THE ROLE OF PPAR-γ IN TROPHOBLAST LAYER DYNAMICS

Based on a Cell Model of Trophoblast Differentiation

by

Khrystyna Levytska

A thesis submitted in conformity with the requirements

for the degree of Master of Science

Graduate Department of Laboratory Medicine and Pathobiology

University of Toronto

©Copyright by Khrystyna Levytska (2013) ABSTRACT

THE ROLE OF PPAR-γ IN TROPHOBLAST LAYER DYNAMICS BASED ON A CELL MODEL OF

TROPHOBLAST DIFFERENTIATION

Degree: Master of Science

Year of Convocation: 2013

Name: Khrystyna Levytska

Graduate Department: Laboratory Medicine and Pathobiology

University: University of Toronto

Severe pre-eclampsia (sPE) is a pregnancy disorder stemming from the placenta which features abnormal trophoblast layer physiology. A transcription factor, glial cell missing-1 (GCM-1), regulates differentiation of cytotrophoblast cells into syncytiotrophoblast (SCT) within the layer.

GCM-1 is regulated by PPAR-γ, a nuclear receptor controlling trophoblast differentiation in mice.

We have previously shown that activation of PPAR-γ ameliorates disease features in a rat model of sPE via an upregulation of heme oxygenase-1 (HO-1), a cytoprotective enzyme. Here, we assessed how manipulating PPAR-γ influences expression and alters proliferation/differentiation balance in a cell line, BeWo, which mimics SCT formation. PPAR-γ activation induced HO-1 expression and led to a transient upregulation of GCM-1 expression. Blocking PPAR-γ activity increased cell proliferation and decreased differentiation, whereas PPAR-γ stimulation did not change this balance. Since HO-1 and GCM-1 are downregulated in sPE, their upregulation via PPAR-

γ offers insights into a pathway relevant to sPE management.

ii

ACKNOWLEDGEMENTS

I would like to thank my supervisor Dr. John CP Kingdom for his guidance, support and contribution to the project design throughout my Master’s degree. Furthermore, I would like to put forth my gratitude to my committee members, Dr. Theodore Brown, Dr. Sarah Keating and Dr. S. Lee Adamson, as well as my international advisor, Dr. Asif Ahmed, for their contribution and helpful discussions.

A special thank you goes out to my outstanding mentors, Mrs. Dora Baczyk and Dr. Sascha Drewlo, who were my inspiration figures and role models since the very beginning of my scientific career. Without their support, encouragement and scientific input I would not have been who and where I am now.

Many thanks go out to members of collaborating laboratories, such as Dr. Lye’s lab, for helping me advance my project swiftly and efficiently. I would also like to thank various funding sources, such as Mount Sinai Hospital, Faculty of Medicine and School of Graduate Studies at the University of Toronto, for providing me with the opportunity and resources necessary for the completion of my degree.

Most importantly, I would like to thank my dear friends and family for their help, understanding and never-ceasing support throughout my graduate career and always.

iii

TABLE OF CONTENTS

ABSTRACT ...... ii ACKNOWLEDGEMENTS ...... iii TABLE OF CONTENTS ...... iv LIST OF TABLES ...... vii LIST OF FIGURES ...... viii LIST OF ABBREVIATIONS ...... x LIST OF APPENDICES ...... xii INTRODUCTION ...... 1 1.1. Healthy pregnancy ...... 2 1.2. Hypertensive disorders of pregnancy ...... 2 1.3. Pre-eclampsia (PE) ...... 3 1.3.1. Disease symptoms ...... 3 1.3.2. Current lack of treatment options for women suffering from PE...... 3 1.4. Placenta ...... 4 1.4.1. Function of the healthy placenta ...... 4 1.4.2. Placental trophoblast lineages ...... 4 1.4.2.1. Extravillous trophoblast (EVT) lineage ...... 5 1.4.2.2. Villous cytotrophoblast (VCT) lineage ...... 6 1.4.3. Pathological features of sPE ...... 7 1.5. Glial cell missing-1 (GCM-1) ...... 9 1.5.1. GCM-1: transcription factor controlling differentiation ...... 9 1.5.2. Role of GCM-1 in placentation and trophoblast differentiation ...... 9 1.6. Peroxisome proliferator-activated receptors (PPARs) ...... 11 1.6.1. Family of peroxisome proliferator-activated receptors ...... 11 1.6.2. PPAR-γ gene and structure ...... 12 1.6.3. Role of PPARs in development ...... 14 1.6.4. Regulation of PPAR-γ activity ...... 14 1.6.5. PPAR-γ expression in healthy pregnancy and pregnancy complications ...... 18 1.6.6. Role of PPAR-γ in trophoblast physiology ...... 20 1.6.7. Modulation of PPAR-γ during pregnancy and models of PE ...... 21

iv

1.6.8. PPAR-γ controls a cytoprotective enzyme, heme oxygenase-1 (HO-1) ...... 22 1.7. Heme oxygenase-1 (HO-1) ...... 22 1.7.1. HO-1 expression during pregnancy ...... 23 1.7.2. Role of HO-1 in placentation ...... 24 1.7.3. Studies of HO-1 in cell-based models...... 26 1.7.4. Molecular manipulation of HO-1 ...... 27 1.8. Rationale of the project ...... 28 1.9. Models used in this study ...... 28 1.10. Hypotheses and specific aims ...... 29 MATERIALS AND METHODS ...... 30 2.1. Mouse model ...... 31 2.1.1. Tissue collection ...... 31 2.1.2. RNA extraction, reverse transcription and qRT-PCR ...... 31 2.2. In vitro model ...... 33 2.2.1. Evaluation of cell models ...... 33 2.2.2. The BeWo cell line ...... 33 2.2.3. Drug treatments of BeWo cells ...... 34 2.2.4. Transfection of cells ...... 36 2.2.5. RNA analysis ...... 37 2.2.5.1. Primer design ...... 37 2.2.5.2. Housekeeping gene stability check ...... 38 2.2.5.3. RNA extraction ...... 39 2.2.5.4. Reverse transcription and quantitative real time-PCR (qRT-PCR) ...... 39 2.2.6. Protein analysis ...... 39 2.2.6.1. Protein isolation ...... 39 2.2.6.2. Western blotting ...... 40 2.2.6.3. Enzyme-linked immunosorbent assay (ELISA) ...... 41 2.2.6.3.1. Free β-hCG ELISA ...... 41 2.2.6.3.2. HO-1 ELISA ...... 41 2.2.6.4. Fluorescent immunohistochemistry (F-IHC) ...... 42 2.2.6.4.1. Validation of E-cadherin as a cell membrane marker ...... 43 2.2.7. Luciferase assay ...... 43

v

2.2.8. Cell proliferation assay ...... 45 2.2.9. Cellular fractionation analysis ...... 46 2.2.10. Statistical analysis ...... 46 RESULTS ...... 47 3.1. Placental in Gcm-1 heterozygous mice ...... 48 3.2. GCM-1 silencing does not affect HO-1 and PPAR-γ expression in the BeWo cell model ...... 49 3.3. Time- and dose- responses to drug treatments in BeWo cells ...... 50 3.3.1. Expression of HO-1 ...... 50 3.3.2. Expression of PPAR-γ ...... 51 3.3.3. Expression of GCM-1...... 52 3.4. Effect of PPAR-γ activity-modulating drugs on HO-1 expression in BeWo cells ...... 53 3.4.1. PPAR-γ activation induces a rapid rise in HO-1 RNA levels ...... 54 3.4.2. Intracellular HO-1 protein expression increases following PPAR-γ activation ...... 54 3.4.3. HO-1 release from BeWo cells increases with PPAR-γ activation ...... 56 3.5. PPAR-γ participates in a negative auto-regulatory feedback mechanism ...... 57 3.5.1. PPAR-γ expression is inversely related to PPAR-γ activity ...... 57 3.5.2. PPAR-γ protein levels correlate with gene expression profiles ...... 58 3.5.3. Phosphorylation of PPAR-γ protein is increased following rosiglitazone treatment .. 59 3.5.4. Fluorescent immunohistochemistry analysis of PPAR-γ localization ...... 60 3.5.5. Nuclear and cytoplasmic (phospho-) PPAR-γ expression...... 61 3.6. Role of PPAR-γ in proliferation/differentiation balance ...... 63 3.6.1. PPAR-γ-mediated induction of GCM-1 expression is transient ...... 63 3.6.2. PPAR-γ activity in BeWo cells ...... 64 3.6.3. Effect of PPAR-γ activity modulation on free β-hCG release ...... 65 3.6.4. Visualization of BeWo cell fusion using fluorescent immunohistochemistry ...... 66 3.6.5. Effect of PPAR-γ activity on BeWo cell proliferation ...... 67 DISCUSSION ...... 68 4.1 Conclusion ...... 76 FUTURE DIRECTIONS ...... 77 REFERENCES ...... 81 APPENDIX ...... 90

vi

LIST OF TABLES

Chapter 1: Introduction

Table 1: Patient characteristics ...... 19

Chapter 2: Materials and Methods

Table 2: Mouse primers ...... 32 Table 3: Drugs and concentrations ...... 35 Table 4: Human primers ...... 37 Table 5: Antibodies and dilutions ...... 40

vii

LIST OF FIGURES

Chapter 1: Introduction

Figure 1: Schematic representation of a placental-decidual interaction in healthy pregnancy...... 5 Figure 2: Schematic representation of a healthy trophoblast turnover...... 7 Figure 3: Schematic representation of a placental-decidual interaction in PE...... 8 Figure 4: Structure of the GCM-1 protein...... 10 Figure 5: Schematic representation of the roles of three PPAR family members...... 12 Figure 6: PPAR-γ isoforms and structure...... 13 Figure 7: PPAR-γ ligands...... 15 Figure 8: Transactivation model of PPAR-γ activity...... 16 Figure 9: Post-translational modifications of PPAR-γ...... 17 Figure 10: Placental expression of PPAR-γ in pathologies and controls...... 19 Figure 11: Schematic representation of heme breakdown by HO-1...... 23

Chapter 2: Materials and Methods

Figure 12: Gene expression in human pregnancy-related cell lines...... 33 Figure 13: Preliminary assessment of gene expression in BeWo cells across passages...... 34 Figure 14: Effects of drugs on cell viability over time...... 35 Figure 15: Location of siRNA sequences in the human GCM-1 mRNA...... 36 Figure 16: Stability of individual housekeeping in BeWo cells under different treatments. .... 38 Figure 17: E-cadherin expression in response to different drug treatments...... 43 Figure 18: Gene map of the 1kb upstream region of the human GCM-1...... 44 Figure 19: Plasmid design and oligonucleotide sequences in the luciferase reporter assay...... 44

Chapter 3: Results

Figure 20: Placental Hmox-1, Hmox-2 or Ppar-γ expression is unaltered in Gcm-1+/- mice...... 48 Figure 21: Downregulation of GCM-1 in BeWo cells has no effect on HO-1 and PPAR-γ levels...... 49 Figure 22: Dose-dependent expression of HO-1 in BeWo cells over time...... 51 Figure 23: Dose-dependent expression of PPAR-γ in BeWo cells over time...... 52 Figure 24: Dose-dependent expression of GCM-1 in BeWo cells over time...... 53 Figure 25: HO-1 mRNA expression levels following modulation of PPAR-γ activity...... 54 Figure 26: Expression of HO-1 protein in BeWo cells...... 55 Figure 27: HO-1 release from BeWo cells...... 56 Figure 28: PPAR-γ mRNA expression levels following modulation of PPAR-γ activity...... 57 Figure 29: Expression of PPAR-γ protein in BeWo cells...... 58 Figure 30: Expression of phosphorylated PPAR-γ protein in BeWo cells...... 59 Figure 31: Effect of PPAR-γ-modulating drugs on its localization in BeWo cells...... 60

viii

Figure 32: Cellular fractionation analysis of PPAR-γ expression in BeWo cells...... 61 Figure 33: Nuclear phosphorylated PPAR-γ expression in BeWo cells...... 62 Figure 34: GCM-1 mRNA expression levels following modulation of PPAR-γ activity...... 63 Figure 35: PPAR-γ activity in BeWo cells...... 64 Figure 36: Effect of PPAR-γ manipulation on free β-hCG release in BeWo cells...... 65 Figure 37: Effect of PPAR-γ modulation on BeWo cell fusion...... 66 Figure 38: Effect of PPAR-γ activity modulation on BeWo cell proliferation...... 67

ix

LIST OF ABBREVIATIONS

Abbreviation Meaning AF-1 Activation function-1 AF-2 Activation function-2 CO Carbon monoxide CT Cytotrophoblast DBD DNA-binding domain DR-1 Direct repeat-1 ERK Extracellular-signal-regulated kinase EVT Extravillous trophoblast GCM-1 Glial cell missing-1 hCG Human chorionic gonadotrophin HDACs Histone deacetylases HELLP Hemolysis, elevated liver enzymes and low platelets HMOX-1 Gene coding for heme oxygenase-1 HO-1 Heme oxygenase-1 HO-2 Heme oxygenase-2 HO-3 Heme oxygenase-3 HTR8 Extra-villous trophoblast cell line HUVECs Human umbilical vein endothelial cells IUGR Intra-uterine growth restriction JEG-3 Human choriocarcinoma cell line LBD Ligand-binding domain LMWH Low molecular weight heparin Nrf-2 Nuclear factor (erythroid-derived 2)-like 2 PE Pre-eclampsia PlGF Placenta-like growth factor PPAR-α Peroxisome proliferator-activated receptor-alpha PPAR-β/δ Peroxisome proliferator-activated receptor-beta (or delta) PPAR-γ Peroxisome proliferator-activated receptor-gamma PPRE PPAR-γ response element RUPP Reduced uterine perfusion pressure RXR Retinoid X receptor

x sFLT-1 Soluble fms-like tyrosine kinase-1 SA Spontaneous abortion sENG Soluble endoglin sIUGR Severe intra-uterine growth restriction SP Spongiotrophoblast sPE Severe pre-eclampsia TIMP Tissue inhibitor of metalloproteinase Treg Regulatory T TS Trophoblast stem uNK Uterine natural killer VEGF Vascular endothelial growth factor VCT Villous cytotrophoblast cell WT Wildtype

xi

LIST OF APPENDICES

Chapter 1: Introduction 1. Drewlo, S., K. Levytska, and J. Kingdom, Revisiting the housekeeping genes of human placental development and insufficiency syndromes. Placenta, 2012. 33(11): p. 952-4. 2. Levytska, K., et al., Heme oxygenase-1 in placental development and pathology. Placenta, 2013. 34(4): p. 291-8.

xii

Chapter 1: INTRODUCTION

1

Successful pregnancy requires several adaptations by the mother aimed to ensure normal fetal development and maternal tolerance of pregnancy. The placenta is a central player; its proper function contributes to successful implantation and subsequent embryonic and fetal growth. The epithelial trophoblast layer covering the placental villi is in direct contact with maternal blood and is central to both maternal adaptation and fetal development. As will be outlined in this chapter, molecular dysregulation of the trophoblast lineages may lead to perinatal complications with adverse outcomes for both mother and baby.

1.1. Healthy pregnancy

Successful human pregnancy is accompanied by hemodynamic changes in the mother including maternal cardiac output and blood volume increase by 30-45% [1, 2]. These changes, accompanied by vasodilation of the uteroplacental arteries, produce an exponential rise in utero- placental blood flow to support the demands of the growing fetus. Higher cardiac output and blood volume would predict a significant elevation in blood pressure were it not for the substantial reduction in systemic vascular resistance that occurs in the second trimester.

1.2. Hypertensive disorders of pregnancy

In the event that maternal hemodynamic adaptations are defective, pregnant women develop hypertension which may co-exist with other conditions. Hypertensive disorders are a group of the most common complications during pregnancy, affecting 5-10% of all pregnancies [3]. These disorders include chronic hypertension, gestational hypertension, pre-eclampsia and chronic hypertension with superimposed pre-eclampsia. Pre-eclampsia, which affects 3-5% of pregnancies worldwide, is a major contributor to maternal, fetal and infant mortality and morbidity due to its downstream effects on the mother’s wellbeing and neonatal development [4].

2

1.3. Pre-eclampsia (PE)

1.3.1. Disease symptoms

Severe early-onset pre-eclampsia (sPE) manifests clinically in the second trimester (after 20 weeks) of gestation. Its key features are hypertension (>140/90 mmHg) and proteinuria (>0.3 g of protein excreted in urine within 24 hours), possibly complicated by edema [5]. sPE is a syndrome affecting multiple organ systems, namely the placenta, the kidneys, the liver, and the brain [6]. The feto-maternal interface is thought to suffer from hypoxia and/or reperfusion injury; it exhibits several morphological abnormalities including impaired placental development and differentiation, placental infarcts, and signs of non-infarct-related maternal vascular underperfusion [7]. Endothelial dysfunction in the kidneys (glomerular endotheliosis) results in high protein levels in the urine. Hepatic ischemia, another hallmark of the syndrome, may lead to the manifestation of the hemolysis, elevated liver enzymes and low platelets (HELLP) syndrome. Lastly, if PE progresses to the most serious stage, it develops into eclampsia. Eclampsia, complicated by cerebral ischemia, poses the largest health risk for the mother and may lead to migraines, seizures, coma, and death. To date, there is no cure for PE other than delivery of the placenta, which in most severe cases necessitates delivery of the fetus prior to 34 weeks of gestation. Several factors have been associated with increased incidence of sPE, such as advanced maternal age at first pregnancy, use of assisted reproductive technologies, obesity and others [4]. Considering that these demographic trends are on the rise, there is a pressing need for extensive research into disease etiology and pathogenesis to alleviate the associated health care burden and improve maternal and neonatal conditions post-partum.

1.3.2. Current lack of treatment options for women suffering from PE

As mentioned above, the only current effective treatment option for sPE is delivery of the placenta, which has severe long-term morbidities for the mother and the newborn thereby increasing the risk of neonatal mortality. Several treatment options have been examined for women suffering from or at risk of developing PE. In most complicated sPE cases, women receive magnesium sulfate treatment at the onset of labor and for the initial 24 hours post-partum, where the drug is thought to contribute to vasorelaxation and prevent eclamptic seizures [8]. Several trials have examined the possibility of managing sPE symptoms with mineral or vitamin supplementation, some with little success. For example, calcium supplementation was shown to be effective in preventing sPE symptoms possibly due to its vasodilatory effects [9], while vitamins C and E and fish oil supplements did not change the risk of developing the disease [10, 11].

3

Low-dose aspirin (commenced prior to 16 weeks of gestation) was shown to decrease severe but not mild PE incidence by inducing vasorelaxation and preventing excessive placental coagulation; however, it was not able to decrease rates of maternal mortality [12]. More recently, the use of low molecular weight heparin (LMWH) has been shown to have positive outcomes for pregnancies complicated by placental insufficiencies via anti-coagulation independent mechanisms [13-16]. Despite some of these studies showing a decrease in the occurrence of sPE, none of these interventions have shown an improvement in function of the organ thought to be a key player in disease development, namely, the placenta.

1.4. Placenta

1.4.1. Function of the healthy placenta

The placenta is a transient organ present only during pregnancy and its main function is the maintenance of a nourishing environment for the developing fetus. undergo hemochorial placentation where maternal blood is separated from the fetus by the chorion, ensuring an efficient transfer of nutrients and waste. The placenta has several functions, including fetal nutrition, respiration, waste excretion and hormone secretion [17, 18]. The placenta is responsible for both active and passive transport of glucose, amino acids and other nutrients to sustain metabolic demands of the growing fetus. It also mediates the transport of oxygen into, and removal of carbon dioxide out of, the fetal circulation. Furthermore, it acts to eliminate waste (such as urea) from the fetal environment [19]. Importantly, the placenta becomes an endocrine organ which is capable of synthesizing glycogen, cholesterol, fatty acids, and hormones (such as estrogen, progesterone, placenta-like growth factor [PlGF] and human chorionic gonadotropin [hCG]) [20].

1.4.2. Placental trophoblast lineages

The key component of the placenta is the trophoblast, responsible for the establishment and maintenance of proper placental physiology throughout gestation and towards term. Two trophoblast lineages exist during human placentation, each with is specific role in proper pregnancy progression and maintenance [20]. The first cell type is the extravillous trophoblast (EVT) responsible for placental invasion into the decidua, spiral artery remodelling and the suppression of maternal immune response. The second cell type is the villous cytotrophoblast (VCT); it provides the epithelial lining of floating placental villi which project into the intervillous space (Figure 1).

4

Figure 1: Schematic representation of a placental-decidual interaction in healthy pregnancy. Extravillous cytotrophoblasts from the anchoring villous columns proliferate and invade the maternal decidua and contribute to spiral artery remodeling. This results in high blood flow from maternal arteries to the intervillous space, ensuring low pressure blood flow and delivery of oxygen to the growing fetus. Placental villi are covered by the trophoblast layer composed of two compartments, the underlying cytotrophoblast cells which replenish the overlying syncytiotrophoblast (appears here with the permission of the publisher from Everett et al, Placenta, 33:893-901, 2012 [21]).

1.4.2.1. Extravillous trophoblast (EVT) lineage

The EVT cells, originating from the columns of the anchoring villi, proliferate and invade the external layer of the uterus, the decidua [20]. EVT cells at the base of the anchoring villus column are of the proliferative phenotype, while those more distal of the column acquire the invasive phenotype and are divided into interstitial and endovascular subtypes. Interstitial subtype invades the maternal decidua; the endovascular subtype migrates toward the lumina of maternal spiral arteries and displaces the endothelial and smooth muscle lining in arterial walls, thus, contributing to blood vessel enlargement and vasorelaxation. Proper invasion of endovascular EVTs is instrumental in the establishment of a high flow/low pressure perfusion in the intervillous space, ensuring minimal shear stress on the villus surface and maximizing delivery of nutrients and oxygen to the fetus [22]. These cells invade maternal arteries as early as 4-6 weeks of gestation, with a marked increase in invasion seen between 10 and 12 weeks [23]; before 8 weeks of gestation, these cells occlude uterine capillaries to limit blood oxygen delivery to the placenta and the fetus which do not yet possess defence mechanisms against oxidative stress [24]. This hypoxia, in turn, serves as a trigger for placental angiogenesis and villous tree branching.

5

To adequately invade the maternal tissue, EVTs must undergo several changes. For example, they must alter their expression of integrins and matrix metalloproteinases (MMPs) in order to co-ordinate decidual degradation and cell migration within maternal tissue [25]. Of note, it has been shown that a tissue inhibitor of metalloproteinase-4 (TIMP-4) is upregulated in sPE placentas, suggesting that this molecule contributes to extracellular matrix degradation in normal pregnancy [26]. Furthermore, invading EVTs must communicate with decidual immune cells (primarily uterine natural killer (uNK) cells); this interaction is critical for proper placental establishment especially during early stages of development (reviewed in [27]).

1.4.2.2. Villous cytotrophoblast (VCT) lineage

The second trophoblast lineage is the villous cytotrophoblast (VCT) which forms a continuous layer covering floating villi and is in direct contact with the maternal blood, creating the interface between maternal and fetal circulations [20]. The trophoblast layer is composed of two compartments: the cytotrophoblast (CT) cells and the syncytiotrophoblast (SCT), or syncytium (Figure 2) [28]. The layer grows throughout gestation to accommodate expansion of the villous surface area which acts to maximize net nutrient and waste exchange across the placenta. The CT cell population is heterogeneous: some cells retain their stem-like potential and divide symmetrically in order to replenish the cytotrophoblast pool; others lose their mitotic potential and differentiate. The plasma membrane of post-mitotic differentiated cells disintegrates and their nuclei fuse with the overlying syncytium layer. Subsequently, as evidence suggests, nuclei which have differentiated apoptose as they aggregate to form syncytial knots which are then shed into the maternal blood stream (as much as 3g per day in healthy pregnancy) [29, 30]. The process of proliferation, differentiation, fusion, apoptosis and shedding within the trophoblast layer is referred to as the trophoblast turnover. It is thought that abnormal trophoblast turnover contributes to placental insufficiency syndromes such as sPE and IUGR (Widdows et al, in preparation).

6

Figure 2: Schematic representation of a healthy trophoblast turnover. Stem (progenitor) cytotrophoblast cells undergo two types of divisions: symmetric and asymmetric. Symmetric divisions produce identical daughter cells and maintain the pool of cytotrophoblast cells. Asymmetric divisions, orchestrated by factors such as the differentiation marker glial cell missing-1, produce one daughter cell with fusogenic potential. This cell moves on to differentiate, and fuse with and donate its nucleus to the overlying syncytiotrophoblast layer. Nuclei within the syncytium subsequently form syncytial knots, undergo apoptosis and shed into the maternal blood stream. The process of cytotrophoblast proliferation, differentiation and fusion into the overlying syncytium is termed as the trophoblast turnover (appears here with the permission of the publisher from Lanoix et al, Molecular and Cellular Endocrinology, 348: 1-11, 2012 [31]).

1.4.3. Pathological features of sPE

Key features of a proper placental establishment and development are defective in placental insufficiency syndromes, such as sPE and intrauterine growth restriction (IUGR). sPE is thought to be a two stage disease: 1) pre-clinical and 2) clinical. The current hypothesis holds that in early stages of sPE, pregnancy is characterized by poor placentation where the infiltration of EVTs into the maternal decidua, and subsequent remodeling of spiral arteries, are inadequate (Figure 3) [32, 33]. The resulting oxidative and nitrosative stress at the placental bed [34, 35] possibly results due to hypoxia or ischemia-reperfusion injury following reduced or intermittent uteroplacental blood flow, respectively [36]. At this stage of disease development, the effects are local and restricted to the feto-maternal interface while the mother remains asymptomatic. By contrast, the second stage of the disease is thought to be characterized by a systemic maternal response leading to a clinical presentation of sPE. While hypoxia and oxidative stress are important contributors to disease pathology, another factor is an exaggerated maternal immune response seen by elevated pro-inflammatory cytokines found in sera of mothers suffering from sPE [37]. The involvement of both hypoxia and inflammation in pathological placental responses has been supported by several findings. For example, the expression of a marker of hypoxia, hypoxia- inducible factor 1-α (HIF1-α), in placental explants increases in low oxygen conditions (reviewed in

7

[38]); interestingly, however, stimulation of explants with angiotensin II (Ang II), a pro- inflammatory peptide, under normoxic conditions has the same effect, in addition to decreasing EVT differentiation [39]. Furthermore, hypoxia induces the release of anti-angiogenic factors (described below) into culture medium from placental explants [40, 41], an effect which is analogous to that seen following stimulaiton of tissue with pro-inflammatory cytokines in normoxia [42, 43]. These findings support the notion that hypoxia is one of several causes of systemic vascular dysfunction, and is most likely complicated by other factors such as wide-spread inflammation and impaired underlying molecular mechanisms. On a microscopic level, villi from PE placentas exhibit excessive formation of structures called syncytial knots which are shed into the maternal blood stream and contribute to wide-spread endothelial cell damage. Interestingly, these syncytial knots are capable of protein synthesis possibly contributing to high serum levels of anti-angiogenic factors in sPE [44-46]. One such factor has been identified to be soluble fms-like tyrosine kinase-1 (sFLT-1), highly upregulated in maternal serum even prior to the development of PE symptoms [47]. sFLT-1, the soluble variant of the membrane-bound vascular endothelial growth factor (VEGF) receptor, acts as a decoy protein for VEGF and PlGF; upon sFLT-1 binding to these factors, their downstream angiogenic functions on the endothelium are disrupted [48]. sFLT-1, along with other anti-angiogenic factors, contributes to maternal vasoconstriction, vascular dysfunction and overt hypertension.

Figure 3: Schematic representation of a placental-decidual interaction in PE. Invasion by the extravillous trophoblasts is shallow leading to the lack of spiral artery remodeling. This is believed to result in unsteady blood flow into the intervillous space, hypoxia or hypoxia- reperfusion injury and detrimental downstream effects on the trophoblast compartment. These contribute to local infarcts, tissue necrosis and excessive shedding of microparticles into maternal bloodstream (appears here with the permission of the publisher from Everett et al, Placenta, 33:893-901, 2012 [21]).

8

As has been discussed thus far, there is evidence implicating that dysregulation of the trophoblast compartment contributes to the pathology seen in PE, leading to the overt manifestation of symptoms, such as systemic endothelial cell damage, hypertentsion, and kidney damage. Several factors contribute to pregnancy maintenance and proper placentation, including glial cell missing-1, peroxisome proliferator-activated receptor-gamma, and heme oxygenase-1. The functions of these have been shown to be altered in pathological placental tissue. Each of these factors and the roles they play in placentation and PE pathophysiology will be discussed in detail in the following sections.

1.5. Glial cell missing-1 (GCM-1)

1.5.1. GCM-1: transcription factor controlling differentiation

Control of the trophoblast turnover has been studied extensively in recent years and several factors have been identified in this physiological process. One of such factors is glial cell missing-1 (Gcm-1), first discovered in where it was shown to regulate neuroblast fate by inducing differentiation of neural precursors into glial cells [49]. Further studies have characterized Gcm-1 as a member of the zinc-containing transcription factor family, possessing a nuclear localization signal, a transactivation domain and a DNA-binding domain (Figure 4A) [50, 51].

1.5.2. Role of GCM-1 in placentation and trophoblast differentiation

The human GCM-1 gene has been localized to 6 and genetic studies have shown that its peptide sequence shares homology across human and rodent species (Figure 4B) [52]. Although first discovered in the brain, GCM-1 is prominently expressed in a few non-neuronal tissues, including the placenta. The crucial role of Gcm-1 in proper placentation in mice was shown by Anson-Cartwright and colleagues in 2000 [53]. First, the group showed that in a healthy placenta, Gcm-1 is localized in a small subset of cells which corresponded to future branch points within the chorioallantoic plate. By creating a double Gcm-1 knock-out mouse, they illustrated that Gcm-1 is responsible for proper branching of the chorioallantoic surface and development of the labyrinth area in the placenta (equivalent of the exchange zone in the human placenta). Due to a defect in branching, mice lacking Gcm-1 died in utero before E10.5. Upon examination, Gcm-1-/- placentas showed lack of differentiated trophoblast cells and did not show signs of cell fusion [53]. Other studies in mice have also identified Gcm-1 as a crucial factor in chorioallantoic interaction during placental development [54].

9

Figure 4: Structure of the GCM-1 protein. (A) Diagrammatic representation of the GCM-1 protein, which contains the general transcription factor domains: DNA-binding domain, nuclear localization signal and transactivation domain. (B) Mouse and human GCM-1 peptide sequences exhibit close homology with highest homology seen in the DNA-binding domain. Dark- and light- shaded regions correspond to identical and similar amino acids, respectively (appears here with the permission of the publisher and adapted from Nait-Oumesmar et al, The Journal of Histochemistry & Cytochemistry, 48(7): 915-922, 2000 [52]).

Studies localizing GCM-1 expression within the human placenta have shown similar expression pattern to the one seen in the mouse [55]. Using in situ hybridization, it was confirmed that GCM-1 expression is restricted to cytotrophoblast subset of cells in the villous trophoblast layer. Furthermore, manipulating GCM-1 expression in the human placenta – downregulating it using siRNA or upregulating its expression with forskolin [56] – has identified GCM-1 as a key regulator of the trophoblast turnover [57]. Downregulation of GCM-1 in placental villi resulted in cytotrophoblast proliferation, whereas forskolin-mediated induction of its expression led to higher levels of differentiation and syncytium formation. Furthermore, GCM-1 was shown to induce differentiation of trophoblast cells along the EVT pathway as assessed by the Matrigel migration assay [57]. Consistent with the role of GCM-1 in trophoblast cell invasion and implantation, as well as in the development of the labyrinth layer in the mouse, evidence shows that GCM-1 protein expression is reduced in placentas with developmental complications, such as in sPE, when

10 compared to age-matched controls [58, 59]. Consequently, a study examining the relationship between Gcm-1 and murine placental morphology confirmed the importance of this transcription factor in proper placental development, specifically villous branching [59]. Using a heterozygous (het) Gcm-1 model (since a double Gcm-1 knockout is lethal in utero), the authors showed that placentas from het conceptuses exhibited higher placental sFlt-1 levels than those from WT mice, as well as abnormalities in the syncytiotrophoblast and thickened inter-hemal membrane separating maternal and fetal circulations, indicating defective maternal-fetal transfer [59]. Complementing the above studies, GCM-1 has been shown to induce the expression of syncytin-1, a fusogenic protein expressed in the human placenta which mediates fusion of cytotrophoblast cells into the overlying syncytiotrophoblast layer [60, 61]. Collectively, data indicate that GCM-1 plays a role in trophoblast turnover and placental morphology and further development.

1.6. Peroxisome proliferator-activated receptors (PPARs)

1.6.1. Family of peroxisome proliferator-activated receptors

Another family of implicated in proper placentation and trophoblast development are peroxisome proliferator-activated receptors (PPARs). PPARs are a family of ligand-activated nuclear receptors and transcription factors. Three members of the PPAR family have been described: PPAR-α, PPAR-β/δ and PPAR-γ (isoforms γ1 and γ2), general functions of which are outlined in Figure 5. Expression of all three members has been assessed in different tissues [62-64]. PPAR-α is mostly expressed in kidney, liver, heart, and muscle cells; PPAR-β/δ is widely expressed, with highest levels found in skin, brain and adipose tissue; and PPAR-γ has been mostly localized to adipocytes, pancreatic β-cells, vascular endothelium, cells of the immune system, and the placenta (reviewed in detail in [65]). The PPAR-γ isoform has been of particular clinical interest, since it was shown to be strongly activated with anti-diabetic glitazones, implying its role as a potential therapeutic target in serious health conditions, such as obesity, metabolic syndrome, and type II diabetes [66]. Initial research on PPARs has shown their prominent role in adipocyte and macrophage differentiation as well as adipogenesis (lipid storage); recent studies have also underlined their functions in processes such as atherosclerosis, cardiovascular disease and inflammation [67].

