RECURRENT MISCARRIAGE - Open Collections - University of British ...

RECURRENT MISCARRIAGE: UNRAVELING THE COMPLEX ETIOLOGY

by Courtney Wood Hanna

B.Sc., The University of British Columbia, 2006

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Medical Genetics)

THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)

April 2013

© Courtney Wood Hanna, 2013

Abstract Recurrent miscarriage (RM), defined as 3 or more consecutive spontaneous losses of pregnancy before 20 weeks gestation, affects 1-2% of couples and has a complex etiology. Half of miscarriages from RM cases are caused by chromosomal abnormalities in the embryo and while there are several associated maternal factors, underlying causes and clinically relevant biomarkers have been elusive. I hypothesized that genetic and/or epigenetic factors associated with maternal meiotic non-disjunction, reproductive aging and endocrinological profile, or placental functioning will contribute to the etiology of RM. In these case-control studies, I investigated the association between RM and 1) maternal mutations in synaptonemal complex protein 3 (SYCP3), 2) maternal telomere lengths, 3) maternal polymorphisms in genes in the hypothalamus-pituitary-ovarian (HPO) axis and 4) placental DNA methylation patterns. The findings suggest that maternal mutations in SYCP3 and polymorphisms in HPO axis genes may not contribute significantly to risk for RM. No mutations in SYCP3 were identified in women with RM with at least one trisomic conception. While associations between polymorphisms within the estrogen receptor β, activin receptor 1, prolactin receptor and glucocorticoid receptor genes and RM were identified, these were not significant after correction for multiple comparisons. Aspects of chromosomal biology may be important factors in the etiology of RM. Women with RM had significantly shorter telomeres compared to controls, suggesting altered rates of biological aging. In the placental villi of RM samples, there were few differences in DNA methylation at targeted sites when compared to isolated miscarriages and elective terminations. However, gene ontology analysis showed that imprinted genes and immune response pathways were overrepresented among those sites differentially methylated between RM and elective termination placentas. The RM group additionally had an increase in the ii

number of outlier cases at a select number of imprinted loci. Furthermore, several placental samples from both cases and controls showed aberrant DNA methylation profiles at many loci investigated, suggesting these samples may have global dysregulation of DNA methylation and/or differences in placental composition/functioning. These studies have improved our understanding of mechanisms involved in RM and will contribute to the direction of future research.

iii

Preface A version of Chapter 2 has been published. Hanna, C.W., Blair, J.D., Stephenson, M.D. and Robinson, W.P. (2012) Absence of SYCP3 mutations in women with recurrent miscarriage with at least one trisomic miscarriage. Reproductive BioMedicine Online. 24(2):251-3. I generated the hypothesis and study design with W.P. Robinson, directly supervised summer student J.D. Blair, and personally wrote the manuscript. J.D. Blair contributed equally to this publication by performing data collection, analyzing the results, and editing the manuscript. M.D. Stephenson ascertained patients. W.P. Robinson supervised the research and edited the manuscript. A version of Chapter 3 has been published. Hanna, C.W., Bretherick, K.L., Gair, J.L., Fluker, M.R., Stephenson, M.D. and Robinson, W.P. (2009) Telomere length and reproductive aging. Human Reproduction. 24(5):1206-11. K.L. Bretherick and I contributed equally to the collection of data, analysis of results and preparation of the manuscript. M.R. Fluker and M.D. Stephenson ascertained patients. W.P. Robinson supervised this project, and generated the hypothesis for this work with J.L. Gair. A version of Chapter 4 has been published. Hanna, C.W., Bretherick, K.L., Liu, C.C., Stephenson, M.D. and Robinson, W.P. (2010) Genetic variation in the hypothalamus-pituitaryovarian axis in women with recurrent miscarriage. Human Reproduction. 25(10):2664-71. I generated the study design, performed ~80% of data collection, analyzed the results, and wrote the manuscript. K.L. Bretherick performed 10% of data collection and assisted with hypothesis formation and manuscript editing. I supervised summer student C.C. Liu in completing the remaining 10% of data collection, in addition to analyzing this subset of the results. M.D.

iv

Stephenson ascertained patients. W.P. Robinson supervised the research and edited the manuscript. A version of Chapter 5 has been published. Hanna, C.W., McFadden, D.E. and Robinson, W.P. (2013) DNA methylation profiling of placental villi from karyotypically normal miscarriage and recurrent miscarriage. American Journal of Pathology. Epub ahead of print 2013 April 09. I generated the hypothesis and study design with W.P. Robinson, completed all data collection, analyzed the results, and wrote the manuscript. D.E. McFadden ascertained patients and did sample collection. W.P. Robinson additionally supervised the research and edited the manuscript. The collection of the samples for these studies was approved by the University of British Columbia Clinical Research Ethics Board, approval number CO1-0460. Copyright permission was obtained for all published figures, tables and texts.

v

Table of Contents Abstract .......................................................................................................................................... ii Preface ........................................................................................................................................... iv Table of Contents ......................................................................................................................... vi List of Tables ..................................................................................................................................x List of Figures ............................................................................................................................... xi List of Abbreviations .................................................................................................................. xii Acknowledgements ......................................................................................................................xv Chapter 1: Introduction ...............................................................................................................1 1.1

Aneuploidy in miscarriage .............................................................................................. 1

1.1.1 Oogenesis and meiosis ................................................................................................ 2 1.1.2 Maternal risk for chromosome missegregation........................................................... 4 1.2

Maternal factors associated with recurrent miscarriage.................................................. 6

1.2.1 Chromosomal .............................................................................................................. 6 1.2.2 Anatomical .................................................................................................................. 7 1.2.3 Immunological ............................................................................................................ 7 1.2.3.1

Uterine natural killer cells ................................................................................... 8

1.2.3.2

Maternal T helper 1 and T helper 2 immune balance ......................................... 9

1.2.3.3

Placental immunity ........................................................................................... 10

1.2.3.4

Autoimmunity ................................................................................................... 11

1.2.3.5

Infection ............................................................................................................ 12

1.2.4 Endocrinological ....................................................................................................... 12 1.2.4.1

Luteal phase defects .......................................................................................... 12 vi

1.2.4.2

Thyroid dysfunction .......................................................................................... 13

1.2.4.3

Polycystic ovarian syndrome ............................................................................ 13

1.2.4.4

Endometrial receptivity..................................................................................... 14

1.2.5 Thrombophilic........................................................................................................... 15 1.2.6 Psychosocial stress .................................................................................................... 16 1.3

Genetic and epigenetic factors contributing to recurrent miscarriage risk ................... 17

1.3.1 Maternal genetic variants .......................................................................................... 17 1.3.1.1

Genes involved in meiosis ................................................................................ 17

1.3.1.2

Genes involved in immune function ................................................................. 18

1.3.1.3

Genes involved in endocrine function .............................................................. 18

1.3.1.4

Thrombophilia associated genes ....................................................................... 19

1.3.2 Telomeres .................................................................................................................. 21 1.3.3 Epigenetics in fetal development .............................................................................. 21 1.3.4 Maternal skewed X-chromosome inactivation ......................................................... 23 1.4

Research objectives ....................................................................................................... 23

Chapter 2: Mutational analysis of the SYCP3 gene .................................................................29 2.1

Introduction ................................................................................................................... 29

2.2

Materials and methods .................................................................................................. 29

2.3

Results ........................................................................................................................... 30

2.4

Discussion ..................................................................................................................... 31

Chapter 3: Telomere length and reproductive aging ..............................................................36 3.1

Introduction ................................................................................................................... 36

3.2

Materials and methods .................................................................................................. 38 vii

3.2.1 Samples ..................................................................................................................... 38 3.2.2 Telomere length ........................................................................................................ 39 3.2.3 Statistical analysis ..................................................................................................... 40 3.3

Results ........................................................................................................................... 40

3.4

Discussion ..................................................................................................................... 41

Chapter 4: Genetic polymorphisms in genes involved in the hypothalamus-pituitaryovarian axis ...................................................................................................................................52 4.1

Introduction ................................................................................................................... 52

4.2

Materials and methods .................................................................................................. 53

4.2.1 Samples ..................................................................................................................... 53 4.2.2 Variant selection ....................................................................................................... 54 4.2.3 Genotyping ................................................................................................................ 54 4.2.4 Statistical analysis ..................................................................................................... 55 4.2.5 Population stratification ............................................................................................ 55 4.3

Results ........................................................................................................................... 55

4.4

Discussion ..................................................................................................................... 57

Chapter 5: Placental DNA methylation associated with pregnancy outcomes .....................70 5.1

Introduction ................................................................................................................... 70

5.2

Materials and methods .................................................................................................. 71

5.2.1 Samples ..................................................................................................................... 71 5.2.2 Array-based quantification of DNA methylation ..................................................... 71 5.2.3 Targeted DNA methylation....................................................................................... 72 5.2.4 Statistical analysis ..................................................................................................... 73 viii

5.3

Results ........................................................................................................................... 74

5.3.1 Array-based quantification of DNA methylation ..................................................... 74 5.3.2 DNA methylation at imprinted genes ....................................................................... 76 5.3.3 ‘Global’ measures of DNA methylation ................................................................... 77 5.4

Discussion ..................................................................................................................... 78

Chapter 6: Discussion .................................................................................................................93 6.1

Summary and significance of findings ......................................................................... 93

6.2

Strengths and limitations............................................................................................... 96

6.3

Future directions ........................................................................................................... 98

6.4

Conclusions ................................................................................................................. 101

References ...................................................................................................................................103 Appendix A: Supplementary tables and figures for Chapter 2 .............................................125 Appendix B: Supplementary tables and figures for Chapter 4 .............................................126 Appendix C: Supplementary tables and figures for Chapter 5 .............................................144

ix

List of Tables Table 2.1 Outcomes of 292 pregnancies from 50 women with recurrent miscarriage. ................ 33 Table 2.2 Minor allele frequencies of noncoding single nucleotide polymorphisms within the SYCP3 gene identified in 50 recurrent miscarriage women. ........................................................ 34 Table 3.1 Rate of telomere loss per year in women with evidence of premature reproductive aging and controls ......................................................................................................................... 46 Table 3.2 Raw and age-adjusted mean telomere length ............................................................... 47 Table 4.1 Summary of 35 polymorphisms assessed in this study................................................. 61 Table 4.2 Allele distributions of short tandem repeat polymorphisms. ........................................ 63 Table 4.3 Genotype distributions of single nucleotide polymorphisms. ...................................... 65 Table 5.1 Comparison of demographics for the recurrent miscarriage, miscarriage and elective termination study groups............................................................................................................... 84 Table 5.2 Frequency of outliers at imprinted loci. ........................................................................ 85 Table 5.3 Patterns of DNA methylation among outlier samples. ................................................. 86

x

List of Figures Figure 1.1 Gametes as the product of meiosis I non-disjunction and meiosis II non-disjunction. 25 Figure 1.2 Number of germ cells with age in females. ................................................................. 26 Figure 1.3 Hormone levels and follicular events during the menstrual cycle. ............................. 27 Figure 1.4 Etiology of recurrent miscarriage. ............................................................................... 28 Figure 2.1 Schematic diagram of the SYCP3 gene and variants. .................................................. 35 Figure 3.1 Telomere-specific qPCR. ............................................................................................ 48 Figure 3.2 Correlation between telomere-specific qPCR and flow-FISH techniques. ................. 49 Figure 3.3 Correlation between telomere length and age in women with evidence of premature reproductive aging and controls. ................................................................................................... 50 Figure 3.4 Correlation between telomere length and age in women with recurrent miscarriage and trisomic pregnancies............................................................................................................... 51 Figure 5.1 Venn diagram of significant Infinium array candidates. ............................................. 87 Figure 5.2 DNA methylation at 4 candidate promoter regions. .................................................... 88 Figure 5.3 Unsupervised clustering of the 20 samples run on the Infinium array. ....................... 89 Figure 5.4 Box plots of DNA methylation at 7 imprinted loci. .................................................... 90 Figure 5.5 Comparison of measures of ‘global’ methylation. ...................................................... 91 Figure 5.6 Principle component plot of all samples. .................................................................... 92

xi

List of Abbreviations ACVR1

activin receptor type 1

ANCOVA

analysis of covariance

APA

antiphospholipid antibody

APC

adenomous polyposis coli

APS

antiphospholipid syndrome

AXL

AXL receptor tyrosine kinase

CD

cluster of differentiation

CI

confidence intervals

CYP1A2

cytochrome P450, subfamily 1A, polypeptide 2

DEFB1

defensin beta 1

DMR

differentially methylated region

ESR1

estrogen receptor α

ESR2

estrogen receptor β

FDR

false discovery rate

FISH

fluorescence in situ hybridization

FSH

follicle-stimulating hormone

GCCR

glucocorticoid receptor

GnRH

gonadotropin releasing hormone

HLA

human leukocyte antigen

HPO

hypothalamus-pituitary-ovarian

HWE

Hardy-Weinberg equilibrium

HY

male-specific histocompatibility antigen xii

IL

interleukin

IVF

in vitro fertilization

kb

kilobases

LD

linkage disequilibrium

LINE-1

long interspersed element

LH

luteinizing hormone

LPD

luteal phase defects

MHC

major histocompatibility complex

M

miscarriage

MI

first meiotic division

MII

second meiotic division

MLPA

multiple ligation-dependent probe amplification

MT

multiple trisomic miscarriages

MTHFR

methylenetetrahydrofolate reductase

NK

natural killer

OR

odds ratio

PCA

principle component analysis

PCOS

polycystic ovarian syndrome

POF

premature ovarian failure

PRLR

prolactin receptor

qPCR

quantitative polymerase chain reaction

RM

recurrent miscarriage

ROS

reactive oxidative species xiii

SD

standard deviation

SLE

systemic lupus erythematosus

SNP

single nucleotide polymorphism

ST

single trisomic miscarriage

STR

short tandem repeat

SYCP3

synaptonemal complex protein 3

TA

termination

T/S ratio

telomere to single copy ratio

Th

T helper

TH

thyroid hormones

uNK

uterine natural killer

UTR

untranslated region

XCI

X-chromosome inactivation

xiv

Acknowledgements I would like to acknowledge my supervisor, Dr. Wendy Robinson, for the encouragement to pursue my research interests, invaluable direction and feedback on project design and manuscript generation, and investing time and energy in my professional development. I would also like to thank my committee, Drs. Carolyn Brown, Michael Kobor and Barbara McGillivray for their advice and support. I owe a particular thanks to Dr. Mary Stephenson, whose clinical expertise has greatly improved the impact of my research, and Dr. Maria Peñaherrera, whose continual advice and encouragement have been so important to my success. To all the current and past lab members who have helped me along the way, Karla, Ruby, Luana, Sara, Ryan, Dan, Magda, John, Kirsten and Irina, thank you. To my family, mom, dad, grandma, grandpa, granny, Shawn, Matt and Amy, thank you for your enduring love and support. Finally Nick, you have invested so much interest in my work and offered support and encouragement through my highs and lows, thank you for believing in me. Thank you to the funding organizations for financial support, including the Canadian Institutes of Health Research, University of British Columbia and Interdisciplinary Women’s Reproductive Health program.

xv

Chapter 1: Introduction Miscarriage, a spontaneous abortion before 20 weeks gestation, occurs in 15% of clinically-recognized pregnancies, making it the most common complication of pregnancy (Warburton and Fraser 1964). Recurrent miscarriage (RM), defined as 3 or more consecutive miscarriages, affects 1-2% of couples trying to conceive (Stirrat 1990), which is greater than expected by chance (0.34%). Approximately 50% of clinically-recognized miscarriages among RM couples have numerical chromosomal abnormalities, with the vast majority being aneuploid (Ogasawara et al. 2000, Stephenson et al. 2002). This rate is even higher among non-recurring miscarriages, with frequencies reported from 50-70% (Hassold and Chiu 1985, Ogasawara et al. 2000). The etiology of RM is complex with many associated maternal factors. The diagnostic value of these associated factors is unclear as they are often identified in women with healthy pregnancies and there is currently few recommended therapeutics for women with RM and a coexisting factor (Tang and Quenby 2010). Consequently, RM is an extremely stressful condition for couples and physicians and is an important area of research. In the introduction to this thesis, the complex etiology of RM will be described as it is currently understood, with an emphasis on the areas of research that can be expanded upon. The main topics of discussion will be: 1) aneuploidy in miscarriage, 2) maternal factors associated with RM, and 3) genetic and epigenetic factors contributing to risk for RM. 1.1

Aneuploidy in miscarriage Aneuploidy, the loss or gain of a chromosome, can arise through missegregation of

chromosomes during meiosis. This can occur through non-disjunction of the homologous chromosome pairs or premature separation of the sister chromatids during the first meiotic 1

division (MI), or missegregation of sister chromatids during the second meiotic division (MII) (Figure 1.1). In addition, errors can arise postzygotically, often in the first few cell divisions after fertilization (Bean et al. 2001). While the contribution of postzygotic, maternal and paternal meiotic errors for each chromosome differs, among miscarriage specimens most are of maternal meiotic origin (Hassold and Hunt 2001). In this section, I will provide an overview of oogenesis and discuss the aspects that make this process particularly susceptible to errors in chromosome segregation. 1.1.1

Oogenesis and meiosis During early female fetal development, primordial germ cells migrate to the gonadal

ridge, which later develops into the fetal ovaries. Oogenesis begins with a vast number of mitotic divisions, giving rise to ~7 million cells by the fifth month of gestation (Baker 1971). At this time, the primary oocytes enter MI and are arrested during the diplotene phase of prophase I. They remain in this state until just before ovulation decades later. Before birth, each primary oocyte is surrounded by a single layer of granulosa cells, forming a primary follicle. By the time a female fetus reaches birth, the number of oocytes has decreased to ~2 million through apoptosis and these cells will continue to deplete until puberty (Figure 1.2) (Baker 1971). Over the course of a woman’s reproductive lifespan, oocytes will be cyclically matured and ovulated until the pool is depleted at menopause. Tilly and coauthors have recently challenged this central dogma, with studies suggesting that the ovarian capacity for oocyte production continues into adulthood (Johnson et al. 2004, White et al. 2012); however, this is highly contested in the scientific community and has yet to be independently reproduced. A secondary follicle is a matured primary follicle and is comprised of a single enlarged oocyte and surrounding layers of differentiated granulosa cells and thecal cells. Granulosa cells 2

are important for generating the protective zona pellucida, providing molecules to the oocyte, and secreting estrogens, while thecal cells have a supportive function and are the ovarian connective tissue (Lunenfeld et al. 1975). The cyclical maturation of the follicle and its contained oocyte are hormonally controlled by the hypothalamus-pituitary-ovarian (HPO) axis through two phases: the follicular phase and luteal phase. Once a follicle starts to develop, it will either reach maturity and be ovulated or degenerate and undergo atresia. The HPO axis is a feedback loop that begins with the production of gonadotropin releasing hormone (GnRH) in the hypothalamus, which in turn promotes the release of both luteinizing hormone (LH) and follicle-stimulating hormone (FSH) from the anterior pituitary gland. During the follicular phase, the action of LH and FSH is to promote the expansion and increased estrogen production of the granulosa cells in a small number of recruited follicles (Figure 1.3) (Sherwood 2004). The corresponding oocytes undergo rapid enlargement, however only one oocyte-containing secondary follicle will usually develop into a mature antral follicle for ovulation (Sherwood 2004). Initiated by the positive feedback of rising estrogen levels, an LH surge signifies the start of ovulation (Figure 1.3). As a result, there are several important changes that occur in the ovary. The follicular cells begin to differentiate into luteal cells. The maturing oocyte completes MI, extruding a polar body, and then arrests in the metaphase of MII (Sherwood 2004). The ovarian wall then ruptures, allowing the release of the mature oocyte into the fallopian tube. The luteal phase is characterized by the formation of the corpus luteum from the remaining follicle after rupture and release of the oocyte. These cells become transformed into an active steroidogenic tissue, producing primarily progesterone (Figure 1.3) (Sherwood 2004). The progesterone secretion promotes morphological and biochemical remodeling 3

(decidualization) of the endometrium in preparation for implantation (Dockery and Rogers 1989). If the oocyte is not fertilized, the corpus luteum deteriorates and the menstrual cycle begins again. If fertilized, the oocyte completes MII, extruding a second polar body. The zygote grows and differentiates into the blastocyst, as it travels down the oviduct to the endometrium for implantation. The corpus luteum produces increasing amounts of progesterone, in response to human chorionic gonadotropin produced by the developing embryo (Sherwood 2004). The production of progesterone during these early weeks of pregnancy is essential, as removal of the corpus luteum during this stage causes the spontaneous loss of pregnancy (Csapo et al. 1972). After the embryo implants, the extraembryonic cells invade the maternal decidualized endometrium. These cells aid in remodeling the maternal arteries and generate the placenta, a site for gas, waste and nutrient exchange between the mother and fetus for the remainder of pregnancy (Sherwood 2004). In the 8th week of pregnancy, the production of progesterone is taken over by the placenta, in the luteoplacental transition (Csapo et al. 1972). 1.1.2

Maternal risk for chromosome missegregation There are two primary risk factors for meiotic non-disjunction in females: 1) advanced

maternal age and 2) aberrant chromosomal recombination (Nagaoka et al. 2012). There have been several advances in the past 10 years in understanding the etiology of these risk factors. However, further investigation in human oocytes is needed to validate and elaborate on these hypotheses. It has long been established that advanced maternal age is associated with increased risk for aneuploidy of most autosomal chromosomes (Hassold and Chiu 1985). This association is hypothesized to be at least partially due to a progressive breakdown of the meiotic machinery 4

during the prolonged prophase arrest, in particular the cohesin protein complexes holding sister chromatids together. Studies using aging animal models observed reduced cohesin proteins in the oocytes of aged females and consequently increased rates of aneuploidy (Liu and Keefe 2008, Subramanian and Bickel 2008). However, there has been recent controversy as to whether the cohesin proteins in adult oocytes are in fact those that were established during fetal development or whether they are replenished during a female’s lifetime. One study found that cohesin proteins are produced in human oocytes during adulthood, suggesting there is the potential for replenishment (Garcia-Cruz et al. 2010). However, a series of experiments in mice tested whether these proteins were replaced upon destruction and found that there was no rescue of the phenotype (Tachibana-Konwalski et al. 2010). Furthermore, mice with heterozygous mutations in cohesin genes showed elevated rates of oocyte aneuploidy that increased with maternal age (Murdoch et al. 2013). Aberrant recombination, which can include both achiasmate and poorly located crossovers in MI, have been shown to result in aneuploidy due to chromosome segregation errors (Lamb et al. 2005). It has been hypothesized that oogenesis may be particularly prone to nondisjunction due to high rates of aberrant recombination in the fetal ovary and low stringency of meiotic check points during oocyte maturation. While linkage studies of chromosomes 18 and 21 suggested that the frequency of achiasmate MI in oocytes is high (Bugge et al. 1998, Oliver et al. 2008), a study of 31 human fetal ovaries found that only 1.4% of oocytes had an achiasmate chromosome pair (Cheng et al. 2009). In mice, it has been shown that meiosis progresses in oocytes despite the presence of double strand breaks due to failed repair during recombination (Kuznetsov et al. 2007). These studies suggest that recombination in fetal oocytes may not be

5

particularly error prone, but those oocytes that do have recombination errors will likely progress through meiosis, thus explaining the maternal origin of many aneuploidies. 1.2

Maternal factors associated with recurrent miscarriage The primary maternal factors associated with RM can be categorized as chromosomal,

anatomical, immunological and endocrinological; however, approximately 50% of cases are idiopathic (Figure 1.4) (Clifford et al. 1994, Stephenson 1996). Variable frequencies and inconsistent associations of these factors and RM are pervasive throughout the literature. This, in part, is not surprising given the complex etiology of RM; however, other contributors include the lack of consistency between clinical evaluations, underpowered studies and the wide interand intra-individual variability of many factors. In particular, hormone levels and immunological cell populations can change with circadian clock, menstrual cycle, pregnancy, age and tissue type; making it very difficult to match case-control populations. Furthermore, all of these abnormalities are identified in a substantial proportion of women with uncomplicated pregnancy histories, which suggests that additional environmental and/or genetic factors may contribute to risk for RM. Despite these difficulties in the study of maternal conditions associated with RM, many factors result in increased risk for pregnancy loss and are important considerations in RM patient management (Rai and Regan 2006). 1.2.1

Chromosomal Approximately 3.5% of couples with RM are carriers of a balanced chromosomal

rearrangement (Clifford et al. 1994, Stephenson 1996). A proportion of the gametes from these couples would thus be unbalanced products of meiosis. Despite the 50% expected risk of an abnormal conception, empirical evidence shows that the frequency of successful pregnancies among these couples is relatively high, with only one third of miscarriages having unbalanced 6

chromosomal rearrangements (Stephenson and Sierra 2006). This is likely because some abnormal conceptuses do not survive to implantation. 1.2.2

Anatomical Uterine structural anomalies have been variably associated with RM, with incidence

ranging from 1.8 to 16% (Clifford et al. 1994, Stephenson 1996). These anomalies can include bicornate uterus, septate uterus, intrauterine adhesions and uterine fibroids, as well as rarer abnormalities. While rates are estimated to be three times higher among women with RM than the general population (Chan et al. 2011), the frequencies reported among RM studies can be erratic due to variable inclusion criteria and imaging technology. The mechanism of how these anomalies may cause miscarriage is unknown; however, physical impedance of the progression of pregnancy or poor implantation at affected regions has been proposed (Chan et al. 2011). Despite the association between uterine anomalies and RM, many affected women do go on to have successful pregnancies (Clifford et al. 1994) and it has not been determined whether surgical treatment of these conditions improves pregnancy rates (Tang and Quenby 2010). 1.2.3

Immunological Pregnancy is accompanied by dramatic changes in the maternal immune system, to allow

the coexistence of a genetically distinct fetus. Not only are there changes in the mother, but the placental barrier also helps to suppress the maternal immune response. Both natural killer (NK) cells and T helper (Th) cells at the feto-maternal interface play a particularly important role in regulating inflammation at the time of implantation and throughout the remainder of pregnancy (Granot et al. 2012). Defects, such as altered feto-maternal immune interactions, autoimmunity and infection, have been suggested to play a role in pregnancy loss, and may be particularly important in women with chromosomally normal RM. As this field of study progresses, the 7

influence of hormones and stress on immune cell populations will need to be better understood and appropriately controlled for in the study of RM. 1.2.3.1

Uterine natural killer cells Uterine NK (uNK) cells have been proposed to be important specifically in implantation

and early pregnancy. During the luteal phase, increasing numbers of uNK cells are observed, which then apoptose before the next follicular phase (King et al. 1991). While the cyclic pattern of uNK cell proliferation implies there is hormonal regulation, the controlling factors have not been determined. Upon fertilization, the number of uNK cells is increased further and maintained throughout the first 20 weeks of pregnancy; with the greatest enrichment at locations of placental invasion (King et al. 1998). This suggests that they may be important for appropriate regulation of placental invasion and decidualization of the endometrium. A large proportion of uNK cells also have distinct characteristics; they express different markers [cluster of differentiation (CD) 56bright, CD16-] than those in peripheral circulation (CD56dim, CD16bright), and have reduced cytotoxic potential with increased secretion of angiogenic factors (Hanna et al. 2006, Nishikawa et al. 1991). The proportion, distribution and number of uNK cells, as well as peripheral NK cells, have been investigated in women with RM. To summarize, RM has been associated with an elevated proportion of peripheral NK cells in blood (Emmer et al. 2000, King et al. 2010, Kwak et al. 1995), an increase in the proportion of uNK cells in the non-pregnant endometrium (Tuckerman et al. 2007) and lower levels of uNK cells in the decidua (Yamamoto et al. 1999a), when compared to healthy controls. Another approach has been to compare the decidua from RM women with chromosomally abnormal miscarriages to those with chromosomally normal miscarriages (Quack et al. 2001, Yamamoto et al. 1999b). The first study found a decrease in 8

uNK cells in the decidua of karyotypically normal miscarriages (Yamamoto et al. 1999b), however the latter found no difference in uNK cells, but an overall increase in activated leukocytes (Quack et al. 2001). Together these studies suggest that changes in the immune cell composition of the maternal endometrium may be important in the predisposition to chromosomally normal RM, but the exact nature of these changes is unclear. There are many challenges and considerations in the design and interpretation of studies investigating NK cells and reproductive pathologies. It is known that the prevalence of uNK cells is strongly associated with levels of progesterone, which vary throughout the menstrual cycle and at the cessation of pregnancy (King et al. 1998). Furthermore, a recent study has found that reproducibility even within the same women from cycle to cycle is poor (Mariee et al. 2012). One group has hypothesized that elevated peripheral NK cells observed among RM women may in fact be due to an acute stress response at the time of blood draw, as levels returned to those consistent with controls upon a second blood draw within 20 minutes; this change was not observed in controls (Shakhar et al. 2006). 1.2.3.2

Maternal T helper 1 and T helper 2 immune balance During pregnancy, there is an essential shift in maternal Th cell balance from cell-

mediated to humoral immunity (Wegmann et al. 1993). The two main players in this balance are Th1 and Th2 cells. Th1 cells drive the cell-mediated response by producing cytokines, including interleukin 2 (IL-2), interferon gamma and transforming growth factor beta, that improve the killing efficacy of macrophages and cytotoxic T cells; while cytokines produced by Th2 cells (IL4, 5, 6, and 10) positively regulate B cells to produce neutralizing antibodies (Laird et al. 2003). A landmark study found that peripheral blood mononuclear cells in 60% of RM women were embryotoxic in vitro, through a cell-mediated Th1 response, and this was not observed in 9

control women (Hill et al. 1995). Since, many studies have validated this finding, showing a predominant Th1 response in peripheral blood (Kheshtchin et al. 2010, Kwak-Kim et al. 2003, Ng et al. 2002) and decidualized endometrium (Michimata et al. 2003) of RM women compared to controls. However, concerns have been raised that these studies were confounded by timing of sample collection (Laird et al. 2006), primarily due to the influence of progesterone on these immune cell populations (Check 2002). There was no evidence of an abnormal increase in cellmediated immunity in the endometrium of non-pregnant women with RM (Shimada et al. 2004). However, studies in mice support an endometrial shift in Th2 to Th1 immunity resulting in increased susceptibility to spontaneous abortion (Clark and Croitoru 2001) and this effect may be mediated by increased trophoblast apoptosis (Lee et al. 2005c). 1.2.3.3

Placental immunity The placenta has specific adaptations to protect itself from the maternal immune

response, primarily involving changes in cell recognition. The major histocompatibility complexes (MHCs) are antigens on the cell surface that present peptides, originating from endogenous or exogenous proteins, to immune cells. In the placenta, there is altered expression of the MHC I genes, also known as human leukocyte antigens (HLA) (Kovats et al. 1990). In normal somatic tissues, the HLA types expressed are the highly variable A, B and C; while the placenta predominantly expresses HLA-E and the non-variable HLA-G (Wei and Orr 1990). Furthermore, the cells that express HLA-G in the extraembryonic tissues are those that come into contact with maternal cells (McMaster et al. 1995), including the invasive extra villous cytotrophoblast, while HLA-E is expressed in all placental cell types, but confined to the cytoplasm (Bhalla et al. 2006). Decreased expression of HLA-G was observed in the cytotrophoblast of RM cases when compared to terminations (Emmer et al. 2002). However, 10

this finding was not replicated in independent studies of cytotrophoblast cells (Bhalla et al. 2006) or the decidua/villi interface (Patel et al. 2003) from women with RM. Polymorphisms in the HLA-G gene have been associated with RM in several studies and are discussed in more detail in section 1.3.1.2 (page 18). 1.2.3.4

Autoimmunity Autoimmunity has been implicated as a risk factor for pregnancy loss, and particularly

RM. While the immunosuppression of pregnancy has been associated with remission of some autoimmune conditions, such as rheumatoid arthritis, others, like systemic lupus erythematosus (SLE), can flare or increase in severity (Buyon 1998). The strongest association with RM has been antiphospholipid syndrome (APS), defined as the presence of autoantibodies to cell membrane phospholipids (APA), present in 14-20% of RM women (Clifford et al. 1994, Stephenson 1996). The rates of APS vary among RM populations, possibly due to erroneous false positives associated with recent infection (Ben-Chetrit et al. 2013); hence, two independent positive tests are recommended for diagnosis (Branch et al. 2010). The typical clinical presentation of APS is an increased incidence of blood clots (thrombosis) with adverse pregnancy outcomes, including miscarriage. It can occur independently or as a systemic autoimmune response, such as in SLE. The incidences of SLE, in addition to other autoimmune disorders, are all elevated among RM patients compared to ethnically and age-matched controls (Christiansen et al. 2008). These autoimmune conditions may cause RM through either thrombotic events in the placental vasculature (De Wolf et al. 1982), or poor placental invasion due to antibodies inhibiting trophoblast function (Yacobi et al. 2002).

