway is activated independently of p53, directly by regulating Rb phosphorylation or indirectly by the positive regulation of p16INK4a. Final result is the cell proliferation inhibition and tumor suppression. MEG3 is also associated with the VEGF and Notch signaling pathways regulating various biological processes like angiogenesis, cell differentiation, proliferation and survival.

MEG3 Modulation in Neoplasia

The first evidence of MEG3 involvement in tumorigenesis was established by Zhang et al. in 2003.22 They identified MEG3 isoform (characterized by the presence of an additional exon) cDNA fragments and MEG3 mRNA being expressed normally or highly in normal pituitary tissue. Same fragments and mRNA expression was lower or absent in clinically nonfunctioning pituitary tumors and in several different human carcinoma cell lines. Re-expression of the imprinted gene in cancer cells was associated with inhibition of cell growth and proliferation conferring the gene a tumor growth suppressor function.22 The same group established that loss of heterozigosity, homozygous deletions or point mutations were not the responsible mechanisms for the decreased MEG3 expression in human clinically nonfunctioning pituitary tumors, but the primary pathogenetic mechanism of MEG3 silencing was promoter CpG hypermethylation especially in the functionally important sequences contained in regions 1 and 4 in the 5′-flanking region of the gene.23 The lateral wings of the anterior pituitary were found to present stronger MEG3 mRNA expression than the other pituitary areas, and mRNA expression was also detected in gonadotrophs, somatotrophs, lactotrophs, thyrotrophs and corticotrophs with growth hormone (GH)-producing cells presenting the strongest and most diffuse MEG3 expression.24 MEG3 expression was detected in secreting tumors but not in clinically nonfunctioning tumors. Notably, DNA methylation of the imprinting control region IG-DMR was significantly higher in nonfunctioning tumors than in secreting tumors and normal pituitary tissues.24

MEG3 possesses cell growth suppression function in human meninges and its mRNA was also detected in normal human meninges but not in the majority of human meningiomas and such decreased mRNA expression was more pronounced in more aggressive cases. Gene copy number loss at 14q32 was observed in grade 2 and grade 3 meningiomas, and MEG3 copy number loss was associated with the absence of mRNA expression. Methylation analysis in the promoter, enhancer and IG-DMR region of the gene revealed an increased rate of CpG methylation in these regions of MEG3 in tumor samples compared to normal human meningeal samples, and the degree of methylation correlated with tumor grade.25

Cigarette smoke condensate treated human bronchial epithelial cells presented downregulation of expression of several genes including the imprinted gene MEG3, compared to control samples. It was also indirectly demonstrated that decreased expression of the MEG3 gene, among others, was attributed to altered methylation levels, as MEG3 time-dependent re-expression was achieved after azacytidine administration. The authors concluded that cigarette smoking might affect the epigenetic machinery of human bronchial epithelial cells and could induce promoter hypermethylation of several imprinted genes including MEG3.26

Meg3 expression resulted in downregulation in a mouse embryonic fibroblast model in which the six imprinted genes studied were severely and coordinately downregulated in contrast with nonimprinted genes which were stably expressed during cell proliferation under stress and during tumorigenesis. Although it was not demonstrated that decreased expression of Meg3 was associated with aberrant CpG methylation, as it was for the Cdkn1c and H19 genes, given that silencing affects the bulk of the cell population over the first cell duplication excludes other mechanisms like genetic modifications followed by a selection process.27

Abnormal methylation profile of a CpG island upstream of MEG3 was observed in renal cell carcinoma tissues and correlated with the expression of the homolog-imprinted gene DLK1. In fact, CpG hypermethylation was consistently observed in renal cell carcinoma tissues with and without 14q32 loss of heterozygosity and was associated with absence of the expression of DLK1 in 78% of the cases. MEG3 expression was negative not only in hypermethylated renal cancer cases but also in normal tissues tested, and azacytidine treatment did not restore MEG3 expression in renal carcinoma cell lines indicating that methylation status of this CpG island does not affect transcription of MEG3. These data suggested that hypermethylation of the specific CpG island upstream of MEG3 is critical for the inactivation of DLK1 in the particular type of tumor.28

