Novel Mechanisms of Pten Dysfunction in Pten Hamartoma Tumor Syndromes

NOVEL MECHANISMS OF PTEN DYSFUNCTION IN PTEN HAMARTOMA TUMOR SYNDROMES

DISSERTATION

Presented in Partial Fulfillment of the Requirements for

the Degree Doctor of Philosophy in the Graduate

School of The Ohio State University

Marcus G. Pezzolesi, B.S.

*****

The Ohio State University 2008

Dissertation Committee: Approved by Professor Allan J. Yates, Advisor

Professor Charis Eng, Co-Advisor ______Professor Wolfgang Sadee Advisor Integrated Biomedical Science Professor Michael C. Ostrowski Graduate Program

Professor Lawrence S. Kirschner

Professor Lei Shen

ABSTRACT

Phosphatase and tensin homolog deleted on chromosome ten (PTEN) encodes a

tumor suppressor phosphatase frequently mutated in both sporadic and heritable forms of

human cancer. Germline mutations in PTEN are associated with a number of heritable cancer syndromes referred to as the PTEN hamartoma tumor syndromes (PHTS) and includes both Cowden syndrome (CS) and Bannayan-Riley-Ruvalcaba syndrome

(BRRS). Data from our laboratory suggests that alternate mechanisms of PTEN deregulation are likely to, at least in part, contribute to dysfunction in patients with these syndromes, particularly in those for whom germline mutations have yet to be identified.

To better understand the mechanism(s) underlying dysregulation of PTEN in these syndromes, we employed a series of genetic and biochemical approaches aimed at investigating novel mechanisms involved in the regulation and deregulation of PTEN.

Using a haplotype-based approach, we identified specific haplotypes and rare alleles within the PTEN locus that contribute to disease susceptibility and the phenotypic complexity of this syndrome. Within a haplotype block associated with PTEN-mutation negative patients, we identified a canonical E-box sequence located upstream of PTEN’s minimal promoter.

We also investigated the role of microRNAs (miRNAs) in regulating PTEN and in PHTS. We show that miR-519e, a miRNA computationally predicted to target PTEN,

specifically interacts with the gene’s 3’ untranslated region (UTR) and down-regulates endogenous PTEN expression in vitro. Subsequently, we show that miR-19a and miR-21, two miRNAs previously shown to target and repress PTEN protein levels, are differentially expressed in CS patients, irrespective of the PTEN mutation status. Our data suggest that these miRNA likely contribute to the phenotypic variability commonly seen in PHTS.

The findings presented in this dissertation contribute significantly to our understanding of the pathogenesis of PHTS in patients in whom traditional screening methodologies have been unable to uncover a genetic cause. We further show that alternate mechanisms of PTEN dysfunction contribute to its deregulation and also to the variable phenotypic spectrum observed in PHTS. It is our hope that these data may lead to improved diagnostic measures and better predictive testing, and ultimately enable

PHTS patients and their family members access to better personalized care.

iii

DEDICATION

To my beautiful wife, Melissa, who has supported me throughout our Ohio adventure, and to my two sons, Quentin and Avery, all three of whom I love so much. And in memory of my father, Gerald G.N. Pezzolesi, who is somewhere smiling proudly.

ACKNOWLEDGMENTS

I would like to thank Dr. Charis Eng for all of her support and mentorship

throughout my graduate training. The time spent I’ve in her laboratory has been

invaluable. She have been instrumental in helping me to think like a scientist and I am

truly grateful for all of the opportunities she has afforded me during this time and for her

unending patience. I also want to thank the members of my dissertation committee, Drs.

Allan J. Yates, Wolfgang Sadee, Michael C. Ostrowski, Lawrence S. Kirschner, and Lei

Shen for their many helpful discussions and for the time that each has devoted in helping

to further my scientific education and my research. I especially would like to thank Dr.

Yates who, despite his many other commitments, has graciously served as my co-advisor over the past two and a half years.

I thank the many member of the Eng laboratory, both past and present, for their scientific support and for their friendships. I especially would like to thank Drs. Kevin

M. Zbuk, Kristin A. Waite, and Attila Patocs and Ms. Nita Williams, Ms. Rosemary

Teresi, Ms. Patricia Kessler, and Mr. Todd Romigh.

I would also like to thank Drs. Andrzej S. Krolewski and James H. Warram for all of their support over many years, without which I most certainly would not be where I am today.

Finally, and most importantly, to wife, Melissa, and my two sons, Quentin and

Avery, who have all inspired me throughout my studies. Melissa, I thank you for your continuous support and for your love. You have always been there for me and I am forever grateful. I look forward to enjoying the rest of our lives together. Quentin and

Avery, you have both inspired me in life and are both so precious to me; my great hope is that life brings you as much happiness and joy as you have brought me.

VITA

March 6, 1972.……………………………….. Born – Leominster, Massachusetts

1994.…………………...…………………….. B.S. Exercise Science, University of Massachusetts, Amherst.

2003 – 2005…………………………………...Graduate Research Associate The Ohio State University.

2005 – 2008………………………………….. Pre-Doctoral Research Fellow Genomic Medicine Institute, Lerner Research Institute, Cleveland Clinic.

PUBLICATIONS

Research Publications

1. Differential expression of PTEN-targeting microRNAs, miR-19a and miR-21, in Cowden syndrome. Pezzolesi MG, Platzer P, Waite KA, Eng C (In Preparation)

2. High-density SNP genome wide linkage scan for susceptibility genes for diabetic nephropathy in type 1 diabetes: Discordant sib-pair approach. Rogus JJ, Poznik GD, Pezzolesi MG, Smiles AM, Dunn JS, Walker W, Wanic K, Canani LH, Araki H, Makita J, Warram JH, Krolewski AS (Diabetes. Submitted)

3. Cowden syndrome-affected patients with PTEN promoter mutations demonstrate abnormal protein Translation. Teresi RE, Zbuk KM, Pezzolesi MG, Waite KA, Eng C (Am J Hum Genet. Oct:81(4):756-767)

4. Comparative genomic and functional analyses reveal a novel cis-acting PTEN regulatory element as a highly conserved functional E-box motif deleted in Cowden syndrome. Pezzolesi MG, Zbuk KM, Waite KA, Eng C (Hum Mol Genet. 2007 May:16(9):1058-1071)

vii

5. Mutation-positive and mutation-negative patients with Cowden and Bannayan- Riley-Ruvalcaba syndromes associated with distinct 10q haplotypes. Pezzolesi MG, Li Y, Zhou XP, Pilarski R, Shen L, Eng C (Am J Hum Genet. 2006 Nov;79(5):923- 934)

6. Examination of PPP1R3B as a candidate gene for the type 2 diabetes and MODY loci on chromosome 8p23. Dunn JS, Mlynarski WM, Pezzolesi MG, Borowiec M, Powers C, Krolewski AS, Doria A (Ann Hum Genet. 2006 Sep;70(Pt 5):587-293)

7. A newly identified mutation in an ipf1 binding site of the insulin gene promoter may predispose to type 2 diabetes mellitus. Malecki MT, Lebrun P, Pezzolesi M, Warram JH, Krolewski AS, Jhala US (Diabetologia 2006 Aug; 49(8):1985-1987)

8. Examination of candidate chromosomal regions for type 2 diabetes mellitus (T2DM) reveals a susceptibility locus on human chromosome 8p23.1. Pezzolesi MG, Nam M, Nagase T, Klupa T, Dunn JS, WM Mlynarski, Rich SS, Warram JH, Krolewski AS (Diabetes. 2004 Feb; 53(2):486-491)

9. Genetic modifiers of the age at diagnosis of diabetes (MODY3) in carriers of hepatocyte nuclear factor-1alpha mutations map to chromosomes 5p15, 9p22, and 14q24. Kim SH, Ma X, Klupa T, Powers C, Pezzolesi M, Warram JH, Rich SS, Krolewski AS, Doria A. (Diabetes. 2003 Aug; 52(8):2182-2186)

10. Determinants of the development of diabetes (maturity-onset diabetes of the young-3 [MODY3] in carriers of HNF-1a mutations. Klupa T, Warram JH, Antonellis A, Pezzolesi M, Nam M, Malecki MT, Doria A, Rich SS, Krolewski AS (Diabetes Care. 2002 Dec; 25(12):2292-2301)

11. A method for developing high density snp maps and its application at the type 1 angiotensin ii receptor (AGTR1) gene locus. Antonellis A, Rogus JJ, Canani LH, Makita Y, Pezzolesi MG, Nam M, Ng D, Moczulski D, Warram JH, Krolewski AS (Genomics. 2002 Mar; 79(3):326-332)

12. Further evidence for a susceptibility locus for type 2 diabetes mellitus on chromosome 20q13.1-q13.2. Klupa T, Malecki MT, Pezzolesi M, Ji L, Curtis S, Rich SS, Warram JH, Krolewski AS (Diabetes. 2000 Dec; 49(12):2212-2216)

viii

FIELDS OF STUDY

Major Field: Integrated Biomedical Science

TABLE OF CONTENTS

P a g e

ABSTRACT...... ii

DEDICATION...... iv

ACKNOWLEDGMENTS ...... v

VITA...... vii

LIST OF TABLES...... xiii

LIST OF FIGURES ...... xiv

CHAPTER 1 INTRODUCTION ...... 1

1.1 COWDEN SYNDROME AND PTEN ...... 1

1.2 PTEN BIOCHEMISTRY...... 4

1.3 MURINE MODELS OF PTEN INACTIVATION ...... 8

1.4 SOMATIC AND GERMLINE INACTIVATION OF PTEN ...... 10

1.5 PTEN HAMARTOMA TUMOR SYDROMES...... 13

1.6 HYPOTHESIS AND OVERVIEW OF CHAPTERS 2-5 ...... 14

CHAPTER 2 MUTATION-POSITIVE AND MUTATION-NEGATIVE COWDEN AND BANNAYAN-RILEY-RUVALCABA SYNDROME PATIENTS ASSOCIATED WITH DISTINCT 10q-HAPLOTYPES ...... 17

2.1 INTRODUCTION ...... 17

2.2 MATERIAL AND METHODS...... 19

2.3 RESULTS ...... 27

2.4 DISCUSSION...... 45

CHAPTER 3 COMPARATIVE GENOMIC AND FUNCTIONAL ANALYSES REVEAL A NOVEL CIS-ACTING PTEN REGULATORY ELEMENT AS A HIGHLY CONSERVED FUNCTIONAL E-BOX MOTIF DELETED IN COWDEN SYNDROME ...... 52

3.1 INTRODUCTION ...... 52

3.2 MATERIAL AND METHODS...... 54

3.3 RESULTS ...... 62

3.4 DISCUSSION...... 82

CHAPTER 4 miR-519e NEGATIVELY REGULATES THE TUMOR SUPPRESSOR PHOSPHATASE PTEN IN HUMAN CANCER CELLS ...... 89

4.1 INTRODUCTION ...... 89

4.2 MATERIAL AND METHODS...... 91

4.3 RESULTS ...... 95

4.4 DISCUSSION...... 107

CHAPTER 5 DIFFERENTIAL EXPRESSION OF PTEN-TARGETING MICRORNAS, miR-19a AND miR-21, IN COWDEN SYNDROME ...... 113

5.1 INTRODUCTION ...... 113

5.2 MATERIAL AND METHODS...... 115

5.3 RESULTS ...... 121

5.4 DISCUSSION...... 135

CHAPTER 6 DISCUSSION AND FUTURE DIRECTIONS...... 140

LIST OF REFERENCES...... 154

xii

LIST OF TABLES

Table Page

1.1 International Cowden Syndrome Consortium operational diagnostic criteria...... 3

2.1 Primer sequences and genotyping methodologies for all SNP and quantitative real-time PCR reactions...... 22

2.2 Characteristics of 30 SNP panel...... 28

2.3 Summary of SNP allele frequency data for control sample and PHTS patient populations...... 30

2.4 Haplotype blocks across the PTEN locus...... 39

2.5 Extended haplotypes for all 30 SNPs across the PTEN locus...... 42

2.6 Comparative haplotype analysis...... 44

3.1 Mutant oligonucleotide competitor probes/sequence of mutant reporter constructs...... 68

5.1 Summary of control and patient clinical features...... 116

5.2 Clinical features, PTEN protein, and relative PTEN, miR-19a, and miR-21 expression among PTEN mutation-positive patients...... 122

5.3 Summary of comparisons of relative PTEN Transcript, miR-19a, and miR-21 expression levels...... 128

xiii

LIST OF FIGURES

Figure Page

1.1 Schematic representation of the PI3K/Akt pathway ...... 7

1.2 Schematic representation of germline PTEN promoter mutations and polymorphisms found in probands with CS ...... 12

2.1 Schematic diagram of the PTEN locus and SNPs included in the current analysis ...... 27

2.2 Summary of SNP allele frequency P-values for PHTS patient population groups versus control population...... 34

2.3 Hemizygous PTEN deletion analysis ...... 35

2.4 GOLD plot of pairwise LD between 30 SNPs ...... 37

3.1 Comparative genomic analysis reveals a highly conserved E-box upstream of the PTEN locus ...... 63

3.2 Identification of a novel USF-specific DNA-protein interaction upstream of the PTEN promoter...... 66

3.3 The -2262 to -2151 conserved region is involved in transcriptional activation ...... 70

3.4 USF proteins, and not Myc/Max, bind to the -2262 to -2151 conserved region...... 75

3.5 USF1 specifically binds to the -2181 to -2176 PTEN E-box element.....77

3.6 Myc/Max does not bind to the -2181 to -2176 PTEN E-box element.....78

3.7 Identification of a functional hemizygous germline deletion upstream of the PTEN coding sequence...... 80

xiv

4.1 Four miRNAs computationally predicted to target PTEN’s 3’UTR ...... 96

4.2 miR-519e inhibits luciferase reporter activity ...... 100

4.3 miR-519e-mediated luciferase reporter activity is specific to the miR-519e seed site...... 101

4.4 Endogenous PTEN expression is decreased by miR-519e ...... 103

4.5 Anti-miR-519e treatment restores miR-519e-induced PTEN down- regulation...... 105

4.6 Both miR-519e and miR-519d decrease endogenous PTEN expression ...... 108

5.1 Schematic diagram of PTEN’s mRNA sequence ...... 118

5.2 Relative expression values for PTEN transcript, miR-19a, and miR-21 among PTEN mutation-positive samples ...... 129

5.3 Relative expression values for PTEN transcript, miR-19a, and miR-21 among control, PTEN mutation-positive, and PTEN mutation-negative patient samples ...... 132

6.1 Alignment of novel PTEN isoforms...... 144

6.2 Schematic of sequence variations identified in GLTSCR2...... 151

6.3 Mechanisms of PTEN dysfunction in PHTS...... 152

CHAPTER 1

INTRODUCTION

1.1 COWDEN SYNDROME AND PTEN

Cowden syndrome, originally termed Cowden’s disease ([MIM 158350]), was

first described in 1963 by Lloyd and Dennis upon presentation of a 20-year-old female

patient at the Youngstown Hospital Association in Youngstown, Ohio (1). At the time of

her visit, Rachel Cowden, for whom the disease was later named, was described as

having a ‘new symptom complex with multiple system involvement’. Several clinical

features rarely observed in a single patient formed the basis of this unusual complex.

Among the manifestations noted were papillomatosis of the lips and oral pharynx,

multiple thyroid adenomas, advanced fibrocystic breast disease, and early age of onset breast cancer, from which she died several years later. Additionally, several developmental anomalies, including both craniofacial and central nervous system

abnormalities, were also present. Milder forms of disease were also noted in the patient’s

mother, sister, and two maternal aunts.

Despite the lack of any gross chromosomal aberrations, Lloyd and Dennis

speculated that Ms. Cowden’s multi-symptom complex, coupled with her family history,

suggested heritability. Furthermore, they proposed that her condition was likely the

consequence of a single genetic defect.

Cowden syndrome (CS), is a rarely recognized, under-diagnosed disorder, with

an estimated incidence of 1 in 200,000 individuals, likely an underestimate (2). CS is

characterized by multiple hamartomatous lesions, or benign tumor-like masses composed

of malformed tissue, and a high risk of breast and thyroid cancers (3-5). Patients meeting the full operational diagnostic criteria established by the International Cowden Syndrome

Consortium are considered to be classic CS patients (Table 1.1). In addition to those who meet these criteria, a number of CS-like (CSL) patients share some features of CS, however, they fail to meet the full diagnostic criteria. The benign features associated with

CS include various facial and oral mucocutaneous lesions (including trichlemmomas and papillomatous papules), acral and plantar keratoses, lipomas, fibromas, thyroid abnormalities, genito-urinary tumors or malformations (including uterine fibroids), and fibrocystic disease of the breast. Macrocephaly and the adult presentation of Lhermitte-

Duclos disease (LDD, hamartomatous gangliocytoma of the cerebellum) are also major phenotypic features of CS.

Importantly, patients with CS and CSL are also at an increased risk of developing malignancy, particularly of the breast, thyroid (especially follicular thyroid carcinoma), and endometrium. The estimated lifetime risk associated with each of these component cancers in CS is approximately 30-50%, 10%, and 5-10%, respectively, compared to

11%, less than 1%, and 2% in the general population (4). However, due the variable phenotypic spectrum associated with this disorder, its accurate diagnosis is often quite

Pathognomonic: Major Criteria: Minor Criteria: Adult presentation of Breast cancer, thyroid Other thyroid lesions Lhermitte-Duclos disease cancer (especially follicular (adenoma, multinodular Mucocutaneous lesions: thyroid carcinoma), goiter), developmental Facial trichilemmomas, macrocephaly (≥ 97th delay/mental retardation (≤ acral keratoses, percentile), endometrial 75), hamartomatous papillomatous papules cancer intestinal polyps, fibrocystic breast disease, lipomas, fibromas, genito-urinary tumors or malformations Operational Diagnosis - Individual: 1. Pathognomonic mucocutaneous lesions alone if there are: • Six or more facial papules, of which three or more must be trichilemmoma, or • Cutaneous facial papules and oral mucosal papillomatosis, or • Oral mucosal papillomatosis and acral keratoses, or • Six or more palmoplantar keratoses 2. Two major criteria, one must be macrocephaly 3. One major and ≥ three minor criteria 4. ≥ four minor criteria Operational Diagnosis - Individuals with family members diagnosed with CS: 1. A pathognomonic mucocutaneous lesion 2. Any one major criteria with or without minor criteria 3. Two minor criteria 4. History of BRRS * Adapted from NCCN Practice Guidelines in Oncology – v.1.2007 and (3, 4).

Table 1.1. International Cowden Syndrome Consortium operational diagnostic criteria.

challenging. Because of the increased risk of malignancy associated with CS, the proper diagnosis of patients at risk of developing neoplasia is paramount to ensure their appropriate management and surveillance.

Using a genome-wide linkage analysis approach, our lab used 12 highly informative CS families, with 40 affected individuals, to localize the susceptibility locus to a 5 centi-Morgan (cM) region on chromosome bands 10q22-23(6). Interestingly,

evidence for linkage in all 12 families mapped solely to the 10q22-23 region, suggesting

that this region of the genome contained the major CS disease locus, while providing

little evidence in support of genetic heterogeneity.

Although no obvious candidate genes were initially evident within the CS critical

interval, previous reports that the 10q22-23 region was a frequent site loss of

heterozygosity (LOH) in several sporadic cancers, including both follicular thyroid and

uterine carcinoma – two CS component cancers, supported its potential role in CS (7).

Subsequent positional cloning and biochemical analyses quickly led to the identification

and characterization of a novel tumor suppressor gene termed phosphatase and tensin homolog deleted on chromosome 10 (PTEN), also known as mutated in multiple advanced cancers 1 (MMAC1) and transforming growth factor-β [TGF-β] regulated and epithelial cell-enriched phosphatase 1 (TEP1), within the 10q23 region (8-10). Shortly after its isolation, our laboratory identified germline mutations in PTEN in CS families genetically linked to this locus (11).

1.2 PTEN BIOCHEMISTRY

Since identifying PTEN as the gene altered in CS, much research has been done

to elucidate its function. It is now well established that this nine-exon gene encodes a

dual-specificity tumor suppressor phosphatase, able to dephosphorylate both protein and lipid substrates (12, 13). The 403-amino acid PTEN protein is composed of two major functional domains, an N-terminal domain, formed by amino acids 1-185 (encoded by exons 1-6), and a C-terminal domain, formed by amino acids 186-402 (encoded by exons

6-9) (14). PTEN’s C-terminal region contains its lipid-binding C2 domain, as well as two

PEST domains and a PDZ domain, regions thought to be important in maintaining PTEN

stability and in facilitating protein-protein interactions (14, 15). PTEN’s enzymatic activity is ascribed to its phosphatase domain located in the protein’s N-terminal region.

The catalytic core motif, CKAGKGR, within this domain is defined by amino acids 124 through 130. As will be discussed later, germline, as well as somatic, mutations have been found to cluster within PTEN’s phosphatase domain and, more specifically, within this core motif, thereby substantiating its fundamental role in both biology and disease.

PTEN exerts a large part of its role as a tumor suppressor by negatively regulating two pathways integral to cellular proliferation and survival; the phosphatidylinositol-3 kinase (PI3K)/Akt pathway and the mitogen-activated protein kinase (MAPK) pathway.

As a lipid phosphatase, cytoplasmic PTEN is able to oppose PI3K’s activity by dephosphorylating the D-3 position of phosphatidylinositol-3,4,5-triphosphate (PIP3) and converting this to phosphatidylinositol-4,5-bisphosphate (PIP2) (Fig. 1.1) (12). PIP3 functions to recruit Akt/protein kinase B, a protein essential for both cell survival and proliferation, to the plasma membrane by directly interacting with its pleckstrin- homology domain. Once at the plasma membrane, 3-phosphoinostide-dependent-kinase-

1 (PDK1) phosphorylates and activates Akt (16). By antagonizing PI3K, PTEN limits the phosphorylation and activation of Akt and is thereby able to negatively regulate progression of the cell cycle and induce apoptosis.

The survival signals mediated by phosphorylated Akt inhibit several key components of the pro-apoptotic pathway, while activating others, resulting in increased cellular growth, survival, and proliferation. Among the proteins directly affected by

Akt’s kinase activity are BCL2 antagonist of cell death (BAD), the forkhead transcription

factor FOXO1, mouse double minute 2 (MDM2), glycogen synthase kinase-3β (GSK3β)

and mammalian target of rapamycin (mTOR) (17). mTOR can also be indirectly

activated by Akt through its phosphorylation and destabilization of the tuberous sclerosis

complex (TSC). Phospho-Akt’s direct phosphorylation of MDM2 negatively regulates

the tumor suppressor p53(18). Interestingly, as PTEN is a transcriptional target of p53,

Akt’s indirect reduction of p53 is thus able to promote its own activity by negatively regulating PTEN’s transcription (19). Additionally, our laboratory has recently shown that PTEN is similarly able to autoregulate its own expression, through the formation of a

PTEN:p53 complex which stabilizes p53 and subsequently facilitates its transactivation of PTEN (20).

Although not as well studied as its lipid phosphatase activity, PTEN’s protein

phosphatase activity allows it to regulate the MAPK cell survival pathway (21, 22).

PTEN perturbs this signaling cascade by inhibiting the formation of intermediate

complexes required for MAPK activation. By directly dephosphorylating Shc, PTEN

prevents formation of the Shc/Grb-2/Sos complex. Disruption of this complex prevents

activation of Ras/Raf/Mek and, as a downstream consequence, reduces activation of

MAPK. PTEN can also inhibit MAPK independent of Shc, by inhibiting the

phosphorylation of the insulin-receptor substrate 1 (IRS-1) docking protein (23). This

action prevents IRS-1 from complexing with the Grb-2 and Sos adapter proteins and,

similar to PTEN’s dephosphorylation of Shc, subsequently results in MAPK’s

inactivation. PTEN’s protein phosphatase activity has also been shown to inhibit the

focal adhesion kinase (FAK) pathway, suggesting a role in modulating cell-cell

Figure 1.1. Schematic representation of the PI3K/Akt pathway. * Adapted from (17).

interactions (21, 24). As well, PTEN has been shown to modulate G1 arrest by downregulating cyclin D1, a positive regulator of cell cycle progression (25).

PTEN’s protein phosphatase activities, in addition to those mediated by its actions as a lipid phosphatase, play a vital role in tumor suppression by negatively regulating both cell growth and survival. Interestingly, our laboratory has recently shown 7

that bifurcation of these activities is regulated, in part, by PTEN’s sub-cellular

localization (26). While cytoplasmic PTEN negatively regulates Akt and upregulates

p27, thereby inducing apoptosis, nuclear PTEN mediates cell cycle arrest through the

down-regulation of cyclin D1 and phospho-MAPK (26, 27). Although deregulation of

PTEN’s nuclear-cytoplasmic partitioning has not been extensively studied in CS, it is likely that its differential compartmentalization could contribute to pathogenesis in this syndrome.

1.3 MURINE MODELS OF PTEN INACTIVATION

The loss of PTEN function protects cells from the various apoptotic signals

mediated by its negative regulation of the PI3K/Akt and MAPK pathways. When left

unchecked, this loss can drive excess cellular survival and proliferation, become oncogenic, and ultimately result in tumorigenesis. The importance of normal PTEN in this process has been demonstrated through investigations involving Pten knockout mice.

Three separate groups have generated null Pten mutations targeting the protein’s phosphatase domain (28-31). Each study reported that Pten -/- mice are not viable, as all embryos reach lethality by embryonic day 9.5. Mice with a heterozygous knockout of

Pten, on the other hand, do survive and the inactivation of one Pten allele in these mice is sufficient for tumor formation. Pten +/- mice develop neoplasia in multiple organs, including the thyroid (both follicular and papillary thyroid carcinoma), prostate, liver, and colon. These mice, despite not developing the hamartomatous lesion that are characteristic of CS, do share some features commonly observed in this disorder. Pten-

knockout mice develop hyperplastic changes in the skin epidermis, specifically focal

acanthosis, although this is often a precursor to skin cancer in mice (a feature not seen in

CS), and in the endometrium as a result of this haploinsufficiency (28, 29). Suzuki et al.

also reported the presence of hamartomatous polyps in the intestine, a minor criterion

observed in patients with CS, primarily affecting the colon in the majority of mice they

examined (30). A subsequent study by this same group found that while younger Pten

+/- mice did not exhibit features seen in CS, older mice (> 6 months of age) begin to

resemble this phenotype (31). Particularly, these mice develop breast tumors,

endometrial hyperplasia, endometrial cancer, and hamartomatous tumors of the

gastrointestinal tract. Interestingly, as noted in patients with CS, there appears to be vast genotype-phenotype variability among the various mouse models examined. These differences may be due, in part, to the different mutations engineered in each model.

However, additional differences in the genetic backgrounds of the mice used in these studies suggest that other loci may behave as modifiers of the observed phenotypes and, thereby, contribute to this variability. In a recent study aimed at investigating these issues, Freeman et al. demonstrated that the onset and incidence of tumor formation were indeed highly dependent on the genetic background of these mice (32). Moreover, they found that genetic background, including potential modifier genes, and not the specific mutation introduced to the mice, was the major contributor to the observed phenotypic variation. Therefore, it is quite likely that similar differences among individual patients contribute to the difference observed in human disease.

1.4 SOMATIC AND GERMLINE INACTIVATION OF PTEN

In humans, several different mechanisms contribute to impaired PTEN function.

Somatic alterations, including point mutations and LOH at the PTEN locus, have been

noted in several sporadic cancers, including the malignancies commonly associated with

CS. Specifically, somatic inactivation of PTEN has been noted in glioblastoma,

melanoma, breast, prostate, lung, cervical, bladder, renal cell, endometrial, ovarian,

thyroid, pancreatic, and head and neck squamous cell carcinoma (33-46). As alluded to

earlier, mutations in several of these sporadic cancers have been shown to cluster within

PTEN (34, 35, 39, 41-45). More specifically, these primarily occur within exon 5, the

region encoding PTEN’s catalytic core motif, as well as in exons 7 and 8. Epigenetic

silencing, resulting from the hypermethylation of PTEN’s promoter, has also been shown

to play a role in its inactivation in a subset of human tumors (47-49).

The primary mechanism of PTEN inactivation in CS involves point mutations

within PTEN’s coding sequence. To date, 80% of individuals with this disease have been

found to harbor mutations within the gene’s first eight exons, with the notable exception

of exon 9 (2, 50). Approximately two-thirds of the mutations identified in CS localize to

exons 5, 7, and 8, with more than 40% of these occurring in exon 5 alone (51).

Interestingly, exon 5 makes up only 20% of PTEN’s entire coding sequence, despite its

increased mutability. Moreover, data that demonstrate a strong correlation between

missense mutations within the phosphatase core motif contained in exon 5, including

those 5’ of this region, and disease severity in CS provide a compelling link between

PTEN’s curious mutation spectrum and the resultant clinical phenotypes observed in CS

patients (52).

In addition to pathogenic alterations within PTEN’s coding sequence, mutations

within its 600-basepair (bp) promoter region, located at nucleotide position -1344 to -745

relative to its translation start site, have recently been observed in a subset of CS patients

who lack exonic changes (Fig. 1.2) (53, 54). This study revealed that approximately 10%

of PTEN mutation-negative CS patients harbor mutations within this region. Although

the mutations reported in this study did not alter either the p53 or early growth response-1

(Egr-1) binding site, two were adjacent to or within putative Sp1 binding sites (19, 55).

Furthermore, carriers of these PTEN promoter mutations were shown to have reduced

PTEN expression, as well as a lower molecular weight protein band and protein laddering, suggesting that these mutations contribute to reduced protein stability and may also alter the site of translation initiation. A subsequent study investigated a small series of CS patients previously found to be carriers of mutations within or near PTEN’s p53

binding site (20). Among these individuals, 1 carrier of a mutation located within one of

the core p53-binding half-sites showed the greatest reduction of PTEN expression, while

two other mutations observed in these patients were shown to reduce p53’s affinity for

the altered binding sites. Moreover, as noted with other PTEN promoter mutations,

patients with alterations either within or adjacent to the p53-binding site exhibited

significantly less stable PTEN compared to controls or patients with intragenic PTEN

mutations. Most recently, our laboratory has provided additional support linking PTEN

promoter mutations with abnormal PTEN expression (56). In this study, 3 mutations, 2

of which lie within PTEN’s minimal promoter region (nucleotide position -958 to -821),

were shown to alter the normal conformation of PTEN’s mRNA secondary structure,

resulting in the inhibition of its efficient translation and a decrease in PTEN protein

Figure 1.2. Schematic representation of germline PTEN promoter mutations and polymorphisms found in probands with CS (50).

expression. The identification of functional mutations within PTEN’s promoter, resulting in aberrant transcription, translation, or protein stability, increases the prevalence of detectable germline PTEN mutations in patients with CS to approximately 85%.

1.5 PTEN HAMARTOMA TUMOR SYNDROMES

CS, along with Bannayan-Riley-Ruvalcaba syndrome (BRRS [MIM 153480]),

Proteus syndrome (PS [MIM 176920]), and Proteus-like syndrome (PLS) represent a

collection of related, yet phenotypically diverse, disorders that share several overlapping

clinical features. These four related syndromes also share a common genetic etiology

and, hence, are collectively referred to as the PTEN hamartoma tumor syndromes (PHTS)

(57).

Like CS, the other component PHTS syndromes are characterized by multiple

hamartomatous lesions. BRRS, a rare congenital disorder, includes several characteristic

features present in patients with CS, including macrocephaly, lipomas, hemangiomas, and

gastrointestinal polyposis (58). Developmental delay and mental retardation are also

commonly associated with BRRS. Male BRRS patients frequently present with

pigmented maculae of the glans penis. Germline PTEN mutations have been identified in

the majority of patients diagnosed with BRRS, as approximately 65% harbor pathogenic

genetic alterations in this gene (59). Interestingly, identical pathogenic PTEN mutations

have been observed among patients with these two seemingly similar, yet divergent

disorders (57). Similar to the previously described studies involving Pten-deficient mice,

this imprecise genotype-phenotype correlation suggests that additional genetic factors

and/or mechanisms likely contribute to the development, progression, and severity of

disease in these patients.

PS, a third, seemingly unrelated disorder also presents with clinical features

overlapping those seen in both CS and BRRS. In addition to hamartomatous overgrowths, features of PS also include macrocephaly, lipomas, and hemangiomas (60).

Epidermal naevi, hyperostosis, congenital malformations, partial gigantism of the hands or feet, limb asymmetry, and craniofacial disfigurement are among various other manifestations associated with PS. Up to 20% of patients with PS have been found to have germline PTEN mutations (61, 62). Furthermore, germline PTEN mutations have

also been identified in approximately 50% of patients with a less severe PS phenotype, referred to as PLS.

1.6 HYPOTHESIS AND OVERVIEW OF CHAPTERS 2-5

While heritable PTEN mutations have been identified in a significant number of

patients diagnosed with PHTS, for many patients with these related syndromes, the cause

of disease remains unknown. In the case of CS and BRRS, 15% and 35% of patients,

respectively, who meet diagnostic criteria have disease without a recognized genetic

cause. Moreover, PTEN mutations have not been identified in the majority of CSL

patients, as approximately >80% of patients with this phenotype do not appear to have

mutations within PTEN or its promoter (Eng, unpublished observation). The lack of

detectable germline alterations in PTEN in these CS, CSL, and BRRS patients suggests

the need to investigate alternate mechanisms of PTEN dysfunction.

