WNT2 promoter methylation in human placenta is associated with low

RESEARCH PAPER

Epigenetics 6:4; 440-449; April 2011; © 2011 Landes Bioscience

WNT2 promoter methylation in human placenta is associated with low birthweight percentile in the neonate Jose C. Ferreira,1,2 Sanaa Choufani,2 Daria Grafodatskaya,2 Darci T. Butcher,2 Chunhua Zhao,2 David Chitayat,4,5 Cheryl Shuman,4 John Kingdom,5 Sarah Keating3 and Rosanna Weksberg1,2,4,* Institute of Medical Sciences; University of Toronto; 2Genetics and Genome Biology Program; Hospital for Sick Children Research Institute; 3Department of Pathology; Mount Sinai Hospital; 4Division of Clinical and Metabolic Genetics; Hospital for Sick Children; 5Maternal-Fetal Medicine Division; Department of Obstetrics and Gynecology; Mount Sinai Hospital; Toronto, ON Canada

1

Key words: fetal growth, DNA methylation, placenta, WNT2, epigenetics Abbreviations: ACOG, American College of Obstetrics and Gynecology; BW%, birthweight percentile; CGI, CpG island; DNA, deoxyribonucleic acid; GA, gestational age; MeDIP, methyl DNA immunoprecipitation; MI, methylation index; qRT-PCR, quantitative reverse transcription-polymerase chain reaction; RNA, ribonucleic acid; RSS, Russell-Silver syndrome; SGA, small for gestational age; WNT2, wingless-type MMTV integration site family member 2; WNT2PrMe, WNT2 promoter methylation

Neonates with birthweights below the tenth percentile for gestational age are considered small for gestational age (SGA). Such infants have an increased risk for perinatal mortality and morbidity as well as an increased lifetime risk for adult onset disorders. Low birth weight percentile is etiologically heterogeneous and may result from maternal, fetal, placental and environmental factors. However, the molecular determinants of human SGA are not well elucidated. We proposed that fetal growth potential could be negatively impacted by the epigenetic dysregulation of specific genes in the placenta. Using methyl DNA immunoprecipitation coupled with Agilent CpG island microarrays, we analyzed the differences in DNA methylation between placentas of eight SGA neonates and eight controls with birthweight percentiles above the tenth percentile. We identified several candidate genomic regions with differential DNA methylation between the two groups. The DNA methylation differences identified in the promoter of the WNT2 gene were prioritized for further study in an extended cohort of 170 samples given the important function of this gene in mouse placental development and its high expression in human placenta. High WNT2 promoter methylation (WNT2PrMe) was found only in placental tissue and not in the cord blood of the fetus. It was significantly associated with reduced WNT2 expression in placenta and with low birthweight percentile in the neonate. Our results show that WNT2 expression can be epigenetically downregulated in the placenta by DNA methylation of its promoter and that high WNT2PrMe is an epigenetic variant that is associated with reduced fetal growth potential. Note: All of the array data in the manuscript can be accessed from the Gene Expression Omnibus (GEO) NCBI database under GEO accession number GSE22326.

©201 1L andesBi os c i enc e. Donotdi s t r i but e.

Introduction Neonates with birthweights below the tenth percentile for gestational age are considered to be SGA. Such infants have an increased risk for perinatal morbidity and mortality including growth, neurodevelopment and compromised cognitive abilities.1-6 Low birthweight percentile also has a long-term impact on adult health.7-16 Notably, such individuals have an increased risk of coronary artery disease,17 osteoporosis and osteoarthritis,18 hypertension, type 2 diabetes and obesity.19,20 These outcomes associated with low birth weight percentile make it critically important to understand the molecular mechanisms involved in loss of fetal growth potential. Although low birth weight percentile is known

to be etiologically heterogeneous, the molecular determinants have not been well elucidated. Studies in transgenic mice have identified a number of genes for which targeted disruption leads to placental dysfunction and also to poor fetal development and/or perinatal death.21 In humans, a number of genome-wide analyses have identified gene expression changes associated with abnormal placental development and/or reduced fetal growth potential.22-30 However, the causes for these expression differences in the placenta have not been investigated. We propose that some of these expression differences result from genetic and/or environmental factors31,32 that are modulated by epigenetic regulation to impact fetal growth potential. Specifically, we hypothesize that epigenetic

*Correspondence to: Rosanna Weksberg; Email: [email protected] Submitted: 11/17/10; Accepted: 12/21/10 DOI: 10.4161/epi.6.4.14554

