A Role for Mitochondrial Enzymes SDH and SOD2 in Thyroid Cancer

A role for mitochondrial enzymes SDH and SOD2 in thyroid cancer

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy

in the Graduate School of The Ohio State University

Amruta Ashtekar

Graduate Program in Molecular Cellular and Developmental Biology

The Ohio State University

2018

Dissertation Committee

Dr. Lawrence Kirschner, Advisor

Dr. Matt Ringel

Dr. Thomas Ludwig

Dr. Christin Burd

Copyrighted by

Amruta Ashtekar

2018

Abstract

Thyroid cancer is the most common endocrine malignancy, accounting for about

1% of all the newly diagnosed cancer cases. Well-differentiated epithelial thyroid cancer can be divided into papillary thyroid cancer (PTC) and follicular thyroid cancer (FTC).

The work described here characterizes two new mouse models with genetic alterations that mimic alterations in thyroid cancer, focusing on the two mitochondrial enzymes – succinate dehydrogenase (SDH) complex and superoxide dismutase 2 (MnSod, encoded by SOD2) that may cause and/or modify the risk of thyroid cancer.

SDH mutations are predisposing factors in various other hereditary cancers, most notably paragangliomas (PGL) and pheochromocytomas (PHEO) and recently found to be associated with thyroid cancer. In order to gain mechanistic insight into SDH mediated carcinogenesis, we generated mice lacking the SDH subunit D (Sdhd) in the thyroid. We report that these mice develop enlarged thyroid glands with follicle hypercellularity and increased proliferation. Interestingly, SDHD knockdown cells acquire stem-like features which are also observed in the mouse tumors. The stem-like characteristics are reversed by α-ketoglutarate, suggesting that SDHD-associated tumorigenesis results from dedifferentiation driven by an imbalance in cellular metabolites of the TCA cycle.

Oxidative stress (OS) due to generation of reactive oxygen species (ROS) is implicated to alter the behavior of non-medullary thyroid cancers. We used genetic mouse models of FTC to over- or under-express the antioxidant enzyme MnSod, which

iii plays a key role in regulation of ROS. In benign thyroid tumors, Sod2 overexpression led to increased tumor burden accompanied by increased cellular proliferation. In contrast, overexpression of Sod2 reduced the tumor proliferation and mortality of mice with aggressive metastatic FTC. In an analysis of human endocrine cancers, we found downregulation of SOD2/Sod2 gene expression in FTC but not PTC. This finding was recapitulated in mouse models of follicular adenomas and FTC by microarray analysis as well. Overall, our results indicate that SOD2 has dichotomous role on cancer progression and acts in a context specific manner. The results of this study reveal a metabolic vulnerability for potential future treatment of SDH-associated neoplasia and warrant caution in using antioxidants for thyroid cancer treatment.

Acknowledgments

First and foremost, I would like to express my gratitude to my advisor, Lawrence

S. Kirschner for allowing me to join his lab, and for his mentorship and encouragement at all times. I could not have imagined a better mentor. Thank you for your support, and for making me a better scientist and a critical thinker. Special thanks to all those in the lab who helped me along the way: Zahida Qamri, Danielle Huk, Suresh Kari, Jeff Chin,

Alexa Magner and Karthik Chakravarthy. I couldn’t have done it without you.

I also want to acknowledge all the collaborators who have helped with the experiments presented in my thesis: Dr. Krista La-Perle for pathology analysis and her expertise into thyroid carcinogenesis; Xialoi Zhang and Maciej Pietrzak for their excellent statistical and bioinformatics analysis, Dr Xuguang Zhu and Shueu-yann Cheng for performing TSH measurements on the mouse serum; Dr. Sissy Jhiang for performing

NIS quantitative PCR; Vasily Vasko for staining of human tissues; Dr Jean-Christophe

Cocuron and Targeted Metabolomics Laboratory for performing the mass spectrometry analysis of cellular metabolites, Dr. Denis Guttridge for allowing me to use Seahorse; Dr.

Nick Denko for sharing his insights into hypoxia pathways; Dr. Motoyasu Saji and

Samantha McCarty for their insights into thyroid cancer research, as well as providing the human thyroid cancer cell lines. Thank you present and past members of Ringel lab

Steven, Adlina, Christina and Anisley for being extremely helpful whenever I needed you. Additionally, I’d like to thank Sara Cole, Richard Montione and Angela Blissett for their incredible work on TEM imaging. I would like to thank Dr Daret St Clair for

v providing the Sod2-Tg mouse model as well as Dr. Piruat and Dr. López-Barneo for providing SdhdL/L mice.

A special thank you to my graduate thesis committee: Dr. Mike Ostrowski, Dr.

Matt Ringel and Dr. Thomas Ludwig. Thank you to Dr. Christin Burd for joining my committee. Additionally, Dr Ringel provided insightful comment on this work during its development. All of your insights have been exceptionally helpful throughout this process. Thank you Nanci Edgington for your outstanding work in coordinating the thyroid cancer program.

Thank you Pelotonia Fellowship program, for their support of the 2016 Graduate

Pelotonia Fellowship and OSUMC annual research trainee day award to provide me with travel funding.

Sri Karthika, Yajun Liu, Geeta Palsule, and Vivian Jung- I am so lucky to have you as classmates. I survived the graduate school, especially the first year solely because of you.

Special thanks to my extended family in US and abroad to offer me your support.

Lastly, the one person who stuck with me throughout- Omkar. You supported me throughout my PhD years and endured my endless rantings about science.

Vita

Diploma, Sarda KVM High school...... 2000

SSC, RYK College...... 2002

BS in Biotechnology, University of Pune...... 2007

MS in Biodiversity, University of Pune...... 2009

MS in Cell Molecular Biology...... 2011

PhD in Cell Molecular Developmental Biology...... 2013- present

Publications

Ashtekar A, Huk D, Magner A, La Perle K, Zhang X, Piruat J, Lopez-Barneo J, Jhiang S, Kirschner L. (2017) Sdhd ablation promotes thyroid tumorigenesis by inducing a stem-like phenotype. Endocrine-related cancer.

Kirschner LS, Qamri Z, Kari S, Ashtekar A. Molecular and cellular endocrinology. (2016) Mouse models of thyroid cancer: A 2015 update.

Sonn KA, Kannan AS, Bellary SS, Yun C, Hashmi SZ, Nelson JT, Ghodasra JH, Nickoli MS, Parimi V, Ghosh A, Shawen N, Ashtekar A, Stock SR, Hsu EL, Hsu WK. Effect of recombinant human bone morphogenetic protein-2 on a novel lung cancer spine metastasis model in rodents. Journal of orthopaedic research. 2016; 34(7):1274-81.

Ghodasra JH, Nickoli MS, Hashmi SZ, Nelson JT, Mendoza M, Nicolas JD, Bellary SS, Sonn K, Ashtekar A, Park CJ, Babu J, Yun C, Ghosh A, Kannan A, Stock SR, Hsu WK, Hsu EL. Ovariectomy-Induced Osteoporosis Does Not Impact Fusion Rates in a Recombinant Human Bone Morphogenetic Protein-2-Dependent Rat Posterolateral Arthrodesis Model. Global spine journal. 2016; 6(1):60-8

Lee SS, Hsu EL, Mendoza M, Ghodasra J, Nickoli MS, Ashtekar A, Polavarapu M, Babu J, Riaz RM, Nicolas JD, Nelson D, Hashmi SZ, Kaltz SR, Earhart JS, Merk BR, McKee JS, Bairstow SF, Shah RN, Hsu WK, Stupp SI. Gel scaffolds of BMP-2-binding peptide

vii amphiphile nanofibers for spinal arthrodesis. Advanced healthcare materials. 2015; 4(1):131-141.

Hsu EL, Sonn K, Kannan A, Bellary S, Yun C, Hashmi S, Nelson J, Mendoza M, Nickoli M, Ghodasra J, Park C, Mitchell S, Ashtekar A, Ghosh A, Jain A, Stock SR, Hsu WK. A comparative evaluation of factors influencing osteoinductivity among scaffolds designed for bone regeneration. Tissue engineering. Part A. 2013; 19(15-16):1764-72.

Fields of Study

Major Field: Molecular Cellular and Developmental Biology

viii

Table of Contents

Abstract ...... iii Acknowledgments ...... v Vita ...... vii Table of Contents ...... ix List of Tables ...... xii List of Figures ...... xiii List of Abbreviations ...... xiv Chapter 1. Introduction ...... 1 Thyroid Gland ...... 1 Thyroid Cancer ...... 4 Cowden Syndrome...... 6 Carney complex ...... 7 Mouse models of thyroid cancer based on inherited tumor neoplasia syndromes...... 8 Chapter 2. Role of Succinate Dehydrogenase in Thyroid Cancer ...... 11 BACKGROUND ...... 11 Metabolism of cancer cells ...... 11 Consequences of aberrant TCA cycle...... 13 TCA cycle plays a role in maintaining the epigenomic landscape ...... 14 RESULTS ...... 19 Sdhd knockout causes thyroid hyperplasia in mice alone or in combination with Pten- knockout...... 19 SDHD deficiency does not affect tumor progression in mice when combined with knockout of R1a and R1a-Pten...... 24 SDHD deficiency leads to metabolic defects...... 26 SDHD depletion leads to increased migration in thyroid cancer cell lines ...... 30 SDHD depletion promotes stemness in thyrocytes...... 32 SDHD deficiency increases global DNA methylation in thyroid cells...... 35 α-ketoglutarate treatment reverses stem cell-like phenotype caused by mutant SDHD ...... 38 DISCUSSION ...... 42 METHODS ...... 46 Animal Strains, Husbandry, and Maintenance ...... 46 Ultrasonography ...... 46 Follicular area measurement ...... 46 Cell culture and reagents ...... 47 Primary thyrocytes isolation ...... 47 ALDEFLUOR assay ...... 48

Immunoblotting ...... 48 Mass Spectrometry ...... 49 Oxygen consumption rate ...... 49 Cell proliferation/growth ...... 50 Cell migration assay ...... 50 Immunohistochemistry ...... 50 Electron microscopy ...... 51 RNA and Real-Time PCR ...... 51 Genomic DNA isolation ...... 52 Methylation Analysis ...... 52 Detection of ROS ...... 52 Statistics ...... 52 Chapter 3. Role of SOD2 in thyroid cancer ...... 54 BACKGROUND ...... 54 Oxidative stress can play a role in cellular growth and tumor initiation ...... 54 Antioxidant mechanisms for ROS defense ...... 56 Clinical trials of antioxidants in cancer prevention ...... 56 Oxidative stress in thyroid cancer...... 57 RESULTS ...... 59 Effects of over- and under-expression of Sod2 in murine models of thyroid follicular neoplasia ...... 59 Effects of over- and under-expression of Sod2 in a locally invasive FTC model ...... 62 Effects of over- and under-expression of Sod2 in metastatic FTC model ...... 66 DISCUSSION ...... 71 MATERIAL AND METHODS ...... 74 Animal Studies...... 74 Ultrasound...... 74 Histology ...... 75 Statistics ...... 75 Chapter 4: Bioinformatic analysis of SDHx and SOD2 genes in thyroid carcinoma ...... 76 BACKGROUND ...... 76 SDHx mutation in human cancers...... 76 Occurrence of SDHB/D mutation in thyroid cancer ...... 77 Oxidative stress in human endocrine cancers ...... 78 RESULTS ...... 79 Clinical relevance of SDHx mutations in human thyroid cancers...... 79 Altered SOD2 levels in human endocrine cancers...... 86 Oxidative stress in mouse thyroid cancer models...... 88 DISCUSSION ...... 92 MATERIALS AND METHDOS ...... 94 Analysis of TCGA data for SDHx mutations ...... 94

Transcriptome analysis for oxidative stress genes...... 103 Statistics ...... 104 Chapter 5: Concluding Remarks ...... 105 Bibliography ...... 109

List of Tables

Table 1: Gender effect on thyroid volume in Pten-TpoKO mice...... 64 Table 2: Gender effect on thyroid volume in R1a-TpoKO mice...... 65 Table 3: Gender effect on thyroid volume in DRP-TpoKO mice...... 72 Table 4: Germline SHDx variants in TCGA reported by Ni et al...... 79 Table 5: SDHx vs BRAF differential gene expression ...... 84

xii

List of Figures

Figure 1.1: Thyroid gland and hormone regulation...... 2 Figure 1.2: Mouse models of thyroid cancer based on CS and CNC...... 9 Figure 2.1 : The Krebs (TCA) cycle ...... 13 Figure 2.2: The SDH complex ...... 15 Figure 2.3: Mechanism of tumorigenesis by SDHx mutations ...... 17 Figure 2.4: Analysis of thyroid function in Sdhd-null mice...... 20 Figure 2.5: Sdhd deletion leads to enhanced cellularity in vivo...... 21 Figure 2.6: Sdhd-KO enhances proliferation ...... 23 Figure 2.7: Effect of Sdhd-KO in the R1a and R1a-Pten deletion background...... 25 Figure 2.8: SDHD depletion leads to succinate accumulation in vitro...... 27 Figure 2.9: Metabolite levels in SDHD depleted cells...... 28 Figure 2.10 : SDHD-KD leads to lower respiration and decreased spare capacity...... 30 Figure 2.11: SDHD depletion does not affect proliferation but increases the migratory capacity of thyroid derived cell lines...... 32 Figure 2.12: SDHD depletion leads to increased stem-like response in vitro and in vivo. 34 Figure 2.13: SDHD depletion does not induced ROS in vitro...... 36 Figure 2.14: Sdhd deletion activates mTOR pathway in vivo...... 37 Figure 2.15: Hypoxia pathways is not induced in SDHD-kd cells ...... 38 Figure 2.16: DNA methylation analysis of SDHD depleted cells ...... 40 Figure 2.17: Succinate levels can cause gene expression changes via increased methylation...... 41 Figure 2.18: α-KG treatment reverses stem-like phenotype in vitro...... 42 Figure 3.1: Cellular antioxidants (adapted from Chainy et al, 2016) ...... 57 Figure 3.2: Sod2 overexpression increases tumor aggressiveness in AKT induced FA (benign) mouse model...... 62 Figure 3.3: Sod2 deficiency and Sod2 overexpression has mixed effects in a PKA induced non-aggressive FTC tumor model...... 67 Figure 3.4: Sod2 deficiency induces tumor growth in a dual PKA/Akt induced aggressive and metastatic FTC model...... 69 Figure 3.5: Gender effect in Sod2 under- and over-expression models ...... 71 Figure 4.1: Co-occurrence of mutations with SDHx in TCGA dataset ...... 82 Figure 4.2: Analysis of gene expression based on SHDx mutation status...... 83 Figure 4.3: Analysis of "Stem" genes in human PTC ...... 87 Figure 4.4: Oxidative stress pathways in human endocrine neoplasia ...... 89 Figure 4.5: Deregulation of oxidative stress genes in murine tumor progression models.93

xiii

List of Abbreviations

ACC...... Adenocrotical carcinoma

ATC...... Anaplastic thyroid carcinoma cAMP...... Cyclic-AMP

CAT...... Catalase

CNC...... Carney Complex

CREB...... c-AMP response element binding protein

CS...... Cowden Syndrome

DNA...... Deoxyribonucleic acid

NAD...... Nicotinamide adenine dinucleotide

FAD...... Flavin adenine dinucleotide

FA...... Follicular thyroid adenoma

FCCP...... Carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone

FTC...... Follicular thyroid carcinoma

GIST...... Gastrointestinal stromal tumor

GPx...... Glutathione peroxidases HPT...... Hypothalamus-pituitary-thyroid axis

α-KG...... α-ketoglutarate

SDHAF2...... SDH assembly factor 2

xiv

SDH...... Succinate dehydrogenase

SOD2...... Superoxide dismutase 2

MTC...... Medullary thyroid carcinoma

NMTC...... non-medullary thyroid cancer

MnSOD...... Manganese-dependent superoxide dismutase mTOR...... Mammalian target of rapamycin

NIS...... Sodium/iodide symporter

OS...... Oxidative stress

PCR...... Polymerase chain reaction

PGL...... Paraganglioma

PHEO...... Pheochromocytoma

PKA...... Protein kinase A

PTC...... Papillary thyroid carcinoma

PTEN...... Phosphatase and tensin homolog qPCR...... Quantitative PCR

R1a...... PKA regulatory subunit R1A

ROS...... Reactive oxygen species

RT-PCR...... Real time PCR

TCA...... Tricarboxylic acid

TCGA...... The Cancer Genome Atlas

TSH...... Thyroid stimulating hormone

TRH...... Thyrotropin releasing hormone

TPO...... Thyroid peroxidase

Tg...... Thyroglobulin

xvi

Chapter 1. Introduction

Thyroid Gland

The thyroid gland in mammals is located in the neck region, producing two hormonesthyroid hormones and calcitonin. It has two lobes which are connected by the isthmus.

The thyroid gland is composed of thyrocytes of epithelial origin, which form the thyroid follicles, and calcitonin-secreting C-cells which are neuroendocrine-derived cells in the interfollicular spaces. Thyroid follicles are roughly spherical, and store and release thyroid hormones (Figure 1.1 A). In the adult gland, follicle size can vary considerably (1). Follicular cell polarity is important, as hormone synthesis and endocytosis takes place at the apical membrane, while iodine and thyroid hormone storage occurs in the lumen of the follicle (1). As hormones are directly released into the blood stream, the thyroid gland is highly vascular. The parathyroid gland is usually located adjacent to the thyroid and is composed of parathyroid hormone secreting cells.

The hypothalamus-pituitary-thyroid (HPT) axis is a classical feedback system (Figure 1.1

B). Hypothalamic thyrotropin-releasing hormone (TRH) stimulates the pituitary gland to produce thyroid-stimulating hormone (TSH). TSH, in turn, stimulates the thyroid to produce thyroid hormones. The prohormone thyroxin (T4 or 3,5,3',5'

Figure 1.1: Thyroid gland and hormone regulation.

A. Structure of thyroid gland (top) and histology of normal thyroid follicles (bottom) is

shown, in which thyrocytes are arranged in spheres making thyroid follicles (represented

as F). B. HPT axis is neuroendocrine system responsible for the regulation of

metabolism involving hypothalamus, pituitary and thyroid. TSH stimulates the thyroid

to produce thyroid hormone (T3 and T4). These thyroid hormones negatively regulate

TRH and TSH production. Adapted from (2).

tetraiodothyronine) is released into the circulation. Thyroid hormones T3 and T4 are derived from the precursor protein thyroglobulin (Tg) by a combined action of iodine and thyroperoxidase enzyme, and transported from the thyrocyte cytoplasm into the bloodstream.

Although the thyroid gland makes some T3, the majority of T3 is derived from conversion of T4 to the active hormone in peripheral tissue by deiodinase activity. T3 in turn exerts a negative feedback control on the level of the hypothalamus and the pituitary (2, 3).

TSH directly stimulates the proliferation of thyrocytes while maintaining the expression of differentiation through activation of thyroid-specific genes (4). Primitive thyroid cells in human embryos express the transcription factors Hex, NKx2.1, Pax8 and Foxe1, while sodium/iodide symporter (NIS), Tg, TPO, and TSH are considered functional genes contributing to the hormone production and secretion (5). Thyroid development takes place within the first seven weeks of gestation in the human embryo. It is thought that cell turnover takes place once per every 5-10 years for human thyrocytes (1). Evolution of the follicular thyroid gland can be traced back to a non-mammalian species of jawless fishes agnathans (1).

Thyroid hormone (TH) has large and wide ranging effects on the majority of tissues, and influences key metabolic pathways that control energy balance. It’s action is exerted primarily via the thyroid hormone receptor (TR) family of nuclear receptors consisting of two major isoforms TR-α and -β. TH maintains basal metabolic rate, regulates energy storage, expenditure and controls metabolism through its action in the brain, fat cells, skeletal muscle, liver, and pancreas. Thyroid signaling defects can result in a broad range of conditions, including profound mental retardation, obesity and metabolic disorders (6). Hyperthyroidism (increased TH) is a condition which leads to increase in resting energy expenditure, weight loss, reduced cholesterol levels, increased lipolysis, and gluconeogenesis. Conversely, hypothyroidism (decreased TH) has opposite effects (7).

Thyroid Cancer

Thyroid cancer is the most common endocrine cancer with the fastest growing incidence rate, even when accounted for enhanced detection techniques (8-10). According to National

Cancer Institute’s (NCI) Surveillance, Epidemiology and End Results (SEER) program, there will be over 56,000 new cases of thyroid cancer in 2017. Thyroid cancer represents 3.4% of all new cancer cases in the U.S with a median age of diagnosis 51 years. Fortunately, patients generally have a favorable outcome. An exception to this is poorly differentiated thyroid cancer

(PDTC). Epithelial thyroid cancers, also called as non-medullary thyroid cancers (NMTC), are divided into three types based on histology: papillary (PTC), follicular (FTC), and anaplastic

(ATC). Of these subtypes, PTC and FTC are well-differentiated cancers whereas ATC is considered undifferentiated cancer. Amongst well-differentiated thyroid cancers, PTC is the most common subtype, accounting for over 80% of all thyroid cancer cases, while FTC makes up to about 10–15% cases (11). FTC is aggressive in nature, and has a tendency to metastasize to distant tissues by a hematogenous route with a poorer prognosis than the more common papillary type. ATC is a rare but aggressive subtype which accounts for 5-10% cases (12). ATC has the worst prognosis, with very few effective therapies and a median survival from time of diagnosis of 3 to 6 months (13).

Thyroid nodules are very common in the general population, with a majority being benign, and about 5-10 % of detected thyroid nodules are malignant (14). To diagnose malignancy of thyroid nodules, neck ultrasound and fine-needle aspiration (FNA) are performed.

The biopsies are categorized as non-deterministic, malignant, follicular lesions or benign according to the Bethesda criteria (15). Thyroid cancers range from stages I through IV which

4 are determined according to the AJCC (American Joint Committee on Cancer) TNM system. The treatment of thyroid cancer depends on type and stage, but usually includes partial or full thyroidectomy, followed by possible radioactive iodine (RAI) treatment and removal of areas of metastatic spread such as the lymph nodes (14).

One of the most important signaling pathways in PTC is the RAS/RAF/MEK pathway, and the most common genetic alteration is the BRAFV600 mutation which causes constitutive activation of the BRAF kinase (16). Activated tyrosine kinase (RTK) receptor activates MAPK pathway. Mutations in BRAF cause increased kinase activity and constitutive activation of the

ERK pathway. BRAFV600 mutation is found in about 40-60 % cases. In the same pathway, RAS oncogenes are thought to account for approximately 10% of PTC cases (12). Amongst the RAS genes, mutations in N-RAS (8.5%) are more commonly found than H-RAS (3.5%), whereas

KRAS mutations are rarest of the three. RAS mutations are also seen in 40-50% FTC cases, suggesting their involvement in follicular carcinogenesis.

RET-PTC gene fusions, many of which represent intrachromosomal rearrangements activating different donor genes are thought to be responsible for about 20% of PTC (12).

Rearranged during Transfection (RET) proto-oncogene is on chromosome 10q11.2 and encodes for a transmembrane tyrosine-kinase receptor involved in the control of cell differentiation and proliferation. This rearrangement constitutively activates the transcription of the RET tyrosinekinase domain in follicular cell, thus triggering the signaling along the MAPK pathway and an uncontrolled proliferation (17). Rearrangements of the RET proto-oncogene are found in

PTC and have been shown to play a pathogenic role. Up to now, 13 different types of RET/PTC rearrangements have been reported but the two most common are RET/PTC1 and RET/PTC3

(18).

The phosphatidylinositide 3-kinase (PI3K) pathway is involved in both PTC and FTC.

Phosphatase-and-Tensin homolog (PTEN) is a negative regulator of Akt. Inactivating mutations in PTEN are seen in ~1% of PTC, but occur more frequently in FTC (10-30%) (19). Genomic amplification and overexpression in PI3KCA (catalytic subunit of PI3K) is more prevalent than activating mutations associated with PTC (20).

PAX8-PPARγ translocations occur in 35% FTC cases resulting in a fusion protein (11).

The peroxisome proliferator-activated receptor (PPARγ) is tumor suppressor from hormone receptor family of nuclear receptors, whereas PAX8 belongs to the paired box family of transcription factors. PAX8 drives the expression of many thyroid-specific genes (TPO, TG) in mature thyrocytes. The PAX8/PPARG rearrangement is created by a translocation between chromosomal regions 2q13 and 3p25, resulting in a fusion transcript and protein.