11

1.2.3.5

Infection Ascending infection may disrupt the feto-maternal interface, by inducing a stronger cell-

mediated maternal immune response, resulting in poor implantation. Viral or bacterial infections can cause isolated miscarriage, but there are few chronic infections that are candidates for RM (Nigro et al. 2011). One such candidate may be bacterial vaginosis, an overgrowth of anaerobic bacteria within the vagina, which has been associated with late RM (Llahi-Camp et al. 1996); although the benefits of treatment on reproductive outcomes have not been shown (Guise et al. 2001). 1.2.4

Endocrinological The hormonal balance in women is maintained by the HPO axis, which regulates

maturation and ovulation of the oocyte, implantation and early pregnancy. In this section, I will discuss several endocrinological conditions, including luteal phase defects (LPD), thyroid dysfunction, and polycystic ovarian syndrome (PCOS) that can increase risk for RM by disrupting this balance of hormones in early pregnancy. 1.2.4.1

Luteal phase defects LPD are characterized by a lack of physiological changes associated with luteal phase

progesterone, including reduced secretion from the corpus luteum or poor responsiveness of the endometrium (Smith and Schust 2011). LPD can be caused by stress, exercise, weight loss and hyperprolactinemia (Arredondo and Noble 2006). In vitro, the over-expression of prolactin has been shown to inhibit progesterone secretion from granulosa cells (McNatty and Sawers 1975). Among women with RM, LPD have been reported in 17-27% (Li et al. 2000, Stephenson 1996), although the diagnosis of LPD is still controversial. Currently progesterone treatment is not recommended for women with RM; although a meta-analysis of several trials showed a marginal 12

reduction in miscarriage rates (Haas and Ramsey 2008). A concern is that the studies were small and inadequately controlled, suggesting a need for a large-scale, randomized, placebo-controlled trial, assessing live birth rate as the primary outcome (Coomarasamy et al. 2011). 1.2.4.2

Thyroid dysfunction Irregular production of thyroid hormones (TH), which in some cases can be caused by

thyroid autoimmunity, is associated with RM (Smith and Schust 2011). Although the exact mechanism of action in early pregnancy is unknown, it has been hypothesized that excess TH can cross the placental barrier and have a direct toxic effect on fetal growth and development (Anselmo et al. 2004). Contrastingly, reduced levels of TH, due to autoimmunity or underproduction, may impair folliculogenesis by altering granulosa cell function (Wakim et al. 1993). It has been suggested that the association of altered thyroid function with RM may be merely due to increased incidence of thyroid dysfunction in older women (Kaprara and Krassas 2008). 1.2.4.3

Polycystic ovarian syndrome PCOS is a complex condition that is associated with irregular endocrine profiles,

disrupted menstrual cycle, altered metabolic function, and/or obesity. Approximately 60% of women with PCOS have at least one first trimester miscarriage (Glueck et al. 2002), although the cause of this association is unclear. While early reports suggested there was an extremely high prevalence of polycystic ovaries among women with RM (Clifford et al. 1994); using the consensus Rotterdam criteria, the incidence only appears to be around 10% (Cocksedge et al. 2009), which is similar to that reported in the general population (Broekmans et al. 2006) . Women with PCOS have an endocrine profile that is characterized primarily by elevated androgens and high levels of LH. Elevated LH has also been previously identified in women 13

with RM (Regan et al. 1990). While not extensively studied, elevated androgens and/or LH do not appear to negatively affect folliculogenesis or oocyte quality (Gleicher et al. 2011, Gonen et al. 1990). Alternatively, women with PCOS, as well as those with RM, often have insulin resistance (Celik et al. 2011, Craig et al. 2002) and increased incidence of obesity (Boots and Stephenson 2011). Both obesity and metabolic changes have been associated with poor oocyte quality (Purcell and Moley 2011). Furthermore, women with PCOS have a higher risk for thyroid autoimmunity (Janssen et al. 2004) and thrombophilic disorders (Moini et al. 2012). Taken together, the many features associated with PCOS may increase risk for RM independently or in combination. 1.2.4.4

Endometrial receptivity An interesting hypothesis has recently emerged from one group, suggesting that women

with RM may represent a distinct “superfertile” subset of the population and that the cause for recurring miscarriage is impairment in natural embryo selection by the endometrium (Teklenburg et al. 2010a). In other words, embryos that would otherwise fail to implant, such as those with aneuploidy or other chromosomal abnormalities, are not recognized effectively in women with RM, resulting in implantation and subsequent miscarriage. Supporting this hypothesis, women with RM were found to have a short time-to-pregnancy interval, with 40% achieving pregnancy in less than 3 months (Salker et al. 2010). The decidualized endometrium secretes specific factors during the ‘window of implantation’ and these signals are altered in the presence of an arresting blastocyst (Teklenburg et al. 2010b). This observation led to the postulation that the decidualized endometrium acts as a biosensor for abnormally developing embryos, and that it may be perturbed in women with RM (Teklenburg et al. 2010a). This group went on to show that the endometrium from women with RM showed altered expression of 14

genes associated with the ‘window of implantation’ and that this could be corrected through in vitro decidualization of the cells (Salker et al. 2010). This mechanism could also explain some of the immunological and endocrinological associations with RM. 1.2.5

Thrombophilic Thrombophilia is a multifactorial condition that is characterized by an increased risk for

the formation of blood clots. The association and treatment of acquired or inherited thrombophilias among women with RM is controversial (Greer 2011, Krabbendam et al. 2005, McNamee et al. 2012). Inherited thrombophilias refer to mutations and/or polymorphisms in genes involved in or modulating the activity of the coagulation pathway, while acquired thrombophilias generally describe APS or acquired activated protein C (an anti-coagulant) resistance (McNamee et al. 2012). Thrombosis may cause late pregnancy loss through disruption of placental vascularization and blood flow to the developing fetus (Vora et al. 2009, Weiner et al. 2003). Consistently, women with thrombophilias have been found to be at increased risk for stillbirths (Preston et al. 1996). A clear connection between inherited thrombophilias and risk for RM has been elusive (Kovalevsky et al. 2004, Krabbendam et al. 2005, Lund et al. 2010). The strongest candidate associations are summarized in section 1.3.1.4 (page 19). Given the rarity of some of these alleles and the inconsistent associations, testing for these variants is currently not recommended as a clinical assessment in the evaluation of RM (Practice Committee of the American Society for Reproductive Medicine 2012). The current recommended therapy for women with thrombophilia, in the form of APS, and RM is a combination of aspirin and heparin during pregnancy (Practice Committee of the American Society for Reproductive Medicine 2012). However, it has been argued that there is a 15

need for practice of evidence-based medicine in this area, as numerous trials have failed to show the efficacy of treatment (Mantha et al. 2010, Tan et al. 2012). In fact, it has even been suggested that treatment of inherited thrombophilia may cause maternal harm, due to rare but serious bleeding as a side-effect of anticoagulants, discomfort of daily injections, erroneous treatment of patients with false positive tests, and psychosocial stress (Bradley et al. 2012). 1.2.6

Psychosocial stress Psychological stress has been implicated in both pregnancy loss and RM risk. Women

with increased cortisol, a physiological marker of stress, during the first few weeks of pregnancy were greater than two times more likely to miscarry than those women with levels in the normal range (Nepomnaschy et al. 2006). Three independent studies found that supportive care improved successful pregnancy rates among women with RM from 30-50% to over 80% (Clifford et al. 1997, Liddell et al. 1991, Stray-Pedersen and Stray-Pedersen 1984). Women with RM reported higher levels of psychological stress compared to fertile controls, although it was not predictive of positive pregnancy outcomes in these women (Li et al. 2012). One mechanism that has been proposed to link elevated stress to miscarriage is altered immune function. Reduced cytotoxicity of peripheral blood NK cells was observed among RM women with higher depressive symptoms (Andalib et al. 2006), although this may not reflect uNK cell changes. Mice with elevated stress (ultrasonic sound exposure) during pregnancy have higher embryo resorption rates, which was associated with an increase in cell-mediated immune response in the endometrium (Joachim et al. 2001). In addition, psychological stress has been associated with markers of biological aging, such as telomere length and reactive oxidative species (ROS), which will be discussed further in Chapter 3.

16

1.3

Genetic and epigenetic factors contributing to recurrent miscarriage risk RM is likely a multifactorial complex trait, as familial studies have shown that sisters of

patients with RM are 6 times more likely to have RM than control women (Christiansen et al. 1990). Genetic and environmental factors may contribute to the etiology of RM in an additive or synergistic epistatic manner, affecting maternal risk by negatively impacting the progression of oogenesis, implantation or early fetal development. Extensive studies of maternal genetic variants, in pathways already implicated in the etiology of RM including meiosis, immunological, endocrinological and thrombophilic, have been undertaken to identify reliable biomarkers of risk and further elucidate the pathogenesis of RM (Christiansen et al. 2008). While there has been considerable progress in this area, there are many inconsistent associations and are likely attributable to differences in ethnicities and underpowered studies. An additional area of study is aspects of chromosome biology, including telomere length, skewed X chromosome inactivation (XCI) and epigenetic modifications. Aberrant establishment or maintenance of these important processes in the oocyte or embryo may result in miscarriage due to increased risk for non-disjunction or limited cellular capacity for differentiation. In this section, I will summarize the evidence that genetic and epigenetic factors contribute to risk for RM. 1.3.1 1.3.1.1

Maternal genetic variants Genes involved in meiosis Genetic variants in genes involved in meiosis may predispose some women to high rates

of aneuploidy, due to increased rates of non-disjunction. The subset of women with RM and recurrent heterotrisomies at a young age may be strong candidates for this genetic predisposition. To date there has only been one gene investigated for mutations in association with RM, 17

synaptonemal complex protein 3 (SYCP3) (Bolor et al. 2009). SYCP3 is one of several proteins that are essential for tethering of homologous chromosomes together during MI prophase (Yuan et al. 2000). Further assessment of the ~400 genes encoding meiosis-specific proteins (Feichtinger et al. 2012) may be an area of future study in this subset of RM cases. 1.3.1.2

Genes involved in immune function In addition to the assessment of gene expression patterns of HLA-G in RM, genetic

polymorphisms within the gene have been examined in numerous studies. Reproducible associations, among larger studies, have been identified for HLA-G haplotypes *010103 (synonymous) and *0105N (frame shift), comprised of seven coding SNPs within exons 2 and 3, and a 14bp insertion/deletion in the 3’untranslated region (UTR) of the HLA-G gene (Aldrich et al. 2001, Pfeiffer et al. 2001, Vargas et al. 2011, Zhu et al. 2010). It has been proposed that HLA-G polymorphisms (14bp in/del, specifically) may also predispose women to secondary RM, if their first liveborn infant was male, by contributing to an altered maternal immune response to the HY (male-specific histocompatibility antigen) (Christiansen et al. 2012). The pathogenesis of this association needs to be elucidated further, but it is an interesting explanation for the epidemiological finding that secondary RM is more frequent after male births (Ooi et al. 2011). 1.3.1.3

Genes involved in endocrine function While there may be an underlying genetic susceptibility to hormonal imbalance in

women with RM, genetic variation in genes involved in the HPO axis has not been widely studied. A moderate association with a SNP (rs10046) in the aromatase (CYP19A1) gene was identified in a study of the estrogen synthesis pathway (Cupisti et al. 2009, Litridis et al. 2011). In addition, several polymorphisms in the promoter and intron 2, as well as missense mutations with functional consequences, of the chorionic gonadotropin (CGB5/8) genes have been 18

associated with RM (Nagirnaja et al. 2012, Rull et al. 2008). A recent meta-analysis of polymorphisms in the estrogen receptor α (ESR1) and progesterone receptor (PR) genes found no association (Su et al. 2011). While these data suggest genetic variation in receptors and regulatory genes may influence risk for RM, a more comprehensive study is needed. 1.3.1.4

Thrombophilia associated genes There has been extensive study of inherited thrombophilias, with the primary focus being

on the functional polymorphisms in genes involved in three pathways that influence blood clot formation: coagulation, fibrinolysis, and the folate cycle (Krabbendam et al. 2005). Two variants, Factor V Leiden (F5) G1691A and prothrombin (F2) G20210A, have been associated with increased risk for late miscarriage (≥10 weeks gestation) in a cohort of more than 32,000 women (Lissalde-Lavigne et al. 2005). A systematic review and meta-analysis also found that these variants were associated with RM (Bradley et al. 2012, Kovalevsky et al. 2004). In addition, two studies with cohorts of >500 women with RM, both found associations with plasminogen activator inhibitor 1 (PAI-1 or SERPINE1) 4G and methylenetetrahydrofolate reductase (MTHFR) C677T variants (Goodman et al. 2006, Ozdemir et al. 2012). While these data suggest that genetic variation in pathways involved in thrombosis susceptibility contributes to risk for RM, environmental factors may also play a role. The folate cycle involves the conversion of dietary folic acid into intermediate molecules contributing to the methylation and nucleotide synthesis pathways. Numerous variants in this pathway have been studied for their association with disease, particularly in fetal health and survival. The most influential and widely studied is the MTHFR C677T polymorphism, a nonsynonymous change that causes a dramatic reduction in enzyme activity (Frosst et al. 1995). In addition to causing elevated homocysteine, genetic variation within folate cycle enzymes, 19

including MTHFR, can cause decreased production of methyl donors and nucleotide precursors (DeVos et al. 2008). As mentioned above, two large studies identified an association between MTHFR C677T and RM (Goodman et al. 2006, Ozdemir et al. 2012); however, a meta-analysis only found an association with RM in Chinese populations, suggesting its effect may be modified by genetic background and environmental factors, such as diet (Ren and Wang 2006). There has been no other consistent evidence of an association between other genetic variants within the folate cycle and RM. MTHFR C677T may contribute to susceptibility of RM through several potential mechanisms: 1) altered establishment and/or maintenance of DNA and/or histone methylation in the developing oocyte or embryo, 2) aberrant DNA synthesis/repair in the developing oocyte or embryo, 3) placental thrombosis due to elevated homocysteine levels, or 4) impaired ovarian function affecting oocyte maturation. While low folate diet and/or MTHFR C677T polymorphism are known to reduce levels of methyl donors, there have been very few studies in humans demonstrating that maternal diet or polymorphisms alter fetal DNA methylation (Hogg et al. 2012, Park et al. 2008). In addition, folic acid deficiency in cell culture causes uracil to be mis-incorporated into DNA, leading to point mutations and genomic instability (Duthie and Hawdon 1998), suggesting folate levels are important for genomic integrity. Low folate diet and/or MTHFR C677T also cause elevated homocysteine (Guttormsen et al. 1996), a nonessential amino acid that is associated with the risk for thrombosis (den Heijer et al. 1996). Independent of the effects seen in the folate pathway, two separate groups have found that MTHFR genotype can influence ovarian activity, specifically decreasing estrogen synthesis from granulosa cells and increasing serum FSH levels (Hecht et al. 2009, Rosen et al. 2007).

20

1.3.2

Telomeres Telomeres are the TTAGGG repeats that associate with complexes of proteins to protect

the ends of chromosomes in humans. Telomere length is a marker of biological aging, as it declines proportionally with number of cell divisions due to the end replication problem (Harley et al. 1990). Additionally, exposure to oxidative stress within the cell increases the rate of loss (Serra et al. 2000) and may account for telomere length decline in cells that do not replicate, such as oocytes. Cells that express telomerase are able to elongate telomeres through reverse transcription; however the majority of somatic cells lack this enzyme (Kim et al. 1994). Telomeres are not only important in protecting the chromosomes from degradation, but also for positioning in meiosis, by allowing the chromosome pairs to tether together and align appropriately within the cell for recombination (Cooper et al. 1998). In mouse, irregular shortening of telomeres is associated with abnormal recombination and synapses in meioses, particularly in females, mimicking the age-related effects (Liu et al. 2004). This led to the hypothesis that the age-related increase in aneuploidy rates in women may be partly attributable to a decline in telomere length (Keefe et al. 2006). This group also showed that exposing mice to a compound that reduces the effects of oxidative stress, increased telomere length in the ovaries and improved egg quality (Liu et al. 2012). In humans, telomere length in sister oocytes from women undergoing in vitro fertilization (IVF), was also a strong predictor of pregnancy outcome (Keefe et al. 2007), suggesting that telomere length may be important for human reproductive health as well. 1.3.3

Epigenetics in fetal development Epigenetics is defined as mitotically heritable chemical changes that influence gene

expression, without affecting DNA sequence. These changes can include DNA methylation, 21

histone modifications, histone variants and non-coding RNAs. The most extensively studied is DNA methylation, which primarily involves the addition of a methyl group to a cytosine in a CpG dinucleotide. Generally, it is thought that DNA methylation at promoter regions can limit the accessibility of this DNA to transcription machinery, directly and through crosstalk with epigenetic modifiers, and thus reduce expression of the associated gene (Klose and Bird 2006). Epigenetic patterns are essential in development for tissue differentiation and response to environmental cues (Monk 1995). Aberrant establishment or maintenance of epigenetic marks in the developing embryo may be a mechanism for pregnancy loss (Messerschmidt et al. 2012, Pliushch et al. 2010, Yin et al. 2012). Developmentally important imprinted genes, those that are mono-allelically expressed in response to parent-of-origin differentially methylated regions have specifically been examined for an association with miscarriage. The first study to look at DNA methylation in miscarriage samples reported an increase in outliers at several imprinted loci (Pliushch et al. 2010). Since, there have been two studies investigating DNA methylation at specific loci in RM, both with limited sample size and therefore must be interpreted with caution. Aberrant gain of allelic DNA methylation at the CGB5 gene, a non-imprinted gene, in placental trophoblast and loss of H19 methylation in sperm was observed from couples with RM (Ankolkar et al. 2012, Uuskula et al. 2011). Interestingly, a mouse model deficient for an epigenetic modifier gene (Trim28) in oocytes showed preferential loss of DNA methylation at imprinted loci and complete embryonic lethality (Messerschmidt et al. 2012), suggesting that loss of differentially methylation regions in oogenesis may be a mechanism of RM. A comprehensive analysis of genome-wide and sitespecific changes of DNA methylation in miscarriage and RM is needed to evaluate the frequency and nature of epigenetic errors in early pregnancy. 22

1.3.4

Maternal skewed X-chromosome inactivation X-chromosome inactivation (XCI) is the epigenetic silencing of one of the X

chromosomes to allow dosage compensation in females. In humans, there is random XCI in all tissues during development, with each X being inactivated in approximately an equal number of progenitor cells, resulting in a 50:50 distribution (Gartler 1976). Skewed XCI can occur when there is selection against cells that have inactivated a particular X chromosome, for example if there was a chromosomal aberration or mutation, or due to stochastic events in a small number of progenitor cells (Willard 1996). Skewed XCI, measured in peripheral blood, has been previously associated with aging (Hatakeyama et al. 2004) and various diseases, in particular RM (Beever et al. 2003). While there have been conflicting reports in the literature, a recent meta-analysis found a two-fold increased risk for RM among women with skewed XCI (Su et al. 2011). It has been suggested that cryptic rearrangements or mutations on the X chromosome or a restricted stem cell population early in development may explain this association (Robinson et al. 2001), but this remains to be demonstrated. 1.4

Research objectives The purpose of my thesis is to investigate factors that may contribute to the pathogenesis

of RM. I hypothesize that genetic and/or epigenetic factors associated with meiotic nondisjunction, maternal endocrinological profile, reproductive aging in females and/or placental functioning will contribute to the etiology of RM. Therefore the objectives of this study are: 1) To determine whether mutations in the synaptonemal complex protein 3 (SYCP3) gene are associated with increased risk for aneuploidy among women with RM. 2) To compare telomere lengths in peripheral blood between women with RM and healthy controls, as an indicator of reproductive aging. 23

3) To evaluate the frequencies of functional polymorphisms within genes involved in the HPO axis among women with RM and those that are reproductively healthy. 4) To investigate patterns of DNA methylation in placental villi from first trimester, karyotypically normal products of conception from women with RM, a single miscarriage, or an elective termination. This study will improve our understanding of mechanisms involved in RM and will identify markers of potential prognostic value for clinical evaluation.

24

Figure 1.1 Gametes as the product of meiosis I non-disjunction and meiosis II non-disjunction. Diagram shows one pair of homologous chromosomes progressing through the meiotic divisions, missegregating in meiosis I (left) and meiosis II (right). Fertilized gametes would either result in trisomy or monosomy.

25

Figure 1.2 Number of germ cells with age in females. The number of germ cells increases dramatically through mitotic divisions in the female fetus until the peak at ~6 months gestation. These then undergo apoptosis, depleting to ~2 million by birth. At the onset of puberty, oocytes are cyclically recruited until their eventual depletion at menopause, occurring at an average age of 50 years old (based on Baker 1971, with permission). 26

Figure 1.3 Hormone levels and follicular events during the menstrual cycle. The days of the menstrual cycle are divided into three portions: 1) follicular phase, 2) ovulation, and 3) luteal phase. The corresponding relative hormonal levels and follicular events are shown in the panels below (based on Mader 2006, with permission).

27

Figure 1.4 Etiology of recurrent miscarriage. Proportion of recurrent miscarriage patients with parental chromosomal rearrangements (balanced translocations), immunological (anti-phospholipid syndrome), uterine anatomical malformations, endocrine (luteal phase defects and thyroid dysfunction) and unknown etiology (based on averages, wherever possible, from Clifford et al. 1994, Stephenson 1996).

28

Chapter 2: Mutational analysis of the SYCP3 gene 2.1

Introduction Women with RM due to aneuploidy likely have a distinct etiology, as these women may

represent a subset of the population that is at increased risk for meiotic non-disjunction. Genetic variation in genes involved in chromosome pairing, recombination and segregation in meiosis may contribute to this increased risk. SYCP3 is involved in forming the synaptonemal complex in MI, which is a structure that allows for the pairing and recombination of homologous chromosomes. SYCP3 is an essential component of the axial and lateral elements of this complex that holds the chromosomes together (Page and Hawley 2004). In mice deficient in Sycp3, chromosomes fail to synapse (Yuan et al. 2000). Male mice deficient in Sycp3 are infertile due to arrest of meiosis, while female mice are fertile but have decreased litter sizes attributable to increased rates of trisomic fetuses due to abnormal pairing of the chromosomes in meiosis in the germ cells (Yuan et al. 2002). Bolor and coauthors (2009) first suggested a role of mutations in the SYCP3 gene in RM, identifying two of 26 Japanese women with RM with heterozygous mutations within and nearby exon 8. The present study sought to identify novel and previously observed mutations in SYCP3 in women with RM who had at least one documented trisomic miscarriage, a subset most likely to carry mutations in SYCP3. 2.2

Materials and methods A total of 50 women with RM and at least one trisomic conception, were ascertained in

the Recurrent Pregnancy Loss Program at BC Women’s Hospital & Health Centre, Vancouver, British Columbia. The age at time of pregnancy [mean ± SD (range)] was 36.2 ± 5.2 years (2244) with a total of 292 pregnancies, of which 216 (74%) ended in miscarriage. Table 2.1 29

summarizes the pregnancy outcomes within this study population, including the distribution of miscarriage karyotypes. The number of miscarriages was 4.3 ± 1.5 (3-9); with 34 women having a single trisomic miscarriage and 16 women having multiple heterotrisomic miscarriages. Carriers of structural chromosome rearrangements were excluded from this study. DNA was extracted from whole peripheral blood using conventional methods. All coding exons (2-9) of SYCP3, including the intron/exon boundaries, were PCR amplified by conventional PCR, using primer sequences shown in Supplementary Table 2.1. Sequencing was done utilizing a 3130xl genetic analyzer (Applied Biosystems, Melbourne, Australia), with BigDye Terminator sequencing kit version 3.1. Sequence data were analyzed using Chromas 2.33 (Technylisium, Australia) and SeqDoC (Crowe 2005). 2.3

Results In this study, all coding exons (2-9) of the SYCP3 gene, located on chromosome 12q23.2,

were sequenced. To assess intron/exon boundaries, peripheral sequence surrounding each exon was included in the corresponding PCR amplicon [mean ± SD = 161.2 ± 82.1 base pairs (bp)]. No novel or previously reported mutations within the coding exons or intron/exon boundaries were identified in our study population. Four non-coding single nucleotide polymorphisms (SNPs) were present at variable frequencies (2-29%) among these 50 RM women (Figure 2.1). The frequencies of these SNPs were comparable to those reported in the world-wide population, obtained from the UCSC Genome Browser (Table 2.2). However, this finding should be interpreted with caution as the ethnicity of the population and RM groups is likely extremely divergent.

30

2.4

Discussion In the first study to assess SYCP3 in 26 Japanese women with RM, two mutations were

identified: a 4 bp deletion within the splice acceptor site of exon 8 resulting in C terminal truncation and a synonymous change at position 657T>C, which disrupted splicing of intron 8 (Figure 2.1) (Bolor et al. 2009). The C terminal region is of particular importance for SYCP3 function, as it comprises a coil-coiled domain that is highly conserved and has been shown to be necessary in rats for SYCP3 assembly in meiosis (Baier et al. 2007). An additional homozygous variant (666A>G) was identified in two Japanese female with unexplained infertility among 88 investigated (Figure 2.1) (Nishiyama et al. 2011). In contrast, a more recent study found no association between the SYCP3 657T>C variant and RM in 101 Japanese women, nor when assessing the subset of 47 women with at least one karyotypically abnormal miscarriage, although not strictly trisomic (Mizutani et al. 2011). The present study supports this latter finding in an ethnically divergent western Canadian population, which is predominantly Caucasian. Although the sample size is limited, the cohort was well-characterized and included women most likely to be at increased risk of meiotic non-disjunction. An additional strength of this analysis was the investigation of the entire coding region including the intron/exon boundaries in SYCP3 for variants within this population. These findings, therefore, suggest that mutations in SYCP3 are not a common factor contributing to risk for meiotic non-disjunction in human maternal gametogenesis. Given the complexity of meiosis and the many genes involved in this process, it seems unlikely that mutations in a single gene would account for a large number of RM patients. Multiple mutations and/or polymorphisms in a variety of genes may influence risk for non-disjunction and subsequent RM. Further characterization of RM patients 31

and the application of techniques such as whole exome sequencing, which would allow screening of many genes of interest simultaneously, would help clarify the underlying mechanisms involved.

32

Table 2.1 Outcomes of 292 pregnancies from 50 women with recurrent miscarriage. Miscarriages are further subcategorized by karyotype. Pregnancy Outcome

Number

Livebirth

56

Termination

15

Ectopic

5

Miscarriage

216

46, XX or 46, XY

17

Trisomy*

73

Triploidy

3

Other

2

Not karyotyped

121

* This includes 22 cases of trisomy 13-15, 14 cases of trisomy 16, 11 cases of trisomy 21-22, and 26 cases of other trisomies.

33

Table 2.2 Minor allele frequencies of noncoding single nucleotide polymorphisms within the SYCP3 gene identified in 50 recurrent miscarriage women. For each variant, the dbSNP identifier, genic location and heterozygosity are given, if available. Population frequencies, from the UCSC Genome Browser, were not significantly different from those in our RM study group, using Yates chi-square comparison. Population minor

RM women minor

allele frequencies

allele frequencies

Variant

Genic Location Heterozygosity

rs3751248

intron 2

0.205 +/- 0.246

8.00%

5.10%

rs10860779

intron 5

0.444 +/- 0.157

33.29%

29.00%

not reported

8.29%

5.00%

not reported

2.28%

2.00%

rs145003954 intron 6 rs17723833

exon 9 (3’UTR)

34

Figure 2.1 Schematic diagram of the SYCP3 gene and variants. Exons of the SYCP3 gene are denoted by grey boxes and numbered, with wider coding exons than those comprising of the untranslated regions. Introns are marked by the dashed line. The transcription start site is marked with an arrow, showing direction of gene transcription. Mutations previously associated with RM are labeled in black, while SNPs found among this RM study population are labeled in red.

