Hypomethylation of the MEG8:Int2-DMR in patients with pathogenic PLAG1 variants suggests new role of the chr14q32 imprinting cluster in Silver-Russell syndrome

Abstract

Background

Silver-Russell syndrome (SRS) is a clinically and genetically heterogeneous imprinting disorder. The most common molecular defects are loss of methylation of the H19/IGF2:IG-DMR on chromosome 11p15.5, followed by maternal uniparental disomy of chromosome 7. Further molecular lesions are genetic variants in the PLAG1 oncogene, as well as in the transcription factor HMGA2 and the fetal growth factor IGF2. A phenotypic overlap exists between SRS and Temple syndrome (TS14) that is also characterized by growth restriction but associated with abnormalities in the imprinted chromosome 14q32 gene cluster. In TS14 patients, the germline MEG3/DLK1: IG-DMR is hypomethylated and the MEG8:Int2-DMR gains methylation probably as consequence of transcriptional readthrough from the MEG3 promoter on the paternal chromosome. However, the functional role of the MEG8 DMR remains unknown.

Results

We analysed the DNA methylation of 11–12 imprinted regions in 17 cases with clinical SRS features and heterozygous for a PLAG1 variant. We observed a specific loss of methylation of the MEG8:Int2-DMR in the ten cases carrying pathogenic PLAG1 variants that result in stable aberrant proteins. Normal MEG8 methylation was observed in the cases carrying variants of uncertain pathogenicity or gene deletions. Most of the PLAG1 cases are familial and both epigenetic and genetic defects co-segregated within the families. Additionally, we assessed the methylation status of the MEG8:Int2-DMR in several SRS patients with HMGA2 or IGF2 variants, H19/IGF2:IG-DMR-LoM and upd(7)mat and all of them showed normal methylation.

Conclusions

Our results indicate that pathogenic PLAG1 variants leading to stable aberrant PLAG1 proteins and possibly acting in a dominant-negative manner influence methylation of the MEG8 locus. This study suggests a new pathogenetic mechanism of the PLAG1 gene in SRS, involving imprinted genes in the chr14q32 cluster through deregulation of the MEG8:Int2-DMR and provides an epigenetic signature that may be used to assess the damaging potential of the PLAG1 variants.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13148-025-02024-6.

Background

Imprinting disorders (ImpDis) are a group of congenital diseases caused by deregulation of imprinted genes and affecting human growth, metabolism and behaviour [1]. Most imprinted genes are organized in clusters, where their parent of origin-dependent expression is regulated by cis-acting regions exhibiting differential DNA methylation (Differentially Methylated Regions, DMRs) between the maternal and paternal alleles [2]. Imprinted DMRs are categorized into two groups: primary DMRs, also known as germline DMRs, and secondary DMRs, also referred to as somatic DMRs. Germline DMRs acquire their methylation pattern during gametogenesis and are stably maintained through the epigenetic reprogramming occurring during the oocyte-to-embryo transition. In contrast, somatic DMRs acquire their allele-specific methylation post-fertilization and are regulated by the neighbouring germline DMR in a hierarchical manner [3]. ImpDis are associated with genetic and epigenetic alterations involving changes in both gene sequences (genetic mutations) and gene regulation (epigenetic mutations) in imprinted gene clusters [1]. Epigenetic abnormalities can affect a single or multiple germline DMRs (multi-locus imprinting disturbances, MLID) and can be associated with genetic variants acting in cis or in trans [2, 4].

Silver-Russell syndrome (SRS; OMIM #180,860; prevalence at birth 1:30,000/1:100,000) is a clinically and genetically heterogeneous imprinting disorder, characterised by intrauterine and post-natal growth retardation, relative macrocephaly at birth, feeding difficulties, protruding forehead in early life and body asymmetry, along with numerous additional features at lower frequencies. According to the Netchine-Harbison clinical scoring system (NH-CSS), a clinical diagnosis is considered if a patient scores at least four out of six most frequent criteria [5]. However, the definition of Silver-Russell syndrome spectrum (SRSp) has been proposed to include cases with a clinical score < 4 yet still exhibiting clinical or molecular features of SRS [6]. Despite the complexity of its molecular mechanisms, an underlying molecular cause can currently be identified in around 60% of patients with clinical diagnosis of SRS. The most common molecular alterations are loss of methylation (LoM) at the H19/IGF2:IG (intergenic)-DMR, also known as IC1, on chr11p15.5, accounting for 30–60% of patients and maternal uniparental disomy of chromosome 7 (upd(7)mat) detected in 5–10% of the cases [5]. About 10–20% of SRS cases with IC1-LoM show MLID [7, 8]. Further molecular lesions associated with SRS phenotype are represented by genetic variants in PLAG1 (Pleiomorphic Adenoma Gene 1), IGF2 (Insulin Like Growth Factor 2), HMGA2 (High Mobility Group AT-Hook 2) and more rarely CDKN1C (Cyclin Dependent Kinase Inhibitor 1C) gene [9–12].

