MATERIALS AND METHODS

Quantitative RT-PCR

Fibroblast cell lines cultured in DMEM and 10% fetal calf serum were harvested while subconfluent, and total RNA was isolated using a guanidine thiocyanate method (Totally RNA kit, Ambion, Austin, TX, USA). First strand cDNA was synthesised from 4 μg total RNA using RetroScript reverse transcriptase (Ambion) and oligo (dT) primer. Quantitative PCR was carried out on an ABI Prism 7700 Sequence Detector using a TaqMan assay for CDKN1C³³ and internal controls β-actin and GAPDH (Applied Biosystems, Foster City, CA, USA). For each TaqMan probe assay, a single 96 well plate contained triplicate cDNA samples from BWS cases and normal controls, as well as serial dilutions of human fetal kidney cDNA to generate a standard curve.

Ribonuclease protection assay

Radioactive RNA probes were synthesised using the MaxiScript T7/T3 kit (Ambion) and ³²P-UTP (800 Ci/mmol) (Perkin-Elmer Life Sciences, Boston, MA, USA). The probe used for the CDKN1C was from Pharmingen, San Diego, CA, USA and the Tri-Cyclophillin human probe was from Ambion. RPA was carried out with 30 μg total RNA using the RPAIII kit as described by the manufacturer (Ambion) with hybridisation at 55°C. RNase digestion was carried out with a 1/100 dilution of the RNaseA/T1 mix. Conditions were such that all probes were in excess. Protected fragments were electrophoresed in 5% TBE-Urea precast Ready Gels (BioRad, Heracles, CA, USA). The intensity of protected fragments was quantified using a Phosphorimager (Piscataway, NJ, USA) and ImageQuant software (Molecular Dynamics, Piscataway, NJ, USA).

Methylation analysis

DNA (5 μg) from patients and controls was digested with PpuMI alone or in combination with SacII at 37°C for 18 hours. Digests were electrophoresed in 0.9% agarose and transferred to Hybond N⁺ (Amersham, Little Chalfont, Bucks, UK) as suggested by the manufacturer. A hybridisation probe for the CDKN1C promoter region was prepared by PCR using sense primer 5′-CGCAGATAGCGGCTTCAGACTCCAGCTCCAGGG-3′ and antisense primer 5′-TGCTGGCTAGCTCGCTCGCTCAGGCCTGGC-3′, with human genomic DNA as template. These primers amplify a sequence corresponding to positions 116–1007 in accession number D64137.³⁴ The 892 bp PCR product was cloned into pCRII TOPO and labelled by random priming. Washes were twice for 10 minutes at room temperature in 2X SSC, and twice for 30 minutes at 65°C with 0.2X SSC, 0.5% SDS.

RESULTS

To test the hypothesis that LOM at KvDMR1 is associated with downregulation of CDKN1C, we analysed the steady state expression level of this gene in fibroblast cell lines from five persons with BWS who had complete LOM at KvDMR1 (fig 1(B) and data not shown). Since, in principle, CDKN1C could be downregulated by more than one mechanism in BWS, patient expression levels were compared with fibroblast cells from normal individuals instead of those from other BWS subgroups (patients with normal methylation at KvDMR1). Total RNA was extracted from subconfluent cultures of patient and control fibroblasts, and first strand cDNA was synthesised. RT-PCR using a TaqMan assay for CDKN1C³³ and internal controls β-actin and GAPDH (Applied Biosystems) was carried out using the standard curve method. Each experiment for a given TaqMan assay consisted of triplicate samples from subjects with BWS and normal controls, as well as serial dilutions of human fetal kidney cDNA to generate a standard curve, on the same 96 well plate.

