Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Evidence for widespread changes in promoter methylation profile in human placenta in response to increasing gestational age and environmental/stochastic factors

Evidence for widespread changes in promoter methylation profile in human placenta in response to... Background: The human placenta facilitates the exchange of nutrients, gas and waste between the fetal and maternal circulations. It also protects the fetus from the maternal immune response. Due to its role at the feto- maternal interface, the placenta is subject to many environmental exposures that can potentially alter its epigenetic profile. Previous studies have reported gene expression differences in placenta over gestation, as well as inter- individual variation in expression of some genes. However, the factors contributing to this variation in gene expression remain poorly understood. Results: In this study, we performed a genome-wide DNA methylation analysis of gene promoters in placenta tissue from three pregnancy trimesters. We identified large-scale differences in DNA methylation levels between first, second and third trimesters, with an overall progressive increase in average methylation from first to third trimester. The most differentially methylated genes included many immune regulators, reflecting the change in placental immuno-modulation as pregnancy progresses. We also detected increased inter-individual variation in the third trimester relative to first and second, supporting an accumulation of environmentally induced (or stochastic) changes in DNA methylation pattern. These highly variable genes were enriched for those involved in amino acid and other metabolic pathways, potentially reflecting the adaptation of the human placenta to different environments. Conclusions: The identification of cellular pathways subject to drift in response to environmental influences provide a basis for future studies examining the role of specific environmental factors on DNA methylation pattern and placenta-associated adverse pregnancy outcomes. Background syncytiotrophoblast), fibroblasts, mesenchymal cells, as The human placenta is a temporary organ that facilitates well as fetal and maternal vascular tissue and blood the exchange of nutrients, gas and waste between mater- cells. The extra-villous trophoblast cells must first nal and fetal circulations. In order to carry out these invade the maternal decidua and remodel maternal functions, it is comprised of heterogeneous cell types arteries, to allow direct contact between maternal blood and the placental syncytiotrophoblast cell layer [1]. In including several trophoblast cell populations (cytotro- phoblasts, extra-villous trophoblasts and addition to these traditional roles, the placenta is also important in shielding the developing fetus from the maternal immune system [2]. * Correspondence: richard.saffery@mcri.edu.au The placenta also undergoes several physiological † Contributed equally Cancer, Disease and Developmental Epigenetics, Murdoch Childrens changes throughout gestation, with one of the most sig- Research Institute, Royal Children’s Hospital and Department of Paediatrics, nificant being the flooding of placenta villi by maternal University of Melbourne, Parkville, Victoria 3052, Australia blood at the end of the first trimester (~12 weeks Full list of author information is available at the end of the article © 2011 Novakovic et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Novakovic et al. BMC Genomics 2011, 12:529 Page 2 of 14 http://www.biomedcentral.com/1471-2164/12/529 gestation), resulting in a rise in oxygen concentration as between unrelated individuals. We proposed that these well as a decrease in trophoblast invasion. It is believed CpG sites may be especially susceptible to environmen- that the inability of the placenta to respond to this tally-induced changes associated with placental disease change in oxygen concentration can lead to placental [21]. In a follow-up investigation, we also observed a disease, such as preeclampsia [3,4]. gestational age difference in DNA methylation profile in The molecular mechanisms behind these morphologi- the placenta across the third trimester [23], while others have recently reported an increase in global DNA cal and functional changes are now beginning to be methylation levels between pre-term (28 weeks) and full understood at both the gene-specific and genome-wide term placenta (40 weeks) [24]. level. Wide-ranging genome-wide gene expression differ- ences between placentas at different gestational ages Theaim ofthecurrent studywas to buildonrecent were reported in two recent studies [5,6]. Despite sam- knowledge obtained through genome-scale gene expres- pling from different locations within the placenta, many sion [5,6,25] and DNA methylation analysis [24,26] of changes were found in common between the two stu- the human placenta. In the current study, we used the dies, each of which reported changes in expression with Illumina Infinium HumanMethylation27 BeadChip plat- increasing gestational age in genes involved in cell cycle form to identify promoter regions subject to change and immune response. This suggests that gene expres- throughout gestation. In addition we wanted to identify sion changes are needed for physiological needs of the those that become increasingly variable between indivi- developing placenta, such as shielding the fetus from the dual placentas over time, possibly in response to accu- maternal immune system [2]. Genes involved in Wnt mulated environmental exposures. signalling also showed expression changes over time [5,6] that resulted in decreasing levels of b-catenin later Results in gestation, possibly linked to decreasing placental inva- Genome-scale DNA methylation analysis of first, second, siveness [6]. and third trimester placenta The importance of epigenetic factors in placental Genome-scale DNA methylation analysis of 18 first tri- development and function has long been known through mester (8-12 weeks), 10 second trimester (17-24 weeks) the study of imprinted genes [7,8] and it is increasingly and 14 third trimester placenta (34-41 weeks) samples clear that the placenta displays a unique epigenetic pro- was performed using the Illumina Infinium Human- file. However, the extent to which epigenetic modifica- Methylation27 BeadChip (see Additional file 1; Data was tions, specifically DNA methylation, contribute to deposited into the NCBI Gene Expression Omnibus, placental function have only recently been widely exam- accession number: GSE31781). Following normalisation ined (reviewed in [9]. and data cleaning (see Methods), a total 26, 162 probes Due to its role as the interface between the mother were available for subsequent analysis. Correlations (r ) and fetus, the placenta is exposed to a myriad of envir- of average methylation of probes between gestational onmental factors, some of which have been shown to ages ranged from 0.935 (first v third trimester) to 0.97 alter placental gene expression, as well as epigenetic (second v third trimester; see Additional file 2). Unsu- marks [10]. These include diet [11,12], smoking [13], pervised hierarchical clustering of all probes clearly and assisted reproductive techniques [14,15]. Mounting delineates samples according to gestational age (Figure evidence implicates epigenetic marks, such as DNA 1) with further evidence for a closer relationship methylation, in mediating environmentally-induced reg- between second and third trimester profile than with ulation of genome function. More studies into the first trimester. Validation of methylation levels at 12 effects of the environment on the placental epigenome Infinium probes (representing 12 genes) in 9 placental are warranted due the importance of this organ in regu- samples (total data points: 49) using the Sequenom Epi- lating pregnancy development. TYPER platform confirmed the robust nature of the Several genome-scale DNA methylation studies have Infinium data (r = 0.76) (see Additional file 3). The focused on finding tissue-specific differentially methy- probes used for validation were chosen due to their lated regions (tDMRs) between placenta and maternal association with genes with known or predicted impor- blood, as a means of detecting placental pathologies and tant roles in regulating placental function (Additional fetal chromosomal trisomies using non-invasive methods file 4). (reviewed in [16-19]. This strategy has recently resulted in the development of the first non-invasive blood test Transition to normoxia is not associated with major for Down syndrome [20]. However, we, and others have changes in placental DNA methylation profile revealed substantial inter-individual DNA variation in Despite the overall interspersed pattern of clustering of placental methylation profile [21,22], with a subset of first trimester samples of various gestations (8 - 12 CpG sites more likely to be differentially methylated weeks) (Figure 1), we tested to see whether any genomic Novakovic et al. BMC Genomics 2011, 12:529 Page 3 of 14 http://www.biomedcentral.com/1471-2164/12/529 Figure 1 Cluster dendrogram based on all autosomal Infinium probes distinguishes placentas of different gestational age. Dendrogram showing the relationship between placental samples from three gestational ages based on DNA methylation levels (b- values) of all analysable Infinium probes. All samples clustered within their gestational age group, with no overlap between gestations, suggesting there are consistent genome-scale DNA methylation patterns associated with each gestational age. First Figure 2 Average methylation of all samples for first, second trimester samples clustered away from second and third trimester and third trimester. Methylation Index (MI) was calculated for each samples, indicating that overall Infinium methylation patterns are sample by calculating the mean of all analysable Infinium b-values more similar in second and third trimester compared to first (26, 162 probes) for that sample. The MIs were then grouped by trimester. gestation and shown as box and whisker plots. First and second trimester placentas show a similar overall level of methylation (p = 0.46) with median MIs of 0.238 and 0.241, respectively. Third trimester samples show significantly elevated average MI values regions undergo selective changes in methylation during (median = 0.256) relative to both first and second trimester, the transition from early first trimester to late first/early indicating that there is a significant increase in methylation level second trimester. This is the period widely regarded as from second to third trimester. the time when placental intervillous space is flooded with oxygenated maternal blood. Only limited DNA clearly apparent when the relative methylation levels for methylation differences were observed between 8 and 12 all probes are displayed in a scatterplot (Additional file week placentas (Additional file 5). A total of only 12 2). All 3 trimesters have the same proportion of probes probes (3 hypomethylated and 9 hypermethylated at 12 with b-values between 0 and 0.2 (~63% of total probe weeks relative to 8 weeks) showed consistent methyla- number). However, as gestation progresses, there is an tion differences (Δb) of 0.2 between the two gestational increase (from 13% to 17%) of highly methylated probes ages. This suggests that promoter DNA methylation (b > 0.6) (Additional file 6). plays a limited role in any physiological changes in the Absolute differences in mean methylation between placenta that occur in response to altered oxygen status. first and second, first and third, and second and third trimesters were generally small, with only 149, 954 and Gestational age is associated with promoter methylation 157 probes respectively, showing Δb >0.2 (Table 1). levels Further analysis showed that the 883 probes that In order to gauge the effects of gestational age on the increased in methylation from first to third trimester overall methylation level at gene promoter regions were predominantly those with intermediate methylation enriched on the Infinium HumanMethylation27 Bead- Chip, the mean methylation across all probes (Methyla- levels (average b range 0.3 - 0.5) in the first trimester tion Index - MI) was calculated for each sample (Figure (Additional file 7). Figure 3 shows a heat map with 2). The mean MI for first, second and third trimester unsupervised clustering of samples based on the 954 st placenta was 0.240, 0.242 and 0.256, respectively. Thus, probes showing a Δb >0.2between1 trimester and there is an overall increase in methylation between sec- term. Importantly, samples from all three trimesters -5 ondand thirdtrimesters(p=5×10 ,student’s t-test). showed distinct methylation patterns. More specifically, No significant differences were detected between first second trimester samples did not cluster with either the and second trimesters (p = 0.46) (Figure 2). This is first trimester or term samples, suggesting a progressive Novakovic et al. BMC Genomics 2011, 12:529 Page 4 of 14 http://www.biomedcentral.com/1471-2164/12/529 Table 1 Number of probes showing differential methylation between first, second and third trimester placental tissue. Comparison Differentially methylated probes Difference b ≥ 0.1 and adj. p < 0.05 Difference b ≥ 0.2 and adj. p < 0.05 (adj. p < 0.05) First v Second 8, 240 411 ↓ 12 ↓ 1, 077 ↑ 137 ↑ First v Third 8, 298 755 ↓ 71 ↓ 2, 581 ↑ 883 ↑ Second v Third 7, 669 288 ↓ 6 ↓ 1, 515 ↑ 151 ↑ ↑: Increase in methylation; ↓: Decrease in methylation change in methylation across gestation. Interestingly, the major factors contributing to methylation differences none of the probes showed a ‘fluctuating’ pattern of between the two gestational time points. In order to test methylation across gestation (i.e. showed a similar level this, methylation levels between purified first trimester of methylation in first and third trimesters, but hypo/ cytotrophoblasts and first and third trimester placenta, hypermethylation in second trimester) based on our cri- were examined using both Infinium and Sequenom Epi- teria of Δb > 0.2, and only 103 probes showed this pat- TYPER analyses (22 data points) (Additional file 8). tern with a Δb > 0.1. This further suggests that most Methylation levels in purified cytotrophoblasts were more methylation changes occur in a progressive manner. similar to first trimester placental tissue (r =0.96and Changes in cell composition, primarily a decrease in 0.93) than third trimester (r = 0.88 and 0.88) as calculated cytotrophoblasts, from first to third trimester could be one by Infinium and Sequenom EpiTYPER, respectively. Figure 3 Unsupervised clustering based on probes with Δb > 0.2 between First and Third trimester. HeatMap showing unsupervised clustering of all placenta samples (x-axis) based on 954 probes with a Δb > 0.2 between First and Third trimester (y-axis). The majority of differentially methylated probes show higher methylation in third trimester (883 probes) compared to only 71 probes with lower methylation in third trimester. Second trimester placentas cluster as a separate group, and show a methylation profile that is an intermediate of first and third trimesters. Green corresponds to low methylation and Red to high methylation. Novakovic et al. BMC Genomics 2011, 12:529 Page 5 of 14 http://www.biomedcentral.com/1471-2164/12/529 Differentially methylated genes between first trimester, (s <0.009) at alltimepoints, thenumberwith an second trimester and term inter-placental variance of > 0.01 was increased in the Probes that showed differences of Δb > 0.2 (approximat- third trimester relative to the earlier time points (Figure ing 20% differential methylation) between each of the 4). More specifically, the third trimester group was three gestational ages, (Table 1), were analysed using enriched for probes with the