Stat5a Regulation by Serine Phosphorylation in Breast Cancer

Virginia Commonwealth University VCU Scholars Compass

Theses and Dissertations Graduate School

2021

STAT5A REGULATION BY SERINE PHOSPHORYLATION IN BREAST CANCER

Alicia E. Woock Virginia Commonwealth University

Follow this and additional works at: https://scholarscompass.vcu.edu/etd

Part of the Disease Modeling Commons, and the Neoplasms Commons

Downloaded from https://scholarscompass.vcu.edu/etd/6537

This Dissertation is brought to you for free and open access by the Graduate School at VCU Scholars Compass. It has been accepted for inclusion in Theses and Dissertations by an authorized administrator of VCU Scholars Compass. For more information, please contact [email protected].

©Alicia Woock, March 2021

STAT5A REGULATION BY SERINE PHOSPHORYLATION IN BREAST CANCER

A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of

Philosophy at Virginia Commonwealth University

ALICIA E. WOOCK

B.S., Chemistry

Appalachian State University, 2012

Director: Charles V. Clevenger, M.D., Ph.D.,

Chair, Department of Pathology

Virginia Commonwealth University

Richmond, Virginia

March, 2021

Abstract The neuroendocrine hormone prolactin (PRL) and its cognate receptor (PRLr) have been implicated in the pathogenesis of breast cancer. PRL signaling relies on activating kinases such as the tyrosine kinase Jak2 and serine/threonine kinases ERK1/2, NEK3, PI3K, and AKT. In the canonical pathway of PRL signaling, JAK2 phosphorylates the transcription factor STAT5a at tyrosine residue 694 (pY694-STAT5a), preceding STAT5a nuclear translocation and transcriptional activity. However, STAT5a exists with functional duality as a transcription factor, having both pro-differentiative and pro-proliferative target genes. Other STAT family members

(STATs 1, 3, and 6) have been shown to have transcriptional activity in the un-tyrosine- phosphorylated (upY) state, distinct from that of pY-STAT activity. This distinction (upY vs. pY) may underlie the duality of STAT5a, coupled with additional regulatory non-canonical posttranslational modifications. Within this notion, STAT5a contains two serine residues, S726 and

S780, whose phosphorylation are necessary for hematopoietic transformation. However, their functions in PRL-mediated breast cancer pathogenesis have not been examined. We hypothesize that STAT5a serine phosphorylation regulates STAT5a nuclear activity in a non-canonical fashion, contributing to its role in mammary oncogenesis, specifically in luminal breast cancer. As shown in a tissue microarray (TMA), human breast cancer tissues express both pS726- and pS780-

STAT5a. Nuclear Allred score for pS726-STAT5a increases two-fold with increasing tumor grade, with no difference in staining associated with estrogen or progesterone receptor (ER, PR) status, nor other clinical characteristics. Likewise, patient derived xenograft (PDX) tumors of various molecular subtypes express pS726- and pS780-STAT5a. Phosphorylation of S726-STAT5a is

PRL-responsive in vitro. To examine the functional significance of STAT5a serine phosphorylation in vitro, we have performed STAT5a knockdown (KD) in the breast cancer cell

iii

line MCF7. Following STAT5a KD, cells were rescued with phospho-site specific STAT5a mutant constructs targeting Y694, S726 and S780. Characteristics of breast cancer examined in these mutation-carrying cells, including anchorage-independent growth and proliferation, show distinct phenotypes compared to controls. Further, PRL-inducible gene expression changes on the global transcriptomic level showed differential effects of the STAT5a mutations compared to wild type

STAT5a, specifically enrichment of different oncogenic signatures and transcription factor programs. Mechanistic studies examine the ability of these STAT5a mutant proteins to undergo nuclear translocation and their regulatory role on phosphorylation of STAT5a tyrosine residue 694.

Collectively, these studies provide novel insights into the role of the non-canonical pathway of

STAT5a activation in breast cancer pathogenesis.

Acknowledgments “Great things are not done by impulse, but by a series of small things brought together.”-Vincent van Gogh

Completion of this thesis would not have been possible without the help of others. First,

I’d like to acknowledge my committee members, Dr. Larisa Litovchick, Dr. Chuck Harrell, Dr.

Joyce Lloyd, and Dr. Devanand Sarkar, for their support of my project and for pushing my science and understanding further.

I’d also like to acknowledge my co-authors on my manuscript submitted February 5, 2021, who provided tissue samples, performed analyses, and gave constructive feedback on the manuscript. Drs. Michael Idowu and Patricija Zot provided tissues for the tissue microarray and analysis. Dr. Chuck Harrell provided PDX samples, which were prepared by either Tia H. Turner or Mohammad A. Alzubi. Amy L. Olex performed the RNA-seq quality control and alignment and performed/provided guidance on RNA-seq analysis. She was extremely helpful throughout the process of analyzing the RNA-seq. Finally, Jackie M. Grible: not only a co-author and former lab member, but a friend. Jackie helped with technical aspects of my research, performing and analyzing the soft agar assay. But she was also a sounding board for ideas, and I appreciate her letting me hang out with her dogs, Tommy and Sammy, when I needed dog cuddles.

I’d also like to acknowledge our laboratory technicians, Shannon E. Hedrick and Madhavi

Puchalapalli for their technical support. Shannon’s knowledge of molecular biology, and her time, were so appreciated. Her technical support on my project ranged from performing the IHC for the

TMA, to teaching me how to clone, run PCR, and analyze DNA gels. I would not be the scientist

I am today without her.

I’d like to acknowledge my mentor and Principal Investigator, Dr. Charles Clevenger. His guidance and support of my project got it to where it is today. In my time in his lab I have learned so much about science and the process of research. His mentoring allowed me to become an independent researcher, and for that I am thankful.

Lastly, I’d also like to acknowledge my family and friends, who supported me, kept encouraging me, and helped balance me. Our use of FaceTime has been impressive.

Abbreviations

(v/v) Volume per volume ratio (w/v) Weight per volume ratio A Alanine AA Amino acid ABL Abelson AKT Protein kinase B AML Acute myeloid leukemia ANOVA Analysis of variance APC Adenomatous polyposis coli APS Ammonium persulfate AR Androgen receptor ATCC American Type Culture Collection bcl6 B-cell lymphoma 6 BLAST Basic local alignment search tool BLG β-lactoglobulin gene BME β-mercaptoethanol bp Base pairs BRCA Breast cancer BSA Bovine serum albumin C Celsius C- Carboxy c-Src Cellular-Src CBP CREB-binding protein CC3 Cleaved caspase 3 CC7 Cleaved caspase 7 CCND1 Cyclin D1

vii

cDNA Complementary deoxyribonucleic acid ChIP Chromatin immunoprecipitation CISH Cytokine-inducible SH2-containing protein CML Chronic myeloid leukemia CMV Cytomegalovirus Co-IP Co-immunoprecipitation

CO2 Carbon dioxide CTCF CCCTC-binding factor CypA Cyclophilin A d Day DAPI 4',6-Diamidino-2-Phenylindole, Dihydrochloride DCIS Ductal carcinoma in situ DEG Differential gene expression Dex Dexamethasone dH2O Distilled water DMEM Dulbecco’s modified eagle’s medium DMSO Dimethylsulfoxide DN Dominant-negative DNA Deoxyribonucleic acid DNase Deoxyribonuclease DPBS Dulbecco’s phosphate-buffered saline E2 Estrogen ECL Enhanced chemiluminescence ECM Extracellular matrix EDTA Ethylene diamine tetra acetic acid EGFR Epidermal growth factor receptor EGTA Ethylene glycol-bis (aminoethylether) N,N,N',N'-tetraacetic acid

viii

EMSA Electrophoretic mobility shift assay EMT Epithelial-to-mesenchymal transition EPO Erythropoietin ER Estrogen receptor ERE Estrogen response element ERK Extracellular signal-related kinase ES Enrichment score EV Empty vector EZH2 Enhancer of zeste homolog 2 F Phenylalanine FAK Focal adhesion kinase FBS Fetal bovine serum FDR False discovery rate FL Full length For Forward g Gram GAS Gamma-activated sequence GEF Guanine nucleotide exchange factor GFP Green fluorescent protein GH Growth hormone GHR Growth hormone receptor Gly Glycine GM-CSF Granulocyte-macrophage colony-stimulating factor GR Glucocorticoid receptor GSEA Gene set enrichment analysis h Hour H3K27me3 Histone H3 tri-methylation of lysine 27

HCl Hydrochloric acid HDAC6 Histone deacetylase 6 HEK293T Human embryonic kidney HMGN2 High-mobility group N2 HP1 Heterochromatin protein 1 HRP Horseradish peroxidase HUVEC Human umbilical vein endothelial cell IB Immunoblotting ICS-2 Interferon consensus sequence 2 IF Immunofluorescence IFN Interferon IGF Insulin-like growth factor IgG Immunoglobulin G IHC Immunohistochemistry IL Interleukin IP Immunoprecipitation IPA Ingenuity pathway analysis IRF Interferon regulatory factor Jak Janus kinase kb Kilobases KD Knockdown KDa Kilo Dalton KiPIK Kinase inhibitor profiling to identify kinases KO Knockout LB Luria-Bertani broth LSP Leu-Ser-Pro motif LTE2 Long-term estrogen independence

M Molar mAb Monoclonal antibody MAPK Mitogen-activated protein kinase MEF Mouse embryonic fibroblast MEK Mitogen-activated protein kinase kinase mg Milligram MGT Mammary gland tumor min Minutes ml Milliliter mM Millimolar MMTV-PyMT Mammary tumor virus-polyoma middle tumor-antigen mRNA Messenger ribonucleic acid MTT 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide N- amino NaCl Sodium chloride NEB New England Biolabs NEK NIMA-related kinase NEP New England Peptides NES Normalized enrichment score ng Nanogram NHS Nurses’ Health Study NIH National Institutes of Health NIMA Never in mitosis A NK Natural killer NRL Neu-related lipocalin NSG NOD scid gamma NT Non-targeting control

O/N Overnight OD Optical density ORF Open reading frame OS Oncogenic signature P Progesterone p- Phospho- pAb Polyclonal antibody PAK p21-activated kinase PBS Phosphate-buffered saline PCR Polymerase chain reaction PDX Patient-derived xenograft PE Paired-end PI Protease inhibitor PI3K Phosphatidylinositol-3 kinase PL Placental lactogen PMSP Pro-Met-Ser-Pro motif PPI Peptidyl-prolyl isomerase PPI Protein phosphatase inhibitor PR Progesterone receptor PRL Prolactin PRLr Prolactin receptor PRLrI Intermediate prolactin receptor isoform PRLrL Long prolactin receptor isoform Pro Proline pS Phosphorylated serine PSP Pro-Ser-Pro motif PTP Protein tyrosine phosphatase

xii

Puro Puromycin PVDF Polyvinyl difluoride pY Phosphorylated tyrosine r Recombinant Rev Reverse RIPA Radioimmunoprecipitation assay RPM Revolutions per minute rPRL Rat prolactin RR Relative risk RT Reverse transcriptase RT Room temperature RT-PCR Reverse transcriptase polymerase chain reaction S Serine SDS-PAGE Sodium dodecyl sulfate polyacrylamide gel electrophoresis Sec Second S.E. Standard error of the mean Seq Sequencing Ser Serine SFM Serum-free medium SH2 Src Homology 2 SHP SH2-containing phosphatase shRNA Short hairpin RNA siRNA Small interfering RNA STAT Signal transducer and activator of transcription TAD Transactivation domain TAg Simian Virus 40 T antigen TBE Tris/borate/EDTA

xiii

TBS Tris-buffered saline TBST Tris-buffered saline with Tween-20 TCGA The Cancer Genome Atlas TF Transcription factor TGF Transforming growth factor TMA Tissue microarray TNBC Triple negative breast cancer TNF Tumor necrosis factor TPO Thrombopoietin Tris Tris(hydroxymethyl)aminomethane Tyr Tyrosine U Units μg Microgram μl Microliter μm Micrometer μM Micromolar UP- Unphosphorylated UTR Untranslated region UV Ultraviolet V Variable V Volts W Tryptophan WAP Whey acidic protein WCL Whole cell lysate WSXWS Tryptophan-Serine-Xaa-Tryptophan-Serine WT Wild-type Y Tyrosine

xiv

Zeo Zeocin

Dedication

This thesis is dedicated to my parents. My dad’s stalwart support for my education is beyond admirable. Even while going through his own cancer journey, he’s been there for me. It is hard to express how thankful I am for him and his encouragement. My mom has been my number one fan, and she’s always there for me whenever I needed. She has simultaneously been my support system and my dad’s, and I don’t think either of us would be where we are today without her or her cookies. Even if they don’t understand most of what I say when I talk about my research, they still listen and have continuously encouraged me to keep learning and growing.

xvi

Contents Abstract ...... iii Acknowledgments...... v Abbreviations ...... vii Dedication ...... xvi List of Tables ...... xx List of Figures ...... xxii 1. Chapter 1: Introduction ...... 1 1.1 Breast Cancer ...... 1 1.2. Prolactin ...... 2 1.3. Prolactin receptor ...... 3 1.3.1. Isoforms: ...... 4 1.3.2. Downstream signaling ...... 5 1.4. PRL/PRLR in mammary development and pregnancy ...... 8 1.5. PRL/PRLR in breast cancer ...... 9 1.5.1. PRL and breast cancer risk ...... 10 1.5.2. Preclinical studies of PRL in vitro and mouse models ...... 11 1.5.3. In vitro studies of PRL/PRLr signaling mechanisms ...... 15 1.6. STAT5a ...... 17 1.6.1. STAT family ...... 17 1.6.2. Regulation of STAT5a signaling ...... 24 1.6.3. Canonical pathway of JAK/STAT ...... 28 1.6.4. Non-canonical activity of STATs without tyrosine phosphorylation ...... 34 1.7. Goal of this work and hypothesis ...... 44 Aim 1:...... 44 Aim 2: ...... 45 Aim 3: ...... 45 2. Chapter 2: Materials and Methods ...... 46 2.1 Materials ...... 46 2.1.2 Chemicals ...... 46 2.1.3 Buffers, Media, Stock solutions ...... 47

xvii

2.1.4 Kits, Reagents, and Enzymes ...... 48 2.1.5 Prolactin ...... 48 2.1.6 Mammalian Cell Lines and tissue culture ...... 48 2.1.7 Tissue culture reagents and media ...... 49 2.1.8 Plasmids ...... 50 2.1.9 Primers ...... 51 2.1.10 Stat5a shRNA...... 53 2.1.11 Antibodies ...... 54 2.1.12 Secondary Antibodies ...... 54 2.1.13 Software & Online Tools ...... 55 2.2 Methods ...... 56 2.2.2 Molecular Biology methods ...... 56 2.2.3 Protein methods ...... 65 2.2.4 Cell culture methods ...... 69 2.2.5 Tissue methods...... 72 2.2.6 RNA-sequencing methods ...... 73 2.2.7 Statistical analysis ...... 75 3. Chapter 3: Serine residues 726 and 780 have nonredundant roles regulating STAT5a activity in luminal breast cancer ...... 77 3.1. Introduction ...... 77 3.2. Hypothesis ...... 79 3.3. Results ...... 79 3.4. Discussion ...... 95 4. Chapter 4: Studies of double-point STAT5a mutants illustrate requirements for cooperative activation of serine and tyrosine residues ...... 100 4.1. Introduction ...... 100 4.2. Hypothesis ...... 101 4.3. Results ...... 102 4.4. Discussion ...... 108 5. Chapter 5: Global transcriptomic changes with expression of STAT5a single-point mutants by RNA-sequencing ...... 111 5.1. Introduction ...... 111

xviii

5.2. Hypothesis ...... 112 5.3. Results ...... 112 5.4. Discussion ...... 131 6. Chapter 6: Regulation of STAT5a serine phosphorylation by Prolactin ...... 139 6.1. Introduction ...... 139 6.2 Hypothesis ...... 140 6.3 Results ...... 141 6.4 Discussion ...... 142 7. Chapter 7: Discussion ...... 146 7.1. Discussion ...... 146 7.2. Limitations ...... 150 7.3. Future Directions ...... 151 8. Appendices ...... 155 8.1. Supplementary Figures for Chapter 3 ...... 155 8.2. Supplementary Tables ...... 160 8.2. Uncropped Western blots for Woock et al (Chapter 3), submitted February 2021 ...... 206 Bibliography ...... 214 Vita ...... 235

xix

List of Tables Page 1.1 Selection of breast cancer cell lines with published PRLr expression. 12

2.1 Chemicals used in the pursuit of this dissertation. 46

2.2 Buffer Recipes and Suppliers. 47

2.3 Kits and Reagents. 48

2.4 Human cell lines used for in vitro work. 48

2.5 Tissue Culture Reagents. 49

2.6 Primers used in this dissertation work. 52

2.7 shRNA constructs. 53

2.8 Primary antibodies. 54

2.9 Secondary antibodies. 54

2.10 Software and online tools. 55

2.11 Reaction using NEBNext® High-Fidelity 2X PCR Master Mix 61

2.12 Reaction using OneTaq® 2X Master Mix with Standard Buffer 62

2.13 Reaction for Phusion Site-Directed Mutagenesis 62

2.14 Reaction for Q5 Site-Directed Mutagenesis. 64

2.15 qRT-PCR Reaction settings. 64

4.1 Cases with Nuclear Allred Scores for both pS726 and pS780 as scored in the breast cancer

TMA. 102

5.1 Significant differentially expressed genes (DEGs) for indicated contrast analyses performed in

EdgeR filtered for counts per million (CPM)>0.5 to filter lowly expressed genes in the dataset library. 113

5.2 Differentially expressed genes determined by RNA-seq analysis of STAT5a-rescued MCF7 cells treated with PRL. (S2 for Woock et al) 114

S1 for Woock et al. Patient characteristics of TMA. 160

S3 for Woock et al. Quality check on WT-STAT5a (PRL vs untreated) dataset presented compared to microarray/RNA seq published previously. 161

8.1 Differentially expressed genes for MCF7 cells expressing WT-STAT5a, PRL vs untreated, log10(fold change)>0.58, fold change >1.5. 166

8.2 Differentially expressed genes for MCF7 cells expressing Y694F-STAT5a, PRL vs untreated log10(fold change)>0.58, fold change >1.5. 179

8.3 Differentially expressed genes for MCF7 cells expressing S726A-STAT5a, PRL vs untreated log10(fold change)>0.58, fold change >1.5. 184

8.4 Differentially expressed genes for MCF7 cells expressing S780A-STAT5a, PRL vs untreated log10(fold change)>0.58, fold change >1.5. 191

8.5 Details for top 10 enriched oncogenic signatures for each STAT5a mutant in the presence of

PRL. Terms were ranked by normalized enrichment score (NES). 197

8.6 Details for top 10 enriched oncogenic signatures for each untreated STAT5a mutant. Terms were ranked by normalized enrichment score (NES). 200

8.7 Details for top 10 ENCODE TF enriched terms for each STAT5a mutant. Terms were ranked by odds ratio. 203

8.8 Odds ratio of top 10 enriched ENCODE Transcription Factors for each STAT5a mutant in gene set enrichment analysis using Enrichr. Data used for Figure 7 in Woock et al (submitted).

204

xxi

List of Figures Page 1.1. Schema of the PRLr and the conformational change PRLr dimers undergo upon PRL-binding.

1.2 Amino acid sequences of STAT family of proteins illustrate the homology of the transcription factors. 20

1.3. Structural homology of the STAT family and domain functions. 23

1.4 Described roles for pY694-, pS726-, and pS780-STAT5a in regulating STAT5a activity.

3.1. STAT5a S726 and S780 phosphorylation occur in human breast cancer 81

3.2. The human breast cancer cell line MCF7 as a model for STAT5a serine phosphorylation

3.3 Expression of STAT5a phospho-mutants differentially affects PRL-regulated gene expression in MCF7 cells 86

3.4. Effect of single-point phospho-deficient STAT5a mutants on characteristics of MCF7 cells

3.5. Induction of phosphorylation on residue Y694 is independent of either S726 or S780.

3.6. Nuclear translocation of STAT5a occurs in absence of pS726 and/or pS780. 93

3.7. Transcription factor gene set analysis identifies complex transcriptional networks contributing to STAT5a mutant-driven phenotypes 95

4.1 Concept and schematic of MCF7 rescues with STAT5a double point phospho-deficient mutants 103

xxii

4.2 MCF7 cells expressing STAT5a lacking both Y694 and S780 or both S726 and S780 are unable to proliferate, and prolactin does not rescue this phenotype. 105

4.3 Loss of either STAT5a serine residue in combination with loss of Y694 or loss of both serine residues decreases MCF7 ability to grow in soft agar as measured by both colony number and colony size. 107

4.4 Loss of both STAT5a serine residues in combination expedites MCF7 migration. 108

5.1 Significant differentially expressed genes in MCF7 cells expressing STAT5a phospho-point mutants are distinct from WT-STAT5a expression 115

5.2 Confirmation of STAT5a point mutations by RNA-sequencing 117

5.3 STAT5a signaling is predicted to be downregulated in MCF7 cells expressing the empty vector

(EV) control by Ingenuity Pathway Analysis (IPA). 119

5.4 IPA functional pathway enrichment analysis of MCF7s expressing EV compared to WT-

STAT5a largely shows inhibition of chosen pathways 120

5.5 Predicted functional effects of the change in magnitude in PRL-induced genes in MCF7 cells expressing the STAT5a mutants compared to WT-STAT5a. 122

5.6 MCF7 cells expressing S726A-STAT5a or Y694F-STAT5a share significant enriched oncogenic signatures 124

5.7 Transcription factor (TF) enrichment shows distinct transcriptional patterns in MCF7 cells expressing STAT5a mutants. 127

5.8 MCF7 cells expressing S726A-STAT5a and Y694F-STAT5a share the most enriched transcription factor programs in the presence of PRL. 128

xxiii

5.9 Pearson’s sample-wise correlation analyses of MCF7 expressing STAT5a phospho-mutants

RNA-seq and RNA-seq from PDX tumors and metastases and the Cancer Genome Atlas (TCGA) datasets. 130

5.10 MCF7 cells expressing STAT5a mutants hierarchically cluster together when combined with

RNA-seq data from PDX mammary gland tumors and data from 817 tumors from the Cancer

Genome Atlas (TCGA). 131

6.1 ERK1/2 interact with STAT5a is dependent on presence of S726 but not S780. 141

6.2 Phosphorylation of S726 and S780 are independent of the MEK pathway. 142

S1. TMA for pY694-STAT5a shows weak nuclear staining 155

S2. Confirmation of STAT5a knockdown and rescue at the mRNA level in MCF7 cells. 156

S3. Gene induction is affected differentially by loss of STAT5a phosphorylation. 157

S4. Effect of single-point phospho-deficient STAT5a mutants on characteristics of MCF7 cells.

158

S5. Effect of single-point phospho-deficient STAT5a mutants on characteristics of MCF7 cells.

159

S6. Validation of specificity of STAT5a staining using empty vector (EV) control cell line.

160

3.1D: Full membrane western blot data 206

3.1E: Full membrane western blot data 207

3.1F: Full membrane western blot data 208

3.2B: Full membrane western blot data 209

3.2D: Full membrane western blot data 210

3.4C: Full membrane western blot data 211

xxiv

3.5A and 3.5B: Full membrane western blot data 212

3.6B: Full membrane western blot data 213

xxv

1. Chapter 1: Introduction

1.1 Breast Cancer

In the United States, breast cancer is the most commonly diagnosed cancer in women and the second most common cause of death from cancer in women.1 In 2021, an estimated 281,550 new cases of invasive carcinoma will be diagnosed in women, while around 43,600 women will die of the disease.2 Relative survival rates of women diagnosed with breast cancer are now 91% five years after diagnosis. Much progress has been made in understanding the disease and in our ability to diagnose breast cancer, and as such the incidence rate of invasive breast cancer rose 0.3% per year during the period from 2012-2016.3

The clinical course of women with breast cancer is dependent upon the molecular subtype of breast cancer that is diagnosed, based on gene expression analysis and particularly the expression of hormone receptors (HR) estrogen receptor and progesterone receptor (ER and PR, respectively) and human epidermal growth factor 2 (HER2). There are five molecular subtypes of breast cancer, namely luminal A, luminal B, HER2-enriched, basal-like, and claudin-low.4

Luminal A breast cancers are the most common subtype of breast cancer and are ER/PR positive (ER+/PR+) and HER2 negative (HER2-). These cancers are the least aggressive, with the most favorable prognosis, in part due to their responsiveness to hormonal therapies. Luminal B breast cancers are both ER+/PR+ and HER2+, are more proliferative than luminal A, and have slightly poorer outcomes. HER2-enriched breast cancers are HR- (ER-/PR-) and HER2+, and are currently treated with HER2-targeting therapies. Basal-like, or triple negative breast cancers

(TNBCs), are both HR- and HER2-, with a worse prognosis compared to luminal A. A further

subtype of TNBCs is the claudin-low subtype of breast cancer, which also expresses low amounts of genes involved in tight-junctions and epithelial cell-cell adhesions (proteins of the claudin family, E-cadherin). Claudin-low tumors are characterized by enrichment of genes involved in epithelial to mesenchymal transition (EMT), and have poor prognosis related to chemotherapy- resistance.5

The molecular subtypes of breast cancer illustrate the importance of understanding genetic alterations that drive mammary oncogenesis. The control of these genetic alterations in breast cancer is complex, with epigenetic changes to the DNA allowing for the activation of tumor- promoting changes or repression of tumor-suppressing genes in response to a multitude of different signaling pathways. These signaling pathways are active in normal epithelial cells to modulate different stages of mammary gland development and mammary function but become dysregulated or hijacked in cancer cells. This leads to alterations in the transcription of genes that control cellular proliferation, apoptosis, differentiation, and motility, contributing to the onset of breast cancer and the heterogeneity of the disease.

1.2.Prolactin

Prolactin (PRL) is a 23 kDa neuroendocrine polypeptide hormone produced mainly in the lactotroph cells of the anterior pituitary but is also produced in various other tissues such as the breast and uterus.6 PRL is a class I helical protein hormone and belongs to the PRL/growth hormone (GH)/placental lactogen (PL) family of protein hormones.7–9 PRL acts in an autocrine, paracrine, and endocrine manner at target tissues when it binds to the prolactin receptor (PRLr).

PRL has multiple target tissues, and in fact it is now appreciated to have more than 300 separate actions, including roles in metabolic homeostasis, and immunomodulation.7,10 However, and most

importantly to this work, its most well-characterized role is for its actions in the mammary gland.

PRL was first identified to have an autocrine/paracrine effect in breast epithelium by Clevenger et al, where it acts in a complex concert along with estrogen and progesterone signaling in normal mammary development.6 PRL is responsible for stimulating lobuloalveolar growth during pregnancy, as well as lactation postpartum. Specifically, PRL (PRL/PRLr signaling) is necessary for the differentiation of mammary epithelial cells into milk-producing cells during late pregnancy.11,12

1.3.Prolactin receptor

The prolactin receptor (PRLr) is a cell-surface receptor belonging in the hematopoietic type

1 cytokine receptor superfamily, along with the growth hormone receptor (GHr), granulocyte- macrophage colony-stimulating factor (GM-CSF) receptor, interleukin (IL) receptors, and the erythropoietin (EPO) receptor.13 The PRLr is a single-pass transmembrane receptor, existing as a dimer pair that lacks intrinsic kinase activity and relies on associated kinases for downstream signaling following ligand binding. Activation of receptor-associated kinases occurs as a consequence of conformational change of the dimer pair upon ligand binding. The conformational change of the PRLr is assisted by the peptidyl prolyl isomerase cyclophilin A (CypA).14,15 The

PRLr itself has three main domains: the extracellular domain (ECD, where PRL binds), the transmembrane domain (TMD), and the intracellular domain (ICD). The ECD contains two type

III fibronectin-like domains, the S1 and S2 domains. The N-terminal S1 domain contains the majority of ligand-binding sites, such that the receptor can bind the ligand in a 2:1 ratio.16,17

Although the S2 domain has a smaller surface area for ligand contact, it does have a highly conserved amino acid (aa) WSXWS motif (that is, Tryptophan-Serine-Xaa-Tryptophan-Serine

motif) that is both necessary for ligand-binding and receptor dimerization.18,19 The TMD, as it spans the cytoplasmic membrane, links the activity of the ECD binding to ligand to intracellular signaling, in an event that is not yet fully understood as the crystal structure of the PRLr has not been fully determined. Instead, we look to studies of the GHr, where ligand binding results in conformational change of the TMD dimers so that the ICD can then undergo conformational change to bring signaling kinases in close proximity to activate downstream signaling.20,21 The

ICD is highly homologous to other cytokine receptors, containing common motifs (from membrane-proximal to C-tail) known as Box 1, Variable (V) Box, Box 2, X Box, and a unique, intrinsically-disordered, C-terminal tail. Briefly, Box 1 is highly proline-rich, and contains the site necessary for JAK2-PRLr association.22,23 The V Box links Box 1 and Box 2, and little is known about its function. Box 2 contains hydrophobic, acidic, amino acid residues. The X Box contains the site necessary for Cyp binding; as a peptidyl prolyl isomerase, CypA is critical for downstream signaling of PRL-PRLr binding and is discussed further below.14,15

1.3.1. Isoforms:

There are seven identified isoforms of the PRLr that have variable expression in both normal and malignant tissue (Figure 1.1A). Dimerized receptor pairs can thus be homo- or heterodimers of these isoforms. Each form has altered downstream signaling from PRL-PRLr binding, although most of our understanding comes from the work done studying the long form.

1. Long: this was the first human PRLr isoform identified, 598 amino acids long. Upon PRL

binding, there is rapid phosphorylation of the ICD and subsequent downstream signaling.24

2. ΔS1: this isoform is generated by alternative splicing with deletion of the S1 motif of the ECD.

Therefore, the affinity for ligand-binding compared to the long form is about 94-fold less (with

less sites for ligand binding available), although signaling remains intact.24

3. Intermediate: this isoform is also generated by an alternative splicing event which induces a

frame shift after amino acid residue 316 with the addition of a novel, 13 amino acid tail due to

a premature stop codon. Functionally, this causes deletion of a portion of the ICD including

the site for STAT5a binding to the PRLr in the C-terminal tail.24 The function of the

intermediate PRLr will be discussed further for its role in breast cancer in Section 1.5.3.

4. Short 1a (S1a): short form of PRLr generated by alternative splicing in which part of exon 11

replaces exon 10, contains both Box 1 and Box 2 motifs, does not have downstream signaling.25

5. Short 1b (S1b): short form of PRLr generated by alternative splicing in which part of exon 11

replaces exon 10, contains Box 1 motif only, does not have downstream signaling.25

6. PRL-binding protein (PRLBP): generated by a proteolytic event, this is a freely circulating

form of the ECD only, and thus can still engage ligand but does not signal. It therefore acts as

a “ligand trap”, and in fact as much as 36% of circulating PRL is bound to PRLBP.26

7. TM-ICD: this fragment lacks the ECD and is possibly generated by the same proteolytic

cleavage event that generates the PRLBP, it can form heterodimers with other forms of the

PRLr and therefore maintains signaling events through JAK2.27

1.3.2. Downstream signaling

As stated above, as ligand binds the pre-dimerized PRLr, the peptidyl prolyl isomerase

CypA induces a conformational change of the ICD that allows associated kinases to come close in proximity to phosphorylate the receptor and themselves (termed autophosphorylation, Figure

1.1B). This starts downstream signaling pathways that alter cellular growth, proliferation, survival, differentiation, and motility.28

Figure 1.1 Schema of the PRLr and the conformational change PRLr dimers undergo upon PRL-binding. A) The PRLr isoforms shown with conserved proximal box motifs in each ICD and unique intracellular tails for the intermediate, S1a, and S1b receptors. B) The long form PRLr exists pre-dimerized. Upon PRL-PRLr binding, CypA facilitates cis-trans isomerization of the ICD, allowing auto-/trans-tyrosine phosphorylation of the kinase JAK2 to initiate downstream signaling events.

1.3.2.1. JAK2

The tyrosine kinase JAK2 is constitutively associated with the PRLr.29 Upon ligand binding to the PRLr, JAK2 is rapidly trans- and auto-phosphorylated within 1 min, and reaching maximum

phosphorylation levels between 10-15 min.24 JAK2 is responsible for phosphorylating the C- terminal tyrosine residues in the PRLr for the subsequent association and tyrosine-phosphorylation of the transcription factor STAT5.24 The role of STAT5 will be discussed further below.

1.3.2.2. RAS/RAF/MAPK

Within 5 minutes of PRL stimulation in both normal mammary epithelial cells and breast cancer cells, the mitogen-activated protein kinase (MAPK) is activated.30,31 Specifically, RAF-1,

MEK1, and the 42 and 44 kDa MAPKs known as extracellular regulated kinases 1 and 2 (ERK1/2) are induced 2-3 fold with PRL treatment, leading to PRL-regulated mitogenesis.30,31 ERK1/2 have some documented effects on STAT5 signaling in response to several cytokines, including PRL and GH.32–34 These interactions will be further detailed below, in the discussion of STAT5 regulation.

1.3.2.3. PI3K/AKT

Upon activation of the PRLr, the p85 subunit of phosphatidylinositol-3 kinase (PI3K) associates with the PRLr, leading to its activation and the generation of metabolites known as phosphoinositides. These metabolites act as docking sites for AKT (also known as protein kinase

B) and kinases upstream of AKT which activate it by serine/threonine phosphorylation.28,35,36 The

AKT pathway initiates survival, inhibits pro-apoptotic signals (including transforming growth factor [TGF]-β-induced apoptosis), and modulates regulators of cell cycle progression including

Cyclin D1.28,36,37

1.3.2.4. NEK3/VAV2/RAC1

The guanine nucleotide exchange factor (GEF) VAV was first found to be activated by

PRL stimulation in T-cells by serine/threonine phosphorylation and resulting in both GEF activity

(that is, activation of RAC), and nuclear translocation of VAV.38 It has since been shown that the

NIMA (never in mitosis A)-related family kinase NEK3 is responsible for the phosphorylation and activation of VAV2 in breast cancer, leading to the exchange of GDP for GTP on RAC1, and its activation. This pathway is ultimately implicated in PRL-stimulated tumor progression and motility through cytoskeletal reorganization and formation of stress fibers and lamellipodia for cellular migration.21,22,23

1.4. PRL/PRLr in mammary development and pregnancy

Mammary gland development takes place in a series of distinct phases controlled by circulating hormones including GH, estrogen (E2), progesterone (P), and PRL.42,43 Although some development occurs in utero, most development occurs primarily after birth specifically at puberty

(ductal development), and pregnancy (ductal/alveolar development in preparation for lactation).11

Along with P, PRL works to stimulate mammary gland proliferation and differentiation of lobules prior to lactation.44,45 In fact, P induces the expression of PRLr and PRL induces the expression of PR in mammary cells, illustrating the complexity of the crosstalk needed between hormones for mammary development.46–48 During pregnancy, epithelial cells proliferate and generate alveolar buds that progressively cleave into distinct alveoli. By late pregnancy these alveoli involve most of the fat pad. By lactation, these alveoli have become milk-secreting lobules.42 At the end of lactation, that is, when the suckling stimulus is removed at weaning, involution remodels the epithelial milk-secreting lobules back to a simple ductal architecture.42

The role for PRL in mammary development has been elegantly studied in Prl and Prlr knockout mice, which have made it clear that PRL is necessary for side-branching and alveolar budding in pregnancy, but that embryonic and postnatal development do not rely on PRL.11,12,49,50

In 1997, Ormandy et al showed that in mice carrying one mutated allele for Prlr, that is a heterozygous null mutation for the Prlr (Prlr+/-), mammary gland proliferation during the first pregnancy was so reduced as to lead to lactational defects, suggesting that epithelial cell proliferation depends on a threshold of Prlr expression that cannot occur even with one functional gene. Interestingly, after further estrous cycles, during a second pregnancy the mammary gland was more fully developed and capable of lactation.11 Targeting the Prl gene by homologous recombination resulted in homozygous female mice (Prl-/-) that were infertile, and although the mammary glands were able to develop a normal ductal tree, lobules failed to develop and the mammary gland resembled that of an adult virgin mouse. Heterozygous females (Prlm+/-) were able to produce Prl at normal bioactive levels, suggesting that loss of one Prl locus can be compensated for in such a way that PRL synthesis is not reduced.12 Rescue of Prl knockout mice with the human transgene, in a humanized PRL model, rescued all reproductive defects and mammary gland development.49 The requirement for PRL/PRLr signaling in alveologenesis was definitively demonstrated when, in 2003, Ormandy et al transplanted Prlr KO epithelium into the stroma of WT mice, which then developed alveolar buds during pregnancy but no lobuloalveolar development. 51 The recombination of WT epithelium and Prlr KO stroma showed normal development, which illustrated that a direct action of PRL on the epithelium is necessary for lobuloalveolar development.

1.5. PRL/PRLr in breast cancer

The function of PRL in human breast cancer has been the subject of debate since it was first recognized to contribute to rodent mammary cancer in the 1970s.52 However, in the 40+ intervening years since this discovery, mounting evidence at the genetic, cellular, and

epidemiologic levels strongly implicate autocrine/paracrine PRL/PRLr signaling in the pathogenesis of breast cancer.

1.5.1. PRL and breast cancer risk

Clinically, evidence for a positive association of plasma PRL levels and human breast cancer was established by examining the Nurses’ Health Study (NHS) and the NHS II. 53–56 The

NHS and NHSII are two large cohorts of registered nurses that were established in 1976 and 1989, respectively. When first started, the NHS cohort of 121,700 US female nurses ranged in age from

30 to 55 years old, and the NHSII cohort of 116,609 female nurses ranged in age from 25 to 42 years old. Follow-up to the NHS was reported in 2000 as 99%.55 Together, the cohorts cover both pre- and post-menopausal women and therefore, risk of breast cancer associated with various life factors. Using these cohorts, prospective, nested case-control studies were performed studying the risk of breast cancer and PRL serum levels at different time points of follow up.53–56

Examination of PRL levels and breast cancer risk in the NHS and NHSII resulted in multiple data sets to compare risk for women who are premenopausal and risk for women who are postmenopausal. A pooled analysis, along with these data sets, provided a summary of about 80% cases of published prospective studies.55 With the power of these studies, relative risk (RR) was calculated and could be associated with risk broken down by menopause status, tumor invasiveness, and lead-time to diagnosis.

First, plasma PRL concentration was associated with breast cancer with a RR of 1.5 in both the NHS and NHSII women who were premenopausal. The risk of breast cancer increased to 60-

70% for premenopausal women older than 45 years old with high serum PRL.55 A positive association for PRL concentration and breast cancer risk for postmenopausal women was also found, confined to invasive ER+ tumors.56

In the pooled analysis (1,500 cases, in which both pre- and postmenopausal women were combined), the risk of breast cancer was 30% comparing the top versus bottom quartile of PRL concentrations, independent of estrogen and testosterone levels.55 In this combined analysis, there were no significant differences for tumor type or invasiveness, although a positive association for

ER+ tumors was observed once more.55 One of the most interesting findings of these studies was a strong relationship between plasma PRL levels and breast cancer in women who were diagnosed less than 4 years after blood collection, suggesting that increased circulating levels of PRL may be due to subclinical breast tumors secreting PRL.55,57,58 Overall, these studies showed that PRL is associated with ER+, post-menopausal disease.53–56

1.5.2. Preclinical studies of PRL in vitro and mouse models

PRL promotes the growth, proliferation, and motility of breast cancer cells in vitro, acting as a mitogen and survival factor.44,52,53,59–63 These activities of PRL have been studied in a range of cultured mammary carcinoma cell lines that vary in their oncogenic mutations and steroid hormone (E2 and P) responsive-ness, illustrating PRLs role in a wide range of breast cancers.

The study of the autocrine/paracrine effects of PRL in breast cancer cells in vitro began after clinical trials treating breast cancer patients using bromocriptine, a dopamine agonist that inhibits PRL secretion from the pituitary, failed.28 The pervading theory at the time was that PRL was secreted exclusively from the pituitary, but the failure of bromocriptine in these trials led researchers in the Clevenger and Vonderhaar laboratories to hypothesize separately that breast epithelium (and other nonendocrine tissues) locally produce PRL.6,28,44,57,64 These studies demonstrated that cultured breast cancer cells, namely MCF7 and MDA-MB-231, could synthesize and secrete PRL.6 Further evidence comes from using in situ hybridization of patient samples:

98% of human breast cancers synthesize PRL mRNA.64

The absolute expression levels (and relative expression levels of isoforms) of PRLr in established culture cell lines is also varied, and has been quantified on both the mRNA and protein levels.28,48 In 1997, Ormandy et al studied the mRNA expression of PRLr of 20 human breast cancer cell lines which model subtypes ranging from luminal A to triple negative, and stratified these lines from non-detected to high expression of PRLr mRNA (those lines which were quantified as to have measurable PRLr expression are listed in Table 1.1). They were further able to correlate the level of PRLr expression to PRLr-binding activity and showed that this was a positively correlated relationship. This suggests that any potential posttranslational modifications of the PRLr do not affect the maintenance of steady state PRLr levels.48 Finally, they showed that

PRL induces the transcriptional upregulation of PRLr, thus potentiating further PRL/PRLr signal transduction.48

Table 1.1 Selection of breast cancer cell lines with published PRLr expression.

Breast cancer cell line Molecular subtype PRLr expression MDA-MB-134 Luminal A65,66 High PRLr48 BT483 Luminal B65,66 High PRLr48 T47D Luminal A/B65,66 High PRLr48 BT474 Luminal B65,66 High PRLr48 MDA-MB-361 Luminal B65,66 High PRLr48 MCF7 Luminal A/B4,65,66 Moderate PRLr48 ZR-75-1 Luminal A65,66 Moderate PRLr48 MDA-MB-175 Luminal A65,66 Moderate PRLr48 MDA-MB-453 HER2 enriched65,66 Moderate PRLr48 BT549 Triple negative65,66 Low PRLr48 SKBR3 HER2 enriched65,66 Low PRLr48 MDA-MB-157 Triple negative65,66 Low PRLr48 MDA-MB-231 Triple negative65,66 Low PRLr48

PRL also plays a role in regulating and enhancing ERα and PR expression in breast cancer cells. In fact, PRL increases ERα levels and E2 responsiveness in MCF7 breast cancer cells.67,68

In one study, PRL induced ERα transcriptional activity in the absence of estrogen binding ERα.69

The authors further went on to show that PRL activated both reporter plasmids containing estrogen response elements (EREs), as well as induced the ERα-target gene pS2, mimicking the effect of

E2 by inducing recruitment of ERα to ERE-containing promoters and resulting in the recruitment of co-activators of transcription.69 Furthermore Ormandy et al found that PRL stimulation of

MCF7 cells increases PR transcript levels and receptor binding activity in a dose-dependent manner.48 Finally, Fiorillo et al showed that the PRLr transactivation domain (TAD) was specifically necessary for regulation of ERα and PR levels in MCF7 breast cancer cells.70 Overall, these findings illustrate how there is complex interplay between these hormone pathways in breast cancer.

Multiple transgenic mouse models have been developed to uncover the role of PRL in breast cancer. First, ubiquitous overexpression of PRL in transgenic mice was found to be sufficient to induce the formation of mammary tumors by activation of the PRLr within 11-15 months.71 In contrast, in the mouse mammary tumor virus-polyoma middle tumor-antigen

(MMTV-PyMT) model, for mice deficient in PRL, tumors appeared later in the mice and grew

30% more slowly.72

Possibly the most considerable in vivo experiments to date utilize rat PRL (rPRL) expression under control of the mammary-specific rat neu-related lipocalin (NRL) promoter, in a model named NRL-PRL.69 Mice with overexpression of PRL within the mammary gland epithelium readily develop mammary neoplasias by 6-9 months and invasive carcinomas within a range of 14-21 months.73,74 While tumors from these mice can be ERα positive (+) or ERα negative

(-), the majority are ERα+, and reflect diverse histological features.73 These characteristics of the

NRL-PRL model make them useful models of aggressive ERα+ clinical disease. In fact, recent genomic analysis of tumors from this model showed that all tumors contained somatic alterations or amplifications of KRAS that were not seen in any of the pre-neoplastic mammary glands.75 In contrast, the pre-neoplastic mammary glands displayed constitutive expression of rPRL along with increased transcripts for inflammatory cytokines. Thus, constitutive PRL signaling in the mammary glands may induce a pro-tumor microenvironment to enable the oncogenic KRAS pathway and spontaneous tumorigenesis. Adding to this proposed role for PRL cooperating with other oncogenes, the NRL-PRL model has also been used to show that along with mutations in the tumor suppressor Adenomatous polyposis coli (APC) PRL facilitates higher cancer stem cell activity dependent on Notch signaling leading to rapidly-proliferating lesions that progress to adenocarcinomas.76

In an effort to establish another in vivo ER+ breast cancer model, the researchers in Hallgeir

Rui’s laboratory have generated a prolactin-humanized mouse. This mouse model was genetically engineered to express physiological levels of human PRL in place of mouse PRL and backcrossed into the immunodeficient NOD scid gamma (NSG) strain.77 Mouse PRL acts as an antagonist at the human PRLr, and so the researchers hypothesized that the lack of human PRL activity selected against the establishment of transplanted ER+ breast cancers.77 The resulting hPRL.NSG mouse model has thus far been validated for its usefulness and improvement over the NSG model in initial tumor establishment with subsequent increased growth rates of the breast cancer xenografts. 77 The mouse model has also shown increased responsiveness to the anti-estrogen tamoxifen, as well as evidence for PRL-dependent progression of ER+ disease.77 Future studies using this model will

most likely help in determining therapeutic targets of the PRL pathway along with other growth signaling pathways.

1.5.3. In vitro studies of PRL/PRLr signaling mechanisms

Given the role of PRL in breast cancer pathogenesis as enumerated above, the PRL signaling pathway deserves investigation to uncover mechanisms and possible clinical targets for future interventions. In vitro models of breast cancer focusing on PRL signaling have been significant in this effort.

It has been through these efforts that PRL has been shown to directly interact with the peptidyl-prolyl isomerase (PPI) CypA.14,15,78 CypA, acting as a molecular switch for potentiating

PRL signaling, mediates a conformational change necessary for JAK2-PRLr binding.15,78 Further, while pharmacologic inhibition of CypA inhibits PRL signaling and decreases PRL-induced gene expression globally in T47D cells, it also inhibits MDA-MB-231 cell migration.78 Thus, PRL signaling relies on the activity of CypA, and CypA has been shown to be a potential therapeutic target in the PRL signaling cascade.

In vitro studies have also shown that PRL stimulates cytoskeletal re-organization and motility in breast cancer cells.40,41 PRL-mediated breast cancer motility relies on the serine/threonine kinase NEK3. When NEK3 was downregulated by small-interfering RNA

(SiRNA), T47D cells exhibited decreased activation of the GTPase RAC1, by way of decreased

VAV2 phosphorylation, as well as decreased cellular migration and invasion.39,40 Specifically,

PRL activation of the ERK pathway regulates NEK3 phosphorylation and activation.41 When activated, NEK3 interacts with the focal adhesion adaptor protein paxillin, such that focal adhesions are modulated for cellular movement in response to PRL.40,41

In studying PRL signaling in vitro, STAT5a has been examined for its role in enhancing proliferation of breast cancer cells by directly upregulating transcription of cytokine-inducible

SH2-containing protein (CISH) and cyclin D1 (CCND1).79–82 STAT5a transcriptional activity in breast cancer cells was first studied using luciferase-reporter assays to monitor signal transduction and gene expression, which relied on exogenous expression of the promoter sequence of a known target gene, such as CISH.83 With PRL stimulation, these assays showed upregulation of CISH by increased bioluminescence, and experimental parameters could be manipulated to show how

STAT5a regulation of CISH was affected. For example, VAV2, but not NEK3, has been shown to contribute to STAT5a signaling by use of luciferase assay; expression of VAV2 led to a PRL- inducible increase in STAT5a-mediated gene transcription that was not affected by expression of

NEK3.39 PRL stimulated STAT5a has also been shown to inhibit expression of the proto- oncogene, BCL6, whereas STAT5a knockdown effectively abrogates this finding.84 The mechanism by which PRL regulates STAT5a transcriptional activity is an active area of research.

To this point, the histone deacetylase HDAC6 has been shown to deacetylate the chromatin-modifying high mobility group N2 protein (HMGN2) such that HMGN2 can be recruited by the PRLr TAD to enable STAT5a-mediated gene expression upon PRL stimulation.79,85 Further, the chromatin linker histone H1 antagonizes STAT5a binding at DNA, which is relieved by HMGN2 upon PRL stimulation.86 In a genomics study utilizing both chromatin immunoprecipitation (ChIP)-sequencing (seq) and RNA-seq methods in MCF7 cells,

Craig et al identified PRL-inducible STAT5a binds at enhancers rather than promoters to convey specificity of PRL-regulated genes, and that these genes are also significantly enriched for cis- regulatory elements bound by HDAC6 and HMGN2.87 Further, the selective HDAC6 inhibitor,

ACY-241, blocked PRL-induced STAT5a chromatin engagement at enhancers along with a

significant proportion of PRL-stimulated gene expression.87 Lastly, this study also identified, through gene set enrichment analysis (GSEA), significant overlap of PRL-regulated genes with genes regulated by ERα, especially PRL-regulated genes with transcriptional regulatory elements co-occupied by both STAT5a and HDAC6.87 Overall, these findings illustrate that PRL-stimulated

STAT5a target genes require cofactors HDAC6 and HMGN2 for transcription.

The role the PRLr isoforms play in breast cancer has also been an area of active research.

The recently published study by Grible et al found that when the intermediate PRLr isoform

(PRLrI) is co-expressed with the wild-type long form (PRLrL) in a partially transformed breast epithelial cell line, MCF10AT, these cells underwent both in vitro and, when transplanted in mice, in vivo transformation.88 In contrast, knockdown of PRLrI in MCF7 cells decreased malignant potential in vitro.88 Interestingly, this study also uncovered an increase in KRAS signaling with increases in the PRLrI:PRLrL ratio, indicating that the potential oncogenic role of PRLrI:PRLrL heterodimers cooperate with KRAS signaling.88 This corroborates the findings in the NRL-PRL mouse model, discussed above.

1.6.STAT5a

1.6.1. STAT family

Signal transducers and activators of transcription (STATs) are latent cytosolic transcription factors (TFs) which transduce extracellular stimuli to activation of target genes in the nucleus.

