NUCLEAR RECEPTOR REGULATION OF A LUMINAL BREAST

CANCER STEM CELL POPULATION

by

LYNSEY MAE FETTIG

B.A., University of North Dakota, 2009

B.S., University of Minnesota, 2011

A thesis submitted to the

Faculty of the Graduate School of the

University of Colorado in partial fulfillment

of the requirements for the degree of

Doctor of Philosophy

Cancer Biology Program

2019

This thesis for the Doctor of Philosophy degree by

Lynsey Mae Fettig

has been approved for the

Cancer Biology Program

by

Rebecca Schweppe, Chair

David Bentley

Peter Kabos

Maranke Koster

Diana Cittelly

Jennifer Richer

Carol A. Sartorius, Advisor

Date: May 17, 2019

ii Fettig, Lynsey Mae (Ph.D., Cancer Biology)

Nuclear Receptor Regulation of a Luminal Breast Cancer Stem Cell Population

Thesis directed by Professor Carol A. Sartorius

ABSTRACT

Crosstalk between nuclear receptors is emerging as important in regulating breast cancer

(BC) treatment response and progression. Estrogen receptor (ER), progesterone receptor (PR), and retinoic acid receptor (RARα) are all co-expressed in luminal subtype BCs that are generally associated with good prognosis. This is mainly due to the success of endocrine therapies targeting ER. However, eventual resistance to endocrine therapy is a long-standing clinical problem that affects ~30% of patients. About half of all luminal tumors contain subpopulations of ER negative cells that express cytokeratin 5 (CK5). CK5+ cells exhibit cancer stem cell (CSC) properties such as resistance to chemo- and endocrine therapies, and enhanced tumor initiating potential. In previous work we identified that progesterone (P4) expands the population of CK5+

BC cells and that retinoids block such expansion. Thus, we decided to investigate the mechanisms by which PR, RARα, and their cognate ligands regulate breast CSC activity, expression of the CK5 gene, and tumor growth. P4-expanded CK5+ cells were more tumorigenic than CK5− cells in vivo, and P4 treated BC cells formed larger mammospheres in vitro. Either silencing of CK5 via shRNA or co-treatment with retinoic acid (RA) abolished this P4 effect in mammospheres. Using promoter deletion constructs, we identified a region 1.1 kb upstream of the CK5 transcriptional start site (TSS) that is necessary for P4 activation. We next performed

ChIP and found that P4 induces association of both PR and RARα with the CK5 promoter at two distinct sites: a distal site (~1.1 kb upstream of the TSS) containing a putative progesterone response element and a proximal site (~130 bp upstream of the TSS) near a known retinoic acid

iii response element. RA blocked recruitment of RARα and other essential co-factors to the CK5 promoter. In ongoing studies, we are assessing genome-wide crosstalk between PR and RARα in

BC cells using RNA-seq and ChIP-seq. Treatment of BC xenografts in vivo with a retinoid reduced the accumulation of CK5+ cells during estrogen depletion. This reduction, together with the inhibition of CK5+ cell expansion through RAR/PR crosstalk, may explain the efficacy of retinoids in prevention of some BC recurrences.

The form and content of this abstract are approved. I recommend its publication.

Approved: Carol A. Sartorius

iv DEDICATION

I wish to thank the incredible people I have met who have encouraged me and helped me get to this point in my life. In particular, my teachers Phil Johnson and Brad Bachmeier who fostered my love of learning and showed me that, even though I am not an artist, I can still think creatively. Also, Dr. Kirsten Nielsen, Dr. Annabell Oh, Dr. Douglas Yee, and Kelly LaPara, who all saw something in me that I did not know was there and stepped in to help guide me through the process of becoming a scientist.

I would especially like to thank my mentor, Dr. Carol Sartorius, for taking a chance on me. Even though I came to you under unusual circumstances, you welcomed me into the lab.

You have not only been supportive of my career and allowed me to explore science in my own way, you have created an environment where I have been allowed to grow as a person and find my confidence. I cannot thank you enough.

I am immensely grateful for my lab mates and graduate school friends. If it were not for you, I would not have survived. I cannot imagine having a more supportive group of people by my side as we tried to find our paths through all of this. You helped me see the good sides and funny sides, and helped pull me through the hard parts. I will forever be thankful for each of our friendships.

I would also like to thank my parents, Kevin and Colleen, and my sister, Annie. Never did you tell me there were things in this life that were beyond my reach; instead you helped me figure out ways to make them happen, regardless of how asinine they initially seemed. The times you have come to my rescue – literally and figuratively – are innumerable. You have been there every time I have asked, and many where I did not have to. Without you wonderful people I would crumble.

v To my babies, Elliot and Wesley – you make life so much better on the hard days. I hope

I can make you proud.

Lastly, I wish to thank my husband, Jayden. You have an incredible talent for knowing how I need to be supported, whether it is letting me work by myself at home while you enjoy your time on the ski hill, celebrating with me when it has been a good day for science, or bringing home ice cream when it has in fact not been a good day for science. Our lives have changed so much since we started this adventure, but six years and two kids later I am so glad that you have been my partner through it all. Here’s to the next chapter in our life together. And I promise, no more school. Really. I’m done this time.

vi ACKNOWLEDGEMENTS

I wish to thank the University of Colorado Cancer Center Flow Cytometry Core,

Biorepository Core, Genomics and Microarray Core, and Tissue Culture Core supported by

P30CA046934 for their technical assistance and services. This work was supported by UL1

TR002535 via the Colorado Clinical and Translational Sciences Institute (LMF) and National

Institutes of Health grants NIH F31 CA210519 (LMF) and NIH 2R01 CA140985 (CAS). For animal work, our IACUC approval number is #000160.

vii TABLE OF CONTENTS

CHAPTER

I. INTRODUCTION ...... 1

Breast Cancer Incidence, Subtypes, and Treatment ...... 1

Cancer Stem Cells ...... 4

Cytokeratin Structure and Function ...... 6

Cytokeratins in Breast Cancer ...... 9

Nuclear Receptors ...... 12

Estrogen Receptors ...... 14

Estrogen in Cancer ...... 14

Progesterone Receptors ...... 15

Progesterone Receptor in Breast Cancer ...... 16

Retinoic Acid Receptors ...... 18

Retinoic Acid Receptors in Cancer ...... 19

Crosstalk Among Nuclear Receptors ...... 20

Thesis Direction and Summary ...... 21

II. MATERIALS AND METHODS ...... 23

Cell Culture and shRNA ...... 23

CK5 Promoter Deletion Constructs ...... 23

Luciferase Reporter Assay and siRNA ...... 23

Chromatin Immunoprecipitation Assay ...... 24

Mammosphere Formation Assay and Proliferation Assays ...... 24

Immunoblotting...... 25

viii Immunohistochemistry, Immunocytochemistry, and Proximity Ligation Assays ...... 25

Quantitative Reverse Transcription PCR and RNA-sequencing ...... 26

Limiting Dilution Analysis and Tumor Growth ...... 26

Statistical Methods ...... 27

III. CYTOKERATIN 5 IS DIRECTLY REGULATED BY PROGESTERONE AND IS

IMPORTANT IN MEDIATING PROGESTERONE-INDUCED CANCER STEM CELL

PROPERTIES ...... 29

Introduction ...... 29

Results ...... 30

Progesterone receptor regulates CK5 transcription through direct binding to the

proximal promoter ...... 30

Progesterone-expanded CK5+ breast cancer cells are more tumorigenic ...... 33

Progesterone-enhanced mammosphere formation of breast cancer cells requires

CK5 ...... 33

Discussion ...... 36

IV. RETINOIC ACID BLOCKS EFFECTS OF PROGESTERONE ON LUMINAL

BREAST CANCER CELLS ...... 40

Introduction ...... 40

Results ...... 41

Retinoids block progestin-induced CK5+ expression and mammosphere growth 41

Retinoic acid receptors are necessary for retinoid antagonism of progesterone-

dependent CK5 expression and mammosphere formation ...... 48

ix Retinoic acid antagonizes progesterone receptor regulation of CK5 by modulating

