Introduction

Today nearly 80 imprinted genes have been identified in humans (http://www.geneimprint.com). Imprinted genes, unlike other genes, are characterised by their mono-allelic expression pattern depending on the parental origin of the allele.1 Almost all imprinted genes have differentially methylated regions (DMR) as a mark to distinguish both alleles.2 For primary imprinted genes, the differential methylation patterns arise in gametes, and after fertilisation they are transmitted to form the somatic imprints of the offspring.2, 3 Secondary imprinted genes gain their differentially methylated patterns after fertilisation. The categories of paternally and maternally expressed genes count a comparable number of genes. For most genes from either category the methylation imprint is derived from the oocyte. Only a few imprinted genes with a paternal methylation imprint have been identified. The best-studied paternally methylated region so far is the Igf2-H19 region, which is located on mouse chromosome 7 and is well conserved on human chromosome 11p15.15. Another pair of reciprocally imprinted genes with features that are both unique and in common with the Igf2-H19 locus lies on mouse chromosome 12 and contains the paternally expressed Delta Like homologue 1 (Dlk1) gene and the maternally non-coding RNA-transcript Gene Trap Locus 2 (Gtl2).4, 5, 6, 7, 8, 9, 10 The Dlk1 gene codes for a transmembrane protein that has six epidermal growth factor like repeats in its extracellular domain and shows homology to proteins of the Notch/Delta signalling pathway. Dlk1 plays an important role in normal cellular differentiation and carcinogenesis.11 Gtl2 probably acts as RNA transcript as no consensus sequence for translation was detected.12 Dlk1 and Gtl2 are located within a 1 Mb imprinted cluster containing other imprinted genes such as the paternally expressed Dio3 gene,13 a retrotransposon-like gene (Rtl1),14 several maternally expressed non-coding RNAs, C/D small nucleolar RNAs15 and numerous microRNAs (miRNAs).14, 16

Paulsen et al9 first identified a CpG-rich tandem repeat ‘repeat area 1’ in the intergenic (IG) region of Dlk1/Gtl2 that was conserved between mouse, human and sheep. This repeat area counts seven 24-bp repeats, nine 18-bp repeats and 16 18-bp repeat motifs in the three species, respectively. The IG region was further investigated by Takada et al,10 who identified an 8 kb long IG-DMR associated with a CpG island located 15 kb upstream of Gtl2 and 70 kb downstream of the Dlk1 promotor. This IG-DMR is unmethylated on the maternal allele and hypermethylated on the paternal allele and can be subdivided into three domains of which the last 3′ domain is repeat area 1 which was shown to be differentially methylated. In the mouse, the IG-DMR is a candidate control element for the whole imprinted cluster on chromosome 12. Deletion of the IG-DMR from the maternal chromosome causes bidirectional loss of imprinting of all genes in the cluster including the miRNAs. When inherited from the paternal chromosome, the deletion of the IG-DMR does not affect imprinted gene expression.16, 17

The mechanisms by which methylation imprints are established and maintained in the human genome remain poorly characterised. One of the main blocking factors in this research is the scarcity of human research material, meaning gametes and embryos.

In this work, we studied the methylation profile of the human orthologous repeat area 1 in the IG-region of DLK1-GTL2, located on chromosome 14q32 (Figure 1). Bisulphite sequencing was used to analyse the methylation imprints of the DMR element of this region in gametes, preimplantation embryos, amniocytes and blood of ICSI babies and adults.

Figure 1
figure 1

Schematic representation of the analysed IG-DMR on human chromosome 14q32. The vertical bars (D to R) each represent a single CpG site. The primers used in the hemi-nested PCR after bisulphite treatment are shown by grey boxes and the position of the nine repeats is indicated by arrows. The binding of the forward primer IGFIII on the second repeat with one mismatch is indicated with a striped box. M, maternal; P, paternal.

Full size image

Materials and methods

Samples

The study protocol was approved by the institutional ethical committee. All samples used in the study were donated for research with informed consent of the patients.

Genomic DNA was directly extracted from peripheral blood of adults (control samples for diagnostic tests) and ICSI babies using a QIAmp Blood Maxikit (Qiagen Benelux BL, Venlo, the Netherlands). The isolation of DNA from amniocytes, obtained by routine amniocentesis carried out for prenatal diagnosis at our genetic department (one from an ICSI pregnancy and two from spontaneous pregnancies), was performed using a standard extraction procedure. The blood samples of ICSI babies were obtained in the course of the follow-up studies of ICSI children.