highest variance (s > Ingenuity Pathways Analysis (IPA). Enrichment of net- 0.02), with 352 such probes compared to 106 (c = works, pathways and gene functions was calculated 133.3, p < 0.001) and 166 (c = 66.7, p < 0.001) in first using the Ingenuity Pathways software. Top enriched and second trimester placentas respectively (Additional canonical pathways for genes differentially methylated file 9). To facilitate the comparison of variation level between first and second trimesters (110 genes), first between trimester groups, we defined sites with variance and third trimesters (654 genes) and between second > 0.02 as highly variable in methylation. and third trimesters (106 genes) are listed in Table 2. Further analysis revealed that most of the gestational ‘Communication between innate and adaptive immune age associated variation was found in probes with cells’ was the most significantly enriched pathway, with intermediate methylation (0.2 < b < 0.6) rather than at least 4 of the top 5 pathways in each comparison low (b <0.2)orhigh(b > 0.6) methylation (Additional being immune-related. file 10; Additional file 9). This suggests that the increased variation observed between different full- Inter-placental methylation variation increases with term placentas is not necessarily a by-product of gestational age increasing methylation across gestation. In addition, DNA methylation may be modulated in part by environ- the increasing methylation across gestation also sug- mental influences and may serve as a mediator between gests that the increasing variability was unlikely caused the environment and genome function (reviewed in by a lack of maintenance of DNA methylation by [27]). We previously investigated the inter-individual DNMTs in the human placenta. variability of DNA methylation in the human placenta While inter-individual variation of DNA methylation [21] and proposed that the highly variable methylation may in part reflect genetic polymorphisms [29], this found in the placenta may be a consequence of cumula- source of variation would be anticipated to be repre- tive response to the intrauterine environmental expo- sented equally across all 3 gestational ages. To investi- sures [28]. To test this further, we calculated the inter- gate the inter-placental variation of methylation in placental variance of all probes within each of the first, more detail, variance level of probes for the third tri- second and third trimester time points. While the vast mester was plotted against the first trimester (Figure majority of probes (95-98%) showed very little variation 5). This revealed that the majority of the highly Table 2 Top Canonical Pathways from IPA for probes showing differential methylation across gestation First v Second trimester p-value # differentially methylated genes/# genes in the pathway Communication between Innate and Adaptive Immune Cells 6.39E-05 6/109 Role of Cytokines in Mediating Communication between Immune Cells 1.53E-03 4/56 Altered T Cell and B Cell Signaling in Rheumatoid Arthritis 6E-03 4/92 Calcium Signaling 6.81E-03 6/204 Crosstalk between Dendritic Cells and Natural Killer Cells 1.04E-02 4/97 Second v Third trimester p-value # differentially methylated genes/# genes in the pathway Systemic Lupus Erythematosus Signaling 4.72E-03 5/166 Crosstalk between Dendritic Cells and Natural Killer Cells 8.2E-03 4/97 Role of NFAT in Regulation of the Immune Response 2.05E-02 5/199 Agrin Interactions at Neuromuscular Junction 2.35E-02 3/69 Wnt/b-catenin Signaling 2.51E-02 5/172 First v Third trimester p-value # differentially methylated genes/# genes in the pathway Communication between Innate and Adaptive Immune Cells 1.69E-07 17/109 Systemic Lupus Erythematosus Signaling 1.9E-06 21/166 Role of Cytokines in Mediating Communication between Immune Cells 1.43E-05 12/56 Altered T Cell and B Cell Signaling in Rheumatoid Arthritis 1.79E-04 13/92 Crosstalk between Dendritic Cells and Natural Killer Cells 2.61E-04 14/97 Novakovic et al. BMC Genomics 2011, 12:529 Page 6 of 14 http://www.biomedcentral.com/1471-2164/12/529 Figure 4 Number of variable probes increases with gestation. The number of probes with high inter-individual variation increases over gestation. Variance (s ) of each probe was calculated for each gestational age. The number of probes (y-axis; log scale) showing a particular level of variance (x-axis) is shown. Most probes (95% in third trimester to 98% in first trimester) show low variation (s2 < 0.009). However, there is an increase in the number of variable probes (s > 0.02) in third trimester placentas, and to a lesser extent second trimester placentas, compared to first trimester. variable sites “gain” variability in the later gestations trimester were not context dependent (c = 2.68, p = (area (c) in Figure 5). Numerically, the probes with 0.14 (Additional file 12). We also examined the genomic the highest variance are found exclusively in the third context of the variable probes falling within the three trimester (223), while far fewer probes show high var- categories shown in Figure 5(A-C). Probes with high iation across all three gestations (47) (Additional file variation in both first and third trimester (Additional 11). file 12-D) are associated with CGIs (p < 0.05). Variable The top pathways for genes that show variable methy- probes in the first, but not the third trimester are lation at each gestastional age are listed in Table 3. The slightly enrichmed for CGI regions (p = 0.05), and those top 5 pathways in the third trimester include ‘amino that were variable in third, but not first trimester show acid metabolism’, while first and second trimester lists a strong association with CGIs (p < 0.001) (Additional both include ‘circadian rhythm signalling’. This supports file 12). We also looked at the relationship between the our hypothesis of an increasing accumulation of epige- distance from transcription start site and DNA methyla- netic variation in response to cumulative environmental tion. There was no association with distance from TSS exposures. and differential or variable methylation (data not shown). CpG density and genomic context influence variability in placental methylation DNA methylation profile influences global gene In order to examine the relationship between CpG den- expression in the placenta sity and methylation status, probe locations were In order to assess the overall effect of DNA methylation assigned to either CpG Island (CGI) or non-CpG Island on gene expression profile, methylation b-values levels (non-CGI) genomic regions, as annotated by Illumina. for first, second and third trimester placental tissue Using a chi square test, we found that probes showing were correlated with publicly available expression data increased methylation across gestation were enriched for for matched gestational age placenta [5,6]. Infinium non-CGI regions (c = 480.2, p < 0.001), while probes probes were quartiled according to methylation level, showing lower methylation in third compared to first with b value ranges of 0.01-0.06 (bottom 25% of probes), Novakovic et al. BMC Genomics 2011, 12:529 Page 7 of 14 http://www.biomedcentral.com/1471-2164/12/529 Figure 5 Variance levels of probes in the third compared to the first trimester. Scatter plot of probe variance (s ) at first trimester (x-axis) and third trimester (y-axis). Dots represent individual probes and the vertical red dotted line marks s = 0.02 for first trimester, while the horizontal red dotted line marks s = 0.02 for third trimester. Probes on the outside of the red line are deemed ‘variable’. This analysis revealed that there are 73 probes (A) which are highly variable in both first and third trimester. Only 33 probes (B) were variable in first, but not third trimester, while 279 probes (C) were variable in third, but not first trimester. This analysis suggests that most of the variable probes become so throughout pregnancy, supporting the hypothesis that accumulating environmental factors contribute to inter-individual variation in DNA methylation, in term placenta. Table 3 Top Canonical Pathways for gene-associated probes that show variable methylation within each gestational age Variable in First trimester p-value genes Arachidonic Acid Metabolism 6.97E-04 CBR1, CYP2E1, GPX7, PNPLA3, PTGS1 Hepatic Fibrosis/Hepatic Stellate Cell Activation 1.36E-02 CYP2E1, ECE1, FGFR1, MYL5 Circadian Rhythm Signaling 1.91E-02 GRIN3A, VIPR2 Calcium Signaling 2.69E-02 CHRNB4, GRIA4, GRIN3A, MYL5 Nitrogen Metabolism 3.54E-02 PTPRG, VNN3 Variable in Second trimester p-value genes Circadian Rhythm Signaling 3.39E-03 GRIN3A, VIPR2, PER1 Glutathione Metabolism 1.62E-02 GPX3, GPX7, GSTO1 Arachidonic Acid Metabolism 1.86E-02 CBR1, CYP2E1, GPX3, GPX7 Sonic Hedgehog Signaling 3.7E-02 DYRK1B, HKR1 Metabolism of Xenobiotics by Cytochrome P450 4.68E-02 ADHFE1, CYP2E1, GSTO1 Variable in Third trimester p-value genes Glutamate Receptor Signaling 6.1E-03 GRID2, GRIK2, GRIN3A, GRM6, SLC1A6 Valine, Leucine and Isoleucine Degradation 1.13E-02 ACADL, ALDH1A3, ELOVL2, IVD, OXCT1 b-alanine Metabolism 2.07E-02 ACADL, ALDH1A3, DPYS, IVD Butanoate Metabolism 3.74E-02 ALDH1A3, ELOVL2, OXCT1, PDHA2 Tyrosine Metabolism 4.19E-02 ADHFE1, ADLH1A3, ELOVL2, MGMT Novakovic et al. BMC Genomics 2011, 12:529 Page 8 of 14 http://www.biomedcentral.com/1471-2164/12/529 0.06-0.15 (25-50%), 0.15-0.47 (50-75%) and 0.47-0.98 Illumina Infinium HumanMethylation27 BeadChip, in (top 25% of probes), and plotted against gene expression placentas of different gestational ages (18 first trimester levels for linked genes data (Additional file 13). A gen- (8-12 weeks), 10 second trimester (17-23 weeks), and 14 eral decrease in median expression level with increasing third trimester samples (34-41 weeks)). We further vali- methylation level was observed for all three gestational dated the array data by targeting 12 CpG sites (covering ages. In particular, there was a distinct down regulation 12 gene promoters) using the Sequenom EpiTYPER in median gene expression between the second (50%) to platform. The correlation of r = 0.76 between the two third quartiles (75%) associated with a change in methy- platforms is comparable to that published for similar lation range from b <0.15 to b > 0.15. In addition the comparisons [30]. We found evidence for both a pro- range of expression was reduced in the higher methyla- grammed change of methylation in gene promoters tion quartiles (Additional file 13). Using the same across gestation and an increase in variability of methy- expression data sets, we plotted methylation change lation as gestation progresses. We predict that this is from first to third trimester against changes in gene directly related to the cumulative effect of intrauterine expression across the same time points (Figure 6). environmental exposure, however, the contribution of While methylation change of less than 0.2 were gener- stochastic events to the observed variation cannot be ally not associated with changes in gene expression, sev- discounted at this time. Further investigation is war- eral genes showed both higher methylation and lower ranted to determine the effects of specific environmental expression in third trimester compared to first trimester. exposures on DNA methylation patterns of the genes These included several immune-regulators ranked highly identified as variable in the present study. by IPA (CCR7 and CCL21) and one with known func- Unsupervised clustering based on all Infinium probes tion in placental development (GNLY; Granulysin) (Fig- (excluding those on the sex chromosomes) clearly sepa- ure 6). Additional file 14 lists additional genes that rated all samples by gestational age, with first trimester showed concordant differences in methylation and samples clustering away from second and third trimester expression between first and third trimesters. samples (Figure 1). This indicates that there are consis- tent, large scale changes in DNA methylation across Discussion gestation. There are several possible explanations for In this study we performed genome-wide DNA methyla- these temporal differences, one of which is the change tion analysis of gene promoter regions, using the in cell composition and differentiation from first to Figure 6 Correlation between methylation and gene expression change between first and third trimester. Methylation difference (Δb) between first and third trimester (x-axis) was plotted against gene expression log fold change (y-axis) between first and third trimester. A positive change in log fold expression indicates higher expression in first trimester, while a positive change in methylation indicates higher methylation in the third trimester. Therefore, the top left panel includes genes which showed lower methylation and higher expression in first compared to third trimester. The three highlighted genes (CCR7, GNLY and CCL21) ranked highly in IPA analysis. Grey dots represent Infinium probes. Black dots represent specific genes of interest. Novakovic et al. BMC Genomics 2011, 12:529 Page 9 of 14 http://www.biomedcentral.com/1471-2164/12/529 third trimester, especially the relative loss of cytotropho- repetitive elements [35,36]. A recent study has reported blasts throughout gestatation, (85% of the total tropho- a positive correlation between global DNA methylation blast population in first trimester but only 15% of the levels (as measured by an ELISA-like assay with 5- trophoblast volume at full term) [31]. In support of this methylcytosine antibody) and gestational age in the pla- playing at least a partial role in the observed change in centa [24], supporting our data for an increasing level of methylation over time, we found a slightly higher corre- promoter-associated DNA methylation during gestation, particularly from second to third trimester (Figure 2). lation between the methylation profile of purified pri- This accumulation of methylation was most apparent in mary cytotrophoblasts with first trimester placental 2 2 genomic regions that showed an intermediate level of tissue (R = 0.96) than with third trimester tissue (R = 0.88; see Additional file 8). Similar data have recently methylation in first trimester, suggesting that these been reported for second trimester placenta tissue genes are the most likely to be epigenetically regulated which has been shown to be more similar to cytotro- by DNA methylation. Furthermore, these CpG sites phoblasts than mesenchyme in terms of methylation were more likely to be in lower CpG density regions profile [32]. However, the greater part of trophoblast (non-CGI) (see Additional file 12) suggesting that volume at full term is syncytiotrophoblast, arising from methylation levels of CpG sites within CpG Islands are the fusion of cytotrophoblast cells into a multinucleated more stable across gestation in this tissue. The higher layer. Thus, the observed trend of increasing methyla- promoter methylation at term could also reflect the end tion may equally be due to differentiation of the cytotro- point of a continual process of re-methylation in the phoblast component, or alternatively, may even be due extra embryonic lineage from the blastocyst stage, at to other aspects of altered placental function known to whichpoint thegenomeisalmostcompletelyhypo- occur as gestation progresses. methylated [34]. Despite the interspersed clustering of first trimester IPA analysis of genes showing differential methylation samples based on the overall Infinium methylation pat- between first and third trimesters indicated that the terns, we were interested in assessing DNA methylation most affected pathways were ‘communication between changes occurring at the transition from the first to the innate and adaptive immune cells’ and other ‘immune- second trimester. This period of placental development related’ pathways (Table 2). Genes in common between is characterised by the loosening of trophoblast plugs two or