35

Chapter 3: Telomere length and reproductive aging 3.1

Introduction Female fertility declines with age due to the combined effects of both a decrease in the

rate of conception and an increase in the rate of pregnancy loss due to aneuploidy. Age-related changes in the human ovary, including depletion of ovarian follicles (Faddy et al. 1992, Faddy 2000) and a decline in oocyte genomic stability leading to aneuploidy (Hassold and Hunt 2001) may contribute to this phenomenon. The rate of female reproductive aging displays a large amount of inter-individual variability. This is reflected in the variability in age of reproductive senescence (menopause), which typically occurs anytime between 40 to 60 years of age (Kato et al. 1998, te Velde and Pearson 2002), as well as in the individual variability in risk of conceiving a trisomic pregnancy (Nicolaides et al. 2005, Warburton et al. 2004). This natural variation in reproductive aging may be the result of environmental and genetic factors that affect individual rates of cellular aging. Both animal models and human epidemiological studies support the suggestion that longevity is associated with an increase in reproductive lifespan. Mice and flies selectively bred for reproductive longevity have an overall increase in total lifespan when compared to unselected controls (Hutchinson and Rose 1991, Nagai et al. 1995). Human population studies have reported that higher total fecundity (Manor et al. 2000, Muller et al. 2002), later age at last reproduction (Doblhammer 2000, Helle et al. 2005, Muller et al. 2002, Smith et al. 2002) and older age at menopause (Cooper and Sandler 1998, Jacobsen et al. 1999, Snowdon et al. 1989) are positively correlated with longevity. A study of female centenarians found that women living to at least 100 are greater than four times more likely to have had a child while in their forties than women living to age 73 (Perls et al. 1997). There are several possible explanations 36

for the relationship between longevity and age at menopause: 1) prolonged estrogen exposure associated with later menopause may have a positive influence on life expectancy (Perls et al. 1997), 2) effective age of the ovary could directly affect longevity (Cargill et al. 2003, Hsin and Kenyon 1999), or 3) selective pressures to maximize a woman’s reproductive years by slow reproductive aging may have positively selected for women with slower rates of cellular aging (Perls et al. 2002, Perls and Fretts 2001). Telomere length exhibits considerable inter-individual variation (Hastie et al. 1990) and may contribute to the observed variability in reproductive aging. Telomere variability may be due to differences in telomere length at conception, telomerase activity during early development, rate of cell division and rate of telomere loss per cell division. Shorter telomeres may limit the mitotic capacity of primordial germ cells during fetal development and therefore restrict the size of the follicular pool (Keefe et al. 2006). Studies examining telomere length and reproductive aging in humans have produced contradictory results in which telomere length has been both positively and negatively associated with different measures of reproductive aging (Aydos et al. 2005, Dorland et al. 1998a, Keefe et al. 2007). Given the links between reproductive aging and biological aging, and the potential influence of telomere length on oocyte quality, I hypothesized that women who display evidence of premature reproductive aging will have a shorter average telomere length than control women. The objective of this study was to assess telomere length in peripheral blood leukocytes in two groups of women with evidence of premature reproductive aging: 1) patients with idiopathic premature ovarian failure (POF) who experienced menopause before 40 years of age, and 2) women with a history of RM. While menopause represents a finite end of the reproductive lifespan, it is preceded by a period of subfertility, in which women have increased susceptibility 37

to miscarriage (Broekmans et al. 2009). These study groups were compared to two control groups: 1) women from the general population not selected on the basis of reproductive history and 2) women who had had a healthy pregnancy after 37 years of age and had not experienced any pregnancy loss. This latter group may represent women with potentially slower rates of reproductive aging, as they have not experienced difficulties conceiving or maintaining pregnancy despite a relatively advanced reproductive age. 3.2 3.2.1

Materials and methods Samples Women with RM (N=95), which includes 47/50 RM cases from Chapter 1, were

ascertained through the Recurrent Pregnancy Loss Clinic at Women’s Health Centre of British Columbia. These 95 women had a total of 458 miscarriages, and of those, 167 were karyotyped. Karyotyped miscarriages consisted of 72 diploid losses, 71 aneuploid losses and 24 other anomalies, including polyploidy, sex chromosome aneuploidies, and translocations. Of those women with aneuploid losses, there were 32 women who had a single trisomic miscarriage (ST), and 17 women who had multiple trisomic miscarriages (MT). POF patients (N=34) with idiopathic secondary amenorrhea were ascertained from the POF Clinic at the Women’s Health Centre of British Columbia. POF diagnosis was made based on the absence of menses for at least 3 months and two serum FSH results of >40 mIU/mL obtained more than one month apart, prior to 40 years of age. Two control groups were used in this study: Control group 1 (N=108) consisted of healthy women of reproductive age, ranging from 17 to 55 years, and Control group 2 (N=41) consisted of women who have had a healthy pregnancy over 37 years of age with no history of infertility or miscarriage. Control group 1 was comprised of anonymous healthy women from 38

the Vancouver area. Similarly, Control group 2 had healthy women from the Vancouver area, ascertained specifically at the British Columbia Women’s Hospital on the basis of a healthy pregnancy after the age of 37. DNA was obtained by standard salt extraction from ~5mL of blood collected in EDTA tubes. 3.2.2

Telomere length Average relative telomere length was determined by quantitative PCR (qPCR) (Cawthon

2002). Amplification of the telomeric repeat region was expressed relative to amplification of 36B4, a single copy housekeeping gene on chromosome 12. This telomere to single copy (T/S) ratio is proportional to the average telomere length of the sample, due to the amplification being proportional to the number of primer binding sites in the first cycle of the PCR reaction (Figure 3.1) (Cawthon 2002). The protocol was performed as previously described (Cawthon 2002) with several modifications; amplifications were carried out in 20uL reaction with approximately 5ng genomic DNA, 0.5uM ROX Reference Dye (Invitrogen, Carlsbad, USA), and 0.2x SYBR Green I nucleic acid gel stain in DMSO (Invitrogen, Carlsbad, USA). Samples were run in triplicate on 96-well plates containing a standard curve constructed with reference DNA serially diluted to concentrations from 10ng to 0.625ng. A no template control and both short- and long-telomere reference samples were run on each plate as quality controls. Dissociation melting curves were run after each sample to ensure amplification of a single species. Replicates of each plate were done to ensure reliable values were ascertained. The values between both runs were significantly correlated, with a correlation coefficient of r=0.49 (p<0.0001). To improve the accuracy of our estimates we averaged the values of the two independent measurements. When values were discrepant between the two runs by more than 0.2 SDs, subsequent runs were done and an average of all values was used in further data analyses. 39

The telomere-qPCR assay was validated using DNA extracted from leukocyte cell pellets following flow fluorescence in situ hybridization (FISH) (N=12) (Baerlocher et al. 2006). There was a strong correlation between the qPCR T/S ratio and the flow-FISH telomere lengths (r=0.96) (Figure 3.2). The strong correlation obtained validates the use of an average measurement of t/s values as an accurate reflection of telomere length. T/S values were converted to kilobases (kb) using the linear equation from this correlation (y = 7.25x + 2.50). As expected the y-intercept is at 2.5 kb since the flow-FISH assay was calibrated using Southern blot telomere restriction fragment lengths, which includes ~ 2.5 kb of subtelomeric repeat (Baerlocher et al. 2006). 3.2.3

Statistical analysis Rate of telomere decline was determined by linear regression analysis, and one-tailed t-

tests were used to determine the significance of the regression because of the a priori hypothesis that telomere length was associated with age. Yearly rates of telomere decline were compared using two-tailed t-tests for comparison of regression slopes. Mean telomere length comparisons between sample groups were determined using pair-wise analysis of covariance (ANCOVA) tests to adjust for differences in ages between sample groups. 3.3

Results Telomere length in Control group 1 significantly declined with age (p=0.001, one-tailed t-

test) at a rate of 40 bp per year [95% confidence interval (CI) =14-66 base pairs], although there was significant variability in telomere length at any given age (R2=0.081, Table 3.1, Figure 3.3). There was also a weak (R2=0.161) but significant negative association between telomere length and age in POF patients (p=0.01 one-tailed t-test), but not in Control group 2 or the RM group as a whole. Subsets of the RM group who have experienced ST or MT are of particular interest, as 40

incidence of trisomic pregnancy increases with age, contributing to the age-related increase in RM. There was a weak (R2=0.130) but significant relationship between telomere length and age in the ST subset of the RM group (p=0.02, one-tailed t-test) but not the MT subset (Table 3.1, Figure 3.4). However, in no sample group was the rate of telomere decline significantly different than that of Control group 1 (two-tailed t-tests for comparison of regression slopes), thus ANCOVA was used to adjust for age effects on mean telomere length for further comparisons of mean telomere length between groups. Mean telomere length and age-adjusted mean telomere length for each sample group are shown in Table 3.2. Although women in Control group 2 had longer age-adjusted mean telomere lengths than those in Control group 1, this difference was not significant. The RM group had shorter age-adjusted mean telomere length than Control group 1 (8.46 vs. 8.92 kb, p=0.0004) and this was also apparent in comparison to Control group 2 (9.11 kb, p=0.02). However, short telomeres were not specifically confined to the subset of this group that had had either a single trisomy or multiple trisomic pregnancies. Contrary to expectation, age-adjusted mean telomere length in the POF patient group was longer than that in Control group 1 (9.58 vs. 8.92 kb, p=0.01), although this was not significant in comparison to Control group 2. 3.4

Discussion Telomere-specific qPCR was used to assess telomere length in groups of women with a

reproductive history suggestive of premature reproductive senescence to determine whether telomere length is associated with reproductive aging. As hypothesized, women experiencing RM had shorter age-adjusted mean telomere length than control women, although this effect was not specifically confined to women with trisomic pregnancies. In contrast, POF patients had a longer age-adjusted mean telomere length than that of controls. The high variability in telomere 41

length at any given age and the rate of telomere length decline with age has been previously reported in many control populations (Benetos et al. 2001, Hastie et al. 1990, Slagboom et al. 1994). In this study, the relationship between telomere length and age was not significantly different than zero in all sample groups, perhaps reflecting the limited age range in some groups. Regardless, none of the groups had a significantly different rate of telomere decline than that of controls. The observed shorter average telomere length in women with RM and the trend of longer telomere lengths in women in Control group 2, who have had viable pregnancies late in their reproductive life, are consistent with the hypothesis that telomere length is a determinant of the rate of reproductive aging in women. Previous studies have reported that telomere length is a strong predictor of developmental potential of sister oocytes from women undergoing IVF (Keefe et al. 2007) and is also correlated with reproductive life span in women (Aydos et al. 2005). Short telomere length in telomerase-deficient mice is associated not only with premature aging but also with reduced fecundity leading to sterility (Liu et al. 2002). These mice exhibit impaired oogenesis and mimic the human age-related decline in oocyte quality, with increased rates of apoptosis of the oocytes, impaired chromosome synapsis and recombination, and increased likelihood of non-disjunction and aneuploidy (Liu et al. 2004). Young mothers of children with Down syndrome have normal telomere lengths (Dorland et al. 1998b), suggesting that predisposition to non-disjunction may not be the only explanation for the finding of shortened telomeres in women with RM; although, the telomere length of peripheral blood cells may not necessarily reflect that of oocytes or embryos. Psychological stress indicated by physiological stress markers has also been shown to negatively influence telomere length (Epel et al. 2004) due to an increased rate of cell turnover and increased exposure to ROS. The shorter 42

telomere lengths in women with RM may therefore reflect higher levels of psychological and physiological stress and/or constitutionally shortened telomeres. The finding of increased telomere length in POF patients does not support the hypothesis that these women have accelerated cellular aging. As this is a relatively small sample size and the findings were not highly significant, it is possible that these results are due to a Type I error (false positive). Although care must be taken in conclusions drawn from these data, and the necessity for additional study in this area must be emphasized, these findings are nonetheless intriguing. One explanation for the increase in telomere length observed in the POF cohort may be a constitutional genetic tendency towards an overall slower rate of cell division, perhaps by predisposing towards a prolonged cell cycle. A slower cell division rate in the developing ovary could lead to the establishment of a reduced follicular pool during early embryonic development, whereas fewer cell divisions in hematopoietic stem cells could result in longer telomeres measured in peripheral blood. If longer telomeres in blood reflect fewer mitotic divisions in the initial germ cell pool, this could explain a smaller follicular pool and early menopause in POF patients (Dorland et al. 1998a). A second possibility is that longer telomeres in the POF patients are a result of autoimmunity in these women. Autoimmune destruction of the ovaries is a common cause of POF (Goswami and Conway 2005), and autoimmunity could conceivably alter blood cell composition (Josefowic et al. 2012) to a cell type with longer telomeres (Rufer et al. 1999). However, the limited existing evidence on telomere length and autoimmunity suggests that autoimmune conditions are associated with shorter rather than longer telomeres (Jeanclos et al. 1998), making this a less likely explanation. Alternatively, longer telomeres in the POF patient group may be the result of abnormal hormone exposure in these women. POF patients in our study may have been exposed to 43

elevated estrogen levels as a result of recruitment of large cohorts of oocytes in menstrual cycles occurring prior to POF onset. Premature follicular pool exhaustion resulting from the continual recruitment of large cohorts of follicles prior to menopause has been proposed as one mechanism for POF (Pal and Santoro 2002). Estradiol is secreted from developing follicles (Havelock et al. 2004) and stimulated estrogen level is correlated with the size of the antral follicle cohort (Scheffer et al. 2003). If abnormally large follicular cohorts were recruited while POF patients were still cycling, this could lead to elevated estrogen levels with a positive influence on telomere length. This positive influence on telomere length prior to the onset of POF may be reflected later in life in the form of long telomeres after POF diagnosis. On the other hand, maintenance of telomere length may be a recent phenomenon in these women, resulting from hormone replacement therapy following POF diagnosis. Although we lack the clinical details to assess this possibility in our POF population, this hypothesis is supported by the finding that long term hormone replacement therapy in postmenopausal women slows the rate of telomere attrition (Lee et al. 2005a). Two mechanisms by which estrogen may positively regulate telomere length have been proposed: 1) estrogen may ameliorate the negative effects of ROS (Aviv et al. 2005) which reduce telomere length by inducing single strand breaks (von Zglinicki 2000) and 2) estrogen may stimulate telomerase activity (Aviv et al. 2005). Ovarian telomerase activity is reportedly low in POF patients with follicular depletion, but high in POF patients with ovarian dysfunction and normal follicle counts (Kinugawa et al. 2000). Since follicle count is correlated with circulating estrogen level (Vital-Reyes et al. 2006) this supports the suggestion that telomerase activity is influenced by estrogen exposure, at least in the ovary. There are several limitations to this study that must be considered. Telomere length varies among blood cell types (Lansdorp 2006, Rufer et al. 1999); therefore variability between 44

individuals in telomere length measured in peripheral whole blood may be a consequence of differences in blood cell composition. Furthermore, telomere length measured in peripheral blood may not necessarily reflect telomere length in the ovary or developing embryo. However, there is a strong correlation between telomere lengths from tissues of a single individual (Butler et al. 1998) suggesting that telomere length measured in whole blood may be an accurate proxy for telomere length at the ovary. Small sample sizes and limited clinical details (including incomplete karyotype information on the losses of RM group and no details on reproductive history of Control group 1) restrict the ability to subdivide sample groups into more homogeneous phenotypes and negatively impact the power of these analyses. RM, trisomic pregnancy, and POF have all been considered measures of premature reproductive aging. However, the observation that RM and POF showed opposite associations with telomere length, and trisomic pregnancy showed no evidence of an association, suggests that these different types of reproductive aging are influenced by unique factors. Further studies are necessary to confirm these findings in larger more precisely defined populations, examine the physiological mechanisms that influence both telomere length and reproductive aging, and investigate the molecular mechanisms responsible for longer telomere lengths in the POF population.

45

Table 3.1 Rate of telomere loss per year in women with evidence of premature reproductive aging and controls Age range

Rate of telomere decline (bp/year)

N

(years)

Mean

Lower 95% CI

Upper 95% CI

R2a

P-valueb

Control group 1

108

17-55

-40

-66

-14

0.081

0.001

Control group 2

46

37-54

26

-56

107

0.009

0.26

POF patients

34

21-50

-98

-178

-17

0.161

0.01

RM

95

24-45

-3

-40

40

0.000

0.44

Single trisomy

32

24-44

-56

-110

-2

0.130

0.02

Multiple trisomy

17

33-44

-23

-150

105

0.010

0.35

Sample group

a

R2 is a measure of the goodness of fit of the regression

a

P values are based on a one-tailed test for significance of the regression based on the t distribution.

46

Table 3.2 Raw and age-adjusted mean telomere length Mean telomere length

a

Mean age

Raw data (± SD)

Age-adjusted P-valuea,b

Sample group

N

(years)

(kb)

(kb)

Control group 1

108

36.3

8.98±1.15

8.92

Control group 2

46

41.5

8.99±1.03

9.11

0.36

POF patients

34

35.4

9.61±1.38

9.58

0.01, 0.32

RM

95

35.8

8.47±0.92

8.46

0.0004, 0.02

Single trisomy

32

36.3

8.80±0.78

8.80

0.39, 0.26

Multiple trisomy

17

39.3

8.42±0.69

8.52

0.11, 0.06

P values for comparison to control group 1, and 2, respectively.

b

ANCOVA (k=2 for comparison to Control group 1 or 2) was used to adjust raw telomere length data by age in comparisons between

groups.

47

Figure 3.1 Telomere-specific qPCR. In summary, primers designed to be complementary to the TTAGGG repeats anneal to the telomere template during the first round of PCR replication. Over consecutive rounds, the preferential amplicon is the shortest possible length, totaling the length of the 2 primers. Subsequently, the relative quantification of the telomere amplification will be proportional to the number of binding sites (ie. the length of the telomere).

48

14 12

Flow FISH tel length (kb)

10 y = 7.2497x + 2.5008 R² = 0.9212

8 6 4 2 0 0.000

0.200

0.400

0.600

0.800

1.000

1.200

1.400

qPCR t/s tel length

Figure 3.2 Correlation between telomere-specific qPCR and flow-FISH techniques.

49

Figure 3.3 Correlation between telomere length and age in women with evidence of premature reproductive aging and controls. Age compared to telomere length (kb) in (a) control group 1, (b) control group 2, (c) POF patients, and (d) RM. 50

Figure 3.4 Correlation between telomere length and age in women with recurrent miscarriage and trisomic pregnancies. Age compared to telomere length (kb) in women with RM with (a) ST and (b) MT.

51

Chapter 4: Genetic polymorphisms in genes involved in the hypothalamuspituitary-ovarian axis 4.1

Introduction Altered levels of hormones and other factors that are involved in maintaining control of

the HPO axis can have negative effects on fertility and pregnancy. As discussed in section 1.1.1 (page 2), the central components of the HPO feedback loop are GnRH, gonadotropins (LH and FSH), and steroid hormones (estradiol and progesterone). Elevated levels of gonadotropins and estradiol have been associated with RM (Gurbuz et al. 2003, Gurbuz et al. 2004, Li et al. 2000). Similarly, elevated FSH is seen with advancing maternal age and is indicative of reduced ovarian responsiveness (Fitzgerald et al. 1998). Furthermore, endocrine disorders such as LPD, thyroid dysfunction and PCOS have also been associated with RM. Together these findings suggest that regulation of the HPO axis may be altered in these women, possibly due to genetic variation affecting the responsiveness or efficiency of receptors, enzymes and regulatory genes. As discussed in section 1.3.1.3 (page 18), there has been some evidence that genetic variation may contribute to susceptibility for RM; however there has been no comprehensive study in this area. I therefore hypothesized that genetic polymorphisms in genes involved in regulating the HPO axis would be associated with RM. To investigate this, we compared allele and genotype frequencies of short tandem repeats (STRs) and SNPs in 20 genes involved in the HPO axis (Table 4.1) among women with RM and controls. Polymorphisms assayed include those that have been previously reported to affect transcription, hormone levels or reproductive outcome.

52

4.2 4.2.1

Materials and methods Samples A total of 357 women were recruited from a Western Canadian population at the BC

Women’s Hospital & Health Centre in Vancouver, British Columbia. The case group consisted of 227 women with RM (all evaluated by a single physician, M.D.S.), which includes 49/50 RM cases from Chapter 1 and 90/95 RM cases from Chapter 2 and 88 new cases. This RM group had a mean age at time of pregnancy (SD; range) of 31.4 (6.1; 15-40) years with a total of 1379 pregnancies, of which 1027 (75%) ended in miscarriage. The mean number of miscarriages (SD; range) was 4.5 (1.9; 3-13). Chromosome results were obtained in 208 of these miscarriages, of which 110 (53%) were euploid, with a 46,XX/46,XY ratio of 0.80 (49/61). Ninety eight (47%) of the miscarriages were karyotypically abnormal, including 70 autosomal trisomies, 16 polyploidies, 3 polyploidies with trisomies, 4 unbalanced translocations, 3 monosomy X (45,X), 1 monosomy X and trisomy 21, and 1 sex chromosome trisomy (47,XXY). Carriers of a structural chromosome rearrangement were excluded from this study. Forty (18%) of the 227 women with RM had concurrent infertility. The control group used in this study consisted of 130 women of reproductive age. Proven fertility and/or regular menstrual cycles were known in 67 of these women with a mean (SD, range) menstrual cycle length of 28.4 (2.1, 23-35) days. Women with a known history of miscarriage, infertility or abnormal cycles were excluded from this study group. Reproductive history was unknown in the remaining 63 women; however, inclusion of these controls will only marginally reduce the power, as few will have irregular cycles and/or RM. On the basis of the 357 subjects, with a power of 0.80 and an α of 0.05, an effect size of 0.16 can be observed in this study (Faul et al. 2007). 53

The collection of the samples for this study was approved by the University of British Columbia Clinical Ethics Review Board. 4.2.2

Variant selection Candidate genes in this study were identified through literature search, using the search

words ‘recurrent miscarriage’, ‘fertility’, and ‘female reproduction’. Genes identified to be involved in female fertility through involvement in or modulation of the HPO axis, were further investigated for potential functional polymorphisms (Supplementary Table 4.1). Polymorphisms chosen are those that have been reported previously to be associated with reduced fertility in women and/or altered HPO axis hormone levels. In some cases, published polymorphisms could not be utilized due to technical constraints on the applied assay design in the current study (Sequenom iPlex) or due to limited available information. To assess the possibility of population stratification, a difference in ethnic distribution between cases and controls, as a confounding factor in this study, 23 ancestral informative SNPs were chosen to assay in cases and controls, as described by (Kosoy et al. 2009). Self-reported ethnicity was available for a subset of cases and controls in this study and was comparable, comprising of predominantly Caucasian women with Asian admixture. 4.2.3

Genotyping DNA was extracted from whole peripheral blood using conventional methods. Thirty-

one SNPs and 21 ancestral informative SNPs were successfully assayed using the Sequenom iPlex Assay (Sequenom Inc., San Diego, CA) by the Génome Québec Innovation Centre at McGill University, Montreal, Canada. STRs near the promoters of, or within, ESR1, ESR2, AR, and SHBG genes were assessed by PCR as previously reported (Bretherick et al. 2008).

54

4.2.4

Statistical analysis Hardy-Weinberg Equilibrium (HWE) was tested for each of the polymorphisms in

controls (Supplementary Table 4.2). Chi-squared analysis was used for comparisons of allele and genotype frequencies for the 35 polymorphisms (31 SNPs and 4 STRs) between the RM cases and controls. Within the RM cases, the comparison of mean number of miscarriages grouped by genotype for each SNP individually was completed using ANCOVA, which also corrected for differences in maternal age between groups (Pineda et al. 2010). The BenjaminiHochberg False Discovery Rate (FDR) model was used to correct for multiple analyses (Benjamini and Hochberg 1995). 4.2.5

Population stratification Twenty-one of the 23 ancestral informative SNPs were genotyped successfully, 1 was

excluded as it was not in HWE, suggesting possible genotyping error and the remaining 20 were analyzed for allele frequencies. There was no significant difference in genotype distribution of the ancestral informative SNPs between control and RM groups (Supplementary Table 4.3) suggesting that population stratification is unlikely to be a confounding factor in this study. 4.3

Results The allele distributions for AR CAG(n), ESR1 TA(n), ESR2 CA(n) and SHBG TAAAA(n)

STRs were compared between women with RM (N=227) and controls (N=130) (Table 4.2). The ESR2 CA(n) allele distribution varied between RM women and controls (p=0.03), however there is no apparent trend based on allele size. The allele and genotype distributions were compared between the RM group and controls for the 31 SNPs assayed in 20 genes (Table 4.3). The genotypes at a C/T SNP (rs37389) within intron 4 of the prolactin receptor (PRLR) gene differed between the RM group and controls with 55

an excess of heterozygotes and deficiency of homozygotes in the RM group (p=0.03). The alleles at a G/C SNP (rs41423247) within intron 2 of the glucocorticoid receptor (GCCR) gene also differed (p=0.04), with a minor allele frequency of 33.7% in RM women compared to 41.5% in controls. The odds ratio (OR) for the GG genotype in the RM group is 1.44 (95% CI, 0.93-2.24). As some effects may be more pronounced among women with multiple miscarriages, we grouped the RM cases by genotype and compared mean number of miscarriages within these groups, correcting for maternal age as a covariant (Supplementary Table 4.4). For a G/T SNP (rs2033962) within the activin receptor type 1 (ACVR1) gene, the presence of the minor T allele was associated with increased number of miscarriages in an additive fashion (p=0.02), with GG genotypes (N=160) having a mean number of miscarriages (SD) of 4.3 (1.6), GT genotypes (N=61) with 5.0 (2.3) and TT genotypes (N=7) with 5.3 (2.7); however, the OR for the presence of the T allele was not higher (1.04, 95% CI 0.65-1.66). The minor G allele for the -351A/G SNP (rs9340799) within the promoter region of the estrogen receptor α (ESR1) gene was not associated with RM. Although, there is a nonsignificant increased frequency in the GG genotype in the RM group (15%) compared to controls (9%) (p=0.11), as well as an increasing number of miscarriages observed with the number of G alleles present (p=0.08) (Supplementary Table 4.4). No difference was observed for the ESR1 397C/T (rs2234693) polymorphisms with RM or number of miscarriages (p=0.23 and p=0.25, respectively), which is in strong linkage disequilibrium (LD) with the -351A/G SNP (van Meurs et al. 2003). After using the Benjamini-Hochberg FDR model to correct for multiple comparisons, none of the associations were found to be statistically significant. 56

4.4

Discussion This study examined 35 polymorphisms within 20 genes that influence the HPO axis

(Table 4.1). I identified several candidate associations; polymorphisms within three genes (ESR2, PRLR and GCCR) were associated with RM and ACVR1 showed an additive trend of increased number of miscarriages with the minor allele. However, after correction for multiple analyses, these associations were not statistically significant. These candidate genes have previously been suggested to have a role in female fertility; therefore, a potential role in RM required investigation. Two independent studies reported that prolactin may play a role in miscarriage, with a reduction in prolactin expression in the endometrium in women with RM (Garzia et al. 2004) and the down-regulation of PRLR in women who underwent in vitro fertilization and miscarried compared to those with ongoing pregnancies (Bersinger et al. 2008). Mouse models have also shown that a lack of Prlr is associated with female infertility due to failure of embryo implantation (Ormandy et al. 1997), suggesting that the PRLR is an essential component for endometrial receptivity. Decreased activin levels have been associated with miscarriage (Prakash et al. 2005). In addition, the G/T SNP (rs2033962) in the ACVR1 gene has been associated with levels of antiMullerian hormone and follicle numbers in women with polycystic ovarian syndrome (Kevenaar et al. 2009), which in turn has been linked to RM (Rai and Regan 2006). The GCCR mediates the activity of cortisol, a marker of elevated stress. The minor allele of the Bc/I (rs41423247) polymorphism within the GCCR gene has been associated with increased cortisol levels in women on oral contraceptives who underwent psychological stress testing (Kumsta et al. 2007). Elevated levels of maternal urinary cortisol prior to 6 weeks of gestation were associated with a higher risk of miscarriage (Nepomnaschy et al. 2006). Lastly, 57

women with self-reported high levels of distress and long menstrual cycles were found to have a higher risk of miscarriage (Hjollund et al. 1999). We found a tendency towards an increased frequency of the G (major) allele of the rs41423247 polymorphism within the GCCR gene in women with RM, with an OR for the GG genotype of 1.44 (95% CI 0.65-1.66). This may suggest a difference in responsiveness to stress between control women and those with RM. Estrogen plays an essential role in follicular development and maintenance of early pregnancy. Esr1 null female mice are infertile, with no corpus luteum formation and altered gonadotropin levels, whereas, Esr2 null female mice have a subfertile phenotype with fewer number of oocytes, which may be due to decreased ovarian responsiveness to gonadotropins (Emmen and Korach 2003). There have been several studies investigating a potential association between the -397T/C and -351A/G SNPs in ESR1 and RM. The -397C allele has been associated with increased expression of the ESR1 gene (Zhai et al. 2006). An effect that may be explained by the creation of a transcription factor binding site or due to the LD with shorter TA(n) alleles in the promoter that may influence expression (Herrington et al. 2002). Alessio and coauthors (2008) assessed both these ESR1 SNPs and the ESR2 STR in 75 Brazilian women with RM and found no association. However, a recent study found an association with an increased number of miscarriages and the ESR1 haplotype composed of the -397T and -351A alleles (Pineda et al. 2010). We did not find such an association, although the observed tendencies in our data suggest that the role of ESR polymorphisms in RM may be of interest to investigate further in a larger study. Contradictory results from these different studies may be due to the differences in ascertainment of women with sporadic and RM. Historically, miscarriage risks were estimated at 15%, because only clinical pregnancies of 6 weeks or greater were included (Jacobs and 58

Hassold 1987). With the inclusion of preclinical pregnancies, miscarriage risks approach 3050% (Edmonds et al. 1982, Wilcox et al. 1988). Many cases of a single preclinical miscarriage may be due to chance rather than an increased susceptibility. This is supported by the finding that rates of chromosome errors, such as trisomy, monosomy and polyploidy, are inversely associated with number of miscarriages (Ogasawara et al. 2000). Therefore, susceptibility due to genetic variability in hormone regulation may be more likely to play a role in women with strictly defined RM. In this study, the mean number of miscarriages within the RM group is higher than most other studies, increasing the likelihood of ascertaining women at an exacerbated risk of miscarriage. RM is known to be heterogeneous in etiology. We did not stratify our sample population for primary (no prior live birth) or secondary (prior live birth) RM, nor for clinical risk factors identified. In addition, many of the miscarriages were not karyotyped; therefore, we could not compare results stratified for euploid and aneuploid miscarriages. We were unable to obtain information on menstrual cycle length or regularity for a subset of the controls and the women with RM. Ensuring all women in the control group had regular cycles would strengthen the study, possibly increasing the significance of true associations. In addition, the role of genetic variation in the HPO axis may be augmented in RM women with irregular menstrual cycles. The selection of only a few polymorphisms for each gene studied in this investigation allows only the assessment of that given site and those in LD with it. It does not, however, capture all of the genetic variation within these genes; therefore, the potential role of other SNPs and rare mutations in the risk for RM cannot be excluded. Furthermore, the synergistic effect of combinations of SNPs, particularly in extremely polymorphic genes, and gene-environmental interactions is difficult to appropriately address in association studies. A more extensive analysis 59

of the genetic variation within these genes is needed in future studies to entirely evaluate the role of the HPO axis in the risk for RM. In conclusion, in this study we investigated the association between genetic polymorphisms affecting the function of genes involved in regulating the HPO axis and RM. We identified candidate associations between RM and genetic variants in ESR2, PRLR, GCCR, and ACVR1. However, these associations were not significant after correcting for multiple comparisons. These findings may suggest that these gene variants have little or no effect on folliculogenesis and/or early maintenance of pregnancy. However, due to the limitation of sample size in this analysis, future studies in a larger, well-characterized group of women with RM are needed to determine whether these candidate genes are associated with RM.

