The Pennsylvania State University

The Graduate School

Eberly College of Science

INTEGRATIVE ANALYSES OF HIPPO PATHWAY MUTATIONS IN HUMAN

CANCER GENOME

A Dissertation in

Molecular, Cellular, and Integrative Biosciences

Tian Yu

 2014 Tian Yu

Submitted in Partial Fulfillment of the Requirements for the Degree of

Doctor of Philosophy

December 2014

The dissertation of Tian Yu was reviewed and approved* by the following:

Zhi-Chun Lai Professor of Biology Professor of Biochemistry and Molecular Biology Chair, Intercollege Graduate Degree Program in Cell and Developmental Biology Dissertation Advisor Chair of Dissertation Committee

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

Graham H. Thomas Associate Professor of Biology, Biochemistry and Molecular Biology

Francisco J. Diaz Associate Professor of Reproductive Biology

Program Chair: Melissa Rolls Associate Professor of Biochemistry and Molecular Biology Chair of the Molecular, Cellular and Integrative Biosciences Graduate Program

*Signatures are on file in the Graduate School.

Abstract

Cancer genome projects identified somatic mutations in Hippo pathways; however, their functional impacts and association with other mutated genes in the signal transduction network is elusive. In this dissertation, I have analyzed the somatic non- synonymous mutations of Hippo Pathway core components, which regulates organ size and tumorigenesis in Drosophila and mammals. First, I used LATS1/2 as an example to evaluate the potential functional importance of mutated residues of Hippo pathway tumor suppressors in human cancers. The study identified the loss-of-function mutations of

LATS1 and LATS2 alleles from human cancer, which promote tumor development.

Second, the mutation spectra of Hippo Pathway core components (LATS1/2, STK3/4,

TEAD2/4, YAP and WWTR1) have been explored based on mutation pathogenicity tools and literature for their domains and functional partners. Last, my genomic analyses focuses on the concurrent and mutually exclusive gene relationship between Hippo pathway mutations and the other gene mutations in cancer. The significant mutated genes are analyzed in the patient pools with damaging mutations from each of the Hippo pathway components alongside sets of non-damaging controls. Significant concurrent and mutually-exclusive relationships were revealed among the mutations of cancer census genes and the interactome of the Hippo core components. The unbiased screening selected a list of genes with strong associations with Hippo components. These work pushed the limit of previous work, and shed light on the mechanisms of Hippo pathway and the development of hippo pathway related diagnostic biomarker panel.

iii

TABLE OF CONTENTS

List of Tables ...... vii List of Figures ...... viii ACKNOWLEDGEMENTS ...... x Chapter 1 Introduction ...... 1 1.1 Organ size as a maze ...... 1 1.2 Cellular control of organ size ...... 2 1.3 Tumor suppressors and oncogenes in cancer...... 4 1.4 Drosophila melanogaster as Cancer Model ...... 6 1.5 Core Components of Hippo Pathway discovered in Drosophila ...... 7 1.6 Core Hippo Pathway components are conserved in Mouse Model ...... 10 1.7 Other components of Hippo Pathway ...... 12 1.8 Hippo Pathway in Human Tumors ...... 13 1.9 Cancer Genome Project...... 13 1.10 Cancer Genome Mutation Analyses ...... 15 1.11 Gene name and abbreviation ...... 18 Chapter 2 Functional Analyses of human LATS1 and LATS2 genes mutations ...... 20 2.1 Introduction ...... 20 2.2 Material and Methods ...... 21 2.2.1 Protein sequence alignment ...... 21 2.2.2 DNA constructs and site-directed mutagenesis ...... 21 2.2.3 Cell culture and luciferase assays ...... 21 2.2.4 Western blot analysis ...... 22 2.2.5 Transgene microinjection and site-specific integration ...... 22 2.2.6 Drosophila genetics and wing analysis ...... 23 2.3 Results ...... 25 2.3.1 Analysis of mutated site conservation and mutation pathogenicity ...... 25 2.3.2 Analysis of expression and kinase activity of LATS1 and LATS2 mutations ... 28

2.3.3 Functional analysis of LATS1 and LATS2 in tissue growth and organ size control through an in vivo over-expression assay ...... 29 2.3.4 Functional analysis of LATS1 and LATS2 through an in vivo phenotypic rescue assay ...... 31 2.3.5 Functional analysis of LATS1 and LATS2 variants for their ability to inhibit YAP/Yki activity ...... 35 2.4 Discussion and Summary ...... 36 Chapter 3 Somatic Mutation Spectrum of Hippo Core Components in Human Cancer Genome ...... 39 3.1 Introductions ...... 39 3.2 Material and Methods ...... 40 3.2.1 Information Sources ...... 40 3.2.2 Evaluation Tools for Missense Amino Acid Mutations ...... 41 3.2.3 Domain and Binding Partner Analysis Tools ...... 42 3.3 Results ...... 42 3.3.1 Somatic mutations of LATS1 and LATS2 genes ...... 42 3.3.2 Somatic mutations of STK3 and STK4 genes ...... 58 3.3.3 Somatic mutations of YAP and WWTR1 genes ...... 65 3.3.4 Somatic mutations of TEAD2 and TEAD4 genes ...... 78 3.3.5 Somatic mutation frequency and damage degree in all hippo core 8 components ...... 85 3.4 Discussion ...... 89 Chapter 4 Hippo Core Components Somatic Mutation and Copy Number Variant in Cancer Genome ...... 91 4.1 Introduction ...... 91 4.2 Materials and Methods ...... 93 4.2.1 Data source ...... 93 4.2.2 Extract Significant Mutated Genes by Mutation Significance (MutSig) ...... 94 4.2.3 Mutation Relation Test by Mutational Significance in Cancer (Genome MuSiC) ...... 94 4.2.4 Visualization by Cytoscape ...... 95 4.3 Results ...... 95

4.3.1 Identification of the most correlated genes with Hippo Core components across their interactome ...... 95 4.3.2 Identification of the significant cancer census genes correlated with Hippo Pathway core components ...... 99 4.3.3 Unbiased screening of genes with most significant correlation with Hippo core components ...... 103 4.3.4 Significantly mutated genes identified in patient genome carrying with the mutations of Hippo Pathway Core Tumor Suppressors...... 104 4.3.5 Significantly mutated genes identified in patient genome carrying with the mutations of Hippo Pathway Core Oncogenes...... 108 4.4 Discussion ...... 109 Chapter 5 Conclusions and Future Work ...... 115 5.1 Conclusions ...... 115 5.2 Future directions ...... 116 5.3 Possible experiments to test hypotheses generated in in silico work ...... 117 5.4 Multi-omics pathway analyses adding DNA copy number, RNA expression alterations...... 118 5.5 Hippo Pathway in tumor evolution and social cell biology ...... 119 References ...... 120

List of Tables

TABLE 1. 1 HIPPO PATHWAY GENE ALTERATION PHENOTYPE IN MAMMALS...... 11

TABLE 2. 1 MUTATION PATHOGENICITY AND CONSERVATION WITH DROSOPHILA HOMOLOGUE ...... 26

TABLE 3. 1 THE TISSUE TYPES WITH THE MOST OR LEAST FREQUENTLY MUTATED HIPPO CORE OMPONENTS...... 87 TABLE 3. 2 RECURRENT MISSENSE (RM) AND NONSENSE OR FRAMESHIFT MUTATIONS (NF) OF HIPPO PATHWAY CORE COMPONENTS INCLUDING SNP...... 88

vii

List of Figures

FIGURE 1. 1 HIPPO SIGNALING PATHWAY CONSERVED IN DROSOPHILA AND MAMMALS. ... 9

FIGURE 2. 1 EVOLUTIONARY CONSERVATION OF AMINO ACIDS THAT ARE MUTATED IN HUMAN LATS1 AND LATS2...... 27 FIGURE 2. 2 WESTERN BLOT ANALYSIS OF HUMAN LATS1 AND LATS2 PROTEINS IN HEK293T CELLS ...... 29 FIGURE 2. 3 IN VIVO ANALYSIS OF HLATS1/2 AND VARIANTS FOR THEIR ACTIVITY TO INHIBIT TISSUE GROWTH IN DROSOPHILA...... 30 FIGURE 2. 4 IN VIVO ANALYSIS OF HLATS1/2 AND MUTANTS FOR THEIR ABILITY TO RESCUE THE LETHALITY INDUCED BY LOSS OF THE ENDOGENOUS WTS GENE FUNCTION...... 32 FIGURE 2. 5 IN VIVO ANALYSIS OF HLATS1/2 AND VARIANTS FOR THEIR ACTIVITY TO CONTROL ORGAN SIZE IN THE ABSENCE OF THE ENDOGENOUS WTS GENE FUNCTION. .. 34 FIGURE 2. 6 ACTIVITIES OF HLATS1/2 AND VARIANTS IN REGULATING YAP/YKI TRANSCRIPTIONAL CO-ACTIVATOR ACTIVITY IN BOTH HUMAN AND DROSOPHILA CELLS...... 36

FIGURE 3. 1 MUTATION SPECTRA OF LATS1 AND LATS2...... 44 FIGURE 3. 2 MUTATIONS IN LATS CONSERVED DOMAIN 1(LCD1) AND PUTATIVE UBIQUITIN ASSOCIATED DOMAIN(UBA) ...... 48 FIGURE 3. 3 MUTATIONS IN LATS CONSERVED DOMAIN 2(LCD2) ...... 50 FIGURE 3. 4 MUTATIONS IN N-TERMINAL REGULATORY DOMAIN ...... 53 FIGURE 3. 5 MUTATIONS IN LATS1/2 KINASE DOMAINS...... 55 FIGURE 3. 6 MUTATIONS IN LATS1/2 C TERMINAL DOMAIN (CTD) AND HYDROPHOBIC MOTIF (HM) ...... 58 FIGURE 3. 7 MUTATION SPECTRA OF STK3 AND/ STK4 DOMAIN...... 60 FIGURE 3. 8 MUTATIONS IN STK3/4 KINASE DOMAINS ...... 62 FIGURE 3. 9 MUTATIONS IN STK3/4 C-TERMINAL DOMAINS ...... 65 FIGURE 3. 10 MUTATION SPECTRA OF YAP AND TAZ...... 68 FIGURE 3. 11 MUTATIONS IN YAP1 N-TERMINAL DOMAINS...... 69 FIGURE 3. 12 A MODEL OF YAP1/WWTR1 REGULATION BY THE HIPPO PATHWAY AND CK1Ε AND Β-TRCP...... 72 FIGURE 3. 13 MUTATIONS AROUND YAP1/WWTR1 WW DOMAINS ...... 74 FIGURE 3. 14 MUTATIONS IN YAP1/WWTR1 C-TERMINAL DOMAINS...... 76 FIGURE 3. 15 MUTATION SPECTRA OF TEAD2 AND TEAD4...... 81 FIGURE 3. 16 MUTATIONS IN TEAD2/4 TRANSCRIPTIONAL ENHANCER ACTIVATOR DOMAIN ...... 83 FIGURE 3. 17 MUTATIONS IN TEAD2/4 TRANSCRIPTIONAL ACTIVATION DOMAINS...... 83

viii

FIGURE 4. 1 MUTATION RELATIONSHIP BETWEEN HIPPO PATHWAY CORE COMPONENTS AND THEIR INTERACTOME ...... 96 FIGURE 4. 2 MUTATION RELATIONSHIPS BETWEEN HIPPO PATHWAY CORE COMPONENTS AND CANCER COSMIC CENSUS GENES ...... 100 FIGURE 4. 3 GENOMIC UNBIASED SCREENING AND MUTATION RELATION PARTNER WITH HIPPO CORE COMPONENTS ...... 105 FIGURE 4. 4 SIGNIFICANTLY MUTATED GENES IN CANCER PATIENTS CARRYING HIPPO CORE 4 TUMOR SUPPRESSOR MUTATIONS ...... 107 FIGURE 4. 5 SIGNIFICANTLY MUTATED GENES IN CANCER PATIENTS CARRYING HIPPO CORE 4 ONCOGENE MUTATIONS ...... 109 FIGURE 4. 6 INTRA-PATHWAY MUTATION RELATIONSHIP OF HIPPO PATHWAY CORE COMPONENTS WITH OTHER COMPONENTS ...... 110 FIGURE 4. 7 SCHEMATIC REPRESENTATION OF MUTATION RELATION BETWEEN HIPPO PATHWAY AND WNT PATHWAY COMPONENTS...... 113

ACKNOWLEDGEMENTS

First of all, my sincere gratitude goes to my advisor Dr. Zhi-Chun Lai, who has offered invaluable help over the past six years. He often encourages me to follow my interests, utilize my strengths, and think independently.

This work cannot be finished with the help from my undergraduate colleagues including John Bachman, Yu-Chun Chang, Nicholas Anzalone, Samar Almarzooqi. I would also like to thank Brenita Jenkins, Priyanka Solanki, and Humza Khalid for participating in my project. Thanks also to my graduate colleagues Lilun Ho, Xin Ye,

Yifan Zhang, Yaoting Deng, and Yurika Matsui for their help and comments.

I am grateful to Dr. Hong Ma for enrolling me into Cell and Developmental

Biology program. I would not have made a smooth transition to the study in Penn State without his advice and encourage. His students also provided lots of help in everyday life in State College and study in the graduate school.

I would also like to show my sincere gratitude to my dissertation committee members: Dr. Robert Paulson, Dr. Francisco Diaz, and Dr. Graham Thomas, for their support and insightful comments in the committee meetings and individual talks.

Collective and individual thanks go to the previous and current members in the first floor of Life Sciences Building for their encouragement, scientific suggestions and daily warm greetings.

I would also like to thank my family and friends for their consistent support, especially my wife Dr. Danhong Chen. To them, I dedicate this dissertation.

Chapter 1 Introduction

1.1 Organ size as a maze

The topic of organ size attracted many scholars from different disciplines. Sizes of organs and species, as the most obvious qualitative index, were discussed in the books of Greek philosophers, such as Aristotle’s History of Animals. Naturalists, like Charles

Darwin, were fascinated by the morphological diversity of organ size in different environment. For physicians, this question arises from abnormal growth, such as cancer and cardiac hypertrophy. Developmental biologists are interested in the accurate development and maintenance of tissues and organs in multi-cellular organisms. In the scope of cell biologists, this is a question of cell proliferation/division, growth, survival, differentiation and communication. For molecular biologists, the sophisticated molecular interaction and signal transduction provide the material, energy and information foundation for the control of organ size.

Along the way of exploration, the discovery in each discipline enlightens and interacts with other roads. Aristotle’s lantern, mouth structure of sea urchin, evoked evolutionary ecologists to consider the phenotypic flexibility (Piersma and Drent, 2003).

The beak of Darwin’s Fitch provides an excellent example for Bmp4-mediated morphological variation (Abzhanov et al., 2004). Both physicians and developmental biologists agree that the normal tissue growth and pattern organization lead to normal development of tissue and organ, while abnormal tissue growth causes hyperplasia, metastasis, and organ hypertrophy. Utilizing molecular genetics and model organism from fruit fly to mice, researchers identified a conserved core signaling pathways

regulating organ size, Hippo Signaling Pathway. The system-wide screening advances in genomics and proteomics convert the linear pathways into more nodes and edges of signaling network.

Most studies found that cell size and cell proliferation are coordinated for determining normal growth pattern and organ size. However, adult C. elegans body grows with the increase in cell size without proliferation (Knight et al 2002). This uncoupling is also found in the poster-mitotic neurons in mammals. During the early embryonic stage of many species, uncoupling of cell growth and division are observed through frequent cell division into small cells without growth (Goranov and Amon 2010).

Moreover, the complicated genome alteration in human cancer challenges the effort of researchers and physicians to improve human health and medicine. The detailed mechanism for the phenomenon like these replies on the further studies for the underlying signaling pathway.

With the achievement for previous understanding and the mysteriousness in this maze, I further explore how the organ size was affected by the dis-regulation of a single signaling pathway utilizing multi-discipline approaches during this inter-disciplinary PhD program training.

1.2 Cellular control of organ size

The size of an organ is determined by cell size, total cell number, and intercellular space. Cell size is determined by cell growth/mass accumulation and degradation. Cell accumulates molecules by self-synthesis and nutrient uptake, while lose them by

metabolism and secretion. Macromolecular syntheses involve the lipid and glycogen synthesis, DNA replication, and mRNA translation and protein modification. Mammalian

Target of Rapamycin (mTOR) is a central player in regulating the macromolecular syntheses in response to the integration of major cues for cell growth: stress, energy status, oxygen, nutrient and growth factors. The insulin-PI3K-Akt axis and EGF-RAS-

MAPK axle both regulates mTORC1, while former one also activates mTORC2

(Laplante and Sabatini, 2012).

Cell number depends on the balance between cell proliferation/division and cell death. Cell division increases cell number. The level of mitogenic signaling triggers the mitosis and uni-directional cell cycle. Programmed cell death (apoptosis) and cellular aging (senescence) cause the reduction of cell numbers. Growth-regulatory mechanism tightly coordinates and regulates cell proliferation and cell death to ensure that a characteristic amount of cell is generated and maintained. The accuracy of this coordination is essential for the maintenance of tissues and organs. Many signaling pathways form an ordered network to mediate these cell fate decisions, including Notch,

BMP, Wnt pathway etc. They form ordered networks to make context-dependent response to various signaling cues. The aberrant regulation leads to pathological conditions such as neoplasm (Danial and Korsmeyer, 2004; Murray 2004; Sherr, 2004).

The delineation of signaling pathways is critical to understanding the intricacy of these conditions. The cross talk between Hippo pathway with BMP/TGF-β, mTOR, GPCR, and

Wnt pathway demonstrated the importance of signal coordination and growth regulation

(Varelas, 2014; Bernascone and Martin-Belmonte, 2014), and will be discussed later. I

focus on the deregulation of Hippo Pathway and mis-coordination with other signaling pathway involved in organ size control.

1.3 Tumor suppressors and oncogenes in cancer.

The direct outcome for loss of growth control is neoplasm. Normal cells accumulate genetic alterations caused from exogenous environment factors (eg. chemicals from smoking, X-ray, tumor virus) and endogenous factors (eg. Reactive

Oxygen Species, DNA replication error). Most of the genetic alterations acquired by the benign neoplasm are still under growth control network, while some develop locally as in situ neoplasm, others further become malignant neoplasm--cancer. As discussed previously, the tissue growth required highly regulated coordination for cell division and size increase. It is hard to imagine that there is no genetic compromise or reorganization for this network in cancer. Cancer became an aberrant system that harbors various genetic alterations involving many genes with different functions.

These cancer-related genes are categorized as oncogene or tumor suppressors, based on whether the genes potentially cause or suppress cancer. The early discovery of these cancer-related genes largely depended on the genetic mutation of these genes in cancer cell lines derived from patients or animals, and experimental model organisms, such as Drosophila. Like other genetic studies, these alterations can usually be classified into loss of function and gain of function, depending on the phenotypical and functional comparison between the alternated gene and wild type genes. The gain of function mutation increases the activity, translation or transcription, or accumulation of a gene product, or even results in new function, while the loss of function mutations decrease the

activity, translation or transcription, or even destroy a gene product. Gain of function of oncogene, such as RAS (Rat Sarcoma gene), increases the selective growth advantage of the cells harbor those changes, and promote tumor growth. This genetic alteration in one of the two alleles in diploid organism usually is dominant to the function of proto- oncogene, wild type oncogene. Wild type oncogenes promote cell division, tissue growth, and/or inhibit apoptosis in most of stage during normal development (Ashworth et al

2011).

For most tumor suppressor genes, such as RB (Retinoblastoma gene), gain of function alteration usually suppress the tumor growth, while loss of function of tumor suppressor can also promote tumor growth. The loss of function of one tumor suppressor allele is usually recessive to the remaining wild type allele of the same gene, which still protect the genetically heterozygous cells from tumor formation. Additional loss of function in the remaining alleles, as the second hit, causes the loss of heterozygosity, and the complete loss of function for both alleles. This is the Two-Hit Hypothesis for most tumor suppressor genes in cancer. However, exception as the majority mutations of

TP53, act oppositely to the normal tumor suppressor function (Muller and Vousden

2013). This dominant-negative effect cause a dosage-dependent haploinsufficiency, which means a single copy of gene product is not sufficient to cover the dosage of two alleles in the diploid cells. Tumor suppressor genes inhibit cell division, tissue growth, and/or promote apoptosis in most of stage during normal development. Usually they are required to maintain growth homeostasis and balance the growth effect of oncogene during developmental stages. Tumor suppressor genes can be classified into three groups.

Gatekeeper genes include all genes that directly inhibit cell growth (Kinzler and

Vogelstein, 1998). Caretaker genes indirectly regulate tumor growth by ensuring genomic stability or repairing DNA damage. Landscaper genes are involved in regulating the microenvironment where tumor cells grow (Macleod, 2000).

Cancer is the second leading cause of death in the United States. It was projected that there would be more than 1.6 million new cases of cancer, and around 0.6 million people will die from cancer in 2014 (Siegel et al., 2014). Understanding the molecular lesions associated with human tumor development is critical to improving the diagnosis and treatment of cancer, and increasing cancer survival rate. However, it is still not fully uncovered how molecular signal transduction pathways are related to tumor development. Better knowledge in these pathways will provide potential diagnostic markers and therapeutic targets for treating cancer. In personalized medicine for cancer patients, identification of an individual’s cancer-related molecular mutations will help improve treatment outcomes (Barron and Kagey, 2014).

1.4 Drosophila melanogaster as Cancer Model

Drosophila melanogaster as an instrumental model organism in genetics, due to its short reproductive cycle and potent fertility. With balancer chromosome and abundant genetic information, various mutant phenotype generation and genetic screens have been performed to investigate the development and tumorigenesis, such as X-ray or mutagen screens, transgenic RNAi screens. Not only utilized the natural P-element transposase in

Drosophila, researchers also implement powerful genetic tools from other species into

Drosophila. UAS/Gal4 binary expression system allows the tissue-specific expression of exogenous genes. Also originally from yeast, FLP-FRT (flippase-flippase recognition

targets) mitotic recombinase system generates genetic mosaic animals (Xu and Rubin,

1993). This largely helps the identification and screening for tumor suppressor genes, because homozygous loss of function of tumor suppressor gene usually cause lethality.

Newly developed phi-C31 integrase system from bacteriophage provide the fixed docking site for exogenous genes (Bischof et al 2007). Drosophila also has a high degree of evolutionary conservations with human signal pathways involving in the various aspect of tumorigenesis, such as apoptosis and cell cycle regulation. Drosophila epithelial tissues in wing and eye also resemble the epithelia in human, where most carcinomas are derived from.

1.5 Core Components of Hippo Pathway discovered in Drosophila

Cellular signaling pathways provide instructions to influence a variety of cellular activities that include cell proliferation, differentiation, and cell death.

The first Hpo pathway component, Warts (Wts)/Drosophila large tumor suppressor

(dLats), was discovered as a tumor suppressor by genetic mosaic screens in Drosophila

(Justin et al. 1995; Xu et al. 1995). Later, genetic analyses identified the WW domain- containing protein Salvador (Sav) (Tapon et al. 2002; Kango-Singh et al. 2002), the

Ste20-like protein kinase Hippo (Hpo) (Harvey et al. 2003; Jia et al. 2003; Pantalacci et al. 2003; UDAN et al. 2003; WU et al. 2003) and a kinase co-activator Mob as tumor suppressor (Mats) (Lai et al. 2005). Loss-of-function mutant clones for any of these four genes display robust tissue overgrowth phenotype due to increased cell proliferation and diminished cell death. Biochemical studies found that these four components form a kinase cascade in which the Hpo kinase associates with Sav to phosphorylate and activate

the Wts kinase-Mats complex (Wu et al. 2003; Wei et al. 2007). Yki, which was identified as a Wts-binding protein in a yeast two-hybrid screen, has been shown to be directly phosphorylated and consequently inactivated by the Wts/Mats kinase complex

(Huang et al. 2005; Wei et al. 2007). Yki is a transcription co-activator for transcription factors such as Scalloped (Sd) to target cycE, diap1 and dmyc genes for promoting cell proliferation and blocking apoptosis (recently reviewed by Pan 2010; Zhao et al. 2011;

Staley and Irvine 2012).

Figure 1. 1 Hippo Signaling Pathway Conserved in Drosophila and Mammals.

The conserved components of Hippo Signaling Pathway were plotted into corresponding region of the cells, and colored with same colors between Drosophila and Mammals (revised from Badouel and McNeill 2011; Johnson and Halder 2013 ). The gene abbreviations are included in the section 1.11.

1.6 Core Hippo Pathway components are conserved in Mouse Model

This core kinase cascade is conserved in mammals such as mouse and human, which includes two Hpo homologs (MST1/STK4 and MST2/STK3), one Sav homolog

(WW45 or SAV1), two Wts homologs (LATS1 and LATS2), and two Mats homologs

(MATS1 and MATS2, or MOBKL1A and MOBKL1B) (Chan et al. 2005; Callus et al.

2006; Praskova et al. 2008; Ye et al. 2009). LATS1/2 phosphorylates and inactivates the mammalian Yki homologs YAP (Dong et al. 2007; Zhao et al. 2007; Hao et al. 2008) and TAZ (WWTR1) (Lei et al. 2008) (Figure 1.1).

Functional conservation is supported by the fact that human YAP, LATS1,

MST2, and MATS1 can functionally rescue the corresponding Drosophila mutants

(Huang et al. 2005; Tao et al. 1999; Wu et al. 2003; Lai et al. 2005). YAP as a transcription co-activator interacts with TEAD family transcription factors (Sd homologs in mammals) to target CTGF and Cyr61 genes (Goulev et al. 2008; Ota et al. 2008; Zhao et al. 2008). The tumor suppressor function of this kinase cascade has been further demonstrated by knock-out studies in mice (Table 1.1). The Lats1 knock-out mice develop soft tissue sarcomas or ovarian stromal cell tumors and are highly sensitive to carcinogens, while Lats2 mutant mice are embryonic lethal (St. John et al. 1999;

McPherson et al. 2004). Knockout mice of WW45 display hyperplasia and differentiation defects in embryonic epithelial (Lee et al. 2008).

Table 1. 1 Hippo Pathway gene alteration phenotype in Mammals.

Study Phenotype LATS1/2 Knock-out mice LATS1 null animals develop soft tissue sarcoma and ovarian tumors Knock-out MEFs Loss of LATS2 leads to loss of contact inhibition Overexpression in cells LATS1 overexpression suppressed tumorigenesis of cancer cells and YAP-driven cellular transformation and EMT RNAi in cells LATS2 expression suppressed tumorigenesis of transformed cells LATS1 depletion leads to anchorage-independent growth of cells Human cancer samplesa LATS1/2 mRNA is downregulated in several human cancersb

STK3/4 STk3/4 double hepatocelular carcinoma, cholangiocarcinoma, colonic adenoma, T cell acute lymphoblastic leukaemia conditional null Overexpression in cells MST2/LATS1 cooperate in suppressing YAP-driven transformation Human cancer samples MST1/2 mRNA is downregulated in sarcomas Loss of cytoplasmic MST1 protein in colorectal cancer

YAP1 Transgenic mice Hepatocelular carcinoma, gastrointestinal dysplesia, pancreatic ductal metaplasia, squamous cell carcinoma Overexpression in cells Overexpression in NIH/3T3 cells leads to loss of contact inhibition Overexpression in MCF10A cells results in anchorage-independent growth and EMT RNAi in cells Depletion of YAP decreses turnorigenicity of cancer cells Human cancer samples YAP overexpression is detected in several cancer types

WWTR1 Overexpression in cells Overexpression in MCF10A cells results in loss of contact inhibition and anchorage-independent growth and EMT transition RNAi in cells Knockdown of TAZ suppresses turmorigenesis of breast cancer cells Human cancer samples TAZ overexpression is observed in breast cancer tumors are homozygous for Nf2 in the hemizygote background. ǂ One Mob1 gene is null, while the other is heterozygous. Tumor lose the last Mob1 allele (Harvey et al., 2013).

