GENETIC CONTRIBUTIONS TO DISPARITIES IN PRETERM BIRTH AMONG AFRICAN ...

The Pennsylvania State University The Graduate School Intercollege Graduate Degree Program in Genetics

GENETIC CONTRIBUTIONS TO DISPARITIES IN PRETERM BIRTH AMONG AFRICAN-AMERICAN WOMEN

A Dissertation in Genetics by Laurel N. Pearson

 2012 Laurel N. Pearson

Submitted in Partial Fulfillment of the Requirements for the Degree of

Doctor of Philosophy

August 2012

ii The dissertation of Laurel N. Pearson was reviewed and approved* by the following:

Mark D. Shriver Professor of Anthropology Thesis Advisor

David J. Vandenbergh Associate Professor of Biobehavioral Health Chair of Committee

Kenneth M. Weiss Evan Pugh Professor of Anthropology and Genetics and Science, Technology, and Society

Nina G. Jablonski Distinguished Professor of Anthropology

Jerome F. Strauss III Dean, School of Medicine Special Member Virginia Commonwealth University

Robert F. Paulson Professor of Veterinary and Biomedical Sciences Chair, Intercollege Graduate Degree Program in Genetics

*Signatures are on file in the Graduate School

ABSTRACT

iii

In the United States African-American women experience the poorest pregnancy outcomes of any ethnic group. Compared to women of European-American ancestry, African-American women have a substantially greater risk of preterm birth, low birth weight neonates, and infant mortality. A variety of factors have been hypothesized to contribute to disparities in these complex pregnancy phenotypes including environment, lifestyle, social factors, stress, and genetics.

This dissertation investigates genetic

ancestry and the role of genes in contributing to risk of poor pregnancy outcomes among African-American women. In the first portion of this research the association between West African genomic ancestry and birth weight and the association between genomic ancestry, skin pigmentation, and serum vitamin D level were investigated. Increasing West African ancestry among female neonates was found to be significantly associated with lower birth weight. Additionally, serum vitamin D level was inversely correlated with increasing West African genomic ancestry and increasing melanin content in the skin. For the next phase of this project an admixture mapping approach was used to help identify novel regions of the genome that are associated with the largest contributor to preterm birth, preterm premature rupture of membranes (PPROM). In this case-control analysis of African-American neonates, regions on five chromosomes were identified to be associated with increased risk of PPROM. Five regions on four chromosomes (5, 8, 11, and 19) were associated with African ancestry and one large region on chromosome 21 was associated with European ancestry. Although these regions are relatively large,

iv

they provide areas for future research into the genetic contributions to risk of preterm birth due to PPROM.

In the final portion of this research, tests for accelerated evolution were conducted to help prioritize candidate genes to investigate the role of genetics in the increased risk of preterm birth among African-American women. Using three test statistics: locusspecific branch length (LSBL), log of the ratio of heterozygosities (lnRH) and normalized Tajima’s D, 90 previously reported preterm birth candidate genes were screened for evidence of accelerated evolution in the parental populations that contribute to AfricanAmerican admixture, European and West African. From these tests, forty-four of the preterm birth candidate genes had evidence of non-neutral evolution. This analysis helped to identify genes that are more likely to contribute to the increased risk of preterm birth in African-American women compared to European-American women. Future work will include a replication of the admixture mapping study to refine the chromosomal regions found to be associated with risk of preterm birth due to PPROM.

Additionally, genotyping of the forty-four preterm birth candidate genes

nominated by the three tests for accelerated evolution is planned to look for risk alleles that contribute to the disparity in preterm birth among African-American women.

v TABLE OF CONTENTS List of Figures .......................................................................................................................... viii List of Tables ........................................................................................................................... ix Acknowledgements .................................................................................................................. x Chapter 1 Introduction ............................................................................................................. 1 Background ...................................................................................................................... 2 Preterm Delivery, PPROM, and Low Birth Weight ................................................. 4 Preterm Delivery and Low Birth Weight in Africa .................................................. 7 Risk Factors Associated with Preterm Birth ............................................................ 8 Genetic Contributions to Preterm Birth.................................................................... 11 Genomic Ancestry Estimation and Admixture Mapping ......................................... 13 Candidate Genes Previously Implicated in Preterm Birth........................................ 16 Screens for Accelerated Evolution ........................................................................... 17 Conclusion ....................................................................................................................... 18 Literature Cited ................................................................................................................ 19 Chapter 2 Genomic Ancestry and Pregnancy-Related Phenotypes ......................................... 24 Abstract ............................................................................................................................ 24 Introduction ...................................................................................................................... 25 Background .............................................................................................................. 26 Genomic Ancestry .............................................................................................. 26 Birth Weight and Genomic Ancestry .................................................................. 28 Skin Pigmentation, Vitamin D, and Genomic Ancestry...................................... 29 Materials and Methods ..................................................................................................... 32 Research Design ....................................................................................................... 32 Genomic Ancestry Estimates ................................................................................... 33 Statistical Analysis ................................................................................................... 35 Results .............................................................................................................................. 35 Genomic Ancestry .................................................................................................... 36 Birth Weight, Gestational Age, and Ancestry .......................................................... 40 Genomic Ancestry, Pigmentation, and Vitamin D ................................................... 43 Discussion ........................................................................................................................ 46 Future Directions ...................................................................................................... 48 Literature Cited ................................................................................................................ 50 Chapter 3 Admixture Mapping to Detect Novel Candidate Regions for Preterm Premature Rupture of Membranes (PPROM) in African-American Women ................................... 54 Abstract ............................................................................................................................ 54 Introduction ...................................................................................................................... 55 Background .............................................................................................................. 56 Previous Studies of the Genetics of PPROM...................................................... 57

Admixture Mapping for the Discovery of Novel Candidate Regions for PPROM .................................................................................................................... 58 Previous Admixture Mapping Studies in Preterm Birth ..................................... 63 Materials and Methods ..................................................................................................... 64 Study Sample ........................................................................................................... 64 Genotyping ............................................................................................................... 65 Data Cleaning ........................................................................................................... 67 Samples Removed from Analysis ........................................................................ 67 Markers Removed from Analysis........................................................................ 67 Departures from Hardy-Weinberg Equilibrium ................................................. 68 Admixture Mapping ................................................................................................. 69 Estimating Genomic Ancestry ................................................................................. 70 Results .............................................................................................................................. 71 Genomic Ancestry Estimates ................................................................................... 72 Admixture Mapping ................................................................................................. 75 Discussion ........................................................................................................................ 80 Future Directions ...................................................................................................... 83 Literature Cited ................................................................................................................ 85 Chapter 4 Investigating Evidence for Accelerated Evolution at Preterm Birth Candidate Genes................................................................................................................................ 89 Abstract ............................................................................................................................ 89 Introduction ...................................................................................................................... 90 Background .............................................................................................................. 91 Materials and Methods ..................................................................................................... 93 Admixture Mapping ................................................................................................. 93 Candidate Genes....................................................................................................... 96 Tests of Accelerated Evolution ................................................................................ 97 Sample Populations and Genotyping Panel ....................................................... 97 Locus-Specific Branch Length (LSBL) ............................................................... 97 Log of the Ratio of Heterozygosities (lnRH) ...................................................... 101 Normalized Tajima’s D ...................................................................................... 102 Evaluating the Significance of the Tests of Accelerated Evolution ......................... 104 Results .............................................................................................................................. 106 Screens for Accelerated Evolution from Admixture Mapping Results for PPROM ............................................................................................................ 106 Tests of Accelerated Evolution in Preterm Birth Candidate Genes ......................... 116 Discussion ........................................................................................................................ 140 Future Directions ...................................................................................................... 145 Literature Cited ................................................................................................................ 146 Chapter 5 Concluding Remarks ............................................................................................... 149 Relationship between Genomic Ancestry and Pregnancy-related Phenotypes ................ 149 Admixture Mapping and Genomic Regions Associated with Risk of PPROM ............... 151 Prioritizing Replication and Genotyping of Candidate Genes for Future Studies ........... 154 Future Directions.............................................................................................................. 155 Genotyping of Additional Mothers and Newborns for Ancestry Association Studies .............................................................................................................. 155 Replication of Admixture Mapping and Investigation of Candidate Genes ............ 156

vi

vii

Conclusion ...................................................................................................................... 157 Literature Cited ................................................................................................................ 158 Appendix A Details on the University of Minnesota 107 Ancestry Informative Marker (AIMs) Panel .................................................................................................................... 160 Appendix B Details on the Illumina African-American Admixture Panel 1,509 Ancestry Informative Marker (AIMs) ............................................................................................. 164 Appendix C Admixture Mapping Results from ADMIXMAP for PPROM.......................... 203 Case-Control Ancestry Association and Allelic Association Values at each AIM .......... 204 Case-Only Ancestry Association Score Maps ................................................................. 241 Case-Only Ancestry Association and Allelic Association Values at each AIM .............. 248 Appendix D Screens for Accelerated Evolution for PPROM Admixture Mapping Results .............................................................................................................................. 281 Full Chromosomal Ancestry Association Peak Screens – West African ......................... 282 Full Chromosomal Ancestry Association Peak Screens – European ............................... 283 Individual AIMs within Chromosomal Ancestry Association Peak Screens – West African ..................................................................................................................... 284 Individual AIMs within Chromosomal Ancestry Association Peak Screens – European ................................................................................................................. 285 Appendix E Screens for Accelerated Evolution in 90 Previously Reported Preterm Birth Candidate Genes .............................................................................................................. 286 West African .................................................................................................................... 287 European .......................................................................................................................... 290

LIST OF FIGURES

viii

Figure 1-1. Infant Mortality in the United States by Ethnicity. ............................................... ....3 Figure 1-2. Pregnancy Outcomes by Ethnicity ........................................................................ …4 Figure 2-1. Distribution of Genomic Ancestry in Research Subjects ...................................... ..38 Figure 2-2. Birth Weight Plotted Against Gestational Age ..................................................... ..41 Figure 2-3. Birth Weight Plotted Against West African Genomic Ancestry........................... ..42 Figure 2-4. Proportion of West African Ancestry, Skin Pigmentation, and Vitamin D .......... ..45 Figure 3-1. Linkage Disequilibrium Created by Admixture .................................................... ..60 Figure 3-2. Illustration of Admixture Mapping Ancestry Association Peak ........................... ..62 Figure 3-3. Illumina African-American Admixture Panel Minor Allele Freqeuncy Differences ....................................................................................................................... ..66 Figure 3-4. Genomic Ancestry Distribution of Study Samples for Admixture Mapping Analysis ........................................................................................................................... ..74 Figure 3-5. Genome-wide Ancestry Association Z-scores Plotted for African Ancestry ....... ..77 Figure 3-6. Ancestry Association Score Maps for Chromosomes with Significant Peaks in PPROM Admixture Mapping .......................................................................................... ..78 Figure 4-1. Illustration of Locus-Specific Branch Length (LSBL).......................................... 100 Figure 4-2. Results of the Tests for Accelerated Evolution in PPROM Admixture Mapping Ancestry-Association Regions.......................................................................... 109 Figure 4-3. Results of the Tests for Accelerated Evolution in Previously Published Preterm Birth Candidate Genes........................................................................................ 118

LIST OF TABLES Table 2-1. Average Allele Frequency Difference Between Parental Populations for 107 Ancestry Informative Marker Panel ................................................................................. ..34 Table 2-2. Maternal Characteristics ......................................................................................... ..39 Table 3-1. Average Delta Calculated between Parental Populations for Illumina AfricanAmerican Admixture Panel. ............................................................................................. ..72 Table 3-2. Average Genomic Ancestry for PPROM (cases) and Controls .............................. ..73 Table 3-3. Ancestry Association Peaks for Admixture Mapping of PPROM ......................... ..79 Table 4-1. Description of Chromosomal Ancestry Association Peaks Identified by Admixture Mapping in PPROM ..................................................................................... ..95 Table 4-2.Empirical Distribution Cut-offs for Tests of Accelerated Evolution in Ancestry Informative Marker Panel used in PPROM Admixture Mapping Study ......................... 105 Table 4-3.Summary of Tests of Accelerated Evolution for PPROM Admixture Mapping Peaks ................................................................................................................................ 114 Table 4-4.Summary of Tests of Accelerated Evolution for Ancestry Informative Markers within PPROM Admixture Mapping Peaks ..................................................................... 114 Table 4-5.Summary of Tests of Accelerated Evolution for Preterm Birth Candidate Genes .. 137 Table 4-6.Significant Results for Screens of Accelerted Evolution in Preterm Birth Candidate Genes Summarized ......................................................................................... 139

ix

ACKNOWLEDGEMENTS

x

There are many people that I would like to thank for making this dissertation possible. First and foremost, I am grateful for the mentorship of my advisor, Mark Shriver. I would also like to thank the members of my dissertation committee – David Vandenbergh, Ken Weiss, Nina Jablonski, and Jerome Strauss – who provided many thoughtful suggestions on improving my research and dissertation. My dissertation could not have been completed without the support of collaborators including Drs. Jerome Strauss, Roberto Romero, and Juan Pedro Kusanovic who generously provided the PPROM data, and Dr. Hyagriv Simhan, Dr. Lisa Bodner, and David Crowe at Magee Womens Hospital. No one at Penn State has been a bigger cheerleader for me than Dr. Rick Ordway, the former chair of the genetics program. His faith in me and encouragement kept me going even when times were tough. The anthropology department at Penn State has been a wonderful home for an interdisciplinary program graduate student. Thank you for treating me like one of your own. I am grateful to have been a member of the Shriver Lab family with Abby Bigham, Ellen Quillen, Denise Liberton, Jen Wagner, Kerri Matthes Rosana, Xianyun Mao, Marc Bauchet, Arslan Zaidi, and Wei Yao. Thank you for all of your advice, support, and fun distractions over the years.