A phenotypic overlap exists between SRS and Temple syndrome (TS14) that is also characterized by growth restriction but associated with abnormalities in the imprinted chr14q32 gene cluster (Fig. 1). This evolutionary conserved imprinted region contains the paternally expressed genes DLK1, RTL1 and DIO3 and the maternally expressed noncoding RNA genes MEG3/GTL2, MEG8, MEG9, and RTL1AS, as well as two large clusters of maternally expressed microRNAs (miRNAs) and small nucleolar RNAs (snoRNAs) [13]. The parent-of-origin-specific expression of the imprinted chr14q32 genes is regulated by two DMRs: the germline MEG3/DLK1:IG-DMR located between the DLK1 and MEG3 genes and the somatic MEG3:TSS-DMR overlapping the MEG3 promoter region [14]. The MEG3/DLK1:IG-DMR acts upstream of the MEG3:TSS-DMR and governs it in a hierarchical fashion [15, 16]. Two further somatic DMRs have been recently described, the MEG8:Int2-DMR located in intron 2 of the MEG8 gene and the DLK1:Int1-DMR overlapping the second exon of the DLK1 gene [17–19]. The function of these two DMRs is currently unknown. In TS14 patients, the paternally methylated MEG3/DLK1:IG-DMR and MEG3:TSS-DMR are hypomethylated leading to upregulation of MEG3 and MEG8 and downregulation of DLK1. In these cases, the MEG8:Int2-DMR, normally methylated on the maternal chromosome, gains methylation probably as consequence of transcriptional readthrough derived from the activated paternal MEG3 promoter [17]. The overlapping phenotypic features between SRS and TS14 can be attributed to the involvement of shared biological pathways and/or gene co-regulation [1]. Several trans-molecular interactions between imprinted genes located on chr11p15 and chr14q32 have been reported. In particular, the concomitant overexpression of MEG3 and MEG8 leads to downregulation of IGF2 in cultured cells [20].

Fig. 1.

Schematic representation of the imprinted gene cluster on human chromosome 14q32. Paternally expressed genes are shown in light blue, while maternally expressed genes are represented in red. DNA methylation is indicated by black lollipops. The IG-DMR (MEG3/DLK1: IG-DMR) and the MEG3-DMR (MEG3: TSS-DMR) are methylated on the paternal allele, whereas the DLK1-DMR (DLK1:Int1-DMR) and the MEG8-DMR (MEG8:Int2-DMR) are methylated on the maternal allele. The dashed red line represents the maternally expressed polycistronic transcript starting from the MEG3 promoter

PLAG1, located on human chromosome 8q12, was initially identified as an oncogene associated with certain types of cancer and is the main translocation target in pleomorphic adenomas of the salivary glands [21]. It belongs to the PLAG family of zinc finger transcription factors, along with PLAG-like 1 (PLAGL1), which is a tumor suppressor, and PLAG-like 2 (PLAGL2), which acts as an oncogene. PLAG1 protein contains seven canonical C2H2 zinc finger domains (ZF) and a serine-rich COOH terminus that exhibits transactivation capacities suggesting that it may act as a transcriptional regulator. Its DNA binding site is composed of the core sequence GRGGC and a G-cluster RGGK separated by seven random nucleotides. The interaction with the core sequence is mediated by fingers 6 and 7, and that with the G-cluster by finger 3 [22]. The three PLAG proteins share the highest degree of homology in their NH₂-terminal zinc finger domain, with PLAGL1 and PLAGL2 showing 73% and 79% identity, respectively. In contrast, the COOH-terminal region is significantly more divergent. The greatest similarity is observed in zinc fingers 6 and 7, suggesting that PLAGL1 and PLAGL2 may also recognize a core motif similar to the PLAG1 core (GRGGC). Zinc fingers 2–5 are less conserved, but key amino acids at critical positions remain preserved [21, 22]. PLAGL1 dimerization through zinc finger 2 enhances transactivation of target genes, suggesting that also PLAG1 may act as a dimer [23].

PLAG1 plays a critical role in the oncogenic HMGA2–PLAG1–IGF2 pathway as it has been shown to be a transcriptional activator of IGF2 that binds its P3 promoter [22]. Genetic defects of this pathway can lead to fetal and postnatal growth restriction [10]. Inherited or de novo alterations involving the PLAG1 gene, including whole-gene deletions and intragenic pathogenic variants, were demonstrated in several cases with phenotypic features of SRS [10, 11, 24–33]. It has been proposed that PLAG1 haploinsufficiency leads to IGF2 repression and growth deficiency [10]. However, if further genes have a role in mediating the effect of PLAG1 on somatic growth is unknown.

To look for a possible dysregulation of imprinted genes linked to PLAG1 defects, we analysed the DNA methylation of 11–12 imprinted regions in 17 cases with clinical SRS features carrying heterozygous PLAG1 variants. We observed loss of methylation of the MEG8:Int2-DMR in all the cases predicted to generate stable aberrant PLAG1 proteins, but not in those with whole gene deletions, VUS or protein variants with C-degron recognized by APPBP2, indicating a functional interaction between PLAG1 and the chr14q32 imprinting cluster.

Patients

The study cohort included 17 patients with clinical diagnosis of SRS based on the Netchine-Harbison clinical scoring system (Additional file 1: Table S1). In these patients IC1 LoM and upd(7)mat as well as molecular TS14 were excluded. As detailed in Table 1, 8 cases were reported in previous publications. All the cases carried a variant involving the PLAG1 gene. In addition, 54 age-matched control samples were included for comparison. This control group consisted of unrelated individuals with no clinical features of the syndrome, analysed following the same molecular conditions as the patient cohort. The group comprised both Italian and French individuals.

Table 1.