440

Epigenetics

Volume 6 Issue 4

RESEARCH PAPER

RESEARCH PAPER

dysregulation of specific genes in the placenta could be a common mechanism determining fetal growth potential. Epigenetic marks modify gene expression via chemical modifications of the DNA and its associated proteins without a change in the primary DNA sequence. These modifications include DNA methylation, histone modifications, chromatin conformation and microRNAs.33 Epigenetic marks participate in normal developmental programming in a tissue- and developmentaltime specific manner. Studying the epigenetic variation in the development of the placenta may provide important insights into the molecular basis of fetal growth dysregulation. Of all the known epigenetic marks DNA methylation is the most well studied. In vertebrates, methylation commonly occurs in cytosines in CpG dinucleotides.34 CpGs are non-randomly distributed throughout the human genome. In 98% of the genome CpG dinucleotides are rare, in the remaining 2% of the genome they are found in selectively conserved clusters called CpG islands (CGI). CGIs, range from 200 bp to several kb in length, and often occur at the 5' ends of genes (promoters) where they are involved in transcriptional regulation. The majority of CpG dinucleotides located in promoters, whether part of CGI or not, are unmethylated.35 Methylation of promoter CpG sites is usually associated with gene silencing, i.e., a decrease in the expression of the gene.36 Epigenetic abnormalities have been associated with reduced fetal growth in mouse models, and also in rare human disorders.37-40 Recently we and others have demonstrated epigenetic alterations in imprinted genes that are associated with syndromes such as Russell-Silver syndrome (RSS), of which one component is low birthweight percentile.39-41 Such epigenetic abnormalities have also been identified at low frequency (less than 1%) in individuals presenting with SGA only at birth.42 Abnormal expression of imprinted genes and dysregulation of the epigenetic marks underlying their imprinting status, have been demonstrated to be associated with phenotypes that often include poor fetal growth, as reviewed in references 43 and 44. To date, common epigenetic variations have not been identified in association with low birthweight percentile, or SGA, in humans or mice. In this study, we undertook a search for epigenetic alterations that negatively impact fetal growth. Using an antibody enrichment microarray assay, we screened for differences in DNA methylation between placentas of eight neonates with birthweights at or below the tenth percentile and eight gestational age-matched controls with birthweight percentiles above the tenth. We identified 21 candidate differentially methylated regions between our cases and controls. Based on the methylation data and a review of targeted disruption of genes in these genomic regions in mouse models, the candidate genomic region overlapping the promoter of WNT2 (wingless-type MMTV integration site family member 2) was prioritized for further study. In an expanded cohort of 170 neonates, WNT2 promoter methylation and expression were assayed by targeted pyrosequencing of bisulfite-modified DNA and by qRT-PCR of cDNA, respectively. A significant association was found between high WNT2 promoter methylation (WNT2PrMe) and reduced WNT2 gene expression in the placenta. Moreover, a significant association was also demonstrated

between high WNT2PrMe and low birthweight percentile. In summary, our results show that WNT2 expression can be epigenetically downregulated in the placenta by DNA methylation of its promoter and, notably, high WNT2PrMe is a common epigenetic variant in placenta associated with an increased risk for low birthweight percentile in the neonate. Results Differential DNA methylation between SGA and controls. We generated DNA methylation profiles in human placenta from eight SGA neonates and eight gestational age-matched controls. As described by Weber and colleagues,45 DNA from these samples were fractionated and enriched by immunoprecipitation for the methylated fraction using an antibody against 5-methyl cytosine. The methylated fraction and the genomic DNA (input) was co-hybridized on the 244 K human Agilent CpG island array. Overall levels of genome-wide methylation were not significantly different between cases and controls. There was a high degree of correlation among all the placental samples 0.90–0.97, suggesting that the overall methylation patterns are consistent across all placentas. Euclidean cluster analysis also showed no clustering by cases versus controls (data not shown). Differential methylation analysis between SGA and controls identified 21 candidate differentially methylated regions. These candidate regions were distributed across the genome and associated with genes having diverse functions. Supplemental Table 1 summarizes the features of these candidate differentially methylated regions including gene function, size of the methylation difference (log2 ratio fold change), its position relative to the transcription start site of the closest gene, the number of probes (genomic length) involved in the difference and the expression level of the gene in human placenta. We prioritized the candidate genomic region associated with the WNT2 gene for further validation based on an assessment of the candidates by a series of selection criteria. The WNT2 associated region had a high average fold change (1.9x) across a large number of probes (12 probes), and the genomic position of the probes demonstrating methylation variation was near the transcription start site (promoter region) of the gene (Chr. 7:116,749, 73–116,751,336, NCBI36/hg18, Mar. 2006) (Fig. 1A). Three cases demonstrated distinctly higher values in comparison with all the other samples. Further, as annotated in expression databases, the highest level of WNT2 expression in humans is found in the placenta (biogps.gnf.org/#goto=welcome).46,47 Finally, and very importantly, WNT2 is known to be critical for vascular patterning of the mammalian placenta.48 Mice carrying an engineered mutation of the Wnt2 gene present with alterations in the size and structure of the placenta and perinatal death occurs in 50% of the offspring.48 Since none of the other 20 candidate regions we considered fulfilled our selection criteria to the same extent as WNT2, we deemed this gene to be the most attractive candidate for further validation. Validation of WNT2 promoter methylation in microarrays by pyrosequencing. In order to validate the differential methylation of the WNT2 promoter in controls vs. SGA cases we developed a pyrosequencing assay. This assay was designed to measure DNA