Representative amplification plots for CDKN1C from one experiment using fibroblast cDNA from normal individuals (green lines) and from persons with BWS who had LOM at KvDMR1 (red lines) are shown in fig 2(A). For illustrative purposes, each amplification plot was normalised with respect to the same internal control Ct value. Note that in all cases, amplification plots corresponding to the persons with BWS crossed the fluorescence threshold at a higher cycle number (lower expression level) than those corresponding to normal individuals. Fig 2(B) shows typical standard curves used for quantification and demonstrates that each PCR assay was linear over four orders of magnitude. The histogram in fig 2(C) shows results from a representative experiment where the expression level of CDKN1C was normalised to β-actin and GAPDH. Despite a wide range of expression levels in fibroblasts from normal individuals, we were able to show that CDKN1C was significantly downregulated in fibroblasts from BWS cases with LOM at KvDMR1. In the experiment shown in fig 2, BWS patient cells exhibited an average 93% reduction in CDKN1C expression when normalised to either β-actin or GAPDH. Five independent experiments were carried out with similar results.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

RT-PCR analysis of BWS patients with LOM at KvDMR1. (A) Corrected amplification plots generated by plotting intensity v cycle number. Each line represents the average of triplicate assays from each normal (green) or patient (red) sample. For illustrative purposes, each plot was normalised to a common internal control signal. (B) Representative standard curves used for quantification generated by plotting the threshold cycle (Ct) for different amounts of input cDNA. (C) Chart showing CDKN1C expression levels determined by RT-PCR normalised with respect to β-actin (blue bars) and GAPDH (red bars). (D) Results of Student’s t-tests comparing normalised expression levels of CDKN1C of normal controls (N) with those of individuals with BWS and LOM at KvDMR1.

These findings were confirmed in a RPA using radioactive single stranded RNA probes for CDKN1C and cyclophilin. Despite roughly equivalent intensities of the cyclophilin protected fragment across all samples, and a large degree of variation in CDKN1C expression among normal samples, an obvious reduction in the intensity of the CDKN1C protected fragment was observed in cases of BWS with LOM at KvDMR1, as evident in fig 3(A). Following quantification by PhosphorImager analysis, and normalisation to the cyclophilin signal, the average CDKN1C expression level in fibroblasts from persons with BWS who had LOM at KvDMR1 was reduced by 86% compared with the level in fibroblasts from normal individuals (fig 3(B)), consistent with the results from the RT-PCR assay. A residual signal indicating low level expression of CDKN1C was observed in RNA from most BWS patient cell lines. Since cell lines from patients’ parents were not available, we could not determine the allelic origin of these transcripts. Although incomplete repression of the maternal allele of CDKN1C in patient fibroblasts cannot be ruled out, the residual signal seen in these samples may represent low level expression of CDKN1C from the paternal allele.^35,36

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Ribonuclease protection assay (RPA) analysis of individuals with BWS who had LOM at KvDMR1. (A) RPAs performed as described above, using a human cyclophilin (CYP) probe to normalise the CDKN1C signals. (B) Results of Student’s t-tests comparing the normalised expression levels of CDKN1C of normal controls (N) with those of persons with BWS and LOM at KvDMR1.

Unlike the mouse Cdkn1c gene, which is differentially methylated on the silent paternal allele, the human locus is devoid of methylation both in the promoter region and in the gene body.³⁵ It has recently been shown, however, that promoter hypermethylation at CDKN1C does occur in a number of malignancies in which this gene is silenced.^37,38 To determine whether promoter methylation at CDKN1C accompanies its silencing in association with LOM at KvDMR1, Southern analysis was carried out using the methylation sensitive restriction endonuclease SacII. A 1.2 kb PpuMI fragment that encompasses the CDKN1C promoter includes two SacII sites which are located in a region containing multiple potential transcription factor binding sites (fig 4). In fibroblast DNA from normal individuals, the 1.2 PpuMI fragment was completely digested by SacII into 500 bp and 700 bp fragments, indicating a lack of methylation at these restriction sites. The same cleavage pattern was observed in DNA from subjects with BWS and LOM at KvDMR1, with no trace of higher molecular weight (methylated) bands detectable. These results demonstrate that, at least for the four CpG’s assessed by the two SacII sites, no promoter methylation changes accompany silencing of CDKN1C in persons with BWS and LOM at KvDMR1.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Methylation analysis of CDKN1C promoter. Fibroblast DNAs from normal individuals, and from persons with BWS and LOM at KvDMR1, were digested with PpuMI alone (first lane only) or with PpuMI plus SacII; electrophoresed in agarose; and transferred to a nylon membrane. The location of the hybridisation probe is shown on the genomic restriction map. Also indicated on the map are the putative transcription start site (arrow), TATA and CAT boxes (open and filled triangles, respectively), and 11 Sp1 binding sites as determined by Tokino et al.³⁴