Summary of the molecular features of the probands and their families

Cases	Proband sex	PLAG1 variant	Pathogenicity evaluation*	Genotype	MEG8:Int2-DMR methylation	Clinical features of proband	Publication
SRS1	Female	NM_002655.3: c.1023 T > A; p.Tyr341* novel nonsense mutation	Likely pathogenic	Proband: het Mother: het Father: wt	Proband: hypomethylation Mother: hypomethylation Father: NA	NH-CSS: 4/6 Small for gestational age, post-natal growth retardation, microcephaly, low body mass index, no protruding forehead	This study
SRS2	Female	NM_002655.3: c.527C > A; p.Ser176* novel nonsense mutation	Likely pathogenic	Proband: het Mother: wt Father: wt	Proband: hypomethylation Mother: normal methylation Father: NA	NH-CSS = 3/6 Small for gestational age, post-natal growth retardation, feeding difficulties during infancy, head circumference (− 1.5 SDS) but no values available	This study
SRS3	Female	NM_002655.2: c.1363del; p.Gln455Serfs*16 novel frameshift mutation-premature stop	Likely pathogenic	Proband: het Mother: wt Father: wt	Proband: hypomethylation Mother: NA Father: NA	NH-CSS = 4/6 Small for gestational age, feeding difficulties during infancy, prominent forehead with triangular face, relative macrocephaly at birth	Abi Habib W et al. [10]
SRS4	Female	NM_002655.2: c.439del; p.Ser147Valfs*82 novel frameshift mutation-premature stop	Pathogenic	Proband: het Mother: het Father: wt Proband’s sister: het	Proband: hypomethylation Mother: hypomethylation Father: NA Proband’s sister: hypomethylation	NH-CSS = 4/6 Small for gestational age, postnatal growth failure, feeding difficulties during infancy, prominent forehead with triangular face	Abi Habib W et al. [10]
SRS5	Female	NM_002655.2: c.599dup; p.Arg201Profs*52 novel frameshift mutation-premature stop	Likely pathogenic	Proband: het Mother: wt Father: wt	Proband: normal methylation Mother: NA Father: NA	NH-CSS = 4/6 Small for gestational age, postnatal growth retardation, feeding difficulties during infancy, prominent forehead	Meyer R., et al., [25]
SRS6	Female	NM_002655.3: c.666del; p.Phe222Leufs*7 novel frameshift mutation-premature stop	Likely pathogenic	Proband: het Mother: NA Father: NA	Proband: hypomethylation Mother: NA Father: NA	NH-CSS: 4/5 Small for gestational age, post-natal growth retardation, relative macrocephaly, feeding difficulties during infancy, protruding forehead noticed at the age of 3y4m	This study
SRS7	Male	NM_002655.3: c.779_780del; p.Val260Alafs*16 novel frameshift mutation-premature stop	Likely pathogenic	Proband: het Mother: het Father: wt	Proband: hypomethylation Mother: hypomethylation Father: normal methylation	NH-CSS = 4/6 Small for gestational age, postnatal growth failure, feeding difficulties during infancy, prominent forehead with triangular face	This study
SRS8	Male	NM_002655.3: c.770del; p.Asn257Metfs*6 novel frameshift mutation-premature stop	Likely pathogenic	Proband: het Mother: het Father: wt	Proband: hypomethylation Mother: hypomethylation Father: normal methylation	NH-CSS = 3/5 Small for gestational age, postnatal growth retardation, triangular face with prominent forehead	This study
SRS9	Female	NM_002655.3: c.1455_1502del; p.Ser485delinsArgAspSerGlyThrTrpIleHisTyrArgAsnValCysValAlaValPro* novel stop-loss mutation; Change of the C-ter end of the peptide, elongated protein (+ 1 aa)	Likely pathogenic	Proband: het Mother:het Father:wt Proband’s brother: het	Proband: hypomethylation Mother: hypomethylation Father: normal methylation Proband’s brother: hypomethylation	NH-CSS = 4/6 Small for gestational age, postnatal growth failure, feeding difficulties during infancy, prominent forehead with triangular face	This study
SRS10	Male	NM_002655.3: c.610_612del; p.Met204del novel in-frame deletion	Likely pathogenic	Proband: het Mother: het Father: wt	Proband: hypomethylation Mother: hypomethylation Father: normal methylation	NH-CSS = 4/6 Small for gestational age, postnatal growth failure, feeding difficulties during infancy, relative macrocephaly, triangular face	Vimercati A. et al.[23]
SRS11	Male	NM_002655.3: c.671G > A; p.(Arg224Gln) novel missense mutation	Likely pathogenic	Proband: het Mother: het Father: wt	Proband: hypomethylation Mother: hypomethylation Father: normal methylation	NH-CSS = 4/6 Small for gestational age, postnatal growth failure, feeding difficulties during infancy, triangular face with protruding forehead	Vimercati A. et al. [23]
SRS12	Male	NM_002655.3: c.545A > T; p.Glu182Val novel missense mutation	VUS Low Pathogenic Support	Proband: het Mother: NA Father: NA	Proband: normal methylation Father: NA Mother: NA	NH-CSS = 3/5 Post-natal growth retardation, feeding difficulties during infancy, prominent forehead	Kessler L et al. [32]
SRS13	Male	NM_002655.3: c.1162A > G; p.Ile388Val rs765459935 AF: 0.00001593 missense mutation	VUS >  > cl2 Uncertain pathogenicity	Proband:het Mother:het Father:NA	Proband: normal methylation Mother: normal methylation Father: NA	NH-CSS = 2/6 Small for gestational age, post-natal growth retardation, low body mass index, microcephaly, hypospadias	This study
SRS14	Female	NM_002655.3: c.-117-5del p.? rs1023307529 AF: 0.0005235 non-coding; (intron–exon boundary)	VUS >  > cl2 Uncertain pathogenicity	Proband: het Mother: het Father: wt Maternal grandmother: het Maternal grandfather: wt	Porband: normal methylation Mother: normal methylation Father: NA Maternal grandmother: NA Maternal grandfather: NA	NH-CSS = 4/6 Small for gestational age, postnatal growth retardation, protruding forehead, feeding difficulties during infancy	This study
SRS15	Female	chr8q12.1 deletion including the PLAG1 gene arr [GRCh37] 8q12.1 (56,834,331_ 58,921,491) × 1	Pathogenic	Proband: het Mother: wt Father: wt	Proband: normal methylation Father: NA Mother: NA	NH-CSS = 4/6 Small for gestational age, moderate postnatal growth retardation, feeding difficulties during infancy, relative macrocephaly, triangular face	Fernández-Fructuoso JR. et al. [27]
SRS16	Male	chr8q12.1 deletion including the PLAG1 gene arr [GRCh37] 8q12.1 (57,079,399_57,155,945) × 1	Pathogenic	Proband: het Mother: wt Father: wt	Proband: normal methylation Father: NA Mother: NA	NH-CSS = 2/6 Feeding difficulties during infancy, triangular face with prominent forehead	Baba N et al. [28]
SRS17	Male	chr8q12.1 deletion including the PLAG1 gene arr[GRCh37]8q12.1 (56,986,129_57,169,684) × 1	Pathogenic	Proband:het Mother:wt Father:wt	Proband: normal methylation Father: NA Mother: NA	NH-CSS = 4/6 Small for gestational age, postnatal growth retardation, low body mass index, relative macrocephaly at birth (but not after birth)	This study