The oocytes, sperm cells and cleavage stage ICSI embryos were obtained and treated as described in Geuns et al.18 The oocytes used were immature at the germinal vesicle (GV) stage or metaphase I (MI) stage, or were left in vitro to spontaneously mature to the metaphase II (MII) stage. Some expanded ICSI blastocysts with a good quality of both inner cell mass (ICM) and trophectoderm were used for laser biopsy of the trophectoderm as described by Cauffman et al 2005.19 Single oocytes or pools of two or three oocytes, single embryos, trophectoderm samples and blanks were transferred to a 1.5 ml tube containing 2 μl alkaline lysis buffer (ALB; 50 mM DTT, 200 mM KOH) and stored at −80°C until use.

Bisulphite treatment

The bisulphite treatment of genomic DNA (blood samples and amniotic cells) was performed according to standard procedures. For the bisulphite treatment of oocytes, sperm cells, embryos and trophectoderm samples a bisulphite sequencing protocol based on low melting point agarose beads was used as described previousls.18

PCR

A hemi-nested PCR protocol was used to amplify the bisulphite converted DNA. In the first round, the forward primer (Eurogentec, Seraing, Belgium) IGFIV (5′-GTGGATTTGTGAGAAATGATTTYGT-3′) (nucleotide position 50975–50999 in GenBank accession number AL117190) labelled with 5′indocarbocyanin (Cy5) was used together with the reverse primer IGRIII (5′-CCATTATAACCAATTACAATACCAC-3′) (51248–51272). The IGFIV primer contained a wobble to allow binding on a methylated as well as on an unmethylated CpG-site. In the second round, the Cy5-labelled forward primer IGFIII (GTTAGTTGTTTGTGGTTTATTAGTTG) (51017–51042) together with the reverse IGRIII primer was used. A PCR-mix containing 0.4 μ M of each primer, 0.2 mM dNTPs (Amersham Pharmacia Biotech, Roosendaal, The Netherlands), 1 × PCR buffer (Applied Biosystems, Nieuwerkerk a/d IJsel, The Netherlands), 2 mM MgCl2 (Applied Biosystems), 1.25 U AmpliTaq DNA Polymerase (Applied Biosystems) in a total volume of 25 μl was used. For, on the one hand genomic DNA and on the other, preimplantation embryos, gametes and trophectoderm cells, the PCR programme in the first round was designed as follows: a denaturation step of 5 min at 94°C followed by 27 and 28 cycles, respectively, of 30 s at 94°C, 30 s at 61.5°C and 30 s at 72°C, and a final extension for 5 min at 72°C. Three microlitres of the first round was used as DNA input for amplification in the second round PCR with the following programme: 5 min denaturation at 94°C followed by 35 cycles of 30 s at 94°C, 30 s at 62°C and 30 s at 72°C, and a final extension step for 5 min at 72°C.

Fragment length analysis of single-cell PCR-products gave two fragments, one of the expected length (256 bp) and another, which is 18 bp or one repeat shorter. For the unmethylated strand, the IGFIII forward primer used in the second round not only anneals (after bisulphite treatment) to the fully orthologous first repeat, but also to the next repeat, despite one mismatch at the 5′ nucleotide of the primer (Figure 1). For the methylated strand, binding at the second repeat would involve two additional mismatches at the methylated CpG-sites D and E and therefore, binding will be much less efficient. In one PCR-cycle, the IGFIII-primers can only bind once to a DNA strand and generate either a full length PCR-fragment or an 18 bp-shorter fragment. No bias will be introduced when considering the totality of short and full-length unmethylated molecules and the totality of (few) short and full-length methylated fragments when calculating the methylation percentage. As further optimisation of the PCR conditions (including testing of other primer sets) was unsuccessful at the single-cell level, both the 256 and 238 bp fragments were cloned; taking into account that the methylation status of CpG-sites D and E could not be determined for the smaller fragment.

The PCR fragments were analysed on an ALFExpress automated sequencer (Amersham Pharmacia Biotech) and positive samples were cloned.

The PCR protocol was validated by the methylation analysis of genomic DNA (adult blood) amplified with a primer set that did not bind to the repeats. Similar results were obtained with this primer set compared to the one we used for our analysis (data not shown).

Cloning and sequencing

The PCR-products were cloned using the TOPO TA cloning kit (Invitrogen, Merelbeke, Belgium). After transformation single colonies were purified on basis of blue/white selection. The length of the insert of 16 colonies was determined by a PCR with M13 primers on lysed bacteria, followed by fragment analysis on a 2% agarose gel. PCR-products of the correct length were then automatically sequenced on the ABI 3100-Avant (Applied Biosystems).