Dlk1/Meg3 was also upregulated in pleiomorphic adenoma gene 1 (PLAG1) transgenic mice models studying PLAG1 involvement in the pathogenesis of adenomyoepitheliomatous lesions and salivary tumor pathogenesis. Igf2 was inactivated proposing molecular pathways other than Igf signaling to be involved in oncogenesis, indicating an indirect involvement of Meg3, mediated by PLAG1, in cancer.29–31

We have studied the methylation status of MEG3 IG-DMR that encompasses the gene promoter in patients with multiple myeloma (MM), acute myeloid leukemia (AML) and myelodysplastic syndromes (MDS). In particular, we observed aberrant methylation pattern in 57% of MM patients, and there was an association between methylation profile and disease subtype and disease stage. Moreover, abnormal methylation was observed in 34.9% of MDS patients and 47.6% AML patients. There was no significant association among methylation status and karyotype, WHO disease subtype and WHO-based prognostic scoring system. However, AML patients harboring aberrant MEG3 promoter methylation had a reduced overall survival (hazard ratio (HR) = 1.98, p = 0.047), and MDS patients had a trend for reduced overall survival (HR = 2.15, p = 0.072) as determined by Kaplan–Meier estimates. Interestingly, patients with AML after MDS presented MEG3 hypermethylation in 50% of the cases compared to the 34.9% of MDS patients suggesting that MEG3 aberrant methylation may be accentuated during disease progression from MDS to AML.32, 33 MEG3 mRNA levels were also quite variable and not significant in AML patient samples and cell lines associated with both monoalleleic and biallelic expression of the paternally expressed DLK1, but DLK1 transcription was established not to be under MEG3 control. Quantitative methylation analysis of three CpG-rich regions upstream or within MEG3 demonstrated hypermethylation but mostly hemimethylation for the region located upstream of MEG3, while the region that crossed the MEG3 start site was slightly hypomethylated with large stretches of hemimethylation. The third region located within the MEG3 gene showed relative hypermethylation, but the statistic analysis did not reveal any significant difference between MEG3 methylation profile and monoallelic or biallelic DLK1 in AML patients expression for any of the CpG regions evaluated.34

Astuti et al. studied the MEG3 promoter methylation profile in neuroblastoma, pheochromocytoma and Wilms' tumors and observed aberrant methylation status in 25%, 10% and 2.5% of the cases, respectively, which correlated also with suppressed MEG RNA expression. MEG3 IG-DMR hypermethylation was observed in neuroblastoma cases showing almost complete CpG methylation. Hypermethylation of the MEG3 promoter in a neuroblastoma cell line was associated with MEG3 silencing and upregulation of DLK1 expression, but no LOI of DLK1 was observed. This association of MEG3 hypermethylation and DLK1 upregulation without LOI is consistent with a model in which MEG3 regulates DLK1 transcription, also providing new insights into the imprinting control although such inverse correlation in their expression might be purely coincidental in oncogenesis.35 MEG3 gene, using pyrosequencing and microarray-based methylation methods, was studied in Wilms' tumour samples by others, but no significant differences in allele-specific expression of the gene were found between groups with or without IGF2 LOI.36

B-cell malignancies harboring a deletion of chromosome 14q involving the immunoglobulin heavy chain locus del(14q/IGH) lose the DLK1/MEG3 domain leading to inactivation and monoallelic expression of the tumor suppressor gene.37 The data obtained in that study did not support a systematic allele loss in tumors but a random choice of deleted chromosome. The methylation status of IG-DMR of the nondeleted allele was altered compared to the corresponding homolog in disomic cells, with an increase of methylation rates in tumors in which the maternal allele was retained, and a decrease in methylation in case of paternal allele retention. On the other hand, methylation profile was identical between tumors with either paternal or maternal allele deletion and controls. The findings reported suggested that the DLK1/MEG3 imprinted domain is not directly involved as a tumor suppressor gene in B-cell malignancies.37

Mice treated with the drug DDC, which is known to induce epigenetic modifications, resulted in formation of Mallory–Denk bodies in hepatocytes which are found in a variety of liver diseases including hepatocellular carcinoma. These mice fed with DCC presented decreased expression of Meg3 and increased expression of H19 and air imprinted genes. Although a possible mechanism for their altered expression was not demonstrated in that report, it was assumed that only global methylation changes could result in altered imprinted gene expression, and that their deregulated expression could initiate tumor formation.38

