Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Dissecting preleukemic mechanisms in a synthetic model of human T-cell acute lymphoblastic leukemia Tyshchenko, Kateryna 2020

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2021_may_tyshchenko_kateryna.pdf [ 5.89MB ]
Metadata
JSON: 24-1.0395080.json
JSON-LD: 24-1.0395080-ld.json
RDF/XML (Pretty): 24-1.0395080-rdf.xml
RDF/JSON: 24-1.0395080-rdf.json
Turtle: 24-1.0395080-turtle.txt
N-Triples: 24-1.0395080-rdf-ntriples.txt
Original Record: 24-1.0395080-source.json
Full Text
24-1.0395080-fulltext.txt
Citation
24-1.0395080.ris

Full Text

DISSECTING PRELEUKEMIC MECHANISMS IN A SYNTHETIC MODEL OF HUMAN T-CELL ACUTE LYMPHOBLASTIC LEUKEMIA by  Kateryna Tyshchenko  B.Sc., The University of British Columbia, 2018  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Interdisciplinary Oncology)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  November 2020  © Kateryna Tyshchenko, 2020  ii  The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, a thesis entitled:  Dissecting preleukemic mechanisms in a synthetic model of human T-cell acute lymphoblastic leukemia   submitted by Kateryna Tyshchenko in partial fulfillment of the requirements for the degree of Master of Science in Interdisciplinary Oncology  Examining Committee: Andrew Weng, Professor, Pathology and Laboratory Medicine, UBC Supervisor  Florian Kuchenbauer, Associate Professor, Medicine, UBC Supervisory Committee Member  Peter Stirling, Associate Professor, Medical Genetics, UBC Additional Examiner  Additional Supervisory Committee Members: Aly Karsan, Professor, Pathology and Laboratory Medicine, UBC Supervisory Committee Member iii  Abstract  T-cell acute lymphoblastic leukemia (T-ALL) is a blood malignancy that arises from T lymphoid progenitor cells. Even though the overall prognosis is favorable with the current treatment, patients who relapse have considerably worse outcomes. Moreover, pediatric patients suffer from significant long-term chemotherapy side effects. Therefore, we are in need of an improved therapy that targets key carcinogenic molecular mechanisms. There are several models that are commonly used to research T-ALL in vitro and in vivo, including cell lines and mouse models. However, all of them present major limitations that hinder the interpretation of the results in a natural human disease. To solve them, we have previously designed a synthetic T-ALL model developed from oncogene-transduced human cord blood cells. It effectively addresses constraints of other research systems and allows to reproducibly study early and late events in leukemia development. To support the notion that the synthetic model recapitulates bona fide T-ALL, I aimed to characterize it transcriptionally. As a result, I showed that NOTCH1-LMO2-TAL1-BMI1 (NLTB) transduced leukemia is similar to a wide range of patient-derived xenograft samples. Moreover, I discovered that HOXB and VEGF pathways (which are involved in various cancers) are also enriched in preleukemic cells compared to their normal counterparts. Interestingly, both of these pathways are clinically relevant in T-ALL patients, as their components negatively affect event-free survival. Using NLTB synthetic model, I further investigated functional roles of each of these pathways. Notably, I discovered that HOXB3 gene expression promotes preleukemic and leukemic cell growth in vitro. This suggests that it is involved in both initiation and maintenance iv  of T-ALL. On the other hand, while VEGFR3 was overexpressed on the surface of preleukemic cells, it did not show a consistent effect neither on their growth nor on the growth of primary leukemia cells. This encourages further research into alternative mechanisms through which VEGFR3 influences cancer progression. It also suggests that cell context determines downstream effects of a particular pathway. Overall, this study provides a valuable insight into cellular mechanisms that are activated in leukemia initiation and maintenance and can possibly be used in targeted therapy for T-ALL patients.   v  Lay Summary  T-cell acute lymphoblastic leukemia (T-ALL) is an aggressive immune system cancer of immature T-lymphocytes. Better understanding of its molecular mechanisms is required to address long-term chemotherapy side effects in children and a high incidence of relapse in adults. A human synthetic T-ALL model allows to study critical leukemogenic events during cancer development. I determined that this system is transcriptionally similar to already established models from patient cells. Moreover, I found that HOXB and VEGF pathways are activated in genetically-defined preleukemic cells and are clinically relevant in natural human malignancy. Functionally, HOXB3 gene significantly promotes expansion of cancer cells during both early and late stages of a synthetic disease. However, VEGFR3 activity does not consistently affect preleukemia or leukemia cell growth. Overall, this work gives an insight into the functional roles of two pathways activated in T-ALL, as well as defines possible new therapeutic targets based on genetics of the disease. vi  Preface  Research from this work has been performed in the laboratory of Dr. Andrew P. Weng at the Terry Fox Laboratory, BC Cancer Agency. Human cord blood cells were used according to the University of British Columbia BC Cancer Research Ethics Board protocols with Ethics Certificate Numbers H13-02326 and H20-00710. Animal use was approved and performed according to the University of British Columbia Animal Care Committee protocols with Animal Care Certificate Numbers A14-0098 and A18-0047. Cell sorting was performed at the Terry Fox Laboratory Flow Cytometry Core Facility. The bioinformatic analyses from Chapter 3 sections 3.1 and 3.2 have been designed by me with guidance from Dr. Manabu Kusakabe and Dr. Andrew P. Weng. The functional experiments from Chapter 3 sections 3.1 and 3.2 have been designed by Dr. Manabu Kusakabe and Dr. Andrew P. Weng. The bioinformatic and functional experiments from Chapter 3 section 3.3 have been designed by me with guidance from Dr. Andrew P. Weng.  A version of Chapter 3 sections 3.1 and 3.2 (and relevant sections in Chapters 1, 2, 4, Appendices) has been published in:  Kusakabe M, Sun AC*, [Tyshchenko K]*, Wong R, Nanda A, Shanna C, Gusscott S, Chavez E, Lorzadeh A, Zhu A, Hill A, Hung S, Brown S, Babaian A, Wang X, Holt RA, Steidl C, Karsan A, Humphries RK, Eaves CJ, Hirst M, and Weng AP. (2019). Synthetic modeling reveals HOXB genes are critical for the initiation and maintenance of human leukemia. Nature Communications, 10(1): 2913, https://doi.org/10.1038/s41467-019-10510-8 (Creative Commons Attribution 4.0 International License https://creativecommons.org/licenses/by/4.0/) * indicates equal contribution vii  The manuscript was written by Dr. Manabu Kusakabe and Dr. Andrew P. Weng; I wrote relevant thesis sections based on the manuscript.  I performed and interpreted all bioinformatic analyses from Chapter 3 except for the following:  Figure 3.4a (PCA plot generated by Rachel Wong)  I also conducted and analyzed all the experiments from Chapter 3 except for the following:   Figure 3.6a, B.1, B.2 (shRNA screen performed and analyzed by Rachel Wong)   Figure 3.6b (#105_LEP tracking performed and analyzed by Rachel Wong)  Figure 3.6c (T-ALL cell lines tracking performed and analyzed by Dr. Samuel Gusscott)   Figure 3.7 (WIA results analyzed by Dr. Andrew P. Weng)   Figure 3.14 (HUVEC shVEGFR3 experiment performed and analyzed by Dr. Samuel Gusscott)  A.2 (mRNA expression data analyzed by Dr. Andrew P. Weng)  In addition, Dr. Manabu Kusakabe collected data for the RNA-Seq synthetic model datasets used in figures in Chapter 3 and A.2, and Ann C. Sun and I collected data for the RNA-Seq primary CB leukemia dataset used in A.7; sequencing was performed by the group of Dr. Martin Hirst and Canada’s Michael Smith Genome Sciences Centre. Ann C. Sun provided raw counts for A.7 analysis. Scott Brown from Dr. Robert A. Holt’s group generated MiXCR results for Figure 3.1b analysis. Dr. Samuel Gusscott made shRNA lentivirus against VEGFR3. viii  Table of Contents  Abstract ......................................................................................................................................... iii Lay Summary .................................................................................................................................v Preface ........................................................................................................................................... vi Table of Contents ....................................................................................................................... viii List of Tables ............................................................................................................................... xii List of Figures ............................................................................................................................. xiii List of Symbols and Abbreviations ............................................................................................xv Acknowledgements .................................................................................................................... xix Dedication .....................................................................................................................................xx Chapter 1: Introduction ................................................................................................................1 1.1 T-cell acute lymphoblastic leukemia (T-ALL) ............................................................... 1 1.1.1 Clinical diagnosis and treatment ............................................................................. 1 1.1.2 Genetic classification of T-ALL ............................................................................. 2 1.1.3 Analysis of main T-ALL models ............................................................................ 2 1.1.3.1 Cell lines ............................................................................................................. 2 1.1.3.2 Transgenic murine models .................................................................................. 3 1.1.3.3 Xenograft models ................................................................................................ 4 1.1.3.4 Synthetic cord blood model ................................................................................ 4 1.2 Thesis hypothesis and objectives .................................................................................... 7 1.3 HOX genes in leukemogenesis ....................................................................................... 8 1.4 Vascular endothelial growth factor pathway in cancer ................................................... 9 ix  1.4.1 VEGFA/VEGFR1 axis.......................................................................................... 10 1.4.2 VEGFR3 axis ........................................................................................................ 11 Chapter 2: Materials and Methods ............................................................................................12 2.1 RNA-Seq analysis ......................................................................................................... 12 2.1.1 Data availability .................................................................................................... 12 2.1.2 Synthetic model of human T-ALL ........................................................................ 12 2.1.2.1 Library preparation ........................................................................................... 12 2.1.2.2 Alignment and QC ............................................................................................ 12 2.1.2.3 Generation of raw counts and downstream normalization ............................... 13 2.1.2.4 Differential expression and pathway analysis................................................... 13 2.1.3 T-ALL patient samples ......................................................................................... 13 2.2 Cell culture .................................................................................................................... 13 2.2.1 Isolation of cord blood progenitor cells ................................................................ 13 2.2.2 NLTB lentiviral construct and transduction ......................................................... 14 2.2.3 Culture of cord blood cells .................................................................................... 15 2.2.4 shRNA constructs and transduction ...................................................................... 15 2.2.5 Culture of adherent cell lines ................................................................................ 16 2.3 Flow cytometry ............................................................................................................. 16 2.3.1 Surface flow cytometry ......................................................................................... 16 2.3.2 Intracellular flow cytometry ................................................................................. 17 2.4 VEGF ligand multiplex assay ....................................................................................... 17 2.5 Cell proliferation assay ................................................................................................. 17 2.6 Well initiation assay ...................................................................................................... 18 x  Chapter 3: Results........................................................................................................................19 3.1 Characterizing transcriptome of the synthetic NLTB model of human T-ALL ........... 19 3.1.1 Validation of the NLTB model ............................................................................. 19 3.1.2 Transcriptome analysis of preleukemic NLTB cells in vitro ................................ 22 3.2 Roles of HOXB gene cluster in preleukemia and established T-ALL .......................... 25 3.2.1 Relevance of HOXB genes in T-ALL patients ..................................................... 25 3.2.2 Impact of HOXB genes on leukemia initiation and maintenance in vitro ............ 31 3.2.2.1 Assessing function of HOXB genes in established T-ALL .............................. 31 3.2.2.2 Assessing function of HOXB3 in preleukemic cells ........................................ 33 3.3 Roles of VEGF pathway in preleukemia and established T-ALL ................................ 36 3.3.1 Relevance of VEGF genes in T-ALL patients ...................................................... 36 3.3.2 Protein expression of VEGF pathway components in a synthetic NLTB model . 41 3.3.3 Selection and validation of shRNAs against VEGFR3 ......................................... 45 3.3.4 Impact of VEGFR3 on leukemia initiation and maintenance in vitro .................. 48 3.3.4.1 Assessing cell growth effects of VEGFR3 in established T-ALL .................... 48 3.3.4.2 Assessing ligand-dependent effect of VEGFR3 on preleukemic cell growth .. 