Mouse genetic experiments demonstrated that Mst1 and Mst2 are important for organ size control and tumor suppression in tissues such as liver (Zhou et al. 2009; Lu et al.

2010; Song et al. 2010). Recently, loss-of-function analysis has also shown that mouse

Mob1a/Mob1b are essential for normal development and tumor suppression by regulating

Hpo pathway activity (Nishio et al. 2012).

1.7 Other components of Hippo Pathway

The intensive studies over the recent ten years discovered over twenty regulators that interact with core components of Hippo Pathway at different levels (Figure 1.1). Most of them are also conserved between Drosophila and mammals (Table 1.2). In mammals, core kinases regulators, actin cytoskeleton composition and regulators, and the cellular junction complex including Crumbs homolog (CRB) complex, are the four major upstream signaling regulate the Hippo Pathway.

The first brand of Hippo Pathway upstream regulators consists of STK3/4 kinases regulators and LATS1/2 kinases regulators. The former one includes NF2/FRMD6/WWC protein complex, TAO kinases, and RASSF protein family members. The later one includes HSP90, AJUBA and NEDD4. Secondly, CRB complex components MPDZ

(Multiple PDZ domain protein 1) recruited angiomotin (AMOT) family protein to activate LATS1 to promote YAP phsosphorylation, or to inhibit YAP directly. Thirdly,

The actin architecture are also regulates LATS and YAP homologues by GPCR pathway.

Fourly, catenins, ZO-1/2, and AJUBA directly or indirectly regulate the YAP1 activity close to the tight junction and cadherin junction (Johnson and Halder 2013)

1.8 Hippo Pathway in Human Tumors

The Hippo pathway components have been found to be inactivated in humans through promoter hypermethylation, allelic deletions. Frequent promoter hypermethylation of MST1/2 (Seidel et al. 2007, Steinmann et al. 2009), as well as decreased expression of MST1/2 are found in colorectal and prostate cancers (Cinar et al.

2007; MINOO et al. 2007). hMATS1ΔS6/7 deletion and mMob1Δ6-216 insertion was detected in human skin melanoma and mouse mammary gland carcinoma, respectively. Both mutations do not have detectable products (Lai et al. 2005). Mob1 mRNA and protein down-regulation have been found in human colorectal and non-small-cell lung cancer samples (Koska et al. 2007; Sasaki et al. 2007). Loss-of-function mutation of WW45 has been observed in several human cancer cell lines (Tapon et al. 2002). NF2, named by the tumor condition neurofibromatosis that it frequently mutated in, are bona fide tumor suppressor genes. In addition, malignant mesothelioma harbors loss function mutations for Merlin, LATS2 or SAV1 (Johnson and Halder 2013). A current model is that loss of

Hippo signaling contributes to human cancer development (Harvey et al. 2013).

1.9 Cancer Genome Project

Cancer is a disease of the genome. The understanding of oncogenes and tumor suppressor genes have been redefined intensively, since the establishment if these two concepts in mid-1980s (Garraway and Lander, 2013). In addition, cancer susceptibility genes like BRCA1/2 (BReast CAncer susceptibility protein 1/2) add more complexity to the concepts of these two cancer-causing genes. The complete sequence of human genome

would be critical for decipher these mysteries, a notion proposed since 1986 by Dulbecco and others, supported in 1990 through the launch of Human Genome Project, and ignited in 2003 by the complete of human genome. Along the fast development of next-generation sequencing technology, Cancer Genome Project (CGP) at the Sanger Institute in the United

Kingdom and The Cancer Genome Altas (TCGA) in the United States are firm steps toward the comprehensive and systematic studies of cancer genomes (Chin et al., 2011). Based on the results of these projects, cBioPortal (Gao et al., 2013) and COSMIC (Forbes et al., 2010) provides convenient interface to access cancer genome information. My work also extracted the information from these databases.

The next-generation sequencing technologies mainly replied on the massively parallel sequence generation after the array or bead-based amplification of individual DNA.

The raw data generated was assembled with a genome reference to map onto chromosomal positions. Then the data will be normalized by correction for background noise. The sequence variant will be called and defined the mutation types and location. The somatic status will be confirmed by comparing the genome sequence of normal tissues and tumor tissues from the same patient. Those interpreted data can be downloaded in TCGA data portal. The common short variants can be annotated in dbSNP (Chin et al 2012).

By the time cancer was diagnostics by morphology and sequenced, there are numerous genetic alteration harbored in the cancer cells. These mutations are categorized into driver and passenger mutations, based on whether the alteration can confer a selective growth advantage or not. Driver mutations are oncogenic to initiate or promote tumorigenesis, so positively selected during tumorigenesis. In contrast, passenger mutations arise from incidental alterations and genome instability, and usually contribute

less to tumorigenesis. Although inevitably containing passenger mutations, cancer genes are mainly defined by the driver mutations (Vogelstein et al 2013).

Genetic alterations, in general called mutations, are DNA base mutation (point mutation, insertion, and deletion), chromosomal copy number alteration (amplification and deletion), structural rearrangement, and epigenetic modification, as well as mRNA transcript expression level up-regulation and down-regulation. The mutations my dissertaion focused are DNA base mutations.

1.10 Cancer Genome Mutation Analyses

A key question for cancer genome is to identify these somatic mutations and their consequences for cancer development. Two major strategies to answer this are to investigate functional impact in molecular level and statistical frequency in genome and population levels. Most pathogenicity evaluation methods defined variant deleteriousness in term of evolutionary conservation, protein structure and biochemical on both quantitative and discrete scales. Evolution conservation provides an excellent way to measure deleteriousness of mutation. During the evolution, natural selection continually remove the deleterious mutation, and keep the conserved position changed more slowly than the neutral mutation rate. So the comparative sequence analyses indirectly predict the deleteriousness of the mutations. For example, the mutation of more conserved amino acid residue is more likely to be deleterious, if the changed residue was not found in the other protein homologues of other specials. Though functional divergence between human genes and other species exists in exception cases, most prediction is efficient and informative

(Cooper and Shendure 2011). The amino acid type and sequence, domain and protein

structure and folding provide predictive features for various bioinformatics methods. Prior knowledge of mutations were also used to train the bioinformatics tools to infer the pathogenicity of new mutations based on machine learning methods (Izarzugaza et al 2012).

To distinguish the functional impact of driver mutations from a large number of missense passenger mutations in cancer genome, various in silico algorithms have been developed, based on the evaluation of evolutionary conservation, protein structure and biochemical features of mutated amino acid residues. For the statistical significance of mutations, the significantly mutated genes could be selected based on their significant mutation rate than the background mutation rate, accounting for sequence context and gene size. For example, the random background mutations will result in expected amount of mutations in a gene, which will be tested with the statistical null hypothesis by the amount observed mutations. One of the important bioinformatics methods, MutSig, effectively identified the true driver mutations while taking into account the probability of background mutations and possible false positives due to other factors, such as gene size and chromosome accessibility (Lawrence et al 2013). Utilizing these and many other methods, many cancer genome studies spend tremendous efforts to identify the key driver mutations and related driver cancer genes from a large number of sequenced patient genome.

However, not all rare driver mutations will be captured by these efforts, contributing to the comprehensive understanding for these genomes. From a single pathway point of view, I want to address that whether Hippo Pathway core components harbor any driver mutations predicted to be deleterious, which may implicate the role of Hippo Pathway in human cancer genome.

Concurrency as a positive correlation, suggests that there may be synergistic or redundant function between two genes or their associated pathways in terms of tumorigenesis (Dees et al 2012). Some of them may be synthetic viable, leading to the survival of tumor cells, although one single gene inactivation may lead to cell apoptosis or senescence. There are two possible mechanisms to explain mutually exclusive relationships between mutated genes. First, the sufficiently positive selection of one mutation in one gene reduces the accumulation of additional mutations in another gene during the tumor evolution.

Second, the co-mutation of these two genes may be synthetic lethal for the mutated cells, so those cells with both genes inactivated are selected out through tumor evolution. By the time of diagnostics, none of the cells will harbor mutations for both genes. Synthetic lethal genetic interactions guide the therapeutically strategy based on the mutation status, such as PARP inhibitor for BRCA mutant tumors (Ashworth et al 2011). The mutation relationship with Hippo Pathway in cancer genome will be an interesting perspective to explore the network around Hippo Pathway and implicate the possible therapeutical strategy.

1.11 Gene name and abbreviation

Human proteins Human protein full name D. Domain composition melanogaster protein Core components STK4/3=MST1/2 Serine/threonine-protein kinase Hpo Kinase domain, 4/3=Mammalian STE20-like SARAH domain protein kinase 1/2 SAV1=WW45 Salvador homolog 1 Sav FERM domain-binding motif, two WW domains, SARAH domain LATS1, LATS2 large tumour suppressor homolog Wts Kinase domain 1/2 MOB1A, MOB1B MOB kinase activator 1A/1B Mats MOB domain YAP, WWTR1=TAZ Yes-associated protein; Yki (yorkie) Two WW domains, 14- WW domain-containing 3-3 binding motif, transcription regulator protein 1; TEAD-binding motif, transcriptional co-activator with PDZ-binding motif PDZ-binding motif TEAD1–TEAD4 TEA domain-containing Sd TEAD DNA-binding sequence-specific transcription (scalloped) domain, vestigial factor 1- 4 binding domain Associate pathway members CRB1/2/3 Crumbs homolog 1/2/3 Crb EGF domains, four laminin AG domains, transmembrane domain MPDZ, PATJ, MUPP1 PALS1-associated tight junction Patj Ribosomal protein L27, protein; eight PDZ domains multiple PDZ domain protein 1 MPP5=PALS1 membrane protein, palmitoylated Sdt Ribosomal protein L27, 5 PDZ domain, SH3 domain, GUK domain AMOT, AMOTL1, angiomotin; angiomotin-like - Coiled coil domain, AMOTL2 protein PDZ binding motif NF2 neurofibromin 2 Merline FERM domain KIBRA kidney and brain protein Kibra Two WW domains, C2 domain FRMD6=EX1=WILLIN Expanded Ex FERM domain TAO1–TAO3 thousand and one amino acid Tao Kinase domain protein MARK1–MARK4 MAP/microtubule affinity- Par-1 Kinase domain regulating kinase AJUBA, LIMD1, WTIP LIM domain-containing protein Jub Three LIM domains 1; Wilms' tumour 1 interacting protein ZYX Zyxin; lipoma-preferred partner; Zyx Three LIM domains thyroid hormone receptor interactor 6

RASSF1–RASSF6 RAS association domain- Rassf RAS association containing family protein domain, SARAH domain PP2A protein phosphatase 2A STRIPAK– Phosphatase domain PP2A complex (dSTRIPAK) SCRIB Scribble Scrib 16 LRR domains, 4 PDZ domains LGL1, LGL2 lethal giant larvae Lgl Four WD40 domains

DLG1–DLG4 discs large Dlg Three PDZ domains, SH3 domain, GUK domain PTPN14 protein tyrosine phosphatase, Pez FERM domain, non-receptor type 14 phosphatase domain CK1 casein kinase 1 Dco Kinase domain FAT1/2/3/4 Protocadherin Fat 1/2/3/4; Fat Cadherin, EGF-like, Cadherin Laminin G-like domains E-cadherin E-cadherin Five cadherin domains, transmembrane domain α-catenin α-catenin VH1–VH3 domains DCHS1/2 Protocadherin-16/23 Ds Cadherin domains FJX1 Four-jointed box protein 1 Fj Proline-rich domain β-TRCP β-transducin repeat-containing E3 Slimb F-box domain, β-TRCP ubiquitin protein ligase domain, WD40 domain HIPK homeodomain-interacting protein Hipk Kinase domain kinase MASK1, MASK2 multiple ankyrin repeats single Mask Two ankyrin domains, KH domain-containing protein KH domain WBP2 WW domain-binding protein 2 Wbp2 GRAM domain

NEDD4 E3 ubiquitin-protein ligase Nedd4 WW domains, HECT NEDD4 domain ITCH E3 ubiquitin-protein ligase Itchy Not WW domains, C2, homolog Available HECT domain WWP1 NEDD4-like E3 ubiquitin-protein Not WW domains, C2, ligase WWP1 Available HECT domain Note: This table was adapted from Harvey et al. (2013) and Johnson and Halder (2013).

Chapter 2 Functional Analyses of human LATS1 and LATS2 genes mutations

This Chapter is mostly derived from my publication: Yu, T., Bachman, J., & Lai,

Z. C. (2013). Evidence for a Tumor Suppressor Role for the Large Tumor Suppressor

Genes LATS1 and LATS2 in Human Cancer. Genetics, 195(3), 1193-1196. All the results shown in this chapter are done by me.

2.1 Introduction

Genetic analysis of Lats/Wts family genes using Drosophila and mice models has revealed their role as a negative growth regulator and tumor suppressor (Justice et al.

1995; Xu et al. 1995; St John et al. 1999). The fact that LATS1 can functionally replace

Wts in Drosophila strongly suggests that LATS may function as a tumor suppressor in human cells (Tao et al. 1999). In this work, I have tested a hypothesis that LATS1 and

LATS2 mutations from human cancer can lead to loss or reduction of their growth- inhibitory activity. Human cancer studies also support that LATS1/2 functions as tumor suppressors. hLATS1 is located at a locus frequently involved in loss of heterozygosity

(LOH) in ovarian (Cooke et al. 1996; Lee et al. 1990), cervical (Mazurenko et al. 1999;

Fuji et al. 1996), and breast cancers (Theile et al. 1996; Noviell et al. 1996). hLATS2 is also located to a frequent LOH locus in breast, ovary and liver cancers (Lee et al. 1988;

SATO et al. 1991; WANG et al. 1988). In addition, allelic loss of hLATS1 has been reported in ovarian cancer (HANSEN et al. 2002), as well as homozygous deletion of hLATS2 in 5 out of 45 pleura mesotheliomas (Murakami et al. 2011). Moreover, hLATS1/2 have been reported to exhibit promoter hypermethylation and mRNA down-regulation in breast cancer (Takahashi et al. 2005), astrocytoma (Jiang et al. 2006), and head and neck squamous cell carcinoma (Steinmann et al. 2009). Recent human cancer genome projects

have begun to produce large amount of sequence data and more point mutations from hLATS1/2 genes have been identified. However, functional significance of these mutations is unclear. In this study, I have focused on LATS1 and LATS2 genes to elucidate functional impact of missense and nonsense mutations identified in human cancers at the organism level through molecular and genetic approaches.

2.2 Material and Methods

2.2.1 Protein sequence alignment

For vertebrate Lats1, they include H. sapiens NP_004681.1, P. troglodytes

XP_001173355.1, M. musculus NP_034820.1 and G. gallus XP_419666.2. Vertebrate

Lats2 includes H. sapiens NP_055387.2, P. troglodytes XP_001149147.1, M. musculus

NP_056586.2 and G. gallus XP_417143.3. For D. melanogaster Lats/Warts:

NP_733403.1. For C. elegans Lats/Wts: NP_492699.1. Protein sequence alignment was performed by using MEGA 5.2.1 (Tamura et al. 2011).

2.2.2 DNA constructs and site-directed mutagenesis

pcDNA3.1-hygro-3xFLAG-hLATS1 and pcDNA3.1-hygro-3xFLAG-hLATS2 (gift from Dr. Xiaolong Yang) were used for site-directed mutagenesis to generate V719I and

R806P mutant constructs for hLATS1 and G40E, G909R, C953* and E1016K mutants for hLATS2. hLATS1/2 wild-type and mutant variants were cloned into pUAST-attB and verified by DNA sequencing.

2.2.3 Cell culture and luciferase assays

HEK293T cells were maintained in Dulbecco’s modified Eagle’s medium containing 10% fetal calf serum and 0.5% penicillin– streptomycin. HEK293T cells were transiently transfected using PolyFect Transfection Reagent (Qiagen) and harvested 36 hours later. Plasmids include wild-type or mutant hLATS1 and hLATS2 in pcDNA3.1-

Hygro-3xFLAG, pCMV-YAP, pCMV-Gal4-TEAD4, pUAS-LUC, and pCMV-Renila (gift of Dr. Kun-Liang Guan). Cells were lysed after 36 hours and luciferase activity was detected using the Dual-Luciferase Reporter assay kit and GloMax-multi detection system (Promega). Drosophila S2R+ cells were cultured in Schneider’s medium (Sigma) containing 10% fetal calf serum. S2 cells were transfected using Effectene (Qiagen) with wild-type or mutant hLATS1/2 in pUAST-attB, pAc-Yki, pUAST-Sd, p3xSd-LUC, and copia-Renila (gift of Dr. Jin Jiang), together with pAc-Gal4. Results represent the mean of standard error of three independent experiments.

2.2.4 Western blot analysis

HEK293T cells were transfected using PolyFect (Qiagen). After 36 hours, cells were treated with 1.0 µM okadaic acid (OA) (Sigma) for 30 min at 37°C, and harvested

36 hours later in lysis buffer (150 mM NaCl, 50 mM Tris-HCl [pH = 7.4], 2 mM EDTA

[pH = 8.0], 1% Triton-X 100, 10% glycerol, 2 mM DTT, 1 mM PMSF, 10 mM NaF, 2 mM Na3VO4, 60 mM Glycerol-2-phosphate, containing Protease inhibitors and phosphatase inhibitor (Sigma). The following antibodies were used: anti-LATS1 (Santa

Cruz), anti-LATS2 and anti-Phospho-LATS1/2 (Thr1079/1041) (Cell Signaling), anti-

FLAG and anti-α-Tubulin (Sigma).

2.2.5 Transgene microinjection and site-specific integration

Embryos from Drosophila attP docking strains y[1] M{vas-int.Dm}ZH-2A w[*];

M{3xP3-RFP.attP}ZH-86Fb or y[1] M{vas-int.Dm}ZH-2A w[*]; M{3xP3-RFP.attP}ZH-

51D (from Bloomington Drosophila Stock Center, BDSC; Bischof et al., 2007) were injected with the pUAST-attB constructs carrying wild-type hLATS1/2 and mutant variants. The phiC-31 intergrase vas-int recognized attP in constructs injected and integrated the constructs into the chromosome attB in chromosome II (51D) or III (86Fb).

Fly embryo injections were done by my hand and Rainbow Transgenic Flies Inc. Orange eye color was a selection marker for isolating transformants. Transgenic lines were established through standard procedures. Flies with pUAST-lacZ-attB inserted in 51D or

86Fb (gift of Dr. Johannes Bischof) were used as controls.

2.2.6 Drosophila genetics and wing analysis

The Drosophila strains that carry wts-RNAi on the second chromosome

(#v106174) and the third chromosome (#34064) were obtained from the Vienna

Drosophila RNAi Center (VDRC) and Bloomington Drosophila Stock Center (BDSC), respectively. The VDRC UAS-wts-RNAi inserted in the second chromosome was combined with UAS-hLATS1/2 in the third chromosome (86Fb), while the BDSC UAS- wts-RNAi inserted in the third chromosome was combined with UAS-hLATS1/2 at the second chromosome (51D). MS1096-Gal4 was used to drive wing-specific expression through upstream activation sequence (UAS) of LATS1/2 and their mutants in the integrated constructs. Adult wings were dissected and analyzed by Image J for size measurement.

2.3 Results

2.3.1 Analysis of mutated site conservation and mutation pathogenicity

When this project was started, only a small number of LATS1/2 mutations were identified, which included V719I from a glioma and R806P from a lung carcinoma for

LATS1, and G40E from a lung carcinoma, G909R from a glioma and C953* from an ovary carcinoma for LATS2. All these five mutations were predicted to disrupt LATS gene function through my bioinformatic analysis (Table 2.1).

Table 2. 1 Mutation pathogenicity and conservation with Drosophila homologue

Gene Sample Name AA AA Mutated Mutated site within Count of SIFT PolyPhen-2 Mutation Name Mutation Mutation site methods Assessor conserved shown to in be Drosophila damaging warts

LATS1 Br27P p.V719I p.V719I Yes protein kinase domain (705-1010) 4 deleterious probably deleterious medium

LATS1 16913 p.R252I p.R252I No 2 tolerated possibly deleterious /benign low

LATS1 16678 p.T255N p.T255N No 2 tolerated benign low

LATS1 16660 p.R502C p.R502C No 4 deleterious probably deleterious medium

LATS1 PD1414a p.M669I p.M669I No 1 tolerated benign neutral

LATS1 PD1364a p.R806P p.R806P Yes protein kinase domain (705-1010) 4 deleterious probably deleterious low

LATS2 NCI-H2126 p.G40E p.G40E No 4 *deleterious probably deleterious low

LATS2 TCGA-36-1568 p.P72L p.P72L No 3 tolerated probably deleterious low

LATS2 TCGA-08-0375 p.R558H p.R558H No 0 tolerated benign neutral

LATS2 Br27P p.G909R p.G909R Yes protein kinase domain (668-973) 4 deleterious probably deleterious medium

LATS2 PD0880a p.C953* p.C953* Yes kinase domain tructation, no AGC C-terminal

LATS2 16929 p.E1016K p.E1016K No AGC-kinase C-terminal (974-1052) 2 tolerated benign low

In silico analysis of LATS1 and LATS2 mutations available in COSMIC database (updated to 2011). Column D shows whether the wild type amino acids are conserved in Drosophila warts, to select the candidates for in vivo transgenic analysis. Column E shows the domain which mutated sites locate. Column F counts the methods in column G-J, which show the mutation damaged protein function, based on conservation and structure information.

Moreover, alignments of amino acid sequence from different species from C. elegans to human have been done to evaluate the evolutionary conservation of these sites.

LATS1-V719, LATS1-R806 and LATS2-G909 are conserved from nematode to vertebrate (Figure 2.1). LATS1-V719, within glycine stretch, and LATS2 G909, located next to the activation segment of kinase domain, are highly conserved in eukaryotic kinases ( HANKE and HUNTER 1995). LATS2-G40 is conserved in vertebrates but not invertebrates (Figure 2.1, the bottom panel). In contrast, LATS2-E1016 is not conserved

(Data not shown) and LATS2-E1016K was due to a single nucleotide polymorphism

(SNP) (Fgiure 4). All of these changes are due to somatic mutations. On the basis of above analyses, I have used V719I, R806P, G40E, G909R, and C953* as representative mutations in this study, to test a hypothesis that loss-of-function mutations in LATS1/2 occur in human cancer.

V719I R806P HsLATS1 A F G E V C L A R E S L A R F Y I A PtLATS1 A F G E V C L A R E N L A R F Y I A MmLATS1 A F G E V C L A R E N L A R F Y I A GgLATS1 A F G E V C L A R E N L A R F Y T A DmLATS A F G E V T L V S E E L A R F Y I A CeLATS A F G K V S L V R E D L A R F Y I A G40E G909R HsLATS2 G L P A G P N S D E M L V G Q P P F PtLATS2 G L P A G P N S D E M L V G Q P P F MmLATS2 G L L V G P N S D E M L V G Q P P F GgLATS2 G L P I G P G S E E M L V G Q P P F DmLATS Q N Y T P L R Y T E M L V G Q P P F CeLATS ------E M V F G R V P F

Figure 2. 1 Evolutionary conservation of amino acids that are mutated in human LATS1 and LATS2.

Human LATS1 (top) and LATS2 (bottom) protein sequences that cover several mutated amino acids are aligned and compared with homologous sequences from other vertebrates, Drosophila and nematode.

2.3.2 Analysis of expression and kinase activity of LATS1 and LATS2 mutations

To address whether LATS1/2 mutations affect their expression and/or activities,

HEK293T cells were transfected with wild type as well as mutant LATS1/2 constructs.

Western blots were done to monitor LATS1/2 protein levels and a phosphor-specific

LATS1/2 antibody was used to identify active LATS1/2 proteins. For LATS1, V719I and

R806P mutations did not significantly alter their expression levels compared to wild-type

(Figure 2.2, the second panel from top). However, V719I moderately and R806P dramatically reduced LATS1 kinase activity (Figure 2.2, the top panel, compare lane 3-4 with lane 2). For LATS2, G40E abolished LATS2 expression as well as activity (Figure

2.2, lane 6); G909R didn’t alter expression level but significantly reduced its activity

(Figure 2.2, lane 7). A nonsense mutation C953* deletes the C-terminus of kinase domain as well as the rest of the C-terminal region. As predicted, C953* led to the production of a truncated inactive protein (Figure 2.2, lane 8); lastly, E1016K reduced its expression somewhat but it has no dramatic change of their activity (Figure 2.2, lane 9). These results indicate that these hLATS1/2 alleles except E1016K are loss-of-function mutations.

Figure 2. 2 Western blot analysis of human LATS1 and LATS2 proteins in HEK293T cells

HEK293T cells were transfected with wild-type (lane 2 and lane 5) or mutant (lane 3-4 and lane 6-9) hLATS1/2 DNA constructs as indicated. Anti-hLATS1 and anti-hLATS2 antibodies were used to specifically detect hLATS1 and hLATS2 expression, respectively (the second and third panels). The endogenous hLATS1but not hLATS2 protein was expressed at a detectable level. An anti-Phospho- LATS1/2 (Thr1079/1041) specifically identified activated hLATS1 and hLATS2 kinase proteins (the top panel). As all hLATS1/2 constructs wee tagged with a Flag sequence, their expressions were all verified by an anti-FLAG antibody (data not shown). Expression of α-Tubulin was monitored to ensure equal loading of protein extracts (the bottom panel). Untransfected HEK293T cells were used as a negative control (lane 1). Protein markers are shown on the left side of the panels.

2.3.3 Functional analysis of LATS1 and LATS2 in tissue growth and organ size control through an in vivo over-expression assay

To test a hypothesis that the human LATS1/2 mutations are loss-of-function mutations that decrease their growth-inhibitory activity, I generated transgenic

Drosophila lines as an in vivo model for measuring activities of wild-type hLATS1/2 and their mutant alleles. To compare the activity of wild-type hLATS1/2 and mutant with the same dosage and chromosome context, each transgene was inserted into the same chromosomal position via the phiC31-based integrase method (BISCHOF et al. 2007).

hLATS1/2 and their variants were selectively expressed in larval wing epithelium under the control of wing-specific drivers such as MS1096-Gal4.

Figure 2. 3 in vivo analysis of hLATS1/2 and variants for their activity to inhibit tissue growth in Drosophila.

(A-H) Images of female wings (anterior to the top) that express the following transgenes under the control of MS1096-Gal4: (A) UAS-86Fb-lacZ as a negative control. (B) UAS-86Fb-hLATS1-wt. (C) UAS-86Fb- hLATS1-V719I. (D) Outlines of the wings shown in (A-C). (E) UAS-86Fb-hLATS2-wt. (F) UAS-86Fb- hLATS2-G909R. (G) UAS-86Fb-hLATS2-C953*. (H) Outlines of the wings shown in (E-G). (I) Quantification of total wing area of the genotypes (n=40 for each genotype) displayed in (A-C and E-G). Results represent mean ± S.E.M. ** indicates p<10-11, *** indicates p<10-53.

As expected, over-expression of wild-type hLATS1 and hLATS2 in Drosophila wing resulted in 34% and 45% reduction in size, respectively (Figure 2.3, A, B, E and I), indicating that human LATS genes are sufficient to reduce tissue growth and organ size.

Interestingly, hLATS1-V719I mutation does not drastically affect hLATS1 activity as the mutant protein can still effectively reduce the organ size (Figure 2.3, C and I). It is appears that Drosophila cells were not sensitive to the structural change caused by V719I alteration. In the case of hLATS2, however, G909R and C953* mutations more effectively reduced the growth-inhibitory activity of hLATS2 as the wings were much less reduced in hLATS2-G909R and hLATS2-C953* transgenic flies (Figure 2.3, E-I).