Thyroid carcinomas can be sporadic or familial. The inherited tumor predisposition syndromes that include thyroid carcinoma as a part of their tumor spectrum are familial adenomatous polyposis, pendred syndrome, Werner syndrome, Cowden Syndrome (CS) and

Carney Complex (CNC) (21).

Cowden Syndrome

Cowden syndrome is a disorder characterized by multiple noncancerous hamartomas and an increased risk of developing breast, thyroid and endometrial cancers. It is a rare disorder inherited in autosomal dominant pattern affecting 1 in 200,000 people (22). CS patients are susceptible to colorectal cancer, kidney cancer and melanomas. Most CS patients have mutations

6 in PTEN (chromosome 10q23.31), which is also one of the most commonly mutated gene in sporadic cancers (23). CS has overlapping symptoms with another syndrome named

BannayanRiley-Ruvalcaba syndrome (together called PTEN hamartoma tumor syndrome as both conditions are associated with mutations in PTEN). CS patients have an early onset of cancers, and can have manifestations other than hamartomas such as macrocephaly, developmental defects and intellectual disabilities. The lifetime risk of differentiated thyroid cancer in CS patients is 3%-10% (24). Patients with features of CS, not meeting diagnostic criteria are termed

CS-like (CSL) (25). Other susceptibility genes for CS/CSL include promoter hypermethylation in gene KLLN (chromosome 10q23.31), and mutations in SDHD (chromosome 11q23.1) and

SDHB (chromosome 1p36.13) (25, 26) [reviewed in chapter 2].

Carney complex

Carney complex is a disorder characterized by an increased risk of myxomas

(noncancerous primary heart tumor, usually irregular in shape) and endocrine tumors such as cancers of adrenal gland, testes, ovaries, pituitary gland and thyroid gland. CNC is inherited in an autosomal dominant manner, and has overlapping features with McCune-Albright syndrome

(MAS) and multiple endocrine neoplasias (MENs). Approximately 750 individuals affected with

CNC have been reported to date. PRKAR1a is the cAMP-dependent protein kinase (protein kinase A, or PKA) type 1A regulatory subunit (27, 28). The gene encoding this subunit, located on the CNC1 locus was identified almost 2 decades ago. Inactivating heterozygous germline mutations in tumor suppressor gene PRKAR1a are found in roughly two thirds of CNC patients

(29, 30). Mutations in this gene are also associated with acrodysostosis, a few cases of pancreatic neoplasms and epithelial tumors (27, 31, 32). 7

Mouse models of thyroid cancer based on inherited tumor neoplasia syndromes.

Mouse models that confirm the genetic drivers of thyroid cancer can provide experimentally tractable systems. We can utilize such mouse models to delineate the function of susceptibility genes, as cancer is thought to be a multiple-hit process (33). Our lab has developed a tumor progression mouse model based on the inherited tumor predisposition syndormes CNC and CS.

In patients, CNC and CS are a mixture of FTC and PTC with overrepresentation of FTC.

However, CNC and CS based models develop FA or FTC in mice (34). Prkar1a heterozygous animals exhibit osteochondromyxomas, thyroid cancer and schwannomas (35). Pten- null conditional mouse models from our lab have resulted in follicular adenomas (34, 36). However, other mouse models with Pten mutations have resulted in variable tumor phenotypes of thyroid cancer, possibly owing to their different genetic backgrounds (36, 37). Transgenic mice in which the TPO promoter drives the expression of Cre recombinase (Tpo-Cre) specific to the thyroid gland starting at embryonic day 14.5 has been described previously (38). Our lab has utilized this model to drive thyroid specific deletion of genes of interest. Our lab has shown previously that of deletion of Pten in the mouse thyroid results mainly in follicular adenomas (with 70% incidence of adenoma), whereas deletion of Prkar1a (abbreviated as R1a) leads to locally invasive FTC without evidence of metastases (with 80% incidence of FC). Double Prkar1a-Pten mice (DRP-TpoKO or RP-TpoKO mice) develop aggressive FTC at 100% penetrance (Figure

1.2) and frequently develop well-differentiated lung metastases. The incidence of lung metastases in these mice has been shown to be about 50% (updated unpublished data) (34, 35).

Figure 1.2: Mouse models of thyroid cancer based on CS and CNC.

Thyroid cancer progression mouse model with thyroid-specific knockout

(Tpo-Cre) of Prkar1a (R1a) and Pten either alone or in combination. Gross

tumor and histology (10x) of thyroid tumors from Pten-TpoKO, R1aTpoKO,

DRP-TpoKO is shown.

Histologic analysis of Pten-TpoKO mice has shown activation of phospho (p)-Akt. In

R1a-TpoKO mice, p-CREB and p-mTOR pathways are involved in tumorigenesis. Lastly,

DRPTpoKO tumors demonstrate p-CREB, p-mTOR and patchy p-ERK staining, indicating activation of these pathways (34). In addition, R1a-TpoKO and DRP-TpoKO mice are

9 hyperthyroid, whereas Pten-TpoKO mice are euthyroid (normal thyroid function) (39). All Pten-

TpoKO and the majority of R1a-TpoKO mice are able to survive past 1 year. In some cases, R1a-

TpoKO animals require removal around 9 months due to dermatitis (unpublished data) and hyperthyroidism- related weight loss (39). In contrast, the majority of DRP-TpoKO mice require euthanasia due to the development of morbidity with labored breathing and/or weight loss

(median age 6 months) (34, 39). Thus, Pten-TpoKO, R1a-TpoKO and DRP-TpoKO models serve as tumor progression systems with experimentally tractable and well-characterized phenotypes.

We have utilized these tumor progression models to understand the role of additional thyroid cancer susceptibly genes, namely Sdhd and Sod2 by using genetic crosses with Pten-TpoKO,

R1a-TpoKO and DRP-TpoKO models. Hence, this study aims at providing in vivo confirmation of the genetic drivers of thyroid cancer as well as providing thyroid cancer models with multiple genetic hits to accelerate the tumor formation. Further, this study describes modulators of the pathways and provides insights that can be extended to other tissue types as well. The work described in Chapter 2 will help in understanding how thyroid tumor cells reprogram their metabolism and the pathways involved in it. In chapter 3, we will characterize the alterations in the oxidative stress landscape of thyroid cancer which has remained largely unknown, using mouse models. Finally in chapter 4, we will describe clinical relevance of oxidative stress and genetic alterations in Sdhd and Sod2 in the context of human thyroid cancers as well as mouse models of thyroid carcinoma.

Chapter 2. Role of Succinate Dehydrogenase in Thyroid Cancer

BACKGROUND

Mitochondria have traditionally been viewed as the powerhouse organelles in eukaryotic cells. Mitochondria play a fundamental role in cellular function, metabolism and apoptosis.

Metabolism is usually divided into two categories: catabolism (the breaking down of organic matter by cellular respiration) and anabolism (synthesize complex molecules such as proteins and nucleic acids). ATP is generated via TCA (tricarboxylic acid cycle, also known as citric acid or Krebs cycle) cycle and the electron transport chain (ETC). The TCA cycle is arguably the most important central metabolic pathway in living cells (Figure 2.1). It utilizes about 66% of total oxygen consumption, generating most of the total energy (40). It only operates under aerobic conditions, producing 12 ATP molecules per molecule of glucose. It constitutes the first stage in cellular respiration in which two-carbon units are oxidized, producing two molecules of

CO2, and high-energy electrons in the form of NADH and FADH2, and GTP (41).

Metabolism of cancer cells

It is known that most cancer cells exhibit increased glycolysis for ATP generation

(known as the Warburg effect). When oxidative phosphorylation is impaired, ATP generated by glycolysis becomes the major source of energy for cells. To meet the needs of proliferation, this metabolic switch alters pathways of energy metabolism and nutrient uptake in cancer cells.

Metabolic flexibility that allows tumor cells to grow under hypoxic conditions and altered energy metabolism are some of the “hallmarks of cancer” (42). Numerous discoveries about the

11 oncogenic role of mitochondrial enzymes have demonstrated that metabolic reprogramming in cancer can be a driver of carcinogenesis (43).

Figure 2.1 : The Krebs (TCA) cycle

TCA cycle is the process by which living cells break down organic molecules in

the presence of oxygen to harvest the energy. In the TCA cycle, SDH facilitates

the conversion of succinate to fumarate. This occurs in the inner mitochondrial

membrane by coupling the TCA cycle and ETC together. Adapted from (41)

In general, oncogene and tumor suppressor networks influence the metabolic shift in cancer. Evidence shows that enhanced PI3K/AKT signaling, mTOR, c-Myc and NF-kB can alter expression of genes that regulate nutrient transport, glucose metabolism and glutaminolysis (44,

45). Moreover, the role of the “oncometabolite” 2-hydroxyglutarate in AML and a subset of gliomas has provided direct evidence linking altered metabolism and cancer (46).

Consequences of aberrant TCA cycle

Aberrant TCA cycle function is implicated in a wide variety of pathological processes, ranging from cancer to neurodegenerative diseases to type II diabetes. Defects in genes involved with oxidative phosphorylation (OXPHOS) have been implicated in diseases such as neurodegeneration, diabetes, developmental delays, Leigh syndrome, language impairment and cardiomyopathy (47). Succinate dehydrogenase (SDH) was one of the first mitochondrial enzymes to implicate the tumor suppressive role of TCA cycle genes, supported by its association with familial neuroendocrine cancers (48). Later, mutations leading to dysfunction of other TCA cycle proteins such as SDHAF2, Fumarate hydratase (FH), and Isocitrate dehydrogenase (IDH) were found to be associated with multiple types of cancer formation (49).

Inherited or somatic mutations in any of the four subunits of SDH can lead to pheochromocytoma (PHEO), paraganglioma (PGL), renal cell carcinoma (RCC), gastrointestinal stromal tumor (GIST), thyroid cancer and breast cancer (50-54). Some other symptoms associated with SDH deficiencies vary from case to case but include growth retardation, pulmonary edema, bronchiolitis, body rigidity or optic atrophy (55). The spectrum of tumors caused by mutations in each of these genes differs, although there may be some overlap among the syndromes.

SDH is a component of the TCA cycle as it oxidizes succinate to fumarate and leads to electron transport to ubiquinone in the electron transport chain (ETC) as shown in Figure 2.2.

SDH is a heterotetrameric nuclear-encoded mitochondrial protein comprised of four subunits encoded by the four autosomal genes SDHA, SDHB, SDHC, and SDHD. Hydrophobic subunits

SDHC (15kDA) and SDHD (12 kDa) anchor the hydrophilic subunits SDHA (70 kDa) and

SDHB (27 kDa) and provide a binding site for ubiquinone. SDHA participates in substrate binding and oxidation while SDHC participates in electron transfer.

Recently, mutations in genes encoding the Succinate dehydrogenase (SDH) subunits were identified as additional susceptibility genes for CS, comprising about 10% of CS and CSL phenotypes (25, 56). Genetic variants in SDHB and SDHD were shown to confer a higher risk of breast, thyroid and other cancers in a subset of CS/CSL patients.

Complex II

SDHC SDHD

Q QH 2 Inner Mitochondrial Membrane SDHB

SDHA

FAD FADH2

Figure 2.2: The SDH complex

The SDH complex connects the TCA cycle to the electron transport chain.

The SDH complex is comprised of four subunits- SDHA, SDHB, SDHC and

SDHD. The latter two subunits are hydrophobic membrane anchor subunits.

TCA cycle plays a role in maintaining the epigenomic landscape

Dysfunction of metabolic enzymes has provided new insight into the observation that the majority, although not all cancers show a preference towards anaerobic glycolysis over oxidative phosphorylation. This metabolic flexibility enables cells to grow under hypoxic conditions (57).

This metabolic alteration was initially thought to be an adaptive mechanism to overcome the hypoxic conditions in tumors, but mounting evidence suggests that metabolic dysfunction is itself a driver of tumorigenesis (58).

Mechanisms that have been proposed to explain how loss of SDH leads to thyroid tumorigenesis include i) an increase in the production of reactive oxygen species (ROS), ii) activation of a hypoxia-like pathway under normoxic conditions (pseudo-hypoxia), and iii) genetic and epigenetics alterations due to the presence of oncometabolites (56, 59-61) (Figure

2.3). Loss of SDH may cause TCA cycle failure and prevents catalytic and electron transport functions, leading to diffusion of free radicals in the cell that cause nuclear DNA damage or induce an acute apoptotic response in cells (62, 63). ROS-induced genomic instability due to

SDHD mutations (conserved homologs Shh4 and Sdh4) was observed in yeast (62). Mutations in

SDHC led to oxidative stress in mouse and hamster fibroblasts (64, 65), whereas certain SDHD variants caused increased ROS in patient lymphoblast cell lines (25). SDH loss may also directly induce Hypoxia-inducible factor 1-alpha (HIF1α) in cells in order to adapt to reduced oxygen

(like) conditions, activate hypoxia genes, increase glycolysis and increase blood supply- all of which are important determinants of growth and survival of tumor cells (66, 67). Alternatively, high levels of ROS may generate a signal that links mitochondrial dysfunction to expression of hypoxia-inducible-genes. Evidence of the hypoxia pathways was provided by miR-210

Figure 2.3: Mechanism of tumorigenesis by SDHx mutations

At least three mechanisms have been proposed to explain the oncogenic role of SDH mutations. Loss of function mutations in SDHD leads to TCA cycle failure and dysfunctional ETC. It causes the accumulation of succinate and e- leakage. Complex II inhibition can also lead to the formation of sub-lethal levels of superoxide, which can have downstream effects such as inflammation, hypoxia, DNA damage and apoptosis.

Complex II of the ETC can act as a sensor for apoptosis induction. On the other hand, accumulation of succinate may cause inhibition of DNA and histone demethylases-

leading to increased methylation (68). Succinate may also inhibit prolyl dehydroxylases

which regulate hypoxia factor HIF1a, thereby inducing a pseudohypoxia response (50,

69).

expression, a key regulator of hypoxia in SDH-deficient PGL, PHEO and GIST (70). Loss of

SDHB caused ROS-dependent hypoxia in mammalian cell lines (71).

In thyroid cancer cell lines, SDHD mutations (G12S and H50R) lead to altered PTEN subcellular localization, resistance to apoptosis and a migratory phenotype (56). SDHD mutations destabilize complex II and show abnormal mitochondrial morphology in head and neck cancer (72). Nevertheless, the actual contribution of each of the proposed mechanisms in thyroid cancer has not been definitively established.

Several lines of genetic and biochemical evidence suggest metabolic rewiring in cancer has profound effects on regulating the gene expression, in particular epigenetic regulation (51,

68, 73). Metabolic rewiring could affect the availability of cofactors required for epigenetic modification enzymes or generate oncometabolites that act as agonists and/or antagonists for epigenetic modification enzymes. Citrate, succinate, fumarate and α-KG (also known as

2oxoglutaric acid or 2-OG) may serve as oncometabolites to promote tumorigenesis by inhibiting a family of epigenetic modulator enzymes (51, 74, 75). The major enzymes responsible for the demethylation and hydroxylation of DNA and histones include α-ketoglutarate (α-KG)dependent dioxygenases. These enzymes require oxygen, iron and α-KG for their activity (61). Succinate and fumarate accumulation as a result of TCA cycle impairment can inhibit the activity of α-KG-

17 dependent dioxygenase family such as TET family of 5-methylcytosine hydroxylases and histone demethylases such as Jumonji C-domain-containing (JMJD) enzymes. Both of these metabolites were shown to inhibit TET-catalyzed hydroxylation of 5mC and the activity of histone demethylases (76). In adddition, it has been proposed that succinate may also inhibit the pro- apotic activity of prolyl-hydroxylases (PHDs), thus setting the stage for neoplastic transformation (45). PHDs are oxygen and α-KG-dependent enzymes that regulate HIF1α activity. Genome-wide DNA methylation and expression profiling in SDHC mutated gastrointestinal stromal tumors have shown CpG island hypermethylation (77). SDHx (mutations in any of the four subunits), in particular SDHB-related paraganglioma metastatic tumors also display a hypermethylator phenotype (78). Hypermethylation phenomenon in these tumors has been implicated in the failure of DNA demethylation maintenance (73, 79).

To evaluate the contribution of Sdhd loss on thyroid tumorigenesis, we generated tissuespecific knockout of Sdhd in the mouse thyroid gland. We chose to study the gene encoding SDH subunit D (Sdhd) as mutations in this gene were observed more frequently than the other subunits in thyroid carcinoma (Reviewed in Chapter 4: Table 1: Germline SHDx variants in TCGA reported by Ni et al. (25). These in vivo studies were complemented by in vitro analyses of human thyroid cancer cells with knockdown of SDHD. Together, these studies reveal the ability of SDHD/Sdhd loss in promoting thyroid neoplasia and provide new mechanistic insights for these observations.

RESULTS

Sdhd knockout causes thyroid hyperplasia in mice alone or in combination with Pten- knockout.

Figure 2.4: Analysis of thyroid function in Sdhd-null mice.

A. Gene expression of sodium iodine symporter (NIS) levels in mouse thyroids by RT-PCR in WT (n=5), Pten-TpoKO (n=3) and SP-TpoKO (n=4).

B. Analysis of thyroid stimulating hormone (TSH) levels in ng/mL in serum of WT (n=13), Sdhd-TpoKO (n=10), Pten-TpoKO (n=12) and SP-TpoKO

(n=11). Graph shows average TSH levels and error bars represents standard deviation (SD). Statistical analyses were performed by two-tailed student’s ttest (ns= non significant).

Figure 2.5: Sdhd deletion leads to enhanced cellularity in vivo.

A. Thyroid volumes of WT (n=12) and Sdhd-TpoKO (n=16) mice analyzed by 3D ultrasonography at 6 months age. B. Thyroid volumes of Pten-TpoKO (n=15) and

SPTpoKO (n=26) mice analyzed by 3D ultrasonography at 6 months of age. C.

Incidence of FA and FC in Sdhd-TpoKO and SP-TpoKO mice compared to controls at

1 year age. D. Representative images of follicles in WT, Sdhd-TpoKO, Pten-TpoKO and SPTpoKO mice (200x magnification) at 1 year age. Quantification of the average

follicular size of WT thyroids compared to Sdhd-TpoKO, and Pten-TpoKO thyroids

compared to SP-TpoKO is on right. E. Representative TEM images are of

mitochondria comparing WT and Sdhd-null thyroids. Qualitative scores of

mitochondrial appearance in on right (n=3 each). Error bars represents standard

deviation (SD). Statistical analyses were performed by two-tailed student’s t-test (*P

value ≤ 0.05, **P value ≤ 0.01, ns=non significant).

To gain insight into the role of Sdhd in thyroid tumorigenesis, we crossed mice carrying a conditional Sdhd null allele (SdhdloxP) to mice expressing Cre recombinase under the control of the thyroid peroxidase (Tpo) promoter in order to generate thyroid- specific Sdhd-KO mice

(denoted as Sdhd-TpoKO). Sdhd-TpoKO mice were viable and fertile with no excess mortality

(data not shown). They also demonstrated normal thyroid function, with normal levels of serum

TSH (WT: 71.8 ± 43.36 ng/ml vs. Sdhd-TpoKO: 57 ± 83.85 ng/ml, p value = 0.7) and no change in the expression of the sodium-iodine symporters (NIS) mRNA compared to controls (Figure

2.4 A, B). Sdhd deletion resulted in a modest increase in the thyroid volumes of Sdhd-TpoKO mice at 6 months compared to controls (Figure 2.5 A), which was sustained for up to 12 months

(p value for the longitudinal analysis = 0.0075). Histopathologic analysis in a subset of mice revealed an incidence of follicular adenoma as 18% (2/11) in Sdhd-TpoKO compared to 0% (0/7) in controls (Figure 2.5 C).

However, the Sdhd-TpoKO thyroids were more cellular, exhibiting both a decrease in follicular area and a 2-fold enhanced rate of proliferation as measured by Ki-67 staining (Figure

2.6). Inflammatory component as well as apoptosis rate examined by immunohistochemistry staining of immune cell markers and cleaved caspase, respectively were unchanged in the

SdhdTpoKO compared to littermate controls (data not shown). Ultrastructure analysis of the thyroids revealed severe degeneration of the mitochondria in Sdhd-TpoKO mice, similar to previous reports (80), suggesting altered bioenergetics (Figure 2.5 E).

Figure 2.6: Sdhd-KO enhances proliferation

Ki-67 staining by immunohistochemistry in formalin fixed thyroid sections of WT,

Sdhd-TpoKO, Pten-TpoKO and SP-TpoKO mice at 1 year age (100x magnification). Multiple areas of stained slides were assessed and average Ki-67 score was calculated. Quantification of

Ki-67 positive nuclei is shown on right. Statistical analyses were performed by two-tailed student’s t-test (*P value ≤ 0.05, **P value ≤ 0.01, ns=nonsignificant).

Similarly, to test whether Sdhd deletion enhances tumorigenesis of the thyroid in cooperation with loss of Pten, we generated Sdhd Pten-TpoKO double knockout mice (denoted as SP-TpoKO) and compared them to mice lacking Pten in their thyroids. We have previously shown that tissue-specific ablation of Pten alone generates mice with enlarged thyroids and follicular adenomas without cancer (34). At 1 year age, mice with both Sdhd and Pten deletion in the thyroid showed no significant difference from Pten- only deletion in terms of thyroid size or the absence of follicular carcinoma (Figure 2.5 B-C). At the cellular level, SP-TpoKO thyroids also showed markedly enhanced proliferation and a reduced follicle size, leading to overall enhanced cellularity in the thyroid, although gross size was not altered (Figure 2.5 D). However, when aged up to 18 months, SP-TpoKO mice demonstrated 81% penetrance of follicular carcinomas, whereas Pten-TpoKO did not undergo malignant transformation (data not shown).

Lung metastases (2 out of 6) were also observed in SP-TpoKO mice (data not shown). Pten-

TpoKO mice were euthyroid (mean TSH 39.2 ± 67.98 ng/mL), and SP-TpoKO mice also exhibited no change in thyroid function (mean TSH 42.2 ± 23.31 ng/mL) (36). Together, these data suggest that Sdhd deletion causes thyroid hyperproliferation as an early sign of tumor initiation, and when combined with Pten-KO, has the potential to progress to advanced disease over time.

SDHD deficiency does not affect tumor progression in mice when combined with knockout of R1a and R1a-Pten.

Figure 2.7: Effect of Sdhd-KO in the R1a and R1a-Pten deletion background.

A. Thyroid volumes of R1a (n=15) and Sdhd R1a-TpoKO (n=14) mice analyzed by 3D

ultrasonography at over 1 year. B. Incidence of FA and FC in Sdhd-null mice in R1a

deletion background. C. Thyroid volumes of DRP-TpoKO (n=25) and SRP-TpoKO

(n=21) mice analyzed by 3D ultrasonography up to 9 months age. D. Kaplan-Meier

curve of DRP-TpoKO (n=66) and SRP-TpoKO (n=30) mice. Statistical analyses were

performed by two-tailed student’s t-test in B; t-test and linear mixed effects models for A

and C; and by log-rank test for survival analysis in D (*P value ≤ 0.05, **P value ≤

0.01, ns=non-significant).

To test whether Sdhd deletion enhances tumorigenesis of the thyroid in cooperation with loss of R1a, we generated Sdhd R1a-TpoKO double knockout mice (denoted as SR-TpoKO) and compared them to mice lacking R1a in their thyroids. We analyzed their thyroid volume by 3D ultrasound every 3 months up to 1 year. A substantial number of mice in this background developed dermatitis around 9 months of age, perhaps due to strain effects, interbreeding, hyperthyroidism or a combination of these factors. Our analysis demonstrated that tumor volume of SR mice was overall similar to R1a mice (Figure 2.7 A). Interestingly, incidence of carcinoma in SR-TpoKO mice was significantly lower than in R1a mice, but it was not controlled for age

(due to weight loss, low body score, dermatitis- related euthanasia according to early removal criteria) (Figure 2.7 B). Moreover, there was no evidence of distant metastases in both these groups (data not shown). Thus, we attribute the difference in incidence of carcinoma but not tumor size to strain effects.

In addition, to test whether Sdhd deletion affects FTC metastasis rate in DRP mice, we generated Sdhd R1a Pten-TpoKO triple knockout mice (denoted as SRP- TpoKO) and compared them to mice lacking both R1a and Pten in their thyroids. As shown in Figure 2.7 C-D, thyroid volume of DRP and SRP mice was similar with similar median survival. The 12 month time point was omitted due to morbidity in these mice past 9 months. Overall, Sdhd did not seem to increase tumorigenesis or metastasis ability in R1a and DRP setting.

