PTK7) in Cancer Development and Cellular Fitness

Characterizing the Role of Protein Tyrosine Kinase 7 (PTK7) in Cancer Development and Cellular Fitness

Sachin Anand Kumar

A thesis submitted in conformity with the requirements for the degree of Master of Science Department of Molecular Genetics University of Toronto

Characterizing the Role of Protein Tyrosine Kinase 7 (PTK7) in Cancer Development and Cellular Fitness

Sachin Anand Kumar

Master of Science

Department of Molecular Genetics University of Toronto

2015

Abstract

Protein Tyrosine Kinase 7 has been recently affiliated with the development and progression of various cancers. Contradictory results about its expression, localization and processing across various cancer types has left a major gap in understanding its molecular mechanism and function. In my thesis, I utilized large data sets from essentiality screens and RNAseq, as well as antibody-related tools, to investigate the role of PTK7 in cellular fitness. I observed a negative correlation between PTK7 processing and its degree of essentiality. Furthermore, I demonstrated that the PTK7 intracellular domain (ICD) is integral to its function. Treatment of mouse xenografts with PTK7 antibodies, capable of blocking PTK7 processing, inhibited tumour growth in an ICD-expression dependent manner. Lastly, mRNA profiling of PTK7-knockdown cells and ICD over-expression cells provided a list of putative downstream gene targets under

PTK7 regulation. Collectively, these findings implicate PTK7 in cellular transformation and provide insight into its signaling mechanism.

Acknowledgments

First and foremost, I would like to thank my supervisor Dr. Jason Moffat. You gave me an opportunity to step outside of my comfort zone and partake in research that pushed the boundary of science, technology and discovery. Your mentorship and creativity have been an inspiration to me – teaching me how to be a critical scientist, think outside of the box and perseverant throughout the hardships of research. Thank you for always being candid with me about my capabilities, strengths and weaknesses and pushing me to achieve more than I knew I was capable of. From you I have learned the tools and foundation I need to pursue my dream career as a physician scientist.

To my committee members Dr. Jeff Wrana and Dr. Igor Stagljar, thank you for your guidance, wealth of knowledge and constructive criticism. The rigorous perspective I have developed in my scientific questioning and approach were an integral part of my success, and will only flourish going forward.

Thank you to my colleagues in the Moffat Lab, past and present. To Andrea, thank you for setting a rigorous example of what scientific questioning should be. Regardless of the ups and downs over the past years, what I have learned and taken away from this experience is invaluable. A special thank you to Peter Xu, who kept me motivated when the multi-day experiments failed and provided insight when I was out of ideas. The late night hours of lab work wouldn’t have been the same. To Megha, Kristin, Michael, Taras, Carly, Hayoung, Clara and Zvezden, thank you for providing countless memories and hilarious discussions – these are the moments that make graduate school a truly wonderful experience. Kevin and Traver, you always indulged my curiosity and pushed me to learn coding to test every single theory I had – you’ve made me a better scientist. Last but not least, to Patti and Christine, thank you for keeping our lab functioning. Whether ordering things last minute or rolling up your sleeves to help us get stuff done, I truly appreciate your commitment to our success and the wealth of knowledge you both have about pretty much every technique ever.

To my friends and supporters (you know who you are), thank you for keeping my spirits high. Research finds unique ways to push you and test your limits. You were always there to help me push back. Last, but certainly not least, thank you to my family. To Mom & Dad, you never

III

really understood what I was working on, but let me bore you with the details anyways because you knew how passionate I was. That support is a huge factor in making me who I am, and where I am, today. To my brother Rishi, you always knew I could achieve anything I put my mind to. Some days that’s the only thing that kept me going. Thank you.

Table of Contents

Acknowledgments ...... III

Table of Contents ...... V

List of Tables ...... VII

List of Figures ...... VIII

List of Abbreviations ...... X

Chapter 1 Introduction ...... 1

1.1 Signaling and activation of receptor tyrosine kinases ...... 1

1.2 Receptor tyrosine kinases in the development and progression of cancer ...... 5

1.3 Screening for genetic-based fitness defects in cancer ...... 6

1.4 Protein tyrosine kinase 7 (PTK7) ...... 10

1.4.1 PTK7 in development and Wnt signaling ...... 11

1.4.2 Post-translational processing of PTK7 ...... 13

1.4.3 PTK7 as a prognostic biomarker in cancer ...... 15

1.5 Thesis objectives ...... 15

Chapter 2 The intracellular domain of PTK7 is integral for proliferation ...... 17

2 Acknowledgements ...... 17

2.1 Materials and Methods ...... 17

2.1.1 Cell Culture ...... 17

2.1.2 Effect of PTK7 knockdown on proliferation ...... 18

2.1.3 Western Blot ...... 19

2.1.4 Over-expression of PTK7 ...... 20

2.1.5 Anchorage independent growth ...... 20

2.1.6 Immunofluorescence of PTK7 fragments ...... 21

2.1.7 ELISA and Cytometric Bead Array ...... 21

2.1.8 Tumour xenografts and antibody treatment ...... 23

2.1.9 Cleavage assessment of recombinant PTK7 protein ...... 23

2.1.10 Cellular RNA isolation and sequencing ...... 24

2.2 Results ...... 25

2.2.1 PTK7 knockdown negatively affects cancer cell fitness ...... 25

2.2.2 PTK7 full-length rescues this proliferation defect ...... 28

2.2.3 PTK7 intracellular domain localizes to the nucleus ...... 30

2.2.4 PTK7 intracellular domain is sufficient to partially rescue the fitness defect observed upon knockdown ...... 32

2.2.5 Processing of PTK7 negatively-correlates to its essentiality ...... 32

2.2.6 PTK7 antibodies demonstrate tumour growth inhibition potential in vivo ...... 34

2.2.7 PTK7 intracellular domain rescues the antibody effect in vivoError! Bookmark not defined.

2.2.8 Antibodies block cleavage of PTK7 in vitro and in vivo ...... 36

2.2.9 RNAseq of PTK7 knockdown and ICD over-expression identify putative targets for PTK7 associated function ...... 39

2.3 Discussion ...... 43

Chapter 3 Summary and Future Directions ...... 49

3 Summary ...... 49

3.1 Future Directions ...... 50

3.1.1 Determining protein interactors and transcriptional targets of the PTK7 intracellular domain ...... 50

3.1.2 Examining the role of PTK7 in cell lines where knockdown increases proliferation ...... 50

3.1.3 Elucidating the role of the shed N-terminus in affecting tumorigenesis ...... 51

References ...... 52

Appendix 1 – Complete RNAseq Data ...... 59

List of Tables

Table 1: List of cell culture conditions used for each cell line…………………………………..27

Table 2: List of plasmid and shRNA sequences…………………………………………………28

Table 3: Study arms of the PTK7 mouse xenograft experiment…………………………………32

Table 4: Top 50 genes with mRNA expression changes upon PTK7 knockdown………………52

Table 5: Top 25 genes based on expression change and correlation with PTK7 essentiality…...54

Table 6: Top 25 genes anti-correlated with PTK7 essentiality…………………………………..56

VII

List of Figures

Figure 1: Schematic of human receptor tyrosine kinase family (Lemmon et al., 2010)………..12

Figure 2: Methods of ligand-mediate RTK dimerization (Lemmon et al., 2010)…...………….13

Figure 3: Small molecule inhibitors affect cancer cell survival and proliferation (Takeuchi & Ito, 2011)…………………………………………………………………………………………….16

Figure 4: Daisy model of gene essentiality (Hart et al., 2013)………………….…...………….17

Figure 5: Schematic of pooled lentiviral shRNA dropout screen protocol……...…...………….18

Figure 6: Protein tyrosine kinase 7 is essential in a variety of primary tumour-derived cell lines………………………………………………………………………………………………19 Figure 7: Three branches of Wnt signaling in cells (Kimoya & Habas, 2008)………………….22

Figure 8: Post-translational processing sites of the PTK7 extracellular domain………………...23

Figure 9: Schematic of methods to detect PTK7 extracellular domain shedding………………..31

Figure 10: Proliferation defect demonstrated by knockdown of PTK7 in HCT116 cell lines…..35

Figure 11: Crystal violet staining of PTK7 knockdown HCT116 TP53R248W/- cells validates fitness defect……………………………………………………………………………………..36

Figure 12: Validation of Bayes factor scores upon PTK7 knockdown across various cancer subtypes…………………………………………………………………………………………..37

Figure 13: Over-expression of PTK7-V5 partially rescues proliferation defect by PTK7 hairpins….………………………………………………………………………………………..38

Figure 14: Anchorage independent growth of HCT116 cells in the presence of PTK7 over- expression………………………………………………………………………………………..40

Figure 15: Localization of PTK7 processed fragments using EGFP tagged truncations………..41

VIII

Figure 16: Over-expression of the PTK7 intracellular domain is sufficient to rescue the proliferation defect observed upon knockdown……………………………………………….....43

Figure 17: Quantification of shed PTK7-ECD reveals a negative correlation with essentiality...45

Figure 18: Treatment of HCT116 subcutaneous xenografts with PTK7 antibody inhibits tumour growth……………………………………………………………………………………………47

Figure 19: Xenografts over-expressing PTK7 ICD negate antibody-induced tumour growth inhibition…………………………………………………………………………………………48

Figure 20: PTK7 IgGs are capable of blocking cleavage of ECD-Fc by MT1-MMP…………...49

Figure 21: RNA sequencing of PTK7 knockdown in HCT116 cell lines reveals novel downstream changes……………………………………………………………………………..51

List of Abbreviations

Acronym Meaning ADAM17 A disintegrin and metalloproteinase 17 ATP Adenosine triphosphate BF Bayes factor BSA Bovine serum albumin CCK-4 Colon carcinoma kinase-4 DAPI 4’6-diamidino-2-phenylindole DNA Deoxyribonucleic acid DTT Dithiothreitol Dvl Dishevelled ECD Extra-cellular domain EDTA Ethylenediaminetetraacetic acid EGFP Enhanced green fluorescent protein EGFR Epidermal growth factor receptor ELISA Enzyme-linked immunosorbent assay ENU N-ethyl-N-nitrosourea Eph Ephrin ErbB Erythroblastic leukemia viral oncogene family ErbB3 Epidermal growth factor receptor 3 FACS Fluorescence activated cell sorting FBS Fetal bovine serum FGFR Fibroblast growth factor receptor Fzd Frizzled receptor GARP Gene activity ranking profile HEK Human embryonic kidney HRP Horseradish peroxidase ICD Intra-cellular domain IGF1R Insulin growth factor I receptor InsR Insulin receptor KIT v-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog KLG Chick kinase-like gene mAb Monoclonal antibody MET MET proto-oncogene MT1-MMP Membrane-type 1 metalloproteinase Otk Off-track PBS Phosphate buffered saline PCP Planar cell polarity PKCδ1 Protein kinase C delta 1 PTK7 Protein tyrosine kinase 7 qzBF Quantile normalized Bayes factor RACK1 Receptor of activated protein kinase C 1 RFP Red fluorescent protein RIPA Radio immunoprecipitation assay

RNA Ribonucleic acid RNAi Ribonucleic acid interference ROR Receptor tyrosine-kinase like orphan receptor RTK Receptor tyrosine kinase RYK Receptor tyrosine-kinase like receptor SDS-PAGE Sodium dodecylsuphate polyacrylamide gel electrophoresis shRNA Short-hairpin ribonucleic acid TCF/LEF T-cell factor/lymphoid enhancer-binding factor TGI Tumour growth inhibition TK Tyrosine kinase TRC The RNAi Consortium Wnt Wingless/Integration-1 ligand

Chapter 1 Introduction

Cancer is a disease of uncontrolled cell growth and proliferation triggered by cellular transformation. Mechanisms are constantly at play within the cell to tightly regulate cell cycle progression, apoptosis, metabolism and survival to maintain healthy homeostasis1. However, aberrations such as mutation, deletion or amplification of key regulatory genes within these pathways position the cell to become cancerous.

As the landscape of cancer research has evolved over the past three decades, characterizing novel biomarkers associated with initiation and progression remains of utmost importance. It is fundamental to cancer biology to elucidate the role of these oncogenes and tumour suppressor genes in the context of downstream signaling and carcinogenesis. This thesis explores and investigates the functional role of PTK7 in cancer. Using an approach of high-throughput analysis concurrently with tool development for molecular characterization it was possible to elucidate downstream signaling and function.

1.1 Signaling and activation of receptor tyrosine kinases

Receptor tyrosine kinases (RTKs) are a subclass of ligand-activated, transmembrane cellular kinases2. Essential for converting extra-cellular stimuli into cellular responses, RTKs serve diverse functions in regulating fundamental processes related to cell growth and survival2. The first RTK to be identified was epidermal growth factor receptor (EGFR), discovered by Cohen et al., after observing increased phosphorylation upon treatment with epidermal growth factor in epidermoid cancer cells3. It was hypothesized that membrane receptors capable of binding extracellular ligands could initiate phosphorylation to activate intracellular effectors for downstream signaling. Studies over the next few decades established tyrosine phosphorylation in these RTKs as a method of rapid cellular signaling and amplification4-6.

Humans have 58 known RTKs that are further classified into 20 subfamilies (Figure 1) based on their structure7. Each RTK contains an intracellular tyrosine-kinase (TK) domain, a regulatory juxtamembrane region, a single-pass α-helix transmembrane region and a ligand-binding extracellular domain7. This general protein architecture of RTKs is highly conserved across vertebrate species, supporting a role for this family of proteins in fundamental cellular processes.

Activation of signaling by RTKs begins with a ligand binding the extracellular domain. Although it has been hypothesized that all RTKs have a ligand, there are several RTKs where ligands have yet to be identified. One such example is Protein Tyrosine Kinase-7 (PTK7) which to-date has no known ligands; it remains unclear if a ligand is yet to be discovered or if PTK7 may act as a

structural co-receptor8. Ligand binding typically induces receptor dimerization, bringing together the TK domains in an active conformation, allowing for trans-phosphorylation and activation7. The phosphorylated TK domain acts as a scaffold for assembly of an array of cytoplasmic effectors for activation of downstream signaling9. In some instances, receptors exist on the cell surface as oligomers, such as InsR or IGF1R. In these contexts, binding of the ligand triggers structural changes that permit TK activity within the dimer10.

Receptor dimerization upon ligand binding can occur through various binding interfaces. In some instances, a bivalent ligand (nerve growth factor) is able to crosslink two receptor molecules together, without any direct interaction between the receptors (Figure 2A). Other complexes such as KIT or FGFR form structures with shared ligand and receptor interfaces (Figure 2B-C). In the case of the ErbB family, receptor dimerization is orchestrated entirely by

the receptor where upon ligand binding the DII domain undergoes a conformational change making available a previously occluded interaction site (Figure 2D).

The intracellular TK domain is comprised of an amino- and carboxy-lobe, as well as the enzymatic active site11. The tyrosine-containing substrate binds to the carboxy-lobe, while the adenosine triphosphate (ATP) donor binds to the cleft created between both lobe structures12. Various residues are conserved within the TK domain that are essential for its catalytic activity. An aspartic acid is required in the active site for the transfer of a phosphate group to the substrate13. Additionally, a conserved DFG domain in the carboxy-lobe is required for binding and orientating the Υ-phosphate of ATP during phosphorylation14. Certain kinases have atypical structure or capacity, but are still members of the RTK family. Members of the RTK family including PTK7, human epidermal growth factor receptor 3 (ErbB3), receptor-like tyrosine kinase (RYK), receptor tyrosine-kinase like orphan 1 (ROR1) and ephrin B6 (EPHB6) lack catalytic activity, but still participate in signal transduction and activation15.