*Pathogenicity assessment based on ACMG criteria. In third column AF: Allele frequency in European population. In fifth column: het = heterozygous; wt = wild type. In both fifth and sixth column: NA = not analysed

Materials and methods

DNA extraction

Genomic DNA of patients and their relatives was isolated from peripheral blood leukocytes (PBL) by standard procedures.

NGS analysis

For SRS3-SRS5, SRS10-SRS12 and their relatives, sequencing analysis has been reported previously [10, 25, 32, 33]. Specifically, for SRS3 and SRS4 library preparation, exome capture, sequencing, and data analysis were performed by IntegraGen SA (Evry, France). For SRS5 DNA was enriched using the Nextera Rapid Capture Exome (v.1.2) (Illumina, San Diego, CA, USA). Sequencing was performed using a NextSeq500 Sequencer (Illumina, San Diego, CA, USA). Targeted sequencing was carried out for cases SRS10 and SRS11. Libraries were prepared using the Nextera XT DNA Library Prep Kit (Illumina, San Diego, CA) and sequenced with an Illumina Miseq sequencer. For patient SRS12, genome sequencing (GS) was conducted by using the DNA PCR-free kit (Illumina Inc. San Diego, CA, USA) and sequencing was performed on a NovaSeq 6000 System (S4 Reagent Kit v1.5) (Illumina Inc.).

For SRS1, SRS2, SRS6, SRS13, SRS14 families, library preparation was performed using a custom sequencing panel designed by Sophia Genetics. Libraries were sequenced on a MiSeq (Illumina, San Diego, CA, USA). For SRS7-SRS9 families, library preparation of single samples was performed using the SureSelect QXT Clinical Research exome v2 and Human All Exon V7 kits compatible with Illumina platform version F0(Agilent Technologies, Santa Clara, CA, USA) following the manufacturer’s instructions. Libraries were sequenced using a NovaSeq6000 system (Illumina, San Diego, CA, USA) [34]. The WES data preprocessing followed the nf-core Sarek pipeline (version 3.3.2) using its default settings [35]. The sequencing reads were mapped to the human reference genome GRCh38 using the BWA-MEM algorithm. The resulting variant call format (VCFs) files were then annotated with Franklin by Genoox.

Validation and segregation analysis of variants in the family members were performed by Sanger sequencing.

Molecular karyotyping

Molecular karyotyping for SRS15 and SRS16 was previously reported [27, 28]. For SRS17, Comparative Genomic Hybridization (CGH) array analysis was carried out by using a 180 k microarray (Agilent Technologies, Santa Clara, CA, USA). Raw data were analysed by the CytoGenomics V3.0 software (Agilent Technologies).

In silico prediction of variant pathogenicity and protein stability

Variants were initially classified based on their predicted impact on the protein. In silico pathogenicity prediction was performed for all Single Nucleotide Variants (SNVs) and indels using CADD v1.7 [36]. For missense variants, additional analyses were conducted using PolyPhen-2 [37], SIFT [38], AlphaMissense [39] and REVEL [40]. Furthermore, all variants, including the three large deletions, were classified according to the ACMG standards and guidelines [41, 42]. Protein stability was investigated using DEGRONOPEDIA [43], which enables the identification of known degron motifs involved in protein degradation pathways.