Various cytokines activate STAT activity, including PRL, interferons (IFNα/β/γ), and interleukins

4 and 6 (IL-4/6), among others.89 Canonically, STAT activity is controlled by tyrosine kinases associated with cytokine receptors: once tyrosine-phosphorylated (pY), STATs dimerize and

translocate to the nucleus, where they interact with other various proteins of transcription to modulate gene expression.90,91 There are 7 known members of the STAT family: STAT1, STAT2,

STAT3, STAT4, STAT5a, STAT5b, and STAT6, and they are involved in a range of regulatory events including hematopoiesis, immunomodulation, and development.92 STAT activity, and therefore biological function, is controlled by receptor-activation and crosstalk, interacting protein partners, and nuclear-cytoplasmic shuttling.

STATs were first identified as mediators of IFN-stimulated gene transcription.93,94

Through molecular and biochemical analysis of the promoter regions of IFN-stimulated genes, it was discovered that specific DNA consensus motifs, namely gamma interferon activation site

(GAS) motifs, required binding with a TF that became known as STAT1.90,93 Identification of the rest of the STAT family subsequently followed using various techniques, one of which utilized purification of proteins bound to GAS motifs found at other gene promoter regions. For example,

STAT5 was isolated from DNA-binding sequences from the β-casein gene.93,95 The GAS motif is a palindromic core motif recognized by all STATs: TT(C/A)YNR(G/T)AA. However, sequence specificity and binding affinity, as well as different STAT protein-protein interactions ensure that

STATs each induce expression of exclusive genes in response to various cytokines.93,94

Genetic mapping of human STATs which encode the TFs shows the family organized as closely linked clusters on three chromosomes: STAT1 and STAT4 on chromosome 2; STAT3,

STAT5a, and STAT5b on chromosome 17; and STAT2 and STAT6 on chromosome 12. These clusters of STAT genes may be related to their functions, and in fact STAT5a and STAT5b have both convergent and divergent functions. The presiding theory is that a series of tandem gene duplications of an ancestral STAT gene, followed by dispersion of linked loci to different

chromosomes, gave rise to the seven STAT genes. Furthermore, the two STAT5 genes arose the most recently with a shared sequence identity of 93.6% at the cDNA level.89,92,96,97

1.6.1.1. STAT structure and isoform homology

The STAT family is highly conserved, with peptide sequence conservation scores between

85-89% (Figure 1.2). Multiple sequence alignment produced by T-coffee and analyzed for conserved residues using the Jalview software illustrates the high conservation of amino acids for the STAT family members, with the greatest difference in amino acids in the C-terminus (Figure

1.2).98–100 STAT5a and STAT5b have high sequence identity at the protein level across species: mouse Stat5a and Stat5b are approximately 95% identical, whereas T-coffee alignment of human

STAT5a and STAT5b show up to 98% conservation.98,99,101,102

Figure 1.2 Amino acid sequences of STAT family of proteins illustrate the homology of the transcription factors. Conserved Y residues and PSP motifs noted in black boxes. Multiple

sequence alignment produced using T-Coffee, and threshold for conservation (=2) set using Jalview software.

The amino acid sequences of the STAT family result in a common set of six domains, each with features that contribute to the function of the TF, as well as, importantly, a conserved tyrosine residue (Tyr, Y) in the C-terminus. As illustrated in Figure 1.3, adapted from Nicholas C., and

Lesinski G., 2011, from N-terminus to C-terminus, these domains are: 103

1. N-terminal domain: together with the Src Homology 2 (SH2) domain, mediates homo-and

hetero-dimerization of STAT monomers during activation, and mediates other protein-

protein interactions.89,104–106

2. Coiled-coil domain: functions as a nuclear localization signal and interacting domain for

other proteins, including IRF-9/p48 for STAT1; c-JUN, STLP1, and GRIM-19 for STAT3;

and SMRT for STAT5a and STAT5b.106,107

3. DNA binding domain: determines specificity of DNA association for each isoform, and is

a highly conserved domain for the STAT family.89,94,105

4. Linker domain: rich in proline (Pro), glycine (Gly), and hydrophilic residues to provide

flexibility to permit the phosphotyrosine (pY) of one STAT to bind to the SH2 domain of

the partner STAT for dimerization during activation.107

5. SH2 domain: responsible for recruitment to, and binding to, receptors at specific tyrosine-

phosphorylated sequences in the receptors; mediates homo- and heterodimerization of the

STATs by binding the pY domain of the partner STAT. STAT1, 3, 4, 5a, and 5b form

homodimers. Depending on the activating ligand, STAT1 can partner with STAT2 or

STAT3 to form heterodimers.108 This domain is highly conserved, and the target for most

STAT inhibitors.89,105,106,109

6. Conserved Y residue: key tyrosine residues on the STAT family are the targets of

phosphorylation by the JAK tyrosine kinases, SRC family kinases, and other tyrosine

kinases. These tyrosine residues are seen in Figure 1.3: Y701 for STAT1, Y690 for

STAT2, Y705 for STAT3, Y693 for STAT4, Y694 for STAT5A, Y699 for STAT5B and

Y641 for STAT6.110 Canonically, phosphorylation of tyrosine residues is understood to be

the activation event necessary for STAT activity and dimerization, and that inactive STATs

are found only in the cytoplasm.105–107 However recent evidence suggests in a non-

canonical model that non-phosphotyrosine-STATs (or unphosphorylated-Y/upY-STATs)

have transcriptional activity which will be discussed further below.105

7. Transactivation domain (TAD): the most carboxy-terminal region of the STATs, and it is

the most diversified peptide sequence between the STATs.107,109 Except for STAT1 and

STAT6, the TAD of all STAT contains at least one conserved phospho-serine (pS) residue,

that, while less understood for its regulation of STAT activity, is appreciated for

influencing STAT activity in some way, possibly by mediating protein-protein

interactions.107 The regulation and activity of the two C-terminal phospho-serine sites in

STAT5a are the focus of this thesis, and thus the regulation of pS-STAT will be discussed

at length below.

Figure 1.3. Structural homology of the STAT family and domain functions. Each isoform length is indicated by amino acids (aa), p=phosphorylation site, Y=conserved tyrosine residue, and S=conserved serine residues. Adapted from (Nicholas C., and Lesinski G., 2011).103

1.6.2. Sites of involvement and homeostatic biological functions

Each STAT family member is differentially activated by specific extracellular ligands, such as cytokines, hormones, or growth factors. This differential activation and downstream signaling lends each STAT its biological role.111 These roles were originally defined by tissue distribution, in vitro studies, and the phenotypes that arose in mouse models lacking their expression.

As stated above, STATs were identified as mediators of IFN-stimulated gene transcription, which are important in immune response. However, although many of their roles are in

hematologic and immunologic development and response, STAT3, STAT5a, and STAT5b have roles in other sites of the body.108 STAT1 and STAT2 chiefly mediate the effects of IFNs; mice that lack STAT1 or STAT2 are prone to microbial infections as well as viral infections.112,113

Targeted loss of STAT1 in natural killer cells (NK cells), impairs NK cytolytic activity, and mice harboring these NK cells more easily grow tumors.114 Gene-targeted mice that are null for STAT3 expression are embryonic lethal, however specific tissue targeting for STAT3 knockout has shown that STAT3 has roles in skin, the thymus, liver, and neurons, as well as in immunological development.108 As discussed below, STAT3 also antagonizes STAT5 action in mammary development.115,116 Loss of STAT5a in mice leads to failure of breast tissue development, which is covered in depth in Section 1.6.4.1. STAT5a and STAT5b have roles in red blood cell production: embryos that are Stat5a-/-5b-/- are severely anemic, due to a decrease in erythroid progenitors ultimately due to a loss of responsiveness to erythropoietin.117,118 STAT5a has also been shown to have a role in immune regulation, from the stimulation of peripheral T cells in response to IL-2 to repopulation of hematopoietic stem cells in bone marrow.119,120 Loss of STAT4 or STAT6 in mice leads to impaired TH1 differentiation, owing to loss of response to IL-12

(STAT4) or IL-4 (STAT6), ultimately leading to impaired immunologic response.108

1.6.3. Regulation of STAT5a signaling

STATs are activated by cytokines, hormones, and growth factors interacting with their respective cell surface receptors. Phosphorylation of tyrosine residues, subsequent dimerization, and translocation to the nucleus all precede canonical STAT transcriptional activity. STAT activity is influenced by modifying protein-protein interactions, expression levels of STATs, and epigenetic modifications that influence STAT chromatin binding.

1.6.3.1. Cytokine and hormone receptors upstream of STAT5a

In normal mammary epithelial cells, STAT5a activation is mediated predominantly by

PRL/PRLr and JAK2.28 However, activation of STAT5a through JAK2-tyorsine phosphorylation has also been shown downstream of the GH receptor, IL-3 receptor, and EPO receptor.121 Further, binding of the insulin-like growth factor (IGF) to the IGF receptor, E2 to the ERα, or epidermal growth factor (EGF) or TGFα to the EGF receptor (EGFR), activates cellular-Src (c-Src) tyrosine phosphorylation of STAT5a.122 Additionally, ErbB4 has been shown to be required for STAT5 activation during lactation.122,123

1.6.3.2. Proteins that modulate STAT5a phosphorylation and dephosphorylation

The protein tyrosine phosphatase (PTP) SH2 domain containing protein tyrosine phosphatase (SHP-2), has been shown to dephosphorylate STAT5a on Y694 in IL-2 stimulated cells.121 In MCF7 and T47D breast cancer cells, knockout of PTP1B sustains pY694-STAT5a, as well as STAT5a transcriptional activity, indicating PTP1B may be responsible for dephosphorylation of STAT5a in malignant breast epithelial tissue.124

In addition to tyrosine phosphorylation, STAT5a is serine phosphorylated on residues S726 and S780, however the kinase responsible is still unclear. The mitogen activated protein kinase/extracellular signal regulated kinase (MAPK/ERK) cascade is often activated concurrently to JAK/STAT signaling by cytokines and hormones, including PRL/PRLr signaling.125 These kinases, which are serine/threonine kinases, have been implicated in phosphorylation of STAT5a

S780.32 However, study findings on ERK1/2-STAT5a interactions, as well as the requirement for active ERK1/2 for pS-STAT5a are inconsistent.32,34,126,127 GH stimulated ERK1/2 is capable of phosphorylating S780-STAT5a in an in vitro kinase assay.32 However, this may not be the case in the cellular environment. Whereas ERK1/2 and STAT5a have been shown to interact by co-

immunoprecipitation, the kinetics of this interaction in response to cytokine stimulation varies. In one study of CHO cells (PRLr/GHr-null) stably expressing GHr, preformed complexes of ERK1/2-

STAT5a dissociated with GH stimulation.32 In a similar study, upon GH stimulation, the association of ERK1/2 and pY- and pS-STAT5a was increased in CHO cells.34 Additionally, studies using pharmacologic inhibitors of the MAPK pathway in cell culture are inconsistent. For example, in Nb2 lymphocytes, the inhibitor PD98059 had no effect on IL-2 induced pS-STAT5a nor its ability to bind DNA.126 In another study of Nb2 lymphocytes, PD98059 had no effect on

PRL-inducible phosphorylation of S726, but decreased PRL-independent, constitutive serine phosphorylation of STAT5a.127 Finally, in an oncogene RhoA-driven model of Madin-Darby

Canine Kidney (MDCK) cells, PD98059 treatment decreased phosphorylation of S726 with no decrease in phosphorylation of S780.128 These findings indicate that ERK1/2 phosphorylation of

STAT5a may be cell-type and cytokine dependent.

1.6.3.3. Transcriptional modulators and epigenetic modifiers

Coactivators of STAT5a transcriptional activity include general transcription factory machinery such as CREB-binding protein (CBP) and p300.121 Our laboratory and others have shown that the PRLr itself translocates to the nucleus upon PRL stimulation and acts as a coactivator of STAT5a.70,85 Further work in our laboratory determined that HMGN2, a chromatin modifying protein, facilitates STAT5a recruitment to the promoter region of the CISH gene, promoting STAT5a-mediated transcription.79,85 In response to PRL, HMGN2 facilitates the displacement of the linker histone H1 at STAT5a GAS consensus sequences in the CISH promoter.86 H1 loss is necessary to allow STAT5a binding and transcriptional activation, and is therefore a critical negative regulator of STAT5a-mediated transcription.86 Thus, chromatin remodeling is an important regulator of STAT5a transcriptional activity.

Epigenetic modifications include DNA methylation and acetylation which play important roles in regulating transcription. The histone-lysine N-methytransferase enzyme enhancer of zeste homolog 2 (EZH2) is responsible for the trimethylation of lysine 27 of histone H3 (H3K27me3).129

Loss of EZH2 in mammary stem cells was shown to result in precocious differentiation of alveolar epithelium.129 This also resulted in increased STAT5 occupancy of mammary-specific STAT5 target genes and their subsequent activation.129 These findings led the authors to propose that the presence of EZH2 enables controlled recruitment of STAT5 and transcriptional co-activators to genes.129

STAT5a transcriptionally upregulates the expression of suppressors of cytokine signaling

(SOCs) proteins and CISH. These proteins act as negative feedback on STAT5a activity.82,121

These proteins can bind directly to tyrosine kinases to deactivate them, or block docking sites on cytokine receptors to inhibit STAT5 activation.121

1.6.3.4. STAT5a and the nuclear receptor family

As shown in models of both normal and malignant breast epithelium, STAT5 physically interacts with PR, glucocorticoid receptor (GR), and ERα, with subsequent effects on transcriptional activity.121,130–132 In cells stimulated with both PRL and the glucocorticoid dexamethasone, STAT5a physically interacts with GR, and STAT5a transcriptional activity is enhanced.121,133,134 This indicates that GR acts as a coactivator for STAT5a gene transcription.

STAT5a also acts as a coactivator in the regulation of some PR genes.135 Conversely, direct physical interaction of ERα and STAT5 has been shown to repress STAT5 activity at some promoters, but enhance STAT5 activity at others, indicating promoter-dependent cross-talk between STAT5 and ERα.131,132 In fact, ERα and STAT5 associate at the promoter of the PRLr in

a PRL-dependent manner to increase expression of PRLr.136 Finally, PRL-induced transcription of

ERα relies on STAT5a binding to GAS elements in the ERα promoter.137

PRL-activated pY-STAT5 has also been shown to cooperatively act with androgen receptor

(AR) in breast cancer cells at the promoter for the carboxypeptidase-D gene.138 While AR is expressed in 70-90% of invasive breast cancers, its role in breast cancer progression is unknown.138

Cross-talk between PRL-STAT5 and AR is much better studied and understood in prostate cancer, where STAT5 has been shown to promote AR signaling and physically interact with AR.139

1.6.4. Canonical pathway of JAK/STAT

1.6.4.1. In normal mammary development

The JAK2/STAT5 pathway downstream of PRL/PRLr binding plays an essential role in establishing alveologenesis during pregnancy. Both Jak2-/- and Stat5-/- (targeting both Stat5a and

Stat5b) mice were shown to phenocopy Prlr-/- mammary gland defects.96,97 Jak2 deletion impaired alveologenesis and pregnancy-induced branching that could not be rescued with multiple pregnancies.140 When Stat5 was conditionally deleted prior to pregnancy, epithelial proliferation and differentiation were inhibited. When Stat5 was deleted during pregnancy, premature cell death occurred, indicating that cell proliferation, differentiation, and survival require STAT5.141

When Stat5 is introduced as a transgene under control of β-lactoglobulin gene (BLG) regulatory sequences to direct it to the mammary epithelium in mice, these BLG/STAT5 mice exhibit larger alveoli in mid-pregnancy and delayed onset of involution, along with elevated levels of β-casein secretion into milk.142 This study, by Iavnilovitch et al, extended the functional observations of STAT5 that had been made in cultured cells and in gene knockout mice studies.

Not only does STAT5 act on normal mammary cell proliferation, but it also acts on milk protein gene expression (i.e. β-casein), and post-lactational involution.

The mechanism by which the STAT family regulate the developmental switch from lactation to involution is mostly by regulating survival signaling through the PI3-kinase/AKT pathway.42,115,143 In response to PRL, STAT5 binds to the consensus motifs within the Akt1 locus, initiating Akt transcription, and generating a strong cell survival signal.143 Indeed, overexpression of Stat5a does not promote aberrant proliferation of epithelial cells, but, in fact, delays mammary gland involution in transgenic mice.143 In contrast STAT3 antagonizes pro-survival STAT5 signals and promotes pro-apoptotic pathways.115

1.6.4.2. STAT5a in breast cancer

The mechanisms through which STATs promote breast cancer are actively being investigated, and while this investigation has included the roles of STAT3 and STAT5b, the focus of this thesis will be on STAT5a. Prior to the early 2000s, most of the pro-tumorigenic gene targets of STATs were identified not necessarily in models of breast cancer, but rather in other cancers such as prostate, or normal mammary epithelium, and extrapolated to breast cancer. The first indications of STAT5a activity in breast cancer came about through the observation that STAT5a is a survival factor in normal mammary epithelium, as discussed above. Evidence for functioning

STAT5a in breast cancer includes its expression in breast cancer tissues and cell lines, mouse models, and target-gene activation. Increasingly, however is evidence that pY-STAT5a is a positive prognostic indicator. This duality in STAT5a in breast cancer warrants thorough investigation.

In order to first explore the in vivo role of STAT5a in establishing mammary hyperplasia and tumor progression, Humphreys and Henninghausen used a mouse model in which neoplasia

is driven by TGFα, which is under the control of whey acidic protein (WAP) promoter for targeting expression to the mammary gland. The authors then took this model and generated WAP-TGFα mice with homozygous loss of Stat5a (Stat5aKOTGFα).144 In the WAP-TGF-α model, mammary targeted TGFα binds EGFR, delaying mammary involution and transforming mammary epithelium in mice bred ad lib after three pregnancies. In this model, during involution, STAT5a was persistently phosphorylated on Y694, stimulated by TGFα-EGFR binding. However, loss of

STAT5a increased the rate of involution in the mammary glands after pregnancy, and increased the levels of apoptotic cells, demonstrating that STAT5a is mandatory to avoid programmed cell death in mammary epithelial cells. Further, loss of STAT5a in the mammary gland delayed any appearance of EGFR-mediated hyperplasia and hypertrophy by 1.5 months and resulted in single gland involvement rather than multiple gland involvement in the TGFα transgenic mice. In this study, persistent STAT5a phosphorylation by TGFα stimulation linked STAT5a to the transformation of the mammary gland, which the authors concluded occurred with loss of cell survival regulation.144

In a similar model of mammary cancer progression initiated by expression of the Simian

Virus 40 T antigen (TAg), i.e. the WAP-TAg transgenic mouse model, 86% of the adenocarcinomas demonstrate activated STAT5a (pY-STAT5a). Ren et al set out to determine if pY-STAT5a was a driver of these cancers or if it was a bystander effect. In this model, hemizygous loss of Stat5a (Stat5a+/- WAP-TAg) significantly reduced the number of mice with palpable tumors after 7 months, reduced the size of tumors overall, and delayed tumor onset from 188 days to 208 days.145 Concordant with delayed mammary cancer progression, Stat5a+/- WAP-TAg tumors had an increased proportion of apoptotic cells. However, there was no change in the

proportion of mitotic cells, nor histologic grade, indicating that proliferation and differentiation of the tumors was not affected by loss of STAT5a.

Rather than STAT5a knockout, Iavnilovitch et al generated a transgenic mouse model in which several variants of STAT5a were expressed in the mammary gland using regulatory sequences of the BLG.146 This study set out to investigate the deregulation of STAT5 in mammary cell transformation. Mammary-directed expression of either WT-STAT5, a constitutively active form of STAT5a (Stat5ca), or a dominant negative form of STAT5a in which the TAD was mostly deleted (Stat5Δ750) in FVB/N mice resulted in mammary tumors of four main phenotypes as analyzed by histology within 8-12 months: undifferentiated carcinoma, adenocarcinoma, papillary adenocarcinoma, and micropapillary adenocarcinoma. Interestingly, of the three transgenic variants of STAT5, BLG/Stat5Δ750 mice more often developed undifferentiated carcinomas, while BLG/Stat5 or BLG/Stat5ca mice developed papillary or micropapillary adenocarcinomas most frequently. Tumors from all three variants expressed high levels of cyclin D1, however tumors from mice bearing the Stat5ca variant expressed three-fold cyclin D1 compared to the other two STAT5 variants. This study concluded that mammary cell transformation and tumor development may by independent from the transactivation of STAT5, however this conclusion relies heavily on the presumption that the Stat5Δ750 transgene functions as a dominant negative transcriptional regulator. The Stat5Δ750 transgene is assumed to be a dominant negative form of

STAT5 because it has been shown to bind canonical STAT5-DNA response elements and block transcription, however, it should be noted that this conclusion came from studies limited by the use of luciferase reporter assays with GAS motif sequences or immunoblot for milk proteins such as β-casein, so global transcriptomic effects are still unknown.142,147 With the truncation of the

TAD of STAT5 in the Stat5Δ750 variant, only one of the phospho-serine residues of interest to

this thesis, S780, is contained in the deleted sequence, and residue S726 is still present and presumptively phosphorylated. These serine residues will be discussed at length.

While these experiments in mouse models demonstrated STAT5a as an important survival factor and possible promoter of mammary cancer progression, it is important to correlate this data to human mammary cancer. One of the first studies to examine the expression of tyrosine- phosphorylated STAT5a in human breast cancer tissue samples, by Cotarla et al in 2003, utilized immunohistochemistry of 78 patient primary breast adenocarcinomas samples compared to normal non-pregnant breast tissue samples obtained from reduction mammoplasties.148

Immunohistochemistry (IHC) of these samples examined extent of nuclear localization, disregarding staining intensity. IHC determined nuclear localized STAT5a in 59 of 78 (76%) of the breast adenocarcinomas, and 38 of 78 (49%) of these tissues exhibited relatively high levels

(per the methodology: ≥51%) of nuclear staining for STAT5a.148 The phosphorylation status of

STAT5a was also examined in these samples, and all samples with nuclear STAT5a exhibited pY-

STAT5a. When the authors correlated STAT5a nuclear localization with patient clinical characteristics, a significant positive association was found between nuclear localized STAT5a and increased levels of histologic differentiation; poorly differentiated cancers were mainly

STAT5a negative.148 No relationship was found with any other clinical outcomes data, i.e., menopausal status, ER status, HER2 status, or lymph node metastases. This study concluded that

STAT5a may be an independent prognostic indicator for breast cancer patients, however limited by a small cohort, a larger study with patient outcomes data would be needed to verify this data.

Yamashita et al and Peck et al would examine this question independently with larger cohorts.149,150 Although Yamashita et al examined a cohort of 517 patients, the authors failed to differentiate 1) between STAT5a and STAT5b, and 2) between nuclear and cytoplasmic staining

of STAT5 for statistical correlations with clinical data (this may be due to their antibody choice, as they found only 18 out of 517 tumors were positive for nuclear STAT5).149 In this study, they found 33.8% of the tumors to be positive for STAT5 by IHC, and this staining was correlated to lower histologic grade, reproducing the findings of Cotarla et al.148,149 STAT5 was also found to be positively associated with ER and PR expression, and increased overall survival in ER+ disease

(but not disease-free survival).149 Further, this study attempted to determine if STAT5 could be a predictive response to endocrine therapy for metastatic breast cancer. However, as has been seen before, the cohort of this study was a limitation: overall 70 patients received endocrine therapy for metastatic breast cancer, 32 of which responded to the therapies they were given in clinic.149 Of the total cohort, 19 out of 70 (27.1%) patient’s primary tumors were STAT5 positive. Of the 19

STAT5 positive primary tumors, 13 of these patients responded to the endocrine therapies they were given, and thus 13 out 32 responders had STAT5 expressed in the primary tumor (or

40.6%).149

Peck et al attempted to overcome the shortcomings of the previous studies by distinguishing expression of STAT5a and STAT5b in clinical patient samples, in a study with five independent cohorts totaling clinical outcomes data for 686 patients.150 Citing the discrepant evidence for a role for STAT5a in promoting tumor initiation but also maintaining tumor differentiation and suppression of disease progression in established human breast cancer, these authors undertook a systematic effort to quantify STAT5a and STAT5b expression during human breast cancer progression relative to normal breast tissue.150 In these cohorts, frequent and selective loss of nuclear and total STAT5a protein was observed in invasive breast cancer and lymph node metastases, whereas expression of STAT5b remained unchanged.150 Nuclear-localized STAT5a was determined to be an independent marker of prognosis in lymph node-negative breast cancer.

This study was powerful in that this finding was validated in multiple patient cohorts of therapy- naïve, lymph node negative disease, as well as in patients who had been treated with antiestrogen therapy. This study also examined two independent cohorts of patients treated with antiestrogen therapy alone and found that low levels of nuclear-STAT5a associated with elevated risk of failure of antiestrogen treatment.150 A limitation of the study was the use of archival cohorts, and that the high and low levels of STAT5a differed between these cohorts for prognostic outcomes and antiestrogen therapy-response. Unfortunately, to date, no follow up studies with randomized prospective trial material have been reported.

1.6.5. Non-canonical activity of STATs without tyrosine phosphorylation

Unphosphorylated STATs (upY-STATs) have long been considered the dormant, non- functioning form of STATs residing in the cytoplasm. Recently, the canonical dogma of STAT activation exclusively through tyrosine-phosphorylation has been challenged as reports have demonstrated STAT actions in non-tyrosine phosphorylated (or “latent”) states. In fact, upY-

STATs are present in the nucleus regardless of cytokine stimulation.151–153 Further, at this time, upY-STAT 1, 3, and 5a have all been reported to have some transcriptional activity distinct from their pY-states.111,154–159

STAT1 is a principal target of type I and type II IFNs implicated in malignancy as a tumor suppressor, and is tyrosine-phosphorylated at residue 701.111 Evidence supporting STAT1 functioning as a transcription factor in the absence of tyrosine phosphorylation comes from several lines of research. First, in a rescue of STAT1-null cells with STAT1 or the tyrosine-to- phenylalanine mutant Y701F, both rescues were sensitive to tumor necrosis factor (TNF)-induced apoptosis, while the STAT1-null cells were resistant. 154 The authors from this study concluded

that STAT1 is required for TNF-mediated apoptosis, and appears to act in the absence of phosphorylation of Y701.154 In another study, upY-STAT1 and interferon regulatory factor 1

(IRF-1) were found to bind to partially overlapping interferon consensus sequence 2 (ICS-2) and

GAS sites in the LMP2 promoter to activate transcription.155 This study used DNA microarrays and revealed that basal expression of many genes was regulated similarly by wild-type unphosphorylated STAT1 (that is, in untreated cells) and mutant Y701F STAT1.155

Activation of STAT3 has been defined by the phosphorylation of tyrosine residue Y705.

However, unphosphorylated STAT3 has been shown to activate gene expression important to oncogenesis, as well as mediate the antiviral and antiproliferative actions of cytokines.111,158,160

In fact, STAT3 serves two distinct roles in cytokine-dependent transcription: during the acute response through the actions of phosphorylated STAT3 dimers, and secondarily as part of the complete response through actions of increased amounts of upY-STAT3.160 Unphosphorylated

STAT3 has been shown to directly interact with other transcriptional co-activators, and has been said to form a “novel transcription factor” that might recognize half of a GAS element plus a

DNA element which the non-STAT partner can bind.111,160 There is also evidence that unphosphorylated STAT3 suppresses tumor formation by influencing heterochromatin dynamics through its association with heterochromatin protein 1 (HP1).159 These data all indicate that upY-

STAT3 has noncanonical functions influencing transcription through a variety of mechanisms.

Although there have only been two studies to date on upY-STAT5a activity, the genome- wide approach taken by Park et al in 2016 supports the idea that upY-STAT5 may act as a partial antagonist of the biological activity of pY-STAT5.157,161,162 Park et al studied STAT5 in a hematopoietic stem cell line that undergoes megakaryocytic differentiation in the presence of thrombopoietin (TPO).161 Prior to TPO treatment, the cells contained nuclear upY-STAT5, with

nuclear accumulation of phospho-tyrosine STAT5 after TPO treatment. ChIP-seq revealed

STAT5 redistribution within the cell genome with TPO treatment and three distinct binding site clusters were identified that represented exclusive upY-STAT5 binding, exclusive pY-STAT5 binding, or binding sites for STAT5 in both unstimulated and TPO-treated cells.161 Further, the authors identified binding cites for CCCTC-binding factor (CTCF) close in proximity to two- thirds of the sites bound by upY-STAT5, and the vast majority of upY-STAT5 binding close to

CTCF was lost with TPO treatment. These regions for upY-STAT5 and CTCF binding were highly enriched for genes involved in megakaryocyte differentiation and platelet development, which were repressed in the presence of upY-STAT5.161 Overall, this data was found to support the concept that upY-STAT5 suppresses transcriptional programming required for megakaryocytic differentiation, while TPO/pY-STAT5 redistribution to promoters with GAS sequences allows for transcription of genes involved in survival of differentiating cells.

The study by Hu et al (2013), found that, in a similar manner to upY-STAT3, upY-STAT5a contributes to heterochromatin formation and maintenance suppressing tumorigenesis by interacting with HP1 in colon cancer cells.157,159 However, these authors concluded that upY-

STAT5a has an exclusive role in heterochromatin formation and therefore transcriptional repression.157 The concept that HP1 interaction with STAT5a is either an alternative mode or related to the biological activity of upY-STAT5a in other cell types was not addressed and remains an open question.

1.6.6. Serine phosphorylation of STATs

The phosphorylation of STAT proteins on serine residues may occur in a manner that is cell type and stimulus dependent. STAT1, STAT3, and STAT4 all include a C-terminal Pro‐Met‐

Ser‐Pro (PMSP) serine phosphorylation motif, whereas STAT5a and STAT5b contain a Pro-Ser-

Pro (PSP) sequence.163

STAT1 is phosphorylated on serine residue 727 (S727) and has been described as playing an important role in some constitutive functions of STAT1.111 STAT1 serine phosphorylation has been found to be independent of tyrosine phosphorylation, and is phosphorylated in response to ultraviolet light (UV), IL-1, or TNF.111,164 Notably, transcriptional activation of Fas and FasL in cardiac myocytes has been found to be dependent on STAT1 S727 phosphorylation but not

Y701.165 In another example, when S727 was mutated to alanine (A, S727A), basal transcription of caspase, GBP1, TAP1, and IFP53 genes we re all effected.111,154 Thus, S727 has been determined to be essential for effective STAT1 transcriptional activity for certain genes.

Like STAT1, STAT3 is phosphorylated on residue S727, and phosphorylation of S727 is independent of phosphorylation of Y705.111 It has been shown to be phosphorylated in response to growth factors, and as necessary for maximal transcriptional activity.111,163 However, this requirement for phosphorylation of S727 on STAT3 may be promoter and cell-type dependent. In one study, activation of the IRF-1 promoter was found dependent on S727.166 Conversely, mutation of S727 to the phospho-deficient alanine (S727A) was not found to have any effect on activation of the haptoglobin acute phase promoter compared to WT-STAT3.167 STAT3 nuclear import has been found to be both serine and tyrosine phosphorylation independent.163,168 There is some evidence that S727 phosphorylation inhibits tyrosine phosphorylation or increases tyrosine dephosphorylation, however the relationship is still unclear.163

Human STAT5a and STAT5b contain serine residues in the highly divergent TAD located in conserved PSP motif at positions 726 (STAT5a) and 730 (STAT5b), as seen in Figure 1.2 and

Figure 1.3.127,169 Further, STAT5a has a unique serine residue, S780, within a Leu-Ser-Pro (LSP)

motif found across species including mouse, rat, and pig.169,170 It should be noted that previous publications regarding human STAT5a used the amino acid sequence numbering for mouse Stat5a, whereas this work used the amino acid sequence for human STAT5a obtained from the current

National Center for Biotechnology Information (NCBI) genome assembly and protein database.107,171,172 The current human STAT5a peptide sequence gives the serine residues of interest the positions 726 and 780, while the mouse Stat5a peptide sequence gives the serine residues positions 725 and 779. Thus, some labelling from the literature uses S725 or S779 for

S726 and S780, respectively. Further, studies showing the phenotypic effects of either S726 or

S780 use models in which the residues are mutated to the phospho-deficient amino acid alanine

(A).

Expression of STAT5a phosphorylated at residues S726 and S780 has been shown in human immune cells and mouse tissues. Phosphorylation of S726 in response to IL-2 was first shown in human T lymphocytes by Kirken et al in 2000, and Nagy et al (2002) replicated this finding.173,174 Phosphorylation of both S725 and S779 in response to IL-3 was shown in the mouse hematopoietic pro-B cell, Ba/F3 cell line.169 Constitutive phosphorylation of S726 was shown in both human chronic myeloid leukemia (CML) and human acute myeloid leukemia (AML) cell lines, whereas constitutive phosphorylation of S780 was restricted to CML cell lines.169 In two separate studies, phosphorylation of S779 and S725 were examined throughout mouse mammary gland development and found to increase throughout pregnancy and lactation, with no expression at involution. 170,175 Evidence for expression of pS779 in the mouse spleen and heart has also been shown.170

PRL-induced serine phosphorylation of STAT5a is less well established. Whereas Kirken et al reported PRL-induced phosphorylation of endogenous STAT5a on S725 in rat Nb2

lymphocytes, Yamashita et al reported in two different studies that pS725 in transfected COS-7 cells was constitutive.127,175,176 Further, Yamashita et al also showed pS779 was constitutive using the same cell line.175 On the other hand, Beuvink et al found that pS779 was induced by PRL in transfected COS-7 cells.170 Each of these reports utilized either audioradiographic phospho-amino analysis, western blotting, or a combination of the two methods. These discrepancies in PRL- inducible phosphorylation highlight the need for understanding cell type-specific signaling events.

1.6.6.1. Hematopoietic models

While there have been studies on the STAT5a serine residues in mammary cells regarding

STAT5a transcriptional activity, the phenotypic effects of mutations of the STAT5a serine residues, either alone or together, have mainly been shown in hematopoietic models.

The study of STAT5a serine residues in hematopoietic systems came about because constitutive activation of STAT5a has been shown to be directly involved with oncogenic transformation, however activating mutations of STAT5a have not been reported. Thus, in studying the function of STAT5a serine residues in leukemogenic transformation, Friedbichler et al first determined that STAT5a was still biochemically functional with serine-to-alanine S725A or S779A mutations; in these STAT5a mutants tyrosine phosphorylation was stimulated with EPO treatment, and these STAT5a mutants exhibited no significant differences in DNA binding properties measured by electrophoretic mobility shift assay (EMSA) using the β-casein response element.169 Further, although STAT5a serine phosphorylation is not essential for mast cell and T- cell proliferation, constitutively active STAT5a confers factor-independent survival of IL-3 dependent pro-B cells (Ba/F3 cells) only in the presence of S725 or S779, showing that the serine residues are necessary for oncogenic transformation.169 Lastly, this study showed that loss of serine phosphorylation delayed disease onset in mice transplanted with leukemic cells expressing serine

mutants of constitutively active STAT5a. Whereas mice transplanted with cells expressing constitutively active STAT5a succumbed to leukemia within 10 weeks, mice with single serine

STAT5a mutants developed signs of disease after 4 months, and throughout the period of analysis

(40 weeks), the double point mutants (S725A + S779A) remained disease-free.169

In CML and AML, STAT5 signaling is often activated downstream of oncogenic tyrosine kinases such as the BCR-ABL oncogene, a constitutively active form of the Abelson (ABL) tyrosine kinase capable of transforming hematopoietic cells.177 Berger et al therefore studied the expression of STAT5a serine mutants S725A-, S779A-, and a double point mutant (termed SASA), in leukemic cell lines expressing BCR-ABL. They found that expression of the STAT5a mutant lacking both the serine residues S725 and S779 did not support transformation and induced apoptosis in the oncogenic cell lines.178 Further, expression of STAT5a lacking either S725 or

S779 alone, or both (the SASA mutant) suppressed the leukemic potential of BCR-ABL cells in vivo. Interestingly, single mutation of S779A prolonged disease latency in mice transplanted with leukemic BCR-ABL cell lines compared to S725A- or WT-STAT5a. Additionally, survival was further enhanced in mice transplanted with the double point mutant.178 Finally, S779A-STAT5a failed to accumulate in the nucleus with BCR-ABL activation, indicating a role for this residue in

STAT5a nuclear translocation (Figure 1.4).178

Figure 1.4 Described roles for pY694-, pS726-, and pS780-STAT5a in regulating STAT5a activity. In breast tissue, phosphorylation of residue Y694 has been ascribed the major canonical role for regulating STAT5a activity, while phosphorylation of the serine residues has been described to have modulatory effects on pY694-STAT5a activity. However, phosphorylation of S726 and S780 is implicated in BCR-ABL leukemogenesis.

1.6.6.2. Normal breast development

Like tyrosine phosphorylation, STAT5a serine phosphorylation increases throughout pregnancy and lactation, however there is a sharp decline of pS- but a slower decline of pY-

STAT5a at involution as examined in mouse mammary gland tissue. 170,175 During pregnancy, levels of circulating PRL also increase, however milk production remains suppressed until late pregnancy. Only at parturition, when circulating progesterone decreases and glucocorticoid levels begin to rise does lactation occur. In fact, GR enhances STAT5 activation, prolonging STAT5a

DNA binding, and phosphorylation of Y694.133 Accordingly, glucocorticoid acts synergistically with PRL-stimulated STAT5a at the promoter for β-casein, a highly abundant protein in milk.134

PRL-inducible reporter gene constructs for β-casein have been heavily relied on to understand the potential functions of STAT5a serine phosphorylation with PRL treatment in normal and malignant breast tissue.127,170,175

Use of similar methodology across laboratories, in which COS-7 cells were transfected with β-casein luciferase reporter gene, the PRL receptor, and S725A-, S779-, or S725+S779-

STAT5a mutants have yielded different results. First, in 1998, the laboratory of Hallgeir Rui showed that the S725A- mutant had no significant change in PRL-inducible β-casein gene expression compared to WT-STAT5a.127 Beuvink et al expanded this nonsignificant (yet significant) finding to include the S725A-, S779-, or S725+S779-STAT5a mutants in 2000.170

Then, in 2001, Rui’s laboratory published another paper in which the S725A- mutant had increased

PRL-induced β-casein gene transcription, and the S725A+S779A mutant had further increased

PRL-induced β-casein gene transcription compared to WT-STAT5a in both COS-7 and MCF-7 cells.175 This finding lent itself to their hypothesis that the combined mutation led to a gain of function for STAT5a transcriptional activity at the β-casein gene. That is, although they could not replicate their earlier findings or the findings of other laboratories, the new observations indicated that functional cooperativity exists between the two serine phosphorylation sites which negatively affect the transcriptional activity of STAT5a.127,169,170,175

Beuvink et al and the work published out of Hallgeir Rui’s laboratory did agree that

STAT5a mutated at both S725 and S779 had longer DNA binding times measured by EMSA up to 30 hours and 24 hours of PRL stimulation, respectively.170,175 Beuvink et al also found that the

S725A mutant decreased Y694 dephosphorylation, which they concluded allowed STAT5a to bind to DNA longer. Together, the findings from Beuvink and Rui suggested serine phosphorylation suppresses STAT5a binding to the GAS element of the β-casein promoter. Finally, Rui’s laboratory showed that costimulation of the GR reversed the inhibitory effect of STAT5a serine phosphorylation.175 Thus, glucocorticoid could act like a time release for STAT5a activity at the beginning of lactation.

Utilization of the β-casein reporter assays, while useful, are limited in scope when discussing STAT5a actions as a transcription factor. While Park et al (2000) found no difference

in transcriptional activity of single-point STAT5a phospho-mutants compared to WT-STAT5a, expression of the double mutant, S725A+S779A-STAT5a, decreased expression of one reporter construct (ntcp) but increased expression of another (β-casein) in the presence of PRL in COS-1 cells.179 Additionally, in GH stimulated HepG2 cells or COS-1 cells expressing the

S725A+S779A-STAT5a and the ntcp reporter gene construct, transcription of the ntcp reporter gene was decreased in the HepG2 cells, but unaffected in the COS-1 cells compared to WT-

STAT5a.179 Thus, serine phosphorylation affects STAT5a transcriptional activity in both an enhancer/promoter context as well as a cell-type context, a concept that was not revealed when only the β-casein promoter was examined previously.

1.6.6.3. Mammary cancer

Other than occasionally using MCF7 or T47D luminal breast cancer cell lines with exogenous over-expression of STAT5a serine mutants and examining transcription of a reporter gene construct, no phenotypic studies have been performed to this date to understand how the serine residues affect STAT5a function.

However, two naturally occurring dominant-negative, C-terminally truncated forms of

STAT5a, STAT5aΔ740 and STAT5aΔ713 have been studied in T47D and MCF7 cells.130 Mutant

STAT5aΔ740 lacks residue S779, but retains S725, while STAT5aΔ713 lacks both serine phospho-sites S725 and S779. Both dominant-negative forms of STAT5a retain the phosphotyrosine 694. Yamashita et al (2003) set out to determine the impact these dominant-negative forms of STAT5a have on ER activity, and found that overexpression of STAT5aΔ740 completely blocked transcriptional activity of endogenous ER in T47D and MCF7 cells, and reduced endogenous PR expression, an ER transcriptional target, in T47D cells.130 Further, dominant- negative STAT5a induced apoptosis through the caspase 3-mediated pathway in T47D cells, but

did not inhibit the growth of MCF7 cells, which are deficient in caspase 3.130,180 The authors concluded that apoptosis of T47D cells may have occurred as an effect of combined inhibition of the STAT5a and ER signaling pathways. These findings indicate that loss of the C-terminal TAD containing both phospho-serine sites of STAT5a may have regulatory functions in ER transcriptional activity and apoptosis in ER+ breast cancer cells.

1.7. Goal of this work and hypothesis

The overall goal of this thesis is to investigate the functional role of non-tyrosine phosphorylated STAT5a as well as serine phosphorylation of STAT5a in luminal breast cancer.

Luminal breast cancer cell lines are used with the aim to model the functional significance of phosphorylation of residues S726 and S780 on upY-STAT5a, individually and together. The central hypothesis is that non-canonical activation of STAT5a by serine phosphorylation in response to PRL regulates STAT5a nuclear activity and modulates transcriptional programming to contribute to mammary oncogenesis. This hypothesis will be tested by the studies in the specific aims that follow:

Aim 1: Characterize the functional significance of STAT5a serine phosphorylation in PRL signaling in breast cancer in vitro. Knockdown/rescue mutagenesis targeting STAT5a in luminal breast cancer cell line (MCF7), is studied and related to PRL signaling, phenotypic hallmarks breast cancer, and transcriptional activity.

Aim 2: Determine the global transcriptomic effects of loss of each phospho-site (p-site) of

STAT5a in luminal breast cancer and relate this to patient datasets. Differential gene expression in MCF7 cells expressing the STAT5a mutants generated in Aim 1 will be determined using RNA- sequencing. We hypothesize that loss of each p-site will differentially affect PRL-induced gene programs. We will compare this data to publish datasets for patient gene expression, as well as patient derived xenografts (PDX), to determine if expression of these STAT5a mutants signals intrinsic subtype-switching.

Aim 3: Determine the mechanism by which PRL regulates serine phosphorylation of STAT5a activity in vitro. We hypothesize that ERK1/2 mediates phosphorylation of S726- and S780-

STAT5a and investigate this by co-immunoprecipitating STAT5a with ERK1/2 and using pharmacologic inhibition of ERK1/2. These methods will be validated in our cell models if phospho-deficient STAT5a is able to interact with ERK1/2 and related to the assays in Aim 1, with the goal to identify rational therapeutic targets through the PRL/PRLr signaling pathway.

2. Chapter 2: Materials and Methods

2.1 Materials

2.1.2 Chemicals

Table 2.1. Chemicals used in the pursuit of this dissertation.

Chemical Supplier Agar Thermo Fisher Scientific (Waltham, MA) Agarose EMD-Millipore (Burlington, MA) Ammonium persulfate Sigma-Aldrich (St. Louis, MO) Ampicillin/carbenicillin Gemini Bio-Products (West Sacramento, CA) β-mercaptoethanol Bio-rad (Hercules, CA) Bovine serum albumin, Fraction V Sigma-Aldrich (St. Louis, MO) DMSO sterile-filtered, hybridoma Sigma-Aldrich (St. Louis, MO) tested (for drug studies) DNase I Promega (Madison, WI) EDTA Quality Biological (Gaithersburg, MD) EGTA Quality Biological (Gaithersburg, MD) Ethanol, 200 proof VWR (Radnor, PA) HEPES Thermo Fisher Scientific (Waltham, MA) Isoproponal WorldWide Medical (Bristol, PA) LB broth, ready pack Thermo Fisher Scientific (Waltham, MA) Methanol WorldWide Medical (Bristol, PA) Phosphatase Inhibitor Cocktail Set II Sigma-Aldrich (St. Louis, MO) Polybrene Sigma-Aldrich (St. Louis, MO) Protease Inhibitor Cocktail Sigma-Aldrich (St. Louis, MO) (cOmplete) Puromycin Gemini Bio-Products (West Sacramento, CA) SDS, 10% (w/v) Bio-Rad (Hercules, CA) Sodium chloride Sigma-Aldrich (St. Louis, MO) Sodium hydroxide Thermo Fisher Scientific (Waltham, MA) TEMED Bio-rad (Hercules, CA) Tris base Thermo Fisher Scientific (Waltham, MA) Triton-X Sigma-Aldrich (St. Louis, MO) Tween-20 Sigma-Aldrich (St. Louis, MO) Zeocin Thermo Fisher Scientific (Waltham, MA)

2.1.3 Buffers, Media, Stock solutions

Table 2.2. Buffer Recipes and Suppliers.

Buffer Recipe Supplier DPBS Sigma-Aldrich (St. Louis, MO) Nuclear Cytoplasmic 20 mM HEPES (pH 7.4) fractionation buffer 10 mM KCl 2 mM MgCl2 1 mM EDTA 1 mM EGTA Fresh before each use: 1:1000 1 mM DTT 1% (v/v) Phosphatase Inhibitor 1x Protease Inhibitor Co-immunoprecipitation Tris-HCl pH 7.4 buffer 150 mM NaCl 1 mM EDTA 0.1% (v/v) Triton X-100 5% (v/v) glycerol 1% (v/v) Phosphatase Inhibitor Cocktail 2 1x Protease Inhibitor 30% Acrylamide/bis (29:1) solution Western Blot Resolving 1.5 M Tris-HCl, pH 6.8 Gel buffer Western Blot Stacking 0.5 M Tris-HCl, pH 8.8 Gel buffer Bio-rad (Hercules, 10% (w/v) SDS solution CA) TEMED Western Blot Running 1X Tris/Glycine/SDS Buffer Western Blot Transfer 1X Turbo-blot Transfer buffer Buffer 20% (v/v) 200 proof Ethanol Western Blot Stripping 1.5% glycine (w/v) Abcam protocol Buffer 0.1% SDS (w/v) (Cambridge, UK) 1% Tween 20 (v/v) pH 2.2 4x Laemmli Sample Bio-rad (Hercules, Buffer CA) 6x DNA Gel Loading NEB (Ipswich,

Buffer MA) 1x TBE LB Medium LB agar

2.1.4 Kits, Reagents, and Enzymes

Table 2.3 Kits and Reagents.

Reagent Supplier QIAprep Spin Miniprep kit Qiagen (Hilden, Germany) Endofree Plasmid Maxi kit Qiagen (Hilden, Germany) Purelink RNA Mini kit (cat# 12183020); Thermo Fisher Scientific (Waltham, MA) Purelink DNase set (cat# 12185010) iScript cDNA synthesis kit Bio-rad (Hercules, CA) SYBR Green PCR master mix Thermo Fisher Scientific (Waltham, MA) Phusion Site-directed mutagenesis kit Thermo Fisher Scientific (Waltham, MA) Q5 Site-directed mutagenesis kit NEB (Ipswich, MA) Kaleidoscope protein ladder Bio-rad (Hercules, CA) Mini-protean TGX gels Bio-rad (Hercules, CA) Immobilon Forte Western HRP substrate Sigma-Aldrich (St. Louis, MO) DNA ladder, 1 kb NEB (Ipswich, MA) DNA ladder, 100 bp NEB (Ipswich, MA) E. cloni competent cells Lucigen (Middleton, WI)

2.1.5 Prolactin

Human recombinant prolactin (PRL) was a gift from Dr. Anthony Kossiakoff (University of

Chicago, Chicago, IL). Aliquots of PRL were stored at -80°C.

2.1.6 Mammalian Cell Lines and tissue culture

Table 2.4 Human cell lines used for in vitro work.

Cell line Description Source MCF-7 Low-invasive human breast cancer cell line; American Type Culture ER+/PR+/Her2- (Luminal A/B) Collection (ATCC; Manassas, VA)

T-47D Low-invasive human breast cancer cell line; American Type Culture ER+/PR+/Her2- (Luminal A) Collection (ATCC; Manassas, VA) HEK293T Human embryonic kidney cell line containing American Type Culture temperature-sensitive SV40 T-antigen for virus Collection (ATCC; Manassas, production VA)

2.1.7 Tissue culture reagents and media

Table 2.5 Tissue Culture Reagents.

Reagent Supplier Bovine serum albumin Thermo Fisher Scientific (Waltham, MA) Gibco DMEM Thermo Fisher Scientific (Waltham, MA) Gibco DMEM, phenol free Thermo Fisher Scientific (Waltham, MA) Corning DMSO (for general tissue Thermo Fisher Scientific (Waltham, MA) culture, ie cryopreservation) DPBS Thermo Fisher Scientific (Waltham, MA) Fetal bovine serum (FBS), heat Thermo Fisher Scientific (Waltham, MA) inactivated Lipofectamine 3000 Thermo Fisher Scientific (Waltham, MA) Mycoalert Mycoplasma detection kit Lonza (Morristown, NJ) Opti-MEM Thermo Fisher Scientific (Waltham, MA) Penicillin/streptomycin Thermo Fisher Scientific (Waltham, MA) Versene Thermo Fisher Scientific (Waltham, MA)

Complete growth media for MCF-7, T-47D, and HEK239T cells: DMEM, 10% (v/v) FBS, 1%

P/S.

Serum starvation media for MCF-7 and T-47D cells: phenol-free DMEM, 0.1% (v/v) BSA.

2.1.8 Plasmids

STAT5a rescue plasmids

STAT5a cDNA was cloned into the pTracer-EF/V5-His mammalian expression vector (v.

