Exposure-Related Placental Methylation Changes Are Associated With Adverse Outcomes By Jennifer Zeynab Joukhadar Maccani B.S., University of Connecticut, 2008 A Dissertation Submitted in Partial Fulfillment of the Requirements for Degree of Doctor of Philosophy in the Division of Biology and Medicine at Brown University Providence, Rhode Island May 2014 © Copyright 2014 by Jennifer Zeynab Joukhadar Maccani! This dissertation by Jennifer Zeynab Joukhadar Maccani is accepted in its present form by the Department of Pathology and Laboratory Medicine as satisfying the dissertation requirement for the degree of Doctor of Philosophy. Date _________________ ______________________________ Karl Kelsey, MD, MOH, Advisor Recommended to the Graduate Council Date _________________ _______________________________ Carmen Marsit, PhD, Reader Date _________________ _______________________________ Murray Resnick, MD, PhD, Reader Date _________________ _______________________________ E. Andrés Houseman, ScD, Reader Approved by the Graduate Council Date _________________ _______________________________ Peter M. Weber, PhD, Dean of the Graduate School iii! Curriculum Vitae Jennifer Zeynab Joukhadar Maccani (born Jennifer Zeynab Joukhadar) was born in Lenox Hill Hospital in New York City, NY, USA on May 13, 1988. After graduating from Fairfield Ludlowe High School in Fairfield, CT, USA in June 2006, she attended Dickinson College in Carlisle, PA, USA until transferring to the University of Connecticut in Storrs, CT, USA in 2007, and graduated from the University of Connecticut’s College of Liberal Arts and Sciences with a Bachelor of Science in “Molecular and Cell Biology” in December 2008. While attending the University of Connecticut Jennifer performed research on the detection of serum biomarkers on Guthrie or blood spot cards using Grating-Coupled Surface Plasmon Resonance Imaging (GCSPRI). Jennifer matriculated into Brown University’s Pathobiology Graduate Program in the Department of Pathology and Laboratory Medicine in Providence, Rhode Island, USA in June 2009 with a summer rotation in Dr. Carmen Marsit’s laboratory. She performed her thesis research under the mentorship of Dr. Karl Kelsey, MD, MOH after discontinuing her research with Dr. Surendra Sharma and joining the Kelsey lab in September of 2011. During her time in the Kelsey lab Jennifer investigated placental DNA methylation biomarkers of in utero exposures such as maternal tobacco smoking and heavy metals and adverse reproductive and developmental outcomes including preterm birth and poor infant neurobehavioral profile. Jennifer passed her preliminary examination in March 2010 and became a doctoral candidate. Jennifer’s first-author manuscripts include “New frontiers in reproductive immunology research: bringing bedside problems to the bench,” published in Expert Rev Clin Immunol in September 2011. She has also designed and completed scientific illustration for Figure 1 of “The iv! epigenetics of maternal cigarette smoking during pregnancy and effects on child development,” a peer-reviewed manuscript by V.S. Knopik and M.A. Maccani, et al published in Dev Psychopathol in November 2012. During her doctoral training, Jennifer has presented her work at national and international scientific conferences, including the 28th Annual New England Membrane Enzyme Group (NUTMEG) Meeting at the Marine Biological Laboratory in Woods Hole, MA, USA in October 2010, the 8th Annual Congress of the European Society for Reproductive Immunology (ESRI) in Munich, Germany in November 2010, the 2nd Jena International Training in Reproductive Sciences and Technologies (InTReST)-Deutsche Gesellschaft für Reproduktionsmedizin (DGRM) in Jena, Germany in November 2010, the 2011 Annual Retreat for Brown University’s Superfund Program in Providence, Rhode Island, USA in January 2011, the 31st Annual Meeting of the American Society for Reproductive Immunology (ASRI) in Salt Lake City, UT, USA in May 2011, and the 60th Annual Meeting of the Society for Gynecologic Investigation (SGI) in Orlando, FL, USA in March 2013. While at Brown, Jennifer received several awards including a Pre-Doctoral Training Fellowship from the National Institute of Environmental Health Sciences (NIEHS) and a Sheridan Teaching Certificate I from Brown University. v! Preface and Acknowledgments I have executed the work presented in this PhD dissertation, including all experiments, analysis, and discussion, and have appropriately acknowledged important collaborations throughout this document. When I entered the Pathobiology Graduate Program four years ago with a summer rotation in Dr. Carmen Marsit’s lab, it was with a fervent desire to perform cutting-edge research in an area that had the potential to improve the quality of life of others. The opportunity to contribute to the scientific and non-scientific communities of which I am a part would not have been possible without the help of many people, and I would like to thank as many of them as sincerely and as deeply as I can. I am grateful to my thesis advisor and mentor, Dr. Karl Kelsey, who has supported my research as well as my development as a young scientist and encouraged me to strive for excellence and take pride in my work. The training I have gained under Karl’s mentorship has been of a truly exceptional caliber, and I will always be grateful that I was blessed with the opportunity to join the Kelsey lab, despite having to make the difficult decision to leave another lab halfway through my time in the program. Thank you to the principal investigator of that lab, Dr. Surendra Sharma; although I did not complete my thesis work in his lab, Dr. Sharma gave me many opportunities to present the research I did during that time, and I learned many valuable things from my experience in his lab. It has been an honor to work with not only Karl, but also Dr. Carmen Marsit, who gave me the opportunity to do a summer rotation in his lab in the summer of 2009 and first introduced me to the field of epigenetics. Carmen has been a vi! source of guidance throughout my graduate training, and has been a crucial member of my thesis committee over these past four years. I am thankful to have been able to work with someone with so much drive and passion for his work. My thesis committee, chaired by Dr. Murray Resnick and including Dr. E. Andrés Houseman, has also been an invaluable resource over the course of my thesis research, and I am grateful for their knowledgeable insights and constructive criticism that has not only improved my work but helped me to think more clearly and critically. I have had the good fortune to work with many other intelligent and driven scientists without whom much of this work would have been impossible, including Dr. Devin Koestler and the past and present members of the Kelsey and Marsit labs, who have helped teach me new techniques or skills or assisted me with protocols or code or sample processing; thank you for the help that you gave me along the way. When I was very small I started asking questions, and I never grew out of it. I loved building models of dinosaurs with my mother, Kristina Joukhadar, catching bugs and butterflies and fish and frogs with my sister, Dana, at my grandparents’ house in Wolfeboro, New Hampshire, collecting rocks and shells, and putting everything under the microscope that I could manage. I want to thank every one of my friends and family who nurtured my enthusiasm for the natural world and my desire to understand everything about it from the very beginning. Thank you to my mother for giving me an education that opened my eyes. Thank you to my father for taking a leap of faith and immigrating to the United States from Syria so that our family could have opportunities that he did not. I am truly blessed to have had the opportunities I have had, including being able to do undergraduate research in the lab of Dr. Michael Lynes at the University of Connecticut, vii! whom I also wish to thank. Thank you, too, to my high school AP Biology teacher, Dr. Andrew Bednarik, who first helped me believe that I could get my doctorate degree, who always went the extra mile to listen to and support his students. Dr. B., if I become a teacher, I hope I can be a teacher like you. Thank you to all my friends around the country and the world, including Whitney Cane, Kristin Krogh, Dr. Kiran Parkhe, Vedant Mehra, Iden Sinai, Dr. Andrew Robson, and Aileen Diaz. Thank you to my family in the United States as well as Syria and beyond. Thank you especially to the family that I have been blessed with over the course of my graduate career—my mother and fathers in law, my grandparents, my aunts and uncles, my cousins both in the United States and Germany. Thank you to the cat that I adopted shortly before I began this journey, Edward Maccani, for all the cuddles. Most of all I want to thank my husband, Dr. Matthew Alan Maccani, whose friendship and love are precious gifts for which I am thankful each and every day. Thank you for being a pillar of patient support and a partner for the journey; thank you for showing me what it is to be a man of both faith and reason; thank you for believing in me when I doubted myself. The journey of graduate school has been at times both intensely challenging and, with the support of all those who were with me along the way, intensely rewarding. As I move on to the next step in life and in my career, I’m thrilled that the journey continues. viii! Table of Contents Page Abstract…………………………………………………………………………………..1 Chapter 1 – Introduction………………………………………………………………..3 Thesis Overview…………………………………………………………………………..4 Fetal origins of adult disease………………………………………………………………5 Placenta……………………………………………………………………………………6 Epigenetics………………………………………………………………………………...7 Imprinting………………………………………………………………………………....7 Non-coding RNAs………………………………………………………………………...9 Histone modifications……………………………………………………………………10 DNA methylation………………………………………………………………………...11 Placental DNA methylation as a biomarker of in utero exposures………………………12 References……………………………………………………………………………......17 Chapter 2 – Smoking-Related Changes to Placental DNA Methylation at the RUNX3 Gene Are Associated With Gestational Age……………………………..…..37 Abstract…………………………………………………………………………………..40 Introduction………………………………………………………………………………41 ix! Methods……………………………………………………………………………........42 Results…………………………………………………………………………………..47 Discussion……………………………………………………………………………….49 Acknowledgments……………………………………………………………………….53 Funding………………………………………………………………………………….54 References…………………………………………………………………………….....55 Chapter 3 – Placental DNA Methylation Changes Within the EMID2 Gene Are Associated With Both Infant Toenail Mercury and Adverse Neurobehavioral Outcomes……………………………………………………………………………......84 Abstract……………………………………………………………………………..........87 Introduction……………………………………………………………………………....88 Methods……………………………………………………………………………..........90 Results……………………………………………………………………………............92 Discussion……………………………………………………………………………......95 Acknowledgments and Funding…………………………………………………………99 References……………………………………………………………………………....100 Chapter 4 – DNA Methylation Changes in the Placenta Associated With Manganese In Utero…………………………………………………………………………….......139 x! Abstract……………………………………………………………………………........142 Introduction……………………………………………………………………………..143 Methods……………………………………………………………………………........144 Results……………………………………………………………………………..........146 Discussion……………………………………………………………………………....147 Acknowledgments and Funding………………………………………………………..150 References……………………………………………………………………….……...151 Chapter 5 – Discussion………………………………………………………………..167 References……………………………………………………………………………....181 xi! List of Tables Pages Chapter 2 Table 1. Study population demographics. ………………………………………………66 Table 2. Two RUNX3 loci significantly associated with maternal smoking during pregnancy were significantly or near-significantly associated with gestational age (<37 weeks vs. ≥ 37 weeks). ………………………………………………………………….68 Table 3. Gestational age <37 weeks was significantly associated with logit-transformed, ComBat-adjusted cg04757093 methylation status (p=0.04) while controlling for potential confounders. ……………………………………………………………………………..69 Table 4. First three PCs are significantly associated with BeadChip (batch effect)…….70 Table 5. First three PCs are no longer significantly associated with BeadChip after ComBat adjustment. ……………………………………………………………………..71 Table 6. Top 50 CpG loci by p value resulting from locus-by-locus analysis for maternal smoking during pregnancy. …………………………………………………………..….72 Chapter 3 Table 1. Study population demographics. ……………………………………………..113 Table 2. Mean ß values and q values of 339 CpG loci associated with infant toenail Hg. ……………………………………………………………………………......................115 xii! Table 3. Mean and standard deviation of 13 NNNS scores in infants with non-high-risk vs high-risk NNNS profiles. …………………………………………………………..134 Table 4. 6 CpG loci associated with infant toenail Hg and high-risk NNNS profile….135 Chapter 4 Table 1. Study population demographics. ……………………………………………..157 Table 2. Mean ß values and q values of 110 infant toenail Mn-associated CpG loci………………………………………………………………………………………159 xiii! List of Figures Pages Chapter 2 Figure 1. Manhattan plot of –log p values for locus-by-locus analysis of methylation and maternal smoking during pregnancy. Logit-transformed, ComBat-adjusted methylation status of 1,918 CpG loci was significantly (p<0.05) associated with maternal smoking during pregnancy, indicated by points above blue line. …………………………………………………………………..75 Figure 2. Smoothed plot across seven RUNX3 CpG loci significantly associated with maternal smoking during pregnancy. Y-axis represents raw beta value. Blue line represents placental methylation in non-smoking mothers; red line represents that of smoking mothers. CpG loci in order of appearance on - strand from 5’ to 3’: 1. cg06037693 2. cg14182690 3. cg24019564 4. cg00117172 5. cg08705994 6. cg00572797 7. cg04757093* (also significantly associated with gestational age). Black arrow indicates transcription start site. Blue P1 and P2 arrows indicate distal and proximal RUNX3 promoters, respectively. Yellow boxes indicate exons. ……………………………………………………………………76 Figure 3. Smoking mothers exhibit significant placental hypermethylation at the cg04757093 locus within the RUNX3 gene compared to non-smoking mothers…………………………………………………………………………………..77 Figure 4. Pyrosequencing confirmed significant placental hypermethylation xiv! at RUNX3 locus cg04757093 (p<0.02) and near-significant placental hypermethylation at RUNX3 locus cg00117172 (p<0.09) in smoking mothers compared to non-smoking mothers. No significant differential methylation was observed at RUNX3 locus cg06037693 between smoking and non-smoking mothers in these 22 placental samples. Mean placental methylation for non-smoking vs smoking mothers: cg06037693, 98 and 97.6, respectively (ranges of 96.3 – 99.5 and 94.3 – 99.7); cg00117172, 12.8 and 21.1, respectively (ranges of 4.5 – 28.3 and 7.7 – 41.3); cg04757093, 64.5 and 69.1, respectively (ranges of 61 – 68 and 60.7 – 79.3). *p<0.05; † p<0.10……………...…...……………………………………..78 Figure 5. Proportion of variance (y-axis) in each PC (x-axis) prior to ComBat adjustment. ..…………………………………………………………...……….79 Figure 6. The first ten PCs (x-axis) account for ~60% of the variance of the data. Y-axis represents cumulative proportion of variance……………………………...80 Figure 7. Plot of the first two PCs, which are significantly associated with BeadChip (batch effect). ...………………………………………………………………81 Figure 8. Proportion of variance (y-axis) in each PC (x-axis) after ComBat adjustment. ………………………………………………………………………………81 Figure 9. The first ten PCs (x-axis) account for ~50% of the variance of the data after ComBat adjustment. Y-axis represents cumulative proportion of variance……………..……………………………………………………………………82 Figure 10. Plot of the first two PCs, which are no longer significantly associated with BeadChip after ComBat adjustment……………………………………83 xv! Chapter 3 Figure 1. Diagram of analysis strategy. 192 placental samples were arrayed on an Illumina Infinium HumanMethylation450 BeadArray. Array data then underwent quality assurance and quality control procedures as detailed in Methods section and sex-linked and SNP- associated CpG loci were removed. 41 placental samples with infant toenail Hg data were analyzed for associations between placental methylation and infant toenail Hg, followed by FDR correction (q<0.2). 339 CpG loci associated with infant toenail Hg were subsequently analyzed for associations with high-risk NNNS profile in an independent set of 151 placental samples with NNNS profile data, followed by FDR correction (q<0.2). …………………………………………………136 Figure 2. Annotated smoothed plot across five EMID2 CpG loci associated with both infant toenail Hg and high-risk neurobehavioral profile in 41 samples with toenail Hg data by toenail Hg tertile. The Y-axis represents methylation β value. Blue line represents placental methylation in low (referent) toenail Hg tertile infants; yellow line represents that of medium toenail Hg tertile infants; red line represents that of high toenail Hg tertile infants. CpG loci in order of appearance on + strand from 5’ to 3’: 1. cg23424003 2. cg27179533 3. cg27528510 4. cg14874750 5. cg13267931. Yellow dot indicates transcription start site. Yellow arrow indicates first exon. Green line indicates location of CpG island. …………….137 Figure 3. Average ß-value across 5 EMID2 CpG loci significantly xvi! associated with both infant toenail Hg and high-risk NNNS profile by non-high-risk vs high-risk profile, in a subset of 151 infants. ……………………..138 Chapter 4 Figure 1. Infants with high toenail Mn have significant placental hypermethylation as compared to infants with medium (referent) toenail Mn at two CpG loci associated with infant toenail Mn in the FILIP1L gene. P values are as follows: cg15963552, p=0.02; cg22092811, p=0.01. …………………………………………………………………..166 xvii! Abstract of “Exposure-Related Placental Methylation Changes Are Associated With Adverse Outcomes” by Jennifer Zeynab Joukhadar Maccani, Ph.D., Brown University, May 2014 Poor-quality in utero environments are associated with adverse pregnancy outcomes and susceptibility to early and later life disease. The overarching hypothesis of this work is that methylation profiling in placental tissue will potentially yield biomarkers of in utero exposures and adverse pregnancy and developmental outcomes. The first aim investigated smoking-associated methylation profiles in the placenta using an Illumina Human Methylation271 BeadArray. In 206 placental samples, an association was observed between maternal smoking during pregnancy, placental hypermethylation of one RUNX3 CpG locus, and preterm birth (<37 weeks gestation2). The second and third aims investigated placental methylation profiles associated with heavy metal exposure. Specifically, the second aim used an Illumina HumanMethylation4503 BeadArray in (n=41) placenta samples and identified 339 CpG loci in the placenta that were significantly associated with infant toenail mercury (Hg), 6 of which were found to be associated in an independent sample set (n=151) with a high- risk NNNS RPMM profile. 110 infant toenail manganese (Mn)-associated CpG loci were also identified in the placenta (n=61) using this array in the third aim, two of which were hypermethylated in the FILIP1L gene, where hypermethylation has been associated in the literature with prostate cancer, a condition also associated with high hair and nail Mn in adults. 1 Together, the three aims of this work better characterize the nature and effects of prenatal exposure to maternal tobacco smoking, Hg, and Mn, provide more complete knowledge of the potential mechanisms that may link prenatal exposures to postnatal outcomes, and contribute to a greater understanding of the developmental origins of health and disease. 2! ! ! CHAPTER 1 Thesis Overview and Introduction 3! ! ! Thesis Overview To date, a large body of literature has shown that a link exists between exposure to environmental toxicants in utero and adverse reproductive outcomes4, 5, infant and childhood disease and development6-9 and disease later in life 10, 11, an idea central to the Developmental Origins of Health and Disease (DOHaD) hypothesis11. Epigenetic mechanisms may play a role in mediating such responses, since epigenetic alterations in the placenta specifically have been associated with in utero exposure to toxicants12 such as cigarette smoke or its components13-15. Epigenetic alterations in the placenta have also been associated with reproductive outcomes16-22, infant or childhood development or disease23, 24, as well as, in some cases, both of these endpoints5, 25. Recent studies have associated methylation changes in the placenta with both in utero toxicant exposure as well as reproductive outcomes such as infant growth5. This thesis aims to identify altered placental methylation patterns associated with both in utero exposure to toxicants, for example maternal cigarette smoking or heavy metals, and reproductive outcomes including decreased gestational age and infant neurobehavioral outcomes. The introductory chapter of this thesis will describe and discuss previous studies in the literature that are relevant to this work, including the association of in utero exposures with later life disease and the importance of fetal programming, the role of the placenta in mediating in utero exposures, epigenetic alterations in the placenta that have been associated with in utero exposures as well as reproductive outcomes, infant and childhood development and disease, and previous studies that have described a role for methylation changes in the placenta in this context. Chapter 2 presents a study of 206 mother-infant pairs and the association between altered placental methylation within the 4! ! ! RUNX3 gene, observed in an Illumina Infinium HumanMethylation271 array platform, and decreased gestational age, specifically with preterm birth (<37 weeks gestation2). Chapter 3 describes the findings of a study of 41 mother-infant pairs, in which mercury (Hg) concentrations in infant toenail clippings were found to be associated with methylation changes at 339 CpG loci in placental samples analyzed on an Illumina Infinium HumanMethylation4503 array platform. 6 of these loci were also found to be associated with neurobehavioral outcomes, specifically with high-risk NNNS23, 24 RPMM class profile, in an independent set of 151 mother-infant pairs not previously analyzed for infant toenail heavy metal concentration. In Chapter 4, infant toenail clippings and placental samples from 61 mother-infant pairs were analyzed for associations between infant toenail manganese (Mn) and methylation changes in the placenta using the same array described in Chapter 3. Using similar methods as in Chapter 3, methylation changes at a set of CpG loci were found to be associated with infant toenail clipping Mn. Chapter 5, the final chapter of this thesis, summarizes the conclusions of Chapters 2-4 and discusses the implications that these studies and their findings may have for future research. Fetal origins of adult disease The Developmental Origins of Health and Disease (DOHaD) hypothesis 26, 27, which states that prenatal or early life exposures have the potential to modulate later life disease susceptibility, was first put forth after studies of a number of birth cohorts found associations between prenatal exposure to famine and adverse outcomes in infancy as well as later in life 10, 11, 28-30. One of these was the Dutch Famine Birth Cohort31, which 5! ! ! studied children born to mothers exposed to famine during the winter of 1944-45 in the Netherlands following a Nazi food blockade. These children were born with reduced birth length. The in utero ‘lean times’ were posited to give rise to an efficient metabolic phenotype that then made these individuals more susceptible to the chronic diseases in adulthood that are associated with ‘times of plenty’ 31. The birth cohort in the Dutch famine were reported to display susceptibility to cardiovascular disease32 and neurologic disorders33 such as schizophrenia34. The persistence of such effects into future generations has been observed with exposures other than famine9. Of particular interest are exposures to environmental toxicants capable of crossing the placenta, such as nicotine from first- or secondhand maternal tobacco smoking and certain heavy metals such as mercury (Hg)35-37 and manganese (Mn). Placenta Proper placentation and placental function are critical in the regulation of fetal growth. This is due to many factors including the importance of placental transport of adequate oxygen and nutrients to the fetus, the production of placental growth factors and hormones as well as placental detoxification activity18. During implantation, placental epithelial stem cells begin to differentiate into multinucleated syncytiotrophoblast cells, which primarily function in secretion and transport processes, or cytotrophoblast cells, which invade the uterine wall and remodel maternal uterine arterioles, eventually resulting in maternal blood flow to the placenta by the end of the first trimester of pregnancy38. 6! ! ! Early in development, fetal growth is outpaced by placental growth, and disturbances in the in utero environment due to toxicant-induced damage to the placenta or inadequate nutrition can affect fetal growth39, 40. These adverse effects may occur in part by exposure-associated changes to the epigenetic state of placental tissue5, 12-15, 41, which may interfere with normal placental function18. These altered placental epigenetic marks may also serve as a record of in utero exposures18. Epigenetics Epigenetics is defined as the study of heritable changes to gene expression without direct changes to DNA sequence42-44. A growing body of research has described a possible role for epigenetics as a mechanism linking environmental exposures, even grandmaternal exposures9, to regulation of gene expression. The four main modes of epigenetic regulation are imprinting, non-coding RNA-mediated regulation such as that by microRNAs, histone modifications, and DNA methylation. Imprinting Genomic imprinting is the expression of genes in a pattern determined by the parent of origin, such that only either the maternal or paternal allele of a gene is expressed 45-49, rather than a Mendelian pattern of expression such that normal alleles from both parents are equally expressed. The silencing of the imprinted allele can be due to deletion of the gene or due to other epigenetic factors such as DNA methylation, histone modification and chromatin conformation47. Genes that are normally imprinted often occur in close proximity to each other46, for example in clusters45 or pairs50, though 7! ! ! they can also occur alone51. Though imprinting of some genes is necessary for normal development, disorders resulting from improper genomic imprinting include the Prader- Willi and Angelman Syndromes, in which only one allele of the chromosome 15q11.2- w13 region is expressed47, 52-55. Lack of expression of the paternal alleles of the genes in this region, including MKRN3, MAGEL2, NECDIN, SNURF-SNRPN and a cluster of small nucleolar RNA genes (snoRNAs), leads to Prader-Willi Syndrome47, 52-55, which is a multisystem disorder characterized by severe hypotonia, delays in motor and language development, cognitive disability, excessive eating in late infancy and early childhood and, without proper dietary control, eventual morbid obesity47. Lack of expression of the maternal alleles of this region, specifically in the UBE3A and ATP10A genes, results in Angelman Syndrome (AS)47, 56, 57. More than 100 imprinted genes have been identified in the mouse placenta58-60, and roles have been hypothesized for some imprinted genes in human pathologies and placental function61, 62. Genome-wide expression analysis has attempted to define the placental ‘imprintome’ in the mouse63, 64, and it has been observed in the literature that the development of glycogen-containing trophoblast cells within the murine spongiotrophoblast layer in the placenta is dependent on the imprinted expression of two genes, IGF265 and CDKN1C66. The imprinting of these two genes is associated with Beckwith-Wiedemann Syndrome in humans, characterized by macroglossia, organomegaly, facial defects, and anomalies of the ear and kidneys67. Thus, imprinting represents a key form of epigenetic regulation in the placenta with implications for later life diseases. 