Targeting the DEK oncogene in head and neck squamous cell carcinoma: functional and transcriptional consequences

A dissertation submitted to the Graduate School of the University of Cincinnati in partial fulfillment of the requirements to the degree of

Doctor of Philosophy (Ph.D.)

in the Department of Cancer and Cell Biology of the College of Medicine March 2015

by

Allie Kate Adams

B.S. The Ohio State University, 2009

Dissertation Committee: Susanne I. Wells, Ph.D. (Chair) Keith A. Casper, M.D. Peter J. Stambrook, Ph.D. Ronald R. Waclaw, Ph.D. Susan E. Waltz, Ph.D. Kathryn A. Wikenheiser-Brokamp, M.D., Ph.D.

Abstract

Head and neck squamous cell carcinoma (HNSCC) is one of the most common malignancies worldwide with over 50,000 new cases in the United States each year. For many years tobacco and alcohol use were the main etiological factors; however, it is now widely accepted that human papillomavirus (HPV) infection accounts for at least one-quarter of all

HNSCCs. HPV+ and HPV- HNSCCs are studied as separate diseases as their prognosis, treatment, and molecular signatures are distinct. Five-year survival rates of HNSCC hover around 40-50%, and novel therapeutic targets and biomarkers are necessary to improve patient outcomes.

Here, we investigate the DEK oncogene and its function in regulating HNSCC development and signaling. DEK is overexpressed in many cancer types, with roles in molecular processes such as transcription, DNA repair, and replication, as well as phenotypes such as apoptosis, senescence, and proliferation. DEK had never been previously studied in this tumor type; therefore, our studies began with clinical specimens to examine DEK expression patterns in primary HNSCC tissue. We find that DEK is highly overexpressed in all subsets of HNSCC, independent of HPV status, stage, or patient demographics. Furthermore, an established transgenic mouse model, crossed with Dek-/- mice, examined the in vivo importance of DEK in

HNSCC development in a 4-NQO model system. Transgenic Dek-/- have improved survival, fewer tumors, and reduced epidermal proliferation compared to Dek+/+ counterparts. Similarly, in vitro systems of HPV+ and HPV- cell lines demonstrated DEK loss reduced cell number and proliferation. Finally, ΔNp63 was identified as a partial mediated of this DEK-dependent cell growth.

We next aimed to identify additional candidates as mediators of DEK-dependent functions. DEK is known to regulate transcription, as either a co-activator or a co-repressor, but the global impact of DEK loss on gene transcription is unknown. Therefore, RNA-Sequencing

i was performed on HPV+ and HPV- HNSCC lines, either proficient or deficient for DEK, to assess the transcriptional consequences of DEK loss. From this data IRAK1 was identified as a candidate down-regulated in the absence of DEK. IRAK1 is a driver of inflammatory signaling, downstream of toll-like and interleukin-1 receptors, and activates many well-known signaling cascades including NF-κB and MAPK. These RNA-Seq results were validated by qRT-PCR and western blot analysis. IRAK1 expression was assessed by mining The Cancer Genome Atlas, which revealed IRAK1 is overexpressed in HNSCC. As IRAK1 has never previously been characterized in HNSCC, experiments were designed to identify the function of IRAK1 in this system. IRAK1 loss reduced downstream TRAF6-ubiquitination and signaling and increased apoptosis, with either shRNA or chemical inhibition. Lastly, DEK and IRAK1 cooperated to regulate ERK1/2 signaling, but independently contributed to HNSCC survival. In summary, these findings implicate two novel oncogenes as mediators in HNSCC growth and survival and that individual or combined targeting of DEK and IRAK1 may be therapeutically beneficial.

ii

iii Preface

The following work in this dissertation is either in preparation for, or previously published, in the following peer-reviewed journals:

1. Adams AK, Wise-Draper TM, Wells, SI. Human Papillomavirus induced transformation in cervical and head and neck cancers. Review for Cancers. 2014 Sep 15;6(3):1793-820. doi: 10.3390/cancers6031793. PMID: 25226287. 2. Adams AK, Hallenbeck GE, Casper KA, Patil YJ, Wilson KM, Kimple RJ, Lambert PF, Witte DP, Xiao W, Gillison ML, Wikenheiser-Brokamp KA, Wise-Draper TM, Wells SI. DEK promotes HPV-positive and –negative head and neck cancer cell proliferation. Oncogene. 2014 Mar 10. doi: 10.1038/onc.2014.15 [Epub ahead of print]. PMID: 24608431. 3. Adams AK, Bolanos LC, Dexheimer PJ, Karns RA, Aronow BJ, Komurov K, Jegga A, Casper KA, Patil YJ, Wilson KM, Starczynowski DT, Wells SI. IRAK1 promotes HNSCC survival through a DEK-regulated network. Manuscript in preparation.

iv Acknowledgements

During the course of my dissertation studies, I have been privileged to collaborate with numerous research scientists and clinicians without whom my work would not have been possible. Firstly, I would like to thank my mentor Dr. Susanne (Susa) Wells at Cincinnati

Children’s Hospital Medical Center. Susa has provided invaluable advice and guidance over the years, both at a scientific and personal level. She has always allowed me to be highly independent in my work and has encouraged me to pursue the experiments I was the most passionate about. I will always be grateful for all of her support and am thankful she asked me to join the Wells lab.

I would also like to thank my thesis committee, Dr. Keith Casper, Dr. Peter Stambrook,

Dr. Ronald Waclaw, Dr. Susan Waltz, and Dr. Kathryn Wikenheiser-Brokamp for their suggestions and feedback on my projects. They have challenged me as a scientist and were an integral part of this work inside and outside of my thesis meetings. Additionally, I thank Dr. Keith

Casper and his colleagues at the University of Cincinnati Department of Otolaryngology for collaborating with our lab on the receipt of primary specimens for our studies. I believe without these tissues our work would not have been nearly as impactful--few graduates have this opportunity and I am eternally grateful for the chance to work with you. I also want to acknowledge Dr. Daniel Starczynowski and his lab for helping me with the IRAK1 story. Without their advice (and reagents) we could not have pursued my final project.

To the Wells lab (past and present): thank you for your advice during the highs and lows of research. I have formed lasting friendships with many of you and my time in this lab has been full of entertainment and laughter. Many of you have been secondary mentors (Elizabeth

Hoskins and Lisa Privette Vinnedge, specifically) and helped me navigate the world that is science.

I dedicate this work to my parents, Randall and Susan Varner, who have provided unconditional love and support these past twenty-eight years. In times of struggle they have

v always been there to listen and have encouraged me to pursue my dreams. My achievements thus far have been possible because of their help, sacrifices, and passion to see their daughter succeed. Thank you for everything!

Finally, to my husband Joshua Adams—words will never be enough to express how thankful I am for all of your emotional support over 5 years of graduate school. You have been my biggest cheerleader and helped me keep the end goal in mind when I struggled with the difficulties of grad school. We met on our first day at Ohio State: two science nerds that have successfully navigated the completion of their degrees, moving to Cincinnati, surviving my qualifying exam, getting married, adopting our pup, Blue, and buying our first home. It has undoubtedly been the greatest journey of my life and I am so glad to have spent it with you.

vi

Table of Contents Abstract ...... i Preface ...... iv Acknowledgements ...... v List of Figures Tables and ...... ix

Chapter 1: Human Papillomavirus Induced Transformation in Cervical and Head

Neck Cancers ...... 1 Abstract ...... 2 Introduction ...... 3 HPV biology ...... 5 Clinical characteristics of HPV+ cancers ...... 9

Prognostic comparison and staging ...... 13 Mutational Treatment analyses variations ...... 2116 HPV Vaccines integration offer ...... protection ...... -­‐ from HPV associated cancers...... 2419 Gene expression in HPV+ cancers ...... 25 Conclusions ...... 28

Chapter 2: Novel oncogenes in HNSCC: an introduction to DEK and IRAK1 ...... 3 0

Part 1. The DEK oncogene is a multifunctional protein ...... 31

Biochemistry of DEK...... 31 Dual roles for DEK in transcription ...... 32 DEK as a chaperone histone ...... 33 Part DEK 2. loss IRAK1 abrogates is a the central, DNA upstream damage response...... hub ...... of inflammatory signaling ...... 3834 DEK functions as an oncogene ...... 34

IRAK1 signaling cascade ...... 38 IRAK1 gene structure ...... 38 Chapter 3: DEK promotes HPV positive and negative head and neck cancer cell Contributions of IRAK1 to cancer ...... 39 proliferation ...... 41 Abstract ...... 42 Introduction ...... 43 Results ...... 46 in vivo

Dek knockout mice exhibit attenuated HNSCC development ...... 46 Dek loss suppresses E7-­‐dependent proliferation in the tongue epidermis ...... 46 ΔDEK protein expression is up-­‐regulated in human HNSCC regardless of HPV status ...... 47 DiscussionDEK expression ...... stimulates ...... growth ...... of HPV positive ...... and ...... negative human HNSCC tumor cells.. 5148 Materials Np63 protein and Methods expression ...... is ...... regulated by ...... DEK and is ...... functionally important ...... 5449

Chapter 4: IRAK1 promotes HNSCC survival through a DEK-­‐regulated network ...... 69

Abstract ...... 70 Introduction ...... 71

vii Results ...... 73

Profiling the DEK-­‐dependent transcriptome in HNSCC...... 73 DEK regulates IRAK1 mRNA and protein levels...... 74 IRAK1 is over-­‐expressed in primary HNSCC...... 74 DiscussionIRAK1 loss ...... increases ...... apoptosis in HNSCC...... 7775 Materials IRAK1 and & DEK Methods ...... independently ...... regulate ...... HNSCC cellular survival...... 8076

Chapter 5: Discussion and Future Directions ...... 105

Functions of DEK and IRAK1 in HNSCC ...... 106 Functional and therapeutic avenues for future studies ...... 109 Summary...... 111

References ...... 112

viii List of Figures and Tables

Chapter 1: Human Papillomavirus Induced Transformation in Cervical and Head and Neck Cancer

Table 1.1 Summary of clinical characteristics 12 Table 1.2 Staging criteria and prognosis for SCC 15 Table 1.3 Mutations identified by whole exome sequencing 23

Chapter 2: Novel oncogenes in HNSCC: DEK and IRAK1

Figure 2.1 DEK structure and identified domains 37 Figure 2.2 IRAK1 is upstream of many oncogenic 40 signaling cascades

Chapter 3: DEK promotes HPV positive and negative head and neck cancer cell proliferation

Figure 3.1 Dek loss protects from HNSCC tumor promotion in vivo 60 Figure 3.2 Dek loss attenuated E7-driver proliferation in tongue epidermis 62 Table 3.1 Clinical Characteristics of primary human tumor samples 63 Figure 3.3 DEK protein expression is universally up-regulated in 64 human HNSCC Figure 3.4 DEK knockdown attenuates the proliferation of HPV positive 65 and negative HNSCC cell lines Figure 3.5 DEK promotes cellular growth by up-regulating ΔNp63 67

Chapter 4: IRAK1 promotes HNSCC survival through a DEK-regulated network

Figure 4.1 Profiling the DEK-dependent transcriptome in HNSCC 85 Figure 4.2 DEK regulates IRAK1 mRNA and protein levels 86 Figure 4.3 IRAK1 is overexpressed in primary HNSCC 87 Figure 4.4 IRAK1 loss increases apoptosis in HNSCC 88 Figure 4.5 IRAK1 and DEK depletion cooperate to increase apoptosis 90 Table S4.1 TF binding site analysis: up-regulated genes 92 Table S4.2 TF binding site analysis: down-regulated genes 94 Table S4.3 TF binding site analysis: oppositely-regulated genes 96 Supp. Figure S4.1 Network analysis heat map 97 Supp. Figure S4.2 Network 1 98 Supp. Figure S4.3 Network 2 99 Supp. Figure S4.4 Network 3 100 Supp. Figure S4.5 Network 4 101 Supp. Figure S4.6 Network 5 102 Supp. Figure S4.7 IRAK1 knockdown cell-cycle profiles 103 Supp. Figure S4.8 IRAK1 overexpression growth curves 104

ix Chapter 1: Human Papillomavirus Induced Transformation in Cervical and Head and Neck Cancers

Allie K. Adams1, Trisha M. Wise-Draper2, and Susanne I. Wells1*

1 Cancer and Blood Diseases Institute, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, 45229, USA 2 Division of Hematology/Oncology, University of Cincinnati Medical Center, University of Cincinnati, Cincinnati, OH, 45229, USA

Published in Cancers. 2014 Sep 15; 6(3):1793-820. doi: 10.3390/cancers6031793. PMID: 25226287

1 Abstract

Human papillomavirus (HPV) is one of the most widely publicized and researched pathogenic DNA viruses. For decades, HPV research has focused on transforming viral activities in cervical cancer. During the past 15 years, however, HPV has also emerged as a major etiological agent in cancers of the head and neck, in particular squamous cell carcinoma. Even with significant strides achieved towards the screening and treatment of cervical cancer, and preventive vaccines, cervical cancer remains the leading cause of cancer-associated deaths for women in developing countries. Furthermore, routine screens are not available for those at risk of head and neck cancer. The current expectation is that HPV vaccination will prevent not only cervical, but also head and neck cancers. In order to determine if previous cervical cancer models for HPV infection and transformation are directly applicable to head and neck cancer, clinical and molecular disease aspects must be carefully compared. In this review, we briefly discuss the cervical and head and neck cancer literature to highlight clinical and genomic commonalities.

Differences in prognosis, staging and treatment, as well as comparisons of mutational profiles, viral integration patterns, and alterations in gene expression will be addressed.

2 Introduction

Human papillomaviruses (HPVs) comprise a large family of viruses, 170 of which are now sequenced and are extensively referenced on the Papillomavirus Episteme1, 2. The high risk

(HR) mucosal HPV types, including the HPV16 prototype, are a subgroup of the alpha papillomaviruses that share the ability to cause cancer in their human host3. These DNA viruses have been extensively investigated for decades for their ability to subvert cellular mechanisms of growth control. In fact, HPV was first implicated in cancer biology approximately forty years ago by Harald zur Hausen, who was awarded the Nobel Prize in Physiology or Medicine in 2008 for the discovery that HPV causes cervical cancer4, 5. Since then, HPV has also been associated with other anogenital cancers including vulvar, penile, and anal, largely due to sexual transmission. More recently, high risk HPV has been clearly detected in head and neck cancers

(HNCs), particularly in the oropharynx, making it an additional etiological factor for a disease previously ascribed most commonly to tobacco and alcohol consumption. Both cervical and head and neck malignancies have poor outcomes when diagnosed at late stages and there is a dire need for clinically effective screening methods for detection of early stage disease and for the identification of novel therapeutic targets.

Cervical cancer in the United States has been on a swift decline due to the general implementation of Papanicolaou (Pap) screening for the detection of pre-malignant lesions of the cervix. However, the expense of treating HPV-related infections is a significant burden to society. Approximately $8 billion was spent for the year 2010, with the majority focused on screening and follow-up visits, and with just over 10% of the total expense on the treatment of cervical cancer6. Many lives are saved in return on a national level, but cervical cancer remains a major cause of death in developing countries where vaccination and screening programs are unavailable. Furthermore, similar screens are not currently available to detect early HPV positive (or negative) head and neck cancers. Clinicians must rely on surrogate markers, such

3 as the cyclin-dependent kinase inhibitor p16 (INK4a), to determine HPV status in the tumor.

HPV negative (HPV–) HNCs have been steadily declining in correlation with decreased tobacco use, while HPV positive (HPV+) head and neck cancers are on a steep, upward trajectory especially in younger populations, and are estimated to surpass cervical cancer rates in the

U.S. by the year 20207. The increasing incidence of HPV+ HNCs is linked to increasing oral HPV exposure via oral sex in younger age groups, and the total number of oral sex and open- mouthed (“French”) kissing partners8. As for cervical cancer, the risk of HPV infection also increases with greater numbers of sexual partners. Given the prospect of a national epidemic of

HPV+ HNCs, it is now critical to test the possibility that protection from cervical cancer through

HPV vaccination will also be protective against HNC. For many years, the study of HPV in cancer biology has been focused on cervical cancers. With a large upward shift in the number of

HPV positive HNCs it will be imperative to adapt the paradigms of HPV function in cervical cancer to HNC wherever possible.

Almost all cervical and at least one-quarter of head and neck cancers share the presence of

HPV genomic DNA and expression of the viral oncogenes. A majority of both cervical and HPV+

HNCs are squamous cell carcinomas (SCCs) due to keratinocyte-specific viral tropism.

Advances in the prevention and treatment of cervical cancer may be an excellent starting point for battling HPV positive HNC, but biological distinctions between these cancer types exist and must be clearly defined in order to apply this knowledge more broadly. For example, HPV types are much more restricted in head and neck cancer when compared to cervical cancer, mostly due to HPV-16 infection varying in its distribution (i.e. prevalence) predominantly as a function of HNC site (e.g., tonsillar vs. oral cavity) 9-11. Furthermore, cervical cancers have shown a dependence on estrogen signaling, whereas hormonal contributions have not been identified in

HNC12-16. Here, we discuss the clinical and molecular commonalities and differences between

HPV+ cervical and head and neck squamous cell carcinomas (HNSCCs).

4 HPV biology

Human papillomavirus is an episomal, double-stranded DNA virus. Almost 200 known types have been identified, only some of which contribute to malignancies. The majority of HPV types infect the nonmucosal, cutaneous epithelium3. The minority of HPV types that infect mucosal tissues and the genital tract are sub-divided into high risk and low risk categories based on their ability to transform the host cell. HPV16, 18, 31, and 45 are examples of high-risk types and are known to cause cervical neoplasias and tumors, while HPV6 and 11 belong to the low-risk subtypes that induce genital warts and respiratory papillomas 17. There are also HPV16 variants that are classified based on sequence alterations in E6 and are detected in different regions across the globe and are named correspondingly. These include: European (EUR), Asian

American (AA), Asian (As), African-1 (AFR1), and African-2 (AFR2) variants18-20. These variants appear to exhibit differences in their relative contributions to cervical malignancies, with Asian

American variants preferentially associated with high-grade CIN lesions and cervical cancer21-25.

Research in HNSCC on HPV16 variants has been more limited, although recent studies implicate certain variants, such as R10G, to be more frequent in HNSCC than in cervical cancer26, 27. New variants have been identified in HNSCC, but most are shared with those already identified9.

The HPV genome is circular with dual promoters that encode two separate groups of viral proteins: the early genes (E1, E2, E4, E5, E6, E7, E8) and the late genes (L1, L2). Some of the early HPV genes are essential for maintaining the viral replicative cycle, while the late HPV genes encode the major (L1) and minor (L2) capsid proteins 28. E1 and E2 are primarily involved in transcription and replication: E1 functions as the viral DNA helicase and E2 as a transcriptional activator and repressor that also complexes with E1 as a critical component of the HPV replisome 29-32. The transforming properties of HPV are a result of E5, E6, and E7 activities, which are discussed in further detail below. The HPV E4 protein is less well

5 characterized, but several studies implicate E4 in virion release via its association with keratin filaments33, 34.

The epithelium is comprised of proliferating and differentiated keratinocytes, which are the host cell for HPV infection and replication, respectively. Keratinocytes that transition into malignant cells via immortalization and transformation initiated by the presence of high-risk HPV frequently result in squamous cell carcinoma (SCC). Transmission of HPV relies on microwounds or abrasions in the epidermis for HPV to gain access to stem and/or proliferating cells in the basal epithelium. Once there, it will utilize the host cell replication machinery to initiate viral DNA replication35. Our understanding of the initial HPV infection process including viral attachment and entry remains incomplete, but interactions with heparan sulfate on the cell surface have been implicated36-42. The HPV life cycle is highly organized, geographically as well as functionally, around the cellular differentiation program43, 44. Under normal circumstances, cell division in the basal cell layer regenerates the stem and transit amplifying cell population, with a subset of cells separating to move upwards, exit the cell cycle and differentiate, thus forming the spinous, granular and cornified layers of the epidermis. In the HPV infected environment, expression of early HPV genes, particularly E7, forces re-entry into the cell cycle, thus enabling

HPV to take advantage of active host replication machinery to amplify its viral genome 45, 46. One mechanism by which HPV regulates viral genome copy number is through an E2 variant,

E8^E2C. E8^E2C is a transcriptional repressor identified in high risk HPV types 31 and 16 that limits viral genome replication and is required for HPV31, but not HPV1647-49. Once HPV infected cells reach the upper layers of the epithelium, late viral genes are expressed for capsid production and virion assembly, and release of infectious particles follows. HPV is shed from the surface of the epidermis in conjunction with squamous flakes in order to maximize local virus concentrations. As such, new infections can occur to repeat the infectious cycle in the same or in a different host.

6 Over the course of viral replication, HPV is maintained as an extrachromosomal, circular element (episome). Partitioning to daughter cells during mitosis can be attributed, at least in part, to the function of E2. The E2 protein is able to tether the HPV genome to DNA through the binding of cellular DNA-associated factors including, but not limited to, Brd450, 51. Episomal HPV is often detected in non-malignant and pre-malignant tissue, while integrated HPV is detected largely in malignancies, but is not necessarily required for oncogenic progression. This is highlighted by one study examining the frequency of HPV integration in precancerous and cancerous cervical lesions. The three most common high risk HPV types (16, 18, and 45) were integrated more often relative to other high risk HPV types. Furthermore, HPV45 was most frequently integrated in precancerous lesions, while HPV18 was most frequently integrated in cervical cancer. Although HPV was present in all specimens, integration was not ubiquitously detected suggesting integration-independent mechanisms of HPV-induced oncogenesis are at play52. HPV has been reported to integrate in the host DNA at common fragile sites, although the copy number and location varies53. However, other mechanisms of HPV integration have been proposed recently, based on sophisticated genome sequencing technology, in cervical and head and neck cancers. These will be described in further detail below.

Integration events can disrupt E2, allowing for the deregulation of the HPV E6 and E7 oncogenes whose upregulation then further promotes growth advantages and genomic instability through their respective cellular targets 54, 55. E2 controls E6 and E7 expression in part through promoter binding and repression56-59. Once released from E2-mediated repression, E6 and E7 activities are stimulated. These proteins bind and target p53 and Rb pocket proteins for degradation, respectively. E7 not only binds to Rb, but also to the Rb family members p107 and p130, ensuring cells progress through S-phase in the absence of proper E2F/Rb cell-cycle control, and gain proliferative characteristics60, 61. E6 also activates telomerase reverse transcriptase (TERT), which contributes to keratinocyte immortalization62-64. More recent studies of these oncogenes have identified new targets in various signaling pathways. For instance, the

7 importance of hormone signaling in the development of cervical cancers is clearly appreciated.

A requirement for estrogen receptor (ER) was discovered in transgenic K14E7 mouse models for the initiation and maintenance of cervical cancer13, 16. Furthermore, microRNAs (miRs) are emerging as an additional mechanism whereby the HPV oncogenes promote SCC65-68. More detailed information on the numerous targets of E6 and E7 can be found in elegant reviews on the biology of HPV 46, 69-71.

Expression of the HPV E5 oncogene is also believed to contribute to the oncogenicity of

HPV, at least in part through signaling from growth factor receptors such as EGFR72. In vivo experiments suggest E5 contributes to hyperplasia, aids in the formation of epithelial tumors and synergizes with E6 and E7 to promote more severe disease phenotypes73-75. Recent work also demonstrated E5 regulates growth and invasion in cervical cancer cell lines76. The interplay between HPV E5, E6 and E7, and their downstream targets is critical for enabling HPV to immortalize and transform keratinocytes, eventually leading to SCC formation.

While there is ample data on HPV viral activities and cellular targets in the literature, details on the viral life cycle including mechanisms of uptake, replication and virion production remain to be discovered. Additionally, evolutionarily conserved properties and critical molecular distinctions between high risk and low risk HPVs require further study for an improved understanding of disease etiology77, 78. The capacity of low-risk HPV E6 and E7 proteins to bind and degrade p53 and Rb is limited when compared to that of the corresponding high-risk HPV proteins. Low risk HPV E6 proteins also do not harbor PDZ binding domains, which contribute to the transforming ability of E679, 80. Furthermore, the characteristics of basal cells that are initially infected by HPV have remained controversial. The basal compartment of the epidermis is believed to contain bona fide stem cells with low proliferative capacity and highly proliferative transit amplifying (TA) cells with less of a self-renewal capacity. The existence of TA cells has been debated, but it is believed these cells will only proliferate for a limited time and eventually undergo differentiation81, 82. It remains to be seen whether and how HPV targets the stem TA

8 cell population in human epidermis44, 83-85. Finally, HPV has been detected in multiple cancer types including cervical, anogenital and head and neck, but the extent to which tissue specific molecular activities drive cancer phenotypes remains unclear.

Clinical characteristics of HPV+ cancers

Cervical cancer is the second leading cause of female cancer deaths worldwide, but due to national recommendations for screening and vaccination the burden of cervical cancer lies largely outside of the United States. HPV accounts for nearly 100% of all cervical cancers, most of which are squamous cell carcinomas due to infection with high risk HPV16; however, cervical adenocarcinomas are most strongly correlated with infection of high risk HPV18 (Table 1.1) 86,

87. Although the main route of HPV infection results from sexual contact, non-sexual transmission (vertical transmission) can occur from mother to child pre- or perinatally. Cesarean births may eliminate vertical transmission, but the surgical risks may outweigh potential benefits88. Papanicolaou (Pap) tests are used to screen the cervix for cellular abnormalities that may be indicative of pre-cancerous or cancerous lesions and these cells can be tested for the presence of HPV DNA through reflex testing by in vitro nucleic acid hybridization assays. Reflex testing is performed only if there are abnormal Pap results, at which time HPV testing is requested and detection of high risk or low risk HPV types is possible 89. Pap tests can detect the early stages of cervical cancer termed cervical intraepithelial neoplasias (CIN). These are pre-malignancies graded on a scale of 1-3, with CIN1 indicating only slight dysplasia while CIN2 and CIN3 indicate moderate to severe disease. Diagnosing CIN lesions prior to progression to cervical cancer drastically improves patient survival and response to treatment90.

The peak age of infection for women is under age 25, with the potential for continual infections by either the same or different HPV types over time91. Continual HPV infection increases the risk of transformation, as the likelihood of genome instability and oncogenic mutations rises when HPV is persistent and disruptions to the cell cycle and many other

9 biological processes occur. However, only a small fraction of women will go on to develop cervical cancer. Immune cells clear these infections in most cases, suggesting immune deficiencies, mutations or inherent genomic instability may be contributors to the development of malignancy. Furthermore, whether viral load and disease severity are positively correlated has been controversial in the literature, and may be dependent on the HPV type; however, viral presence and disease risk can now be reliably assessed by E6 and E7 mRNA expression and may be utilized in conjunction with clinical data to more accurately answer this question7, 90.

Head and neck squamous cell carcinomas (HNSCCs) account for approximately 50,000 new cancer diagnoses in the United States each year92. These tumors are derived from the epithelial cells that line the mucosal surfaces of the head and neck, which include the oral cavity and oropharynx. Previously, common risk factors were tobacco and alcohol use, but more recently

HPV infection has emerged as an additional etiological factor in this disease, particularly in oropharyngeal cancers9. In fact, the number of oropharyngeal cancers is on a steep upwards trajectory, attributable to increased incidence of HPV-associated tumors93. Patients with HPV positive tumors have improved survival rates and respond better to therapies than their HPV negative counterparts94. This had led to the suggestion that HNSCCs may be treated as distinct entities, dependent on the HPV status of the tumor. Treatments for HNSCCs include chemotherapy, radiation, and surgery, all of which can lead to undesirable side effects for patients such as hearing loss, dysphagia, facial disfigurement, and inability to use their natural voice. The combined efforts of clinicians and basic scientists to improve the survival rate of

HNSCC have been relatively unsuccessful overall95. Further identification of how HPV positive and HPV negative HNSCCs differ molecularly will be important for identifying specific therapeutic targets in each subset and better understanding the pathogenesis of this disease for improved targeted therapies.

Table 1.1 depicts a summary of clinical characteristics of cervical and head and neck cancers. Almost all HNSCCs that harbor HPV DNA contain HPV16, whereas in cervical cancer,

10 many more high-risk types of HPV are accountable for the disease. It is not understood why there is such HPV-type specific variability between these cancer types, but perhaps there is a tissue specific or viral exposure preference for HPV16 in the oropharynx over other HPV types.

HPV-infected head and neck cancers are mostly localized to the oropharynx, which includes tonsillar and base of tongue cancers, with some evidence that oral cavity cancers have a low prevalence of HPV96, 97. Unlike cervical cancer, there are no available screening tools such as

Pap smears to aid in the prevention of HNSCC development. Designing such screens will be difficult due to the feasibility of accessing the appropriate tissue where infection occurs98.

Instead, patients must rely on diligently evaluating themselves for persistent symptoms such as sore throats, swollen glands or oral lesions. These seemingly benign symptoms are often overlooked, contributing to the unfortunate fact that HNSCCs are oftentimes not diagnosed until the later stages of disease.

11

Table 1.1. Summary of Clinical Characteristics of Cervical and Head and Neck Cancers

Cervical SCC Cervical HPV+ HNSCC HPV- References Adenocarcinoma HNSCC HPV Types 16 (>50%) 18, 45, 31, 18 (38%), 16 (12%), 45, 16 (90%), 18, N/A , , 9 (decreasing 33 31 few other HR- 87 86 prevalence) HPV types Prevalence ~73% of all invasive ~14% of all invasive >25% of all ≤75% of , 9 cervical carcinomas cervical carcinomas HNSCC HNSCC 99 Over 99% HPV+ Over 99% HPV+ Additional Smoking, HIV infection, Chlamydia, Oral Tobacco Usage, Alcohol, Paan , Risk Factors Contraceptives (>5 years) (Asia), Maté (South America) 100 92 Incidence Decreasing USA; High Increasing Incidence Increasing; Decreasing; , Incidence in younger older patients 95 90 Developing Countries patients

12 Prognostic comparison and staging

In general, the prognosis of cervical cancer is excellent at early stages, often with surgical resection alone. Cervical cancer screening in developed countries has led to early detection, thus dramatically reducing cervical cancer mortality. Unfortunately, locally advanced cervical cancer and metastatic disease is much more common in developing countries making it the second leading cause of cancer related death in women worldwide101. Women with cervical cancer may present with abnormal or post-coital vaginal bleeding, but most are asymptomatic and therefore, without proper screening detection is delayed. Cervical cancer is the only cancer type that is still staged clinically and criteria are mostly based on local and distant extension of the tumor (Table 1.2)102. Cervical cancer patients with stage I disease have five-year overall survival rates of 80-99%; however, survival is greatly reduced to only 15-16% in those with stage IV disease, even with aggressive treatment103.