51 3.3.4.3 Assessing ligand-independent effect of VEGFR3 on preleukemic cell growth 56 Chapter 4: Discussion ..................................................................................................................58 4.1 Summary and significance of research ......................................................................... 58 4.2 Future directions ........................................................................................................... 62 References .....................................................................................................................................64 Appendices ....................................................................................................................................77 Appendix A Supplementary Figures ......................................................................................... 77 xi  A.1 NLTB and HOXB gene expression in T-ALL cell lines .......................................... 77 A.2 Expression of HOXB and flanking genes in CB cell populations in vitro ............... 78 A.3 Testing covariate linearity assumption of Cox regression for VEGF genes in T-ALL patients .................................................................................................................................. 79 A.4 Maxstat cutpoint selection for VEGF genes in T-ALL patients ............................... 80 A.5 VEGFR3 surface protein level variation in preleukemic cells in vitro ..................... 81 A.6 VEGFR3 intracellular protein levels in preleukemic cells in vitro ........................... 82 A.7 VEGF gene expression in primary CB leukemias .................................................... 83 A.8 VEGFR3 surface protein level upon addition of 100 ng/mL VEGFC in preleukemic cells in vitro........................................................................................................................... 84 Appendix B Supplementary Tables .......................................................................................... 85 B.1 List of shRNA clones against HOX genes, FLT4 and non-targeting controls ......... 85 B.2 List of primers used for NGS library construction ................................................... 88  xii  List of Tables  Table 3.1: Pathway analysis using RNA-Seq data from CB cell populations in vitro ................. 24  xiii  List of Figures  Figure 1.1: Functional characterization of in vitro preleukemic NLTB cells ................................. 7 Figure 3.1: Transcriptional characterization of the synthetic NLTB model ................................. 22 Figure 3.2: Differential expression analysis between CB cell populations in vitro...................... 23 Figure 3.3: HOXB gene expression analysis of T-ALL patients .................................................. 26 Figure 3.4: Characterization of T-ALL patients with high HOXB levels based on known prognostic factors .......................................................................................................................... 28 Figure 3.5: Characterization of T-ALL patients with high HOXB levels based on normal T-cell development .................................................................................................................................. 31 Figure 3.6: Effect of HOXB3 on cell growth of established NLTB leukemias ............................ 33 Figure 3.7: Effect of HOXB3 on clonogenic activity of preleukemic NLTB cells ...................... 35 Figure 3.8: VEGF gene expression analysis of T-ALL patients ................................................... 36 Figure 3.9: Association of continuous VEGF gene expression with event-free survival (EFS) .. 37 Figure 3.10: Association of dichotomized VEGF gene expression with event-free survival (EFS)....................................................................................................................................................... 39 Figure 3.11: Association of dichotomized VEGF gene expression with minimal residual disease levels (MRD) on day 29 ................................................................................................................ 40 Figure 3.12: Protein levels of VEGF receptors in CB cells in vitro ............................................. 42 Figure 3.13: Secretion of VEGF ligands by CB and feeder cells in vitro .................................... 44 Figure 3.14: Test of VEGFR3 knockdown in HUVEC cells ........................................................ 46 Figure 3.15: Test of select shRNAs against VEGFR3 in preleukemic NLTB cells ..................... 47 Figure 3.16: Protein levels of VEGFR3 in primary CB leukemias in vitro .................................. 49 xiv  Figure 3.17: Cell growth of established NLTB leukemias upon VEGFR3 knockdown .............. 50 Figure 3.18: Addition of VEGF ligands to CB cells ..................................................................... 51 Figure 3.19: Cell growth of preleukemic NLTB cells upon VEGFR3 knockdown with exogenous ligand present ................................................................................................................................ 53 Figure 3.20: Effect of VEGFR3 ligand on clonogenic activity of preleukemic NLTB cells ....... 56 Figure 3.21: Cell growth of preleukemic NLTB cells upon VEGFR3 knockdown ..................... 57  xv  List of Symbols and Abbreviations  ALL  Acute lymphoblastic leukemia AML  Acute myeloid leukemia BFP  Blue fluorescent protein BMI1  B cell-specific Moloney murine leukemia virus integration site 1 bp  Base pair CB  Cord blood CD  Cluster of differentiation CDKN2A Cyclin-dependent kinase inhibitor 2A CI  Confidence interval CNS  Central nervous system DAPI  4′,6-diamidino-2-phenylindole dbGaP  Database of genotypes and phenotypes DE  Differential expression DNA  Deoxyribonucleic acid DL1  Delta-like 1 E2A  Equine rhinitis A virus 2A EdU  5-Ethynyl-2´-deoxyuridine EFS  Event-free survival ETP  Early T-cell precursor ELDA  Extreme limiting dilution analysis F2A  Foot-and-mouth disease virus 2A xvi  FACS  Fluorescence activated cell sorting FAK  Focal adhesion kinase FBS  Fetal bovine serum FDR  False discovery rate FLT1  FMS-related tyrosine kinase 1 FLT4  FMS-related tyrosine kinase 4 FLT3L  FMS-like tyrosine kinase 3 ligand FPKM  Fragments per kilobase of exon model per million reads mapped GFP  Green fluorescent protein GSEA  Gene set enrichment analysis h  Human HIF-1 Hypoxia inducible factor 1 alpha HOX  Homeobox HSC  Hematopoietic stem cell HTS  High throughput sampler HUVEC Human umbilical vein endothelial cell IgG1  Immunoglobulin G1 IL  Interleukin LMO2  LIM domain only 2 LSGS  Low serum growth supplement LYL1  Lymphoblastic leukemia derived sequence 1 mL  Milliliter mm  Millimeter xvii  MRD  Minimal residual disease mRNA  Messenger ribonucleic acid NCBI   National center for biotechnology information NES  Normalized enrichment score ng  Nanogram NGFR  Nerve growth factor receptor NLTB  NOTCH1-LMO2-TAL1-BMI1 NOD  Non-obese diabetic NOTCH1 Notch homolog 1, translocation-associated NSG  NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ PCA  Principal component analysis PBS  Phosphate-buffered saline PDX  Patient-derived xenograft PGF  Placental growth factor PI3K  Phosphoinositide 3-kinase PlGF  Placental growth factor QC  Quality control Rlog  Regularized logarithm RNA  Ribonucleic acid RNA-Seq Ribonucleic acid-sequencing SCF  Stem cell factor SDF1 Stromal cell-derived factor 1 alpha Scr  Scramble xviii  sh  Short hairpin shRNA Short hairpin ribonucleic acid SRA  Sequence read archive STR  Short tandem repeats T2A  Thosea asigna virus 2A TAL1  T-cell acute lymphocytic leukemia 1 T-ALL  T-cell acute lymphoblastic leukemia TRA  T-cell receptor alpha TRB  T-cell receptor beta TRD  T-cell receptor delta TRG  T-cell receptor gamma M  Micromolar m  Micrometer VEGF  Vascular endothelial growth factor VEGFA Vascular endothelial growth factor A VEGFB Vascular endothelial growth factor B VEGFC Vascular endothelial growth factor C VEGFD Vascular endothelial growth factor D VEGFR1 Vascular endothelial growth factor receptor 1 VEGFR2 Vascular endothelial growth factor receptor 2 VEGFR3 Vascular endothelial growth factor receptor 3 WIA   Well initiation assay xix  Acknowledgements  I would like to thank my supervisor, Dr. Andrew P. Weng, for his guidance and advice during my studies. Graduate school under his supervision was an invaluable experience, and I very much appreciate opportunities he has given me. I would also like to extend my gratitude to all Weng lab members. I especially thank Dr. Samuel Gusscott and Gabriela C. Segat, who kindly offered me their technical help, as well as moral support, throughout my degree. Special thank you goes to my supervisory committee members, Dr. Aly Karsan and Dr. Florian Kuchenbauer, for their useful insights and suggestions for my project. I would also like to acknowledge the help of Vivian Lam with the VEGF ligand multiplex assays, and assistance of Dr. Wenbo Xu, Dr. Guillermo Simkin and Jubin Kim with cell sorting and flow cytometry assays.  Importantly, I am very grateful to my family. I cannot begin to express my thanks for their wholehearted support and encouragement; I would not be where I am today without them. xx  Dedication     To my parents 1  Chapter 1: Introduction 1.1 T-cell acute lymphoblastic leukemia (T-ALL) 1.1.1 Clinical diagnosis and treatment T-cell acute lymphoblastic leukemia (T-ALL) is an aggressive blood malignancy defined by uncontrolled proliferation of immature T-lymphocytes. It accounts for 12%-15% of ALL cases in children [1] and around 25% in adults [2]. Patients usually display cytopenia and high number of leukemic blasts, and might have enlarged organs (for example, spleen and/or lymph nodes) [2,3]. Microscopy and immunophenotypic assays are required to characterize cancer cells and correctly diagnose T-ALL [4]. Cytogenetics can also be used to determine genetic subtype of a patient’s disease. While overall prognosis for T-ALL patients is favorable with a 5-year event-free survival of ~85% [1], with current treatment around 15% of children and 40% of adults suffer from relapse with inferior outcomes [1,5]. Administered regimen includes a three-phase chemotherapy over the course of around 2 years [4]. The first phase after diagnosis is induction, when patients receive a glucocorticoid, vincristine, asparaginase and, if required, anthracycline. The second phase, intensification, aims to prevent expansion of the remaining cancer cells. It includes methotrexate and mercaptopurine, as well as drugs from the first phase. Minimal residual disease levels are tracked by physicians to predict the possibility of relapse. The third, longest, phase is continuation, when methotrexate and mercaptopurine are administered along with other medications if required. It is also important to note that standard chemotherapy treatment leads to a currently unsolved issue of long-term side effects in pediatric patients [5]. In addition, highly toxic cranial irradiation might be used to reach leukemic cells in the central nervous system (CNS). 2  1.1.2 Genetic classification of T-ALL There are several common key genetic lesions in T-ALL that were recognized over the years. NOTCH1 activating mutations are found in more than half of the patients [6] and are able to generate disease in mice [7]. In addition, deactivating mutations in FBXW7, an ubiquitin ligase involved in NOTCH1 degradation, promotes downstream pathway effects in around 10% of patients [8]. Moreover, more than 70% of samples have tumor suppressor CDKN2A/CDKN2B gene deletions, responsible for cell cycle regulation [9].  Furthermore, T-ALL has been divided into subgroups based on oncogene activation. They are either identified through gene rearrangement, gene expression or both [10]. Most commonly observed activations include those of transcription factors TAL1/ TAL2, TLX1/TLX3, NKX2.1, LYL1 and LMO1/LMO2 [10]. Notably, TAL1 and LMO2 act together during DNA binding both in normal hematopoiesis and leukemogenesis [11,12]. However, while in general oncogene subtypes discerned by different research groups are similar, there are still some important discrepancies as to how T-ALL is classified [10,13]. 1.1.3 Analysis of main T-ALL models 1.1.3.1 Cell lines There are several main models that are used to explore molecular mechanisms behind T-ALL pathogenesis. For instance, human cell lines are established in culture from primary cells of a patient. These cells are immortalized and can be kept in vitro or injected into immunodeficient mice as xenograft models [14–16]. There are multiple T-ALL cell lines that are commonly used to study various features of this disease, including Jurkat and HPB-ALL [16]. While this method provides an efficient way to study leukemic mechanisms, it is important to consider its limitations. Firstly, it has been found that T-ALL cell lines are highly heterogeneous [17] – thus 3  negatively contributing to reproducibility of the results. Secondly, cell lines are cultured in vitro for a long time. While this is advantageous from a technical point of view, at the same time these cells might not reflect real biochemical mechanisms happening during cancer due to selection in a dish over the course of many years.  1.1.3.2 Transgenic murine models Mus musculus is a commonly used organism for modelling cancer, and in particular T-ALL. It allows to study natural leukemogenic processes in vivo, removing the caveat of artificial cell selection. Mice also provide multiple technical advantages to researchers, such as small size (which allows for easier handling than in the case of larger animals) and short life span (which allows to obtain faster results) [18]. Moreover, in general, there are numerous genetic and physiological similarities between mice and humans [18]. We can thus apply various research findings from mice to human biology.  Multiple models have been successfully designed to study T-ALL, such as Lck-driven TAL1 or LMO2 transgenic mice [19]. However, in the context of blood cancers they also exhibit several important limitations that affect the way we interpret the data. For instance, there are major immunophenotype differences between human and mice hematopoietic stem cell (HSC) populations [20]. Thus, cross-reference between the species is not always possible. Moreover, 41%-89% of transcription factor (TF) binding events differ between the species [21]. This suggests that in the case of T-ALL, possible targets of TFs (such as LMO2) that were identified in mice would need to be rigorously checked for relevance in patients. Overall, depending on a researcher’s question of interest, transgenic mouse models might not be able to accurately translate obtained findings to human T-ALL. Furthermore, there are cancer mechanisms that are 4  only observed in human disease progression [22] and would thus never be detected in mouse experiments. 1.1.3.3 Xenograft models Patient-derived xenografts (PDXs) is another standard way to study T-ALL in vivo. Notably, injection of cancer cells into mice addresses some of the issues displayed by transgenic mouse models. For instance, we are able to effectively determine which mechanisms are activated in patient cells themselves, without the need to worry about species differences. Because of that, PDXs are often used for drug studies and other translational research [23].  