8! ! ! Non-coding RNAs Non-coding RNAs (ncRNAs) have been shown to regulate gene expression throughout many aspects of differentiation and development68 in many important signaling pathways, including the Sonic hedgehog69 and Notch70 pathways. Small non- coding RNA-mediated regulation, such as that by microRNAs (miRNAs), is a form of epigenetic regulation that has been shown to be of particular importance in the placenta. The first-identified and possibly best understood miRNA, lin-4, was discovered in C. elegans by Victor Ambros and colleagues in 1993 and described as encoding a small RNA with imperfect base-pairing complementarity to target mRNAs that was capable of blocking their translation71. Research has since shown that miRNAs are typically 21-25 nucleotides in length and may regulate gene expression post-transcriptionally by binding to the 3’ UTR of their mRNA targets71, 72. Depending on the degree of complementarity, this binding may lead to either mRNA degradation or, in the case of partial complementarity, translational repression, such that a single miRNA can regulate multiple genes22, 72. MiRNAs are involved in regulating many normal cellular processes including division, migration, differentiation and death73, and are expressed in a tissue-specific manner74. Alterations to normal patterns of miRNA expression in the placenta have been associated in the literature with in utero toxicant exposures such as bisphenol A (BPA)12 and cigarette smoke14, as well as with adverse pregnancy outcomes such as pre- eclampsia75 and fetal growth restriction76, highlighting their importance in placental tissue and demonstrating the potentially crucial role that miRNAs may play in regulating 9! ! ! early and later life disease susceptibilities associated with adverse in utero exposures or environments. Histone modifications A third mechanism of epigenetic regulation is through the post-translational modification of amino acids within the histone proteins to which DNA is bound. Chromatin is made up of nucleosomes, in which 147 base pairs of DNA are wound around histone proteins77. These histone cores are octamers made up of two proteins each of H2A, H2B, H3, and H4 histone proteins77. Post-translational modifications to the amino acid tails or core regions of these proteins78, such as ubiquitination, phosphorylation, acetylation, and methylation79-83, can affect the density and dynamics of chromatin, and thereby alter the accessibility of DNA to transcription factors84. These changes, in turn, can affect gene expression77, whether by changing the electrostatic charge of histones and altering the binding of DNA to these proteins, or through a binding mechanism whereby bromodomain or chromodomain molecules may recognize and bind to acetylated or methylated lysines85. Histone modifications have been identified in the literature that are associated with various aspects of development as well as pathologic conditions such as cancer. Normal post-translational histone modification processes and molecules involved in these processes can be disrupted or deregulated by perturbations in the in utero environment, such as protein deprivation86 or pre-eclampsia87. Cancer-associated “onco- modifications”85 to histone proteins have also been identified83, 88, and deregulation of histone-modifying enzymes have also been described in cancer cells85. In addition, 10! ! ! grandmaternal smoking has been linked to asthma, possibly through a mechanism involving histone modifications89. Histone modifications thus represent an important aspect of epigenetic regulation that may be dysregulated not only during development, but in the context of later life diseases. DNA methylation A fourth and final form of epigenetic regulation exists by which the accessibility of DNA to transcription factors and other proteins may be modulated. DNA methylation involves the addition of methyl groups to the 5’ position of cytosines within the genome, and occurs almost exclusively in the context of CpG dinucleotides90. These methylated cytosine resides are more rare than might be expected, and constitute about 3-5% of the human genome91. Methylation of these cytosine resides, particularly within gene promoter regions92, can induce a transcriptionally inactive conformation in that genomic region, and can silence gene expression; however, recent literature has described controversial findings suggesting that DNA methylation within the gene body can enhance gene transcription93-97 DNA methylation patterns are important for normal growth and cellular differentiation, and are established de novo early in fetal development following a post-fertilization wave of genomic demethylation98, 99. Due to the stability of DNA methylation patterns and their functional relevance, they have been demonstrated in the literature to be useful biomarkers, for example in the context of prostate cancer, where diagnostic methylation profiles have been described100. Importantly, a new form of DNA methylation has recently been identified called hydroxymethylation, which involves the addition of a hydroxymethyl group to the 5’ 11! ! ! position of cytosine101. Like DNA methylation, DNA hydroxymethylation plays an important role in development, but DNA hydroxymethylation profiles associated with specific cell types and cellular processes are only just being identified. An association has been observed in the literature between DNA 5-hydroxymethylcytosine enrichment and demethylation and reactivation of genes, particularly in genomic regions that are involved in pluripotency101, as well as in exons and untranslated regions (UTRs)102. DNA hydroxymethylation may also be involved in enhancer activation103. Fetal-specific differentially hydroxymethylated regions (DhMRs) have also been described102; however, bisulfite modification cannot distinguish between DNA hydroxymethylation and DNA methylation 104, and is a limitation of some techniques commonly employed to measure DNA methylation such as pyrosequencing and bisulfite sequencing. A technique called “oxidative bisulfite” sequencing (oxBS-Seq) can, however, be used to detect hydroxymethylation marks and distinguish them from methylation marks on DNA. oxBS-Seq functions by oxidating 5-hmC to uracil using potassium perruthenate; this product is then bisulfite modified so that 5-fC is converted to uracil 104. Both DNA methylation and hydroxymethylation may play critical roles during development, particularly, in the case of DNA methylation, in the placenta, and with the discovery of hydroxymethylation and its distinct functions, care must be taken in future studies to distinguish DNA hydroxymethylation from DNA methylation. Placental DNA methylation as a biomarker of in utero exposures It has been proposed that epigenetic marks, such as DNA methylation patterns, in the placenta can serve as a record of the quality of the in utero environment18. Exposure 12! ! ! to environmental toxicants, such as tobacco smoke from maternal first- or secondhand smoking, can result in changes to normal placental epigenetic marks17, including DNA methylation patterns13, 15, 105. These changes can, in turn, affect normal placental functioning and development, thereby either directly or indirectly affecting fetal growth and development18. Nicotine from maternal tobacco smoke, for example, is known to cross the placenta14 and accumulate in fetal blood and amniotic fluid106, and fetal nicotine levels have been shown to be 15% higher than those in the mother107. Unfortunately, maternal tobacco smoking during pregnancy is a relatively common exposure, as 12-15% of mothers smoke during pregnancy 108, 109. Maternal tobacco smoking during pregnancy is associated with perinatal and later life morbidity and mortality, including premature birth 110, 111, low birth weight4, 112, adverse neurobehavioral outcomes113, childhood obesity114-116, sudden infant death syndrome117, asthma and airway hyperresponsiveness6, 7, 118-122 . As maternal tobacco smoking during pregnancy is also associated with altered placental DNA methylation patterns13, 15, 105, and placental DNA methylation patterns have been identified that are associated with gestational age19, specific patterns of DNA methylation in the placenta may exist that are associated with both maternal tobacco smoking and altered gestational age or preterm birth. In addition to maternal tobacco smoke, fetal exposures may include other environmental toxicants such as Hg and other heavy metals. Maternal dietary Hg intake from contaminated seafood may be one common route of prenatal exposure to Hg123. Atmospheric Hg on the surfaces of bodies of water is biomethylated by microorganisms and, in turn, biomagnified in fish123 such as salmon, tuna and swordfish124. Methyl- mercury (MeHg) from contaminated seafood can accumulate in the placenta in double the 13! ! ! concentration found in maternal blood125. Placental Hg accumulation can also occur as a result of maternal dental amalgam restorations, as these amalgams often contain inorganic Hg123, 126. Having a single amalgam restoration has been associated with an increase in placental Hg levels of 3-6 times126. Hg is known to cross the placenta and accumulate in fetal organs, such as the kidneys and brain126, and cord blood127. The fetal brain is a known target of prenatal Hg exposure and is particularly vulnerable to Hg exposure35, 123 as elemental Hg is converted into neurotoxic Hg2+ in neuronal tissue128. Adverse neurologic effects of prenatal exposure to Hg have been observed in infants and children for over four decades, beginning with studies of MeHg poisoning from industrial wastewater in Japan. These infants developed cerebral palsy, mental retardation, and seizures associated with their prenatal exposure to MeHg123, 129. A second study of Iraqi children exposed to Hg through bread made from seed grain contaminated by an Hg-containing fungicide130 supported the association of early-life Hg exposures with adverse neurologic effects despite the relatively low doses to which these children were exposed131. More recently, a study of children in the Faeroe Islands, born to mothers who ate seafood-heavy diets, found associations in these children at 7 years of age between Hg exposure and adverse neurodevelopmental outcomes including memory, attention, language, and visual spatial perception132. In 2000, the National Research Council 133 recommended that the United States Food and Drug Administration change its Hg reference dose based on findings from this cohort study35. That same year, a study of a second cohort of Faeroese children reported associations of prenatal Hg exposure with adverse neurologic outcomes134, and additional studies have observed similar associations 14! ! ! with size of cerebellum135, damage to the central nervous system136, behavioral137 and psychomotor138 deficits, delays in cognitive development in children139, and even later life effects such as Type II diabetes susceptibility140. Some of these studies and others have used fingernail or toenail Hg as surrogate measures of Hg levels in the body141-143. These measures are non-invasive and are becoming increasingly common in the literature. Other heavy metals in addition to Hg, such as Mn, are known to cross the placenta144. Mn is a micronutrient present in microgram (µg) levels in the human body145, and is obtained primarily from grains146. For adults, the Estimated Safe and Adequate Daily Dietary Intake (ESADDI) of Mn is 2-5 mg Mn/day147, and daily intake has been found to range from 0.67 – 4.99 mg with an average daily intake of 2.21 mg148. Though adequate levels of Mn are necessary for normal growth and development, high levels of Mn have been observed in rat models to inhibit growth in offspring; the doses at which this effect was observed were 3-5 times higher than normal rat dietary Mn requirements149. Due to the adverse effects of both increased and decreased levels of Mn, it has been observed that Mn has an inverted U-shaped mechanism of effect150, by which both low151 and high144, 152 Mn can be detrimental, especially for fetal growth and development. Just as fingernail or toenail Hg has been used as a noninvasive surrogate measure of body Hg levels, nail Mn has also been reported in the literature to be informative, and high nail Mn is associated with various cancers153, 154. No nail Hg- or Mn-associated altered patterns of placental DNA methylation have yet been described, but placental epigenetic effects remain a possible link between heavy metal exposures in 15! ! ! utero and postnatal neurobehavioral outcomes. Placental epigenetic associations have been observed in the literature with adverse neurobehavioral outcomes155. Taken together, the literature described above has demonstrated that a link may exist between epigenetic alterations in the placenta and in utero toxicant exposures, such as maternal cigarette smoking and heavy metals, that may in turn be associated with early and later life diseases and disorders, such as preterm birth and adverse neurobehavioral outcomes. In this thesis, the following chapters describe placental methylation profiles which have been identified that are associated with in utero exposure to maternal tobacco smoke and preterm birth, infant toenail Hg and neurobehavior, and infant toenail Mn. 16! ! ! REFERENCES 1. Illumina. Infinium HumanMethylation27 BeadChip Kits. 2011. 2. Cheshire M, Kingston M, McQuillan O, Gittins M. Are HIV-related factors associated with pre-term delivery in a UK inner city setting? Journal of the International AIDS Society 2012; 15:18223. 3. Illumina. Infinium HumanMethylation450 BeadChip Kits. 2011. 4. Miller HC, Hassanein K, Hensleigh PA. Fetal growth retardation in relation to maternal smoking and weight gain in pregnancy. Am J Obstet Gynecol 1976; 125:55-60. 5. Wilhelm-Benartzi CS, Houseman EA, Maccani MA, Poage GM, Koestler DC, Langevin SM, et al. In utero exposures, infant growth, and DNA methylation of repetitive elements and developmentally related genes in human placenta. Environmental health perspectives 2012; 120:296-302. 6. Gilliland FD, Berhane K, Li YF, Rappaport EB, Peters JM. Effects of early onset asthma and in utero exposure to maternal smoking on childhood lung function. American journal of respiratory and critical care medicine 2003; 167:917-24. 7. Lannero E, Wickman M, Pershagen G, Nordvall L. Maternal smoking during pregnancy increases the risk of recurrent wheezing during the first years of life (BAMSE). Respiratory research 2006; 7:3. 8. Haberg SE, Stigum H, Nystad W, Nafstad P. Effects of pre- and postnatal exposure to parental smoking on early childhood respiratory health. Am J Epidemiol 2007; 166:679-86. 17! ! ! 9. Li YF, Langholz B, Salam MT, Gilliland FD. Maternal and grandmaternal smoking patterns are associated with early childhood asthma. Chest 2005; 127:1232-41. 10. Barker DJ. In utero programming of chronic disease. Clin Sci (Lond) 1998; 95:115-28. 11. Silveira PP, Portella AK, Goldani MZ, Barbieri MA. Developmental origins of health and disease (DOHaD). Jornal de pediatria 2007; 83:494-504. 12. Avissar-Whiting M, Veiga KR, Uhl KM, Maccani MA, Gagne LA, Moen EL, et al. Bisphenol A exposure leads to specific microRNA alterations in placental cells. Reprod Toxicol 2010; 29:401-6. 13. Suter M, Abramovici A, Showalter L, Hu M, Shope CD, Varner M, et al. In utero tobacco exposure epigenetically modifies placental CYP1A1 expression. Metabolism: clinical and experimental 2010; 59:1481-90. 14. Maccani MA, Avissar-Whiting M, Banister CE, McGonnigal B, Padbury JF, Marsit CJ. Maternal cigarette smoking during pregnancy is associated with downregulation of miR-16, miR-21, and miR-146a in the placenta. Epigenetics : official journal of the DNA Methylation Society 2010; 5:583-9. 15. Suter M, Ma J, Harris A, Patterson L, Brown KA, Shope C, et al. Maternal tobacco use modestly alters correlated epigenome-wide placental DNA methylation and gene expression. Epigenetics : official journal of the DNA Methylation Society 2011; 6:1284-94. 16. Banister CE, Koestler DC, Maccani MA, Padbury JF, Houseman EA, Marsit CJ. Infant growth restriction is associated with distinct patterns of DNA methylation 18! ! ! in human placentas. Epigenetics : official journal of the DNA Methylation Society 2011; 6:920-7. 17. Nelissen EC, van Montfoort AP, Dumoulin JC, Evers JL. Epigenetics and the placenta. Human reproduction update 2011; 17:397-417. 18. Maccani MA, Marsit CJ. Epigenetics in the placenta. Am J Reprod Immunol 2009; 62:78-89. 19. Novakovic B, Yuen RK, Gordon L, Penaherrera MS, Sharkey A, Moffett A, 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. 20. Filiberto AC, Maccani MA, Koestler D, Wilhelm-Benartzi C, Avissar-Whiting M, Banister CE, et al. Birthweight is associated with DNA promoter methylation of the glucocorticoid receptor in human placenta. Epigenetics : official journal of the DNA Methylation Society 2011; 6:566-72. 21. Lambertini L, Marsit CJ, Sharma P, Maccani M, Ma Y, Hu J, et al. Imprinted gene expression in fetal growth and development. Placenta 2012; 33:480-6. 22. Maccani MA, Padbury JF, Marsit CJ. miR-16 and miR-21 expression in the placenta is associated with fetal growth. PLoS One 2011; 6:e21210. 23. Marsit CJ, Lambertini L, Maccani MA, Koestler DC, Houseman EA, Padbury JF, et al. Placenta-imprinted gene expression association of infant neurobehavior. The Journal of pediatrics 2012; 160:854-60 e2. 19! ! ! 24. Marsit CJ, Maccani MA, Padbury JF, Lester BM. Placental 11-beta hydroxysteroid dehydrogenase methylation is associated with newborn growth and a measure of neurobehavioral outcome. PLoS One 2012; 7:e33794. 25. Maccani MA, Marsit CJ. Exposure and fetal growth-associated miRNA alterations in the human placenta. Clinical epigenetics 2011; 2:401-4. 26. Barker DJ. The origins of the developmental origins theory. Journal of internal medicine 2007; 261:412-7. 27. Gluckman PD, Hanson MA. Developmental origins of disease paradigm: a mechanistic and evolutionary perspective. Pediatric research 2004; 56:311-7. 28. Barker DJ. Maternal nutrition, fetal nutrition, and disease in later life. Nutrition 1997; 13:807-13. 29. Hales CN, Barker DJ. Type 2 (non-insulin-dependent) diabetes mellitus: the thrifty phenotype hypothesis. Diabetologia 1992; 35:595-601. 30. Burdge GC, Lillycrop KA. Nutrition, epigenetics, and developmental plasticity: implications for understanding human disease. Annual review of nutrition 2010; 30:315-39. 31. Roseboom T, de Rooij S, Painter R. The Dutch famine and its long-term consequences for adult health. Early human development 2006; 82:485-91. 32. Roseboom TJ, van der Meulen JH, Ravelli AC, Osmond C, Barker DJ, Bleker OP. Effects of prenatal exposure to the Dutch famine on adult disease in later life: an overview. Twin research : the official journal of the International Society for Twin Studies 2001; 4:293-8. 20! ! ! 33. de Rooij SR, Wouters H, Yonker JE, Painter RC, Roseboom TJ. Prenatal undernutrition and cognitive function in late adulthood. Proceedings of the National Academy of Sciences of the United States of America 2010; 107:16881- 6. 34. Brown AS, Susser ES. Prenatal nutritional deficiency and risk of adult schizophrenia. Schizophrenia bulletin 2008; 34:1054-63. 35. Council NR. Toxicological Effects of Methylmercury. Washington, DC: National Academy Press, 2000. 36. Ilback NG, Sundberg J, Oskarsson A. Methyl mercury exposure via placenta and milk impairs natural killer (NK) cell function in newborn rats. Toxicology letters 1991; 58:149-58. 37. Yang J, Jiang Z, Wang Y, Qureshi IA, Wu XD. Maternal-fetal transfer of metallic mercury via the placenta and milk. Annals of clinical and laboratory science 1997; 27:135-41. 38. Genbacev O, Miller RK. Post-implantation differentiation and proliferation of cytotrophoblast cells: in vitro models--a review. Placenta 2000; 21 Suppl A:S45- 9. 39. Dandrea J, Wilson V, Gopalakrishnan G, Heasman L, Budge H, Stephenson T, et al. Maternal nutritional manipulation of placental growth and glucose transporter 1 (GLUT-1) abundance in sheep. Reproduction 2001; 122:793-800. 40. Desai M, Hales CN. Role of fetal and infant growth in programming metabolism in later life. Biological reviews of the Cambridge Philosophical Society 1997; 72:329-48. 21! ! ! 41. Burris HH, Rifas-Shiman SL, Baccarelli A, Tarantini L, Boeke CE, Kleinman K, et al. Associations of LINE-1 DNA Methylation with Preterm Birth in a Prospective Cohort Study. Journal of developmental origins of health and disease 2012; 3:173-81. 42. Rakyan VK, Down TA, Balding DJ, Beck S. Epigenome-wide association studies for common human diseases. Nature reviews Genetics 2011; 12:529-41. 43. Bird A. Perceptions of epigenetics. Nature 2007; 447:396-8. 44. Prescott S, Saffery R. The role of epigenetic dysregulation in the epidemic of allergic disease. Clinical epigenetics 2011; 2:223-32. 45. Reik W, Walter J. Imprinting mechanisms in mammals. Current opinion in genetics & development 1998; 8:154-64. 46. Prickett AR, Oakey RJ. A survey of tissue-specific genomic imprinting in mammals. Molecular genetics and genomics : MGG 2012; 287:621-30. 47. Cassidy SB, Schwartz S, Miller JL, Driscoll DJ. Prader-Willi syndrome. Genetics in medicine : official journal of the American College of Medical Genetics 2012; 14:10-26. 48. Horsthemke B, Buiting K. Genomic imprinting and imprinting defects in humans. Advances in genetics 2008; 61:225-46. 49. Horsthemke B. Mechanisms of imprint dysregulation. American journal of medical genetics Part C, Seminars in medical genetics 2010; 154C:321-8. 50. Wood AJ, Bourc'his D, Bestor TH, Oakey RJ. Allele-specific demethylation at an imprinted mammalian promoter. Nucleic acids research 2007; 35:7031-9. 22! ! ! 51. Hagiwara Y, Hirai M, Nishiyama K, Kanazawa I, Ueda T, Sakaki Y, et al. Screening for imprinted genes by allelic message display: identification of a paternally expressed gene impact on mouse chromosome 18. Proceedings of the National Academy of Sciences of the United States of America 1997; 94:9249-54. 52. Christian SL, Robinson WP, Huang B, Mutirangura A, Line MR, Nakao M, et al. Molecular characterization of two proximal deletion breakpoint regions in both Prader-Willi and Angelman syndrome patients. American journal of human genetics 1995; 57:40-8. 53. Amos-Landgraf JM, Ji Y, Gottlieb W, Depinet T, Wandstrat AE, Cassidy SB, et al. Chromosome breakage in the Prader-Willi and Angelman syndromes involves recombination between large, transcribed repeats at proximal and distal breakpoints. American journal of human genetics 1999; 65:370-86. 54. Chai JH, Locke DP, Greally JM, Knoll JH, Ohta T, Dunai J, et al. Identification of four highly conserved genes between breakpoint hotspots BP1 and BP2 of the Prader-Willi/Angelman syndromes deletion region that have undergone evolutionary transposition mediated by flanking duplicons. American journal of human genetics 2003; 73:898-925. 55. Glenn CC, Saitoh S, Jong MT, Filbrandt MM, Surti U, Driscoll DJ, et al. Gene structure, DNA methylation, and imprinted expression of the human SNRPN gene. American journal of human genetics 1996; 58:335-46. 56. Lossie AC, Whitney MM, Amidon D, Dong HJ, Chen P, Theriaque D, et al. Distinct phenotypes distinguish the molecular classes of Angelman syndrome. Journal of medical genetics 2001; 38:834-45. 23! ! ! 57. Williams CA, Driscoll DJ, Dagli AI. Clinical and genetic aspects of Angelman syndrome. Genetics in medicine : official journal of the American College of Medical Genetics 2010; 12:385-95. 58. Lefebvre L. The placental imprintome and imprinted gene function in the trophoblast glycogen cell lineage. Reproductive biomedicine online 2012; 25:44- 57. 59. Coan PM, Burton GJ, Ferguson-Smith AC. Imprinted genes in the placenta--a review. Placenta 2005; 26 Suppl A:S10-20. 60. Fowden AL, Coan PM, Angiolini E, Burton GJ, Constancia M. Imprinted genes and the epigenetic regulation of placental phenotype. Progress in biophysics and molecular biology 2011; 106:281-8. 61. Frost JM, Moore GE. The importance of imprinting in the human placenta. PLoS genetics 2010; 6:e1001015. 62. Piedrahita JA. The role of imprinted genes in fetal growth abnormalities. Birth defects research Part A, Clinical and molecular teratology 2011; 91:682-92. 63. Cooper WN, Constancia M. How genome-wide approaches can be used to unravel the remaining secrets of the imprintome. Briefings in functional genomics 2010; 9:315-28. 64. Henckel A, Arnaud P. Genome-wide identification of new imprinted genes. Briefings in functional genomics 2010; 9:304-14. 65. Redline RW, Chernicky CL, Tan HQ, Ilan J, Ilan J. Differential expression of insulin-like growth factor-II in specific regions of the late (post day 9.5) murine placenta. Molecular reproduction and development 1993; 36:121-9. 24! ! ! 66. Georgiades P, Watkins M, Burton GJ, Ferguson-Smith AC. Roles for genomic imprinting and the zygotic genome in placental development. Proceedings of the National Academy of Sciences of the United States of America 2001; 98:4522-7. 67. Maher ER, Reik W. Beckwith-Wiedemann syndrome: imprinting in clusters revisited. The Journal of clinical investigation 2000; 105:247-52. 68. Amaral PP, Mattick JS. Noncoding RNA in development. Mammalian genome : official journal of the International Mammalian Genome Society 2008; 19:454-92. 69. Kohtz JD, Fishell G. Developmental regulation of EVF-1, a novel non-coding RNA transcribed upstream of the mouse Dlx6 gene. Gene expression patterns : GEP 2004; 4:407-12. 70. Tsutsumi M, Itoh M. Novel transcript nort is a downstream target gene of the Notch signaling pathway in zebrafish. Gene expression patterns : GEP 2007; 7:227-32. 71. Lee RC, Feinbaum RL, Ambros V. The C. elegans heterochronic gene lin-4 encodes small RNAs with antisense complementarity to lin-14. Cell 1993; 75:843-54. 72. Du T, Zamore PD. Beginning to understand microRNA function. Cell research 2007; 17:661-3. 73. Miska EA. How microRNAs control cell division, differentiation and death. Current opinion in genetics & development 2005; 15:563-8. 74. Bentwich I, Avniel A, Karov Y, Aharonov R, Gilad S, Barad O, et al. Identification of hundreds of conserved and nonconserved human microRNAs. Nature genetics 2005; 37:766-70. 25! ! ! 75. Pineles BL, Romero R, Montenegro D, Tarca AL, Han YM, Kim YM, et al. Distinct subsets of microRNAs are expressed differentially in the human placentas of patients with preeclampsia. Am J Obstet Gynecol 2007; 196:261 e1- 6. 76. Mouillet JF, Chu T, Hubel CA, Nelson DM, Parks WT, Sadovsky Y. The levels of hypoxia-regulated microRNAs in plasma of pregnant women with fetal growth restriction. Placenta 2010; 31:781-4. 77. Chang CP, Bruneau BG. Epigenetics and cardiovascular development. Annual review of physiology 2012; 74:41-68. 78. Cosgrove MS, Boeke JD, Wolberger C. Regulated nucleosome mobility and the histone code. Nature structural & molecular biology 2004; 11:1037-43. 79. Strahl BD, Allis CD. The language of covalent histone modifications. Nature 2000; 403:41-5. 80. Tan M, Luo H, Lee S, Jin F, Yang JS, Montellier E, et al. Identification of 67 histone marks and histone lysine crotonylation as a new type of histone modification. Cell 2011; 146:1016-28. 81. Jaenisch R, Bird A. Epigenetic regulation of gene expression: how the genome integrates intrinsic and environmental signals. Nature genetics 2003; 33 Suppl:245-54. 82. Jenuwein T, Allis CD. Translating the histone code. Science 2001; 293:1074-80. 83. Kouzarides T. Chromatin modifications and their function. Cell 2007; 128:693- 705. 26! ! ! 84. Greer EL, Shi Y. Histone methylation: a dynamic mark in health, disease and inheritance. Nature reviews Genetics 2012; 13:343-57. 85. Fullgrabe J, Kavanagh E, Joseph B. Histone onco-modifications. Oncogene 2011; 30:3391-403. 86. Gheorghe CP, Goyal R, Mittal A, Longo LD. Gene expression in the placenta: maternal stress and epigenetic responses. The International journal of developmental biology 2010; 54:507-23. 87. Chen CP, Wang KG, Chen CY, Yu C, Chuang HC, Chen H. Altered placental syncytin and its receptor ASCT2 expression in placental development and pre- eclampsia. BJOG : an international journal of obstetrics and gynaecology 2006; 113:152-8. 88. Sharma S, Kelly TK, Jones PA. Epigenetics in cancer. Carcinogenesis 2010; 31:27-36. 89. Rehan VK, Liu J, Naeem E, Tian J, Sakurai R, Kwong K, et al. Perinatal nicotine exposure induces asthma in second generation offspring. BMC medicine 2012; 10:129. 90. Bird AP. Use of restriction enzymes to study eukaryotic DNA methylation: II. The symmetry of methylated sites supports semi-conservative copying of the methylation pattern. Journal of molecular biology 1978; 118:49-60. 91. Ehrlich M, Gama-Sosa MA, Huang LH, Midgett RM, Kuo KC, McCune RA, et al. Amount and distribution of 5-methylcytosine in human DNA from different types of tissues of cells. Nucleic acids research 1982; 10:2709-21. 27! ! ! 92. Ameri A, Alidoosti A, Hosseini SY, Parvin M, Emranpour MH, Taslimi F, et al. Prognostic Value of Promoter Hypermethylation of Retinoic Acid Receptor Beta (RARB) and CDKN2 (p16/MTS1) in Prostate Cancer. Chinese journal of cancer research = Chung-kuo yen cheng yen chiu 2011; 23:306-11. 93. Zilberman D, Gehring M, Tran RK, Ballinger T, Henikoff S. Genome-wide analysis of Arabidopsis thaliana DNA methylation uncovers an interdependence between methylation and transcription. Nature genetics 2007; 39:61-9. 94. Zhang X, Yazaki J, Sundaresan A, Cokus S, Chan SW, Chen H, et al. Genome- wide high-resolution mapping and functional analysis of DNA methylation in arabidopsis. Cell 2006; 126:1189-201. 95. Ball MP, Li JB, Gao Y, Lee JH, LeProust EM, Park IH, et al. Targeted and genome-scale strategies reveal gene-body methylation signatures in human cells. Nature biotechnology 2009; 27:361-8. 96. Hellman A, Chess A. Gene body-specific methylation on the active X chromosome. Science 2007; 315:1141-3. 97. Rauch TA, Wu X, Zhong X, Riggs AD, Pfeifer GP. A human B cell methylome at 100-base pair resolution. Proceedings of the National Academy of Sciences of the United States of America 2009; 106:671-8. 98. Cedar H, Bergman Y. Epigenetics of haematopoietic cell development. Nature reviews Immunology 2011; 11:478-88. 99. North ML, Ellis AK. The role of epigenetics in the developmental origins of allergic disease. Annals of allergy, asthma & immunology : official publication of 28! ! ! the American College of Allergy, Asthma, & Immunology 2011; 106:355-61; quiz 62. 100. Van Neste L, Herman JG, Otto G, Bigley JW, Epstein JI, Van Criekinge W. The epigenetic promise for prostate cancer diagnosis. The Prostate 2012; 72:1248-61. 101. Gao Y, Chen J, Li K, Wu T, Huang B, Liu W, et al. Replacement of Oct4 by Tet1 during iPSC Induction Reveals an Important Role of DNA Methylation and Hydroxymethylation in Reprogramming. Cell stem cell 2013. 102. Wang T, Pan Q, Lin L, Szulwach KE, Song CX, He C, et al. Genome-wide DNA hydroxymethylation changes are associated with neurodevelopmental genes in the developing human cerebellum. Human molecular genetics 2012; 21:5500-10. 103. Serandour AA, Avner S, Oger F, Bizot M, Percevault F, Lucchetti-Miganeh C, et al. Dynamic hydroxymethylation of deoxyribonucleic acid marks differentiation- associated enhancers. Nucleic acids research 2012; 40:8255-65. 104. Olkhov-Mitsel E, Bapat B. Strategies for discovery and validation of methylated and hydroxymethylated DNA biomarkers. Cancer medicine 2012; 1:237-60. 105. Wilhelm-Benartzi CS, Christensen BC, Koestler DC, Andres Houseman E, Schned AR, Karagas MR, et al. Association of secondhand smoke exposures with DNA methylation in bladder carcinomas. Cancer causes & control : CCC 2011; 22:1205-13. 106. Koren G, Klein J, Forman R, Graham K, Phan MK. Biological markers of intrauterine exposure to cocaine and cigarette smoking. Developmental pharmacology and therapeutics 1992; 18:228-36. 29! ! ! 107. Lambers DS, Clark KE. The maternal and fetal physiologic effects of nicotine. Seminars in perinatology 1996; 20:115-26. 108. Cnattingius S. The epidemiology of smoking during pregnancy: smoking prevalence, maternal characteristics, and pregnancy outcomes. Nicotine & tobacco research : official journal of the Society for Research on Nicotine and Tobacco 2004; 6 Suppl 2:S125-40. 109. Goodwin RD, Keyes K, Simuro N. Mental disorders and nicotine dependence among pregnant women in the United States. Obstetrics and gynecology 2007; 109:875-83. 110. Shiono PH, Klebanoff MA, Rhoads GG. Smoking and drinking during pregnancy. Their effects on preterm birth. JAMA : the journal of the American Medical Association 1986; 255:82-4. 111. Simpson WJ. A preliminary report on cigarette smoking and the incidence of prematurity. Am J Obstet Gynecol 1957; 73:807-15. 112. Olsen J. Cigarette smoking in pregnancy and fetal growth. Does the type of tobacco play a role? Int J Epidemiol 1992; 21:279-84. 113. Stroud LR, Paster RL, Goodwin MS, Shenassa E, Buka S, Niaura R, et al. Maternal smoking during pregnancy and neonatal behavior: a large-scale community study. Pediatrics 2009; 123:e842-8. 114. Toschke AM, Montgomery SM, Pfeiffer U, von Kries R. Early intrauterine exposure to tobacco-inhaled products and obesity. Am J Epidemiol 2003; 158:1068-74. 30! ! ! 115. von Kries RB, G; Baghi, L.; Toschke, A. MGME Study Group. Prenatal smoking and childhood obesity--is maternal smoking in pregnancy the critical exposure? Int J Epidemiol 2008; 37:7. 116. Oken E, Levitan EB, Gillman MW. Maternal smoking during pregnancy and child overweight: systematic review and meta-analysis. Int J Obes (Lond) 2008; 32:201-10. 117. Tong VT, Jones JR, Dietz PM, D'Angelo D, Bombard JM. Trends in smoking before, during, and after pregnancy - Pregnancy Risk Assessment Monitoring System (PRAMS), United States, 31 sites, 2000-2005. MMWR Surveill Summ 2009; 58:1-29. 118. Lux AL, Henderson AJ, Pocock SJ. Wheeze associated with prenatal tobacco smoke exposure: a prospective, longitudinal study. ALSPAC Study Team. Archives of disease in childhood 2000; 83:307-12. 119. Magnusson LL, Olesen AB, Wennborg H, Olsen J. Wheezing, asthma, hayfever, and atopic eczema in childhood following exposure to tobacco smoke in fetal life. Clinical and experimental allergy : journal of the British Society for Allergy and Clinical Immunology 2005; 35:1550-6. 120. Singh SP, Barrett EG, Kalra R, Razani-Boroujerdi S, Langley RJ, Kurup V, et al. Prenatal cigarette smoke decreases lung cAMP and increases airway hyperresponsiveness. American journal of respiratory and critical care medicine 2003; 168:342-7. 31! ! ! 121. Prabhu N, Smith N, Campbell D, Craig LC, Seaton A, Helms PJ, et al. First trimester maternal tobacco smoking habits and fetal growth. Thorax 2010; 65:235-40. 122. Wongtrakool C, Wang N, Hyde DM, Roman J, Spindel ER. Prenatal nicotine exposure alters lung function and airway geometry through alpha7 nicotinic receptors. American journal of respiratory cell and molecular biology 2012; 46:695-702. 123. Davidson PW, Myers GJ, Weiss B. Mercury exposure and child development outcomes. Pediatrics 2004; 113:1023-9. 124. Fletcher AM, Gelberg KH. An Analysis of Mercury Exposures Among the Adult Population in New York State. Journal of community health 2012. 125. Ask K, Akesson A, Berglund M, Vahter M. Inorganic mercury and methylmercury in placentas of Swedish women. Environmental health perspectives 2002; 110:523-6. 126. Takahashi Y, Tsuruta S, Hasegawa J, Kameyama Y, Yoshida M. Release of mercury from dental amalgam fillings in pregnant rats and distribution of mercury in maternal and fetal tissues. Toxicology 2001; 163:115-26. 127. Vahter M, Akesson A, Lind B, Bjors U, Schutz A, Berglund M. Longitudinal study of methylmercury and inorganic mercury in blood and urine of pregnant and lactating women, as well as in umbilical cord blood. Environmental research 2000; 84:186-94. 128. Counter SA, Buchanan LH. Mercury exposure in children: a review. Toxicology and applied pharmacology 2004; 198:209-30. 32! ! ! 129. Harada Y, Miyamoto Y, Nonaka I, Ohta S, Ninomiya T. Electroencephalographic studies of Minamata disease in children. Developmental medicine and child neurology 1968; 10:257-8. 130. Bakir F, Damluji SF, Amin-Zaki L, Murtadha M, Khalidi A, al-Rawi NY, et al. Methylmercury poisoning in Iraq. Science 1973; 181:230-41. 131. Cox C, Clarkson TW, Marsh DO, Amin-Zaki L, Tikriti S, Myers GG. Dose- response analysis of infants prenatally exposed to methyl mercury: an application of a single compartment model to single-strand hair analysis. Environmental research 1989; 49:318-32. 132. Grandjean P, Weihe P, White RF, Debes F, Araki S, Yokoyama K, et al. Cognitive deficit in 7-year-old children with prenatal exposure to methylmercury. Neurotoxicology and teratology 1997; 19:417-28. 133. Hinners T, Tsuchiya A, Stern AH, Burbacher TM, Faustman EM, Marien K. Chronologically matched toenail-Hg to hair-Hg ratio: temporal analysis within the Japanese community (U.S.). Environmental health : a global access science source 2012; 11:81. 134. Steuerwald U, Weihe P, Jorgensen PJ, Bjerve K, Brock J, Heinzow B, et al. Maternal seafood diet, methylmercury exposure, and neonatal neurologic function. The Journal of pediatrics 2000; 136:599-605. 135. Cace IB, Milardovic A, Prpic I, Krajina R, Petrovic O, Vukelic P, et al. Relationship between the prenatal exposure to low-level of mercury and the size of a newborn's cerebellum. Med Hypotheses 2011; 76:514-6. 