more of the top 5 immune-related pathways and the associated rise in oxygen concentration from 2- included immune regulators CCR7, CD28, CSF2, 3%,to7-8%[33]. Therapid increase in oxygen levels IFNA17, IFNA2, IFNA21, IFNB1, IL1F7, IL2, IL3, IL5, LTA, TLR6, TLR9, TNF, TNFRSF13B and TNFSF13B. can lead to the accumulation of reactive oxygen species (ROS), which might be linked to the reduction in tro- This is in accordance with previous gene expression stu- phoblast invasion and migration observed at about this dies, which also showed substantial enrichment of time. Furthermore, it has been suggested that the inabil- immune regulators amongst the most differentially ity of the placenta to adapt to this increase in oxygen expressed genes in placenta at different gestational ages concentration may lead to the development of pree- [5,6]. Several immune-regulators showed a strong corre- clampsia [3]. In this study, we found only 12 CpG sites lation between methylation and expression change from with Δb > 0.2 between these time points, suggesting first to third trimester, including Granulysin (GNLY), that DNA methylation changes are unlikely to play a which has previously been implicated in spontaneous major role in the physiological changes in placentation abortions [37] (Figure 6). Given the critical role of the associated with transition from low oxygen to a normal placenta in modulating the maternal immune response oxygen environment. It must be noted however, that we to the developing pregnancy, and mounting evidence for do not have data on the level of vascularisation for our a role of environmental exposures in controlling both placental samples, so it is possible that flooding of the immune-system development and epigenetic profile, it is maternal blood may not have occurred at the time the not surprising that immune regulators are amongst the 12 week tissue was collected. most variably methylated gene groups. The separation of the trophectoderm from the inner In addition to identifying CpG sites that consistently cell mass at the blastocyst stage occurs at a time when change over gestation, we were also interested in CpG genome-wide DNA methylation levels are at their lowest sites that show inter-individual variability within each [34]. The subsequent re-establishment of methylation gestational age. Genes that show inter-individual differ- ences in expression in placenta have previously been marks occurs at a slower rate, and to a lesser extent, in described by Sood et al. [25]. We have previously the extra-embryonic lineage compared to somatic tissues reported a subset of CpG sites showing variable methy- [34]. This accounts for the low global DNA methylation in human placenta, which is more similar to human lation in term placenta between unrelated individuals, tumours, and is mostly due to hypomethylation of and suggested that these CpG sites may be more Novakovic et al. BMC Genomics 2011, 12:529 Page 10 of 14 http://www.biomedcentral.com/1471-2164/12/529 susceptible to change under adverse conditions [21]. We expression levels at all three gestational time points. have now extended these findings by demonstrating that However, thelarge rangein gene expressionin all four although ~95% of all probes showed very little variation methylation quartiles, and the lack of strong correlation (s < 0.01) between individuals, the number of highly between methylation and expression change across variable probes (s > 0.02) increased with gestational gestation (Figure 6), support previous data that promo- age. Our data support a model whereby increasing varia- ter DNA methylation is only one of the mechanisms controlling gene expression, with many other factors bility in methylation profile between placentas arises in combining to regulate the expression of genes, including response to cumulative differences in environmental histone modifications, non-coding RNAs, and transcrip- exposure rather than sequence polymorphisms (Figure 5). The most variable probes in the first trimester were tion factors. associated with genes involved in, ‘calcium signaling’ and ‘nitrogen metabolism’,while ‘circadian rhythm sig- Conclusions nalling’ and ‘arachidonic acid metabolism’ were In the present study, we performed genome-scale DNA enriched in both first and second trimester placenta methylation analysis of human placenta at three distinct (Table 3). The most enriched variable pathways at term gestational ages, using the Infinium HumanMethyla- included ‘glutamate receptor signaling’, ‘valine, leucine tion27 array. This platform targets over 14, 000 genes, and isoleucine degradation’,and ‘b-alanine’, ‘butanoate’, however, is limited to the 5’ promoter region. Our find- and ‘tyrosine’ metabolism. The enrichment of pathways ings support the hypothesis that DNA methylation levels involved in metabolism further supports our hypothesis in the human placenta are dynamic and change over that the methylation variation is occurring in response gestation, possibly in response to changing cellular com- to environmental influences, potentially as part of an position and/or cumulative environmental influences. ‘adaptive’ response of the developing pregnancy, thereby The identification of pathways that are likely to be allowing changes in gene expression that may be more affected by the latter (i.e. those that show higher varia- beneficial under altered environmental conditions [38]. tion at full term) will provide valuable candidates for The placental epigenome regulates placental gene testing in studies examining placenta-associated adverse expression and function, and any disruption in placental pregnancy outcomes. epigenetics (via either stochastic or environmental per- turbation) has the potential to affect the developing Methods fetus. Therefore it will be interesting in future to exam- Sample collection ine the potential for role of placental epigenome varia- Use of cells and DNA isolated from 8- and 12-week tion with the now widely accepted phenomenon of ‘fetal gestational tissue of normal pregnancies was approved programming’ first described in the context of long by the Cambridgeshire Research Ethics committee st term effects of intrauterine environment on offspring (CREC 04/Q0108/23). Additional 1 trimester (8-12 nd [39] and subsequently developed by Barker and collea- week gestation) and 2 trimester (17-24 week gestation) gues in the early 1990s [40-42]. placental tissue was collected from elective abortions Finally, contrary to CpG sites that increase in methyla- with the approval from the ethics committees of the tion across gestation, variable CpG sites are more likely University of British Columbia and the Children’s& to be located within CpG islands (Additional file 12). Women’s Health Centre of British Columbia. Term pla- This coupled with the finding that hypermethylated cental tissue was collected as previously described probes (b > 0.6) do not show higher variation compared [21,23,43]. Detailed information of the placental samples to probes with low methylation levels (b <0.2)(see can be found in Additional file 1. Additional file 10) suggests that increased variation in late gestation is not a by-product of increasing methyla- DNA extraction tion across gestation. Tissue samples were incubated at 50°C overnight with The direct role of DNA methylation in controlling shaking in DNA extraction buffer (100 mM NaCl, 10 placental global gene expression levels was examined by mM Tris.HCl pH8, 25 mM EDTA, 0.5% (w/v) SDS), comparing our DNA methylation data with published containing 200 μg/ml proteinase K. DNA was isolated gene expression data for placenta tissue of the same by two rounds of phenol:chloroform extraction, followed gestational age (Figure 6 and Additional files 13 and 14). by RNAse A treatment, precipitation in absolute ethanol A general trend of decreasing expression in response to containing 10% (v/v) sodium acetate (3 M, pH 5.2), and increasing DNA methylation was found as expected. resuspended in 100 μl nuclease-free water (Ambion, Further, the range of expression also decreased in the Austin, TX, USA) or using salting out method followed highly methylated probe group. These findings confirm by purification with Qiagen blood and tissue kit (Qia- that promoter DNA methylation influences global gene gen, Mississauga, ON, USA). DNA was stored at -20°C. Novakovic et al. BMC Genomics 2011, 12:529 Page 11 of 14 http://www.biomedcentral.com/1471-2164/12/529 Infinium DNA methylation analysis Raw data obtained from MassArray EpiTYPING was Infinium arrays were hybridized and scanned as per cleaned systematically using an R-script with the follow- manufacturer’s instructions (Illumina, San Diego, USA). ing three criteria (in order): (1) removal of samples that Individual probe b-values (range 0-1) were are approxi- failed across 100% of CpG sites, (2) removal of CpG mate representations of the absolute methylation per- sites that failed (for example due to low or high mass, centage of specific CpG sites within the sample silent peaks, or low intensity readings) across more than population. The values were derived by comparing the 40% of samples, and (3) removal of samples that failed ratio of intensities between the methylated and to generate data for more than 70% of CpG sites tested. unmethylated alleles using the following formula: Publicly available gene expression microarray data Max Signal B,0 analysis β value = [Max(Signal A,0) + Max(Signal B,0)] Gene expression data was downloaded from Gene Expression Omnibus (Barrett and Edgar, 2006, Sayers et WhereSignalBis the arrayintensityvaluefor the al., 2009), (Database issue D885-D890). CEL files were methylated allele and Signal A is the non-methylated downloaded from series GSE9984 [6] and GSE5999 [5] allele. Samples were processed using the Bioconductor and processed with the Bio-conductor package ‘gcrma’ package lumi, which is specifically designed for Illumina [47]. DNA methylation and gene expression data for the data [44,45]. Samples were assessed for quality, color- corresponding gestational age was compared as pre- adjusted to take into account the difference between viously described [48]. The two expression matrices batches, background-corrected and ssn normalized. Any were merged based on probe ID as both were generated probe within a sample with a detection p value of 0.05 using the same array platform, the Affymetrix Human or greater was excluded from further analysis. Probes on Genome U133A Array. Gene expression data was then the × and Y chromosomes were removed from further linked with the methylation data according to gene analysis to eliminate sex-specific differences in methyla- name. Sample quartiles were produced from the methy- tion, leaving 26, 162 analysable probes. A batch correc- lation data. The expression values of the genes in each tion effect was applied to the data in order to remove quartile were then plotted as box and whisker plots. noise produced by processing samples at different times. Differentially methylated probes were defined as having Additional material a Δb > 0.2 between groups, with an adjusted (Benjamini) p value < 0.1. Additional file 1: Summary of placental samples analysed in the study. Gene Ontology and Pathway analysis Additional file 2: Correlations between average methylation in first, Data sets were interrogated using the Ingenuity Path- second and third trimesters. Correlations between average methylation in (A) first and second trimesters, (B) second and third trimesters, and (C) ways Analysis (IPA) application (Ingenuity Systems, first and third trimesters. This analysis revealed that first and third Redwood City, CA; http://www.ingenuity.com). IPA was trimester methylation was most discordant, as expected, while second used to identify enriched canonical pathways, gene net- trimester placenta is more similar to third trimester placenta in terms of overall promoter DNA methylation. Furthermore, this figure visually works, functional classes, and toxicity lists (molecules shows the increase in methylation in third trimester compared to first involved in known toxicity processes). (C). Additional file 3: Correlation between Infinium and Sequenom Locus-specific methylation analysis methylation levels. Correlation between Infinium HumanMethylation27 BeadChip and Sequenom EpiTYPER locus-specific methylation analysis. Sequenom MassARRAY EpiTYPING was performed to Methylation levels in 12 genes were measured using Sequenom validate Infinium methylation, as previously described MassARRAY Epityping targeting the same CpG sites interrogated on the [46]. Sequenom assays were designed to target specific Infinium BeadChip Arrays. Correlation between platforms was 0.76, supporting the use of the Infinium HumanMethylation27 BeadChip for Infinium probes. Genomic sequences for assay design profiling DNA methylation in this study. Genes interrogated are listed in were extracted from the UCSC genome browser http:// Additional file 4. www.genome.ucsc.edu/. Primer pairs for amplification Additional file 4: Sequenom EpiTYPER primer sequences. were designed using EpiDesigner web tool http://www. Additional file 5: Unsupervised clustering of first trimester placenta epidesigner.com/. The primers are listed in Additional based on differentially methylated probes between 8 and 12 weeks gestation. HeatMap showing unsupervised clustering of 8 and 12 week file 4. Amplification was performed after bisulfite con- placenta samples based on 12 probes that showed a Δb > 0.2 between version of genomic DNA with the MethylEasy Xceed 8 week and 12 week placenta. The 12 probes were associated with 11 bisulphite conversion kit (Human Genetic Signatures, genes, with two probes associated with the BTG4 gene. White corresponds to low methylation, and black to high methylation. North Ryde, Australia). Amplification conditions were Additional file 6: Proportion of probes within a particular 40 cycles: 95°C for 5 min, 56°C for 1 min 30 sec and 72° methylation level. Pie Charts showing the proportion of probes within a C for 1 min 30 sec, then 72°C for 7 min. particular methylation level for first, second and third trimester. The Novakovic et al. BMC Genomics 2011, 12:529 Page 12 of 14 http://www.biomedcentral.com/1471-2164/12/529 percentage of probes with a ‘b < 0.02’ is the same in all three HumanMethylation BeadChip probes were quartiled into 4 groups (0- gestational ages (63%), suggesting that probes with low methylation in 25%, 25-50%, 50-75%, 75-100%) based on methylation level, with the first trimester remain low over placental development. Furthemore our same number of probes in each quartile. The quartiles for each data suggests that probes with an intermediate methylation in first gestational age were plotted on the x-axis with the corresponding gene trimester are the ones that increase over gestation, with a lower expression values obtained from publically available first, second and proportion of probes in the ‘b = 0.2 - 0.6 group’ in second (22%) and third trimester data (y-axis). This analysis shows a decreasing median st third (20%) compared to 1 trimester (24%); and a higher proportion of gene expression level with increasing DNA methylation, highlighting the nd probes in the b > 0.6 group in 2 (15%) and term (17%) placenta functional relevance of DNA methylation in placenta at all three st compared to 1 trimester (13%). gestational ages. Additional file 7: Average methylation level of probes that increase Additional file 14: Correlation between methylation and expression in methylation over gestation. Box plot showing average methylation change between first and third trimester with more genes of (A) all probes (n = 26, 162) in first, second and third trimester highlighted. Methylation difference (Δb) between first and third placenta, and (B) probes that shown an increase in methylation from first trimester (x-axis) was plotted against gene expression log fold change (y- to