xi

The friends that I made at Penn State and in State College – Abby Bigham, Jason De Leon, Ellen Quillen, Geoff Vasile, Holly Dunsworth, Kevin Stacey, Sam and Jen Sholtis, Kirk Straight, Maria Inclan, Dawn Miller, Sharon DeWitte, Eric Jones, Logan Kistler, Anna Sewell, Chris Percival, Carolyn Keagel, Erick and Sarah Rochette, Ryan Peterson, and the ladies from book club and knitting group – helped me keep my sanity and made graduate school much more fun. I would like to thank my parents, Ken and Dinah Pearson, and my sister, Erin, for their love and support. Finally, none of this would have been possible without the love, encouragement, and patience of my fiancé, Kirk French. Thank you for all of the great adventures. I look forward to many more in the years to come.

1 Chapter 1 Introduction

African-American women experience the poorest pregnancy outcomes of any U.S. ethnic group. Disparities in pregnancy outcomes for these women include much higher rates of preterm birth, low birth weight neonates for gestational age, and increased risk of infant mortality compared to women of self-described European-American ancestry. Many possible explanations have been suggested for these apparent disparities in pregnancy phenotypes among African-American women, including environment, lifestyle, access to healthcare, stress, and genetics. The current research will investigate the genetic and gene-environment factors that may influence pregnancy outcomes from the perspective of genomic ancestry. It will also exploit recent admixture in African Americans to help identify novel genomic regions contributing to risk for preterm birth due to preterm premature rupture of membranes (PPROM) by admixture mapping. Finally, tests for accelerated evolution in West African and European parental populations will be used to prioritize previously reported preterm birth candidate genes and ancestry-associated risk-regions for PPROM for future studies of genetic contributions to disparity in preterm birth risk in African Americans. This research addresses four main questions: 1) Is West African genomic ancestry associated with low birth weight for gestational age among African-American neonates 2) Is West African genomic ancestry associated with serum vitamin D level, a possible

contributor to preterm birth?

2

3) Can admixture mapping be used to identify novel

candidate regions of the genome for risk of premature birth due to PPROM in AfricanAmerican women? 4) Can screens for accelerated evolution help prioritize previously reported candidate genes for preterm birth and genomic regions identified as risk regions for PPROM by admixture mapping for future investigation of disparity in preterm birth risk among African-American women?

Background

Preterm birth (PTB) is the leading cause of neonatal mortality and infant morbidity in the United States [1]. The annual economic burden of preterm birth was estimated at $26.2 billion in 2005 by the Institute of Medicine [2]. This figure does not include the lifelong healthcare costs attributable to chronic illness associated with preterm delivery [3]. The rate of PTB in the U.S. is higher than most other developed nations and appears to be rising [4-6]. The rise in PTB is partially attributable to the increased use of assisted reproductive technologies (ART) and the increased number of multiple gestations [7]. However, after correcting for these factors, PTB rate is still increasing. Causes of PTB are numerous with 80% considered spontaneous due to either preterm labor (PTL) or preterm premature rupture of membranes (PPROM) [5]. The remaining 20% of preterm births are medically induced due to factors such as fetal malformation, multiple gestations, or maternal health factors, like pregnancy-induced hypertension (preeclampsia).

3

Within the framework of PTB there is considerable disparity between AfricanAmerican and European-American women in terms of risk. African-American women exhibit a significantly higher rate of PTB compared to European-American women, with 18.5% of neonates born to African-American women delivered preterm compared to 11.6% for European-American women [8, 9]. Additionally, African-American women have the highest rate of low birth weight and very low birth weight infants of any racial/ethnic group and the highest rate of infant mortality in the United States (Figures 11 and 1-2) [10]. Many factors have been implicated to explain these differences, including behavior, environment, social factors, risk of infection, and genetics. This research is focused on investigating the contribution of genetic ancestry to disparity in pregnancy outcomes and exploiting the relatively recent admixture in African-Americans to help identify novel candidate genes for preterm delivery due to PPROM.

Figure 1-1. Infant mortality per 1,000 live births by ethnicity. Figure from the 2004 National Vital Statistics Reports.

4

Figure 1-2. Pregnancy outcomes by ethnicity: preterm birth and low birth weight. Adapted from Martin et al. (2009) [9].

Preterm Delivery, PPROM, and Low Birth Weight

In humans, normal term gestation for singleton pregnancies is 37 to 41 weeks [10]. Changes in maternal and fetal physiology near the end of pregnancy are thought to promote the normal progression of parturition at term, through both hormonal and mechanical processes. These changes promote the formerly quiescent uterus to start contracting and the breakdown of the fetal membranes that together facilitate

5

spontaneous birth. There are a variety of factors thought to contribute to early changes in these pathways that can result in spontaneous preterm birth [11].

Preterm delivery (PTB) is defined as birth prior to 37 weeks gestation, with gestational age determined from first day of the last menstrual period or early second trimester ultrasound estimation, the “gold standard” for confirming gestational age [12]. Preterm delivery falls into two categories, spontaneous and medically induced. Spontaneous PTB is caused by preterm labor or preterm premature rupture of membranes (PPROM). In the United States, spontaneous preterm birth occurs in approximately 12.8% of all deliveries. Among African-American women, the frequency is 18.7%, much greater than the 11.7% incidence of PTB among European-American women [9]. Additionally, African-American women are at disproportionately higher risk of very preterm delivery ( 0.5 and an average distance between markers less than 3 centimorgans (cM) [28]. Genotyping arrays like the Illumina (San Diego, California) African-American Admixture Panel have been specifically designed to meet these standards.

With AIMs spaced evenly across the genome it is assumed that genes functionally affecting disease with risk will be in linkage disequilibrium with some of the ancestry markers that are typed. In the case of a dichotomous disease phenotype, like PPROM, it is expected that those affected by the disease will show a different pattern of ancestry in the area surrounding risk regions of the genome compared to controls.

Figure 3-2

illustrates the expected association of ancestry with a risk region in cases compared to controls [29]. In a phenotype like PPROM where there is increased risk in AfricanAmericans compared to European-Americans, this peak representing an increased risk of disease in cases might be expected to be associated with increased West African ancestry.

62

Figure 3-2. This drawing illustrates the potential usefulness of admixture mapping with a large panel of ancestry informative markers (AIMs) to identify chromosomal regions that are associated with a disease phenotype of interest. Surrounding a disease-associated locus there will be an excess of ancestry from one of the parental populations included in the model. Figure from Patterson et al. (2004) [29].

The potential success of admixture mapping is dependent upon the time since admixture. If the population has only been admixed for a few generations, LD blocks are expected to be large, resulting in large regions of association between ancestry and the phenotype of interest. In this situation, isolating a potentially causative gene will be much more difficult to achieve due to the large size of the region that would require finescale mapping. In the inverse scenario, if many generations have passed since admixture began, linkage disequilibrium blocks will be small and many more markers will be required.

African-American and Hispanic populations are well suited to admixture

mapping studies because they have undergone a modest number of generations of admixture. For example, estimates from large genotyping arrays suggest an average

63

number of six generations, approximately 150 years, of admixture among AfricanAmericans [29, 30]. Bonilla et al. (2004) report slightly greater number estimate of seven generations (~ 175 years) for Hispanics in the San Luis Valley of Colorado [31].

Using a genome-wide panel of AIMs has similar power as much larger panels of tag SNPs used in genome-wide association studies (GWAS), but at a substantially reduced cost [32]. The Illumina African-American Admixture panel that was genotyped in this study has been used by other researchers to help identify novel candidate genes associated with asthma [29, 33], cardiovascular disease [34-36], multiple sclerosis [37], prostate cancer [38-40], obesity [41, 42], and recently premature birth [43].

Previous Admixture Mapping Studies in Preterm Birth

To date, only one research report has been published using an admixture mapping approach to investigate preterm birth. Last year, Manuck et al. (2011) reported a caseonly study of 177 African-American women with one or more spontaneous preterm births. Using the same genotyping platform as the present research, the Illumina AfricanAmerican Admixture Panel, and the ANCESTRYMAP analysis package, the authors identified a region on chromosome 7 (7q21-7q22) associated with risk of preterm birth. This peak is associated with a chromosomal region of increased West African ancestry. The number of significant SNPs in this chromosomal region found to be associated with the phenotype increased when the severity of prematurity was considered. Compared to the three SNPs identified with suggestive Logarithm of Odds (LOD) scores using the full

64

sample of 177 women with gestation < 37 weeks, using only the 106 women with previous very premature births (< 32 weeks gestation), fifteen significant SNPs were reported that reached a suggestive LOD score of 1.50 or greater.

The number of

significant polymorphisms is reduced to six for gestation < 32 weeks when only those that reach a significance threshold greater than a LOD score of three are considered [43]. The current research looks to improve upon the Manuck et al. study by including a larger sample and a more specific preterm birth phenotype, preterm premature rupture of membranes (PPROM).

Materials and Methods

Study Sample

Study subjects included in this research are fetuses/neonates born from singleton pregnancies to women of self-reported African-American ancestry. Pregnant women were recruited to participate in the study at the time of admission to the hospital for delivery from three study sites: Virginia Commonwealth University Health System (Richmond, Virginia), Hutzel Hospital (Detroit, Michigan), and the University of Pennsylvania Health System (Philadelphia, Pennsylvania) by Dr. Jerome Strauss, Dr. Roberto Romero or Dr. Juan Pedro Kusanovic. Case subjects are those deliveries complicated by preterm premature rupture of membranes (PPROM) prior to 37 complete weeks of gestation without evidence of major fetal malformation, genetic diseases known

65

to contribute to prematurity, trauma, maternal drug use during pregnancy, or preeclampsia. Criteria used to confirm the case state (PPROM) included pooling of

amniotic fluid in the vagina, a characteristic ferning pattern of the amniotic fluid, and a positive nitrazine test [4]. The control subjects in this analysis consist of neonates from singleton pregnancies born at term, after 37 weeks of gestation, without known complications. Gestational age was determined by the last menstrual period (LMP) method and confirmed by ultrasound. DNA was extracted from cord blood collected at the time of delivery and whole genome amplified to accommodate genetic analyses.

Genotyping

All samples were genotyped on the Illumina (San Diego, California) AfricanAmerican Admixture Panel at the Hershey Genome Sciences Core Facility at the Pennsylvania State University Medical Center. This panel uses GoldenGate® Assay technology to genotype 1,509 single nucleotide polymorphisms (SNPs) that show large allele frequency differences between West African and European populations, the two parental groups with the largest genetic contribution to the modern African-American population [44]. These SNPs are referred to as ancestry informative markers (AIMs). This platform is a valuable tool for admixture mapping because it covers the entire genome in just over 1,500 markers, unlike GWAS panels that require hundreds of thousands of markers. It is estimated that this targeted panel contains 75-80% of the power to detect associations as found with in much larger, and substantially more expensive, genome-wide panels of polymorphic markers that contain 300,000 to one

million markers [28, 44].

66

Details on the AIMs included in the Illumina African-

American Admixture Panel are in Appendix B.

Figure 3-3. Illumina African-American Admixture Panel. Minor allele frequency (MAF) differences between West African v. European (orange) and West African v. Asian (blue). Modified from Illumina SNP genotyping datasheet for the African-American Admixture Panel [44].