RARα recruitment near hormone response elements in the CK5 promoter ...... 50

Discussion ...... 53

V. RETINOIDS BLOCK ACCUMULATION OF CK5+ CELLS DURING ENDOCRINE

THERAPY TREATMENT ...... 56

Introduction ...... 56

Results ...... 57

Discussion ...... 59

VI. RETINOID ANTAGONISM OF PROGESTERONE RECEPTOR ALTERS

GENOMIC RECRUITMENT AND TRANSCRIPTIONAL PROGRAMS ...... 61

Introduction ...... 61

Results ...... 61

Discussion ...... 65

VII. DISCUSSION AND FUTURE DIRECTIONS ...... 69

REFERENCES ...... 74

APPENDICES ...... 94

A. Colorado Clinical and Translational Sciences Institute Shadowing Summary ...... 94

B. Supplemental Table 1 ...... 97

x LIST OF TABLES

Table 2.1 Antibodies ...... 28

Table 2.2 qPCR Primers (5’ to 3’) ...... 28

Table 3.1 Tumor-initiating capacity of P4-induced CK5+ compared to CK5- T47D breast cancer

cells...... 35

Table 6.1 Top 40 genes inversely regulated by P4 and RA by absolute expression...... 64

xi TABLE OF FIGURES

Figure 1.1 Clinical outcomes of breast cancer patients by intrinsic subtype...... 2

Figure 1.2 Cytokeratin structure...... 7

Figure 1.3 Nuclear receptor structure...... 11

Figure 1.4 Progesterone receptor regulation and structure...... 17

Figure 3.1 PR is recruited to the proximal CK5 promoter near a PRE that is necessary for P4

transcriptional activation...... 31

Figure 3.2 CK5+ cells are more tumor initiating...... 34

Figure 3.3 The P4 dependent increase in mammosphere size requires expression of CK5...... 37

Figure 4.1 T47D proliferation and CK5 reporter respond to RA in a dose-dependent manner. .. 42

Figure 4.2 RA blocks P4-mediated CK5 expression and P4 induction of large mammospheres. 43

Figure 4.3 Mammosphere formation is inhibited by retinoids in a concentration-dependent

manner...... 46

Figure 4.4 RA inhibition of P4-dependent expression of CK5 requires RARs...... 47

Figure 4.5 Retinoid blockade of P4-induced CK5 expression and mammosphere formation

occurs through RARs...... 49

Figure 4.6 PR and RARα recruitment to the CK5 promoter at the distal PRE region and a

proximal RARE-containing region is P4-dependent and is blocked by RA...... 51

Figure 4.7 Diagram of proposed coactivator bridging between promoter elements...... 54

Figure 5.1 Co-treatment with retinoids during estrogen depletion reduces accumulation of CK5+

breast cancer cells...... 58

Figure 6.1 RNA-seq data: P4 regulated transcripts and downstream pathways...... 62

Figure 6.2 Upstream regulators of P4 and RA transcriptomes...... 66

xii Figure 6.3 PR and RARα are recruited to tRNA promoters with P4 and RA cotreatment...... 66

Figure 6.4 PR antibody optimization for ChIP...... 68

xiii ABBREVIATIONS

3'UTR 3'-untranslated region AF Activation function AI Aromatase inhibitor ALDH1 Aldehyde dehydrogenase AML Acute myeloid leukemia AR Androgen receptor ATRA All-trans retinoic acid BrdU 5’-bromo-2’-deoxyuridine ChIP Chromatin immunoprecipitation ChIP-seq Chromatin immunoprecipitation sequencing CK Cytokeratin CK5 Cytokeratin 5 CSC Cancer stem cell DAPI 4',6'-diamidino-2-phenylindole DBD DNA binding domain DR Direct repeat E2 Estrogen ER Estrogen receptor ER-8 Everted repeat - 8 ERE Estrogen response element ERR Estrogen related receptor EV Empty vector Fen Fenretinide, 4-hydroxy(phenyl)retinamide, 4-HPR GR Glucocorticoid receptor HER2 Human epidermal growth factor receptor 2 HRE Hormone response element HSP Heat shock protein ICC Immunocytochemistry ICI ICI 182 780, Fulvestrant IHC Immunohistochemistry IR Inverted repeat ISH In situ hybridization LBD Ligand binding domain miRNA microRNA MR Mineralocorticoid receptor mRNA Messenger RNA OE Overexpression P4 Progesterone PI Propidium iodide PR Progesterone receptor PRE Progesterone response element

xiv qRT-PCR Quantitative real time polymerase chain reaction RA Retinoic acid, 9-cis retinoic acid RAR Retinoic acid receptor RARE Retinoic acid response element RNA-seq Ribonucleic acid sequencing RXR Retinoid X receptor Ser Serine SERD Selevtive estrogen receptor degrader SERM Selective estrogen receptor modulator SP Side population Tam Tamoxifen Thr Threonine TRADD TNFR1-associated death domain TSS Transcriptional start site TTNPB Arotinoid acid, 4-[(E)-2-(5,6,7,8-Tetrahydro-5,5,8,8-tetramethyl-2- naphthalenyl)-1-propenyl]benzoic acid Tyr Tyrosine

xv CHAPTER I

INTRODUCTION

Breast Cancer Incidence, Subtypes, and Treatment

Breast cancer has both the highest incidence and cancer related mortality of any cancer in women worldwide1. In the United States, it is the most commonly diagnosed cancer in women, accounting for 30% of all cancer diagnoses, and is the second leading cause of cancer related deaths at about 14%2.

Breast cancer is a heterogeneous disease, which is often categorized by molecular profiling into six intrinsic molecular subtypes (Figure 1.1), each with differing prognoses:

Luminal A, Luminal B, HER2-enriched, Claudin-low, Basal-like, and Normal Breast-like3. The most common subtype is Luminal A, which is defined by expression of estrogen receptor (ER) and progesterone receptor (PR), and has the best prognosis. Luminal B tumors also express ER but, compared to Luminal A tumors, have poorer survival, higher expression of proliferation genes, lower expression of PR, and a higher likelihood of expressing human epidermal growth factor receptor 2 (HER2). HER2-enriched tumors have overexpression of HER2, often lack expression of ER and PR, and have the highest mutational burden of all the subtypes2. They also have among the shortest relapse-free survival interval and lowest overall survival4. Claudin-low and basal-like are both triple-negative tumors that lack expression of ER, PR, and HER2.

Claudin-low tumors are enriched in biological properties associated with normal mammary stem cells, have a lower response rate to standard neoadjuvent chemotherapy, and a worse prognosis than luminal tumors3. Basal-like breast tumors have similarly poor survival, high expression of

proliferative genes, and express cytokeratins (CK) that are normally present in the basal layer of

skin, such as CK5, CK14, and CK172.

1

Figure 1.1 Clinical outcomes of breast cancer patients by intrinsic subtype. Kaplan-Meier analysis of relapse-free survival and overall survival curves using the UNC337 data set with Normal Breast-like samples excluded. Figure and legend adapted from Pratt and Perou3.

2 While this specific molecular subtyping is not currently used clinically in the United

States, expression of ER, PR, and HER2 is assessed by immunohistochemistry (IHC) for each

patient. The hormone receptor status coupled with other factors such as tumor grade, lymph node

status, and overall patient health, are then used to make treatment decisions. ER positive tumors

are treated with endocrine therapy, such as selective estrogen receptor modulators (SERMs),

selective estrogen receptor degraders (SERDs), or aromatase inhibitors (AIs), which prevent the

conversion of testosterone to estrogen. For certain ER positive tumors, molecular profiling using

PAM50, Oncotype DX, or Mammaprint can be used to determine if the patient is at high or low

risk for recurrence, an indicator of whether they would benefit from chemotherapy5, 6. For tumors

that have overexpression of HER2 by IHC or in situ hybridization (ISH), patients are treated with

HER2-targeting antibodies such as trastuzumab and pertuzumab in addition to chemotherapy.