third trimester of b > 0.2 (n = 883). This analysis shows that probes axis) between first and third trimester. A positive change in log fold with intermediate levels of methylation in first trimester are the ones expression indicates higher expression in first trimester, while a positive that increase over time. On the other hand, probes with low methylation change in methylation indicates higher expression in third trimester. (b < 0.2) in the first trimester do not appear to increase in methylation in Highlighted genes are those that show a correlation between the third trimester placenta. methylation and expression level. Grey dots represent Infinium probes, black dots represent most differentially methylated and expressed genes. Additional file 8: Correlation between methylation levels in purified first trimester cytotrophoblasts and first and third trimester placenta. Scatter plot showing correlation (r ) between first and third trimester placenta and purified first trimester cytotrophoblasts methylation, based on Infinium HumanMethylation27 BeadChip and Acknowledgements Sequenom EpiTYPER analysis. The correlations between first trimester We would like to thank Dr. Dan Diego-Alvarez and Ruby Jiang for their placenta and cytotrophoblasts were (A) 0.96 and (C) 0.93, and between technical assistance and Kristal Louie for placenta donor recruitment in third trimester placenta and cytotrophoblasts were (B) 0.88 and (D) 0.88, Vancouver. using Infinium and Sequenom platforms, respectively. This finding Funding suggests that both first and third trimester placenta methylation levels BN is supported by an NHMRC (Australia) Dora Lush Biomedial Postgraduate are indicative of cytotrophoblast levels. The lower correlation in third Scholarship. RKCY is supported by a graduate student scholarship from the trimester is likely due to both lower numbers of villous cytotrophoblasts Child & Family Research Institute. The Vancouver data was funded by a and their differentiation into the syncytiotrophoblast layer. Canadian Institutes for Health Research Grant (to WPR). The Murdoch Childrens Research Institute is supported by the Victorian Government’s Additional file 9: Number of probes showing variation at each Operational Infrastructure Support Program. gestational age. Additional file 10: Relationship between probe methylation level Author details and variation. Relationship between methylation level and inter- 1 Cancer, Disease and Developmental Epigenetics, Murdoch Childrens individual variation at each gestational age (s ). Probes were separated Research Institute, Royal Children’s Hospital and Department of Paediatrics, into three groups: (A) low methylation (b < 0.2), (B) intermediate 2 University of Melbourne, Parkville, Victoria 3052, Australia. Early Life methylation (0.2 < b > 0.6) and (C) high methylation (b > 0.6). The Epigenetics, Murdoch Childrens Research Institute, Royal Children’s Hospital number of probes was plotted on the y-axis (in log scale) and the and Department of Paediatrics, University of Melbourne, Parkville, Victoria variance (s ) on the x-axis. Probes with an intermediate methylation level 3 3052, Australia. Department of Medical Genetics, University of British were most likely to show inter-individual variation (B), while probes with Columbia, Child & Family Research Institute, 950 West 28th Ave., Vancouver, a high methylation level were least likely to show inter-individual 4 BC, Canada. Bioinformatics Unit, Murdoch Children’s Research Institute, variation. In fact, most of the probes with a variance of > 0.02 were from Royal Children’s Hospital, Flemington Road, Parkville, Victoria 3052, Australia. the intermediate methylation level (88/106 first trimester, 119/166 second 5 Department of Pathology, University of Cambridge, Cambridge, CB2 1QP, trimester and 255/352 term), even though most of the Infinium probes UK. (63%) are actually in the low methylation group. Additional file 11: Number of probes showing variation at each Authors’ contributions gestational age. Venn diagram of variable probes (s > 0.02) in each BN analysed the data, participated in critical discussion, and wrote the draft gestational age. This analysis revealed that the vast majority of variable manuscript. LG processed the array data and performed data analysis. RKCY, probes are only variable in third trimester, while 52 were only variable in JMC, RS and WR designed the study, participated in critical discussion and second, and 21 only in first trimester. A total of 47 probes were variable wrote the manuscript. MSP set-up experiments and processed array data. AS across all gestational ages. and AM provided first trimester villi samples. All authors approved of the final manuscript. Additional file 12: Relationship between DNA methylation and genomic context. Probes were separated into two groups based on Received: 25 May 2011 Accepted: 28 October 2011 their genomic location - CpG Island (CGI) or non-CpG Island (non-CGI). Published: 28 October 2011 The expected frequency was based on the proportion of all analysable Infinium probes (A) within a CGI or outside a CGI (0.76 and 0.24, respectively). Probes that increased in methylation over gestation were References predominantly in non-CGI regions (B), while probes that decreased in 1. Huppertz B: The anatomy of the normal placenta. Journal of clinical methylation over gestation showed the expected proportions. pathology 2008, 61:1296-1302. Furthermore, probes that showed inter-individual variation in (D) both 2. von Rango U: Fetal tolerance in human pregnancy–a crucial balance first and third trimesters, (E) first trimester only, and (F) third trimester between acceptance and limitation of trophoblast invasion. Immunology only, were predominantly associated with CGIs. This finding further letters 2008, 115:21-32. suggests that probes that change over gestation are not the same as 3. Huppertz B: Placental origins of preeclampsia: challenging the current those that show inter-individual variation within each gestation. hypothesis. Hypertension 2008, 51:970-975. 4. Jauniaux E, Poston L, Burton GJ: Placental-related diseases of pregnancy: Additional file 13: Relationship between DNA methylation and gene Involvement of oxidative stress and implications in human evolution. expression levels. Relationship between DNA methylation and gene Hum Reprod Update 2006, 12:747-755. expression levels in first, second and third trimester placenta. Infinium Novakovic et al. BMC Genomics 2011, 12:529 Page 13 of 14 http://www.biomedcentral.com/1471-2164/12/529 5. Winn VD, Haimov-Kochman R, Paquet AC, Yang YJ, Madhusudhan MS, hypomethylation of multiple genes in early-onset preeclampsia. Eur J Gormley M, Feng KT, Bernlohr DA, McDonagh S, Pereira L, et al: Gene Hum Genet . expression profiling of the human maternal-fetal interface reveals 27. Loke YW, Gardner L, Grabowska A: Isolation of human extravillous dramatic changes between midgestation and term. Endocrinology 2007, trophoblast cells by attachment to laminin-coated magnetic beads. 148:1059-1079. Placenta 1989, 10:407-415. 6. Mikheev AM, Nabekura T, Kaddoumi A, Bammler TK, Govindarajan R, 28. Robinson CJ, Alanis MC, Wagner CL, Hollis BW, Johnson DD: Plasma 25- Hebert MF, Unadkat JD: Profiling gene expression in human placentae of hydroxyvitamin D levels in early-onset severe preeclampsia. American different gestational ages: an OPRU Network and UW SCOR Study. journal of obstetrics and gynecology 2010, 203:366 e361-366. Reproductive sciences (Thousand Oaks, Calif 2008, 15:866-877. 29. Kerkel K, Spadola A, Yuan E, Kosek J, Jiang L, Hod E, Li K, Murty VV, 7. Frost JM, Moore GE: The importance of imprinting in the human Schupf N, Vilain E, et al: Genomic surveys by methylation-sensitive SNP placenta. PLoS genetics 2010, 6:e1001015. analysis identify sequence-dependent allele-specific DNA methylation. 8. Coan PM, Burton GJ, Ferguson-Smith AC: Imprinted genes in the Nature genetics 2008, 40:904-908. placenta–a review. Placenta 2005, 26(Suppl A):S10-20. 30. Bibikova M, Le J, Barnes B, Saedinia-Melnyk S, Zhou L, Shen R, 9. Maccani MA, Marsit CJ: Epigenetics in the placenta. Am J Reprod Immunol Gunderson KL: Genome-wide DNA methylation profiling using Infinium 2009, 62:78-89. assay. Epigenomics 2009, 1:177-200. 10. Roberts CT: Review: Complicated interactions between genes and the 31. Benerischke K, Kaufmann P: Pathology of the Human Placenta Springer; environment in placentation, pregnancy outcome and long term health. 2006. Placenta 2010. 32. Grigoriu A, Ferreira JC, Choufani S, Baczyk D, Kingdom J, Weksberg R: Cell 11. Waterland RA, Jirtle RL: Transposable elements: targets for early specific patterns of methylation in the human placenta. Epigenetics 2011, nutritional effects on epigenetic gene regulation. Mol Cell Biol 2003, 6:368-379. 23:5293-5300. 33. Lash GE, Otun HA, Innes BA, Bulmer JN, Searle RF, Robson SC: Low oxygen 12. Wolff GL, Kodell RL, Moore SR, Cooney CA: Maternal epigenetics and concentrations inhibit trophoblast cell invasion from early gestation methyl supplements affect agouti gene expression in Avy/a mice. Faseb placental explants via alterations in levels of the urokinase plasminogen J 1998, 12:949-957. activator system. Biol Reprod 2006, 74:403-409. 13. Bruchova H, Vasikova A, Merkerova M, Milcova A, Topinka J, Balascak I, 34. Monk M, Boubelik M, Lehnert S: Temporal and regional changes in DNA Pastorkova A, Sram RJ, Brdicka R: Effect of maternal tobacco smoke methylation in the embryonic, extraembryonic and germ cell lineages exposure on the placental transcriptome. Placenta 2010, 31:186-191. during mouse embryo development. Development 1987, 99:371-382. 14. Zechner U, Pliushch G, Schneider E, El Hajj N, Tresch A, Shufaro Y, 35. Gama-Sosa MA, Midgett RM, Slagel VA, Githens S, Kuo KC, Gehrke CW, Seidmann L, Coerdt W, Muller AM, Haaf T: Quantitative methylation Ehrlich M: Tissue-specific differences in DNA methylation in various analysis of developmentally important genes in human pregnancy mammals. Biochim Biophys Acta 1983, 740:212-219. losses after ART and spontaneous conception. Mol Hum Reprod 2009. 36. Gama-Sosa MA, Wang RY, Kuo KC, Gehrke CW, Ehrlich M: The 5- 15. Zhang Y, Cui Y, Zhou Z, Sha J, Li Y, Liu J: Altered global gene expressions methylcytosine content of highly repeated sequences in human DNA. of human placentae subjected to assisted reproductive technology Nucleic Acids Res 1983, 11:3087-3095. treatments. Placenta 2010, 31:251-258. 37. Nakashima A, Shiozaki A, Myojo S, Ito M, Tatematsu M, Sakai M, Takamori Y, 16. Tsui DW, Chiu RW, Lo YD: Epigenetic approaches for the detection of Ogawa K, Nagata K, Saito S: Granulysin produced by uterine natural killer fetal DNA in maternal plasma. Chimerism 2010, 1:30-35. cells induces apoptosis of extravillous trophoblasts in spontaneous 17. Chim SS, Jin S, Lee TY, Lun FM, Lee WS, Chan LY, Jin Y, Yang N, Tong YK, abortion. Am J Pathol 2008, 173:653-664. Leung TY, et al: Systematic search for placental DNA-methylation 38. Symonds ME, Sebert SP, Hyatt MA, Budge H: Nutritional programming of markers on chromosome 21: toward a maternal plasma-based the metabolic syndrome. Nat Rev Endocrinol 2009, 5:604-610. epigenetic test for fetal trisomy 21. Clinical chemistry 2008, 54:500-511. 39. Koldovsky O: Development of sucrase activity: effect of maternal 18. Chu T, Burke B, Bunce K, Surti U, Allen Hogge W, Peters DG: A microarray- hormonal status and fetal programming of jejuno-ileal differences. Ciba based approach for the identification of epigenetic biomarkers for the Found Symp 1979, 147-168. noninvasive diagnosis of fetal disease. Prenatal diagnosis 2009. 40. Barker DJ, Osmond C, Simmonds SJ, Wield GA: The relation of small head 19. Papageorgiou EA, Fiegler H, Rakyan V, Beck S, Hulten M, Lamnissou K, circumference and thinness at birth to death from cardiovascular Carter NP, Patsalis PC: Sites of differential DNA methylation between disease in adult life. BMJ 1993, 306:422-426. placenta and peripheral blood: molecular markers for noninvasive 41. Barker DJ: The fetal and infant origins of adult disease. Bmj 1990, prenatal diagnosis of aneuploidies. Am J Pathol 2009, 174:1609-1618. 301:1111. 20. Papageorgiou EA, Karagrigoriou A, Tsaliki E, Velissariou V, Carter NP, 42. Osmond C, Barker DJ: Ischaemic heart disease in England and Wales Patsalis PC: Fetal-specific DNA methylation ratio permits noninvasive around the year 2000. J Epidemiol Community Health 1991, 45:71-72. prenatal diagnosis of trisomy 21. Nat Med 17:510-513. 43. Novakovic B, Sibson M, Ng HK, Manuelpillai U, Rakyan V, Down T, Beck S, 21. Yuen RK, Avila L, Penaherrera MS, von Dadelszen P, Lefebvre L, Kobor MS, Fournier T, Evain-Brion D, Dimitriadis E, et al: Placenta-specific methylation Robinson WP: Human placental-specific epipolymorphism and its of the vitamin D 24-hydroxylase gene: implications for feedback association with adverse pregnancy outcomes. PloS one 2009, 4:e7389. autoregulation of active vitamin D levels at the fetomaternal interface. J 22. Turan N, Katari S, Gerson LF, Chalian R, Foster MW, Gaughan JP, Coutifaris C, Biol Chem 2009, 284:14838-14848. Sapienza C: Inter- and Intra-Individual Variation in Allele-Specific DNA 44. Du P, Kibbe WA, Lin SM: lumi: a pipeline for processing Illumina Methylation and Gene Expression in Children Conceived using Assisted microarray. Bioinformatics 2008, 24:1547-1548. Reproductive Technology. PLoS genetics 2010, 6:e1001033. 45. Du P, Zhang X, Huang CC, Jafari N, Kibbe WA, Hou L, Lin SM: Comparison 23. Yuen RK, Penaherrera MS, von Dadelszen P, McFadden DE, Robinson WP: of Beta-value and M-value methods for quantifying methylation levels DNA methylation profiling of human placentas reveals promoter by microarray analysis. BMC bioinformatics 2010, 11:587. hypomethylation of multiple genes in early-onset preeclampsia. Eur J 46. Wong NC, Novakovic B, Weinrich B, Dewi C, Andronikos R, Sibson M, Hum Genet 2010, 18:1006-1012. Macrae F, Morley R, Pertile MD, Craig JM, Saffery R: Methylation of the 24. Chavan-Gautam P, Sundrani D, Pisal H, Nimbargi V, Mehendale S, Joshi S: adenomatous polyposis coli (APC) gene in human placenta and Gestation-dependent changes in human placental global DNA hypermethylation in choriocarcinoma cells. Cancer letters 2008, 268:56-62. methylation levels. Mol Reprod Dev 2011, 78:150. 47. James JL, Stone PR, Chamley LW: The isolation and characterization of a 25. Sood R, Zehnder JL, Druzin ML, Brown PO: Gene expression patterns in population of extravillous trophoblast progenitors from first trimester human placenta. Proc Natl Acad Sci USA 2006, 103:5478-5483. human placenta. Hum Reprod 2007, 22:2111-2119. 26. Yuen RK, Penaherrera MS, von Dadelszen P, McFadden DE, Robinson WP: 48. Novakovic B, Gordon L, Wong NC, Moffett A, Manuelpillai U, Craig JM, DNA methylation profiling of human placentas reveals promoter Sharkey A, Saffery R: Wide ranging DNA methylation differences of Novakovic et al. BMC Genomics 2011, 12:529 Page 14 of 14 http://www.biomedcentral.com/1471-2164/12/529 primary trophoblast cell populations and derived-cell lines: implications and opportunities for understanding trophoblast function. Mol Hum Reprod 2011, 17:344-353. doi:10.1186/1471-2164-12-529 Cite this article as: Novakovic et al.: Evidence for widespread changes in promoter methylation profile in human placenta in response to increasing gestational age and environmental/stochastic factors. BMC Genomics 2011 12:529. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png BMC Genomics Springer Journals

Evidence for widespread changes in promoter methylation profile in human placenta in response to increasing gestational age and environmental/stochastic factors

Loading next page...
 
/lp/springer-journals/evidence-for-widespread-changes-in-promoter-methylation-profile-in-WYOpyH2G2i

References (106)

Publisher
Springer Journals
Copyright
Copyright © 2011 by Novakovic et al; licensee BioMed Central Ltd.