Using various methods of cis-autoinhibition unique to each receptor, the cell is able to regulate and avoid constitutive tyrosine kinase activity. For example, the insulin receptor (InsR) is autoinihibited by an activation loop. A key tyrosine residue in the loop (Y1162) projects into the active site of the TK domain, blocking accessibility to both the ATP donor and tyrosine substrate. Upon insulin binding and receptor dimerization, the Y1162 residue is trans- phosphorylated by the partner TK inducing a conformational change to the active state, freeing it for substrate binding and activity11. Phosphorylation of the activation loop is critical for regulating most kinases by further stabilizing the active conformation of the TK domain. Some kinases are also cis-autoinhibited by tyrosine residues within the juxtamembrane region outside of the TK domain. Tyrosine residues stabilize the activation loop in an autoinhibitory conformation. Receptors such as KIT and Eph family members use this form of regulation, where upon dimerization the tyrosine residues within the juxtamembrane region are phosphorylated inducing a conformational change that promotes a stable activation state16. Similarly, the c-terminal tail of certain RTKs like Tie2 can stabilize the inactive conformation by interacting with the active site17. Collectively, these three mechanisms demonstrate the complex and crucial role of efficient RTK regulation.

1.2 Receptor tyrosine kinases in the development and progression of cancer

As a central regulator of many essential processes within the cell, strict RTK regulation is of utmost importance. However, many diseases are characterized by aberrant RTK signaling, including but not limited to gain-of-function mutations that render the kinase constitutively active, gene amplification and abnormal paracrine or autocrine secretion of ligands18. Disruption of these integral cell functions creates an oncogenic phenotype, which is why RTK dysregulation has been linked with various types of cancer. Aberrations in RTKs provide cells with transforming potential; somatic and germline mutations that are associated with cancer development have been found in at least ten of the RTK subfamilies19. Mutations or deletions in regulatory regions result in a constitutively active RTK. In some instances, mutations in the transmembrane region, such as in ErbB2, can render the kinase active independent of ligand binding20.

Although cancer cells possess various genetic abnormalities, it is possible that their survival is predicated on a specific oncogene that promotes cell survival and proliferation. This principle is called oncogene addiction and is the basis for successful targeted therapeutics against RTKs over the past two decades18. For example, ErbB2-positive breast cancers are highly dependent on the activity of ErbB2, and direct targeting of ErB2 via the monoclonal antibodies (mAbs) Trastuzumab or Pertuzumab has emerged as an effective treatment21. Similarly, EGFR is over- expressed in a variety of solid tumours where it acts as a critical driver of oncogenic transformation. The mAb Cetuximab targets EGFR and has been effective at improving clinical outcomes in patients with metastatic non-small-cell lung cancer among other cancers22,23. Other tyrosine kinase inhibitors (both small molecules and antibodies) have also emerged over the past decade against receptors such as MET, FGFR, and IGF1R which disrupt proliferation and survival signaling (Figure 3)18. Tyrosine-kinase inhibitors are often successful when the target is responsible for regulating many critical signaling pathways18.

Deregulation of RTKs in a majority of the subfamilies is associated with cancer development and poor prognosis. Better understanding and characterization of RTKs, and their downstream signal effectors provides a basis for successful development of targeted therapies. Over the past two- decades, the generation of antibodies and small molecules against cell surface markers has had a

positive impact on improving clinical outcomes. However, aside from Her3, few small molecules or antibodies have been developed to target kinase dead RTKs21,24. The value of targeting this population of atypical receptors remains to be further investigated.

1.3 Screening for genetic-based fitness defects in cancer

Analysis of cancer genetics through a variety of functional genomics and proteomics approaches has begun to provide mechanistic insight into oncogenic cell biology. As a consequence of subverting normal cell growth processes, cancer cells begin to accumulate novel vulnerabilities that are unique to their genetic state. This genetic state can be exploited to cause a reduction in cellular fitness or growth, which we loosely define as genetic essentiality25. It is important to note that for the purposes of my investigation, I will be referring to essential genes as any gene

that negatively affects cellular fitness or growth when perturbed, as opposed to genes that exclusively trigger lethality. We believe that essentiality is not a binary feature of cells and that most genes are neither absolutely essential nor absolutely non-essential. Instead, a quantitative measure of their contribution to a cell’s growth defines their level of essentiality26. By investigating these cellular sensitivities in comparison to healthy cells, it is possible to elucidate context-specific essential genes as well as core essential genes. The daisy model of gene essentiality states that each cell-line or tissue provides a unique context where certain genes may be essential26. However, amongst these cell lines there is an overlap of genes that are ubiquitously expressed and ubiquitously essential across all cell types. These are the core essential genes, often housekeeping genes that are fundamental to cellular processes such as splicing, transcription and translation, to name a few26. In contrast, context-dependent genes are genes that are differentially essential based on the entirety of the cell’s genetic background (Figure 4).

Using a high-throughput RNA interference (RNAi) approach, it is possible to identify essential genes across a variety of cancer contexts27. Characterizing these genetic interactions provides three major insights into cell biology. The first is a list of core essential genes, which are important for basal cell function across multiple lines. New genes in this category will indicate some putative housekeeping role for genes of unknown function. Secondly, screening will provide a list of novel driver genes important for cancer development and progression. Identifying these biomarkers establishes a prioritized list of genes or complexes that may play a role in tumorigenesis, especially those without any current cancer related annotations.

Investigating these genes may lead to putative therapies to treat malignancies. Lastly, this data provides information regarding the appropriate cell-line models to investigate certain genes or signaling pathways. Cell lines extremely sensitive or resistant to knockdown of a gene can be compared to further elucidate molecular-genetic differences, as well as phenotypic distinctions between the two groups. Collectively, this approach is highly effective at generating genetic- interaction networks and has been effectively used over the past decade25,27-29.

The lentiviral-based TRC libraries were developed for high-throughput genetics in human and murine cells27. In recent years, multiple primary tumour derived cell lines have been generated and screened for cancer cell-line specific gene essentiality25. Briefly, a pooled library of 80,000 hairpins, with five-fold coverage of 16,000 human protein-coding genes, was used to screen 72 primary tumour derived cell lines from four different subtypes – breast, colon, pancreatic and ovarian (Figure 5). Relative hairpin prevalence was taken over a time course, and genes were ranked for essentiality based on their rate of hairpin drop out, as well as consistency across multiple hairpins (i.e. GARP score).

By tracking hairpin abundance, one can generate a confidence score of whether a gene is essential or not in a specific cell line. Subsequently, a Bayesian classifier was developed using training sets of essential and non-essential genes to establish a quality score for screens and an improved gene essentiality score called the Bayes Factor (BF) score26. The BF provides us a confidence indicator of a gene’s essentiality in a cell line, with scores greater than two being hits of very high confidence26.

From this analysis emerged a list of putative cancer driver genes which when perturbed could negatively affect cellular fitness. One candidate that emerged as widely varying in essentiality across all of the cancer subtypes was Protein Tyrosine Kinase 7 (Figure 6). PTK7 is a poorly characterized RTK and a pseudokinase with very little known about its downstream signaling function. Thus, PTK7 represented an ideal candidate for further investigation and we set out to use the data generated from the RNAi screens, including the list of sensitive or resistant cell lines

to PTK7 perturbation, to elucidate the molecular function of PTK7 and the role it plays in promoting cellular fitness.

1.4 Protein tyrosine kinase 7 (PTK7)

Protein tyrosine kinase 7 is also known as colon carcinoma kinase 4 (CCK-4) and is a single-pass transmembrane protein that is part of the receptor tyrosine kinase family. PTK7 was discovered over 2 decades ago as a highly expressed protein in colon adenocarcinoma tissue when compared to normal adjacent tissue; its molecular function and role in cancer development remain unclear to this day30. Located on chromosome 6p21.1, the gene contains 21 exons and the full-length isoform is 1070 amino acids in length31. There are 6 isoforms of PTK7, with isoforms 2 through 5 missing exons but remaining in frame, resulting in truncated proteins15. Isoform 6 specifically initiates translation at an alternate start codon producing a longer protein with a distinct N- terminus15. With 72% sequence identity to the chick kinase-like gene (KLG), and orthologs in many species such as off-track (otk) in Drosophila or Lemon in Hydra, PTK7 represents its own unique subfamily of the RTKs32-34. Structurally PTK7 contains an N-terminal secretion signal, seven Ig-like loops, a transmembrane region, and a catalytically inactive cytoplasmic TK

domain. The TK domain is missing key conserved residues, mainly a missense mutation in the DFG motif required for correctly orienting the donor ATP during phosphorylation15,30,32.

1.4.1 PTK7 in development and Wnt signaling

Since its discovery over 2 decades ago, multiple studies have demonstrated a role of PTK7 and its orthologs in development and morphogenesis. Tracking PTK7 mRNA expression during mouse embryo development revealed a dynamic localization pattern suggesting multiple roles in development of the tail, limbs, somites, gut and spinal cord15. In Drosophila, otk is required for axon guidance and vertebrates rely on PTK7 for cardiac morphogenesis35,36. Mice with null mutations in PTK7 display severe defects in neural tube closure, neural crest cell migration as well as polarity defects of the inner ear36-38. Furthermore, PTK7 is necessary for convergent extension during gastrulation in Xenopus, zebrafish and mice and the deletion of PTK7 in mice results in an embryonic lethal phenotype36,39.

PTK7 acts as a regulator of development through the wingless/integration-1 (Wnt) signaling pathway. Wnt signaling is a highly conserved pathway capable of transducing cues across cells and tissues to regulate proliferation, migration, polarity, and cell fate determination40. The pathway is characterized by the secretion of active Wnt ligands, a group of lipid-modified glycoproteins that are bound by their primary receptors, the Frizzled (Fzd) family41,42. Wnt binding and downstream signaling can be broadly separated into three distinct branches. First, the canonical Wnt pathway revolves around β-catenin stability and transcriptional activation of genes affecting cell growth, stemness and proliferation42. Second, the Wnt/Ca2+ pathway functions by stimulating intracellular calcium release via trimeric G- proteins38,43. This orchestrated calcium ion release is essential for patterning during early gastrulation and embryonic pattern formation43. Third, is the planar cell polarity (PCP) pathway, which is essential during tissue and organ development. Epithelial cells can achieve a polarity that is orthogonal to their apical-basal axis. PCP is the polarization of these cells or structures providing positional information across the plane to enable structural orientation. Common examples of PCP include the orientation of hairs on Drosophila wings, patterning of mammalian fur, arrangement of stereocilia within the inner ear and convergent extension within vertebrates44. PCP signaling is dependent on the asymmetrical organization of the Frizzled-Dishevelled-Diego and Vangl-Prickle axes, along with cadherin EGF LAG seven-pass transmembrane G-type

receptor (CELSR)45. Intracellular signaling through small GTPases results in changes in cell movement and actin cytoskeleton rearrangement40,46. A schematic of these branches of Wnt signaling, detailing the molecular effectors of downstream signaling are shown in Figure 740,46.

Although the molecular mechanisms of canonical and non-canonical Wnt signaling are well characterized, there remains a lot to be understood about the specification of downstream signals by receptor complexes. In humans, there are 19 Wnt ligands and 10 Fzd receptors41. Furthermore, there exists a group of Wnt co-receptors, including PTK7, receptor tyrosine kinase- like orphan receptor (ROR1, ROR2), related-to-receptor tyrosine kinase (RYK) and lipoprotein- like-receptor related protein (LRP5, LRP6)36. It is hypothesized that these receptors and co- receptors act as molecular switches to control specific downstream Wnt signaling responses, the specificities of which remain largely unstudied.

Much of the mechanism for PTK7-Wnt signaling is unknown. Recent studies have implicated PTK7 as a molecular switch between both the canonical and non-canonical PCP pathways39,47,48. PTK7 assists in the recruitment of the Dishevelled (Dvl) protein to the plasma membrane, mediated by the receptor of activated protein kinase C (RACK1) and protein kinase C (PKCδ1)47. Additionally, PTK7 has been shown to directly impact canonical Wnt signaling39,47. Using a TOPflash luciferase-based reporter as a readout for transcriptional activity of TCF/LEF associated genes, PTK7 knockdown has been shown to both inhibit and activate canonical signaling in Xenopus and mammalian cells39,47. More recently, Hayes et al. demonstrated that maternal zygotic PTK7 mutants had increased canonical Wnt signaling, suggesting an inhibitory role for PTK748. Using exogenous Wnt8 and an eyeless phenotype associated with activated Wnt/β-catenin signaling, they demonstrated that PTK7 inhibits canonical signaling – a phenotype dependent on the presence of a tethered extra-cellular domain. Striking, the authors of this study found that full-length Ptk7 and Ptk7/dICD constructs had equivalent abilities to rescue maternal zygotic Ptk7 mutant defects, whereas deletion of the Ptk7 extracellular domain eliminated protein function48. Additionally, exogenous treatment with Wnt5 and Wnt11 in the presence of PTK7 enhanced PCP signaling suggesting a role for potentiation of non-canonical signaling. Although these findings implicate PTK7 as an inhibitor of Wnt/β-catenin signaling and activator of PCP signaling, the molecular mechanisms remain ill defined. The contrary findings about PTK7, and its central role in regulating two major branches of the Wnt signaling pathway, highlight the importance of further investigating its molecular function.

1.4.2 Post-translational processing of PTK7

Adding to the complexity of PTK7 biology are the numerous post-translational modifications that occur on PTK7. PTK7 is N-glycosylated at 10 sites, and sequentially cleaved by three different proteases15,49,50. Notably, shedding of RTKs is thought to be important for their function and regulation in some cases. For example, processing of Ephrins removes the ligand- binding fragment rendering the receptor signaling impaired51. Likewise, shedding of the vascular endothelial growth factor receptor VEGFR generates soluble receptors that saturate ligand to inhibit intracellular signaling52. For PTK7, the first processing step is cleavage of PTK7 by membrane-type 1 matrix metalloproteinase (MT1-MMP) in the seventh Ig-like loop (Figure 8)49. MT1-MMP cleaves at the consensus sequence PKP621LI releasing the N-terminal soluble fragment into the extracellular space49. A mutant construct containing a L622àD did not demonstrate processing by MT1-MMP over-expression, thereby confirming the cleavage site49. The shedding of the extracellular-domain (ECD) has been associated with collagen invasion and

knockdown of MT1-MMP or over-expression of the mutant PTK7L622D decreases PTK7 shedding leading to impaired collagen invasion49.

In colon cancer cells, processing of PTK7 was demonstrated by a disintegrin and metalloproteinase 17 (ADAM17) and subsequently by Y-secretase50. Using Edman-degradation of the C-terminal fragments released within the cell, it was possible to detect the exact cleavage sites for both proteases50. ADAM17 processes PTK7 at the base of the seventh Ig-loop at E689. The resulting small fragment embedded in the membrane is then a target for detection and processing by Y-secretase at G721 within the transmembrane region (Figure 8)50. Cleavage at these sites results in the release of two fragments containing the intracellular kinase domain (ICD). After full processing of PTK7, the ICD product was shown to traffic towards the nucleus in SW480 colon cancer cells50. Further analysis on the effect of over-expression of the ICD fragment showed an increase in anchorage independent growth and migration in SW480 cells50. More recently, a study was published that looked for PCP related defects in N-ethyl-N- nitrosourea (ENU) treated mouse embryos. A specific chuzhoi mutant of PTK7 was discovered, with severe embryonic lethal defects in convergent extension and neural tube closure53. The mutation provided a 9 amino acid insertion within the fifth Ig-loop of PTK753. The newly generated sequence, QVL – ANP – EKLK, produces a new MT1-MMP consensus sequence

PXXL that is processed by MT1-MMP54. The introduction of this novel cleavage site results in the shedding of an incomplete ECD fragment, and overall release of the C-terminal fragments54. Collectively these results implicate PTK7 processing as essential to its function in development and potentially oncogenesis. The exact mechanism downstream of this processing remains to be further investigated.