Therefore, human cancer mutations such as hLATS2-G909R and hLATS2-C953* are loss-of-function mutations that reduce their growth inhibitory function.

2.3.4 Functional analysis of LATS1 and LATS2 through an in vivo phenotypic rescue assay

To further investigate the growth regulatory function of hLATS1/2 mutants, I used

RNAi lines for warts (wts) or dLats, Drosophila ortholog of human LATS1/2, to test the

LATS1/2 function in a wts loss-of-function mutant background. Previous studies have shown that expression of the hLATS1 transgene in homozygous wts mutants rescued lethality (TAO et al. 1999). Under the control of a wing-specific driver MS1096-Gal4,

UAS-wts-RNAi resulted in lethality in the late pupae stage (Figure 2.4). This provided an ideal assay to investigate the difference between the function of hLATS1/2 wild-type and mutant proteins.

To determine the relative survival rate of strains expressing different hLATS1/2 transgenes, I crossed XMS1096-Gal4Y males with transgenic UAS-hLATS1/2 females to generate female offspring with induced hLATS expression, and male offspring without any hLATS expression, in the presence or absence of wts-RNAi. As controls, expression of hLATS1/2 and variants in a wild-type background caused no defects in viability

(Figure 2.4). Compared to wild-type hLATS1, hLATS1-V719I did not fully rescue the lethality caused by wts-RNAi (Figure 2.4). Moreover, unlike wild-type hLATS2, hLATS2 mutants G909R and C953* failed to fully rescue the lethality of wts-RNAi flies (Figure

2.4). These results are consistent with the idea that V719I, G909R and C953* are loss-of- function mutations and were unable to fully replace the function of endogenous wts for normal development.

Figure 2. 4 in vivo analysis of hLATS1/2 and mutants for their ability to rescue the lethality induced by loss of the endogenous wts gene function.

Relative survival rate is measured by the number of adult females divided by the number of adult males produced by parental flies of males MS1096-Gal4 (on the X chromosome) crossed with females with genotype as indicated. The number of offspring males serves as a control to represent the expected number

of female flies that express the UAS transgenes driven by MS1096-Gal4. In a wtsRNAi (II) background, the survival rate in the absence of LATS transgenic expression was zero (the control column). Between 40-148 offspring males were scored from each cross.

To investigate this further, the wings of female offspring were analyzed. In a wts-

RNAi background, both hLATS1 and hLATS2 still exhibited potent growth-inhibiting activities (Figure 2.5, compare 2.5B and 2.5E with 2.5A). Compared with wild-type hLATS1, hLATS1-V719I was less capable in blocking tissue growth (Figure 2.5I, also compared 2.5C with 2.5B). Moreover, hLATS1-R806P also appeared to be less active

(Figure 2.5J). Interestingly, hLATS2 mutant G909R was inactive in growth inhibition as there was no longer reduction in wing size (Figure 2.5I, compared 2.5F with 2.5E). The

G909R wing in a wts-RNAi background was actually 11% larger than wild type wings, possibly due to a dominant-negative effect of the mutant protein. LATS2-C953* failed to maintain the normal wing morphology in wts mutant tissues (Figure 2.5G), which makes it difficult to accurately measure wing size. These results are consistent with what was observed in other assays described in Figure 2.3 and Figure 2.4.

Figure 2. 5 in vivo analysis of hLATS1/2 and variants for their activity to control organ size in the absence of the endogenous wts gene function.

Representative images of female adult wings (anterior to the top): (A) MS1096-Gal4/+; UAS-lacZ/+, which only expresses a UAS-lacZ -Galactosidase expression in the wing has no impact on wing development. (B) MS1096-Gal4/+; UAS-wtsRNAi(II)/+; UAS-hLATS1-wt. (C) MS1096-

Gal4/+; UAS-wtsRNAi(II)/+; UAS-hLATS1-V719I. (E) MS1096-Gal4/+; UAS-wtsRNAi(II)/+; UAS- hLATS2-wt. (F) MS1096-Gal4/+; UAS-wtsRNAi(II)/+; UAS-hLATS2-G909R. (G) MS1096-Gal4/+; UAS- wtsRNAi(II)/+; UAS-hLATS2-C953*. Wing size comparison of samples from (A-C) is shown in (D), and those from (A, E and F) is shown in (H). (I and J) display wing size data from statistic analysis of flies with genotypes described at the bottom portion of the panels (n=40 for each genotype). (J) UAS-wtsRNAi

(III, inserted on the third chromosome) driven by MS1096-Gal4 didn’t cause an obvious lethality and therefore UAS-hLATS1-R806P (II, inserted on the second chromosome) was not tested in a survival rate experiment. The activities of LATS1-R806P in overexpresion and wing-size rescue experiments were analyzed and shown in (J). Results represent mean ± S.E.M. ** indicates p<10-4, *** indicates p<10-22.

2.3.5 Functional analysis of LATS1 and LATS2 variants for their ability to inhibit

YAP/Yki activity

Mammalian Lats1/2 genes mediate growth control by directly phosphorylating and inactivating Yap/Taz transcriptional co-activators (ZHAO et al. 2007; DONG et al.

2007; LEI et al. 2008; HAO et al. 2008). To investigate how the mutations in hLATS1/2 might affect their ability to negatively regulate YAP, I have used luciferase reporter assays to directly measure YAP/Yki transcriptional activities. In cultured HEK293T cells, wild-type hLATS1 and hLATS2 effectively inhibited the transcriptional activity of YAP

(Figure 2.6A). V719I and R806P mutations in hLATS1 and G909R and C953* in hLATS2 all decreased the activity of hLATS1/2 as inhibitors of YAP (Figure 2.6A). hLATS2-G40E mutation causes protein destabilization as no mutant protein can be detected (Figure 2.2). Indeed, YAP activity was not obviously affected in cells expressing hLATS2-G40E (Figure 2.6A). Similar results were observed using Drosophila S2 cells

(Figure 2.6B). Therefore, these data further support that the hLATS1/2 mutations cause loss or reduction of their activity in mediating Hippo signaling.

A B

Figure 2. 6 Activities of hLATS1/2 and variants in regulating YAP/Yki transcriptional co-activator activity in both human and Drosophila cells.

(A) Relative luciferase activity in HEK293T cells that over-expressed Renila luciferase as an internal control, UAS-luciferase, CMV-YAP (except the first column), and fusion construct TEAD4-Gal4 that fuses

TEAD4 with Gal4 DNA-Binding Domain (except the second column). (B) Relative luciferase activity of

UAS-LATS1/2-wild type and mutants in S2 cells that over-expressed Renila luciferase as an internal control, 3xSd (Scalloped)-luciferase, Ac-Yki (except the first and second columns), and UAS-Sd (except the first column).

2.4 Discussion and Summary

To experimentally examine the function of hLATS1/2 and their variants in vivo, I have used Drosophila as a model to monitor the ability of hLATS1/2 to regulate tissue growth and organ size using an over-expression assay as well as a rescue assay to

compensate for the loss of the endogenous wts gene function. Several hLATS non- synonymous mutations identified at an earlier stage of this project have been used in these tests: V719I and R806P for hLATS1, G40E, G909R and C953* for hLATS2. While

G40E is a protein-null mutation, all other mutations do not affect their protein stability. In all the in vivo function assays, these alleles have been shown to be loss or reduction of hLATS1/2 function mutations. These results support a model that LATS1 and LATS2 are tumor suppressors in human cells and loss or reduction of their functions contributes to tumorigenesis. Therefore, human LATS1 and LATS2 mutations are expected to confer tissue growth advantage to drive tumor development.

Although hLATS mutations such as hLATS2-G909R and hLATS2-C953* are loss- of-function alleles, they exist in a heterozygous state in somatic diploid cancer cells. Loss or reduction of function of the wild-type hLATS allele through mechanisms like promoter hypermethylation in the heterozygous cells would be a way to further lose hLATS function to make the cells defective in growth control. Another mechanism could be mediated through concurrent mutations in other genes in the Hippo pathway.

Interestingly, hLATS1-V719I and hLATS2-G909R were detected as heterozygous mutations in a glioblastoma from the same patient. In a preliminary analysis with over

3,000 samples, I found that concurrency of non-silent mutations in hLATS1 and hLATS2 is significantly higher than what is expected if it occurs randomly (Data not shown).

Therefore, simultaneous mutations in hLATS1 and hLATS2 may allow cells to gain more growth advantage for tumor development.

Previous studies have shown that Lats is a critical component of the Hippo pathway and Yap is a key target of Lats kinase. Indeed, the hLATS1/2 mutations decrease

their ability to inhibit YAP transcriptional activity. As YAP functions to promote tissue growth and appears to be an oncogenic protein, my results suggest that hLATS1/2 mutations contribute to tumor development due to their reduced activity to negatively regulate YAP function.

Chapter 3 Somatic Mutation Spectrum of Hippo Core Components in Human Cancer Genome

3.1 Introductions

In recent years, human cancer genome projects provide unprecedented opportunities for discoveries of cancer genes and signaling pathways that contribute to tumor development. While numerous gene mutations can been identified from each cancer genome, what these mutations mean for cancer is a challenging question to address, especially for those from less understood putative new cancer genes. As a powerful approach, in silico bioinformatics analysis could efficiently sort out mutations that are predicted to interrupt gene function. Such mutations can be valuable for elucidating mechanisms of cancer gene action.

Hippo signaling plays a crucial role in organ development and tumorigenesis.

However, few somatic mutations of its components in human have been identified in the primary tumor tissues. A few single nucleotide variants of LATS2 have been identified by a candidate sequencing approach, including Q566R and D1002* in soft tissue sarcomas (Hisaoka, et al. 2002), S1073R in non-small cell lung carcinomas (Stražišar, et al. 2009), Y649X in mesothelioma (Murakami, et al., 2011). With the technological advance of the next generation sequencing, Cancer Genome Project (CGP) at the Sanger

Institute and The Cancer Genome Altas (TCGA) discovered more DNA mutations in

Hippo pathway components. The mutation characteristics of those mutations are largely unknown.

In general, the point mutations in oncogenes recurrently locate in few amino acid positions. These mutations usually became the driver mutation for tumorigenesis, such as

codon 12 and 13 of KRAS. Tumor suppressor genes more tend to be truncated, or mutated with equal distribution throughout their length (Vogelstein et al 2013). The question I am curious to know here are whether any these rules applied for core hippo components. The quantitative evaluation for Hippo core components may help to understand their mutation distribution and characteristics. In addition, I also wonder if any mutation site could be explained by any functional information available in literature or database, such as post-translational modification or protein partner binding.

I focused on the core 4 tumor suppressors in human hippo pathway: two pairs of kinase homologues, human STK3 (MST2) and STK4 (MST1), as well as human LATS1 and LATS2. The putative oncogenes are two homologs of transcriptional co-activators

YAP and WWTR1 (TAZ), as well as two homologs of transcriptional activators TEAD2 and TEAD4. TEAD proteins are transcription factors that can play a key role in development and the hippo pathway. In human, there are four TEAD proteins (TEAD1-

4). My study focused more upon TEAD 2 and 4, because of their relative higher mutation frequency shown on cbioportal website. In this chapter, I will discuss the mutation frequency in different tissue types, mutation residue distribution and proximity to critical sites in protein domains, and their proven or possible functional impact.

3.2 Material and Methods

3.2.1 Information Sources

Searches through the following information sources were completed in April,

2014. Mutation data and total patient numbers were collected from Catalogue of Somatic

Mutations In Cancer COSMIC

(http://cancer.sanger.ac.uk/cancergenome/projects/cosmic) and cBioPortal for Cancer

Genomics (http://www.cbioportal.org/public-portal/). The gene names used were LATS1,

LATS2, STK3, STK4, YAP1, WWTR1, TEAD2, and TEAD4. Human LATS1 and

LATS2 mutation data was collected from Catalogue of Somatic Mutations in Cancer

(COSMIC) and TCGA databases. LATS1 data includes LATS1,

LATS1_ENST00000543571, LATS1_ENST00000253339, and LATS2 data includes

LATS2, LATS2_ENST00000382592. YAP1 data contain YAP1 and

YAP1_ENST00000282441, among which P46937-1 was used for the figure. All non- synonymous single nucleotide mutations are analyzed, using Uniprot identifier O95835 or Ensembl Protein ID ENSP00000253339 and ENSP00000437550. For humans, LATS1

- H. sapiens NP_004681.1. and LATS2 - H. sapiens NP_055387.2 were used. All mutations were explored in dbSNP (http://www.ncbi.nlm.nih.gov/SNP/index.html).

3.2.2 Evaluation Tools for Missense Amino Acid Mutations

The severity of missense mutations were analyzed by SIFT (Sorting Tolerant

From Intolerant) (Kumar et al., 2009; http://sift.jcvi.org/), PROVEAN (Protein Variation

Effect Analyzer)(Choi et al., 2012; http://provean.jcvi.org/index.php), PolyPhen-2

(Adzhubei et al., 2010; http://genetics.bwh.harvard.edu/pph2/), and Mutation Assessor

(Reva et al., 2011; http://mutationassessor.org/?set=tcga-gbm-nov.2009). Mutation severity was graded on a scale of 0-4 (green to red) depending on how many methods designated the mutation as deleterious. Input for SIFT and Provean methods included the use of an ENSP number for gene identification purposes. ENSP numbers for LATS1

were ENSP00000253339, LATS2- ENSP00000441817, STK3- ENSP00000390500,

STK4- ENSP00000361892, TEAD2- ENSP00000310701, TEAD4- ENSP00000352926,

YAP1- ENSP00000282441, and WWTR1- ENSP00000353847. Input for Polyphen and

Mutation Assessor followed a similar pattern utilizing the gene name and species (ex

LATS1_HUMAN).

3.2.3 Domain and Binding Partner Analysis Tools

Domain information was collected from literature and databases including

PhosphoSitePlus (http://www.phosphosite.org/homeAction.do) and UniProt

(http://www.uniprot.org).

3.3 Results

3.3.1 Somatic mutations of LATS1 and LATS2 genes

In the Catalogue of Somatic Mutation in Cancer (COSMIC) database, 113 non- synonymous LATS1 somatic mutations have been identified from 9183 unique human tumor samples (Figure 3.1). Similarly, there were 80 LATS2 non-synonymous mutations out of 9516 samples. Therefore, an overall mutation rate was about 1.10% for LATS1 and

0.84% for LATS2. In the cBioPortal database, LATS1 and LATS2 exhibited overall mutation rates, 1.83% (173/7390) and 1.50% (111/7390) respectively. As shown in the cBioPortal database, the top three highest mutation rates for LATS1 occurred in stomach adenocarcinoma (5.91%, n=220), uterine corpus endometrial carcinoma (4%, n= 248), and bladder urothelial carcinoma (3.1%, n=130). There were no LATS1 mutations found in thyroid carcinoma (n=401), glioblastoma multiforme (n=283), and prostate

adenocarcinoma (n=261). Moreover, the highest LATS2 non-synonymous mutation rate occurred in uterine corpus endometrial carcinoma (5.2%, n = 248), stomach adenocarcinoma (4.1%, n=220), and lung adenocarcinoma (3.9%, n=229). Lastly, There were no mutations found in LATS2 thyroid carcinoma (0%, n=401), glioblastoma multiforme (0%, n=283), and acute myeloid leukemia (0%, n=196).

Figure 3. 1 Mutation spectra of LATS1 and LATS2.

Putative ubiquitin associated domain(UBA) (LATS1:98-139, LATS2: 98-139), Proline rich segment (P- rich) (LATS1: 236-266), PAPA motif (Proline-alanine repeats) (LATS2: 403-463), PPPY motif (LATS1:

373-376 and 556-559, LATS2: 515-518 ), N terminal Regulatory Domain(NTR)(LATS1: 642-704, LATS2:

603-667 ), Kinase domain I-XI (LATS1: 705-1010, LATS2: 668-973), Insert (LATS1:861-905, LATS2:

824-868), Activation segment(AS) (LATS1: 906-922, LATS2: 824-868), C terminal domain(CTD)(LATS1: 1011-1090 ,LATS2: 974-1052), and Hydrophobic motif (HM) (LATS1: 1075-1085,

LATS2: 1037-1047). The blue triangles are towards the phosphorylation sites by upstream kinases. The blue bars represent the binding sites or baits to bind with the partners shown below the bars.

To determine the mutation distribution across different domains, analysis through

Fisher's exact test showed that both the kinase domain (p=.01075) and proline-rich

(p=.0312) of LATS1 displayed the highest mutation frequency among all the LATS1 domains. The proline-rich domain had 7 mutations in a 31-amino acid (aa) region (2.2 mutations/10 aa), and the kinase domain had 43 mutations in a 306-aa region (1.41 mutations/10 aa). In LATS2, the kinase domain (p=5.66e-05) and insertion domain

(p=0.03121) had the highest mutation frequency. LATS2 kinase domain had 44 mutations in a 306-aa region (1.48 mutation/10 aa) and LATS2 insertion domain had 9 in

45 aa (2 mutation/10 aa). These data support that selections have occurred to enrich mutations in functionally significant regions such as kinase domain to facilitate tumorigenesis.

Among all the mutations, nonsense and frame-shift mutations clearly disrupted

LATS1/2 gene function. Thirteen and 10 unique nonsense mutations were found in LATS1 and LATS2, respectively. Moreover, 12 and 7 unique frame-shift mutations were detected in LATS1 and LATS2, respectively. The pattern of equal distribution of LATS1/2 nonsense or frameshift mutations was consistent with the idea that LATS1 and LATS2 are tumor

suppressor genes. The percentage for amino acid residues, carrying either nonsense or frameshift mutations for LATS1 and LATS2, was 17.43% and 10.69% respectively.

To predict the functional relevance of other non-synonymous mutations in

LATS1/2, I performed analyses of evolutionary conservation and protein structure through four different mutation-assessing methods: SIFT, PROVEAN, PolyPhen-2, and

Mutation Assessor (Ng et al., 2003; Adzhubei et al., 2010; Reva et al., 2011; Choi et al.,

2012). I found that 73.85% (161/218) of missense mutations from LATS1 and 77.98%

(124/159) from LATS2 were predicted to be deleterious by at least one method. In regards to mutations that were labeled severe (deleterious across all four methods), LATS1 had a percentage of 25.69% (56/218) and LATS2 had 31.44% (50/159). Total deleterious counts, including both nonsense and missense mutations, were also calculated for LATS1 and LATS2. LATS1 had a cumulative deleterious count of 91.28% (199/218) and LATS2 had 88.68% (141/159). Interestingly, mutations predicted to be more deleterious were mostly located in the C-terminal portion where the kinase domain is located. These data support that selections must have occurred to enrich mutations in functionally significant regions such as the kinase domain to facilitate cancer development.

LATS1/2 proteins share LATS conserved domain 1 (LCD1) and LATS conserved domain 2 (LCD2), which are highly conserved in all vertebrate LATS1/2 homologues.

LCD1 and LCD2 domains are critical for LATS1/2 function and regulation. The deletions of either LCD1 or LCD2 in mouse Lats2 abolished its tumor suppressor activity in immortalized mouse cell line (Li, et al., 2003). Lats1 LCD1 knockout mice were born with a low birth rate, from which the mouse embryonic fibroblasts displayed chromosomal instability and tumorigenesis (Yabuta, et al. 2013). A possible reason was

that this region binds with CDK2 to stabilize BRCA2-mediated replication stalling, and maintain genomic fidelity (Pefani et al 2014).

Within LCD1, a short segment called Conserved N-terminal Motif (CNM) (amino acids 72-89 for LATS1 and 71-88 for LATS2) has been recently shown to be important for membrane recruitment and activation of LATS1/2 by Merlin/NF2. The alterations of three highly conserved residues in LATS1/2-A77/76P-I81/80T-L85/84P prevented its interaction with Merlin/NF2, membrane localization and activation (Yin et al. 2013).

LATS1-I81M, LATS1-R82Q/*, LATS2- P72L, LATS2-L77fs, and LATS2-I80fs mutations here might fail to interact with Merlin/NF2 and consequently cannot be activated for growth inhibition (Figure 3.2). The protein fragment products of LATS1 heterozygous DNA mutation L78fs, R82*, E100* and G231* may compete with the wild type LATS1/2 for the binding partners of LCD1. The conservation of certain amino acids between LATS1/2 in the LCD1 domain can be vital for function. Mutations were found at certain conserved sites in both LATS1 and LATS2. These include the sites LATS1-

R28/LATS2-R16, S45/S33, L78/L77, and I81/I80.

Figure 3. 2 Mutations in LATS conserved Domain 1(LCD1) and Putative ubiquitin associated domain(UBA)

(UBA) (LATS1:98-139, LATS2: 98-139). The blue triangles are towards the phosphorylation sites by upstream kinases. The blue bars represent the binding sites or baits to bind with the partners shown below the bars.

Meanwhile, sitting in this CNM, LATS2-S83 was phosphorylated by Aurora-A, which regulates the centrosomal localization and mitotic activity of LATS2 (Toji et al

2004). LATS2-L77fs, and LATS2-I80fs will abolish this phosphorylation site. The kinase assay, tested by the same research group, proved that S91C and T93D did not affect

LATS2-S83 phosphorylation. Here, S91L (found in dbSNP) and G92S also showed to be less deleterious.

Moreover, N463S, P468S, H475Y, A483T, P493S, R502C, P506R/L and W519C

in LCD2 for LATS1 are predicted to be damaging (Figure 3.3). P468S is close to the phosphorylation site LATS1-S464 by NUAK family SNF1-like kinase 1 (NUAK1/2), which promotes LATS1 degradation (Humbert et al., 2010). Additionally, CDC2 phosphorylates S613 of LATS1. CDC2 forms a complex with LATS1 in the centrosome and phosphorylation of S613 occurs during mitosis. (Morisaki, et al. 2002). Mutations near the S613 phosphorylation site, such as K607M and I615T, can possibly interfere with phosphorylation at this site. Three lesser deleterious LATS2 mutations were found in LCD2, which contains the phosphorylation sites S408 and S446 by Chk1/2 in response to UV damage (Okada, et al. 2011). Next to LCD2, LATS2-S380 was phosphorylated by

Aurora A during mitosis, which is a critical step for Aurora A-LATS1/2-Aurora-B axis to regulate accurate mitotic progression (Yabuta et al 2011). S380 is also phosphorylated by PKA and CHK1/2. In addition, LATS2 S380 sits in one of the Ajuba binding regions of LATS2, which regulates the spindle apparatus formation (Abe et al. 2006). So Q345*,

H391H and A392fs, H393fs will disrupt those interactions and related LATS2 functions.

Figure 3. 3 Mutations in LATS Conserved Domain 2(LCD2)

(LATS2: 403-463), PPPY motif (LATS1: 373-376). The blue triangles are towards the phosphorylation sites by upstream kinases. The blue bars represent the binding sites or baits to bind with the partners shown below the bars.

LATS1 has a unique Proline-rich region, containing 6 out the total 30 amino acid residue sites among 22 cases of mutations (Figure 1A). Previous phosphor-proteomics studies have detected the phosphorylated status of T246 and T255 in the P-stretch motif, as well as S336 in LATS1 (Hornbeck et al 2012). One S336G, one T255N and two

T255A mutations in LATS1 clearly abolished the ability of these residues to be

phosphorylated, and one R2552 and two R252G next to those phosphorylation sites may also change the preference of substrate recognition. The wild-type residue of Y277C and

S278C in LATS1 were detected to be phosphorylated in proteomics studies according to

PhosphorSitePlus (Hornbeck et al 2012). There are located within the LIMK-binding portion of hLATS1 (Yang et al. 2004). The multitude of mutations occurring at the P237

(found in the database of SNP) site could implicate this location as a mutation hotspot. 13 unique cases of mutations were discovered at this site, 11 P237Q and 2 P237fs mutations.

The P237Q mutations were found in a variety of different tissue types, including three lung, three haematopoietic, two ovary, two stomach, and one breast. Out of the eleven

P237fs mutations, five were diploid and four were heterozygous loss..The LATS1 N- terminal fragments of P266fs and W268*, R287*, P292fs may affect the wild type binding with those N-terminal binding partners.

The interaction between the Proline-Proline-Proline-Tyrosine (PPPY) motif and the hydrophobic pocket of WW domain is critical for the LATS1/2 binding with either their substrate YAP or their activator KIBRA. It has been reported that LATS1-Y559F and Y376A abrogated the binding with YAP(Hao, et al., 2008; Oka, et al., 2008), while

LATS2-Y518A partially abolish the binding with KIBRA (Xiao, et al., 2011). These interaction may be affected by the third proline mutation P375S in the first PPPY of

LATS1 and the second proline mutation P516L in the only PPPY motif of hLATS2. In addition, G554E is two amino acids prior to the second PPPY.

The N-terminal regulatory domain (NTR) adjacent to the kinase domain is required for LATS1/2 activation by MOB, AMOTL2 and KIBRA (Bothos, et al., 2005;

Paramasivam, et al., 2011; Xiao, et al., 2011) Positively charged residues such as

LATS1-R657/LATS2-K620, LATS1-R660/LATS2-R623, LATS1-R682/LATS2-R645,

LATS1-R694/LATS2-R657, and LATS1-R697/LATS2-R660 in LATS1/2 N-terminal regulatory domain (NTR) are conserved in the NDR protein subfamily from yeast to human. Previous studies have demonstrated that LATS1-R657A and R684A result in the complete loss of their interaction with the negatively charged surface of hMOB1A/B protein and loss of LATS1 kinase activity (Bothos, et al., 2005; Hergovich, et al., 2006).

Here in cancer patients, LATS1 mutants R657C, Q678H, R694C, R697C, and LATS2 mutants R623W, R645L, which are close to or on these sites, probably cause LATS1/2 inactivation in tumor tissues due to the lack of Mob-binding (Figure 3.4). Moreover,

K662fs and E689*/FS in LATS1 and E652* in LATS2 truncated NTR and totally lose their kinase domain. The Drosophila Wts-R702 residue is equivalent to LATS1/2-

R694/R657 and critical for Mob-binding, kinase catalytic activity, and inhibition of tissue growth in development (Ho et al., 2010).

Figure 3. 4 Mutations in N-terminal Regulatory Domain

705-1010, LATS2: 668-973. The blue triangles are towards the phosphorylation sites by upstream kinases.

The blue bars represent the binding sites or baits to bind with the partners shown below the bars.

LATS1/2 are members of the extensively studied AGC (first named after PKA,

PKG, PKC) protein kinase family, which are based on the catalytic kinase domain alignment. Although the crystal structures of some AGC protein members have been deciphered, there is a lack of information available for the kinase domain of LATS1/2. To further estimate mutation-induced structural changes, I did the protein sequence alignment for LATS1/2 based on two AGC family proteins, ROCK and PKC. Conserved in most eukaryotic protein kinases, the N-terminal catalytic domain of LATS1/2 interacts with the phosphate donor ATP through a crucial network, composed of GxGxxGxV loop

(Gly-x-Val LATS1- 712-719/ LATS2-675-682), K734/697, E753/716, DxKxxN (Asp-x-

Lys-x-x-Asn 828-833/791-795) and DFG motif (Asp-Phe-Gly 846-848/809-811) (“x” represents any amino acid) (Endicott, et al., 2012; Hanks and Hunter, 1995). In addition, previous experimental studies also have verified that designed mutants LATS1-

K734M/LATS2-K697A/mLats2-K655M are all kinase-dead (Kamikubo, et al., 2003; Li, et al., 2003; Wei, et al., 2007; Xia, et al., 2002). Mutated within or close to these conserved catalytic elements, LATS1 V719I/A, R737*, R744Q/L, A748T, R827T,

R837H, L844M and LATS2 G675W, A678S, V682L, H691fs, L693M, L699V, D800Y,

G803C are most likely to disrupt ATP binding and catalysis (Figure 3.5). Five cases of

R737*, four R744Q*, one R806* in LATS1 and one cases of H691fs and E722 will truncate LATS1/2 kinase activity severely. LATS1-R995C/L/H and LATS2-G909R also change the highly conserved residue in most eukaryotic protein kinases. Therefore, these cancer mutants may greatly affect their kinase activity.