For patients with triple negative tumors, there are currently few targeted therapies though there

are several in clinical trials7. These clinical trials include targeting of androgen receptor (AR),

which is highly expressed in triple negative breast cancers, as well as immunotherapies, though

these have not been widely effective due to the relatively low mutational burden of these

tumors8, 9. Additionally, PARP inhibitors have shown efficacy when the patient also carries a

BRCA mutation10. Due to the aggressive nature of these tumors, patients are often treated with

chemotherapy and radiotherapy.

Regardless of treatment, many patients exhibit de novo or acquired resistance. In ER positive tumors, development of resistance to endocrine therapies occurs most often in one of four ways: 1. de novo resistance in which the patient presents with resistance to endocrine therapy or recurs soon after initiation of hormonal therapy without responding to additional endocrine therapies; 2. de novo resistance to some but not all endocrine therapies; 3. acquired

3 resistance developed after having an initial response, followed by temporary response to

additional hormonal therapies leading to eventual resistance to all endocrine therapies; 4.

progression several years after cessation of a hormonal therapy, followed by a transient or lack of

a response to the same agent11. Despite the relatively favorable initial prognosis, risk of

recurrence for patients with ER positive breast cancer continues as far out as 20 years, with risk

of distant recurrence between 10-26% depending on initial tumor grade12.

There are several predominant theories that may explain the ability of these tumors to

adapt and evade treatment-induced death. One hypothesis suggests that cancers dependent on

growth factor signaling, such as HER-2 driven breast cancers, may be able to switch their growth

factor dependency in the face of targeted therapy13. The clonal evolution hypothesis speculates

that, because cancers are heterogeneous by nature, selective pressure induced by targeted or

chemotherapies allows the destruction of some clones within the tumor while others continue to

survive14. Another potential explanation for relapse is the cancer stem cell hypothesis, discussed

in detail below.

Cancer Stem Cells

The cancer stem cell (CSC) hypothesis posits that in a tumor there is a minor

subpopulation of cells that is more quiescent, more therapy resistant, and has the capacity to

repopulate a tumor15. These attributes can be tested in vitro using a variety of methods.

Quiescence can be assessed by investigating cell cycle status via flow cytometry after staining

for DNA content with propidium iodide (PI), 4',6'-diamidino-2-phenylindole (DAPI) or 5’-

bromo-2’-deoxyuridine (BrdU) versus expression of cyclin D, cyclin E, cyclin A or cyclin B116.

This bivariate analysis allows the delineation of G0 from G1 cells. Capacity to repopulate a tumor is often tested in vitro via tumorsphere formation assays where cells are plated in ultra low

4 attachment conditions and assessed for their ability to form 3D spheroids. In vivo this capability is tested via a limiting dilution tumor initiation xenograft study where decreasing concentrations of cells are injected into mice and assessed for their ability to form tumors at the lowest concentrations.

John Dick and colleagues were the first group in the modern era to show experimental evidence for the CSC hypothesis in acute myeloid leukemia (AML). They discovered that only cells expressing CD34 on their surface are capable of initiating AML xenografts in immunocompromised mice17. Following this discovery it was hypothesized that CSCs also existed in solid tumors. The first support for the CSC hypothesis in solid tumors was identified in breast cancer. Al-Hajj et al. showed that as few as 100 CD44+CD24-/low cells from patient breast tumor samples are able to form tumors in immunocompromised mice, whereas tens of thousands of cells of alternate phenotypes failed to form tumors18. This was subsequently described using various markers in several other cancer types including multiple myeloma19, melanoma20, prostate cancer21, colorectal cancer22, and pancreatic cancer23.

Since this discovery, additional putative markers of the breast CSC population have been identified. It is important to note that several groups have found that these markers may have overlapping populations but they are not identical24, 25. Both normal and cancerous human breast mammary epithelial cells with increased aldehyde dehydrogenase (ALDH1) activity have stem- like properties24. Another marker, Sox2, is typically expressed during development then shut off in adult tissues, but is commonly reexpressed in cancer26, 27. Sox2 expression is enriched in both breast cancer cell lines as well as patient tumor samples grown in 3D culture compared to the same samples grown in adherent culture27. Further, Sox2 overexpression is able to increase the number of tumorspheres in 3D culture, while silencing of Sox2 expression decreases the capacity

5 of cells to form tumorspheres27. Yet another fraction of putative breast CSCs has been identified by their ability to pump out dyes such as Hoechst 33342 via the ATP-binding cassette transporter

ABCG2, and has been denoted the side population (SP)28, 29. This SP has been shown in breast cancer cell lines to display CSC characteristics such as enriched CSC gene expression, tumor initiation, and colony formation30, 31. Cells that express an additional cell surface marker, CD133, sorted from mouse mammary tumors have a greater capacity for tumor formation as well as the ability to reconstitute both cell fractions from the original cell population unlike their CD133- counterparts32. Additionally, expression of cell surface marker CD49f (α 6-integrin) was significantly higher in breast cancer cells grown in 3D tumorspheres compared to 2D growth control33. In mice, CD61 expression marks a population of cells with higher tumor initiating potential in tumors driven by MMTV-Wnt-1 or that are heterozygous for p53 mutation34.

While these markers have been successful in identifying tumor-initiating cells within the highly aggressive triple negative tumor types and murine models, several are not reflected in luminal tumors. Another putative marker of the CSC population that is found in many triple negative and luminal tumors, however, is CK5. CK5 is a marker of the committed stem cell and progenitor cell populations in the normal human breast35. We and others have shown in luminal cell lines in vitro that CK5 expressing cells are relatively more quiescent, therapy resistant, and tumor initiating than their CK5 negative counterparts36-38. However, these characteristics do not hold for populations expressing other markers of the normal breast stem cell, indicating that CK5 may be a rational marker for the luminal breast CSC population.

Cytokeratin Structure and Function

Cytokeratins (CKs) are intermediate filament proteins that play important roles in cell structure and motility, and are the major cytoskeletal component in epithelial cells. In humans,

6

Figure 1.2 Cytokeratin structure. (A) All cytokeratins share a common structure, with a head, rod, and tail domain. The α-helical rod domain is composed of four helices (coil 1A, coil 1B, coil 2A, and coil 2B) separated by three linker domains (L1, L12, and L2). Adapted from Snider et al.39 (B) Cytokeratins are assembled first by the heterodimerization of one type 1 and one type 2 cytokeratin, followed by the formation of staggered, antiparallel tetramers, then end-to-end binding of tetramers creating protofilaments. Finally, the assembly of 8 protofilaments form a full intermediate filament. Adapted from Thomas40.

7 there are 54 known functional cytokeratins that are differentially expressed based on tissue type,

cell type, developmental stage, and differentiation status (reviewed in Bragulla et al.41). These

are categorized into two subtypes: Type 1 cytokeratins (CK9-28, 31-40) are relatively small and

acidic, and primarily found on human chromosome 17, with the exception of CK18 42; Type 2

keratins (CK1-8, and 71-86) are larger, more basic, and clustered with CK18 on chromosome 12

42. All CKs share the same three domain structure: an N-terminal head domain, a central α- helical rod domain, and a C-terminal tail domain (Figure 1.2A)39. While there is substantial variability in the head and tail domains among CKs, the central α-helical rod domain is highly conserved and comprised of four helices (coil 1A, coil 1B, coil 2A, and coil 2B) separated by three linker domains (L1, L12, and L2)39. When assembling into filaments, CKs form coiled parallel heterodimers each consisting of a type 1 and type 2 CK (Figure 1.2B). Two heterodimers then go on to assemble in a staggered, anti-parallel manner to form tetramers, which then bind end to end to form protofilaments. Eight protofilaments are then assembled to form a full ~10 nm intermediate filament.

CK filaments are highly dynamic – they nucleate toward the periphery of cells at focal adhesions then, as they are elongated, move inward toward the nucleus via actin-dependent transport43. During this time, they are integrated into CK networks and when they reach the nucleus, mature CK filament bundles either remain as part of the nest-like perinuclear cytoskeleton or are disassembled into soluble tetramers43.