C, Invitrogen #V88720, Thermo Fisher Scientific, Waltham, MA) such that it was in nucleotide sequence with the C-terminal V5 tag. This plasmid allows high level constitutive expression of the inserted gene under the control of the EF-1α promoter. The pTracer plasmid also has a constitutively expressed Cycle 3 GFP-Zeocin fusion gene for selection in mammalian cells. For this work, Zeocin was added to complete media for 1 to 2 weeks after viral transduction into MCF-

7 cells.

Plasmid Transfection and Viral Transduction

For transient transfections, the Lipofectamine 3000 transfection reagent protocol was used and scaled according to manufacturer’s instructions (Thermo Fisher Scientific, Waltham, MA).

Prior to transfection, adherent cells were washed with PBS at 37oC, and media was replaced with

Opti-MEM reduced-serum media (Thermo Fisher Scientific, Waltham, MA).

For stable viral transductions, viral particles were generated using HEK293T cells (ATCC).

For lentiviral production, HEK293T cells in Opti-mem media were transfected with 4.5 μg of Lenti construct DNA, 1.5 μg pMDG plasmid (containing VSV-G envelope) and 6 μg pCMVΔR8.91

(packaging construct) using Lipofectamine 3000 (Thermo Fisher Scientific, Waltham, MA) following manufacturer’s protocol. After transfection complexes were added, the cells were returned to the incubator O/N. The following morning, Opti-mem media was replaced with complete growth media (DMEM supplemented with 10% (v/v) FBS and 1% (v/v) P/S), and plates were returned to the incubator for 24-30 h. The following day, the supernatant was pipetted away

from the HEK293T cells and centrifuged at 1000 g for 5 min and filtered through a 0.45 μm filter.

This supernatant contains the viral particles for stable transduction of mammalian cell lines, and aliquots were stored at -80oC until use.

Breast cancer cell lines, namely T47D and MCF-7, were “spinfected” in 6-well plates at a cell density of 25-50% confluency. To each well of a 6-well plate, 2 ml of viral particles were mixed with 1 ml complete media, and then 8 μg/ml polybrene was added. Plates were then spinfected at 500 g for 2 h at 32oC, and plates were placed back in the tissue culture incubator O/N.

Viral media was replaced with complete growth media the next morning, and selective drugs were added 24 h after that.

Antibiotic selection

Antibiotics were used to selective for cells that were stably transduced with viral particles.

For vectors containing a puromycin resistance gene, puromycin (Gemini Bio-Products, West

Sacramento, CA) was used to select cells at 2 μg/ml for 1 week. After 1 week, puromycin was delivered to a concentration of 1 μg/ml for maintenance. Zeocin (Thermo Fisher, Waltham, MA) was used to select for cells transduced with pTracer viral particles. For MCF-7 cells, 400 μg/ml zeocin was added to media for 2 weeks, and after 2 weeks cells were maintained in 200 μg/ml zeocin. For T47D cells, 200 μg/ml zeocin was added to media for 2 weeks, and subsequently maintained in 100 μg/ml zeocin.

2.1.9 Primers

Primers for sequencing, PCR, and cloning were designed using Integrated DNA

Technologies online OligoAnalyzer Tool (IDT, Coralville, IA). Primers for qRT-PCR were

validated from literature sources. Primers for site-directed mutagenesis were designed using the

Phusion Site-Directed Mutagenesis Kit (Thermo Fisher Scientific, Waltham, MA) guidelines for primer design or the NEBase Changer v1.2.9 (NEB, Ipswich, MA) online tool for Q5 site-directed mutagenesis. All primers were synthesized by Eurofins Genomics (Louisville, KY), and lyophilized powder was resuspended in dH2O for a stock of 100 μM. Primer stocks were diluted

1:10 for working concentration of 10 μM. All primers were stored at -20°C.

Table 2.6 Primers used in this dissertation work.

Site-directed mutagenesis primers Targeted amino acid change For and Rev primers (5’-3’) Y694F* For GCTGTTGATGGATTTGTGAAACCACAG Rev TTTAGCCAGCACAGGAGTGTAGTACTTGG S726A* For CAGGCCCCCGCCCCAGCT Rev GTCCATGTACGTGGCGCTGCTGCC S780A For CTCCCGCCTCGCGCCCCCTGC Rev TCAAGACTGTCCATTGGTCGGCGTAAGAGTTC Cloning primers For and Rev primers (5’-3’) pTracer Multiple Cloning Site (MCS) T7 promoter For TAATACGACTCACTATAGGG BGH Rev TAGAAGGCACAGTCGAGG STAT5a KpnI For AAAAGGTACCATGGCGGGCTGGATC STAT5a NotI Rev AAAAGCGGCCGCTGAGAGGGAGCCTCTGGCAGA Sequencing primers STAT5a sequencing primer 1 For CCGAGAAGCACCAGAAGACCC Rev GGGTCTTCTGGTGCTTCTCGG STAT5a sequencing primer 2 For GCCTGACCAAGGAGAACC Rev GGTTCTCCTTGGTCAGGC STAT5a sequencing primer 3 For CCATCATCAGTGAGCAGCAGGC Rev GCCTGCTGCTCACTGATGATGG qRT-PCR primers Gene Target For and Rev primers (5’-3’) BCL6 For CTGCAGATGGAGCATGTTGT Rev TCTTCACGAGGAGGCTTGAT

CCND1 For CCGTCCATGCGGAAGATC Rev GAAGACCTCCTCCTCGCACTT CEBPβ For AGAACGAGCGGCTGCAGAAGA Rev CAAGTTCCGCAGGGTGCTGA CISH For AGAGGAGGATCTGCTGTGCAT Rev GGAACCCCAATACCAGCCAG ERα For GGAAGCTACTGTTTGCTCCTAACTTG Rev AGATCTCCACCATGCCCTCTAC GAPDH For CATGAGAAGTATGACAACAGCCT Rev AGTCCTTCCACGATACCAAAGT ID1 For CTGGACGAGCAGCAGGTAA Rev CTCCAACTGAAGGTCCCTGA PR For TCAGTGGGCAGATGCTGTATTT Rev GCCACATGGTAAGGCATAATGA STAT5a For GTTCAGTGTTGGCAGCAATGAGC Rev AGCACAGTAGCCGTGGCATTGT ZYX For AGGACATGGAGCATCCTCAGA Rev GCATCGGCCGCAGAGTT *: 5’ phosphorylated

2.1.10 Stat5a shRNA

Table 2.7 shRNA constructs.

Target Type Name Supplier Clone ID Sequence (5’-3’) STAT5a shRNA MISSION TRC2 Sigma- NM_003152. CCGGCCCACGTTT pLKO-puro shStat5a Aldrich 2-3761s21c1 CCCGGGATATATC 3’UTR TCGAGATATATCC CGGGAAACGTGG GTTTTTG Control shRNA MISSION TRC2 Sigma- SHC202 CCGGCAACAAGA pLKO.5-puro Non- Aldrich TGAAGAGCACCA Mammalian shRNA ACTCGAGTTGGTG NT control CTCTTCATCTTGT TGTTTTT

2.1.11 Antibodies

Table 2.8 Primary antibodies.

Antibody Host Application Dilution Supplier (product number) α-pY694- rabbit IB, IP, IHC 1:1000 Cell Signaling Technology STAT5a (9359S) α-pS726-STAT5a rabbit IB, IHC 1:1000 Abcam (ab128896) α-pS780-STAT5a rabbit IB, IHC 1:1000 Abcam (ab30649) α-STAT5a mouse IB, IF 1:1000, 1:50 Santa Cruz (sc-271542) α-STAT5a rabbit IB, IHC 1:1000 Bioss Antibodies (bs-1142R) α-STAT5 rabbit IB 1:1000 Cell Signaling Technology (94205) α-pERK1/2 rabbit IB 1:1000 Cell Signaling Technology (9101S) α-ERK1/2 rabbit IB 1:1000 Cell Signaling Technology (9102S) α -cleaved rabbit IB 1:1000 Cell Signaling Technology caspase 3 (9661) α -cleaved rabbit IB 1:1000 Cell Signaling Technology caspase 7 (9491) α-tubulin mouse IB 1:1000 Invitrogen/Thermo Fisher Scientific (32-2500) α-V5 tag mouse IB, IP, IF 1:1000, 2.5 Invitrogen/Thermo Fisher μg, 1:100 Scientific (R960-25) α-vinculin mouse IB 1:2000 Bio-rad (MCA465GA)

2.1.12 Secondary Antibodies

Table 2.9 Secondary antibodies.

Antibody Application Dilution Supplier (product number) Goat anti-rabbit IgG IB 1:2000 Cell Signaling Technology (7074) HRP Horse anti-mouse IB 1:2000- Cell Signaling Technology (7076) IgG HRP 1:5000 Alexa Fluor 568 Goat IF 1:2000 Invitrogen/Thermo Fisher anti-mouse IgG Scientific (A-11031)

2.1.13 Software & Online Tools

Table 2.10 Software and online tools.

Software/Online Tool Source Website ApE (v2.0.60) M. Wayne Davis, University of https://jorgensen.biology.utah.e Utah du/wayned/ape/ BLAST NCBI, NIH https://blast.ncbi.nlm.nih.gov/Bl ast.cgi CorelDraw CorelDraw https://www.coreldraw.com/en/ DeepVenn 2020 Tim Hulsen https://www.deepvenn.com/ Enrichr Ma'ayan Lab, Icahn School of https://maayanlab.cloud/Enrichr Medicine, Mount Sinai / Galaxy Part of the Center for https://usegalaxy.org/ Comparative Genomics and Bioinformatics at Penn State, the Department of Biology at Johns Hopkins University and the Computational Biology Program at Oregon Health & Science University Geneious Geneious https://www.geneious.com/ GraphPad Prism 7 and GraphPad https://www.graphpad.com/ 8 and 9 GSEA Broad Institute, Inc., https://www.gsea- Massachusetts Institute of msigdb.org/gsea/index.jsp Technology, and Regents of the University of California ImageJ NIH https://imagej.nih.gov/ij/ ImageQuant TL GE Healthcare Life Sciences https://www.gelifesciences.com /en/us/shop/protein- analysis/molecular-imaging-for- proteins/imaging- software/imagequant-tl-8-1-p- 00110 Incucyte Sartorius, Essen Bioscience https://www.essenbioscience.co m/en/contact/our-offices/ Integrated Genome BioViz https://bioviz.org/ Browser Jalview Geoff Barton's Group at the https://www.jalview.org/ University of Dundee

LI-COR Odyssey Fc LI-COR Biosciences https://www.licor.com/bio/odys sey-fc/ NEBase Changer NEB https://nebasechanger.neb.com/ (v1.2.9) NEBCloner (v1.4.0) NEB http://nebcloner.neb.com/#!/ OligoAnalyzer Integrated DNA Technologies https://www.idtdna.com/pages T-Coffee Comparative Bioinformatics http://tcoffee.crg.cat/ Group Bioinformatics and Genomics Programme Center for Genomic Regulation (CRG) University of Barcelona xCELLigence Acea Biosciences https://www.aceabio.com/produ cts/xcelligence-rtca/

2.2 Methods

2.2.2 Molecular Biology methods

Cloning and subcloning

DNA gel electrophoresis

DNA was analyzed on 0.8-1% agarose gels, with gel percentage chosen in order to optimize DNA fragment resolution based on size. Agarose was mixed with 1X TBE buffer and heated in a microwave to dissolve. Once the mixture cooled to ~60oC, the gel was poured into a horizontal gel tray and set for ~1 h at RT. The gel was then placed in the ENDURO Gel XL

Electrophoresis System (Labnet International Inc, Edison, NJ), and covered with 1X TBE buffer.

Samples were diluted with 6x DNA Gel Loading Buffer (NEB, Ipswich, MA), and 5-10 μl were loaded into each well. Gels were run at constant voltage (120 mV) for 45 m-1 h. DNA fragments visualized using the Licor Odyssey Fc or, for gel purification purposes, UV-light using a Kodak

Electrophoresis Documentation and Analysis System (EDAS) 290 system (Kodak, Rochester,

NY).

Gel purification

While visualizing DNA fragments under UV-light using a Kodak EDAS 290 system, DNA fragments were excised from the gel using a clean razor blade. The E.Z.N.A.® Poly-Gel DNA

Extraction Kit (Omega Bio-Tek, Norcross, Georgia) was used following manufacturer’s protocol to extract and purify DNA fragments and PCR products. Final purified DNA was eluted in 25 μl dH2O.

Restriction enzyme digestion

Restriction digests were set up as 50 μl reactions following the NEBCloner (v1.4.0; http://nebcloner.neb.com/#!/) protocol for double digests. 1 μg of DNA was incubated with 20 U of each restriction enzyme in NEB CutSmart Buffer for 15 min at 37oC.

Ligation

After digestion of vector and intended gene insert, ligation was accomplished using T4

DNA Ligase in a 5 min reaction at RT following manufacturer’s protocol (Thermo Fisher

Scientific, Waltham, MA), prior to transformation of competent cells.

Transformation of competent cells

E. cloni 10G Chemically Competent Cells (Lucigen, Middleton, WI) were thawed on ice briefly. To each vial of competent cells, 5 μl of ligation reaction was added. Reactions were tapped briefly to mix, and incubated on ice for 30 min. Then, the cells were heat-shocked in a 42oC water bath for 45 s, followed by 2 min on ice. 960 μl of RT Recovery media (antibiotic-free) was added to the cells, and tubes of cells were incubated at 37oC with shaking at 250 rpm for 1 h. 100-200 μl of these cells were added to LB plates containing appropriate antibiotic for selection, and incubated upside down O/N at 37oC.

Glycerol stocks

Individual plasmid clones were stored in glycerol stocks prepared with a 1:1 ratio of LB media containing the clone and a 50% (v/v) sterile glycerol and dH2O mixture. Glycerol stocks were stored at -80oC. For recovery of the glycerol stock, a sterile loop was inoculated by scraping the top of the glycerol stock and streaking a LB agar plate for O/N growth at 37oC.

Plasmid DNA purification (mini- and maxi-prep)

Individual clones of plasmid DNA was cultured O/N in 4 ml of LB media with appropriate antibiotic at 250 rpm at 37oC. About 2-4 ml of the O/N culture was purified using QIAprep Spin

Miniprep Kit (Qiagen, Hilden, Germany) following the manufacturer's protocol, and final DNA was eluted in ~45 μl dH2O. For larger yields of plasmid DNA, individual clones were cultured in

~250 ml of LB media with appropriate antibiotic at 250 rpm at 37oC. About 200 ml of O/N culture

was purified using the Endofree Plasmid Maxi Kit (Qiagen, Hilden, Germany) following manufacturer’s protocol. Final DNA was eluted in ~250 μl dH2O.

DNA purity and concentration

To determine DNA purity and concentration, the NanoDrop 2000 spectrophotometer

(Thermo Fisher Scientific, Waltham, MA) was utilized. DNA purity and concentration were determined at an absorbance at 260 nm. A 260/280 ratio of ~1.8, and a 260/230 ratio of ~2.0 are generally accepted as indications of DNA purity.

DNA sequencing

Once DNA was purified after mini- or maxi-preparation, aliquots were taken for sequencing using the Eurofins Genomics sequencing services. Pre-designed sequencing primers were sent with the samples, or standard sequencing primers were used to cover up to 1000 bases.

Sequencing results were analyzed against the known sequence of the gene or vector using the

Geneious Bioinformatics Software for Sequence Data Analysis (https://www.geneious.com/).

Ethanol precipitation of DNA

To recover DNA in aqueous solutions to better concentrate it for downstream applications,

DNA was ethanol precipitated. First, 3M sodium acetate (pH 5), was added to DNA to a final volume ratio of 1:10. Glycogen was then added as a carrier to a final concentration of 1 µg/µl.

100% ethanol was added to a final concentration >80% (v/v), and the DNA-ethanol solution was then stored at --20°C overnight for best recovery. The following day, the solution was centrifuged

at 13000 g for 30 min at 4°C. Supernatant was then removed, and the pellet was washed with ice- cold 75% ethanol (v/v), and centrifuged again at 13000 g for 30 min at 4°C. The supernatant was removed, and the pellet was pulse-centrifuged for 5-10 s, and residual ethanol was evaporated off by allowing the pellet to air dry for 10 min. The pellet was then resuspended in nuclease-free dH2O, and the concentration and purity were measured by NanoDrop 2000 (Thermo Fisher Scientific,

Waltham, MA).

RNA extraction

To avoid RNase contamination, all benchtop surfaces and pipettes for RNA extraction were cleaned with RNaseZap RNase Decontamination Solution (Thermo Fisher Scientific, Waltham,

MA). Cells were washed once with 37oC PBS, and then harvested in 500 μl ice-cold TRI Reagent

Solution (Thermo Fisher Scientific, Waltham, MA). Samples were pipetted up and down ~2 dozen times to lyse the cells and transferred to a 1.5 ml DNase/RNase/pyrogen-free microcentrifuge tube.

An additional 500 μl of 4oC TRI Reagent was added to the 1.5 ml tube, and samples were incubated at RT for 5 min.200 μl chloroform (Sigma-Aldrich, St. Louis, MO), was then added to each sample, and inverted several times before shaking each tube vigorously for 15 s. Samples were then centrifuged at 10,000 rpm for 20 m at 4oC. The aqueous layer was then removed (~550 μl) and transferred to a new 1.5 ml microcentrifuge tube. 550 μl of 100% isopropanol (WorldWide

Medical, Bristol, PA) per 1 ml TRI Reagent originally used was added to this tube, mixed gently, and incubated at RT for 5 min. These samples were then centrifuged at 14,000 rpm for 20 min at

4oC. The isopropanol supernatant was then discarded, and pellets were gently washed with 1 ml

o 75% ethanol in DEPC-treated H2O, before being centrifuged at 9,500 rpm for 10 min at 4 C.

Supernatant was then discarded, and pellets were allowed to air dry for 10 min, making care not to let the pellet dry completely as this would decrease solubility. RNA pellets were resuspended

in 15-20μl nuclease free (DEPC-treated) H2O. The samples of RNA were then measured for concentration and purity using the Nanodrop, and for downstream processing the 260/280 ratio was greater than 1.8. RNA was kept at -80oC until further processing.

cDNA synthesis

cDNA was synthesized using the iScript cDNA synthesis kit (Bio-rad, Hercules, CA) following manufacturer’s protocol and 1 μg of purified RNA as the template for the reaction.

Reactions were incubated in a thermal cycler using the following conditions: 5 min at 25oC, 20 min at 46oC, 1 min at 95oC, and final hold at 4oC. For qRT-PCR, 100 ng of the first strand synthesis reaction was used as a template.

Polymerase chain reaction (PCR)

Cloning PCR

For cloning PCR, NEBNext® High-Fidelity 2X PCR Master Mix (NEB, M0541) or

OneTaq® 2X Master Mix with Standard Buffer (NEB, M0482S) were used according to manufacturer’s protocol (NEB, Ipswitch, MA). In general, 1-10 ng of template DNA was added to each reaction.

Table 2.11 Reaction using NEBNext® High-Fidelity 2X PCR Master Mix (NEB, M0541):

STAT5a in plasmid, ie pTracer Step Temperature Time Cycles Initial denaturation 98oC 30 s 1 Denaturation 98oC 10 s

Annealing 72oC 30 s 25 Extension 72oC 5 m Final extension 72oC 2 m 1 4oC hold

Table 2.12 Reaction using OneTaq® 2X Master Mix with Standard Buffer (NEB, M0482S):

STAT5a in plasmid, ie pTracer Step Temperature Time Cycles Initial denaturation 94oC 30 s 1 Denaturation 94oC 10 s Annealing 70oC 60 s 30 Extension 68oC 5 m Final extension 68oC 5 m 4oC hold 1

Site-directed mutagenesis

Phusion Site-Directed Mutagenesis (Thermo Fisher Scientific, Waltham, MA)

The Phusion Site-Directed Mutagenesis kit and protocol were utilized for single-point mutations in non-GC rich areas of the STAT5a sequence (ie, for Y694F and S726A point mutations). Primers for these reactions are listed in section 2.1.8. 1 ng of plasmid template was used in each reaction, along with 0.5 μM of forward and reverse primers. Reactions were subjected to the following conditions:

Table 2.13 Reaction for Phusion Site-Directed Mutagenesis:

Y694F Step Temperature Time Cycles Initial denaturation 98oC 30 s 1 Denaturation 98oC 10 s Annealing 68oC 30 s

Extension 72oC 5 m 25 Final extension 72oC 10 m 4oC hold 1 S726A Step Temperature Time Cycles Initial denaturation 98oC 30 s 1 Denaturation 98oC 10 s Annealing 72oC 30 s 25 Extension 72oC 5 m Final extension 72oC 10 m 4oC hold 1

After performing the Phusion Site-Directed mutagenesis PCR reaction, 1 μl of FastDigest

DpnI enzyme was added directly to the reaction mixture to digest Dam-methylated parental plasmid DNA and allowed to react for 15 m at 37oC. 5 μl of this reaction product were ligated following the manufacturer’s protocol (Thermo Fisher Scientific, Waltham, MA) using T4 DNA

Ligase in a 5 min reaction at RT. Following this reaction, 5 μl of ligated product were transformed in competent cells.

Q5 Site-Directed Mutagenesis (NEB, Ipswitch, MA)

The Q5 Site-Directed Mutagenesis kit and protocol were utilized for single-point mutations in GC rich areas of the STAT5a sequence (ie, for the S780A point mutation). Primers for these reactions are listed in section 2.1.8. 1 ng of plasmid template was used in each reaction, along with

0.5 μM of forward and reverse primers. Reactions were subjected to the following conditions:

Table 2.14 Reaction for Q5 Site-Directed Mutagenesis.

S780A Step Temperature Time Cycles Initial denaturation 98oC 30 s 1 Denaturation 98oC 10 s Annealing 72oC 30 s 25 Extension 72oC 6 m Final extension 72oC 2 m 4oC hold 1

Following the Q5 Site-Directed mutagenesis PCR reaction, 1 μl of PCR product was subjected to kinase, ligase and DpnI (KLD) treatment following the Q5 NEB protocol. The reaction was incubated at RT for 5 min prior to transformation of competent cells.

Quantitative real time PCR (qRT-PCR)

For quantitative real time PCR (qRT-PCR), Power SYBR Green PCR Master Mix (Thermo

Fisher Scientific, Waltham, MA) was used. For each desired PCR amplicon, forward and reverse primers were brought to a final concentration of 900 nM, along with a total of 100 ng cDNA template in 1X Power SYBR Green Master Mix (Thermo Fisher Scientific, Waltham, MA). Each reaction was run in triplicate in a MicroAmp Optical 96-well reaction plate (Applied Biosystems,

Foster, CA) in a CFX96 thermal cycler (Bio-rad, Hercules, CA) using the following settings:

Table 2.15 qRT-PCR Reaction settings.

Step Temperature Time Cycles Initial denaturation 95oC 10 m 1 Denaturation 95oC 15 s Annealing/Extension 60oC 1 m 40

Melt curve analysis 65-95oC 0.5oC increment at 1 2-5 s/step

2.2.3 Protein methods

Protein lysis

For detection of endogenous or rescued proteins, adherent cells were first rinsed in ice-cold

PBS and scraped in ice-cold RIPA buffer containing 1x protease inhibitor cocktail and 1% (v/v) phosphatase inhibitor cocktail. Samples were sonicated in two cycles of 10 1s pulses, with 1 min rest between each cycle. Lysates were then incubated on ice for 10 min, and subsequently centrifuged at 12,000 g for 10 min at 4°C. Samples were then diluted in 4x Laemmli buffer and run on SDS-PAGE gel followed by WB analysis.

Nuclear-cytoplasmic fractionation

Separation of nuclear and cytoplasmic fractions of cell cultures was performed following the Abcam protocol for subcellular fractionation (Abcam, Cambridge, UK). Briefly, after PRL stimulation, cells were washed with ice-cold PBS, and harvested in 500 μL fractionation buffer by scraping. After incubating for 15 min on ice, the cell suspension was passed through a 27-gauge needle 10 times. After another 20 min incubation on ice, the samples were centrifuged at 720 g for

5 min at 4oC. At this point, the pellet contained the nuclei, and the supernatant (containing the cytoplasm, membrane, and mitochondria) was transferred to a fresh tube and kept on ice. The nuclear pellet was then washed with 500 μL fractionation buffer, and the pellet was passed through a 25-gauge needle 10 times. These nuclear samples were centrifuged again at 720 g for 10 min at

4oC. The supernatant was discarded, and the pellet was then resuspended in TBS with 0.1% (v/v)

SDS. The nuclei were then sonicated briefly to shear genomic DNA and homogenize the lysate (3 s on ice at a power setting of 2-continuous). The cytoplasmic/membrane/mitochondrial supernatant was then centrifuged at 10,000 g for 5 min at 4o, and the resulting supernatant containing the cytoplasm and membrane fractions was transferred to a new tube on ice. Both the cytoplasmic and nuclear fractions were then placed in 4x Laemmli buffer and run on a SDS-PAGE gel followed by

WB analysis.

Transient transfection and harvest for apoptosis assay

MCF7 or T47D cells with endogenous STAT5a KD were assessed for signs of apoptosis after transient transfection with the STAT5a rescue constructs. After cells adhered in 6- or 12- well- plates, the media was changed to Opti-mem, and cells were transfected using Lipofectamine 3000

(Thermo Fisher Scientific, Waltham, MA). 24 hours after transfection, cells were scraped in media and all media was collected and centrifuged for 5 min at 1000 g. Cells were resuspended in ice- cold PBS to wash off any remaining media, and centrifuged. The PBS supernatant was discarded, and cells were then resuspended in RIPA buffer containing 1x protease inhibitor cocktail and 1%

(v/v) phosphatase inhibitor cocktail. Samples were sonicated in two cycles of 10 1s pulses, with 1 min rest between each cycle. Lysates were then incubated on ice for 10 min, and subsequently centrifuged at 12,000 g for 10 min at 4°C. Samples were then diluted in 4x Laemmli buffer and run on SDS-PAGE gel followed by WB analysis. Cleaved caspase 3 (CC3) antibody or cleaved caspase 7 antibody (CC7; Cell Signaling Technology, #9661, #9491) were used to probe samples for indications of apoptosis.

Immunoprecipitation

For immunoprecipitation experiments, adherent cells were first rinsed in ice-cold PBS, and scraped in ice-cold co-immunoprecipitation buffer (Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA,

0.1% Triton X-100, 5% glycerol, 1% Phosphatase Inhibitor Cocktail 2, and 1x Protease Inhibitor).

Cells were lysed and pelleted at 12,000 g for 10 min at 4°C. Supernatants were transferred to a new tube. In all cases, 5% input was saved for WB analysis before 4 μg of α-V5 antibody (or α-

IgG2a antibody control) was added and incubated overnight. 50 µl of Dynabead protein G magnetic beads (Thermo Fisher Scientific, Waltham, MA) were added for 1 h at 4°C and samples were collected on a magnetic particle concentrator (DynaMag™-2 Magnet, Themo Fisher Scientific) and washed three times in fresh IP buffer. Bound proteins were recovered in 2x Laemmli buffer and run on a 7% SDS-PAGE gel followed by WB analysis.

SDS-PAGE electrophoresis for protein analysis

Handcasting SDS-PAGE gels

In the case that protein detection required an SDS-PAGE gel of a single percent, gels were prepared by handcasting following the Bio-rad “Handcasting Polyacrylamide gels” protocol online. In brief, resolving gels were prepared using 30% acrylamide/bis (29:1) solution, 10% (w/v)

SDS, dH2O, and Resolving Gel buffer (1.5 M Tris-HCl, pH 8.8; Bio-rad, #1610798), de-gassed for 15 min, and 10% (w/v) APS and TEMED were added. Gels were poured into casting modules, and topped with water-saturated n-butanol, and allowed to set for 1h. The stacking layer was prepared using 30% acrylamide/bis (29:1) solution, 10% (w/v) SDS, dH2O, and Stacking Gel buffer (0.5 M Tris-HCl, pH 6.8; Bio-rad, #1610799), and de-gassed for 15 min, with subsequent

addition of 10% (w/v) APS and TEMED. Before adding the stacking layer, the n-butanol was discarded, and the top of the resolving layer was rinsed with dH2O. After adding the stacking layer, a comb was placed into the casting module and the gel was set to polymerize for 1 h.

SDS-PAGE electrophoresis

Protein separation was achieved under reducing and denaturing conditions using sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). Protein samples were mixed with 4x Laemmli sample buffer (Bio-rad, Hercules, CA) containing 20% (v/v) β-mercaptoethanol, and boiled at 100oC for 10 min. Equal amounts of protein were run along with 10 μl Precision Plus

Protein™ Kaleidoscope™ Prestained Protein Standards on 7%, 10%, or 4-20% Mini-PROTEAN

TGX acrylamide gels to optimize size separation utilizing the Mini-PROTEAN 3 System (Bio- rad, Hercules, CA). 1x Tris/Glycine/SDS running buffer was used for electrophoresis. Gels were run under constant voltage (105 mV) until the dye front was run off the gel.

Western blot

Following protein separation as described above, proteins were transferred to PVDF membranes using Trans-blot Turbo system (Bio-rad, Hercules, CA) following the manufacturer's instructions. After transfer, membranes were blocked in 5% milk (w/v) in TBST, or, for phospho- specific antibodies, 5% BSA in TBST (w/v). After 1h, membranes were incubated overnight (O/N) in the appropriate antibody diluted in blocking buffer at 4oC. The following day, primary antibody was removed and membranes were washed in TBST 3 times for 5 mins/wash. Secondary antibody diluted in blocking buffer was added to membranes for 1 h incubation at room temperature,

followed by 3 washes of 5 min/wash. Membranes were developed with Immobilon Forte ECL substrate (Sigma-Aldrich, St. Louis, MO) for 5 min. Membranes were imaged using the GE

ImageQuant LAS 4000 (Chicago, IL). Images were analyzed by densitometric analysis in the

ImageQuant TL software.

Western blot stripping

To remove primary and secondary antibodies from the membranes for re-probing, blots were washed for 15 min in Western blot stripping buffer, then washed twice in TBS for 10 min/wash, followed by two 5 min washes with TBST. Membranes were then blocked in blocking buffer for 1h at RT, followed by primary antibody diluted in blocking buffer O/N at 4oC for continuation of the Western blot protocol.

Western blot analysis and quantification

Preliminary densitometric analysis of western blots was performed using ImageQuant TL

Toolbox software. Normalization to loading controls was performed in Excel.

2.2.4 Cell culture methods

Maintenance of human cell lines

o All cell lines were maintained in a humidified 37 C incubator with 5% CO2.

Serum starvation and prolactin stimulation

For experiments requiring serum starvation and prolactin (PRL) stimulation, cells were washed with 37oC DPBS, and the media was replaced with serum free media [phenol-free DMEM,

0.1% (v/v) BSA] for 20-24h. PRL was added at 250 ng/ml at the indicated time point.

Soft agar colony growth

A base agar layer [0.6% noble agar, 10% serum (v/v)] was established in a 6-well plate.

The bottom layer was overlaid with 0.5 mL of top agar solution [0.3% noble agar, 10% serum

(v/v)] containing 1.0x106 cells in a single-cell suspension. Cells were incubated at 37°C for 2 weeks and stained using MTT. 10 images/well were obtained for subsequent quantification.

Quantification was performed using CellProfiler (cellprofiler.org). 2.5x103 µm2 was the defined cut-off for colony calling.

Proliferation assay utilizing the xCELLigence apparatus

Real-time monitoring of the effects of prolactin on proliferation of breast cancer cell lines in vitro was measured using the xCELLigence Real-Time Cell Analyzer (RTCA; Acea

Biosciences, San Diego, CA). 50 μl of complete media was first added to the E-Plate 16 and a baseline reading was obtained on the RTCA software. Breast cancer cells were seeded at a density of 5,000 cells/50 μl per well in the E-Plate 16, such that 100 μl of media total was added. The plate was set at RT for 30 min before being placed back in the xCELLigence for 20-24 h in a humidified

o incubator at 37 C with 5% CO2, until a Cell Index (CI) between 1-2 was reached. At that point, the experiment was paused and the E-Plate 16 was removed from the apparatus, wells were washed

twice with PBS, and 100 μl serum free media (phenol-free DMEM containing 0.1% BSA) was added to the cells and the E-Plate was placed back in the xCELLigence for 24 h as the CI leveled off. After 24 h, the experiment was again paused, and 50 μl of media was replaced with either 50

μl of serum-free media as described above, or with serum-media containing 2x prolactin for a final concentration of 250 ng/ml in each well. The plate was placed back on the apparatus and the experiment was resumed for at least 72 hours.

Scratch wound assay utilizing the Incucyte WoundMaker

To assess cellular migration, the Incucyte WoundMaker (Sartorius, Essen Bioscience, Ann

Arbor, MI) was used to monitor real-time scratch wound migration in a 96 well plate containing the cells of interest. Following the manufacturer protocol, cells were seeded onto the 96 well plate at a density of 40,000 cells per well in 100 μl. Cells were allowed to adhere at 37oC for about 20-

24 h until 100% confluency was reached. A scratch was made in all wells simultaneously using a

WoundMaker (Essen Bioscience) according to the manufacturer’s instructions. The cells were washed with culture media twice, and complete media was added to the wells before placing the plate in the Incucyte and programming image acquisition at 10x every 3 hours for 72 hours (1 image/well). Image data were analyzed using the Incucyte software and quantification of relative wound width closure over time was performed in Excel.

Immunofluorescence (IF) slide preparation

Single cell suspensions were seeded onto sterile glass microscope slides at a density of

50,000 cells. After 24 h, complete media was gently aspirated, wells were washed with DMEM,

and replaced with serum free media. After 24 h of serum starvation, slides were stimulated with

250 ng/ml PRL for indicated time points. Cells were then fixed with 4% paraformaldehyde in PBS, and permeabilized in PBS containing 0.1% (v/v) Triton X-100. Cells were labeled with α-Stat5a antibody (Santa Cruz, Dallas, TX) at a concentration of 1:50 in PBS containing 0.1%(v/v) Triton

X-100 and visualized with Alexa Fluor 594 nm donkey anti-mouse fluorescein (FITC) conjugated secondary antibody at 1:2000 in PBS containing 0.1%(v/v) Triton X-100. Nuclear staining was done with 1 μg/ml Hoechst (Thermo Fisher Scientific, Waltham, MA) in PBS before mounting in

ProLong Gold antifade reagent without DAPI (Thermo Fisher Scientific, Waltham, MA) and curing for 24 h.

2.2.5 Tissue methods

Tissue microarray (TMA)

Human breast cancer tissues were obtained from VCU Anatomic Pathology in the form of a TMA. All tissues were stripped of all patient identifiers before use and were therefore anonymous to investigators. The TMA consisted of 47 human unmatched breast cancer samples including invasive ductal and lobular carcinoma, and ductal carcinoma in situ (DCIS). Each sample has triplicate cores to mitigate against possible tumor heterogeneity. The presence and integrity of tumor in each sample was confirmed by hematoxylin and eosin (H & E) staining. Samples of the

TMA were stained for pY694-, pS726-, pS780-, and total STAT5a following standard IHC protocol. The TMA was scored independently by a clinical pathologist in the VCU Department of

Pathology and scored for both Proportion Score and Intensity Score for Allred score calculations in regards to nuclear staining.181

2.2.6 RNA-sequencing methods

RNA isolation for RNA sequencing (RNA-seq)

Cells were plated on 15 cm tissue culture dishes to about 70% confluency in complete media before incubation in serum starvation media for 20-24 hours in phenol-free DMEM (Life

Technologies) with 0.1% FBS, and subsequently treated with PRL (250 ng/ml in PBS) or PBS control for two hours. After treatment, cells were washed with 37oC PBS and mRNA was isolated with PureLink™ RNA Mini Kit (cat #12183018A) with on-column DNase treatment (PureLink™

DNase Set, cat#12-185-010, Thermo Fisher Scientific) according to manufacturer’s instructions.

Each rescue (EV, WT-, Y694F-, S726A-, or S780A-STAT5a) treated with or without PRL

(NO/PRL) was assessed in three independent biological replicates, for a total of 30 RNA samples.

RNA purity was determined by spectrophotometry at 260, 270, and 280 nm. RNA integrity number

(RIN) was assessed on the Agilent 2200 TapeStation Bioanalyzer (Agilent Technologies), all samples possessed RIN ≥9. mRNA library preparation and sequencing were performed the

Genomics and Microarray Core at the University of Colorado Anschutz Medical Campus (CU

Anschutz). mRNA library was prepared by polyA selection on ≥500 ng RNA using the Universal

Plus mRNA-Seq Library Preparation Kit (NuGEN) following manufacturer’s protocol.

Sequencing was performed on the S4 flow cell platform of the Illumina NovaSEQ6000 System

(Illumina, Inc), generating ≥50 million 150 bp paired-end (PE) reads per sample. All libraries were prepared and sequenced in a single batch/flow cell lane.

Bioinformatics analysis

FastQ files were analyzed with FastQC, and read trimming was done with CutAdapt to trim adapter contamination.182,183 Reads were then aligned to the latest version of the human genome, GRCh38 distributed by Gencode. The primary genome fasta file was downloaded from ftp://ftp.ebi.ac.uk/pub/databases/gencode/Gencode_human/release_29/GRCh38.primary_assembl y.genome.fa.gz. The matching gene annotation GTF file was downloaded from ftp://ftp.ebi.ac.uk/pub/databases/gencode/Gencode_human/release_29/gencode.v29.primary_asse mbly.annotation.gtf.gz. STAR v2.7.3a was used to align the fastq files to the human reference genome.184 STAR log statistics were compiled with the MultiQC tool, and report that over 90% of reads in each sample were uniquely mapped, with less than 5% reads in each sample being left unmapped. Prior to transcript quantification, the GFFRead utility was used to create a transcriptome fasta file from the main reference genome fasta file.185 Salmon v0.12.0 utilized the transcript fasta file to quantify each transcript from the STAR-aligned data using “quant” mode and library type set to “ISF”.186 Salmon output included the raw count estimate, TPM value, and estimated transcript length for each gene. The tximport R package was used to import the Salmon quantifications from each sample into R v3.6.0, identify the official gene symbol, and merge raw count and TPM values into a single matrix file.187,188 Trimmed BAM files of Chr17:42,287,547-

42,311,943 for GRCh38 human genome were generated to confirm point-mutations of the STAT5a mutants in Integrated Genome Browser (IGB, BioViz).189

Differential expression analyses were performed in EdgeR using the Galaxy web platform, specifically, raw estimated read counts from Salmon were analyzed using the public server at usegalaxy.org.190–192 Genes with zero expression in all samples included in the contrasts were removed with a CPM filter >0.5. Genes were filtered on p value of <0.05, and differentially

expressed genes (DEGs) were determined using a cutoff of absolute log fold change >0.58 (fold change >1.5). DEGs were determined for the following contrasts: each rescue line PRL vs NO

(untreated, ie WT_PRL-WT_NO). Heatmaps for DEG expression were generated using heatmap2 in the Galaxy web platform. Lists of DEGs, with gene identifiers and corresponding expression values (in log fold change) and p values, were uploaded into Ingenuity Pathway Analysis software

(IPA 2020, Qiagen) for pathway predictions. DeepVenn (Tim Hulsen, 2020) was used for generating all proportional venn diagrams.193 Enrichr analysis (https://maayanlab.cloud/Enrichr/) of the ENCODE gene sets library was performed on significant DEGs for the PRL vs untreated contrast of each STAT5a mutant.194 Gene set enrichment analysis (GSEA) of annotated gene sets from the Molecular Signatures Database (MSigDB) was used on expression values of all DEGs from the PRL versus untreated contrast of each STAT5a mutant as well as each STAT5a mutant

+ PRL versus WT-STAT5a + PRL.195,196

TCGA and PDX datasets were integrated with the MCF7 STAT5a phospho-deficient mutant dataset by Amy L. Olex, Senior Bioinformatics Specialist affiliated with C. Kenneth and

Dianne Wright Center for Clinical and Translational Research as defined previously.197

All data generated and analyzed in this study are included in this published article. RNA- seq datasets have been submitted to the National Center for Biotechnology Information Gene

Expression Omnibus (GEO) and are accessible through GEO Series accession number GSE165678

(https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc= GSE165678).

2.2.7 Statistical analysis

Statistical analysis was performed using one-way or two-way analysis in GraphPad Prism

7 (San Diego, CA). Results are reported as means 土 SEM (standard error of the mean). P< 0.05

was considered statistically significant. All experiments were done three times unless otherwise noted.

3. Chapter 3: Serine residues 726 and 780 have nonredundant roles

regulating STAT5a activity in luminal breast cancer

3.1. Introduction

In mammary epithelium, the polypeptide prolactin (PRL) acts through its cognate receptor

(PRLr), and is responsible for terminal maturation and differentiation of lactating glands during pregnancy. 6,11,12,44,45,49,51 As a potent mitogenic factor, PRL/PRLr signaling is also associated with mammary tumorigenesis. 6,73,85,86,198,199 Canonical PRL/PRLr signaling involves the tyrosine janus kinase 2 (JAK2), which phosphorylates the transcription factor (TF) signal transducer and activator of transcription 5a (STAT5a) on tyrosine 694 (Y694). The resulting phosphorylated STAT5a

(pY694-STAT5a) dimerizes and translocates to the nucleus.

STAT5a is one of seven members of the STAT family of transcription factors. The STAT family is highly conserved, each having a tyrosine residue near residue position 700 that is phosphorylated by JAK in the cytoplasm. This phosphorylation event has historically been considered as the activation switch for STAT (pY-STAT), whereas dormant, unphosphorylated

STAT (upY-STAT), has long been considered to have no significant functions. However, accumulating evidence shows that upY-STATs are present in the nucleus regardless of cytokine stimulation. 157,200 Studies have illustrated that upY-STATs actively bind chromatin to drive gene expression distinct from that of the pY-STAT forms. 111,152,155,158,168,200,201 Notably, other functions described of upY-STATs include promotion of heterochromatin and global downregulation of transcription. 157,159 These differing reports highlight the possibility of stimulus- and tissue-specific functions of upY-STAT vs. pY-STAT.

Compounding this growing understanding of the regulation of STAT activity, STATs (with the exception of STATs 2 and 6) also have serine residues in the Transactivation Domain (TAD) that are capable of being phosphorylated, with known function in regulating STAT activity.

163,166,202–205 To this point, STAT5a has two serine residues, S726 and S780, which regulate

STAT5a-dependent malignant hematopoietic transformation. 169,178 The function of phosphorylation of either S726 or S780 in PRL-responsive normal or malignant breast tissue, however, is less well understood.

Several studies regarding STAT5a serine phosphorylation have focused on PRL-induced

β-casein expression using luciferase gene reporter assays. 127,170,175 However, the results of these studies are inconsistent regarding the loss of the serine residues on PRL-stimulated β-casein gene induction. Beuvink et al (2000) reported that loss of S726 or S780 had no effect on PRL-stimulated

β-casein promoter activity. 170 While Yamashita et al (2001) also observed that loss of S780 had no effect on PRL-stimulated β-casein promoter activity, in contrast they found that expression of phospho-deficient S726A-STAT5a in COS-7 and MCF7 cells increased the PRL-stimulated β- casein promoter activity compared to WT-STAT5a. 127,175 These studies did not examine the physiologic effect of phospho-deficient STAT5a expression.

In this study, we examine the expression of serine-phosphorylated (pS-)STAT5a in human breast cancer, describe the effects of tyrosine and serine phospho-deficient STAT5a mutants on breast cancer characteristics in vitro, and show for the first time how these residues differentially affect transcriptional programs in luminal breast cancer.

3.2. Hypothesis

We hypothesize that loss of the STAT5a phospho-sites will differentially affect STAT5a activity that do not indicate regulation of pY694-STAT5a but indicate non-redundant, discrete roles as evidenced by changes in the transcriptome and phenotypic characteristics of breast cancer.

3.3. Results

Stat5a serine residues 726 and 780 are phosphorylated in human breast cancer cell lines and patient tumor samples

Serine phosphorylation of STAT5a has been observed in mouse mammary epithelial cells throughout mammary development, however this has yet to be clinically characterized in human breast cancer. 170,175 We performed immunohistochemical (IHC) staining of a breast cancer tissue microarray (TMA) using phospho-specific antibodies for pY694-, pS726-, and pS780-STAT5a, as well as total STAT5a. The TMA consisted of 47 clinically staged cancers I-III, with diagnoses ranging from ductal carcinoma in situ (DCIS) to invasive ductal or lobular carcinoma

(Supplemental Table S1, Appendix). Visual scoring of STAT5a S726 phosphorylation revealed a significant increase in nuclear intensity in tumor samples grade III compared to grade I (nuclear

Allred scores of 6 and 2, respectively; Figure 3.1A). Notably, STAT5a S780 phosphorylation was also observed in the nucleus of tissue samples, however there was no significant association of expression with either tumor grade or proliferative status (Ki67 staining; Figure 3.1A). While total nuclear STAT5a expression remained high throughout malignant progression in these breast tissue samples, a weak signal for nuclear pY694-STAT5a was observed in the tissue samples independent of tumor grade, hormone receptor status, or Ki67 staining, consistent with previous reports (Supplemental Figure S1). 150

Patient derived xenograft (PDX) whole cell extract (WCE) samples (Figure 3.1B) and established breast cancer cell lines (Figure 3.1C) spanning the intrinsic molecular subtypes were probed for STAT5a S726 and S780 phosphorylation. 206 HCI-011 (luminal B, ER+/PR+/HER2-) showed expression of both pS726- and pS780-STAT5a (Figure 3.1B), in concordance with the luminal A/B breast cancer cell lines MCF7 and T47D (Figure 3.1C). 4,197 Interestingly, the claudin-low breast cancer cell line MDA-MB-231 exclusively exhibited pS780-STAT5a as well

(Figure 3.1C). 4 Ex vivo analysis of HCI-011 lysates demonstrated pY694- and pS726-STAT5a expression following PRL stimulation (Figure 3.1D). Likewise, PRL stimulation of both T47D cells and MCF7 cells results in robust S726-STAT5a phosphorylation (Figure 3.1E, F). However, phosphorylation of STAT5a at S780 was observed to be constitutive and independent of PRL stimulation, consistent with literature findings (Figure 3.1E, F). 170,175

Figure 3.1. STAT5a S726 and S780 phosphorylation occur in human breast cancer. A)

Representative anti-pS726- and anti-pS780-STAT5a IHC images from unmatched Grade III, ER- /PR-/Her2 amplified primary tumors included in the TMA (scale bars, 100 μm). Note the nuclear accumulation of pS726- and pS780-STAT5a. Graphs show quantification of nuclear Allred scores according to the tumor grade, Ki67 status, and molecular subtypes. Panels of PDX samples (B) and established breast cancer cell lines (C) immunoblotted for pY694-, pS726- and pS780- STAT5a. Note the presence of pS726/S780-STAT5a in the luminal B subtype (ER+/PR+/Her2-) sample HCI-011, and in the luminal B cell lines T47D and MCF7, along with pY694-STAT5a. Lysates for C) were run in triplicate and representative blots for STAT5a and tubulin are shown. D) HCI-011 lysates were subjected to PRL stimulation and analysis for pS726-STAT5a and pY694-STAT5a induction. PRL stimulation of the human breast cancer cell lines T47D (E), and MCF7 (F), show induction of pS726-STAT5a but constitutive pS780-STAT5a, quantified by densitometric analysis. n=3, *p≤0.03, **p=0.001, ***p≤0.0004 compared to untreated, time 0.

STAT5a serine residues have nonredundant transcriptional roles in luminal breast cancer cells Previous studies have shown that STAT5a serine phosphorylation affects PRL-induced

STAT5a DNA binding kinetics (as measured by electrophoretic mobility shift assay, or EMSA) and phosphorylation kinetics of residue Y694, however the role of these phospho-sites in PRL- regulated gene expression has not been investigated on a global scale.170 We hypothesized that expression of serine phospho-deficient STAT5a mutants would affect breast cancer transcriptomic programs and these alterations in gene expression patterns would lead to changes on a phenotypic scale. To test this hypothesis, we utilized MCF7 cells stably transduced with either STAT5a shRNA (STAT5a KD) or nontargeting (NT) control shRNA (Figure 3.2A, B). The STAT5a shRNA targeted the 3’untranslated region (UTR) of Stat5a on chromosome 17 such that rescue of

STAT5a by cDNA was possible. Pools of MCF7 cells with stable STAT5a KD were rescued with lentiviral transduction of WT STAT5a, phospho-deficient single-point mutants (containing a C- terminal V5-tag), or empty vector (EV), such that stable novel MCF7 cell lines were established.

Hereafter, these transfectants are referred to as WT-STAT5a, Y694F-STAT5a, S726A-STAT5a,

S780A-STAT5a, or EV (Figure 3.2A, C). Relative STAT5a KD and rescue protein expression was confirmed by Western blot (WB; Figure 3.2B, C, D). Endogenous STAT5a KD was determined as being ~50% effective by qRT-PCR, and rescue STAT5a transcript expression was

250-400-fold higher than that of the nontargeting control (Supplemental Figure S2). However, on the protein level, STAT5a was equally expressed by the rescues (Figure 3.2D). Therefore, quantification of all subsequent studies using these MCF7 pooled rescue cell lines is reported normalized to V5-STAT5a expression.

Figure 3.2. The human breast cancer cell line MCF7 as a model for STAT5a serine phosphorylation. A) Schema for generating of stable knockdown and rescue cell lines. B) STAT5a knockdown in MCF7 cells after antibiotic selection confirmed by WB. C) Representations of WT-, Y694F-, S726A-, and S780A-STAT5a constructs with C-terminal V5 tags used to generate MCF7 rescue cell lines. Confirmation of STAT5a rescue expression in MCF7 cells by D) protein analysis with appropriate antibodies targeting each phospho-site of STAT5a.