Methylation analysis

Methylation-specific multiple ligation-dependent probe amplification (MS-MLPA) targeting 11 or 12 imprinted loci was performed using the SALSA MS-MLPA Kit (MRC-Holland, Amsterdam, The Netherlands) according to the manufacturer’s instructions. Probemix ME034-C1 or Probemix ME034-D1, designed for multi-locus imprinting analysis, were used. Raw data were analysed using Coffalyser.Net software (MRC Holland, Amsterdam, The Netherlands). Pyrosequencing analysis was carried out as described in Sparago et al. [44] and it was conducted in separate batches of experiments. The 13 analysed CpGs are included in the following coordinates: chr14:100,904,601–100,904,702 (GRCh38/hg38). ImprintCap, a NGS-based methodology for large-scale methylation studies at several imprinted loci was performed as described in Brioude et al. [45] on 13 control samples to explore the methylation of MEG8:Int2-DMR. The 42 analysed CpGs are included in the following coordinates: chr14: 100,904,423- 100,905,080 (GRCh38/hg38).

Results

DNA methylation analyses

We investigated the DNA methylation of 11–12 imprinted regions in 17 SRS cases carrying rare PLAG1 variants in heterozygosity (Table 1) and 54 control individuals, by using the multi-locus MS-MLPA C1 or D1 kit. Compared to the mean of the controls, we found a partial loss of methylation of the MEG8:Int2-DMR in the cases SRS1-SRS4,SRS6-SRS11 and SRS16. Most of these cases were familial and the imprinting defect segregated within the family, being detected in all the members who carried the PLAG1 variant (Fig. 2). Conversely, MEG8:Int2-DMR methylation was normal in SRS5, SRS12-SRS15 and SRS17. The MS-MLPA results are reported in the Additional file 1: Table S2. In all cases, the methylation levels of the other tested DMRs, including the MEG3: TSS-DMR, were comparable to those of the controls (Fig. 2). In addition, we evaluated the methylation status of the MEG8:Int2-DMR in 7 SRS patients carrying HMGA2 variants and 8 with IGF2 variants and all of them showed normal methylation (Additional file 1: Table S3).

Fig. 2.

Representative MS-MLPA results in one of the families. Copy number and DNA methylation were analysed for each member of the family SRS9 by ME034-C1 multilocus kit. The mean values of control subjects were used for assessment of relative copy number and methylation percentage. Red arrows indicate the probes detecting the methylation defect. *No clinical information was available for this patient

To further explore MEG8 methylation in other molecular subgroups of SRS, we performed multi-locus MS-MLPA in 28 patients with IC1 loss of methylation (IC1-LoM), 4 with maternal uniparental disomy of chromosome 7, and 31 idiopathic cases (NH-CSS ≥ 4). All of these also exhibited normal methylation levels (Additional file 1: Table S4). Finally, 2/54 of the healthy control individuals showed lower methylation levels at the MEG8:Int2-DMR, according to the MS-MLPA results (Additional file 1: Table S5).

To validate the MS-MLPA results, we analysed MEG8:Int2-DMR methylation through the more quantitative pyrosequencing assay in the patients carrying the PLAG1 variants and in several controls. With this assay, we tested the methylation level of 13 CpGs across the MEG8:Int2-DMR. The results were consistent with the data obtained by MS-MLPA in 10/11 patients and confirmed the co-segregation of the MEG8:Int2-DMR hypomethylation with the PLAG1 variants (Fig. 3A–G).

Fig. 3.

Pyrosequencing analyses of DNA methylation at the MEG8:Int2-DMR in cases with 677 PLAG1 variants. H shows data from additional patients (SRS3, SRS5, SRS10-SRS12, SRS14-SRS17) for whom no family member analysis was possible. DNA of patient SRS6 was not available for pyrosequencing analysis. The exact genomic positions of the 13 CpGs analysed are provided in Additional file 1: Table S6. Gray-background symbols indicate individuals with a milder phenotype; while symbols with black dots indicate that those individuals are carriers of the genetic variant although their clinical phenotype remains unclear. *No clinical information

Furthermore, the pyrosequencing results demonstrated that the hypomethylation affected the entire MEG8:Int2-DMR. The reduction in methylation levels was between 15 and 33% compared to the mean of the controls (Additional file 1: Table S6; Additional file 2: Fig. S1). Patient SRS16 and the two healthy individuals found hypomethylated with MS-MLPA showed a more modest (3–6%) methylation defect restricted to only a few CpGs located at the 5’ part of the DMR (Fig. 3H; Additional file 1: Table S7; Additional file 3: Fig. S2). From ImprintCap analysis on 13 control samples, those latter 5’-CpGs were shown to be highly variable, making those 5’-CpGs not suitable to define gain or loss-of-methylation in patients (Additional file 1: Table S8; Additional file 4: Fig. S3).

In summary, these results indicate a specific loss of methylation of the entire MEG8:Int2-DMR in 10 out of 17 SRS cases carrying a genetic variant in PLAG1.

Genetic alterations in PLAG1

Since the MEG8 methylation defect was not consistently present in all the 17 SRS cases with PLAG1 variants, we assessed the pathogenic potential of the identified genetic variants by in silico analysis (Table 1). Figure 4 shows all the variants identified in our cohort. The clinical features and PLAG1 variants of 8 patients (SRS3-SRS5, SRS10-SRS12 and SRS15, SRS16) have been described in previous studies [10, 25, 27, 28, 32, 33], while the remaining nine are novel cases.

Fig. 4.