SDHD deficiency leads to metabolic defects.

Because mouse thyroid primary cells are difficult to manipulate in primary culture, we selected human thyroid cell lines and used lentiviral shRNAs to generate stable SDHD knockdowns using Pten-null FTC133 follicular thyroid cancer cell line and the non-malignant

Figure 2.8: SDHD depletion leads to succinate accumulation in vitro.

A. Quantification of SDHD knockdown in FTC133 by RT-PCR. Gene expression is

normalized to beta-2-Microglobulin (B2M). B. Quantification of SDHD knockdown in

NthyOri 3.1 by RT-PCR. Gene expression is normalized to B2M. C. Immunoblot showing

SDHD knockdown in lentivirus transduced thyroid cell lines FTC133 and NthyOri 3.1. D.

Quantification of relative succinate levels in FTC133 cells analyzed by mass spectrometry. Data

represents average of four biological replicates. E. Quantification of relative succinate levels in

NthyOri 3.1 cells analyzed by mass spectrometry. Data represents average of four biological

replicates. Error bars represents standard deviation (SD). Statistical analyses were performed

by two-tailed student’s t-test (*P value ≤ 0.05, **P value ≤ 0.01, ns=nonsignificant).

NthyOri 3.1 cells as model systems. The efficiency of SDHD knockdown was tested by quantitative RT-PCR and verified by Western blot (Figure 2.8 A-C). Interestingly, SDHD- deficient cells showed normal expression levels of genes encoding the other SDH subunits (data not shown), indicating a lack of co-regulation of the genes encoding this multi subunit protein.

Figure 2.9: Metabolite levels in SDHD depleted cells.

A. Mass spectrometry analysis of TCA cycle metabolites in SDHD depleted FTC133

cells. B. Mass spectrometry analysis of TCA cycle metabolites in SDHD depleted

NthyOri 3.1 cells. C. Relative levels of ATP in FTC133 cells as analyzed by mass

spectrometry. D. Relative levels of ATP in NthyOri 3.1 cells as analyzed by mass

spectrometry. Graphs represent mean± SD of four biological replicates. Statistical

analyses were performed by two-tailed student’s t-test (*P value ≤ 0.05, **P value ≤ 27

0.01, ns=non-significant).

To assess the metabolic effects of SDHD knockdown, we performed quantitative mass spectrometry of small organic acids from the cells. This analysis demonstrated that as expected,

SDHD downregulation led to an accumulation of succinate in both cell lines following transfection (Figure 2.8 D-E). Changes in other metabolites of the TCA cycle were inconsistent in the two cell lines (Figure 2.9).

As SDHD is involved in metabolism through its role in TCA cycle and associated ETC, we investigated whether SDHD dysfunction results in defects in cellular respiration and metabolism by seahorse assay mitochondrial stress test. Oxygen consumption rate (OCR) is determined by sequential addition of modulators that perturb mitochondrial activity in cells.

Addition of oligomycin, an ATP synthase inhibitor, blocks oxidative phosphorylation

(OXPHOS). FCCP (Carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone) artificially increases OCR by uncoupling the respiratory chain and OXPHOS. Finally, rotenone is a complex I inhibitor that completely blocks mitochondrial respiration. Interestingly, SDHD knockdown did not completely abrogate the basal oxygen consumption rate (OCR) in either cell lines (Figure 2.10). However, SDHD knockdown cell lines responded differently to FCCP treatment. SDHD depletion thus caused a decrease in spare capacity and rendered cells more sensitive to energy stress (Figure 2.10 E, F). Moreover, SDHD deficient cells showed less

Figure 2.10: SDHD-KD leads to lower respiration and decreased spare capacity. A. Seahorse respiratory profile analysis of SDHD-deficient FTC133 cells. B. Seahorse

respiratory profile analysis of SDHD-deficient NthyOri 3.1 cells. C-F. SDHD

knockdown cells FTC133 (left) and NthyOri (right) show a corresponding increase in

non-mitochondrial respiration and decrease in spare capacity. Data represents one of

three representative experiments. Error bars represent standard deviation (SD).

Statistical analyses were performed by two-tailed student’s t-test (*P value ≤ 0.05, **P value

≤ 0.01, ns=non-significant).

29 sensitivity in response to oligomycin treatment, indicating a greater reliance on non- mitochondrial respiration. SDHD knockdown cells had increased non-mitochondrial respiration in comparison to control cells in both cell lines (Figure 2.10 C, D). Overall, knockdown of

SDHD decreased mitochondrial reserve respiratory capacity.

SDHD depletion leads to increased migration in thyroid cancer cell lines

To assess the effect of SDHD knockdown on cellular markers of cancer behaviors, we first measured cell proliferation. Over the course of 4 days of growth, there was no difference in cell numbers between control and SDHD-KD cells for either cell line (Figure 2.11 A, B). Next, we measured migratory ability using a 2-D scratch assay. In both cell lines, SDHD-KD cells demonstrated an increased migratory capacity compared to control cells (Figure 2.11 C, D). This migratory phenotype was also confirmed by Boyden chamber assay (Figure 2.11 E, F).

Figure 2.11: SDHD depletion does not affect proliferation but increases the migratory capacity of thyroid derived cell lines.

A,B) Proliferation of WT or SDHD-knockdown FTC133 and NthyOri 3.1 cells analyzed by crystal violet assay. C,D) Effect of SDHD-knockdown on FTC133 and

NthyOri 3.1 cell migration at 24 hours as measured by 2D scratch assay. E,F) Effect of SDHD-knockdown on FTC133 and NthyOri 3.1 cell migration at 24 hours as measured by a Boyden chamber assay. Each graph represents data from three

independent experiments. Error bars represents standard deviation (SD). Statistical

analyses were performed by two-tailed student’s t-test (*P value ≤ 0.05, **P value ≤

0.01, ns=non-significant).

SDHD depletion promotes stemness in thyrocytes.

Collectively, our data suggest that loss of SDHD leads to increased cellularity in vivo, and a migratory phenotype in vitro. SDHD dysfunction also lead to altered metabolic properties, mainly reduced oxidative phosphorylation in vitro. These characteristics are associated with a socalled ‘stemness phenotype’ (81-84). We further tested if SDHD dysfunction had any impact on cellular differentiation. We utilized our SDHD knockdown human thyroid cancer cell lines to test the expression of stem cell transcription factors. This analysis demonstrated that mRNA levels of transcription factors Nanog and Oct-4 were upregulated in SDHD knockdown NthyOri

3.1 cells relative to control cells (Figure 2.12 A). We next analyzed the cells for expression of intracellular aldehyde dehydrogenases (ALDH) activity, which has been proposed to play a key role in stem/progenitor cell expansion and differentiation as well as tumor initiation and progression (85). In agreement with our RT-PCR data, we observed a marked increase in the fraction of ALDH positive cells in SDHD depleted NthyOri 3.1 cells (Figure 2.12 B).

Unlike NthyOri 3.1 cells, FTC133 cells (which lack Pten) showed no ALDH activity nor increased “stem” gene expression as these cells are known to lack ALDH activity (86). With the identification of stem-like features in the cell lines, we wanted to see if the same observations were applicable to murine tumors in vivo, as acquisition of this stem-like phenotype would be a significant indicator of tumor initiation and neoplasia. To test this hypothesis, we isolated P and

Figure 2.12: SDHD-depletion leads to increased stem-like response in vitro and in vivo.

A. Quantitative gene expression analysis in NthyOri 3.1 cells. Gene expression is normalized to B2M. Graph represents one of three representative experiments. B.

ALDH activity in SDHD-depleted NthyOri 3.1 cells by ALDHFLUOR assay. The percentage of ALDHhigh cells from a single experiment is shown in each panel (left). On 33

the right, the mean of four independent experiments measuring % ALDH positive cells

gated according to DEAB treated negative controls is quantified. Error bars represents

standard deviation (SD). C. ALDH activity in primary mouse thyrocytes. The percentage

of ALDHhigh cells from a single experiment is shown in each panel (left) and the mean of

three independent experiment measuring % ALDH positive cells is on the right. Error bars

represents standard deviation (SD). Statistical analyses were performed by two-tailed

student’s t-test (*P value ≤ 0.05, **P value ≤ 0.01, ns=non-significant).

Figure 2.13: SDHD depletion does not induced ROS in vitro.

A. ROS levels in FTC133 cells as analyzed by flow cytometry using 1µM CMH2DCFA dye. B. ROS levels in NthyOri 3.1 cells as analyzed by flow cytometry. C. Graph represents mean percent FITC positive cells calculated from three independent experiments gated at 5% of untreated Sh-control cells. NAC (1mM) treatment was used as a negative control and H2O2 (1mM) treatment was used as a positive control. Gating

is according to 5% of negative control. X-axis represents florescent intensity; Y-axis

represents cell count. Statistical analyses were performed by two-tailed student’s t-test

(*P value ≤ 0.05, **P value ≤ 0.01, ns=non-significant).

SP thyrocytes for measurement of their ALDH activity. As shown in Figure 2.12 C, the population of ALDH positive cells in SP-null thyroid tumors was significantly elevated compared to those in Pten-null only mice. These data indicate that the ability of Sdhd ablation to promote tumorigenesis is significantly reliant on promotion of this stem-like phenotype.

SDHD deficiency increases global DNA methylation in thyroid cells.

Figure 2.14: Sdhd deletion activates mTOR pathway in vivo.

A. Immunohistochemistry of p-mTOR (phospho S2448), pAKT (Ser473) and p-ERK

(Thr202/Tyr204) staining (brown) in formalin fixed thyroid sections of WT, SdhdTpoKO,

Pten-TpoKO and SP-TpoKO mice at 1 year age (scale bar = 100 μ magnification). B.

Immunoblot of p-mTOR (phospho S2448) in SDHD-KD cell lines NthyOri 3.1 and

FTC133.

As described above, three mechanisms have been proposed to account for the tumorigenic effect of TCA cycle gene mutations: generation of free radicals, hypoxia/pseudohypoxia, and epigenetic alterations caused by metabolites excess. To assess free radical generation, we measured ROS in control and SDHD-deficient cells. While ROS levels could be manipulated by altering the oxidizing environment of the cells, no differences were detected between control and KD cells at baseline or under H2O2 stimulation (Figure 2.13 A-C).

We also examined the ERK, PI3K, and mTOR to evaluate activation of cancer pathways in SDHD-deficient cells. While p-AKT and p-ERK levels remained unchanged both in vivo and in vitro, we observed mildly increased levels of p-mTOR in Sdhd-TpoKO and SP- TpoKO mouse thyroids (Figure 2.14 A). However, the ratio of p-mTOR to total mTOR was unaltered in vitro (Figure 2.14 B). Hence, it is unclear whether PI3K/ERK signaling changes under

SDHDdeficient conditions. To produce a pseudohypoxic state, it has been suggested that succinate accumulation can inhibit prolyl hydroxylase (PHD) leading to a hypoxic response

(69). However, we did not observe induction of HIF1α or HIF2a proteins (Figure 2.15), suggesting that this pathway is not induced in these cells.

It has been shown that TCA cycle intermediates fumarate and succinate can inhibit histone demethylases of the Jumonji C class in a dose dependent manner (54). Also, SDHC mutations give rise to the hypermethylator phenotype in PGL (51). Thus, we tested whether

SDHD depletion directly affected DNA methylation in thyroid cell lines and mouse tumors. In

Figure 2.15: Hypoxia pathways is not induced in SDHD-kd cells

Immunoblot of HIF proteins in SDHD-KD cell lines cultured under normal oxygen and standard culture conditions. Positive control is cell lysate from CoCl2 (at 1mM) treated cells, while

GAPDH is used as loading control.

NthyOri 3.1 cells, there was an approximately 3-fold increase in 5-methylcytosine (5-mC) levels after depletion of SDHD (Figure 2.16 A). In contrast, DNA methylation levels in FTC133 cells seemed to be unaffected by SDHD loss. It is unclear if the lack of change is similar to (or caused by) the same defect that prevents ALDH expression, or if this is a limitation of the sensitivity of the assay. In mouse tumors, there was a clear trend towards an increase in 5-mC levels between

SP tumors and Pten-only tumors although these results did not reach significance, perhaps due to high variability observed between mice (Figure 2.16 C).

α-ketoglutarate treatment reverses stem cell-like phenotype caused by mutant SDHD

It is increasingly appreciated that DNA methylation has important role in cancer development and it helps maintain transcriptional silencing of some genes (87). Epigenetic modifications are reversible and can be therapeutically targeted. The phenotype caused by

SDHD loss can be possibly explained by altered αKG/succinate ratio, as we have reported increased succinate levels in our model. If an alteration of the relative amounts of TCA cycle intermediates is responsible for this effect, then reestablishing the balance should revert this effect (Figure 2.17). To test this possibility, we treated cells with cell permeable α-KG and measured the phenotypic characteristics as described above. Indeed, treatment with low dose of

α-KG for three days fully reversed the migratory phenotype as well as stem-like response evidenced by ALDH assay and reduced the ALDH levels back to that of control cells (Figure

2.18 A, C).

Figure 2.16: DNA methylation analysis of SDHD depleted cells

A. 5-mC levels in NthyOri 3.1 cells DNA analyzed by enzyme-linked immunosorbent

assay (ELISA). Graph represents mean ± SD of three biological replicates. B. 5-mC

levels in FTC133 cells DNA analyzed by ELISA. Graphs represents mean ± SD of

three biological replicates. C. 5-mC levels in DNA isolated from SP-TpoKO and

PtenTpoKO tumors analyzed by ELISA. Graphs represents mean ± SD of five mice.

Statistical analyses were performed by two-tailed student’s t-test (*P value ≤ 0.05, **P

value ≤ 0.01, ns=non-significant).

To examine whether unavailability of α-KG is also responsible for increased levels of

DNA methylation observed with SDHD deficiency, we measured 5-mC levels in NthyOri 3.1 cells treated with α-KG. Indeed, treatment with high dose of α-KG could rescue the hypermethylation phenotype in NthyOri cells (Figure 2.18 B).

Figure 2.17: Succinate levels can cause gene expression changes via increased methylation.

Inactivating mutations in SDH lead to imbalance in the αKG/Succinate ratio. Succinate can competitively inhibit α-KG dependent dioxygenase, leading to epigenetic changes. Succinate and fumarate behave as α-KG-competitive antagonists for inhibiting TETs and JMJDs.

Figure 2.18: α-KG treatment reverses stem-like phenotype in vitro.

A. SDHD knockdown NthyOri 3.1 cells with aKG (0.1 mM) treatment for 3 days were subjected to FACS analysis to measure ALDH activity. The percentage of ALDHhigh cells from a single experiment is shown in each panel. B. Quantification of three independent experiment measuring % ALDH positive cells treated with a-KG (blue bars, statistical comparison is to the untreated Sh-Ctrl). Gating is according to DEAB treated negative control is shown as C. DNA methylation levels of NthyOri 3.1 cells after treatment with 1mM α-KG for 3 days. Graphs represents mean ± SD of five

40 biological replicates. D. Migration in thyroid cell lines treated with aKG (0.1 mM) at

12 hours performed by wound healing assays. Graph represents data from three independent experiments. Error bars represents standard deviation (SD). Statistical analyses by unpaired t test for ALDH assay and migration, and by Mann-Whitney test in case of DNA methylation. (*P value ≤ 0.05, **P value ≤ 0.01, ns=non-significant).

DISCUSSION

Although mutations in SDH genes are known to cause pre-disposition to human cancers, analysis of the mechanism by which this occurs has been hampered by a lack of robust model systems. Previous efforts to study tumorigenesis in vivo did not identify an effect of mutant IDH,

FH, or SDH genes, including in tissues clearly affected in human patients (52, 80, 88), perhaps due to late onset of carcinogenesis. In this report, we address the role of SDHD in thyroid cancer by ablating Sdhd in the thyroid gland. Because it has been suggested that there may be an interaction between loss of the PTEN tumor suppressor and mutations in SDH genes, we also generated mice lacking both of these genes and compared them to Pten-KO thyroids, which only develop thyroid adenomas. Studying the role of tumor suppressors and oncogenes in the thyroid provides an advantageous model system to study pre-neoplastic and neoplastic changes because of the unique architecture of the thyroid follicle, as well as the opportunity to study the differentiated function of thyroid hormone secretion.

In contrast to the prior studies looking at the role of SDHD in neural crest-derived tissue

(including the adrenal medulla), we observed that Sdhd null thyroids show enlarged glands, a potential indicator of early neoplasia. Examination of the tissue revealed hypercellularity caused by enhanced proliferation, further supporting the notion that Sdhd ablation promotes tumor initiation in the thyroid gland. In fact, when combined with Pten KO, these mice developed FTC with metastasis at advanced age. Although Sdhd ablation from the thyroid causes enhanced proliferation in vivo, we did not observe this same effect in vitro. We attribute the differences either to the fact that in vitro studies were carried out in immortalized cell lines, which may

42 already have altered proliferation properties, or to the lack of a normal 3D tissue environment which exists in vivo.

One of the most striking observations from this study is that KO of Sdhd from the thyroid in vivo or knockdown of the gene in vitro promotes a shift to a stem-like phenotype, including the expression of both stem cell transcription factors and production of the thyroid stem cell marker aldehyde dehydrogenase (ALDH). Although the tumor cells retained sufficient differentiated function to maintain normal thyroid function, the acquisition of stem-like features is likely to play an important role in the ability of the hyperplastic glands to eventually progress to thyroid cancer.

Three mechanisms have been proposed to account of the mechanism by which mutations in SDH (or other TCA cycle genes) may account for the acquisition of a malignant phenotype. In the thyroid gland, we could find no evidence for an excess of reactive oxygen species nor for significant induction of a pseudohypoxia signature. A number of studies on TCA cycle tumor suppressors that have resulted in inconsistent reports on HIF and ROS involvement, casting doubts on pseudo-hypoxia and free-radical mechanisms (52, 69, 89, 90). Based on our observations, we focused on the role of succinate as an oncometabolite (abnormal accumulation of a metabolite causing both metabolic and nonmetabolic dysregulation and potential transformation to malignancy). Levels of succinate were elevated in SDHD-KD cells, although there were minimal effects on other TCA cycle intermediates. Succinate has been proposed to affect DNA demethylases, and, indeed, knockdown cells exhibited increased DNA methylation, a trend which was seen in vivo but which did not reach statistical significance due to high variability. The in vivo studies may reflect not only variability in the thyrocytes themselves, but

43 also contributions from stromal cells, which would be expected to have normal SDH function and thus reduce the magnitude of any alterations restricted to thyroid epithelial cells.

It has been proposed that the effect of oncometabolites is mediated by an imbalance altering the ratio of various allosteric enzyme modulators (46, 88, 91). In the case of succinate, it has been suggested that the effects are mediated by an imbalance in the ratio of succinate to its precursor, alpha-ketoglutarate (α-KG) (68). We tested this experimentally by treating cells with

α-KG and demonstrated a nearly complete reversal of phenotypes with this treatment. The fact that SDH-associated changes can be reversed by excess α-KG suggests that altered α-

KG/succinate ratio, rather than absolute levels likely contributes to tumorigenesis by enzymatic inhibition of α-KG-dependent dioxygenases. The differential response of SDH dysfunction in different cell systems may be attributed to the strength of epigenetic effects which often varies between cell-types (92). This observation may have clinical implications, although one would need to be cautious about manipulation of these metabolites, lest other intracellular processes were disturbed.

As expected, Sdhd-KO thyroids exhibited aberrant mitochondria, and studies in vitro demonstrated altered respiratory function, consistent with the metabolic remodeling. In the case of SDHD mutations, our data is consistent with the notion that the metabolic abnormalities can drive tumorigenesis, rather than occur as a secondary effect. The thyroid gland is a unique environment, in that it is a tissue that is highly reliant on intracellular oxidation (through the action of thyroid peroxidase) for its biological function of generating thyroid hormone. This fact may render thyrocytes either susceptible or more resistant to changes that affect reactive oxygen species. In our hands, we did not observe a noticeable effect on reactive oxygen species.

In summary, Sdhd knockout in the thyrocytes is sufficient to cause excess cell growth in mice, which may play a role in tumor initiation. In addition, the present results unveil a role for an aberrant TCA cycle in the generation of a stem-like phenotype in vivo and in vitro. Together, our data tie metabolic dysfunction with tumorigenic response, and warrant further studies for indepth analysis of the regulation of gene expression due to epigenetic changes in each model.

These data also suggest that therapeutic reversal of DNA methylation may arise as an attractive approach to add to existing treatments of SDHx mutated tumors. The identification of a hypermethylator phenotype, albeit observed in single non-tumor derived NthyOri cell line helps in explaining both the tumor-suppressive role of SDH and the context-specific phenotypic characteristics. It warrants further studies of the impact of these mutations on specific genomic regions which will help us identify signaling pathways that play a role in the SDH related oncogenesis.

METHODS

Animal Strains, Husbandry, and Maintenance

The use of animals was in compliance with federal and Ohio State University Laboratory Animal

Resources regulations. SdhdloxP/loxP mice were provided by Jose Piruat and Jose López-Barneo

(52). SdhdloxP/loxP (C57BL/6 background) were crossed with Thyroid Peroxidase (TPO)–cre (38) to generate Sdhd-TpoKO. In addition, PtenloxP/loxP mice (Pten-TpoKO), R1aloxP/loxP mice

(R1aTpoKO) and double PtenloxP/loxP R1aloxP/loxP mice (DRP-TpoKO) were generated as described previously (39). Tpo-cre, Sdhd loxP/loxP were mated with these mice to generate SP-TpoKO,

SRTpoKO, and SRP-TpoKO mice. The experiments were performed using littermate mice from a mixed C57BL/6 and FVB genetic background.

Ultrasonography

All mice were imaged with a VisualSonics Vevo 2100 (VisualSonics, Toronto, CA) every 3 months up to 1 year age. A MS550D transducer (22–55 MHz) was used with 3D-Mode imaging to determine thyroid size. The volume of both thyroid lobes was calculated using Vevolab 2.1 software in a blinded fashion.

Follicular area measurement

The H&E-stained sections were imaged at 20x magnification. The area of follicles was determined by measuring the luminal surface using ImageJ software as described previously

(36).

Cell culture and reagents

FTC133 (human follicular thyroid carcinoma) cells were maintained in DMEM media with 10%

FBS and penicillin/streptomycin. NthyOri 3.1 (human follicular epithelial thyroid cells) were maintained in RPMI media with 10% FBS and penicillin/streptomycin. Octyl-α-ketoglutarate was purchased from Cayman Chemicals (Ann Arbor, MI). Lentiviral based shRNA for SDHD knockdown was obtained from Sigma Aldrich (St. Louis, MO). Lentiviral supernatant was produced from phoenix cells transfected with Fugene (Promega, Madison WI), packaging mix

(Sigma, St. Louis, MO) and viral vector. Stable SDHD-knockdown thyroid cells were generated by puromycin selection after transducing with shRNA lentivirus. shRNA sequences were as follows:

Sh1-SDHD CCGGGCCGAGCTCTGTTGCTTCGAACTCGAGTTCGAAG

CAACAGAGCTCGGCTTTTTG

Sh2-SDHD CCGGTCTGCTTCCGGCTGCTTATTTCTCGAGAAATAAGC

AGCCGGAAGCAGATTTTTG

Primary thyrocytes isolation

Primary thyroid cells were isolated from mouse tumors by enzymatic dissociation using collagenase/hyaluronidase solution, dispase and dnase using protocols adapted from manufacturer (StemCell Technologies, Vancouver, Canada). Briefly, the thyroid was dissected and chopped using automatic tissue chopper and placed in collagenase/hyaluronidase solution for

3 hours at 37o C with agitation. The sample was centrifuged at 350g for 5 mins and the pellet resuspended in 0.25% trypsin-EDTA on ice for 1 hour. After centrifugation, the cell pellet was

47 resuspended in ice-cold HBSS with 2% serum. After centrifugation, cells were dissociated with pre-warmed Dispase and Dnase I, resuspended ice cold HBSS with 2% serum and filtered through 40 µ cell strainer. Cells were kept on ice until use.