HNSCC is a much more heterogeneous group of cancers, and therefore, the prognosis varies with site and stage of disease. Much like cervical cancer, however, many localized

HNSCCs (stage I and II) are cured with radiation or surgical resection alone. As in cervical cancer, HNSCC survival decreases with advancing stage. Importantly, the prognosis of oropharyngeal HNSCC has greatly improved over the past several decades (Table 1.2). In fact, for tonsil cancer, five-year survival rates have improved dramatically from the 1980s to 2006 in local disease, locally advanced and distant disease, from 62% to 85.9%, 38.2% to 73%, and

21.2% to 41.5%, respectively (Table 1.2). These improvements are likely reflective of more sophisticated surgical and radiation techniques, increasing HPV prevalence, and better supportive care. However, a more detailed understanding of disease mechanisms will help elucidate strategies to continue to improve these outcomes. In this regard, cooperative group studies investigating treatment paradigms for the earlier and intermediate stage cancers are ongoing. Unfortunately, up to 60% of HNSCCs present locally advanced, and more than 50% of

13 HNSCC develop locoregional recurrence or distant metastasis even with initially aggressive treatment. Upon recurrence or metastasis, the prognosis is dismal at only 10-15% with a median overall survival of 10-12 months104. These cancers prove the most challenging, and successful treatment will thus require intense investigation.

14

Table 1.2. Staging Criteria and Prognosis for Squamous Cell Carcinomas

Cervical Cancer HNSCC* Prognosis

***** Stage Staging Criteria Treatment Prognosis (5yr OS) Staging Criteria Treatment (5yr OS) I

Carcinoma confined Surgical resection 80-­‐93% Tumor less than 2cm, 85.9% for *** Single modality to cervix (stage IB may no LN involvement local ** (radiation or benefit by -­‐ R+/ C ) disease resection); II adverse Extends beyond Stage IIA: Surgical 58-­‐63% Tumor between 2-­‐ features at cervix, but not pelvic resection 4cm, no LN resection=adju wall w/adjuvant R+/-­‐ C; involvement vant R+/-­‐C III Stage IIB: R+/-­‐C Carcinoma extends to Combined 32-­‐35% Tumor >4cm w/o LN Chemoradiatio 73% for pelvic wall and/or chemoradiation involvement; or n; if residual regional involves lower third tumor <4cm with disease, salvage disease (III, of vagina and/or single ipsilateral LN surgery or Iva, IVb) causes <3cm involvement resection with hydronephrosis or adjuvant R+/-­‐ C non-­‐functioning IV kidney IVa: Extends beyond IVa: Combined 15-­‐16% Invasion of tumor in IVa: 41.5% true pelvis or has chemoradiation surrounding chemoradiation distant IVc clinically involved the IVb: Palliative structures with or , induction mucosa of bladder or chemotherapy without LN chemotherapy **** rectum; IVb : involvement; or any or resection metastatic disease tumor with +/-­‐ adjuvant contralateral or >6cm chemoradiation LN; IVc: metastatic ; IVb : clinical disease trials vs. chemoradiation ; IVc : palliative chemotherapy * oropharynx, other sites will vary slightly ** R+/-­‐C: radiation +/-­‐ chemotherapy *** LN: Lymph node **** + ipsilateral or bilateral neck dissection based on LN involvement required 105 ***** Tonsillar Cancer from 2002-­‐2006

15 Cervical cancers, whether SCC or adenocarcinoma, have a very high prevalence of HPV.

Therefore, it is difficult to determine whether HPV infection per se affects prognosis in these populations. Recently published data suggests HPV is a prognostic factor in anal cancer, predicting both overall and disease-specific survival106. However, there have been reports that cervical cancers positive for HPV18 carry a poor prognosis when compared to those positive for

HPV16 107. In contrast, HPV is associated with fewer HNSCC cases, only around 25%, and almost exclusively with the HPV16 subtype and predominantly in the oropharynx. Substantial evidence has recently emerged demonstrating greater survival for HPV positive HNSCC patients; however this protective effect is lessened in those who report tobacco and alcohol use94, 108.

Treatment variations

In general, early stage cervical cancer (Stage IA or IB) is treated with resection alone either by conization for those with early stromal invasion or extrafascial hysterectomy for those with lymphovascular invasion. Patients who desire pregnancy may also be candidates for radical trachelectomy with pelvic lymphadenectomy. Modified radical hysterectomy is often offered for patients with stage IA2 or microscopic IB disease. However, stage IB-IIA with high risk pathologic features (positive surgical margin, parametrial invasion, ≥ 3 positive pelvic lymph nodes or poorly differentiated tumor) requires radical hysterectomy and adjuvant concurrent radiation and chemotherapy (CRT). Adjuvant radiation is recommended for patients with intermediate risk factors (≥ 50% stromal invasion, lymph-vascular invasion, and tumor size ≥

4cm). Radiation therapy alone is also an acceptable option for early stage disease.

In contrast, locally advanced cervical cancer (Stage IIB-IVA), is treated with radiation alone or CRT. Combined treatment has greatly increased survival in locally advanced disease. The addition of cisplatin treatment was shown to increase overall survival (OS) in several studies including RTOG 90-01 demonstrating 67% OS in chemotherapy arm versus 41% in radiation

16 alone arm at 8 years, GOG-120 with 65% in chemotherapy vs. 47% in radiation alone, and

GOG-123 with 83% OS for chemotherapy versus 74% with radiation alone 109-111.

Patients that develop recurrent disease often undergo either radiation or surgery depending on previous treatments, although some cases may not be amenable to either option. Therefore, these patients and those with metastatic disease are considered incurable and many opt for palliative chemotherapy. In general, chemotherapy is cisplatin-based but many trials have demonstrated increased response rates for combination therapy combining cisplatin and most commonly, paclitaxel. Unfortunately, despite combinations of chemotherapeutics response rates remain low at 25-35% in recurrent and metastatic disease highlighting a dire need for novel agents112, 113.

Like cervical cancer, HNSCC is often curable with surgical resection or radiation therapy alone for those with early stage disease. However, many of these patients will develop recurrence and oftentimes present with locally advanced disease. Patients with locally advanced disease may undergo surgery, radiation, or CRT with each treatment carrying certain consequences. Surgical resection can result in disfigurement, poor wound healing and at times loss of voice (in the case of laryngeal tumors). However, CRT may result in dysphagia, mucositis, neurotoxicity and renal failure. Despite aggressive treatments, many tumors recur, and patients then undergo the treatment modality not previously offered, or re-resection.

Despite improvements of surgical technique, the rate of relapse for both local and metastatic disease after surgery alone is high, particularly in patients with high-risk features such as perineural invasion, multiple lymph node involvement, extracapsular spread and positive surgical margins. Two pivotal studies demonstrated improved progression-free survival (PFS) and overall survival (OS) in patients with high-risk features treated with adjuvant CRT, albeit with greater toxicity over treatment with radiation alone. However, PFS and OS were only 47% and 53%, respectively at 5 years, suggesting that new treatments are necessary in order to improve outcomes in high-risk patients114, 115.

17 Combined CRT is frequently offered upfront in locally advanced disease. In 2003, a pivotal phase III trial showed that concurrent chemotherapy and radiation using cisplatin resulted in superior outcomes when compared to radiation alone in unresectable, locally advanced disease. Three year overall survival was 23% in the radiation alone arm compared to 37% in the

CRT arm (p=0.014)116. As in cervical cancer, cisplatin is the most commonly used chemotherapeutic with the highest response rate. This therapy remains a standard of care for patients with unresectable, locally advanced disease and is also offered for organ preservation.

Since the original report, many retrospective studies suggest improved survival in these patients with survival rates as high as 90% at 3 years in some instances117, 118. Interestingly, the EGFR directed monoclonal antibody, cetuximab, and concurrent radiation also resulted in an increase in survival in HNSCC of 45.6% versus 36.4% for radiation alone; thus, cetuximab is an acceptable agent oftentimes associated with less severe toxicities119. Most of the patients in the latter study had oropharyngeal HNSCC, but unfortunately HPV status was unknown. Therefore, it is unclear whether improved survival may have been due to a higher proportion of HPV positive tumors.

Many HNSCC relapses occur locally despite successful initial definitive treatment. Upon local relapse, standard of care is often salvage surgical resection which confers local control rates of 33-50% and a long term overall survival of 20-40%120-122. Many patients, however, are not candidates for resection due to multiple comorbidities. Like cervical cancer, palliative chemotherapy, either single or multiple agent, is regularly offered to these patients as well as to those with metastatic disease. Unfortunately, response rates are poor at 15-36% with increased toxicity and overall survival rates of only 5-11 months123-126.

Our ability to determine which patients (cervical cancer and HNSCC) will recur and which will do well with less intensive therapy and therefore less treatment related toxicity is limited. Active research is ongoing to develop model systems that resemble human HNSCC and that may be useful platforms for the development of alternative treatment strategies127, 128. As mentioned

18 above, it is unclear whether HPV has an influence on cervical cancer survival as almost all of these tumors are HPV positive, but certainly HPV status has an established impact on the prognosis of HNSCC. In fact, a new report suggests that HPV is not only protective against initial disease progression, but also improves survival in patients with recurrent disease129.

Patients with HPV positive disease may be able to undergo less intensive therapy, thereby decreasing the incidence of toxicity, at least partly due to their increased sensitivity to radiotherapy 130, 131. Several clinical trials are ongoing including Radiation Therapy Oncology

Group (RTOG) Trial 1016 comparing cetuximab and radiation versus cisplatin and radiation in locally advanced disease. The benefit of cetuximab in cervical cancer is unknown and actively investigated. However, it is clear that not all HPV positive HNSCC tumors behave equally.

Positive smoking history, likely resulting in increased genomic mutations, may explain why some HPV positive tumors do not fare as well with therapy. Accordingly, the improved survival of HPV positive patients is negated by tobacco use108. As the incidence of this group of cancers grows, it is important to further classify these tumors and elucidate optimal targets that are specific to each subset.

Vaccines offer protection from HPV-associated cancers

Cancers often arise due to mutations that are either inherited, occur de novo, or are acquired from exposure to environmental factors. As cancers are quite heterogeneous in nature, with a multitude of mutations, identifying therapies that work on entire tumor cell populations including tumor-initiating or cancer stem cells has been difficult132. Current therapeutic intervention includes radiation, chemotherapy, and surgery, with the major caveat that many of these treatments cause major disruptions in a patient’s quality of life. The transforming ability of human papillomavirus in the cervix, anogenital tract, and oropharynx suggests that prevention of

HPV infection may ultimately prevent HPV-associated cervical and head and neck cancers.

Vaccination has been successful at preventing infection of other viruses including influenza,

19 rotavirus, and varicella, setting the precedent and opportunity for scientists to create a vaccine that prevents infection with HPV and therefore, HPV-associated cancers.

There are currently two vaccines on the market: Cervarix® (GlaxoSmithKline) and Gardasil®

(Merck & Co.). These vaccines cover the most common oncogenic HPV types known to cause cervical cancer, while Gardasil® also includes those responsible for genital warts. Cervarix® is a bivalent vaccine that protects against high risk HPV16 and 18, while Gardasil® is quadrivalent and protects against HPV16 and 18, as well as the low risk HPV6 and 11 types. Additionally, protection for both vaccines has been demonstrated to be ≥5 years133-135. These vaccines are protective against cervical, vaginal and vulvar cancers, as well as genital warts in the case of

Gardasil®. The current recommendation by the Food and Drug Administration (FDA) is for men and women ages 9-26 years to be vaccinated. This is the age range where HPV infection is most likely to occur. It is imperative to begin the vaccination series at the earliest recommended ages, prior to the start of sexual activity, to offer maximal protection against HPV infection; vaccination after infection does not result in enhanced protection from or clearance of the virus.

Both Cervarix® and Gardasil® utilize virus-like particles (VLP) formed by self-assembly of the HPV capsid protein L1 to elicit the immune response, as opposed to live or attenuated virus.

Antibodies are then produced against the L1-specific to each HPV type included in the vaccine

136. Although capsid proteins are highly homologous amongst HPV types, it is unclear how much protection is offered against other HPV types not covered by vaccination; however, a recent study with Cervarix® suggests it may also cross-protect against the oncogenic HPV types 31,

33, 45 and 51137. Since the introduction of these vaccines in the mid-2000s, wide spread studies of their efficacy in various populations and countries has been performed. These studies have focused on anogenital cancers and genital warts, but the vaccines are not yet approved as preventives against HPV-infected HNCs, partially impeded by the lack of information on HNCs required by the FDA for future studies138, 139. Unlike cervical cancer where there is still a large prevalence of other high-risk HPV types contributing to disease, over 90% of all head and neck

20 cancers are HPV16 positive. Even though not yet approved by the FDA for protection against

HNCs, these vaccines may be highly efficacious as they cover the HPV type that is overwhelmingly implicated in HNC carcinogenesis.

Mutational analyses

The advent of whole exome and whole genome sequencing and other “omics” technologies has enabled scientists to identify specific mutations in tumor specimens that may lead to targeted and personalized therapies. This is especially important in cancers such as cervical and head and neck where a diagnosis is often not made until the advanced stages of disease.

These late stage tumors are not as treatable and have few targetable molecules identified.

Previously, common chromosomal aberrations in both tumor types had been identified in the presence of HPV. These include loss at 13q and 11q22.3-25 and gains at 20q and 3q24-29; albeit unique chromosomal alterations also exist for cervical and HNC140. This indicates the existence of specific genomic alterations which are caused by the presence of high risk HPV independent of anatomical site. Only recently have samples from these two cancer types been analyzed by whole exome sequencing, which revealed profiles of somatic mutations and globally de-regulated pathways.

A recent publication by Ojesina, et.al. compared mutations in cervical squamous cell carcinoma and adenocarcinoma against normal controls to identify novel mutations. In this study the authors identified multiple new mutations in cervical cancer including MAPK1,

FBXW7, and EP300141. Additionally, pathway analysis revealed that genes associated with the immune response were most significantly mutated in cervical SCC, while cervical adenocarcinoma had recurrent mutations in the PIK3CA/PTEN pathway.

Similar publications from exome sequencing of HNSCCs were reported in 2011. Only a few somatic mutations match those discovered in the cervical cancers142, 143, including FBXW7 and

PIK3CA. The head and neck studies were further segmented into HPV+ and HPV- categories,

21 with far fewer genes mutated in HPV+ as compared to HPV- tumors. Not surprisingly, HPV- tumors harbored higher global mutation frequencies compared to the HPV+ group, presumably due to tobacco use and highlighted by TP53 mutations in almost all HPV- tumors. Table 1.3 depicts prominent examples of mutated genes in both the cervical and head and neck cancer studies. A complete list can be obtained from the original research papers. The data suggest that although HPV can be the etiological cause in both cancer types, associated cellular mutations are variable. This may be due to other etiological influences that drive mutations including tobacco use or to the differential anatomical locations of the tumors.

22

Table 1.3. Mutations identified by whole exome sequencing. Mutated genes

that are shared between cancer types

Cervical SCC141 HPV+ HNSCC142, 143 FBXW7¢ NOTCH1 FBXW7¢ PIK3CA¢ SYNE1 CDKN2A MAPK1 HRAS NOTCH1 HLA-B PIK3CA¢ PIK3CA¢ STK11 MED1 Agrawal, et.al. EP300 MLL2 NFE2L2 EZH2 SYNE2 NOTCH3 Stransky, et.al.

23 HPV integration

HPV integration in cervical cancer can lead to the disruption of E2 as HPV enters common fragile or other sites in the host genome. This is an important event during the progression of cells to a transformed state, albeit not necessarily required, which provides a growth advantage given the control of E6/E7 expression by E2 under normal circumstances. Transformation is continuously dependent upon E6/E7 expression and can be reversed by the reintroduction of

E2144-146. Research related to the mechanisms and sites of HPV integration has been limited to cervical cancers, with nascent studies of integration in HNCs. Integration is one of the more poorly understood aspects of HPV biology, with controversy over where integration events occur: either randomly through the genome or at specific “hotspots”. These ‘hotspots’ include cytogenetic bands 3q28, 4q13.3, 8q24.21, 13q22.1, and 17q21.2 accounting for integration sites of 22% of cervical tumors analyzed147, 148. This study also included strong evidence for HPV

DNA near various miRNA sites, some of which are already published as known regulators in cervical cancer. On the other hand, recent work demonstrates HPV integrates randomly across the genome, but tends to cluster near sites of structural alterations in the genome149. This report included not only cervical cancer lines, but also HNC cell lines and primary tumor specimens, all exhibiting a similar pattern of HPV integration in regions of genomic instability. An elegant mechanism to explain this phenomenon was proposed, with a viral genome looping model to explain HPV-driven amplifications and rearrangements that occur at sites of integration, which may be continually propagated throughout the genome149, 150. Additionally, the authors identified numerous breakpoints in the integrated HPV genome that are outside of the E2 open reading frame, thus questioning the longstanding model of preferential E2 disruption through integration.

Even with some disparate findings between the above reports, both identify TP63 as a gene affected by integration events found in both cervical and HNCs. Schmitz, et.al. defined the chromosomal hotspot 3q28, where the TP63 gene is located, and demonstrated TP63 was

24 disrupted by HPV integration in multiple cervical tumors. Likewise, Akagi, et.al. identified HPV integration within the TP63 gene, resulting in the loss of normal TP63 gene expression and the creation of a novel protein isoform, along with viral-host fusion transcripts. Furthermore, c-myc was also amplified and overexpressed in both cancer types in the presence of integrated

HPV151, 152. Although E6 and E7 are sufficient for immortalization in vitro, the accumulation of oncogenic mutations in cellular tumor drivers or suppressors is necessary to achieve neoplastic transformation and cancer development46. In depth studies enabled by next-generation genomics and functional studies will dissect the contribution of HPV integration and genomic rearrangements to carcinogenesis. Elucidating the role of the full complement of human genes whose activation or inactivation initiate and drive transformation will identify new targets and therapies tailored to the mutational landscape of individual tumors and patients.

Gene expression in HPV+ cancers

It is clear that HPV infection and transformation de-regulates gene expression patterns in keratinocytes. For many years, studies of HPV-driven host gene expression involved either the introduction of the HPV genome or of the HPV E6 and E7 oncogenes in existing cell lines. More recently, studies of primary tumors compared to control specimens have provided additional insights in clinically relevant model systems. To create a database of reliable genomics and transcriptomics information for many cancer types, the National Cancer Institute (NCI) along with the National Human Genome Research Institute (NHGRI) collaborated on the ambitious

‘The Cancer Genome Atlas’ (TCGA) project 153. The goal of this project was to create freely available tumor datasets (with appropriately matched normal controls) subjected to genomic studies including exome sequencing, single nucleotide polymorphism (SNP) analysis, and mRNA expression by RNA-Sequencing, together with extensive clinical annotations. Currently, information from over 30 tumor types has been deposited in the TCGA, with a collection target of at least 500 samples for each cancer type153. The TCGA will provide massive datasets for

25 broad integration with basic and clinical sciences. This includes the opportunity for personalized medicine, where new diagnostics, therapeutic targets, and individualized treatments may eventually be possible. Both cervical squamous cell carcinoma and adenocarcinoma, along with

HNSCC are part of the TCGA; however, the cervical data will not be available for data mining until later in 2014. Once these data are available, it will allow for groundbreaking opportunities to compare HPV+ cervical with HPV+ HNCs. Given the large number of specimens already banked for both cancer types, clear trends in gene expression should delineate common molecular and genomic occurrences that are due to HPV activities.

In the absence of published TCGA based comparisons, smaller studies that profile HPV- related gene expression are publically available. One of the first papers published comparing cervical cancer (SCC and adenocarcinoma) to normal cervical keratinocytes by Santin et.al. identified approximately 500 genes up- or down-regulated in the presence of HPV. Up-regulated genes included CDKN2A (p16), TOP2A, E2F1, FOXM1, MCM2/4/5, and PTGES, with a high prevalence of genes involved in cell cycle regulation. Down-regulated genes including TGFβ1,

TGFα, and KRT16, were clustered in differentiation and tumor suppressor pathways154. This report also demonstrated substantial similarities in expression profiles of cervical SCC and adenocarcinomas, despite the fact that SCCs were HPV-16 positive and adenocarcinomas were

HPV-18 positive. This suggested distinct high-risk HPV types evoke similar transcriptome alterations.

Analyses by Mattarocci, et.al. in 2013 utilized a retrospective approach to identify novel genes involved in dysplastic and neoplastic cervical lesions, based on targets previously published in the literature155. The authors confirmed earlier results where FOXM1, TOP2A, and

E2F1 were not only overexpressed, but where their respective expression levels gradually increased as cervical lesions progressed from low-grade to high-grade lesions and eventually to carcinomas. Other de-regulated genes included CDC25A, TP73, and CCNE2. Conversely,

26 genes that demonstrated reduced expression in the progression from dysplastic tissue to carcinoma included PTGS2, BCL2, and FOS155.

A multitude of publications comparing the gene expression profiles of HPV+ to HPV- HNSCCs are available. A study by Martinez, et.al. compared HPV+ vs. HPV- vs. normal oral mucosa and analyzed these specimens in a number of ways. The authors identified many of the same gene expression changes initially reported by Santin et.al. including the up-regulation of CDKN2A,

TOP2A, and MCM2/3 in HPV-infected HNCs compared to normal mucosa. Other up-regulated genes reflected pathways such as DNA replication, cell cycle progression, DNA repair and differentiation, while down-regulated genes reflected processes such as proteolysis and chemotaxis. Additional comparisons between HPV+ and HPV- HNCs identified up-regulated genes involved in nuclear structure, mitosis, DNA repair and methylation, and down-regulated genes involved in signal transduction and proteolysis156. Finally, Pyeon, et.al. directly compared gene expression of HPV+ cervical and head and neck cancers and discovered similar trends where cell cycle-related genes are highly up-regulated157. Other datasets confirm these findings where HPV drastically affects gene expression patterns involved in cell cycle, repair and replication157-160. This is not necessarily surprising given the known function of E6 and E7 in the inhibition and degradation of critical tumor suppressors such as p53 and Rb.

p53 and Rb-dependent functions are regulated by p14ARF and p16, respectively, which are encoded by the same gene, CDKN2A161, 162. This is one of the most frequently overexpressed genes in HPV positive tumors, largely due to regulatory interference by E6 and E7. When E7 inactivates Rb, E2F1 is free to initiate transcription of its targets that includes the CDKN2A locus163. p16 is so consistently and highly overexpressed in HPV+ cancers that it is used clinically as a surrogate marker for HPV positivity during cancer diagnosis.

Epigenetics is another mechanism whereby gene expression patterns can be altered in normal and transformed cells. This is a complex field that investigates DNA changes that do not affect the inherent nucleotide sequence, but rather processes such as methylation patterns and

27 histone modifications. The presence of HPV greatly affects methylation signatures across the genome, promoting hyper- and hypomethylation. For example, tumor suppressor genes, such as TERT, exhibit promoter hypermethlyation and thus gene silencing in CIN lesions and cervical cancers164, 165. The methyltransferases responsible for methylation, DNMT1, DNMT3A, and

DNMT3B, are also highly overexpressed in HPV positive cervical and HNSCC cells90, 166, 167.

Conversely, hypomethylation increases during the transition of dysplastic to cancer cells and results in aberrant gene activation, suggesting that hyper- and hypomethylation coordinate for the evolution of an oncogenic state168. Comparisons of HPV positive and negative HNSCC samples also reveal the extent to which HPV alters methylation patterns, research which contributes substantially to the identification of novel target genes169.

Initial studies comparing cervical and HNCs to normal controls suggested HPV elicits similar gene expression alterations between tumor types. Publication and analysis of the TCGA data will likely confirm this trend, but novel details are also expected which should reflect differential tumor origins in various tumor types, data which may be exploited for the targeting of tumor cells by novel agents. The large number of samples combined with thorough bioinformatics analyses and ultimately confirmatory studies will help validate genes of interest shared or unique to cervical and HNCs. Further separation of these groups based on clinical, environmental and demographic data is bound to enrich and surprise the HPV community.

Conclusions

Human papillomavirus is the most common sexually transmitted viral pathogen and causes approximately 5% of the global cancer burden170. Even after decades of intense research of the epidemiology and molecular biology of HPV infection and transformation, the aggressive nature of HPV-related late stage cancers and novel molecular targets for new therapies do not meet clinical needs. Cellular events downstream of HPV infection and integration such as the induction of genomic instability and selection for oncogenic mutations must occur in order to

28 complete transformation process that is initiated by the expression of the E6 and E7 oncogenes.

Cervical cancer was the first tumor type attributed to the effects of HPV infection and has since been the standard for studying HPV biology. In fact, one of the most widely used cell lines in cancer biology is the HeLa cell line derived from an HPV18 cervical adenocarcinoma. The discovery that HPV infection is also an underlying cause of head and neck squamous cell carcinomas has opened the door for comparisons that will ultimately inform the treatment of each tumor type towards improved diagnostics, therapeutic options, and patient survival.

Acknowledgements

The authors would like to thank Dr. John C. Morris at the University of Cincinnati and Eric A.

Smith for their critical reading of this manuscript. Work on HPV-related carcinogenesis was supported by grant funding from the NIH: RO1 CA102357 and CA116316 (S.I.W.),

T32ES007250 (A.K.A.), University of Cincinnati MERF (T.W.D.).

29 Chapter 2: Novel oncogenes in HNSCC: an introduction to DEK and IRAK1

30 Part 1. The DEK oncogene is a multifunctional protein

Biochemistry of DEK

DEK is a highly conserved and unique protein that has little homology to other proteins, but is most closely related to the HMG family. DEK is 375 amino acids in length with three DNA- binding domains: the pseudo-SAP, the SAF/SAP (scaffold attachment factor-box), and the C- terminal multimerization domain.171-173 It also harbors multiple acidic regions and a putative nuclear localization signal (NLS), but has no enzymatic activity (Figure 2.1).174 Most of DEK is localized in the nucleus, bound to chromatin, but a small percentage of DEK binds RNA.175

Recent work suggests recombinant DEK is internalized by multiple cell types and this is dependent on heparan sulfate proteoglycans (HSPGs). Additionally, DEK is secreted by macrophages and can be taken up by adjacent cells. This supports the role of DEK as an autoantigen in various autoimmune disorders.176

DEK does not bind DNA in a sequence-specific manner, but preferentially binds to particular DNA structures such as supercoiled, helical, and four-way junction DNA.177 While the protein prefers associating with negatively supercoiled DNA, it can also introduce positive supercoils into SV40 minichromosomes in cell-free assays.178 In mammalian cells there is conflicting data regarding the preference of DEK for heterochromatin or euchromatin. Hu et al. reported DEK to be excluded from heterochromatin and concentrated at regions undergoing active transcription.179 On the other hand, DEK was found to be essential for heterochromatin integrity and enhanced H3K9me3 foci through binding to heterochromatin protein 1α.180 From these findings, it is likely DEK regulates gene expression at least in part through modulating chromatin topology. However, a link between this role for DEK and transcription remains to be established.

DEK undergoes a variety of post-translational modifications that can influence its activity, including phosphorylation, acetylation, and poly(ADP-ribosyl)ation. Phosphorylation decreases

31 the affinity of DEK for DNA in vitro and stimulates multimerization of DEK with itself. The main kinase identified in regulating DEK DNA-binding activity is Casein-Kinase 2 (CK2), but DEK harbors phosphorylation sites for PKC and GSK-3 phosphorylation as well.181 Acetylation by

CBP, p300, and P/CAF also decreases the affinity of DEK for DNA and re-localizes DEK within the nucleus to interchromatin granules (ICGs), which contain other RNA-processing and transcription factors.182 Interestingly, autoantibodies from the synovial fluid of patients with juvenile idiopathic arthritis preferentially recognize acetylated DEK over the phosphorylated form.183 DEK can also be modified with poly(ADP-ribose) (PAR) by PARP-1, which releases

DEK into the cytoplasm.184, 185 Finally, in cells undergoing apoptosis, DEK is both phosphorylated and PARylated, and excreted into the extracellular space.186 Although these modifications are established in the literature, little data is available on the functional importance of these modifications in the DEK-dependent phenotypes observed in various malignancies.

Dual roles for DEK in transcription

Given the close relationship of DEK and chromatin, it is not surprising that the extent of

DEK expression is important for cellular gene expression, but the role of DEK in transcription is not clearly defined. Paradoxically, DEK has been implicated as either a co-activator or a co- repressor, depending on the system. As mentioned above, DEK is a substrate for PARP-1, and

PARP-1 in combination with SET functions to remove DEK from chromatin and permit assembly of the preinitiation complex for transcriptional activation.185 Supporting this role for DEK as a transcriptional repressor, DEK represses p300 and P/CAF histone acetyltransferase-dependent transcription in leukemia cells in a manner dependent upon the acidic regions within DEK.187

DEK also negatively regulates NF-κB-dependent transcription through interactions with the p65/RelA protein.188 On the other hand, DEK has also been described as a transcriptional co- activator, as seen by its interaction with AP-2α and stimulation of transcriptional activity in

32 glioblastoma, and its co-activation of nuclear receptors in Drosophila.189, 190 Lastly, C/EBPα and

DEK functionally cooperate to activate genes associated with myeloid differentiation.191

Global profiling of transcriptional networks associated with DEK loss or overexpression in normal or transformed cells has only recently been initiated. When DEK-knockdown lung cancer cells were compared to non-targeting controls, metabolism, nucleosome assembly, and transcription factor expression were significantly altered.192 Later, chromatin immunoprecipitation combined with microarray analysis in DEK-depleted leukemia cells revealed that DEK represses hTERT transcription in a manner that was dependent on the posttranslational modifications of DEK.193 Finally, in myeloid cells DEK preferentially binds open chromatin and is enriched at the promoters of highly transcribed genes. The localization of DEK at these promoters was found to either activate or repress transcription, depending upon the site.194 Taken together, the relationship between DEK loss and transcriptional regulation remains poorly defined. The effects of DEK overexpression on transcription have yet to be studied and may provide possible explanation for how DEK contributes to the transformation of normal cells.