Figure 3. 5 Mutations in LATS1/2 Kinase domains.

COSMIC and TCGA data is allocated to the top and bottom of their respective charts. Missense mutations were analyzed using four methods, SIFT, Provean, Polyphen, and MutationAssessor. Included Kinase domain I-XI (LATS1: 705-1010, LATS2: 668-973), Insert (LATS1:861-905, LATS2: 824-868), Activation segment(AS) (LATS1: 906-922, LATS2: 824-868). The blue triangles are towards the phosphorylation sites by upstream kinases. The blue bars represent the binding sites or baits to bind with the partners shown below the bars.

Certain genes interact with the LATS2 kinase domains such as AJUBA, KIBRA,

Androgen Receptor (AR), MOB, and AMOTL2. AR interacts with LATS2 from aa 636 to aa 799. “LATS2 directly binds the androgen receptor transcription factor and inhibits its ability to regulate transcription of several genes including probasin and PSA” (Visser, et al. 2010). Deleterious mutations (E652*, H691fs, E722*) within the AR binding region can result in the androgen receptor not being inhibited and upregulation of genes such as probasin and PSA.

Conserved sites between the kinase domain of LATS1 and LATS2 may be vital for function of kinase and gene function. Mutations occurring on these conserved sites for both LATS1/2 included LATS1-N762/LATS2-725, R806/R769, A810/A773,

R827/R790, D837/D800, R854/R817, and R995/R958. These mutations were all regarded as deleterious by at least one method. The mutations occurring at the 806/769,

810/773, 827/790, 837/800, and 854/817 sites were marked as deleterious across all four methods, resulting in a severely damaging classification.

Unlike the other AGC family members, LATS1/2 and the NDR subfamily members have an insert between the DFG motif at the end of the kinase subdomain VII and the activation segment in the kinase subdomain VIII. The basic amino acid residues inhibit the activation segment, which function as auto-inhibitory sequence (Hergovich, et al., 2006). LATS1-D871Y and P882H abolished the acidic residues or introduced the basic residues, which may enhance the auto-inhibition of LATS1/2 and deregulation of

YAP. In contrast, LATS1-Q863E, LATS2-R849L, G851E, and K863E may have opposite effects.

The phospho-S909/S872 and other residues in the activation segment (AS) organize interaction between the ATP in network and the substrate of LATS1/2. Previous study has demonstrated that human LATS1 mutant S909A/D/E cannot be activated by

MST2. Therefore, LATS2-S872L would abolish the kinase activity, and LATS1- A899fs,

Q903*, T913I and LATS2-T876N, A881V also likely damage catalytic activity. In the C- terminal of kinase IX and XI subdomain, the LATS2-G909 and LATS1-R995 are highly conserved among eukaryotic kinases (Endicott, et al., 2012). The five cases of R995 on these two sites probably will damage the kinase domain structure or function. While

LATS2-G909R and C953* has been proven to be damaging mutation for LATS2 kinase activity and YAP regulation. Similar to LATS2-C953*, LATS1-K1005*, F1010fs may also damage the last kinase subdomain and the following domains.

In the C-terminal domain (CTD), NFD (Asn-Phe-Asp) Conserved phenylalanine residues in hydrophobic motif(HM) and interacts with hydrophobic pocket in the N- terminus of kinase domain to facilitate the activation of PKC, a AGC family kinase protein (Pearce, et al., 2010). In the conserved NFD motif of LATS1, N1038H may disrupt the activation of LATS1 kinase domain (Figure 3.6). In addition, LATS1-

R1020T, P1028A/T, S1023C and LATS2-P996L, D998G may also disrupt this activation of LATS1/2. LATS1/2 activation also requires phosphorylation of T1079/T1041 by

MST1/2 in hydrophobic motif (HM), which has consensus F-x-x-Y/F-T-Y/F-K/R in the

NDR protein subfamily (Hergovich et al., 2006). Therefore, two T1041I and one T1041P of LATS2 are the phosphorylation deficient, and LATS1-R1082K, D1086Y and LATS2-

E1039K, R1043L, D1048N next to it may disrupt the kinase activation as well. Mutations at conserved sites between LATS1/2 such as LATS1-F1015/LATS2-F978, R1020/R983,

F1039/F1001, and D1086/D1048 can disrupt function at the C-terminus.

Figure 3. 6 Mutations in LATS1/2 C terminal domain (CTD) and Hydrophobic motif (HM)

1075-1085, LATS2: 1037-1047). The blue triangles are towards the phosphorylation sites by upstream kinases. The blue bars represent the binding sites or baits to bind with the partners shown below the bars.

3.3.2 Somatic mutations of STK3 and STK4 genes

In the COSMIC database, 42 non-synonymous STK3 mutations have been identified out of 9169 unique samples. In addition, there are 42 unique non-synonymous

STK4 mutations out of a unique sample pool of 8956. This data reveals a mutation rate of .46% for STK3 and .47% for STK4. The cancer genomic database, cBioPortal, was also used to determine which tissue types had the highest and lowest mutation rates for specific genes. In cBioPortal, STK3 and STK4 exhibited overall mutation rates of .70%

(52/7390) and .81% (60/7390) respectively. The three highest tissue types that had non-

synonymous STK3 mutations present were uterine corpus endometrial carcinoma (4%, n=248), stomach adenocarcinoma (4.1%, n=220), and uterine carcinosarcoma (1.8%, n=57). The three lowest tissue types with the lowest mutation rate for STK3 were kidney renal cell carcinoma (0%, n=424), thyroid carcinoma (0%, n=401), and ovarian serous cystadenocarcinoma (0%, n=316). The analysis of tissue types and mutations rates was also done for STK4. Uterine corpus endometrial carcinoma (3.2%, n=248), and bladder urothelial carcinoma (2.3%, n=130), and uterine carcinosarcoma (1.8%, n=57) had the highest mutation rates for non-synonymous mutations in STK4. Lastly, kidney renal cell carcinoma (0%, n=424), thyroid carcinoma (0%, n=401), and head and neck squamous cell carcinoma (0%, n=306) had the lowest mutation rates for STK4.

The amount of nonsense and frameshift mutations was also examined for STK3 and STK4. These mutations are important because of the negative effect they can have on gene’s function. A total of 7 nonsense mutations were found in STK3 and 6 mutations in

STK4. In addition, 3 frame shift mutations were found in STK3 whereas there were no frame shift mutations present in the STK4 data. The percentage for amino acid residues, carrying either nonsense or frameshift mutations for STK3 and STK4, was 15.39% and

8.11% respectively.

In order to assess the functional significance of these mutations, four different mutation-assessing methods were used. Using both the data from COSMIC and cBioPortal, overall mutation rates were determined for both STK3 and STK4 (Figure

3.7). I observed that 76.06% (54/71) of missense mutations for STK3 and 85.13% (63/74) for STK4 were predicted to be damaging by at least one of the methods. Also, 36.62%

(26/71) of the STK3 mutations and 40.54% (30/74) of the STK4 mutations were labeled

damaging by all four of the methods. Total damaging counts was also measured for

STK3/4 including nonsense and missense mutations. The total damaging percentages for

STK3 and STK4 was 91.55% (65/71) and 93.24% (69/74) respectively.

Figure 3. 7 Mutation spectra of STK3 and/ STK4 domain.

COSMIC and TCGA data is allocated to the top and bottom of their respective charts. Missense mutations were analyzed using four methods, SIFT, Provean, Polyphen, and MutationAssessor. Included are Kinase domain (STK3: 27-278, STK4: 30-281), Coiled coil(CC) (STK3: 287-328), Nuclear export signal(NES) 1 and 2 (STK3: 363-369 and 438-447, STK4: 365-370 and 436-444), Inhibitory domain (STK4: 331-394),

Sav/Rassf/Hpo (SARAH) domain (STK3: 437-484, STK4: 432-480), and Nuclear localization sequence(NLS) (STK3: 473-487, STK4: 469-483). The blue triangles are towards the phosphorylation sites by upstream kinases. The blue bars represent the binding sites or baits to bind with the partners shown below the bars.

The kinase domain of STK4 has the highest mutation frequency with statistical significance (p < .05 by fisher’s exact test). In this domain, there were 35 mutations in a

252 amino acid span. This equates to a ratio of 1.4 mutations per 10 amino acids.

The kinase domain is located from 27-278 AA and 30-281 AA in STK3/4 respectively (Figure 3.8). K56/59 is the conserved ATP binding site in STK 3/4 respectively. K56/59R mutation has been shown to disrupt ATP binding (Callus and

Verhagen, 2006). A number of mutations were found around the ATP binding site in both

STK3/4. These include Q60H, V61H in STK3 and A54fs, I55V in STK4. It is possible that any of these deleterious mutations could affect ATP binding in the respective protein.

STK3 is phosphorylated at Y81 and T117 by c-Abl and Akt respectively, along with an auto-phosphorylation site T180. In the vicinity of Y81, a couple of mutations were found that could interrupt c-Abl interaction. Two deleterious mutations were found at R144

(R114Q/*), which could interfere with Akt phosphorylation. The autoactivation site was interesting because of the number of mutations that were found its vicinity. These include

N179S, T180I, and V181L with T180I being the most suspect because it occurs directly on the auto-phosphorylation site (Kim et al 2010). Any of these could cause interference with the auto-phosphorylation mechanism. STK4 shares a number of conserved sites with its homolog, including the auto-phosphorylation site T183 and AKt phosphorylation site at T120. Mutation T183A has been shown to cause STK4 to lose activity (Praskova, et al

2004). That particular mutation was not found in this study, but there were a couple in the vicinity (V184M and R181Q). It is possible that either of these could disrupt auto- phosphorylation and halt STK4 activity. At the Akt phosphorylation site, a non- deleterious mutation was discovered. Additionally, multiple deleterious mutations were found around this site including R117Q, T120M and E123K, although R117Q and

T120M were found in the SNP database. A difference between STK3 and 4 is the presence of JNK which phosphorylates STK4 at the S82 site. A Mutation that could disrupt this interaction is P83L (Bi et al., 2010).

Figure 3. 8 Mutations in STK3/4 Kinase domains

COSMIC and TCGA data is allocated to the top and bottom of their respective charts. Missense mutations were analyzed using four methods, SIFT, Provean, Polyphen, and MutationAssessor. Included are Kinase domain (STK3: 27-278, STK4: 30-281). The blue triangles are towards the phosphorylation sites by upstream kinases. The blue bars represent the binding sites or baits to bind with the partners shown below the bars.

Coiled coil domain is only present in STK3 and the only notable mutation occurs at S316. At this site, a phosphor-serine is present with two different mutations on the site

(S316L/*). The functional significance of this site is unknown as it is unclear what phosphorylates it.

Caspase-3 cleaves STKs on site DELD321 (STK4) and DELD321 (STK3) under proapoptotic stimuli, such as H2O2. Additionally, caspase 6/7 cleaves STK4 on site

TMTD349 (Avruch et al. 2012). The subcellular localization and substrate selectivity are different between full-length ST3/4 polypeptides and caspase cleaved STK3/4 (Avruch et al. 2012). The mutation D322N in STK3 has been shown to make the gene resistant to proteolytic cleavage (Lee et al. 2001). Mutations found within this vicinity include

E318K, D319H, and D322Y. It is possible that any of these could interrupt caspase-3 mediated cleavage. In STK4, D326N and D349N cause STK4 to be resistant to proteolysis cleavage by caspase during apoptosis when found in conjunction with one another (Lee et al. 2001). A couple of mutations were found near these two sites that can possibly have a similar effect. These mutations include D323N, T346I, and G350* and their presence could bring about similar effects brought on by D326N and D349N.

Along with the sites T117/120 in STK3/4, AKt phosphorylates another site further downstream. This site is T384 in STK3 and T387 in STK4. T384 occurs within the inhibitory domain of STK4. In regards to the Akt phosphorylation site in STK3, only one mutation was found in this vicinity. This was V388I and it is possible that it could interfere with AKt phosphorylation. In STK4, the closest mutation was found eight amino acids away from the phosphorylation site (T387).

Akt is probably the negative regulator of STK3/4, a notion supported by a series of publications. Both STK3/4 have one canonical consensus sequence for the phosphorylation by Akt RLRNKT120/117L. Akt phosphorylates STK4-T120 in response to

IGF-1, which is functionally proven by phosphomimetic T120D mutant and phosphorylation-deficient T120A mutation (Avruch et al, 2012). There are a couple of mutations around this site (R117Q, K119R, E123K) and one is directly on it (T120M).

T120M is nonpolar, much like T120A, so maybe this mutation will also bring upon a higher level of caspase cleavage and nuclear organization. Additionally, T387 was identified as a frequent site of Akt phosphorylation in vitro (Jang et al. 2007). No mutations were discovered close to this site, with the closest being G379E.

Both STK3 (438-447 AA) and STK4 (436-444) share this second nuclear export signal domain (Figure 3.9). Within STK3, only one mutation occurs in this region

(N442fs). In STK4, it has been shown that mutation L444P can result in the loss of homodimerization, activation, and autophosphorylation (Praskova et al., 2004). This mutation occurs within the NES2 domain and a couple mutations were found within its vicinity. These include D443N and R447T (although this one is not in NES2). Perhaps the devastating effects brought on by L444P can be mimicked by one of these two.

Figure 3. 9 Mutations in STK3/4 C-terminal Domains

3.3.3 Somatic mutations of YAP and WWTR1 genes

In Catalogue of Somatic Mutation in Cancer (COSMIC) database, 27 non- synonymous YAP1 somatic mutations have been identified from 8845 unique human tumor samples. Also, there are 33 WWTR1 non-synonymous mutations out 8294 samples. This gives a mutation rate of about .31% for YAP1 and .40% for WWTR1.

Within the cBioPortal database, YAP1 had a mutation rate of .5% (37/7390) and

WWTR1had .46% (34/7390). Using cBioPortal, bladder urothelial carcinoma (2.3%,

n=130), uterine corpus endometrial carcinoma (1.6%, n=248), and kidney chromophobe

(1.6%, n=66) displayed the highest rate of non-synonymous mutations for YAP1. The lowest three mutation rates for YAP1 were kidney renal cell carcinoma (0%, n=424), thyroid carcinoma (0%, n=301), and brain lower grade glioma (0%, n=289). In respect to

WWTR1, the highest rates of non-synonymous mutations were present in pancreatic adenocarcinoma (13.2%, n=91), chromophone renal cell carcinoma (3%, n=66), and uterine corpus endometrial carcinoma (1.6%, n=248). Lastly, kidney renal cell carcinoma

(0%, n=424), thyroid carcinoma (0%, n= 401), and ovarian serous cystadenocarcinoma

(0%, n=316) displayed the lowest non-synonymous mutation rate in WWTR1.

Nonsense and frame shift mutations can also impact the function of genes such as

WWTR1 and YAP1 (Figure 3.10). A total of 4 nonsense mutations (W199*, L161*,

R411*, and 505Q) were found in YAP1 and 4 (R248*, Q353*, G366*, and G372*) were found in WWTR1. In addition, six frame shift mutations (A47, P48, A50, Q431, G432, and M463) were found in YAP1 along. There were no frame shift mutations found in

WWTR1 using COSMIC and cBioPortal. The percentage for amino acid residues, carrying either nonsense or frameshift mutations for YAP1 and WWTR1, was 18.52% and 15.91% respectively.

To predict the functional relevance of the non-synonymous in YAP1 and

WWTR1, four different mutation-assessing methods were used: SIFT, Provean, Mutation

Assessor, and Polypen-2. I observed that 74.07% (40/54) of missense mutations from

YAP1 and 72.72% (32/44) mutations from WWTR1 were predicted by at least one method to be damaging. 24.07% (13/54) of the mutations in YAP1 and 15.9% (7/44) of the mutations in WWTR1 were observed to be severely damaging (registered as

damaging across all four methods). Lastly, when including both nonsense and damaging missense mutations, YAP1 had a percentage of 92.59% (50/54) and WWTR1 had

88.63% (39/44).

The PDZ domain of YAP1 (p=.00116) and 14-3-3 binding domain of WWTR1

(p=.0389) displayed the highest mutation frequency. The PDZ domain had 4 mutations in

6 amino acids (6.7 mutation/10 AA), and 14-3-3 of WWTR1 had 3 mutations in an 11 amino acid span (2.72 mutations/10 amino acid).

Figure 3. 10 Mutation spectra of YAP and TAZ.

Proline-rich region (YAP: 3-49), TEAD binding domain (YAP: 47-111, TAZ: 13-57), 14-3-3 binding domain(14-3-3) (YAP: 122-132, TAZ: 84-94), WW domain (YAP: 174-200 and 231-263, TAZ: 127-153),

SH3 binding motif(SH3) (YAP: 278-287), Coiled coil domain (YAP: 298-359, and TAZ: 225-259),

Transcriptional activation domain (TAD) (YAP: 276-472, and TAZ: 208-381), and PDZ binding motif(PDZ) (YAP: 499-504, TAZ: 394-400). The blue triangles are towards the phosphorylation sites by upstream modifiers. The blue bars represent the binding sites or baits to bind with the partners shown below the bars.

YAP1 is evolutionary conserved, with the exception of nematodes, and YAP orthologs are found in all metazoans and even in prematazoans (Hilman and Gat, 2011).

However, WWTR1 is only found in vertebrates. Both of these genes are oncogenes and thus implicated in some cancers. In human liver tumors, high levels of YAP are observed and it is thought that YAP is the driver of this tumorigenesis (Xu, et al. 2009). In 15% of ovarian cancers, a high YAP level is identified (Zhang, et al. 2011). Studies have shown that WWTR1 is overexpressed in about 20% of human breast cancers (Chan, et al. 2008).

It has also been implicated in the carcinogenesis of non-small lung cancer (Zhou, et al.

2011). Although YAP functions as an oncogenes in most scenario, there is a few evidence suggests that YAP can function as a tumor suppressor in breast and intestine

(Wang et al., 2013 Li et al., 2012). YAP expression has been found to be decreased in head and neck cancers (HNSCC). Also, YAP knockdown displays enhanced cell proliferation and survival in HNSSC cell lines.

YAP1 has a proline rich domain located on its N-terminus whereas it is lacking in

WWTR1(Figure 3.11). However, in YAP1 this region interacts with heterogenous nuclear ribonuclear protein U (HnRNPU) (Howell, et al. 2004). HnRNPU is a nuclear matrix RNA-binding protein that plays a role in mRNA processing (Varelas 2014). It is possible that one of the two frame shift mutations found in the proline region could affect

HnPNPU binding,

Figure 3. 11 Mutations in YAP1 N-terminal domains.

Proline-rich region (YAP: 3-49), TEAD binding domain (YAP: 47-111, TAZ: 13-57), 14-3-3 binding domain(14-3-3) (YAP: 122-132, TAZ: 84-94), The blue triangles are towards the phosphorylation sites by upstream modifiers. The blue bars represent the binding sites or baits to bind with the partners shown below the bars.

YAP1/WWTR1 both bind TEAD (YAP: AA 47- 111, WWTR1: AA 13-57) at the

N-terminus. Previous studies, have demonstrated that mutation S94A causes the loss of interaction with TEAD1-4 (Li, et al. 2010). Additionally, F95A and F96A cause loss of interaction with TEAD1. P98L was found in the vi cinity of the aforementioned mutations and may play a role in disrupting TEAD interaction. Additionally, another study demonstrated that the mutations to the YAP conserved sites L54, F54, M71, R72,

L76, P77, F80, and F81 (L31, F32, W43, R44, L48, P49, F52, and F53 in WWTR1) could reduce binding to TEADs (Chen, et al. 20 10). In WWTR1, the mutation S51A/D results in the loss of interaction with TEAD4 (Zhang et al 2009). The mutation K54E was found in relative vicinity to this S51 site, and it could also disrupt WWTR1/ TEAD binding, which mediates the expression of pro-proliferative genes. Within the TEAD binding domain, LATS also phosphorylates the S61 and S109 site in YAP1 and the S66 site in WWTR1. No mutations were found within a five amino acid vicinity of these phosphorylation sites.

YAP and WWTR1 share a 14-3-3 binding domain, resulting in cytoplasmic sequestration and ubiquitin-dependent degradation. The phosphorylation of site YAP-

S127/WWTR1-S89 by LATS creates this 14-3-3 binding site (see side figure below,

Zhao, et al. 2010). Previous studies have demonstrated that mutations such as

H122A/R/N/K can result in loss phosphorylation by LATS and decreased interaction with YWHAB (14-3-3 beta) (Zhao, et al. 2007). Additionally, the same group showed that H122L/Y mutations can also result in decreased interaction with YWHAB and significantly decreased phosphorylation at S-127. The three mutations R124A, S127A,

and P129D also result in phosphorylation and interaction with YWHAB being affected

(Hao, et al. 2008). It is possible that the deleterious mutations, R124Q, H126R, and

H126Y may interfere with the phosphorylation at site S127 and mimic the results of the preceding sentences, ultimately resulting in deregulation of YAP and increasing its oncogenic properties. Additionally, P129S, on the +2 position on LATS2 consensus phosphorylation target sequence, is more preferred than wild type. It may lead to more phosphorylation at the S127 site. Moreover, the phosphorylation of YAP equivalence

S127T by LATS1 reduced 33% compared with wild-type (Hao, et al. 2008). Comparing to wild type YAP1 peptide, the phosphorylation of R124A and R124K reduced modestly, while the mutation R124A and R127A fully abolished its phosphorylation by LATS1.

(Hao, et al. 2008). In regards to WWTR1, the mutation S89A has been shown to result in significant resistance to inhibition by STK3 and LATS2 (Lei, et al. 2008). There were three mutations in proximity of the S89 phosphorylation site, S89T, S90P, and P91H.

S89T occurs directly on the phosphorylation site. S89T or a combination of it and S90P and P91H can possibly result in WWTR1 not being phosphorylated at the S89T site.

Figure 3. 12 A model of YAP1/WWTR1 regulation by the Hippo pathway and CK1ε and β-TRCP.

YAP1 contains two WW domains in certain isoforms (AA 174-200 and AA 231-

263) whereas WWTR1 only has one (AA 127-153) (Figure 3.13). WW domains in YAP and WWTR1 have two different functions. First is to bind to PPXY motifs of LATS resulting in LATS-mediated phosphorylation and cytoplasmic retention of

YAP/WWTR1. The second function is to interact with PPXY-containing transcription factors, such as RUNX1/2, ErbB4, p73, NFE2, KLF5, and PML, to promote their targeted gene expression in the nucleus (Wang et al. 2013). It has been shown that the

WW domains of YAP are not involved in TEAD binding (Tian, et al. 2010). YAP interaction with some of the aforementioned binding partners, PML and p73, can induce apoptosis in response to DNA damage (Wang et al 2013). YAP1 binds p73 to regulate transcriptional activity and cell death induction (Strano et al. 2013). This interaction can be negatively mediated by Akt-phosphorylation and negatively so by DNA damage. PML plays a role in transcriptional regulation, DNA damage response, and DNA repair. Data suggests that YAP1 may link the two tumor suppressor pathways of PML and p73 in response to DNA damage (Lapi, et al. 2008). YAP1 can also bring upon oncogenic consequences by interacting with certain binding partners, such as ErbB4. YAP1 binds with ErbB4 to promote its transcriptional activity (Wang 2013). RUNX1/2 shows reciprocal antagonism with ER alpha, so that it can act in tumor suppressor or oncogenic role depending upon the situation (Chimge and Frenkel 2012). Previous studies have demonstrated the effects that certain mutations have on interactiosn between the PPXY motif and YAP WW domain. For example, W199A in YAP can result in a loss of interaction with ERBB4 (Komuro, et al. 2003). The mutation W199* was also located at this site that could result in the same fate as that brought on by W199A. Other deleterious mutations such as S183F S227L, W236L, and G244A can interfere with YAP binding in the WW domains. Mutations in the second WW domain can especially interfere with the binding of PTPN14 because the second WW domain is more important for PTPN14 binding (Wang, et al. 2012).

Figure 3. 13 Mutations around YAP1/WWTR1 WW domains

3 binding domain (14-3-3) (YAP: 122-132, TAZ: 84-94), WW domain (YAP: 174-200 and 231-263, TAZ:

127-153). The blue triangles are towards the phosphorylation sites by upstream modifiers. The blue bars represent the binding sites or baits to bind with the partners shown below the bars.

RUNX1/2 and KLF5 bind WWTR1 WW binding domain, as well as YAP WW domains. NKX2, PAX3, and KLF5 are unique to interacting with WWTR1 WW domains. KLF5 is transcription factor implicated in breast cell proliferation (Zhao, et al.

2011). Zhao et al has demonstrated that the interaction between KLF5 and WWTR1 may protect KLF5 from WWP1-meidated ubiquitination and degradation. They suggest that

KLF5 could interact with WWTR1 to promote breast cell proliferation and tumor growth.

PAX3 and WWTR1 has been shown to be co-expressed in paraxial mesoderm, limb

buds, and neural tube during embryogenesis (Murakami, et al. 2005). Murakami et al demonstrates that WWTR1 serves as a transcriptional co-activator of PAX3, and this interaction is important for embryonic development. NKX2 has been found to be co- expressed with WWTR1 in the epithelial cells of the fetal lung (Park, et al. 2004).

Deleterious mutations present in the WW domain of WWTR1 could interfere with the interaction between the PPxY motif and said WW domain. These mutations include

A136T, K148N, and I149N.

A coiled coil domain is present in both YAP (AA 298-359) and WWTR1 (AA

225-259). Two interesting mutations occurred in the coiled coil domain of WWTR1. The first occurred at the 233 site with Q233del and Q233H. There were five unique cases of these two mutations at this site. All four of the Q233del mutations occurred in the pancreas whereas the Q233H mutation was in the bladder. The other site was the 248 site which had four cases of a R248* nonsense mutation. Two of these R248* mutations occurred in the colon and the other two were in the urinary system. In the YAP coiled coil domain, there were no mutations that showed up in multiple cases.

Figure 3. 14 Mutations in YAP1/WWTR1 C-terminal domains.

(YAP: 276-472, and TAZ: 208-381), and PDZ binding motif (PDZ) (YAP: 499-504, TAZ: 394-400). The blue triangles are towards the phosphorylation sites by upstream modifiers. The blue bars represent the binding sites or baits to bind with the partners shown below the bars.

Both YAP/TAZ have a transcriptional activation domain (TAD) located in the C- terminus (Figure 3.14). Within the TAD of YAP are a couple of important phosphorylation sites. In this domain, CDK1 phosphorylates S367, LATS phosphorylates

YAP-S397/TAZ-S311 and CK1 phosphorylates S400 and S403. The phosphorylation of

YAP-S397 (TAZ-S311) by LATS provides a docking site, which allows CK1 to phosphorylate YAP-S400 and S403 (TAZ S311 and S314). The latter phosphorylated serines create phosphodegron, which can recruit SCF to degrade YAP. In TAZ, LATS2

phosphorylates the S311 site. A mutation at this site (S311A) brings about partial resistance to MST2 and LATS2 (Lei, et al. 2008). Therefore, E396D and T398A in YAP,

H306L and S307L in WWTR1 may polish the inhibition of LATS1 and CK1, and transform WWTR1 to be more oncogenic.