ALDEFLUOR assay

The ALDEFLUOR kit (StemCell Technologies, Vancouver, Canada) was used to isolate cell populations with a high ALDH activity. Cells were obtained from cultured cells or mouse thyroid tumors. Optimal conditions for the experiment were determined by titrating cell number and incubation time. Cultured cells and primary tumor cells were resuspended in ALDEFLUOR assay buffer containing ALDH substrate (BAAA, 5uL per 1×105 cells/mL) and incubated for 20 minutes at 37°C. An aliquot of each sample was treated with diethylaminobenzaldehyde (DEAB,

5uL per 1×105 cells/mL), a specific ALDH inhibitor as a negative control. Propidium iodide was used for cell viability. Cells were kept on ice after the ALDEFLUOR reaction was completed until analyzed on BD LSR II flow cytometer. A DEAB treated negative control sample was used as a gating control and the gate was set to include 0.5% ALDHbright cells on DEAB treatment for each sample.

Immunoblotting

Proteins were run on a SDS-PAGE gel by standard procedure, and the membranes were probed with the indicated antibodies: Millipore-SDHD (ABT110), p-mTOR (ab109268), mTOR (CST

2972), BD biosciences- HIF1a (610959), Novus- HIF2a (NB100), Cell Signaling Technology-

GAPDH (2118). All antibodies were used at 1:1000 dilution. Signals were detected with

SuperSignal™ West Femto Maximum Sensitivity Substrate (Thermo Fisher). 48

Mass Spectrometry

Organic acid and phosphorylated compounds were extracted from 1×106 cultured NthyOri 3.1 and FTC133 cells using boiling water and subjected to LC-MS/MS analysis using an AS-11 column (Dionex, 2.1 x 250mm). 13C-fumarate was used as internal standard. Relative succinate levels were calculated by normalizing the succinate levels to sum of all measured metabolites as described previously (93).

Oxygen consumption rate

FTC133 and NthyOri 3.1 cells were plated at a density of 20,000 cells per well (XF24 cell culture microplate; Seahorse Biosciences, North Billerica, MA). The cells were allowed to grow for 24 h, following which the cells were washed with XF Assay media (with 20mM glucose,

1mM sodium pyruvate, no sodium bicarbonate at pH 7.4). The cells were incubated for 1 h at 37

°C in a non-CO2 incubator. The assay was normalized to protein and analyzed using the XFe 2.3 software. Optimal seeding density and concentrations of oligomycin, rotenone and FCCP

(purchased from Sigma) were determined experimentally by plotting standard curves for each cell line. For mitochondrial stress testing, oxygen consumption rate (OCR) was measured with sequential addition of oligomycin, rotenone and FCCP. Spare reserve capacity was calculated as

FCCP induced-maximum OCR relative to baseline OCR, whereas non-mitochondrial respiration was calculated from the mitochondrial stress test based on residual respiration in response to rotenone.

Cell proliferation/growth

5000 cells were plated in each well of a 48-well plate for growth curve by crystal violet assay. At

each time point, triplicate wells were stained with 0.05% crystal violet, 0.1% formalin for 20

mins and extracted with 10% acetic acid. Absorbance was measured at 590 nm on Dynatech

MR7000 platform.

Cell migration assay

Cell migration was assessed by scratch wound healing assay. For the measurement of cell migration, control and SDHD-knockdown cells were cultured in individual wells of a 6- well plate. After reaching a confluent state, cell layers were scratched with a 200 uL plastic micropipette tip. The medium was aspirated away and replaced by 1-2 ml of fresh serum-free medium. Cells were allowed to migrate in serum free medium for 24 hours. Images were obtained at 0 and 24 hours by phase contrast microscopy. For evaluation of scratch closure, the horizontal distance of migrating cells from the initial wound was measured at 2 points along each scratch using ImageJ software.

Immunohistochemistry

Dissected mouse thyroid tissues were fixed in 10% neutral-buffered formalin solution. Tissues were processed, embedded in paraffin, cut in 5 μm sections, placed on positively charged slides, deparaffinized, rehydrated, and stained with hematoxylin and eosin (H&E). For immunohistochemistry, all sections were stained using a Bond Rx autostainer (Leica Biosystems,

Richmond, IL) with Ki67 (Abcam, Cambridge, MA) antibody. Briefly, slides were baked at

65°C for 15 minutes, and automated software performed dewaxing, rehydration, antigen retrieval, blocking, primary antibody incubation, post primary antibody incubation, detection

(DAB), and counterstaining using Bond reagents (Leica). Samples were then removed from the machine, dehydrated through ethanols and xylenes, mounted, and coverslipped. For quantification of the DAB staining, the images were reviewed and analyzed using ImmunoRatio application. For p-mTOR detection, p-mTOR antibody (ab51044) was used at 1/100 dilution.

Electron microscopy

Mouse thyroid tissue was fixed in 2.5 % gluteraldehyde, 0.1 M phosphate buffer at pH 7.4 for

TEM microscopy. The tissue sections were imaged on FEI Tecnai G2 Spirit TEM microscope.

RNA and Real-Time PCR

Total RNA from cells and tissue was isolated from Trizol using the Qiagen RNeasy kit. RNA quality was assessed by Nanodrop ND-1000 (Thermo Scientific, Waltham, MA). cDNA was prepared using an iScript cDNA Synthesis Kit (BioRad Laboratories, Hercules, CA) from 200 ng

RNA and subject to qRT-PCR using the iQ SYBR Green Supermix Kit (BioRad) as per manufacturer's instructions. Primer sets were validating by performing melt curves and standard curves. RT-PCR reactions were performed in triplicate. RT-PCR primers sequences were as follows.

Hs Oct-4 -F GACAGGGGGAGGGGAGGAGCTAGG

Hs Oct-4 -F CTTCCCTCCAACCAGTTGCCCCAA AC

Hs Nanog-F ATGCCTCACACGGAGACTGT

Hs Nanog-F AAGTGGGTTGTTTGCCTTTG 51

Hs-B2M-F TGCTGTCTCCATGTTTGATGTATCT

Hs-B2M-R TCTCTGCTCCCCACCTCTAAGT Hs-SDHD-F

ATACACTTGTCACCGAGCCAC

Hs-SDHD-R AAGGCCCCAGTGACCATGAAG

Genomic DNA isolation

Genomic DNA was isolated from control and SDHD-knockdown cells, and mouse thyroids using

DNeasy Blood & Tissue Kit (Qiagen, Germantown, MD) according to the manufacturer's protocol. The concentration and purity were determined by measuring the absorbance at 230,

260, and 280 nm using a Nanodrop ND-1000 (Thermo Scientific).

Methylation Analysis

Methylated DNA was quantified from 100 ng DNA using an ELISA for methylated DNA according to the manufacturer's directions (Abcam). Absolute and relative methyl-cytosine content was then calculated using the supplied formula.

Detection of ROS

o Cells were loaded with 1µM CMH2DCFA for 45 mins in dark at 37 C. After spinning down and washing with PBS twice, control tubes were treated with NAC or H202. Propidium iodide was used for cell viability. Cells were kept on ice until analyzed on BD LSR II flow cytometer.

Statistics

All data was analyzed via student's t-test, Mann Whitney test or ANOVA using GraphPad Prism software, p-values less than 0.05 were considered significant. Statistical analyses were 52 performed by two-tailed student’s t-test (*P value ≤ 0.05, ** P value ≤ 0.01, ns= non-significant) for all experiments, except otherwise mentioned in each figure legend.

Chapter 3. Role of SOD2 in thyroid cancer

BACKGROUND

Oxidative stress can play a role in cellular growth and tumor initiation

Reactive oxygen species (ROS) are produced by organisms as a result of normal cellular metabolism. Redox balance is essential in maintaining a healthy cellular microenvironment and cells employ natural antioxidant enzymes to this end. ROS includes superoxide anions, hydroxyl radicals, nitric oxides, singlet oxygen molecules, hydroperoxyl radicals, peroxyl radicals, hydrogen peroxides, and hypochlorous acid. The mitogenic effects of ROS on cancer cells has been the focus of interest in recent studies

(94). Mitochondria are the chief source of ROS, mainly at complex I and III of the electron transport chain (ETC). Excess ROS can be mutagenic and cause downstream effects, whereas low levels of intracellular ROS can act as signaling molecules (95-101).

Oxidative stress (OS) due to the inactivation or dysregulation of antioxidant enzymes is thought to be involved in the pathophysiology of many types of cancers

(102). Accumulation of reactive oxygen species (ROS) can lead to OS and cause multiple types of damage to the cell. Studies have shown that ROS can play a cell- autonomous role in tumor initiation and survival by mediating signal transduction pathways of cell growth and cell death (103, 104) and can also function in a cell non- autonomous role by

Figure 3.1: Cellular antioxidants (adapted from Chainy et al, 2016)

Reactive oxygen species can cause oxidative damage to DNA, proteins, and membrane lipids. Superoxide dismutase family consists of extracellular SOD, cytosolic SOD, and mitochondrially located SOD and plays a major role in the formation of hydrogen peroxide. Glutathione peroxidases (GPxs), catalase

(CAT), and peroxiredoxins all play a role in the enzymatic catabolism of this

- ROS. SODs can convert O2 to O2 or H2O2. Catalase (CAT) in the mitochondria and glutathione peroxidase (GPx) in the cytoplasm subsequently convert the

H2O2 to H2O and O2.

altering the tumor microenvironment (105). ROS such as superoxides can be produced either in the cytoplasm or in mitochondria. Cells possess elaborate antioxidant defense

55 systems amongst which superoxide dismutases (SOD), glutathione peroxidases (GPx), and catalase (CAT) play an important role (Figure 3.1).

Antioxidant mechanisms for ROS defense

The manganese-dependent superoxide dismutase, MnSod, is encoded by the gene

SOD2, and it is the major antioxidant enzyme that scavenges free radicals in

- mitochondria. MnSod converts superoxide radicals (O2 ) to hydrogen peroxide (H2O2), which is subsequently detoxified to water by GPx or CAT. Whereas animals with knockouts of cytosolic Sod1 and extracellular Sod3 are viable with mild phenotypes

(106, 107), Sod2 seems to be essential for cell survival as homozygous knockout of this gene causes embryonic lethality in mice (108, 109). Sod1 deficiency causes liver cancer in mice, whereas Sod2 heterozygous mice form lymphomas and pituitary adenomas

(110). Extracellular superoxide dismutase SOD3 has anti-oxidative, anti-inflammatory, antiapoptotic, and growth promoting characteristics (107). While Sod3 deficient mice do not form tumors, the overexpression of this enzyme reduces tumor formation (110).

Clinical trials of antioxidants in cancer prevention

As ROS is thought to be involved in aging and cancer, dietary antioxidants were naturally proposed to aid cancer treatment. Dietary antioxidant uptake has been demonstrated to affect the course of some diseases and certain conditions as diabetes, asthma, hemodialysis, thalassemia, rheumatoid arthritis, systemic attack, menopause, schizophrenia, depression, and leukemia (111). A number of studies were done to test the efficacy of antioxidants such as carotene, selenium (a component of glutathione 56 peroxidase and thioredoxin reductase), vitamin E, NAC in reducing the incidence of various types of cancers in patients (111). Although not successful in reducing skin cancer incidence (basal cell or squamous cell skin cancer), selenium with combination vitamin E, B-carotene was shown to improve cancer outcome and mortality (110, 112).

However, these reports were inconsistent between different studies (113). Moreover, although antioxidant therapy showed promises in pre-clinical studies, a majority of them have been inefficient or even failed in clinical trials (114). Thus, antioxidant therapy for cancer treatment remains a very controversial issue.

Oxidative stress in thyroid cancer

The association of OS with thyroid cancer progression has been suggested by some studies, although the source of ROS has not been defined (115). Thyroid hormone regulates oxidative metabolism as well as synthesis and degradation of many antioxidant enzymes (102). The thyroid gland itself is also a unique endocrine organ that utilizes hydrogen peroxide for thyroid hormone synthesis (116). Low total antioxidant levels and an altered OS index has been found in the serum of thyroid cancer patients, with evidence of DNA damage and lipid peroxidation found in thyroid adenoma (FA), follicular thyroid cancer (FTC) and papillary thyroid cancer (PTC) (80, 102, 117).

However, the mechanistic link between OS and thyroid cancer remains poorly understood (115). The studies associating OS with thyroid disorders and cancer have relied on clinical retrospective studies, and the causal relationship between the two remains to be directly investigated (118). Because of this association, anti-oxidant

57 therapy has been proposed as an attractive approach to mitigate cancer growth and progression.

To address SOD2’s direct role in thyroid cancer, we analyzed the effect of its forced overexpression or reduction on the behavior of established murine models of thyroid cancer. The results from this study suggest that the role of ROS is complex and context dependent in thyroid cancer genesis and progression.

RESULTS

Effects of over- and under-expression of Sod2 in murine models of thyroid follicular neoplasia

Oxidative stress is proposed as an important factor in thyroid cancer; however, the expression of SOD2 has been suggested to correlate to the degree of differentiation

(119). Thus far, the effect of Sod2 manipulation has not been experimentally investigated in FTC mouse models. To this end, we genetically manipulated OS in our FTC mouse models as described in Chapter 1 by either over- or under-expressing the gene. Because

Sod2-/- mice are not viable, we used Sod2+/- mice to achieve a reduction in Sod2, as these animals have been shown to exhibit about 50% of the MnSod activity of the wild-type

(120). We used a transgenic line overexpressing Sod2 globally as a means to upregulate expression of the enzyme (121). Notably, Sod2+/- mice have significantly elevated measures of oxidative stress and increased incidence of spontaneous cancers (108, 122,

123), whereas Sod2-Tg mice have been shown to have enhanced mitochondrial oxidative capacity and decreased incidence of cancers such as pancreatic tumors (121, 124). Pten-,

R1a -, and DRP-TpoCre mice were bred with Sod2+/- and Sod2-Tg animals to generate thyroid-specific tumors with up- or down-regulation of Sod2 as a means to understand the effect of altered oxidative stress in FTC disease progression.

All of these mouse models described above were subjected to ultrasonography at three month intervals for up to 1 year to record thyroid growth over time. At study endpoint (reaching 1 year age or meeting early removal criteria), mice were subjected to

Figure 3.2: Sod2 overexpression increases tumor aggressiveness in AKT induced

FA (benign) mouse model.

A. 3D rendering of ultrasonographic images of Pten-TpoKO with Sod2-wt, Sod2+/- and

Sod2-Tg at 12 months (top). Average thyroid volumes determined by 3D ultrasonography at 3, 6, 9 and 12 month intervals of Pten-TpoKO mice with Sod2-wt

(black line, n= 12), Sod2+/- (red line, n= 22) and Sod2-Tg (green line, n= 18) mice is shown at the bottom. Graphs showing mean ± SD, n is indicated at the 9 month time point. B. The incidence of thyroid adenoma or carcinoma in Pten-TpoKO mice with

Sod2-wt, Sod2+/- and Sod2-Tg is shown. C. Representative 40x images of Ki67 staining 60 in mouse thyroid tumors at 12 months of age is shown at the top. Quantification of proliferation represented as the DAB to nuclear ratio is shown at the bottom. Thyroid volume analysis was done using linear mixed model among three different genotype groups (Sod2-wt, Sod2+/- and Sod2-tg) averaged across months and gender. Ki67 score was compared using unpaired t-test. Holm’s procedure was used to adjust for multiple comparisons. Graphs showing mean ± SD. *p ≤ 0.05, **P ≤ 0.01

histological examination by a veterinary pathologist to analyze incidence of adenoma or carcinoma. For baseline comparison, control groups with Sod2-wt (Pten-TpoKO,

R1aTpoKO or DRP-TpoKO) were used.

All Pten-TpoKO mice with Sod2- wild type (wt), deficiency and overexpression were viable with no excess mortality observed up to 1 year. However, ultrasound analysis revealed that Sod2 overexpression (Sod2-Tg) consistently increased the tumor volume, whereas Sod2 underexpression (Sod2+/-) did not affect gross tumor size of Pten-

TpoKO mice (Figure 3.2 A). There were no sex differences in the Sod2-wt; Pten-TpoKO group (Figure 3.5 A and Table 1). Interestingly, male mice had generally higher tumor volume than female mice in case of Sod2 over- and under-expression Pten-TpoKO groups (overall P value 0.053), especially in the Sod2-Tg mice (Figure 3.5 C, E and

Table 1). To better understand the role Sod2 plays in the reduction of tumor size in Pten- null mice, proliferation markers were examined in the thyroid tissue. The increased tumor burden in the Sod-Tg group was supported by increased proliferation in Sod2 overexpression tumors as evidenced by Ki-67 staining compared to controls (Figure 3.2

C). Proliferation in Sod2+/-; Pten-TpoKO mice was not altered. Consistent with increased tumor burden, Pten mice with Sod2 overexpression mice had a significantly higher incidence of carcinoma, compared to Sod2-wt or Sod2+/- mice (P-value 0.007 by χ2 analysis) (Figure 3.2 B). One Sod2-Tg; Pten-TpoKO mouse developed lung metastasis, whereas none of the other groups exhibited lung metastasis. Hence, higher expression of

Sod2 correlated with advanced cancer in Pten-KO background.

Table 1: Gender effect on thyroid volume in Pten-TpoKO mice.

PTEN : vol

Genotype Sex N Obs N Mean Std Dev Minimum Maximum

Pten -TpoKO F 36 36 46.93 34.70 6.97 163.47

M 19 19 45.28 21.97 8.61 82.06 Sod2(+/-); Pten -TpoKO F 40 40 35.82 18.41 11.27 98.32

M 45 45 52.20 27.10 10.30 147.78 Sod2-Tg; Pten -TpoKO F 36 36 49.16 34.52 6.67 200.99

M 43 43 66.23 42.91 15.20 240.37

Effects of over- and under-expression of Sod2 in a locally invasive FTC model

All R1a-TpoKO mice with Sod2- wt, Sod2 deficiency and Sod2 overexpression were viable. Ultrasonography revealed that thyroids from Sod2+/-; R1a-TpoKO mice had volumes comparable to R1a-TpoKO mice at all time points studied (Figure 3.3 A).

Overall, Sod2 deficient R1a mice had lower tumor volume compared to R1a-only, but this observation did not reach statistical significance. Similar to the benign (Pten-

TpoKO) group, R1a-null male mice had significantly higher thyroid volume than female

R1a mice with overall P value 0.02 (Figure 3.5 B, D, F, Table 2). Sod2 deficient mice 62 had significantly lower proliferation compared to R1a-TpoKO mice; whereas Sod2-Tg did not affect the proliferation (Figure 3.3 C). However, the incidence of cancer was similar in all groups with no evidence of lung metastases (Figure 3.3 B). The results revealed no significant difference in thyroid tumor volume among the groups averaged across time or within same gender.

Table 2: Gender effect on thyroid volume in R1a-TpoKO mice. Analysis Variable : vol

Genotype Sex month N Obs N Mean Std Dev Minimum Maximum

R1a-TpoKO F 3 7 7 20.94 4.53 13.21 26.16

6 6 6 41.66 11.90 29.41 62.21

9 6 6 69.35 13.94 52.06 85.60

12 1 1 63.32 . 63.32 63.32

M 3 11 11 24.88 7.91 13.92 36.23

6 9 9 58.80 26.03 15.11 107.78

9 5 5 51.06 21.06 26.00 73.10

12 4 4 66.70 29.17 39.37 98.81 Sod2(+/-); R1a -TpoKO F 3 12 12 18.28 6.25 5.10 27.24

6 10 10 32.57 14.59 11.58 58.36

9 9 9 42.76 20.69 10.44 71.71

12 6 6 57.75 24.69 15.22 89.04

M 3 7 7 33.22 39.41 7.38 121.37

6 6 6 38.10 17.15 17.44 60.51

9 5 5 63.93 29.14 24.79 102.79

12 1 1 81.99 . 81.99 81.99 Sod2-Tg; R1a -TpoKO F 3 22 22 22.50 8.09 8.94 43.04

6 18 18 38.83 13.81 18.34 72.75

9 13 13 65.41 15.44 41.92 83.39

12 12 12 75.72 23.20 33.75 111.86

Analysis Variable : vol

Genotype Sex month N Obs N Mean Std Dev Minimum Maximum

M 3 11 11 24.40 6.28 17.88 36.28

6 10 10 43.93 13.73 31.99 79.00

9 7 7 59.76 6.94 47.68 67.21

12 4 4 87.47 5.64 83.51 95.83

Figure 3.3: Sod2 deficiency and Sod2 overexpression has mixed effects in a PKA induced non-aggressive FTC tumor model.

A. 3D rendering of ultrasonographic images of R1a-TpoKO with Sod2-wt, Sod2+/- and

Sod2-Tg at 12 months (top). Average thyroid volumes as determined by 3D ultrasonography at 3, 6, 9 and 12 month intervals of R1a -TpoKO mice with Sod2-wt

(black line, n= 11), Sod2+/- (red line, n= 14) and Sod2-Tg (green line, n= 19) is shown

(bottom). Graphs showing mean ± SD, n is indicated at 9 month time point. B. The incidence of thyroid carcinoma in R1a -TpoKO mice with Sod2-wt, Sod2+/- and Sod2Tg

65 is shown. C. Representative 40x images of Ki67 staining in thyroid sections of mice at

12 months of age is shown at the top. Quantification of proliferation represented as the

DAB to nuclear ratio is shown at bottom. Ki67 score was compared using unpaired t- test. Holm’s procedure was used to adjust for multiple comparisons to analyze tumor volumes. Graphs showing mean ± SD. *p ≤ 0.05, **P ≤ 0.01

Effects of over- and under-expression of Sod2 in metastatic FTC model

We next analyzed the effect of Sod2 overexpression and underexpression in a more aggressive tumor model. We have previously shown that DRP-TpoKO mice form aggressive thyroid tumors with capsular fibrosis and local invasion, and they have the ability to form distant metastases (34). In this model, Sod2 deficiency consistently promoted tumor growth and its overexpression was protective against tumor growth

(Figure 3.4 A). All three groups of mice developed carcinomas with 100% penetrance, but the gross tumor size in Sod2+/-, DRP-TpoKO mice was increased compared to controls (P value 0.06), and significantly higher than Sod2-Tg; DRP-TpoKO (P value

0.006). Student’s t-test at individual time points indicated significantly increased tumor volume in Sod2+/-, DRP-TpoKO at 3 and 6 months compared to the control group and no significant effect of Sod2-Tg group. A 12 month time point was omitted as the majority of mice were deceased past 9 months. The proportion of DRP-TpoKO mice surviving at

9 months was 26% in Sod2-wt group, 15% in Sod2 deficient group and 60% in Sod2 overexpression group. Overall, thyroid tumors of Sod2+/- mice had increased

66 proliferation compared control that did not reach statistical significance. The reduced tumor size in Sod2-Tg; DRP-TpoKO was however supported by significantly

Figure 3.4: Sod2 deficiency induces tumor growth in a dual PKA/Akt induced aggressive and metastatic FTC model.

A. 3D rendering of ultrasonographic images of thyroid glands in DRP-TpoKO with

Sod2-wt, Sod2+/- and Sod2-Tg mice at 12 months (top). Average thyroid volumes were determined by 3D ultrasonography at 3, 6, 9 and 12 month in DRP-TpoKO mice with

Sod2-wt (black line, n= 16), Sod2+/- (red line, n= 12) and Sod2-Tg (green line, n= 20) 67

(bottom). Graphs show mean ± SD, n is indicated at 6 month time point. B.

KaplanMeier survival curves of DRP-TpoKO mice with Sod2-wt (black line), Sod2 deficiency (red line), and Sod2 overexpression (green line). C. Representative 40x images of Ki67 staining at 12 months of age is shown at top. Quantification of proliferation represented as percent DAB to nuclear ratio is shown below. D. Incidence of lung metastases in DRP-TpoKO mice with Sod2-wt, Sod2+/- and Sod2-Tg is shown.

Ki67 score was compared using unpaired t-test. Holm’s procedure was used to adjust for multiple comparisons for tumor volume. Graphs showing mean ± SD. *p ≤ 0.05,

**P ≤ 0.01*.

decreased proliferation, as evidenced by Ki67 staining (Figure 3.4 C). Moreover,

Sod2Tg; DRP-TpoKO mice had significantly longer lifespans (median survival of 52 weeks compared to 23 weeks in controls) despite having similar cancer incidence, whereas Sod2+/-; DRP-TpoKO did not seem to decrease the lifespan of mice (Figure 3.4

C). We hypothesized that loss of Sod2 might correlate with advanced disease progression and metastasis, hence we screened lungs of these mice for metastatic spread. The frequency of lung metastases in all groups of DRP-TpoKO mice was found to be similar

(Figure 3.4 D). There was no gender effect in the DRP-TpoKO mice (Table 3).

