Investigating the Role of DDX28 As a Novel Regulator of the Hypoxic Cap- Dependent Translation Initiation Complex

Investigating the role of DDX28 as a novel regulator of the hypoxic cap- dependent translation initiation complex

Sonia Evagelou

A Thesis presented to The University of Guelph

In partial fulfilment of requirements for the degree of Masters of Science in Molecular and Cellular Biology

Guelph, Ontario, Canada

INVESTIGATING THE ROLE OF DDX28 AS A NOVEL REGULATOR OF THE HYPOXIC CAP-DEPENDENT TRANSLATION INITIATION COMPLEX

Sonia Evagelou Advisor: University of Guelph, 2018 Dr. James Uniacke

The most common form of translation initiation in eukaryotes requires recruitment of the eIF4F complex to the m7GTP cap located at the 5’ end of all cellular mRNAs. eIF4F is composed of the cap-binding factor eIF4E, the RNA helicase eIF4A, and the scaffolding protein eIF4G, which together function to initiate translation. However, in response to hypoxia, a common feature of several physiological and pathological processes, eIF4E is inhibited. Hypoxic cells maintain translation of select transcripts by utilizing the eIF4E homologue, eIF4E2, as an alternative cap- binding protein as part of the eIF4FH complex. We have identified DDX28 as a novel regulator of eIF4FH by demonstrating that knocking-down DDX28 results in an overall increase in cap- bound eIF4E2 and eIF4E2-mediated translation under hypoxia. Additionally, depletion of

DDX28 confers a proliferative advantage to cells grown in hypoxic conditions, which we suspect is a consequence of the translational upregulation of a subset of hypoxic mRNAs.

Dedication

In loving memory of my late grandmother. Brevis a naturâ nobis vita data est; at memoria bene redditæ vitæ est sempiterna.

iii

Acknowledgements

First and foremost I would like to thank my advisor Dr. James Uniacke for his excellent mentorship over the past several years, in addition to providing me with the liberty to explore this project as an independent researcher. My development as a scientist in his lab has far exceeded what I imagined was possible in such a short period of time, and the knowledge I have gained will not only benefit me in the field of research, but in everyday life as well. I would also like to thank Erin Specker whose extensive help and generosity has contributed significantly to my positive experience in the lab. My gratitude also extends to past and present members of the lab: Sara Timpano, Dr. Phil Medeiros, Gaelan Melanson, Andrea Brumwell, Sydney Pascetta,

Nicole Kelly, Joszef Varga, Brianna Guild and Olivia Bebenek for their continued assistance and friendship over the years. To my advisory committee, Dr. Ray Lu and Dr. Jnanankur Bag, thank you for your insight and advice throughout the course of my project.

I would like to give a special thanks to my family for their incredible support throughout this journey, none of this would have been possible without you. Additionally, to my wonderful boyfriend Daniel Prevedel, thanks for always being a shoulder to cry on and letting me talk about my project for hours when no one else wanted to listen!

Declaration of Work Performed

Olivia Bebenek performed the hypoxic polysome profiling for RNA analysis and the normoxic and hypoxic polysome RT-PCR, and qPCR. Bromodeoxyuridine labeling experiments were performed by Erin Specker. All other experiments were performed by the author.

Table of Contents

ABSTRACT ...... II

DEDICATION ...... III

ACKNOWLEDGEMENTS ...... IV

DECLARATION OF WORK PERFORMED ...... V

LIST OF FIGURES ...... VIII

LIST OF ABBREVIATIONS ...... IX Chapter 1. Introduction ...... 1 1.1 Hypoxia in Development and Disease ...... 1 1.2 The HIF Program ...... 2 1.3 Cap-Dependent Translation Initiation and eIF4F ...... 8 1.4 mTOR and Hypoxic Regulation of eIF4F ...... 11 1.5 IRES-Mediated Translation ...... 14 1.6 Hypoxic Cap-Dependent Translation and eIF4FH ...... 15 1.7 Regulation of eIF4FH by DDX28 ...... 17 1.8 Hypothesis and Research Objectives ...... 22 Chapter 2. Materials and Methods ...... 24 2.1 Cell Culture and Reagents ...... 24 2.2 Generation of Stable Cell Lines ...... 24 2.2.1 DDX28 Knockdown Cells ...... 24 2.2.2 DDX28-FLAG Cells ...... 25 2.3 Western Blot Analysis and Antibodies ...... 25 2.4 Polysome Profiling ...... 26 2.4.1 Polysome Protein Analysis ...... 27 2.4.2 Polysome RNA Isolation, RT-PCR, and quantitative PCR ...... 27 2.5 GFP-Trap Immunoprecipitation ...... 28 2.6 FLAG Immunoprecipitation ...... 29 2.7 Analysis of Cap-Binding Proteins ...... 30 2.8 Crystal Violet Growth Curves ...... 31

2.9 Bromodeoxyuridine Labeling ...... 32 Chapter 3. Results ...... 33 3.1 HIF-2α mediates the interaction between DDX28 and eIF4E2 ...... 33 3.2 Knockdown of DDX28 enhances the cap-binding affinity of eIF4E2 ...... 35 3.3 Knockdown of DDX28 increases eIF4E2-mediated translation ...... 38 3.3 Knockdown of DDX28 increases translation of EGFR mRNA ...... 42 3.4 Knockdown of DDX28 Increases Cell Viability in Hypoxia ...... 44 3.5 Knockdown of DDX28 Increases Cell Proliferation Rates in Hypoxia ...... 46 Chapter 4. Discussion ...... 49 Chapter 5. Conclusions and Future Directions ...... 61 References ...... 63

vii

List of Figures

Figure 1. Alignment of HIF-1α and HIF-2α structural domains...... 5

Figure 2. Overview of the regulation of HIF-α in response to changing oxygen concentrations..6

Figure 3. Proposed model of the eukaryotic translation initiation complex mediating cap- dependent translation during normoxic and hypoxic conditions...... 12

Figure 4. Immunoprecipitation of FLAG-eIF4E2 from normoxic and hypoxic U-87 MG cell lysates...... 18

Figure 5. Schematic diagram of the conserved motifs of the DEAD-box protein superfamily...19

Figure 6. Standard workflow of a cap-affinity assay ...... 22

Figure 7. Isolating polyribosomes using polysome profiling ...... 23

Figure 8. DDX28 interacts with HIF-2α in U-87 MG ...... 34

Figure 9. Knockdown of DDX28 increases the cap-binding affinity of eIF4E2 in U-87 MG.....36

Figure 10. Knockdown of DDX28 in U-87 MG significantly increases polysome-associated eIF4E2 in hypoxia...... 40

Figure 11. Knockdown of DDX28 in U-87 MG significantly increases translation of EGFR mRNA in hypoxia...... 43

Figure 12. Knockdown of DDX28 increases U-87 MG viability in hypoxia...... 45

Figure 13. Knockdown of DDX28 increases U-87 MG proliferation rates in hypoxia...... 47

Figure 14. Proposed model of DDX28 regulation of the hypoxic translation initiation complex eIF4FH...... 57

Figure 15. Human Protein Atlas DDX28 protein expression summary...... 59

viii

List of Abbreviations

4E-BP eIF4E-binding protein aHIF antisense HIF

AMPK adenosine monophosphate-activated protein kinase

ATCC American type culture collection

ATP adenosine triphosphate

BCA bicinchoninic acid bHLH/PAS basic helix-loop-helix per-arnt-sim

BrdU bromodeoxyuridine

BSA bovine serum albumin

CBP creb-binding protein

CCRCC clear cell renal cell carcinoma

CDK4/6 cyclin-dependent kinase 4/6 cDNA complementary DNA

CHX cycloheximide

CO2 carbon dioxide

CTAD carboxy-terminal activation domain

DDX28 DEAD-box protein 28

DDX3 DEAD-box protein 3

DDX6 DEAD-box protein 6

DMEM dulbecco’s modified eagle medium

E embryonic day

EDTA ethylenediaminetetraacetic acid

EGFP enhanced green fluorescent protein

EGFR epidermal growth factor receptor eIF1A eukaryotic translation initiation factor 1A eIF3 eukaryotic translation initiation factor 3 eIF4A eukaryotic translation initiation factor 4A eIF4B eukaryotic translation initiation factor 4B eIF4E eukaryotic translation initiation factor 4E eIF4E2 eukaryotic translation initiation factor 4E2 eIF4F eukaryotic translation initiation factor 4F eIF4FH eukaryotic translation initiation factor 4FH eIF4G eukaryotic translation initiation factor 4G eIF4G3 eukaryotic translation initiation factor 4G3 eIF4H eukaryotic translation initiation factor 4H eIF5 eukaryotic translation initiation factor 5

EPAS1 endothelial PAS-containing protein 1

EPO erythropoietin

FBS fetal bovine serum

Fe(II) ferrous

FIH factor inhibiting HIF

GAP GTPase activating protein

GFP green fluorescent protein

GLUT-1 glucose transporter 1

HAF HIF-associated factor

HCl hydrogen chloride

HIF-1α hypoxia inducible factor 1 alpha

HIF-2α hypoxia inducible factor 2 alpha

HIF-β hypoxia inducible factor beta

HRE hypoxic-response element

HRP horseradish peroxidase

HRPTEC human renal proximal tubular epithelial cells

HSP90ab heat-shock protein 90ab1

IGF1R insulin-like growth factor 1 receptor

IP immunoprecipitation

IRES internal ribosome entry site

KD knockdown m7GTP 7-methyguanosine 5’ triphosphate

Met-tRNAi methionyl initiator transfer RNA mRNA messenger RNA mTOR mammalian target of rapamycin

N2 nitrogen

NaCl sodium chloride

NLS nuclear localization signal

NP-40 nonyl phenoxypolyethoxylethanol

NTAD amino-terminal activation domain

O2 oxygen

ODD oxygen degredation domain

P13K/Akt phosphoinositide-3-kinase

P4H-TD transmembrane-prolyl-4-hydroxylase

PABP poly-A binding protein

PAR-CLIP photoactivatable ribonucleoside-enhanced crosslinking and IP

PAS A per-arnt-sim A

PAS B per-arnt-sim B

PBS phosphate buffered saline

PBS-t PBS-tween 20

PCR polymerase chain reaction

PDGF platelet derived growth factor

PDGFRA platelet derived growth factor receptor alpha

PEI polyethylenimine

PHD prolyl hydroxylase domain

PIC pre-initiation complex

PKB/AKT protein kinase B poly(A) polyadenyl

PTEN phosphatase and tensin homolog

PVDF polyvinylidene fluoride pVHL von Hippel-Lindau protein

xii

qPCR quantitative PCR

RBM4 RNA-binding motif protein 4

RecA recombinase A rHRE RNA hypoxic response element

RNA ribonucleic acid

RNP ribonucleoprotein rRNA ribosomal RNA

RT-PCR reverse-transcription PCR

SDS sodium dodecyl sulfate

SDS-PAGE SDS-polyacrylamide gel electrophoresis

SEM standard error of the mean shRNA short-hairpin RNA

TBS tris buffered saline

TCA trichloroacetic acid

TSC1/TSC2 tuberous sclerosis protein 1/2

UTR untranslated region

VEGF vascular endothelial growth factor

xiii Chapter 1. Introduction

1.1 Hypoxia in Development and Disease

The requirement for oxygen is a fundamental aspect of survival for aerobic organisms in all domains of life. Mammals have evolved complex circulatory, respiratory, and neuroendocrine systems to satisfy the need for molecular oxygen as the primary electron acceptor in oxidative phosphorylation, which supplies energy in the form of ATP1–3. Environmental concentrations of oxygen, however, are subject to change; therefore, each nucleated cell within the human body is capable of sensing and responding to fluctuations in oxygen availability in order to maintain homeostasis4. The ability of a cell to acclimate to low oxygen tension (hypoxia), which usually arises due to an imbalance in supply and demand, is essential from the earliest stages of life5,6.

This is because a significant proportion of mammalian development occurs in a hypoxic environment. Prior to the establishment of uteroplacental circulation, embryonic cells receive only as much as 2% O2, and following oxygenation by maternal blood, the embryo still contains discrete regions of hypoxia7,8. This form of physiological hypoxia actually helps govern the process of development and numerous studies have emphasized the role of differing oxygen gradients in cell fate determination, angiogenesis, placentation, cardiogenesis, bone formation, and adipogenesis6,9–13. However, following birth, hypoxia is rarely experienced by the healthy adult, and rather is associated more frequently with pathogenesis2,14. Aside from high-altitude adaption, hypoxia encountered in adulthood is usually a consequence of atherosclerosis, resulting in ischemic diseases such as myocardial infarction and stroke1.

Hypoxia is also a common feature of the tumor microenvironment, and plays a key role in both the development and progression of the malignant state1,15. The onset of hypoxia within a tumor usually begins early on as rapidly dividing cells quickly outgrow the ten cell-layer

diffusion limit of oxygen16,17. As the concentration of oxygen begins to decrease, cells respond by excreting pro-angiogenic factors such as vascular endothelial growth factor (VEGF) and platelet derived growth factors (PDGF), thereby inducing local angiogenesis to re-vascularize the oxygen-depleted tissue18. Normally, the majority of the vasculature is established in a highly regulated process during embryogenesis, and remains largely quiescent throughout adulthood18.

Therefore, aside from transient physiological processes such as female reproductive cycling and wound healing, the requirement for new vessel growth is low18. The activation of angiogenesis initiated by hypoxic tumor cells thus results in the development of structurally and functionally abnormal vessels that are characterized by excessive or convoluted branching, distorted and enlarged passageways, and erratic blood flow3,18. This aberrant vessel growth results in the formation of highly vascularized tumors that still contain regions of severe hypoxia3. These regions help to select for highly aggressive cells that have undergone significant adaptive and genetic alterations, conferring on them the ability to survive in oxygen deprived microenvironments16,19. For this reason, tumors that display a high degree of hypoxia often correlate with a more malignant and invasive phenotype, and generally do not respond well to therapy or surgical interventions20.

1.2 The HIF Program

A large part of the physiological response to hypoxia is mediated by hypoxia-inducible transcription factors known as HIFs. HIFs activate the transcription of numerous genes, including those involved in metabolism and erythropoiesis, in order to simultaneously reduce the activity of oxygen expensive processes and promote the increased uptake of oxygen7. The activity of HIFs is also co-opted by cells during pathological bouts of hypoxia, such as during the

oxygen deprivation resulting from tumorigenesis; however, the response in these circumstances is severely dysregulated16.

Functionally active hypoxia-inducible transcription factors are heterodimeric complexes that consist of an oxygen labile HIF-α subunit and a constitutively expressed nuclear HIF-β subunit1. HIF-α subunits are classified as part of the basic helix-loop-helix/Per-Arnt-Sim

(bHLH/PAS) family of transcription factors, which are comprised of a bHLH DNA-binding region adjacent to degenerate PAS A and PAS B dimerization domains21. Currently, there are three known genes encoding for α subunits in humans: HIF-1α, HIF-2α (EPAS1), and HIF-3α

(IPAS)1,22,23. Of the three homologs, HIF-1α and HIF-2α have been the most extensively characterized to date, as the contribution of HIF-3α to hypoxic transcription is somewhat ambiguous due to the proteins lack of a C-terminal activation domain (CTAD)24. HIF-1α and

HIF-2α on the other hand, which share approximately 48% protein sequence similarity, are integral to the cellular response to hypoxia, and contribute significantly to both embryonic development and the cancer phenotype (Figure 1)1,23. While early reports of the two proteins emphasized their redundancy, the ubiquitous expression pattern of HIF-1α in contrast to the more localized expression of HIF-2α to distinct cell types, such as those of myeloid lineage, hepatocytes, and endothelial cells, suggested that the proteins might in fact support unique roles23,25,26. Indeed it has since been shown that HIF-1α and HIF-2α are capable of regulating both overlapping and distinct target genes despite their similarities27,28. This finding is also supported by knockout studies in mice, where HIF-1α -/- embryos display significant cardiovascular and neural defects resulting in death around embryonic day (E) 10.5, whereas

HIF-2α-null embryos suffer from vascular and lung defects resulting in death around E13.5, but occasionally survive until birth29–31.