C-Abl phosphorylates Y407 and promotes the expression of p73-mediated proapoptotic and growth arrest target genes through the binding and coactivation of p73 by YAP in response to DNA damage (Levy et al. 2008). Additionally, Set7 methylates

K494 near the PDZ domain. The mutation S397A has been shown to result in the loss of phosphorylation by LATS1 (Hao, et al. 2008). Also, the mutation Y407E has demonstrated that it brings about enhances interaction with tp73 whereas Y407F results in no phosphorylation by Abl1 and partial loss of binding to tp73 (Levy, et al. 2008).

Although these mutations were not found in this study, a number of them were discovered around the phosphorylation sites. These include E396D, T398A, and

R411*/Q. These mutations could bring about changes resulting in a lack of phosphorylation by LATS, Abl1, and CK1. If any of these mutations interferes with the

S397 site, it can result in SCF not being recruited and ultimately YAP does not get degraded. If YAP is not degraded, it can progress to the nucleus to interact with TEAD to upregulate gene expression. Mutations R411* and R411Q can possibly affect phosphorylation by c-Abl. This could reduce p73-mediated apoptosis and growth arrest.

At the 299 site, three unique mutations were discovered to be heterozygous in 7 cases.

The F299L and F299S mutations were discovered twice in the large intestine and F299V three times in breast tissue.

Both YAP/TAZ share a PDZ binding motif at the C-terminus. Deletion of the

PDZ-binding motif inhibited the nuclear translocation of YAP (Shimomura et al,. 2013).

This resulted in YAP not being able to interact with TEAD in the nucleus, reducing its oncogenic activity. Additionally, this study showed that cells expressing YAP with the

PDZ motif deletion had a decreased CTGF expression. It was postulated that the lack of

PDZ domain disrupted TEAD/YAP interaction so the CTFG gene expression was decreased. The PDZ domain of YAP also binds EBP50, which is involved in the compartmentalization of YAP at the apical membrane (Mohler, et al. 1999). Mutants lacking the EBP50 motif are mislocalized when expressed in airway epithelial cells. TAZ binds NHERF-2 and PALS1 at the PDZ domain. NHERF-2 is a molecule that tethers plasma membrane ion channels and receptors to cytoskeleton actin (Kanai, et al. 2000).

PALS1 is an associated tight junction protein, which along with PAR3-PAR6-aPKC regulates apicobasal polarity of epithelial cells (Duning, et al. 2010). There were no mutations found in the PDZ domain of TAZ.

3.3.4 Somatic mutations of TEAD2 and TEAD4 genes

Using COSMIC, the catalogue of somatic mutations in cancer, I found 59 non- synonymous TEAD2 mutations were identified out of 8294 unique samples.

Additionally, 35 non-synonymous TEAD4 mutations were found out of 8295 unique samples. Using solely the COSMIC data, this gives a mutation rate of about .71% for

TEAD2 and .42% for TEAD4. cBioPortal data revealed a mutation rate of .51%

(38/7390) for TEAD2 and .62% (46/7390) for TEAD4. As shown in the cBioPortal database, the tissue types with the highest mutation rate for TEAD2 are uterine corpus

endometrial carcinoma (3.2%, n=248), cervical squamous cell carcinoma and endocervical adenocarcinoma (2.6%, n=39), and skin cutaneous melanoma (2.5%, n=

278). The three tissue types with the lowest mutation rate for TEAD2 are thyroid carcinoma (0%, n=401), head and neck squamous cell carcinoma (0%, n= 306), and prostate adenocarcinoma (0%, n=261). The relationship between mutation rates and tissue types was also examined for TEAD2. Pancreatic adenocarcinoma (4.4%, n=91), uterine corpus endometrial carcinoma (3.6%, n=248), and skin cutaneous melanoma

(2.9%, n=278) display the three highest non-synonymous mutation rates for TEAD4.

Lastly, the three lowest non-synonymous mutation rates for TEAD4 are present in kidney renal cell carcinoma (0%, n=424), ovarian serous cystadenocarcinoma (0%, n=316), and brain lower grade glioma (0%, n=289).

Using cosmic and cBioPortal, the number of nonsense and frame shift mutations was also quantified for tead2/4. There were 4 nonsense mutations (R95*, Q116*, W172*, and E356*) for TEAD2 and 3 (Q188*, W299*, K333*) for TEAD4. I found 7 frame shift mutations in TEAD2 and 5 in TEAD4. It is interesting to note that position 295 of

TEAD2 showed one of the highest mutation rates out of the eight genes. The mutation

H295fs was found in 22 unique patients. The majority of these samples (19/22) were from the large intestine, mainly the colon. The percentages for amino acid residues, carrying either nonsense or frameshift mutations for TEAD2 and TEAD4, were 42.31% and 14.04% respectively.

Using both COSMIC and cBioPortal data, I analyzed the unique mutations with a set of four different mutation-assessing methods. I observed that 52.56% (41/78) of mutations for TEAD2 and 75.44% (43/57) for TEAD4 mutations were labeled as

deleterious by at least one of the methods. 34.62% (27/78) of TEAD2 mutations and

42.11% (24/57) of TEAD4 mutations were labeled as severely deleterious (being labeled as deleterious across all four methods). Deleterious missense mutations and nonsense mutations were also added together to get a total deleterious count. For TEAD2, the counts were 94.87% (74/78) and 89.47% (51/57) for TEAD4.

The TEAD proteins contain both a TEA domain, which binds certain DNA elements, and a transactivation domain, which binds co-activators (Figure 3.15). The

TEAD proteins interact with several co-activators, including YAP/TAZ, VG11 proteins, and p160s (Pobbati, Hong. 2013). The TEAD/YAP interaction can promoting the expression of downstream genes, such as CTGF, which are involved in proliferation and implicated in some cancers. The domains that showed the highest mutation frequency within TEAD2/4 were the transcriptional activation domain of TEAD2 (p=.01919) and the TEA domain of TEAD2 (p=.01517). The transcriptional activation domain had a total of 28 mutations within 231 amino acids (1.21 mutations/10AA). The TEA domain of

TEAD4 had 10 mutations in a 68 amino acid span (1.47 mutations/10 AA).

Figure 3. 15 Mutation spectra of TEAD2 and TEAD4

COSMIC and TCGA data is allocated to the top and bottom of their respective charts. Missense mutations were analyzed using four methods, SIFT, Provean, Polyphen, and MutationAssessor. Included are Transcriptional Enhancer Activator domain (TEA) (TEAD2: 40-107, TEAD4: 38-105), Proline rich region (P-Rich) (TEAD2: 152- 216), and Transcriptional activation site (TEAD2: 217-447, TEAD4: 213-434). The blue triangles are towards the phosphorylation sites by upstream kinases. The blue bars represent the binding sites or baits to bind with the partners shown below the bars.

The TEA domain in TEAD is used for DNA binding (Figure 3.16). In TEAD2 it is located from 40-107 AA, and from 28-105 in TEAD4. I found 11 TEAD2 mutations

and 10 TEAD4 in this region in my study. Studies have revealed that this TEA domain binds DNA elements such as 5′-GGAATG-3’ and this is seen in the SV40 enhancer and promoter regions of TEAD target genes (Pobbati, Hong. 2013). The TEA domain is made up of a hydrophobic core which is used for protein stability and DNA interactions.

One of the amino sites involved in this hydrophobic core is Ala-74 of TEAD2. In this study, I found a mutation at this site A84V. In previous work, a disruption between

TEAD and SRF (serum response factor) was found in a A48, K52 double mutant (Gupta et al,. 2001). This may have resulted in the tertiary structure being destabilized because

A48 is located in the hydrophobic core. It can be deduced that perhaps similar destabilization may occur in the A84V mutation. In TEAD2 data, the R95 site was mutated into both a nonsense stop codon and glutamine in separate cases. This was the only site in both TEAD2/4 that had two mutations in the TEA domain.

Figure 3. 16 Mutations in TEAD2/4 Transcriptional Enhancer Activator domain

COSMIC and TCGA data is allocated to the top and bottom of their respective charts. Missense mutations were analyzed using four methods, SIFT, Provean, Polyphen, and MutationAssessor. Included are Transcriptional Enhancer Activator domain (TEA) (TEAD2: 40-107, TEAD4: 38-105. The blue triangles are towards the phosphorylation sites by upstream kinases. The blue bars represent the binding sites or baits to bind with the partners shown below the bars.

Proline rich region is only found in the TEAD2 (152-216). Multiple mutations were discovered in this region including three nonsense mutations, W172fs, W172*, and

Q203fs. The functional significance of mutations in this region is unknown.

Figure 3. 17 Mutations in TEAD2/4 Transcriptional Activation Domains

COSMIC and TCGA data is allocated to the top and bottom of their respective charts. Missense mutations were analyzed using four methods, SIFT, Provean, Polyphen, and MutationAssessor. Included are Transcriptional activation site (TEAD2: 217-447, TEAD4: 213-434). The blue triangles are towards the phosphorylation sites by upstream kinases. The blue bars represent the binding sites or baits to bind with the partners shown below the bars.

The TEAD proteins also share a transcriptional activation domain (TAD) at the

C-terminus (Figure 3.17). It is here that TEAD binds co-activators such as YAP/TAZ and

Vgll proteins. Among the Vgll proteins, Vgll1 is a prognostic marker for bladder cancer

(Pobbati and Hong, 2013). Both of these cofactors have been implicated in some cancers.

Mutations, such as L299K/K301E, V427A/E429K, and Y442H, in the TAD of TEAD2 disrupt YAP1 binding and fail to activate t heir downstream genes (Tian, et al. 2010).

L299P, F302L, and V445A occur on or close to the aforementioned sites and may result in YAP not being able to bind. Additionally, L299A/K301E, E356R/K357E, and

E404K/N405A have been shown to severely disrupt binding activity between TEAD2 and YAP (Tian, et al. 2010). E404K was found as well, so it can be expected to alter

YAP/TEAD2 binding. At the L299 and E356 sites, the mutations L299P and E356* can possibly reduce relative binding activity. Tian, et al also demonstrated that S349A/F305A mutations retained YAP binding. It is possible the mutation S349S can have a similar effect and possibly enhance YAPA/TEAD binding. An interesting mutation occurring in the TEAD2 TAD is at the H295 site. A non-sense H295 frame shift mutation occurred in

22 unique samples. The fact that these mutations occurred in 22 samples raises the question if this is a possible mutation hotspot. In 18 of the samples, c.833delC occurred primarily in the large intestine (16/18) with the pancreas and ovary each having one mutation. With the exception of three, these mutations all showed a heterozygous zygosity. The other four mutations were made up of a c.882_882delCC and three

883_884insC. Again these mutations primarily occurred in the large intestine. The truncation caused by H295fs could result in the inactivation of the transcriptional activation domain.

In TEAD4, previous studies have been undertaken to analyze the functional significance of certain mutations within this domain. The D266A mutation has been shown to reduce transforming ability in TEAD4 (Chen, et al. 2010). There were three mutations found nearby this site that could induce similar results, E263K, V265M, and

R268C. The mutations K297A and W299A result in the loss of interaction with YAP1 and complete loss of transforming ability (Chen, et al. 2010). Two mutations (K297T and

W299*) were found at these two sites that could very well induce a similar loss of interaction with YAP1. F337A has the significance in that it reduces interactions with

YAP1 (Chen, et al. 2010). Again, a mutation was discovered on this same site (F337Y) that could interfere with YAP binding. Lastly, F391A and F393A have been shown to reduce transforming ability of TEAD4 (Chen, et al. 2010). The closest mutation to these two sites was E392K and it could also affect transforming ability of TEAD4. The Chen study has also demonstrated that certain sites in TEAD4 interact with YAP1 through van der Waal forces. These interactions occur with TEAD4 E266/E384, D265/E384 and

YAP1 M71 and R74 respectively. The mutation V265M occurs near these sites in

TEAD4 and could result in van der Waal forces being disrupted.

3.3.5 Somatic mutation frequency and damage degree in all hippo core 8 components

In combined cBioPortal data, the overall mutation rate for all tissue types was

5.45% (403/7390) for the eight hippo core components. The mutation rate for four Hippo

Pathway tumor suppressors was more than double than what was shown for four Hippo

Pathway oncogenes (3.99% vs 1.92%). Out of all tissue types, pancreatic

adenocarcinoma had the highest rate of mutation at 19.8% (n=91), followed by stomach adenocarcinoma (13.6%, n=220), and bladder urothelial carcinoma (13.1%, n=130). The tissues with the lowest mutation rates were acute myeloid leukemia (0%, n=196), medulloblastoma (0%, n=125), and adenoid cystic carcinoma (0%, n=60). Core 8 damaging rate is 91.11% (687/754) and 30.90% (233/754) for severely damaging. The combined COSMIC Core 8 mutation rate was 3.59% (330/9169). The three tissue types with the largest percentages of Core 8 mutations was the endometrium at 13.88%

(80/641), large intestine at 12.48% (80/641), and urinary tract at 11.43% (12/105).

Table 3. 1 The tissue types with the most or least frequently mutated Hippo core omponents.

To evaluate the characteristics of the mutations of Hippo Pathway core 8 components, the proportion of different mutations are calculated. The four core tumor suppressors (LATS1, LATS2, STK3, STK4) have higher recurrent missense mutation rates than nonsense or frameshift mutations, while opposite trends are found in TEAD2 and YAP (Table 3.2).

Table 3. 2 Recurrent Missense (RM) and nonsense or frameshift mutations (NF) of Hippo Pathway core components including SNP.

LATS1 LATS2 STK3 STK4 TEAD2 TEAD4 YAP1 WWTR1 Total counts for all unique mutated cases 218 159 71 74 78 57 54 44 Non-Damaging (0 methods) 19 18 6 5 4 6 4 5 Severe Damaging counts (4 methods) 56 50 26 30 27 24 13 7 % Severly Damaging 25.7% 31.4% 36.6% 40.5% 34.6% 42.1% 24.1% 15.9% Damaging Missense counts (no */FS) 161 124 54 63 42 43 39 32 % Damaging Missense Only 73.9% 78.0% 76.1% 85.1% 53.8% 75.4% 72.2% 72.7% Frame-Shift (FS) 14 7 3 0 5 28 6 0 nonsense(*) 24 10 8 6 3 4 5 7 Trucating mutations (FS/*) 38 17 11 6 32 8 11 7 % Trucating mutations (FS/*) 17.4% 10.7% 15.5% 8.1% 41.0% 14.0% 20.4% 15.9% Damaging counts (2-4 methods, and fs/*) 199 141 65 69 74 51 50 39 % Damaging (Missense and FS/*) 91.3% 88.7% 91.5% 93.2% 94.9% 89.5% 92.6% 88.6% recurrent mutations(>2/site) case counts 67 40 14 19 10 10 4 12 % recurrent mutations(>2/site) 30.7% 25.2% 19.7% 25.7% 12.8% 17.5% 7.4% 27.3% Recurrent Missense (RM) counts include missense mutations at the same amino acid divided by total number of mutations. NF count was number of nonsense or frameshift mutations (*/FS) divided by total number of mutations

3.4 Discussion

Genetic analysis of Lats/Wts family genes using Drosophila and mice models has revealed their role as negative growth regulators and tumor suppressors (Harvey et al., 2013).

The fact that human LATS1 can functionally replace Wts in Drosophila strongly suggests that

LATS may function as a tumor suppressor in human cells. While in vivo functional significance of some of these LATS1/2 mutations have been tested (Yu et al., 2013), results reported here further support that LATS1/2 normally act as tumor suppressors and loss of their functions contributes to human tumorigenesis. This may shed a light for investigate the other hippo tumor suppressors, including STK3 and STK4. Importantly, specific structural alterations determined through mutant analyses should facilitate characterization of mechanisms of hippo core components mutations in tumorigenesis.

Although previous work has shown single Hippo Pathway mutations through candidate approach (Seidel et al. 2007, Steinmann et al. 2009), I first analyzed the mutations data of all eight core components of Hippo Pathway for all the available point mutation data in cancer genome database. The mutation statistics and spectra across core 8 components have been illustrated. I also explore their mutation deleteriousness by multiple pathogeneicity evaluation tools, as well as the damaging mutation rate across Hippo Pathway core component proteins or their domains.

All the mutations have been intensively curated by the literature information. Recurrent deleterious mutations have been identified, which could be the possible driver mutations to experimentally test. This work demonstrates the characteristic of Hippo Pathway mutations, and indirectly illustrates the role of Hippo Pathway core components in carcinogenesis. I also annotated the post-translational site for Hippo Pathway core 8 components. These rich

information will help the interpretation of the mutations detected by the current update of cancer genome. And they will be helpful for the more newly discovered mutations in the near future.

My mutation deleteriousness analyses across protein domains could also improve the way to define oncogene and tumor suppressor genes. In 2013, Vogelstein, et al classified the oncogenes and tumor suppressors based on their RM ratio or NF ratio with a 20% minimal cutoff. According to this and the functional property, only WWTR1 are more likely to be oncogenes. However, given the limited sample size and vague cutoff boundary, it cannot support the tumor suppressive function roles of hippo components in tumorigenesis based on the even distribution across tumor suppressor proteins. A deeper view for mutation function outcome prediction or even experiment should be carried on to make more cautious cutoff and criteria.

Their tissue specificity could indict the context-dependent role of Hippo Pathway activity in different tissue types, although the variation of background mutation rate for different tissue types should be excluded first. Although Hippo tumor suppressors are not frequently mutated genes across all tissue types, their expression and function can be effected through a variety of mechanisms including promoter methylation. Further exploration in cbioPortal website could be done to evaluate the total mutations rate including the RNA expression data and DNA copy number variants.

Chapter 4 Hippo Core Components Somatic Mutation and Copy Number Variant in Cancer

Genome

4.1 Introduction

Many avenues have been explored to target the growth control function of Hippo

Pathway after its discovery as organ size control pathway. There are 4 core tumor suppressors in human hippo pathway: two pairs of kinase homologues, human STK3 (MST2) and STK4

(MST1), as well as human LATS1 and LATS2. The putative oncogenes include two homologs of transcriptional co-activators YAP and WWTR1 (TAZ), and two homologs of transcriptional activators TEAD2 and TEAD4. Extensive evidences have demonstrated the tumor suppressive role of Hippo pathway 4 core tumor suppressors in controlling downstream core oncogenes in mice and human cancer (Reviewed by Johnson and Halder 2013). The mutation patterns of

Hippo core components and their correlation in cancer genome are largely unknown.

Cancer Genome Project (CGP) at the Sanger Institute in the United Kingdom and The

Cancer Genome Altas (TCGA) in the United States systematically sequenced and explored the genome of various types of tumor, especially for the intragenic point mutations. A key question for cancer genome is to identify these somatic mutations and their consequences for cancer development. These mutations are categorized into driver and passenger mutations, based on whether the alteration can confer a selective growth advantage or not. Driver mutations are oncogenic to initiate or promote tumorigenesis, so positively selected during tumorigenesis. In contrast, passenger mutations arise from incidental alterations and genome instability, and usually contribute less to tumorigenesis. Although inevitably containing passenger mutations, cancer genes are mainly defined by the functional impact and statistical significance of driver mutations.

To distinguish the functional impact of driver mutations from a large number of missense passenger mutations, various in silico algorithms have been developed, based on the evaluation of evolutionary conservation, protein structure and biochemical features of mutated amino acid residues. For the statistical significance of mutations, the significantly mutated genes could be selected based on their significant mutation rate than the background mutation rate, accounting for sequence context and gene size. For example, the random background mutations will result in an expected amount of mutations in a gene, which will be tested in the statistical null hypothesis by the amount observed mutations. One of the important bioinformatics methods, MutSig, effectively identified the true driver mutations while taking into account the probability of background mutations and possible false positives due to other factors, such as gene size and chromosome accessibility (Lawrence et al 2013). Utilizing these and many other methods, many cancer genome studies spend tremendous efforts to identify the key driver mutations and related driver cancer genes from a large number of sequenced patient genome. However, not all rare driver mutations will be captured by these efforts, contributing to the comprehensive understanding for these genomes. From a single pathway point of view, I want to address that whether Hippo Pathway core components harbor any driver mutations predicted to be deleterious, which may implicate the role of Hippo Pathway in human cancer genome.

The complexity of cancer genome mutation does not come from the independent event of a large number of different genes, but rather is the outcome of extensive genetic interaction among them. The concurrent and mutually exclusive correlations reflect the degree to which two genes or more occurs together in the same cancer patient. Concurrency as a positive correlation, suggests that there may be synergistic or redundant function between two genes or their associated pathways in terms of tumorigenesis. Some of them may be synthetic viable, leading to

the survival of tumor cells, although one single gene inactivation may lead to cell apoptosis or senescence. There are two possible mechanisms to explain mutually exclusive relationships between mutated genes. First, the sufficiently positive selection of one mutation in one gene reduces the accumulation of additional mutations in another gene during the tumor evolution.

In this study, significant concurrent and mutually-exclusive relationships were investigated among the mutations of the interactome of the Hippo core components and cancer census genes in

3,720 genomes. My unbiased screening identified a list of genes with strong associations with the core components of the Hippo pathway. The significantly mutated genes are analyzed in the patient pools with deleterious mutations from each of the Hippo pathway components along with sets of non-damaging controls. These genomic analyses shed light on mechanisms of Hippo pathways and the concurrence and mutually exclusivity with its associated genes in cancer, which could aid in the development of a cancer biomarker panels for prognoses and treatments.

4.2 Materials and Methods

4.2.1 Data source

Somatic mutation data was gathered from tumor data supplied by The Cancer Genome

Atlas (TCGA) primarily through an open sourced data portal (cBioPortal) as well as TCGA’s Data

Matrix. Selected mutational data encompassed 10 selected carcinomas - BLCA, BRCA, COAD,

HNSC, LUAD, LUSC, OV, READ, STAD, and UCEC. Data were downloaded in groups based on specific qualifications such as deleterious or non-deleterious mutations. Patients with non- somatic mutation data were occasionally utilized as controls.

4.2.2 Extract Significant Mutated Genes by Mutation Significance (MutSig)

Cancer patient genome data sets were downloaded from TCGA using patient IDs carrying the damaging or non-damaging mutations based on the analysis results from mutation pathogenicity test. MutSig (Lawrence et al., 2013) was used primarily in analyzing the damaging and non-damaging somatic mutations to locate possible gene relationships for the Hippo gene pathway. The input file requires a minimum of four data inputs: “gene” (Hugo Symbol of the mutated gene), “patient” (patient ID), “effect” (the effect of the mutation on the gene), and

“categ”. The Matlab-based program, MutSigCV, consists of four components including

“mutations.maf”, “coverage.txt”, “covariates.txt”, and “output.txt”. “mutations.maf” is the input file and “output.txt” returns the results. Coverage and covariates is provided by the Broad

Institute. Since MutSig builds a model that analyzes the mutations of each gene to identify genes that were mutated more often than expected by chance, the results would differentiate between mutations having functional impact and those without. Data sets were saved as “.txt” and analyzed with MutSigCV command. Outputted results for the deleterious and non-deleterious data of same tissue types were then compared to locate significant and deleterious mutations.

4.2.3 Mutation Relation Test by Mutational Significance in Cancer (Genome MuSiC)

Mutation Relation Test (MRT) is a function of Genome Mutational Significance in

Cancer (Genome MuSiC) that acts as a downstream analysis tool that utilizes permutation tests

in order to establish relationships among genes while also highlighting significantly mutated ones (Dees et al 2012). I use MuSiC 0.4 compatible with the Ubuntu system 10.1. The program command I used are music mutation-relation --bam-list.csv -- directory for maf-file.csv –-output- file.csv --permutations 10000. Default value is “true” for skip-non-coding Boolean. I used

Genomic MuSiC for several sets of data: unbiased Gene List composed of all genes available in

TCGA data portal and known Hippo Pathway Core 8 components interactome.

4.2.4 Visualization by Cytoscape

Cytoscape is an open-sourced platform for the dynamic visualization of complex networks with a detailed Protocol (Saito et al 2012). Cytoscape charts are used to show the significant gene relationships determined by the aforementioned tests. With high degrees of concurrency and mutually exclusivity, Cytoscape is also used for the visualization of the most significant gene pairs (lowest p-value) shown in the broadest lines.

4.3 Results

4.3.1 Identification of the most correlated genes with Hippo Core components across their interactome

To systematically identify most significant mutation relationships among the known interaction relationships of hippo pathways in cancer genome with Hippo mutations, I first collected the binding partners for each Hippo Pathway core 8 component in Entrez Gene as well as those found in four Hippo Pathway proteomic studies. Then this interactome for the Hippo

Pathway core 8 components were tested by MuSiC mutation relation tests. The results were then sorted through and selected for the most significant correlations (p<0.01) among the core Hippo components and their known associated genes (Figure 4.1).

Figure 4. 1 Mutation relationship between Hippo Pathway core components and their Interactome

Node Color: Green nodes are Hippo Pathway core 8 components. Blue Hues represent the interactive degree of mutually exclusive correlations. Yellow and Red shades represent interactive degree of concurrent correlations. Darker colored genes have more links with Core 8 genes. Node Size: Larger bubble sizes have greater number of cases containing mutation of genes. Line Color: Represents the Core 8 group in which the gene is interacting with. Line Width: Wider lines have smaller (more significant) p-values. Line Type: Solid lines show concurrent correlations, dot lines show mutually exclusivity. Green boxes are the genes mentioned in texts.

The upstream regulators of Hippo pathway were shown to have significant correlation with Hippo Pathway core components. HSP90AB1 were significantly co-mutated with STK3

(p=0.0019) and TEAD4 (p=0.0024). Molecular chaperone HSP90, encoded by this gene, suppressed deleterious effects of mutated proteomes in tumor cells (Ashworth et al., 2011).

However, the inhibition of HSP90 not only leads to depletion of LATS1/2 and reduction in catalytic activity, which promotes YAP mediated oncogenic gene expression (Huntoon et al.,

2010), but also prevents the formation of cell-cell junction and contact guidance (Raghunathan et al 2014). The mutation of both HSP90 and LATS1/2 may be the two hits of tumor suppression activity of Hippo pathway.

Similar mechanism may explain the concurrency between MPDZ and RASSF4 with

STK3. The mutations of another Hippo Pathway upstream regulator MPDZ was also concurrent with STK3 (p = 0.0066). MPDZ formed a portion of the Crumbs Polarity Complex, which reported cell density information by increasing the phosphorylation of TAZ/YAP and cytoplasmic retention (Varelas et al., 2010).

The significant interactome MRT results shows two members of the Ras associating domain family (RASSF) members being associated with its binding partner STK3. There exists a concurrency between RASSF4 and STK3 (p = 0.0021). RASSF1-6 are all tumor suppressors that can interact with MST (Steinmann et al., 2009). DNA Methylation of RASSF4 appears to be higher in patients with recurrent HNSCC (Steinmann et al., 2009). In addition, RASSF4 binds

MST1 or activated KRAS when exogenously expressed together leading to cell apoptosis

(Eckfeld et al., 2004; van der Weyden et al., 2007). The mutation of RASSF4 and STK3 may hit this growth control pathway twice, and lead to the dysregulation of this growth control pathway.

However, the mutations of same family protein RASSF2 were mutual exclusivity with TEAD4

(p = 0.0076). RASSF2 had been shown to activate and stabilize MST2 to inhibit cell growth

(Cooper et al., 2009; van der Weyden et al., 2007). Furthermore, exogenous RASSF2 induced changes coupled with altered actin polymerization and decreased RHOA expression, which led to cell death (van der Weyden et al., 2007). The oncogenic mutations of TEAD4 may be sufficient to promote the tumorigenesis without the loss of function of RASSF2, or vice versa.