Subject
Life Sciences; Life Sciences, general; Microarrays; Proteomics; Animal Genetics and Genomics; Microbial Genetics and Genomics; Plant Genetics & Genomics
eISSN
1471-2164
DOI
10.1186/1471-2164-12-529
pmid
22032438
Publisher site
See Article on Publisher Site

Abstract

Background: The human placenta facilitates the exchange of nutrients, gas and waste between the fetal and maternal circulations. It also protects the fetus from the maternal immune response. Due to its role at the feto- maternal interface, the placenta is subject to many environmental exposures that can potentially alter its epigenetic profile. Previous studies have reported gene expression differences in placenta over gestation, as well as inter- individual variation in expression of some genes. However, the factors contributing to this variation in gene expression remain poorly understood. Results: In this study, we performed a genome-wide DNA methylation analysis of gene promoters in placenta tissue from three pregnancy trimesters. We identified large-scale differences in DNA methylation levels between first, second and third trimesters, with an overall progressive increase in average methylation from first to third trimester. The most differentially methylated genes included many immune regulators, reflecting the change in placental immuno-modulation as pregnancy progresses. We also detected increased inter-individual variation in the third trimester relative to first and second, supporting an accumulation of environmentally induced (or stochastic) changes in DNA methylation pattern. These highly variable genes were enriched for those involved in amino acid and other metabolic pathways, potentially reflecting the adaptation of the human placenta to different environments. Conclusions: The identification of cellular pathways subject to drift in response to environmental influences provide a basis for future studies examining the role of specific environmental factors on DNA methylation pattern and placenta-associated adverse pregnancy outcomes. Background syncytiotrophoblast), fibroblasts, mesenchymal cells, as The human placenta is a temporary organ that facilitates well as fetal and maternal vascular tissue and blood the exchange of nutrients, gas and waste between mater- cells. The extra-villous trophoblast cells must first nal and fetal circulations. In order to carry out these invade the maternal decidua and remodel maternal functions, it is comprised of heterogeneous cell types arteries, to allow direct contact between maternal blood and the placental syncytiotrophoblast cell layer [1]. In including several trophoblast cell populations (cytotro- phoblasts, extra-villous trophoblasts and addition to these traditional roles, the placenta is also important in shielding the developing fetus from the maternal immune system [2]. * Correspondence: richard.saffery@mcri.edu.au The placenta also undergoes several physiological † Contributed equally Cancer, Disease and Developmental Epigenetics, Murdoch Childrens changes throughout gestation, with one of the most sig- Research Institute, Royal Children’s Hospital and Department of Paediatrics, nificant being the flooding of placenta villi by maternal University of Melbourne, Parkville, Victoria 3052, Australia blood at the end of the first trimester (~12 weeks Full list of author information is available at the end of the article © 2011 Novakovic et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Novakovic et al. BMC Genomics 2011, 12:529 Page 2 of 14 http://www.biomedcentral.com/1471-2164/12/529 gestation), resulting in a rise in oxygen concentration as between unrelated individuals. We proposed that these well as a decrease in trophoblast invasion. It is believed CpG sites may be especially susceptible to environmen- that the inability of the placenta to respond to this tally-induced changes associated with placental disease change in oxygen concentration can lead to placental [21]. In a follow-up investigation, we also observed a disease, such as preeclampsia [3,4]. gestational age difference in DNA methylation profile in The molecular mechanisms behind these morphologi- the placenta across the third trimester [23], while others have recently reported an increase in global DNA cal and functional changes are now beginning to be methylation levels between pre-term (28 weeks) and full understood at both the gene-specific and genome-wide term placenta (40 weeks) [24]. level. Wide-ranging genome-wide gene expression differ- ences between placentas at different gestational ages Theaim ofthecurrent studywas to buildonrecent were reported in two recent studies [5,6]. Despite sam- knowledge obtained through genome-scale gene expres- pling from different locations within the placenta, many sion [5,6,25] and DNA methylation analysis [24,26] of changes were found in common between the two stu- the human placenta. In the current study, we used the dies, each of which reported changes in expression with Illumina Infinium HumanMethylation27 BeadChip plat- increasing gestational age in genes involved in cell cycle form to identify promoter regions subject to change and immune response. This suggests that gene expres- throughout gestation. In addition we wanted to identify sion changes are needed for physiological needs of the those that become increasingly variable between indivi- developing placenta, such as shielding the fetus from the dual placentas over time, possibly in response to accu- maternal immune system [2]. Genes involved in Wnt mulated environmental exposures. signalling also showed expression changes over time [5,6] that resulted in decreasing levels of b-catenin later Results in gestation, possibly linked to decreasing placental inva- Genome-scale DNA methylation analysis of first, second, siveness [6]. and third trimester placenta The importance of epigenetic factors in placental Genome-scale DNA methylation analysis of 18 first tri- development and function has long been known through mester (8-12 weeks), 10 second trimester (17-24 weeks) the study of imprinted genes [7,8] and it is increasingly and 14 third trimester placenta (34-41 weeks) samples clear that the placenta displays a unique epigenetic pro- was performed using the Illumina Infinium Human- file. However, the extent to which epigenetic modifica- Methylation27 BeadChip (see Additional file 1; Data was tions, specifically DNA methylation, contribute to deposited into the NCBI Gene Expression Omnibus, placental function have only recently been widely exam- accession number: GSE31781). Following normalisation ined (reviewed in [9]. and data cleaning (see Methods), a total 26, 162 probes Due to its role as the interface between the mother were available for subsequent analysis. Correlations (r ) and fetus, the placenta is exposed to a myriad of envir- of average methylation of probes between gestational onmental factors, some of which have been shown to ages ranged from 0.935 (first v third trimester) to 0.97 alter placental gene expression, as well as epigenetic (second v third trimester; see Additional file 2). Unsu- marks [10]. These include diet [11,12], smoking [13], pervised hierarchical clustering of all probes clearly and assisted reproductive techniques [14,15]. Mounting delineates samples according to gestational age (Figure evidence implicates epigenetic marks, such as DNA 1) with further evidence for a closer relationship methylation, in mediating environmentally-induced reg- between second and third trimester profile than with ulation of genome function. More studies into the first trimester. Validation of methylation levels at 12 effects of the environment on the placental epigenome Infinium probes (representing 12 genes) in 9 placental are warranted due the importance of this organ in regu- samples (total data points: 49) using the Sequenom Epi- lating pregnancy development. TYPER platform confirmed the robust nature of the Several genome-scale DNA methylation studies have Infinium data (r = 0.76) (see Additional file 3). The focused on finding tissue-specific differentially methy- probes used for validation were chosen due to their lated regions (tDMRs) between placenta and maternal association with genes with known or predicted impor- blood, as a means of detecting placental pathologies and tant roles in regulating placental function (Additional fetal chromosomal trisomies using non-invasive methods file 4). (reviewed in [16-19]. This strategy has recently resulted in the development of the first non-invasive blood test Transition to normoxia is not associated with major for Down syndrome [20]. However, we, and others have changes in placental DNA methylation profile revealed substantial inter-individual DNA variation in Despite the overall interspersed pattern of clustering of placental methylation profile [21,22], with a subset of first trimester samples of various gestations (8 - 12 CpG sites more likely to be differentially methylated weeks) (Figure 1), we tested to see whether any genomic Novakovic et al. BMC Genomics 2011, 12:529 Page 3 of 14 http://www.biomedcentral.com/1471-2164/12/529 Figure 1 Cluster dendrogram based on all autosomal Infinium probes distinguishes placentas of different gestational age. Dendrogram showing the relationship between placental samples from three gestational ages based on DNA methylation levels (b- values) of all analysable Infinium probes. All samples clustered within their gestational age group, with no overlap between gestations, suggesting there are consistent genome-scale DNA methylation patterns associated with each gestational age. First Figure 2 Average methylation of all samples for first, second trimester samples clustered away from second and third trimester and third trimester. Methylation Index (MI) was calculated for each samples, indicating that overall Infinium methylation patterns are sample by calculating the mean of all analysable Infinium b-values more similar in second and third trimester compared to first (26, 162 probes) for that sample. The MIs were then grouped by trimester. gestation and shown as box and whisker plots. First and second trimester placentas show a similar overall level of methylation (p = 0.46) with median MIs of 0.238 and 0.241, respectively. Third trimester samples show significantly elevated average MI values regions undergo selective changes in methylation during (median = 0.256) relative to both first and second trimester, the transition from early first trimester to late first/early indicating that there is a significant increase in methylation level second trimester. This is the period widely regarded as from second to third trimester. the time when placental intervillous space is flooded with oxygenated maternal blood. Only limited DNA clearly apparent when the relative methylation levels for methylation differences were observed between 8 and 12 all probes are displayed in a scatterplot (Additional file week placentas (Additional file 5). A total of only 12 2). All 3 trimesters have the same proportion of probes probes (3 hypomethylated and 9 hypermethylated at 12 with b-values between 0 and 0.2 (~63% of total probe weeks relative to 8 weeks) showed consistent methyla- number). However, as gestation progresses, there is an tion differences (Δb) of 0.2 between the two gestational increase (from 13% to 17%) of highly methylated probes ages. This suggests that promoter DNA methylation (b > 0.6) (Additional file 6). plays a limited role in any physiological changes in the Absolute differences in mean methylation between placenta that occur in response to altered oxygen status. first and second, first and third, and second and third trimesters were generally small, with only 149, 954 and Gestational age is associated with promoter methylation 157 probes respectively, showing Δb >0.2 (Table 1). levels Further analysis showed that the 883 probes that In order to gauge the effects of gestational age on the increased in methylation from first to third trimester overall methylation level at gene promoter regions were predominantly those with intermediate methylation enriched on the Infinium HumanMethylation27 Bead- Chip, the mean methylation across all probes (Methyla- levels (average b range 0.3 - 0.5) in the first trimester tion Index - MI) was calculated for each sample (Figure (Additional file 7). Figure 3 shows a heat map with 2). The mean MI for first, second and third trimester unsupervised clustering of samples based on the 954 st placenta was 0.240, 0.242 and 0.256, respectively. Thus, probes showing a Δb >0.2between1 trimester and there is an overall increase in methylation between sec- term. Importantly, samples from all three trimesters -5 ondand thirdtrimesters(p=5×10 ,student’s t-test). showed distinct methylation patterns. More specifically, No significant differences were detected between first second trimester samples did not cluster with either the and second trimesters (p = 0.46) (Figure 2). This is first trimester or term samples, suggesting a progressive Novakovic et al. BMC Genomics 2011, 12:529 Page 4 of 14 http://www.biomedcentral.com/1471-2164/12/529 Table 1 Number of probes showing differential methylation between first, second and third trimester placental tissue. Comparison Differentially methylated probes Difference b ≥ 0.1 and adj. p < 0.05 Difference b ≥ 0.2 and adj. p < 0.05 (adj. p < 0.05) First v Second 8, 240 411 ↓ 12 ↓ 1, 077 ↑ 137 ↑ First v Third 8, 298 755 ↓ 71 ↓ 2, 581 ↑ 883 ↑ Second v Third 7, 669 288 ↓ 6 ↓ 1, 515 ↑ 151 ↑ ↑: Increase in methylation; ↓: Decrease in methylation change in methylation across gestation. Interestingly, the major factors contributing to methylation differences none of the probes showed a ‘fluctuating’ pattern of between the two gestational time points. In order to test methylation across gestation (i.e. showed a similar level this, methylation levels between purified first trimester of methylation in first and third trimesters, but hypo/ cytotrophoblasts and first and third trimester placenta, hypermethylation in second trimester) based on our cri- were examined using both Infinium and Sequenom Epi- teria of Δb > 0.2, and only 103 probes showed this pat- TYPER analyses (22 data points) (Additional file 8). tern with a Δb > 0.1. This further suggests that most Methylation levels in purified cytotrophoblasts were more methylation changes occur in a progressive manner. similar to first trimester placental tissue (r =0.96and Changes in cell composition, primarily a decrease in 0.93) than third trimester (r = 0.88 and 0.88) as calculated cytotrophoblasts, from first to third trimester could be one by Infinium and Sequenom EpiTYPER, respectively. Figure 3 Unsupervised clustering based on probes with Δb > 0.2 between First and Third trimester. HeatMap showing unsupervised clustering of all placenta samples (x-axis) based on 954 probes with a Δb > 0.2 between First and Third trimester (y-axis). The majority of differentially methylated probes show higher methylation in third trimester (883 probes) compared to only 71 probes with lower methylation in third trimester. Second trimester placentas cluster as a separate group, and show a methylation profile that is an intermediate of first and third trimesters. Green corresponds to low methylation and Red to high methylation. Novakovic et al. BMC Genomics 2011, 12:529 Page 5 of 14 http://www.biomedcentral.com/1471-2164/12/529 Differentially methylated genes between first trimester, (s <0.009) at alltimepoints, thenumberwith an second trimester and term inter-placental variance of > 0.01 was increased in the Probes that showed differences of Δb > 0.2 (approximat- third trimester relative to the earlier time points (Figure ing 20% differential methylation) between each of the 4). More specifically, the third trimester group was three gestational ages, (Table 