These generated stable MCF7 cells were then stimulated with PRL (250ng/mL) and subjected to RNA sequencing (RNA-seq), to assess global transcriptomic changes following

STAT5a mutagenesis. These analyses revealed that following PRL treatment, MCF7-WT-

STAT5a cells had 510 differentially expressed genes (DEGs), and each STAT5a phospho- deficient mutant had fewer DEGs than the WT-STAT5a (ranging from 171 [Y694F-STAT5a] to

285 DEGs [S726A-STAT5a]; Supplemental Table S2). Next, the top 50 DEGs from the four

STAT5a rescues when treated with PRL vs untreated (for a total of ~200 DEGs) were hierarchical clustered (Figure 3.3A). Results from Y694F-STAT5a MCF7 cells treated with PRL were most similar to untreated WT-STAT5a MCF7 cells, indicating that loss of Y694-STAT5a is sufficient to inhibit most PRL-induced transcript expression. When treated with PRL, S726A- and S780A-

STAT5a MCF7 cells had similar expression of the top 50 DEGs from each contrast, visualized by the clustering of the S726A + PRL and S780A + PRL samples. Interestingly, when untreated, the three STAT5a point mutants top 50 DEGs cluster away from the untreated WT-STAT5a MCF7 cells. This indicates that in these cells STAT5a may drive expression of a subset of genes in the absence of tyrosine or serine phosphorylation. This would be a distinct function for STAT5a that is independent from its role of inducible, phosphorylation-dependent gene expression, similar to descriptions of STAT3. 158,160

WT-STAT5a PRL-induced or inhibited genes were compared to previous PRL- transcriptomic data released by our laboratory, such that after matching the criteria that the DEG be present in at least two datasets, 105 putative PRL-regulated genes were identified

(Supplemental Table S2). 70,78 A subset of these genes was used to identify functionally-enriched pathways with PRL treatment that may be changed by expression of the STAT5a mutants through

Ingenuity Pathway Analysis (IPA, Qiagen, Figure 3.3B). The PRL treatment of WT-STAT5a

MCF7 cells significantly activated pathways responsible for cellular proliferation and survival, colony formation, migration, and invasion. Interestingly, expression of the Y694F-STAT5a mutant

had an opposite effect on 10 of the 13 (76.9%) queried functional predicted pathways compared to WT-STAT5a, including cell survival and colony formation pathways (Figure 3.3C). Expression of the S726A-STAT5a mutant similarly affected 7 of the same pathways that Y694F-STAT5a expression did (7/13, 53.8%), and S780A-STAT5a affected the senescence pathway only (1/13,

7.7%; Figure 3.3D, E).

Figure 3.3 Expression of STAT5a phospho-mutants differentially affects PRL-regulated

gene expression in MCF7 cells. A) Hierarchical clustering of each STAT5a species’ top 50 DEGs (≥1.5-fold change) from RNA-seq analysis of PRL treated versus untreated MCF7 cells expressing STAT5a rescues. B) IPA functional pathway enrichment analysis shows upstream regulators and predicted pathway activation or inhibition with PRL treatment of WT-STAT5a-expressing MCF7 cells. C-E) IPA functional pathway analysis of indicated STAT5a phospho-mutants overlaid with the regulatory network shown in panel B to predict activation or inhibition. p value calculated by Fisher’s exact test.

Confirmation of the gene expression changes observed in the RNA-seq data was done with

MCF7 cells similarly stimulated with PRL. qRT-PCR was performed on a panel of genes, including the gene encoding cytokine inducible SH2 containing protein (CISH), a known STAT5a- target gene induced by PRL in breast epithelial cells and in mammary carcinoma. 83,207 As seen in

Supplemental Figure S3, loss of pY694 completely ablated PRL-induced CISH expression compared to WT-STAT5a. S726 and S780 point mutagenesis similarly were sufficient to significantly reduce CISH expression following PRL stimulation, compared to that of WT-

STAT5a, albeit to a lesser extent than was observed for the tyrosine mutant. These results indicate some role for serine phosphorylation in modulating STAT5a transcriptional activation. Further,

Estrogen Receptor α (ESR1) expression was shown to rely on phosphorylation of the phospho- serine residues differentially. Loss of pS726 increased PRL-induced ESR1 expression relative to

WT-STAT5a (an increase that is more pronounced when the expression of total STAT5a is considered; Supplemental Figure S3), whereas loss of pS780 did not significantly affect ESR1 expression.

Function of the STAT5a phospho-serine residues in vitro are independent of pY694

To confirm these functional predicted effects observed with expression of these STAT5a mutants, we next assessed the loss of the STAT5a phospho-sites on cancer characteristics in vitro.

The ability of transformed cells to grow independently and survive without extracellular matrix attachment can be used to understand cell differentiation, transformation and tumorigenesis. 208,209

Therefore, we performed a quantitative evaluation of soft agar assay with various growth conditions. In quantifying three independent experiments compared to WT-STAT5a rescue colonies, MCF7 cells rescued with either Y694F- or S780A-STAT5a had fewer colonies (Figure

3.4A, Supplemental Figure S4). This indicates that not only phosphorylation of Y694, but also phosphorylation of S780 on STAT5a is involved in the clonogenicity of breast cancer, in that fewer

MCF7 colonies were established with loss of the phospho-sites. When analyzed for the size of colonies established, total STAT5a knockdown had increased colony size compared to WT-

STAT5a rescue (p=0.02, Figure 3.4A). Together, these results illustrate that STAT5a expression restricts the extent of breast cancer outgrowth and clonogenicity, and that loss of any individual phospho-regulatory site on STAT5a does not mimic total loss of STAT5a. 210

To further study the roles of each phospho-site of STAT5a, proliferation in 2D culture of the MCF7 rescues was measured on an xCELLigence apparatus (Agilent Technologies). In the presence of PRL, the S726A-STAT5a rescues showed decreased proliferation compared to WT-

STAT5a rescues from 12 to 21 hours (Figure 3.4B, Supplemental Figure S4). This change from

WT-STAT5a indicates a role for the phosphorylation of STAT5a on S726 mediating PRL-induced proliferation.

Figure 3.4. Effect of single-point phospho-deficient STAT5a mutants on characteristics of MCF7 cells. A) Representative images and quantification of number and size of the colonies formed in soft agar by indicated MCF7 cells with knockdown (KD) of STAT5a rescued by phospho-deficient mutants. n=3, *p<0.02 compared to WT-STAT5a. B) Proliferation in serum free media (SFM) supplemented with PRL was assessed using XCELLigence, cell index indicates proliferation. n=3, p<0.02 compared to WT-STAT5a. C) Immunoblots of T47D cells (top) and MCF7 (bottom) cells carrying transient rescues of STAT5a point mutants assessed for apoptosis by expression of cleaved caspase 3 or 7.

The migratory ability of MCF7 cells in 2D culture was analyzed by real time imaging using a scratch wound assay. 72 hours after establishing a wound in a confluent cell layer, all cell lines

containing STAT5a rescues were 100% confluent once more (Supplemental Figure S5).

Unremarkably, the knockdown of STAT5a, or the rescue with various phospho-deficient point mutations, had no effect on the ability of MCF7 cells to migrate on the 2D plastic layer, indicating single-phospho-point mutants are not sufficient to elicit a phenotypic effect on migration.

STAT5a and PRL are both implicated in dysregulating apoptotic machinery in breast cancer cells, which was confirmed in our pathway analysis of the MCF7 STAT5a mutants RNA- seq data (Figure 3.3B-E). 24,62–64 Further, it was observed that the T47D breast cancer cell line, when either transiently transfected or stably transduced with the STAT5a rescue constructs, had variable success in establishing 2D colonies; ultimately only transient expression was viable for performing quantitative assays. MCF7 cells do not express caspase 3, therefore to investigate this phenomenon, we analyzed protein expression of cleaved caspases 3 (CC3) and 7 (CC7) in T47D and MCF7 cells, respectively, with endogenous STAT5a knockdown and transient re-expression of the STAT5a rescues (Figure 3.4C).180,212 T47D cells rescued with S726A-STAT5a were highly positive for CC3, indicating that this serine residue contributes to the survival of the T47D cell line. On the other hand, MCF7 cells exhibited no CC7, indicating that these cells do not rely on

STAT5a for survival. This difference in STAT5a regulation of apoptosis may be due to the difference in apoptotic pathways used in T47D and MCF7 cells, or by the reported reliance of

T47D cells on STAT5a for survival, and now, specifically pS726-STAT5a.130 Overall, these studies illustrate that phosphorylation of S726 may be responsible for STAT5a involvement in evasion of apoptosis, while phosphorylation of S780 is responsible for STAT5a involvement in tumor clonogenicity and establishment of anchorage-independent colonies.

The canonical pathway of STAT5a is independent of phosphorylation of S726 or S780

To determine the role of pS726 and pS780 in regulation of the canonical pathway of

STAT5a signaling, the kinetics of phosphorylation of Y694-STAT5a was analyzed in the corresponding rescue cell lines. Previously it was reported that loss of S726 stabilized PRL- induction of pY694, while WT-STAT5a loses pY694 by 5 h of PRL stimulation. 170 However,

MCF7 cells re-expressing either WT-STAT5a or S726A-STAT5a have comparable induction of pY694 up to 6 h (Figure 3.5A). Similarly, we found that after 2 h and 6 h of PRL stimulation, the

S780A-STAT5a mutant MCF7 cells had no effect on PRL-induced phosphorylation of the JAK2- target Y694 (Figure 3.5B), indicating that neither phospho-serine contributes to maintenance of pY694.

Figure 3.5. Induction of phosphorylation on residue Y694 is independent of either S726 or S780. MCF7 cells with stable re-expression of either WT-STAT5a, S726A-STAT5a (A), or S780A-STAT5a (B) were treated with PRL up to 6 h. Representative blots of 3 independent experiments for pY694-STAT5a induction shown, with quantification by densitometric analysis.

Unphosphorylated STAT5a undergoes nuclear translocation

To determine if differences in transcriptional activity between WT-STAT5a and the phospho-site STAT5a mutants is due to altered nuclear localization of STAT5a, we performed

both immunofluorescence (IF) imaging and WB analysis of nuclear-cytoplasmic fractionation with

PRL stimulation (Figure 3.6, Supplemental Figure S6). Nuclear-localized WT-STAT5a began to increase by 30 minutes and was significantly increased by 60 minutes of PRL stimulation. While there was an upward trend for Y694F-STAT5a translocating to the nucleus, there was a no net change in translocation into or out of the nucleus with PRL stimulation (Figure 3.6A). The loss of

S726 phosphorylation increased STAT5a nuclear translocation compared to WT over time with

PRL stimulation (Figure 3.6A) and retained STAT5a in the nucleus up to 60 minutes following

PRL stimulation. The loss of S780 phosphorylation shifted the kinetics of STAT5a nuclear translocation such that the amount of STAT5a peaked in the nucleus with 15 minutes of PRL treatment, and then decreased to close to baseline after 30 minutes (Figure 3.6A). WB analysis of nuclear and cytoplasmic fractions after 15 minutes of PRL treatment illustrates a snapshot of these trends (Figure 3.6B). In agreement with previous data, Y694F-STAT5a was present in the nucleus independent of PRL stimulation, confirming reports that have also found upY694-STAT5a is continuously shuttled into and out of the nucleus (Figure 3.6B). 157,200 These results indicate that each phospho-serine could play an independent role in STAT5a nuclear translocation, degradation, or export.

Figure 3.6. Nuclear translocation of STAT5a occurs in absence of pS726 and/or pS780. MCF7 cells with stable re-expression of STAT5a carrying specific phospho-deficient point mutations were imaged with immunofluorescence with up to 60 min PRL stimulation, representative merged Hoechst and STAT5a images shown, and quantified for nuclear translocation (A) or analyzed by WB nuclear cytoplasmic fractionation after 15 min PRL stimulation (B). *p<0.04 compared to untreated, time 0.

In order to examine how STAT5a phospho-mutants regulate the functional phenotypes uncovered in this study, we used our RNA-seq data in combination with the open-source Enrichr web-based platform.194 Enrichr transcriptional gene set analysis (using ENCODE project chromatin immunoprecipitation [ChIP]-seq data) confirmed STAT5a is one of the top transcriptional regulators in the WT-STAT5a MCF7 cells when treated with PRL (OR 2.05, p=0.03; Figure 3.7A), along with HDAC2, a class IIb histone deacetylase similar to HDAC6 (a known STAT5a cofactor).79 Further, amongst the top 10 enriched TF terms, the AP-1 subunits

(JUN and FOS), as well as the FOX TF family (FOXA1), were also present, confirming our laboratory’s recent findings that STAT5a shares chromatin binding sites with these TF families as well as the ETS, AP-2, SP/KLF, CREB, and Nuclear Receptor (NR, e.g. ESR1) TF families.87

Conversely, Enrichr analysis of Y694F-STAT5a or S780A-STAT5a data did not identify STAT5a as one of the top significant transcriptional regulators with PRL treatment (Figure 3.7B).

Furthermore, the STAT5a transcriptional signature with PRL treatment was significantly enriched in the case of S726A-STAT5a Enrichr analysis (OR 1.21, p=0.007). Intriguingly, expression of

Y694F- or S726A-STAT5a increased the involvement of the SREBF (Sterol regulatory element- binding factor, or protein [SREBP]) TF family (Figure 3.7B, C), that was not seen with the expression of S780A-STAT5a (Figure 3.7D). Expression of S780A-STAT5a with PRL treatment, however, showed involvement of STAT3, SP1, and two of the FOX TF family members (Figure

3.7D). Overall, these data indicate that various transcription factor families may serve to

compensate for the loss of pY694- and pS-STAT5a differentially to uphold oncogenic signals in

MCF7 cells.

Figure 3.7. Transcription factor gene set analysis identifies complex transcriptional networks contributing to STAT5a mutant-driven phenotypes. TF analysis of the DEGs from MCF7 cells treated with PRL for A) WT-STAT5a, B) Y694F-STAT5a, C) S726A-STAT5a, and D) S780A- STAT5a. The ENCODE project ChIP-seq data set library was used to find the top-ranking transcription factors by odds ratio (OR), p<0.05.

3.4. Discussion

The phospho-regulation of STAT5a by PRL in breast tissues has long been thought to only encompass the phosphorylation of the residue Y694. However, it is becoming clearer that regulation of this transcription factor is more nuanced. Here, we demonstrate for the first time, phosphorylation of S726 and S780 in human breast carcinoma samples. While both residues are

phosphorylated in tumor grades I-III, pS726 increases with tumor grade, and is PRL-responsive.

Further, pS726 expression is restricted to luminal breast cancer cell lines and PDXs. Conversely, pS780 is PRL-independent, confirming earlier reports that it is constitutively phosphorylated, and occurs in both luminal breast cancer cell lines and PDX lines, but also in the basal-like breast cancer cell line MDA-MB-436. 170 The constitutive, wide-ranging expression of pS780 suggests that it could result from a non-specific phosphorylation event in breast cancer pathogenesis, however studies performed herein determined that it has a non-redundant role for STAT5a in

MCF7 breast cancer characteristics.

It has been assumed that pY694-STAT5a is necessary for most, if not all, PRL-induced gene expression changes. This assumption prior to this study had not been directly assessed by qRT-PCR, or, more globally, RNA-sequencing, chromatin immunoprecipitation (ChIP), or ChIP- sequencing. 85,86,136,213–216 Rather, pharmacologic inhibition or total knockdown of STAT5a had been the focus of previous work to understand STAT5a transcriptional functions, ignoring the gradation of STAT5a transcriptional function that may be in play in breast cancer and other pathologies. 86,213,214 In this study we performed RNA-sequencing of MCF7 cells expressing single-point phospho-deficient STAT5a mutants treated with PRL in attempts to identify differentially expressed genes and functional pathways for cancer outgrowth. Examination of

PRL-regulated DEGs from the WT-STAT5a expressing cells compared to previous studies allowed us to first identify core genes responsive to PRL in this study. Then we were able to examine the activation or inhibition of upstream regulators and downstream functional pathways.

While loss of pY694-STAT5a had the largest effect on these functional pathways, loss of either of the pS-sites also had some modulatory effect on these pathways which did not mimic the loss of pY694.

For the first time, we have analyzed STAT5a serine phosphorylation on breast cancer phenotypic characteristics that fall within the hallmarks of cancer.217,218 As presented here, the roles for the serine residues of STAT5a are nonredundant, exhibiting different effects on cancer characteristics, and different regulatory roles on the canonical pathway of PRL/STAT5a signaling.

Our results indicate a role for STAT5a in establishing and maintaining anchorage-independent growth that may rely on phosphorylation of S780. Expression of Y694F-STAT5a led to a decrease in colony number compared to MCF cells expressing WT-STAT5a. However, the expression of

S780A-STA5a in MCF7 rescue cells led to a further decrease in colony number. Therefore, phosphorylation of S780 may contribute to the potential for breast cancer cells to establish viable colonies, indicating a delicate balance for STAT5a regulation. On the other hand, loss of pS726-

STAT5a had no effect on colony number or size at the end point for the soft agar assay, which indicates that S726 does not singularly contribute to the overall regulatory effect of STAT5a on anchorage-independent growth. Therefore, it is possible that targeting pS780 on STAT5a would decrease tumorigenicity, leading to better prognostic outcomes for patients.

The biologic affects from loss of either phospho-serine residue regarding proliferation and viability were likewise nonredundant. Loss of pS726 decreased PRL-stimulated proliferation in

MCF7 cells, and increased the expression of cleaved caspase 3, an effector of apoptosis, in T47D cells. Interestingly, the expression of cleaved caspase 7, another effector molecule of apoptosis, was not increased in MCF7 cells. With the loss of caspase 3, MCF7 cells may be at an advantage when faced with yet another hit to cell viability, i.e. the loss of pS726-STAT5a, compared to T47D cells. 219 Increased apoptosis in T47D cells, on the other hand, illustrates the reliance on intact compensatory mechanisms to evade apoptosis, which is lost with expression of the pS726-STAT5a

mutant. Loss of pS780-STAT5a was unremarkable on these characteristics, indicating roles for each phospho-serine residue independent of each other.

We sought to determine the mechanism by which loss of STAT5a phospho-residues affected breast cancer characteristics differentially by examining canonical STAT5a activity, and found 1) the canonical phosphorylation of pY694 is independent of either pS residue, and 2) there is nuclear localized STAT5a with loss of pY694, pS726, or pS780, with altered kinetics for translocation. These results determined that the phospho-serine residues of STAT5a perform roles independent from pY694. Taken together with the RNA-seq data, the STAT5a serine residues appear to regulate functional phenotypes at the transcriptional level.

Accordingly, in silico analysis using TF gene set enrichment analysis was used to establish how these residues alter STAT5a function at the transcriptional level. Our data from this analysis indicates that an array of TF families may be involved as compensation with loss of pY- or pS-

STAT5a function. In this study, the SREBP TF family had indicated involvement in both Y694F- and S726A-STAT5a PRL-regulated gene expression changes. As a TF, SREBP has been found to play a critical role in breast cancer migration and invasion, and its involvement in conjunction with the loss of pY694 or pS726 may indicate a role in oncogenic signal switches that confound the loss of STAT5a. 220 Activation of SREBPs are required for growth factor-independent growth and proliferation, which could explain our phenotypic findings with our functional studies. Further,

SREBP acts downstream of oncogenic PI3K or K-Ras, and our laboratory recently found that expression of K-Ras is a necessary co-factor for PRLr-mediated oncogenic transformation (Grible et al, in press). This illustrates the complexity of transcription factor activity in breast cancer, and the homeostatic mechanisms that may be at play for oncogenic growth.

Future studies of the loss of these phospho-sites on STAT5a are warranted. With loss of one phospho-site, compensatory mechanisms for recovering STAT5a activity may still be in effect.

However, if two or more of these phospho-residues are targeted, we may be able to study how only one remaining phospho-site regulates STAT5a activity. Further studies of how loss of these residues affects STAT5a binding to transcriptional cofactors, or to binding sites on chromatin, will uncover mechanisms responsible for STAT5a activity.

4. Chapter 4: Studies of double-point STAT5a mutants illustrate requirements for cooperative activation of serine and tyrosine residues

4.1. Introduction The results for STAT5a single point phospho-deficient mutants indicate non-redundant roles for the serine residues S726 and S780 in luminal breast cancer. However, these residues may also act together to affect STAT5a activity and phenotypic characteristics of breast cancer.

Studies showing phenotypic effects of mutations of both serine residues have mainly been performed in hematopoietic models. In the mouse interleukin (IL)-3 dependent, pro-B cell, cell line Ba/F3, it was shown that phosphorylation of either serine residue is required to abolish IL-3 dependence of Ba/F3 cells.169 Further, this finding was exacerbated in cells expressing a double phospho-serine STAT5a mutant. Importantly, this contrasts to constitutively active STAT5, which provides survival and proliferation stimuli resulting in cytokine-independent cell expansion.169 In another study, the loss of both serine residues of STAT5a hampered both v-ABLp160+ and BCR-

ABLp185+ leukemic cell viability in vitro. Further, in both of these studies, transplantation of leukemic cells expressing double-serine STAT5a phospho-mutants into lethally irradiated or nod scid gamma (NSG) mice significantly increased disease latency.169,178

Like the literature examining single-point STAT5a phospho-mutants and cytokine stimulation, the literature for the double-point STAT5a phospho-mutants (i.e., S726A + S780A-

STAT5a) has focused mainly on reporter assays for known STAT5a target genes or STAT5a promoter sequences. Beuvink et al (2000) found that loss of both serine residues did not decrease

PRL-stimulated β-casein promoter transcription via luciferase activity compared to WT-STAT5a in COS-7 cells.170 Conversely, Yamashita et al (2001) found that loss of both serine residues increased PRL-stimulated β-casein expression via luciferase activity compared to WT-STAT5a

100

in both COS-7 and MCF7 cells. Further, the double mutant STAT5a led to elevated luciferase activity above either single point phospho-mutant, which the authors concluded suggested functional cooperativity between the two serine sites.175 Interestingly, both studies used the same concentration and timing for PRL treatment, so it may be possible that the difference in these findings lies with the β-casein reporter construct.170,175

Park et al (2001) found that expression of the double mutant of STAT5a decreased expression of one reporter construct but increased expression of another compared to WT-

STAT5a in the presence of PRL in COS-1 cells.179 Adding to this complex finding, when the authors stimulated HepG2 cells or COS-1 cells with growth hormone (GH) and examined the luciferase activity (i.e. gene expression) of the same reporter construct, they found the double point mutant in HepG2 cells decreased luciferase activity, whereas luciferase activity was not affected in the COS-1 cells compared to WT-STAT5a.179 These findings lend to the argument that serine phosphorylation affects STAT5a transcriptional action that is dependent upon both the enhancer/promoter as well as the cell-type.

Overall, the evidence regarding loss of both serine residues on STAT5a transcriptional activity is confounding, and there remains a lack of evidence for any phenotypic effects this phospho-mutant has on breast cancer cells. Therefore, in this Aim, we sought to determine if loss of these serine residues of STAT5a, together or in conjunction with loss of tyrosine residue 694

(Y694) affected phenotypic characteristics of breast cancer in a way similar to the hematopoietic studies.

4.2. Hypothesis We hypothesize that STAT5a serine residues S726 and S780 cooperatively act with Y694 to regulate STAT5a activity and can be visualized at the phenotypic level of cancer characteristics.

101

4.3. Results Immunohistochemical (IHC) staining and scoring for pS726-, and pS780-STAT5a of the breast cancer tissue microarray (TMA) we reported on in Aim 1 was used to determine how concordant phosphorylation of both S726- and S780- STAT5a were in the samples. As shown in

Table 4.1, 77% of the 47 cancer tissues were highly concordant between pS726- and pS780-

STAT5a nuclear staining (with a difference of ≤1 modified nuclear Allred score). Specifically, 22 cases, or 47%, had equal modified nuclear Allred scores for both phospho-antibodies. Conversely, only 5 cases, or 11%, were met the criteria of having a modified nuclear Allred score of ≥5, indicating either specifically staining for pS726- or pS780-STAT5a and excluding staining for the other phospho-antibody. These results indicate that phosphorylation of both serine residues occurs clinically in breast cancer.

Table 4.1 Cases with Nuclear Allred Scores for both pS726 and pS780 as scored in the breast cancer TMA. pS726 and pS780 occurred together in 77% of the tissues. Percentage (of 47 total Criteria Number of cases cases) With ≥5 Nuclear Allred 5 11% Score difference Between 2 and 5 Nuclear 6 13% Allred Score difference With ≤1 Nuclear Allred 36 77% Score difference With Equal Allred Scores 22 47%

In a manner similar to Aim 1, we sought to utilize mutant constructs of STAT5a to determine the effects of loss of the phospho-sites. Figure 4.1A shows a schematic of the structure of STAT5a with double point mutants. MCF7 cells were stably transduced with either STAT5a shRNA (STAT5a KD) or nontargeting (NT) control shRNA (Figure 4.1B). Pools of MCF7 cells with stable STAT5a KD were rescued with lentiviral transduction of WT STAT5a, phospho-

102

deficient double-point mutants (containing a C-terminal V5-tag), or empty vector (EV), such that stable novel MCF7 cell lines were established. Hereafter, these transfectants, which were confirmed by WB, are referred to as WT-STAT5a, Y694F+S726A-STAT5a, Y694F+S780A-

STAT5a, S726A+S780A-STAT5a, or EV (Figure 4.1B, C).

Figure 4.1 Concept and schematic of MCF7 rescues with STAT5a double point phospho- deficient mutants. A) Structure of STAT5a protein with indicated point mutations, in which the peptide tyrosine (Y) is mutated to phenylalanine (F) and serine (S) is mutated to alanine (A). B) Schema for stable knockdown and rescue using antibiotic selection. C) Confirmation of double point phospho-deficient mutants by WB.

To study the effects of the expression of the STAT5a double point mutants on cancer characteristics, proliferation in 2D culture of the MCF7 rescues was measured on an xCELLigence

103

apparatus (Agilent Technologies). As seen in Figure 4.2A, the proliferative ability of MCF7 cells expressing Y694F+S780A-STAT5a or S726A+S780A-STAT5a without PRL treatment was significantly decreased from WT-STAT5a. This decrease in proliferation was not rescued with

PRL treatment. Conversely, MCF7 cells expressing Y694F+S726A-STAT5a had no difference in proliferation compared to WT-STAT5a with or without PRL treatment. Altogether, MCF7 expressing Y694F+S780A-STAT5a or S726A+S780A-STAT5a had decreased proliferation compared to WT-STAT5a regardless of PRL treatment (Figure 4.2B).

104

Figure 4.2 MCF7 cells expressing STAT5a lacking both Y694 and S780 or both S726 and S780 are unable to proliferate, and prolactin does not rescue this phenotype. Proliferation was assessed using XCELLigence, cell index (CI) indicates proliferation. A) Individual graphs of CI for cells grown in serum free media (SFM) or with PRL treatment (SFM+PRL), significance compared to cells expressing WT-STAT5a without PRL or with PRL treatment, respectively. B) Combined graph of MCF7 cells grown in SFM and with PRL treatment shows significance compared to cells expressing WT-STAT5a without PRL (SFM). n=3 **p=0.003, ***p≤0.001, ****p≤0.0001.

105

To further study the effects of the STAT5a phospho-sites together, we performed quantitative evaluation of the MCF7 rescues grown in soft agar. The ability of cancer cells to grow independently and survive without attachment to extracellular matrix is a hallmark of carcinogenesis.209 Compared to WT-STAT5a, loss of either pS726 or pS780 in combination with loss of pY694 (Y694F+S726A-, or Y694F+S780A-STAT5a) or together (S726A+S780A-

STAT5a) significantly decreased MCF7 colony growth in soft agar measure by both colony number and size (Figure 4.3). Further, the double point phospho-mutants significantly differed from each other in colony number. MCF7 cells expressing the phospho-serine double mutant,

S726A+S780A-STAT5a, formed the least number of colonies in soft agar compared to MCF7 cells expressing the other two phospho-mutants. Alternatively, MCF7 cells expressing Y694F+S780A-

STAT5a formed the most colonies compared to the MCF7 cells expressing the other phospho- mutants. In fact, these cells were the least significantly different from WT-STAT5a in size and number (p≤0.04).

106

Figure 4.3 Loss of either STAT5a serine residue in combination with loss of Y694 or loss of both serine residues decreases MCF7 ability to grow in soft agar as measured by both colony number and colony size. Representative images and quantification of number and size of the colonies formed in soft agar by indicated MCF7 cells with knockdown (KD) of STAT5a rescued by double phospho-deficient mutants. n=3, *p≤0.04, ***p=0.0002, ****p<0.0001 compared to WT-STAT5a.

Metastatic cancer is responsible for 90% of human cancer deaths.217 Invasion and metastasis rely on cancer cells ability to migrate from the primary tumor. Cell migration is easily studied in vitro using a scratch wound assay with real-time imaging.221,222 12 hours after establishing a scratch wound, MCF7 cells expressing the phospho-serine double point mutant,

S726A+S780A-STAT5a. had migrated significantly further than WT-STAT5a (Figure 4.4). These

107

cells continued to migrate further than WT-STAT5a, or any other cell type, up to 30 hours after the scratch wound was established.

Figure 4.4 Loss of both STAT5a serine residues in combination expedites MCF7 migration. MCF7 cells expressing indicated STAT5a phospho-deficient mutants were subjected to scratch wound assay and imaged at indicated time points for fraction of wound closure. n=3, *p=0.02, **p≤0.002, ***p=0.0004 compared to WT-STAT5a.

4.4. Discussion The data presented in this chapter indicate cooperative actions for the STAT5a phospho- sites that result in distinct phenotypic breast cancer characteristics. Whereas loss of Y694 and S726 decreased anchorage-independent growth, this combination of loss did not affect proliferation or migration of MCF7 cells. Loss of Y694 and S780 affected anchorage-independent growth, as well as proliferation. However, loss of these phospho-sites did not affect MCF7 cell migration. Loss of both phospho-serine sites, S726 and S780, decreased anchorage-independent growth and proliferation, and increased the migratory ability of MCF7 cells. Additionally, loss of both

108

phospho-serine residues led to the greatest decrease in MCF7 cells establishing colony number in an anchorage-independent manner.

Overall, this data indicates that loss of both phospho-serine residues modulates STAT5a activity leading to decreased tumorigenesis, and that intact phosphorylation of S726 and S780 increases tumorigenesis. However, the migration data indicate that intact phosphorylation of S726 and S780 on STAT5a decreases breast cancer cells migratory programs. These data suggest that these residues cooperatively modulate STAT5a activity at different gene loci responsible for different tumorigenic pathways. Moreover, these results are similar to the results in the hematopoietic studies which showed decreased leukemic potential both in vitro and in vivo.169,178

It would be of interest to perform IHC on a TMA panel that includes matched primary and metastatic tissue examining the concordance of pS726/pS780 expression to determine if migration and therefore metastatic potential is indeed decreased with these STAT5a phospho-sites intact.

Previous literature has shown when STAT5a lacks both S726 and S780, it is still capable of being phosphorylated at residue Y694.130,169,170,175 Therefore, because the canonical pathway of

STAT5a tyrosine phosphorylation is not affected, future mechanistic studies can focus elsewhere.

Complex formation of STAT5a and other transcriptional activators and repressors will give insight into how serine phosphorylation regulates STAT5a interactions in the nucleus. Additionally, RNA- sequencing, and confirmatory qRT-PCR, coupled with ChIP-sequencing, will better determine how STAT5a actions are modulated by phospho-serine and phospho-tyrosine at the transcriptional level in an enhancer/promoter dependent context.

Integration of the single-point and double-point STAT5a phospho-deficient mutant data was not performed due to technical limitations. The rescue of STAT5a in MCF7 cells as WT-

STAT5a was performed alongside the single- and double-point STAT5a mutants, ultimately

109

leading to two WT-STAT5a MCF7 cell lines. Thus, analyses comparing the STAT5a mutants to

WT-STAT5a was performed separately for the single-point STAT5a mutants and double-point

STAT5a mutants.

This study was limited to in vitro characterization of the double point phospho-mutants of

STAT5a. Therefore, it is limited in the conclusions we can draw about how these phospho-serine residues affect STAT5a in a physiologic setting. Thus, future studies in vivo are warranted, with implantation of the MCF7 cells expressing the STAT5a double point phospho-mutants in the mammary gland. An interesting, and warranted study, for the development of spontaneous metastasis with IHC examination of tissue from the primary and, if applicable, metastatic tumors.

As a corollary to the TMA proposed above, this would allow for the study of whether the double phospho-serine STAT5a mutants influence on migration is sufficient to drive metastasis in vivo.

110

5. Chapter 5: Global transcriptomic changes with expression of STAT5a single-point mutants by RNA-sequencing

5.1. Introduction PRL-induced gene expression changes have largely been assumed to necessitate only the pY694 residue of STAT5a, but not the pS726 or pS780 residues. This assumption has been upheld by studies of phospho-point mutant STAT5a transcriptional activity limited to utilizing readouts relying solely on the STAT5a GAS consensus sequence or the canonical β-casein gene reporter assays (i.e. transfection-based luciferase activity assays or EMSAs).127,170,175,179 In fact, this assumption has not been directly assessed by qRT-PCR, or, more globally, RNA-sequencing

(RNA-seq), chromatin immunoprecipitation (ChIP), or ChIP-sequencing.85,86,136,213–216 When studies of global transcriptomics have been undertaken, the approach often employed relies either on pharmacologic inhibition or total knockdown of STAT5a, therefore focusing on the role of pY694-STAT5a and ignoring the gradation of STAT5a activity that may occur in breast cancer and other pathologies.86,213,214 This has led to a general lack of understanding for the biology of endogenous upY694-, upS726-, or upS780-STAT5a.

The phenotypic characteristic changes observed in MCF7 cells harboring STAT5a serine phospho-mutants in outlined in Aim 1 led us to hypothesize that the phospho-serine residues on

STAT5a may regulate PRL-induced gene changes throughout the transcriptome. Therefore, in this study we performed RNA-seq of MCF7 cells expressing single-point phospho-deficient STAT5a mutants treated with PRL in attempts to identify differentially expressed genes, functional pathways for cancer outgrowth, and changes in transcription factor programs in response to loss of STAT5a phospho-sites. Additionally, this data was integrated with RNA-seq data from patient derived xenographs (PDX) and The Cancer Genome Atlas breast cancer samples (TCGA-BRCA)

111

to determine if phospho-mutants of STAT5a were sufficient to significantly change the intrinsic molecular subtype of the MCF7 cells harboring these mutants.197

5.2. Hypothesis

We hypothesize that STAT5a phosphorylation of S726 and/or S780 affect STAT5a transcriptional activity in response to PRL and that loss of these phospho-sites will result in global transcriptomic changes that are independent from the loss of the Y694 residue.

5.3. Results

Stable MCF7 single point phospho-mutant cells described in Aim 1 (Chapter 3) were stimulated with PRL (250 ng/ml) and subjected to RNA-seq to assess global transcriptomic changes following STAT5a mutagenesis.

112

Table 5.1 Significant differentially expressed genes (DEGs) for indicated contrast analyses performed in EdgeR filtered for counts per million (CPM)>0.5 to filter lowly expressed genes in the dataset library. No filter for fold change was applied.

Contrast (PRL vs 0) # Genes

WT 1906 EV 4987 S726A 1795 S780A 315

Y649F 1365 Contrast (Mut_PRL-Mut_0)-(WT_PRL-WT_0) # Genes

EV 219 S726A 1431 S780A 159 Y649F 1464 Contrast (Mut_0 to WT_0) # Genes

EV 1497 S726A 4637

S780A 996 Y649F 3621

Contrast (Mut_PRL vs WT_PRL) # Genes EV 2947 S726A 3010

S780A 1186 Y649F 800

Multiple analyses were performed prior to filtering differentially expressed genes (DEGs) based on fold change of at least 1.5, as seen in Table 5.1. However, a threshold for 1.5-fold change allowed us to filter biologically relevant PRL-induced gene expression changes for each STAT5a mutant (Table 5.2). Following PRL treatment, MCF7-WT-STAT5a had 510 differentially expressed genes (DEGs); specifically, 235 genes were upregulated and 275 genes were downregulated (Figure 5.1, Table 8.1). Each STAT5a phospho-deficient mutant had fewer DEGs

113

than the WT-STAT5a. MCF7 cells expressing Y694F-STAT5a had 171 DEGs: 90 genes were upregulated and 81 were downregulated. (Table 8.2) MCF7-S726A-STAT5a had the most DEGs of the mutant-carrying MCF7s: 285 DEGs total, of which 178 genes were upregulated and 107 genes were downregulated (Table 8.3). MCF7-S780A-STAT5a cells had 208 DEGs: 182 genes were upregulated and 26 were downregulated (Table 8.4). Importantly, when compared to MCF7 cells expressing WT-STAT5a, most of the DEGs in the MCF7 cells expressing the STAT5a mutants were distinct (Figure 5.1).

Table 5.2 Differentially expressed genes determined by RNA-seq analysis of STAT5a-rescued MCF7 cells treated with PRL. Included are the PRL vs untreated contrast filtered for fold change of at least 1.5. PRL treatment vs Upregulated Genes Downregulated Genes untreated

WT-STAT5a 235 275 Y694F-STAT5a 90 81 S726A-STAT5a 178 107 S780A-STAT5a 182 26

114

Figure 5.1 Significant differentially expressed genes in MCF7 cells expressing STAT5a phospho-point mutants are distinct from WT-STAT5a expression. Venn Diagrams of upregulated DEGs (A), and downregulated DEGs (B).

115

Quality control on this dataset was performed in several aspects. First, the STAT5a phospho-point mutations were confirmed utilizing RNA-seq BAM files. BAM files are output files from the bioinformatics analysis pipeline that contain aligned sequences. Therefore, cropped BAM files containing the sequencing for chromosome 17 were aligned to the human reference genome

(GRCh38/hg38, Dec 2013) in Integrated Genome Browser (IGB 9.1).189 Notably, the expected point mutations were intact in the sequenced reads for each STAT5a mutant (Figure 5.2).

Specifically, for the amino acid mutation Y694F, the adenine (A) nucleotide was replaced with a thymine (T; Figure 5.2A). For the amino acid mutations S726A and S780A, the T nucleotide was replaced with a guanine (G; Figure 5.2B, C, respectively).

116

Figure 5.2 Confirmation of STAT5a point mutations by RNA-sequencing. BAM files containing sequencing reads for STAT5a were aligned to chromosome 17 in Integrated Genome Browser (IGB) for STAT5a sequence. Sequenced reads were upload to IGB and aligned to the human reference genome, with labels for the reads on the left of the snipped screenshots. A) Y694F- B), S726A-, and C) S780A-STAT5a mutants reads aligned with chromosome 17 show expected genetic point mutations for each STAT5a mutant.

117

RNA-seq was also performed on the MCF7 cells harboring the empty vector (EV) as a control for this experiment. As shown in Figure 5.3, Ingenuity Pathway Analysis (IPA, Qiagen) of the EV RNA-seq dataset showed decreased measurement of PRL-induced STAT5a-regulated genes, and therefore predicted inhibition of STAT5a activity. Indeed, STAT5a expression in the

MCF7 EV cells was 6-fold less than MCF7 cells expressing WT-STAT5a (p=1.04x10-17, false discovery rate [FDR]=1.60x10-13). To further affirm quality of the dataset, WT-STAT5a PRL- induced or inhibited genes were compared to previous PRL-regulated transcriptomic data released by our laboratory, such that after matching the criteria that the DEG be present in at least two datasets, 105 putative PRL-regulated genes were identified (Supplemental Table S2,

Appendix).70,78 As shown in Chapter 3, a subset of these genes were used to identify functionally- enriched pathways with PRL treatment that may be changed by expression of the STAT5a mutants through IPA (Figure 3.3B, Chapter 3). DEGs from contrasting untreated MCF7 cells harboring

EV to untreated MCF7 cells expressing WT-STAT5a were overlaid on the IPA pathway and showed inhibition of 7 of the 13 queried pathways, including cellular transformation, colony formation, proliferation, migration, and cell cycle progression (Figure 5.4A). Inhibition of 10 of the 13 pathways was seen when MCF7 cells harboring EV and treated with PRL were compared to MCF7 cells expressing WT-STAT5a treated with PRL (Figure 5.4B). Further, this contrast showed predicted activation of pathways leading to apoptosis of breast cancer cell lines and cell death. Overall, MCF7-EV cells had decreased predicted tumorigenicity compared to cells expressing WT-STAT5a.

118

Figure 5.3 STAT5a signaling is predicted to be downregulated in MCF7 cells expressing the empty vector (EV) control by Ingenuity Pathway Analysis (IPA). As illustrated here, decreased expression of STAT5a target genes was used to predict that STAT5a activity in EV control cells was downregulated.

119

Figure 5.4 IPA functional pathway enrichment analysis of MCF7s expressing EV compared to WT-STAT5a largely shows inhibition of chosen pathways. IPA functional pathway analysis of the DEGs comparing untreated EV with untreated WT-STAT5a (A) or comparing PRL-treated EV with PRL-treated WT-STAT5a (B) overlaying the regulatory network shown in to predict activation or inhibition. p value calculated by Fisher’s exact test.

120

As stated previously, multiple contrasts between the MCF7 RNA-seq datasets were performed to determine DEGs across the various conditions in this study. One contrast resulted in differentially expressed genes by comparing mutant PRL-regulated changes to STAT5a PRL- regulated changes, ultimately illustrating the change of PRL-regulated gene expression with

STAT5a phospho-mutants. This contrast for each STAT5a mutant, namely (Mut_PRL-Mut_0)-

(WT_PRL-WT_0) in Table 5.1, was explored in IPA as well. It is important to note that this contrast uses the untreated control for each mutant compared to WT-STAT5a, since baseline these cells may express genes differentially as well. As shown in Figure 5.5A, expression of the Y694F-

STAT5a mutant in MCF7 cells caused increased expression of BCL6 compared to WT-STAT5a in the presence of PRL. With PRL treatment, BCL6 expression is decreased with WT-STAT5a.

Further, CISH expression was decreased from WT-STAT5a. Together, these changes in expression illustrate the role of loss of pY694 on STAT5a canonical function. These findings were confirmed by qRT-PCR as illustrated in Chapter 3 (Supplemental Figure S3). Interestingly, each mutant in the presence of PRL had an opposite effect on all 13 queried functional predicted pathways compared to WT-STAT5a. However, each STAT5a phospho-mutant changed the expression of the illustrated PRL-inducible DEGs to different degrees (as based on shades of measurement and activation), showing that the STAT5a phospho-sites do not affect PRL-regulated gene expression equivalently. Overall, using this subset of putative PRL-regulated DEGs derived from experiments with endogenous STAT5a expression may only reflect canonical PRL/STAT5a-regulated gene expression and may not elucidate the functions of the phospho-serine residues.

121

Figure 5.5 Predicted functional effects of the change in magnitude in PRL-induced genes in MCF7 cells expressing the STAT5a mutants compared to WT-STAT5a. DEGs of PRL stimulated Y694F- (A), S726A- (B), and S780A-STAT5a (C) compared to the DEGs of PRL stimulated WT-STAT5a illustrate overall decrease in PRL-responsive functional pathways.

Gene set enrichment analysis (GSEA), is a powerful tool for analyzing molecular profiling data to gain insights into biological mechanisms that may be underlying RNA-seq results.195

Various GSEA were performed on the PRL-stimulated gene expression of each STAT5a mutant to determine if and how STAT5a mutations were functionally changing cellular processes. Gene sets were obtained from the Molecular Signatures Database (MSigDB). These gene set signatures

122

were generated from microarray data from NCBI GEO, internal unpublished profiling experiments involving perturbation of known cancer genes, or from scientific publications.195,223

The oncogenic signatures (OS) collection of gene sets represents signatures of cellular pathways often dysregulated in cancer. In the presence of PRL, MCF7 cells expressing WT-

STAT5a were enriched in 18 OS gene sets, while MCF7 cells expressing Y694F-STAT5a had 19 enriched OS gene sets, and MCF7 cells expressing S726A-STAT5a had 30 enriched OS gene sets

(FDR<0.25). None of the OS gene sets for MCF7 cells expressing S780A-STAT5a met the FDR criteria for significance, however, 6 OS gene sets were significant at p<0.05 (Figure 5.6).

123

Figure 5.6 MCF7 cells expressing S726A-STAT5a or Y694F-STAT5a share significant enriched oncogenic signatures. Venn Diagram comparing enriched oncogenic signature (OS) gene sets for MCF7 cells expressing WT-STAT5a or mutants of STAT5a in the presence of PRL. These gene sets represent signatures of cellular pathways often dysregulated in cancer. Significantly enriched OS for WT-, Y694F-, and S726A-STAT5a FDR <0.25, for S780A-STAT5a p<0.05.

All ten of the top 10 significantly enriched OS for MCF7 cells expressing WT-STAT5a and treated with PRL were unique, that is, none of these enriched OS appeared in the top 10 for the MCF7 cells expressing STAT5a phospho-mutants (Table 8.5). Further, of the top 10

124

significantly enriched OS for MCF7 cells expressing WT-STAT5a in the presence of PRL, 5 of these terms (50%) involved upregulation of the KRAS pathway. Conversely, MCF7 cells expressing Y694F-STAT5a treated with PRL shared 9 out of the top 10 enriched OS with either

S726A-STAT5a or S780A-STAT5a, and the only unique OS for these MCF7-Y694F cells was

STK33_SKM_UP (normalized enrichment score [NES]=1.76, FDR q value=0.009).

MCF7 cells expressing S726A-STAT5a shared 8 out of 10 of the enriched OS with either

Y694F-STAT5a or S780A-STAT5a, and was uniquely enriched for LTE2_UP.V1_DN

(NES=1.71, FDR q value=0.017) and EGFR_UP.V1_UP (NES=1.73, FDR q value=0.019).

Interestingly, these two gene sets were generated in the same study and share only 10 genes in common out of about 195 genes each.224However, 4 of these shared genes make up the leading edge of both sets, that is, a subset of genes in a gene list that account for the gene set’s enrichment signal.195 Therefore, the significant enrichment of both gene sets in the MCF7-S726A-STAT5a cells treated with PRL must depend on the other core genes (numbering ~70) in the leading edge subset for these gene sets.

MCF7 cells expressing S780A-STAT5a only shared 4 out of 10 of the enriched OS with

Y694F-STAT5a or S726A-STAT5a, however these 4 shared OS were the most significant in the top 10 for S780A-STAT5a (FDR<0.89, p<0.023). One OS, PIGF_UP.V1_UP, was shared amongst all three STAT5a phospho-mutants in the presence of PRL (NES≥1.58, S780A-STAT5a p=0.001,

S726A-/Y694F-STAT5a FDR q value≤0.008) for as well as the untreated MCF7 cells expressing

WT-STAT5a (Tables 7.5 and 7.6). For the STAT5a phospho-mutants treated with PRL, the majority of the leading edge genes in the PIGF_UP.V1_UP gene set were shared amongst all three phospho-mutants, indicating that all three phospho-sites cooperatively regulate these genes.

125

Utilizing the open-source Enrichr web-based platform, we were able to examine how the

STAT5a phospho-mutants affect PRL-responsive transcriptional machinery.194 Enrichr includes gene set lists that are putative targets for transcription factors (TF) which have been profiled by

ChIP-seq. Specifically, Enrichr uses data from the Encyclopedia of DNA Elements (ENCODE) project where peaks were first identified in the aligned files of 646 experiments profiling TF in mammalian cells, and then sorted by distance to the transcription factor start site (TSS). The top

2000 target genes for each ChIP-seq experiment were retained as gene sets in the library.194,225

Hierarchical clustering of the odds ratios (OR) of the 812 ENCODE transcriptional gene sets for the DEGs from MCF7 cells expressing WT-STAT5a or the phospho-mutants in the presence of

PRL showed distinct transcriptional patterns (Figure 5.7). Normalized ORs determined that

S726A-STAT5a and Y694F-STAT5a were enriched in similar TF gene sets. Whereas S780A-

STAT5a and WT-STAT5a enriched similar TF gene sets, there was a distinct proportion of TF gene sets in which the normalized ORs for the two were opposite on the -1 to +1 scale (Figure

5.7). For WT-, Y694F-, and S726A-STAT5a, the top 10 enriched TF gene sets all had p≤0.05 and

OR≥1.39 (Table 8.7). The top 10 enriched TF gene sets for S780A-STAT5a were p≤0.07 and

OR≥1.29 (Table 8.7). The top 10 enriched TF gene sets were examined in Figure 3.7 of Chapter

3 (Table 8.8). STAT5a was confirmed as one of the top transcriptional regulators for PRL-treated

WT-STAT5a MCF7 cells (OR 2.05, p=0.03; Figure 3.7), and was also enriched significantly in

S726A-STAT5a MCF7 cells (OR 1.21, p=0.007). However, STAT5a was not identified as a top significant TF with PRL treatment for Y694F- or S780A-STAT5a MCF7 cells. Further analysis of the top 10 enriched TF gene sets showed that S726A- and Y694F-STAT5a MCF7 cells shared

6 out of 10 TF gene sets as seen in Figure 5.8, similar to what was seen on a larger scale in Figure

5.7. The one unique enriched TF for PRL-stimulated MCF7 cells expressing S726A-STAT5a was

126

ESSRA (OR 1.59, p=7.74E-09), whereas cells expressing Y694F-STAT5a were uniquely enriched in TF THAP1, IRF3, and FOSL1 (OR ≥ 1.59, p≤2.88E-04). MCF7 cells expressing S780A-

STAT5a had the least enriched TF in common with WT-, Y694F-, or S726A-STAT5a, only sharing TF SP1 with S726A-STAT5a and TF FOXA1 with WT-STAT5a (Table 8.7). Further, these S780A-STAT5a cells were most significantly and uniquely enriched for STAT3 as a TF (OR

1.37, p=0.02).

Figure 5.7 Transcription factor (TF) enrichment shows distinct transcriptional patterns in MCF7 cells expressing STAT5a mutants. Hierarchical clustering of the odds ratios for 812 enriched ENCODE ChIP-seq project transcription factor terms from RNA-seq analysis of PRL treated versus untreated MCF7 cells expressing STAT5a rescues.

127

Figure 5.8 MCF7 cells expressing S726A-STAT5a and Y694F-STAT5a share the most enriched transcription factor programs in the presence of PRL. Venn Diagram of the top 10 enriched ENCODE project TF terms shows distinct and shared TF terms in the various STAT5a species.