11

Figure 5: Schematic representation of the roles of three PPAR family members. Roles of PPAR-α, PPAR-β/δ and PPAR-γ in development, inflammation, lipid metabolism and atherosclerosis are described (appears here with the permission of the publisher and adapted from Michalik and Wahli, Current Opinion in Biotechnology, 10: 564-570, 1999 [68]).

1.6.2. PPAR-γ gene and protein structure

Four splice variants of PPAR-γ have been identified: γ1, γ2, γ3 and γ4; when translated, these result in two protein isoforms: PPAR-γ1 and PPAR-γ2 (due to alternative splicing, γ1, γ3 and γ4 all produce the same peptide sequence, PPAR-γ1) (Figure 6A). Mouse, rat and bovine PPAR-γ1 and

PPAR-γ2 proteins are highly conserved (PPAR-γ1 and PPAR-γ2 share 97-98% and 95-96% homology with humans, respectively) [69]. Functional studies of this nuclear receptor have identified the following five domains within its structure: 1) the “A/B” [or activation function-1 (AF-1)] domain which has ligand-independent activity; 2) the highly conserved “C” [or DNA-binding] domain which possesses classical two zinc-finger-like motifs with α-helical elements that bind to the PPAR-γ response element (PPRE); 3) the “D” [or hinge] domain which interacts with the extension site outside of the PPRE; and 4) the “E/F” [or activation function-2 (AF-2)] domain, which is the ligand- binding domain (Figure 6B). Domains D and E/F are involved in dimerization of PPAR-γ with its transactivation partner and cofactor, retinoid X receptor (RXR) [70]. The PPRE response element to which the heterodimer complex binds has been characterized as a direct repeat 1 (DR1) motif consisting of two copies of the consensus AGGTCA sequence separated by one nucleotide (Figure 6C) [71].

12

Figure 6: PPAR-γ isoforms and structure. (A) Diagrammatic representation of PPAR-γ isoforms at the RNA and protein level. Due to alternative splicing, γ1, γ3 and γ4 all produce the same peptide sequence – PPAR-γ1 (a 477 amino acid protein), while γ2 isoform results in a 505 amino acid peptide (appears here with the permission of the publisher and adapted from Zieleniak et al, Arch. Immunol. Ther. Exp., 56: 331– 345, 2008 [72]). (B) Schematic representation of PPAR-γ domains A/B, C, D, and E/F (appears here with the permission of the publisher and adapted from Gelman et al, Current Opinion in Cell Biology, 17:216-222, 2005 [73]). (C) Representation of the PPAR-γ response element (PPRE) which is a direct repeat 1 motif consisting of two repeats of AGGTCA separated by one nucleotide. Below is a computer generated model of how PPAR-γ interacts with its cofactor and binds to the DNA (appears here with the permission of the publisher from Chandra et al, Nature, 456(7220): 350- 356, 2008 [70]).

13

1.6.3. Role of PPARs in development

While initially discovered in their involvement with lipid metabolism and adipogenesis, PPAR-β/δ and PPAR-γ have been shown to be important in placentation and embryogenesis. Although PPAR-δ null pups are viable, these pregnancies are characterized by lower litter sizes when compared to wildtype (WT) matings, primarily due to their developmental abnormalities [74]. To dissect the contributions of fetal vs maternal PPAR-δ expression to pregnancy outcomes, Wang et al performed cross-implantation experiments by introducing WT or null blastocysts into WT or PPAR-δ null mothers. These experiments allowed the group to conclude that maternal PPAR- δ was critical for on-time implantation and decidual angiogenesis, while the embryonic copy was essential for successful placental development (placentas of null pups showed impaired ecto- placental trophoblast development and chorioallantoic fusion) [74]. Contrary to the outcomes from PPAR-δ null pregnancies, double PPAR-γ mutations resulted in embryonic death at E10 [75]. Interestingly, Barak et al found that lethality of these embryos was observed in two stages: an initial defect was seen in trophoblast differentiation and placental vascularization (1), which subsequently contributed to a myocardial defect (2). To prove the connection between placental malformation and improper subsequent heart development, the group performed chimera experiments where PPAR-γ null embryos were supplemented with WT placentas. Researchers found that although the cardiac defect could be rescued, pups surviving to term exhibited other lethal complications and died shortly after birth (within one week). The study reinforced the importance of PPAR-γ in placentation and embryo development.

1.6.4. Regulation of PPAR-γ activity

Being a ligand-activated nuclear receptor, PPAR-γ is a useful target for pharmacological manipulation. Furthermore, regulation of PPAR-γ expression and activity can be achieved at many levels, such as control of its gene transcription, pharmacologic control of its activity with receptor agonists and antagonists, and post-translational modifications (ex. phosphorylation, sumoylation and ubiquitination). This section will focus on control of PPAR-γ activity with several known drugs (agonists and antagonists) and phosphorylation. PPAR-γ is a member of a family of ligand-activated nuclear receptors and a number of its natural and synthetic ligands have been described (Figure 7A, B) [68, 72, 76]. PPAR-γ can be activated by peroxisome proliferators (ex. xenobiotics) as well as long-chain fatty acid and their derivatives [77-79]. Compounds, such as arachidonic acid and its prostaglandin metabolites of D2 and J2 families (ex. 15-deoxy-Δ12,14-prostaglandin J2) have been identified as powerful natural PPAR-

14

γ ligands (Figure 7A) [78, 79]. In addition to known natural ligands of PPAR-γ, several synthetic ligands have been described (Figure 7B) [66, 72]. Thiazolidinediones, a class of drugs used to treat diabetes and insulin resistance, are well-characterized synthetic ligands highly specific for the gamma member of the PPAR family (reviewed in [80]). The most common thiazolidinediones are pioglitazone and rosiglitazone, both of which are approved for use in the USA [66, 81]. It has been shown that upon binding to PPAR-γ, these agonists induce conformational and other changes of the receptor [63, 70], leading to transactivation (a DNA-dependent process) or transrepression (a DNA- independent process).

Figure 7: PPAR-γ ligands. Natural (A) and synthetic (B) PPAR-γ agonists. (C) Synthetic PPAR-γ antagonists (appears here with the permission of the publisher and adapted from Zieleniak et al, Arch. Immunol. Ther. Exp., 56: 331–345, 2008 [72]).

A model of transactivation has been summarized by Feige et al, 2006 (Figure 8) [82]. It proposes that in the absence of a ligand, PPAR-γ is coupled with its co-factor RXR-α and both are bound to the PPRE in the target gene promoter. The heterodimer also interacts with several co- repressors, histone deacetylases (HDACs) and nuclear co-repressor exchange factors (such as TBL1 and TBLR1), which results in a block of target gene transcription. Binding of an agonist to PPAR-γ at the α-helix 12 (H12) within the carboxyl terminus (AF-2 domain) leads to conformational changes, where H12 along with H3, 4 and 5 form a hydrophobic pocket which allows the recruitment of co- activators. Furthermore, nuclear co-repressor exchange factors facilitate the degradation of co- repressors and HDACs, while at the same time inducing a tighter binding of the heterodimer to the

15

PPRE and recruitment of co-activators and elements of the transcriptional machinery to initiate transcription of downstream factors.

Figure 8: Transactivation model of PPAR-γ activity. In the absence of an agonist, PPAR-γ and RXR are bound to the PPRE, along with co-repressors, histone deacetylases (HDACs) and nuclear co-repressor exchange factors (TBL1 and TBLR1), blocking gene expression. Upon ligand binding, repressors dissociate and are replaced by co- activators; transcriptional machinery is recruited to induce transcription (appears here with the permission of the publisher and adapted from Feige et al, Progress in Lipid Research, 45: 120-159, 2006 [82]).

Another effect that agonist binding may have on gene expression is transrepression (summarized in [82]). In this DNA-independent mechanism, activation of the PPAR-γ/RXR-α heterodimer results in transrepression of target genes, thus, leading to a decrease in gene expression. Transrepression can occur via three different pathways. First, agonist-binding to PPAR- γ may lead to transcription factor tethering; under such circumstances, the PPAR-γ/RXR-α heterodimer interacts with other transcription factors, preventing them from binding to their response elements and inducing transcription of genes. In another mechanism, squelching, PPAR-γ activation results in its competitive interaction (and sequestering) of co-activators shared between both the transcription factor and PPAR-γ. Lastly, agonist binding might lead to a redirection of common co-repressors to the other transcription factor in exchange for the factor’s co-activators; this process is called inverse squelching. Irrespective of their mechanism, all three processes result in repression of target gene expression without a direct interaction between PPAR-γ and the target gene promoter (ie, independent of DNA binding). Just as there are synthetic PPAR-γ agonists, several synthetic compounds have been identified as PPAR-γ antagonists, among these, GW9662 (which also weakly antagonizes PPAR-α) and T0070907 (Figure 7C) [83, 84]. T0070907 is an irreversible, non-competitive inhibitor of

16

PPAR-γ which binds covalently to the cysteine 313 residue of PPAR-γ2, located within helix 3 (H3). This binding interferes with the proper alignment of helices 12, 3, 4, and 5, as well as induces the recruitment of nuclear co-repressors (NCoRs) to the heterodimer [83, 84]. T0070907 can outcompete the effects of rosiglitazone (an agonist) in biochemical assays and cell-based systems. The overall effect that this antagonist binding has on receptor activity is an inhibition of PPAR-γ activity and a decrease of target gene expression. PPAR-γ can also undergo a series of post-translational modifications, including phosphorylation, ubiquitination and sumoylation (Figure 9) (reviewed in [73, 85]). To date, one phosphorylation site has been identified (located within the A/B domain), corresponding to serine residue 84 in PPAR-γ1 or 112 in PPAR-γ2. Phosphorylation of this site with ERK1/2 kinases leads to inhibition of PPAR-γ transcriptional activity as has been assessed in JEG-3 [86], 293K [87], and fibroblast [88] cell models. This loss of transcriptional activity is accomplished either by the recruitment of repressor proteins to the AF-1 domain, release of the activator or decrease in the ligand binding efficiency, indicating that phosphorylation might impede transcription by conformational changes within the LBD of PPAR-γ [86, 89]. Alternatively, phosphorylation of PPAR- γ by complexes containing cdk7 or cdk9 kinases was shown to lead to activation of the receptor [90, 91]. Although it appears that phosphorylation can either induce or inhibit PPAR-γ activity, it remains as one of the most important ways of modulating receptor activity in a ligand- independent manner.

Figure 9: Post-translational modifications of PPAR-γ. (A) Phosphorylation along with sumolyation within the activation function-1 (AF-1) domain results in reduced PPAR-γ activity. (B) Phosphorylation of PPAR-γ can also lead to higher receptor activity, for example, when phosphorylated by Cdk7 and Cdk9. (C) Ubiquitination is hypothesized to be related to phosphorylation of the PPAR-γ protein (although no direct ubiquitin-acceptor sites have been identified so far). Hypo-phosphorylated PPAR-γ was shown to have higher activity with respect to the adipogenic genes transcription and was associated with higher receptor degradation. (D) Lastly, sumoylation of PPAR-γ on K107 residue inhibits PPAR-γ activity in the context of inflammation (appears here with the permission of the publisher from van Beekum et al, Obesity, 17: 213-219, 2009 [85]).

17

1.6.5. PPAR-γ expression in healthy pregnancy and pregnancy complications

PPAR-γ expression patterns in human and mouse placentas have been studied extensively [75, 92-94]. In the human placenta, PPAR-γ was found to localize in cytotrophoblast and syncytiotrophoblast cells, as well as invading trophoblast columns [92-95]. In the mouse placenta, PPAR-γ was detected as early as E8.5 and persisted throughout gestation. Its expression was localized to the spongiotrophoblast and labyrinthine trophoblast, as well as secondary giant cells [75]. Expression of PPAR-γ in placental insufficiency syndromes has not been well established. Some reports have shown no differences in PPAR-γ mRNA and protein expression in sPE and IUGR [92], while our preliminary results show an increase in mRNA expression and an upregulation of PPAR-γ protein in pathologies when compared to controls (Figure 10) [patient population is the same cohort as was used in Drewlo et al, Placenta, 33: 952-954, 2012 [96] (Appendix 1); Table 1, appears here with permission of the publisher]. A study assessing expression of PPAR-γ during the first trimester of pregnancy showed that its protein levels were highly upregulated in EVT cells derived from placentas of spontaneous abortions (miscarriages) compared to those obtained from terminations of pregnancy [95]. Furthermore, it was shown that activators of PPAR-γ are significantly decreased in sera from PE pregnancies (implying lower PPAR-γ activity), and this reduction is seen prior to onset of PE symptoms [97].

18

Figure 10: Placental expression of PPAR-γ in pathologies and controls. (A) PPAR-γ mRNA expression was assessed in different pathologies (PE and IUGR) and over gestation (early and late first trimester, second trimester and term). Patient characteristics are listed in Table 1. PPAR-γ protein expression in control (B) and PE (C) placentas was assessed using immunohistochemistry (400X magnification). Values represented as mean ± SEM; ***p≤0.001, *p≤0.05 (n=9-11). PTC, pre-term control; IUGR, intrauterine growth restriction; PE, pre-eclampsia.

Table 1: Patient characteristics

Severe intrauterine growth restriction (sIUGR): Birth weight ≤10th percentile, abnormal umbilical artery Doppler (absent end diastolic flow velocity (AEDFV) or reversed end diastolic flow velocity (REDFV)), normotensive women. Severe pre-eclampsia (sPE): Birth weight >10th percentile, BP >140/90, proteinuria >300 mg/day or ≥1+ on dipstick. Mixed IUGR (MIUGR) group contained patients with co-existing features of the two groups above. aPre-term control (PTC): Birth weight >20th percentile, no abnormal umbilical artery Doppler measurements, BP <140/90, no history of gestational diabetes, no histologic evidence of chorio-amnionitis. bHealthy term delivery by caesarian section. cHealthy term vaginal delivery, (>37+n weeks gestation).

19

Several animal studies have looked at how placental insufficiency affects PPAR-γ expression in tissues other than the placenta, such as the adipose tissue and lungs. Interestingly, in the rat utero-placental insufficiency model of IUGR, PPAR-γ expression in male and female neonatal lungs was decreased [98] while it was increased in male visceral adipose tissue in the same model [99], indicating that placental insufficiency plays a differential role on PPAR-γ expression in different tissues. Additionally, in rats that were food-restricted during the second half of pregnancy (another model of IUGR), an upregulation of PPAR-γ expression in adipose tissues of male pups was seen, and these early programming events influenced rats’ development later in life, seen by adipocyte hypertrophy and higher fat accumulation at 9 months of age [100]. Collectively, studies described above support the role of PPAR-γ as an important modulator of pregnancy maintenance, as well as its role in post-natal development.

1.6.6. Role of PPAR-γ in trophoblast physiology

Studies have shown that PPAR-γ plays a critical role in placentation, consistent with findings by Barak et al in PPAR-γ null mice [75]. Linking the role of this nuclear receptor to trophoblast layer physiology was a study done by Parast et al looking at the role of this nuclear receptor in trophoblast differentiation [101]. Parast et al used PPAR-γ-null mouse trophoblast stem (TS) cells to study the role of this transcription factor in normal trophoblast turnover. Of note is their finding that PPAR-γ deficient TS cells showed decreased differentiation along the labyrinthine trophoblast lineage and premature differentiation into giant cells (equivalent of human EVTs). Equally as important was the finding that this transcription factor was able to directly regulate the expression of Gcm-1, a differentiation marker described previously. Other studies have established the regulation of Gcm-1 and syncytin-1 by PPAR-γ, supporting the role of these factors in normal trophoblast differentiation and trophoblast turnover [102]. The role of PPAR-γ in cell proliferation and differentiation in terms of trophoblast physiology has been explored using different models. As mentioned previously, lack of PPAR-γ in murine trophoblast stem cells impairs their potential to differentiate [101]. Consistent with these findings, addition of a specific PPAR-γ agonist to WT mouse TS cells induces their differentiation along the labyrinthine pathway and upregulates expression of downstream markers such as Gcm-1 [101]. Further supporting their data, the group found that adenoviral re-introduction of PPAR-γ into null TS cells decreased cell invasion (assessed using a Matrigel assay) and induced expression of syncytiotrophoblast markers. A similar approach was taken when studying isolated human primary trophoblast cells. It has been shown that in cultured primary cytotrophoblasts, synthetic ligand-

20 mediated stimulation of PPAR-γ induced their differentiation into syncytiotrophoblasts [93]. Additionally, in a follow-up study it was shown that cultured trophoblast cells treated with another synthetic PPAR-γ agonist increased their lipid accumulation and free fatty acid uptake [103]. Lastly, a study completed by Fournier and colleagues found that stimulation of PPAR-γ activity with either natural or synthetic ligands decreased primary trophoblast invasion while having no effect on cell proliferation [104]. Altogether, studies show that activation of PPAR-γ inhibits cytotrophoblast cell invasion, which has important implications for sPE where EVT cell invasion is inadequate.

1.6.7. Modulation of PPAR-γ during pregnancy and models of PE

Several groups have studied how modulation of PPAR-γ activity during pregnancy impacts placental development in rodents. Schaiff et al studied how the administration of rosiglitazone throughout the last 8 days of pregnancy (E10.5-18.5 at 30mg/kg body weight) to WT mice mated with WT or PPAR-γ+/- males affects placental morphology. They found that administration of PPAR- γ agonist decreased fetoplacental weight, decreased spongiotrophoblast layer thickness and reduced labyrinthine vasculature surface area in WT placentas, while it did not adversely impact PPAR-γ+/- placentas [105]. The group, therefore, concluded that lower PPAR-γ expression was protective against the effects of rosiglitazone during pregnancy. Furthermore, in a study conducted by Nadra et al, rosiglitazone was administered to mice at a dose of 100mg/kg body weight for 12 days (E4.5-16.5) [106]. The group found that following rosiglitazone treatment, placental structure exhibited overall disorganization, including features such as spongiotrophoblast thinning and presence of glycogen cells within the labyrinth, as well as decreased expression of pro-agniogenic genes. The study suggested that treatment of normal pregnancies with rosiglitazone negatively influences placental development. Contrary to the studies mentioned above which found that rosiglitazone has several negative impacts on placental development, a study by McCarthy et al found a positive effect of rosiglitazone treatment on PE-like pregnancies in rats [107]. Using the established reduced uterine perfusion pressure (RUPP) rat model of PE [108], the group showed that administration of PPAR-γ agonist during GD16-18 at a dose of 5mg/kg, ameliorated hypertension, endothelial dysfunction and kidney function which were present in RUPP-treated animals receiving the vehicle, indicating a beneficial effect of rosiglitazone on pregnancy complicated by placental insufficiency. Furthermore, by blocking activity of PPAR-γ with a specific antagonist T0070907, the authors were able to induce PE-like features, such as hypertension, higher plasma sFlt-1levels as well as lower levels of Vegf, in healthy rats, indicating impaired endothelial function [109]. Additionally, blocking PPAR-γ activity

21 disrupted placental development assessed by lower rate of labyrinthine differentiation and higher proliferation. Studies described above indicate an important role of PPAR-γ in placental development and pregnancy. Although results are contradictory, it should be mentioned that dosage (20mg/kg per day dose in mice is the equivalent of the clinically relevant dose in humans), timing and duration of drug administration differed in these studies. Thus, these factors should be strongly considered when translating studies on rosiglitazone treatment to human pregnancies. Since it has been shown that rosiglitazone is able to cross the placenta [110], the use of this drug as a therapeutic agent should be carefully examined in the human placental explant model. Several case studies of women being exposed to rosiglitazone (5mg/day) in first or second trimester at of pregnancy have not shown any adverse pregnancy outcomes; however, these findings need to be confirmed in a large- population study [111, 112].

1.6.8. PPAR-γ controls a cytoprotective enzyme, heme oxygenase-1 (HO-1)

Because of its role in placental development, and the ability of using pharmacological tools to modulate its activity, PPAR-γ has been proposed as a potential therapeutic target for the treatment of PE [107, 109, 113]. Interestingly, PPAR-γ is able to ameliorate disease characteristics in the rat PE model by acting via its downstream target, heme oxygenase-1 (HO-1), an enzyme responsible for heme breakdown within the cell [107]. Studies have shown that PPAR-γ is able to directly regulate HO-1 expression in vascular smooth muscle cells and human umbilical vein endothelial cells (HUVECs) [114]. Combined, these findings indicate that this enzyme, which possesses multiple anti-inflammatory, cytoprotective and anti-oxidant functions, may play a role in the pathogenesis of sPE.

1.7. Heme oxygenase-1 (HO-1)

Note: the following subsections have been published in Placenta journal by Levytska et al in 2013 (Appendix 2) [115]. Heme oxygenase (HO) family comprises three enzyme isoforms: HO-1, HO-2 and HO-3 [116, 117]. HO-1, the only inducible HO isoform, is highly expressed in tissues responsible for heme metabolism, such as spleen and liver [118]. HO-2 is constitutively-expressed and mostly found in the brain and testes [116]. The function of HO-3 is presently unknown, but the enzyme is believed to be inactive. HO-1, a type I integral protein embedded in smooth ER membrane, was first isolated from spleen and liver microsomal fractions and shown to catalyze the rate-limiting step of intracellular heme breakdown [119]. Using oxygen, the enzyme metabolizes heme into three components produced in equimolar amounts: carbon monoxide (CO), biliverdin and iron (Fe+2)

22

[120]. Under physiological conditions, biliverdin is immediately converted into bilirubin by biliverdin reductase. The two metabolites of heme breakdown, CO and bilirubin, have important cellular functions giving HO-1 its vasodilatory, anti-inflammatory, anti-apoptotic, anti-oxidant, and cytoprotective properties (Figure 11) [121-128].

Figure 11: Schematic representation of heme breakdown by HO-1. HO-1 is responsible for heme metabolism within the cell. The process requires NADPH-cytochrome- P450 reductase and molecular oxygen (O2) which act as a co-enzyme and a substrate, respectively, for HO-1. HO-1 breaks down heme into three equimolar components: iron (Fe2+), biliverdin (immediately converted to bilirubin by biliverdin reductase), and CO. These metabolites are responsible for vasodilatory, anti-oxidant and other protective properties of HO-1 (appears here with the permission of the publisher from Levytska et al, Placenta, 34: 291-298, 2013 [115]).

1.7.1. HO-1 expression during pregnancy

Studies of HO-1 and HO-2 expression in mouse, rat and human placentas have been numerous [129-133]. However, since the first report localizing HO to human placental tissue, studies of HO-1 expression throughout human gestation and in pathology have been inconsistent. McLean and colleagues were the first group to demonstrate HO-1 and HO-2 expression in human placenta, showing that HO-2 protein was more abundant than HO-1 at term [134]. Yoshiki et al reported that HO-1 and HO-2 proteins are localized in different cell types in the human placenta: HO-1 immunohistochemistry revealed positive staining in the syncytiotrophoblast layer with traces

23 in the underlying cytotrophoblasts, while HO-2 was mostly expressed in endothelial and smooth muscle cells [135]. This group also showed that HO-1 expression did not change throughout gestation, while HO-2 expression increased towards term. Differential expression of HO-1 in the invading cytotrophoblast cell columns has been noted; HO-1 was found to be high in distal, but drastically decreased, in the proximal cell layers [133]. The first study looking at the expression of HO-1 in pregnancy complications showed lower HO-1 protein expression in pre-eclamptic (PE) placentas; this was in contrast to a positive association between placental HO-1 expression and feto-placental vascularisation in normal pregnancy [122]. In support of pathologic sPE observations, human in vivo studies demonstrate that while exhaled CO levels increase throughout gestation in healthy women, end-tidal CO measurements are lower in hypertensive pregnant women, especially those affected by sPE, supporting the role of CO as a physiologic vasodilator implicated in gestational hemodynamic changes [136, 137]. Furthermore, Zenclussen et al. found lower HO-2 protein levels in spontaneous abortion (SA) cases (vs normal pregnancy) and in PE (vs age-matched controls), while HO-1 levels were low across all pathologies without significant differences [138]. Lash et al found that infarcted villi in PE placentas had lower HO activity and lower expression of HO-2 while HO-1 expression was unaffected [139]. In contrast to above studies, Barber et al showed that HO-1 protein was undetectable in human placenta, and there was no difference in HO-2 expression between PE and growth restricted pregnancies and preterm controls [140]. These studies show that discrepancies regarding the expression of HO-1 and HO-2 in disease states and throughout gestation exist and may be due to intra- and inter-tissue heterogeneity.

1.7.2. Role of HO-1 in placentation

The first description of an HO-1 knockout system was published by Poss and Tonegawa in 1997 [141, 142]. The authors found that HO-1 is required for proper iron reutilization, since adult mice lacking HO-1 developed anemia and were susceptible to oxidative tissue damage and inflammation [142]. Furthermore, the group showed that murine embryonic fibroblast cells lacking HO-1 were sensitive to oxidative stress [141]. Kapturczak et al showed that mice lacking Hmox-1 had impaired immune responses and higher levels of pro-inflammatory Th1 factors [143]. Zhao and colleagues were the first group to provide evidence for HO-1 as a critical player in proper placental development [130, 144, 145]. The group demonstrated that in normal pregnancy, mice exhibit an increase in utero-placental blood flow accompanied by a significant increase in maternal abdominal aorta diameter (adaptations similar to those observed in humans). These changes were concurrent

24 with higher endogenous CO production and elevated plasma bilirubin levels, and placental HO-1 expression which peaked at E13.5 [130, 145]. In a follow-up study, Zhao et al characterized the effect of HO-1 deficiency on placental development [144]. Crossing Hmox-1 heterozygous (+/-) mice resulted in an altered Mendelian ratio of Hmox-1-/- pups (2.4% instead of the expected 25% ratio), most of which aborted before E10.5. Hmox-1+/- placentas presented with a thinner spongiotrophoblast (SP) layer due to increased apoptosis at the junction zone, supporting the role of HO-1 as an anti-apoptotic player [144]. The implantation rate in heterozygous crossbreedings was lower, and, at E12.5 and E15.5, the placentas were lighter and smaller (compared to wildtype breedings). Furthermore, heterozygous mothers had higher plasma levels of sFlt-1, one of the key players in vascular endothelial cell dysfunction, and elevated diastolic blood pressure, implying that Hmox-1 is involved in the manifestation of sPE-like symptoms in mice, mimicking complications seen in humans [146]. It was later shown that inadequate placentation of Hmox-1-/- placentas resulted in IUGR and subsequent fetal death [147]. Exogenous application of CO, a product of heme metabolism by HO-1, to Hmox-1-deficient mice during the implantation window decreased fetal death in these pregnancies. It was also shown that Hmox-1+/+ blastocysts isolated from mice on E3.5 attached sooner than Hmox-1+/- blastocysts while blastocysts lacking Hmox-1 failed to attach altogether [147]. This study also characterized the effect of Hmox-1 deficiency on placentation showing that Hmox-1 is necessary for the differentiation of trophoblast stem cells to giant cells which are critical for proper placental development. Early placental vascularisation is also affected by Hmox-1 as seen from studies of crossbreeding Hmox-1 wildtype mice and Hmox-1 heterozygous mice [146]. Breeding female mice partially deficient in Hmox-1 with males of the same genotype resulted in feto-maternal interface malformations; namely, lower microvasculature blood vessel volumes, less sinusoid spaces in the labyrinth and disorganized capillaries, when compared to wildtype breedings. Interestingly, these alterations in placental morphology were independent of the placental genotypes and occurred in both, Hmox-1+/+ and Hmox-1+/- placentas, as long as both parents were Hmox-1 deficient. Since these impaired features of placental development are similar to those observed in pregnancy complications such as sPE, the role of HO-1 in implantation and placental development is further supported.

25

1.7.3. Studies of HO-1 in cell-based models

Cell models have also served as a useful tool to study HO-1 properties, which include protection against cytotoxicity and cell death, induction of an angiogenic phenotype, and others. HO-1 has been studied in cell models related to pregnancy and placental development, such as HUVECs, HTR8 cells (a model for extravillous trophoblast cells, EVTs), and BeWo cells (a model for syncytiotrophoblast formation). The activation of HO-1 by nuclear factor (erythroid-derived 2)-like 2 (Nrf-2), a prominent protector against oxidative stress, was described in the BeWo cell model, where it protected cells against hydrogen peroxide- and Fe+2-induced stress [148]. Interestingly, a cleaved isoform of HO-1 has been shown to have transcription factor-like activity by contributing to the expression of cytoprotective genes in cultured fibroblast and Hepa cells [149]. Genetic profiling of highly migratory EVTs and villous cytotrophoblast (CT) cells (with a low migratory potential) isolated from cultured explants and first trimester human placentas, respectively, showed a lower expression of HO-1 in EVTs vs CTs [150]. Confirming these findings, the group also showed that CT cells had higher HO-1 RNA expression compared to EVT cells, as well as higher HO-1 protein expression in proliferating CTs (vs EVTs) in first trimester placental explants, as seen by immunohistochemistry. Furthermore, the study showed that retroviral induction of HO-1 in naturally-migratory HTR-8 cells reduced their motility, while HO-1 silencing in non-migratory BeWo cells increased their migration, suggesting that lower HO-1 expression is associated with a lower cell motility and trophoblast invasion. Interestingly, the mechanism responsible for increasing cell migration by HO-1 was via the regulation of PPAR-γ. These findings are in contrast to a study by McCaig et al who showed that inhibition of global HO activity with zinc protoporphyrin IX and antibody-mediated inhibition of HO-2 reduced invasion of primary CT cells in vitro [151]. Discrepancies in studies of how HO affects cell invasion and motility may be due to the use of different cell models; CT cells isolated from term placentas (when invasion is no longer prominent) in the study by McCaig et al provide a different model compared to Bilban et al, who used CT cells isolated from first trimester tissue (when invasion is high). Furthermore, cell models used by Bilban et al are representative of the early placental development stages and, thus, might reflect the actions of HO-1 in placental invasion more accurately. Although differential mechanisms of HO-1 regulation in several cell types have been shown, collectively, studies suggest that HO-1 contributes to cell migration and invasion. These features are important in sPE since inadequate invasion of the uterus by EVTs, together with defective formation of the outer syncytiotrophoblast, imply dual defects in each type of the differentiated trophoblast [152].

26

1.7.4. Molecular manipulation of HO-1

The role of HO-1 in underlying mechanisms of placental insufficiency syndromes has been studied extensively. Considering the prominent role of anti-angiogenic factors in the manifestation of endothelial cell damage, it is natural to study the connection between the protective role of HO-1 and these disease markers. Implicating the role of HO-1 in sPE, it was shown that HO-1 induction or CO administration in HUVECs inhibits their release of sFLT-1 and sEndoglin (sENG) [153, 154]. Both of these anti-angiogenic factors contribute to endothelial dysfunction in sPE; thus, reinforcing the hypothesis that HO-1 and heme metabolites promote a pro-angiogenic phenotype [153]. This same molecular pathway was studied in rats which have undergone the RUPP surgery; they had elevated mean arterial pressure and exhibited lower fetal weight, suggesting that hypoxia contributes to fetal growth restriction [155]. RUPP animals also have elevated plasma sFlt-1 and higher superoxide production, both of which can be attenuated with an HO-1 activity inducer or by the administration of bilirubin or CO. HO-1 placental expression is decreased in RUPP animals, suggesting that in rat tissue, hypoxia negatively regulates HO-1 [156]. As mentioned previously, HO-1 has the ability to decrease sFLT-1 release; interestingly, HO-1 is also able to offer protection against sFlt-1 damage while bypassing this effect (ie, when plasma sFlt-1 levels are held constantly high) [153, 157]. George et al showed that increasing HO-1 activity in rats with sFlt-1-induced hypertension ameliorated elevated blood pressure despite the constantly elevated sFlt-1 levels [157]. HO-1 induction improved endothelial function, decreased blood pressure and increased VEGF levels. Supporting the anti-apoptotic role of HO-1, it was shown that HO-1 induction in the RUPP-treated animals shifted the balance from pro-injury to pro-survival phenotype measured by the restoration of intracellular ATP levels and increased phosphorylation of survival regulators [158]. Collectively, studies described above support the hypothesis that pharmacologic interventions aimed at increasing HO-1 expression and activity, particularly within the placenta, may help restore maternal cardiovascular and endothelial functions which are defective in sPE.

27

1.8. Rationale of the project

sPE is a serious complication of pregnancy without a cure, other than delivery of the placenta which often necessitates premature birth. Because of this observation, the placenta, specifically the trophoblast layer, has been thought to be the key player in PE where altered trophoblast turnover is observed. Several molecules localized to the trophoblast, such as GCM-1, PPAR-γ and HO-1, have been shown to contribute to placental dysfunction. GCM-1, a trophoblast differentiation marker, and HO-1, an enzyme with anti-oxidant and vasodilatory properties, are both downregulated in PE placentas. It is speculated that decreased levels of these molecules lead to shallow placental invasion, inadequate spiral artery remodeling, vasoconstriction and oxidative damage.

In mice, the expression of both GCM-1 and HO-1 is regulated by another transcription factor, PPAR-γ, the lack of which leads to loss of differentiation potential of trophoblast stem cells [101]. PPAR-γ is a ligand-activated nuclear receptor whose expression can be regulated with specific natural and synthetic agonists and antagonists. Recently, we have shown that upregulating PPAR-γ activity with rosiglitazone ameliorates disease features in a rat model of PE via its induction of HO- 1 [107]. Therefore, we decided to study the role of PPAR-γ in a human cell model. Our ability to pharmacologically upregulate PPAR-γ activity poses a potential avenue for improving placental trophoblast physiology (via upregulation of GCM-1 and differentiation) and lowering systemic vascular resistance (via upregulation of HO-1 and vasodilatation).