1.4.3 PTK7 as a prognostic biomarker in cancer

As PTK7 was originally discovered for its differential expression in colon carcinoma tissue when compared to normal adjacent colorectal tissue, it is not surprising that PTK7 has been associated with the development and progression of various cancers30. More surprising however, are the number of published studies that report opposing findings on the role of PTK7 as a tumour suppressor gene or oncogene8,55-61.

Clinically, PTK7 has been found as a biomarker of progressive cancers. Acute myeloid leukemia cells expressing PTK7 show increased resistance to anthracycline chemotherapy treatments, as well as pro-migratory and anti-apoptotic phenotypes8. In colorectal cancer, PTK7 over- expression is associated with pro-migratory and pro-metastatic phenotypes, resulting in poor clinical outcomes56. Similarly, in breast cancer PTK7 has been used as a predictive marker for poor prognosis and resistance to adjuvant chemotherapy57,62. Patients with PTK7 over-expression were more likely to develop nodal metastasis, and predicted shorter disease-free survival58.

On the contrary, many papers have shown a clinical association between low PTK7 expression and poor prognosis, identifying it as a tumour suppressor gene. An analysis of 204 cases of ovarian cancer revealed that negative expression of PTK7 has a significantly poorer outcome compared to patients whose expression was high55. Additionally, in lung squamous cell carcinoma, PTK7 expression at the mRNA and protein level was low, and over-expression of PTK7 inhibited cell growth. This function was attributed to the inactivation of AKT and ERK signaling59. Further clinical studies have identified high PTK7 expression in gastric cancer as favourable for patient outcome61,63.

1.5 Thesis objectives

It is clear that PTK7 signaling and processing is complex. Our screening for PTK7 has identified it as a differentially essential gene across various cancer subtypes. Multiple studies have

implicated PTK7 in Wnt signaling, but the exact mechanism by which it specifies canonical or non-canonical signaling remains unclear. Moreover, contradictory studies have been published on its role for activating or inhibiting β-catenin mediated-signaling. Processing of PTK7 is clearly linked to its function, as demonstrated by the chuzhoi mutant and migration defects. However, the molecular mechanism by which the intracellular or extracellular fragments signal remains to be elucidated. Lastly, over-expression of PTK7 has opposing roles on patient prognosis dependent on the subtype of cancer. It is crucial to elucidate the role of PTK7 in cancer development, and specifically the molecular mechanism by which it’s processing regulates its downstream signaling effectors. The goal of this thesis is to provide insight into the role of PTK7 in cancer cell proliferation, as well as how enzymatic processing plays a role in its molecular function. Eventually, with the mechanism better understood, I aim to delineate downstream targets of PTK7, as well as the gene contexts that contribute to its oncogenic or tumour suppressing role.

Chapter 2 The intracellular domain of PTK7 is integral for proliferation

As PTK7 continues to become a well-established prognostic biomarker of various cancer subtypes, understanding its function in cancer development and progression is increasingly important. Developing a phenotypic profile of PTK7 based on cell proliferation, migration and transformation will provide key insights into its oncogenic function. Furthermore, characterizing the cleavage of PTK7 will help establish a mechanism for PTK7 regulated cell signaling – either as a secreted product or as an active cytoplasmic effector. Here I aim to investigate these aspects of PTK7 cell biology in the context of carcinogenesis.

2 Acknowledgements

The majority of the work done on the PTK7 project was done in collaboration with a postdoctoral fellow in the lab, Dr. Andrea Uetrecht. Dr. Traver Hart conducted analysis of data generated from the essentiality screens. Dr. Kevin Brown mapped and analyzed data collected from RNAseq of PTK7 knockdown and over-expression. The concepts and experimental designs for the project were a collaborative effort with Dr. Moffat and Dr. Uetrecht. Mrs. Patricia Mero provided protocols and assistance throughout the project. Flow cytometry was performed with the support of Dionne White. Xiaowei Wang performed all tumour xenografts and dissections. I, unless otherwise stated, completed all other work described herein.

2.1 Materials and Methods

2.1.1 Cell Culture

All cell lines were grown in culture according to their recommended conditions in their ATCC annotations (Table 1). Unless otherwise stated, all culture media was supplemented with 5% penicillin/streptomycin and 10% fetal bovine serum (FBS). Cells were passaged at with 0.25% trypsin-EDTA consistently at 80% confluence unless otherwise stated. Media collected for analysis was lifted from cells in culture and spun down to remove cell debris. Supernatants were transferred to Eppendorfs and frozen at -20oC for later testing.

Table 1: List of cell culture conditions used for each cell line. Name of Cell Line Culture Medium Tissue Origin HCT116 TP53R248W/- McCoy’s 5A Modified Media Colon HCT116 TP53-/- McCoy’s 5A Modified Media Colon MDA-MB-231 Dulbecco’s Modified Eagle Media (DMEM) Breast MDA-MB-157 DMEM Breast KP-4 DMEM Pancreas BT549 Roswell Park Memorial Institute (RPMI) Breast OVCAR3 RPMI Ovarian HPAC DMEM + 5% FBS Pancreas T47D RPMI Breast HCT116 McCoy’s 5A Modified Media Colon HT29 McCoy’s 5A Modified Media Colon SW480 DMEM + 20% FBS Colon MDA-MB-468 DMEM Breast MX1 RPMI + 5% FBS Breast MACLS2 RPMI Breast MiaPaca2 DMEM Pancreas Su.86.86 RPMI Pancreas HEK 293T DMEM Kidney

2.1.2 Effect of PTK7 knockdown on proliferation

Knockdown of PTK7 was performed using lentiviral short-hairpin RNAs (Table 2). All hairpins were expressed in the pLKO.1 vector containing a puromycin resistance cassette. Negative control shRNAs against LacZ or RFP were used for comparison. HEK 293T cells were co- transfected with lentiviral packaging plasmids pMD2.G and psPAX2. Briefly, 1.0 x 106 cells were seeded in a 6-well plate and left to attach overnight. The following day, cells were transfected with 1800ng of psPAX2, 200ng of pMD2.G, 2µg of the corresponding shRNA pLKO.1 plasmid with 10uL X-tremeGENE DNA Transfection Reagent, in 200µL of Opti- MEM® I Reduced Serum Media. 24 hours post-transfection, media was exchanged with 2mL of DMEM + 20% bovine serum albumin (BSA) for viral harvesting. Viral media was collected at 24 hours and 48 hours and frozen in 2mL aliquots at -80oC for further use. Fresh virus was used to infect cell lines to validate PTK7 fitness defects. Cells were tryspinized and counted to a concentration of 150,000/mL and were seeded in a 12-well plate concurrently with 8µg of polybrene and 200µL of virus. Conditions were performed in triplicate wells, along with a no puromycin triplicate to gauge viral titers. Replicate plates were infected for counting on day 1, day 3 and day 6 post-selection (Figure 10A). Counting was performed either manually by whole

cell counts via the Beckman Coulter Counter, stained for visualization and imaging by crystal violet, or measured with the automated Zoom Incucyte imager for confluence on an hourly basis.

Table 2: List of plasmids and shRNA Plasmid/shRNA/Primer Target Sequence/Product Number TRCN0000002210 LacZ CCGTCATAGCGATAACGAGTT TRCN0000002235 RFP CGTAATGCAGAAGAAGACCAT TRCN0000006431 PTK7 3’ UTR CACAGGGTTAATGAGTCTCTT TRCN0000006433 PTK7 CDS CCACAGCACAAGTGATAAGAT TRCN0000006434 PTK7 CDS CCTCATGTTCTACTGCAAGAA TRCN0000199565 PTK7 3’ UTR GCCACTCATCTGCCAACTTTG pMD2.G Lentiviral Packaging AddGene #12259 psPAX2 Lentiviral Packaging AddGene #12260 pLKO.1 shRNA Expression AddGene #10878 pLX304 Destination Gateway-V5 AddGene #25890 pLL3.7-PTK7-EGFP PTK7 ORF Produced by Dr. Andrea Uetrecht pLL3.7-ICD-EGFP PTK7 721-1070 Produced by Dr. Andrea Uetrecht pLV417G-H2B TurboGFP H2B ORF Provided by Dr. Craig Strathdee pLX304-PTK7-V5-Blast PTK7 ORF Produced by Sachin Kumar pLX304-ICD-V5-Blast PTK7 721-1070 Produced by Sachin Kumar pLX304-LacZ-V5-Blast LacZ Produced by Sachin Kumar pENTR223.1 Assorted Entries Invitrogen M13 Forward (-20) pLX304 GTAAAACGACGGCCAG M13 Reverse pLX304 CAGGAAACAGCTATGA

2.1.3 Western Blot

Western blots (WB) were performed using sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). Lysates were generated from cells in culture. Briefly, cells were gently washed with ice-cold phosphate buffered saline (PBS) to remove exogenous salts, protein and media. Radio Immunoprecipitation Assay Buffer (RIPA) was added to samples and adherent cells were scraped from the plate. RIPA buffer is composed of the following components: 150 mM sodium chloride, 1.0%Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS (sodium dodecyl sulphate), and 50 mM Tris maintained at pH 8.0. Protease and phosophatase inhibitors were supplemented in RIPA buffer at a final concentration of 2µg/mL. Whole lysates were transferred to an Eppendorf tube and left rotating at 4oC for 1 hour. Protein samples were centrifuged at 13,200rpm for 10 minutes to separate non-soluble cell debris and DNA. Supernatants were carefully transferred to new Eppendorfs and the pellets were discarded. Protein concentrations were measured by Bradford Assay and standardized to a final concentration of 2µg/uL in RIPA buffer and 4x NuPAGE LDS sample buffer. Dithiothreitol

(DTT) was added to samples that were run in a reduced condition, and samples were boiled at 90oC for 5 minutes. A total of 20µg of each sample was electrophoresed in a 10% Tris-glycine gel in a MOPS running buffer at 100V for 2 hours. Proteins were transferred to a polyvinylidene fluoride (PVDF) membrane, pre-activated in methanol for 2 minutes for 1 hour at 120V. After transfer, the SDS-PAGE gels were discarded and the membranes were blocked with 5% (BSA) in Tris-buffered saline with 0.5% Tween 20 (TBS-T) for 30 minutes. Primary antibodies against raised against the carboxy-terminus of PTK7 (Cell Signaling Technology, Cat #11926) were used at a concentration of 1:1000 in blocking solution overnight. Membranes were washed with TBS-T for 3 x 5 minute washes and incubated with secondary Goat anti Rabbit antibody, conjugated with Horseradish peroxidase (HRP) at a dilution of 1:5000 for one hour. Membranes were again washed with TBS-T (3x5min) and developed using SuperSignal PicoWest Enhanced ChemiLuminescence by Pierce and exposed to film for detection.

2.1.4 Over-expression of PTK7

PTK7 over-expression constructs containing EGFP tags were cloned by Dr. Uetrecht and protein expression was confirmed by Western blot. Cells were transduced and sorted using fluorescence activated cell sorting (FACS) using a FACSAria (BD Biosceinces) with assistance from Dionne White, director of the University of Toronto Flow Cytometry Facility. Over-expression constructs containing V5-tags were cloned using the gateway cloning system into pLX304 plasmids64. Cells were transduced with pLX304-derived lentiviruses and selected with blasticidin for 10 days (2µg/mL).

2.1.5 Anchorage independent growth

HCT116 TP53-/- and TP53R248W/- cell lines were plated on soft agarose layers to test for colony formation without adhesion. In 96-well plates, a bottom layer of 1% agarose was created followed by a top layer of 0.4% agarose. 60,000 cells were added in 2mL of McCoy’s medium to each well, and cells were observed for 72 hours. Brightfield images were taken at 40x magnification. Colony formation was counted from representative images from triplicate samples.

2.1.6 Immunofluorescence of PTK7 fragments

Cells over-expressing various constructs (section 2.1.4) were seeded onto glass coverslips in 12- well plates. After 24 hours, media was aspirated and coverslips were incubated in freezing cold - 20oC methanol for 10 minutes to fix cells. Carefully, coverslips were removed and permeablized in PBS + 0.01% Triton-X-100 for 15 minutes. EGFP-tagged constructs went straight to the counterstain step below, whereas cells with V5 tagged constructs were probed with a primary antibody at 1:1000 (ab9116) for one hour and subsequently washed gently with PBS + 0.01% Tween 20 (PBS-T). A secondary goat-anti-rabbit AlexaFluor488 secondary was used for visualization (Life Technologies, A2124, 1:1000). The coverslips were gently rinsed with PBS and incubated with 4,6-diamidino-2-phenylindole (DAPI) to counterstain nuclei (1:10,000) for 5 minutes. Cells were again rinsed gently in PBS-T and left to air dry. Dako immunofluorescence mounting media (Cat #S3023) was added dropwise onto a glass slide and the coverslip was gently flipped and overlayed avoiding any bubbles. Slides were imaged using a Quorum WaveFX spinning disc confocal microscope via Velocity and analyzed/coloured with ImageJ.

2.1.7 ELISA and Cytometric Bead Array

An enzyme-linked immunosorbent assay (ELISA) was used to quantify shed PTK7 from cell supernatants (Figure 9A). Briefly, a capture antibody E1 was used to coat the bottom of a high- binding 384-well plate at a concentration of 10µg/mL for 1 hour at room temperature. Plates were then blocked with 60µL of PBS-T + 2% BSA for 30 minutes to saturate the plate. Samples of recombinant PTK7 extracellular domain protein (obtained from the Toronto Recombinant Antibody Centre or TRAC) or cell culture supernatants were added to the wells in triplicate (50µL). Plates were thoroughly washed with PBS-T and a secondary Biotinylated Fab (antibody fragment containing variable region) called E6 was added to the plate at a concentration of 10µg/mL. After 1 hour, the plate was again rigorously washed with PBS-T, and a tertiary streptavidin-HRP (Sigma, S5512) was added at a concentration of 1:10,000 in PBS-T for 30 minutes. Plates were washed with PBS-T and colourized with the addition of the substrate 3,3’,5,5’-tetramethylbenzidine (TMB). The reaction was kept for 5 minutes and neutralized by the addition of H3PO4 phosphoric acid. Plates were read for absorbance at 450nM using a BioTek Synergy 2 plate reader.

To quantify shed PTK7 in larger volumes of supernatants, a cytometric bead array (CBA) was developed (Figure 9B). Fluorescent A4 beads were obtained from Beckton Dickinson Biosciences (558578, 558556) along with a functional bead conjugation set. E1 and E6 antibodies were covalently crosslinked to the beads using a Sulfo-SMCC and NEM conjugation. Using the provided Human Soluble Protein Master Buffer Set (558264), supernatants were incubated with antibody-conjugated beads for capture for 30 minutes. Secondary biotinylated Fab targeting a unique epitope (E6 for E1 beads, and vice versa for E6 beads) was added at

5µg/mL. A tertiary Streptavidin-PE (BD Biosciences, 554061) was added and samples were washed twice in flow tubes. Samples were run on the BD FacsCalibur and 300 events were collected for analysis. All analysis was performed on FlowJo v9.7.1 and raw median fluorescence intensity (MFI) values were plotted.