However, multiple disadvantages still exist in these models. First of all, as recipient mice are immunodeficient, we are not able to study interactions of cancer cells with immune cells [18]. Secondly, there is a large genetic variation between T-ALL patients [24]. Therefore, it may negatively affect reproducibility of the data and also makes it more difficult to dissect molecular mechanisms and their genetic regulation. Thirdly, these models have several technical issues, such as difficulties with human cell engraftment and availability of patient cells. For example, the engraftment frequency could differ depending on patient mutations [25]. Moreover, around 30% of T-ALL samples do not engraft in mice in the first 50 days [25]. This, in turn, contributes to the second point, as it is harder to obtain a large number of replicates. Hence, while currently PDXs remain the most effective way to study natural human disease, there are still major limitations remaining for T-ALL research. 1.1.3.4 Synthetic cord blood model To address the crucial shortcomings of other T-ALL models, we sought to develop a novel human model using cord blood CD34+ progenitor cells [26]. Through lentiviral constructs, we are able to define a specific type of T-ALL by overexpressing select genes. Cells are 5  transduced with chosen oncogenes, such as TAL1/LMO2 or LYL1 [2], based on a disease class of interest. Moreover, we are able to model commonly observed NOTCH1 activation by its overexpression, as well as deletion of CDKN2A in patients through addition of BMI1 [27–29]. For the purpose of this thesis, I will therefore further focus on NOTCH1-LMO2-TAL1-BMI1 (NLTB) model and its applications. This combination represents common gene dysregulations found in groups of T-ALL patients as mentioned above. Transduced cord blood cells are cultured on OP9-DL1 mouse feeder cells that provide a consistent source of NOTCH1 ligand, Delta-like 1 (DL1) [30]. Therefore, this culture system supports generation of T-cells from CD34+ progenitors due to active NOTCH1 signaling at physiological levels [31,32]. By using such a model, in vitro cord blood populations recapitulate several cell states, including both preleukemic state (i.e. transduced oncogene-overexpressing cells) and normal state (i.e. non-transduced cells that differentiate into T-cells). As expected, the NLTB population is able to expand in a dish (Figure 1.1b) and outcompete all other groups in culture already by day 24 (Figure 1.1a). Moreover, they retain their immature phenotype, suggesting that oncogene combination imposes a differentiation block (Figure 1.1c). In vitro studies on these populations, therefore, allow to efficiently investigate cell mechanisms behind T-ALL development. In particular, this is an effective way to determine which previously unidentified leukemic processes may happen in patients before cancer is first diagnosed.  6   7  Furthermore, cultured cells can be injected into immunodeficient NSG mice, producing a serially transplantable leukemia. Tissue histology and cancer immunophenotyping support the notion that the disease is indeed T-ALL [26]. Hence, various imaginative studies can be performed with this method to explore human cell transformation in vivo. Overall, synthetic NLTB cord blood model is a genetically defined system indistinguishable from spontaneous human tumors. Importantly, its uniqueness lies in the fact that we are able to evaluate events that happen during preleukemia. This is not easily possible with other systems. Moreover, it provides solutions to multiple disadvantages of other available platforms, such as reproducibility and technical limitations (for example, engraftment rate). 1.2 Thesis hypothesis and objectives To improve currently available regimen with regard to outcome and detrimental side effects, we need to better understand disease mechanisms. This will allow to establish more specific molecular targets for therapy.  Notably, use of the previously mentioned synthetic NLTB cord blood model provides an efficient platform to determine molecules that affect the process of leukemogenesis. For instance, we are able to study downstream effects of chosen transcription factors, known to be involved in Figure 1.1: Functional characterization of in vitro preleukemic NLTB cells (a) GFP/Cherry flow cytometry tracking of live CD45+ cells in NLTB synthetic model. CB cells were transduced with NLTB lentiviral constructs with GFP/Cherry markers on day 0 and day 5. Cells were cultured on OP9-DL1 feeders over the whole duration of the experiments. (b) Cell growth of NLTB GFP+Cherry+ cells (n=3) calculated based on fold increase between passages. (c) Comparison of immunophenotypes of NLTB and normal cells on day 19 and day 33. MeanSD is plotted based on flow cytometry results. Each point is an independent culture. *p < 0.05; **p < 0.01 (by two-tailed t test with Holm−Sidak correction for multiple comparisons) G+C+: GFP+Cherry+, G-C-: GFP-Cherry-. Figure modified from Synthetic modeling reveals HOXB genes are critical for the initiation and maintenance of human leukemia (Kusakabe et al. Nat Comms 2019) https://doi.org/10.1038/s41467-019-10510-8, accessed August 2020. Used under Creative Commons Attribution 4.0 International License. 8  T-ALL. Specifically, comparison of in vitro NLTB cells before transplantation and their normal counterparts serves as a novel way to identify previously undescribed mechanisms that could promote T-ALL progression during initial transformation. In this work, I hypothesized that NLTB promotes cell growth by upregulating HOXB and VEGF pathways in preleukemic cells in vitro.  Thesis objectives: 1. Transcriptionally characterize synthetic NLTB cord blood model Aim 1.1: Compare gene expression of primary NLTB leukemia to bona fide T-ALL Aim 1.2: Identify differentially upregulated pathways in preleukemic NLTB cells in vitro 2. Determine if top upregulated pathways in preleukemic cells affect T-ALL leukemogenesis Aim 2.1: Establish association between expression of pathway genes and clinical features in T-ALL patients Aim 2.2: Determine the effect of select pathway components on preleukemic and leukemic NLTB cell growth in vitro 1.3 HOX genes in leukemogenesis Homeobox (HOX) genes are a family of conserved transcription factors. In humans, there are four clusters (HOXA/B/C/D), each of which have anterior and posterior genes based on the embryo areas they regulate during its development [33]. Furthermore, such genes as HOXB3 and HOXB4 promote blood cell growth, specifically hematopoietic stem cell self-renewal [34]. However, besides normal development, HOX regions are also implicated in various cancers. Multiple HOX genes are highly involved in the process of leukemogenesis [35]. For example, dysregulation of HOXA cluster happens in patient subgroups of both acute myeloid leukemia (AML) and T-ALL [36,37]. In AML, these genes promote cell growth and 9  chemoresistance [38]. In T-ALL, samples with HOXA overexpression are enriched in alterations of several oncogenic pathways, such as JAK-STAT and Ras [10]. Importantly, JAK3 and HOXA9 together functionally contribute to in vivo leukemic progression [39]. On the other hand, there are conflicting reports on the effect of HOXB cluster overexpression in leukemia. Some findings suggest that upregulation of the anterior genes is indicative of reduced AML aggressiveness in subgroups of patients [40,41], while others – that it promotes myeloid leukemogenesis and/or chemoresistance [42,43]. The discrepancies between these studies could possibly be attributed to genetic heterogeneity of AML patients [41,44]. Interestingly, there is also a significant correlation between HOXB gene expression [41], suggesting that the whole cluster might be regulated by the same mechanism.  Unlike myeloid malignancies, there is currently no published research on the prognostic or functional roles of HOXB genes in T-ALL besides this study [26]. However, based on the HOXB-dependent proliferation observed in HSCs and activated T-lymphocytes [34,45,46], we can hypothesize that anterior genes might have a positive effect on leukemic T-cells. 1.4 Vascular endothelial growth factor pathway in cancer Vascular endothelial growth factor (VEGF) pathway is one of the crucial mechanisms activated during normal development of blood and lymphatic vessels. The pathway consists of three receptor tyrosine kinases (VEGFR1/2/3) and five ligands (VEGFA/B/C/D and PlGF) [47]. Specifically, VEGFR1 binds VEGFA, VEGFB and PlGF; VEGFR2 binds VEGFA, VEGFC and VEGFD; VEGFR3 binds VEGFC and VEGFD. Ligand attachment then leads to receptor dimerization and phosphorylation [47], a well described mechanism for the molecules of this family. However, in addition to canonical homodimerization, VEGF receptors can also 10  heterodimerize with each other – both VEGFR1/VEGFR2 and VEGFR2/VEGFR3 have been observed by multiple groups [47]. VEGF pathway activation also plays a role in malignant development. For the purpose of this introduction, I will be focusing on downstream signaling effects of VEGFR1 and VEGFR3 receptors in cancer, as in general VEGFR2 expression was not detected in T-ALL cells [48,49]. 1.4.1 VEGFA/VEGFR1 axis VEGFA/VEGFR1 loop as well as downstream components of this pathway are indicative of worse clinical prognosis in solid and blood cancers [49–51]. Particularly in hematologic malignancies, both autocrine and paracrine signaling are commonly dysregulated. Cancer itself and normal stroma from tumor microenvironment are able to produce and secrete VEGFA, which then successfully binds VEGFR1 receptor on malignant cells and activates signaling molecules. They are known to further promote cell growth and migration in several types of blood cancer, such as myeloma and AML [52,53]. Notably, well established downstream targets of VEGFR1 include PTK7, AKT and FAK [54].  With respect to ALL, it has been found that hypoxia-inducible factor HIF-1 is often coexpressed with VEGFA, which is in turn associated with worse survival and therapy response [50]. Specifically, one of the roles of VEGFA produced by cancer cells is to promote progression of leukemia into the central nervous system through affecting endothelial cells [55]. Moreover, VEGFA-dependent secretion of downstream matrix metalloproteinases [49] could also promote disease aggressiveness. These results are also indirectly supported by another in vivo study, where gradients of two of the ligands (VEGFA and PlGF) determine the location of leukemic cells through paracrine VEGFR1 activation [56]. In particular, VEGFR1 positively affects ALL migration (notably, exit out of the bone marrow into circulation) and protects cells from 11  apoptosis [56]. Therefore, we can conclude that both VEGFR1 and VEGFA play a significant role in leukemogenesis through regulation of cell localization and survival. 1.4.2 VEGFR3 axis VEGFR3 is another commonly dysregulated component of the VEGF pathway that promotes cancer progression. For instance, autocrine and paracrine VEGFC/VEGFR3 signaling leads to increased cell invasion and migration in several solid cancers [57,58]. Most frequently described downstream targets that have been shown to contribute to these processes are FAK and PI3K pathways [59–61]. Furthermore, non-canonical VEGFR3 signaling has also been observed in multiple, normal and carcinogenic, contexts [62–64]. It similarly promotes cell migration, in addition to cell survival in breast cancer. It is also important to note that both VEGFC and VEGFR3 are associated with advanced metastasis in lung cancer patients [65,66]. Hence, both VEGFC/VEGFR3 and VEGFR3 on its own are capable of making cancer more aggressive. While the role of VEGFR3 and its signaling in leukemia has not been researched as extensively as in solid cancers, there are independent reports supporting its role in blood cancer development. For instance, AML patients have higher cell protein levels of both VEGFR3 and VEGFC compared to normal control samples [67]. However, no survival difference was observed between patients with high and low expression for either component [67]. Nevertheless, it has been functionally observed that paracrine VEGFC/VEGFR3 loop in AML protects cells from apoptosis and promotes proliferation, as well as angiogenesis through COX2 and JNK [68,69]. Thus, based on previous reports, we can hypothesize that VEGFR3 activity could likewise positively affect cancer cell growth in T-ALL. 12  Chapter 2: Materials and Methods 2.1 RNA-Seq analysis 2.1.1 Data availability RNA-Seq data of NLTB synthetic model is available in the European Genome-Phenome Archive (EGA) database (EGAS00001003627, dataset EGAD00001004993); RNA-Seq data for primary CB leukemias in A.7 will be uploaded to an appropriate public repository in the future. RNA-Seq data of PDX samples was obtained from the NCBI SRA database (SRP103099). RNA-Seq data and clinical annotation of T-ALL patients from the COG TARGET study were obtained from the database of Genotypes and Phenotypes (dbGaP) (phs000218/000464). RNA-Seq data of normal human hematopoietic samples was obtained from the National Center for Biotechnology Information Gene Expression Omnibus (GEO) database (GSE69239). RNA-seq data for T-ALL cell lines was obtained from the EGA database (EGAS00001000536).  2.1.2 Synthetic model of human T-ALL 2.1.2.1 Library preparation For EGAS00001003627, TRIzol and PureLink RNA Mini Kit columns (Invitrogen/Thermo Fisher Scientific) were used to isolate total RNA from sorted live cells. PolyA-enriched or ribosomal RNA-depleted (NEBNext rRNA Depletion Kit; New England Biolabs, cat# E6310) strand-specific library construction protocol was performed, followed by sequencing on Illumina HiSeq 2500 instrument (8 samples/lane; paired-end 75bp or 125bp reads were obtained). 2.1.2.2 Alignment and QC For EGAS00001003627, Partek Flow software (version 6.0.17.0503; Partek Inc, St Louis, MO) was used as a QC step to trim the reads from the 3′ end based on quality score (end 13  min quality level (Phred) = 20). Reads were then aligned to GRCh37/hg19 reference human genome with STAR 2.5.2b [70] in Partek Flow software. 2.1.2.3 Generation of raw counts and downstream normalization For EGAS00001003627, raw gene counts were obtained with featureCounts function in Rsubread v1.24.2 [71] (R version 3.3.1 [72]). GRCh37 Ensemble release 75 was used as annotation. Genes with total raw counts of equal to or less than 1 across all samples were filtered out. Rlog function in DESeq2 [73] was used to obtain log2 values from raw counts and normalize for library size. If different datasets were combined, ComBat function in sva package was used to correct for batch effects (with non-parametric adjustments and biological covariates) [74]. 2.1.2.4 Differential expression and pathway analysis Filtered raw counts were used as an input for DESeq2 to perform differential expression analysis. The design formula was dependent on a comparison of interest. Differentially expressed genes were then input into Reactome pathway database online tool [75]. 2.1.3 T-ALL patient samples Rlog values were obtained following methods from section 2.1.2.3. FPKM values were obtained from Liu et al (2017) [10]. VEGF clinical analysis was performed with R version 3.6.1. Packages survival v3.1-11 [76,77] and survminer v0.4.7 [78] were used for survival analysis. 