The major regulators of CK function, stability, and solubility are posttranslational modifications, which include phosphorylation, O-linked glycosylation, ubiquitination,

SUMOylation, acetylation, and transamidation (Figure 1.2A)39. Phosphorylation of serine (Ser)

or threonine (Thr) residues, for example, primarily promotes CK solubility, which is important

8 for the dynamic subunit exchange seen in the mature perinuclear CK cytoskeleton.

Phosphorylation of tyrosine (Tyr) 267 on CK8, however, promotes proper filament assembly and

subunit insolubility44. Posttranslational modifications can also be induced by cellular stress, allowing for quick alterations to CK network dynamics and organization. Under stress conditions, JNK and p38 MAPK are rapidly activated and recruited to CKs, allowing them to phosphorylate Ser74 on CK8 or the analogous Thr residue on CK5 or CK6 leading to filament disassembly45, 46. Because CKs are able to integrate a multitude of cellular signals via alterations in their posttranslational modifications and expression, they are exquisite sensors of cellular conditions and the microenvironment, which may explain their altered expression in some cancers.

Cytokeratins in Breast Cancer

Expression of particular CKs has been linked to differences in prognosis in multiple solid tumors47-50. In breast tumors, IHC analysis of 564 patient samples with follow up showed that positive staining for CK5/6 and/or CK17 was associated with significantly worse prognosis than those patients who had negative staining for both, independent of tumor size, tumor grade, or

ER, HER2, or GATA-3 expression status47. Additional studies have also seen this association between myoepithelial markers and basal/myoepithelial cell cytokeratin expression and poorer prognosis, early relapse, and reduced overall survival51-53. In a large 1917 patient cohort, CK5/6 expression was significantly positively correlated with histological grade, tumor size, local and regional recurrence, distant metastasis, and death from breast cancer48. While many of these studies show that CK5 in particular is a signature marker of poor prognosis in basal-like breast cancer, it is also associated with worse prognosis within luminal ER+ tumors54,55.

9 In line with their functional importance in cell structure, CKs have been shown to play a

role in tumor migration. Cells that express CK14, a known binding partner of CK5, are enriched

in lung metastases in luminal murine mammary tumor models, and have been found to lead

collective invasion in luminal A, luminal B, and triple negative murine mammary tumor models;

this invasion is lost with knockdown of CK14 though the exact mechanism driving this was not

explored56.

Beyond their role as biomarkers of prognosis and importance in cell structure, CKs also

play a role in other cell processes through alterations in protein trafficking or signaling.

Modulation of their expression can have important consequences; for example, loss of CK8 has

been shown to predispose cells to apoptosis via increased expression of Fas, while

overexpression of CK8 leads to resistance to multiple types of chemotherapies including

dihydrofolate inhibitors, DNA alkylating agents, and microtubule destabilizers57, 58. Interestingly

this desensitization to chemotherapy occurs without affecting drug uptake or efflux57. CK18 and

CK14 have also both been shown to suppress caspase 8 activation through TNFR1 via

association with the C-terminal portion of TNFR1-associated death domain protein (TRADD), a key downstream mediator of TNFR1-induced cytotoxicity59. Even more simplistically, CKs have been shown to act as phosphate sponges, whereby kinases that would typically phosphorylate and activate mediators of apoptosis instead phosphorylate CKs, dampening pro-apoptotic signals60. Further, CK17 and CK5 in particular have been shown to interact with 14-3-3 proteins, which can affect Akt signaling and progression through the cell cycle61, 62. Together these studies indicate that CK5 could be more than just a marker of the CSC population; it may have a functional role in determining breast cancer sensitivity to a variety of treatments, including endocrine therapy.

10 A Response Element Receptor Sequences Progesterone receptor GnACAnnnTGTTCT

Estrogen receptor AGGTCAnnnTGACCT

Retinoic acid receptor AGGTCAnnnnnAGGTCA

Retinoid X receptor AGGTCAnTGACCT

B

Figure 1.3 Nuclear receptor structure. (A) Most nuclear receptors contain a variable N-terminal region (A/B), a DNA binding domain (DBD) (C), a variable hinge region (D), a ligand binding domain (E), and a variable C-terminal region. Differences in the DBD lead to binding to similar but relatively selective response elements. (B) Functional properties of each nuclear reception region. Adapted from Tata et al.63

11 Nuclear Receptors

Nuclear receptors are members of a superfamily of DNA-binding transcription factors.

Structurally, nuclear receptors consist of five regions: the A/B, C, D, E, and F regions (Figure

1.3A, reviewed in Germain et al.64). The N-terminal A/B region is highly variable in sequence and length among nuclear receptors, and contains the ligand independent activation domain activation function 1 (AF1) (Figure 1.3B). It is also the target of several posttranslational modifications. The C region is the most highly conserved region and contains a DNA-binding domain (DBD) comprised of two cysteine-rich zinc finger motifs, two α helices, and a COOH extension. This DBD allows the receptors to recognize and bind specific sequences in the DNA termed hormone response elements (HREs) (Figure 1.3A). Similar to the A/B region, the D region is poorly conserved and is a target for posttranslational modifications. It also serves as a hinge region to permit the DBD and ligand binding domain (LBD) to adopt different conformations. This region importantly also contains the nuclear localization sequence. Region E contains several important functions. The LBD is contained in region E, as is a ligand-dependent activation domain, AF2. This region, in conjunction with the DBD, also allows receptor dimerization. The final region, the F region, is not contained in all nuclear receptors and its function is poorly understood.

Traditionally, nuclear receptors have been classified into four types according to their mode of action (reviewed by Sever and Glass65). Type I nuclear receptors, such as ER and PR, are activated through binding of their ligand which, in turn, allows dissociation from chaperones, dimerization, and binding to HREs in DNA to facilitate transcriptional activation65. Depending on the receptor and cell type, these receptors either reside in the cytoplasm associated with heat shock proteins and translocate to the nucleus upon ligand binding, and/or primarily reside in the

12 nucleus whereby the addition of ligand promotes a change in intranuclear localization and

binding to HREs66. Most often these HREs are inverted repeat sequences. Type II nuclear

receptors, such as retinoic acid receptor (RAR), are bound to DNA in the absence of ligand and

form heterodimers, most often with another type II receptor, retinoid X receptor (RXR). This

leads to transcriptional repression via recruitment of corepressors such as NCoR and SMRT65.

Type III nuclear receptors act similarly to type I nuclear receptors but bind direct repeat HREs rather than inverted65. Type IV nuclear receptors also act like type I receptors, however they act

as monomers instead of dimers, binding to half site HREs65.

More recently, however, nuclear receptors have been reclassified into a newer system

based on sequence alignment and phylogenetic tree construction67. Broadly, nuclear receptors are

categorized into six subfamilies (NR1 through NR6), most of which have additional group divisions. These groups are defined by at least 80-90% homology in the DBD and at least 40-

60% homology in the LBD. There is an additional subfamily (NR0) that is used as a catchall for other unusual nuclear receptors irrespective of their evolutionary origin so long as they contain either a DBD or LBD. The NR1 subfamily includes RARs (RARα (NR1B1), RARβ (NR1B2),

RARγ (NR1B3)) and 24 other nuclear receptors that primarily heterodimerize with RXRs68. The

NR2 subfamily contains 11 receptors including RXRα (NR2B1), RXRβ (NR2B2), and RXRγ

(NR2B3). The NR3 subfamily includes the six steroid receptors (ER (ERα (NR3A1), ERβ

(NR3A2)), PR (NR3C3), androgen receptor (AR, NR3C4), glucocorticoid receptor (GR,

NR3C1), mineralocorticoid receptor (MR, NR3C2)), and two orphan estrogen related receptors

(ERRα (NR3B1), ERRβ (NR3B2)). The NR4 subfamily contains the four members of the nerve

growth factor-induced clone B group of orphan receptors. The NR5 subfamily has four group

members - the steroidogenic factor 1 and the receptors related to the Drosophila FTZ-F1. The

13 NR6 subfamily only contains the GCNF1 receptor. Below we focus and expand on the nuclear

receptors most relevant to this project: ER, PR, and RAR.