Figure 3.5: Gender effect in Sod2 under- and over-expression models

Thyroid volume of Pten-TpoKO mice with Sod2-wt background (A), Sod2+/- background (C) and Sod2-Tg background (E) at 3, 6 9 and 12 months grouped by gender. Thyroid volume of R1a-TpoKO mice with Sod2-wt background (B), Sod2+/- background (D) and Sod2-Tg background (F) at 3, 6 9 and 12 months grouped by gender. Graphs showing mean ± SD.

Table 3: Gender effect on thyroid volume in DRP-TpoKO mice.

Analysis Variable : vol

Genotype Sex month N Obs N Mean Std Dev Minimum Maximum

RP-TpoKO F 3 15 15 52.41 29.16 12.64 124.15

6 11 11 72.17 46.62 35.44 199.90

9 4 4 122.52 70.85 30.77 186.72

M 3 15 15 70.95 44.09 16.58 160.55

6 7 7 93.22 68.47 36.93 243.87

9 4 4 109.88 12.05 92.75 120.79 Sod2(+/-); RP-TpoKO F 3 11 11 88.60 34.79 36.81 148.39

6 7 7 117.27 45.55 48.42 182.06

9 3 3 144.97 38.60 111.86 187.37

M 3 9 9 75.97 33.44 43.21 141.98

6 5 5 114.06 35.06 85.42 155.01 Sod2-Tg; RP -TpoKO F 3 17 17 51.46 30.21 24.70 149.60

6 13 13 56.38 16.59 32.37 88.57

9 10 10 93.17 29.06 57.97 146.63

M 3 8 8 63.33 31.05 20.09 127.49

6 7 7 77.03 59.45 20.34 189.24

9 5 5 84.90 58.79 36.37 181.46

DISCUSSION

Alterations in the cellular response to oxidative stress have been proposed to affect the behavior of thyroid disorders, including thyroid cancer. The thyroid has unique requirements for the handling of reactive oxygen species, as hormone generation requires specific peroxidases. However, the role of mitochondrial antioxidants in thyroid cancer progression has been unexplored. As mitochondria are the chief source of ROS, the mitochondrial superoxide dismutase SOD2 (MnSOD) is assumed to be the major antioxidant enzyme, and we have focused on this enzyme in this report.

The use of mouse models enables us to assess directly the effects of manipulating

Sod2 levels on tumor behavior. Sod2 overexpression in FA (Pten-null mice) led to a more aggressive cancer phenotype, which was accompanied by increased thyrocyte proliferation. In contrast, the same increase in Sod2 reduced cell proliferation in the metastatic FTC model (R1a Pten- null mice). Despite the fact that tumor size was not significantly decreased (although there was a trend towards smaller size) and there was no reduction in metastases, the mice had increased survival, pointing to a less aggressive cancer phenotype. When Sod2 levels were decreased in this model, tumors were larger with a trend towards enhanced proliferation, although there was no change in overall survival of the animals. Lastly, in mice with an intermediate phenotype of locally invasive but not metastatic FTC (R1a null mice), up- or down-regulation of Sod2 levels did not have a significant effect on tumor behavior.

Based on these observations, we conclude that the effects of SOD2 alteration seem to be context dependent and complex. This conclusion is supported by data in other 71 systems. Initially, Sod2 was thought to act as a tumor suppressor, as heterozygous ablation of this gene led to increased incidence of cancer in aged mice, whereas its overexpression led to suppression of tumor promotion in a stromal microenvironment and suppression of tumor formation in xenograft studies in several cancer types (125-

127). However, other studies have pointed to a dichotomous role Sod2 (128-130). For example, high Sod2 expression can lead to increased peroxidase activity with enhanced mutagenesis and tumorigenesis, as demonstrated in recent studies (131). We too observed that Sod2 acts in a pro-transformation function in Pten-null tumors. It has been shown that H202 accumulation leads to Pten inactivation resulting in angiogenic response

(132). Thus one possible explanation of increased tumor burden in Pten-KO mice due to

Sod2 overexpression could be H202 induced angiogenesis in tumors. Pten-TpoKO tumors behave like early stage benign tumors with increased Sod2/GPx1 and Sod2/Catalase ratios. When Sod2 is overexpressed, we hypothesize that it leads to peroxide accumulation which makes tumor more aggressive. In R1a mice, which develop nonmetastatic FTC, these effects already occur outside the stimulation by ROS, and thus there is no effect on the phenotype.

In our aggressive FTC mouse model, Sod2 deficiency correlated with poorer outcome and larger tumors; conversely, increasing Sod2 improved survival and reduced proliferation, confirming a protective effect of Sod2 in this model. A possible explanation for the discrepancy in the cancer outcome depending on the genetic model is tumorigenesis due to activation of metabolism regulating oncogenic pathways (34).

Reduced Sod2/Gpx1 and Sod2/catalase ratio DRP-TpoKO mice suggests inability to scavenge ROS. In this mouse model, the aggressive nature of the tumor combined with

Sod2 deficiency leads to tumor growth due to increased ROS, whereas efficient H2O2 scavenging by artificial Sod2 overexpression can decrease tumor cell proliferation and improve the overall outcome. Akt itself can be regulated by ROS, such as through Ras proteins as well as forkhead box O (Foxo) family of transcription factors (110, 133).

Hyperthyroidism in R1a- and DRP-TpoKO mice may be another confounding variable in how Sod2 modulates the tumor phenotype, as R1a- and DRP-TpoKO mice are hyperthyroid, whereas Pten-TpoKO mice are euthyroid (34, 39).

Experimental evidence from this study indicated that increased Sod2 levels in benign tumor types promote tumors progression, whereas Sod2 deficiency in aggressive and metastatic tumor models leads to increased tumor size. These different and contrasting outcomes could possibly be explained by excess free radical generation or excess peroxide levels- a hypothesis that requires validation. Overall, Sod2 has shown both tumor suppressive and promoting functions. Further studies of spatial-temporal changes in MnSod expression over the course of cancer progression in different cancer models and in vitro studies are required to better understand these redox-driven signaling events.

MATERIAL AND METHODS

Animal Studies

All animal experimentation described was conducted in accordance with accepted standards of humane animal care and in compliance with Ohio State University

Laboratory Animal Resources regulations. First, we generated targeted deletion of Pten and Prkar1a (R1a) alleles either separately or in combination (double R1a-Pten [DRP] knockouts) under control of the Thyroid peroxidase (TPO)-Cre in mice with FVB background as described previously (34). Next, we created Sod2 haploinsufficiency in these mice by crossing them with mice carrying a null allele of Sod2- (Jackson Labs stock

002973, C57Bl/6 background (134)). As a result, we generated three models with Sod2 haploinsufficency as i) Sod2+/-; Pten-TpoKO ii) Sod2+/-; R1a-TpoKO and iii) Sod2+/-;

DRP-TpoKO mice. To analyze the effect of overexpression of Sod2, we introduced

Sod2Tg allele in hemizygous state (C57Bl/6 background) in our established tumor progression model (121). As a result, we generated three mouse models with Sod2 overexpression as i) Sod2-Tg; Pten-TpoKO ii) Sod2-Tg; R1a-TpoKO and iii) Sod2-Tg;

DRP-TpoKO mice. As a result, all mice had FVB and C57Bl/6 mixed background and both sexes were included in the study.

Ultrasound

Three-dimensional thyroid ultrasonography was performed at 3 month intervals on live animals using a VisualSonics Vevo2100 ultrasound equipped with an MS550D

74 transducer. Images were acquired in 3D mode and volumes calculated using Vevolab 2.1 software in a blinded fashion.

Histology

Dissected mouse thyroid tissues were fixed in 10% buffered formalin solution and embedded in paraffin. Tissue sections were cut in 5 µm sections and stained with hematoxylin and eosin (H&E). Immunohistochemistry was performed and quantified as previously described with a Ki-67 antibody (Abcam ab15580) (135).

Statistics

All data, except Kaplan Meier curves and cancer incidence, were analyzed via paired or unpaired t-test using Prism GraphPad software. Thyroid volume analysis was done using linear mixed model among three different genotype groups (Sod2-wt, Sod2+/- and Sod2-tg) averaged across months and gender. Holm’s procedure was used to adjust for multiple comparisons. P values less than 0.05 were considered significant.

Chapter 4: Bioinformatic analysis of SDHx and SOD2 genes in thyroid carcinoma

BACKGROUND

SDHx mutation in human cancers

About two decades ago, the study of inherited predisposition to head and neck paragangliomas lead to the discovery of PGL/PHEO syndrome caused by succinate dehydrogenase (SDH) genes, namely SDHB, SDHC and SDHD through linkage analysis

(136). Germline mutations in SDHA genes are associated with Leigh syndrome, a rare but fatal neurodegenerative disease. Interestingly, SDHA mutations are associated with nonsyndromic PGL/PHEO with markedly different phenotypes than the rest, and some cases of pituitary adenomas (137). The incidence of cancers caused by germline mutations in SDH genes was found to be 80% in familial PGL/PHEO (136). The major

PGL and PHEO sites are carotid body and adrenal medulla, respectively (138). The hereditary syndrome, PGL1 (OMIM ID: 168000), is caused by mutations in the SDHD gene. Following studies found that mutations in the SDHC gene (PGL3), SDHB gene

(PGL4) and SDHAF2 gene (PGL2) can also cause familial PGLs (139). The tumor spectrum of germline SDH mutations has now been expanded to Carney-Stratakis syndrome, renal cell carcinoma, GIST and thyroid cancer (25, 53, 73, 140).

Occurrence of SDHB/D mutation in thyroid cancer

Ni et al. in 2012 reported additional susceptibility genes in CS/CSL patients as discussed in Chapter 1. The frequency of thyroid and breast tumors in CS/CSL patients with SDHx mutations (mutations in any of the four subunits) was 51.5% and 57.4% respectively, compared to PTEN-only mutations (breast cancer 32.4%, thyroid cancer

25.7%). In CS or CSL cases, renal cell carcinoma was present in 20% patients with SDH variants (25). This study reported germline SDHx variants in the TCGA dataset as well as CS cohort (25). The summary of SDHB and SDHD mutations in the TCGA thyroid carcinoma dataset is shown in Table 4.

Table 4: Germline SHDx variants in TCGA reported by Ni et al. Gene Variation N

SDHB (n=13) c.8C>G, p.Arg3Gly 1 c.158G>A, p.Gly53Glu 1 c.178A>G, p.Thr60Ala 1 c.425A>T, p.Asp142Val 1 c.487T>C, p.Ser163Pro 9 SDHD (n=15) c.34G>A, p.Gly12Ser 10 c.149A>G, p.His50Arg 5

Further, downregulation of SDH subunits was observed in both PTC and FTC and shown to correlate with poorer prognosis (141). Interestingly, CS/CSL patients with variants in SDHD exhibit an increased risk for PTC, while individuals with mutations in

PTEN have a predilection to both PTC and FTC (141). Based on epigenetic changes seen in our in vitro data as well as different clinical histopathologies of the resulting tumors, we hypothesized that SDHx mutations have a distinct molecular signature. The majority of papillary thyroid carcinomas are driven by mutations in either BRAF (specifically 77

BRAFV600E) or RAS. Therefore, we hypothesized that SDHx mutant samples in the

TCGA thyroid carcinoma dataset will show a distinct gene signature compared to BRAF mutated samples.

Oxidative stress in human endocrine cancers

The study by Wang et al (2011) supports the notion of oxidative stress in thyroid cancer patients by demonstrating a high total oxidant status and OS index (102). Further, increased OS confers patients with thyroid cancer an elevated risk for other conditions, such as cardiovascular diseases, degenerative neurological disorders and other cancers

(115). Studies have indicated that Mnsod deficiency is correlated with decreased complex II function leading to oxidative stress (142). Many studies have linked OS to thyroid cancer by showing its association with the deregulation of oxidant or antioxidant molecules.

In the case of thyroid cancer, limited data is available on SODs as a biomarker to predict tumor progression and how its direct manipulation affects patient outcome. As such, we sought to investigate SOD2’s role using human and mouse mRNA expression datasets.

RESULTS

Clinical relevance of SDHx mutations in human thyroid cancers.

We selected patient cohorts from TCGA thyroid cancer dataset based on SDHx mutant/variants status as reported in Table 1. First, we looked at the correlation of gene expression of various SDH subunits in thyroid cancer patients. The correlation coefficients of SDHD with other subunits- SDHA, SDHB and SDHC were 0.64, 0.67 and

0.673 respectively, indicating that there is a weak correlation. In contrast, our in vitro data indicates that shRNA knockdown of SDHD in human thyroid cancer cell lines does not affect gene expression of other SDH subunits (data not shown). This suggests that in human cancers, OXPHOS function may be reduced in response to either tumor biology, altered metabolism or other oncogenic drivers. The study by Ni et al. also reported increased ROS levels in patient-derived lymphoblast cultures with SDHB Ser163Pro,

SDHD Gly12Ser, or His50Arg variants. In contrast, SDHB Ala3Gly and SDHD

His145Asn had normal ROS levels (141). This may be indicative of a differential degree of pathogenicity of these variants.

We analyzed the co-occurrence of other mutations with SDHx mutations and found that they fall in expected normal frequencies (Figure 4.1 A, C, D). The number of mutations co-occurring with either SDHB or SDHD was similar, suggesting that both may have same degree of pathogenicity. This analysis revealed that the vast majority of

SDHB/D mutation co-occurred with BRAF mutation (64%). These data sets were used to perform principal component analysis (PCA) to identify trends in the data (Figure 4.1 B).

Figure 4.1: Co-occurrence of mutations with SDHx in TCGA dataset

A. Frequency of mutations co-occurring with SDHB and SDHD. B. PCA of gene expression comparing SDHx+BRAF mutations (blue symbols) to BRAF-only mutations (green symbols). C. Co-occurrence of mutations with SDHD in TCGA dataset (filtered at frequency below 2 mutations). D. Co-occurrence of mutations with SDHB in TCGA dataset (filtered at frequency below 2 mutations).

Figure 4.2: Analysis of gene expression based on SHDx mutation status.

A. Multi-Dimensional Scaling (MDS) plot shows clustering of SHDx+BRAF samples

(shown in blue) and BRAF-only samples (shown in green). B. Hierarchical clustering of the samples and genes. Distance matrix was calculated by Pearson’s correlation.

SHDx+BRAF samples are shown in blue and BRAF-only samples are shown in green.

It revealed lack of distinct clustering between BRAF-only and SDHx+BRAF mutant patients, suggesting that either all mutations may not be pathogenic or the effect of SDHx mutations may be more subtle. However, hierarchical clustering by Ward’s minimum

81 variance method of the same dataset revealed that SDHx mutant samples form somewhat distinct clusters than BRAF-only mutations, suggesting that there may be subtle signaling differences between SDHx-mutant/variant and BRAF-only group (Figure 4.2).

Analysis of differential gene expression was performed between BRAF + SDHx and

BRAF-only samples under relaxed conditions. A list of top hundred upregulated or downregulated genes is shown in Table 5.

Table 5: SDHx vs BRAF differential gene expression Ensembl_gene_id hgnc_symbol logFC logCPM PValue FDR ENSG00000198074 AKR1B10 -6.45943 1.480861 1.55E-93 3.27E-90 ENSG00000108849 PPY -5.92959 1.268029 1.24E-60 2.18E-58 1.49E- ENSG00000162510 MATN1 -5.00298 1.744906 1.76E-116 112 ENSG00000156076 WIF1 -4.92027 1.098234 1.45E-37 4.49E-36 ENSG00000075673 ATP12A -4.91336 1.457916 5.93E-81 4.18E-78 ENSG00000100362 PVALB -4.84839 1.141591 2.88E-96 6.94E-93 2.36E- ENSG00000224940 PRRT4 -4.75254 1.251069 1.40E-145 141 ENSG00000092758 COL9A3 -3.99065 6.537499 2.27E-66 6.08E-64 ENSG00000122585 NPY -3.8938 1.486849 4.72E-39 1.63E-37 ENSG00000100867 DHRS2 -3.71193 1.309517 6.57E-72 2.52E-69 ENSG00000151632 AKR1C2 -3.45613 2.557694 6.14E-69 2.03E-66 ENSG00000167244 IGF2 -3.35601 6.681319 2.01E-58 2.87E-56 ENSG00000257017 HP -3.24998 1.620811 1.44E-45 8.48E-44 ENSG00000229457 LINC01789 -3.24017 2.059429 6.98E-78 4.36E-75 3.58E- ENSG00000154269 ENPP3 -3.21855 1.135258 8.49E-108 104 ENSG00000118137 APOA1 -3.13061 1.582545 1.78E-70 6.28E-68 ENSG00000132874 SLC14A2 -3.08581 1.952182 2.48E-56 3.12E-54 ENSG00000104332 SFRP1 -3.05179 4.978698 6.03E-51 5.41E-49 ENSG00000123560 PLP1 -3.01802 1.60645 1.46E-44 8.02E-43 ENSG00000205358 MT1H -2.97051 4.531821 1.33E-39 4.90E-38 ENSG00000266968 -2.96442 1.132414 1.29E-49 1.06E-47 ENSG00000130294 KIF1A -2.95565 2.25174 1.48E-65 3.61E-63 ENSG00000175097 RAG2 -2.95342 1.511411 2.94E-26 3.66E-25

ENSG00000122756 CNTFR -2.94946 2.06041 5.12E-43 2.53E-41 ENSG00000125144 MT1G -2.88861 7.632451 1.73E-43 8.98E-42 ENSG00000257501 -2.84881 1.323347 1.31E-67 3.88E-65 ENSG00000164326 CARTPT -2.84147 3.357955 1.85E-18 1.22E-17 ENSG00000167749 KLK4 -2.81591 1.759448 4.85E-41 2.05E-39 ENSG00000174469 CNTNAP2 -2.62049 1.088478 2.96E-49 2.35E-47 ENSG00000259459 -2.60263 2.66495 6.04E-67 1.70E-64 ENSG00000143512 HHIPL2 -2.56986 1.068311 3.69E-39 1.30E-37 ENSG00000186510 CLCNKA -2.55438 2.114714 3.25E-55 3.84E-53 ENSG00000211452 DIO1 -2.45997 6.248035 1.53E-35 3.99E-34 ENSG00000159307 SCUBE1 -2.44746 2.694019 1.54E-72 6.34E-70 ENSG00000279905 -2.44683 2.22114 1.22E-37 3.80E-36 ENSG00000144908 ALDH1L1 -2.44016 1.865472 2.43E-59 3.82E-57 ENSG00000114200 BCHE -2.43852 1.204044 1.42E-43 7.42E-42 ENSG00000162552 WNT4 -2.43613 2.000197 2.83E-74 1.29E-71 ENSG00000126500 FLRT1 -2.38143 1.944454 5.27E-50 4.48E-48 ENSG00000224568 LINC01886 -2.36748 2.828003 6.21E-35 1.56E-33 ENSG00000124780 KCNK17 -2.36204 1.411563 4.37E-61 7.78E-59 ENSG00000006071 ABCC8 -2.3145 2.364146 2.23E-46 1.38E-44 ENSG00000185038 MROH2A -2.31228 1.294952 1.34E-18 8.89E-18 ENSG00000111249 CUX2 -2.30583 2.267693 2.38E-22 2.12E-21 ENSG00000010282 HHATL -2.23118 5.668501 2.12E-62 4.12E-60 ENSG00000134121 CHL1 -2.22153 1.846681 1.30E-34 3.17E-33 ENSG00000233705 SLC26A4-AS1 -2.2112 7.977586 7.35E-42 3.32E-40 ENSG00000198417 MT1F -2.14002 6.446411 1.30E-49 1.06E-47 ENSG00000178602 OTOS -2.12061 3.804729 1.44E-54 1.55E-52 ENSG00000165061 ZMAT4 -2.11118 2.625518 7.20E-22 6.17E-21 ENSG00000184156 KCNQ3 1.927732 4.344943 1.88E-50 1.63E-48 ENSG00000123500 COL10A1 1.966408 2.850761 6.77E-23 6.31E-22 ENSG00000041353 RAB27B 1.968877 2.323332 6.78E-26 8.10E-25 ENSG00000172061 LRRC15 1.988469 2.631212 1.21E-18 8.05E-18 ENSG00000067715 SYT1 1.989319 2.482928 1.87E-27 2.59E-26 ENSG00000101197 BIRC7 1.99148 2.56501 2.20E-19 1.56E-18 ENSG00000143217 NECTIN4 1.997099 4.246346 4.95E-48 3.59E-46 ENSG00000115602 IL1RL1 2.019132 4.956539 7.94E-21 6.25E-20 ENSG00000145864 GABRB2 2.02388 5.949763 4.44E-34 1.03E-32 ENSG00000148346 LCN2 2.032693 5.0394 3.55E-27 4.80E-26

ENSG00000105664 COMP 2.034854 5.024577 1.43E-25 1.65E-24 ENSG00000272482 2.058146 2.537854 1.39E-43 7.28E-42 ENSG00000115194 SLC30A3 2.065218 2.258218 8.18E-25 8.91E-24 ENSG00000169594 BNC1 2.096909 1.046892 4.41E-38 1.40E-36 ENSG00000075461 CACNG4 2.097938 2.072416 7.78E-28 1.12E-26 ENSG00000128564 VGF 2.099548 2.462538 9.33E-15 4.53E-14 ENSG00000167105 TMEM92 2.10066 1.698371 2.63E-24 2.76E-23 ENSG00000155897 ADCY8 2.101523 2.623475 1.73E-18 1.14E-17 ENSG00000169035 KLK7 2.106767 4.714689 3.09E-21 2.51E-20 ENSG00000117152 RGS4 2.107061 2.897888 6.62E-38 2.10E-36 ENSG00000124493 GRM4 2.128455 2.408703 6.21E-30 1.06E-28 ENSG00000129451 KLK10 2.167062 5.893514 3.23E-28 4.83E-27 ENSG00000117069 ST6GALNAC5 2.176375 3.222882 1.10E-26 1.42E-25 ENSG00000183317 EPHA10 2.17952 2.678772 9.21E-60 1.57E-57 ENSG00000175315 CST6 2.241798 6.36046 3.36E-40 1.31E-38 ENSG00000109255 NMU 2.24727 3.620827 1.13E-24 1.22E-23 ENSG00000223573 TINCR 2.253457 1.086799 1.95E-34 4.69E-33 ENSG00000087128 TMPRSS11E 2.260792 1.05982 1.68E-20 1.29E-19 ENSG00000054179 ENTPD2 2.263803 2.170517 3.17E-51 2.88E-49 ENSG00000135925 WNT10A 2.281238 2.513103 3.89E-41 1.67E-39 ENSG00000137648 TMPRSS4 2.2941 5.492223 3.88E-32 7.81E-31 ENSG00000173227 SYT12 2.321682 5.745333 1.58E-30 2.84E-29 ENSG00000184292 TACSTD2 2.327141 8.02339 7.74E-46 4.62E-44 ENSG00000115414 FN1 2.354998 13.14064 9.51E-42 4.24E-40 ENSG00000157765 SLC34A2 2.374301 10.73774 9.01E-40 3.38E-38 ENSG00000124467 PSG8 2.439854 1.101096 8.70E-16 4.57E-15 ENSG00000163817 SLC6A20 2.47537 2.739883 1.07E-28 1.68E-27 ENSG00000060718 COL11A1 2.482333 3.165733 3.32E-21 2.68E-20 ENSG00000166897 ELFN2 2.492639 1.683951 1.13E-29 1.89E-28 ENSG00000267206 LCN6 2.510693 2.453677 2.20E-37 6.67E-36 ENSG00000198535 C2CD4A 2.524347 2.430861 5.82E-39 1.98E-37 ENSG00000214797 2.529354 1.797018 8.41E-32 1.65E-30 ENSG00000134258 VTCN1 2.635443 2.141203 7.19E-27 9.44E-26 ENSG00000167755 KLK6 2.668757 1.75515 4.42E-24 4.57E-23 ENSG00000241644 INMT 2.70683 5.543978 1.00E-65 2.49E-63 ENSG00000237463 2.816544 3.171801 4.24E-46 2.60E-44 ENSG00000172020 GAP43 2.926746 2.842361 3.80E-38 1.23E-36

ENSG00000187045 TMPRSS6 3.116727 5.188111 4.31E-50 3.70E-48 ENSG00000164935 DCSTAMP 3.378859 6.291653 1.65E-49 1.33E-47 ENSG00000105492 SIGLEC6 3.71011 2.134192 5.58E-47 3.60E-45

Figure 4.3: Analysis of "Stem" genes in human PTC Gene expression in SDHx+BRAF mutant patient group is compared to BRAF-only mutant group. A. Normalized RSEM gene expression of SOX2, MYC and KLF4 genes

B. Normalized RSEM gene expression of OCT4 and CD44. Comparisons using unpaired t-test. Box plots showing upper and lower quartiles; median is indicated by the horizontal line. *p ≤ 0.05, ns= non-significant.