2.2 Cell culture 2.2.1 Isolation of cord blood progenitor cells CD34+ CB cells were purified with Rosette-Sep/EasySep human CD34-positive selection kit (StemCell Technologies). They were then plated into 96-well round-bottom plates with 5 14  μg/cm2 fibronectin (StemCell Technologies) in the following media for 16-24 hours: StemSpan SFEM II (StemCell Technologies) with 100 ng/mL human SCF, 50 ng/mL human TPO, 100 ng/mL human FLT3L and 20 g/mL human LDL; or with 10 ng/mL human SCF, 20 ng/mL human TPO, 20 ng/mL human IGF2 and 10 ng/mL human FGF-acidic (Peprotech) with or without 8M cyclosporin H (Toronto Research Chemicals Inc) for additional 16-17 hours. In some experiments, 2.5 g/mL fungizone (Thermo Fisher Scientific) was added to coculture for the first 5 days. 2.2.2 NLTB lentiviral construct and transduction Human NOTCH1ΔE allele, TAL1, BMI1, and mouse Lmo2 cDNAs were obtained from Dr. J Aster (Boston), Harvard PlasmID, and Dr. E. Lawlor (UCLA). Equine rhinitis A virus 2A (E2A) peptide [79] was used to link NOTCH1ΔE and GFP cDNAs. Thosea asigna virus 2A (T2A), foot-and-mouth disease virus 2A (F2A), and E2A peptides were likewise used for LMO2, TAL1, BMI1 and Cherry cDNAs, respectively. They were then cloned into pRRL-cPPT/CTS-MNDU3-PGK-GFP-WPRE [80]. 293T cells were transfected with second-generation packaging/envelop vectors pCMV dR8.74 (Addgene #22036), pRSVRev (Addgene #12253), and pCMV VSV-G (Addgene #8454), using polyethyleneimine HCl MAX (Polysciences). The virus supernatants were then obtained by ultracentrifugal concentration: 25000 rpm for 90 min at 4°C using Beckman SW32Ti rotor. Concentrated lentiviral supernatants were added on top of cord blood cells in 96-well plates for 6 hours. Cells were then transferred on plates with OP9-DL1 feeders. 15  2.2.3 Culture of cord blood cells CB cells were cultured in MEM and 20% FBS (Gibco/Thermo Fisher Scientific) media with 1mM sodium pyruvate, 2mM GlutaMAX (Gibco/Thermo Fisher Scientific) and antibiotics (penicillin and streptomycin), as well as the following cytokines: 10 ng/mL human SCF, 5 ng/mL human FLT3L and 3 ng/mL mouse IL-7 (Peprotech). The cells were kept on top of confluent adherent OP9-DL1 feeders. 2.2.4 shRNA constructs and transduction shRNA constructs targeting genes of interest were prepared using 5mL overnight bacterial cultures (B.1). Clones were pLKO-based (from RNAi Consortium shRNA library (GPP Web Portal). They were sourced from the University of British Columbia Centre for High-Throughput Biology (CHiBi) or cloned via annealing oligos. For HOX shRNA screen, a single plasmid DNA prep (Qiagen) was sequenced on an Illumina MiSeq to detect individual clones. Lentiviral constructs for the screen were prepared separately, combined and added to the cells. We aimed to have a single lentiviral integration in most cells (target multiplicity of infection = 0.3). Each shRNA clone was represented at ~1000-fold. Logarithmic growth phase was maintained through regular cell passaging.  RNAi Consortium (GPP Web Portal) protocols were used to PCR amplify shRNA hairpins from DNAzol extracted genomic DNA (Thermo Fisher Scientific). Q5 High-Fidelity 2X Master Mix (New England Biolabs) was used for amplification (30 cycles total; B.2). Precast E-Gel EX 2% agarose gels (Invitrogen/Thermo Fisher Scientific) using QIAquick gel extraction kit (Qiagen) were used to gel purify 300bp amplicons. Illumina MiSeq instrument was used for paired-end sequencing. Reads were aligned to a shRNA reference sequences list with 0 16  mismatches allowed. EdgeR package [81] was used for differential representation analysis, with ggplot2 package [82] used for plotting.  For shVEGFR3 experiments, CB cells were spinfected with shRNA viral supernatants in 4 ug/mL polybrene (Sigma-Aldrich): 500g for 2 hours at 37°C, or for 3.5 hours at 35°C in the case of CB leukemias. HUVEC cells were kept overnight with unconcentrated virus, with 1 g/mL puromycin added for three days the next day. 2.2.5 Culture of adherent cell lines OP9-DL1 feeder cells were obtained from J.C. Zuniga-Pflucker (University of Toronto) and cultured in MEM and 10-20% FBS (Gibco/Thermo Fisher Scientific) media with 1mM sodium pyruvate, 2mM GlutaMAX (Gibco/Thermo Fisher Scientific) and antibiotics. 293T cells were obtained from J. Aster (Brigham & Women’s Hospital, Boston) and cultured in DMEM and 10% FBS (Gibco/Thermo Fisher Scientific) media with 1mM sodium pyruvate, 2mM GlutaMAX (Gibco/Thermo Fisher Scientific) and antibiotics. HUVEC cells were purchased from Gibco and cultured in Medium 200 with Low Serum Growth Supplement (LSGS) (Gibco/Thermo Fisher Scientific). 2.3 Flow cytometry 2.3.1 Surface flow cytometry DAPI (Invitrogen) or LIVE/DEAD Aqua (Invitrogen/Thermo Fisher Scientific) stain was used to gate out dead cells. AccuCheck counting beads (Invitrogen/Thermo Fisher Scientific) were used to calculate absolute cell numbers. Detection of NGFR marker was performed through anti-hCD271 antibody (BD Biosciences). Mouse IgG1, kappa antibody (BioLegend, R&D Systems) was used as isotype control. Samples were run on LSRFortessa and FACSymphony 17  instruments (BD Biosciences). Flow cytometry data was analyzed with FlowJo software (TreeStar). Cell tracking results were plotted with Microsoft Excel. 2.3.2 Intracellular flow cytometry Samples were sorted on FACSAria III and run on LSRFortessa and FACSymphony instruments (BD Biosciences). 1.5% paraformaldehyde was added for 15 minutes to fix the cells, and then cells were permeabilized with ice-cold methanol for 10 minutes at -80C. Mouse IgG1, kappa antibody (R&D Systems) was used as isotype control. Flow cytometry data was analyzed with FlowJo software (TreeStar). 2.4 VEGF ligand multiplex assay Supernatants from CB cell culture were collected to perform MILLIPLEX MAP Human and Mouse Angiogenesis/Growth Factor Magnetic Bead Panel 96-well plate assay (MilliporeSigma). Recommended guidelines from the manufacturer were followed to prepare the samples which were run on Luminex 100 machine with Bio-Plex Manager 6.1 software (Bio-Rad). Results were plotted with Microsoft Excel. 2.5 Cell proliferation assay Click-iT EdU Pacific Blue Flow Cytometry Assay Kit (Invitrogen/Thermo Fisher Scientific) was used to measure proliferative fraction of cells. 20M EdU was added to the samples, which were then incubated for 2 hours at 37C. Recommended guidelines from the manufacturer were further followed to prepare the samples which were run on LSRFortessa instrument (BD Biosciences). Flow cytometry data was analyzed with FlowJo software (TreeStar). Results were plotted with Microsoft Excel.  18  2.6 Well initiation assay Samples were sorted on FACSAria3 (BD Biosciences) into a 96-well plate with feeder OP9-DL1 cells seeded at 5000 cells/well. Upon collection post-sort, cells were trypsinized (Life Technologies) and washed with respective media. AccuCheck counting beads (Invitrogen/Thermo Fisher Scientific) were used to calculate absolute cell numbers. Samples were run on LSRFortessa instrument with HTS option (BD Biosciences). Flow cytometry data was analyzed with FlowJo software (TreeStar). Cell numbers data was plotted with GraphPad Prism 7.02 and 8.0.1 software. Well initiating frequencies were calculated with ELDA tool at http://bioinf.wehi.edu.au/software/elda/ [83].19  Chapter 3: Results 3.1 Characterizing transcriptome of the synthetic NLTB model of human T-ALL 3.1.1 Validation of the NLTB model Our group has previously shown that injection of NLTB-transduced human CB cells leads to development of serially transplantable T-ALL in mice. As our next step, we were interested in further characterizing the model by determining how similar this synthetic disease is to established T-ALL patient-derived xenografts (PDXs). Consequently, I combined and analyzed two different RNA-Seq datasets – one containing GFP+Cherry+, GFP+Cherry- CB leukemias and five T-ALL PDXs from the Weng lab; and the other one containing 17 T-ALL PDXs from PRoXe repository (NCBI SRA Accession SRP103099) [84]. The top 1000 variable genes between PDXs were identified to cluster and calculate the distance between samples with a (1 – Spearman correlation) formula (Figure 3.1a). The resulting analysis suggests that CB leukemias are as transcriptionally similar to a wide range of PDXs as they are to each other. In addition, the obtained results demonstrate that CB NLTB leukemias, even if they come from different donors as shown by STR profiling, are greatly similar to each – supporting the notion that the model is highly reproducible. Interestingly, GFP+Cherry- leukemias clustered apart from GFP+Cherry+ samples. This reveals that the overexpression of LMO2, TAL1 and BMI1 oncogenes in addition to NOTCH1 affects genome-wide gene expression and results in a separate type of synthetic T-ALL. It is known that there are frequent TCR rearrangements in T-ALL [85], and that clonal evolution and/or expansion is observed in relapsed malignancy [86,87]. To investigate whether synthetic T-ALL recapitulates real disease in this aspect, I evaluated TCR clonality of synthetic leukemia RNA-Seq samples (Figure 3.1b) by using the MiXCR tool [88]. Mature T-cells were 20  used as a control, demonstrating polyclonal TRA and TRB rearrangements as expected. In contrast, TCR repertoires of synthetic leukemias were dominated by a single clone. Specifically, CD4-CD8- GFP+Cherry+ leukemias had a major dominant clone of TRG/TRD, while CD4+CD8+ GFP+Cherry+ and GFP+Cherry- leukemias – of TRB. These results suggest that, with respect to clonality, T-ALL arising from CB cells injected into immunodeficient mice is similar to malignancies commonly observed in patients. 21   22  3.1.2 Transcriptome analysis of preleukemic NLTB cells in vitro NLTB-transduced CB cells cultured in vitro provide an efficient way to study the process of leukemia initiation. For instance, GFP+Cherry+ cells undergoing malignant transformation can be compared to their normal counterparts cultured in the same conditions. As a result, we would be able to describe what leukemogenic events happen before the cells are transplanted into mice. Therefore, we prepared an RNA-Seq dataset containing CB cell subsets from in vitro samples. Upon comparing GFP+Cherry+ NLTB preleukemic population to GFP-Cherry- normal cells, I observed 468 genes that were differentially expressed between the two groups (Figure 3.2). In particular, 243 of those were upregulated in NLTB cells compared to normal. I have further performed Reactome Pathway analysis [75] to identify pathways that are enriched in this upregulated gene list (Table 3.1). The first three significant results were associated with NOTCH signaling, which could be explained by supraphysiological expression of NOTCH1 in GFP+Cherry+ cells. Interestingly, the other top two pathways besides NOTCH were related to the activation of HOX and VEGF genes. Hence, I hypothesized that these two pathways are required for the process of leukemogenesis in CB NLTB cells in vitro.  Figure 3.1: Transcriptional characterization of the synthetic NLTB model (a) Unsupervised hierarchical clustering of 17 CB leukemias and 22 T-ALL PDXs. RNA-Seq data was batch-corrected with ComBat. (1-Spearman correlation) distance (shown on the color scale) was calculated by using top 1000 variable genes between 22 PDX samples. STR profiling was used to determine CB donors in synthetic leukemias. n=17 PDX dataset is from PRoXe repository (NCBI SRA Accession SRP103099). (b) TCR clonality of RNA-Seq samples. MiXCR tool was used to show individual CDR3 regions (indicated by colors, which are independent for each sample). G+C+: GFP+Cherry+; G+C-: GFP+Cherry-. 23  Figure 3.2: Differential expression analysis between CB cell populations in vitro mRNA expression heatmap of differentially expressed (DE) genes between GFP+Cherry+ NLTB cells and GFP-Cherry- normal cells. CD45+ cells were sorted based on GFP/Cherry markers expression on day 14 and day 24 of coculture in several independent experiments. DE genes (243 up, 225 down, 468 total) were obtained using log2 fold change threshold of 1 and FDR<0.1. Normalized expression was scaled by gene with mean of 0 and standard deviation of 1. G+C+: GFP+Cherry+; G+C-: GFP+Cherry-. 24  Table 3.1: Pathway analysis using RNA-Seq data from CB cell populations in vitro Reactome pathway analysis of DE list of 243 upregulated genes for GFP+Cherry+ NLTB versus GFP-Cherry- normal cells comparison (log2 fold change threshold = 1, FDR<0.1). List was sorted by increasing FDR. HOX and VEGF pathways are highlighted as bold text. 25  3.2 Roles of HOXB gene cluster in preleukemia and established T-ALL 3.2.1 Relevance of HOXB genes in T-ALL patients As a first step in establishing the role of HOXB genes in T-ALL, we wanted to confirm the relevance of this gene cluster in natural human disease. To address this question, I analyzed an RNA-Seq dataset of 264 T-ALL patients (COG TARGET study, dbGaP phs000218/000464) [10]. A group of these samples expressed HOXB2, HOXB3 and HOXB4 – all of which were significantly correlated to each other (Figure 3.3a). This indicates that if a patient has high levels of HOXB2, they are likely to have high HOXB3 and HOXB4 expression as well. In addition, there was only a small number of patients that expressed HOXB5 gene. To elucidate whether HOXB genes expression is important in T-ALL from the clinical standpoint, 252 patients were split into two groups – HOXB_high and HOXB_low – based on the FPKM values. Specifically, patients that had gene level for HOXB2 or HOXB3 or HOXB4 or HOXB5 greater than 80th percentile and had FPKM>1 for those genes were considered as “high”. I then performed survival analysis by comparing event-free survival of HOXB_high and HOXB_low groups. Interestingly, HOXB_high patients had a significantly worse prognosis compared to HOXB_low patients (Figure 3.3b). Moreover, when I investigated the contribution of each of the HOXB genes to the survival difference, I noted that HOXB4 had the biggest effect on the curve (Figure 3.3c), while HOXB3 also had the same trend. It, however, was not significant with a p-value of 0.06 (Figure 3.3d). HOXB2, on the other hand, did not display any effect on patient event-free survival (Figure 3.3e). HOXB5 was not evaluated due to a low number of patients in the “high” group. Overall, these results suggest that anterior HOXB genes can predict patient survival in T-ALL. 26   Figure 3.3: HOXB gene expression analysis of T-ALL patients RNA-Seq samples from the COG TARGET study (dbGaP phs000218/000464). (a) mRNA expression heatmap of 264 patients. Log2(FPKM+1) values are shown. Pearson correlation matrix is depicted below the plot. Samples are sorted left to right by decreasing HOXB3 expression values.  (b)-(e) Analysis of HOXB gene expression and their association with event-free survival (EFS) in 252 T-ALL patients. Twelve patients were excluded from the analysis due to withdrawn consent, second malignancy or were inevaluable. Log-rank test p-value is shown. (b) HOXB_high defined as patients that had gene levels greater than 80th percentile and FPKM>1 for HOXB2 or HOXB3 or HOXB4 or HOXB5. (c) HOXB4 expression and EFS (d) HOXB3 expression and EFS (e) HOXB2 expression and EFS. NS: not significant. 