Data Cleaning

67

Samples Removed from Analysis

Prior to analysis, the genotype data was cleaned to remove subjects that did not fit the criteria for either case or control. Subjects who were excluded as cases included those that were classified as preterm but not PPROM, medically induced preterm births, and pregnancies with multiple gestations. Additionally, duplicate subjects were removed from the analysis, and those with less than a 70% genotyping call rate were also excluded due to presumed poor DNA quality. Finally, neonates from women who did not selfreport African-American ancestry or with very low West African genomic ancestry (< 20%) were removed prior to admixture mapping analysis. Of the 642 research subjects that were genotyped, 616 met the criteria for inclusion in the study.

Markers Removed from Analysis

The data were also cleaned to remove genotyping markers that did not meet a minimum threshold of 70% of genotypes called. Additionally, all genotyping markers on the X chromosome (n=54) were excluded from these analyses due to lack of complete data on the sex of the neonates to include in the model and the known deficiency of ADMIXMAP to accommodate analysis of associations on the X chromosome [40]. Finally, some SNPs were excluded due to their departure from Hardy-Weinberg

68

equilibrium, as described in the next section. In total, 1,398 SNP genotyping markers were included in the analysis.

Departures from Hardy-Weinberg Equilibrium

An initial analysis of the data included screening for departures from HardyWeinberg equilibrium (HWE).

Preliminarily, both case and control samples were

evaluated using both ADMIXMAP and PLINK [45, 46]. ADMIXMAP utilizes the entire sample without regard to case or control status to test HWE. It is important to consider the cases and controls separately because alleles with significant risk effects may exhibit departures from HWE in a combined sample of cases and controls. Removing SNPs from further analysis due to a departure from HWE seen in the sample as a whole might risk the loss of a highly significant disease-associated marker prior to testing for association. Unlike ADMIXMAP, PLINK tests for deviations from Hardy-Weinberg equilibrium in cases-only, controls-only and for the entire sample. A departure from HWE in cases only was not considered sufficient for exclusion of a genotyping marker from inclusion in the admixture mapping analysis, as explained above. All markers identified as departing significantly from HWE (p < 0.01) in the control-only PLINK output were removed prior to admixture mapping analysis. The tests for departure from Hardy-Weinberg equilibrium resulted in the exclusion of 93 SNPs from the admixture mapping study.

69

Admixture Mapping

Admixture mapping analysis was performed with the Bayesian statistical program ADMIXMAP

v.3.8

for

Windows

(available

for

download

at

http://www.homepages.ed.ac.uk/pmckeigu/admixmap/index.html) [28, 47]. This program tests the association between a phenotype (continuous or dichotomous) and locus ancestry by conditioning on parental admixture. A three-way model of ancestry was specified with prior allele frequencies included in the analysis from three parental populations, West African, European, and East Asian. The phenotype was specified as a dichotomous variable where a PPROM case subject was given a value of 1 and a control subject a value of 0. The program also permits the inclusion of covariates; however, none were included in this analysis due to lack of information for all of the subjects included in the study. The program was run with 500 burn-in and 10,000 iterations, with samples thinned to record to the output files every 5 iterations. The adequacy of the burn-in and number of iterations was established by investigating the diagnostic plots generated by the ADMIXMAP program. A smoothing of the line over the course of the run indicates that sufficient iterations have been used in the analysis. Both a case-control model and an affected-only model were run. Although the affected-only model is considered more powerful for rare diseases, the frequency of preterm birth is substantial (12.5% of births in the United States) and therefore output from only the case-control analysis is included in the results section of this chapter.

The output of both affected-only and the case-

control analysis are included in the appendix (Appendix C).

70

Ancestry associations are reported as z-scores with their associated p-values for each locus included in the model. The z-score is a test statistic for association of the phenotype (PPROM in this case) with ancestry at each locus. In viewing the overall plot of ancestry association presented in the results, a positive z-score represents an association with West African ancestry, while a negative score is an association with nonWest African ancestry, either European or East Asian. Individual plots of the ancestry associations by parental population are available in the appendix (Appendix C).

Estimating Genomic Ancestry

Genomic ancestry for all research subjects was estimated using a three-way model of admixture. Prior allele frequencies derived from three parental populations (West African, European, and East Asian) were included in the ADMIXMAP analysis to improve the model as well as to estimate individual genetic ancestry. The Hershey Genome Sciences Core Facility at Penn State Medical Center provided the populationspecific prior allele frequencies that are included for analysis by Illumina [44]. However, the allele frequencies for the West African and European parental populations were originally described in Smith et al. (2004) [30]. East Asian parental frequencies are derived from the International HapMap Project [48]. The method employed in the ADMIXMAP software package has been validated as achieving equivalent ancestry estimates to those calculated using the maximum likelihood estimation (MLE) method and with the software program Structure, especially when a large panel of ancestry informative markers and a large sample size are included in the model [28, 47, 49-52].

Results

71

From the initial 642 DNA samples that were genotyped, a total of 616 samples with 1,398 genotypes met all of the inclusion criteria for analysis in this study. This included 352 case samples with preterm premature rupture of membranes (PPROM) and 264 control samples.

Additionally, 111 of the 1,509 ancestry informative markers

(AIMs) that were genotyped on the African-American Admixture panel were excluded from the study for reasons detailed in the methods. The average difference in parental allele frequency for each pairwise comparison (West African to European, West African to East Asian, and European to East Asian), delta (δ), was not affected by the removal of these markers as seen in Table 3-1. However, the average distance between genotyping markers, measured in centimorgans (cM), increased from an average of 1.95 ∆cM for the full panel of 1,509 AIMs to 2.75 ∆cM for the reduced panel of 1,398 markers. Figure 3-5 contains information on the average distance between genotyping markers calculated for each chromosome.

72

Table 3-1. Average delta (δ) calculated between each parental population included in the full Illumina African-American Admixture Panel compared to the reduced panel used in the admixture mapping analysis for this study.

Delta (δ) Number of SNPs

West African European

West African East Asian

European East Asian

Full Ancestry Panel

1,509

0.738

0.568

0.193

Analysis Panel

1,398

0.736

0.565

0.194

Genomic Ancestry Estimates

Genomic ancestry was calculated as a parameter of the ADMIXMAP analysis from user specified prior allele frequencies. The estimates are more robust with larger sample sizes and increased numbers of ancestry informative markers (AIMs).

On

average, cases and controls have similar levels of European admixture, 17.3% and 17.0% respectively, as represented in Table 3-2. The distribution of West African ancestry ranges are 42%-97% in cases and 48%-98% in controls (Figure 3-4). The mean genomic ancestry estimate for the study sample is 80.8% West African, 17.2% European and 2.1% East Asian. As expected, the East Asian average was very low for the sample as a whole. The small East Asian component may more likely reflect a low level of Native American admixture that was not assessed due to lack of available prior allele frequencies for all of the loci included in the analyses. Additionally, the sum of intensities parameter gives an

73

estimated average time since admixture (τ) of 6.7 (with a 2.5% bound of 6.59 and 97.5% bound of 6.83) generations. At a generation time of 25 years, this is approximately 168 years.

Table 3-2. Average genetic ancestry calculated for cases (PPROM) and controls in a 3way admixture model (West African, European, and East Asian) using ADMIXMAP. Standard deviations included in parentheses. Average Ancestry Sample

West African

European

East Asian

PPROM

352

80.4 (±8.0)

17.3 (±7.2)

2.3 (±3.1)

Control

264

81.3 (±7.8)

17.0 (±6.7)

1.7 (±3.0)

Total

616

80.8 (±7.9)

17.2 (±7.0)

2.1 (±3.1)

74

Figure 3-4. Genomic ancestry of the study sample. Triangle plots of three-way admixture for cases, controls and the full sample are shown in the top of the figure. Each circle represents the admixture proportions of a single subject. The histogram compares the distribution of West African genomic ancestry between case and control subjects. In both portions of the figure, controls are labeled in orange and cases in blue.

75 Admixture Mapping

Genome-wide ancestry association z-scores for the case-control analysis in ADMIXMAP are plotted by chromosome and shown in Figure 3-5. Chromosomes with ancestry association peaks above |z| = 2.5 (p < 0.01) are shown in Figure 3-6. These peaks occur on chromosomes 5, 8, 11, 19, and 21. Each of these chromosomes has one ancestry association peak, with the exception of chromosome 11 with a peak on either end of the chromosome. The genetic ancestry association peaks on chromosomes 5, 8, 11, and 19 are in the West African direction; however the peak on chromosome 21 is associated with European ancestry. Ancestry association z-scores and their associated pvalues are shown in Table 3-3.

The SNP with the largest ancestry association is

rs2833775, located on chromosome 21. This SNP shows an increased European ancestry with a z-score of -3.636 (p= 0.00028). It is located among a cluster of eight other SNPs (rs9977512, rs380417, rs2070398, rs2832643, rs2284473, rs2834670, and rs718387) that fall within a large peak on chromosome 21.

The strongest West African ancestry

association is found on chromosome 19, rs10405317 (z = 3.263, p = 0.0011). This SNP along with one other (rs2285972, z = 2.767, p = 0.0057) form a small peak near the telomeric end of the short (p) arm of chromosome 19. In total, there are twenty-nine SNPs with ancestry association z-scores greater than 2.5 (|z| > 2.5, p-value < 0.01) under six peak on five chromosomes. In addition to the plots shown in this chapter, ancestry association score maps displaying the ancestry association z-scores plotted by location along each chromosome for West African and European ancestry are shown in Appendix C for both the case-

76 control and affected-only admixture mapping analyses. Tables containing complete lists of the ancestry association and allelic association values (z-score and p-value) by genotyping marker for the case-control and affected-only analyses are also included in the appendix (Appendix C).

77

(a)

Distance (cM) (b)

Figure 3-5. a) Genome-wide ancestry associations from case-control admixture mapping analysis. Z-scores show associations with West African ancestry. b) Number of SNPs per chromosome and the average physical distance in centimorgans (cM) between SNPs for each chromosome included in the admixture mapping analysis.

78

Figure 3-6. Ancestry association score maps for chromosomes with significant peaks (Chromosomes 5, 8, 11, 19, and 21). Ancestry association z-scores for each locus are plotted along the chromosome in centimorgans (cM). a) Ancestry association maps for West African ancestry. b) Ancestry association map for European ancestry.

79

Table 3-3. Ancestry association peaks for case-control admixture mapping. The z-scores and p-values for each SNP within the ancestry association peaks are shown. Changes in the shading are used to distinguish the SNPs within each chromosomal peak of ancestry association. Locus Physical Chromosome Position Name rs40030 5 36212631 rs930072 5 36701828 rs4869577 5 38254937 rs12657366 5 39207788 rs6883840 5 40322167 rs16932440 8 67113059 rs7933164 11 4050526 rs10768634 11 5149986 rs10765838 11 11326144 rs2403595 11 12341885 rs6486270 11 15927515 rs4148636 11 17383939 rs11024739 11 18602419 rs647756 11 107040968 rs566552 11 108726089 rs4622301 11 110301295 rs6589360 11 112555502 rs7934726 11 113415486 rs2507874 11 114433145 rs2285972 19 1309726 rs10405317 19 1642725 rs9977512 21 24556659 rs380417 21 26194030 rs2070398 21 29800235 rs2832643 21 30450088 rs2284473 21 31450991 rs2833775 21 32634146 rs2834670 21 35202246 rs718387 21 36049947

Ancestry Association Ancestral z -score p -value Population 2.635 0.0084077 African 2.793 0.00522288 African 2.738 0.00618999 African 2.909 0.00363159 African 2.881 0.00396513 African 2.941 0.00327326 African 2.591 0.00957268 African 2.639 0.00831996 African 2.751 0.00594973 African 2.708 0.00677201 African 2.994 0.00275669 African 2.956 0.00311595 African 2.857 0.00427769 African 2.471 0.0134832 African 2.808 0.00498026 African 2.864 0.00417852 African 2.933 0.00335893 African 2.996 0.00273367 African 2.488 0.0128593 African 2.767 0.00565273 African 3.263 0.00110079 African -2.737 0.00619357 European -2.8 0.00511056 European -3.603 0.00031422 European -3.602 0.00031533 European -3.542 0.00039638 European -3.636 0.00027695 European -2.938 0.00330705 European -2.581 0.00984816 European

Discussion

80

Average genomic ancestry of the study subjects is similar to the average reported by Parra et al. (1998) and Smith et al. (2004) for U.S. African-Americans, approximately 80% West African and 20% European. Additionally, the time since admixture reported from the sum of intensities for this study was τ = 6.7, approximately 168 years. While this value may seem low given the known historical introduction of enslaved West Africans to the United State over 400 years ago (< 20 generations), it is within the ranges previously reported, and similar to those found by Patterson et al. (2004) and Smith et al. (2004) using similar high density SNP panels, τ = 6.0 and τ = 6.3, respectively. Power calculations reported by Hoggart et al. (2004) suggest that the cutoff for statistical significance for ADMIXMAP results is a z-score of 4.27 with a corresponding p-value of 10-5. They determined that for an African-American population with average admixture proportions approximating those found in this study (80% West African and 20% European) a sample of 800 subjects would be needed [28]. Not surprisingly, given the much smaller sample available for this analysis, none of the ancestry associations reported in this study reach this level of significance. The most significant result is the large peak on chromosome 21 that is associated with excess European ancestry. This peak is composed of eight SNPs that span a nearly 11.5 Mb region on chromosome 21q21-21q22. The center of the peak is composed of four SNPs (rs2070398, rs2832643, rs2284473, and rs2833775) with z-scores near 3.6 (p-values < 0.003). This large peak of European ancestry suggests a region where European ancestry is contributing to risk of PPROM in the African-American research subjects.