As a breast cancer cell line, MCF7 cells are classified as luminal A/B.4,65 To determine if expression of phospho-point mutants of STAT5a in MCF7 cells could cause intrinsic subtype switching, the RNA-seq data were merged with mammary gland tumor (MGT) transcripts from

PDXs as well as 817 TCGA breast cancer samples as described in Alzubi et al.197 Expression of the top 5% most variable genes was used to determine Pearson’s sample-wise correlations for the

128

MCF7 dataset and the PDX samples (Figure 5.9A). The MCF7 STAT5a samples correlated closely with the PDX MGT samples, ranging from 0.91-0.96. However, there was no pattern of greater correlation of the MCF7 samples to luminal B PDX samples (that is, HCI03, HCI09,

HCI11, HCI13) compared to the basal-like PDX samples (HCI01, HCI02, HCI04, HCI10, HCI16,

UCD18, UCD52, WHIM2, WHIM30; Figure 5.9A).197 Expression of 1,770 genes related to breast cancer was used to determine Pearson’s sample-wise correlations for the MCF7 dataset and PDX

MGT in addition to RNA-seq from 817 TCGA-BRCA samples (Figure 5.9B).197,206,226 The MCF7

STAT5a samples ranged from about 0.73-0.94 in correlating to the TCGA-BRCA and PDX samples in this comparison (Figure 5.9B). The samples that least correlated to the MCF7 samples included one luminal B PDX (HCI13), three TCGA-BRCA triple negative breast cancers, and three TCGA-BRCA samples of histologically mixed invasive ductal carcinoma that have reported areas of small cell neuroendocrine carcinoma.227,228

129

Figure 5.9 Pearson’s sample-wise correlation analyses of MCF7 expressing STAT5a phospho-mutants RNA-seq and RNA-seq from PDX tumors and metastases and the Cancer Genome Atlas (TCGA) datasets. A) The top 5% most variable gene expression levels from the MCF7 dataset and the PDX dataset from Alzubi et al were used to determine Pearson’s sample- wise correlation .197 B) 1,770 genes related to breast cancer were used to determine correlation between the MCF7 dataset and a dataset consisting of 817 TCGA-BRCA samples and PDX samples as described in Alzubi et al.197 Black bars to the left of the analyses indicate MCF7 samples.

This combined dataset was then analyzed further for PAM50 gene expression (Figure

5.10).197,226 The PAM50 is used to define the four intrinsic subtypes of breast cancer (i.e., luminal

A/B, HER2-enriched, and basal-like) based on differential expression of 50 genes, has been shown to be predictive of treatment benefits and recurrence risk, and has been used to characterize breast cancer cell lines.4,206,226,229 Figure 5.10A shows hierarchical clustering of the full dataset. Figure

5.10B, a magnified view of the heatmap surrounding the MCF7 dataset, illustrates how the MCF7

STAT5a samples (luminal A) hierarchically cluster together closely with luminal A TCGA-BRCA samples. Altogether, the data presented in Figures 5.9 and 5.10 illustrate how STAT5a phospho-

130

point mutants do not cause a large enough change in gene expression to elicit an intrinsic subtype switch for MCF7 cells.

Figure 5.10 MCF7 cells expressing STAT5a mutants hierarchically cluster together when combined with RNA-seq data from PDX mammary gland tumors and data from 817 tumors from the Cancer Genome Atlas (TCGA).197 The PAM50 genes were used to hierarchical cluster the combined dataset.4,197,206 A) Complete combined dataset. B) Magnified view of the boxed area in (A) surrounding the MCF7 dataset.

5.4. Discussion

This study was limited by the low number of biological replicates performed. While previous genomic studies in our laboratory using microarray successfully identified DEGs with three replicates, recommendations for RNA-seq call for at least six replicates per condition.230 This may have led to downstream issues with gene expression quantification.

Technical bias may have been introduced by mRNA library construction using polyadenylated (poly(A)) RNA selection. Although poly(A) selection is known for its low-noise rate, the protocol enriches poly(A) transcripts including mRNA and non-coding RNAs.231 The

131

other library selection method, ribosomal RNA (rRNA) depletion, selectively removes rRNA, enriching samples with the poly(A) mRNA, protein-coding mRNAs that are non-poly(A), and non- coding RNAs.232 These different RNA sample preparation methods may lead to significant variations in quantification of gene expression.231,233 In the case of our RNA-seq dataset, the ribosomal 5.8S rRNA was found to be a significant confounder in analysis of DEGs. While FastQ screening did not show any contamination from this rRNA, review of the BAM file using IGB showed about 80,000 reads aligning to the rRNA region. Therefore, it is likely that rRNA contamination of our sequencing samples led to lower read counts for protein-coding mRNA during sequencing. This led to downstream issues with analysis, specifically low read counts that subsequently decreased the power of identifying significant DEGs. To partially overcome this limitation, rRNA genes were removed from the gene matrix by Amy Olex, collaborator and bioinformatician. However, this issue ultimately led to the use of p values for identifying significance rather than the false discovery rate. To confirm significant DEGs, an analysis comparing past genomic datasets from our laboratory to this dataset was performed

(Supplementary Table S3, Appendix). Consistency across these datasets overall allowed for confidence in our data.

In this chapter, we examined RNA-seq of MCF7 cells expressing single-point phospho- deficient STAT5a mutants treated with PRL in attempts to identify differentially expressed genes and functional pathways for cancer outgrowth. While expression of the STAT5a phospho-mutants had less PRL-induced DEGs compared to WT-STAT5a, the DEGs were mostly distinct, and led to differences in the functional pathways analyzed.

Quality of the RNA-seq dataset was performed both by analysis of BAM files for intact

STAT5a point mutations and examination of MCF7 cells expressing EV. While examination of

132

BAM files confirmed STAT5a point mutations, study of EV showed 6-fold decrease in STAT5a expression and an overall decrease in tumorigenicity compared to WT-STAT5a when DEGs and functional downstream pathways of STAT5a were analyzed.

Comparison of the WT-STAT5a dataset to previous microarray and RNA-seq studies identified 105 putative PRL-regulated DEGs.70,78,87 These DEGs were used to analyze how the

STAT5a phospho-mutants changed the expression of PRL-regulated genes in IPA and led to predicted changes in the activation or inhibition of upstream regulators and downstream functional pathways. However, this gene set, derived from experiments with WT-STAT5a expression, may reflect only canonical PRL/STAT5a-regulated gene expression, and consequently be a fraction of the overall affect the phospho-serine residues have on STAT5a transcriptional function.

To this point, single-gene analysis may miss effects on pathways, as cellular processes often affect sets of genes acting in concert.195 That is, smaller increases or decreases in all the genes encoding members of a specific cellular process may dramatically change the biology of the pathway and may be more important than a large increase in a single gene. Therefore, GSEA for oncogenic signatures was undertaken to examine how the STAT5a phospho-mutants may affect cellular pathways dysregulated in cancer. Importantly, the top 10 most significantly enriched OS for PRL-treated WT-STAT5a were unique, and 50% of these terms involved the upregulation of the KRAS pathway. Further, the unique enriched OS for PRL-treated Y694F-STAT5a was

STK33_SKM_UP, a gene set which consists of upregulated genes after the knockdown of the serine/threonine kinase STK33.234,235 STK33 is required for survival and proliferation of KRAS- dependent cancer cells.234,235 Therefore, expression of Y694F-STAT5a functionally upregulates the same genes that are upregulated upon knockdown of STK33, showing that transcriptional activity of pY694-STAT5a affects the KRAS pathway. These findings contribute to the growing

133

literature surrounding the cooperative role for KRAS and PRL in breast cancer pathogenesis. Most recently, Grible et al found that PRLr long and intermediate (PRLrL+I) hetero-dimers cooperate with KRAS signal transduction to promote cellular transformation.(Grible et al, in press 2021)

Further, with GSEA, Grible et al uncovered a significant association of expression of oncogenic

KRAS in an analysis of TCGA PRLrIhi/PRLrLlo breast cancer samples.(Grible et al, in press 2021)

In another study of a PRL-driven mammary cancer mouse line, genomic analyses found activating mutations and copy number amplifications of KRAS in established tumors that were absent in preneoplastic tissues.75 These studies illustrate the need to characterize the complex interplay between the PRL/STAT5a pathway and KRAS pathway.

MCF7-S726A-STAT5a cells were uniquely enriched for LTE2_UP.V1_DN and

EGFR_UP.V1_UP, gene sets which were derived from the same study.224 LTE2_UP.V1_DN is a set of genes downregulated in MCF7 cells that have been adapted for long-term estrogen- independent growth. EGFR_UP.V1_UP consists of a set of genes upregulated in MCF7 cells which, through expression of ligand-activatable EGFR, have hyperactivation of the MAPK pathway, and have adapted to estrogen-independent growth with loss of ERα expression.224

Interestingly, while the gene signature of upregulated genes in estrogen-independent MCF7 cells derived from overactive growth factor signaling is different from the up- or downregulated genes in MCF7 cells which have been cultured to adapt to long-term estrogen-independent growth,

S726A-STAT5a MCF7 cells treated with PRL have genes upregulated and downregulated that fit both.224 Therefore, while enrichment of these two gene sets seems contrary, altogether they point to a bigger picture: that PRL-induced phosphorylation of S726-STAT5a may contribute to MCF7 estrogen-dependent growth. In fact, PRL has been shown to play a role in regulating and enhancing

ERα expression in breast cancer and increases estrogen responsiveness in breast cancer cells.67

134

Additionally, PRL can induce ER transcriptional activity in breast cancer cells independent of estrogen ligand binding ER, as shown by induction of the ER-target gene pS2.69 Thus, loss of pS726-STAT5a may in fact mimic loss of ER on a transcriptomic scale.

The enriched OS that were shared amongst at least two of the STAT5a phospho-mutants show how these phospho-residues cooperatively regulate STAT5a activity. To that point, the OS

PIGF_UP.V1_UP was enriched in all three STAT5a phospho-mutants when treated with PRL, as well as in untreated WT-STAT5a. The PIGF_UP.V1_UP gene set consists of genes upregulated in human umbilical cord vein endothelial cells (HUVECs) when treated with the angiogenic factor placental growth factor (PIGF).236 Convergence on this pathway may indicate cooperative regulation of STAT5a activity by all three phospho-sites. Increased circulating levels of PIGF have been found to be associated with nodal and distant metastasis in breast cancer, and increased tissue expression of PIGF is found in TNBC cases compared to other intrinsic subtypes.237 Therefore, when the phospho-sites of STAT5a are intact, their cooperative role regulating gene expression in this case may be antagonistic to PIGF-associated tumorigenicity.

In order to understand how globally the transcriptional machinery may have been affected by loss of each of the STAT5a phospho-sites, in silico TF gene set enrichment analysis was performed. Our data from this analysis indicates that an array of TF families may act specifically as compensation to loss of pY- or pS-STAT5a function. For example, loss of phosphorylation of

S726 resulted in one unique enriched TF, ESSRA. ESSRA is the gene encoding estrogen-related receptor α (ERRα, NR3B1). Intriguingly, through competitive genomic crosstalk, ERRα interferes with ERα repression of ERBB2, the gene encoding HER2.238 ERRα regulates HER2 expression, and transcriptional activity of ERRα is positively modulated by EGFR/ERBB2 signaling in a feed- forward loop.238 This competitive crosstalk exhibited by increased ERRα transcriptional activity,

135

and therefore associated decrease in ERα activity, along with the finding that S726A-STAT5a cells were enriched for estrogen-independence (by GSEA) bolsters the concept that loss of phosphorylation of S726-STAT5a mimics the loss of ERα.

Similarly, this study also indicated that the SREBP TF family was involved in both Y694F- and S726A-STAT5a PRL-regulated gene expression changes. As a TF, SREBP plays a critical role in breast cancer migration and invasion, and its involvement in conjunction with the loss of pY694 or pS726 may indicate a role in oncogenic signal switches to overcome the loss of

STAT5a.220 Activation of SREBPs are required for growth factor-independent growth and proliferation, which could explain our phenotypic findings with our functional studies. Lastly,

SREBP acts downstream of oncogenic PI3K or KRAS.239 Along with the findings of enriched OS, this finding illustrates the complexity of transcription factor activity in breast cancer, and the homeostatic mechanisms that may be at play for oncogenic growth.

Finally, loss of pS780-STAT5a resulted in enrichment for STAT3. It has been shown previously that STAT5 and STAT3 mediate opposing effects on target gene expression and cellular phenotypes in breast cancer.240–242 This finding reported here indicates that phosphorylation of the

S780 residue modulates the balance between STAT5 and STAT3 transcriptional activity.

Overall, when taken together these findings indicate that PRL-stimulated STAT5a transcriptional activity is affected differentially by loss of each phospho-site. Further, signaling pathways that have been implicated acting in cooperation with PRL rely on the phospho-sites of

STAT5a to various degrees. For instance, GSEA of oncogenic signatures indicates that transcriptional activity of pY694-STAT5a could affect the KRAS pathway. That is to say, loss of pY694-STAT5a enriches a gene set in a manner similar to loss of STK33, a protein that is required for KRAS-dependent cancer cells for proliferation and survival.234,235 This indicates that in cancers

136

expressing upY694-STAT5a that is still phosphorylated on pS726 and pS780, there is decreased

KRAS pathway activity. Conversely, loss of pS726-STAT5a enriched gene sets which indicate that intact phosphorylation of S726-STAT5a may contribute to MCF7 estrogen-dependent growth.

Concurrently, loss of pS726-STAT5a leads to enrichment for an ERRα signature, a TF that is known to compete with ERα, indicating that phosphorylation of S726-STAT5a may reinforce ERα transcriptional activity. Finally, as discussed above, phosphorylation of S780-STAT5a may modulate the balance between STAT5 and STAT3. Thus, while each phospho-site has distinct roles in regulating transcriptional machinery, they also act together to regulate STAT5a activity as seen in the GSEA where all three phospho-mutants were enriched in the PIGF gene set.

Inclusion of this dataset with RNA-seq from PDX MGT and TCGA-BRCA samples was done to determine if the changes seen on the transcriptional level translated to stratification of the samples in the intrinsic subtypes of breast cancer.197 Subtype switching and gene expression changes are seen in metastatic tumors compared to primary breast tumors, and have been reported after treatment.243–246 A distribution of tumor samples reflecting the different intrinsic subtypes is necessary to reveal the predicted intrinsic subtype using the PAM50 gene profile panel.4,206 Thus, hierarchical clustering of the integrated STAT5a phospho-mutant, PDX, and TCGA-BRCA datasets showed that the regardless of the phospho-point mutation, the STAT5a dataset clustered together and did not elicit an intrinsic phenotype switch.

In total, the RNA-seq results presented indicate that the phospho-sites of STAT5a affect

STAT5a-mediated gene expression differentially, with specific roles in regulating STAT5a activity in various oncogenic pathways. Further, loss of these phospho-sites elicits disparate TF programs. Therefore, it can be concluded that rather than the phospho-serine residues having only a role in modulating the canonical pY694-STAT5a transcriptional activity, phosphorylation of

137

these residues likely regulates STAT5a activity at different genomic regulatory elements. While these changes in gene expression were not large enough to elicit intrinsic subtype switching of the

MCF7 cells, it is possible that phosphorylation of the serine residues could be used as biomarkers for stratifying patients and determining outcomes. Future studies of immunohistochemical analysis for pS726- and pS780-STAT5a of primary tumors compared to metastases correlated to patient outcomes data would verify this hypothesis.

138

6. Chapter 6: Regulation of STAT5a serine phosphorylation by Prolactin

6.1. Introduction

PRL/PRLr signaling initiates the mitogen activated protein kinase/extracellular signal regulated kinase (MAPK/ERK) cascade concurrent to JAK2/STAT5a signaling, and the kinases in the MAPK pathway are known serine/threonine kinases.125 ERK kinases regulate the activity of transcriptional factors involved in the control of the cell cycle.34,247 ERK1/2 is known to target

PMSP consensus motifs. STAT5a residue S726 lies within a PSP motif, missing the methionine residue, and so may be a potential weak ERK1/2 target.127,169,174 Further, residue S780 is contained within a LSP motif, which may also act as a weak ERK1/2 target.170

Commonly, to establish if proteins interact, co-immunoprecipitation for protein complexes is performed. ERK1/2 and STAT5a have been shown to complex in several reports in the context of growth hormone (GH), a member of the same cytokine super family as PRL. In a study by

Pircher et al, CHO cells (PRLr/GHr-null) stably expressing GHr (the cognate GH receptor) were shown to have preformed complexes of ERK1/2-STAT5a which dissociated with GH stimulation.32 However ERK1/2:STAT complex formation or dissociation upon stimulation, as well as any interdependence of these two proteins, is controversial.248 In fact, Dinerstein-Cali and co-authors showed that upon GH stimulation, the association of ERK1/2 and pY- and pS-STAT5a was increased in CHO cells.34

Further work for understanding the relationship between ERK1/2 and STAT5a has utilized in vitro kinase assays as well as pharmacological inhibition of the MAPK pathway. Pircher et al established that ERK1/2, activated in response to GH, was capable of phosphorylating S780-

STAT5a using an in vitro kinase assay.32 The mitogen-activated protein kinase kinase (MEK)

139

inhibitor PD98059 has been used in various studies to determine if STAT5a serine phosphorylation was downstream of the MAPK pathway in various cell types upon cytokine stimulation.126–128

Inhibition of MEK inhibits ERK1/2 phosphorylation and activation, and therefore possible downstream targets.249 Interestingly, findings for inhibition of STAT5a serine phosphorylation may be cell-type and cytokine-specific. In Nb2 lymphocytes, PD98059 had no effect on IL-2 induced STAT5a serine phosphorylation nor its ability to bind DNA.126 In another study using Nb2 lymphocytes and PRL, pretreatment of the cells with PD98059 had no effect on inducible phosphorylation of S726, but decreased PRL-independent, constitutive serine phosphorylation of

STAT5a.127 This study, it must be mentioned, did not examine phosphorylation of S780, which we have found to be independent of PRL stimulation, and therefore may in fact be the constitutively phosphorylated serine residue in that study. However, in an oncogene RhoA-driven model of

Madin-Darby Canine Kidney (MDCK) cells also treated with PD98059, phosphorylation of S726 was decreased with no decrease in phosphorylation of S780.128

These studies are evidence of the controversial data that exists around STAT5a serine phosphorylation. Therefore, we set out to determine if ERK1/2 associated with STAT5a in preformed complexes or associated upon PRL stimulation, and if ERK1/2 phosphorylated STAT5a serine residues in breast cancer specifically.

6.2 Hypothesis

We hypothesize that ERK1/2 is responsible for phosphorylation of at least one serine residue of

STAT5a, and that ERK1/2 associates with STAT5a in PRL-responsive breast cancer cells.

140

6.3 Results

We first sought to determine if ERK1/2 and STAT5a associate in a PRL-dependent manner, and if this association depended on the presence of the STAT5a phospho-serine residues

S726 or S780. As seen in Figure 6.1, with immunoprecipitation targeting V5-STAT5a, STAT5a-

ERK1/2 complexes were present before PRL stimulation. After 30 minutes PRL treatment, there was an observable, albeit slight, difference in complex formation (WT-STAT5a). Interestingly, in the same co-immunoprecipitation experiment, there is a loss of ERK1/2 complexing with STAT5a when pS726 is lost (Figure 6.1A). Further, with loss of pS780, there is an increase in ERK1/2 complexing with STAT5a (Figure 6.1B).

Figure 6.1 ERK1/2 interact with STAT5a is dependent on presence of S726 but not S780. MCF7 cells rescued with A) S726- or B) S780-STAT5a or WT-STAT5a were immunoprecipitated with a V5-specific antibody and protein lysates were analyzed for V5-STAT5a, and total ERK1/2. WCE were probed for pERK1/2, total ERK1/2, V5-STAT5a, and vinculin loading control.

141

Our laboratory has previously shown that with PRL stimulation, ERK1/2 mediates phosphorylation of another serine/threonine kinase, NEK3.41 In those studies, the ERK1/2 inhibitor, U0126, was shown to specifically inhibit ERK1/2 and NEK3 phosphorylation in response to PRL. Therefore, we sought to determine if this inhibitor would inhibit phosphorylation of either STAT5a serine residue. U0126 treatment completely inhibited phosphorylation of

ERK1/2, as expected (Figure 6.2). However, neither the PRL-inducible phosphorylation of S726, nor the constitutive phosphorylation of S780 was inhibited. (Figure 6.2A and B, respectively).

Figure 6.2 Phosphorylation of S726 and S780 are independent of the MEK pathway. MCF7 cells were treated with the MEK inhibitor for 2 hours and then stimulated with PRL for 30 minutes. Analysis by WB for pS726 (A) or pS780 (B) was performed along with pERK1/2.

6.4 Discussion

Our findings in this Aim show that ERK1/2-STAT5a complexing is dependent on the presence of S726 but inhibited by S780. Further, we show that pharmacologic inhibition of the

MAPK pathway does not inhibit phosphorylation of either of these STAT5a serine residues.

Like Pircher and coauthors, we found that ERK1/2 and STAT5a are complexed prior to

PRL stimulation.32 However, there was only a slight dissociation of the complex after 30 minutes of PRL stimulation. ERK1/2 is known to have other targets which bind more strongly to inactive

ERK1/2 than active ERK1/2, including MAPK-interacting kinase 1 (Mnk1) and ribosomal S6

142

kinase (RSK).250,251 Therefore it is likely that STAT5a is also bound to inactive ERK1/2, with dissociation upon mitogen activation.

Intriguingly, the pattern of STAT5a-ERK1/2 association in the serine phospho-point mutants illustrates how this interaction relies differentially on both residues. The loss of pS726 decreased ERK1/2-STAT5a association, indicating that the presence of S726 may increase the affinity of this interaction. Conversely, the loss of S780 increased the association of ERK1/2-

STAT5a, consistent with previous literature, and suggesting that the phosphorylation of S780 may, in fact, decrease the affinity for STAT5a and ERK1/2 interaction.32

Using the MEK1/2 inhibitor U0126, we found no change in either PRL-stimulated pS726 nor constitutive pS780. Our results are most consistent with previous studies which found no change in either IL-2- or PRL-induced STAT5a serine phosphorylation when cells were treated with PD98059, another MEK inhibitor, in different cell types.126,127 While Pircher et al found that

ERK1/2 directly phosphorylated S780-STAT5a in an in vitro kinase assay, this may not be the case in the cellular environment.32 Our results, as well as the studies done by Kirken and

Yamashita, illustrate that the relationship that exists between ERK1/2 and STAT5a may be cell- and cytokine-specific.126,127

This study was biased and thus leaves the question of which kinase, in response to PRL, phosphorylates the STAT5a serine residue S726, unanswered. PRL-PRLr binding concomitantly activates the c-Src, JAK2/STAT5, PI3K/AKT, VAV2/RAC1, and MAPK signaling cascades.39,125

Several lines of work may point to the serine/threonine p21-activated kinase (PAK), downstream of RAC1. First, in 2005, Miller et al found that expression of the GEF VAV2 increased STAT5- mediated gene transcription that was independent of the serine/threonine kinase NEK3.39 This indicated that VAV2/RAC1 pathway may be involved in regulating STAT5.39 The VAV/RAC

143

pathway has been previously reported as upstream of phosphorylation of STAT3 on S727 in response to IL-6.252 Further, in a hematopoietic model of leukemogenic transformation, STAT5 transactivation was shown to be downstream of the focal adhesion kinase (FAK)/RAC1-

Tiam1/PAK1 axis, although this study examined only STAT5 nuclear translocation and did not examine the role serine phosphorylation has on nuclear translocation.253 In yet another model of

BCR-ABL-driven leukemogenesis, PAK was found to associate with STAT5 and phosphorylate

S779 (that is, S780).178 Finally, in a study of normal mammary gland development, PAK was found to be necessary for phosphorylation of S779 (S780) in mouse HC11 mammary cells and normal mammary development.254 However, this study did not examine the PRL-inducible serine phosphorylation for either S726 or S780. Therefore, it is apparent that there is a case for studying

PAK-STAT5a association and phosphorylation in response to PRL in breast cancer cells.

The results in this study also raise a new question: what is the significance of ERK1/2-

STAT5a association, if not for serine phosphorylation? As stated previously, the ERK1/2-STAT5a complex was preformed prior to PRL stimulation, with slight dissociation following PRL treatment. It has been previously reported that inactive ERK1/2 is in a complex with other kinases, including the serine/threonine kinase RSK, which is activated downstream of ERK1/2.251,255 RSK, and any other serine/threonine kinases downstream of ERK1/2 activity are effectively precluded from being the possible kinase responsible for STAT5a serine phosphorylation given our findings.

However, it is possible that ERK1/2 and STAT5a exist in a complex with other, as of yet unidentified, kinases that could phosphorylate STAT5a.

Our work was limited by using pharmacologic agents to determine if ERK1/2 phosphorylates STAT5a serine residues. Use of ERK1/2 genetic knockdown or knockout would confirm these findings. This aim took a biased approach targeted at confirming ERK1/2 as the

144

kinase responsible for phosphorylating the STAT5a serine residues. Future directions for this aim would be to take a non-biased approach and examine other PRL-induced serine/threonine kinases and their association with STAT5a. A kinase inhibitor assay with phospho-serine STAT5a as a readout would help to narrow the field of serine/threonine kinases that are active downstream of

PRL.

145

7. Chapter 7: Discussion

7.1. Discussion

While evidence for STAT5a having a functional role as a survival factor and promoter of breast cancer progression has been shown in cell lines and mouse models, in histologic examination of patient tumor samples, nuclear localized pY-STAT5a has been found to correlate with lower histologic grade, as well as ER and PR expression, whereas loss of nuclear and total

STAT5a is observed in invasive breast cancer and lymph node metastases, and is associated with elevated risk of failure of antiestrogen treatment.143–146,148–150 The dual functionality of STAT5a, as a proliferation and survival factor as well as a differentiation factor, is not unheard of, as more cancer “gene-chameleons” have been described in which function depends on overall genome context and pathway disturbances.256,257

Phosphorylation of STAT5a residue Y694 in response to cytokine stimulation has long been thought to be the only post-translational modification regulating STAT5a activity, however a role for STAT5a in the absence of tyrosine phosphorylation is beginning to be appreciated. To this point, we know that in the absence of tyrosine phosphorylation STAT5a shuttles in and out of the nucleus, which our findings show as well.151–153 (Woock et al, manuscript in review). Further, using genomic studies, Park et al supports the idea that upY-STAT5 may act as a partial antagonist of the biological activity of pY-STAT5.157,161,162 In fact, their study showed that that upY-STAT5 suppressed transcriptional programming required for megakaryocytic differentiation, while

TPO/pY-STAT5 redistribution to promoters with GAS sequences allowed for transcription of genes involved in survival of differentiating cells.161 The work done in pursuit of this thesis joins

146

a growing literature showing that regulation of this transcription factor is more nuanced. Thus, determining the regulation of STAT5a activity in the absence of tyrosine phosphorylation, may allow us to understand how STAT5a may act as a cancer “gene-chameleon”.

The phosphorylation of STAT5a on serine residues S726 and S780 had previously been shown in human immune cells and mouse tissues169,173,174 Further, in the mouse mammary gland, phosphorylation of S780 and S726 increase throughout pregnancy and lactation, but this phosphorylation is lost at involution.170,175 While functions have been ascribed for these serine residues in hematopoietic models of transformation, their functions are less clear in normal or malignant breast tissue.127,130,169,170,175,178,179In hematopoietic studies, mouse transplant studies determined that loss of STAT5a serine phosphorylation in leukemic cells delayed disease onset in mice.169,178 No such phenotypic study has been performed to understand STAT5a serine phosphorylation in breast development or cancer. Instead, the focus has been on STAT5a-target gene induction studies, with mixed results dependent on cell type and reporter gene construct.127,130,170,175,179 The consensus from these studies had been that the phosphorylation of the serine residues mainly negatively regulate canonical phosphorylation of residue Y694 when analyzing STAT5a activity on the β-casein gene.127,169,170,175 However, the reliance of these studies on the use of the β-casein reporter assay, while useful, are limited in scope when discussing

STAT5a as a transcription factor with multiple gene targets.

In this thesis, we show that the phosphorylation of STAT5a on serine residues S726 and

S780 in both human breast carcinoma cell lines and patient tissue samples in a TMA, and that while pS726 increased from tumor grade I-III, pS726 and pS780 co-occurred in 77% of the tissues.

Phosphorylation of S726 is dependent on PRL, whereas pS780 is PRL-independent and constitutively expressed in a wider range of intrinsic subtypes of breast cancer.

147

Using MCF7 cells expressing single-point and double-point phospho-deficient STAT5a, we determined that the loss of STAT5a serine phosphorylation differentially affects breast cancer phenotypic characteristics that fall within the hallmarks of cancer.217,218 In MCF7 cells, single- point mutations of the phospho-sites led to modest, yet significant decreases in soft agar growth

(Y694F-, S780A-STAT5a), or PRL-induced proliferation (S726A-STAT5a). Expression of

S726A-STAT5a also led to an increase in the apoptosis protein cleaved caspase 3 in T47D cells.

These findings indicate that intact expression of each phospho-site contributes to tumorigenicity of luminal breast cancer cells in a non-redundant manner. Further, expression of the double-point mutant S726A+S780A-STAT5a significantly decreased MCF7 cell proliferation, which was not rescued by PRL treatment, decreased the ability of the cells to grow in an anchorage-independent manner, and increased the rate of migration. These studies show that intact expression of pS726 and pS780 is necessary for MCF7 cellular proliferation and the ability to grow without an extracellular substrate, and that intact expression pY694 is not sufficient to overcome the loss of these phospho-serine residues. Conversely, intact expression of these phospho-serine residues may restrict migratory ability of MCF7 cells, ultimately affecting the possibility of metastatic disease.

Together, the data from both the single-point and double-point phospho-deficient STAT5a studies indicate that while each phospho-site has a role in regulating STAT5a, there is cooperative action as well. These findings are similar to the those from the hematopoietic studies which showed loss of the serine residues decreased leukemic potential both in vitro and in vivo.169,178

For the first time, the loss of single phospho-sites on STAT5a have been studied for their effects on the transcriptome in breast cancer. Previously, the assumption that pY694-STAT5a was necessary for most PRL-induced gene expression changes had only been indirectly assessed with pharmacologic inhibition (targeting JAK2 tyrosine-phosphorylation) or total knockdown of

148

STAT5a and the subsequent transcriptional activity of select reporter genes. 85,86,136,213–216 The overwhelming use of reporter genes in the study of the phospho-serine residues of STAT5a supported this assumption. 127,130,170,175,179

Thus, RNA-sequencing of MCF7 cells expressing single-point phospho-deficient STAT5a mutants treated with PRL was performed in attempts to identify differentially expressed genes and functional pathways for cancer outgrowth. Expression of the STAT5a phospho-mutants led to distinct PRL-induced DEGs compared to WT-STAT5a, which ultimately led to differences in the functional pathways analyzed. To this end, GSEA for oncogenic signatures was undertaken to examine how the STAT5a phospho-mutants may affect cellular pathways dysregulated in cancer.

GSEA goes beyond analysis of comparing single gene changes from a list of DEGs, so that smaller increases or decreases in all genes encoding members of a cellular process may be considered.

These small changes may represent dramatic changes in biology of a cellular process or pathway that may be more important than the change in a single gene.195 The GSEA analysis revealed that signaling pathways, such as KRAS and ERα, which have been implicated in cooperating with PRL signaling rely on the phospho-sites differentially.67,69,75 (Grible et al, in press 2021) Additional examination of the RNA-seq data using GSEA determined that loss of each STAT5a phospho-site led to enrichment for various transcription factors in MCF7 cells. Interestingly, some of the same enriched pathways implicated by oncogenic signature analysis were also implicated in examination of transcription factor enrichment.

Overall, the phospho-sites of STAT5a have been shown to mediate gene expression differentially, and rather than the phospho-serine residues only modulating canonical pY694-

STAT5a transcriptional activity, most likely these phospho-serine residues regulate STAT5a at different regulatory elements in the genome.

149

Comparison of the MCF7 STAT5a phospho-point mutant RNA-seq data with PDX MGT and TCGA-BRCA samples led to the findings that the gene expression changes observed in the

RNA-seq were not large enough to elicit intrinsic subtype switching. However, given the data we’ve presented in this thesis, in which pS726 and pS780 are present in tumors spanning grades

I-III, and the fact that pY694 is lost with breast cancer progression, it is possible that the phospho- serine residues of STAT5a could be biomarkers to stratify patients and determine outcomes.143,148–

150

In Aim 3 of this thesis, the targeted focus was to determine if ERK1/2 was the serine/threonine kinase responsible for phosphorylation of either S726 or S780 on STAT5a. Given the various previous reports showing ERK1/2 associating with STAT5a, and the apparent in vitro kinase activity which had been described in response to GH, we hypothesized that ERK1/2 would also serine-phosphorylate STAT5a in response to PRL.32,34,127,128 Pharmacologic inhibition of the

MAPK pathway did not support this hypothesis in luminal breast cancer cells. However, we found that ERK1/2 and STAT5a are complexed prior to PRL stimulation, and ERK1/2-STAT5a complex association was found to depend on the phospho-serine residues in a contrasting manner; the intact expression of S726 may increase the affinity of the interaction, while the presence of S780 may decrease the affinity of the interaction. Thus, the kinase responsible for phosphorylating either serine residue, and the role of the association of ERK1/2 and STAT5a, remains to be determined.

7.2. Limitations

This thesis was limited to the in vitro characterization of STAT5a phospho-serine residues.

While in vitro approaches are powerful by providing a relatively quick and efficient means to study cellular phenotypes and function, they are limited in the ability to recapitulate disease progression and biology that occurs in living systems. Additionally, the stable expression of the STAT5a

150

phospho-deficient mutants was restricted to the luminal cell line, MCF7. Expression of the S726A-

STAT5a mutant in the luminal T47D cell line led to uncontrolled apoptosis, precluding studies of the phospho-deficient mutants in T47D cells. Ideally, a range of breast cancer subtypes would have been examined for the effects of expression of STAT5a phospho-deficient mutants, to better understand how these phospho-sites affect tumorigenicity throughout breast cancer progression.

Limitations and technical bias were also introduced in performing the RNA-seq, which has been detailed in section 5.4 in Chapter 5. Briefly, a limited number of biological replicates were performed, following guidelines for microarray high-throughput sequencing rather than the suggested six replicates per condition for RNA-seq.230 This limitation, along with using the poly(A)-selection method for preparing the mRNA library for sequencing, ultimately led to the use of p values rather than the FDR to identify significant DEGs. To overcome this limitation, comparison of the WT-STAT5a dataset to past genomic datasets from our laboratory was performed to confirm significant DEGs.

The targeted approach to Aim 3 was limited by primarily examining the relationship of

STAT5a with one serine/threonine kinase, ERK1/2. Further, this study employed pharmacologic inhibition of ERK1/2 but did not validate these findings using genetic knockdown of ERK1/2 to study the effects of ERK1/2 on STAT5a serine phosphorylation. Often, genetic knockdown or knockout by shRNA or siRNA, respectively, is utilized for such validation to mitigate any speculation of off-target effects.258

7.3. Future Directions

Our data show that upY694-STAT5a has intact phosphorylation of the serine residues S726 and S780. Further, examining phospho-site mutants by RNA-seq in conjunction with functional studies, we have shown that the phospho-serine residues have non-redundant roles distinct from

151

regulating the phosphorylation of Y694. Phospho-S726 and pS780 have been shown to be more influential on proliferation and soft agar growth, respectively, than pY694 in luminal breast cancer cells. Thus, it is likely these residues function in breast cancer when pY694-STAT5a is lost and contribute to the proliferative, more aggressive functions of STAT5a in breast cancer progression.

However, the influence of these STAT5a phospho-sites on more aggressive intrinsic subtypes of breast cancer, i.e. basal-like or claudin-low, remains to be determined. Therefore a future direction of this research would be to express the STAT5a phospho-mutants in a basal-like breast cancer cell line such as MDA-MB-231 or MDA-MB-436, which have low levels of endogenous STAT5a expression.259,260 Functional studies and RNA-seq of these basal-like breast cancer cells expressing the phospho-site mutants could give insight into how the phospho-serine residues regulate STAT5a activity in more aggressive breast cancers. Given our findings in luminal breast cancer cells, we hypothesize that loss of the phospho-serine residues will lead to decreased tumorigenicity in the basal-like subtype, whereas loss of pY694 will be less effective in eliciting a change in phenotype or predicted oncogenic pathways. Transplantation of these cells, along with the MCF7 expressing STAT5a phospho-site mutants into the mammary glands of NSG mice will allow for examination of the functions of these phospho-serine residues in vivo. Examining the ability of these phospho-serine residues to affect tumor growth and possible metastatic growth will increase our understanding of the function of STAT5a in the progression of more aggressive breast cancer. Loss of these phospho-serine residues should lead to decreased tumorigenicity in vivo as evidenced by decreased tumor burden and increased overall survival, as seen in the hematopoietic mouse models expressing phospho-deficient STAT5a leukemic cells.169,178

Another future direction will be to examine primary breast cancer tissues and matched metastatic tissues for expression of pY694-, pS726-, and pS780-STAT5a. It is hypothesized that

152

examining a TMA for these STAT5a phospho-sites will show differential expression of the phospho-sites. Specifically, we hypothesize that loss of pY694 will occur in the metastases, whereas there will be similar or increased expression of pS726 and pS780 in the metastatic tissue compared to the primary tumor. This future finding would support the idea that phosphorylation of these serine residues could be used as a biomarker in primary tumors for potential metastatic progression.

Finally, the kinase responsible for phosphorylating either of the STAT5a serine residues is still unknown. Therefore, we propose as a future direction an inhibition assay for potential serine/threonine kinases to determine which kinase phosphorylates S726 and S780. Specifically, a recent technique was developed, termed kinase inhibitor profiling to identify kinases (KiPIK), to identify kinases for specific phosphorylation sites using in vitro kinase assays on cellular extracts in the presence of kinase inhibitors, which would be applicable to determining potential kinases responsible for STAT5a serine phosphorylation.261 If potential kinases are identified, then we will utilize both pharmacologic inhibition (if targeted drugs exist) and genetic knockdown/knockout approaches to validate the inhibition assay. We hypothesize that any kinase downstream of

ERK1/2 will not be responsible for phosphorylation of S726 or S780. This includes the NEK3 kinase, which is activated by ERK1/2 phosphorylation in response to PRL.39–41 Conversely, PAK is potentially responsible for phosphorylating either serine residue, and should therefore be included in this study.178,253,254

While ERK1/2 and downstream kinases are most likely not responsible for STAT5a serine phosphorylation, it is evident from our work that ERK1/2 and STAT5a are complexed. A future direction of this work is to therefore uncover the function of this association. ERK1/2 and STAT5a may be complexed with other kinases that may be responsible for phosphorylation of STAT5a,

153

since ERK1/2 is known to have complex formation with various other proteins.250,251 Further,

ERK1/2 is largely located in the cytoplasm in quiescent cells, but there is evidence for ERK1/2 in the nucleus prior to cytokine stimulation.262 Upon growth-factor or cytokine stimulation, ERK1/2 translocates to the nucleus, and is known to interact with other proteins, phosphorylate substrates, or direct protein-DNA interactions.263 Therefore, it could be hypothesized that ERK1/2 and

STAT5a interact in the nucleus. Our data indicated that there is ERK1/2-STAT5a complexed both prior to and following stimulation, with slight dissociation upon stimulation, however it is still possible that there could be nuclear interaction between the two, given upY-STAT5a is shuttled into and out of the nucleus.111,152,155,157–159,162,201 Evidence to support this hypothesis would come from a nuclear-cytoplasmic cellular fractionation assay followed by co-immunoprecipitation.

Thus, it could be possible that ERK1/2 acts to direct STAT5a nuclear translocation, or even influences STAT5a interactions with cofactors.

154

8. Appendices

8.1. Supplementary Figures for Chapter 3

Supplementary Figure S1. TMA for pY694-STAT5a shows weak nuclear staining.

Representative anti-pY694 and anti-STAT5a IHC images from unmatched Grade III, ER-/PR-

155

/Her2 amplified primary tumors included in the TMA (scale bars, 100 μm). Graphs show quantification of modified Nuclear Allred scores according to the tumor grade, Ki67 status, and molecular subtypes.

Supplementary Figure S2. Confirmation of STAT5a knockdown and rescue at the mRNA level in MCF7 cells. MCF7 cells with stable knockdown (KD) and rescue of STAT5a by indicated constructs analyzed by qRT-PCR for STAT5a mRNA. **p<0.001, ****p<0.0001.

156

Supplementary Figure S3. Gene induction is affected differentially by loss of STAT5a phosphorylation. A) Expression log ratios of select PRL-inducible or repressed genes from the

157

RNA-sequencing visualized in IPA. B) Expression log ratios of the same panel of PRL-induced

DEGs of each mutant STAT5a compared to WT-STAT5a + PRL DEGs visualized in IPA, illustrating how each mutant specifically increases or decreases expression of these genes compared to WT-STAT5a + PRL. C)-F) Confirmatory qRT-PCR analysis of the same panel of

PRL-inducible or PRL-repressed genes, normalized to total STAT5a expression and GAPDH for each sample. **p=0.005, ***p<0.0001 vs. WT-STAT5a + PRL. #p<0.00005 rescue PRL vs. untreated.

Supplemental Figure S4. Effect of single-point phospho-deficient STAT5a mutants on characteristics of MCF7 cells. 2D proliferation measured by XCELLigence for each MCF7 cell

158

type assayed untreated (serum-free media, SFM) or treated with PRL. A) All rescues in SFM or

SFM + PRL (n=3, p<0.02 compared to WT-STAT5a; see Figure 3.4, Chapter 3) B) Each rescue graphed individually comparing SFM to SFM + PRL. n=3, Nonsignificant.

Supplementary Figure S5. Effect of single-point phospho-deficient STAT5a mutants on characteristics of MCF7 cells. A) Representative micrographs of anchorage-independent growth assayed by soft agar for WT-STAT5a, EV, or S726A-STAT5a at endpoint. B) Scratch wound assay for migration, *p≤0.04, ***p=0.008 compared to WT-STAT5a.

159

Supplementary Figure S6. Validation of specificity of STAT5a staining using empty vector

(EV) control cell line. MCF7 cells transduced with EV pTracer were stained with anti-STAT5a antibody and imaged as in Figure 6. Representative merged Hoechst and α-STAT5a images from time course illustrate lack of STAT5a staining.

8.2. Supplementary Tables

Supplementary Table S1 for Woock et al. Patient characteristics of TMA. Characteristic TMA (N=47) Age-yr. Median 55 Range 25-90

160

Race-n (%) Black 27 (57.4) Hispanic 1 (2.1) White 17 (36.2) Other 2 (4.3) Histologic Grade-n (%) I 6 (12.8) II 10 (21.3) III 27 (57.4) Not reported 4 (8.5) Ki67 status-n (%) Low 18 (38.3) Intermediate 4 (8.5) High 8 (17.0) Not reported 16 (34.0) ER Status-n (%) Negative 16 (34.0) Positive 31 (65.9) PR Status-n (%) Negative 26 (55.3) Positive 21 (44.7) HER2 amplified Status-n (%) Negative 38 (80.9) Positive 9 (19.1) Hormone Receptor Status-n (%) ER+/ PR- /Her2- 8 (17.0) ER+ /PR+/ Her2- 19 (40.4) +/+/+ 2 (4.3) ER- /PR- /Her2+ 5 (10.6) -/-/- 11 (23.4) ER+ /PR- /Her2+ 2 (4.3)

Supplementary Table S2 for Woock et al. is included in Chapter 5 (Table 5.2)

Supplementary Table S3 for Woock et al. Quality check on WT-STAT5a (PRL vs untreated) dataset presented compared to microarray/RNA seq published previously.

Fiorillo et al70 Hakim et al78 Craig et al87 Woock et al

161

Gene FC p value FC p value logFC p value logFC p value

BCL6 -3.088 0.00828 -3.085 1.29E-19 -2.267 7.58E-10 -2.511 7.23E- 86

HBP1 -1.72 0.00261 -0.61 1.15E-14 -0.567 0.000266 -0.327 0.00056 8

FBXO32 -1.577 0.0192 -0.293 0.000216 -0.467 0.00017 1

YPEL2 -1.544 0.0324 -0.323 0.00322

TSC22D3 -1.543 0.0281 -0.743 7.63E-16 -0.54 5.96E-07

TXNIP -1.489 0.047 -0.615 1.39E-13 -0.528 3.51E-06

ZNF627 -1.422 0.0275 -0.39 0.00221

FAM217B -1.417 0.0171 -0.25 0.0159

EFNA1 -1.398 0.0336 -0.524 9.55E-11 -0.218 0.00086

PAQR6 -1.359 0.00055 -0.203 0.00624 0.339 0.000841

TTLL3 -1.287 0.0388 -0.22 0.00731 0.43 0.000623

NPHP3 -1.264 0.0259 -0.447 0.006

FRAT1 -1.203 0.00489 -0.691 2.03E-12 -0.412 0.000263

PYCR3 1.222 0.019 0.165 0.00408 0.207 0.0242

JMJD6 1.238 0.0137 0.737 1.23E-11 0.446 5.07E-05 0.327 0.00774

STX1A 1.242 0.00406 0.295 0.0308

SPHK1 1.243 0.0389 0.426 5.52E-09 0.366 0.00106 0.217 0.0151

CHKA 1.25 0.0143 0.327 0.000156 0.27 0.000416

OSGIN1 1.252 0.0349 0.458 4.16E-08 0.485 0.00106 0.339 0.00084 2

ANKRD27 1.265 0.0341 0.236 3.55E-06 -0.255 0.00613

SENP5 1.273 0.0269 0.39 6.66E-08 -0.226 0.0395

GCNT1 1.275 0.0414 0.667 2.67E-08 0.601 0.00034 7

162

ICOSLG/L 1.287 0.00244 0.662 4.5E-07 0.589 2.31E- OC102 06 723996

TRMT61A 1.291 0.0143 0.177 0.00433 0.212 0.0325

ETS2 1.309 0.00607 0.696 1.64E-10 0.239 0.0247

CD3EAP 1.313 0.00169 0.428 3.31E-08 0.346 0.000262 0.403 0.00014 2

RPP25 1.314 0.0465 0.301 4.81E-06 0.307 0.00261

ASB7 1.333 0.0016 0.202 0.000531 -0.28 0.0359

USP36 1.333 0.00442 0.849 8.59E-08 0.441 6.78E-06 0.229 0.01

PLA2G15 1.337 0.0162 0.362 1.24E-06 0.34 0.00115 0.225 0.0477

RRP1 1.337 0.0299 0.458 1.58E-08 0.265 0.00356

PPAN 1.346 0.00442 0.447 9.98E-10 0.44 0.000804 0.388 5.23E- 05

PDGFB 1.348 0.0148 0.217 0.00404 0.223 0.0178

LRRC8E 1.361 0.00953 0.479 4.07E-08 0.245 0.031

ZNF324 1.368 0.0344 0.203 0.000122 0.33 0.0012

ABCF2 1.369 0.045 0.338 1.91E-07 0.204 0.0226

G6PD 1.371 0.0355 0.177 0.0446

TXNRD1 1.383 0.0388 0.119 0.000864 -0.196 0.0232

MINPP1 1.385 0.0498 0.234 0.000107 -0.297 0.0407

PA2G4 1.387 0.0295 0.109 0.00489 0.197 0.021

DDX28 1.405 0.0111 0.16 0.0036 0.309 0.00676

RCL1 1.411 0.0167 0.411 2E-09 0.407 1.11E-05 0.392 0.00064 3

KLF4 1.415 0.00312 0.392 0.000122 -0.262 0.005

TMEM158 1.415 0.000137 0.497 0.000588 0.904 0.00435

ADRM1 1.419 0.0315 0.225 4.72E-06 0.181 0.0372

163

DDX56 1.432 0.0457 0.147 0.00476 0.212 0.0155

TSPAN14 1.437 0.0264 0.158 0.00176 0.237 0.0058

MAT2A 1.437 0.0271 0.625 4.87E-14 0.331 4.06E-05 0.276 0.00133

CISH 1.44 0.00202 0.707 2.04E-11 1.318 2.96E-11 0.645 7.74E- 10

RRP7A 1.45 0.0445 0.224 0.00011 0.331 0.000269 0.218 0.0201

CSRP1 1.466 0.0151 0.322 7.19E-08 0.226 0.00835

NOTCH1 1.471 0.00367 0.473 3.21E-07 1.381 0.00058 4

JMJD4 1.476 0.000247 0.338 1.2E-06 0.234 0.0304

DHCR7 1.487 0.0144 0.172 0.000741 0.201 0.0187

PUS1 1.495 0.0423 0.395 2.88E-09 0.372 0.000418 0.238 0.0141

GRPEL1 1.515 0.0248 0.565 3.44E-14 0.25 0.0109

MIR22HG 1.516 0.000606 0.892 6.15E-14 0.357 0.0464

NOP56 1.526 0.00315 0.306 0.000061 0.229 0.0101

SLC20A1 1.531 0.0126 0.46 3.49E-09 0.264 0.00748

DOLK 1.533 0.004 0.155 0.00355 0.286 0.00328

BYSL 1.556 0.0126 0.515 7.16E-13 0.451 5.61E-05 0.332 0.00291

RCAN1 1.572 0.00661 0.57 1.38E-07 0.771 3.2E-07 0.667 1.17E- 09

CCDC86 1.584 0.0107 0.42 4.51E-11 0.428 5.74E-05 0.34 0.00111

POLR1C 1.584 0.0011 0.438 1.65E-11 0.467 2.82E-05 0.31 0.00628

GRWD1 1.589 0.0423 0.252 4.79E-05 0.221 0.0152

RNF126 1.589 0.017 0.39 2.58E-06 0.231 0.0208

TWNK 1.602 0.00806 0.449 8.96E-07 0.305 0.00012 0.232 0.0232

CEBPA 1.628 0.0234 0.585 1.88E-07 0.314 0.00204

164

MSRB1 1.638 0.016 0.374 3.51E-10 0.395 0.000237 0.431 2.33E- 06

NOP2 1.644 0.0166 0.349 1.07E-06 0.289 0.000286 0.186 0.04

AUNIP 1.649 0.0107 0.408 2.46E-07 0.376 0.00931

GEMIN4 1.671 0.0345 0.37 3.9E-07 0.262 0.000329 0.263 0.00396

SLC2A1 1.678 0.0272 0.202 0.000548 0.226 0.00818

BRPF1 1.707 0.00448 0.347 8.76E-09 0.218 0.0163

CNN2 1.711 0.00473 0.267 9.19E-05 0.171 0.047

NOL6 1.722 0.0314 0.272 9.44E-05 0.215 0.0174

NXT1 1.775 0.00217 0.433 3.4E-11 0.359 0.000282 0.25 0.0252

NOP16 1.817 0.000846 0.641 3.69E-12 0.478 4.15E-05 0.294 0.00239

AEN 1.829 0.0206 0.516 9.69E-11 0.36 9.44E-06 0.309 0.00185

FJX1 1.885 0.000274 0.538 9.96E-11 0.309 0.0112

RIOX1 1.906 0.0022 0.251 2.32E-05 0.236 0.0148

CEBPB 1.932 0.000818 0.804 5.82E-17 0.264 0.00347

PPRC1 2.007 0.000593 0.53 1.31E-09 0.462 2.56E-05 0.333 0.00019 3

RRS1 2.053 0.00289 0.359 8.47E-10 0.23 0.000894 0.263 0.00412

ZYX 2.076 0.000448 0.854 9.08E-11 0.511 1.67E-05 0.186 0.0308

RARA 2.166 0.000791 0.838 4.18E-10 0.639 4.77E-08 0.436 4.1E-07

SPATA2L 2.326 0.00281 0.556 1.48E-09 0.423 0.00105 0.294 0.00309

TUFT1 2.351 0.000781 1.259 1.85E-17 0.215 0.0278

PHLDA2 2.384 0.000907 1.187 4.65E-13 0.699 3.23E-06 0.952 4.1E-08

IER3 2.417 0.000551 0.777 8.23E-13 0.444 5.27E-07 0.382 4.08E- 05

TNFRSF12 3.04 8.74E-05 1.331 2.09E-10 0.443 1.12E-05 0.259 0.00255 A

165

MYADM 3.398 5.61E-07 0.352 0.000167 0.366 0.00004 1

BTG1 -0.729 6.27E-13 -0.467 0.000066 -0.19 0.0282

NR2F2 -0.545 1.27E-06 -0.495 0.00095 -0.289 0.00086 9

STARD13 -0.543 1.96E-09 -0.556 0.000314 -0.292 0.00877

ACKR3 -0.538 5.04E-08 -0.606 0.000585 -0.238 0.00694

TSC22D1 -0.534 1.6E-12 -0.516 0.000148 -0.256 0.005

CDKN1B -0.491 2.67E-08 -0.661 4.44E-05 -0.303 0.00103

CITED2 -0.465 2.48E-06 -0.575 0.000143 -0.325 0.00015 9

NAB2 0.206 0.000994 0.417 4.74E-07 0.289 0.00104

PNP 0.317 5.32E-10 0.327 6.27E-05 0.228 0.0303

MRTO4 0.317 8.66E-05 0.327 0.000624 0.253 0.00729

MYBBP1A 0.347 7.32E-06 0.278 5.17E-05 0.198 0.0232

LOC102724 0.472 2.66E-10 0.329 8.06E-05 0.216 0.0441 159/P WP2

JUN 1.466 3.43E-14 0.301 0.00102 0.207 0.0403

Table 8.1 Differentially expressed genes for MCF7 cells expressing WT-STAT5a, PRL vs

untreated, log10(fold change)>0.58, fold change >1.5.