Estrogen Receptors

There are two paralogs of ER, ERα and ERβ, which are transcribed from genes located on two different chromosomes69. ERα and ERβ share very similar sequence and structural

homology in their DBD and LBD with their primary difference being in the N-terminal

domain66. Both ERs bind 17β-estradiol and recognize the same estrogen response element (ERE)

in the DNA: AGGTCAnnnTGACCT, where n is any nucleotide70. While co-expression of ERα and ERβ does occur in some tissue and cell types, they also exhibit differential tissue and cell expression patterns, are regulated independently of each other, and are functionally distinct71, 72.

ERs are heavily regulated by posttranslational modifications. ERα in particular has about 22 sites that are subject to modification including phosphorylation, methylation, acetylation,

SUMOylation, and ubiquitination73. These modifications alter ER hormone sensitivity, transcriptional activity, and stability74, 75. Posttranslational modifications are also key in mediating nongenomic effects of ER. Interestingly, ER is targeted for localization to the plasma membrane via a palmitoylation site in its LBD (Cys 447)76. Further, methylation of ERα in the

DBD allows interaction with PI3K in the cytoplasm, which further recruits Src and FAK, and leads to downstream signaling77.

Estrogen in Cancer

ER expression has been reported in almost 30 types of cancer, though this has

predominantly been in hormone-sensitive tumors such as breast, ovarian, endometrial, and

prostate78, 79. In breast cancer, estrogen through ERα acts in multiple ways to support tumor

survival. ERα drives proliferation by upregulation of c-Myc and cyclin D180, 81. Additionally,

14 82 ERα inhibits apoptosis through upregulation of Bcl-2 and Bcl-XL . The role of ERβ in breast

cancer, however, is still controversial in part due to lack of a clinically standardized antibody,

and in part due to conflicting in vitro studies showing it to be both pro-survival and pro-

apoptotic83. Regardless, the ability of estrogen to support tumor development and progression in

breast cancer has made targeting ERs wildly successful, though resistance to ER-targeting

therapies often develops.

In addition to their direct effects on the tumor cells themselves, ERs also have effects on

the tumor microenvironment. Indeed, estrogen is important in regulating immune response via

transcriptional regulation within immune cells, regulation of lymphopoiesis, and regulation of

immune cell differentiation84-86. Further, estrogen induces secretion of IL-8 and vascular

endothelial growth factor to induce angiogenesis, providing additional nutrients to support tumor

growth87.

Progesterone Receptors

PR has two isoforms (PR-A and PR-B), which are produced from different promoters within the same gene88. PR-A differs from PR-B by truncation of the first 164 amino acid

residues. This N-terminal region contains a third activation domain, AF-3, which contributes to

the strong transcriptional activity of PR-B. As the DBD is identical, PRs bind the same

progesterone response element (PRE) in DNA (GnACAnnnTGTTCT), identified using

mutational studies and later verified by ChIP-seq89, 90. Despite this, the two isoforms have

overlapping but unique transcriptomes90, 91. Interestingly, while PR-A and PR-B are both usually

expressed in target tissues, the ratio of PR-A to PR-B can vary substantially leading to tissue and cell specific activity. Genetic experiments in mice have shown that in the uterus, PR-A facilitates the majority of progesterone effects, while in the breast PR-B is the predominant isoform92, 93.

15 Similar to ERs, PRs are regulated in large part by posttranslational modifications (Figure

1.4A). Full length PR-B has a total of 14 phosphorylation sites, four of which are particularly

important in breast cancer – serines (Ser) 81, 294, 345, and 400. Ser81 is involved in gene

regulation; Ser294 is involved in PR shuttling, gene regulation, ligand preference, and receptor

turnover; Ser345 is involved in gene regulation via SP1 tethering; Ser400 is a basal

phosphorylation site that has been shown to enhance ligand-independent transcriptional activity

(reviewed in Hagan et al.94). These sites are phosphorylated by kinases such as CDK2 and

MAPK. Additional sites are SUMOylated, acetylated and ubiquitinated. All of these marks

integrate cellular signals to modify the responsiveness of PR with or without ligand.

Progesterone Receptor in Breast Cancer

In patient tumors, PR expression has been assessed as a surrogate for ER activity,

meaning coexpression of ER and PR indicates a good prognosis and a higher likelihood of

response to ER-targeting therapies. However, PR has been identified as a therapeutic target in its

own right, though the PR field has increasingly noted a dichotomy in progestin action that is

summarized as “context dependent” 95. Clinically, both agonists (ligands that activate canonical

PR regulated genes) and antagonists (ligands that repress canonical PR regulated genes) have

shown efficacy in a subset of patients (reviewed in Carroll et al.96). However, additional studies

link the use of synthetic progestins to increased breast cancer risk97-99. In our own studies of

patient-derived xenograft models, we have found a similar bivariate response to progestins in our

ER+ tumors – some tumors have their growth repressed by the addition of progestins, while

others have increased growth response (data not shown). The mechanism behind this is still

under investigation. This contradictory role of PR in breast cancer warrants further investigation

so we can better understand the differential downstream regulators of its effects.

16 A AF3

B

Figure 1.4 Progesterone receptor regulation and structure. (A) Posttranslational modifications affect PR function and its ability to interact with different ligands. Adapted from Grimm et al.100 (B) Structure of dimerized PR DNA binding domain (DBD) and C-terminal extension (CTE) in complex with DNA. Adapted from Hill et al.101, 102.

17 One important aspect of PR biology that emerged in the last decade is its ability to

expand breast cancer stem cells (CSCs). Such cells have increased tumor initiating and

tumorsphere forming potential, drug insensitivity, and relative quiescence. Breast cancer cells

that are CD44+/CD24-/low, ALDH1+, or CK5+ all correspond with increased breast CSC activity, and progestins (both natural hormone progesterone and synthetic analogs) increase each of these populations in ER+/PR+ breast cancer cell lines, while antagonists, where tested, block these actions (reviewed in Bernstein103). Of note, recent studies suggest that the different PR isoforms may support different functional CSC characteristics104.

Retinoic Acid Receptors

The retinoic acid receptor subfamily is comprised of three members, RARα, RARβ, and

RARγ, which are each transcribed from different genes that likely resulted from a genomic duplication event105. All three RARs form heterodimers with RXRs and can recognize a large number of retinoic acid response elements (RARE) in DNA: an inverted repeat (IR) 0

(AGGTCATGACCT), an everted repeat (ER)-8 (TGACCTnnnnnnnnACTGGA), a direct repeat

(DR)-1 (AGGTCAnAGGTCA), a DR-2 (AGGTCAnnAGGTCA) or, most commonly identified, a DR-5 (AGGTCAnnnnnAGGTCA)70, 106. As previously discussed, RARs are generally thought to sit on these RAREs in the absence of ligand and recruit co-repressors such as NCoR and

SMRT107, 108. Upon ligand binding, conformational changes induce loss of repressors and recruitment of co-activators, which in turn recruit larger complexes containing chromatin remodelers such as CBP108, 109. New studies have uncovered that this may not be the only way in which RARs regulate transcription. They may, in fact, be recruited to DNA as cofactors with activation of other nuclear receptors110. Natural ligands for RARs are the vitamin A derivatives all-trans retinoic acid (ATRA) and 9-cis retinoic acid (9-cis RA, RA); an additional vitamin A

18 derivative, 13-cis retinoic acid, selectively binds and activates RXRs. Several synthetic ligands with varying degrees of specificity and affinity for the receptors have also been developed

(reviewed in di Masi106).

Similar to other nuclear receptors, RARs can exert non-genomic effects. In multiple cancerous and non-cancerous cell types, ATRA stimulation has been shown to activate p38

MAPK signaling111. RARα in particular has also been shown to associate with lipid rafts in the cell membrane and induce phosphorylation of MAPK via PI3K112. The primary focus of our studies, however, is on the genomic effects of RARs.