The proteomic degradation components BTRC in Hippo Pathway also showed to be concurrent with LATS1 (p = 0.0073) and STK3 (p = 0.0016). β-TRCP, encoded by BTRC, which formed part of the SCF complex that bind and ubiquitinate YAP1 following its phosphorylation by LATS1 and CSNK1E to degrades YAP1 (Zhao et al., 2010). PSMD2, a subunit of the 26s subunit of proteasome, was also concurrently mutated with LATS1.

Furthermore, PSMD2 physically interacted with the Hippo core component STK4 in a proteomic study (Varjosalo et al., 2013). These concurrencies may damage the tumor suppressive function of Hippo Pathway to degrade YAP1.

RHOA (encodes Ras homolog gene family member A) and ACTB (encodes β-Actin) have significant mutually exclusive correlations with LATS1/2 and one of STK3/4. RHOA is responsible for the mediation of a multitude of morphologic and transcriptional responses by regulating the formation of F-actin stress fibers (Regué et al., 2013). F-actin subunit β-Actin encoded by ACTB is a binding partner of LATS1 (Visser-Grieve et al., 2011). The Rho mediated-actin cytoskeleton change led to an inhibition of LATS1/2 that in turn reduced the phosphorylation of YAP and WWTR1 (Lin et al., 2013). Other studies suggest its implications in the regulation of downstream YAP response to mechanical stresses (Smirnov et al., 2009).

Considering these evidences and its concurrency with Hippo tumor suppressors, it is possible that mutation on both sides may lead to the instability for the coupling of actin architecture and

Hippo signaling to sense the local environment, which may bring the unfitness of cancer cells that harbor both mutations.

NOTCH1 is mutually exclusive with WWTR1, YAP1, and LATS1. Notch pathway ligands, DELTA and JAGGED family, bind to receptors and cleaves the intracellular domains of

NOTCH to form NICD, which proceeds to regulate YAP. Notch signaling inhibition was

sufficiently compensated by YAP1 expression (Li et al., 2012). In addition, YAP1-TEAD4 up- regulates the downstream target of Jagged-1 ligand in carcinoma cells (Tschaharganeh et al.,

2013). These mutual regulations suggested that the mutations on one side may not be required for the additional mutations on the other sides in term of their shared functions.

SMAD5, co-mutated with STK3 (p = 0.0067) as well as STK4 (p = 0.005), is a receptor- regulated SMAD (R-SMAD) that is part of the BMP pathway, functioning as a link between the

Hippo and BMP pathways (Huntoon et al., 2010). WWTR1 has been implicated in the promotion of SMAD5 intracellular signaling through BMP4. WWTR1 forms a complex with

TEAD that leads to the up-regulation of BMP4 expression and subsequent phosphorylation of

SMAD5 to promote cell migration (Lai et al., 2013). Two of Wnt Pathway components APC also exhibit concurrent mutation relation with Hippo core components, including APC for

LATS1 (p=0.0193) and LATS2 (p=0.0001), CTNNB1 for LATS1 (p = 0.0046).

4.3.2 Identification of the significant cancer census genes correlated with Hippo Pathway core components

A census of human cancer summarized the genes whose mutations have been causally implicated in human cancers (Futreal et al, 2004). To investigate the correlation between Hippo

Pathway and these cancer census gene, I performed the Genomic MuSiC Mutation Relation Test for all cancer census genes and Hippo Pathway core components, and demonstrated the statistically significant (p<0.01) correlations with Hippo Pathway core components (Figure 4.2).

Mutually exclusivity as a negative correlation requires larger sample sizes of the patients carrying mutations, so I combined top ten caricnoma genome with the highest mutated frequency of the mutations in Hippo Pathway components. TP53 mutations are mutually exclusive with

most of Hippo Pathway core 8 components genes other than WWTR1, which is consistent with

Hippo Pathway core 8 genes MRT results. This is followed by PIK3CA with LATS1/2, STK4,

TEAD2/4 and CDKN2A to LATS1/2 and TEAD2/4.

Figure 4. 2 Mutation Relationships between Hippo Pathway core components and Cancer Cosmic Census Genes

The mutations of TP53 gene, encoded p53 protein as the subunit of a tetramer, mostly are loss of function mutations, and dominant negatively suppress the wild types function (Muller and

Vousden 2013). The mutually exclusive relationship between TP53 and Hippo Pathway may be explained by their functional redundancy in apoptosis regulation. LATS2-MDM2-p53 axis and

100

LATS2-ASPP1-p53 axis promote the post-mitotic G1 arrest by up-regulation of p21 for tetraploid cells, or apopotosis by up-regulation of BAX and PIG3 (Reviewed by Yabuta and

Nojima 2013). In addition, YAP-PML-p53 axis also regulates the cell proliferation and cellular senescence (Fauti et al 2013). It may also be possible that the coupling of mutant p53 and Hippo

Pathway components leads to the lethality for the tumor cells, resulting in few cells harboring both mutations. In addition, most of the p53 mutant are gain of tumor function.

The mutually exclusive relationship between the mutations of PIK3CA-LATS1 is one of the most statistically significant. PIK3CA, the catalytic domain of PIK3, is associated with the transduction of EGFR signaling via small GTPases. The blockade of PI3K signaling leads to an increase in cytoplasmic phospho-Mst1-Thr183 (Collak et al., 2012). PI3K inhibits the proapoptotic functions of MST (Yuan et al., 2010; Kim et al., 2010). Furthermore, the PI3K-PDK1 pathway mediates the inactivation of LATS as well as YAP nuclear translocation, and target gene transcription (Fan et al., 2012). Once PIK3CA is constitutively activated, LATS1/2, STK4 and TEAD2/4 mutations may not be required.

Other high degrees of mutually exclusivity include CDKN2A with LATS1/2 and

TEAD2/4. The mutations of CDKN2A (encodes p16INK4a and p14ARF) are mutually exclusive with LATS1/2 and TEAD2/4, from unbiased MRT and Cosmic Census results. The mutually exclusivity between LATS2 and CDKN2A is surprising, considering LATS2 locus 13q12.11 is physically linked to RB1 on chromosome13q14.2. This could be explained by a large degree of functional overlapping among CDKN2A, LATS1/2 and TEAD 2/4. In Rb-mediated cellular senescence, the sustained activation of Rb/p16 pathway and mitogenic signaling leads to the overall reduction of LATS1 (Takahashi et al., 2006). In mice, both LATS1-/- phenocopied and

CDKN2A-/- phenotype lead to the development of spontaneous tumors (St John et al., 1999).

101

The partial knockdown of LATS2 also suppresses portions of the Rb pathway (Tschöp et al.,

2011). Furthermore, cellular senescence from YAP-TEAD deficiency is primarily due to

Rb/p16/p53 pathways and their downstream targets CDK6 (Xie et al., 2013). CDKN2A expression is regulated by either YAP or TEAD1 detected by microarray (Zhao et al 2008). In sum, this mutation exclusivity results highlight the possible mutually dispensable manner regarding common consequences of CDKN2A and Hippo Pathway mutations.

RAF1, MSH6, FGFR1 represent the highest degree of concurrency with three, three, and two Core 8 genes respectively. RAF1 or C-Raf is a proto-oncogene. The significance of RAF1 and STK3 mutation concurrency is further supported by the binding and inhibition of MST2 kinase by RAF1. The disassociation of the RAF1-MST2 thereby promotes mitogenic signaling and increased apoptotic risk (Romano et al., 2010). The mutations of LATS2 and TEAD2 may work synergistically with oncogenic mutations of RAF1 to promote tumorigenesis. The correlations with MSH6 and FGFR1 the Hippo pathways are less well known. MSH6 (mutS homolog 6), involved in DNA damage repair, may work synergistically with LATS2 to maintain genome stability (Visser et al., 2010). FGFR1 may relate with Hippo Pathway through its ligand

FGF1, transcription target of YAP1 (Badouel and McNeill 2011).

In addition to high degrees of concurrency and mutually exclusivity, Cytoscape also allows for the visualization of the most significant gene pairs (lowest p-value) shown in the broadest lines. In this case the top interactions STK11-LATS2 have the most mutually exclusive relationship while NUMA1-TEAD2 exhibit one of the strongest concurrent relationships

(Mohseni et al., 2014). The mutation of both NUMA1 and inactivation of LATS1 lead to spindle dysfunction (Palazzo et al., 2014). Furthermore, NUMA1 binds to Pins and the resulting conformational change allows for association with Gαi that in turn interacts with Hippo core

102

associated components. A mutation in NUMA1 would therefore lead to the disruption of epithelial sheets, allowing epithelial cells to detach and induce anoikis (McCaffrey et al., 2011;

Zhao et al., 2012).

4.3.3 Unbiased screening of genes with most significant correlation with Hippo core components

In order to unbiasedly identify the genes that have the most significant correlations with

Hippo core 8 component mutations, I used Hippo core 8 component mutations as baits to screen

16385 genes across genome through Mutation Relation Tests. There is a large degree of overlapping among the mutated genes correlated with Hippo Pathway core 8 genes in the most significant (p < 0.01) gene relationships. Among them, FAM69A (Family With Sequence

Similarity 69, Member A) concurrently mutated with LATS2, STK3, STK4, and TEAD2.

CENPC1 mutations are concurrent with the ones of all the tumor suppressors as well as TEAD2.

FAM69A, CENPC1 was found to be the downstream target of either YAP or TEAD1, together with SLC6A12, which has strongest concurrency with STK3 (Zhao et al., 2008). NBPF10, one of the two neuroblastoma breakpoint family members identified through this screen, showed the most number of mutually exclusive mutation correlations, together with CDKN2A. In addition to

CDKN2A, this unbiased MRT test also validates other highly significant correlations in previous tests, such as RHOA, APC, PSMD2 and NOTCH1.

PPP1R3A (encodes protein phosphatase 1 regulatory subunit alpha) mutations are identified in this unbiased screen to be concurrent with the mutations of LATS1 (p=0.0017) and

STK3 (p=0.0033), which is consistent with the concurrency of PPP2R2A (encodes protein phosphatase 2 regulatory subunit) and STK3 mutations found in previous Hippo interactome test.

103

While PP2A dephosphorylates Raf-1 to suppress STK3, PP1A dephosphorylates WWC1 and

WWTR1 to suppress the phosphorylation and activation of LATS2 and antagonize Hippo signaling tumor suppression (Reviewed by Hata et al 2013). The oncogenic function of PP1A and PP2A mutations may require the additional loss of function mutations Hippo Pathway tumor suppressors.

NEDD4, is concurrent with LATS1 (p = 0.006), whereas the concurrency with LATS2 has a p-value of 0.005. NEDD4 E3 ubiquitin ligase binds and degrades LATS1 to subsequently increase YAP1 nuclear localization and YAP1 transcriptional activity (Salah et al., 2013). The over-expression of NEDD4 in various cancers has demonstrated proto-oncogenic properties. It seems LATS1/2 mutations are necessary for NEDD4 oncogenic activity. It also suggested that

NEDD4 has a more important role among other similar regulators for LATS1/2 degradation, such as WWP1 and NUAK1.

4.3.4 Significantly mutated genes identified in patient genome carrying with the mutations of Hippo Pathway Core Tumor Suppressors.

To further study these mutations of Hippo pathway in cancer genome, I collected genome of cancer patients who carrying Hippo pathway mutations from top ten carcinoma tissues, which have higher mutation rates for Hippo core components and larger sample sizes. I collected the patients carrying deleterious mutation and non- deleterious mutations for each of the core tumor suppressor genes of hippo pathway (LATS1/2, STK3/4) separately. Deleterious mutation includes truncation and frame shift mutations, and the mutations predicted to be deleterious by at least one method. Non-deleterious mutations include the synonymous mutations and non- damaging mutations. I compared MutSig results to extract the unique co-mutated genes within

104

Hippo Pathway damaging mutation carriers, while not identified in non-damaging carrier pools.

There is a large degree of overlapping for the significant mutated genes identified in the patient pools enriched with deleterious mutations of Hippo core tumor suppressors (Figure 4.3), which indicates the collaborative role of Hippo core tumor suppressors’ mutations in tumorigenesis.

Figure 4. 3 Genomic unbiased Screening and mutation relation partner with Hippo core Components

Up left corner shows the unbiased screen procedure. Node Color: Green nodes are Hippo Pathway core 8 components. Blue Hues represent the interactive degree of mutually exclusive correlations. Yellow and Red shades represent interactive degree of concurrent correlations. Darker colored genes have more links with Core 8 genes. Node Size: Larger bubble sizes have greater number of cases containing mutation of genes. Line Color: Represents the Core 8 group in which the gene is interacting with. Line Width: Wider lines have smaller (more significant) p- values. Line Type: Solid lines show concurrent correlations, dot lines show mutually exclusivity. Green boxes are the genes mentioned in texts.

There are three shared significant mutated genes found in patient genomes with the deleterious mutations in either three Hippo Pathway tumor suppressors: a calmodulin- and actin-

105

binding protein CALD1, zinc-figure protein ZNF330 and DNA Replication Complex GINS

Protein GINS1. CALD1 regulates cell division and size control upon its activation of CDC2, which also phosphorylated LATS1 during mitosis (Li et al 2003). In addition, LATS1 also negatively regulate CDC2/cyclin A to inhibit cell proliferation (Tao et al 1999). Moreover,

CALD1 expression was reduced by knockdown Hippo core oncogene TEAD4 (Lim et al., 2009).

These evidences suggested that CALD1 and Hippo Core components mutations may work synergistically as two hit to jeopardize the regulation of tumor cell proliferation. TAOK1 was identified as significant mutated genes in both LATS1 and LATS2 lesion patients, which are supported by the fact that TAOK1-STK3/4-LATS1/2 kinase cascade in Hippo pathway (Poon et al 2011, Boggiano et al 2011).

However, the most significantly mutated genes (with the smallest p-value) are more uniquely identified in deleterious LATS2 or LATS1 mutation MutSig tests, such as BCL9 and

KLF3. Wnt pathway regulator BCL9 in LATS2 result is consistent with the functional study that

LATS2 can inhibit the interaction between BCL9 and β-catenin (Li et al 2013). The functional hint for transcription factor KLF3 could be LATS1 downstream component YAP can bind and stabilize another Kruppel-like factors family member KLF5 to promote breast cell proliferation and survival (Zhi et al 2012).

RUNX2, known transcription factor activated by YAP1 and WWTR1, has been found to be significantly mutated in LATS1 (p=0.001, Figure 4.4) and LATS2 (p=0.02) patient pools. The mechanism of these correlation has been partially illustrated by the experimental evidence that enhancement of YAP1-TEAD oncogenic activity upon the “two hit” of Hippo pathway components by the loss of function of RASSF1 and RUNX2 (van der Weyden et al., 2012).

Similarly, upon the coupled loss of function mutations of LATS1/2 and RUNX2, YAP1 may be

106

more disposed to associating with TEAD and subsequent tumorgenesis in certain tumor genetic compositions. However, RUNX2, as a context-dependent transcription factors, has a more complicated role in various tissue types of tumor cells (Westendorf , 2006; Chimge et al., 2012).

RUNX2 directly regulates its transcription target UGCG, a sphingolipid metabolism enzyme, to promote cell survival in rodent fibroblast cells (Kilbey et al., 2010). UGCG are also significantly mutated in LATS1 deleterious mutations-specific patients, and consistent with their concurrency identified by unbiased MRT results. It is possible that the gain of function mutation of UGCG work synergistically with LATS1 loss of function mutations to promote cell survival. Similar pairs identified by both methods are LATS2-RUNDC3B, SMAD5-STK4, and PSMD5-STK3.

Figure 4. 4 Significantly mutated genes in cancer patients carrying Hippo core 4 tumor suppressor mutations

Node Color: Green nodes are Hippo Pathway core 4 tumor suppressors. Yellow and Red shades represent the interactive degree of concurrent correlations. Darker colored genes have more links with the Hippo Pathway genes. Line Color: Represents the gene was identified in the specific patient pool carrying the damaging mutations of certain Hippo pathway core tumor suppressor. Line Width: Wider lines have smaller (more significant) p-values. Line Type:

107

Solid lines connects Hippo pathway core tumor suppressors with the significantly mutated genes identified in the specific patient pools by MutSig. The number of Hippo core tumor suppressor deleterious/damaging mutation carriers are listed at the bottom left corner. 4.3.5 Significantly mutated genes identified in patient genome carrying with the mutations of Hippo Pathway Core Oncogenes.

To identify the significant mutated genes specific to the patient pools enriched with deleterious mutations Hippo Pathway core 4 oncogenes, I compared the MutSig results for the patient pools enriched with deleterious mutations with the ones with non-deleterious controls.

Compared with the tumor suppressor genes result (Figure 4.4), significant mutated genes found in the patients carrying with hippo 4 putative oncogene mutations only shared low degree of overlapping, except PSMA1 and GSTZ1 (Figure 4.5). This demonstrates that the Hippo Pathway oncogenes mutations may use various paths with other genes to contribute for tumorigenesis.

108

Figure 4. 5 Significantly mutated genes in cancer patients carrying Hippo core 4 oncogene mutations

Node Color: Green nodes are Hippo Pathway core 4 oncogenes. Yellow and Red shades represent the interactive degree of concurrent correlations. Darker colored genes have more links with the Hippo Pathway genes. Line Color: Represents the gene was identified in the specific patient pool carrying the damaging mutations of certain Hippo pathway core oncogenes. Line Width: Wider lines have smaller (more significant) p-values. Line Type: Solid lines show concurrent correlations. Slashed lines show correlations involving both MutSig and MRT results.

PSMA1 (Proteasome Subunit, Alpha Type, 1) forms proteasome, which involved in YAP degradation. Although I cannot tell the deleterious mutation for YAP and PSMA1 are either simple loss of function or oncogenic mutations, there is a possibility that the oncogenic YAP variants may accumulate and contribute more to tumorigenesis, if their degradation is reduced by the malfunction of proteasome caused by the PSMA1 mutations. Not only found in TEAD2 or

WWTR1 lesion patients, GSTZ1 (glutathione S-transferase zeta 1) were also identified in

LATS2 and STK3 lesion patients (figure 4.4). It may worth to investigate its role in Hippo pathway with a proteomic study shown ABL1 as the binding partners with both YAP1 and

GSTZ1 (Wu et al 2007). CYP51A1 and LRP2BP are two significant mutated genes identified by both MutSig and unbiased MRT. CYP51A1 are concurrently mutated with LATS1 (p=0.0364) and STK3 (p<0.0001) in unbiased screens. LRP2BP are also co-mutated with LATS2

(p=0.0017), TEAD2 (p=0.0201), and TEAD4 (p=0.0016). Their correlations with Hippo

Pathway are largely unknown.

4.4 Discussion

The discovery of the strong concurrency between Hippo Pathway core tumor suppressors and upstream regulators, such as RASSF4, TAOK1, and MPDZ, are supported by the previous experiment evidences. Considering prevalence of heterozygous loss of function mutations of

Hippo Pathway core tumor suppressors mutations, this suggests the direct two-hit model of

Hippo Pathway tumor suppressor mutations in loss of YAP-TEAD control during tumorigenesis.

109

Another two-hit model was illustrated by the loss of LATS1/2 upstream regulator, RASSF1A and the absence of the other transcription factor of YAP, RUNX2 together lead to higher levels of YAP1-TEAD association and subsequent proliferation (van der Weyden et al., 2012). The significant concurrency of RUNX2 mutations and LATS1/2 mutations I identified, may provide additional path to alter the distribution of YAP towards to TEAD to promote tumorigenesis.

Figure 4. 6 Intra-pathway mutation relationship of Hippo Pathway core components with other components

110

My multiple mutation relation tests highlight the significant concurrent relationships between

Hippo core components and protein stability components such as NEDD4, BTRC, and

PSMD2/5. The significant concurrencies for proteasome subunits PSMA1, PSMD2 and PSMD5 with Hippo core 4 tumor suppressors and 2 oncogenes suggests the role of proteasome in Hippo

Pathway function in cancer genome.

The mutation relationship between Hippo Pathway with the mutated components of other signal transduction pathway in cancer genome were also supported by the previous work on the crosstalk of Hippo Pathway with Wnt, BMP and NOTCH pathway. Previous studies have shown the inhibition of the Wnt pathway by Hippo pathway components (Figure 4.6). Following the initial phosphorylation of YAP by LATS, CSNK1E (CK1ε) will attach at the S381 site and continue the phosphorylation of nearby sites (pD). This allows the subsequent attachment of SCF leading to the degradation of YAP (Zhao et al., 2010). In addition, STK3 also binds with CK1ε to inhibit the beta-catenin degradation (Xu et al., 2014). The gene encoding CK1 ε, CSNK1E is found to be a significantly mutated gene in LATS2 patient pool. LATS2 loss of function mutations may work synergistically with oncogenic CSNK1E for the loss of inhibition of β- catenin. In addition, the mutations of CTNNB1, a beta-catenin gene, are mutually exclusive with

LATS1 mutations. The presence and interaction of YAP1 with CTNNB1 and TEAD in the nucleus initiates a pro-growth expression program. Furthermore, the mutation of CTNNB1 in the nucleus is enough to promote tumorigenesis in the Wnt pathway. In intestinal epithelial cells, an abundance of YAP in the cytoplasm represses Wnt signaling leading to loss of ISCs. This in turn leads to the degeneration of intestinal epithelium (Barry et al., 2013). APC (Figure 4.1) exhibits mutational concurrency in LATS2 patient pools (p = 0.0058). APC is also Wnt related. In the absence of Wnt signaling, APC forms a portion of the “Active Destruction Complex” (degrades

111

YAP1) also composed of CSNK1E (Bernascone et al., 2013). BCL9 binds to CTNNB1 in order to promote transcriptional activity (Figure 4.2). In addition to Wnt Pathway, components of

Notch and BMP also correlated with Hippo core components, such as SMAD5 with STK3/4,

NOTCH1 with YAP1/WWTR1. In sum, these correlations reflected the co-mutated inter- pathway relationships in human tumors, which may facilitate further understanding of detailed mechanisms that underlie the correlation. I found more and stronger correlations among tumor suppressors than oncogenes. It may illustrate more independent role of oncogenic role of Hippo

Pathway oncogenes than tumor suppressors. Along with more cancer genome sequenced, I may find more mutually exclusivity of oncogenes which is limited here by the mutations numbers of them.

112

Figure 4. 7 Schematic representation of mutation relation between Hippo Pathway and Wnt Pathway components.

This study also identified the cancer census genes that mutated with Hippo Pathway core components in concurrent manner, such as RAF1, MSH6, FGFR, and PIK3R1; or mutually exclusive manner, such as TP53, PIK3CA, CDKN2A, and BRAF for the latter. This information draw the landscape for the cooperative or separated role that Hippo Pathway play with these

113

frequently mutated cancer driver genes. In addition, multiple tests also demonstrated the significant relationships among p53-most hippo core components, TP53BP2-STK3. The multiple relationships like this may also provide the clue for the possible mechanism that how Hippo

Pathway contribute to carcinogenesis in the background of these cancer gene mutations in cancer patients. In addition to the mutual exclusivity between TP53 and Hippo Core components, concurrency correlation, ATM-YAP1, CHEK2-LATS2/STK4 which may elucidate the possible involvement of Hippo Pathway in DNA damage response network underlie those correlations.

The rescue of YAP1 by STK3 mutations has been proven to induce apoptosis during DNA damage (Cottini et al., 2014). This synthetic lethal strategy provided another perspective to intervene the Hippo pathway in cancer genome. My study illustrated the gene relationships in the cancer genome, which provides rich information to study similar lethality and viability of cells carrying Hippo pathway mutations in cancer. This study has identified the genetic alterations of

Hippo signaling pathway in cancer, highlighted the important regulations of these pathways, and shed light on the potential therapeutic strategies.

114

Chapter 5 Conclusions and Future Work

5.1 Conclusions

Within this thesis, I describe the loss of growth control function of LATS1/2 mutations identified in cancer genome utilizing the model organism Drosophila. These experimental results support a model that LATS1 and LATS2 are tumor suppressors in human cells and loss or reduction of their functions contributes to tumorigenesis. Therefore, human LATS1 and LATS2 mutations are expected to confer tissue growth advantage to drive tumor development. Previous studies have shown that LATS is a critical kinase of the Hippo pathway to phosphorylate YAP.

Indeed, the hLATS1/2 mutations decrease their ability to inhibit YAP transcriptional activity. As

YAP functions to promote tissue growth and appears to be an oncogenic protein, my results suggest that hLATS1/2 mutations contribute to tumor development due to their reduced activity to negatively regulate YAP function.

In addition, my in silico analyses have evaluated the mutations of Hippo Pathway core components in a larger scale. The mutation statistics and spectra across core 8 components have been illustrated. I also explore their mutation deleteriousness by multiple pathogeneicity evaluation tools, as well as the damaging mutation rate across Hippo Pathway core component proteins or their domains. All the mutations have been intensively curated by literature information. Some possible driver mutations have been identified. This work demonstrates the characteristic of Hippo Pathway mutations, and indirectly illustrates the role of Hippo Pathway core components in carcinogenesis.

115

The discovery of the significant correlation between Hippo Pathway core components and its associated components by genomic approaches first demonstrates the critical signaling information flow in cancer genome within Hippo Pathway. The concurrency has been identified between Hippo Pathway core tumor suppressors and their upstream regulators, such as RASSF4,

TAOK1, and MPDZ. The mutation relationship analyses also highlights the significant concurrent relationships between hippo core components and protein stability components such as NEDD4, BTRC, and PSMD2. Secondly, my work also demonstrates the important role of inter-pathway cross-talk between Hippo Pathway and Wnt Pathway in human cancer, as well as possible cross-talk with other transduction pathway. These findings lend support to the conclusion that cancerous cells generally require mutations in several pathway (Hanahan and

Weingerg 2011). Thirdly, this study also identifies the cancer census genes, which mutated with

Hippo Pathway core components in concurrent or mutually exclusive manners.

5.2 Future directions

5.2.1 Explore mutation relationship of the mutations Hippo regulators

An immediate next step is to explore mutated relationship from the perspectives of significant relationship between the Hippo Pathway-associated components. From the lesson I learned from two-hit model examples with Hippo core components, it is very likely there is a concurrent mutated pattern between Hippo Pathway-associated components. The concurrency may rely on the coordinated regulation of Hippo Pathway core components. Exclusivity may also suggest the uncoupled independency of Hippo Pathway regulators.

116

5.2.2 Identification of Hippo core components in the reciprocal significant mutated tests

Both MutSig and MRT tests identified mutated genes with significant concurrency with the mutations of Hippo core components. The reciprocal MutSig tests in patient pools carrying identified these genes could further screen these genes for the mutually concurrent genes, which are significant mutated excludes the variations introduced by the gene size, chromosome accessibility, and local background mutation rates. Since MRT is only based on statistical permutation tests to exclude the distribution of mutated genes across samples. Preliminary tests have been carried to test some of previous identified genes with experimental data support. I found out LATS2 in patients specifically carrying CHEK2 or NEFH mutations, as well as

TEAD2 in mutated NOTCH1 patients. Further tests could be performed for the most significant genes with unknown interaction with Hippo core components.

5.3 Possible experiments to test hypotheses generated in in silico work

The current mutation relationship of LATS-RUNX2 could be tested by the viability or proliferation could be tested by flow cytometry to compare the single gene mutations with both or none gene mutations. So the synergistic role in promoting cell fitness or causing cell death could be determined. The preliminary tests for loss of function of both genes could be tested by

RNA interference or gain of function, while exogenous overexpression of mutated gene for gain of function tests. In addition, the genomic editing could be used to mimic the normal expression levels and network. In Drosophila, transgenic overexpression assay or genomic editing by

CRISPR/CAS9 could be used to validate the results of the mutated site conserved between

117

human and Drosophila in vivo , while mosaic experiment to mimic the tumor environment in the background of cells carrying single gene mutations.