DEK as a histone chaperone

Histone modifications and nucleosome assembly play crucial roles in chromatin structure and signaling transcriptional activity; therefore, it is not surprising a chromatin-associated protein like DEK is involved in this process. Histone chaperones are proteins that bind histones and regulate their loading onto DNA. DEK is a histone chaperone and interacts with histones, including H2B.195 In Drosophila, DEK facilitated the assembly of histone H3.3 into nucleosomes during transcription activation; however, in human cells DEK limits H3.3 incorporation into chromatin to the telomeres, strengthening the role of DEK as a mediator of heterochromatin integrity.190, 196 This notion coincides with published reports that H3.3 is found near active transcriptional sites, but also at telomeric and heterochromatic regions.197 DEK also associates

33 with Daxx, a transcriptional co-repressor, HDACII, and the core histones (H2A, H2B, H3 and

H4) and contributes to transcriptional repression.198

DEK loss abrogates the DNA damage response

In addition to chromatin structure and transcription DEK also is involved in the response to DNA damage, DNA repair, and DNA replication. DEK loss sensitized human and mouse cells to DNA damage and also significantly reduced repair by non-homologous end joining (NHEJ) in

Dek knockout mouse embryonic fibroblasts (MEFs). This was supported by delayed Ku70/80 localization to double-strand breaks and reduced DNA-PK signaling upon DEK loss.186, 199 The sensitivity of melanoma cells to clastogenic chemotherapies could be enhanced by DEK loss, suggesting inhibition of DEK may be a useful clinical target.200 DEK also counteracts replication stress from stalled replication forks, resulting in fork progression and cellular proliferation, and may be one mechanism for how DEK contributes to cancer progression and chemoresistance.201

DEK functions as an oncogene

DEK was initially discovered as a DEK-CAN fusion protein in acute myeloid leukemia, and this fusion protein supports leukemia in vivo by targeting the long-term hematopoietic stem cell population.202, 203 Since then, DEK overexpression has been described in most malignancies including breast, melanoma, bladder, and hepatocellular carcinoma.200, 204-206 DEK was not defined as a bona fide oncogene until 2009, when a report demonstrated that DEK knockout mice exhibited reduced papilloma formation in a chemical carcinogenesis mouse model, and that DEK overexpression could stimulate keratinocyte transformation in soft agar and xenograft assays in vivo.207 Earlier work demonstrated DEK overexpression inhibited normal cellular differentiation of keratinocytes and increased hyperplasia of 3D epidermal models.208

Additionally, DEK inhibited senescence and apoptosis, the latter being partially attributable to

34 the destabilization of p53. Together, these observations support a role for DEK as an oncogene.209, 210

The overexpression of DEK in many tumor types may be attributable to its regulation by the E2F family of transcription factors, which are aberrantly activated when Rb is deregulated.211

Promoter studies of DEK revealed it also contains YY1 and NF-Y transcription factor binding sites that are implicated in transformation.212 In breast cancer, chromatin immunoprecipitation identified DEK as a target of estrogen receptor α, which contributes to ER+ breast cancer pathogenesis.213 Outside of transcriptional regulation, the E3 ubiquitin ligase, Fbxw7, was identified to target DEK for degradation in intestinal models of carcinogenesis.214

DEK targets multiple proteins and signaling pathways, including, p53, p63, Wnt/β- catenin signaling, and mTOR which, in part, contribute to DEK-associated phenotypes including proliferation, invasion, and stemness.204, 208, 210, 215, 216 In hematopoietic systems, DEK promotes cell survival and is important for hematopoietic stem cell engraftment as measured by competitive bone marrow transplant.217, 218 In breast cancer and melanoma, DEK expression levels correlate with disease progression, with lowest DEK expression in benign tumors and with highest expression in advanced disease.204, 219, 220 Normal DEK expression is low in primary and differentiated cells and highest in malignant cells, which should be advantageous to patients should a DEK inhibitor be developed or discovered. Because DEK plays important roles in regulating oncogenic phenotypes, and because secreted DEK protein can be detected in urine, it is currently being investigated as a biomarker for bladder cancer and has been described as a potential therapeutic target there.221

Beyond cancer, DEK autoantibodies are found in various autoimmune diseases including juvenile rheumatoid arthritis, juvenile idiopathic arthritis, systemic lupus erythematosus, and sarcoidosis.183, 222-226 DEK was demonstrated to be a pro-inflammatory protein and is secreted by macrophages, but DEK also attracts neutrophils, natural killer cells,

35 and CD8+ T lymphocytes.227 This data suggests DEK is involved in multiple aspects of inflammation and has important cellular functions outside of its nuclear localization. However, a defined role for DEK in inflammation has yet to be revealed.

36

Figure 2.1. DEK structure and identified domains. DEK is 375 amino acids in length and contains multiple DNA binding domains: the ψ-SAP, SAP, and the DEK-C. DEK-C is also the multimerization domain. Residues 205-221 harbor a putative nuclear localization signal (NLS).

DEK also carries multiple acidic regions, highlighted in pink.

37 Part 2. IRAK1 is a central, upstream hub of inflammatory signaling

IRAK1 signaling cascade

Interleukin-1 receptor-associated kinase-1 (IRAK1) is part of the IRAK family of kinases that also includes IRAK2, IRAK3, and IRAK4. These kinases are essential for driving inflammatory signaling through the toll-like (TLR) and interleukin-1 receptors (IL-1R). As part of the innate immune system, these receptors recognize foreign pathogens and initiate downstream signaling through IRAK1. This begins with the recruitment of adaptor proteins, such as MyD88, which interact with the N-terminal death-domain of IRAK1.228, 229 IRAK4 then complexes with MyD88 and IRAK1, initiating the phosphorylation of IRAK1. Phosphorylation of

IRAK1 on Threonine-209 is the most critical event and T209 phosphorylation is used as a marker for activation.230 This event then signals IRAK1 auto-phosphorylation and release.

Phospho-IRAK1 then binds to TRAF6, an E3 ubiquitin ligase, that autoubiquitinates to enable binding to TAK1 and its two binding partners, TAB1 and TAB2.231-233 Together these kinases phosphorylate and activate the IKK complex to initiate NF-κB signaling through the translocation of p50/p65 heterodimer into the nucleus for transcriptional activation.234 TAK1 can also phosphorylate MAP3Ks, turning on JNK, ERK1/2, or p38 signaling, driving subsequent translocation of AP-1 into the nucleus.235 This cascade is summarized in Figure 2.2.

In addition to phosphorylation, IRAK1 can also be modified by ubiquitination and sumoylation, which alters the localization of IRAK1 within the cell. K48-ubiquitination of IRAK1 results in proteasomal degradation, while K63-ubiquitination recruits the NEMO complex and aids in the activation of NF-κB.236, 237 Finally, sumoylation translocates IRAK1 to the nucleus and may activate transcription in monocytes.238

IRAK1 gene structure

38 The IRAK1 gene is located on the X chromosome and is 712 amino acids in length. As mentioned previously, IRAK1 contains an N-terminal death domain, a kinase domain, an activation loop, and a C1 and C2 domain at the C-terminus.239 Although IRAK family members have similar structures, they only share around 40% homology. Only IRAK1 and IRAK4 have kinase activity and IRAK4 lacks the C-terminal domain that is crucial to the interaction with

TRAF6.240-242 Additionally, IRAK3 is only expressed in monocytes and macrophages and may negatively regulate IRAK1/IRAK4 signaling.243, 244

Contributions of IRAK1 to cancer

The contributions of IRAK1 to the formation of malignancies, particularly hematological disease, have only recently been reported. Overexpression and activation of IRAK1 has been observed in acute myeloid leukemia (AML), myelodysplastic syndrome (MDS), and melanoma.242, 245, 246 MDS patients are predisposed to the development of AML, and previous work suggested that TRAF6 and IRAK1 are overexpressed in these diseases through loss of miR-146a. miR-146a is deleted in at least 80% of all del(5q)MDS and AML, and since TRAF6 and IRAK1 are both targets of miR-146a, these candidates were examined for their function in

MDS and AML.247, 248 In particular, IRAK1 was pursued as a novel drug target due to both its overexpression in MDS and central importance in innate immune signaling disease pathology. A small molecular inhibitor of IRAK1, termed IRAK-1/4, blocks kinase activity and as a result, reduced TRAF6 polyubiquitination, NF-kB activation, and progenitor cell function and increased apoptosis in MDS.245 This inhibitor is a benzimidazole that is highly specific for IRAK1 and

IRAK4 and has little effect on other kinases. Likewise, in melanoma IRAK inhibition in combination with vinblastine (a chemotherapy for melanoma treatment) increases apoptosis.

Together, this data suggests inhibition of inflammatory signaling, specifically IRAK, may be a novel approach for treating multiple cancer types.

39

Figure 2.2 IRAK1 is upstream of many oncogenic signaling cascades. Activation of TLR and IL-1R signaling initiates a cascade where MyD88 adaptor proteins complex with IRAK1 and

IRAK4. IRAK4 phosphorylates IRAK1 on a critical Thr209 residue, followed by hyperphosphyorlation of IRAK1 that releases it from MyD88. pIRAK1 now binds to TRAF6 where ubiquitination enables interaction with the TAK1/TAB complexes. These kinases can now phosphorylate IKK or MAP3K to trigger signaling that will allow the entry of transcription factors into the nucleus.

40 Chapter 3: DEK promotes HPV positive and negative head and neck cancer cell proliferation

Allie K. Adams1, Grace E. Hallenbeck1, Keith A. Casper2, Yash J. Patil2, Keith M. Wilson2, Randall J. Kimple3, Paul F. Lambert3, David P. Witte4, Weihong Xiao5, Maura L. Gillison5, Kathryn A. Wikenheiser-Brokamp6, Trisha M. Wise-Draper1,7#, and Susanne I. Wells1#

1. Cancer and Blood Disease Institute, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, USA 2. Department of Otolaryngology, Head and Neck Surgery, University of Cincinnati, Cincinnati, OH, USA 3. McArdle Laboratory for Cancer Research, University of Wisconsin School of Medicine and Public Health, Madison, WI, USA 4. Division of Pathology & Laboratory Medicine, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, USA 5. Viral Oncology, The Ohio State University Comprehensive Cancer Center, Columbus, OH, USA 6. Pathology & Laboratory Medicine and Pulmonary Biology, Cincinnati Children’s Hospital Medical Center/University of Cincinnati, Cincinnati, OH, USA 7. Division of Hematology/Oncology, University Hospital, University of Cincinnati, Cincinnati, OH, USA #Corresponding Authors

Published in Oncogene. 2014 Mar 10; doi: 10.1038/onc.2014.15. [Epub ahead of print] PMID: 246084317

41 Abstract

Head and neck squamous cell carcinoma (HNSCC) is the sixth most common malignancy worldwide, and patient outcomes using current treatments remains poor. Tumor development is etiologically associated with tobacco or alcohol use and/or HPV infection. HPV positive HNSCCs, which frequently harbor wild-type p53, carry a more favorable prognosis and are a biologically distinct subgroup when compared to their HPV negative counterparts. HPV E7 induces expression of the human DEK gene, both in vitro and in vivo. In keratinocytes, DEK overexpression is sufficient for causing oncogenic phenotypes in the absence of E7.

Conversely, DEK loss results in cell death in HPV positive cervical cancer cells at least in part through p53 activation, and Dek knockout mice are relatively resistant to the development of chemically induced skin papillomas. Despite the established oncogenic role of DEK in HPV associated cervical cancer cell lines and keratinocytes, a functional role of DEK has not yet been explored in HNSCC. Using an established transgenic mouse model of HPV16 E7 induced

HNSCC, we demonstrate that Dek is required for optimal proliferation of E7-transgenic epidermal cells and for the growth of HNSCC tumors. Importantly, these studies also demonstrate that DEK protein is universally up-regulated in both HPV positive and negative human HNSCC tumors relative to adjacent normal tissue. Furthermore, DEK knockdown inhibited the proliferation of HPV positive and negative HNSCC cells, establishing a functional role for DEK in human disease. Mechanistic studies reveal that attenuated HNSCC cell growth in response to DEK loss was associated with reduced expression of the oncogenic p53 family member, ΔNp63. Exogenous ΔNp63 expression rescued the proliferative defect in the absence of DEK, thereby establishing a functional DEK-ΔNp63 oncogenic pathway that promotes

HNSCC. Taken together, our data demonstrate that DEK stimulates HNSCC cellular growth and identify ΔNp63 as a novel DEK effector.

42 Introduction

Head and neck cancer is a devastating disease with approximately 60% of patients presenting with locally advanced disease.249 Over 600,000 new head and neck cancers are diagnosed each year worldwide, the vast majority representing squamous cell carcinomas

(SCCs).101 HNSCC has traditionally been associated with risk factors such as tobacco and alcohol consumption; however, in recent years human papillomavirus (HPV), a well-known cause of cervical cancer, has emerged as a new etiological pathogen. It is now established that at least 25% of all HNSCCs carry HPV genomic DNA, predominantly of the high risk HPV16 subtype.9 HPV positive HNSCCs are associated with improved survival for affected patients, and such tumors are more sensitive to available therapies. This increased tumor response is due at least in part to p53 tumor suppressor activation despite the presence of E6, a viral oncogene which targets p53 through physical interactions and degradation.94, 130 Thus, HPV positive and negative head and neck tumors are biologically distinct subtypes of HNSCC.

Regardless of HPV status however, achieving efficacy with radiation and chemotherapy treatments remains notoriously difficult. This is due to the inherent molecular heterogeneity of

HNSCC, combined with patients presenting at late stage disease where there is a high likelihood of local recurrence and metastasis. Thus, survival rates for patients with HNSCC have not significantly improved in decades, highlighting the urgent need for novel biomarkers and therapeutic targets.250 Recent limited analysis of HNSCC tissue arrays by our laboratory had suggested that DEK expression was up-regulated in the majority of HPV negative HNSCCs.221

Herein we extend these studies by analyzing a panel of primary human tumors to determine

DEK expression levels in both HPV positive and negative HNSCCs, and define the functional role of DEK in regulating HNSCC growth in vitro and in vivo.

The human DEK oncogene was first identified as a fusion with the CAN nucleoporin in acute myeloid leukemia, and the DEK protein was subsequently purified based on its ability to

43 modulate DNA topology.195, 202 DEK is located primarily in the nucleus where it is bound to chromatin and is rarely amplified or mutated, but frequently transcriptionally up-regulated in a variety of malignancies. Transcriptional induction of DEK can occur through Rb-dependent mechanisms in several tumor types; a finding which is consistent with the original observation that DEK is induced by high risk HPV E7 expression in human keratinocytes and up-regulated in

HPV positive cervical cancer cells.209 DEK knockdown in cervical cancer cells expressing the

HPV E6 and E7 oncogenes, results in apoptosis that was at least partially mediated through the stabilization of the p53 tumor suppressor.210 Conversely, DEK overexpression in near-diploid immortalized keratinocytes that form skin (NIKS) stimulated proliferation and suppressed differentiation in organotypic epithelial rafts through a p53 independent mechanism.208 Effects of

DEK loss beyond those dependent on p53, however, remain undefined in human SCCs.

Dek-/- mice are viable and are relatively resistant to benign papilloma formation in a chemically induced skin carcinogenesis model, thus implicating DEK in tumor initiation.207 In order to test the requirement for Dek in a malignant SCC model system in vivo, we utilized a transgenic HNSCC mouse model. Therein, HPV E7 is targeted to the stratified squamous epithelium with the keratin 14 promoter (K14E7) and mice develop rapid and highly penetrant tongue and esophageal tumors upon addition of the mutagen 4-nitroquinoline-1-oxide (4-NQO) to the drinking water.251 To determine whether Dek expression is important for HNSCC development, K14E7 mice in a Dek proficient and deficient background were subjected to 4-

NQO treatment. Transgenic K14E7 Dek–/– mice exhibited decreased epidermal cell proliferation and attenuated tumor growth as compared to K14E7 Dek+/+ mice. Importantly, Dek was not required for proliferation in non-transgenic mice, indicating that cellular growth suppression by

Dek is specific to the oncogenic stimulus induced by E7. Complementary studies in primary human HNSCC tumor tissue and cells demonstrate that DEK protein expression is universally up-regulated regardless of HPV status and that DEK supports tumor cell proliferation. Finally,

44 we show for the first time that DEK promotes HNSCC growth through a ΔNp63 dependent mechanism, thus identifying ΔNp63 as a novel downstream effector of DEK function.

45 Results

Dek knockout mice exhibit attenuated HNSCC development in vivo

High risk E7 protein was previously shown to stimulate DEK expression in vitro and in vivo, and DEK up-regulation was sufficient to stimulate oncogenic phenotypes in cellular models of SCC.207, 209 To directly determine the functional role of Dek in HPV E7 driven HNSCC, we tested the requirement for Dek in an established K14E7-driven mouse model of HNSCC.252

K14E7 transgenic mice were crossed into a Dek knockout background to generate K14E7 Dek-/- which were compared to K14E7 Dek+/+ mice. Non-transgenic Dek wild-type and knockout mice were generated as controls. All mice were treated with 4-NQO for 16 weeks followed by normal drinking water for 8 weeks. At this 24 week time point, all mice were sacrificed and analyzed for tumor development. K14E7 Dek-/- mice trended towards increased survival compared to the

K14E7 Dek+/+ mice; 82% of the K14E7 Dek+/+ mice died prior to sacrifice, compared to 43% of the K14E7 Dek-/- mice (Figure 3.1a). The cause of premature death for the K14E7 Dek+/+ and one of three K14E7 Dek-/- mice was likely due to tumor burden. The cause of death for the remaining two K14E7 Dek-/- mice is unknown. A representative section from a K14E7 Dek+/+ mouse confirms Dek is highly expressed in the tumor (Figure 3.1b). The results of the pathological analyses for these tumors are summarized in Figure 3.1c. As expected from the literature,252 all 11 of the K14E7 Dek+/+ mice developed SCCs of the tongue and esophagus

(Figure 3.1c), whereas the non-transgenic mice had no macroscopic tumors at the time of sacrifice. In contrast to K14E7 Dek+/+ mice, only one of seven K14E7 Dek-/- mice developed a visible tumor. K14E7 Dek-/- mice did have microscopic tumors indicating that Dek is not required for tumor initiation in this model (Figure 3.1c). Importantly, Dek loss was associated with smaller tumors indicating that Dek promotes HNSCC growth.

Dek loss suppresses E7-dependent proliferation in the tongue epidermis

46 We next sought to determine the underlying cellular mechanism responsible for the observed HNSCC growth suppression in K14E7 Dek–/– mice. DEK was shown to inhibit apoptosis and stimulate proliferation in vitro, depending on the cellular and experimental context. Accordingly, we assessed for both biological processes in E7-transgenic Dek proficient and deficient epidermis. Apoptotic cells were not detected by cleaved caspase-3 detection (data not shown) indicating that increased cellular death did not account for the decrease in tumor growth in Dek knockout mice. As was previously published, K14E7 transgene expression was associated with hyperplasia and an increase in BrdU positive cells in both the basal and suprabasal compartments as compared to the non-transgenic controls (Figure 3.2a and 3.2b).253

Importantly, K14E7 Dek–/– mice displayed decreased proliferation when compared to K14E7

Dek+/+ epidermis, with proliferative rates repressed to baseline levels observed in non- transgenic mice (Figure 3.2b). Dek loss alone was not sufficient to decrease cellular proliferation in the absence of E7 expression. Taken together, these findings support the notion that Dek up- regulation is required for E7 driven epidermal proliferation, and that cellular growth suppression in response to Dek loss in vivo occurs specifically in the oncogenic environment induced by E7.

DEK protein expression is up-regulated in human HNSCC regardless of HPV status

DEK mRNA and protein expression is up-regulated in multiple cancer types including breast cancer, melanoma, and hepatocellular carcinoma200, 204, 211, 254 and is also induced in response to E7 expression.209 We previously reported DEK up-regulation in high risk HPV positive and HPV negative HNSCC tissue microarrays; however, analyses were limited for HPV positive tumors since only 4/44 (9%) of the examined samples were HPV positive by in situ hybridization.221 Additionally, potential associations with demographic and clinical information were not examined. The current analysis extends these prior studies by performing rigorous

HPV testing with maximal sensitivity for the detection of HPV16 genome sequences. HPV16 is

47 the most prevalent HPV type in HNSCCs. We also analyzed relevant clinical characteristics including parameters characteristically associated with HPV status. Additionally, optimal tissue preservation was ensured by analyzing human HNSCC specimens that were freshly biopsied or resected with immediate processing and uniform preparation for IHC analysis to detect DEK protein expression. Table 3.1 depicts HPV status as determined by in situ hybridization by two independent institutions, along with highly sensitive RT-PCR for HPV16 sequences, as well as primary tumor site, tumor stage, p16 status, history of cigarette and alcohol consumption, and determination of DEK protein expression. HPV presence and DEK expression are quantified in the table, and a detailed description of the relevant values is provided in the Materials and

Methods. Expression of the cyclin-dependent kinase inhibitor p16 is a clinical marker for HPV presence. Chart review indicated that 10/18 (56%) of HNSCCs tested were positive for p16 expression. HPV positive status was detected in 8/18 (44%) specimens by in situ hybridization at Site 1 (Figure 3.3a), which detects multiple high-risk HPV types, with examples of HPV negative and HPV positive specimens shown. Highly sensitive in situ hybridization for HPV16 at

Site 2 identified two additional HPV positive specimens (10/18 or (56%)). Finally, RT-PCR identified one additional HPV16 positive specimen, (11/18) resulting in 61% of the tumors being

HPV positive. All samples were subjected to IHC for DEK expression (Figure 3.3a). The intensity of DEK protein expression and the proportion of tumor cells positive for DEK were determined relative to adjacent normal tissue using the grading system depicted in Figure 3.3b and summarized in Table 3.1. Interestingly, strong DEK expression was detected in all tumors regardless of HPV status, tumor stage or demographic information. Neither the level of DEK expression nor the DEK-positive proportion of tumor cells correlated with HPV status (Table 3.1,

Figure 3.3a, b). Taken together, these data indicate that DEK protein is universally over- expressed in HPV16 positive and negative HNSCCs.

DEK expression stimulates growth of HPV positive and negative human HNSCC tumor cells

48 In order to determine the importance of DEK expression in human HNSCC cells, we next analyzed a panel of HNSCC cell lines that were transduced with non-targeting or DEK knockdown vectors. One HPV positive line, 93VU147T, and two HPV negative lines, UMSCC1 and UMSCC6, were analyzed following DEK knockdown confirmed by western blot analysis

(Figure 3.4a-c). DEK knockdown consistently suppressed cellular growth in both HPV positive and HPV negative cell lines (Figures 3.4a-c). Moreover, growth suppression was associated with reduced proliferation in vitro as was observed in the E7-positive epidermis in the in vivo mouse model. BrdU incorporation was significantly reduced in all cell lines after DEK knockdown, indicating decreased progression into S phase in the absence of DEK (Figure 3.4a- c). Together, these data demonstrate that DEK loss in human and murine HNSCC cells attenuates tumor cell growth through suppression of cellular proliferation.

ΔNp63 protein expression is regulated by DEK and is functionally important

Our previous studies uncovered p53 dependent oncogenic function for DEK in HPV positive cervical cancer cells. p53 remains intact, yet is bound and degraded by E6 leading to suppressed p53 activity in HPV positive cells. In line with previous results in HeLa cells, DEK knockdown in HPV positive, primary CCHMC-HNSCC1 cells resulted in the up-regulation of p53 protein levels, as well as increased phosphorylation of Serine15 on p53 indicating p53 activation

(Figure 3.5a). In contrast, p53 is mutated in the majority of HPV negative HNSCCs and is not expressed in HPV negative UMSCC1 and UMSCC6 cells (data not shown, and 130, 255). Given that the growth of all three HNSCC cell lines was suppressed by DEK knockdown, p53 activity alone could not account for the proliferative defects associated with DEK depletion. The p63 family of proteins is related to p53 and expressed from two promoters, resulting in two distinct groups of p63 isoforms, TAp63 and ΔNp63, which have opposing functions. TAp63 functions as a tumor suppressor much like p53, while ΔNp63 has oncogenic functions. ΔNp63 is the predominant p63 isoform in keratinocytes and is overexpressed in the majority of HNSCCs.256

49 Mechanisms driving ΔNp63 up-regulation are complex and incompletely understood.256-258 Our previous data demonstrated that DEK overexpression in human keratinocytes increases ΔNp63 protein expression and the numbers of ΔNp63 positive cells in the basal layer of organotypic epithelial rafts.208 We therefore determined the effect of DEK loss in HNSCCs on ΔNp63.

Lentiviral DEK knockdown in UMSCC1 and 93VU147T cells decreased ΔNp63 protein levels

(Figure 3.5b). Furthermore, DEK loss in HPV positive or negative primary HNSCC cell populations also reduced ΔNp63 protein expression (Figure 3.5c) and IHC on consecutive

CCHMC-HNSCC1 human tissue demonstrates DEK and ΔNp63 are strongly expressed in similar areas of the tumor (Figure 3.5d). Reduction of ΔNp63 was not associated with decreased mRNA levels indicating that DEK dependent control of ΔNp63 levels is not mediated through transcriptional regulation (Figure 3.5e); a finding which implicates DEK in a post-transcriptional mechanism controlling ΔNp63 levels. To determine whether the observed ΔNp63 repression in

DEK-depleted cells was functionally relevant, HA-tagged human ΔNp63 was exogenously expressed in adenovirally DEK-depleted UMSCC1 cells to determine if the proliferation defect in

DEK deficient cells could be rescued by restoring ΔNp63 levels (Figure 3.5f). AdDEKsh infection alone resulted in significantly reduced cell growth similar to the effects of lentiviral knockdown in

Figure 4 and published data in cervical cancer cells.208 Expression of HA-ΔNp63 rescued the cellular growth defect in DEK-deficient cells, but did not increase the growth of DEK-proficient cells (Figure 3.5g). These data demonstrate that DEK expression in HNSCC cells is required for maintenance of ΔNp63 protein expression and indicate that DEK-dependent cellular growth in

HNSCC cells is at least in part mediated by ΔNp63.

50 Discussion

HNSCC is a devastating disease causing approximately 350,000 deaths per year.259

Thus, development of novel therapies and identification of clinically relevant disease biomarkers and molecular targets is of paramount importance. Current chemotherapeutic strategies result in major toxicity, and surgical resection leads to debilitating physiological side effects and permanent disfigurement. Although there have been promising advancements in the field, it is unlikely that any singular treatment will drastically increase survival, and early tumor detection remains a priority and major challenge. While HPV positive tumors are associated with an improved prognosis over their HPV negative counterparts, outcomes still remain less than optimal. Biological and molecular commonalities and differences between HPV positive and

HPV negative HNSCCs are only beginning to emerge.95, 130 Thus, studies of molecular pathways that mark and drive HNSCC need to be examined in parallel in these two biologically distinct subtypes. Therefore, our analyses of the expression and role of the human DEK protein were carried out in tumors that were carefully analyzed for the presence of high-risk HPV, with a focus on HPV16, together with associated patient and tumor characteristics. Expression of p16 is considered an excellent clinical marker of HPV positive tumor status. Indeed, positive staining for p16 was in agreement with the detection of HPV sequences by in situ hybridization in 70%

(7/10) and 80% (8/10) of the specimens at two testing sites. However, one of ten p16 positive specimens (CCHMC-HNSCC13) was HPV16 negative by three different analyses, suggesting a need for independent confirmation of HPV status to avoid false positive results. Two specimens of unknown p16 status were additionally identified as HPV positive. All together, 61% of the

HNSCC tumor specimens analyzed were determined to harbor high risk HPV16 DNA sequences. Importantly, 100% of the tumor samples showed strong DEK over-expression regardless of HPV status. However, it is important to note that our studies were primarily limited to oropharyngeal HNSCC tumors and may not represent HNSCCs as a whole. Comparisons

51 with published DEK expression studies in breast cancer suggest some notable distinctions in

HNSCCs which need to be investigated further. One report on breast cancer by our laboratory demonstrated that the proportion of DEK-expressing cells, but not per-cell staining intensity, increases with disease progression from normal breast epithelium to hyperplastic, and finally to malignant tumors.204 A second report by Liu et al. associated increasing DEK expression levels with increasing tumor stage.219 We note that for stage I-III specimens, the proportion of DEK positive tumor cells, as well as the staining intensity of the DEK oncogene, were already maximal (Table 3.1), thus suggesting that DEK may be particularly useful as a biomarker to detect early stage HNSCCs where identification is most beneficial for the patient. Regarding the molecular mediators driving DEK protein over-expression, transcriptional up-regulation of DEK message has been reported for a multitude of human tumors and has been ascribed to distinct transcriptional mechanisms in different cell and tumor types.209, 212, 213 In HPV positive cervical cancer cells, DEK expression is driven by the high risk HPV E7 oncoprotein through Rb pocket protein dependent mechanisms. It is likely that in HPV positive HNSCCs, E7-dependent DEK regulation occurs in a similar fashion, whereas in their HPV negative counterparts, Rb mutations might be at play.

A requirement for Dek expression in genetic mouse models of cancer, and specifically during SCC development in vivo, has remained unexplored. We utilized an established E7- driven, 4-NQO dependent mouse model of HNSCC and demonstrate that Dek expression is important for the proliferation of epidermal keratinocytes and for sustained tumor cell growth.