1.9. Models used in this study

One of the objectives of our study was to test whether there is a connection between GCM-1 and HO-1 using two models. In the first model, we studied HO-1 expression in Gcm-1 heterozygous mouse placentas. In the second model, we studied gene expression in GCM-1 silenced BeWo cells originating from the human choriocarcinoma. Of the cell lines assessed, BeWo cells are the only cell type which expresses all three genes of interest, namely, HO-1, GCM-1 and PPAR-γ. Furthermore, while BeWo cells have an inherent ability to fuse spontaneously in culture, there is also a possibility of inducing cell fusion pharmacologically. Because of these features, the BeWo cell model was used to study key events associated with trophoblast function, namely, cell proliferation and differentiation.

28

1.10. Hypotheses and specific aims

Our studies were based on the following hypotheses. First, we hypothesized that the transcription factor GCM-1 controls the expression of HO-1 in the mouse placenta and in the human choriocarcinoma-derived cell model. Second, we hypothesized that PPAR-γ-mediated expression of GCM-1 and HO-1 regulates key events associated with trophoblast physiology, assessed in the BeWo cell line.

Our hypotheses were tested in the following aims:

1. Analysis of gene expression in Gcm-1 heterozygous mice. 2. Analysis of gene expression in GCM-1-silenced BeWo cells. 3. Establishment of the role of PPAR-γ in trophoblast physiology using the BeWo cell model. a. Establishing the effect of PPAR-γ agonist and antagonist on expression of HO-1. b. Establishing the effect of PPAR-γ activity-modulating drugs on PPAR-γ mRNA and protein expression and localization. c. Establishing the effect of PPAR-γ agonist and antagonist on the differentiation axis, including the effect on GCM-1 expression, cell proliferation, differentiation and fusion.

29

Chapter 2 : MATERIALS AND METHODS

30

2.1. Mouse model

A mouse model was used to study the effects of Gcm-1 deletion on Hmox-1 expression. Because the homozygous deletion of Gcm-1 is lethal in mice [53] and GCM-1 expression in human PE placentas is hypomorphic [58], the heterozygous Gcm-1 mouse model was used (mouse breeding and tissue collection, as well as mRNA isolation, was performed by members of Dr. Adamson’s laboratory, University of Toronto, and kindly donated for our study [59]). Virgin 6-10 week old wildtype CD1 females (Gcm-1+/+) were mated with heterozygous males (Gcm-1+/-), which resulted in an approximate 1:1 Mendelian ratio of Gcm-1 wildtype and heterozygous conceptuses.

2.1.1. Tissue collection

Once removed from the uterus on embryonic days 13.5 (E13.5) and E17.5, placentas were dissected to remove decidual tissue; then they were further microdissected to enrich for the labyrinth region. Samples were stored in RNAlater (Ambion, Life Technologies, Burlington, ON, Canada) and frozen at -80oC for further RNA analysis by quantitative real time-polymerase chain reaction (qRT-PCR). All conceptuses were genotyped at Transnetyx, Inc. (Cordova, Tennessee, USA).

2.1.2. RNA extraction, reverse transcription and qRT-PCR

RNA was isolated from mouse labyrinth-enriched placental samples using the TRIzol method (Invitrogen, Burlington, ON, Canada) according to manufacturer’s instructions (n=6 per genotype). Genomic DNA was removed using DNase treatment (RNase-free DNase Set, Qiagen, Toronto, ON, Canada). RNA was further purified using the RNeasy Mini-elute Cleanup Kit (Qiagen, Toronto, ON, Canada). RNA concentration was measured using NanoDrop 1000 (Thermo Scientific, Ottawa, ON, Canada). One microgram (µg) of each sample was reverse transcribed to cDNA using iScript™ Reverse Transcription Supermix (Bio-Rad, Mississauga, ON, Canada) according to the following protocol: 5 minutes at 25OC, followed by 30 minutes at 42OC, and 5 minutes at 85OC. Gene expression was measured using qRT-PCR and run on the CFX384 Real-Time PCR Detection System (Bio-Rad, Mississauga, ON, Canada) with LuminoCt® SYBR® Green qPCR ReadyMix™ (Sigma- Aldrich, Oakville, ON, Canada). qRT-PCR reactions were performed according to the following protocol: initial activation at 95OC for 5 minutes, followed by 38 thermal cycles of denaturation at 95OC for 5 seconds and annealing/extension at 60OC for 20 seconds, followed by a melting curve (from 65OC to 95OC in 0.5OC increments for 5 seconds). Primers were designed using the Primer-

31

BLAST tool (http://www.ncbi.nlm.nih.gov/tools/primer-blast/) for the following genes: murine glial cell missing-1 (Gcm-1), heme oxygenase-1 (Hmox-1), heme oxygenase-2 (Hmox-2), peroxisome proliferator-activated receptor-gamma (Ppar-γ). Primers were manufactured by Integrated DNA Technologies, Inc (Skokie, IL, USA). Housekeeping genes [TATA box binding protein (Tbp), β-actin (BAct), and glyceraldehyde-3-phosphate dehydrogenase (Gapdh)] were chosen according to published literature [59]. Primer efficiency and specificity checks were performed on the CFX96 Real-Time PCR Detection System (Bio-Rad, Mississauga, ON, Canada) with LuminoCt® SYBR® Green qPCR ReadyMix™ (Sigma-Aldrich, Oakville, ON, Canada). E15 placental cDNA was used as the standard in 1:4 decreasing dilution series with the starting concentration of 50 x 10-9 ng/µL; qRT- PCR reactions were performed as described above. Primer sequences and efficiencies are summarized in Table 2. CFX ManagerTM Software, Version 2.1 (Bio-Rad, Mississauga, ON, Canada) was used to analyze gene expression. Expression of genes of interest (ie, Gcm-1, Ppar-γ, Hmox-1 and Hmox-2) was normalized to the geometric mean of three house-keeping genes (Tbp, BAct and Gapdh). To compare gene expression between wildtype and heterozygous Gcm-1 placentas, expression in heterozygous samples was expressed as fold change vs wildtype placentas (set as 1).

Table 2: Mouse primers

Gene Primer sequence (5’ → 3’) Number of bases Primer pair efficiency Gcm-1 Forward TGA AGC TTA TTC CCT GCC GA 20 113.9% Reverse TGA TCA TGC TCG CCT TTG G 19

Ppar-γ Forward TCA CAA TGC CAT CAG GTT TGG 21 106.8% Reverse AAT GCT TTG CCA GGG CTC GCA 21

Hmox-1 Forward GGC TTT AAG CTG GTG ATG GCT TC 23 113.0% Reverse TTC GGT GCA GCT CCT CAG GGA A 22

Hmox-2 Forward ATG AGC CAA GGA CCA GAG TGA G 22 108.5% Reverse TCC GAG GTC TCC ACC TCT GAA GA 23

BAct Forward TCG TGC GTG ACA TCA AAG AGA 21 108.2% Reverse GAA CCG CTC GTT GCC AAT A 19

Gapdh Forward AGG AGT AAG AAA CCC TGG ACC 21 106.6% Reverse AGG CCC CTC CTG TTA TTA TGG 21

Tbp Forward CGG ACA ACT GCG TTG ATT TTC 21 108.6% Reverse AGC CCA ACT TCT GCA CAA CTC 21

32

2.2. In vitro model

2.2.1. Evaluation of cell models

Several pregnancy-related cell lines were evaluated to select a cell-based model that expresses as many target genes as possible. Four cell models were evaluated: the human choriocarcinoma cell lines JEG-3 and BeWo, human transformed extravillous trophoblast cell line HTR8 cells and human umbilical vascular endothelial cells (HUCECs). JEG-3 and BeWo cells were obtained from ATCC (Burlington, ON, Canada); HUVECs were purchased from Lonza (Allendale, NJ, USA); and HTR8 cells were a kind gift from Dr. Graeme Smith (Queen’s University, Kingston, ON, Canada). Based on gene expression analysis (Figure 12), the human choriocarcinoma cell line BeWo was selected as the cell model of choice.

1.5

1.0 GCM-1 PPAR- HO-1

0.5 sFLT-1 Gene expression Gene

0.0 BeWo HTR8 JEG-3 HUVEC Cell line

Figure 12: Gene expression in human pregnancy-related cell lines. GCM-1, HO-1, PPAR-γ and sFLT-1 expression was assessed in BeWo, HTR8, JEG-3 and HUVE cells. BeWo cells were chosen as the cell model of choice (n=1).

2.2.2. The BeWo cell line

Prior to performing any experiments, the BeWo cell line was fingerprinted at The Centre for Applied Genomics (SickKids, Toronto, ON, Canada); markers were found to be identical to those in the ATCC database. Gene expression of PPAR-γ, HMOX-1 and GCM-1 in BeWo cells across passages was evaluated (Figure 13). Based on the results, only BeWo cells between passages 10-20 were used in all experiments. Cells were maintained in F12K medium (Wisent Inc, St. Bruno, QC, Canada), supplemented with 10% fetal bovine serum (FBS) (Canadian grade, heat-inactivated, Invitrogen, Burlington, ON, Canada), 100 units/mL penicillin, 100 µg/mL streptomycin and 2nM L-glutamine

(Life Technologies, Burlington, ON, Canada), in atmospheric O2/5% CO2 at 37OC.

33

2.5 2.5 A B 2.0 2.0

1.5 1.5

1.0 1.0

(normalized to HKGs) to (normalized (normalized to HKGs) to (normalized

0.5 0.5 mRNA expression across expression passages mRNA

mRNA expression across expression passages mRNA HO-1

PPAR- 0.0 0.0 8 10 15 17 19 20 22 8 10 15 17 19 20 22 Passage Passage

C 0.4

0.3

0.2

(normalized to HKGs) to (normalized 0.1 mRNA expression across expression passages mRNA

GCM-1 0.0 8 10 15 17 19 20 22 Passage

Figure 13: Preliminary assessment of gene expression in BeWo cells across passages. mRNA expression of PPAR-γ (A), HO-1 (B), and GCM-1 (C) was assessed in BeWo cell passages 8, 10, 15, 17, 19, 20, and 22, and normalized to expression of housekeeping genes (HKGs). Cells between passages 10 and 20 were used in subsequent studies (n=1).

2.2.3. Drug treatments of BeWo cells

BeWo cells were seeded at 100,000 cells per well in 24-well culture plates and allowed to attach for 24 hours. Cells were treated the following day with a variety of drugs (Table 3). First, cells were pretreated with the inhibitor (T0070907) and/or vehicles for 30 minutes. Then, medium containing treatments (ie, agonist, antagonist, and/or vehicles) was applied. Cell viability under all treatments was assessed at 48 hours of culture using CytoTox-ONETM Homogeneous Membrane Integrity Assay (Promega, Madison, WI, USA). No drug treatments resulted in significant cell toxicity at 48 hours, except for T0070907 at 10µM (Figure 14). The chosen drug concentrations were based on our preliminary results and/or published literature. Rosiglitazone and forskolin in vitro administration is more common than that of T0070907, which has been mostly used in animal models [159, 160]. The group which first described T0070907 as a PPAR-γ antagonist have

34 identified its IC50 as 1nM and showed that 1µM is able to bind irreversibly and fully inhibit the receptor activity; thus, we based our dose-response studies on this value [161].

Table 3: Drugs and concentrations

Chemical Mode of action Concentration(s) Vehicle Supplier T0070907 specific antagonist of PPAR-γ; 0.1µM DMSO Cayman Chemical binds irreversibly to cysteine in helix 3, 1µM Ann Arbor, MI, thus, changing conformation of helix 12 10µM USA of the PPAR-γ LBD

Rosiglitazone selective synthetic PPAR-γ agonist; 10µM DMSO Enzo Life Sciences induces PPAR-γ translocation into the 100µM Burlington, ON, nucleus; Canada increases transcriptional activity

Forskolin increases cellular concentration of cAMP; 20µM Ethanol Sigma-Aldrich indirectly induces expression of GCM-1 25µM Oakville, ON, and activity of PPAR-γ Canada

1.5 a b b b

1.0 24 hours 48 hours

0.5 72 hours

Relative cell toxicity cell Relative (compared to vehicle) to (compared 0.0 Vehicle Rosi (10M) Rosi (100M) T007 (1M) T007 (10M)

Treatment

Figure 14: Effects of drugs on cell viability over time. Cell viability was measured using the CytoTox-ONE Homogeneous Membrane Integrity Assay at 24, 48 and 72 hours post-treatment. No significant differences on cell viability were found at 24 and 48 hours of culture (with the exception of T0070907 at 10µM concentration). Data represented as mean ± SEM; a,b, p≤0.05 vs vehicle control, set as 1 (n=4-5). Rosi, rosiglitazone; T007, T0070907.

Cells were treated for 3, 6, 24 or 48 hours, then washed in ice-cold Dulbecco's phosphate- buffered saline (D-PBS) (Wisent Inc, St. Bruno, QC, Canada), and collected according to different downstream applications. For RNA analysis, cells were collected into RLT Plus buffer (Qiagen, Toronto, ON, Canada) with 10% β-mercaptoethanol (Fisher Scientific, Ottawa, ON, Canada). For protein analysis, cells were scraped and collected either in RIPA Buffer (Thermo Scientific, Ottawa, ON, Canada) with phosphatase (Phosphatase Inhibitor Cocktail 2, Sigma-Aldrich, Oakville, ON,

35

Canada) and protease inhibitors (cOmplete Mini, EDTA-free Protease Inhibitor Cocktail Tablets, Roche Applied Science, Laval, QC, Canada) diluted according to manufacturer’s instructions; or in HO-1 Extraction Buffer (Enzo Life Sciences, Burlington, ON, Canada) with the same inhibitors added. Conditioned medium was collected at 24 and 48 hour time points, centrifuged for 5 minutes at 425g at room temperature (RT) to remove cellular debris. All collected samples were stored at - 80OC for further analysis.

2.2.4. Transfection of cells

Four different siRNA sequences targeting the human GCM-1 gene were assessed (unpublished observations); two of them, siRNA 815 and 201, were found to have the strongest effects on GCM-1 expression [57]. Furthermore, the combination of both oligonucleotide sequences led to an even more pronounced gene downregulation; thus, the combination of siRNA 815 and 201 was used for our subsequent experiments in BeWo cells. The cells were co-transfected with two double-stranded siRNA oligonucleotides (21mer), named 815 and 201, against the human GCM-1 sequence (Qiagen, Toronto, ON, Canada). 815 siRNA was designed to 5’-AACCTACAGTAGTGGAGACCT-3’ and 201 siRNA was designed to 5’- AACTCCCGCATCCTCAAGAAG-3’ DNA targets (see Figure 15 for gene map). Non-silencing control (sequence: 5’-TTCTCCGAACGTGTCACGT-3’) was used as a negative control. At 40-50% confluency, BeWo cells were transfected with 20µM of a 1:1 cocktail of 815 and 201 GCM-1 siRNA or non- silencing control siRNA with Lipofectamine Reagent (Invitrogen, Burlington, ON, Canada) overnight for 24 hours, in accordance with the manufacturer’s instructions. The following day, media was changed and cells were treated with 25µM forskolin or vehicle control for another 48 hours. Experiments with fluorescent-labeled siRNA established 80-90% transfection efficiency (data not shown). Toxicity of siRNA treatment was monitored with Human Interferon Alpha ELISA kit (PBL Biomedical Laboratories, Brussels, Belgium).

Figure 15: Location of siRNA sequences in the human GCM-1 mRNA. Two sequences, 201 and 815, were designed against human GCM-1 mRNA transcript.

36

2.2.5. RNA analysis

2.2.5.1. Primer design

Human primers were designed using the online Primer-BLAST tool (http://www.ncbi.nlm.nih.gov/tools/primer-blast/) for genes listed in Table 4. Primers were produced by Integrated DNA Technologies, Inc (Skokie, IL, USA). Primer efficiency was tested using RNA combined from samples from first and second trimester human placental tissue, placental pathologies (PE, IUGR and mixed PE-IUGR) and age-matched controls. Placental tissue was crushed with mortar and pestle in liquid nitrogen. RNA from these samples was isolated using the TRIzol Reagent (Invitrogen, Burlington, ON, Canada) according to manufacturer’s instructions. Samples were further purified using RNeasy Plus Mini Kit (Qiagen, Toronto, ON, Canada) according to manufacturer’s specifications. RNA quality and amount was measured using the Experion RNA StdSens Analysis Kit (Bio-Rad, Mississauga, ON, Canada); only samples with RQI>7.5 were used to create a pooled cDNA sample. One microgram of pooled RNA sample was reverse transcribed using iScript™ Reverse Transcription Supermix (Bio-Rad, Mississauga, ON, Canada) according to manufacturer’s instructions. Primer efficiency and specificity checks were performed according to the same protocol as summarized in section 2.1.2 (see Table 4 for human primer efficiencies).

Table 4: Human primers

Gene Primer sequence (5’ → 3’) Number of Primer pair bases efficiency GCM-1 Forward ATG GCA CCT CTA GCC CCT ACA 21 102.5% Reverse GCT CTT CTT GCC TCA GCT TCT AA 23

PPAR-γ Forward CTC AGT GGA GAC CGC CCA GG 20 109.2% Reverse GCT CCA GGG CTT GTA GCA GG 20

HMOX-1 Forward CGG CTT CAA GCT GGT GAT GGC 21 110.6% Reverse CCT GCT CCA GGG CAG CCT TG 20

GAPDH Forward AGA TCA TCA GCA ATG CCT CC 20 108.2% Reverse CAT GAG TCC TCC CAC GAT AC 20

YWHAZ Forward ACT TTT GGT ACA TTG TGG CTT CAA 24 95.3% Reverse CCG CCA GGA CAA ACC AGT AT 20

TBP Forward TGC ACA GGA GCC AAG AGT GAA 21 110.4% Reverse CAC ATC ACA GCT CCC CAC CA 20

SDHA Forward TGG GAA CAA GAG GGC ATC TG 20 101.7% Reverse CCA CCA CTG CAT CAA ATT CAT G 22

HPRT Forward TGA CAC TGG CAA AAC AAT GCA 21 95.7% Reverse GGT CCT TTT CAC CAG CAA GCT 21

37

2.2.5.2. Housekeeping gene stability check

Stability of five human housekeeping genes across different drug treatments was tested in BeWo cells using the GeNorm method, which is part of BiogazelleTM software. The software calculates relative housekeeping gene stability in relation to other housekeeping genes. The gene stability value, M, indicates an average pairwise variation of the control housekeeping gene with other housekeeping genes. Thus, higher M value indicates high variation in control gene expression between treatments, and, conversely, lower M value indicates high gene stability (for a detailed explanation, see [162]). Based on results obtained from the preliminary experiments, HPRT, YWHAZ and TBP were chosen as housekeeping genes for subsequent gene analyses (Figure 16).

Figure 16: Stability of individual housekeeping genes in BeWo cells under different treatments. Lower GeNorm M value reflects a higher stability of the gene across different treatments. TBP, YWHAZ and HPRT were the most stable housekeeping genes and were used in subsequent qRT-PCR reactions.

38

2.2.5.3. RNA extraction

BeWo cell RNA was extracted using the RNeasy Plus Mini Kit (Qiagen, Toronto, ON, Canada) according to manufacturer’s instructions. Briefly, cells in the 24-well plate were washed with ice- cold D-PBS (Wisent Inc, St. Bruno, QC, Canada), and collected with 350µL of RLT Plus Buffer with 10% β-mercaptoethanol. After the removal of genomic DNA, RNA was precipitated using 70% ethanol. Following several washes with RW1 and RPE buffers, total RNA was eluted with 30µL of RNase-free water. RNA concentration was measured immediately using NanoDrop 1000 (Thermo Scientific, Ottawa, ON, Canada).

2.2.5.4. Reverse transcription and quantitative real time-PCR (qRT-PCR)

Five hundred nanograms (ng) of each sample were reverse transcribed to cDNA using iScript™ Reverse Transcription Supermix (Bio-Rad, Mississauga, ON, Canada) according to the following protocol: 5 minutes at 25OC, 30 minutes at 42OC, and followed by 5 minutes at 85OC. Gene expression was measured using quantitative real time-PCR and run on the CFX384 Real-Time PCR Detection System (Bio-Rad, Mississauga, ON, Canada) with LuminoCt® SYBR® Green qPCR ReadyMix™ (Sigma-Aldrich, Oakville, ON, Canada). qRT-PCR reactions were performed according to the following protocol: initial activation at 95oC for 5 minutes, followed by 38 thermal cycles of denaturation at 95oC for 5 seconds and annealing/extension at 60oC for 20 seconds. Gene expression was normalized to the geometric mean of three housekeeping genes (HPRT, TBP and YWHAZ). Gene of interest expression in each treatment was expressed as fold change relative to its respective vehicle (set as 1).

2.2.6. Protein analysis

2.2.6.1. Protein isolation

Following treatment for 24 or 48 hours, BeWo cells were collected in 250µL of lysis buffer with phosphatase and protease inhibitors added (as described above). Samples were homogenized using the Bullet Blender Blue Homogenizer with Air Cooling (Ideal Scientific, Ancaster, ON, Canada) for 3 minutes. Samples were then placed on a mutator for 1 hour at 4OC, after which they were spun at top speed for 10 minutes at 4OC; the supernatant was transferred to a fresh tube. Protein concentration was measured using Pierce BCA Protein Kit (Thermo Scientific, Ottawa, ON, Canada), according to manufacturer’s instructions.

39

2.2.6.2. Western blotting

Twenty-five µg of total protein [diluted in RIPA Buffer with phosphatase and protease inhibitors added (as described above), and 4X Loading Dye (Invitrogen, Burlington, ON, Canada) with 10% β-mercaptoethanol (Fisher Scientific, Ottawa, ON, Canada)] were electrophoresed in 1x TG-SDS Buffer (Wisent Inc., St. Bruno, QC, Canada) at 50-100V, on 4-20% Mini-PROTEAN TGXTM pre-cast Gels, (Bio-Rad, Mississauga, ON, Canada). Following electrophoresis, proteins were transferred onto 0.2µm PVDF membrane, Trans-Blot TurboTM Transfer Pack, Midi format (Bio-Rad, Mississauga, ON, Canada) using the Trans-Blot Turbo Transfer System (Bio-Rad, Mississauga, ON, Canada). Immediately following the transfer, membranes were blocked in 5% milk/TBS-T (Blotting-Grade Blocker, Bio-Rad, Mississauga, ON, Canada; TWEEN®20, Sigma-Aldrich, Oakville, ON, Canada) for 1 hour at room temperature. All primary antibodies were applied overnight at 4OC (see Table 5 for dilutions) in 5% milk/TBS-T. The following day, membranes were washed with 0.001% TBST, 3x 20 minutes. Membranes were incubated for 1 hour at RT in respective secondary antibodies (GE Healthcare UK Limited, UK) diluted 1:3,000 in 5% milk-TBST. Following three washes in TBST (20 minutes each), membranes were developed using Western Lightning Plus-ECL (Thermo Scientific, Ottawa, ON, Canada) on Premium Autoradiography Film (Denville Scientific, South Plainfield, NJ, USA). Band intensities within linear range were quantified using Quantity One software (Bio-Rad, Mississauga, ON, Canada). Protein of interest levels were normalized to a housekeeping protein (α-tubulin or β-actin). Protein expression under each treatment was further compared to its respective vehicle control (set as 1).

Table 5: Antibodies and dilutions

Antigen Origin Size (kDa) Dilution Supplier

HO-1 Rabbit 32 1:500 (western) Enzo Life Sciences Burlington, ON, Canada PPAR-γ Rabbit 54-57 1:100 (western) SantaCruz Biotechnology 1:100 (F-IHC) Dallas, TX, USA Phospho-PPAR-γ Rabbit 57 1:500 (western) Sigma-Aldrich Oakville, ON, Canada E-cadherin Mouse 110 1:1000 (western) Abcam 1:100 (F-IHC) Toronto, ON, Canada Lamin B Mouse 67 1:500 (western) SantaCruz Biotechnology Dallas, TX, USA α-Tubulin Mouse 50 1:2000 (western) Sigma Oakville, ON, Canada β-Actin Mouse 42 1:2000 (western) Abcam Toronto, ON, Canada

40

2.2.6.3. Enzyme-linked immunosorbent assay (ELISA)

Protein release from BeWo cells into the media was quantified using enzyme-linked immunosorbent assays (ELISA). Conditioned media was collected into 96-well plates and was first centrifuged at 425g for 5 minutes at RT after which the supernatant was transferred to a new plate. Protein concentration in the media was measured using Pierce BCA Protein Kit (Thermo Scientific, Ottawa, ON, Canada), according to manufacturer’s instructions. Samples were stored at -80OC until further analyses.

2.2.6.3.1. Free β-hCG ELISA

Free β-hCG ELISA Kit (Phoenix Pharmaceuticals, Inc., Burlingame, CA, USA) was used to measure β-hCG release in conditioned media following 48 hours of culture. The ELISA was conducted according to manufacturer’s instructions. Briefly, 45µg of total protein (or 15µg of total protein from forskolin-treated samples) was used per reaction, diluted in D-PBS (Wisent Inc, St. Bruno, QC, Canada). Following incubation of samples and standards, the plate was washed with deionized water (3X) and incubated with Enzyme Conjugate Reagent. The plate was developed using the TMB Reagent; the reaction was stopped with the Stop Solution provided. Endpoint optical density was measured at 450nm with a reference wavelength of 655nm. β-hCG protein was quantified using the standard curve and protein levels were normalized to total released protein. β- hCG release under different treatments was further normalized to respective vehicle(s) (set as 1).

2.2.6.3.2. HO-1 ELISA

Human ImmunoSet HO-1 ELISA development set was purchased from Enzo Life Sciences, Burlington, ON, Canada; the final assay protocol was adapted from manufacturer’s instructions. Briefly, black bottom, half-volume, 96-well plates were used to plate the HO-1 capture antibody overnight. On the following day, plates were blocked for at least 1 hour at RT. Subsequent incubations were performed on a shaker at RT. Standards and samples diluted in Assay Buffer were incubated for 1 hour. After washing with Wash Buffer (4X), samples were incubated with detection antibody for 45 minutes. Following another set of washes, the plate was incubated for 20 minutes with SA-HRP conjugate. The plate was developed using QuantaBluTM Fluorogenic Peroxidase Substrate Kit (Thermo Scientific, Ottawa, ON, Canada). After 30 minutes of incubation with QuantaBlu working solution, stop solution was applied and excitation and emission maxima were measured at 325nm and 420nm, respectively. Amount of HO-1 protein in ng/mL was calculated

41 using the standard curve. HO-1 protein levels were normalized to total released protein. HO-1 release under different treatments was further normalized to the corresponding vehicle (set as 1).

2.2.6.4. Fluorescent immunohistochemistry (F-IHC)

Immunofluorescence was used to visualize PPAR-γ protein localization and expression in BeWo cells under different treatments and to study cell fusion using e-cadherin as a cell membrane marker. The following protocol was used for the experiment. Coverslips were submerged into 100% ethanol and left to dry under UV light for 30 minutes. BeWo cells were plated on coverslips placed into 6-well plates at a density of 150,000 cells per well. Cells were treated for 1, 3, 6, 24, and 48 hours the next day. Following treatment, cells were washed in ice-cold D-PBS (Wisent Inc, St. Bruno, QC, Canada) and fixed in 1:1 methanol:acetone solution for 3 minutes on ice. Cells were permeabilized in 0.2% TritonX-100 (Fisher Scientific, Ottawa, ON, Canada) for 5 minutes on ice, washed, and blocked for 1 hour at RT in Protein Block Serum-Free, Ready-to-use (DAKO, Carpinteria, CA, USA). Primary antibodies for PPAR-γ and e-cadherin (see Table 5 for dilutions) were diluted in the same blocking solution and incubated overnight at 4OC. The following day, PPAR-γ signal was amplified with anti-rabbit biotinylated antibody (diluted 1:300 in blocking solution) for 1 hour at RT. Lastly, DAPI (1:1,000; Sigma-Aldrich, Oakville, ON, Canada), anti-mouse Alexa546 antibody (1:200; Invitrogen, Burlington, ON, Canada) and SA-Alexa488 (1:1,000; Invitrogen, Burlington, ON, Canada), all diluted in blocking solution, were incubated for 1 hour at RT. Coverslips were mounted on slides using a Immu-Mount mounting medium (Thermo Scientific, Ottawa, ON, Canada). Slides were left to dry in the dark and stored at 4OC until further analysis. Fluorescent microscopy was performed using the Spinning Disc Confocal Microscope (DMI6000B, Leica Microsystems, Concord, ON, Canada). Z stacks were taken using Volocity software, Version 5.3.0 (PerkinElmer, Woodbridge, ON, Canada) and deconvolved using Huygens Essential software, Version 4.2.2 (Scientific Volume Imaging, Hilversum, The Netherlands). All images were taken on the same day, under the same acquisition settings, normalized to the highest PPAR-γ expression under T0070907 treatment, to minimize variability and allow parallel comparison in protein expression across time points and treatments. For fusion visualization, images were taken at 200X magnification; for PPAR-γ localization analysis, z stacks were taken at 630X magnification.

42

2.2.6.4.1. Validation of E-cadherin as a cell membrane marker

E-cadherin was chosen as a cell surface marker used to visualize cell fusion in BeWo cells. Although its expression has been shown to decrease upon BeWo cell differentiation [163], we did not see a difference in E-cadherin protein expression under rosiglitazone, T0070907 or forskolin treatments (when normalized to α-tubulin as a housekeeping protein; Figure 17).

1.5 1.5

1.0 1.0

0.5 0.5

(compared to vehicle) to (compared

(compared to vehicle) to (compared

Relative e-cadherin expression e-cadherin Relative Relative e-cadherin expression e-cadherin Relative

0.0 0.0 Vehicle Ro10M T1M For20M Vehicle Ro10M T1M For20M Treatment Treatment

Figure 17: E-cadherin expression in response to different drug treatments. E-cadherin protein expression was assessed at 24 (A) and 48 (B) hours of treatment. Protein expression was normalized to α-tubulin levels; expression under each drug treatment was compared to respective vehicle control (set as 1, red dashed line). No significant differences were observed. Values expressed as mean±SEM (n=3). Ro, rosiglitazone; T, T0070907; For, forskolin.

2.2.7. Luciferase assay

Two putative PPAR-γ binding consensus sequences were identified within 250bp upstream region of the human GCM-1 promoter using the Gene2Promoter software (Version 6.3, Genomix, Germany; see Figure 18 for gene map). Oligonucleotides of these sites were synthesised and linkers added for cloning into the pGL4.10[luc2] cloning vector (binding sequences 1 and 2; Promega, Madison, WI, USA). Sites were mutated to generate sequence-based controls (mutated sequences 1 and 2). Binding and mutated sequences are outlined in Figure 19. Oligonucleotides and vectors were digested independently, purified and ligated. One-Shot TOP10 Chemically Competent E. coli cells (Invitrogen, Burlington, ON, Canada) were transformed and plated on ampicillin-positive plates. Antibiotic-resistant clones were selected and grown; plasmids (including the control renilla gene-containing plasmid) were isolated using Plasmid Plus Midi Kit (Qiagen, Toronto, ON, Canada) and sequenced to confirm sequence insertion. Quality and amount of plasmid DNA was confirmed using the NanoDrop 1000 (Thermo Scientific, Ottawa, ON, Canada). All plasmids were brought to similar concentrations in TE Buffer.

43

Figure 18: Gene map of the 1kb upstream region of the human GCM-1. The 1kb region upstream of the human GCM-1 gene was analyzed for PPAR-γ binding sites. Two putative PPREs were found within 250 base pairs from the transcription start site (sequence 1 and sequence 2). The gene map does not show the full GCM-1 DNA sequence; only exons 1 and 2 are displayed (map not to scale).

Figure 19: Plasmid design and oligonucleotide sequences in the luciferase reporter assay. (A) Plasmid map of the pGL4.10[luc2] vector (adapted from Promega Inc. protocol). (B) Duplex oligonucleotide sequences for binding sequence 1 (top) and mutated sequence 1 (bottom). (C) Duplex oligonucleotide sequences for binding sequence 2 (top) and mutated sequence 2 (bottom). Nucleotides conserved between the consensus PPRE (shown above for comparison) and our identified binding sites are bolded. Mutated nucleotides are shown in red. PPAR-γ binding site is highlighted in pink, RXR-α binding site shown in blue.

For transfection experiments, BeWo cells were seeded at 12,500 cells/well in clear-bottom tissue culture-suitable 96-well plates (Greiner bio-one, Monroe, NC, USA) to obtain 30% confluency. The following day, cells were transfected using ExGene 500 transfection reagent (Fermentas, Pittsburgh, PA, USA) and transfection efficiency was confirmed using a GFP-expressing control

44 plasmid. Transfection optimization experiments established that 200-300ng of plasmid DNA was optimal for 70-90% transfection efficiency using 0.7µL of transfection reagent. Transfection mixes were generated for binding sequence 1, binding sequence 2, mutated sequence 1 and mutated sequence 2. Each mix contained a 10:1 composition of experimental vector to a co-reporter vector (renilla). Following transfection mix application to the cells, culture plates were centrifuged for 5 minutes at 300g to improve transfection efficiency, according to the manufacturer’s protocol. Eight hours following transfection, media containing different chemicals was applied (rosiglitazone, T0070907, both or vehicle). Cells were cultured for additional 24 hours, following which media was removed, cells were lysed using Passive Lysis Buffer and the assay was performed according to Dual-Luciferase Reporter Assay System protocol (Promega, Madison, WI, USA). Cell extracts were kept on ice and luciferase activity was measured using an automated photometer. First, following Luciferase Assay Reagent II application, firefly luciferase activity was measured. Then, well contents were quenched with the Stop & Glo Reagent which also allowed the measurement of renilla luciferase activity. Renilla luciferase activity was an internal calibrator used to correct for differences in cell density and transfection efficiency between wells. Relative luciferase activity under the treatment with each plasmid (binding sequences 1 and 2) and drug was normalized to its corresponding vehicle control. Furthermore, to reduce random allosteric sequence-dependent background signal, signal obtained with each binding sequence was normalized to the corresponding mutated sequence.