33! ! ! 136. Choi BH. The effects of methylmercury on the developing brain. Progress in neurobiology 1989; 32:447-70. 137. Gao Y, Yan CH, Tian Y, Wang Y, Xie HF, Zhou X, et al. Prenatal exposure to mercury and neurobehavioral development of neonates in Zhoushan City, China. Environmental research 2007; 105:390-9. 138. Llop S, Guxens M, Murcia M, Lertxundi A, Ramon R, Riano I, et al. Prenatal exposure to mercury and infant neurodevelopment in a multicenter cohort in Spain: study of potential modifiers. Am J Epidemiol 2012; 175:451-65. 139. Fukui AF, A.; Yokota, M.; Yamada, K.; Nakamua, R.; Fukuhara, R.; Kimura, H.; Mizunuma, H. Uterine and circulating natural killer cells and their roles in women with recurrent pregnancy loss, implantaton failure and preeclampsia. Journal of Reproductive Immunology 2011; 90:6. 140. Rice DC. Evidence for delayed neurotoxicity produced by methylmercury. Neurotoxicology 1996; 17:583-96. 141. He K, Xun P, Liu K, Morris S, Reis J, Guallar E. Mercury Exposure in Young Adulthood and Incidence of Diabetes Later in Life: The CARDIA trace element study. Diabetes care 2013. 142. Wickre JB, Folt CL, Sturup S, Karagas MR. Environmental exposure and fingernail analysis of arsenic and mercury in children and adults in a Nicaraguan gold mining community. Archives of environmental health 2004; 59:400-9. 143. Xun P, Liu K, Morris JS, Jordan JM, He K. Distributions and determinants of mercury concentrations in toenails among American young adults: the CARDIA 34! ! ! Trace Element Study. Environmental science and pollution research international 2013; 20:1423-30. 144. Misselwitz B, Muhler A, Weinmann HJ. A toxicologic risk for using manganese complexes? A literature survey of existing data through several medical specialties. Investigative radiology 1995; 30:611-20. 145. Nielsen FH. How should dietary guidance be given for mineral elements with beneficial actions or suspected of being essential? The Journal of nutrition 1996; 126:2377S-85S. 146. Pennington JA, Young BE. Total diet study nutritional elements, 1982-1989. Journal of the American Dietetic Association 1991; 91:179-83. 147. Greger JL. Dietary standards for manganese: overlap between nutritional and toxicological studies. The Journal of nutrition 1998; 128:368S-71S. 148. Tripathi RM, Mahapatra S, Raghunath R, Sastry VN, Krishnamoorthy TM. Daily intake of manganese by the adult population of Mumbai. The Science of the total environment 2000; 250:43-50. 149. Antonova MV. [Effect of different amounts of manganese in the daily food ration on the growth and development of the progeny of white rats]. Voprosy pitaniia 1978:65-8. 150. Zota AR, Ettinger AS, Bouchard M, Amarasiriwardena CJ, Schwartz J, Hu H, et al. Maternal blood manganese levels and infant birth weight. Epidemiology 2009; 20:367-73. 151. Mistry HD, Williams PJ. The importance of antioxidant micronutrients in pregnancy. Oxidative medicine and cellular longevity 2011; 2011:841749. 35! ! ! 152. Takser L, Mergler D, Hellier G, Sahuquillo J, Huel G. Manganese, monoamine metabolite levels at birth, and child psychomotor development. Neurotoxicology 2003; 24:667-74. 153. Wozniak A, Kujawa A, Senczuk-Przybylowska M, Kulza M, Gawecki W, Szybiak B, et al. [Physiological metals in the serum, hair and nails of patients with head and neck cancer]. Przeglad lekarski 2012; 69:785-97. 154. Karimi G, Shahar S, Homayouni N, Rajikan R, Abu Bakar NF, Othman MS. Association between trace element and heavy metal levels in hair and nail with prostate cancer. Asian Pacific journal of cancer prevention : APJCP 2012; 13:4249-53. 155. Maccani MA, Padbury JF, Lester BM, Knopik VS, Marsit CJ. Placental miRNA expression profiles associated with measures of infant neurobehavioral outcomes. Pediatric research In Press. ! ! 36! ! ! ! CHAPTER 2 Smoking-Related Changes to Placental DNA Methylation at the RUNX3 Gene are Associated with Gestational Age Maccani, Jennifer Z.J., Koestler, Devin C., Houseman, E. Andrés, Marsit, Carmen J.; and Kelsey, Karl T. (Manuscript in preparation.) ! 37! Smoking-Related Changes to Placental DNA Methylation at the RUNX3 Gene are Associated with Gestational Age Jennifer Z. J. Maccani, B.S.1, Devin C. Koestler, PhD2, E. Andrés Houseman, ScD3, Carmen J. Marsit, PhD4, and Karl T. Kelsey, MD, MOH1,5 1 Department of Pathology and Laboratory Medicine, Brown University, Providence, RI, USA 2 Section of Biostatistics & Epidemiology, Dartmouth Medical School, Lebanon, NH, USA 3 College of Public Health and Human Sciences, Oregon State University, Corvallis, OR, USA 4 Departments of Pharmacology and Toxicology and Community and Family Medicine, Geisel School of Medicine at Dartmouth, Hanover, New Hampshire, USA 5 Department of Epidemiology, Brown University, Providence, RI, USA Corresponding Author: Carmen J. Marsit, Associate Professor, Departments of Pharmacology & Toxicology and Community and Family Medicine, Geisel School of Medicine at Dartmouth, HB 7650, Pharmacology & Toxicology, Hanover, NH, 03755, USA; phone: 603-650-1825; fax: 603-650-1129; email: Carmen.J.Marsit@Dartmouth.edu RUNNING TITLE: Smoking, RUNX3 methylation and gestational age KEY WORDS: epigenetics, gestational age, in utero, methylation, placenta, pregnancy, RUNX3, smoking ! 38! ABBREVIATIONS: single nucleotide polymorphism (SNP), small-for-gestational age (SGA), intrauterine growth restriction (IUGR), Principal Components Analysis (PCA), Principal Components (PCs), polymerase chain reaction (PCR) COMPETING FINANCIAL INTERESTS DECLARATION: The authors declare that there are no competing financial interests. ! 39! ABSTRACT Background: The Developmental Origins of Health and Disease hypothesis (DOHaD) states that later life disease may be influenced by the quality of the in utero environment. Environmental toxicants can have detrimental effects on fetal development potentially through effects on placental development and function. Maternal smoking during pregnancy is associated with low birth weight, preterm birth, and other complications and exposure to cigarette smoke in utero has been linked to gross, pathologic, and molecular changes to the placenta including differential DNA methylation in placental tissue. Objectives: To investigate the relationship between maternal smoking during pregnancy, methylation changes in the placenta, and gestational age. Methods: We used Illumina’s Human Methylation27 BeadChip technology platform to interrogate the methylation status of 21,551 autosomal, non-single nucleotide polymorphism (SNP)-associated CpG loci in DNA extracted from 206 human placentas and examined loci whose variation in methylation was associated with maternal smoking during pregnancy. Results: We found that methylation patterns of a number of loci within the RUNX3 gene were significantly associated with smoking during pregnancy, and one of these loci was associated with decreased gestational age (p=0.04). Conclusions: Our findings, demonstrating maternal smoking-induced changes in DNA methylation at specific loci, suggest a mechanism by which in utero tobacco smoke exposure could exert its detrimental effects upon the health of the fetus. ! 40! INTRODUCTION It has long been suspected that environmental exposures in utero can increase susceptibility to adult disease. A growing body of literature suggests that the in utero environment may play a role in the development of cardiovascular disease, diabetes, and certain cancers (Burdge 2010; Kulis 2010; Herz 2000; Prescott 2009; Hollingsworth 2008). The Barker hypothesis posits that the in utero environment can affect offspring, altering risk for the development of disease throughout life (Barker 1997, 1998; Hales 1992; Silveira et al. 2007). Indeed, it has been suggested that even grand-maternal exposures can affect future disease susceptibility (Li et al. 2005). Fetal programming in response to the in utero environment is thought to be of epigenetic origin, where heritable changes to gene expression occur without direct changes to DNA sequence (Rakyan 2011; Bird 2007; Prescott and Saffery 2011). The placenta plays an important role in regulating fetal growth and development, as it produces a number of growth factors and hormones. Additionally, the placenta exhibits a significant degree of metabolic activity, including the metabolism of potentially toxic compounds (Maccani and Marsit 2009). However, many toxicants are capable of crossing the placenta, acting directly or potentially by altering the metabolic function of the placenta. Environmental toxicants that cross the placenta may affect placental function by modifying the epigenetic state of the tissue, including altering DNA methylation (Hoyo et al. 2012; Burris et al. 2012). Thus, epigenetic marks in the placenta can serve as a record of in utero exposure (Maccani and Marsit 2009). ! 41! Maternal tobacco smoking during pregnancy is associated with significant morbidity and mortality, both perinatally and later in life. Several chemicals found in tobacco smoke, including nicotine, can cross the placenta and negatively impact the fetus (Maccani 2010). Nicotine accumulates in fetal blood and amniotic fluid (Koren 1992), and fetal nicotine levels have been shown to be 15% higher than maternal levels (Lambers and Clark 1996). The detrimental effects of maternal tobacco smoking include premature birth (Simpson 1957; Shiono et al. 1986), low birth weight (Olsen 1992; Miller et al. 1976), abnormal neurobehavioral outcomes (Stroud 2009), childhood obesity (Toschke 2003; von Kries 2008; Oken 2008), respiratory tract diseases and sudden infant death syndrome (Tong et al. 2009). Prenatal tobacco exposure can have damaging effects through both genetic and epigenetic mechanisms (Maccani and Marsit 2009; Nelissen et al. 2011), and maternal tobacco smoking during pregnancy is associated with altered DNA methylation patterns in the placenta (Wilhelm-Benartzi 2011; Suter et al. 2010; Suter 2011). Additionally, DNA methylation profiles associated with gestational age have been identified (Novakovic et al. 2011). We hypothesized that maternal smoking during pregnancy is associated with changes to DNA methylation in the placenta and that smoking-associated DNA methylation alterations are in turn associated with altered gestational age, thereby providing a biological mechanism linking this exposure to important reproductive outcomes. METHODS Study Design. 206 placental samples were collected from infants delivered at Women and Infants Hospital in Providence, RI, USA between September 2008 and September 2009 (Banister et al. 2011), using Institutional Review Board (IRB)-approved ! 42! protocols at all involved institutions. This cohort oversampled for small for gestational age (SGA, <10th percentile of birthweight) infants and clinical intrauterine growth restricted (IUGR) diagnoses. Placental samples were collected within 2 hours of birth and full thickness sections were taken from the maternal side of the placenta 2 cm from the umbilical cord insertion site. The samples were placed in RNAlaterTM (Applied Biosystems, Inc., AM7020) immediately after collection. After being stored at 4ºC for at least 72 hours, the samples were blotted dry of RNAlater and then frozen at -80ºC prior to processing. Smoking status at any time during pregnancy was recorded from patient charts, and was analyzed as a dichotomous variable. DNA extraction and bisulfite modification. As previously described (Banister et al. 2011), DNA extractions were performed using the QIAmp DNA Mini Kit (Qiagen, Inc., 51304) and quantified by NanoDrop ND-1000 spectrophotometer (NanoDrop). Bisulfite modification using the EZ DNA Methylation Kit (Zymo Research, D5008) was performed on 1 µg of extracted placental DNA. DNA from one peripheral blood sample from an adult not included in the study was also extracted and bisulfite modified; the bisulfite-modified DNA from this blood sample was run on each BeadChip to control for inter-array variability. DNA methylation profiling. Bisulfite modified DNA samples were arrayed using the Illumina Infinium HumanMethylation27 BeadArray (Illumina 2011) at the University of California San Francisco (UCSF) Institute for Human Genetics Genomic Core Facility (Banister et al. 2011). β values representing the methylation status at each CpG locus were calculated from the intensity of the methylated (M) and unmethylated (U) alleles, where the ratio of fluorescent signals β = Max(M,0)/[Max(M,0) + Max(U,0) + 100] and 0 ! 43! < β < 1. β values near 1 indicate complete methylation and values near 0 indicate absence of methylation. Array quality assurance was assessed according to the method described previously (Banister et al. 2011), and 21,551 autosomal, non-single nucleotide polymorphism (SNP)-associated CpG loci were utilized in this analysis. Statistical analysis. As the β values of these loci were non-normally distributed, they were logit transformed (Kuan et al. 2010). To adjust the data for batch effect (BeadChip), the ComBat procedure (Johnson et al. 2007) was applied to the methylation array data. After logit-transforming the β values of 21,551 autosomal, non-single nucleotide polymorphism (SNP)-associated loci, a Principal Components Analysis (PCA) was performed to analyze the data for associations between Principal Components (PC) and variables of interest, such as: age, sex, smoking during pregnancy case-control status and BeadChip. The analysis of sources of variability within array data is crucial in determining whether certain aspects of the data, for example batch effects due to BeadChip, may contribute to a significant proportion of the variance of the data and thereby may confound analysis results. PCA facilitates an analysis of the predominant sources of variability in methylation across the array; if a significant amount of variability is associated with technical features of the array/assay processing (i.e., batch, plate, BeadChip), the data may need to be controlled or adjusted for those effects. Figures 5-10 and Tables 5 and 6 demonstrate this technical variability prior to and after adjustment, and describe the relationship of the variability identified in the PCA with technical features of the array as well as demographic features of the population. Post- ComBat univariate analyses of the first three PCs with BeadChip and other variables of interest are summarized in Table 5, with significant associations between the first and ! 44! third PCs and gestational age identified. As the racial breakdown of the mothers within this study could have impacted data variance and, in turn, could have influenced the findings resulting from our analysis, maternal race was analyzed following ComBat adjustment. No significant associations between the first and third PCs and maternal race, IVF, or alcohol use during pregnancy were identified. Also, there were no differences in the proportion of smokers between Caucasian and non-Caucasian mothers. After ComBat adjustment, associations between differential methylation at each of 21,551 CpG loci and maternal smoking during pregnancy were investigated using a locus-by-locus approach. This consisted of a series of independent linear regression models modeling logit- transformed, ComBat-adjusted methylation values as the dependent variable and smoking status during pregnancy as the independent variable. False discovery rate estimation was used to control for the large number of tests performed (Benjamini 1995). RUNX3 CpG loci exhibiting associations with maternal smoking during pregnancy were subsequently analyzed for associations with gestational age (dichotomized, <37 weeks gestation vs. ≥ 37 weeks gestation, as this is considered the clinical threshold for preterm birth) using both two-tailed Student’s t-tests and multivariable logistic regression models, controlled for potential confounders (maternal age, infant gender, birth weight, and delivery method (vaginal vs. Caesarean section)). Bisulfite pyrosequencing DNA methylation analysis. Pyrosequencing was performed to confirm array findings for specific CpG loci of interest. After bisulfite modifying DNA from 22 placental samples (11 samples from smoking mothers and 11 samples from non-smoking mothers), pyrosequencing was performed on polymerase chain reaction (PCR)-amplified product. Pyrosequencing assays were designed using ! 45! Pyromark Assay Design 2.0 software (Qiagen, Inc., 9019077) and ordered from Invitrogen. Pyrosequencing was performed on a Pyromark MD pyrosequencing instrument running Pyromark qCpG 1.1.11 software (Qiagen, Inc.). PCR was performed using HotStarTaq DNA Polymerase (Qiagen, Inc., 203205). cg06037693, cg00117172, and cg04757093 were assessed by pyrosequencing. For cg06037693, the following forward and biotinylated reverse primers were used for amplification: RUNX3 cg06037693-FW, 5’- TTTTTGGTAATAATGGTGGTGGATAATGG -3’ and RUNX3 cg06037693-Biot-RV, 5’-biotin- ACTCCTAAATAAAAAACTCCTTCTTAAT -3’. For cg00117172, the following forward and biotinylated reverse primers were used: RUNX3 cg00117172-FW, 5’- GTTTTGGATGAGTTTTGTAGGAATGAT -3’ and RUNX3 cg00117172-Biot-RV, 5’-biotin- CCCTATCCCCAAATCCTCTTCTCC -3’. For cg04757093, the following biotinylated forward and reverse primers were used: RUNX3 cg04757093-Biot-FW, 5’-biotin- GATGGGTTTTGGGAATTAGAGTTTAAG -3’ and RUNX3 cg04757093-RV, 5’- ACTAACATAACCCCCAAATAATACATCCTA -3’. Cycling conditions for all primer sets were 95°C for 15 minutes followed by 40 cycles of 94°C for 30 seconds, an assay-dependent annealing temperature for 30 seconds and 72°C for 1 minute with a final 10 minute extension at 72°C. Annealing temperatures were 52.2°C for cg06037693; 59.0°C for cg00117172; 55.9°C for cg04757093. The following sequencing primers for each assay were used for pyrosequencing: RUNX3 cg06037693- Seq, 5’- AATGGTAGGGAGTTAG -3’; RUNX3 cg00117172-Seq, 5’- GTTGTTTAGTTTTATTTGGGTTT -3’; RUNX3 cg04757093-Seq, 5’- AAAAAAAAATCAATTCCAACT -3’. ! 46! RESULTS' The demographics of the study population are described in Table 1. The population was over-sampled for small-for-gestational age (SGA) infants (28%) and the majority of the infants in this study were born at or near term, with a mean gestational age of 38.2 weeks. Nearly 11% of mothers reported smoking during pregnancy. The DNA methylation status of the 206 placenta samples in this study was interrogated using the Illumina Infinium HumanMethylation27 BeadChip, which examines 27,578 loci. Poor-performing loci, those associated with the sex chromosomes, and those whose probe contained a SNP were removed, leaving 21,551 loci in 206 samples for study. Principal Components Analysis (PCA) was performed to test for associations between the first three Principal Components (PCs) with several variables including BeadChip. Since BeadChip was significantly associated with the top three PCs, which represent the maximal variation in methylation across the array, ComBat (Johnson et al. 2007) was used to normalize the array methylation data according to BeadChip. A locus-by-locus analysis assessed possible associations between smoking during pregnancy and differential methylation status at each of the 21,551 autosomal CpG loci. Analysis revealed that 1,918 of these loci had differential methylation patterns associated with maternal smoking during pregnancy (p<0.05; the top 50 CpG loci and p values are given in Table 6), although the lowest q-value observed was 0.3. A Manhattan plot describing the distribution of p values derived from these associations by chromosomal location is shown in Figure 1. CpG loci of interest, described below, were pyrosequenced ! 47! to confirm methylation differences in the placental tissue of smoking and non-smoking mothers. Seven loci residing within the intronic and promoter regions of the RUNX3 gene displayed differential methylation patterns that were significantly associated with maternal smoking during pregnancy. Since the RUNX3 gene has been associated with airway hyperresponsiveness and asthma (Haley et al. 2011; Fainaru et al. 2005; Fainaru et al. 2004; Wongtrakool et al. 2012; Lux et al. 2000; Haberg et al. 2007; Lannero et al. 2006; Magnusson et al. 2005; Gilliland et al. 2003; Stein et al. 1999; Prabhu et al. 2010; Singh et al. 2003) and these conditions are also associated with maternal smoking, this locus was chosen for followup bisulfite pyrosequencing. The locations of these seven loci within the RUNX3 gene and their methylation status in the placentas of smoking and non-smoking mothers are illustrated in Figure 2. One locus is situated immediately preceding the first promoter region (P1), and another in the first intronic region after P1; another locus also situated in the first intronic region precedes the second promoter region (P2); and the remaining four loci significantly associated with maternal smoking during pregnancy are located in intronic regions 2, 4 and 5 of this gene. As maternal smoking during pregnancy is associated with decreased gestational age and preterm birth (Kyrklund-Blomberg and Cnattingius 1998; Shah and Bracken 2000; Shiono et al. 1986), we investigated the methylation status of each RUNX3 CpG locus identified from the smoking analysis for associations with gestational age. Gestational age was modeled as a dichotomous variable consisting of infants of <37 ! 48! weeks gestation and of ≥37 weeks gestation (Cheshire et al. 2012). Univariate analysis revealed that two of the seven CpG loci within RUNX3 that were associated with smoking during pregnancy, cg04757093 and cg14182690, were significantly or near- significantly associated with gestational age (p=0.07 and p=0.01, respectively; Table 2). Multivariable logistic regression models were used to test for associations between the methylation status of these two CpG loci with gestational age, controlling for maternal age, infant gender, birth weight, and delivery method. One CpG locus, cg04757093, was significantly associated with gestational age in the model (Table 3). This locus exhibited significant hypermethylation in the placentas of smoking mothers (n=22) compared to non-smoking mothers (n=184) (p=0.04; Figure 3) and a 1-logit increase in methylation was associated with a 10-fold increased risk for preterm birth (OR 10.2, 95% CI 1.1, 103.3; p<0.04). As orthogonal confirmation of the array results, we assessed methylation by bisulfite pyrosequencing for this loci with the strong association to gestational age (cg04757093 and two neighboring loci (cg06037693, cg00117172) in a subset of placenta samples (total n=22, Figure 4). Pyrosequencing confirmed that the placental tissue from infants exposed to maternal tobacco smoke had significantly higher DNA methylation at cg04757093 (p<0.02) and showed the same trend as the array analysis at cg00117172 (p<0.09), while the trend was not confirmed at cg06037693 by pyrosequencing. DISCUSSION In a sample of 206 human placentas, we observed a significant association between maternal smoking during pregnancy and placental hypermethylation of ! 49! cg04757093, a CpG locus within the gene body of the RUNX3 gene. A growing body of literature suggests that RUNX3 plays an important role in normal immune system development (Woolf et al. 2007; Zamisch et al. 2009), susceptibility to early life disease as a result of in utero exposures (Haley et al. 2011), and many cancers (Chen 2012; Chen et al. 2011; He et al. 2012; Ito et al. 2008; Ito et al. 2005; Li et al. 2002; Lu et al. 2012a; Tang et al. 2012; Tokunaga et al. 2009; Vogiatzi et al. 2006; Xiao and Liu 2004). The runt-related transcription factor 3 (RUNX3) is important for normal cellular differentiation and development, including T-cell differentiation (Woolf et al. 2007; Zamisch et al. 2009; Tokunaga et al. 2009; Klunker et al. 2009), macrophage differentiation (Sanchez-Martin et al. 2011), neuronal cell development (Inoue et al. 2002), and cell-cycle progression (Li et al. 2002), and is known to negatively regulate dendritic cell maturation (Puig-Kroger et al. 2010). RUNX3 is a tumor suppressor gene (Chen 2012; He et al. 2012; Ito et al. 2005; Vogiatzi et al. 2006; Bae and Choi 2004; Tanaka et al. 2007) and interacts with β-catenin (Ito et al. 2008). When upregulated, RUNX3 is known to inhibit cyclins D1 and E and increase p27, Rb and TIMP-1 expression (Chen et al. 2011; He et al. 2012). RUNX3 expression changes are associated with clear cell renal carcinoma (CCRC) (He et al. 2012) and hepatocellular carcinoma (HCC) (Xiao and Liu 2004). RUNX3 hypermethylation is associated with cancers of the breast (Chen 2012), stomach (Li et al. 2002; Tang et al. 2012; Lu et al. 2012), prostate (Mahapatra et al. 2012), and lung (Yanagawa et al. 2003), as well as many others (Puig- Kroger and Corbi 2006). It has also been demonstrated that hypermethylation of RUNX3 in smokers is associated with bladder cancer (Wolff et al. 2008). The precise function of RUNX3 in placenta has not been described. ! 50! Prenatal exposure to tobacco smoke is highly prevalent (Tong et al. 2009) and is associated with the development of childhood asthma (Wongtrakool et al. 2012; Lux et al. 2000; Haberg et al. 2007; Lannero et al. 2006; Magnusson et al. 2005; Gilliland et al. 2003; Stein et al. 1999; Prabhu et al. 2010), in part by increasing airway hyperresponsiveness (Singh et al. 2003). In murine models, the RUNX family of transcription factors has been shown to play a role in the development of prenatal smoke exposure-induced airway hyperresponsiveness (Haley et al. 2011). This is supported by research demonstrating that RUNX3-knockout mice spontaneously develop asthma-like disease (Fainaru et al. 2005; Fainaru et al. 2004). Increased expression of CCR7 in the absence of RUNX3 expression has been shown to allow increased dendritic cell migration to draining lymph nodes, a feature associated with an asthma-like phenotype including airway hyperresponsiveness (Fainaru et al. 2005). Further investigation is necessary to determine the precise nature of the influence of prenatal smoke exposure on placental RUNX3 expression that may result from epigenetic changes, especially considering the limitations of this research, which has observed differential methylation at a single CpG locus within the gene body. More comprehensive coverage of RUNX3 CpG loci, for example via Illumina’s now-available Human Methylation450 BeadChip technology or through comprehensive next-gen sequencing, may reveal relevant methylation changes within RUNX3 in more detail. A larger sample size would also help to investigate the effects of smoking during pregnancy on placental RUNX3 methylation, as a limitation of this research was the relatively small number of smoking mothers in this study population (n=22). We also are limited in our examination of the dosage of exposure, which may also provide additional information on the mechanisms underlying ! 51! these effects. This research, though, does point to a possible role for RUNX3 in tobacco- associated adverse pregnancy outcome and, as RUNX3 expression is controlled by a retinoic acid-sensitive signaling pathway in some cell types, may suggest novel interventional approaches which could be utilized (Puig-Kroger and Corbi 2006). An extensive literature has reported associations between maternal tobacco smoking during pregnancy and the gestational age at birth (Simpson 1957; Kullander 1971; Stillman et al. 1986; Kyrklund-Blomberg and Cnattingius 1998; Shah and Bracken 2000), and a recent study by Joubert et al. (2012) linked maternal tobacco smoking during pregnancy to altered DNA methylation in infant cord blood, suggesting epigenetic mechanisms for tobacco smoke toxicity. Our data provides further evidence for epigenetic toxicity of tobacco smoke exposures in utero, and a potential role for this alteration in preterm birth, which is supported by other studies which have linked epigenetic alterations to this outcome (Burris et al. 2012). Further, this work suggests that altered placental DNA methylation of this locus, and potentially others, should be examined in studies of preterm birth and other adverse pregnancy outcomes as this may provide novel avenues for intervention. Additional research is warranted to elucidate the molecular pathways involved in altered gestational age resulting from maternal smoking during pregnancy and the role that DNA methylation changes in the placenta may play in this process. Further investigation will lead to a greater understanding of the roles of placental DNA methylation in the context of the developmental origins of health and disease. ! 52! ACKNOWLEDGMENTS: The authors wish to thank Joyce Lee, Gilda Ferro, and Keila Veiga for their collection of placental samples and recruitment of patients to this study. ! 53! FUNDING: This research was funded by grants R01MH094609 from NIH-NIMH, P20GM103537 from NIH-NIGMS, and T32ES007272 from the NIH-NIEHS. ! 54! REFERENCES 1. Burdge GCL, K.A. 2010. Nutrition, epigenetics, and developmental plasticity: implications for understanding human disease. Annu Rev Nutr 30: 315-331. 2. Kulis ME, M. 2010. DNA methylation and cancer. Adv Genet 70: 27-56. 3. Herz UJ, R., Ahrens, B., Scheffold, A., Radbruch, A., Renz, H. 2000. Prenatal sensitization in a mouse model. Am J Respir Crit Care Med 162: S62-65. 4. Prescott SC, V. 2009. Asthma and pregnancy: emerging evidence of epigenetic interactions in utero. Curr Opin Allergy Clin Immunol 9: 417-426. 5. Hollingsworth JWM, S., Boon, K., et al. 2008. In utero supplementation with methyl donors enhances allergic airway disease in mice. J Clin Invest 118: 3462- 3469. 6. Barker DJ. 1997. Maternal nutrition, fetal nutrition, and disease in later life. Nutrition 13: 807-813. 7. Barker DJ. 1998. In utero programming of chronic disease. Clin Sci (London) 95: 115-128. 8. Hales CN, Barker DJ. 1992. Type 2 (non-insulin-dependent) diabetes mellitus: the thrifty phenotype hypothesis. Diabetologia 35(7): 595-601. 9. Silveira PP, Portella AK, Goldani MZ, Barbieri MA. 2007. Developmental origins of health and disease (DOHaD). Jornal de pediatria 83(6): 494-504. 10. Li YF, Langholz B, Salam MT, Gilliland FD. 2005. Maternal and grandmaternal smoking patterns are associated with early childhood asthma. Chest 127(4): 1232- 1241. ! 55! 11. Rakyan VKD, T.A., Balding, D.J., Beck, S. 2011. Epigenome-wide association studies for common human diseases. Nat Rev Genet 12: 529-541. 12. Bird A. 2007. Perceptions of epigenetics. Nature 447: 396-398. 13. Prescott S, Saffery R. 2011. The role of epigenetic dysregulation in the epidemic of allergic disease. Clinical epigenetics 2(2): 223-232. 14. Maccani MA, Marsit CJ. 2009. Epigenetics in the placenta. Am J Reprod Immunol 62(2): 78-89. 15. Hoyo C, Fortner K, Murtha AP, Schildkraut JM, Soubry A, Demark-Wahnefried W, et al. 2012. Association of cord blood methylation fractions at imprinted insulin-like growth factor 2 (IGF2), plasma IGF2, and birth weight. Cancer causes & control : CCC 23(4): 635-645. 16. Burris HH, Rifas-Shiman SL, Baccarelli A, Tarantini L, Boeke CE, Kleinman K, et al. 2012. Associations of LINE-1 DNA Methylation with Preterm Birth in a Prospective Cohort Study. Journal of developmental origins of health and disease 3(3): 173-181. 17. Maccani, M.A., Avissar-Whiting, M., Banister, C.E., McGonnigal, B., Padbury, J.F., Marsit, C.J. 2010. Maternal cigarette smoking during pregnancy is associated with downregulation of miR-16, miR-21, and miR-146a in the placenta. Epigenetics 5(7): 583-589. 18. Koren GK, J., Forman, R., Graham, K., Phan, M.K. 1992. Biological markers of intrauterine exposure to cocaine and cigarette smokig. Dev Pharmacol Ther 18(3- 4): 228-236. ! 56! 19. Lambers DS, Clark KE. 1996. The maternal and fetal physiologic effects of nicotine. Seminars in perinatology 20(2): 115-126. 20. Nelissen EC, van Montfoort AP, Dumoulin JC, Evers JL. 2011. Epigenetics and the placenta. Human reproduction update 17(3): 397-417. 21. Simpson WJ. 1957. A preliminary report on cigarette smoking and the incidence of prematurity. Am J Obstet Gynecol 73(4): 807-815. 22. Shiono PH, Klebanoff MA, Rhoads GG. 1986. Smoking and drinking during pregnancy. Their effects on preterm birth. JAMA : the journal of the American Medical Association 255(1): 82-84. 23. Cheshire M, Kingston M, McQuillan O, Gittins M. Are HIV-related factors associated with pre-term delivery in a UK inner city setting? Journal of the International AIDS Society 2012, 15:18223. 24. Olsen J. 1992. Cigarette smoking in pregnancy and fetal growth: Does the type of tobacco play a role? Int J Epidemiol 21(2): 279-284. 25. Miller HC, Hassanein K, Hensleigh PA. 1976. Fetal growth retardation in relation to maternal smoking and weight gain in pregnancy. American journal of obstetrics and gynecology 125(1): 55-60. 26. Stroud LRP, R.L., Goodwin, M.S., et al. 2009. Maternal smoking during pregnancy and neonatal behavior: a large-scale community study. Pediatrics 123(5): e842-848. 27. Toschke AMM, S.M., Pfeiffer, U., von Kries, R. 2003. Early intrauterine exposure to tobbaco-inhaled products and obesity. Am J Epidemiol 158(11): 1068-1074. ! 