2.1.8 Tumour xenografts and antibody treatment

For all experiments containing EGFP tags, pure populations of sorted cells over-expressing EGFP were used. HCT116 TP53-/- and TP53R248W cells were cultured in 15cm plates, checked for mycoplasma and harvested, counted and re-suspended in PBS for injection. A total of 120 severe combined immunodeficiency (SCID) mice were separated into 10 study arms (Table 3). Due to differences in in vivo growth rates of the TP53-/- and TP53R248W cell lines, cells were injected at 1.0x106 and 4.0x106 cells respectively. Cells were injected in PBS subcutaneously at a volume of 100µL and mice were carried forward for approximately 35 days. Measurements of tumour size/volume were calculated every three to four days until sacrifice. Antibody treatment was given in PBS by intraperitoneal injection at a concentration of 30mg/kg starting five days after implantation. Tumours were dissected upon sacrifice, and separated for protein lysate and histology analysis. The TP53-/- cells over-expressing H2B-GFP as a control were not used for injections due to an unidentified fungal contamination days prior to injection.

Table 3: Study arms of the PTK7 mouse xenograft experiment Cell-Line Genetic Background Treatment HCT116 TP53-/- Wild-type PBS HCT116 TP53-/- Wild-type 30mg/kg PTK7-E11 Antibody HCT116 TP53-/- PTK7 ICD-EGFP over-expressing PBS HCT116 TP53-/- PTK7 ICD-EGFP over-expressing 30mg/kg PTK7-E11 Antibody HCT116 TP53R248W Wild-type PBS HCT116 TP53R248W Wild-type 30mg/kg PTK7-E11 Antibody HCT116 TP53R248W PTK7 ICD-EGFP over-expressing PBS HCT116 TP53R248W PTK7 ICD-EGFP over-expressing 30mg/kg PTK7-E11 Antibody HCT116 TP53R248W H2B-EGFP over-expressing 30mg/kg PTK7-E11 Antibody

2.1.9 Cleavage assessment of recombinant PTK7 protein

To assess the in vitro cleavage capacity of MT1-MMP and other proteases, recombinant enzymes were used. The PTK7 extracellular domain (PTK7-ECD) was Fc tagged and purified at the TRAC, containing sequence 1aa – 721aa. Recombinant MT1-MMP was purchased from R&D Systems (918-MP-010) along with its pro-enzyme activator rhFurin (1503-SE). Enzymatically

active MMP-2 and MMP-9 were purchased from EMD Millipore (PF-023, PF-024 respectively). MT1-MMP was activated according to the product sheet, and 200ng of enzyme was used to treat 2µg of PTK7-ECD. Similarly, 150 ng of MMP-2 and MMP-9 were used to cleave recombinant PTK7-ECD. The reactions proceeded for 2 hours at 37oC. The entire reaction volume was loaded into an SDS-PAGE gel in non-reducing conditions following addition of 4x NuPAGE LDS buffer, and ran on a 12%-4% Bis-Tris gradient gel at 100V for 2 hours. The gel was incubated with Coomassie Blue stain (1g Coomassie Brilliant Blue, 50% methanol, 40% H2O, 10% acetic acid) for 1 minute in the microwave and then destained overnight (20% methanol, 10% acetic acid, 70% H2O). Antibody treatments to block cleavage were done by pre-incubation of the PTK7-ECD for 15 minutes at room temperature (2µg/mL), while maintaining the remainder of the protocol consistent with that described above. To confirm the enzymatic activity of MT1- MMP, a fluorogenic peptide was used, as provided by R&D Systems (ES010).

2.1.10 Cellular RNA isolation and sequencing

HCT116 cell lines were treated with shRNAs and placed under puromycin selection after 24 hours. Within 48 hours of infection, cells were harvested for RNA isolation. Briefly, 10 cm plates were grown until approximately 80% confluence. Cells were washed twice with cold PBS and a total of 2mL of TRIzol® Reagent (Life Technologies) was added to the plate. Cells were scraped and transferred to 15mL conical tubes. Tubes were incubated at room temperature for 5 minutes and 0.4mL of chloroform was then added. Tubes were then shaken and centrifuged at 12,000g for 15 minutes at 4oC. After phase separation, the top aqueous phase was collected and transferred to a new Eppendorf tube. Samples were washed with 1mL of isopropanol to precipitate RNA and centrifuged at 12,000g for 10 minutes at 4oC. The resulting pellet was washed twice with 75% ethanol and centrifuged at 7500g for 5 minutes at 4oC. Samples were allowed to air dry and then resuspended in diethylpyrocarbonate (DEPC) treated RNase free water. The NanoDrop Spectrometer was used to determine the purity and concentration of all RNA samples. Sequencing libraries were prepared using the Illumina TruSeq mRNA v2 kit and sequencing was performed on the Illumina HiSeq 2500. Dr. Kevin Brown performed mapping and genomic analysis of the resulting sequence files. Briefly, genomic alignment was performed using the STAR aligner (v2.3.0)65. Default STAR parameters were chosen, except – outSAMstrandField was set to intronMotif. Reads were aligned to the NCBI Build 36 reference human genome, using Gencode V19 transcript models. Gene expression levels were estimated

using Cufflinks (v.2.2.1) using the default parameters and Gencode V19 GTF file66. All resulting cufflinks output files were merged using a bespoke script written in R (v.3.0.3). A total of 6 samples were tested for each cell line. Controls included shRFP infected cells, as well as uninfected sparse and dense isolates. Knockdown samples included three hairpins targeting PTK7 – sh6431, sh6433, sh199565 (Table 2).

2.2 Results

2.2.1 PTK7 knockdown negatively affects cancer cell fitness

PTK7 was found to be required for cell proliferation across several different cancer cell lines in a large-scale screening study led by the Moffat lab (Figure 6; BF > 2). Due to the fact that the screens were performed in a high-throughput fashion, my first objective was to validate the growth defect across multiple cell lines. I chose two sensitive isogenic colorectal cancer cell lines, HCT116 TP53-/- and TP53R248W, to validate the results of the high-throughput screens. Utilizing three unique shRNAs towards PTK7 including PTK7-sh6431, PTK7-sh6433, and PTK7-sh199565, as well as a control shRFP hairpin, I performed proliferation assays by counting cell numbers on days 1, 3 and 6 following transduction with (Figure 10A). Both cell lines demonstrated a significant reduction in cell numbers across all three hairpins (Figure 10B) compared to cells treated with the control shRFP hairpin (P<0.01). Validation was also done using crystal violet stain in TP53R248W cells (Figure 11) and knockdown of protein expression was confirmed by Western blot (Figure 10C). It is important to note that although protein level knockdown was equally effective across all three hairpins based on my Western blotting data (Figure 10C), not all shRNAs caused the same anti-proliferative effect in HCT116 cells (at a similar MOI), likely due to off-target effects67,68.

To assess the reliability of the screening data and relative importance of PTK7 for cellular fitness, I supplemented the above results by repeating the same proliferation assays (ie. -/+ PTK7) across a panel of cell lines that gave a wide range of PTK7 Bayes Factor scores (Figure 12). Specifically, I selected cell lines with BF scores between 2 and -1, as I was curious to validate the predictive ability of the BF scores below the high-confidence cutoff of BF > 2 and repeated the cell counting experiments described above in additional cell lines including KP-4, BT549, OVCAR3, HPAC and T47D cells. Notablly, I observed a correlation of relative cell counts across different cell lines associated with their respective BF scores (Figure 12). As predicted, KP-4 and BT549 cells were indeed sensitive to PTK7 perturbation, and OVCAR3 cell lines were not (Figure 12, see table). Unexpectedly, a significant increase in proliferation for HPAC and T47D cell lines, which had low Bayes factor scores, was also observed. These observations support the importance of PTK7 levels and the idea that PTK7 might act as a tumour suppressor gene in certain genetic contexts, as knockdown in these cell lines induced a significant increase in proliferation.

2.2.2 PTK7 full-length rescues this proliferation defect

Although the hairpins recapitulated the screening data, it is still possible that off-target effects of the hairpins could be responsible for, or confound the severity of, the defect. Importantly, HCT116 TP53-/- cells over-expressing PTK7 showed enhanced proliferation (Figure 13A,B) compared to control cells treated with hairpins PTK7-sh6433 and PTK7-sh199565 (P = 0.0035, P=0.0202 respectively), whereas PTK7 overexpression did not enhance proliferation in cell expressing PTK7-sh6431. However, a postdoctoral fellow in the lab, Dr. Andrea Uetrecht, performed similar ‘rescue’ experiments by co-transducing PTK7 over-expression constructs and PTK7 shRNAs, and observed significant ‘rescue’ with full-length PTK7-V5 for all three of the PTK7 shRNAs (unpublished results).

In addition to examining cell proliferation in 2-dimensions, I collaborated with Dr. Andrea Uetrecht to perform anchorage independent growth assays for cells over-expressing full-length PTK7. In HCT116 TP53-/- cells, neither control nor over-expressing cells were able to generate substantial numbers of colonies (Figure 14). However, HCT116 TP53R248W/- cells, which are more mesenchymal in morphology, were able to form colonies (average = 29). In this genetic background, over-expression of PTK7 significantly increased the number of colonies formed (average = 64.5, P= 0.036), suggesting a role for PTK7 in promoting anchorage-independent growth (Figure 14).

2.2.3 PTK7 intracellular domain enriches to the nucleus

It was previously shown by Na et al. that the cytoplasmic domain of PTK7 localizes to the nucleus after processing by Y-secretase50. Plasmids were designed to over-express the cleaved Y-secretase fragment of PTK7 based on the published processing site. A schematic of the construct used for over-expression and localization is shown in Figure 15A. The native ICD fragment exhibited localization to both the cytoplasm and nucleus in HCT116 TP53R248W/- and

KP-4 cell lines, confirming previous reports. As mentioned previously, PCP based defects were rescued in zebrafish using a tethered and secreted ECD fragment of the PTK7 ECD48. I attempted to treat cells with the purified version of the PTK7-ECD to rescue a proliferation defect but observed no changes. With evidence of nuclear enrichment, we hypothesized that perhaps the PTK7-ICD fragment was responsible for the proliferation-associated effects we observed.

2.2.4 PTK7 intracellular domain is sufficient to partially rescue the fitness defect observed upon knockdown

To test whether the intracellular domain is providing the PTK7 activity required for cellular fitness, HCT116 cells were transduced with the ICD-EGFP construct and sorted for a high- expressing population using FACS. Cells were transduced with two different 3’ UTR targeting PTK7 hairpins and compared to shRFP controls in counting assays. Infections were done concurrently in both the over-expression HCT116 cell lines as well as control untransduced cell lines, and cells were counted on day 6 post-infection. All three hairpins significantly reduced proliferation in both HCT116-TP53-/- and HCT116-TP53pm/- cells, and these effects were reversed with expression of the PTK7-ICD (Figure 16A). I also tested the breast cancer cell line BT549 to determine if the rescue effect occurred in a cell line derived from a different tumor tissue type. BT549 cells knocked down for PTK7 had reduced proliferation, which was only partially rescued by expression of PTK7-ICD in the PTK7-sh6431 expressing cells (Figure 16B). Although the rescue was not a complete return to the shRFP control level, over-expression of the PTK7-ICD fragment significantly increased cellular fitness in the presence of PTK7 knockdown with PTK7-sh6431.

2.2.5 Processing of PTK7 negatively-correlates to its essentiality

The data described above and unpublished results from the Moffat lab suggest that the intracellular domain of PTK7 is important for proliferation in PTK7-sensitive cell lines, and may explain the lack of correlation between PTK7 sensitivity and transcript expression. That is, PTK7 mRNA expression does not correlate with PTK7 essentiality as measured by Bayes Factors (Figure 17A). If processing of PTK7 and subsequent release of the ICD are integral for its function, then the amount of processing may be more indicative of its essentiality. As PTK7 processing is a sequential event caused by the three enzymes MT1-MMP, ADAM17 and γ- secretase, I reasoned that quantifying the level of ECD shed into the cell supernatant would be indicative of the amount of ICD released from the membrane. To test this hypothesis, I developed a method to quantify the amount of ICD present in a solution. Utilizing two human synthetic antibodies previously generated against PTK7 in collaboration with the TRAC, I developed a sandwich ELISA to quantify the amount of shed extracellular PTK7.

The sandwich ELISA made use of the antibodies E1 and E6, which targeted two unique epitopes of the PTK7 extracellular domain (Figure 9A). Using recombinant PTK7-ECD protein, I determined the detection range of the ELISA by a dilution series experiment (Figure 17B). The dilution series revealed that the linear detection range went as low as 3.12nM, 3-fold above background as determined by capture with a control maltose-binding protein (MBP) antibody (Figure 17B). Preliminary testing of conditioned media from SW480, JHU and HT29 colon

cancer cells showed high background noise and large variability in 450nm absorbance. To improve the sensitivity and specificity of the assay, I developed a fluorescence based cytometric bead array (Figure 9B). This method of analysis allowed for faster and more sensitive screening of conditioned media from different cell lines. Breast cancer lines tested were: BT549, MDAMB157, MDAMB231, MDAMB468, T47D, MX1; pancreatic cancer lines tested were KP- 4, MIAPACA2, Su.86.86, HPAFII, HPAC, GP3A and ASPC1. Interestingly, a negative correlation between the relative quantity of soluble PTK7 detected in conditioned media and PTK7 essentiality as measured by Bayes Factors, was observed (Figure 17C).In other words, cell lines where PTK7 was highly processed were less likely to be affected by PTK7 knockdown.

2.2.6 PTK7-ICD can rescue the anti-tumour effects of PTK7 antibodies in vivo

It was previously observed in our lab that anti-PTK7 synthetic human antibodies developed in the Moffat and Sidhu labs have inhibitory effects on tumour xenograft growth (J. Moffat & S. Sidhu, unpublished). Since perturbation of PTK7 by shRNA was able to consistently induce a fitness defect in HCT116 cell lines, we were curious if treatment of cells with antibodies against PTK7 could be rescued by the ICD in vivo. To test this hypothesis we designed a 10-arm study (Table 3) to test human synthetic antibodies against PTK7 in vivo. Both strains of HCT116 cells were prepared for subcutaneous xenograft in SCID mice. Mice with HCT116 TP53-/- cells treated with an antibody (E11) against PTK7 showed a marked tumour growth inhibition (TGI = 29.5%, P =0.0219) after 35 days post-implantation (Figure 18A). These results have also been with the TP53R248W/- HCT116 cell line.

To test whether cells over-expressing the PTK7-ICD were susceptible to antibody-dependent tumour growth inhibition, cells over-expressing the ICD were used for xenograft experiments

and treatment with the PTK7-E11 antibody. To control for the nuclear EGFP, cells with H2B- EGFP were also xenografted (Table 3). I hypothesized that cells that exogenously over-express the functional ICD fragment would be resistant to antibody treatment. Accordingly, we observed that H2B-EGFP mice treated with the antibody had tumour growth inhibition of 49.2% (P=0.0004). In contrast, mice with ICD over-expressing tumours (Figure 19) exhibited minimal attenuation of tumour growth (TGI = 12.7%, P=0.4466). HCT116 TP53-/- tumours grew almost identical to PBS treated controls (TGI=3.6%, P=0.8298), a stark decrease in comparison to the previous 29.5% (Figure 18B). These comparisons support the idea that PTK7-ICD has an important role in proliferation in these cells. Additionally, this result suggests that the antibody is acting on the cell at a level that disrupts PTK7-ICD accumulation – either indirectly by causing a decrease in total PTK7 expression or directly by disrupting PTK7 processing and ICD release.