2.2.8. Cell proliferation assay

Relative BeWo cell numbers were assessed using the CellTiter-FluorTM Cell Viability Assay (Promega, Madison, WI, USA). In 96-well cell culture-suitable plates, cells were seeded at 10,000 cells per well, in 50µL of media. Cells were treated the following day with drugs diluted in 50µL of media. Cell number was measured at 48 hours of culture. The manufacturer’s protocol was adapted as follows. The GF-AFC Substrate was diluted 1:1,000 in Assay Buffer to make CellTiter-FluorTM Reagent. Next, 15µL of D-PBS and 65µL of CellTiter-FluorTM Reagent were added to each well and incubated for 45 minutes at 37OC, protected from light. The following controls were included in the measurement: 1) maximum lysis control (to validate the absence of a false positive result), 2) no cell, media only control (to use as a background fluorescence measurement), and 3) no cell, media with vehicle control (to account for the effect of vehicle on fluorescence). The plate was read with excitation and emission maxima set at 400nm and 505nm, respectively. Recorded optical density (OD) values were blanked using the media only/no cell control to account for background

45 fluorescence; no differences were found between controls 2) and 3). Fluorescent measurements obtained with this assay represent an indication of the relative BeWo cell number, such that cell proliferation can be assessed between different drug treatments relative to their respective controls (set as 1).

2.2.9. Cellular fractionation analysis

Cellular fractionation was performed using the Nuclear Extract Kit (Active Motif, Burlington, ON, Canada). The protocol was adapted as follows. Cells in a 24-well plate were collected 1, 6, 24 hours post-treatment in 180µL PBS + Phosphatase inhibitor cocktail. Cells were centrifuged for 5 minutes at 500rpm at 4OC. After discarding the supernatant, the pellet was resuspended in 300µL 1X hypotonic buffer. Following a 15-min incubation on ice, 15µL of detergent were added and mixture was vortexed. Then it was centrifuged for 30 seconds at 14,000g at 4OC and the supernatant (ie, the cytoplasmic fraction) was collected. The remaining pellet was resuspended in 50µL of Complete Lysis Buffer, vortexed, and incubated for 30 minutes on a rocking platform at 4OC. The suspension was centrifuged for 10 minutes at 14,000g at 4OC. The supernatant (ie, the nuclear fraction) was collected. Protein concentration was measured using the Pierce BCA Protein Kit (Thermo Scientific, Ottawa, ON, Canada). Protein amount was normalized and prepared for further analysis using Western blotting [for phospho-PPAR-γ, PPAR-γ, α-tubulin (cytoplasmic protein) and Lamin B (nuclear protein)].

2.2.10. Statistical analysis

Experiments were performed in technical duplicates of at least three biological replicates. Data are represented as mean ± standard error of the mean. Student’s t-test was used to compare drug treatments to corresponding vehicle controls. One-way ANOVA and Bonferroni's Multiple Comparison Test were used to compare between treatment groups. Data in treatment groups were normalized relative to corresponding vehicle controls. All statistical calculations were performed using GraphPad Prism 5.2 software. P values <0.05 were considered significant.

46

Chapter 3: RESULTS

47

3.1. Placental gene expression in Gcm-1 heterozygous mice

To study the effect of Gcm-1 downregulation on the expression levels of Ppar-γ and Hmox-1, we utilized the Gcm-1 heterozygous CD1 mouse model. Labyrinth regions of wildtype and heterozygous placentas were collected at days E13.5 and E17.5 corresponding to mid- and late gestation, respectively. To dissect the connection between Ppar-γ, Hmox-1 and Gcm-1, mRNA levels of Ppar-γ, Hmox-1, Hmox-2 and Gcm-1 were assessed using qRT-PCR; expression of these genes in heterozygous placentas was normalized to wildtype placental expression (set as 1). Firstly, levels of Gcm-1 were compared between heterozygous and wildtype placentas. Then, Hmox-1 expression was studied to determine whether a decrease in Gcm-1 had an effect on Hmox-1 expression in the placenta. Furthermore, levels of Hmox-2 were assessed to study possible compensatory effect in the event of Hmox-1 expression being altered (as has been described previously [144]) (Figure 20). As expected, Ppar-γ expression was not altered in heterozygous placentas at E13.5 and E17.5 despite a 50% downregulation of Gcm-1 expression in these placentas (p<0.0001, n=6). Contrary to our expectations, levels of Hmox-1 and Hmox-2 also did not differ between wildtype and heterozygous placentas at E13.5 and E17.5. Our findings suggest that, in the mouse placenta, Gcm-1 does not regulate Hmox-1 or Ppar-γ levels.

Figure 20: Placental Hmox-1, Hmox-2 or Ppar-γ expression is unaltered in Gcm-1+/- mice. Placentas from Gcm-1 +/- mice were tested for their expression of Hmox-1, Hmox-2 and Ppar-γ. Relative mRNA levels of genes of interest were measured at E13.5 (A) and E17.5 (B) using qRT- PCR. All values are compared to corresponding wildtype (WT) pups from the same mother. Values represented as mean ± SEM; ****p<0.0001 (n=6).

48

3.2. GCM-1 silencing does not affect HO-1 and PPAR-γ expression in the BeWo cell model

The human choriocarcinoma cell line BeWo was used as an in vitro system for studying how GCM-1 downregulation influences expression of PPAR-γ and HO-1, based on the finding that BeWo cells express all three genes (see Figure 12). mRNA expression was measured following 48 hours of silencing and/or treatment (Figure 21). Using a combination of two siRNA sequences (815+201), GCM-1 expression was downregulated by 62±0.1% compared to non-silencing (NS) control, set as 1 (p≤0.05, n=3) (Figure 21A). Treatment with forskolin induced GCM-1 expression 1.8±0.3-fold in NS- transfected cells (p≤0.05, n=4); when the drug was combined with GCM-1-targeting siRNA, the levels of GCM-1 did not differ compared to NS. Expression of HO-1 did not change with any of the treatments (Figure 21B). PPAR-γ expression showed a mild (~22%) downregulation following siRNA treatment, but this did not reach statistical significance. Addition of forskolin decreased PPAR-γ levels by 46±10.0% (p≤0.05, n=3; Figure 21C). Confirming our findings from the Gcm-1 heterozygous mouse model, we did not see a GCM-1-dependent HO-1 expression mechanism in the BeWo cell line.

A 2.5 *** B 1.5 * 2.0 1.0

1.5 expression 1.0 expression

0.5

HO-1

GCM-1 (normalized to NS control) NS to (normalized (normalized to NS control) NS to (normalized 0.5 *

0.0 0.0 NS siRNA NS+Forsk siRNA+Forsk NS siRNA NS+Forsk siRNA+Forsk

Treatment Treatment

C 1.5

1.0 expression  *

0.5

PPAR- (normalized to NS control) NS to (normalized

0.0 NS siRNA NS+Forsk siRNA+Forsk

Treatment

Figure 21: Downregulation of GCM-1 in BeWo cells has no effect on HO-1 and PPAR-γ levels. BeWo cells were treated with siRNA against human GCM-1, non-silencing control, and forskolin for 48 hours. GCM-1 (A), HO-1 (B) and PPAR-γ (C) mRNA levels were measured and compared to NS control (set as 1, red line). Values represented as mean ± SEM, n=3-4; *p≤0.05 vs NS control; ***p≤0.001. NS, non-silencing; Forsk, forskolin.

49

3.3. Time- and dose- responses to drug treatments in BeWo cells

The effects of rosiglitazone (PPAR-γ agonist), T0070907 (PPAR-γ antagonist) and forskolin (cAMP inducer), either alone or in combinations, were tested for their efficacy and toxicity to BeWo cells over time. Toxicity results are shown in Figure 14. Gene expression profiles for GCM-1, PPAR-γ and HO-1 at 3, 6 and 24 hours of treatment are shown in Figures 22-24. Concentrations tested were chosen either based on previously published work or our initial experiments. The following concentrations were used: for rosiglitazone, 10µM and 100µM; for T0070907, 0.1µM, 1µM and 10µM; and for forskolin, 20µM [57]. Based on our findings, for subsequent experiments, rosiglitazone and T0070907 were primarily used at 10µM and 1µM, respectively.

3.3.1. Expression of HO-1

HO-1 mRNA expression in response to PPAR-γ activity-modulating drugs was assessed at 3, 6 and 24 hours post-treatment and compared to vehicle control (set as 1; Figure 22 A-C). Lower concentration of rosiglitazone induced HO-1 expression by 2.4±0.2, 2.6±0.1 and 2.7±0.1-fold at 3, 6 and 24 hours, respectively (p≤0.05, n=3-4). Analogous findings were obtained following treatment with the higher dose, where HO-1 was induced more than 2-fold at all three time points studied (p≤0.05, n=3-4). Activation of HO-1 expression with rosiglitazone was a robust and maintained event, happening as early as 3 hours and persisting until 24 hours. All concentrations of T0070907 (0.1µM, 1µM, 10µM) significantly blocked HO-1 expression as early as 3 hours of culture (down to 59±2.8%, 66±9.3% and 65±4.4%, from lowest to highest dose) and decreased HO-1 expression by 68±3.9%, 67±2.7% and 34±6.9% (lowest to highest dose) by 24 hours (p≤0.05). By 24 hours, all treatments of T0070907 were able to outcompete the activation of PPAR-γ induced by rosiglitazone; however, 1µM concentration was the most effective dose decreasing HO-1 expression by 62±0.6% (vs vehicle) at 24 hours (p≤0.05). Forskolin blocked the expression of HO-1 at 3 and 6 hours by 40±4.6% and 68±3.5%, respectively (p≤0.05), but this effect diminished by 24 hours of treatment. Additive effects on HO-1 expression were seen at 24 hours when forskolin was combined with T0070907 or rosiglitazone, where gene expression was further downregulated by 60±4.3% or upregulated by 48±13.0%, respectively (p≤0.05).

50

3 B 3 A 3hrs * 6hrs * * *

2 2

mRNA

mRNA

HO1 HO1

1

1 levels expression

expression levels expression Relative Relative * * * * * * * * * * * 0 0 M M M M M M M M M M M M M M M M M M M M M M                       T1 T1 Vehicle T0.1 T10 Vehicle T0.1 T10 Ro10 M+T1 For20 M+T1 Ro10 M+T1 For20 M+T1 Ro100 M+T0.1  M+T10  Ro100  M+T10    M+T0.1  M+Ro100  M+Ro100   Ro10 For20 Ro10 Ro10 Ro10 Ro10 For20 Ro10 For20 For20

C 3 * 24hrs * 2

mRNA * HO1

1 expression levels expression

Relative * * * * * * * 0

M M M M M M M M M M M            T1 T0.1 T10 Vehicle Ro10 M+T1 For20 M+T1 Ro100  M+T10  M+T0.1   M+Ro100  Ro10 For20 Ro10 Ro10 For20

Figure 22: Dose-dependent expression of HO-1 in BeWo cells over time. Time- and concentration-course experiments were performed to study expression of HO-1 at 3 (A), 6 (B) and 24 (C) hours post-treatment. All treatments are compared to their respective vehicle controls (set as 1, red dashed line). Values are represented as mean ± SEM; *p≤0.05 vs vehicle control (n=3-4). Ro, rosiglitazone; T, T0070907; For, forskolin.

3.3.2. Expression of PPAR-γ

PPAR-γ expression in response to drug treatments was assessed at 3, 6 and 24 hours of treatment and compared to vehicle control (Figure 23 A-C). Although changes at 3 hours of treatment were less pronounced, by 6 and 24 hours, responses to agonist and antagonist treatments were significantly different when compared to vehicle control. Following activation of PPAR-γ with rosiglitazone (10µM), PPAR-γ expression decreased to 47±3.4% by 24 hours (p≤0.05, n=4). Conversely, blocking PPAR-γ activity with T0070907 (1µM) resulted in a significant upregulation of its expression by 24 hours (2.0±0.1-fold, p≤0.05, n=4). Forskolin alone did not have a significant effect on PPAR-γ expression at 24 hours, while forskolin with T0070907 increased gene expression 1.9±0.1-fold and addition of rosiglitazone significantly decreased expression by 60±4.1% (p≤0.05, n=4).

51

A 2.5 B 3hrs 2.5 * 6hrs * 2.0 2.0

* * * mRNA mRNA * 

1.5  1.5 PPAR-

1.0 PPAR- 1.0

expressionlevels

expressionlevels Relative 0.5 Relative 0.5 * * * 0.0 0.0

M M M M M M M M M M M M M M M M M M M M M M                       T1 T1 Vehicle T0.1 T10 T0.1 T10 Ro10 M+T1 For20 M+T1 Vehicle Ro10 M+T1 For20 M+T1 Ro100  M+T10  Ro100  M+T10  M+T0.1  M+T0.1   M+Ro100  M+Ro100   Ro10 For20 Ro10 For20 Ro10 Ro10 Ro10 Ro10 For20 For20

C 2.5 24hrs * * * 2.0 *

* * mRNA  1.5

PPAR- 1.0 expression levels expression Relative 0.5 * * *

0.0

M M M M M M M M M M M            T1 T0.1 T10 Vehicle Ro10 M+T1 For20 M+T1 Ro100  M+T10  M+T0.1   M+Ro100  Ro10 For20 Ro10 Ro10 For20

Figure 23: Dose-dependent expression of PPAR-γ in BeWo cells over time. Time- and concentration-course experiments were performed to study expression of PPAR-γ at 3 (A), 6 (B) and 24 (C) hours post-treatment. All treatments are compared to their respective vehicle controls (set as 1, red dashed line). Values are represented as mean ± SEM; *p≤0.05 vs vehicle control (n=3-4). Ro, rosiglitazone; T, T0070907; For, forskolin.

3.3.3. Expression of GCM-1

Lastly, the effect of PPAR-γ activity modulation on GCM-1 expression levels was tested at 3, 6 and 24 hours (Figure 24 A-C). It has been previously described that Ppar-γ regulates Gcm-1 expression in mouse trophoblast stem cells [101]. Therefore, we studied GCM-1 response to rosiglitazone and T0070907 treatments at different time points. Activation of PPAR-γ with the agonist (at 10µM concentration) led to a transient and weak induction of GCM-1 expression, evident only at 3 hours of treatment (1.5±0.1-fold, p≤0.05, n=4). Blocking PPAR-γ activity with T0070907 (1µM) produced a more robust and prolonged reduction of GCM-1 mRNA expression which persisted towards 24 hours and led to a 60±5.4% decrease (p≤0.05). Furthermore, when both rosiglitazone and T0070907 were combined, there was a 34±8.2% reduction in GCM-1 mRNA expression at 24 hours vs the vehicle (p≤0.05). As expected, forskolin alone significantly

52 upregulated GCM-1 expression in BeWo cells as early as 3 hours of treatment (2.9±0.2-fold increase, p≤0.05); this effect was also evident at 6 hours (2.3±0.1-fold rise, p≤0.05) and 24 hours (2.9±0.4- fold change, p≤0.05).

A B 4 3hrs 4 6hrs

* * *

3 * 3 mRNA mRNA *

2 2 GCM-1

GCM-1 * *

expressionlevels

expression levels expression Relative Relative 1 1 * * * * *

0 0

M M M M M M M M M M M M M M M M M M M M M M                       T1 T1 T0.1 T10 Vehicle T0.1 T10 Vehicle Ro10 M+T1 For20 M+T1 Ro10 M+T1 For20 M+T1 Ro100  M+T10  Ro100 M+T0.1  M+T10  M+T0.1     M+Ro100 M+Ro100   Ro10 For20 Ro10 For20 Ro10 Ro10 Ro10 Ro10 For20 For20

C 4 24hrs *

3 mRNA

2

GCM-1 expressionlevels

Relative 1 * * * * * 0

M M M M M M M M M M M            T1 T0.1 T10 Vehicle Ro10 M+T1 For20 M+T1 Ro100  M+T10  M+T0.1   M+Ro100  Ro10 For20 Ro10 Ro10 For20

Figure 24: Dose-dependent expression of GCM-1 in BeWo cells over time. Time- and concentration-course experiments were performed to study expression of GCM-1at 3 (A), 6 (B) and 24 (C) hours post-treatment. All treatments are compared to their respective vehicle controls (set as 1, red dashed line). Values are represented as mean ± SEM; *p≤0.05 vs vehicle control (n=3-4). Ro, rosiglitazone; T, T0070907; For, forskolin.

3.4. Effect of PPAR-γ activity-modulating drugs on HO-1 expression in BeWo cells

The effect of manipulating PPAR-γ activity on HO-1 expression was assessed. First, the expression of HO-1 was tested following 3, 6 and 24 hours of treatment with the stimulators and inhibitors at optimal concentrations. Second, HO-1 protein expression in BeWo cell lysates was assessed using Western blotting under the same conditions. Lastly, HO-1 protein release from BeWo cells was studied using ELISA.

53

3.4.1. PPAR-γ activation induces a rapid rise in HO-1 RNA levels

Using rosiglitazone as a stimulator and T0070907 as an inhibitor of PPAR-γ activity, we tested the response of HO-1 to PPAR-γ activity-modulating drugs at the mRNA level. HO-1 mRNA responses to drugs at their optimal concentrations at 3, 6 and 24 hours are compiled in Figure 25 (results are described in detail in section 3.3.1). Briefly, PPAR-γ stimulation induced HO-1 expression in BeWo cells as has been described in other systems [114]. This response was robust and prolonged, as it was maintained at 24 hours of treatment. T0070907 decreased gene expression on its own, as well as outcompeted rosiglitazone when both drugs were co-administered at 3, 6 and 24 hours (p≤0.001 vs rosiglitazone alone). Optimal drug concentrations, selected based on mRNA studies, were used in subsequent protein analyses, namely, cellular HO-1 protein expression and HO-1 release.

*** *** 3 *** # # #

2 3hrs mRNA 6hrs

HO-1 24hrs 1 #

expression levels # Relative # # # #

0 Rosi(10M) T007 (1M) Rosi+T007 Forskolin

Figure 25: HO-1 mRNA expression levels following modulation of PPAR-γ activity. Compiled HO-1 expression levels at optimal drug concentrations overtime are shown. All treatments are compared to their respective vehicle controls (set as 1, red dashed line). Values are represented as mean ± SEM; #p≤0.05 vs vehicle control; ***p≤0.001 (n=3-4). Ro, rosiglitazone; T, T0070907; For, forskolin.

3.4.2. Intracellular HO-1 protein expression increases following PPAR-γ activation

To validate our mRNA findings, we assessed intracellular HO-1 protein expression at 24 and 48 hours of treatment using Western blotting (Figure 26 A,B). Consistent with gene expression results, HO-1 protein levels increased significantly following rosiglitazone treatment at 24 (Figure 26 A) and 48 (Figure 26 B) hours. At 24 hours, lower dose of rosiglitazone upregulated HO-1 expression by 54±14.8% (p≤0.05, n=4), while co-administration of T0070907 decreased HO-1

54 protein levels by 40±5.5% compared to baseline (p≤0.001 vs rosiglitazone alone). Higher agonist dose also upregulated HO-1 expression by 69±20.7% (p≤0.05). Conversely, inhibition of PPAR-γ activity (at 1µM dose of T0070907) led to a downregulation of HO-1 protein expression by 83±16.8% (p≤0.05). Forskolin treatment decreased HO-1 expression by 38±13.1%, however, this did not reach significance (p=0.1011). At 48 hours of treatment, analogous results were attained. Lower and higher rosiglitazone concentrations increased HO-1 expression by 49±25.2% (p=0.1926) and 86±24.6% (p≤0.05, n=4), respectively. T0070907 (1µM) decreased protein levels by 83±16.2% (p≤0.05); and the combination of rosiglitazone and T0070907 was significantly different from antagonist treatment alone (p≤0.01). Although HO-1 expression decreased following forskolin treatment at 24 hours, this effect was diminished by 48 hours.

A 24 hours B 48 hours

2.5 2.5 *** * 2.0 * 2.0 * ** 1.5 1.5

1.0 1.0

(relative to vehicle) to (relative (relative to vehicle) to (relative

Cellular HO-1 expression HO-1 Cellular * Cellular HO-1 expression HO-1 Cellular 0.5 * 0.5 * * *

0.0 0.0

M M M M M M M M M M M M M M               T1 T1 T10 T10 Vehicle Ro10 Vehicle Ro10 M+T1 For20 Ro100 M+T1 For20 Ro100  M+T10  M+T10   Ro10 Ro10 Ro10 Ro10 Treatment Treatment C D

Figure 26: Expression of HO-1 protein in BeWo cells. BeWo cells were treated for 24 (A) and 48 (B) hours. HO-1 protein levels were assessed using Western blotting and normalized to α-tubulin; each treatment was expressed as fold-change relative to its vehicle control (set as 1, red dashed line). Representative Western blots for expression at 24 (C) and 48 (D) hours are shown. Values are represented as mean ± SEM; ***p≤0.001; **p≤0.01; *p≤0.05 vs vehicle control (n=4). Ro, rosiglitazone; T, T0070907; For, forskolin.

55

3.4.3. HO-1 release from BeWo cells increases with PPAR-γ activation

Based on previous work in our laboratory, HO-1 is released into the medium from placental explants (unpublished observations). We tested whether this observation holds true in BeWo cells. BeWo cell conditioned medium was collected following 24 and 48 hours of treatment with PPAR-γ activity-modulating drugs and HO-1 protein content was measured by ELISA (Figure 27 A,B). By 48 hours of rosiglitazone exposure, HO-1 release from BeWo cells into the media was significantly upregulated by 57±24.8% and 88±26.7% corresponding to lower and higher drug concentrations, respectively (p≤0.05, n=4; Figure 27B). As expected, HO-1 release was decreased by 34±6.7% following T0070907 treatment (1µM, p≤0.05). Furthermore, the presence of T0070907 significantly blocked the induction associated with rosiglitazone (p≤0.05 vs rosiglitazone alone) by 48 hours. Consistent with mRNA and cellular protein expression profiles, forskolin inhibited HO-1 release from BeWo cells at 48 hours (down by 39±7.4%, p≤0.05). This effect was exacerbated with the addition of T0070907 to forskolin treatment (45±13.5% reduction, p≤0.05). Rosiglitazone, in combination with forskolin, led to a significantly higher HO-1 release when compared to forskolin alone (p≤0.01).

A 24 hours B 48 hours

2.5 2.5 * ** # # # 2.0 2.0 #

1.5 1.5

1.0 1.0 #

# # # # #

(relative to vehicle) to (relative

(relative to vehicle) to (relative

HO-1 release HO-1 HO-1 release HO-1 0.5 0.5

0.0 0.0

M M M M M M M M M M M M M M               T1 T1 Vehicle Vehicle Ro10 M+T1 For20 Ro10 M+T1 For20 Ro100  For+T1 Ro100  For+T1 For+Ro100 For+Ro100 Ro10 Ro10

Treatment Treatment

Figure 27: HO-1 release from BeWo cells. HO-1 release into culture media was assessed at 24 (A) and 48 (B) hours of treatment. HO-1 protein release was normalized to total protein content in conditioned media. Each drug treatment was compared to its respective vehicle control (set as 1, red dashed line). Values are represented as mean ± SEM; **p≤0.01; *p≤0.05; #p≤0.05 vs vehicle control (n=4). Ro, rosiglitazone; T, T0070907; For, forskolin.

56

3.5. PPAR-γ participates in a negative auto-regulatory feedback mechanism

Based on our time- and dose-response experiments, we observed that PPAR-γ participates in a negative auto-regulatory feedback, such that induction of its activity with rosiglitazone resulted in decreased expression, and blocking receptor activity led to increased receptor levels. To study this further, we assessed PPAR-γ mRNA expression and cellular PPAR-γ protein expression, receptor phosphorylation and localization.

3.5.1. PPAR-γ expression is inversely related to PPAR-γ activity

The expression of PPAR-γ was studied at 3, 6 and 24 hours following drug treatment; levels at optimized drug concentrations are shown in Figure 28. Results are also summarized in section 3.3.2. Briefly, induction of PPAR-γ activity resulted in a decrease of its expression, while treatment with T0070907 had the opposite effect, with most pronounced changes observed 24 hours after treatment. Co-administration of T0070907 with rosiglitazone significantly ameliorated rosiglitazone-induced PPAR-γ downregulation at 6 (p≤0.01) and 24 (p≤0.001) hours of treatment (vs rosiglitazone alone).

** ***

2.5 * #

2.0 # #

mRNA 3hrs  1.5 # 6hrs 24hrs PPAR- 1.0

expression levels #

0.5 # Relative

0.0 Rosi(10M) T007 (1M) Rosi+T007 Forskolin

Figure 28: PPAR-γ mRNA expression levels following modulation of PPAR-γ activity. Compiled PPAR-γ expression levels at optimal drug concentrations are shown overtime (3, 6, 24 hours). All treatments are compared to their respective vehicle controls (set as 1, red dashed line). Values are represented as mean ± SEM; #p≤0.05 vs vehicle control; ***p≤0.001; **p≤0.01; *p≤0.05 (n=3-4). Ro, rosiglitazone; T, T0070907; For, forskolin.

57

3.5.2. PPAR-γ protein levels correlate with gene expression profiles

To confirm our gene expression findings, we examined the cellular PPAR-γ protein expression at 24 and 48 hours following treatment (Figure 29 A,B). Total PPAR-γ protein levels were decreased by 58±5.3% and 66±9.8% at 24 and 48 hours, respectively, by rosiglitazone (10µM) treatment (p≤0.05, n=4). PPAR-γ expression was increased following treatment with T0070907 (1µM) 2.1±0.4-fold (p=0.0759) and 1.8±0.2-fold (p≤0.05) at 24 and 48 hours, respectively. Co- administration of the inhibitor with rosiglitazone significantly induced PPAR-γ expression at 24 hours (p≤0.01 vs rosiglitazone alone). PPAR-γ expression decreased by 30±9.2% following a 24- hour treatment with forskolin (p≤0.05). By 48 hours, effect of rosiglitazone was more pronounced when combined with T0070907, resulting in decreased PPAR-γ expression (p≤0.001 vs T0070907 alone).

A 24 hours B 48 hours 4 8 *** * ** 7 3 **

6

expression

expression  2  5

2 * (relative to vehicle) to (relative

(relative to vehicle) to (relative * 1 * 1 *

Cellular PPAR- * * Cellular PPAR- * * 0 0

M M M M M M M M M M M M M M               T1 T1 Vehicle T10 T10 Ro10 M+T1 For20 Vehicle Ro10 M+T1 For20 Ro100  M+T10 Ro100  M+T10   Ro10 Ro10 Ro10 Ro10 Treatment Treatment C D

Figure 29: Expression of PPAR-γ protein in BeWo cells. PPAR-γ protein levels were assessed 24 (A) and 48 (B) hours using Western blotting and normalized to α-tubulin; each treatment was further normalized to its representative vehicle control (set as 1, red dashed line). Representative Western blots for expression at 24 (C) and 48 hours (D) are shown. Values are represented as mean ± SEM; ***p≤0.001; **p≤0.01; *p≤0.05 vs vehicle control (n=4). Ro, rosiglitazone; T, T0070907; For, forskolin.

58

3.5.3. Phosphorylation of PPAR-γ protein is increased following rosiglitazone treatment

To further study the inverse relationship between PPAR-γ expression and activity, we looked at phosphorylation of the PPAR-γ protein (Figure 30 A,B). Levels of phosphorylated PPAR-γ were compared to total cellular PPAR-γ protein levels at 24 and 48 hours; the effects were more pronounced at the later time point (Figure 30B). By 48 hours of treatment, both lower and higher doses of rosiglitazone led to a rise of PPAR-γ phosphorylation by 2.3±0.9-fold (p=0.2772) and 4.5±0.1-fold (p≤0.05), respectively (n=4). Inhibition of PPAR-γ activity with T0070907 (1µM) resulted in a decrease of receptor phosphorylation at 48 hours by 17±1.5% (p≤0.05). By 48 hours of treatment, forskolin also decreased PPAR-γ phosphorylation levels by 39±5.6% (p≤0.05).

A 24 hours B 48 hours

6 6 ** *

*

expression) expression)  4

 4

protein expression

protein expression

 

2 2 *

* * * *

(normalized to total PPAR- total to (normalized

(normalized to total PPAR- total to (normalized Phospho-PPAR- Phospho-PPAR- 0 0

M M M M M M M M M M M M M M               T1 T1 Vehicle T10 T10 Ro10 M+T1 For20 Vehicle Ro10 M+T1 For20 Ro100  M+T10 Ro100  M+T10   Ro10 Ro10 Ro10 Ro10 Treatment Treatment C D

Figure 30: Expression of phosphorylated PPAR-γ protein in BeWo cells. Phospho-PPAR-γ (Ph-PPAR-γ) protein levels were assessed at 24 (A) and 48 (B) hours using Western blotting and normalized to total PPAR-γ protein levels. Each treatment was further represented as fold change relative to vehicle control (set as 1, red dashed line). Representative Western blots for expression at 24 (C) and 48 hours (D) are shown. Values are represented as mean ± SEM; ***p≤0.001; **p≤0.01; *p≤0.05 vs vehicle control (n=4). Ro, rosiglitazone; T, T0070907; For, forskolin.

59

3.5.4. Fluorescent immunohistochemistry analysis of PPAR-γ localization

PPAR-γ cellular localization and relative expression was assessed using fluorescent immunohistochemistry at 3, 6 and 24 hours of treatment with rosiglitazone (10µM), T0070907 (1µM) or vehicle (Figure 31). Our studies showed that PPAR-γ localized to the nucleus, with some expression seen in the cytoplasm. Concurrent with our PPAR-γ protein analysis using Western blotting, we saw increased expression of PPAR-γ after T0070907 treatment and a decrease in staining levels after rosiglitazone treatment. This trend persisted and became even more prominent at 48 hours of treatment (data not shown).

Figure 31: Effect of PPAR-γ-modulating drugs on its localization in BeWo cells. PPAR-γ expression was visualized at 3 (top), 6 (middle) and 24 (bottom) hours following treatment with vehicle, rosiglitazone or T0070907. PPAR-γ shown in green, e-cadherin (cell surface marker) in red, DAPI (nuclear marker) in blue (630X magnification).

60

3.5.5. Nuclear and cytoplasmic (phospho-) PPAR-γ expression

Cellular fractionation analysis was used to study potential changes in PPAR-γ protein localization at 1, 6 and 24 hours of treatment (Figure 32 A,B). Overall, expression of PPAR-γ was stronger in the nucleus compared to the cytoplasm as was expected based on our fluorescent studies. Within the nucleus, there was a trend towards a lower PPAR-γ protein expression with rosiglitazone treatment: a 55±13.3% decrease at 6 hours post-treatment (p≤0.05, n=4) and a 30±18.1% decrease following 24 hours (p=0.1). By contrast, T0070907 treatment led to a significant increase in nuclear PPAR-γ expression after 1 hour of treatment (2.3±0.6-fold, p≤0.05, n=4); although the trend persisted, the results were not significant at 6 (2.6±0.9-fold increase, p=0.1487) and 24 (1.8±0.5-fold induction, p=0.2309) hours. Combination of both chemicals did not alter nuclear PPAR-γ localization at either time point. Cytoplasmic PPAR-γ expression did not show pronounced changes (Figure 32B). Most of the treatments did not show significant results at 1, 6 and 24 hours (possibly because of n=3). There was a trend towards a lower PPAR-γ expression under rosiglitazone treatment at 6 hours (34±6.3% decrease, p≤0.05, n=3) and 24 hours (28±35.2% downregulation, p=0.5166). T0070907 results were not as pronounced in the cytoplasm as they were in the nucleus; there was no change in PPAR-γ expression at 1, 6 and 24 hours. A B 4 4 *

3 # 3 expression

 Vehicle

expression  Rosi(10M) 2 2 T007 (1M)

Rosi+T007 (compared to vehicle) to (compared 1 vehicle) (compared to 1 #

# # Relative nuclear PPAR- nuclear Relative

0 PPAR- cytoplasmic Relative 0 1hr 6hrs 24hrs 1hr 6hrs 24hrs Time Time C D

Figure 32: Cellular fractionation analysis of PPAR-γ expression in BeWo cells. Nuclear (A) and cytoplasmic (B) PPAR-γ protein expression was assessed at 1, 6 and 24 hours of treatment using Western blotting; nuclear expression was normalized to lamin B; cytoplasmic expression normalized to α-tubulin. Each treatment was further normalized to DMSO (vehicle, set as 1, red dashed line). Representative Western blots for nuclear (C) and cytoplasmic (D) expressions are shown. Values are represented as mean ± SEM; *p≤0.05; #p≤0.05 vs vehicle control (n=4). Rosi, rosiglitazone; T007, T0070907.

61

Lastly, levels of nuclear phosphorylated PPAR-γ were assessed at 1 and 24 hours of treatment (Figure 33). No significant changes were found in phosphorylated PPAR-γ protein expression at either time point.

1.5

expression 

1.0 Vehicle /PPAR-  Rosi(10M) T007 (1M)

0.5 Rosi+T007 (compared to vehicle) to (compared

0.0 Nuclear Phospho-PPAR- Nuclear 1hr 24hrs Time

Figure 33: Nuclear phosphorylated PPAR-γ expression in BeWo cells. Nuclear phosphorylated PPAR-γ (phospho-PPAR-γ) protein expression was assessed at 1 and 24 hours of treatment and normalized to total nuclear PPAR-γ expression using Western blotting. Each treatment was normalized to vehicle (set as 1, red dashed line). Values are represented as mean ± SEM (n=4). Rosi, rosiglitazone; T007, T0070907.