60

Table 4.1 Summary of 35 polymorphisms assessed in this study. Gene

Description

Polymorphisms assayed

Activin receptor type 1

rs2033962

Androgen receptor

CAG repeat, rs6152

CBG

Corticosteroid-binding globulin

rs2281517

CGB5

Chorionic gonadotropin beta polypeptide 5

rs4801789

CYP17

Steroid 17-hydrolase

rs743572

CYP19

Aromatase

rs10046

ESR1

Estrogen receptor α

TA repeat, rs2234693, rs9340799

ESR2

Estrogen receptor β

CA repeat, rs1256049

FBLN1

Fibulin 1

rs9682

FSHR

Follicle-stimulating hormone receptor

rs1394205, rs6166

GCCR

Glucocorticoid receptor

rs41423247, rs6198

INHA

Inhibin α

rs35118453

LHR

Luteinizing hormone receptor

rs2293275, rs12470652

ACVR1 AR

61

Gene

Description

Polymorphisms assayed

PAPPA

Pregnancy-associated plasma protein A

rs7020782

PGR

Progesterone receptor

rs518162, rs1042838

PRL

Prolactin

rs1341239, rs2244502

PRLR

Prolactin receptor

rs9292573, rs37389, rs13354826

SHBG

Sex hormone-binding globulin

TAAAA repeat, rs6259, rs1799941, rs6257

THRB

Thyroid hormone receptor β

rs3752874

TSHR

Thyroid stimulating hormone receptor

rs2234919, rs1991517

62

Table 4.2 Allele distributions of short tandem repeat polymorphisms. Comparison of allele distributions between women with RM (N=227) and controls (N=130) for STR polymorphisms within hormone receptors.

Allele

RM Observed

Controls

(frequency)

Observed (frequency)

AR (Androgen receptor) CAG repeat

p-value* 0.631

≤20

124 (0.27)

85 (0.33)

21

81 (0.18)

44 (0.17)

22

53 (0.12)

26 (0.10)

23

62 (0.14)

31 (0.12)

≥24

134 (0.30)

74 (0.28)

ESR1 (Estrogen receptor α) TA repeat

0.250

≤13

47 (0.10)

24 (0.10)

14

135 (0.30)

87 (0.34)

15

54 (0.12)

19 (0.07)

16

13 (0.03)

11 (0.04)

17-20

46 (0.10)

21 (0.08)

21

45 (0.10)

23 (0.09)

22

28 (0.06)

26 (0.10)

23

37 (0.08)

25 (0.10)

≥24

49 (0.11)

24 (0.09)

ESR2 (Estrogen receptor β) CA repeat

0.026

≤18

74 (0.16)

33 (0.13)

19

25 (0.06)

20 (0.08)

20

11 (0.02)

15 (0.06)

21

32 (0.07)

17 (0.07) 63

RM Observed

Controls

Allele

(frequency)

Observed (frequency)

22

52 (0.12)

45 (0.17)

23

164 (0.36)

76 (0.29)

≥24

96 (0.21)

54 (0.21)

SHBG (Sex hormone-binding globulin) TAAAA repeat

a

≤6

121 (0.27)

61 (0.23)

7

31 (0.07)

23 (0.09)

8

149 (0.33)

95 (0.38)

9

111 (0.24)

63 (0.24)

≥10

42 (0.10)

18 (0.07)

p-value*

0.511

Chi-square analysis

64

Table 4.3 Genotype distributions of single nucleotide polymorphisms. Comparison of genotype distributions between women with RM (N=227a) and controls (N=130 a) for hormone pathway gene polymorphisms.

SNP

Genotype

RM Observed

Controls

P

P

(frequency)

Observed (frequency)

genotypesb

allelesb

0.896

ACVR1 (Activin receptor type 1) rs2033962

GG

159 (0.70)

92 (0.71)

GT

61 (0.27)

33 (0.25)

TT

7 (0.03)

5 (0.04)

GG

161 (0.71)

95 (0.73)

GA

65 (0.29)

33 (0.25)

AA

1 (0.00)

2 (0.02)

0.920

AR (Androgen receptor) rs6152

0.752

0.920

0.767

0.699

0.803

1.000

0.323

0.320

0.307

0.663

CBG (Corticosteroid-binding globulin) rs2281517

TT

141 (0.62)

79 (0.61)

TC

76 (0.34)

43 (0.33)

CC

10 (0.04)

8 (0.06)

CGB5 (Chorionic gonadotropin β polypeptide 5) rs4801789

CC

123 (0.55)

69 (0.53)

CT

72 (0.32)

46 (0.35)

TT

29 (0.13)

15 (0.12)

AA

77 (0.34)

54 (0.42)

AG

105 (0.46)

51 (0.39)

GG

45 (0.20)

25 (0.19)

62 (0.27)

33 (0.25)

CYP17 (Steroid 17-hydrolase) rs743572

CYP19 (Aromatase) rs10046

TT

65

RM Observed SNP

Controls

Genotype

(frequency)

Observed (frequency)

TC

108 (0.48)

72 (0.55)

CC

57 (0.25)

25 (0.19)

TT

70 (0.31)

43 (0.33)

TC

103 (0.45)

66 (0.51)

CC

54 (0.24)

21 (0.16)

AA

101 (0.45)

57(0.44)

AG

90 (0.40)

62 (0.48)

GG

35 (0.15)

11 (0.09)

GG

199 (0.88)

115 (0.88)

GA

24 (0.11)

14 (0.11)

AA

4 (0.02)

1 (0.01)

CC

91 (0.40)

40 (0.31)

CT

109 (0.48)

71 (0.55)

TT

27 (0.12)

19 (0.15)

P genotypes

P b

allelesb

ESR1 (Estrogen receptor α) rs2234693

rs9340799

0.231

0.230

0.113

0.450

1.000

0.764

0.208

0.130

0.677

0.454

0.831

0.624

ESR2 (Estrogen receptor β) rs1256049

FBLN1 (Fibulin 1 ) rs9682

FSHR (Follicle-stimulating hormone receptor) rs1394205

rs6166

GG

112 (0.50)

69 (0.53)

GA

93 (0.41)

52 (0.40)

AA

21 (0.09)

9 (0.07)

AA

67 (0.30)

41 (0.32)

AG

118 (0.52)

68 (0.52)

GG

42 (0.19)

21 (0.16)

GCCR (Glucocorticoid receptor) 66

RM Observed

Controls

P

P b

allelesb

SNP

Genotype

(frequency)

Observed (frequency)

genotypes

rs41423247

GG

102 (0.45)

47 (0.36)

0.120

0.044

GC

97 (0.43)

58 (0.45)

CC

28 (0.12)

25 (0.19)

AA

164 (0.73)

88 (0.69)

0.381

0.269

AG

55 (0.25)

34 (0.27)

GG

5 (0.02)

6 (0.05)

CC

147 (0.65)

81 (0.62)

0.878

0.671

CT

68 (0.30)

41 (0.32)

TT

12 (0.05)

8 (0.06)

0.148

0.279

1.000

1.000

0.947

0.842

0.387

0.584

rs6198

INHA (Inhibin α ) rs35118453

LHR (Luteinizing hormone receptor) rs2293275

rs12470652

GG

92 (0.41)

41 (0.32)

GA

90 (0.40)

65 (0.50)

AA

41 (0.18)

23 (0.18)

TT

200 (0.88)

115 (0.88)

TC

27 (0.12)

15 (0.12)

CC

0 (0.00)

0 (0.00)

PAPPA (Pregnancy-associated plasma protein A) rs7020782

AA

100 (0.44)

56 (0.43)

AC

105 (0.46)

60 (0.46)

CC

22 (0.10)

14 (0.11)

CC

190 (0.84)

114 (0.88)

CT

36 (0.16)

14 (0.11)

TT

1 (0.00)

2 (0.02)

PGR (Progesterone receptor) rs518162

67

RM Observed

Controls

P

P b

allelesb

SNP

Genotype

(frequency)

Observed (frequency)

genotypes

rs1042838

GG

176 (0.78)

92 (0.71)

0.409

0.282

GT

46 (0.20)

34 (0.26)

TT

5 (0.02)

3 (0.02)

GG

94 (0.42)

52 (0.40)

0.923

0.752

GT

102 (0.45)

59 (0.45)

TT

30 (0.13)

19 (0.15)

AA

105 (0.47)

69 (0.53)

0.340

0.446

AT

104 (0.46)

49 (0.38)

TT

17 (0.08)

11 (0.09)

TT

100 (0.44)

54 (0.40)

0.304

0.842

TC

94 (0.41)

63 (0.48)

CC

33 (0.15)

13 (0.12)

CC

178 (0.78)

108 (0.83)

0.028

0.920

CT

45 (0.20)

15 (0.12)

TT

4 (0.02)

7 (0.05)

TT

105 (0.47)

57 (0.44)

0.807

0.572

TC

92 (0.41)

53 (0.41)

CC

28 (0.12)

19 (0.15)

0.842

0.823

0.733

0.920

PRL (Prolactin) rs1341239

rs2244502

PRLR (Prolactin receptor) rs9292573

rs37389

rs13354826

SHBG (Sex hormone-binding globulin) rs6259

rs1799941

GG

171 (0.75)

96 (0.74)

GA

51 (0.23)

31 (0.24)

AA

3 (0.01)

2 (0.02)

GG

138 (0.61)

77 (0.59)

68

RM Observed SNP

rs6257

Controls

Genotype

(frequency)

Observed (frequency)

GA

75 (0.33)

47 (0.36)

AA

14 (0.06)

6 (0.05)

TT

194 (0.85)

109 (0.84)

TC

32 (0.14)

19 (0.15)

CC

1 (0.00)

2 (0.02)

P genotypes

P b

allelesb

0.791

0.617

0.425

0.377

0.286

0.357

0.277

0.224

THRB (Thyroid hormone receptor β) rs3752874

CC

172 (0.76)

91 (0.70)

CT

49 (0.22)

36 (0.28)

TT

6 (0.03)

3 (0.02)

TSHR (Thyroid stimulating hormone receptor) rs2234919

rs1991517

a

CC

196 (0.86)

118 (0.91)

CA

30 (0.13)

11 (0.08)

AA

1 (0.00)

1 (0.01)

CC

184 (0.81)

112 (0.86)

CG

39 (0.17)

17 (0.13)

GG

4 (0.02)

1 (0.01)

N is the total number of samples run on the platform, the number of successful genotype calls

may be less for some SNPs. b

Chi-square analysis

69

Chapter 5: Placental DNA methylation associated with pregnancy outcomes 5.1

Introduction During development, changes in DNA methylation are associated with the stable

differentiation of cell types from a single totipotent zygote to the embryonic and extra-embryonic tissues (Monk 1995). However, subtle environmental perturbations during embryonic development or gametogenesis can disrupt the setting of these epigenetic marks (Hogg et al. 2012, Velker et al. 2012). Imprinted genes have been shown to be environmentally sensitive and may be particularly susceptible to disruptions in DNA methylation (Faulk and Dolinoy 2011). As many imprinted genes have an essential role in placental growth and differentiation (Reik and Walter 2001), they are attractive candidates to test for abnormalities associated with poor placental function. Furthermore DNA methylation may be inherently more variable in the placenta that in other tissues, possibly due to its need to be responsive to a variety of signals in its function as a mediator of exchange between the fetus and mother (Yuen and Robinson 2011). Aberrant DNA methylation, arising during embryo development or gametogenesis, has been suggested as a potential cause of pregnancy loss. Extreme DNA methylation values at several imprinted loci were more frequent in the muscle tissue of stillborns and spontaneous abortions than in induced abortions (Pliushch et al. 2010). In addition, aberrant hemimethylation and mono-allelic expression of the maternal CBG5 gene, a component of the placental hormone human chorionic gonadotropin, was seen in trophoblast from 2 RM cases and one elective termination, but not in healthy pregnancies (Ankolkar et al. 2012, Uuskula et al. 2011). A comprehensive analysis of genome-wide and site-specific patterns of DNA methylation in miscarriage and RM is needed to evaluate the frequency and nature of epigenetic errors in early pregnancy. In this study, we evaluated patterns in DNA methylation in chorionic 70

villus samples from the products of conception of women with RM, isolated miscarriage, or elective termination. We hypothesized that placental villi of karyotypically normal miscarriages, particularly those occurring in women with RM, would exhibit aberrant DNA methylation globally and/or at specific loci. Using both microarray and targeted approaches we assessed 1) differences at specific candidate loci between groups; 2) overall differences and individual outliers at imprinted loci; and 3) “global” alterations in DNA methylation based on long interspersed element (LINE-1) sequences and all CpG sites interrogated. 5.2 5.2.1

Materials and methods Samples Placental chorionic villus samples were obtained anonymously from miscarriage samples

evaluated through the Embryopathology Laboratory at the BC Children’s and Women’s Hospital. Cases were comprised of karyotypically normal miscarriage samples from an independent cohort of women with a history of recurrent miscarriage (RM; N=33), and women with a single miscarriage (M; N=21). As part of routine clinical workup, all miscarriage specimens with culture failure or in which the karyotype was 46,XX, were further assessed with comparative genome hybridization. First trimester chromosomally normal control samples were separately ascertained from anonymous 8-12 week elective terminations (TA; N=16). Chromosome constitution of TA samples was assessed with multiple ligation-dependent probe amplification. Table 5.1 describes and compares the demographics for each study group. 5.2.2

Array-based quantification of DNA methylation DNA from placental chorionic villus samples was purified using the DNeasy Blood and

Tissue Kit (Qiagen, Hilden, Germany), and 750ng of DNA for each sample was bisulfite converted using the EZ DNA Methylation Kit (Zymo Research Corporation, Irvine, USA). A 71

subset of 10 RM and 10 TA villi samples were run independently on the Infinium HumanMethylation27 BeadChip array (Illumina Inc., San Diego, USA). Each probes in this array interrogates DNA methylation at one of 27,578 CpG sites throughout the genome, and is enriched for sites within gene promoters. Samples and arrays were prepared as per the manufacturer’s specifications (Illumina Inc., San Diego, USA). Data were normalized to background intensity levels using GenomeStudio Software 2011 (Illumina Inc., San Diego, USA). Probes on the X and Y chromosomes (N=1092) and with bad detection p-values (p>0.01 in any sample; N=424) were omitted, leaving a total of 26,062 probes for analysis. The array data from this study are publicly available at NCBI Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo) under accession number #16704108. 5.2.3

Targeted DNA methylation Candidate CpG sites from 4 significant Infinium array probes, 7 imprinted genes (Table

5.2), and LINE-1 sequences were assayed using bisulfite pyrosequencing. This was performed on a Pyromark MD machine using the PyroGold SQA reagent kit (Qiagen, Hilden, Germany). Primers were designed using the PSQ Assay Design Software: Version 1.0.6 or selected from published studies (Supplementary Table 5.1). Each 15 µL PCR reaction contained 1xPCR Buffer (Qiagen, Hilden, Germany), 3 mM Gibco dNTPs (Invitrogen, Carlsbad, USA), 0.9 U HotStart Taq DNA polymerase (Qiagen, Hilden, Germany), 6 µM of each of the forward and reverse primers, and ~15 ng of bisulfite converted DNA. Cycling conditions were: 95°C for 15 min, 95°C for 30 s, 55°C for 30 s, 72°C for 30 s x40, with a final extension of 72°C for 10 min. The correlation between the Infinium beta values and percent methylation measured by bisulfite pyrosequencing was highly significant for all candidate CpG sites (p<0.0001; Supplementary Figure 5.1). PCR cycling conditions for LINE-1 primers were those from the commercially 72

available LINE-1 assay (Qiagen, Hilden, Germany). DNA methylation of LINE-1 sequences has been used extensively in the literature as a proxy for global methylation, as they are CG rich and distributed widely throughout the genome (Price et al. 2012). 5.2.4

Statistical analysis Student’s two-tailed t-test was used to compare study group demographics, including

maternal age, gestational age, and number of gestations, parity and miscarriages. Fisher’s Exact Probability Test was used to compare the male to female ratio among the study groups. For the 26,062 quality Infinium array probes, beta values were corrected using M-value conversion (Du et al. 2010) and colour channel normalization in R Statistical Software 2.12.0 (The R Project for Statistical Computing, Auckland, New Zealand). Significance Analysis of Microarrays (Stanford University, Stanford, USA) was utilized on M-values to select significant candidate CpG sites. An FDR of less than 0.05 was used in conjunction with an absolute difference (delta beta) of greater than 0.05 average beta values between the RM and TA groups (Figure 5.1). Within the candidate probe list, probes containing SNPs and/or showing bimodal distribution of beta values (N=3) or cross-hybridizing to multiple locations within the genome (N=0) were eliminated from analysis (Supplementary Table 5.2). Gene ontology analysis was completed using ermineJ version 2.1.21 (Lee et al. 2005b). A gene score resampling method was done using delta beta values as the probe score for each of the 26,062 Infinium array probes. The ‘Best’ score for gene replicates was used, with 10,000 iterations to obtain corrected p-values. In addition to standard biological processes gene ontology groups, custom gene sets included in this analysis were: 1) genes previously associated with RM (Baek 2004, Rull et al. 2012), 2) imprinted genes (http://igc.otago.ac.nz/), which was

73

further subdivided into 2a) maternally expressed imprinted genes and 2b) paternally expressed imprinted genes. Average DNA methylation of target regions, assessed by bisulfite pyrosequencing, was compared between groups using the non-parametric Mann-Whitney (k=2) and Kruskal-Wallis (k=3) tests. The post hoc Dunn’s Multiple Comparisons Test was used to further assess which pair-wise comparisons were contributing to significance in the Kruskal-Wallis analysis. Correction for multiple comparisons was done using the Benjamini Hochberg FDR method (Benjamini and Hochberg 1995). The relationship between DNA methylation (%) by pyrosequencing and Infinium average beta values or gestational age was evaluated using linear correlation. Fisher’s Exact test was used to compare the number of outliers for DNA methylation at imprinted genes between the RM, M and TA groups. Principle component analysis (PCA) utilizing DNA methylation (%) values for all first trimester placental samples (N=70) at the 12 targeted loci assessed by bisulfite pyrosequencing in this study (Supplementary Table 5.1), including LINE-1 sequences, was done to identify outlier samples. GraphPad Prism 4 (GraphPad Software, Inc., La Jolla, USA), VassarStats (Vassar College, Poughkeepsie, USA) and R Statistical Software 2.12.0 (The R Project for Statistical Computing, Auckland, New Zealand) were used for statistical analyses and graphing. 5.3 5.3.1

Results Array-based quantification of DNA methylation The Illumina Infinium HumanMethylation27 BeadChip array quantifies DNA

methylation of CpG sites within the proximal promoter regions of almost 15,000 genes throughout the genome. Using a criteria of an FDR<0.05 and delta beta>0.05, 14 differentially methylated candidate CpG sites were identified from the comparison of 10 RM and 10 TA 74

placental samples. Three of these were omitted due to the presence of a SNP in the probe binding region and/or bimodal distribution of beta values (suggestive of being influenced by a SNP). Five of the remaining 11 candidates were more highly methylated in cases compared to controls, while 6 were less methylated (Supplementary Table 5.2). Four of these candidates were selected for confirmatory follow up based on functional relevance of the associated gene, including cytochrome P450, subfamily 1A, polypeptide 2 (CYP1A2) cg04968473, defensing β 1 (DEFB1) cg24292612, adenomatous polyposis coli (APC) cg20311501, and AXL tyrosine kinase receptor (AXL) cg14892768. These sites were further assessed in the larger sample populations of 33 RM, 21 M, and 16 TA samples, with bisulfite pyrosequencing. There was a significant difference in methylation at CYP1A2 promoter region between groups (p=0.002; Figure 5.2A), which post hoc analysis identified as a significant increase in average methylation in M (64.4%) compared to TA (50.6%; p<0.01); while the RM group was also increased relative to the TA group, this was not significant. DEFB1 showed a difference in methylation between groups (p=0.008, Figure 5.2B), in which both RM (9.3%) and M (7.9%) had marginally decreased methylation compared to TA (11.3%; p<0.05). The result for APC was not confirmed in this larger sample set (Figure 5.2C). Finally, altered methylation was observed at the AXL promoter (p=0.02; Figure 5.2D), with an increase in RM (59.1%) compared to TA (52.0%; p<0.05) in post hoc analysis. Previous studies have shown that gestational age has a strong influence on DNA methylation at many sites throughout the genome in placental villi (Novakovic et al. 2011). Unsupervised clustering showed that the TA samples do not cluster separately and are thus of similar nature and gestational age as the RM samples (Figure 5.3). However, as gestational age is not known for the remainder of the follow up group of TAs, the influence of gestational age 75

was considered in further analyses. Using the RM and M samples (N=54), we identified a significant positive correlation between gestational age and DNA methylation at the CYP1A2 promoter region (r=0.58, p<0.0001; Supplementary Figure 5.2). It is thus possible that the observed increase in methylation in the RM and M groups compared to the TA group (Figure 5.2A), may be confounded by differences in gestational age. There was no significant correlation between maternal age and DNA methylation at any of the assayed regions (Supplementary Figure 5.3). ErmineJ gene ontology analysis, of the 26,062 Infinium array probes, was utilized to assess whether certain gene ontologies were enriched among those sites showing the largest delta betas between the RM and TA groups. In other words, this analysis is not based on the small subset of probes we identified as candidates, but on the distribution of differences between the two groups relative to the gene content on the array. There was a highly significant enrichment of imprinted genes (p=9.53E-10) and genes previously associated with RM (p=9.51E-06; Supplementary Table 5.3). When subcategorizing the imprinted genes by parental origin of expression, maternally expressed genes (paternally methylated) were more significantly enriched (p=1.90E-09) than paternally expressed genes (p=7.98E-06). Notably, there were a higher percentage of gene ontology biological processes involved in immune response among those identified as significantly enriched in this dataset (18.4%). 5.3.2

DNA methylation at imprinted genes DNA methylation was assessed at imprinted loci using bisulfite pyrosequencing (Figure

5.4), due to the previously reported association with pregnancy loss (Pliushch et al. 2010) and the observed enrichment in the ErmineJ gene ontology analysis. These 7 loci, including maternally methylated PLAGL1, SGCE, KvDMR1 and SNRPN and paternally methylated H19/IGF2 ICR1, 76

CDKN1C, and MEG3, were selected because they have been previously demonstrated to maintain their imprints in placenta (Bourque et al. 2011, Yuen et al. 2011). After correction for multiple comparisons, there was significantly increased average methylation in M (51.7%) compared to RM (48.2%) and TA (47.8%) at the H19/IGF2 ICR1 (p<0.0001; Supplementary Figure 5.4). In addition, there was no correlation between DNA methylation at any of the 7 imprinted loci and gestational age (Supplementary Figure 5.5). As we may expect only a subset of miscarriages to be attributed to aberrant DNA methylation due to the heterogeneous etiology, we sought to identify individual samples that display values outside of the normal range. It was previously demonstrated that spontaneous abortion and stillbirth were associated with increased number of outliers, defined as greater than 1.5x the inter-quartile range, for DNA methylation at imprinted loci (Pliushch et al. 2010). Using this criterion, we observed a significant increase in the number of outliers for DNA methylation in the RM (3.9%) group compared to M (0.0%) and TA (0.9%; p=0.02; Table 5.2). 5.3.3

‘Global’ measures of DNA methylation To assess whether there was overall dysregulation of DNA methylation in any sample, or

groups as a whole, two ‘global’ measures, using representative dispersed sequences, were used: 1) the average of the 26,062 Infinium array probes, and 2) the average methylation at consensus LINE-1 sequences. There were no significant differences in average methylation observed between groups, after correction for multiple comparisons (Figure 5.5). To investigate individual samples, principle components analysis (PCA) was utilized to identify those that show distinct patterns of DNA methylation at the 12 targeted loci within this study, including LINE-1 sequences. In a PCA plot comparing the two primary principle components, attributing 44% of the variance within the dataset, outlier samples were identified (Figure 5.6). The identified 77

outlier samples show altered DNA methylation at multiple loci (Table 5.3), defined as greater or less than one or more standard deviations from the mean measured in all first trimester chorionic villi (N=70), although no consistent pattern of dysregulation was evident. 5.4

Discussion In this study, we assessed DNA methylation globally and at targeted loci in placental

samples from first trimester RM, M and TA pregnancies. This was used to address whether there were differences between these groups and whether a subset of pregnancies showed distinct epigenetic patterns. Using both candidacy and gene ontology approaches, several differences in DNA methylation were associated with RM and/or isolated miscarriage. Two candidate CpG sites, near the promoters of DEFB1 and AXL, were identified as differentially methylated between RM and TA, while DEFB1 and CYP1A2 were differentially methylated between M and TA. As a subset of the RM and TA groups did not cluster separately using Infinium array profiles, it does not appear that mode of fetal demise is associated with gross differences in cell composition or epigenetic gene regulation in the placenta. The gene ontology analysis of differential methylation on the Infinium array showed an enrichment of genes previously associated with RM, imprinted loci and immunological pathways. In addition, there were an increased number of outliers for DNA methylation at 7 imprinted loci among RM placentae. While we did not observe an overall trend of altered ‘global’ DNA methylation in any group, specific placental samples in each of the three comparison groups showed altered methylation at multiple loci. The CYP1A2 gene promoter region showed an increase in DNA methylation in the M placental samples compared to TA. However, the contribution of gestational age on the observed association cannot be addressed, as information was not available for the TA samples. 78

There is a strong positive correlation between DNA methylation at this site and gestational age in RM and M samples, therefore the association would need to be independently validated in a well-characterized population. CYP1A2 is of particular interest, given its role in caffeine and drug metabolism. High CYP1A activity, a combination of both isoform 1 and 2, in first trimester placenta correlates strongly with maternal age and is associated with maternal smoking and alcohol consumption (Collier et al. 2002). Maternal intake of caffeine in conjunction with genetically altered metabolic activity of CYP1A2 has been associated with both karyotypically normal miscarriage and RM (Sata et al. 2005, Signorello et al. 2001). Together these findings raise the possibility that altered expression of CYP1A2, may be reflective of genetic and environmental influences that contribute to risk for miscarriage. At the promoter of DEFB1 (hBD-1) there was an incremental decrease in methylation in RM and M placental villi compared to TA. DEFB1 encodes for an antimicrobial peptide involved in the innate immune response, which is expressed from placental tissues (King et al. 2007). Increased placental expression of DEFB1 was observed in HIV-positive women (Aguilar-Jimenez et al. 2011), and a trend towards increased levels was observed in women with preterm premature rupture of membranes and chorioamnionitis (Polettini et al. 2011). The placenta provides an immunological barrier between the mother and fetus, protecting the genetically distinct fetus from the maternal immune system. Furthermore, the process of implantation is mediated by the immune system and it has been suggested that miscarriage may be a process of an exaggerated inflammatory response by the mother (Christiansen 2012). There is conflicting evidence as to whether infection during pregnancy is associated with RM (Rai and Regan 2006); however gene expression studies (Baek 2004, Krieg et al. 2012), as well as the

79

observed enrichment of gene ontology groups involved in immune response in our gene ontology analysis, support some role of immune function in risk for RM. CpG sites within the promoter regions of APC and AXL, two putative imprinted genes, were also identified in the candidate analysis (Choufani et al. 2011, Yuen et al. 2011). In follow up, only AXL showed a consistent increase in methylation in the RM compared to the TA samples. The paternally methylated AXL functions to promote cell proliferation, although its role in the placenta has not been studied. Interestingly, a knockout of 3 tyrosine kinases in mice, including Axl, resulted in lupus erythematosus and recurrent fetal loss (Lu and Lemke 2001). A threshold mechanism has been proposed; suggesting that an accumulation of aberrant DNA methylation at developmentally important loci, such as imprinted genes, passed a tolerated threshold may result in the miscarriage of pregnancy (Pliushch et al. 2010). Supporting this hypothesis, Pliushch and coauthors identified outliers in 4.6% and 1.0% of muscle samples from spontaneous abortions and induced abortions, respectively, at 6 imprinted genes (Pliushch et al. 2010). Using the same definition of outlier DNA methylation, but placental samples, we report similar differences between the RM (3.9%) and TA (0.9%) groups at 7 imprinted genes (4/6 from the Pluishch study), while we observed no outliers in M group. As the previous study did not stratify cases based on pregnancy history, it is possible these represent a similar subset of patients at increased risk for miscarriage. Using ErmineJ gene ontology analysis, we additionally observed an enrichment of imprinted genes in those sites showing greater differences in beta values between RM and TA groups. Deletion of a single gene, Trim28, in the oocytes of female mice resulted in RM with no liveborns and the corresponding fetuses showed widespread loss of DNA methylation at imprinted loci, particularly those which are paternally methylated (Messerschmidt et al. 2012). 80

Interestingly, we also observed a stronger enrichment of paternally imprinted (maternally expressed) genes in the gene ontology analysis. However, there was no apparent increase in the number of outliers at paternally methylated imprinted loci using a targeted approach (Table 2); however, this may be due to the very small number of outliers identified in this study and the limited analysis of only 3 paternally methylated genes. It has been hypothesized that maternal protein complexes within the oocyte provide protection of germline differentially methylated regions near imprinted loci, before embryonic transcription initiates (Messerschmidt et al. 2012). Therefore, dysregulation of these maternal effect genes, either genetically or environmentally, may contribute to risk for RM with corresponding DNA methylation abnormalities in the embryo. These types of genes may be potential candidates for future study in women with RM with no liveborns and evidence for dysregulation of methylation at imprinted genes in these miscarriages. There are several limitations to this study. First, there was incomplete information on obstetrical history and clinical investigations of the women with RM. A well-defined population would allow comparisons of DNA methylation levels with specific clinical features. Obtaining exact gestational ages for the TA cohort would also improve the study power, allowing statistical correction for this covariate. Assessment of gene expression corresponding to the observed DNA methylation changes would also strengthen the findings; however, due the complex nature of sample collection from spontaneous or scheduled abortions, the duration of time for placental tissue degradation is extensive and this is detrimental to the integrity of placental RNA (Avila et al. 2010). Also, a criticism has been that DNA methylation may not be a stable epigenetic mark at imprinted loci in the placenta (Lewis et al. 2004); to address this, we specifically targeted sites that show maintenance of DNA methylation throughout gestation (Bourque et al. 2011). 81

Furthermore, the observed frequency of outliers at imprinted loci was similar between first trimester placenta and fetal somatic tissue (Pliushch et al. 2010). The observed differences in DNA methylation between RM, M and TA groups appear to be limited to specific loci, as ‘global’ DNA methylation was not altered. This contrasts with a recent study that found decreased average methylation of all genomic CpG sites and altered expression of DNA methyltransferases in placental villi of early pregnancy losses, although with no correction for gestational age (Yin et al. 2012). The targeted differences combined with findings of the gene ontology analysis suggest that changes in placental DNA methylation of genes involved in environmental adaptation, immune response and imprinted genes, may contribute to the etiology of RM. It is, however, difficult to determine whether these differences are causal, or a consequence of placental adaptation to an unhealthy embryo. Evidence suggesting that aberrant establishment or maintenance of DNA methylation in the embryo may contribute to miscarriage is mounting (Messerschmidt et al. 2012, Pliushch et al. 2010, Yin et al. 2012). Studies from mouse suggest that altered DNA methylation in the embryo may impair implantation and normal growth (Yin et al. 2012). Alternatively, the demise of pregnancy is marked by declining progesterone and altered uterine immune cell composition (King et al. 1989), suggesting that apoptosis of placental cells and/or a cellular response to the termination of pregnancy is possible and may be reflected by changes in DNA methylation. Future studies with a larger, well-characterized sample population will allow for a more comprehensive assessment of small differences in DNA methylation between groups. Several samples showed distinct patterns of altered DNA methylation, not only at imprinted loci, but at several of the 12 targeted loci investigated. A more extensive genomic analysis of these dysregulated samples may further elucidate the nature of these altered patterns. 82

In addition, the clinical relevance of these findings will need to be determined, elucidating whether these differences in DNA methylation are more common in placentae associated with adverse pregnancy outcomes.