57! 28. von Kries RB, G., Baghi, L., Toschke, A. 2008. MGME Study Group. Parental smoking and childhood obesity--is maternal smoking in pregnancy the critical exposure? Int J Epidemiol 37(1): 210-216. 29. Oken EL, E.B., Gillman, M.W. 2008. Maternal smoking during pregnancy and child overweight: sytematic review and meta-analysis. Int J Obes (Lond) 32(2): 201-210. 30. Tong VT, Jones JR, Dietz PM, D'Angelo D, Bombard JM. 2009. Trends in smoking before, during, and after pregnancy - Pregnancy Risk Assessment Monitoring System (PRAMS), United States, 31 sites, 2000-2005. MMWR Surveill Summ 58(4): 1-29. 31. Wilhelm-Benartzi CSH, E.A., Maccani, M.A., Poage, G.M., Koestler, D.C., Langevin, S.M., Gagne, L.A., Banister, C., Padbury, J.F., Marsit, C.J. 2011. In utero exposures, infant growth, and DNA methylation of repetitive element and developmentally related genes in human placenta. Environ Health Perspect. 32. Suter M, Abramovici A, Showalter L, Hu M, Shope CD, Varner M, et al. 2010. In utero tobacco exposure epigenetically modifies placental CYP1A1 expression. Metabolism: clinical and experimental 59(10): 1481-1490. 33. Suter MM, J., Harris, A.S., Patterson, L., Brown, K.A., Shope, C., Showalter, L., Abramovici, A., Aagaard-Tillery, K.M. 2011. Maternal tobacco use modestly alters correlated epigenome-wide placental DNA methylation and gene expression. Epigenetics 6(11). 34. Novakovic B, Yuen RK, Gordon L, Penaherrera MS, Sharkey A, Moffett A, et al. 2011. Evidence for widespread changes in promoter methylation profile in human ! 58! placenta in response to increasing gestational age and environmental/stochastic factors. BMC genomics 12: 529. 35. Banister CE, Koestler DC, Maccani MA, Padbury JF, Houseman EA, Marsit CJ. 2011. Infant growth restriction is associated with distinct patterns of DNA methylation in human placentas. Epigenetics : official journal of the DNA Methylation Society 6(7): 920-927. 36. Illumina I. 2011. Infinium HumanMethylation27 BeadChip Kits. Available: http://www.illumina.com/products/infinium_humanmethylation27_beadchip_kits. ilmn [accessed 20 Dec 2011]. 37. Kuan PF, Wang S, Zhou X, Chu H. 2010. A statistical framework for Illumina DNA methylation arrays. Bioinformatics 26(22): 2849-2855. 38. Johnson WE, Li C, Rabinovic A. 2007. Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics 8(1): 118-127. 39. Benjamini YH, Yosef. 1995. Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society, Series B (Methodological) 57(1): 12. 40. Kyrklund-Blomberg NB, Cnattingius S. 1998. Preterm birth and maternal smoking: risks related to gestational age and onset of delivery. American journal of obstetrics and gynecology 179(4): 1051-1055. 41. Shah NR, Bracken MB. 2000. A systematic review and meta-analysis of prospective studies on the association between maternal cigarette smoking and preterm delivery. American journal of obstetrics and gynecology 182(2): 465-472. ! 59! 42. Woolf E, Brenner O, Goldenberg D, Levanon D, Groner Y. 2007. Runx3 regulates dendritic epidermal T cell development. Developmental biology 303(2): 703-714. 43. Zamisch M, Tian L, Grenningloh R, Xiong Y, Wildt KF, Ehlers M, et al. 2009. The transcription factor Ets1 is important for CD4 repression and Runx3 up- regulation during CD8 T cell differentiation in the thymus. The Journal of experimental medicine 206(12): 2685-2699. 44. Haley KJ, Lasky-Su J, Manoli SE, Smith LA, Shahsafaei A, Weiss ST, et al. 2011. RUNX transcription factors: association with pediatric asthma and modulated by maternal smoking. American journal of physiology Lung cellular and molecular physiology 301(5): L693-701. 45. Chen LF. 2012. Tumor suppressor function of RUNX3 in breast cancer. Journal of cellular biochemistry 113(5): 1470-1477. 46. Chen Y, Wei X, Guo C, Jin H, Han Z, Han Y, et al. 2011. Runx3 suppresses gastric cancer metastasis through inactivation of MMP9 by upregulation of TIMP- 1. International journal of cancer Journal international du cancer 129(7): 1586- 1598. 47. He L, Zhao X, Wang H, Zhang P, Guo C, Huang C, et al. 2012. RUNX3 mediates suppression of tumor growth and metastasis of human CCRCC by regulating cyclin related proteins and TIMP-1. PloS one 7(3): e32961. 48. Ito K, Lim AC, Salto-Tellez M, Motoda L, Osato M, Chuang LS, et al. 2008. RUNX3 attenuates beta-catenin/T cell factors in intestinal tumorigenesis. Cancer cell 14(3): 226-237. ! 60! 49. Ito K, Liu Q, Salto-Tellez M, Yano T, Tada K, Ida H, et al. 2005. RUNX3, a novel tumor suppressor, is frequently inactivated in gastric cancer by protein mislocalization. Cancer research 65(17): 7743-7750. 50. Li QL, Ito K, Sakakura C, Fukamachi H, Inoue K, Chi XZ, et al. 2002. Causal relationship between the loss of RUNX3 expression and gastric cancer. Cell 109(1): 113-124. 51. Lu XX, Yu JL, Ying LS, Han J, Wang S, Yu QM, et al. 2012a. Stepwise cumulation of RUNX3 methylation mediated by Helicobacter pylori infection contributes to gastric carcinoma progression. Cancer 118(22): 5507-5517. 52. Tang GH, Sun SW, He XS. 2012. [Correlation of CpG methylation status of Runx3 with pathogenesis of gastric carcinoma]. Zhonghua bing li xue za zhi Chinese journal of pathology 41(5): 314-319. 53. Tokunaga T, Hayashi A, Kadota Y, Shiono H, Inoue M, Sawabata N, et al. 2009. Regulation of Th-POK and Runx3 in T cell development in human thymoma. Autoimmunity 42(8): 653-660. 54. Vogiatzi P, De Falco G, Claudio PP, Giordano A. 2006. How does the human RUNX3 gene induce apoptosis in gastric cancer? Latest data, reflections and reactions. Cancer biology & therapy 5(4): 371-374. 55. Xiao WH, Liu WW. 2004. Hemizygous deletion and hypermethylation of RUNX3 gene in hepatocellular carcinoma. World journal of gastroenterology : WJG 10(3): 376-380. 56. Klunker S, Chong MM, Mantel PY, Palomares O, Bassin C, Ziegler M, et al. 2009. Transcription factors RUNX1 and RUNX3 in the induction and suppressive ! 61! function of Foxp3+ inducible regulatory T cells. The Journal of experimental medicine 206(12): 2701-2715. 57. Sanchez-Martin L, Estecha A, Samaniego R, Sanchez-Ramon S, Vega MA, Sanchez-Mateos P. 2011. The chemokine CXCL12 regulates monocyte- macrophage differentiation and RUNX3 expression. Blood 117(1): 88-97. 58. Inoue K, Ozaki S, Shiga T, Ito K, Masuda T, Okado N, et al. 2002. Runx3 controls the axonal projection of proprioceptive dorsal root ganglion neurons. Nature neuroscience 5(10): 946-954. 59. Puig-Kroger A, Aguilera-Montilla N, Martinez-Nunez R, Dominguez-Soto A, Sanchez-Cabo F, Martin-Gayo E, et al. 2010. The novel RUNX3/p33 isoform is induced upon monocyte-derived dendritic cell maturation and downregulates IL-8 expression. Immunobiology 215(9-10): 812-820. 60. Bae SC, Choi JK. 2004. Tumor suppressor activity of RUNX3. Oncogene 23(24): 4336-4340. 61. Tanaka Y, Imamura J, Kanai F, Ichimura T, Isobe T, Koike M, et al. 2007. Runx3 interacts with DNA repair protein Ku70. Experimental cell research 313(15): 3251-3260. 62. Lu XX, Yu JL, Ying LS, Han J, Wang S, Yu QM, et al. 2012b. Stepwise cumulation of RUNX3 methylation mediated by Helicobacter pylori infection contributes to gastric carcinoma progression. Cancer. 63. Mahapatra S, Klee EW, Young CY, Sun Z, Jimenez RE, Klee GG, et al. 2012. Global methylation profiling for risk prediction of prostate cancer. Clinical cancer ! 62! research : an official journal of the American Association for Cancer Research 18(10): 2882-2895. 64. Yanagawa N, Tamura G, Oizumi H, Takahashi N, Shimazaki Y, Motoyama T. 2003. Promoter hypermethylation of tumor suppressor and tumor-related genes in non-small cell lung cancers. Cancer science 94(7): 589-592. 65. Puig-Kroger A, Corbi A. 2006. RUNX3: a new player in myeloid gene expression and immune response. Journal of cellular biochemistry 98(4): 744-756. 66. Wolff EM, Liang G, Cortez CC, Tsai YC, Castelao JE, Cortessis VK, et al. 2008. RUNX3 methylation reveals that bladder tumors are older in patients with a history of smoking. Cancer research 68(15): 6208-6214. 67. Wongtrakool C, Wang N, Hyde DM, Roman J, Spindel ER. 2012. Prenatal nicotine exposure alters lung function and airway geometry through alpha7 nicotinic receptors. American journal of respiratory cell and molecular biology 46(5): 695-702. 68. Lux AL, Henderson AJ, Pocock SJ. 2000. Wheeze associated with prenatal tobacco smoke exposure: a prospective, longitudinal study. ALSPAC Study Team. Archives of disease in childhood 83(4): 307-312. 69. Haberg SE, Stigum H, Nystad W, Nafstad P. 2007. Effects of pre- and postnatal exposure to parental smoking on early childhood respiratory health. American journal of epidemiology 166(6): 679-686. 70. Lannero E, Wickman M, Pershagen G, Nordvall L. 2006. Maternal smoking during pregnancy increases the risk of recurrent wheezing during the first years of life (BAMSE). Respiratory research 7: 3. ! 63! 71. Magnusson LL, Olesen AB, Wennborg H, Olsen J. 2005. Wheezing, asthma, hayfever, and atopic eczema in childhood following exposure to tobacco smoke in fetal life. Clinical and experimental allergy : journal of the British Society for Allergy and Clinical Immunology 35(12): 1550-1556. 72. Gilliland FD, Berhane K, Li YF, Rappaport EB, Peters JM. 2003. Effects of early onset asthma and in utero exposure to maternal smoking on childhood lung function. American journal of respiratory and critical care medicine 167(6): 917- 924. 73. Stein RT, Holberg CJ, Sherrill D, Wright AL, Morgan WJ, Taussig L, et al. 1999. Influence of parental smoking on respiratory symptoms during the first decade of life: the Tucson Children's Respiratory Study. American journal of epidemiology 149(11): 1030-1037. 74. Prabhu N, Smith N, Campbell D, Craig LC, Seaton A, Helms PJ, et al. 2010. First trimester maternal tobacco smoking habits and fetal growth. Thorax 65(3): 235- 240. 75. Singh SP, Barrett EG, Kalra R, Razani-Boroujerdi S, Langley RJ, Kurup V, et al. 2003. Prenatal cigarette smoke decreases lung cAMP and increases airway hyperresponsiveness. American journal of respiratory and critical care medicine 168(3): 342-347. 76. Fainaru O, Shseyov D, Hantisteanu S, Groner Y. 2005. Accelerated chemokine receptor 7-mediated dendritic cell migration in Runx3 knockout mice and the spontaneous development of asthma-like disease. Proceedings of the National Academy of Sciences of the United States of America 102(30): 10598-10603. ! 64! 77. Fainaru O, Woolf E, Lotem J, Yarmus M, Brenner O, Goldenberg D, et al. 2004. Runx3 regulates mouse TGF-beta-mediated dendritic cell function and its absence results in airway inflammation. The EMBO journal 23(4): 969-979. 78. Kullander SK, B. 1971. A prospective study of smoking and pregnancy. Acta Obstet Gynecol Scand 50: 11. 79. Stillman RJ, Rosenberg MJ, Sachs BP. 1986. Smoking and reproduction. Fertility and sterility 46(4): 545-566. ! ! ! 65! TABLES Table 1. Study population demographics. Total, n (%) 206 (100%) Infant gender Male, n (%) 99 (48.1%) Female, n (%) 107 (51.9%) Maternal age (years) Mean (SD) 27.9 ± 5.9 Median (Range) 27.5 (18-43) Tobacco use Smokers (lifetime), n (%) 25 (12.1%) Smokers (during pregnancy), n (%) 22 (10.7%) Birth weight status Appropriate for Gestational Age, n (%) 139 (67.5%) Small for Gestational Age, n (%) 58 (28.2%) Large for Gestational Age, n (%) 9 (4.4%) Gestational age (weeks) Mean (SD) 38.2 ± 2.0 Median (Range) 39 (28.0-41.2) Birth weight (grams) Mean (SD) 2937.4 ± 599.0 Median (Range) 2892.5 (890-4270) ! 66! Maternal race Caucasian, n (%) 106 (51.5%) Non-Caucasian, n (%) 100 (48.5%) Maternal insurance Private, n (%) 106 (51.5%) Other, n (%) 100 (48.5%) C-section Yes, n (%) 63 (30.6%) No, n (%) 140 (68.0%) Maternal recreational drug use Yes (lifetime), n (%) 11 (5.3%) Yes (during pregnancy), n (%) 6 (2.9%) ! 67! Table 2. Two RUNX3 loci significantly associated with maternal smoking during pregnancy were significantly or near-significantly associated with gestational age (<37 weeks vs. ≥ 37 weeks). CpG Locus of Interest p value cg00117172 0.25 cg00572797 0.50 cg04757093 0.07† cg06037693 0.23 cg08705994 0.37 cg14182690 0.01* cg24019564 0.26 * p<0.05; † p<0.10. ! 68! Table 3. Gestational age <37 weeks was significantly associated with logit-transformed, ComBat-adjusted cg04757093 methylation status (p=0.04) while controlling for potential confounders. Covariate Odds Ratio (95% CI) p value Logit-transformed, adjusted 10.15 (1.07, 103.30) 0.04* cg04757093 methylation Maternal age (years) 1.09 (0.99, 1.19) 0.08† Infant gender 0.59 Male Ref Female 1.36 (0.45, 4.28) Birth weight (grams) 0.997 (0.996, 0.998) 3.19e-6* Delivery method 0.57 Vaginal Ref C-section 0.70 (0.19, 2.29) * p<0.05; † p<0.10. ! 69! Table 4. First three PCs are significantly associated with BeadChip (batch effect). Associations with PCs (p value) Variable PC1 PC2 PC3 BeadChip 5.35E-14* 1.80E-04* 8.31E-15* Infant gender 0.12 0.94 0.37 Gestational age (weeks) 0.01* 0.09 6.70E-06* Maternal age (years) 0.12 0.15 0.86 Birth weight (grams) 0.68 0.01* 1.35E-05* Smoking during pregnancy 0.30 0.40 0.60 * p<0.05. ! 70! Table 5. First three PCs are no longer significantly associated with BeadChip after ComBat adjustment. Associations with PCs (p value) Variable PC1 PC2 PC3 BeadChip 1.00 1.00 0.99 Infant gender 0.07 0.52 0.10 Gestational age (weeks) 0.02* 0.74 9.13E-4* Maternal age (years) 0.22 0.17 0.65 Birth weight (grams) 0.25 0.21 1.90E-3* Smoking during pregnancy 0.13 0.38 0.17 Maternal Race 0.86 0.50 0.95 In Vitro Fertilization 0.87 0.20 0.16 Alcohol use during pregnancy 0.21 0.55 0.78 Recreational drug use during pregnancy 0.02* 0.82 0.64 *p<0.05. ! 71! Table 6. Top 50 CpG loci by p value resulting from locus-by-locus analysis for maternal smoking during pregnancy. Illumina CpG P value for association With Maternal Rank Designation Smoking During Pregnancy 1 cg16626670 3.36E-05 2 cg14511156 9.15E-05 3 cg27631817 0.000104959 4 cg24660086 0.000114616 5 cg22233974 0.000126895 6 cg00090147 0.000142274 7 cg27478659 0.000149333 8 cg11263296 0.000158502 9 cg22740835 0.000201507 10 cg22449114 0.000266674 11 cg06784466 0.000341032 12 cg20050113 0.000407446 13 cg21126707 0.000475157 14 cg27398499 0.000615361 15 cg17836145 0.000635855 16 cg18621299 0.000745391 17 cg25033993 0.000759293 18 cg23732024 0.000806806 ! 72! 19 cg24046474 0.000814949 20 cg13504059 0.000828018 21 cg12387247 0.000882135 22 cg15052335 0.000908434 23 cg27050793 0.000931415 24 cg17974185 0.001023562 25 cg21148892 0.001223422 26 cg01248426 0.001260675 27 cg22549408 0.001292275 28 cg05829479 0.001330475 29 cg10894512 0.001389597 30 cg04019407 0.001393156 31 cg04396791 0.00140267 32 cg04076481 0.001469523 33 cg17683775 0.001505767 34 cg24447890 0.001563299 35 cg02210887 0.001582425 36 cg24939733 0.001612411 37 cg22039846 0.001673108 38 cg16220183 0.00168329 39 cg12061236 0.001768373 40 cg07074571 0.001831098 41 cg13269964 0.001911704 ! 73! 42 cg16434306 0.001953222 43 cg16152813 0.001998496 44 cg23752985 0.002022433 45 cg15383087 0.002089883 46 cg15275890 0.002159844 47 cg11074362 0.002197525 48 cg03294491 0.002212847 49 cg01273150 0.002236531 50 cg19592945 0.002254264 ! 74! FIGURES Figure 1. Manhattan plot of –log p values for locus-by-locus analysis of methylation and maternal smoking during pregnancy. Logit-transformed, ComBat-adjusted methylation status of 1,918 CpG loci was significantly (p<0.05) associated with maternal smoking during pregnancy, indicated by points above blue line. ! 75! Figure 2. Smoothed plot across seven RUNX3 CpG loci significantly associated with maternal smoking during pregnancy. Y-axis represents raw ß value. Blue line represents placental methylation in non-smoking mothers; red line represents that of smoking mothers. CpG loci in order of appearance on - strand from 5’ to 3’: 1. cg06037693 2. cg14182690 3. cg24019564 4. cg00117172 5. cg08705994 6. cg00572797 7. cg04757093* (also significantly associated with gestational age). Black arrow indicates transcription start site. Blue P1 and P2 arrows indicate distal and proximal RUNX3 promoters, respectively. Yellow boxes indicate exons. ! 76! Figure 3. Smoking mothers exhibit significant placental hypermethylation at the cg04757093 locus within the RUNX3 gene compared to non-smoking mothers. ! 77! Percent Methylation in Placentas of Non- Percent Methylation in Placentas of Non- Smoking and Smoking Mothers at Smoking and Smoking Mothers at cg06037693 Determined by Pyrosequencing cg00117172 Determined by Pyrosequencing 100 100 Percent Methylation Determined by Pyrosequencing † p<0.09 Percent Methylation Determined by Pyrosequencing 98 80 96 60 † 94 40 92 20 p=0.44 90 0 Non-Smokers Smokers Non-Smokers Smokers n = 11 n = 11 n = 11 n = 11 Percent Methylation in Placentas of Non- Smoking and Smoking Mothers at cg04757093 Determined by Pyrosequencing 100 Percent Methylation Determined by Pyrosequencing 90 * 80 70 60 * p<0.02 50 Non-Smokers Smokers n = 11 n = 11 Figure 4. Pyrosequencing confirmed significant placental hypermethylation at RUNX3 locus cg04757093 (p<0.02) and near-significant placental hypermethylation at RUNX3 locus cg00117172 (p<0.09) in smoking mothers compared to non-smoking mothers. No significant differential methylation was observed at RUNX3 locus cg06037693 between smoking and non-smoking mothers in these 22 placental samples. Mean placental methylation for non-smoking vs smoking mothers: cg06037693, 98 and 97.6, respectively (ranges of 96.3 – 99.5 and 94.3 – 99.7); cg00117172, 12.8 and 21.1, respectively (ranges of 4.5 – 28.3 and 7.7 – 41.3); cg04757093, 64.5 and 69.1, respectively (ranges of 61 – 68 and 60.7 – 79.3). *p<0.05; † p<0.10. ! 78! Figure 5. Proportion of variance (y-axis) in each PC (x-axis) prior to ComBat adjustment. ! 79! Figure 6. The first ten PCs (x-axis) account for ~60% of the variance of the data. Y-axis represents cumulative proportion of variance. ! 80! Figure 7. Plot of the first two PCs, which are significantly associated with BeadChip (batch effect). Figure 8. Proportion of variance (y-axis) in each PC (x-axis) after ComBat adjustment. ! 81! Figure 9. The first ten PCs (x-axis) account for ~50% of the variance of the data after ComBat adjustment. Y-axis represents cumulative proportion of variance. ! 82! Figure 10. Plot of the first two PCs, which are no longer significantly associated with BeadChip after ComBat adjustment. ! 83! CHAPTER 3 Placental DNA Methylation Changes Within the EMID2 Gene Are Associated With Both Infant Toenail Mercury and Adverse Neurobehavioral Outcomes Maccani, Jennifer Z.J.; Koestler, Devin C.; Lester, Barry; Kelsey, Karl T.; and Marsit, Carmen J. (Manuscript in preparation.) ! 84! Placental DNA Methylation Changes Within the EMID2 Gene Are Associated With Both Infant Toenail Mercury and Adverse Neurobehavioral Outcomes Jennifer Z. J. Maccani, B.S.1, Devin C. Koestler, PhD2, Barry Lester, PhD4, Karl T. Kelsey, MD, MOH1,5, and Carmen J. Marsit, PhD2,3 1 Department of Pathology and Laboratory Medicine, Brown University, Providence, Rhode Island, USA 2 Section of Biostatistics & Epidemiology, Department of Community and Family Medicine, and 3 Department of Pharmacology and Toxicology, Geisel School of Medicine at Dartmouth, Hanover, New Hampshire, USA 4 Departments of Psychiatry and Human Behavior and Pediatrics, Brown University, Providence, Rhode Island, USA 5 Department of Epidemiology, Brown University, Providence, Rhode Island, USA Corresponding Author: Carmen J. Marsit, Associate Professor, Departments of Pharmacology & Toxicology and Community and Family Medicine, Geisel School of Medicine at Dartmouth, HB 7650, Pharmacology & Toxicology, Hanover, NH, 03755, USA; phone: 603-650-1825; fax: 603-650-1129; email: Carmen.J.Marsit@Dartmouth.edu RUNNING TITLE: EMID2 methylation, mercury and neurobehavior KEY WORDS: epigenetics, in utero, neurobehavior, methylation, mercury, placenta, pregnancy, EMID2 COMPETING FINANCIAL INTERESTS DECLARATION: The authors declare that there are no competing financial interests. ! 85! ABBREVIATIONS: single nucleotide polymorphism (SNP), mercury (Hg), methyl- mercury (Me-Hg), NICU Network Neurobehavioral Scales (NNNS), False Discovery Rate (FDR), Recursively Partitioned Mixture Modeling (RPMM), transcription start site (TSS), untranslated region (UTR) ! 86! ABSTRACT Background: Prenatal exposure to mercury (Hg) is associated with adverse neurobehavioral outcomes in infants and children. As Hg can interfere with normal placental functioning, as well as cross the placenta and target the fetal brain, prenatal Hg exposure can negatively impact fetal growth and development both directly and indirectly. Objectives: To examine potential associations between placental DNA methylation changes, infant toenail Hg, and neurobehavioral outcomes. Methods: The DNA methylation status of >485,000 CpG loci was interrogated in 192 placental samples using Illumina’s Infinium HumanMethylation450 array platform. Hg data was analyzed in toenail clipping samples from a subset of 41 of these 192 infants, and neurobehavior was assessed using the NICU Network Neurobehavioral Scales (NNNS) in an independent subset of 151 of these infants. Results: 339 CpG loci associated with infant toenail Hg were identified, and then investigated for association with high-risk profile of neurodevelopment based on the NNNS data in an independent sample set. 6 CpG loci associated with both infant toenail Hg and high-risk neurodevelopmental profile were subsequently identified. 5 of these 6 CpG loci reside in the EMID2 gene and were hypomethylated in the placental tissue of high-risk class infants compared to non-high risk class infants. Conclusions: The observed hypomethylation at these EMID2 CpG loci may represent a novel mechanism linking in utero Hg exposure and adverse infant neurobehavioral outcomes. ! 87! INTRODUCTION Multiple studies have found associations between in utero, childhood, or early adulthood mercury (Hg) exposure and later neurologic and psychological impairment. One of the most cited is a study of children in the Faeroe Islands who were predominantly exposed to Hg through a seafood-heavy diet, and were examined for neurobehavioral outcomes at 7 and 14 years of age (Grandjean et al. 1997). Many other groups have also reported findings that support the conclusion that early life exposure to Hg is associated with adverse neurodevelopmental outcomes (Counter and Buchanan 2004; Clarkson 1997), including size of cerebellum (Cace et al. 2011), adverse behavioral outcomes (Gao et al. 2007), central nervous system damage (Choi 1989) and negative effects on psychomotor development in infants (Llop et al. 2012), delays in cognitive development in children (Freire et al. 2010), and other effects which may not appear until later in life (Rice 1996), such as increased susceptibility to Type II diabetes (He et al. 2013). Several of these and other studies (He et al. 2013; Wickre et al. 2004; Xun et al. 2013; Hinners et al. 2012) have assessed levels of fingernail or toenail Hg to provide an integrated estimate of exposure to Hg. The placenta plays a crucial role in regulating fetal growth and development, including neurodevelopment (Lester and Padbury 2009). In utero exposures to environmental toxicants may disrupt normal placental function, impacting the production of placental growth factors and hormones as well as placental detoxification activity (Maccani and Marsit 2009). One mode through which toxicants may interfere with placental function is through epigenetic alterations, including changes to normal placental DNA methylation patterns (Hoyo et al. 2012; Burris et al. 2012; Wilhelm-Benartzi et al. ! 88! 2012; Suter et al. 2010; Suter et al. 2011), which themselves control genes involved in key cellular processes of the placenta. Hg is known to cross the placenta (Council 2000; Yang et al. 1997; Ilback et al. 1991) and interfere with normal placental function (Boadi et al. 1992). Twice the concentration of methyl-mercury (Me-Hg), the predominant form of Hg exposure, has been found in placenta as in maternal blood (Ask et al. 2002). A common source of this in utero Hg exposure is maternal dietary intake of fish, (Davidson et al. 2004), although maternal dental amalgams, which often contain inorganic Hg (Davidson et al. 2004; Takahashi et al. 2001), can also increase levels of Hg in the placenta. A single maternal amalgam restoration has been associated with an increase in placental Hg levels of 3-6 fold (Takahashi et al. 2001). Me-Hg exposure has recently been associated with DNA hypomethylation (Goodrich et al. 2013), suggesting that DNA methylation alterations may play a mechanistic role in producing these adverse neurological and behavioral outcomes. There has been limited examination of the association of Hg exposure to the developing placenta and any of the multiple adverse consequences of exposure to Hg. The placenta is an active organ during development, with alterations in DNA methylation in the placenta having been previously associated with fetal growth and development and neurobehavioral outcomes (Wilhelm-Benartzi et al. 2012; Bromer et al. 2012; Marsit et al. 2012a; Marsit et al. 2012b; Banister et al. 2011; Filiberto et al. 2011). Thus, it is plausible that Hg-associated alterations of the placenta could play a role in mediating neurobehavioral outcomes associated with Hg exposure. Here, we hypothesized that placental DNA methylation changes are associated with infant Hg exposure accessed ! 89! through toenail Hg levels, and that these altered placental DNA methylation patterns are, in turn, associated with adverse infant neurobehavioral outcomes. METHODS Study Design. This study investigated placental methylation in the first 192 placental samples obtained from infants enrolled in the ongoing Rhode Island Child Health Study, a birth cohort of non-pathologic term pregnancies delivered at Women and Infants’ Hospital in Providence, RI, USA. These samples were collected according to protocols and an informed consent process approved by the Institutional Review Boards of Women and Infants’ Hospital and Dartmouth College. Within 2 hours of birth, full thickness biopsy sections were taken from the maternal side of the placenta, 2 cm from the umbilical cord insertion site and free of maternal decidua. These sections were immediately placed in RNAlaterTM (Applied Biosystems, Inc., AM7020). Following at least 72 hours of storage at 4ºC, samples were blotted dry of RNAlater, snap-frozen in liquid nitrogen, pulverized to homogenize the samples and stored at -80ºC until analysis. Toenail clippings are requested from mothers and infants following discharge, and were available for 41 of these 192 infants. Infants enrolled in the study also undergo a newborn neurobehavioral assessment, the NICU Network Neurobehavioral Scales (NNNS) (Lester and Tronick 2004), after the first 24 hours of life but before hospital discharge. In this sample, NNNS data was available for 151 of these 192 infants. DNA extraction and bisulfite modification. Placental DNA was extracted using the QIAmp DNA Mini Kit (Qiagen, Inc., 51304) according to manufacturer’s protocol, and quantified through the use of a NanoDrop ND-1000 spectrophotometer (NanoDrop). ! 90! DNA methylation profiling. Placental DNA methylation was assessed at the University of Minnesota Genomics Center, on the Illumina Infinium HumanMethylation450 BeadArray (Illumina 2013), following sodium bisulfite modification. From methylated (M) and unmethylated (U) allele intensity, β values representing the methylation status at each CpG locus were calculated, where the ratio of fluorescent signals β = Max(M,0)/[Max(M,0) + Max(U,0) + 100] and 0 < β < 1. β values range from 0-1, with complete methylation indicated by values near 1 and absence of methylation indicated by values near 0. Array quality assurance was assessed and poor- performing loci were removed following previously described methods (Banister et al. 2011). X- and Y-linked loci were also removed from analysis, as were all CpG loci within 100bp of a known SNP based on data from the Illumina annotation file. This resulted in a dataset of methylation consisting of 384,474 autosomal, CpG loci from this array for 192 infants. Statistical analysis. A diagram of the analysis strategy is presented in Figure 1. DNA methylation array data was adjusted for the presence of plate-effects by applying the ComBat method (Johnson et al. 2007), shown to perform effectively and efficiently compared to competing batch/plate-adjustment methodologies. The resulting plate- adjusted methylation data were then investigated to ensure that variation in methylation across the array induced by plate-effects had been successfully attenuated using a Principal Components Analysis. As we did not want to assume linear response in the association between toenail Hg levels and placental methylation, as well as to assure that any outlying values were not unduly skewing results, Hg data were categorized as tertiles, which were used for subsequent analysis (Kuan et al. 2010). Potential ! 91! associations between methylation status at each of the 384,474 loci and infant toenail Hg tertile were investigated by fitting a series of independent linear regression models for each of the 384,474 logit-transformed, ComBat-adjusted CpG methylation values using infant toenail Hg as the independent variable in the 41 infants where toenail Hg data was available. An assessment of false discovery rate (Benjamini 1995) was used to control for Type I error, and only loci exhibiting differences in the methylation β value>0.125 between any tertiles, and a q-value<0.2 were selected for subsequent analysis. In order limit the number of tests performed when comparing between loci associated with both infant toenail Hg and methylation and NNNS outcomes, we used a latent profile analysis similar to that previously described for use on NNNS scores (Liu et al. 2010) based on recursively partitioned mixture models to define profiles or classes of neurobehavior based on the 13 summary scores (Appleton et al. 2013). For the loci identified to be associated with Hg, we fit a series of logistic regression models with high-risk neurobehavioral class as the dependent variable and methylation at each of the selected loci as the independent variable, controlled for the same potential confounders as above, namely, maternal age, birth weight percentile, delivery method, and infant gender. Control of Type I error was again performed using an assessment of false discovery rate (Benjamini 1995), and loci were selected for presentation with q values <0.2. RESULTS Table 1 describes the study population demographics and compares them between the two subsets of samples examined. Overall, by design, all infants were born at 37 weeks or greater gestation, and there is oversampling for small and large for ! 92! gestational age infants. The two subsets did not differ by maternal age, infant gender, infant birth weight or gestation time. There were no reported smokers among the mothers of infants used in the Hg analyses; among the infants used in the NNNS analysis, there was a greater proportion of non-Caucasian mothers and a greater proportion of infants born through Caesarean section. Using the available methylation array data, mean methylation β values were calculated for tertiles of toenail Hg level, with the low, or referent, infant toenail Hg tertile ranging from 0.005 µg Hg per gram of toenail to 0.031; the medium tertile from 0.032 to 0.076; and the high tertile from 0.077 to 0.425. The delta-β values were calculated for each CpG locus by taking the difference between the mean β values between each of the pairs of tertiles. To limit further analysis to CpG loci with potentially more biological relevance, CpG loci with delta-β values >0.125, were included in the analysis to examine associations between methylation and Hg exposure. From these analyses, 339 CpG loci were identified at q<0.2, for an association between placental methylation and toenail Hg level. A table of these 339 loci, and their mean methylation β values by toenail Hg tertile are given in Table 2. 79 loci exhibited a positive and 34 loci a negative relationship across the 3 tertiles of Hg exposure level, while 226 loci exhibited a non-monotonic relationship. The resulting 339 CpG loci associated with infant toenail Hg were then assessed for their association with neurobehavioral outcomes in an independent set of 151 placenta samples with placenta DNA methylation data available. These infants were also subjected to the NICU Network Neurobehavioral Scales (NNNS) assessment during their newborn hospital stay. We used a latent profiling strategy as in Liu et al (2010) to classify these ! 