Introduction

Alarming figures indicate that an estimated 48.5 million couples worldwide are unable to conceive after 5 years of unprotected sex1. For these couples, medically assisted reproductive technologies (ARTs) represent the best chance to conceive. However, when treatment is successful (<40%), there is an increased risk of perinatal complications even within singletons, including preterm birth, intrauterine growth restriction, low birth weight1,2,3 and the genomic imprinting disorders Beckwith-Wiedemann Syndrome (BWS)4,5,6,7,8, Angelman Syndrome (AS)6,8,9,10,11 and Silver-Russell Syndrome (SRS)12,13,14,15,16,17,18.

Genomic imprinting is an epigenetic phenomenon that restricts expression to one parental allele while the other allele is in an inactivated state. Imprinted genes are regulated by a master control switch known as a gametic differentially methylated region (gDMR) or imprinting control region (ICR). Importantly, abnormal cytosine methylation levels at the ICR can lead to imprinting disorders such as BWS, AS and SRS.

Risk association studies have found increased risks of imprinting disorders in ART children. The risk of BWS is 3–16 times greater in children in the ART population compared to those in the general population4,5,7,15,19,20,21,22,23,24. Epigenetic errors at KCNQ1OT1, namely maternal hypomethylation, are observed in more than 90% of ART BWS cases compared to 50% in the general population4,5,7,15,19,20,21,22,23,25,26, while H19 maternal hypermethylation occurs in 17% of ART BWS cases compared to 5% in the general population4,21,25,27. Of the small number of patients analyzed, 46% of AS patients conceived by ARTs possessed imprinting defects at the SNRPN ICR compared to 5% in the general population9,10,11,28, while 92% of SRS patients conceived by ARTs harboured H19 hypomethylation compared to 40% in the general population12,13,14,15,16,17,18, The overall risk for an imprinting disorder such as BWS, AS or SRS in ART children is estimated to be around 1 in 5,0003. Thus, disparity has arisen concerning the frequency of imprinting errors produced by ARTs in humans compared to mice, as mouse studies have identified between 10% to 90% of treated preimplantation embryos showing abnormal imprint maintenance29,30,31,32,33. However, one key difference in studies between these species is the time of analysis. The majority of mouse studies have focused on preimplantation or mid-gestation development, while human studies are primarily retrospective studies of ART children with imprinting disorders. Consequently, we sought to determine whether donated human ART-produced preimplantation embryos harbour aberrant imprinted methylation at similar incidences to that observed in the mouse29,30,31,32,33. Additionally, we analyzed whether short or extended culture produces a greater frequency of imprinted methylation errors and whether aberrant imprinted methylation correlates with parental biometrics or clinical treatment. We analyzed methylation levels at SNRPN, KCNQ1OT1 and H19 ICRs in individual good to high quality day 3 cleavage and blastocyst stage ART-produced human embryos.

Results

Imprinted methylation in control samples

Imprinted DNA methylation at the SNRPN, KCNQ1OT1 and H19 ICR was first assessed in untreated human buccal cell (Bu) samples from 4 young, non-patient adults. Bisulfite clonal sequencing was used to analyze 20–24 CpGs per gene. For all controls, a total of 30–65 clones were sequenced to obtain representative DNA strands. Sequences with identical CpG methylation profiles and unconverted cytosines were considered to be identical and were included once to eliminate clonal bias. Each region of analysis included a single nucleotide polymorphism(s) (SNP) that when present in heterozygous samples could distinguish between parental alleles (Supplementary Table 1 online). Since we did not have access to patient samples, we consider the methylated strands as the presumptive paternal H19, maternal SNRPN and maternal KCNQ1OT1 alleles and the unmethylated strands as the maternal H19, paternal SNRPN and paternal KCNQ1OT1 alleles, as was done in previous studies34,35.

For the SNRPN ICR, a 360 bp-region was analyzed comprising 24 CpGs and a G/A SNP (Rs220029) that occurs at a general population frequency of 84.8% and 15.2%, respectively (Fig. 1A). All control samples were homozygous at this SNP (Supplementary Table 1 online) and thus no allelic assignment could be made. Total SNRPN methylation levels in buccal cell controls (~1000 cells) were Bu1-1000 46%, Bu2-1000 45%, Bu3-1000 43% and Bu4-1000 40% (Fig. 1B). Since buccal samples exhibited a mean SNRPN methylation level less than anticipated (43%), we analyzed SNRPN methylation in human embryonic stem cells (hESCs), an undifferentiated cell type that more closely matched preimplantation embryos. In hESCs, SNRPN methylation levels were 41% (Fig. 1B), consistent with those in buccal cells. To assess cell numbers similar to blastocyst and day 3 embryos, methylation levels were analyzed in approximately 100, 50 (Fig. 1C) and 5–10 cells (denoted hereafter as 10 cells) (Fig. 1D) for Bu1 and Bu3 samples. Total SNRPN methylation levels were Bu1-100 39%, Bu1-50 41% (Fig. 1C), Bu1-10 44% and 41% (Fig. 1D) and Bu3-100 49%, Bu3-50 44% (Fig. 1C), Bu3-10 38% and 42% (Fig. 1D). Thus within sample, methylation level mean and standard deviation were 42.2 ± 2.8 for Bu1 and 43.2 ± 4.0 for Bu3.

Figure 1
figure 1

SNRPN imprinted methylation in buccal cell and human embryonic stem cell (hESC) control samples.