27   Moreover, we wanted to determine whether HOXB genes can be put in the context of other, already established, prognostic features. For instance, ETP-ALL is a subtype of T-ALL which was described as a poor prognostic factor [89]. Thus, I explored whether patients with high expression of HOXB genes were enriched in ETP-ALL subgroup. Notably, a PCA plot showed that there was little overlap between the two (Figure 3.4a). Statistically, I also observed an opposite, though non-significant (p-value = 0.0681), trend - indicating that HOXB_high samples are distinct from ETP-ALL patients (Figure 3.4b). Furthermore, I performed a similar analysis using previously defined classification by transcription factors [10] (Figure 3.4c). HOXB_high patients were found to be significantly associated with TAL1 and NKX2.1, and also with “unknown” subgroup. On the other hand, HOXB_low patients were enriched in TLX1 and TLX3 subgroups. Therefore, there is a possibility that known genetic features of T-ALL play an important role in HOXB gene upregulation in patients. 28    Figure 3.4: Characterization of T-ALL patients with high HOXB levels based on known prognostic factors 252 RNA-Seq samples from the COG TARGET study (dbGaP phs000218/000464). (a) Principal Component Analysis (PCA) plot of T-ALL patients, showing HOXB, ETP-ALL and near ETP-ALL subgroup classifications. (b) Correlation between patients classified into HOXB and ETP-ALL subgroups. Fisher’s exact test was used to calculate a two-tailed p-value. (c) Correlation between patients classified into HOXB and transcription factor subgroups. Fisher’s exact test and Benjamini-Hochberg multiple testing correction were used to calculate p-values.    29  Previous research has shown that HOXB genes (in particular, HOXB4) are crucial for the process of self-renewal in hematopoietic stem cells, HSCs [45,90,91]. Based on these studies, I hypothesized that patients with high levels of HOXB genes have HSC-like signatures. To test my hypothesis, I used an RNA-Seq dataset which includes normal blood progenitor cells and T-cells. I divided samples into four groups based on PCA plot dimensions (Figure 3.5a): HSC/LMPP, CLP/BCP, CD34+ early thymocytes, and late thymocytes (CD4+CD8-, CD4-CD8+ or CD4+CD8+). As expected, these groups make sense from our knowledge on T-cell development stages. After performing pairwise differential expression between three relevant subsets – HSC/LMPP, early thymocytes and late thymocytes – I used the obtained gene lists as signatures for Gene Set Enrichment Analysis (GSEA) on T-ALL patients (Figure 3.5b). Interestingly, patients with high levels of HOXB genes were found to be significantly enriched in signatures of late thymocytes versus early thymocytes or HSC/LMPP, as well as in early thymocytes versus HSC/LMPP. These findings are the opposite of what was expected, and they indicate that HOXB_high samples are associated with gene signatures of T-cells at later stages of differentiation instead of the HSC stage. However, this is also supported by the transcription factor subgroup analysis above, as TAL1 and NKX2.1 are considered late cortical subtypes [10]. Overall, more research is needed to address which populations are affected by HOXB overexpression and what is the functional role of these genes on leukemia initiation and maintenance. 30   31  3.2.2 Impact of HOXB genes on leukemia initiation and maintenance in vitro 3.2.2.1 Assessing function of HOXB genes in established T-ALL To understand whether HOXB genes functionally contribute to progression of T-ALL, we performed a pooled shRNA screen on a primary CB leukemia, #105_LEP. The screen targeted four anterior HOXB (HOXB2, HOXB3, HOXB4, HOXB5) and four HOXA (HOXA5, HOXA7, HOXA9, HOXA10) genes with 5-10 shRNAs per gene. Additionally, there were three non-targeting controls. The screen showed a significant depletion of four shRNAs against HOXB3 and two shRNAs against HOXB5 (Figure 3.6a), which indicates that CB leukemia cell growth is negatively affected by lower levels of these two genes. We focused on HOXB3 gene for further experiments, as HOXB5 was not expressed in the majority of patient samples (Figure 3.3a). We performed bulk cell number tracking of two primary CB leukemias, #105_LEP and m160. These experiments supported the findings of the shRNA screen. In contrast to shScr control, there was a larger depletion of cells with a knockdown of HOXB3 (Figure 3.6b). In addition, we selected three different T-ALL cell lines – HSB2, PF382, PEER – with high expression of HOXB3 and/or TAL1/LMO2 (in comparison to other cell lines) (A.1). Results similar to primary CB leukemia experiments were observed Figure 3.5: Characterization of T-ALL patients with high HOXB levels based on normal T-cell development (a) Principal Component Analysis (PCA) of 20 normal samples from human bone marrow and thymus (GEO GSE69239) based on protein-coding genes from somatic chromosomes. Four groups obtained through PC1 and PC2 axes are indicated on the plot. (b) Preranked Gene Set Enrichment Analysis (GSEA) of 252 RNA-Seq patient samples from the COG TARGET study (dbGaP phs000218/000464). 1000 permutations were used. Genes were ordered based on (sign(log2 fold change)*(−log10(p-value))) formula, calculated after differential expression analysis between HOXB_high and HOXB_low patients. Gene signatures were obtained after pairwise differential expression analysis with log2 fold change threshold of 2 between HSC/LMPP, early thymocytes and late thymocytes groups. 32  (Figure 3.6c). Thus, we concluded that HOXB3 is required for promoting cell growth during T-ALL maintenance.  33  3.2.2.2 Assessing function of HOXB3 in preleukemic cells HOXB genes were found to be upregulated in preleukemic cells compared to normal (section 3.1.2, A.2). However, it is not clear whether these genes also have a functional effect on leukemia initiation similarly to established leukemia maintenance (section 3.2.2.1). Therefore, I assessed the role of HOXB3, shown to be relevant in established T-ALL (section 3.2.2.1), on clonogenic activity of preleukemic cells. I used a well initiation assay (WIA), which is similar to commonly used colony forming cell assays but is optimized for cells cultured on feeders. I confirmed that the total number of live CD45+GFP+Cherry+NGFR+ cells per well in the shScr-NGFR control was higher in comparison to shHOXB3-NGFR conditions (shHOXB3_643 and shHOXB3_644) (Figure 3.7a). Specifically, shHOXB3_644 cell number was significantly lower even at 10 input cells per well, while shHOXB3_643 – at 50 input cells per well as compared to the control. Furthermore, a threshold of 500 CD45+GFP+Cherry+NGFR+ cells was used to define a positive well. Therefore, a well initiation frequency was calculated for each condition based on this threshold. shScr control had a frequency of ~1 in 41 cells (95% CI: 1 in 30-56 cells) – around 1 in 41 cells is able to expand a well (Figure 3.7b). On the other hand, shHOXB3_643 and shHOXB3_644 had a significantly lower frequency of well initiating cells, ~1 in 240 (95% CI: 1 in 140-430 cells) and ~1 in 510 (95% CI: 1 in 230-1100 cells), respectively (Figure 3.7b). Figure 3.6: Effect of HOXB3 on cell growth of established NLTB leukemias (a) Volcano plot of HOXA/HOXB shRNA screen results. Two replicates of a primary NLTB CB leukemia (Spearman correlation of 0.58) were transduced with shRNA pool against eight HOXA/HOXB genes and three non-targeting controls. Cells were cultured on OP9-DL1 feeders and sorted on day 2 (t0) and days 9/11 (t1) for genomic DNA. Each shRNA is shown as a datapoint. If a gene was targeted by two or more shRNAs with FDR<0.05, it is shown in color. (b)-(c) Tracking of shRNA-transduced cultured populations by flow cytometry. For experiments performed at least in triplicate, meanSD of the initial transduction value is shown. (b) Two primary NLTB CB leukemia samples (c) Three human T-ALL cell lines. 34  These observations imply that HOXB3 expression not only promotes cell growth during leukemia maintenance but is also advantageous for clonogenic expansion of preleukemic cells in the context of NLTB genes.  35   Figure 3.7: Effect of HOXB3 on clonogenic activity of preleukemic NLTB cells Well initiation assay results with shHOXB3 knockdown. CB cells were transduced with NLTB lentiviral constructs on day 0 and day 4. Cells were cultured on OP9-DL1 feeders over the whole duration of the experiment. Cells were transduced with shRNA-NGFR (shHOXB3 or shScr) on day 19. On day 25, CD45+GFP+Cherry+NGFR+ cells were sorted into a 96-well plate. After ~3 weeks of culture, cells were collected with trypsin and analyzed by flow cytometry. (a) Total number of live CD45+GFP+Cherry+NGFR+ cells per each well (n=12-16 for 50, 100 cells/well; n=24-32 for 1, 3, 10 and 25 cells/well). 500 cells threshold for well initiation frequency calculations is indicated as a dotted line. Tukey was used to define box and whiskers. ****p < 0.0001; ***p < 0.001; ns not significant (by Kruskal−Wallis test with Dunn’s correction for multiple comparisons). (b) Well initiating frequency calculations with a threshold of 500 cells. 95% CI are shown as dotted lines. ****p < 0.0001 (by chi-square test). WIC: well-initiating cell. 36  3.3 Roles of VEGF pathway in preleukemia and established T-ALL 3.3.1 Relevance of VEGF genes in T-ALL patients Besides the HOXB cluster, VEGF pathway was also upregulated in NLTB preleukemic cells compared to their normal counterparts (section 3.1.2). Therefore, I performed a similar analysis looking at the clinical relevance of VEGF genes in patient T-ALL samples. A heatmap of gene expression for all eight of VEGF pathway components showed that five of these genes are expressed in the cohort (PGF, FIGF and KDR were not found in >95% of the samples) (Figure 3.8). For further analyses, I focused only on FLT1 (codes for VEGFR1 protein), FLT4 (codes for VEGFR3 protein) and VEGFA, as these genes were significantly upregulated in preleukemic cells versus normal control. Figure 3.8: VEGF gene expression analysis of T-ALL patients RNA-Seq samples from the COG TARGET study (dbGaP phs000218/000464). mRNA expression heatmap of 264 patients. Log2(FPKM+1) values are shown. 37   Cox Proportional Hazards model allows to predict survival using gene levels as a continuous variable. It is a more accurate method than commonly used dichotomization [92]. Univariate analysis for FLT1, FLT4 and VEGFA showed that only FLT1 had a statistically significant hazard ratio greater than 1 (1.32, 95% CI: 1.00-1.75) (Figure 3.9). This indicates that higher expression of FLT1 increases the hazard by 32% and is thus associated with shorter event-free survival. However, it is important to note that Cox linearity assumption was not held for FLT4 and VEGFA (A.3). Therefore, non-significant results for these two genes might be Figure 3.9: Association of continuous VEGF gene expression with event-free survival (EFS)  261 RNA-Seq samples from the COG TARGET study (dbGaP phs000218/000464). Three patients were excluded from the analysis due to being inevaluable. Cox Proportional Hazards method was utilized for each gene separately, using rlog values from DESeq2. Likelihood ratio test p-value is shown. 95% CI is indicated in brackets and by whiskers. 38  false negative. To account for this issue in the future, one option would be to use spline regression. An alternative method that does not make a linearity assumption regarding the data is maximally selected rank statistics [93]. It should be emphasized, however, that one of the main limitations of this method is that it transforms continuous gene expression data into a categorical variable. This tool selects a cutpoint with the most significant separation between two groups based on standardized log-rank statistic. I used it on the three genes of interest to pick an appropriate value for each (A.4). After dichotomization, event-free survival curves for FLT1 indicated that patients with high levels of this gene had worse prognosis (Figure 3.10a). This corroborates Cox regression analysis. Moreover, high expression of FLT4 was also associated with shorter event-free survival (Figure 3.10b). On the other hand, there was no value of VEGFA that led to a significant difference in survival in patients (Figure 3.10c). Interestingly, neither of the three genes was correlated with higher levels of minimal residual disease (MRD) on day 29 (Figure 3.11), which is prevalently used as a prognostic factor [94]. Hence, I concluded that VEGF receptors 1 and 3 can be used as potential biomarkers in T-ALL in addition to MRD quantification. 39   Figure 3.10: Association of dichotomized VEGF gene expression with event-free survival (EFS)  261 RNA-Seq samples from the COG TARGET study (dbGaP phs000218/000464). Three patients were excluded from the analysis due to being inevaluable. Cutpoint for each gene was selected using maximally selected rank statistics. Log-rank test p-value is shown. (a) FLT1 (b) FLT4 (c) VEGFA 40  Figure 3.11: Association of dichotomized VEGF gene expression with minimal residual disease levels (MRD) on day 29 261 RNA-Seq samples from the COG TARGET study (dbGaP phs000218/000464). Three patients were excluded from the analysis due to being inevaluable. Cutpoint for each gene was selected using maximally selected rank statistics. Welch’s t-test p-value is shown. (a) FLT1 (b) FLT4 (c) VEGFA 41  3.3.2 Protein expression of VEGF pathway components in a synthetic NLTB model As VEGF components are correlated with worse event-free survival in clinical samples (section 3.3.1), I hypothesized that they play a functional role in leukemia based on RNA-Seq differential expression analysis (section 3.1.2). As a first step, I needed to determine whether protein levels corroborate transcriptome results in vitro. Therefore, I checked surface expression of both differentially expressed receptors in preleukemic and normal cells. Flow cytometry assay showed that NLTB GFP+Cherry+ cells do not have higher levels of surface VEGFR1 compared to the GFP-Cherry- control (Figure 3.12a). However, preleukemic cells overexpress VEGFR1 intracellularly instead (Figure 3.12b). Two possible explanations for this are that there is a rapid turnover of the receptor on the surface due to the presence of the ligand in coculture, or that there is intracrine signaling happening [95] and the protein does not get to the surface of the cell. Furthermore, preleukemic cells overexpress VEGFR3 on the surface compared to the normal control (Figure 3.12a). Nonetheless, it is important to note that surface levels of this protein are variable in coculture, as there was little difference in its expression in GFP+Cherry+ cells and GFP-Cherry- cells at some timepoints (A.5). Intracellular flow cytometry, however, still indicated that preleukemic cells have higher VEGFR3 levels than normal - even if the protein is not present on the surface (A.6). Possible explanations of this phenomenon are that VEGFR3 surface amount changes based on ligand amount or CB cell density. 42   To test whether VEGF ligands are secreted in coculture, I performed multiplex assays that detect human and mouse VEGFA and PlGF (VEGFR1 ligands), as well as VEGFC and VEGFD (VEGFR3 ligands). The results showed that both of VEGFR1 ligands are secreted by Figure 3.12: Protein levels of VEGF receptors in CB cells in vitro  (a) Surface day 21 VEGFR1 and day 14, 19 VEGFR3 or (b) intracellular day 21 VEGFR1 expression in GFP+Cherry+ NLTB (blue) and GFP-Cherry- normal (red) cells as measured by flow cytometry after fixation and permeabilization. CB cells were transduced with NLTB lentiviral constructs on day 0 in the presence of cyclosporin H (trial 101) or on day 0 and day 5 (trial 73). Cells were cultured on OP9-DL1 feeders over the whole duration of the experiments. Trial 101 (but not trial 73) cells were sorted into CD45+GFP+Cherry+ and CD45+GFP-Cherry- populations. Mouse IgG1, kappa was used as isotype control (orange). VEGFR3 plots are representative of several independent experiments. G+C+: GFP+Cherry+; G-C-: GFP-Cherry-, c: condition. 