PLAG1 mutations. The top panel shows the exon/intron structure of the PLAG1 gene, with exons 4 and 5 containing the coding sequences (highlighted in dark grey). The bottom panel illustrates the PLAG1 protein, with seven zinc fingers (ZF1-ZF7) represented in blue. Red lollipops indicate the variants associated with hypomethylation at the MEG8:Int2-DMR, while green lollipops denote the variants associated with normal methylation. The green upper lines highlight three cases with chr8q deletions encompassing the PLAG1 gene, which are associated with normal methylation at the MEG8:Int2-DMR

The patients SRS1-SRS11 (Table 1) carried heterozygous SNVs or indels of PLAG1 that were classified as pathogenic or likely pathogenic according to the ACMG guidelines and gene variant interpretation (Additional file 1: Table S9). In particular, two nonsense variants were identified in cases SRS1 and SRS2. The variant p.Tyr341* resulting in loss of the C-terminal activation domain was found in both the SRS1 proband and her mother, while the variant p.Ser176* causes the loss of ZF6, ZF7 and the entire C-terminal region of PLAG1 in patient SRS2. Frameshift variants were identified in the cases SRS3-SRS8. These pathogenic variants lead to putative truncated proteins lacking key functional domains (Additional file 5: Fig. S4). In particular, the variant p.Gln455Serfs*16 identified in SRS3 results in a truncated protein with 496 amino acids [10]. The p.Ser147Valfs*82 mutation found in SRS4 causes a frameshift before the ZF5 domain, generating an abnormal peptide of 227 residues [10]. In patient SRS5, the p.Arg201Profs*52 variant leads to a truncated protein of 251 amino acids [25]. Additionally, the p.Phe222Leufs*7 mutation carried by patient SRS6 causes a frameshift starting from ZF7 and likely results in a truncated peptide composed of 227 amino acid residues (221 wild-type and 6 mutated), while the p.Val260Alafs*16 and p.Asn257Metfs*6 variants, inherited maternally by the SRS7 and SRS8 patients, respectively, lead to loss of the entire C-terminal domain, producing abnormal proteins of 274 and 261 amino acids. Notably, all the truncating variants occur in the last two exons of PLAG1.A novel 48 bp deletion was identified in the SRS9 proband, as well as in her mother and brother, but not in her father. This mutation likely results in a protein of 501 amino acids, with 484 wild-type, 16 mutated and one additional residue at the C-terminus (p.Ser485delinsArgAspSerGlyThrTrpIleHisTyrArgAsnValCysValAlaValPro*). Finally, SRS10 and SRS11 carried an in-frame deletion and a missense mutation in the ZF6 and ZF7 sequences, respectively [33].

To assess the stability of the 11 predicted aberrant proteins resulting from deleterious PLAG1 mutations, we analysed the presence of known degron motifs involved in degradation pathways [43]. Notably, a RxxGxx motif was identified at the C-terminus of the PLAG1 mutant p.Arg201Profs*52 (SRS5). APPBP2, a substrate receptor of CRL2 (Cullin-RING E3 ubiquitin ligase 2) complexes, is known to recognize C-degrons defined by a distinctive C-terminal Arg-x-x-Gly sequence [46]. This specific degron motif was absent in the other mutated proteins that were subjected to analysis, suggesting a potential importance of the R-x-x-G motif in substrate recognition and subsequent ubiquitination processes within CRL2-mediated proteolysis for the PLAG1 mutant p.Arg201Profs*52 (Additional file 5: Fig. S4).

According to the results of the in-silico analysis and the ACMG criteria, the patients SRS12 [32], SRS13 and SRS14 carried Variants of Uncertain Significance (VUS) in the PLAG1 gene (Table 1, Additional file 1: Table S9). In particular, an intronic variant (NM_002655.3:c.-117-5del;p.?) occurring in a homopolymeric region of PLAG1 intron 3 was found in SRS14, her mother and her healthy maternal grandmother, while the missense variants p.Glu182Val and p.Ile388Val were identified in cases SRS12 and SRS13, respectively. In particular, the p.Ile388Val variant involved a non-conserved residue and, in both SRS12 and SRS13 cases, the functional PLAG1 domains were not affected. Thus, their pathogenicity remains uncertain.

Finally, patients SRS15 [27], SRS16 [28], and SRS17 carried heterozygous deletions on chromosome 8q (Table 1; Additional file 6: Fig. S5). Among them, patient SRS16 exhibited the smallest deletion, affecting only PLAG1 and CHCHD7 genes. In contrast, SRS15 presented with a larger de novo deletion of approximately 2.1 Mb at 8q12, encompassing 32 genes, including 9 OMIM-annotated genes. In SRS17, molecular karyotyping identified a ~ 183 kb de novo deletion (arr[GRCh37]8q12.1(56,986,129_57,169,684) × 1) involving PLAG1 but also CHCHD7, MOS, and SDR16C5. In all three cases, PLAG1 was proposed as the primary candidate gene responsible for the observed SRS phenotype [27, 28]. According to the ACMG criteria, all three deletions are classified as pathogenic.

Overall, we collected a wide spectrum of PLAG1 variants: two nonsense variants (SRS1-SRS2), six frameshift (SRS3-SRS8), one stop-loss (SRS9), one in-frame deletion (SRS10), three missense variants (SRS11-SRS13), one intronic variant (SRS14), and three large deletions (SRS15-SRS17) affecting the entire PLAG1 gene on chromosome 8q. Evaluation of their potential pathogenicity led us to observe that 11 out of 14 SNVs/indels were predicted to be damaging, whereas the remaining 3 were classified as VUS. Among these 11 damaging variants, one harbours a degron motif likely associated with protein degradation. The copy number variants were also classified as pathogenic.