Kim et al. observed that MEG3 presented variable allelic expression in different human embryonic stem cell (hESC) lines and some lines presented fully methylated or hemimethylated CpG islands of the CCCTC-binding factor (CTCF) binding protein that did not correlate with interline allelic expression differences. However, a second CpG island was fully hypermethylated in all hESC lines and was also associated with monoallelic expression suggesting that CpG methylation does not regulate the monoallelic expression of this particular imprinted gene but it plays a role in more differentiated tissues. Authors concluded that because of the established role of imprinted genes in cell differentiation, proliferation and tumorigenesis, routine screening of cell lines would be wise given the probability that imprinting disruption could result in phenotypic abnormalities including cancer.39 Studies reporting on MEG3 involvement in tumorigenesis are described in Table 1.


It has long been proposed that chromosome 14q32 represents a locus in which resides a tumor suppressor gene whose allelic loss is implicated in the pathogenesis of solid and hematological malignancies.40, 41 The studies reported here support the notion that MEG3 could represent the tumor suppressor gene in discussion mainly because: (i) it is located in the 14q32 locus, (ii) it is expressed in several types of normal tissues but its expression is decreased/absent in malignant tissues, (iii) gene copy number loss and CpG methylation result in MEG3 silencing, (iv) MEG3 expression has been associated with tumor cell growth suppression, and finally (v) MEG3 interacts with the p53/MDM2 autoregulatory loop thus keeping cell proliferation under control. As an imprinted gene MEG3 is more sensitive to epigenetic inactivation than the remaining genes in the genome27 and an overall hypermethylation in specific MEG3 regions might result in permanent gene transcriptional silencing, consequent loss of its antiproliferative function contributing to oncogenesis. Recently, it has been demonstrated that the maternally inherited deletion of Meg3 resulted in increased IG-DMR methylation gained during embryogenesis, which in turn caused methylation of the remaining Meg3-DMR and further repressed expression of maternally expressed genes. Moreover, it is probable that the Meg3 promoter truncation could abolish the expression of the maternally expressed genes located downstream of Meg3, including Mirg, Rian, anti-Rtl1, miR-127 and miR-410 while it could activate the expression of the paternally expressed genes Dlk1, Rtl1 and Dio3.42 These data indicated that activation of Meg3 and its downstream maternally expressed genes including some tumor suppressive miRNAs plays a pivotal role in regulating Dlk1/Meg3 imprinting probably by maintaining active status of IG-DMR. In fact, MEG3 and DLK1 genes are in tight association and alteration in the expression of either gene may influence/modulate the expression of the imprinted neighbor altering the susceptibility of diseases including cancer,17, 35 although there conflicting data on their association.34 Moreover, MEG3 expression could be upregulated by other genes like PLAG1 but no mechanism for such deregulation has been reported. Several imprinted miRNA genes are located at the DLK1/MEG3 imprinted domain in two main clusters: one containing five to six miRNAs genes overlapping the RTL1 gene in the opposite orientation, and a second one containing more than 40 miRNAs genes (one of the larger miRNAs cluster genes in the genome) mapped downstream of the C/D RNA gene cluster.43, 44 One of the most important miRNAs cluster hosted in MEG3 is downregulated in various human cancer tissues acting as a tumor suppressor itself.45 Genetic and epigenetic changes are hallmarks of cancer, and MEG3 is submitted to these changes in both solid and hematological tumors. The above-mentioned findings render MEG3 a very important tumor suppressor gene as its reduced expression is characteristic in some cancers and correlates with overall prognosis. It is also capable of modulating paternally and maternally expressed genes situated downstream of Meg3 and more importantly maintains the homeostasis of Dlk1/Meg3 imprinted domain. Such homeostasis is disrupted in solid and hematological malignancies. MEG3 also hosts miRNAs with an established role in tumorigenesis whose expression seems to be under MEG3 control. Moreover, it is possible that MEG3 interacts with the MAPK signaling pathway thus inhibiting cell proliferation and growth. The importance of MEG3 in cellular biology and its involvement in pathways with an established role in tumorigenesis is under continuous investigation.1

Table 1. Studies reporting MEG3 involvement in various neoplasms
inline image