81

There are a number of genes within the large ancestry association peak on chromosome 21. While none of these genes had been identified as a candidate gene for preterm birth or PPROM, two related genes in the ADAMTS (ADAM metallopeptidase with thrombospondin) family, ADAMTS1 (21q21.2) and ADAMTS5 (21q21.3), were notable.

This family of genes has been previously reported to be associated with

parturition and inflammation. Specifically, increased expression of ADAMTS1 has been shown in mice to be involved in cervical ripening around the time of birth [53]. Given their known involvement in inflammation and parturition, these genes may warrant further investigation for their potential role in preterm birth. Another potentially interesting result comes from the small ancestry association peak on chromosome 8. This peak shows an increase in West African ancestry and includes only one SNP, rs16932440. While the z-score associated with this SNP (z = 2.941, p = 0.0033) does not meet the significance cutoff of z > 4.27, it is located near the corticotropin releasing hormone gene (CRH) at 8q13. With the expected long stretches of linkage disequilibrium found in admixed populations, the proximity of rs16932440 to CRH is worth investigating. Corticotropin (CRH) is produced by the placenta and fetal membranes during pregnancy [54, 55]. The level of CRH increases over the course of pregnancy, with high levels detected in the maternal blood and amniotic fluid near the time of birth [56, 57]. CRH is thought to be important in the onset of uterine contractions and has also been implicated in the rupture of fetal membranes through the activation of the matrix metalloproteinase MMP-9 [57, 58]. Additionally, women who experience preterm birth have higher levels of CRH than those that deliver at or after term [59].

82

Although CRH is not listed as a preterm birth candidate gene, corticotropin releasing hormone binding protein (CRHBP) was found to be strongly associated with preterm birth risk by Velez et al. (2008) [60]. With the known role of corticotropin in birth timing, the ancestry association near the CRH gene is intriguing and will be investigated further in the planned replication of this study. In addition to investigating the results of the admixture mapping ancestry associations with known candidate genes, the results of the current study were compared to the findings from Manuck et al. (2011) for evidence of overlap in significant ancestry associations [43].

While Manuck et al. found an ancestry association peak on

chromosome 7, this was not replicated in the current admixture mapping analysis. Additionally, none of the fifteen SNPs identified by Manuck et al. showed evidence of significantly increased West African ancestry in the current study. There may be several reasons that the current analysis failed to replicate the results found by Manuck et al. First, the phenotype used in this study is a sub-phenotype of preterm birth. Manuck et al. used only African-American ancestry and previous preterm delivery as criteria for inclusion of samples in their study.

There was no

discussion of analyzing samples by cause of preterm delivery. It is possible that different genes may contribute to PPROM than to other mechanisms related to preterm birth, like preterm labor or preeclampsia. Additionally, the Manuck et al. study used maternal DNA and this study uses neonate DNA. The varying contributions of the maternal and fetal genomes to parturition are not fully understood. It is likely that different genes in mothers and fetuses play a role in both normal birth timing and preterm birth.

83

The use of admixture mapping to understand the genetics of preterm birth is still in its infancy. With the availability of inexpensive methods for genotyping large panels of polymorphisms, the ability to use admixture mapping to locate novel candidate regions and to help identify genes contributing to the increased risk of preterm birth seen among African-American women is promising.

Future Directions

Covariates were not included in the current analysis. This was due to a lack of information for all genotyped individuals at the time of analysis. When these data are available, the results presented in this chapter will be repeated to include the investigation of covariates such as number of weeks gestation at the time of delivery (particularly to consider early v. late preterm delivery), age of the mother, smoking status of the mother, pre-pregnancy BMI, infection status, and parity, all known to potentially influence risk of preterm delivery. To improve the power to detect novel candidate regions for PPROM, a continuation of this study is planned to increase the sample size. Presently, additional PPROM cases are being collected to facilitate a replication of this work. In addition to implementing an admixture mapping approach with the replication sample, tag SNPs (polymorphisms used to infer variation across a region of the genome due to strong linkage disequilibrium) will be included in future genotyping efforts to evaluate

84

previously implicated candidate genes for PPROM as well as putative candidate genes that show evidence for accelerated evolution as described in Chapter 4. Given that the Illumina (San Diego, California) African-American Admixture Panel is no longer available, the use of a similarly powerful platform for admixture mapping will be needed [44].

At present, Affymetrix (Santa Clara, California) is

marketing a whole genome (Axiom) exom genotyping array that contains over 318,000 markers [61]. In addition to its applications to genome-wide association studies, this array has been validated for use on admixed populations with the inclusion of a large panel of ancestry informative markers (AIMs). Another appealing aspect to this platform is the ability to add a panel of up to 100,000 customer-selected markers to the off-theshelf exom array. This will facilitate the addition of SNPs identified as significant in the current admixture mapping study that might not be represented on the basic exom array, or the inclusion of SNPs within candidate genes previously implicated in PPROM.

Literature Cited

1. 2.

3.

4.

5.

6.

7. 8.

9. 10. 11.

12.

13. 14.

15.

85

Parry, S. and J.F. Strauss, 3rd, Premature rupture of the fetal membranes. N Engl J Med, 1998. 338(10): p. 663-70. Ferrand, P.E., et al., A polymorphism in the matrix metalloproteinase-9 promoter is associated with increased risk of preterm premature rupture of membranes in African Americans. Mol Hum Reprod, 2002. 8(5): p. 494-501. Ahern, J., et al., Preterm birth among African American and white women: a multilevel analysis of socioeconomic characteristics and cigarette smoking. J Epidemiol Community Health, 2003. 57(8): p. 606-11. Wang, H., et al., A functional SNP in the promoter of the SERPINH1 gene increases risk of preterm premature rupture of membranes in African Americans. Proc Natl Acad Sci U S A, 2006. 103(36): p. 13463-7. Meis, P.J., et al., Preterm prediction study: is socioeconomic status a risk factor for bacterial vaginosis in Black or in White women? Am J Perinatol, 2000. 17(1): p. 41-5. Evangelou, E., et al., Impact of phenotype definition on genome-wide association signals: empirical evaluation in human immunodeficiency virus type 1 infection. Am J Epidemiol, 2011. 173(11): p. 1336-42. Frayling, T.M., Genome-wide association studies provide new insights into type 2 diabetes aetiology. Nat Rev Genet, 2007. 8(9): p. 657-62. Winkvist, A., I. Mogren, and U. Hogberg, Familial patterns in birth characteristics: impact on individual and population risks. Int J Epidemiol, 1998. 27(2): p. 248-54. Carr-Hill, R.A. and M.H. Hall, The repetition of spontaneous preterm labour. Br J Obstet Gynaecol, 1985. 92(9): p. 921-8. Hoffman, J.D. and K. Ward, Genetic factors in preterm delivery. Obstet Gynecol Surv, 1999. 54(3): p. 203-10. Barabas, A.P., Ehlers-Danlos syndrome: associated with prematurity and premature rupture of foetal membranes; possible increase in incidence. Br Med J, 1966. 2(5515): p. 682-4. Lind, J. and H.C. Wallenburg, Pregnancy and the Ehlers-Danlos syndrome: a retrospective study in a Dutch population. Acta Obstet Gynecol Scand, 2002. 81(4): p. 293-300. Anum, E.A., et al., Connective tissue and related disorders and preterm birth: clues to genes contributing to prematurity. Placenta, 2009. 30(3): p. 207-15. Fujimoto, T., et al., A single nucleotide polymorphism in the matrix metalloproteinase-1 (MMP-1) promoter influences amnion cell MMP-1 expression and risk for preterm premature rupture of the fetal membranes. J Biol Chem, 2002. 277(8): p. 6296-302. Wang, H., et al., Functionally significant SNP MMP8 promoter haplotypes and preterm premature rupture of membranes (PPROM). Hum Mol Genet, 2004. 13(21): p. 2659-69.

16.

17.

18.

19. 20. 21.

22. 23. 24. 25. 26.

27.

28. 29. 30. 31.

32.

33. 34.

86

Romero, R., et al., A genetic association study of maternal and fetal candidate genes that predispose to preterm prelabor rupture of membranes (PROM). Am J Obstet Gynecol, 2010. 203(4): p. 361 e1-361 e30. Ferrand, P.E., et al., The CARD15 2936insC mutation and TLR4 896 A>G polymorphism in African Americans and risk of preterm premature rupture of membranes (PPROM). Mol Hum Reprod, 2002. 8(11): p. 1031-4. Roberts, A.K., et al., Association of polymorphism within the promoter of the tumor necrosis factor alpha gene with increased risk of preterm premature rupture of the fetal membranes. Am J Obstet Gynecol, 1999. 180(5): p. 1297-302. Anum, E.A., et al., Genetic contributions to disparities in preterm birth. Pediatr Res, 2009. 65(1): p. 1-9. Nei, M. and W.H. Li, Linkage disequilibrium in subdivided populations. Genetics, 1973. 75(1): p. 213-9. Chakraborty, R. and K.M. Weiss, Admixture as a tool for finding linked genes and detecting that difference from allelic association between loci. Proc Natl Acad Sci U S A, 1988. 85(23): p. 9119-23. Nordborg, M. and S. Tavare, Linkage disequilibrium: what history has to tell us. Trends Genet, 2002. 18(2): p. 83-90. Shriver, M.D., et al., Skin pigmentation, biogeographical ancestry and admixture mapping. Hum Genet, 2003. 112(4): p. 387-99. Darvasi, A. and S. Shifman, The beauty of admixture. Nat Genet, 2005. 37(2): p. 118-9. Parra, E.J., et al., Estimating African American admixture proportions by use of population-specific alleles. Am J Hum Genet, 1998. 63(6): p. 1839-51. Lautenberger, J.A., et al., Significant admixture linkage disequilibrium across 30 cM around the FY locus in African Americans. Am J Hum Genet, 2000. 66(3): p. 969-78. Pfaff, C.L., et al., Population structure in admixed populations: effect of admixture dynamics on the pattern of linkage disequilibrium. Am J Hum Genet, 2001. 68(1): p. 198-207. Hoggart, C.J., et al., Design and analysis of admixture mapping studies. Am J Hum Genet, 2004. 74(5): p. 965-78. Patterson, N., et al., Methods for high-density admixture mapping of disease genes. Am J Hum Genet, 2004. 74(5): p. 979-1000. Smith, M.W., et al., A high-density admixture map for disease gene discovery in african americans. Am J Hum Genet, 2004. 74(5): p. 1001-13. Bonilla, C., et al., Admixture in the Hispanics of the San Luis Valley, Colorado, and its implications for complex trait gene mapping. Ann Hum Genet, 2004. 68(Pt 2): p. 139-53. Carlson, C.S., et al., Additional SNPs and linkage-disequilibrium analyses are necessary for whole-genome association studies in humans. Nat Genet, 2003. 33(4): p. 518-21. Barnes, K.C., et al., African Americans with asthma: genetic insights. Proc Am Thorac Soc, 2007. 4(1): p. 58-68. Zhu, X., et al., Admixture mapping for hypertension loci with genome-scan markers. Nat Genet, 2005. 37(2): p. 177-81.

35.

36.

37. 38.

39. 40. 41. 42.

43.

44.

45. 46. 47. 48. 49. 50.

51.

52. 53.