2.2.7 Antibodies directly block cleavage of PTK7 in vitro

To test the possibility that PTK7 antibodies can block MMP-mediated cleavage of PTK7 in vitro, I established an in vitro PTK7 cleavage assay and began testing whether some of the synthetic human anti-PTK7 antibodies previously generated in our lab could impact this assay. I designed an experiment to assess the ability of PTK7 antibodies to block cleavage of the recombinant PTK7-ECD Fc-fusion protein (Figure 20A-B). The human synthetic antibodies targeting PTK7 were created previously in the Moffat and Sidhu labs using Cellectseq, a method for generating antibodies from phage-displayed antibody libraries directly on cells without the need for purified antigen. A panel of anti-PTK7 human synthetic antibodies with defined affinities, specificities, and epitope groupings were provided to me by the TRAC for studying PTK7 processing. Importantly, recombinant active MT1-MMP was able to cleave ECD-Fc with high efficiency (Figure 20C), corroborating previous reports conducted by over-expression or knockdown of MT1-MMP49,54,69. Treatment with a control pooled human IgG did not affect the efficiency of ECD-Fc cleavage. However, treatment with PTK7 specific IgGs, especially IgG #3 (E6), was able to attenuate MT1-MMP dependent cleavage. The remaining three IgGs did not demonstrate a direct blocking effect. This result provides the first evidence of direct processing of PTK7 by MT1-MMP and suggests that at least in the case of IgG#3, the antibody can block processing in vitro. However, no conclusions can be made about the mechanism of action of E11 specifically, for which the TGI was observed in vivo.

In the interest of identifying other putative proteases responsible for processing, the protein sequence was entered into the PROSPER database70. From this consensus sequence analysis, matrix metalloproteinase 2 (MMP-2) and 9 (MMP-9) emerged as capable of processing PTK7 at the reported L621 site where MT1-MMP cleaves. Thus, catalytically active recombinant MMP-2 and MMP-9 were tested for their ability to cleave the ECD-Fc protein. Treatment with MMP-2 resulted in cleavage of the ECD-Fc, which could be attenuated upon pre-incubation with PTK7 IgG#3 (Figure 20D). However, MMP-9 did not show any cleavage potential for recombinant ECD-Fc suggesting it might not process PTK7, at least within this in vitro setting. Discovery and validation of MMP-2 as a putative enzyme for PTK7 processing suggests alternative players involved in PTK7 regulation. It remains unclear whether MMP-2 processing of PTK7 is an observation unique to my in vitro setting or if secreted MMP-2 is capable of regulating PTK7 processing in its native cellular context. Regardless, it is clear that PTK7 processing is complex with multiple potential proteases that act directly on the protein, and should be further investigated to better understand the molecular mechanism of its post-translational regulation.

2.2.8 RNAseq of PTK7 knockdown and ICD over-expression identify putative targets for PTK7 associated function

Thus far I have demonstrated that PTK7 is important for cellular proliferation and the intracellular domain of PTK7 is at least partially required for this effect. In order to test the

hypothesis that PTK7 has a function in nuclear transcription, RNA sequencing (RNAseq) experiments were performed following PTK7 knockdown or in HCT116 cells over-expressing the intracellular domain. Testing was completed in both TP53-/- and TP53R248W/- cell lines, as I hoped we could determine genes that consistently changed between both contexts. My goal was to identify common pathways or complexes that significantly changed upon perturbation of PTK7 or over-expression of the ICD. With the help of Dr. Kevin Brown, I compared the knockdown of PTK7 across 3 hairpins to the shRFP hairpin and untreated controls. As expected, messenger RNA expression of PTK7 was significantly decreased across all 3 hairpins (Figure 21A). The genes were plotted in a volcano plot to identify differentially expressed genes (Figure 21B). Surprisingly, a total of 318 genes emerged as having changed significantly upon knockdown (Padjusted < 0.05, log2FoldChange[FPKM] ≥ 1.0 or ≤ -1.0). The top 50 hits based on significance are presented in Table 4, with known genes associated with the Wnt signaling pathway highlighted in yellow. Further highlights have been made in green for genes associated with cancer development based on their GeneCards annotations (pathway categorization includes “Molecular Mechanisms of Cancer”). Interestingly, two known Wnt pathway genes changed significantly in expression. First, Dickkopf-1 or DKK1 is a potent, secreted inhibitor of canonical Wnt signaling. DKK1 is also a target of the β-catenin/T-cell factor transcriptional activation, as part of a feedback loop71. In contrast, a second Wnt pathway associated gene, axis inhibition protein 2 (AXIN2), increased in expression upon PTK7 knockdown. AXIN2 is a member of the destruction complex responsible for β-catenin degradation and down-regulating canonical signaling72. Similar to DKK1, the transcription of AXIN2 was previously shown to be up- regulated by active β-catenin-TCF/LEF signaling as part of a feedback loop73.

To further resolve the data, I attempted to use gene set enrichment analysis to identify specific pathways or clusters from the RNA-seq data. Hits identified as significant above were enlisted in gene set enrichment analysis using the MSigDB database specifically for BioCarta, KEGG and Reactome curated gene sets. Unfortunately, the results did not provide any significant enrichment of pathways or complexes. Instead I chose to analyze the data utilizing the diverse amount of essentiality data we have from the screens. I hypothesized that if PTK7 was truly regulating transcription downstream, and these gene targets were involved in growth, they should also emerge to some degree as correlated to PTK7 essentiality. That is to say, genes with Bayes factor scores that correlate well with PTK7 – meaning they are essential in the sensitive lines and non-essential in the resistant lines – may be downstream transcriptional targets.

Table 4: Top 50 genes with mRNA expression changes upon PTK7 knockdown

Gene Symbol Log2FoldChange Padjusted PTK7 -3.45416359 1.10E-36 COL5A3 3.909025594 4.04E-20 SEMA7A 2.014146597 2.53E-18 HMGCS1 1.803040525 8.89E-15 THBS1 1.908520562 3.73E-13 VGF -1.633098758 6.67E-12 H1F0 -1.815935391 1.34E-11 MSMO1 1.561184513 3.82E-11 BMP4 1.566052301 4.17E-11 TRIB3 -1.799720345 4.17E-11 NRL -1.714134391 4.18E-11 EREG -1.613927541 9.23E-11 PCK2 -1.693031229 9.93E-11 TRIML2 2.237842099 1.23E-10 SUSD2 1.49569297 8.20E-10 FDFT1 1.450143076 1.07E-09 PCSK9 1.417001482 1.87E-09 DKK1 -1.630068888 2.43E-09 ACAT2 1.458656189 2.77E-09 FAM129A -2.155213176 5.46E-09 CTGF 1.687512869 9.75E-09 YOD1 1.373771647 1.42E-08 PEX19 -1.50674589 1.42E-08 PRSS22 1.592352852 3.16E-08 PLK2 1.292168562 4.46E-08 BMF 2.192622812 5.68E-08 LAMP3 -2.392107766 6.06E-08 SMAD7 2.315441728 6.07E-08 ARID5B 1.630093864 6.07E-08 ASNS -1.407772671 7.23E-08 RABL6 1.14892906 7.72E-08 CSTA -2.751930255 1.03E-07 SLC7A1 -1.367506855 1.37E-07 RBCK1 -1.519043516 1.37E-07 MFSD2B -1.44326029 1.48E-07 COL13A1 1.224364562 1.82E-07 CYP24A1 -1.257020706 2.49E-07 INSIG1 1.213403986 3.37E-07 PSAT1 -1.306442297 4.79E-07 HMOX1 1.270679538 4.91E-07 BIRC3 1.530003424 7.26E-07 KLHDC7B -4.251155327 8.36E-07 CTNNAL1 1.239795702 1.13E-06 LINC00152 1.349439944 1.14E-06 AXIN2 1.560721337 1.27E-06 ZNF367 1.169079323 1.54E-06 CDH23 1.181510402 2.35E-06 SH3RF2 1.292457756 2.36E-06 FERMT2 1.165518761 2.56E-06 GABRB3 -1.494395814 2.64E-06

By correlating Bayes factor scores for all genes across the original 68 cell lines, I selected the correlation scores for the 318 genes that emerged previously. I ranked them by their correlation to PTK7 essentiality (with PTK7 being ranked first, a correlation score of 1). Next, to further resolve the data I ranked the genes by the absolute value log2FoldChange to determine which genes had the largest changes in expression. A new column (sum of ranks) shows the sum of ranks for both expression change and correlation to PTK7 essentiality (Table 5,6). These preliminary lists provide a new ranked order of genes whose expression changed substantially and that were highly correlated or anti-correlated with PTK7 essentiality. One example of a gene that emerged to the top of the list is Lipocalin 2 (LCN2). This gene has annotations associated with epithelial-to-mesenchymal transition, angiogenesis and has been linked to iron-dependent apoptosis74. With essentiality and expression change correlated with PTK7 knockdown, this provides a possible candidate gene for downstream PTK7 signaling and function. A new diverse list of gene candidates emerged after this analysis, and genes with cancer related annotations were highlighted in green. I have since completed some RNAseq of HCT116 cell lines over- expressing the ICD domain. Unfortunately not enough replicates have been performed to generate a list of genes that had significant changes in expression (data not shown). A list of common genes that have altered and opposite expression patterns upon ICD over-expression compared to PTK7 knockdown should help to further refine the list of candidate genes.

2.3 Discussion

In this chapter, I aimed to validate the role of PTK7 in cellular proliferation and elucidate the impact of PTK7 processing towards that function. I began by demonstrating the ability of multiple shRNAs against PTK7 to induce a defect in cellular fitness across 7 breast, colon, pancreatic and ovarian cancer cell lines. This was in accordance with the Bayes factor scores generated from high-throughput functional genetic screening of primary tumour-derived cancer cell lines. The screens were performed to identify genes important for cancer cell proliferation at the genome-scale. It is evident that PTK7 has a complex function in cells that may be pro- or anti-proliferative. One explanation for this is that the underlying expression or activation of other key pathways upstream or downstream of PTK7, influences its end phenotype. PTK7 has been implicated as a Wnt co-receptor in various studies36,39,47,75,76. It is unclear whether the role as PTK7 being pro- or anti-proliferative is associated with Wnt signaling within in the cell line.

Further analyses of the genetic contexts where PTK7 enhances or suppresses growth are necessary to provide insight into its functional role.

Table 5: Top 25 genes based on expression change and correlation with PTK7 essentiality

Gene Symbol Log2FoldChange Essentiality Correlation Sum of Ranks PTK7 -3.45416359 1.0 3 LCN2 -4.070343871 0.310647471 3 MCF2 2.742969286 0.246439451 8 LOC389791 2.704905519 0.184104076 17 OR51B2 -1.871231913 0.227236279 17 LURAP1L -1.51845435 0.254675396 19 GPR111 2.183723559 0.170841267 23 IL20RB -1.842467357 0.185183317 23 SIDT1 1.594239017 0.212547557 25 LAT2 -1.304819119 0.225150948 27 LCP1 1.916974444 0.15455387 30 TUBE1 -1.273431518 0.207455275 32 SLC35F1 -1.148328027 0.227434544 35 SCEL 1.925316271 0.137191831 36 SKIDA1 1.783335133 0.153605497 36 TRIB3 -1.799720345 0.146822379 37 OR51B4 -1.081859535 0.253946337 37 LITAF -1.091243268 0.219421197 40 PHLDA1 -1.352070349 0.150005921 41 H1F0 -1.815935391 0.129968584 43 TRIML2 2.237842099 0.106562704 45 CD68 -1.087069207 0.179980849 46 ANKRD1 2.15197277 0.100204541 49 CYP24A1 -1.257020706 0.141737597 49 GLIS3 1.175293771 0.141773003 51

To confirm the specificity of the PTK7 knockdown defect, rescues were performed with full- length PTK7. With the help of Dr. Andrea Uetrecht, we were able to significantly rescue the defects across multiple hairpins. Rescues were only partial, rather than completely back to shRFP infected controls. One possibility is that the hairpins were having off-target effects that enhance the defect aside from PTK7 knockdown. It is also possible that cells are extremely sensitive to PTK7 expression and it is tightly regulated. Expression of the full-length PTK7 construct may need to occur at a specific concentration, where too little or too much protein may cause issues in PTK7 trafficking or processing, affecting the rescue. In my hands, I have seen that cells are capable or rewiring over multiple passages and growing even with PTK7 knockdown. Thus, cellular reprogramming after over-expression but before knockdown may

affect our ability to rescue the phenotype. Finally, another possibility is that the exogenously expressed protein is incapable of being processed efficiently due to the C-terminal tag. The PTK7 ICD is important for cellular fitness, and immunofluorescence of the C-terminal V5- tagged full-length PTK7 (Figure 14) shows minimal nuclear localization. To address these possibilities, rescue experiments should be performed so transduction of hairpins and over- expression constructs happens concurrently (Figure 13D). Furthermore, creating a native expression construct with no tags may provide a complete rescue. Taken together, this data suggests that PTK7 is important for fitness in the cellular contexts investigated.

Over-expression of the full-length PTK7 was able to produce an anchorage independent growth (AIG) phenotype in HCT116 TP53R248W/- cells. Anchorage independent growth is a hallmark of transformed cells, and is frequently used to test the oncogenic potential of a gene77. PTK7 over- expression was capable of drastically increasing the ability for cells to form colonies without an adherent surface. However, this increase was not observed in the TP53-/- cells. It remains unclear if this phenotype is dependent on the presence of p53 within the cell, or rather has to do with the mesenchymal state that TP53R248W/- cells already exist in. To test the true potential of PTK7 to be transformative, it would be worthwhile to test over-expression in non-transformed fibroblasts, which are a gold standard to test cellular transformation for adhesion independent colony formation1,77. It would also be interesting to test the transformative potential of the intracellular domain alone, across sensitive and insensitive lines, to validate a positive AIG result shown by Na et al50. There is limited data that addresses the role of PTK7 in various cellular locations. This observation brings to light the possibility that PTK7 signaling for some functions may be sufficient at the membrane whereas localization of the ICD may have unique signaling implications.

Expression of the intracellular domain of PTK7 exhibits an enriched nuclear localization. The localization of the catalytically inactive ICD to the nucleus suggests a possible method for cellular signaling independent of kinase activity. A similar mechanism exists for multiple members of the RTK family78. One such example is EGFR, where following endocytosis the full-length receptor is trafficked to the nucleus to induce transcriptional activation, DNA repair and replication independent of kinase activity79,80.