(A) Map of the SNRPN region analyzed. Accession numbers are located below genes, primer locations are marked with arrows and SNPs are indicated by arrowheads. Methylation analyses in (B) four buccal cell (Bu1-4) and human embryonic stem cell (hESC) control samples with ~1000 cells, (C) in buccal cell samples with ~100 and ~50 cells, representing blastocysts and (D) with buccal cell samples ~10 cells, representing day 3 cleavage embryos. Each group of circles represents an individual human sample. Each line is an individual DNA strand. Methylated CpGs are filled black circles and unmethylated CpGs are open circles. Percent methylation is indicated above each set of DNA strands for a gene or parental allele and was calculated as the number of methylated CpGs divided by the total number of CpG dinucleotides.

For the KCNQ1OT1 ICR, a 265 bp-region was analyzed encompassing 22 CpGs34 and a G (94.7%)/A (6.3%) SNP (Rs56134313), that eliminated the first CpG (Fig. 2A). All controls were homozygous at the KCNQ1OT1 SNP (Supplementary Table 1 online). Total KCNQ1OT1 methylation levels in control samples were Bu1-1000 63%, Bu2-1000 57%, Bu3-1000 58% and Bu4-1000 65% (Fig. 2B). Since the mean KCNQ1OT1 methylation level was greater than anticipated (60%), KCNQ1OT1 methylation was assessed in hESCs. KCNQ1OT1 methylation levels were hESC-1000 65% (Fig. 2B), consistent with those in buccal cells. At cell numbers similar to blastocyst and day 3 embryos, KCNQ1OT1 methylation levels were Bu1-100 57%, Bu1-50 65%, (Fig. 2C), Bu1-10 64% and 64% (Fig. 2D) and Bu3-100 54%, Bu3-50 57% (Fig. 2C), Bu3-10 54% and 57% (Fig. 2D). Thus within sample, methylation level mean and standard deviation were 62.6 ± 3.2 for Bu1 and 56.0 ± 1.8 for Bu3.

Figure 2
figure 2

KCNQ1OT1 imprinted methylation in buccal cell and hESC control samples.

(A) Map of the KCNQ1OT1 region analyzed. Methylation analyses in (B) buccal cell (Bu) and human embryonic stem cell (hESC) control samples with ~1000 cells, (C) in buccal cell samples with ~100 and ~50 cells, representing blastocysts and (D) with buccal cell samples ~10 cells, representing day 3 cleavage embryos. See Figure legend 1 for details.

Samples assessed for KCNQ1OT1 methylation levels were also analyzed for DNA methylation at the H19 ICR. We initially began our analysis for a 234 bp-region within the H19 imprinting control region that included 18 CpGs34 and a common A (33.6%)/C (66.4%) SNP (Rs2071094) (Fig. 3A). However, we observed biased allelic recovery and subsequently found two additional SNPs present in the forward and reverse inner nested primers. Thus, we designed new internal primers for a larger 249 bp-region within the H19 ICR containing 20 CpGs, the Rs2071094 (A, 33.6%; C, 66.4%) and the Rs2107425 SNP (G, 55.5%; A, 44.5%) SNP (Fig. 3A). For buccal cell samples, Bu3 was heterozygous at both H19 SNPs, Bu4 was heterozygous at one SNP, while Bu1 and Bu2 were homozygous for both H19 SNPs (Supplementary Table 1 online). Samples Bu1-1000 and Bu2-1000 had total H19 methylation levels of 57% and 61%, respectively. Sample Bu3-1000 had 96% methylation on the presumptive paternal H19 allele and 11% methylation on the presumptive maternal H19 allele (56% total methylation), while Bu4-1000 had 94% and 11% methylation on the presumptive paternal and maternal H19 alleles, respectively (52% total methylation) (Fig. 3B). Thus, total methylation levels fell within a mean (56%) expected for paternally methylated and maternally unmethylated alleles. For smaller cell numbers, total H19 methylation levels were Bu1-100 55%, Bu1-50 60% (Fig. 3C), Bu1-10 63% and 53% (Fig. 3D) and Bu3-100 59%, Bu3-50 50% (Fig. 3C), Bu3-10 52% and 54% (Fig. 3D), with 94–98% and 3–12% methylation on the presumptive paternal and maternal H19 alleles, respectively. Thus within sample, methylation level mean and standard deviation were 57.6 ± 4.0 for Bu1 and 54.2 ± 3.5 (Pat 95.6 ± 1.5; Mat 8.8 ± 3.6) for Bu3.

Figure 3
figure 3

H19 imprinted methylation in buccal cell control samples.

(A) Map of the H19 region analyzed. (Methylation analyses in (B) buccal cell (Bu) and human embryonic stem cell (hESC) control samples with ~1000 cells, (C) in buccal cell samples with ~100 and ~50 cells, representing blastocysts and (D) with buccal cell samples ~10 cells, representing day 3 cleavage embryos. Grey circles are not included in methylation analyses as they represent a C/T SNP that cannot be distinguished following bisulfite conversion. Alleles are separated into presumptive maternal (Mat) and paternal (Pat) strands in samples with heterozygous SNPs. See Figure legend 1 for details.

Given the SNRPN, KCNQ1OT1 and H19 methylation levels in all control samples, conservatively, we considered a methylation range of 4 times the standard deviations above/below the mean as a normal methylation level. For SNRPN, the mean methylation level was 42.2% ± 3.0, generating a 30%-54% normal methylation range. For KCNQ1OT1, the mean methylation level was 60.0% ± 4.4, giving a normal methylation range of 42%–78%. The mean methylation level for H19 was 56.0% ± 4.1, generating a 40%–72% normal methylation range. For embryos with heterozygous SNPs, conservatively ≥70% methylation on the presumptive maternal SNRPN, maternal KCNQ1OT1 and paternal H19 alleles and ≤20% methylation on the presumptive paternal SNRPN, paternal KCNQ1OT1 and maternal H19 alleles were considered as normal methylation levels.