The molecular mechanism related with Hippo Pathway transcription targets underlies the discovered correlation could be validated first with their mRNA expression alteration in the background of Hippo Pathway lesion. The first batch of candidates could be the YAP-TEAD downstream targets CDKN2A, FAM69A, CENPC1, and SLC6A12. RT-PCR could be used to confirm their expression alteration by RNA interference or overexpression loss of function mutations of Hippo tumor suppressor. Another candidate could be RUNX2 transcriptional target

UGCG, the expression alteration of which by YAP and LATS1/2 should be tested first. Further function assay similar to RUNX2-LATS1 could be followed for concurrency pairs, while the cell apoptosis assay could be performed to validate the synthetic lethal hypothesis for those mutations suggested by the mutually exclusivity in cancer genome.

5.4 Multi-omics pathway analyses adding DNA copy number, RNA expression alterations.

To identify the significant mutations patterns of Hippo Pathway in additional to DNA mutation sequencing data, more layer of data could be acquired through control access, and analyzed for the patient pools carrying various types of hippo pathway mutations, including copy number variant and RNA expression alterations. The tools such as MEMo (Mutual Exclusivity

Modules, Ciriello 2012) to MEMo can identify the significant recurrently altered genes with point mutation, copy number variant and mRNA expression change. It also identifies the mutually exclusive patterns of gene pairs likely to be involved in the same pathway based on the

Human Reference Network (HRN Wu et al 2010). In addition, a de novo driver pathway

118

detection method, Dendrix (De novo driver exclusivity), will be implemented for patient pools with Hippo pathway lesion to validate the MEMo results and identify the rare driver mutation.

Similar to Dendrix, Multi-Dendrix (De novo driver exclusivity, Leiserson et al., 2013) could de novo explore the pathway or interaction information specific for input genome. Those tools could polish the landscape of altered signaling pathway related with Hippo Pathway.

5.5 Hippo Pathway in tumor evolution and social cell biology

Although I evaluate all the oncogenes and tumor suppressor mutations in the tumor tissue of each patient as single unit, I am also aware of the heterogeneity of both mutation pattern and cell types, as well as non-cell-autonoumous process of those mutations in tumorigenesis. My work adds to the literature in mutation data in Hippo Pathway. Further work could be done in this area using available single cell sequencing data and heterogeneity data to explore the role of hippo pathway in tumor evolution and social cell biology. The possible discovery may facilitate the in-silico identification of new components and cross-talk pathway, and in-vivo validation with Drosophila genetic tools.

119

References

Abe, Y., Ohsugi, M., Haraguchi, K., Fujimoto, J., & Yamamoto, T. (2006). LATS2–Ajuba

complex regulates γ-tubulin recruitment to centrosomes and spindle organization during

mitosis. FEBS letters, 580(3), 782-788.

Abzhanov, A., Protas, M., Grant, B. R., Grant, P. R., & Tabin, C. J. (2004). Bmp4 and

morphological variation of beaks in Darwin's finches. Science, 305(5689), 1462-1465.

Ashworth, A., Lord, C. J., & Reis-Filho, J. S. (2011). Genetic interactions in cancer progression

and treatment. Cell, 145(1), 30-38.

Avruch, J., Zhou, D., Fitamant, J., Bardeesy, N., Mou, F., & Barrufet, L. R. (2012, September).

Protein kinases of the Hippo pathway: regulation and substrates. In Seminars in cell &

developmental biology (Vol. 23, No. 7, pp. 770-784). Academic Press.

Aylon, Y., Ofir-Rosenfeld, Y., Yabuta, N., Lapi, E., Nojima, H., Lu, X., & Oren, M. (2010). The

Lats2 tumor suppressor augments p53-mediated apoptosis by promoting the nuclear

proapoptotic function of ASPP1. Genes & development, 24(21), 2420-2429.

Badouel, C., & McNeill, H. (2011). SnapShot: The hippo signaling pathway. Cell, 145(3), 484-

484

Barron, D. A., & Kagey, J. D. (2014). The role of the Hippo pathway in human disease and

tumorigenesis. Clinical and translational medicine, 3(1), 25.

Barry, E. R., & Camargo, F. D. (2013). The Hippo superhighway: signaling crossroads

converging on the Hippo/Yap pathway in stem cells and development. Current opinion in

cell biology, 25(2), 247-253.

Basu, S., Totty, N. F., Irwin, M. S., Sudol, M., & Downward, J. (2003). Akt phosphorylates the

120

Yes-associated protein, YAP, to induce interaction with 14-3-3 and attenuation of p73-

mediated apoptosis. Molecular cell, 11(1), 11-23.

Bernascone, I., & Martin-Belmonte, F. (2013). Crossroads of Wnt and Hippo in epithelial

tissues. Trends in cell biology, 23(8), 380-389.

Bi, W., Xiao, L., Jia, Y., Wu, J., Xie, Q., Ren, J., ... & Yuan, Z. (2010). c-Jun N-terminal kinase

enhances MST1-mediated pro-apoptotic signaling through phosphorylation at serine 82.

Journal of biological chemistry, 285(9), 6259-6264.

Bischof, J., Maeda, R. K., Hediger, M., Karch, F., & Basler, K. (2007). An optimized

transgenesis system for Drosophila using germ-line-specific φC31 integrases.

Proceedings of the National Academy of Sciences, 104(9), 3312-3317.

Boggiano, J. C., Vanderzalm, P. J., & Fehon, R. G. (2011). Tao-1 phosphorylates Hippo/MST

kinases to regulate the Hippo-Salvador-Warts tumor suppressor pathway. Developmental

cell, 21(5), 888-895.

Bothos, J., Tuttle, R. L., Ottey, M., Luca, F. C., & Halazonetis, T. D. (2005). Human LATS1 is a

mitotic exit network kinase. Cancer research, 65(15), 6568-6575.

Bykov, V. J., & Wiman, K. G. (2014). Mutant p53 reactivation by small molecules makes its

way to the clinic. FEBS letters.

Callus, B. A., Verhagen, A. M., & Vaux, D. L. (2006). Association of mammalian sterile twenty

kinases, Mst1 and Mst2, with hSalvador via C‐terminal coiled‐coil domains, leads to its

stabilization and phosphorylation. FEBS Journal, 273(18), 4264-4276.

Chan, E. H., Nousiainen, M., Chalamalasetty, R. B., Schäfer, A., Nigg, E. A., & Silljé, H. H.

(2005). The Ste20-like kinase Mst2 activates the human large tumor suppressor kinase

121

Lats1. Oncogene, 24(12), 2076-2086.

Chan, S. W., Lim, C. J., Guo, K., Ng, C. P., Lee, I., Hunziker, W., ... & Hong, W. (2008). A role

for TAZ in migration, invasion, and tumorigenesis of breast cancer cells. Cancer

research, 68(8), 2592-2598.

Chan, S. W., Lim, C. J., Huang, C., Chong, Y. F., Gunaratne, H. J., Hogue, K. A., ... & Hong, W.

(2011). WW domain-mediated interaction with Wbp2 is important for the oncogenic

property of TAZ. Oncogene, 30(5), 600-610.

Chen, D., Sun, Y., Wei, Y., Zhang, P., Rezaeian, A. H., Teruya-Feldstein, J., ... & Ma, L. (2012).

LIFR is a breast cancer metastasis suppressor upstream of the Hippo-YAP pathway and a

prognostic marker. Nature medicine, 18(10), 1511-1517.

Chen, L., Chan, S. W., Zhang, X., Walsh, M., Lim, C. J., Hong, W., & Song, H. (2010).

Structural basis of YAP recognition by TEAD4 in the hippo pathway. Genes &

development, 24(3), 290-300.

Chimge, N. O., & Frenkel, B. (2013). The RUNX family in breast cancer: relationships with

estrogen signaling. Oncogene, 32(17), 2121-2130.

Cinar, B., Fang, P. K., Lutchman, M., Di Vizio, D., Adam, R. M., Pavlova, N., ... & Freeman, M.

R. (2007). The pro‐apoptotic kinase Mst1 and its caspase cleavage products are direct

inhibitors of Akt1. The EMBO journal, 26(21), 4523-4534.

Cline, M. S., Smoot, M., Cerami, E., Kuchinsky, A., Landys, N., Workman, C., ... & Bader, G.

D. (2007). Integration of biological networks and gene expression data using Cytoscape.

Nature protocols, 2(10), 2366-2382.

Cottini, F., Hideshima, T., Xu, C., Sattler, M., Dori, M., Agnelli, L., ... & Tonon, G. (2014).

122

Rescue of Hippo coactivator YAP1 triggers DNA damage-induced apoptosis in

hematological cancers. Nature medicine, 20(6), 599-606.

Collak, F. K., Yagiz, K., Luthringer, D. J., Erkaya, B., & Cinar, B. (2012). Threonine-120

phosphorylation regulated by phosphoinositide-3-kinase/Akt and mammalian target of

rapamycin pathway signaling limits the antitumor activity of mammalian sterile 20-like

kinase 1. Journal of Biological Chemistry, 287(28), 23698-23709.

Cooke, I. E., Shelling, A. N., Le Meuth, V. G., Charnock, F. M. L., & Ganesan, T. S. (1996).

Allele loss on chromosome arm 6q and fine mapping of the region at 6q27 in epithelial

ovarian cancer. Genes Chromosomes and Cancer, 15(4), 223-233.

Cooper, W. N., Hesson, L. B., Matallanas, D., Dallol, A., von Kriegsheim, A., Ward, R., ... &

Latif, F. (2009). RASSF2 associates with and stabilizes the proapoptotic kinase

MST2. Oncogene, 28(33), 2988-2998.

Das Thakur, M., Feng, Y., Jagannathan, R., Seppa, M. J., Skeath, J. B., & Longmore, G. D.

(2010). Ajuba LIM proteins are negative regulators of the Hippo signaling pathway.

Current Biology, 20(7), 657-662.

Dees, N. D., Zhang, Q., Kandoth, C., Wendl, M. C., Schierding, W., Koboldt, D. C., ... & Ding,

L. (2012). MuSiC: identifying mutational significance in cancer genomes. Genome

research, 22(8), 1589-1598.

Densham, R. M., O'Neill, E., Munro, J., König, I., Anderson, K., Kolch, W., & Olson, M. F.

(2009). MST kinases monitor actin cytoskeletal integrity and signal via c-Jun N-terminal

kinase stress-activated kinase to regulate p21Waf1/Cip1 stability. Molecular and cellular

123

biology, 29(24), 6380-6390.

Dephoure, N., Zhou, C., Villén, J., Beausoleil, S. A., Bakalarski, C. E., Elledge, S. J., & Gygi, S.

P. (2008). A quantitative atlas of mitotic phosphorylation. Proceedings of the National

Academy of Sciences, 105(31), 10762-10767.

Dong, J., Feldmann, G., Huang, J., Wu, S., Zhang, N., Comerford, S. A., ... & Pan, D. (2007).

Elucidation of a Universal Size-Control Mechanism in Drosophila and

Mammals. Cell, 130(6), 1120-1133.

Duning, K., Rosenbusch, D., Schlüter, M. A., Tian, Y., Kunzelmann, K., Meyer, N., ... & Weide,

T. (2010). Polycystin-2 activity is controlled by transcriptional coactivator with PDZ

binding motif and PALS1-associated tight junction protein. Journal of biological

chemistry, 285(44), 33584-33588.

Eckfeld, K., Hesson, L., Vos, M. D., Bieche, I., Latif, F., & Clark, G. J. (2004). RASSF4/AD037

is a potential ras effector/tumor suppressor of the RASSF family. Cancer research,

64(23), 8688-8693.

Ellis, M. J., Ding, L., Shen, D., Luo, J., Suman, V. J., Wallis, J. W., ... & Watson, M. (2012).

Whole-genome analysis informs breast cancer response to aromatase inhibition. Nature,

486(7403), 353-360.

Espanel, X., & Sudol, M. (2001). Yes-associated protein and p53-binding protein-2 interact

through their WW and SH3 domains. Journal of Biological Chemistry, 276(17), 14514-

14523.

Fan, R., Kim, N. G., & Gumbiner, B. M. (2013). Regulation of Hippo pathway by mitogenic

growth factors via phosphoinositide 3-kinase and phosphoinositide-dependent kinase-

124

1. Proceedings of the National Academy of Sciences, 110(7), 2569-2574.

Fausti, F., Di Agostino, S., Cioce, M., Bielli, P., Sette, C., Pandolfi, P. P., ... & Blandino, G.

(2013). ATM kinase enables the functional axis of YAP, PML and p53 to ameliorate loss

of Werner protein-mediated oncogenic senescence. Cell Death & Differentiation, 20(11),

1498-1509.

Fei, X., Yan-lin, W., Jiao-jiao, C., Si-chen, D., Lei, D., Nan, J., ... & Jin, Z. (2013). Mammalian

sterile 20-like kinase 1/2 inhibit Wnt/beta-catenin signaling pathway by directly binding

Casein kinase 1 epsilon. Biochemical Journal.

Fujii, H., Zhou, W., & Gabrielson, E. (1996). Detection of frequent allelic loss of 6q23–q25. 2 in

microdissected human breast cancer tissues. Genes, Chromosomes and Cancer, 16(1),

35-39.

Garraway, L. A., & Lander, E. S. (2013). Lessons from the cancer genome. Cell, 153(1), 17-37.

Gavva, N. R., Gavva, R., Ermekova, K., Sudol, M., & Shen, C. K. J. (1997). Interaction of WW

domains with hematopoietic transcription factor p45/NF-E2 and RNA polymerase II.

Journal of Biological Chemistry, 272(39), 24105-24108.

Goranov, A. I., & Amon, A. (2010). Growth and division—not a one-way road. Current opinion

in cell biology, 22(6), 795-800.

Goulev, Y., Fauny, J. D., Gonzalez-Marti, B., Flagiello, D., Silber, J., & Zider, A. (2008).

SCALLOPED Interacts with YORKIE, the Nuclear Effector of the Hippo Tumor-

Suppressor Pathway in Drosophila. Current Biology, 18(6), 435-441.

Gupta, M., Kogut, P., Davis, F. J., Belaguli, N. S., Schwartz, R. J., & Gupta, M. P. (2001).

125

Physical interaction between the MADS box of serum response factor and the

TEA/ATTS DNA-binding domain of transcription enhancer factor-1. Journal of

Biological Chemistry, 276(13), 10413-10422.

Hanks, S. K., & Hunter, T. (1995). Protein kinases 6. The eukaryotic protein kinase superfamily:

kinase (catalytic) domain structure and classification. The FASEB journal, 9(8), 576-596.

Hansen, L. L., Jensen, L. L., Dimitrakakis, C., Michalas, S., Gilbert, F., Barber, H. R., ... &

Arzimanoglou, I. I. (2002). Allelic imbalance in selected chromosomal regions in ovarian

cancer. Cancer genetics and cytogenetics, 139(1), 1-8.

Hao, Y., Chun, A., Cheung, K., Rashidi, B., & Yang, X. (2008). Tumor suppressor LATS1 is a

negative regulator of oncogene YAP. Journal of biological chemistry, 283(9), 5496-5509.

Hata, Y., Timalsina, S., & Maimaiti, S. (2013). Okadaic Acid: A Tool to Study the Hippo

Pathway. Marine drugs, 11(3), 896-902.

Harvey KF, Pfleger CM, Hariharan IK. 2003. The Drosophila Mst ortholog, hippo, restricts

growth and cell proliferation and promotes apoptosis. Cell 114:457-467

Harvey, K. F., Zhang, X., & Thomas, D. M. (2013). The Hippo pathway and human cancer.

Nature Reviews Cancer, 13(4), 246-257.

Hilman, D., & Gat, U. (2011). The evolutionary history of YAP and the Hippo/YAP pathway.

Molecular biology and evolution, 28(8), 2403-2417.

Hirota, T., Morisaki, T., Nishiyama, Y., Marumoto, T., Tada, K., Hara, T., ... & Saya, H. (2000).

Zyxin, a regulator of actin filament assembly, targets the mitotic apparatus by interacting

with h-warts/LATS1 tumor suppressor. The Journal of cell biology, 149(5), 1073-1086.

Hori, T., Takaori-Kondo, A., Kamikubo, Y., & Uchiyama, T. (2000). Molecular cloning of a

126

novel human protein kinase, kpm, that is homologous to warts/lats, a Drosophila tumor

suppressor. Oncogene, 19(27).

Hornbeck, P. V., Kornhauser, J. M., Tkachev, S., Zhang, B., Skrzypek, E., Murray, B., ... &

Sullivan, M. (2012). PhosphoSitePlus: a comprehensive resource for investigating the

structure and function of experimentally determined post-translational modifications in

man and mouse. Nucleic acids research, 40(D1), D261-D270.

Howell, M., Borchers, C., & Milgram, S. L. (2004). Heterogeneous nuclear ribonuclear protein

U associates with YAP and regulates its co-activation of Bax transcription. Journal of

Biological Chemistry, 279(25), 26300-26306.

Huang, H., Sun, F., Owen, D. M., Li, W., Chen, Y., Gale, M., & Ye, J. (2007). Hepatitis C virus

production by human hepatocytes dependent on assembly and secretion of very low-

density lipoproteins. Proceedings of the National Academy of Sciences, 104(14), 5848-

5853.

Huang, J., Wu, S., Barrera, J., Matthews, K., & Pan, D. (2005). The Hippo Signaling Pathway

Coordinately Regulates Cell Proliferation and Apoptosis by Inactivating Yorkie, the

Drosophila Homolog of YAP. Cell, 122(3), 421-434.

Humbert, N., Navaratnam, N., Augert, A., Da Costa, M., Martien, S., Wang, J., ... & Bernard, D.

(2010). Regulation of ploidy and senescence by the AMPK‐related kinase NUAK1. The

EMBO journal, 29(2), 376-386.

Huntoon, C. J., Nye, M. D., Geng, L., Peterson, K. L., Flatten, K. S., Haluska, P., ... & Karnitz,

L. M. (2010). Heat shock protein 90 inhibition depletes LATS1 and LATS2, two

regulators of the mammalian hippo tumor suppressor pathway. Cancer research, 70(21),

127

8642-8650.

Iida, S. I., Hirota, T., Morisaki, T., Marumoto, T., Hara, T., Kuninaka, S., ... & Saya, H. (2004).

Tumor suppressor WARTS ensures genomic integrity by regulating both mitotic

progression and G1 tetraploidy checkpoint function. Oncogene, 23(31), 5266-5274.

Jang, S. W., Yang, S. J., Srinivasan, S., & Ye, K. (2007). Akt phosphorylates MstI and prevents

its proteolytic activation, blocking FOXO3 phosphorylation and nuclear translocation.

Journal of biological chemistry, 282(42), 30836-30844.

Jia, J., Zhang, W., Wang, B., Trinko, R., & Jiang, J. (2003). The Drosophila Ste20 family kinase

dMST functions as a tumor suppressor by restricting cell proliferation and promoting

apoptosis. Genes & development, 17(20), 2514-2519.

Jiang, Z., Li, X., Hu, J., Zhou, W., Jiang, Y., Li, G., & Lu, D. (2006). Promoter

hypermethylation-mediated down-regulation of LATS1 and LATS2 in human

astrocytoma. Neuroscience research, 56(4), 450-458.

Johnson R., Halder G. (2013). The two faces of Hippo: targeting the Hippo pathway for

regenerative medicine and cancer treatment. Nature Reviews Drug Discovery. Advanced

Online.

Justice, R. W., Zilian, O., Woods, D. F., Noll, M., & Bryant, P. J. (1995). The Drosophila tumor

suppressor gene warts encodes a homolog of human myotonic dystrophy kinase and is

required for the control of cell shape and proliferation. Genes & development, 9(5), 534-

546.

Kanai, F., Marignani, P. A., Sarbassova, D., Yagi, R., Hall, R. A., Donowitz, M., ... & Yaffe, M.

128

B. (2000). TAZ: a novel transcriptional co‐activator regulated by interactions with 14‐3‐3

and PDZ domain proteins. The EMBO journal, 19(24), 6778-6791.

Kaneko, K. J., & DePamphilis, M. L. (1998). Regulation of gene expression at the beginning of

mammalian development and the TEAD family of transcription factors. Developmental

genetics, 22(1), 43-55.

Kango-Singh, M., Nolo, R., Tao, C., Verstreken, P., Hiesinger, P. R., Bellen, H. J., & Halder, G.

(2002). Shar-pei mediates cell proliferation arrest during imaginal disc growth in

Drosophila. Development, 129(24), 5719-5730.

Kilbey, A., Terry, A., Jenkins, A., Borland, G., Zhang, Q., Wakelam, M. J., ... & Neil, J. C.

(2010). Runx regulation of sphingolipid metabolism and survival signaling. Cancer

research, 70(14), 5860-5869.

Kim, D., Shu, S., Coppola, M. D., Kaneko, S., Yuan, Z. Q., & Cheng, J. Q. (2010). Regulation of

Proapoptotic Mammalian ste20–Like Kinase MST2 by the IGF1-Akt Pathway. PloS one,

5(3), e9616.

Knight, C. G., Patel, M. N., Azevedo, R. B., & Leroi, A. M. (2002). A novel mode of

ecdysozoan growth in Caenorhabditis elegans. Evolution & development, 4(1), 16-27.

Komuro, A., Nagai, M., Navin, N. E., & Sudol, M. (2003). WW domain-containing protein YAP

associates with ErbB-4 and acts as a co-transcriptional activator for the carboxyl-terminal

fragment of ErbB-4 that translocates to the nucleus. Journal of Biological Chemistry,

278(35), 33334-33341.

Kosaka, Y., Mimori, K., Tanaka, F., Inoue, H., Watanabe, M., & Mori, M. (2007). Clinical

significance of the loss of MATS1 mRNA expression in colorectal cancer. International

journal of oncology, 31(2), 333-338.

129

Kuninaka, S., Nomura, M., Hirota, T., Iida, S. I., Hara, T., Honda, S., ... & Saya, H. (2005). The

tumor suppressor WARTS activates the Omi/HtrA2-dependent pathway of cell death.

Oncogene, 24(34), 5287-5298.

Lai, D., & Yang, X. (2013). BMP4 is a novel transcriptional target and mediator of mammary

cell migration downstream of the Hippo pathway component TAZ.Cellular

signalling, 25(8), 1720-1728.

Lai, Z. C., Wei, X., Shimizu, T., Ramos, E., Rohrbaugh, M., Nikolaidis, N., ... & Li, Y. (2005).

Control of cell proliferation and apoptosis by mob as tumor suppressor, mats. Cell,

120(5), 675-685.

Lapi, E., Di Agostino, S., Donzelli, S., Gal, H., Domany, E., Rechavi, G., ... & Blandino, G.

(2008). PML, YAP, and p73 are components of a proapoptotic autoregulatory feedback

loop. Molecular cell, 32(6), 803-814.

Laplante, M., & Sabatini, D. M. (2012). mTOR signaling in growth control and disease. Cell,

149(2), 274-293.

Lawrence, M. S., Stojanov, P., Polak, P., Kryukov, G. V., Cibulskis, K., Sivachenko, A., ... &

Onofrio, R. (2013). Mutational heterogeneity in cancer and the search for new cancer-

associated genes. Nature.

Lee, E. Y., To, H., Shew, J. Y., Bookstein, R., Scully, P., & Lee, W. H. (1988). Inactivation of

the retinoblastoma susceptibility gene in human breast cancers. Science, 241(4862), 218-

221.

Lee, J. H., Kavanagh, J. J., Wildrick, D. M., Wharton, J. T., & Blick, M. (1990). Frequent loss of

heterozygosity on chromosomes 6q, 11, and 17 in human ovarian carcinomas. Cancer

130

research, 50(9), 2724-2728.

Lee, J. H., Kim, T. S., Yang, T. H., Koo, B. K., Oh, S. P., Lee, K. P., ... & Lim, D. S. (2008). A

crucial role of WW45 in developing epithelial tissues in the mouse. The EMBO journal,

27(8), 1231-1242.

Lee, K. K., & Yonehara, S. (2001). Phosphorylation and dimerization regulate

nucleocytoplasmic shuttling of mammalian STE20-like kinase (MST). Journal of

Biological Chemistry, 277(14), 12351-12358.

Lei, Q. Y., Zhang, H., Zhao, B., Zha, Z. Y., Bai, F., Pei, X. H., ... & Guan, K. L. (2008). TAZ

promotes cell proliferation and epithelial-mesenchymal transition and is inhibited by the

hippo pathway. Molecular and cellular biology, 28(7), 2426-2436.

Levy, D., Adamovich, Y., Reuven, N., & Shaul, Y. (2008). Yap1 phosphorylation by c-Abl is a

critical step in selective activation of proapoptotic genes in response to DNA damage.

Molecular cell, 29(3), 350-361.

Li, J., Chen, X., Ding, X., Cheng, Y., Zhao, B., Lai, Z. C., ... & Wang, C. Y. (2013). LATS2

Suppresses Oncogenic Wnt Signaling by Disrupting β-Catenin/BCL9 Interaction. Cell

reports, 5(6), 1650-1663.

Li, Y., Hibbs, M. A., Gard, A. L., Shylo, N. A., & Yun, K. (2012). Genome‐wide analysis of

N1ICD/RBPJ targets in vivo reveals direct transcriptional regulation of Wnt, SHH, and

hippo pathway effectors by Notch1. Stem Cells, 30(4), 741-752.

Li, Y., Pei, J., Xia, H., Ke, H., Wang, H., & Tao, W. (2003). Lats2, a putative tumor suppressor,

inhibits G1/S transition. Oncogene, 22(28), 4398-4405.

Li, Y., Wessels, D., Wang, T., Lin, J. L., Soll, D. R., & Lin, J. J. (2003). Regulation of

131

caldesmon activity by Cdc2 kinase plays an important role in maintaining membrane

cortex integrity during cell division. Cellular and Molecular Life Sciences CMLS, 60(1),

198-211.

Li, Z., Zhao, B., Wang, P., Chen, F., Dong, Z., Yang, H., ... & Xu, Y. (2010). Structural insights

into the YAP and TEAD complex. Genes & development, 24(3), 235-240.

Liekens, A. M., De Knijf, J., Daelemans, W., Goethals, B., De Rijk, P., & Del-Favero, J. (2011).

BioGraph: unsupervised biomedical knowledge discovery via automated hypothesis

generation. Genome Biol, 12(6), R57.

Liekens, A. M., De Knijf, J., Daelemans, W., Goethals, B., De Rijk, P., & Del-Favero, J. (2011).

BioGraph: unsupervised biomedical knowledge discovery via automated hypothesis

generation. Genome Biol, 12(6), R57.

Lim, B., Park, J. L., Kim, H. J., Park, Y. K., Kim, J. H., Sohn, H. A., ... & Kim, S. Y. (2013).

Integrative genomics analysis reveals the multi-level dysregulation and oncogenic

characteristics of TEAD4 in gastric cancer.Carcinogenesis, bgt409.

Lin, J. I., Poon, C. L., & Harvey, K. F. (2013). The Hippo Size Control Pathway--Ever

Expanding. Science signaling, 6(259), pe4.

Liu, C. Y., Zha, Z. Y., Zhou, X., Zhang, H., Huang, W., Zhao, D., ... & Guan, K. L. (2010). The

hippo tumor pathway promotes TAZ degradation by phosphorylating a phosphodegron

and recruiting the SCFβ-TrCP E3 ligase. Journal of biological chemistry, 285(48),

37159-37169.

Liu, W., Wu, J., Xiao, L., Bai, Y., Qu, A., Zheng, Z., & Yuan, Z. (2012). Regulation of neuronal

cell death by c-Abl-Hippo/MST2 signaling pathway. PloS one, 7(5), e36562.