In order to investigate potential mechanisms of PTEN deregulation in patients

with CS, CSL, and BRRS, our laboratory has been meticulously documenting PTEN

dysfunction in patients with these disorders both with and without detectable PTEN mutations. Through these efforts, we have identified decreased PTEN expression in a large number of patients found to be negative for both PTEN mutations and deletions

(Waite and Eng, unpublished observation). Additionally, variable PTEN levels have also been observed in CS, CSL, and BRRS patients sharing identical PTEN mutations, suggesting that alternate mechanisms are likely to, at least in part, contribute to dysfunction in these patients. Based on these observations and given that the etiology of disease remains unknown for many CS, CSL, and BRRS patients, a general hypothesis in our laboratory has been that alternate mechanisms of PTEN dysfunction underlie disease susceptibility in patients lacking germline PTEN mutations. Efforts to explore these molecular mechanisms are at the heart of our research and the overall objective of the dissertation presented herein.

In Chapters 2 through 5, we continued investigating novel mechanisms of PTEN dysfunction in PHTS patients. In Chapter 2, we describe a haplotype-based approach aimed at investigating the association of specific genomic regions of the PTEN locus in

CS and BRRS. We found this locus to be characterized by 3 distinct haplotype blocks

and that the distribution of these blocks differed significantly among PHTS patients and

controls. More specifically, PTEN mutation-negative patients were found to be strongly

associated with a haplotype block spanning a region upstream of PTEN and including its

first intron. In Chapter 3, we employed a comparative genomic approach combined with

molecular biology techniques to identify a highly conserved region upstream of PTEN’s

promoter. This region contained a canonical E-box motif located more than 800bp upstream of the PTEN core promoter and was shown to be recognized by upstream stimulatory factor 1 (USF1) and USF2. Moreover, this region was found to be deleted in a subset of CS and CSL PTEN mutation-negative patients. In Chapter 4, we examined the functionality of a subset of microRNAs (miRNAs) computationally predicted to target and repress PTEN. Through this investigation, we identified one miRNA, miR-

519e, that specifically interacts with PTEN’s 3’ untranslated region (UTR) and down- regulates endogenous PTEN expression in vitro. In Chapter 5, we characterized the relative expression of two miRNAs previously shown to down-regulate PTEN, miR19a and miR-21, in PTEN mutation-positive and PTEN mutation-negative PHTS patients compared to control subjects. Interestingly, we observed that miRNA levels appear to be correlated with differential PTEN protein expression among PTEN mutation-positive patients sharing identical truncating mutations. Our data also show that miR-19a levels are over-expressed in both PHTS patient cohorts relative to controls, regardless of their mutation status, and that miR-21 is over-expressed in PTEN mutation-positive patients.

Taken together, the findings presented in this dissertation contribute significantly to our understanding of the pathogenesis of PHTS in patients where traditional screening methodologies have been unable to uncover a genetic cause. It is our hope that the findings from our examination of alternate mechanisms of PTEN dysfunction in CS and

BRRS may lead to improved diagnostic measures and better predictive testing in these under-recognized syndromes.

CHAPTER 2

MUTATION-POSITIVE AND MUTATION-NEGATIVE COWDEN AND BANNAYAN-RILEY-RUVALCABA SYNDROME PATIENTS ASSOCIATED WITH DISTINCT 10q-HAPLOTYPES

2.1 INTRODUCTION

Phosphatase and tensin homolog deleted on chromosome ten (PTEN [MIM

601728]) (also known as mutated in multiple advanced cancers 1 (MMAC1) and tensin-

like phosphatase 1 (TEP1)) encodes a tumor suppressor phosphatase that signals down

the phosphoinositol-3-kinase (PI3K)/AKT pathway, effecting apoptosis and cell cycle

arrest (63, 64). Germline PTEN mutations are primarily associated with a number of

apparently clinically distinct heritable cancer syndromes jointly referred to as the PTEN

hamartoma tumor syndrome (PHTS) (57). These include Cowden syndrome (CS [MIM

158350]), Bannayan-Riley-Ruvalcaba syndrome (BRRS [MIM 153480]), Proteus

syndrome (PS [MIM 176920]), and Proteus-like syndrome (PLS). All four syndromes

are characterized by multiple hamartomatous lesions affecting derivatives of all three

germ cell layers. In CS, patients are also at an increased risk of developing breast, thyroid, and endometrial cancer (4). To date, germline PTEN mutations have been

identified in 85% of patients diagnosed with CS and 65% of patients diagnosed

with BRRS (50, 57). Additionally, 20% and 50% of patient with PS and PLS, respectively, have also been shown to carry PTEN germline mutations (4).

Mutation scanning of PTEN has primarily focused on the gene’s 9 exons and intron/exon boundaries, which span approximately 103 kilo-basepair (kb) on chromosome sub-band 10q23.3. Germline mutations have been reported throughout

PTEN, with the exception of exon 9, and the majority of these localize to its phosphatase catalytic core located in exon 5 (7, 65). More recently, mutations in PTEN’s core promoter region have also been identified and found to be associated with CS and increased phosphorylated AKT levels (50). However, despite the significant proportion of patients with known PTEN mutations, there are still many individuals with classic

PHTS diagnostic features for whom mutations have yet to be identified. Notably, CS is believed to be linked to the PTEN region, without genetic heterogeneity (6). In BRRS, on the other hand, the extent of genetic heterogeneity is unknown. Other mechanisms, such as modifiers of PTEN or another gene (or genes), which have yet to be identified, may be causal of this syndrome (57, 66). For individuals with PHTS, particularly those with CS, and without identifiable germline mutations, therefore, it is likely that the molecular mechanism(s) underlying their disease involves genetic alteration outside of the PTEN coding sequence, possibly involving elements associated in its trans-regulation, or deregulation, and which may lie upstream, downstream, or intronic of PTEN.

Identifying the mechanism of PTEN dysfunction in these patients is critical and of significant importance to the practice of personalized genetic healthcare.

To aid in identifying these genetic alterations, we used a haplotype-based approach to investigate the association of specific genomic regions of the PTEN locus

with disease. Through this approach, we demonstrate that specific haplotypes, perhaps

acting as low-penetrance susceptibility loci, are associated with PHTS in PTEN mutation-

negative samples. In addition to furthering our understanding of the role PTEN has in patients without detectable mutations, we have also identified specific haplotypes which

may act as low-penetrance alleles, or modifying factors, which could influence phenotypic expression in a subset of CR/BRRS patients with known germline PTEN

mutations.

2.2 MATERIALS AND METHODS

Study Subjects

A total of 447 unrelated subjects were included in the current analysis: 94 white

control subjects, 148 white PHTS patients without detectable germline PTEN mutations

(i.e. PTEN mutation-negative patients), and 205 white PHTS patients with previously

identified germline PTEN mutations/variations (i.e. PTEN mutation/variation-positive

patients). DNA for control subjects (Utah residents with ancestry from northern and

western Europe) was acquired from the Coriell Institute for Medical Research (Camden,

NJ). All PHTS samples were enrolled by referral from centers located throughout the

United States, Canada and Europe. Informed consent was acquired for all referred

subjects in accordance with procedures approved by the Human Subjects Protection

Committees of each respective institution.

Among the PTEN mutation-negative patients, 94 were classic CS, 10 patients

were classic BRRS, 4 patients exhibited features of both CS and BRRS (termed CS-

BRRS overlap), and 39 patients exhibited a CS-like phenotype (i.e. patients with some

features of CS, but not meeting operational diagnostic criteria). We were unable to

classify 1 PTEN mutation-negative patient.

The cohort of PTEN mutation/variation-positive patients included 103 mutation- positive samples (i.e. samples with pathogenic heterozygous missense or nonsense mutations) and 102 variation-positive samples. This latter group consists primarily of individuals with identified variants of unknown significance (VUS) located in the PTEN core promoter region or within potential splice donor/acceptor sites. Among the PTEN mutation-positive samples, 34 were classic CS, 18 were classic BRRS, 10 exhibited features of CS-BRRS overlap, and 40 were classified as CS-like, while 1 patient could not be classified. The PTEN variation-positive samples included 39 patients with classic

CS, 2 samples with classic BRRS, 6 samples with both CS and BRRS features, and 52

CS-like samples. Among our PTEN variation-positive patients, 3 could not be classified.

All patients classified as CS in the current study meet operational criteria

established by the International Cowden Consortium and curated by the National

Comprehensive Cancer Network (www.nccn.org) (4).

SNP Genotyping

SNPs spanning the PTEN locus and located approximately 1 every 5 kb were selected from the dbSNP database (http://www.ncbi.nlm.nih.gov/projects/SNP/) for validation and estimation of minor allele frequency in a 10-sample screening set consisting of 5 white control subjects and 5 white patient samples. 24 screened SNPs were found to have a minor allele frequency ≥ 0.10, and met our criteria for inclusion in this study. To achieve a uniformly spaced SNP map, 6 additional SNPs with a minor

allele frequency ≥ 0.10 were identified by DNA resequencing in our screening set. All

30 SNPs were genotyped in our 447 sample cohort. Polymerase chain reactions (PCRs) included 12.5µl HotStarTaq Master Mix (Qiagen, Valencia, CA), 10µM forward primer,

10µM reverse primer, and 20ng of template DNA and used the following thermal cycling conditions: 95°C for 15 min, 34 cycles of 95°C for 30 s, 50-58°C for 45 s, and 72°C

for 1 min, followed by a 72°C final extension for 10 min. 29 SNPs were genotyped

using either restriction fragment length polymorphism (RFLP), SNaPshot (Applied

Biosystems, Foster City, CA), or fragment analysis. SNaPshot and fragment analysis

products were electorphoresed using an ABI 3730 DNA Analyzer (Applied Biosystems,

Foster City, CA) and analyzed using GeneMapper v3.5 software (Applied Biosystems,

Foster City, CA). rs12573787 was genotyped by direct DNA resequencing. Primer

sequences and genotyping methodologies are provided in Table 2.1.

Hemizygous PTEN Deletion Analysis

Real-time quantitative PCR was used to investigate potential micro-deletions in

both control (n=4) and PTEN mutation-negative patient samples (n=14) where

homozygosity was observed for all 30 SNPs. 15 PTEN mutation/variation-positive

samples were also homozygous for SNPs assayed in this region, however, by virtue of

their heterozygous mutations/variations, these samples are assumed to carry two copies of

the PTEN allele. Copy number determinations were made for our target gene, PTEN

exons 2 and 5, and a control reference gene, GAPDH exon 7. 4 homozygous control

samples and 4 homozygous PTEN mutation/variation-positive samples were used as

Genotyping SNP Forward Primer Reverse Primer Methodology 1 GATAGAGTCTTGCTCTGTAG ACCATACAATATCTGCCTTG SNaPshot SBE primer: tgccacgtcgtgaaagtctgacaaGAGTAGCTGGGACTACAG 2 GCTGTGGTATGTACTTTCTG ATGCATGAAACAGCTACTTG RFLP (BanI) 3 TAAGTGGATCATGCCTGTAG CTTAATGGATGCAGACTCAG RFLP (BsiHKAI) 4 CATTCTCAAGCAGGACTCAG AATCCACCTGCTTCAGCTTC RFLP (HincII) 5 ACTGCAACTTTGACCTCCTG GCAGAATCTCACTCTGTCAG RFLP (DpnII) 6 GCTGTGGTTGCTCATCATTC CAATAGGAAGATACCCTGAC RFLP (AciI) 7 CCTGATGTTTAGAGAAGCAG CTTAGATTGCTGATCTTGTCTCC RFLP (BfaI) 8 ACTGGGCATGCTCAGTAGAG AGACCAACTCTCCGGCGTTC DNA resequencing 9 TTACTAAGGCTAAACTGGAC /FAM/-gcgaatcGTCATGTCACAGCTCACATG Fragment Analysis 10 GGATCACAGATGTAGGCTTG /FAM/- Fragment catcgccTAGCTGAGAGTGTACTAGAC Analysis

22 11 AGTTGAGAAGTCTAGTACAC ATCCTGTAATCCCACTCTAG SNaPshot SBE primer: atcgagatcgacccacaatccactggtcCTATAGTTGTGAATATGTTTAT 12 GCAAGATAGCTAGTACCATG AATGCCATATGCTAGCACAG RFLP (MboII) 13 AGGAATTCATGTCTGATGTG GTGACTGTACTGCTCACTTC SNaPshot SBE primer: gtgcAATCAAATTTTTGTACCTACAA 14 /HEX/- TAAACAGTCCTTCTGGCATC Fragment cgtccgaCATTATGCAGATGTAGACTC Analysis 15 TAGCATATTCTGACTCCTTC GATTAGCCCAAGAGTTGTAC SNaPshot SBE primer: agtcttcgagatccagccatcatcgactggtcAGTGCTGGGATTATAGGC 16 TGTAACCTGCAGGAGGCATC AAAGCAGAGAGGTAATACTC SNaPshot SBE primer: attacgtaGACTACGACCCAGGTAGG

Continued

Table 2.1. Primer sequences and genotyping methodologies for all SNP and quantitative real-time PCR reactions.

Table 2.1. Continued.

17 ACAGTTGTTCACAGTGGTAG /FAM/-gtaccgtTCCTAAGCAGATTGCTCCTG Fragment Analysis 18 TGCTTGTTAGAGTGAGGTAG CTAGCTCTATCAATCAGGTG RFLP (NcoI)

19 AGGTAGGTATGAATGTACTG /HEX/-agtcgatATCAGACTCCTCTTATCAAC Fragment Analysis 20 ACTGCAACCTCTACCTCCTG /FAM/- Fragment cgtccgcAGCTCAATGAACTCATGTAC Analysis 21 GCAACTGAATAGATGCGTAG ATAACTAACACCATCGTCAC 26 SNaPshot

SBE primer: cttaatccgtagtcaCCATTACTTCACCTCATCT 22 GGTACACTACTAATCACTTG TCACCGTGTTAGCCAGGATG RFLP (DraI) 23 GGAAGACTAGGTATTGACAG AAAGAGCATCAATGAGACTC RFLP (NlaIII) 24 AGAAACTGGAGCTTCTCATG AAGGCAATCTGAGTTATCTG RFLP (HpyCH4V)

23 25 AAGACAAAGCCAACCGATACTT GGAAAGACTAGAAGAGGCAGAAGC RFLP (HincII) 26 Same as SNP25 Same as SNP25 RFLP (BsaXI) 27 CATAATACCTGCTGTGGATG TCAGACCACAGCTAGTGAAC SNaPshot

SBE primer: aagctaggtgccacgacgagatagtctgagaaCCGAGTTGGGACTAGGGC 28 ATTGCTTCGCTCACCTGCTC CCTTTGAGATCCTCAGTAAG RFLP (HpyCH4 IV) 29 TAATTCTGGAGCTTCCTGAG CTGACTCTATACTCTGTGAG SNaPshot

SBE primer: atctagatccacccatactccgactatcAGGCTGAGGCATGAGAAT 30 TTGGCTACAAATGTCTCTAG GGTGCTGCTGTTTACTGAG RFLP (Bsu36 I)

Continued

Table 2.1. Continued.

Quantitative Real-time PCR Primers GAPDH GTATCGTGGAAGGACTCATG GGAAATTATGGGAAAGCCAG

PTEN GTTTGATTGCTGCATATTTCAG CCTGTATACGCCTTCAAGTC

PTEN CGAACTGGTGTAATGATATG TCCAGGAAGAGGAAAGGAAA

negative controls. Additionally, 2 samples previously determined to have PTEN deletions (1 spanning the entire PTEN locus, the other spanning both the PTEN and

BMPR1A genes) were assayed as positive controls. PCR efficiencies for each amplicon

were determined by standard curve analysis using serial dilutions of genomic DNA from a control sample (100ng, 50ng, 25ng, and 12.5ng, respectively). The calculated PCR efficiencies for these amplicons ranged from 76-81%.

Determination of gene copy number was assayed using 12.5µl iQ SYBR Green

Supermix (Bio-Rad Laboratories, Hercules, CA), 10µM forward primer, 10µM reverse

primer, and 20ng of template DNA. Thermal cycling conditions comprised of 95°C for 3

min and 40 cycles at 95°C for 30 s followed by 58°C for 30 s and 72°C for 30 s using

an ABI 7700 Sequence Detection System (Applied Biosystems, Foster City, CA). Target

and reference genes were assayed in triplicate for each sample and subject to meltcurve

analysis in order to determine amplicon specificity. The relative quantification of gene

copy number for both PTEN amplicons was determined using the comparative delta Ct

method (2-∆∆Ct) as described by Livak et al. (67).

Linkage Disequilibrium and Haplotype Analysis

Following assessment of Hardy-Weinberg equilibrium at each polymorphic locus,

pairwise LD coefficients (Lewontin’s D’) were estimated using the LDmax software program and visualized using the GOLD graphical interface (68). D’ was calculated and plotted separately for each sample population (control subjects, PTEN mutation-negative patients, and PTEN mutation/variation-positive patients). LD blocks were determined

using data from the control population and the dynamic programming algorithms

implemented in the HapBlock software program (Empirical LD method, D’ >0.90 for

strong LD) (69, 70). Following block partitioning, haplotype phase was reconstructed for

each block and all genotyped samples using the SNPHap software program, based on

pair-wise LD measurements and the expectation-maximization (EM) algorithm, and the

PHASE v2.1 software program, based on a Bayesian approach (71, 72). Additionally,

haplotype phase was reconstructed for the extended 30 SNP haplotype for all samples.

Statistical Analysis

Allele and genotype frequencies were computed for each SNP. P-values for

Hardy-Weinberg equilibrium (HWE) were obtained and Bonferroni adjustment was

applied to control the overall type-I error rate at 0.05. Each patient group (sharing the

same mutation status) was compared to the controls in their allele and genotype

distributions for each SNP. Following haplotype reconstruction, haplotypes from

PHASE were selected for comparisons. For each block and the extended block, a number

of tests were performed. First, haplotype frequencies in all phenotype groups with

distinct mutation statuses were compared using a Pearson χ2 test, where rare haplotypes

(expected frequency less than 5 for any group) were pooled together to make the χ2

approximation accurate as determined by the criterion of Cochran (73). We applied

Bonferroni adjustment to the 4 overall tests, using the significance level of 0.05/4

(0.0125) for each test. Each pair of groups was then compared using a Pearson χ2 test with the same criterion of pooling rare haplotypes. If the result of the overall test is statistically significant (P-value < 0.0125), the subsequent pairwise tests provide more

specific comparisons between groups. The first χ2 test controls the overall type-I error rate, but we further adjust for multiple tests between pairs of groups by using 0.05/6

(0.0083) as the significant level for each such test. Following this, we compared groups with different clinical features in terms of the haplotype frequencies using the same approach of an overall Pearson χ2 tests and subsequent comparisons of each group (one at

a time) with the controls, pooling rare haplotypes in each test as described above. The

same set of tests was also performed for the controls and the subset of patients classified

as mutation-positive or mutation-negative. Similarly to the first group of tests, we used

0.0125 as the significance level for each overall test to adjust for the total number of

blocks (4, including 3 haplotype blocks and the extended block), and 0.0125 as the

significance level for each subsequent pairwise comparison to adjust for the number of

groups being compared with the control group in turn.

Figure 2.1. Schematic diagram of the PTEN locus and SNPs included in the current analysis.

2.3 RESULTS

SNP Analysis and Identification of Hemizygous Deletions

We developed an informative marker set comprised of 30 relatively evenly spaced

SNPs (1 SNP every 5.6 kb, with a minor allele frequency greater than 10%) across a 163 kb region spanning the entire PTEN locus and including 30 kb of flanking sequence (Fig.

2.1 and Table 2.2). The majority of identified SNPs are intronic (18/30); 11 are outside of the gene (7 upstream and 4 downstream), and 1 SNP is located in PTEN’s 5’ untranslated region (UTR). These include 19 transitions, 5 transversions, and 6 insertion/deletion polymorphisms. Table 2.3 shows the allele frequencies for all 30 polymorphisms genotyped in the control and PHTS patient populations. No significant departures from HWE were observed. Figure 2.2 summarizes the -log10 P-values from comparisons of allele frequencies among PTEN mutation-negative, PTEN mutation-

Variation (Major/Minor Minor-Allele SNP dbSNP ID Positiona Alleles) Frequencyb Locationc 1 rs7085791 89583605 G/T 0.12 -30602 2 rs10887756 89587630 A/T 0.15 -26577 3 rs10887758 89593295 T/C 0.20 -20912 4 rs11202585 89598759 G/C 0.19 -15448 5 ss52090924d 89603299 T/C 0.20 -10908 6 rs11202590 89607699 C/T 0.14 -6508 7 rs1903860 89610190 T/C 0.13 -4017 8 rs12573787 89613696 G/A 0.14 -510 9 rs3216482 89616359 ins/del A 0.20 IVS1+2074 10 rs11355437 89629037 del/ins G 0.40 IVS1-14725 11 rs2673836 89629942 A/G 0.29 IVS1-13820 12 ss52090925d 89634206 C/G 0.21 IVS1-9556 13 rs10887763 89645216 A/G 0.14 IVS2+1370 14 rs3831732 89645229 ins/del A 0.39 IVS2+1382 15 rs12569872 89655492 G/A 0.14 IVS2+11645 16 rs1234224 89665276 A/G 0.32 IVS2-9974 17 ss52090926d 89666296 del/ins 32nt 0.39 IVS2-8954 18 rs10490920 89675623 T/C 0.14 IVS3+329 19 rs3830675 89680936 ins/del TCTTA 0.31 IVS4+109 20 ss52090927d 89689289 del/ins 16nt 0.15 IVS5+6300 21 rs2299941 89694699 A/G 0.12 IVS5-7156 22 ss52090928d 89699396 T/C 0.21 IVS5-2459 23 rs2673832 89702453 A/G 0.14 IVS6+457 24 ss52090929d 89710231 T/C 0.22 IVS7-400 25 rs555895 89710887 T/G 0.31 IVS8+32 26 rs926091 89711392 C/T 0.14 IVS8+537 27 rs701848 89716725 T/C 0.39 *614 28 rs10509532 89727534 C/T 0.14 *12325 29 rs7908337 89743671 T/C 0.24 *28462 30 rs11202614 89745623 C/T 0.14 *30414 a SNP position on chromosome 10 according to March 2006 human genome assembly, NCBI Build 36.1 (hg18). b Frequency in control population. c Location relative to translation start codon (-), PTEN exons (IVS), or translation stop codon (*). d SNPs identified by DNA resequencing in our screening set.

Table 2.2. Characteristics of 30 SNP panel.

positive, and PTEN variation-positive groups versus the control population. Overall, results from 13/90 comparisons (14%) were significant at the 0.05 level. Specifically, the allele frequency of SNP2 differed significantly among PTEN mutation-positive samples

and control samples (P-value = 0.0083). More strikingly, the allele frequencies of SNPs

10, 12, 14, 19, 24, 25, and 27 were all significantly different from the control population

among the PTEN variation-positive group (P-values < 0.01). Additionally, SNPs 16 and

17 both achieved statistical significance for this same comparison (P-values = 0.0127 and

0.0123, respectively).

We found 33/447 samples (7.4%) to be homozygous for all 30 SNPs in our panel, including; 4/94 control samples (4.3%), 14/148 PTEN mutation-negative samples (9.5%),

and 15/205 PTEN mutation/variation-positive samples (7.3%). Because heterozygosity

has previously been identified in the PTEN mutation/variation-positive samples, PTEN

copy number determinations were only made for the control and PTEN mutation-negative

samples. Previously we reported that 2-∆∆Ct values close to 1 indicates the presence of two PTEN alleles, while values close to 0.5 are indicative of hemizygous PTEN deletions

(50). As shown in Figure 2.3, the control samples were found to have average 2-∆∆Ct

values of 1.09±0.14 for PTEN exon 2 and 1.06±0.20 for PTEN exon 5, confirming that these samples retain two copies of PTEN. Similarly, a subset of PTEN mutation/variation-positive samples had average 2-∆ ∆Ct values of 0.94±0.14 for PTEN exon 2 and 0.97±0.12 for PTEN exon 5. Two samples known to harbor hemizygous germline deletions spanning the entire PTEN locus displayed average values of 0.67 and

0.53 for the two PTEN amplicons, respectively. 12 homozygous PTEN mutation-

SNP Samples n Allele Frequency P-value 1 G T Control 94 0.88 0.12 --- Mut- 146 0.81 0.19 0.0739 Mut+ 103 0.81 0.19 0.0920 Var+ 102 0.87 0.13 0.8844 2 A T Control 94 0.85 0.15 --- Mut- 146 0.76 0.24 0.0219 Mut+ 103 0.74 0.26 0.0083 Var+ 102 0.79 0.21 0.1807 3 T C Control 94 0.80 0.20 --- Mut- 146 0.78 0.22 0.6749 Mut+ 103 0.80 0.20 0.9607 Var+ 102 0.79 0.21 0.9735 4 G C Control 94 0.81 0.19 --- Mut- 146 0.78 0.22 0.4491 Mut+ 103 0.80 0.20 0.7522 Var+ 102 0.82 0.18 0.9062 5 T C Control 94 0.80 0.20 --- Mut- 146 0.78 0.22 0.6368 Mut+ 103 0.80 0.20 0.9431 Var+ 102 0.79 0.21 0.8278 6 C T Control 94 0.86 0.14 --- Mut- 146 0.84 0.16 0.5202 Mut+ 103 0.84 0.16 0.6405 Var+ 102 0.88 0.12 0.7544 7 T C Control 94 0.86 0.14 --- Mut- 146 0.84 0.16 0.5202 Mut+ 103 0.84 0.16 0.6405 Var+ 102 0.88 0.12 0.6450 8 G A Control 94 0.86 0.14 --- Mut- 146 0.83 0.17 0.5615 Mut+ 103 0.83 0.17 0.6542 Var+ 102 0.87 0.13 0.8598 9 D I Control 94 0.80 0.20 --- Mut- 146 0.78 0.22 0.6368 Mut+ 103 0.80 0.20 0.9607 Var+ 102 0.78 0.22 0.6498

Continued

Table 2.3. Summary of SNP allele frequency data for control and PHTS patient samples. 30

Table 2.3. Continued.

10 D I Control 94 0.60 0.40 --- Mut- 146 0.54 0.46 0.2033 Mut+ 103 0.62 0.38 0.7570 Var+ 102 0.73 0.27 0.0091 11 A G Control 94 0.70 0.30 --- Mut- 146 0.79 0.21 0.0351 Mut+ 103 0.78 0.22 0.0914 Var+ 102 0.74 0.26 0.5368 12 G C Control 94 0.79 0.21 --- Mut- 146 0.79 0.21 0.9388 Mut+ 103 0.83 0.17 0.3401 Var+ 102 0.90 0.10 0.0026 13 A G Control 94 0.86 0.14 --- Mut- 146 0.84 0.16 0.7762 Mut+ 103 0.83 0.17 0.6542 Var+ 102 0.90 0.10 0.2832 14 I D Control 94 0.61 0.39 --- Mut- 146 0.54 0.46 0.1526 Mut+ 103 0.62 0.38 0.9257 Var+ 102 0.74 0.26 0.0090 15 G A Control 94 0.86 0.14 --- Mut- 146 0.84 0.16 0.7762 Mut+ 103 0.84 0.16 0.7512 Var+ 102 0.89 0.11 0.445 16 A G Control 94 0.69 0.31 --- Mut- 146 0.66 0.34 0.5814 Mut+ 103 0.62 0.38 0.2137 Var+ 102 0.56 0.44 0.0127 17 D I Control 94 0.61 0.39 --- Mut- 146 0.54 0.46 0.1526 Mut+ 103 0.62 0.38 0.9257 Var+ 102 0.74 0.26 0.0123 18 T C Control 94 0.86 0.14 --- Mut- 146 0.83 0.17 0.5615 Mut+ 103 0.84 0.16 0.7512 Var+ 102 0.88 0.12 0.5397

Continued

Table 2.3. Continued.

19 D I Control 94 0.69 0.31 --- Mut- 146 0.66 0.34 0.5814 Mut+ 103 0.64 0.36 0.3446 Var+ 102 0.55 0.45 0.0073 20 I D Control 94 0.85 0.15 --- Mut- 146 0.84 0.16 0.9743 Mut+ 103 0.84 0.16 0.9815 Var+ 102 0.89 0.11 0.2885 21 A G Control 94 0.87 0.13 --- Mut- 146 0.88 0.12 0.9112 Mut+ 103 0.86 0.14 0.9259 Var+ 102 0.89 0.11 0.6513 22 C T Control 94 0.79 0.21 --- Mut- 146 0.79 0.21 0.9893 Mut+ 103 0.84 0.16 0.2256 Var+ 102 0.87 0.13 0.0340 23 G A Control 94 0.86 0.14 --- Mut- 146 0.91 0.09 0.1572 Mut+ 103 0.88 0.12 0.6188 Var+ 102 0.93 0.07 0.0538 24 C T Control 94 0.79 0.21 --- Mut- 146 0.79 0.21 0.9893 Mut+ 103 0.84 0.16 0.2256 Var+ 102 0.90 0.10 0.0026 25 T G Control 94 0.69 0.31 --- Mut- 146 0.65 0.35 0.5299 Mut+ 103 0.63 0.37 0.2961 Var+ 102 0.54 0.46 0.0054 26 C T Control 94 0.86 0.14 --- Mut- 146 0.83 0.17 0.5615 Mut+ 103 0.84 0.16 0.7512 Var+ 102 0.88 0.12 0.5397 27 T C Control 94 0.61 0.39 --- Mut- 146 0.55 0.45 0.2260 Mut+ 103 0.63 0.37 0.8474 Var+ 102 0.74 0.26 0.0090

Continued

Table 2.3. Continued.

28 C T Control 94 0.86 0.14 --- Mut- 146 0.83 0.17 0.5615 Mut+ 103 0.84 0.16 0.7512 Var+ 102 0.89 0.11 0.4450 29 T C Control 94 0.76 0.24 --- Mut- 146 0.72 0.28 0.4920 Mut+ 103 0.73 0.27 0.6173 Var+ 102 0.75 0.25 0.9071 30 C T Control 94 0.86 0.14 --- Mut- 146 0.84 0.16 0.6292 Mut+ 103 0.84 0.16 0.7512 Var+ 102 0.85 0.15 0.9282

negative samples exhibited 2-∆∆Ct values similar to those observed in the control and

PTEN mutation/variation-positive samples (1.14-1.66 for PTEN exon 2 and 0.95-1.51 for

PTEN exon 5). Two samples, 1582-02 (0.46 for PTEN exon 2 and 0.21 for PTEN exon

5) and 2849-01 (0.72 for PTEN exon 2 and 0.57 for PTEN exon 5) had 2-∆∆Ct values that

were consistent with hemizygous deletions. Because of their hemizygous status at this locus, both 1582-02 and 2849-01 were excluded from the subsequent LD and haplotype analyses.

Linkage Disequilibrium along the PTEN Locus

We found 3 distinct haplotype blocks characterized by strong LD in the control

population (Fig. 2.4A). Block 1 spans SNP1 (-30602 G/T) to SNP9 (IVS1+2074insA)

(33 kb), block 2 spans SNP11 (IVS1-13820 A/G) to SNP21 (IVS5-7156 A/G) (65 kb),

and block 3 spans SNP23 (IVS6+457 A/G) to SNP30 (*30414 C/T) (43 kb). Adjacent to

each partitioned block, LD decays. SNP10 (IVS1-14725delG) displayed average D’

Figure 2.2. Summary of SNP allele frequency P-values for PHTS patient population groups versus control population. Allele frequencies among 3 PHTS patient populations

(PTEN mutation-negative, PTEN mutation-positive, and PTEN variation-positive) were

compared to the control population for all 30 SNPs using a Pearson χ2 test. –log10 of the

P-values were plotted for each comparison and for all SNPs. Note: -log10 P-value 1 =

P-value 0.1, -log10 P-value 2 = P-value 0.01, and -log10 P-value 3 = P-value 0.001.

Figure 2.3. Hemizygous PTEN deletion analysis. PTEN copy number was estimated at

exons 2 and 5 using the Livak method for control (n=4), PTEN mutation/variation-

positive (n=4), and PTEN mutation-negative samples (n=14) found to be homozygous for

all 30 genotyped SNPs, as well as for known PTEN deletion positive samples (n=2). 2-

∆∆Ct values for the control samples ranged from 0.87 to 1.38. PTEN mutation/variation-

positive samples (known to have heterozygous PTEN mutations/variations) displayed

values between 0.75 and 1.13. PTEN deletion positive samples had average 2-∆∆Ct values of 0.67 and 0.53 for exons 2 and 5, respectively. 12 PTEN mutation-negative samples

had values similar to the control and PTEN mutation-positive samples (0.95 to 1.66). 2

PTEN mutation-negative samples (1582-02 and 2849-01) displayed 2-∆∆Ct values similar

to the PTEN deletion positive samples, ranging from 0.21 to 0.72.

values of 0.75 and 0.85 with blocks 1 and 2, respectively, and could not be assigned to

either block. Similarly, SNP22 (IVS5-2459 T/C) had an average D’ < 0.90 and was not in strong LD with either adjacent block, suggesting that both SNPs lie in/near putative recombination hot-spots. The PTEN haplotype structure in two PHTS patient populations

(146 unrelated PTEN mutation-negative and 205 unrelated PTEN mutation/variation-

positive PHTS patient samples) are shown in Figures 2.4B and 2.4C, respectively.

Similar to the control population, significant LD was observed for the entire region.

However, compared to controls, the overall LD patterns observed in the PHTS patient

samples appear to be distinct. LD in these samples suggests less recombination of the

adjacent blocks and the presence of extended haplotypes across this locus.