Table 6: Top 25 genes anti-correlated with PTK7 essentiality

Gene Symbol Log2FoldChange Essentiality Correlation Sum of Ranks CSTA -2.751930255 -0.272268845 5 CEACAM1 1.874114637 -0.186450241 12 ARID5B 1.630093864 -0.24543599 12 LIPG 1.945160718 -0.174699596 13 HMGA1 -1.488030154 -0.200819571 17 PCSK9 1.417001482 -0.278663732 17 WDR49 3.189156214 -0.144630375 19 SELPLG 1.808362258 -0.146307512 23 GABRB3 -1.494395814 -0.163645122 23 IQGAP2 1.681479339 -0.150011167 24 PLK2 1.292168562 -0.175222227 28 RGCC 1.413817125 -0.162457522 29 MSMO1 1.561184513 -0.13394132 30 HBA2 1.548251593 -0.136827382 30 BTG2 1.19161055 -0.187177873 30 NRL -1.714134391 -0.129443461 31 ASNS -1.407772671 -0.156567025 32 STMN2 3.056846587 -0.103590878 38 R3HDM2 -1.16959666 -0.160141372 38 KIF1A 1.440847979 -0.127581079 40 CREB5 1.442506493 -0.124404464 41 HSD17B8 -1.378499597 -0.130758336 41 HCN3 1.056397494 -0.199867373 41 MRPL24 -1.095245069 -0.163707745 43 WNT7A 1.272325841 -0.130741822 44

A total of 18 RTKs have been shown to traffic to the nucleus, and in many instances the intracellular domain fragments. These fragments are achieved through proteolysis, alternate splice variants or internal translation sites78. As an atypical RTK without kinase activity, it is not surprising for the ICD to traffic to the nucleus to execute its function. To this end, I observed that the ICD alone was sufficient to rescue the fitness defect observed by PTK7 knockdown. This finding emphasizes the importance of efficient PTK7 processing and ICD release. The rescue was performed in both HCT116 cell lines along with BT549 breast cancer lines, suggesting that PTK7 is important across cell lines from multiple cancer types. It remains to be tested whether over-expression of the ICD can rescue the defect across multiple sensitive lines. It would also be interesting to test the potential of the ICD to attenuate cell proliferation in cell lines with high levels of soluble PTK7.

Based on various recent contradictory reports, it is clear that PTK7 functionality depends on more than just expression55-61. Here, I have shown that PTK7 essentiality does not correlate with its mRNA expression, based on comparing transcript and functional genetic information. Importanly, essentiality is associated with processing of PTK7, based on the quantification of the shed PTK7 extracellular domain. This observation is paramount in understanding the sensitivity of a cancer cells to PTK7 perturbation. It may be possible to characterize a cell’s sensitivity to PTK7 simply from sampling culture medium. Looking forward, it is possible that quantifying PTK7 processing in a cell line where proteases such as MT1-MMP, ADAM17 or γ-secretase are inhibited may reveal novel players capable of processing PTK7. Currently no evidence of PTK7 directly binding a ligand, and it remains formally possible that ligand binding or structural changes are responsible for triggering cleavage by membrane proteases. Screening of conditioned media from different cell lines provides the possibility to test various ligands, inhibitors and antibodies to screen for antagonists of PTK7. Finally, there is potential to use this detection method to screen patient serum samples to quantify the concentration of shed PTK7 to determine the possibility to use PTK7 proteolysis as a marker for prognosis or treatment options.

The current landscape of PTK7 research has postulated that targeting the RTK might be an effective therapeutic. Using human synthetic antibodies developed in-house, we were able to show moderate tumour growth inhibition in vivo. This finding, collectively with data from ICD rescues in vivo, implicates the released intracellular domain as an important player in cellular fitness. To generate further evidence of this, an experiment in which a cleavage-incapable mutant (L621D) full-length PTK7 is used to rescue the hairpin defects should be conducted. One of the antibodies (IgG#3) showed direct ability to block processing of PTK7 by MT1-MMP. It is unclear the effect inhibitors of MT1-MMP, ADAM17 and γ-secretase would have collectively. One hypothesis is that combinatorial inhibition will eliminate PTK7 processing completely, resulting in a synergistic effect. Alternatively, extra processing of the extracellular domain, such as what’s seen in the Chuzhoi mutant may be another method for disrupting PTK7 function53,54.

The mechanism by which PTK7 enacts downstream signaling to promote a proliferative phenotype is poorly understood. To generate a better understanding of this problem, I performed RNAseq of HCT116 cells facing PTK7 knockdown. Analysis of this list returned a list of candidate genes which may act downstream of PTK7. One pair of genes that emerged included DKK1 and AXIN2, known members of the Wnt pathway. Changes in these genes suggest a role

for PTK7 in either activating or inhibiting canonical Wnt signaling. HCT116 cells are known to carry a mutant β-catenin hypermorphic allele that promotes growth and survival81. This may be one reason certain cell lines are susceptible to PTK7 perturbation, and it remains to be investigated if there is a preference for the cell lines identified as sensitive to preferentially signal via the β-catenin pathway. Many other candidate genes emerged that were associated with oncogenesis. Candidate genes such as epiregulin (EREG), tribble pseudokinase 3 (TRIB3) and asparagine synthetase (ASNS) were markedly down regulated upon PTK7 knockdown. Epiregulin is a member of the EGF family of ligands that binds EGFR and ErbB2 and stimulates cell proliferation82. EREG expression and secretion are often deregulated in various malignancies82. TRIB3 is a poor prognostic marker for colorectal cancer, and siRNA inhibition has resulted in significant reductions in cell growth83. ASNS inhibition in breast cancer significantly decreased cellular proliferation and anchorage independent colony formation84. Collectively, these results and gene candidates behave consistently with the phenotypes observed upon PTK7 knockdown and may be direct downstream targets of transcriptional activation. Although much still remains to be investigated and understood, the RNAseq data provided herein provides a candidate list to further investigate as direct targets of PTK7.

Many of the targets that emerged from the RNAseq data have been implicated in cancer development and progression. Thus, PTK7 may play a role upstream of many of these driver genes, acting as either a repressor or activator of expression depending on specific cellular contexts. It remains unclear what specifically defines the role of PTK7 as a possible tumour suppressor gene or oncogene, but the data presented in this thesis suggests that abundance of processing and ICD release have an important role in PTK7 activity. By continuing to investigate PTK7 for putative transcription targets and protein-protein interactions, a path for targeting PTK7 in a therapeutic setting may become reality..

Chapter 3 Summary and Future Directions 3 Summary

Receptor tyrosine kinases play an integral role in regulating many diverse and crucial processes for cellular growth, migration and division. Although many members of the RTK family have been well studied and characterized, there are many members who belong to the atypical family of pseudokinases that remain poorly understood. Specifically, it is unclear the mechanism by which these RTKs may signal within the cell without catalytic activity to phosphorylate downstream effectors. One such pseudokinase is PTK7, which has been implicated heavily in early embryonic development and a variety of malignancies. However, conflicting reports have emerged as to the role of PTK7 in cancer development and progression, specifically its role as an oncogene or tumour suppressor gene. Due to its abundance across many cancers, and its complex nature, it is essential to investigate PTK7 in a variety of genetic and cancer contexts to resolve its activities and roles.

In this thesis I have demonstrated the ability for PTK7 to affect cellular growth and proliferation, as well as stimulate anchorage independent growth. Additionally, I have observed the importance of the intracellular domain for enacting this function, and developed antibody-based assays to quantify the release of this fragment. Some of these antibodies are capable of attenuating PTK7 processing and need to be further investigated for their capacity to act as a therapeutic lead. Finally, an analysis of RNAseq data generated from PTK7 knockdown revealed a list of candidate genes which may be regulated by PTK7. Further investigation into these candidates, as well as confirming the role of the ICD and transcription will provide a refined understanding of PTK7 function.

In summary, I have better defined the role of PTK7 in cellular fitness and implicated the processing of the receptor, more specifically the release of the intracellular domain and localization to the nucleus, as an important part of its function. These findings provide a framework to further investigate and analyze cellular targets and interactors of PTK7, as well as unique genetic contexts which dictate how PTK7 functions inside of the cell.

3.1 Future Directions

3.1.1 Determining protein interactors and transcriptional targets of the PTK7 intracellular domain

Sequencing of the PTK7 knockdown data alone provided a complete list of 318 genes that significantly changed in expression. However, this data only provides half of the story. I hypothesize that over-expression of the ICD will provide a novel set of putative downstream targets. Taken together, genes of overlap should be further validated. Changes of top hits should be confirmed at the protein level by Western blot, as mRNA expression changes are not necessarily indicative of protein abundance. Genes that have been validated should then be subjected to shRNA knockdown to assess if the phenotype is consistent. It is possible that PTK7 is a master regulator of a variety of genes, and accordingly knockdown of individual targets may not recapitulate the fitness defect. To address this, I think it is also critical to perform protein- protein and protein-DNA interaction screens. First I would attempt an affinity-purification/mass spectrometry (AP/MS) approach with the ICD fragment. This could provide a list of candidate transcription factors or complexes it may associate with. Depending of whether known transcription factors emerged, I would recommend a ChIP-Seq experiment to determine if PTK7 is actually acting directly at the DNA level. If the intracellular domain is behaving as a co- activator or co-repressor of transcription, identifying its partners will help refine the list of candidate genes generated from RNAseq. Performing these experiments in the context of cell lines that are known to be sensitive to perturbation or those which are more resistant may again add another layer of understanding the function of the ICD.

3.1.2 Examining the role of PTK7 in cell lines where knockdown increases proliferation

The purpose of my study was to determine the role PTK7 played in cellular growth and fitness, specifically how it promotes proliferation. However, as I discovered from analyzing the spectrum of cell lines with low essentiality scores, it is possible that PTK7 is in fact inhibiting growth. I am curious to further examine these cell lines and understand the genetic contexts that may shuttle PTK7 signaling towards one pathway versus another. Checking the endogenous expression of the ICD across these various lines relative to full-length expression should provide some useful information in better elucidating the intracellular domain function. In cell lines that demonstrate increased growth upon PTK7 knockdown, such as HPAC and T47D, I would start

by over-expressing PTK7. If full-length or ICD over-expression was capable of inhibiting growth the same way knockdown was in the sensitive lines, I could confirm that there are contexts where PTK7 over-expression is toxic. If it is true that ICD over-expression inhibits growth, I would like to assess cellular behaviour in scenarios where I stimulate proteolysis of PTK7 such as the addition of phorbol myristate acetate. I hypothesize that cells that are not sensitive to PTK7 perturbation rely on a completely independent pathway to stimulate growth. Analysis of genes that are essential in this grouping may provide insight into possible signatures that distinguish these growth patterns.

3.1.3 Elucidating the role of the shed N-terminus in affecting tumorigenesis

To date there is no published ligand of PTK7. However, multiple studies have implicated the N- terminal fragment shed after PTK7 processing as having a functional role54,69,85,86. Importantly, it has been hypothesized that the extracellular domain may be able to dimerize with PTK7 receptors or act as a saturator of an undetermined ligand. The focus of my thesis has been centered on the PTK7-ICD, but of equal importance is understanding the role of the ECD. In particular, as the rescue experiments did not return full function, there may be a possible role for the ECD in promoting cellular growth. To elucidate the effect of the extracellular domain, I propose to treat cells with exogenous ECD-Fc and determine changes to known phenotypes of PTK7. This includes measuring effects on proliferation, migration and collagen invasion. If phenotype changes were observed, it would be worthwhile to identify putative protein interactors for the ECD. Treating cells with ECD-Fc, it is possible to pull down the Fc tag and perform AP/MS on both the supernatant, as well as cellular lysate. Possible interactors or ligands could provide insight into PTK7 signaling. One hypothesis that hasn’t been well discussed is the ability for shed PTK7 to interact with and block proteases on the cell surface. If shed PTK7-ECD is still capable of binding proteases such as MT1-MMP, rendering the enzyme occupied but incapable of cutting, it could represent a natural regulatory feedback mechanism to ensure PTK7 processing is kept in check. To directly assess this effect, individual immunoprecipitation-WB for all of MT1-MMP, ADAM17 and γ-secretase could be tested. Lastly, if any effect is seen by treatment of cells with the PTK7-ECD-Fc, one could test proteolysis inhibitors to determine if the effects were additive or independent.

References

1 Hanahan, D. & Weinberg, R. A. Hallmarks of cancer: the next generation. Cell 144, 646- 674, doi:10.1016/j.cell.2011.02.013 (2011).

2 Gschwind, A., Fischer, O. M. & Ullrich, A. The discovery of receptor tyrosine kinases: targets for cancer therapy. Nature reviews. Cancer 4, 361-370, doi:10.1038/nrc1360 (2004).

3 Carpenter, G., King, L., Jr. & Cohen, S. Epidermal growth factor stimulates phosphorylation in membrane preparations in vitro. Nature 276, 409-410 (1978).

4 Ushiro, H. & Cohen, S. Identification of phosphotyrosine as a product of epidermal growth factor-activated protein kinase in A-431 cell membranes. The Journal of biological chemistry 255, 8363-8365 (1980).

5 Kasuga, M., Zick, Y., Blithe, D. L., Crettaz, M. & Kahn, C. R. Insulin stimulates tyrosine phosphorylation of the insulin receptor in a cell-free system. Nature 298, 667-669 (1982).

6 Cooper, J. A., Bowen-Pope, D. F., Raines, E., Ross, R. & Hunter, T. Similar effects of platelet-derived growth factor and epidermal growth factor on the phosphorylation of tyrosine in cellular proteins. Cell 31, 263-273 (1982).

7 Lemmon, M. A. & Schlessinger, J. Cell signaling by receptor tyrosine kinases. Cell 141, 1117-1134, doi:10.1016/j.cell.2010.06.011 (2010).

8 Prebet, T. et al. The cell polarity PTK7 receptor acts as a modulator of the chemotherapeutic response in acute myeloid leukemia and impairs clinical outcome. Blood 116, 2315-2323, doi:10.1182/blood-2010-01-262352 (2010).

9 Ullrich, A. & Schlessinger, J. Signal transduction by receptors with tyrosine kinase activity. Cell 61, 203-212 (1990).

10 Ward, C. W., Lawrence, M. C., Streltsov, V. A., Adams, T. E. & McKern, N. M. The insulin and EGF receptor structures: new insights into ligand-induced receptor activation. Trends in biochemical sciences 32, 129-137, doi:10.1016/j.tibs.2007.01.001 (2007).

11 Huse, M. & Kuriyan, J. The conformational plasticity of protein kinases. Cell 109, 275- 282 (2002).

12 Hubbard, S. R. & Till, J. H. Protein tyrosine kinase structure and function. Annual review of biochemistry 69, 373-398, doi:10.1146/annurev.biochem.69.1.373 (2000).

13 Knighton, D. R. et al. Structural features that specify tyrosine kinase activity deduced from homology modeling of the epidermal growth factor receptor. Proceedings of the National Academy of Sciences of the United States of America 90, 5001-5005 (1993).

14 Hanks, S. K., Quinn, A. M. & Hunter, T. The protein kinase family: conserved features and deduced phylogeny of the catalytic domains. Science 241, 42-52 (1988).

15 Jung, J. W., Shin, W. S., Song, J. & Lee, S. T. Cloning and characterization of the full- length mouse Ptk7 cDNA encoding a defective receptor protein tyrosine kinase. Gene 328, 75-84, doi:10.1016/j.gene.2003.12.006 (2004).

16 Hubbard, S. R. Juxtamembrane autoinhibition in receptor tyrosine kinases. Nature reviews. Molecular cell biology 5, 464-471, doi:10.1038/nrm1399 (2004).

17 Shewchuk, L. M. et al. Structure of the Tie2 RTK domain: self-inhibition by the nucleotide binding loop, activation loop, and C-terminal tail. Structure 8, 1105-1113 (2000).