Correlation of MEG8 methylation with PLAG1 gene variants

To further investigate the relationship between MEG8:Int2-DMR methylation and PLAG1 function, we compared the methylation of this locus among the 17 SRS patients carrying different PLAG1 variants (Table 1). Interestingly, loss of MEG8:Int2-DMR methylation was detected in 10 out of 11 cases carrying pathogenic or likely pathogenic single nucleotide variants or indels, with SRS5 (p.Arg201Profs*52) representing the only exception. In particular, this mutated protein uniquely harbours the C-terminal RxxGxx motif, which facilitates protein degradation [46, 47], suggesting a distinct impact on protein stability compared with the other pathogenic mutations. Conversely, all the 6 cases carrying VUS or large deletions encompassing the PLAG1 gene showed normal methylation levels. Notably, MEG8:Int2-DMR hypomethylation was not detected in any of the SRS patients with molecular alterations in other genes of the HMGA2–PLAG1–IGF2 pathway, in those with IC1-LoM, upd(7)mat and in idiopathic SRS cases.

In summary, MEG8 hypomethylation is associated with PLAG1 variants that are predicted to produce stable abnormal proteins, but not with whole PLAG1 gene deletions, variants of uncertain pathogenicity or variants leading to protein degradation.

Discussion

The observed association between damaging PLAG1 variants and MEG8:Int2-DMR hypomethylation in SRS indicates a possible role of PLAG1 in regulating the expression of chr14q32 imprinted genes controlling somatic growth.

In the present study, MEG8:Int2-DMR hypomethylation was demonstrated in cases with pathogenic or likely pathogenic PLAG1 variants that are predicted to produce abnormal proteins. In contrast, normal MEG8 methylation was observed in the cases with missense/intronic variants classified as VUS according to the ACMG guidelines and PLAG1 deletions. In addition, most of the pathogenic PLAG1 cases are familial and both epigenetic and genetic defects co-segregated within the families, further supporting the association between damaging PLAG1 variants and MEG8 hypomethylation. This epigenetic defect appears to be specific for PLAG1, because it was not found in the cases with HMGA2 and IGF2 variants, patients with IC1-LoM, upd(7)mat cases and idiopathic cases.

Apart from MEG8, the methylation of other 10–11 imprinted loci was found to be normal in the patients studied, suggesting a specific interaction between PLAG1 and MEG8-DMR. Analysis of chromatin immunoprecipitation sequencing (ChIP-seq) data indicates that PLAG1 normally interacts with the MEG8:Int2-DMR, although it is unclear if its binding is influenced by DNA methylation [48]. Interestingly, PLAGL2, a homologue of PLAG1, is also predicted to bind the same DMR, as suggested by data from JASPAR CORE 2024—Predicted Transcription Factor Binding Sites [49]. Particularly, PLAGL2 shares 79% identity with PLAG1 in its NH₂-terminal zinc finger domain, suggesting that the zinc finger domain may be involved in recognizing the MEG8:Int2-DMR. In humans, ENCODE ChIP-seq data revealed a conserved CTCF binding site within the MEG8-DMR [19]. In mice, however, CTCF binds to the Meg8-DMR in a non-allele-specific manner in vivo [50]. CTCF (CCCTC-binding factor) is a well-characterized vertebrate protein with eleven zinc fingers, the first ten of which are C2H2-type, similar to those found in PLAG1, while the last one is of C2HC-type [51]. Notably, PLAG1 and CTCF recognize similar consensus sequences, as both bind to G-rich regions.

In the present SRS patients, the truncating PLAG1 variants map in the last two exons of the gene and should therefore escape the nonsense-mediated decay (NMD) pathway [52]. Also, they likely lead to aberrant PLAG1 proteins, which could influence methylation at the MEG8:Int2-DMR because of reduced DNA binding. Indeed, because zinc finger 2 is retained in all these variants, they may dimerize similar to PLAGL1 and exert a dominant-negative effect on the wildtype allele [23]. Consistent with this hypothesis, the single nucleotide deletion (SRS10) and the missense variant (SRS11) in the zinc fingers 6 and 7, that are responsible for specific DNA motif recognition [33], result in MEG8 hypomethylation. In contrast, the SRS5 variant that contains a C-terminal degron motif and likely results in an unstable aberrant protein is associated with normal MEG8 methylation [46, 47]. Also, the variants of uncertain pathogenicity are unlikely to disrupt the DNA binding function of PLAG1 and thereby maintain normal MEG8 methylation. Nevertheless, the mechanism by which reduced PLAG1 binding may interfere with MEG8 methylation remains to be defined. The absence of methylation defects in the cases with whole gene deletion is consistent with the dominant-negative hypothesis [53].

In SRS16 and the two healthy individuals who scored hypomethylated at the MEG8:Int2-DMR according to the MS-MLPA results, further analysis by pyrosequencing showed that this methylation defect was less severe compared to that detected in all the other patients and restricted to a few CpGs located at the 5’ part of the DMR, which was shown to be highly variable in a control population. This finding suggests that while MS-MLPA may indicate MEG8:Int2-DMR hypomethylation, it does not always reflect the full methylation pattern, and further confirmation using more sensitive and extensive techniques like pyrosequencing or NGS-based technologies like ImprintCap is essential for its accurate assessment.