93! infants based on the similarity of their NNNS scores. This resulted in the identification of 7 classes of NNNS behavioral profile, and further dichotomized the sample comparing a “high-risk class” which had the same characteristics as the high risk class described in Liu et al (Liu et al. 2010) to all other infants. Descriptive statistics on the individual NNNS summary scores between the two groups is provided in Table 3. Infants in this high-risk neurobehavioral profile exhibited hyperarousal and increased excitability, with reduced self-regulation and poorer lethargy scores. From the 339 CpG loci associated with infant toenail Hg levels, we identified 6 CpG loci that were associated (q<0.2) with high-risk NNNS profile, and these results are described in Table 4. Five of these six loci were found to be associated with the EMID2 gene. These five EMID2 CpG loci were located within the same CpG island region of the gene. Four of the five loci were located within 200 base pairs of the EMID2 transcription start site (TSS) on the + strand, and one CpG locus (cg13267931) is located within the 5’ untranslated region (UTR) of EMID2 just upstream of the first exon. The locations of these five CpG loci within the EMID2 gene and their placental methylation by toenail Hg tertile are illustrated in Figure 2; the average placental methylation ß values across these five EMID2 CpG loci by NNNS profile is illustrated in Figure 3. All five EMID2 CpG loci identified were hypomethylated in the placental tissue of infants exhibiting the high- risk profile compared to infants with low-risk profiles. The sixth CpG locus associated with both infant toenail Hg and high-risk neurobehavioral profile is located on chromosome 14 within the T-cell receptor alpha locus gene. ! 94! DISCUSSION In this analysis, we observed altered patterns of DNA methylation in the placenta that are significantly associated with both infant toenail Hg and high-risk infant neurobehavioral profile. The 339 CpG loci significantly associated with infant toenail Hg (see Table 2) reside in genes involved in diverse biological functions and aspects of development, and some reside in genes associated with disorders including type 2 diabetes (ZBED3) (Ohshige et al. 2011), asthma (EMID2) (Pasaje et al. 2012; Pasaje et al. 2011), and many types of cancer (FBXO3, HOOK2, MT2A, EIF3E, RPH3AL, PTRF, MT1M, STK32A) (Cha et al. 2011; Shimada et al. 2005; Krzeslak et al. 2013; Zhou et al. 2012; SL Liu et al. 2012; YP Liu et al. 2012; Lee et al. 2009; Ji et al. 2006; Mao et al. 2012). However, a large number of these CpG loci reside in genes involved in neurodevelopment (DIXDC1, NRBP2, KIF26B, FEZF1, DMRTA2, ACTN1, MYO10, LRFN1) (Singh et al. 2010; Kivimae et al. 2011; Larsson et al. 2008; Heinrich et al. 2012; Eckler et al. 2011; Shimizu and Hibi 2009; Watanabe et al. 2009; Shimizu et al. 2010; Yoshizawa et al. 2011; Konno et al. 2012; Kremerskothen et al. 2002; Silver et al. 2012; Ju et al. 2013; Raines et al. 2012; Yu et al. 2012; Morimura et al. 2006), neurobehavior (CPLX1, LMX1B, ADD2) (Drew et al. 2007; Glynn et al. 2007; Barreto- Valer et al. 2013; Porro et al. 2010), and neurological disorders, including schizophrenia (DIXDC1, ARVCF, MAGI2, ZIC2) (Bradshaw and Porteous 2012; Sim et al. 2012; Mas et al. 2010; Mas et al. 2009; Chen et al. 2005), ADHD (TCERG1L) (Neale et al. 2010; Karlsson et al. 2012; Hatayama et al. 2011), movement disorders (NOL3, TP53INP2) (Russell et al. 2012; Bennetts et al. 2007), Creutzfeldt-Jakob disease (CHN2) (Mead et al. ! 95! 2012), Huntington’s disease (H2AFY2, AGPAT1) (Hu et al. 2011; Cong et al. 2012), Parkinson’s disease (LMX1B) (Tian et al. 2012), and autism (PLXNA4, WNT2) (Suda et al. 2011; Lin et al. 2012; Kalkman 2012). Due to the number of genes involved in neurological or neuropathologic processes and as there are well defined links between Hg exposure and neurodevelopmental deficits, we further examined these identified loci for their association with newborn neurobehavioral profiles defined by the NNNS (Lester and Tronick 2004). We used a latent profile methodology to reduce the dimensionality of the data, and as Liu et al reported, such profiles can be associated with later childhood outcomes including acute medical and behavior problems, school readiness and IQ through 4 ½ years of age (Liu et al. 2010). In this analysis, we identified six loci that were associated infant toenail Hg and were associated with a risk profile similar to that in Liu et al, and five of the six CpG loci reside in the EMID2 gene. The sixth CpG locus is located on chromosome 14 within the T-cell receptor alpha locus gene, in an enhancer region. Although the function of EMID2 in the placenta has not yet been studied, EMID2 has been associated in the literature with asthma (Pasaje et al. 2012; Pasaje et al. 2011), as well as a long-range cis-regulatory mutation of the sonic hedgehog (SHH) gene resulting in its enhancer adoption, such that ectopic expression of SHH is driven by an enhancer not located within the SHH gene but instead within an intronic region of the EMID2 gene (Lettice et al. 2011). Although these loci are not located within this specific intronic region of EMID2, and roles in enhancer adoption in either SHH or other genes have not yet been observed in the literature for these five CpG loci, it is possible that ! 96! DNA methylation alterations to these loci could have consequences for expression changes, either in the EMID2 gene or a different gene, that have yet to be discovered. Future investigation is warranted, particularly due to the associations in the literature between SHH and neurodevelopment, including neural tube patterning, proliferation of neural stem cells, and neural cell survival (Herrmann et al. 2008; McCarthy and Argraves 2003; Ho and Scott 2002). Interestingly, we observed that it was hypomethylation of this gene associated with this potential risk neurobehavioral profile. Hypomethylation occurred at these loci in the low to medium tertiles of Hg exposure, with the most extreme hypomethylation occurring in the mid-range exposure group. This would suggest a non-monotonic relationship with exposure and a potentially complicated relationship between exposure, DNA methylation alteration, and outcome. We were limited in this study in our ability to specifically address these relationships in the same individuals, and so we urge caution in this interpretation until these results can be expanded and validated in a larger sample population. Limitations to this study include using placental tissue only from term infants, as well as the study’s relatively small sample size. Here, 41 placental samples were analyzed for associations between DNA methylation changes and infant toenail Hg, and 151 placental samples were analyzed for associations between these Hg-associated CpG loci and high-risk neurobehavioral profile. This represents one of the largest studies, to date, to link epigenetic alterations in placenta to either Hg exposure or neurodevelopmental outcome. Still, it is important that these findings be replicated in additional studies, and that the true function of methylation of the identified loci ! 97! including those in EMID2 in the placenta be elucidated in more focused, mechanistic experimental systems. This study provides evidence for a potential role for placental epigenetic alteration as a potential mechanism linking Hg exposure and adverse infant neurodevelopment. It also suggests a possible role for the EMID2 gene and its protein product in the placenta, or potentially an exposure-associated role for enhancer adoption of SHH or another gene. Overall, these results highlight an association between in utero exposure to mercury, placental DNA methylation changes and the developmental origins of health and disease. ! 98! ACKNOWLEDGMENTS AND FUNDING: This research was funded by grants R01MH094609 from NIH-NIMH and T32ES007272 from the NIH-NIEHS. ! 99! REFERENCES 1. Grandjean P, Weihe P, White RF, Debes F, Araki S, Yokoyama K, et al. 1997. Cognitive deficit in 7-year-old children with prenatal exposure to methylmercury. Neurotoxicology and teratology 19(6): 417-428. 2. Counter SA, Buchanan LH. 2004. Mercury exposure in children: a review. Toxicology and applied pharmacology 198(2): 209-230. 3. Clarkson TW. 1997. The toxicology of mercury. Critical reviews in clinical laboratory sciences 34(4): 369-403. 4. Cace IB, Milardovic A, Prpic I, Krajina R, Petrovic O, Vukelic P, et al. 2011. Relationship between the prenatal exposure to low-level of mercury and the size of a newborn's cerebellum. Med Hypotheses 76(4): 514-516. 5. Gao Y, Yan CH, Tian Y, Wang Y, Xie HF, Zhou X, et al. 2007. Prenatal exposure to mercury and neurobehavioral development of neonates in Zhoushan City, China. Environmental research 105(3): 390-399. 6. Choi BH. 1989. The effects of methylmercury on the developing brain. Progress in neurobiology 32(6): 447-470. 7. Llop S, Guxens M, Murcia M, Lertxundi A, Ramon R, Riano I, et al. 2012. Prenatal exposure to mercury and infant neurodevelopment in a multicenter cohort in Spain: study of potential modifiers. Am J Epidemiol 175(5): 451-465. 8. Freire C, Ramos R, Lopez-Espinosa MJ, Diez S, Vioque J, Ballester F, et al. 2010. Hair mercury levels, fish consumption, and cognitive development in preschool children from Granada, Spain. Environmental research 110(1): 96-104. ! 100! 9. Rice DC. 1996. Evidence for delayed neurotoxicity produced by methylmercury. Neurotoxicology 17(3-4): 583-596. 10. He K, Xun P, Liu K, Morris S, Reis J, Guallar E. 2013. Mercury Exposure in Young Adulthood and Incidence of Diabetes Later in Life: The CARDIA trace element study. Diabetes care. 11. Wickre JB, Folt CL, Sturup S, Karagas MR. 2004. Environmental exposure and fingernail analysis of arsenic and mercury in children and adults in a Nicaraguan gold mining community. Archives of environmental health 59(8): 400-409. 12. Xun P, Liu K, Morris JS, Jordan JM, He K. 2013. Distributions and determinants of mercury concentrations in toenails among American young adults: the CARDIA Trace Element Study. Environmental science and pollution research international 20(3): 1423-1430. 13. Hinners T, Tsuchiya A, Stern AH, Burbacher TM, Faustman EM, Marien K. 2012. Chronologically matched toenail-Hg to hair-Hg ratio: temporal analysis within the Japanese community (U.S.). Environmental health : a global access science source 11: 81. 14. Lester BM, Padbury JF. 2009. Third pathophysiology of prenatal cocaine exposure. Developmental neuroscience 31(1-2): 23-35. 15. Maccani MA, Marsit CJ. 2009. Epigenetics in the placenta. Am J Reprod Immunol 62(2): 78-89. 16. Hoyo C, Fortner K, Murtha AP, Schildkraut JM, Soubry A, Demark-Wahnefried W, et al. 2012. Association of cord blood methylation fractions at imprinted ! 101! insulin-like growth factor 2 (IGF2), plasma IGF2, and birth weight. Cancer causes & control : CCC 23(4): 635-645. 17. Burris HH, Rifas-Shiman SL, Baccarelli A, Tarantini L, Boeke CE, Kleinman K, et al. 2012. Associations of LINE-1 DNA Methylation with Preterm Birth in a Prospective Cohort Study. Journal of developmental origins of health and disease 3(3): 173-181. 18. Wilhelm-Benartzi CS, Houseman EA, Maccani MA, Poage GM, Koestler DC, Langevin SM, et al. 2012. In utero exposures, infant growth, and DNA methylation of repetitive elements and developmentally related genes in human placenta. Environmental health perspectives 120(2): 296-302. 19. Suter M, Abramovici A, Showalter L, Hu M, Shope CD, Varner M, et al. 2010. In utero tobacco exposure epigenetically modifies placental CYP1A1 expression. Metabolism: clinical and experimental 59(10): 1481-1490. 20. Suter M, Ma J, Harris A, Patterson L, Brown KA, Shope C, et al. 2011. Maternal tobacco use modestly alters correlated epigenome-wide placental DNA methylation and gene expression. Epigenetics : official journal of the DNA Methylation Society 6(11): 1284-1294. 21. Council NR. Toxicological Effects of Methylmercury., 2000. Washington, DC. National Academy Press. 22. Yang J, Jiang Z, Wang Y, Qureshi IA, Wu XD. 1997. Maternal-fetal transfer of metallic mercury via the placenta and milk. Annals of clinical and laboratory science 27(2): 135-141. ! 102! 23. Ilback NG, Sundberg J, Oskarsson A. 1991. Methyl mercury exposure via placenta and milk impairs natural killer (NK) cell function in newborn rats. Toxicology letters 58(2): 149-158. 24. Boadi WY, Urbach J, Brandes JM, Yannai S. 1992. In vitro exposure to mercury and cadmium alters term human placental membrane fluidity. Toxicology and applied pharmacology 116(1): 17-23. 25. Ask K, Akesson A, Berglund M, Vahter M. 2002. Inorganic mercury and methylmercury in placentas of Swedish women. Environmental health perspectives 110(5): 523-526. 26. Davidson PW, Myers GJ, Weiss B. 2004. Mercury exposure and child development outcomes. Pediatrics 113(4 Suppl): 1023-1029. 27. Takahashi Y, Tsuruta S, Hasegawa J, Kameyama Y, Yoshida M. 2001. Release of mercury from dental amalgam fillings in pregnant rats and distribution of mercury in maternal and fetal tissues. Toxicology 163(2-3): 115-126. 28. Goodrich JM, Basu N, Franzblau A, Dolinoy DC. 2013. Mercury biomarkers and DNA methylation among Michigan dental professionals. Environmental and molecular mutagenesis 54(3): 195-203. 29. Bromer C, Marsit CJ, Armstrong DA, Padbury JF, Lester B. 2012. Genetic and epigenetic variation of the glucocorticoid receptor (NR3C1) in placenta and infant neurobehavior. Developmental psychobiology. 30. Marsit CJ, Maccani MA, Padbury JF, Lester BM. 2012a. Placental 11-beta hydroxysteroid dehydrogenase methylation is associated with newborn growth and a measure of neurobehavioral outcome. PLoS One 7(3): e33794. ! 103! 31. Marsit CJ, Lambertini L, Maccani MA, Koestler DC, Houseman EA, Padbury JF, et al. 2012b. Placenta-imprinted gene expression association of infant neurobehavior. The Journal of pediatrics 160(5): 854-860 e852. 32. Banister CE, Koestler DC, Maccani MA, Padbury JF, Houseman EA, Marsit CJ. 2011. Infant growth restriction is associated with distinct patterns of DNA methylation in human placentas. Epigenetics : official journal of the DNA Methylation Society 6(7): 920-927. 33. Filiberto AC, Maccani MA, Koestler D, Wilhelm-Benartzi C, Avissar-Whiting M, Banister CE, et al. 2011. Birthweight is associated with DNA promoter methylation of the glucocorticoid receptor in human placenta. Epigenetics : official journal of the DNA Methylation Society 6(5): 566-572. 34. Lester BM, Tronick EZ. 2004. History and description of the Neonatal Intensive Care Unit Network Neurobehavioral Scale. Pediatrics 113(3 Pt 2): 634-640. 35. Illumina. 2013. Illumina HumanMethylation450 BeadChip Kit. Available: http://www.illumina.com/products/methylation_450_beadchip_kits.ilmn [accessed 2013 Mar 7 2013]. 36. Johnson WE, Li C, Rabinovic A. 2007. Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics 8(1): 118-127. 37. Kuan PF, Wang S, Zhou X, Chu H. 2010. A statistical framework for Illumina DNA methylation arrays. Bioinformatics 26(22): 2849-2855. 38. Benjamini YH, Yosef. 1995. Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society, Series B (Methodological) 57(1): 12. ! 104! 39. Liu J, Bann C, Lester B, Tronick E, Das A, Lagasse L, et al. 2010. Neonatal neurobehavior predicts medical and behavioral outcome. Pediatrics 125(1): e90- 98. 40. Appleton AA, Koestler DC, Murphy M, Lester BM, Marsit CJ. 2013. Profiles of Neurobehavioral Development in a Healthy Term Cohort. Submitted. 41. Ohshige T, Iwata M, Omori S, Tanaka Y, Hirose H, Kaku K, et al. 2011. Association of new loci identified in European genome-wide association studies with susceptibility to type 2 diabetes in the Japanese. PLoS One 6(10): e26911. 42. Pasaje CF, Bae JS, Park BL, Cheong HS, Kim JH, Jang AS, et al. 2012. Possible role of EMID2 on nasal polyps pathogenesis in Korean asthma patients. BMC medical genetics 13: 2. 43. Pasaje CF, Kim JH, Park BL, Cheong HS, Kim MK, Choi IS, et al. 2011. A possible association of EMID2 polymorphisms with aspirin hypersensitivity in asthma. Immunogenetics 63(1): 13-21. 44. Cha JD, Kim HJ, Cha IH. 2011. Genetic alterations in oral squamous cell carcinoma progression detected by combining array-based comparative genomic hybridization and multiplex ligation-dependent probe amplification. Oral surgery, oral medicine, oral pathology, oral radiology, and endodontics 111(5): 594-607. 45. Shimada H, Nakashima K, Ochiai T, Nabeya Y, Takiguchi M, Nomura F, et al. 2005. Serological identification of tumor antigens of esophageal squamous cell carcinoma. International journal of oncology 26(1): 77-86. 46. Krzeslak A, Forma E, Chwatko G, Jozwiak P, Szymczyk A, Wilkosz J, et al. 2013. Effect of metallothionein 2A gene polymorphism on allele-specific gene ! 105! expression and metal content in prostate cancer. Toxicology and applied pharmacology. 47. Zhou CF, Ji J, Yuan F, Yu YY, Liu BY, Zhang J, et al. 2012. [Expression of metastasis associated 1 family member 2 (MTA2) in gastric cancer and its relationship with transcription factor Sp1]. Zhonghua Zhong Liu Za Zhi 34(8): 592-595. 48. Liu SL, Han Y, Zhang Y, Xie CY, Wang EH, Miao Y, et al. 2012. Expression of metastasis-associated protein 2 (MTA2) might predict proliferation in non-small cell lung cancer. Targeted oncology 7(2): 135-143. 49. Liu YP, Shan BE, Wang XL, Ma L. 2012. Correlation between MTA2 overexpression and tumour progression in esophageal squamous cell carcinoma. Experimental and therapeutic medicine 3(4): 745-749. 50. Lee H, Ryu SH, Hong SS, Seo DD, Min HJ, Jang MK, et al. 2009. Overexpression of metastasis-associated protein 2 is associated with hepatocellular carcinoma size and differentiation. Journal of gastroenterology and hepatology 24(8): 1445-1450. 51. Ji Y, Zhang P, Lu Y, Ma D. 2006. Expression of MTA2 gene in ovarian epithelial cancer and its clinical implication. Journal of Huazhong University of Science and Technology Medical sciences = Hua zhong ke ji da xue xue bao Yi xue Ying De wen ban = Huazhong keji daxue xuebao Yixue Yingdewen ban 26(3): 359-362. 52. Mao J, Yu H, Wang C, Sun L, Jiang W, Zhang P, et al. 2012. Metallothionein MT1M is a tumor suppressor of human hepatocellular carcinomas. Carcinogenesis 33(12): 2568-2577. ! 106! 53. Singh KK, Ge X, Mao Y, Drane L, Meletis K, Samuels BA, et al. 2010. Dixdc1 is a critical regulator of DISC1 and embryonic cortical development. Neuron 67(1): 33-48. 54. Kivimae S, Martin PM, Kapfhamer D, Ruan Y, Heberlein U, Rubenstein JL, et al. 2011. Abnormal behavior in mice mutant for the Disc1 binding partner, Dixdc1. Translational psychiatry 1: e43. 55. Larsson J, Forsberg M, Brannvall K, Zhang XQ, Enarsson M, Hedborg F, et al. 2008. Nuclear receptor binding protein 2 is induced during neural progenitor differentiation and affects cell survival. Molecular and cellular neurosciences 39(1): 32-39. 56. Heinrich J, Proepper C, Schmidt T, Linta L, Liebau S, Boeckers TM. 2012. The postsynaptic density protein Abelson interactor protein 1 interacts with the motor protein Kinesin family member 26B in hippocampal neurons. Neuroscience 221: 86-95. 57. Eckler MJ, McKenna WL, Taghvaei S, McConnell SK, Chen B. 2011. Fezf1 and Fezf2 are required for olfactory development and sensory neuron identity. The Journal of comparative neurology 519(10): 1829-1846. 58. Shimizu T, Hibi M. 2009. Formation and patterning of the forebrain and olfactory system by zinc-finger genes Fezf1 and Fezf2. Development, growth & differentiation 51(3): 221-231. 59. Watanabe Y, Inoue K, Okuyama-Yamamoto A, Nakai N, Nakatani J, Nibu K, et al. 2009. Fezf1 is required for penetration of the basal lamina by olfactory axons ! 107! to promote olfactory development. The Journal of comparative neurology 515(5): 565-584. 60. Shimizu T, Nakazawa M, Kani S, Bae YK, Shimizu T, Kageyama R, et al. 2010. Zinc finger genes Fezf1 and Fezf2 control neuronal differentiation by repressing Hes5 expression in the forebrain. Development 137(11): 1875-1885. 61. Yoshizawa A, Nakahara Y, Izawa T, Ishitani T, Tsutsumi M, Kuroiwa A, et al. 2011. Zebrafish Dmrta2 regulates neurogenesis in the telencephalon. Genes to cells : devoted to molecular & cellular mechanisms 16(11): 1097-1109. 62. Konno D, Iwashita M, Satoh Y, Momiyama A, Abe T, Kiyonari H, et al. 2012. The mammalian DM domain transcription factor Dmrta2 is required for early embryonic development of the cerebral cortex. PLoS One 7(10): e46577. 63. Kremerskothen J, Teber I, Wendholt D, Liedtke T, Bockers TM, Barnekow A. 2002. Brain-specific splicing of alpha-actinin 1 (ACTN1) mRNA. Biochemical and biophysical research communications 295(3): 678-681. 64. Silver M, Janousova E, Hua X, Thompson PM, Montana G. 2012. Identification of gene pathways implicated in Alzheimer's disease using longitudinal imaging phenotypes with sparse regression. NeuroImage 63(3): 1681-1694. 65. Ju XD, Guo Y, Wang NN, Huang Y, Lai MM, Zhai YH, et al. 2013. Both Myosin-10 Isoforms Are Required for Radial Neuronal Migration in the Developing Cerebral Cortex. Cereb Cortex. 66. Raines AN, Nagdas S, Kerber ML, Cheney RE. 2012. Headless Myo10 is a negative regulator of full-length Myo10 and inhibits axon outgrowth in cortical neurons. The Journal of biological chemistry 287(30): 24873-24883. ! 108! 67. Yu H, Wang N, Ju X, Yang Y, Sun D, Lai M, et al. 2012. PtdIns (3,4,5) P3 recruitment of Myo10 is essential for axon development. PLoS One 7(5): e36988. 68. Morimura N, Inoue T, Katayama K, Aruga J. 2006. Comparative analysis of structure, expression and PSD95-binding capacity of Lrfn, a novel family of neuronal transmembrane proteins. Gene 380(2): 72-83. 69. Drew CJ, Kyd RJ, Morton AJ. 2007. Complexin 1 knockout mice exhibit marked deficits in social behaviours but appear to be cognitively normal. Human molecular genetics 16(19): 2288-2305. 70. Glynn D, Sizemore RJ, Morton AJ. 2007. Early motor development is abnormal in complexin 1 knockout mice. Neurobiology of disease 25(3): 483-495. 71. Barreto-Valer K, Lopez-Bellido R, Rodriguez RE. 2013. Cocaine modulates the expression of transcription factors related to the dopaminergic system in zebrafish. Neuroscience 231: 258-271. 72. Porro F, Rosato-Siri M, Leone E, Costessi L, Iaconcig A, Tongiorgi E, et al. 2010. beta-adducin (Add2) KO mice show synaptic plasticity, motor coordination and behavioral deficits accompanied by changes in the expression and phosphorylation levels of the alpha- and gamma-adducin subunits. Genes, brain, and behavior 9(1): 84-96. 73. Bradshaw NJ, Porteous DJ. 2012. DISC1-binding proteins in neural development, signalling and schizophrenia. Neuropharmacology 62(3): 1230-1241. 74. Sim K, Chan WY, Woon PS, Low HQ, Lim L, Yang GL, et al. 2012. ARVCF genetic influences on neurocognitive and neuroanatomical intermediate ! 109! phenotypes in Chinese patients with schizophrenia. The Journal of clinical psychiatry 73(3): 320-326. 75. Mas S, Bernardo M, Gasso P, Alvarez S, Garcia-Rizo C, Bioque M, et al. 2010. A functional variant provided further evidence for the association of ARVCF with schizophrenia. American journal of medical genetics Part B, Neuropsychiatric genetics : the official publication of the International Society of Psychiatric Genetics 153B(5): 1052-1059. 76. Mas S, Bernardo M, Parellada E, Garcia-Rizo C, Gasso P, Alvarez S, et al. 2009. ARVCF single marker and haplotypic association with schizophrenia. Progress in neuro-psychopharmacology & biological psychiatry 33(6): 1064-1069. 77. Chen HY, Yeh JI, Hong CJ, Chen CH. 2005. Mutation analysis of ARVCF gene on chromosome 22q11 as a candidate for a schizophrenia gene. Schizophrenia research 72(2-3): 275-277. 78. Neale BM, Medland S, Ripke S, Anney RJ, Asherson P, Buitelaar J, et al. 2010. Case-control genome-wide association study of attention-deficit/hyperactivity disorder. Journal of the American Academy of Child and Adolescent Psychiatry 49(9): 906-920. 79. Karlsson R, Graae L, Lekman M, Wang D, Favis R, Axelsson T, et al. 2012. MAGI1 copy number variation in bipolar affective disorder and schizophrenia. Biological psychiatry 71(10): 922-930. 80. Hatayama M, Ishiguro A, Iwayama Y, Takashima N, Sakoori K, Toyota T, et al. 2011. Zic2 hypomorphic mutant mice as a schizophrenia model and ZIC2 mutations identified in schizophrenia patients. Scientific reports 1: 16. ! 110! 81. Russell JF, Steckley JL, Coppola G, Hahn AF, Howard MA, Kornberg Z, et al. 2012. Familial cortical myoclonus with a mutation in NOL3. Annals of neurology 72(2): 175-183. 82. Bennetts JS, Rendtorff ND, Simpson F, Tranebjaerg L, Wicking C. 2007. The coding region of TP53INP2, a gene expressed in the developing nervous system, is not altered in a family with autosomal recessive non-progressive infantile ataxia on chromosome 20q11-q13. Developmental dynamics : an official publication of the American Association of Anatomists 236(3): 843-852. 83. Mead S, Uphill J, Beck J, Poulter M, Campbell T, Lowe J, et al. 2012. Genome- wide association study in multiple human prion diseases suggests genetic risk factors additional to PRNP. Human molecular genetics 21(8): 1897-1906. 84. Hu Y, Chopra V, Chopra R, Locascio JJ, Liao Z, Ding H, et al. 2011. Transcriptional modulator H2A histone family, member Y (H2AFY) marks Huntington disease activity in man and mouse. Proceedings of the National Academy of Sciences of the United States of America 108(41): 17141-17146. 85. Cong WN, Cai H, Wang R, Daimon CM, Maudsley S, Raber K, et al. 2012. Altered hypothalamic protein expression in a rat model of Huntington's disease. PLoS One 7(10): e47240. 86. Tian LP, Zhang S, Zhang YJ, Ding JQ, Chen SD. 2012. Lmx1b can promote the differentiation of embryonic stem cells to dopaminergic neurons associated with Parkinson's disease. Biotechnology letters 34(7): 1167-1174. ! 111! 87. Suda S, Iwata K, Shimmura C, Kameno Y, Anitha A, Thanseem I, et al. 2011. Decreased expression of axon-guidance receptors in the anterior cingulate cortex in autism. Molecular autism 2(1): 14. 88. Lin PI, Chien YL, Wu YY, Chen CH, Gau SS, Huang YS, et al. 2012. The WNT2 gene polymorphism associated with speech delay inherent to autism. Research in developmental disabilities 33(5): 1533-1540. 89. Kalkman HO. 2012. A review of the evidence for the canonical Wnt pathway in autism spectrum disorders. Molecular autism 3(1): 10. 90. Lettice LA, Daniels S, Sweeney E, Venkataraman S, Devenney PS, Gautier P, et al. 2011. Enhancer-adoption as a mechanism of human developmental disease. Human mutation 32(12): 1492-1499. 91. Herrmann M, King K, Weitzman M. 2008. Prenatal tobacco smoke and postnatal secondhand smoke exposure and child neurodevelopment. Current opinion in pediatrics 20(2): 184-190. 92. McCarthy RA, Argraves WS. 2003. Megalin and the neurodevelopmental biology of sonic hedgehog and retinol. Journal of cell science 116(Pt 6): 955-960. 93. Ho KS, Scott MP. 2002. Sonic hedgehog in the nervous system: functions, modifications and mechanisms. Current opinion in neurobiology 12(1): 57-63. ! 112! TABLES Table 1. Study population demographics. SUBSET 1 SUBSET 2 Infants With Toenail Infants With NNNS Hg Data Data Total, n (%) 41 (100%) 151 (100%) Infant gender Male, n (%) 20 (49%) 74 (49%) Female, n (%) 21 (51%) 77 (51%) Maternal age (years) Mean (±SD) 32 ± 3.6 28.3 ± 6.0 Median (Range) 32 (23 – 39) 29 (18 – 40) Tobacco use during pregnancy† Yes, n (%) 0 (0%) 9 (6%) No, n (%) 40 (98%) 142 (94%) Birth weight (grams) Mean (±SD) 3591.3 ± 628.5 3460.9 ± 739.6 Median (Range) 3590 (2160 – 4530) 3415 (1705 – 5465) Gestational age (weeks) Mean (±SD) 39.3 ± 1.1 39.3 ± 1.1 Median (Range) 39.0 (37.0 – 41.0) 39.3 (37.0 – 41.9) ! 113! Maternal race Caucasian, n (%) 37 (90.2%) 109 (72.2%) Non-Caucasian, n (%) 4 (9.8%) 42 (27.8%) C-section Yes, n (%) 14 (34.1%) 73 (48.3%) No, n (%) 27 (65.9%) 78 (51.7%) Recreational drug use during pregnancy Yes, n (%) 1 (2.4%) 4 (2.6%) No, n (%) 40 (97.6%) 147 (97.4%) † 1 sample with infant toenail Hg data is missing tobacco smoking during pregnancy data. ! 114! Table 2. Mean ß values and q values of 339 CpG loci associated with infant toenail Hg. Mean ß Mean ß q Value for Value for Value for Mean ß Association Low Infant Medium Value for With Hg in Illumina Illumina Toenail Hg Infant High infant Infant Gene CpG Tertile Toenail Hg Toenail Hg Toenail Symbol Rank Designation (Referent) Tertile Tertile Clippings Annotation 1 cg09415955 0.631092544 0.532473108 0.504205451 0.000485514 CASZ1 2 cg00054702 0.187934009 0.379224176 0.344347811 0.001080475 LOC389332 3 cg02124514 0.171872828 0.359782728 0.334838473 0.001080475 LOC389332 4 cg06680397 0.112729002 0.275140248 0.25500808 0.001080475 LOC389332 5 cg09142848 0.460736818 0.599009598 0.544544235 0.001080475 6 cg10454766 0.257016803 0.396607773 0.385508956 0.001080475 C11orf87 7 cg12195446 0.621778853 0.424271114 0.854316061 0.001080475 IRS2 8 cg21158476 0.106260066 0.249864563 0.222615542 0.001080475 LOC389332 9 cg21852408 0.370261182 0.513649667 0.498733556 0.001080475 C11orf87 10 cg05430997 0.602683151 0.556057613 0.694593254 0.001080475 11 cg06430688 0.677788664 0.695877789 0.482643148 0.001080475 12 cg14048874 0.226422851 0.200159608 0.341625397 0.001080475 EMID2 13 cg19190016 0.417554438 0.356238118 0.28986407 0.001080475 ! 115! 14 cg19497031 0.061049321 0.129769598 0.200703502 0.001080475 POU4F1 15 cg23119063 0.772592688 0.7012043 0.626832668 0.001080475 16 cg22728830 0.489660746 0.447719561 0.349867843 0.001082624 17 cg00146240 0.762412459 0.481193224 0.584656974 0.001099849 18 cg04861929 0.283119597 0.396211119 0.411816755 0.001201267 C11orf87 19 cg05704942 0.203260577 0.428814655 0.394259616 0.001201648 MRTO4 20 cg05956126 0.126016211 0.271156162 0.251197874 0.001201648 LOC389332 21 cg07374961 0.475048454 0.622495283 0.542942419 0.001201648 22 cg21422164 0.446498562 0.590602893 0.51485988 0.001201648 RASA3 23 cg13267931 0.2601824 0.247476811 0.381779715 0.001201648 EMID2 24 cg02511231 0.405638104 0.310024903 0.276698557 0.001201648 TLX3 TP53TG1; 25 cg02673015 0.22326221 0.239038834 0.352706778 0.001201648 CROT 26 cg07180307 0.444956378 0.567815342 0.574470698 0.001201648 C11orf87 C11orf87; 27 cg06719900 0.27539615 0.43315318 0.451498839 0.001242325 C11orf87 28 cg15892115 0.263769683 0.262198511 0.410239683 0.001242325 ONECUT2 29 cg08768621 0.347674078 0.330591316 0.457428964 0.00129876 ONECUT2 30 cg09696939 0.339892717 0.262123503 0.210959792 0.00129876 BICC1 ! 116! TP53TG1; 31 cg12969170 0.166176068 0.193552656 0.301581346 0.00129876 CROT 32 cg12288267 0.545279202 0.591974614 0.43197279 0.001307687 33 cg05986417 0.495548637 0.668035257 0.592013164 0.001365141 TP53TG1; 34 cg16063666 0.223366267 0.245549829 0.405259014 0.001424856 CROT 35 cg05454501 0.078148766 0.215003335 0.265749692 0.001427794 ZBED3 36 cg13965724 0.031211688 0.171382974 0.073868964 0.001427794 P4HTM 37 cg09856068 0.37761381 0.556959742 0.484729286 0.001482542 FGF18 38 cg24642825 0.170846351 0.364264807 0.231822865 0.001482542 39 cg03494795 0.22635923 0.12235401 0.268819301 0.001482542 ABL1 40 cg11464064 0.192745706 0.095881695 0.238016807 0.001482542 ABL1 41 cg03517284 0.285239937 0.36704156 0.41980724 0.001482542 42 cg27179533 0.313557766 0.281122087 0.426255709 0.001524254 EMID2 43 cg11144103 0.419118753 0.243619686 0.479496237 0.001606978 PTRF 44 cg13225272 0.706885433 0.549922871 0.644891369 0.001606978 45 cg24900563 0.767217725 0.632083062 0.738574624 0.001606978 46 rs4742386 0.727302219 0.520210535 0.400186239 0.001606978 47 cg10288784 0.15641827 0.062914082 0.201984604 0.001606978 ABL1 ! 117! SNORA15; 48 cg14995921 0.759212765 0.649913814 0.800269476 0.001606978 CCT6A 49 cg11199014 0.246009366 0.269006158 0.395465243 0.001648469 LRFN1 50 cg20174000 0.226362373 0.142691912 0.27344344 0.001710776 ABL1 51 rs939290 0.757130472 0.424761 0.539292123 0.001789813 52 cg17339147 0.156986192 0.124685586 0.251353276 0.001789813 WNT2 53 cg27528510 0.366449741 0.341019039 0.485340798 0.00182074 EMID2 54 cg00695177 0.646077261 0.723528956 0.813909738 0.00182074 55 cg15436096 0.043098394 0.129237028 0.241128456 0.00183007 GPR135 56 cg03920544 0.568671944 0.605226477 0.460009284 0.001850169 57 cg02720091 0.599788469 0.518536851 0.456954789 0.001850169 CASZ1 58 cg21882477 0.255951684 0.267877824 0.381944255 0.001850169 STK32A 59 cg16100355 0.619376787 0.744601553 0.644112585 0.002043701 MCF2L 60 cg19587838 0.495927849 0.626745303 0.542692748 0.002043701 61 cg21864016 0.808891587 0.645349872 0.732522096 0.002043701 C14orf132 62 cg04032226 0.310599143 0.341367925 0.472607246 0.002043701 PYY;NAGS 63 cg07820332 0.620996969 0.662859895 0.519607216 0.002043701 64 cg14874750 0.321421239 0.292220879 0.431074572 0.002043701 EMID2 65 cg06183338 0.336062523 0.432380202 0.488989425 0.002043701 COL23A1 ! 118! 66 cg17163751 0.135381372 0.177274849 0.269460409 0.002043701 H2AFY2 67 cg23448505 0.324941798 0.192257747 0.266748888 0.002104078 HKR1 68 rs133860 0.520487029 0.788692699 0.72867357 0.002104078 FSCN2;FSC 69 cg02090762 0.37664555 0.357250425 0.492212086 0.002104078 N2 70 cg01393340 0.44716939 0.29794509 0.32913091 0.002178773 CD83 71 cg06711418 0.31626449 0.18915032 0.45241593 0.002178773 MT2A 72 cg13133387 0.684454477 0.820450527 0.782277827 0.002178773 73 cg13506281 0.691144694 0.516226417 0.515255131 0.002178773 MTUS2 74 cg13755866 0.468181123 0.60412492 0.580663919 0.002178773 75 cg18391209 0.671572409 0.41333768 0.516907808 0.002178773 CAPN8 76 cg20919942 0.065891435 0.199460347 0.146302814 0.002178773 SSH2 77 cg26734888 0.232693486 0.090863221 0.151707714 0.002178773 HKR1 78 rs1416770 0.314665198 0.549676499 0.430455328 0.002178773 79 cg09088988 0.23587056 0.217941053 0.394476054 0.002178773 STK32A LOC100130 522; 80 cg13590055 0.661602136 0.609360417 0.735934612 0.002178773 PARD6G 81 cg16024904 0.294618995 0.293422534 0.166059094 0.002178773 B3GNT3 82 cg23424003 0.488149865 0.44762481 0.588810542 0.002178773 EMID2 ! 