The presence of microscopic tumors in the Dek knockout tongues shows that Dek is not required for tumor initiation in this 4-NQO-dependent system. The fact that Dek loss attenuated cellular proliferation specifically in the E7 oncogene expressing, but not in control epidermis, indicates the existence of a therapeutic window for targeting DEK. DEK depletion may represent a reasonably tumor specific approach that can be exploited for tumor prevention and cancer therapy in HNSCC patient populations and beyond. Along these lines, a degree of specificity for

52 targeting DEK functions has been supported by earlier in vitro studies, where DEK expression was highest in HPV positive cervical cell lines, lower in normal, primary keratinocytes, and lowest in differentiated keratinocytes.208, 209 Broad HNSCC cancer cell addiction to DEK over- expression was observed in both HPV positive and negative HNSCC lines. Previously published data demonstrate that in HeLa cells, DEK loss was associated with apoptosis and senescence by a mechanism that at least partially involved up-regulation and activation of p53.210 Similarly,

DEK depletion in HPV positive HNSCCs led to p53 up-regulation. However, in light of the fact that HPV negative HNSCCs respond to DEK knockdown, we sought to identify p53-independent

DEK effectors.

Expression of the ΔNp63 member of the p53 protein family has been implicated in the maintenance of stem and progenitor cells in various epithelia. In fact, ΔNp63 was recently shown to be essential for reprogramming mouse embryonic fibroblasts as measured by pluripotent stem cell phenotypes such as colony formation by alkaline phosphatase stain and teratoma formation.260 Such activities in cellular stemness are supported by the restricted expression of ΔNp63 in the basal cell layer of the epidermis where keratinocyte stem and progenitor cells reside. On the other hand, ΔNp63 has been implicated in triggering terminal differentiation.261, 262 The role of ΔNp63 in stem cell functions is therefore complex, and the extent to which ΔNp63 regulates stem cell proliferation, differentiation or a combination of both remains controversial. Apart from pluripotent stem cells, ΔNp63 is now recognized as a somatic stem cell marker in other types of epithelium, highlighting the importance of this protein in multiple developmental contexts.263 p63-/- mice are well characterized for their severe developmental defects at birth including the lack of appendages, squamous epithelium, and hair follicles, and not surprisingly, recent work implicates the loss of the ΔNp63 isoform as the largest contributor to these phenotypes.264-266 Not surprisingly, maintaining appropriate levels of

ΔNp63 is critical for preserving epithelial integrity and suppressing transformation. Indeed,

53 overexpression of ΔNp63 is considered oncogenic in HNSCC, and p63 amplification and mutation have been reported in a subset of HNSCCs.142, 256, 257 Regulation of ΔNp63 expression is complex and can occur at both transcriptional and post-transcriptional levels. Here, we report that DEK controls ΔNp63 expression via a post-transcriptional mechanism. Importantly, exogenous ΔNp63 expression in DEK-depleted cells rescued cancer cell growth demonstrating that DEK-dependent ΔNp63 expression is functionally important for promoting HNSCC growth.

DEK activities can thus be placed upstream of ΔNp63, where targeting DEK results in decreased levels of this recognized oncogene that contributes to SCC pathogenesis.256 Future work will examine the mechanism(s) whereby DEK regulates ΔNp63. Given that ΔNp63 expression in DEK deficient tumor cells did not completely restore proliferation to the level of control cells, we cannot rule out the possibility that other DEK effectors exist and contribute to tumor cell growth in vitro and in vivo. Taken together, our results suggest the existence of a new ΔNp63-DEK signaling axis, which promotes proliferation in HNSCC.

Materials and Methods

Cell culture. HPV negative UMSCC1 and UMSCC6 head and neck cancer cells were maintained in Dulbecco’s modified Eagle medium (DMEM, Gibco, New York, USA) supplemented with 1% hydrocortisone, 10% fetal bovine serum (FBS), antibiotics and antifungals. HPV positive 93VU147T head and neck cancer cell lines were maintained in the same media but without hydrocortisone.

Lentiviral transduction. Cells were transduced at 30% confluency with lentiviral pLKO.1 non- targeting control (NTsh) or pLKO.1 DEK832 (DEKsh2) (Sigma-Aldrich Mission shRNA library) as previously described.186, 204 A final concentration of 8 µg/mL polybrene was added for each transduction. Cells were selected and carried in 1ug/mL of puromycin for primary cultures or in

1.25 µg/mL for established cell lines, and were independently transduced for each experiment.

54 Adenoviral transduction. Cells were transduced at 40% confluency with adenoviral control vector (AdGFP) or adenoviral DEK knockdown (AdDEKsh) at 10 infectious units per cell as previously published.210 Protein lysates were collected 72 hours post-infection.

Primary human head and neck cancer specimens. Human HNSCC tissues were freshly obtained with IRB approval at the time of surgical resection or biopsy. Chart review was performed to determine patient demographics, risk factors and histopathological characteristics.

Tissues were immediately placed in F-media.208 A portion of the tissues was fixed in 4% paraformaldehyde and embedded in paraffin for immunohistochemical analysis. A second portion of each specimen was also cultured in F-media on irradiated J2-3T3 mouse fibroblasts for primary cell culture as described previously.267 Resulting cell populations were considered primary if passaged fewer than 15 times. Tissue sections were subjected to hemotoxylin and eosin (H&E) staining to confirm the presence of squamous cell carcinoma.

HPV detection. For in-house HPV analysis (Site 1), tissue sections were deparaffinized and subjected to in situ hybridization using standard DNA probes specific for 12 high risk HPV types including 16 and 18 types provided by Ventana Medical Systems (INFORM HPV III Family 16

Probe (B)). Sections were then incubated with a fluorescein-tagged DNA probe and counterstained using the automated Ventana BenchMark instrument (Ventana Medical System,

Tucson, AZ, USA). Further HPV detection by in situ hybridization and real-time TaqMAN PCR assays (Site 2) was previously described by the Gillison laboratory.268 Briefly, all 21 CCHMC-

HNSCC specimens were evaluated for HPV16 using a biotinlyated probe (Genpoint, Dako,

Copenhagen, Denmark) and were considered positive based on nuclear staining in tumor cells.

Real-time PCR was performed on formalin-fixed, paraffin-embedded specimen (FFPE) extracted DNA, and interrogated for HPV16 sequences. Human endogenous retrovirus-3

(ERV3) was used as the internal control. Threshold for positivity is any value above 1 HPV copy as determined by calculating the ratio of HPV16/ERV3 and accounting for total percentage of tumor present.

55 Evaluation of patient clinical characteristics. Chart review was performed to extract relevant clinical information for each patient. For Table 3.1, any patient who currently smokes or who smoked previously was designated as a smoker. Alcohol consumption was assigned to patients who self-reported moderate to heavy drinking and not social or occasional drinkers. Patients who answered no to either smoking or drinking were designated as such. Expression of p16, a marker of HPV positivity, was also noted if clinically determined. Quantitation of DEK positive staining by immunohistochemistry was carried out as follows: The number of cells with positive staining for DEK was quantified as: 3= >90% positive tumor cells, 2= 10-50% positive tumor, 1=

<10% positive tumor cells. The overall intensity of DEK staining was determined as W (weak), V

(variable) or S (strong) as compared to surrounding non-neoplastic squamous epithelium. The overall DEK status is then labeled by a combination of these values. For instance, ‘3S’ in Table

3.1 indicates this tumor specimen is maximally positive for DEK staining intensity and for the proportion of DEK-expressing cells.

Transgenic and knockout mouse models. The K14E7 head and neck cancer mouse model and Dek-/- mouse were previously described.253,207 Dek−/− mice were backcrossed into an F/VBN background and E7 was maintained in a heterozygous state. For generation of K14E7 Dek−/− mice, female Dek−/− mice were crossed with K14E7 males. The following primer sets were used for genotyping: Dek wild-type allele (5’-CGA ACT CGT GAA GAG GAT CTT GA-3′, 5’-ATG

TGT CAG GCT GCA TCT CCA ATG-3’), Dek knockout allele (5’-ATC CAT CAT GGC TGA TGC

AAT GCG-3′, 5’-TGG AAG GTA AAG AGT GGC CCT TA-3’), E7 transgene (5’-ACT CTA

CGC TTC GGT TGT GCG TA-3’, 3’-GCA CAC AAT TCC TAG TGT GCC CAT-5’). Mouse use and handling was carried out in accordance with the American Association for Accreditation of

Laboratory Animal Care and the Cincinnati Children's Hospital Veterinary Care Facility according to a protocol approved by the Institutional Animal Care and Use Committee to S.I.

Wells.

56 Head and neck cancer induction. Eight week old transgenic and non-transgenic mice were administered the chemical carcinogen and tumor initiator 4-NQO in the drinking water at 10

µg/ml as previously published251 for a period of 16 weeks and then returned to normal water for

8 weeks. One hour prior to sacrifice, mice were injected intraperitoneally with 6µg/ gram of mouse weight of bromodeoxyuridine (BrdU) (BD Biosciences, San Jose, CA, USA). Mice were analyzed either upon premature death, or at 24 weeks after timed sacrifice. Sacrifice occurred upon moribund appearance due to the tumor burden which prevented the mice from eating and drinking. In the case of the K14E7 Dek-/- mice, two of three mice died prematurely from unknown causes. Gross and microscopic tumors were recorded. Tongue and esophagus were collected for histological analyses.

Histological analysis and immunohistochemistry. Mouse tissues were fixed in 4% paraformaldehyde, embedded in paraffin, sectioned at 5 µm thickness, and fixed onto slides.

Routine H&E stained sections were analyzed for histopathology.207 Paraffin sections were deparaffinized in xylene and rehydrated for antigen retrieval in sodium citrate. Sections were then treated with the Mouse on Mouse peroxidase immunostaining kit (Vector Labs,

Burlingame, CA, USA). Sections were stained with diaminobenzidine (DAB) and counterstained with Vector Laboratories hematoxylin QS or Nuclear Fast Red (Poly Scientific, Bay Shore, NY,

USA) and mounted with Permount (Fisher Scientific, Pittsburgh, PA, USA). Staining was detected with a Leica DM2500 microscope and LAS software (Leica Microsystems Inc., Buffalo

Grove, IL, USA), and images were captured at the indicated magnifications. Antibody dilutions were used as follows: BrdU (1:100, Invitrogen), p63α (1:100, Santa Cruz (H-129), Dallas, TX,

USA) and DEK (1:50 and 1:100, BD Transduction).

Western blotting. Whole cell lysates were harvested with RIPA buffer containing protease and phosphatase inhibitors or Laemmli buffer containing phosphatase inhibitors. 20 µg total protein was analyzed by western blotting as previously described.210 Membranes were probed with DEK

(1:1000, BD Biosciences), p53 (1:1000, Cell Signal (1C12), Boston, MA, USA), phospho-

57 serine15 p53 (1:1000, Cell Signal), ΔNp63 (1:200, Santa Cruz (4A4)) or actin (1:10,000; a gift from James Lessard, Cincinnati Children’s Hospital Medical Center).

Growth curves. Following selection, lentivirally transduced cells were plated in triplicate at an equal density of 100,000 cells/well. The cells were then trypsinized, counted on a hemacytometer and total cell number was calculated for each time point. Data represent two independent experiments.

Flow cytometry. BrdU incorporation was assessed as described previously.213 Briefly, HNSCC cells were plated in equal numbers and labeled with 10 µM BrdU for 2 hours. Collected cells were prepared following the BD Pharmingen APC BrdU Flow kit instructions, analyzed on a BD

FacsCalibur, and data further analyzed with FlowJo software (Tree Star, Ashland, OR).

Samples were analyzed in triplicate and data represents two independent experiments.

Quantitative RT-PCR. mRNA was harvested using Trizol reagent (Invitrogen, Carlsbad, CA,

USA) from transduced cells immediately following puromycin selection. 1 µg of RNA was reverse transcribed to cDNA using the QuantiTect Reverse Transription kit (Qiagen, Valencia,

CA, USA). cDNA expression was detected using SYBR green master mix (Applied Biosystems,

Carlsbad, CA, USA) on an ABI7300 Real Time PCR machine (Applied Biosystems). Data was analyzed using the ΔΔCt method. Primers were as follows: (1) ΔNp63 Forward: 5’-

GGAAAACAATGCCCAGACTC-3’ Reverse: 5’-GTGGAATACGTCCAGGTGGC-3’ (2) DEK

Forward: 5’-TGT TAAGAAAGCAGATAGCAGCACC-3’ Reverse: 5’-

ATTAAAGGTTCATCATCTGAACTATCCTC-3’ (3) GAPDH Forward: 5’-

GGTCTCCTCTGACTTCAACA-3’ Reverse: 5’-ATACCAGGAAATGAGCTTGA-3’. ΔNp63 primers were previously published.269

Cellular proliferation rescue experiments. 400,000 cells were plated and transduced with adenovirus on the following day, as described above. Twenty-four hours post-infection, cells were transfected in duplicate with 4 ug of the control (pcDNA) or ΔNp63 construct (human HA-

58 ΔNp63), along with 1 ug of dsRed as a transfection efficiency control. DNA was added to Opti-

MEM (Gibco), followed by DharmaFECT 1 Transfection Reagent (Fisher) and added drop-wise to the cells. Total cell counts were determined by hemacytometer counts 72 hours post- infection. This represents three independent experiments.

Statistics. Statistical significance was calculated using GraphPad Prism 6 software (La Jolla,

CA, USA). Student’s two-tailed t-test was used for experiments where noted, with **=p .01 and

***=p .001. ≤

≤

Acknowledgements

This work was supported by the Clinical Scientist Training Program at the University of

Cincinnati (TWD), Public Health Service Grant R01 CA116316 from the NIH (SIW), NIH Training

Grant T32ES007250 (AKA) and R00 CA160639 from the NIH (RJK). All flow cytometric data were acquired using equipment in the Research Flow Cytometry Core in the Division of

Rheumatology at Cincinnati Children’s Hospital Medical Center, supported in part by NIH AR-

47363, NIH DK78392, and NIH DK90971. We would like to acknowledge the CCHMC pathology core with excellent technical support by Meredith Taylor for Ventana HPV in situ hybridization,

Dr. Madhavi Kadakia for providing the HA-ΔNp63 construct, Marie Matrka for providing protein lysates, and expert technical support by Dr. Ronald Waclaw for use of his Leica microscope.

59

Figure 3.1. Dek loss protects from HNSCC tumor promotion in vivo. (a) Dek loss prolongs survival in a murine model of HNSCC. Nine of eleven K14E7 Dek+/+ were sacrificed prematurely due to moribund appearance, whereas only three out of seven K14E7 Dek-/- mice were sacrificed early. (b) Immunohistochemistry depicting a K14E7 Dek+/+ mouse tumor that stains positive for DEK, with arrows representing examples of positive cells. (c) Mice were sacrificed at 24 weeks for gross macroscopic and microscopic tongue and esophagus lesion analysis.

Tumors grade was determined histologically as previously describe for this 4-NQO-driven

HNSCC models270 and representative images are shown. One K14E7 Dek-/- mouse had microscopic tumors in both the tongue and esophagus, hence the number of microscopic

60 tumors in this group totals 8 for the 7 mice. Original magnifications are at 200x. p=.0893 as determined by log-rank (Mantel-Cox) test.

61

Figure 3.2. Dek loss attenuates E7-driven proliferation in tongue epidermis. (a) H&E stained images of Dek-proficient and Dek-deficient E7-trangenic tongue epidermis showing epithelial hyperplasia are shown at a total magnification of 200x. (b) Mice were injected with

BrdU prior to sacrifice, and immunohistochemistry done to detect BrdU positive cells in the pre- malignant tongue epidermis. Representative images with arrows denoting BrdU positive cells are shown. The percentage of epidermal BrdU positive cells was determined by counting >1000 cells in sections from 4 K14E7 Dek+/+, 4 K14E7 Dek–/–, 4 nontransgenic Dek+/+, and 3 nontransgenic Dek–/– mice. Tongues from K14E7 Dek+/+ mice harbored significantly more BrdU positive cells as compared to K14E7 Dek-/- mice. **=p≤0.01 by two-tailed t-test.

62

Clinical characteristics of primary human tumor samples Table 3.1

63

Figure 3.3. DEK protein expression is universally up-regulated in human HNSCC. (a)

Tumor specimens were sectioned, H&E stained for morphology to confirm the diagnosis of

HNSCC, and subjected to in situ hybridization for dectection of high-risk HPV sequences.

Representative HPV positive (Specimen B) and HPV negative (Specimen A) tumors are shown.

Strong DEK expression was observed in both HPV positive and negative tumors as compared to adjacent non-neoplastic epithelium. (b) The percentage of DEK positive cells on each section was estimated and staining intensity determined as described in the Materials & Methods.

Representative images and scores are shown along with the corresponding H&E image.

Original magnification 200x; insets 1000x.

64

Figure 3.4. DEK knockdown attenuates the proliferation of HPV positive and negative

HNSCC cell lines. (a-c, left panels) HPV positive and HPV negative cell lines were lentivirally transduced with control or DEK-specific shRNA, selected in puromycin, and plated in equal cell numbers for cell counts taken over the course of four days. Western blot analysis indicates DEK knockdown is below the level of detection in all cell lines, with actin used as a loading control.

DEK loss substantially decreased total cell numbers on day 3 and 4, regardless of HPV status.

Data represents the average of two independent experiments with each sample measured in triplicate. Error bars indicate standard deviation (SD). (a-c, right panels) Control and DEK

65 knockdown cells were incubated with BrdU for 2 hours and BrdU positive cells determined by flow cytometry. Representative contour plots show BrdU positive gating for NTsh and DEKsh cells. Data represents the average of two independent experiments with each sample measured in triplicate. Error bars indicate standard deviation (SD).

66

Figure 3.5. DEK promotes cellular growth by up-regulating ΔNp63. (a) Western blot analysis using CCHMC-HNSCC1 cell lysates shows phospho-ser15 p53 and total p53 up- regulation after DEK knockdown with shRNA as compared to control shRNA. (b) Western blot analysis of UMSCC1 and 93VU147T cells shows reduced ΔNp63 protein levels in DEKshRNA compared to control shRNA lentiviral infected cells and similarly with adenovirally infected

UMSCC1 cells. (c) Primary human-derived cell populations CCHMC-HNSCC1 (HPV positive) and CCHMC-HNSCC3 (HPV negative) also show reduced levels of ΔNp63 with DEKshRNA as compared to NTsh controls, with densitometry depicted in the graph to the right for CCHMC-

67 HNSCC1. (d) Immunohistochemistry for DEK and ΔNp63 have similar expression patterns in the CCHMC-HNSCC1 tumor specimen. Dashed boxes indicate expanded area with representative pictures below. Original magnifications at 200x. (e) qRT-PCR indicates ΔNp63 mRNA levels are not reduced with DEK loss in either cell line. GAPDH was used as the internal control with relative expression determined by the ΔΔCt method. (f) UMSCC1 cells were transduced with AdDEKsh or control AdGFP adenovirus and transfected with control (pcDNA) or human HA-ΔNp63 constructs on the following day. dsRed was used in all samples as a transfection efficiency control. DEK and ΔNp63 expression levels were verified by western blot analysis. (g) Viable cells from panel (f) were counted on day 3. Exogenously expressed ΔNp63 increased the total number of cells in the presence of AdDEKsh but not in control cells. Error bars indicate standard error of the mean (SEM) from 3 independent experiments. NS=not significant, **=p .01 and ***=p .001 as determined by paired two-tail t-test.

≤ ≤

68 Chapter 4: IRAK1 promotes HNSCC survival through a DEK- regulated network

Allie K. Adams1, Lyndsey C. Bolanos2, Phillip J. Dexheimer3, Rebekah A. Karns3, Bruce J. Aronow 3, Kakajan Komurov2, Anil Jegga3, Keith A. Casper4, Yash J. Patil4, Keith M. Wilson4, Daniel T. Starczynowski2, and Susanne I. Wells1#

1Division of Oncology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, USA

2Division of Experimental Hematology and Cancer Biology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, USA

3Division of Biomedical Informatics, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, USA

4Department of Otolaryngology, Head and Neck Surgery, University of Cincinnati, Cincinnati, OH, USA

#Corresponding author

Manuscript in preparation for submission to Oncogene.

69 Abstract

The chromatin-binding DEK protein was recently reported to promote the growth of

HPV+ and HPV- head and neck squamous cell carcinomas (HNSCCs). However, the relevant cellular and molecular mechanism(s) remain poorly understood. While DEK is known to activate and repress specific transcriptional targets, the global DEK-dependent transcriptome in HNSCC is unknown. In order to identify DEK transcriptional signatures and functional targets, we carried out RNA-Sequencing (RNA-Seq) in head and neck cancer cell lines that were either proficient or deficient for DEK. Bioinformatic analyses and subsequent validation revealed that the interleukin-1 receptor activated kinase (IRAK1) was significantly repressed upon DEK knockdown. IRAK1 is best known for its role in inflammatory signaling, and was previously shown to have tumor suppressive activities in oral squamous cell carcinoma. Surprisingly, datamining of The Cancer Genome Atlas (TCGA) revealed 14% of HNSCC patients overexpress IRAK1, thus supporting possible oncogenic activities. Furthermore, IRAK1 knockdown in HPV+ and HPV- HNSCC cell lines was sufficient to increase HNSCC cell death by apoptosis and to reduce signaling through downstream pathways such as ERK1/2. Finally, targeting DEK and IRAK1 simultaneously enhanced cell death over the targeting of either. We propose a model wherein IRAK1 stimulates oncogenic signaling and phenotypes both independently and in conjunction with DEK.

70 Introduction

Head and neck squamous cell carcinoma (HNSCC) is a disease comprised of two distinct entities: human papillomavirus (HPV) positive and HPV negative. HPV- disease is attributable to tobacco and alcohol use, and its declining incidence in the US has been ascribed to the well-publicized health risks of these activities. In stark contrast, HPV+ disease is on the rise, particularly in younger patient populations.7 While improved response to traditional chemotherapies and thus favorable long-term survival is observed in HPV+ patients, prognoses remain grim for patients with advanced and metastatic tumors.94 Furthermore, major quality of life issues arise due to treatment-related tissue damage.271 Therefore, the need for novel therapeutic targets and biomarkers cannot be underestimated. Our previous work highlighted the oncogenic functions of DEK in in both HPV+ and HPV- human HNSCCs, wherein DEK was highly overexpressed and required for optimal growth and proliferation.272 Dek loss of function in mice attenuated the proliferation of HPV16 E7 expressing, but not normal, epidermis and inhibited overt tumor growth in a chemically induced model of HNSCC. In view of the observed specificity of DEK targeting for premalignancies and malignancies, this molecule has been reported as a potential therapeutic target. However, DEK-dependent signaling pathways and molecular mediators of DEK-dependent tumor phenotypes in HNSCC are limited. Our previous work implicated ΔNp63 as a down-stream DEK target that regulated DEK-dependent proliferation. In our studies we aimed to identify additional signaling pathways dependent on

DEK expression.

DEK is well published for its importance in various cancer types, including breast and bladder cancer, melanoma, and most recently, HNSCCs.200, 204, 205, 213, 272 This is a versatile nuclear protein, with functions that range from chromatin modifier and histone chaperone to modulator of DNA repair, replication, and transcription.190, 194, 195, 199 For example, DEK represses transcription in leukemia cells through inhibition of p300 and P/CAF.187 DEK also

71 activates transcription via interaction with AP-2α in glioblastoma.189 Although DEK has been published as a co-activator or co-repressor of transcription in various systems, transcriptome data to determine the role of DEK in global transcriptional regulation in solid tumors is scarce.188,

189, 192

In this study, we performed transcriptome profiling by RNA-Seq to define the broader

DEK-dependent transcriptome in HNSCC cells and to identify novel targets, which may themselves be functionally relevant for the growth of these tumors. We focused on both subsets of HNSCC, HPV- and HPV+, to identify targets that may be beneficial to patients regardless of

HPV status. Following gene ontological analysis, biological processes involved in the immune response were strongly implicated. DEK has previously been published as an autoantigen in autoimmune diseases and it can function as a pro-inflammatory protein, suggesting it may regulate inflammatory signaling.183, 227 We pursued the IRAK1 serine/threonine kinase, which mediates signaling from the toll-like receptor (TLR) and interleukin-1 receptors (IL1R).242 The

IRAK1 cascade includes the E3 ubiquitin ligase TRAF6, which engages, among other pathways,

NF-κB and MAPK signaling. IRAK1 was recently implicated as a novel therapeutic target in myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML), but its function in most solid tumors remains unknown.245 We first characterized expression patterns of IRAK1 in

HNSCC with data from The Cancer Genome Atlas (TCGA) and found it is overexpressed by genomic amplification and transcriptional up-regulation in a portion of tumors. IRAK1 inhibition attenuated downstream signaling through TRAF6 and increased apoptosis, suggesting IRAK1 inhibition may be a new therapeutic avenue for patients with HNSCC. Finally, DEK and IRAK1 contributed to HNSCC survival independently, and targeting them jointly enhanced HNSCC cell death over the targeting of either. Taken together, these data reveal IRAK1 as a component of the DEK transcriptome, and a targetable oncogene in HNSCC.

72 Results

Profiling the DEK-dependent transcriptome in HNSCC

Little is known about the global impact of DEK loss on gene expression, and relevant transcriptional targets are largely uncharacterized.192 DEK plays dual roles in transcription, either as a co-activator or a co-repressor depending on the cellular context, and was recently published to bind transcriptional start sites of some of the activated or repressed target genes.194 In order to define the consequences of DEK loss on transcription in HNSCC, we used a well-established lentiviral approach previously published in this model system.272 The HPV- and HPV+ cell lines, UMSCC1 and UMSCC47, respectively, were transduced with DEK versus control knockdown vector and selected. Successful knockdown was confirmed by western blot analysis (Figure 4.1A). Messenger RNA (mRNA) was collected and subjected to RNA-Seq.

Independent analyses were performed on UMSCC1 and UMSCC47 RNA-Seq data to identify fold changes for differentially expressed genes. Venn diagrams highlight over 2,000 genes that were differentially expressed upon DEK loss and similarly regulated in these cell lines (Figure

4.1B). Ontology analyses revealed clear trends in altered biological processes. Among the significantly regulated processes identified in this way was the immune response (Figure 4.1C).

DEK has previously been implicated in immune system regulation, possibly in part through

NFκB signaling, but few DEK targets have been described as mediators of these processes.227

In addition to ontological analysis, we assessed common transcription factor binding sites among the overlapping gene set, and found new DEK targets, along with published ones such as p53 (Supplementary Tables S4.1-4.3).210 To further define those targets, we used network analysis software to highlight nodes of interest. A portion of the down-regulated gene network for UMSCC1 cells is shown (Figure 4.1D), with arrows highlighting two nodes of interest.

Expanded versions of this network map, along with others, are found in Supplementary Figures

S4.1-4.6. As expected, DEK was significantly down-regulated and connected to many other

73 genes of interest, including TNFAIP3, IL6, and MAPKs. As these genes are well established for their contribution to inflammation and immune signaling, but are downstream mediators of receptor-initiated signaling, we focused on IRAK1 that lies upstream. Since IRAK1 is a known regulator of toll-like receptor signaling in the innate immune system it may be a potential candidate for DEK-dependent processes.

DEK regulates IRAK1 mRNA and protein levels

IRAK1 is an important mediator of immune signaling, but little is known about its regulation outside of post-translational modifications. IRAK1 expression was decreased by 1.4- to 3.2-fold upon DEK loss, as determined by RNA-Seq (Figure 4.2A). To validate these findings from the RNA-Seq data, we utilized qRT-PCR in multiple cell lines (Figure 4.2B-4.2D) to confirm that IRAK1 expression was reduced following DEK depletion. As expected, IRAK1 mRNA levels were reduced in UMSCC1 and UMSCC47 cell lines, as well as an additional HPV- HNSCC cell line. Similarly, DEK knockdown in these cell lines reduced IRAK1 protein expression levels, along with known downstream MAPK signaling (Figure 4.2E). Reduced expression of IRAK1 in the absence of DEK was correlated with reduced expression of downstream pathway components, thus suggesting IRAK1 may a functionally relevant DEK target.

IRAK1 is over-expressed in primary HNSCC

Based on the observed IRAK1 transcriptional regulation in HNSCC cell lines, we probed available TCGA databases to determine possible IRAK1 alterations in primary HNSCCs. This data mining revealed IRAK1 is altered in 14% of HNSCCs, predominantly as a result of gene amplification or mRNA up-regulation (Figure 4.3A). This was observed in HPV+ and HPV- tumor subsets. To confirm that IRAK1 protein is expressed in HNSCC compared to normal tissue, we performed immunohistochemistry for IRAK1 on primary HNSCC tissue samples, which were previously described.272 Examples of HPV+ and HPV- specimens are shown (Figure 4.3B), with strong IRAK1 protein expression detectable in the cytoplasm as expected, and some additional

74 nuclear staining. This data suggests IRAK1 is overexpressed in HNSCC in line with possible oncogenic activities.

IRAK1 loss increases apoptosis in HNSCC

Having identified IRAK1 as a candidate effector in HNSCC we next sought to characterize its function in this tumor type. Previously, IRAK1 was proposed to have a tumor suppressive role in HNSCCs.273 Therefore, we aimed to determine the contribution of IRAK1 to

HNSCC phenotypes. We utilized a published IRAK1 shRNA construct to deplete IRAK1 levels in both HNSCC cell lines.245 IRAK1 knockdown resulted in decreased total and activated IRAK1, as measured by phosphorylation of residue Thr209, as well as reduced activation of downstream pathway components. These include a reduction in NF-κB (pIKKα/β) and MAPK

(p38 and ERK1/2) signaling (Figure 4.4A), both well-known cascades triggered by IRAK1 activation. Chemical inhibition of IRAK1 was also carried out with the IRAK-1/4 inhibitor which has been shown to increase apoptosis in melanoma in vitro and in vivo and to inhibit signaling and cell viability in MDS.245, 246 This inhibitor is a benzimidazole that is selective for IRAK1 and

IRAK4 and shows little specificity for other kinases.246 Similar to these published studies, 10 µM concentrations of IRAK-1/4 attenuated activation of IRAK1 at 24-72 hours post-treatment

(Figure 4.4B) in UMSCC1 and UMSCC47 cells. Activated phospho-IRAK1 then complexes with

TRAF6, which undergoes ubiquitination, thereby initiating downstream signaling cascades. To verify that signaling effects observed with IRAK1 loss were a result of a reduction in TRAF6- ubiquitination, TRAF6 was immunoprecipitated and subsequently probed for ubiquitin. TRAF6 ubiquitination was decreased in the absence of IRAK1, suggesting that NF-κB and MAPK signaling is mediated through TRAF6 (Figure 4.4C-4.4D). Finally, cellular proliferation and death were quantified upon IRAK1 inhibition in order to probe for a potential functional role in HNSCC cells. Significantly increased apoptosis was observed in the absence of IRAK1, either with shRNA or chemical inhibition (Figure 4E-4G). However, we did not observe any differences in

75 cell cycle profiles (Supplemental Figure S4.7). These results demonstrate that IRAK1 promotes the survival of HNSCC cells and that IRAK1 inhibition may be a novel therapeutic strategy to enhance cell death in this tumor type.