Retinoic Acid Receptors in Cancer

Retinoids have been extensively investigated for the treatment or prevention of cancer,

predominantly due to their ability to induce differentiation and therefore halt proliferation. RA,

in fact, is potently antiproliferative in breast cancers in vitro113. While in some hematological

malignancies, such as APL and certain T-cell lymphomas, retinoids have been overwhelmingly

successful, in solid tumors their efficacy has been very mixed (reviewed in Connolly et al.114). In

head and neck cancer, the use of retinoids after completion of surgery or radiation showed little

promise115. In lung cancer, retinoids have been disappointing in the preventative setting, however

they have shown some benefit in response rate and progression free survival when combined

with standard chemotherapy116-118.

In breast cancer, results have been equally contradictory. In a primary prevention study

using the RAR-specific agonist fenretinide (Fen), either low dose tamoxifen (Tam) or low dose

Fen both led to a reduction in annual odds versus placebo in high risk women119. Similarly, in a

secondary prevention study, there was no difference in rates of breast cancer in the total study

population with Fen treatment, however there was a 35% reduction in occurrence in

19 premenopausal women specifically120. In studies in women with metastatic disease, either 13-cis

RA (isotretinoin) or 9 cis-RA (alitretinoin) was used in combination with Tam. Interestingly,

there was no difference in response between patients receiving Tam alone versus Tam plus 13-

cis RA, which only targets RXR121; however patients who received 9-cis RA, which targets RAR

and RXR, in combination with Tam saw a benefit122. While this study was small (12 patients), of

the 9 assessable patients, two had durable response, one had partial response with stable disease

for 18 months, and one had complete remission. Further, in a recent metabolomics study looking

at 46 patients with clinically localized breast cancer, 41 metastatic patients, and 49 normal

samples, it was shown that serum levels of 9-cis RA in particular is decreased significantly with

breast cancer progression123. Additionally the authors found that treatment with RA or

knockdown of the intracellular enzyme that processes RA to an inactive form (aldehyde

dehydrogenase 1 family member A1) decreases the ability of the highly metastatic breast cancer

cell line MDA-MB-231 to migrate and invade. Taken together these studies indicate that there

may be benefit to targeting RAR in conjunction with endocrine therapy, which warrants further

investigation.

Crosstalk Among Nuclear Receptors

Potential interaction between nuclear receptors has been studied for decades by assessing the effects of activation of multiple receptors on transcription124, 125. Recently interest in crosstalk among these receptors has been renewed, and mounting evidence suggests crosstalk among these receptors is far more prevalent than previously thought. One such example is ER and PR. Unlike in the normal human breast where ER and PR are typically expressed in separate cell populations, in breast cancer their expression becomes tightly correlated126. This led to several

20 studies that showed, in luminal breast cancer cells, ER and PR physically interact and are both required for regulation of certain progestin or estrogen regulated genes127-130.

Indeed interaction between other steroid receptors has been seen as well. AR has been shown to both antagonize and cooperate with ER function in a similar manner to PR131-133. In breast cancer, glucocorticoid receptor (GR) and mineralocorticoid receptor (MR) have been shown to interact with PR; addition of their ligands mimics the effects of progesterone, however this effect is ablated with loss of PR expression134. Steroid receptors have also been shown to interact with other types of nuclear receptors. RARα in particular is positively regulated by estrogens in breast cancer cells and is a necessary cofactor for ER transcriptional programs 110,

135. Together, these studies show that there is great need to consider the larger nuclear receptor context when investigating their mechanism of action.

Thesis Direction and Summary

ER positive breast cancer may initially have a more favorable prognosis than other molecular subtypes, however the risk of recurrence can go out as far as 20 years. The presence of a persistent cancer cell population poses a therapeutic challenge that needs to be addressed.

Previously our group and others have shown that a subpopulation of cells expressing CK5, a marker of the progenitor cell population in the normal human breast, has properties in line with the cancer stem cell hypothesis – they are more tumor initiating and more endocrine and chemotherapy resistant than their CK- counterparts – and they are enriched in patients samples following endocrine therapy treatment. Gaining a better understanding of how these cells are regulated is needed to develop therapies to target this population. In response, Chapter III will directly examine how PR regulates expression of CK5, and if CK5 itself is important in facilitating the tumor initiating characteristics of these cells in response to progesterone (P4)

21 stimulation. Using chromatin immunoprecipitation (ChIP), we show that PR directly regulates

transcription of CK5 via recruitment to a response element in the CK5 promoter. Additionally

we show that knockdown of CK5 expression is sufficient to attenuate the increase in

mammosphere potential induced by P4. Thus we hypothesized that suppression of CK5 may be

utilized to prevent the accumulation of CK5+ cells seen in response the endocrine therapy. To

address this hypothesis we first needed to identify negative regulators of CK5 expression. As we

had previously found that retinoids were able to suppress CK5 expression in response to P4, we

validated these data. In Chapter IV we show that RA can block both CK5 protein and message.

Additionally, we show that RA blocks P4-induced mammosphere formation, as can the RAR- specific agonist TTNPB. In order to elucidate how RA is able to antagonize PR function, we performed ChIP studies and found that, while RA is unable to affect PR binding to its response element in the CK5 promoter, is does seem to act as a cofactor allowing activation of CK5 expression perhaps by facilitating promoter looping, though additional studies are needed to confirm this hypothesis. In order to more directly test our previously stated hypothesis, in

Chapter V we will show that co-treatment of luminal breast cancer xenografts with the synthetic retinoid fenretinide (Fen) suppresses the accumulation of CK5 cells that occurs with estrogen withdrawal. Finally, to further investigate the genome-wide effects of PR/RAR interaction and antagonism, in Chapter IV we present global RNA-sequencing and ChIP-sequencing experiments that identify important pathways being modulated by these two receptors. Taken together, these findings elucidate an important, novel interaction between PR and RAR. Further, this interaction may highlight a strategy for treatment by which the CSC population in luminal breast cancers may be suppressed and eliminated, thereby reducing breast cancer recurrence.

22 CHAPTER II

MATERIALS AND METHODS

Cell Culture and shRNA

Breast cancer cell lines (T47D, MCF7, BT-474) were obtained from the University of

Colorado Cancer Center Tissue Culture core. Cells were maintained in minimal Eagle’s medium,

5% fetal bovine serum (FBS), 1X NEAA, 1x10-9 M insulin, 0.1 mg/mL sodium pyruvate, and 2 mM L-glutamine. Cell lines were authenticated using short tandem repeat (STR) analysis, and tested negative for mycoplasma using the MycoAlert mycoplasma detection kit (Lonza, Basel,

Switzerland). shRNAs targeting CK5 (TRCN0000425222, TRCN0000433559,

TRCN0000083878) and a non-targeting clone (SHC0002) were purchased from Sigma Mission shRNA library (Functional Genomics Facility, University of Colorado, Boulder, CO, USA).

Cells were transduced with virus containing the shRNAs and stable pools selected with puromycin.

CK5 Promoter Deletion Constructs

Constructs were created from the previously described 6 kb fragment of the human CK5 promoter cloned into a lentiviral vector upstream of firefly luciferase136 using the restriction enzymes listed in Supplemental Table 1. All constructs were analyzed via gel electrophoresis to verify size deletions and sequenced to confirm appropriate deletion. Cells were transduced with virus containing each construct and stable pools selected with puromycin.

Luciferase Reporter Assay and siRNA

T47D cells stably expressing CK5 promoter-luciferase reporter constructs were grown in media containing charcoal stripped serum for 24 hours. Cells were then transfected with

DharmaFECT 1 (T-2001-02, Dharmacon, Lafayette, CO, USA) non-targeting siRNA (D-

0081810-10), or 10 nM siRNA targeting RARα (L-003437-00, Dharmacon), RARγ (L-003439-

23 00, Dharmacon), or both together for 24 hours, followed by treatment with vehicle (EtOH) or P4

(100 nM) for an additional 24 hours. Lysates were harvested and assayed using the Luciferase

Assay System (Promega, Madison, WI, USA).

Chromatin Immunoprecipitation Assay

For ChIP experiments, T47D cells were grown to 70-80% confluency in 15 cm2 dishes in

phenol red free media containing charcoal stripped serum. The next day, cells were pretreated for

30 minutes with 100 nM RA or vehicle then treated with 100 nM P4 or vehicle for one hour.