43  feeder OP9-DL1 cells, but not CB (Figure 3.13a, b). On the other hand, none of VEGFR3 ligands were detected in coculture (Figure 3.13c, d). Therefore, there is a possibility that mouse VEGFA/PlGF bind to human VEGFR1; but VEGFR3 level variation cannot be attributed to ligand secretion.  44   Figure 3.13: Secretion of VEGF ligands by CB and feeder cells in vitro  Levels of secreted ligands on day 24 of coculture with OP9-DL1 feeders as measured by a multiplex assay. CB cells were transduced with NLTB lentiviral constructs on day 0 in the presence of cyclosporin H. Cells were cultured on OP9-DL1 feeders over the whole duration of the experiment. Concentration was calculated using standards provided in the kit (red). Dotted line indicates the lowest standard concentration and defines level of detectability. Human and mouse assays were performed on two replicates of CB cells with OP9-DL1 feeder samples, as well as two replicates of OP9-DL1 feeder-only samples. (a) Amount of mouse VEGFR1 ligands, VEGFA and PlGF, in coculture supernatant. (b) Amount of human VEGFR1 ligands, VEGFA and PlGF, in coculture supernatant. (c) Amount of mouse VEGFR3 ligands, VEGFC and VEGFD, in coculture supernatant. (d) Amount of human VEGFR3 ligands, VEGFC and VEGFD, in coculture supernatant. Std: standard. 45  3.3.3 Selection and validation of shRNAs against VEGFR3 For the purpose of this thesis, I chose to study VEGFR3 signaling. As a method to research its role in T-ALL, we tested eight available shRNAs against this gene in HUVEC cells. The top three constructs with the best knockdown efficiency were then selected – shVEGFR3_637, shVEGFR3_640 and shVEGFR3_656 (Figure 3.14). They are further referred to as sh37, sh40 and sh56. Next, I investigated whether these shRNAs work in the context of NLTB CB model. Therefore, I transduced preleukemic cells with the above vectors, as well as shScr control. As expected, surface flow cytometry assay demonstrated that cells with shRNA against VEGFR3 have lower levels of the receptor compared to the non-transduced cells from the same culture (Figure 3.15). On the other hand, VEGFR3 expression in the control is the same between transduced and non-transduced cells. These data confirm that sh37, sh40 and sh56 can be used to successfully reduce surface VEGFR3 levels in NLTB cells.  46   Figure 3.14: Test of VEGFR3 knockdown in HUVEC cells Surface VEGFR3 receptor expression on day 4 after shRNA transfection in HUVEC cells as measured by flow cytometry. Values were normalized to shScramble control and background was subtracted. Controls shLuciferase-BFP and shScramble-BFP are shown in red, eight shRNA-BFP constructs targeting VEGFR3 are shown in blue. 47  Figure 3.15: Test of select shRNAs against VEGFR3 in preleukemic NLTB cells Surface VEGFR3 receptor expression in GFP+Cherry+ shRNA-transduced cells (blue) and GFP+Cherry+ shRNA non-transduced control cells (red) on day 34 of coculture as measured by flow cytometry. CB cells were transduced with NLTB lentiviral constructs on day 0 in the presence of cyclosporin H. Cells were cultured on OP9-DL1 feeders over the whole duration of the experiment. shRNA-NGFR spinfection was performed on day 23 of coculture. Sh37, sh40 and sh56 target VEGFR3. Mouse IgG1, kappa was used as isotype control (orange). G+C+: GFP+Cherry. 48  3.3.4 Impact of VEGFR3 on leukemia initiation and maintenance in vitro 3.3.4.1 Assessing cell growth effects of VEGFR3 in established T-ALL To determine whether VEGFR3 affects cell growth in established T-ALL, I knocked down its expression through shRNAs in four primary CB NLTB leukemias. The samples were selected based on FLT4 transcript levels – m156, m101, m107 and m160 had high amount of FLT4 transcript compared to other leukemias (A.7). However, flow cytometry indicated that after cells were thawed, VEGFR3 was not present on the surface of the cells (Figure 3.16a). Interestingly, one of the four leukemias (m107) gained surface expression of the protein by day 2 post-thaw (Figure 3.16b). One of the explanations for this phenomenon could be the internalization of the receptor due to a likely presence of a mouse VEGFR3 ligand in vivo – in contrast to its absence during in vitro culture. Nevertheless, it remains unclear why the protein was not detected in other samples in this case – it is possible that because cells have undergone freeze-thaw cycle, regulation of receptor levels was affected. 49    Cell number tracking experiment in the four leukemias showed that there was no consistent effect of VEGFR3 expression on leukemic cell growth in vitro. While m156 and m107 were both negatively affected by the knockdown of the receptor, m101 and m160 did not display a reliable difference between shScr control and all three of shRNAs against VEGFR3 (Figure 3.17). This suggests that while CB leukemias are highly similar to each other as shown in Figure 3.16: Protein levels of VEGFR3 in primary CB leukemias in vitro Surface VEGFR3 receptor expression in GFP+Cherry+ NLTB primary leukemia cells (blue) on (a) day 0 or (b) day 2 of coculture as measured by flow cytometry. Cells were thawed on day 0 and cultured on OP9-DL1 feeders over the whole duration of the experiment. Mouse IgG1, kappa was used as isotype control (orange). G+C+: GFP+Cherry. 50  previous sections, there could still be important differences in biochemical mechanisms between transplanted replicates.  Figure 3.17: Cell growth of established NLTB leukemias upon VEGFR3 knockdown Bulk cell growth tracking of primary NLTB leukemia cells as measured by flow cytometry. Cells were cultured on OP9-DL1 feeders over the whole duration of the experiment. Cells were spinfected with shRNA-BFP vectors on day 2 of coculture post-thawing. shScr control (red) and shVEGFR3 (blue) are depicted. G+C+: GFP+Cherry+. 51  3.3.4.2 Assessing ligand-dependent effect of VEGFR3 on preleukemic cell growth To elucidate the role of canonical VEGFR3 signaling in leukemia initiation, I added human VEGFC to NLTB cultures (as it was not produced by human or mouse cells). As expected, surface levels of the receptor decreased in the presence of its ligand, in contrast to no addition and 50 ng/mL VEGFA conditions (Figure 3.18, A.8). This could indicate that the receptor is internalized due to the presence of VEGFC, possibly leading to downstream pathway activation; however, VEGFR3 phosphorylation status is required to confirm this hypothesis.  To ensure that any possible effect of ligand addition happens due to its interaction with my receptor of interest, I transduced NLTB cells with shRNAs (selected in section 3.3.3) five days before adding VEGFC. I further tracked the total number of cells over the course of 19 Figure 3.18: Addition of VEGF ligands to CB cells Surface VEGFR3 receptor expression in GFP+Cherry+ NLTB cells on day 27 of coculture with OP9-DL1 feeders as measured by flow cytometry. CB cells were transduced with NLTB lentiviral constructs on day 0 and day 5. Cells were cultured on OP9-DL1 feeders over the whole duration of the experiment. Exogenous human ligand was added on day 19 of coculture. Left to right: no exogenous ligand added, cells cultured according to standard protocol; exogenous 50 ng/mL hVEGFA addition; exogenous 10 ng/mL hVEGFC addition. G+C+: GFP+Cherry+; G-C-: GFP-Cherry-. 52  days. I observed that there was no difference between shScr and shVEGFR3 conditions (Figure 3.19a). Additionally, EdU proliferation assay support these results, as none of shVEGFR3 samples had a significant change in cell proliferation compared to the control on day 34 (Figure 3.19b). Overall, these data suggest that VEGFR3 activity in the presence of its ligand VEGFC does not affect cell growth or proliferation of preleukemic NLTB cells in vitro. 53   Figure 3.19: Cell growth of preleukemic NLTB cells upon VEGFR3 knockdown with exogenous ligand present Tracking of CB GFP+Cherry+ population. CB cells were transduced with NLTB lentiviral constructs on day 0 in the presence of cyclosporin H. Cells were cultured on OP9-DL1 feeders over the whole duration of the experiment. Cells were spinfected with shRNA-NGFR vectors on day 23 of coculture. VEGFC was exogenously maintained over the course of experiment at 10ng/mL. shScr control (red) and shVEGFR3 (blue) are depicted. (a) Bulk cell growth of preleukemic cells as measured by flow cytometry cell number tracking for 19 days. (b) Proliferation activity of preleukemic cells as measured by incorporation of EdU on day 34. Two-tailed two-sample unequal variance t-test p-values are shown.  54  Furthermore, I performed a WIA comparing two conditions – NLTB and NLTB+10 ng/mL VEGFC. With a well initiation threshold of 1000 (Figure 3.20a), there was no significant difference between samples. NLTB cells with exogenous addition of VEGFC had a well initiation frequency of ~1 in 79 cells (95% CI: 1 in 59-106 cells), while NLTB-only condition - ~1 in 84 cells (95% CI: 1 in 63-113 cells) (Figure 3.20b). This experiment indicates that presence of VEGFC in coculture does not affect clonogenic potential of preleukemic cells.  55   56  3.3.4.3 Assessing ligand-independent effect of VEGFR3 on preleukemic cell growth To investigate whether VEGFR3 expression is functionally relevant in preleukemic cells in the absence of its canonical ligands, I transduced NLTB cells with shRNAs against the receptor in accordance with section 3.3.4. There was no difference between cell numbers of control and shVEGFR3 conditions over the course of coculture (Figure 3.21a). Similarly, shVEGFR3 cells had the same percentage of proliferative fraction as shScr (Figure 3.21b). Hence, non-canonical ligand-independent VEGFR3 activity does not play a functional role in preleukemic cell growth and proliferation in vitro.  Figure 3.20: Effect of VEGFR3 ligand on clonogenic activity of preleukemic NLTB cells CB cells were transduced with NLTB lentiviral constructs on day 0 in the presence of cyclosporin H. Cells were cultured on OP9-DL1 feeders over the whole duration of the experiment. For the condition with exogenous ligand, VEGFC was maintained over the course of coculture at 10 ng/mL post-sort. On day 29, CD45+GFP+Cherry+ cells were sorted into a 96-well plate. After 19 days of culture, cells were collected with trypsin and analyzed by flow cytometry. (a) Total number of live CD45+GFP+Cherry+ cells per each well (n=23-24 for each condition). 1000 cells threshold for well initiation frequency calculations is indicated as a dotted line. (b) Well initiating frequency calculations with a threshold of 1000 cells. 95% CI are shown as dotted lines. Chi-square test p-value is shown. G+C+: GFP+Cherry+. 57   Figure 3.21: Cell growth of preleukemic NLTB cells upon VEGFR3 knockdown Tracking of CB GFP+Cherry+ population. CB cells were transduced with NLTB lentiviral constructs on day 0 in the presence of cyclosporin H. Cells were cultured on OP9-DL1 feeders over the whole duration of the experiment. Cells were spinfected with shRNA-NGFR vectors on day 23 of coculture. shScr control (red) and shVEGFR3 (blue) are depicted. (a) Bulk cell growth of preleukemic cells as measured by flow cytometry cell number tracking for 19 days. (b) Proliferation activity of preleukemic cells as measured by incorporation of EdU on day 34. Two-tailed two-sample unequal variance t-test p-values are shown.   58  Chapter 4: Discussion 4.1 Summary and significance of research Due to a lack of efficient and reproducible human T-ALL models, a synthetic NLTB model has been developed in our group to study this disease. Through functional assays, we previously established that human NLTB cells progress to T-ALL in immunodeficient mice. In the first part of my project, I aimed to transcriptionally characterize primary NLTB leukemia cells and determine their similarity to bona fide disease. I demonstrated that NLTB samples gene expression is similar to that of patient-derived xenografts in gene expression. In addition, these leukemias show clonal rearrangements as expected in natural T-ALL [86,87]. Importantly, while this analysis was performed on a transcript level, which could differ from that of a protein, they were also corroborated by a clinical BIOMED-2 assay [26]. Overall, my data support prior experimental results, suggesting that NLTB cells can be effectively used to research critical leukemogenic events. Hence, I concluded that our synthetic model is a highly effective tool to study established disease and uniquely provides a way to dissect leukemia initiation processes in a human cell context. In particular, it can be successfully applied to investigate biochemical mechanisms relevant in T-ALL in genetically defined conditions. Characterization of in vitro cultured NLTB cells prior to in vivo mouse transplantation allows to determine which cellular mechanisms are important in T-ALL establishment. Thus, I next performed differential expression analysis between preleukemic and normal cells to investigate pathways activated during initial stages of the disease. Anterior HOXB and VEGF genes were significantly upregulated in NLTB cells, suggesting that these pathways might be functionally contributing to T-ALL progression. 59  It has been shown that dysregulation of HOX genes is involved in carcinogenesis, particularly in acute myeloid and lymphoblastic leukemias [96,97]. Hence, I asked a question of whether anterior HOXB genes are associated with clinical features in the context of T-ALL. Overexpression of at least one of these genes was associated with worse prognosis, with HOXB4 being the main contributor. However, it should be pointed out that the patients were divided into two separate groups; thus, dichotomization of the dataset is a caveat of this analysis.  