In order to regulate the expression of downstream genes, HIF-1α and HIF-2α (herein referred to as HIF-α) must first be stabilized in a hypoxic-dependent fashion. Regulation of HIF-

α subunits largely occurs at the post-translational level to allow for rapid accumulation in response to sudden decreases in oxygen availability. Under ambient oxygen concentrations, or what is referred to as normoxia, HIF-α subunits are targeted for 26S proteasomal degradation by the von Hippel-Lindau (pVHL) tumor suppressor protein24,32,33. This occurs through a mechanism that involves the hydroxylation of one or both of two highly conserved proline residues within HIF-α’s oxygen degradation domain (ODD): P402 and P564 in HIF-1α, and

P405 and P531 in HIF-2α24 (Figure 1). Hydroxylation of these residues in the presence of oxygen is catalyzed by a family of oxygen-sensing prolyl-hydroxylase domain (PHD) proteins, and an endoplasmic reticulum-resident non-PHD family member known as transmembrane- prolyl-4-hydroxylase (P4H-TD)2434. Once hydroxylated, these residues serve as a recognition site for the binding of pVHL, a component of the VCB-Cul2 E3 ubiquitin ligase complex, which stimulates the polyubiquitination and subsequent destruction of the HIF-α subunits24,32,35 (Figure

2). Additional forms of hydroxylation within the CTAD of HIF-α are carried out by asparaginyl hydroxylase factor-inhibiting HIF (FIH) under normoxia, and do not lead to protein degradation.

Instead, this form of hydroxylation inhibits the recruitment of co-activators CBP/p300 to the promoters of target genes, thereby suppressing transcription32,34. While the utility of FIH may not be readily apparent given that hydroxylation by PHDs is most dominant, it has been shown that

FIH enzyme activity is more robust at lower oxygen concentrations compared to that of the

PHDs24,34. This suggests that while PHDs are responsible for the vast majority of HIF-α regulation under ambient oxygen conditions, FIH serves to fine-tune the activity of the HIFs in the intermediate conditions between normoxia and hypoxia. Additionally, FIH displays increased

substrate affinity for HIF-1α over HIF-2α, further implicating a more intricate role for the protein in the hypoxic response34.

Figure 1. Alignment of HIF-1α and HIF-2α structural domains. Comparison between the HIF isoforms HIF-1α and HIF-2α reveals the greatest regions of homology are contained within the basic helix-loop-helix (bHLH) and per-arnt-sin (PAS) A and B domains (99%, 91% and 96% respectively). Both proteins also share a less conserved oxygen-dependent degradation domain (ODD) and a C-terminal activation domain (C-TAD). Hydroxylation at one or both of two highly conserved proline residues within the ODD results in proteasomal degradation. Figure adapted from 22.

Regardless of their differing functions, PHDs, P4H-TD, and FIH all belong to the Fe(II) and 2-oxogluterate-dependent oxygenase superfamily, and therefore require molecular oxygen as a substrate for hydroxylation24,32. Consequently, the enzymatic hydroxylase activity of these proteins becomes impaired as the availability of oxygen decreases. This results in the rapid stabilization of HIF-α subunits under hypoxic conditions. At this point, the nuclear localization signal (NLS) present in the CTAD of HIF-α directs the constitutive translocation of the subunits from the cytoplasm into the nucleus, where they are free to heterodimerize with HIF-β via their

PAS domains25,36. These newly formed HIF-α/HIF-β dimers can recognize and bind to specialized hypoxic response elements (HREs) present in the promoters or enhancers of target

genes to stimulate their transcription21,23,37,38 (Figure 2).

Figure 2. Overview of the regulation of HIF-α in response to changing oxygen concentrations. Under normal oxygen conditions, HIF-α subunits are hydroxylated by a family of prolyl-hydroxylase domain (PHD) proteins at two conserved proline residues. This hydroxylation catalyzes Von Hippel-Lindau (pVHL)-mediated polyubiquitination, targeting the subunits for proteasomal degradation. When oxygen levels significantly decrease, inhibition of PHDs permits HIF-α stabilization and nuclear translocation. Once in the nucleus, HIF-α subunits heterodimerize with HIF-β to form functionally active transcription factors. HIF-α/HIF-β dimers initiate the transcription of hundreds of target genes containing hypoxic response elements (HREs) within their promoters.

Intriguingly, genome-wide chromatin immunoprecipitation studies have shown that both

HIF-α subunits share a common G/ACGTG consensus binding sequence, and consequently

interact with the majority of the same regions in the genome37. Given that HIF-1α and HIF-2α share 83% identity in their DNA binding regions, this observation is perhaps not surprising, but instead introduces into question what determines transcriptional preference for one HIF-α subunit over the other27? While there are multiple factors that contribute to α-subunit specificity, the proteins’ less conserved N-terminal activation domains (NTAD) appear to interact differently with transcriptional cofactors, resulting in distinct activation of target genes27. Differing temporal control of HIF-1α versus HIF-2α in response to hypoxia also likely plays an important role in hypoxic gene activation. It is generally well accepted that HIF-1α protein expression peaks early in response to more severe hypoxic conditions (0-2% O2); but begins to decrease over time, while HIF-2α expression remains relatively consistent throughout chronic bouts of

19,39 more moderate oxygen-deprivation (2-5% O2) . The reduction in HIF-1α levels during extended periods of hypoxia occurs as a result of several negative-regulatory mechanisms, and would suggest that a more dominant HIF-2α-regulated response is favourable in this context.

One such mechanism results in the degradation of HIF-1α mRNA by means of the production of a natural antisense transcript (aHIF) that specifically targets the 3’ untranslated region (UTR) of

HIF-1α mRNA, culminating in the exposure of an AU-rich element39,40. HIF-2α appears to lack this stretch of AU bases, and generally shares low 3’ UTR homology with HIF-1α, likely explaining the feature of HIF-2α mRNA to evade aHIF-mediated degradation at later time points39,40. Additionally, the interaction between HIF-1α and an atypical E3 ubiquitin ligase known as HIF-associated factor (HAF) has also recently been shown to induce polyubiquitination and proteasomal destruction of the protein, while HAF binding to HIF-2α promotes transcriptional transactivation33,41. Undoubtedly, further investigation of the unique

features of HIF-1α and HIF-2α and their divergent roles in the cell will reveal novel insights into the dynamic response to hypoxia.

Activation of the HIF transcriptional program in cancer occurs not only as a result of intratumoral hypoxia, but also through mutations in oncogenes and tumor suppressors that promote the stability of HIF-α in the absence of hypoxia as well2,42. Perhaps the most common of these mutations found in cancer is the loss of pVHL activity, but they can also manifest as inactivating or activating mutations in key signaling proteins such as phosphatase and tensin homolog (PTEN) and protein kinase B (PKB/AKT), respectively24. In these circumstances, the constitutive activity of HIFs promotes the synthesis of critical hypoxic mRNAs such as VEGF, epidermal growth factor receptor (EGFR), glucose transporter 1 (GLUT-1), and PDGF receptor alpha (PDGFRA)1,43. While expression of these genes during periods of physiological hypoxia aids in cell survival, cancer cells exploit their activity to drive vascularization, proliferation, invasion, migration, immune evasion, metabolic reprogramming and the overall aggressiveness of the disease1–3. This is possible due to the capacity of many cancer cells to engage in intensive protein synthesis allowing for the consistent translation of tumorigenic mRNAs, which may also serve as an explanation as to why tumors expressing high levels of HIFs are generally more difficult to treat17,24.

1.3 Cap-Dependent Translation Initiation and eIF4F

The central dogma of molecular biology illustrates the importance of mRNA translation as part of a fundamental process of life that allows the information encoded in genes to dictate the phenotype of an organism44. The process of translation generally consists of three stages, initiation, elongation, and termination, and requires the coordinated interaction of a number of proteins, RNAs, and amino acids45,46. While each step of translation is subject to various forms

of regulation, the greatest amount of control is exerted at the rate-limiting step of initiation46,47.

The most common form of translation initiation in eukaryotes occurs in a cap-dependent fashion, which relies on the specialized m7GTP cap-structure present at the 5’ end of all nuclear- encoded cellular mRNAs46,48,49. In this process, recruitment of the 40S ribosomal subunit to mRNA is dependent on the presence of the heterotrimeric eukaryotic translation initiation factor

4F (eIF4F) at the 5’ cap47,49. This requirement was established over thirty years ago when it was discovered that addition of eIF4F, the composition of which was unknown at the time, was necessary for the translation of capped globin mRNA in a cell-free system50. This was in addition to the observation that cap-dependent translation of poliovirus-infected lysates could be restored when complemented with eIF4F in vitro50. It is now known that encompassed within eIF4F is the cap binding protein eIF4E, the ATP-dependent RNA helicase eIF4A, and the large scaffold protein eIF4G47,49. While there are multiple theories that exist to explain how eIF4F recruitment to the 5’ cap occurs in vivo, there is strong evidence supporting a model where eIF4F is pre- formed prior to its interaction with mRNA, opposed to the sequential binding of each individual protein. This is because interaction with eIF4G as a constituent of eIF4F has been shown to induce conformational changes in both eIF4E and eIF4A that increases the intrinsic activities of both proteins51,52. For example, eIF4E has approximately 20-fold higher affinity for the cap as a component of eIF4F than as an individual subunit, and eIF4A displays increased RNA binding and ATP-dependent helicase activity when bound by eIF4G47,51–54.

Regardless of the mode of recruitment, once the eIF4F complex is tethered to the 5’ end of mRNA by means of eIF4E, eIF4A begins to denature local RNA secondary structure in the 5’

UTR upstream of the 40S ribosome46,47,55. eIF4E also plays an important role in maintaining eIF4A’s localization at the cap, as the nonprocessive nature of the enzyme ultimately leads to

mRNA dissociation54,56. eIF4A is further stimulated by RNA-binding proteins eIF4B and/or eIF4H, which possess no catalytic function of their own, but may aid in stabilizing single- stranded RNA following duplex separation56. This activity is required to allow the 43S pre- initiation complex (PIC), which is composed of a 40S subunit bound by initiation factors eIF1, eIF1A, eIF3, eIF5 and a charged eIF2-GTP-Met-tRNAi ternary complex, to bind the 5’-proximal region and begin scanning the mRNA46,54. The interaction between the 43S PIC and the mRNA- bound initiation complex is mediated by eIF4G, which makes critical contacts with the 40S- associated eIF3 multi-subunit complex54. Once the 48S PIC encounters an initiation codon in a favourable context, it is joined by the 60S subunit and elongation commences46.

Subsequent rounds of cap-dependent translation initiation are aided by eIF4G’s interaction with the RNA-binding protein, poly(A)-binding protein (PABP). Multiple copies of

PABP, as the name implies, can be found along the 3’ poly(A) tract of an mRNA57. By interacting with the 5’-eIF4F complex, PABP promotes mRNA circularization and as a result translation54. This occurs by several different means; the formation of a closed-loop structure encourages ribosome recycling, but has also been shown to promote recruitment of both 40S and

60S ribosomal subunits de novo by localizing eIF4G in the vicinity of the cap54,57. The interaction between eIF4G and PABP also ensures that the initiation complex remains associated with a particular mRNA even if attachments at the 5’-end are transiently lost following a round of translation54.

The entire process of cap-dependent translation accounts for up to 70% of a cell’s total

ATP consumption, making it the most energy-expensive biosynthetic pathway58. Consequently, regulation of translation initiation is exerted to account for fluctuations in energy availability, and is a process that is tightly regulated in order to ensure cell survival.

1.4 mTOR and Hypoxic Regulation of eIF4F

In response to the energy depletion that coincides with a number of cellular stressors, multiple signaling pathways converge upon the serine/threonine kinase mammalian target of rapamycin

(mTOR), in order to modulate their effects on the rate of translation59,60. Hypoxia is a particularly potent inhibitor of protein synthesis that has been observed to down-regulate mTOR signaling even prior to the point of detectable ATP depletion, as well as into the stages of chronic oxygen deprivation61–64. This is possible due to the diversity of hypoxic effector proteins functioning both up and downstream of the mTOR axis point. Perhaps the most extensively studied regulators of translation initiation during hypoxia downstream of mTOR are the family of eIF4E binding proteins (4E-BPs)49. Under normal oxygen tension, 4E-BPs are maintained in their inactive conformation due to upstream signaling and phosphorylation by mTOR65. This permits the 4E-BP substrate, eIF4E, to remain unbound by its inhibitor and therefore remain functionally active as part of the eIF4F translation initiation complex65,66. However, in response to oxygen deprivation mTOR activity is inhibited, resulting in hypophophorylation and activation of 4E-BPs65,66. These inhibitory proteins can then bind eIF4E, which precludes the ability of eIF4G from interacting with eIF4E, and consequently prevents initiation of cap- dependent translation49,66 (Figure 3).

Multiple interconnected and independent signaling pathways that lead to the inhibition of mTOR in response to hypoxia have been elucidated in recent years, while some remain to be further investigated. HIF-1α-dependent hypoxic inhibition of mTOR is frequently reported in the literature, and occurs as a result of the transcriptional upregulation of the REDD1 gene by the

HIF-1α transcription factor in response to low oxygen availability64,65. As a result, the REDD1 protein competitively binds to and sequesters the small protein inhibitor 14-3-3, which is

normally bound to the tuberous sclerosis 2 (TSC2) component of the TSC1/TSC2 tumor suppressor complex64,65. Once relieved of inhibition, TSC1/TSC2 GTPase activating protein

(GAP) activity is restored and the complex can act upon the small G-protein Rheb, an upstream activator of mTOR, by converting the protein to its inactive GDP-bound state64.

eIF4F-dependent translation eIF4FH-dependent translation

Normal [O2] Low [O2]

HIF-2α HIF-2α

RBM4

7 m -G 7 AAAAA m -G rHRE AAAAA

P P P P P

4E-BP 4E-BP 4E

AAAAA rHRE AAAAA PABP RBM4 Translation

7 eIF4A m -G4E HIF-2α eIF4G Translation

eIF4A 4E2m7-G eIF4G3

Figure 3. Proposed model of the eukaryotic translation initiation complex mediating cap- dependent translation during normoxic and hypoxic conditions. Select mRNA containing RNA hypoxic response elements (rHREs) in their 3’ UTRs are destined for translation by the eIF4FH complex, which only functions in hypoxia due to the continuous degradation of HIF-2α at normal oxygen conditions. These transcripts constitutively interact with the RNA-binding protein RBM4, which interacts with the hypoxic inducible factor HIF-2α. RBM4-bound HIF-2α aids in the recruitment of the cap-binding protein eIF4E2, which together with other members of the eIF4FH complex promotes the selective translation of rHRE-containing transcripts. mRNAs lacking rHREs are unable to undergo selective hypoxic translation by eIF4FH. Rather, these transcripts are translated by the eIF4F-complex, which only functions in normoxic conditions due to inhibition of the cap binding protein, eIF4E, under hypoxia.

While this model has been generally well accepted, some argue that a decrease in mTOR activity observed after only approximately fifteen minutes of moderate hypoxia cannot be attributed to HIF-1α activity due to the delay in transcriptional activation and translation of

REDD160,63. This has led to the identification of HIF-1α-independent signaling pathways that

instead rely on the activity of the serine/threonine AMP-activated protein kinase (AMPK)63.

AMPK activity in humans is stimulated by phosphorylation of Thr-172, resulting from increases in cellular AMP/ATP ratios initiated by numerous metabolic stressors, including hypoxia62,63,65.

AMPK induces its downstream effects on mTOR via TSC2/Rheb signaling, likewise to HIF-1α- dependent pathways, by activating the GAP activity of the TSC1/TSC2 complex; however, this is accomplished by direct phosphorylation of TSC2 at Thr-122763,66. It has also been demonstrated that AMPK activation can occur independently of serum starvation, suggesting that moderate or short-term hypoxia alone is sufficient to trigger the inhibition of mTOR downstream of AMPK, likely due to energy starvation63. This model supports a bi-phasic response of mTOR regulated hypoxic translation initiation inhibition, where AMPK mediates acute and moderate hypoxic responses, and HIF-1α chronic responses.

However, herein lies a paradox; despite the fact that hypoxia impedes the cells primary mode of translation, stress response proteins, including those that contribute to the cancer phenotype, continue to be synthesized17. This suggests that other mechanisms are likely responsible for the maintenance of stress response protein synthesis during oxygen deprivation.

1.5 IRES-Mediated Translation

Once it was concluded that classical cap-dependent translation activity could be reduced by up to

70% during hypoxia, researchers began to investigate other mechanisms that could potentially be responsible for the maintenance of stress response protein synthesis during oxygen deprivation67.

The idea that these hypoxic transcripts were being translated via initiation at an internal ribosome entry site (IRES) quickly became the favoured hypothesis among the scientific community.