Immunoprecipitation for PR was performed using antibody PR Ab-8 (recognizes both PR-A and

PR-B, MS-298-P, ThermoFisher, Grand Island, NY, USA), RARα, p300, PolR3a, (ab41934, ab54984, ab96328, Abcam, Cambridge, MA, USA), or CBP (7389, Cell Signaling, Danvers,

MA, USA). Cells were processed using the ChIP-IT Express kit (Active Motif, Carlsbad, CA,

USA) for qPCR analysis and ChIP-IT High Sensitivity kit (Active Motif) for sequencing.

Chromatin was sheared using an S220 Focused Ultrasonicator (Covaris, Woburn, MA, USA).

Library preparation for sequencing experiments was performed using the Illumina TruSeq ChIP

Library Preparation kit (Illumina, San Diego, CA, USA).

Mammosphere Formation Assay and Proliferation Assays

ZsGreen-labeled cells were plated at a density of 100 cells/well in 96-well ultra-low attachment plates in quintuplicate in 100µL MammoCult Media (STEMCELL Technologies,

Vancouver, BC, Canada) containing 1% methylcellulose plus indicated treatments. Cells were grown for two weeks, fed with additional media without methylcellulose containing treatment once per week, imaged by whole well scan, and analyzed for sphere number and size using parameters on the FITC channel using the IncuCyte ZOOM Live Cell Analysis System (Essen

Bioscience, Ann Arbor, MI, USA). For proliferation assays, 5000 cells were plated in 96-well

24 plates and imaged in the IncuCyte. Cells were labeled with constitutively expressed nuclear mCherry and a GFP reporter driven by a 6kb fragment of the CK5 promoter136. Cell number was derived from nuclear red object count.

Immunoblotting

Whole cell lysates were collected in RIPA buffer following indicated treatments. 100µg of lysate were denatured, separated on SDS-PAGE gels, and transferred to polyvinylidene fluoride membranes. After blocking in 5% BSA in TBST, membranes were incubated with primary antibodies suspended in 3% BSA in TBST overnight at 4 °C. Proteins were immunoblotted with primary antibodies (listed in Table 2.1) followed by IRDye 800CW Goat-

Anti-Mouse IgG (926-32210, Li-Cor Biosciences, Lincoln, NE, USA) and IRDye 680LT Goat-

Anti-Rabbit IgG (926-68021, LiCor Biosciences). The Odyssey Infrared Imaging System (Li-

Cor Biosciences) was used to image immunoblots, and Image Studio Lite (Li-Cor) was used for blot analysis.

Immunohistochemistry, Immunocytochemistry, and Proximity Ligation Assays

IHC and ICC were performed essentially as previously described137, 138. For IHC, tumors were removed from animals and fixed in 10% buffered formalin. Tissue was processed, paraffin embedded, and cut into 5 µm sections. Antigen retrieval was performed in high-temperature citrate buffer followed by blocking with 2.5% normal horse serum. Primary antibodies were applied (listed in Table 2.1), followed by secondary antibodies, and developed using ImmPRESS

Peroxidase detection kit (Vector Laboratories, Burlingame, CA, USA). Slides were scanned into the Aperio digital pathology system (Leica Biosystems) and whole sections analyzed for percent of positive cells using an algorithm tuned for cytoplasmic CK5 staining, limiting bias and therefore blinding was not necessary. For ICC, cells were plated on cover slips and treated as

25 indicated for 24 hours. Cells were then fixed with methanol/acetone for 5 minutes on ice or

paraformaldehyde for 15 minutes on ice. Cells were then incubated with secondary fluorescent

antibodies (A11029, A11037, Invitrogen, Grand Island, NY, USA). Proximity ligation assay

(Duolink, Sigma-Aldrich, St. Louis, MO) was performed per the manufacturer’s protocol.

Primary antibodies (listed in Table 2.1) were incubated at room temp for one hour.

Quantitative Reverse Transcription PCR and RNA-sequencing

For qRT-PCR, total RNA was harvested using QIAzol lysis reagent (Qiagen, Venlo,

Netherlands), and qRT-PCR was performed using Absolute Blue Sybr Green (Thermo-Fisher). cDNA was made using the Verso cDNA synthesis kit (ThermoFisher). For RNA-seq, total RNA was harvested using the RNeasy Mini kit (Quiagen). Analysis was performed using the Pfaffl method for qPCR and ΔΔCT for ChIP 139. Primers targeting mRNA and genomic DNA are in

Table 2.2. 2µg of RNA was prepared for sequencing using the Illumina TruSeq Stranded mRNA

Library Prep Kit (Illumina, San Diego, CA). Resulting libraries were sequenced on an Illumina

HiSeq 4000 (1 × 150 bp). After demultiplexing, the reads were trimmed with cutadapt140 to

remove 3′ adaptor sequences and low quality 3′ bases (q < 10). The trimmed reads were then

aligned to the human (GRCh38) genome using hisat2141. Aligned reads were assigned to features

using Rsubread’s featureCounts142. Differential gene expression testing was performed using

limma143 and voom,144 and visualized using degust (https://github.com/drpowell/degust).

Limiting Dilution Analysis and Tumor Growth

Experiments involving animals were performed under an approved University of

Colorado Institutional Animal Care and Use Committee protocol. For limiting dilution analysis,

T47D cells harboring the previously described CK5 promoter driven GFP reporter were treated with 100 nM P4 for 24, GFP+ and GFP− cells collected by FACS, serially diluted into Cultrex

26 (Trevigen, Gaithersburg, MD, USA) and an equal number of GFP+ and GFP− cells as indicated were injected bilaterally into opposing fourth mammary fat pads of 8 week old female nu/nu

mice (Jackson labs, Bar Harbor, ME). All animals were supplemented with silastic pellets

containing 17β-estradiol (1 mg). Tumors were palpated twice weekly for six weeks by a blinded

researcher. Results were analyzed using extreme limiting dilution analysis (ELDA) software 145.

For treatment experiments, T47D xenografts were developed by injecting 1x106 T47D cells in

Cultrex (Trevigen) bilaterally into the fourth mammary fat pads of 8-week-old female

NOD/SCID mice (Jackson labs) supplemented with silastic pellets containing 17β-estradiol (E2,

1 mg). When tumors reached an average of 75 mm2, mice were stratified into six treatment

groups with equal average tumor volume (n=5 animals each). Treatments were as follows:

continued on E2, E2 plus fenretinide (Fen), estrogen withdrawn (EWD), EWD plus Fen, EWD

plus fulvestrant (ICI) plus one dose of Fen, and EWD plus ICI plus continued Fen treatment. Fen

(2.5 mg/mouse) or peanut oil vehicle was administered subcutaneously 2x per week. ICI (5

mg/mouse) or peanut oil vehicle was administered 1x per week. EWD was performed by surgical

removal of the silastic E2 pellet. Tumors were measured 2x per week and volumes estimated by

the formula l(w2)/2. After three weeks of treatment, mice were sacrificed, and tumors collected and analyzed by IHC. Sample number was calculated at 80% power and α = 0.05.

Statistical Methods

Data are represented ± s.e.m. unless otherwise noted and analyzed using a two-tailed

Student’s t-test or one-way ANOVA followed by either a Tukey or Dunnett multiple comparison post-hoc test as indicated. Prism 6.0-8.0 (GraphPad Software, La Jolla, CA, USA) was used for statistical analyses when samples met variance and normality tests. P<0.05 were considered significant.