Haplotype Association Analysis at the PTEN locus

Having identified 3 regions of strong LD flanked by two apparent recombination hot-spots, we next proceeded to investigate the haplotypes contained within each LD block. Haplotype phase was reconstructed using both the SNPHap (http://www-

gene.cimr.cam.ac.uk/clayton/software/) and PHASE

(http://www.stat.washington.edu/stephens/software.html) software programs. The two algorithms performed similarly, agreement was reached for 98.8% of the reconstructed haplotype blocks and for 96.5% of the reconstructed chromosomes (i.e. extended haplotypes) (data not shown). PHASE haplotype blocks and haplotype block frequencies for all chromosomes are shown in Table 2.4. The number of common haplotypes accounting for >80% of the observed chromosomes varied among the 3 blocks. We identified 5 common haplotypes for both blocks 1 and 2 and a total of 7 common

Figure 2.4. GOLD plot of pairwise LD between 30 SNPs. D’ values are reported for all sample groups: A) 94 control samples, B) 146 PTEN mutation-negative samples, and C)

205 PTEN mutation/variation-positive samples. The control samples display 3 distinct haplotype blocks: block 1 from SNP1 (-30602 G/T) to SNP9 (IVS1+2074insA), block 2 from SNP11 (IVS1-13820 A/G) to SNP21 (IVS5-7156 A/G), and block 3 from SNP23

(IVS6+457 A/G) to SNP30 (*30414 C/T). SNP10 (IVS1-14725delG) and SNP22 (IVS5-

2459 T/C) appear to lie near/within areas of historical recombination. Both the PTEN mutation-negative and the PTEN mutation/variation-positive samples display varied LD

patterns across this locus compared to the control population.

haplotypes for block 3. For block 3, the number of common haplotypes also varied

among sample groups. The haplotype distributions for each block differed significantly

among the examined groups (Table 2.4).

The distribution of the 5 block 1-haplotypes amongst controls, PTEN mutation-

negative patients, mutation-positive patients and variation-positive patients was

significantly different (χ2 = 30.66; P-value = 0.0098). Haplotype 1 was found to be

under-represented in PTEN mutation-negative samples (49.7%) and over-represented in

the control population (63.8%). Haplotype 2 was over-represented in PTEN mutation- negative and PTEN mutation-positive samples compared to both control and PTEN variation-positive samples, 18.2% and 16.5% versus 12.2% and 12.3%, respectively.

Interestingly, the percentage of low frequency haplotypes was also over-represented among both PTEN mutation-negative and PTEN variation-positive samples (10.3% and

8.8%, respectively) compared to controls (2.7%).

Statistically significant differences were also observed for the haplotype

distributions of blocks 2 and 3 between the examined sample populations (χ2 = 45.31 and

62.53, respectively; P-values < 0.0001 for both comparisons). For block 2, haplotype 1

was under-represented in both the PTEN mutation-negative samples (19.2%) and the

PTEN mutation-positive samples (21.4%) compared to control subjects (29.3%).

Haplotype 2 was the most frequent haplotype among the PTEN variation-positive

samples (32.4%) and over-represented in this group compared to both the control and

PTEN mutation-negative samples (15.4% and 16.4%, respectively). The converse was

observed for haplotype 4; a 9.8% haplotype frequency was seen in the PTEN variation- 38

PTEN PTEN PTEN Controls Mutation - Mutation + Variation + Block and Haplotypes (n = 188) a (n = 292) a (n = 206) a (n = 204) a Block 1 haplotypesb: 1. GATGTCTGD 0.638 0.497 0.549 0.559 2. TTTGTCTGD 0.122 0.182 0.165 0.123 3. GACCCTCAI 0.138 0.120 0.141 0.108 4. GTTGTCTGD 0.027 0.055 0.073 0.074 5. GACCCCTGI 0.048 0.045 0.044 0.049 Low Frequency 0.027 0.103 0.029 0.088 Block 2 haplotypesc: 1. GCADGAITDIA 0.293 0.192 0.214 0.255 2. ACADGGITIIA 0.154 0.164 0.199 0.324 3. ACAIGADTDIA 0.176 0.240 0.204 0.162 4. AGAIGADTDIA 0.213 0.202 0.165 0.098 5. ACGDAGICIDG 0.128 0.113 0.126 0.103 Low Frequency 0.037 0.089 0.092 0.059 Block 3 haplotypesd: 1. ATTCCCTC 0.176 0.226 0.214 0.157 2. ACTCCCTC 0.213 0.205 0.160 0.098 3. ATTCTCTC 0.160 0.123 0.136 0.216 4. ATGTTTCT 0.144 0.154 0.150 0.098 5. ATGCTCCC 0.101 0.110 0.107 0.118 6. ATGCTCTC 0.069 0.065 0.097 0.191 7. GTTCTCTC 0.138 0.089 0.117 0.069 Low Frequency 0.000 0.027 0.024 0.054 a n = Number of haplotypes. b χ2 = 30.66; P = 0.0098. c χ2 = 45.31; P < 0.0001. d χ2 = 62.53; P < 0.0001.

Table 2.4. Haplotype blocks across the PTEN locus.

positive samples compared to 21.3% and 20.2% for the control and PTEN mutation- negative samples, respectively.

As observed for block 1, low-frequency haplotypes were also over-represented in

PHTS samples. These haplotypes were over-represented in both PTEN mutation- negative and PTEN mutation-positive samples compared to controls for block 2: 8.9% and 9.2% versus 3.7%. For block 3, low frequency haplotypes are only represented in the

3 PHTS sample groups (2.7% in PTEN mutation-negative samples, 2.4% in PTEN mutation positive samples, and 5.4% in PTEN variation-positive samples).

Block 3-haplotype 2 was under-represented in PTEN variation-positive samples

(9.8%) and over-represented in the control (21.3%) and PTEN mutation-negative populations (20.5%). As discussed above for block 2-haplotypes 2 and 4 among these same 3 sample populations, block 3-haplotype 6 also displayed an inverse relationship with block3-haplotype 2: PTEN variation-positive samples (19.1%) compared to the control (6.9%) and PTEN mutation-negative (6.5%) samples. This observation suggests that a founder haplotype is formed by the extended haplotype between blocks 2 and 3

(haplotypes 4 and 2, respectively). Furthermore, an extended haplotype may also exist between block 2-haplotype 2 and block 3-haplotype 6, however, the former appears to be associated with more haplotype diversity (Table 2.5).

To explore genetic associations pertaining to extended haplotypes, we also reconstructed haplotypes spanning all 30 SNPs (Table 2.5). 10 extended haplotypes represented 81.9% of all haplotypes observed in our cohort, while 71 additional ‘rare’ extended haplotypes accounted for the remaining 18.1% (data not shown). Statistically significant differences were observed between the sample populations (χ2 = 77.64; P-

value = 0.0001). Haplotype 2 was observed to be under-represented in both the PTEN

mutation-negative (8.6%) and PTEN mutation-positive (8.7%) samples. This same

haplotype was over-represented in the PTEN variation-positive samples (18.6%).

Haplotype 5 was over-represented in the control population, 13.8%, and under- represented in both the PTEN mutation-negative and PTEN variation-positive groups,

7.5% and 5.9% respectively. Interestingly, extended haplotype 1, the most frequent

haplotype observed in all sampled chromosomes (16.0%), was under-represented in

PTEN variation-positive samples (9.3%) compared to both control (18.6%) and PTEN

mutation-negative (19.2%) samples. This haplotype is comprised of block 2-haplotype 4

and block 3-haplotype 2, as well as block 1-haplotype 1 (the most common haplotype

observed in this block, ≥ 50% in all sample populations). This strongly suggests that,

despite the presence of two recombination hot-spots, a founder haplotype likely exists for

this region of 10q. Two additional extended haplotypes, 2 and 5, were also observed to

be over-represented in the control population (13.3% and 13.8%, respectively) compared

to the PTEN mutation-negative group (8.6% and 7.5%, respectively). Haplotype 2 was

also under-represented in PTEN mutation-positive samples (8.7%).

Additionally, as observed for each of the 3 individual blocks, the frequencies of

‘rare’ extended haplotypes were different among the different sample populations,

accounting for only 12.8% of control chromosomes, compared to 22.6% and 18.6% of

PTEN mutation-negative and PTEN variation-positive chromosomes, respectively. These

data suggest that rare alleles may underlie the disease etiology in these sample

populations and, more specifically in the case of the PTEN mutation-negative group, may

harbor pathogenic variant(s) which escaped detection by ‘standard’ PTEN mutation

PTEN PTEN PTEN Total Controls Mutation - Mutation + Variation + PTEN Extended Haplotypes (n = 890) a (n = 188) a (n = 292) a (n = 206) a (n = 204) a 1. GATGTCTGDDAGAIGADTDIACACTCCCTC 0.160 0.186 0.192 0.155 0.093 2. GATGTCTGDIGCADGAITDIATATTCTCTC 0.119 0.133 0.086 0.087 0.186 3. TTTGTCTGDDACAIGADTDIATATTCCCTC 0.113 0.101 0.137 0.121 0.083 4. GACCCTCAIIACGDAGICIDGTATGTTTCT 0.099 0.117 0.082 0.117 0.088 5. GATGTCTGDIGCADGAITDIATGTTCTCTC 0.092 0.138 0.075 0.107 0.059 6. GATGTCTGDIACADGGITIIATATGCTCTC 0.064 0.027 0.031 0.073 0.137 7. GATGTCTGDIACADGGITIIATATGCTCCC 0.054 0.048 0.055 0.063 0.049 8. GACCCCTGIIACADGGITIIATATGCTCCC 0.044 0.048 0.038 0.044 0.049 9. GATGTCTGDDACAIGADTDIATATTCCCTC 0.039 0.059 0.048 0.029 0.020 10. GTTGTCTGDDACAIGADTDIATATTCCCTC 0.035 0.016 0.031 0.044 0.049 Low Frequency 0.181 0.128 0.226 0.160 0.186 2

4 NOTE – χ = 77.64; P < 0.0001. 2 a n = Number of haplotypes.

Table 2.5. Extended haplotypes for all 30 SNPs across the PTEN locus.

scanning methodologies.

To examine these associations further, we performed a series of comparative

haplotype analyses among PHTS and control samples for haplotype blocks and the

extended haplotypes (Table 2.6). A significant difference was observed for block 1

between the PTEN mutation-negative and control samples (χ2 = 18.20; P-value =

0.0027). For PTEN variation-positive samples, block 2, block 3, and the extended

haplotype all differed significantly from the control population (χ2 = 22.06; P-value

=0.0005, χ2 = 37.96; P-value = <0.0001, and χ2 = 38.84; P-value = <0.0001, respectively). Notably, the allele frequencies of several individual SNPs comprising these haplotype blocks were significantly different among these same two groups (Table 2.6 and Fig. 2.2). A comparison among PTEN mutation-negative and PTEN variation-

positive samples revealed significant differences at these same genomic regions: block 2

(χ2 = 28.65; P-value = <0.0001), block 3 (χ2 = 39.97; P-value = <0.0001), and the

extended haplotype (χ2 = 44.13; P-value = <0.0001). In a comparison based on stratification by clinical diagnoses, block 2, block 3, and the extended haplotype were also associated with CS-like patients, reaching statistical significance for each of these comparisons (χ2 = 18.46; P-value = 0.0024. χ2 = 24.35; P-value = 0.0010, and χ2 =

28.02; P-value = 0.0018, respectively). A similar trend was observed for this phenotype when the PTEN mutation-negative and PTEN mutation-positive groups were combined

(block 2: χ2 = 13.60; P-value = 0.0587, block 3: χ2 = 12.61; P-value = 0.0273, and the extended haplotype: χ2 = 21.81; P-value = 0.0095). While interesting, only the

Block 1 Block 2 Block 3 Extended Haplotype Comparison χ2 P χ2 P χ2 P χ2 P Mutation statusa: PTEN Mut – vs. control 18.20 0.0027 12.03 0.0614 10.44 0.1649 17.27 0.0447 PTEN Mut + vs. control 6.78 0.2376 9.66 0.0854 8.67 0.2771 13.34 0.2054 PTEN Var + vs. control 12.34 0.0304 22.06 0.0005 37.96 <0.0001 38.84 <0.0001 PTEN Mut – vs. PTEN Mut + 10.91 0.0531 3.41 0.7566 3.83 0.7987 13.05 0.2899 PTEN Mut – vs. PTEN Var + 5.02 0.5415 28.65 <0.0001 39.97 <0.0001 44.13 <0.0001 PTEN Mut + vs. PTEN Var + 8.38 0.1364 13.82 0.0318 21.65 0.0029 20.31 0.0161 Clinical featuresb: Overall 9.32 0.3162 29.76 0.0193 26.42 0.0484 7.98 0.0924 CS vs. control 12.36 0.0302 7.61 0.1788 10.08 0.1841 15.51 0.1147 BRRS vs. control 1.57 0.6667 9.87 0.0789 10.03 0.1233 5.07 0.4065 CS/BRRS vs. control 1.87 0.3932 9.19 0.0564 1.31 0.8600 0.49 0.4825

44 CS-like vs. control 12.94 0.0240 18.46 0.0024 24.35 0.0010 28.02 0.0018 Mutation status and clinical featuresc: Overall 8.82 0.0659 13.41 0.0984 23.70 0.0220 3.58 0.4700 PTEN Mut – and Mut + CS vs. control 14.16 0.0146 12.40 0.0883 10.12 0.0720 11.98 0.1519 PTEN Mut – and Mut + BRRS vs. control 0.96 0.8107 9.04 0.1715 9.02 0.1083 4.11 0.5339 PTEN Mut – and Mut + CS/BRRS vs. control 0.70 0.4027 0.04 0.8415 4.32 0.1155 0.29 0.5890 PTEN Mut – and Mut + CS-like vs. control 11.35 0.0449 13.60 0.0587 12.61 0.0273 21.81 0.0095 NOTE – Significant results are indicated in bold. Mut+ = mutation-positive; Mut – = mutation-negative; Var+ = variation-positive. a Patients with PHTS were stratified based on their PTEN mutation status and were compared with controls as well as each other. b An overall comparison was made on the basis of stratification of clinical features followed by comparisons based on clinical diagnoses (CS, BRRS, CS/BRRS, or CS-like phenotype) for all patients, irrespective of mutation status, compared with controls. c Comparisons of patient clinical diagnoses among PTEN Mut– and Mut+ samples followed by comparisons versus control samples.

Table 2.6. Comparative haplotype analysis.

comparison of the extended haplotype was statistically significant. Additionally, among

PTEN mutation-negative and PTEN mutation-positive CS patients, block 1 appeared to

show an association with this phenotype (χ2 = 14.16; P-value = 0.0146), although this result did not reach statistical significance following Bonferroni adjustment.

2.4 DISCUSSION

PHTS represents an assemblage of phenotypically diverse syndromes manifested by germline pathogenic mutations in the PTEN gene. Standard germline mutation scanning has identified causal variants in a majority of patients diagnosed with this complex disorder, particularly for patients diagnosed with CS or BRRS (2, 7). Despite extensive mutation scanning, however, the etiologic variant(s) have yet to be identified in

15% and 35% of patients with these syndromes, respectively. To investigate genetic associations with PTEN in this subset of patients, as well as to characterize the haplotype architecture of this locus, we chose to utilize a case-control haplotype-based approach.

Similar approaches have been used to examine genetic associations at a growing number of candidate genes (74-76). Haplotype-based approaches are of particular interest as most reports of disease-associated mutations describe variants that directly alter the protein coding sequence of a gene. These studies fail to consider other mechanisms that may alter gene function and, where mutations are not found, may overlook polymorphisms that reside outside of the coding region. Such mechanisms include alterations of gene regulation through the disruption of trans-acting factor(s) and cis-acting sequence element interactions, resulting in a pathologic state (77).

While the mutation spectrum of PTEN in PHTS has been well studied, its

haplotype architecture has not. The extent of LD across this regions has been examined

in three previous studies (78-80). Hamilton et al. first reported the existence of two distinct four-marker haplotypes in the general population, but found no association with prostate cancer and benign prostatic hyperplasia (79). A study by Zhang et al. examined the association of this same locus with smoking initiation and nicotine addiction using 5 haplotype tagging SNPs (htSNPs) selected using the SNPbrowser software program

(Applied Biosystems, Forster City, CA) (80). In this study, three haplotype blocks were

observed; block 1 spanned 41 kb (from nucleotide position 89,606,485 to 89,647,130),

block 2 spanned 16 kb (from nucleotide position 89,679,301 to 89,695,409), and block 3

included a single SNP located at position 89,716,724. As the authors noted, this differed

slightly from the PTEN haplotype structure observed by the International HapMap

Project (http://www.hapmap.org/). Most recently, Haiman et al. investigated the influence of common variations across this region and the risk of sporadic breast and prostate cancer (78). Also employing a htSNP approach, these authors identified 9 common haplotypes representing >87% of all chromosomes across 123 kb of the PTEN

locus. Among these common haplotypes, no strong association was found with either

sporadic cancer.

For the present study, haplotype phase was reconstructed for our samples using

the SNPHap software program, based on pair-wise LD measurements and the EM

algorithm (81, 82). Previous studies have demonstrated the appropriateness of the EM

algorithm for inferring haplotypes from data obtained from unrelated individuals (81, 83-

85). Because our analysis relied on statistical inferences of haplotypes from unphased

data, we chose to validate this reconstruction using a second algorithm based on a

Bayesian approach as implemented in the PHASE software program (72, 86). Although the two programs rely on different mathematical approaches, both algorithms performed remarkably similarly.

Our analysis of the LD structure across this region of 10q revealed 3 distinct

haplotype blocks; block 1 spans 33 kb (from nucleotide position 89,583,605 to

89,616,359), block 2 spans 65 kb (from nucleotide position 89,629,942 to 89,694,699),

and block 3 spans 43 kb (from nucleotide position 89,702,453 to 89,745,623). Block 2 is

flanked by regions of decreased LD, suggesting that SNPs at these sites lie within areas of chromosome recombination. Our block partitioning, based on the method by Gabriel et al., partially agreed with that described by Zhang et al. However, based on our data, block 1 described by Zhang et al. is actually made up of two distinct blocks. As previously mentioned, these authors defined this region using two htSNPs. To ensure the accurate characterization of this region, we chose to empirically assess its haplotype architecture using a high-density set of polymorphic markers. Because the extent of LD is variable in this region, the htSNP approach failed to capture all pertinent information regarding the locus in question, specifically regarding the breakdown of LD observed at

SNP10 (IVS1-14725delG) and SNP22 (IVS5-2459 T/C). Therefore, a more dense marker set is required. htSNP approaches are capable of capturing most haplotype diversity within a population, i.e. approximately 90% of all chromosomes in a given population (70). However, for uncommon haplotypes, particularly in cases where the causal allele is under-represented, this approach is limited. Our finding that ‘rare’ haplotype blocks account for 2- to 3-fold more PHTS chromosomes compared to control

chromosomes and ‘rare’ extended haplotypes account for nearly 2-fold more PTEN mutation-negative and PTEN variation-positive chromosomes, suggests that for rare diseases, such as PHTS, low frequency, or ‘rare’, haplotypes are the ones associated with disease and may harbor pathogenic variants.

In our effort to characterize the haplotype architecture of the PTEN locus, we identified two PHTS patients, 1582-02 and 2849-01, with hemizygous micro-deletions.

Each sample retained only a single copy of the PTEN allele; 1582-02 retained extended haplotype 4 and 2849-01 retained extended haplotype 5. These haplotypes had allele frequencies of 9.9% and 9.2%, respectively, in the entire sample population, resulting in less than a 1% chance of homozygosity for these alleles. By contrast, 3 of the 4 homozygous control samples were homozygous for the most frequent haplotype observed in our study. Based on the analysis of microsatellite markers, these deletions span less than approximately 312 kb to 390 kb, respectively (data not shown). Previously, we identified PTEN deletions in only 3 PHTS patients, all of whom were clinically diagnosed with BRRS or CS/BRRS overlap (50). The patients identified in the current study have diagnoses of classic CS (2849-01) and CS-like (1582-02). Implications from these data extend to the clinical realm, suggesting that PTEN deletion analysis is warranted in all PHTS patients with CS, BRRS, CS/BRRS, and CS-like phenotypes who lack apparent germline mutations.

Interestingly, 1 PTEN mutation-negative sample was homozygous for a ‘rare’ extended haplotype with an allele frequency <0.7% in the entire study population. Close inspection of this haplotype revealed that blocks 2 and 3 were relatively common, while block 1 consisted of a low frequency block. This low frequency haplotype block,

GACCCTCGI, was only observed in 8 samples; 7 PTEN mutation-negative samples and

1 PTEN variation-positive sample. Carriers of this allele include 4 CS patients, 3 CS-like patients, and 1 CS/BRRS patient. For our homozygous sample, this suggests that, because of the locations of our amplicons, our deletion analysis may have been unable to detect a possible deletion of the 5’ region of this locus. This data implicates the

GACCCTCGI block as a low frequency, highly penetrant PHTS susceptibility allele.

Furthermore, all 8 samples have similar ‘rare’ extended haplotypes; 5 (3 CS and 2 CS- like) share the same haplotype, 1 (CS/BRRS) deviates from this haplotype by a single variation in block 2, and 2 (1 CS and 1 CS-like) are variable for both blocks 2 and 3.

Although the SNPs which make up this block and extended haplotype are not causal

(based on their frequency in the control population), they are likely in LD with an unknown functional variant conferring disease susceptibility. This further supports the notion that ‘rare’, low frequency alleles (LD blocks and/or extended haplotypes) may be associated with disease and should therefore be considered as candidate susceptibility alleles in rare disorders.

In addition to an association with rare haplotypes, our analysis of haplotype blocks and extended haplotypes revealed significant differences among the control group and various patient sample populations. The number and frequency of common haplotypes needed to cover >80% of the observed chromosomes varied for each of the 3 blocks and the extended haplotype. Similar to the association with rare alleles, these data suggest greater haplotype diversity among the PHTS patient populations compared to the control group and are indicative of a higher degree of recombination of the ‘ancient haplotype’. Interestingly, the overall LD pattern observed in our patient samples appears

to indicate the presence of extended haplotypes. This effect seemed most apparent when

PTEN variation-positive patients were compared to controls, revealing significant differences between these groups for blocks 2 and 3, as well as for the extended haplotype, and suggesting less recombination among PHTS patients. Furthermore, pairwise comparisons between groups revealed that the PTEN mutation-negative and

PTEN mutation-positive groups were most similar, suggesting that different pathogenic variants may have arisen from similar haplotypic backgrounds. Taken together, these data indicate that some PHTS patients, i.e. PTEN mutation-positive individuals, and perhaps PTEN variation-positive individuals, exhibit a haplotype-founder effect, while others, i.e. PTEN mutation-negative individuals, harbor rare extended haplotypes which have undergone extensive ‘shuffling’ of the LD blocks across this region.

Interestingly, among PTEN mutation-negative samples, the strongest genetic

effect appears to be associated with haplotypes forming block 1 (a block spanning at least

30 kb upstream of PTEN and which includes several kilo-basepairs of the gene’s first

intron). With the exception of PTEN’s core promoter and exon 1, this region has not

been well characterized. Screening efforts which have failed to identify

mutations/variations at these sites in this group of patients suggest that alterations in this

region may have a role in PTEN’s regulation. These likely involving novel regulatory

elements and contribute to its deregulation.

Various PHTSs, such as BRRS and CS, appear to be caused by the same PTEN

mutations, despite clear differences in phenotypic presentation (7). The R130X mutation

in exon 5, for example, occurs in 8 PTEN mutation-positive patients included in this

study. Among these individuals, 3 have a clinical diagnosis of CS, 2 have a clinical

diagnosis of BRRS, and 3 have a clinical diagnosis of CS/BRRS. Both BRRS individuals are carriers of extended haplotypes 3 and 10 and exhibit classic features of

BRRS including macrocephaly, lipomas, and pigmented macules of the penis. The probability of this genotype in the general population is <0.3%, suggesting that this infrequent allelic combination likely contributes to their phenotype and that low- penetrant functional variants reside on these loci. Furthermore, although stratification by clinical phenotype was only minimally associated with our haplotypes, correlations from these data become more apparent when the patient’s mutation status is considered.

In addition to providing a panel of informative markers for testing genetic associations at the PTEN locus, our data strongly suggest that specific haplotypes along this region are associated with increased PHTS susceptibility. ‘PTEN mutation-negative’ samples lacking traditional mutations in the PTEN coding sequence possess a significantly different haplotype architecture compared to control samples. Along with an association to block 1 of this locus, ‘rare’ alleles comprise this architecture and may underlie the disease etiology in these patients. Furthermore, haplotype profiles in PHTS patients with known mutations/variations contribute to the phenotypic complexity of this syndrome.

Although the mechanisms underlying these relationships have yet to be elucidated, these data suggest that associated chromosomal segments likely harbor variants, potentially involved in the transcriptional regulation of PTEN, which are both pathogenic and/or modifying in nature, and manifest as low-penetrant disease susceptibility alleles.

CHAPTER 3

COMPARATIVE GENOMIC AND FUNCTIONAL ANALYSES REVEAL A NOVEL CIS-ACTING PTEN REGULATORY ELEMENT AS A HIGHLY CONSERVED FUNCTIONAL E-BOX MOTIF DELETED IN COWDEN SYNDROME

3.1 INTRODUCTION

Germline mutations in the 10q23-located gene encoding phosphatase and tensin

homolog deleted on chromosome ten (PTEN [MIM 601728]) have been found in 85% of probands with Cowden syndrome (CS [MIM 158350]) and 65% of probands with

Bannayan-Riley-Ruvalcaba syndrome (BRRS [MIM153480]) (50, 52, 57). PTEN, a ubiquitously expressed tumor suppressor dual-specificity phosphatase, antagonizes the phosphatidylinositol-triphosphate kinase (PI3K) signaling pathway through its lipid phosphatase activity, resulting in the subsequent inhibition of the Akt proto-oncogene (7,

87-90). PTEN’s protein phosphatase activity negatively regulates the mitogen-activated protein kinase (MAPK) pathway (13, 21). Inactivation or loss of PTEN function results

in increased cell survival and uncontrolled cellular proliferation mediated by these

pathways and, as is seen in many human cancers, ultimately resulting in neoplasia.

PTEN is believed to be a constitutively active protein, whose sufficient activity is

dependent upon protein levels (91). Despite this fact, and PTEN’s significant roles in regulating the cell cycle and in the malignant transformation associated with carcinogenesis, relatively little is known about the mechanisms that govern transcriptional regulation of PTEN expression, and virtually nothing is known about its

transcriptional regulation in human heritable disorders such as CS and BRRS. Previous

in vitro studies have identified functional consensus binding sites for transcriptional

activators p53 and early growth response-1 (EGR1) in the PTEN core promoter region

(nucleotide [nt] position -1344 to -745) (19, 55). Additional transcription factors have

been shown to be involved in regulating PTEN transcription, including peroxisome

proliferator-activated receptor gamma (PPARγ), nuclear factor kappaB (NFκB), c-Jun,

and, most recently, CBF-1 (92-95). However, for most of these, the precise mechanisms

of transcriptional regulation remain unclear.

Together with data from the Human Genome Project, sequence information from

several non-human vertebrate genomes is being used to identify novel regulatory

elements in previously uncharacterized, noncoding DNA (96-98). Comparative sequence

analysis approaches have identified highly conserved regions that contain functionally important elements involved in the regulation of several human genes, including IL-4, IL-

13, IL-5, SCL, IFN-γ, and BRCA1 (99-102). Using a similar approach, here, we set out to

identify novel functional cis-acting regulatory elements along and around the PTEN

locus, an exercise which is directly germane to the observation that germline mutations in

the PTEN promoter occur in 10% of mutation-positive CS, and large deletions, favoring

the 5’ end of PTEN (exon 1) and upstream of the gene (ie, the promoter region), occur in

11% of BRRS patients (50).

In using a comparative genomic approach, we identified a highly conserved

sequence, sharing 80% sequence identity, among the Homo sapiens, Mus musculus, and

Rattus norvegicus PTEN locus. Within this region, we identified a canonical E-box

sequence (CACGTG) located at position -2181 to -2176, approximately 800bp upstream

of the PTEN core promoter and more than 1.1 kb upstream of its minimal promoter

region (located at -958 to -821). In vitro assays suggest that this motif is recognized and

bound by members of the basic region-helix-loop-helix-leucine-zipper (bHLH-LZ)

transcription factor family, upstream stimulatory factor 1 (USF1) and USF2, and is

involved in the transcriptional activation of PTEN. Furthermore, we identified 1 CS

patient with a hemizygous germline deletion which localizes exclusively to this highly

conserved region upstream of PTEN.

3.2 MATERIALS AND METHODS

Comparative Genomic Analysis

Sequence data spanning 163 kb of the PTEN locus, to include the entire PTEN

gene (103 kb) and 30 kb of flanking sequence, for Homo sapiens (chromosome 10,

position 89,583,175-89,746,111, March 2006 Human Genome Assembly, NCBI Build

36.1 (hg18)) was obtained from the UCSC Genome Browser (http://genome.ucsc.edu/).

Pair-wise sequence comparisons were carried out using the mVISTA software program

(http://genome.lbl.gov/vista/index.shtml) for syntenic regions from Mus musculus

(chromosome 19, position 32,793,494-32,916,025, February 2006 Mouse Genome

Assembly [mm8]), and Rattus norvegicus (chromosome 1, position 236,741,027-

236,867,261, November 2004 Rat Genome Assembly [rn4]) (103). For these comparisons, Homo sapiens was considered the base sequence and, prior to alignment, this sequence was masked for interspersed and simple repeat elements using the

RepeatMasker software program (http://www.repeatmasker.org/cgi- bin/WEBRepeatMasker). A sliding window 100bp in length was utilized to identify all contiguous subsegments that had a minimum sequence identity ≥70%.

Cell Lines and Culture

MCF-7 breast cancer cell lines and HeLa cervical cancer cell lines were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 100 units/ml each of Penicillin and Streptomycin and maintained at

37˚C with 5% CO2. Nuclear protein extracts were isolated from MCF-7 cells with a high-salt method using NE-PER nuclear and cytoplasmic extraction reagents according to the manufacturer’s protocol (Pierce, Rockford, IL). Proteasome and phosphatase inhibitors were added to the extraction buffers in the following concentrations: 2µg/ml aprotinin, 2µg/ml leupeptin, 0.75mg/ml PMSF, 0.2mM sodium orthovanadate, 25mM sodium fluoride, 10mM β-glycerophosphate, and 2µg/ml pepstatin A.

Lymphoblastoid cell lines from 2 control samples and 2 CS patients with previously measured decreased PTEN protein (1 with a known PTEN deletion and 1 with a newly identified deletion upstream of PTEN) were cultured in RPMI-1640 media supplemented with 20% FBS and 100 units/ml each of Penicillin and Streptomycin and maintained at 37˚C with 5% CO2. Total protein extracts were then isolated using M-PER 55

Mammalian Protein Extraction Reagent (Pierce) supplemented with protease and phosphatase inhibitors.

Electrophoretic Mobility Shift Assay (EMSA)

The 112bp region spanning position -2262 to -2151 was subjected to PCR amplification using the following oligonucleotide primers containing EcoRI restriction

sites (indicated in lowercase); forward 5’-

GTCAgaattcCCCGAGCAAAGGAAGAAGAC-3’ and reverse 5’-

GTCAgaattcGTCGGAACTACTTTCCGAAG-3’. The resultant amplicon was digested

with the EcoRI enzyme, purified, and radiolabeled by incubation with Klenow fragment

and alpha-32P dATP (3,000Ci/mmol, 10mCi/ml). 2µg of either HeLaScribe (Promega,

Madison, WI) or isolated MCF-7 cell nuclear extract was then incubated with 1ng of end- labeled DNA in binding buffer (10mM HEPES (pH 7.9), 4% glycerol, 50mM NaCl,

2.5mM MgCl2, 0.5mM DTT, 1µg/ml BSA, and 1µg poly dI-dC) for 20 min at room temperature. The resulting DNA-protein complexes were resolved on a 4% non- denaturing polyacrylamide gel for 3.5 hrs at 150V at 4˚C, dried, and exposed to autoradiographic film with intensifying screens at -80°C. Specific and non-specific cold competitors, comprised of a 100-molar excess of unlabeled probe and an amplicon spanning exon 14 of PHLPP (an unrelated DNA fragment located on chromosome 18; forward 5’-TTGCATGCAAAGAGTAGGAG-3’ and reverse 5’-

TATGAATCCCATTGCCAGTG-3’), respectively, were added to subsequent reactions, as indicated, in order to determine binding specificity. Additional EMSAs were carried out using radiolabeled 68bp amplicons generated using the following sets of primer pairs:

A) forward 5’- GTCAgaattcCCCGAGCAAAGGAAGAAGAC -3’ and reverse 5’-

GTCAgaattcGGAACTTTCCAAATTCCCAC -3’ and B) forward 5’-

GTCAgaattcGGGAGTGGGAATTTGGAAAG -3’ and reverse 5’-

GTCAgaattcGTCGGAACTACTTTCCGAAG -3’. Nucleotides represented in lowercase indicate the EcoRI sites. A series of competition experiments were performed using a

100-molar excess of double-stranded cold oligonucleotide probes: specific oligonucleotide probe (-2200 to -2167, 34bp) 5’-

AAGTTCCCCAACTAGGGACACACGTGACCTCCTT-3’, non-specific oligonucleotide probe (PTEN exon 5, 25bp) 5’-

GTAATGATATGTGCATATTTATTAC-3’, Myc-Max consensus oligonucleotide probe

5’-GGAAGCAGACCACGTGGTCTGCTTCC-3’ (Santa Cruz Biotechnology Inc., Santa

Cruz, CA), and Myc-Max mutant oligonucleotide probe 5’-

GGAAGCAGACCACGGAGTCTGCTTCC -3’ (Santa Cruz Biotechnology Inc.). As well, we constructed various mutant forms of the specific oligonucleotide probe (Table

3.1) and performed additional competition experiments. Subsequent EMSAs were also performed using recombinant USF1, Myc, and Max protein (Protein One, Bethesda, MD) as indicated in the corresponding figure legends.