IRES translation initiation was first identified over two decades ago in picornaviruses, when Jang et al. (1988) were attempting to determine how the family of viruses maintained

efficient translation of their naturally uncapped mRNAs68,69. While most IRES have been found to vary between organisms, they generally are defined as mRNA 5’ UTR sequences and secondary structures that directly recruit the 40S ribosome to the vicinity of the start codon in the absence of eIF4F67,70. Further investigation revealed that the use of IRES elements is presumably an adaptive mechanism utilized by picornaviruses to promote selective translation of viral mRNAs during the process of infection71. This is because host protein synthesis is frequently inhibited during viral infection, either by cleavage of eIF4G by viral proteases, or enhancement of 4E-BPs, resulting in the failure of proper eIF4F complex formation71.

While the majority of protein-coding mRNAs in eukaryotes are translated by cap- dependent mechanisms, it has since been discovered that 3-5% of cellular mRNAs, including

VEGF and HIF-1α, can be translated by IRES-mediated initiation in normal conditions67. This observation led to the development of the hypothesis that IRES-mediated translation was most likely the means by which proteins could be continually synthesized during hypoxic stress67. It was later observed however that the basal levels of IRES-mediated translation of critical hypoxic mRNAs, such as VEGF and HIF-1α, did not increase during oxygen deprivation when the requirement for these proteins is significantly increased67. This suggested that IRES mechanisms were in fact unlikely to perform critical roles in the synthesis of hypoxic response proteins67.

1.6 Hypoxic Cap-Dependent Translation and eIF4FH

Until recently, the method by which cells continued to engage in translation under hypoxic conditions perplexed the scientific community. However, in 2012, Uniacke and colleagues elucidated a function for the eIF4E homolog, eIF4E2, and managed to shed some light on the long disputed topic72. It was found that cells can initiate a molecular switch in the translation apparatus under hypoxia, whereby eIF4E2, which shares 30% amino acid identity

with eIF4E, is utilized as the cap-binding protein to initiate translation in the absence of eIF4E72,73. The hypoxic specificity of eIF4E2-mediated cap-dependent translation is attributed to the requirement for HIF-2α as part of initiation complex, which is exemplified by the suppression of hypoxic translation in HIF-2α but not HIF-1α depleted cell lines72. The association between HIF-2α and target transcripts was mapped using photoactivatable ribonucleoside-enhanced crosslinking and immunoprecipitation (PAR-CLIP) to the 3’ UTR, concentrating in regions that contain CGG trinucleotides, referred to as RNA hypoxic response elements (rHREs)72. Since HIF-2α does not contain a classical RNA binding domain, this interaction requires mediation by RNA binding motif protein 4 (RBM4), which displays no dependence on oxygen for rHRE association72. Therefore, it is hypothesized that upon hypoxic induction, stabilized HIF-2α subunits associate with a pool of critical hypoxic mRNAs containing RMB4-bound rHREs in their 3’ UTRs72. HIF-2α then recruits eIF4E2 to the 5’-cap of these transcripts through direct protein-protein interactions to promote cap-dependent translation, which is further augmented by tethering of the 3’ and 5’ ends72 (Figure 3). Some additional members of this hypoxic cap-dependent translation initiation complex, referred to as eIF4FH, have recently been identified. These include the prototypical RNA helicase, eIF4A, and the eIF4G homolog eIF4G3, which presumably assumes the role of protein scaffold for the complex as eIF4E2 does not interact with eIF4G72–74. This novel mechanism has been shown to initiate cap-dependent translation of potentially hundreds of mRNAs that contribute to hypoxic survival, many of which are upregulated by the HIF transcriptional program72.

While initial reports describing the capacity for eIF4E2 to bind the cap were published as early as 1998, research on the topic was slow to progress73. This was likely due to the fact that the contribution of eIF4E2 to cap-dependent translation under normal conditions was presumed

to be minimal, as its cap-binding affinity is estimated to be 30-100 fold less than that of eIF4E75.

Sequence alignment and crystallographic studies have shown that this feature is likely attributed to two amino acid alterations in eIF4E2 (R112−>K and K162−>I), which correspond to evolutionarily conserved positively-charged residues in eIF4E that form hydrogen bonds with the phosphate groups of m7GTP73,75. This would imply that even though eIF4E2 possesses the ability to bind the cap, a situation in which this interaction would not be outcompeted by eIF4E would be unlikely to occur. This assumption was further validated by the fact that there is approximately five to ten-fold more eIF4E than eIF4E2 in human cells73.

Unlike eIF4E however, eIF4E2 is able to evade total hypoxic inhibition by 4EBPs.

Characterization of the interaction between eIF4E2 and 4EBPs revealed that 4EBPs use the same binding cleft to interact with both eIF4E and eIF4E2; however, a single point mutation (Y54A or

L59A) within this region is sufficient to abolish the association with eIF4E2 but not eIF4E76.

Additionally, standard IP of 4EBP failed to capture eIF4E2, but can readily detect eIF4E interactions under hypoxic conditions73,76. These data suggest that 4EBPs have an overall reduced affinity for eIF4E2, allowing eIF4E2 to function as a potent translation initiator under low oxygen conditions when eIF4E competition is minimized72,76.

While there is currently some understanding of how this non-canonical translation initiation complex is functioning, little is known about the entirety of its composition and, aside from its hypoxic induction, how it is regulated.

1.7 Regulation of eIF4FH by DDX28

Research carried out using normal human renal proximal tubular epithelial cells (HRPTEC) and a number of malignant cancer cell lines revealed that the shift from the use of eIF4E to eIF4E2 under hypoxia is a survival mechanism used by both healthy cells and tumorigenic cells17,77. This

suggests that eIF4E2-mediated translation may have evolved as a means to combat periods of normal physiological hypoxia, and is misappropriated by cancer cells thriving in the hypoxic region of a tumor core17. However, since healthy adults rarely experience hypoxia, targeting eIF4E2 as an anticancer therapeutic would provide a unique opportunity to selectively inhibit the translational apparatus of malignant cells2,17. This would serve as a valuable treatment method as many current cancer therapies, including those that take advantage of tumor overexpression of eIF4E, still harbour the potential to damage healthy tissues and as a result cause severe side effects17,47. Early research has indeed corroborated this hypothesis, as depletion of eIF4E2 from both developing and established tumors in mice significantly decreased tumorigenic potential and tumor mass, respectively, without affecting normoxic healthy cells17. The elucidation of novel components of eIF4FH, and regulatory mechanisms that govern its activity could therefore reveal additional methods by which protein synthesis in cancer cells could be selectively inhibited.

IP of recombinant FLAG-eIF4E2 from human U-87 MG glioblastoma cells exposed to normoxic and hypoxic conditions followed by mass-spectrometry analysis has lead to the identification of a candidate interactor and potential regulator of eIF4FH (Uniacke, unpublished)

(Figure 4).

1% O2 21% O2

IP Ctrl eIF4E2 Ctrl eIF4E2

1A 1B 2A 2B 3A 3B Band from Protein Coomassie 4A 4B gel in Eukaryotic translation initiation factor 4G3 1A, 1B 5A 5B Hypoxia Inducible Factor 2-alpha 2A Splicing Factor Proline.Glutamine-Rich 2A Grb10-interacting GYF protein 2 2B Eukaryotic translation initiation factor 4B 3A, 3B DEAD Box Polypeptide 28 4A, 4B 6A 6B Eukaryotic translation initiation factor 3L 4A Non-POU Domain Containing Octamer-Binding 4A Paraspeckle Component 1 4A Serine/threonine-protein kinase 38 4A RNA Binding Motif Protein 4 5A Eukaryotic translation initiation factor 4A 5A, 5B Eukaryotic translation initiation factor 4E2 6A, 6B

Figure 4. Immunoprecipitation of FLAG-eIF4E2 from normoxic and hypoxic U-87 MG cell lysates. A) SDS-PAGE gel stained with coomassie-blue for visualization of FLAG-eIF4E2 IP from U-87 MG cells cultured in 1% O2 (hypoxia) and 21% O2 (normoxia), using cells transfected with an empty FLAG vector as a control (Ctrl). B) Bands (1A-6A and 1B-6B) were excised and identified by liquid chromatography-mass spectrometry (Uniacke, unpublished).

In this assay, eIF4E2 was found to associate with the DEAD box protein, DDX28. DDX28 was first cloned and characterized in 2001, when it was isolated from a human testis cDNA library78.

Sequence alignment revealed that the protein belongs to the largest family of RNA helicases, known as the DEAD box proteins, of which eIF4A is the archetypal member79,80. DEAD box proteins are defined by the presence of two core recombinase A (RecA)-like helicase domains, which encompass at least twelve highly conserved sequence motifs that function in RNA binding and ATP-hydrolysis81 (Figure 5). The family derives its name from motif II, also known as the

Walker-B motif, which contains the amino acid sequence Asp-Glu-Ala-Asp (DEAD)79. Despite the high conservation found in their helicase domains, these proteins have found unique roles in almost every aspect of cellular biology concerning RNA80. This is possible due to the

appreciable variation in both amino and carboxy-terminal domains of these proteins, which can participate in interactions with an array of unique protein and/or RNA targets81.

Figure 5. Schematic diagram of the conserved motifs of the DEAD-box protein superfamily Proteins within the DEAD-box superfamily are characterized by a large helicase core composed of two recombinase A (Rec-A)-like domains (domains 1 and 2). Contained within these domains are at least twelve conserved motifs that function in ATP binding and hydrolysis (red), RNA binding (blue), or both RNA and ATP binding and communication (yellow). Figure adapted from 81.

There are currently 37 known members of the DEAD box family in humans, many of which still remain to be biochemically characterized81. This is due to the observation that many of these helicases are capable of performing multifaceted roles within the cell, determined by differences in localization and/or ribonucleoprotein (RNP)-complex components81. In accordance with this, the currently known functions of DDX28 are limited, and mainly focus on the proteins contribution to mitoribosome large subunit biogenesis by means of its interaction with 16S rRNA82,83. However, DDX28’s ability to interact with eIF4E2 indicates that it may also be capable of functioning as part of the eIF4FH complex, or regulating its activity through eIF4E2. In fact, substantial evidence exists to suggest that DEAD-box proteins perform crucial regulatory roles in multiple forms of translation initiation. DDX3 for example has been implicated in the repression of cap-dependent translation by interacting with eIF4E through a conserved eIF4E-binding motif, enhancing its affinity for the cap, but preventing eIF4G

binding84. Intriguingly, the reduction in cap-dependent translation observed during DDX3 overexpression is accompanied by an increase in IRES-mediated translation of cellular mRNAs84. DDX3 has also been described as an integral component of the initiation machinery during the cap-dependent translation of a subset of transcripts containing complex 5’ UTRs85. It is interesting to note that while these reports appear to be in conflict with one another, translational activation by DDX3, in contrast to its repressional activity, only depends on interactions with eIF4G and PABP, and not eIF4E85. This emphasizes how the dynamic interplay between DEAD box helicases and auxiliary factors can alter the role of these proteins even within a single biological process, such as translation.

Another DEAD box protein, DDX6, through an unknown mechanism, specifically inhibits the IRES-mediated translation of VEGF mRNA under normoxic conditions by binding to its 5’ UTR86. Under hypoxic conditions however, a natural decrease in total DDX6 protein levels leads to VEGF derepression and enhanced IRES translation86. In light of the observation that

DDX28 also appears to be down-regulated under 1% O2, it is tempting to speculate that perhaps

DDX28 functions in a method akin to both DDX3 and DDX6, wherein its binding to eIF4E2 serves to negatively regulate the activity of eIF4FH under normoxia, but relieves its inhibition under hypoxia.

Although recent research has revealed that cells can maintain cap-dependent translation of select mRNAs during hypoxic stress by exploiting the eIF4FH complex, the mechanisms that govern this activity are still largely unknown. By investigating DDX28 and the roles it may perform in controlling the activity of eIF4FH, we stand to gain valuable insights into a number of biological phenomena that are characterized by low oxygen availability, including embryogenesis and cancer.

1.8 Hypothesis and Research Objectives

Given the preliminary evidence provided by the Uniacke lab regarding the identification of novel eIF4E2 interactors, as reviewed above, and the current understanding of the contribution of

DEAD box proteins to translational regulation, we hypothesize that the interaction between

DDX28 and eIF4E2 curbs the activity of eIF4FH-mediated translation initiation. In order to test this hypothesis, three main research objectives were addressed.

The first objective of this investigation was to characterize the interaction between eIF4E2 and DDX28, as validation of this interaction is integral to the analysis of the potential effect(s) DDX28 may have on eIF4FH. This was accomplished through the employment of a number of IP reactions of either DDX28 itself, or core components of the eIF4FH machinery.

The second objective of this research was to determine what effect(s) DDX28 has on the involvement of eIF4E2 in translation initiation under 21% and 1% O2 conditions. Given that the primary role of eIF4E2 as part of the eIF4FH complex is to directly recognize and bind the 5’ cap, we tested how shRNA-mediated depletion of DDX28 from human U-87 MG cells affects the affinity of eIF4E2 for m7GTP-conjugated sepharose resin (Figure 6). Given that cap-binding proteins have been observed to interact with the cap in a translation-inhibitory fashion, we further addressed the effect of DDX28 knockdown on eIF4E2-mediated translation initiation by performing polysome profiling experiments to monitor the association of eIF4E2 with translationally active ribosomal complexes (Figure 7).

A) B)

D) C)

Figure 6. Standard workflow of a cap-affinity assay. A) Cells are lysed with a non-denaturing buffer. B) Lysates are pre-cleared with “blank” agarose beads, supernatants are collected by centrifugation. C) The pre-cleared lysates are incubated with m7GTP-conjugated agarose beads to capture cap-binding proteins. D) The m7GTP-agarose beads are collected by centrifugation and the supernatant is discarded. The beads are washed to remove any contaminants and captured proteins are eluted by denaturation. E) The eluates are analyzed by SDS-PAGE and Western blotting.

The third and final objective we addressed was how stable shRNA-based depletion of

DDX28 affected U-87 MG cell viability and proliferation under 21% and 1% O2 conditions.

Since the overexpression of bona fide eIF4FH target transcripts, such as the tyrosine receptor kinase EGFR, have been associated with the acquisition of cancer traits such as sustained proliferation and growth, we reasoned that deregulation of eIF4E2-mediated translation may have consequences on cell survival87. To this end, we used crystal violet staining and bromodeoxyuridine (BrdU) labeling to monitor average cell viability over a 72-hour growth period, and cellular proliferation rates, respectively.

80S Add CHX 40S Lyse & load Ultracentrifuge Frac9ona9on Measure UV absorbance Analyze Polysomes Absorbanceat254 nm

Fractions

Figure 7. Isolating polyribosomes using polysome profiling. Following treatment with the translational inhibitor cycloheximide (CHX), cells are lysed and loaded onto a 7-47% sucrose density gradient and ultracentrifuged for 90 minutes at 39, 000 rpm in order to separate the ribosomal constituents. The gradient is then fractionated using the Brandel density gradient fractionation system, which simultaneously detects the UV absorbance of each sample at 254 nm.

Chapter 2. Materials and Methods

2.1 Cell Culture and Reagents

U-87 MG human glioblastoma cells were obtained from the American Type Culture Collection

(ATCC) and cultured as suggested in Dulbecco’s Modified Eagles Medium (DMEM) supplemented with 7.5% fetal bovine serum (FBS), and 1% penicillin streptomycin. In the case of shRNA treated and FLAG-DDX28/control stable lines, cells were grown in the presence of 1 ug/mL puromycin or 200 ug/mL G418 respectively. Normoxic cells were maintained at 37 °C in ambient O2 levels (21%) and 5% CO2 in a humidified incubator. Hypoxia was induced by culturing at 1% O2 and 5% CO2 at 37 °C for 24 h, unless otherwise stated, using an N2-balanced

Whitley H35 Hypoxystation (Don Whitley Scientific Ltd., Shipley, UK).

2.2 Generation of Stable Cell Lines

2.2.1 DDX28 Knockdown Cells

OmicsLink™ shRNA expression clones were used to target the coding sequence of human

DDX28 [HSH014712-3-nU6 sequence 5’-ggtggactacatcttagag-3’ (shRNA clone 3), HSH014712-

3-nU6 sequence 5’-acgctgcaagattacatcc-3’ (shRNA clone 4)]. A non-targeting sequence shRNA

was used as a control (Genecopoeia, Rockville, USA). For each shRNA expression and control clone, a 10 cm culture dish was transfected with 24 µg of plasmid DNA using 40 µL of

Lipofectamine™ 2000, following the manufacturers protocol (Fisher Scientific Ltd., Toronto,

CAD). Selection was initiated 48 h post-transfection using DMEM supplemented with FBS, penicillin streptomycin, and 1 µg/mL puromycin. Stable cell lines, which are identified by their shRNA clone number followed by their colony number (eg. 3.1, 4.9), were generated by picking single colonies after approximately seven days in selection medium.