62

3.6. Role of PPAR-γ in proliferation/differentiation balance

The BeWo cell model was used to study the effect of pharmacological PPAR-γ activity modulation on the proliferation/differentiation balance, an important physiologic process within the trophoblast layer [57]. We assessed these events by studying markers of syncytial differentiation (such as GCM-1 expression and free β-hCG release), visualizing cell fusion and measuring BeWo cell proliferation.

3.6.1. PPAR-γ-mediated induction of GCM-1 expression is transient

GCM-1 mRNA expression levels were assessed at 3, 6 and 24 hours following treatment; mRNA responses at optimal drug concentrations are summarized in Figure 34 and discussed in detail in section 3.3.4. Briefly, forskolin was used as a positive control to show that GCM-1 induction was possible in the BeWo cell model. Treatment with rosiglitazone did not have a pronounced effect on GCM-1 expression except at 3 hours of culture. Conversely, blocking PPAR-γ activity with T0070907 led to a downregulation of GCM-1 expression which persisted towards 24 hours of treatment.

4 ** # ** # 3 3hrs mRNA # 6hrs 2 # GCM-1 24hrs

expression levels 1 # # # Relative #

0 Rosi(10M) T007 (1M) Rosi+T007 Forskolin

Figure 34: GCM-1 mRNA expression levels following modulation of PPAR-γ activity. Compiled GCM-1 expression levels at optimal drug concentrations are shown overtime (3, 6, 24 hours). All treatments are compared to their respective vehicle controls (set as 1, red dashed line). Values are represented as mean ± SEM; #p≤0.05 vs vehicle control; **p≤0.01; *p≤0.05 (n=3-4). Ro, rosiglitazone; T, T0070907; For, forskolin.

63

3.6.2. PPAR-γ activity in BeWo cells

In our mRNA expression experiments, we observed that although it was possible to block the activity of PPAR-γ and, consequently, decrease the expression of downstream targets, it was considerably more difficult to induce its activity and, thus, upregulate the expression of its target genes, particularly GCM-1. Thus, because of our inability to induce GCM-1 expression over a prolonged time in BeWo cells, we decided to study the transcriptional activity of PPAR-γ using the luciferase reporter assay (Figure 32 A,B). We identified two PPREs in the human GCM-1 promoter; these sequences were cloned into plasmids and transfected into BeWo cells (see Figure 17). PPAR-γ activity was studied at 24 hours following treatment. Relative promoter activity upon binding to sequence 1 and sequence 2 are shown in Figures 35A and 35B, respectively. Luciferase activity following its transcription under the control of binding sequence 1 was increased by 32±12.7% following treatment with rosiglitazone at 10µM, however, this failed to reach statistical significance (p=0.1274, n=4; Figure 35A). Higher dose of rosiglitazone (100µM) increased luciferase transcription levels by 46±11.8% (p≤0.05, n=4). Sequence 1 did not show a response to T0070907 alone; however, in combination with rosiglitazone (100µM), T0070907 significantly blocked the effect of rosiglitazone (p≤0.01 vs rosiglitazone alone).

A Binding sequence 1 B Binding sequence 2

2.0 2.0 ** * 1.5 1.5 *** 1.0 1.0

* (norm alized to vehicle)

0.5 (norm alized to vehicle)0.5

Relative promoter activity Relative promoter activity 0.0 0.0

M M M M M M M M M M          

Ro10 T0.1 T0.5 T0.1 T0.5 Vehicle Ro100 Vehicle Ro10 M+T0.5 Ro100 M+T0.5  

Ro100 Ro100 Treatment Treatment

Figure 35: PPAR-γ activity in BeWo cells. PPAR-γ activity in BeWo cells was assessed using the luciferase reporter assay. Two sequences identified as PPAR-γ binding sites in the GCM-1 promoter were used to test the activity of PPAR-γ in response following 24 hours post-treatment. PPAR-γ transcription of sequence 1 (A) and sequence 2 (B) following drug treatments are shown. Treatments are compared to vehicle (set as 1, red dashed line). Values are represented as mean ± SEM; ***p≤0.001; **p≤0.01; *p≤0.05 vs vehicle control (n=3-4). Ro, rosiglitazone; T, T0070907.

64

Luciferase activity under the control of sequence 2 was quite different from that observed under sequence 1 (Figure 35B). Here, rosiglitazone treatment did not have an effect on PPAR-γ activity, but T0070907 treatment led to a 38±3.6% reduction in transcription of the reporter gene (p≤0.05, n=4). Interestingly, a combination of rosiglitazone and T0070907 significantly increased receptor activity when compared to T0070907 alone (p≤0.001) indicating that rosiglitazone was still able to exert its effects on transcription.

3.6.3. Effect of PPAR-γ activity modulation on free β-hCG release

To further study the role of PPAR-γ on the maintenance of proliferation/differentiation balance, we assessed differentiation of BeWo cells by using free β-hCG release as a marker (Figure 36). In our experiments, forskolin was used as a positive control for cell differentiation since it is known to upregulate free β-hCG secretion. We saw a 20.9±2.8-fold induction in β-hCG release following forskolin treatment at 48 hours (p≤0.05, n=7). Lower dose of rosiglitazone (10µM) did not affect free β-hCG release, while higher concentration of the drug showed a trend toward an increase in β-hCG release (34±20.4% increase, p=0.1914, n=4). On the contrary, blocking PPAR-γ activity with T0070907 significantly downregulated β-hCG secretion by 41±7.3% (p≤0.05, n=7). Co- administration of both the inhibitor and the activator resulted in no change of β-hCG protein release.

25 #

20

15

2.0

1.5 * -hCG secretion -hCG

 1.0

# Free

(relative to vehicle control) 0.5

0.0

M M M M M      T1 Vehicle Ro10 M+T1 For20 Ro100 

Ro10 Treatment

Figure 36: Effect of PPAR-γ manipulation on free β-hCG release in BeWo cells. β-hCG release was measured at 48 hours post-treatment. β-hCG levels were normalized to total protein content in the media. Each drug treatment was compared to its respective vehicle control (set as 1, red dashed line). Values are represented as mean ± SEM *p≤0.05; #p≤0.05 vs vehicle control (n=4-7). Ro, rosiglitazone; T, T0070907; For, forskolin.

65

3.6.4. Visualization of BeWo cell fusion using fluorescent immunohistochemistry

To visualize cell fusion in BeWo cells, we used E-cadherin as the cell surface marker (Figure 37). Loss of membrane integrity, seen as pigmented E-cadherin staining, and a rise in aggregation of nuclei was observed following forskolin treatment (positive control) at 48 hours of treatment (Figure 37B). Compared to vehicle (Figure 37A), rosiglitazone treatment did not appear to induce cell fusion (Figure 37C). As expected, T0070907 did not increase cell fusion either (Figure 37D); however, it appeared that BeWo cell numbers were altered under this treatment (seen as a high incidence of condensed chromatin in diving cells).

Figure 37: Effect of PPAR-γ modulation on BeWo cell fusion. Cell fusion was assessed at 48 hours post-treatment. Cells were treated with forskolin (B) as a positive cell-fusion control, rosiglitazone (C), and T0070907 (D) and compared to vehicle control (A). E-cadherin (red) was used as a cell surface marker; DAPI (blue) was used as the nuclear stain (200X magnification).

66

3.6.5. Effect of PPAR-γ activity on BeWo cell proliferation

The role of PPAR-γ on BeWo cell number was assessed at 48 hours using the CellTiter-Fluo Cell Viability Assay (Figure 38). Forskolin, because of its effect on cell differentiation, was expected to decrease cell proliferation [57]. We found that following 48 hours of treatment, forskolin decreased BeWo cell numbers by 16±1.4% (p≤0.05, n=4). Confirming our initial observations from fluorescent immunohistochemistry studies (see section 3.6.4 above), treatment of BeWo cells with T0070907 significantly induced cell proliferation by 39±12.8% when compared to vehicle control (p≤0.05). Rosiglitazone did not have an effect on BeWo cell number.

2.0

1.5 *

1.0 *

0.5

(normalized to vehicle control) Relative cell proliferation cell Relative 0.0

M M M M M      T1 Vehicle Ro10 M+T1 For20 Ro100 

Ro10 Treatment

Figure 38: Effect of PPAR-γ activity modulation on BeWo cell proliferation. BeWo cell numbers were measured using CellTiter-Fluo Cell Viability Assay at 48 hours of treatment. Relative cell number was recorded in each treatment and compared to its respective vehicle control (set as 1, red dashed line). Values are represented as mean ± SEM; *p≤0.05 vs vehicle control (n=4). Ro, rosiglitazone; T, T0070907; For, forskolin.

67

Chapter 4: DISCUSSION

68

Using the Gcm-1 heterozygous mouse model and the GCM-1-silenced BeWo cell model, we showed that HO-1 expression is not GCM-1-dependent. Furthermore, we showed that there is no feedback mechanism between GCM-1 and PPAR-γ. Additionally, we confirmed that PPAR-γ positively regulates HO-1 expression in the BeWo cell line, as assessed on mRNA and protein (cellular and conditioned media content) levels. When studying the effects of PPAR-γ activity- modulating drugs on key events of trophoblast physiology (proliferation and differentiation), we found that inducing PPAR-γ activity in BeWo cells with rosiglitazone did not alter cell proliferation and did not have a profound effect on GCM-1 expression and cell differentiation as assessed by free β-hCG release. Conversely, blocking PPAR-γ activity with T0070907 significantly upregulated cell proliferation and led to lower GCM-1 levels and decreased β-hCG release. Collectively, we showed that modulating PPAR-γ activity with an antagonist had a stronger effect on the cell proliferation/differentiation balance. These findings have important implications for further studies on the role of PPAR-γ in healthy pregnancy and placental insufficiency syndromes.

Transcription factor GCM-1 does not regulate HO-1 expression

Our initial work aimed to elucidate the relationship between the transcription factor and differentiation marker GCM-1 and HO-1, an enzyme responsible for heme breakdown within the cell. We undertook this task using two different models. First, we utilized an animal model where heterozygous Gcm-1 mouse placentas were subjected to gene analysis. Second, we studied gene expression following GCM-1 silencing in vitro using the BeWo cell line.

Gcm-1 heterozygous placentas were analyzed at two developmental stages: E13.5 (corresponding to mid gestation) and E17.5 (corresponding to late gestation). Maximum placental Gcm-1 expression in the mouse is around E14-15 [164, 165], thus, we expected that if a regulatory mechanism between GCM-1 and HO-1 was present, the differences would be most striking at this stage. Contrary to our initial hypothesis, no differences in Hmox-1 mRNA expression levels were seen at E13.5 or E17.5 in Gcm-1 heterozygous placentas when compared to wildtype. Furthermore, when studying the expression of Hmox-2, another isoform of the heme oxygenase family, we did not detect differences in gene expression between the two groups. Hmox-2 has been previously shown to be higher in placentas with Hmox-1 downregulation, possibly as part of a compensatory mechanism [144]. The fact that its expression is not altered in heterozygous placentas offers additional evidence for lack of Gcm-1-controlled Hmox-1 regulation. Lastly, we did not see differences in Ppar-γ mRNA levels, indicating that a decrease in Gcm-1 expression does not affect

69 the expression of its upstream regulator, Ppar-γ. Collectively, our studies show that, in mice, Gcm-1 does not regulate the expression of Ppar-γ, suggesting the lack of a potential regulatory feedback mechanism between these two transcription factors in the mouse labyrinth.

Acknowledging the possibility that there is a potential differential gene regulation in mice when compared to that of humans, we assessed the response of HO-1 to GCM-1 silencing in the human choriocarcinoma cell line, BeWo. A significant downregulation of GCM-1 was accomplished using a combination of siRNA sequences targeted to the human GCM-1 gene. GCM-1 levels were downregulated by 62% after 48 hours of silencing; however, we found no differences in expression levels of our genes of interest, namely, HO-1 and PPAR-γ. Therefore, as we had seen previously in the mouse model, GCM-1 downregulation did not affect expression levels of HO-1 implying the lack of transcriptional control of HO-1 by GCM-1. Furthermore, since levels of PPAR-γ remained unchanged, we concluded that GCM-1 also does not control PPAR-γ. Collectively, our animal and the BeWo cell models indicate that HO-1 and PPAR-γ are not under transcriptional control of GCM-1.

We acknowledge that based on our Gcm-1 heterozygous mouse model and GCM-1 silencing experiments in BeWo cells, we cannot eliminate the possibility of GCM-1 controlling HO-1 expression. To definitively conclude that this connection is absent, subsequent studies, such as chromatin immunoprecipitation (ChIP) or electrophoretic mobility shift assay (EMSA), are necessary to show that GCM-1 does not bind to promoters of genes such as HO-1 and PPAR-γ. However, our experiments do strongly support the lack of transcriptional control of HO-1 by GCM-1, as well as the lack of any feedback mechanisms between GCM-1 and PPAR-γ.

Our subsequent studies focused on pharmacologic manipulation of PPAR-γ and the effect it had on key features of the trophoblast turnover. We used the BeWo cell model to study how PPAR-γ activity induction or inhibition influences gene and protein expression, protein localization, cell proliferation and differentiation.

70

The PPAR-γ/HO-1 molecular axis is conserved and functional in the BeWo cell line

The relationship between HO-1 and PPAR-γ has been studied in several systems. Since the initial proof that PPAR-γ directly binds to the HMOX-1 promoter in HUVECs and vascular smooth muscle cells, studies have looked at how these molecules interact in other systems [114]. Considering that the connection between HO-1 and PPAR-γ was shown to be important in cell migration, we decided to study how this molecular axis behaves in the BeWo cell line [150]. Our experiments indicated that stimulation of PPAR-γ with rosiglitazone significantly increased HO-1 transcript expression, intra-cellular protein expression and protein release into the culture medium. HO-1 mRNA induction was a rapid and persistent event where rosiglitazone showed an effect as early as 3 hours post-treatment. Cellular protein expression was significantly upregulated at 24 hours of treatment, while HO-1 protein content in the BeWo cell-conditioned media was elevated at 48 hours of treatment. The time frames for these events support the efficacy of our findings, such that changes at the mRNA level were followed by alternations of the intracellular protein expression, which was further followed by changes in protein release. Conversely, opposite effects on HO-1 expression were seen following T0070907 treatment. In experiments where both drugs were combined, pre-treatment with T0070907 skewed gene and protein expression profiles to resemble those consistent with antagonist treatment alone, implicating that T0070907 is a more potent drug, even when the dose used is ten times lower. Although both synthetic ligands (rosiglitazone and T0070907) are highly specific for PPAR-γ (and especially the gamma isoform), the observed effect can be attributed to the ability of T0070907 to bind irreversibly within the ligand-binding pocket of the receptor. Surprisingly, when treating cells with forskolin, there was a decrease in HO-1 mRNA and protein expression, despite the fact that forskolin is known to mildly increase the activity of PPAR-γ [166], and has also been shown to increase HO-1 expression in smooth muscle cells [167]. These changes in gene expression studies were performed at 3 and 6 hours of treatment; the effect diminished by 24 hours of treatment. This explains the lack of HO-1 mRNA response in our GCM-1 silencing experiments on BeWo cells since they were assessed after 48 hours of treatment with forskolin. Furthermore, this finding has implications for our study of whether and how GCM-1 controls HO-1 expression, considering that forskolin is the primary available inducer of GCM-1 expression. Thus, this observation must be taken into account when stimulating cells with this drug, as it might override a possible connection between GCM-1 and HO-1.

71

PPAR-γ participates in a negative autoregulatory feedback mechanism

When studying the effect of PPAR-γ activity modulation on its expression, we observed that this transcription factor is able to participate in a negative auto-regulation feedback mechanism. We found that a rise in PPAR-γ activity (with treatment of an agonist) was accompanied by a decrease in its expression, while the opposite was true with the treatment of cells with the antagonist. Although forskolin slightly and transiently upregulated PPAR-γ mRNA expression, it led to a modest downregulation of the receptor protein expression. Since PPAR-γ is an important transcription factor with a wide range of actions, several mechanisms have evolved to modulate its activity; one such mechanism is phosphorylation (reviewed in [85]). We studied PPAR-γ phosphorylation at 24 and 48 hours of treatment. Interestingly, when comparing phosphorylated PPAR-γ (phos-PPAR-γ) levels to total PPAR-γ protein expression, we found that levels of phos-PPAR-γ are higher in rosiglitazone-treated cells, while the opposite is seen when treating BeWo cells with the antagonist. These findings allow us to speculate that this is another part of the feedback mechanism in which PPAR-γ participates. Whether this happens directly or indirectly is yet to be elucidated, as is the identity of factors which phosphorylate this protein. Until recently, most studies indicated that PPAR-γ phosphorylation results in protein inactivation; however, there is a possibility that the opposite occurs, depending on which kinases phosphorylate the receptor (ex. Cdk7 and 9 phosphorylation of PPAR-γ was shown to increase receptor activity). Thus, although we cannot definitively state that phosphorylation under our treatments leads to receptor inactivation, if it does, we can propose it to be another part of the feedback mechanism aimed at stabilizing the activity of PPAR-γ when the system is disrupted. Several points should be addressed regarding our studies of PPAR-γ phosphorylation. First, we measured phosphorylation of the receptor at 24 and 48 hours. Phosphorylation is a rapid event, thus, measuring the levels of phosphorylated receptor at earlier time points (such as 1 and 3 hours) in whole tissue lysates may produce different results. Second, the antibody used to detect phosphorylated PPAR-γ protein is polyclonal; this might impair its ability to detect only the phosphorylated isoform of the receptor. Therefore, we cannot conclusively state whether the phosphorylated PPAR-γ isoform corresponds only to the top band (at 57 kDa) or to both bands seen on the blot (corresponding to PPAR-γ1 and PPAR-γ2).

72

Our findings of PPAR-γ expression indicate that certain molecular mechanisms in BeWo cells are in place (and possibly in other systems) to ensure fine-tuning of PPAR-γ activity which has important implications in cell lipid metabolism, differentiation and immune response. The induction of activity is compensated for by a decrease in transcription factor expression and a rise in receptor phosphorylation (and possibly inactivation), while a decrease in its activity is complimented with a rise in receptor levels and a decrease in protein phosphorylation. To complete our study of PPAR-γ expression in BeWo cells, we analyzed its localization using fluorescent IHC and cellular fractionation technique. Using fluorescent IHC, we showed that PPAR-γ was predominantly localized in the nucleus with low levels found in the cytoplasm. This finding was expected as PPAR-γ is a transcription factor and, thus, should localize to the nucleus; since it is a protein, it should be present in the cytoplasm in trace amounts where it has to be translated. After performing a series of cellular fractionation analyses, we observed that changes in PPAR-γ expression were mostly restricted to the nucleus; while cytoplasmic expression fluctuations were much less prominent (although it should be acknowledged that n of 3 might contribute to this result). Consistent with our findings from whole cell lysate experiments, PPAR-γ expression decreased following rosiglitazone, and increased after T0070907, treatment in the nuclear compartment; these effects were less striking in the cytoplasm. Interestingly, the levels of nuclear phos-PPAR-γ when compared to total nuclear PPAR-γ expression remain unchanged following different treatments. Western blots illustrate that only one isoform of PPAR-γ is detected in the cell nucleus, while both isoforms are present in the cytoplasm (as was the case in whole cell lysate analyses described above). This suggests that expression and localization of one of the isoforms is affected more than that of the other. Additionally, our data indicate that although synthesis of the PPAR-γ protein may not be significantly affected (as seen by lack of prominent changes at the protein synthesis site, the cytoplasm), changes in PPAR-γ protein expression are mediated at the level of the nucleus. One way we can test this hypothesis is by blocking protein synthesis while treating cells with chemicals to study whether the system will respond the same way to these manipulations. It can be further speculated that changes in the nucleus, but not in the cytoplasm, are due to altered protein stability (which may be affected by ubiquitination or unfolded protein response) or shunting between the two cell compartments. Collectively, our findings suggest that there are differential PPAR-γ protein regulation mechanisms in the nucleus and the cytoplasm.

73

Baseline activity of PPAR-γ on the GCM-1 promoter is high

In the next set of experiments, we aimed to study how molecular manipulation of PPAR-γ affects the cascade of events associated with BeWo cellular dynamics, representing key features of the trophoblast turnover, such as proliferation, differentiation and fusion. First, we studied how PPAR-γ activation influences the expression of GCM-1. We found that rosiglitazone-mediated increase in GCM-1 expression is transient and modest, where it is only seen at 3 hours of treatment. In contrast, T0070907-mediated inhibition of PPAR-γ stably blocked the expression of GCM-1, which persisted at 24 hours of treatment. Forskolin, a drug used to induce GCM-1 expression, was used as a positive control to confirm that induction of GCM-1 was possible. These findings indicate that induction of GCM-1 in BeWo cells is possible; however, it is not primarily mediated via the transcription factor PPAR-γ. Since it has been shown that PPAR-γ regulates GCM-1 expression in mouse trophoblast stem cells [101], we found it surprising that our modulation of this axis could not be easily accomplished in the BeWo cell model. To offer a possible explanation for this finding, we performed a series of luciferase reporter assays where we studied PPAR-γ interaction with the GCM-1 promoter in vitro, in the BeWo cell line. Upon treatment of cells with vehicle control, we observed a high baseline activity of PPAR-γ; this activity was not greatly different from that observed following treatment of cells with rosiglitazone. Furthermore, when treating cells with PPAR-γ antagonist, luciferase expression was greatly downregulated. These finding suggest that baseline interaction of PPAR-γ with the GCM-1 promoter in the BeWo cell line is strong, and rosiglitazone is not able to induce it even further. The ability of inducing PPAR-γ activity is possible as seen by the rise of HO-1 transcription levels; however, data suggest that the PPAR-γ/GCM-1 axis does not have the potential to be stimulated much further. Since it is easier to inhibit the activity of a highly active transcription factor, we saw a robust decrease in luciferase gene expression with T0070907 treatment. These reporter assay findings are useful in explaining the next set of experiments on cell proliferation and differentiation.

Blocking PPAR-γ activity influences key features of the trophoblast physiology

To study the features of trophoblast turnover, we examined how manipulating PPAR-γ activity affects cell proliferation and differentiation. In our experiments, we showed that cell proliferation was unaffected by rosiglitazone stimulation at 48 hours; however, relative BeWo cell numbers were significantly increased following T0070907 treatment. This further reinforced our

74 initial hypothesis that T0070907 is able to influence the PPAR-γ/GCM-1 axis more than rosiglitazone. Forskolin, which is known to induce differentiation of BeWo cells, was used as a positive control; its treatment resulted in a decrease in BeWo cell numbers. To study differentiation, we used a well-characterized assay of free β-hCG release, where a rise in free β-hCG secretion indicates cell differentiation in our model. Upon treatment of cells with rosiglitazone, we did not see a significant increase in β-hCG release; this was also true after using a higher dose of the drug. This finding was expected based on the transient and weak induction of GCM-1 expression and our findings of weak luciferase reporter signals in BeWo cells following rosiglitazone treatment. As a control, we showed that induction of differentiation was possible in our system upon treatment with forskolin, which led to significantly higher levels of β-hCG in the culture medium. As was predicted, PPAR-γ antagonist had a more prominent effect than the agonist and significantly decreased β-hCG secretion. Our studies of BeWo cell differentiation were confirmed with fusion studies (fluorescent IHC) using e-cadherin as a cell surface marker. Cell fusion can be visualized as aggregation of nuclei and disintegration of the cell surface in forskolin- treated cells. These characteristics were absent in rosiglitazone- and T0070907-treated cells. Collectively, our studies indicate that baseline PPAR-γ activity in our BeWo cell model is relatively high; inducing it even further is challenging, but blocking it with an antagonist has prominent downstream effects on events such as GCM-1 expression, cell proliferation and differentiation.

Implications for treatment and/or management of sPE

PPAR-γ is a key regulator of many physiological processes, including trophoblast and adipocyte differentiation, lipid metabolism, inflammation and others. Our studies have shown that its expression and activity are inversely related, such that inducing its activity with an agonist leads to a lower expression, while blocking its activity results in higher receptor levels. Importantly, we observed a trend towards increased PPAR-γ mRNA and protein levels in sPE placentas; this might reflect that the activity of this transcription factor and nuclear receptor is decreased in pathological tissue. This also implies that the trophoblast differentiation axis, as controlled by PPAR-γ, is disrupted in disease. Therefore, pharmacological upregulation of PPAR-γ activity has the potential to restore placental physiology (via an upregulation of GCM-1) as well as ameliorate maternal hypertension (via the induction of HO-1, and, subsequently, CO). Although manipulation of PPAR-γ activity with rosiglitazone appears to be a promising tool for the management and/or prevention of sPE, the possibility of treating women at risk of sPE with this drug has to be approached with caution, considering that rosiglitazone exerts its effects on

75 multiple tissues in the human body. When implementing rosiglitazone into treatment plans for women identified at risk of disease development, the dosage as well as start and duration of treatment should be carefully determined. Lastly, and most importantly, since this thiazolidinedione can cross the human placenta, the potential harmful effects of this drug on the developing fetus deserve considerable attention and should undergo extensive research to warrant its safe use in human pregnancy.

4.1 Conclusion

Our studies support the role of PPAR-γ in the regulation of several factors involved in trophoblast layer function. We have shown that this transcription factor is able to regulate the expression of two important molecules, HO-1 and GCM-1. Activation of PPAR-γ with an agonist increased the levels of HO-1 on transcriptional and translational levels, without having a strong effect on cellular GCM-1 levels, cell proliferation, differentiation and fusion. In comparison, blocking the activity of PPAR-γ using T0070907 resulted in a prominent block of HO-1 and GCM-1 expression, decreased cell differentiation (as seen by a decrease in β-hCG release), and increased cell proliferation. Our findings have important implications for further study of PPAR-γ in the placenta and its role in sPE pathogenesis. Since both HO-1 and GCM-1 are downregulated in sPE placentas [58, 122], our ability to induce their expression via an activation of an upstream factor, PPAR-γ, is an invaluable tool for the development of possible prophylactic and/or treatment options for women at risk of developing and suffering from sPE.

76

Chapter 5: FUTURE DIRECTIONS

77

The current study aimed to (1) establish the relationship between GCM-1 and HO-1, and (2) assess the effects of PPAR-γ activity modulation on trophoblast layer-related events. Our initial findings open several avenues which can be explored in further research. To address our first aim and conclusively state that GCM-1 does not directly regulate expression of HO-1 in BeWo cells, we propose to perform EMSA or ChIP assays to prove that there is no interaction between the GCM-1 protein and the human HO-1 promoter. The BeWo cell line is a useful model to study the molecular axes (PPAR-γ/HO-1 and PPAR- γ/GCM-1) which are the focus of our study. Using this model, we will address several outstanding questions from our second aim. We propose to study whether the effects of PPAR-γ activity modulation on cell proliferation and differentiation are HO-1-dependent as well as determine the role of this enzyme in trophoblast physiology. This can be approached using two different models. Firstly, HO-1 gene expression can be downregulated using commercially available siRNA sequences. Alternatively, HO-1 activity can be decreased using several antagonists, such as ZnPP-9 (zinc protoporphyrin-9) or SnPP (tin protoporphyrin). These experiments will be done concurrently with rosiglitazone treatment to assess the contribution of the enzyme’s activity to any of HO-1- (in)dependent effects seen in vitro. Furthermore, to bypass PPAR-γ in the induction of HO-1 expression, drugs which directly stimulate HO-1 mRNA expression such as hemin or statins, as well as those that stimulate HO-1 activity such as CoPP (cobalt protoporphyrin), can be used to study the role of HO-1 in trophoblast physiology. Alternatively, to prove that these effects are specific to PPAR-γ-associated activation, PPAR- γ levels can be downregulated using siRNA or lentiviral vectors targeting human PPAR-γ expression. Using either of these techniques will help us define the direct contribution of PPAR-γ to events related to the trophoblast physiology when cells are treated with rosiglitazone or T0070907. Although both drugs are specific to PPAR-γ, these experiments will also clarify whether the chemicals have any secondary effects. A potential limitation of this technique must be acknowledged: it appears that there are mechanisms in place which constantly attempt to compensate for fluctuations in receptor activity. To account for this and monitor whether PPAR-γ activity is decreased following gene downregulation, PPAR-γ trans-activation assay will be performed. Another avenue worth exploring is the role of PPAR-γ co-factor, RXR-α, in cell proliferation and differentiation. This transcription factor can be activated with specific ligands, such as 9-cis- retinoic acid, and a combination of RXR-α and PPAR-γ agonists might have additive positive effects on trophoblast layer dynamics and integrity.

78

Although the BeWo cell model is a useful model to study trophoblast function in vitro, it has limitations. The largest drawback is the lack of sFLT-1 expression in BeWo cells. As mentioned previously, sFLT-1, an anti-angiogenic protein, is presently believed to be one of the key players in maternal endothelial cell dysfunction in sPE. Because these cells do not express sFLT-1, downstream effects of rosiglitazone on the placenta can only be speculated. Therefore, to study how rosiglitazone affects the physiology of the organ itself, the established first trimester human floating villous explant model will be used. Analogous experiments to those performed on BeWo cells in our study, as well as experiments outlined above in this section, will be done using this explant model. Firstly, we will study how GCM-1 and HO-1 mRNA and protein levels change in response to rosiglitazone and T0070907 treatments. Secondly, we will assess the balance between cytotrophoblast cell proliferation and differentiation using Ki67 and/or BrdU markers and β-hCG release, respectively. Additionally, we will assess explant morphology in semi-thin sections and by electron microscopy. Last, and most importantly, we will study levels of sFLT-1 in the conditioned media following treatment with PPAR-γ agonists and antagonists. A decrease in the levels of sFLT-1 release following rosiglitazone treatment would validate the idea of using PPAR-γ as a potential treatment option for women suffering from sPE. The ability to lower circulating serum sFLT-1 levels in the mother will help treat the systemic endothelial cell dysfunction and ameliorate disease symptoms such as hypertension and proteinuria. To conclude our studies on the role of PPAR-γ/HO-1 axis in placental physiology, we will assess the activity of this enzyme in first trimester placental explants. HO-1 is an enzyme that breaks down heme into three components, one of which is CO; assays are available to study the concentration of CO in the gaseous phase above a homogenized sample. Furthermore, a new technique is available to detect bilirubin levels using the UnaG protein which possesses a high affinity for, and fluoresces upon the binding of, this metabolite. An interesting study will be to assess whether drugs modulating PPAR-γ activity also influence HO-1 activity, as measured by CO and/or bilirubin levels. A rise in their concentrations following rosiglitazone treatment would have direct implications for the efficacy of the use of this drug for management of sPE. This finding would suggest that rosiglitazone may help treat maternal hypertension and decrease systemic vascular resistance by increasing the expression and activity of an enzyme responsible for secretion of a potent vasodilator (CO) and an anti-oxidant (bilirubin). The response of early- and late-onset sPE placentas to drugs of interest should be assessed. Whether sPE placentas respond similarly and positively to treatment with rosiglitazone will determine how useful this drug can be to treat the diseased organ. It must be mentioned that

79 whether or not rosiglitazone will exert its effects on the pathological tissue will depend on the severity of placental pathology. It is possible that tissue response to rosiglitazone may not be strongly pronounced, considering that at the time of delivery, sPE placentas are beyond the point of when treatment can reverse or treat the pathology. If this will be the case, the study of the response of first trimester explant tissue to rosiglitazone becomes even more urgent in order to deliver rosiglitazone as a prophylactic treatment to women identified at risk of developing sPE. Identifying the responses of healthy first trimester and sPE explants to drugs regulating PPAR-γ activity will offer insights into the actions of this transcription factor during healthy pregnancy progression and in placental insufficiency syndromes. In the event of a finding that rosiglitazone treatment decreases secretion of sFLT-1 and induces the activity of HO-1 (seen as an upregulation of CO levels), this drug will gain further support as a treatment option for women at risk of or suffering from sPE.