WT-STAT5a Gene Log10(Fold Change) p value padj BCL6 -2.511 7.2E-86 0.0000 CISH 0.645 7.7E-10 0.0000 DYNC2H1 1.925 1.1E-09 0.0000 RCAN1 0.667 1.2E-09 0.0000 FAM83A 1.078 1.3E-08 0.0002 PHLDA2 0.952 4.1E-08 0.0004 RGPD2 -3.439 1.8E-07 0.0014

166

BTBD8 -2.979 2.3E-07 0.0015 RPL23AP47 -6.244 9.5E-07 0.0051 FP565260.3 0.589 2.3E-06 0.0085 P2RX2 -0.776 2.5E-06 0.0085 AC011468.2 -2.938 2.9E-06 0.0094 SOCS2 0.594 5.1E-06 0.0137 AL603962.1 6.043 5.3E-06 0.0137 AL121845.2 6.040 6.2E-06 0.0143 GCOM1 -0.896 6.3E-06 0.0143 AP001057.1 0.823 7.7E-06 0.0161 INPP5F -0.718 8.7E-06 0.0175 CU633906.1 -5.112 1.1E-05 0.0200 AC243772.4 -5.992 1.3E-05 0.0216 AL162151.2 -0.832 1.3E-05 0.0216 SLC7A5P2 -5.976 1.4E-05 0.0222 AC012531.2 -5.932 2.0E-05 0.0307 CHURC1- -0.950 2.1E-05 0.0307 FNTB AGAP11 -1.466 2.5E-05 0.0362 AL021546.1 3.131 2.9E-05 0.0400 CFAP206 -0.881 3.1E-05 0.0419 MSH5-SAPCD1 -1.249 3.8E-05 0.0506 Z83844.3 0.685 4.6E-05 0.0547 AC027644.4 -1.456 5.9E-05 0.0647 AP000873.1 -2.372 7.0E-05 0.0701 TNS4 1.123 7.1E-05 0.0701 IFIT1 -1.363 7.3E-05 0.0708 AL049650.1 5.701 7.5E-05 0.0718 ZNF625-ZNF20 -0.922 9.5E-05 0.0858 AL445363.3 1.123 1.0E-04 0.0883 RPPH1 -0.622 1.5E-04 0.1210 PTGES3L- 1.262 1.7E-04 0.1240 AARSD1 FGF7P5 -5.580 1.7E-04 0.1240 AL136982.4 1.583 2.2E-04 0.1520 UGT1A8 2.377 2.5E-04 0.1586 AL662884.2 -5.556 2.7E-04 0.1698 PLGLB2 -2.891 3.1E-04 0.1891 ZNF490 -0.610 3.3E-04 0.1927 AL355987.3 -2.606 3.4E-04 0.1951 GCNT1 0.601 3.5E-04 0.1951 AC011511.4 5.413 3.5E-04 0.1951

167

RASL11A 0.743 3.5E-04 0.1951 H2BFS -3.861 3.9E-04 0.2068 ZNF511-PRAP1 1.276 4.0E-04 0.2098 AMT -3.743 4.3E-04 0.2185 LINC01260 5.358 4.5E-04 0.2222 AC106886.2 -1.542 4.6E-04 0.2222 AC012254.2 -0.732 5.0E-04 0.2335 AC211476.3 -5.955 5.5E-04 0.2563 NOTCH1 1.381 5.8E-04 0.2651 ZRSR2P1 -5.380 5.9E-04 0.2658 GOLGA8O 5.294 6.3E-04 0.2739 LIMS4 2.533 6.8E-04 0.2837 Z84488.2 5.286 7.0E-04 0.2837 AC104532.1 -5.325 7.2E-04 0.2874 BPTFP1 1.246 9.1E-04 0.3308 AL049795.2 -5.255 9.4E-04 0.3368 RHOQP3 -2.218 9.8E-04 0.3425 EGR4 0.938 9.8E-04 0.3425 RPL23AP60 -5.251 1.1E-03 0.3535 AL021920.3 -5.207 1.1E-03 0.3596 PVRIG2P -5.996 1.2E-03 0.3741 AC087289.3 -0.874 1.2E-03 0.3741 AC027796.3 -6.916 1.2E-03 0.3741 AL691442.1 1.222 1.3E-03 0.3881 CACNA1I 1.153 1.4E-03 0.4060 CBSL -1.310 1.5E-03 0.4174 AL137247.1 5.085 1.5E-03 0.4174 AL136295.1 -5.151 1.5E-03 0.4174 ACP5 -5.124 1.6E-03 0.4341 FAM156B -0.635 1.7E-03 0.4543 AC159540.2 2.717 1.8E-03 0.4611 DUSP2 0.813 1.8E-03 0.4706 AC073111.4 3.103 1.9E-03 0.4706 TERC 5.029 1.9E-03 0.4715 GLDN -0.603 1.9E-03 0.4726 AL512356.1 1.026 2.0E-03 0.4802 AC068533.4 -1.910 2.0E-03 0.4802 CORT -1.912 2.1E-03 0.4802 LUZP6 4.999 2.1E-03 0.4802 MFSD2A 0.796 2.2E-03 0.5023 AC092143.3 1.488 2.3E-03 0.5070

168

PCDHA8 -1.119 2.4E-03 0.5280 PLCXD2 -0.809 2.5E-03 0.5287 AC013271.1 4.933 2.6E-03 0.5427 AL021368.3 4.919 2.7E-03 0.5588 AC004224.2 1.374 2.9E-03 0.5797 HIST1H3D 4.916 2.9E-03 0.5806 TVP23C -0.717 3.1E-03 0.5966 AC083899.1 -0.649 3.2E-03 0.5990 TAF1A -0.721 3.3E-03 0.6004 RPL23AP24 4.857 3.5E-03 0.6100 Z84492.2 -4.943 3.5E-03 0.6100 NUTM2E 4.848 3.5E-03 0.6123 AL139353.1 0.844 4.0E-03 0.6414 COX6B2 4.805 4.1E-03 0.6473 AC025183.2 -2.581 4.2E-03 0.6514 AC073343.2 4.792 4.2E-03 0.6550 ZBTB45P1 -4.861 4.3E-03 0.6550 LINC02138 4.785 4.3E-03 0.6566 AP005263.1 4.783 4.3E-03 0.6566 TMEM158 0.904 4.3E-03 0.6566 PCDHGB3 -0.588 4.5E-03 0.6682 ZNF816- 4.767 4.6E-03 0.6706 ZNF321P AC020891.2 -1.937 4.6E-03 0.6706 EDARADD 0.644 4.6E-03 0.6730 TMEM110- 0.930 4.7E-03 0.6755 MUSTN1 FAM157C 1.268 4.7E-03 0.6755 AL079301.1 1.838 4.8E-03 0.6768 AL031123.2 -0.633 4.9E-03 0.6768 BCLAF1P2 -1.094 4.9E-03 0.6768 AC008758.6 -4.816 5.0E-03 0.6768 IFIT3 -1.073 5.0E-03 0.6768 STOX2 -0.698 5.2E-03 0.6892 CASS4 1.214 5.4E-03 0.7140 AC134772.1 1.883 5.5E-03 0.7157 AL137786.1 -4.786 5.7E-03 0.7295 AL109766.1 -4.783 5.7E-03 0.7301 AL121790.2 -4.774 5.8E-03 0.7395 FOXQ1 -2.520 5.9E-03 0.7408 NUTM2A -0.633 5.9E-03 0.7408 OR7E39P -4.759 6.0E-03 0.7426

169

AL139393.2 0.779 6.1E-03 0.7471 AC146944.1 -2.782 6.2E-03 0.7481 HOXC5 0.714 6.3E-03 0.7481 CPEB1 -2.500 6.6E-03 0.7745 RWDD4P1 4.655 6.6E-03 0.7793 MCRIP2P1 -4.720 6.7E-03 0.7803 AC008763.2 4.641 6.7E-03 0.7803 CA9 1.660 6.8E-03 0.7819 AC242376.2 -1.482 6.8E-03 0.7868 AC007238.1 0.815 6.9E-03 0.7868 UGT1A10 -4.703 7.0E-03 0.7953 STARD9 -1.100 7.1E-03 0.7966 VN1R108P 4.618 7.1E-03 0.7966 AL357558.2 1.601 7.1E-03 0.7966 KRT18P13 4.621 7.1E-03 0.7966 AC012146.3 4.614 7.2E-03 0.7979 AL159163.1 4.614 7.6E-03 0.8130 CORO7- -4.697 7.6E-03 0.8132 PAM16 AL357673.1 -4.696 7.6E-03 0.8132 WDR7-OT1 -4.668 7.8E-03 0.8171 AL157392.3 -0.613 8.0E-03 0.8321 WDR63 -2.760 8.1E-03 0.8356 BRCC3P1 -4.665 8.2E-03 0.8405 AC015712.4 -4.660 8.5E-03 0.8527 AC091167.1 -4.656 8.6E-03 0.8552 AC025279.2 -4.640 8.6E-03 0.8552 AC023509.3 1.390 8.8E-03 0.8633 SERPINA11 1.239 9.1E-03 0.8730 AL139352.1 -1.418 9.1E-03 0.8774 FAM78B -0.850 9.3E-03 0.8828 AC131160.1 -0.813 9.5E-03 0.8947 AL136295.5 0.782 9.8E-03 0.9028 AC119674.2 -4.591 1.0E-02 0.9097 OXCT2 0.945 1.0E-02 0.9104 AC093904.2 -0.688 1.0E-02 0.9342 SCARNA9 -1.893 1.0E-02 0.9342 BBS12 -0.628 1.0E-02 0.9347 GTF2IRD2P1 -1.702 1.1E-02 0.9454 DUSP19 -0.835 1.1E-02 0.9460 AC125611.3 0.907 1.1E-02 0.9502

170

AL121748.2 4.476 1.1E-02 0.9502 CSPG4P10 4.454 1.1E-02 0.9555 SLC22A13 4.465 1.1E-02 0.9555 AC105046.1 0.811 1.1E-02 0.9555 AC073263.2 4.461 1.1E-02 0.9555 SYT15 -1.110 1.1E-02 0.9555 CD164L2 0.772 1.1E-02 0.9555 AC092115.3 -1.054 1.1E-02 0.9555 AC007229.1 4.454 1.1E-02 0.9555 U47924.1 4.453 1.1E-02 0.9555 AC012615.6 3.309 1.1E-02 0.9555 AC124283.5 4.449 1.1E-02 0.9555 SNORA50C 4.451 1.1E-02 0.9555 ENO1P4 -4.542 1.2E-02 0.9617 GAPDHP72 3.299 1.2E-02 0.9617 CT45A1 -4.538 1.2E-02 0.9617 AL137129.1 -4.539 1.2E-02 0.9617 C9orf153 4.430 1.2E-02 0.9640 CLEC4A 4.430 1.2E-02 0.9640 AC011498.1 4.426 1.2E-02 0.9665 AC072061.1 -0.854 1.2E-02 0.9678 RPL23AP88 4.414 1.2E-02 0.9790 AL358074.1 -4.514 1.2E-02 0.9873 SNHG28 -4.512 1.2E-02 0.9905 UCN -0.778 1.3E-02 0.9905 AC025871.2 -4.506 1.3E-02 0.9905 ADPRM -0.588 1.3E-02 0.9940 INSL4 -3.159 1.3E-02 1.0000 IQUB 4.403 1.3E-02 1.0000 MIA-RAB4B 2.983 1.3E-02 1.0000 AC092143.2 0.934 1.3E-02 1.0000 CLEC3B -1.800 1.3E-02 1.0000 AC008608.2 0.633 1.3E-02 1.0000 AC005041.1 1.147 1.3E-02 1.0000 CROCCP3 -0.867 1.3E-02 1.0000 AC246785.2 -1.751 1.3E-02 1.0000 ZNF790 -1.422 1.4E-02 1.0000 LINC02166 1.667 1.4E-02 1.0000 KISS1 0.810 1.4E-02 1.0000 PRR34 0.787 1.4E-02 1.0000 HERC2P3 -1.012 1.5E-02 1.0000

171

Z84723.1 -2.546 1.5E-02 1.0000 CU638689.1 -3.233 1.5E-02 1.0000 KCNK4 1.551 1.6E-02 1.0000 TRIM31 -0.920 1.6E-02 1.0000 PEBP4 0.937 1.6E-02 1.0000 AL390195.1 2.272 1.6E-02 1.0000 AF111169.1 4.290 1.6E-02 1.0000 SNORD3A -0.782 1.7E-02 1.0000 AC005154.6 -4.394 1.7E-02 1.0000 AP000866.1 -1.097 1.7E-02 1.0000 CHAC1 0.610 1.7E-02 1.0000 RWDD4P2 -1.920 1.7E-02 1.0000 AL772155.3 3.196 1.7E-02 1.0000 STON1- -1.406 1.7E-02 1.0000 GTF2A1L SPRN 1.425 1.7E-02 1.0000 POTEE 0.907 1.7E-02 1.0000 AC092902.2 -0.681 1.7E-02 1.0000 AL162430.1 4.278 1.7E-02 1.0000 AC109587.1 1.682 1.8E-02 1.0000 OR2L2 3.169 1.8E-02 1.0000 AC005829.2 -4.362 1.8E-02 1.0000 DPY19L1P2 4.256 1.8E-02 1.0000 AP000654.1 -4.362 1.8E-02 1.0000 HMGN1P4 4.259 1.8E-02 1.0000 ACTN1-AS1 -4.359 1.8E-02 1.0000 TMEM249 0.811 1.8E-02 1.0000 AC010240.3 4.264 1.8E-02 1.0000 CTAGE3P -4.351 1.8E-02 1.0000 FSIP2 -4.349 1.8E-02 1.0000 KHDC1L -4.346 1.9E-02 1.0000 LINC02204 -4.339 1.9E-02 1.0000 LINC02327 -4.338 1.9E-02 1.0000 AC023137.1 4.243 1.9E-02 1.0000 DNAJC19P5 4.243 1.9E-02 1.0000 AL023806.1 -0.998 1.9E-02 1.0000 CTAGE15 4.253 1.9E-02 1.0000 FXYD4 4.235 1.9E-02 1.0000 GSAP -0.619 1.9E-02 1.0000 GNA15 4.231 1.9E-02 1.0000 TRGV10 4.231 1.9E-02 1.0000

172

AC046168.1 4.230 1.9E-02 1.0000 PNMT -4.322 1.9E-02 1.0000 MMP13 -0.682 1.9E-02 1.0000 AL713922.1 4.226 1.9E-02 1.0000 MRVI1 -4.316 1.9E-02 1.0000 AC106028.2 4.222 1.9E-02 1.0000 TBC1D3P1 4.222 1.9E-02 1.0000 MTND4P14 4.222 1.9E-02 1.0000 C8orf46 1.601 2.0E-02 1.0000 IL24 4.218 2.0E-02 1.0000 RPL36A- 4.240 2.0E-02 1.0000 HNRNPH2 AL356309.3 4.214 2.0E-02 1.0000 FOXD4L6 2.866 2.0E-02 1.0000 AC133551.1 -4.319 2.0E-02 1.0000 SP140 2.664 2.0E-02 1.0000 AC006252.1 1.185 2.0E-02 1.0000 AC092653.2 -3.010 2.0E-02 1.0000 MMP23A 4.216 2.0E-02 1.0000 AC243919.1 -2.580 2.0E-02 1.0000 GUSBP5 -1.185 2.1E-02 1.0000 TP53TG3B -1.175 2.1E-02 1.0000 AC104506.1 0.853 2.1E-02 1.0000 MYLK3 -1.304 2.1E-02 1.0000 RBM26-AS1 1.279 2.1E-02 1.0000 DNAJB4 -0.822 2.1E-02 1.0000 KRT85 3.078 2.1E-02 1.0000 LINC01119 1.428 2.1E-02 1.0000 AC138466.1 2.171 2.1E-02 1.0000 KLHL13 -0.785 2.2E-02 1.0000 U73166.1 -0.822 2.2E-02 1.0000 AC138894.3 -4.262 2.2E-02 1.0000 ABCA11P -0.593 2.2E-02 1.0000 ZBED8 -0.636 2.2E-02 1.0000 AC009961.1 1.147 2.2E-02 1.0000 HMGN1P3 -2.058 2.3E-02 1.0000 PTH1R 2.625 2.3E-02 1.0000 AL445309.1 -1.288 2.3E-02 1.0000 SYS1-DBNDD2 -0.853 2.3E-02 1.0000 AC135506.1 -1.377 2.3E-02 1.0000 NODAL 2.619 2.3E-02 1.0000

173

Z97633.1 0.619 2.4E-02 1.0000 AP000553.1 0.984 2.4E-02 1.0000 VAX1 1.459 2.4E-02 1.0000 AC069368.1 -4.241 2.4E-02 1.0000 ANKRD20A21P -2.014 2.4E-02 1.0000 MEF2B -0.690 2.4E-02 1.0000 AC105036.3 2.354 2.4E-02 1.0000 AC024940.6 -4.223 2.4E-02 1.0000 AC009093.8 -2.955 2.5E-02 1.0000 BMS1P11 -4.213 2.5E-02 1.0000 ARMC4 3.054 2.5E-02 1.0000 AL138966.2 1.090 2.5E-02 1.0000 AC008026.1 -4.205 2.5E-02 1.0000 ZNF230 -0.635 2.5E-02 1.0000 GLUD1P2 -2.188 2.6E-02 1.0000 AC093673.1 0.592 2.6E-02 1.0000 AC012309.1 -4.205 2.6E-02 1.0000 AL109613.1 3.037 2.6E-02 1.0000 AC023509.4 0.715 2.6E-02 1.0000 AGAP7P -0.591 2.6E-02 1.0000 AL031598.1 -1.829 2.7E-02 1.0000 NOXRED1 -1.616 2.7E-02 1.0000 H2AFB3 -1.975 2.7E-02 1.0000 C1orf127 -2.933 2.7E-02 1.0000 AC083843.3 -0.845 2.7E-02 1.0000 ARHGAP9 -4.181 2.7E-02 1.0000 RGS5 1.518 2.7E-02 1.0000 CAP2P1 -4.177 2.7E-02 1.0000 GNL3LP1 -0.743 2.7E-02 1.0000 AC244517.1 3.018 2.7E-02 1.0000 PRAMEF12 -4.174 2.7E-02 1.0000 LINC01143 0.588 2.7E-02 1.0000 GGTLC1 -4.173 2.7E-02 1.0000 LINC00899 -4.172 2.7E-02 1.0000 GLIPR1L2 -4.169 2.7E-02 1.0000 AC004466.2 -4.168 2.8E-02 1.0000 AC004554.2 -4.168 2.8E-02 1.0000 AC009955.1 3.013 2.8E-02 1.0000 AC008870.2 0.959 2.8E-02 1.0000 DUTP6 -4.167 2.8E-02 1.0000 NLRP2 -4.166 2.8E-02 1.0000

174

TMPRSS7 3.011 2.8E-02 1.0000 SAPCD1 1.169 2.8E-02 1.0000 HMGB1P11 -4.159 2.8E-02 1.0000 AL589666.1 -0.587 2.8E-02 1.0000 AC093908.1 1.110 2.8E-02 1.0000 AC015726.1 -1.294 2.8E-02 1.0000 AC135178.2 -1.257 2.8E-02 1.0000 CEP83-AS1 -0.674 2.8E-02 1.0000 PLAC1 0.876 2.8E-02 1.0000 AL353795.2 -4.152 2.8E-02 1.0000 AL358075.2 -2.909 2.8E-02 1.0000 ITIH6 -0.647 2.8E-02 1.0000 AC010761.1 -2.902 2.9E-02 1.0000 TMEM239 2.991 2.9E-02 1.0000 SNCAIP -4.128 2.9E-02 1.0000 AC092634.4 -1.054 2.9E-02 1.0000 LIMS3 -1.011 2.9E-02 1.0000 CATSPERZ 1.195 3.0E-02 1.0000 SNHG22 -1.029 3.0E-02 1.0000 AL645728.1 0.754 3.0E-02 1.0000 AC004854.2 0.928 3.0E-02 1.0000 AC156455.1 1.505 3.0E-02 1.0000 AL354993.1 1.323 3.0E-02 1.0000 SYNC -0.653 3.0E-02 1.0000 AC002310.2 -0.821 3.0E-02 1.0000 CPHL1P -0.733 3.0E-02 1.0000 FAM95C -0.706 3.1E-02 1.0000 NHP2P1 0.905 3.1E-02 1.0000 C20orf197 4.028 3.1E-02 1.0000 C9orf139 -1.138 3.1E-02 1.0000 AC027702.1 0.978 3.1E-02 1.0000 AL049836.1 4.014 3.1E-02 1.0000 MKRN9P 4.014 3.1E-02 1.0000 CLEC18C 4.010 3.1E-02 1.0000 AC006030.1 4.001 3.2E-02 1.0000 MTX1P1 4.000 3.2E-02 1.0000 AC217774.1 4.001 3.2E-02 1.0000 BCRP5 4.001 3.2E-02 1.0000 PELI2 4.001 3.2E-02 1.0000 AL049597.1 4.001 3.2E-02 1.0000 BTN3A1 -1.229 3.2E-02 1.0000

175

BEST4 3.992 3.2E-02 1.0000 HSFX3 3.992 3.2E-02 1.0000 SUCLA2-AS1 3.992 3.2E-02 1.0000 AP000757.1 3.992 3.2E-02 1.0000 AC022414.1 1.619 3.2E-02 1.0000 HLA-DRB5 2.527 3.2E-02 1.0000 PRAMEF5 3.984 3.2E-02 1.0000 PMS2P9 -4.093 3.2E-02 1.0000 AL627309.3 -1.582 3.2E-02 1.0000 LMO7-AS1 2.523 3.2E-02 1.0000 SYNPO2L -0.642 3.2E-02 1.0000 CSTA 3.979 3.3E-02 1.0000 RF00017 -1.113 3.3E-02 1.0000 AC106028.3 0.946 3.3E-02 1.0000 AC104596.1 -0.950 3.3E-02 1.0000 AL031133.1 3.975 3.3E-02 1.0000 ZNF442 3.975 3.3E-02 1.0000 AL158211.3 3.974 3.3E-02 1.0000 RPSAP31 3.974 3.3E-02 1.0000 TCL6 3.974 3.3E-02 1.0000 PRR4 -1.067 3.3E-02 1.0000 SIM1 3.966 3.3E-02 1.0000 LRRC4 1.177 3.3E-02 1.0000 SHBG 1.262 3.3E-02 1.0000 AC099668.1 3.983 3.3E-02 1.0000 AC018512.1 1.162 3.4E-02 1.0000 BTN3A2 -0.975 3.4E-02 1.0000 PRSS29P 1.805 3.4E-02 1.0000 AL353764.1 -2.111 3.4E-02 1.0000 KRT18P29 0.936 3.4E-02 1.0000 AL390195.2 -1.919 3.4E-02 1.0000 AC139272.1 -4.055 3.5E-02 1.0000 PDE1A -0.897 3.5E-02 1.0000 PIWIL4 -1.498 3.5E-02 1.0000 GLIS1 1.793 3.5E-02 1.0000 CHCHD4P3 -1.992 3.5E-02 1.0000 RGS22 -1.644 3.5E-02 1.0000 CA4 0.713 3.6E-02 1.0000 HIST1H2AK -1.547 3.6E-02 1.0000 SLC2A12 -0.597 3.6E-02 1.0000 NPIPB7 -2.017 3.6E-02 1.0000

176

LINC01134 1.895 3.6E-02 1.0000 UGT1A7 -0.882 3.7E-02 1.0000 SCARNA12 -2.052 3.7E-02 1.0000 AL022312.1 1.690 3.7E-02 1.0000 SPI1 -1.540 3.7E-02 1.0000 RASL10A 0.600 3.7E-02 1.0000 COG8 3.574 3.7E-02 1.0000 AC138969.2 -0.905 3.7E-02 1.0000 AC239800.4 -4.011 3.7E-02 1.0000 AC069287.2 1.460 3.8E-02 1.0000 LINC02170 -1.460 3.8E-02 1.0000 AC004130.1 -2.808 3.8E-02 1.0000 HSPA8P15 -1.242 3.8E-02 1.0000 AP000346.2 1.360 3.8E-02 1.0000 AC092111.1 0.881 3.8E-02 1.0000 AC006033.2 -2.796 3.9E-02 1.0000 FGB 1.119 3.9E-02 1.0000 DRICH1 2.897 3.9E-02 1.0000 ZSCAN12 -0.605 3.9E-02 1.0000 AC073896.2 -0.840 3.9E-02 1.0000 AC106869.1 -1.238 3.9E-02 1.0000 NDUFB2-AS1 0.646 3.9E-02 1.0000 AC010422.5 -3.988 4.0E-02 1.0000 AC008894.3 -0.655 4.0E-02 1.0000 AMY2B -3.974 4.0E-02 1.0000 USP32P2 -0.997 4.0E-02 1.0000 KIAA0319 -0.658 4.1E-02 1.0000 AC112503.1 -3.973 4.1E-02 1.0000 HNRNPA1P54 0.809 4.1E-02 1.0000 Z98884.2 -3.970 4.1E-02 1.0000 AC011468.1 -3.969 4.1E-02 1.0000 DAPL1 2.861 4.1E-02 1.0000 LINC01661 -3.965 4.1E-02 1.0000 AP004782.1 -3.963 4.1E-02 1.0000 AC019163.1 -3.962 4.1E-02 1.0000 LINC00654 -3.962 4.1E-02 1.0000 FAM225B -3.963 4.1E-02 1.0000 TKTL2 -3.961 4.1E-02 1.0000 AC025884.2 2.860 4.1E-02 1.0000 GAS5-AS1 -1.036 4.1E-02 1.0000 KLK8 -3.956 4.2E-02 1.0000

177

AC003101.3 2.857 4.2E-02 1.0000 TMEM179 -3.955 4.2E-02 1.0000 AC006160.1 -3.953 4.2E-02 1.0000 SLC26A10 -1.083 4.2E-02 1.0000 AC093627.5 -0.829 4.2E-02 1.0000 ISCA2P1 -3.948 4.2E-02 1.0000 EI24P2 -3.948 4.2E-02 1.0000 TUBB7P -1.844 4.2E-02 1.0000 ARHGAP31- -3.945 4.2E-02 1.0000 AS1 FAM96AP2 -3.945 4.2E-02 1.0000 AC069503.2 2.072 4.2E-02 1.0000 AC005358.2 -3.938 4.2E-02 1.0000 FLRT3 -3.938 4.2E-02 1.0000 AL645941.3 -3.937 4.2E-02 1.0000 SLC4A9 -3.937 4.2E-02 1.0000 OSGEPL1-AS1 -2.296 4.3E-02 1.0000 TREX1 0.986 4.3E-02 1.0000 DOCK10 -0.627 4.3E-02 1.0000 C11orf96 1.748 4.3E-02 1.0000 AC019322.3 -3.930 4.3E-02 1.0000 AC068544.1 -3.922 4.3E-02 1.0000 AL645941.1 -3.922 4.3E-02 1.0000 TPSB2 -3.922 4.3E-02 1.0000 AC009163.4 2.413 4.3E-02 1.0000 PPIAP29 -3.922 4.4E-02 1.0000 AC138035.2 -3.916 4.4E-02 1.0000 PDE4B -0.863 4.4E-02 1.0000 LAMC3 -0.873 4.4E-02 1.0000 GCNT2 2.431 4.4E-02 1.0000 AC010422.2 2.158 4.4E-02 1.0000 AC007613.1 0.885 4.5E-02 1.0000 AL451164.3 -3.911 4.5E-02 1.0000 AC064807.2 -1.703 4.5E-02 1.0000 AC021087.3 -1.494 4.5E-02 1.0000 MEMO1P1 0.806 4.6E-02 1.0000 FSTL1 -0.650 4.6E-02 1.0000 AC026202.3 2.142 4.6E-02 1.0000 FSTL5 1.313 4.6E-02 1.0000 AC004801.6 2.139 4.6E-02 1.0000 PLAG1 -1.410 4.7E-02 1.0000

178

IDH1-AS1 -0.749 4.7E-02 1.0000 AL138781.1 0.670 4.7E-02 1.0000 AL161729.1 -0.795 4.7E-02 1.0000 CACNA1A 1.712 4.7E-02 1.0000 AC093591.1 -3.878 4.7E-02 1.0000 AC011825.3 -3.885 4.8E-02 1.0000 AC241584.1 1.822 4.8E-02 1.0000 MUSTN1 -3.877 4.8E-02 1.0000 AC005154.3 -1.523 4.8E-02 1.0000 ANKRD10-IT1 2.380 4.8E-02 1.0000 HPCAL4 1.815 4.9E-02 1.0000 AC139491.6 3.744 4.9E-02 1.0000 EREG -0.714 4.9E-02 1.0000 SEMA5B -0.612 5.0E-02 1.0000 PNMA5 3.746 5.0E-02 1.0000 AC007744.1 3.740 5.0E-02 1.0000 NPIPA5 1.820 5.0E-02 1.0000

Table 8.2 Differentially expressed genes for MCF7 cells expressing Y694F-STAT5a, PRL vs untreated log10(fold change)>0.58, fold change >1.5.

Y694F-STAT5a Gene Log10(Fold Change) p value padj BCL6 -0.932 2.4E-06 0.0365 NR5A2 1.048 2.7E-04 0.4708 RIMBP3B -1.199 3.3E-04 0.4708 AC093668.1 -3.691 6.2E-04 0.4708 RCAN1 0.749 7.3E-04 0.4708 NABP1 0.745 2.0E-03 0.4708 AP003071.5 1.690 2.1E-03 0.4708 ACKR2 -1.021 2.2E-03 0.4708 AC010468.1 0.725 2.3E-03 0.4708 AL035078.4 1.084 2.4E-03 0.4708 ALOX12B -1.324 3.0E-03 0.4708 NR4A3 0.806 3.1E-03 0.4708 ZNF669 0.628 3.2E-03 0.4708 YEATS2-AS1 0.993 3.4E-03 0.4708 AC011005.1 -1.259 3.8E-03 0.4708 AC012676.5 1.332 3.9E-03 0.4708 AC069503.1 0.718 4.0E-03 0.4708

179

AL731684.2 1.318 4.0E-03 0.4708 NRTN -0.687 4.3E-03 0.4708 C16orf86 -0.806 4.4E-03 0.4708 CYP4F26P -1.333 4.7E-03 0.4708 MTCP1 0.766 4.8E-03 0.4708 WDR93 -0.785 4.8E-03 0.4708 RPL17-C18orf32 3.453 4.9E-03 0.4708 AC020910.5 0.957 5.4E-03 0.4708 AL137002.2 -0.682 5.7E-03 0.4708 PSCA -0.608 5.8E-03 0.4708 LRP5L 0.877 7.4E-03 0.4708 IGFLR1 -0.641 7.5E-03 0.4708 C22orf23 -0.677 7.6E-03 0.4708 AC036214.2 0.946 7.7E-03 0.4708 LINC01508 0.750 7.7E-03 0.4708 TYMP -0.691 8.5E-03 0.4708 DOCK9-AS2 -0.961 8.7E-03 0.4708 AC145207.2 -0.913 8.7E-03 0.4708 MIR503HG 0.982 8.9E-03 0.4708 DLEU2 0.614 9.1E-03 0.4708 GUSBP3 -1.722 9.1E-03 0.4708 RNU4-1 -1.780 9.2E-03 0.4708 KCNH1 1.255 9.5E-03 0.4708 POLR3G 0.717 1.0E-02 0.4708 AC010616.1 -0.654 1.0E-02 0.4708 AC131160.1 1.863 1.1E-02 0.4708 LINC01106 0.883 1.1E-02 0.4708 AC120114.4 0.668 1.1E-02 0.4708 ODF3B -0.818 1.1E-02 0.4708 AC245041.1 1.258 1.1E-02 0.4708 NUDT4B 2.013 1.1E-02 0.4708 PELI1 0.609 1.2E-02 0.4708 AC118553.2 -1.094 1.2E-02 0.4708 TLX1NB 1.312 1.2E-02 0.4708 AC100860.1 -1.106 1.3E-02 0.4708 AC068946.2 -0.904 1.3E-02 0.4708 ZNF550 0.611 1.3E-02 0.4708 NELFA 0.651 1.3E-02 0.4708 AC010894.2 -0.769 1.4E-02 0.4708

180

PPP5D1 0.792 1.4E-02 0.4708 MIF-AS1 -1.012 1.4E-02 0.4708 RN7SK -1.197 1.4E-02 0.4708 AC093827.5 1.368 1.5E-02 0.4708 C16orf87 0.640 1.5E-02 0.4708 P2RX5- -1.011 1.5E-02 0.4708 TAX1BP3 XKR9 1.182 1.5E-02 0.4708 SLC22A18AS -0.766 1.6E-02 0.4708 AC245389.2 0.856 1.6E-02 0.4708 AL021546.1 7.236 1.6E-02 0.4708 BZW1P2 0.651 1.6E-02 0.4708 PRSS53 1.596 1.6E-02 0.4708 MAK16 0.604 1.6E-02 0.4708 C4orf36 -0.736 1.6E-02 0.4708 CU633906.1 3.692 1.7E-02 0.4708 UNC13A 0.789 1.7E-02 0.4708 LINC01132 0.763 1.7E-02 0.4708 AL121900.2 2.317 1.8E-02 0.4708 AC090527.2 -8.073 1.8E-02 0.4708 PCDHA4 -1.151 1.8E-02 0.4708 LINC01144 -0.825 1.8E-02 0.4708 UBE2Q2P1 0.876 1.9E-02 0.4708 AC012531.2 6.647 1.9E-02 0.4708 AL161421.1 0.722 2.0E-02 0.4708 FAM76A 0.612 2.0E-02 0.4708 AC138904.1 -0.703 2.0E-02 0.4708 LINC00672 0.772 2.0E-02 0.4708 TCEA1P2 0.675 2.1E-02 0.4708 RGMA -0.641 2.1E-02 0.4708 RN7SL1 -1.069 2.1E-02 0.4708 SLC4A11 -0.690 2.1E-02 0.4708 RAB11FIP2 0.662 2.1E-02 0.4708 C1orf229 -0.874 2.1E-02 0.4708 AC007405.3 -0.757 2.2E-02 0.4708 FP236383.3 -1.023 2.2E-02 0.4708 TEN1-CDK3 -0.989 2.2E-02 0.4708 AL591684.2 0.903 2.3E-02 0.4708 C1QTNF3 1.002 2.3E-02 0.4708 LINC02256 0.918 2.4E-02 0.4708

181

AL355075.4 -1.463 2.4E-02 0.4708 ANKRD29 0.941 2.4E-02 0.4708 RN7SL2 -1.010 2.4E-02 0.4708 HSPA6 1.024 2.5E-02 0.4708 BASP1-AS1 -0.673 2.5E-02 0.4708 AL139289.2 -0.611 2.5E-02 0.4708 SNORD3A -1.479 2.6E-02 0.4708 GK 0.587 2.6E-02 0.4708 LIG4 0.604 2.7E-02 0.4708 MATN4 -0.911 2.7E-02 0.4708 TAF1A-AS1 0.745 2.7E-02 0.4708 RPPH1 -0.958 2.7E-02 0.4708 AL358113.1 1.663 2.8E-02 0.4708 C9orf50 -0.838 2.8E-02 0.4708 HOTAIRM1 0.690 2.8E-02 0.4708 XK 0.726 2.8E-02 0.4708 IGF2 -6.285 2.8E-02 0.4708 PNCK -0.616 2.9E-02 0.4708 MAP4K1 -0.637 2.9E-02 0.4708 HIST4H4 -0.761 2.9E-02 0.4708 AC010207.1 0.782 3.0E-02 0.4708 TENM4 0.929 3.0E-02 0.4708 RNU4-2 -1.527 3.0E-02 0.4708 NPIPA2 -2.588 3.0E-02 0.4708 AL139393.2 0.716 3.0E-02 0.4708 STAMBPL1 0.792 3.0E-02 0.4708 ZNF793 0.681 3.1E-02 0.4708 SNHG25 -0.671 3.1E-02 0.4708 CCT6P3 0.664 3.1E-02 0.4708 RN7SL3 -0.984 3.1E-02 0.4708 FP236383.2 -1.062 3.2E-02 0.4708 AC064799.1 0.956 3.3E-02 0.4708 GGN -0.681 3.3E-02 0.4708 AL354714.2 1.122 3.4E-02 0.4708 GPR156 0.624 3.4E-02 0.4708 ZNF519 0.616 3.5E-02 0.4708 AC118344.2 0.720 3.5E-02 0.4708 HERC2P5 0.951 3.5E-02 0.4708 TRIM71 0.742 3.5E-02 0.4708

182

IL17C -0.709 3.5E-02 0.4708 AOC3 0.866 3.6E-02 0.4708 AC009779.3 0.889 3.6E-02 0.4708 CXCR5 -0.751 3.6E-02 0.4708 SLC6A12 -1.775 3.7E-02 0.4708 FSTL1 0.735 3.8E-02 0.4708 RETREG1 0.655 3.8E-02 0.4708 CAT 2.232 3.8E-02 0.4708 CFL1P5 -0.633 3.9E-02 0.4708 AC026471.6 -1.127 4.0E-02 0.4708 VWCE 0.867 4.0E-02 0.4708 AC023509.3 1.787 4.0E-02 0.4708 GGT1 -1.827 4.0E-02 0.4708 FAM227A 0.706 4.0E-02 0.4708 ZUFSP 0.585 4.0E-02 0.4708 AC027575.2 -0.833 4.1E-02 0.4708 AC242376.2 1.013 4.1E-02 0.4708 AC105046.1 0.715 4.2E-02 0.4708 C1orf116 -0.716 4.2E-02 0.4708 TMEM191C -0.911 4.3E-02 0.4708 DDIT4-AS1 -0.897 4.3E-02 0.4708 LINC01124 -0.726 4.5E-02 0.4708 CASC11 0.882 4.5E-02 0.4708 FAM157C 1.422 4.5E-02 0.4708 CARD9 -0.697 4.6E-02 0.4708 ALG1L9P -0.631 4.7E-02 0.4708 VILL -0.683 4.7E-02 0.4708 AC009133.2 -0.763 4.7E-02 0.4708 RMRP -2.168 4.7E-02 0.4708 AC011603.2 -0.597 4.8E-02 0.4708 ADRA2B -0.816 4.8E-02 0.4708 CDKN2B-AS1 0.684 4.8E-02 0.4708 AL358781.1 -0.601 4.9E-02 0.4708 SDHAP2 -7.790 4.9E-02 0.4708 TMEM151A -0.704 4.9E-02 0.4708 PIK3CD-AS2 -0.895 4.9E-02 0.4708 AL121839.2 0.909 5.0E-02 0.4708

183

Table 8.3 Differentially expressed genes for MCF7 cells expressing S726A-STAT5a, PRL vs untreated log10(fold change)>0.58, fold change >1.5.

S726A-STAT5a Gene Log10(Fold Change) p value padj BCL6 -2.120 0.000 0.0000 NR4A3 2.102 0.000 0.0147 SRSF5 0.828 0.000 0.0501 RASL11A 1.292 0.000 0.0501 ADORA2A 8.622 0.000 0.1365 RCAN1 0.994 0.000 0.1365 MB21D2 0.919 0.000 0.1481 CISH 0.625 0.000 0.1841 MFSD2A 1.480 0.000 0.1841 NR5A2 1.155 0.000 0.2468 SH3RF1 0.808 0.000 0.2998 NT5DC3 0.619 0.001 0.4059 WNT6 -0.672 0.001 0.4059 AP001057.1 0.690 0.002 0.4059 NKX3-2 -1.084 0.002 0.4059 AC124798.1 1.694 0.002 0.4059 CD164L2 -1.553 0.002 0.4059 AC027307.2 -0.645 0.002 0.4059 FAM156B -0.988 0.002 0.4059 TP53TG3C -6.523 0.002 0.4059 AL139393.2 1.285 0.002 0.4059 TMCC2 -0.847 0.002 0.4059 TNFAIP8 0.846 0.002 0.4059 UBE2V1P2 -1.546 0.003 0.4059 GFOD1 0.625 0.003 0.4059 NPIPB13 0.935 0.003 0.4059 AL050341.2 0.904 0.003 0.4059 RAET1G -0.940 0.003 0.4059 FOSL1 0.613 0.003 0.4059 MOGS -1.039 0.004 0.4059 TERT -1.107 0.004 0.4059 MELTF-AS1 -0.640 0.004 0.4059 SCART1 0.928 0.004 0.4059 FFAR2 -0.713 0.004 0.4059

184

GPR156 0.969 0.004 0.4059 KLF6 0.789 0.004 0.4059 AL109840.2 -0.714 0.004 0.4059 ZNF775 -0.597 0.005 0.4059 NKX2-8 -0.855 0.005 0.4059 AL442663.3 1.020 0.005 0.4059 MAP1LC3B2 1.133 0.005 0.4059 AMD1 0.599 0.005 0.4059 SIX2 -0.746 0.005 0.4059 AC009118.2 1.432 0.005 0.4059 AL117332.1 -0.790 0.005 0.4059 SMIM27 -0.615 0.005 0.4059 NABP1 0.707 0.005 0.4059 ARRDC4 0.713 0.005 0.4059 C3orf80 1.356 0.005 0.4059 EPGN 1.453 0.005 0.4059 AC113935.1 0.596 0.006 0.4059 VTCN1 0.622 0.006 0.4059 COL28A1 0.952 0.006 0.4059 AP000438.1 1.096 0.006 0.4059 FRMD6 0.699 0.006 0.4059 ATXN7 1.497 0.006 0.4059 C5orf30 0.686 0.006 0.4059 AL138976.2 1.133 0.006 0.4059 SGPP1 0.665 0.006 0.4059 ELOVL4 0.794 0.007 0.4059 AC138932.2 0.722 0.007 0.4059 AC105339.2 1.795 0.007 0.4059 IL12A 1.336 0.007 0.4059 NXT2 0.694 0.007 0.4059 RPP40 0.631 0.007 0.4059 FRMD4B 0.616 0.008 0.4059 SMIM13 0.604 0.008 0.4059 AC110285.2 0.975 0.008 0.4059 RIMBP3 -0.668 0.008 0.4059 LINC01836 -0.594 0.008 0.4059 ZBTB8A 0.744 0.008 0.4059 MT-ND6 0.697 0.008 0.4059 MIR503HG 0.930 0.008 0.4059

185

GABRD -0.803 0.008 0.4059 AC024592.3 -0.741 0.009 0.4059 AC113189.4 -0.765 0.009 0.4059 AC105052.4 0.766 0.009 0.4059 AC010531.6 -0.901 0.009 0.4059 ZNF815P -0.691 0.010 0.4059 Z83844.3 -0.960 0.010 0.4059 PDE1A -1.166 0.010 0.4059 ZNF559 0.664 0.010 0.4059 TUBB8P12 -1.582 0.010 0.4059 AC002350.1 1.086 0.010 0.4059 FAM157C -1.865 0.010 0.4059 GPR68 -0.621 0.010 0.4059 TAF1A-AS1 0.842 0.010 0.4059 PCDHAC2 0.802 0.011 0.4059 LINC00982 -0.980 0.011 0.4059 CALHM3 -1.119 0.011 0.4059 PSCA -0.600 0.011 0.4059 MIR22HG 0.690 0.011 0.4059 AC099336.2 -0.927 0.011 0.4059 THSD8 -0.661 0.012 0.4059 SPECC1L- -9.189 0.012 0.4059 ADORA2A GGT1 -1.491 0.012 0.4059 ZNF57 0.608 0.012 0.4059 CCT6B 0.644 0.012 0.4059 GDF1 -0.646 0.012 0.4059 MST1P2 1.281 0.012 0.4059 CSPG4 -1.241 0.012 0.4059 NR4A1 0.759 0.012 0.4059 MZT1 0.661 0.012 0.4059 PPAT 0.654 0.012 0.4059 EGR4 0.868 0.012 0.4059 KLF9 0.684 0.013 0.4059 AL139353.1 -1.217 0.013 0.4059 IGF2 4.557 0.013 0.4059 PRSS36 -0.971 0.013 0.4059 AC084018.2 0.815 0.013 0.4059 AC145207.3 -1.125 0.013 0.4059 BAG2 0.946 0.014 0.4059

186

CASS4 0.884 0.014 0.4059 PFKFB4 -0.582 0.014 0.4059 ARG2 0.798 0.014 0.4059 DYNC1I1 0.957 0.014 0.4059 AC004264.1 0.746 0.015 0.4059 ETV2 -0.625 0.015 0.4059 BSCL2 -0.583 0.015 0.4059 HOXA13 1.795 0.015 0.4059 GNL3LP1 0.938 0.015 0.4059 AC015871.3 0.769 0.016 0.4059 LINC01132 0.884 0.016 0.4059 CDRT4 -2.642 0.016 0.4059 IKBKGP1 -0.691 0.016 0.4059 AC233280.1 1.141 0.016 0.4059 AC091057.1 0.937 0.016 0.4059 SMG1P2 0.603 0.017 0.4059 ZNF23 0.671 0.017 0.4059 PSD2 0.881 0.017 0.4059 ZNF81 0.666 0.017 0.4059 AC087632.1 -0.661 0.017 0.4059 ITPKA -0.601 0.018 0.4059 ZDHHC11B 0.880 0.018 0.4059 PSPN -0.723 0.018 0.4059 C4orf48 -0.587 0.018 0.4059 AC048341.1 0.962 0.018 0.4059 REM2 0.672 0.018 0.4059 AC106038.1 0.585 0.018 0.4059 AL117339.5 1.321 0.018 0.4059 AC124944.3 0.655 0.018 0.4059 ARHGAP26 -1.024 0.018 0.4059 ANKRD6 0.584 0.018 0.4059 WDR43 0.602 0.019 0.4059 AC120114.4 0.904 0.019 0.4059 EGR2 0.871 0.019 0.4059 TREX2 -0.828 0.019 0.4059 LINC00672 1.007 0.019 0.4059 KCTD12 0.796 0.019 0.4059 WDFY4 -0.787 0.019 0.4059 LINC01816 -0.602 0.019 0.4059

187

AP000812.5 -2.567 0.020 0.4059 HOTAIRM1 0.785 0.020 0.4059 AL035563.1 1.091 0.020 0.4059 AC009779.3 1.039 0.020 0.4059 AC009053.2 -0.837 0.020 0.4059 RNFT1-DT 0.634 0.021 0.4059 ZNF587B 0.597 0.021 0.4059 AC022210.1 -1.116 0.021 0.4059 CHRM1 -0.599 0.021 0.4059 TGFBR3L -0.649 0.021 0.4059 AC016027.1 0.635 0.022 0.4059 MSH5-SAPCD1 2.042 0.022 0.4059 RGPD2 2.086 0.022 0.4059 AC131009.3 0.889 0.022 0.4059 AC108488.1 1.434 0.023 0.4059 AC005829.1 -1.194 0.024 0.4059 SLC26A2 0.659 0.024 0.4059 GCNT1 0.759 0.024 0.4059 THUMPD2 0.597 0.024 0.4059 ZNF433 0.586 0.025 0.4059 RPPH1 -0.749 0.025 0.4059 DRAXIN -0.794 0.025 0.4059 PAQR5 0.640 0.025 0.4059 LYPLA1 0.645 0.025 0.4059 FNDC4 -0.612 0.025 0.4059 SLC25A16 0.598 0.025 0.4059 MAMSTR -0.783 0.025 0.4059 PCDHGA7 1.140 0.025 0.4059 B3GNT6 0.885 0.025 0.4059 AP001486.2 0.910 0.025 0.4059 BACH1 0.699 0.026 0.4059 AL356019.2 0.935 0.026 0.4059 LPAR5 -0.675 0.026 0.4059 SLC25A21 0.586 0.027 0.4059 CYP24A1 1.170 0.028 0.4059 GOLGA6L10 0.968 0.028 0.4059 C21orf91 0.800 0.029 0.4059 SOWAHA 0.743 0.029 0.4059 CCDC151 -0.613 0.029 0.4059

188

NIPAL1 0.757 0.029 0.4059 MAGED4 -2.548 0.029 0.4059 CABCOCO1 0.786 0.030 0.4059 SLX1A -1.023 0.031 0.4059 ERC2 0.736 0.031 0.4059 DLEU2 0.738 0.031 0.4059 AC092279.2 0.872 0.032 0.4059 AC020917.4 0.991 0.032 0.4059 NALT1 0.685 0.032 0.4059 CDH15 -0.736 0.032 0.4059 PEG13 -0.602 0.032 0.4059 GOLGA2P5 0.599 0.033 0.4059 AP003071.5 1.035 0.033 0.4059 AL731569.1 0.733 0.033 0.4059 FAM81A 0.745 0.033 0.4059 AL359258.3 0.693 0.033 0.4059 ZC3H12C 0.821 0.033 0.4059 MINOS1-NBL1 2.106 0.034 0.4059 AC004997.1 -1.554 0.034 0.4059 NOXO1 -0.646 0.035 0.4059 AC004877.1 0.766 0.035 0.4059 AC025594.2 -0.614 0.035 0.4059 PSMC6 0.789 0.035 0.4059 RND3 0.603 0.035 0.4059 BCDIN3D-AS1 -0.926 0.035 0.4059 AC022968.1 0.901 0.036 0.4059 AL354714.2 0.951 0.036 0.4059 STK38L 0.635 0.036 0.4059 AL353763.2 0.763 0.036 0.4059 PRKAA2 0.823 0.036 0.4059 ROBO1 1.435 0.037 0.4059 AKR7A3 0.915 0.037 0.4059 CRYBG2 -0.623 0.037 0.4059 AL451062.3 -0.881 0.037 0.4059 TENM4 -1.062 0.037 0.4059 LINC00641 0.742 0.037 0.4059 AL133523.1 0.605 0.037 0.4059 AL137025.1 -0.771 0.037 0.4059 TCF7 -0.855 0.038 0.4059

189

AC106820.4 1.005 0.038 0.4059 AC079848.1 0.962 0.038 0.4059 FSTL1 -0.581 0.039 0.4059 AC093525.8 0.759 0.039 0.4059 AC005034.3 0.620 0.039 0.4059 TLR5 -1.103 0.039 0.4059 LINC00852 0.783 0.039 0.4059 MYLK4 0.855 0.039 0.4059 GIN1 0.620 0.039 0.4059 TCTE3 0.848 0.040 0.4059 SLC12A9-AS1 -0.627 0.040 0.4059 LINC02288 0.774 0.040 0.4059 NELFA -0.592 0.040 0.4059 FP236383.2 -0.638 0.041 0.4059 AC126177.4 -0.623 0.041 0.4059 COG8 2.652 0.041 0.4059 AC099778.1 0.691 0.041 0.4059 GPCPD1 0.590 0.042 0.4059 CYP4F22 -0.859 0.042 0.4059 SYT5 -0.886 0.042 0.4059 ULK4P1 2.728 0.043 0.4059 C3orf52 0.677 0.043 0.4059 AC006330.1 0.841 0.043 0.4059 ZNF596 0.833 0.043 0.4059 GOLGA8A 0.868 0.043 0.4059 ARC 0.618 0.043 0.4059 LRRC4B -0.786 0.043 0.4059 AC131160.1 0.955 0.044 0.4059 TMEM270 -0.966 0.044 0.4059 FOXA2 -0.615 0.045 0.4059 FKBP1C -0.657 0.045 0.4059 BOLA3-AS1 -0.750 0.045 0.4059 AL355075.4 -1.139 0.045 0.4059 RDM1 -0.601 0.046 0.4059 LETM2 -0.612 0.046 0.4059 ADRA2B -0.642 0.046 0.4059 SYNC 1.170 0.047 0.4059 C20orf144 -0.776 0.047 0.4059 GPR75 0.899 0.047 0.4059

190

TEX48 -0.869 0.047 0.4059 ALDH1L2 0.838 0.048 0.4059 AP003469.2 -1.066 0.048 0.4059 FAM95C 0.636 0.048 0.4059 AC105046.1 0.830 0.048 0.4059 NOTCH1 1.446 0.048 0.4059 SMIM22 -0.638 0.049 0.4059 LRP12 0.690 0.049 0.4059 MMP9 -0.612 0.049 0.4059 PPTC7 0.609 0.049 0.4059 KATNBL1 0.599 0.049 0.4059 ULK4P2 -2.382 0.049 0.4059 NFKBIZ 0.599 0.049 0.4059 AC012321.1 -0.980 0.049 0.4059 AC005225.4 0.626 0.049 0.4059 CCDC103 -1.006 0.050 0.4059 FHAD1 0.745 0.050 0.4059

Table 8.4 Differentially expressed genes for MCF7 cells expressing S780A-STAT5a, PRL vs untreated log10(fold change)>0.58, fold change >1.5.