DISCUSSION

Multiple genetic and epigenetic mechanisms affecting the integrity or expression of imprinted genes in human band 11p15.5 can give rise to BWS.^3,4 Most of these mechanisms result in either two active copies of the IGF2 gene which encodes a potent fetal growth factor—for example, paternal UPD, paternal duplication, LOI—or the absence (paternal UPD) or mutation of the active maternal allele of CDKN1C, a cyclin dependent kinase inhibitor. Animal models involving increased expression of Igf2 or null mutation in Cdkn1c suggest that abnormal expression of these two genes is the major contributor to the BWS phenotype. Indeed, in one mouse model, Igf2 and Cdkn1c appeared to function in an antagonistic manner during the development of certain organ systems.³⁹

KvDMR1 is normally methylated on the maternally derived chromosome and unmethylated on the paternal allele.^7,19–21 This locus has been shown to function as an ICR in 11p15.5 (and in the mouse counterpart distal chromosome 7), regulating the imprinted expression of at least six paternally silenced genes.^26,27 The most frequent alteration (genetic or epigenetic) associated with BWS is an absence or LOM at the maternal allele of KvDMR1.^22–25 Here we show, both by RT-PCR and RPA, that LOM at KvDMR1 is associated with a dramatic reduction in the steady state expression level of CDKN1C. Our results are consistent with the recent demonstration that Cdkn1c was silenced in ES cells deficient for Dnmt3L, a DNA methyltransferase family member shown to be essential for the establishment of maternal imprints.⁴⁰ Roughly 50% of familial cases and 5% of sporadic cases of BWS are due to maternally inherited mutations of CDKN1C.²⁰ Assuming that expression levels in fibroblasts mirror those in fetal tissues during development, the epigenetic repression of CDKN1C observed in fibroblasts from individuals with BWS who have LOM at KvDMR1 could be functionally equivalent to an inactivating mutation and, if so, is the likely cause of the syndrome. These results also explain the high frequency of exomphalos observed in cases of BWS with LOM at KvDMR1 or with germline mutations in CDKN1C.^15,23,25

KvDMR1 has been shown to function as a chromatin insulator and/or bidirectional silencer in cell culture,^28,30,31 and at least the insulator function is abrogated by methylation.³⁰ However, at present, it is not known whether this locus operates in a similar fashion to silence paternal copies of maternal specific genes. KvDMR1 also contains the promoter for KCNQ1OT1/Kcnq1ot which appears to be functionally separable from its insulator/silencer activity (our unpublished results).^26–28 The involvement of non-coding RNAs in gene silencing has been shown for genes on the mammalian inactive X-chromosome⁴¹ and at the Igf2r locus.⁴² Although the functional significance of KCNQ1OT1 in gene silencing has yet to be demonstrated, it is possible that KvDMR1 is a bipartite regulatory locus which utilises more than one mechanism to—for example—silence different genes. Regardless of the manner in which the unmethylated copy of KvDMR1 results in gene silencing, our present findings support the model that loss of methylation at KvDMR1 inappropriately activates this repressive function on the maternal chromosome which, in turn, leads to the pathological silencing of the imprinted candidate tumour suppressor gene, CDKN1C.³ This model is also supported by the finding that two other genes normally repressed by unmethylated KvDMR1 in the mouse (TSSC3 and SLC22A1L)²⁷ are at least partially silenced in cases of BWS with LOM at KvDMR1 (ND-M’s and MJH’s unpublished results). However, we cannot exclude the possibility that LOM at KvDMR1 and decreased expression of CDKN1C and other genes are manifestations of a widespread epimutation in the 11p15.5 imprinted domain that are not directly related mechanistically. In this regard, however, it is significant that a preliminary analysis demonstrated no aberrant methylation at the CDKN1C promoter in cells from individuals with BWS and LOM at KvDMR1 (fig 4). Thus, unlike the situation in a variety of cancers,^37,38 repression of CDKN1C in cases of BWS with LOM at KvDMR1 does not appear to be associated with promoter methylation. Nevertheless, formal proof of a causative relationship between LOM at KvDMR1 and silencing of CDKN1C and other genes awaits the generation of a mouse model with biallelic unmethylated copies of KvDMR1 and features of BWS.