80

Chapter 6: REFERENCES

81

1. Metcalfe, J. and K. Ueland, Maternal cardiovascular adjustments to pregnancy. Prog Cardiovasc Dis, 1974. 16(4): p. 363-74. 2. Abbas, A.E., S.J. Lester, and H. Connolly, Pregnancy and the cardiovascular system. Int J Cardiol, 2005. 98(2): p. 179-89. 3. Roberts CL, F.J., Algert CS, et al, Population-based trends in pregnancy hypertension and pre- eclampsia: an international comparative study. BMJ Open 2011. 4. Hutcheon, J.A., S. Lisonkova, and K.S. Joseph, Epidemiology of pre-eclampsia and the other hypertensive disorders of pregnancy. Best Pract Res Clin Obstet Gynaecol, 2011. 25(4): p. 391-403. 5. Practice, A.C.o.O., ACOG practice bulletin. Diagnosis and management of preeclampsia and eclampsia. Number 33, January 2002. American College of Obstetricians and Gynecologists. Int J Gynaecol Obstet, 2002. 77(1): p. 67-75. 6. Parikh, S.M. and S.A. Karumanchi, Putting pressure on pre-eclampsia. Nat Med, 2008. 14(8): p. 810-2. 7. Franco, C., et al., Placental infarction and thrombophilia. Obstet Gynecol, 2011. 117(4): p. 929-34. 8. Witlin, A.G. and B.M. Sibai, Magnesium sulfate therapy in preeclampsia and eclampsia. Obstet Gynecol, 1998. 92(5): p. 883-9. 9. Hofmeyr, G.J., A.N. Atallah, and L. Duley, Calcium supplementation during pregnancy for preventing hypertensive disorders and related problems. Cochrane Database Syst Rev, 2006(3): p. CD001059. 10. Zhou, S.J., et al., Fish-oil supplementation in pregnancy does not reduce the risk of gestational diabetes or preeclampsia. Am J Clin Nutr, 2012. 95(6): p. 1378-84. 11. Basaran, A., M. Basaran, and B. Topatan, Combined vitamin C and E supplementation for the prevention of preeclampsia: a systematic review and meta-analysis. Obstet Gynecol Surv, 2010. 65(10): p. 653-67. 12. Roberge, S., et al., Early administration of low-dose aspirin for the prevention of severe and mild preeclampsia: a systematic review and meta-analysis. Am J Perinatol, 2012. 29(7): p. 551-6. 13. Rey, E., et al., Dalteparin for the prevention of recurrence of placental-mediated complications of pregnancy in women without thrombophilia: a pilot randomized controlled trial. J Thromb Haemost, 2009. 7(1): p. 58-64. 14. Kingdom, J.C., et al., Unfractionated heparin for second trimester placental insufficiency: a pilot randomized trial. J Thromb Haemost, 2011. 9(8): p. 1483-92. 15. Drewlo, S., et al., Heparin promotes soluble VEGF receptor expression in human placental villi to impair endothelial VEGF signaling. J Thromb Haemost, 2011. 9(12): p. 2486-97. 16. Sobel, M.L., J. Kingdom, and S. Drewlo, Angiogenic response of placental villi to heparin. Obstet Gynecol, 2011. 117(6): p. 1375-83. 17. Sibley, C.P., et al., Mechanisms of maternofetal exchange across the human placenta. Biochem Soc Trans, 1998. 26(2): p. 86-91. 18. Gude, N.M., et al., Growth and function of the normal human placenta. Thromb Res, 2004. 114(5-6): p. 397-407. 19. Burton, G.J., J. Hempstock, and E. Jauniaux, Nutrition of the human fetus during the first trimester--a review. Placenta, 2001. 22 Suppl A: p. S70-7. 20. Benirschke K, K.P., Baergen R., Pathology of the human placenta. 2005, New York, NY: Springer. 21. Everett, T.R. and C.C. Lees, Beyond the placental bed: placental and systemic determinants of the uterine artery Doppler waveform. Placenta, 2012. 33(11): p. 893-901. 22. Pijnenborg, R., et al., Uteroplacental arterial changes related to interstitial trophoblast migration in early human pregnancy. Placenta, 1983. 4(4): p. 397-413.

82

23. Pijnenborg, R., et al., The pattern of interstitial trophoblastic invasion of the myometrium in early human pregnancy. Placenta, 1981. 2(4): p. 303-16. 24. Burton, G.J., E. Jauniaux, and A.L. Watson, Maternal arterial connections to the placental intervillous space during the first trimester of human pregnancy: the Boyd collection revisited. Am J Obstet Gynecol, 1999. 181(3): p. 718-24. 25. Bischof, P. and A. Campana, A model for implantation of the human blastocyst and early placentation. Hum Reprod Update, 1996. 2(3): p. 262-70. 26. Drewlo, S., et al., Glial cell missing-1 mediates over-expression of tissue inhibitor of metalloproteinase-4 in severe pre-eclamptic placental villi. Hum Reprod, 2011. 26(5): p. 1025-34. 27. Moffett-King, A., Natural killer cells and pregnancy. Nat Rev Immunol, 2002. 2(9): p. 656-63. 28. Morrish, D.W., J. Dakour, and H. Li, Functional regulation of human trophoblast differentiation. J Reprod Immunol, 1998. 39(1-2): p. 179-95. 29. Mayhew, T.M., Villous trophoblast of human placenta: a coherent view of its turnover, repair and contributions to villous development and maturation. Histol Histopathol, 2001. 16(4): p. 1213-24. 30. Huppertz, B., et al., Villous cytotrophoblast regulation of the syncytial apoptotic cascade in the human placenta. Histochem Cell Biol, 1998. 110(5): p. 495-508. 31. Lanoix, D., et al., Melatonin: the smart killer: the human trophoblast as a model. Mol Cell Endocrinol, 2012. 348(1): p. 1-11. 32. Kaufmann, P., S. Black, and B. Huppertz, Endovascular trophoblast invasion: implications for the pathogenesis of intrauterine growth retardation and preeclampsia. Biol Reprod, 2003. 69(1): p. 1-7. 33. Pijnenborg, R., et al., Placental bed spiral arteries in the hypertensive disorders of pregnancy. Br J Obstet Gynaecol, 1991. 98(7): p. 648-55. 34. Matsubara, K., et al., Role of nitric oxide and reactive oxygen species in the pathogenesis of preeclampsia. J Obstet Gynaecol Res, 2010. 36(2): p. 239-47. 35. Myatt, L., et al., Nitrotyrosine residues in placenta. Evidence of peroxynitrite formation and action. Hypertension, 1996. 28(3): p. 488-93. 36. Hung, T.H. and G.J. Burton, Hypoxia and reoxygenation: a possible mechanism for placental oxidative stress in preeclampsia. Taiwan J Obstet Gynecol, 2006. 45(3): p. 189-200. 37. Jonsson, Y., et al., Cytokine mapping of sera from women with preeclampsia and normal pregnancies. J Reprod Immunol, 2006. 70(1-2): p. 83-91. 38. Patel, J., et al., Regulation of hypoxia inducible factors (HIF) in hypoxia and normoxia during placental development. Placenta, 2010. 31(11): p. 951-7. 39. Araki-Taguchi, M., et al., Angiotensin II mimics the hypoxic effect on regulating trophoblast proliferation and differentiation in human placental explant cultures. Life Sci, 2008. 82(1-2): p. 59-67. 40. Nevo, O., et al., Increased expression of sFlt-1 in in vivo and in vitro models of human placental hypoxia is mediated by HIF-1. Am J Physiol Regul Integr Comp Physiol, 2006. 291(4): p. R1085-93. 41. Yinon, Y., et al., Severe intrauterine growth restriction pregnancies have increased placental endoglin levels: hypoxic regulation via transforming growth factor-beta 3. Am J Pathol, 2008. 172(1): p. 77-85. 42. Ahmad, S. and A. Ahmed, Elevated placental soluble vascular endothelial growth factor receptor-1 inhibits angiogenesis in preeclampsia. Circ Res, 2004. 95(9): p. 884-91. 43. Cudmore, M., et al., Negative regulation of soluble Flt-1 and soluble endoglin release by heme oxygenase-1. Circulation, 2007. 115(13): p. 1789-1797.

83

44. Rajakumar, A., et al., Transcriptionally active syncytial aggregates in the maternal circulation may contribute to circulating soluble fms-like tyrosine kinase 1 in preeclampsia. Hypertension, 2012. 59(2): p. 256-64. 45. Ellery, P.M., et al., Evidence for transcriptional activity in the syncytiotrophoblast of the human placenta. Placenta, 2009. 30(4): p. 329-34. 46. Fogarty, N.M., et al., A quantitative analysis of transcriptionally active syncytiotrophoblast nuclei across human gestation. J Anat, 2011. 219(5): p. 601-10. 47. Levine, R.J., et al., Circulating angiogenic factors and the risk of preeclampsia. N Engl J Med, 2004. 350(7): p. 672-83. 48. Ferrara, N., H.P. Gerber, and J. LeCouter, The biology of VEGF and its receptors. Nat Med, 2003. 9(6): p. 669-76. 49. Hosoya, T., et al., glial cells missing: a binary switch between neuronal and glial determination in Drosophila. Cell, 1995. 82(6): p. 1025-36. 50. Schreiber, J., E. Sock, and M. Wegner, The regulator of early gliogenesis glial cells missing is a transcription factor with a novel type of DNA-binding domain. Proceedings of the National Academy of Sciences of the United States of America, 1997. 94(9): p. 4739-4744. 51. Schreiber, J., J. Enderich, and M. Wegner, Structural requirements for DNA binding of GCM proteins. Nucleic Acids Research, 1998. 26(10): p. 2337-2343. 52. Nait-Oumesmar, B., A.B. Copperman, and R.A. Lazzarini, Placental expression and chromosomal localization of the human Gcm 1 gene. J Histochem Cytochem, 2000. 48(7): p. 915-22. 53. Anson-Cartwright, L., et al., The glial cells missing-1 protein is essential for branching morphogenesis in the chorioallantoic placenta. Nat Genet, 2000. 25(3): p. 311-4. 54. Stecca, B., et al., Gcm1 expression defines three stages of chorio-allantoic interaction during placental development. Mech Dev, 2002. 115(1-2): p. 27-34. 55. Baczyk, D., et al., Complex patterns of GCM1 mRNA and protein in villous and extravillous trophoblast cells of the human placenta. Placenta, 2004. 25(6): p. 553-559. 56. Chang, C.-W., et al., Stimulation of GCMa transcriptional activity by cyclic AMP/protein kinase A signaling is attributed to CBP-mediated acetylation of GCMa. Molecular and cellular biology, 2005. 25(19): p. 8401-8414. 57. Baczyk, D., et al., Glial cell missing-1 transcription factor is required for the differentiation of the human trophoblast. Cell death and differentiation, 2009. 16(5): p. 719-727. 58. Chen, C.P., et al., Decreased placental GCM1 (glial cells missing) gene expression in pre- eclampsia. Placenta, 2004. 25(5): p. 413-421. 59. Bainbridge, S.A., et al., Effects of reduced Gcm1 expression on trophoblast morphology, fetoplacental vascularity, and pregnancy outcomes in mice. Hypertension, 2012. 59(3): p. 732-9. 60. Yu, C., et al., GCMa regulates the syncytin-mediated trophoblastic fusion. J Biol Chem, 2002. 277(51): p. 50062-8. 61. Lin, C., M. Lin, and H. Chen, Biochemical characterization of the human placental transcription factor GCMa/1. Biochem Cell Biol, 2005. 83(2): p. 188-95. 62. Mukherjee, R., et al., Identification, characterization, and tissue distribution of human peroxisome proliferator-activated receptor (PPAR) isoforms PPARgamma2 versus PPARgamma1 and activation with retinoid X receptor agonists and antagonists. J Biol Chem, 1997. 272(12): p. 8071-6. 63. Elbrecht, A., et al., Molecular cloning, expression and characterization of human peroxisome proliferator activated receptors gamma 1 and gamma 2. Biochem Biophys Res Commun, 1996. 224(2): p. 431-7. 64. Barbier, O., et al., Pleiotropic actions of peroxisome proliferator-activated receptors in lipid metabolism and atherosclerosis. Arterioscler Thromb Vasc Biol, 2002. 22(5): p. 717-26.

84

65. Abbott, B.D., Review of the expression of peroxisome proliferator-activated receptors alpha (PPAR alpha), beta (PPAR beta), and gamma (PPAR gamma) in rodent and human development. Reprod Toxicol, 2009. 27(3-4): p. 246-57. 66. Lehmann, J.M., et al., An antidiabetic thiazolidinedione is a high affinity ligand for peroxisome proliferator-activated receptor gamma (PPAR gamma). J Biol Chem, 1995. 270(22): p. 12953-6. 67. Lehrke, M. and M.A. Lazar, The many faces of PPARgamma. Cell, 2005. 123(6): p. 993-9. 68. Michalik, L. and W. Wahli, Peroxisome proliferator-activated receptors: three isotypes for a multitude of functions. Curr Opin Biotechnol, 1999. 10(6): p. 564-70. 69. Marceau, G., Blanchon, L., Lobaccaro, J.-M., Sapin V. , Role of Nuclear Receptors Peroxisome Proliferator-Activated Receptors (PPARs) and Liver X Receptors (LXRs) in the Human Placental Pathophysiology, in Recent Advances in Research on the Human Placenta, J. Zheng, Editor. 2012. 70. Chandra, V., et al., Structure of the intact PPAR-gamma-RXR- nuclear receptor complex on DNA. Nature, 2008. 456(7220): p. 350-6. 71. A, I.J., et al., Polarity and specific sequence requirements of peroxisome proliferator-activated receptor (PPAR)/retinoid X receptor heterodimer binding to DNA. A functional analysis of the malic enzyme gene PPAR response element. J Biol Chem, 1997. 272(32): p. 20108-17. 72. Zieleniak, A., M. Wojcik, and L.A. Wozniak, Structure and physiological functions of the human peroxisome proliferator-activated receptor gamma. Arch Immunol Ther Exp (Warsz), 2008. 56(5): p. 331-45. 73. Gelman, L., et al., Kinase signaling cascades that modulate peroxisome proliferator-activated receptors. Curr Opin Cell Biol, 2005. 17(2): p. 216-22. 74. Wang, H., et al., Stage-specific integration of maternal and embryonic peroxisome proliferator- activated receptor delta signaling is critical to pregnancy success. J Biol Chem, 2007. 282(52): p. 37770-82. 75. Barak, Y., et al., PPAR gamma is required for placental, cardiac, and adipose tissue development. Mol Cell, 1999. 4(4): p. 585-95. 76. Willson, T.M. and W. Wahli, Peroxisome proliferator-activated receptor agonists. Curr Opin Chem Biol, 1997. 1(2): p. 235-41. 77. Issemann, I. and S. Green, Activation of a member of the steroid hormone receptor superfamily by peroxisome proliferators. Nature, 1990. 347(6294): p. 645-50. 78. Forman, B.M., et al., 15-Deoxy-delta 12, 14-prostaglandin J2 is a ligand for the adipocyte determination factor PPAR gamma. Cell, 1995. 83(5): p. 803-12. 79. Kliewer, S.A., et al., A prostaglandin J2 metabolite binds peroxisome proliferator-activated receptor gamma and promotes adipocyte differentiation. Cell, 1995. 83(5): p. 813-9. 80. Yki-Jarvinen, H., Thiazolidinediones. N Engl J Med, 2004. 351(11): p. 1106-18. 81. Sakamoto, J., et al., Activation of human peroxisome proliferator-activated receptor (PPAR) subtypes by pioglitazone. Biochem Biophys Res Commun, 2000. 278(3): p. 704-11. 82. Feige, J.N., et al., From molecular action to physiological outputs: peroxisome proliferator- activated receptors are nuclear receptors at the crossroads of key cellular functions. Prog Lipid Res, 2006. 45(2): p. 120-59. 83. Leesnitzer, L.M., et al., Functional consequences of cysteine modification in the ligand binding sites of peroxisome proliferator activated receptors by GW9662. Biochemistry, 2002. 41(21): p. 6640-50. 84. Lee, G., et al., T0070907, a selective ligand for peroxisome proliferator-activated receptor gamma, functions as an antagonist of biochemical and cellular activities. J Biol Chem, 2002. 277(22): p. 19649-57.

85

85. van Beekum, O., V. Fleskens, and E. Kalkhoven, Posttranslational modifications of PPAR- gamma: fine-tuning the metabolic master regulator. Obesity (Silver Spring), 2009. 17(2): p. 213-9. 86. Adams, M., et al., Transcriptional activation by peroxisome proliferator-activated receptor gamma is inhibited by phosphorylation at a consensus mitogen-activated protein kinase site. J Biol Chem, 1997. 272(8): p. 5128-32. 87. Camp, H.S. and S.R. Tafuri, Regulation of peroxisome proliferator-activated receptor gamma activity by mitogen-activated protein kinase. J Biol Chem, 1997. 272(16): p. 10811-6. 88. Hu, E., et al., Inhibition of adipogenesis through MAP kinase-mediated phosphorylation of PPARgamma. Science, 1996. 274(5295): p. 2100-3. 89. Shao, D., et al., Interdomain communication regulating ligand binding by PPAR-gamma. Nature, 1998. 396(6709): p. 377-80. 90. Compe, E., et al., Dysregulation of the peroxisome proliferator-activated receptor target genes by XPD mutations. Mol Cell Biol, 2005. 25(14): p. 6065-76. 91. Iankova, I., et al., Peroxisome proliferator-activated receptor gamma recruits the positive transcription elongation factor b complex to activate transcription and promote adipogenesis. Mol Endocrinol, 2006. 20(7): p. 1494-505. 92. Rodie, V.A., et al., Human placental peroxisome proliferator-activated receptor delta and gamma expression in healthy pregnancy and in preeclampsia and intrauterine growth restriction. J Soc Gynecol Investig, 2005. 12(5): p. 320-9. 93. Schaiff, W.T., et al., Peroxisome proliferator-activated receptor-gamma modulates differentiation of human trophoblast in a ligand-specific manner. J Clin Endocrinol Metab, 2000. 85(10): p. 3874-81. 94. Waite, L.L., et al., Placental peroxisome proliferator-activated receptor-gamma is up- regulated by pregnancy serum. J Clin Endocrinol Metab, 2000. 85(10): p. 3808-14. 95. Toth, B., et al., Peroxisome proliferator-activated receptor-gamma in normal human pregnancy and miscarriage. Acta Histochem, 2009. 111(4): p. 372-8. 96. Drewlo, S., K. Levytska, and J. Kingdom, Revisiting the housekeeping genes of human placental development and insufficiency syndromes. Placenta, 2012. 33(11): p. 952-4. 97. Waite, L.L., R.E. Louie, and R.N. Taylor, Circulating activators of peroxisome proliferator- activated receptors are reduced in preeclamptic pregnancy. J Clin Endocrinol Metab, 2005. 90(2): p. 620-6. 98. Joss-Moore, L.A., et al., IUGR decreases PPARgamma and SETD8 Expression in neonatal rat lung and these effects are ameliorated by maternal DHA supplementation. Early Hum Dev, 2010. 86(12): p. 785-91. 99. Joss-Moore, L.A., et al., Uteroplacental insufficiency increases visceral adiposity and visceral adipose PPARgamma2 expression in male rat offspring prior to the onset of obesity. Early Hum Dev, 2010. 86(3): p. 179-85. 100. Desai, M., et al., Programmed upregulation of adipogenic transcription factors in intrauterine growth-restricted offspring. Reprod Sci, 2008. 15(8): p. 785-96. 101. Parast, M.M., et al., PPARgamma regulates trophoblast proliferation and promotes labyrinthine trilineage differentiation. PLoS One, 2009. 4(11): p. e8055. 102. Ruebner, M., et al., Regulation of the human endogenous retroviral Syncytin-1 and cell-cell fusion by the nuclear hormone receptors PPARgamma/RXRalpha in placentogenesis. J Cell Biochem, 2012. 113(7): p. 2383-96. 103. Schaiff, W.T., et al., Peroxisome proliferator-activated receptor-gamma and retinoid X receptor signaling regulate fatty acid uptake by primary human placental trophoblasts. J Clin Endocrinol Metab, 2005. 90(7): p. 4267-75. 104. Fournier, T., et al., Involvement of PPARgamma in human trophoblast invasion. Placenta, 2007. 28 Suppl A: p. S76-81.

86

105. Schaiff, W.T., et al., Ligand-activated peroxisome proliferator activated receptor gamma alters placental morphology and placental fatty acid uptake in mice. Endocrinology, 2007. 148(8): p. 3625-34. 106. Nadra, K., et al., PPARgamma in placental angiogenesis. Endocrinology, 2010. 151(10): p. 4969-81. 107. McCarthy, F.P., et al., Peroxisome proliferator-activated receptor-gamma as a potential therapeutic target in the treatment of preeclampsia. Hypertension, 2011. 58(2): p. 280-6. 108. Granger, J.P., et al., Reduced uterine perfusion pressure (RUPP) model for studying cardiovascular-renal dysfunction in response to placental ischemia. Methods Mol Med, 2006. 122: p. 383-92. 109. McCarthy, F.P., et al., Evidence implicating peroxisome proliferator-activated receptor-gamma in the pathogenesis of preeclampsia. Hypertension, 2011. 58(5): p. 882-7. 110. Chan, L.Y., J.H. Yeung, and T.K. Lau, Placental transfer of rosiglitazone in the first trimester of human pregnancy. Fertil Steril, 2005. 83(4): p. 955-8. 111. Yaris, F., et al., Normal pregnancy outcome following inadvertent exposure to rosiglitazone, gliclazide, and atorvastatin in a diabetic and hypertensive woman. Reprod Toxicol, 2004. 18(4): p. 619-21. 112. Kalyoncu, N.I., et al., A case of rosiglitazone exposure in the second trimester of pregnancy. Reprod Toxicol, 2005. 19(4): p. 563-4. 113. McCarthy, F.P., et al., PPAR-gamma -- a possible drug target for complicated pregnancies. Br J Pharmacol, 2013. 168(5): p. 1074-85. 114. Kronke, G., et al., Expression of heme oxygenase-1 in human vascular cells is regulated by peroxisome proliferator-activated receptors. Arterioscler Thromb Vasc Biol, 2007. 27(6): p. 1276-82. 115. Levytska, K., et al., Heme oxygenase-1 in placental development and pathology. Placenta, 2013. 34(4): p. 291-8. 116. Maines, M.D., G.M. Trakshel, and R.K. Kutty, Characterization of two constitutive forms of rat liver microsomal heme oxygenase. Only one molecular species of the enzyme is inducible. J Biol Chem, 1986. 261(1): p. 411-9. 117. Shibahara, S., et al., Functional analysis of cDNAs for two types of human heme oxygenase and evidence for their separate regulation. J Biochem, 1993. 113(2): p. 214-8. 118. Yoshida, T., et al., Human heme oxygenase cDNA and induction of its mRNA by hemin. Eur J Biochem, 1988. 171(3): p. 457-61. 119. Tenhunen, R., H.S. Marver, and R. Schmid, Microsomal heme oxygenase. Characterization of the enzyme. J Biol Chem, 1969. 244(23): p. 6388-94. 120. Montellano, P.R., The mechanism of heme oxygenase. Curr Opin Chem Biol, 2000. 4(2): p. 221-7. 121. Neuzil, J. and R. Stocker, Free and albumin-bound bilirubin are efficient co-antioxidants for alpha-tocopherol, inhibiting plasma and low density lipoprotein lipid peroxidation. J Biol Chem, 1994. 269(24): p. 16712-9. 122. Ahmed, A., et al., Induction of placental heme oxygenase-1 is protective against TNFalpha- induced cytotoxicity and promotes vessel relaxation. Mol Med, 2000. 6(5): p. 391-409. 123. Brouard, S., et al., Carbon monoxide generated by heme oxygenase 1 suppresses endothelial cell apoptosis. J Exp Med, 2000. 192(7): p. 1015-26. 124. Sarady, J.K., et al., Carbon monoxide modulates endotoxin-induced production of granulocyte macrophage colony-stimulating factor in macrophages. Am J Respir Cell Mol Biol, 2002. 27(6): p. 739-45. 125. Coceani, F., Carbon monoxide and dilation of blood vessels. Science, 1993. 260(5109): p. 739. 126. Morita, T., et al., Smooth muscle cell-derived carbon monoxide is a regulator of vascular cGMP. Proc Natl Acad Sci U S A, 1995. 92(5): p. 1475-9.

87

127. Acevedo, C.H. and A. Ahmed, Hemeoxygenase-1 inhibits human myometrial contractility via carbon monoxide and is upregulated by progesterone during pregnancy. J Clin Invest, 1998. 101(5): p. 949-55. 128. Bainbridge, S.A., et al., Carbon monoxide decreases perfusion pressure in isolated human placenta. Placenta, 2002. 23(8-9): p. 563-9. 129. Kreiser, D., et al., Gestational pattern of heme oxygenase expression in the rat. Pediatr Res, 2003. 54(2): p. 172-8. 130. Zhao, H., et al., Regulation of maternal and fetal hemodynamics by heme oxygenase in mice. Biol Reprod, 2008. 78(4): p. 744-51. 131. Watanabe, S., et al., Marked developmental changes in heme oxygenase-1 (HO-1) expression in the mouse placenta: correlation between HO-1 expression and placental development. Placenta, 2004. 25(5): p. 387-95. 132. McLaughlin, B.E., et al., Heme oxygenase expression in selected regions of term human placenta. Exp Biol Med (Maywood), 2003. 228(5): p. 564-7. 133. Lyall, F., et al., Hemeoxygenase expression in human placenta and placental bed implies a role in regulation of trophoblast invasion and placental function. FASEB J, 2000. 14(1): p. 208-19. 134. McLean, M., et al., Expression of the heme oxygenase-carbon monoxide signalling system in human placenta. J Clin Endocrinol Metab, 2000. 85(6): p. 2345-9. 135. Yoshiki, N., T. Kubota, and T. Aso, Expression and localization of heme oxygenase in human placental villi. Biochem Biophys Res Commun, 2000. 276(3): p. 1136-42. 136. Baum, M., et al., End-tidal carbon monoxide measurements in women with pregnancy-induced hypertension and preeclampsia. Am J Obstet Gynecol, 2000. 183(4): p. 900-3. 137. Kreiser, D., et al., End tidal carbon monoxide levels are lower in women with gestational hypertension and pre-eclampsia. J Perinatol, 2004. 24(4): p. 213-7. 138. Zenclussen, A.C., et al., Heme oxygenases in pregnancy II: HO-2 is downregulated in human pathologic pregnancies. Am J Reprod Immunol, 2003. 50(1): p. 66-76. 139. Lash, G.E., et al., Relationship between tissue damage and heme oxygenase expression in chorionic villi of term human placenta. Am J Physiol Heart Circ Physiol, 2003. 284(1): p. H160-7. 140. Barber, A., et al., Heme oxygenase expression in human placenta and placental bed: reduced expression of placenta endothelial HO-2 in preeclampsia and fetal growth restriction. FASEB J, 2001. 15(7): p. 1158-68. 141. Poss, K.D. and S. Tonegawa, Reduced stress defense in heme oxygenase 1-deficient cells. Proc Natl Acad Sci U S A, 1997. 94(20): p. 10925-30. 142. Poss, K.D. and S. Tonegawa, Heme oxygenase 1 is required for mammalian iron reutilization. Proc Natl Acad Sci U S A, 1997. 94(20): p. 10919-24. 143. Kapturczak, M.H., et al., Heme oxygenase-1 modulates early inflammatory responses: evidence from the heme oxygenase-1-deficient mouse. Am J Pathol, 2004. 165(3): p. 1045-53. 144. Zhao, H., et al., Effect of heme oxygenase-1 deficiency on placental development. Placenta, 2009. 30(10): p. 861-8. 145. Wong, R.J., H. Zhao, and D.K. Stevenson, A deficiency in haem oxygenase-1 induces foetal growth restriction by placental vasculature defects. Acta Paediatr, 2012. 101(8): p. 827-34. 146. Zhao, H., et al., Maternal heme oxygenase 1 regulates placental vasculature development via angiogenic factors in mice. Biol Reprod, 2011. 85(5): p. 1005-12. 147. Zenclussen, M.L., et al., Haem oxygenase-1 dictates intrauterine fetal survival in mice via carbon monoxide. J Pathol, 2011. 225(2): p. 293-304. 148. Kweider, N., et al., Interplay between vascular endothelial growth factor (VEGF) and nuclear factor erythroid 2-related factor-2 (Nrf2): implications for preeclampsia. J Biol Chem, 2011. 286(50): p. 42863-72.

88

149. Lin, Q., et al., Heme oxygenase-1 protein localizes to the nucleus and activates transcription factors important in oxidative stress. J Biol Chem, 2007. 282(28): p. 20621-33. 150. Bilban, M., et al., Identification of novel trophoblast invasion-related genes: heme oxygenase-1 controls motility via peroxisome proliferator-activated receptor gamma. Endocrinology, 2009. 150(2): p. 1000-13. 151. McCaig, D. and F. Lyall, Heme oxygenase expression in human placental villous tissue in response to exposure to in vitro ischemia-reperfusion injury. Hypertens Pregnancy, 2009. 28(3): p. 256-72. 152. Fitzgerald, B., et al., Villous trophoblast abnormalities in extremely preterm deliveries with elevated second trimester maternal serum hCG or inhibin-A. Placenta, 2011. 32(4): p. 339-45. 153. Cudmore, M., et al., Negative regulation of soluble Flt-1 and soluble endoglin release by heme oxygenase-1. Circulation, 2007. 115(13): p. 1789-97. 154. Cudmore, M.J., et al., Resveratrol inhibits the release of soluble fms-like tyrosine kinase (sFlt-1) from human placenta. Am J Obstet Gynecol, 2012. 206(3): p. 253 e10-5. 155. Sedeek, M., et al., Role of reactive oxygen species in hypertension produced by reduced uterine perfusion in pregnant rats. Am J Hypertens, 2008. 21(10): p. 1152-6. 156. George, E.M., et al., Induction of heme oxygenase 1 attenuates placental ischemia-induced hypertension. Hypertension, 2011. 57(5): p. 941-8. 157. George, E.M., et al., Induction of heme oxygenase-1 attenuates sFlt-1-induced hypertension in pregnant rats. Am J Physiol Regul Integr Comp Physiol, 2011. 301(5): p. R1495-500. 158. George, E.M. and I. Arany, Induction of heme oxygenase-1 shifts the balance from proinjury to prosurvival in the placentas of pregnant rats with reduced uterine perfusion pressure. Am J Physiol Regul Integr Comp Physiol, 2012. 302(5): p. R620-6. 159. Schaefer, K.L., et al., PPARgamma inhibitors reduce tubulin protein levels by a PPARgamma, PPARdelta and proteasome-independent mechanism, resulting in cell cycle arrest, apoptosis and reduced metastasis of colorectal carcinoma cells. Int J Cancer, 2007. 120(3): p. 702-13. 160. McCarthy, F., et al., Evidence implicating peroxisome proliferator-activated receptor-γ in the pathogenesis of preeclampsia. Hypertension, 2011. 58(5): p. 882-887. 161. Lee, G., et al., T0070907, a selective ligand for peroxisome proliferator-activated receptor gamma, functions as an antagonist of biochemical and cellular activities. The Journal of biological chemistry, 2002. 277(22): p. 19649-19657. 162. Vandesompele, J., et al., Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol, 2002. 3(7): p. RESEARCH0034. 163. Coutifaris, C., et al., E-cadherin expression during the differentiation of human trophoblasts. Development, 1991. 113(3): p. 767-77. 164. Basyuk, E., et al., Murine Gcm1 gene is expressed in a subset of placental trophoblast cells. Dev Dyn, 1999. 214(4): p. 303-11. 165. Dawson, P.A., J. Rakoczy, and D.G. Simmons, Placental, renal, and ileal sulfate transporter gene expression in mouse gestation. Biol Reprod, 2012. 87(2): p. 43. 166. Watanabe, M., et al., Regulation of PPAR gamma transcriptional activity in 3T3-L1 adipocytes. Biochem Biophys Res Commun, 2003. 300(2): p. 429-36. 167. Durante, W., et al., cAMP induces heme oxygenase-1 gene expression and carbon monoxide production in vascular smooth muscle. The American journal of physiology, 1997. 273(1 Pt 2): p. 23.