Aberrant imprinted methylation in day 3 embryos

During fertility treatment, embryos were cultured to day 3, after which embryos were transferred to the mother, cryopreserved and stored for future cycles, or cultured to the blastocyst stage then cryopreserved and stored for future cycles. For identification purposes, embryos were given an alphanumeric ID that included patient number (1–23), freeze stage [day 3 cleavage (C) or blastocyst (B)] and embryo number (1–6), for example “9C2” represents patient 9, day 3 cleavage embryo 2. Individual, cryopreserved day 3 cleavage embryos were analyzed for maintenance of imprinted methylation. For all day 3 and blastocyst-stage embryos, a total of 30–65 clones were sequenced to obtain representative DNA strands and to sequence all possible unique DNA strands following thawing and bisulfite treatment. Data were obtained for 9 out of 12 day 3 embryos for SNRPN; 7 out of 12 day 3 embryos for KCNQ1OT1; and 7 out of 12 day 3 embryos for H19.

SNRPN is normally methylated on the silent maternal allele, while the paternal allele is unmethylated. All day 3 embryos were homozygous at the Rs220029 SNP (Supplementary Table 1 online) and thus were examined for total methylation levels. Of the 9 day 3 cleavage embryos analyzed, normal methylation levels were observed for 4 embryos (Fig. 4A). By comparison, 5 embryos had abnormal SNRPN methylation levels, with 4 embryos exhibiting aberrant hypermethylation (1C1, 62%; 1C5, 67%; 1C6, 59%; 18C1, 62%) and 1 embryo (21C1) displaying aberrant hypomethylation of 18%. Overall, 56% of day 3 cleavage embryos had abnormal SNRPN imprinted methylation.

Figure 4
figure 4

Methylation of the (A) SNRPN, (B) KCNQ1OT1 and (C) H19 ICRs in day 3 human cleavage-stage embryos. Each group of DNA strands is an individual day 3 embryo with embryo ID (top left) and percent methylation and presumptive maternal/paternal allele designation (top right) indicated. Normal (N) and abnormal (A) embryos are designated after percent methylation (top right). The pre-freeze and post-thaw cell numbers, respectively, are indicated in brackets beside each embryo name. See Figure legend 1 for details.

KCNQ1OT1 is also normally methylated on the silent maternal allele, while the paternally inherited allele is unmethylated. All 7 day 3 embryos were homozygous at the Rs56134313 SNP (Supplementary Table 1 online) and thus, total methylation levels were analyzed. One embryo had methylation levels within the normal range (Fig. 4B). Of the remaining 6 embryos, 1 embryo had abnormal hypermethylation (9C1, 80%) while 5 embryos exhibited aberrant KCNQ1OT1 hypomethylation (12C1, 19%; 7C1, 33%; 7C2, 35%; 6C1, 22%; 4C1, 19%). In total, 86% of day 3 cleavage embryos had aberrant KCNQ1OT1 imprinted methylation.

H19 is normally methylated on the paternal allele, while the maternally inherited allele is unmethylated. Three day 3 cleavage embryos (4C1, 7C1, 7C2) were heterozygous at both Rs2071094 and Rs2107425, 1 embryo (6C2) was heterozygous at Rs2071094 and 2 embryos (6C1, 9C1) were heterozygous at Rs2107425 (Supplementary Table 1 online), allowing for allelic assignment. Only one embryo (3C1) was homozygous at the Rs2071094 and Rs2107425 SNPs and was examined for total methylation levels. Out of 7 day 3 embryos, 2 had a normal methylation pattern with methylation ≥70% on the presumptive paternal allele and hypomethylation on the presumptive maternal allele (Fig. 4C). Of the remaining 5 embryos, 2 showed loss of methylation on the presumptive paternal H19 allele (6C1, 35%; 6C2, 61%) and 2 displayed a gain of methylation on the presumptive maternal allele (7C1, 85%; 7C2, 71%). Finally, for the homozygous embryo (3C1), there was a loss of total H19 methylation (38%). Overall, 71% of day 3 cleavage embryos were abnormally hypo- and/or hypermethylated at H19. Furthermore, of the 6 embryos successfully assessed for both KCNQ1OT1 and H19 methylation, 3 embryos (50%) displayed aberrant methylation levels at both genes (7C1; 7C2; 6C1).

Abnormal imprinted methylation in blastocyst stage embryos

Individual, cryopreserved blastocysts were also analyzed for maintenance of imprinted methylation. Data were obtained for 12 out of 15 blastocysts for SNRPN; 13 out of 14 blastocysts for KCNQ1OT1; and 14 out of 14 blastocysts for H19. For SNRPN, 3 blastocyst-stage embryos (22B1, 9B2, 17B1) were heterozygous at Rs220029, while the remaining 9 embryos were homozygous at the Rs220029 SNP (Supplementary Table 1 online). Four embryos had total methylation levels within the normal range (30%–54%) (Fig. 5). Of the 8 remaining embryos, 3 homozygous embryos showed a gain of total SNRPN methylation (10B3, 63%; 14B3, 57%; and 14B4, 62%) and 2 homozygous blastocysts exhibited SNRPN hypomethylation (16B2, 28%; and 23B1, 15%), while 1 heterozygous blastocyst (9B2) exhibited a gain of paternal SNRPN methylation (24% Pat) and 2 heterozygous blastocysts possessed both a loss of maternal SNRPN methylation and a gain of paternal SNRPN methylation (17B1, 65% Mat, 26% Pat; and 22B1, 48% Mat, 26% Pat) (Fig. 3). In total, 67% of blastocyst embryos exhibited abnormal SNRPN imprinted methylation.