83

Table 5.1 Comparison of demographics for the recurrent miscarriage, miscarriage and elective termination study groups. Student’s two-tailed t-test was used to statistically compare groups for each variable, unless otherwise denoted. RM (N=33) M (N=21)

TA (N=16)

P-value

Maternal age (years) (Mean±SD)

33.7±5.0

31.1±8.7

NA

0.17

Fetal male:female ratio

17:16

11:10

9:7

0.95*

Gestational age (weeks) (Mean ±SD)

9.5±2.4

12.6±3.2

1st trimester

0.001

Gestations [Median (range)]

4 (3-9)

1 (1-4)

NA

<0.0001

Parity [Median (range)]

0 (0-2)

0 (0-2)

NA

0.48

Miscarriages [Median (range)]

3 (3-9)

1 (1-1)

NA

<0.0001

SD = standard deviation; NA = not available *

Fisher’s Exact Probability Test.

84

Table 5.2 Frequency of outliers at imprinted loci. Comparison of the number of outliers (defined as greater than 1.5x the interquartile range) for DNA methylation at 7 imprinted loci between RM, M and TA groups (p=0.02). Gene/

Location

Methylated

Average methylation (%)

RM

M

TA

allele

(N=70)

(N=33)

(N=21)

(N=16)

Region PLAGL1

6q24.2

M

53.7

3

0

0

SGCE

7q12.3

M

48.1

1

0

0

KvDMR1

11p15.5

M

61.4

0

0

0

SNRPN

15q11.2

M

48.7

1

0

0

H19/IGF2

11p15.5

P

49.2 2

0

1

ICR1 CDKN1C

11p15.5

P

24.8

1

0

0

MEG3

14q32.3

P

36.3

1

0

0

9/231 3.9%

0/147 0.0%

1/112 0.9%

Total number of outliers Percentage M = maternal; P = paternal

85

Table 5.3 Patterns of DNA methylation among outlier samples. Variability of DNA methylation observed at 12 loci among samples identified in the Principle Component Analysis (PCA) as outliers. PCA

Sample

plot

Gestational Karyotype age (wks)

DNA methylation range PLAGL1

SGCE

KvDMR1

SNRPN

#

H19/IGF2

CDKN1C

MEG3

CYP1A2

DEFB1

APC

AXL

LINE1

ICR1

8

RM27

13.6

46,XY

++

+

+

++

N

N

+

+

N

N

N

N

16

RM43

11.1

46,XX

N

N

+

N

+

--

++

N

++

--

--

++

35

M2

13.5

46,XY

+

N

+

N

N

N

+

N

N

+

N

N

36

M4

6

46,XX

N

-

N

N

N

-

+

+

+

--

--

++

39

M18

10

46,XX

++

N

++

N

N

N

N

N

N

+

N

N

59

TA6

NI

46,XX

N

N

N

N

N

--

+

+

+

--

--

++

61

TA9

NI

46,XX

N

N

N

-

N

-

+

N

+

--

--

++

N represents a normal range of methylation (within one standard deviation [SD] of the mean in all first trimester placental samples [N=70]); +/- is more/less than 1 SD from the mean; ++/-- is more/less than 2 SD from the mean.

86

Figure 5.1 Venn diagram of significant Infinium array candidates. Venn diagram of the Illumina Infinium HumanMethylation27 BeadChip probes identified using either a false discovery rate (FDR) <0.05 or an absolute difference of beta values (Delta beta) >0.05 between the RM (N=10) and TA (N=10) groups, and those in common.

87

Figure 5.2 DNA methylation at 4 candidate promoter regions. Box plots comparing DNA methylation (%) at the promoter regions of A) CYP1A2 (p=0.002), B) DEFB1 (p=0.008), C) APC (p=0.18), and D) AXL (p=0.02) genes between RM (N=33), M (N=21) and TA (N=16) groups. The box plot whiskers indicate 1.5x the inter-quartile range, while the open circles are outlier values. The horizontal bars with asterisk indicate which comparisons were statistically significant in post hoc pair wise analysis (* p<0.05; ** p<0.01).

88

Figure 5.3 Unsupervised clustering of the 20 samples run on the Infinium array. Unsupervised clustering of RM (N=10) and TA (N=10) samples run on the Infinium array. Gestational ages are denoted for RM samples (blue).

89

Figure 5.4 Box plots of DNA methylation at 7 imprinted loci. Box plots of DNA methylation (%) at 7 imprinted loci for all first trimester placental samples (N=70). The box plot whiskers indicate 1.5x the inter-quartile range, while outlier values are denoted for each group: RM (circle), M (square), TA (triangle). 90

Figure 5.5 Comparison of measures of ‘global’ methylation. Comparison of measures of ‘global’ DNA methylation using: A) average methylation at LINE-1 consensus sequences (p=0.03) between RM (N=33), M (N=21), and TA (N=16) groups and B) Infinium array probe average (p=0.19) between RM (N=10) and TA (N=10). The box plot whiskers indicate 1.5x the inter-quartile range, while the open circles are outlier values.

91

Figure 5.6 Principle component plot of all samples. PCA plot of component 1 vs. component 2 (44% of variance) for DNA methylation (%) at 12 targeted loci among RM (N=33, blue), M (N=21, green) and TA (N=16, black) placental samples. Red arrows represent the vectors for each of the 12 assays and outliers are circled.

92

Chapter 6: Discussion In this thesis, I investigated several genetic and epigenetic factors that may contribute to the etiology of RM. These included the mutational analysis of the synaptonemal complex gene SYCP3, measurement of telomeres as a representation of biological aging, genotyping of polymorphisms in genes involved in the HPO axis and assessing DNA methylation patterns in placental villi. In this discussion, I will summarize the main findings and their significance, highlight the strengths and limitations, and discuss future directions for this research. 6.1

Summary and significance of findings RM is a heterogeneous, multifactorial trait and despite expecting small contributions of

genetic and epigenetic factors to risk, identifying associations has proven challenging. In this thesis, I have found that genetic variants in SYCP3 and HPO axis genes likely do not contribute significantly to the etiology of RM. However, associations with aspects of chromosome biology, such as maternal telomere length and placental DNA methylation, suggest that biological aging and placental development are important areas of future research. The results from Chapter 2 contradict earlier findings of an association between RM and mutations in the SYCP3 gene, as no mutations were identified among a population of 50 women with RM and an aneuploid loss. This finding has been further supported by a recent publication that also found no mutations in SYCP3 among 56 women with RM or an aneuploid loss (LopezCarrasco et al. 2012). To date, there have been no examples of mutations leading to aneuploid miscarriage or RM in humans. This is despite several studies in mouse that have identified candidate genes involved in meiosis, that when mutated result in increased rates of aneuploidy in oocytes (Murdoch et al. 2013, Shin et al. 2010, Yuan et al. 2002). While further research may

93

reveal this as a contributing risk factor in isolated cases, the present study provides evidence that this is not a common risk factor for RM and is not justified to be assessed routinely in the clinic. The identification of shorter telomeres among women with RM supports the hypothesis that there may be an altered rate of biological aging among these women. While there was no significant change in the rates of telomere decline, average telomere lengths were shorter among women with RM compared to control women from the general population and those ascertained on the basis of advanced reproductive health. Although shorter telomere lengths in the oocyte have been suggested to predispose to non-disjunction (Treff et al. 2011), there was not a pronounced effect observed among those women in the RM group with aneuploid losses. The impact of this study is emphasized by its already 20 citations in the literature. While maternal telomere length cannot be used as a clinical prognostic test, the observations in this study may hint at underlying factors that may be associated with both shortened telomeres and RM. These underlying factors may include increased exposure to stress (Epel et al. 2004), altered hormonal profile (Lee et al. 2005a), autoimmunity (Jeanclos et al. 1998) or, in fact, truly reflect reproductive aging (Aydos et al. 2005). The investigation of 35 functional polymorphisms in genes involved in the HPO axis identified several associations with RM; however there is a need for independent verification, as these were not significant after correction for multiple comparisons. These weak associations may suggest that disruptions of HPO axis gene function or expression may individually have a small contribution to risk for RM. The CA(n) STR in the ESR2 gene showed altered allele distribution in RM relative to controls, however no trend was apparent. The heterozygous C/T genotype in the PRLR gene polymorphism (rs37389) was overrepresented among women with RM. The G allele in the GCCR gene polymorphism (rs41423247), which has been previously 94

associated with altered response to stress, was also associated with RM. Finally, the T allele in the ACVR1 gene polymorphism (rs2033962) was associated with increased number of miscarriages, in an additive manner. This is the first study to perform a more comprehensive investigation of genetic variation in the HPO axis and lays the framework for future studies. If the candidates identified are validated and assessed for their contribution to changes in hormone levels, these may provide markers that can be tested in conjunction with endocrine profiles to allow for more personalized hormone treatments with improved efficacy. The assessment of global and targeted DNA methylation in RM, miscarriage and elective termination placental villi in this thesis has been an important scientific contribution. Several groups have suggested that dysregulation of DNA methylation, especially at imprinted loci, may be a cause of pregnancy loss (Messerschmidt et al. 2012, Pliushch et al. 2010); although no comprehensive study had been done. Using the Infinium array, 11 candidate loci were identified with differential DNA methylation between RM and elective terminations. Despite the identification of a limited number of candidates, using gene ontology analysis, I inferred that there may be altered methylation profiles at imprinted genes, genes previously associated with RM and immune response genes in placental villi of RM cases. While these changes may be indicative of causal factors, more likely they represent changes in vascularization and immune response at the maternal-fetal interface commonly associated with RM and miscarriage. Targeted assaying of imprinted genes showed an increase in the number of outlier methylation values among RM cases, consistent with previous reports in miscarriages and stillbirths (Pliushch et al. 2010). However, given that these outliers were identified in <5% of cases it may not be a valuable prognostic marker for routine clinical use. Using ‘global’ measures of DNA methylation, no difference was observed between groups; however, a subset 95

of samples, not restricted to the RM group, showed altered methylation profiles at multiple loci. This may suggest that there is dysregulated growth or development of these samples. While errors in DNA methylation may not be a significant contributor to chromosomally normal miscarriage, aberrant patterns of DNA methylation are observed in a subset of cases. Whether these changes are associated with a pathological phenotype is yet to be determined. 6.2

Strengths and limitations Previous studies of RM have used varied patient inclusion criteria, which can contribute

to contradictory associations and unclear findings. All women have a risk of miscarriage and to identify a distinct subset of women at increased risk requires stringent criteria. In these studies I have used the definition of 3 or more consecutive miscarriages, as recommended by the European Society of Human Reproduction and Embryology (Daya 2005). The American Society for Reproductive Medicine has recently defined RM as two or more non-consecutive miscarriages (Practice Committee of the American Society for Reproductive Medicine 2013). However, using this criterion, studies are likely including many women who have had two miscarriages by chance rather than due to an underlying predisposition. In support of this view, it has been reported that rates of aneuploidy decrease as the number of consecutive miscarriages increases (Ogasawara et al. 2000); suggesting that women with higher rates of miscarriage have differing contributing etiological factors. However, this has been contested by a study that found no difference in the frequency of associated factors between those women with two or more, versus three or more miscarriages (Jaslow et al. 2010). To further delineate the differences in etiology, a more comprehensive analysis may be needed using an RM cohort with 5 or more losses, as this appears to be a threshold where a dramatic shift in the rate of aneuploidy occurs (Ogasawara et al. 2000). An additional consideration is the 96

age of the women; as the rates of aneuploidy among RM patients >35 years old, even with 5 more losses, is more similar to that of women with isolated miscarriage (Christiansen et al. 2008). Our total study cohort is a particular strength of these studies, given the large proportion of women with ≥5 miscarriages (35%) and whose age at first miscarriage was ≤35 (75%). This therefore enriched our patient cohort with women likely to have chromosomally normal miscarriages, despite only having karyotypic information on 20% of miscarriages. Furthermore, all patients have been evaluated by one clinician, allowing consistent assessment of associated clinical factors and reproductive histories. In addition to the variable definition of RM, there are many small studies in different populations that have reported conflicting genetic associations. Our investigation assessing the association between genetic polymorphisms and RM (Chapter 3) was enhanced by the use of ancestral informative SNPs to address population stratification as a confounder. Minor allele frequencies can vary dramatically depending on the population and given the diversity and admixture of most urban centres, particularly Vancouver, this is an important consideration for any association study using these types of populations. Furthermore, the power to assess a difference was increased by using control women not only from the general population, but selected to be reproductively healthy, with no history of infertility and/or miscarriage and at least one pregnancy after the age of 37. As we would expect women from the general population to contain a variety of reproductive profiles, these reproductively healthy women represent the opposing end of a spectrum as our RM cohort. There are however several important limitations to these studies. Despite our relatively large cohort of women with RM, the sample size limits our ability to subcategorize women based 97

on clinical characteristics for a more refined analysis of interactions between genetic and clinical features. Such analyses may provide clinically valuable biomarkers for specific risk groups. Furthermore, due to the complex nature of RM, genetic and environmental factors may have small cumulative contributions to risk, and larger cohorts will be needed to assess these and their interactions. Another limitation with this cohort of women with RM is the incomplete karyotypic information of all miscarriages. However, this is a common challenge among RM studies, as routine clinical assessment of fetal chromosomal constitution is usually only done after the third miscarriage, if at all. As women with a miscarriage resulting from meiotic nondisjunction represent a distinct etiological group from those with euploid miscarriages, it is likely that there is misclassification of some women in our case population, reducing the study power. Investigating aspects of chromosome biology, such as telomere length or DNA methylation, among RM patients presents certain challenges, particularly regarding tissue- and cell type-specific differences. The measurement of telomere length in whole peripheral blood is based on an average of all chromosomes and all cell types in this sample. Furthermore, I was unable to delineate whether shortened telomere lengths in RM women was indicative of limited oocyte viability due to systemically shortened telomere lengths, elevated stress due to the condition, or associated with a coexisting factor. Similarly, the changes in DNA methylation in placental villi may be reflecting tissue composition differences, a response to an unhealthy embryo or maternal factor, or a causal epigenetic defect. Some of these questions may be answered with the analysis of additional tissues at several time points in future studies. 6.3

Future directions Future studies need to be designed specifically to help unravel the maternal versus fetal

causes of RM. While maternal causes are largely speculative at this time, associations with 98

immunological, endocrinological, and thrombophilic factors suggest that the remodeling and maintenance of the maternal endometrium is essential. Fetal causes generally refer to chromosomal imbalances, such as aneuploidy, but may also include changes in DNA methylation, telomeres, or recessive or de novo lethal mutations. While fetal factors may be expected to have a larger contribution to the etiology of isolated miscarriage, maternal predisposition may result in recurrence of these errors, leading to RM. Despite this complexity, in my opinion there are two primary research outcomes that should be strived for: 1) identify women predisposed to RM due to a maternal factor and 2) assess mechanistically how maternal and fetal factors detrimentally impact pregnancy. There are several considerations and exciting areas of future study that can enable the field to work towards these goals. An important enhancement that would improve the ability to detect genetic associations is refinement of the RM and control cohorts. Christiansen and coauthors (2008) suggested that the inclusion of women with ≥5 miscarriages and ≤35 years of age would enhance risk estimates by reducing the contribution of miscarriages caused by chromosomal abnormalities and other fetal factors (Christiansen et al. 2008). While increasing these thresholds would further the homogeneity of this RM group, attaining appropriate sample sizes would become more challenging. In our study, reproductively healthy controls were defined as no history of miscarriage and a healthy pregnancy late in reproductive life; however, this group could be further refined by selecting only women with regular menstrual cycles and at least two live births. The addition of these two criterion would help eliminate women with endocrinological conditions and those women who may be susceptible to secondary RM. Much of current genetics research on the etiology of RM is centred on case control studies that are primarily candidate-driven investigations in the areas of thrombophilia, 99

immunology, and endocrinology. While these have provided valuable insight into the pathways involved in RM, there has been limited progress in identifying genetic biomarkers of risk. It is likely that many genetic variants may have a small contribution to overall risk for RM and there are challenges in obtaining adequate sample sizes to detect these associations. Synthesizing findings from larger studies and meta-analyses may lead to the eventual characterization of sets of risk biomarkers (Christiansen et al. 2008). Progress in this area may be expedited by subclassifying patients based on clinical features and performing association studies in each subset separately. Psychosocial stress is a potential maternal risk factor for RM, supported by the associations of RM with shortened telomeres and a glucocorticoid receptor polymorphism identified in this thesis (Chapters 2 and 3). Several older studies have shown that supportive care among RM patients improves pregnancy success rates (Clifford et al. 1997, Liddell et al. 1991, Stray-Pedersen and Stray-Pedersen 1984). Managing patient stress is a relatively accessible and non-invasive clinical intervention, thus making this an exciting area for future research. As there is a complex interaction between environmental, genetic and epigenetic susceptibilities in multifactorial diseases, designing a study to evaluate all three will be important. Chronic stress can be indirectly measured using hair cortisol measurements (Vanaelst et al. 2012). The benefits of using this method, as compared to blood, serum or urine measurements, are that it is noninvasive, unlikely to cause acute stress and not susceptible to daily fluctuations (Russell et al. 2012). Evaluating levels of chronic stress in conjunction with genetic variants among women with RM would be a novel investigation of the impact of stress on pregnancy outcomes in a high risk population. Furthermore, the evaluation of DNA methylation changes in the maternal

100

endometrium and placentae from miscarriages among these RM women may provide insight into the pathological mechanisms of stress in pregnancy and RM risk. Alternatively or in addition to being a consequence of stress, shortened telomeres could reflect altered rates of reproductive aging among these RM women. Recent studies investigating reproductive aging in women with BRCA1/2 mutations found an association with earlier age at natural menopause (Lin et al. 2013), decreased ovarian reserve (Titus et al. 2013) and decreased telomere lengths (Martinez-Delgado et al. 2011). BRCA proteins are involved in the double strand break (DSB) repair pathway, which is essential for recombination in meiosis and alternative telomere lengthening in the early embryo (Johnson and Keefe 2013). Together these data further support the link between reproductive aging and telomere attrition. A new strongly predictive marker of biological aging is the level of DNA methylation at specific genomic loci (Hannum et al. 2013). These authors identified individuals that have faster or slower aging rates than their chronological age. These epigenetic markers of aging may be valuable as an independent measure of biological aging rates in women with evidence of premature reproductive aging, such as RM cases. 6.4

Conclusions Recurrent miscarriage is a complex condition, in which almost half of all patients have no

associated risk factor and those who do, have limited and often experimental available treatment options. A subset of women with aneuploid losses late in their reproductive life would benefit from education and planning for families earlier; however, there is a need for improved understanding of maternal factors that contribute to idiopathic RM and identification of genetic biomarkers to direct treatment and counseling for these susceptible groups of women. While the genetic and epigenetic factors associated with RM in this thesis cannot be directly used in the 101

clinic, this work lays the framework for future directions and furthers our understanding of the pathogenesis of RM.

102

References Aguilar-Jimenez W, Zapata W and Rugeles MT. Differential expression of human beta defensins in placenta and detection of allelic variants in the DEFB1 gene from HIV-1 positive mothers. Biomedica 2011:31:44-54. Aldrich CL, Stephenson MD, Karrison T, Odem RR, Branch DW, Scott JR, Schreiber JR and Ober C. HLA-G genotypes and pregnancy outcome in couples with unexplained recurrent miscarriage. Mol Hum Reprod 2001:7:1167-1172. Alessio AM, Siqueira LH, de Carvalho EC, Barini R, Mansur Ade P, Hoehr NF and AnnichinoBizzacchi JM. Estrogen receptor alpha and beta gene polymorphisms are not risk factors for recurrent miscarriage in a Brazilian population. Clin Appl Thromb Hemost 2008:14:180-185. Andalib A, Rezaie A, Oreizy F, Shafiei K and Baluchi S. A study on stress, depression and NK cytotoxic potential in women with recurrent spontaneous abortion. Iran J Allergy Asthma Immunol 2006:5:9-16. Ankolkar M, Patil A, Warke H, Salvi V, Kedia Mokashi N, Pathak S and Balasinor NH. Methylation analysis of idiopathic recurrent spontaneous miscarriage cases reveals aberrant imprinting at H19 ICR in normozoospermic individuals. Fertil Steril 2012:98:1186-1192. Anselmo J, Cao D, Karrison T, Weiss RE and Refetoff S. Fetal loss associated with excess thyroid hormone exposure. JAMA 2004:292:691-695. Arredondo F and Noble LS. Endocrinology of recurrent pregnancy loss. Semin Reprod Med 2006:24:33-39. Avila L, Yuen RK, Diego-Alvarez D, Penaherrera MS, Jiang R and Robinson WP. Evaluating DNA methylation and gene expression variability in the human term placenta. Placenta 2010:31:1070-1077. Aviv A, Shay J, Christensen K and Wright W. The longevity gender gap: are telomeres the explanation?. Sci Aging Knowledge Environ 2005:2005:pe16. Aydos SE, Elhan AH and Tukun A. Is telomere length one of the determinants of reproductive life span?. Arch Gynecol Obstet 2005:272:113-116. Baek KH. Aberrant gene expression associated with recurrent pregnancy loss. Mol Hum Reprod 2004:10:291-297. Baerlocher GM, Vulto I, de Jong G and Lansdorp PM. Flow cytometry and FISH to measure the average length of telomeres (flow FISH). Nat Protoc 2006:1:2365-2376.

103

Baier A, Alsheimer M and Benavente R. Synaptonemal complex protein SYCP3: Conserved polymerization properties among vertebrates. Biochim Biophys Acta 2007:1774:595-602. Baker TG. Radiosensitivity of mammalian oocytes with particular reference to the human female. Am J Obstet Gynecol 1971:110:746-761. Bean CJ, Hunt PA, Millie EA and Hassold TJ. Analysis of a malsegregating mouse Y chromosome: evidence that the earliest cleavage divisions of the mammalian embryo are nondisjunction-prone. Hum Mol Genet 2001:10:963-972. Beever CL, Stephenson MD, Penaherrera MS, Jiang RH, Kalousek DK, Hayden M, Field L, Brown CJ and Robinson WP. Skewed X-chromosome inactivation is associated with trisomy in women ascertained on the basis of recurrent spontaneous abortion or chromosomally abnormal pregnancies. Am J Hum Genet 2003:72:399-407. Ben-Chetrit E, Wiener-Well Y, Fadeela A and Wolf DG. Antiphospholipid antibodies during infectious mononucleosis and their long term clinical significance. J Clin Virol 2013:. Benetos A, Okuda K, Lajemi M, Kimura M, Thomas F, Skurnick J, Labat C, Bean K and Aviv A. Telomere length as an indicator of biological aging: the gender effect and relation with pulse pressure and pulse wave velocity. Hypertension 2001:37:381-385. Benjamini Y and Hochberg Y. Controlling the false disovery rate: A practical and powerful approach to multiple testing. J R Statist Soc 1995:57:289-300. Bersinger NA, Wunder DM, Birkhauser MH and Mueller MD. Gene expression in cultured endometrium from women with different outcomes following IVF. Mol Hum Reprod 2008:14:475-484. Bhalla A, Stone PR, Liddell HS, Zanderigo A and Chamley LW. Comparison of the expression of human leukocyte antigen (HLA)-G and HLA-E in women with normal pregnancy and those with recurrent miscarriage. Reproduction 2006:131:583-589. Bolor H, Mori T, Nishiyama S, Ito Y, Hosoba E, Inagaki H, Kogo H, Ohye T, Tsutsumi M, Kato T et al. Mutations of the SYCP3 gene in women with recurrent pregnancy loss. Am J Hum Genet 2009:84:14-20. Boots C and Stephenson MD. Does obesity increase the risk of miscarriage in spontaneous conception: a systematic review. Semin Reprod Med 2011:29:507-513. Bourque DK, Penaherrera MS, Yuen RK, Van Allen MI, McFadden DE and Robinson WP. The utility of quantitative methylation assays at imprinted genes for the diagnosis of fetal and placental disorders. Clin Genet 2011:79:169-175.

104

Bradley LA, Palomaki GE, Bienstock J, Varga E and Scott JA. Can Factor V Leiden and prothrombin G20210A testing in women with recurrent pregnancy loss result in improved pregnancy outcomes?: Results from a targeted evidence-based review. Genet Med 2012:14:3950. Branch DW, Silver RM and Porter TF. Obstetric antiphospholipid syndrome: current uncertainties should guide our way. Lupus 2010:19:446-452. Bretherick KL, Hanna CW, Currie LM, Fluker MR, Hammond GL and Robinson WP. Estrogen receptor alpha gene polymorphisms are associated with idiopathic premature ovarian failure. Fertil Steril 2008:89:318-324. Broekmans FJ, Knauff EA, Valkenburg O, Laven JS, Eijkemans MJ and Fauser BC. PCOS according to the Rotterdam consensus criteria: Change in prevalence among WHO-II anovulation and association with metabolic factors. BJOG 2006:113:1210-1217. Broekmans FJ, Soules MR and Fauser BC. Ovarian aging: mechanisms and clinical consequences. Endocr Rev 2009:30:465-493. Bugge M, Collins A, Petersen MB, Fisher J, Brandt C, Hertz JM, Tranebjaerg L, de LozierBlanchet C, Nicolaides P, Brondum-Nielsen K et al. Non-disjunction of chromosome 18. Hum Mol Genet 1998:7:661-669. Butler MG, Tilburt J, DeVries A, Muralidhar B, Aue G, Hedges L, Atkinson J and Schwartz H. Comparison of chromosome telomere integrity in multiple tissues from subjects at different ages. Cancer Genet Cytogenet 1998:105:138-144. Buyon JP. The effects of pregnancy on autoimmune diseases. J Leukoc Biol 1998:63:281-7. Cargill SL, Carey JR, Muller HG and Anderson G. Age of ovary determines remaining life expectancy in old ovariectomized mice. Aging Cell 2003:2:185-190. Cawthon RM. Telomere measurement by quantitative PCR. Nucleic Acids Res 2002:30:e47. Celik N, Evsen MS, Sak ME, Soydinc E and Gul T. Evaluation of the relationship between insulin resistance and recurrent pregnancy loss. Ginekol Pol 2011:82:272-275. Chan YY, Jayaprakasan K, Zamora J, Thornton JG, Raine-Fenning N and Coomarasamy A. The prevalence of congenital uterine anomalies in unselected and high-risk populations: a systematic review. Hum Reprod Update 2011:17:761-771. Check JH. Th1 and Th2 cytokine profiles in recurrent aborters may merely reflect the progesterone status. Hum Reprod 2002:17:1669-70; author reply 1670-1.