119! 83 cg26402828 0.470863017 0.397974381 0.528809446 0.002178773 PCK2 84 cg27256423 0.723779307 0.653181222 0.786386007 0.002178773 85 cg27331524 0.130785441 0.089904334 0.234523223 0.002178773 WNT2 86 cg00002033 0.101003295 0.143266373 0.245678327 0.002178773 LRFN1 87 cg00695416 0.348028935 0.388993331 0.505941081 0.002178773 CBR1 88 cg02096887 0.652513959 0.570800088 0.488779652 0.002178773 CASZ1 89 cg03841977 0.286825087 0.340832779 0.441216278 0.002178773 CBR1 90 cg08475953 0.243645433 0.361108922 0.457896308 0.002178773 COL23A1 91 cg14562786 0.632309914 0.64939181 0.76469107 0.002178773 FAM75A3 92 cg16657538 0.546348433 0.482641037 0.421021127 0.002178773 ZNF397OS 93 cg17240976 0.412286972 0.491600729 0.570755441 0.002178773 HDLBP 94 cg17882660 0.160399645 0.250370832 0.291477191 0.002178773 ZIC2 95 cg20606489 0.81704016 0.791385623 0.676471612 0.002178773 96 cg23169584 0.715167528 0.735190416 0.851284637 0.002178773 EXOC4 97 cg27307781 0.314143965 0.37835182 0.502352299 0.002178773 CBR1 98 cg22244940 0.557331192 0.694021762 0.497036594 0.002189631 MMP17 99 cg24651215 0.621026438 0.559008616 0.692399239 0.002194078 100 cg05280698 0.225852951 0.073147524 0.138922462 0.002247092 HKR1 101 cg24834889 0.259863091 0.127804427 0.189010333 0.002247092 HKR1 ! 120! 102 cg07697895 0.130922044 0.074650219 0.248912376 0.002247092 WNT2 103 cg12451631 0.241557926 0.339366523 0.408937944 0.002247092 RASGEF1A 104 cg07015190 0.239921044 0.394664964 0.260329241 0.002256842 RPS6KA2 105 cg13687570 0.274266712 0.129219304 0.187160608 0.002256842 HKR1 106 cg14166009 0.26666097 0.134532431 0.19542877 0.002256842 HKR1 107 cg03315469 0.450986114 0.61013043 0.532658598 0.002331281 108 cg24385580 0.068822326 0.19541165 0.144056048 0.002331281 SSH2 109 cg03224005 0.571128173 0.514503456 0.389378238 0.002331281 110 cg04517722 0.289919528 0.297639622 0.163120844 0.002331281 B3GNT3 111 cg07229186 0.329998572 0.393865462 0.26871746 0.002331281 MT1M 112 cg26166804 0.193304835 0.225245188 0.356453926 0.00238738 H2AFY2 113 cg22655988 0.20224374 0.359587855 0.280179188 0.002398623 CRMP1 114 cg12024906 0.292374677 0.12100908 0.202548466 0.002407207 HKR1 115 cg12948621 0.322063297 0.158428971 0.253746128 0.002407207 HKR1 116 cg19080354 0.447290384 0.308952028 0.403678906 0.002407207 ATHL1 117 cg16322262 0.159964376 0.208804652 0.294848 0.002407207 GPR135 118 cg16662267 0.195653775 0.256864391 0.345581748 0.002407207 CBR1 119 cg21893185 0.038078028 0.134230175 0.222358634 0.002407207 GPR135 120 cg05237503 0.676668569 0.464395081 0.527843432 0.002407407 FBXO3 ! 121! 121 cg08858160 0.489695558 0.621980903 0.526657033 0.002407407 122 cg10320659 0.288424643 0.418394038 0.375987023 0.002407407 123 cg26910511 0.216626574 0.215346534 0.355499838 0.002407407 LRFN1 124 cg22455450 0.296360477 0.339002804 0.423090479 0.002407407 ZNF808 125 cg10165894 0.343371131 0.45106821 0.470887077 0.002412151 126 cg07686394 0.727579062 0.569848365 0.577876814 0.002414372 127 cg05712748 0.746320213 0.60587713 0.565456768 0.002435193 128 cg00321709 0.38756312 0.306578598 0.441680633 0.002435193 CYP2E1 129 cg16744531 0.359401256 0.367971751 0.21143103 0.002435193 B3GNT3 130 cg12386614 0.631289829 0.684446065 0.77936292 0.002437957 131 cg10014563 0.147513866 0.205493201 0.288903715 0.002560643 GPR135 132 cg05115233 0.508769357 0.646690784 0.569480037 0.002568458 RASA3 133 cg16891431 0.529143648 0.393188081 0.443645343 0.002568458 TCERG1L 134 cg21149357 0.517625023 0.39219906 0.575065305 0.002568458 135 cg11009736 0.57279779 0.470174804 0.431836362 0.002568458 MARCO 136 cg15210622 0.527750986 0.629711789 0.661228097 0.002568458 PPP2R3A 137 cg22821834 0.13682943 0.206508003 0.297194699 0.002568458 GPR135 138 cg15165122 0.291417589 0.425918499 0.349061251 0.002589039 ANKRD53 139 cg15351736 0.301622883 0.161676712 0.218356507 0.002589039 DIXDC1 ! 122! 140 cg00243527 0.601589911 0.615337385 0.759499705 0.002589039 LIF 141 cg27467876 0.717509017 0.661553956 0.534596423 0.002589039 SLC39A14 142 cg22848598 0.430444826 0.31930533 0.276324699 0.002589039 ADAM32 143 cg15425276 0.748557001 0.62332373 0.668217275 0.002675386 ACTN1 144 cg08422420 0.318670674 0.192957956 0.391509406 0.0026991 SDHAP3 145 cg03022609 0.277569113 0.145198057 0.206608168 0.002723801 DIXDC1 146 cg08778598 0.348998356 0.203584943 0.43454172 0.002723801 SDHAP3 147 cg11484348 0.490268423 0.616113261 0.553425025 0.002723801 DEFB116 148 cg09310644 0.493512075 0.395353646 0.527643621 0.002723801 ZFP90 149 cg10470368 0.686332406 0.600710771 0.741335352 0.002723801 150 cg02636041 0.229879934 0.320035771 0.367258175 0.002723801 RASGEF1A 151 cg04186360 0.038163238 0.096585578 0.167377185 0.002723801 GPR135 152 cg06721546 0.320494061 0.210659996 0.372390171 0.002736657 SDHAP3 153 cg10485664 0.17540832 0.352979609 0.224864479 0.00281911 FEZF1 154 cg17951978 0.81658544 0.67873175 0.771476995 0.00281911 HSD3B2 155 cg10172783 0.257837245 0.312278685 0.387764627 0.00281911 NAGS;PYY 156 cg09859398 0.039958463 0.124722786 0.229108137 0.002912042 GPR135 157 cg13401703 0.409647782 0.271546799 0.392814264 0.002936682 TTC23 158 cg20206437 0.276036093 0.405541884 0.302620578 0.002936682 RPS6KA2 ! 123! 159 cg24139169 0.328551457 0.19014993 0.259628269 0.002936682 DIXDC1 160 cg25385940 0.439098722 0.301200336 0.407155987 0.002936682 TTC23 161 cg11093548 0.243261104 0.17451025 0.335750406 0.002936682 CREB5 MIR654; MIR376B; MIR376A1; 162 cg19653246 0.508525082 0.59660774 0.673294208 0.002936682 MIR376A2; MIR376C C20orf166; 163 cg04990378 0.571485789 0.424516243 0.469713277 0.002976689 MIR133A2 164 cg27149073 0.394481189 0.264937184 0.459722418 0.002977437 SDHAP3 165 cg10441365 0.096438366 0.113710883 0.263917027 0.002977437 SAR1B 166 cg21488538 0.339929255 0.42360086 0.298577783 0.003036991 CNGA3 167 cg10512745 0.184919512 0.312784674 0.201190765 0.003074752 DMRTA2 168 cg24603464 0.357431916 0.432145899 0.306245477 0.003136178 MLPH OR5R1;OR 169 cg05114178 0.452147719 0.577194036 0.417909144 0.003153333 8U8 170 cg17974166 0.44572974 0.570879253 0.503955979 0.003166504 171 cg03976877 0.298116757 0.409280963 0.264650942 0.003166504 VIPR2 172 cg12556569 0.214288505 0.185221285 0.355978528 0.003166504 APOA5 173 cg00543972 0.302727547 0.148323613 0.211444583 0.003241447 DIXDC1 ! 124! 174 cg03084184 0.473796637 0.609809438 0.436818449 0.003241447 JAKMIP3 175 cg06520095 0.591864781 0.753691388 0.585007914 0.003241447 176 rs2208123 0.459568375 0.673392113 0.516708974 0.003241447 177 cg20250269 0.493893564 0.571703925 0.388228366 0.003241447 178 cg24960960 0.265884542 0.17087508 0.348199156 0.003241447 SDHAP3 LCLAT1;L 179 cg15652532 0.246148239 0.314360265 0.384087031 0.003241447 CLAT1 180 cg23136139 0.361662552 0.447173544 0.493495295 0.003241447 RASGEF1A 181 cg27378537 0.374214561 0.24221486 0.423322313 0.003284622 SDHAP3 182 cg21931717 0.264722876 0.170216597 0.338289917 0.003284622 SDHAP3 183 cg05593887 0.544371802 0.586036447 0.677002072 0.003284622 MAGI2 184 cg02113055 0.380319368 0.681747372 0.439770699 0.003296129 185 cg08900396 0.584244592 0.713929321 0.785428886 0.003296129 RPH3AL 186 cg25885322 0.302599085 0.165776405 0.235069774 0.003296129 DIXDC1 187 cg25465065 0.385144959 0.442176165 0.605742815 0.003320896 PMF1 188 cg21543270 0.567049526 0.66894691 0.705706751 0.003320896 PACS2 189 cg21558508 0.037477872 0.056623043 0.169520189 0.003320896 SAR1B 190 cg05714155 0.604522843 0.466064851 0.510397941 0.003485387 191 cg16511983 0.293327659 0.413286528 0.258178185 0.003485387 ! 125! 192 cg17128947 0.600672811 0.430189752 0.483251678 0.003496806 CPLX1 193 cg12624040 0.639577838 0.733088905 0.600126745 0.003496806 B3GALT1 194 cg12466610 0.200719861 0.189030547 0.435041746 0.003513314 MOSC2 195 cg02589828 0.130827894 0.156048034 0.259970709 0.003594819 SAR1B 196 cg24714905 0.292459203 0.370584431 0.418603758 0.003594819 ZIC2 197 cg14056849 0.443896326 0.345574212 0.315239266 0.003715017 LOC169834 198 cg00303341 0.778566383 0.819073319 0.692408759 0.003782188 199 cg26472636 0.408319587 0.261607544 0.487565314 0.003906568 SDHAP3 200 cg18828306 0.552633043 0.485697997 0.645986598 0.003906568 USH1C 201 cg17662493 0.536870298 0.645431874 0.704138367 0.003906568 SMC1B 202 cg00332745 0.264041048 0.202310409 0.343678412 0.003948605 NOL3 203 cg05134500 0.144263745 0.227108635 0.10159534 0.003948605 MFI2 204 cg07875818 0.324884393 0.266597046 0.433734568 0.003948605 PLXNA4 SDHA; 205 cg23490161 0.277212604 0.163390923 0.31550464 0.003948605 CCDC127 206 cg15385386 0.095082943 0.116638343 0.231125819 0.003948605 SAR1B 207 cg23892028 0.469093664 0.383189767 0.316543036 0.003948605 208 cg01356752 0.630478788 0.711124075 0.775534187 0.003965653 209 rs739259 0.309980213 0.461188601 0.415529993 0.004131195 ! 126! 210 cg12031275 0.734595111 0.72665374 0.597915214 0.004131195 211 cg00980980 0.554747103 0.678665555 0.71205988 0.004131195 212 rs2125573 0.732886534 0.537298693 0.553029446 0.004139999 213 cg16744961 0.506672483 0.589495811 0.657245005 0.004238913 CLECL1 ADAMTS1 214 cg22664298 0.288736168 0.154171784 0.165831123 0.004284665 9 215 cg27452255 0.736402226 0.553660833 0.711323738 0.004284665 216 cg16519587 0.355968546 0.280381465 0.406009345 0.004284665 LOC100134 217 cg20381372 0.625096208 0.690370689 0.780455873 0.004284665 317 218 cg00689685 0.549487102 0.632566523 0.712156279 0.004333304 AGPAT1 219 cg08049519 0.35507184 0.271331899 0.510798497 0.004459853 220 cg03349922 0.15690623 0.201860303 0.285204819 0.004482026 H2AFY2 221 cg01336390 0.210547283 0.362189174 0.288477667 0.004509196 222 cg13697578 0.595341081 0.61902234 0.493661059 0.004509196 223 cg14175932 0.674129059 0.555220928 0.710359489 0.004518979 ADAMTS1 224 cg03358735 0.845479369 0.798738902 0.69486862 0.004524287 7 225 cg12454169 0.183095556 0.271557858 0.322440823 0.004555929 LCLAT1 226 rs9363764 0.663237735 0.443913109 0.491681543 0.004556711 ! 127! 227 cg14264023 0.486074324 0.450728267 0.334130617 0.004556711 COLEC11 228 cg24836826 0.688716082 0.770872763 0.836821717 0.004569261 NCOR2 229 cg17635970 0.646851244 0.503633706 0.700655533 0.004571469 HHLA1 230 cg26531700 0.41399914 0.345423024 0.473101917 0.004602699 231 cg13251750 0.459512734 0.350666198 0.328642663 0.004662784 SERPINA9 232 cg14317384 0.728800097 0.62304208 0.598660891 0.004707385 233 cg18260973 0.054202489 0.08086361 0.227763397 0.004709682 SAR1B 234 rs4331560 0.548536451 0.414857988 0.503255064 0.004838011 235 cg02658043 0.636115315 0.503346122 0.593383608 0.004838382 NRBP2 236 cg18237047 0.664111821 0.514668854 0.588505072 0.004838382 237 cg08849813 0.628859527 0.728584059 0.554282113 0.004881026 SERGEF 238 cg25420101 0.493519785 0.34380088 0.403599752 0.004915316 WDR41 239 cg23318063 0.307464132 0.252918157 0.380427519 0.004957813 240 rs966367 0.501947007 0.289177618 0.43697435 0.005069574 241 cg11497957 0.342859097 0.206240012 0.247707801 0.005194146 ZNF572 242 cg19245335 0.740018119 0.589980889 0.534920287 0.005194146 PRDM15 243 cg01127608 0.510496078 0.597506155 0.457687564 0.005194146 LMX1B 244 cg05062612 0.320179054 0.461628153 0.399089279 0.005218549 WSCD2 245 cg09481537 0.469579974 0.526603031 0.369611711 0.005218549 ! 128! 246 cg10156366 0.246263923 0.311647122 0.17547372 0.005218549 247 cg27591450 0.305488329 0.299362521 0.178925392 0.005218549 248 cg05900567 0.403200148 0.454225654 0.586346663 0.005418323 249 cg19726179 0.143297272 0.184111335 0.286908443 0.005506328 H2AFY2 250 rs264581 0.271559878 0.41946527 0.45965481 0.00550984 251 cg05279330 0.651646619 0.580554062 0.715651747 0.005548037 252 rs1941955 0.59350538 0.604397192 0.801837551 0.00569672 253 rs10774834 0.60515335 0.799310982 0.585534559 0.0057626 254 cg01948148 0.074160129 0.085497369 0.223360052 0.0057626 SAR1B 255 cg17323488 0.413555265 0.382561273 0.288527986 0.0057626 256 cg13236378 0.309457981 0.441282464 0.378042927 0.006169403 WSCD2 257 cg14554788 0.836279843 0.710971361 0.842552251 0.006203344 C3orf26 258 cg15170605 0.248028214 0.237742657 0.448154053 0.006203344 ADD2 259 cg05519582 0.568802801 0.639092186 0.713037314 0.006203457 GRAMD4 SCARNA16 260 cg08024264 0.462975863 0.307449522 0.324829081 0.006334369 ; C17orf86 261 cg26293310 0.059093517 0.078487985 0.237303816 0.006334369 SAR1B 262 cg12894709 0.435766623 0.573960483 0.482872522 0.006385575 PURA 263 cg15845365 0.285187158 0.151640014 0.28361411 0.006385575 MYO10 ! 129! 264 cg20673830 0.477748723 0.422414035 0.599936842 0.006385575 265 cg27230769 0.509470721 0.493918475 0.681902926 0.006385575 LOC285830 266 cg13275129 0.634376098 0.741151029 0.772609326 0.006385575 MAML2 267 cg20187719 0.316219104 0.259134545 0.169728428 0.006385575 LOC285375 268 cg06417478 0.342720675 0.501116804 0.355061735 0.006401946 HOOK2 269 rs798149 0.586843596 0.476173382 0.415942271 0.006401946 270 cg04194432 0.40555425 0.247956329 0.382613167 0.006404992 271 rs715359 0.790575011 0.718434717 0.589973868 0.006484454 272 cg23160717 0.43951081 0.568318589 0.479449911 0.006532577 PURA 273 cg07021532 0.20703454 0.349499534 0.377492495 0.006569902 ZFP2 274 rs1510480 0.376532333 0.55707236 0.466918111 0.006569902 275 rs2385226 0.55070542 0.330231761 0.472193858 0.006569902 276 rs6982811 0.504160657 0.3021701 0.420054163 0.006794112 277 cg03384579 0.140644533 0.13698084 0.27883501 0.006794112 ADD2 278 rs951295 0.401729086 0.295302261 0.441579606 0.006884119 279 rs5936512 0.181179256 0.338849961 0.316378436 0.00689174 280 cg02343823 0.500741405 0.629812105 0.467748877 0.006932119 ZNF300 281 cg11593111 0.449251626 0.579060295 0.485924124 0.006932119 PURA 282 cg12469381 0.505215267 0.490819171 0.671639496 0.007018988 CHN2 ! 130! 283 cg24694833 0.571838194 0.566949014 0.744412653 0.007127042 284 cg18318704 0.04782446 0.063552792 0.191383113 0.007227564 SAR1B 285 rs6546473 0.31865592 0.122325871 0.250351186 0.007409111 286 cg04453550 0.581482267 0.515623772 0.693542209 0.007409111 TNFRSF17 287 cg25649515 0.532616701 0.582813223 0.44719326 0.007409111 288 rs654498 0.621674964 0.541597053 0.678016448 0.007409111 289 cg05971102 0.495603204 0.627852237 0.522511686 0.007521181 290 cg02546601 0.602855644 0.677739973 0.525028121 0.007532243 291 rs2857639 0.535956336 0.524480499 0.305845259 0.007578609 292 cg08603678 0.535774151 0.705184608 0.580607053 0.007665978 EIF3E 293 cg13092806 0.314613028 0.399191469 0.233997033 0.007665978 294 cg05659187 0.220623489 0.195568664 0.353999367 0.007733861 ADD2 295 cg13568515 0.923740349 0.798723351 0.931052333 0.007754941 JAK2 296 cg04675542 0.476299673 0.584853372 0.455613002 0.007754941 ZNF300 297 cg23427269 0.208979358 0.174832902 0.349923507 0.007754941 ADD2 298 cg19863210 0.27193342 0.398539204 0.316502537 0.00784209 DENND3 299 cg15454726 0.180156771 0.246063807 0.318601526 0.00784209 300 cg07501029 0.564659205 0.377441882 0.490470183 0.008001505 KIF26B 301 cg20592836 0.294002773 0.400328874 0.26603274 0.008030452 TP53INP2 ! 131! 302 cg18816122 0.394374521 0.493855531 0.551267126 0.008030452 PLEKHG4B 303 cg21203249 0.560407658 0.496621495 0.647244294 0.008120643 SEMA4D 304 rs9292570 0.608846411 0.591586084 0.42600055 0.008364919 305 cg19783435 0.166974854 0.293372096 0.209735813 0.008392078 ARVCF 306 cg22109827 0.50765037 0.58090069 0.652126828 0.008424608 307 cg18237551 0.418939066 0.545035183 0.360917102 0.008427132 ZNF300 308 cg07258916 0.15744265 0.055335971 0.191148554 0.008454567 PLXNA4 309 cg02907150 0.562965714 0.456855666 0.601103973 0.008544788 PCNX 310 rs10033147 0.508760542 0.653261184 0.622018846 0.008646536 311 cg26864826 0.420389924 0.323685366 0.285041887 0.008646536 312 cg21913319 0.286354338 0.18177412 0.324579242 0.008703407 COL24A1 313 cg12568536 0.520347954 0.380940601 0.53263771 0.008749979 314 cg11291313 0.404031399 0.496033447 0.365362191 0.008749979 ZNF300 315 cg01891583 0.536919262 0.599479021 0.376511604 0.008792625 USP7 316 cg24347663 0.192770865 0.168078678 0.323415077 0.008814128 ADD2 317 cg03609614 0.299480709 0.419324772 0.284100453 0.009096981 ZNF300 318 cg27639199 0.273446014 0.380909587 0.39897105 0.009255808 TMC3 319 cg03988092 0.189882011 0.205303233 0.331289277 0.009395828 320 cg08857144 0.206457895 0.184627189 0.332179798 0.009501983 ADD2 ! 132! 321 cg22851875 0.581109243 0.59478548 0.439452157 0.009527093 322 cg04462931 0.681111156 0.566295273 0.711276254 0.009633798 323 cg21164300 0.326845748 0.450133081 0.29250976 0.009812308 324 cg04610028 0.642909509 0.584365541 0.746197737 0.009822127 RAB11B 325 rs1484127 0.519068759 0.704435811 0.580590697 0.010204674 326 cg10376408 0.484191769 0.411189952 0.543570548 0.010238515 SCUBE2 327 rs5987737 0.621976162 0.490939064 0.635754251 0.010400797 328 rs13369115 0.605699286 0.716544097 0.530692169 0.010400797 329 cg03449867 0.420098061 0.34557929 0.535567841 0.011199566 OCA2 330 cg00988148 0.358641645 0.248133014 0.398021331 0.011364745 331 cg21228005 0.445855973 0.530744006 0.401251987 0.011364745 ZNF300 332 cg26846609 0.478578721 0.534934278 0.609954562 0.011485804 THUMPD1 333 cg24844518 0.542317553 0.42714852 0.585064849 0.011674069 CYFIP2 334 cg16043227 0.533307801 0.599993184 0.459144249 0.011965712 CCR6 335 cg17138852 0.393595902 0.303978453 0.451214007 0.011965712 336 cg24506221 0.474660082 0.392348056 0.526912184 0.012681723 GSTM1 337 rs3818562 0.556814275 0.408733353 0.532703339 0.012858393 338 cg20696478 0.237080178 0.306492598 0.167532476 0.013950948 ZNF175 339 cg07319199 0.692970843 0.781828365 0.645292359 0.014243711 ANTXR1 ! 133! Table 3. Mean and standard deviation of 13 NNNS scores in infants with non-high-risk vs high-risk NNNS profiles. Non-High-Risk NNNS High-Risk NNNS NNNS Score Profile Infants (n=135) Profile Infants (n=16) Habituation, Mean (±SD) † 7.2 (± 1.6) 7.5 (± 0.7) Attention, Mean (±SD) †† 3.8 (± 1.1) 3.4 (± 1.2) Handling, Mean (±SD) 0.4 (± 0.2) 0.6 (± 0.2) Quality of movement, Mean (±SD) 4.1 (± 0.6) 3.0 (± 0.5) Regulation, Mean (±SD) ††† 4.8 (± 0.8) 3.5 (± 0.5) Nonoptimal reflexes, Mean (±SD) 6.3 (± 2.0) 5.6 (± 3.0) Asymmetric reflexes, Mean (±SD) 1.9 (± 1.4) 1.8 (± 1.3) Stress/abstinence, Mean (±SD) 0.2 (± 0.1) 0.3 (± 0.1) Arousal, Mean (±SD) 4.0 (± 0.8) 5.3 (± 0.4) Hypertonicity, Mean (±SD) 0.3 (± 0.6) 1.2 (± 1.3) Hypotonicity, Mean (±SD) 0.7 (± 1.0) 0.6 (± 1.2) Excitability, Mean (±SD) 4.4 (± 2.5) 10.3 (± 1.2) Lethargy, Mean (±SD) 7.0 (± 2.4) 5.0 (± 2.4) †60 samples missing habituation data; 53 non-high-risk samples, 7 high-risk samples. ††16 samples missing attention data; 14 non-high-risk samples, 2 high-risk samples. †††1 non-high-risk sample missing regulation data. ! 134! Table 4. 6 CpG loci associated with infant toenail Hg and high-risk NNNS profile. Illumina Illumina q Value for q Value for Relation CpG Gene Association With Hg in Association With to UCSC Designation Symbol Infant Toenail High-Risk NNNS CpG Annotation Clippings Profile Island cg13267931 EMID2 0.001 0.111 Island cg14874750 EMID2 0.002 0.111 Island cg23424003 EMID2 0.002 0.111 Island cg27179533 EMID2 0.002 0.111 Island cg27528510 EMID2 0.002 0.111 Island cg14175932 0.005 0.111 ! 135! FIGURES 192$Placenta$Samples$ Illumina$450K$Array$ (>485,000$CpG$loci)$ QA/QC, Sex-linked loci removed, SNP- associated loci removed 41$Placenta$Samples$with$ toenail$Hg$data$(384,474$ CpG$loci$examined)$ Locus x Locus analysis on Hg tertiles with FDR correction (q<0.2) 151$Placenta$Samples$ with$NNNS$Profiles$(339$ CpG$loci$examined)$ Locus x Locus analysis on NNNS profile with FDR correction (q<0.2) Figure 1. Diagram of analysis strategy. 192 placental samples were arrayed on an Illumina Infinium HumanMethylation450 BeadArray. Array data then underwent quality assurance and quality control procedures as detailed in Methods section and sex-linked and SNP-associated CpG loci were removed. 41 placental samples with infant toenail Hg data were analyzed for associations between placental methylation and infant toenail Hg, followed by FDR correction (q<0.2). 339 CpG loci associated with infant toenail Hg were subsequently analyzed for associations with high-risk NNNS profile in an ! 136! independent set of 151 placental samples with NNNS profile data, followed by FDR correction (q<0.2). Figure 2. Annotated smoothed plot across five EMID2 CpG loci associated with both infant toenail Hg and high-risk neurobehavioral profile in 41 samples with toenail Hg data by toenail Hg tertile. The Y-axis represents methylation β value. Blue line represents placental methylation in low (referent) toenail Hg tertile infants; yellow line represents that of medium toenail Hg tertile infants; red line represents that of high toenail Hg tertile infants. CpG loci in order of appearance on + strand from 5’ to 3’: 1. cg23424003 2. cg27179533 3. cg27528510 4. cg14874750 5. cg13267931. Yellow dot indicates transcription start site. Yellow arrow indicates first exon. Green line indicates location of CpG island. ! 137! Figure 3. Average ß value across 5 EMID2 CpG loci significantly associated with both infant toenail Hg and high-risk NNNS profile by non-high-risk vs high-risk profile, in a subset of 151 infants. ! 138! CHAPTER 4 DNA Methylation Changes in the Placenta Associated With Manganese in Utero Maccani, Jennifer Z.J.; Koestler, Devin C.; Marsit, Carmen J.; and Kelsey, Karl T. (Manuscript in preparation.) 139! DNA Methylation Changes in the Placenta Associated With Manganese in Utero Jennifer Z.J. Maccani, B.S.1, Devin C. Koestler, PhD2, Carmen J. Marsit, PhD2,3, and Karl T. Kelsey, MD, MOH1,4 1 Department of Pathology and Laboratory Medicine, Brown University, Providence, Rhode Island, USA 2 Section of Biostatistics & Epidemiology, Department of Community and Family Medicine, and 3 Department of Pharmacology and Toxicology, Geisel School of Medicine at Dartmouth, Hanover, New Hampshire, USA 4 Department of Epidemiology, Brown University, Providence, Rhode Island, USA Corresponding Author: Carmen J. Marsit, Associate Professor, Departments of Pharmacology & Toxicology and Community and Family Medicine, Geisel School of Medicine at Dartmouth, HB 7650, Pharmacology & Toxicology, Hanover, NH, 03755, USA; phone: 603-650-1825; fax: 603-650-1129; email: Carmen.J.Marsit@Dartmouth.edu RUNNING TITLE: Manganese-associated placental methylation changes KEY WORDS: epigenetics, in utero, methylation, manganese, FILIP1L, placenta, pregnancy COMPETING FINANCIAL INTERESTS DECLARATION: The authors declare that there are no competing financial interests. 140! ABBREVIATIONS: single nucleotide polymorphism (SNP), manganese (Mn), False Discovery Rate (FDR) 141! ABSTRACT Background: Adequate intake of micronutrients, such as manganese (Mn), is important for normal fetal development, and nutritional deficiencies have been associated with susceptibility to later life disease. However, excess exposures have been associated with neurodevelopmental deficiencies in children, although the mechanisms for this toxicity are unclear. The appropriate function of the placental is critical for successful fetal development, and alterations to placental function may be involved in altering fetal programming and lifelong health. Mn is known to cross and accumulate in the placenta during in utero development, and exposures during this period may alter placental function through epigenetic mechanisms contributing to long-term health effects. Objectives: This study aimed to investigate potential associations between placental DNA methylation changes and infant toenail Mn levels as a marker of in utero exposure. Methods: Using Illumina’s Infinium HumanMethylation450 BeadArray platform, the DNA methylation status of CpG loci genome-wide was interrogated in 61 placental samples and the associations between methylation and infant Mn exposure measured in toenails was assessed. Results: 110 CpG loci associated with infant toenail Mn were identified, including CpG loci residing in genes linked to cancer and neurodevelopmental health. Conclusions: Our observation of altered placental DNA methylation at 110 CpG loci associated with infant toenail Mn suggests a link between prenatal micronutrient levels and epigenetic mechanisms in the placenta. 142! INTRODUCTION Proper placental growth and function are important for normal fetal growth and development. This is particularly true early in gestation, when inadequate nutrition or exposure to environmental toxicants may disturb placental development, thus affecting the growth of the fetus (Dandrea et al. 2001; Desai and Hales 1997; Mattison 2010). The growth and function of the placenta is controlled by the complex interplay of various cellular pathways, which themselves can be altered at the molecular level through epigenetic mechanisms. These epigenetic mechanisms may be susceptible to environmental exposures, and a number of adverse environmental exposures have been shown to alter DNA methylation patterns, representing an epigenetic mechanism in the placenta (Hoyo et al. 2012; Burris et al. 2012; Wilhelm-Benartzi et al. 2012; Suter et al. 2010; Suter et al. 2011). Manganese (Mn) is a micronutrient, and is found in the body in microgram (µg) levels (Nielsen 1996). Adequate amounts of Mn in the diet, predominantly from grain sources (Pennington and Young 1991), are important for normal growth and development; however, too much Mn can inhibit growth (Antonova 1978). During pregnancy, decreased levels of Mn have different mechanisms of effect from increased levels of Mn (Zota et al. 2009), although both low (Mistry and Williams 2011) and high (Misselwitz et al. 1995; Takser et al. 2003) Mn levels in the in utero environment can have negative results. Previous research has suggested that Mn crosses (Misselwitz et al. 1995) and also concentrates in (Miller et al. 1987) the placenta. High hair and nail Mn levels have been observed to be associated with various cancers (Wozniak et al. 2012; 143! Karimi et al. 2012), demonstrating that nail Mn can be utilized as a surrogate measure of Mn in the body. Taking into account the growing body of literature implicating epigenetic mechanisms in the development of later life diseases associated with prenatal exposures, literature associating epigenetic alterations with exposure to manganese and other heavy metals (Cantone et al. 2011; Goodrich et al. 2013; Baccarelli and Bollati 2009), and the associations of prenatal Mn perturbations with childhood neurodevelopmental problems (Claus Henn et al. 2010; Roels et al. 2012; Ericson et al. 2007) and later life diseases such as cancer (Karimi et al. 2012; Bunin et al. 2013), we hypothesized that alterations to normal placental DNA methylation patterns are associated with infant toenail Mn. METHODS Study Design. In this study, the methylation status of >485,000 CpG loci in 61 placental samples was interrogated using an Illumina Infinium HumanMethylation450 BeadArray (Illumina 2013). These placental samples were collected from infants enrolled in an ongoing birth cohort of non-pathologic term pregnancies delivered at Women and Infants’ Hospital in Providence, RI, USA, according to protocols approved by both Women and Infants’ Hospital and Dartmouth College Institutional Review Boards. Full thickness sections of fetal placental tissue were collected 2 cm from the umbilical cord insertion site, free of maternal decidua, within 2 hours of birth. These sections were immediately placed in RNAlaterTM (Applied Biosystems, Inc., AM7020). Samples were blotted dry of RNAlater after being stored at 4ºC for at least 72 hours, snap-frozen in liquid nitrogen and homogenized into a powder and stored at -80ºC prior to processing. 144! Following hospital discharge, toenail clippings from mothers and infants were requested, and were available for 61 of these infants. Infant toenail clippings were taken from all toes and were analyzed for µg Mn per g of toenail material following methods previously described (Karagas et al. 2012) in the Dartmouth trace element analysis laboratory. DNA extraction and bisulfite modification. The QIAmp DNA Mini Kit (Qiagen, Inc., 51304) was used for placental DNA extraction according to manufacturer’s protocol; once extracted, quantification of placental DNA was achieved with a NanoDrop ND-1000 spectrophotometer (NanoDrop). 1 µg of placental DNA was bisulfite modified using an EZ DNA Methylation Kit (Zymo Research, D5008) for subsequent genome- wide DNA methylation analysis. DNA methylation profiling. The Illumina Infinium HumanMethylation450 BeadChip array platform (Illumina 2013) was utilized to interrogate >485,000 CpG loci in the placental DNA samples. This assessment was performed at the University of Minnesota Genomics Center following standardized protocols of the manufacturer. β values representing the methylation status at each CpG locus were calculated from methylated (M) and unmethylated (U) allele intensity; the ratio of fluorescent signals β = Max(M,0)/[Max(M,0) + Max(U,0) + 100] and 0 < β < 1. Complete methylation is indicated by β values near 1 and absence of methylation is indicated by β values near 0. Following array quality assurance (Banister et al. 2011), 384,474 autosomal, non-single nucleotide polymorphism (SNP)-associated CpG loci were subsequently analyzed. Statistical analysis. As β values of the 384,474 included CpG loci were also non- normally distributed, these β values were logit-transformed (Kuan et al. 2010). To adjust 145! for the presence of plate-effects, the ComBat method (Johnson et al. 2007) was applied to the methylation array data; this method has been shown to perform effectively and efficiently compared to competing batch or plate-adjustment methodologies. Successful attenuation of plate-effects was assessed using a Principle Components Analysis of the plate-adjusted methylation data following ComBat. To avoid assuming a linear response in the association between infant toenail Mn and placental methylation, tertiles were generated for infant toenail Mn data and used for subsequent analysis (Kuan et al. 2010). Referent tertile for Mn analyses was medium Mn, due to the fact that Mn has been shown to have an inverted U-shaped mechanism of effect, exhibiting different effects at low doses than at high doses (Zota et al. 2009; Claus Henn et al. 2010). Potential associations between the methylation status of each of 384,474 autosomal, non-single nucleotide polymorphism (SNP)-associated CpG loci and infant toenail Mn were tested by fitting a series of independent linear regression models with logit-transformed, ComBat-adjusted methylation as the dependent variable and tertiles of infant toenail Mn as the key explanatory variable, controlled for maternal age, birth weight percentile, delivery method, and infant gender. False discovery rate (Benjamini 1995) was assessed to control for Type I error, and CpG loci with differences in methylation β value >0.125 and q- value <0.2 were selected. RESULTS Table 1 describes the demographics of the population of mother-infant pairs included in this study. The mothers who participated in this study were predominantly Caucasian and the infants were term births and had a median birth weight percentile of 51. None of the mothers reported smoking during their pregnancies. Mn levels ranged 146! from 0.131 to 5.666 µg Mn per gram of toenail, and these levels were categorized into tertiles for subsequent analyses. The low infant toenail Mn tertile ranged from 0.131 µg Mn per gram of toenail to 0.374; the medium, or referent, tertile from 0.394 to 0.771; and the high tertile from 0.858 to 5.666. Genome-wide DNA methylation status of placenta-derived DNA from 61 placental samples was assessed and the association between methylation at 384,474 CpG loci and infant toenail Mn levels was examined. This resulted in the identification of 110 CpG loci demonstrating a difference in DNA methylation β value of >0.125 between tertiles and FDR-adjusted q-value <0.2 examining the association between methylation and exposure. A full list of these 110 CpG sites, their mean ß values in each infant toenail Mn tertile, and q values for associations with infant toenail Mn is provided in Table 2. 54 of these CpG loci were located in CpG island regions, 14 were located in shore regions, and 10 were located in shelf regions. 