18 Takeuchi, K. & Ito, F. Receptor tyrosine kinases and targeted cancer therapeutics. Biological & pharmaceutical bulletin 34, 1774-1780 (2011).

19 Robertson, S. C., Tynan, J. & Donoghue, D. J. RTK mutations and human syndromes: when good receptors turn bad. Trends in genetics : TIG 16, 368 (2000).

20 Bargmann, C. I., Hung, M. C. & Weinberg, R. A. Multiple independent activations of the neu oncogene by a point mutation altering the transmembrane domain of p185. Cell 45, 649-657 (1986).

21 Xie, T. et al. Pharmacological targeting of the pseudokinase Her3. Nature chemical biology 10, 1006-1012, doi:10.1038/nchembio.1658 (2014).

22 Wang, Y., Schmid-Bindert, G. & Zhou, C. Erlotinib in the treatment of advanced non- small cell lung cancer: an update for clinicians. Therapeutic advances in medical oncology 4, 19-29, doi:10.1177/1758834011427927 (2012).

23 Riely, G. J. & Yu, H. A. EGFR: The Paradigm of an Oncogene-Driven Lung Cancer. Clinical cancer research : an official journal of the American Association for Cancer Research 21, 2221-2226, doi:10.1158/1078-0432.CCR-14-3154 (2015).

24 Gaborit, N. et al. Examination of HER3 targeting in cancer using monoclonal antibodies. Proceedings of the National Academy of Sciences of the United States of America 112, 839-844, doi:10.1073/pnas.1423645112 (2015).

25 Marcotte, R. et al. Essential gene profiles in breast, pancreatic, and ovarian cancer cells. Cancer discovery 2, 172-189, doi:10.1158/2159-8290.CD-11-0224 (2012).

26 Hart, T., Brown, K. R., Sircoulomb, F., Rottapel, R. & Moffat, J. Measuring error rates in genomic perturbation screens: gold standards for human functional genomics. Molecular systems biology 10, 733, doi:10.15252/msb.20145216 (2014).

27 Moffat, J. et al. A lentiviral RNAi library for human and mouse genes applied to an arrayed viral high-content screen. Cell 124, 1283-1298, doi:10.1016/j.cell.2006.01.040 (2006).

28 Vizeacoumar, F. J. et al. A negative genetic interaction map in isogenic cancer cell lines reveals cancer cell vulnerabilities. Molecular systems biology 9, 696, doi:10.1038/msb.2013.54 (2013).

29 Duan, Z. et al. Lentiviral short hairpin RNA screen of genes associated with multidrug resistance identifies PRP-4 as a new regulator of chemoresistance in human ovarian cancer. Molecular cancer therapeutics 7, 2377-2385, doi:10.1158/1535-7163.MCT-08- 0316 (2008).

30 Mossie, K. et al. Colon carcinoma kinase-4 defines a new subclass of the receptor tyrosine kinase family. Oncogene 11, 2179-2184 (1995).

31 Banga, S. S., Ozer, H. L., Park, S. K. & Lee, S. T. Assignment of PTK7 encoding a receptor protein tyrosine kinase-like molecule to human chromosome 6p21.1-->p12.2 by fluorescence in situ hybridization. Cytogenetics and cell genetics 76, 43-44 (1997).

32 Park, S. K., Lee, H. S. & Lee, S. T. Characterization of the human full-length PTK7 cDNA encoding a receptor protein tyrosine kinase-like molecule closely related to chick KLG. Journal of biochemistry 119, 235-239 (1996).

33 Lu, X. et al. PTK7/CCK-4 is a novel regulator of planar cell polarity in vertebrates. Nature 430, 93-98, doi:10.1038/nature02677 (2004).

34 Miller, M. A. & Steele, R. E. Lemon encodes an unusual receptor protein-tyrosine kinase expressed during gametogenesis in Hydra. Developmental biology 224, 286-298, doi:10.1006/dbio.2000.9786 (2000).

35 Winberg, M. L. et al. The transmembrane protein Off-track associates with Plexins and functions downstream of Semaphorin signaling during axon guidance. Neuron 32, 53-62 (2001).

36 Peradziryi, H., Tolwinski, N. S. & Borchers, A. The many roles of PTK7: a versatile regulator of cell-cell communication. Archives of biochemistry and biophysics 524, 71- 76, doi:10.1016/j.abb.2011.12.019 (2012).

37 Montcouquiol, M., Crenshaw, E. B., 3rd & Kelley, M. W. Noncanonical Wnt signaling and neural polarity. Annual review of neuroscience 29, 363-386, doi:10.1146/annurev.neuro.29.051605.112933 (2006).

38 Kohn, A. D. & Moon, R. T. Wnt and calcium signaling: beta-catenin-independent pathways. Cell calcium 38, 439-446, doi:10.1016/j.ceca.2005.06.022 (2005).

39 Puppo, F. et al. Protein tyrosine kinase 7 has a conserved role in Wnt/beta-catenin canonical signalling. EMBO reports 12, 43-49, doi:10.1038/embor.2010.185 (2011).

40 Komiya, Y. & Habas, R. Wnt signal transduction pathways. Organogenesis 4, 68-75 (2008).

41 Mikels, A. J. & Nusse, R. Wnts as ligands: processing, secretion and reception. Oncogene 25, 7461-7468, doi:10.1038/sj.onc.1210053 (2006).

42 Janda, C. Y., Waghray, D., Levin, A. M., Thomas, C. & Garcia, K. C. Structural basis of Wnt recognition by Frizzled. Science 337, 59-64, doi:10.1126/science.1222879 (2012).

43 Slusarski, D. C. & Pelegri, F. Calcium signaling in vertebrate embryonic patterning and morphogenesis. Developmental biology 307, 1-13, doi:10.1016/j.ydbio.2007.04.043 (2007).

44 Seifert, J. R. & Mlodzik, M. Frizzled/PCP signalling: a conserved mechanism regulating cell polarity and directed motility. Nature reviews. Genetics 8, 126-138, doi:10.1038/nrg2042 (2007).

45 Devenport, D. The cell biology of planar cell polarity. The Journal of cell biology 207, 171-179, doi:10.1083/jcb.201408039 (2014).

46 Wang, Y. Wnt/Planar cell polarity signaling: a new paradigm for cancer therapy. Molecular cancer therapeutics 8, 2103-2109, doi:10.1158/1535-7163.MCT-09-0282 (2009).

47 Peradziryi, H. et al. PTK7/Otk interacts with Wnts and inhibits canonical Wnt signalling. The EMBO journal 30, 3729-3740, doi:10.1038/emboj.2011.236 (2011).

48 Hayes, M., Naito, M., Daulat, A., Angers, S. & Ciruna, B. Ptk7 promotes non-canonical Wnt/PCP-mediated morphogenesis and inhibits Wnt/beta-catenin-dependent cell fate decisions during vertebrate development. Development 140, 1807-1818, doi:10.1242/dev.090183 (2013).

49 Golubkov, V. S. et al. The Wnt/planar cell polarity protein-tyrosine kinase-7 (PTK7) is a highly efficient proteolytic target of membrane type-1 matrix metalloproteinase: implications in cancer and embryogenesis. The Journal of biological chemistry 285, 35740-35749, doi:10.1074/jbc.M110.165159 (2010).

50 Na, H. W., Shin, W. S., Ludwig, A. & Lee, S. T. The cytosolic domain of protein- tyrosine kinase 7 (PTK7), generated from sequential cleavage by a disintegrin and metalloprotease 17 (ADAM17) and gamma-secretase, enhances cell proliferation and migration in colon cancer cells. The Journal of biological chemistry 287, 25001-25009, doi:10.1074/jbc.M112.348904 (2012).

51 Reiss, K. & Saftig, P. The "a disintegrin and metalloprotease" (ADAM) family of sheddases: physiological and cellular functions. Seminars in cell & developmental biology 20, 126-137, doi:10.1016/j.semcdb.2008.11.002 (2009).

52 Rahimi, N., Golde, T. E. & Meyer, R. D. Identification of ligand-induced proteolytic cleavage and ectodomain shedding of VEGFR-1/FLT1 in leukemic cancer cells. Cancer research 69, 2607-2614, doi:10.1158/0008-5472.CAN-08-2905 (2009).

53 Paudyal, A. et al. The novel mouse mutant, chuzhoi, has disruption of Ptk7 protein and exhibits defects in neural tube, heart and lung development and abnormal planar cell polarity in the ear. BMC developmental biology 10, 87, doi:10.1186/1471-213X-10-87 (2010).

54 Golubkov, V. S., Aleshin, A. E. & Strongin, A. Y. Potential relation of aberrant proteolysis of human protein tyrosine kinase 7 (PTK7) chuzhoi by membrane type 1 matrix metalloproteinase (MT1-MMP) to congenital defects. The Journal of biological chemistry 286, 20970-20976, doi:10.1074/jbc.M111.237669 (2011).

55 Wang, H. et al. PTK7 protein is decreased in epithelial ovarian carcinomas with poor prognosis. International journal of clinical and experimental pathology 7, 7881-7889 (2014).

56 Lhoumeau, A. C. et al. Overexpression of the Promigratory and Prometastatic PTK7 Receptor Is Associated with an Adverse Clinical Outcome in Colorectal Cancer. PloS one 10, e0123768, doi:10.1371/journal.pone.0123768 (2015).

57 Ataseven, B. et al. PTK7 as a potential prognostic and predictive marker of response to adjuvant chemotherapy in breast cancer patients, and resistance to anthracycline drugs. OncoTargets and therapy 7, 1723-1731, doi:10.2147/OTT.S62676 (2014).

58 Zhang, H. et al. Protein tyrosine kinase 7 (PTK7) as a predictor of lymph node metastases and a novel prognostic biomarker in patients with prostate cancer. International journal of molecular sciences 15, 11665-11677, doi:10.3390/ijms150711665 (2014).

59 Kim, J. H. et al. Protein tyrosine kinase 7 plays a tumor suppressor role by inhibiting ERK and AKT phosphorylation in lung cancer. Oncology reports 31, 2708-2712, doi:10.3892/or.2014.3164 (2014).

60 Shin, W. S. et al. Oncogenic role of protein tyrosine kinase 7 in esophageal squamous cell carcinoma. Cancer science 104, 1120-1126, doi:10.1111/cas.12194 (2013).

61 Lin, Y. et al. PTK7 as a novel marker for favorable gastric cancer patient survival. Journal of surgical oncology 106, 880-886, doi:10.1002/jso.23154 (2012).

62 Ataseven, B. et al. PTK7 expression in triple-negative breast cancer. Anticancer research 33, 3759-3763 (2013).

63 Gorringe, K. L., Boussioutas, A., Bowtell, D. D. & Melbourne Gastric Cancer Group, P. M. M. A. F. Novel regions of chromosomal amplification at 6p21, 5p13, and 12q14 in gastric cancer identified by array comparative genomic hybridization. Genes, chromosomes & cancer 42, 247-259, doi:10.1002/gcc.20136 (2005).

64 Technologies, L. Vol. MAN0000282 (Invitrogen, 2015).

65 Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15-21, doi:10.1093/bioinformatics/bts635 (2013).

66 Trapnell, C. et al. Transcript assembly and quantification by RNA-Seq reveals unannotated transcripts and isoform switching during cell differentiation. Nature biotechnology 28, 511-515, doi:10.1038/nbt.1621 (2010).

67 Moffat, J., Reiling, J. H. & Sabatini, D. M. Off-target effects associated with long dsRNAs in Drosophila RNAi screens. Trends in pharmacological sciences 28, 149-151, doi:10.1016/j.tips.2007.02.009 (2007).

68 Kaelin, W. G., Jr. Molecular biology. Use and abuse of RNAi to study mammalian gene function. Science 337, 421-422, doi:10.1126/science.1225787 (2012).

69 Golubkov, V. S. & Strongin, A. Y. Insights into ectodomain shedding and processing of protein-tyrosine pseudokinase 7 (PTK7). The Journal of biological chemistry 287, 42009- 42018, doi:10.1074/jbc.M112.371153 (2012).

70 Song, J. et al. PROSPER: an integrated feature-based tool for predicting protease substrate cleavage sites. PloS one 7, e50300, doi:10.1371/journal.pone.0050300 (2012).

71 Niida, A. et al. DKK1, a negative regulator of Wnt signaling, is a target of the beta- catenin/TCF pathway. Oncogene 23, 8520-8526, doi:10.1038/sj.onc.1207892 (2004).

72 Katoh, M. & Katoh, M. WNT signaling pathway and stem cell signaling network. Clinical cancer research : an official journal of the American Association for Cancer Research 13, 4042-4045, doi:10.1158/1078-0432.CCR-06-2316 (2007).

73 Jho, E. H. et al. Wnt/beta-catenin/Tcf signaling induces the transcription of Axin2, a negative regulator of the signaling pathway. Molecular and cellular biology 22, 1172- 1183 (2002).

74 Yang, J., McNeish, B., Butterfield, C. & Moses, M. A. Lipocalin 2 is a novel regulator of angiogenesis in human breast cancer. FASEB journal : official publication of the Federation of American Societies for Experimental Biology 27, 45-50, doi:10.1096/fj.12- 211730 (2013).

75 Linnemannstons, K. et al. The PTK7-related transmembrane proteins off-track and off- track 2 are co-receptors for Drosophila Wnt2 required for male fertility. PLoS genetics 10, e1004443, doi:10.1371/journal.pgen.1004443 (2014).

76 Shnitsar, I. & Borchers, A. PTK7 recruits dsh to regulate neural crest migration. Development 135, 4015-4024, doi:10.1242/dev.023556 (2008).

77 Mori, S. et al. Anchorage-independent cell growth signature identifies tumors with metastatic potential. Oncogene 28, 2796-2805, doi:10.1038/onc.2009.139 (2009).

78 Carpenter, G. & Liao, H. J. Receptor tyrosine kinases in the nucleus. Cold Spring Harbor perspectives in biology 5, a008979, doi:10.1101/cshperspect.a008979 (2013).

79 Wang, Y. N. & Hung, M. C. Nuclear functions and subcellular trafficking mechanisms of the epidermal growth factor receptor family. Cell & bioscience 2, 13, doi:10.1186/2045- 3701-2-13 (2012).

80 Wang, S. C. & Hung, M. C. Nuclear translocation of the epidermal growth factor receptor family membrane tyrosine kinase receptors. Clinical cancer research : an official journal of the American Association for Cancer Research 15, 6484-6489, doi:10.1158/1078- 0432.CCR-08-2813 (2009).

81 Kaler, P., Augenlicht, L. & Klampfer, L. Activating mutations in beta-catenin in colon cancer cells alter their interaction with macrophages; the role of snail. PloS one 7, e45462, doi:10.1371/journal.pone.0045462 (2012).

82 Riese, D. J., 2nd & Cullum, R. L. Epiregulin: roles in normal physiology and cancer. Seminars in cell & developmental biology 28, 49-56, doi:10.1016/j.semcdb.2014.03.005 (2014).

83 Miyoshi, N. et al. Abnormal expression of TRIB3 in colorectal cancer: a novel marker for prognosis. British journal of cancer 101, 1664-1670, doi:10.1038/sj.bjc.6605361 (2009).

84 Yang, H. et al. Down-regulation of asparagine synthetase induces cell cycle arrest and inhibits cell proliferation of breast cancer. Chemical biology & drug design 84, 578-584, doi:10.1111/cbdd.12348 (2014).