S780A-STAT5a Gene Log10(Fold Change) p value padj BCL6 -2.423 0.0000 0.0002 AC105046.1 1.589 0.0002 1.0000 FAM157C 2.935 0.0003 1.0000 SOCS2 0.803 0.0003 1.0000 RASL11A 1.047 0.0005 1.0000 NPIPB13 1.167 0.0006 1.0000 MIR503HG 1.575 0.0006 1.0000 RCAN1 0.935 0.0007 1.0000 C8orf59 0.703 0.0008 1.0000 COX7B 0.675 0.0009 1.0000 CISH 0.747 0.0011 1.0000 RSL24D1 0.893 0.0011 1.0000 AC093668.1 -4.056 0.0012 1.0000 AL355075.4 1.917 0.0017 1.0000 EEF1E1 0.816 0.0018 1.0000 CRISPLD2 0.824 0.0019 1.0000

191

MRPL1 0.860 0.0028 1.0000 LINC01132 1.060 0.0029 1.0000 MND1 0.676 0.0031 1.0000 CABCOCO1 0.773 0.0033 1.0000 RN7SL1 1.194 0.0034 1.0000 SH3RF1 0.742 0.0035 1.0000 C15orf41 2.350 0.0036 1.0000 RPPH1 1.542 0.0037 1.0000 ITGB3BP 0.946 0.0038 1.0000 SLC22A20P 1.253 0.0038 1.0000 LINC00345 1.045 0.0039 1.0000 JKAMP 0.586 0.0041 1.0000 TXNDC9 0.916 0.0044 1.0000 PSMA2 0.674 0.0045 1.0000 DNAJC25-GNG10 0.974 0.0045 1.0000 C18orf65 -1.276 0.0046 1.0000 CCNC 0.702 0.0047 1.0000 SNORA73B 1.827 0.0048 1.0000 ETFRF1 0.874 0.0051 1.0000 RPL7P1 0.678 0.0053 1.0000 CAPZA2 0.900 0.0058 1.0000 MRPL13 0.721 0.0059 1.0000 AC092687.3 1.015 0.0059 1.0000 COMMD8 0.868 0.0060 1.0000 RPL21P119 -8.213 0.0064 1.0000 AC107068.1 1.190 0.0068 1.0000 EIF3E 0.725 0.0071 1.0000 AP000525.1 0.895 0.0072 1.0000 RPL5P34 0.763 0.0073 1.0000 SYT15 2.309 0.0075 1.0000 NR4A3 1.072 0.0075 1.0000 AL662899.2 0.866 0.0076 1.0000 AL590652.1 1.051 0.0081 1.0000 RPL22L1 0.582 0.0088 1.0000 ACKR2 -1.011 0.0091 1.0000 EPHX4 0.807 0.0092 1.0000 ICOSLG 0.585 0.0095 1.0000 ANAPC10 0.688 0.0096 1.0000 AL121790.1 0.910 0.0099 1.0000

192

KBTBD8 1.460 0.0100 1.0000 NABP1 1.089 0.0103 1.0000 XRCC4 0.983 0.0105 1.0000 TPRKB 0.581 0.0112 1.0000 AC073842.2 1.023 0.0112 1.0000 RN7SK 1.342 0.0114 1.0000 AC244197.3 0.612 0.0115 1.0000 PYGM 0.791 0.0115 1.0000 RPL26P19 0.621 0.0118 1.0000 AC092139.1 1.425 0.0120 1.0000 SYNJ2BP-COX16 1.379 0.0121 1.0000 COPS2 0.834 0.0123 1.0000 AL139393.2 1.053 0.0124 1.0000 SERPINF1 0.607 0.0127 1.0000 AL009174.1 0.716 0.0128 1.0000 ZNF138 0.946 0.0129 1.0000 AL162151.2 0.875 0.0130 1.0000 CENPQ 0.986 0.0130 1.0000 AL121845.3 0.688 0.0131 1.0000 GCNT1 0.846 0.0134 1.0000 COG8 8.991 0.0134 1.0000 RN7SL2 1.129 0.0138 1.0000 RUSC1-AS1 0.711 0.0143 1.0000 ZNF98 0.998 0.0143 1.0000 TSHZ3 0.693 0.0144 1.0000 TDH 1.188 0.0144 1.0000 AC099568.2 0.637 0.0145 1.0000 AC011497.2 0.797 0.0148 1.0000 Z95331.1 -1.238 0.0149 1.0000 KIAA1586 0.893 0.0152 1.0000 AL031009.1 1.496 0.0156 1.0000 EIF4BP7 -0.739 0.0164 1.0000 AL121845.2 -6.910 0.0164 1.0000 LRRC37A11P -0.844 0.0165 1.0000 VN1R1 1.007 0.0166 1.0000 AC159540.2 -3.949 0.0171 1.0000 PDZD7 -1.087 0.0174 1.0000 AC126474.2 0.908 0.0174 1.0000 AC027097.1 0.959 0.0175 1.0000

193

AC073585.1 1.196 0.0176 1.0000 GPC2 0.671 0.0181 1.0000 LINC02163 -1.054 0.0182 1.0000 AC008560.1 1.009 0.0196 1.0000 SPECC1L-ADORA2A -8.442 0.0201 1.0000 LINC01376 0.836 0.0206 1.0000 STYK1 1.193 0.0213 1.0000 HIST2H3C 8.892 0.0213 1.0000 FAM171B 0.854 0.0217 1.0000 AC018638.2 1.018 0.0219 1.0000 MSMB 0.641 0.0219 1.0000 AC113189.2 6.765 0.0220 1.0000 MZT1 0.818 0.0221 1.0000 ACAP1 0.784 0.0226 1.0000 TNFAIP8L3 -1.080 0.0230 1.0000 CU633904.1 7.192 0.0231 1.0000 SHISA2 -0.608 0.0231 1.0000 AL021368.2 1.813 0.0241 1.0000 EDN1 -0.710 0.0243 1.0000 WASH4P 0.690 0.0244 1.0000 AL137002.2 -0.661 0.0244 1.0000 SNORD3A 1.021 0.0246 1.0000 STX17-AS1 0.864 0.0250 1.0000 AC040162.1 0.590 0.0257 1.0000 MBLAC2 0.839 0.0259 1.0000 RPL23AP53 1.000 0.0262 1.0000 CCDC112 0.642 0.0267 1.0000 HAT1 0.652 0.0267 1.0000 MFSD2A 0.875 0.0268 1.0000 HIST1H4E 1.476 0.0275 1.0000 ID2-AS1 0.939 0.0278 1.0000 SCOC 0.597 0.0279 1.0000 FBLN2 -0.658 0.0281 1.0000 NPHP3-ACAD11 7.734 0.0282 1.0000 GGT5 1.088 0.0286 1.0000 C20orf204 0.867 0.0287 1.0000 AC141586.3 -0.868 0.0287 1.0000 AL603750.1 0.730 0.0290 1.0000 AC148477.2 0.732 0.0292 1.0000

194

GNG10 0.621 0.0297 1.0000 ALOX12B 1.494 0.0298 1.0000 GNL3LP1 1.131 0.0300 1.0000 JAGN1 1.069 0.0307 1.0000 CT62 -0.819 0.0308 1.0000 RAD51AP1 0.875 0.0308 1.0000 C12orf29 0.660 0.0314 1.0000 TGDS 0.644 0.0315 1.0000 AL359878.1 0.949 0.0319 1.0000 AC083843.3 1.666 0.0322 1.0000 TTC34 0.800 0.0328 1.0000 CU634019.1 -7.172 0.0328 1.0000 DVL3 0.803 0.0331 1.0000 MAB21L3 1.105 0.0334 1.0000 AL080276.2 0.672 0.0337 1.0000 SPTB 1.708 0.0339 1.0000 FHAD1 0.931 0.0340 1.0000 LINC00346 -0.743 0.0345 1.0000 DLL4 0.594 0.0346 1.0000 GIN1 0.839 0.0360 1.0000 TNFAIP8 0.808 0.0361 1.0000 ZNF195 0.676 0.0363 1.0000 TYW3 0.595 0.0371 1.0000 AC079781.5 -0.905 0.0371 1.0000 AL122058.1 0.968 0.0373 1.0000 AC027644.4 1.298 0.0379 1.0000 RPL31P63 0.829 0.0379 1.0000 PRDM12 0.770 0.0380 1.0000 PDIA2 0.836 0.0380 1.0000 TRAPPC13 0.747 0.0380 1.0000 DBF4 0.756 0.0381 1.0000 RN7SL3 1.253 0.0383 1.0000 ITGB2-AS1 1.035 0.0387 1.0000 Z93930.2 0.783 0.0390 1.0000 AC005520.1 0.883 0.0391 1.0000 AC004951.4 0.768 0.0394 1.0000 AC138811.2 6.239 0.0398 1.0000 AL139011.2 0.904 0.0406 1.0000 MYNN 0.628 0.0409 1.0000

195

BBIP1 0.637 0.0413 1.0000 GOLGA8H 0.790 0.0417 1.0000 MT-ND6 0.648 0.0419 1.0000 AC215522.2 0.669 0.0425 1.0000 AC004477.1 0.763 0.0425 1.0000 RGPD1 3.280 0.0428 1.0000 AC138932.2 0.668 0.0439 1.0000 AC008543.1 1.141 0.0441 1.0000 DUSP10 -0.630 0.0443 1.0000 AC093525.8 0.729 0.0444 1.0000 AC006330.1 1.030 0.0444 1.0000 FAM226B 1.211 0.0445 1.0000 AC105052.1 2.729 0.0446 1.0000 IFIH1 2.067 0.0447 1.0000 LCA5L 0.671 0.0449 1.0000 AC122688.3 0.594 0.0453 1.0000 SSX2IP 0.902 0.0456 1.0000 AP001486.2 1.423 0.0458 1.0000 EMILIN1 -0.582 0.0462 1.0000 SNHG22 0.810 0.0463 1.0000 SPEF2 0.691 0.0467 1.0000 AC128689.1 -0.923 0.0469 1.0000 AC146944.4 1.703 0.0475 1.0000 VIM-AS1 0.942 0.0476 1.0000 OTUD6B 0.763 0.0477 1.0000 IL1R1 0.619 0.0478 1.0000 POTEKP 1.623 0.0479 1.0000 TMED5 0.613 0.0481 1.0000 TMEM161B 0.693 0.0484 1.0000 AL645728.1 0.600 0.0485 1.0000 PHOSPHO1 1.706 0.0489 1.0000 AC011379.2 0.774 0.0494 1.0000 AL353763.2 0.861 0.0494 1.0000 AC073610.3 1.970 0.0496 1.0000 AC083799.1 0.591 0.0496 1.0000 WFDC2 0.614 0.0498 1.0000

196

Table 8.5 Details for top 10 enriched oncogenic signatures for each STAT5a mutant in the presence of PRL. Terms were ranked by normalized enrichment score (NES).

Enriched Oncogenic Signatures WT-STAT5a + SIZE ES NES NOM FDR RANK LEADING PRL treatment p-val q-val AT EDGE MAX KRAS.600.LUNG 103 0.536 1.762 0.000 0.059 1226 tags=27%, .BREAST_UP.V1 list=9%, _UP signal=30% KRAS.BREAST_ 54 0.568 1.662 0.000 0.075 1226 tags=33%, UP.V1_UP list=9%, signal=36% IL15_UP.V1_UP 132 0.438 1.521 0.000 0.096 2459 tags=36%, list=17%, signal=43% KRAS.600_UP.V 115 0.462 1.551 0.000 0.102 1288 tags=25%, 1_UP list=9%, signal=28% IL2_UP.V1_UP 121 0.438 1.486 0.008 0.103 2280 tags=31%, list=16%, signal=37% BCAT.100_UP.V 24 0.602 1.527 0.035 0.105 1068 tags=25%, 1_DN list=8%, signal=27% MEL18_DN.V1_ 95 0.479 1.593 0.007 0.115 2287 tags=34%, UP list=16%, signal=40% CTIP_DN.V1_UP 48 0.499 1.487 0.026 0.116 284 tags=15%, list=2%, signal=15% KRAS.LUNG.BR 51 0.528 1.555 0.012 0.124 1604 tags=35%, EAST_UP.V1_U list=11%, P signal=40% KRAS.50_UP.V1 20 0.590 1.441 0.089 0.124 957 tags=30%, _UP list=7%, signal=32% Y694F-STAT5a + SIZE ES NES NOM FDR RANK LEADING PRL treatment p-val q-val AT EDGE MAX PIGF_UP.V1_UP 152 0.537 1.939 0.000 0.002 4789 tags=71%, list=34%, signal=106%

197

ERBB2_UP.V1_ 187 0.501 1.848 0.000 0.007 4312 tags=62%, DN list=30%, signal=88% STK33_UP 199 0.480 1.756 0.000 0.009 3911 tags=47%, list=28%, signal=64% TBK1.DF_DN 263 0.465 1.748 0.000 0.009 4930 tags=59%, list=35%, signal=89% CSR_EARLY_U 139 0.488 1.768 0.000 0.010 5121 tags=56%, P.V1_UP list=36%, signal=87% VEGF_A_UP.V1 165 0.484 1.770 0.000 0.012 4785 tags=62%, _DN list=34%, signal=92% STK33_NOMO_ 206 0.476 1.772 0.000 0.015 3939 tags=44%, UP list=28%, signal=60% JAK2_DN.V1_D 108 0.487 1.674 0.003 0.021 4548 tags=53%, N list=32%, signal=77% RB_P130_DN.V1 118 0.465 1.633 0.000 0.028 5067 tags=55%, _DN list=36%, signal=85% STK33_SKM_UP 192 0.446 1.640 0.000 0.029 2661 tags=35%, list=19%, signal=42% S726A-STAT5a + SIZE ES NES NOM FDR RANK LEADING PRL treatment p-val q-val AT EDGE MAX ERBB2_UP.V1_ 187 0.516 1.964 0.000 0.002 4158 tags=58%, DN list=29%, signal=81% JAK2_DN.V1_D 110 0.532 1.871 0.000 0.006 2588 tags=38%, N list=18%, signal=46% STK33_UP 201 0.487 1.852 0.000 0.006 3792 tags=50%, list=27%, signal=68% PIGF_UP.V1_UP 151 0.516 1.882 0.000 0.008 4788 tags=64%, list=34%, signal=96% CSR_EARLY_U 138 0.485 1.761 0.000 0.014 4506 tags=49%, P.V1_UP list=32%, signal=71%

198

STK33_NOMO_ 207 0.453 1.725 0.000 0.017 3642 tags=44%, UP list=26%, signal=58% LTE2_UP.V1_D 179 0.446 1.711 0.000 0.017 3804 tags=42%, N list=27%, signal=57% VEGF_A_UP.V1 164 0.450 1.686 0.000 0.018 5316 tags=63%, _DN list=38%, signal=100% TBK1.DF_DN 264 0.436 1.695 0.000 0.018 4870 tags=57%, list=34%, signal=85% EGFR_UP.V1_U 172 0.464 1.728 0.000 0.019 3270 tags=42%, P list=23%, signal=55% S780A-STAT5a + SIZE ES NES NOM FDR RANK LEADING PRL treatment p-val q-val AT EDGE MAX ERBB2_UP.V1_ 188 0.522 1.547 0.000 0.230 3174 tags=48%, DN list=22%, signal=61% PIGF_UP.V1_UP 153 0.540 1.578 0.001 0.285 3987 tags=57%, list=28%, signal=78% VEGF_A_UP.V1 166 0.499 1.460 0.006 0.514 3694 tags=46%, _DN list=26%, signal=61% RB_P130_DN.V1 119 0.491 1.387 0.023 0.886 2113 tags=20%, _DN list=15%, signal=23% WNT_UP.V1_UP 100 0.496 1.378 0.030 0.784 3130 tags=25%, list=22%, signal=32% P53_DN.V1_DN 108 0.472 1.332 0.046 0.879 618 tags=12%, list=4%, signal=12% IL15_UP.V1_UP 132 0.450 1.287 0.067 0.929 3141 tags=34%, list=22%, signal=43% CRX_NRL_DN.V 87 0.482 1.321 0.070 0.850 1978 tags=16%, 1_DN list=14%, signal=19% ESC_V6.5_UP_E 122 0.437 1.248 0.082 0.910 2969 tags=26%, ARLY.V1_UP list=21%, signal=33%

199

PTEN_DN.V1_D 73 0.483 1.300 0.095 0.913 2213 tags=29%, N list=16%, signal=34%

Table 8.6 Details for top 10 enriched oncogenic signatures for each untreated STAT5a mutant. Terms were ranked by normalized enrichment score (NES).

Enriched Oncogenic Signatures WT-STAT5a SIZE ES NES NOM FDR RANK LEADING Untreated p-val q-val AT EDGE MAX BCAT_BILD_ET_ 42 -0.623 -1.614 0.013 0.233 3544 tags=55%, AL_DN list=25%, signal=73% PIGF_UP.V1_UP 152 -0.505 -1.553 0.003 0.266 4628 tags=51%, list=33%, signal=74% LEF1_UP.V1_DN 134 -0.477 -1.438 0.005 0.426 2786 tags=30%, list=20%, signal=37% CAHOY_OLIGO 73 -0.506 -1.419 0.042 0.432 1672 tags=18%, DENDROCUTIC list=12%, signal=20% TGFB_UP.V1_DN 147 -0.476 -1.456 0.006 0.432 2729 tags=27%, list=19%, signal=33% CRX_DN.V1_UP 87 -0.462 -1.312 0.068 0.478 3009 tags=34%, list=21%, signal=44% JAK2_DN.V1_DN 108 -0.446 -1.312 0.059 0.511 3562 tags=41%, list=25%, signal=54% SRC_UP.V1_DN 117 -0.437 -1.295 0.061 0.520 1644 tags=20%, list=12%, signal=22% JNK_DN.V1_UP 93 -0.512 -1.466 0.011 0.521 2978 tags=37%, list=21%, signal=46% PRC2_SUZ12_UP 100 -0.450 -1.314 0.070 0.545 3442 tags=36%, .V1_UP list=24%, signal=47% Y694F-STAT5a SIZE ES NES NOM FDR RANK LEADING Untreated p-val q-val AT EDGE MAX

200

KRAS.300_UP.V1 65 -0.584 -2.078 0.000 0.000 2179 tags=48%, _DN list=15%, signal=56% RELA_DN.V1_D 61 -0.552 -1.909 0.000 0.006 1578 tags=34%, N list=11%, signal=39% KRAS.600_UP.V1 117 -0.497 -1.938 0.000 0.006 2210 tags=42%, _DN list=16%, signal=49% KRAS.LUNG_UP. 67 -0.514 -1.850 0.000 0.010 2042 tags=46%, V1_DN list=14%, signal=54% KRAS.LUNG.BR 68 -0.492 -1.803 0.000 0.010 2210 tags=43%, EAST_UP.V1_DN list=16%, signal=50% P53_DN.V2_UP 61 -0.495 -1.778 0.000 0.010 2048 tags=36%, list=14%, signal=42% KRAS.600.LUNG. 126 -0.460 -1.814 0.000 0.011 2210 tags=39%, BREAST_UP.V1_ list=16%, DN signal=46% KRAS.BREAST_ 62 -0.483 -1.719 0.003 0.016 2973 tags=45%, UP.V1_DN list=21%, signal=57% KRAS.50_UP.V1_ 21 -0.585 -1.646 0.021 0.027 2099 tags=52%, DN list=15%, signal=61% ATM_DN.V1_UP 66 -0.455 -1.623 0.000 0.033 1763 tags=30%, list=12%, signal=34% S726A-STAT5a SIZE ES NES NOM FDR RANK LEADING Untreated p-val q-val AT EDGE MAX P53_DN.V2_UP 64.00 -0.553 -2.019 0.000 0.001 1310 tags=36%, 0 list=9%, signal=39% KRAS.AMP.LUN 80.00 -0.572 -2.170 0.000 0.001 2231 tags=48%, G_UP.V1_DN 0 list=16%, signal=56% KRAS.50_UP.V1_ 20.00 -0.648 -1.863 0.003 0.006 1626 tags=45%, DN 0 list=11%, signal=51% KRAS.LUNG_UP. 67.00 -0.465 -1.694 0.003 0.028 1930 tags=42%, V1_DN 0 list=14%, signal=48%

201

PRC2_EED_UP.V 173.0 -0.410 -1.710 0.000 0.029 4219 tags=57%, 1_DN 00 list=30%, signal=80% PTEN_DN.V1_UP 81.00 -0.439 -1.654 0.000 0.030 1296 tags=26%, 0 list=9%, signal=28% RELA_DN.V1_D 59.00 -0.462 -1.655 0.003 0.033 1132 tags=27%, N 0 list=8%, signal=29% PIGF_UP.V1_DN 112.0 -0.442 -1.749 0.000 0.033 2211 tags=28%, 00 list=16%, signal=33% KRAS.KIDNEY_ 49.00 -0.491 -1.663 0.009 0.034 2593 tags=39%, UP.V1_DN 0 list=18%, signal=47% SRC_UP.V1_UP 103.0 -0.444 -1.711 0.000 0.035 2535 tags=44%, 00 list=18%, signal=53% S780A-STAT5a SIZE ES NES NOM FDR RANK LEADING Untreated p-val q-val AT EDGE MAX PDGF_UP.V1_DN 63 -0.411 -1.275 0.082 0.471 1196 tags=16%, list=8%, signal=17% KRAS.AMP.LUN 50 -0.431 -1.244 0.134 0.498 1019 tags=22%, G_UP.V1_UP list=7%, signal=24% IL2_UP.V1_DN 98 -0.390 -1.279 0.050 0.506 2096 tags=32%, list=15%, signal=37% KRAS.KIDNEY_ 49 -0.437 -1.293 0.075 0.512 965 tags=22%, UP.V1_DN list=7%, signal=24% LEF1_UP.V1_UP 117 -0.365 -1.227 0.112 0.521 1292 tags=20%, list=9%, signal=21% PRC2_EZH2_UP. 116 -0.345 -1.188 0.083 0.528 1474 tags=20%, V1_UP list=10%, signal=22% ESC_J1_UP_EAR 137 -0.350 -1.195 0.082 0.532 1293 tags=15%, LY.V1_DN list=9%, signal=16% ATM_DN.V1_DN 58 -0.413 -1.247 0.131 0.535 1760 tags=26%, list=12%, signal=29%

202

KRAS.600.LUNG. 102 -0.369 -1.212 0.100 0.537 1155 tags=19%, BREAST_UP.V1_ list=8%, UP signal=20% ATM_DN.V1_UP 66 -0.391 -1.202 0.126 0.539 1680 tags=23%, list=12%, signal=26%

Table 8.7 Details for top 10 ENCODE TF enriched terms for each STAT5a mutant. Terms were ranked by odds ratio.

Top 10 ENCODE TF terms Odds Overlap p-value padj WT-STAT5a Ratio MAX 56/2000 0.00752 1 1.39 HDAC2 29/913 0.01114 1 1.57 FOSL2 16/469 0.02967 1 1.69 TEAD4 17/493 0.02334 1 1.71 NFIC 24/663 0.00447 1 1.79 FOXA1 16/435 0.01605 1 1.82 JUND 9/241 0.05678 1 1.85 NFE2 11/287 0.03222 1 1.90 STAT5A 9/217 0.03274 1 2.05 TCF12 5/104 0.05979 1 2.38 Top 10 ENCODE TF terms Odds Overlap p-value padj Y694F-STAT5a Ratio THAP1 191/1775 6.49E-11 1.77E-08 1.58 IRF3 82/755 1.80E-05 1.94E-04 1.59 SIX5 138/1258 1.21E-08 7.62E-07 1.61 SREBF2 45/405 8.42E-04 0.004187834 1.63 SREBF1 138/1234 3.57E-09 3.24E-07 1.64 PBX3 238/2098 4.03E-16 3.29E-13 1.66 JUND 14/123 0.0405 0.077219684 1.67 NR3C1 26/211 0.00246 0.009216281 1.81 FOSL1 28/203 2.88E-04 0.001867847 2.02 SUPT20H 26/172 1.06E-04 8.09E-04 2.22 Top 10 ENCODE TF terms Odds Overlap p-value padj S726A-STAT5a Ratio NR3C1 28/211 0.02307 0.06255534 1.48 SUPT20H 23/172 0.03404 0.083160569 1.49 SREBF1 167/1234 2.83E-08 1.93E-06 1.51 JUND 110/809 5.84E-06 1.04E-04 1.52

203

TEAD4 61/447 6.13E-04 0.004068339 1.52 SIX5 172/1258 8.24E-09 1.12E-06 1.53 SP1 171/1249 8.29E-09 9.66E-07 1.53 PBX3 291/2098 2.98E-15 2.43E-12 1.55 ESRRA 146/1023 7.74E-09 1.26E-06 1.59 TCF12 15/104 0.04407 0.101577738 1.61 Top 10 ENCODE TF terms Odds Overlap p-value padj S780A-STAT5a Ratio SP1 35/1727 0.07116 1 1.29 TRIM28 42/2000 0.03152 1 1.34 USF2 42/2000 0.03152 1 1.34 STAT3 43/2000 0.02102 1 1.37 MYBL2 35/1595 0.02772 1 1.40 FOXA1 20/889 0.06862 1 1.43 TAF1 13/512 0.06164 1 1.62 FOXP2 11/423 0.07127 1 1.66 RXRA 5/139 0.06839 1 2.29 ZZZ3 11/293 0.00685 1 2.39

Table 8.8 Odds ratio of top 10 enriched ENCODE Transcription Factors for each STAT5a mutant in gene set enrichment analysis using Enrichr. Data used for Figure 7 in Woock et al

(submitted).

Top 10 ENCODE ENRICHED TF terms WT- Y694F- S726A- S780A- for WT-STAT5a STAT5a STAT5a STAT5a STAT5a MAX 1.39 1.42 1.33 0.86 HDAC2 1.57 1.46 1.00 1.33 FOSL2 1.69 0.91 1.05 0.41 TEAD4 1.71 1.13 1.11 1.42 NFIC 1.79 1.08 0.99 1.15 FOXA1 1.82 1.25 1.36 0.59 JUND 1.85 1.34 1.39 0.53 NFE2 1.90 1.43 1.32 1.11 STAT5A 2.05 1.15 1.29 0.29 TCF12 2.38 0.56 1.61 0.61 Top 10 ENCODE ENRICHED TF terms WT- Y694F- S726A- S780A- for Y694F-STAT5a STAT5a STAT5a STAT5a STAT5a THAP1 0.50 1.58 1.29 0.65 IRF3 0.79 1.59 1.45 0.67

204

SIX5 0.75 1.61 1.53 0.91 SREBF2 0.73 1.63 1.30 0.79 SREBF1 1.12 1.64 1.51 0.57 PBX3 0.90 1.66 1.55 0.88 JUND 2.01 1.67 1.27 0.08 NR3C1 0.47 1.81 1.48 0.30 FOSL1 1.46 2.02 1.32 1.57 SUPT20H 0.58 2.22 1.49 0.74 Top 10 ENCODE ENRICHED TF terms WT- Y694F- S726A- S780A- for S726A-STAT5a STAT5a STAT5a STAT5a STAT5a NR3C1 0.47 1.81 1.48 0.30 SUPT20H 0.58 2.22 1.49 0.74 SREBF1 1.12 1.64 1.51 0.57 JUND 1.10 1.47 1.52 0.63 TEAD4 1.11 1.41 1.52 0.85 SIX5 0.75 1.61 1.53 0.91 SP1 0.95 1.56 1.53 0.87 PBX3 0.90 1.66 1.55 0.88 ESRRA 0.63 1.53 1.59 0.56 TCF12 2.38 0.56 1.61 0.61 Top 10 ENCODE ENRICHED TF terms WT- Y694F- S726A- S780A- for S780A-STAT5a STAT5a STAT5a STAT5a STAT5a SP1 1.03 1.16 1.28 1.29 TRIM28 0.64 1.17 1.09 1.34 USF2 1.06 1.42 1.27 1.34 STAT3 0.99 1.10 1.13 1.37 MYBL2 0.87 1.21 1.08 1.40 FOXA1 1.17 1.16 1.09 1.43 TAF1 0.97 0.80 1.00 1.62 FOXP2 1.40 1.46 1.27 1.66 RXRA 1.78 1.16 1.04 2.29 ZZZ3 0.68 1.40 1.14 2.39

205

8.2. Uncropped Western blots for Woock et al (Chapter 3), submitted February 2021 Figure 3.1D: Full membrane western blot data

206

Figure 3.1E: Full membrane western blot data

207

Figure 3.1F: Full membrane western blot data

208

Figure 3.2B: Full membrane western blot data

209

Figure 3.2D: Full membrane western blot data

210

Figure 3.4C: Full membrane western blot data

211

Figure 3.5A and 3.5B: Full membrane western blot data

212

Figure 3.6B: Full membrane western blot data

213

Bibliography 1. Breast Cancer Statistics. Centers for Disease Control. Published May 28, 2019. https://www.cdc.gov/cancer/breast/statistics/

2. Siegel RL, Miller KD, Fuchs HE, Jemal A. Cancer Statistics, 2021. CA Cancer J Clin. 2021;71(1):7-33. doi:https://doi.org/10.3322/caac.21654

3. DeSantis CE, Ma J, Gaudet MM, et al. Breast cancer statistics, 2019. CA Cancer J Clin. 2019;69(6):438-451. doi:10.3322/caac.21583

4. Prat A, Karginova O, Parker JS, et al. Characterization of cell lines derived from breast cancers and normal mammary tissues for the study of the intrinsic molecular subtypes. Breast Cancer Res Treat. 2013;142(2):237-255. doi:10.1007/s10549-013-2743-3

5. Dias K, Dvorkin-Gheva A, Hallett RM, et al. Claudin-Low Breast Cancer; Clinical & Pathological Characteristics. PloS One. 2017;12(1):e0168669-e0168669. doi:10.1371/journal.pone.0168669

6. Clevenger CV, Chang WP, Ngo W, Pasha TL, Montone KT, Tomaszewski JE. Expression of prolactin and prolactin receptor in human breast carcinoma. Evidence for an autocrine/paracrine loop. Am J Pathol. 1995;146(3):695-705.

7. Freeman ME, Kanyicska B, Lerant A, Nagy G. Prolactin: structure, function, and regulation of secretion. Physiol Rev. 2000;80(4):1523-1631. doi:10.1152/physrev.2000.80.4.1523

8. Brooks CL. Molecular Mechanisms of Prolactin and Its Receptor. Endocr Rev. 2012;33(4):504-525. doi:10.1210/er.2011-1040

9. Huising MO, Kruiswijk CP, Flik G. Phylogeny and evolution of class-I helical cytokines. J Endocrinol. 2006;189(1):1-25. doi:10.1677/joe.1.06591

10. Bernard V, Young J, Binart N. Prolactin — a pleiotropic factor in health and disease. Nat Rev Endocrinol. 2019;15(6):356-365. doi:10.1038/s41574-019-0194-6

11. Ormandy CJ, Camus A, Barra J, et al. Null mutation of the prolactin receptor gene produces multiple reproductive defects in the mouse. Genes Dev. 1997;11(2):167-178. doi:10.1101/gad.11.2.167

12. Horseman ND, Zhao W, Montecino-Rodriguez E, et al. Defective mammopoiesis, but normal hematopoiesis, in mice with a targeted disruption of the prolactin gene. EMBO J. 1997;16(23):6926-6935. doi:10.1093/emboj/16.23.6926

13. Baker SJ, Rane SG, Reddy EP. Hematopoietic cytokine receptor signaling. Oncogene. 2007;26(47):6724-6737. doi:10.1038/sj.onc.1210757

14. Syed F, Rycyzyn MA, Westgate L, Clevenger CV. A Novel and Functional Interaction Between Cyclophilin A and Prolactin Receptor. Endocrine. 2003;20(1-2):83-90. doi:10.1385/ENDO:20:1-2:83

214

15. Zheng J, Koblinski JE, Dutson LV, Feeney YB, Clevenger CV. Prolyl Isomerase Cyclophilin A Regulation of Janus-Activated Kinase 2 and the Progression of Human Breast Cancer. Cancer Res. 2008;68(19):7769-7778. doi:10.1158/0008-5472.CAN-08-0639

16. Rozakis-Adcock M. Identification of Ligand Binding Determinantsof the Prolactin Receptor. :6.

17. Rozakis-Adcock M, Kelly P. Mutational Analysis of the Ligand-binding Domainof the Prolactin Receptor. :6.

18. Dagil R, Knudsen MJ, Olsen JG, et al. The WSXWS Motif in Cytokine Receptors Is a Molecular Switch Involved in Receptor Activation: Insight from Structures of the Prolactin Receptor. Structure. 2012;20(2):270-282. doi:10.1016/j.str.2011.12.010

19. Liu Y, Jiang J, Lepik B, Zhang Y, Zinn KR, Frank SJ. Subdomain 2, Not the Transmembrane Domain, Determines the Dimerization Partner of Growth Hormone Receptor and Prolactin Receptor. Endocrinology. 2017;158(10):3235-3248. doi:10.1210/en.2017-00469

20. Biener-Ramanujan E, Ramanujan VK, Herman B, Gertler A. Spatio-temporal kinetics of growth hormone receptor signaling in single cells using FRET microscopy. Growth Horm IGF Res Off J Growth Horm Res Soc Int IGF Res Soc. 2006;16(4):247-257. doi:10.1016/j.ghir.2006.06.001

21. Waters MJ, Brooks AJ, Chhabra Y. A new mechanism for growth hormone receptor activation of JAK2, and implications for related cytokine receptors. JAK-STAT. 2014;3(2). doi:10.4161/jkst.29569

22. Lebrun JJ, Ali S, Ullrich A, Kelly PA. Proline-rich sequence-mediated Jak2 association to the prolactin receptor is required but not sufficient for signal transduction. J Biol Chem. 1995;270(18):10664-10670. doi:10.1074/jbc.270.18.10664

23. Pezet A, Buteau H, Kelly PA, Edery M. The last proline of Box 1 is essential for association with JAK2 and functional activation of the prolactin receptor. Mol Cell Endocrinol. 1997;129(2):199-208. doi:10.1016/s0303-7207(97)00063-4

24. Clevenger CV, Kline JB. Prolactin receptor signal transduction. Lupus. 2001;10(10):706- 718. doi:10.1191/096120301717164949

25. Hu Z-Z, Meng J, Dufau ML. Isolation and Characterization of Two Novel Forms of the Human Prolactin Receptor Generated by Alternative Splicing of a Newly Identified Exon 11. J Biol Chem. 2001;276(44):41086-41094. doi:10.1074/jbc.M102109200

26. Kline JB, Clevenger CV. Identification and characterization of the prolactin-binding protein in human serum and milk. J Biol Chem. 2001;276(27):24760-24766. doi:10.1074/jbc.M011786200

27. Gadd SL, Clevenger CV. Ligand-Independent Dimerization of the Human Prolactin Receptor Isoforms: Functional Implications. Mol Endocrinol. 2006;20(11):2734-2746. doi:10.1210/me.2006-0114

215

28. Clevenger CV, Furth PA, Hankinson SE, Schuler LA. The Role of Prolactin in Mammary Carcinoma. Endocr Rev. 2003;24(1):1-27. doi:10.1210/er.2001-0036

29. Lebrun JJ, Ali S, Sofer L, Ullrich A, Kelly PA. Prolactin-induced proliferation of Nb2 cells involves tyrosine phosphorylation of the prolactin receptor and its associated tyrosine kinase JAK2. J Biol Chem. 1994;269(19):14021-14026.

30. Das R, Vonderhaar BK. Activation of raf-1, MEK, and MAP kinase in prolactin responsive mammary cells. Breast Cancer Res Treat. 1996;40(2):141-149. doi:10.1007/bf01806209

31. Das R, Vonderhaar BK. Tamoxifen inhibits prolactin signal transduction in ER− NOG-8 mammary epithelial cells. Cancer Lett. 1997;116(1):41-46. doi:10.1016/S0304- 3835(97)04750-2

32. Pircher TJ, Petersen H, Gustafsson JA, Haldosén LA. Extracellular signal-regulated kinase (ERK) interacts with signal transducer and activator of transcription (STAT) 5a. Mol Endocrinol Baltim Md. 1999;13(4):555-565. doi:10.1210/mend.13.4.0263

33. Gao J, Horseman ND. Prolactin-Independent Modulation of the β-Casein Response Element by Erk2 MAP Kinase1MAP–mitogen-activated protein; PRL–Prolactin; IFN– interferons; IL–interleukins; GH–growth hormone; EGF–epidermal growth factor; PDGF– platelet-derived growth factor; EPO–erythropoietin; GM-CSF–granulocyte-macrophage colony stimulating factor; CSF-1–colony-stimulating factor-1; HTLV-I–human T cell lymphotrophic virus I; PLSP–Pro-Leu-Ser-Pro; EGF–epidermal growth factor.1. Cell Signal. 1999;11(3):205-210. doi:10.1016/S0898-6568(98)00067-9

34. Dinerstein-Cali H, Ferrag F, Kayser C, Kelly PA, Postel-Vinay M. Growth hormone (GH) induces the formation of protein complexes involving Stat5, Erk2, Shc and serine phosphorylated proteins. Mol Cell Endocrinol. 2000;166(2):89-99. doi:10.1016/s0303- 7207(00)00277-x

35. al-Sakkaf KA, Dobson PR, Brown BL. Prolactin induced tyrosine phosphorylation of p59fyn may mediate phosphatidylinositol 3-kinase activation in Nb2 cells. J Mol Endocrinol. 1997;19(3):347-350. doi:10.1677/jme.0.0190347

36. Bailey JP, Nieport KM, Herbst MP, Srivastava S, Serra RA, Horseman ND. Prolactin and transforming growth factor-beta signaling exert opposing effects on mammary gland morphogenesis, involution, and the Akt-forkhead pathway. Mol Endocrinol Baltim Md. 2004;18(5):1171-1184. doi:10.1210/me.2003-0345

37. Muise-Helmericks RC, Grimes HL, Bellacosa A, Malstrom SE, Tsichlis PN, Rosen N. Cyclin D expression is controlled post-transcriptionally via a phosphatidylinositol 3- kinase/Akt-dependent pathway. J Biol Chem. 1998;273(45):29864-29872. doi:10.1074/jbc.273.45.29864

38. Clevenger CV, Ngo W, Sokol DL, Luger SM, Gewirtz AM. Vav is necessary for prolactin- stimulated proliferation and is translocated into the nucleus of a T-cell line. J Biol Chem. 1995;270(22):13246-13253. doi:10.1074/jbc.270.22.13246

216

39. Miller SL, DeMaria JE, Freier DO, Riegel AM, Clevenger CV. Novel association of Vav2 and Nek3 modulates signaling through the human prolactin receptor. Mol Endocrinol Baltim Md. 2005;19(4):939-949. doi:10.1210/me.2004-0443

40. Miller SL, Antico G, Raghunath PN, Tomaszewski JE, Clevenger CV. Nek3 kinase regulates prolactin-mediated cytoskeletal reorganization and motility of breast cancer cells. Oncogene. 2007;26(32):4668-4678. doi:10.1038/sj.onc.1210264

41. Harrington KM, Clevenger CV. Identification of NEK3 Kinase Threonine 165 as a Novel Regulatory Phosphorylation Site That Modulates Focal Adhesion Remodeling Necessary for Breast Cancer Cell Migration. J Biol Chem. 2016;291(41):21388-21406. doi:10.1074/jbc.M116.726190

42. Macias H, Hinck L. Mammary Gland Development. Wiley Interdiscip Rev Dev Biol. 2012;1(4):533-557. doi:10.1002/wdev.35

43. Berryhill GE, Trott JF, Hovey RC. Mammary gland development—It’s not just about estrogen. J Dairy Sci. 2016;99(1):875-883. doi:10.3168/jds.2015-10105

44. Clevenger CV, Plank TL. Prolactin as an Autocrine/Paracrine Factor in Breast Tissue. :10.

45. Ormandy CJ, Sutherland RL. Mechanisms of prolactin receptor regulation in mammary gland. Mol Cell Endocrinol. 1993;91(1):C1-C6. doi:10.1016/0303-7207(93)90247-H

46. Goldhar AS, Duan R, Ginsburg E, Vonderhaar BK. Progesterone induces expression of the prolactin receptor gene through cooperative action of Sp1 and C/EBP. Mol Cell Endocrinol. 2011;335(2):148-157. doi:10.1016/j.mce.2011.01.004

47. Lee HJ, Ormandy CJ. Interplay between progesterone and prolactin in mammary development and implications for breast cancer. Mol Cell Endocrinol. 2012;357(1):101- 107. doi:10.1016/j.mce.2011.09.020

48. Ormandy CJ, Hall RE, Manning DL, et al. Coexpression and Cross-Regulation of the Prolactin Receptor and Sex Steroid Hormone Receptors in Breast Cancer. J Clin Endocrinol Metab. 1997;82(11):3692-3699. doi:10.1210/jcem.82.11.4361

49. Christensen HR, Murawsky MK, Horseman ND, Willson TA, Gregerson KA. Completely Humanizing Prolactin Rescues Infertility in Prolactin Knockout Mice and Leads to Human Prolactin Expression in Extrapituitary Mouse Tissues. Endocrinology. 2013;154(12):4777- 4789. doi:10.1210/en.2013-1476

50. Oakes SR, Rogers RL, Naylor MJ, Ormandy CJ. Prolactin Regulation of Mammary Gland Development. J Mammary Gland Biol Neoplasia. 2008;13(1):13-28. doi:10.1007/s10911- 008-9069-5

51. Ormandy CJ. Investigation of the Transcriptional Changes Underlying Functional Defects in the Mammary Glands of Prolactin Receptor Knockout Mice. Recent Prog Horm Res. 2003;58(1):297-323. doi:10.1210/rp.58.1.297

52. Welsch CW, Nagasawa H. Prolactin and murine mammary tumorigenesis: a review. Cancer Res. 1977;37(4):951-963.

217

53. Hankinson SE, Willett WC, Michaud DS, et al. Plasma prolactin levels and subsequent risk of breast cancer in postmenopausal women. J Natl Cancer Inst. 1999;91(7):629-634. doi:10.1093/jnci/91.7.629

54. Tworoger SS, Sluss P, Hankinson SE. Association between plasma prolactin concentrations and risk of breast cancer among predominately premenopausal women. Cancer Res. 2006;66(4):2476-2482. doi:10.1158/0008-5472.CAN-05-3369

55. Tworoger SS, Eliassen AH, Sluss P, Hankinson SE. A prospective study of plasma prolactin concentrations and risk of premenopausal and postmenopausal breast cancer. J Clin Oncol Off J Am Soc Clin Oncol. 2007;25(12):1482-1488. doi:10.1200/JCO.2006.07.6356

56. Tworoger SS, Eliassen AH, Rosner B, Sluss P, Hankinson SE. Plasma prolactin concentrations and risk of postmenopausal breast cancer. Cancer Res. 2004;64(18):6814- 6819. doi:10.1158/0008-5472.CAN-04-1870

57. Ginsburg E, Vonderhaar BK. Prolactin synthesis and secretion by human breast cancer cells. Cancer Res. 1995;55(12):2591-2595.

58. Bhatavdekar JM, Patel DD, Shah NG, et al. Prolactin as a local growth promoter in patients with breast cancer: GCRI experience. Eur J Surg Oncol J Eur Soc Surg Oncol Br Assoc Surg Oncol. 2000;26(6):540-547. doi:10.1053/ejso.2000.0943

59. Das R, Vonderhaaru BK. Prolactin as a Mitogen in Mammary Cells. :11.

60. Malarkey WB, Kennedy M, Allred LE, Milo G. Physiological concentrations of prolactin can promote the growth of human breast tumor cells in culture. J Clin Endocrinol Metab. 1983;56(4):673-677. doi:10.1210/jcem-56-4-673

61. Manni A, Wright C, Davis G, Glenn J, Joehl R, Feil P. Promotion by prolactin of the growth of human breast neoplasms cultured in vitro in the soft agar clonogenic assay. Cancer Res. 1986;46(4 Pt 1):1669-1672.

62. Naylor MJ, Lockefeer JA, Horseman ND, Ormandy CJ. Prolactin regulates mammary epithelial cell proliferation via autocrine/paracrine mechanism. Endocrine. 2003;20(1- 2):111-114. doi:10.1385/ENDO:20:1-2:111

63. Maus MV, Reilly SC, Clevenger CV. Prolactin as a chemoattractant for human breast carcinoma. Endocrinology. 1999;140(11):5447-5450. doi:10.1210/endo.140.11.7245

64. Reynolds C, Montone KT, Powell CM, Tomaszewski JE, Clevenger CV. Expression of Prolactin and Its Receptor in Human Breast Carcinoma. Endocrinology. 1997;138(12):5555-5560. doi:10.1210/endo.138.12.5605

65. Dai X, Cheng H, Bai Z, Li J. Breast Cancer Cell Line Classification and Its Relevance with Breast Tumor Subtyping. J Cancer. 2017;8(16):3131-3141. doi:10.7150/jca.18457

66. Smith SE, Mellor P, Ward AK, et al. Molecular characterization of breast cancer cell lines through multiple omic approaches. Breast Cancer Res BCR. 2017;19. doi:10.1186/s13058- 017-0855-0

218

67. Gutzman JH, Miller KK, Schuler LA. Endogenous human prolactin and not exogenous human prolactin induces estrogen receptor α and prolactin receptor expression and increases estrogen responsiveness in breast cancer cells. J Steroid Biochem Mol Biol. 2004;88(1):69-77. doi:10.1016/j.jsbmb.2003.10.008

68. Shafie S, Brooks SC. Effect of Prolactin on Growth and the Estrogen Receptor Level of Human Breast Cancer Cells (MCF-7). Cancer Res. 1977;37(3):792-799.

69. González L, Zambrano A, Lazaro-Trueba I, et al. Activation of the unliganded estrogen receptor by prolactin in breast cancer cells. Oncogene. 2009;28(10):1298-1308. doi:10.1038/onc.2008.473

70. Fiorillo AA, Medler TR, Feeney YB, Wetz SM, Tommerdahl KL, Clevenger CV. The Prolactin Receptor Transactivation Domain Is Associated with Steroid Hormone Receptor Expression and Malignant Progression of Breast Cancer. Am J Pathol. 2013;182(1):217- 233. doi:10.1016/j.ajpath.2012.09.021

71. Wennbo H, Gebre-Medhin M, Gritli-Linde A, Ohlsson C, Isaksson OG, Törnell J. Activation of the prolactin receptor but not the growth hormone receptor is important for induction of mammary tumors in transgenic mice. J Clin Invest. 1997;100(11):2744-2751.