For supershift EMSAs (SS-EMSAs), antibodies against USF1 (Santa Cruz

Biotechnology Inc., sc-229), USF2 (Santa Cruz Biotechnology Inc., sc-862), Myc (Santa

Cruz Biotechnology Inc., sc-764), Max (Santa Cruz Biotechnology Inc., sc-765), and IgG

(Santa Cruz Biotechnology Inc., sc-2027) were obtained. SS-EMSA binding reactions were performed as described above with the following modification: for each reaction

2µg of antibody was pre-incubated with 4µg of nuclear extract for 20 min at 4°C prior to

the addition of the appropriate radiolabeled probe and binding buffer. Following addition of the radiolabeled probe and binding buffer, the resulting reaction was continued for an additional 20 min at 4°C.

Luciferase Gene Reporter Constructs and Assays

The full-length PTEN promoter region located at position -1344 to -1 was PCR amplified and subcloned upstream of the firefly luciferase gene and into the NheI/XhoI sites of the pGL3.1-Basic vector (Promega) (pGL3-B-FL) as previously described (20).

In order to interrogate the conserved region located at position -2262 to -2151, an additional construct was made by inserting this region upstream of the full-length PTEN promoter in the pGL3-B-FL construct using the KpnI/NheI sites (pGL3-B-FL-2262-WT).

The -2262 to -2151 insert was amplified from genomic DNA using the following primers: 5’-GTCAggtaccCCCGAGCAAAGGAAGAAGAC-3’ and 5’-

GTCAgctagcGTCGGAACTACTTTCCGAAG-3’. KpnI and NheI restriction sites, repesctively, are indicated in lowercase for each primer. Additionally, the GeneTailor

Site-Directed Mutagenesis System (Invitrogen, Carlsbad, CA) was used to generate mutant reporter constructs containing mutated half-sites, core sequences, or alterations in the flanking E-box motif (pGL3-B-FL-2262-MT1 through pGL3-B-FL-2262-MT6)

(Table 3.1). All constructs were resequenced to confirm orientation and sequence integrity.

MCF-7 and HeLa cells were seeded using 1ml DMEM in 12-well culture plates

24 hrs prior to transient transfection such that they were 50 to 60% confluent at the time of transfection. Cells were cotransfected with 0.5µg reporter construct and 10ng pRL-TK

Renilla luciferase vector (Promega) using 1.5µl of FuGENE 6 (Roche Diagnostics,

Indianapolis, IN) for each transfection. pRL-TK Renilla luciferase activity was used to

control for transfection efficiency. 24-48 hrs post-transfection, cells were washed twice

with PBS and harvested using passive lysis buffer as described by the manufacturer

(Promega). Samples were analyzed for both firefly and Renilla luciferase activity by luminometry (Molecular Devices, Sunnyvale, CA) using Dual-Luciferase Reporter Assay

reagents according to the manufacturer’s protocol (Promega) and normalized to Renilla luciferase expression. For each construct, three independent transfection experiments were performed.

Mutation and Deletion Analysis

We screened 30 patients (15 with a diagnosis of CS and 15 classified as ‘CS-like’,

i.e., had some features of CS but did not meet operational diagnostic criteria) without

detectable germline PTEN mutations and with previously measured decreased PTEN

protein expression for genetic alterations at the PTEN E-box site. Briefly, polymerase

chain reactions (PCRs) were carried out using 12.5µl HotStarTaq Master Mix (Qiagen,

Valencia, CA), 10µM forward primer (5’-TCTCAGCATTTCCGAATCAG-3’), 10µM

reverse primer (5’-CTGATGATGAAAGCTGAGATGG -3’), and 20ng of template DNA

and used the following thermal cycling conditions: 95°C for 15 min, 34 cycles of 95°C

for 30 s, 55°C for 45 s, and 72°C for 2 min, followed by a 72°C final extension for 10

min. Purified PCR product was then subject to direct DNA resequencing (Lerner

Research Institute, Genomics Core) and analyzed using Lasergene 7.0 software

(DNASTAR, Inc., Madison, WI).

Real-time quantitative PCR was used to investigate potential micro-deletions in

30 PTEN mutation-negative patient samples. 5 normal control samples and 1 previously

determined deletion positive sample (spanning both the PTEN and BMPR1A genes) were assayed as negative and positive controls, respectively. Copy number determinations were made for the region flanking the PTEN E-box (position -2237 to -2058, F: 5’-

TGCCTCCGGAGCTATCACTG-3’ and R: 5’-TACGGAACGGTAGGAAGCTG-3’)

and for exon 7 of a control reference gene, GAPDH (F: 5’-

ATGCCTCCTGCACCACCAAC-3’ and R: 5’-AGTCTTGGATGAGAAAGGTG-3’).

Determination of gene copy number was assayed using 12.5µl SYBR Green PCR Master

Mix (Applied Biosystems, Foster City, CA), 10µM forward primer, 10µM reverse

primer, and 20ng of template DNA. Thermal cycling conditions comprised of 50°C for 2

min, 95°C for 10 min, and 40 cycles at 95°C for 15 s followed by 58°C for 1 min using

an ABI 7500 Sequence Detection System (Applied Biosystems). PCR efficiencies for

each amplicon were determined by standard curve analysis using serial dilutions of

genomic DNA from a control sample and ranged from 90-92% for these amplicons.

Copy number determinations for PTEN exon 1 were performed as previously described

(50). Additionally, gene copy number was also assessed at PTEN exons 2 and 5 as

previously described (104). Target and reference genes were assayed in duplicate for

each sample and subject to meltcurve analysis and subsequent gel electrophoresis in order

to determine amplicon specificity. Positive values were further assayed in at least two

additional independent experiments. Gene copy number was determined using the

comparative delta Ct method (2-∆∆Ct) as described by Livak et al. (67).

All subjects were enrolled by referral from centers throughout the United States,

Canada, and Europe following informed consent in accordance with the procedures

approved by the human subjects protection committees of each respective institution. CS

patients were classified in accordance with criteria established by the International

Cowden Consortium and curated by the National Comprehensive Cancer Network (4).

Western analysis

15µg of protein from each sample was separated on a 10% SDS-PAGE gel,

transferred to a nitrocellulose membrane, and subsequently blocked for nonspecific

binding using 5% milk in 1% Tris-buffered saline containing 0.1% Triton X-100 (TBST).

Membranes were incubated overnight with the following primary antibodies:

phosphorylated p44/p42 MAPK (Cell Signaling, Danvers, MA), phosphorylated Akt

(Cell Signaling), both diluted 1:1000 in 3% BSA, and β-actin (diluted 1:5000 in 3%

BSA; Sigma, St. Louis, MO). After incubation, membranes were washed with TBST, incubated with either anti-mouse IgG or anti-rabbit IgG secondary antibody (diluted

1:2500 in 5% milk; Promega), and the resulting protein bands were visualized by enhanced chemiluminescence (Amersham Pharmacia Corp., Piscataway, NJ).

Statistical Analysis

All results are expressed at the mean ± standard deviation from three separate experiments. Statistical analysis was performed using Student’s t test and results are considered significant at the P < 0.05 level.

3.3 RESULTS

Identification of Novel DNA-Protein Interactions at a Highly Conserved E-box Element

Upstream of the PTEN Promoter

Multi-species comparative genomic analysis was used to identify highly

conserved regions across a 163 kb region spanning the entire PTEN gene and including

30 kb of flanking sequence in three distantly related species, Homo sapiens, Mus

musculus, and Rattus norvegicus. This analysis yielded several highly conserved regions

(>70% identity, across 100bp), with the majority (20/25) localized to a region upstream

of the PTEN promoter and within the gene’s first intron (Fig. 3.1A). Of the regions

displaying sequence homology, one particular region, located at position -2250 to -2151,

exhibited approximately 80% sequence identity across 102bp of DNA among all three

species (Fig. 3.1B), compared to 70-77% for the other regions. Because of this extensive

conservation, we chose to examine this region further.

To identify potential DNA-protein complexes formed at this conserved site, we

performed mobility shift assays, utilizing HeLa nuclear extract and a PCR-amplified

DNA probe which spans the conserved site, inclusive of nucleotides -2262 to -2151. We

found that this region could indeed bind to nuclear protein, as a single retarded band,

indicative of the formation of a DNA-protein complex, was observed (Fig. 3.2A). The

formation of this complex was specifically inhibited in the presence of 100-molar excess

of the unlabeled DNA probe, but not by an excess of a non-specific unlabeled DNA

probe (Fig. 3.2A, lanes 3 and 4). In order to better localize the nucleotides involved in

this complex, we performed a subsequent mobility shift assay using overlapping DNA

probes which span -2262 to -2195 (A) and -2218 to -2151 (B) (Fig. 3.2B). A complex is

Figure 3.1. Comparative genomic analysis reveals a highly conserved E-box upstream of the PTEN locus. A) Schematic diagram of the PTEN gene and the mVISTA alignment from our three-species comparative genomic analysis. A 24 kb region, including 3 kb upstream of PTEN and approximately 20 kb of intron 1, is provided. The circled area indicates the region corresponding to -2262 to -2151. B) Three-species nucleotide alignment of this region. The highly conserved E-box element (located at position -2181 to -2176) is underlined. Asterisks indicate conserved nucleotides.

formed using the -2218 to -2151 probe, however no DNA-protein complex formation is

observed using the -2262 to -2195 probe (Fig. 3.2B, lanes 6 and 2, respectively). The

DNA-protein complex formed with the -2218 to -2151 probe is specifically competed with an excess of the unlabeled DNA probe, but not by an excess of the non-specific unlabeled DNA probe (Fig. 3.2B, lanes 7 and 8). These data strongly suggest that the

DNA-protein binding observed at this conserved site is localized to the nucleotides

spanning position -2218 to -2151.

Several putative transcription factor recognition sequences exist within the -2218

to -2151 region. Among these is a putative canonical E-box (CACGTG), located at

position -2181 to -2176, to which several proteins are known to bind. To aid in

identifying the most likely candidate proteins involved in the formation of this complex,

we relied on the TESS transcription factor prediction software program

(http://www.cbil.upenn.edu/cgi-bin/tess). Among the most significant predictions were members of the basic helix-loop-helix leucine zipper (bHLH-LZ) family of transcription factors; including USF, Myc, Myc-associated factor X (Max), and transcription factor E3

(TFE-3). bHLH-LZ proteins, also known as E proteins, specifically bind to DNA sequences containing the E-box consensus sequence, which is minimally defined by the hexameric CANNTG motif (105).

To determine whether the binding we observed along this fragment was occurring at this E-box element, we performed a series of mobility shift assays using various non- specific cold competitor probes. For the initial experiments, we utilized a 34-nt specific oligonucleotide probe (spanning position -2200 to -2167) containing the E-box and commercially available Myc-Max consensus and mutant oligonucleotide probes, the

former of which contains the consensus CACGTG E-box element, while the latter

contains mutations of the two 3’ nucleotides (CACGTG to CACGGA). In the presence

of an excess of the -2200 to -2167 specific probe, formation of the DNA-protein complex

was inhibited (Fig. 3.2C, lane 4). Futhermore, an excess of the Myc-Max consensus

oligonucleotide also efficiently inhibited formation of the DNA-protein complex at this

site, while an excess of the mutant oligonucleotide did not (Fig. 3.2C, lanes 7 and 8).

These data suggest that the complex formed along this fragment occurs at the putative E-

box located upstream of the PTEN promoter at position -2181 to -2176.

Identification of USF-specific Binding at the -2181 to -2176 E-box Element

It has previously been shown that the binding preferences of USF and Myc/Max

to DNA are determined by the sequence flanking the core consensus E-box element (106-

108). Using a specific competitor oligonucleotide probe, containing the core consensus

E-box flanked by native sequence, and various mutant oligonucleotide competitor probes

(Table 3.1), we chose to exploit this feature in order to discriminate which bHLH-LZ family members are likely involved in the formation of the DNA-protein complex observed at position -2181 to -2176. The mutant competitor probes contained mutated nucleotides either within the core E-box sequence or in the nucleotides flanking the consensus motif. Mutant oligonucleotides MT1, MT2, and MT3 contain mutated 5’ half- sites (AattGTGA), 3’ half-sites (ACACattA), or complete core (AattattA) sequences,

respectively. MT4 (ACAgcTGA) contains an inversion of the core sequence’s two

central nucleotides, a nucleotide change tolerated by USF but not Myc/Max dimers (106).

Both MT5 (tCACGTGA) and MT6 (cCtCACGTGACg) contain mutations in regions

Figure 3.2. Identification of a novel USF-specific DNA-protein interaction upstream of the PTEN promoter. A) EMSA using a probe spanning the -2262 to -2151 region. Lane

1: radiolabeled probe only (-), lane 2: radiolabeled probe plus HeLa nuclear extract (+), lane 3: radiolabeled probe, HeLa nuclear extract, and 100-molar excess of unlabeled specific competitor probe (Sp), lane 4: radiolabeled probe, HeLa nuclear extract, and

100-molar excess of unlabeled non-specific (PHLPP exon 14) competitor probe (NS). B)

EMSA using probes spanning -2262 to -2195 (A) and -2218 to -2151 (B). Lanes 1: and

5: radiolabeled probe only (-), lanes 2: and 6: radiolabeled probe plus HeLa nuclear

extract (+), lanes 3: and 7: radiolabeled probe, HeLa nuclear extract, and 100-molar

excess of unlabeled specific competitor probe (Sp), lanes 4: and 8: radiolabeled probe,

HeLa nuclear extract, and 100-molar excess of unlabeled non-specific (PHLPP exon 14,

274bp amplicon) competitor probe (NS). Continued

Figure 3.2. Continued. C) Binding observed at -2262 to -2151 was assessed using a series of specific and non-specific competitor probes. Lane 1: radiolabeled probe only

(-), lane 2: radiolabeled probe plus HeLa nuclear extract (+). In lanes 3-8, radiolabeled probe, HeLa nuclear extract, plus 100-molar excess of the following unlabeled competitors were added: lane 3: -2262 to -2151 (specific, 112bp [Sp]), lane 4: -2200 to -

2167 (specific, 34bp [Sp]), lane 5: PHLPP exon 14 (non-specific, 274bp [NS]), lane 6:

PTEN exon 5 (non-specific, 25bp [NS]), lane 7: Myc-Max consensus (specific, 26bp

[Con]), lane 8: Myc-Max mutant (non-specific, 26bp [MT]). D) Binding at the putative

E-box located at position -2181 to -2176 was evaluated using the -2262 to -2151 radiolabeled probe and mutated competitor probes. Lane 1: radiolabeled probe only (-), lane 2: radiolabeled probe plus HeLa nuclear extract (+). In lanes 3-9, radiolabeled probe, HeLa nuclear extract, plus 100-molar excess of unlabeled specific competitor (Sp) and mutated competitor probes MT1 through MT6 were added, respectively.

Designation Oligonucleotide Sequencea Wild-type (WT) 5’- AAGTTCCCCAACTAGGGACACACGTGACCTCCTT-3’

Mutant 1 (MT1) 5’- AAGTTCCCCAACTAGGGACAattGTGACCTCCTT-3’

Mutant 2 (MT2) 5’-AAGTTCCCCAACTAGGGACACACattACCTCCTT-3’

Mutant 3 (MT3) 5’- AAGTTCCCCAACTAGGGACAattattACCTCCTT-3’

Mutant 4 (MT4) 5’- AAGTTCCCCAACTAGGGACACAgcTGACCTCCTT-3’

Mutant 5 (MT5) 5’- AAGTTCCCCAACTAGGGACtCACGTGACCTCCTT-3’

Mutant 6 (MT6) 5’- AAGTTCCCCAACTAGGGcCtCACGTGACgTCCTT-3’ a E-box (position -2181 to -2176) is indicated in italics. Mutated nucleotides are indicated in lowercase.

Table 3.1. Mutant oligonucleotide competitor probes/sequence of mutant reporter constructs.

flanking the core sequence which disfavors the binding of Myc/Max and Max/Max complexes (109). An excess of either of the E-box half-site mutant oligonucleotides are able to diminish binding at this site (Fig. 3.2D, lanes 4 and 5), while MT3, with a completely mutated E-box motif, is unable to compete with the -2262 to -2151 radiolabeled probe. In contrast, the MT4, MT5, and MT6 mutant oligonucleotides, which favor USF binding, all efficiently inhibit formation of the DNA-protein complex, suggesting that the complex formed at the -2181 to -2176 E-box is due to the binding of

USF.

The -2181 to -2176 E-box Element is Involved in Transcriptional Activation

To determine the functional significance of the -2181 to -2176 E-box, we chose to

examine the conserved region (position -2262 to -2151) using a luciferase reporter assay.

The full-length PTEN promoter was subcloned into the pGL3.1-Basic vector (pGL3-B-

FL). Additionally, the -2262 to -2151 conserved region was subcloned upstream of the

full-length PTEN promoter and into the pGL3-B-FL vector (pGL3-B-FL-2262-WT) [Fig.

3.3A]. Both constructs, along with the pGL3-B empty vector, were transiently transfected

into HeLa and MCF-7 cells and assayed for luciferase activity. The luciferase activity of

the pGL3-B-FL and pGL3-B-FL-2262-WT constructs did not differ significantly

following transfection in MCF-7 cells (P = 0.132) [Fig. 3.3B]. However, a significant

increase in reporter gene activity was observed following transfection of these constructs

in HeLa cells (P = 0.013). These results indicate that, in HeLa cells, a cis-acting element(s) contained within the -2262 to -2151 conserved site is able to induce reporter

gene transcription approximately 60% above that of the full-length PTEN promoter alone.

Interestingly, it has previously been reported that MCF-7 cells, along with other breast

cancer cell lines, express USF but that, in these particular cells, this protein lacks

transcriptional activity (110). Our observation that this conserved region failed to induce

transcription in MCF-7 cells supports this report and also further indicates that USF

proteins are involved in the transactivation of PTEN.

We also generated 6 mutant reporter constructs, each containing mutations to either the -2181 to -2167 E-box consensus sequence or to its adjacent flanking sequence

(Table 3.1 and Fig. 3.3A). All 6 constructs, along with the wild-type construct, were transfected into HeLa cells. pGL3-B-FL-2262-MT1, pGL3-B-FL-2262-MT2, and pGL3-

B-FL-2262-MT3, which contain mutated half-sites or a completely mutated core,

displayed a significant reduction in luciferase activity compared to the wild-type

construct (P ≤ 0.02) (Fig. 3.3C). Similarly, pGL3-B-FL-2262-MT4 also resulted in a

marked decrease. pGL3-B-FL-2262-MT5 and pGL3-B-FL-2262-MT6, both of which

contain alterations flanking the E-box motif, did not differ significantly from the wild-

type construct (P > 0.05).

Together, these data further suggest that the -2181 to -2167 E-box is involved in

the transactivation of PTEN and that this activity is mediated primarily by USF proteins.

Figure 3.3. The -2262 to -2151 conserved region is involved in transcriptional activation.

A) Schematic diagram of constructs used to assess the transcriptional activity of the -

2262 to -2151 region: pGL3-B (empty vector), pGL3-B-FL (full-length PTEN promoter

upstream of pGL3-B), pGL3-B-FL-2262-WT (-2262 to -2151 region upstream of the full-

length PTEN promoter and pGL3-B), and pGL3-B-FL-2262-MT1 through pGL3-B-FL-

2262-MT6 (constructs containing mutant -2181 to -2176 E-box/flanking sequences

within pGL3-B-FL-2262-WT). Continued

Figure 3.3. Continued. B) All three constructs were transiently cotransfected in either

MCF-7 or HeLa cells along with the pRL-TK internal control plasmid and assayed for luciferase activity. Firefly luciferase measurements from three independent experiments were corrected for transfection efficiency using the Renilla luciferase internal control, averaged, and then normalized to the full-length PTEN promoter construct (pGL3-B-FL).

Error bars represent the standard deviation of the three independent experiments. *P- value for pGL3-B-FL versus pGL3-B-FL-2262-WT in MCF-7 cells = 0.132. **P-value for pGL3-B-FL versus pGL3-B-FL-2262-WT in HeLa cells = 0.013. Continued

Figure 3.3. Continued. C) Mutant reporter constructs (pGL3-B-FL-2262-MT1 through pGL3-B-FL-2262-MT6) were transfected into HeLa cells along with pRL-TK and assayed for firefly and Renilla luciferase activity. pGL3-B-FL-2262-MT1, pGL3-B-FL-

2262-MT2, and pGL3-B-FL-2262-MT3 displayed a significant reduction in luciferase activity compared to the wild-type construct (P ≤ 0.02). Similarly, pGL3-B-FL-2262-

MT4 also resulted in a marked decrease. pGL3-B-FL-2262-MT5 and pGL3-B-FL-2262-

MT6, did not differ significantly from the wild-type construct (P > 0.05).

USF1 and USF2, and not Myc/Max, Bind the -2181 to -2176 PTEN E-box Element

The absence of multiple shifted bands in Figures 3.1 and 3.2 suggests that a single

DNA-protein complex is formed at the -2262 to -2151 site. Taken together with data demonstrating that the sequence flanking the -2181 to -2176 E-box element favors USF binding, this suggests that this DNA-protein complex is formed by USF protein binding.

To examine this further, we performed mobility shift assays where antibodies raised against members of the bHLH-LZ family of transcription factors, specifically; anti-USF1, anti-USF2, anti-Myc, and anti-Max, were pre-incubated with either HeLa (Fig. 3.4A) or

MCF-7 (Fig. 3.4B) nuclear extract. The USF protein complex can function either as a homodimer, consisting of either USF1 or USF2 dimers, or as a USF1/USF2 heterodimer, with the heterodimeric complex being predominant (111). Myc, on the other hand, exerts is transcriptional activity as a heterodimer in complex with Max (112). Because of this, each antibody was assayed singly as well as in their respective heterodimeric combination (i.e. anti-USF1/anti-USF2 and anti-Myc/anti-Max). The addition of anti-

USF1, anti-USF2, and anti-USF1/anti-USF2 each resulted in a shift of the specific band formed by this DNA-protein complex in the presence of HeLa nuclear extract compared to the complex formed in the absence of antibody (Fig. 3.4A, lanes 3, 4, and 5 versus lane

2). No shift was observed with the addition of anti-Myc, anti-Max, anti-Myc/anti-Max

(Fig. 3.4A, lanes 6 through 8). As expected, a shift in the DNA-protein complex was not observed when anti-IgG was added to the reaction (Fig 3.4A, lane 9). Similar results were observed in the presence of MCF-7 nuclear extract (Fig. 3.4B). These results demonstrate that in vitro, the native PTEN E-box allows for binding of the USF1 and

USF2 proteins in both HeLa and MCF-7 nuclear extract.

Our analysis suggests that the USF proteins, and not Myc/Max, bind upstream of

the PTEN promoter and transactivates its gene expression. However, the Myc/Max heterodimer can be difficult to assay in nuclear extract isolated from mammalian cells

(113, 114). Furthermore, Myc is a low abundant protein in many mammalian cells and tissues (115). Because of this, and to better understand which bHLH-LZ protein(s) binds the -2181 to -2176 E-box, we chose to further assess this region using USF1, Myc, and

Max recombinant proteins. USF1 specifically binds to the -2181 to -2176 E-box (Fig.

3.5, lane 2 and Fig. 3.6, lane 2). Formation of this DNA-protein complex is abolished with the addition of excess specific competitor (Fig. 3.5, lane 3 and Fig. 3.6, lane 3).

Contrary to this, mutated competitor probes MT1, MT2, and MT3 do not compete with this binding reaction (Fig. 3.5, lanes 4, 5, and 6). The MT4 oligonucleotide probe, in which the two central nucleotides of the consensus E-box are inverted (CG to GC) is able to partially compete this binding, however, because USF1 has a higher affinity for the consensus sequence, a retarded band is observed (Fig. 3.5, lane 7). MT5 and MT6, both of which retain the consensus E-box element but contain mutated nucleotides in the flanking sequence to favor USF binding, are able to efficiently compete with the formation of this complex (Fig. 3.5, lanes 8 and 9). Moreover, the addition of anti-USF1 antibody to this reaction resulted in a super-shifted DNA-protein complex (Fig. 3.5, lanes

10 and 12), while the addition of the control anti-USF2 antibody did not (Fig. 3.5, lane

11).

Next, the formation of a DNA-protein complex was assayed using Myc and Max recombinant proteins, both singly and in combination (i.e. Myc/Max) to interrogate formation of the Myc/Max heterodimer (Fig. 3.6, lanes 4-19). In contrast to the DNA-

Figure 3.4. USF proteins, and not Myc/Max, bind to the -2262 to -2151 conserved region. SS-EMSAs were performed using the -2262 to -2151 region probe and antibodies against USF1, USF2, Myc, Max, and IgG. A) Lane 1: radiolabeled probe only (-), lane 2: radiolabeled probe plus HeLa nuclear extract (+). In lanes 3-9, radiolabeled probe, HeLa nuclear extract, plus α-USF1, α-USF2, α-USF1/α-USF2, α-Myc, α-Max, α-Myc/α-

Max, and α-IgG antibodies were added, respectively. B) Lane 1: radiolabeled probe only

(-), lane 2: radiolabeled probe plus MCF-7 nuclear extract (+). In lanes 3-9, radiolabeled probe, MCF-7 nuclear extract, plus α-USF1, α-USF2, α-USF1/α-USF2, α-Myc, α-Max,

α-Myc/α-Max, and α-IgG antibodies were added, respectively.

protein complex observed with USF1 recombinant protein (Fig. 3.5, lane 2 and Fig. 3.6,

lane 2), no complex was observed in the presence of recombinant Myc (Fig. 3.6, lanes 4 and 5). Interestingly, a faint, apparently specific band was observed in the presence of recombinant Max and Myc/Max proteins (Fig. 3.6, lanes 6-9), suggesting that these proteins may bind to the -2181 to -2176 E-box and compete for binding at this site.

However, this interaction appears to have significantly less affinity than that observed with USF1. Additionally, Max and Myc/Max protein were each able to bind to a radiolabeled Myc-Max consensus oligonucleotide probe (Fig. 3.6, lanes 12 and 14).

Formation of this complex failed to occur in the presence of the radiolabeled Myc-Max mutant oligonucleotide probe (Fig. 3.6, lanes 18 and 19).

Taken together, these data provide further evidence that the highly conserved

PTEN E-box is specifically and primarily bound by USF.

Mutation Analysis of E-box Region in CS/CS-like Patients

To begin to assess the pathogenic role of the PTEN E-box element in CS, we

analyzed this site along with 1.6 kb of flanking sequence (position -2895 to -1295) for

germline genetic alterations in 30 previously identified germline PTEN mutation-negative patients but with decreased PTEN protein levels and with classic CS (N=15) or CS-like

(N=15) diagnostic features. No nucleotide variants were identified within the -2180 to -

2176 E-box motif or within the adjacent sequence in these samples. Based on these data, we suspected that germline point mutations in the E-box consensus binding motif or

Figure 3.5. USF1 specifically binds to the -2181 to -2176 PTEN E-box element. EMSA

and SS-EMSA assays were carried out using recombinant USF1 protein. Lane 1:

radiolabeled probe only (-), lane 2: radiolabeled probe plus recombinant USF1 protein

(+). In lanes 3-9, radiolabeled probe, recombinant USF1 protein, plus 100-molar excess

of unlabeled specific competitor (Sp) and mutated competitor probes MT1 through MT6

were added, respectively. In lanes 10-12, SS-EMSAs were carried out using α-USF1, α-

USF2, and α-USF1/α-USF2 antibodies, respectively. 77

Figure 3.6. Myc/Max does not bind to the -2181 to -2176 PTEN E-box element.

EMSAs were carried out using either recombinant USF1, Myc, Max, or Myc/Max

protein. Lane 1: radiolabeled probe only (-2262 to -2151) [-]. Lanes 2, 4, 6, and 8

radiolabled probe (-2262 to -2151) were incubated using USF1, Myc, Max, or Myc/Max

recombinant protein, respectively (+). In lanes 3, 5, 7, and 9, these same reactions were

carried out in the presence of 100-molar excess of unlabeled specific competitor probe

(Sp). Lanes 10 and 16: radiolabeled Myc-Max consensus and Myc-Max mutant probes,

respectively (-). Lanes 11, 12, 14 and lanes 17, 18, and 19 each were incubated with

radiolabled probe (Myc-Max consensus: lanes 11, 12, and 14; Myc-Max mutant: lanes

17, 18, 19) and Myc, Max, or Myc/Max recombinant protein, respectively. In lanes 13 and 15, reactions in the presence of Max and Myc/Max recombinant protein, respectively, were carried out in the presence of 100-molar excess of unlabeled Myc-Max consensus probe (Sp).

within its adjacent flanking sequences are likely to be rare events in CS and CS-like

patients.

Deletions of PTEN have previously been identified only in BRRS which

encompass the whole gene or localize to its 5’ end, typically including exon 1 and often

extending through exon 5 (50, 104). To date, all germline deletions include at least exon

1 of PTEN. In other words, no deletions exclusively involving the highly conserved

region upstream of PTEN, which includes both the PTEN full-length promoter and a

putative CpG island (spanning position -2453 to -99), have been described. To assess

whether deletions localize to this region, we performed real-time quantitative PCR in this

same 30 PTEN mutation-negative CS/CS-like patients. Copy number determinations

were estimated for all samples, including 5 controls and 1 positive control sample with a

known deletion, for regions targeting position -2237 to -2058 and PTEN exon 1 (c.52 to

c.79+57). 2-∆∆Ct values among the control samples were 1.01 ± 0.09 and 1.00 ± 0.05 at

each of the two regions, respectively, compared to 0.57 ± 0.05 and 0.60 ±0.03 for the

known deletion positive sample. Among the 30 CS/CS-like patients, 3 were found to

have 2-∆∆Ct values suggestive of a hemizygous deletion at both the upstream E-box region

and exon 1 (data not shown). Among 15 CS patients, 1 (7%) (11099-01) was found to be

hemizygous only at the region encompassing the PTEN E-box element (0.52 ± 0.06 for

the region spanning -2237 to -2058, compared to 0.89 ± 0.06 for exon 1) (Fig. 3.7A).

Subsequent copy number determinations targeting PTEN exons 2 (c.80-51 to c.141, 2-∆∆Ct

= 1.02 ± 0.08) and 5 (c.388 to c.492+40, 2-∆∆Ct = 1.06 ± 0.03) in this patient indicate that the entire PTEN gene proper is biallelic. Furthermore, we demonstrated that sample

Figure 3.7. Identification of a functional hemizygous germline deletion upstream of the

PTEN coding sequence. A) Real-time quantitative PCR was used to investigate potential

micro-deletions across the PTEN locus. Copy number determinations at position -2237

to -2058 and PTEN exons 1, 2, and 5 among 5 controls samples were indicative of two

alleles at each respective location (2-∆∆Ct = 1.00 ± 0.03 to 1.05 ± 0.12). 1 known deletion positive sample displayed 2-∆∆Ct values consistent with a hemizygous deletion across all 4 regions (0.52 ± 0.04 to 0.60 ± 0.03). 1 PTEN mutation-negative sample exhibited 2-∆∆Ct values similar to those observed in the control samples for exons 1, 2, and 5 (0.89 ± 0.06 to 1.05 ± 0.03). In this same sample, copy number determinations targeting position -

2237 to -2058 were indicative of a hemizygous deletion exclusively at this upstream region (2-∆∆Ct = 0.52 ± 0.06). Continued

Figure 3.7. Continued. B) Western analysis reveals that the upstream deletion results in an increase in P-AKT levels, the downstream target of PTEN’s lipid phosphatase activity, compared to normal control samples. Additionally, P-p42/44-MAPK, the downstream target of PTEN’s protein phosphatase activity, is also slightly increased in the sample harboring this deletion. Samples from normal control subjects (Control 1 and Control 2), a known PTEN deletion positive sample (Del. Positive), and the mutation-negative sample with a novel deletion upstream of PTEN (11099-01) were assayed.

11099-01 exhibited an increase in P-Akt, the downstream target of PTEN’s lipid phosphatase activity, and a slight increase in P-p44/42-MAPK, the target of PTEN’s protein phosphatase activity, compared to controls (Fig. 3.7B). These data further suggest that aberrant regulation of PTEN is a manifestation of this deletion of the E-box

region, affirming the integral role of this highly conserved upstream region in PTEN

regulation, and points to a novel mechanism of disease etiology in CS patients.