2.2.2 DDX28-FLAG Cells

U-87 MG cells stably expressing C-terminal 3x FLAG tagged DDX28 were generated by transfecting cells as stated above with the OmicsLink™ pEZ-M14 EX-A3144-M14 expression vector encoding the human DDX28 coding sequence and 3x FLAG peptide (Genecopoeia,

Rockville, USA). Vehicle control cells were transfected with the pcDNA3 vector obtained from the Jones Lab (University of Guelph) encoding the 3x FLAG peptide with no gene insertion.

Selection was initiated 48 h post-transfection using DMEM supplemented with FBS, and 400

µg/mL G418. Stable cell lines were generated by selecting single colonies after approximately seven days in selection medium.

2.3 Western Blot Analysis and Antibodies

Whole cell lysate samples, unless otherwise indicated, were collected by washing cells twice with phosphate-buffered saline (PBS) and lysing with 4% SDS in PBS. Samples were separated by SDS-PAGE following standard techniques and transferred to polyvinylidene difluoride

(PVDF) membrane for 1.5 h at 100V using Towbin transfer buffer containing 10% methanol and

0.01% SDS. Membranes were blocked in 5% (w/v) skim milk prepared in 0.02% PBS-tween

(PBS-T) for 1 hour at room temperature and incubated with primary anti-eIF4E2 (1:1000;

Genetex, GTX82524), anti-DDX28 (1:1000; Abcam, AB70821), anti-eIF4E (1:1000; Cell

Signaling Technology, C46H6) anti-GAPDH (1:30,000; Cell Signaling Technology, D16H11), anti-RPL5 (1:1000; Abcam, ab137617), anti-HIF-2α (1:500; Novus Biologicals, NB100-122), or anti-FLAG (1:1000; Sigma, F1804) antibodies overnight at 4 °C. Membranes were washed 3 times for 10 min in 0.02% PBS-T and incubated with horseradish peroxidase (HRP) conjugated anti-rabbit (Promega, PR-W4011) or anti-mouse (Promega, PR-W4012) secondary antibody for

1 hour at room temperature. All primary and secondary antibody dilutions were prepared in 5%

(w/v) skim milk in 0.02% PBS-T. After washing 3 times for 10 min in 0.02% PBS-T, membranes were incubated with enhanced chemiluminescent HRP substrate (EMD Millipore,

Etobicoke, CAN) and signals were detected using the ChemiDoc MP imaging system.

2.4 Polysome Profiling

Cells were plated in four 15 cm culture dishes and cultured at their indicated oxygen concentrations for 24 h. Cycloheximide (CHX) was added to each dish of cells at a final concentration of 0.1 mg/mL for 10 min at 37 °C. Cells were washed 2 times with ice-cold PBS containing 0.1 mg/mL CHX and harvested by scraping and centrifugation for 2 min at 400 rpm.

Cells were lysed in 1 mL of RNA lysis buffer [0.3M NaCl, 15mM MgCl2, 15mM Tris-HCl (pH

7.4), 1% Triton X-100, 0.1 mg/mL CHX, 100 units/mL RNase inhibitor] for 10 min on ice, and pre-cleared by centrifugation at 5,000 rpm for 10 min and subsequently 14,000 rpm for 10 min at

4 °C. The A260 of each sample was determined using the NanoDrop™ 2000 spectrophotometer

(Fisher Scientific Ltd., Toronto, CAD). Using the A260 measurement, the lysate volume

corresponding to 1000 µg of RNA was loaded onto a 7-47% sucrose density gradient containing

0.1 mg/mL CHX, and subjected to ultracentrifugation at 39,000 rpm for 90 min at 4 °C in a SW-

41 Ti rotor (Beckman Coulter, Mississauga, CAN). Gradients were fractionated into nine 1 mL samples, while the absorbance of each fraction was simultaneously measured at 245 nm using the Brandel density gradient fractionation system (Brandel, Gaithersburg, USA).

2.4.1 Polysome Protein Analysis

Total protein was isolated from each fraction using trichloroacetic acid (TCA)/acetone precipitation. Briefly, one volume of TCA was added to four volumes of heterogeneous sample and incubated at 4 °C for 10 min to precipitate protein. Protein was collected by centrifugation at

14,000 rpm for 5 min at 4 °C and washed twice with ice-cold acetone. Excess acetone was evaporated at room temperature. Proteins were solubilized in Laemmli sample buffer and equal volumes of sample were analyzed by SDS-PAGE and Western blot. Blotting for RPL5 was used as a marker of protein integrity. The total eIF4E2 or eIF4E signal was quantified by densitometry using Bio-Rad Image Lab software, and the percentage of eIF4E2 or eIF4E present in monosome and polysome fractions relative to the total signal was calculated. All averages depict the mean ± standard error of the mean (SEM) for at least three independent biological replicates. Differences between group means were assessed with a one-way ANOVA and Tukey’s honestly significant difference (HSD) post-hoc test using GraphPad Software.

2.4.2 Polysome RNA Isolation, RT-PCR, and quantitative PCR

Proteinase K was added to each fraction to digest proteins, and total RNA was extracted using standard phenol-chloroform procedure. RNA was precipitated with 3M sodium acetate (pH 5.2)

and 95% ethanol overnight at -20 °C. RNA was collected by centrifugation at 15,000 rpm for 30 min at 4 °C, and washed once with 70% ice-cold ethanol. Excess ethanol was evaporated at room temperature and RNA pellets corresponding to monosomes and polysomes were resuspended in equal volumes of ultrapure water. Equal volumes of total RNA were then reverse transcribed using the high-capacity cDNA reverse transcription kit (Applied Biosystems, Foster City, USA).

Cycling conditions were as follows: 10 min at 25 °C, 120 min at 37 °C, 5 min at 85 °C.

Quantitative PCR (qPCR) reactions were performed using Sso Advanced Universal SYBR Green

Supermix (Bio-Rad, Hercules, USA). Cycling conditions were as follows: initial denaturation for

30 seconds at 95°C, followed by 39 cycles of 15 seconds at 95°C and 30 seconds at 60 °C.

Primer sequences used (5’-3’): EGFR, GGA GAA CTG CCA GAA ACT GAC (forward) and

GGG GTT CAC ATC CAT CTG (reverse); HSP90ab1, TGT CCC TCA TCA TCA ATA CC

(forward) and TCT TTA CCA CTG TCC AAC TT (reverse); RPLP0, AAC ATC TCC CCC

TTC TCC (forward) and CCA GGA AGC GAG AAT GC (reverse); and GAPDH, GTC AAG

GCT GAG AAC GGG A (forward) and CAA ATG AGC CCC AGC CTT C (reverse). Relative fold change in expression was calculated using the ΔΔCT method, and transcript levels were normalized to RPLP0 and GAPDH. All averages depict the mean ± standard error of the mean

(SEM) for at least three independent biological replicates. Differences between group means were assessed with a one-way ANOVA and Tukey’s honestly significant difference (HSD) post- hoc test using GraphPad Software.

2.5 GFP-Trap Immunoprecipitation

Three 10 cm culture dishes at approximately 70% confluence were each transfected with 4 µg of pAdlox-GFP vector DNA encoding a constitutively active variant of HIF-2α as previously

described 72. Briefly, DNA and 20 µg of polyethylenamine (PEI) were diluted in 600 µL of lactate buffered saline (pH 4.0) and incubated at room temperature for 20 min. The DNA/PEI complexes were diluted in 2.4 mL DMEM without FBS or antibiotics and added drop-wise to the cells and incubated at 37 °C and 5% CO2 for 8 h, after which cells were washed with PBS and fresh medium was added. Three 10 cm dishes of cells were mock-transfected in tandem with no

DNA as a negative control. 24 h post-transfection cells were placed in 1% O2 for an additional

24 h. 48 h post-transfection, cells were washed twice with PBS and lysed in 200 µL of ice-cold

IP lysis buffer [10mM Tris-HCl (pH 7.4), 150mM NaCl, 0.5mM EDTA, 0.5% NP-40, 1X protease inhibitor cocktail (New England Biolabs, Whitby, CAN)]. Lysates were collected by centrifugation at 12,000 rpm for 10 min at 4 °C and diluted with 500 µL of dilution buffer

[10mM Tris-HCl (pH 7.4), 150mM NaCl, 0.5mM EDTA, 1 X protease inhibitor cocktail].

Lysates were quantified by BCA protein assay and 500 µg (1 mg/mL) of total protein was incubated with 25 µL of GFP-Trap magnetic micro beads (ChromoTek GmbH, Munich, GER) pre-blocked with 3% BSA and washed as per the manufacturers instructions.

Immunoprecipitation was carried out for 1 hour at 4 °C with agitation. Beads were collected using a magnetic stand, and the supernatant was discarded. The beads were washed 4 times for 5 min at 4 °C with wash buffer [10mM Tris-HCl (pH 7.4), 500mM NaCl, 0.5mM EDTA, 0.1%

NP-40], and proteins were eluted at 95 °C for 10 min in 1 X Laemmli sample buffer. 25 µg of whole cell lysate was used as the input.

2.6 FLAG Immunoprecipitation

Three 10 cm dishes of FLAG-DDX28 and vector-control FLAG stable cells were cultured at their indicated oxygen concentrations for 24 h. For FLAG-eIF4E2 IPs, three 10 cm dishes of U-

87 MG cells were transfected as described above with either the empty pcDNA3 vector obtained from the Jones Lab (University of Guelph), or with the p3XFLAG-CMV-10 vector encoding human eIF4E2, generated in the Uniacke lab (University of Guelph). Briefly, the human eIF4E2 sequence was subcloned from the pFLAG vector, described previously 88, into the p3XFLAG-

CMV-10 vector using the restriction enzymes HindIII and BamHI. Cells were washed twice with

PBS and lysed with 200 µL of ice-cold IP lysis buffer. Lysates were pre-cleared by centrifugation at 12, 000 rpm for 15 min and diluted with 500 µL of dilution buffer. Total protein concentration was quantified by BCA protein assay. For immunoprecipitation, 1 mg (1 mg/mL) of total protein was incubated for 1 hour at 4 °C with 25 µL of anti-FLAG M2 magnetic agarose beads (Sigma, Oakville, CAN), pre-blocked with 3% bovine serum albumin (BSA) made in PBS.

Anti-FLAG beads were separated with the use of a magnetic stand and supernatants were discarded. Beads were washed 4 X with ice-cold TBS (50 mM Tris, 150 mM NaCl), resuspended in 1 X Laemmli sample buffer, and proteins were eluted at 95 °C for 10 min. 25 µg of whole cell lysate was used as the input.

2.7 Analysis of Cap-Binding Proteins

Cells were maintained in two 15 cm culture dishes for 24 h at their indicated oxygen concentrations prior to lysis in 1 mL of ice cold cap lysis buffer [20mM Tris-HCl (pH 8.0),

137mM NaCl, 0.5% sodium deoxycholate, 1% NP-40, 10% glycerol, 2 mM EDTA, 1X protease inhibitor cocktail]. Cell debris was collected by centrifugation at 10,000 x g for 10 min at 4 °C, and total protein concentration of the supernatant was quantified by BCA protein assay. 35 µg of whole cell lysate was kept for input analysis, and 1 mg (1 mg/mL) of protein lysate was pre- cleared with 50 µL of unmodified agarose beads (Jena Bioscience GmbH, Jena, GER) for 10 min

at 4 °C with agitation. The supernatant was collected by centrifugation at 500 x g for 30 sec and transferred to a new tube containing 50 µL of immobilized γ-aminophenyl-m7GTP (Jena

Bioscience GmbH, Jena, GER). Cap-binding proteins were captured by incubation for 2 h at 4

°C with agitation, and pelleted beads were washed 4 X with ice cold TBS. Beads were resuspended in 1 X Laemmli sample buffer and captured proteins were eluted at 70 °C for 10 min.

Following elution, beads were discarded and the supernatants and input samples were analyzed by SDS-PAGE and Western blot. The eIF4E2 signal in input and cap-elution lanes was quantified by densitometry using Bio-Rad Image Lab software. The eIF4E2 signal in the DDX28 knockdown cap-elution lane relative to control was normalized to the eIF4E2 input signal ratio.

All averages depict the mean ± standard error of the mean (SEM) for at least three independent biological replicates. Differences between group means were assessed with an unpaired two- sample t-test using GraphPad Software.

2.8 Crystal Violet Growth Curves

For each indicated time point, 10,000 cells per well were plated in triplicate in a 24-well plate.

The following day (day 0), cells were incubated at their indicated oxygen concentrations, and following each 24 hour increment, cells were washed once with PBS and stained with 400 µL of

1% crystal violet solution prepared in 20% methanol, with gentle rocking for 20 min at room temperature. The crystal violet solution was aspirated and cells were gently washed with water to remove excess stain. Plates were air-dried overnight, and 400 µL of 10 % acetic acid was added to each well and incubated on a shaker for 20 min at room temperature to de-stain cells. The absorbance at 595 nm of each well was measured using the ThermoMax microplate reader. All averages depict the mean ± standard error of the mean (SEM) for at least three independent

biological replicates. Differences between group means were assessed with an unpaired two- sample t-test using GraphPad Software.

2.9 Bromodeoxyuridine Labeling

250,000 cells seeded on three coverslips in a 6 cm dish were incubated at their indicated oxygen concentrations for 24 h prior to treatment with cell proliferation labeling reagent (1:1000; GE

Healthcare Bio-Sciences, Pittsburgh, USA) for one hour. Cells were washed once with PBS and fixed in ice-cold methanol for 10 min. Excess methanol was removed by washing 3 times for 5 min with PBS, and coverslips were incubated with primary anti-bromodeoxyuridine antibody

(1:100; GE Healthcare Bio-Sciences, Pittsburgh, USA) in the dark for one hour at room temperature. Coverslips were washed 3 times for 5 min with PBS and incubated with goat anti- mouse Alexa Fluor 555 secondary antibody (1:1000; Invitrogen, Waltham, USA), in the dark for

1 hour at 37 °C. Cells were counterstained with Hoechst (1:50 000) for 6 min, and coverslips were mounted on microscope slides using prolong gold antifade reagent (Invitrogen, Waltham,

USA). Cells were visualized with the Nikon eclipse Ti-S inverted microscope at 60X magnification. An average of 200 cells were assessed for positive BrdU labeling per each independent biological replicate. All averages depict the mean ± standard error of the mean

(SEM) for at least three independent biological replicates. Differences between group means were assessed with an unpaired two-sample t-test using GraphPad Software.

Chapter 3. Results

3.1 HIF-2α mediates the interaction between DDX28 and eIF4E2

DDX28 was first identified as a potential eIF4E2 interactor by mass spectrometry analysis following IP of recombinant FLAG-eIF4E2 from human U-87 MG glioblastoma cells cultured under normoxic (21% O2) and hypoxic (1% O2) conditions (Figure 4). This interaction however was never validated experimentally by Western blot; therefore, we attempted to capture eIF4E2 via IP of FLAG-DDX28. This study was conducted entirely with the use of U-87 MG, as not only was preliminary research supporting the investigation of DDX28 carried out with this cell line, but so was the initial characterization of the eIF4FH complex72. Due to the previous observation that protein levels of DDX28 substantially decreased under hypoxia, it was decided that expression and IP of recombinant FLAG-peptide C-terminal tagged DDX28 would provide the most favourable context in which the interaction between eIF4E2 and DDX28 could be captured. The choice for the use of a C-terminal rather than an N-terminal epitope tag was based on a previous study, in which it was concluded that fusion of EGFP to the C-terminus of DDX28 did not appear to affect the localization of the protein or ATPase activity 89. Despite the fact that exogenous DDX28 was readily immunoprecipitated from both normoxic and hypoxic cell lysates, eIF4E2 co-IP failed to be detected above the background levels observed in control IPs in either condition (Figure 8A). Since a large proportion of DDX28 localizes to mitochondria to fulfill its role in mitoribosome biogenesis, we reasoned that perhaps the percentage of DDX28 interacting with eIF4E2 at a given time within the cell was too low to be detected beyond background levels through IP of DDX28 itself 82. We therefore attempted to replicate the initial

IP following transfection of U-87 MG with recombinant FLAG-eIF4E2; however, co-IP of

DDX28 was still not obtained (Figure 8B). We then considered the possibility that the

previously observed interaction between eIF4E2 and DDX28 was not a result of direct protein- protein binding, but instead required mediation by another eIF4FH member, making the interaction more precarious to maintain ex vivo. Given that HIF-2α is a known eIF4E2-binding partner, and an integral component of eIF4FH, we chose to IP exogenously expressed recombinant GFP-HIF-2α in cells stably expressing FLAG-DDX28. Taking advantage of the affinity tag to maximize protein precipitation, GFP-HIF-2α successfully co-precipitated with both eIF4E2 and FLAG-DDX28 under hypoxic conditions (Figure 8C). This suggests that the interaction between eIF4E2 and DDX28 is potentially mediated by HIF-2α.