105

Cheng EY, Hunt PA, Naluai-Cecchini TA, Fligner CL, Fujimoto VY, Pasternack TL, Schwartz JM, Steinauer JE, Woodruff TJ, Cherry SM et al. Meiotic recombination in human oocytes. PLoS Genet 2009:5:e1000661. Choufani S, Shapiro JS, Susiarjo M, Butcher DT, Grafodatskaya D, Lou Y, Ferreira JC, Pinto D, Scherer SW, Shaffer LG et al. A novel approach identifies new differentially methylated regions (DMRs) associated with imprinted genes. Genome Res 2011:21:465-476. Christiansen OB. Reproductive immunology. Mol Immunol 2012:. Christiansen OB, Kolte AM, Dahl M, Larsen EC, Steffensen R, Nielsen HS and Hviid TV. Maternal homozygocity for a 14 base pair insertion in exon 8 of the HLA-G gene and carriage of HLA class II alleles restricting HY immunity predispose to unexplained secondary recurrent miscarriage and low birth weight in children born to these patients. Hum Immunol 2012:73:699705. Christiansen OB, Mathiesen O, Lauritsen JG and Grunnet N. Idiopathic recurrent spontaneous abortion. Evidence of a familial predisposition. Acta Obstet Gynecol Scand 1990:69:597-601. Christiansen OB, Steffensen R, Nielsen HS and Varming K. Multifactorial etiology of recurrent miscarriage and its scientific and clinical implications. Gynecol Obstet Invest 2008:66:257-267. Clark DA and Croitoru K. TH1/TH2,3 imbalance due to cytokine-producing NK, gammadelta T and NK-gammadelta T cells in murine pregnancy decidua in success or failure of pregnancy. Am J Reprod Immunol 2001:45:257-265. Clifford K, Rai R and Regan L. Future pregnancy outcome in unexplained recurrent first trimester miscarriage. Hum Reprod 1997:12:387-389. Clifford K, Rai R, Watson H and Regan L. An informative protocol for the investigation of recurrent miscarriage: preliminary experience of 500 consecutive cases. Hum Reprod 1994:9:1328-1332. Cocksedge KA, Saravelos SH, Metwally M and Li TC. How common is polycystic ovary syndrome in recurrent miscarriage?. Reprod Biomed Online 2009:19:572-576. Collier AC, Tingle MD, Paxton JW, Mitchell MD and Keelan JA. Metabolizing enzyme localization and activities in the first trimester human placenta: the effect of maternal and gestational age, smoking and alcohol consumption. Hum Reprod 2002:17:2564-2572. Coomarasamy A, Truchanowicz EG and Rai R. Does first trimester progesterone prophylaxis increase the live birth rate in women with unexplained recurrent miscarriages?. BMJ 2011:342:d1914.

106

Cooper GS and Sandler DP. Age at natural menopause and mortality. Ann Epidemiol 1998:8:229-235. Cooper JP, Watanabe Y and Nurse P. Fission yeast Taz1 protein is required for meiotic telomere clustering and recombination. Nature 1998:392:828-831. Craig LB, Ke RW and Kutteh WH. Increased prevalence of insulin resistance in women with a history of recurrent pregnancy loss. Fertil Steril 2002:78:487-490. Crowe ML. SeqDoC: rapid SNP and mutation detection by direct comparison of DNA sequence chromatograms. BMC Bioinformatics 2005:6:133. Csapo AI, Pulkkinen MO, Ruttner B, Sauvage JP and Wiest WG. The significance of the human corpus luteum in pregnancy maintenance. I. Preliminary studies. Am J Obstet Gynecol 1972:112:1061-1067. Cupisti S, Fasching PA, Ekici AB, Strissel PL, Loehberg CR, Strick R, Engel J, Dittrich R, Beckmann MW and Goecke TW. Polymorphisms in estrogen metabolism and estrogen pathway genes and the risk of miscarriage. Arch Gynecol Obstet 2009:280:395-400. Daya S. Methodological issues in the evaluation of treatment efficacy in recurrent miscarriage. Special Interest Group Early Pregnancy, ESHRE 2005:. De Wolf F, Carreras LO, Moerman P, Vermylen J, Van Assche A and Renaer M. Decidual vasculopathy and extensive placental infarction in a patient with repeated thromboembolic accidents, recurrent fetal loss, and a lupus anticoagulant. Am J Obstet Gynecol 1982:142:829834. den Heijer M, Koster T, Blom HJ, Bos GM, Briet E, Reitsma PH, Vandenbroucke JP and Rosendaal FR. Hyperhomocysteinemia as a risk factor for deep-vein thrombosis. N Engl J Med 1996:334:759-762. DeVos L, Chanson A, Liu Z, Ciappio ED, Parnell LD, Mason JB, Tucker KL and Crott JW. Associations between single nucleotide polymorphisms in folate uptake and metabolizing genes with blood folate, homocysteine, and DNA uracil concentrations. Am J Clin Nutr 2008:88:11491158. Doblhammer G. Reproductive history and mortality later in life: a comparative study of England and Wales and Austria. Popul Stud (Camb) 2000:54:169-176. Dockery P and Rogers AW. The effects of steroids on the fine structure of the endometrium. Baillieres Clin Obstet Gynaecol 1989:3:227-248. Dorland M, van Kooij RJ and te Velde ER. General ageing and ovarian ageing. Maturitas 1998a:30:113-118. 107

Dorland M, van Montfrans JM, van Kooij RJ, Lambalk CB and te Velde ER. Normal telomere lengths in young mothers of children with Down's syndrome. Lancet 1998b:352:961-962. Du P, Zhang X, Huang CC, Jafari N, Kibbe WA, Hou L and Lin SM. Comparison of Beta-value and M-value methods for quantifying methylation levels by microarray analysis. BMC Bioinformatics 2010:11:587. Duthie SJ and Hawdon A. DNA instability (strand breakage, uracil misincorporation, and defective repair) is increased by folic acid depletion in human lymphocytes in vitro. FASEB J 1998:12:1491-1497. Edmonds DK, Lindsay KS, Miller JF, Williamson E and Wood PJ. Early embryonic mortality in women. Fertil Steril 1982:38:447-453. Emmen JM and Korach KS. Estrogen receptor knockout mice: phenotypes in the female reproductive tract. Gynecol Endocrinol 2003:17:169-176. Emmer PM, Nelen WL, Steegers EA, Hendriks JC, Veerhoek M and Joosten I. Peripheral natural killer cytotoxicity and CD56(pos)CD16(pos) cells increase during early pregnancy in women with a history of recurrent spontaneous abortion. Hum Reprod 2000:15:1163-1169. Emmer PM, Steegers EA, Kerstens HM, Bulten J, Nelen WL, Boer K and Joosten I. Altered phenotype of HLA-G expressing trophoblast and decidual natural killer cells in pathological pregnancies. Hum Reprod 2002:17:1072-1080. Epel ES, Blackburn EH, Lin J, Dhabhar FS, Adler NE, Morrow JD and Cawthon RM. Accelerated telomere shortening in response to life stress. Proc Natl Acad Sci U S A 2004:101:17312-17315. Faddy MJ. Follicle dynamics during ovarian ageing. Mol Cell Endocrinol 2000:163:43-48. Faddy MJ, Gosden RG, Gougeon A, Richardson SJ and Nelson JF. Accelerated disappearance of ovarian follicles in mid-life: implications for forecasting menopause. Hum Reprod 1992:7:13421346. Faul F, Erdfelder E, Lang AG and Buchner A. G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 2007:39:175191. Faulk C and Dolinoy DC. Timing is everything: the when and how of environmentally induced changes in the epigenome of animals. Epigenetics 2011:6:791-797. Feichtinger J, Aldeailej I, Anderson R, Almutairi M, Almatrafi A, Alsiwiehri N, Griffiths K, Stuart N, Wakeman JA, Larcombe L et al. Meta-analysis of clinical data using human meiotic 108

genes identifies a novel cohort of highly restricted cancer-specific marker genes. Oncotarget 2012:3:843-853. Fitzgerald C, Zimon AE and Jones EE. Aging and reproductive potential in women. Yale J Biol Med 1998:71:367-381. Frosst P, Blom HJ, Milos R, Goyette P, Sheppard CA, Matthews RG, Boers GJ, den Heijer M, Kluijtmans LA and van den Heuvel LP. A candidate genetic risk factor for vascular disease: a common mutation in methylenetetrahydrofolate reductase. Nat Genet 1995:10:111-113. Garcia-Cruz R, Brieno MA, Roig I, Grossmann M, Velilla E, Pujol A, Cabero L, Pessarrodona A, Barbero JL and Garcia Caldes M. Dynamics of cohesin proteins REC8, STAG3, SMC1 beta and SMC3 are consistent with a role in sister chromatid cohesion during meiosis in human oocytes. Hum Reprod 2010:25:2316-2327. Gartler SM. X-chromosome inactivation and selection in somatic cells. Fed Proc 1976:35:21912194. Garzia E, Borgato S, Cozzi V, Doi P, Bulfamante G, Persani L and Cetin I. Lack of expression of endometrial prolactin in early implantation failure: a pilot study. Hum Reprod 2004:19:19111916. Gleicher N, Weghofer A and Barad DH. The role of androgens in follicle maturation and ovulation induction: friend or foe of infertility treatment?. Reprod Biol Endocrinol 2011:9:1167827-9-116. Glueck CJ, Wang P, Goldenberg N and Sieve-Smith L. Pregnancy outcomes among women with polycystic ovary syndrome treated with metformin. Hum Reprod 2002:17:2858-2864. Gonen Y, Balakier H, Powell W and Casper RF. Use of gonadotropin-releasing hormone agonist to trigger follicular maturation for in vitro fertilization. J Clin Endocrinol Metab 1990:71:918922. Goodman CS, Coulam CB, Jeyendran RS, Acosta VA and Roussev R. Which thrombophilic gene mutations are risk factors for recurrent pregnancy loss?. Am J Reprod Immunol 2006:56:230-236. Goswami D and Conway GS. Premature ovarian failure. Hum Reprod Update 2005:11:391-410. Granot I, Gnainsky Y and Dekel N. Endometrial inflammation and effect on implantation improvement and pregnancy outcome. Reproduction 2012:144:661-668. Greer IA. Antithrombotic treatment for recurrent pregnancy loss?. J Thromb Haemost 2011:9 Suppl 1:302-305. 109

Guise JM, Mahon SM, Aickin M, Helfand M, Peipert JF and Westhoff C. Screening for bacterial vaginosis in pregnancy. Am J Prev Med 2001:20:62-72. Gurbuz B, Yalti S, Ficicioglu C, Ozden S, Yildirim G and Sayar C. Basal hormone levels in women with recurrent pregnancy loss. Gynecol Endocrinol 2003:17:317-321. Gurbuz B, Yalti S, Ozden S and Ficicioglu C. High basal estradiol level and FSH/LH ratio in unexplained recurrent pregnancy loss. Arch Gynecol Obstet 2004:270:37-39. Guttormsen AB, Ueland PM, Nesthus I, Nygard O, Schneede J, Vollset SE and Refsum H. Determinants and vitamin responsiveness of intermediate hyperhomocysteinemia (> or = 40 micromol/liter). The Hordaland Homocysteine Study. J Clin Invest 1996:98:2174-2183. Haas DM and Ramsey PS. Progestogen for preventing miscarriage. Cochrane Database Syst Rev 2008:(2):CD003511. doi:CD003511. Hanna J, Goldman-Wohl D, Hamani Y, Avraham I, Greenfield C, Natanson-Yaron S, Prus D, Cohen-Daniel L, Arnon TI, Manaster I et al. Decidual NK cells regulate key developmental processes at the human fetal-maternal interface. Nat Med 2006:12:1065-1074. Hannum G, Guinney J, Zhao L, Zhang L, Hughes G, Sadda S, Klotzle B, Bibikova M, Fan JB, Gao Y et al. Genome-wide Methylation Profiles Reveal Quantitative Views of Human Aging Rates. Mol Cell 2013:49:359-367. Harley CB, Futcher AB and Greider CW. Telomeres shorten during ageing of human fibroblasts. Nature 1990:345:458-460. Hassold T and Chiu D. Maternal age-specific rates of numerical chromosome abnormalities with special reference to trisomy. Hum Genet 1985:70:11-17. Hassold T and Hunt P. To err (meiotically) is human: the genesis of human aneuploidy. Nat Rev Genet 2001:2:280-291. Hastie ND, Dempster M, Dunlop MG, Thompson AM, Green DK and Allshire RC. Telomere reduction in human colorectal carcinoma and with ageing. Nature 1990:346:866-868. Hatakeyama C, Anderson CL, Beever CL, Penaherrera MS, Brown CJ and Robinson WP. The dynamics of X-inactivation skewing as women age. Clin Genet 2004:66:327-332. Havelock JC, Rainey WE and Carr BR. Ovarian granulosa cell lines. Mol Cell Endocrinol 2004:228:67-78. Hecht S, Pavlik R, Lohse P, Noss U, Friese K and Thaler CJ. Common 677C-->T mutation of the 5,10-methylenetetrahydrofolate reductase gene affects follicular estradiol synthesis. Fertil Steril 2009:91:56-61. 110

Helle S, Lummaa V and Jokela J. Are reproductive and somatic senescence coupled in humans? Late, but not early, reproduction correlated with longevity in historical Sami women. Proc Biol Sci 2005:272:29-37. Herrington DM, Howard TD, Brosnihan KB, McDonnell DP, Li X, Hawkins GA, Reboussin DM, Xu J, Zheng SL, Meyers DA et al. Common estrogen receptor polymorphism augments effects of hormone replacement therapy on E-selectin but not C-reactive protein. Circulation 2002:105:1879-1882. Hill JA, Polgar K and Anderson DJ. T-helper 1-type immunity to trophoblast in women with recurrent spontaneous abortion. JAMA 1995:273:1933-1936. Hjollund NH, Jensen TK, Bonde JP, Henriksen TB, Andersson AM, Kolstad HA, Ernst E, Giwercman A, Skakkebaek NE and Olsen J. Distress and reduced fertility: a follow-up study of first-pregnancy planners. Fertil Steril 1999:72:47-53. Hogg K, Price EM, Hanna CW and Robinson WP. Prenatal and perinatal environmental influences on the human fetal and placental epigenome. Clin Pharmacol Ther 2012:92:716-726. Hsin H and Kenyon C. Signals from the reproductive system regulate the lifespan of C. elegans. Nature 1999:399:362-366. Hutchinson EW and Rose MR. Quantitative genetics of postponed aging in Drosophila melanogaster. I. Analysis of outbred populations. Genetics 1991:127:719-727. Jacobs PA and Hassold TJ. Chromosome abnormalities: origin and etiology in abortions and live births. In Vogal F and Sperling K (eds) Human Genetics. 1987. Springer-Verlag, Berlin, pp. 233244. Jacobsen BK, Knutsen SF and Fraser GE. Age at natural menopause and total mortality and mortality from ischemic heart disease: the Adventist Health Study. J Clin Epidemiol 1999:52:303-307. Janssen OE, Mehlmauer N, Hahn S, Offner AH and Gartner R. High prevalence of autoimmune thyroiditis in patients with polycystic ovary syndrome. Eur J Endocrinol 2004:150:363-369. Jaslow CR, Carney JL and Kutteh WH. Diagnostic factors identified in 1020 women with two versus three or more recurrent pregnancy losses. Fertil Steril 2010:93:1234-1243. Jeanclos E, Krolewski A, Skurnick J, Kimura M, Aviv H, Warram JH and Aviv A. Shortened telomere length in white blood cells of patients with IDDM. Diabetes 1998:47:482-486. Joachim RA, Hildebrandt M, Oder J, Klapp BF and Arck PC. Murine stress-triggered abortion is mediated by increase of CD8+ TNF-alpha+ decidual cells via substance P. Am J Reprod Immunol 2001:45:303-309. 111

Johnson J, Canning J, Kaneko T, Pru JK and Tilly JL. Germline stem cells and follicular renewal in the postnatal mammalian ovary. Nature 2004:428:145-150. Johnson J and Keefe DL. Ovarian aging: breaking up is hard to fix. Sci Transl Med 2013:5:172fs5. Josefowic SZ, Niec RE, Kim HY, Treuting P, Chinen T, Zheng Y, Umetsu DT and Rudensky AY. Extrathymically generated regulatory T cells control mucosal TH2 inflammation. Nature 2012:482:395-9. Kaprara A and Krassas GE. Thyroid autoimmunity and miscarriage. Hormones (Athens) 2008:7:294-302. Kato I, Toniolo P, Akhmedkhanov A, Koenig KL, Shore R and Zeleniuch-Jacquotte A. Prospective study of factors influencing the onset of natural menopause. J Clin Epidemiol 1998:51:1271-1276. Keefe DL, Liu L and Marquard K. Telomeres and aging-related meiotic dysfunction in women. Cell Mol Life Sci 2007:64:139-143. Keefe DL, Marquard K and Liu L. The telomere theory of reproductive senescence in women. Curr Opin Obstet Gynecol 2006:18:280-285. Kevenaar ME, Themmen AP, van Kerkwijk AJ, Valkenburg O, Uitterlinden AG, de Jong FH, Laven JS and Visser JA. Variants in the ACVR1 gene are associated with AMH levels in women with polycystic ovary syndrome. Hum Reprod 2009:24:241-249. Kheshtchin N, Gharagozloo M, Andalib A, Ghahiri A, Maracy MR and Rezaei A. The expression of Th1- and Th2-related chemokine receptors in women with recurrent miscarriage: the impact of lymphocyte immunotherapy. Am J Reprod Immunol 2010:64:104-112. Kim NW, Piatyszek MA, Prowse KR, Harley CB, West MD, Ho PL, Coviello GM, Wright WE, Weinrich SL and Shay JW. Specific association of human telomerase activity with immortal cells and cancer. Science 1994:266:2011-2015. King A, Balendran N, Wooding P, Carter NP and Loke YW. CD3- leukocytes present in the human uterus during early placentation: phenotypic and morphologic characterization of the CD56++ population. Dev Immunol 1991:1:169-190. King A, Burrows T, Verma S, Hiby S and Loke YW. Human uterine lymphocytes. Hum Reprod Update 1998:4:480-485. King A, Wellings V, Gardner L and Loke YW. Immunocytochemical characterization of the unusual large granular lymphocytes in human endometrium throughout the menstrual cycle. Hum Immunol 1989:24:195-205. 112

King AE, Kelly RW, Sallenave JM, Bocking AD and Challis JR. Innate immune defences in the human uterus during pregnancy. Placenta 2007:28:1099-1106. King K, Smith S, Chapman M and Sacks G. Detailed analysis of peripheral blood natural killer (NK) cells in women with recurrent miscarriage. Hum Reprod 2010:25:52-58. Kinugawa C, Murakami T, Okamura K and Yajima A. Telomerase activity in normal ovaries and premature ovarian failure. Tohoku J Exp Med 2000:190:231-238. Klose RJ and Bird AP. Genomic DNA methylation: the mark and its mediators. Trends Biochem Sci 2006:31:89-97. Kosoy R, Nassir R, Tian C, White PA, Butler LM, Silva G, Kittles R, Alarcon-Riquelme ME, Gregersen PK, Belmont JW et al. Ancestry informative marker sets for determining continental origin and admixture proportions in common populations in America. Hum Mutat 2009:30:6978. Kovalevsky G, Gracia CR, Berlin JA, Sammel MD and Barnhart KT. Evaluation of the association between hereditary thrombophilias and recurrent pregnancy loss: a meta-analysis. Arch Intern Med 2004:164:558-563. Kovats S, Main EK, Librach C, Stubblebine M, Fisher SJ and DeMars R. A class I antigen, HLA-G, expressed in human trophoblasts. Science 1990:248:220-223. Krabbendam I, Franx A, Bots ML, Fijnheer R and Bruinse HW. Thrombophilias and recurrent pregnancy loss: a critical appraisal of the literature. Eur J Obstet Gynecol Reprod Biol 2005:118:143-153. Krieg SA, Fan X, Hong Y, Sang QX, Giaccia A, Westphal LM, Lathi RB, Krieg AJ and Nayak NR. Global alteration in gene expression profiles of deciduas from women with idiopathic recurrent pregnancy loss. Mol Hum Reprod 2012:18:442-450. Kumsta R, Entringer S, Koper JW, van Rossum EF, Hellhammer DH and Wust S. Sex specific associations between common glucocorticoid receptor gene variants and hypothalamus-pituitaryadrenal axis responses to psychosocial stress. Biol Psychiatry 2007:62:863-869. Kuznetsov S, Pellegrini M, Shuda K, Fernandez-Capetillo O, Liu Y, Martin BK, Burkett S, Southon E, Pati D, Tessarollo L et al. RAD51C deficiency in mice results in early prophase I arrest in males and sister chromatid separation at metaphase II in females. J Cell Biol 2007:176:581-592. Kwak JY, Beaman KD, Gilman-Sachs A, Ruiz JE, Schewitz D and Beer AE. Up-regulated expression of CD56+, CD56+/CD16+, and CD19+ cells in peripheral blood lymphocytes in pregnant women with recurrent pregnancy losses. Am J Reprod Immunol 1995:34:93-99. 113

Kwak-Kim JY, Chung-Bang HS, Ng SC, Ntrivalas EI, Mangubat CP, Beaman KD, Beer AE and Gilman-Sachs A. Increased T helper 1 cytokine responses by circulating T cells are present in women with recurrent pregnancy losses and in infertile women with multiple implantation failures after IVF. Hum Reprod 2003:18:767-773. Laird SM, Tuckerman EM, Cork BA, Linjawi S, Blakemore AI and Li TC. A review of immune cells and molecules in women with recurrent miscarriage. Hum Reprod Update 2003:9:163-174. Laird SM, Tuckerman EM and Li TC. Cytokine expression in the endometrium of women with implantation failure and recurrent miscarriage. Reprod Biomed Online 2006:13:13-23. Lamb NE, Sherman SL and Hassold TJ. Effect of meiotic recombination on the production of aneuploid gametes in humans. Cytogenet Genome Res 2005:111:250-255. Lansdorp PM. Stress, social rank and leukocyte telomere length. Aging Cell 2006:5:583-584. Lee DC, Im JA, Kim JH, Lee HR and Shim JY. Effect of long-term hormone therapy on telomere length in postmenopausal women. Yonsei Med J 2005a:46:471-479. Lee HK, Braynen W, Keshav K and Pavlidis P. ErmineJ: tool for functional analysis of gene expression data sets. BMC Bioinformatics 2005b:6:269. Lee J, Choi BC, Cho C, Hill JA, Baek KH and Kim JW. Trophoblast apoptosis is increased in women with evidence of TH1 immunity. Fertil Steril 2005c:83:1047-1049. Lewis A, Mitsuya K, Umlauf D, Smith P, Dean W, Walter J, Higgins M, Feil R and Reik W. Imprinting on distal chromosome 7 in the placenta involves repressive histone methylation independent of DNA methylation. Nat Genet 2004:36:1291-1295. Li TC, Spuijbroek MD, Tuckerman E, Anstie B, Loxley M and Laird S. Endocrinological and endometrial factors in recurrent miscarriage. BJOG 2000:107:1471-1479. Li W, Newell-Price J, Jones GL, Ledger WL and Li TC. Relationship between psychological stress and recurrent miscarriage. Reprod Biomed Online 2012:25:180-189. Liddell HS, Pattison NS and Zanderigo A. Recurrent miscarriage--outcome after supportive care in early pregnancy. Aust N Z J Obstet Gynaecol 1991:31:320-322. Lin WT, Beattie M, Chen LM, Oktay K, Crawford SL, Gold EB, Cedars M and Rosen M. Comparison of age at natural menopause in BRCA1/2 mutation carriers with a non-clinic-based sample of women in northern California. Cancer 2013:. Lissalde-Lavigne G, Fabbro-Peray P, Cochery-Nouvellon E, Mercier E, Ripart-Neveu S, Balducchi JP, Daures JP, Perneger T, Quere I, Dauzat M et al. Factor V Leiden and prothrombin 114

G20210A polymorphisms as risk factors for miscarriage during a first intended pregnancy: the matched case-control 'NOHA first' study. J Thromb Haemost 2005:3:2178-2184. Litridis I, Kapnoulas N, Natisvili T, Agiannitopoulos K, Peraki O, Ntostis P, Pantos K and Lamnissou K. Genetic variation in the CYP17 gene and recurrent spontaneous abortions. Arch Gynecol Obstet 2011:283:289-293. Liu J, Liu M, Ye X, Liu K, Huang J, Wang L, Ji G, Liu N, Tang X, Baltz JM et al. Delay in oocyte aging in mice by the antioxidant N-acetyl-L-cysteine (NAC). Hum Reprod 2012:27:14111420. Liu L, Blasco M, Trimarchi J and Keefe D. An essential role for functional telomeres in mouse germ cells during fertilization and early development. Dev Biol 2002:249:74-84. Liu L, Franco S, Spyropoulos B, Moens PB, Blasco MA and Keefe DL. Irregular telomeres impair meiotic synapsis and recombination in mice. Proc Natl Acad Sci U S A 2004:101:64966501. Liu L and Keefe DL. Defective cohesin is associated with age-dependent misaligned chromosomes in oocytes. Reprod Biomed Online 2008:16:103-112. Llahi-Camp JM, Rai R, Ison C, Regan L and Taylor-Robinson D. Association of bacterial vaginosis with a history of second trimester miscarriage. Hum Reprod 1996:11:1575-1578. Lopez-Carrasco A, Oltra S, Monfort S, Mayo S, Rosello M, Martinez F and Orellana C. Mutation screening of AURKB and SYCP3 in patients with reproductive problems. Mol Hum Reprod 2012:. Lu Q and Lemke G. Homeostatic regulation of the immune system by receptor tyrosine kinases of the Tyro 3 family. Science 2001:293:306-311. Lund M, Nielsen HS, Hviid TV, Steffensen R, Nyboe Andersen A and Christiansen OB. Hereditary thrombophilia and recurrent pregnancy loss: a retrospective cohort study of pregnancy outcome and obstetric complications. Hum Reprod 2010:25:2978-2984. Lunenfeld B, Kraiem Z and Eshkol A. The function of the growing follicle. J Reprod Fertil 1975:45:567-574. Mader SS. Figure 15.9. In Human Biology. 2006. McGraw-Hill Companies, New York, NY, pp. 299. Manor O, Eisenbach Z, Israeli A and Friedlander Y. Mortality differentials among women: the Israel Longitudinal Mortality Study. Soc Sci Med 2000:51:1175-1188.

115

Mantha S, Bauer KA and Zwicker JI. Low molecular weight heparin to achieve live birth following unexplained pregnancy loss: a systematic review. J Thromb Haemost 2010:8:263-268. Mariee N, Tuckerman E, Ali A, Li W, Laird S and Li TC. The observer and cycle-to-cycle variability in the measurement of uterine natural killer cells by immunohistochemistry. J Reprod Immunol 2012:95:93-100. Martinez-Delgado B, Yanowsky K, Inglada-Perez L, Domingo S, Urioste M, Osorio A and Benitez J. Genetic anticipation is associated with telomere shortening in hereditary breast cancer. PLoS Genet 2011:7:e1002182. McMaster MT, Librach CL, Zhou Y, Lim KH, Janatpour MJ, DeMars R, Kovats S, Damsky C and Fisher SJ. Human placental HLA-G expression is restricted to differentiated cytotrophoblasts. J Immunol 1995:154:3771-3778. McNamee K, Dawood F and Farquharson R. Recurrent miscarriage and thrombophilia: an update. Curr Opin Obstet Gynecol 2012:24:229-234. McNatty KP and Sawers RS. Relationship between the endocrine environment within the Graafian follicle and the subsequent rate of progesterone secretion by human granulosa cells in vitro. J Endocrinol 1975:66:391-400. Messerschmidt DM, de Vries W, Ito M, Solter D, Ferguson-Smith A and Knowles BB. Trim28 is required for epigenetic stability during mouse oocyte to embryo transition. Science 2012:335:1499-1502. Michimata T, Sakai M, Miyazaki S, Ogasawara MS, Suzumori K, Aoki K, Nagata K and Saito S. Decrease of T-helper 2 and T-cytotoxic 2 cells at implantation sites occurs in unexplained recurrent spontaneous abortion with normal chromosomal content. Hum Reprod 2003:18:15231528. Mizutani E, Suzumori N, Ozaki Y, Oseto K, Yamada-Namikawa C, Nakanishi M and SugiuraOgasawara M. SYCP3 mutation may not be associated with recurrent miscarriage caused by aneuploidy. Hum Reprod 2011:26:1259-1266. Moini A, Tadayon S, Tehranian A, Yeganeh LM, Akhoond MR and Yazdi RS. Association of thrombophilia and polycystic ovarian syndrome in women with history of recurrent pregnancy loss. Gynecol Endocrinol 2012:28:590-593. Monk M. Epigenetic programming of differential gene expression in development and evolution. Dev Genet 1995:17:188-197. Muller HG, Chiou JM, Carey JR and Wang JL. Fertility and life span: late children enhance female longevity. J Gerontol A Biol Sci Med Sci 2002:57:B202-6. 116

Murdoch B, Owen N, Stevense M, Smith H, Nagaoka S, Hassold T, McKay M, Xu H, Fu J, Revenkova E et al. Altered cohesin gene dosage affects Mammalian meiotic chromosome structure and behavior. PLoS Genet 2013:9:e1003241. Nagai J, Lin CY and Sabour MP. Lines of mice selected for reproductive longevity. Growth Dev Aging 1995:59:79-91. Nagaoka SI, Hassold TJ and Hunt PA. Human aneuploidy: mechanisms and new insights into an age-old problem. Nat Rev Genet 2012:13:493-504. Nagirnaja L, Venclovas C, Rull K, Jonas KC, Peltoketo H, Christiansen OB, Kairys V, Kivi G, Steffensen R, Huhtaniemi IT et al. Structural and functional analysis of rare missense mutations in human chorionic gonadotrophin beta-subunit. Mol Hum Reprod 2012:18:379-390. Nepomnaschy PA, Welch KB, McConnell DS, Low BS, Strassmann BI and England BG. Cortisol levels and very early pregnancy loss in humans. Proc Natl Acad Sci U S A 2006:103:3938-3942. Ng SC, Gilman-Sachs A, Thaker P, Beaman KD, Beer AE and Kwak-Kim J. Expression of intracellular Th1 and Th2 cytokines in women with recurrent spontaneous abortion, implantation failures after IVF/ET or normal pregnancy. Am J Reprod Immunol 2002:48:77-86. Nicolaides KH, Spencer K, Avgidou K, Faiola S and Falcon O. Multicenter study of firsttrimester screening for trisomy 21 in 75 821 pregnancies: results and estimation of the potential impact of individual risk-orientated two-stage first-trimester screening. Ultrasound Obstet Gynecol 2005:25:221-226. Nigro G, Mazzocco M, Mattia E, Di Renzo GC, Carta G and Anceschi MM. Role of the infections in recurrent spontaneous abortion. J Matern Fetal Neonatal Med 2011:24:983-989. Nishikawa K, Saito S, Morii T, Hamada K, Ako H, Narita N, Ichijo M, Kurahayashi M and Sugamura K. Accumulation of CD16-CD56+ natural killer cells with high affinity interleukin 2 receptors in human early pregnancy decidua. Int Immunol 1991:3:743-750. Nishiyama S, Kishi T, Kato T, Suzuki M, Bolor H, Nishizawa H, Iwata N, Udagawa Y and Kurahashi H. A rare synaptonemal complex protein 3 gene variant in unexplained female infertility. Mol Hum Reprod 2011:17:266-271. Novakovic B, Yuen RK, Gordon L, Penaherrera MS, Sharkey A, Moffett A, Craig JM, Robinson WP and Saffery R. 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. Ogasawara M, Aoki K, Okada S and Suzumori K. Embryonic karyotype of abortuses in relation to the number of previous miscarriages. Fertil Steril 2000:73:300-304. 117