IRAK1 and DEK independently regulate HNSCC cellular survival

To assess if IRAK1 is functionally important downstream of DEK, and if these two proteins coordinate to regulate oncogenic phenotypes, IRAK1 was overexpressed in the presence and absence of DEK in an attempt to rescue deficiencies induced by DEK loss.

(Figure 4.5A). Interestingly, IRAK1 overexpression rescued phospho-ERK1/2 signaling (Figure

4.5A), but reconstitution of this pathway was not sufficient to rescue cell death (Figure 4.5B), cell cycle arrest (Figure 4.5C), or total cell number (Figure 4.5D) caused by DEK loss. This suggests DEK and IRAK1 independently contribute to HNSCC cell survival and that targeting both may improve apoptosis efficacy. We used a dual approach of infecting stably transduced

IRAK1 knockdown cells with adenovirus to deplete DEK (AdDEKsh). Either DEK or IRAK1 knockdown alone could induce apoptosis as expected, but the combined effect of IRAK1 shRNA with AdDEKsh infection was greater than that of the respective control cells (Figure 4.5E).

Taken together, these data support a model wherein DEK and IRAK1 function in parallel pathways that control apoptosis, and highlight an additive relationship that may be beneficial for therapeutic intervention (Figure 4.5F).

76 Discussion

A majority of patients with head and neck squamous cell carcinoma present at advanced stages of disease, contributing to the poor survival outcomes observed. These tumors also notoriously recur, despite aggressive treatment modalities including surgery, chemotherapy and/or radiation therapy, that have frequent side effects that can dramatically and permanently decrease patient quality of life. This suggests these cells have a high proliferative and survival capacity that is necessary for sustained growth of these tumors. Therefore, understanding the relevant targetable mediators of these phenotypes is of the utmost importance. Here we addressed this clinical need by first profiling the transcriptome of HNSCC cell lines that are dependent on the DEK oncogene. DEK is an important regulator of HNSCC growth, and is up- regulated in >90% of primary HNSCCs tested to date. Although some transcriptional DEK targets have been described, the DEK-dependent transcriptome in squamous cell carcinomas remains unknown. Ontology analysis revealed biological processes significantly altered in the absence of DEK, including adhesion, differentiation, immune regulation, and development. This is in contrast to transcriptional data from neuroendocrine carcinoma of the lung with DEK loss.

Shibata et al. revealed in their analysis that steroid metabolism, nucleosome assembly, and lipid synthesis and metabolism were altered most often in the absence of DEK. This suggests the effects of DEK loss in malignancies are cell-type dependent, which is similar to the divergence of gene expression observed in various normal cellular systems.

Here we have identified DEK-dependent gene expression that supports phenotypes previously defined, along with new systems of interest for future studies. Although DEK overexpression is published to promote the migration and invasion of breast and HNSCC cells, alterations in cellular adhesion have not been pursued.215 Our data suggests this may be one mechanism by which DEK promotes invasion. Furthermore, we have correlated the DEK- dependent transcriptome with common transcription factor binding sites that are associated with

77 DEK loss (Supplementary Tables S4.1-4.3). Many notable transcription factor sites and known targets of DEK were identified, including p53, CEBP, and p65 validating the importance of DEK in multiple cancer types.188, 191 For example, control of p53 binding sites was expected given that

DEK loss was reported to lead to the stabilization of p53.210 This correlated with the up- regulated induction of several p53 regulated genes including BCL11B, KRT15, and PIM1.

Herein we focused on the regulation of genes with roles in immune cell signaling. These included NF-κB and MAPK driven genes such as MAPK14, TNFAIP3, IL6, and IRAK1. We chose to probe the role of the IRAK1 serine/threonine kinase, a driver of inflammatory pathways in hematological disease, based on its function as a central signaling hub in the cytoplasm, and it is a targetable molecule in MDS and AML.

The role of IRAK1 in solid tumors has not been explored extensively, but our data suggest oncogenic and potentially targetable activities in HNSCCs. IRAK1 was transcriptionally up-regulated and amplified in a proportion of HNSCCs in the TCGA, in line with a newly discovered functional requirement for maximal survival of HPV positive and negative HNSCC cell lines. Such a role was uncovered through IRAK1 knockdown using stable lentiviral vectors, as well as through chemical inhibition. Downstream IRAK1 signaling was suppressed through

TRAF6, attenuating activation of NF-κB and MAPK and stimulated cancer cell death, thus highlighting the potential use of IRAK1 inhibitors in the treatment of HNSCC.

Our data identified IRAK1 as a component of the DEK-dependent transcriptome whose expression in HNSCC contributes to tumor cell survival. IRAK1 drives ERK1/2 signaling, but this alone was not sufficient to rescue cell growth in the absence of DEK. This is not surprising given the large network of genes DEK regulates and suggests multiple genes are required to maintain a proliferative state. It is also possible that the ERK1/2 signaling may be unimportant or may modify the response to DEK loss. Although IRAK1 was not by itself sufficient to rescue DEK- deficiencies, the combined targeting of DEK and IRAK1 demonstrated an additive relationship.

78 This additive effect emphasizes the large network of signaling hubs through which DEK functions, independent of IRAK1, and which can be further explored from the transcriptional data in future experiments.

Importantly, this work has defined IRAK1 as one functionally important driver of HNSCC survival. Interestingly, tumor suppressive functions of IRAK1 have been proposed in a recent publication in oral squamous cell carcinoma cells (OSCC). Hung et al. described miR-146a, a known regulator of IRAK1, as overexpressed in OSCC. Other publications define miR-146a as tumor suppressive, where its loss hyper-activates IRAK1 and may be one mechanism for IRAK1 overexpression.248 In OSCC, exogenous miR-146a expression increased orthotopic tumors and metastasis of SAS cells and reduced IRAK1 protein levels. In these same cells, IRAK1 knockdown combined with TRAF6 knockdown by siRNA increased invasion and tumor volume, but IRAK1 knockdown alone had few effects.273 Here, we utilized oropharyngeal HNSCC cells lines wherein IRAK1 surprisingly exhibited oncogenic functions in this tumor type. These opposing findings may be related to the site of origin for each cell line, as their primary focus was oral cancers compared to the more encompassing HNSCC. Additionally, the TCGA data wherein IRAK1 is largely overexpressed in HNSCCs supports the hypothesis that IRAK1 contributes to oncogenic phenotypes.273 Many possibilities exist to explain these discrepancies.

IRAK1 functions may be anatomically or cell line dependent or the method of inhibition may be important (acute (siRNA) versus stable (shRNA)). Additionally, IRAK1 expression may be a double-edged sword and careful balance of its expression must be maintained. TCGA data in other squamous tumors, such as cervical cancer, identified one patient with homozygous deletion of IRAK1 and two with truncated mutations. The other twenty cases where IRAK1 was altered were due to copy number amplification and mRNA up-regulation. Even in one HNSCC case from Figure 3, there is a patient with IRAK1 mRNA down-regulated. Overexpression versus loss of IRAK would then be a predictive marker of response to IRAK1 inhibitors and may be a fruitful biomarker for squamous cancers of multiple anatomical origins.

79 Materials & Methods

Cell culture. HPV negative UMSCC1 and UMSCC6, and HPV positive UMSCC47 head and neck cancer cell lines were cultured in DMEM (Gibco, New York, NY, USA) supplemented with

1% hydrocortisone (HPV- only), 10% fetal bovine serum, antibiotics and antifungals.

Lentiviral transduction. Cell lines were transduced with lentiviral pLKO.1 vectors for either nontargeting control shRNA (NTsh), IRAK1sh (TRCN0000000543, OpenBiosystems, Lafayette,

CO, USA), or DEK832 (DEKsh, Sigma-Aldrich Mission shRNA library, St Louis, MO, USA) in the presence of polybrene (8 µg/mL). Cells were selected in puromycin at a final concentration of 1

µg/mL.

Adenoviral transduction. Cells transduced with control (NTsh) and IRAK1 knockdown

(IRAK1sh) lentiviruses were plated in equal densities and kept under puromycin selection. 48 hours post-plating cells were transduced with control (AdGFP) or DEK knockdown (AdDEKsh) adenoviral vectors at 10 infectious units per cell as previously published.210 72 hours post- transduction, cells and media were collected and fixed to analyze for flow cytometry (see below). IRAK1 and DEK knockdown were confirmed by western blot analysis.

IRAK1 overexpression. UMSCC1 cells were lentivirally transduced with control (LeGo-iG2) or

IRAK1 overexpression (LeGO-IRAK1) vectors in the presence of 8 ug/mL polybrene. Cells were sorted based on GFP-positivity and expanded for experiments post-sorting. Creation of these vectors has been described previously.274

80 cBioPortal Analysis. The results depicted here are in whole or part based upon data generated by the TCGA Research Network: http://cancergenome.nih.gov.275, 276 For IRAK1 expression z-score thresholds were set at 2.0.

RNA-Sequencing. Transduced and selected UMSCC1 and UMSCC47 NTsh and DEKsh cells were collected and processed with a ZR RNA MiniPrep kit (R1064, Zymo Research, Irvine, CA,

USA), per kit instructions. A portion of the final RNA isolate for each sample was submitted for quality assurance prior to RNA-Sequencing. RNA-Sequencing was performed by the CCHMC

DNA Sequencing and Genotyping Core on an Illumina HiSeq2500 for single-end sequencing.

(Will insert GEO accession number once data is uploaded).

GeneSpring NGS Analysis. RNA-Seq files were imported into GeneSpring Multi-Omic Analysis

Software V12.6 (Agilent, Santa Clara, CA, USA) and sequences were aligned to the reference genome, hg19GFCh37, which efficiently aligns reads spanning known or novel splice junctions.

The reference annotations were produced by the Ensembl project.277 Aligned reads were filtered on base quality, with a quality threshold >=30. The aligned gene read counts were quantified and used to compute reads per kilobase per million reads (RPKMs) for each transcript in each sample. Raw counts were normalized using the DESeq algorithm and threshold set to 1.

Subsequent filtrations removed all genes with fewer than 3 reads. Fold change was calculated as DEKsh/NTsh with a cut-off of 1.4. Venn diagrams were created in GeneSpring and entity lists were translated from UMSCC1. Gene lists were submitted to ToppGene

(http://toppgene.cchmc.org) for ontological analysis.278

Network Analysis: All of the network analyses and visualizations were performed using the

279 NetWalker software and were described in detail previously.

81 Transcription factor binding site analysis. In order to identify enriched (p value <0.05) putative transcription factor binding sites within the up- and down-regulated genes, we mined the catalog of human, mouse, rat, and dog conserved regulatory motifs in promoters using the

ToppGene server.278, 280

Quantitative RT-PCR. RNA was collected with Trizol (Invitrogen, Grand Island, NY, USA) and reverse transcribed to cDNA using the QuantiTect Reverse Transcription kit (Qiagen, Valencia,

CA, USA). cDNA expression was detected with TaqMan Gene Expression Master Mix and probes (Applied Biosystems). Data was analyzed using the ΔΔCt method and values calculated relative to GAPDH. TaqMan probes were as follows: DEK (Hs01078267_m1), IRAK1

(Hs0108347_m1), and GAPDH (Hs02758991_g1).

Western blotting. Whole cell lysates were harvested using Laemmli buffer and a total of 20 µg of protein was analyzed as described previously.210 Membranes were probed with DEK (1:1000,

BD Biosciences, San Jose, CA, USA), IRAK1 (1:1000, Santa Cruz Biotechnology, Dallas, TX,

USA (sc-7883)), IRAK4 (1:1000, Cell Signaling Technologies, Danvers, MA, USA (4363)),

TRAF6 (1:1000, Santa Cruz (sc-7221)), phospho-IRAK1 (Thr209, 1:800, Assay Biotechnology

Company (A1074), Sunnyvale, CA, USA), phospho-p38 (1:500, Cell Signaling Technologies

(4361)), p38 (1:1000, Cell Signaling Technologies (9212)), phospho-ERK1/2 (1:1000, Cell

Signaling Technologies (4377)), ERK1/2 (1:1000, Cell Signaling Technologies (4695)), phospho-IKKα/β (1:500, Cell Signaling Technologies (2697)), IKKα (1:1000, Cell Signaling

Technologies (2682)), actin (1:10,000 a gift from James Lessard), and GAPDH (1:1000, Cell

Signaling Technologies (5174)).

Flow cytometry for Cleaved Caspase-3. Lentivirally transduced cells were plated at equal densities and collected 48 hours later. Cells were fixed and prepped following the BD FITC

82 Active Caspase-3 Apoptosis kit protocol (BD Biosciences). Adenovirally transduced cells were prepped with an Alexa-Fluor 647 conjugated cleaved-caspase 3 antibody to account for GFP- positivity (Cell Signaling Technologies (9602)). Analysis was performed on a BD FacsCanto and data analyzed on FlowJo software (Tree Star, Ashland, OR, USA). Experiments were performed

3 times with standard error of the mean (SEM) represented.

Flow cytometry for cell cycle analysis. Lentivirally transduced cells were plated in equal numbers and 48 hours later were pulsed with 10 µM BrdU for 45 minutes. Cells were collected and prepped following the BD Pharmigen APC BrdU flow kit and analyzed on a BD FacsCanto.

Data was analyzed on FlowJo Software as above. Experiments were performed 3 times and

SEM represented.

IRAK1 inhibitor. UMSCC1 cells were plated at equal densities and IRAK1 inhibitor (IRAK-1/4 inhibitor, I5409, Sigma-Aldrich, St. Louis, MO, USA) added the same day for caspase-3 flow cytometry experiments. DMSO was added in equal volume for a control. Cells and media were collected 72 hours later and prepared and analyzed as above. For the time-course experiment, cells were plated and inhibitor was added the following day, with protein collected at 24, 48, and

72 hours and analyzed by western blot for IRAK1 inhibition.

Immunohistochemistry. Primary human tissue specimens and staining protocol were previously described272 but with IRAK1 antibody (1:50, sc-7883, Santa Cruz). Images were captured with a Leica DM2500 microscope and LAS software (Leica Microsystems Inc., Buffalo

Grove, IL, USA) at the indicated magnifications.

TRAF6 Immunoprecipitation. Samples were lysed using RIPA buffer containing protease and phosphatase inhibitors and protein concentration determined using BCA Protein Assay Kit

83 (Pierce 23225). 600 µg of each sample was used to perform TRAF6 immunoprecipitation.

Lysates were incubated with A/G beads (sc-2003, Santa Cruz, Dallas, TX, USA) and incubated with TRAF6 antibody (sc-7221, Santa Cruz). Samples were loaded onto a 4-15% gradient Mini-

PROTEAN TGX Precast Gel (BioRad, Hercules, CA, USA) and proteins separated by SDS-

PAGE electrophoresis. Membranes were probed with ubiquitin primary antibody (sc-8017,

Santa Cruz). Protein from the original lysis, prior to immunoprecipitation, was run following the above western blot protocol. Membranes were probed with IRAK1, TRAF6, and GAPDH.

Densitometry was performed using ImageJ software.

Growth Curves. Control and IRAK1 overexpressing cells were plated at equal densities, in triplicate, and total cell number counted over 3 days. Experiments were performed twice with SD represented.

Statistics. Statistical analysis was performed using GraphPad Prism 6 software (La Jolla, CA,

USA). Student’s t-test was used to calculate p-values, where *=p .05 and **=p .01.

≤ ≤

Conflict of Interest. The authors declare no conflict of interest.

Acknowledgements. We are grateful to Dr. James Lessard of CCHMC and SevenHills

Bioresearch (Cincinnati, OH, USA) for the gift of the C4 pan-actin monoclonal antibody used in this work. Thank you to Eric Smith and Lisa Privette Vinnedge for critical reading of this manuscript. All flow cytometric data were acquired using equipment maintained by the

Research Flow Cytometry Core in the Division of Rheumatology at Cincinnati Children’s

Hospital Medical Center, supported in part by NIH AR-47363, NIH DK78392 and NIH DK90971.

Support for this work was provide by the NIH: T32ES007250 (AKA) and Public Health Service

Grant R01 CA116316 (SIW).

84

Figure 4.1. Profiling the DEK-dependent transcriptome in HNSCC. (a) UMSCC1 and

UMSCC47 were lentivirally depleted for DEK, as confirmed by western blot analysis, and mRNA from these lines was submitted for RNA-Sequencing (RNA-Seq). (b) GeneSpring NGS analysis was performed on genes differentially expressed 1.4 fold or greater (DEKsh/NTsh). GeneSpring derived Venn diagrams depict genes altered in both cell lines and overlap highlights genes common to both UMSCC1 and UMSCC47. (c) Overlapping genes (from panel B, 2282) from

UMSCC1 and UMSCC47 were analyzed on TOPPGene to identify significantly altered biological processes. The top 10 biological processes are represented. (d) A portion of the down-regulated gene networks analyzed on NetWalker highlights nodes containing IRAK1 and

DEK in UMSCC1 cells. A full view of view can be found in Supplementary figure S4.2.

85

Figure 4.2. DEK regulates IRAK1 mRNA and protein levels. (a) Table depicts fold changes for DEK and IRAK1 in UMSCC1 and UMSCC47 from GeneSpring NGS analysis. (b, c, d)

IRAK1 mRNA is reduced following DEK depletion. IRAK1 and DEK mRNA levels were confirmed by TaqMan PCR to validate RNA-Seq results in three cell lines: UMSCC1,

UMSCC47, and UMSCC6 (HPV negative). Experiments were performed twice and standard deviation (SD) depicted. (e) Western blot analysis confirms IRAK1 protein levels and downstream signaling are also depleted in the absence of DEK.

86

Figure 4.3. IRAK1 is overexpressed in primary HNSCC. (a) cBioPortal analysis of the TCGA

HNSCC database reveals IRAK1 alterations occur in 14% of HNSCC. A total of 279 samples were analyzed and were further broken down into HPV- and HPV+ subsets. (b) IRAK1 is highly expressed in primary HNSCC tissues. CCHMC-HNSCC1 (HPV+) and CCHMC-HNSC18 (HPV-) were stained for IRAK1 by immunohistochemistry, with intense staining in both the nucleus and cytoplasm. Images were taken at 5x and 20x magnification.

87

Figure 4.4. IRAK1 loss increases apoptosis in HNSCC. (a) IRAK1 loss attenuates activation of downstream signaling pathways in HNSCC. UMSCC1 and UMSCC47 were transduced with control (NTsh) or IRAK1 knockdown (IRAK1sh) vectors and protein was collected following selection in puromycin. Whole cell lysates were analyzed by western blot analysis to confirm

IRAK1 knockdown, along with reduction in IRAK1 activation (pIRAK1Thr209) and MAPK pathways. GAPDH was used as a loading control. (b) IRAK1-inhibitor reduces IRAK1 activation in HPV- and HPV+ cell lines. UMSCC1 and UMSCC47 cells were plated and DMSO (control) or

88 IRAK1-inhibitor was added the following day at 1 µM or 10 µM final concentration. Cells were then collected for western blot analysis at indicated time-points. Inhibition of IRAK1 was confirmed by western blot analysis, as measured by phosphorylation of IRAK1. GAPDH was used as a loading control. (c, d) TRAF6 ubiquitination is reduced following IRAK1 loss.

Immunoprecipitation was performed on RIPA lysates with the TRAF6 antibody. Western blot was then performed for ubiquitin. Remaining whole cell lysates were analyzed by western blot for TRAF6, IRAK1, and GAPDH. (e, f) IRAK1 loss increases cellular death via apoptosis. Cells were analyzed by flow cytometry for cleaved-caspase 3 conjugated to FITC. Experiments were performed in triplicate with SEM depicted. (g) IRAK1-inhibitor increases apoptosis. UMSCC1 cells were plated and DMSO or IRAK1-inhibitor was added after cells attached. Cells and media were collected 72 hours later and analyzed for cleaved caspase-3 by flow cytometry.

Experiment was performed three times, with SEM depicted. (*=p<.05).

89

Figure 4.5. IRAK1 and DEK depletion cooperate to increase apoptosis. (a) IRAK1 overexpression rescues ERK1/2 signaling. Sorted control or IRAK1 overexpressing cells were transduced with control (NTsh) or DEK knockdown (DEKsh) vector. After selection was complete protein was collected and analyzed by western blot. GAPDH was used as a loading control. Growth curves of iG2 versus IRAK1 overexpressing cells can be found in

Supplementary S4.8. (b) IRAK1 and DEK regulate cell growth and viability independently. Cells from (A) were used to analyze apoptosis (b), cellular cycle (c), and cellular proliferation. (d)

90 IRAK1 overexpression did not rescue the phenotypes observed with DEK loss. Experiments were performed twice and SD is represented. (e) Combined IRAK1 and DEK loss increases cell death. Control and IRAK1 knockdown cells were transduced with control (AdGFP) or DEK knockdown (AdDEKsh) adenovirus. Three days post-adenoviral infection cells were collected and later analyzed for cleaved caspase-3 by flow cytometry. Graph represents fold change compared to NTsh AdGFP samples. Experiments were performed twice and SD represented. (f)

Molecular model. IRAK1 is a novel target in HNSCC. DEK is required for efficient expression of

IRAK1 and DEK and IRAK1 also independently contribute to HNSCC cell survival and depletion of either promotes apoptosis and this may be a result of ERK1/2, p38 and/or NF-κB signaling.

91 TF binding site ID log p- Up-regulated Genes value V$P53_DECAMER_Q2 2.972 ABCA12, AKAP6, ARHGEF37, ARHGEF6, BCL11B, DOCK9, FLRT1, GJB4, HCAR2, HOXB3, KRT15, LRP1, PIM1, PITPNC1, PPL, PRDM12, PTAFR, PTPRG, RAP2B, RARB, ST5, TBC1D8, TECTA V$E12_Q6 2.610 ABLIM3, ALDH1A2, ARVCF, CAMK2A, COL4A3, CORO2B, CPZ, CRELD1, GFAP, GJB2, HIF3A, ITGB4, MDGA1, MYH14, NFATC4, NPEPPS, POU2F3, SASH1, SELL, SORCS2, SPTB RTAAACA_V$FREAC2_01 2.452 ABCA1, ABLIM3, ACVR1C, ADAMTS13, ARHGEF6, ARID4A, ASAH1, BACH2, BCL11B, BIK, C1QTNF3, CCND1, CDKN1B, DAPP1, DOCK3, EMP1, EPHB3, FBXW7, GATS, GCNT2, GPR110, GPR4, GRK5, HCAR2, HCAR3, HOXA5, HSPG2, IL16, L3MBTL1, LEMD1, LRP1, MAML3, MGAT4B, MMP13, MN1, NEBL, NGEF, NLGN3, NR4A2, PBXIP1, PCDHGA5, PHEX, PIK3IP1, PITPNC1, PLAG1, PLAU, PPAP2B, PRDM1, PROX1, RP1, RPS6KA5, SDCBP2, SLC34A3, SLC4A1, SOCS1, SPRY4, TCF7L2 V$TAL1BETAE47_01 2.110 ABHD16A, ASB4, DTX2, HOXB6, HRK, KCNIP2, NPR2, OMG, PBXIP1, PCDH1, PCDH12, PHEX, PLXNA3, PPAP2B, PYY, RCAN2, RNF19A, SCN1B, UPK2, WNT9A V$IRF_Q6 1.932 ACSL5, ARHGEF6, CCND1, DAPP1, ESR1, HLA-C, HLA-F, HOXB3, MLLT3, MUSK, NCF1, NPEPPS, PLXNC1, PSMB10, SELL, SOCS1, TCF7L2, ZBP1 V$AREB6_01 1.876 CDKN1B, CEL, CRELD1, CSAD, CYP26B1, FBXO24, FXYD3, GAD1, HOXB6, IQGAP2, MAP3K5, PCDHGA5, PLAG1, PROX1, RRAD, S100A9, SLC4A11, STRC, STX1A, VWF KRCTCNNNNMANAGC_UNKNOWN 1.816 HIST1H1C, HIST1H2AC, HIST1H2AJ, HIST1H2BD, HIST1H2BJ, HIST2H2AC, HIST2H2BE, HIST3H2BB V$P53_02 1.809 ABCA12, ACSL5, AKAP6, ASIC4, BCL11B, FLRT1, GAD1, GCNT2, HCAR2, KRT15, NR4A3, PIK3R2, PITPNC1, PPL, PRDM12, PTPRG, RARB, ST5, TMC7 TGANNYRGCA_V$TCF11MAFG_01 1.770 ABCC6, ANGPTL4, ATL1, BCL6, CBX6, CLCN5, CPZ, DRP2, DTX2, ESR1, FBLN2, LRP1, NR4A3, OMG, PFKFB1, PIM1, PRDM1, RAG1, SLC22A18, TFEC, TREML2 TGTTTGY_V$HNF3_Q6 1.760 ALDH1A2, ARHGEF6, ASB4, BCL11B, BCL6, CBX6, CDH12, CNTNAP2, CSAD, DOCK9, DTX2, EFNA1, ESR1, FCGBP, FLRT1, GJB4, HCAR1, HCAR2, HCAR3, HIST1H1C, HIST1H2BD, HOXB3, HS3ST5, IL16, IQGAP2, ITM2B, MAST4, MGAT4B, MUSK, NFATC4, NOS3, NR4A3, OMG, PDGFRA, PHEX, PRDM1, PTPRG, SHC3, SLC4A11, SNCG, SPAG8, ST5, TCF7L2, TFEC V$AML1_01 1.701 ACSL5, AKAP3, ANXA8L2, CD6, EMP1, GRK5, MMP13, MMP14, MPL, MR1, PDZK1, PLXNC1, RAG1, S100A9, SCEL, SLC37A2, STRC, TLL2 V$AML1_Q6 1.701 ACSL5, AKAP3, ANXA8L2, CD6, EMP1, GRK5, MMP13, MMP14, MPL, MR1, PDZK1, PLXNC1, RAG1, S100A9, SCEL, SLC37A2, STRC, TLL2 YATGNWAAT_V$OCT_C 1.680 ACBD4, ALDH1A2, ARHGAP4, CCND1, DNAH5, EPHB3, FBXO24, GPR4, HIST1H2AC, HIST2H2AC, HIST2H2BE, HOXA5, HOXB6, MMP17, NR4A3, PFKFB1, PIWIL4, POU2F3, PRDM1, RARB, RNASE4, SASH1, SEMA6C, SH3BGRL, WNT9A V$PPARA_02 1.644 AHCYL2, CBX6, CDH17, EPHB1, HOXA5, MNT, NLGN3, PSCA, RTN2, SERPINF2, UCN2

V$PPARA_01 1.640 CD36, PAPLN, PDZK1,PTPRG, ZNF547 V$FREAC4_01 1.603 ADAMTS13, ALDH1A2, EMP1, HSPG2, MAST4, PHEX, PRDM1, PTAFR, SLC34A3, SPRY4, TCF7L2, TECTA RYCACNNRNNRNCAG_UNKNOWN 1.543 AHCYL2, FYB, IL17RE, MB, PBXIP1, RHOF, UPK2 V$NFKAPPAB65_01 1.480 COL11A2, CSF1R, CYP2D6, DSC2, GNG4, GRK5, KCNN2, KRT23, PCDH12, POU2F3, PRDM12, PTGES, SDC4, TCEA2, TRIB2, TSLP, ZMYND15 V$GATA_Q6 1.475 ABCA12, COL4A3, DENND1B, EPB42, FMO1, GPR116, HOXB6, HS3ST5, PDGFRA, PFKFB1, PLAG1, PVRL4, SLC4A1, SPRY4, SYT7 TGCCAAR_V$NF1_Q6 1.470 ABCA9, ABLIM2, AGER, AQP11, ARHGEF6, BACH2, BCL6, C1QTNF3, C4A, C4B, C6orf223, CLCN5, COL4A3, DDO, EIF4EBP3, ESR1, FLRT1, GCNT2, GFAP, GNAT2, IL16, INHBC, MAST4, MPL, MUSK, NAPB, NLGN3, PAPLN, PAQR6, PBXIP1, PHEX, PLXNC1, PROX1, RARB, RPS6KA5, RRAD, SPRY4, SYTL2, TSLP, VNN2, XDH V$PAX4_03 1.442 AGER, ANXA4, CBX6, COL11A2, FBXO24, HIF3A, HOXB6, KCNN2, LRP1, MMP14, MNT, MPL, NFATC4, NLGN3, PPAP2B, PRDM1, SERPINI1, TRIM46 RTTTNNNYTGGM_UNKNOWN 1.390 ALDH1A2, ESR1, FHIT, LRP1, MARCO, MAST4, NOS3, PHEX, PTPRCAP, PTPRG, STX6, TECTA TGACATY_UNKNOWN 1.376 ACVR1C, AKAP3, AOC2, ARHGAP4, ARHGEF6, ASB11, ASB4, CLCN5, DSC2, FAM71F1, FBXO32, HOXB3, ITM2B, KCNN2, LEMD1, LRP1, MLLT3, MNT, NEBL, NFATC4, PBXIP1, PDGFRA, PITPNC1, PLAG1, PRDM1, PTPRC, RAPSN, RARB, RCAN2, RNF19A, RRAD, SASH1, SH3BGR, SLC4A1, ST5, STX6, TBC1D8, TECTA V$CEBP_01 1.358 BCL11B, CLDN16, CSAD, CYP2D6, DOCK9, EVI2A, GCNT2, GSTA4, HOXB3, HOXB6, MNT, NFATC4, PFKFB1, PLXNC1, PRDM1, RNF19A, S100A9, ST5 V$OCT_C 1.358 ARHGAP4, EPHB3, FBXO24, GCNT2, GPR4, HIST1H2AC, HIST2H2AC, HIST3H2BB, HOXB3, MMP17, PFKFB1, POU2F3, PRDM1, SASH1, SEMA6C, SH3BGRL, STAT4, TLL2 GATAAGR_V$GATA_C 1.336 ABCA12, ADPRHL1, EPB42, FCGBP, FMO1, HS3ST5, HYAL3, KRT15, KRT23, LEMD1, MLLT3, MST1, NEBL, PFKFB1, PLAG1, PTPRC, SLC34A3, SLC4A1, SPRY4, SYT7 V$COREBINDINGFACTOR_Q6 1.326 ADD1, ANXA8L2, B3GNT5, CD6, CYGB, EMP1, EVI2A, HOXB6, IL17RE, MMP13, MPL, NR4A3, PDZK1, PITPNC1, RAG1, S100A9, SLC37A2, STRC CCANNAGRKGGC_UNKNOWN 1.323 ACBD4, GAD1, GRK5, LRP1, MB, PCDH12, POU2F3, SORCS2, UPK2 V$HNF4_Q6 1.307 ABCA12, ARHGEF37, C4A, C4B, EMP1, FBXW7, HOXA5, PDZK1, PRDM1, PROX1, PTPRH, RARB, SLC5A2, SLPI, SP140, SPINT1, SULT2B1