Owing to the bisulphite sequencing procedure and repeats present in the IG-DMR sequence some incomplete or erroneous sequences can be generated. CpG-sites in sequence parts that could not be unequivocally interpreted were indicated as not analysed (NA) in Figure 2.

Figure 2
figure 2figure 2

Methylation analysis of the IG-region in (a) oocytes, (b) sperm cells, (c) trophectoderm cells, (d) amniocytes, (e) blood of ICSI-babies and (f) blood of adults. The 15 analysed CpG-dinucleotides are given in the top row of the graph (D–R). The number of clones analysed is mentioned and clones with the same pattern are grouped. For the oocytes and sperm samples, the number of analysed cells in each independent experiment is mentioned in the first column. White boxes are unmethylated CpGs, black boxes are methylated CpGs and NA indicates CpG-sites that could not be analysed. Amniocyte sample 1 and 3 are from spontaneous pregnancies whereas amniocyte sample is from an ICSI pregnancy. GV, germinal vesicle; MI, metaphase I oocyte; MII, metaphase II oocyte; SP, sperm cell.

Full size image

Checking the intermediate patterns

PCR-products were loaded on a 10% non-denaturating acrylamide gel, and after electrophoresis only bands of the correct length (256 bp) were purified using a standard protocol;20 heteroduplex bands were not cut. Afterwards the purified PCR-product was cloned.

Results

Preliminary tests

Bisulphite sequencing was carried out to analyse the methylation status of the CpG-island in gametes, preimplantation embryos and differentiated cells. The bisulphite sequencing protocol was first optimised for analysis of a small number of cells by embedding the cells in Low Melting Point (LMP) agarose beads. The PCR-protocol was adapted to the single-cell level to allow for analysis of oocytes, sperm cells and preimplantation embryos. With our single-cell PCR-protocol we were able to analyse the methylation status of 15 CpG sites (named D → R) out of 17 CpG sites of the CpG-island. The amplification efficiency of the single-cell protocol, assessed on single-cells, was 17.6% (6/34). Blanks were included in each bisulphite sequencing experiment. In total, 29 blanks were tested and none of them showed amplification. Further optimisation was not successful and this protocol was applied for the methylation analysis of human gametes and embryos, despite its low efficiency.

Gametes

When collecting oocytes, special attention was paid to the complete removal of the zona pellucida and surrounding cumulus cells. Results of the methylation profile of the human IG-DMR in different developmental stages of oogenesis (GV, MI and MII oocytes) were obtained in three independent experiments of single oocytes or pools of two or three oocytes. In the GVs, MI oocytes and MII oocytes a largely unmethylated pattern of 0.0% (0 methylated CpG sites/364 analysed CpG sites), 0.23% (1/433) and 3.87% (12/310) respectively, was found (Figure 2a).

Methylation analysis of four pools of eight or 10 sperm cells showed an overall methylated pattern of 99.3% (421/424) (Figure 2b). Pools with fewer sperm cells did not amplify in our hands, probably because of the more difficult lysis.

Preimplantation embryos

Of the 49 research embryos that were carefully selected (derived from normally fertilised ICSI oocytes and showing normal morphology and developmental timing at the moment of use), 25 cleavage stage embryos (2–10 cells) gave an amplification signal after bisulphite treatment and single-cell PCR. An average methylation pattern of 45.7% (205.8/450.2) was found with 44.4% (114/257) of the clones showing a hypermethylated pattern (74–100%), 49.4% (127/257) a hypomethylated (0–26%) and 6.2% (17/257) an intermediate pattern (26–74%) (Tables 1 and 2). The average methylation percentage was corrected for several factors (see Supplementary Materials) and clones in which four or more CpG-sites out of 15 show a different methylation status were defined as intermediate. Strikingly, the intermediate patterns appeared only after the second cleavage stage (Figure 3). When the results of the methylation analysis of a blastocyst, a compacting and a compacted embryo were pooled, an average methylation pattern of 53.3% (209/392) was found with 37.0% (10/27) of the clones showing a hypermethylated pattern, 25.9% (7/27) showing an unmethylated pattern and 37.0% (10/27) of the clones showing an intermediate pattern. This result was set apart from the mean methylation percentage of the cleavage stage embryos (2–10 cells), because their higher cell number and related higher F-factor (see Supplementary Materials) would bias the average results towards the results of the blastocyst, compacting and compacted embryo.

Table 1 Methylation analysis of 25 cleavage stage embryos
Full size table
Table 2 Overview of the methylation analysis results
Full size table
Figure 3
figure 3

Schematic overview of the methylation patterns in the 25 cleavage stages embryos, the two morulas and blastocyst. Each bar represents a single embryo and embryos with the same number of blastomeres are grouped. The number under each bar refers to the embryo numbers in Table 1. M, Morula; Bl, Blastocyst.