89

Chapter 7: APPENDIX

90

Placenta 33 (2012) 952e954

Contents lists available at SciVerse ScienceDirect

Placenta

journal homepage: www.elsevier.com/locate/placenta

Technical note Revisiting the housekeeping genes of human placental development and insufficiency syndromes

S. Drewlo a,*, K. Levytska a, J. Kingdom a,b a Centre for Research in Women’s and Infant’s health, Samuel Lunenfeld Research Institute, Mount Sinai Hospital, University of Toronto, 25 Orde Street #6-1020, Toronto, ON, Canada M5T 3H7 b Department of Obstetrics & Gynaecology, Mount Sinai Hospital, University of Toronto, Canada article info abstract

Article history: Gene expression analysis using semi-quantitative RT-PCR is a common tool in placental research. Accepted 11 September 2012 However the comparison of steady-state gene transcription between different clinical groups is dependent upon comparison of target mRNA data with mRNA obtained from so-called housekeeping Keywords: (HK) genes whose steady-state transcription does not differ significantly between the groups. In this House keeping genes communication, we evaluated the performance of candidate HK genes across nine clinical groups Placenta commonly used in placental research. We used the GeNorm method to evaluate qRT-PCR data to qRTPCR determine the performance of candidate HKs. Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction samples. In this way, the gene of interest can be normalized to the HK gene’s expression. Use of the geometric mean of multiple The quantification of mRNA gene expression via qRT-PCR is one controls of at least two, and ideally three, HK genes is needed for of the most important tools used in placental research. It offers the optimal precision [2]. Thus far a number of groups have addressed opportunity to compare the expression levels of numerous genes this issue albeit with different results [3e5]. Different house- using small amounts of tissue generated from specific clinical keeping genes have been proposed for use in human placental groups of pathologies and control cases. As little as 5 pg of cDNA is tissue, but thus far, none of the current studies have evaluated HK sufficient for a successful PCR reaction. Although qRT-PCR is genes for potential changes across normal gestation or in the range widely-used, many researchers do not follow established rules to of pregnancy complications attributed to placental dysfunction. ensure the validity of their data. These rules may not be critical when high-fold changes of a target gene are found in comparison 2. Material and methods with controls, but become critical at the lower end of relevant (2e 4-fold) changes in gene expression (Supplemental Fig. 1 [1]). We isolated a total of 90 samples of RNA from 9 different clinical groups of Inappropriate choice of housekeeping gene(s) could lead to wrong placental material to search for ideal housekeeping genes applicable to all condi- tions. These groups are summarized in Table 1; healthy first trimester (early and conclusions, even when sample sizes create mean fold changes that late), healthy second trimester, preterm controls, term controls (vaginal and are statistically significant. In such circumstances, careful assess- caesarian delivery), severe early onset intrauterine growth restriction (IUGR) and ment of potential variation in selected HKG expression could either severe pre-eclampsia (PET) as well as a mixed PET-IUGR group (MIUGR). These cause, or obscure these smaller, but potentially relevant target gene tissues specimens were obtained by the Research Centre for Women’s and Infants’ transcription data. Health (RCWIH) BioBank program of Mount Sinai Hospital, in accordance with the policies of the Mount Sinai Hospital Research Ethics Board (MSH,10-0128-E). The quality of housekeeping gene data is highly dependent on Placentas from singleton pathologic pregnancies and age matched controls of live RNA purity and integrity, as well as the cDNA conversion rate and born infants delivered between 24 þ 0 and 34 þ 6 weeks of gestation. They were the efficiency of the PCR reaction. To compare gene expression in classified into different groups, with characteristics summarized in Table 1: Severe different samples, it is important to use internal controls, described intrauterine growth restriction (IUGR): Birth weight 10%ile, abnormal umbilical artery Doppler (absent end diastolic flow velocity (AEDFV) or reversed end diastolic as housekeeping (HK) genes. Ideally, these genes should be stable flow velocity (REDFV)), normotensive women. Severe pre-eclampsia (PET): Birth with similar abundance as the target genes expressed in all weight >10%ile, BP > 140/90, proteinuria >300 mg/day or 1þ on dipstick. The Mixed IUGR (MIUGR) group contained patients with co-existing features of the two groups above. Pre-term control (PTC): Birth weight >20th centile, no abnormal umbilical artery Doppler measurements, BP < 140/90, no history of gestational * Corresponding author. Tel.: þ1 416 586 4800x8322; fax: þ1 416 586 5116. diabetes, no histologic evidence of chorio-amnionitis. First and early second E-mail address: [email protected] (S. Drewlo). trimester villous tissues from elective social terminations of pregnancy as well as

0143-4004/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.placenta.2012.09.007 S. Drewlo et al. / Placenta 33 (2012) 952e954 953

Abbreviation MSH Mount Sinai Hospital OD optical density AEDFV absent end diastolic flow velocity PBS phosphate-buffered saline REDFV reversed end diastolic flow velocity sPE severe pre-eclampsia Beta2M beta-2-microglobulin PTC pre-term control CYC1 cytochrome c-1 qRT-PCR quantitative real-time polymerase chain reaction cDNA complimentary deoxyribonucleic acid RCWIH Research Centre for Women’s and Infants’ Health GAPDH glyceraldehyde-3-phosphate dehydrogenase TBP TATA box-binding protein HK housekeeping TC term control HPRT hypoxanthine phosphoribosyltransferase TOP1 DNA topoisomerase 1 sIUGR Severe intrauterine growth restriction SDHA succinate dehydrogenase complex, subunit A, MIUGR mixed pre-eclampsia and intrauterine growth flavoprotein (Fp) restriction YWHAZ tyrosine 3-monooxygenase/tryptophan 5- mRNA messenger ribonucleic acid monooxygenase activation protein, zeta polypeptide

healthy term vaginal deliveries as well as group which underwent caesarian section were pipetted with a robot from Perkin Elmer (total volume 7 ml, with LuminoCT (c/s), control placentas (>37 þ n weeks gestation) were also collected for gestational from Sigma Mastermix ¼ 3.5 ml per PCR reaction) and run on the CFX Bio-Rad age profile studies. 384 PCR machine. All samples were run at the same time per gene to avoid Immediately after delivery, placental weight was recorded and four 2e3cm2 inter-assay variations with the following PCR protocol: initial 95 Cfor5min tissue cores through the full thickness of the placenta were obtained from a site followed by 38 cycles of 95 Cfor15sand60C for 20 s. We evaluated 8 within each quadrant, avoiding areas with obvious evidence of thrombosis or commonly-used HG genes in placental research: TBP, HPRT, GAPDH, SDHA, TOP1, other gross abnormalities. Chorionic plate tissue was removed. The resultant CYC, YWHAZ, and beta2M and designed or used previously published primers cores of villous tissue were rinsed briefly in chilled phosphate-buffered saline with an efficiency of at least 95% (sequences are summarized in Supplemental (PBS) to remove residual blood and further dissected to generate 0.5e1cm3 Table 1). We isolated a total of 90 samples of RNA from 9 different clinical pieces. Pooled samples, from individual placentas were generated containing groups of placental material to search for ideal housekeeping genes applicable to one piece of tissue from each of the four cores then snap-frozen. RNA was iso- all conditions. These groups are summarized in Table 1 and encompass healthy Ò lated using the QiaZol extraction kit (Qiagen), then quantified and quality- first trimester (early 6e8weeksandlate9e12 weeks), healthy second trimester, controlled using Nanodrop (Thermo Fisher) and Experion RNA chips (Bio-Rad). preterm controls (before 34 weeks), term controls, severe early onset intra- Respectively, 1 mg of total RNA from all samples was reverse-transcribed simul- uterine growth restriction (IUGR) and severe pre-eclampsia (PET) as well as taneously using the BIORAD i-Script Supermix (Bio-Rad). cDNA samples (10 ng) a mixed PE-IUGR group (MIUGR).

Table 1 Patient data summary.

Groupsa TCb vTermc MIUGR sIUGR sPE Early pregnancy group

PTC 5e7 8e12 13e18 Weeks Weeks Weeks N 1011119109101010 Maternal age 28.6 6.4 36.6 3.4 32.4 9.8 32.3 7.9 31 4.1 33 4.2 (years) Gravida 2 0.7 3 1.8 3 1.4 2.3 2.1 2 1.0 2 1.0 Para 0.4 0.7 1 0.8 1 1.0 1 1.1 1 11 0.8 Gestational age 32 2.1 39 0.8 40 129 2.2 31 2.8 29.7 1.6 6 0.9 10 1.1 15 1.5 (weeks) Ethnicity Black 2 2 1 0 2 5 Caucasian 7 9 9 6 1 2 Asian 0 1 1 2 5 1 Other 0 0 0 0 2 1 N/A 1 0 0 1 0 0 Systolic BP > 140mmHg No 8 4 11 0 10 0 Yes 0 2 0 8 0 9 N/A 1 5 0 1 0 0 Diastolic BP > 90 mmHg No 8 11 11 2 10 0 Yes 0 0 0 5 0 9 N/A 1 0 0 1 0 0 Abnormal umbilical 0009100 cord doppler (AEDV,REDFV) Sex Male 7 3 8 3 6 4 Female 2 6 3 6 4 5 N/A 0 2 0 0 0 0 Birth weight (g) 2057 505.7 3576 112.4 3482 115 804 259.0 933 278.8 1337 480 Placental weight (g) Wet 449 137.7 589 178.5 580 113 209 73.4 300 143.4 338 58

Severe intrauterine growth restriction (IUGR): Birth weight 10%ile, abnormal umbilical artery Doppler (absent end diastolic flow velocity (AEDFV) or reversed end diastolic flow velocity (REDFV)), normotensive women.. Severe pre-eclampsia (PET): Birth weight >10%ile, BP > 140/90, proteinuria >300 mg/day or 1 þ on dipstick. The Mixed IUGR (MIUGR) group contained patients with co-existing features of the two groups above. a Pre-term control (PTC): Birth weight >20th centile, no abnormal umbilical artery Doppler measurements, BP < 140/90, no history of gestational diabetes, no histologic evidence of chorio-amnionitis. b Healthy term delivery by caesarian section. c Healthy term vaginal delivery, (>37 þ n weeks gestation). 954 S. Drewlo et al. / Placenta 33 (2012) 952e954

Table 2 stable genes for analysing gestational changes were YWHAZ, TBP Choice of housekeeping gene by clinical groupings. and CYC1. If PTC was included in this group, the following HK genes Sample groups HKG. 3 HKG. 2 HKG. 1 were identified as appropriate controls: TOP1, TBP and CYC1. When First, second and term non-labour CS þ TOP1 YWHZ CYC1 comparing pathologies (sPE and sIUGR) with preterm and term vaginal delivery controls, TOP1, TBP and CYC1 again gave the best results, which did First, second and term non-labour CS YWHAZ TBP CYC1 not change with the addition of the MIUGR group to the sample First, second, PTC and term non-labour CS TOP1 TBP CYC1 pool. We furthermore investigated if labour has an influence on sPE, sIUGR, PTC and term non-labour CS TOP1 TBP CYC1 sPE, sIUGR, MIX, PTC, and term non-labour CS TOP1 TBP CYC1 housekeeping gene stability by comparing vaginal deliveries versus All groups combined TOP1 YWHAZ CYC1 caesarian sections. We were able to show differential regulation of Vaginal delivery versus CS at term TOP1 CYC1 YWHAZ TBP and HPRT in these groups (Supplemental Table 2). The most HKG ¼ Housekeeping genes. HKG.1 most stable > HKG.2 > HKG.3; HKG1 and 2 stable housekeeping genes were TOP1, CYC1 and YWAHZ. provide a variation index of <0.15 in each group. Finally, if all nine groups are assessed our data indicates that sPE ¼ severe pre-eclampsia, sIUGR ¼ severe intra uterine growth restriction, TOP1, CYC1 and YWHAZ are the most stable HK genes. ¼ ¼ MIX mixed elements of sIUR and sPE, CS Cesarean section. Our data will assist researchers in choosing the appropriate HKG 3. Results and discussion for their experimental setup, although an initial validation of the house keeping genes in a specific sample population is highly recommended. We were able to show that freshly extracted RNA from patho- logical as well as normal tissues can be partially degraded under our standardized collection conditions, when analysed with the Appendix A. Supplementary data Experion RNA chip. Even if the OD values from the Nanodrop provided perfect results (OD ¼ 2), up to 29% of all samples collected Supplementary data related to this article can be found at http:// showed at least partial degradation (Experion QC values below dx.doi.org/10.1016/j.placenta.2012.09.007. <6.5 were excluded). These changes have previously been shown to potentially influence cDNA conversion and PCR performance [6]. References We highly suggest monitoring RNA integrity with a bioanalyzer or comparable method, at least in a selected number of samples. [1] Drewlo S, Levytska K, Sobel M, Baczyk D, Lye SJ, Kingdom JCP. Heparin promotes After the PCR run, we tested the expression stability of our HK soluble VEGF receptor expression in human placental villi to impair endothelial VEGF signaling. Journal of Thrombosis and Haemostasis: JTH 2011-Dec;9(12): genes in all nine groups. We used the GeNorm method published by 2486e97. Vandesompele (2002)(2), which is part of the BiogazelleÔ software [2] Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, De Paepe A, et al. package. The software calculates the stability of each HK gene M for Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes [Research Support, Non-U.S. Gov’t]. a reference gene as the average pairwise variation V for that gene Genome Biology 2002 Jun 18;3(7). RESEARCH0034. with all other tested reference genes. Stepwise exclusion of the [3] Meller M, Vadachkoria S, Luthy DA, Williams MA. Evaluation of housekeeping gene with the highest M value allows ranking of the tested genes genes in placental comparative expression studies. Placenta 2005 SepeOct; e e according to their expression stability. This results in different V 26(8 9):601 7. [4] Murthi P, Fitzpatrick E, Borg AJ, Donath S, Brennecke SP, Kalionis BGAPDH. scores. A V score 0.15 is desired as a cut-off and was determined 18S rRNA and YWHAZ are suitable endogenous reference genes for relative by the fact that the addition of another HK gene would not signif- gene expression studies in placental tissues from human idiopathic fetal ’ icantly improve the sensitivity of the normalization [2]. We fol- growth restriction [Research Support, Non-U.S. Gov t]. Placenta 2008 Sep; 29(9):798e801. lowed these rules to determine the best HK gene combinations for [5] Patel P, Boyd CA, Johnston DG, Williamson C. Analysis of GAPDH as a standard the different groups and summarized these data in Table 2. for gene expression quantification in human placenta [Letter Research Support, We were able to extend previously published data as follows. As Non-U.S. Gov’t]. Placenta 2002 SepeOct;23(8e9):697e8. fl fi [6] Fleige S, Pfaf MW. RNA integrity and the effect on the real-time qRT-PCR predicted, speci c HK genes were differentially regulated in the performance [Review]. Molecular Aspects of Medicine 2006 ApreJun;27(2e3): different groups and, thus, did not meet HK gene criteria. The most 126e39. Placenta 34 (2013) 291e298

Contents lists available at SciVerse ScienceDirect

Placenta

journal homepage: www.elsevier.com/locate/placenta

Current topic Heme oxygenase-1 in placental development and pathology

K. Levytska a,b, J. Kingdom a,b,c,d, D. Baczyk a, S. Drewlo a,* a Program in Development and Fetal Health, Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Canada b Department of Laboratory Medicine and Pathobiology, University of Toronto, Canada c Maternal-Fetal Medicine Division, Department of Obstetrics and Gynecology, Mount Sinai Hospital, Canada d Department of Obstetrics and Gynecology, University of Toronto, Toronto, Canada article info abstract

Article history: Pregnancy is accompanied by several adaptations in the mother, such as increased blood volume, higher Accepted 12 January 2013 cardiac output and reduced peripheral vascular resistance. Inability to accomplish these changes places both her and her pregnancy at risk of major placental complications such severe pre-eclampsia (sPE) or Keywords: severe intra-uterine growth restriction (sIUGR). sPE is characterized by wide-spread maternal vascular Heme oxygenase-1 dysfunction expressed as increased systemic vascular resistance; this state is accompanied by elevated Pre-eclampsia levels of anti-angiogenic factors and lower production of vasodilatory gases. One of the key molecules Placenta implicated in sPE pathogenesis is heme oxygenase-1 (HO-1), a rate-limiting enzyme that breaks down sFLT-1 heme into carbon monoxide (CO), biliverdin and free iron. CO and bilirubin (a downstream product of biliverdin processing) account for the angiogenic, vasodilatory and anti-oxidant properties of HO-1. These collective actions of the heme breakdown metabolites generated by HO-1 offer protection against cytotoxicity, inflammation, hypoxia and other forms of cellular stress that are central to the pathogenesis of sPE. Placental HO-1 expression and exhaled CO levels are both lower in women with sPE, consistent with a pathogenic role of HO-1. In vitro experiments demonstrate that induction of HO-1 downregulates secretion of the anti-angiogenic factor soluble fms-like tyrosine kinase-1 (sFLT-1) and increases CO production. Advancing our understanding of regulatory pathways promoting placental HO- 1 expression may offer new pharmacological tools to reduce maternal and perinatal morbidity in severe placental insufficiency syndromes, especially in women at greatest risk of developing sPE. Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction decreased, cardiac output and a large increase in peripheral vascular resistance [3]. High peripheral vascular resistance implies systemic Successful human pregnancy is accompanied by hemodynamic endothelial cell dysfunction in the mother and precedes the and immune changes in the mother to ensure proper embryo im- development of overt hypertension, one of the hallmarks of sPE. plantation and further fetal development. In normal pregnancy, Systemic vasoconstriction and blood volume contraction, in tandem both maternal cardiac output and blood volume increase by 30e with diseased utero-placental arteries, lead to unstable blood flow 45%; these changes support an exponential rise in utero-placental to the intervillous space [4]. The resultant ischemia-reperfusion blood flow that is needed to support the demands of the growing injury suppresses cytotrophoblast proliferation leading to syncy- fetus [1,2]. These cardiovascular changes alone would predict tiotrophoblast damage and placental infarcts, eventually resulting a significant elevation in blood pressure were it not for the sub- in intrauterine growth restriction (IUGR) [5,6]. Ischemia- stantial reduction in systemic vascular resistance that occurs in the reperfusion injury may create a potent feed-forward loop for vas- second trimester. These maternal hemodynamic adaptations are cular dysfunction and severe hypertension in sPE that, at present, often defective in placental complications of pregnancy, especially can only be reversed by delivery of the fetus and placenta. in severe pre-eclampsia (sPE), where women exhibit unaltered, or The manifestation of maternal vascular dysfunction and hy- pertension stems from the imbalance between pro- and anti- angiogenic factors in maternal blood, and partially due to a decrease in production of vasodilatory gases (such as nitric oxide, * Corresponding author. Samuel Lunenfeld Research Institute, Mount Sinai Hos- carbon monoxide and hydrogen sulfide) [7e9]. The wide-spread pital, 25 Orde St., Room 6-1020, Toronto, ON M5T 3H7, Canada. Tel.: þ1 416 586 4800x8322. endothelial cell dysfunction resulting in maternal hypertension, E-mail addresses: [email protected], [email protected] (S. Drewlo). proteinuria and generalized edema is, in part, attributed to the

0143-4004/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.placenta.2013.01.004 292 K. Levytska et al. / Placenta 34 (2013) 291e298 soluble fms-like tyrosine kinase-1 (sFLT-1) [10]. Excess levels of sFLT-1, a decoy protein for the vascular endothelial growth factor (VEGF), shift the net angiogenic factor signaling response towards an anti-angiogenic phenotype [11]. sFLT-1 levels are elevated in women destined to develop sPE, implicating a role for this mole- cule in disease pathogenesis [12]. Ameliorating the negative effects of excess sFLT-1 via pharmacological and other means has been the focus of several studies [13e15]. Recently, it has been shown that the induction of a cytoprotective enzyme heme oxygenase-1 (HO- 1) may help ameliorate the wide-spread anti-angiogenic pheno- type by decreasing the levels of anti-angiogenic factors, such as sFLT-1 and others, possibly via the actions of carbon monoxide (CO) [15,16]. Properties of HO-1, as discussed in this review, have led researchers to suggest that HO-1 and its downstream factors are potential therapeutic targets in sPE treatment and prevention [17e19].

2. Heme oxygenases

Three isoforms of heme oxygenase have been described: HO-1, HO-2 and HO-3 [20,21]. HO-1, the only inducible HO isoform, is highly expressed in tissues responsible for heme metabolism, such as spleen and liver [22]. HO-2 is constitutively-expressed and mostly found in the brain and testes [20]. The function of HO-3 is presently unknown, but the enzyme is believed to be inactive. HO- Fig. 1. Schematic representation of heme breakdown by HO-1. HO-1 is responsible for heme metabolism within the cell. The process requires NADPH-cytochrome-P450 1, a type I integral protein embedded in smooth ER membrane, was reductase and molecular oxygen (O2) which act as a co-enzyme and a substrate, first isolated from spleen and liver microsomal fractions and shown respectively, for HO-1. HO-1 breaks down heme into three equimolar components: þ to catalyze the rate-limiting step of intracellular heme breakdown iron (Fe2 ), biliverdin (immediately converted to bilirubin by biliverdin reductase), and [23]. Using oxygen, the enzyme metabolizes heme into three CO. These metabolites are responsible for vasodilatory, anti-oxidant and other pro- tective properties of HO-1. components produced in equimolar amounts: carbon monoxide þ (CO), biliverdin and iron (Fe 2) [24]. Under physiological condi- tions, biliverdin is immediately converted into bilirubin by bili- exploring a potential association between HMOX-1 promoter verdin reductase. The two metabolites of heme breakdown, CO and polymorphisms and placental insufficiency syndromes. bilirubin, have important cellular functions giving HO-1 its vaso- dilatory, anti-inflammatory, anti-apoptotic, anti-oxidant, and cytoprotective properties (Fig. 1) [25e32]. 3. Heme oxygenase expression in pregnancy Analyses of HO-1 gene deficiency and promoter polymorphisms have emphasized the importance of this enzyme in growth and Studies of HO-1 and HO-2 expression in mouse, rat and human homeostasis. The first case of HO-1 deficiency in humans was placentas have been numerous [38e42]. However, since the first reported in a 6-year old boy who had both alleles of the gene report localizing HO to human placental tissue, studies of HO-1 mutated [33]. He presented with growth restriction, anemia, ele- expression throughout human gestation and in pathology have vated iron deposition in renal and hepatic tissues, and susceptibility been inconsistent. The expression patterns and activity of HO-1 in to oxidative stress. These symptoms were accompanied by low the rodent and human placenta are summarized in Tables 1 and 2, bilirubin plasma levels and profound endothelial cell damage, respectively. McLean and colleagues were the first group to dem- illustrating the cytoprotective qualities of HO-1. Subsequent genetic onstrate HO-1 and HO-2 expression in the human placenta, studies of HMOX-1 (gene coding for HO-1) have identified an as- showing that HO-2 protein was more abundant than HO-1 at term sociation between the number of nucleotide repeats [(GT)n repeats] [43]. Yoshiki et al. reported that HO-1 and HO-2 proteins are in the promoter region and either cardiovascular diseases or preg- localized in different cell types in the human placenta: HO-1 nancy complications [34e37]. Generally, longer (GT)n repeats in the immunohistochemistry revealed positive staining in the syncytio- promoter correlate with lower HO-1 expression and activity and trophoblast layer with traces in the underlying cytotrophoblasts, have been shown to be associated with a higher risk of recurrent while HO-2 was mostly expressed in endothelial and smooth miscarriage [34]. Conversely, short (25 (GT)n) repeats in the pro- muscle cells [44]. This group also showed that HO-1 expression did moter protect against oxidative stress and lower the risk of coronary not change throughout gestation, while HO-2 expression increased heart disease [37]. Some studies do not support this general hy- towards term. Differential expression of HO-1 in the invading pothesis. For example, the presence of a medium or longer-sized cytotrophoblast cell columns has been noted; HO-1 was found to be (GT)n repeat variant in at least one of the alleles was shown to high in distal, but drastically decreased, in the proximal cell layers correlate with a lower risk of chronic hypertension and a lower [41]. The first study looking at the expression of HO-1 in pregnancy systolic blood pressure in the Chinese Han population [35]. Sur- complications showed lower HO-1 protein expression in pre- prisingly, one study found that the AA genotype of T(-413)A locus eclamptic (PE) placentas; this was in contrast to a positive associ- (vs. TT and TA genotype), which is associated with an increased ation between placental HO-1 expression and feto-placental vas- expression of HO-1, was more common in hypertensive women of cularization in normal pregnancy [26]. In support of pathologic sPE Japanese descent [36]. Although it appears that disrupted function observations, human in vivo studies demonstrate that while of the HMOX-1 gene may participate in the pathogenesis of car- exhaled CO levels increase throughout gestation in healthy women, diovascular and reproductive diseases, at present, no data exist end-tidal CO measurements are lower in hypertensive pregnant K. Levytska et al. / Placenta 34 (2013) 291e298 293

Table 1 Summary of HO-1 and HO-2 expression and function in rodents.

Category Description References Expression over gestation Placental HO-1 expression and HO activity peaks at E14.5 and decreases towards Watanabe et al., 2004 term (mouse) Zhao et al., 2008 Placental HO activity peaks on d21 of pregnancy; HO-1 and HO-2 expression Kreiser et al., 2003 highest on d19, HO-1 mRNA highest on d16 (rat) Placental localization HO-1 protein localized to junction zone, predominantly to spongiotrophoblast cells Watanabe et al., 2004 (mouse) Zhao et al., 2009 HO-1 and HO-2 protein primarily expressed in the syncytiotrophoblast region (rat) Kreiser et al., 2003 Expression in hypoxia or HO-1 downregulated in placental tissue following 48 h of hypoxia (rat) George et al., 2012 hypoxia-reperfusion injury HO-1 downregulated in plasma of RUPP-treated rats George et al., 2011 McCarthy et al., 2011 Expression in pregnancy Reduced placental HO-1 expression in model of spontaneous expression (mouse) Sollwedel et al., 2005 complications Zenclussen et al., 2005 Zenclussen et al., 2006 Reduced HO-1 expression in IL-12-induced model of abortion (mouse) Zenclussen et al., 2002 Genetic studies First HO-1 knock-out model (mouse) Poss & Tonegawa, 1997 Effect of HO-1 deficiency on placentation (mouse) Zhao et al., 2009 Effect of maternal HO-1 deficiency on placental development (mouse) Zhao et al., 2011 Effect of HO-1 deletion on inflammatory responses (mouse) Kapturczak et al., 2004 HO roles in placenta and Inflammation, immunity Schumacher et al., 2012 related models George et al., 2012 Kapturczak et al., 2004 Cytoprotection George et al., 2012 Brouard et al., 2000 Tachibana et al., 2011 Sollwedel et al., 2005 Muscle relaxation, vasodilatation Zhao et al., 2008 Morita et al., 1995

women, especially those affected by sPE, supporting the role of CO [56]. Hypoxia in the established placental ischemia (or reduced as a physiologic vasodilator implicated in gestational hemodynamic uterine perfusion pressure, RUPP) model in the rat has also resulted changes [9,45]. Furthermore, Zenclussen et al. found lower HO-2 in a decrease of circulating plasma HO-1 levels [57]. Thus, studies protein levels in spontaneous abortion (SA) cases (vs. normal on human placental explants challenge the central hypothesis that pregnancy) and in PE (vs. age-matched controls), while HO-1 levels hypoxia and/or hypoxia-reperfusion injury alone is a good ex vivo were low across all pathologies without significant differences [46]. model for sPE, since while placental expression of HO-1 is reduced Lash et al. found that infarcted villi in PE placentas had lower HO in sPE, this finding cannot be recapitulated in hypoxic tissue culture activity and lower expression of HO-2 while HO-1 expression was experiments. Possible explanations include explant gestational age unaffected [47]. In contrast to above studies, Barber et al. showed and tissue heterogeneity, relevance of a simple hypoxic protocol vs. that HO-1 protein was undetectable in human placenta, and there the more likely ischemia-reperfusion injury model (that may be was no difference in HO-2 expression between PE and growth more potent at reducing syncytiotrophoblast protein translation), restricted pregnancies and preterm controls [48]. These studies and others [6]. Hypoxia is the postulated feature of sPE and oxygen show that discrepancies regarding the expression of HO-1 and HO- is an important factor in human and murine placental develop- 2 in disease states and throughout gestation exist and may be due ment. Prolonged hypoxia in disease states leads to reactive oxygen to intra- and inter-tissue heterogeneity. species accumulation and oxidative damage and contributes to placental insufficiency. Although hypoxic stress is not the only 4. HO-1 in hypoxic models of “utero-placental vascular cause of placental syndromes, it is one of the factors in the etiology insufficiency” of disease; thus, hypoxia remains as one of the best models avail- able to researchers in their attempts to understand the physiology Maternal vasoconstriction characteristic of sPE leads to inad- of the placenta and the underlying mechanisms of disease. equate blood supply to the intervillous space, hypoxia-reperfusion injury and tissue necrosis (infarcts) [5,49]. In attempts to recapit- 5. Role of HO-1 in models of spontaneous abortion ulate these conditions ex vivo, placental tissue hypoxia has been the model of choice. It is well-established that HO-1 activity decreases Pregnancy is a state of balance between immune tolerance and with decreased oxygen availability since molecular oxygen is one of immune suppression in the mother to allow her to tolerate the the substrates required for heme breakdown [50]. However, presence of the “foreign” material, that is, the fetus (reviewed in although HO-1 expression has been shown to increase in response Ref. [58]). Several immune responses occur during pregnancy, to hypoxia in several organ systems, placental HO response to including the shift in balance of pro- and anti-inflammatory cyto- hypoxia has not been established [51e53]. For example, Appleton kines towards the latter [59,60]. The balance between anti- et al. showed that HO-2 expression in term placental tissue is un- inflammatory helper T type 2 (Th2) cytokines (such as IL-4, IL-10, affected by different oxygen tensions and HO-1, which showed TGF-b) and pro-inflammatory helper T type 1 (Th1) cytokines (such a very low expression overall, also was unaffected by manipulating as IL-2, IFN-g, TNF-a) has been studied in pregnancy complications oxygen tension [54]. Furthermore, HO-1 and HO-2 expression in and models; while normal pregnancy has been labeled as a Th2 response to hypoxia-reperfusion injury was measured in term phenomenon, this observation does not hold true in pregnancy placental explants and no differences in HO-1 or HO-2 levels were complications [60e64]. Consequently, it has been shown that fol- found [55]. In contrast, in d19 rat placental tissue, HO-1 expression lowing exposure to trophoblast cell-conditioned media, the secre- decreased following a 48-h exposure to hypoxic culture conditions ted ratio of Th1/Th2 cytokines from T lymphocytes decreases, 294 K. Levytska et al. / Placenta 34 (2013) 291e298

Table 2 Summary of HO-1 and HO-2 expression and function in humans and human cell-based models.

Category Description References Expression over gestation First localizations of HO-1 and HO-2 proteins in placental tissue McLean et al., 2000 No differences in HO-1 RNA or protein expression between first Yoshiki et al., 2000 trimester and term placentas; HO-2 RNA and protein increased towards term HO-1 protein expression increases towards term Ahmed et al., 2000 HO-2 protein expression higher in endothelial cells at term Lyall et al., 2000 Placental localization HO-1 found in syncytiotrophoblast with traces in cytotrophoblast; Yoshiki et al., 2000 HO-2 expressed in endothelial and smooth muscle cells HO-2 expressed in syncytiotrophoblast, endothelial cells, Lyall et al., 2000 cytotrophoblast cells in cell columns and shell; HO-1 strongly expressed in distal, but not proximal, layers of invading cytrotrophoblast cell columns HO-2 protein expressed in vascular endothelium, smooth muscle McLean et al., 2000 cells and syncytiotrophoblast; HO-1 expression in syncytiotrophoblast and vascular endothelium, less abundant than HO-2 Expression in hypoxia or Lower HO-1 expression than HO-2; neither is affected by hypoxia Appleton et al., 2003 hypoxia-reperfusion injury in term explants