Figure 5
figure 5

Methylation of the SNRPN ICR in human blastocyst-stage embryos.

Each group of DNA strands is an individual blastocyst with embryo ID (top left) and percent methylation and presumptive maternal/paternal allele designation (top right) indicated. Normal (N) and abnormal (A) embryos are designated after percent methylation (top right). See Figure legend 1 for details.

For KCNQ1OT1, all embryos were homozygous at the Rs56134313 SNP (Supplementary Table 1 online), allowing total methylation levels to be determined. Normal KCNQ1OT1 methylation levels (42%–78%) were observed in 9 blastocysts (Fig. 6). For the remaining 4 blastocysts, a loss of KCNQ1OT1 methylation was observed (14B2, 16%; 11B1, 19%; 19B1, 37%; and 2B2, 39%). Overall, 4 out of 13 (31%) blastocysts had abnormal KCNQ1OT1 methylation levels.

Figure 6
figure 6

Methylation of the KCNQ1OT1 ICR in human blastocyst-stage embryos.

Each group of DNA strands is an individual blastocyst with embryo ID (top left) and percent methylation and presumptive maternal/paternal allele designation (top right) indicated. Normal (N) and abnormal (A) embryos are designated after percent methylation (top right). See Figure legend 1 for details.

The same 14 embryos analyzed for KCNQ1OT1 imprinted methylation were assessed for H19 imprinted methylation. Three blastocysts (4B1, 8B1, 14B2) were heterozygous at Rs2071094 and Rs2107425, 3 blastocysts (9B1, 19B1, 2B1) were heterozygous at Rs2071094 and 2 blastocysts (2B2 and 13B1) were heterozygous for Rs2107425 (Supplementary Table 1 online). The remaining 6 blastocysts (14B1, 11B1, 15B1, 21B1, 4B2 and 20B1) were homozygous for both H19 SNPs (Supplementary Table 1 online). All blastocysts, except 2, fell within the normal H19 methylation range (40%–72%) (Fig. 7). One blastocyst displayed a loss of total H19 methylation (20B1, 29%) and one displayed abnormal gain of maternal H19 methylation (14B2, 87% Pat, 36% Mat). Overall, 14% of blastocysts had an abnormal H19 methylation profile. Blastocyst 20B1, with aberrant H19 methylation, had normal KCNQ1OT1 methylation, while blastocyst 14B2 had abnormal methylation at both H19 and KCNQ1OT1. In total for all three genes, 76% day 3 embryos and 50% blastocysts exhibited abnormal imprinted methylation (Supplemental Figure 1 online).

Figure 7
figure 7

Methylation of the H19 ICR in human blastocyst-stage embryos.

Each group of DNA strands is an individual blastocyst with embryo ID (top left) and percent methylation and presumptive maternal/paternal allele designation (top right) indicated. Normal (N) and abnormal (A) embryos are designated after percent methylation (top right). See Figure legend 1 for details.

Intra-patient comparison of imprinted methylation in embryos at different preimplantation stages

The design of this study allowed multiple embryos from the same patient to be compared for their imprinted methylation status. Out of 22 patients for whom data were obtained, 10 patients had more than one embryo analyzed (Supplementary Table 2 online). For two patients, 6 and 7, all in vitro-produced embryos experienced perturbations in imprinted methylation (KCNQ1OT1/H19 or H19). The remaining 8 patients had a portion of embryos with normal and a portion of embryos with abnormal methylation levels. For patient 1, 3/6 day 3 embryos had aberrant SNRPN imprinted methylation. For patients 2, 10, 14 and 16, 1/2 (abnormal KCNQ1OT1), 1/3 (abnormal SNRPN), 3/5 (aberrant KCNQ1OT1/aberrant H19; aberrant SNRPN) and 1/2 (abnormal SNRPN) blastocysts had aberrant methylation levels, respectively. Finally, three patients had both day 3 cleavage and blastocyst-stage embryos. For patient 21 and 4, the day 3 embryos had aberrant methylation (abnormal SNRPN; abnormal KCNQ1OT1), while the blastocysts displayed normal methylation levels. Finally, for patient 9, 1 day 3 embryo and 1 blastocyst possessed normal methylation levels, while 1 day 3 embryo and 1 blastocyst had perturbed methylation (abnormal KCNQ1OT1; abnormal SNRPN). Overall, all 10 patients had at least one embryo with aberrant imprinted methylation. Since there were embryos with and without imprinted methylation errors from the same patient and there were genes with and without aberrant imprinted methylation in the same embryo, methylation errors were likely stochastic in nature. Furthermore, the presence of methylation errors in both day 3 cleavage and blastocyst-stage embryos indicates that methylation errors arise as early as the 6–8 cell stage and that extended culture does not pose a greater risk for imprinting errors than short culture.