87

Cheng, C.Y., et al., Admixture mapping scans identify a locus affecting retinal vascular caliber in hypertensive African Americans: the Atherosclerosis Risk in Communities (ARIC) study. PLoS Genet, 2010. 6(4): p. e1000908. Deo, R.C., et al., Single-nucleotide polymorphisms in LPA explain most of the ancestry-specific variation in Lp(a) levels in African Americans. PLoS One, 2011. 6(1): p. e14581. Reich, D., et al., A whole-genome admixture scan finds a candidate locus for multiple sclerosis susceptibility. Nat Genet, 2005. 37(10): p. 1113-8. Freedman, M.L., et al., Admixture mapping identifies 8q24 as a prostate cancer risk locus in African-American men. Proc Natl Acad Sci U S A, 2006. 103(38): p. 14068-73. Haiman, C.A., et al., Multiple regions within 8q24 independently affect risk for prostate cancer. Nat Genet, 2007. 39(5): p. 638-44. Bock, C.H., et al., Results from a prostate cancer admixture mapping study in African-American men. Hum Genet, 2009. 126(5): p. 637-42. Cossrow, N. and B. Falkner, Race/ethnic issues in obesity and obesity-related comorbidities. J Clin Endocrinol Metab, 2004. 89(6): p. 2590-4. Cheng, C.Y., et al., Admixture mapping of obesity-related traits in African Americans: the Atherosclerosis Risk in Communities (ARIC) Study. Obesity (Silver Spring), 2010. 18(3): p. 563-72. Manuck, T.A., et al., Admixture mapping to identify spontaneous preterm birth susceptibility loci in African Americans. Obstet Gynecol, 2011. 117(5): p. 107884. Illumina. Illumina African American Admixture Panel. 2010. http://www.illumina.com/products/african_american_admixture_panel.ilmn. February 29, 2012. Purcell, S. PLINK. v1.07. http://pngu.mgh.harvard.edu/purcell/plink/. Purcell, S., et al., PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet, 2007. 81(3): p. 559-75. Hoggart, C.J., et al., Control of confounding of genetic associations in stratified populations. Am J Hum Genet, 2003. 72(6): p. 1492-1504. The International HapMap Project. Nature, 2003. 426(6968): p. 789-96. Chakraborty, R. and K.M. Weiss, Frequencies of complex diseases in hybrid populations. Am J Phys Anthropol, 1986. 70(4): p. 489-503. Hanis, C.L., et al., Individual admixture estimates: disease associations and individual risk of diabetes and gallbladder disease among Mexican-Americans in Starr County, Texas. Am J Phys Anthropol, 1986. 70(4): p. 433-41. McKeigue, P.M., et al., Estimation of admixture and detection of linkage in admixed populations by a Bayesian approach: application to African-American populations. Ann Hum Genet, 2000. 64(Pt 2): p. 171-86. Pritchard, J.K., et al., Association mapping in structured populations. Am J Hum Genet, 2000. 67(1): p. 170-81. Ruscheinsky, M., C. De la Motte, and M. Mahendroo, Hyaluronan and its binding proteins during cervical ripening and parturition: dynamic changes in size, distribution and temporal sequence. Matrix Biol, 2008. 27(5): p. 487-97.

54. 55. 56.

57. 58.

59. 60. 61.

88

McLean, M., et al., A placental clock controlling the length of human pregnancy. Nat Med, 1995. 1(5): p. 460-3. McLean, M. and R. Smith, Corticotropin-releasing Hormone in Human Pregnancy and Parturition. Trends Endocrinol Metab, 1999. 10(5): p. 174-178. Frim, D.M., et al., Characterization and gestational regulation of corticotropinreleasing hormone messenger RNA in human placenta. J Clin Invest, 1988. 82(1): p. 287-92. Grammatopoulos, D.K. and E.W. Hillhouse, Role of corticotropin-releasing hormone in onset of labour. Lancet, 1999. 354(9189): p. 1546-9. Li, W. and J.R. Challis, Corticotropin-releasing hormone and urocortin induce secretion of matrix metalloproteinase-9 (MMP-9) without change in tissue inhibitors of MMP-1 by cultured cells from human placenta and fetal membranes. J Clin Endocrinol Metab, 2005. 90(12): p. 6569-74. Kalantaridou, S.N., et al., Corticotropin-releasing hormone, stress and human reproduction: an update. J Reprod Immunol, 2010. 85(1): p. 33-9. Velez, D.R., et al., Preterm birth in Caucasians is associated with coagulation and inflammation pathway gene variants. PLoS One, 2008. 3(9): p. e3283. Affymetrix. Axiom Exome Genotyping Array Data Sheet. http://www.affymetrix.com/support/technical/datasheets/axiom_exome_array_dat asheet.pdf. March 12, 2012.

89

Chapter 4

Investigating Evidence for Accelerated Evolution at Preterm Birth Candidate Genes

Abstract

Preterm birth is a complex phenotype that is a leading cause of infant mortality in the United States. Although social and environmental factors play a role in risk of prematurity, it is increasingly understood that genes likely contribute to risk. AfricanAmerican women experience the greatest risk of preterm birth among U.S. populations, nearly twice that of European-American women.

In this study, tests to screen for

accelerated evolution of both the regions identified by admixture mapping and or previously reported preterm birth candidate genes were used to prioritize a panel of genomic regions and candidate genes to better understand the role of genetics in the increased risk of preterm birth among African-American women. In Chapter 3, admixture mapping analysis was conducted using a sample of 616 neonates born to women of self-reported African-American ancestry. Genotyping of DNA from the newborns of 352 women with confirmed preterm premature rupture of membranes (PPROM) and neonates from 264 women with normal term pregnancy outcomes was completed using a genome-wide panel of ancestry informative markers (AIMs) designed for admixture mapping analysis in African-American populations. Bayesian admixture mapping identified six regions on five chromosomes (5, 8, 11, 19 and 21) that contribute to risk of PPROM. These ancestry association peaks as well as

90

90 previously reported candidate genes for preterm birth were tested for signatures of accelerated evolution using locus specific branch length (LSBL), log of the ratio of heterozygosities (lnRH), and normalized Tajima’s D. All of the genomic regions found by admixture mapping to be associated with risk of PPROM among African-American women had significant evidence of accelerated evolution. Additionally, 31 of the 90 candidate genes screened were found to show signatures of accelerated evolution in West Africans and 24 in Europeans. For phenotypes that vary among populations, like preterm birth, tests for accelerated evolution may be a useful method for nominating candidate regions and candidate genes for further study.

Introduction

In the United States incidence of preterm birth is substantially greater among African-American women compared to women of any other ancestry.

It has been

suggested that while social and environmental factors may account for a portion of the increased incidence of preterm birth seen in African-American women, there is likely a genetic component to risk. To date, numerous studies have been undertaken to identify genes that contribute to risk of preterm delivery with limited success in identifying genetic variants that explain the disparity in risk experienced by African-American women [1].

In an attempt to prioritize candidate genes that influence risk of preterm

birth in the admixed African-American population, previously identified regions of the genome found to be associated with preterm premature rupture of membranes (PPROM)

91

(Chapter 3) and previously published preterm birth candidate genes were screened for evidence of accelerated evolution in West African and European parental populations.

Background

Modern human populations exhibit a large amount of phenotypic variation both in many normal traits and in susceptibility to various diseases. Evolutionary forces have shaped the genes contributing to this variation over the past 100-200 thousand years as anatomically modern humans evolved in Africa and small populations began migrating to inhabit a variety of environments outside of Africa [2, 3]. As large genotyping and sequencing platforms have become available, testing the genome for signatures of departures from neutrality, or accelerated evolution, has become possible. A variety of statistical methods have been created to localize genes shaped by evolution due to random change such as genetic drift, demographic effects such as population bottlenecks or expansions, and directional selection (purifying or positive) resulting from adaptation to varying environments and pathogens.

These tests exploit

predictable patterns in the genome that are likely to occur when evolution is acting. As selection acts to increase the frequency of the advantageous allele a reduction in heterozygosity at the gene under selection and at linked loci in the area surrounding the gene will occur. This phenomenon is referred to as a “selective sweep”. Additionally, large differences in allele frequency between populations are an indication that evolution has been acting differentially among populations.

92

Some of the strongest signals of evolution in the genome appear to be in genes that confer a fitness advantage to the population under study.

Evidence for strong

selection has been reported for variation in normal human phenotypes like skin pigmentation, genes that confer resistance to pathogens such as malaria, and even among genes associated with diet like the convergent evolution of lactase persistence at the LCT gene among Northern Europeans and some African populations [4-6]. Complex disease phenotypes that show differential risk among human populations are likely the result of the interplay between environment, lifestyle, and genetic factors, as is likely the case with preterm birth. Although natural selection is unlikely to contribute directly to disparity in preterm birth, the pathways that have been implicated in birth timing, like inflammation and endocrine, affect many tissues and systems of the body. Due to the pleiotropic effects of these genes, if selection has acted for a different phenotype, it is possible that changes in these genes have affected preterm birth risk. Identifying genomic regions and candidate genes that have signatures of evolution may provide insight into the genetic contributions to disparity in disease risk. For this study three tests for accelerated evolution, locus specific branch length (LSBL), log of the ratio of heterozygosities (lnRH), and normalized Tajima’s D, were conducted on the six chromosomal regions previously reported to contribute to risk of preterm premature rupture of membranes (PPROM) among Africa-American women (see Chapter 3) and ninety previously reported candidate genes for preterm birth [7-9]. The goal of this research is to prioritize genomic regions and candidate genes for future genotyping efforts that are more likely to explain the difference in risk of preterm birth experienced by African-American women.

Materials and Methods

93

Admixture Mapping

As described in Chapter 3, DNA samples from 616 newborns of self-identified African-American women 352 with preterm premature rupture of membranes (PPROM), the largest contributor to preterm birth, and 264 with normal pregnancy outcomes were analyzed on the Illumina (San Diego, California) African-American Admixture Panel [10, 11]. The panel is comprised of 1,509 single nucleotide polymorphisms (SNPs) spanning the human genome. These markers were specifically selected for admixture mapping because they are ancestry informative, that is, they have large allele frequency differences between the West African and European parental populations known to contribute to African-American admixture [12-14].

These SNPs are called ancestry

informative markers and are referred to as AIMs in this chapter. Admixture mapping analysis was conducted in ADMIXMAP using the 1,398 AIMs that passed quality controls and met the other inclusion criteria previously described in Chapter 3 [15-17].

The analysis resulted in the identification of five

chromosomes with peaks of significant ancestry association with PPROM (|z| > 2.5, pvalue < 0.01). Chromosomes 5, 8, 11 (two peaks), and 19 had evidence of association with West African ancestry while a large peak on chromosome 21 was associated with European ancestry. In total six chromosomal regions on five chromosomes were found to have ancestry association with the phenotype of interest, PPROM. These peaks and

94

the AIMs that were within the peaks were analyzed for evidence of accelerated evolution using three tests, locus-specific branch length (LSBL), log of the ratio of heterozygosities (lnRH) and normalized Tajima’s D. For the analysis of the ancestry associated peaks a window beginning 200 kilobases (kb) before the first AIM in the ancestry peak and ending 200 kb beyond the last AIM in the peak were analyzed. The start and end positions of the first and last AIM within each ancestry associated peak are shown in Table 4-1 along with the AIMs that are found within the ancestry peak and the ancestral population of ancestry association.

It is important to note that the first ancestry

association peak on chromosome 11 described in Chapter 3 has been divided into two different peaks for the analyses included in this study because not all of the SNPs within the peak had significant ancestry association values (p < 0.01). The two test regions created from this one ancestry association peak have continuous significant SNPs from the African-American Admixture panel; they are labeled 11 Peak 1 and 11 Peak 2. The third peak analyzed in this study, 11 Peak 3, is the second peak on chromosome 11 described in Chapter 3. In addition to the analysis of each chromosomal region found to be associated with PPROM by admixture mapping, each AIM within the peaks was investigated for evidence of accelerated evolution. As with the peak analysis, a 200 kb window upstream and downstream of the AIM position was used for the screens. There is no overlap between test windows for AIMs from the same ancestry association peak.

The

chromosomal position and test window location for each of the AIMs is shown in Table 4-3 of the results.

95

Table 4-1. Description of chromosomal peaks identified by admixture mapping as associated with PPROM. The peak “start” and “end” positions are determined by the subtracting 200kb from the first SNP found within the peak and adding 200kb to the last ancestry-associated SNP within the peak. The ancestry association of each peak is listed in the “Assoc. Pop.” column of this table.