Full size image

Trophectoderm cells, amniocytes, genomic DNA isolated from peripheral blood of ICSI babies and adults

As imprinted genes have been reported to play an important role in placental function the selected region was also studied in trophectoderm cells.21, 22, 23 Four trophectoderm samples of good quality blastocysts were separately analysed for their methylation status at the IG-DMR. An average methylation pattern of 27.9% (198/710) was found with 14.3% (7/49) showing a hypermethylated pattern, 61.2% (30/49) of the clones showing a hypomethylated pattern, and 24.5% (12/49) showing an intermediate pattern (Figure 2c, Table 2).

To see whether the loss of the strict differential methylation pattern during the early embryonic period was permanent, further analyses of differentiated cells at different stages of the life cycle (prenatal, postnatal and adult stages) were carried out. In amniocytes samples, an average methylation pattern of 51.6% (222/430) was found, with 41.4% (12/29) of the clones showing a hypermethylated pattern, 44.8% (13/29) a hypomethylated pattern and 13.8% (4/29) an intermediate pattern. Intermediate patterns were seen in all three samples (Figure 2d, Table 2).

The analysed blood samples from ICSI babies, gave an average methylation pattern of the IG-DMR of 48.6% (215/442). Thirty percent (9/30) of the clones showed a hypermethylated pattern, 16.7% (5/30) a hypomethylated pattern and 53.3% (16/30) an intermediate pattern (Figure 2e, Table 2).

The methylation pattern of the IG-DMR in adults was tested on genomic DNA isolated from blood of four healthy adult donors, each analysed in two independent bisulphite sequencing experiment. An average methylation pattern of the IG-DMR of 51.2% (513/1003) was detected with 35.3% (24/68) of the clones showing a hypermethylated pattern, 33.8% (23/68) a hypomethylated pattern and 30.9% (21/68) an intermediate pattern (Figure 2f, Table 2).

Conversion rate

To ensure that the intermediate patterns were not due to incomplete conversion after bisulphite, the conversion rates were determined. The conversion rate for genomic DNA (amniocytes, adult and ICSI blood) was 99.77% (11888 correctly converted cytosines/11915 cytosines). At the single cell level (embryos, oocytes and sperm cells) the conversion rate was 98.81% (25415/25720).

Checking the intermediate patterns

To prove that the intermediate patterns found are a biological phenomenon and not a technical artefact of the bisulphite sequencing technique as described by Sandovici et al,21 further experiments were performed. In a parallel experiment, the PCR-products of bisulphite-treated genomic DNA isolated from blood were either directly used for cloning or were first purified on an acrylamide gel, and only the band of 256 bp was cloned. The acrylamide gel showed a band representing the 238 bp PCR-fragment (18 bp smaller) and other weaker bands, most probably heteroduplexes in addition to the 256 bp fragment. Direct cloning produced three clones out of eight (37.50%) with intermediate patterns. The experiment with the purified 256 bp fragment produced an intermediate pattern for four out of 12 clones (33.33%). All clones in this latter experiment contained the first repeat (CpG sites D and E) as expected for the 256 bp fragment.

Discussion

In this study, we report the methylation status of the IG-DMR of DLK1-GTL2 in human gametes, preimplantation embryos, amniocytes and blood samples (adults and ICSI babies). The development of the single-cell PCR for the IG-DMR was complex, and a rather low amplification efficiency (17.6%) was obtained, compared to similar protocols developed for the SNRPN-gene (28.7%) and the LIT1-gene (25.0%).18, 24 This was most probably due to the numerous tandem repeats present in the sequence of the IG-DMR.

The IG-DMR of DLK1-GTL2 of oocytes at different developmental stages was analysed using the adapted bisulphite sequencing protocol and showed a mainly unmethylated pattern. In spermatozoa on the other hand, a mainly methylated pattern was found. These methylation analysis results indicate that the region under study carries a germ-line mark. The results in human gametes are in agreement with a previous study of the orthologous region in the mouse, in which a germ-line specific methylation mark was also detected, with methylation of the paternal allele.10 After fertilisation, the strict differential methylation status of the IG-DMR was maintained in the embryos until the sixth cell stage. After the second cleavage stage intermediate methylation patterns (26–74%) were found in 6.2% of the cleavage stage embryos (2–10 cells) and in 51.8% of embryos at the morula and blastocyst stage. As imprinted genes have been reported to play an important role in placental functions,22, 23, 25 we also carried out a methylation analysis of the IG-DMR of DLK1-GTL2 in trophectoderm cells biopsied from blastocysts. Here again, part of the clones analysed (28.6%) showed intermediate methylation patterns. This indicates that the intermediate patterns are present both in the embryonic and the extra-embryonic lineage. It is not clear why the average methylation percentage in trophectoderm cells was lower than in embryos and blood samples. Further analyses of extra-embryonic tissues at different developmental stages are required to clarify this.