Correlation between parental biometrics, clinical treatment and aberrant imprinted methylation

Medical records were examined for parental biometrics, clinical treatment and pregnancy outcomes. Clinical pregnancy rates for fresh embryo transfers as determined by gestational sac by ultrasound were 65% for the same cycle in which the surplus embryos were cryopreserved and donated. Live birth rate was 61% and live births/embryo transfer was 36% (Supplementary Table 3 online). Of all live births, 45% (9/20) of newborns (2 singletons, 3 sets of twins and 1 of the triplets) were outside clinically normal birth weight, with 1 high birth weight (>4000 g), 5 low birth weight (<2500), 1 very low birth weight (<1500 g) and 2 extremely low birth weight (<1000 g). To discern any confounding factors related to parental biometrics or clinical treatment, embryos with methylation levels in the normal range were compared to embryos with aberrant methylation for maternal age, patient diagnosis, induction method, hormone dose, stimulation response (E2 levels), fertilization method (IVF/ICSI) and embryo grade (Supplementary Table 4 online). Note that for all embryos, the same conditions and reagents were used for in vitro culture and slow-freezing cryopreservation and thus no comparison could be made. For this analysis, the premise was that each embryo could have a different response to influences/exposures, although we acknowledge that embryos from the same mother may have similar exposures to maternal factor treatment. To make a comparison at the patient level for maternal age, hormone dose and estrogen response, separate analyses was also done for patients with only one embryo (12/22), since the remaining 10 patients with more than one embryo had a least one embryo with abnormal methylation. Data from both stages were combined for analyses, except for embryo grade.

Maternal age range for patients in this study was 23–42 years. Mean maternal age for embryos with normal methylation levels was 34 years while that for embryos with aberrant methylation was 33 years (Fig. 8A), which was not statistically different (p = 0.21). Excluding patients with more than one embryo, maternal age for embryos with normal methylation levels was 33 years while that for embryos with abnormal methylation was 30 years (results not shown) (p = 0.38). Multiple etiologies contributing to infertility were diagnosed in patients. The four most common patient diagnoses were bilateral tubal obstruction/occlusion (BTO, 29.4% normal, 26.9% abnormal), male factor (MF, 17.6% normal, 15.4% abnormal), blocked tubes with endometriosis (BTO + ENDO, 11.8% normal, 15.4% abnormal) and polycystic ovarian syndrome (PCOS, 11.8% normal, 11.5% abnormal) (Fig. 8B). Thus, patient diagnoses were not statistically different between embryos with normal and abnormal methylation levels (p > 0.99). For induction method, Nafarelin (Synarel®) and Follitropin-alpha (Gonal-F®) was the most common hormone combination for patients in both normal (70.6%) and abnormal (69.2%) embryo groups, followed by Urofollitropin (Bravelle®) and Ganirelix Acetate (Orgalutran®) (11.8% normal and 7.7% abnormal) (Fig. 8C). Thus, no significant difference was observed for hormone induction method (p = 0.80). Mean hormone dose and estrogen response (E2) was calculated at 2894.1 IU and 15084.4 pM/L for the normal group and at 2361.5 IU and 12484.7 pM/L for the abnormal group (Fig. 8D,E), which was not significantly different (p = 0.18 and 0.20, respectively). Excluding patients with more than one embryo, dose and estrogen response (E2) was 4150 IU and 15394.3 pM/L for the normal group and 2233.3 IU and 11546.6 pM/L for the abnormal group (results not shown), which was not significantly different (p = 0.06 and p = 0.43, respectively). For fertilization method, percentage of embryos in the normal group was 62.5% IVF and 47.5% ICSI and in the abnormal group was 57.7% IVF and 42.3% ICSI (Fig. 8F), which did not differ statistically (p = 0.33). For day 3 embryo grade, embryos with normal methylation levels exhibited a grade of slight C/G2 (slC/G2) (3 embryos) and C/G3 (1 embryo) while those with abnormal methylation levels had a grade of A/B/G1 (4 embryos), slC/G2 (8 embryos) and C (1 embryo) (Fig. 8G). Importantly, embryos transferred to patients (Supplementary Table 3 online, 28 A/B/G1, 15 slC/G2 and 12 C/G3) had similar grading information to those that were frozen. For blastocysts, 10 of the 13 embryos with normal methylation levels had grading information; 3 were AA, 1 AB, 1 BA, 2 BB, 1 BC, 1CA and 1 CB (Fig. 8H). For embryos with abnormal methylation levels, 6 of the 13 had grading information: 5 were AA and 1 BA. These grades were not statistically different (p = 0.25). A comparison of these grading criteria separately showed that for stage (all 26 embryos included), embryos with normal methylation (5 stage 2, 2 stage 3, 6 stage 4) were not significantly different (p > 0.99) from embryos with abnormal methylation (1 stage 1, 7 stage 2, 3 stage 3, 2 stage 4). For ICM grade, embryos with normal methylation (4 A, 4 B and 2 C) were not statistically different (p = 0.40) than embryos with abnormal methylation levels (5 A, 1 B). For TE grade, embryos with normal methylation (5 A, 4 B, 1 C) were not statistically different (p = 0.60) from embryos with abnormal methylation levels (6 A). Overall, no specific parameter was identified to have an association with abnormal imprinted methylation. Importantly, we found that embryos of the highest quality with day 3 A/B/G1 and blastocyst AA grading can have with abnormal methylation.

Figure 8
figure 8

Patient characteristics and embryo outcome for embryos with normal and abnormal imprinted methylation.