As in vitro and mouse data of SDHD-depletion pointed towards the acquisition of a stem-like phenotype, we next analyzed expression of “stem” genes in human tumors.

To delineate the effect of SHDB/D mutation on stem gene expression, we compared the

SDHx+BRAF mutant group to the BRAF-only group. This analysis revealed that three of four important Yamanaka factors (namely SOX2, c-MYC, KLF-4 and OCT3/4) were upregulated in SHDx mutant tumors (Figure 4.3 A) (143). In particular, mRNA levels of

SOX2, c-MYC and KLF-4 were upregulated in SHDx mutant tumors. OCT4 showed a similar trend that did not reach statistical significance; whereas CD44 expression was downregulated (Figure 4.3 B).

Altered SOD2 levels in human endocrine cancers.

SOD2 gene expression exhibits complex patterns with down-regulation in some cancers and up-regulation in others, but a comprehensive picture of thyroid carcinoma in regards to OS status and SOD2 expression is lacking (97, 144, 145). Given the importance of ROS signaling on thyroid function, we examined publically available and previously published datasets examining mRNA expression patterns in thyroid cancer

(146-149). PTC is the most common type of thyroid cancer and generally has a good prognosis. FTC is more likely to undergo hematogneous spread to distant organs than

PTC, whereas the aggressive ATC almost always develops metastases with a median survival of 3 to 5 months (150, 151). Analysis of oxidative stress genes in previously published human microarray data revealed that the oxidative pathway was altered in

FTC compared to FA (Figure 4.4 A) (152). Expression of SOD2 was downregulated in

FTC (fold change 0.7) compared to FA. Contrary to other cancers, analysis of PTC from

The Cancer Genome Atlas (TCGA) data revealed that SOD2 was neither significantly altered in primary tumor nor in metastatic tissue compared to normal and matched

86 primary tumor, respectively (Figure 4.4 B). However, this could be a limitation of the number of available sample with metastases (153, 154). In more aggressive anaplastic cancer,

Figure 4.4: Oxidative stress pathways in human endocrine neoplasia

Oxidative pathway and SOD2 expression were analyzed in different endocrine tumors.

A. Deregulation of oxidative stress genes in FTC compared to FA analyzed from microarray data. Significantly altered genes above cutoff (P value < 0.01) are indicated by red symbols and genes below cutoff are indicated by grey symbols; SOD2 is indicated by an arrow. B. Expression of SOD2 in PTC compared to normal tissue on left, and matched primary tissue vs metastatic tissue on right. C. Analysis of SOD2 expression in ATC compared to poorly differentiated thyroid carcinoma. D. Kaplan

Meier survival curves of ACC patients based on low (red line) and high (black line)

SOD2 expression at median cutoff. P-value was calculated using a log rank test in survival analysis. **P value < 0.01, ns = not significant.

expression of SOD2 was significantly increased compared to poorly differentiated thyroid cancer (PDTC) (147) (Figure 4.4 C).

To further test if SOD2 expression influences survival outcomes in aggressive endocrine cancers, we examined available endocrine cancer datasets. As limited survival data is available for publically available PTC and FTC datasets, we evaluated the data for adrenocortical carcinoma (ACC), another hormone secreting endocrine cancer which has poor overall outcome (149). We observed that patients with high SOD2 expression had increased survival; whereas low SOD2 levels poorly correlated with survival status in ACC (Figure 4.4 D).

Oxidative stress in mouse thyroid cancer models

Frequent MnSOD deregulation in cancers suggest that it may play an important role in cancer progression. According to some studies, SOD2 shows a positive correlation with aggressive cancer phenotypes (155-157). We therefore investigated the relationship between Sod2 expression and histopathological features using mouse models of FTC. Previous mouse modeling data from our lab has shown that the activation of AKT (arising from Pten loss) results in follicular adenomas, whereas activation of PKA signaling (arising from loss of R1a) is sufficient to promote a locally invasive FTC phenotype [reviewed in Chapter 1: Mouse models of thyroid cancer based

88 on inherited tumor neoplasia syndromes.] (34). When there is dual activation of these pathways (via combined loss of R1a and Pten), mice develop an aggressive FTC which exhibits metastatic behavior similar to that seen in human FTCs. Thus, we utilized this

FTC disease progression mouse model to analyze the oxidative stress in the slow growing tumors in Pten-TpoKO (70% benign tumors), intermediate follicular carcinoma

(80% carcinoma) in R1a-TpoKO, and 100% penetrant metastatic FTC in DRP-TpoKO

(Double R1a-Pten TpoKO) mice (Figure 1.2).

We analyzed the expression of genes in OS pathways in our FTC disease progression mouse models by mining previously published microarray data (34). As expected, the proportion of significantly altered genes in the oxidative stress pathway was higher in the more aggressive DRP-TpoKO model than R1a-TpoKO and Pten-

TpoKO (Figure 4.5 A-C). mRNA expression of Sod2 was significantly reduced in DRP-

TpoKO and R1a-TpoKO but not in Pten-TpoKO models (Figure 4.5 D) (34, 39). Thus, similar to the human FTC data discussed earlier, Sod2 expression was unchanged in adenomas, and downregulated in aggressive types of murine thyroid cancer models.

Further, the ratio of Sod2 and H2O2 scavenging enzyme levels is an important biomarker of tumor progression (158). Our microarray analysis revealed that PtenTpoKO mice had

Sod2/GPx1 ratio of 2.7 and Sod2/Catalase ratio of 1. R1a-TpoKO mice had Sod2/GPx1 ratio of 0.53 and Sod2/Catalase of 0.78. Similar to R1a mice, Sod2/GPx1 and

Sod2/Catalase ratio was 0.65 and 0.78 respectively in DRP-TpoKO tumors. Overall,

SOD2/Sod2 genes were similarly downregulated in human FTC and murine FTC model, but not in murine adenomas or human PTC. The OS gene signature as well as ratios of antioxidant enzymes indicate a progression of gene expression changes as the tumors

89 become more aggressive, thus setting the stage for OS stress in different thyroid carcinoma models.

Figure 4.5: Deregulation of oxidative stress genes in murine tumor progression models.

Microarray data comparing Pten-, R1a-, and DRP-TpoKO tumors. Oxidative pathways genes were analyzed in A. Pten-TpoKO tumors B. R1a-TpoKO tumors and C.

DRPTpoKO tumors. Significantly altered genes above cutoff (P value < 0.01) are indicated by red symbols and below cutoff are indicated by grey symbols; Sod2 indicated by arrow. WT Cre –negative littermates used for comparison for each model.

D. Expression of Sod2 in each tumor model compared to WT control. Dotted line represents fold change of Cre-negative control tissue for each group. **P value < 0.01.

DISCUSSION

The present chapter provides detailed information about clinical manifestations related to SDHx mutations and OS landscape, and serves to provide a starting point for further exploration of the possible relevance of each variant. A striking feature of this analysis is the prevalence of reported mutations in SDHB and SDHD compared to SDHA and SDHC. We performed bioinformatics analysis to identify clusters in TCGA patients cohorts based on their SHDx mutation status. Although we did not identify distinct clusters in PCA analysis (Figure 4.2), we found that SDHx variant samples cluster together by hierarchical clustering. This could be a limitation of a very small number of patients with SDHx mutations, or the functional relevance of the mutations themselves.

However consistent with our mouse and in vitro data, gene expression analysis showed a consistent upregulation of some, but not all stem genes in PTC patients with SDHx mutations (Figure 4.3).

In an analysis of SOD2/Sod2 mRNA expression in human and mouse differentiated thyroid cancer, we observed a consistent downregulation of the gene in

FTC. These transcriptional changes were not observed in human or mouse follicular adenomas, suggesting that the change is correlated with cellular transformation.

Interestingly, PTC tumors in the TCGA dataset did not exhibit alterations in SOD2, consistent with different molecular behaviors of these tumors. SOD2 levels were increased as tumors became less differentiated, with anaplastic thyroid cancers having the highest levels.

Because the mortality of thyroid cancer is low, we analyzed human ACC data to look for a correlation between Sod2 expression and survival. We chose to study adrenal cancer because the adrenal gland is another endocrine organ which relies heavily on the cAMP/PKA pathway for trophic growth and physiologic hormone secretion, similar to signaling in the thyroid gland. Unlike differentiated thyroid cancer, ACC is aggressive and there is mortality data to correlate with expression changes. As would be predicted from the effects in our aggressive mouse thyroid cancer model, tumors that expressed higher levels of SOD2 showed improved patient survival.

During recent years, a growing body of evidence has been suggestive of antioxidant enzymes playing crucial roles in cancer development and behavior. The results on SOD2 gene expression in human thyroid and other endocrine tumors demonstrate the dichotomous effects of this antioxidant enzyme. The context-dependent changes in Sod2 expression may be linked to changes in its expression as tumor cells progress from initial benign stage to acquisition of more invasive and metastatic phenotype, or it may be dependent on the underlying oncogenic drivers (144, 153).

Moreover, limited research is available concerning OS and use of antioxidants in thyroid disorders such as hypothyroidism and autoimmune thyroiditis (159, 160); and the research is controversial in case of cancer treatment (145, 161, 162). As Sod2 seems to have a context-dependent role, it is evident that a better understanding of REDOX balance on pharmacogenomics- as well as case-by-case basis is needed to avoid risky treatments for cancer patients.

MATERIALS AND METHDOS

Analysis of TCGA data for SDHx mutations

Normalized RNA-seq read counts for genes of interest based on mutation groups

(namely SDHx+ BRAF vs BRAF-only) were extracted using Firebrowse (163). Patient

IDs with SDHx mutations were provided by Charis Eng. Gene expression data was extracted for SHDx mutant group (based on 28 patients with SDHA/D mutations reported by Ni et al). This TCGA dataset has a total patient sample size of 484, the majority of which are PTC cases. 28 patients had SHDx variants/mutations. Out of 28 SDHx (SDHB and SDHD) mutation cases, 18 patients had co-mutations with BRAF. Hence,

SHDx+BRAF mutation group was compared with BRAF-only group by principle component analysis and hierarchical clustering. In addition, gene expression of “stem” genes was compared between these two groups.

Following R script was used for this analysis.

library(RJSONIO) fb <- function() { paste("http://gdac.broadinstitute.org/runs/stddata__2016_01_28/data/THCA/20160128/g dac.broadinstitute.org_THCA.Merge_rnaseqv2__illuminahiseq_rnaseqv2__unc_edu__L evel_3__RSEM_genes_normalized__data.Level_3.2016012800.0.0.tar.gz", sep="") } fetch <- function() { dataurl <- fb() fname <- basename(dataurl) download.file(dataurl, fname) cat("uncompressing\n", file=stderr()) untar(fname, compressed="gzip") # gunzip and untar unlink(fname) # remove the tar.gz version

localdir <- sub(".tar.gz", "", fname) cat("Creating a binary R object\n", file=stderr()) datasrc <- file.path(localdir, dir(localdir, pattern="RSEM")) dumbheads <- read.table(datasrc, sep="\t", header=TRUE, nrows=1, row.names=1) ncounts <- read.table(datasrc, sep="\t", header=FALSE, skip=2, row.names=1) colnames(ncounts) <- colnames(dumbheads) ncounts <- as.matrix(ncounts) rm(localdir, datasrc, dumbheads) ncounts } setwd("/Users/amruta/Research") f <- "thyroid.Rda" if (file.exists(f)) { load(f) } else { thyroid <- fetch()

save(thyroid, file=f) } rm(f) dim(thyroid) ## [1] 20531 568 #reading data from text files patientIDBRAFMutation<-read.table("/Users/amruta/Research/braf.txt", header = FALSE) patientIDSDHBRAFMutation <- read.table("/Users/amruta/Research/sdh_braf.txt", header = FALSE) #patientIDBRAFMutation<-read.table("/Users/amruta/Research/ras_sdh.txt", header = FALSE) #patientIDSDHBRAFMutation <- read.table("/Users/amruta/Research/ras.txt", header = FALSE)

#converting data to vector thyroidBRAF<-thyroid[,patientIDBRAFMutation[,1]] thyroidSDHBRAF<-thyroid[,patientIDSDHBRAFMutation[,1]]

#shortening patient names BRAF thyroidSampleType_BRAF <- substring(colnames(thyroidBRAF), 14, 15) table(thyroidSampleType_BRAF) ## thyroidSampleType_BRAF ## 01 06 11 ## 242 1 29

#shortening patient names SHD thyroidSampleType_SDHBRAF <- substring(colnames(thyroidSDHBRAF), 14, 15) table(thyroidSampleType_SDHBRAF) ## thyroidSampleType_SDHBRAF ## 01 ## 18 #thyroidSampleID <- substring(colnames(thyroid), 1, 12) #colnames(thyroid)<-thyroidSampleID is.matrix(thyroid) ## [1] TRUE dim(thyroidBRAF) ## [1] 20531 272 dim(thyroidSDHBRAF) ## [1] 20531 18 #selecting 01= primary tumor samples and combing braf and sdh together solidTumorThyroid <- cbind(thyroidBRAF[, thyroidSampleType_BRAF == "01"], thyroidSDHBRAF[, thyroidSampleType_SDHBRAF == "01"]) dim(solidTumorThyroid) ## [1] 20531 260 #creating labels as BRF and SHD stsrc <- factor(rep(c("BRF", "SDH"),times=c(sum(thyroidSampleType_BRAF == "01"), sum(thyroidSampleType_SDHBRAF == "01")))) #Log transforming ST <- log2(1 + solidTumorThyroid) names(stsrc) <- colnames(ST) <- paste(stsrc, substring(colnames(solidTumorThyroid), 1, 12), sep='.')

# Getting the gene names out of the rows. In the process, we remove any duplicate gene names. gn <- unlist(lapply(strsplit(rownames(ST), "\\|"), function(x) x[1])) dup <- duplicated(gn) || gn == '\\?' ST <- ST[!dup,] gn <- gn[!dup]

#Removing zeros M <- apply(ST, 1, max) ST <- ST[M > 0,] gn <- gn[M > 0] dim(ST) ## [1] 20044 260 length(gn) ## [1] 20044

# Color coding data srcColors <- c("green","blue") names(srcColors) <- c("BRF", "SDH") srcVec <- srcColors[stsrc] boxplot(ST, col=srcVec) library(ClassComparison) ## Loading required package: oompaBase library(ClassDiscovery) ## Loading required package: cluster hc <- hclust(distanceMatrix(ST, "pearson"), "ward.D2") plotColoredClusters(hc, lab=stsrc, col=srcVec) spca <- SamplePCA(ST, split=stsrc) plot(spca, col=srcColors) compareTwoMutations <- function(file1,file2,mutation1Name,mutation2Name){ #reading data from text files mutation1Data<-read.table(file1, header = FALSE) mutation2Data <- read.table(file2, header = FALSE)

#converting data to vector mutation1DataVector<-thyroid[,mutation1Data[,1]] mutation2DataVector<-thyroid[,mutation2Data[,1]]

#shortening patient names BRAF thyroidSampleType_mutation1 <- substring(colnames(mutation1DataVector), 14, 15) table(thyroidSampleType_mutation1)

#shortening patient names SHD thyroidSampleType_mutation2 <- substring(colnames(mutation2DataVector), 14, 15) table(thyroidSampleType_mutation2) dim(mutation1DataVector) dim(mutation2DataVector)

#creating labels as BRF and SHD stsrc <- factor(rep(c(mutation1Name, mutation2Name),times=c(sum(thyroidSampleType_mutation1 == "01"), sum(thyroidSampleType_mutation2 == "01")))) #Log transforming ST <- log2(1 + solidTumorThyroid) names(stsrc) <- colnames(ST) <- paste(stsrc, substring(colnames(solidTumorThyroid), 1, 12), sep='.')

#Removing zeros M <- apply(ST, 1, max) ST <- ST[M > 0,] gn <- gn[M > 0] dim(ST) length(gn)

# Color coding data srcColors <- c("green","blue") names(srcColors) <- c(mutation1Name, mutation2Name) srcVec <- srcColors[stsrc] boxplot(ST, col=srcVec) library(ClassComparison) library(ClassDiscovery) hc <- hclust(distanceMatrix(ST, "pearson"), "ward.D2") plotColoredClusters(hc, lab=stsrc, col=srcVec) spca <- SamplePCA(ST, split=stsrc) plot(spca, col=srcColors) } compareTwoMutations("/Users/amruta/Research/braf.txt","/Users/amruta/Research/sdh_ braf.txt","BRAF","SDH_BRAF")

98 compareTwoMutations("/Users/amruta/Research/ras.txt","/Users/amruta/Research/sdh_r as.txt","RAS","SDH_RAS") compareTwoMutations("/Users/amruta/Research/sdh_braf.txt","/Users/amruta/Research/ sdh_ras.txt","SDH_BRAF","SDH_RAS") compareTwoMutations("/Users/amruta/Research/braf.txt","/Users/amruta/Research/ras.t xt","BRAF","RAS") compareThreeMutations <- function(file1,file2,file3,mutation1Name,mutation2Name,mutation3Name){ #reading data from text files mutation1Data<-read.table(file1, header = FALSE) mutation2Data <- read.table(file2, header = FALSE) mutation3Data <- read.table(file3, header = FALSE)

#converting data to vector mutation1DataVector<-thyroid[,mutation1Data[,1]] mutation2DataVector<-thyroid[,mutation2Data[,1]] mutation3DataVector<-thyroid[,mutation3Data[,1]]

#shortening patient names BRAF thyroidSampleType_mutation1 <- substring(colnames(mutation1DataVector), 14, 15) table(thyroidSampleType_mutation1)

#shortening patient names SHD thyroidSampleType_mutation2 <- substring(colnames(mutation2DataVector), 14, 15) table(thyroidSampleType_mutation2) thyroidSampleType_mutation3 <- substring(colnames(mutation3DataVector), 14, 15) table(thyroidSampleType_mutation3) dim(mutation1DataVector) dim(mutation2DataVector) dim(mutation3DataVector)

#selecting 01= primary tumor samples and combing braf and sdh together solidTumorThyroid <- cbind(mutation1DataVector[, thyroidSampleType_mutation1 == "01"], mutation2DataVector[, thyroidSampleType_mutation2 == "01"],mutation3DataVector[, thyroidSampleType_mutation3 == "01"]) dim(solidTumorThyroid)

#creating labels as BRF and SHD stsrc <- factor(rep(c(mutation1Name, mutation2Name,mutation3Name),times=c(sum(thyroidSampleType_mutation1 == "01"),sum(thyroidSampleType_mutation2 == "01"),sum(thyroidSampleType_mutation3 == "01")))) #Log transforming ST <- log2(1 + solidTumorThyroid) names(stsrc) <- colnames(ST) <- paste(stsrc, substring(colnames(solidTumorThyroid), 1, 12), sep='.')

#Removing zeros M <- apply(ST, 1, max) ST <- ST[M > 0,] gn <- gn[M > 0] dim(ST) length(gn)

# Color coding data srcColors <- c("green","blue","red") names(srcColors) <- c(mutation1Name, mutation2Name,mutation3Name) srcVec <- srcColors[stsrc] boxplot(ST, col=srcVec) library(ClassComparison) library(ClassDiscovery) hc <- hclust(distanceMatrix(ST, "pearson"), "ward.D2") plotColoredClusters(hc, lab=stsrc, col=srcVec) spca <- SamplePCA(ST, split=stsrc) plot(spca, col=srcColors) } compareThreeMutations("/Users/amruta/Research/braf.txt","/Users/amruta/Research/SD HB_mutations.txt","/Users/amruta/Research/SDHD_mutations.txt","BRAF","SDHB","S DHD")

100 compareFourMutations <- function(file1,file2,file3,file4,mutation1Name,mutation2Name,mutation3Name,mutation 4Name){ #reading data from text files mutation1Data<-read.table(file1, header = FALSE) mutation2Data <- read.table(file2, header = FALSE) mutation3Data <- read.table(file3, header = FALSE) mutation4Data <- read.table(file4, header = FALSE)

#converting data to vector mutation1DataVector<-thyroid[,mutation1Data[,1]] mutation2DataVector<-thyroid[,mutation2Data[,1]] mutation3DataVector<-thyroid[,mutation3Data[,1]] mutation4DataVector<-thyroid[,mutation4Data[,1]]

#shortening patient names BRAF thyroidSampleType_mutation1 <- substring(colnames(mutation1DataVector), 14, 15) table(thyroidSampleType_mutation1)

#shortening patient names SHD thyroidSampleType_mutation2 <- substring(colnames(mutation2DataVector), 14, 15) table(thyroidSampleType_mutation2) thyroidSampleType_mutation3 <- substring(colnames(mutation3DataVector), 14, 15) table(thyroidSampleType_mutation3) thyroidSampleType_mutation4 <- substring(colnames(mutation4DataVector), 14, 15) table(thyroidSampleType_mutation4) dim(mutation1DataVector) dim(mutation2DataVector) dim(mutation3DataVector) dim(mutation4DataVector)

101

#creating labels as BRF and SHD stsrc <- factor(rep(c(mutation1Name, mutation2Name,mutation3Name,mutation4Name),times=c(sum(thyroidSampleType_mu tation1 == "01"),sum(thyroidSampleType_mutation2 == "01"),sum(thyroidSampleType_mutation3 == "01"),sum(thyroidSampleType_mutation4 == "01"))))

#Log transforming ST <- log2(1 + solidTumorThyroid) names(stsrc) <- colnames(ST) <- paste(stsrc, substring(colnames(solidTumorThyroid), 1, 12), sep='.')

#Removing zeros M <- apply(ST, 1, max) ST <- ST[M > 0,] gn <- gn[M > 0] dim(ST) length(gn)

# Color coding data srcColors <- c("green","blue","red","yellow") names(srcColors) <- c(mutation1Name, mutation2Name,mutation3Name,mutation4Name) srcVec <- srcColors[stsrc] boxplot(ST, col=srcVec) library(ClassComparison) library(ClassDiscovery) hc <- hclust(distanceMatrix(ST, "pearson"), "ward.D2") plotColoredClusters(hc, lab=stsrc, col=srcVec) spca <- SamplePCA(ST, split=stsrc) plot(spca, col=srcColors) } compareFourMutations("/Users/amruta/Research/braf.txt","/Users/amruta/Research/ras.t xt","/Users/amruta/Research/SDHB_mutations.txt","/Users/amruta/Research/SDHD_mu tations.txt","BRAF","RAS","SDHB","SDHD")

102

#all.equal(rownames(thyroidBRAF),rownames(thyroidSDH)) #library(corrplot) #mycorr <- cor(thyroidBRAF) #range(mycorr) #corrplot(mycorr,method="number",number.cex=0.6) solidTumorThyroid_transpose <- t(solidTumorThyroid) d <- dist(solidTumorThyroid_transpose) # euclidean distances between the rows fit <- cmdscale(d,eig=TRUE, k=2) # k is the number of dim fit # view results ## $points ## [,1] [,2] ## TCGA.EL.A3T7.01A.11R.A22L.07 -571947.306 -45058.617 .112440e-03 -4.146866e-03 -9.973287e-03 ## ## $x ## NULL ## ## $ac ## [1] 0 ## ## $GOF ## [1] 0.8567618 0.8567618 save(d, file = "distances_transpose.rda") save(fit, file = "fit_transpose.rda")

# plot solution x <- fit$points[,1] y <- fit$points[,2] plot(x, y, xlab="Coordinate 1", ylab="Coordinate 2", main="Metric MDS", type="n") #text(x, y, labels = row.names(solidTumorThyroid_transpose), cex=.7, col=srcVec) text(x, y, labels = stsrc,cex=.7, col=srcVec)

Transcriptome analysis for oxidative stress genes

The Cancer Genome Atlas and Gene Expression Omnibus were used to screen for SOD2 expression in human endocrine cancers. Three representative data sets are shown in Figure 4.4 B, C and D. (TCGA thyroid carcinoma, GEO Accession no.