www.landesbioscience.com Epigenetics

441


Figure 1. DNA methylation at the WNT2 promoter in the placenta from CGI microarray and pyrosequencing. (A) Shows representative data for high methylation promoter (top) and low methylation (bottom) of the WNT2 promoter in two different SGA placental samples. Data is shown as log2 of the ratio methylated DNA/total DNA of each of the array probes covering the region. Each vertical bar corresponds to a probe. The length of the bar corresponds to the log2 ratios, whose values are in the Y axis. The data presented is a screenshot from UCSC genome browser. The horizontal bars at the bottom of the figure shows, from top to bottom the region targeted by the pyrosequencing assay, the location of the first exon and intron of the WNT2 gene as well as the CpG island targeted by this group of array probes. (B) Pyrosequencing data of the samples shown in (A). The shadowed area corresponds to a CG dinucleotide. The ratio between C and the C + T give the absolute methylation level for each cytosine in a CpG.

methylation levels of the promoter region of WNT2 in a genomic region overlapping the candidate CpG island identified by microarray. Bisulfite converted genomic DNA obtained from placental biopsies of the same SGA and control samples were tested for methylation changes at five CpG sites partially overlapping one of the array probes (Chr. 7: 116,750,697–116,750,747, NCBI36/ hg18, Mar. 2006) found to be differentially methylated between cases and controls. The data from the pyrosequencing assay (Fig. 1B) validated the methylation values from the arrays with a Pearson correlation (r) value of 0.95 and essentially discriminated the same set of three SGA samples with increased methylation at the promoter of WNT2 as the arrays. WNT2 promoter methylation and WNT2 expression in the placenta. We predicted that an increase in methylation of the promoter of the WNT2 gene would be associated with reduced WNT2 expression and possibly with reduced fetal growth. Thus, to better understand the biological role of the variation of WNT2 promoter methylation (WNT2PrMe) in placenta, we explored several aspects of this finding including its distribution, effect on expression, placental distribution and organ specificity, as well as its clinical significance, that is, association with birthweight percentile. We tested 170 samples for WNT2PrMe and WNT2 expression. Table 1 provides details of our sample cohort. Birthweight percentile was correlated with the presence of placental lesions but not with sex, gestational age, ethnicity and preeclampsia.

442

To assess the distribution of WNT2PrMe in the larger cohort of placental samples, we performed bisulfite pyrosequencing on 170 placental samples and observed a bimodal distribution of DNA methylation (Fig. 2). Thirty-five samples (21%) were found to have WNT2PrMe values in a distinct range of the distribution of values across all samples (i.e., methylation index (MI) 40–58%). Samples with this epigenetic variant were classified as high WNT2PrMe; the remaining 135, with MI = 3–24% were classified as low WNT2PrMe. We then tested whether high WNT2PrMe was associated with reduced expression of WNT2 using quantitative real time PCR. A statistically significant difference was found between the levels of WNT2 expression (here presented in a log transformed format, as explained in Methods) in the high (mean = -0.83, n = 35) versus low (mean = -0.51, n = 135) WNT2PrMe samples (two sided t-test of log transformed expression data, p < 0.0005), with an effect size, measured by Glass’s Δ, of 0.78 (Fig. 3A). Non-parametric test analysis of the non-logged transformed data using Mann-Whitney U test gave the same result (p < 0.0005). Notably, when WNT2PrMe is low, there is a wide range of expression, whereas with high WNT2PrMe the range of expression is much lower (Fig. 3B). WNT2 promoter methylation and expression in multiple samples from the same placenta. We also considered whether the variation in WNT2 promoter methylation among placentas

Epigenetics

Volume 6 Issue 4

Table 1. Description of the 170 samples Descriptive

Association with BW% (p value)

Median = 21 Mean = 29 (SD = 26.7)

BW% (n = 170)

Range = 93 (1 to 94)*

Not applicable

IQR = 6 to 51 Preeclampsia (PE) (n = 170)

Without PE = 119 (70%)

Placental lesions a ssociated with low birth weight (n = 159)**

Without lesions = 68 (42.8%)

With PE = 51 (30%)

With lesions = 91 (57.2%)

0.148MW