Supplementary Table S4.1. Transcription factor binding site analysis for genes up-regulated in UMSCC1 and UMSCC47

93 log p- TF binding site ID Down-regulated Genes value V$SRF_C 3.170 ACVR1, CYR61, DIXDC1, EMILIN2, HOXA3, MAPK14, MYL9, NR2F2, PPP2R3A, RAB30, SCOC, TAGLN, THBS1, TNNC1, ZAK V$SRF_Q4 2.907 DIXDC1, EDN1, EMILIN2, HOXA3, MAPK14, MYL9, NR2F2, PPP2R3A, RAB30, SCOC, TAGLN, THBS1, TNNC1, VGF, ZAK CTTTAAR_UNKNOWN 2.802 ANKRD28, ATP5SL, AXL, CREB3L1, CYR61, DCX, DOCK11, EDN1, EGLN3, ETHE1, FPGT, GAN, GNG11, GNGT1, HOXA3, HOXA4, HPS3, KCTD8, KIRREL3, KTN1, MAPKAPK3, NELL2, NR2F2, NR5A2, PDE3B, PLXDC2, PPM1E, PPP2R3A, PXK, RBM24, SCD, SCML1, SMARCA1, SOCS2, STEAP2, TAGLN, TBX2, TCF4, TGFB2, VGF, ZNF593 TGGAAA_V$NFAT_Q4_01 2.689 ANKRD28, ANTXR1, APOM, ARHGEF25, CALB1, CCNI, CD86, CDC42EP3, CKS1B, CNIH1, COL2A1, CXCL10, CYR61, DNMT3B, EFNB3, ERBB4, FAM155B, FZD7, GDA, GFOD1, GPR150, HOXA13, HOXA3, HOXA4, HSPH1, HTR7, IL6, IL7R, KCNQ5, KIRREL3, LRRC2, MGP, MTX2, NGF, NINJ2, NR5A2, NRAS, PAK1IP1, PDE3B, PIGW, POLG2, PPP2CA, PPP2R3A, RAB30, RCN3, RNF128, SCRN3, SKP2, SLA, SLC43A1, SMYD2, SNX12, SOCS2, STEAP2, SV2A, TBX2, TCF12, TCF4, TGFB2, TJP2, TNFRSF11B, TUB, VEGFC, VGF, XPNPEP1, ZIC5, ZNF593 CCAWWNAAGG_V$SRF_Q4 2.622 DIXDC1, RAB30, SCOC, TAGLN, TCF4, THBS1, TNNC1, ZAK V$COUP_01 2.506 APOM, CKS1B, CNIH1, DOCK11, EFNB3, FAM155B, HOXA3, IL21R, NR2F2, NYAP1, PDE3B, PPP2R3A, RPL34, SMYD5, STEAP2 V$ER_Q6_02 2.506 ADAMTS15, CAPN12, CD37, DACT2, DCX, FHL1, HOXA3, HPCAL4, KIRREL3, MTX2, NR2F6, PPM1E, SMARCA1, STEAP2, TOMM40 V$EN1_01 2.073 ERBB4, FHL1, HOXA4, HTR7, MAPK14, NRAS, PEX2, TCF4 V$MYC_Q2 2.068 CA14, COL2A1, HPCAL4, LDHA, NRAS, PFDN2, SC5D, SHMT2, SLC43A1, TGFB2, ZNF593 V$NMYC_01 1.978 ANGPT2, BCL2, CA14, COL2A1, GJA1, MAPKAPK3, NRAS, RAB30, RNF128, SC5D, SLC43A1, SOCS5, VGF, ZCCHC7 V$OCT1_01 1.978 CDC42EP3, CYR61, EGLN3, HOXA3, HPCAL4, IRAK1, LHX6, NR2F2, PPP2R3A, PYGO1, SLC19A3, SLC7A11, TCF12, TCF4 RYTGCNNRGNAAC_V$MIF1_01 1.826 BTG4, EFNB3, HMGCS1, HSF1, LHX6, ZNF593 V$SMAD_Q6 1.813 BCL2, CCNI, CKS1B, GARNL3, KCNQ5, LHX6, NYAP1, PRX, SMARCA1, SMYD2, SNX12, SYT11, TAGLN V$NFAT_Q4_01 1.690 ANKRD28, HOXA3, KCNQ5, KIRREL3, MRPL30, NR5A2, PAK1IP1, PDE3B, PPP2R3A, RNF128, SCRN3, SLA, TJP2 V$NRF2_01 1.680 BCL2, C11orf84, CKS1B, E2F4, FBXO22, HSPH1, NRAS, PAFAH1B2, RPL34, TIMM8B, TOMM40, ZNF22 CACGTG_V$MYC_Q2 1.618 ANGPT2, BEX2, CA14, COL2A1, DZIP1, GAR1, GJA1, HOXA3, HOXA4, HPCAL4, HPS3, HSPH1, KCNQ5, KIAA1033, LDHA, MAPKAPK3, NPTX1, NRAS, PDP2, PDPR, PFDN2,

94 PIGW, RAB30, RNF128, RRS1, SC5D, SHMT2, SLC43A1, SOCS2, SOCS5, TBC1D5, TCF4, TGFB2, VGF, ZCCHC7, ZNF593 V$MYCMAX_02 1.603 ANGPT2, APLN, COL2A1, DZIP1, HOXA3, HSPH1, KCNQ5, LRRC2, RAB30, RNF128, TCF4, TGFB2, VGF AAGWWRNYGGC_UNKNOWN 1.583 ANKRD28, HOXC10, IRAK1, NRAS, PPM1E, RFC1, SMYD5 TAAWWATAG_V$RSRFC4_Q2 1.577 BEX2, GRB14, HOXA4, KCNQ5, KTN1, PPP2R3A, SV2A, TCF4, TNNC1 V$CEBP_Q3 1.516 CYR61, DYRK3, EFNA5, GPLD1, NR2F2, PPM1E, RAB30, SNX12, TCF12, TGFB2, WNT10B, ZAK V$RSRFC4_Q2 1.501 ADAM11, BNIP3, CDC42EP3, KCNQ5, KTN1, MRPS23, PTPN1, SCML1, SLCO2A1, SV2A, TNNC1 V$POU3F2_02 1.460 BBX, CYR61, DIXDC1, HOXA3, HOXA4, HPCAL4, IRAK1, LHX6, NRAS, OLFML2B, PAK1IP1, PDE3B V$STAT5A_03 1.460 BBX, GPR150, HOXA3, HOXA4, HOXC10, MAPK14, PDE3B, PLXDC2, PPP2R3A, SV2A, TCF4, TNFRSF11B V$MEF2_Q6_01 1.460 ADAM11, ANGPT2, CDC42EP3, GAN, GRB14, KCNQ5, KTN1, LRRC2, PRX, SMARCA1, SV2A, TNNC1 V$ELK1_02 1.457 CKS1B, DDIAS, FBXO22, MTX2, NRAS, OGG1, RFC4, TIMM8B, TOMM40, TRO, ZNF22 WWTAAGGC_UNKNOWN 1.456 BBX, DCX, ETHE1, IRAK1, MYL9, NR2F2, NRAS, TBX2 V$CEBP_Q2 1.442 EDN2, LHX6, MAPK14, NR2F2, PDE3B, PTX3, PYGO1, RRM2B, SKP2, TCF12, TCF4 GAANYNYGACNY_UNKNOWN 1.430 CKS1B, GDA, NR2F2, SUMO1, TCF12 V$FOXD3_01 1.428 ANKRD28, CYR61, DIXDC1, HOXA3, KCNQ5, NRAS, PRICKLE2, RNF128, SLC39A8, TGFB2 V$SRF_Q6 1.420 DIXDC1, EMILIN2, HOXA3, LDHA, MYL9, NR2F2, PPP2R3A, SCOC, TAGLN, THBS1, TNNC1, ZAK V$OCT1_05 1.394 BCL2, EDN1, HOXA3, IL6, LHX6, NR2E1, PPM1E, PPP2R3A, SCML1, SLC19A3, SLC7A11, TCF12 V$USF_Q6_01 1.372 APLN, HOXA3, KIAA1033, NPTX1, NRAS, PDP2, PFDN2, SC5D, SLC43A1, SOCS5, VGF V$ATF1_Q6 1.372 CCNI, CD37, DOK1, DYRK3, HOXA4, HOXC10, KCTD8, LDHA, NR2E1, RRM2B, VGF V$CP2_02 1.358 CDC42EP3, DCX, DDIT4, FOXL2, RAB30, RNF121, TBC1D5, TBX2, TNNC1, WNT10B, XPNPEP1 V$USF_02 1.342 APLN, CA14, CYR61, FAM155B, NPTX1, NYAP1, PFDN2, RFC1, RNF128, TCF12, TGFB2, UBE2L6 V$SRF_Q5_01 1.331 DIXDC1, EMILIN2, MAPK14, MYL9, NR2F2, NRAS, PPP2R3A, SCOC, TAGLN, THBS1, ZAK V$IK1_01 1.317 CD86, IL6, LHX6, MMP16, NEK4, NR2F6, RNF128, SMYD2, SOCS2, TNFSF18, VGF, ZIC5

Supplementary Table S4.2. Transcription factor binding site analysis for genes down-regulated in UMSCC1 and UMSCC47

95

TF binding log p- Oppositely-Regulated Genes site ID value

V$GATA6_01 1.567 BCL6, BTG2, CLEC18C, COL4A3, DENND1B, FMO1, HOXB3, HOXB6, HS3ST5, KRT15, KRT23, PDGFRA, PFKFB1, PTPRG, PYY, RNF112, SPRY4, SYT7

Supplementary Table S4.3. Transcription factor binding site analysis for genes differentially expressed in both UMSCC1 and UMSCC47, but oppositely regulated (up vs. down).

96

Supplementary Figure S4.1. Network analysis was performed with Netwalker software on

UMSCC1 and UMSCC47 RNA-Seq data. Labeled Networks 1-5 can be seen in expanded views in Supplementary figures 2-6.

97

Supplementary Figure S4.2. Network 1 is represented, with genes down-regulated in

UMSCC1, with log fold change represented. This is the expanded view of the network map highlighted in Figure 4.1.

98

Supplementary Figure S4.3. Network 2 is represented, with genes up-regulated in UMSCC47, with log fold change represented.

99

Supplementary Figure S4.4. Network 3 is represented, with genes down-regulated in

UMSCC1, with log fold change represented.

100

Supplementary Figure S4.5. Network 4 is represented, with genes up-regulated in UMSCC1, with log fold change represented.

101

Supplementary Figure S4.6. Network 5 is represented, with genes up-regulated in UMSCC47, with log fold change represented.

102

Supplementary Figure S4.7. UMSCC1 and UMSCC47 cells with IRAK1 knockdown reveals no alterations in cell cycle profiles. BrdU incorporation was measured by flow cytometry to determine G1, S, and G2/M populations. Experiments were performed in triplicate.

103

Supplementary Figure S4.8. UMSCC1 IRAK1 overexpression cells (IRAK1) do not exhibit cell growth difference over controls (iG2). Cells were plated at equal densities and counted over three days. Experiments were performed in duplicate.

104 Chapter 5: Discussion and Future Directions

105 Functions of DEK and IRAK1 in HNSCC

Identifying novel biomarkers or targets for cancer therapies is of the utmost importance for improving patient outcomes in poorly understood tumor types and advanced stage malignancies. Among these cancer types is head and neck squamous cell carcinoma, where patients often present at advanced stage disease and have an average predicted 5-year survival rate of 40-50%.104 Additionally, genomic profiling of HNSCC has not fulfilled the promise of many new targetable molecules. Our work aimed to identify new oncogenes and signaling pathways of importance in HNSCC that have the potential to translate into new therapeutic interventions. Based on previous studies and the literature, we hypothesized a protein of interest, DEK, would be functionally implicated in regulating HNSCC phenotypes. Our approach utilized targeting DEK with shRNA to determine the effects of DEK loss in HNSCC. We focused experiments on both HPV+ and HPV- HNSCC subsets to determine if the effects of DEK loss were similar among these subtypes. Although HPV+ patients have improved survival, late stage disease among all HNSCC patients remains a problem. Parallel methods were employed to 1) define the functional role for DEK in HNSCC in vitro and in vivo and 2) determine the global transcriptional effects of DEK loss.

Our work has uncovered two oncogenes, DEK and IRAK1, which are important for the growth and survival of HNSCCs. The first portion of our work demonstrated the in vitro and in vivo consequences of DEK loss in head and neck model systems (Chapter 3). Transgenic

K14E7 mice were crossed with Dek-/- mice and used in an established mouse model of head and neck cancer. We found K14E7 Dek-/- mice had prolonged survival and significantly fewer macroscopic tumors compared to K14E7 Dek+/+ mice. However, all mice still formed microscopic tongue tumors, suggesting Dek loss inhibits tumor progression but not tumor initiation in this chemically and genetically initiated model. It would be interesting to determine in similar 4-NQO mouse experiments, whether tumor development in the Dek-/- mice is delayed but

106 can still occur after the 24-week experimental end-point or whether such tumor development is permanently inhibited. To further define the contributions of DEK to HNSCC tumorigenesis, our laboratory has generated a transgenic mouse model wherein inducible DEK-overexpression was targeted to the epidermis. These mice will be utilized for 4-NQO driven head and neck cancer development in order to determine whether DEK over-expression is sufficient to stimulate tumor growth. In this mouse model, a tetracycline response element regulates two bi- directional CMV promoters that control the expression of DEK and luciferase, respectively. The mice were crossed with tetracycline transactivator (tTa) keratin 5 (K5) mice to drive DEK over- expression to the basal layer of mouse epithelium. Importantly, the over-expression of DEK can be repressed with the addition of doxycycline to either the food or drinking water. This allows for determination of the potential therapeutic impact whereby DEK loss is predicted to cause tumor regression.

We built on previous tumor microarray studies that suggested DEK is overexpressed in

HNSCC.221 To determine whether DEK gene expression translated into protein accumulation, a panel of twenty primary HNSCC specimens, largely oropharyngeal, was evaluated for DEK protein expression. The data was then correlated with HPV status and patient demographics.

The results of these studies revealed that DEK is highly overexpressed even at the earliest stages of HNSCC development and that expression levels were independent of HPV status or patient characteristics. We also validated the use of p16 as a surrogate marker for HPV positivity. One limitation of these studies was the exclusive focus on oropharyngeal specimens; therefore, future studies should determine how DEK expression compares in other sites defined as head and neck cancer, including oral cavity, laryngeal and nasopharyngeal.

The final portion of this project utilized in vitro methods to profile DEK knockdown in established and primary head and neck cancer cell lines. DEK loss attenuated cellular proliferation and growth of these cells and ΔNp63 was implicated as a relevant downstream

DEK effector. DEK does not appear to regulate ΔNp63 transcription, suggesting post-

107 transcriptional modes of regulation. DEK may regulate ΔNp63 protein stability, and future work should determine the exact mechanism. We speculate Fbxw7, an E3 ubiquitin ligase that is mutated in some HNSCCs, may be a potential candidate as it induces degradation of both DEK and p63.214, 281, 282 Furthermore, levels of another p63 variant, TAp63, are increased during apoptosis suggesting DEK may also regulate this functionally antagonistic isoform. Mechanisms of regulation for both isoforms, and the relevance of TAp63 regulation should be evaluated for their contributions to other DEK-dependent phenotypes.

The second approach profiled the impact of DEK loss on transcription in HNSCC cells

(Chapter 4). We utilized the same cell lines and knockdown strategies as described above to identify the DEK-dependent transcriptome by RNA-Sequencing. Differentially expressed genes common to the HPV+ and HPV- cell lines were analyzed to identify novel mediators of DEK- dependent phenotypes. These screens revealed gene signatures associated with inflammatory pathways and functional studies focused on an upstream regulator of immune signaling and transcription, IRAK1. This gene was consistently repressed upon DEK knockdown and its function in HNSCC is unknown. IRAK1 drives inflammatory signaling downstream of toll-like and interleukin-1 receptors and can activate multiple signaling cascades. IRAK1 mRNA and protein repression were validated upon DEK loss, with both decreased as predicted. Data mined from

The Cancer Genome Atlas found IRAK1 is altered, and predominantly up-regulated, in 14% of

HNSCCs. Utilizing the primary human tumor specimens from our earlier work, we confirmed

IRAK1 is expressed in human HNSCCs. Beyond melanoma, this is the first report in solid tumors to our knowledge wherein IRAK1 is implicated as an oncogene. To determine the functional role of IRAK1 in HNSCC, knockdown was performed and validated by the observed repression of TRAF6-ubiquitination and attenuated downstream signaling. Importantly, IRAK1 repression by knockdown or chemical inhibition stimulated apoptosis. Lastly, we assessed if

IRAK1 overexpression could rescue the effects of DEK loss. Although DEK loss reduced cell number, cell cycle progression, and increased apoptosis, IRAK1 overexpression could not

108 reverse these phenotypes. This occurred despite a rescue of ERK1/2 activation following IRAK1 overexpression and DEK depletion, thus indicating that ERK1/2 reconstitution is not sufficient for rescue from DEK loss. However, the combined inhibition of IRAK1 and DEK resulted in significant cell killing over the knockdown of either alone. Taken together, our work has identified DEK and IRAK1 as independent contributors to HNSCC growth and survival.

Functional and therapeutic avenues for future studies

This dissertation work has established DEK as an important regulator of HNSCC growth and proliferation and has uncovered a new role for IRAK1 in HNSCC survival. Additional follow- up of the DEK knockdown transcriptome results should reveal other functionally important regulators of tumor phenotypes. Many other interesting targets were identified that may also contribute to DEK-dependent phenotypes. These include IL6, a cytokine well published for its roles in inflammation in head and neck cancer.283 Transcriptome profiling of additional cell lines, and particularly primary cell populations, is beneficial to determine global gene expression consequences in response to DEK loss. In targeting of DEK genetically, we have taken the first necessary steps to assess the potential therapeutic value of chemical or genetic DEK inhibition.

Whether DEK over-expressing compared to control cells will exhibit inverse regulation remains to be determined.

Although we have established that DEK is required for the expression of IRAK1, it remains unknown if such regulation is direct or indirect. Based on the well accepted role of DEK in transcription (Chapter 2), DEK may be a co-factor at IRAK1 promoters or DEK loss may alter chromatin topology in a way that hinders IRAK1 transcription. There is little published work on the regulation of IRAK1, implicating DEK as a potential transcriptional control mechanism. One well-published regulator of IRAK1 expression is miR-146a. Most often, miR-146a is tumor suppressive and its loss results in hyperactivation of IRAK1. A published report by Hung et al. suggests miR-146a functions oppositely in oral carcinoma and is oncogenic, thus suppressing

109 IRAK1. Since our data demonstrates IRAK1 is oncogenic in HNSCC, future studies should pursue the dual role of miR-146a mediated IRAK1 expression and function in our model system.

Western blot analysis revealed a reduction in NF-κB and MAPK activation upon the loss of IRAK1; however, activation should be measured in more detail and with the use of luciferase reporters. DEK loss also resulted in reduced MAPK signaling, but we have not obtained clear results for how DEK impacts NF-κB. DEK is published to inhibit NF-κB signaling through interactions with p65 in MEFs, but mechanisms whereby DEK affects NF-κB signaling in transformed cells remain uncertain. While IRAK1 rescued ERK1/2 signaling in the absence of

DEK, we have yet to define phenotypic consequences of this rescue, if any. We have published

DEK overexpression stimulates invasion of breast and HNSCC cells, and this is a potentially relevant and unexplored phenotype that may result as a consequence of DEK and IRAK1 activation.215

Since DEK or IRAK1 loss alone stimulates apoptosis and their dual inhibition increases this further, it is possible they may be useful therapeutic targets. Currently, no DEK inhibitors exist, but IRAK1 inhibitors are available and have proven useful in recapitulating the phenotypes observed with shRNA depletion. IRAK1 has already been published as a treatment strategy for

MDS/AML and studies are ongoing to further improve this inhibitor for clinical use. In order to pursue IRAK1 as a candidate therapy in HNSCC, mouse studies will be imperative. Monitoring

HNSCC xenograft growth in the presence or absence of the IRAK1 inhibitor will be the first step in determining if turning off this pathway is therapeutically relevant through prevention and/or remission of established tumors.

During the course of patient treatment many drugs are often given simultaneously.

Therefore, continuing in vitro cell line studies with IRAK1-inhibitors in combination with the currently used treatment modalities for HNSCC will determine if there is improved killing efficacy

110 in the absence of IRAK1 signaling. Of critical importance will be defining how normal cells respond to IRAK1 inhibition, as that may indicate undesired consequences of targeting IRAK1.

The pursuit of DEK as a therapeutic target will be more difficult. This molecule has no known enzymatic activity and a complete crystal structure has yet to be reported. In the interim, domains and residues most crucial to DEK-dependent phenotypes in cancer may inform the best targeting strategies. Currently, DEK is a useful biomarker in other cancers. A clinical trial is underway (ID: NCT01563796) to measure DEK protein levels in the urine of patients suspected of having bladder cancer as a potential diagnostic tool. This ELISA technique may prove fruitful for other cancer types and work is ongoing in other laboratories to detect DEK in the serum of breast and head and neck cancer patients. Beyond these clinical considerations of DEK as a biomarker or drug target, the observed complete penetrance of DEK overexpression in HNSCC suggests the oncogene is central to the development of this malignancy. As such, future studies to identify the full complement of DEK activities in HNSCC may identify other targets and thus additional avenues for the development of next-generation therapies.

Summary

This dissertation has defined the consequences of targeting DEK and/or IRAK1 in

HNSCCs. We uncovered two proteins important for HNSCC survival that have not previously been reported in the head and neck cancer literature. Our work suggests that co-targeting of these proteins may be a fruitful therapeutic avenue. Future studies should better define the structural properties of DEK in the hope that small molecule inhibitors can be identified for clinical purposes. Further assessment of the contribution of DEK and IRAK1 to HNSCC development will be crucial to better understanding their roles in this malignancy.

111 References 1. McBride A. Papillomavirus Episteme [1 May 2014]. Available from: . 2. Van Doorslaer K, Tan Q, Xirasagar S, Bandaru S, Gopalan V, Mohamoud Y, et http://pave.niaid.nih.gov/al. The Papillomavirus Episteme: a central resource for papillomavirus sequence data and analysis. Nucleic Acids Res. 2013 Jan;41:D571-8. 3. de Villiers EM, Fauquet C, Broker TR, Bernard HU, zur Hausen H. Classification of papillomaviruses. Virology. 2004 Jun 20;324:17-27. 4. de Villiers EM, Gissmann L, zur Hausen H. Molecular cloning of viral DNA from human genital warts. J Virol. 1981 Dec;40:932-5. 5. Gissmann L, zur Hausen H. Partial characterization of viral DNA from human genital warts (Condylomata acuminata). Int J Cancer. 1980 May 15;25:605-9. 6. Chesson HW, Ekwueme DU, Saraiya M, Watson M, Lowy DR, Markowitz LE. Estimates of the annual direct medical costs of the prevention and treatment of disease associated with human papillomavirus in the United States. Vaccine. 2012 Sep 14;30:6016-9. 7. Chaturvedi AK, Engels EA, Pfeiffer RM, Hernandez BY, Xiao W, Kim E, et al. Human papillomavirus and rising oropharyngeal cancer incidence in the United States. J Clin Oncol. 2011 Nov 10;29:4294-301. 8. D'Souza G, Agrawal Y, Halpern J, Bodison S, Gillison ML. Oral sexual behaviors associated with prevalent oral human papillomavirus infection. The Journal of infectious diseases. 2009 May 1;199:1263-9. 9. Gillison ML, Koch WM, Capone RB, Spafford M, Westra WH, Wu L, et al. Evidence for a causal association between human papillomavirus and a subset of head and neck cancers. J Natl Cancer Inst. 2000 May 3;92:709-20. 10. Klussmann JP, Weissenborn SJ, Wieland U, Dries V, Kolligs J, Jungehuelsing M, et al. Prevalence, distribution, and viral load of human papillomavirus 16 DNA in tonsillar carcinomas. Cancer. 2001 Dec 1;92:2875-84. 11. Klussmann JP, Mooren JJ, Lehnen M, Claessen SM, Stenner M, Huebbers CU, et al. Genetic signatures of HPV-related and unrelated oropharyngeal carcinoma and their prognostic implications. Clin Cancer Res. 2009 Mar 1;15:1779-86. 12. Langevin SM, Grandis JR, Taioli E. Female hormonal and reproductive factors and head and neck squamous cell carcinoma risk. Cancer letters. 2011 Nov 28;310:216-21. 13. Chung SH, Wiedmeyer K, Shai A, Korach KS, Lambert PF. Requirement for estrogen receptor alpha in a mouse model for human papillomavirus-associated cervical cancer. Cancer Res. 2008 Dec 1;68:9928-34. 14. Spurgeon ME, Chung SH, Lambert PF. Recurrence of cervical cancer in mice after selective estrogen receptor modulator therapy. Am J Pathol. 2014 Feb;184:530-40. 15. Cortes-Malagon EM, Bonilla-Delgado J, Diaz-Chavez J, Hidalgo-Miranda A, Romero-Cordoba S, Uren A, et al. Gene expression profile regulated by the HPV16 E7 oncoprotein and estradiol in cervical tissue. Virology. 2013 Dec;447:155-65. 16. Son J, Park JW, Lambert PF, Chung SH. Requirement of estrogen receptor alpha DNA-binding domain for HPV oncogene-induced cervical carcinogenesis in mice. Carcinogenesis. 2014 Feb;35:489-96.

112 17. Epidemiology and Prevention of Vaccine-Preventable Diseases: The Pink Book [1 May 2014]. Available from: . 18. Hang D, Gao L, Sun M, Liu Y, Ke Y. Functional effects of sequence variations in the E6 and E2 genes of humanhttp://www.cdc papillomavirus.gov/vaccines/pubs/pinkbook/hpv.html 16 European and Asian variants. Journal of medical virology. 2014 Apr;86:618-26. 19. Cornet I, Gheit T, Iannacone MR, Vignat J, Sylla BS, Del Mistro A, et al. HPV16 genetic variation and the development of cervical cancer worldwide. British journal of cancer. 2013 Jan 15;108:240-4. 20. Yamada T, Manos MM, Peto J, Greer CE, Munoz N, Bosch FX, et al. Human papillomavirus type 16 sequence variation in cervical cancers: a worldwide perspective. J Virol. 1997 Mar;71:2463-72. 21. Freitas LB, Chen Z, Muqui EF, Boldrini NA, Miranda AE, Spano LC, et al. Human papillomavirus 16 non-European variants are preferentially associated with high-grade cervical lesions. PloS one. 2014;9:e100746. 22. Berumen J, Ordonez RM, Lazcano E, Salmeron J, Galvan SC, Estrada RA, et al. Asian-American variants of human papillomavirus 16 and risk for cervical cancer: a case-control study. J Natl Cancer Inst. 2001 Sep 5;93:1325-30. 23. Villa LL, Sichero L, Rahal P, Caballero O, Ferenczy A, Rohan T, et al. Molecular variants of human papillomavirus types 16 and 18 preferentially associated with cervical neoplasia. The Journal of general virology. 2000 Dec;81:2959-68. 24. Xi LF, Koutsky LA, Hildesheim A, Galloway DA, Wheeler CM, Winer RL, et al. Risk for high-grade cervical intraepithelial neoplasia associated with variants of human papillomavirus types 16 and 18. Cancer Epidemiol Biomarkers Prev. 2007 Jan;16:4-10. 25. Niccoli S, Abraham S, Richard C, Zehbe I. The Asian-American E6 variant protein of human papillomavirus 16 alone is sufficient to promote immortalization, transformation, and migration of primary human foreskin keratinocytes. J Virol. 2012 Nov;86:12384-96. 26. Betiol J, Villa LL, Sichero L. Impact of HPV infection on the development of head and neck cancer. Brazilian journal of medical and biological research = Revista brasileira de pesquisas medicas e biologicas / Sociedade Brasileira de Biofisica [et al]. 2013 Mar;46:217-26. 27. Du J, Nordfors C, Nasman A, Sobkowiak M, Romanitan M, Dalianis T, et al. Human papillomavirus (HPV) 16 E6 variants in tonsillar cancer in comparison to those in cervical cancer in Stockholm, Sweden. PloS one. 2012;7:e36239. 28. Zheng ZM, Baker CC. Papillomavirus genome structure, expression, and post- transcriptional regulation. Front Biosci. 2006;11:2286-302. 29. Bernard BA, Bailly C, Lenoir MC, Darmon M, Thierry F, Yaniv M. The human papillomavirus type 18 (HPV18) E2 gene product is a repressor of the HPV18 regulatory region in human keratinocytes. J Virol. 1989 Oct;63:4317-24. 30. Demeret C, Desaintes C, Yaniv M, Thierry F. Different mechanisms contribute to the E2-mediated transcriptional repression of human papillomavirus type 18 viral oncogenes. J Virol. 1997 Dec;71:9343-9. 31. Phelps WC, Howley PM. Transcriptional trans-activation by the human papillomavirus type 16 E2 gene product. J Virol. 1987 May;61:1630-8.