85 CpG loci displayed a U-shaped methylation curve, with hypermethylation (55 loci) or hypomethylation (30 loci) in both low and high infant toenail Mn tertiles as compared to referent (medium) tertile. However, 10 CpG loci displayed increasing methylation with increasing infant toenail Mn tertile, and 15 CpG loci displayed decreasing methylation with increasing infant toenail Mn tertile. DISCUSSION The 110 CpG loci we have found to be associated with infant toenail Mn reside in genes that play a myriad of roles in normal biological functioning as well as several diseases and disorders. Some of these genes have been associated in the literature with neurodevelopment (NRBP2) (Larsson et al. 2008) and neurobehavior (LYNX1, GPC5) 147! (Miwa and Walz 2012; Joslyn et al. 2011), and some have been associated with the development of or risk for autism (PLXNA4) (Suda et al. 2011). Roles for other genes have been observed in lung cancer (STK32A) (Dong et al. 2012) as well as the development of multi-drug resistance in tumors (PTRF) (Yi et al. 2013). For one gene in particular, FILIP1L, CpG hypermethylation has been associated with prostate cancer (Desotelle et al. 2013). Interestingly, at two CpG loci within the FILIP1L gene associated with infant toenail Mn, we observed significant (p=0.02 for cg15963552, and p=0.01 for cg22092811) hypermethylation in the placental tissue of infants with high toenail Mn tertile as compared to infants with medium (referent) toenail Mn tertile. This finding, which is illustrated in Figure 1, may represent a novel mechanism or risk factor for later life disease susceptibility, potentially including prostate or other cancers. As high levels of Mn in hair and nails have been associated in the literature with prostate cancer (Karimi et al. 2012) and hypermethylation of CpG loci residing within the FILIP1L gene has been observed in both prostate tumors as well as prostate cancer cell lines (Desotelle et al. 2013), our observation of hypermethylation within the FILIP1L gene in the placental tissue of infants with high toenail Mn tertile may represent an association with mechanistic or risk-related implications for the susceptibility to later life disease in infants exposed to high levels of Mn in utero. Although the function of FILIP1L in the placenta has not yet been described, recent research has demonstrated that epigenetic reprogramming of placental genes is associated with aggressive lung cancer (Rousseaux et al. 2013). Our data identifying epigenetically altered loci associated with in utero exposures in the placenta may aid in identifying such targets for future cancer development. Additional research is warranted to explore if similar profiles of alteration 148! can be observed in the target tissue, in order to examine if such an approach may be warranted; further research into other potential outcomes of interest is also necessary to elucidate whether placental epigenetic effects may link in utero Mn exposure with adverse reproductive outcomes such as preterm birth or low birth weight. Due to this association between paternal Mn intake and cancer in offspring (Bunin et al. 2013), paternal Mn intake prior to conception, which likely would not be reflected in the infant toenail sample, is a potential confounder we have not been able to control. A second limitation of this study is its relatively small sample size (n=61); further investigation of altered placental methylation patterns in a larger sample set with complete data on Mn in corresponding infant hair samples or toenail clipping samples would greatly aid in illuminating the nature of the association between placental methylation changes and in utero exposure to Mn. Such replication would also serve to validate these findings in an independent data set. This study represents an important first step in identifying altered placental methylation patterns associated with differential prenatal exposure to Mn. The infant toenail-Mn-associated differential placental methylation we have observed at these 110 CpG loci, and particularly at cg15963552 and cg22092811 within the FILIP1L gene, may function as a record of fetal exposure (Maccani and Marsit 2009) to increased levels of Mn. Future research aimed at further elucidating the mechanism of the association between Mn and other micronutrients and placental epigenetic effects will add to the understanding of how perturbations in prenatal Mn levels may contribute to the developmental origins of health and disease. 149! ACKNOWLEDGMENTS AND FUNDING: This research was funded by grants R01MH094609 from NIH-NIMH, P01ES022832 and T32ES007272 from the NIH- NIEHS. 150! REFERENCES 1. Dandrea J, Wilson V, Gopalakrishnan G, Heasman L, Budge H, Stephenson T, et al. 2001. Maternal nutritional manipulation of placental growth and glucose transporter 1 (GLUT-1) abundance in sheep. Reproduction 122(5): 793-800. 2. Desai M, Hales CN. 1997. Role of fetal and infant growth in programming metabolism in later life. Biological reviews of the Cambridge Philosophical Society 72(2): 329-348. 3. Mattison DR. 2010. Environmental exposures and development. Current opinion in pediatrics 22(2): 208-218. 4. Hoyo C, Fortner K, Murtha AP, Schildkraut JM, Soubry A, Demark-Wahnefried W, et al. 2012. Association of cord blood methylation fractions at imprinted insulin-like growth factor 2 (IGF2), plasma IGF2, and birth weight. Cancer causes & control : CCC 23(4): 635-645. 5. Burris HH, Rifas-Shiman SL, Baccarelli A, Tarantini L, Boeke CE, Kleinman K, et al. 2012. Associations of LINE-1 DNA Methylation with Preterm Birth in a Prospective Cohort Study. Journal of developmental origins of health and disease 3(3): 173-181. 6. Wilhelm-Benartzi CS, Houseman EA, Maccani MA, Poage GM, Koestler DC, Langevin SM, et al. 2012. In utero exposures, infant growth, and DNA methylation of repetitive elements and developmentally related genes in human placenta. Environmental health perspectives 120(2): 296-302. 151! 7. Suter M, Abramovici A, Showalter L, Hu M, Shope CD, Varner M, et al. 2010. In utero tobacco exposure epigenetically modifies placental CYP1A1 expression. Metabolism: clinical and experimental 59(10): 1481-1490. 8. Suter M, Ma J, Harris A, Patterson L, Brown KA, Shope C, et al. 2011. Maternal tobacco use modestly alters correlated epigenome-wide placental DNA methylation and gene expression. Epigenetics : official journal of the DNA Methylation Society 6(11): 1284-1294. 9. Nielsen FH. 1996. How should dietary guidance be given for mineral elements with beneficial actions or suspected of being essential? The Journal of nutrition 126(9 Suppl): 2377S-2385S. 10. Pennington JA, Young BE. 1991. Total diet study nutritional elements, 1982- 1989. Journal of the American Dietetic Association 91(2): 179-183. 11. Antonova MV. 1978. [Effect of different amounts of manganese in the daily food ration on the growth and development of the progeny of white rats]. Voprosy pitaniia(1): 65-68. 12. Zota AR, Ettinger AS, Bouchard M, Amarasiriwardena CJ, Schwartz J, Hu H, et al. 2009. Maternal blood manganese levels and infant birth weight. Epidemiology 20(3): 367-373. 13. Mistry HD, Williams PJ. 2011. The importance of antioxidant micronutrients in pregnancy. Oxidative medicine and cellular longevity 2011: 841749. 14. Misselwitz B, Muhler A, Weinmann HJ. 1995. A toxicologic risk for using manganese complexes? A literature survey of existing data through several medical specialties. Investigative radiology 30(10): 611-620. 152! 15. Takser L, Mergler D, Hellier G, Sahuquillo J, Huel G. 2003. Manganese, monoamine metabolite levels at birth, and child psychomotor development. Neurotoxicology 24(4-5): 667-674. 16. Miller RK, Mattison DR, Panigel M, Ceckler T, Bryant R, Thomford P. 1987. Kinetic assessment of manganese using magnetic resonance imaging in the dually perfused human placenta in vitro. Environmental health perspectives 74: 81-91. 17. Wozniak A, Kujawa A, Senczuk-Przybylowska M, Kulza M, Gawecki W, Szybiak B, et al. 2012. [Physiological metals in the serum, hair and nails of patients with head and neck cancer]. Przeglad lekarski 69(10): 785-797. 18. Karimi G, Shahar S, Homayouni N, Rajikan R, Abu Bakar NF, Othman MS. 2012. Association between trace element and heavy metal levels in hair and nail with prostate cancer. Asian Pacific journal of cancer prevention : APJCP 13(9): 4249-4253. 19. Cantone L, Nordio F, Hou L, Apostoli P, Bonzini M, Tarantini L, et al. 2011. Inhalable metal-rich air particles and histone H3K4 dimethylation and H3K9 acetylation in a cross-sectional study of steel workers. Environmental health perspectives 119(7): 964-969. 20. Goodrich JM, Basu N, Franzblau A, Dolinoy DC. 2013. Mercury biomarkers and DNA methylation among Michigan dental professionals. Environmental and molecular mutagenesis 54(3): 195-203. 21. Baccarelli A, Bollati V. 2009. Epigenetics and environmental chemicals. Current opinion in pediatrics 21(2): 243-251. 153! 22. Claus Henn B, Ettinger AS, Schwartz J, Tellez-Rojo MM, Lamadrid-Figueroa H, Hernandez-Avila M, et al. 2010. Early postnatal blood manganese levels and children's neurodevelopment. Epidemiology 21(4): 433-439. 23. Roels HA, Bowler RM, Kim Y, Claus Henn B, Mergler D, Hoet P, et al. 2012. Manganese exposure and cognitive deficits: a growing concern for manganese neurotoxicity. Neurotoxicology 33(4): 872-880. 24. Ericson JE, Crinella FM, Clarke-Stewart KA, Allhusen VD, Chan T, Robertson RT. 2007. Prenatal manganese levels linked to childhood behavioral disinhibition. Neurotoxicology and teratology 29(2): 181-187. 25. Bunin GR, Li Y, Ganguly A, Meadows AT, Tseng M. 2013. Parental nutrient intake and risk of retinoblastoma resulting from new germline RB1 mutation. Cancer causes & control : CCC 24(2): 343-355. 26. Illumina. 2013. Illumina HumanMethylation450 BeadChip Kit. Available: http://www.illumina.com/products/methylation_450_beadchip_kits.ilmn [accessed 2013 Mar 7 2013]. 27. Karagas MR, Andrew AS, Nelson HH, Li Z, Punshon T, Schned A, et al. 2012. SLC39A2 and FSIP1 polymorphisms as potential modifiers of arsenic-related bladder cancer. Human genetics 131(3): 453-461. 28. Banister CE, Koestler DC, Maccani MA, Padbury JF, Houseman EA, Marsit CJ. 2011. Infant growth restriction is associated with distinct patterns of DNA methylation in human placentas. Epigenetics : official journal of the DNA Methylation Society 6(7): 920-927. 154! 29. Kuan PF, Wang S, Zhou X, Chu H. 2010. A statistical framework for Illumina DNA methylation arrays. Bioinformatics 26(22): 2849-2855. 30. Johnson WE, Li C, Rabinovic A. 2007. Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics 8(1): 118-127. 31. Benjamini YH, Yosef. 1995. Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society, Series B (Methodological) 57(1): 12. 32. Larsson J, Forsberg M, Brannvall K, Zhang XQ, Enarsson M, Hedborg F, et al. 2008. Nuclear receptor binding protein 2 is induced during neural progenitor differentiation and affects cell survival. Molecular and cellular neurosciences 39(1): 32-39. 33. Miwa JM, Walz A. 2012. Enhancement in motor learning through genetic manipulation of the Lynx1 gene. PLoS One 7(11): e43302. 34. Joslyn G, Wolf FW, Brush G, Wu L, Schuckit M, White RL. 2011. Glypican Gene GPC5 Participates in the Behavioral Response to Ethanol: Evidence from Humans, Mice, and Fruit Flies. G3 (Bethesda) 1(7): 627-635. 35. Suda S, Iwata K, Shimmura C, Kameno Y, Anitha A, Thanseem I, et al. 2011. Decreased expression of axon-guidance receptors in the anterior cingulate cortex in autism. Molecular autism 2(1): 14. 36. Dong J, Hu Z, Wu C, Guo H, Zhou B, Lv J, et al. 2012. Association analyses identify multiple new lung cancer susceptibility loci and their interactions with smoking in the Chinese population. Nature genetics 44(8): 895-899. 155! 37. Yi JS, Mun DG, Lee H, Park JS, Lee JW, Lee JS, et al. 2013. PTRF/Cavin-1 is Essential for Multidrug Resistance in Cancer Cells. Journal of proteome research 12(2): 605-614. 38. Desotelle J, Truong M, Ewald J, Weeratunga P, Yang B, Huang W, et al. 2013. CpG island hypermethylation frequently silences FILIP1L isoform 2 expression in prostate cancer. The Journal of urology 189(1): 329-335. 39. Rousseaux S, Debernardi A, Jacquiau B, Vitte AL, Vesin A, Nagy-Mignotte H, et al. 2013. Ectopic activation of germline and placental genes identifies aggressive metastasis-prone lung cancers. Science translational medicine 5(186): 186ra166. 40. Maccani MA, Marsit CJ. 2009. Epigenetics in the placenta. Am J Reprod Immunol 62(2): 78-89. 156! TABLES Table 1. Study population demographics. Total, n (%) 61 (100%) Infant gender Male, n (%) 27 (44.3%) Female, n (%) 34 (55.7%) Maternal age (years) Mean (±SD) 32.2 ± 3.7 Median (Range) 32 (23 – 39) Tobacco use during pregnancy† Yes, n (%) 0 (0%) No, n (%) 60 (98.4%) Birth weight (grams) Mean (±SD) 3426.9 ± 678.3 Median (Range) 3460 (2160 – 4530) Gestational Age (weeks) Mean (±SD) 39.0 ± 1.2 Median (Range) 39 (37 – 41) Maternal race Caucasian, n (%) 53 (86.9%) Non-Caucasian, n (%) 8 (13.1%) 157! C-section†† Yes, n (%) 19 (31.1%) No, n (%) 41 (67.2%) Recreational drug use during pregnancy Yes, n (%) 1 (1.6%) No, n (%) 60 (98.4%) †1 sample missing tobacco smoking during pregnancy data. ††1 sample missing delivery method data. 158! Table 2. Mean ß values and q values of 110 infant toenail Mn-associated CpG loci. Mean Methylation Mean ß Value of Mean q Value for Methylation Medium Methylation Association ß Value of (Referent) ß Value of With Mn in Illumina Relation Illumina Low Infant Infant High Infant Infant Gene to UCSC CpG Toenail Mn Toenail Mn Toenail Mn Toenail Symbol CpG Rank Designation Tertile Tertile Tertile Clippings Annotation Island 1 cg08422599 0.249863202 0.152052083 0.285932903 0.009339013 KCNIP1 Island 2 cg10903116 0.185156544 0.16387791 0.321558358 0.009339013 MECOM Island 3 cg24458896 0.188341552 0.209146339 0.38238738 0.009339013 MECOM Island 4 cg15958576 0.2398755 0.217889294 0.4281704 0.014358617 MECOM Island 5 cg10953508 0.12493674 0.144556612 0.268462801 0.015867528 MECOM Island 6 cg11782635 0.334069394 0.192081122 0.387441031 0.016939865 KCNIP1 N_Shore 7 cg04248157 0.29660673 0.155577369 0.336756655 0.017565688 KCNIP1 N_Shore 8 cg15713707 0.431022526 0.309276029 0.470614313 0.017565688 KCNIP1 N_Shore 9 cg07875818 0.449964518 0.246510284 0.412612549 0.040986142 PLXNA4 Island C9orf130; 10 cg14274895 0.042505855 0.173223616 0.124273541 0.040986142 C9orf102 Island 11 cg19986202 0.176802839 0.169320852 0.326137205 0.040986142 MECOM Island 159! 12 cg15585036 0.316637302 0.163312511 0.23780218 0.042551972 PLXNA4 Island 13 cg24496475 0.238740697 0.189719704 0.324708606 0.043390718 Island 14 cg17658874 0.67962044 0.806469472 0.785059746 0.047361438 RBMS3 15 cg08355157 0.646909661 0.655718663 0.494295267 0.047361438 LOC154822 16 cg20949107 0.755842696 0.62696048 0.708477889 0.052481223 S_Shore 17 cg11784887 0.318855014 0.489267658 0.342687081 0.054847227 MFHAS1 18 cg03951394 0.642703093 0.846699605 0.769768667 0.057881191 DEAF1 Island 19 rs1416770 0.563584268 0.296347081 0.506910678 0.057881191 20 cg18326657 0.350933767 0.317573548 0.444665102 0.060699787 HLA-L Island 21 cg04467162 0.409237324 0.276219235 0.257381181 0.06142773 Island 22 cg15385386 0.22182139 0.086231817 0.183740166 0.06701333 SAR1B Island 23 cg01948148 0.208892498 0.057682598 0.169588127 0.070751958 SAR1B Island 24 cg04610028 0.529330688 0.71510126 0.662864462 0.070751958 RAB11B N_Shore 25 cg05865327 0.638907598 0.786373592 0.760012367 0.070751958 PPP2R5C 26 cg05873889 0.247647642 0.120464869 0.165258822 0.070751958 SERP2 Island 27 cg06183338 0.494014014 0.354240406 0.445681212 0.070751958 COL23A1 Island 28 cg07258916 0.260795098 0.072868514 0.136626644 0.070751958 PLXNA4 Island 29 cg10441365 0.239384432 0.088223744 0.19849288 0.070751958 SAR1B Island 30 cg12379940 0.191357644 0.058152464 0.111111596 0.070751958 SERP2 Island 160! 31 cg14317384 0.615309446 0.753678396 0.606307787 0.070751958 Island 32 cg21468420 0.360550214 0.18938082 0.23355534 0.070751958 Island 33 cg24079361 0.263590685 0.113052331 0.174544254 0.070751958 SERP2 Island 34 cg07519816 0.15137116 0.128041406 0.270589763 0.070751958 Island 35 cg17738613 0.37684222 0.259353326 0.396916456 0.070751958 GPC5 36 cg18093448 0.716282544 0.692435495 0.840930327 0.070751958 WWC2 C3orf26; FILIP1L; 37 cg22092811 0.229699959 0.168280815 0.333707532 0.070751958 MIR548G Island 38 cg21847720 0.198089923 0.132505445 0.061495341 0.070751958 MYOM2 Island 39 cg18260973 0.203031041 0.041042647 0.171577035 0.074599992 SAR1B Island 40 cg26209005 0.288655583 0.161068906 0.204998522 0.07681301 PLXNA4 Island 41 cg09993319 0.246284472 0.392446059 0.294003139 0.082161035 MGMT 42 cg19357499 0.428322346 0.354242885 0.257749515 0.082161035 NAPRT1 Island 43 cg00345083 0.563391191 0.624792944 0.389132601 0.086475692 AJAP1 N_Shore 44 rs739259 0.514643177 0.34513586 0.293797748 0.087172089 45 cg16142707 0.166203438 0.2635224 0.292319411 0.087172089 LEKR1 Island 46 cg02346492 0.307967225 0.150736582 0.22461069 0.092763829 NTN1 Island 47 cg00243527 0.603590096 0.586168599 0.743272995 0.092763829 LIF N_Shore 161! 48 cg22232327 0.153169987 0.124505282 0.266620388 0.092763829 Island 49 cg09088988 0.362181865 0.203556331 0.33482894 0.097243764 STK32A Island 50 cg23698271 0.575305561 0.76431553 0.724824858 0.097243764 TIAL1 51 cg01997599 0.572289349 0.461589996 0.420999825 0.102155057 TRPM4 S_Shelf 52 cg18318704 0.192246899 0.039787278 0.134584531 0.10725978 SAR1B Island 53 cg18391209 0.44891165 0.634882083 0.478187947 0.108710107 CAPN8 S_Shelf 54 rs3818562 0.509581273 0.553976785 0.345220301 0.108710107 55 cg02956194 0.528582128 0.489480039 0.396856716 0.108710107 S_Shelf 56 cg15829294 0.186378049 0.057016698 0.138962761 0.117022628 ZNF709 Island 57 cg26293310 0.216785983 0.047511456 0.176169691 0.11775395 SAR1B Island 58 cg10363118 0.670134457 0.618630801 0.53489034 0.119470373 S_Shelf 59 cg01356752 0.714205017 0.681892998 0.812248868 0.119653528 S_Shore 60 cg13275129 0.703501117 0.778102297 0.562169073 0.119653528 MAML2 C3orf26; FILIP1L; 61 cg15963552 0.218834194 0.150439561 0.300287108 0.119653528 MIR548G N_Shore 62 cg02507579 0.537690211 0.322515454 0.458243814 0.122107885 OR5H15 63 cg06264882 0.668914495 0.524365833 0.670983336 0.122107885 N_Shore 64 cg08002883 0.418411993 0.547060237 0.495001451 0.122107885 FGF12 Island 162! 65 cg08475953 0.432372006 0.273570527 0.344281219 0.122107885 COL23A1 Island 66 cg25673075 0.454379765 0.582278766 0.53040605 0.122107885 67 cg27307781 0.444033927 0.300905806 0.44399679 0.122107885 CBR1 Island 68 cg16120147 0.632934758 0.62174686 0.481255802 0.122107885 LYNX1 69 cg09561458 0.166534368 0.288949458 0.294084603 0.122107885 ZNF417 S_Shore 70 cg16043227 0.42700609 0.538351407 0.560373937 0.122107885 CCR6 71 cg19738233 0.612649804 0.716349365 0.738195054 0.122107885 DNAJC24 S_Shelf 72 cg19863210 0.239874533 0.345698082 0.389781243 0.122107885 DENND3 Island 73 cg07506153 0.559488925 0.690720361 0.689693957 0.125162743 EBF3 N_Shore 74 cg18096251 0.440397091 0.293963523 0.343447964 0.125162743 75 cg17560327 0.27741889 0.43232743 0.411874522 0.125785159 ZNF417 Island 76 cg02658043 0.52020216 0.68835243 0.516973216 0.13756803 NRBP2 N_Shore 77 cg15029183 0.506897 0.480128186 0.631246498 0.13756803 AIG1 78 cg08289409 0.164663889 0.310177722 0.301015065 0.139427235 ZNF417 Island 79 cg16316162 0.505079614 0.440220658 0.344284131 0.139427235 NAPRT1 Island 80 cg12342501 0.680823141 0.641024972 0.544253964 0.142385599 81 cg11390073 0.342444603 0.487145372 0.393922851 0.152930168 Island 82 cg15561493 0.622915519 0.753519299 0.644907564 0.154377079 VWDE 83 cg26035071 0.566580609 0.581146372 0.705819555 0.155114157 S_Shelf 163! 84 cg08508337 0.448543919 0.369940828 0.292431368 0.156208064 NAPRT1 Island 85 cg07673740 0.450233489 0.32055842 0.363658988 0.157423452 Island 86 cg24694833 0.713186121 0.53660077 0.725149435 0.157423452 S_Shelf 87 cg02604524 0.416840198 0.366165086 0.508909835 0.158703274 CCDC48 Island 88 cg00713939 0.444165014 0.371855494 0.302777261 0.158703274 NAPRT1 Island 89 cg01876809 0.413897408 0.505207546 0.552570692 0.158703274 S_Shelf 90 cg09219280 0.597712027 0.717689686 0.73415745 0.158703274 NFIA 91 cg12471283 0.479571571 0.620792886 0.610439723 0.16195422 92 cg09737095 0.414539919 0.302002078 0.45816984 0.16195422 KCNJ5 93 cg22664298 0.196874801 0.263368178 0.133205581 0.16195422 ADAMTS19 Island 94 cg10902825 0.291579373 0.43464538 0.34827024 0.164290635 Island 95 cg11144103 0.310799185 0.447212371 0.307554454 0.164983279 PTRF S_Shelf 96 cg12624040 0.687925859 0.666639926 0.542691431 0.164983279 B3GALT1 97 rs2208123 0.631519123 0.494614803 0.49947291 0.165792368 98 cg21149357 0.567576864 0.520507838 0.418336502 0.16815874 S_Shelf 99 cg10171063 0.254135762 0.413197129 0.346430763 0.169906912 Island 100 rs4331560 0.575572199 0.436185477 0.394003448 0.173138283 101 rs5926356 0.358557597 0.511637079 0.354942935 0.173138283 164! 102 rs1484127 0.476829311 0.668362444 0.665793234 0.17459099 103 cg23727079 0.647991682 0.567850146 0.702562248 0.181830192 104 cg25094735 0.714005836 0.567942419 0.61689433 0.192034216 NAPSB 105 cg05971102 0.518187582 0.50981094 0.639755354 0.192034216 S_Shore 106 cg00316839 0.417420839 0.251071629 0.303100908 0.196547304 Island 107 cg09232430 0.310461988 0.182083695 0.196837818 0.196547304 ECE1 Island 108 rs133860 0.727658814 0.545223831 0.720378364 0.196547304 109 rs1495031 0.497104562 0.621986273 0.451502228 0.196547304 110 rs1510480 0.579215368 0.396832728 0.55184914 0.198833165 165! FIGURE Figure 1. Infants with high toenail Mn have significant placental hypermethylation as compared to infants with medium (referent) toenail Mn at two CpG loci associated with infant toenail Mn in the FILIP1L gene. P values are as follows: cg15963552, p=0.02; cg22092811, p=0.01. 166! CHAPTER 5 Discussion 167! ! ! This thesis aimed to elucidate placental DNA methylation biomarkers of adverse in utero exposures as well as reproductive outcomes such as preterm birth and neurodevelopment. In this chapter, the findings of the previous four chapters will be summarized and contextualized by the existing literature, and future experiments will be proposed that may further reveal the mechanistic nature of these associations and their implications. It is widely recognized in the literature that epigenetic mechanisms may underlie susceptibilities to many diseases and disorders in humans. It is important to investigate the role that epigenetic processes, such as alterations to normal DNA methylation patterns in the placenta, may play in response to adverse exposures during pregnancy. A greater knowledge of such mechanisms may assist in the discovery of novel interventions to improve the quality of life of infants suffering from certain diseases or disorders. Chapter 1 of this thesis introduced many of the previously reported links between adverse in utero exposures, placental DNA methylation changes, and negative reproductive outcomes. In Chapter 2, the results of a study of 206 human placental samples were described in which an association was observed between maternal tobacco smoking during pregnancy, alterations to normal placental DNA methylation patterns and preterm birth. Specifically, cg04757093 within the gene body of the runt-related transcription factor 3 (RUNX3) gene was found to be hypermethylated in the placental tissue of infants born to smoking mothers as compared to the placental tissue of infants of non-smoking mothers, both in data from an Illumina Infinium HumanMethylation27 BeadArray (Illumina 2011) as well as in pyrosequencing data. Associations between maternal 168! ! ! tobacco smoking during pregnancy and gestational age at birth (Simpson 1957; Kullander 1971; Stillman et al. 1986; Kyrklund-Blomberg and Cnattingius 1998; Shah and Bracken 2000) have been previously reported in the literature. Epigenetic alterations have also been described in the literature to be associated with both maternal tobacco smoking during pregnancy (Joubert et al. 2012) and preterm birth (Burris et al. 2012). A potential role for the RUNX3 gene in the modulation of effects of maternal smoking during pregnancy may exist, as RUNX3 is known to be involved in in utero exposure-associated early life disease susceptibility (Haley et al. 2011). RUNX3 is involved in immune system development, including T cell (Woolf et al. 2007; Zamisch et al. 2009; Tokunaga et al. 2009; Klunker et al. 2009) and macrophage differentiation (Sanchez-Martin et al. 2011). RUNX3 has also been described in the literature as a tumor suppressor gene associated with several cancers (Tokunaga et al. 2009; Chen et al. 2011; Chen 2012; He et al. 2012; Ito et al. 2008; Li et al. 2002; Lu et al. 2012; Tang et al. 2012; Vogiatzi et al. 2006; Xiao and Liu 2004). Specifically, RUNX3 hypermethylation is associated with breast (Chen 2012), stomach (Li et al. 2002; Lu et al. 2012; Tang et al. 2012), prostate (Mahapatra et al. 2012), and lung cancers (Yanagawa et al. 2003), as well as bladder cancer in smokers (Wolff et al. 2008). Though a role has not yet been described in the literature for RUNX3 in placental tissue, it has been shown in mice to be associated with prenatal smoke exposure-induced increased airway hyperresponsiveness (Haley et al. 2011) as well as with asthma (Fainaru et al. 2005; Fainaru et al. 2004) through alterations to dendritic cell function and associated changes in lung alveoli. Increased airway hyperresponsiveness is known to be associated with childhood asthma (Singh et al. 2010; Singh et al. 2003), which is associated with prenatal exposure to tobacco smoke 169! ! ! (Wongtrakool et al. 2012; Lux et al. 2000; Haberg et al. 2007; Lannero et al. 2006; Magnusson et al. 2005; Gilliland et al. 2003; Stein et al. 1999; Prabhu et al. 2010), still a highly prevalent in utero exposure (Tong et al. 2009). Also, as RUNX3 expression is controlled by a retinoic acid-sensitive signaling pathway in some cell types (Puig-Kroger et al. 2010), retinoic acid may be used in future in vitro or laboratory-based studies to further explore the mechanistic nature of the methylation changes observed in this study and any expression changes that may be associated with them. An interesting aspect of the implications of these findings is that the hypermethylation we observed was located specifically in the gene body of RUNX3. Controversial findings exist in the literature that DNA methylation within the gene body region of genes may enhance, rather than silence, gene transcription (Zilberman et al. 2007; Zhang et al. 2006; Ball et al. 2009; Hellman and Chess 2007; Rauch et al. 2009). Since this study has focused on DNA methylation changes, it is important for future studies to determine the precise nature of the effect that these methylation changes have on RUNX3 gene expression. Hydroxymethylation, as compared to methylation, has also been associated in the literature with enhanced gene expression rather than gene silencing (Serandour et al. 2012), and a potential limitation of commonly used techniques such as pyrosequencing is that these two types of methylation cannot be distinguished following bisulfite modification, since amplification of product can continue and render 5- hydroxymethylcytosine indistinguishable from methylated cytosine, in spite of a distinct adduct being formed for 5-hydroxymethylcytosine (Olkhov-Mitsel and Bapat 2012). Future studies desiring to distinguish DNA hydroxymethylation from DNA methylation may, however, use the technique of oxidative bisulfite modification, or oxBS-Seq, to 170! ! ! convert hydroxymethyl marks to uracil and detect them using bisulfite sequencing or pyrosequencing (Olkhov-Mitsel and Bapat 2012). As RUNX3 silencing in mice has been shown to result in asthma-like disease (Fainaru et al. 2005; Fainaru et al. 2004), it is important to determine whether silencing or enhancement of RUNX3 gene expression results from this hypermethylation at cg04757093, as well as what consequences the silencing or enhancement of RUNX3 gene expression in placental tissue may have as compared to that in other tissues. DNA methylation changes associated with both smoking during pregnancy and preterm birth, such as those which have been observed in this study, may be investigated in future studies using a comparatively more high-throughput genome-wide array, such as the Illumina Infinium HumanMethylation450 BeadArray (Illumina 2013). The Illumina Infinium HumanMethylation450 BeadArray analyzes more than 450,000 CpG loci, providing greater coverage than the Illumina Infinium HumanMethylation27 BeadArray (Illumina 2011) employed in this study, which analyzed more than 27,000 CpG loci. The additional coverage provided by the HumanMethylation450 BeadArray would allow the DNA methylation status of many more loci within RUNX3 to be assessed. This would allow for other nearby CpG sites to be analyzed, perhaps including more CpG loci specifically located within the gene body of RUNX3. In tandem with gene expression analysis, such a study would greatly aid in elucidating the potential effects that DNA methylation changes in placental tissue may have. Specifically, DNA methylation changes within the gene body of RUNX3, such as those observed at cg04757093, may play novel roles in the response to in utero exposure to tobacco smoke, as well as in 171! ! ! preterm birth, and it will be important for future research to investigate these mechanisms. Taken together, the results of the study described in Chapter 2 of this thesis suggest that CpG locus cg04757093, within the RUNX3 gene, may be involved in the development of preterm birth associated with maternal tobacco smoking during pregnancy and should be investigated in future studies of preterm birth as well as other adverse reproductive outcomes. Results from future analyses may aid in the discovery of novel ways to improve the quality of life of infants affected by such exposures and outcomes. Chapter 3 described placental patterns of DNA methylation that were observed to be associated with infant toenail mercury (Hg) as well as high-risk infant neurobehavioral profile. 339 CpG loci were found to be associated with infant toenail Hg, and several of these loci have been implicated in the literature with many disorders, including type 2 diabetes (ZBED3) (Ohshige et al. 2011), cancers (FBXO3, HOOK2, MT2A, EIF3E, RPH3AL, PTRF, MT1M, STK32A) (Cha et al. 2011; Shimada et al. 2005; Krzeslak et al. 2013; Zhou et al. 2012; SL Liu et al. 2012; YP Liu et al. 2012; Lee et al. 2009; Ji et al. 2006; Mao et al. 2012), schizophrenia (DIXDC1, ARVCF, MAGI2, ZIC2) (Bradshaw and Porteous 2012; Sim et al. 2012; Mas et al. 2010; Mas et al. 2009; Chen et al. 2005), ADHD (TCERG1L) (Neale et al. 2010; Karlsson et al. 2012; Hatayama et al. 2011), movement disorders (NOL3, TP53INP2) (Russell et al. 2012; Bennetts et al. 2007), Creutzfeldt-Jakob disease (CHN2) (Mead et al. 2012), Huntington’s disease (H2AFY2, AGPAT1) (Hu et al. 2011; Cong et al. 2012), Parkinson’s disease (LMX1B) (Tian et al. 2012), and autism (PLXNA4, WNT2) (Suda et al. 2011; Lin et al. 2012; Kalkman 2012). 172! ! ! Six of these 339 loci were found to also be associated with high-risk infant neurobehavioral profile. For this analysis, infant neurobehavioral profiles were assessed using a latent profiling strategy (Liu et al. 2010) which clustered infants by NICU Network Neurobehavioral Scales (NNNS) scores assessed during their hospital stay. This approach generated seven classes from which four profiles were defined, one of which was a high-risk infant neurobehavioral profile with the same characteristics as the high- risk class in the study by Liu, et al (Liu et al. 2010). Five of the six CpG loci associated with both infant toenail Hg and high-risk NNNS RPMM class were found to reside in the EMID2 gene. A function for the EMID2 gene in the placenta has not yet been described in the literature. However, a long-range cis-regulation mutation within EMID2 that results in enhancer adoption of the sonic hedgehog (SHH) gene has been described in the literature (Lettice et al. 2011). This enhancer is located within an intronic region of the EMID2 gene, and drives the ectopic expression of SHH (Lettice et al. 2011). Despite the fact that these five EMID2 CpG loci have not been previously described as associated with enhancer adoption of SHH, this possible link warrants further investigation, particularly in light of literature associating SHH with neurodevelopmental processes such as neural tube patterning, stem cell proliferation, and cell survival (Herrmann et al. 2008; McCarthy and Argraves 2003; Ho and Scott 2002). Although limitations of this study include its relatively small sample size (41 samples were analyzed for infant toenail Hg-associated placental DNA methylation changes in an Illumina Infinium HumanMethylation450 BeadArray (Illumina 2013)) and the inclusion of placental tissue only from term infants, this study comprises an important 173! ! ! first step in investigating placental DNA methylation as a potential link between in utero Hg exposure and infant neurobehavioral outcomes. In addition, it may be important for future studies aimed at further exploring the association between in utero exposure to Hg and adverse neurobehavioral outcomes to examine these five EMID2 CpG loci and their methylation status in the placenta, as these altered patterns of DNA methylation may play a role in Hg-associated neurobehavioral deficits in exposed infants. A future study, for example, could include a prospective cohort of mother-infant pairs, from which placental tissue, toenail clippings, and infant NNNS scores could be collected for future analysis using a modified methodology from previously reported studies (Marsit et al. 2012; Watson et al. 2012). These infants could then be followed throughout childhood and into adulthood, and annual neurobehavioral measures could be recorded including standardized test scores as well as other measures of neurobehavior, such as ADHD or autism diagnoses, hand-eye coordination tests, etc. Such a study could then investigate associations of placental DNA methylation changes, infant toenail Hg, and adverse neurobehavioral outcomes not only at birth but later into childhood and adulthood as well. However, it would be prudent for such a future study to control for seafood in the diet, which is a known source of Hg exposure (Clarkson 1997). Collectively, the findings of Chapter 3 have helped to elucidate the nature of the association between in utero Hg exposure and adverse neurobehavioral outcomes in infants, as well as further describe the role that placental DNA methylation changes may play in modulating that relationship. Although a role for EMID2 in the placenta has yet to be described in the literature, these findings may contribute to a greater understanding of the role of epigenetics in the developmental origins of health and disease. 