1), were analysed using enriched for probes with the highest variance (s > Ingenuity Pathways Analysis (IPA). Enrichment of net- 0.02), with 352 such probes compared to 106 (c = works, pathways and gene functions was calculated 133.3, p < 0.001) and 166 (c = 66.7, p < 0.001) in first using the Ingenuity Pathways software. Top enriched and second trimester placentas respectively (Additional canonical pathways for genes differentially methylated file 9). To facilitate the comparison of variation level between first and second trimesters (110 genes), first between trimester groups, we defined sites with variance and third trimesters (654 genes) and between second > 0.02 as highly variable in methylation. and third trimesters (106 genes) are listed in Table 2. Further analysis revealed that most of the gestational ‘Communication between innate and adaptive immune age associated variation was found in probes with cells’ was the most significantly enriched pathway, with intermediate methylation (0.2 < b < 0.6) rather than at least 4 of the top 5 pathways in each comparison low (b <0.2)orhigh(b > 0.6) methylation (Additional being immune-related. file 10; Additional file 9). This suggests that the increased variation observed between different full- Inter-placental methylation variation increases with term placentas is not necessarily a by-product of gestational age increasing methylation across gestation. In addition, DNA methylation may be modulated in part by environ- the increasing methylation across gestation also sug- mental influences and may serve as a mediator between gests that the increasing variability was unlikely caused the environment and genome function (reviewed in by a lack of maintenance of DNA methylation by [27]). We previously investigated the inter-individual DNMTs in the human placenta. variability of DNA methylation in the human placenta While inter-individual variation of DNA methylation [21] and proposed that the highly variable methylation may in part reflect genetic polymorphisms [29], this found in the placenta may be a consequence of cumula- source of variation would be anticipated to be repre- tive response to the intrauterine environmental expo- sented equally across all 3 gestational ages. To investi- sures [28]. To test this further, we calculated the inter- gate the inter-placental variation of methylation in placental variance of all probes within each of the first, more detail, variance level of probes for the third tri- second and third trimester time points. While the vast mester was plotted against the first trimester (Figure majority of probes (95-98%) showed very little variation 5). This revealed that the majority of the highly Table 2 Top Canonical Pathways from IPA for probes showing differential methylation across gestation First v Second trimester p-value # differentially methylated genes/# genes in the pathway Communication between Innate and Adaptive Immune Cells 6.39E-05 6/109 Role of Cytokines in Mediating Communication between Immune Cells 1.53E-03 4/56 Altered T Cell and B Cell Signaling in Rheumatoid Arthritis 6E-03 4/92 Calcium Signaling 6.81E-03 6/204 Crosstalk between Dendritic Cells and Natural Killer Cells 1.04E-02 4/97 Second v Third trimester p-value # differentially methylated genes/# genes in the pathway Systemic Lupus Erythematosus Signaling 4.72E-03 5/166 Crosstalk between Dendritic Cells and Natural Killer Cells 8.2E-03 4/97 Role of NFAT in Regulation of the Immune Response 2.05E-02 5/199 Agrin Interactions at Neuromuscular Junction 2.35E-02 3/69 Wnt/b-catenin Signaling 2.51E-02 5/172 First v Third trimester p-value # differentially methylated genes/# genes in the pathway Communication between Innate and Adaptive Immune Cells 1.69E-07 17/109 Systemic Lupus Erythematosus Signaling 1.9E-06 21/166 Role of Cytokines in Mediating Communication between Immune Cells 1.43E-05 12/56 Altered T Cell and B Cell Signaling in Rheumatoid Arthritis 1.79E-04 13/92 Crosstalk between Dendritic Cells and Natural Killer Cells 2.61E-04 14/97 Novakovic et al. BMC Genomics 2011, 12:529 Page 6 of 14 http://www.biomedcentral.com/1471-2164/12/529 Figure 4 Number of variable probes increases with gestation. The number of probes with high inter-individual variation increases over gestation. Variance (s ) of each probe was calculated for each gestational age. The number of probes (y-axis; log scale) showing a particular level of variance (x-axis) is shown. Most probes (95% in third trimester to 98% in first trimester) show low variation (s2 < 0.009). However, there is an increase in the number of variable probes (s > 0.02) in third trimester placentas, and to a lesser extent second trimester placentas, compared to first trimester. variable sites “gain” variability in the later gestations trimester were not context dependent (c = 2.68, p = (area (c) in Figure 5). Numerically, the probes with 0.14 (Additional file 12). We also examined the genomic the highest variance are found exclusively in the third context of the variable probes falling within the three trimester (223), while far fewer probes show high var- categories shown in Figure 5(A-C). Probes with high iation across all three gestations (47) (Additional file variation in both first and third trimester (Additional 11). file 12-D) are associated with CGIs (p < 0.05). Variable The top pathways for genes that show variable methy- probes in the first, but not the third trimester are lation at each gestastional age are listed in Table 3. The slightly enrichmed for CGI regions (p = 0.05), and those top 5 pathways in the third trimester include ‘amino that were variable in third, but not first trimester show acid metabolism’, while first and second trimester lists a strong association with CGIs (p < 0.001) (Additional both include ‘circadian rhythm signalling’. This supports file 12). We also looked at the relationship between the our hypothesis of an increasing accumulation of epige- distance from transcription start site and DNA methyla- netic variation in response to cumulative environmental tion. There was no association with distance from TSS exposures. and differential or variable methylation (data not shown). CpG density and genomic context influence variability in placental methylation DNA methylation profile influences global gene In order to examine the relationship between CpG den- expression in the placenta sity and methylation status, probe locations were In order to assess the overall effect of DNA methylation assigned to either CpG Island (CGI) or non-CpG Island on gene expression profile, methylation b-values levels (non-CGI) genomic regions, as annotated by Illumina. for first, second and third trimester placental tissue Using a chi square test, we found that probes showing were correlated with publicly available expression data increased methylation across gestation were enriched for for matched gestational age placenta [5,6]. Infinium non-CGI regions (c = 480.2, p < 0.001), while probes probes were quartiled according to methylation level, showing lower methylation in third compared to first with b value ranges of 0.01-0.06 (bottom 25% of probes), Novakovic et al. BMC Genomics 2011, 12:529 Page 7 of 14 http://www.biomedcentral.com/1471-2164/12/529 Figure 5 Variance levels of probes in the third compared to the first trimester. Scatter plot of probe variance (s ) at first trimester (x-axis) and third trimester (y-axis). Dots represent individual probes and the vertical red dotted line marks s = 0.02 for first trimester, while the horizontal red dotted line marks s = 0.02 for third trimester. Probes on the outside of the red line are deemed ‘variable’. This analysis revealed that there are 73 probes (A) which are highly variable in both first and third trimester. Only 33 probes (B) were variable in first, but not third trimester, while 279 probes (C) were variable in third, but not first trimester. This analysis suggests that most of the variable probes become so throughout pregnancy, supporting the hypothesis that accumulating environmental factors contribute to inter-individual variation in DNA methylation, in term placenta. Table 3 Top Canonical Pathways for gene-associated probes that show variable methylation within each gestational age Variable in First trimester p-value genes Arachidonic Acid Metabolism 6.97E-04 CBR1, CYP2E1, GPX7, PNPLA3, PTGS1 Hepatic Fibrosis/Hepatic Stellate Cell Activation 1.36E-02 CYP2E1, ECE1, FGFR1, MYL5 Circadian Rhythm Signaling 1.91E-02 GRIN3A, VIPR2 Calcium Signaling 2.69E-02 CHRNB4, GRIA4, GRIN3A, MYL5 Nitrogen Metabolism 3.54E-02 PTPRG, VNN3 Variable in Second trimester p-value genes Circadian Rhythm Signaling 3.39E-03 GRIN3A, VIPR2, PER1 Glutathione Metabolism 1.62E-02 GPX3, GPX7, GSTO1 Arachidonic Acid Metabolism 1.86E-02 CBR1, CYP2E1, GPX3, GPX7 Sonic Hedgehog Signaling 3.7E-02 DYRK1B, HKR1 Metabolism of Xenobiotics by Cytochrome P450 4.68E-02 ADHFE1, CYP2E1, GSTO1 Variable in Third trimester p-value genes Glutamate Receptor Signaling 6.1E-03 GRID2, GRIK2, GRIN3A, GRM6, SLC1A6 Valine, Leucine and Isoleucine Degradation 1.13E-02 ACADL, ALDH1A3, ELOVL2, IVD, OXCT1 b-alanine Metabolism 2.07E-02 ACADL, ALDH1A3, DPYS, IVD Butanoate Metabolism 3.74E-02 ALDH1A3, ELOVL2, OXCT1, PDHA2 Tyrosine Metabolism 4.19E-02 ADHFE1, ADLH1A3, ELOVL2, MGMT Novakovic et al. BMC Genomics 2011, 12:529 Page 8 of 14 http://www.biomedcentral.com/1471-2164/12/529 0.06-0.15 (25-50%), 0.15-0.47 (50-75%) and 0.47-0.98 Illumina Infinium HumanMethylation27 BeadChip, in (top 25% of probes), and plotted against gene expression placentas of different gestational ages (18 first trimester levels for linked genes data (Additional file 13). A gen- (8-12 weeks), 10 second trimester (17-23 weeks), and 14 eral decrease in median expression level with increasing third trimester samples (34-41 weeks)). We further vali- methylation level was observed for all three gestational dated the array data by targeting 12 CpG sites (covering ages. In particular, there was a distinct down regulation 12 gene promoters) using the Sequenom EpiTYPER in median gene expression between the second (50%) to platform. The correlation of r = 0.76 between the two third quartiles (75%) associated with a change in methy- platforms is comparable to that published for similar lation range from b <0.15 to b > 0.15. In addition the comparisons [30]. We found evidence for both a pro- range of expression was reduced in the higher methyla- grammed change of methylation in gene promoters tion quartiles (Additional file 13). Using the same across gestation and an increase in variability of methy- expression data sets, we plotted methylation change lation as gestation progresses. We predict that this is from first to third trimester against changes in gene directly related to the cumulative effect of intrauterine expression across the same time points (Figure 6). environmental exposure, however, the contribution of While methylation change of less than 0.2 were gener- stochastic events to the observed variation cannot be ally not associated with changes in gene expression, sev- discounted at this time. Further investigation is war- eral genes showed both higher methylation and lower ranted to determine the effects of specific environmental expression in third trimester compared to first trimester. exposures on DNA methylation patterns of the genes These included several immune-regulators ranked highly identified as variable in the present study. by IPA (CCR7 and CCL21) and one with known func- Unsupervised clustering based on all Infinium probes tion in placental development (GNLY; Granulysin) (Fig- (excluding those on the sex chromosomes) clearly sepa- ure 6). Additional file 14 lists additional genes that rated all samples by gestational age, with first trimester showed concordant differences in methylation and samples clustering away from second and third trimester expression between first and third trimesters. samples (Figure 1). This indicates that there are consis- tent, large scale changes in DNA methylation across Discussion gestation. There are several possible explanations for In this study we performed genome-wide DNA methyla- these temporal differences, one of which is the change tion analysis of gene promoter regions, using the in cell composition and differentiation from first to Figure 6 Correlation between methylation and gene expression change between first and third trimester. Methylation difference (Δb) between first and third trimester (x-axis) was plotted against gene expression log fold change (y-axis) between first and third trimester. A positive change in log fold expression indicates higher expression in first trimester, while a positive change in methylation indicates higher methylation in the third trimester. Therefore, the top left panel includes genes which showed lower methylation and higher expression in first compared to third trimester. The three highlighted genes (CCR7, GNLY and CCL21) ranked highly in IPA analysis. Grey dots represent Infinium probes. Black dots represent specific genes of interest. Novakovic et al. BMC Genomics 2011, 12:529 Page 9 of 14 http://www.biomedcentral.com/1471-2164/12/529 third trimester, especially the relative loss of cytotropho- repetitive elements [35,36]. A recent study has reported blasts throughout gestatation, (85% of the total tropho- a positive correlation between global DNA methylation blast population in first trimester but only 15% of the levels (as measured by an ELISA-like assay with 5- trophoblast volume at full term) [31]. In support of this methylcytosine antibody) and gestational age in the pla- playing at least a partial role in the observed change in centa [24], supporting our data for an increasing level of methylation over time, we found a slightly higher corre- promoter-associated DNA methylation during gestation, particularly from second to third trimester (Figure 2). lation between the methylation profile of purified pri- This accumulation of methylation was most apparent in mary cytotrophoblasts with first trimester placental 2 2 genomic regions that showed an intermediate level of tissue (R = 0.96) than with third trimester tissue (R = 0.88; see Additional file 8). Similar data have recently methylation in first trimester, suggesting that these been reported for second trimester placenta tissue genes are the most likely to be epigenetically regulated which has been shown to be more similar to cytotro- by DNA methylation. Furthermore, these CpG sites phoblasts than mesenchyme in terms of methylation were more likely to be in lower CpG density regions profile [32]. However, the greater part of trophoblast (non-CGI) (see Additional file 12) suggesting that volume at full term is syncytiotrophoblast, arising from methylation levels of CpG sites within CpG Islands are the fusion of cytotrophoblast cells into a multinucleated more stable across gestation in this tissue. The higher layer. Thus, the observed trend of increasing methyla- promoter methylation at term could also reflect the end tion may equally be due to differentiation of the cytotro- point of a continual process of re-methylation in the phoblast component, or alternatively, may even be due extra embryonic lineage from the blastocyst stage, at to other aspects of altered placental function known to whichpoint thegenomeisalmostcompletelyhypo- occur as gestation progresses. methylated [34]. Despite the interspersed clustering of first trimester IPA analysis of genes showing differential methylation samples based on the overall Infinium methylation pat- between first and third trimesters indicated that the terns, we were interested in assessing DNA methylation most affected pathways were ‘communication between changes occurring at the transition from the first to the innate and adaptive immune cells’ and other ‘immune- second trimester. This