103

GSE76039 and TCGA Adrenocortical carcinoma). The dataset in Fig. 1A was downloaded from the EMBL ArrayExpress repository (www.ebi.ac.uk/arrayexpress) using accession ID EMEXP-97 (152). A panel of OS genes were selected from microarray data based on a P value cutoff at 0.01. OS genes were selected from previously published microarray data for human and murine FTC (34, 164). Kaplan-

Meier curve was analyzed using log-rank test from adrenocortical carcinoma using low and high expression of Sod2 at median cutoff (149).

Statistics

All data, except Kaplan Meier curves and cancer incidence, were analyzed via paired or unpaired t-test using Prism GraphPad software. P values less than 0.05 were considered significant.

104

Chapter 5: Concluding Remarks

A plethora of work has focused on BRAFV600 mutations and the RAS/RAF/MEK pathway in thyroid cancers. Our lab has published on the significance of PKA pathway in thyroid carcinogenesis, focusing on FTC development (34). The PI3K/AKT pathway has also been extensively studied and shown to be important in the development of PTC (37).

However, little work has been focused on mechanisms by which metabolic remodeling and oxidative stress leads to thyroid cancer development. The majority of CS patients have mutations in PTEN, but a subset of CS patients do not meet the diagnostic criteria of classic CS and are evaluated for additional mutations. Mutations in SDHx represent such a susceptibility gene for CS-like individuals.

This work describes two new mouse models for studying how SDHD, an additional susceptibility gene associated with CS/CSL cases may contribute to thyroid cancer initiation and progression. We have used four Sdhd-null mouse models, either alone or in combination with R1a, Pten and double R1a-Pten to understand how Sdhd mutations play a role in cancer progression. Additionally, we show how oxidative stress contributes to thyroid cancer development and metastasis.

In Chapter 2, we demonstrated that deletion of Sdhd in mouse thyroids leads to the formation of adenoma with an increased rate of proliferation. Double deletion of

Sdhd and Pten was carcinogenic at advanced age and lead to distant metastases, indicating the possibility that Sdhd may be acting as a co-modulator of the PI3K pathway. Although we did not see inactivation of AKT or ERK in vitro, deletion of Sdhd

105 in vivo resulted in activation of the p-mTOR pathway. mTOR is a known downstream target of AKT and we have previously published that DRP-TpoKO mice harbor activation of mTOR leading to aggressive and metastatic FTC (34). Additionally, we have shown that in vitro knockdown of SDHD leads to an altered metabolic profile, rendering cells more sensitive to sudden energy demands. Most importantly, we have shown that SDHD deletion both in vitro and in vivo leads to the acquisition of a stem- like phenotype. This stem-like phenotype is believed to be caused by increased DNA methylation due to the altered ratio of TCA cycle metabolites (α-KG/succinate), as re- balancing the ratio reverses all the phenotypes seen with SDHD deficiency. It is unclear whether histone methylation and hydroxyl-methylation are affected by increased α-

KG/succinate ratio. Further studies will include the analysis of genomic regions or promoters susceptible to epigenetic changes due to TCA cycle disruptions and potential drug targets such as azacytidine to treat the resulting tumors.

In Chapter 3, we described mouse models of oxidative stress based on genetic manipulation of antioxidant MnSod. We studied tumor formation in Pten-KO (adenoma),

R1a-KO (intermediate carcinoma) and DRP-KO (aggressive and metastasis FTC) with

Sod2 upregulation or downregulation. Interestingly, we showed that Sod2 overexpression in FA led to a more aggressive cancer phenotype, accompanied by increased thyrocyte proliferation. In contrast, the same Sod2 overexpression improved mortality and reduced cell proliferation in the metastatic FTC model (R1a Pten- null mice). Lastly, in mice with an intermediate phenotype of locally invasive FTC, up- or down-regulation of Sod2

106 levels did not have a significant effect on tumor behavior. Based on data from multiple mouse models, we attribute these complex and context-dependent effects of Sod2 to steady-state levels of peroxide. In addition to providing insights into how Sod2 might be regulating cell proliferation and metastasis, these models also help us to understand the differential outcomes of antioxidant therapies. Future work would include in vitro manipulation of Sod2 in primary thyroid cell lines derived from Pten, R1a, DRP-null mouse tumors to understand the changes in signaling pathways as well as to understand how peroxide levels in different types of cancers regulate cell growth and metastasis.

In Chapter 4, we looked for bioinformatics insights to understand the molecular signatures of SDHx mutations. We compared patient cohorts carrying SDHx-variants plus

BRAF mutations to those carrying BRAF-only mutations. This analysis showed that samples with SDHx variants form somewhat distinct clusters than those with BRAF mutations. Further, this analysis corroborated our in vitro finding that SDHD deficiency leads to activation of a stem-like phenotype. To investigate the oxidative stress in endocrine cancers, we looked at relationships between Sod2 expression and histopathological features using publically available human endocrine cancers as well as mouse models of FTC. Future work would include epigenetic analysis of SDHx mutated thyroid tissues to identify the expected phenotype of enhanced DNA methylation and to identify genomic regions affected by SDH loss.

Where low SOD2 levels poorly correlated with survival status in ACC, they were found to be downregulated in FTC patients as well as intermediate (R1a-KO) and aggressive (DRP-KO) FTC mouse models. PTC patients did not have alerted levels of

107

Sod2 in primary tumor or regional metastases. Similarly, a mouse model of FA (PtenKO) did not have alterations in Sod2 levels. Conversely, the aggressive ATC tumors had significantly increased Sod2 levels than PDTC. Overall, these analyses demonstrate that

Sod2 deregulation is complex and context dependent. However, our mouse models of

FTC recapitulate many features of human cancers, including the up- or down-regulation of Sod2. Future studies would include understating how Sod2 levels correlate with the thyroid differentiation score (TDS) in the TCGA thyroid cancer dataset. The work described here as well as future studies could potentially lead to the development of viable drug combinations as well as predictive screening to determine whether antioxidant treatment would be beneficial.

Together, the information gathered so far from mouse models of Sdhd in FTC tumor progression model has varied. Thyroid-specific models of Sdhd-KO alone and in combination with Pten-KO are good models to study epigenetic mechanisms of carcinogenesis and test potential new therapies. In contrast, the phenotypes associated with Sdhd-KO in combination with more aggressive FTC models (R1a-KO and DRP-

KO) have remained unimpressive. Genetic modeling of Sod2 has proven more difficult to interpret with contrasting outcomes and requires future studies to fully understand the cellular behaviors associated with OS. How oxidative stress alters the genetic signaling pathways driving cancer formation and progression have proven more difficult to fully elucidate.

108

Bibliography

1. C Maenhaut , D Christophe , Gilbert Vassart , Jacques Dumont , P.P Roger , Opitz R 2015 Ontogeny, Anatomy, Metabolism and Physiology of the Thyroid. Endotext. 2. L M, G. VdB 2009 The hypothalamus-pituitary-thyroid axis in critical illness. Vol 67. Neth J Med, 332-340. 3. Ortiga-Carvalho TM, Chiamolera MI, Pazos-Moura CC, Wondisford FE 2016 Hypothalamus-Pituitary-Thyroid Axis. Compr Physiol 6:1387-1428. 4. Lu M, Lin RY 2008 TSH stimulates adipogenesis in mouse embryonic stem cells. J Endocrinol 196:159-169. 5. Abu-Khudir R, Larrivée-Vanier S, Wasserman JD, Deladoëy J 2017 Disorders of thyroid morphogenesis. Best Pract Res Clin Endocrinol Metab 31:143-159. 6. Brent GA 2012 Mechanisms of thyroid hormone action. J Clin Invest 122:30353043. 7. Mullur R, Liu YY, Brent GA 2014 Thyroid hormone regulation of metabolism. Physiol Rev 94:355-382. 8. Leenhardt L, Grosclaude P, Cherie-Challine L 2004 Increased incidence of thyroid carcinoma in france: a true epidemic or thyroid nodule management effects? Report from the French Thyroid Cancer Committee. Thyroid 14:10561060. 9. Li N, Du XL, Reitzel LR, Xu L, Sturgis EM 2013 Impact of enhanced detection on the increase in thyroid cancer incidence in the United States: review of incidence trends by socioeconomic status within the surveillance, epidemiology, and end results registry, 1980-2008. Thyroid 23:103-110. 10. Mao Y, Xing M 2016 Recent incidences and differential trends of thyroid cancer in the USA. Endocr Relat Cancer 23:313-322. 11. Song YS, Lim JA, Min HS, Kim MJ, Choi HS, Cho SW, Moon JH, Yi KH, Park DJ, Cho BY, Park YJ 2017 Changes in the clinicopathological characteristics and genetic alterations of follicular thyroid cancer. Eur J Endocrinol 177:465-473. 12. Kirschner LS, Qamri Z, Kari S, Ashtekar A 2016 Mouse models of thyroid cancer: A 2015 update. Mol Cell Endocrinol 421:18-27. 13. Lim SM, Shin SJ, Chung WY, Park CS, Nam KH, Kang SW, Keum KC, Kim JH, Cho JY, Hong YK, Cho BC 2012 Treatment outcome of patients with anaplastic thyroid cancer: a single center experience. Yonsei Med J 53:352-357. 14. Nguyen QT, Lee EJ, Huang MG, Park YI, Khullar A, Plodkowski RA 2015 Diagnosis and treatment of patients with thyroid cancer. Am Health Drug Benefits 8:30-40.

109

15. Hambleton C, Kandil E 2013 Appropriate and accurate diagnosis of thyroid nodules: a review of thyroid fine-needle aspiration. Int J Clin Exp Med 6:413- 422. 16. Gilfillan CP 2010 Review of the genetics of thyroid tumours: diagnostic and prognostic implications. ANZ J Surg 80:33-40. 17. Fusco A, Grieco M, Santoro M, Berlingieri MT, Pilotti S, Pierotti MA, Della Porta G, Vecchio G 1987 A new oncogene in human thyroid papillary carcinomas and their lymph-nodal metastases. Nature 328:170-172. 18. Romei C, Elisei R 2012 RET/PTC Translocations and Clinico-Pathological Features in Human Papillary Thyroid Carcinoma. Front Endocrinol (Lausanne) 3:54. 19. Nikiforova MN, Nikiforov YE 2009 Molecular diagnostics and predictors in thyroid cancer. Thyroid 19:1351-1361. 20. Hou P, Liu D, Shan Y, Hu S, Studeman K, Condouris S, Wang Y, Trink A, ElNaggar AK, Tallini G, Vasko V, Xing M 2007 Genetic alterations and their relationship in the phosphatidylinositol 3-kinase/Akt pathway in thyroid cancer. Clin Cancer Res 13:1161-1170. 21. Richards ML 2010 Familial syndromes associated with thyroid cancer in the era of personalized medicine. Thyroid 20:707-713. 22. Eng C 2003 PTEN: one gene, many syndromes. Hum Mutat 22:183-198. 23. Chalhoub N, Baker SJ 2009 PTEN and the PI3-kinase pathway in cancer. Annu Rev Pathol 4:127-150. 24. Nagy R, Ganapathi S, Comeras I, Peterson C, Orloff M, Porter K, Eng C, Ringel MD, Kloos RT 2011 Frequency of germline PTEN mutations in differentiated thyroid cancer. Thyroid 21:505-510. 25. Ni Y, Zbuk KM, Sadler T, Patocs A, Lobo G, Edelman E, Platzer P, Orloff MS, Waite KA, Eng C 2008 Germline mutations and variants in the succinate dehydrogenase genes in Cowden and Cowden-like syndromes. Am J Hum Genet 83:261-268. 26. Eng C 2001 PTEN Hamartoma Tumor Syndrome. GeneReviews. 27. Espiard S, Bertherat J 2013 Carney complex. Front Horm Res 41:50-62. 28. Stratakis CA, Kirschner LS, Carney JA 2001 Clinical and molecular features of the Carney complex: diagnostic criteria and recommendations for patient evaluation. J Clin Endocrinol Metab 86:4041-4046. 29. Stratakis CA, Carney JA, Lin JP, Papanicolaou DA, Karl M, Kastner DL, Pras E, Chrousos GP 1996 Carney complex, a familial multiple neoplasia and lentiginosis syndrome. Analysis of 11 kindreds and linkage to the short arm of chromosome 2. J Clin Invest 97:699-705. 30. Liu Q, Tong D, Liu G, Yi Y, Zhang D, Zhang J, Zhang Y, Huang Z, Li Y, Chen R, Guan Y, Yi X, Jiang J 2017 Carney complex with PRKAR1A gene mutation: A case report and literature review. Medicine (Baltimore) 96:e8999.

110

31. Wang S, Cheng Y, Zheng Y, He Z, Chen W, Zhou W, Duan C, Zhang C 2016 PRKAR1A is a functional tumor suppressor inhibiting ERK/Snail/E-cadherin pathway in lung adenocarcinoma. Sci Rep 6:39630. 32. Ueyama K, Namba N, Kitaoka T, Yamamoto K, Fujiwara M, Ohata Y, Kubota T, Ozono K 2017 Endocrinological and phenotype evaluation in a patient with acrodysostosis. Clin Pediatr Endocrinol 26:177-182. 33. Knudson AG 1971 Mutation and cancer: statistical study of retinoblastoma. Proc Natl Acad Sci U S A 68:820-823. 34. Pringle DR, Vasko VV, Yu L, Manchanda PK, Lee AA, Zhang X, Kirschner JM, Parlow AF, Saji M, Jarjoura D, Ringel MD, La Perle KM, Kirschner LS 2014 Follicular thyroid cancers demonstrate dual activation of PKA and mTOR as modeled by thyroid-specific deletion of Prkar1a and Pten in mice. J Clin Endocrinol Metab 99:E804-812. 35. Kirschner LS, Kusewitt DF, Matyakhina L, Towns WH, Carney JA, Westphal H, Stratakis CA 2005 A mouse model for the Carney complex tumor syndrome develops neoplasia in cyclic AMP-responsive tissues. Cancer Res 65:4506-4514. 36. Yeager N, Klein-Szanto A, Kimura S, Di Cristofano A 2007 Pten loss in the mouse thyroid causes goiter and follicular adenomas: insights into thyroid function and Cowden disease pathogenesis. Cancer research 67:959-966. 37. Halachmi N, Halachmi S, Evron E, Cairns P, Okami K, Saji M, Westra WH, Zeiger MA, Jen J, Sidransky D 1998 Somatic mutations of the PTEN tumor suppressor gene in sporadic follicular thyroid tumors. Genes Chromosomes Cancer 23:239-243. 38. Kusakabe T, Kawaguchi A, Kawaguchi R, Feigenbaum L, Kimura S 2004 Thyrocyte-specific expression of Cre recombinase in transgenic mice. Genesis 39:212-216. 39. Pringle DR, Yin Z, Lee AA, Manchanda PK, Yu L, Parlow AF, Jarjoura D, La Perle KM, Kirschner LS 2012 Thyroid-specific ablation of the Carney complex gene, PRKAR1A, results in hyperthyroidism and follicular thyroid cancer. Endocr Relat Cancer 19:435-446. 40. Jeremy B, John T, Lubert S Biochemistry, New York. 41. Akram M 2014 Citric acid cycle and role of its intermediates in metabolism. Cell Biochem Biophys 68:475-478. 42. Hanahan D, Weinberg RA 2011 Hallmarks of cancer: the next generation. Cell 144:646-674. 43. Frezza C 2014 The role of mitochondria in the oncogenic signal transduction. Int J Biochem Cell Biol 48:11-17. 44. Yang L, Moss T, Mangala LS, Marini J, Zhao H, Wahlig S, Armaiz-Pena G, Jiang D, Achreja A, Win J, Roopaimoole R, Rodriguez-Aguayo C, Mercado-Uribe I,

111

Lopez-Berestein G, Liu J, Tsukamoto T, Sood AK, Ram PT, Nagrath D 2014 Metabolic shifts toward glutamine regulate tumor growth, invasion and bioenergetics in ovarian cancer. Mol Syst Biol 10:728. 45. Chang B, Yang H, Jiao Y, Wang K, Liu Z, Wu P, Li S, Wang A 2016 SOD2 deregulation enhances migration, invasion and has poor prognosis in salivary adenoid cystic carcinoma. Sci Rep 6:25918. 46. Cairns RA, Mak TW 2013 Oncogenic isocitrate dehydrogenase mutations: mechanisms, models, and clinical opportunities. Cancer Discov 3:730-741. 47. Smith AC, Robinson AJ 2011 A metabolic model of the mitochondrion and its use in modelling diseases of the tricarboxylic acid cycle. BMC Syst Biol 5:102. 48. King A, Selak MA, Gottlieb E 2006 Succinate dehydrogenase and fumarate hydratase: linking mitochondrial dysfunction and cancer. Oncogene 25:46754682. 49. Cantor JR, Sabatini DM 2012 Cancer cell metabolism: one hallmark, many faces. Cancer discovery 2:881-898. 50. Bardella C, Pollard PJ, Tomlinson I 2011 SDH mutations in cancer. Biochimica et biophysica acta 1807:1432-1443. 51. Letouze E, Martinelli C, Loriot C, Burnichon N, Abermil N, Ottolenghi C, Janin M, Menara M, Nguyen AT, Benit P, Buffet A, Marcaillou C, Bertherat J, Amar L, Rustin P, De Reynies A, Gimenez-Roqueplo AP, Favier J 2013 SDH mutations establish a hypermethylator phenotype in paraganglioma. Cancer cell 23:739-752. 52. Millan-Ucles A, Diaz-Castro B, Garcia-Flores P, Baez A, Perez-Simon JA, Lopez-Barneo J, Piruat JI 2014 A conditional mouse mutant in the tumor suppressor SdhD gene unveils a link between p21(WAF1/Cip1) induction and mitochondrial dysfunction. PloS one 9:e85528. 53. Williamson SR, Eble JN, Amin MB, Gupta NS, Smith SC, Sholl LM, Montironi R, Hirsch MS, Hornick JL 2015 Succinate dehydrogenase-deficient renal cell carcinoma: detailed characterization of 11 tumors defining a unique subtype of renal cell carcinoma. Mod Pathol 28:80-94. 54. Xiao M, Yang H, Xu W, Ma S, Lin H, Zhu H, Liu L, Liu Y, Yang C, Xu Y, Zhao S, Ye D, Xiong Y, Guan KL 2012 Inhibition of alpha-KG-dependent histone and DNA demethylases by fumarate and succinate that are accumulated in mutations of FH and SDH tumor suppressors. Genes & development 26:1326-1338. 55. P R, T B, B P, D C, A M, . RA 1997 Inborn errors of the Krebs cycle: a group of unusual mitochondrial diseases in human. Biochim Biophys Acta 1361. 56. Yu W, He X, Ni Y, Ngeow J, Eng C 2015 Cowden syndrome-associated germline SDHD variants alter PTEN nuclear translocation through SRC-induced PTEN oxidation. Human molecular genetics 24:142-153. 57. Archetti M 2015 Heterogeneity and proliferation of invasive cancer subclones in game theory models of the Warburg effect. Cell proliferation 48:259-269. 58. Wong CC, Qian Y, Yu J 2017 Interplay between epigenetics and metabolism in oncogenesis: mechanisms and therapeutic approaches. Oncogene 36:3359-3374. 112

59. Jardim-Messeder D, Caverzan A, Rauber R, de Souza Ferreira E, Margis-Pinheiro M, Galina A 2015 Succinate dehydrogenase (mitochondrial complex II) is a source of reactive oxygen species in plants and regulates development and stress responses. The New phytologist 208:776-789. 60. MacKenzie ED, Selak MA, Tennant DA, Payne LJ, Crosby S, Frederiksen CM, Watson DG, Gottlieb E 2007 Cell-permeating alpha-ketoglutarate derivatives alleviate pseudohypoxia in succinate dehydrogenase-deficient cells. Molecular and cellular biology 27:3282-3289. 61. Salminen A, Kaarniranta K, Hiltunen M, Kauppinen A 2014 Krebs cycle dysfunction shapes epigenetic landscape of chromatin: novel insights into mitochondrial regulation of aging process. Cellular signalling 26:1598-1603. 62. Chang YL, Hsieh MH, Chang WW, Wang HY, Lin MC, Wang CP, Lou PJ, Teng SC 2015 Instability of succinate dehydrogenase in SDHD polymorphism connects reactive oxygen species production to nuclear and mitochondrial genomic mutations in yeast. Antioxid Redox Signal 22:587-602. 63. Lemarie A, Huc L, Pazarentzos E, Mahul-Mellier AL, Grimm S 2011 Specific disintegration of complex II succinate:ubiquinone oxidoreductase links pH changes to oxidative stress for apoptosis induction. Cell Death Differ 18:338-349. 64. Ishii T, Yasuda K, Akatsuka A, Hino O, Hartman PS, Ishii N 2005 A mutation in the SDHC gene of complex II increases oxidative stress, resulting in apoptosis and tumorigenesis. Cancer Res 65:203-209. 65. Slane BG, Aykin-Burns N, Smith BJ, Kalen AL, Goswami PC, Domann FE, Spitz DR 2006 Mutation of succinate dehydrogenase subunit C results in increased O2., oxidative stress, and genomic instability. Cancer Res 66:7615- 7620. 66. Wang YM, Gu ML, Ji F 2015 Succinate dehydrogenase-deficient gastrointestinal stromal tumors. World J Gastroenterol 21:2303-2314. 67. Eales KL, Hollinshead KE, Tennant DA 2016 Hypoxia and metabolic adaptation of cancer cells. Oncogenesis 5:e190. 68. Benit P, Letouze E, Rak M, Aubry L, Burnichon N, Favier J, Gimenez-Roqueplo AP, Rustin P 2014 Unsuspected task for an old team: succinate, fumarate and other Krebs cycle acids in metabolic remodeling. Biochim Biophys Acta 1837:1330-1337. 69. Briere JJ, Favier J, Benit P, El Ghouzzi V, Lorenzato A, Rabier D, Di Renzo MF, Gimenez-Roqueplo AP, Rustin P 2005 Mitochondrial succinate is instrumental for HIF1alpha nuclear translocation in SDHA-mutant fibroblasts under normoxic conditions. Human molecular genetics 14:3263-3269. 70. Tsang VH, Dwight T, Benn DE, Meyer-Rochow GY, Gill AJ, Sywak M, Sidhu S, Veivers D, Sue CM, Robinson BG, Clifton-Bligh RJ, Parker NR 2014 Overexpression of miR-210 is associated with SDH-related pheochromocytomas, paragangliomas, and gastrointestinal stromal tumours. Endocr Relat Cancer 21:415-426. 113

71. Guzy RD, Sharma B, Bell E, Chandel NS, Schumacker PT 2008 Loss of the SdhB, but Not the SdhA, subunit of complex II triggers reactive oxygen speciesdependent hypoxia-inducible factor activation and tumorigenesis. Mol Cell Biol 28:718-731. 72. Szarek E, Ball ER, Imperiale A, Tsokos M, Faucz FR, Giubellino A, Moussallieh FM, Namer IJ, Abu-Asab MS, Pacak K, Taïeb D, Carney JA, Stratakis CA 2015 Carney triad, SDH-deficient tumors, and Sdhb+/- mice share abnormal mitochondria. Endocr Relat Cancer 22:345-352. 73. Killian JK, Miettinen M, Walker RL, Wang Y, Zhu YJ, Waterfall JJ, Noyes N, Retnakumar P, Yang Z, Smith WI, Killian MS, Lau CC, Pineda M, Walling J, Stevenson H, Smith C, Wang Z, Lasota J, Kim SY, Boikos SA, Helman LJ, Meltzer PS 2014 Recurrent epimutation of SDHC in gastrointestinal stromal tumors. Sci Transl Med 6:268ra177. 74. Carey BW, Finley LW, Cross JR, Allis CD, Thompson CB 2015 Intracellular alpha-ketoglutarate maintains the pluripotency of embryonic stem cells. Nature 518:413-416. 75. Sciacovelli M, Frezza C 2016 Oncometabolites: Unconventional triggers of oncogenic signalling cascades. Free radical biology & medicine 100:175-181. 76. Xiao M, Yang H, Xu W, Ma S, Lin H, Zhu H, Liu L, Liu Y, Yang C, Xu Y, Zhao S, Ye D, Xiong Y, Guan KL 2012 Inhibition of α-KG-dependent histone and DNA demethylases by fumarate and succinate that are accumulated in mutations of FH and SDH tumor suppressors. Genes Dev 26:1326-1338. 77. J. Keith Killian MM, Robert L. Walker, Yonghong Wang, Yuelin Jack Zhu, Joshua J. Waterfall, Natalia Noyes, Parvathy Retnakumar, Zhiming Yang, William I. Smith Jr., M. Scott Killian, C. Christopher Lau, Marbin Pineda, Jennifer W 2014 Recurrent epimutation of SDHC in gastrointestinal stromal tumors. Sci Transl Med 6. 78. Letouzé E MC, Loriot C, Burnichon N, Abermil N, Ottolenghi C, Janin M, Menara M, Nguyen AT, Benit P, Buffet A, Marcaillou C, Bertherat J, Amar L, Rustin P, De Reyniès A, Gimenez-Roqueplo AP, Favier J. 2013 SDH Mutations Establish a Hypermethylator Phenotype in Paraganglioma. Cancer Cell 23. 79. Killian JK1 KS, Miettinen M, Smith C, Merino M, Tsokos M, Quezado M, Smith WI Jr, Jahromi MS, Xekouki P, Szarek E, Walker RL, Lasota J, Raffeld M, Klotzle B, Wang Z, Jones L, Zhu Y, Wang Y, Waterfall JJ, O'Sullivan MJ, Bibikova M, Pacak K, Stratakis 2013 Succinate Dehydrogenase Mutation Underlies Global Epigenomic Divergence in Gastrointestinal Stromal Tumor. Cancer Discov 6. 80. Szarek E, Ball ER, Imperiale A, Tsokos M, Faucz FR, Giubellino A, Moussallieh FM, Namer IJ, Abu-Asab MS, Pacak K, Taieb D, Carney JA, Stratakis CA 2015 Carney triad, SDH-deficient tumors, and Sdhb+/- mice share abnormal mitochondria. Endocrine-related cancer 22:345-352.