Oliver TR, Feingold E, Yu K, Cheung V, Tinker S, Yadav-Shah M, Masse N and Sherman SL. New insights into human nondisjunction of chromosome 21 in oocytes. PLoS Genet 2008:4:e1000033. Ooi PV, Russell N and O'Donoghue K. Secondary recurrent miscarriage is associated with previous male birth. J Reprod Immunol 2011:88:38-41. Ormandy CJ, Camus A, Barra J, Damotte D, Lucas B, Buteau H, Edery M, Brousse N, Babinet C, Binart N et al. Null mutation of the prolactin receptor gene produces multiple reproductive defects in the mouse. Genes Dev 1997:11:167-178. Ozdemir O, Yenicesu GI, Silan F, Koksal B, Atik S, Ozen F, Gol M and Cetin A. Recurrent pregnancy loss and its relation to combined parental thrombophilic gene mutations. Genet Test Mol Biomarkers 2012:16:279-286. Page SL and Hawley RS. The genetics and molecular biology of the synaptonemal complex. Annu Rev Cell Dev Biol 2004:20:525-558. Pal L and Santoro N. Premature ovarian failure (POF): discordance between somatic and reproductive aging. Ageing Res Rev 2002:1:413-423. Park HM, Shin SJ, Choi DH, Oh D, Lee S and Kim NK. Association between folate metabolismrelated gene polymorphisms and methylation of p16(INK4A) and hMLH1 genes in spontaneously aborted embryos with normal chromosomal integrity. Fertil Steril 2008:90:16051610. Patel RN, Quack KC, Hill JA and Schust DJ. Expression of membrane-bound HLA-G at the maternal-fetal interface is not associated with pregnancy maintenance among patients with idiopathic recurrent pregnancy loss. Mol Hum Reprod 2003:9:551-557. Perls T, Levenson R, Regan M and Puca A. What does it take to live to 100?. Mech Ageing Dev 2002:123:231-242. Perls TT, Alpert L and Fretts RC. Middle-aged mothers live longer. Nature 1997:389:133. Perls TT and Fretts RC. The evolution of menopause and human life span. Ann Hum Biol 2001:28:237-245. Pfeiffer KA, Fimmers R, Engels G, van der Ven H and van der Ven K. The HLA-G genotype is potentially associated with idiopathic recurrent spontaneous abortion. Mol Hum Reprod 2001:7:373-378. Pineda B, Hermenegildo C, Tarin JJ, Laporta P, Cano A and Garcia-Perez MA. Alleles and haplotypes of the estrogen receptor alpha gene are associated with an increased risk of spontaneous abortion. Fertil Steril 2010:93:1809-1815. 118

Pliushch G, Schneider E, Weise D, El Hajj N, Tresch A, Seidmann L, Coerdt W, Muller AM, Zechner U and Haaf T. Extreme methylation values of imprinted genes in human abortions and stillbirths. Am J Pathol 2010:176:1084-1090. Polettini J, Takitane J, Peracoli JC and Silva MG. Expression of beta defensins 1, 3 and 4 in chorioamniotic membranes of preterm pregnancies complicated by chorioamnionitis. Eur J Obstet Gynecol Reprod Biol 2011:157:150-155. Practice Committee of the American Society for Reproductive Medicine. Definitions of infertility and recurrent pregnancy loss: a committee opinion. Fertil Steril 2013:99:63. Practice Committee of the American Society for Reproductive Medicine. Evaluation and treatment of recurrent pregnancy loss: a committee opinion. Fertil Steril 2012:98:1103-1111. Prakash A, Laird S, Tuckerman E, Li TC and Ledger WL. Inhibin A and activin A may be used to predict pregnancy outcome in women with recurrent miscarriage. Fertil Steril 2005:83:17581763. Preston FE, Rosendaal FR, Walker ID, Briet E, Berntorp E, Conard J, Fontcuberta J, Makris M, Mariani G, Noteboom W et al. Increased fetal loss in women with heritable thrombophilia. Lancet 1996:348:913-916. Price EM, Cotton AM, Penaherrera MS, McFadden DE, Kobor MS and Robinson W. Different measures of "genome-wide" DNA methylation exhibit unique properties in placental and somatic tissues. Epigenetics 2012:7:652-663. Purcell SH and Moley KH. The impact of obesity on egg quality. J Assist Reprod Genet 2011:28:517-524. Quack KC, Vassiliadou N, Pudney J, Anderson DJ and Hill JA. Leukocyte activation in the decidua of chromosomally normal and abnormal fetuses from women with recurrent abortion. Hum Reprod 2001:16:949-955. Rai R and Regan L. Recurrent miscarriage. Lancet 2006:368:601-611. Regan L, Owen EJ and Jacobs HS. Hypersecretion of luteinising hormone, infertility, and miscarriage. Lancet 1990:336:1141-1144. Reik W and Walter J. Genomic imprinting: parental influence on the genome. Nat Rev Genet 2001:2:21-32. Ren A and Wang J. Methylenetetrahydrofolate reductase C677T polymorphism and the risk of unexplained recurrent pregnancy loss: a meta-analysis. Fertil Steril 2006:86:1716-1722.

119

Robinson WP, Beever C, Brown CJ and Stephenson MD. Skewed X inactivation and recurrent spontaneous abortion. Semin Reprod Med 2001:19:175-181. Rosen MP, Shen S, McCulloch CE, Rinaudo PF, Cedars MI and Dobson AT. Methylenetetrahydrofolate reductase (MTHFR) is associated with ovarian follicular activity. Fertil Steril 2007:88:632-638. Rufer N, Brummendorf TH, Kolvraa S, Bischoff C, Christensen K, Wadsworth L, Schulzer M and Lansdorp PM. Telomere fluorescence measurements in granulocytes and T lymphocyte subsets point to a high turnover of hematopoietic stem cells and memory T cells in early childhood. J Exp Med 1999:190:157-167. Rull K, Nagirnaja L and Laan M. Genetics of recurrent miscarriage: challenges, current knowledge, future directions. Front Genet 2012:3:34. Rull K, Nagirnaja L, Ulander VM, Kelgo P, Margus T, Kaare M, Aittomaki K and Laan M. Chorionic gonadotropin beta-gene variants are associated with recurrent miscarriage in two European populations. J Clin Endocrinol Metab 2008:93:4697-4706. Russell E, Koren G, Rieder M and Van Uum S. Hair cortisol as a biological marker of chronic stress: current status, future directions and unanswered questions. Psychoneuroendocrinology 2012:37:589-601. Salker M, Teklenburg G, Molokhia M, Lavery S, Trew G, Aojanepong T, Mardon HJ, Lokugamage AU, Rai R, Landles C et al. Natural selection of human embryos: impaired decidualization of endometrium disables embryo-maternal interactions and causes recurrent pregnancy loss. PLoS One 2010:5:e10287. Sata F, Yamada H, Suzuki K, Saijo Y, Kato EH, Morikawa M, Minakami H and Kishi R. Caffeine intake, CYP1A2 polymorphism and the risk of recurrent pregnancy loss. Mol Hum Reprod 2005:11:357-360. Scheffer GJ, Broekmans FJ, Looman CW, Blankenstein M, Fauser BC, teJong FH and teVelde ER. The number of antral follicles in normal women with proven fertility is the best reflection of reproductive age. Hum Reprod 2003:18:700-706. Serra V, Grune T, Sitte N, Saretzki G and von Zglinicki T. Telomere length as a marker of oxidative stress in primary human fibroblast cultures. Ann N Y Acad Sci 2000:908:327-330. Shakhar K, Rosenne E, Loewenthal R, Shakhar G, Carp H and Ben-Eliyahu S. High NK cell activity in recurrent miscarriage: what are we really measuring?. Hum Reprod 2006:21:24212425. Sherwood L. The reproductive system. In Julet M, Rose N and Arbogast M (eds) Human physiology: from cells to systems. 2004. Thompson Learning, Inc., Belmont, CA, pp. 748-802. 120

Shimada S, Kato EH, Morikawa M, Iwabuchi K, Nishida R, Kishi R, Onoe K, Minakami H and Yamada H. No difference in natural killer or natural killer T-cell population, but aberrant Thelper cell population in the endometrium of women with repeated miscarriage. Hum Reprod 2004:19:1018-1024. Shin YH, Choi Y, Erdin SU, Yatsenko SA, Kloc M, Yang F, Wang PJ, Meistrich ML and Rajkovic A. Hormad1 mutation disrupts synaptonemal complex formation, recombination, and chromosome segregation in mammalian meiosis. PLoS Genet 2010:6:e1001190. Signorello LB, Nordmark A, Granath F, Blot WJ, McLaughlin JK, Anneren G, Lundgren S, Ekbom A, Rane A and Cnattingius S. Caffeine metabolism and the risk of spontaneous abortion of normal karyotype fetuses. Obstet Gynecol 2001:98:1059-1066. Slagboom PE, Droog S and Boomsma DI. Genetic determination of telomere size in humans: a twin study of three age groups. Am J Hum Genet 1994:55:876-882. Smith KR, Mineau GP and Bean LL. Fertility and post-reproductive longevity. Soc Biol 2002:49:185-205. Smith ML and Schust DJ. Endocrinology and recurrent early pregnancy loss. Semin Reprod Med 2011:29:482-490. Snowdon DA, Kane RL, Beeson WL, Burke GL, Sprafka JM, Potter J, Iso H, Jacobs DR,Jr and Phillips RL. Is early natural menopause a biologic marker of health and aging?. Am J Public Health 1989:79:709-714. Stephenson MD. Frequency of factors associated with habitual abortion in 197 couples. Fertil Steril 1996:66:24-29. Stephenson MD, Awartani KA and Robinson WP. Cytogenetic analysis of miscarriages from couples with recurrent miscarriage: a case-control study. Hum Reprod 2002:17:446-451. Stephenson MD and Sierra S. Reproductive outcomes in recurrent pregnancy loss associated with a parental carrier of a structural chromosome rearrangement. Hum Reprod 2006:21:10761082. Stirrat GM. Recurrent miscarriage. Lancet 1990:336:673-675. Stray-Pedersen B and Stray-Pedersen S. Etiologic factors and subsequent reproductive performance in 195 couples with a prior history of habitual abortion. Am J Obstet Gynecol 1984:148:140-146. Su MT, Lin SH and Chen YC. Association of sex hormone receptor gene polymorphisms with recurrent pregnancy loss: a systematic review and meta-analysis. Fertil Steril 2011:96:14351444.e1. 121

Subramanian VV and Bickel SE. Aging predisposes oocytes to meiotic nondisjunction when the cohesin subunit SMC1 is reduced. PLoS Genet 2008:4:e1000263. Tachibana-Konwalski K, Godwin J, van der Weyden L, Champion L, Kudo NR, Adams DJ and Nasmyth K. Rec8-containing cohesin maintains bivalents without turnover during the growing phase of mouse oocytes. Genes Dev 2010:24:2505-2516. Tan WK, Lim SK, Tan LK and Bauptista D. Does low-molecular-weight heparin improve live birth rates in pregnant women with thrombophilic disorders? A systematic review. Singapore Med J 2012:53:659-663. Tang AW and Quenby S. Recent thoughts on management and prevention of recurrent early pregnancy loss. Curr Opin Obstet Gynecol 2010:22:446-451. te Velde ER and Pearson PL. The variability of female reproductive ageing. Hum Reprod Update 2002:8:141-154. Teklenburg G, Salker M, Heijnen C, Macklon NS and Brosens JJ. The molecular basis of recurrent pregnancy loss: impaired natural embryo selection. Mol Hum Reprod 2010a:16:886895. Teklenburg G, Salker M, Molokhia M, Lavery S, Trew G, Aojanepong T, Mardon HJ, Lokugamage AU, Rai R, Landles C et al. Natural selection of human embryos: decidualizing endometrial stromal cells serve as sensors of embryo quality upon implantation. PLoS One 2010b:5:e10258. Titus S, Li F, Stobezki R, Akula K, Unsal E, Jeong K, Dickler M, Robson M, Moy F, Goswami S et al. Impairment of BRCA1-Related DNA Double-Strand Break Repair Leads to Ovarian Aging in Mice and Humans. Sci Transl Med 2013:5:172ra21. Treff NR, Su J, Taylor D and Scott RT,Jr. Telomere DNA deficiency is associated with development of human embryonic aneuploidy. PLoS Genet 2011:7:e1002161. Tuckerman E, Laird SM, Prakash A and Li TC. Prognostic value of the measurement of uterine natural killer cells in the endometrium of women with recurrent miscarriage. Hum Reprod 2007:22:2208-2213. Uuskula L, Rull K, Nagirnaja L and Laan M. Methylation allelic polymorphism (MAP) in chorionic gonadotropin beta5 (CGB5) and its association with pregnancy success. J Clin Endocrinol Metab 2011:96:E199-207. van Meurs JB, Schuit SC, Weel AE, van der Klift M, Bergink AP, Arp PP, Colin EM, Fang Y, Hofman A, van Duijn CM et al. Association of 5' estrogen receptor alpha gene polymorphisms with bone mineral density, vertebral bone area and fracture risk. Hum Mol Genet 2003:12:17451754. 122

Vanaelst B, Huybrechts I, Bammann K, Michels N, de Vriendt T, Vyncke K, Sioen I, Iacoviello L, Gunther K, Molnar D et al. Intercorrelations between serum, salivary, and hair cortisol and child-reported estimates of stress in elementary school girls. Psychophysiology 2012:49:10721081. Vargas RG, Sarturi PR, Mattar SB, Bompeixe EP, Silva Jdos S, Pirri A and Bicalho Mda G. Association of HLA-G alleles and 3' UTR 14 bp haplotypes with recurrent miscarriage in Brazilian couples. Hum Immunol 2011:72:479-485. Velker BA, Denomme MM and Mann MR. Embryo culture and epigenetics. Methods Mol Biol 2012:912:399-421. Vital-Reyes V, Chhieng D, Rodriguez-Burford C, Tellez-Velasco S, Grizzle W, Chavarria-Olarte ME and Reyes-Fuentes A. Ovarian biopsy in infertile patients with ovarian dysfunction. Int J Gynecol Pathol 2006:25:90-94. von Zglinicki T. Role of oxidative stress in telomere length regulation and replicative senescence. Ann N Y Acad Sci 2000:908:99-110. Vora S, Shetty S, Khare M and Ghosh K. Placental histomorphology in unexplained foetal loss with thrombophilia. Indian J Med Res 2009:129:144-149. Wakim AN, Polizotto SL, Buffo MJ, Marrero MA and Burholt DR. Thyroid hormones in human follicular fluid and thyroid hormone receptors in human granulosa cells. Fertil Steril 1993:59:1187-1190. Warburton D, Dallaire L, Thangavelu M, Ross L, Levin B and Kline J. Trisomy recurrence: a reconsideration based on North American data. Am J Hum Genet 2004:75:376-385. Warburton D and Fraser FC. Spontaneous Abortion Risks in Man: Data from Reproductive Histories Collected in a Medical Genetics Unit. Am J Hum Genet 1964:16:1-25. Wegmann TG, Lin H, Guilbert L and Mosmann TR. Bidirectional cytokine interactions in the maternal-fetal relationship: is successful pregnancy a TH2 phenomenon?. Immunol Today 1993:14:353-356. Wei XH and Orr HT. Differential expression of HLA-E, HLA-F, and HLA-G transcripts in human tissue. Hum Immunol 1990:29:131-142. Weiner Z, Younis JS, Blumenfeld Z and Shalev E. Assessment of uterine placental circulation in thrombophilic women. Semin Thromb Hemost 2003:29:213-218. White YA, Woods DC, Takai Y, Ishihara O, Seki H and Tilly JL. Oocyte formation by mitotically active germ cells purified from ovaries of reproductive-age women. Nat Med 2012:18:413-421. 123

Wilcox AJ, Weinberg CR, O'Connor JF, Baird DD, Schlatterer JP, Canfield RE, Armstrong EG and Nisula BC. Incidence of early loss of pregnancy. N Engl J Med 1988:319:189-194. Willard HF. X chromosome inactivation and X-linked mental retardation. Am J Med Genet 1996:64:21-26. Yacobi S, Ornoy A, Blumenfeld Z and Miller RK. Effect of sera from women with systemic lupus erythematosus or antiphospholipid syndrome and recurrent abortions on human placental explants in culture. Teratology 2002:66:300-308. Yamamoto T, Takahashi Y, Kase N and Mori H. Decidual natural killer cells in recurrent spontaneous abortion with normal chromosomal content. Am J Reprod Immunol 1999a:41:337342. Yamamoto T, Takahashi Y, Kase N and Mori H. Role of decidual natural killer (NK) cells in patients with missed abortion: differences between cases with normal and abnormal chromosome. Clin Exp Immunol 1999b:116:449-452. Yin LJ, Zhang Y, Lv PP, He WH, Wu YT, Liu AX, Ding GL, Dong MY, Qu F, Xu CM et al. Insufficient maintenance DNA methylation is associated with abnormal embryonic development. BMC Med 2012:10:26-7015-10-26. Yuan L, Liu JG, Hoja MR, Wilbertz J, Nordqvist K and Hoog C. Female germ cell aneuploidy and embryo death in mice lacking the meiosis-specific protein SCP3. Science 2002:296:11151118. Yuan L, Liu JG, Zhao J, Brundell E, Daneholt B and Hoog C. The murine SCP3 gene is required for synaptonemal complex assembly, chromosome synapsis, and male fertility. Mol Cell 2000:5:73-83. Yuen RK, Jiang R, Penaherrera MS, McFadden DE and Robinson WP. Genome-wide mapping of imprinted differentially methylated regions by DNA methylation profiling of human placentas from triploidies. Epigenetics Chromatin 2011:4:10. Yuen RK and Robinson WP. Review: A high capacity of the human placenta for genetic and epigenetic variation: implications for assessing pregnancy outcome. Placenta 2011:32 Suppl 2:S136-41. Zhai Y, Zhou G, Deng G, Xie W, Dong X, Zhang X, Yu L, Yang H, Yuan X, Zhang H et al. Estrogen receptor alpha polymorphisms associated with susceptibility to hepatocellular carcinoma in hepatitis B virus carriers. Gastroenterology 2006:130:2001-2009. Zhu Y, Huo Z, Lai J, Li S, Jiao H, Dang J and Jin C. Case-control study of a HLA-G 14-bp insertion-deletion polymorphism in women with recurrent miscarriages. Scand J Immunol 2010:71:52-54. 124

Appendix A: Supplementary tables and figures for Chapter 2 Supplementary Table 2.1 Primer sequences used for sequencing analysis of the coding exons (2-9) of the SYCP3 gene. SYCP3 region

Primer Sequences

Exon 2

F 5’- TCCTTGTTCGATATCTCCTTTGA -3’ R 5’- CCGTGTCAGCAGGTTCTGTA -3’

Exon 3, 4

F 5’- AACCCAGGGAGACTTGAAAAA -3’ R 5’- TGTGAGAACAAGGCATTAAATAACA -3’

Exon 5

F 5’- ACACATTGTTTTGTTTATTAGCTCTTTTT -3’ R 5’- AGGACTATCATACTTAGAGAAAAATCAAGC -3’

Exon 6

F 5’- TTTTGGTTTCCCATCAGAAGA -3’ R 5’- TTTAAAACACATGGCCAGCA -3’

Exon 7

F 5’- GCATTGATTTTTAACACTTTCTTTT -3’ R 5’- TCCCAACAAAACCATTTGAA -3’

Exon 8

F 5’- ACCTATTTCAGCAAATAAAAT -3’ R 5’- CAAATAGATGAGCATTTGAA -3’

Exon 9

F 5’- TGGAAACTGTAAGTGATCATATTGAA -3’ R 5’- ATGTAAAATAGATTTTGTATTCCGTTT -3’

125

Appendix B: Supplementary tables and figures for Chapter 4 Supplementary Table 4.1 Polymorphisms within genes involved in hypothalamus-pituitary-ovarian axis regulation selected for investigation in this study. Gene

Description

Polymorphism

Location

UCSC code

Heterozygosity

G/T

intron

rs2033962

0.281 +/- 0.248

A/T

intron

rs1220134*

0.489 +/- 0.074

Protein change

Associations

References

Disturbed folliculogenesis in ACVR1

Activin receptor type 1

Kevenaar et al. 2009 PCOS patients

T/C

CAG repeat AR

intron

rs10497189*

0.104 +/- 0.203

exon 1

Expanded polyglutamine

PCOS; premature sexual

Hickey et al. 2002;

repeat

maturation

Lappalainen et al. 2008

Recurrent spontaneous abortion;

Karvela et al. 2008; Yang et

endometrial cancer risk

al. 2009

Androgen receptor

1733 G/A

exon 1

rs6152

0.383 +/- 0.212

synonymous

Corticosteroid-binding

Upregulated in endometrium of Misao et al. 1995

CBG globulin

T/C

promoter

rs2281517

0.333 +/- 0.236

infertile women

C/T

promoter

rs4801789

unknown

SNPs flank region associated

A/G

5’ UTR

rs710899*

0.180 +/- 0.240

with RM

C/T

intron 2

rs34335161*

0.172 +/-0.237

Protective effect toward RM

A/G

5’ UTR

rs743572

0.475 +/- 0.109

Short menstrual cycles

T/C

3' UTR

rs10046

0.483 +/- 0.091

Chorionic gonadotropin CGB5

Rull et al. 2008

beta polypeptide 5

CYP17

Steroid 17-hydrolase

Henningson et al. 2007 Dunning et al. 2004; He et al.

Estrodiol levels; Age at natural CYP19

Aromatase

2007; Guo et al. 2006; Cupisti menopause; risk for miscarriage

ESR1

A/G

exon 3

TA repeat

promoter

PvuII T/C

intron 1

rs700518*

0.473 +/- 0.112

synonymous (Val-Val)

et al. 2009 Premature ovarian failure

Bretherick et al. 2008

Endometriosis; ovarian hyper-

Hsieh et al. 2007; Georgiou et

Estrogen receptor α rs2234693

0.497 +/- 0.038

126

Gene

ESR2

Description

Estrogen receptor β

Polymorphism

Location

XbaI A/G

intron 1

CA repeat

intron 5

RsaI G/A

coding

UCSC code

rs9340799

rs1256049

Heterozygosity

Protein change

0.399 +/- 0.205

0.276 +/- 0.249

Associations

References

stimulation response; IVF

al. 1997; Sundarrajan et al.

pregnancy outcome; risk for

1999; Pineda et al. 2009

spontaneous abortion

synonymous (Val-Val)

Breast cancer risk

Tsezou et al. 2008

Ovulatory dysfunction

Sundarrajan et al. 1999

Abnormal expression of FBLN1 FBLN1

is associated with abnormal

Fibulin 1 C/T

exon 9

rs9682

0.417 +/- 0.186

synonymous

Singh et al. 2006

placenta in mice Disruption of potential TF Perez-Mayorga et al. 2000; de binding site; serum FSH levels & Castro et al. 2003

FSHR

Follicle-stimulating hormone receptor

-29 G/A

promoter

rs1394205

0.461 +/- 0.134

sensitivity of FSHR in vivo Severity of clinical features in Valkenburg et al. 2009

GNRH

Gonadotropin releasing hormone

919 A/G

exon 10

rs6166

0.473 +/- 0.112

Asn680Ser

PCOS

C/G

exon 1

rs6185*

0.415 +/- 0.188

Trp16Ser

Breast cancer adverse outcomes

Piersma et al. 2007

GC sensitivity ; hypothalamusKumsta et al. 2007 BclI G/C

intron B

rs41423247

0.441 +/- 0.162

pituitary-adrenal axis response GRβ unable to bind ligand; dom -

GCCR

Glucocorticoid receptor

A/G

3’ UTR

rs6198

0.164+/- 0.235

ve effect; psychological stress is

Hjollund et al. 1999;

associated with reduced fertility

Nepomnaschy et al. 2006

and risk for pregnancy loss Woad et al. 2009; Harris et al.

INHA

Inhibin α

Premature ovarian failure 129 C/T

promoter

rs35118453

unknown

linked with TG repeat

2005; Marozzi et al. 2002

127

Gene

Description

Polymorphism

Location

UCSC code

Heterozygosity

Protein change

Associations

References Themmen and Huhtaniemi. 2000; Huhtaniemi and

Luteinizing hormone β

Hyperfunctional promoter;

subunit

infertility

Themmen. 2005; Nagirnaja et

LHB

al. 2010; Okuno et al. 2001; T/C

exon 2

rs1800447*

0.419 +/- 0.184

Trp8Arg

Liu et al. 2005

Luteinizing hormone

G/A

exon 10

rs2293275

0.469 +/- 0.121

Ser312Asn

Breast cancer risk

receptor

T/C

exon 10

rs12470652

0.022 +/- 0.103

Asn291Ser

Increased receptor sensitivity

Pregnancy-associated plasma protein A

A/C

exon 14

rs7020782

0.477 +/- 0.105

Tyr/Ser

Recurrent pregnancy loss

Piersma et al. 2007

LHR

PAPPA

Suzuki et al. 2006

Uterine fibroids and Govindan et al. 2007 +44 C/T

PGR

promoter

rs518162

0.224 +/- 0.249

endometriosis

Progesterone receptor G/T

exon 4

rs1042838

0.144 +/- 0.226

G/T

promoter

rs1341239

0.305 +/- 0.244

Val660Leu (PROGINS)

Less responsive to progestin;

Romano et al. 2007; De

endometriosis

Carvalho et al. 2007

Altered promoter activity ; PRL Stevens et al. 2001; Lee et al.

PRL Prolactin

levels in plasma; breast cancer 2007; Vaclavicek et al. 2006 A/T

intron 1

rs2244502

0.458 +/- 0.139

T/C

intron 1

rs9292573

0.492 +/- 0.063

C/T

intron 4

rs37389

0.385 +/- 0.210

T/C

intron 1

rs13354826

0.262 +/- 0.250

risk

PRLR Prolactin receptor

Breast cancer risk

Vaclavicek et al. 2006

Xita et al. 2003; Eriksson et PCOS; serum SHBG levels TAAAA repeat

promoter

al. 2006

SHBG Sex hormone-binding globulin G/A

exon 8

rs6259

0.188 +/- 0.242

G/A

5' UTR

rs1799941

0.329 +/- 0.237

Asp327Asn

SHBG and estradiol levels; Age

Cousin et al. 2004; Eriksson et

at menopause

al. 2006; Xita et al. 2005

SHBG levels

Eriksson et al. 2006

128

Gene

Description

Polymorphism

Location

UCSC code

Heterozygosity

T/C

intron 1

rs6257

0.128 +/- 0.218

Protein change

Associations

References

Serum SHBG levels

Riancho et al. 2008

Higher serum TSH; mutation Sorenson et al. 2008; Anselmo THRB

Thyroid hormone receptor β

TSHR Thyroid stimulating hormone receptor

associated with increased rate of et al. 2004 C/T

exon 7

rs3752874

0.192 +/- 0.243

synonymous

miscarriage

C/A

exon 1

rs2234919

unknown

Pro52Thr

Reduced receptor function

Loos et al. 1995

C/G

exon 10

rs1991517

0.184 +/- 0.241

Asp727Glu

Lower levels of TSH in plasma

Peeters et al. 2003

*Due to technical limitations, assays could not be designed for these SNPs

129

Supplementary Table 4.2 Assessing single nucleotide polymorphisms for Hardy-Weinberg Equilibrium within controls. Using observed and expected genotype frequencies for controls (N=130), Hardy-Weinberg Equilibrium was calculated for all assessed SNPs; N may be less if genotyping calls failed. Alleles Observed HPO Axis Polymorphisms rs10046

Expected

CC

25

29

CT

72

65

TT rs1042838

33

37

GG

92

92

GT

34

34

3

3

TT

115

115

TC

15

14

0

0

GG

115

114

GA

14

15

1

0

TT

57

54

CT

53

59

CC rs1341239

19

16

TT

19

18

TG

59

61

GG rs1394205

52

51

GG

69

69

GA

52

51

9

9

GG

77

78

GA

47

46

6

7

TT rs12470652

CC rs1256049

AA rs13354826

AA rs1799941

AA rs1991517

p-valueb

0.644

0.888

1.000

0.863

0.719

0.966

0.995

0.956

130

Alleles

Observed

Expected

p-valueb 1.000

CC

112

112

CG

17

18

1

1

GG

92

91

GT

33

36

5

4

0.887

CC

43

44

0.951

CT

66

63

TT rs2234919

21

22

CC

118

117

AC

11

12

1

0

AA

69

68

AT

49

51

TT rs2281517

11

10

TT

79

78

TC

43

46

8

7

AA

23

24

AG

65

63

GG rs35118453

41

42

CC

81

79

TC

41

45

8

6

CC

108

103

TC

15

26

7

2

CC

91

91

TC

36

35

3

3

25

22

GG rs2033962

TT rs2234693

AA rs2244502

CC rs2293275

TT rs37389

TT rs3752874

TT rs41423247 CC

0.842

0.956

0.919

0.970

0.779

0.054

1.000

0.783

131

Alleles

Observed

Expected

CG

58

63

GG rs4801789

47

44

CC

69

65

CT

46

54

TT rs518162

15

11

CC

114

113

CT

14

17

2

1

GG

95

96

GA

33

32

2

3

AA

41

43

AG

68

63

GG rs6198

21

23

AA

88

86

GA

34

38

6

4

TT

109

108

CT

19

21

2

1

GG

96

96

GA

31

30

2

2

AA

56

57

AC

60

58

CC rs743572

14

15

AA

54

49

AG

51

62

GG rs9292573

25

20

CC

13

15

CT

63

59

TT rs6152

AA rs6166

GG rs6257

CC rs6259

AA rs7020782

p-valueb

0.504

0.863

0.920

0.848

0.723

1.000

1.000

0.961

0.393

0.856

132

Alleles

Observed

Expected

TT rs9340799

54

56

AA

57

60

AG

62

57

GG rs9682

11

14

CC

40

44

CT

71

63

p-valueb

0.726

0.592

19 23 TT Ancestral Informative Polymorphisms rs4908343 AA

89

86

AG

34

39

7

4

TT

110

110

CT

19

19

1

1

TT

110

104

TG

13

24

7

1

TT

65

65

CT

54

54

CC rs10007810

11

11

CC

84

86

CT

43

40

3

5

TT

121

121

GT

8

8

GG rs870347

0

0

TT

115

112

GT

10

17

4

1

GG

73

75

AG

52

47

AA

5

7

GG rs3737576

CC rs260690

GG rs6548616

TT rs7657799

GG rs6451722

0.546

0.863

0.019

1.000

0.730

0.791

0.584

0.737

133

Alleles rs6422347

Observed

Expected

TT

106

104

CT

21

24

3

1

TT

81

79

CT

41

45

8

6

GG

55

54

AG

58

59

AA rs10108270

17

16

CC

49

49

CA

62

62

AA rs2416791

19

19

GG

103

103

AG

25

26

2

2

GG

108

109

AG

22

20

0

1

TT

111

109

CT

16

20

3

1

TT

109

107

CT

18

22

3

1

GG

99

98

AG

28

30

3

2

CC

68

64

CT

47

54

TT rs3784230

15

11

AA

39

44

CC rs1040045

CC rs7803075

AA rs772262

AA rs9319336

CC rs7997709

CC rs9530435

AA rs9522149

p-valueb 1.000

0.779

0.980

1.000

1.000

1.000

0.863

0.863

1.000

0.543

0.403

134

Alleles

Expected

AG

74

63

GG rs11652805

17

22

AA

86

82

AG

35

42

9

5

AA

111

110

AG

17

19

GG

2

1

GG rs4891825

b

Observed

p-valueb

0.393

1.000

Chi-square analysis, genotypes were combined where necessary to meet the analysis

requirements.