In mouse, the occurrence of intermediate methylation patterns for this region has not been previously reported. Analysis of mouse embryos (embryonic day 13.5) with maternal and paternal uniparental disomy (UPD) for chromosome 12 revealed strictly unmethylated and methylated patterns, respectively.10

Bisulphite sequencing of the IG-DMR of DLK1-GTL2 at later developmental stages (genomic DNA isolated from amniocytes, blood of ICSI-babies or from adults) also produced clones with intermediate methylation patterns. The percentage of intermediate patterns varied among the samples; this most probably only reflects the limitations of the methodology. It was not possible to determine the parental origin of the intermediate patterns. Although there was one SNP (RS967189, C/G polymorphism) available in the studied region, only the G allele was observed in our samples. In humans, there are three papers describing the methylation status of repeat area 1 of the IG-DMR. The publication by Lin et al,17 investigated the methylation status of a flanking region (including CpG-site T) using methylation-sensitive restriction enzymes and a Southern blot probe encompassing our selected region. The authors' analysis of DNA from mUPD14 and pUPD14 patients showed a paternal methylation mark, but intermediate patterns were not reported. These results were similar to the data obtained from mouse studies.10 The second paper involved the methylation analysis of CpG-sites D to K in four normal control blood samples using the bisulphite sequencing technique. Here, an overall methylation percentage of 32% was found; however, the presentation of the data did not allow checking individual clones for intermediate patterns.26 The third paper examined the methylation status of the IG-DMR (CpG site D-R) in four peripheral blood samples using the bisulphite sequencing technique. Detailed information about one representative sample is given and showed intermediate patterns in three of the eight clones analysed.27

The possibility exists that the intermediate patterns represent bacterial recombination artefacts. The amplification of bisulphite-treated DNA can generate heteroduplexes, as the parental template strands only differ at specific CpG-sites. Upon cloning, the heteroduplexes can be converted into a single hybrid sequence by the host's mismatch repair mechanism.21 We were able to exclude this possibility by demonstrating that cloning of purified PCR-samples as well as non-purified PCR samples produced a similar percentage of intermediate methylation patterns. It was also ruled out that the intermediate patterns stemmed from incomplete bisulphite conversion.

Other reports have also shown that the maintenance of paternal and maternal methylation patterns was not absolute. Studies at the imprint control region of H19/Igf2 in mice as well as in the human showed around 5 and 6% intermediate methylation patterns (40–50%) respectively.28, 29 These percentages are lower than the relaxation of allele-specific methylation observed at the IG-DMR.

Relaxation of imprinting, indicating biallelic expression of a gene that is normally expressed from one parental allele, and associated altered DNA methylation patterns at DMRs has been described as a polymorphic trait among individuals, being tissue specific or age related.27, 30, 31, 32, 33 Our data rather indicate a common phenomenon as intermediate patterns of the IG-DMR have been observed in different tissue samples and in all the different subjects analysed. Relaxation of the IG-DMR methylation patterns may not necessarily imply relaxation of imprinting expression at the cluster since other epigenetic modifications can contribute to the mono-allelic expression of these imprinted genes. We hypothesise that a differential methylation pattern is needed in the early cleavage stage embryo for the correct imprint establishment of (an) other imprinted gene(s) in the DLK1-GTL2 locus. Once correct imprinting of the involved genes is acquired, the IG-DMR may become redundant and a more relaxed state may be allowed.

Several studies have demonstrated that in vitro culture systems and embryo manipulations have a detrimental influence on imprinting mechanisms in animal models.34, 35, 36 The influence of in vitro culture conditions and the ICSI-procedure on the methylation patterns of the oocytes and embryos in our experiments cannot be excluded. However, in humans these embryos are the only source of research material that can be used. The proportion of intermediate patterns in the preimplantation embryos increases with time in culture. Similarly, analysis of somatic cells of pre- and postnatal stages revealed intermediate methylation patterns similar to those in the preimplantation embryos, which led us to believe that these patterns indeed arise in the preimplantation embryo physiologically, and are not an artefact of the in vitro manipulations. Recent publications about a higher incidence of BWS and AS, both imprinting disorders with a low incidence, after ICSI and IVF37, 38, 39, 40, 41, 42, 43, 44, 45, 46 address the risks of ART and highlight the need for basic molecular studies of the imprinting processes in the human.