72. Vomachka AJ, Pratt SL, Lockefeer JA, Horseman ND. Prolactin gene-disruption arrests mammary gland development and retards T-antigen-induced tumor growth. Oncogene. 2000;19(8):1077-1084. doi:10.1038/sj.onc.1203348

73. Rose-Hellekant TA, Arendt LM, Schroeder MD, Gilchrist K, Sandgren EP, Schuler LA. Prolactin induces ERalpha-positive and ERalpha-negative mammary cancer in transgenic mice. Oncogene. 2003;22(30):4664-4674. doi:10.1038/sj.onc.1206619

74. O’Leary KA, Shea MP, Schuler LA. Modeling Prolactin Actions in Breast Cancer in vivo: Insights from the NRL-PRL Mouse. Adv Exp Med Biol. 2015;846:201-220. doi:10.1007/978-3-319-12114-7_9

75. Campbell KM, O’Leary KA, Rugowski DE, et al. A Spontaneous Aggressive ERα+ Mammary Tumor Model Is Driven by Kras Activation. Cell Rep. 2019;28(6):1526-1537.e4. doi:10.1016/j.celrep.2019.06.098

76. O’Leary KA, Rugowski DE, Shea MP, Sullivan R, Moser AR, Schuler LA. Prolactin synergizes with canonical Wnt signals to drive development of ER+ mammary tumors via activation of the Notch pathway. Cancer Lett. 2021;503:231-239. doi:10.1016/j.canlet.2021.01.012

77. Rui H, Utama FE, Yanac AF, et al. Abstract S1-8: Prolactin-humanized mice: an improved animal recipient for therapy response-testing of patient-derived breast cancer xenotransplants. Cancer Res. 2012;72(24 Supplement):S1-S1-8. doi:10.1158/0008- 5472.SABCS12-S1-8

78. Hakim S, Craig JM, Koblinski JE, Clevenger CV. Inhibition of the Activity of Cyclophilin A Impedes Prolactin Receptor-Mediated Signaling, Mammary Tumorigenesis, and Metastases. iScience. 2020;23(10):101581. doi:10.1016/j.isci.2020.101581

219

79. Medler TR, Craig JM, Fiorillo AA, Feeney YB, Harrell JC, Clevenger CV. HDAC6 Deacetylates HMGN2 to Regulate Stat5a Activity and Breast Cancer Growth. Mol Cancer Res MCR. 2016;14(10):994-1008. doi:10.1158/1541-7786.MCR-16-0109

80. Borgés S, Moudilou E, Vouyovitch C, et al. Involvement of a JAK/STAT pathway inhibitor: cytokine inducible SH2 containing protein in breast cancer. Adv Exp Med Biol. 2008;617:321-329. doi:10.1007/978-0-387-69080-3_30

81. Brockman JL, Schroeder MD, Schuler LA. PRL Activates the Cyclin D1 Promoter Via the Jak2/Stat Pathway. Mol Endocrinol. 2002;16(4):774-784. doi:10.1210/mend.16.4.0817

82. Matsumoto A, Masuhara M, Mitsui K, et al. CIS, a cytokine inducible SH2 protein, is a target of the JAK-STAT5 pathway and modulates STAT5 activation. Blood. 1997;89(9):3148-3154.

83. Fang F, Antico G, Zheng J, Clevenger CV. Quantification of PRL/Stat5 signaling with a novel pGL4-CISH reporter. BMC Biotechnol. 2008;8:11. doi:10.1186/1472-6750-8-11

84. Tran TH, Utama FE, Lin J, et al. Prolactin inhibits BCL6 expression in breast cancer through a Stat5a-dependent mechanism. Cancer Res. 2010;70(4):1711-1721. doi:10.1158/0008-5472.CAN-09-2314

85. Fiorillo AA, Medler TR, Feeney YB, Liu Y, Tommerdahl KL, Clevenger CV. HMGN2 Inducibly Binds a Novel Transactivation Domain in Nuclear PRLr to Coordinate Stat5a- Mediated Transcription. Mol Endocrinol. 2011;25(9):1550-1564. doi:10.1210/me.2011- 0106

86. Schauwecker SM, Kim JJ, Licht JD, Clevenger CV. Histone H1 and Chromosomal Protein HMGN2 Regulate Prolactin-induced STAT5 Transcription Factor Recruitment and Function in Breast Cancer Cells. J Biol Chem. 2017;292(6):2237-2254. doi:10.1074/jbc.M116.764233

87. Craig JM, Turner TH, Harrell JC, Clevenger CV. Prolactin Drives a Dynamic STAT5A/HDAC6/HMGN2 Cis-Regulatory Landscape Exploitable in ER+ Breast Cancer. Endocrinology. Published online February 16, 2021. doi:10.1210/endocr/bqab036

88. Grible JM, Zot P, Olex AL, et al. The human intermediate prolactin receptor is a mammary proto-oncogene. Npj Breast Cancer. 2021;7(1):1-11. doi:10.1038/s41523-021-00243-7

89. Ihle JN. STATs: Signal Transducers and Activators of Transcription. Cell. 1996;84(3):331- 334. doi:10.1016/S0092-8674(00)81277-5

90. Schindler C, Shuai K, Prezioso VR, Darnell JE. Interferon-dependent tyrosine phosphorylation of a latent cytoplasmic transcription factor. Science. 1992;257(5071):809- 813. doi:10.1126/science.1496401

91. Shuai K, Ziemiecki A, Wilks AF, et al. Polypeptide signalling to the nucleus through tyrosine phosphorylation of Jak and Stat proteins. Nature. 1993;366(6455):580-583. doi:10.1038/366580a0

220

92. Wang Y, Levy DE. Comparative evolutionary genomics of the STAT family of transcription factors. JAK-STAT. 2012;1(1):23-33. doi:10.4161/jkst.19418

93. Clevenger CV. Roles and Regulation of Stat Family Transcription Factors in Human Breast Cancer. Am J Pathol. 2004;165(5):1449-1460.

94. Ehret GB, Reichenbach P, Schindler U, et al. DNA Binding Specificity of Different STAT Proteins COMPARISON OF IN VITRO SPECIFICITY WITH NATURAL TARGET SITES. J Biol Chem. 2001;276(9):6675-6688. doi:10.1074/jbc.M001748200

95. Wakao H, Gouilleux F, Groner B. Mammary gland factor (MGF) is a novel member of the cytokine regulated transcription factor gene family and confers the prolactin response. EMBO J. 1994;13(9):2182-2191.

96. Copeland NG, Gilbert DJ, Schindler C, et al. Distribution of the mammalian Stat gene family in mouse chromosomes. Genomics. 1995;29(1):225-228. doi:10.1006/geno.1995.1235

97. Ambrosio R, Fimiani G, Monfregola J, et al. The structure of human STAT5A and B genes reveals two regions of nearly identical sequence and an alternative tissue specific STAT5B promoter. Gene. 2002;285(1-2):311-318. doi:10.1016/s0378-1119(02)00421-3

98. Kemena C, Notredame C. Upcoming challenges for multiple sequence alignment methods in the high-throughput era. Bioinforma Oxf Engl. 2009;25(19):2455-2465. doi:10.1093/bioinformatics/btp452

99. Di Tommaso P, Moretti S, Xenarios I, et al. T-Coffee: a web server for the multiple sequence alignment of protein and RNA sequences using structural information and homology extension. Nucleic Acids Res. 2011;39(Web Server issue):W13-17. doi:10.1093/nar/gkr245

100. Waterhouse AM, Procter JB, Martin DMA, Clamp M, Barton GJ. Jalview Version 2—a multiple sequence alignment editor and analysis workbench. Bioinformatics. 2009;25(9):1189-1191. doi:10.1093/bioinformatics/btp033

101. Liu X, Robinson GW, Gouilleux F, Groner B, Hennighausen L. Cloning and expression of Stat5 and an additional homologue (Stat5b) involved in prolactin signal transduction in mouse mammary tissue. Proc Natl Acad Sci U S A. 1995;92(19):8831-8835.

102. Teglund S, McKay C, Schuetz E, et al. Stat5a and Stat5b Proteins Have Essential and Nonessential, or Redundant, Roles in Cytokine Responses. Cell. 1998;93(5):841-850. doi:10.1016/S0092-8674(00)81444-0

103. Nicholas C, Lesinski G. The Jak-STAT Signal Transduction Pathway in Melanoma. In: ; 2011. doi:10.5772/18876

104. Vinkemeier U, Moarefi I, Darnell JE, Kuriyan J. Structure of the Amino-Terminal Protein Interaction Domain of STAT-4. Science. 1998;279(5353):1048-1052. doi:10.1126/science.279.5353.1048

221

105. Recio C, Guerra B, Guerra-Rodríguez M, et al. Signal transducer and activator of transcription (STAT)-5: an opportunity for drug development in oncohematology. Oncogene. 2019;38(24):4657-4668. doi:10.1038/s41388-019-0752-3

106. Verhoeven Y, Tilborghs S, Jacobs J, et al. The potential and controversy of targeting STAT family members in cancer. Semin Cancer Biol. 2020;60:41-56. doi:10.1016/j.semcancer.2019.10.002

107. Lim CP, Cao X. Structure, function, and regulation of STAT proteins. Mol Biosyst. 2006;2(11):536-550. doi:10.1039/B606246F

108. Levy DE, Darnell JE. STATs: transcriptional control and biological impact. Nat Rev Mol Cell Biol. 2002;3(9):651-662. doi:10.1038/nrm909

109. Miklossy G, Hilliard TS, Turkson J. Therapeutic modulators of STAT signalling for human diseases. Nat Rev Drug Discov. 2013;12(8):611-629. doi:10.1038/nrd4088

110. Dorritie KA, McCubrey JA, Johnson DE. STAT transcription factors in hematopoiesis and leukemogenesis: opportunities for therapeutic intervention. Leukemia. 2014;28(2):248-257. doi:10.1038/leu.2013.192

111. Yang J, Stark GR. Roles of unphosphorylated STATs in signaling. Cell Res. 2008;18:443- 451. doi:https://doi-org.proxy.library.vcu.edu/10.1038/cr.2008.41

112. Durbin JE, Hackenmiller R, Simon MC, Levy DE. Targeted disruption of the mouse Stat1 gene results in compromised innate immunity to viral disease. Cell. 1996;84(3):443-450. doi:10.1016/s0092-8674(00)81289-1

113. Park C, Li S, Cha E, Schindler C. Immune response in Stat2 knockout mice. Immunity. 2000;13(6):795-804. doi:10.1016/s1074-7613(00)00077-7

114. Lee CK, Rao DT, Gertner R, Gimeno R, Frey AB, Levy DE. Distinct requirements for IFNs and STAT1 in NK cell function. J Immunol Baltim Md 1950. 2000;165(7):3571-3577. doi:10.4049/jimmunol.165.7.3571

115. Clarkson RWE, Boland MP, Kritikou EA, et al. The Genes Induced by Signal Transducer and Activators of Transcription (STAT)3 and STAT5 in Mammary Epithelial Cells Define the Roles of these STATs in Mammary Development. Mol Endocrinol. 2006;20(3):675-685. doi:10.1210/me.2005-0392

116. Abell K, Bilancio A, Clarkson RWE, et al. Stat3-induced apoptosis requires a molecular switch in PI(3)K subunit composition. Nat Cell Biol. 2005;7(4):392-398. doi:10.1038/ncb1242

117. Socolovsky M, Fallon AE, Wang S, Brugnara C, Lodish HF. Fetal anemia and apoptosis of red cell progenitors in Stat5a-/-5b-/- mice: a direct role for Stat5 in Bcl-X(L) induction. Cell. 1999;98(2):181-191. doi:10.1016/s0092-8674(00)81013-2

118. Socolovsky M, Nam H, Fleming MD, Haase VH, Brugnara C, Lodish HF. Ineffective erythropoiesis in Stat5a(-/-)5b(-/-) mice due to decreased survival of early erythroblasts. Blood. 2001;98(12):3261-3273. doi:10.1182/blood.v98.12.3261

222

119. Bunting KD, Bradley HL, Hawley TS, Moriggl R, Sorrentino BP, Ihle JN. Reduced lymphomyeloid repopulating activity from adult bone marrow and fetal liver of mice lacking expression of STAT5. Blood. 2002;99(2):479-487. doi:10.1182/blood.v99.2.479

120. Moriggl R, Topham DJ, Teglund S, et al. Stat5 is required for IL-2-induced cell cycle progression of peripheral T cells. Immunity. 1999;10(2):249-259. doi:10.1016/s1074- 7613(00)80025-4

121. Able AA, Burrell JA, Stephens JM. STAT5-Interacting Proteins: A Synopsis of Proteins that Regulate STAT5 Activity. Biology. 2017;6(1). doi:10.3390/biology6010020

122. Furth PA, Nakles RE, Millman S, Diaz-Cruz ES, Cabrera MC. Signal transducer and activator of transcription 5 as a key signaling pathway in normal mammary gland developmental biology and breast cancer. Breast Cancer Res BCR. 2011;13(5):220. doi:10.1186/bcr2921

123. Jones FE, Welte T, Fu X-Y, Stern DF. Erbb4 Signaling in the Mammary Gland Is Required for Lobuloalveolar Development and Stat5 Activation during Lactation. J Cell Biol. 1999;147(1):77-88. doi:10.1083/jcb.147.1.77

124. Johnson KJ, Peck AR, Liu C, et al. PTP1B Suppresses Prolactin Activation of Stat5 in Breast Cancer Cells. Am J Pathol. 2010;177(6):2971-2983. doi:10.2353/ajpath.2010.090399

125. Aksamitiene E, Achanta S, Kolch W, Kholodenko BN, Hoek JB, Kiyatkin A. Prolactin- stimulated activation of ERK1/2 mitogen-activated protein kinases is controlled by PI3- kinase/Rac/PAK signaling pathway in breast cancer cells. Cell Signal. 2011;23(11):1794- 1805. doi:10.1016/j.cellsig.2011.06.014

126. Kirken RA, Malabarba MG, Xu J, et al. Two Discrete Regions of Interleukin-2 (IL2) Receptor β Independently Mediate IL2 Activation of a PD98059/Rapamycin/Wortmannin- insensitive Stat5a/b Serine Kinase. J Biol Chem. 1997;272(24):15459-15465. doi:10.1074/jbc.272.24.15459

127. Yamashita H, Xu J, Erwin RA, Farrar WL, Kirken RA, Rui H. Differential control of the phosphorylation state of proline-juxtaposed serine residues Ser725 of Stat5a and Ser730 of Stat5b in prolactin-sensitive cells. J Biol Chem. 1998;273(46):30218-30224. doi:10.1074/jbc.273.46.30218

128. Benitah SA, Valerón PF, Rui H, Lacal JC. STAT5a Activation Mediates the Epithelial to Mesenchymal Transition Induced by Oncogenic RhoA. Mol Biol Cell. 2003;14(1):40-53. doi:10.1091/mbc.E02-08-0454

129. Yoo KH, Oh S, Kang K, Hensel T, Robinson GW, Hennighausen L. Loss of EZH2 results in precocious mammary gland development and activation of STAT5-dependent genes. Nucleic Acids Res. 2015;43(18):8774-8789. doi:10.1093/nar/gkv776

130. Yamashita H, Iwase H, Toyama T, Fujii Y. Naturally occurring dominant-negative Stat5 suppresses transcriptional activity of estrogen receptors and induces apoptosis in T47D breast cancer cells. Oncogene. 2003;22(11):1638-1652. doi:10.1038/sj.onc.1206277

223

131. Faulds MH, Pettersson K, Gustafsson J-Å, Haldosén L-A. Cross-Talk Between ERs and Signal Transducer and Activator of Transcription 5 Is E2 Dependent and Involves Two Functionally Separate Mechanisms. Mol Endocrinol. 2001;15(11):1929-1940. doi:10.1210/mend.15.11.0726

132. Bjornstrom L, Kilic E, Norman M, Parker MG, Sjoberg M. Cross-talk between Stat5b and estrogen receptor-alpha and -beta in mammary epithelial cells. J Mol Endocrinol. 2001;27(1):93-106. doi:10.1677/jme.0.0270093

133. Wyszomierski SL, Yeh J, Rosen JM. Glucocorticoid receptor/signal transducer and activator of transcription 5 (STAT5) interactions enhance STAT5 activation by prolonging STAT5 DNA binding and tyrosine phosphorylation. Mol Endocrinol Baltim Md. 1999;13(2):330-343. doi:10.1210/mend.13.2.0232

134. Stöcklin E, Wissler M, Gouilleux F, Groner B. Functional interactions between Stat5 and the glucocorticoid receptor. Nature. 1996;383(6602):726-728. doi:10.1038/383726a0

135. Obr AE, Grimm SL, Bishop KA, Pike JW, Lydon JP, Edwards DP. Progesterone receptor and Stat5 signaling cross talk through RANKL in mammary epithelial cells. Mol Endocrinol Baltim Md. 2013;27(11):1808-1824. doi:10.1210/me.2013-1077

136. Kavarthapu R, Tsai Morris C-H, Dufau ML. Prolactin induces up-regulation of its cognate receptor in breast cancer cells via transcriptional activation of its generic promoter by cross-talk between ERα and STAT5. Oncotarget. 2014;5(19):9079-9091. doi:10.18632/oncotarget.2376

137. Frasor J, Gibori G. Prolactin regulation of estrogen receptor expression. Trends Endocrinol Metab. 2003;14(3):118-123. doi:10.1016/S1043-2760(03)00030-4

138. Koirala S, Thomas LN, Too CKL. Prolactin/Stat5 and Androgen R1881 Coactivate Carboxypeptidase-D Gene in Breast Cancer Cells. Mol Endocrinol. 2014;28(3):331-343. doi:10.1210/me.2013-1202

139. Hoang DT, Iczkowski KA, Kilari D, See W, Nevalainen MT. Androgen receptor-dependent and -independent mechanisms driving prostate cancer progression: Opportunities for therapeutic targeting from multiple angles. Oncotarget. 2016;8(2):3724-3745. doi:10.18632/oncotarget.12554

140. Impaired Alveologenesis and Maintenance of Secretory Mammary Epithelial Cells in Jak2 Conditional Knockout Mice. Accessed May 12, 2020. https://www-ncbi-nlm-nih- gov.proxy.library.vcu.edu/pmc/articles/PMC419899/

141. Cui Y, Riedlinger G, Miyoshi K, et al. Inactivation of Stat5 in Mouse Mammary Epithelium during Pregnancy Reveals Distinct Functions in Cell Proliferation, Survival, and Differentiation. Mol Cell Biol. 2004;24(18):8037-8047. doi:10.1128/MCB.24.18.8037- 8047.2004

142. Iavnilovitch E, Groner B, Barash I. Overexpression and Forced Activation of Stat5 in Mammary Gland of Transgenic Mice Promotes Cellular Proliferation, Enhances Differentiation, and Delays Postlactational Apoptosis11 German Israeli Research Fund

224

(GIF), and a grant from the Chief Scientist, Israeli Ministry of Agriculture. Mol Cancer Res. 2002;1(1):32-47.

143. Creamer BA, Sakamoto K, Schmidt JW, Triplett AA, Moriggl R, Wagner K-U. Stat5 Promotes Survival of Mammary Epithelial Cells through Transcriptional Activation of a Distinct Promoter in Akt1. Mol Cell Biol. 2010;30(12):2957-2970. doi:10.1128/MCB.00851- 09

144. Humphreys RC, Hennighausen L. Signal Transducer and Activator of Transcription 5a Influences Mammary Epithelial Cell Survival and Tumorigenesis. Cell Growth Differ. 1999;10(10):685.

145. Ren S, Cai HR, Li M, Furth PA. Loss of Stat5a delays mammary cancer progression in a mouse model. Oncogene. 2002;21(27):4335-4339. doi:10.1038/sj.onc.1205484

146. Iavnilovitch E, Cardiff RD, Groner B, Barash I. Deregulation of Stat5 expression and activation causes mammary tumors in transgenic mice. Int J Cancer. 2004;112(4):607-619. doi:10.1002/ijc.20484

147. Moriggl R, Gouilleux-Gruart V, Jähne R, et al. Deletion of the carboxyl-terminal transactivation domain of MGF-Stat5 results in sustained DNA binding and a dominant negative phenotype. Mol Cell Biol. 1996;16(10):5691-5700.

148. Cotarla I, Ren S, Zhang Y, Gehan E, Singh B, Furth PA. Stat5a is tyrosine phosphorylated and nuclear localized in a high proportion of human breast cancers. Int J Cancer. 2004;108(5):665-671. doi:10.1002/ijc.11619

149. Yamashita H, Nishio M, Ando Y, et al. Stat5 expression predicts response to endocrine therapy and improves survival in estrogen receptor-positive breast cancer. Endocr Relat Cancer. 2006;13(3):885-893. doi:10.1677/erc.1.01095

150. Peck AR, Witkiewicz AK, Liu C, et al. Low levels of Stat5a protein in breast cancer are associated with tumor progression and unfavorable clinical outcomes. Breast Cancer Res BCR. 2012;14(5):R130. doi:10.1186/bcr3328

151. Marg A, Shan Y, Meyer T, Meissner T, Brandenburg M, Vinkemeier U. Nucleocytoplasmic shuttling by nucleoporins Nup153 and Nup214 and CRM1-dependent nuclear export control the subcellular distribution of latent Stat1. J Cell Biol. 2004;165(6):823-833. doi:10.1083/jcb.200403057

152. Brown S, Zeidler MP. Unphosphorylated STATs go nuclear. Curr Opin Genet Dev. 2008;18(5):455-460. doi:10.1016/j.gde.2008.09.002

153. Reich NC. STATs get their move on. JAK-STAT. 2013;2(4). doi:10.4161/jkst.27080

154. Kumar A, Commane M, Flickinger TW, Horvath CM, Stark GR. Defective TNF-α-Induced Apoptosis in STAT1-Null Cells Due to Low Constitutive Levels of Caspases. Science. 1997;278(5343):1630-1632. doi:10.1126/science.278.5343.1630

225

155. Chatterjee-Kishore M, Wright KL, Ting JP-Y, Stark GR. How Stat1 mediates constitutive gene expression: a complex of unphosphorylated Stat1 and IRF1 supports transcription of the LMP2 gene. EMBO J. 2000;19(15):4111-4122. doi:10.1093/emboj/19.15.4111

156. Ramana CV, Gil MP, Schreiber RD, Stark GR. Stat1-dependent and -independent pathways in IFN-γ-dependent signaling. Trends Immunol. 2002;23(2):96-101. doi:10.1016/S1471-4906(01)02118-4

157. Hu X, Dutta P, Tsurumi A, et al. Unphosphorylated STAT5A stabilizes heterochromatin and suppresses tumor growth. Proc Natl Acad Sci U S A. 2013;110(25):10213-10218. doi:10.1073/pnas.1221243110

158. Pfeffer SR, Fan M, Du Z, Yang CH, Pfeffer LM. Unphosphorylated STAT3 regulates the antiproliferative, antiviral, and gene-inducing actions of Type I interferons. Biochem Biophys Res Commun. 2017;490(3):739-745. doi:10.1016/j.bbrc.2017.06.111

159. Dutta P, Zhang L, Zhang H, et al. Unphosphorylated STAT3 in heterochromatin formation and tumor suppression in lung cancer. BMC Cancer. 2020;20. doi:10.1186/s12885-020- 6649-2

160. Yang J, Chatterjee-Kishore M, Staugaitis SM, et al. Novel Roles of Unphosphorylated STAT3 in Oncogenesis and Transcriptional Regulation. Cancer Res. 2005;65(3):939-947.

161. Hj P, J L, R H, et al. Cytokine-induced megakaryocytic differentiation is regulated by genome-wide loss of a uSTAT transcriptional program. EMBO J. 2015;35(6):580-594. doi:10.15252/embj.201592383

162. Decker T. Emancipation from transcriptional latency: unphosphorylated STAT5 as guardian of hematopoietic differentiation. EMBO J. 2016;35(6):555-557. doi:10.15252/embj.201693974

163. Decker T, Kovarik P. Serine phosphorylation of STATs. Oncogene. 2000;19(21):2628- 2637. doi:10.1038/sj.onc.1203481

164. Zhu X, Wen Z, Xu LZ, Darnell JE. Stat1 serine phosphorylation occurs independently of tyrosine phosphorylation and requires an activated Jak2 kinase. Mol Cell Biol. 1997;17(11):6618-6623.

165. Stephanou A, Scarabelli TM, Brar BK, et al. Induction of apoptosis and Fas receptor/Fas ligand expression by ischemia/reperfusion in cardiac myocytes requires serine 727 of the STAT-1 transcription factor but not tyrosine 701. J Biol Chem. 2001;276(30):28340-28347. doi:10.1074/jbc.M101177200

166. Wen Z, Zhong Z, Darnell JE. Maximal activation of transcription by statl and stat3 requires both tyrosine and serine phosphorylation. Cell. 1995;82(2):241-250. doi:10.1016/0092- 8674(95)90311-9

167. Kim H, Baumann H. The Carboxyl-terminal Region of STAT3 Controls Gene Induction by the Mouse Haptoglobin Promoter*. J Biol Chem. 1997;272(23):14571-14579. doi:10.1074/jbc.272.23.14571

226

168. Liu L, McBride KM, Reich NC. STAT3 nuclear import is independent of tyrosine phosphorylation and mediated by importin-α3. Proc Natl Acad Sci U S A. 2005;102(23):8150-8155. doi:10.1073/pnas.0501643102

169. Friedbichler K, Kerenyi MA, Kovacic B, et al. Stat5a serine 725 and 779 phosphorylation is a prerequisite for hematopoietic transformation. Blood. 2010;116(9):1548-1558. doi:10.1182/blood-2009-12-258913

170. Beuvink I, Hess D, Flotow H, Hofsteenge J, Groner B, Hynes NE. Stat5a Serine Phosphorylation SERINE 779 IS CONSTITUTIVELY PHOSPHORYLATED IN THE MAMMARY GLAND, AND SERINE 725 PHOSPHORYLATION INFLUENCES PROLACTIN-STIMULATEDIN VITRO DNA BINDING ACTIVITY. J Biol Chem. 2000;275(14):10247-10255. doi:10.1074/jbc.275.14.10247

171. STAT5A signal transducer and activator of transcription 5A [Homo sapiens (human)] - Gene - NCBI. Accessed April 17, 2020. https://www-ncbi-nlm-nih- gov.proxy.library.vcu.edu/gene/6776

172. NCBI RefSeq genes, curated subset (NM_*, NR_*, NP_* or YP_*) - NM_003152.3. Accessed April 17, 2020. https://genome.ucsc.edu/cgi- bin/hgc?hgsid=824343311_Na0Q4l4qd7WQ2FrIOBHpf6Z1347X&c=chr17&l=42287866&r= 42310860&o=42287546&t=42311943&g=ncbiRefSeqCurated&i=NM_003152.3

173. Kirken RA, Erwin RA, Wang L, Wang Y, Rui H, Farrar WL. Functional Uncoupling of the Janus Kinase 3-Stat5 Pathway in Malignant Growth of Human T Cell Leukemia Virus Type 1-Transformed Human T Cells. J Immunol. 2000;165(9):5097-5104. doi:10.4049/jimmunol.165.9.5097

174. Nagy ZS, Wang Y, Erwin-Cohen RA, et al. Interleukin-2 family cytokines stimulate phosphorylation of the Pro-Ser-Pro motif of Stat5 transcription factors in human T cells: resistance to suppression of multiple serine kinase pathways. J Leukoc Biol. 2002;72(4):819-828.

175. Yamashita H, Nevalainen MT, Xu J, et al. Role of serine phosphorylation of Stat5a in prolactin-stimulated β-casein gene expression. Mol Cell Endocrinol. 2001;183(1):151-163. doi:10.1016/S0303-7207(01)00546-9

176. Kirken RA, Malabarba MG, Xu J, et al. Prolactin Stimulates Serine/Tyrosine Phosphorylation and Formation of Heterocomplexes of Multiple Stat5 Isoforms in Nb2 Lymphocytes *. J Biol Chem. 1997;272(22):14098-14103. doi:10.1074/jbc.272.22.14098

177. Sillaber C, Gesbert F, Frank DA, Sattler M, Griffin JD. STAT5 activation contributes to growth and viability in Bcr/Abl-transformed cells. Blood. 2000;95(6):2118-2125.

178. Berger A, Hoelbl-Kovacic A, Bourgeais J, et al. PAK-dependent STAT5 serine phosphorylation is required for BCR-ABL-induced leukemogenesis. Leukemia. 2014;28(3):629-641. doi:10.1038/leu.2013.351

179. Park SH, Yamashita H, Rui H, Waxman DJ. Serine phosphorylation of GH-activated signal transducer and activator of transcription 5a (STAT5a) and STAT5b: impact on STAT5

227

transcriptional activity. Mol Endocrinol Baltim Md. 2001;15(12):2157-2171. doi:10.1210/mend.15.12.0746

180. Jänicke RU, Sprengart ML, Wati MR, Porter AG. Caspase-3 is required for DNA fragmentation and morphological changes associated with apoptosis. J Biol Chem. 1998;273(16):9357-9360. doi:10.1074/jbc.273.16.9357

181. Allred DC, Harvey JM, Berardo M, Clark GM. Prognostic and predictive factors in breast cancer by immunohistochemical analysis. Mod Pathol Off J U S Can Acad Pathol Inc. 1998;11(2):155-168.

182. Babraham Bioinformatics - FastQC A Quality Control tool for High Throughput Sequence Data. Accessed November 4, 2020. https://www.bioinformatics.babraham.ac.uk/projects/fastqc/

183. Martin M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.journal. 2011;17(1):10-12. doi:10.14806/ej.17.1.200

184. STAR: ultrafast universal RNA-seq aligner | Bioinformatics | Oxford Academic. Accessed November 4, 2020. https://academic.oup.com/bioinformatics/article/29/1/15/272537

185. Pertea G. Gpertea/Gffread.; 2020. Accessed November 4, 2020. https://github.com/gpertea/gffread

186. Salmon provides fast and bias-aware quantification of transcript expression | Nature Methods. Accessed November 4, 2020. https://www.nature.com/articles/nmeth.4197

187. Soneson C, Love MI, Robinson MD. Differential analyses for RNA-seq: transcript-level estimates improve gene-level inferences. F1000Research. 2016;4:1521. doi:10.12688/f1000research.7563.2

188. R: The R Project for Statistical Computing. Accessed November 4, 2020. https://www.r- project.org/

189. Freese NH, Norris DC, Loraine AE. Integrated genome browser: visual analytics platform for genomics. Bioinformatics. 2016;32(14):2089-2095. doi:10.1093/bioinformatics/btw069

190. Chen Y, Lun ATL, Smyth GK. From reads to genes to pathways: differential expression analysis of RNA-Seq experiments using Rsubread and the edgeR quasi-likelihood pipeline. F1000Research. 2016;5:1438. doi:10.12688/f1000research.8987.2

191. Afgan E, Baker D, Batut B, et al. The Galaxy platform for accessible, reproducible and collaborative biomedical analyses: 2018 update. Nucleic Acids Res. 2018;46(W1):W537- W544. doi:10.1093/nar/gky379

192. Galaxy Community Hub. Accessed November 4, 2020. https://galaxyproject.org/

193. Hulsen T, de Vlieg J, Alkema W. BioVenn – a web application for the comparison and visualization of biological lists using area-proportional Venn diagrams. BMC Genomics. 2008;9(1):488. doi:10.1186/1471-2164-9-488

228

194. Kuleshov MV, Jones MR, Rouillard AD, et al. Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic Acids Res. 2016;44(W1):W90-W97. doi:10.1093/nar/gkw377

195. Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles | PNAS. Accessed February 10, 2021. https://www.pnas.org/content/102/43/15545.abstract

196. Mootha VK, Lindgren CM, Eriksson K-F, et al. PGC-1α-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. Nat Genet. 2003;34(3):267-273. doi:10.1038/ng1180

197. Alzubi MA, Turner TH, Olex AL, et al. Separation of breast cancer and organ microenvironment transcriptomes in metastases. Breast Cancer Res BCR. 2019;21. doi:10.1186/s13058-019-1123-2

198. Clevenger CV, Gadd SL, Zheng J. New mechanisms for PRLr action in breast cancer. Trends Endocrinol Metab. 2009;20(5):223-229. doi:10.1016/j.tem.2009.03.001

199. Arendt LM, Rugowski DE, Grafwallner-Huseth TA, Garcia-Barchino MJ, Rui H, Schuler LA. Prolactin-induced mouse mammary carcinomas model estrogen resistant luminal breast cancer. Breast Cancer Res BCR. 2011;13(1):R11. doi:10.1186/bcr2819

200. Iyer J, Reich NC. Constitutive nuclear import of latent and activated STAT5a by its coiled coil domain. FASEB J Off Publ Fed Am Soc Exp Biol. 2008;22(2):391-400. doi:10.1096/fj.07-8965com

201. Yue H, Li W, Desnoyer R, Karnik SS. Role of nuclear unphosphorylated STAT3 in angiotensin II type 1 receptor-induced cardiac hypertrophy. Cardiovasc Res. 2010;85(1):90-99. doi:10.1093/cvr/cvp285

202. Putz EM, Gotthardt D, Hoermann G, et al. CDK8-Mediated STAT1-S727 Phosphorylation Restrains NK Cell Cytotoxicity and Tumor Surveillance. Cell Rep. 2013;4(3):437-444. doi:10.1016/j.celrep.2013.07.012

203. Pilz A, Kratky W, Stockinger S, et al. Dendritic Cells Require STAT-1 Phosphorylated at Its Transactivating Domain for the Induction of Peptide-Specific CTL. J Immunol. 2009;183(4):2286-2293. doi:10.4049/jimmunol.0901383

204. Varinou L, Ramsauer K, Karaghiosoff M, et al. Phosphorylation of the Stat1 Transactivation Domain Is Required for Full-Fledged IFN-γ-Dependent Innate Immunity. Immunity. 2003;19(6):793-802. doi:10.1016/S1074-7613(03)00322-4

205. Qin HR, Kim H-J, Kim J-Y, et al. Activation of Stat3 through a Phosphomimetic Serine727 Promotes Prostate Tumorigenesis Independent of Tyrosine705 phosphorylation. Cancer Res. 2008;68(19):7736-7741. doi:10.1158/0008-5472.CAN-08-1125

206. Perou CM, Sørlie T, Eisen MB, et al. Molecular portraits of human breast tumours. Nature. 2000;406(6797):747-752. doi:10.1038/35021093

229

207. Fang F, Rycyzyn MA, Clevenger CV. Role of c-Myb during Prolactin-Induced Signal Transducer and Activator of Transcription 5a Signaling in Breast Cancer Cells. Endocrinology. 2009;150(4):1597-1606. doi:10.1210/en.2008-1079

208. Borowicz S, Van Scoyk M, Avasarala S, et al. The soft agar colony formation assay. J Vis Exp JoVE. 2014;(92):e51998. doi:10.3791/51998

209. Du F, Zhao X, Fan D. Soft Agar Colony Formation Assay as a Hallmark of Carcinogenesis. BIO-Protoc. 2017;7(12). doi:10.21769/BioProtoc.2351

210. Casetti L, Martin-Lannerée S, Najjar I, et al. Differential Contributions of STAT5A and STAT5B to Stress Protection and Tyrosine Kinase Inhibitor Resistance of Chronic Myeloid Leukemia Stem/Progenitor Cells. Cancer Res. 2013;73(7):2052-2058. doi:10.1158/0008- 5472.CAN-12-3955

211. Bertucci PY, Quaglino A, Pozzi AG, Kordon EC, Pecci A. Glucocorticoid-induced impairment of mammary gland involution is associated with STAT5 and STAT3 signaling modulation. Endocrinology. 2010;151(12):5730-5740. doi:10.1210/en.2010-0517

212. Jänicke RU. MCF-7 breast carcinoma cells do not express caspase-3. Breast Cancer Res Treat. 2009;117(1):219-221. doi:10.1007/s10549-008-0217-9

213. Kanai T, Seki S, Jenks JA, et al. Identification of STAT5A and STAT5B Target Genes in Human T Cells. PLoS ONE. 2014;9(1). doi:10.1371/journal.pone.0086790

214. Nelson EA, Walker SR, Alvarez JV, Frank DA. Isolation of Unique STAT5 Targets by Chromatin Immunoprecipitation-based Gene Identification. J Biol Chem. 2004;279(52):54724-54730. doi:10.1074/jbc.M408464200

215. Wingelhofer B, Neubauer HA, Valent P, et al. Implications of STAT3 and STAT5 signaling on gene regulation and chromatin remodeling in hematopoietic cancer. Leukemia. 2018;32(8):1713-1726. doi:10.1038/s41375-018-0117-x

216. Li P, Mitra S, Spolski R, et al. STAT5-mediated chromatin interactions in superenhancers activate IL-2 highly inducible genes: Functional dissection of the Il2ra gene locus. Proc Natl Acad Sci U S A. 2017;114(46):12111-12119. doi:10.1073/pnas.1714019114

217. Hanahan D, Weinberg RA. The Hallmarks of Cancer. Cell. 2000;100(1):57-70. doi:10.1016/S0092-8674(00)81683-9

218. Hanahan D, Weinberg RA. Hallmarks of Cancer: The Next Generation. Cell. 2011;144(5):646-674. doi:10.1016/j.cell.2011.02.013

219. Devarajan E, Sahin AA, Chen JS, et al. Down-regulation of caspase 3 in breast cancer: a possible mechanism for chemoresistance. Oncogene. 2002;21(57):8843-8851. doi:10.1038/sj.onc.1206044

220. Bao J, Zhu L, Zhu Q, Su J, Liu M, Huang W. SREBP-1 is an independent prognostic marker and promotes invasion and migration in breast cancer. Oncol Lett. 2016;12(4):2409-2416. doi:10.3892/ol.2016.4988

230

221. Liang C-C, Park AY, Guan J-L. In vitro scratch assay: a convenient and inexpensive method for analysis of cell migration in vitro. Nat Protoc. 2007;2(2):329-333. doi:10.1038/nprot.2007.30

222. Pijuan J, Barceló C, Moreno DF, et al. In vitro Cell Migration, Invasion, and Adhesion Assays: From Cell Imaging to Data Analysis. Front Cell Dev Biol. 2019;7. doi:10.3389/fcell.2019.00107

223. GSEA. Accessed February 10, 2021. https://www.gsea-msigdb.org/gsea/index.jsp

224. Creighton CJ, Hilger AM, Murthy S, Rae JM, Chinnaiyan AM, El-Ashry D. Activation of mitogen-activated protein kinase in estrogen receptor alpha-positive breast cancer cells in vitro induces an in vivo molecular phenotype of estrogen receptor alpha-negative human breast tumors. Cancer Res. 2006;66(7):3903-3911. doi:10.1158/0008-5472.CAN-05-4363

225. Sloan CA, Chan ET, Davidson JM, et al. ENCODE data at the ENCODE portal. Nucleic Acids Res. 2016;44(D1):D726-732. doi:10.1093/nar/gkv1160

226. Parker JS, Mullins M, Cheang MCU, et al. Supervised Risk Predictor of Breast Cancer Based on Intrinsic Subtypes. J Clin Oncol. 2009;27(8):1160-1167. doi:10.1200/JCO.2008.18.1370

227. Jiang Y-Z, Ma D, Suo C, et al. Genomic and Transcriptomic Landscape of Triple-Negative Breast Cancers: Subtypes and Treatment Strategies. Cancer Cell. 2019;35(3):428-440.e5. doi:10.1016/j.ccell.2019.02.001

228. Balanis NG, Sheu KM, Esedebe FN, et al. Pan-cancer Convergence to a Small-Cell Neuroendocrine Phenotype that Shares Susceptibilities with Hematological Malignancies. Cancer Cell. 2019;36(1):17-34.e7. doi:10.1016/j.ccell.2019.06.005

229. Wallden B, Storhoff J, Nielsen T, et al. Development and verification of the PAM50-based Prosigna breast cancer gene signature assay. BMC Med Genomics. 2015;8. doi:10.1186/s12920-015-0129-6

230. Schurch NJ, Schofield P, Gierliński M, et al. How many biological replicates are needed in an RNA-seq experiment and which differential expression tool should you use? RNA. 2016;22(6):839-851. doi:10.1261/rna.053959.115

231. Chen L, Yang R, Kwan T, et al. Paired rRNA-depleted and polyA-selected RNA sequencing data and supporting multi-omics data from human T cells. Sci Data. 2020;7(1):376. doi:10.1038/s41597-020-00719-4

232. RIKEN Genome Exploration Research Group, Genome Science Group (Genome Network Project Core Group), the FANTOM Consortium, et al. Antisense Transcription in the Mammalian Transcriptome. Science. 2005;309:1564-1566. doi:10.1126/science.1112009

233. Cui P, Lin Q, Ding F, et al. A comparison between ribo-minus RNA-sequencing and polyA- selected RNA-sequencing. Genomics. 2010;96(5):259-265. doi:10.1016/j.ygeno.2010.07.010

231

234. Scholl C, Fröhling S, Dunn IF, et al. Synthetic Lethal Interaction between Oncogenic KRAS Dependency and STK33 Suppression in Human Cancer Cells. Cell. 2009;137(5):821-834. doi:10.1016/j.cell.2009.03.017

235. Azoitei N, Hoffmann CM, Ellegast JM, et al. Targeting of KRAS mutant tumors by HSP90 inhibitors involves degradation of STK33. J Exp Med. 2012;209(4):697-711. doi:10.1084/jem.20111910

236. Schoenfeld J, Lessan K, Johnson N, et al. Bioinformatic analysis of primary endothelial cell gene array data illustrated by the analysis of transcriptome changes in endothelial cells exposed to VEGF-A and PlGF. Angiogenesis. 2004;7(2):143-156. doi:10.1007/s10456- 004-1677-0

237. Naik A, Al-Zeheimi N, Bakheit CS, et al. Neuropilin-1 Associated Molecules in the Blood Distinguish Poor Prognosis Breast Cancer: A Cross-Sectional Study. Sci Rep. 2017;7. doi:10.1038/s41598-017-03280-0

238. Deblois G, Chahrour G, Perry M-C, Sylvain-Drolet G, Muller WJ, Giguère V. Transcriptional control of the ERBB2 amplicon by ERRalpha and PGC-1beta promotes mammary gland tumorigenesis. Cancer Res. 2010;70(24):10277-10287. doi:10.1158/0008- 5472.CAN-10-2840

239. Ricoult SJH, Yecies JL, Ben-Sahra I, Manning BD. Oncogenic PI3K and K-Ras stimulate de novo lipid synthesis through mTORC1 and SREBP. Oncogene. 2016;35(10):1250- 1260. doi:10.1038/onc.2015.179

240. Walker SR, Nelson EA, Zou L, et al. Reciprocal Effects of STAT5 and STAT3 in Breast Cancer. Mol Cancer Res. 2009;7(6):966-976. doi:10.1158/1541-7786.MCR-08-0238

241. Walker SR, Nelson EA, Yeh JE, Pinello L, Yuan G-C, Frank DA. STAT5 Outcompetes STAT3 To Regulate the Expression of the Oncogenic Transcriptional Modulator BCL6. Mol Cell Biol. 2013;33(15):2879-2890. doi:10.1128/MCB.01620-12

242. Walker SR, Xiang M, Frank DA. Distinct roles of STAT3 and STAT5 in the pathogenesis and targeted therapy of breast cancer. Mol Cell Endocrinol. 2014;382(1). doi:10.1016/j.mce.2013.03.010

243. Stefanovic S, Deutsch TM, Wirtz R, et al. Impact of mRNA-Assessed Molecular Subtype Conversion, Intact and Apoptotic Circulating Tumor Cells on Survival of Metastatic Breast Cancer Patients: Proof of Principle. Diagn Basel Switz. 2020;10(6). doi:10.3390/diagnostics10060369

244. Klebe M, Fremd C, Kriegsmann M, et al. Frequent Molecular Subtype Switching and Gene Expression Alterations in Lung and Pleural Metastasis From Luminal A–Type Breast Cancer. JCO Precis Oncol. Published online July 24, 2020. doi:10.1200/PO.19.00337

245. Priedigkeit N, Hartmaier RJ, Chen Y, et al. Intrinsic Subtype Switching and Acquired ERBB2/HER2 Amplifications and Mutations in Breast Cancer Brain Metastases. JAMA Oncol. 2017;3(5):666-671. doi:10.1001/jamaoncol.2016.5630

232

246. Need EF, Selth LA, Trotta AP, et al. The unique transcriptional response produced by concurrent estrogen and progesterone treatment in breast cancer cells results in upregulation of growth factor pathways and switching from a Luminal A to a Basal-like subtype. BMC Cancer. 2015;15:791. doi:10.1186/s12885-015-1819-3

247. Treisman R. Regulation of transcription by MAP kinase cascades. Curr Opin Cell Biol. 1996;8(2):205-215. doi:10.1016/S0955-0674(96)80067-6

248. David M, Petricoin E, Benjamin C, Pine R, Weber MJ, Larner AC. Requirement for MAP kinase (ERK2) activity in interferon alpha- and interferon beta-stimulated gene expression through STAT proteins. Science. 1995;269(5231):1721-1723. doi:10.1126/science.7569900

249. Favata MF, Horiuchi KY, Manos EJ, et al. Identification of a novel inhibitor of mitogen- activated protein kinase kinase. J Biol Chem. 1998;273(29):18623-18632. doi:10.1074/jbc.273.29.18623

250. Waskiewicz AJ, Flynn A, Proud CG, Cooper JA. Mitogen-activated protein kinases activate the serine/threonine kinases Mnk1 and Mnk2. EMBO J. 1997;16(8):1909-1920. doi:10.1093/emboj/16.8.1909

251. Hsiao KM, Chou SY, Shih SJ, Ferrell JE. Evidence that inactive p42 mitogen-activated protein kinase and inactive Rsk exist as a heterodimer in vivo. Proc Natl Acad Sci U S A. 1994;91(12):5480-5484.

252. Schuringa JJ, Jonk LJ, Dokter WH, Vellenga E, Kruijer W. Interleukin-6-induced STAT3 transactivation and Ser727 phosphorylation involves Vav, Rac-1 and the kinase SEK- 1/MKK-4 as signal transduction components. Biochem J. 2000;347 Pt 1:89-96.

253. Chatterjee A, Ghosh J, Ramdas B, et al. Regulation of Stat5 by FAK and PAK1 in Oncogenic FLT3- and KIT-Driven Leukemogenesis. Cell Rep. 2014;9(4):1333-1348. doi:10.1016/j.celrep.2014.10.039

254. Wang R-A, Vadlamudi RK, Bagheri-Yarmand R, Beuvink I, Hynes NE, Kumar R. Essential functions of p21-activated kinase 1 in morphogenesis and differentiation of mammary glands. J Cell Biol. 2003;161(3):583-592. doi:10.1083/jcb.200212066

255. Anjum R, Blenis J. The RSK family of kinases: emerging roles in cellular signalling. Nat Rev Mol Cell Biol. 2008;9(10):747-758. doi:10.1038/nrm2509

256. Stepanenko AA, Vassetzky YS, Kavsan VM. Antagonistic functional duality of cancer genes. Gene. 2013;529(2):199-207. doi:10.1016/j.gene.2013.07.047

257. Huebner K, Procházka J, Monteiro AC, Mahadevan V, Schneider-Stock R. The activating transcription factor 2: an influencer of cancer progression. Mutagenesis. 2019;34(5-6):375- 389. doi:10.1093/mutage/gez041

258. Weiss WA, Taylor SS, Shokat KM. Recognizing and exploiting differences between RNAi and small-molecule inhibitors. Nat Chem Biol. 2007;3(12):739-744. doi:10.1038/nchembio1207-739

233

259. Broad Institute Cancer Cell Line Encyclopedia (CCLE). Accessed February 24, 2021. https://portals.broadinstitute.org/ccle

260. Ghandi M, Huang FW, Jané-Valbuena J, et al. Next-generation characterization of the Cancer Cell Line Encyclopedia. Nature. 2019;569(7757):503-508. doi:10.1038/s41586- 019-1186-3

261. Watson NA, Cartwright TN, Lawless C, et al. Kinase inhibition profiles as a tool to identify kinases for specific phosphorylation sites. Nat Commun. 2020;11(1):1684. doi:10.1038/s41467-020-15428-0

262. Chen RH, Sarnecki C, Blenis J. Nuclear localization and regulation of erk- and rsk- encoded protein kinases. Mol Cell Biol. 1992;12(3):915-927.

263. Wortzel I, Seger R. The ERK Cascade. Genes Cancer. 2011;2(3):195-209. doi:10.1177/1947601911407328

234

Vita Alicia Woock was born on January 7, 1990 in New Bern, North Carolina. Alicia attended Appalachian State University in the Blue Ridge Mountains. There, under the mentorship of Dr. Jennifer Perry Cecile, Alicia performed research and wrote her undergraduate honors thesis titled “Organic Anion Transport in live Caenorhabditis elegans viewed by fluorescence microscopy”. Highlights of her time at Appalachian State include receiving the Sigma Xi Outstanding Interdisciplinary Research Award for presenting her work at the Annual Celebration of Student Research and Creative Endeavors, as well as serving as the President of the Appalachian Chemical Society. Alicia graduated summa cum laude with a B.S. in Honors Chemistry with a minor in Biology in 2012. Alicia received a post-baccalaureate Intramural Training Research Award from the National Institutes of Health after graduating. She spent two years as a research assistant in the laboratory of Dr. Robert Innis in the Positron Emission Tomography (PET) Imaging Neurosciences section of the Molecular Imaging Branch of the National Institute of Mental Health. While there, Alicia co-authored three peer-reviewed publications. In 2014, Alicia was accepted into the MD-PhD program at Virginia Commonwealth University. In 2016, she was accepted into the Clinical and Translational Sciences doctoral program with a concentration in cancer and molecular medicine. As a doctoral student, she has been co-author on one publication, and has one first-author publication in review. During this time, she was awarded a poster award for her presentation at the FASEB Scientific Research Conference: GH/PRL Family in Biology & Disease (2019), as well as an Early Career Forum Travel Award from the Endocrine Society for her abstract for the ENDO 2020 Conference. After completing her PhD, Alicia will return to medical school and complete her clinical rotations.

235