96

Candidate Genes

Genes that have been previously described in candidate gene studies of preterm birth were chosen for investigation in this study. This included 84 genes listed in the Preterm

Birth

Gene

database

(PTBGene

database)

(http://bioinformatics.aecom.yu.edu/ptbgene/index.html) and in Preterm Birth: Causes, Consequences and Prevention (Behrman and Butler 2007) [18, 19]. Dr. Jerome Strauss of Virginia Commonwealth University suggested an additional six genes from more recent publications (ENPP1, SERPINH1, FSHR, TIMP2, IGF2, and COL4A3) for inclusion in this analysis (personal communication with Dr. Strauss). The candidate genes listed in the PTBGene database and Behrman and Butler (2007) are not limited to any specific phenotype associated with preterm birth, such as preterm premature rupture of membranes or preterm labor. These genes are found in a variety of pathways thought to be associated with birth timing and possibly preterm birth, including inflammatory, uteroplacental, endocrine, uterine contraction, coagulation, metabolic, and matrix metabolism pathways. In total ninety genes were tested for accelerated evolution. For each gene analyzed the transcription start and end positions were determined using the UCSC Genome Browser (http://genome.ucsc.edu/cgi-bin/hgGateway) [20]. For the tests of accelerated evolution a window was created to include the gene plus 200 kilobases (kb) upstream and downstream of the gene. A full list of the candidate genes included in this analysis and their chromosomal positions are shown in Appendix E.

Tests for Accelerated Evolution

97

Sample Populations and Genotyping Panel

The data set used for this study consists of genome-wide genotype data from four parental populations known to contribute to admixture in the United States, European, West African, East Asian, and Native American. Three of the populations included are from the International HapMap Consortium [21]. These include CEPH European individuals from Utah (n = 60), Yoruba individuals from Ibidan, Nigeria (n = 60), and East Asians from Beijing, China (Han Chinese) and Tokoyo, Japan (n = 90). Additionally, Indigeous American (n = 88) individuals from four Central and South American locations were also analyzed but not used in the current study. Genotyping of DNA from these individuals was performed on the Affymetrix (Santa Clara, California) Genome-wide Human SNP Array 6.0 [22]. This genotyping panel included 906,600 single nucleotide polymorphism distributed across the nuclear and mitochondrial genomes at an average marker spacing of ~1.7 kilobases (kb). Genotyping markers and individuals with low call rates (< 95%) were excluded from the analysis.

Locus Specific Branch Length (LSBL)

The first test for accelerated evolution assessed in this research is locus-specific branch length (LSBL). This test uses pairwise measurements of FST in three populations

98

to identify directional change in frequency of an allele of interest. F-statistics were first described by Sewall Wright as a way to measure inbreeding by looking at reductions in heterozygosity [23].

The FST statistic determines how much of the variation exists

within subpopulations (S) compared to the total population (T).

Pairwise FST is

calculated using the following equation:

Fst =

σp 2 pq

where, σ is the variance in allele frequencies in a single population and p and q (q = 1 - p) are the mean allele frequencies in the whole population. Values of FST range from 0 to 1. A value of 0 indicates there is no difference between populations whereas FST equal to 1 indicates that the variation all of the variation is between populations and that an allele is fixed in one population and lost in the other. Large values of FST can occur due to natural selection, however they can also indicate demographic factors like population bottlenecks or genetic drift. Weir and Cockerham (1984) developed an unbiased FST estimator that is used in this analysis [24]. The unbiased FST (θ) is a more appropriate test than Wright’s FST for human data because it accounts for variation in sample size between populations and the inability in natural populations to sample all of the genetic variation. Weir and Cockerham’s unbiased FST equation is:

99

where, a is the variance between populations, b is the variance between individuals within a population, and c is the variance between gametes within individuals. To calculate locus specific branch length (LSBL) first pairwise FST was calculated between each parental population, West Africans and Europeans (dWE), West Africans and East Asians (dWA), and Europeans and East Asians (dEA) [7] (see Figure 4-1). Then each branch length was calculated using the following formulas: West African

LSBLW = (dWE + dWA – dEA)/2

European

LSBLE = (dWE + dEA – dWA)/2

East Asian

LSBLA = (dWA + dEA – dWE)/2

100

Figure 4-1. Diagram of Locus Specific Branch Length (LSBL) using the three populations modeled in this study: West African (W), European (E), and East Asian (A). LSBL is calculated from the genetic distances (D) between the population of interest and two other populations. For this study, LSBLW was calculated as (dWE + dWA – dEA) / 2 and LSBLE was calculated as (dWE + dEA – dWA)/2. Adapted from Shriver et al. (2004) [7].

A long branch length in a particular population can be interepreted in two ways, either that the change occurred specifically in that population or that the change occurred in the other two populations after their divergence from the common ancestor of the population of interest. For example, a large branch length in West Africans could indicate evolution in Africa after Europeans and East Asians migrated out of Africa or evolution occurring in European and East Asian populations, or the common ancestor of these populations, since their split with Africa. Large positive values of LSBL at a locus reflects dramatic

101

change in allele frequency either due to the action of natural selection or random genetic drift. Empirical distributions of LSBL values were generated for West Africans and Europeans and SNPs that fell within the top 5% of the distribution were considered significant. The cut-off for the West African sample was 0.3664 and the value for Europeans was 0.2232.

Log of the Ratio of Heterozygosity (lnRH)

Under the neutral expectation, two populations should maintain the same level of genetic diversity that existed prior to their divergence. In this case, the level of heterozygosity in both populations should be the same. If, however, genetic drift or natural selection has acted at a locus, a reduction in heterozygosity is expected. The lnRH statistic was developed by Schlotterer and Dieringer (2005) to measure this difference in genetic variation between two populations [8, 25]. Using this statistic, a reduction in heterozygosity in population 1 compared to population 2 will result in a negative lnRH value and indicate that an evolutionary force, like natural selection or genetic drift, has acted on the tested locus in the first population and not in the second. The equation for lnRH is shown below. In this equation H is the expected heterozygosity determined for each of the two populations from the known allele frequencies in the populations.

   2   1  E   −1    1 − H Pop1         = ln Gene Diversity Population 1  ln RH = ln    2    Gene Diversity Population 2  1  −1   E      1 − H Pop 2 

102

In this study the lnRH values were calculated for the ratio of West African/European (for West Africans) and also European/West African (for Europeans) for each SNP in the specified gene window. As described previously, negative values indicate accelerated evolution in the population of interest (population 1) compared to a reference population (population 2). Significance cut-off values for lnRH determined established from the bottom 5% of the empirical distribution for West Africans is -3.5837 and -6.9975 for Europeans.

Normalized Tajima’s D

The Tajima’s D statistic was developed to detect regions of the genome that depart from the neutral expectation. Using sequence data two different measure of gene diversity (θ) are compared, the number of segregating sites (S) and the average number of pairwise differences (π) [9]. In the absence of evolution or changes in population size, the values of S and π will be equal. However, if differences in these values exist D will become either positive or negative. Positive values of D occur when π is greater than S and indicate either balancing selection or a reduction in population size. A large S value

103

will result in a negative Tajima’s D. Negative values of D can indicate either positive or purifying selection or a population expansion. equation below.

Tajima’s D is calculated using the

The values of the variables a1, e1 and e2 are related to the number of

sequences included in the calculation.

D=

π − Sa 1 e1S + e2 S(S −1)

For this study, Tajima’s D was calculated using overlapping sliding windows of 100 kb in 25 kb increments to accommodate the use of SNPs instead of sequence data [26]. The number of SNPs in each window varies as a result of the genotyping platform used for this analysis. Windows containing fewer than five successfully genotyped SNPs were not considered.

Tajima’s D is sensitive to demographic factors, especially to

population expansion after a bottleneck which leads to an excess of rare variants. To account for the varying effects of demographic factors across populations, the Tajima’s D values used in this analysis were normalized by dividing the D value calculated in each window by the average genome-wide D value for the population of interest, West African and European in this analysis. For the purposes of this study, only negative D values, evidence of directional selection, were considered. The 5% cutoffs established by the empirical distributions for normalized Tajima’s D were -1.7708 and -1.9209 for the West African and European populations, respectively.

Evaluating the Significance of the Tests of Accelerated Evolution

104

For each PPROM ancestry association peak or preterm birth candidate gene a 200 kilobase (kb) window surrounding the SNP or gene was tested for evidence of accelerated evolution using three tests, locus specific branch length (LSBL), log of the ratio of heterozygosities (lnRH) and normalized Tajima’s D in this analysis. Tests of accelerated evolution were conducted for both West African and European parental populations. Figures were generated for each test of evolution for each chromosomal region or candidate gene that was tested for both West Africans and Europeans. Each blue or orange vertical line represents an individual SNP (LSBL and lnRH) or sliding window (Tajima’s D) tested within the peak of interest. SNPs or windows that fall in the appropriate 5% tail of the distribution are considered significant for the test, as described in the methods, and are displayed in orange. A consideration when conducting the screens of accelerated evolution on the regions and SNPs identified by admixture mapping is that all of the markers used on the Illumina African-American Admixture Panel are ancestry informative and may be enriched for areas of the genome that are not neutral. To verify that there was not increased evidence for accelerated evolution at the panel AIMs, a 200kb region surrounding each marker in the admixture mapping panel was tested in West Africans for each of the screens of accelerated evolution. The empirical distribution of the SNPs (LSBL and lnRH) or windows (normalized Tajima’s D) was generated and the 5% cutoff was established for each test. In all of the tests, the 5% cut-off for the AIMs panel is less stringent than for the panel of SNPs used to evaluate accelerated evolution. That is

105

to say, the cut-off for lnRH and normalized Tajima’s D are less negative for the AIMs panel than for the SNP panel and more positive for LSBL. Table 4-2 displays the 5% empirical cut-off values for the full SNP panel and the admixture mapping AIMs panel.

Table 4-2. The 5% empirical distribution cut-off values for each test of accelerated evolution in the regions surrounding the Ancestry Informative Markers in the array used for admixture mapping compared to the 5% cut-off for the SNP panel used for the tests of accelerated evolution.

West African #SNPs/Windows 5% AIMs Distribution

5% SNPs Cutoff

LSBL

174526

0.4516

0.3664

lnRH

206989

-6.2630

-6.8090

Tajima's D

22455

-1.4690

-1.7708

To summarize the tests of accelerated evolution a Perl script was used to calculate the proportion of significant SNPs/windows for each test in each chromosomal region or candidate gene tested. Nearly every chromosomal region or candidate gene analyzed had significant values for at least one of the tests of evolution. To avoid reporting spurious results and to help prioritize a manageable number of candidate regions and genes for future analysis a significance threshold was established. Preterm birth candidate genes and PPROM ancestry association regions were considered significant if there was evidence of departure from the neutral expectation, no evolution, in all three of the tests

106

performed (LSBL, lnRH, and normalized Tajima’s D) or if more than 5% of the SNPs or windows in the candidate gene were significant in two of the tests. Setting a threshold of significance may remove candidate regions and genes that have experienced accelerated evolution, however it reduces the probability of prioritizing false-positive results negatively impacting future genotyping efforts.

Results

Screens for Accelerated Evolution from Admixture Mapping Results for PPROM

In the current study, the chromosomal regions found to have ancestry association in the admixture mapping analysis were tested for evidence of accelerated evolution in both West Africans and Europeans. This included one region on each of chromosomes 5, 8, 19 and 21, and three peaks on chromosome 11. Plots of the screens for accelerated evolution are shown in Figure 4-2. In the West African tests for accelerated evolution, each of the seven PPROM ancestry association peaks, except for the small region on chromosome 8, showed evidence for accelerated evolution in West African populations for all three tests (LSBL, lnRH and normalized Tajima’s D). The strongest evidence of accelerated evolution among the West African tests was in chromosome 11 peak 1 (11 Peak 1). This peak has significance values at greater than 5% of the SNPs or windows for each of the three tests (LSBL 5.9%, lnRH 6.13%, and normalized Tajima’s D 8.33%).

107

For the European screens, all of the ancestry association peaks evaluated were

significant for at least two of the tests of accelerated evolution. Of note is the small peak on chromosome 8 (8 Peak) that is the only chromosomal segment tested that has greater than 5% significant SNPs or windows for each of the three tests of accelerated evolution (LSBL 6.82%, lnRH 7.84%, and normalized Tajima’s D 25.0%).