A) Input IP HCtrl NFLAG HFLAG NCtrl H FLAG HFLAG NCtrl NFLAG HCtrl FLAG (DDX28) eIF4E2 (SE) C) Input GFP IP eIF4E2 (LE) GFP-HIF-2α (−) (+) (−) (+)

HIF-2α

FLAG B) (DDX28) Input IP eIF4E2 N FLAG NFLAG HFLAG HCtrl NCtrl H FLAG HFLAG NCtrl NFLAG HCtrl FLAG (eIF4E2)

DDX28

Figure 8. DDX28 interacts with HIF-2α in U-87 MG. A) FLAG co-immunoprecipitation of recombinant FLAG-DDX28 from normoxic (N) and hypoxic (H) U-87 MG cell lysates. Cells expressing an empty FLAG vector was used as a control (Ctrl). 25 µg of whole cell lysate was used as the input. SE, short exposure. LE, long exposure. B) FLAG co-immunoprecipitation of recombinant FLAG-eIF4E2 from normoxic and hypoxic U-87 MG lysates. Cells expressing an empty FLAG vector was used as a control. 25 µg of whole cell lysate was used as the input. C) GFP co-immunoprecipitation of recombinant GFP-HIF-2α from hypoxic U-87 MG lysates stably expressing FLAG-DDX28. Cells mock-transfected with no DNA was used as a negative control (−). 25 µg of whole cell lysate was used as the input.

3.2 Knockdown of DDX28 enhances the cap-binding affinity of eIF4E2

The specialized function of eIF4E2 and other cap-binding proteins to specifically recognize and bind the cap, and orient other initiation factors at an mRNAs 5’ end, is the rate limiting and arguably most critical step of cap-dependent translation initiation. However, not all cap-binding proteins are created equally. eIF4E’s significantly increased affinity over eIF4E2 for m7GTP under normoxic conditions allows the protein to successfully compete for the binding of available capped transcripts, minimizing the activity of eIF4E275. These differences in cap-

binding affinities help to determine how the cell utilizes a particular protein in different circumstances. Consequently, the intrinsic affinity of cap-binding proteins can be modified by their interactions with different regulatory proteins, to help elicit the appropriate cellular response. It is therefore extremely beneficial to determine how depletion of DDX28 affects eIF4E2’s ability to interact with the cap. To accomplish this, U-87 MG cells stably expressing one of two commercially available shRNA-expression constructs targeting different regions of the DDX28 coding sequence, or a non-targeting control, were generated (Figure 9B). For each shRNA-expression vector a single clone was selected, and is herein referred to as knockdown

(KD) 3.1 and KD4.9. Following exposure of KD3.1, KD4.9 and control cell lines to 21% or 1%

O2 for 24 hours, equal samples of whole cell protein lysate were used in a cap affinity assay to quantitatively assess eIF4E2’s ability to bind the cap, the workflow to which is described in

Figure 6. Using densitometry to quantify band intensity, it was found that under hypoxic conditions eIF4E2 bound the cap with an average of 1.6-fold greater efficiency in KD3.1 cells, and 1.5-fold more in KD4.9 cells, compared to control (Figure 9D). Surprisingly, depletion of

DDX28 in cells cultured in normoxic conditions had an even more profound effect on eIF4E2’s ability to bind the cap. In this condition, an average of 2.9- and 2.5-fold more eIF4E2 was found to bind to the cap in KD3.1 and KD4.9 cells, respectively, compared to the control (Figure 9C).

Statistical analyses confirmed that not only did depletion of DDX28 significantly increase the amount of cap-bound eIF4E2 compared to control cells in both normoxic and hypoxic conditions, but there was also a significant increase in the amount of cap-associated eIF4E2 relative to control in KD cells cultured in 21% O2, compared to 1% O2 (Figure 9E-F). This suggests that the effect(s) of DDX28 depletion on the cap-binding activity of eIF4E2 may be more pronounced under normoxic conditions.

B) 21% O2 1% O2 KD4.9 Ctrl Ctrl KD3.1 KD3.1 KD4.9 DDX28 (SE) A) 2.0 2.5 DDX28 (LE) 2.0 1.5 21% O2 1% O2 1.5 HIF-2α 1.0 HIF-2α 0.5 1.0 eIF4E 0.08 DDX28 0.06 0.5 eIF4E2 0.04 GAPDH 0.0 Actin 0.02

DDX28/GAPDH Expression 0.00 100% 14% DDX28/Actin Expression

N WT H WT N Ctrl H Ctrl N KD3.1N KD4.9 H KD3.1H KD4.9 C) Normoxia (21% O2) GTP GTP 7 GTP GTP 7 GTP GTP 7 GTP GTP 4.0 7 3.0 * CtrlInput KD3.1Input CtrlBlank KD3.1Blank Ctrlm KD3.1m *

3.0 DDX28 CtrlInput KD4.9Input CtrlBlank KD4.9Blank Ctrlm KD4.9m 2.0 DDX28 eIF4E2 2.0 eIF4E2 1.0 eIF4E2 (LE) 1.0 eIF4E2 (LE) eIF4E 0.0 0.0

KD3.1 KD4.9 eIF4E2 Captured Relative to Control Control eIF4E2 Captured Relative to Control Control

D) Hypoxia (1% O2) GTP GTP 7 GTP GTP 7 GTP GTP 7 GTP GTP 7

CtrlInput KD3.1Input CtrlBlank KD3.1Blank Ctrlm KD3.1m 2.0 2.0 * * DDX28

1.5 CtrlInput KD4.9Input CtrlBlank KD4.9Blank Ctrlm KD4.9m 1.5 eIF4E2 1.0 DDX28 1.0 eIF4E2 (LE) eIF4E2 0.5 0.5 eIF4E eIF4E2 (LE) 0.0 0.0

KD3.1 KD4.9 eIF4E2 Captured Relative to Control Control

eIF4E2 Captured Relative to Control Control E) F)

4 3 * *

3 2

1 1

0 0 Captured eIF4E2 KD3.1/Ctrl Captured eIF4E2 KD4.9/Ctrl

N KD3.1 H KD3.1 N KD4.9 H KD4.9

Figure 9. Knockdown of DDX28 increases the cap-binding affinity of eIF4E2 in U-87 MG. A) Comparison and quantification of DDX28 protein levels in U-87 MG cultured in normoxic (21% O2) and hypoxic (1% O2) conditions for 24 hours. GAPDH was used as a loading control. B) Comparison and quantification of DDX28 protein levels in normoxic or hypoxic U-87 MG stably expressing one of two shRNAs (KD3.1 or KD4.9) targeting the coding region of DDX28, or a non-targeting shRNA as a control (Ctrl). Actin was used as a loading control. C) Comparison and quantification of eIF4E2 cap-binding affinity in control and DDX28 knockdown (KD3.1 left, KD4.9 right) U-87 MG cultured in normoxic or D) hypoxic conditions. 35 µg of whole cell lysate was used as the input. LE, long exposure E) Comparison between the ratio of cap-bound eIF4E2 in KD3.1/control cells or F) KD4.9/control cells in normoxic (N) versus to hypoxic (H) conditions. Bio-Rad Image Lab software was used to quantify band intensity by densitometry. All averages depict the mean ± standard error of the mean (SEM) for at least three independent biological replicates. Differences between group means were assessed with an unpaired two-sample t-test using GraphPad Software. α=0.05. * represents p-values < 0.05.

Given that eIF4E is the predominant cap-binding protein in normoxia, we began to consider whether the increase in cap-bound eIF4E2 in KD cells would have an effect on the proportion of cap-bound eIF4E. Stripping and re-probing a subset of the membranes from experiments described above for eIF4E revealed that when eIF4E2 binding to the cap was increased, eIF4E appeared to decrease (Figure 9C-D). However, upon inspection of the whole cell lysate input lanes, which were equally loaded, eIF4E also appeared to be decreased in KD cells compared to control cells. While compelling, further investigation is still required at this time as only DDX28 KD3.1 samples were analyzed. Intriguingly, in all described conditions,

DDX28 was not detected in m7GTP-captured elution lanes, implying that the indirect association between DDX28 and eIF4E2 does not occur while eIF4E2 is interacting with the cap. Taken together, these data suggest that DDX28 decreases the affinity of eIF4E2 for the m7GTP cap structure.

3.3 Knockdown of DDX28 increases eIF4E2-mediated translation

Alteration to the intrinsic affinity of cap-binding proteins often results in changes to protein function; however, an increase in cap-binding affinity does not always directly correlate with an

increase in translation initiation. For example, binding of the regulatory protein 4E-BP to eIF4E is known to increase eIF4E’s affinity for the cap, yet the inability of eIF4G to interact with 4E-

BP-bound eIF4E renders the associated transcripts translationally inhibited53. So, although depletion of DDX28 was observed to increase eIF4E2’s affinity for the cap, this does not exclusively imply a resulting increase in eIF4FH –mediated translation. Therefore, to directly address how DDX28 affects eIF4E2’s role in translation initiation, we monitored the association of eIF4E2 with polyribosomes obtained from control and KD cells cultured in both normoxic and hypoxic conditions.

Polysome profiling is a technique used to capture multiple actively translating ribosomes

(polysomes) and their bound mRNAs, effectively separating them from free ribosomal subunits and inactive ribosomal complexes, known as monosomes90. By exploiting the feature of mRNAs undergoing active translation to have a larger mass than their translationally inactive counterparts, as a result of ribosomal occupancy, polysomes can successfully be isolated from monosomes using sucrose density gradient centrifugation90. The subsequent measurement of sample absorption at 260 nm while the sucrose gradient is fractionated generates a characteristic absorption profile, wherein early peaks correspond to ribosomal subunits and monosomes, and later peaks to light and heavy polysomes (Figure 7). Isolation of total protein from individual fractions and subsequent Western blotting analysis allows for the identification of initiation factors and other proteins within the discrete regions of the corresponding polysome profile.

Therefore, a high monosome association for a particular translation factor would indicate a minimal contribution to translation, whereas a high polysome association would indicate the opposite.

Using densitometry to quantify the total eIF4E2 signal for each sample, it was found that on average 95% of the eIF4E2 signal from control cells cultured under normoxic conditions was present in monosome fractions (Figure 10A). This result was expected given that the contribution of eIF4E2 to translation is known to be minimal at 21% O2. Upon depletion of

DDX28, there was an increase in the proportion of polysome-associated eIF4E2 even under 21%

O2, with only an average of 78% of eIF4E2 present in monosomes; however, this difference was not statistically significant (Figure 10A and 10C). Control cells cultured under hypoxic conditions for 24 hours also saw a dramatic decrease in the amount of eIF4E2 present in monosomes, comparable to that of normoxic KD cells, with an average of 79% (Figure 10A-B).

While the substantial increase in polysome-associated eIF4E2 from 5% in normoxic control cells to 21% in hypoxic control cells was not surprising given the current knowledge regarding the function of eIF4FH, the difference was considered statistically non-significant (Figure 10C).

Knockdown of DDX28 in hypoxia had an even greater effect on eIF4E2’s participation in translation, with an average of 42% of eIF4E2 associating with polysomes, and correspondingly only 58% with monosomes (Figure 10B). Not only are these findings in agreement with previous reports describing eIF4E2’s increased contribution to translation in hypoxic conditions, but taken together they also show that depletion of DDX28 significantly increases the proportion eIF4E2 participating in active translation in hypoxic, but not normoxic conditions. Furthermore, quantification of the percentage of eIF4E in monosome and polysome fractions in all conditions revealed that depletion of DDX28 has no significant effect on eIF4E’s participation in translation (Figure 10E).

Normoxia (21% O A) 2) Control DDX28 KD 1 1

0.9 0.9 40S 0.8 0.8 80S 40S 0.7 0.7 80S 0.6 0.6 0.5 0.5 Polysomes Polysomes 0.4 0.4

0.3 0.3

0.2 0.2 Absorbance254at nm Absorbance254at nm 0.1 0.1

0 0 Fraction 81 2 161 3 241 4 321 5 401 6 481 7 561 8 641 9 72110 801 Fraction 161 2 241 3 3214 401 5 481 6 561 7 641 8 721 9 80110 881 DDX28 DDX28 eIF4E2 eIF4E2 eIF4E eIF4E RPL5 RPL5

Hypoxia (1% O B) 2) Control DDX28 KD 1 1

0.9 0.9

0.8 0.8

0.7 0.7

0.6 80S 0.6 80S at 254at nm at 254at nm 0.5 0.5 Polysomes 0.4 0.4 Polysomes 0.3 0.3

0.2 0.2 Absorbance Absorbance 0.1 0.1

0 0 Fraction 81 2 161 3 241 4 321 5 401 6 481 7 561 8 641 9 72110 801 Fraction 81 2 161 3 241 4 321 5 401 6 481 7 561 8 641 9 72110 801 DDX28 DDX28 eIF4E2 eIF4E2 eIF4E eIF4E RPL5 RPL5

C) D)

* 60 * 100 80 100 * * * * 60 40

50 50 40 20 20

0 0 0 0 % eIF4E in Polysomes % eIF4E in Monosomes % eIF4E2 in Polysomes % eIF4E2 in Monosomes

N Ctrl H Ctrl N Ctrl H Ctrl N Ctrl H Ctrl N Ctrl H Ctrl N KD3.1 H KD3.1 N KD3.1 H KD3.1 N KD3.1 H KD3.1 N KD3.1 H KD3.1

Figure 10. Knockdown of DDX28 in U-87 MG significantly increases polysome-associated eIF4E2 in hypoxia. Polysome fractions were collected from U-87 MG shRNA control (left) and DDX28-knockdown (KD, right) cells maintained in A) 21% O2 and B) 1% O2 for 24 hours. Total protein was isolated from individual fractions and subjected to SDS-PAGE and Western blot analysis for the indicated proteins. C) The cumulative eIF4E2 or D) eIF4E signal from normoxic (N) and hypoxic (H) profiles was quantified by densitometry using Image Lab software, and the % of eIF4E2 or eIF4E in monosome and polysome fractions relative to the total signal was calculated. All averages depict the mean ± standard error of the mean (SEM) for at least three independent biological replicates. Differences between group means were assessed with a one- way ANOVA and Tukey’s honestly significant difference (HSD) post-hoc test using GraphPad Software. α=0.05. * represents p-values < 0.05.

3.3 Knockdown of DDX28 increases translation of EGFR mRNA

In addition to isolating ribosomal proteins and translation factors participating in translation, polysome profiling can also be used to analyze the translatome of the cell. In this scenario, total

RNA, rather than protein, is extracted from monosome and polysome fractions. Reverse transcription (RT) and quantitative PCR (qPCR) can then be used to determine the relative quantity of specific mRNAs undergoing translation at the time of lysis, based on their association with polysomes. By probing for transcripts that are translated by distinct machineries, we can gain insights into the activity of the translation factors that govern their expression. The shift of

EGFR into polysomes for example is a good indicator of eIF4E2’s translational activity, as it was the first rHRE-containing mRNA shown to be translated in an eIF4E2-dependent manner72. Heat shock protein 90ab1(HSP90ab1) on the other hand does not contain an rHRE, and instead displays preference towards eIF4E for its translation77. Consequently, the presence of HSP90ab1 in polysomes is indicative of eIF4Es translational activity, and theoretically should not be affected by alterations to eIF4E2-mediated translation. We therefore performed polysome profiling using control and KD3.1 cells grown in both normoxic and hypoxic conditions followed by qRT-PCR to probe for EGFR and HSP90ab1 to corroborate the observation that knockdown of DDX28 increases eIF4E2, but not eIF4E-mediated translation in hypoxic

conditions. Normalizing to GAPDH and RPLP0, we observed a significant increase in the fold expression of EGFR in polysomes relative to monosomes from 4.46-fold in control cells to

11.26-fold in KD cells under hypoxia (Figure 11A). Under normoxic conditions, the fold expression of EGFR in polysomes relative to monosomes in DDX28 KD cells was substantially lower at 3.82-fold. This was comparable to that of the control cells at 3.47-fold (Figure 11A).