The observed MEG8 hypomethylation in the present study suggests that chr14q32 genes have a role in PLAG1-dependent pathogenesis of SRS. Transcription starting from the MEG3 promoter is crucial for establishing the methylation imprint at the MEG8-DMR. In the TS14 cases with MEG3: TSS-DMR hypomethylation, the MEG3 long noncoding RNA is activated and transcription through the MEG8:Int2-DMR leads to its hypermethylation [17, 19]. Unfortunately, MEG3 expression levels could not be assessed in our cases as RNA was not available. However, in all the cases with PLAG1 mutations, MEG3 methylation is unaffected, suggesting that alternative molecular mechanisms lead to MEG8:Int2-DMR hypomethylation. In a recent study, Baena et al. [19] described a family in which MEG8 but not MEG3 exhibits a methylation pattern dependent on the parental origin of a DLK1 deletion, suggesting an interaction between these two loci. Also, a conserved CTCF binding site is present and therefore an insulator may be formed within the MEG8-DMR [19]. Interestingly, several studies indicate a regulatory function of PLAG1 on the DLK1/MEG3 domain. Declercq et al. [54] demonstrated that targeted PLAG1 overexpression in murine salivary glands induces pleomorphic adenoma-like tumors accompanied by strong upregulation of Dlk1. Furthermore, recent transcriptomic profiling of Central Nervous System (CNS) embryonal tumors with PLAG1 fusions demonstrated upregulation of DLK1, supporting the role of PLAG1 in the deregulation of the DLK1/MEG3 domain [55]. In human hematopoietic stem and progenitor cells (HSPCs) PLAG1 has been shown to activate this locus at the transcriptional level. Specifically, within the DLK1/MEG3 region on chr14, PLAG1 overexpression induces the expression of multiple imprinted microRNAs, including miR-770, miR-433, miR-127, and miR-370 [56, 57]. However, these reports did not evaluate the methylation of the different DMRs of the DLK1/MEG3 locus. The MEG8 DMR overlaps the MEG8 lncRNA and two clusters of miRNAs and snoRNAs, all of which are deregulated in cancer [58, 59]. It is possible that the disrupted PLAG1 binding to MEG8 interferes with expression of one or more of these genes. Nevertheless, the molecular mechanisms by which PLAG1 controls the expression of the chr 14 imprinted genes is unknown.

Finally, the normal methylation in the cases with HMGA2 and IGF2 variants suggests that MEG8 works as PLAG1 target upstream of HMGA2 and IGF2 in the PLAG1–HMGA2-IGF pathway.

SRS3-SRS5, SRS10-SRS12, SRS15, SRS16 are previously published cases in which the identified PLAG1 variant was considered the main cause of the SRS phenotype. However, the new molecular analysis on the entire cohort suggests that the VUS present in SRS12 and those of SRS13 and SRS14 may not be damaging. Thus, the variation is unlikely to be causal of the SRS phenotype. Dominant-negative mutations are generally expected to cause more severe phenotypes, because they interfere with the function of the wild-type allele [53]. In the present study, we cannot confidently conclude whether there are significant phenotypic differences between patients carrying PLAG1 variants associated with MEG8-DMR hypomethylation and patients with PLAG1 variants exhibiting MEG8-DMR normal methylation. Within our cohort, only two individuals (SRS13 and SRS16) exhibited a NH-CSS score < 3. They carry a VUS and whole gene deletion, respectively, and both are associated with normal MEG8 methylation. Unfortunately, the small cohort size limits the possibility to clearly establish a role of MEG8 methylation in the phenotype of our patients.

Conclusions

The present study identifies MEG8:Int2-DMR hypomethylation as a specific epigenetic signature of damaging PLAG1 variants resulting in stable aberrant proteins, because they likely act in a dominant-negative manner. These findings support a novel pathogenetic mechanism whereby impaired PLAG1 function alters the methylation and possibly the expression of imprinted genes within the chr14q32 region. Furthermore, MEG8:Int2-DMR methylation studies might be useful in the future as a functional test for PLAG1 VUS.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

DMRs

Differentially methylated regions

ImpDis

Imprining disorders

LoM

Loss of methylation

MLID

Multi-locus imprinting disturbances

NH-CSS

Netchine-Harbison clinical scoring system

NMD

Nonsense-mediated decay

PBL

Peripheral blood leukocytes

SRS

Silver-Russell syndrome

SRSp

Silver-Russell syndrome spectrum

SNVs

Single nucleotide variants

TS

Temple syndrome

UPD

Uniparental disomy

VUS

Variant of uncertain significance

ZF

Zinc Finger

Author contributions

Conceptualization, FB, AR, FlCe; investigation, EDA,LP,FrCe,AV,MVC, AS, CG, NT, SR, TE,IN, JRF; writing—draft preparation EDA, FlCe, AR, FB; supervision, FB, AR. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by grants from the Associazione Italiana Ricerca sul Cancro (AIRC; IG 2020 ID 24405) and Fondazione Telethon (GMR23T1062) awarded to AR and Italian Ministry of University and Research PRIN 2022B2N2BY awarded to AR and MVC. TE is supported by the Deutsche Forschungsgemeinschaft (EG 115/13–1). ImprintCap experiments were funded with the French Agence Nationale pour la Recherche (ANR) grant no. ANR-22-CE14-0021.

Data availability

The datasets supporting the conclusions of this article are included within the article and its additional files.

Declarations

Ethics approval and consent to participate

The study was approved by the ethical committees of the University of Campania Luigi Vanvitelli (Naples, Italy; approval number: 10423, 5 May 2020), Istituto Auxologico Italiano, Hôpital Trousseau (Paris), Center for Human Genetics and Genome Medicine (Aachen), Hospital General Universitario de Santa Lucía. A written informed consent was obtained from subjects or legals for molecular studies and publication of the data.

Consent for publication

The families agreed for publication by signing an informed consent template.

Competing interests

The authors declare no competing interests.