3.4 DISCUSSION

Germline mutations in PTEN, the second most commonly mutated gene in all human cancers, are primarily associated with a number of clinically distinct heritable cancer syndromes, collectively referred to as PTEN hamartoma tumor syndrome (PHTS)

(57). PHTS is characterized by multiple hamartomatous lesions affecting derivatives of all three germ cell layers and includes both CS and BRRS (51). Germline PTEN mutations have been identified in 85% of patients diagnosed with CS and in 65% of patients diagnosed with BRRS (50, 57). In a subset of classic CS patients lacking mutations in the PTEN coding sequence, approximately 10% were found to harbor

mutations in the gene’s core promoter (50). Furthermore, we also identified large

germline hemizygous PTEN deletions, spanning the 5’ region of PTEN and extending to

its upstream region, in 11% of mutation-negative BRRS patients. All previously

identified deletions have included at least 1 exon of PTEN. For the remaining 15% and

35% of CS and BRRS patients, respectively, the etiology of their disease remains

unknown.

In order to better understand the mechanism(s) underlying the deregulation of

PTEN in patients lacking mutations, in those with 5’ deletions and mutations in the

promoter, as well as those contributing to its phenotypic complexity, we sought to

identify functional cis-regulatory elements involved in its regulation. We have

previously shown that CS and BRRS patients lacking mutations in the PTEN gene and its

core promoter are strongly associated with a haplotype block encompassing the region

upstream of this gene and including a portion of its first intron (104). Because mutation

scanning has failed to identify pathogenic mutations in these individuals, we

hypothesized that other mechanisms potentially exist which can disrupt normal PTEN

function and result in pathogenesis. Genetic alterations at cis-regulatory sites, either structural or functional in nature, could potentially alter the normal regulation of PTEN

and, thus, represent one such mechanism.

Comparative genomic analysis has been shown to be a powerful method for

identifying potential novel regulatory elements in genomic sequence. Using such an

approach combined with cellular biology techniques, we demonstrate here that both

USF1 and USF2 can bind to a highly conserved, novel PTEN E-box element. This

element is located at position -2181 to -2176 relative to the PTEN translation initiation

codon and within a conserved 102-bp fragment upstream of the PTEN core promoter.

USF binding at the PTEN E-box element is both specific and functional, as we have

demonstrated that, in cooperation with the full-length PTEN promoter, this region is able

to mediate a 60% increase in reporter gene transactivation, compared to the full-length

PTEN promoter alone. Interestingly, this response was only observed in HeLa cells and

not in MCF-7 cells. This observation supports the findings of Ismail et al., who described

the loss of USF transcriptional activity in several breast cancer cell lines, including MCF-

7 cells, and provides additional evidence supporting the notion that the transactivation we

observed was indeed mediated by USF proteins (110). Furthermore, mutation of the

putative E-box site abolished its ability to bind USF. In addition, mutation of the -2181

to -2176 E-box core sequence in the pGL3-B-FL-2262 reporter construct results in a

significant decrease in reporter gene activity compared to the wild-type construct.

USF1 and USF2 are ubiquitously expressed transcription factors comprised of

highly conserved basic regions and both helix-loop-helix (HLH) and leucine zipper (LZ)

domains (105, 116, 117). Residues forming the basic region enable USF proteins to bind

DNA at consensus E-box motifs (CANNTG), while the HLH and LZ domains are

primarily involved in its dimerization. Although several members of the bHLH-LZ

family can recognize this motif, previous studies have shown that specific binding is

dependent upon the flanking sequence surrounding the E-box site (106, 108). Based on

these studies, the sequence flanking the -2181 to -2176 E-box element appears to favor

USF-specific binding. Through a series of EMSA experiments using mutant competitor

probes designed in consideration of this evidence, we show that the -2181 to -2176 E-box is preferentially bound by USF. Moreover, using antibodies to specific bHLH-LZ proteins, SS-EMSAs further support these findings. Both USF1 and Myc/Max recombinant protein experiments provide clear evidence that, despite the redundancy with which proteins are able to bind to this motif, the DNA-protein complex formed at the -2181 to -2176 E-box is USF-specific.

USF proteins are involved in the transactivation of many human genes involved in a variety of cellular processes, including regulation of the cell cycle. In addition to regulating the expression of genes involved in the transition from G1/S and G2/M, including both cyclin B1 and Cdk1, USFs have also been shown to regulate known tumor suppressor genes, including p53, BRCA2, and APC (118-122). In the present study, we

show that PTEN, a tumor suppressor phosphatase involved in regulating the cell cycle, is

also transcriptionally regulated by USF.

p53 has been shown to inhibit USF-dependent BRCA2 promoter activation by preventing USF from binding to BRCA2’s minimal promoter region, resulting in a

decrease in both BRCA2 mRNA and protein levels (123). Although the precise

mechanism by which p53 inhibits USF binding to the BRCA2 promoter has yet to be

identified, p53’s ability to regulate BRCA2’s transcriptional activity suggests a

regulatory loop among these important tumor suppressor genes. We and others have

previously shown similar co-regulation exists among p53 and PTEN (19, 20, 124, 125).

In addition to regulating PTEN gene expression, through the p53-binding site located in

its promoter, PTEN and p53 also physically interact, forming a complex which serves to

autoregulate PTEN’s own expression. p53 and BRCA2’s USF-mediated interaction

suggests an additional mechanism by which p53 may regulate PTEN mRNA and protein

expression.

A recent study by Nowak et al. described the involvement of PI3K in regulating

USF-mediated transactivation of APOA5 (126). Together with our evidence suggesting

that USF transactivates PTEN expression, this presents a potential self-regulatory

mechanism involving the PI3K/Akt pathway. PTEN, the major 3-phosphatase in the

PI3K/Akt pathway, antagonizes PI3K’s activation of Akt by dephosphorylating

phosphatidylinositol 3,4,5-trisphosphate (PIP3) to phosphatidylinositol 4,5-bisphosphate

(PIP2) (12). Phosphorylation of USF1 by PI3K prevents it from binding to the APOA5

E-box motif, thereby modulating its ability to transactivate APOA5 expression.

Similarly, PI3K-dependent phosphorylation of USF1 could modulate binding at the

PTEN E-box, resulting in PI3K-mediated transrepression of PTEN, however this remains to be examined and is beyond the scope of the present study.

In the present study, we provide evidence suggesting that USF proteins regulate

the PTEN tumor suppressor gene. The precise mechanism by which USF1 and USF2

regulate PTEN expression through the -2181 to -2176 E-box is unknown. USF1 has been

shown to interact with other proteins involved in the assembly of the transcription

preinitiation complex. Specifically, it has been described that USF1 can interact with

TFII-I, an initiator-binding protein, and also bind the TFIID complex (127-129).

Importantly, Roy et al. also demonstrated that TFII-I can recruit the TFIID complex onto

a TATA-less promoter (130). As has been postulated with the TATA-less promoter of

APC, USF1 and USF2 may also interact through the TFII-I/TFIID complex at PTEN’s

TATA-less promoter and thereby facilitate its transactivation (121).

Based on previous studies describing USFs role in regulating other tumor

suppressor genes and its anti-proliferative properties, deregulation of members of the

USF protein family potentially has a role in tumorigenesis. USF genetic or epigenetic

alterations may not only be important in carcinogenesis, but may also contribute to the

phenotypic diversity observed in complex cancer syndromes. To date, however, no

mutations have been described linking USF to cancer susceptibility. The only identified association has been with a familial combined hyperlipidemia (FCHL) risk haplotype involving two single nucleotide polymorphisms (SNPs) located on the USF1 locus (131).

However, given their involvement in regulating cellular growth, genetic alterations in the

USF genes may contribute to cancer susceptibility, either directly or perhaps through a modifying effect, by altering tumor suppressor gene regulation. The observation that

cancer cell lines frequently lose USF transcriptional activity supports the potential role of

their deregulation in carcinogenesis (110).

Given these data, we had hypothesized that alterations, whether structural or functional, at the PTEN E-box could contribute to altered PTEN expression and underlie disease in CS/CS-like patients with previously unidentified mutations and contribute to its phenotypic complexity. We have shown here that at least a subset of CS/CS-like patients without previously identified intragenic PTEN and promoter mutations harbor germline mutations of the E-box located more than 2 kb upstream of the gene’s translation start site, but does not include its coding sequence. The pathogenicity is

confirmed by an associated concomitant decrease in PTEN and upregulation of both

downstream targets of PTEN’s lipid and protein phosphatase activities. The CS patient with a germline hemizygous deletion of the E-box region, without involvement of PTEN or its promoter, has clinical features that include early onset breast cancer and macrocephaly. Interestingly, so far, germline promoter mutations in CS patients are strongly associated with breast pathology. Here, this patient with the E-box region deletion shows this similar phenotype.

Given PTEN‘s paramount role in both the cell cycle’s pro-apoptotic pathway and

in carcinogenesis, the intricacies of its transcriptional regulation necessitate further

elucidation. We have identified a novel hexanucleotide E-box motif (CACGTG) located

upstream of PTEN, which is specifically bound by the USF transcription factors USF1

and USF2 and is involved in the transactivation of PTEN. Using a multi-species

comparative genomic approach approach, this study offers the first detailed look at

evolutionarily conserved sequence elements along and flanking the PTEN locus and

suggests that a comprehensive investigation of other conserved regions is warranted to better understand PTEN’s complex regulatory pathways. Because there appears to be a range of transcription factor binding sites within and even upstream of PTEN’s promoter region, as evidenced by this and previous reports, we can speculate that differential relative utilization of these different transcription factor binding sites, either because of structural variation or other unknown non-genetic mechanisms, can contribute to the diverse phenotypes of PHTS (20, 50, 104).

CHAPTER 4

miR-519e NEGATIVELY REGULATES THE TUMOR SUPPRESSOR PHOSPHATASE PTEN IN HUMAN CANCER CELLS

4.1 INTRODUCTION

Phosphatase and tensin homolog deleted on chromosome ten (PTEN [MIM

601728]) {also known as mutated in multiple advanced cancers 1 (MMAC1) and tensin- like phosphatase 1 (TEP1)} encodes a tumor suppressor phosphatase that antagonizes both the phosphoinositol-3-kinase (PI3K)/Akt and the mitogen-activated protein kinase

(MAPK) pathways causing apoptosis and cell cycle arrest (7, 12, 13, 21, 89, 90).

Germline mutations in PTEN are associated with several distinct disorders characterized by multiple hamartomatous lesions, fibrocystic disease, macrocephaly, and developmental delay. These include both Cowden syndrome (CS [MIM 158350]) and

Bannayan-Riley-Ruvalcaba syndrome (BRRS [MIM153480]) (4, 51). In addition to its benign manifestations, patients with CS are at increased risk of developing malignant neoplasias of the breast, thyroid (especially follicular thyroid carcinoma), and endometrium. Germline PTEN mutations have been identified in the majority of patients diagnosed with CS (85%) and BRRS (65%). However, for a subset of patients sharing

similar phenotypic features and meeting diagnostic criteria, the pathogenic cause(s)

remains unknown (50). Moreover, many of these patients without germline mutations

exhibit decreased PTEN protein expression, suggesting that alternate mechanisms may

underlie their dysfunction (Waite and Eng, unpublished data).

A growing list of non-coding gene regulators, microRNAs (miRNAs), have been

shown to inhibit gene expression post-transcriptionally, through either mRNA

degradation or translational repression (132). The involvement of these small non-coding

RNAs has been implicated in human disease, including a number of cancers (133).

Mounting evidence also suggests an involvement in cardiovascular disease, Alzheimer

disease, and several metabolic diseases, including type 2 diabetes mellitus (134-139).

Furthermore, miRNAs are thought to impact phenotypic variation and, therefore, may

modulate phenotypic expression even in the presence of identical germline mutations.

Such is the case in both CS and BRRS (140). However, the identification of

miRNA:target gene interactions has been challenging. Currently, there are 475 human

miRNAs listed in miRBase (http://microrna.sanger.ac.uk/), the miRNA curation database

(141). While predictions estimate that these miRNAs likely target the majority of protein-coding genes, only a handful of these biological targets have been experimentally

verified (142).

In order to better understand both the mechanisms involved in PTEN’s regulation and the potential causes of its dysregulation, we sought to assess the functionality of miRNAs computationally predicted to target and regulate PTEN.

4.2 MATERIALS AND METHODS

Cell Lines and Culture

MCF-7 breast cancer and WRO82-1 follicular thyroid cancer (FTC) cell lines

were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10%

fetal bovine serum (FBS) and 100 units/ml each of Penicillin and Streptomycin and

maintained at 37˚C with 5% CO2. MDA-MB-231 and MDA-MB-435 breast cancer cell

lines were cultured in RPMI-1640 media supplemented with 10% FBS, 10mM HEPES,

0.2 units/ml bovine insulin, and 100 units/ml each of Penicillin and Streptomycin and maintained at 37˚C with 5% CO2.

Luciferase Gene Reporter Constructs

PTEN’s 3’UTR, from position +10 to +930, was PCR amplified using the following primer: 5’-AGTCtctagaTATCAAGAGGGATAAAACAC-3’ and 5’-

AGTCtctagaGGAGATGGAGAAGTCGTTAC-3’ (nucleotides (nt) in lowercase indicate

the XbaI restriction sites of each primer). The resulting 931-basepair (bp) fragment was

then subcloned downstream of the firefly luciferase gene and upstream of the SV40 late

poly(A) signal using the XbaI site of the pGL3.1-Promoter vector (Promega, Madison,

WI). PTEN’s 3’UTR sequence was subcloned in both the sense (pGL3-PTEN3’UTR/5-

3) and antisense orientation (pGL3-PTEN3’UTR/3-5). Additionally, the GeneTailor

Site-Directed Mutagenesis System (Invitrogen, Carlsbad, CA) was used to generate a

mutant reporter construct containing a 7-bp deletion of the miR-519e seed binding site

(GGCACTT, position +271 to +277) using the following primers: 5’-

CAATTAGGATTAATAAAGATTCCCGTTTTA-3’ and 5’-

ATCTTTATTAATCCTAATTGAATTTTAAAT-3’ (pGL3-PTEN3’UTR/5-3∆519e).

All constructs were resequenced to confirm sequence integrity and ensure the desired

orientation.

Transient Transfection of miRNAs

Precursor miRNAs for hsa-miR-26a, hsa-miR-519e, hsa-miR-20a, hsa-miR-22, or miRNA Negative Control #1 (Ambion, Austin, TX) were transiently co-transfected into

MCF-7 cells along with either the pGL3-PTEN3’UTR/5-3 or pGL3-PTEN3’UTR/3-5 reporter construct. 24 hrs prior to transfection, cells were seeded using 1ml DMEM in

12-well culture plates such that they were 50 to 60% confluent at the time of transfection.

Cells were transfected with 0.5µg reporter construct, 10ng pRL-TK Renilla luciferase

vector (Promega), and 10nM precursor miRNA using 5.0µl of DMRIE-C (Invitrogen) for

each transfection. Subsequent experiments also include the transfection of the hsa-miR-

519d precursor miRNA molecule (Ambion). pRL-TK Renilla luciferase activity was used to control for transfection efficiency. 48 hrs post-transfection, cells were washed twice with PBS and harvested using passive lysis buffer as described by the manufacturer

(Promega). Samples were analyzed for both firefly and Renilla luciferase activity by luminometry (Molecular Devices, Sunnyvale, CA) using Dual-Luciferase Reporter Assay reagents according to the manufacturer’s protocol (Promega) and normalized to Renilla

Luciferase expression. For each construct and miRNA, three independent transfection experiments were performed.

Additionally, MCF-7 cells were seeded using 2ml DMEM in 6-well culture plates

24 hrs prior to transient transfection such that they were 50 to 60% confluent at the time

of transfection. Cells were co-transfected with 10nM of either hsa-miR-26a, hsa-miR-

519e, hsa-miR-20a, hsa-miR-22, or miRNA Negative Control #1 precursor miRNAs using 10.0µl of DMRIE-C (Invitrogen). As well, 10nM of anti-miR-519e miRNA inhibitor (Ambion) were subsequently co-transfected along with either hsa-miR-519e or miRNA Negative Control #1 precursor miRNAs. 72 hrs post-transfection, cells were washed twice with PBS and total protein extracts were isolated using M-PER Mammalian

Protein Extraction Reagent (Pierce, Rockford, IL) according to the manufacturer’s

protocol and supplemented with the following protease and phosphatase inhibitors:

2µg/ml aprotinin, 2µg/ml leupeptin, 0.75mg/ml PMSF, 0.2mM sodium orthovanadate,

25mM sodium fluoride, 10mM β-glycerophosphate, and 2µg/ml pepstatin A. Similar

experiments were subsequently performed using WRO82-1, MDA-MB-231, and MDA-

MB435 cells as indicated in the figure legends. Additionally, the hsa-miR-519d

precursor miRNA was assessed. All experiments were performed a minimum of three

times.

Western Analysis

15µg of protein from each sample was separated on a 10% SDS-PAGE gel,

transferred to a nitrocellulose membrane, and subsequently blocked for nonspecific

binding using 5% milk in 1% Tris-buffered saline containing 0.1% Triton X-100 (TBST).

Membranes were incubated overnight with the following primary antibodies: PTEN

(Cascade Biosciences, Waltham, MA), phosphorylated p44/p42 MAPK (P-MAPK) (Cell

Signaling, Danvers, MA), phosphorylated Akt (P-Akt) (Cell Signaling), Akt (Cell

Signaling) p44/p42 MAPK (Santa Cruz, Biotechnology Inc., Santa Cruz, CA) all diluted

1:1000 in 3% BSA, and β-actin (diluted 1:5000 in 3% BSA; Sigma, St. Louis, MO).

After incubation, membranes were washed with TBST, and incubated with the appropriate HRP-conjugated secondary antibody (diluted 1:2500 in 5% milk; Promega).

The resulting protein bands were visualized by enhanced chemiluminescence (Amersham

Pharmacia Corp., Piscataway, NJ) and subsequently quantified by densitometric analysis.

RNA Isolation, RT-PCR, and Real-time Quantitative PCR

Total RNA was isolated from cell lines using the mirVana miRNA isolation kit

(Ambion) and subjected to DNase treatment using the DNA-free kit (Ambion) according to the manufacturer’s protocol. 10ng of DNase-treated RNA was then reverse transcribed using SuperScript II (Invitrogen). Relative gene expression was assayed by real-time

quantitative PCR using 12.5µl SYBR Green PCR Master Mix (Applied Biosystems,

Foster City, CA), 10µM forward primer, 10µM reverse primer, and 20ng of template

cDNA. Gene expression determinations were made for PTEN (F: 5’-

TCCACAAACAGAACAAGATG-3’ and R: 5’-CTGGTCCTGGTATGAAGAAT- 3’)

and GAPDH, a normalization control, (F: 5’-GGGCTGCTTTTAACTCTGGTAA-3’ and

R: 5’-ATGGGTGGAATCATATTGGAAC -3’). Thermal cycling conditions comprised

of 50°C for 2 min, 95°C for 10 min, and 40 cycles at 95°C for 15 s followed by 60°C

for 1 min using an ABI 7500 Sequence Detection System (Applied Biosystems). All

reactions were performed in triplicate. Subsequent relative gene expression

determinations were made using the comparative delta CT method (2-ddCT) as described

by Livak et al. (67).

Statistical Analysis

All results are expressed as the mean ± standard deviation from a minimum of three separate experiments. Statistical analysis was performed using Student’s t test and results are considered significant at the P-value < 0.05 level.

4.3 RESULTS

Identification and selection of potential PTEN-targeting miRNAs

In order to select miRNAs predicted to putatively target the 3’UTR of PTEN, we

relied on the computational miRNA:target prediction algorithms as implemented in the

TargetScan 3.1 (http://www.targetscan.org/), miRANDA (http://microrna.sanger.ac.uk/), and PicTar (http://pictar.bio.nyu.edu/) software programs and limited our search to those predicted within the 902-bp 3’UTR of the published PTEN mRNA sequence

(NM_000314.2). Interestingly, among the three websites, more than 25 miRNAs were predicted to target this region. Furthermore, the highest-ranking miRNAs varied considerably. In order to define a more finite list of putative miRNAs to assess functionally, we restricted our efforts to those located within regions of extensive sequence conservation and, more specifically, to those sharing strong conservation among the nucleotides located at the putative miRNA:target binding site (positions 2 to 8 of the mature miRNA) as well as those predicted by at least 2 of the 3 computational algorithms. Using these criteria, we selected four miRNAs for further investigation: miR-

26a, miR-519e, miR-20a, and miR-22 which target positions +17 to +48, +247 to +278,

+249 to +279, and +673 to +696 of PTEN’s 3’UTR, respectively (Fig. 4.1A, 4.1B, and

4.1C).

Figure 4.1. Four miRNAs computationally predicted to target PTEN’s 3’UTR.

A) Schematic diagram of 4 putative miRNAs predicted to target PTEN’s 3’UTR. The coding portion of exon 9 is depicted (gray) along with its stop codon (TGA) and positions

+1 to +902 of PTEN’s 3’UTR. miR-26a, miR-519e, miR-20a, and miR-22 are shown

relative to their predicted target regions on PTEN’s mRNA sequence (NM_000314.2).

B) Alignment of miR-26a, miR-519e, miR-20a, and miR-22 to their respective target

sequences in PTEN’s 3’UTR. Continued 96

Figure 4.1. Continued. C) Multi-species alignment of the highly conserved regions of

PTEN’s 3’UTR predicted to be targeted by miR-26a, miR-519e, miR-20a, and miR-22.

Alignment includes sequence from human, mouse, rat, opossum, and chicken. Asterisks

(*) indicate nucleotides conserved among all 5 species.

miR-519e interacts with PTEN’s 3’UTR

To investigate whether the selected miRNAs interact with PTEN’s 3’UTR

sequence, we examined the effects of their over-expression using a reporter construct

containing PTEN’s wild-type (WT) 3’UTR sequence from position +10 to +930

subcloned downstream of the firefly luciferase gene and upstream of the SV40 late

poly(A) signal (pGL3-PTEN3’UTR/5-3) (Fig. 4.2A). Transient transfection of precursor 97

molecules for miR-26a, miR-20a, and miR-22 into MCF-7 cells did not significantly alter luciferase activity relative to our mock transfection or transfection with the control precursor miRNA (P > 0.09) [Fig. 4.2B]. Conversely, a significant decrease in reporter gene activity (~35% relative to the mock transfection) was observed following transfection of the miR-519e precursor (P = 0.001). This effect was not observed following co-transfection of the miR-519e precursor with a reporter construct containing

PTEN’s 3’UTR in the antisense orientation (P > 0.15) [Fig. 4.2A and 4.2B], suggesting that miR-519e’s repression is specific to both PTEN’s WT 3’UTR sequence and orientation.

Previous studies have demonstrated that the majority of miRNA:target interactions rely on complementary basepairing between positions 2-8 at the miRNA’s 5’ end and the putative mRNA target sites (142-144). This so called ‘seed site’ is minimally required for miRNA-target recognition. Therefore, to determine whether the observed miR-519e-mediated repression of our reporter construct was specific to the putative binding site located at position +271 to +277 of PTEN’s 3’UTR, we generated a mutant construct lacking miR-519e’s 7-bp seed site (GGCACTT) [Fig. 4.3A]. As observed previously, co-transfection of the miR-519e precursor along with the pGL3-

PTEN3’UTR/5-3 construct resulted in a significant decrease in reporter activity relative to pGL3-PTEN3’UTR/5-3 alone (P = 0.004) [Fig. 4.3B]. However, transfection of the pGL3-PTEN3’UTR/5-3∆519e construct both alone and when co-transfected along with the miR-519e precursor did not repress luciferase activity (Fig. 4.3B).

Taken together, these data strongly suggest that miR-519e is able to bind to a specific target sequence located in PTEN’s 3’UTR and that the interaction at this binding

site mediates a significant inhibition of reporter gene expression.

miR-519e down-regulates PTEN protein

To examine the effects of predicted PTEN-targeting miRNAs on PTEN protein expression, we subsequently transfected each miRNA precursor molecule into MCF-7 cells, isolated total protein lysates, and assayed PTEN expression by Western blot analysis. Similar to our reporter assay experiments, no significant differences in PTEN expression were observed following transfection of miR-26a, miR-20a, and miR-22 while

over-expression of miR-519e resulted in a ~25% decrease in PTEN protein levels relative

to the mock transfection (P = 0.008) [Fig. 4.4A and 4.4B]. These observations are

concordant with our reporter assay experiments and provide further evidence supporting

miR-519e’s role in the regulation of PTEN protein expression.

Additionally, we assayed the downstream read-out of PTEN’s lipid phosphatase

and protein phosphatase activities, P-Akt and P-MAPK, respectively. Interestingly, we

did not observe increases in P-Akt and P-MAPK levels that are generally associated with

decreased PTEN levels. Moreover, total Akt and MAPK levels were also unchanged

(data not shown).

We also assessed the effects of miR-519e over-expression in 3 additional cell lines, which are among the component cancers associated with CS, including an FTC- derived cell line (WRO82-1) and 2 additional breast cancer lines (MDA-MB-231 and

100

Figure 4.2. miR-519e inhibits luciferase reporter activity. A) Schematic representation of reporter constructs used to assess the activity of miRNAs predicted to target PTEN. The 3’UTR of PTEN (position +10 to +930) was inserted into the pGL3.1-Promoter vector in both the sense (pGL3-PTEN3’UTR/5-3) and antisense orientation (pGL3-PTEN3’UTR/3-5). B) MCF-7 cells were cotransfected with either pGL3-PTEN3’UTR/5-3 or pGL3-PTEN3’UTR/3-5 and the pRL-TK internal control plasmid. Additionally, cells were cotransfected with precursor miRNAs (as indicated in figure) and subsequently assayed for luciferase activity. Firefly luciferase measurements from three independent experiments were corrected for transfection efficiency using the Renilla luciferase internal control, averaged, and then normalized to the mock transfection. Error bars represent the standard deviation of the three independent experiments. *P-value < 0.01 versus mock transfection.

101

Figure 4.3. miR-519e-mediated luciferase reporter activity is specific to the miR-519e seed site. A) Schematic diagram of the pGL3- PTEN3’UTR/5-3∆519e reporter construct. Site-directed mutagenesis was used to generate a mutant reporter construct containing a 7- bp deletion of the miR-519e seed site (GGCACTT, position +271 to +277). B) MCF-7 cells were cotransfected with either the wild- type (WT) pGL3-PTEN3’UTR/5-3 construct or the mutant pGL3-PTEN3’UTR/5-3∆519e construct and pRL-TK in either the presence or absence of the miR-519e precursor molecule. Firefly luciferase measurements were assayed from three independent experiments, corrected for transfection efficiency using the Renilla luciferase internal control, averaged, and then normalized to the WT construct. Error bars represent the standard deviation of the three independent experiments. *P-value < 0.01 versus WT pGL3- PTEN3’UTR/5-3.

MDA-MB-435). As observed in MCF-7 cells, over-expression of miR-519e resulted in significantly reduced PTEN expression in both WRO82-1 (P = 0.005) and MDA-MB-

435 cells (P = 0.009) [Fig. 4.4C]. Modest reductions in PTEN expression were also noted in MDA-MB-231 cells.

Anti-miR-519e treatment restores miR-519e-induced PTEN down-regulation

In order to determine if the observed decrease in PTEN expression was specific to

miR-519e over-expression, we transfected MCF-7 cells with an antisense miRNA inhibitor complementary to miR-519e and assayed PTEN expression by Western blot analysis. As in our previous experiments, we observed a significant decrease in PTEN expression following transfection with the miR-519e precursor molecule (28.4 ± 5.6% relative to the mock transfection, P = 0.018) [Fig. 4.5A and 4.5B]. P-Akt and P-MAPK levels were similar to those shown in Fig. 4.4 (data not shown). Inhibition of miR-519e using the anti-miR-519e miRNA inhibitor restored PTEN levels to endogenous level

(96.6 ± 6.4% relative to the mock transfection, P = 0.636).

Over-expression of miR-519e reduces expression of PTEN mRNA

In order to better understand the mode by which miR-519e represses PTEN, we

assessed PTEN mRNA levels using total RNA isolated from transiently transfected MCF-

7 cells. Using real-time PCR, we observed a significant decrease in PTEN mRNA levels

among cells in which miR-519e was over-expressed (24.7 ± 3.8% relative to the mock

transfection, P < 0.001) [Fig. 4.5C]. This effect was countered by co-transfection of the

anti-miR-519e miRNA inhibitor molecule. Moreover, no reduction was observed 102

103

Figure 4.4. Endogenous PTEN expression is decreased by miR-519e. A) Following transfection of miRNA precursor molecules into

MCF-7 cells, PTEN expression was assayed by Western blot. As well, P-Akt and P-MAPK (p42/44), the downstream targets of

PTEN’s lipid phosphatase and protein phosphatase activity, receptively, and β-actin were also assayed. B) Quantification of PTEN

protein expression normalized to β-actin levels. Western blot results are presented relative to the mock transfection control. Error

bars represent the standard deviation of the three independent experiments. *P-value < 0.01 versus mock transfection. Continued

Figure 4.4. Continued. C) WRO82-1, MDA-MB-231, and MDA-MB-435 cell lines were transfected with either the negative control miR or the miR-519e precursor molecule and assayed for PTEN and β-actin expression by Western blot analysis.

following transfection with the mock or control miRNA precursor, either in the presence or absence of the anti-miR-519e miRNA inhibitor. These data suggest that miR-519e’s negative regulation of PTEN is likely the result of miR-519e-mediated PTEN mRNA degradation.

miR-519d over-expression is also able to down-regulate PTEN

miR-519e is one member of a large non-conserved miRNA cluster, which has been grouped into four highly related miRNA families, recently identified on

104

105

Figure 4.5. Anti-miR-519e treatment restores miR-519e-induced PTEN down-regulation. A) MCF-7 cells were transfected with the miR-519e precursor molecule and either with or without an antisense miRNA inhibitor complementary to miR-519e (as indicated in figure) and assayed for PTEN expression by Western blot analysis. A decrease in PTEN expression was observed following transfection with the miR-519e precursor molecule. Inhibition of miR-519e expression using the anti-miR-519e miRNA inhibitor restored PTEN expression to endogenous level. B) Quantification of PTEN protein expression normalized to β-actin levels. Western blot results are presented relative to the mock transfection control. Error bars represent the standard deviation of the three independent experiments. *P-value < 0.01 versus mock transfection. **P-value > 0.05 versus mock transfection. Continued

Figure 4.5. Continued. C) Real-time PCR was used to assay PTEN mRNA expression following the transfection of MCF-7 cells with the miR-519e precursor molecule and either with or without the anti-miR-519e inhibitor as indicated. miR-519e induced a reduction in PTEN mRNA levels relative to the mock or Control miRNA transfected

samples, while the anti-miR-519e miRNA inhibitor restored PTEN mRNA levels to those

similar to mock and Control miRNA transfected samples. Real-time PCR results are

depicted as fold change (2-ddCT) relative to the mock transfection control. Error bars

represent the standard deviation of the three independent experiments. *P-value < 0.01

versus mock transfection. **P-value > 0.05 versus mock transfection.

106

chromosome 19 (145). Among the family containing miR-519e, a second miRNA, miR-

519d, shares the same seed sequence and, as well, is predicted to bind to the same region of PTEN’s mRNA (Fig. 4.6A). Because of this, we chose to investigate whether over- expression of miR-519d could similarly down-regulate PTEN. Following transfection of the miR-519d precursor, we observed a ~23% decrease in PTEN expression relative to the mock transfection experiment, an effect comparable to that of miR-519e, while no reduction was observed following transfection with either the mock or control miRNA precursor (Fig. 4.6B). These data suggest that both miRNAs target PTEN and, therefore, likely contribute to the regulation of its expression.

4.4 DISCUSSION

Mechanisms of regulation and dysregulation beyond structural genetic alterations of

PTEN are increasingly pertinent given PTEN’s role in heritable and sporadic carcinogenesis. As such, we have identified the role of PTEN’s nuclear-cytoplasmic partitioning, a process likely altered in carcinogenesis, identified the deletion of a novel cis-acting regulatory element upstream of PTEN, and, recently, uncovered variants associated with aberrant PTEN translation (56, 146, 147).

Because of PTEN’s significant roles in both regulating the cell cycle and in malignant transformation, and in an attempt to better understand the mechanisms that govern PTEN expression, we chose to investigate the functional involvement of miRNAs in regulating this vital tumor suppressor.

107

Figure 4.6. Both miR-519e and miR-519d decrease endogenous PTEN expression.

A) Sequence alignment of the mature miRNA sequences for miR-519e and miR-519d.

The shared seed sequences (nucleotides ~2-8) among the two related miRNAs are shown

in gray. B) MCF-7 cells were transfected with either the miR-519e or the miR-519d

precursor molecule and assayed for PTEN expression by Western blot analysis. Similar

decreases in PTEN expression were observed following transfection with both miR-519

precursor molecules relative to the mock and Control miRNA transfected cells.

108

Our data provide several lines of evidence supporting miR-519e’s role in modulating PTEN’s expression. Interestingly, upon over-expression of miR-519e, we did not observe increases in P-Akt and P-MAPK expression as generally associated with decreased PTEN levels. Additionally, total Akt and MAPK levels were also unchanged.

This observation is not completely unexpected, as most miRNAs are believed to affect many target mRNAs. In accordance with this, both miRANDA and TargetScan list several additional targets of miR-519e, more than 1,300 in total. Included among these putative targets are several member of the Ras-Raf signaling cascade, a pathway known to activate both the MEK/MAPK and PI3K/Akt pathways (148-150). Thus, it is plausible that other pathways that modulate Akt and MAPK phosphorylation are affected and may reflect an important feedback regulatory loop. This is an interesting area of investigation but is beyond the scope of this paper.

The importance of miRNAs in several biological processes, including cell differentiation, control of the cell cycle, and apoptosis, is well appreciated (134).