Somewhat surprisingly, the fold expression of HSP90ab1 significantly increased in KD polysomes relative to monosomes in hypoxia compared to both normoxic control and KD cells; however, the difference in expression of HSP90ab1 between control and KD cells in both conditions was not statistically significant (Figure 11B). These results are in agreement with the polysome data observed at the protein level, and together indicate that depletion of DDX28 has no effect on eIF4E-mediated translation, but increases the translational activity eIF4E2 under hypoxic conditions.

A) B) EGFR Expression HSP90ab1 Expression

* * * 15 * 5 *

4 10 3

2 5 1 Expression Ratio Expression Ratio

0 0 Polysome to Monosome Polysome to Monosome

N Ctrl H Ctrl N Ctrl H Ctrl N KD3.1 H KD3.1 N KD3.1 H KD3.1

Figure 11. Knockdown of DDX28 in U-87 MG significantly increases translation of EGFR mRNA in hypoxia. Polysome fractions were collected from U-87 MG shRNA control (Ctrl) and DDX28-knockdown (KD) cells maintained in normoxic (N) or hypoxic (H) conditions for 24 hours. Total RNA was isolated from individual fractions, reverse transcribed, and amplified by quantitative PCR (qPCR) using primers for A) EGFR or B) HSP90ab. The fold change in expression in polysome and monosome fractions was calculated using the ΔΔCT method, and transcript levels were normalized to RPLP0 and GAPDH expression. Bar graphs represent the normalized fold expression of A) EGFR or B) HSP90ab in polysomes relative to the normalized fold expression in monosomes. All averages depict the mean ± standard error of the mean (SEM) for at least three independent biological replicates. Differences between group means were assessed with a one-way ANOVA and Tukey’s honestly significant difference (HSD) post-hoc test using GraphPad Software. α=0.05. * represents p-values < 0.05.

3.4 Knockdown of DDX28 Increases Cell Viability in Hypoxia

Dysregulation of translation often ultimately manifests as changes to cellular growth and/or proliferation, and can therefore be considered a hallmark of cancer91. This phenomenon is exemplified by the deletion of raptor, the subunit of mTOR responsible for phosphorylating 4E-

BPs, which results in increased programmed cell death and decreased cellular proliferation, as a result of decreased translation92. Alternatively, overexpression of eIF4E and increased translation of eIF4F-dependent mRNAs has been linked to increased rates of proliferation, and neoplastic transformation91. We therefore sought to determine if depletion of DDX28, and the resultant

changes to eIF4FH-mediated translation, have any effect on the viability of cells grown in 21% or

1% O2. To assess this, we monitored the average number of viable KD3.1, KD4.9, and control cells in culture over a 72 hour growth period using crystal violet staining. This basic technique relies on the staining of DNA and proteins of adherent cells with crystal violet dye to indirectly determine cell counts using a pre-generated standard curve. The characteristic of dead cells to lose adherence in culture is exploited by this assay, as dead cells are washed away prior to staining, and therefore only the number of viable cells are reflected by the cell counts. All absolute cell count values obtained were normalized to the number of cells present on day 0 for each individual independent experiment, representing the fold change in the number of cells at each time point relative to day 0. When cells were grown in 21% O2, there was no significant difference in cell viability between either KD3.1 or KD4.9 cells compared to control cells following 24, 48, or 72 hours of growth (Figure 12A). However, when cells were grown in 1%

O2, both KD3.1 and KD4.9 cells had significantly increased viability compared to control cells following 24, 48, and 72 hours of growth (Figure 12B). It is also interesting to note that the largest disparity between KD and control cell viability was apparent on day three, suggesting that the effects of DDX28 depletion on cell viability may be more pronounced following more chronic exposures to hypoxia.

1.0 A) 0.8 0.6 Normoxia (21% O2) 0.4 Control DDX28 KD3.1 24 Hours 48 Hours 0.2 72 Hours DDX28 KD4.9 log (Cell Count) 0.0 1.0 3 5 6 -0.2 0.8 0 24 48 72 4 5 Hours 0.6 2 4 3 0.4 Control 3 DDX28 KD3.1 2 0.2 1 2 DDX28 KD4.9 log (Cell Count) Fold Cell Count Fold Cell Count Fold Cell Count 1 1 0.0

0 0 0 -0.2 0 24 48 72 Ctrl Ctrl Ctrl KD3.1 KD4.9 KD3.1 KD4.9 KD3.1 KD4.9 Hours

1.0 B) 0.8 0.6 Hypoxia (1% O2) 0.4 Control DDX28 KD3.1 24 Hours 48 Hours 0.2 72 Hours 0.8 DDX28 KD4.9 log (Cell Count) 0.0 4 5 * * 7-0.2 * 0.6 * 6 0 * 24 48 72 * 4 3 5 Hours Control 3 4 0.4 2 DDX28 KD3.1 2 3 0.2 1 2 DDX28 KD4.9 log (Cell Count) Fold Cell Count Fold Cell Count

1 Fold Cell Count 1 0 0 0 0.0 0 24 48 72 Ctrl Ctrl Ctrl KD3.1 KD4.9 KD3.1 KD4.9 KD3.1 KD4.9 Hours

Figure 12. Knockdown of DDX28 increases U-87 MG viability in hypoxia. Crystal violet staining was used to determine average viable cell counts of U-87 MG stably expressing one of two DDX28-targeting shRNAs (KD3.1 or KD4.9), or an shRNA control (Ctrl), grown in either A) 21% O2 or B) 1% O2 every 24 hours over a total growth period of 72 hours. All averages depict the mean ± standard error of the mean (SEM) for at least three independent biological replicates. Differences between group means were assessed with an unpaired two-sample t-test using GraphPad Software. α=0.05. * represents p-values < 0.05.

3.5 Knockdown of DDX28 Increases Cell Proliferation Rates in Hypoxia

Using crystal violet staining to estimate viable cell counts is a simple and rapid way to determine growth rates of adherent cells; however, substantive information regarding the underlying reason for observed differences in cell viability cannot be ascertained using this assay. For example, the increased growth rates of KD cells observed in hypoxia compared to control could be a result of decreased cell death, increased rates of proliferation, or a combination of the two. Therefore, to gain further insights into the observed increase in cell viability of KD cells compared to control cells, we used immunofluorescence to measure the incorporation of bromodeoxyuridine (BrdU)

into the nuclei of actively dividing cells. Following incubation in either 21% or 1% O2 for 24 hours, cells were treated with BrdU, and the percentage of Brdu-containing KD3.1, KD4.9 and control cells were quantified to determine rates of proliferation. Under normoxic conditions, control and KD4.9 proliferation rates were comparable, at 29.80% and 29.38%, respectively.

However, KD3.1 cells proliferated significantly more at 41.67% (Figure 13A and 13C). Under hypoxic conditions, KD3.1 and KD4.9 cells proliferated at a more comparable rate of 31.08% and 28.67%, respectively, both of which were significantly greater than the proliferation rate of control cells at 23.24% (Figure 13B-C). Taken together, these data indicate that knockdown of

DDX28 increases the proliferation of cells cultured under hypoxic conditions, which is in agreement with the data obtained from crystal violet viability experiments. Since KD3.1 but not

KD4.9 cells cultured under normoxic conditions displayed a significant increase in proliferation compared to control, it cannot unequivocally be stated that the observed results were due to the reduction in DDX28.

A) B) Normoxia (21% O2) Hypoxia (1% O2) BrdU/Hoechst BrdU BrdU/Hoechst BrdU Control Control KD3.1 KD4.9

C) D) 50 * 50 * 40 40 *

30 30

20 20

10 10 % BrdU Incorporation 0 % BrdU Incorporation 0

N Ctrl H Ctrl N KD3.1 N KD4.9 H KD3.1 H KD4.9

Figure 13. Knockdown of DDX28 increases U-87 MG proliferation rates in hypoxia. U-87 MG stably expressing one of two DDX28-targeting shRNAs (KD3.1 or KD4.9), or an shRNA control (Ctrl) were cultured in A) normoxic or B) hypoxic conditions for 24 hours and stained with anti-BrdU. C) Quantification of BrdU incorporation in normoxic (N) and D) hypoxic (H) conditions. All averages depict the mean ± standard error of the mean (SEM) for at least three independent biological replicates. Differences between group means were assessed with an unpaired two-sample t-test using GraphPad Software. α=0.05. * represents p-values < 0.05.

Chapter 4. Discussion

Despite the advances made in the fields of molecular and cancer biology over the past decade, many patients diagnosed with cancer today are still receiving long-established treatments such as chemo and radiotherapy, which are known to have substantial side-effects due to their indiscriminate nature93. The identification of mechanisms that are unique to cancer metabolism and growth are therefore pivotal to the development of new cancer therapeutics that are not only effective, but also selective. The discovery of eIF4E2 as a physiologically relevant alternate cap- binding factor that is hijacked by highly aggressive cancer cells residing in the hypoxic tumor microenvironment has provided us with one such potential therapeutic target. It has also provided us with the opportunity to investigate the complex underpinnings of a novel translational machinery that will undoubtedly contribute to our knowledge of how cells cope with hypoxic stress in both physiological, and pathological contexts. This study in particular has aimed to investigate the potential role of the DEAD-box helicase, DDX28, as a novel interactor and negative regulator of eIF4E2 in the human glioblastoma cell line, U-87 MG.

The first objective of this research was to confirm the interaction between DDX28 and eIF4E2, as previously observed by mass spectrometry analysis following IP of FLAG-eIF4E2 from U-87 MG in the Uniacke lab (unpublished). Although IP of recombinant DDX28 or eIF4E2 was unable to recapture the interaction, IP of the known eIF4E2 binding partner and eIF4FH component, HIF-2α, from hypoxic cell lysates successfully co-immunoprecipitated both eIF4E2 and DDX28 (Figure 8C). This result suggests that perhaps DDX28 and eIF4E2 do not associate by direct protein-protein interactions, but rather interact indirectly through HIF-2α. The mode by which this may occur however is still elusive, particularly due to the nature of DEAD box proteins to interact with both RNA and protein substrates81. Since IP was not carried out in the

presence of external RNAses, such as RNAse A, it is conceivable that HIF-2α was able to co- precipitate DDX28 through an RNA intermediate. So although we have provided evidence to support our hypothesis that eIF4E2 and DDX28 interact, albeit indirectly, further characterization of the nature of the interaction between HIF-2α, DDX28, and eIF4E2 should be a main topic of investigation for future studies.

Next we sought to determine what effect(s) DDX28 has on the involvement of eIF4E2 in translation initiation under 21% and 1% O2 conditions. To this end, we generated DDX28 KD U-

87 MG cell lines through the stable integration of DDX28-targetting shRNAs, and monitored the affect this had on eIF4E2’s cap-binding and translational activity. Using m7GTP-conjugated sepharose resin, we observed an overall increase in the amount of cap-bound eIF4E2 when

DDX28 was depleted, compared to control cells, regardless of O2 concentration (Figure 9C-9F).

Analysis of DDX28 KD and control whole cell lysates by Western blotting for eIF4E2 revealed that knocking down DDX28 does not affect the abundance of eIF4E2, which excludes the possibility that eIF4E2 bound to the cap resin more as a result of elevated protein levels (Figure

9A). Instead, this suggests that DDX28 normally functions to reduce the affinity of eIF4E2 for the cap.

While depletion of DDX28 was observed to significantly increase the cap-affinity of eIF4E2 in both normoxic and hypoxic conditions, it was somewhat surprising to find that the increase in cap-bound eIF4E2 under normoxic conditions was significantly greater than that observed in hypoxic conditions (Figure 9E-9F). Initially, we perceived this as an indication that

DDX28 has increased relevance as a regulator of eIF4E2 under normoxic conditions; however, we now speculate that that the observed results were a consequence of the differing levels of

DDX28 in 21% and 1% O2 conditions. Comparison of DDX28 in wild-type U-87 MG

maintained at 21% O2 versus 1% O2 for 24 hours by Western blotting revealed that hypoxia reduces levels of the protein by approximately 86% (Figure 9A). We observed this same phenomenon in control and DDX28 KD cells; therefore, we predict that the effect of DDX28 knockdown on eIF4E2 cap-binding is more apparent in normoxia, in comparison to hypoxia when levels of DDX28 are already naturally depleted. This is supported by quantification of

DDX28 levels using densitometry in control and DDX28 KD cells maintained in normoxic and hypoxic conditions, which shows that the level of DDX28 depletion achieved in comparison to each respective control is greater in normoxic conditions for both DDX28 KD3.1 and 4.9 cells

(Figure 9B). In accordance with this, we also speculate that the somewhat heightened effects observed in DDX28 KD3.1 versus 4.9 cells in both conditions is a direct result of the increased shRNA-mediated reduction in DDX28 achieved in this clonal cell line (Figure 9B).

Under no circumstance was DDX28 detected in cap-bound elution lanes. This implies that the mechanism by which DDX28 alters eIF4E2’s ability to interact with the cap does not require an association between DDX28 and the cap, or cap-bound eIF4E2. Rather, we hypothesize that DDX28 exerts its regulation on “free” eIF4E2, prior to the point of cap association. Since many characterized negative regulators of eIF4E have actually been found to enhance retention of eIF4E on m7GTP resins by interacting directly with the protein through a conserved binding motif, we postulate that DDX28 functions by means of a mechanism that is unique to eIF4E2-mediated translation53,84. One point of interest for future investigation in regards to this is the decrease in cap-bound eIF4E accompanying the increase in cap-bound eIF4E2 observed in DDX28 KD3.1 cells (Figure 9C and 9D). While it is difficult at this time to draw any definitive conclusions, it is tempting to speculate that in the absence of DDX28, eIF4E2 acquires an augmented affinity for the cap that allows the protein to more efficiently

compete against eIF4E for binding of available capped transcripts, resulting in decreased levels of cap-bound eIF4E.

To corroborate that the observed increase in binding of eIF4E2 to the cap in DDX28 KD cells results in an increase in eIF4FH- mediated translation, we monitored the association of both eIF4E2, and the eIF4E2-dependent transcript EGFR, with polysomes isolated from DDX28 KD and control cells cultured in 21% or 1% O2. Depletion of DDX28 resulted in an increase in the proportion of eIF4E2 associated with polysomes in both normoxic and hypoxic conditions, seemingly confirming that DDX28 negatively regulates the cap-binding affinity and consequently translational activity of eIF4E2. This increase however was only considered to be statistically significant in hypoxia (Figure 10C). This instead suggests that DDX28 only functions as a negative regulator of eIF4E2 under 1% O2. Analysis of EGFR transcript abundance in polysomes by qPCR further validated these results, as depletion of DDX28 in hypoxia, but not normoxia, led to a significant increase in the fold expression of EGFR in polysomes relative to monosomes, compared to control cells (Figure 11A).

The discrepancy observed in normoxia between the cap-binding and translational activity of eIF4E2 can be attributed to differences in the nature of the two techniques used. Cap assays are performed in vitro, and therefore can only provide us with information regarding the ability of eIF4E2 to interact with the cap in a controlled environment. Polysome profiling on the other hand more accurately depicts translation as it occurs in vivo. Nevertheless, the results from the cap assay are still beneficial as they reveal that DDX28’s ability to alter eIF4E2’s affinity for the cap is unlikely to be regulated by changes in O2 concentration. However, for eIF4E2 to function as a translation initiator in vivo, the protein must act in concert with other members of the eIF4FH complex. This includes the hypoxia-dependent subunit, HIF-2α72. Accordingly, even though

depletion of DDX28 results in increased binding of eIF4E2 to the cap, the rapid degradation of

HIF-2α in normoxic conditions prevents eIF4E2 from initiating translation, which is reflected by the polysome profiling results. With that said, many cancers often have a severely dysregulated hypoxic response due to underlying mutations2,16,42. It is therefore important to take into consideration the highly transformed state of the cell line used in these experiments when analyzing results. Multiple groups have reported in the literature that HIF-2α can be detected at low levels in the absence of hypoxia in several malignant cell lines, including U-87 MG28,39,94,95.

H This could contribute to background activation of the eIF4F complex even at 21% O2, and may serve as an explanation for the observed increase in polysome-associated eIF4E2 in DDX28 KD cells under normoxic conditions (Figure 10A and 10C). Nevertheless, the levels of HIF-2α stabilized under these conditions are still unlikely to compare with those induced by hypoxic stress and probably explains why the shift of eIF4E2 from monosomes to polysomes was considered to be statistically significant in hypoxia, but not normoxia (Figure 10C).