113 32. Chiang CM, Ustav M, Stenlund A, Ho TF, Broker TR, Chow LT. Viral E1 and E2 proteins support replication of homologous and heterologous papillomaviral origins. Proc Natl Acad Sci U S A. 1992 Jul 1;89:5799-803. 33. Doorbar J. The E4 protein; structure, function and patterns of expression. Virology. 2013 Oct;445:80-98. 34. Doorbar J, Ely S, Sterling J, McLean C, Crawford L. Specific interaction between HPV-16 E1-E4 and cytokeratins results in collapse of the epithelial cell intermediate filament network. Nature. 1991 Aug 29;352:824-7. 35. Schiller JT, Day PM, Kines RC. Current understanding of the mechanism of HPV infection. Gynecologic oncology. 2010 Jun;118:S12-7. 36. Giroglou T, Florin L, Schafer F, Streeck RE, Sapp M. Human papillomavirus infection requires cell surface heparan sulfate. J Virol. 2001 Feb;75:1565-70. 37. Johnson KM, Kines RC, Roberts JN, Lowy DR, Schiller JT, Day PM. Role of heparan sulfate in attachment to and infection of the murine female genital tract by human papillomavirus. J Virol. 2009 Mar;83:2067-74. 38. Abban CY, Meneses PI. Usage of heparan sulfate, integrins, and FAK in HPV16 infection. Virology. 2010 Jul 20;403:1-16. 39. Kumar A, Jacob T, Abban CY, Meneses PI. Intermediate Heparan Sulfate Binding During HPV-16 Infection in HaCaTs. American journal of therapeutics. 2014 Mar 11. 40. Aksoy P, Abban CY, Kiyashka E, Qiang W, Meneses PI. HPV16 infection of HaCaTs is dependent on beta4 integrin, and alpha6 integrin processing. Virology. 2014 Jan 20;449:45-52. 41. Surviladze Z, Dziduszko A, Ozbun MA. Essential roles for soluble virion- associated heparan sulfonated proteoglycans and growth factors in human papillomavirus infections. PLoS pathogens. 2012 Feb;8:e1002519. 42. Cruz L, Meyers C. Differential dependence on host cell glycosaminoglycans for infection of epithelial cells by high-risk HPV types. PloS one. 2013;8:e68379. 43. Fehrmann F, Laimins LA. Human papillomaviruses: targeting differentiating epithelial cells for malignant transformation. Oncogene. 2003 Aug 11;22:5201-7. 44. Bodily J, Laimins LA. Persistence of human papillomavirus infection: keys to malignant progression. Trends Microbiol. 2011 Jan;19:33-9. 45. Cheng S, Schmidt-Grimminger DC, Murant T, Broker TR, Chow LT. Differentiation-dependent up-regulation of the human papillomavirus E7 gene reactivates cellular DNA replication in suprabasal differentiated keratinocytes. Genes Dev. 1995 Oct 1;9:2335-49. 46. Moody CA, Laimins LA. Human papillomavirus oncoproteins: pathways to transformation. Nat Rev Cancer. 2010 Aug;10:550-60. 47. Stubenrauch F, Hummel M, Iftner T, Laimins LA. The E8E2C protein, a negative regulator of viral transcription and replication, is required for extrachromosomal maintenance of human papillomavirus type 31 in keratinocytes. J Virol. 2000 Feb;74:1178-86. 48. Straub E, Dreer M, Fertey J, Iftner T, Stubenrauch F. The viral E8^E2C repressor limits productive replication of human papillomavirus 16. J Virol. 2014 Jan;88:937-47. 49. Lace MJ, Anson JR, Thomas GS, Turek LP, Haugen TH. The E8--E2 gene product of human papillomavirus type 16 represses early transcription and replication

114 but is dispensable for viral plasmid persistence in keratinocytes. J Virol. 2008 Nov;82:10841-53. 50. McBride AA, Oliveira JG, McPhillips MG. Partitioning viral genomes in mitosis: same idea, different targets. Cell Cycle. 2006 Jul;5:1499-502. 51. You J, Croyle JL, Nishimura A, Ozato K, Howley PM. Interaction of the bovine papillomavirus E2 protein with Brd4 tethers the viral DNA to host mitotic chromosomes. Cell. 2004 Apr 30;117:349-60. 52. Vinokurova S, Wentzensen N, Kraus I, Klaes R, Driesch C, Melsheimer P, et al. Type-dependent integration frequency of human papillomavirus genomes in cervical lesions. Cancer Res. 2008 Jan 1;68:307-13. 53. Smith PP, Friedman CL, Bryant EM, McDougall JK. Viral integration and fragile sites in human papillomavirus-immortalized human keratinocyte cell lines. Genes, chromosomes & cancer. 1992 Sep;5:150-7. 54. Pett MR, Alazawi WO, Roberts I, Dowen S, Smith DI, Stanley MA, et al. Acquisition of high-level chromosomal instability is associated with integration of human papillomavirus type 16 in cervical keratinocytes. Cancer Res. 2004 Feb 15;64:1359-68. 55. Schwarz E, Freese UK, Gissmann L, Mayer W, Roggenbuck B, Stremlau A, et al. Structure and transcription of human papillomavirus sequences in cervical carcinoma cells. Nature. 1985 Mar 7-13;314:111-4. 56. Tan SH, Leong LE, Walker PA, Bernard HU. The human papillomavirus type 16 E2 transcription factor binds with low cooperativity to two flanking sites and represses the E6 promoter through displacement of Sp1 and TFIID. J Virol. 1994 Oct;68:6411-20. 57. Dong G, Broker TR, Chow LT. Human papillomavirus type 11 E2 proteins repress the homologous E6 promoter by interfering with the binding of host transcription factors to adjacent elements. J Virol. 1994 Feb;68:1115-27. 58. Romanczuk H, Thierry F, Howley PM. Mutational analysis of cis elements involved in E2 modulation of human papillomavirus type 16 P97 and type 18 P105 promoters. J Virol. 1990 Jun;64:2849-59. 59. Thierry F, Howley PM. Functional analysis of E2-mediated repression of the HPV18 P105 promoter. The New biologist. 1991 Jan;3:90-100. 60. Dyson N, Howley PM, Munger K, Harlow E. The human papilloma virus-16 E7 oncoprotein is able to bind to the retinoblastoma gene product. Science. 1989 Feb 17;243:934-7. 61. Dyson N, Guida P, Munger K, Harlow E. Homologous sequences in adenovirus E1A and human papillomavirus E7 proteins mediate interaction with the same set of cellular proteins. J Virol. 1992 Dec;66:6893-902. 62. Howie HL, Katzenellenbogen RA, Galloway DA. Papillomavirus E6 proteins. Virology. 2009 Feb 20;384:324-34. 63. McMurray HR, McCance DJ. Human papillomavirus type 16 E6 activates TERT gene transcription through induction of c-Myc and release of USF-mediated repression. J Virol. 2003 Sep;77:9852-61. 64. Kiyono T, Foster SA, Koop JI, McDougall JK, Galloway DA, Klingelhutz AJ. Both Rb/p16INK4a inactivation and telomerase activity are required to immortalize human epithelial cells. Nature. 1998 Nov 5;396:84-8.

115 65. McKenna DJ, Patel D, McCance DJ. miR-24 and miR-205 expression is dependent on HPV onco-protein expression in keratinocytes. Virology. 2014 Jan 5;448:210-6. 66. Yu Y, Zhang Y, Zhang S. MicroRNA-92 regulates cervical tumorigenesis and its expression is upregulated by human papillomavirus-16 E6 in cervical cancer cells. Oncology letters. 2013 Aug;6:468-74. 67. Martinez I, Gardiner AS, Board KF, Monzon FA, Edwards RP, Khan SA. Human papillomavirus type 16 reduces the expression of microRNA-218 in cervical carcinoma cells. Oncogene. 2008 Apr 17;27:2575-82. 68. Wald AI, Hoskins EE, Wells SI, Ferris RL, Khan SA. Alteration of microRNA profiles in squamous cell carcinoma of the head and neck cell lines by human papillomavirus. Head Neck. 2011 Apr;33:504-12. 69. Doorbar J, Quint W, Banks L, Bravo IG, Stoler M, Broker TR, et al. The biology and life-cycle of human papillomaviruses. Vaccine. 2012 Nov 20;30 Suppl 5:F55-70. 70. Roman A, Munger K. The papillomavirus E7 proteins. Virology. 2013 Oct;445:138-68. 71. McLaughlin-Drubin ME, Munger K. Biochemical and functional interactions of human papillomavirus proteins with polycomb group proteins. Viruses. 2013 May;5:1231-49. 72. DiMaio D, Mattoon D. Mechanisms of cell transformation by papillomavirus E5 proteins. Oncogene. 2001 Nov 26;20:7866-73. 73. DiMaio D, Petti LM. The E5 proteins. Virology. 2013 Oct;445:99-114. 74. Genther Williams SM, Disbrow GL, Schlegel R, Lee D, Threadgill DW, Lambert PF. Requirement of epidermal growth factor receptor for hyperplasia induced by E5, a high-risk human papillomavirus oncogene. Cancer Res. 2005 Aug 1;65:6534-42. 75. Maufort JP, Shai A, Pitot HC, Lambert PF. A role for HPV16 E5 in cervical carcinogenesis. Cancer Res. 2010 Apr 1;70:2924-31. 76. Liao S, Deng D, Zhang W, Hu X, Wang W, Wang H, et al. Human papillomavirus 16/18 E5 promotes cervical cancer cell proliferation, migration and invasion in vitro and accelerates tumor growth in vivo. Oncology reports. 2013 Jan;29:95-102. 77. White EA, Kramer RE, Tan MJ, Hayes SD, Harper JW, Howley PM. Comprehensive analysis of host cellular interactions with human papillomavirus E6 proteins identifies new E6 binding partners and reflects viral diversity. J Virol. 2012 Dec;86:13174-86. 78. White EA, Sowa ME, Tan MJ, Jeudy S, Hayes SD, Santha S, et al. Systematic identification of interactions between host cell proteins and E7 oncoproteins from diverse human papillomaviruses. Proc Natl Acad Sci U S A. 2012 Jan 31;109:E260-7. 79. Nguyen ML, Nguyen MM, Lee D, Griep AE, Lambert PF. The PDZ ligand domain of the human papillomavirus type 16 E6 protein is required for E6's induction of epithelial hyperplasia in vivo. J Virol. 2003 Jun;77:6957-64. 80. Munger K, Baldwin A, Edwards KM, Hayakawa H, Nguyen CL, Owens M, et al. Mechanisms of human papillomavirus-induced oncogenesis. J Virol. 2004 Nov;78:11451-60. 81. Potten CS. Cell replacement in epidermis (keratopoiesis) via discrete units of proliferation. International review of cytology. 1981;69:271-318.

116 82. Alonso L, Fuchs E. Stem cells of the skin epithelium. Proc Natl Acad Sci U S A. 2003 Sep 30;100 Suppl 1:11830-5. 83. Jones PH, Simons BD, Watt FM. Sic transit gloria: farewell to the epidermal transit amplifying cell? Cell stem cell. 2007 Oct 11;1:371-81. 84. Ramdass B, Chowdhari A, Koka P. Cancer-initiating cells as target for prevention of recurring disease etiology: role of these malignant putative progenitor cells in relapse or metastasis of human cervical carcinoma. Journal of stem cells. 2013;8:233-51. 85. Villanueva-Toledo J, Ponciano-Gomez A, Ortiz-Sanchez E, Garrido E. Side populations from cervical-cancer-derived cell lines have stem-cell-like properties. Molecular biology reports. 2014 Apr;41:1993-2004. 86. Clifford GM, Smith JS, Plummer M, Munoz N, Franceschi S. Human papillomavirus types in invasive cervical cancer worldwide: a meta-analysis. British journal of cancer. 2003 Jan 13;88:63-73. 87. Munoz N, Bosch FX, de Sanjose S, Herrero R, Castellsague X, Shah KV, et al. Epidemiologic classification of human papillomavirus types associated with cervical cancer. N Engl J Med. 2003 Feb 6;348:518-27. 88. Freitas AC, Mariz FC, Silva MA, Jesus AL. Human papillomavirus vertical transmission: review of current data. Clinical infectious diseases : an official publication of the Infectious Diseases Society of America. 2013 May;56:1451-6. 89. Clavel C, Masure M, Bory JP, Putaud I, Mangeonjean C, Lorenzato M, et al. Hybrid Capture II-based human papillomavirus detection, a sensitive test to detect in routine high-grade cervical lesions: a preliminary study on 1518 women. British journal of cancer. 1999 Jul;80:1306-11. 90. Woodman CB, Collins SI, Young LS. The natural history of cervical HPV infection: unresolved issues. Nat Rev Cancer. 2007 Jan;7:11-22. 91. Franceschi S, Herrero R, Clifford GM, Snijders PJ, Arslan A, Anh PT, et al. Variations in the age-specific curves of human papillomavirus prevalence in women worldwide. Int J Cancer. 2006 Dec 1;119:2677-84. 92. Institute NC. National Cancer Institute: Head and Neck Cancers [3 December 2013]. Available from: . 93. Stein AP, Saha http://www.cancer.gov/cancertopics/faS, Yu M, Kimple RJ, Lambert PF. Prevalencectsheet/Sites of human-­‐Types/head -­‐ papillomavirusand-­‐neck -­‐ r30 in oropharyngeal squamous cell carcinoma in the United States across time. Chemical research in toxicology. 2014 Apr 21;27:462-9. 94. Ang KK, Harris J, Wheeler R, Weber R, Rosenthal DI, Nguyen-Tan PF, et al. Human papillomavirus and survival of patients with oropharyngeal cancer. N Engl J Med. 2010 Jul 1;363:24-35. 95. Leemans CR, Braakhuis BJ, Brakenhoff RH. The molecular biology of head and neck cancer. Nat Rev Cancer. 2011 Jan;11:9-22. 96. Chaturvedi AK, Anderson WF, Lortet-Tieulent J, Curado MP, Ferlay J, Franceschi S, et al. Worldwide trends in incidence rates for oral cavity and oropharyngeal cancers. J Clin Oncol. 2013 Dec 20;31:4550-9. 97. Combes JD, Franceschi S. Role of human papillomavirus in non-oropharyngeal head and neck cancers. Oral Oncol. 2014 May;50:370-9.

117 98. Fakhry C, Rosenthal BT, Clark DP, Gillison ML. Associations between oral HPV16 infection and cytopathology: evaluation of an oropharyngeal "pap-test equivalent" in high-risk populations. Cancer prevention research. 2011 Sep;4:1378-84. 99. Wang SS, Sherman ME, Hildesheim A, Lacey JV, Jr., Devesa S. Cervical adenocarcinoma and squamous cell carcinoma incidence trends among white women and black women in the United States for 1976-2000. Cancer. 2004 Mar 1;100:1035-44. 100. Society AC. Cervical Cancer [1 June 2014]. Available from:

. 101.http://www.cancer.org Ferlay J, Shin HR,/cancer/cervicalcancer/ Bray F, Forman D, Mathers C, Parkin DM. Estimates of worldwide burden of cancer in 2008: GLOBOCAN 2008. Int J Cancer. 2010 Dec 15;127:2893-917. 102. Pecorelli S. Revised FIGO staging for carcinoma of the vulva, cervix, and endometrium. International journal of gynaecology and obstetrics: the official organ of the International Federation of Gynaecology and Obstetrics. 2009 May;105:103-4. 103. Society AC. Survival rates for cervical cancer by stage [1 June 2014]. Available from: . 104. http://www.cancer.org/cancer/cervicalcancer/detailedguide/cervicalPatel PR, Salama JK. Reirradiation for recurrent head and neck cancer.-­‐cancer Expert-­‐ reviewsurvival of anticancer therapy. 2012 Sep;12:1177-89. 105. Pulte D, Brenner H. Changes in survival in head and neck cancers in the late 20th and early 21st century: a period analysis. The oncologist. 2010;15:994-1001. 106. Serup-Hansen E, Linnemann D, Skovrider-Ruminski W, Hogdall E, Geertsen PF, Havsteen H. Human papillomavirus genotyping and p16 expression as prognostic factors for patients with American Joint Committee on Cancer stages I to III carcinoma of the anal canal. J Clin Oncol. 2014 Jun 10;32:1812-7. 107. Schwartz SM, Daling JR, Shera KA, Madeleine MM, McKnight B, Galloway DA, et al. Human papillomavirus and prognosis of invasive cervical cancer: a population- based study. J Clin Oncol. 2001 Apr 1;19:1906-15. 108. Gillison ML, Zhang Q, Jordan R, Xiao W, Westra WH, Trotti A, et al. Tobacco smoking and increased risk of death and progression for patients with p16-positive and p16-negative oropharyngeal cancer. J Clin Oncol. 2012 Jun 10;30:2102-11. 109. Eifel PJ, Winter K, Morris M, Levenback C, Grigsby PW, Cooper J, et al. Pelvic irradiation with concurrent chemotherapy versus pelvic and para-aortic irradiation for high-risk cervical cancer: an update of radiation therapy oncology group trial (RTOG) 90-01. J Clin Oncol. 2004 Mar 1;22:872-80. 110. Rose PG, Bundy BN, Watkins EB, Thigpen JT, Deppe G, Maiman MA, et al. Concurrent cisplatin-based radiotherapy and chemotherapy for locally advanced cervical cancer. N Engl J Med. 1999 Apr 15;340:1144-53. 111. Keys HM, Bundy BN, Stehman FB, Muderspach LI, Chafe WE, Suggs CL, 3rd, et al. Cisplatin, radiation, and adjuvant hysterectomy compared with radiation and adjuvant hysterectomy for bulky stage IB cervical carcinoma. N Engl J Med. 1999 Apr 15;340:1154-61. 112. Moore DH, Blessing JA, McQuellon RP, Thaler HT, Cella D, Benda J, et al. Phase III study of cisplatin with or without paclitaxel in stage IVB, recurrent, or

118 persistent squamous cell carcinoma of the cervix: a gynecologic oncology group study. J Clin Oncol. 2004 Aug 1;22:3113-9. 113. Long HJ, 3rd, Bundy BN, Grendys EC, Jr., Benda JA, McMeekin DS, Sorosky J, et al. Randomized phase III trial of cisplatin with or without topotecan in carcinoma of the uterine cervix: a Gynecologic Oncology Group Study. J Clin Oncol. 2005 Jul 20;23:4626-33. 114. Cooper JS, Pajak TF, Forastiere AA, Jacobs J, Campbell BH, Saxman SB, et al. Postoperative concurrent radiotherapy and chemotherapy for high-risk squamous-cell carcinoma of the head and neck. N Engl J Med. 2004 May 6;350:1937-44. 115. Bernier J, Domenge C, Ozsahin M, Matuszewska K, Lefebvre JL, Greiner RH, et al. Postoperative irradiation with or without concomitant chemotherapy for locally advanced head and neck cancer. N Engl J Med. 2004 May 6;350:1945-52. 116. Adelstein DJ, Li Y, Adams GL, Wagner H, Jr., Kish JA, Ensley JF, et al. An intergroup phase III comparison of standard radiation therapy and two schedules of concurrent chemoradiotherapy in patients with unresectable squamous cell head and neck cancer. J Clin Oncol. 2003 Jan 1;21:92-8. 117. Ley J, Mehan P, Wildes TM, Thorstad W, Gay HA, Michel L, et al. Cisplatin versus cetuximab given concurrently with definitive radiation therapy for locally advanced head and neck squamous cell carcinoma. Oncology. 2013;85:290-6. 118. Riaz N, Sherman E, Koutcher L, Shapiro L, Katabi N, Zhang Z, et al. Concurrent Chemoradiotherapy With Cisplatin Versus Cetuximab for Squamous Cell Carcinoma of the Head and Neck. American journal of clinical oncology. 2014 Jan 7. 119. Bonner JA, Harari PM, Giralt J, Cohen RB, Jones CU, Sur RK, et al. Radiotherapy plus cetuximab for locoregionally advanced head and neck cancer: 5-year survival data from a phase 3 randomised trial, and relation between cetuximab-induced rash and survival. The lancet oncology. 2010 Jan;11:21-8. 120. Parsons JT, Mendenhall WM, Stringer SP, Cassisi NJ, Million RR. Salvage surgery following radiation failure in squamous cell carcinoma of the supraglottic larynx. Int J Radiat Oncol Biol Phys. 1995 Jun 15;32:605-9. 121. Bachar GY, Goh C, Goldstein DP, O'Sullivan B, Irish JC. Long-term outcome analysis after surgical salvage for recurrent tonsil carcinoma following radical radiotherapy. Eur Arch Otorhinolaryngol. 2010 Feb;267:295-301. 122. Zafereo ME, Hanasono MM, Rosenthal DI, Sturgis EM, Lewin JS, Roberts DB, et al. The role of salvage surgery in patients with recurrent squamous cell carcinoma of the oropharynx. Cancer. 2009 Dec 15;115:5723-33. 123. Vermorken JB, Stohlmacher-Williams J, Davidenko I, Licitra L, Winquist E, Villanueva C, et al. Cisplatin and fluorouracil with or without panitumumab in patients with recurrent or metastatic squamous-cell carcinoma of the head and neck (SPECTRUM): an open-label phase 3 randomised trial. The lancet oncology. 2013 Jul;14:697-710. 124. Jacobs C, Lyman G, Velez-Garcia E, Sridhar KS, Knight W, Hochster H, et al. A phase III randomized study comparing cisplatin and fluorouracil as single agents and in combination for advanced squamous cell carcinoma of the head and neck. J Clin Oncol. 1992 Feb;10:257-63. 125. Burtness B, Goldwasser MA, Flood W, Mattar B, Forastiere AA, Eastern Cooperative Oncology G. Phase III randomized trial of cisplatin plus placebo compared

119 with cisplatin plus cetuximab in metastatic/recurrent head and neck cancer: an Eastern Cooperative Oncology Group study. J Clin Oncol. 2005 Dec 1;23:8646-54. 126. Forastiere AA, Metch B, Schuller DE, Ensley JF, Hutchins LF, Triozzi P, et al. Randomized comparison of cisplatin plus fluorouracil and carboplatin plus fluorouracil versus methotrexate in advanced squamous-cell carcinoma of the head and neck: a Southwest Oncology Group study. J Clin Oncol. 1992 Aug;10:1245-51. 127. Keysar SB, Astling DP, Anderson RT, Vogler BW, Bowles DW, Morton JJ, et al. A patient tumor transplant model of squamous cell cancer identifies PI3K inhibitors as candidate therapeutics in defined molecular bins. Molecular oncology. 2013 Aug;7:776- 90. 128. Kimple RJ, Harari PM, Torres AD, Yang RZ, Soriano BJ, Yu M, et al. Development and characterization of HPV-positive and HPV-negative head and neck squamous cell carcinoma tumorgrafts. Clin Cancer Res. 2013 Feb 15;19:855-64. 129. Fakhry C, Zhang Q, Nguyen-Tan PF, Rosenthal D, El-Naggar A, Garden AS, et al. Human Papillomavirus and Overall Survival After Progression of Oropharyngeal Squamous Cell Carcinoma. J Clin Oncol. 2014 Jun 23. 130. Kimple RJ, Smith MA, Blitzer GC, Torres AD, Martin JA, Yang RZ, et al. Enhanced radiation sensitivity in HPV-positive head and neck cancer. Cancer Res. 2013 Aug 1;73:4791-800. 131. Rieckmann T, Tribius S, Grob TJ, Meyer F, Busch CJ, Petersen C, et al. HNSCC cell lines positive for HPV and p16 possess higher cellular radiosensitivity due to an impaired DSB repair capacity. Radiother Oncol. 2013 May;107:242-6. 132. Chinn SB, Darr OA, Owen JH, Bellile E, McHugh JB, Spector ME, et al. Cancer stem cells: Mediators of tumorigenesis and metastasis in head and neck squamous cell carcinoma. Head Neck. 2014 Jan 10. 133. Einstein MH, Baron M, Levin MJ, Chatterjee A, Edwards RP, Zepp F, et al. Comparison of the immunogenicity and safety of Cervarix and Gardasil human papillomavirus (HPV) cervical cancer vaccines in healthy women aged 18-45 years. Human vaccines. 2009 Oct;5:705-19. 134. Keam SJ, Harper DM. Human papillomavirus types 16 and 18 vaccine (recombinant, AS04 adjuvanted, adsorbed) [Cervarix]. Drugs. 2008;68:359-72. 135. Siddiqui MA, Perry CM. Human papillomavirus quadrivalent (types 6, 11, 16, 18) recombinant vaccine (Gardasil). Drugs. 2006;66:1263-71; discussion 72-3. 136. Stanley M, Lowy DR, Frazer I. Chapter 12: Prophylactic HPV vaccines: underlying mechanisms. Vaccine. 2006 Aug 31;24 Suppl 3:S3/106-13. 137. Wheeler CM, Castellsague X, Garland SM, Szarewski A, Paavonen J, Naud P, et al. Cross-protective efficacy of HPV-16/18 AS04-adjuvanted vaccine against cervical infection and precancer caused by non-vaccine oncogenic HPV types: 4-year end-of- study analysis of the randomised, double-blind PATRICIA trial. The lancet oncology. 2012 Jan;13:100-10. 138. Bosch FX, Broker TR, Forman D, Moscicki AB, Gillison ML, Doorbar J, et al. Comprehensive control of human papillomavirus infections and related diseases. Vaccine. 2013 Dec 31;31 Suppl 7:H1-31. 139. Gillison ML, Castellsague X, Chaturvedi A, Goodman MT, Snijders P, Tommasino M, et al. Eurogin Roadmap: comparative epidemiology of HPV infection and

120 associated cancers of the head and neck and cervix. Int J Cancer. 2014 Feb 1;134:497- 507. 140. Wilting SM, Smeets SJ, Snijders PJ, van Wieringen WN, van de Wiel MA, Meijer GA, et al. Genomic profiling identifies common HPV-associated chromosomal alterations in squamous cell carcinomas of cervix and head and neck. BMC medical genomics. 2009;2:32. 141. Ojesina AI, Lichtenstein L, Freeman SS, Pedamallu CS, Imaz-Rosshandler I, Pugh TJ, et al. Landscape of genomic alterations in cervical carcinomas. Nature. 2014 Feb 20;506:371-5. 142. Stransky N, Egloff AM, Tward AD, Kostic AD, Cibulskis K, Sivachenko A, et al. The mutational landscape of head and neck squamous cell carcinoma. Science. 2011 Aug 26;333:1157-60. 143. Agrawal N, Frederick MJ, Pickering CR, Bettegowda C, Chang K, Li RJ, et al. Exome sequencing of head and neck squamous cell carcinoma reveals inactivating mutations in NOTCH1. Science. 2011 Aug 26;333:1154-7. 144. Thierry F, Yaniv M. The BPV1-E2 trans-acting protein can be either an activator or a repressor of the HPV18 regulatory region. EMBO J. 1987 Nov;6:3391-7. 145. Jeon S, Allen-Hoffmann BL, Lambert PF. Integration of human papillomavirus type 16 into the human genome correlates with a selective growth advantage of cells. J Virol. 1995 May;69:2989-97. 146. Goodwin EC, DiMaio D. Repression of human papillomavirus oncogenes in HeLa cervical carcinoma cells causes the orderly reactivation of dormant tumor suppressor pathways. Proc Natl Acad Sci U S A. 2000 Nov 7;97:12513-8. 147. Schmitz M, Driesch C, Jansen L, Runnebaum IB, Durst M. Non-random integration of the HPV genome in cervical cancer. PloS one. 2012;7:e39632. 148. Adey A, Burton JN, Kitzman JO, Hiatt JB, Lewis AP, Martin BK, et al. The haplotype-resolved genome and epigenome of the aneuploid HeLa cancer cell line. Nature. 2013 Aug 8;500:207-11. 149. Akagi K, Li J, Broutian TR, Padilla-Nash H, Xiao W, Jiang B, et al. Genome-wide analysis of HPV integration in human cancers reveals recurrent, focal genomic instability. Genome research. 2014 Feb;24:185-99. 150. Peter M, Stransky N, Couturier J, Hupe P, Barillot E, de Cremoux P, et al. Frequent genomic structural alterations at HPV insertion sites in cervical carcinoma. J Pathol. 2010 Jul;221:320-30. 151. Peter M, Rosty C, Couturier J, Radvanyi F, Teshima H, Sastre-Garau X. MYC activation associated with the integration of HPV DNA at the MYC locus in genital tumors. Oncogene. 2006 Sep 28;25:5985-93. 152. Kim SH, Koo BS, Kang S, Park K, Kim H, Lee KR, et al. HPV integration begins in the tonsillar crypt and leads to the alteration of p16, EGFR and c-myc during tumor formation. Int J Cancer. 2007 Apr 1;120:1418-25. 153. Institute NC. The Cancer Genome Atlas [10 June 2014]. Available from: . 154. Santin AD, Zhan F, Bignotti E, Siegel ER, Cane S, Bellone S, et al. Gene expressionhttp://cancergenome.nih.gov/ profiles of primary HPV16- and HPV18-infected early stage cervical cancers and normal cervical epithelium: identification of novel candidate molecular markers for cervical cancer diagnosis and therapy. Virology. 2005 Jan 20;331:269-91.