My results also demonstrated that patients with high HOXB genes correlate with late T-cell subgroups and thymic signatures. Interestingly, this suggests that anterior HOXB genes might affect cellular events in late T-cell progenitor cells instead of hematopoietic stem cells (HSCs), as was expected based on previous knowledge [45,91,98]. These results are supported by data from another group, which depicted functional effects of HOXB4 transduction in progenitors but not stem cells [99]. According to several studies, HOXB genes (in particular, HOXB3 and HOXB4) can play a role in cell growth [34,96,100]. Therefore, we investigated whether they also contribute to this process in T-ALL. Our functional experiments demonstrated that HOXB3 and HOXB5 negatively affected in vitro cell growth of a primary leukemia. As HOXB5 wasn’t highly expressed in the patient dataset, we focused our efforts on HOXB3. Cell growth of established leukemia samples (tested in two different platforms) and clonogenic activity of preleukemic cells significantly decreased with lower levels of HOXB3. These results from distinct T-ALL models indicate that elevated HOXB3 contributes to progression of the malignancy, and that it was not an artifact of in vitro preleukemic NLTB culture. Overall, this work gives insight into a previously unknown role of HOXB genes in T-ALL cell growth during both disease initiation and maintenance. Moreover, patient and synthetic model analyses highlight a possibility of 60  upstream HOXB cluster regulation through overexpression of TAL1 oncogene in T-ALL. Therefore, this knowledge can be potentially used to develop targeted therapies and prevent leukemia relapse in a susceptible subset of patients. VEGF receptors (FLT1 and FLT4) were also associated with lower event-free survival in T-ALL patients. In particular, FLT1 was significant as shown by two separate survival analysis methods. This corroborates previous data, illustrating the relationship of VEGF genes with negative clinical features in multiple cancers, including leukemia [50,51,101,102]. However, to my knowledge, correlation of these particular receptors with worse prognosis in T-ALL has not been shown before. Therefore, my work illustrates two novel biomarkers in the context of this disease, which could be used to predict outcome and design more effective treatments. To investigate these receptors functionally during leukemogenic process, I performed further experiments in the NLTB model in vitro. I found that VEGFR1 is overexpressed intracellularly in preleukemic cells compared to normal population. Moreover, both of its ligands were secreted by feeder cells in culture, suggesting that their interaction with VEGFR1 could lead to receptor internalization. As concluded by a previous study [56], both of these mouse ligands are likely able to activate the human receptor. However, it is important to note that, to my knowledge, there have been no studies directly showing cross-species binding. These protein expression results suggest that CB cells might depend on feeder cells to activate VEGFR1 signaling. Furthermore, secretion of VEGFA by such cell types as osteoblasts [103] in mice could contribute to NLTB cell localization in vivo similarly to what was already established in literature [56]. However, it is also possible that CB cells themselves produce VEGFA protein (as its transcript is detected by RNA-Seq) but it is not secreted – there have been reports on VEGFA/VEGFR1 intracrine signaling [95]. 61  Furthermore, I found that VEGFR3 was higher on the surface of preleukemic cells than normal, and that none of its ligands were present in coculture. Unlike VEGFR1 results, these data suggest that CB system provides a controlled environment to study VEGFR3 downstream effects, as we can regulate the amount of ligand available for uptake. In particular, addition of exogenous VEGFC will allow us to look at canonical signaling, which might be relevant in vivo after CB cell injection. For instance, osteoclasts secrete VEGFC for bone resorption [104], and it has been shown that mouse ligand is able to bind to human VEGFR3 [105]. Therefore, human NLTB cells could use secreted mouse VEGFC to promote leukemogenesis in vivo. Moreover, we are also able to investigate the role of ligand-independent VEGFR3 signaling in the CB system. It has been previously shown to play a role in cancer cell growth and motility [62]. Thus, in this work I prioritized studying VEGFR3 activity and not VEGFR1.  According to several research groups, both canonical and ligand-independent VEGFR3 signaling are involved in promoting cell growth [62,68,106]. In particular, AML cells displayed an increase in proliferation due to VEGFC/VEGFR3 signaling [68]. Therefore, I hypothesized that VEGFR3 similarly mediates cell expansion in preleukemic and leukemic cells in the context of T-ALL.  Firstly, I conducted a bulk cell growth assay using four different primary CB leukemias. While two of the samples showed that VEGFR3 knockdown has a negative effect on leukemic cell growth, the other two replicates did not recapitulate that data. Therefore, I concluded that VEGFR3 expression does not have a consistent effect on leukemia maintenance. This highlights intrinsic differences between primary leukemias which could affect downstream protein functions. VEGFR3 could play various roles in T-ALL malignancy (including but not limited to promoting cell growth), depending on currently undetermined variables in cancer cells. It is also 62  important to note that due to culturing of leukemias in vitro, major factors (such as exogenous VEGFR3 ligand from mouse cells) are likely missing in the system – therefore raising the possibility that cell response to knockdown in culture could greatly differ from that in mice. In addition, I performed several growth and proliferation assays in preleukemic cells – both in the presence and absence of VEGFC. None of the experimental results on the effect of VEGFR3 on leukemia initiation showed a significant effect, rejecting my hypothesis. Nevertheless, this does not mean that the receptor is not functional in NLTB cells. It must be noted that variation in its surface levels, in addition to unaccounted activity of neuropilins [107–109], could have affected overall results. Moreover, bulk cell number tracking and EdU incorporation assays present a major limitation – both of them provide only a snapshot of the overall situation at a specific timepoint. Additionally, it is possible that surface receptor loss only has an effect on a subset of preleukemic cells, or that other pathways are able to rescue the phenotype. For instance, as shRNAs had NGFR reporter, it could have possibly interfered with cell growth assays by activating similar downstream genes – analogous to findings on overlap between other growth factor pathways [110]. However, my main interpretation of the data is that VEGFR3 activity does not influence preleukemic NLTB cell growth – in contrast to what was expected from literature.  Therefore, overall results from leukemia initiation and maintenance experiments suggest that different outcomes of VEGFR3 signaling are observed depending on cell and developmental context. This notion is especially important while designing therapies for distinct types of cancer. 4.2 Future directions While assays on HOXB3 in preleukemic and leukemic NLTB contexts demonstrated that this gene contributes to the process of cell growth in T-ALL, more information is needed on the 63  role of genetic features on HOXB gene expression. For instance, upstream mechanisms of regulation (possibly, through TAL1 oncogene) can be explored to determine why some patients overexpress HOXB gene cluster, while others don’t. In the case of VEGFR3, it is critical to investigate its alternative possible effects on malignant cells besides cell growth. It has to be noted that VEGFR3 in cancer is upstream of multiple other cellular processes – for instance, motility [58,62,66] and angiogenesis [69,111].  Thus, we can ask whether NLTB cells with higher VEGFR3 levels are able to migrate towards its ligand in contrast to their normal counterparts. This process could be particularly relevant in vivo, where preleukemic cells might be migrating towards higher ligand levels and establishing a leukemic niche similarly to CXCR4/CXCL12 mechanism [112–114]. Moreover, we could further elucidate if exogenous VEGFC could support angiogenesis through VEGFR3 activity in T-ALL as it does in AML [69]. In addition, investigation of the role of VEGFR1 in leukemia initiation and maintenance is also of interest. In particular, presence of its secreted mouse ligands in coculture could be indicative of its downstream activity in preleukemic cells. Therefore, performing cell number tracking and apoptosis assays on NLTB model in vitro can help determine whether there is a major effect of this receptor on preleukemic (and leukemic) cell survival. To account for the uncontrolled ligand presence in coculture, feeders with VEGFA/PlGF deletion, ligand-binding antibodies such as bevacizumab or shRNAs against VEGFR1 could be used. Exploration of downstream VEGFR1 functions will also help answer the question of whether VEGFR1 and VEGFR3 play similar roles in NLTB cells by affecting the same cellular processes. 64  References 1.  Raetz, E. A. & Teachey, D. T. T-cell acute lymphoblastic leukemia. Hematol. Am. Soc. Hematol. Educ. Program 2016, 580–588 (2016). 2.  Ferrando, A. A. et al. Gene expression signatures define novel oncogenic pathways in T cell acute lymphoblastic leukemia. Cancer Cell 1, 75–87 (2002). 3.  Onciu, M. Acute Lymphoblastic Leukemia. Hematol. Oncol. Clin. North Am. 23, 655–674 (2009). 4.  Inaba, H., Greaves, M. & Mullighan, C. G. Acute lymphoblastic leukaemia. Lancet 381, (2013). 5.  Durinck, K. et al. Novel biological insights in T-cell acute lymphoblastic leukemia. Exp. Hematol. 43, 625–639 (2015). 6.  Weng, A. P. et al. Activating mutations of NOTCH1 in human T cell acute lymphoblastic leukemia. Science 306, 269–271 (2004). 7.  Pear, W. S. et al. Exclusive development of T cell neoplasms in mice transplanted with bone marrow expressing activated Notch alleles. J. Exp. Med. 183, 2283–2291 (1996). 8.  O’Neil, J. et al. FBW7 mutations in leukemic cells mediate NOTCH pathway activation and resistance to γ-secretase inhibitors. J. Exp. Med. 204, 1813–1824 (2007). 9.  Hebert, J., Cayuela, J. M., Berkeley, J. & Sigaux, F. Candidate tumor-suppressor genes MTS1 (p16INK4A) and MTS2 (p15INK4B) display frequent homozygous deletions in primary cells from T- but not from B-cell lineage acute lymphoblastic leukemias. Blood 84, 4038–4044 (1994). 10. Liu, Y. et al. The genomic landscape of pediatric and young adult T-lineage acute lymphoblastic leukemia. Nat. Genet. 49, 1211–1218 (2017). 65  11. Stanulović, V. S., Cauchy, P., Assi, S. A. & Hoogenkamp, M. LMO2 is required for TAL1 DNA binding activity and initiation of definitive haematopoiesis at the haemangioblast stage. Nucleic Acids Res. 45, 9874–9888 (2017). 12. Kennedy, J. A., Barabe, F. & Dick, J. E. LMO2 and TAL1 Cooperate in a Model of Human T−Cell Leukemogenesis. Blood 108, 773–773 (2006). 13. Homminga, I. et al. Integrated Transcript and Genome Analyses Reveal NKX2-1 and MEF2C as Potential Oncogenes in T Cell Acute Lymphoblastic Leukemia. Cancer Cell 19, 484–497 (2011). 14. You, Q. et al. Animal models of T-cell acute lymphoblastic leukemia: mimicking the human disease. J. Bio-X Res. 1, 32–40 (2018). 15. Saland, E. et al. A robust and rapid xenograft model to assess efficacy of chemotherapeutic agents for human acute myeloid leukemia. Blood Cancer J. 5, e297 (2015). 16. Squiban, B., Ahmed, S. T. & Frazer, J. K. Creation of a Human T-ALL Cell Line Online Database Human T-ALL cell line database. Leuk. Lymphoma 58, 2728–2730 (2017). 17. Burger, R., Hansen-Hagge, T. E., Drexler, H. G. & Gramatzki, M. Heterogeneity of T-acute lymphoblastic leukemia (T-ALL) cell lines: Suggestion for classification by immunophenotype and T-cell receptor studies. Leuk. Res. 23, 19–27 (1999). 18. Kohnken, R., Porcu, P. & Mishra, A. Overview of the Use of Murine Models in Leukemia and Lymphoma Research. Front. Oncol. 7, (2017). 19. Jacoby, E., Chien, C. D. & Fry, T. J. Murine Models of Acute Leukemia: Important Tools in Current Pediatric Leukemia Research. Front. Oncol. 4, (2014). 20. Parekh, C. & Crooks, G. M. Critical Differences in Hematopoiesis and Lymphoid Development Between Humans and Mice. J. Clin. Immunol. 33, 711–715 (2013). 66  21. Odom, D. T. et al. Tissue-Specific Transcriptional Regulation Has Diverged Significantly between Human and Mouse. Nat. Genet. 39, 730–732 (2007). 22. Rangarajan, A. & Weinberg, R. A. Comparative biology of mouse versus human cells: modelling human cancer in mice. Nat. Rev. Cancer 3, 952–959 (2003). 23. Lai, Y. et al. Current status and perspectives of patient-derived xenograft models in cancer research. J. Hematol. Oncol.J Hematol Oncol 10, (2017). 24. Girardi, T., Vicente, C., Cools, J. & De Keersmaecker, K. The genetics and molecular biology of T-ALL. Blood 129, 1113–1123 (2017). 25. Wang, K. et al. Patient-derived xenotransplants can recapitulate the genetic driver landscape of acute leukemias. Leukemia 31, 151–158 (2017). 26. Kusakabe, M. et al. Synthetic modeling reveals HOXB genes are critical for the initiation and maintenance of human leukemia. Nat. Commun. 10, 2913 (2019). 27. Jacobs, J. J. L., Kieboom, K., Marino, S., DePinho, R. A. & van Lohuizen, M. The oncogene and Polycomb-group gene bmi-1 regulates cell proliferation and senescence through the ink4a locus. Nature 397, 164–168 (1999). 28. Jacobs, J. J. L. et al. Bmi-1 collaborates with c-Myc in tumorigenesis by inhibiting c-Myc-induced apoptosis via INK4a/ARF. Genes Dev. 13, 2678–2690 (1999). 29. Miyazaki, M. et al. Thymocyte Proliferation Induced by Pre-T Cell Receptor Signaling Is Maintained through Polycomb Gene Product Bmi-1-Mediated Cdkn2a Repression. Immunity 28, 231–245 (2008). 30. Holmes, R. & Zúñiga-Pflücker, J. C. The OP9-DL1 System: Generation of T-Lymphocytes from Embryonic or Hematopoietic Stem Cells In Vitro. Cold Spring Harb. Protoc. 2009, pdb.prot5156 (2009). 67  31. Pui, J. C. et al. Notch1 Expression in Early Lymphopoiesis Influences B versus T Lineage Determination. Immunity 11, 299–308 (1999). 32. Radtke, F. et al. Deficient T Cell Fate Specification in Mice with an Induced Inactivation of Notch1. Immunity 10, 547–558 (1999). 33. Quinonez, S. C. & Innis, J. W. Human HOX gene disorders. Mol. Genet. Metab. 111, 4–15 (2014). 34. Björnsson, J. M. et al. Reduced Proliferative Capacity of Hematopoietic Stem Cells Deficient in Hoxb3 and Hoxb4. Mol. Cell. Biol. 23, 3872–3883 (2003). 35. Eklund, E. The role of HOX proteins in leukemogenesis; insights into key regulatory events in hematopoiesis. Crit. Rev. Oncog. 16, 65–76 (2011). 36. Jung, N., Dai, B., Gentles, A. J., Majeti, R. & Feinberg, A. P. An LSC epigenetic signature is largely mutation independent and implicates the HOXA cluster in AML pathogenesis. Nat. Commun. 6, 8489 (2015). 37. Soulier, J. et al. HOXA genes are included in genetic and biologic networks defining human acute T-cell leukemia (T-ALL). Blood 106, 274–286 (2005). 38. Dickson, G. J. et al. HOXA/PBX3 knockdown impairs growth and sensitizes cytogenetically normal acute myeloid leukemia cells to chemotherapy. Haematologica 98, 1216–1225 (2013). 39. Bock, C. E. de et al. HOXA9 Cooperates with Activated JAK/STAT Signaling to Drive Leukemia Development. Cancer Discov. 8, 616–631 (2018). 40. Umeda, S. et al. Prognostic significance of HOXB4 in de novo acute myeloid leukemia. Hematology 17, 125–131 (2012). 68  41. Lindblad, O. et al. The role of HOXB2 and HOXB3 in acute myeloid leukemia. Biochem. Biophys. Res. Commun. 467, 742–747 (2015). 42. Wang, H. et al. HOXB4 knockdown reverses multidrug resistance of human myelogenous leukemia K562/ADM cells by downregulating P-gp, MRP1 and BCRP expression via PI3K/Akt signaling pathway. Int. J. Oncol. 49, 2529–2537 (2016). 43. Qian, P. et al. Retinoid-Sensitive Epigenetic Regulation of the Hoxb Cluster Maintains Normal Hematopoiesis and Inhibits Leukemogenesis. Cell Stem Cell 22, 740-754.e7 (2018). 44. Liu, H.-C. et al. Expression of HOXB genes is significantly different in acute myeloid leukemia with a partial tandem duplication of MLL vs. a MLL translocation: a cross-laboratory study. Cancer Genet. 204, 252–259 (2011). 45. Antonchuk, J., Sauvageau, G. & Humphries, R. K. HOXB4 overexpression mediates very rapid stem cell regeneration and competitive hematopoietic repopulation. Exp. Hematol. 