However, despite strong evidence associating their deregulation with carcinogenesis, the precise contributions of miRNAs to disease are not well known, nor are they understood in other diseases, such as cardiovascular disease and type 2 diabetes. Initial miRNA studies have primarily taken global approaches to implicate these small non-coding

RNAs in disease pathogenesis, identifying mis-regulated miRNA by assessing their expression using microarrays. Efforts to decipher the specific relationships between miRNAs and their targets have lagged.

PTEN is among the few genes for which miRNA:target interactions have been reported. At the onset of this project, miR-19a was the only miRNA experimentally

109

shown to target PTEN (143). The miR-19a binding site, based on predictions using

TargetScan and miRANDA, is located at position +1208 to +1228. Interestingly, the

binding site for miR-19a lies outside of the published PTEN mRNA sequence used in the

selection of putative miRNAs in the present study. Since the inception of this project, the

PTEN mRNA sequence as maintained by NCBI (www.ncbi.nlm.nih.gov/) has undergone two version updates. In support of this report by Lewis et al., the latest version, NM_000314.4 (released June 2007), contains a lengthened 3’UTR which now spans a total of 3,329-bp. This updated version contains a second poly-adenylation

signal, located at position +3,283 to +3,288, in addition to the one previously annotated

(position +880 to +885), suggesting that PTEN’s mRNA can possess at least 2 alternative

3’UTR sequences. In addition to providing additional binding sites for regulatory

proteins which may modulate PTEN, this lengthened alternative 3’UTR likely provides

additional recognition sites for several potential miRNAs, thereby increasing the number

of miRNAs potentially involved in its regulation.

More recently, miR-21 has been shown to specifically repress PTEN translation

(151). In contrast to miR-19a and miR-519e, miR-21 is not predicted by any of the computational miRNA:target prediction algorithms used in this study to target PTEN. A scan of its seed sequence suggests that miR-21 binds PTEN’s 3’UTR at approximately position +831 to +838.

The deregulation of miRNAs that target tumor suppressors, such as PTEN, and oncogenes provide one mechanism by which miRNAs likely contribute to carcinogenesis. As PTEN is the second most mutated gene in all human cancers, exceeded only by mutations in TP53, it is plausible that miRNAs involved in its

110

regulation may be among those most dysregulated in cancer (152). Of the miRNAs known to target PTEN, both miR-21 and miR-19a have been linked with cancer (153-

155). It is likely that the pathogenicity associated with mis-expression of these miRNAs is due in large part to the effect they impart to their target genes. Interestingly, aberrant miR-21 expression has been observed in breast carcinoma, a component cancer of CS, while in B cell chronic lymphocytic leukemia, the over-expression of miR-19a has been correlated with decreased PTEN expression. To date, the role of miR-519e in cancer has not been established. miR-519e is one of 89 cloned miRNAs recently identified by

Bentwich et al. (145). Its recent discovery may, in part, contribute to the lack of reports describing any association with cancer, as this miRNA was not examined in many of the early studies reporting these associations.

miR-519e was identified as part of a large non-conserved miRNA cluster on chromosome 19, containing four highly related miRNA families. We examined a second miRNA from the same family as miR-519e, miR-519d, and which shares its same seed sequence. Similarly to its family member, over-expression of miR-519d also induced the down-regulation PTEN, suggesting that both have a role in regulating PTEN. In their study, Bentwich et al. examined the expression of their newly identified miRNAs in five tissues (placenta, testis, thymus, brain, and prostate) and only noted significant expression of miR-519e and miR-519d in placenta. This feature was conserved among all

54 miRNAs in this cluster, suggesting that miR-519e and miR-519d may have roles in fetal development. Of note, nominal expression of both miRNAs was also detected in prostate.

111

In the present study, we have provided functional evidence that PTEN is

negatively regulated by miR-519e, a miRNA that is highly expressed in placenta and a member of a large non-conserved miRNA cluster. The work presented here brings the total number of miRNAs experimentally proven to target PTEN to 3 and, equally

importantly, adds to the handful of biological targets that have been experimentally

verified to date. Although the role of miR-519e has yet to be examined in cancer, our data

suggest that modulation of its expression, as well as other miRNAs shown to target

known tumor suppressors and/or oncogenes, may prove to be therapeutically useful in the

management of patients suffering from malignancy. Despite the arduous task,

experimental validation of miRNA targets is critical towards understanding their roles in

both normal cellular processes and in human disease. We speculate that miRNAs that

repress PTEN, including miR-519e, and possibly miR-19a and miR-21, may have a role in

CS and BRRS pathogenesis, potentially acting as genetic modifiers, and may likely aid in

elucidating novel mechanisms of PTEN dysfunction in patients for whom the cause of

disease has proven enigmatic.

112

CHAPTER 5

DIFFERENTIAL EXPRESSION OF PTEN-TARGETING MICRORNAS, miR-19a AND miR-21, IN COWDEN SYNDROME

5.1 INTRODUCTION

Cowden syndrome (CS [MIM 158350]) is an under-diagnosed autosomal

dominant disorder characterized by multiple hamartomatous lesions and a vast

phenotypic spectrum that includes predisposition to malignancies of the breast, thyroid,

and endometrium (4). The majority of patients with CS (85%) have been found to harbor

pathogenic germline mutations in the gene encoding phosphatase and tensin homolog

deleted on chromosome ten (PTEN [MIM 601728]), a tumor suppressor gene located on

10q23 (50). The inactivation of PTEN protein, an antagonist of the phosphatidylinositol-

triphosphate kinase (PI3K) signaling pathway, in CS results in the constitutive activation

of Akt (7, 63, 64, 87-90). As a consequence, hyperactive Akt-mediated signaling of

several cascades results in increased and uncontrolled cellular survival and proliferation.

Germline mutations in PTEN also cause a subset of Bannayan-Riley-Ruvalcaba syndrome (BRRS [MIM 153480]), a related hamartomatous tumor syndrome that, in

113

addition to its genetic etiology, shares several of the clinical manifestations seen in CS

(7). Interestingly, identical pathogenic PTEN mutations have been observed among

patients with these seemingly disparate disorders (57). Furthermore, these same identical

mutations have also been observed in a subset of patients referred to as CS-like (CSL)

patients who do not meet the full diagnostic criteria for CS. While germline PTEN

mutations have been identified in the majority of patients diagnosed with CS and BRRS,

for approximately 15% and 35%, respectively, as well as for 90% of CSL patients, the pathogenic mutations have yet to be identified (50, 57). Interestingly, decreased PTEN

protein expression has been observed in a large number of these mutation-negative

patients (Waite and Eng, unpublished observation). The lack of detectable germline

PTEN mutations in subsets of CS, CSL, and BRRS patients with altered PTEN expression, as well as the imprecise genotype-phenotype correlation associated with these syndromes, suggests the presence of alternate mechanisms that contribute to PTEN dysfnction and to the development and progression of the disease.

MicroRNAs (miRNAs), a novel class of negative gene regulators, have recently

been shown to regulate the expression of several tumor suppressors and oncogenes and

contribute to carcinogenesis (154-158). To date, two miRNAs, miR-19a and miR-21,

have been reported to specifically target and down-regulated PTEN (143, 151).

Furthermore, the over-expression of each of these miRNAs has been correlated with

decreased PTEN levels in human cancer (153, 159).

Because PTEN dysfunction is common in patients with CS, CSL, and BRRS,

irrespective of their mutation status, we hypothesized that aberrant miR-19a and/or miR-

21 expression could modulate PTEN protein levels and, thereby, modulate phenotypic

114

expression in patients with these syndromes. To investigate this hypothesis, we chose to characterize the relative expression levels of both miR-19a and miR-21 in PTEN mutation-positive and PTEN mutation-negative CS, CSL, and BRRS patients relative to healthy control subjects.

5.2 MATERIALS AND METHODS

Study Subjects

A total of 178 unrelated subjects were included in the present study, including 20 control subjects and 158 CS/CSL/BRRS patients [Table 5.1]. Mutation analysis of the entire PTEN coding sequence and its exon/intron boundaries and promoter region was performed for all CS/CSL/BRRS patients included in the present study (50).

Additionally, all patient samples were also screened for PTEN protein expression, by

Western blot analysis as previously described (63). PTEN protein levels were compared to controls and scored, in a blinded fashion, as either normal, decreased (< 50% loss, compared to controls), half (~ 50% of controls) or faint (> 50% loss, compared to controls) by two individuals (Dr. Kristin A. Waite and Ms, Tammy M. Sadler). PTEN mutation-negative patients with decreased, half or faint PTEN protein levels were also scanned for large deletions and rearrangements and the latter excluded. Among the

CS/CSL/BRRS patients, 28/158 of the individuals selected for inclusion were previously found to harbor pathogenic germline mutations (i.e. PTEN mutation-positive patients).

Mutation analysis in the remaining 130/158 patients revealed no detectable germline

PTEN mutations (i.e. PTEN mutation-negative patients). In order to interrogate miR-19a and miR-21 expression in CS/CSL/BRRS patients without germline PTEN mutations yet

115

PTEN Mutation- PTEN Mutation- Controls positive Patients negative Patients (n = 20) (n = 28) (n = 130) Gender - n(%) Male 7 (35.0) 12 (42.9) 10 (7.7) Female 13 (65.0) 16 (57.1) 120 (92.3)

Phenotype - n(%) CS --- 17 a (60.7) 63 (48.5) CSL --- 5 (17.9) 62 (47.7) BRRS --- 5 (17.9) 5 (3.8)

Mutation - n(%) R130X --- 13 (46.4) --- R233X --- 5 (17.9) --- R335X --- 10 (35.7) ---

Clinical Features - n(%) Macrocephaly --- 25 (89.3) 53 (40.8) Pathognomonic --- 16 (57.1) 28 (21.5) Breast CA --- 7 (43.8) b 80 (66.7) b Thyroid CA --- 2 (7.1) 32 (24.6) Endometrial CA --- 2 (12.5) b 14 (11.7) b

NOTE – Pathognomonic features include the mucocutaneous lesions associated with CS (trichilemmomas, acral keratoses, and papillomatous papules). a One PTEN mutation-positive sample had features consistent with both CS and BRRS. b Frequency among female patients.

Table 5.1. Summary of control and patient clinical features.

116

with deregulated PTEN protein levels, all 130 mutation-negative patients selected for inclusion in the current study were previously found to have decreased PTEN protein expression relative to normal controls (data not shown). All subjects were enrolled by referral from centers throughout the United States, Canada, and Europe following informed consent in accordance with procedures approved by the human subjects protection committees of each respective institution.

Among the cohort of PTEN mutation-positive patients included in our analysis were 13/28 patients with R130X (c.388 C/T) mutations, 5/28 patients with R233X (c.697

C/T) mutations, and 10/28 patients with R335X (c.1003 C/T) mutations (Fig. 5.1 and

Table 5.1). Of those with R130X mutations, 9/13 were classic CS patients, 1/13 patients exhibited a CSL phenotype, and 2/13 were classic BRRS patients. Additionally, 1/13 patients with an R130X mutation had phenotypic features consistent with both CS and

BRRS, and thus was considered as a CS/BRRS overlap patient. Among the patients with an R233X mutation, 1/5 patients met classic CS criteria, 3/5 patients were considered

CSL, and 1/5 patients was classic BRRS. And lastly, 7/10 patients with R335X mutations met classic CS criteria, 1/10 was a CSL patient, and 2/10 were BRRS patients.

Among the 130 PTEN mutation-negative patients with decreased PTEN expression, 63 were classic CS, 62 were CSL, and 5 were diagnosed with BRRS. All PTEN mutation- positive and PTEN mutation-negative CS patients included in the present study were classified in accordance with criteria established by the International Cowden Consortium and curated by the National Comprehensive Cancer Network (http://www.nccn.org) (4).

117

Figure 5.1. Schematic diagram of PTEN’s mRNA sequence. PTEN’s full length mRNA

sequence (NM_000314.4), including its 1,032 nucleotide (nt) 5’ UTR, 1,212 nt coding

sequence, and 3,302 nt 3’ UTR, is shown along with the miR-19a binding site (position

*1208 to *1228). miR-21’s binding site has not been reported. Additionally, positions

for the 3 truncating mutations (R130X, R233X, and R335X) observed in the PTEN

mutation-positive cohort are shown. PTEN’s translation start site and the region

encoding PTEN’s phosphatase core are also annotated.

Cell Lines and Culture

Lymphoblastoid cell lines (LBCLs) from all healthy control samples and

CS/CSL/BRRS patient samples were generated by the Genomic Medicine Biorepository

(http://www.lerner.ccf.org/gmi/gmb/methods.php). LBCLs were cultured in RPMI-1640 media supplemented with 20% FBS and 100 units/ml each of Penicillin and Streptomycin and maintained at 37˚C with 5% CO2.

118

Isolation of Total RNA and cDNA Synthesis

Total RNA was isolated from control and patient LBCLs using the miRVana

miRNA Isolation Kit (Ambion, Austin, TX) according to the manufacturer’s protocol

and, subsequently, subjected to DNase treatment using DNA-free (Ambion). For analysis of PTEN expression levels, 1µg of DNase-treated RNA was converted to cDNA using

Superscript II Reverse Transcriptase (Invitrogen, Carlsbad, CA), as specified by the

manufacturer.

In order to assay mature miR-19a and miR-21 expression in our control and

patient samples, stem-loop reverse transcription (RT) was performed using TaqMan

MicroRNA Assays (Applied Biosystems, Foster City, CA). For each miRNA, individual

RT reactions were carried out using 10ng of DNase-treated RNA and either miR-19a,

miR-21, or U6 small nuclear RNA (RNU6) specific TaqMan MicroRNA RT primers

(Applied Biosystems). 20µl RT reactions were performed using the TaqMan MicroRNA

Reverse Transcription Kit (Applied Biosystems) according to the manufacturer’s

protocol.

Real-time Quantitative PCR

Relative expression of the full-length PTEN transcript was assayed for all samples

using 12.5µl SYBR Green PCR Master Mix (Applied Biosystems), 10µM forward primer

(5’- TCCACAAACAGAACAAGATG -3’), 10µM reverse primer (5’-

CTGGTCCTGGTATGAAGAAT -3’), and 20ng of template cDNA. Thermal cycling

conditions comprised of 50°C for 2 min, 95°C for 10 min, and 40 cycles at 95°C for 15

s followed by 60°C for 1 min using an ABI 7500 Sequence Detection System (Applied 119

Biosystems). Real-time PCR reactions were also performed for GAPDH, a normalization

control (F: 5’-GGGCTGCTTTTAACTCTGGTAA-3’ and R: 5’-

ATGGGTGGAATCATATTGGAAC -3’).

Mature miR-19a and miR-21 expression was assayed using a 20µl reaction

containing 10.0µl TaqMan 2X Universal PCR Master Mix, No AmpErase UNG (Applied

Biosystems), 1.0µl 20X TaqMan MicroRNA assay (Applied Biosystems), and 1.3µl of

RT product. Real-time PCR conditions were performed as follows: 50°C for 2 min,

95°C for 10 min, and 40 cycles at 95°C for 15 s followed by 60°C for 1 min using an

ABI 7500 Sequence Detection System (Applied Biosystems). Additionally, RNU6

expression was assayed for normalization.

All reactions were performed in triplicate and relative gene expression

-∆∆CT determinations were made using the comparative delta-delta CT method (2 ) as described by Livak et al. (67).

Statistical Analysis

Following expression analysis, statistical comparisons of relative PTEN, miR-19a,

and miR-21 expression levels between all patient groups and control samples were

performed using Welch’s t-test. All results were considered significant at the P < 0.05

level. In addition, we preformed classification and regression tree (CART) analysis as

implemented in the R software package (http://cran.r-project.org).

120

5.3 RESULTS

Differential Expression of miR-19a and miR-21 in 3 Groups of CS/CSL/BRRS Patients with Germline PTEN Mutations R130X, R233X and R335X

In the present study, we sought to investigate whether miRNAs previously shown

to target and regulate PTEN protein expression are differentially regulated in patients

with CS, CSL, and BRRS relative to healthy controls and, more specifically, whether

these miRNAs are modifiers of these phenotypes. To address this question, we initially

chose to assess the relative expression of miR-19a and miR-21 in a series of 28 PTEN

mutation-positive patients sharing identical truncating mutations located in each of

PTEN’s 3 most frequently mutated exons (Fig. 5.1) and 20 normal controls (7). Included

among these patients were 13 carriers of R130X mutations, 5 carriers of R233X

mutations, and 10 carriers of R335X mutations. Previous studies have reported a broad

clinical spectrum in patients even within each of these 3 mutation groups (52, 57).

Similar observations were apparent in this collection as 9 CS, 1 CSL, 1 CS/BRRS, and 2

BRRS patients were among those with R130X mutations, 1 CS, 3 CSL, and 1 BRRS

patient were among those with R233X mutations, and 7 CS, 1 CSL, and 2 BRRS patients

were among those with R335X mutations (Table 5.2).

In addition to the variable clinical spectrum observed among patients with

identical mutations, PTEN mRNA expression levels were also found to be variable.

Among all 28 PTEN mutation-positive patients, expression of PTEN transcript levels

were significantly reduced relative to controls (Table 5.3, P = 0.015). However, when we

divided this group based on the 3 different germline PTEN mutations found in these

patients, significantly decreased PTEN transcript expression was only observed in

121

Mutation/ Age at PTEN PTEN miR-19a miR-21 Sample Gender Diagnosis Phenotype Expression Expression Expression Expression Clinical Manifestations R130X 0047-01 M 46 CS Faint 0.893 1.567 2.128 Macrocephaly, trichilemmomas, acral keratoses, papillomatous papules, lipomas, GI cancer 0491-01 F 70 CS Normal 0.870 1.510 1.181 Macrocephaly, fibrocystic breast disease, breast cancer, endometrial fibroids, endometrial cancer 0521-01 M 15 CS Decrease 0.805 1.880 2.937 Macrocephaly, LDD,

122 benign thyroid, multiple papules on lips, tongue, and esophagus 0857-01 F 63 CS Half 0.604 1.876 1.381 Macrocephaly, benign thyroid, colon polyps, GU abnormality 1111-01 M 56 CS Half 1.020 3.900 2.532 Macrocephaly, LDD, thyroid cancer

Continued

Table 5.2. Clinical features, PTEN protein, and relative PTEN, miR-19a, and miR-21 expression among PTEN mutation-positive

patients.

Table 5.2. Continued.

1432-01 M 59 CS Normal 0.637 0.829 0.918 Macrocephaly, lipomas, trichilemmomas, papillomatous papules, benign thyroid 4396-01 F 47 CS Normal 0.817 1.231 0.646 Macrocephaly, lipomas, benign thyroid, breast cancer, endometrial fibroids 11578-01 F 61 CS Half 0.634 1.861 1.447 Macrocephaly, papillomatous papules, breast cancer, hemangioma, lipomas,

123 benign thyroid, endometrial fibroids 12644-01 F 47 CS Decrease 0.828 1.456 1.502 Macrocephaly, trichilemmomas, papillomatous papules, breast cancer, benign thyroid 2381-01 F 2 CSL Normal 0.984 1.158 1.271 Macrocephaly, lipomas, developmental delay

Continued

Table 5.2. Continued.

15413-01 M 50 CS/BRRS Decrease 0.913 1.510 2.249 Macrocephaly, papillomatous papules, lipomas, hemangioma, benign thyroid, LDD, pigmented macules on penis 0268-01 M 9 BRRS Normal 1.219 0.959 1.315 Macrocephaly, lipomas, pigmented macules on penis 4498-01 M 9 BRRS Normal 0.932 1.016 1.006 Macrocephaly, oral papillomatous papules, pigmented macules on

124 penis R233X 12961-01 F 48 CS Half 0.528 1.419 1.421 Macrocephaly, trichilemmomas, papillomatous papules, acral keratoses, lipomas, hemangioma, thyroid cancer 4026-01 F 3 CSL Normal 0.514 1.696 1.625 Macrocephaly, developmental delay 4389-01 F 40 CSL Half 0.527 2.078 2.180 Macrocephaly, breast cancer, lipomas

Continued

Table 5.2. Continued.

17727-01 F 6 CSL Decrease 0.619 1.310 1.133 Macrocephaly, lipomas, hemangioma, developmental delay 972-01 M 7 BRRS Decrease 0.602 1.443 1.028 Macrocephaly, developmental delay, hemangioma R335X 1113-01 F 32 CS Faint 1.275 1.073 1.235 Macrocephaly, trichilemmomas, acral keratoses, lipomas, fibromas, breast cancer 1194-01 F 60 CS Faint 1.146 2.210 1.537 Macrocephaly,

125 papillomatous papules, lipomas, benign thyroid, developmental delay 1495-01 F 50 CS Decrease 1.202 2.225 1.964 Fibrocystic breast disease, papillomatous papules, lipomas, benign thyroid, endometrial fibroids 4292-01 F 56 CS Half 1.381 0.463 0.950 Fibrocystic breast disease, acral keratoses, lipomas, trichilemmomas, papillomatous papules, endometrial fibroids

Continued

Table 5.2. Continued.

15394-01 F 50 CS Decrease 1.284 1.549 1.929 Fibrocystic breast disease, trichilemmomas, papillomatous papules, lipomas, trichilemmomas, breast cancer, benign thyroid, endometrial cancer, endometrial fibroids 16224-01 M 38 CS Decrease 1.661 1.628 1.609 Macrocephaly, trichilemmomas, papillomatous papules,

126 lipomas 19885-01 F 10 CS Decrease 1.067 1.562 1.222 Macrocephaly, acral keratoses, lipomas, hemangioma 6230-01 M 39 CSL Half 1.260 2.118 1.574 Macrocephaly, lipomas

0232-01 M 44 BRRS Normal 1.008 0.678 0.597 Macrocephaly, trichilemmomas, papillomatous papules, lipomas, benign thyroid, pigmented macules on penis 0531-01 M 1 BRRS Faint 1.333 0.762 1.107 Macrocephaly, developmental delay

patients with the R130X and R233X mutations (Table 5.3, P <0.001) and not in carriers

of the R335X mutation (P = 0.287). Similarly, we observed differential expression of miR-19a and miR-21 among our mutation-positive samples. Overall, the average relative

expression of both miR-19a and miR-21 was higher in PTEN mutation-positive patients

compared to controls (Fig. 5.2A and Table 5.3, 2-∆∆CT = 1.54 versus 1.08, P = 0.003 and

2-∆∆CT = 1.49versus 1.15, P = 0.006, respectively). For PTEN mutation-positive patients

with R130X and R233X mutations, miR-19a was found to be over-expressed compared

to controls (P = 0.037 and P = 0.015, respectively), while miR-21 was over-expressed

only in carriers of the R130X mutation (P = 0.044). miR-19a and miR-21 were not differentially expressed in carriers of the R335X mutation relative to our control population. Interestingly, 90% of patients with this mutation had decreased PTEN protein levels, compared to only 54% of those with R130X mutations. Moreover, while miR-19a and miR-21 were not differentially expressed in carriers of the R335X mutation,

the majority of these patients, 70%, exhibit classic CS features, compared to only 56% of

R130X and R233X patients.

miR-19a and miR-21 Over-Expression are Associated with Decreased PTEN Protein

Levels and Clinical Phenotype in PTEN Mutation-Positive CS/CSL/BRRS Patients

Our data show that PTEN transcript and PTEN protein levels, as determined by

Western blot analysis, do not correspond well among many of the mutation-positive

patients included in this analysis, suggesting that other biological processes are likely

involved in the regulation of these products (Table 5.2). Based on this observation, we

next chose to examine the relationship among the relative expression of each miRNA and

127

PTEN miR-19a miR-21 PTEN + 0.015 0.003 0.006 R130X <0.001 0.037 0.044 R233X <0.001 0.015 0.184 R335X 0.287 0.137 0.147 PTEN + dec. vs. PTEN 0.336 0.009 0.003 + norm. R130X/R233X dec. vs. 0.234 0.020 0.008 R130X/R233X norm. PTEN + CS 0.069 0.006 0.013 PTEN + CSL 0.055 0.034 0.085 PTEN + BRRS 0.209 0.932 0.772

PTEN - 0.969 <0.001 0.141 PTEN - CS 0.900 <0.001 0.835 PTEN - CSL 0.544 <0.001 0.018 PTEN - BRRS 0.148 0.492 0.215

PTEN + vs. 0.001 0.977 <0.001 PTEN - PTEN + CS vs. PTEN - 0.020 0.878 0.015 CS PTEN + CSL vs. PTEN 0.079 0.368 0.029 - CSL PTEN + BRRS vs. 0.067 0.496 0.326 PTEN - BRRS NOTE – P-values were determined from t tests comparing the relative expression of PTEN, miR-19a, and miR-21 between controls and patients and patient subgroups, except where indicated. Significant results are indicated in bold. PTEN + = PTEN mutation-positive patients. PTEN - = PTEN mutation-negative patients. PTEN + dec. = PTEN + patients with decreased PTEN protein expression. PTEN + norm. = PTEN + patients with normal PTEN protein expression. R130X/R233X dec. = R130X and R233X patients with decreased PTEN protein expression. R130X/R233X norm. = R130X and R233X patients with normal PTEN protein expression.

Table 5.3. Summary of comparisons of relative PTEN transcript, miR-19a, and miR-21

expression levels. 128

2 9

Figure 5.2. Relative expression values for PTEN transcript, miR-19a, and miR-21 among PTEN mutation-positive samples. Relative

expression values for PTEN transcript and each miRNA are shown for A) controls and all PTEN mutation-positive patients, B)

PTEN mutation-positive patients who express normal PTEN protein and those who express decreased PTEN protein and C) controls

and PTEN mutation-positive patients subdivided based on their clinical phenotypes. All relative expression values are expressed as

fold change (2-∆∆CT). PTEN + = PTEN mutation-positive patients. Error bars represent the standard deviation within each subgroup.

PTEN protein levels in these patients. More specifically, we examined miR-19a and miR-21 expression levels in mutation-positive patients with normal PTEN protein

expression compared to those with decreased PTEN protein levels. Interestingly, both miRNAs were significantly over-expressed in PTEN mutation-positive patients with decreased PTEN protein levels compared to those with normal PTEN protein levels (miR-

19a: 2-∆∆CT = 1.69 versus 1.13, P = 0.009 and miR-21: 2-∆∆CT = 1.65 versus 1.07, P =

0.003) (Fig. 5.2B and Table 5.3). Moreover, R130X and R233X patients with decreased

PTEN protein expressed approximately 54% higher levels of miR-19a (P = 0.020) and

60% more miR-21 (P = 0.008) relative to R130X/R223X patients with normal PTEN

protein levels (Table 5.3).

Next, we subdivided our PTEN mutation-positive group based on their clinical

diagnosis (CS, CSL or BRRS) and compared their relative miR-19a, miR-21, and PTEN

expression levels to those of the control group. PTEN mutation-positive CS patients

displayed relative over-expression of both miR-19a and miR-21 compared to controls

(Fig. 5.2C and Table 5.3, P = 0.006 and P = 0.013, respectively). A modest association

was observed between PTEN mutation-positive CSL patients and miR-19a (P = 0.034),

while miR-21 over-expression in this same group did not reach statistical significance

when compared to controls (P = 0.085), due in part to the small sample size in this latter

subgroup. No differences in PTEN transcript expression were observed in either

phenotypic subgroup. Lastly, PTEN transcript, miR-19a, and miR-21 expression did not

differ between our PTEN mutation-positive BRRS patients and the control samples (P >

0.20).

130

Differential Expression of miR-19a and miR-21 in CS/CSL/BRRS PTEN Mutation- negative Patients

Differential expression of miR-19a and miR-21 in PTEN mutation-positive

CS/CSL/BRRS patients with variable PTEN protein expression suggests that these

miRNAs play a role in modulation of both PTEN protein expression and the disease

phenotype in patients with these syndromes. Based on these results, we further

hypothesized that these miRNA may play a similar role in CS/CSL/BRRS patients who

lack detectable PTEN mutations. To investigate this, we, therefore, chose to assess the

relative expression of miR-19a and miR-21 in 130 selected CS/CSL/BRRS individuals

with both undetectable PTEN mutations and decreased PTEN protein expression.

In contrast to the PTEN mutation-positive patients, decreased expression of PTEN

transcript levels was not observed in PTEN mutation-negative patients (P = 0.969).

Interestingly, similar to our mutation-positive patients, PTEN mutation-negative patients

displayed relative over-expression of miR-19a when compared to our control population

(Fig. 5.3 and Table 5.3, P < 0.001). However, contrary to what was observed in PTEN

mutation-positive patients, no difference in miR-21 expression was observed in this group

(P = 0.141).

Following this analysis, PTEN mutation-negative patients were subsequently

subdivided based on their clinical diagnoses (CS, CSL or BRRS). We then carried out

genotype-phenotype association analyses based on these groupings. Similar to the PTEN

mutation-positive cohort, miR-19a over-expression was observed in both mutation-

negative CS and CSL patients (Table 5.3, P < 0.001 in both groups). In contrast, miR-21

131

Figure 5.3. Relative expression values for PTEN transcript, miR-19a, and miR-21 among

control, PTEN mutation-positive, and PTEN mutation-negative patient samples. All

relative expression values are expressed as fold change (2-∆∆CT). PTEN + = PTEN

mutation-positive patients. PTEN - = PTEN mutation-negative patients. Error bars

represent the standard deviation within each subgroup.

132

was under-expressed in PTEN mutation-negative CSL patients compared to controls (P =

0.018), while no difference was observed in CS patients from this same group (P =

0.835). We did not detect any difference in miR-19a and miR-21 expression among our

PTEN mutation-negative BRRS patients (P > 0.215).

miR-21 is Differentially Expressed Between PTEN Mutation-Positive and PTEN

Mutation-Negative Patients

Because miR-19a is over-expressed in CS/CSL patients, irrespective of PTEN

mutation status, while over-expressed miR-21 occurs only in PTEN mutation-positive

patients, we next chose to compare the relative expression of each miRNA between

patients with and without PTEN mutations. miR-21 was significantly over-expressed in

PTEN mutation-positive patients relative to patients without PTEN mutations (Fig. 5.3

and Table 5.3, 2-∆∆CT = 1.48 versus 1.05, P = <0.001), while miR-19a expression did not

differ between the two groups (P = 0.977). PTEN transcript levels were significantly

lower in the PTEN mutation-positive patient group (P = 0.001). When subdivided based

on their clinical diagnosis, miR-21 over-expression was observed in both PTEN mutation-

positive CS and CSL patients relative to mutation-negative patients with these same

phenotypes (P = 0.015 and P = 0.029, respectively), while miR-19a expression did not

differ among these subgroups (P > 0.496).

miR-21 Expression May Contribute to Phenotypic Features Associated with CS

To investigate whether differential expression of miR-19a and miR-21 was

associated with the cancers commonly seen in CS, as well as other clinical features of this

133

syndrome, we compared their relative expression between patient groups with and

without each of the key phenotypic features of CS. These comparisons failed to yield any

significant associations with breast cancer, thyroid cancer, or macrocephaly between

miRNA or PTEN expression among all patient samples and patient subgroups (P > 0.05),

likely due to the small sample sizes in these subgroup analyses. We did observe a trend

of over-expression of miR-21 in patients diagnosed with endometrial cancer compared to

those not diagnosed with this cancer, irrespective of mutation status (2-∆∆CT = 1.39 versus

1.06 for mutation-positive patients and 2-∆∆CT = 1.37 vs 1.01 for mutation-negative; P =

0.12-0.15). Despite our small sample size, we did find that miR-21 was significantly

over-expressed among CS patients with one or more pathognomonic feature (adult

Lhermitte-Duclos disease, trichilemmomas, acral keratoses, and papillomatous papules)

relative to those without any of these features (P = 0.02).

To examine the predictive value each miRNA may contribute to the CS

phenotypes observed in PTEN mutation-negative patients, we performed CART analysis

in an attempt to identify subgroups of patients at higher risk of developing 4 major

clinical features associated with this syndrome (breast cancer, endometrial cancer, thyroid

cancer, and macrocephaly) using relative expression values for both miR-19a and miR-

21. Additional CART analysis was performed for the CS and CSL phenotypes.

Approximately 76% of PTEN mutation-negative patient samples with relative miR-19a

expression > 2.26 had developed breast cancer. Similarly, 70% of those with high miR-

21 expression (2-∆∆CT >1.81) also developed breast cancer. Although the total number of

PTEN mutation-negative patients included in our study who developed endometrial

cancer was small (N = 16), none of these patients were among the subgroups identified

134

with relatively low miR-19a and miR-21 expression (2-∆∆CT <0.87 and <0.64, respectively).

Macrocephaly was observed in 64% of PTEN mutation-negative patients with

miR-19a expression > 2.60, while this feature was observed in 71% of patients with miR-

21 expression >1.94. Analysis of the CS and CSL phenotypes revealed that 72% of

PTEN mutation-negative patients with high miR-19a expression (2-∆∆CT >2.32) were

classic CS patients, while 67% of those with low miR-19a expression (2-∆∆CT <0.75) were

CSL patients. Similarly, 79% of patients with high miR-21 expression (2-∆∆CT >1.56) were CS patients. CART analysis was not able to identify any unique clusters of patients with thyroid cancer among the PTEN mutation-negative patients using either miRNA as predictors.

Each phenotype was also assessed using the relative expression of both miRNAs

jointly as phenotypic predictors. While expression of neither miRNA was predictive of

thyroid cancer, as was the case when each was considered separately, high miR-21

expression proved to be the single strongest predictor for endometrial cancer,

macrocephaly, and the CS/CSL phenotype as the joint analysis using both miR-19a and

miR-21 together identified the same clusters as did miR-21 when considered

independently.