There are several other interesting points that can be made based on the results of the polysome profiling experiments that are worth noting. For instance, hypoxic treatment alone, as expected, led to an increase in the proportion of eIF4E2 in polysomes in both control and

DDX28 KD cells, which corresponded well with an increase EGFR mRNA translation (Figure

10C and 11A). Unexpectedly, this increase was only considered to be statistically significant in

DDX28 KD cells. This suggests that hypoxia has a greater stimulatory effect on eIF4E2- mediated translation in the absence of DDX28, further corroborating the hypothesis that DDX28 negatively regulates eIF4E2 under hypoxia. It is also interesting to note that the levels of EGFR expression and eIF4E2 in polysomes are strikingly similar between the normoxic DDX28 KD and hypoxic control cells (Figure 10C). Based on the previous observation that DDX28 protein

levels are reduced under hypoxia, we suspect that knockdown of DDX28 in normoxia mimics the wild-type hypoxic phenotype to some extent in U-87 MG.

To confirm that knockdown of DDX28 had no effect eIF4E-mediated translation, we also analyzed the association of eIF4E, and the eIF4E-dependent transcript HSP90ab1, with polysomes isolated from control and DDX28 KD cells grown in 21% or 1% O2. No significant difference was detected between DDX28 KD and the respective control cells in either condition, suggesting that DDX28’s regulation of cap-dependent translation is specific to the eIF4FH complex (Figure 10D and 11B). It is important to address that in these experiments hypoxia failed to downregulate eIF4E-mediated translation, as would normally be expected due to the inhibition of eIF4E by 4E-BPs49 (Figure 10D). This can be explained by the PTEN inactivating mutation characteristic of U-87 MG94,96. The phosphatase PTEN functions as an antagonist of mTOR activity by dephosphorylating key activating molecules involved the PI3K/Akt nutrient- signaling pathway97,98. The homozygous mutation of PTEN in U-87 MG however renders the cell line PTEN null, allowing for the persistent activation of mTOR by PI3K/Akt signaling94,96,98.

While there is still some degree of hypoxia-induced mTOR inhibition by AMPK and REDD1 signaling in these cells, eIF4E continues to contribute to cap-dependent translation irrespective of O2 concentration.

Inactivation of PTEN could also help explain the atypical increase in the expression of

HSP90ab1 in hypoxic DDX28 KD polysomes compared to both normoxic control and normoxic

DDX28 KD cells, which was considered to be statistically significant (Figure 11B). This is because activation of EGFR by mitogens directly converges on PI3K/Akt signaling; therefore, the hypoxia-induced upregulation of EGFR observed in DDX28 KD cells likely results in further stimulation of mTOR and eIF4E activity. So even though DDX28 depletion alone does not

appear to affect eIF4E-mediated translation, it may do so indirectly in hypoxic U-87 MG as a result of the increased expression of EGFR and loss of PTEN.

The final objective of this study was to determine how DDX28 knockdown affected U-87

MG cell viability and proliferation under normoxic and hypoxic conditions. We found that compared to the control, DDX28 KD cells had a significant increase in both cell proliferation, and viability when cultured under 1% O2 (Figure 12B and 13D). We hypothesize that this is a direct consequence of the hypoxic-specific increase in EGFR translation and expression observed in DDX28 KD cells. EGFR is a potent growth factor receptor that signals to a network of downstream signaling pathways involved in an array of unique cellular processes including cell proliferation, and growth98. One of the primary ways in which EGFR signaling promotes proliferation is through the synthesis of cyclin D1, which together with cyclin-dependent kinase

4/6 (CDK4/6) initiates progression of the cell cycle past G198. In fact, hyperactivation of EGFR signaling through a number of different means occurs frequently during the development of cancer, and often correlates with increased aggressiveness of the disease, and reduced patient survival98,99. It is interesting to note that prior to the elucidation of the eIF4FH complex, Franovic et al. (2007) published a study in which they discovered that translational upregulation of EGFR in U-87 MG occurred in a HIF-2α-dependent manner, and was a precursor for autonomous proliferation87. Two years later the same group published a follow-up article, in which they reported that hypoxic expression of insulin-like growth factor 1 receptor (IGF1R) was required in addition to EGFR for cells to reach their full proliferative potential94. It is now known that

IGF1R, like EGFR, possesses an rHRE and is a verified target of eIF4FH-mediated translation72.

We therefore speculate that IGF1R translation would also be upregulated in response to the hypoxic knockdown of DDX28, contributing to the observed proliferative advantage of these

cells.

The same effect on cell viability and proliferation in DDX28 KD cells was not observed under normoxic conditions, presumably due to the lack of eIF4FH-mediated translational upregulation of EGFR. There was however a discrepancy between the proliferation rate and viability of DDX28 KD3.1 cells grown in 21% O2, the former data representing a significant increase compared to control, while the latter showed no difference (Figure 12A and 13C).

Taking into consideration the capability of the crystal violet assay to account for dead cells, it is probable that DDX28 KD3.1 cells proliferated significantly faster, but were not more viable due to increased cell death. In these circumstances, the observed effects are more likely a consequence of acquired clonal variation rather than knockdown of DDX28, as DDX28 KD4.9 cells did not respond in a similar manner (Figure 13C). This suggests that knockdown of

DDX28 has no effect on the proliferation rate or viability of cells grown under normoxic conditions. While the conclusions drawn from this objective are valid, the limitations of the assays employed are also quite apparent. Future studies should therefore take into consideration how DDX28 depletion affects apoptosis to gain further insight into why these cells are more viable under hypoxia. This would be particularly relevant as EGFR signaling initiates the expression of anti-apoptotic genes, in addition to those that promote proliferation98.

Based on the data gathered in this study, we propose that DDX28 acts as a negative regulator of eIF4E2-mediated hypoxic translation through interaction with the eIF4FH complex component HIF-2α. While our knowledge of how this mechanism might operate is limited, we do not anticipate that DDX28 functions by inhibiting the interaction between HIF-2α and eIF4E2. This is because we cannot exclude the possibility that HIF-2α interacts with eIF4E2 and

DDX28 simultaneously, as IP of HIF-2α from hypoxic lysates successfully co-

immunoprecipitated both DDX28 and eIF4E2 (Figure 8C). Rather, in this model we hypothesize that DDX28 impedes or restricts HIF-2α’s ability to interact with rHRE-containing mRNAs, either by affecting the interaction between RNA and HIF-2α–bound RBM4, or between HIF-2α and RBM4 itself (Figure 14). Since HIF-2α’s association with rHREs aids in the selective binding of eIF4E2 to the 5’cap of critical hypoxic mRNAs, disruption of this interaction in either of these scenarios would interfere with the recruitment of eIF4E2 to the cap. This could also serve as an explanation for how DDX28 affects eIF4E2’s binding to the cap without directly or indirectly associating with the cap itself. The disruption of eIF4E2’s recruitment would culminate in an overall decrease in translation of rHRE-containing mRNAs, such as EGFR.

eIF4FH-dependent translation eIF4FH-dependent translation Low [DDX28] High [DDX28]

7 m -G rHRE AAAAA m7-G rHRE AAAAA

RBM4 RBM4

HIF-2α HIF-2α DDX28

4E2 4E2

AAAAA rHRE

RBM4

RBM4 HIF-2α No Translation Translation m7-G rHRE AAAAA eIF4A 4E2m7-G eIF4G3

Figure 14. Proposed model of DDX28 regulation of the hypoxic translation initiation complex eIF4FH. In the presence of HIF-2α, the cap binding protein eIF4E2 and other members of the eIF4FH initiation complex are recruited to the 5’ end of a distinct pool of critical hypoxic mRNAs containing RNA hypoxic response elements (rHREs) in their 3’ untranslated regions (UTRs). In the presence of high intracellular concentrations of the DEAD-box protein DDX28, eIF4FH-mediated translation is downregulated resulting in the decreased expression of hypoxic mRNAs. Interactions between DDX28 and HIF-2α may hinder recruitment of the initiation complex to rHRE-containing mRNAs by affecting HIF-2α’s association with the RNA-binding protein RBM4.

This brings into to question why a negative regulator of eIF4FH-mediated translation would be beneficial for a cell, given that the expression of rHRE-containing mRNAs is critical for hypoxic survival. As a consequence of the drastic reduction in DDX28 protein abundance that occurs, we hypothesize that the antagonistic effect of DDX28 on eIF4E2-mediated translation is normally minimal under low O2 conditions. Rather, we speculate that DDX28

functions to prevent aberrant activation of eIF4FH in conditions when the atypical expression of proto-oncogenic mRNAs such as EGFR would be unfavourable. For example, mutations acquired during neoplastic transformation frequently promote the normoxic stabilization of

HIFs2,24,42. So while degradation of HIF-2α would normally prevent the activation of eIF4FH complex, these abnormal cells are capable of initiating translation of potentially dozens of mRNAs that further promote oncogenesis. We speculate that in these conditions DDX28 may help to reduce translation of such transcripts. In accordance with this, data made available by the human protein atlas indicates that partial or complete loss of DDX28 at the protein level is associated with multiple types of cancer100 (Figure 15). DDX28 is also listed as a favourable prognostic marker in cases of renal cell carcinoma (available from v18.proteinatlas.org and https://www.proteinatlas.org/ENSG00000182810-DDX28/pathology). This data suggests that

DDX28 potentially possesses tumor suppressor activity, which corresponds well with our hypothesis that DDX28 helps to curb the translational activity of eIF4FH. Of all the cancer types assessed in this databank, it is compelling that DDX28 expression was found to be a favourable prognostic marker specifically in renal cell carcinoma. Upwards of 80% of all clear cell renal cell carcinomas (CCRCC), the most common form of renal cancer, contain inactivating mutations in

101 VHL . This means that many of these tumors express HIF-α subunits irrespective of O2 concentration, due to pVHL’s pivotal role as part of the VCB-Cul2 E3 degradation complex101.

The hyperactivation of HIF-2α that occurs in cases of CCRCC is therefore a prototypical example of when DDX28’s antagonistic effect on eIF4FH would be beneficial for the cell. This data also highlights the relevancy of DDX28 in other cell types, suggesting the effects observed in this study may not be unique to U-87 MG. For this reason it would be interesting in the future to investigate how the effects of DDX28 KD differ between a primary cell line such as

HRPTECs, in which HIF-α regulation is more typical, compared to a VHL-null cell line like

786-O.

Figure 15. Human Protein Atlas DDX28 protein expression summary. A) Average level of DDX28 protein expression in 44 normal human tissue types. B) Percentage of analyzed patient tumour samples displaying medium to high DDX28 protein expression for the indicated cancer type, white bars indicate low or no protein expression. At least three independent patient tumour samples were analyzed for each cancer type. Colours correspond to the normal organ type from which the cancer originates. All images and data available from v18.proteinatlas.org and https://www.proteinatlas.org/ENSG00000182810-DDX28/pathology.

Chapter 5. Conclusions and Future Directions

This study has provided evidence to support the hypothesis that DDX28 functions as a negative regulator of the eIF4FH complex resulting in the reduced translation of critical hypoxic mRNAs such as EGFR. There is still much to be learned about the precise role of DDX28 in hypoxic cap- dependent translation, and given the novelty of this research, possible future directions are endless. Precedence should be given to the elucidation of the mechanism by which DDX28’s interaction with HIF-2α impedes eIF4FH-mediated translation. Initial experiments could explore if depletion of DDX28 increases HIF-2α’s association with RBM4 through coIP. Given that

DEAD box proteins act as remodelers of RNA-protein complexes, it would also be beneficial to investigate if knockdown of DDX28 increases RBM4’s association with rHRE-containing RNAs through RNA IP81. Additionally, reintroduction of exogenous DDX28 containing various truncations or point mutations into DDX28 KD cells could be performed to assess the functional requirement of the proteins helicase activity in translational regulation. Long-term objectives could include the generation of DDX28 overexpression cell lines to test if increasing levels of

DDX28 has the opposite effect of depletion on eIF4FH-mediated translation.

DDX28’s potential tumour suppressor activity and relevance in cancer could also be further assessed through the employment of number of rudimentary experiments. DDX28 KD cells could be tested for their ability to autonomously proliferate and migrate by measuring BrdU incorporation and performing scratch wound closure experiments in the absence of growth serum. Invasive capabilities of DDX28 KD cells could also be assessed through the use of matrigel chambers, which require cells to pass through a basement membrane mimic to reach a high serum environment.

Regardless of their tissue of origin, solid tumors often contain regions of hypoxia that are

associated with poor patient prognosis16. Increasing evidence suggests that aggressive cancer cells are dependent on eIF4E2-mediated translation to maintain synthesis of hundreds of proteins required for survival during tumor progression17,72. The emerging interrelationships between

DDX28, eIF4E2, HIF-2α, and EGFR provides a promising indication of the relevance DDX28 may have in the ongoing pursuit to further our understanding of the complexities that underlie cancer development. Further investigation of DDX28 and its role eIF4FH-mediated translation will aid in our basic understanding of how this alternate machinery functions in response to cellular stress, and potentially aid in the development of cancer therapeutics that are not only effective, but also selective.

References

1. Majmundar, A. J., Wong, W. J. & Simon, M. C. Hypoxia-inducible factors and the response to hypoxic stress. Mol. Cell 40, 294–309 (2010). 2. Semenza, G. L., Kappler, M., Taubert, H. & Eckert, A. W. Oxygen sensing, homeostasis, and disease. N. Engl. J. Med. 365, 537–547 (2011). 3. Semenza, G. L. Hypoxia-inducible factors: mediators of cancer progression and targets for cancer therapy. Trends Pharmacol. Sci. 33, 207–214 (2012). 4. Semenza, G. L. HIF-1⍺: mediator of physiological and pathophysiological responses to hypoxia. J. Appl. Physiol. 88, 1447–1480 (2002). 5. Yeh, J. J. & Kim, W. Y. Targeting tumor hypoxia with hypoxia-activated prodrugs. J. Clin. Oncol. 33, 1505–1508 (2015). 6. Simon, M. C. & Keith, B. The role of oxygen availability in embryonic development and stem cell function. Nat Rev Mol Cell Biol 9, 285–296 (2008). 7. Dunwoodie, S. L. The role of hypoxia in development of the mammalian embryo. Dev. Cell 17, 755–773 (2009). 8. Withington, S. L. et al. Loss of Cited2 affects trophoblast formation and vascularization of the mouse placenta. Dev. Biol. 294, 67–82 (2006). 9. Amarilio, R. et al. HIF-1⍺ regulation of Sox9 is necessary to maintain differentiation of hypoxic prechondrogenic cells during early skeletogenesis. Development 134, 3917–28 (2007). 10. Licht, A. H., Müller-Holtkamp, F., Flamme, I. & Breier, G. Inhibition of hypoxia- inducible factor activity in endothelial cells disrupts embryonic cardiovascular development. Blood 107, 584–590 (2006). 11. Watson, E. & Cross, J. Development of structures and transport functions in the mouse placenta. Physiology 20, 180–193 (2005). 12. Yun, Z., Maecker, H. L., Johnson, R. S. & Giaccia, A. J. Inhibition of PPARγ2 gene expression by the HIF-1-regulated gene DEC1/Stra13: A mechanism for regulation of adipogenesis by hypoxia. Dev. Cell 2, 331–341 (2002). 13. Jögi, A. et al. Hypoxia alters gene expression in human neuroblastoma cells toward an immature and neural crest-like phenotype. Proc. Natl. Acad. Sci. U. S. A. 99, 7021–6 (2002). 14. Bangoura, G., Yang, L., Huang, G. & Wang, W. Expression of HIF-2⍺/EPAS1 in hepatocellular carcinoma. World J. Gastroenterol. 10, 525–30 (2004). 15. Cairns, R., Papandreou, I. & Denko, N. Overcoming physiologic barriers to cancer treatment by molecularly targeting the tumor microenvironment. Mol. Cancer Res. 4, 61– 70 (2006). 16. Giaccia, A. J., Simon, M. C. & Johnson, R. The biology of hypoxia: the role of oxygen sensing in development, normal function, and disease. Genes Dev. 18, 2183–2194 (2004). 17. Uniacke, J., Perera, J. K., Lachance, G., Francisco, C. B. & Lee, S. Cancer cells exploit eIF4E2-directed synthesis of hypoxia response proteins to drive tumor progression. Cancer Res. 74, 1379–1389 (2014). 18. Hanahan, D. & Weinberg, R. A. Hallmarks of cancer: the next generation. Cell 144, 646– 674 (2011). 19. Keith, B., Johnson, R. S. & Simon, M. C. HIF1α and HIF2α: sibling rivalry in hypoxic