Day 3 cleavage and blastocyst-stage embryos exhibiting normal imprinted methylation (purple bars; n = 17) were compared to those with abnormal methylation (orange bars; n = 26) for (A) maternal age (t-test), (B) patient diagnosis (KS test), (C) induction method (KS test), (D) hormone dose (t-test), (E) estrogen levels (t-test) and (F) fertilization method (KS test). Means are indicated by black line for maternal age, hormone dose and estrogen levels. (G) Day 3 embryos with normal (n = 4) and abnormal methylation (n = 13) (no statistical analysis). (H) Blastocysts with normal [n = 13 (stage), n = 10 (grade)] and abnormal methylation [n = 13 (stage), n = 6 (grade)] were compared for embryo stage and grade (KS test). No significant difference was observed for any parameter between embryos with normal and abnormal methylation. BTO, bilateral tubal obstruction/occlusion; MF, male factor; ENDO, endometriosis; PCOS, polycystic ovarian syndrome; AMA, advanced maternal age; IDIO, idiopathic; ANOV, anovulatory; TD, tubal disease; (donor), donor sperm; FPES Fresh/frozen percutaneous epididymal/testicular sperm aspiration sample; Syn, Synarel® (Nafarelin); G-F, Gonal-F® (Follitropin-alpha); Brav, Bravelle® (Urofollitropin); Org, Orgalutran® (Ganirelix Acetate); Lup, Lupron® (Leuprolide Acetate); Rep, Repronex® (Menotropins); Men, Menopur® (Menotropins); Pur, Puregon® (Follitropin-beta); Fert, Fertinorm® (Urofollitrophin); IVF, in vitro fertilization; ICSI intracytoplasmic sperm injection. See methods for embryo grades.

Discussion

Although mouse models have been instrumental in analyzing the effects of ARTs on genomic imprinting in oocytes and early embryos, it is important to assess the effects of these technologies in donated human counterparts. This is especially important, as imprinting errors were perceived to be more common in mouse preimplantation embryos than in ART-conceived children. In this study, we observed that 76% day 3 embryos exhibited perturbed imprinted methylation, with 56%, 86% and 71% day 3 embryos possessing aberrant SNRPN, KCNQ1OT1 and H19 imprinted methylation, respectively. Furthermore, 50% blastocyst-stage embryos exhibited abnormal methylation levels with 67%, 31% and 14% blastocysts having aberrant SNRPN, KCNQ1OT1 and H19 imprinted methylation, respectively. Both losses and gains of imprinted methylation were observed and in some case, both within the same embryo (ex. 17C1, 22B1). Additionally, 50% of day 3 and one blastocyst embryo exhibited both KCNQ1OT1 and H19 imprinted methylation perturbations (6C1, 7C1, 7C2, 14B2). This is similar to the multi-locus loss of imprinting we previously observed in the mouse29 and that others have reported in BWS and SRS children4,15,21,22,27,36,37,38,39.

Very few studies have examined the effects of ARTs on genomic imprinting in donated human preimplantation embryos34,35,40,41,42,43. Moreover, these studies were primarily performed on poor quality embryos that were unsuitable for transfer. Nevertheless, their results were similar to what is reported here. For SNRPN, 8/9 day 3 embryos (89%) possessed a loss or gain of methylation42. For KCNQ1OT1, 7/67 day 3 embryos (10%)40 and 9/16 poor quality blastocysts (56%) harboured aberrant methylation34. Finally for H19, 3 studies reported aberrant imprinted methylation in 6/32 day 3 embryos (17%)41, 9/21 poor quality morula-blastocysts (43%)35 and 5/60 blastocysts (8%)40, while the remaining study did not observe any alterations in H19 imprinted methylation in 8 low quality blastocysts (0%)34. In addition to these genes, previous studies identified 11/65 day 3 embryos (17%) with abnormal PEG1 imprinted methylation40 and 18/24 day 3 embryos (75%) with aberrant GTL2 imprinted methylation43. All together, our study along with previous publications demonstrates that the frequency of imprinting errors in human donated preimplantation embryos (6–89%) occurs at a similar frequency to that produced in mouse preimplantation embryos (10–90%)29,30,31,33.

Of the above studies, two examined imprinted methylation in good quality, in vitro produced embryos. For KCNQ1OT1, 2/5 high quality (40%) blastocysts harboured aberrant methylation34, which was similar to what we report here (4/13; 31%). For H19, 0/5 high quality (0%) morula-blastocysts35 and 0/5 high quality blastocysts (0%) possessed aberrant methylation34. This contrasted with our study where we observed 2/14 blastocysts (14%) with aberrant H19 methylation. This discrepancy may relate to the number of embryos analyzed in these studies.

The design of our study allowed comparison of short culture to day 3 cleavage stages and extended culture to the blastocyst stage. Our data together with previous studies found imprinted methylation errors at both stages; SNRPN day 3 (56%, 89%) versus blastocysts (67%); KCNQ1OT1 day 3 (86%, 10%) versus blastocysts (31%, 40%, 56%); and H19 day 3 (71%, 17%) versus blastocysts (14%, 8%, 0%)34,35,40,41. Thus, the presence of methylation errors in embryos undergoing both short (55% embryos) and extended (31% embryos) culture indicates that methylation errors arise as early as the 6–8 cell stage. Furthermore, extending culture from day 3 to the blastocyst stage does not appear to pose any greater risk for imprinting errors. Consequently, our study offers additional support for extended culture to the blastocyst stage to select the most developmentally competent embryos.

Although the frequency of imprinting errors was similar between mice and human preimplantation embryos, disparity still exists between frequencies of imprinting errors in human preimplantation embryos compared to frequencies of imprinting errors reported in ART children. One explanation for this discrepancy may be that imprinting errors in the early embryo lead to reduced levels of implantation or pregnancy failure. Alternatively, blastomeres with aberrant imprinted methylation may be preferentially relegated to the extraembryonic lineages. Previous studies in the mouse provide support for the latter explanation, since we and others have observed a selective loss of imprinting in the placenta compared to the embryo in midgestation mouse embryos following preimplantation development in culture32,44,45.