This peak is the

smallest chromosomal region identified by admixture mapping with ancestry association with PPROM and contains only one AIM (rs16932440). A summary of the significant tests of accelerated evolution by ancestry association peak and population is displayed in Table 4-3. A table with the percentage of significant SNPs or windows for each test for each ancestry association peak tested is shown in Appendix D. The chromosomal regions identified by admixture mapping to have significant ancestry association with PPROM are not uniform in size and therefore will not contain the same number of SNPs or windows for the tests of evolution. For example, the largest ancestry association peak is on Chromosome 21 and contains eight ancestry associated AIMs over 11.5 megabases (Mb). By contrast, the smallest peak is on chromosome 8 which contains just one ancestry associated AIM. To evaluate the composition of these peaks more consistently, each AIM within the ancestry association peaks was also tested for evidence of accelerated evolution. Table 4-4 contains a summary of the significant tests for each AIM within the PPROM ancestry association peaks. The individual AIM values for each test and the plots of the accelerated evolution screens can be found in Appendix D.

108

Analysis of the individual AIMs within the PPROM ancestry association peaks

revealed that while the entire peak may show evidence for accelerated evolution, when evaluated separately, not all of the AIMs within the peak are significant. Additionally, in some cases, when an ancestry association peak appears to be positive for accelerated evolution in both West Africans and Europeans, the individual AIM analysis suggests population-specific patterns of evolution within the peak. One example of this is peak 2 on chromosome 11 (11 Peak 2). When the full peak is tested for evidence of accelerated evolution, both populations (West African and European) appear to have significant values for the three tests evaluated. However, screening of the individual AIMs within this peak shows a shifting pattern, with significant evolution in Europeans at the first three AIMs (rs10765838, rs2403595, and rs6486270) in the peak followed by two AIMs (rs4148636 and rs11024739) that are significant in West Africans. A similar shifting trend of accelerated evolution between West Africans and Europeans is seen in the ancestry association peak on chromosome 5 and the second peak on chromosome 11.

(a) West African Admixture Mapping Peaks Chromosome 5

109

110 Chromosome 11 – Peak 2

Chromosome 11 – Peak 3

111 Chromosome 21

(b) European Admixture Mapping Peaks Chromosome 5

Chromosome 8

112 Chromosome 11 – Peak 1

Chromosome 11 – Peak 2

Chromosome 11 – Peak 3

113 Chromosome 19

Chromosome 21

Figure 4-2. Results of the tests for accelerated evolution in all of the chromosomal regions found to be associated with preterm premature rupture of membranes (PPROM) by admixture mapping. Each bar represents a SNP (LSBL and lnRH) or window (Tajima’s D) plotted along the chromosome. Blue bars are not significant and orange bars fall in the 5% tail of the empirical distribution for the given test in the population under study. The populations presented in this figure are (a) West African (b) European.

114

Table 4-3. Summary of the tests for accelerated evolution in the PPROM ancestry association peaks identified by admixture mapping. The position of the test window was determined by subtracting 200kb from the first AIM in the ancestry association peak and adding 200kb to the last AIM in the peak. An ‘X’ represents a positive test either because greater than 5% of the SNPs or windows within the region are significant (bold “X”) or because all three tests have evidence for accelerated evolution. Gray regions of the plot represent ancestry association peaks that are not significant for the population listed.

Table 4-4. Summary of the tests for accelerated evolution in the AIMs that make-up the PPROM ancestry association peaks identified by admixture mapping. The start position of the test window was determined by subtracting 200kb from the AIM position and the end of the window was likewise by adding 200kb to physical position of the AIM. An ‘X’ represents a positive test either because greater than 5% of the SNPs or windows within the region are significant (bold “X”) or because all three tests have evidence for accelerated evolution. Gray regions of the plot represent AIMs that are not significant for the population listed.

115

Tests of Accelerated Evolution in Preterm Birth Candidate Genes

116

In addition to investigating accelerated evolution in the chromosomal regions identified as significantly associated with preterm premature rupture of membranes by admixture mapping, previously published candidate genes for preterm birth were also evaluated. Figure 4-3 displays plots of each test for the candidate genes that were found to be significant in each population, and a summary of the significant results is shown in Table 4-5. In total 90 genes were screened in both West African and European parental populations for evidence of accelerated evolution. Forty-four of the candidate genes tested were found to be significant in at least one of the populations studied. Of these, 31 were significant in West Africans and 24 in Europeans. Twelve genes showed evidence of evolution in both populations.

Forty-six of the candidate genes included in the

analysis were not found to be significant for either West Africans or Europeans. Table 4-6 summarizes the genes that were found to be significant for each population. Among the 44 preterm birth candidate genes identified in West Africans, six have greater than 5% significant SNPs/windows for all three of the tests. These genes are PTGER3, AGTR1, FGB, VEGF, FL, and CBS.

In Europeans only two of the 24

significant candidate genes have greater than 5% significance for each of the three tests of accelerated evolution, CYP2E1 and CYP1A1. Of the twelve candidate genes that overlap between the tests in West Africans and Europeans, only three genes have evidence for accelerated evolution in the three screens performed in this analysis, VEGF,

117

F2 and CBS. However, none of these candidate genes has values for all three tests in both populations with greater than 5% significant SNPs/windows.

Plots for the genes with non-significant tests for accelerated evolution are shown in Appendix E, as is a table with the detailed information of each test for all of the candidate genes included in this study.

(a) West African Candidate Gene Screens

TNFRSF1B – 1p36.22

PTGER3 – 1p31.2

SELE – 1q22-q25

118

119 PPARG – 3p25

AGTR1 – 3q24

ADD1 4q16.3

120 TLR2 – 4q32

FGB – 4q28

ITGA2 – 5q11.2

121 PART1 – 5q12.1

CRHBP – 5q11.2-q13.3

ADRB2 – 5q31-q32

122 VEGF – 6p12

LTA – 6q21.3

ENPP1 – 6q22-q23

123 SERPINE1 – 7q21.3-q22

PLAT – 8q12

CYP2C19 – 10q24

124 F2 – 11p11

PGR – 11q22-q23

GNB3 – 12q3

125 ALOX5AP – 13q12

F7 – 13q34

NQO1 – 16q22.1

126 NOS2A – 17q11.2-q12

ITGB3 – 17q21.32

ACE – 17q23.3

127 TIMP2 – 17q25

ICAM1 – 19p13.3-p13.2

CBS – 21q22.3

128 GSTT1 – 22q11.23

(b) European Candidate Gene Screens MTHFR – 1p36.3

TNFRSF1B – 1p36.22

129 FASLG– 1q23

IL1B – 2q14

IL1RN – 2q14.2

130 PPARG – 3p25

TLR2 – 4q32

FGB – 4q28

131 IL5 – 5q31.1

IL4 – 5q31.1

ENPP1 – 6p22-q23

132 VEGF – 6p12

TREM1 – 6q21.1

PLAT – 8q12

133 CYP2E1 – 10q2.3-qter

IGF2 – 11p15.5

F2 – 11p11

134 SERPINH1 – 11q13.5

TNFRSF1A – 12p13.2

GNB3 – 12q13

135 KL – 13q12

F7 – 13q34

CYP1A1 – 15q24.1

136 NQO1 – 16q22.1

Figure 4-3. Results of the significant tests for accelerated evolution in previously published candidate genes for preterm birth. Each bar represents a SNP (LSBL and lnRH) or window (Tajima’s D) plotted along the chromosome. Blue bars are not significant and orange bars fall in the 5% tail of the empirical distribution for the given test in the population under study. The populations shown in this figure are (a) West African (b) European.

137

Table 4-5. Summary of the genes that are significant for tests of accelerated evolution. An “X” signifies at least 5% of the SNPs or windows tested are significant (bold “X”), or in the case that all three tests have significant values, there may be an “X” listed even if the test does not contain at least 5% significant SNPs or windows. Gray bars denote that the tests were not found to be significant in the population shown.

138

139

Table 4-6. Significant results of screens for accelerated evolution surrounding previously reported candidate genes for preterm birth. Both West African (a) and European (b) analyses are displayed in this figure.

Population

West African

European

Both Populations

# Significant Genes

Genes

31

TNFRSF1B, PTGER3, SELE, PPARG, AGTR1, ADD1, TLR2, FGB, ITGA2, PART1, CRHBP, ADRB2, VEGF, LTA, ENPP1, SERPINE1, PLAT, CYP2C19, F2, PGR, GNB3, ALOX5AP, F7, NQO1, NOS2A, ITGB3, ACE, TIMP2, ICAM1, CBS, GSTT1

24

MTHFR, TNFRSF1B, FASLG, IL1B, IL1RN, PPARG, TLR2, FGB, IL5, IL4, ENPP1, VEGF, TREM1, PLAT, CYP2E1, IGF2, F2, SERPINH1, TNFRSF1A, GNB3, KL, F7, CYP1A1, NQO1

12

TNFRSF1B, PPARG, TLR2, FGB, ENPP1, VEGF, PLAT, F2, GNB3, F7, NQO1, CBS

A summarized list of the forty-four candidate genes with significant evidence for accelerated evolution in at least one of the parental populations investigated in this analysis includes:

MTHFR, TNFRSF1B, PTGER3, SELE, FASLG, IL1B, IL1RN,

PPARG, AGTR1, ADD1, TLR2, FGB, ITGA2, PART1, CRHBP, IL5, IL4, ADRB2, LTA, TREM1, ENPP1, VEGF, SERPINE1, PLAT, CYP2C19, CYP2E1, IGF2, F2, SERPINH1, PGR, TNFRSF1A, GNB3, ALOX5AP, KL, F7, CYP1A1, NQO1, NOS2A, ITGB3, ACE, TIMP2, ICAM1, CBS, and GSTT1.

Discussion

140

This research uses tests of accelerated evolution to investigate regions of the genome identified by admixture mapping (see Chapter 3) that contribute to risk of preterm premature rupture of membranes (PPROM) among African-American women. Additionally, ninety previously reported preterm birth candidate genes were screened for evidence of evolution in the parental populations that contribute to African-American admixture, West African and European. The goal of this study is to inform the planned replication of the PPROM study reported in Chapter 3 and to create a list of prioritized candidate genes for future investigations of disparity in preterm birth risk found among U.S. African-American women. Analysis of chromosomal regions found in the previously described admixture mapping study revealed that all of the ancestry-associated peaks had evidence of evolution in at least one of the populations considered.

These ancestry-associated

chromosomal regions vary in size with different numbers of AIMs within each peak. The windows generated to test the ancestry association peaks range in size from 400 kilobases (kb), with only one AIM in the peak plus 200kb on either side, to nearly 12 megabases (Mb) on chromosome 21. Within these peaks there may be evidence for varying effects of accelerated evolution across the chromosomal region. By evaluating the individual AIMs within the ancestry association peaks, variation in the patterns of evolution became apparent and that different regions had evidence in Europeans compared to West Africans. That is to say, AIMs in one region of a peak may have evidence of accelerated

141

evolution in Europeans while other AIMs in the same peak are significant in West Africans. One of the most dramatic ancestry associations with preterm birth due to preterm premature rupture of membranes (PPROM) found in the admixture mapping results (reported in Chapter 3) was the very large peak on chromosome 21 spanning 11.5 megabases (Mb) that was associated with European ancestry.

In the screens for

accelerated evolution presented here, both West African and European parental populations were significant with a slightly different pattern of SNPs within the peak for each population.

Interestingly, only one of the previously reported candidate genes for

preterm birth falls on chromosome 21, CBS. This gene showed evidence for accelerated evolution in both West Africans and Europeans in the current analysis. CBS has been reported in two previous studies of preterm birth; however, both studies were conducted in populations of European ancestry, one in the United States and the other in Australia [27, 28].