tumour growth and progression. Nat. Rev. Cancer 12, 9–22 (2012). 20. Rockwell, S., Dobrucki, I. T., Kim, E. Y., Marrison, S. T. & Vu, V. T. Hypoxia and radiation therapy: past history, ongoing research, and future promise. Curr. Mol. Med. 9, 442–458 (2009). 21. Kewley, R. J., Whitelaw, M. L. & Chapman-Smith, A. The mammalian basic helix-loop- helix/PAS family of transcriptional regulators. International Journal of Biochemistry and Cell Biology 36, 189–204 (2004). 22. Zhao, J., Du, F., Shen, G., Zheng, F. & Xu, B. The role of hypoxia-inducible factor-2 in digestive system cancers. Cell Death Dis. 6, e1600 (2015). 23. Tian, H., Mcknight, S. L. & Russell, D. W. Endothelial PAS domain protein 1 (EPAS1) a transcription factor selectively expressed in endothelial cells. Genes Dev. 11, 72–82 (1997). 24. Kaelin, W. G. & Ratcliffe, P. J. Oxygen sensing by metazoans: the central role of the HIF hydroxylase pathway. Mol. Cell 30, 393–402 (2008). 25. Franke, K., Gassmann, M. & Wielockx, B. Erythrocytosis: the HIF pathway in control. Blood 122, 1122–1128 (2013). 26. Wiesener, M. S. et al. Widespread hypoxia-inducible expression of HIF-2⍺ in distinct cell populations of different organs. FASEB J. 17, 271–273 (2003). 27. Hu, C.-J., Sataur, A., Wang, L., Chen, H. & Simon, M. C. The N-terminal transactivation domain confers target gene specificity of hypoxia-inducible factors HIF-1⍺ and HIF-2⍺. Mol. Biol. Cell 18, 4528–42 (2007). 28. Holmquist-Mengelbier, L. et al. Recruitment of HIF-1α and HIF-2α to common target genes is differentially regulated in neuroblastoma: HIF-2α promotes an aggressive phenotype. Cancer Cell 10, 413–423 (2006). 29. Iyer, N. V et al. Cellular and developmental control of O2 homeostasis by hypoxia- inducible factor 1⍺. Genes Dev. 12, 149–162 (1998). 30. Peng, J., Zhang, L., Drysdale, L. & Fong, G. H. The transcription factor EPAS-1/hypoxia- inducible factor 2⍺ plays an important role in vascular remodeling. Proc. Natl. Acad. Sci. U. S. A. 97, 8386–91 (2000). 31. Compernolle, V. et al. Loss of HIF-2α and inhibition of VEGF impair fetal lung maturation, whereas treatment with VEGF prevents fatal respiratory distress in premature mice. Nat. Med. 8, 702–710 (2002). 32. Lando, D. Asparagine hydroxylation of the HIF transactivation domain: a hypoxic switch. Science 295, 858–861 (2002). 33. Koh, M. Y., Darnay, B. G. & Powis, G. Hypoxia-associated factor, a novel E3-ubiquitin ligase, binds and ubiquitinates hypoxia-inducible factor 1 , leading to Its oxygen- independent degradation. Mol. Cell. Biol. 28, 7081–7095 (2008). 34. Koivunen, P., Hirsilä, M., Günzler, V., Kivirikko, K. I. & Myllyharju, J. Catalytic properties of the asparaginyl hydroxylase (FIH) in the oxygen sensing pathway are distinct from those of its prolyl 4-hydroxylases. J. Biol. Chem. 279, 9899–9904 (2004). 35. Sowter, H. M., Raval, R., Moore, J., Ratcliffe, P. J. & Harris, A. L. Predominant role of hypoxia-inducible transcription factor (HIF)-1α versus HIF-2α in regulation of the transcriptional response to hypoxia. Cancer Res. 63, 6130–6134 (2003). 36. Luo, J. C. & Shibuya, M. A variant of nuclear localization signal of bipartite-type is required for the nuclear translocation of hypoxia inducible factors (1α, 2α and 3α). Oncogene 20, 1435–44 (2001).

37. Mole, D. R. et al. Genome-wide association of hypoxia-inducible factor (HIF)-1 α and HIF-2α DNA binding with expression profiling of hypoxia-inducible transcripts. J. Biol. Chem. 284, 16767–16775 (2009). 38. Wang, G. L. & Semenza, G. L. General involvement of hypoxia-inducible factor 1 in transcriptional response to hypoxia. Proc. Natl. Acad. Sci. U. S. A. 90, 4304–8 (1993). 39. Uchida, T. et al. Prolonged hypoxia differentially regulates hypoxia-inducible factor (HIF)-1α and HIF-2α expression in lung epithelial cells: implication of natural antisense HIF-1α. J. Biol. Chem. 279, 14871–14878 (2004). 40. Rossignol, F., Vache, C. & Clottes, E. Natural antisense transcripts of hypoxia-inducible factor 1⍺ are detected in different normal and tumour human tissues. Gene 299, 135–140 (2002). 41. Mei Yee Koh, Robert Lemos Jr, Xiuping Liu, and G. P. The hypoxia-associated factor switches cells from HIF-1α– to HIF-2α–dependent signaling promoting stem cell characteristics, aggressive tumor growth and invasion. 71, 4015–4027 (2012). 42. Maxwell, P. H., Pugh, C. W. & Ratcliffe, P. J. Activation of the HIF pathway in cancer. Curr. Opin. Genet. Dev. 11, 293–299 (2001). 43. Carroll, V. A. & Ashcroft, M. Role of hypoxia-inducible factor (HIF)-1α versus HIF-2α in the regulation of HIF target genes in response to hypoxia, insulin-like growth factor-I, or loss of von hippel-lindau function: implications for targeting the HIF pathway. Cancer Res. 66, 6264–6270 (2006). 44. Crick, F. Central dogma of molecular biology. Nature 227, 561–563 (1970). 45. Gradi, A. et al. A novel functional human eukaryotic translation initiation factor 4G. Mol. Cell. Biol. 18, 334–42 (1998). 46. Topisirovic, I., Svitkin, Y. V, Sonenberg, N. & Shatkin, A. J. Cap and cap-binding proteins in the control of gene expression. Wiley Interdiscip. Rev. RNA 2, 277–298 (2011). 47. Sonenberg, N. & Hinnebusch, A. G. Regulation of translation initiation in eukaryotes: mechanisms and biological targets. Cell 136, 731–745 (2009). 48. Ramanathan, A., Robb, G. B. & Chan, S.-H. mRNA capping: biological functions and applications. Nucleic Acids Res. 44, 7511–7526 (2016). 49. Richter, J. D. & Sonenberg, N. Regulation of cap-dependent translation by eIF4E inhibitory proteins. Nature 433, 477–480 (2005). 50. Grifo, J. A., Tahara, S. M., Morgan, M. A., Shatkin, A. J. & Merrick, W. C. New initiation factor activity required for globin mRNA translation. J. Biol. Chem. 258, 5804–5810 (1983). 51. Haghighat, A. & Sonenberg, N. eIF4G dramatically enhances the binding of eIF4E to the mRNA 5’-cap structure. J. Biol. Chem. 272, 21677–21680 (1997). 52. Schutz, P. et al. Crystal structure of the yeast eIF4A-eIF4G complex : an RNA-helicase controlled by protein – protein interactions. Proc. Natl. Acad. Sci. U. S. A. 105, 9564– 9569 (2008). 53. Volpon, L., Osborne, M. J., Topisirovic, I., Siddiqui, N. & Borden, K. L. B. Cap-free structure of eIF4E suggests a basis for conformational regulation by its ligands. EMBO J. 25, 5138–5149 (2006). 54. Jackson, R. J., Hellen, C. U. T. & Pestova, T. V. The mechanism of eukaryotic translation initiation and principles of its regulation. Nat. Rev. Mol. Cell Biol. 11, 113–127 (2010). 55. Gingras, A., Raught, B. & Sonenberg, N. eIF4 initiation factors: effectors of mRNA recruitment to ribosomes and regulators. Annu. Rev. Biochem. 68, 913–963 (1999).

56. Rogers, G. W., Richter, N. J., Lima, W. F. & Merrick, W. C. Modulation of the helicase activity of eIF4A by eIF4B, eIF4H, and eIF4F. J. Biol. Chem. 276, 30914–30922 (2001). 57. Kahvejian, A., Svitkin, Y. V, Sukarieh, R., Boutchou, M. M. & Sonenberg, N. Mammalian poly ( A ) -binding protein is a eukaryotic translation initiation factor , which acts via multiple mechanisms. Genes Dev. 19, 104–113 (2005). 58. Buttgereit, F. & Brand, M. D. A hierarchy of ATP-consuming processes in mammalian cells. Biochem. J. 312, 163–167 (1995). 59. Hay, N. & Sonenberg, N. Upstream and downstream of mTOR. Genes and Development 18, 1926–1945 (2004). 60. Arsham, A. M., Howell, J. J. & Simon, M. C. A novel hypoxia-inducible factor- independent hypoxic response regulating mammalian target of rapamycin and its targets. J. Biol. Chem. 278, 29655–29660 (2003). 61. Moore, C. E. J. et al. Elongation factor 2 kinase Is regulated by proline hydroxylation and protects cells during hypoxia. Mol. Cell. Biol. 35, 1788–1804 (2015). 62. Mungai, P. T. et al. Hypoxia triggers AMPK activation through reactive oxygen species- mediated activation of calcium release-activated calcium channels. Mol. Cell. Biol. 31, 3531–3545 (2011). 63. Liu, L. et al. Hypoxia-induced energy stress regulates mRNA translation and cell growth. Mol. Cell 21, 521–531 (2006). 64. Deyoung, M. P., Horak, P., Sofer, A., Sgroi, D. & Ellisen, L. W. Hypoxia regulates TSC1 / 2 – mTOR signaling and tumor suppression through REDD1-mediated 14 – 3 – 3 shuttling. Genes Dev. 22, 239–251 (2008). 65. Brugarolas, J. et al. Regulation of mTOR function in response to hypoxia by REDD1 and the TSC1 / TSC2 tumor suppressor complex. Genes Dev. 18, 2893–2904 (2004). 66. Connolly, E., Braunstein, S., Formenti, S. & Schneider, R. J. Hypoxia inhibits protein synthesis through a 4E-BP1 and elongation factor 2 kinase pathway controlled by mTOR and uncoupled in breast cancer cells. Mol. Cell. Biol. 26, 3955–3965 (2006). 67. Young, R. M. et al. Hypoxia-mediated selective mRNA translation by an internal ribosome entry site-independent mechanism. J. Biol. Chem. 283, 16309–16319 (2008). 68. Jang, S. K. et al. A segment of the 5’ nontranslated region of encephalomyocarditis virus RNA directs internal entry of ribosomes during in vitro translation. J. Virol. 62, 2636–43 (1988). 69. Stein, I. et al. Translation of vascular endothelial growth factor mRNA by internal ribosome entry: implications for translation under hypoxia. Mol Cell Biol 18, 3112–3119 (1998). 70. van den Beucken, T., Koritzinsky, M. & Wouters, B. G. Translational control of gene expression during hypoxia. Cancer Biol. Ther. 5, 749–755 (2006). 71. Borer, R. A. et al. Internal ribosome entry sites in eukaryotic mRNA molecules. J. Virol. 85, 49–63 (2001). 72. Uniacke, J. et al. An oxygen-regulated switch in the protein synthesis machinery. Nature 486, 126–9 (2012). 73. Rom, E. et al. Cloning and characterization of 4EHP, a novel mammalian eIF4E-related cap-binding protein. J. Biol. Chem. 273, 13104–13109 (1998). 74. Ho, J. J. D. et al. Systemic reprogramming of translation efficiencies on oxygen stimulus. Cell Rep. 14, (2016). 75. Zuberek, J. et al. Weak binding affinity of human 4EHP for mRNA cap analogs. RNA 13,

691–697 (2007). 76. Tee, A. R., Tee, J. A. & Blenis, J. Characterizing the interaction of the mammalian eIF4E- related protein 4EHP with 4E-BP1. FEBS Lett. 564, 58–62 (2004). 77. Timpano, S. & Uniacke, J. Human cells cultured under physiological oxygen utilize two cap-binding proteins to recruit distinct mRNAs for translation. J. Biol. Chem. 291, 10772– 10782 (2016). 78. Valgardsdottir, R., Brede, G., Eide, L. G., Frengen, E. & Prydz, H. Cloning and characterization of MDDX28, a putative DEAD-box helicase with mitochondrial and nuclear localization. J. Biol. Chem. 276, 32056–32063 (2001). 79. Rocak, S. & Linder, P. DEAD-box proteins: the driving forces behind RNA metabolism. Nat. Rev. Mol. Cell Biol. 5, 232–241 (2004). 80. Linder, P. Dead-box proteins: a family affair - active and passive players in RNP- remodeling. Nucleic Acids Res. 34, 4168–4180 (2006). 81. Linder, P. & Jankowsky, E. From unwinding to clamping - the DEAD box RNA helicase family. Nat. Rev. Mol. Cell Biol. 12, 505–516 (2011). 82. Tu, Y. T. & Barrientos, A. The human mitochondrial DEAD-box protein DDX28 resides in RNA granules and functions in mitoribosome assembly. Cell Rep. 10, 854–864 (2015). 83. Antonicka, H. & Shoubridge, E. A. Mitochondrial RNA granules are centers for posttranscriptional RNA processing and ribosome biogenesis. Cell Rep. 10, 920–932 (2015). 84. Shih, J.-W., Tsai, T.-Y., Chao, C.-H. & Lee, W. Candidate tumor suppressor DDX3 RNA helicase specifically represses cap-dependent translation by acting as an eIF4E inhibitory protein. Oncogene 27, 700–714 (2008). 85. Soto-Rifo, R. et al. DEAD-box protein DDX3 associates with eIF4F to promote translation of selected mRNAs. EMBO J. 31, 3745–3756 (2012). 86. De Vries, S. et al. Identification of DEAD-box RNA helicase 6 (DDX6) as a cellular modulator of vascular endothelial growth factor expression under hypoxia. J. Biol. Chem. 288, 5815–5827 (2013). 87. Franovic, A. et al. Translational up-regulation of the EGFR by tumor hypoxia provides a nonmutational explanation for its overexpression in human cancer. Proc. Natl. Acad. Sci. 104, 13092–13097 (2007). 88. Okumura, F., Zou, W. & Zhang, D. E. ISG15 modification of the eIF4E cognate 4EHP enhances cap structure-binding activity of 4EHP. Genes Dev. 21, 255–260 (2007). 89. Valgardsdottir, R. & Prydz, H. Transport signals and transcription-dependent nuclear localization of the putative DEAD-box helicase MDDX28. J. Biol. Chem. 278, 21146– 21154 (2003). 90. Mašek, T., Valášek, L. & Pospíšek, M. in RNA: Methods and Protocols (ed. Nielsen, H.) 293–309 (Humana Press, 2011). doi:10.1007/978-1-59745-248-9_20 91. Bhat, M. et al. Targeting the translation machinery in cancer. Nature Reviews Drug Discovery 14, 261–278 (2015). 92. Dowling, R. J. O. et al. mTORC-mediated cell proliferation, but not cell growth, controlled by the 4E-BPs. Science 328, 1172–1177 (2010). 93. Baskar, R., Lee, K. A., Yeo, R. & Yeoh, K. W. Cancer and radiation therapy: current advances and future directions. Int. J. Med. Sci. 9, 193–199 (2012). 94. Franovic, A., Holterman, C. E., Payette, J. & Lee, S. Human cancers converge at the HIF- 2α oncogenic axis. Proc. Natl. Acad. Sci. 106, 21306–21311 (2009).

95. Wiesener, M. S. et al. Induction of endothelial PAS domain protein-1 by hypoxia: characterization and comparison with hypoxia-inducible factor-1⍺. Blood 92, 2260–2268 (1998). 96. Clark, M. J. et al. U87MG decoded: the genomic sequence of a cytogenetically aberrant human cancer cell line. PLoS Genet. 6, (2010). 97. Porta, C., Paglino, C. & Mosca, A. Targeting PI3K/Akt/mTOR signaling in cancer. Front. Oncol. 4, 1–11 (2014). 98. Wee, P. & Wang, Z. Epidermal growth factor receptor cell proliferation signaling pathways. Cancers (Basel). 9, 1–45 (2017). 99. Yarden, Y. & Sliwkowski, M. X. Untangling the ErbB signalling network. Nat. Rev. Mol. Cell Biol. 2, 127–137 (2001). 100. Uhlen, M. et al. A pathology atlas of the human cancer transcriptome. Science 357, (2017). 101. Razafinjatovo, C. et al. Characterization of VHL missense mutations in sporadic clear cell renal cell carcinoma: hotspots, affected binding domains, functional impact on pVHL and therapeutic relevance. BMC Cancer 16, 1–12 (2016).