period of placental development related’ pathways (Table 2). Genes in common between is characterised by the loosening of trophoblast plugs two or more of the top 5 immune-related pathways and the associated rise in oxygen concentration from 2- included immune regulators CCR7, CD28, CSF2, 3%,to7-8%[33]. Therapid increase in oxygen levels IFNA17, IFNA2, IFNA21, IFNB1, IL1F7, IL2, IL3, IL5, LTA, TLR6, TLR9, TNF, TNFRSF13B and TNFSF13B. can lead to the accumulation of reactive oxygen species (ROS), which might be linked to the reduction in tro- This is in accordance with previous gene expression stu- phoblast invasion and migration observed at about this dies, which also showed substantial enrichment of time. Furthermore, it has been suggested that the inabil- immune regulators amongst the most differentially ity of the placenta to adapt to this increase in oxygen expressed genes in placenta at different gestational ages concentration may lead to the development of pree- [5,6]. Several immune-regulators showed a strong corre- clampsia [3]. In this study, we found only 12 CpG sites lation between methylation and expression change from with Δb > 0.2 between these time points, suggesting first to third trimester, including Granulysin (GNLY), that DNA methylation changes are unlikely to play a which has previously been implicated in spontaneous major role in the physiological changes in placentation abortions [37] (Figure 6). Given the critical role of the associated with transition from low oxygen to a normal placenta in modulating the maternal immune response oxygen environment. It must be noted however, that we to the developing pregnancy, and mounting evidence for do not have data on the level of vascularisation for our a role of environmental exposures in controlling both placental samples, so it is possible that flooding of the immune-system development and epigenetic profile, it is maternal blood may not have occurred at the time the not surprising that immune regulators are amongst the 12 week tissue was collected. most variably methylated gene groups. The separation of the trophectoderm from the inner In addition to identifying CpG sites that consistently cell mass at the blastocyst stage occurs at a time when change over gestation, we were also interested in CpG genome-wide DNA methylation levels are at their lowest sites that show inter-individual variability within each [34]. The subsequent re-establishment of methylation gestational age. Genes that show inter-individual differ- ences in expression in placenta have previously been marks occurs at a slower rate, and to a lesser extent, in described by Sood et al. [25]. We have previously the extra-embryonic lineage compared to somatic tissues reported a subset of CpG sites showing variable methy- [34]. This accounts for the low global DNA methylation in human placenta, which is more similar to human lation in term placenta between unrelated individuals, tumours, and is mostly due to hypomethylation of and suggested that these CpG sites may be more Novakovic et al. BMC Genomics 2011, 12:529 Page 10 of 14 http://www.biomedcentral.com/1471-2164/12/529 susceptible to change under adverse conditions [21]. We expression levels at all three gestational time points. have now extended these findings by demonstrating that However, thelarge rangein gene expressionin all four although ~95% of all probes showed very little variation methylation quartiles, and the lack of strong correlation (s < 0.01) between individuals, the number of highly between methylation and expression change across variable probes (s > 0.02) increased with gestational gestation (Figure 6), support previous data that promo- age. Our data support a model whereby increasing varia- ter DNA methylation is only one of the mechanisms controlling gene expression, with many other factors bility in methylation profile between placentas arises in combining to regulate the expression of genes, including response to cumulative differences in environmental histone modifications, non-coding RNAs, and transcrip- exposure rather than sequence polymorphisms (Figure 5). The most variable probes in the first trimester were tion factors. associated with genes involved in, ‘calcium signaling’ and ‘nitrogen metabolism’,while ‘circadian rhythm sig- Conclusions nalling’ and ‘arachidonic acid metabolism’ were In the present study, we performed genome-scale DNA enriched in both first and second trimester placenta methylation analysis of human placenta at three distinct (Table 3). The most enriched variable pathways at term gestational ages, using the Infinium HumanMethyla- included ‘glutamate receptor signaling’, ‘valine, leucine tion27 array. This platform targets over 14, 000 genes, and isoleucine degradation’,and ‘b-alanine’, ‘butanoate’, however, is limited to the 5’ promoter region. Our find- and ‘tyrosine’ metabolism. The enrichment of pathways ings support the hypothesis that DNA methylation levels involved in metabolism further supports our hypothesis in the human placenta are dynamic and change over that the methylation variation is occurring in response gestation, possibly in response to changing cellular com- to environmental influences, potentially as part of an position and/or cumulative environmental influences. ‘adaptive’ response of the developing pregnancy, thereby The identification of pathways that are likely to be allowing changes in gene expression that may be more affected by the latter (i.e. those that show higher varia- beneficial under altered environmental conditions [38]. tion at full term) will provide valuable candidates for The placental epigenome regulates placental gene testing in studies examining placenta-associated adverse expression and function, and any disruption in placental pregnancy outcomes. epigenetics (via either stochastic or environmental per- turbation) has the potential to affect the developing Methods fetus. Therefore it will be interesting in future to exam- Sample collection ine the potential for role of placental epigenome varia- Use of cells and DNA isolated from 8- and 12-week tion with the now widely accepted phenomenon of ‘fetal gestational tissue of normal pregnancies was approved programming’ first described in the context of long by the Cambridgeshire Research Ethics committee st term effects of intrauterine environment on offspring (CREC 04/Q0108/23). Additional 1 trimester (8-12 nd [39] and subsequently developed by Barker and collea- week gestation) and 2 trimester (17-24 week gestation) gues in the early 1990s [40-42]. placental tissue was collected from elective abortions Finally, contrary to CpG sites that increase in methyla- with the approval from the ethics committees of the tion across gestation, variable CpG sites are more likely University of British Columbia and the Children’s& to be located within CpG islands (Additional file 12). Women’s Health Centre of British Columbia. Term pla- This coupled with the finding that hypermethylated cental tissue was collected as previously described probes (b > 0.6) do not show higher variation compared [21,23,43]. Detailed information of the placental samples to probes with low methylation levels (b <0.2)(see can be found in Additional file 1. Additional file 10) suggests that increased variation in late gestation is not a by-product of increasing methyla- DNA extraction tion across gestation. Tissue samples were incubated at 50°C overnight with The direct role of DNA methylation in controlling shaking in DNA extraction buffer (100 mM NaCl, 10 placental global gene expression levels was examined by mM Tris.HCl pH8, 25 mM EDTA, 0.5% (w/v) SDS), comparing our DNA methylation data with published containing 200 μg/ml proteinase K. DNA was isolated gene expression data for placenta tissue of the same by two rounds of phenol:chloroform extraction, followed gestational age (Figure 6 and Additional files 13 and 14). by RNAse A treatment, precipitation in absolute ethanol A general trend of decreasing expression in response to containing 10% (v/v) sodium acetate (3 M, pH 5.2), and increasing DNA methylation was found as expected. resuspended in 100 μl nuclease-free water (Ambion, Further, the range of expression also decreased in the Austin, TX, USA) or using salting out method followed highly methylated probe group. These findings confirm by purification with Qiagen blood and tissue kit (Qia- that promoter DNA methylation influences global gene gen, Mississauga, ON, USA). DNA was stored at -20°C. Novakovic et al. BMC Genomics 2011, 12:529 Page 11 of 14 http://www.biomedcentral.com/1471-2164/12/529 Infinium DNA methylation analysis Raw data obtained from MassArray EpiTYPING was Infinium arrays were hybridized and scanned as per cleaned systematically using an R-script with the follow- manufacturer’s instructions (Illumina, San Diego, USA). ing three criteria (in order): (1) removal of samples that Individual probe b-values (range 0-1) were are approxi- failed across 100% of CpG sites, (2) removal of CpG mate representations of the absolute methylation per- sites that failed (for example due to low or high mass, centage of specific CpG sites within the sample silent peaks, or low intensity readings) across more than population. The values were derived by comparing the 40% of samples, and (3) removal of samples that failed ratio of intensities between the methylated and to generate data for more than 70% of CpG sites tested. unmethylated alleles using the following formula: Publicly available gene expression microarray data Max Signal B,0 analysis β value = [Max(Signal A,0) + Max(Signal B,0)] Gene expression data was downloaded from Gene Expression Omnibus (Barrett and Edgar, 2006, Sayers et WhereSignalBis the arrayintensityvaluefor the al., 2009), (Database issue D885-D890). CEL files were methylated allele and Signal A is the non-methylated downloaded from series GSE9984 [6] and GSE5999 [5] allele. Samples were processed using the Bioconductor and processed with the Bio-conductor package ‘gcrma’ package lumi, which is specifically designed for Illumina [47]. DNA methylation and gene expression data for the data [44,45]. Samples were assessed for quality, color- corresponding gestational age was compared as pre- adjusted to take into account the difference between viously described [48]. The two expression matrices batches, background-corrected and ssn normalized. Any were merged based on probe ID as both were generated probe within a sample with a detection p value of 0.05 using the same array platform, the Affymetrix Human or greater was excluded from further analysis. Probes on Genome U133A Array. Gene expression data was then the × and Y chromosomes were removed from further linked with the methylation data according to gene analysis to eliminate sex-specific differences in methyla- name. Sample quartiles were produced from the methy- tion, leaving 26, 162 analysable probes. A batch correc- lation data. The expression values of the genes in each tion effect was applied to the data in order to remove quartile were then plotted as box and whisker plots. noise produced by processing samples at different times. Differentially methylated probes were defined as having Additional material a Δb > 0.2 between groups, with an adjusted (Benjamini) p value < 0.1. Additional file 1: Summary of placental samples analysed in the study. Gene Ontology and Pathway analysis Additional file 2: Correlations between average methylation in first, Data sets were interrogated using the Ingenuity Path- second and third trimesters. Correlations between average methylation in (A) first and second trimesters, (B) second and third trimesters, and (C) ways Analysis (IPA) application (Ingenuity Systems, first and third trimesters. This analysis revealed that first and third Redwood City, CA; http://www.ingenuity.com). IPA was trimester methylation was most discordant, as expected, while second used to identify enriched canonical pathways, gene net- trimester placenta is more similar to third trimester placenta in terms of overall promoter DNA methylation. Furthermore, this figure visually works, functional classes, and toxicity lists (molecules shows the increase in methylation in third trimester compared to first involved in known toxicity processes). (C). Additional file 3: Correlation between Infinium and Sequenom Locus-specific methylation analysis methylation levels. Correlation between Infinium HumanMethylation27 BeadChip and Sequenom EpiTYPER locus-specific methylation analysis. Sequenom MassARRAY EpiTYPING was performed to Methylation levels in 12 genes were measured using Sequenom validate Infinium methylation, as previously described MassARRAY Epityping targeting the same CpG sites interrogated on the [46]. Sequenom assays were designed to target specific Infinium BeadChip Arrays. Correlation between platforms was 0.76, supporting the use of the Infinium HumanMethylation27 BeadChip for Infinium probes. Genomic sequences for assay design profiling DNA methylation in this study. Genes interrogated are listed in were extracted from the UCSC genome browser http:// Additional file 4. www.genome.ucsc.edu/. Primer pairs for amplification Additional file 4: Sequenom EpiTYPER primer sequences. were designed using EpiDesigner web tool http://www. Additional file 5: Unsupervised clustering of first trimester placenta epidesigner.com/. The primers are listed in Additional based on differentially methylated probes between 8 and 12 weeks gestation. HeatMap showing unsupervised clustering of 8 and 12 week file 4. Amplification was performed after bisulfite con- placenta samples based on 12 probes that showed a Δb > 0.2 between version of genomic DNA with the MethylEasy Xceed 8 week and 12 week placenta. The 12 probes were associated with 11 bisulphite conversion kit (Human Genetic Signatures, genes, with two probes associated with the BTG4 gene. White corresponds to low methylation, and black to high methylation. North Ryde, Australia). Amplification conditions were Additional file 6: Proportion of probes within a particular 40 cycles: 95°C for 5 min, 56°C for 1 min 30 sec and 72° methylation level. Pie Charts showing the proportion of probes within a C for 1 min 30 sec, then 72°C for 7 min. particular methylation level for first, second and third trimester. The Novakovic et al. BMC Genomics 2011, 12:529 Page 12 of 14 http://www.biomedcentral.com/1471-2164/12/529 percentage of probes with a ‘b < 0.02’ is the same in all three HumanMethylation BeadChip probes were quartiled into 4 groups (0- gestational ages (63%), suggesting that probes with low methylation in 25%, 25-50%, 50-75%, 75-100%) based on methylation level, with the first trimester remain low over placental development. Furthemore our same number of probes in each quartile. The quartiles for each data suggests that probes with an intermediate methylation in first gestational age were plotted on the x-axis with the corresponding gene trimester are the ones that increase over gestation, with a lower expression values obtained from publically available first, second and proportion of probes in the ‘b = 0.2 - 0.6 group’ in second (22%) and third trimester data (y-axis). This analysis shows a decreasing median st third (20%) compared to 1 trimester (24%); and a higher proportion of gene expression level with increasing DNA methylation, highlighting the nd probes in the b > 0.6 group in 2 (15%) and term (17%) placenta functional relevance of DNA methylation in placenta at all three st compared to 1 trimester (13%). gestational ages. Additional file 7: Average methylation level of probes that increase Additional file 14: Correlation between methylation and expression in methylation over gestation. Box plot showing average methylation change between first and third trimester with more genes of (A) all probes (n = 26, 162) in first, second