121 155. Mattarocci S, Abbruzzese C, Mileo AM, Carosi M, Pescarmona E, Vico C, et al. Identification of pivotal cellular factors involved in HPV-induced dysplastic and neoplastic cervical pathologies. Journal of cellular physiology. 2014 Apr;229:463-70. 156. Martinez I, Wang J, Hobson KF, Ferris RL, Khan SA. Identification of differentially expressed genes in HPV-positive and HPV-negative oropharyngeal squamous cell carcinomas. Eur J Cancer. 2007 Jan;43:415-32. 157. Pyeon D, Newton MA, Lambert PF, den Boon JA, Sengupta S, Marsit CJ, et al. Fundamental differences in cell cycle deregulation in human papillomavirus-positive and human papillomavirus-negative head/neck and cervical cancers. Cancer Res. 2007 May 15;67:4605-19. 158. Slebos RJ, Yi Y, Ely K, Carter J, Evjen A, Zhang X, et al. Gene expression differences associated with human papillomavirus status in head and neck squamous cell carcinoma. Clin Cancer Res. 2006 Feb 1;12:701-9. 159. Lohavanichbutr P, Houck J, Fan W, Yueh B, Mendez E, Futran N, et al. Genomewide gene expression profiles of HPV-positive and HPV-negative oropharyngeal cancer: potential implications for treatment choices. Archives of otolaryngology--head & neck surgery. 2009 Feb;135:180-8. 160. Schlecht NF, Burk RD, Adrien L, Dunne A, Kawachi N, Sarta C, et al. Gene expression profiles in HPV-infected head and neck cancer. J Pathol. 2007 Nov;213:283- 93. 161. Sano T, Masuda N, Oyama T, Nakajima T. Overexpression of p16 and p14ARF is associated with human papillomavirus infection in cervical squamous cell carcinoma and dysplasia. Pathology international. 2002 May-Jun;52:375-83. 162. Quelle DE, Zindy F, Ashmun RA, Sherr CJ. Alternative reading frames of the INK4a tumor suppressor gene encode two unrelated proteins capable of inducing cell cycle arrest. Cell. 1995 Dec 15;83:993-1000. 163. Khleif SN, DeGregori J, Yee CL, Otterson GA, Kaye FJ, Nevins JR, et al. Inhibition of cyclin D-CDK4/CDK6 activity is associated with an E2F-mediated induction of cyclin kinase inhibitor activity. Proc Natl Acad Sci U S A. 1996 Apr 30;93:4350-4. 164. Steenbergen RD, Snijders PJ, Heideman DA, Meijer CJ. Clinical implications of (epi)genetic changes in HPV-induced cervical precancerous lesions. Nat Rev Cancer. 2014 Jun;14:395-405. 165. Fang J, Zhang H, Jin S. Epigenetics and cervical cancer: from pathogenesis to therapy. Tumour Biol. 2014 Jun;35:5083-93. 166. Leonard SM, Wei W, Collins SI, Pereira M, Diyaf A, Constandinou-Williams C, et al. Oncogenic human papillomavirus imposes an instructive pattern of DNA methylation changes which parallel the natural history of cervical HPV infection in young women. Carcinogenesis. 2012 Jul;33:1286-93. 167. Sartor MA, Dolinoy DC, Jones TR, Colacino JA, Prince ME, Carey TE, et al. Genome-wide methylation and expression differences in HPV(+) and HPV(-) squamous cell carcinoma cell lines are consistent with divergent mechanisms of carcinogenesis. Epigenetics : official journal of the DNA Methylation Society. 2011 Jun;6:777-87. 168. Duenas-Gonzalez A, Lizano M, Candelaria M, Cetina L, Arce C, Cervera E. Epigenetics of cervical cancer. An overview and therapeutic perspectives. Molecular cancer. 2005;4:38.

122 169. Kostareli E, Holzinger D, Bogatyrova O, Hielscher T, Wichmann G, Keck M, et al. HPV-related methylation signature predicts survival in oropharyngeal squamous cell carcinomas. The Journal of clinical investigation. 2013 Jun 3;123:2488-501. 170. Parkin DM. The global health burden of infection-associated cancers in the year 2002. Int J Cancer. 2006 Jun 15;118:3030-44. 171. Devany M, Kappes F, Chen KM, Markovitz DM, Matsuo H. Solution NMR structure of the N-terminal domain of the human DEK protein. Protein science : a publication of the Protein Society. 2008 Feb;17:205-15. 172. Devany M, Kotharu NP, Matsuo H. Solution NMR structure of the C-terminal domain of the human protein DEK. Protein science : a publication of the Protein Society. 2004 Aug;13:2252-9. 173. Kappes F, Scholten I, Richter N, Gruss C, Waldmann T. Functional domains of the ubiquitous chromatin protein DEK. Mol Cell Biol. 2004 Jul;24:6000-10. 174. Broxmeyer HE, Mor-Vaknin N, Kappes F, Legendre M, Saha AK, Ou X, et al. Concise review: role of DEK in stem/progenitor cell biology. Stem cells. 2013 Aug;31:1447-53. 175. Kappes F, Burger K, Baack M, Fackelmayer FO, Gruss C. Subcellular localization of the human proto-oncogene protein DEK. J Biol Chem. 2001 Jul 13;276:26317-23. 176. Saha AK, Kappes F, Mundade A, Deutzmann A, Rosmarin DM, Legendre M, et al. Intercellular trafficking of the nuclear oncoprotein DEK. Proc Natl Acad Sci U S A. 2013 Apr 23;110:6847-52. 177. Waldmann T, Baack M, Richter N, Gruss C. Structure-specific binding of the proto-oncogene protein DEK to DNA. Nucleic Acids Res. 2003 Dec 1;31:7003-10. 178. Waldmann T, Eckerich C, Baack M, Gruss C. The ubiquitous chromatin protein DEK alters the structure of DNA by introducing positive supercoils. J Biol Chem. 2002 Jul 12;277:24988-94. 179. Hu HG, Scholten I, Gruss C, Knippers R. The distribution of the DEK protein in mammalian chromatin. Biochemical and biophysical research communications. 2007 Jul 13;358:1008-14. 180. Kappes F, Waldmann T, Mathew V, Yu J, Zhang L, Khodadoust MS, et al. The DEK oncoprotein is a Su(var) that is essential to heterochromatin integrity. Genes Dev. Apr 1;25:673-8. 181. Kappes F, Damoc C, Knippers R, Przybylski M, Pinna LA, Gruss C. Phosphorylation by protein kinase CK2 changes the DNA binding properties of the human chromatin protein DEK. Mol Cell Biol. 2004 Jul;24:6011-20. 182. Cleary J, Sitwala KV, Khodadoust MS, Kwok RP, Mor-Vaknin N, Cebrat M, et al. p300/CBP-associated factor drives DEK into interchromatin granule clusters. J Biol Chem. 2005 Sep 9;280:31760-7. 183. Mor-Vaknin N, Kappes F, Dick AE, Legendre M, Damoc C, Teitz-Tennenbaum S, et al. DEK in the synovium of patients with juvenile idiopathic arthritis: characterization of DEK antibodies and posttranslational modification of the DEK autoantigen. Arthritis and rheumatism. 2011 Feb;63:556-67. 184. Fahrer J, Popp O, Malanga M, Beneke S, Markovitz DM, Ferrando-May E, et al. High-affinity interaction of poly(ADP-ribose) and the human DEK oncoprotein depends upon chain length. Biochemistry. 2010 Aug 24;49:7119-30.

123 185. Gamble MJ, Fisher RP. SET and PARP1 remove DEK from chromatin to permit access by the transcription machinery. Nat Struct Mol Biol. 2007 Jun;14:548-55. 186. Kappes F, Fahrer J, Khodadoust MS, Tabbert A, Strasser C, Mor-Vaknin N, et al. DEK is a poly(ADP-ribose) acceptor in apoptosis and mediates resistance to genotoxic stress. Mol Cell Biol. 2008 May;28:3245-57. 187. Ko SI, Lee IS, Kim JY, Kim SM, Kim DW, Lee KS, et al. Regulation of histone acetyltransferase activity of p300 and PCAF by proto-oncogene protein DEK. FEBS letters. 2006 May 29;580:3217-22. 188. Sammons M, Wan SS, Vogel NL, Mientjes EJ, Grosveld G, Ashburner BP. Negative regulation of the RelA/p65 transactivation function by the product of the DEK proto-oncogene. J Biol Chem. 2006 Sep 15;281:26802-12. 189. Campillos M, Garcia MA, Valdivieso F, Vazquez J. Transcriptional activation by AP-2alpha is modulated by the oncogene DEK. Nucleic Acids Res. 2003 Mar 1;31:1571-5. 190. Sawatsubashi S, Murata T, Lim J, Fujiki R, Ito S, Suzuki E, et al. A histone chaperone, DEK, transcriptionally coactivates a nuclear receptor. Genes Dev. 2010 Jan 15;24:159-70. 191. Koleva RI, Ficarro SB, Radomska HS, Carrasco-Alfonso MJ, Alberta JA, Webber JT, et al. C/EBPalpha and DEK coordinately regulate myeloid differentiation. Blood. 2012 May 24;119:4878-88. 192. Shibata T, Kokubu A, Miyamoto M, Hosoda F, Gotoh M, Tsuta K, et al. DEK oncoprotein regulates transcriptional modifiers and sustains tumor initiation activity in high-grade neuroendocrine carcinoma of the lung. Oncogene. 2010 Aug 19;29:4671-81. 193. Karam M, Thenoz M, Capraro V, Robin JP, Pinatel C, Lancon A, et al. Chromatin redistribution of the DEK oncoprotein represses hTERT transcription in leukemias. Neoplasia. 2014 Jan;16:21-30. 194. Sanden C, Jarvstrat L, Lennartsson A, Brattas PL, Nilsson B, Gullberg U. The DEK oncoprotein binds to highly and ubiquitously expressed genes with a dual role in their transcriptional regulation. Molecular cancer. 2014;13:215. 195. Alexiadis V, Waldmann T, Andersen J, Mann M, Knippers R, Gruss C. The protein encoded by the proto-oncogene DEK changes the topology of chromatin and reduces the efficiency of DNA replication in a chromatin-specific manner. Genes Dev. 2000 Jun 1;14:1308-12. 196. Ivanauskiene K, Delbarre E, McGhie JD, Kuntziger T, Wong LH, Collas P. The PML-associated protein DEK regulates the balance of H3.3 loading on chromatin and is important for telomere integrity. Genome research. 2014 Oct;24:1584-94. 197. Szenker E, Ray-Gallet D, Almouzni G. The double face of the histone variant H3.3. Cell research. 2011 Mar;21:421-34. 198. Hollenbach AD, McPherson CJ, Mientjes EJ, Iyengar R, Grosveld G. Daxx and histone deacetylase II associate with chromatin through an interaction with core histones and the chromatin-associated protein Dek. Journal of cell science. 2002 Aug 15;115:3319-30. 199. Kavanaugh GM, Wise-Draper TM, Morreale RJ, Morrison MA, Gole B, Schwemberger S, et al. The human DEK oncogene regulates DNA damage response signaling and repair. Nucleic Acids Res. Jun 7.

124 200. Khodadoust MS, Verhaegen M, Kappes F, Riveiro-Falkenbach E, Cigudosa JC, Kim DS, et al. Melanoma proliferation and chemoresistance controlled by the DEK oncogene. Cancer Res. 2009 Aug 15;69:6405-13. 201. Deutzmann A, Ganz M, Schonenberger F, Vervoorts J, Kappes F, Ferrando-May E. The human oncoprotein and chromatin architectural factor DEK counteracts DNA replication stress. Oncogene. 2014 Oct 27. 202. von Lindern M, Fornerod M, van Baal S, Jaegle M, de Wit T, Buijs A, et al. The translocation (6;9), associated with a specific subtype of acute myeloid leukemia, results in the fusion of two genes, dek and can, and the expression of a chimeric, leukemia- specific dek-can mRNA. Mol Cell Biol. 1992 Apr;12:1687-97. 203. Oancea C, Ruster B, Henschler R, Puccetti E, Ruthardt M. The t(6;9) associated DEK/CAN fusion protein targets a population of long-term repopulating hematopoietic stem cells for leukemogenic transformation. Leukemia. Nov;24:1910-9. 204. Privette Vinnedge LM, McClaine R, Wagh PK, Wikenheiser-Brokamp KA, Waltz SE, Wells SI. The human DEK oncogene stimulates beta-catenin signaling, invasion and mammosphere formation in breast cancer. Oncogene. Jun 16;30:2741-52. 205. Datta A, Adelson ME, Mogilevkin Y, Mordechai E, Sidi AA, Trama JP. Oncoprotein DEK as a tissue and urinary biomarker for bladder cancer. BMC cancer. 2011;11:234. 206. Kondoh N, Wakatsuki T, Ryo A, Hada A, Aihara T, Horiuchi S, et al. Identification and characterization of genes associated with human hepatocellular carcinogenesis. Cancer Res. 1999 Oct 1;59:4990-6. 207. Wise-Draper TM, Mintz-Cole RA, Morris TA, Simpson DS, Wikenheiser-Brokamp KA, Currier MA, et al. Overexpression of the cellular DEK protein promotes epithelial transformation in vitro and in vivo. Cancer Res. 2009 Mar 1;69:1792-9. 208. Wise-Draper TM, Morreale RJ, Morris TA, Mintz-Cole RA, Hoskins EE, Balsitis SJ, et al. DEK proto-oncogene expression interferes with the normal epithelial differentiation program. Am J Pathol. 2009 Jan;174:71-81. 209. Wise-Draper TM, Allen HV, Thobe MN, Jones EE, Habash KB, Munger K, et al. The human DEK proto-oncogene is a senescence inhibitor and an upregulated target of high-risk human papillomavirus E7. J Virol. 2005 Nov;79:14309-17. 210. Wise-Draper TM, Allen HV, Jones EE, Habash KB, Matsuo H, Wells SI. Apoptosis inhibition by the human DEK oncoprotein involves interference with p53 functions. Mol Cell Biol. 2006 Oct;26:7506-19. 211. Carro MS, Spiga FM, Quarto M, Di Ninni V, Volorio S, Alcalay M, et al. DEK Expression is controlled by E2F and deregulated in diverse tumor types. Cell Cycle. 2006 Jun;5:1202-7. 212. Sitwala KV, Adams K, Markovitz DM. YY1 and NF-Y binding sites regulate the transcriptional activity of the dek and dek-can promoter. Oncogene. 2002 Dec 12;21:8862-70. 213. Privette Vinnedge LM, Ho SM, Wikenheiser-Brokamp KA, Wells SI. The DEK oncogene is a target of steroid hormone receptor signaling in breast cancer. PloS one. 2012;7:e46985. 214. Babaei-Jadidi R, Li N, Saadeddin A, Spencer-Dene B, Jandke A, Muhammad B, et al. FBXW7 influences murine intestinal homeostasis and cancer, targeting Notch,

125 Jun, and DEK for degradation. The Journal of experimental medicine. 2011 Feb 14;208:295-312. 215. Privette Vinnedge LM, Benight NM, Wagh PK, Pease NA, Nashu MA, Serrano- Lopez J, et al. The DEK oncogene promotes cellular proliferation through paracrine Wnt signaling in Ron receptor-positive breast cancers. Oncogene. 2014 Jun 23. 216. Sanden C, Ageberg M, Petersson J, Lennartsson A, Gullberg U. Forced expression of the DEK-NUP214 fusion protein promotes proliferation dependent on upregulation of mTOR. BMC cancer. 2013;13:440. 217. Secchiero P, Voltan R, di Iasio MG, Melloni E, Tiribelli M, Zauli G. The oncogene DEK promotes leukemic cell survival and is downregulated by both Nutlin-3 and chlorambucil in B-chronic lymphocytic leukemic cells. Clin Cancer Res. Mar 15;16:1824- 33. 218. Broxmeyer HE, Kappes F, Mor-Vaknin N, Legendre M, Kinzfogl J, Cooper S, et al. DEK regulates hematopoietic stem engraftment and progenitor cell proliferation. Stem cells and development. 2012 Jun 10;21:1449-54. 219. Liu S, Wang X, Sun F, Kong J, Li Z, Lin Z. DEK overexpression is correlated with the clinical features of breast cancer. Pathology international. 2012 Mar;62:176-81. 220. Kappes F, Khodadoust MS, Yu L, Kim DS, Fullen DR, Markovitz DM, et al. DEK expression in melanocytic lesions. Human pathology. 2011 Jul;42:932-8. 221. Wise-Draper TM, Draper DJ, Gutkind JS, Molinolo AA, Wikenheiser-Brokamp KA, Wells SI. Future directions and treatment strategies for head and neck squamous cell carcinomas. Transl Res. Sep;160:167-77. 222. Murray KJ, Szer W, Grom AA, Donnelly P, Levinson JE, Giannini EH, et al. Antibodies to the 45 kDa DEK nuclear antigen in pauciarticular onset juvenile rheumatoid arthritis and iridocyclitis: selective association with MHC gene. The Journal of rheumatology. 1997 Mar;24:560-7. 223. Sierakowska H, Williams KR, Szer IS, Szer W. The putative oncoprotein DEK, part of a chimera protein associated with acute myeloid leukaemia, is an autoantigen in juvenile rheumatoid arthritis. Clinical and experimental immunology. 1993 Dec;94:435-9. 224. Szer IS, Sierakowska H, Szer W. A novel autoantibody to the putative oncoprotein DEK in pauciarticular onset juvenile rheumatoid arthritis. The Journal of rheumatology. 1994 Nov;21:2136-42. 225. Haber PL, Osborn TG, Moore TL. Antinuclear antibody in juvenile rheumatoid arthritis sera reacts with 50-40 kDa antigen(s) found in HeLa nuclear extracts. The Journal of rheumatology. 1989 Jul;16:949-54. 226. Dong X, Michelis MA, Wang J, Bose R, DeLange T, Reeves WH. Autoantibodies to DEK oncoprotein in a patient with systemic lupus erythematosus and sarcoidosis. Arthritis and rheumatism. 1998 Aug;41:1505-10. 227. Mor-Vaknin N, Punturieri A, Sitwala K, Faulkner N, Legendre M, Khodadoust MS, et al. The DEK nuclear autoantigen is a secreted chemotactic factor. Mol Cell Biol. 2006 Dec;26:9484-96. 228. Wesche H, Henzel WJ, Shillinglaw W, Li S, Cao Z. MyD88: an adapter that recruits IRAK to the IL-1 receptor complex. Immunity. 1997 Dec;7:837-47. 229. Neumann D, Kollewe C, Resch K, Martin MU. The death domain of IRAK-1: an oligomerization domain mediating interactions with MyD88, Tollip, IRAK-1, and IRAK-4. Biochemical and biophysical research communications. 2007 Mar 23;354:1089-94.

126 230. Kollewe C, Mackensen AC, Neumann D, Knop J, Cao P, Li S, et al. Sequential autophosphorylation steps in the interleukin-1 receptor-associated kinase-1 regulate its availability as an adapter in interleukin-1 signaling. J Biol Chem. 2004 Feb 13;279:5227- 36. 231. Takaesu G, Kishida S, Hiyama A, Yamaguchi K, Shibuya H, Irie K, et al. TAB2, a novel adaptor protein, mediates activation of TAK1 MAPKKK by linking TAK1 to TRAF6 in the IL-1 signal transduction pathway. Mol Cell. 2000 Apr;5:649-58. 232. Jiang Z, Ninomiya-Tsuji J, Qian Y, Matsumoto K, Li X. Interleukin-1 (IL-1) receptor-associated kinase-dependent IL-1-induced signaling complexes phosphorylate TAK1 and TAB2 at the plasma membrane and activate TAK1 in the cytosol. Mol Cell Biol. 2002 Oct;22:7158-67. 233. Deng L, Wang C, Spencer E, Yang L, Braun A, You J, et al. Activation of the IkappaB kinase complex by TRAF6 requires a dimeric ubiquitin-conjugating enzyme complex and a unique polyubiquitin chain. Cell. 2000 Oct 13;103:351-61. 234. Qian Y, Commane M, Ninomiya-Tsuji J, Matsumoto K, Li X. IRAK-mediated translocation of TRAF6 and TAB2 in the interleukin-1-induced activation of NFkappa B. J Biol Chem. 2001 Nov 9;276:41661-7. 235. Kopp E, Medzhitov R, Carothers J, Xiao C, Douglas I, Janeway CA, et al. ECSIT is an evolutionarily conserved intermediate in the Toll/IL-1 signal transduction pathway. Genes Dev. 1999 Aug 15;13:2059-71. 236. Yamin TT, Miller DK. The interleukin-1 receptor-associated kinase is degraded by proteasomes following its phosphorylation. J Biol Chem. 1997 Aug 22;272:21540-7. 237. Conze DB, Wu CJ, Thomas JA, Landstrom A, Ashwell JD. Lys63-linked polyubiquitination of IRAK-1 is required for interleukin-1 receptor- and toll-like receptor- mediated NF-kappaB activation. Mol Cell Biol. 2008 May;28:3538-47. 238. Su J, Richter K, Zhang C, Gu Q, Li L. Differential regulation of interleukin-1 receptor associated kinase 1 (IRAK1) splice variants. Molecular immunology. 2007 Feb;44:900-5. 239. Gottipati S, Rao NL, Fung-Leung WP. IRAK1: a critical signaling mediator of innate immunity. Cellular signalling. 2008 Feb;20:269-76. 240. Cao Z, Henzel WJ, Gao X. IRAK: a kinase associated with the interleukin-1 receptor. Science. 1996 Feb 23;271:1128-31. 241. Li S, Strelow A, Fontana EJ, Wesche H. IRAK-4: a novel member of the IRAK family with the properties of an IRAK-kinase. Proc Natl Acad Sci U S A. 2002 Apr 16;99:5567-72. 242. Rhyasen GW, Starczynowski DT. IRAK signalling in cancer. British journal of cancer. 2014 Oct 7. 243. Wesche H, Gao X, Li X, Kirschning CJ, Stark GR, Cao Z. IRAK-M is a novel member of the Pelle/interleukin-1 receptor-associated kinase (IRAK) family. J Biol Chem. 1999 Jul 2;274:19403-10. 244. Kobayashi K, Hernandez LD, Galan JE, Janeway CA, Jr., Medzhitov R, Flavell RA. IRAK-M is a negative regulator of Toll-like receptor signaling. Cell. 2002 Jul 26;110:191-202. 245. Rhyasen GW, Bolanos L, Fang J, Jerez A, Wunderlich M, Rigolino C, et al. Targeting IRAK1 as a therapeutic approach for myelodysplastic syndrome. Cancer Cell. 2013 Jul 8;24:90-104.

127 246. Srivastava R, Geng D, Liu Y, Zheng L, Li Z, Joseph MA, et al. Augmentation of therapeutic responses in melanoma by inhibition of IRAK-1,-4. Cancer Res. 2012 Dec 1;72:6209-16. 247. Gondek LP, Tiu R, O'Keefe CL, Sekeres MA, Theil KS, Maciejewski JP. Chromosomal lesions and uniparental disomy detected by SNP arrays in MDS, MDS/MPD, and MDS-derived AML. Blood. 2008 Feb 1;111:1534-42. 248. Starczynowski DT, Kuchenbauer F, Argiropoulos B, Sung S, Morin R, Muranyi A, et al. Identification of miR-145 and miR-146a as mediators of the 5q- syndrome phenotype. Nat Med. 2010 Jan;16:49-58. 249. Forastiere AA, Ang KK, Brizel D, Brockstein BE, Burtness BA, Cmelak AJ, et al. Head and neck cancers. J Natl Compr Canc Netw. 2008 Aug;6:646-95. 250. Gupta S, Kong W, Peng Y, Miao Q, Mackillop WJ. Temporal trends in the incidence and survival of cancers of the upper aerodigestive tract in Ontario and the United States. Int J Cancer. 2009 Nov 1;125:2159-65. 251. Jabbar S, Strati K, Shin MK, Pitot HC, Lambert PF. Human papillomavirus type 16 E6 and E7 oncoproteins act synergistically to cause head and neck cancer in mice. Virology. Nov 10;407:60-7. 252. Strati K, Lambert PF. Role of Rb-dependent and Rb-independent functions of papillomavirus E7 oncogene in head and neck cancer. Cancer Res. 2007 Dec 15;67:11585-93. 253. Herber R, Liem A, Pitot H, Lambert PF. Squamous epithelial hyperplasia and carcinoma in mice transgenic for the human papillomavirus type 16 E7 oncogene. J Virol. 1996 Mar;70:1873-81. 254. Lu ZL, Luo DZ, Wen JM. Expression and significance of tumor-related genes in HCC. World J Gastroenterol. 2005 Jul 7;11:3850-4. 255. Bradford CR, Zhu S, Ogawa H, Ogawa T, Ubell M, Narayan A, et al. P53 mutation correlates with cisplatin sensitivity in head and neck squamous cell carcinoma lines. Head Neck. 2003 Aug;25:654-61. 256. Deyoung MP, Ellisen LW. p63 and p73 in human cancer: defining the network. Oncogene. 2007 Aug 9;26:5169-83. 257. Moll UM, Slade N. p63 and p73: roles in development and tumor formation. Molecular cancer research : MCR. 2004 Jul;2:371-86. 258. Su X, Chakravarti D, Flores ER. p63 steps into the limelight: crucial roles in the suppression of tumorigenesis and metastasis. Nat Rev Cancer. 2013 Feb;13:136-43. 259. Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA: a cancer journal for clinicians. 2005 Mar-Apr;55:74-108. 260. Alexandrova EM, Petrenko O, Nemajerova A, Romano RA, Sinha S, Moll UM. DeltaNp63 regulates select routes of reprogramming via multiple mechanisms. Cell Death Differ. 2013 Dec;20:1698-708. 261. Truong AB, Kretz M, Ridky TW, Kimmel R, Khavari PA. p63 regulates proliferation and differentiation of developmentally mature keratinocytes. Genes Dev. 2006 Nov 15;20:3185-97. 262. Koster MI, Dai D, Marinari B, Sano Y, Costanzo A, Karin M, et al. p63 induces key target genes required for epidermal morphogenesis. Proc Natl Acad Sci U S A. 2007 Feb 27;104:3255-60.

128 263. Pignon JC, Grisanzio C, Geng Y, Song J, Shivdasani RA, Signoretti S. p63- expressing cells are the stem cells of developing prostate, bladder, and colorectal epithelia. Proc Natl Acad Sci U S A. 2013 May 14;110:8105-10. 264. Mills AA, Zheng B, Wang XJ, Vogel H, Roop DR, Bradley A. p63 is a p53 homologue required for limb and epidermal morphogenesis. Nature. 1999 Apr 22;398:708-13. 265. Yang A, Schweitzer R, Sun D, Kaghad M, Walker N, Bronson RT, et al. p63 is essential for regenerative proliferation in limb, craniofacial and epithelial development. Nature. 1999 Apr 22;398:714-8. 266. Romano RA, Smalley K, Magraw C, Serna VA, Kurita T, Raghavan S, et al. DeltaNp63 knockout mice reveal its indispensable role as a master regulator of epithelial development and differentiation. Development. 2012 Feb;139:772-82. 267. Hoskins EE, Morris TA, Higginbotham JM, Spardy N, Cha E, Kelly P, et al. Fanconi anemia deficiency stimulates HPV-associated hyperplastic growth in organotypic epithelial raft culture. Oncogene. 2009 Feb 5;28:674-85. 268. Chaturvedi AK, Engels EA, Pfeiffer RM, Hernandez BY, Xiao W, Kim E, et al. Human papillomavirus and rising oropharyngeal cancer incidence in the United States. J Clin Oncol. Nov 10;29:4294-301. 269. Rocco JW, Leong CO, Kuperwasser N, DeYoung MP, Ellisen LW. p63 mediates survival in squamous cell carcinoma by suppression of p73-dependent apoptosis. Cancer Cell. 2006 Jan;9:45-56. 270. Strati K, Pitot HC, Lambert PF. Identification of biomarkers that distinguish human papillomavirus (HPV)-positive versus HPV-negative head and neck cancers in a mouse model. Proc Natl Acad Sci U S A. 2006 Sep 19;103:14152-7. 271. Marur S, Forastiere AA. Head and neck cancer: changing epidemiology, diagnosis, and treatment. Mayo Clinic proceedings. 2008 Apr;83:489-501. 272. Adams AK, Hallenbeck GE, Casper KA, Patil YJ, Wilson KM, Kimple RJ, et al. DEK promotes HPV-positive and -negative head and neck cancer cell proliferation. Oncogene. 2014 Mar 10;0:1. 273. Hung PS, Liu CJ, Chou CS, Kao SY, Yang CC, Chang KW, et al. miR-146a enhances the oncogenicity of oral carcinoma by concomitant targeting of the IRAK1, TRAF6 and NUMB genes. PloS one. 2013;8:e79926. 274. Fang J, Rhyasen G, Bolanos L, Rasch C, Varney M, Wunderlich M, et al. Cytotoxic effects of bortezomib in myelodysplastic syndrome/acute myeloid leukemia depend on autophagy-mediated lysosomal degradation of TRAF6 and repression of PSMA1. Blood. 2012 Jul 26;120:858-67. 275. Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, et al. Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Science signaling. 2013 Apr 2;6:pl1. 276. Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer discovery. 2012 May;2:401-4. 277. Flicek P, Amode MR, Barrell D, Beal K, Brent S, Carvalho-Silva D, et al. Ensembl 2012. Nucleic Acids Res. 2012 Jan;40:D84-90.

129 278. Chen J, Bardes EE, Aronow BJ, Jegga AG. ToppGene Suite for gene list enrichment analysis and candidate gene prioritization. Nucleic Acids Res. 2009 Jul;37:W305-11. 279. Komurov K, Dursun S, Erdin S, Ram PT. NetWalker: a contextual network analysis tool for functional genomics. BMC genomics. 2012;13:282. 280. Xie X, Lu J, Kulbokas EJ, Golub TR, Mootha V, Lindblad-Toh K, et al. Systematic discovery of regulatory motifs in human promoters and 3' UTRs by comparison of several mammals. Nature. 2005 Mar 17;434:338-45. 281. Galli F, Rossi M, D'Alessandra Y, De Simone M, Lopardo T, Haupt Y, et al. MDM2 and Fbw7 cooperate to induce p63 protein degradation following DNA damage and cell differentiation. Journal of cell science. 2010 Jul 15;123:2423-33. 282. Cancer Genome Atlas N. Comprehensive genomic characterization of head and neck squamous cell carcinomas. Nature. 2015 Jan 29;517:576-82. 283. Yadav A, Kumar B, Datta J, Teknos TN, Kumar P. IL-6 promotes head and neck tumor metastasis by inducing epithelial-mesenchymal transition via the JAK-STAT3- SNAIL signaling pathway. Molecular cancer research : MCR. 2011 Dec;9:1658-67.

130