114

81. Burgess RJ, Agathocleous M, Morrison SJ 2014 Metabolic regulation of stem cell function. Journal of internal medicine 276:12-24. 82. Hermann PC, Huber SL, Herrler T, Aicher A, Ellwart JW, Guba M, Bruns CJ, Heeschen C 2007 Distinct populations of cancer stem cells determine tumor growth and metastatic activity in human pancreatic cancer. Cell stem cell 1:313323. 83. Li Q, Ye L, Zhang X, Wang M, Lin C, Huang S, Guo W, Lai Y, Du H, Li J, Song L, Peng X 2017 FZD8, a target of p53, promotes bone metastasis in prostate cancer by activating canonical Wnt/beta-catenin signaling. Cancer letters 402:166-176. 84. Pardal R, Molofsky AV, He S, Morrison SJ 2005 Stem cell self-renewal and cancer cell proliferation are regulated by common networks that balance the activation of proto-oncogenes and tumor suppressors. Cold Spring Harbor symposia on quantitative biology 70:177-185. 85. Rodriguez-Torres M, Allan AL 2016 Aldehyde dehydrogenase as a marker and functional mediator of metastasis in solid tumors. Clinical & experimental metastasis 33:97-113. 86. Nagayama Y, Shimamura M, Mitsutake N 2016 Cancer Stem Cells in the Thyroid. Frontiers in endocrinology 7:20. 87. Baylin SB 2005 DNA methylation and gene silencing in cancer. Nature clinical practice Oncology 2 Suppl 1:S4-11. 88. Lu C, Venneti S, Akalin A, Fang F, Ward PS, Dematteo RG, Intlekofer AM, Chen C, Ye J, Hameed M, Nafa K, Agaram NP, Cross JR, Khanin R, Mason CE, Healey JH, Lowe SW, Schwartz GK, Melnick A, Thompson CB 2013 Induction of sarcomas by mutant IDH2. Genes Dev 27:1986-1998. 89. Aspuria PJ, Lunt SY, Varemo L, Vergnes L, Gozo M, Beach JA, Salumbides B, Reue K, Wiedemeyer WR, Nielsen J, Karlan BY, Orsulic S 2014 Succinate dehydrogenase inhibition leads to epithelial-mesenchymal transition and reprogrammed carbon metabolism. Cancer & metabolism 2:21. 90. Edalat A, Schulte-Mecklenbeck P, Bauer C, Undank S, Krippeit-Drews P, Drews G, Dufer M 2015 Mitochondrial succinate dehydrogenase is involved in stimulussecretion coupling and endogenous ROS formation in murine beta cells. Diabetologia 58:1532-1541. 91. Nowicki S, Gottlieb E 2015 Oncometabolites: tailoring our genes. Febs j 282:2796-2805. 92. Tabansky I, Stern JN, Pfaff DW 2015 Implications of Epigenetic Variability within a Cell Population for "Cell Type" Classification. Front Behav Neurosci 9:342. 93. Denkert C, Budczies J, Weichert W, Wohlgemuth G, Scholz M, Kind T, Niesporek S, Noske A, Buckendahl A, Dietel M, Fiehn O 2008 Metabolite profiling of human colon carcinoma--deregulation of TCA cycle and amino acid turnover. Molecular cancer 7:72.

115

94. Son Y, Cheong YK, Kim NH, Chung HT, Kang DG, Pae HO 2011 Mitogen- Activated Protein Kinases and Reactive Oxygen Species: How Can ROS Activate MAPK Pathways? J Signal Transduct 2011:792639. 95. Harris IS, Blaser H, Moreno J, Treloar AE, Gorrini C, Sasaki M, Mason JM, Knobbe CB, Rufini A, Hallé M, Elia AJ, Wakeham A, Tremblay ML, Melino G, Done S, Mak TW 2014 PTPN12 promotes resistance to oxidative stress and supports tumorigenesis by regulating FOXO signaling. Oncogene 33:1047-1054. 96. Bell EL, Emerling BM, Ricoult SJ, Guarente L 2011 SirT3 suppresses hypoxia inducible factor 1α and tumor growth by inhibiting mitochondrial ROS production. Oncogene 30:2986-2996. 97. Klaunig JE, Kamendulis LM, Hocevar BA 2010 Oxidative stress and oxidative damage in carcinogenesis. Toxicol Pathol 38:96-109. 98. Guo YL, Chakraborty S, Rajan SS, Wang R, Huang F 2010 Effects of oxidative stress on mouse embryonic stem cell proliferation, apoptosis, senescence, and self-renewal. Stem Cells Dev 19:1321-1331. 99. Schieke SM, Ma M, Cao L, McCoy JP, Liu C, Hensel NF, Barrett AJ, Boehm M, Finkel T 2008 Mitochondrial metabolism modulates differentiation and teratoma formation capacity in mouse embryonic stem cells. J Biol Chem 283:2850628512. 100. Bigarella CL, Liang R, Ghaffari S 2014 Stem cells and the impact of ROS signaling. Development 141:4206-4218. 101. Pinegin B, Vorobjeva N, Pashenkov M, Chernyak B 2017 The role of mitochondrial ROS in antibacterial immunity. J Cell Physiol. 102. Wang D, Feng JF, Zeng P, Yang YH, Luo J, Yang YW 2011 Total oxidant/antioxidant status in sera of patients with thyroid cancers. Endocr Relat Cancer 18:773-782. 103. Reuter S, Gupta SC, Chaturvedi MM, Aggarwal BB 2010 Oxidative stress, inflammation, and cancer: how are they linked? Free Radic Biol Med 49:16031616. 104. Huang J, Lam GY, Brumell JH 2011 Autophagy signaling through reactive oxygen species. Antioxid Redox Signal 14:2215-2231. 105. Costa A, Scholer-Dahirel A, Mechta-Grigoriou F 2014 The role of reactive oxygen species and metabolism on cancer cells and their microenvironment. Semin Cancer Biol 25:23-32. 106. Sentman ML, Granström M, Jakobson H, Reaume A, Basu S, Marklund SL 2006 Phenotypes of mice lacking extracellular superoxide dismutase and copper- and zinc-containing superoxide dismutase. J Biol Chem 281:6904-6909. 107. Carlsson LM, Jonsson J, Edlund T, Marklund SL 1995 Mice lacking extracellular superoxide dismutase are more sensitive to hyperoxia. Proc Natl Acad Sci U S A 92:6264-6268.

116

108. Vincent AM, Russell JW, Sullivan KA, Backus C, Hayes JM, McLean LL, Feldman EL 2007 SOD2 protects neurons from injury in cell culture and animal models of diabetic neuropathy. Exp Neurol 208:216-227. 109. Li Y, Huang TT, Carlson EJ, Melov S, Ursell PC, Olson JL, Noble LJ, Yoshimura MP, Berger C, Chan PH, Wallace DC, Epstein CJ 1995 Dilated cardiomyopathy and neonatal lethality in mutant mice lacking manganese superoxide dismutase. Nat Genet 11:376-381. 110. Gill JG, Piskounova E, Morrison SJ 2016 Cancer, Oxidative Stress, and Metastasis. Cold Spring Harb Symp Quant Biol 81:163-175. 111. Mut-Salud N, Álvarez PJ, Garrido JM, Carrasco E, Aránega A, RodríguezSerrano F 2016 Antioxidant Intake and Antitumor Therapy: Toward Nutritional Recommendations for Optimal Results. Oxid Med Cell Longev 2016:6719534. 112. Clark LC, Combs GF, Turnbull BW, Slate EH, Chalker DK, Chow J, Davis LS, Glover RA, Graham GF, Gross EG, Krongrad A, Lesher JL, Park HK, Sanders BB, Smith CL, Taylor JR 1996 Effects of selenium supplementation for cancer prevention in patients with carcinoma of the skin. A randomized controlled trial. Nutritional Prevention of Cancer Study Group. JAMA 276:1957-1963. 113. Vance TM, Su J, Fontham ET, Koo SI, Chun OK 2013 Dietary antioxidants and prostate cancer: a review. Nutr Cancer 65:793-801. 114. Steinhubl SR 2008 Why have antioxidants failed in clinical trials? Am J Cardiol 101:14D-19D. 115. Xing M 2012 Oxidative stress: a new risk factor for thyroid cancer. Endocr Relat Cancer 19:C7-11. 116. Ohye H, Sugawara M 2010 Dual oxidase, hydrogen peroxide and thyroid diseases. Exp Biol Med (Maywood) 235:424-433. 117. Young O, Crotty T, O'Connell R, O'Sullivan J, Curran AJ 2010 Levels of oxidative damage and lipid peroxidation in thyroid neoplasia. Head Neck 32:750756. 118. Lassoued S, Mseddi M, Mnif F, Abid M, Guermazi F, Masmoudi H, El Feki A, Attia H 2010 A comparative study of the oxidative profile in Graves' disease, Hashimoto's thyroiditis, and papillary thyroid cancer. Biol Trace Elem Res 138:107-115. 119. Nishida S, Akai F, Iwasaki H, Hosokawa K, Kusunoki T, Suzuki K, Taniguchi N, Hashimoto S, Tamura TT 1993 Manganese superoxide dismutase content and localization in human thyroid tumours. J Pathol 169:341-345. 120. Williams MD, Van Remmen H, Conrad CC, Huang TT, Epstein CJ, Richardson A 1998 Increased oxidative damage is correlated to altered mitochondrial function in heterozygous manganese superoxide dismutase knockout mice. J Biol Chem 273:28510-28515. 121. Yen HC, Oberley TD, Vichitbandha S, Ho YS, St Clair DK 1996 The protective role of manganese superoxide dismutase against adriamycin-induced acute cardiac toxicity in transgenic mice. J Clin Invest 98:1253-1260. 117

122. Oh SS, Sullivan KA, Wilkinson JE, Backus C, Hayes JM, Sakowski SA, Feldman EL 2012 Neurodegeneration and early lethality in superoxide dismutase 2deficient mice: a comprehensive analysis of the central and peripheral nervous systems. Neuroscience 212:201-213. 123. Van Remmen H, Ikeno Y, Hamilton M, Pahlavani M, Wolf N, Thorpe SR, Alderson NL, Baynes JW, Epstein CJ, Huang TT, Nelson J, Strong R, Richardson A 2003 Life-long reduction in MnSOD activity results in increased DNA damage and higher incidence of cancer but does not accelerate aging. Physiol Genomics 16:29-37. 124. Lark DS, Kang L, Lustig ME, Bonner JS, James FD, Neufer PD, Wasserman DH 2015 Enhanced mitochondrial superoxide scavenging does not improve muscle insulin action in the high fat-fed mouse. PLoS One 10:e0126732. 125. Zhao Y, Xue Y, Oberley TD, Kiningham KK, Lin SM, Yen HC, Majima H, Hines J, St Clair D 2001 Overexpression of manganese superoxide dismutase suppresses tumor formation by modulation of activator protein-1 signaling in a multistage skin carcinogenesis model. Cancer Res 61:6082-6088. 126. Zhong W, Oberley LW, Oberley TD, St Clair DK 1997 Suppression of the malignant phenotype of human glioma cells by overexpression of manganese superoxide dismutase. Oncogene 14:481-490. 127. Weydert C, Roling B, Liu J, Hinkhouse MM, Ritchie JM, Oberley LW, Cullen JJ 2003 Suppression of the malignant phenotype in human pancreatic cancer cells by the overexpression of manganese superoxide dismutase. Mol Cancer Ther 2:361369. 128. Hart PC, Mao M, de Abreu AL, Ansenberger-Fricano K, Ekoue DN, Ganini D, Kajdacsy-Balla A, Diamond AM, Minshall RD, Consolaro ME, Santos JH, Bonini MG 2015 MnSOD upregulation sustains the Warburg effect via mitochondrial ROS and AMPK-dependent signalling in cancer. Nat Commun 6:6053. 129. Zhang Y, Zhang HM, Shi Y, Lustgarten M, Li Y, Qi W, Zhang BX, Van Remmen H 2010 Loss of manganese superoxide dismutase leads to abnormal growth and signal transduction in mouse embryonic fibroblasts. Free Radic Biol Med 49:1255-1262. 130. Hemachandra LP, Shin DH, Dier U, Iuliano JN, Engelberth SA, Uusitalo LM, Murphy SK, Hempel N 2015 Mitochondrial Superoxide Dismutase Has a Protumorigenic Role in Ovarian Clear Cell Carcinoma. Cancer Res 75:49734984. 131. Ansenberger-Fricano K, Ganini D, Mao M, Chatterjee S, Dallas S, Mason RP, Stadler K, Santos JH, Bonini MG 2013 The peroxidase activity of mitochondrial superoxide dismutase. Free Radic Biol Med 54:116-124. 132. Connor KM, Subbaram S, Regan KJ, Nelson KK, Mazurkiewicz JE, Bartholomew PJ, Aplin AE, Tai YT, Aguirre-Ghiso J, Flores SC, Melendez JA 2005 Mitochondrial H2O2 regulates the angiogenic phenotype via PTEN oxidation. J Biol Chem 280:16916-16924. 118

133. Dolado I, Nebreda AR 2008 AKT and oxidative stress team up to kill cancer cells. Cancer Cell 14:427-429. 134. Lebovitz RM, Zhang H, Vogel H, Cartwright J, Dionne L, Lu N, Huang S, Matzuk MM 1996 Neurodegeneration, myocardial injury, and perinatal death in mitochondrial superoxide dismutase-deficient mice. Proc Natl Acad Sci U S A 93:9782-9787. 135. Ashtekar A, Huk D, Magner A, La Perle K, Zhang X, Piruat J, Lopez-Barneo J, Jhiang S, Kirschner L 2017 Sdhd ablation promotes thyroid tumorigenesis by inducing a stem-like phenotype. Endocr Relat Cancer. 136. Pasini B, Stratakis CA 2009 SDH mutations in tumorigenesis and inherited endocrine tumours: lesson from the phaeochromocytoma-paraganglioma syndromes. J Intern Med 266:19-42. 137. Dwight T, Mann K, Benn DE, Robinson BG, McKelvie P, Gill AJ, Winship I, Clifton-Bligh RJ 2013 Familial SDHA mutation associated with pituitary adenoma and pheochromocytoma/paraganglioma. J Clin Endocrinol Metab 98:E1103-1108. 138. Tevosian SG, Ghayee HK 2018 Pheochromocytoma/Paraganglioma: A Poster Child for Cancer Metabolism. J Clin Endocrinol Metab. 139. Taïeb D, Pacak K 2018 Molecular imaging and theranostic approaches in pheochromocytoma and paraganglioma. Cell Tissue Res. 140. Eva Szarek ERB, Alessio Imperiale , Maria Tsokos , Fabio R. Faucz , Alessio Giubellino , François-Marie Moussallieh , Izzie-Jacques Namer , Mones S. Abu-Asab , Karel Pacak , David Taïeb , J. Aidan Car Eva Szarek , Evan R. Ball , Alessio Imperiale , Maria Tsokos , Fabio R. Faucz , Alessio Giubellino , François-Marie Moussallieh , Izzie-Jacques Namer , Mones S. Abu-Asab , Karel Pacak , David Taïeb , J. Aidan Carney , and Constantine A. Stratakis ney , and Constantine A. Stratakis 2015 Carney Triad, SDH- deficient tumors, and Sdhb +/− mice share abnormal mitochondria. Endocrine Related Cancer. 141. Ni Y, Seballos S, Ganapathi S, Gurin D, Fletcher B, Ngeow J, Nagy R, Kloos RT, Ringel MD, LaFramboise T, Eng C 2015 Germline and somatic SDHx alterations in apparently sporadic differentiated thyroid cancer. Endocrine-related cancer 22:121-130. 142. Lustgarten MS, Jang YC, Liu Y, Qi W, Qin Y, Dahia PL, Shi Y, Bhattacharya A, Muller FL, Shimizu T, Shirasawa T, Richardson A, Van Remmen H 2011 MnSOD deficiency results in elevated oxidative stress and decreased mitochondrial function but does not lead to muscle atrophy during aging. Aging Cell 10:493-505. 143. Liu X, Huang J, Chen T, Wang Y, Xin S, Li J, Pei G, Kang J 2008 Yamanaka factors critically regulate the developmental signaling network in mouse embryonic stem cells. Cell Res 18:1177-1189.

119

144. Dhar SK, Tangpong J, Chaiswing L, Oberley TD, St Clair DK 2011 Manganese superoxide dismutase is a p53-regulated gene that switches cancers between early and advanced stages. Cancer Res 71:6684-6695. 145. Le Gal K, Ibrahim MX, Wiel C, Sayin VI, Akula MK, Karlsson C, Dalin MG, Akyürek LM, Lindahl P, Nilsson J, Bergo MO 2015 Antioxidants can increase melanoma metastasis in mice. Sci Transl Med 7:308re308. 146. Aldred MA, Huang Y, Liyanarachchi S, Pellegata NS, Gimm O, Jhiang S, Davuluri RV, de la Chapelle A, Eng C 2004 Papillary and follicular thyroid carcinomas show distinctly different microarray expression profiles and can be distinguished by a minimum of five genes. J Clin Oncol 22:3531-3539. 147. Landa I, Ibrahimpasic T, Boucai L, Sinha R, Knauf JA, Shah RH, Dogan S, Ricarte-Filho JC, Krishnamoorthy GP, Xu B, Schultz N, Berger MF, Sander C, Taylor BS, Ghossein R, Ganly I, Fagin JA 2016 Genomic and transcriptomic hallmarks of poorly differentiated and anaplastic thyroid cancers. J Clin Invest 126:1052-1066. 148. Network CGAR 2014 Integrated genomic characterization of papillary thyroid carcinoma. Cell 159:676-690. 149. Zheng S, Cherniack AD, Dewal N, Moffitt RA, Danilova L, Murray BA, Lerario AM, Else T, Knijnenburg TA, Ciriello G, Kim S, Assie G, Morozova O, Akbani R, Shih J, Hoadley KA, Choueiri TK, Waldmann J, Mete O, Robertson AG, Wu HT, Raphael BJ, Shao L, Meyerson M, Demeure MJ, Beuschlein F, Gill AJ, Sidhu SB, Almeida MQ, Fragoso MCBV, Cope LM, Kebebew E, Habra MA, Whitsett TG, Bussey KJ, Rainey WE, Asa SL, Bertherat J, Fassnacht M, Wheeler DA, Hammer GD, Giordano TJ, Verhaak RGW, Network CGAR 2016 Comprehensive Pan-Genomic Characterization of Adrenocortical Carcinoma. Cancer Cell 30:363. 150. Song YS, Jung CK, Jung KC, Park YJ, Won JK 2017 Rare Manifestations of Anaplastic Thyroid Carcinoma: the Role of BRAF Mutation Analysis. J Korean Med Sci 32:1721-1726. 151. Besic N, Gazic B 2013 Sites of metastases of anaplastic thyroid carcinoma: autopsy findings in 45 cases from a single institution. Thyroid 23:709-713. 152. Weber F, Shen L, Aldred MA, Morrison CD, Frilling A, Saji M, Schuppert F, Broelsch CE, Ringel MD, Eng C 2005 Genetic classification of benign and malignant thyroid follicular neoplasia based on a three-gene combination. J Clin Endocrinol Metab 90:2512-2521. 153. Termini L, Fregnani JH, Boccardo E, da Costa WH, Longatto-Filho A, Andreoli MA, Costa MC, Lopes A, da Cunha IW, Soares FA, Villa LL, Guimarães GC 2015 SOD2 immunoexpression predicts lymph node metastasis in penile cancer. BMC Clin Pathol 15:3. 154. Ye H, Wang A, Lee BS, Yu T, Sheng S, Peng T, Hu S, Crowe DL, Zhou X 2008 Proteomic based identification of manganese superoxide dismutase 2 (SOD2) as a

120

metastasis marker for oral squamous cell carcinoma. Cancer Genomics Proteomics 5:85-94. 155. Loo SY, Hirpara JL, Pandey V, Tan TZ, Yap CT, Lobie PE, Thiery JP, Goh BC, Pervaiz S, Clément MV, Kumar AP 2016 Manganese Superoxide Dismutase Expression Regulates the Switch Between an Epithelial and a Mesenchymal-Like Phenotype in Breast Carcinoma. Antioxid Redox Signal 25:283-299. 156. Chen PM, Wu TC, Shieh SH, Wu YH, Li MC, Sheu GT, Cheng YW, Chen CY, Lee H 2013 MnSOD promotes tumor invasion via upregulation of FoxM1-MMP2 axis and related with poor survival and relapse in lung adenocarcinomas. Mol Cancer Res 11:261-271. 157. Kamarajugadda S, Cai Q, Chen H, Nayak S, Zhu J, He M, Jin Y, Zhang Y, Ai L, Martin SS, Tan M, Lu J 2013 Manganese superoxide dismutase promotes anoikis resistance and tumor metastasis. Cell Death Dis 4:e504. 158. Miar A, Hevia D, Muñoz-Cimadevilla H, Astudillo A, Velasco J, Sainz RM, Mayo JC 2015 Manganese superoxide dismutase (SOD2/MnSOD)/catalase and SOD2/GPx1 ratios as biomarkers for tumor progression and metastasis in prostate, colon, and lung cancer. Free Radic Biol Med 85:45-55. 159. Chakrabarti SK, Ghosh S, Banerjee S, Mukherjee S, Chowdhury S 2016 Oxidative stress in hypothyroid patients and the role of antioxidant supplementation. Indian J Endocrinol Metab 20:674-678. 160. Köhrle J 2013 Selenium and the thyroid. Curr Opin Endocrinol Diabetes Obes 20:441-448. 161. Mendelsohn AR, Larrick JW 2014 Paradoxical effects of antioxidants on cancer. Rejuvenation Res 17:306-311. 162. Supabphol A, Supabphol R 2012 Antimetastatic potential of N-acetylcysteine on human prostate cancer cells. J Med Assoc Thai 95 Suppl 12:S56-62. 163. Deng M, Brägelmann J, Kryukov I, Saraiva-Agostinho N, Perner S 2017 FirebrowseR: an R client to the Broad Institute's Firehose Pipeline. Database (Oxford) 2017. 164. Marsh DJ, Coulon V, Lunetta KL, Rocca-Serra P, Dahia PL, Zheng Z, Liaw D, Caron S, Duboue B, Lin AY, Richardson AL, Bonnetblanc JM, Bressieux JM, Cabarrot-Moreau A, Chompret A, Demange L, Eeles RA, Yahanda AM, Fearon ER, Fricker JP, Gorlin RJ, Hodgson SV, Huson S, Lacombe D, Eng C, et al. 1998 Mutation spectrum and genotype-phenotype analyses in Cowden disease and Bannayan-Zonana syndrome, two hamartoma syndromes with germline PTEN mutation. Hum Mol Genet 7:507-515.

121