and third trimester highlighted. Methylation difference (Δb) between first and third placenta, and (B) probes that shown an increase in methylation from first trimester (x-axis) was plotted against gene expression log fold change (y- to third trimester of b > 0.2 (n = 883). This analysis shows that probes axis) between first and third trimester. A positive change in log fold with intermediate levels of methylation in first trimester are the ones expression indicates higher expression in first trimester, while a positive that increase over time. On the other hand, probes with low methylation change in methylation indicates higher expression in third trimester. (b < 0.2) in the first trimester do not appear to increase in methylation in Highlighted genes are those that show a correlation between the third trimester placenta. methylation and expression level. Grey dots represent Infinium probes, black dots represent most differentially methylated and expressed genes. Additional file 8: Correlation between methylation levels in purified first trimester cytotrophoblasts and first and third trimester placenta. Scatter plot showing correlation (r ) between first and third trimester placenta and purified first trimester cytotrophoblasts methylation, based on Infinium HumanMethylation27 BeadChip and Acknowledgements Sequenom EpiTYPER analysis. The correlations between first trimester We would like to thank Dr. Dan Diego-Alvarez and Ruby Jiang for their placenta and cytotrophoblasts were (A) 0.96 and (C) 0.93, and between technical assistance and Kristal Louie for placenta donor recruitment in third trimester placenta and cytotrophoblasts were (B) 0.88 and (D) 0.88, Vancouver. using Infinium and Sequenom platforms, respectively. This finding Funding suggests that both first and third trimester placenta methylation levels BN is supported by an NHMRC (Australia) Dora Lush Biomedial Postgraduate are indicative of cytotrophoblast levels. The lower correlation in third Scholarship. RKCY is supported by a graduate student scholarship from the trimester is likely due to both lower numbers of villous cytotrophoblasts Child & Family Research Institute. The Vancouver data was funded by a and their differentiation into the syncytiotrophoblast layer. Canadian Institutes for Health Research Grant (to WPR). The Murdoch Childrens Research Institute is supported by the Victorian Government’s Additional file 9: Number of probes showing variation at each Operational Infrastructure Support Program. gestational age. Additional file 10: Relationship between probe methylation level Author details and variation. Relationship between methylation level and inter- 1 Cancer, Disease and Developmental Epigenetics, Murdoch Childrens individual variation at each gestational age (s ). Probes were separated Research Institute, Royal Children’s Hospital and Department of Paediatrics, into three groups: (A) low methylation (b < 0.2), (B) intermediate 2 University of Melbourne, Parkville, Victoria 3052, Australia. Early Life methylation (0.2 < b > 0.6) and (C) high methylation (b > 0.6). The Epigenetics, Murdoch Childrens Research Institute, Royal Children’s Hospital number of probes was plotted on the y-axis (in log scale) and the and Department of Paediatrics, University of Melbourne, Parkville, Victoria variance (s ) on the x-axis. Probes with an intermediate methylation level 3 3052, Australia. Department of Medical Genetics, University of British were most likely to show inter-individual variation (B), while probes with Columbia, Child & Family Research Institute, 950 West 28th Ave., Vancouver, a high methylation level were least likely to show inter-individual 4 BC, Canada. Bioinformatics Unit, Murdoch Children’s Research Institute, variation. In fact, most of the probes with a variance of > 0.02 were from Royal Children’s Hospital, Flemington Road, Parkville, Victoria 3052, Australia. the intermediate methylation level (88/106 first trimester, 119/166 second 5 Department of Pathology, University of Cambridge, Cambridge, CB2 1QP, trimester and 255/352 term), even though most of the Infinium probes UK. (63%) are actually in the low methylation group. Additional file 11: Number of probes showing variation at each Authors’ contributions gestational age. Venn diagram of variable probes (s > 0.02) in each BN analysed the data, participated in critical discussion, and wrote the draft gestational age. This analysis revealed that the vast majority of variable manuscript. LG processed the array data and performed data analysis. RKCY, probes are only variable in third trimester, while 52 were only variable in JMC, RS and WR designed the study, participated in critical discussion and second, and 21 only in first trimester. A total of 47 probes were variable wrote the manuscript. MSP set-up experiments and processed array data. AS across all gestational ages. and AM provided first trimester villi samples. All authors approved of the final manuscript. Additional file 12: Relationship between DNA methylation and genomic context. Probes were separated into two groups based on Received: 25 May 2011 Accepted: 28 October 2011 their genomic location - CpG Island (CGI) or non-CpG Island (non-CGI). Published: 28 October 2011 The expected frequency was based on the proportion of all analysable Infinium probes (A) within a CGI or outside a CGI (0.76 and 0.24, respectively). Probes that increased in methylation over gestation were References predominantly in non-CGI regions (B), while probes that decreased in 1. Huppertz B: The anatomy of the normal placenta. Journal of clinical methylation over gestation showed the expected proportions. pathology 2008, 61:1296-1302. Furthermore, probes that showed inter-individual variation in (D) both 2. von Rango U: Fetal tolerance in human pregnancy–a crucial balance first and third trimesters, (E) first trimester only, and (F) third trimester between acceptance and limitation of trophoblast invasion. Immunology only, were predominantly associated with CGIs. This finding further letters 2008, 115:21-32. suggests that probes that change over gestation are not the same as 3. Huppertz B: Placental origins of preeclampsia: challenging the current those that show inter-individual variation within each gestation. hypothesis. Hypertension 2008, 51:970-975. 4. Jauniaux E, Poston L, Burton GJ: Placental-related diseases of pregnancy: Additional file 13: Relationship between DNA methylation and gene Involvement of oxidative stress and implications in human evolution. expression levels. Relationship between DNA methylation and gene Hum Reprod Update 2006, 12:747-755. expression levels in first, second and third trimester placenta. Infinium Novakovic et al. BMC Genomics 2011, 12:529 Page 13 of 14 http://www.biomedcentral.com/1471-2164/12/529 5. Winn VD, Haimov-Kochman R, Paquet AC, Yang YJ, Madhusudhan MS, hypomethylation of multiple genes in early-onset preeclampsia. Eur J Gormley M, Feng KT, Bernlohr DA, McDonagh S, Pereira L, et al: Gene Hum Genet . expression profiling of the human maternal-fetal interface reveals 27. Loke YW, Gardner L, Grabowska A: Isolation of human extravillous dramatic changes between midgestation and term. Endocrinology 2007, trophoblast cells by attachment to laminin-coated magnetic beads. 148:1059-1079. Placenta 1989, 10:407-415. 6. Mikheev AM, Nabekura T, Kaddoumi A, Bammler TK, Govindarajan R, 28. Robinson CJ, Alanis MC, Wagner CL, Hollis BW, Johnson DD: Plasma 25- Hebert MF, Unadkat JD: Profiling gene expression in human placentae of hydroxyvitamin D levels in early-onset severe preeclampsia. American different gestational ages: an OPRU Network and UW SCOR Study. journal of obstetrics and gynecology 2010, 203:366 e361-366. Reproductive sciences (Thousand Oaks, Calif 2008, 15:866-877. 29. Kerkel K, Spadola A, Yuan E, Kosek J, Jiang L, Hod E, Li K, Murty VV, 7. Frost JM, Moore GE: The importance of imprinting in the human Schupf N, Vilain E, et al: Genomic surveys by methylation-sensitive SNP placenta. PLoS genetics 2010, 6:e1001015. analysis identify sequence-dependent allele-specific DNA methylation. 8. Coan PM, Burton GJ, Ferguson-Smith AC: Imprinted genes in the Nature genetics 2008, 40:904-908. placenta–a review. Placenta 2005, 26(Suppl A):S10-20. 30. Bibikova M, Le J, Barnes B, Saedinia-Melnyk S, Zhou L, Shen R, 9. Maccani MA, Marsit CJ: Epigenetics in the placenta. Am J Reprod Immunol Gunderson KL: Genome-wide DNA methylation profiling using Infinium 2009, 62:78-89. assay. Epigenomics 2009, 1:177-200. 10. Roberts CT: Review: Complicated interactions between genes and the 31. Benerischke K, Kaufmann P: Pathology of the Human Placenta Springer; environment in placentation, pregnancy outcome and long term health. 2006. Placenta 2010. 32. Grigoriu A, Ferreira JC, Choufani S, Baczyk D, Kingdom J, Weksberg R: Cell 11. Waterland RA, Jirtle RL: Transposable elements: targets for early specific patterns of methylation in the human placenta. Epigenetics 2011, nutritional effects on epigenetic gene regulation. Mol Cell Biol 2003, 6:368-379. 23:5293-5300. 33. Lash GE, Otun HA, Innes BA, Bulmer JN, Searle RF, Robson SC: Low oxygen 12. Wolff GL, Kodell RL, Moore SR, Cooney CA: Maternal epigenetics and concentrations inhibit trophoblast cell invasion from early gestation methyl supplements affect agouti gene expression in Avy/a mice. Faseb placental explants via alterations in levels of the urokinase plasminogen J 1998, 12:949-957. activator system. Biol Reprod 2006, 74:403-409. 13. Bruchova H, Vasikova A, Merkerova M, Milcova A, Topinka J, Balascak I, 34. Monk M, Boubelik M, Lehnert S: Temporal and regional changes in DNA Pastorkova A, Sram RJ, Brdicka R: Effect of maternal tobacco smoke methylation in the embryonic, extraembryonic and germ cell lineages exposure on the placental transcriptome. Placenta 2010, 31:186-191. during mouse embryo development. Development 1987, 99:371-382. 14. Zechner U, Pliushch G, Schneider E, El Hajj N, Tresch A, Shufaro Y, 35. Gama-Sosa MA, Midgett RM, Slagel VA, Githens S, Kuo KC, Gehrke CW, Seidmann L, Coerdt W, Muller AM, Haaf T: Quantitative methylation Ehrlich M: Tissue-specific differences in DNA methylation in various analysis of developmentally important genes in human pregnancy mammals. Biochim Biophys Acta 1983, 740:212-219. losses after ART and spontaneous conception. Mol Hum Reprod 2009. 36. Gama-Sosa MA, Wang RY, Kuo KC, Gehrke CW, Ehrlich M: The 5- 15. Zhang Y, Cui Y, Zhou Z, Sha J, Li Y, Liu J: Altered global gene expressions methylcytosine content of highly repeated sequences in human DNA. of human placentae subjected to assisted reproductive technology Nucleic Acids Res 1983, 11:3087-3095. treatments. Placenta 2010, 31:251-258. 37. Nakashima A, Shiozaki A, Myojo S, Ito M, Tatematsu M, Sakai M, Takamori Y, 16. Tsui DW, Chiu RW, Lo YD: Epigenetic approaches for the detection of Ogawa K, Nagata K, Saito S: Granulysin produced by uterine natural killer fetal DNA in maternal plasma. Chimerism 2010, 1:30-35. cells induces apoptosis of extravillous trophoblasts in spontaneous 17. Chim SS, Jin S, Lee TY, Lun FM, Lee WS, Chan LY, Jin Y, Yang N, Tong YK, abortion. Am J Pathol 2008, 173:653-664. Leung TY, et al: Systematic search for placental DNA-methylation 38. Symonds ME, Sebert SP, Hyatt MA, Budge H: Nutritional programming of markers on chromosome 21: toward a maternal plasma-based the metabolic syndrome. Nat Rev Endocrinol 2009, 5:604-610. epigenetic test for fetal trisomy 21. Clinical chemistry 2008, 54:500-511. 39. Koldovsky O: Development of sucrase activity: effect of maternal 18. Chu T, Burke B, Bunce K, Surti U, Allen Hogge W, Peters DG: A microarray- hormonal status and fetal programming of jejuno-ileal differences. Ciba based approach for the identification of epigenetic biomarkers for the Found Symp 1979, 147-168. noninvasive diagnosis of fetal disease. Prenatal diagnosis 2009. 40. Barker DJ, Osmond C, Simmonds SJ, Wield GA: The relation of small head 19. Papageorgiou EA, Fiegler H, Rakyan V, Beck S, Hulten M, Lamnissou K, circumference and thinness at birth to death from cardiovascular Carter NP, Patsalis PC: Sites of differential DNA methylation between disease in adult life. BMJ 1993, 306:422-426. placenta and peripheral blood: molecular markers for noninvasive 41. Barker DJ: The fetal and infant origins of adult disease. Bmj 1990, prenatal diagnosis of aneuploidies. Am J Pathol 2009, 174:1609-1618. 301:1111. 20. Papageorgiou EA, Karagrigoriou A, Tsaliki E, Velissariou V, Carter NP, 42. Osmond C, Barker DJ: Ischaemic heart disease in England and Wales Patsalis PC: Fetal-specific DNA methylation ratio permits noninvasive around the year 2000. J Epidemiol Community Health 1991, 45:71-72. prenatal diagnosis of trisomy 21. Nat Med 17:510-513. 43. Novakovic B, Sibson M, Ng HK, Manuelpillai U, Rakyan V, Down T, Beck S, 21. Yuen RK, Avila L, Penaherrera MS, von Dadelszen P, Lefebvre L, Kobor MS, Fournier T, Evain-Brion D, Dimitriadis E, et al: Placenta-specific methylation Robinson WP: Human placental-specific epipolymorphism and its of the vitamin D 24-hydroxylase gene: implications for feedback association with adverse pregnancy outcomes. PloS one 2009, 4:e7389. autoregulation of active vitamin D levels at the fetomaternal interface. J 22. Turan N, Katari S, Gerson LF, Chalian R, Foster MW, Gaughan JP, Coutifaris C, Biol Chem 2009, 284:14838-14848. Sapienza C: Inter- and Intra-Individual Variation in Allele-Specific DNA 44. Du P, Kibbe WA, Lin SM: lumi: a pipeline for processing Illumina Methylation and Gene Expression in Children Conceived using Assisted microarray. Bioinformatics 2008, 24:1547-1548. Reproductive Technology. PLoS genetics 2010, 6:e1001033. 45. Du P, Zhang X, Huang CC, Jafari N, Kibbe WA, Hou L, Lin SM: Comparison 23. Yuen RK, Penaherrera MS, von Dadelszen P, McFadden DE, Robinson WP: of Beta-value and M-value methods for quantifying methylation levels DNA methylation profiling of human placentas reveals promoter by microarray analysis. BMC bioinformatics 2010, 11:587. hypomethylation of multiple genes in early-onset preeclampsia. Eur J 46. Wong NC, Novakovic B, Weinrich B, Dewi C, Andronikos R, Sibson M, Hum Genet 2010, 18:1006-1012. Macrae F, Morley R, Pertile MD, Craig JM, Saffery R: Methylation of the 24. Chavan-Gautam P, Sundrani D, Pisal H, Nimbargi V, Mehendale S, Joshi S: adenomatous polyposis coli (APC) gene in human placenta and Gestation-dependent changes in human placental global DNA hypermethylation in choriocarcinoma cells. Cancer letters 2008, 268:56-62. methylation levels. Mol Reprod Dev 2011, 78:150. 47. James JL, Stone PR, Chamley LW: The isolation and characterization of a 25. Sood R, Zehnder JL, Druzin ML, Brown PO: Gene expression patterns in population of extravillous trophoblast progenitors from first trimester human placenta. Proc Natl Acad Sci USA 2006, 103:5478-5483. human placenta. Hum Reprod 2007, 22:2111-2119. 26. Yuen RK, Penaherrera MS, von Dadelszen P, McFadden DE, Robinson WP: 48. Novakovic B, Gordon L, Wong NC, Moffett A, Manuelpillai U, Craig JM, DNA methylation profiling of human placentas reveals promoter Sharkey A, Saffery R: Wide ranging DNA methylation differences of Novakovic et al. BMC Genomics 2011, 12:529 Page 14 of 14 http://www.biomedcentral.com/1471-2164/12/529 primary trophoblast cell populations and derived-cell lines: implications and opportunities for understanding trophoblast function. Mol Hum Reprod 2011, 17:344-353. doi:10.1186/1471-2164-12-529 Cite this article as: Novakovic et al.: Evidence for widespread changes in promoter methylation profile in human placenta in response to increasing gestational age and environmental/stochastic factors. BMC Genomics 2011 12:529. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit

Journal

BMC GenomicsSpringer Journals

Published: Oct 28, 2011

There are no references for this article.