Infertility rates have increased around the world1,46. Advanced maternal age (>35 years) is directly related to this rise, consequently leading to the question of whether delayed childbearing in ART women may contribute to increased imprinting errors in ART children. Additionally, current evidence indicates that the supra-physiological hormonal milieu of ovarian stimulation may produce adverse outcomes in ART pregnancies. For example, similar incidences of low birth weight and preterm low birth weight were present in ART children produced from donor oocytes from fertile women compared to oocytes from women compared to oocytes from subfertile mothers47. This birth weight variation in in vitro-conceived children may be explained by alterations in DNA methylation levels at growth-related genes, as detected in newborn cord blood and placenta48. With respect to imprinting disorders, ovarian stimulation has also been linked to BWS and AS in ART-conceived children7,11,38 and for some of these children, the only procedure used was ovarian stimulation11,38. Our comparison of maternal age, induction method, hormone dosage levels and stimulation response in embryos with and without aberrant methylation revealed no significant difference between these groups. These results were not all that surprising, since embryos with and without methylation errors may have had similar exposures to maternal factor treatment and/or parental biometrics; and all embryos were generated using supra-physiological hormone dosages and the same conditions for in vitro culture and slow-freezing cryopreservation. Similarly, no significant difference in fertilization method (IVF/ICSI) or blastocyst grade was observed between embryos with normal or abnormal imprinted methylation. However, it should be noted that even the highest quality day 3 cleavage (A/B/G1) and blastocyst-stage (AA) embryos harbour abnormal methylation levels. This finding has significant bearing on future studies employing high quality embryos as their control group. One further observation of note was that two embryos (19B1, 20B1), produced via donor sperm for male factor infertility, possessed abnormal imprinted methylation. This suggests that imprinting errors in these embryos cannot be explained by inherent infertility, but instead may point to ART-induced errors. Further studies are required to investigate imprinted methylation errors in in vitro-produced embryos using donor oocytes and sperm.

There were several limitations of this study. Similar to other studies on ART human embryos, our investigation lacks naturally conceived controls, which is unavoidable. Additionally, due to limited availability of donated embryos, this study and others employed small numbers in analyses. However, the statistical analyses used in this type of study remains valid within the embryo population analyzed and may allow cumulative analysis of larger sample sizes in the future. Finally, although our study controlled for operating procedure in the clinic, donated embryos analyzed here were obtained from a single fertility clinic.

Going forward, future research should focus on determining differences between human embryos with and without imprinting errors with respect to embryo properties, the timing and origin of these errors, as well as the molecular factors responsible for inducing imprinted methylation errors in ART embryos. Animal models will be instrumental in these studies prior to investigation in human embryos.

Materials and Methods

Donated human embryos

Twenty-three patients who had completed their fertility treatment at The Fertility Clinic at London Health Sciences Centre donated for research 24 day 3 cleavage and 29 blastocyst-stage human embryos that were no longer needed for their treatment. Buccal cells (B1-B4) were obtained from 4 healthy, non-patient adults (<30 years old). Research ethics approval was obtained through the Western University’s Health Science Research Ethics Board (102659) and the methods were carried out in accordance with the approved guidelines. Informed consent was obtained from patients donating embryos and non-patient adults providing buccal cell samples. All embryos were cultured in the glucose/phosphate-free preimplantation stage 1 (P1) culture medium (Irvine Scientific, California) to day 3, then in Blastocyst Medium (BM) in a sequential media protocol (Irvine Scientific, California) to the blastocyst stage. Embryos were slow frozen between the years 2000–2007 and thawed between October 2013-August 2014. Slow freezing was performed according to the Testart’s (propanediol) freezing method49 using Sydney IVF Cryopreservation Kits.

Day 3 human embryos were graded by blastomere number and morphological fragmentation levels by either the former A through F grading system or the currently used G1 through G6 system: A, even, regular, no fragments; B, uneven, irregular, no fragments; slight C (slC), slight fragmentation; C, minor (<25%) fragmentation; D, major (between 25–50%) fragmentation; E, extensive (>50%) fragmentation; F, degenerate; or by fragmentation levels: G1, <5% fragmentation; G2, 5–10% fragmentation; G3, 11–25% fragmentation; G4, 26–50% fragmentation; G5, >50% fragmentation; and G6, degenerate50,51,52. Following thawing, the majority of embryos were G1-G3 grade and had an average of 4 cells (data not shown).

Blastocyst grading was according to blastocyst cavity size/hatching, inner cell mass characteristics and trophoblast characteristics giving a numeric-alpha-alpha score based on the Gardner and Schoolcraft scoring system53. Cavity size or hatching score was graded as 1, early blastocyst with cavity less than half the embryo volume; 2, blastocyst with cavity greater than half the embryo volume; 3, full blastocyst, cavity full; 4, expanded blastocyst, cavity expanded beyond earlier embryo size with thinning zona; 5, hatching blastocyst; 6, hatched blastocyst. Inner cell mass (ICM) grading was A, tightly packed ICM, many cells; B, loosely grouped ICM, several cells; and C, very few cells and trophectoderm was graded as A, many cells with cohesive epithelium; B, few cells with loose epithelium; and C, very few large cells. All embryos were immediately processed for methylation analysis following thawing.

Isolation of Control Cells

Buccal cells were collected using the end of a sterile 20 μL pipet tip and diluted into approximately 1000, 100, 50 and 5–10 cells in 20 μL of 1 X PBS (Phosphate-Buffered Saline). Buccal cells were then embedded into a 2:1 3% LMP agarose and lysis solution and then subjected to imprinted DNA methylation analysis. One confluent well of a 6-well dish (~1 × 106 cells) of HES2 human ESCs (WiCell Research Institute Inc.) was washed once with 1X PBS (Sigma) and incubated in TrypLE Express (GIBCO) in Dulbecco’s PBS (DPBS). Trypsin was inactivated by addition of DMEM and 10% Fetal Bovine Serum (FBS) medium. Detached hESCs were collected, pelleted gently, washed with 1X PBS and re-suspended in 1000 μL of 1X DPBS. Approximately 1 μL of cells (~1000 cells) was embedded into a 2:1 3% LMP agarose and lysis solution, then subjected to bisulfite mutagenesis.