85 Golubkov, V. S. & Strongin, A. Y. Downstream signaling and genome-wide regulatory effects of PTK7 pseudokinase and its proteolytic fragments in cancer cells. Cell communication and signaling : CCS 12, 15, doi:10.1186/1478-811X-12-15 (2014).

86 Golubkov, V. S. et al. Protein-tyrosine pseudokinase 7 (PTK7) directs cancer cell motility and metastasis. The Journal of biological chemistry 289, 24238-24249, doi:10.1074/jbc.M114.574459 (2014).

Appendix 1 – Complete RNAseq Data

Gene Symbol Log2FoldChange Padjusted HRNR 6.026803813 6.07E-06 COL5A3 3.909025594 4.04E-20 SERTAD4-AS1 3.663115569 0.000371321 SYNDIG1L 3.537932083 0.002022825 C8orf31 3.43605114 0.000572586 FOXJ1 3.227460486 6.52E-06 WDR49 3.189156214 0.007907692 KIAA1755 3.18394917 0.00136152 STMN2 3.056846587 0.010797176 ABCD2 2.76664068 0.030637485 MCF2 2.742969286 0.034967935 SERPINB7 2.71575786 0.001802585 LOC389791 2.704905519 0.008875948 CXCR1 2.69814804 0.037668791 GZMB 2.691591918 0.000535152 SLCO1A2 2.545766346 0.008424161 CD14 2.537762614 0.009796759 SLCO1B1 2.47694931 0.021739965 SLC22A1 2.377441021 0.04283003 C2orf54 2.341966725 0.029396679 SMAD7 2.315441728 6.07E-08 TRIML2 2.237842099 1.23E-10 GCNT4 2.207513807 0.013515057 BMF 2.192622812 5.68E-08 GPR111 2.183723559 0.027174568 HTR1B 2.16998068 0.023180374 ANKRD1 2.15197277 0.004608809 SDPR 2.145054672 0.004659668 GLIS3-AS1 2.136000988 0.038044204 NXNL2 2.127984289 0.001336841 MIR663B 2.098228427 0.007123969 EDN1 2.081186375 0.048043764 SEMA7A 2.014146597 2.53E-18 LIPG 1.945160718 5.90E-05 SCEL 1.925316271 0.015718691 CCDC19 1.918818922 0.002724275 C9orf163 1.917774957 0.030427799 LCP1 1.916974444 0.010121453 THBS1 1.908520562 3.73E-13 ZNF252P-AS1 1.893861973 0.049670732 CEACAM1 1.874114637 6.79E-05 SPEF1 1.872161091 0.038891992 IFNE 1.833075445 0.046658616 CD53 1.824910401 0.026024431

Gene Symbol Log2FoldChange Padjusted CCDC80 1.823590394 0.017123875 APLN 1.81689291 0.006869667 NHLH1 1.811511902 0.02776271 SELPLG 1.808362258 0.01192233 HMGCS1 1.803040525 8.89E-15 SKIDA1 1.783335133 0.026024431 SNCA 1.757597957 0.010540854 TGFB2 1.750239247 0.038686272 SIX2 1.745190492 0.032990924 IGDCC3 1.732147762 0.009269757 LOC100216546 1.724221471 0.000150193 CREB3L1 1.692896832 0.021289201 CTGF 1.687512869 9.75E-09 TCN1 1.685093679 0.006318167 IQGAP2 1.681479339 0.000162198 RDH16 1.640930572 0.034102699 ARID5B 1.630093864 6.07E-08 SIDT1 1.594239017 0.046445846 PRSS22 1.592352852 3.16E-08 BMP4 1.566052301 4.17E-11 MSMO1 1.561184513 3.82E-11 AXIN2 1.560721337 1.27E-06 HBA2 1.548251593 0.003576302 HBA1 1.532923732 0.004175245 BIRC3 1.530003424 7.26E-07 SUSD2 1.49569297 8.20E-10 SCG5 1.478295247 0.000359265 ACAT2 1.458656189 2.77E-09 LYPD3 1.457373421 0.000169183 FDFT1 1.450143076 1.07E-09 CREB5 1.442506493 0.037816645 KIF1A 1.440847979 0.011950191 CDH17 1.439659958 0.002766193 PRSS30P 1.43179951 0.015389409 PCSK9 1.417001482 1.87E-09 RGCC 1.413817125 0.010540854 MEX3B 1.412524247 0.015244328 MAP3K8 1.380415949 0.00400609 YOD1 1.373771647 1.42E-08 LOC100216545 1.372113747 0.018759474 DNAJB5 1.357957179 2.91E-05 LINC00152 1.349439944 1.14E-06 SCHIP1 1.349416951 8.17E-06 SMAD6 1.346090963 0.000119357 HAP1 1.342554138 0.00719102 IQCJ-SCHIP1 1.337556454 9.29E-06 MYBL1 1.323695272 3.25E-06

Gene Symbol Log2FoldChange Padjusted TNNC1 1.308010831 0.015548259 PALM3 1.307259028 0.000885739 APOBEC3A 1.301698181 0.034869624 HTRA1 1.292894842 0.002482943 SH3RF2 1.292457756 2.36E-06 PLK2 1.292168562 4.46E-08 KCNQ1OT1 1.282201811 0.000120098 YPEL3 1.280175696 0.004409606 WNT7A 1.272325841 0.01422198 HMOX1 1.270679538 4.91E-07 KHDC1L 1.255552875 0.002885723 MIR22HG 1.255141861 0.003775261 RUSC1-AS1 1.25298856 1.76E-05 APOBEC3A_B 1.251464268 3.07E-05 APOBEC3B 1.241100387 7.08E-05 CTNNAL1 1.239795702 1.13E-06 ST8SIA6 1.237077936 0.046546514 FRZB 1.23320312 0.017210164 HCN2 1.226072269 0.000708396 MAGEB17 1.224542682 0.007207804 COL13A1 1.224364562 1.82E-07 SLC2A3 1.221160409 0.000105027 LATS2 1.214279743 5.30E-06 INSIG1 1.213403986 3.37E-07 FAT4 1.213357598 0.017184239 HIST2H2AA4 1.213072369 3.92E-05 MVK 1.205548945 6.52E-06 SCML1 1.203837463 0.002053141 CLU 1.197053293 1.44E-05 PRF1 1.193795887 0.004375113 BTG2 1.19161055 3.25E-06 FAM46B 1.190457624 3.64E-05 CDH23 1.181510402 2.35E-06 GLIS3 1.175293771 0.000966608 STMN1 1.173702828 2.65E-05 ZNF367 1.169079323 1.54E-06 FN1 1.167217489 3.42E-05 DHCR7 1.166595866 4.92E-06 FDPS 1.165667072 4.65E-06 FERMT2 1.165518761 2.56E-06 TGFBR2 1.164515049 3.92E-05 DACT1 1.160249716 0.024662584 NPPA-AS1 1.155862708 0.002522946 TNFSF18 1.150291325 0.038286093 RABL6 1.14892906 7.72E-08 TFAP2C 1.145643929 0.000535152 MAP3K14 1.143761754 2.84E-05

Gene Symbol Log2FoldChange Padjusted C10orf54 1.141129624 4.99E-06 MIR3917 1.139663121 0.000141869 OSCP1 1.13487962 0.039185537 HOXB6 1.129468461 9.03E-06 FGF9 1.127156717 0.028558207 HIST1H2BE 1.124927064 0.031537451 FBXL8 1.115958155 0.046658616 ANGPTL4 1.114632229 0.000603426 ZNF804A 1.108144162 0.012700659 ATOX1 1.10795201 0.000222977 AIM1L 1.107638248 0.002926957 LOC541471 1.101489506 1.25E-05 MVD 1.0983503 6.33E-06 UTRN 1.097355964 8.11E-06 RHBDF1 1.093494769 0.000560668 CLCN6 1.081943398 0.000304497 HIC2 1.078093224 0.000261925 DACT3 1.073851108 0.018286985 GPR37 1.070171895 0.049465888 MOK 1.067324095 0.012077083 NEDD4L 1.066968256 3.64E-05 GALC 1.066645261 0.035317051 GRIK4 1.063637344 0.046775472 COL17A1 1.063184885 0.006898334 BTBD11 1.06193054 0.022852381 HIST1H2BK 1.059389609 2.34E-05 HCN3 1.056397494 0.017696784 ZC3H12A 1.055701355 0.01422198 TMCC2 1.054568751 0.006869667 NUAK2 1.049708839 0.026630528 HMGCR 1.047461383 6.83E-05 CORO2A 1.045949892 0.000202494 PCYT2 1.043090346 4.53E-05 MAGED2 1.04177321 0.000102088 CNN3 1.037040981 2.88E-05 NAV3 1.02996747 0.000441349 RBM19 1.028727296 0.000543941 TBC1D2 1.023947018 0.000165198 RAD51 1.020997165 0.001430014 SYDE1 1.020676201 0.025369314 COL1A1 1.019604853 0.001438951 MICA 1.015917041 0.000122226 SLC22A4 1.015026191 0.037926268 HOXB-AS3 1.014599797 0.000101764 WDR62 1.008652939 0.000626 CCDC64 1.007358752 0.00129815 SAMD4A 1.006915018 0.000163276

Gene Symbol Log2FoldChange Padjusted SLC12A4 1.005935812 0.000141869 GCSAM 1.005614499 0.042840542 SNRPA1 1.002223371 0.000156727 CDC25A 1.002210455 0.001111287 TET3 1.000756162 8.07E-05 PIGV -1.007695836 0.018374243 TMEM218 -1.014264652 0.027174568 SLC38A2 -1.022489066 0.000875566 EPDR1 -1.024460898 0.034295955 DARS2 -1.025137456 0.00160888 SLC1A4 -1.030176297 0.030931845 TMEM144 -1.032509259 0.006502963 CCDC3 -1.032678615 0.041640797 PEF1 -1.053822406 0.001802585 THNSL1 -1.055513573 0.014516827 SLC3A1 -1.056885094 0.00517568 GTPBP2 -1.062515937 0.000601133 HCRTR1 -1.065027408 0.004528253 FERMT1 -1.072591901 3.92E-05 CKMT2 -1.073502971 0.001267073 ULBP1 -1.075684098 0.003190966 ITPR1 -1.077488134 0.007508068 OR51B4 -1.081859535 0.032219739 DNAH2 -1.082974998 0.015023227 ZFAS1 -1.083651252 0.000325757 CD68 -1.087069207 0.001732083 GRB10 -1.089983005 0.000565881 LITAF -1.091243268 0.000169183 UCA1 -1.09387229 0.001743485 MRPL24 -1.095245069 0.000164496 LOC253039 -1.109276751 0.006723544 RPS12 -1.109279786 0.000800321 TRIM25 -1.110153857 0.000114575 LOC100130691 -1.113756564 0.01699272 NCS1 -1.119237916 4.54E-05 DDIT4 -1.130930208 0.000174289 SLC24A1 -1.13490178 0.002420998 SNORA33 -1.140240031 0.00012136 BEX2 -1.141227621 0.000102088 RPS6KA5 -1.141424522 0.00023194 RPL29 -1.142968847 0.000162198 FKTN -1.144996216 0.000745936 EPB41L4A -1.145314113 0.00129815 IFI35 -1.146647112 0.010024299 SLC35F1 -1.148328027 5.04E-05 NNT -1.152076589 2.34E-05 ATF5 -1.152526808 0.001672219

Gene Symbol Log2FoldChange Padjusted BBS10 -1.152713814 8.07E-05 KITLG -1.154082016 7.11E-05 ALPK1 -1.156574875 0.0009584 DCBLD2 -1.16346807 0.000102088 PCDH7 -1.168236645 3.73E-05 R3HDM2 -1.16959666 6.79E-05 TMEM246 -1.173480212 0.002558826 RPL29P2 -1.177215754 0.008283225 GAS5 -1.188066616 6.99E-05 ITGA3 -1.198993442 0.000165198 GPRC5A -1.20256373 1.44E-05 RHBDD1 -1.20574209 2.34E-05 S100A14 -1.206185745 0.000164888 SLC7A11 -1.211372482 8.57E-06 AKAP12 -1.212822002 0.003211413 SNHG8 -1.214170381 1.82E-05 LPHN3 -1.220108962 0.000122226 CCR3 -1.222093264 0.002278328 C1orf116 -1.23349988 0.000875566 ID1 -1.234086618 4.57E-06 SLC35A5 -1.236596865 0.001221232 FGF19 -1.242650616 2.88E-05 PGAP3 -1.25301882 0.013683589 CYP24A1 -1.257020706 2.49E-07 DEPTOR -1.259534668 0.016992622 SLC7A11-AS1 -1.265255546 4.92E-06 LOC201651 -1.270898486 0.000886954 TUBE1 -1.273431518 6.84E-05 ERRFI1 -1.273544497 6.61E-05 GAS5-AS1 -1.276941881 0.001038963 AREG -1.28326392 7.29E-06 LAT2 -1.304819119 0.003658918 PSAT1 -1.306442297 4.79E-07 EPB41L4A-AS1 -1.309199316 0.00030989 C6orf48 -1.320673287 1.33E-05 ACSL5 -1.325278951 2.85E-06 PHLDA1 -1.352070349 6.07E-06 SLC7A1 -1.367506855 1.37E-07 DUSP19 -1.377500279 0.049670732 HSD17B8 -1.378499597 0.000563789 SLFN5 -1.380940343 5.30E-06 C2orf81 -1.392534252 6.16E-06 ASNS -1.407772671 7.23E-08 LONRF2 -1.413062997 0.027404145 TEX19 -1.42645663 0.009751368 MFSD2B -1.44326029 1.48E-07 FONG -1.475222061 0.014021587

Gene Symbol Log2FoldChange Padjusted HMGA1 -1.488030154 2.62E-05 GABRB3 -1.494395814 2.64E-06 PEX19 -1.50674589 1.42E-08 BEST1 -1.51665984 0.014516827 LURAP1L -1.51845435 0.000601133 RBCK1 -1.519043516 1.37E-07 KRT13 -1.568260777 0.032596874 MUC12 -1.588641421 0.005129664 EREG -1.613927541 9.23E-11 DKK1 -1.630068888 2.43E-09 VGF -1.633098758 6.67E-12 FCGBP -1.650917391 0.001127461 FAM19A3 -1.663799785 0.049027209 DMGDH -1.668679332 0.009796759 GATM -1.670318054 0.043849964 PCK2 -1.693031229 9.93E-11 PRPH -1.713306244 0.003245804 NRL -1.714134391 4.18E-11 ANGPT4 -1.780143089 0.000537495 TRIB3 -1.799720345 4.17E-11 LOC100287314 -1.800111511 0.000274033 H1F0 -1.815935391 1.34E-11 IL20RB -1.842467357 0.001850485 OR51B2 -1.871231913 0.001438951 TRPV6 -1.928309998 0.035738427 FAM25C -1.960983508 0.041771029 FAM129A -2.155213176 5.46E-09 WTAPP1 -2.193937965 0.000886122 LAMP3 -2.392107766 6.06E-08 COL6A3 -2.405849725 7.14E-05 SNORD15A -2.607773246 0.029396679 CSTA -2.751930255 1.03E-07 MMP1 -2.952719134 3.56E-06 NUPR1 -3.072886888 0.032918502 LOC100132354 -3.189029466 0.000164496 PTK7 -3.45416359 1.10E-36 DYNAP -3.717797392 0.007907692 LCN2 -4.070343871 1.69E-05 LOC442132 -4.206345458 0.01327055 KLHDC7B -4.251155327 8.36E-07