174! ! ! Chapter 4 described a similar study to that in Chapter 3, and observed associations between infant toenail manganese (Mn) and placental DNA methylation alterations. Mn has been shown in the literature to have an inverse U-shaped mechanism of effect (Zota et al. 2009) in which both low (Mistry and Williams 2011) and high (Misselwitz et al. 1995; Takser et al. 2003) levels of Mn can have detrimental effects on human health. Hair and nail Mn have been used as surrogate measures of bodily Mn levels in the literature, and increased levels of Mn have been associated with adverse effects on childhood neurodevelopment (Claus Henn et al. 2010; Roels et al. 2012; Ericson et al. 2007) and later life diseases including several types of cancer (Wozniak et al. 2012); (Karimi et al. 2012). Mn is capable of crossing both the placenta and the intact blood brain barrier (Misselwitz et al. 1995), and like several other heavy metals (Goodrich et al. 2013; Baccarelli and Bollati 2009), exposure to Mn has been associated in the literature with epigenetic alterations (Cantone et al. 2011; Goodrich et al. 2013; Baccarelli and Bollati 2009). In the study detailed in Chapter 4, 110 CpG loci were found to be significantly associated with infant toenail Mn, and two of these 110 CpG loci (cg15963552 and cg22092811) were found to reside within the FILIP1L gene and were hypermethylated in the placental tissue of infants with high toenail Mn tertile as compared to the placental tissue of infants with medium (referent) toenail Mn tertile. High hair and nail Mn levels have been associated in the literature with the development of prostate cancer (Karimi et al. 2012), and the FILIP1L gene has been found to be hypermethylated in prostate tumors and prostate cancer cell lines (Desotelle et al. 2013), although its function in placental tissue has not yet been described in the literature. Considering the association observed 175! ! ! between increased Mn intake and prostate cancer (Karimi et al. 2012), the hypermethylation of these loci within the FILIP1L gene in the placental tissue of infants with high toenail Mn tertile is interesting and warrants further investigation, particularly in light of studies demonstrating that epigenetic reprogramming of placental genes is associated with aggressive cancers of the lung (Rousseaux et al. 2013). These findings may suggest a novel mechanism involved in, or a potential risk-associated factor for, later life diseases that may include prostate or other cancers. Future studies might include prospective cohort studies in which placental tissue and toenail clippings are collected at birth, and participants are followed into adulthood and rates of prostate cancer are analyzed with relation to the placental methylation status of the FILIP1L gene, specifically the two loci observed to be hypermethylated in this study, cg15963552 and cg22092811. Such a prospective study could help to further elucidate the nature of the association between in utero Mn exposure and risk for later- life prostate cancer. However, associations have also been observed between increased paternal Mn intake prior to conception and an increased risk of retinoblastoma in offspring (Bunin et al. 2013), so a limitation of the study in Chapter 4 is that paternal Mn intake prior to conception has not been controlled for. This may be a potential confounder, as we have observed associations between increased infant toenail Mn levels and placental epigenetic alterations in a gene previously associated with cancer risk in the literature. A prospective cohort study like the one outlined above would ideally include paternal toenail Mn measurements, perhaps by enrolling parents who are attempting pregnancy and by collecting toenail clippings at time of enrollment and at additional time-points throughout the study period, thus better controlling for paternal Mn levels 176! ! ! prior to conception. This approach would ensure a more accurate determination of the potential contribution of paternal Mn levels to fetal development and later life disease in offspring. A second limitation of this study is its relatively small sample size (n=61), and a prospective cohort study would ideally be composed of a much larger sample set, perhaps enrolling couples at pre-natal checkups through local doctors, as enrollment through fertility clinics may bias such a study toward infants born to couples with fertility problems, which could introduce additional confounding factors. For enrolled mothers who eventually become pregnant, information and samples to be collected at birth could include infant toenail clippings, NNNS scores, and placental samples. Placental samples could then be processed according to the methods described in Chapter 2, and DNA methylation could be assessed using Illumina’s Infinium HumanMethylation450 BeadArray (Illumina 2013) described above. Conducting a study in such a sample set with complete data on Mn in infant toenail clippings, along with methylation profiling in placental tissue, would aid in more fully describing the potential role that DNA methylation may play in the association between increased prenatal Mn levels and later life prostate cancer, and may also function to replicate the findings detailed in Chapter 4 in an independent sample set, thereby addressing a third limitation of the current study – a lack of replication in an independent sample set. Despite these limitations and the vast potential for future research, this study is a first step in attempting to discover epigenetic mechanisms associated with in utero exposure to increased levels of Mn. The altered placental methylation patterns associated with prenatal exposure to Mn may influence future research aimed at uncovering 177! ! ! relationships between prenatal micronutrient levels and the developmental origins of later life disease. Future investigations should focus on the important role that micronutrient intake during pregnancy may play in programming offspring for a healthy infancy, childhood, and adult life, and hopefully such research can influence recommendations to pregnant women for micronutrient intake during pregnancy, thus helping to improve the health and quality of life for infants and their families. Taken together, this thesis has identified placental DNA methylation biomarkers of in utero exposures, such as maternal tobacco smoking and heavy metal exposure, as well as adverse reproductive outcomes, including preterm birth and high-risk neurobehavioral profile. Although there is a large body of literature describing in utero exposures to such toxicants, much research remains to be done on the mechanisms linking these exposures to adverse reproductive outcomes. Epigenetic mechanisms such as DNA methylation have recently been proposed to play a role in the development of later life disease associated with prenatal exposures, and although this thesis work has not elucidated specific mechanisms of disease, it has described associations between prenatal toxicant exposures, DNA methylation alterations in the placenta, and adverse reproductive outcomes in infants. This thesis work has suggested that additional research is warranted to further reveal the nature of these associations, which may someday aid in the development of novel interventions or diagnostics for adverse outcomes associated with such exposures. Rather than being hypothesis-driven, this thesis research has employed a hypothesis-generating approach through the use of microarray technology, specifically by assessing the DNA methylation status in placental tissue of thousands of CpG loci. This 178! ! ! approach allowed for the high-throughput screening of these CpG loci for placental DNA methylation biomarkers of in utero toxicant exposures, such as maternal tobacco smoke or heavy metals, as well as for placental DNA methylation biomarkers of preterm birth or high-risk neurobehavioral profile. By identifying biomarkers of both exposures and outcomes, this thesis described potential roles for epigenetic mechanisms in the context of these associations and contributed to better characterizing the nature of the developmental origins of health and disease. This work might be expanded upon in several ways. First, the array methods as well as bisulfite pyrosequencing methods employed in this work are unable to distinguish between DNA methylation and DNA hydroxymethylation, and due to the distinct functions of these two marks that have only very recently been described, future studies could focus on attempting to distinguish them from each other, as well as assess the effects on gene expression that these DNA methylation and/or DNA hydroxymethylation marks might have. Such research would help to better elucidate the nature of these epigenetic mechanisms in the context of in utero exposure to maternal tobacco smoke and/or heavy metals as well as the role that they may play in the development of later life disease. Additionally, prospective cohort studies, such as those described above, may aid in investigating the associations between in utero exposures and later life diseases such as asthma, which has been associated with maternal tobacco smoking during pregnancy as well as the RUNX3 gene, and prostate cancer, which has been associated both with increased Mn intake (Karimi et al. 2012) as well as FILIP1L hypermethylation (Desotelle et al. 2013). 179! ! ! Finally, a more complete understanding of how multiple epigenetic mechanisms may contribute to the developmental origins of health and disease may be achieved by investigating other epigenetic mechanisms in the context of prenatal maternal tobacco smoking or heavy metal exposures and outcomes such as preterm birth or neurobehavioral profile. While this thesis has assessed DNA methylation alterations as biomarkers of these exposures and outcomes, microRNAs (miRNAs) may also play a role in these associations, as placental miRNAs have recently been observed to be associated with infant neurobehavior (Maccani et al. In Press.). Ultimately, this research identified placental DNA methylation biomarkers of detrimental prenatal exposures that have been associated with adverse reproductive outcomes and later life disease. The discovery of these biomarkers suggests that future research is warranted to further describe the nature of the role that epigenetic mechanisms may play in potentially linking prenatal toxicant exposures to later life disease. These biomarkers may also serve to assist in the development of novel interventions or diagnostics for infants exposed to poor in utero environments and who may be more susceptible to the development of diseases or disorders. This thesis work has sought to generate a greater knowledge of the developmental origins of health and disease, and this knowledge, in turn, will contribute to improving the quality of life of infants at greater risk for later life disease. 180! ! ! REFERENCES 1. Illumina. 2011. Infinium HumanMethylation27 BeadChip Kits. Available: http://www.illumina.com/products/infinium_humanmethylation27_beadchip_kits. ilmn [accessed 2011 20 Dec 2011]. 2. Simpson WJ. 1957. A preliminary report on cigarette smoking and the incidence of prematurity. Am J Obstet Gynecol 73(4): 807-815. 3. Kullander SK, B. 1971. A prospective study of smoking and pregnancy. Acta Obstet Gynecol Scand 50: 11. 4. Stillman RJ, Rosenberg MJ, Sachs BP. 1986. Smoking and reproduction. Fertility and sterility 46(4): 545-566. 5. Kyrklund-Blomberg NB, Cnattingius S. 1998. Preterm birth and maternal smoking: risks related to gestational age and onset of delivery. Am J Obstet Gynecol 179(4): 1051-1055. 6. Shah NR, Bracken MB. 2000. A systematic review and meta-analysis of prospective studies on the association between maternal cigarette smoking and preterm delivery. Am J Obstet Gynecol 182(2): 465-472. 7. Joubert BR, Haberg SE, Nilsen RM, Wang X, Vollset SE, Murphy SK, et al. 2012. 450K epigenome-wide scan identifies differential DNA methylation in newborns related to maternal smoking during pregnancy. Environmental health perspectives 120(10): 1425-1431. 8. Burris HH, Rifas-Shiman SL, Baccarelli A, Tarantini L, Boeke CE, Kleinman K, et al. 2012. Associations of LINE-1 DNA Methylation with Preterm Birth in a 181! ! ! Prospective Cohort Study. Journal of developmental origins of health and disease 3(3): 173-181. 9. Haley KJ, Lasky-Su J, Manoli SE, Smith LA, Shahsafaei A, Weiss ST, et al. 2011. RUNX transcription factors: association with pediatric asthma and modulated by maternal smoking. American journal of physiology Lung cellular and molecular physiology 301(5): L693-701. 10. Woolf E, Brenner O, Goldenberg D, Levanon D, Groner Y. 2007. Runx3 regulates dendritic epidermal T cell development. Developmental biology 303(2): 703-714. 11. Zamisch M, Tian L, Grenningloh R, Xiong Y, Wildt KF, Ehlers M, et al. 2009. The transcription factor Ets1 is important for CD4 repression and Runx3 up- regulation during CD8 T cell differentiation in the thymus. The Journal of experimental medicine 206(12): 2685-2699. 12. Tokunaga T, Hayashi A, Kadota Y, Shiono H, Inoue M, Sawabata N, et al. 2009. Regulation of Th-POK and Runx3 in T cell development in human thymoma. Autoimmunity 42(8): 653-660. 13. Klunker S, Chong MM, Mantel PY, Palomares O, Bassin C, Ziegler M, et al. 2009. Transcription factors RUNX1 and RUNX3 in the induction and suppressive function of Foxp3+ inducible regulatory T cells. The Journal of experimental medicine 206(12): 2701-2715. 14. Sanchez-Martin L, Estecha A, Samaniego R, Sanchez-Ramon S, Vega MA, Sanchez-Mateos P. 2011. The chemokine CXCL12 regulates monocyte- macrophage differentiation and RUNX3 expression. Blood 117(1): 88-97. 182! ! ! 15. Chen JS, Kuo YB, Chou YP, Chan CC, Fan CW, Chen KT, et al. 2011. Detection of autoantibodies against Rabphilin-3A-like protein as a potential biomarker in patient's sera of colorectal cancer. Clinica chimica acta; international journal of clinical chemistry 412(15-16): 1417-1422. 16. Chen LF. 2012. Tumor suppressor function of RUNX3 in breast cancer. Journal of cellular biochemistry 113(5): 1470-1477. 17. He L, Zhao X, Wang H, Zhang P, Guo C, Huang C, et al. 2012. RUNX3 mediates suppression of tumor growth and metastasis of human CCRCC by regulating cyclin related proteins and TIMP-1. PLoS One 7(3): e32961. 18. Ito K, Lim AC, Salto-Tellez M, Motoda L, Osato M, Chuang LS, et al. 2008. RUNX3 attenuates beta-catenin/T cell factors in intestinal tumorigenesis. Cancer cell 14(3): 226-237. 19. Li QL, Ito K, Sakakura C, Fukamachi H, Inoue K, Chi XZ, et al. 2002. Causal relationship between the loss of RUNX3 expression and gastric cancer. Cell 109(1): 113-124. 20. Lu XX, Yu JL, Ying LS, Han J, Wang S, Yu QM, et al. 2012. Stepwise cumulation of RUNX3 methylation mediated by Helicobacter pylori infection contributes to gastric carcinoma progression. Cancer 118(22): 5507-5517. 21. Tang GH, Sun SW, He XS. 2012. [Correlation of CpG methylation status of Runx3 with pathogenesis of gastric carcinoma]. Zhonghua bing li xue za zhi Chinese journal of pathology 41(5): 314-319. 183! ! ! 22. Vogiatzi P, De Falco G, Claudio PP, Giordano A. 2006. How does the human RUNX3 gene induce apoptosis in gastric cancer? Latest data, reflections and reactions. Cancer biology & therapy 5(4): 371-374. 23. Xiao WH, Liu WW. 2004. Hemizygous deletion and hypermethylation of RUNX3 gene in hepatocellular carcinoma. World journal of gastroenterology : WJG 10(3): 376-380. 24. Mahapatra S, Klee EW, Young CY, Sun Z, Jimenez RE, Klee GG, et al. 2012. Global methylation profiling for risk prediction of prostate cancer. Clin Cancer Res 18(10): 2882-2895. 25. Yanagawa N, Tamura G, Oizumi H, Takahashi N, Shimazaki Y, Motoyama T. 2003. Promoter hypermethylation of tumor suppressor and tumor-related genes in non-small cell lung cancers. Cancer science 94(7): 589-592. 26. Wolff EM, Liang G, Cortez CC, Tsai YC, Castelao JE, Cortessis VK, et al. 2008. RUNX3 methylation reveals that bladder tumors are older in patients with a history of smoking. Cancer Res 68(15): 6208-6214. 27. Fainaru O, Shseyov D, Hantisteanu S, Groner Y. 2005. Accelerated chemokine receptor 7-mediated dendritic cell migration in Runx3 knockout mice and the spontaneous development of asthma-like disease. Proceedings of the National Academy of Sciences of the United States of America 102(30): 10598-10603. 28. Fainaru O, Woolf E, Lotem J, Yarmus M, Brenner O, Goldenberg D, et al. 2004. Runx3 regulates mouse TGF-beta-mediated dendritic cell function and its absence results in airway inflammation. The EMBO journal 23(4): 969-979. 184! ! ! 29. Singh KK, Ge X, Mao Y, Drane L, Meletis K, Samuels BA, et al. 2010. Dixdc1 is a critical regulator of DISC1 and embryonic cortical development. Neuron 67(1): 33-48. 30. Singh SP, Barrett EG, Kalra R, Razani-Boroujerdi S, Langley RJ, Kurup V, et al. 2003. Prenatal cigarette smoke decreases lung cAMP and increases airway hyperresponsiveness. American journal of respiratory and critical care medicine 168(3): 342-347. 31. Wongtrakool C, Wang N, Hyde DM, Roman J, Spindel ER. 2012. Prenatal nicotine exposure alters lung function and airway geometry through alpha7 nicotinic receptors. American journal of respiratory cell and molecular biology 46(5): 695-702. 32. Lux AL, Henderson AJ, Pocock SJ. 2000. Wheeze associated with prenatal tobacco smoke exposure: a prospective, longitudinal study. ALSPAC Study Team. Archives of disease in childhood 83(4): 307-312. 33. Haberg SE, Stigum H, Nystad W, Nafstad P. 2007. Effects of pre- and postnatal exposure to parental smoking on early childhood respiratory health. Am J Epidemiol 166(6): 679-686. 34. Lannero E, Wickman M, Pershagen G, Nordvall L. 2006. Maternal smoking during pregnancy increases the risk of recurrent wheezing during the first years of life (BAMSE). Respiratory research 7: 3. 35. Magnusson LL, Olesen AB, Wennborg H, Olsen J. 2005. Wheezing, asthma, hayfever, and atopic eczema in childhood following exposure to tobacco smoke in 185! ! ! fetal life. Clinical and experimental allergy : journal of the British Society for Allergy and Clinical Immunology 35(12): 1550-1556. 36. Gilliland FD, Berhane K, Li YF, Rappaport EB, Peters JM. 2003. Effects of early onset asthma and in utero exposure to maternal smoking on childhood lung function. American journal of respiratory and critical care medicine 167(6): 917- 924. 37. Stein RT, Holberg CJ, Sherrill D, Wright AL, Morgan WJ, Taussig L, et al. 1999. Influence of parental smoking on respiratory symptoms during the first decade of life: the Tucson Children's Respiratory Study. Am J Epidemiol 149(11): 1030- 1037. 38. Prabhu N, Smith N, Campbell D, Craig LC, Seaton A, Helms PJ, et al. 2010. First trimester maternal tobacco smoking habits and fetal growth. Thorax 65(3): 235- 240. 39. Tong VT, Jones JR, Dietz PM, D'Angelo D, Bombard JM. 2009. Trends in smoking before, during, and after pregnancy - Pregnancy Risk Assessment Monitoring System (PRAMS), United States, 31 sites, 2000-2005. MMWR Surveill Summ 58(4): 1-29. 40. Puig-Kroger A, Aguilera-Montilla N, Martinez-Nunez R, Dominguez-Soto A, Sanchez-Cabo F, Martin-Gayo E, et al. 2010. The novel RUNX3/p33 isoform is induced upon monocyte-derived dendritic cell maturation and downregulates IL-8 expression. Immunobiology 215(9-10): 812-820. 186! ! ! 41. Zilberman D, Gehring M, Tran RK, Ballinger T, Henikoff S. 2007. Genome-wide analysis of Arabidopsis thaliana DNA methylation uncovers an interdependence between methylation and transcription. Nature genetics 39(1): 61-69. 42. Zhang X, Yazaki J, Sundaresan A, Cokus S, Chan SW, Chen H, et al. 2006. Genome-wide high-resolution mapping and functional analysis of DNA methylation in arabidopsis. Cell 126(6): 1189-1201. 43. Ball MP, Li JB, Gao Y, Lee JH, LeProust EM, Park IH, et al. 2009. Targeted and genome-scale strategies reveal gene-body methylation signatures in human cells. Nature biotechnology 27(4): 361-368. 44. Hellman A, Chess A. 2007. Gene body-specific methylation on the active X chromosome. Science 315(5815): 1141-1143. 45. Rauch TA, Wu X, Zhong X, Riggs AD, Pfeifer GP. 2009. A human B cell methylome at 100-base pair resolution. Proceedings of the National Academy of Sciences of the United States of America 106(3): 671-678. 46. Serandour AA, Avner S, Oger F, Bizot M, Percevault F, Lucchetti-Miganeh C, et al. 2012. Dynamic hydroxymethylation of deoxyribonucleic acid marks differentiation-associated enhancers. Nucleic acids research 40(17): 8255-8265. 47. Olkhov-Mitsel E, Bapat B. 2012. Strategies for discovery and validation of methylated and hydroxymethylated DNA biomarkers. Cancer medicine 1(2): 237- 260. 48. Illumina. 2013. Illumina HumanMethylation450 BeadChip Kit. Available: http://www.illumina.com/products/methylation_450_beadchip_kits.ilmn [accessed 2013 Mar 7 2013]. 187! ! ! 49. Ohshige T, Iwata M, Omori S, Tanaka Y, Hirose H, Kaku K, et al. 2011. Association of new loci identified in European genome-wide association studies with susceptibility to type 2 diabetes in the Japanese. PLoS One 6(10): e26911. 50. Cha JD, Kim HJ, Cha IH. 2011. Genetic alterations in oral squamous cell carcinoma progression detected by combining array-based comparative genomic hybridization and multiplex ligation-dependent probe amplification. Oral surgery, oral medicine, oral pathology, oral radiology, and endodontics 111(5): 594-607. 51. Shimada H, Nakashima K, Ochiai T, Nabeya Y, Takiguchi M, Nomura F, et al. 2005. Serological identification of tumor antigens of esophageal squamous cell carcinoma. International journal of oncology 26(1): 77-86. 52. Krzeslak A, Forma E, Chwatko G, Jozwiak P, Szymczyk A, Wilkosz J, et al. 2013. Effect of metallothionein 2A gene polymorphism on allele-specific gene expression and metal content in prostate cancer. Toxicology and applied pharmacology. 53. Zhou CF, Ji J, Yuan F, Yu YY, Liu BY, Zhang J, et al. 2012. [Expression of metastasis associated 1 family member 2 (MTA2) in gastric cancer and its relationship with transcription factor Sp1]. Zhonghua Zhong Liu Za Zhi 34(8): 592-595. 54. Liu SL, Han Y, Zhang Y, Xie CY, Wang EH, Miao Y, et al. 2012. Expression of metastasis-associated protein 2 (MTA2) might predict proliferation in non-small cell lung cancer. Targeted oncology 7(2): 135-143. 188! ! ! 55. Liu YP, Shan BE, Wang XL, Ma L. 2012. Correlation between MTA2 overexpression and tumour progression in esophageal squamous cell carcinoma. Experimental and therapeutic medicine 3(4): 745-749. 56. Lee H, Ryu SH, Hong SS, Seo DD, Min HJ, Jang MK, et al. 2009. Overexpression of metastasis-associated protein 2 is associated with hepatocellular carcinoma size and differentiation. Journal of gastroenterology and hepatology 24(8): 1445-1450. 57. Ji Y, Zhang P, Lu Y, Ma D. 2006. Expression of MTA2 gene in ovarian epithelial cancer and its clinical implication. Journal of Huazhong University of Science and Technology Medical sciences = Hua zhong ke ji da xue xue bao Yi xue Ying De wen ban = Huazhong keji daxue xuebao Yixue Yingdewen ban 26(3): 359-362. 58. Mao J, Yu H, Wang C, Sun L, Jiang W, Zhang P, et al. 2012. Metallothionein MT1M is a tumor suppressor of human hepatocellular carcinomas. Carcinogenesis 33(12): 2568-2577. 59. Bradshaw NJ, Porteous DJ. 2012. DISC1-binding proteins in neural development, signalling and schizophrenia. Neuropharmacology 62(3): 1230-1241. 60. Sim K, Chan WY, Woon PS, Low HQ, Lim L, Yang GL, et al. 2012. ARVCF genetic influences on neurocognitive and neuroanatomical intermediate phenotypes in Chinese patients with schizophrenia. The Journal of clinical psychiatry 73(3): 320-326. 61. Mas S, Bernardo M, Gasso P, Alvarez S, Garcia-Rizo C, Bioque M, et al. 2010. A functional variant provided further evidence for the association of ARVCF with schizophrenia. American journal of medical genetics Part B, Neuropsychiatric 189! ! ! genetics : the official publication of the International Society of Psychiatric Genetics 153B(5): 1052-1059. 62. Mas S, Bernardo M, Parellada E, Garcia-Rizo C, Gasso P, Alvarez S, et al. 2009. ARVCF single marker and haplotypic association with schizophrenia. Progress in neuro-psychopharmacology & biological psychiatry 33(6): 1064-1069. 63. Chen HY, Yeh JI, Hong CJ, Chen CH. 2005. Mutation analysis of ARVCF gene on chromosome 22q11 as a candidate for a schizophrenia gene. Schizophrenia research 72(2-3): 275-277. 64. Neale BM, Medland S, Ripke S, Anney RJ, Asherson P, Buitelaar J, et al. 2010. Case-control genome-wide association study of attention-deficit/hyperactivity disorder. Journal of the American Academy of Child and Adolescent Psychiatry 49(9): 906-920. 65. Karlsson R, Graae L, Lekman M, Wang D, Favis R, Axelsson T, et al. 2012. MAGI1 copy number variation in bipolar affective disorder and schizophrenia. Biological psychiatry 71(10): 922-930. 66. Hatayama M, Ishiguro A, Iwayama Y, Takashima N, Sakoori K, Toyota T, et al. 2011. Zic2 hypomorphic mutant mice as a schizophrenia model and ZIC2 mutations identified in schizophrenia patients. Scientific reports 1: 16. 67. Russell JF, Steckley JL, Coppola G, Hahn AF, Howard MA, Kornberg Z, et al. 2012. Familial cortical myoclonus with a mutation in NOL3. Annals of neurology 72(2): 175-183. 68. Bennetts JS, Rendtorff ND, Simpson F, Tranebjaerg L, Wicking C. 2007. The coding region of TP53INP2, a gene expressed in the developing nervous system, 190! ! ! is not altered in a family with autosomal recessive non-progressive infantile ataxia on chromosome 20q11-q13. Developmental dynamics : an official publication of the American Association of Anatomists 236(3): 843-852. 69. Mead S, Uphill J, Beck J, Poulter M, Campbell T, Lowe J, et al. 2012. Genome- wide association study in multiple human prion diseases suggests genetic risk factors additional to PRNP. Human molecular genetics 21(8): 1897-1906. 70. Hu Y, Chopra V, Chopra R, Locascio JJ, Liao Z, Ding H, et al. 2011. Transcriptional modulator H2A histone family, member Y (H2AFY) marks Huntington disease activity in man and mouse. Proceedings of the National Academy of Sciences of the United States of America 108(41): 17141-17146. 71. Cong WN, Cai H, Wang R, Daimon CM, Maudsley S, Raber K, et al. 2012. Altered hypothalamic protein expression in a rat model of Huntington's disease. PLoS One 7(10): e47240. 72. Tian LP, Zhang S, Zhang YJ, Ding JQ, Chen SD. 2012. Lmx1b can promote the differentiation of embryonic stem cells to dopaminergic neurons associated with Parkinson's disease. Biotechnology letters 34(7): 1167-1174. 73. Suda S, Iwata K, Shimmura C, Kameno Y, Anitha A, Thanseem I, et al. 2011. Decreased expression of axon-guidance receptors in the anterior cingulate cortex in autism. Molecular autism 2(1): 14. 74. Lin PI, Chien YL, Wu YY, Chen CH, Gau SS, Huang YS, et al. 2012. The WNT2 gene polymorphism associated with speech delay inherent to autism. Research in developmental disabilities 33(5): 1533-1540. 191! ! ! 75. Kalkman HO. 2012. A review of the evidence for the canonical Wnt pathway in autism spectrum disorders. Molecular autism 3(1): 10. 76. Liu J, Bann C, Lester B, Tronick E, Das A, Lagasse L, et al. 2010. Neonatal neurobehavior predicts medical and behavioral outcome. Pediatrics 125(1): e90- 98. 77. Lettice LA, Daniels S, Sweeney E, Venkataraman S, Devenney PS, Gautier P, et al. 2011. Enhancer-adoption as a mechanism of human developmental disease. Human mutation 32(12): 1492-1499. 78. Herrmann M, King K, Weitzman M. 2008. Prenatal tobacco smoke and postnatal secondhand smoke exposure and child neurodevelopment. Current opinion in pediatrics 20(2): 184-190. 79. McCarthy RA, Argraves WS. 2003. Megalin and the neurodevelopmental biology of sonic hedgehog and retinol. Journal of cell science 116(Pt 6): 955-960. 80. Ho KS, Scott MP. 2002. Sonic hedgehog in the nervous system: functions, modifications and mechanisms. Current opinion in neurobiology 12(1): 57-63. 81. Marsit CJ, Lambertini L, Maccani MA, Koestler DC, Houseman EA, Padbury JF, et al. 2012. Placenta-imprinted gene expression association of infant neurobehavior. The Journal of pediatrics 160(5): 854-860 e852. 82. Watson GE, Evans K, Thurston SW, van Wijngaarden E, Wallace JM, McSorley EM, et al. 2012. Prenatal exposure to dental amalgam in the Seychelles Child Development Nutrition Study: associations with neurodevelopmental outcomes at 9 and 30 months. Neurotoxicology 33(6): 1511-1517. 192! ! ! 83. Clarkson TW. 1997. The toxicology of mercury. Critical reviews in clinical laboratory sciences 34(4): 369-403. 84. Zota AR, Ettinger AS, Bouchard M, Amarasiriwardena CJ, Schwartz J, Hu H, et al. 2009. Maternal blood manganese levels and infant birth weight. Epidemiology 20(3): 367-373. 85. Mistry HD, Williams PJ. 2011. The importance of antioxidant micronutrients in pregnancy. Oxidative medicine and cellular longevity 2011: 841749. 86. Misselwitz B, Muhler A, Weinmann HJ. 1995. A toxicologic risk for using manganese complexes? A literature survey of existing data through several medical specialties. Investigative radiology 30(10): 611-620. 87. Takser L, Mergler D, Hellier G, Sahuquillo J, Huel G. 2003. Manganese, monoamine metabolite levels at birth, and child psychomotor development. Neurotoxicology 24(4-5): 667-674. 88. Claus Henn B, Ettinger AS, Schwartz J, Tellez-Rojo MM, Lamadrid-Figueroa H, Hernandez-Avila M, et al. 2010. Early postnatal blood manganese levels and children's neurodevelopment. Epidemiology 21(4): 433-439. 89. Roels HA, Bowler RM, Kim Y, Claus Henn B, Mergler D, Hoet P, et al. 2012. Manganese exposure and cognitive deficits: a growing concern for manganese neurotoxicity. Neurotoxicology 33(4): 872-880. 90. Ericson JE, Crinella FM, Clarke-Stewart KA, Allhusen VD, Chan T, Robertson RT. 2007. Prenatal manganese levels linked to childhood behavioral disinhibition. Neurotoxicology and teratology 29(2): 181-187. 193! ! ! 91. Wozniak A, Kujawa A, Senczuk-Przybylowska M, Kulza M, Gawecki W, Szybiak B, et al. 2012. [Physiological metals in the serum, hair and nails of patients with head and neck cancer]. Przeglad lekarski 69(10): 785-797. 92. Karimi G, Shahar S, Homayouni N, Rajikan R, Abu Bakar NF, Othman MS. 2012. Association between trace element and heavy metal levels in hair and nail with prostate cancer. Asian Pacific journal of cancer prevention : APJCP 13(9): 4249-4253. 93. Cantone L, Nordio F, Hou L, Apostoli P, Bonzini M, Tarantini L, et al. 2011. Inhalable metal-rich air particles and histone H3K4 dimethylation and H3K9 acetylation in a cross-sectional study of steel workers. Environmental health perspectives 119(7): 964-969. 94. Goodrich JM, Basu N, Franzblau A, Dolinoy DC. 2013. Mercury biomarkers and DNA methylation among Michigan dental professionals. Environmental and molecular mutagenesis 54(3): 195-203. 95. Baccarelli A, Bollati V. 2009. Epigenetics and environmental chemicals. Current opinion in pediatrics 21(2): 243-251. 96. Desotelle J, Truong M, Ewald J, Weeratunga P, Yang B, Huang W, et al. 2013. CpG island hypermethylation frequently silences FILIP1L isoform 2 expression in prostate cancer. The Journal of urology 189(1): 329-335. 97. Rousseaux S, Debernardi A, Jacquiau B, Vitte AL, Vesin A, Nagy-Mignotte H, et al. 2013. Ectopic activation of germline and placental genes identifies aggressive metastasis-prone lung cancers. Science translational medicine 5(186): 186ra166. 194! ! ! 98. Bunin GR, Li Y, Ganguly A, Meadows AT, Tseng M. 2013. Parental nutrient intake and risk of retinoblastoma resulting from new germline RB1 mutation. Cancer causes & control : CCC 24(2): 343-355. 99. Maccani MA, Padbury JF, Lester BM, Knopik VS, Marsit CJ. In Press. Placental miRNA expression profiles associated with measures of infant neurobehavioral outcomes. Pediatric research. 195! ! !