5.4 DISCUSSION

Alternate mechanisms of PTEN dysfunction are becoming increasingly germane

in the PTEN hamartoma tumor syndromes, particularly in CS and BRRS (56, 146, 147,

160). Despite a better understanding of these genetic and molecular mechanisms, the

135

factors that underlie disease susceptibility and contribute to the phenotypic diversity in these syndromes remain to be fully elucidated. An area of research that has garnered considerable interest of late is the role of miRNAs in various human diseases, including cardiovascular disease, Alzheimer’s disease, psoriasis, and several metabolic conditions, including type 2 diabetes mellitus (135-139, 161). These negative gene regulators are also known to be involved in regulating the expression of several tumor suppressors and oncogenes, including PTEN (143, 151, 153). However, these have yet to be examined in

CS and BRRS.

Our study suggests that differential expression of both PTEN-targeting miRNAs modulate PTEN protein levels and the CS/CSL phenotype, irrespective of patient mutation status. Variable PTEN protein levels were inversely correlated with miR-19a and miR-21 expression levels in PTEN mutation-positive patients and, more specifically, only in patients with R130X and R233X mutations. In R335X, where the relationship among miR expression, PTEN protein levels, and clinical diagnosis was absent, 7 of 10 patients had classic CS with full-blown phenotypic features. This observation suggests that the R335X PTEN genotype strongly influences phenotype, and indeed, this is corroborated by the miR-19a/miR-21-independent overall decreased PTEN protein levels in this R335X group of patients. In contrast, we believe that differential expression of miR-19a and miR-21, which is directly correlated with peripheral PTEN protein levels in

R130X and R233X mutation carriers, may help modulate phenotype.

Similarly, miR-19a and miR-21 was differentially expressed in a series of mutation-negative CS and CSL patients with variable clinical phenotypes and decreased expression of PTEN protein. Importantly, decreased expression of PTEN transcript

136

levels was not observed in these mutation-negative patients, suggesting that their decreased PTEN protein expression is likely due to dysregulation at the protein level.

Among the various clinical features examined in this data set, we only detected an association with the presence of pathognomonic features among patients in whom miR-21 was over-expressed. Our data also suggest that miR-21 over-expression may contribute to the risk of endometrial cancer in CS, however, a larger sample size would be needed to affirm this potential association.

Taken together, our data in both PTEN mutation-positive and mutation-negative patients demonstrate that miR-19a and miR-21 can modulate PTEN protein levels.

Because PTEN’s sufficient activity is dependent on its protein levels, perturbation of its expression can enhance disease progression and facilitate tumorigenesis (91). Therefore, our data suggest essential roles for miR-19a and miR-21 in modulating the diverse clinical and molecular phenotypes observed in CS.

The identification of clear genotype-phenotype correlations has, for the most part, proven elusive in CS and prompted speculation that other loci contribute to the variable clinical spectrum observed in this syndrome (52, 57). Despite a lack of genetic heterogeneity in CS, recent studies have suggested that other genetic factors, such as modifier loci, contribute to disease susceptibility and the variable phenotypes observed in patients with this disorder (4, 6, 32, 57). Interestingly, a study by Freeman et al demonstrated that phenotypic differences, including the onset and incidence of tumor formation, differed among mice with identical PTEN mutations yet with different genetic backgrounds (32). These differences correspond well with the vast clinical spectrum common in patients with CS and BRRS, suggesting that genetic modifiers, likely

137

including other genes, miRNAs, and proteins (for example PICT-1), account for the observed differences in human patients (162). Our data are highly suggestive that miR-

19a and miR-21, at least in part, contribute to this phenotypic variability.

In accordance with this, neither miR-19a nor miR-21 were differentially expressed in patients with the BRRS phenotype. While we interpret this with caution, due to the small number of BRRS patients included in our study, our observation that these miRNAs are not differentially expressed in BRRS, while being associated with CS (in both patients with and without germline PTEN mutations), suggests their role as modifiers in these related syndromes. Together, these data support miR-19a and miR-

21’s role as modifiers in CS and CSL, and not in BRRS. While these disorders are allelic, their underlying modifiers differ. If this observation can be replicated in a larger series of BRRS, then it is tempting to speculate that miR-19a/miR-21 modulation of

PTEN leads to CS/CSL while its absence of modulatory influence, likely together with other mechanisms, results in the BRRS phenotype.

Our investigation of miR-19a and miR-21 in patients with CS and CSL is highly suggestive of their role in these diseases. The mechanism by which these miRNAs are deregulated in these patients, however, is not addressed in the current study and, thus, remains unknown. Potential mechanisms that may contribute to the aberrant expression of miR-19a and miR-21 in CS and CSL include gene amplification, genetic polymorphisms at miRNA-binding sites within PTEN’s 3’UTR, and alterations within the promoter region of each miRNA. Based on our findings, further investigation to better understand these potential mechanisms in patients with CS and CSL is warranted.

138

Our study is the first patient-based study to examine miR-19a and miR-21 expression in CS, CSL, and BRRS. Along with other, yet to be identified, genetic modifiers, these miRNAs may contribute to disease susceptibility and phenotypic modulation and may serve as potential biomarkers of the risks associated with each syndrome. It is our hope that these findings will improve our understanding of the pathogenesis of CS, CSL, and BRRS in patients, both in those with defined germline

PTEN mutations and in those where traditional mutational scanning methodologies have been unable to uncover a genetic cause. We believe an improved understanding of the role of these, and other, modifier loci in CS, CSL, and BRRS will enable more accurate clinical and molecular phenotyping and lead to advances in both the diagnostic and preventive care afforded to patients afflicted with these syndromes.

139

CHAPTER 6

DISCUSSION AND FUTURE DIRECTIONS

PHTS is a collection of phenotypically diverse syndromes that share many

common clinical manifestations and a common genetic etiology. In addition to being at an increased risk of developing neoplasia, particularly in the case of patients diagnosed with CS, PHTS’s benign features are also associated with increased patient morbidity.

Because CS presents with a complex array of benign and malignant clinical manifestations, this syndrome often goes unrecognized, leaving patients with this syndrome and their family members unknowingly at risk of disease and without proper risk management and cancer surveillance.

Despite the successes of PTEN mutation scanning in CS (focused exclusively on

the gene’s coding sequence and promoter region), for a subset of patients, pathogenic

mutations have yet to be identified. This is particularly troubling for patients who do not

meet the full diagnostic criteria of this syndrome, as >80% of these CSL patients have

disease with an unknown etiology. Work done in our laboratory has shown that despite

the lack of identifiable mutations in these patients, a significant proportion have PTEN

dysfunction, thereby suggesting that mechanisms of PTEN dysfunction beyond

140

structural genetic alterations within PTEN and its promoter are likely contributors in

these syndromes. In this dissertation, we hypothesized that such alternate mechanisms

underlie PTEN dysfunction and contribute to disease susceptibility in these patients. We

addressed this hypothesis using several approaches aimed at investigating novel genetic

and biochemical mechanisms involved in the regulation and potential deregulation of

PTEN in PHTS patients lacking PTEN mutations, as well as in a series of patients with previously identified PTEN mutations.

Our initial investigation to elucidate novel causes of PTEN dysfunction in PHTS patients utilized a haplotype-based approach across the entire PTEN locus. In this study

(Chapter 2), we identified significant differences among mutation-negative PHTS

patients and controls for the 163-kb haplotype spanning this entire locus, as well as

among the 3 haplotype blocks that form this extended haplotype. Interestingly, the

strongest association among all PTEN mutation-negative patients was identified at a haplotype block spanning a region upstream of PTEN and including a portion of the

gene’s first intron, suggesting that this region likely contains, or is in strong LD with,

genetic alterations or elements that contribute to PTEN’s deregulation in these patients.

Closer inspection of the haplotype blocks and extended haplotypes revealed an excess of

‘rare’ haplotypes in PHTS patients. This finding is potentially very significant,

particularly as data from the HapMap project and advances in genotyping platforms now

allow investigators to interrogate up to 1,000,000 SNPs across the human genome. Given

that most haplotype-based studies focus on identifying common haplotypes, in cases

where the causal allele is likely to be under-represented, uncommon haplotypes may

actually be more informative. This notion poses additional challenge to such studies, as

141

current designs are likely to be underpowered to detect these more modest effects.

Despite this challenge, however, our study was able to show that specific haplotypes, as well as rare alleles, across the PTEN locus are associated with disease susceptibility in a subset of PHTS patients. Within this population, these haplotypes and rare alleles likely constitute low-penetrance, modifier loci, and in the case of patients with PHTS for whom traditional mutations have yet to be identified, these haplotypes may harbor pathogenic variants that have escaped detection by standard PTEN mutation-scanning technologies.

Having identified a region strongly associated with PHTS in patients negative for

PTEN mutations, we subsequently sought to examine potential functional cis-regulatory elements contained within this region and that may be involved in PTEN dysfunction and disease pathogenesis. Using a comparative genomic approach combined with molecular genetic techniques, in Chapter 3, we identified a highly conserved E-box motif upstream of the PTEN promoter that we show to be specifically bound by USF1 and USF2, two members of the bHLH-LZ transcription factor family not previously known to regulate

PTEN’s transcription. Although mutation analysis failed to identify genetic alterations either within or adjacent to this motif, deletion analysis allowed us to identify one classic

CS patient with a germline deletion that localized to this E-box region and, thereby provided evidence that this element is important, and necessary, in regulating PTEN and that alterations at this region likely contribute to pathogenesis in CS.

In this study, we interrogated 1 of 20 highly conserved regions found to localize either upstream of PTEN or within the gene’s first intron, the same region our haplotype study found to be associated with disease in PHTS patients without PTEN mutations.

These other regions potentially contain additional elements with regulatory potential. To

142

look at this further, we supplemented our initial multi-species comparison with sequence

data available from 6 other species, bringing the total number of compared species to 9.

Comparisons with mouse and rat showed the most extensive conservation throughout this

region, however, inclusion of opossum identified several highly conserved non-coding

elements that localized primarily within intron 1 of PTEN, approximately 14,500bp

downstream of the gene’s first exon. This finding prompted us to examine this region a

bit more closely and, upon further inspection, revealed that 3 spliced expressed sequence tags (ESTs), BG218135, AI378794, and BG540908, localize to this intron, with 1

(BG540908) mapping in close proximity to these conserved regions. Sequencing of the inserts contained within EST clones AI378794 (869bp insert) and BG540908 (1464bp insert) suggest the existence of two novel PTEN isoforms (Fig. 6.1). AI378794 appears

to share exon 1 with the full-length PTEN isoform and has an additional novel exon located at position IVS1+2125 to IVS1+2501. This 377bp exon splices appropriately

(i.e. with consensus donor and acceptor sites) and contains a single polyadenylation site.

Interestingly, it appears that this transcript does not contain an open reading frame,

suggesting that it is a non-coding RNA, and therefore may potentially have some as yet

unknown regulatory function. BG540908 contains 3 novel exons located at positions

IVS1+14143 to IVS1+14475, IVS1-13953 to IVS1-13915, and IVS1-13725 to IVS-

13587, while lacking exon 6 of the full-length transcript. RT-PCR analysis using cDNA

derived from LBCLs and primers located within these novel exons further confirmed

their expression and suggest that they are naturally occurring PTEN splice variants. And,

although EST clone BG218135 is proprietary, RT-PCR analysis suggests that this EST is

also expressed and that a novel exon resides at position IVS1-5503 to IVS1-5752. While

143

Figure 6.1. Alignment of novel PTEN isoforms. Alignment of re-sequenced AI378794 and BG540908 clone inserts (869bp and 1464bp, respectively) and RT-PCR product obtained from BG218135 (304bp) with respect to PTEN using the UCSC Genome

Browser (http://genome.ucsc.edu).

these data suggest the existence of 3 novel PTEN isoforms, further characterization of these transcripts is needed.

Based on these findings, however, it would be interesting to examine the relative gene expression of these 3 novel isoforms in CS, CSL, and BRRS PTEN mutation- negative patients. Indeed, previous work in our laboratory has shown that naturally occurring PTEN splice variants are differentially expressed in PTEN mutation-negative

144

CS and CSL patients compared to normal controls (160). In this study, a total of 8 splice variants were examined. Many of these variants have been shown to function similarly to full-length PTEN, while others function in opposition to this (163). One such splice variant, SV-5b, increases cyclin D1 promoter activity, an activity typically decreased by

PTEN’s full-length transcript. This particular isoform was also shown to be over- expressed in several breast cancer cell lines and, perhaps more importantly, was found to be significantly over-expressed in PTEN mutation-negative CS and CSL patients compared to normal controls (160). Similar studies to examine the relative expression of the 3 novel PTEN isoforms partially located within intron 1 of PTEN will likely add to this work and facilitate a better understanding of the role of these splice variants in regulating PTEN and in these various syndromes.

In the studies described in Chapters 2 and 3, we also identified germline PTEN deletions in 6 CS/CSL patients previously found to be negative for mutations within the

PTEN’s coding sequence and its promoter region. Prior to this work, germline deletions had only been reported in patients with BRRS (50). The deletions we identified ranged in size; including 2 that spanned the entire PTEN locus, 3 that included exon 1 of PTEN but not the region 3’, and 1 that was shown to localize to a region upstream of PTEN’s promoter. The detection of germline deletions in this subset of CS/CSL patients has significant clinical implications, suggesting that PTEN deletion analysis is warranted in all mutation-negative PHTS patients, in addition to those with the BRRS phenotype.

An area of research that has garnered much attention over the past few years is the investigation of small non-coding regulatory RNAs, particularly miRNAs, and their role in disease pathogenesis. These negative gene regulators have been examined in a

145

variety of human diseases, however, their deregulation has most frequently been

described in cancer. Specifically, aberrant miRNA expression has been reported in chronic lymphocytic leukemia, glioblastoma, lymphoma, breast, colorectal, and lung cancer (154-158, 164-166). Additionally, our laboratory has recently shown that miR-

197 and miR-346 are over-expressed in follicular thyroid carcinoma (167). Evidence to

further support their role in malignancy comes from studies that have found that miRNAs

are frequently located either within or near regions thought to contain cancer

susceptibility loci (156, 168). Given their increasingly important roles in carcinogenesis

and disease susceptibility, and as negative regulators of gene expression, in Chapters 4

and 5 we sought to investigate these important molecules relative to PTEN and PHTS.

In Chapter 4, we discussed findings that demonstrate that miR-519e, a miRNA

computationally predicted to target PTEN, can bind to this gene’s 3’UTR, repress its

expression, and, thereby, serve as a negative regulator of its tumor suppressor activity. At

the onset of this study, our hope was to identify novel PTEN-targeting miRNAs and

assess the relative expression of these miRNAs in RNA derived from PTEN mutation-

positive and mutation-negative PHTS patients. Because of miR-519e’s low expression in

LBCLs, we were not able to extend this study further and interrogate this miRNA in our

patient samples. However, as previous studies have shown that 2 other miRNAs, miR-

19a and miR-21, also bind to PTEN and repress expression of its protein, our study was able to increase the total number of experimentally validated PTEN-targeting miRNAs.

This study highlights several important challenges facing this field. Firstly, while

we were able to experimentally validate miR-519e’s involvement in the regulation of

PTEN in vivo, 3 additional miRNAs computationally predicted to target PTEN had no

146

affect on PTEN’s expression in our assay. Although this apparently negative result could be an artifact of our assay, given the vast number of miRNAs predicted to target each mRNA, in many cases more than 1,000, many of these prediction interactions are certainly likely to be spurious. Secondly, in addition to the excessive number of putative miRNA:target interactions predicted by these algorithms, there is very little overlap among the predictions they generate. These discrepancies complicate the selection of the most appropriate putative miRNA:target interactions with which investigators can then use for subsequent experimental validation. And lastly, estimates of the total number of putative miRNA:target interactions in humans, approximately 38,000-47,500, reveal that the current number of experimentally validated interactions is grossly undersized (169).

Presently, only 461 interactions have been reported

(http://www.diana.pcbi.upenn.edu/tarbase/html. Given their importance in many biological processes, there is a vast need for the expansion of these efforts. Undoubtedly, as this field matures, these challenges are likely to be met and researchers will then be able to better examine their role in human biological and disease processes.

In Chapter 5, we examined the relative expression of miR-19a and miR-21 in

RNA derived from LBCLs in PTEN mutation-positive and mutation-negative CS, CSL, and BRRS patients. As part of these efforts, we compared the relative expression of both miRNAs among CS, CSL, CS/BRRS, and BRRS patients with R130X, R233X, or

R335X mutations and, more specifically, among those with variable PTEN protein expression despite sharing identical PTEN mutations. Interestingly, our data demonstrate that miR-19a and miR-21 expression levels are inversely correlated with PTEN protein levels in PTEN mutation-positive carriers of R130X and R233X mutations and,

147

moreover, that these levels modulate PTEN protein levels and contribute to the

phenotypic heterogeneity among this subgroup of patients. Additionally, we observed

that both miRNAs were differentially expressed in CS and CSL patients who lacked

detectable PTEN mutations. Together, these data provide initial support that miR-19a and miR-21 over-expression manifest in the deregulation of PTEN protein in patients with CS and CSL and that both act as genetic modifiers of these phenotypes.

While our investigation of miR-19a and miR-21 in patients with CS is highly

suggestive of their role in these diseases, the mechanism by which these miRNAs are

deregulated in these patients is currently unknown. miR-19a and miR-21 localize to

chromosome sub-bands 13q31.3 and 17q23.1, respectively. While LOH has been

described in each of these regions, these loci have not been genetically linked to CS,

CSL, or BRRS to date (6, 59, 170, 171). However, given the limited power of linkage

studies to detect genes or loci of modest affect, this does not rule out the possibility that

alterations at these sites could contribute to their aberrant expression in these patients.

Additionally, genetic polymorphisms at miRNA binding sites have recently been shown

to alter miRNA:target gene interactions and, although no genetic alterations have been

identified in PTEN’s 3’UTR near miR-19a’s reported binding site (miR-21’s precise

binding site in unknown), it is interesting to speculate whether genetic variations in

PTEN’s 3’UTR further contribute to the variable phenotypic spectrum of CS and CSL by

altering the binding of other potential miRNAs. Furthermore, dysregulation of miR-19a

and miR-21 may be caused by factors that modulate expression of each miRNA’s primary

miRNA or, perhaps, the processing of their mature miRNA. Based on our findings

148

detailed in this chapter, further investigation to better understand these potential

mechanisms in patients with CS and CSL is warranted.

Another area of research in PHTS that has not been extensively investigated is the likelihood that its component syndromes may be caused by mutations in genes other than

PTEN. While genetic evidence gathered in mapping the CS disease locus suggests that this syndrome is linked to the 10q23 region without genetic heterogeneity, these studies were appropriately powered to identify genes with a major genetic contribution and, thereby, more modest affects may have gone undetected (6). Furthermore, it is not currently known whether other loci contribute to BRRS. Potentially, mutations in other genes may contribute to disease susceptibility in these syndromes. For example, our group has recently identified genetic variants in SDHB and SDHD, two genes involved in

the mitochondrial complex II, in PTEN mutation-negative patients with CS and CSL

(Zbuk et al., unpublished data). Interestingly, carriers of these variants have an increased

incidence of renal cell and thyroid carcinoma compared to PTEN mutation-positive CS

patients.

One recently characterized candidate gene that could contribute to CS and BRRS susceptibility is the glioma tumor suppressor critical region 2 (GLTSCR2

[MIM#605691]) gene (172). This 13 exon gene located on chromosome 19q was initially characterized through efforts aimed at identifying putative candidate genes within a region of frequent LOH in human gliomas. Subsequent studies revealed that GLTSCR2’s protein product, later named protein interacting with carboxyl terminus 1 (PICT-1), binds to and stabilizes PTEN protein (162). Based on this evidence, mutations in the GLTSCR2 gene could contribute to CS and BRRS, particularly in patients without germline PTEN

149

mutations and with deficient PTEN protein levels. To investigate this possibility, we

examined all exons and intron/exon boundaries of GLTSCR2 for genetic alterations in 48

PTEN mutation-negative CS/BRRS patients, 50% of whom had decreased PTEN protein

expression. In total, we identified 25 sequence variations (Fig. 6.2A). Of these sequence

differences, 15 were within the gene’s introns, 6 were synonymous changes, and 4 were

non-synonymous changes. All 4 of these non-synonymous changes were identified in

classic CS patients. One of these changes, S6S, was previously reported in dbSNP to

have a minor allele frequency of 11% in a European population, while the other 3

variants had not been described in this database. To examine the D31H, A260V, and

T284M variations further, we sequenced each of these changes in 100 healthy control

subjects. No differences in allele frequencies were noted for both the D31H and T284M

variants, however we did not observe the A260V variant in the 200 control chromosomes we examined. This variant was present in only 1 of our patients whose history included

breast cancer, fibrocystic breast disease, multi-nodule goiter, endometrial fibroids, and

gastrointestinal hamartomas. Moreover, this patient also exhibited a 50% reduction in

PTEN protein levels and increased p-Akt protein relative to healthy control subjects.

Interestingly, all 3 of these changes are predicted by the SIFT algorithm

(http://www.blocks.fhcrc.org/sift) to impact function of the PICT-1 protein , however additional experiments are necessary to further characterize any potential role these

variants may have in disease (Fig. 6.2B).

150

Figure 6.2. Schematic of sequence variations identified in GLTSCR2 and SIFT predictions. A) 25 sequence variations that were identified in PTEN mutation-negative

CS/BRRS patients are shown. Non-synonymous changes not listed in dbSNP are indicated in red. B) SIFT predictions for D31H, A260V, and T284M variations.

Coupled with the results from our studies detailed in Chapters 2-5, the preliminary data introduced in this chapter highlights the need to investigate novel mechanisms of PTEN dysfunction in patients with CS, CSL, and BRRS. Through the course of these studies, we have employed a series of genetic and biochemical approaches in order to improve our understanding of the mechanisms that contribute to

PTEN’s regulation and deregulation in PHTS and have tried to translate these efforts into 151

Figure 6.3. Mechanisms of PTEN dysfunction in PHTS.

clinically meaningful results. Our data provide evidence that support novel mechanisms of PTEN dysfunction in patients who lack mutations in PTEN’s coding sequence and promoter region. While mutations in these regions account for disease in the majority of patients affected with these syndromes, our data point to genetic alterations outside of these routinely screened regions as contributors of disease in patients whose pathogenicity was previously unexplained. Our work has identified several alternate

152

mechanisms that are likely to contribute to PTEN dysfunction in patients who lack PTEN

mutations (Fig. 6.3). As well, many of these factors are also likely to contribute to disease in patients with known mutations, including the phenotypic variability associated with these complex human disorders.

A goal of genomic medicine is to predict disease susceptibility and to personalize healthcare through the tailored management of these risks. In order to achieve this goal, careful analysis of risk genotypes and disease phenotypes is required. In many cases, as in the case of CS and BRRS, the correlations between these are not always clear, and, as such, require looking beyond ‘traditional’ causes of disease. Our data suggest that alternate mechanisms of PTEN dysfunction do contribute to disease in patients with these syndromes. It is our hope that these studies will enable more accurate molecular phenotyping in CS and BRRS and, through these efforts, enable patients and their family members access to better diagnostic and personalized care.

153

LIST OF REFERENCES

1. Lloyd, K.M., 2nd and Dennis, M. (1963) Cowden's disease. A possible new symptom complex with multiple system involvement. Ann Intern Med, 58, 136- 42.

2. Pilarski, R. and Eng, C. (2004) Will the real Cowden syndrome please stand up (again)? Expanding mutational and clinical spectra of the PTEN hamartoma tumour syndrome. J Med Genet, 41, 323-6.

3. Gustafson, S., Zbuk, K.M., Scacheri, C. and Eng, C. (2007) Cowden syndrome. Semin Oncol, 34, 428-34.

4. Zbuk, K.M. and Eng, C. (2007) Cancer phenomics: RET and PTEN as illustrative models. Nat Rev Cancer, 7, 35-45.

5. Walter, J.B. and Israel, M.S. (1975) General Pathology. 4 ed. Churchill Livingstone, London.

6. Nelen, M.R., Padberg, G.W., Peeters, E.A., Lin, A.Y., van den Helm, B., Frants, R.R., Coulon, V., Goldstein, A.M., van Reen, M.M., Easton, D.F. et al. (1996) Localization of the gene for Cowden disease to chromosome 10q22-23. Nat Genet, 13, 114-6.

7. Eng, C. (2003) PTEN: one gene, many syndromes. Hum Mutat, 22, 183-98.

8. Li, J., Yen, C., Liaw, D., Podsypanina, K., Bose, S., Wang, S.I., Puc, J., Miliaresis, C., Rodgers, L., McCombie, R. et al. (1997) PTEN, a putative protein tyrosine phosphatase gene mutated in human brain, breast, and prostate cancer. Science, 275, 1943-7.

154

9. Li, D.M. and Sun, H. (1997) TEP1, encoded by a candidate tumor suppressor locus, is a novel protein tyrosine phosphatase regulated by transforming growth factor beta. Cancer Res, 57, 2124-9.

10. Steck, P.A., Pershouse, M.A., Jasser, S.A., Yung, W.K., Lin, H., Ligon, A.H., Langford, L.A., Baumgard, M.L., Hattier, T., Davis, T. et al. (1997) Identification of a candidate tumour suppressor gene, MMAC1, at chromosome 10q23.3 that is mutated in multiple advanced cancers. Nat Genet, 15, 356-62.

11. Liaw, D., Marsh, D.J., Li, J., Dahia, P.L., Wang, S.I., Zheng, Z., Bose, S., Call, K.M., Tsou, H.C., Peacocke, M. et al. (1997) Germline mutations of the PTEN gene in Cowden disease, an inherited breast and thyroid cancer syndrome. Nat Genet, 16, 64-7.

12. Maehama, T. and Dixon, J.E. (1998) The tumor suppressor, PTEN/MMAC1, dephosphorylates the lipid second messenger, phosphatidylinositol 3,4,5- trisphosphate. J Biol Chem, 273, 13375-8.

13. Myers, M.P., Stolarov, J.P., Eng, C., Li, J., Wang, S.I., Wigler, M.H., Parsons, R. and Tonks, N.K. (1997) P-TEN, the tumor suppressor from human chromosome 10q23, is a dual-specificity phosphatase. Proc Natl Acad Sci U S A, 94, 9052-7.

14. Lee, J.O., Yang, H., Georgescu, M.M., Di Cristofano, A., Maehama, T., Shi, Y., Dixon, J.E., Pandolfi, P. and Pavletich, N.P. (1999) Crystal structure of the PTEN tumor suppressor: implications for its phosphoinositide phosphatase activity and membrane association. Cell, 99, 323-34.

15. Georgescu, M.M., Kirsch, K.H., Akagi, T., Shishido, T. and Hanafusa, H. (1999) The tumor-suppressor activity of PTEN is regulated by its carboxyl-terminal region. Proc Natl Acad Sci U S A, 96, 10182-7.

16. Stokoe, D., Stephens, L.R., Copeland, T., Gaffney, P.R., Reese, C.B., Painter, G.F., Holmes, A.B., McCormick, F. and Hawkins, P.T. (1997) Dual role of phosphatidylinositol-3,4,5-trisphosphate in the activation of protein kinase B. Science, 277, 567-70.

17. Vivanco, I. and Sawyers, C.L. (2002) The phosphatidylinositol 3-Kinase AKT pathway in human cancer. Nat Rev Cancer, 2, 489-501.

155

18. Mayo, L.D. and Donner, D.B. (2001) A phosphatidylinositol 3-kinase/Akt pathway promotes translocation of Mdm2 from the cytoplasm to the nucleus. Proc Natl Acad Sci U S A, 98, 11598-603.

19. Stambolic, V., MacPherson, D., Sas, D., Lin, Y., Snow, B., Jang, Y., Benchimol, S. and Mak, T.W. (2001) Regulation of PTEN transcription by p53. Mol Cell, 8, 317-25.

20. Tang, Y. and Eng, C. (2006) PTEN autoregulates its expression by stabilization of p53 in a phosphatase-independent manner. Cancer Res, 66, 736-42.

21. Gu, J., Tamura, M. and Yamada, K.M. (1998) Tumor suppressor PTEN inhibits integrin- and growth factor-mediated mitogen-activated protein (MAP) kinase signaling pathways. J Cell Biol, 143, 1375-83.

22. Weng, L., Brown, J. and Eng, C. (2001) PTEN induces apoptosis and cell cycle arrest through phosphoinositol-3-kinase/Akt-dependent and -independent pathways. Hum Mol Genet, 10, 237-42.

23. Weng, L.P., Smith, W.M., Brown, J.L. and Eng, C. (2001) PTEN inhibits insulin- stimulated MEK/MAPK activation and cell growth by blocking IRS-1 phosphorylation and IRS-1/Grb-2/Sos complex formation in a breast cancer model. Hum Mol Genet, 10, 605-16.

24. Tamura, M., Gu, J., Matsumoto, K., Aota, S., Parsons, R. and Yamada, K.M. (1998) Inhibition of cell migration, spreading, and focal adhesions by tumor suppressor PTEN. Science, 280, 1614-7.

25. Weng, L.P., Brown, J.L. and Eng, C. (2001) PTEN coordinates G(1) arrest by down-regulating cyclin D1 via its protein phosphatase activity and up-regulating p27 via its lipid phosphatase activity in a breast cancer model. Hum Mol Genet, 10, 599-604.

26. Chung, J.H. and Eng, C. (2005) Nuclear-cytoplasmic partitioning of phosphatase and tensin homologue deleted on chromosome 10 (PTEN) differentially regulates the cell cycle and apoptosis. Cancer Res, 65, 8096-100.

156

27. Liu, J.L., Sheng, X., Hortobagyi, Z.K., Mao, Z., Gallick, G.E. and Yung, W.K. (2005) Nuclear PTEN-mediated growth suppression is independent of Akt down- regulation. Mol Cell Biol, 25, 6211-24.

28. Di Cristofano, A., Pesce, B., Cordon-Cardo, C. and Pandolfi, P.P. (1998) Pten is essential for embryonic development and tumour suppression. Nat Genet, 19, 348-55.

29. Podsypanina, K., Ellenson, L.H., Nemes, A., Gu, J., Tamura, M., Yamada, K.M., Cordon-Cardo, C., Catoretti, G., Fisher, P.E. and Parsons, R. (1999) Mutation of Pten/Mmac1 in mice causes neoplasia in multiple organ systems. Proc Natl Acad Sci U S A, 96, 1563-8.

30. Suzuki, A., de la Pompa, J.L., Stambolic, V., Elia, A.J., Sasaki, T., del Barco Barrantes, I., Ho, A., Wakeham, A., Itie, A., Khoo, W. et al. (1998) High cancer susceptibility and embryonic lethality associated with mutation of the PTEN tumor suppressor gene in mice. Curr Biol, 8, 1169-78.

31. Stambolic, V., Tsao, M.S., Macpherson, D., Suzuki, A., Chapman, W.B. and Mak, T.W. (2000) High incidence of breast and endometrial neoplasia resembling human Cowden syndrome in pten+/- mice. Cancer Res, 60, 3605-11.

32. Freeman, D., Lesche, R., Kertesz, N., Wang, S., Li, G., Gao, J., Groszer, M., Martinez-Diaz, H., Rozengurt, N., Thomas, G. et al. (2006) Genetic background controls tumor development in PTEN-deficient mice. Cancer Res, 66, 6492-6.

33. Aveyard, J.S., Skilleter, A., Habuchi, T. and Knowles, M.A. (1999) Somatic mutation of PTEN in bladder carcinoma. Br J Cancer, 80, 904-8.

34. Cairns, P., Evron, E., Okami, K., Halachmi, N., Esteller, M., Herman, J.G., Bose, S., Wang, S.I., Parsons, R. and Sidransky, D. (1998) Point mutation and homozygous deletion of PTEN/MMAC1 in primary bladder cancers. Oncogene, 16, 3215-8.

35. Cairns, P., Okami, K., Halachmi, S., Halachmi, N., Esteller, M., Herman, J.G., Jen, J., Isaacs, W.B., Bova, G.S. and Sidransky, D. (1997) Frequent inactivation of PTEN/MMAC1 in primary prostate cancer. Cancer Res, 57, 4997-5000.

157

36. Celebi, J.T., Shendrik, I., Silvers, D.N. and Peacocke, M. (2000) Identification of PTEN mutations in metastatic melanoma specimens. J Med Genet, 37, 653-7.

37. Halachmi, N., Halachmi, S., Evron, E., Cairns, P., Okami, K., Saji, M., Westra, W.H., Zeiger, M.A., Jen, J. and Sidransky, D. (1998) Somatic mutations of the PTEN tumor suppressor gene in sporadic follicular thyroid tumors. Genes Chromosomes Cancer, 23, 239-43.

38. Hosoya, Y., Gemma, A., Seike, M., Kurimoto, F., Uematsu, K., Hibino, S., Yoshimura, A., Shibuya, M. and Kudoh, S. (1999) Alteration of the PTEN/MMAC1 gene locus in primary lung cancer with distant metastasis. Lung Cancer, 25, 87-93.

39. Kurose, K., Gilley, K., Matsumoto, S., Watson, P.H., Zhou, X.P. and Eng, C. (2002) Frequent somatic mutations in PTEN and TP53 are mutually exclusive in the stroma of breast carcinomas. Nat Genet, 32, 355-7.

40. Kurose, K., Zhou, X.P., Araki, T., Cannistra, S.A., Maher, E.R. and Eng, C. (2001) Frequent loss of PTEN expression is linked to elevated phosphorylated Akt levels, but not associated with p27 and cyclin D1 expression, in primary epithelial ovarian carcinomas. Am J Pathol, 158, 2097-106.