Imprinted DNA Methylation Analysis

Immediately following thawing individual embryos were embedded under mineral oil (Sigma) into 10 μL of a 2:1 mixture of 3% LMP agarose (Sigma) and lysis solution [100 mM Tris–HCl, pH 7.5 (Bioshop), 500 mM LiCl (Sigma), 10 mM EDTA, pH 8.0 (Sigma), 1% LiDS (Bioshop) and 5 mM DTT (Sigma), 1 μl 2 mg/ml proteinase K (Sigma), and1 μl 10% Igepal (Sigma)]. DNA methylation analysis was performed using the bisulfite mutagenesis and clonal sequencing method as previously described54. Samples were placed on ice for 10 minutes to generate an agarose/lysis bead and subsequently incubated overnight in SDS lysis buffer for 20 hours in a 50 °C water bath. Lysis buffer was removed and replaced with 300 μL of mineral oil and embryos were either frozen at −20 °C for a maximum of 3 days or immediately processed for bisulfite mutagenesis. Briefly, for bisulfite treatment, samples were incubated at 90 °C to inactivate proteinase K (Sigma) for 2.5 minutes and transferred to ice for 10 minutes. DNA denaturation was performed in 1 mL of 0.1 M NaOH at 37 °C for 15 minutes. Samples were covered with 300 μL of mineral oil and 500 μL of 2.5 M bisulfite solution for a 3.5-hour bisulfite conversion at 50 °C. After conversion, desulfonation was performed in 1 mL of 0.3 M NaOH at 37 °C for 15 minutes. Negative controls (beads containing no embryo or buccal cell sample) were processed with each bisulfite reaction. For first round PCR amplification, agarose bead with bisulfite converted DNA (10 μL) was added directly to 15 μL of Hot Start Ready-To-Go (RTG) (GE Healthcare) PCR bead that contained 0.5 μL of each 10 μM gene-specific external primer, 1 μL of 240 ng/mL transfer RNA and water with a 25 μL mineral oil overlay. Multiplexing of H19 and KCNQ1OT1 was performed during the first round of PCR. SNRPN amplification was performed individually. Five microliters of first round PCR product was added to 20 μL of RTG beads mixed with 19 μL 0.5 μL of each 10 μM internal primer and water for nested PCR. Separate second round PCR reactions were performed for H19 and KCNQ1OT1.

The KCNQ1OT1 PCR bisulfite primers were described previously34,35. The KCNQ1OT1 region analyzed contained a G (94.7%)/A (6.3%) SNP (Rs56134313). For the H19 region (GenBank Af087017, 6161-6409), external primers used were as described previously34. Due to SNPs residing in the previously described inner primers34, newly designed forward inner primer 5′-TTGGTTGTAGTTGTGGAAT-3′ and H19 reverse inner primer 5′-AACCATAACACTAAAACCCT-3′ were used for nested PCR, amplifying a 249 base pair sequence encompassing 20 CpGs and Rs2071094 A (33.6%)/C (66.4%) and Rs2107425 G (55.5%)/A (44.5%) common SNPs. For SNRPN, nested primers (UCSC, chr15:25, 200, 009-25, 200, 379) were designed to amplify a 360 base pair region encompassing 24 CpGs and a G (84.8%)/A (15.2%) SNP (Rs220029) within the ICR as follows, SNRPN outer forward, 5′-TAGTGTTGTGGGGTTTTAGGG-3′; SNRPN outer reverse, 5′-TACCCACCTCCACCCATATC-3′; SNRPN inner forward, 5′-AGGGAGGGAGTTGGGATTT-3′; SNRPN inner reverse, 5′-CACAACAACAAACCTCTAAACATTC-3′. All PCR reactions were performed as previously described55, 94 °C for 10 minutes followed by 55 cycles of 94 °C for 15 seconds, 56 °C for 20 seconds and 72 °C for 20 seconds, with a final 72 °C for 10 minute extension.

PCR products were ligated into the pGEM-T EASY vector system (Promega), transformed into Z-competent DH5α Escherichia coli cells (Zymo Research) and following blue/white selection and colony PCR, samples were sent for sequencing at Bio Basic Inc. (Markham, ON, Canada)29. For both day 3 and blastocyst-stage embryos, 32–64 clones were sequenced per embryo per gene. Methylation patterns were determined using online software (BISMA). Identical clones (identical location and number of unconverted CpG-associated cytosines and identical location and number of unconverted non-CpG-associated cytosines) were included only once and represented one individual DNA strand. Only clones with ≥85% conversion rates were included. Total DNA methylation for each gene, or for each allele of a gene, if parental identity was assigned, was calculated as a percentage of the total number of methylated CpG/the total number of CpG dinucleotides.

Statistical Analysis

Student’s t-test was used to examine significance between embryos with normal methylation and those with abnormal methylation for maternal age, hormone dose and stimulation response (E2 levels). Statistical analyses for patient diagnosis, hormone induction method, fertilization method and embryo grade was determined using the nonparametric Kolmogorov-Smirnov (KS) test to analyze differences between groups. A p-value of <0.05 was considered to be statistically different.

Additional Information

How to cite this article: White, C. R. et al. High Frequency of Imprinted Methylation Errors in Human Preimplantation Embryos. Sci. Rep. 5, 17311; doi: 10.1038/srep17311 (2015).