Of the two, only Velez et al. (2008) found an association between a

polymorphism in CBS among European-American neonates and risk of preterm birth. With the large ancestry association peak found on this chromosome by admixture mapping and evidence of accelerated evolution in both the ancestry association peak and the preterm birth candidate gene CBS, chromosome 21 warrants further study for its potential role in preterm birth risk. Of the forty-four preterm birth candidate genes with positive screens for signatures of accelerated evolution in this study, only five are previously reported candidates for preterm premature rupture of membranes (PPROM). These genes include SERPINH1, IL1B, TIMP2, PLAT, and LTA [1, 29-31]. Of these genes, only SERPINH1

142

has been previously demonstrated to contribute to the disparity in preterm birth risk

among African-American women. These five genes are of particular interest for the planned replication of the PPROM study reported in Chapter 3. Interestingly, the proportion of candidate genes that were identified using the screens for signatures of accelerated evolution was higher than expected: 44 of 90 candidate genes in either the West African or European parental population. This large number is likely due to the liberal cutoffs chosen for evidence of accelerated evolution. By definition, alleles that contribute to difference in genetic risk of preterm birth between African-American and European-American women will show a substantial frequency difference between the parental populations that contribute to these groups. The tests performed in this analysis help identify candidate genes and genomic regions with evidence of ancestry-informative differences, those that are more likely to contribute to differences in risk of preterm birth. For this analysis the primary goal was to prioritize candidate genes and genomic regions for future evaluation of contribution to disparity in preterm birth among African-American women. To that end, it was important to use a liberal cutoff to reduce the number of genes for genotyping and to maximize statistical power by excluding candidate genes that are unlikely to contribute to the disparity in risk by virtue of having insufficient evolution since the divergence of the parental populations in question. Signatures of accelerated evolution detected in preterm birth candidate genes and genomic regions identified by admixture mapping may indicate natural selection (either ecological selection or sexual selection), genetic drift, or the effects of demographic factors such as population expansion. While it is unlikely that natural selection has acted

143

to increase disease phenotypes like preterm birth, it is possible that the role of these candidate genes in other important biological functions has been the target of natural selection. It could also be postulated that natural selection has acted to favor preterm birth in the African environment where there is limited medical intervention to deliver neonates that are too large to pass through the pelvis, a situation that could reduce the fitness of the mother by interfering with future fertility or by causing death. Additionally, selection on a complex phenotype that was advantageous in the past African environment may be maladaptive in an industrialized environment like the U.S. with different environmental pressures. This idea has been proposed to explain the increased risk of metabolic disorders like diabetes and obesity among individuals of Native American ancestry living in the U.S. today. Adaptive strategies that helped prevent starvation in the past are detrimental to populations who have continual access to high calorie foods. To further evaluate the large proportion of candidate genes found to have signatures of accelerated evolution, future work is planned to test similar size sets of randomly chosen genes. It is possible that similar proportions of genes in the randomly selected sets will be found to contain signatures of accelerated evolution. This outcome might be expected given the relaxed thresholds used to define significant signatures of accelerated evolution used in this analysis. Even if preterm birth genes have a similar patterns of significance to randomly chosen genes, the analysis has still accomplished the goal of drawing attention to the genes that are more likely to contribute to difference in risk and removing from consideration those that could not be responsible for the variation in preterm birth observed between populations. However, if randomly selected sets of

144

genes reveal smaller proportions of significant tests, it could be argued that preterm birth candidate genes have experienced greater average levels of accelerated evolution.

In

humans, primates, and most other organisms, research to identify signatures of natural selection routinely point to genes involved in reproduction as producing among the strongest signals. Therefore, it may not be a surprise to find enrichment for signatures of accelerated evolution in preterm birth candidate genes given their role in the biology of reproduction. In this case, a more stringent cut-off might be warranted to identify the most important and probable candidate genes contributing to increased risk of preterm birth among African-American women. Although not considered in the current study, recent evolution among AfricanAmericans could be a source of variation seen in disease risk in the United States. An attempt to identify evidence of natural selection in African-Americans was recently reported by Jin et al. (2012) [32].

They postulated that the extreme change in

environment and exposure to new pathogens faced by West Africans who arrived in the Americas as slaves would result in selection that could be detected in modern AfricanAmerican individuals. Their study revealed six regions of the genome that have excess of either West African or European ancestry and fourteen regions that are highly differentiated between African-Americans and the Yoruba West African parental population. These regions of natural selection may reveal interesting candidate genes for a variety of diseases where there is increased risk among African-Americans. However, in the current study, none of the admixture mapping peaks or the preterm birth candidate genes with evidence of accelerated evolution overlap with the genomic regions of natural selection among African-Americans reported by Jin et al (2012).

145

Tests of accelerated evolution have identified genes that are known to contribute to variation in phenotypes across human populations, especially those that are related to adaptation to varying environments and pathogens, like skin pigmentation change or resistance to malaria. It is unlikely that positive selection is acting directly on preterm birth to generate the demonstrated population differences in susceptibility. However, genes contributing to preterm birth may have experienced accelerated evolution due to their role in other phenotypes (pleiotropy). Genes that contribute to differences in disease risk may have evidence of a population-specific evolutionary history that can be exploited to prioritize genotyping efforts. Using the results of the current study will complement planned genotyping strategies for identifying genes that contribute to disparities in risk of preterm birth and preterm premature rupture of membranes.

Future Directions

As described in Chapter 3, a replication of the admixture mapping results is planned with the possible use of the newly available Affymetrix (Santa Clara, California) Axiom exom array [33]. In addition to the 318,000 markers included on the array, there is the option to include up to 100,000 additional customer-specified markers to the genotyping platform.

With the information provided by the screens of accelerated

evolution conducted in this study, SNPs in five previously reported candidate genes for PPROM were found to have significant evidence of evolution in West Africans or

Europeans.

146

These genes, SERPINH1, IL1B, TIMP2, PLAT, and LTA, will also be

prioritized in the replication effort and included on the custom portion of the exom array. Using the results of the current study, another area of future research could include the development of an accelerated evolution nominated panel of candidate genes to facilitate the investigation of the genetic contributions to the disparity of preterm birth risk among African-American women.

Literature Cited

1.

2. 3. 4. 5. 6. 7. 8.

9. 10. 11.

Wang, H., et al., A functional SNP in the promoter of the SERPINH1 gene increases risk of preterm premature rupture of membranes in African Americans. Proc Natl Acad Sci U S A, 2006. 103(36): p. 13463-7. Jorde, L.B., M. Bamshad, and A.R. Rogers, Using mitochondrial and nuclear DNA markers to reconstruct human evolution. Bioessays, 1998. 20(2): p. 126-36. Cann, R.L., M. Stoneking, and A.C. Wilson, Mitochondrial DNA and human evolution. Nature, 1987. 325(6099): p. 31-6. Quillen, E.E., et al., OPRM1 and EGFR contribute to skin pigmentation differences between Indigenous Americans and Europeans. Hum Genet, 2011. Sabeti, P.C., et al., Positive natural selection in the human lineage. Science, 2006. 312(5780): p. 1614-20. Tishkoff, S.A., et al., Convergent adaptation of human lactase persistence in Africa and Europe. Nat Genet, 2007. 39(1): p. 31-40. Shriver, M.D., et al., The genomic distribution of population substructure in four populations using 8,525 autosomal SNPs. Hum Genomics, 2004. 1(4): p. 274-86. Schlotterer, C. and D. Dieringer, A Novel Test Statistic for the Identification of Local Selective Sweeps Based on Microsatellite Gene Diversity, in Selective Sweep. 2005. p. 55-64. Tajima, F., Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics, 1989. 123(3): p. 585-95. Parry, S. and J.F. Strauss, 3rd, Premature rupture of the fetal membranes. N Engl J Med, 1998. 338(10): p. 663-70. Illumina. Illumina African American Admixture Panel. 2010. http://www.illumina.com/products/african_american_admixture_panel.ilmn. February 29, 2012.

12. 13. 14. 15. 16. 17.

18. 19.

20. 21. 22.

23. 24. 25. 26. 27. 28. 29.

30.

31.

147

Patterson, N., et al., Methods for high-density admixture mapping of disease genes. Am J Hum Genet, 2004. 74(5): p. 979-1000. Smith, M.W., et al., A high-density admixture map for disease gene discovery in african americans. Am J Hum Genet, 2004. 74(5): p. 1001-13. Parra, E.J., et al., Estimating African American admixture proportions by use of population-specific alleles. Am J Hum Genet, 1998. 63(6): p. 1839-51. Hoggart, C.J., et al., Control of confounding of genetic associations in stratified populations. Am J Hum Genet, 2003. 72(6): p. 1492-1504. Hoggart, C.J., et al., Design and analysis of admixture mapping studies. Am J Hum Genet, 2004. 74(5): p. 965-78. Chakraborty, R. and K.M. Weiss, Admixture as a tool for finding linked genes and detecting that difference from allelic association between loci. Proc Natl Acad Sci U S A, 1988. 85(23): p. 9119-23. PTBGene. cited Access Acess Date. Avaiable from: http://bioinformatics.aecom.yu.edu/ptbgene/index.htmlI. Behrman, R.E., A.S. Butler, and Institute of Medicine (U.S.). Committee on Understanding Premature Birth and Assuring Healthy Outcomes., Preterm birth : causes, consequences, and prevention. 2007, Washington, D.C.: National Academies Press. xvi, 772 p. Kent, W.J., et al., The human genome browser at UCSC. Genome Res, 2002. 12(6): p. 996-1006. The International HapMap Project. Nature, 2003. 426(6968): p. 789-96. Affymetrix. Genome-Wide Human SNP Array 6.0 Data Sheet. http://www.affymetrix.com/support/technical/datasheets/genomewide_snp6_datas heet.pdf. April 2, 2012. Wright, S., Genetical structure of populations. Nature, 1950. 166(4215): p. 247-9. Weir, B.S. and C.C. Cockerham, Estimating F-Statistics for the Analysis os Population Sturcture. Evolution, 1984. 38(6): p. 13. Schlotterer, C., A microsatellite-based multilocus screen for the identification of local selective sweeps. Genetics, 2002. 160(2): p. 753-63. Carlson, C.S., et al., Genomic regions exhibiting positive selection identified from dense genotype data. Genome Res, 2005. 15(11): p. 1553-65. Velez, D.R., et al., Preterm birth in Caucasians is associated with coagulation and inflammation pathway gene variants. PLoS One, 2008. 3(9): p. e3283. Gibson, C.S., et al., Genetic polymorphisms and spontaneous preterm birth. Obstet Gynecol, 2007. 109(2 Pt 1): p. 384-91. Wang, H., et al., Genetic and epigenetic mechanisms combine to control MMP1 expression and its association with preterm premature rupture of membranes. Hum Mol Genet, 2008. 17(8): p. 1087-96. Bitner, A. and J. Kalinka, IL-1beta, IL-6 promoter, TNF-alpha promoter and IL1RA gene polymorphisms and the risk of preterm delivery due to preterm premature rupture of membranes in a population of Polish women. Arch Med Sci, 2010. 6(4): p. 552-7. Romero, R., et al., A genetic association study of maternal and fetal candidate genes that predispose to preterm prelabor rupture of membranes (PROM). Am J Obstet Gynecol, 2010. 203(4): p. 361 e1-361 e30.

32. 33.

148

Jin, W., et al., Genome-wide detection of natural selection in African Americans pre- and post-admixture. Genome Res, 2012. 22(3): p. 519-27. Affymetrix. Axiom Exome Genotyping Array Data Sheet. http://www.affymetrix.com/support/technical/datasheets/axiom_exome_array_dat asheet.pdf. March 12, 2012.

149

Chapter 5 Concluding Remarks

The purpose of this research was to investigate the disparity that exists in pregnancy outcomes in African-American women compared to other U.S. populations. African-American women have a significantly increased risk of experiencing preterm birth and their neonates are more likely to be low birth weight for gestational age and have higher mortality. The results of a considerable amount of research focused on identifying the causes of disparity in pregnancy outcomes for African-American women suggest that there are likely environmental, social, and genetic factors that contribute. The goal of the research presented in the previous three chapters was to gain better understanding of genetic and gene-environment interactions that contribute to disparity among African-Americans. The results of this research are reviewed below and future directions are discussed.

Relationship between Genomic Ancestry and Pregnancy-related Phenotypes

To date, very few studies of disparities in pregnancy related phenotypes have addressed genomic ancestry contributions to admixture in African-Americans as potential risk factors.

Unlike most European-American populations, a significant degree of

variation exists in European genomic contribution to admixture within and among African-American populations and across the United States [1].

Classification of

150

individuals in studies of health disparities by “race” without accounting for variation in genomic West African admixture assumes a degree of biological homogeneity within the African-American community that is inaccurate. To better understand how genomic ancestry contributes to phenotypic variation genomic ancestry can be estimated using genotype data from modest panels of ancestry informative markers and tested for association with variation in the phenotype of interest.

For example, greater West

African ancestry is associated with increased skin pigmentation, lower vitamin D level, and recently has been reported to be associated with increased risk of preterm birth [2-5]. The research presented in Chapter 2 evaluates the relationship between genomic ancestry and two pregnancy related phenotypes, low birth weight and serum vitamin D level (a suspected contributor to preterm birth risk). Using pregnant women of selfreported European-American and African-American ancestry living in Western Pennsylvania a considerable degree of variation in West African genomic ancestry was seen among African-American women (range 15%-100%, mean = 20%) that was not observed in the European-American women.

West African genomic ancestry was

correlated with birth weight adjusted for gestational age for female neonates (p2.5, p2.5, p