41. Kurose, K., Zhou, X.P., Araki, T. and Eng, C. (2000) Biallelic inactivating mutations and an occult germline mutation of PTEN in primary cervical carcinomas. Genes Chromosomes Cancer, 29, 166-72.

42. Perren, A., Komminoth, P., Saremaslani, P., Matter, C., Feurer, S., Lees, J.A., Heitz, P.U. and Eng, C. (2000) Mutation and expression analyses reveal differential subcellular compartmentalization of PTEN in endocrine pancreatic tumors compared to normal islet cells. Am J Pathol, 157, 1097-103.

43. Poetsch, M., Lorenz, G. and Kleist, B. (2002) Detection of new PTEN/MMAC1 mutations in head and neck squamous cell carcinomas with loss of chromosome 10. Cancer Genet Cytogenet, 132, 20-4.

44. Risinger, J.I., Hayes, A.K., Berchuck, A. and Barrett, J.C. (1997) PTEN/MMAC1 mutations in endometrial cancers. Cancer Res, 57, 4736-8.

158

45. Wang, S.I., Puc, J., Li, J., Bruce, J.N., Cairns, P., Sidransky, D. and Parsons, R. (1997) Somatic mutations of PTEN in glioblastoma multiforme. Cancer Res, 57, 4183-6.

46. Mutter, G.L., Lin, M.C., Fitzgerald, J.T., Kum, J.B., Baak, J.P., Lees, J.A., Weng, L.P. and Eng, C. (2000) Altered PTEN expression as a diagnostic marker for the earliest endometrial precancers. J Natl Cancer Inst, 92, 924-30.

47. Alvarez-Nunez, F., Bussaglia, E., Mauricio, D., Ybarra, J., Vilar, M., Lerma, E., de Leiva, A. and Matias-Guiu, X. (2006) PTEN promoter methylation in sporadic thyroid carcinomas. Thyroid, 16, 17-23.

48. Khan, S., Kumagai, T., Vora, J., Bose, N., Sehgal, I., Koeffler, P.H. and Bose, S. (2004) PTEN promoter is methylated in a proportion of invasive breast cancers. Int J Cancer, 112, 407-10.

49. Zhou, X.P., Gimm, O., Hampel, H., Niemann, T., Walker, M.J. and Eng, C. (2000) Epigenetic PTEN silencing in malignant melanomas without PTEN mutation. Am J Pathol, 157, 1123-8.

50. Zhou, X.P., Waite, K.A., Pilarski, R., Hampel, H., Fernandez, M.J., Bos, C., Dasouki, M., Feldman, G.L., Greenberg, L.A., Ivanovich, J. et al. (2003) Germline PTEN promoter mutations and deletions in Cowden/Bannayan-Riley- Ruvalcaba syndrome result in aberrant PTEN protein and dysregulation of the phosphoinositol-3-kinase/Akt pathway. Am J Hum Genet, 73, 404-11.

51. Eng, C. (2000) Will the real Cowden syndrome please stand up: revised diagnostic criteria. J Med Genet, 37, 828-30.

52. Marsh, D.J., Coulon, V., Lunetta, K.L., Rocca-Serra, P., Dahia, P.L., Zheng, Z., Liaw, D., Caron, S., Duboue, B., Lin, A.Y. 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, 507-15.

53. Sheng, X., Koul, D., Liu, J.L., Liu, T.J. and Yung, W.K. (2002) Promoter analysis of tumor suppressor gene PTEN: identification of minimum promoter region. Biochem Biophys Res Commun, 292, 422-6.

159

54. Zhou, X.P., Marsh, D.J., Morrison, C.D., Chaudhury, A.R., Maxwell, M., Reifenberger, G. and Eng, C. (2003) Germline inactivation of PTEN and dysregulation of the phosphoinositol-3-kinase/Akt pathway cause human Lhermitte-Duclos disease in adults. Am J Hum Genet, 73, 1191-8.

55. Virolle, T., Adamson, E.D., Baron, V., Birle, D., Mercola, D., Mustelin, T. and de Belle, I. (2001) The Egr-1 transcription factor directly activates PTEN during irradiation-induced signalling. Nat Cell Biol, 3, 1124-8.

56. Teresi, R.E., Zbuk, K.M., Pezzolesi, M.G., Waite, K.A. and Eng, C. (2007) Cowden syndrome-affected patients with PTEN promoter mutations demonstrate abnormal protein translation. Am J Hum Genet, 81, 756-67.

57. Marsh, D.J., Kum, J.B., Lunetta, K.L., Bennett, M.J., Gorlin, R.J., Ahmed, S.F., Bodurtha, J., Crowe, C., Curtis, M.A., Dasouki, M. et al. (1999) PTEN mutation spectrum and genotype-phenotype correlations in Bannayan-Riley-Ruvalcaba syndrome suggest a single entity with Cowden syndrome. Hum Mol Genet, 8, 1461-72.

58. Gorlin, R.J., Cohen, M.M., Jr., Condon, L.M. and Burke, B.A. (1992) Bannayan- Riley-Ruvalcaba syndrome. Am J Med Genet, 44, 307-14.

59. Marsh, D.J., Dahia, P.L., Zheng, Z., Liaw, D., Parsons, R., Gorlin, R.J. and Eng, C. (1997) Germline mutations in PTEN are present in Bannayan-Zonana syndrome. Nat Genet, 16, 333-4.

60. Wiedemann, H.R., Burgio, G.R., Aldenhoff, P., Kunze, J., Kaufmann, H.J. and Schirg, E. (1983) The proteus syndrome. Partial gigantism of the hands and/or feet, nevi, hemihypertrophy, subcutaneous tumors, macrocephaly or other skull anomalies and possible accelerated growth and visceral affections. Eur J Pediatr, 140, 5-12.

61. Smith, J.M., Kirk, E.P., Theodosopoulos, G., Marshall, G.M., Walker, J., Rogers, M., Field, M., Brereton, J.J. and Marsh, D.J. (2002) Germline mutation of the tumour suppressor PTEN in Proteus syndrome. J Med Genet, 39, 937-40.

62. Zhou, X., Hampel, H., Thiele, H., Gorlin, R.J., Hennekam, R.C., Parisi, M., Winter, R.M. and Eng, C. (2001) Association of germline mutation in the PTEN tumour suppressor gene and Proteus and Proteus-like syndromes. Lancet, 358, 210-1. 160

63. Waite, K.A. and Eng, C. (2003) BMP2 exposure results in decreased PTEN protein degradation and increased PTEN levels. Hum Mol Genet, 12, 679-84.

64. Weng, L.P., Smith, W.M., Dahia, P.L., Ziebold, U., Gil, E., Lees, J.A. and Eng, C. (1999) PTEN suppresses breast cancer cell growth by phosphatase activity- dependent G1 arrest followed by cell death. Cancer Res, 59, 5808-14.

65. Bonneau, D. and Longy, M. (2000) Mutations of the human PTEN gene. Hum Mutat, 16, 109-22.

66. Carethers, J.M., Furnari, F.B., Zigman, A.F., Lavine, J.E., Jones, M.C., Graham, G.E., Teebi, A.S., Huang, H.J., Ha, H.T., Chauhan, D.P. et al. (1998) Absence of PTEN/MMAC1 germ-line mutations in sporadic Bannayan-Riley-Ruvalcaba syndrome. Cancer Res, 58, 2724-6.

67. Livak, K.J. and Schmittgen, T.D. (2001) Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods, 25, 402-8.

68. Abecasis, G.R. and Cookson, W.O. (2000) GOLD--graphical overview of linkage disequilibrium. Bioinformatics, 16, 182-3.

69. Zhang, K., Deng, M., Chen, T., Waterman, M.S. and Sun, F. (2002) A dynamic programming algorithm for haplotype block partitioning. Proc Natl Acad Sci U S A, 99, 7335-9.

70. Gabriel, S.B., Schaffner, S.F., Nguyen, H., Moore, J.M., Roy, J., Blumenstiel, B., Higgins, J., DeFelice, M., Lochner, A., Faggart, M. et al. (2002) The structure of haplotype blocks in the human genome. Science, 296, 2225-9.

71. Clayton, D., Chapman, J. and Cooper, J. (2004) Use of unphased multilocus genotype data in indirect association studies. Genet Epidemiol, 27, 415-28.

72. Stephens, M., Smith, N.J. and Donnelly, P. (2001) A new statistical method for haplotype reconstruction from population data. Am J Hum Genet, 68, 978-89.

73. Cochran, W. (1954) Some Methods of Strengthening the Common Chi-square Tests. Biometrics, 10, 417-451. 161

74. Drysdale, C.M., McGraw, D.W., Stack, C.B., Stephens, J.C., Judson, R.S., Nandabalan, K., Arnold, K., Ruano, G. and Liggett, S.B. (2000) Complex promoter and coding region beta 2-adrenergic receptor haplotypes alter receptor expression and predict in vivo responsiveness. Proc Natl Acad Sci U S A, 97, 10483-8.

75. Greenwood, T.A. and Kelsoe, J.R. (2003) Promoter and intronic variants affect the transcriptional regulation of the human dopamine transporter gene. Genomics, 82, 511-20.

76. Yu, C., Zhou, Y., Miao, X., Xiong, P., Tan, W. and Lin, D. (2004) Functional haplotypes in the promoter of matrix metalloproteinase-2 predict risk of the occurrence and metastasis of esophageal cancer. Cancer Res, 64, 7622-8.

77. Kleinjan, D.A. and van Heyningen, V. (2005) Long-range control of gene expression: emerging mechanisms and disruption in disease. Am J Hum Genet, 76, 8-32.

78. Haiman, C.A., Stram, D.O., Cheng, I., Giorgi, E.E., Pooler, L., Penney, K., Le Marchand, L., Henderson, B.E. and Freedman, M.L. (2006) Common Genetic Variation at PTEN and Risk of Sporadic Breast and Prostate Cancer. Cancer Epidemiol Biomarkers Prev, 15, 1021-5.

79. Hamilton, J.A., Stewart, L.M., Ajayi, L., Gray, I.C., Gray, N.E., Roberts, K.G., Watson, G.J., Kaisary, A.V. and Snary, D. (2000) The expression profile for the tumour suppressor gene PTEN and associated polymorphic markers. Br J Cancer, 82, 1671-6.

80. Zhang, L., Kendler, K.S. and Chen, X. (2006) Association of the phosphatase and tensin homolog gene (PTEN) with smoking initiation and nicotine dependence. Am J Med Genet B Neuropsychiatr Genet, 141, 10-4.

81. Excoffier, L. and Slatkin, M. (1995) Maximum-likelihood estimation of molecular haplotype frequencies in a diploid population. Mol Biol Evol, 12, 921- 7.

82. Lewontin, R.C. (1964) The Interaction Of Selection And Linkage. Ii. Optimum Models. Genetics, 50, 757-82.

162

83. Bonnen, P.E., Wang, P.J., Kimmel, M., Chakraborty, R. and Nelson, D.L. (2002) Haplotype and linkage disequilibrium architecture for human cancer-associated genes. Genome Res, 12, 1846-53.

84. Niu, T., Qin, Z.S., Xu, X. and Liu, J.S. (2002) Bayesian haplotype inference for multiple linked single-nucleotide polymorphisms. Am J Hum Genet, 70, 157-69.

85. Tishkoff, S.A., Pakstis, A.J., Ruano, G. and Kidd, K.K. (2000) The accuracy of statistical methods for estimation of haplotype frequencies: an example from the CD4 locus. Am J Hum Genet, 67, 518-22.

86. Stephens, M. and Donnelly, P. (2003) A comparison of bayesian methods for haplotype reconstruction from population genotype data. Am J Hum Genet, 73, 1162-9.

87. Maehama, T. and Dixon, J.E. (1999) PTEN: a tumour suppressor that functions as a phospholipid phosphatase. Trends Cell Biol, 9, 125-8.

88. Myers, M.P., Pass, I., Batty, I.H., Van der Kaay, J., Stolarov, J.P., Hemmings, B.A., Wigler, M.H., Downes, C.P. and Tonks, N.K. (1998) The lipid phosphatase activity of PTEN is critical for its tumor supressor function. Proc Natl Acad Sci U S A, 95, 13513-8.

89. Stambolic, V., Suzuki, A., de la Pompa, J.L., Brothers, G.M., Mirtsos, C., Sasaki, T., Ruland, J., Penninger, J.M., Siderovski, D.P. and Mak, T.W. (1998) Negative regulation of PKB/Akt-dependent cell survival by the tumor suppressor PTEN. Cell, 95, 29-39.

90. Waite, K.A. and Eng, C. (2003) From developmental disorder to heritable cancer: it's all in the BMP/TGF-beta family. Nat Rev Genet, 4, 763-73.

91. Waite, K.A. and Eng, C. (2002) Protean PTEN: form and function. Am J Hum Genet, 70, 829-44.

92. Hettinger, K., Vikhanskaya, F., Poh, M.K., Lee, M.K., de Belle, I., Zhang, J.T., Reddy, S.A. and Sabapathy, K. (2007) c-Jun promotes cellular survival by suppression of PTEN. Cell Death Differ, 14, 218-29.

163

93. Patel, L., Pass, I., Coxon, P., Downes, C.P., Smith, S.A. and Macphee, C.H. (2001) Tumor suppressor and anti-inflammatory actions of PPARgamma agonists are mediated via upregulation of PTEN. Curr Biol, 11, 764-8.

94. Vasudevan, K.M., Gurumurthy, S. and Rangnekar, V.M. (2004) Suppression of PTEN expression by NF-kappa B prevents apoptosis. Mol Cell Biol, 24, 1007-21.

95. Whelan, J.T., Forbes, S.L. and Bertrand, F.E. (2007) CBF-1 (RBP-Jkappa) binds to the PTEN promoter and regulates PTEN gene expression. Cell Cycle, 6, 80-4.

96. Dubchak, I., Brudno, M., Loots, G.G., Pachter, L., Mayor, C., Rubin, E.M. and Frazer, K.A. (2000) Active conservation of noncoding sequences revealed by three-way species comparisons. Genome Res, 10, 1304-6.

97. Gilligan, P., Brenner, S. and Venkatesh, B. (2002) Fugu and human sequence comparison identifies novel human genes and conserved non-coding sequences. Gene, 294, 35-44.

98. Hardison, R.C. (2000) Conserved noncoding sequences are reliable guides to regulatory elements. Trends Genet, 16, 369-72.

99. Gottgens, B., Barton, L.M., Gilbert, J.G., Bench, A.J., Sanchez, M.J., Bahn, S., Mistry, S., Grafham, D., McMurray, A., Vaudin, M. et al. (2000) Analysis of vertebrate SCL loci identifies conserved enhancers. Nat Biotechnol, 18, 181-6.

100. Loots, G.G., Locksley, R.M., Blankespoor, C.M., Wang, Z.E., Miller, W., Rubin, E.M. and Frazer, K.A. (2000) Identification of a coordinate regulator of interleukins 4, 13, and 5 by cross-species sequence comparisons. Science, 288, 136-40.

101. Shnyreva, M., Weaver, W.M., Blanchette, M., Taylor, S.L., Tompa, M., Fitzpatrick, D.R. and Wilson, C.B. (2004) Evolutionarily conserved sequence elements that positively regulate IFN-gamma expression in T cells. Proc Natl Acad Sci U S A, 101, 12622-7.

102. Wardrop, S.L. and Brown, M.A. (2005) Identification of two evolutionarily conserved and functional regulatory elements in intron 2 of the human BRCA1 gene. Genomics, 86, 316-28.

164

103. Mayor, C., Brudno, M., Schwartz, J.R., Poliakov, A., Rubin, E.M., Frazer, K.A., Pachter, L.S. and Dubchak, I. (2000) VISTA : visualizing global DNA sequence alignments of arbitrary length. Bioinformatics, 16, 1046-7.

104. Pezzolesi, M.G., Li, Y., Zhou, X.P., Pilarski, R., Shen, L. and Eng, C. (2006) Mutation-positive and mutation-negative patients with cowden and bannayan- riley-ruvalcaba syndromes associated with distinct 10q haplotypes. Am J Hum Genet, 79, 923-34.

105. Baxevanis, A.D. and Vinson, C.R. (1993) Interactions of coiled coils in transcription factors: where is the specificity? Curr Opin Genet Dev, 3, 278-85.

106. Bendall, A.J. and Molloy, P.L. (1994) Base preferences for DNA binding by the bHLH-Zip protein USF: effects of MgCl2 on specificity and comparison with binding of Myc family members. Nucleic Acids Res, 22, 2801-10.

107. Blackwell, T.K., Huang, J., Ma, A., Kretzner, L., Alt, F.W., Eisenman, R.N. and Weintraub, H. (1993) Binding of myc proteins to canonical and noncanonical DNA sequences. Mol Cell Biol, 13, 5216-24.

108. Fisher, F., Crouch, D.H., Jayaraman, P.S., Clark, W., Gillespie, D.A. and Goding, C.R. (1993) Transcription activation by Myc and Max: flanking sequences target activation to a subset of CACGTG motifs in vivo. Embo J, 12, 5075-82.

109. Solomon, D.L., Amati, B. and Land, H. (1993) Distinct DNA binding preferences for the c-Myc/Max and Max/Max dimers. Nucleic Acids Res, 21, 5372-6.

110. Ismail, P.M., Lu, T. and Sawadogo, M. (1999) Loss of USF transcriptional activity in breast cancer cell lines. Oncogene, 18, 5582-91.

111. Viollet, B., Lefrancois-Martinez, A.M., Henrion, A., Kahn, A., Raymondjean, M. and Martinez, A. (1996) Immunochemical characterization and transacting properties of upstream stimulatory factor isoforms. J Biol Chem, 271, 1405-15.

112. Blackwood, E.M. and Eisenman, R.N. (1991) Max: a helix-loop-helix zipper protein that forms a sequence-specific DNA-binding complex with Myc. Science, 251, 1211-7.

165

113. Eisenman, R.N., Tachibana, C.Y., Abrams, H.D. and Hann, S.R. (1985) V-myc- and c-myc-encoded proteins are associated with the nuclear matrix. Mol Cell Biol, 5, 114-26.

114. Miltenberger, R.J., Sukow, K.A. and Farnham, P.J. (1995) An E-box-mediated increase in cad transcription at the G1/S-phase boundary is suppressed by inhibitory c-Myc mutants. Mol Cell Biol, 15, 2527-35.

115. Waters, C.M., Littlewood, T.D., Hancock, D.C., Moore, J.P. and Evan, G.I. (1991) c-myc protein expression in untransformed fibroblasts. Oncogene, 6, 797- 805.

116. Ferre-D'Amare, A.R., Pognonec, P., Roeder, R.G. and Burley, S.K. (1994) Structure and function of the b/HLH/Z domain of USF. Embo J, 13, 180-9.

117. Atchley, W.R. and Fitch, W.M. (1997) A natural classification of the basic helix- loop-helix class of transcription factors. Proc Natl Acad Sci U S A, 94, 5172-6.

118. Cogswell, J.P., Godlevski, M.M., Bonham, M., Bisi, J. and Babiss, L. (1995) Upstream stimulatory factor regulates expression of the cell cycle-dependent cyclin B1 gene promoter. Mol Cell Biol, 15, 2782-90.

119. North, S., Espanel, X., Bantignies, F., Viollet, B., Vallet, V., Jalinot, P., Brun, G. and Gillet, G. (1999) Regulation of cdc2 gene expression by the upstream stimulatory factors (USFs). Oncogene, 18, 1945-55.

120. Davis, P.L., Miron, A., Andersen, L.M., Iglehart, J.D. and Marks, J.R. (1999) Isolation and initial characterization of the BRCA2 promoter. Oncogene, 18, 6000-12.

121. Jaiswal, A.S. and Narayan, S. (2001) Upstream stimulating factor-1 (USF1) and USF2 bind to and activate the promoter of the adenomatous polyposis coli (APC) tumor suppressor gene. J Cell Biochem, 81, 262-77.

122. Reisman, D. and Rotter, V. (1993) The helix-loop-helix containing transcription factor USF binds to and transactivates the promoter of the p53 tumor suppressor gene. Nucleic Acids Res, 21, 345-50.

166

123. Wu, K., Jiang, S.W. and Couch, F.J. (2003) p53 mediates repression of the BRCA2 promoter and down-regulation of BRCA2 mRNA and protein levels in response to DNA damage. J Biol Chem, 278, 15652-60.

124. Mayo, L.D., Dixon, J.E., Durden, D.L., Tonks, N.K. and Donner, D.B. (2002) PTEN protects p53 from Mdm2 and sensitizes cancer cells to chemotherapy. J Biol Chem, 277, 5484-9.

125. Tang, Y. and Eng, C. (2006) p53 down-regulates phosphatase and tensin homologue deleted on chromosome 10 protein stability partially through caspase- mediated degradation in cells with proteasome dysfunction. Cancer Res, 66, 6139-48.

126. Nowak, M., Helleboid-Chapman, A., Jakel, H., Martin, G., Duran-Sandoval, D., Staels, B., Rubin, E.M., Pennacchio, L.A., Taskinen, M.R., Fruchart-Najib, J. et al. (2005) Insulin-mediated down-regulation of apolipoprotein A5 gene expression through the phosphatidylinositol 3-kinase pathway: role of upstream stimulatory factor. Mol Cell Biol, 25, 1537-48.

127. Chiang, C.M. and Roeder, R.G. (1995) Cloning of an intrinsic human TFIID subunit that interacts with multiple transcriptional activators. Science, 267, 531-6.

128. Roy, A.L., Du, H., Gregor, P.D., Novina, C.D., Martinez, E. and Roeder, R.G. (1997) Cloning of an inr- and E-box-binding protein, TFII-I, that interacts physically and functionally with USF1. Embo J, 16, 7091-104.

129. Roy, A.L., Meisterernst, M., Pognonec, P. and Roeder, R.G. (1991) Cooperative interaction of an initiator-binding transcription initiation factor and the helix-loop- helix activator USF. Nature, 354, 245-8.

130. Roy, A.L., Carruthers, C., Gutjahr, T. and Roeder, R.G. (1993) Direct role for Myc in transcription initiation mediated by interactions with TFII-I. Nature, 365, 359-61.

131. Pajukanta, P., Lilja, H.E., Sinsheimer, J.S., Cantor, R.M., Lusis, A.J., Gentile, M., Duan, X.J., Soro-Paavonen, A., Naukkarinen, J., Saarela, J. et al. (2004) Familial combined hyperlipidemia is associated with upstream transcription factor 1 (USF1). Nat Genet, 36, 371-6.

167

132. Pillai, R.S., Bhattacharyya, S.N. and Filipowicz, W. (2007) Repression of protein synthesis by miRNAs: how many mechanisms? Trends Cell Biol, 17, 118-26.

133. Calin, G.A. and Croce, C.M. (2006) MicroRNA signatures in human cancers. Nat Rev Cancer, 6, 857-66.

134. Bartel, D.P. (2004) MicroRNAs: genomics, biogenesis, mechanism, and function. Cell, 116, 281-97.

135. Cheng, Y., Ji, R., Yue, J., Yang, J., Liu, X., Chen, H., Dean, D.B. and Zhang, C. (2007) MicroRNAs are aberrantly expressed in hypertrophic heart: do they play a role in cardiac hypertrophy? Am J Pathol, 170, 1831-40.

136. Lukiw, W.J. (2007) Micro-RNA speciation in fetal, adult and Alzheimer's disease hippocampus. Neuroreport, 18, 297-300.

137. Martin, M.M., Buckenberger, J.A., Jiang, J., Malana, G.E., Nuovo, G.J., Chotani, M., Feldman, D.S., Schmittgen, T.D. and Elton, T.S. (2007) The human angiotensin II type 1 receptor +1166 A/C polymorphism attenuates microRNA- 155 binding. J Biol Chem.

138. Baroukh, N., Ravier, M.A., Loder, M.K., Hill, E.V., Bounacer, A., Scharfmann, R., Rutter, G.A. and Van Obberghen, E. (2007) MicroRNA-124a Regulates Foxa2 Expression and Intracellular Signaling in Pancreatic beta-Cell Lines. J Biol Chem, 282, 19575-88.

139. Esau, C., Davis, S., Murray, S.F., Yu, X.X., Pandey, S.K., Pear, M., Watts, L., Booten, S.L., Graham, M., McKay, R. et al. (2006) miR-122 regulation of lipid metabolism revealed by in vivo antisense targeting. Cell Metab, 3, 87-98.

140. Georges, M., Coppieters, W. and Charlier, C. (2007) Polymorphic miRNA- mediated gene regulation: contribution to phenotypic variation and disease. Curr Opin Genet Dev, 17, 166-176.

141. Griffiths-Jones, S., Grocock, R.J., van Dongen, S., Bateman, A. and Enright, A.J. (2006) miRBase: microRNA sequences, targets and gene nomenclature. Nucleic Acids Res, 34, D140-4.

168

142. Brennecke, J., Stark, A., Russell, R.B. and Cohen, S.M. (2005) Principles of microRNA-target recognition. PLoS Biol, 3, e85.

143. Lewis, B.P., Shih, I.H., Jones-Rhoades, M.W., Bartel, D.P. and Burge, C.B. (2003) Prediction of mammalian microRNA targets. Cell, 115, 787-98.

144. Lai, E.C. (2004) Predicting and validating microRNA targets. Genome Biol, 5, 115.

145. Bentwich, I., Avniel, A., Karov, Y., Aharonov, R., Gilad, S., Barad, O., Barzilai, A., Einat, P., Einav, U., Meiri, E. et al. (2005) Identification of hundreds of conserved and nonconserved human microRNAs. Nat Genet, 37, 766-70.

146. Minaguchi, T., Waite, K.A. and Eng, C. (2006) Nuclear localization of PTEN is regulated by Ca(2+) through a tyrosil phosphorylation-independent conformational modification in major vault protein. Cancer Res, 66, 11677-82.

147. Pezzolesi, M.G., Zbuk, K.M., Waite, K.A. and Eng, C. (2007) Comparative genomic and functional analyses reveal a novel cis-acting PTEN regulatory element as a highly conserved functional E-box motif deleted in Cowden syndrome. Hum Mol Genet, 16, 1058-71.

148. Rodriguez-Viciana, P., Warne, P.H., Dhand, R., Vanhaesebroeck, B., Gout, I., Fry, M.J., Waterfield, M.D. and Downward, J. (1994) Phosphatidylinositol-3-OH kinase as a direct target of Ras. Nature, 370, 527-32.

149. Dent, P., Haser, W., Haystead, T.A., Vincent, L.A., Roberts, T.M. and Sturgill, T.W. (1992) Activation of mitogen-activated protein kinase kinase by v-Raf in NIH 3T3 cells and in vitro. Science, 257, 1404-7.

150. Kyriakis, J.M., App, H., Zhang, X.F., Banerjee, P., Brautigan, D.L., Rapp, U.R. and Avruch, J. (1992) Raf-1 activates MAP kinase-kinase. Nature, 358, 417-21.

151. Meng, F., Henson, R., Lang, M., Wehbe, H., Maheshwari, S., Mendell, J.T., Jiang, J., Schmittgen, T.D. and Patel, T. (2006) Involvement of human micro- RNA in growth and response to chemotherapy in human cholangiocarcinoma cell lines. Gastroenterology, 130, 2113-29.

169

152. Greenblatt, M.S., Bennett, W.P., Hollstein, M. and Harris, C.C. (1994) Mutations in the p53 tumor suppressor gene: clues to cancer etiology and molecular pathogenesis. Cancer Res, 54, 4855-78.

153. Calin, G.A., Liu, C.G., Sevignani, C., Ferracin, M., Felli, N., Dumitru, C.D., Shimizu, M., Cimmino, A., Zupo, S., Dono, M. et al. (2004) MicroRNA profiling reveals distinct signatures in B cell chronic lymphocytic leukemias. Proc Natl Acad Sci U S A, 101, 11755-60.

154. Chan, J.A., Krichevsky, A.M. and Kosik, K.S. (2005) MicroRNA-21 is an antiapoptotic factor in human glioblastoma cells. Cancer Res, 65, 6029-33.

155. Iorio, M.V., Ferracin, M., Liu, C.G., Veronese, A., Spizzo, R., Sabbioni, S., Magri, E., Pedriali, M., Fabbri, M., Campiglio, M. et al. (2005) MicroRNA gene expression deregulation in human breast cancer. Cancer Res, 65, 7065-70.

156. Calin, G.A., Dumitru, C.D., Shimizu, M., Bichi, R., Zupo, S., Noch, E., Aldler, H., Rattan, S., Keating, M., Rai, K. et al. (2002) Frequent deletions and down- regulation of micro- RNA genes miR15 and miR16 at 13q14 in chronic lymphocytic leukemia. Proc Natl Acad Sci U S A, 99, 15524-9.

157. Metzler, M., Wilda, M., Busch, K., Viehmann, S. and Borkhardt, A. (2004) High expression of precursor microRNA-155/BIC RNA in children with Burkitt lymphoma. Genes Chromosomes Cancer, 39, 167-9.

158. Michael, M.Z., SM, O.C., van Holst Pellekaan, N.G., Young, G.P. and James, R.J. (2003) Reduced accumulation of specific microRNAs in colorectal neoplasia. Mol Cancer Res, 1, 882-91.

159. Meng, F., Henson, R., Wehbe-Janek, H., Ghoshal, K., Jacob, S.T. and Patel, T. (2007) MicroRNA-21 regulates expression of the PTEN tumor suppressor gene in human hepatocellular cancer. Gastroenterology, 133, 647-58.

160. Sarquis, M.S., Agrawal, S., Shen, L., Pilarski, R., Zhou, X.P. and Eng, C. (2006) Distinct expression profiles for PTEN transcript and its splice variants in Cowden syndrome and Bannayan-Riley-Ruvalcaba syndrome. Am J Hum Genet, 79, 23- 30.

170

161. Sonkoly, E., Wei, T., Janson, P.C., Saaf, A., Lundeberg, L., Tengvall-Linder, M., Norstedt, G., Alenius, H., Homey, B., Scheynius, A. et al. (2007) MicroRNAs: novel regulators involved in the pathogenesis of Psoriasis? PLoS ONE, 2, e610.

162. Okahara, F., Ikawa, H., Kanaho, Y. and Maehama, T. (2004) Regulation of PTEN phosphorylation and stability by a tumor suppressor candidate protein. J Biol Chem, 279, 45300-3.

163. Agrawal, S. and Eng, C. (2006) Differential expression of novel naturally occurring splice variants of PTEN and their functional consequences in Cowden syndrome and sporadic breast cancer. Hum Mol Genet, 15, 777-87.

164. Eis, P.S., Tam, W., Sun, L., Chadburn, A., Li, Z., Gomez, M.F., Lund, E. and Dahlberg, J.E. (2005) Accumulation of miR-155 and BIC RNA in human B cell lymphomas. Proc Natl Acad Sci U S A, 102, 3627-32.

165. Hayashita, Y., Osada, H., Tatematsu, Y., Yamada, H., Yanagisawa, K., Tomida, S., Yatabe, Y., Kawahara, K., Sekido, Y. and Takahashi, T. (2005) A polycistronic microRNA cluster, miR-17-92, is overexpressed in human lung cancers and enhances cell proliferation. Cancer Res, 65, 9628-32.

166. van den Berg, A., Kroesen, B.J., Kooistra, K., de Jong, D., Briggs, J., Blokzijl, T., Jacobs, S., Kluiver, J., Diepstra, A., Maggio, E. et al. (2003) High expression of B-cell receptor inducible gene BIC in all subtypes of Hodgkin lymphoma. Genes Chromosomes Cancer, 37, 20-8.

167. Weber, F., Teresi, R.E., Broelsch, C.E., Frilling, A. and Eng, C. (2006) A limited set of human MicroRNA is deregulated in follicular thyroid carcinoma. J Clin Endocrinol Metab, 91, 3584-91.

168. Calin, G.A., Sevignani, C., Dumitru, C.D., Hyslop, T., Noch, E., Yendamuri, S., Shimizu, M., Rattan, S., Bullrich, F., Negrini, M. et al. (2004) Human microRNA genes are frequently located at fragile sites and genomic regions involved in cancers. Proc Natl Acad Sci U S A, 101, 2999-3004.

169. Sethupathy, P., Corda, B. and Hatzigeorgiou, A.G. (2006) TarBase: A comprehensive database of experimentally supported animal microRNA targets. Rna, 12, 192-7.

171

170. Eccles, D.M., Russell, S.E., Haites, N.E., Atkinson, R., Bell, D.W., Gruber, L., Hickey, I., Kelly, K., Kitchener, H., Leonard, R. et al. (1992) Early loss of heterozygosity on 17q in ovarian cancer. The Abe Ovarian Cancer Genetics Group. Oncogene, 7, 2069-72.

171. Lin, Y.W., Sheu, J.C., Liu, L.Y., Chen, C.H., Lee, H.S., Huang, G.T., Wang, J.T., Lee, P.H. and Lu, F.J. (1999) Loss of heterozygosity at chromosome 13q in hepatocellular carcinoma: identification of three independent regions. Eur J Cancer, 35, 1730-4.

172. Smith, J.S., Tachibana, I., Pohl, U., Lee, H.K., Thanarajasingam, U., Portier, B.P., Ueki, K., Ramaswamy, S., Billings, S.J., Mohrenweiser, H.W. et al. (2000) A transcript map of the chromosome 19q-arm glioma tumor suppressor region. Genomics, 64, 44-50.

172