HO-1 and HO-2 expression unaffected by fluctuating O2 McCaig & Lyall, 2009 concentrations in term explants Expression in pregnancy Decreased HO-1 protein expression in PE Ahmed et al., 2000 complications Lower HO activity in infacted regions of PE placentas; HO-2 levels Lash et al., 2008 decreased in infracted PE regions, while HO-1 protein expression did not change Reduced HO-2 expression in SA vs. normal pregnancy, and Zenchlussen et al., 2003 PE vs. age-matched controls; low HO-1 protein expression across pathologies and not different from age-matched controls HO activity did not differ between mild PE and normotensive McLaughlin et al., 2003 pregnancies; higher HO-1 protein expression in chorionic villi; no differences in HO-2 expression No differences in placental bed expression of HO-1 and HO-2 in Barber et al., 2001 PE and fetal growth restricted (FGR) cases; HO-2 lower in endothelial cells in PE and FGR Genetic studies Human HO-1 gene mutation Yachi et al., 1999 HMOX-1 promoter polymorphisms Denschlag et al., 2004 Lin et al., 2011 Ono et al., 2003 Chen et al., 2012 HO roles in placenta and Inflammation, immunity Neuzil et al., 1994 related models Ahmed et al., 2000 Kweider et al., 2011 Cell migration and fusion, angiogenesis Cudmore et al., 2007 Cudmore et al., 2011 Bilban et al., 2009 McCaig et al., 2009 Muscle relaxation, vasodilatation Bainbridge et al., 2002 Acevedo & Ahmed, 1998 Coceani et al., 1993 Kreiser et al., 2004 Baum et al., 2000 suggesting that the presence of trophoblast cells promotes an anti- Aberrant immune response is one of the key features in murine inflammatory environment [63]. Several studies have looked at the models of spontaneous abortion (SA); conversely, the introduction balance between Th1 and Th2 cytokines during pregnancy and of pro-inflammatory cytokines can be used to induce abortion in showed, for example, that administration of IL-10, a Th2 cytokine, mice [70]. The role of HO-1 in human and murine models of to pregnant rats suffering from endothelial cell dysfunction and abortion as well as its role in the maternal immune response has hypertension improved pregnancy outcomes [65]. Connection has been explored recently. Women with pregnancies ending in been shown between Th2 cytokines and HO-1, for example, where spontaneous abortions have a specific HMOX-1 promoter poly- IL-10 increased the expression of HO-1 in murine macrophages and morphism leading to a lower expression of HO-1 [34]. Studies have exerted its anti-inflammatory effects via the production of CO [66]. also shown reduced levels of placental and decidual HO-1 expres- Furthermore, pregnancy is characterized by an expansion of anti- sion in a murine SA model [62,71,72]. Furthermore, it has been inflammatory regulatory T (Treg) cell population, which is known shown that adenoviral delivery of HO-1 to SA mice is able to to contribute to implantation and early pregnancy maintenance; ameliorate the abortion rates by increasing the IL-4/IFN-g ratio in implicating their role in pregnancy complications is the finding that spleen lymphocytes, as well as decreasing apoptosis at the fetoe Treg cell numbers are lower in human and murine abortion-prone maternal interface [73]. Additionally, it was shown that the trans- pregnancies [67e69]. Collectively, the placenta and maternal fer of pregnancy-induced Treg cells (which are particularly helper T cells contribute to the creation of a favorable, tolerant important for early stages of pregnancy) from healthy pregnant maternal environment which will support the developing fetus. mice into abortion-prone mice decreased fetal rejection rates and These and other adaptations in the maternal immune response increased the levels of HO-1 and TGF-b at the fetoematernal contribute to proper embryo implantation, establishment and interface, without affecting the Th1/Th2 ratio, however [71].In development. another study, inducing HO-1 activity pharmacologically led to K. Levytska et al. / Placenta 34 (2013) 291e298 295 a rise in the expression of an anti-apoptotic factor Bag-1 at the microvasculature blood vessel volumes, less sinusoid spaces in the fetoematernal interface and increased the activation of Treg cells, labyrinth and disorganized capillaries, when compared to wildtype collectively resulting in a lower SA rate [62]. Further supporting the breedings. Interestingly, these alterations in placental morphology role of HO-1 in pregnancy maintenance was a study showing that were independent of the placental genotypes and occurred in both, þ þ þ blocking the activity of HO-1 with zinc protoporphyrin IX (ZnPPIX) Hmox-1 / and Hmox-1 / placentas, as long as both parents were increased abortion rates in normal as well as in abortion-prone Hmox-1 deficient. Since these impaired features of placental pregnancies and abrogated the protective effects of Treg cell development are similar to those observed in pregnancy compli- transfer into abortion-prone mice [74]. Furthermore, induction of cations such as sPE, the role of HO-1 in implantation and placental abortion in mice with IL-12, a cytokine which upregulates the development is further supported. secretion of Th1 cytokines, decreased HO-1 and HO-2 expression, Cell models have also served as a useful tool to study HO-1 led to placental necrosis and hemorrhage and, subsequently, fetal properties which include protection against cytotoxicity and cell rejection [70]. Collectively, studies suggest that HO-1 contributes to death induction of an angiogenic phenotype, and others. HO-1 has proper placental development and pregnancy maintenance in mice been studied in cell models related to pregnancy and placental via its anti-apoptotic properties and plays a role in the induction of development, such as human umbilical vein endothelial cells a tolerant, favorable environment throughout gestation. (HUVECs), HTR8 cells (a model for extravillous trophoblast cells, EVTs), and BeWo cells (a model for syncytiotrophoblast formation). 6. Physiology of HO-1 in placentation and cell-based models The activation of HO-1 by nuclear factor (erythroid-derived 2)-like 2 (Nrf-2), a prominent protector against oxidative stress, was The first description of an HO-1 knockout system was published described in the BeWo cell model, where it protected cells against þ by Poss and Tonegawa in 1997 [75,76]. The authors found that HO-1 hydrogen peroxide- and Fe 2-induced stress [81]. Interestingly, is required for proper iron reutilization, since adult mice lacking a cleaved isoform of HO-1 has been shown to have transcription HO-1 developed anemia and were susceptible to oxidative tissue factor-like activity by contributing to the expression of cytopro- damage and inflammation [76]. Furthermore, the group showed tective genes in cultured fibroblast and Hepa cells [82]. Genetic that murine embryonic fibroblast cells lacking HO-1 were sensitive profiling of highly migratory EVTs and villous cytotrophoblast (CT) to oxidative stress [75]. Kapturczak et al. showed that mice lacking cells (with a low migratory potential) isolated from cultured ex- HO-1 had impaired immune responses and higher levels of pro- plants and first trimester human placentas, respectively, showed inflammatory Th1 factors [77]. Zhao and colleagues were the first a lower expression of HO-1 in EVTs vs. CTs [83]. Confirming these group to provide evidence for HO-1 as a critical player in proper findings, the group also showed that CT cells had higher HO-1 RNA placental development [40,78,79]. The group demonstrated that in expression compared to EVT cells, as well as higher HO-1 protein normal pregnancy, mice exhibit an increase in utero-placental expression in proliferating CTs (vs. EVTs) in first trimester placental blood flow accompanied by a significant increase in maternal explants, as seen by immunohistochemistry. Furthermore, the abdominal aorta diameter (adaptations similar to those observed in study showed that retroviral induction of HO-1 in naturally- humans). These changes were concurrent with higher endogenous migratory HTR-8 cells reduced their motility, while HO-1 silenc- CO production, elevated plasma bilirubin levels, and placental HO-1 ing in non-migratory BeWo cells increased their migration, sug- expression which peaked at E13.5 [40,79]. In a follow-up study, gesting that lower HO-1 expression is associated with a lower cell Zhao et al. characterized the effect of HO-1 deficiency on placental motility and trophoblast invasion. Interestingly, the mechanism development [78]. Crossing Hmox-1 heterozygous (þ/) mice responsible for increasing cell migration by HO-1 was via the reg- resulted in an altered Mendelian ratio of Hmox-1 / pups (2.4% ulation of another factor, a nuclear receptor peroxisome instead of the expected 25% ratio), most of which aborted before proliferator-activated receptor-gamma (PPAR-g) which, as will be þ E10.5. Hmox-1 / placentas presented with a thinner spongio- discussed shortly, has been shown to affect cell migration and in- trophoblast (SP) layer due to increased apoptosis at the junction vasion [83]. These findings are in contrast to the study by McCaig zone, supporting the role of HO-1 as an anti-apoptotic player [78]. et al. who showed that inhibition of global HO activity with zinc The implantation rate in heterozygous crossbreedings was lower, protoporphyrin IX and antibody-mediated inhibition of HO-2 and, at E12.5 and E15.5, the placentas were lighter and smaller reduced invasion of primary CT cells in vitro [55]. Discrepancies in (compared to wildtype breedings). Furthermore, heterozygous studies of how HO affects cell invasion and motility may be due to mothers had higher plasma levels of sFlt-1, one of the key players in the use of different cell models; CT cells isolated from term pla- vascular endothelial cell dysfunction, and elevated diastolic blood centas (when invasion is no longer prominent) in the study by pressure, implying that Hmox-1 is involved in the manifestation of McCaig et al. provide a different model compared to Bilban et al., sPE-like symptoms in mice, mimicking complications seen in who used CT cells isolated from first trimester tissue (when inva- humans [80]. It was later shown that inadequate placentation of sion is high). Furthermore, cell models used by Bilban et al. are Hmox-1 / placentas resulted in IUGR and subsequent fetal death representative of the early placental development stages and, thus, [18]. Exogenous application of CO, a product of heme metabolism might reflect the actions of HO-1 in placental invasion more by HO-1, to Hmox-1-deficient mice during the implantation win- accurately. Although differential mechanisms of HO-1 regulation in dow decreased fetal death in these pregnancies. It was also shown several cell types have been shown, collectively, studies suggest þ þ that Hmox-1 / blastocysts isolated from mice on E3.5 attached that HO-1 contributes to cell migration and invasion. These features þ sooner than Hmox-1 / blastocysts while blastocysts lacking Hmox- are important in sPE since inadequate invasion of the uterus by 1 failed to attach altogether [18]. This study also characterized the EVTs, together with defective formation of the outer syncytio- effect of Hmox-1 deficiency on placentation showing that Hmox-1 trophoblast, imply dual defects in each type of the differentiated is necessary for the differentiation of trophoblast stem cells to trophoblast [84]. giant cells which are critical for proper placental development. Early placental vascularization is also affected by Hmox-1 as seen 7. Molecular manipulations of HO-1 from studies of crossbreeding Hmox-1 wildtype mice and Hmox-1 heterozygous mice [80]. Breeding female mice partially deficient The role of HO-1 in underlying mechanisms of placental insuf- in Hmox-1 with males of the same genotype resulted in ficiency syndromes has been studied extensively. Considering the fetoematernal interface malformations; namely, lower prominent role of anti-angiogenic factors in the manifestation of 296 K. Levytska et al. / Placenta 34 (2013) 291e298 endothelial cell damage, it is natural to study the connection be- resistance and vasoconstriction, all of which are the hallmark fea- tween the protective role of HO-1 and these disease markers. tures of sPE. In normal pregnancy, the physiologic placental Implicating the role of HO-1 in sPE, it was shown that HO-1 in- expression of HO-1 may effectively ameliorate the symptoms of sPE duction or CO administration in HUVECs inhibits their release of and prevent systemic vasculopathy by lowering the release of anti- sFLT-1 and sEndoglin (sEng) [15,16]. Both of these anti-angiogenic angiogenic proteins. The rescue of placental function via the factors contribute to endothelial dysfunction in sPE; thus, rein- manipulation of HO-1 may restore the physiology of the tropho- forcing the hypothesis that HO-1 and heme metabolites promote blast layer by inhibiting the rise in plasma sFLT-1 levels seen in sPE a pro-angiogenic phenotype [15]. This same molecular pathway as well as inducing the production of CO, a potent vasodilator. was studied in a model of placental hypoxia in rats, or the RUPP Caution should be taken when interpreting these data, however, model [85]. Animals which have undergone the RUPP surgery have since the contribution of placental HO-1 to the disease is unknown. elevated mean arterial pressure and exhibit lower fetal weight, Novel interventions exploiting the properties of HO-1 may give suggesting that hypoxia contributes to fetal growth restriction [86]. clinicians the capacity to significantly improve perinatal outcomes, RUPP animals also have elevated plasma sFlt-1 and higher super- should they prove capable of safely advancing pregnancies com- oxide production, both of which can be attenuated with an HO-1 plicated by sPE for even 2e3 weeks. Furthermore, advancing in vitro activity inducer or by the administration of bilirubin or CO. HO-1 pharmacologic tools to the clinical area may provide an opportu- placental expression is decreased in RUPP animals, suggesting nity to move beyond the removal of the placenta as the only that in rat tissue, hypoxia negatively regulates HO-1 [57].As definitive treatment for established sPE. Indeed, manipulation of mentioned previously, HO-1 has the ability to decrease sFLT-1 HO-1 using statins as potential treatment of sPE is already being release; interestingly, HO-1 is also able to offer protection against explored in a clinical trial [95]. sFlt-1 damage while bypassing this effect (i.e., when plasma sFlt-1 levels are held constantly high) [14,15]. George et al. showed that Acknowledgments increasing HO-1 activity in rats with sFlt-1-induced hypertension ameliorated elevated blood pressure despite the constantly high The authors thank Drs. Sarah Keating, Theodore Brown, S. Lee sFlt-1 levels [14]. HO-1 induction improved endothelial function, Adamson (University of Toronto) and Asif Ahmed (University of decreased blood pressure and increased VEGF levels. Supporting Edinburgh) for their helpful discussions and supervision of Ms. the anti-apoptotic role of HO-1, it was shown that HO-1 induction Levytska in her MSc studies. Dr. Kingdom’s research is supported by in the RUPP-treated animals shifted the balance from pro-injury to CIHR (No. 64302) and the Rose Torno Chair at Mount Sinai Hospital, pro-survival phenotype measured by the restoration of intra- Toronto, ON, Canada. cellular ATP levels and increased phosphorylation of survival reg- ulators [87]. References Studies of molecular regulation of HO-1 in several models have fi identi ed a prominent transcriptional factor and nuclear receptor [1] Metcalfe J, Ueland K. Maternal cardiovascular adjustments to pregnancy. Prog PPAR-g as one of the regulators of HO-1 [88e90]. Upon ligand Cardiovasc Dis 1974;16(4):363e74. binding, PPAR-g translocates into the nucleus, binds to the PPAR-g [2] Abbas AE, Lester SJ, Connolly H. Pregnancy and the cardiovascular system. Int J Cardiol 2005;98(2):179e89. response element (PPRE), and induces transcription of genes [3] Melchiorre K, Sutherland GR, Liberati M, Thilaganathan B. Maternal car- involved in fat metabolism and lipid peroxidation. The role of PPAR- diovascular impairment in pregnancies complicated by severe fetal growth g in trophoblast layer dynamics has been explored; some groups restriction. Hypertension 2012;60(2):437e43. g [4] Burton GJ, Woods AW, Jauniaux E, Kingdom JC. Rheological and physiological have shown that activation of PPAR- activity controlled the consequences of conversion of the maternal spiral arteries for uteroplacental expression of genes critical for trophoblast formation, such as glial blood flow during human pregnancy. Placenta 2009;30(6):473e82. cell missing-1 and syncytin-1, while others demonstrated its [5] Franco C, Walker M, Robertson J, Fitzgerald B, Keating S, McLeod A, et al. involvement in cell invasion and migration, under the influence of Placental infarction and thrombophilia. Obstet Gynecol 2011;117(4): 929e34. HO-1 induction [83,91e94]. More importantly, in line with the roles [6] Yung HW, Calabrese S, Hynx D, Hemmings BA, Cetin I, Charnock-Jones DS, of HO-1 and PPAR-g in sPE pathogenesis, PPAR-g has been sug- et al. Evidence of placental translation inhibition and endoplasmic reticulum gested to be one of the key targets for the treatment of sPE based on stress in the etiology of human intrauterine growth restriction. Am J Pathol g 2008;173(2):451e62. the rat RUPP model [19,90]. Induction of PPAR- activity increased [7] Holwerda KM, Bos EM, Rajakumar A, Ris-Stalpers C, van Pampus MG, circulating HO-1 levels and was able to ameliorate PE-like symp- Timmer A, et al. Hydrogen sulfide producing enzymes in pregnancy and toms (i.e., high blood pressure, endothelial cell dysfunction and preeclampsia. Placenta 2012;33(6):518e21. [8] Kulandavelu S, Whiteley KJ, Qu D, Mu J, Bainbridge SA, Adamson SL. Endo- high microalbumin:creatine ratio) in RUPP-treated animals [19]. thelial nitric oxide synthase deficiency reduces uterine blood flow, spiral ar- Further supporting the role of PPAR-g in sPE pathogenesis was the tery elongation, and placental oxygenation in pregnant mice. Hypertension finding that blocking PPAR-g activity increased circulating sFlt-1 2012;60(1):231e8. [9] Baum M, Schiff E, Kreiser D, Dennery PA, Stevenson DK, Rosenthal T, et al. levels, an analogous feature to the one found in mice with End-tidal carbon monoxide measurements in women with pregnancy- decreased Hmox-1 expression [78,90]. These studies support the induced hypertension and preeclampsia. Am J Obstet Gynecol 2000;183(4): hypothesis that pharmacologic interventions aimed at increasing 900e3. [10] Diagnosis and management of preeclampsia and eclampsia. Number 33 ACOG HO-1 expression and activity in the placenta may help restore practice bulletin. American college of obstetricians and gynecologists. Int J maternal cardiovascular function which is defective in sPE. Gynaecol Obstet January 2002;77(1):67e75. [11] Maynard SE, Min JY, Merchan J, Lim KH, Li J, Mondal S, et al. Excess placental 8. Conclusions and future directions soluble fms-like tyrosine kinase 1 (sFlt1) may contribute to endothelial dys- function, hypertension, and proteinuria in preeclampsia. J Clin Invest 2003; 111(5):649e58. sPE pathology originates as a disorder of placental, specifically [12] Levine RJ, Maynard SE, Qian C, Lim KH, England LJ, Yu KF, et al. Circulating angiogenic factors and the risk of preeclampsia. N Engl J Med 2004;350(7): trophoblast, function leading to the clinical manifestations of e “ fi ” 672 83. placental insuf ciency. Developmentally-abnormal sPE placenta [13] Thadhani R, Kisner T, Hagmann H, Bossung V, Noack S, Schaarschmidt W, et al. is characterized by defective trophoblast layer turnover and syn- Pilot study of extracorporeal removal of soluble fms-like tyrosine kinase 1 in cytiotrophoblast formation as well as abnormal shedding of anti- preeclampsia. Circulation 2011;124(8):940e50. [14] George EM, Arany M, Cockrell K, Storm MV, Stec DE, Granger JP. Induction of angiogenic proteins. Their excess release into maternal serum heme oxygenase-1 attenuates sFlt-1-induced hypertension in pregnant rats. leads to endothelial cell damage, systemic vascular peripheral Am J Physiol Regul Integr Comp Physiol 2011;301(5):R1495e500. K. Levytska et al. / Placenta 34 (2013) 291e298 297

[15] Cudmore M, Ahmad S, Al-Ani B, Fujisawa T, Coxall H, Chudasama K, et al. [42] McLaughlin BE, Lash GE, Smith GN, Marks GS, Nakatsu K, Graham CH, et al. Negative regulation of soluble Flt-1 and soluble endoglin release by heme Heme oxygenase expression in selected regions of term human placenta. Exp oxygenase-1. Circulation 2007;115(13):1789e97. Biol Med (Maywood) 2003;228(5):564e7. [16] Cudmore MJ, Ramma W, Cai M, Fujisawa T, Ahmad S, Al-Ani B, et al. [43] McLean M, Bowman M, Clifton V, Smith R, Grossman AB. Expression of the Resveratrol inhibits the release of soluble fms-like tyrosine kinase (sFlt-1) heme oxygenase-carbon monoxide signalling system in human placenta. from human placenta. Am J Obstet Gynecol 2012;206(3). 253 e10-5. J Clin Endocrinol Metab 2000;85(6):2345e9. [17] El-Mousleh T, Casalis PA, Wollenberg I, Zenclussen ML, Volk HD, Langwisch S, [44] Yoshiki N, Kubota T, Aso T. Expression and localization of heme oxygenase in et al. Exploring the potential of low doses carbon monoxide as therapy in human placental villi. Biochem Biophys Res Commun 2000;276(3):1136e42. pregnancy complications. Med Gas Res 2012;2(1):4. [45] Kreiser D, Baum M, Seidman DS, Fanaroff A, Shah D, Hendler I, et al. End tidal [18] Zenclussen ML, Casalis PA, El-Mousleh T, Rebelo S, Langwisch S, Linzke N, carbon monoxide levels are lower in women with gestational hypertension et al. Haem oxygenase-1 dictates intrauterine fetal survival in mice via carbon and pre-eclampsia. J Perinatol 2004;24(4):213e7. monoxide. J Pathol 2011;225(2):293e304. [46] Zenclussen AC, Lim E, Knoeller S, Knackstedt M, Hertwig K, Hagen E, et al. [19] McCarthy FP, Drewlo S, Kingdom J, Johns EJ, Walsh SK, Kenny LC. Peroxisome Heme oxygenases in pregnancy II: HO-2 is downregulated in human patho- proliferator-activated receptor-gamma as a potential therapeutic target in the logic pregnancies. Am J Reprod Immunol 2003;50(1):66e76. treatment of preeclampsia. Hypertension 2011;58(2):280e6. [47] Lash GE, McLaughlin BE, MacDonald-Goodfellow SK, Smith GN, Brien JF, [20] Maines MD, Trakshel GM, Kutty RK. Characterization of two constitutive Marks GS, et al. Relationship between tissue damage and heme oxygenase forms of rat liver microsomal heme oxygenase. Only one molecular species of expression in chorionic villi of term human placenta. Am J Physiol Heart Circ the enzyme is inducible. J Biol Chem 1986;261(1):411e9. Physiol 2003;284(1):H160e7. [21] Shibahara S, Yoshizawa M, Suzuki H, Takeda K, Meguro K, Endo K. Functional [48] Barber A, Robson SC, Myatt L, Bulmer JN, Lyall F. Heme oxygenase expression analysis of cDNAs for two types of human heme oxygenase and evidence for in human placenta and placental bed: reduced expression of placenta endo- their separate regulation. J Biochem 1993;113(2):214e8. thelial HO-2 in preeclampsia and fetal growth restriction. FASEB J 2001;15(7): [22] Yoshida T, Biro P, Cohen T, Müller RM, Shibahara S. Human heme oxy- 1158e68. genase cDNA and induction of its mRNA by hemin. Eur J Biochem 1988; [49] Hung TH, Skepper JN, Charnock-Jones DS, Burton GJ. Hypoxia-reoxygenation: 171(3):457e61. a potent inducer of apoptotic changes in the human placenta and possible [23] Tenhunen R, Marver HS, Schmid R. Microsomal heme oxygenase. Character- etiological factor in preeclampsia. Circ Res 2002;90(12):1274e81. ization of the enzyme. J Biol Chem 1969;244(23):6388e94. [50] Appleton SD, Marks GS, Nakatsu K, Brien JF, Smith GN, Graham CH. Heme [24] Montellano PR. The mechanism of heme oxygenase. Curr Opin Chem Biol oxygenase activity in placenta: direct dependence on oxygen availability. 2000;4(2):221e7. Am J Physiol Heart Circ Physiol 2002;282(6):H2055e9. [25] Neuzil J, Stocker R. Free and albumin-bound bilirubin are efficient co- [51] Panchenko MV, Farber HW, Korn JH. Induction of heme oxygenase-1 by antioxidants for alpha-tocopherol, inhibiting plasma and low density lip- hypoxia and free radicals in human dermal fibroblasts. Am J Physiol Cell oprotein lipid peroxidation. J Biol Chem 1994;269(24):16712e9. Physiol 2000;278(1):C92e101. [26] Ahmed A, Rahman M, Zhang X, Acevedo CH, Nijjar S, Rushton I, et al. In- [52] Motterlini R, Foresti R, Bassi R, Calabrese V, Clark JE, Green CJ. Endothelial duction of placental heme oxygenase-1 is protective against TNFalpha- heme oxygenase-1 induction by hypoxia. Modulation by inducible nitric- induced cytotoxicity and promotes vessel relaxation. Mol Med 2000;6(5): oxide synthase and S-nitrosothiols. J Biol Chem 2000;275(18):13613e20. 391e409. [53] Carraway MS, Ghio AJ, Carter JD, Piantadosi CA. Expression of heme [27] Brouard S, Otterbein LE, Anrather J, Tobiasch E, Bach FH, Choi AM, Soares MP. oxygenase-1 in the lung in chronic hypoxia. Am J Physiol Lung Cell Mol Carbon monoxide generated by heme oxygenase 1 suppresses endothelial cell Physiol 2000;278(4):L806e12. apoptosis. J Exp Med 2000;192(7):1015e26. [54] Appleton SD, Marks GS, Nakatsu K, Brien JF, Smith GN, Graham CH, et al. Ef- [28] Sarady JK, Otterbein SL, Liu F, Otterbein LE, Choi AM. Carbon monoxide fects of hypoxia on heme oxygenase expression in human chorionic villi ex- modulates endotoxin-induced production of granulocyte macrophage plants and immortalized trophoblast cells. Am J Physiol Heart Circ Physiol colony-stimulating factor in macrophages. Am J Respir Cell Mol Biol 2002; 2003;284(3):H853e8. 27(6):739e45. [55] McCaig D, Lyall F. Heme oxygenase expression in human placental villous [29] Coceani F. Carbon monoxide and dilation of blood vessels. Science 1993; tissue in response to exposure to in vitro ischemia-reperfusion injury. 260(5109):739. Hypertens Pregnancy 2009;28(3):256e72. [30] Morita T, Perrella MA, Lee ME, Kourembanas S. Smooth muscle cell-derived [56] George EM, Colson D, Dixon J, Palei AC, Granger JP. Heme oxygenase-1 at- carbon monoxide is a regulator of vascular cGMP. Proc Natl Acad Sci U S A tenuates hypoxia-induced sFlt-1 and oxidative stress in placental villi through 1995;92(5):1475e9. its metabolic products CO and bilirubin. Int J Hypertens 2012;2012:486053. [31] Acevedo CH, Ahmed A. Hemeoxygenase-1 inhibits human myometrial con- [57] George EM, Cockrell K, Aranay M, Csongradi E, Stec DE, Granger JP. Induction tractility via carbon monoxide and is upregulated by progesterone during of heme oxygenase 1 attenuates placental ischemia-induced hypertension. pregnancy. J Clin Invest 1998;101(5):949e55. Hypertension 2011;57(5):941e8. [32] Bainbridge SA, Farley AE, McLaughlin BE, Graham CH, Marks GS, Nakatsu K, [58] Leber A, Zenclussen ML, Teles A, Brachwitz N, Casalis P, El-Mousleh T, et al. et al. Carbon monoxide decreases perfusion pressure in isolated human pla- Pregnancy: tolerance and suppression of immune responses. Methods Mol centa. Placenta 2002;23(8e9):563e9. Biol 2011;677:397e417. [33] Yachie A, Niida Y, Wada T, Igarashi N, Kaneda H, Toma T, et al. Oxidative stress [59] Gonzalez IT, Barrientos G, Freitag N, Otto T, Thijssen VL, Moschansky P, et al. causes enhanced endothelial cell injury in human heme oxygenase-1 defi- Uterine NK cells are critical in shaping DC immunogenic functions compatible ciency. J Clin Invest 1999;103(1):129e35. with pregnancy progression. PLoS One 2012;7(10):e46755. [34] Denschlag D, Marculescu R, Unfried G, Hefler LA, Exner M, Hashemi A, et al. [60] Wegmann TG, Lin H, Guilbert L, Mosmann TR. Bidirectional cytokine in- The size of a microsatellite polymorphism of the haem oxygenase 1 gene is teractions in the maternal-fetal relationship: is successful pregnancy a TH2 associated with idiopathic recurrent miscarriage. Mol Hum Reprod 2004; phenomenon? Immunol Today 1993;14(7):353e6. 10(3):211e4. [61] Tachibana M, Hashino M, Nishida T, Shimizu T, Watarai M. Protective role of [35] Lin R, Fu W, Zhou W, Wang Y, Wang X, Huang W, et al. Association of heme heme oxygenase-1 in Listeria monocytogenes-induced abortion. PLoS One oxygenase-1 gene polymorphisms with essential hypertension and blood 2011;6(9):e25046. pressure in the Chinese Han population. Genet Test Mol Biomarkers 2011; [62] Sollwedel A, Bertoja AZ, Zenclussen ML, Gerlof K, Lisewski U, Wafula P, et al. 15(1e2):23e8. Protection from abortion by heme oxygenase-1 up-regulation is associated [36] Ono K, Mannami T, Iwai N. Association of a promoter variant of the haeme with increased levels of Bag-1 and neuropilin-1 at the fetal-maternal inter- oxygenase-1 gene with hypertension in women. J Hypertens 2003;21(8): face. J Immunol 2005;175(8):4875e85. 1497e503. [63] Liu F, Guo J, Tian T, Wang H, Dong F, Huang H, et al. Placental trophoblasts [37] Chen M, Zhou L, Ding H, Huang S, He M, Zhang X, et al. Short (GT) ( n ) repeats shifted Th1/Th2 balance toward Th2 and inhibited Th17 immunity at feto- in heme oxygenase-1 gene promoter are associated with lower risk of coro- maternal interface. APMIS 2011;119(9):597e604. nary heart disease in subjects with high levels of oxidative stress. Cell Stress [64] Dong M, He J, Wang Z, Xie X, Wang H. Placental imbalance of Th1- and Th2- Chaperones 2012;17(3):329e38. type cytokines in preeclampsia. Acta Obstet Gynecol Scand 2005;84(8):788e [38] Kreiser D, Kelly DK, Seidman DS, Stevenson DK, Baum M, Dennery PA. Ges- 93. tational pattern of heme oxygenase expression in the rat. Pediatr Res 2003; [65] Tinsley JH, South S, Chiasson VL, Mitchell BM. Interleukin-10 reduces 54(2):172e8. inflammation, endothelial dysfunction, and blood pressure in hypertensive [39] Watanabe S, Akagi R, Mori M, Tsuchiya T, Sassa S. Marked developmental pregnant rats. Am J Physiol Regul Integr Comp Physiol 2010;298(3):R713e9. changes in heme oxygenase-1 (HO-1) expression in the mouse placenta: [66] Lee TS, Chau LY. Heme oxygenase-1 mediates the anti-inflammatory effect of correlation between HO-1 expression and placental development. Placenta interleukin-10 in mice. Nat Med 2002;8(3):240e6. 2004;25(5):387e95. [67] Somerset DA, Zheng Y, Kilby MD, Sansom DM, Drayson MT. Normal human [40] Zhao H, Wong RJ, Doyle TC, Nayak N, Vreman HJ, Contag CH, et al. Regulation pregnancy is associated with an elevation in the immune suppressive CD25þ of maternal and fetal hemodynamics by heme oxygenase in mice. Biol Reprod CD4þ regulatory T-cell subset. Immunology 2004;112(1):38e43. 2008;78(4):744e51. [68] Sasaki Y, Sakai M, Miyazaki S, Higuma S, Shiozaki A, Saito S. Decidual and [41] Lyall F, Barber A, Myatt L, Bulmer JN, Robson SC. Hemeoxygenase peripheral blood CD4þCD25þ regulatory T cells in early pregnancy subjects expression in human placenta and placental bed implies a role in regu- and spontaneous abortion cases. Mol Hum Reprod 2004;10(5):347e53. lation of trophoblast invasion and placental function. FASEB J 2000;14(1): [69] Zenclussen AC, Gerlof K, Zenclussen ML, Sollwedel A, Bertoja AZ, Ritter T, et al. 208e19. Abnormal T-cell reactivity against paternal antigens in spontaneous abortion: 298 K. Levytska et al. / Placenta 34 (2013) 291e298

adoptive transfer of pregnancy-induced CD4þCD25þ T regulatory cells pre- [83] Bilban M, Haslinger P, Prast J, Klinglmüller F, Woelfel T, Haider S, et al. vents fetal rejection in a murine abortion model. Am J Pathol 2005;166(3): Identification of novel trophoblast invasion-related genes: heme oxygenase-1 811e22. controls motility via peroxisome proliferator-activated receptor gamma. [70] Zenclussen AC, Joachim R, Hagen E, Peiser C, Klapp BF, Arck PC. Heme oxy- Endocrinology 2009;150(2):1000e13. genase is downregulated in stress-triggered and interleukin-12-mediated [84] Fitzgerald B, Levytska K, Kingdom J, Walker M, Baczyk D, Keating S. Villous murine abortion. Scand J Immunol 2002;55(6):560e9. trophoblast abnormalities in extremely preterm deliveries with elevated [71] Zenclussen AC, Gerlof K, Zenclussen ML, Ritschel S, Zambon Bertoja A, Fest S, second trimester maternal serum hCG or inhibin-A. Placenta 2011;32(4): et al. Regulatory T cells induce a privileged tolerant microenvironment at the 339e45. fetal-maternal interface. Eur J Immunol 2006;36(1):82e94. [85] Granger JP, LaMarca BB, Cockrell K, Sedeek M, Balzi C, Chandler D, et al. [72] Zenclussen AC, Sollwedel A, Bertoja AZ, Gerlof K, Zenclussen ML, Reduced uterine perfusion pressure (RUPP) model for studying Woiciechowsky C, et al. Heme oxygenase as a therapeutic target in cardiovascular-renal dysfunction in response to placental ischemia. Methods immunological pregnancy complications. Int Immunopharmacol 2005;5(1): Mol Med 2006;122:383e92. 41e51. [86] Sedeek M, Gilbert JS, LaMarca BB, Sholook M, Chandler DL, Wang Y, et al. Role of [73] Zenclussen ML, Anegon I, Bertoja AZ, Chauveau C, Vogt K, Gerlof K, et al. Over- reactive oxygen species in hypertension produced by reduced uterine perfusion expression of heme oxygenase-1 by adenoviral gene transfer improves in pregnant rats. Am J Hypertens 2008;21(10):1152e6. pregnancy outcome in a murine model of abortion. J Reprod Immunol 2006; [87] George EM, Arany I. Induction of heme oxygenase-1 shifts the balance from 69(1):35e52. proinjury to prosurvival in the placentas of pregnant rats with reduced [74] Schumacher A, Wafula PO, Teles A, El-Mousleh T, Linzke N, Zenclussen ML, uterine perfusion pressure. Am J Physiol Regul Integr Comp Physiol 2012; et al. Blockage of heme oxygenase-1 abrogates the protective effect of regu- 302(5):R620e6. latory T cells on murine pregnancy and promotes the maturation of dendritic [88] Kronke G, Kadl A, Ikonomu E, Blüml S, Fürnkranz A, Sarembock IJ, et al. cells. PLoS One 2012;7(8):e42301. Expression of heme oxygenase-1 in human vascular cells is regulated by [75] Poss KD, Tonegawa S. Reduced stress defense in heme oxygenase 1-deficient peroxisome proliferator-activated receptors. Arterioscler Thromb Vasc Biol cells. Proc Natl Acad Sci U S A 1997;94(20):10925e30. 2007;27(6):1276e82. [76] Poss KD, Tonegawa S. Heme oxygenase 1 is required for mammalian iron [89] Cheng CF, Lian WS, Chen SH, Lai PF, Li HF, Lan YF, et al. Protective effects of reutilization. Proc Natl Acad Sci U S A 1997;94(20):10919e24. adiponectin against renal ischemia-reperfusion injury via prostacyclin- [77] Kapturczak MH, Wasserfall C, Brusko T, Campbell-Thompson M, Ellis TM, PPARalpha-heme oxygenase-1 signaling pathway. J Cell Physiol 2011; Atkinson MA, et al. Heme oxygenase-1 modulates early inflammatory re- 227(1):239e49. sponses: evidence from the heme oxygenase-1-deficient mouse. Am J Pathol [90] McCarthy FP, Drewlo S, English FA, Kingdom J, Johns EJ, Kenny LC, et al. Ev- 2004;165(3):1045e53. idence implicating peroxisome proliferator-activated receptor-gamma in the [78] Zhao H, Wong RJ, Kalish FS, Nayak NR, Stevenson DK. Effect of heme pathogenesis of preeclampsia. Hypertension 2011;58(5):882e7. oxygenase-1 deficiency on placental development. Placenta 2009;30(10): [91] Fournier T, Handschuh K, Tsatsaris V, Evain-Brion D. Involvement of PPAR- 861e8. gamma in human trophoblast invasion. Placenta 2007;28(Suppl. A):S76e81. [79] Wong RJ, Zhao H, Stevenson DK. A deficiency in haem oxygenase-1 induces [92] Fournier T, Guibourdenche J, Handschuh K, Tsatsaris V, Rauwel B, Davrinche C, foetal growth restriction by placental vasculature defects. Acta Paediatr 2012; et al. PPARgamma and human trophoblast differentiation. J Reprod Immunol 101(8):827e34. 2011;90(1):41e9. [80] Zhao H, Azuma J, Kalish F, Wong RJ, Stevenson DK. Maternal heme oxy- [93] Parast MM, Yu H, Ciric A, Salata MW, Davis V, Milstone DS. PPARgamma genase 1 regulates placental vasculature development via angiogenic fac- regulates trophoblast proliferation and promotes labyrinthine trilineage dif- tors in mice. Biol Reprod 2011;85(5):1005e12. ferentiation. PLoS One 2009;4(11):e8055. [81] Kweider N, Fragoulis A, Rosen C, Pecks U, Rath W, Pufe T, et al. Interplay [94] Ruebner M, Langbein M, Strissel PL, Henke C, Schmidt D, Goecke TW, et al. between vascular endothelial growth factor (VEGF) and nuclear factor Regulation of the human endogenous retroviral Syncytin-1 and cell-cell erythroid 2-related factor-2 (Nrf2): implications for preeclampsia. J Biol Chem fusion by the nuclear hormone receptors PPARgamma/RXRalpha in pla- 2011;286(50):42863e72. centogenesis. J Cell Biochem 2012;113(7):2383e96. [82] Lin Q, Weis S, Yang G, Weng YH, Helston R, Rish K, et al. Heme oxygenase-1 [95] Ahmed A. New insights into the etiology of preeclampsia: identification of key protein localizes to the nucleus and activates transcription factors impor- elusive factors for the vascular complications. Thromb Res 2011;127(Suppl. tant in oxidative stress. J Biol Chem 2007;282(28):20621e33. 3):S72e5.