29, 1125–1134 (2001). 46. Carè, A. et al. Coordinate expression and proliferative role of HOXB genes in activated adult T lymphocytes. Mol. Cell. Biol. 14, 4872–4877 (1994). 47. Karaman, S., Leppänen, V.-M. & Alitalo, K. Vascular endothelial growth factor signaling in development and disease. Dev. Camb. Engl. 145, (2018). 48. Revskoy, S., Mizgirev, I. & Corey, S. J. Bioimaging of Vascular Microenvironment Reveals Presence of Fli1a, Not VEGFR2, in Acute Lymphoblastic Leukemia Model in Syngeneic Zebrafish. Blood 118, 2551–2551 (2011). 49. Poyer, F. et al. Secretion of MMP-2 and MMP-9 induced by VEGF autocrine loop correlates with clinical features in childhood acute lymphoblastic leukemia. Leuk. Res. 33, 407–417 (2009). 69  50. Wellmann, S. et al. Activation of the HIF pathway in childhood ALL, prognostic implications of VEGF. Leukemia 18, 926–933 (2004). 51. Tokunaga, T. et al. Vascular endothelial growth factor (VEGF) mRNA isoform expression pattern is correlated with liver metastasis and poor prognosis in colon cancer. Br. J. Cancer 77, 998–1002 (1998). 52. Pillozzi, S. et al. VEGFR-1 (FLT-1), β1 integrin, and hERG K+ channel for a macromolecular signaling complex in acute myeloid leukemia: role in cell migration and clinical outcome. Blood 110, 1238–1250 (2007). 53. Vincent, L. et al. Fetal stromal-dependent paracrine and intracrine vascular endothelial growth factor-a/vascular endothelial growth factor receptor-1 signaling promotes proliferation and motility of human primary myeloma cells. Cancer Res. 65, 3185–3192 (2005). 54. Lee, H. K., Chauhan, S. K., Kay, E. & Dana, R. Flt-1 regulates vascular endothelial cell migration via a protein tyrosine kinase-7-dependent pathway. Blood 117, 5762–5771 (2011). 55. Münch, V. et al. Central nervous system involvement in acute lymphoblastic leukemia is mediated by vascular endothelial growth factor. Blood 130, 643–654 (2017). 56. Fragoso, R. et al. VEGFR-1 (FLT-1) activation modulates acute lymphoblastic leukemia localization and survival within the bone marrow, determining the onset of extramedullary disease. Blood 107, 1608–1616 (2006). 57. Decio, A. et al. Vascular endothelial growth factor c promotes ovarian carcinoma progression through paracrine and autocrine mechanisms. Am. J. Pathol. 184, 1050–1061 (2014). 70  58. Yeh, Y.-W. et al. Targeting the VEGF-C/VEGFR3 axis suppresses Slug-mediated cancer metastasis and stemness via inhibition of KRAS/YAP1 signaling. Oncotarget 8, 5603–5618 (2017). 59. Chen, H., Suo, K., Cheng, Y., Zheng, B. & Xu, L. Vascular endothelial growth factor C enhances cervical cancer migration and invasion via activation of focal adhesion kinase. Gynecol. Endocrinol. Off. J. Int. Soc. Gynecol. Endocrinol. 29, 20–24 (2013). 60. Beierle, E. A. et al. Inhibition of the focal adhesion kinase and vascular endothelial growth factor receptor-3 interaction leads to decreased survival in human neuroblastoma cell lines. Mol. Carcinog. 53, 230–242 (2014). 61. Coso, S., Zeng, Y., Opeskin, K. & Williams, E. D. Vascular endothelial growth factor receptor-3 directly interacts with phosphatidylinositol 3-kinase to regulate lymphangiogenesis. PloS One 7, e39558 (2012). 62. Kurenova, E. V. et al. Vascular endothelial growth factor receptor-3 promotes breast cancer cell proliferation, motility and survival in vitro and tumor formation in vivo. Cell Cycle Georget. Tex 8, 2266–2280 (2009). 63. Galvagni Federico et al. Endothelial Cell Adhesion to the Extracellular Matrix Induces c-Src–Dependent VEGFR-3 Phosphorylation Without the Activation of the Receptor Intrinsic Kinase Activity. Circ. Res. 106, 1839–1848 (2010). 64. Wang, J. F., Zhang, X. F. & Groopman, J. E. Stimulation of beta 1 integrin induces tyrosine phosphorylation of vascular endothelial growth factor receptor-3 and modulates cell migration. J. Biol. Chem. 276, 41950–41957 (2001). 71  65. Ko, Y. H. et al. Clinical Significance of Vascular Endothelial Growth Factors (VEGF)-C and -D in Resected Non-Small Cell Lung Cancer. Cancer Res. Treat. Off. J. Korean Cancer Assoc. 40, 133–140 (2008). 66. Su, J.-L. et al. The VEGF-C/Flt-4 axis promotes invasion and metastasis of cancer cells. Cancer Cell 9, 209–223 (2006). 67. Liersch, R. et al. Expression of VEGF-C and its receptor VEGFR-3 in the bone marrow of patients with acute myeloid leukaemia. Leuk. Res. 32, 954–961 (2008). 68. Dias, S., Choy, M., Alitalo, K. & Rafii, S. Vascular endothelial growth factor (VEGF)–C signaling through FLT-4 (VEGFR-3) mediates leukemic cell proliferation, survival, and resistance to chemotherapy. Blood 99, 2179–2184 (2002). 69. Chien, M.-H. et al. Vascular endothelial growth factor-C (VEGF-C) promotes angiogenesis by induction of COX-2 in leukemic cells via the VEGF-R3/JNK/AP-1 pathway. Carcinogenesis 30, 2005–2013 (2009). 70. Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013). 71. Liao, Y., Smyth, G. K. & Shi, W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinforma. Oxf. Engl. 30, 923–930 (2014). 72. R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria, http://www.R-project.org/ (2013). 73. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014). 72  74. Leek, J. T., Johnson, W. E., Parker, H. S., Jaffe, A. E. & Storey, J. D. The sva package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinforma. Oxf. Engl. 28, 882–883 (2012). 75. Fabregat, A. et al. The Reactome Pathway Knowledgebase. Nucleic Acids Res. 46, D649–D655 (2018). 76. Therneau, T. A Package for Survival Analysis in R. R package version 3.1-11, https://CRAN.R-project.org/package=survival (2020). 77. Therneau, T. M. & Grambsch, P. M. Modeling Survival Data: Extending the Cox Model. (Springer New York, 2000). doi:10.1007/978-1-4757-3294-8. 78. Kassambara, A., Kosinski, M. & Biecek, P. survminer: Drawing Survival Curves using ‘ggplot2’. R package version 0.4.7. https://CRAN.R-project.org/package=survminer (2020). 79. Szymczak, A. L. et al. Correction of multi-gene deficiency in vivo using a single ‘self-cleaving’ 2A peptide–based retroviral vector. Nat. Biotechnol. 22, 589–594 (2004). 80. Giambra, V. et al. NOTCH1 promotes T cell leukemia-initiating activity by RUNX-mediated regulation of PKC-θ and reactive oxygen species. Nat. Med. 18, 1693–1698 (2012). 81. Robinson, M. D., McCarthy, D. J. & Smyth, G. K. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139–140 (2010). 82. Wickham, H. ggplot2: Elegant Graphics for Data Analysis. (Springer International Publishing, 2016). doi:10.1007/978-3-319-24277-4, https://ggplot2.tidyverse.org. 83. Hu, Y. & Smyth, G. K. ELDA: Extreme limiting dilution analysis for comparing depleted and enriched populations in stem cell and other assays. J. Immunol. Methods 347, 70–78 (2009). 73  84. Townsend, E. C. et al. The Public Repository of Xenografts Enables Discovery and Randomized Phase II-like Trials in Mice. Cancer Cell 29, 574–586 (2016). 85. Pilozzi, E. et al. Gene rearrangements in T-cell lymphoblastic lymphoma. J. Pathol. 188, 267–270 (1999). 86. Kunz, J. B. et al. Pediatric T-cell lymphoblastic leukemia evolves into relapse by clonal selection, acquisition of mutations and promoter hypomethylation. Haematologica 100, 1442–1450 (2015). 87. Mullighan, C. G. et al. Genomic analysis of the clonal origins of relapsed acute lymphoblastic leukemia. Science 322, 1377–1380 (2008). 88. Bolotin, D. A. et al. Antigen receptor repertoire profiling from RNA-seq data. Nat. Biotechnol. 35, 908–911 (2017). 89. Coustan-Smith, E. et al. Early T-cell precursor leukaemia: a subtype of very high-risk acute lymphoblastic leukaemia. Lancet Oncol. 10, 147–156 (2009). 90. Beslu, N. et al. Molecular interactions involved in HOXB4-induced activation of HSC self-renewal. Blood 104, 2307–2314 (2004). 91. Thorsteinsdottir, U., Sauvageau, G. & Humphries, R. K. Enhanced in vivo regenerative potential of HOXB4-transduced hematopoietic stem cells with regulation of their pool size. Blood 94, 2605–2612 (1999). 92. Raman, P. et al. A comparison of survival analysis methods for cancer gene expression RNA-Sequencing data. Cancer Genet. 235–236, 1–12 (2019). 93. Hothorn, T. & Lausen, B. On the exact distribution of maximally selected rank statistics. Comput. Stat. Data Anal. 43, 121–137 (2003). 74  94. van Dongen, J. J. et al. Prognostic value of minimal residual disease in acute lymphoblastic leukaemia in childhood. Lancet Lond. Engl. 352, 1731–1738 (1998). 95. Gerber, H.-P. et al. VEGF regulates haematopoietic stem cell survival by an internal autocrine loop mechanism. Nature 417, 954–958 (2002). 96. Bi, L. et al. A novel miR-375-HOXB3-CDCA3/DNMT3B regulatory circuitry contributes to leukemogenesis in acute myeloid leukemia. BMC Cancer 18, 182 (2018). 97. Ferrando, A. A. et al. Gene expression signatures in MLL-rearranged T-lineage and B-precursor acute leukemias: dominance of HOX dysregulation. Blood 102, 262–268 (2003). 98. Schiedlmeier, B. et al. High-level ectopic HOXB4 expression confers a profound in vivo competitive growth advantage on human cord blood CD34+ cells, but impairs lymphomyeloid differentiation. Blood 101, 1759–1768 (2003). 99. Lim, H. et al. Committed Hemopoietic Progenitors, Not Stem Cells, are the Principal Responders to Hox Gene Transduction. bioRxiv 174490 (2017) doi:10.1101/174490. 100. Krosl, J. et al. Cellular proliferation and transformation induced by HOXB4 and HOXB3 proteins involves cooperation with PBX1. Oncogene 16, 3403–3412 (1998). 101. Bando, H. et al. Association between intratumoral free and total VEGF, soluble VEGFR-1, VEGFR-2 and prognosis in breast cancer. Br. J. Cancer 92, 553–561 (2005). 102. Martins, S. F. et al. Clinicopathological Correlation and Prognostic Significance of VEGF-A, VEGF-C, VEGFR-2 and VEGFR-3 Expression in Colorectal Cancer. Cancer Genomics - Proteomics 10, 55–67 (2013). 103. Hu, K. & Olsen, B. R. Osteoblast-derived VEGF regulates osteoblast differentiation and bone formation during bone repair. J. Clin. Invest. 126, 509–526. 75  104. Zhang, Q. et al. VEGF-C, a Lymphatic Growth Factor, Is a RANKL Target Gene in Osteoclasts That Enhances Osteoclastic Bone Resorption through an Autocrine Mechanism. J. Biol. Chem. 283, 13491–13499 (2008). 105. Kukk, E. et al. VEGF-C receptor binding and pattern of expression with VEGFR-3 suggests a role in lymphatic vascular development. Development 122, 3829–3837 (1996). 106. Salameh, A., Galvagni, F., Bardelli, M., Bussolino, F. & Oliviero, S. Direct recruitment of CRK and GRB2 to VEGFR-3 induces proliferation, migration, and survival of endothelial cells through the activation of ERK, AKT, and JNK pathways. Blood 106, 3423–3431 (2005). 107. Ou, J.-J. et al. Neuropilin-2 mediates lymphangiogenesis of colorectal carcinoma via a VEGFC/VEGFR3 independent signaling. Cancer Lett. 358, 200–209 (2015). 108. Xu, Y. et al. Neuropilin-2 mediates VEGF-C–induced lymphatic sprouting together with VEGFR3. J. Cell Biol. 188, 115–130 (2010). 109. Mercurio, A. M. VEGF/Neuropilin Signaling in Cancer Stem Cells. Int. J. Mol. Sci. 20, (2019). 110. Fambrough, D., McClure, K., Kazlauskas, A. & Lander, E. S. Diverse Signaling Pathways Activated by Growth Factor Receptors Induce Broadly Overlapping, Rather Than Independent, Sets of Genes. Cell 97, 727–741 (1999). 111. Laakkonen, P. et al. Vascular endothelial growth factor receptor 3 is involved in tumor angiogenesis and growth. Cancer Res. 67, 593–599 (2007). 112. Pitt, L. A. et al. CXCL12-Producing Vascular Endothelial Niches Control Acute T Cell Leukemia Maintenance. Cancer Cell 27, 755–768 (2015). 113. Passaro, D. et al. CXCR4 Is Required for Leukemia-Initiating Cell Activity in T Cell Acute Lymphoblastic Leukemia. Cancer Cell 27, 769–779 (2015). 76  114. Sipkins, D. A. et al. In vivo imaging of specialized bone marrow endothelial microdomains for tumor engraftment. Nature 435, 969–973 (2005).  77  Appendices Appendix A  Supplementary Figures  A.1 NLTB and HOXB gene expression in T-ALL cell lines Data obtained and reanalyzed from EGA database (EGAD00001000849). mRNA expression heatmap of 18 human T-ALL cell lines. Normalized expression was scaled by gene with mean of 0 and standard deviation of 1.  78  A.2 Expression of HOXB and flanking genes in CB cell populations in vitro mRNA expression heatmap of transduced and non-transduced cell populations cultured in vitro. CD45+ cells were sorted based on GFP/Cherry markers expression on day 14 and day 24 of coculture in several independent experiments. Rlog values with mean  SD are shown. *p < 0.05; **p < 0.01; ***p < 0.001; ns not significant (by two-tailed t test with Holm−Sidak correction for multiple comparisons). 79  A.3 Testing covariate linearity assumption of Cox regression for VEGF genes in T-ALL patients 261 RNA-Seq samples from the COG TARGET study (dbGaP phs000218/000464). Fitted line should be linear for this assumption. Rlog values were used.  80  A.4 Maxstat cutpoint selection for VEGF genes in T-ALL patients 261 RNA-Seq samples from the COG TARGET study (dbGaP phs000218/000464). Three patients were excluded from the analysis due to being inevaluable. Cutpoint for each gene was selected using maximally selected rank statistics. Rlog values were used.  81  A.5 VEGFR3 surface protein level variation in preleukemic cells in vitro Surface day 21 VEGFR3 expression in GFP+Cherry+ NLTB (blue) and GFP-Cherry- normal (red) cells as measured by flow cytometry. CB cells were transduced with NLTB lentiviral constructs on day 0 in the presence of cyclosporin H. Cells were cultured on OP9-DL1 feeders over the whole duration of the experiments. They were then sorted into live CD45+GFP+Cherry+ and CD45+GFP-Cherry- populations. Mouse IgG1, kappa was used as isotype control (orange). G+C+: GFP+Cherry+; G-C-: GFP-Cherry-.        82  A.6 VEGFR3 intracellular protein levels in preleukemic cells in vitro Intracellular day 21 VEGFR3 expression in GFP+Cherry+ NLTB (blue) and GFP-Cherry- normal (red) cells as measured by flow cytometry after fixation and permeabilization. CB cells were transduced with NLTB lentiviral constructs on day 0 in the presence of cyclosporin H. Cells were cultured on OP9-DL1 feeders over the whole duration of the experiments. They were then sorted into live CD45+GFP+Cherry+ and CD45+GFP-Cherry- populations. Mouse IgG1, kappa was used as isotype control (orange). G+C+: GFP+Cherry+; G-C-: GFP-Cherry-.  83  A.7 VEGF gene expression in primary CB leukemias mRNA expression heatmap of 31 primary CB NLTB and NOTCH1E leukemias. Normalized expression was scaled by gene with mean of 0 and standard deviation of 1. 84  A.8 VEGFR3 surface protein level upon addition of 100 ng/mL VEGFC in preleukemic cells in vitro Surface VEGFR3 receptor expression in GFP+Cherry+ NLTB cells on day 27 of coculture with OP9-DL1 feeders as measured by flow cytometry. CB cells were transduced with NLTB lentiviral constructs on day 0 and day 5. Cells were cultured on OP9-DL1 feeders over the whole duration of the experiment. Exogenous 100 ng/mL human VEGFC ligand was added on day 19 of coculture. G+C+: GFP+Cherry+; G-C-: GFP-Cherry-. 85  Appendix B  Supplementary Tables  B.1 List of shRNA clones against HOX genes, FLT4 and non-targeting controls Clones obtained from the RNAi Consortium (TRC), Addgene, and Sigma-Aldrich (SHC).  86  87  88  B.2 List of primers used for NGS library construction 89   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0395080/manifest

Comment

Related Items