135

Supplementary table 4.3 Genotype distributions for controls and recurrent miscarriage women for 21 ancestral informative single nucleotide polymorphisms. Genotype

RM (N=227)a

Controls (N=130)a

N

Frequency

N

AA

136

0.60

89

AG

79

0.35

34

GG

12

0.05

7

TT

195

0.86

CT

30

0.13

CC rs6548616

2

0.01

p-valueb

Frequency

rs4908343 0.68 0.26 0.05 0.233

rs3737576

TT

107

CT

104

CC rs10007810

110

0.85

19

0.15

1

0.01 0.863

0.48

65

0.50

0.46

54

0.42

0.06

11

0.08

13

0.502

146

0.64

84

0.65

69

0.30

43

0.33

12

0.05

3

0.02

CC CT 0.379

TT rs7657799 202

0.91

121

0.94

20

0.09

8

0.06

0

0.00

0

0.00

TT GT 0.462

GG rs870347 182

0.81

115

0.89

38

0.17

10

0.08

4

0.02

4

0.03

TT GT 0.071

GG rs6451722 141

0.62

73

0.56

72

0.32

52

0.40

GG AG

0.228

136

Genotype

RM (N=227)a

Controls (N=130)a

N 14

Frequency 0.06

N 5

Frequency 0.04

178

0.78

106

0.82

44

0.19

21

0.16

5

0.02

3

0.02

p-valueb

AA rs6422347 TT CT 0.748

CC rs1040045 144

0.63

81

0.62

72

0.32

41

0.32

11

0.05

8

0.06

TT CT 0.869

CC rs7803075 112

0.49

55

0.42

80

0.35

58

0.45

35

0.15

17

0.13

GG AG 0.217

AA rs10108270 115

0.51

49

0.38

83

0.37

62

0.48

29

0.13

19

0.15

CC CA 0.056

AA rs2416791 165

0.73

103

0.79

56

0.25

25

0.19

6

0.03

2

0.02

GG AG 0.368

AA rs772262 187

0.82

108

0.83

37

0.16

22

0.17

3

0.01

0

0.00

GG AG 1.000

AA rs9319336 180

0.79

111

0.85

37

0.16

16

0.12

10

0.04

3

0.02

TT CT CC

0.323

rs7997709

137

Genotype

RM (N=227)a

Controls (N=130)a

N 174

Frequency 0.77

N 109

Frequency 0.84

44

0.20

18

0.14

7

0.03

3

0.02

p-valueb

TT CT 0.338

CC rs9530435 165

0.73

99

0.76

50

0.22

28

0.22

11

0.05

3

0.02

GG AG 0.472

AA rs9522149 108

0.48

68

0.52

71

0.31

47

0.36

47

0.21

15

0.12

CC CT 0.084

TT rs3784230 90

0.40

39

0.30

105

0.46

74

0.57

32

0.14

17

0.13

AA AG 0.131

GG rs11652805 155

0.68

86

0.66

63

0.28

35

0.27

9

0.04

9

0.07

AA AG 0.470

GG rs4891825 189

0.83

111

0.85

37

0.16

17

0.13

1

0.00

2

0.02

AA AG GG

0.708

*N may be less for individual SNPs if genotype calls failed **Chi-square analysis

138

Supplementary Table 4.4 Mean number of miscarriages within the recurrent miscarriage group subdivided by genotype for each of 31 single nucleotide polymorphisms within genes involved in the hypothalamus-pituitary-ovarian axis. Age

Mean miscarriages

Standard Deviation

P-valueb

CC

32.9

4.42

1.58

0.757

CT

34.6

4.53

1.81

TT

33.7

4.68

2.36

GG

33.9

4.49

1.92

GT

34.0

4.76

1.91

TT

34.5

3.33

0.52

TT

34.1

4.56

1.83

TC

33.0

4.25

2.32

GG

33.6

4.59

1.97

GA

35.4

3.92

1.18

AA

41.8

5.00

0.71

CC

33.8

4.46

2.17

CT

34.2

4.51

1.99

TT

33.8

4.49

1.72

TT

33.5

4.27

1.74

TG

33.1

4.29

1.60

Genotype rs10046

rs1042838 0.211

rs12470652 0.438

rs1256049 0.217

rs13354826 0.995

rs1341239 0.106

139

GG

35.1

4.84

2.20

Genotype

Age

Mean miscarriages

Standard Deviation

P-valueb

GG

33.8

4.43

1.74

0.490

GA

34.1

4.54

1.83

AA

33.5

4.95

2.82

GG

34.2

4.47

1.70

GA

33.6

4.67

2.33

AA

33.6

4.21

1.12

CC

34.0

4.59

1.98

CG/GGa

34.0

4.20

1.49

TT

33.9

5.29

2.69

TG

33.7

5.02

2.34

GG

34.1

4.29

1.62

CC

33.4

4.89

2.10

CT

34.3

4.39

1.81

TT

33.8

4.44

1.86

CC

33.9

4.48

1.81

CA/AAa

34.1

4.77

2.40

34.1

4.59

1.77

rs1394205

rs1799941 0.627

rs1991517 0.225

rs2033962 0.021

rs2234693 0.253

rs2234919 0.423

rs2244502 AA

0.194

140

AT

33.4

4.26

1.66

Genotype

Age

Mean miscarriages

Standard Deviation

TT

34.5

4.75

2.12

TT

33.9

4.41

1.99

TC

34.0

4.70

1.71

CC

34.0

4.70

1.95

GG

34.9

4.76

2.05

GA

33.5

4.28

1.62

AA

33.4

4.56

2.13

TT

32.7

5.08

2.07

TC

33.8

4.62

2.07

CC

34.1

4.43

1.80

TT/TC

33.4

3.65

1.67

CC

34.1

4.48

1.96

TT

34.5

5.17

3.92

CT

32.8

4.53

1.94

CC

34.3

4.49

1.80

CC

32.1

4.50

1.95

CG

34.2

4.64

1.90

GG

34.2

4.40

1.89

P-valueb

rs2281517 0.541

rs2293275 0.186

rs35118453 0.434

rs37389 0.557

rs3752874 0.694

rs41423247 0.673

141

rs4801789 Genotype

Age

Mean miscarriages

Standard Deviation

P-valueb

CC

34.1

4.52

1.87

0.676

CT

34.3

4.58

2.05

TT

33.2

4.21

1.47

CC

34.0

4.53

1.87

CT/TTa

33.5

4.43

2.09

GG

33.9

4.55

1.90

GA/AAa

34.1

4.45

1.87

AA

33.7

4.43

1.74

AG

34.3

4.44

1.88

GG

33.4

4.88

2.17

GG

32.0

4.00

0.71

GA

34.3

4.38

1.91

AA

33.9

4.58

1.93

CC/CTa

34.9

4.64

1.75

TT

33.8

4.50

1.93

GG

34.0

4.62

1.83

GA/AAa

33.8

4.25

2.09

rs518162 0.779

rs6152 0.715

rs6166 0.378

rs6198 0.676

rs6257 0.723

rs6259 0.220

rs7020782

142

CC

33.6

4.64

2.57

0.406

Genotype

Age

Mean miscarriages

Standard Deviation

P-valueb

CA

34.1

4.68

1.94

AA

33.9

4.33

1.67

AA

32.7

4.52

1.83

AG

34.6

4.49

2.02

GG

34.4

4.58

1.78

CC

34.4

4.48

2.00

CT

33.9

4.47

1.68

TT

33.9

4.57

2.07

AA

33.8

4.21

1.65

AG

34.5

4.67

1.96

GG

32.9

4.94

2.21

CC

33.7

4.37

1.56

CT

34.4

4.72

2.19

TT

33.0

4.19

1.66

rs743572 0.963

rs9292573 0.933

rs9340799 0.077

rs9682

a

0.279

Less than 5 individuals homozygous for the minor allele, therefore combined with heterozygotes

for analysis b

ANCOVA

143

Appendix C: Supplementary tables and figures for Chapter 5 Supplementary Table 5.1 Primers used for assessment of DNA methylation by bisulfite pyrosequencing. Gene/Region APC

PLAGL1

SGCE

H19/IGF2 ICR1

CDKN1C

KvDMR1

MEG3

SNRPN

Primers F TTTTTTGTTTGTTGGGGATTG R Biotin/AATCCRACAACACCTCCATTCTAT S TTTGTTGGGGATTGG F Biotin/GAYGGGTTGAATGATAAATGGTAGATG R TCRACRCAACCATCCTCTTAACTAC S ACRCAACCATCCTCTTA F TGGTGTGTGTYGAAGAAATTTGATTG R Biotin/CAAACRCRATCTCCACTAAATAC S TGTGTGTYGAAGAAATTTGAT F Biotin/ACAATACAAACTCACACATCACAAC R TGAGTGTTTTATTTTTAGATGATTTT S GTGGTTTGGGTGATT F Biotin/TATTATATTATGTTAATTGTGGTTGGG R CAACAAACACTAATACACACTAATA S AACACTAATACACACTAATACTAAA F TTAGTTTTTTGYGTGATGTGTTTATTA R Biotin/CCCACAAACCTCCACACC S TTGYGTGATGTGTTTATTA F Biotin/GGTTTATATTTGGGAATTAGTTATGT R CCCCCAAATTCTATAACAAATTA S AATACTTTTTCCCTAC F Biotin/TATGTTTAGGYGGGGATGTGTG R AAAAACCACCRACACAACTAACCTTAC S CAAATACRTCAAACATCT

Reference (if applicable) Avila et al. 2010

Bourque et al. 2010

Peñaherrera et al. 2010

Horike et al. 2009

N/A

Bourque et al. 2010

N/A

Bourque et al. 2010

144

Gene/Region AXL

CYP1A2

DEFB1

LINE-1

Primers F TTTGAGGAAAGTTTGGTATTTATG R Biotin/CACTCACCCCTAAAAACCAT S TAGGATGGGTAGGGTT F TGGGGATTTGGGTTGAAAATTAG R Biotin/AAACTTCTTTCCCACTACACACATAA S GATTTGGGTTGAAAATTA F GGATTTTAGGAATTGGGGAGA R Biotin/CCTTAACTATAACACCTCCCTTCA S AGGTTTTTAGAGGTTGGA F TTTTGAGTTAGGTGTGGGATATA R Biotin/AAAATCAAAAAATTCCCTTTC S AGTTAGGTGTGGGATATAGT

Reference (if applicable) N/A

N/A

N/A

Bollati et al. 2007

145

Supplementary Table 5.2 List of 14 candidate CpG sites identified by the Illumina Infinium HumanMethylation27 BeadChip analysis, using a false discovery rate<0.05 and a Delta beta>0.05. Hg18 Location

RM

TA

Y

110888725

0.190

0.280

0.027

-0.090

Y

12q24

N

121824850

0.333

0.397

0.038

-0.064

Y

Leukocyte cellderived chemotaxin 2

5q31

N

135318187

0.628

0.689

0.011

-0.062

N

Inflammatory response (Yamagoe et al. 1996)

ALX4

Aristaless-like 4, mouse, homolog of

11p11

Y

44283975

0.399

0.460

0.039

-0.060

N

N/A

cg05316065

GSDMC

Gasdermin C

8q24

N

130868189

0.206

0.263

0.046

-0.057

N

cg06812977

RNLS

Renalase

10q23

Y

90333016

0.266

0.320

0.050

-0.055

N

N/A Regulates blood pressure and cardiac function (Xu et al. 2005)

cg13262687

POU4F2

Pou domain, class 4, transcription factor 2

4q31

Y

147779029

0.272

0.325

0.011

-0.053

N

N/A

N

Antimicrobial peptide active in epithelia of the female reproductive tract (Bensch et al. 1995); increased expression in the endometrium of women with infertility (Das et a. 2007)

Probe

Gene

Gene name

Chr

cg22879515

BTG4

B-cell translocation gene 4

11q23

cg08775774

CCDC62

Coiled-coil domain-containing protein 62

cg21783004

LECT2

cg04970352

cg24292612

DEFB1

Defensin, beta, 1

8p23

CGI

N

6722882

0.129

0.180

q-value

0.032

Delta beta

-0.052

SNP/bimodal

Evidence for functional candidacy Growth inhibitor; highly expressed in the oocyte & preimplantation embryos in mice (Buanne et al. 2000) Interacts with ERα and β, modifies expression of ER targets (Chen et al. 2009)

146

Hg18 Location

RM

TA

N

112101401

0.558

0.505

0.039

0.054

N

19q13

N

46417172

0.671

0.614

0.039

0.058

N

Vanin 2

6q23

N

133126335

0.539

0.469

0.027

0.071

N

ASB2

Ankyrin repeatand socs boxcontaining protein 2

14q23

N

93493457

0.721

0.624

0.048

0.097

Y

N/A

cg04968473

CYP1A2

Cytochrome P450, subfamily I, polypeptide 2

15q24

N

72827787

0.504

0.382

0.026

0.123

N

Caffeine & drug metabolism (Shimada et al. 2004); associated with RM (Sata et al. 2005)

cg05294455

MYL4

Myosin, light chain 4, alkali, atrial, embryonic

17q21

N

42641608

0.668

0.538

0.027

0.131

N

N/A

Probe

Gene

Gene name

Chr

cg20311501

APC

APC gene

5q22

cg14892768

AXL

AXL receptor tyrosine kinase

cg17836145

VNN2

cg19949550

CGI

q-value

Delta beta

SNP/bimodal

Evidence for functional candidacy Putative imprinted gene in placenta (Yuen et al. 2011; Guilleret et al. 2009) Mice deficient in 3 tyrosine kinases, including AXL, had systemic lupus erythematosus and recurrent fetal loss (Lu and Lemke. 2001) Increased expression in autoimmune disease (Bovin et al. 2007)

147

Supplementary Table 5.3 Significant gene ontology groups from ErmineJ analysis of 10 recurrent miscarriage and 10 elective termination samples using the Illumina Infinium HumanMethylation27 array, listed by corrected p-value. Name muscle system process muscle contraction Imprinted genes xenobiotic metabolic process cellular response to xenobiotic stimulus regulation of heart contraction Maternally expressed imprinted genes adenylate cyclase-activating G-protein coupled receptor signaling pathway circulatory system process blood circulation cellular metal ion homeostasis calcium ion homeostasis cellular divalent inorganic cation homeostasis inflammatory response metal ion homeostasis divalent inorganic cation homeostasis regulation of response to external stimulus cellular calcium ion homeostasis regulation of muscle system process leukocyte activation cell-cell adhesion leukocyte migration platelet activation

Number ID Probes of Genes GO:0003012 135 135 GO:0006936 122 122 IM Genes 68 68 GO:0006805 99 99 GO:0071466 100 100 GO:0008016 64 64 MEG Genes 36 36

Raw Score p-value 0.02775651 1.00E-12 0.0283608 1.00E-12 0.03970521 1.00E-12 0.02916292 1.87E-12 0.02907881 2.44E-12 0.03485897 1.00E-12 0.04346512 1.00E-12

GO:0007189 GO:0003013 GO:0008015 GO:0006875 GO:0055074 GO:0072503 GO:0006954 GO:0055065 GO:0072507 GO:0032101 GO:0006874 GO:0090257 GO:0045321 GO:0016337 GO:0050900 GO:0030168

0.03846768 0.02655087 0.02655087 0.02389958 0.02501162 0.02490388 0.0234339 0.02344973 0.02446927 0.02317321 0.02533335 0.02856518 0.02329665 0.02330821 0.02561222 0.02377188

31 132 132 186 143 140 200 197 149 198 135 76 174 170 123 154

31 132 132 186 143 140 200 197 149 198 135 76 174 170 123 154

1.00E-12 3.93E-11 3.93E-11 1.02E-10 1.58E-10 2.32E-10 6.66E-10 6.26E-10 1.06E-09 1.84E-09 2.12E-09 2.50E-09 3.04E-09 2.91E-09 2.83E-09 4.28E-09

Corrected p-value 6.35E-10 7.62E-10 9.52E-10 1.02E-09 1.16E-09 1.27E-09 1.90E-09 3.81E-09 1.50E-08 1.66E-08 3.53E-08 5.00E-08 6.79E-08 1.69E-07 1.70E-07 2.52E-07 4.11E-07 4.48E-07 5.01E-07 5.26E-07 5.28E-07 5.38E-07 7.08E-07

148

Name regulation of epithelial cell proliferation elevation of cytosolic calcium ion concentration regulation of homeostatic process regulation of muscle contraction response to bacterium cellular response to cytokine stimulus negative regulation of multicellular organismal process digestion adenylate cyclase-modulating G-protein coupled receptor signaling pathway cytosolic calcium ion homeostasis lymphocyte activation steroid metabolic process epidermis development immune effector process Paternally expressed imprinted genes G-protein coupled receptor signaling pathway, coupled to cyclic nucleotide second messenger Genes associated with RM regulation of leukocyte activation regulation of lymphocyte activation positive regulation of secretion regulation of neurological system process regulation of nucleotide metabolic process regulation of synaptic transmission cytokine-mediated signaling pathway gland development visual perception

Number ID Probes of Genes GO:0050678 124 124 GO:0007204 79 79 GO:0032844 140 140 GO:0006937 68 68 GO:0009617 185 185 GO:0071345 166 165 GO:0051241 166 166 GO:0007586 69 69

Raw Score p-value 0.02525684 8.07E-09 0.02804934 8.54E-09 0.02373767 1.20E-08 0.02957137 1.26E-08 0.02288847 1.31E-08 0.02346636 1.19E-08 0.02337247 1.62E-08 0.0279473 1.90E-08

Corrected p-value 1.28E-06 1.30E-06 1.69E-06 1.71E-06 1.72E-06 1.75E-06 2.06E-06 2.33E-06

GO:0007188 GO:0051480 GO:0046649 GO:0008202 GO:0008544 GO:0002252 PEG Genes

76 91 141 154 146 120 32

76 91 141 154 146 120 32

0.02764782 0.02724974 0.0234194 0.02302849 0.02326541 0.02450512 0.03547532

2.16E-08 2.30E-08 3.28E-08 4.91E-08 5.28E-08 6.72E-08 7.97E-08

2.57E-06 2.65E-06 3.67E-06 5.34E-06 5.58E-06 6.92E-06 7.98E-06

GO:0007187 RM Genes GO:0002694 GO:0051249 GO:0051047 GO:0031644 GO:0006140 GO:0050804 GO:0019221 GO:0048732 GO:0007601

100 60 200 176 133 132 184 114 131 163 116

100 60 200 176 133 132 184 114 130 162 116

0.02531593 0.02926887 0.02203887 0.02214319 0.02380548 0.02362232 0.02193497 0.02389627 0.0235637 0.02234582 0.02382121

9.68E-08 9.99E-08 1.11E-07 1.61E-07 1.83E-07 3.00E-07 3.14E-07 3.38E-07 3.51E-07 3.95E-07 4.10E-07

9.45E-06 9.51E-06 1.03E-05 1.46E-05 1.62E-05 2.60E-05 2.66E-05 2.80E-05 2.84E-05 3.13E-05 3.19E-05

149

Name epithelial cell differentiation actin filament-based process developmental maturation regulation of MAP kinase activity regulation of purine nucleotide metabolic process sensory perception of light stimulus response to oxidative stress protein activation cascade regulation of transmission of nerve impulse cell junction organization activation of immune response secretion by cell regulation of ion homeostasis endocytosis negative regulation of transport response to nutrient response to metal ion regulation of blood pressure humoral immune response regulation of membrane potential positive regulation of neurological system process cellular response to growth factor stimulus sensory perception of chemical stimulus regulation of MAPK cascade regulation of cytokine production sodium ion transport glycerolipid metabolic process negative regulation of kinase activity

Number ID Probes of Genes GO:0030855 152 152 GO:0030029 183 183 GO:0021700 78 78 GO:0043405 150 150 GO:1900542 182 182 GO:0050953 117 117 GO:0006979 144 144 GO:0072376 43 43 GO:0051969 121 121 GO:0034330 99 99 GO:0002253 159 159 GO:0032940 199 199 GO:2000021 71 71 GO:0006897 156 156 GO:0051051 146 146 GO:0007584 153 153 GO:0010038 151 151 GO:0008217 69 69 GO:0006959 60 60 GO:0042391 111 111 GO:0031646 33 33 GO:0071363 109 109 GO:0007606 49 49 GO:0043408 180 180 GO:0001817 196 196 GO:0006814 73 73 GO:0046486 116 116 GO:0033673 94 94

Raw Score p-value 0.02255615 4.30E-07 0.02178972 4.95E-07 0.02618592 5.08E-07 0.02248285 5.30E-07 0.02175086 5.58E-07 0.02369091 5.71E-07 0.02244905 5.83E-07 0.03154828 6.25E-07 0.02364677 6.39E-07 0.02444168 7.56E-07 0.02208538 8.41E-07 0.02140264 8.89E-07 0.02607518 9.31E-07 0.02201122 1.04E-06 0.02213919 1.38E-06 0.02190577 1.40E-06 0.0221242 1.43E-06 0.0257222 1.83E-06 0.02748179 2.29E-06 0.02342726 2.40E-06 0.03283363 2.38E-06 0.02340325 2.54E-06 0.02858786 2.67E-06 0.02116248 3.24E-06 0.02097018 3.35E-06 0.02529259 4.06E-06 0.02287256 4.18E-06 0.02366534 4.13E-06

Corrected p-value 3.28E-05 3.69E-05 3.72E-05 3.81E-05 3.94E-05 3.96E-05 3.96E-05 4.17E-05 4.19E-05 4.88E-05 5.34E-05 5.55E-05 5.72E-05 6.28E-05 8.19E-05 8.20E-05 8.27E-05 1.04E-04 1.28E-04 1.31E-04 1.31E-04 1.36E-04 1.41E-04 1.69E-04 1.72E-04 2.06E-04 2.07E-04 2.07E-04

150

Name morphogenesis of an epithelium response to growth factor stimulus muscle organ development regulation of peptidyl-tyrosine phosphorylation positive regulation of MAP kinase activity cell chemotaxis positive regulation of cytokine production positive regulation of protein kinase activity regulation of hormone levels positive regulation of protein serine/threonine kinase activity negative regulation of hydrolase activity cellular defense response growth response to corticosteroid stimulus positive regulation of cell activation extracellular matrix organization embryonic organ development organophosphate metabolic process respiratory system development negative regulation of immune system process response to lipopolysaccharide potassium ion transport cell junction assembly defense response to bacterium leukocyte chemotaxis regulation of T cell activation response to hypoxia

Number ID Probes of Genes GO:0002009 184 184 GO:0070848 120 120 GO:0007517 141 141 GO:0050730 83 83 GO:0043406 108 108 GO:0060326 58 58 GO:0001819 104 104 GO:0045860 194 194 GO:0010817 113 113

Raw Score p-value 0.02104682 4.50E-06 0.02280165 4.93E-06 0.02161454 5.52E-06 0.02490981 6.08E-06 0.02290668 7.59E-06 0.02671957 7.77E-06 0.0228866 7.92E-06 0.02066089 8.26E-06 0.02285105 8.55E-06

GO:0071902 GO:0051346 GO:0006968 GO:0040007 GO:0031960 GO:0050867 GO:0030198 GO:0048568 GO:0019637 GO:0060541 GO:0002683 GO:0032496 GO:0006813 GO:0034329 GO:0042742 GO:0030595 GO:0050863 GO:0001666

0.02226148 0.023245 0.03044358 0.02058592 0.02321601 0.02134234 0.02456241 0.02059873 0.02228751 0.0229996 0.02299012 0.02222971 0.02220614 0.02400723 0.02422679 0.02890583 0.02183027 0.02102689

128 96 36 189 96 152 79 181 121 97 94 115 120 89 76 41 134 148

128 96 36 189 96 152 79 181 121 97 94 115 120 89 76 41 134 148

8.98E-06 9.87E-06 1.03E-05 1.02E-05 1.05E-05 1.10E-05 1.15E-05 1.55E-05 1.57E-05 1.61E-05 1.64E-05 1.78E-05 1.88E-05 1.92E-05 2.07E-05 2.09E-05 2.39E-05 2.37E-05

Corrected p-value 2.20E-04 2.38E-04 2.63E-04 2.86E-04 3.52E-04 3.57E-04 3.59E-04 3.70E-04 3.78E-04 3.93E-04 4.27E-04 4.34E-04 4.38E-04 4.38E-04 4.54E-04 4.69E-04 6.28E-04 6.29E-04 6.40E-04 6.46E-04 6.92E-04 7.21E-04 7.30E-04 7.81E-04 7.82E-04 8.75E-04 8.77E-04

151

Name neuropeptide signaling pathway nucleotide biosynthetic process response to oxygen levels glycerophospholipid metabolic process regionalization regulation of ion transport lipid catabolic process cellular aromatic compound metabolic process response to molecule of bacterial origin negative regulation of epithelial cell proliferation

Number ID Probes of Genes GO:0007218 58 58 GO:0009165 109 109 GO:0070482 158 158 GO:0006650 73 73 GO:0003002 160 160 GO:0043269 120 120 GO:0016042 115 115 GO:0006725 135 135 GO:0002237 125 125 GO:0050680 50 50

Raw Score p-value 0.02595758 2.47E-05 0.02234566 2.45E-05 0.02080428 2.54E-05 0.02424943 2.52E-05 0.02079875 2.57E-05 0.02204689 2.64E-05 0.02204796 2.64E-05 0.02178681 2.63E-05 0.02202947 2.74E-05 0.02679098 2.98E-05

Corrected p-value 8.88E-04 8.89E-04 8.94E-04 8.96E-04 8.98E-04 8.99E-04 9.05E-04 9.11E-04 9.24E-04 9.96E-04

Italics = Custom groups; Bold = gene ontology classifications involved in immune response

152

Supplementary Figure 5.1 Correlation between Infinium average beta values and bisulfite pyrosequencing methylation (%) values for all samples run on the array (N=20) at candidate CpG sites selected for follow up: A) CYP1A2 cg04968473 (r=0.98, p<0.0001), B) DEFB1 cg24292612 (r=0.89, p<0.0001), C) APC cg20311501* (r=0.96, p<0.0001), and D) AXL cg14892768 (r=0.82, p<0.0001). *Note: the pyrosequencing assay assessed a nearby CpG, but not the exact same site as the Infinium probe. 153

Supplementary Figure 5.2 Correlation between gestational age and DNA methylation (%), as measured by bisulfite pyrosequencing, at each of the candidate regions identified from the Infinium analysis: A) CYP1A2 (r=0.58, p<0.0001), B) DEFB1 (r=-0.05, p=0.74), C) APC (r=0.03, p=0.83), D) AXL (r=0.03, p=0.83) in RM and M placental samples (N=54).

154

Supplementary Figure 5.3 Correlation between maternal age (years) and DNA methylation (%) at 12 targeted loci in RM and M placental samples (N=54): A) H19/IGF2 ICR1 (r=0.02, p=0.87), B) SNPRN (r=-0.06, p=0.64), C) PLAGL1 (r=-0.15, p=0.29), D) SGCE (r=0.08, p=0.55), E) CDKN1C (r=-0.25, p=0.06), F) KvDMR1 (r=-0.21, p=0.13), G) MEG3 (r=0.12, p=0.37), H) APC (r=-0.19, p=0.17), I) AXL (r=-0.13, p=0.33), J) CYP1A2 (r=-0.06, p=0.69), K) DEFB1 (r=0.24, p=0.08), and L) LINE1 (r=0.17, p=0.23).

155

Supplementary Figure 5.4 Comparison of average DNA methylation (%) at 7 imprinted loci between RM (N=33), M (N=21) and TA (N=16) groups: A) PLAGL1 (p=0.34), B) SGCE (p=0.006), C) H19/IGF2 ICR1 (p<0.0001), D) CDKN1C (p=0.77), E) KvDMR1 (p=0.14), F) MEG3 (p=0.93), G) SNRPN (p=0.07).

156

Supplementary Figure 5.5 Linear correlation between gestational age and DNA methylation at imprinted loci among RM and M placental samples (N=54): A) PLAGL1 (r=-0.00, p=0.98), B) SGCE (r=-0.01, p=0.96), C) H19/IGF2 ICR1 (0.04, p=0.75), D) CDKN1C (r=0.13, p=0.35), E) KvDMR1 (0.14, p=0.32), F) MEG3 (r=0.03, p=0.86), G) SNRPN (r=0.11, p=0.46).

157