132

Lu L, Li Y, Kim SM, Bossuyt W, Liu P, Qiu Q, Wang Y, Halder G, Finegold MJ, Lee JS,

Johnson RL. 2010. Hippo signaling is a potent in vivo growth and tumor suppressor

pathway in the mammalian liver. Proc Natl Acad Sci USA 107:1437-1442.

Macleod, K. (2000). Tumor suppressor genes. Current opinion in genetics & development, 10(1),

81-93.

Mazurenko, N. A. T. A. L. I. A., Attaleb, M. O. H. A. M. E. D., Gritsko, T., Semjonova, L. J. U.

D. M. I. L. A., Pavlova, L. A. R. I. S. A., Sakharova, O. L. G. A., & Kisseljov, F. J. O. D.

O. R. (1999). High resolution mapping of chromosome 6 deletions in cervical cancer.

Oncology reports, 6(4), 859-922.

McCaffrey, L. M., & Macara, I. G. (2011). Epithelial organization, cell polarity and

tumorigenesis. Trends in cell biology, 21(12), 727-735.

McPherson, J. P., Tamblyn, L., Elia, A., Migon, E., Shehabeldin, A., Matysiak‐Zablocki, E., ... &

Hakem, R. (2004). Lats2/Kpm is required for embryonic development, proliferation

control and genomic integrity. The EMBO journal, 23(18), 3677-3688.

Minoo, P., Zlobec, I., Baker, K., Tornillo, L., Terracciano, L., Jass, J. R., & Lugli, A. (2007).

Prognostic significance of mammalian sterile20-like kinase 1 in colorectal cancer.

Modern pathology, 20(3), 331-338.

Mohler, P. J., Kreda, S. M., Boucher, R. C., Sudol, M., Stutts, M. J., & Milgram, S. L. (1999).

Yes-associated protein 65 localizes p62c-Yes to the apical compartment of airway

epithelia by association with EBP50. The Journal of cell biology, 147(4), 879-890.

Mohseni, M., Sun, J., Lau, A., Curtis, S., Goldsmith, J., Fox, V. L., ... & Camargo, F. D. (2014).

A genetic screen identifies an LKB1–MARK signalling axis controlling the Hippo–YAP

133

pathway. Nature cell biology, 16(1), 108-117.

Morisaki, T., Hirota, T., Iida, S. I., Marumoto, T., Hara, T., Nishiyama, Y., ... & Saya, H. (2002).

WARTS tumor suppressor is phosphorylated by Cdc2/cyclin B at spindle poles during

mitosis. FEBS letters, 529(2), 319-324

Muller, P. A., & Vousden, K. H. (2013). p53 mutations in cancer. Nature cell biology, 15(1), 2-8.

Murakami, H., Mizuno, T., Taniguchi, T., Fujii, M., Ishiguro, F., Fukui, T., ... & Sekido, Y.

(2011). LATS2 is a tumor suppressor gene of malignant mesothelioma. Cancer research,

71(3), 873-883.

Murakami, M., Tominaga, J., Makita, R., Uchijima, Y., Kurihara, Y., Nakagawa, O., ... &

Kurihara, H. (2006). Transcriptional activity of Pax3 is co-activated by TAZ.

Biochemical and biophysical research communications, 339(2), 533-539.

Nishio, M., Hamada, K., Kawahara, K., Sasaki, M., Noguchi, F., Chiba, S., ... & Suzuki, A.

(2012). Cancer susceptibility and embryonic lethality in Mob1a/1b double-mutant mice.

The Journal of clinical investigation, 122(12), 4505.

Noviello, C., Courjal, F., & Theillet, C. (1996). Loss of heterozygosity on the long arm of

chromosome 6 in breast cancer: possibly four regions of deletion. Clinical cancer

research, 2(9), 1601-1606.

Okada, N., Yabuta, N., Suzuki, H., Aylon, Y., Oren, M., & Nojima, H. (2011). A novel Chk1/2–

Lats2–14-3-3 signaling pathway regulates P-body formation in response to UV damage.

Journal of cell science, 124(1), 57-67.

Ota, M., & Sasaki, H. (2008). Mammalian Tead proteins regulate cell proliferation and contact

inhibition as transcriptional mediators of Hippo signaling. Development, 135(24), 4059-

134

4069.

Oudhoff, M. J., Freeman, S. A., Couzens, A. L., Antignano, F., Kuznetsova, E., Min, P. H., ... &

Zaph, C. (2013). Control of the Hippo pathway by Set7-dependent methylation of Yap.

Developmental cell, 26(2), 188-194.

Palazzo, Luca, et al. "ATM controls proper mitotic spindle structure." Cell Cycle 13 (2014):

1091-100.

Pan, D. (2010). The hippo signaling pathway in development and cancer. Developmental cell,

19(4), 491-505.

Pantalacci, S., Tapon, N., & Léopold, P. (2003). The Salvador partner Hippo promotes apoptosis

and cell-cycle exit in Drosophila. Nature cell biology, 5(10), 921-927.

Paramasivam, M., Sarkeshik, A., Yates, J. R., Fernandes, M. J., & McCollum, D. (2011).

Angiomotin family proteins are novel activators of the LATS2 kinase tumor suppressor.

Molecular biology of the cell, 22(19), 3725-3733.

Park, K. S., Whitsett, J. A., Di Palma, T., Hong, J. H., Yaffe, M. B., & Zannini, M. (2004). TAZ

interacts with TTF-1 and regulates expression of surfactant protein-C. Journal of

Biological Chemistry, 279(17), 17384-17390.

Pobbati, A. V., & Hong, W. (2013). Emerging roles of TEAD transcription factors and its

coactivators in cancers. Cancer biology & therapy, 14(5), 390

Poon, C. L., Lin, J. I., Zhang, X., & Harvey, K. F. (2011). The sterile 20-like kinase Tao-1

controls tissue growth by regulating the Salvador-Warts-Hippo pathway. Developmental

cell, 21(5), 896-906.

135

Piersma, T., & Drent, J. (2003). Phenotypic flexibility and the evolution of organismal design.

Trends in Ecology & Evolution, 18(5), 228-233

Powzaniuk, M., McElwee-Witmer, S., Vogel, R. L., Hayami, T., Rutledge, S. J., Chen, F., ... &

Bai, C. (2004). The LATS2/KPM tumor suppressor is a negative regulator of the

androgen receptor. Molecular Endocrinology, 18(8), 2011-2023..

Praskova, M., Khoklatchev, A., Ortiz-Vega, S., & Avruch, J. (2004). Regulation of the MST1

kinase by autophosphorylation, by the growth inhibitory proteins, RASSF1 and NORE1,

and by Ras. Biochem. J, 381, 453-462.

Qiao, M., Wang, Y., Xu, X., Lu, J., Dong, Y., Tao, W., ... & Pardee, A. B. (2010). Mst1 is an

interacting protein that mediates PHLPPs' induced apoptosis. Molecular cell, 38(4), 512-

523.

Raghunathan, V. K., Dreier, B., Morgan, J. T., Tuyen, B. C., Rose, B. W., Reilly, C. M., ... &

Murphy, C. J. (2014). Involvement of YAP, TAZ and HSP90 in Contact Guidance and

Intercellular Junction Formation in Corneal Epithelial Cells. PloS one, 9(10), e109811.

Regué, L., Mou, F., & Avruch, J. (2013). G protein‐coupled receptors engage the mammalian

Hippo pathway through F‐actin. Bioessays, 35(5), 430-435.

Renner, O., Fominaya, J., Alonso, S., Blanco-Aparicio, C., Leal, J. F., & Carnero, A. (2007).

Mst1, RanBP2 and eIF4G are new markers for in vivo PI3K activation in murine and

human prostate. Carcinogenesis, 28(7), 1418-1425.

Romano, D., Matallanas, D., Weitsman, G., Preisinger, C., Ng, T., & Kolch, W. (2010).

Proapoptotic kinase MST2 coordinates signaling crosstalk between RASSF1A, Raf-1,

and Akt. Cancer research, 70(3), 1195-1203.

136

Roth, R. B., Hevezi, P., Lee, J., Willhite, D., Lechner, S. M., Foster, A. C., & Zlotnik, A. (2006).

Gene expression analyses reveal molecular relationships among 20 regions of the human

CNS. Neurogenetics, 7(2), 67-80.

Roy, D., Sarkar, S., & Felty, Q. (2006). Levels of IL-1 beta control stimulatory/inhibitory growth

of cancer cells. Front Biosci, 11, 889-898.

Salah, Z., Cohen, S., Itzhaki, E., & Aqeilan, R. I. (2013). NEDD4 E3 ligase inhibits the activity

of the Hippo pathway by targeting LATS1 for degradation. Cell Cycle, 12(24), 3817-

3823.

Salah, Z., Melino, G., & Aqeilan, R. I. (2011). Negative regulation of the Hippo pathway by E3

ubiquitin ligase ITCH is sufficient to promote tumorigenicity. Cancer research, 71(5),

2010-2020.

Saito R, Smoot ME, Ono K, Ruscheinski J, Wang PL, Lotia S, Pico AR, Bader GD, Ideker T. A

travel guide to Cytoscape plugins. Nature Methods. 2012 Nov;9 (11):1069-76.

Sasaki, H., Kawano, O., Endo, K., Suzuki, E., Yukiue, H., Kobayashi, Y., ... & Fujii, Y. (2007).

Human MOB1 Expression in Non–Small-Cell Lung Cancer. Clinical lung cancer, 8(4),

273-276.

Sato, T., Saito, H., Morita, R., Koi, S., Lee, J. H., & Nakamura, Y. (1991). Allelotype of human

ovarian cancer. Cancer research, 51(19), 5118-5122.

Seidel, C., Schagdarsurengin, U., Blümke, K., Würl, P., Pfeifer, G. P., Hauptmann, S., ... &

Dammann, R. (2007). Frequent hypermethylation of MST1 and MST2 in soft tissue

sarcoma. Molecular carcinogenesis, 46(10), 865-871.

137

Shimomura, T., Miyamura, N., Hata, S., Miura, R., Hirayama, J., & Nishina, H. (2014). The

PDZ-binding motif of Yes-associated protein is required for its co-activation of TEAD-

mediated CTGF transcription and oncogenic cell transforming activity.

Biochemical and biophysical research communications, 443(3), 917-923.

Siegel, R., Ma, J., Zou, Z., & Jemal, A. (2014). Cancer statistics, 2014. CA: a cancer journal for

clinicians, 64(1), 9-29.

Smirnov, D. A., Morley, M., Shin, E., Spielman, R. S., & Cheung, V. G. (2009). Genetic analysis

of radiation-induced changes in human gene expression. Nature, 459(7246), 587-591.

Song, H., Mak, K. K., Topol, L., Yun, K., Hu, J., Garrett, L., ... & Yang, Y. (2010). Mammalian

Mst1 and Mst2 kinases play essential roles in organ size control and tumor suppression.

Proceedings of the National Academy of Sciences, 107(4), 1431-1436.

St John MA, Tao W, Fei X, Fukumoto R, Carcangiu ML, Brownstein DG, Parlow AF, McGrath

J, Xu T. 1999. Mice deficient of Lats1 develop soft-tissue sarcomas, ovarian tumours and

pituitary dysfunction. Nat Genet 21:182-186.

Staley, B. K., & Irvine, K. D. (2012). Hippo signaling in Drosophila: recent advances and

insights. Developmental Dynamics, 241(1), 3-15.

Steinmann, K., Sandner, A., Schagdarsurengin, U., & Dammann, R. H. (2009). Frequent

promoter hypermethylation of tumor-related genes in head and neck squamous cell

carcinoma. Oncology reports, 22(6), 1519-1526.

Strano, S., Munarriz, E., Rossi, M., Castagnoli, L., Shaul, Y., Sacchi, A., ... & Blandino, G.

(2001). Physical interaction with Yes-associated protein enhances p73 transcriptional

138

activity. Journal of Biological Chemistry, 276(18), 15164-15173.

Sudol, M., Shields, D. C., & Farooq, A. (2012, September). Structures of YAP protein domains

reveal promising targets for development of new cancer drugs. In Seminars in cell &

developmental biology (Vol. 23, No. 7, pp. 827-833). Academic Press.

Suzuki, H., Yabuta, N., Okada, N., Torigata, K., Aylon, Y., Oren, M., & Nojima, H. (2013).

Lats2 phosphorylates p21/CDKN1A after UV irradiation and regulates apoptosis. Journal

of cell science, 126(19), 4358-4368.

Takahashi, A., Ohtani, N., Yamakoshi, K., Iida, S. I., Tahara, H., Nakayama, K., ... & Hara, E.

(2006). Mitogenic signalling and the p16INK4a–Rb pathway cooperate to enforce

irreversible cellular senescence. Nature cell biology, 8(11), 1291-1297.

Takahashi, Y., Miyoshi, Y., Takahata, C., Irahara, N., Taguchi, T., Tamaki, Y., & Noguchi, S.

(2005). Down-regulation of LATS1 and LATS2 mRNA expression by promoter

hypermethylation and its association with biologically aggressive phenotype in human

breast cancers. Clinical Cancer Research, 11(4), 1380-1385.

Tamura, K., Peterson, D., Peterson, N., Stecher, G., Nei, M., & Kumar, S. (2011). MEGA5:

molecular evolutionary genetics analysis using maximum likelihood, evolutionary

distance, and maximum parsimony methods. Molecular biology and evolution, 28(10),

2731-2739.

Tao, W., Zhang, S., Turenchalk, G. S., Stewart, R. A., St, M. A., Chen, W., & Xu, T. (1999).

Human homologue of the Drosophila melanogaster lats tumour suppressor modulates

CDC2 activity. Nature genetics, 21(2), 177-181.

Tapon, N., Harvey, K. F., Bell, D. W., Wahrer, D. C., Schiripo, T. A., Haber, D. A., &

Hariharan, I. K. (2002). salvador Promotes Both Cell Cycle Exit and Apoptosis

139

in Drosophila and Is Mutated in Human Cancer Cell Lines. Cell, 110(4), 467-

478.

Theile, M., Seitz, S., Arnold, W., Jandrig, B., Frege, R., Schlag, P. M., ... & Scherneck, S.

(1996). A defined chromosome 6q fragment (at D6S310) harbors a putative tumor

suppressor gene for breast cancer. Oncogene, 13(4), 677-685.

Tian, W., Yu, J., Tomchick, D. R., Pan, D., & Luo, X. (2010). Structural and functional analysis

of the YAP-binding domain of human TEAD2. Proceedings of the National Academy of

Sciences, 107(16), 7293-7298.

Toji, S., Yabuta, N., Hosomi, T., Nishihara, S., Kobayashi, T., Suzuki, S., ... & Nojima, H.

(2004). The centrosomal protein Lats2 is a phosphorylation target of Aurora‐A kinase.

Genes to Cells, 9(5), 383-397.

Tomlinson, V., Gudmundsdottir, K., Luong, P., Leung, K. Y., Knebel, A., & Basu, S. (2010).

JNK phosphorylates Yes-associated protein (YAP) to regulate apoptosis. Cell death &

disease, 1(2), e29.

Tschaharganeh, D. F., Chen, X., Latzko, P., Malz, M., Gaida, M. M., Felix, K., ... & Breuhahn,

K. (2013). Yes-associated protein up-regulates Jagged-1 and activates the Notch pathway

in human hepatocellular carcinoma.Gastroenterology, 144(7), 1530-1542.

Tschöp, K., Conery, A. R., Litovchick, L., DeCaprio, J. A., Settleman, J., Harlow, E., & Dyson,

N. (2011). A kinase shRNA screen links LATS2 and the pRB tumor suppressor. Genes &

development, 25(8), 814-830.

Udan, R. S., Kango-Singh, M., Nolo, R., Tao, C., & Halder, G. (2003). Hippo promotes

proliferation arrest and apoptosis in the Salvador/Warts pathway. Nature cell biology,

140

5(10), 914-920.

Ueyama, T., Zhu, C., Valenzuela, Y. M., Suzow, J. G., & Stewart, A. F. (2000). Identification of

the functional domain in the transcription factor RTEF-1 that mediates α1-adrenergic

signaling in hypertrophied cardiac myocytes. Journal of Biological Chemistry, 275(23),

17476-17480. van der Weyden, L., & Adams, D. J. (2007). The Ras-association domain family (RASSF)

members and their role in human tumourigenesis. Biochimica et Biophysica Acta (BBA)-

Reviews on Cancer, 1776(1), 58-85. van der Weyden, L., Papaspyropoulos, A., Poulogiannis, G., Rust, A. G., Rashid, M., Adams, D.

J., ... & O'Neill, E. (2012). Loss of rassf1a synergizes with deregulated runx2 signaling in

tumorigenesis. Cancer research, 72(15), 3817-3827.

Varelas, X. (2014). The Hippo pathway effectors TAZ and YAP in development, homeostasis

and disease. Development, 141(8), 1614-1626.

Varelas, X., Samavarchi-Tehrani, P., Narimatsu, M., Weiss, A., Cockburn, K., Larsen, B. G., ...

& Wrana, J. L. (2010). The Crumbs complex couples cell density sensing to Hippo-

dependent control of the TGF-β-SMAD pathway.Developmental cell, 19(6), 831-844.

Varjosalo, M., Sacco, R., Stukalov, A., van Drogen, A., Planyavsky, M., Hauri, S., ... & Superti-

Furga, G. (2013). Interlaboratory reproducibility of large-scale human protein-complex

analysis by standardized AP-MS. Nature methods, 10(4), 307-314.

Visser, S., & Yang, X. (2010). LATS tumor suppressor: a new governor of cellular homeostasis.

Cell Cycle, 9(19), 3922-3933.

141

Visser, S., & Yang, X. (2010). Identification of LATS transcriptional targets in HeLa cells using

whole human genome oligonucleotide microarray. Gene,449(1), 22-29.

Visser-Grieve, S., Zhou, Z., She, Y. M., Huang, H., Cyr, T. D., Xu, T., & Yang, X. (2011).

LATS1 tumor suppressor is a novel actin-binding protein and negative regulator of actin

polymerization. Cell research, 21(10), 1513.

Vogelstein, B., Papadopoulos, N., Velculescu, V. E., Zhou, S., Diaz, L. A., & Kinzler, K. W.

(2013). Cancer genome landscapes. Science, 339(6127), 1546-1558.

Wang, H. P., & Rogler, C. E. (1988). Deletions in human chromosome arms 11p and 13q in

primary hepatocellular carcinomas. Cytogenetic and Genome Research, 48(2), 72-78.

Wang, J., Huo, K., Ma, L., Tang, L., Li, D., Huang, X., ... & Yang, X. (2011). Toward an

understanding of the protein interaction network of the human liver.Molecular systems

biology, 7(1).

Wang, W., Huang, J., Wang, X., Yuan, J., Li, X., Feng, L., ... & Chen, J. (2012). PTPN14 is

required for the density-dependent control of YAP1. Genes & development, 26(17),

1959-1971.

Wei, X., Shimizu, T., & Lai, Z. C. (2007). Mob as tumor suppressor is activated by Hippo kinase

for growth inhibition in Drosophila. The EMBO journal, 26(7), 1772-1781.

Westendorf, J. J. (2006). Transcriptional co-repressors of Runx2. Journal of cellular

biochemistry, 98(1), 54-64.

Wu, C., Ma, M. H., Brown, K. R., Geisler, M., Li, L., Tzeng, E., ... & Li, S. S. C. (2007).

Systematic identification of SH3 domain‐mediated human protein–protein interactions

by peptide array target screening. Proteomics, 7(11), 1775-1785.

142

Wu, S., Huang, J., Dong, J., & Pan, D. (2003). hippo Encodes a Ste-20 Family Protein

Kinase that Restricts Cell Proliferation and Promotes Apoptosis in Conjunction with

salvador and warts. Cell, 114(4), 445-456.

Xiao, L., Chen, Y., Ji, M., & Dong, J. (2011). KIBRA regulates Hippo signaling activity via

interactions with large tumor suppressor kinases. Journal of biological chemistry,

286(10), 7788-7796.

Xie, Q., Chen, J., Feng, H., Peng, S., Adams, U., Bai, Y., ... & Yuan, Z. (2013). YAP/TEAD–

Mediated Transcription Controls Cellular Senescence. Cancer research, 73(12), 3615-

3624.

Xu, J., Yin, Z., Cao, S., Gao, W., Liu, L., Yin, Y., ... & Shu, Y. (2013). Systematic review and

meta-analysis on the association between IL-1B polymorphisms and cancer risk. PloS

one, 8(5), e63654.

Xu, M. Z., Yao, T. J., Lee, N. P., Ng, I. O., Chan, Y. T., Zender, L., ... & Luk, J. M. (2009). Yes

associated protein is an independent prognostic marker in hepatocellular carcinoma.

Cancer, 115(19), 4576-4585.

Xu, T., Wang, W., Zhang, S., Stewart, R. A., & Yu, W. (1995). Identifying tumor suppressors in

genetic mosaics: the Drosophila lats gene encodes a putative protein kinase.

Development, 121(4), 1053-1063.

Yabuta, N., Mukai, S., Okada, N., Aylon, Y., & Nojima, H. (2011). The tumor suppressor Lats2

is pivotal in Aurora A and Aurora B signaling during mitosis. Cell Cycle, 10(16), 2724-

2736.

143

Yabuta, N., & Nojima, H. (2013). Hippo in Cell Cycle and Mitosis. In The Hippo Signaling

Pathway and Cancer (pp. 199-221). Springer New York.

Yagi, R., Kohn, M. J., Karavanova, I., Kaneko, K. J., Vullhorst, D., DePamphilis, M. L., &

Buonanno, A. (2007). Transcription factor TEAD4 specifies the trophectoderm lineage at

the beginning of mammalian development. Development, 134(21), 3827-3836.

Yang, S., Zhang, L., Liu, M., Chong, R., Ding, S. J., Chen, Y., & Dong, J. (2013). CDK1

Phosphorylation of YAP Promotes Mitotic Defects and Cell Motility and Is Essential for

Neoplastic Transformation. Cancer research, 73(22), 6722-6733.

Yang, X., Yu, K., Hao, Y., Li, D. M., Stewart, R., Insogna, K. L., & Xu, T. (2004). LATS1

tumour suppressor affects cytokinesis by inhibiting LIMK1. Nature cell biology, 6(7),

609-617.

Ye, X., Nikolaidis, N., Nei, M., & Lai, Z. C. (2009). Evolution of the mob Gene Family. The

Open Cell Signaling Journal, 1, 1-11.

Yin, F., Yu, J., Zheng, Y., Chen, Q., Zhang, N., & Pan, D. (2013). Spatial organization of Hippo

signaling at the plasma membrane mediated by the tumor suppressor Merlin/NF2. Cell,

154(6), 1342-1355.

Yuan, Z., Kim, D., Shu, S., Wu, J., Guo, J., Xiao, L., ... & Cheng, J. Q. (2010). Phosphoinositide

3-kinase/Akt inhibits MST1-mediated pro-apoptotic signaling through phosphorylation of

threonine 120. Journal of biological chemistry, 285(6), 3815-3824.

Zhang, H., Liu, C. Y., Zha, Z. Y., Zhao, B., Yao, J., Zhao, S., ... & Guan, K. L. (2009). TEAD

transcription factors mediate the function of TAZ in cell growth and epithelial-

mesenchymal transition. Journal of biological chemistry, 284(20), 13355-13362.

Zhang, L., Iyer, J., Chowdhury, A., Ji, M., Xiao, L., Yang, S., ... & Dong, J. (2012). KIBRA

144

regulates aurora kinase activity and is required for precise chromosome alignment during

mitosis. Journal of Biological Chemistry, 287(41), 34069-34077.

Zhang, X., George, J., Deb, S., Degoutin, J. L., Takano, E. A., Fox, S. B., ... & Harvey, K. F.

(2011). The Hippo pathway transcriptional co-activator, YAP, is an ovarian cancer

oncogene. Oncogene, 30(25), 2810-2822.

Zhao, B., Li, L., Wang, L., Wang, C. Y., Yu, J., & Guan, K. L. (2012). Cell detachment activates

the Hippo pathway via cytoskeleton reorganization to induce anoikis. Genes &

development, 26(1), 54-68.

Zhao B, Ye X, Yu J, Li L, Li W, Li S, Yu J, Lin JD, Wang CY, Chinnaiyan AM, Lai ZC, Guan

KL. 2008. TEAD mediates YAP-dependent gene induction and growth control. Genes

Dev 22:1962-1971.

Zhao B. Tumaneng K. Guan KL. 2011. The Hippo pathway in organ size control, tissue

regeneration and stem cell self-renewal. Nat Cell Biol 13:877-883.

Zhao, B., Kim, J., Ye, X., Lai, Z. C., & Guan, K. L. (2009). Both TEAD-binding and WW

domains are required for the growth stimulation and oncogenic transformation activity of

yes-associated protein. Cancer research, 69(3), 1089-1098.

Zhao, B., Li, L., Tumaneng, K., Wang, C. Y., & Guan, K. L. (2010). A coordinated

phosphorylation by Lats and CK1 regulates YAP stability through SCFβ-TRCP. Genes &

development, 24(1), 72-85.

Zhao, B., Wei, X., Li, W., Udan, R. S., Yang, Q., Kim, J., ... & Guan, K. L. (2007). Inactivation

of YAP oncoprotein by the Hippo pathway is involved in cell contact inhibition and

tissue growth control. Genes & development, 21(21), 2747-2761.

145

Zhao, B., Ye, X., Yu, J., Li, L., Li, W., Li, S., ... & Guan, K. L. (2008). TEAD mediates YAP-

dependent gene induction and growth control. Genes & development, 22(14), 1962-1971.

Zhao, D., Zhi, X., Zhou, Z., & Chen, C. (2012). TAZ antagonizes the WWP1-mediated KLF5

degradation and promotes breast cell proliferation and tumorigenesis. Carcinogenesis,

33(1), 59-67.

Zhi, X., Zhao, D., Zhou, Z., Liu, R., & Chen, C. (2012). YAP promotes breast cell proliferation

and survival partially through stabilizing the KLF5 transcription factor. The American

journal of pathology, 180(6), 2452-2461.

Zhou, D., Conrad, C., Xia, F., Park, J. S., Payer, B., Yin, Y., ... & Bardeesy, N. (2009). Mst1 and

Mst2 maintain hepatocyte quiescence and suppress hepatocellular carcinoma

development through inactivation of the Yap1 oncogene. Cancer cell, 16(5), 425-438.

Zhou, Z., Hao, Y., Liu, N., Raptis, L., Tsao, M. S., & Yang, X. (2011). TAZ is a novel oncogene

in non-small cell lung cancer. Oncogene, 30(18), 2181-2186.

146

Curriculum Vitae

Tian Yu

Education

Ph.D. Molecular, Cellular and Integrative Biosciences 2008-2014

Huck Institutes of the Life Sciences, the Pennsylvania State University

B.S. B i o l o g y , Nanjing Agricultural University, China 2 0 0 4 - 2 0 0 8

Publications

Yu T, Bachman JF, Lai ZC. (2013) Evidence for a Tumor Suppressor Role for the Large Tumor

Suppressor genes LATS1 and LATS2 in human cancer. Genetics 195:1193-1196.

Yu T, Bachman JF, Lai ZC. (2013). Mutation spectra and pathogenicity analyses of LATS1/2 mutations in cancer genome. In Prep.

Yu T, Chang Y, Bachman JF, Lai ZC. Integrative analyses of Hippo pathway components in human cancer genome. In Prep.

Teaching Experience

BIOL 240W Function and Development of Organisms laboratory 2009-2013

BIOL 416 Biology of Cancer 2011

BIOL 230W Molecular and Cells laboratory 2009

Awards

J. Ben and Helen D. Hill Memorial Fund Award, Penn State University 2013, 2010

Graduate Student Travel Award, Penn State 2013, 2014