27 Table 2.1 Antibodies Antibody Company Catalog # Assay α-tubulin Sigma ST1568 Immunoblot CBP Cell Signaling 7389 ChIP CK5 Leica NCL-L-CK5 Immunoblot, IHC, ICC IgG Abcam ab10485 ChIP p300 Abcam ab54984 ChIP PolR3a Abcam ab96328 ChIP PR Dako PgR 1294 Immunoblot, IHC, ICC, PLA PR Ab8 ThermoFisher MS-298-P ChIP PR F4 Santa Cruz sc166169 ChIP PR 6A Active Motif 61023 ChIP RARα Abcam ab54984 ChIP RARα Abcam sc551 Immunoblot RARα Abcam ab28767 ICC, PLA

Table 2.2 qPCR Primers (5’ to 3’) mRNA Primers Forward Reverse KRT5 GGAGAAGGAGTTGGACCAGTCAAC CTACCTCCGGCAAGACCTCCAC ACTB GATCATTGCTCCTCCTGAGC ACTCCTGCTTGCTGATCCAC

Genomic Primers Forward Reverse KRT5 -1.1kb (set 1) GAGTGGGTGTGGTTTAGAACAG GTCTATGGATTGTCCTGCCAG KRT5 -1.1kb (set 2) CTGGCAGGACAATCCATAGAC CCAGCAAGCTCTATTCCACTAG KRT5 -130bp CCAAGAGATCAGTGCTGCAAGG GTTACCCAGGAACGGTGATGC KRT5 gene negative GCCGTTGTTCGAGCAGTACA CTCCACCCCAACTCACTTGTT KRT5 promoter negative CTGAGGCTAGGGCTTTGTGAA GCACAAAAAGTGGGGAGCAAT tRNA Primers Forward Reverse Arg-TCT-1-1 GTGGCGCAATGGATAGC CTTCCTTTGAATGCCTTCAGC Ile-TAT-1-1 GTGGCGCAATCGGTTAGC GCATTGCTCCGCTCGC Met-CAT-1-1 GCCTCGTTAGCGCAGTAG CTCACGACCTTCAGATTATG

28 CHAPTER III

CYTOKERATIN 5 IS DIRECTLY REGULATED BY PROGESTERONE AND IS

IMPORTANT IN MEDIATING PROGESTERONE-INDUCED CANCER STEM CELL

PROPERTIES*

Introduction

Greater than 70% of all breast cancers express estrogen receptor alpha (ER) at diagnosis and display various degrees of dependency on estrogens for proliferation146. While ER targeted endocrine therapies have greatly improved survival for patients with ER+ disease, intrinsic or acquired resistance still accounts for half of all breast cancer deaths147. Furthermore, recurrences can occur after an extended remission (>5 years), suggesting cell populations in ER+ tumors can survive initial treatment and a prolonged dormancy148, 149. One possible explanation for this recurrence is the cancer stem cell (CSC) theory, which posits that tumors contain a small population of cells that exhibit characteristics of the normal stem cell population including drug resistance, quiescence, and replicative immortality, allowing tumors to reform150. It is important to note that breast cancer cells can acquire a CSC phenotype through signaling or therapeutic pressure, therefore prevention of the CSC phenotype may be equally as important as targeting existing CSCs and paramount to developing new treatment strategies151, 152.

Progesterone receptors (PR) are co-expressed in the majority of ER+ breast cancers and are associated with positive response to endocrine therapy153. The role of PR itself is complex; it can exert autonomous proliferative signals or oppose the mitogenic effects of estrogens in a context-dependent manner127, 129, 154, 155. In particular, we and others have shown that progesterone (P4) increases a population of ER−, cytokeratin 5 (CK5)+ breast cancer cells55, 156.

* Portions of this chapter were originally published in Fettig et al.37 and are republished with permission.

29 In the normal human breast, CK5 is expressed in the committed stem cell population, as well as

the ER− luminal progenitor cell populations that give rise to ER+PR+ luminal cells35,157. CK5+ compared with CK5− breast cancer cells have enhanced mammosphere and tumor forming potential, and are chemo- and endocrine therapy resistant36, 136, 158. Furthermore, breast cancer cell lines with larger P4-dependent CK5+ populations following suppression of microRNAs

(miR)29 and miR141 had increased tumor initiating ability138, 159. While the use of CK5 as a marker of this population of cells with CSC properties is important, identifying if CK5 has a role to play in actively facilitating these properties is also necessary.

Results

Progesterone receptor regulates CK5 transcription through direct binding to the proximal

promoter

While it was established that P4 increases expression of the CK5 protein and regulates an

exogenous fragment of the CK5 promoter, whether this is a direct or indirect regulation needed

further investigation136. To narrow down the region(s) of the CK5 promoter required for P4

responsiveness, we constructed a series of 5’ and internal deletion mutants using a 6 kb region of

the human proximal CK5 promoter, previously identified as sufficient for expression of CK5 in

keratinocytes, cloned upstream of luciferase in a lentiviral vector (Figure 3.1A-C)160. Constructs

were stably integrated into T47D cells. Thee was some variability in basal reporter expression,

however most were expressed at a similar level (Figure 3.1B). P4 increased luciferase reporter

expression by about 6-fold, however P4-induced luciferase activity was significantly decreased

in four constructs with deletions incorporating a region 1.1 kb upstream of the transcriptional

start site (TSS) (Figure 3.1C). This region contains a sequence (GGAACAgggTGGTTC, -1098

bp from the TSS) with extensive identity to an optimal progesterone response element (PRE)

30

B A 800000 ***

600000

400000 *** ** RLU **

200000

** 0 6 4.6 3.9 3.4 2 1.3 1 0.2 Δ0.2 Δ1 Δ3.4 Prom. Neg. -1.2 -4.3 -4.6

C 6kb + 6kb 6kb + 4.6kb 9kb + 3.9kb

4kb + 3.4kb ** 2kb + 2kb = PRE majority conc. seq. 3kb + 1.3kb = PRE half site 1kb + 1kb **

2kb + 0.1kb *** 0.2- ** 1.0- + ** 3.4- + ** 0 2 4 6 8 Fold Change Luciferase (RLU) 6kb 5kb 4kb 3kb 2kb 1kb TSS

E 1.6 *** 0.11 *** 1.4 0.10 0.09 1.2 D 0.08 1.0 0.0004 % Input % Input 0.08 0.0003 0.0002 Primer Set 2 0.04 Primer Set 1 0.0001 0.00 0.0000 Veh P4 Veh P4 Veh P4 Veh P4 IP: PR IP: IgG IP: PR IP: IgG Primer set 1 Primer set 2

Figure 3.1 PR is recruited to the proximal CK5 promoter near a PRE that is necessary for P4 transcriptional activation.

31 Figure 3.1 PR is recruited to the proximal CK5 promoter near a PRE that is necessary for P4 transcriptional activation. CK5 reporter constructs. (A) As previously described, a 6 kb region of the CK5 promoter was cloned in front of both a GFP and luciferase reporter. Adapted from Axlund et al.136 (B) Basal expression of CK5 reporter deletion constructs. Deletion constructs were engineered from the 6 kb fragment of the CK5 promoter cloned upstream of luciferase in a lentiviral vector using existing restriction sites. Constructs were stably integrated via lentiviral transduction into T47D cells. Cell lysates were collected without treatment and analyzed using the Luciferase Assay Kit (Promega). Basal reporter expression of each cell line is shown as relative light units (RLU). Data represent mean ± s.e.m. **P< 0.01, ***P< 0.001 (ANOVA followed by Dunnett’s compared to full 6 kb construct.) (C) P4 action requires a region containing a putative PRE 1098 bp upstream of the TSS. The left side of the graph indicates relative size of the promoter construct with 5’ or internal deletions, and the location of putative PRE half-sites, or full majority consensus sequence based on that reported by Lieberman et al 33 and Graham et al 34. Constructs were transduced into T47D cells and stable puromycin resistant pools selected. Cells were then seeded at 5000 cells per well in 96 well plates and treated with vehicle or 100 nM P4 for 24 hours. Lysates were collected and luciferase activity analyzed using the Luciferase Assay Kit (Promega). Relative fold changes are indicated for each construct over vehicle control. Data represent mean ± s.e.m. **P<0.01, ***P<0.001 ANOVA followed by Tukey post-hoc comparing to the full length 6 kb promoter construct. (D) Diagram showing location of primer sets used in E. (E) T47D cells were treated with vehicle or 100 nM P4, fixed with formalin, and chromatin harvested. ChIP using antibodies against PR or IgG was performed. Data represent mean (percent input) ± s.e.m. *P< 0.05, **P< 0.01, ***P< 0.001 (Student’s t-test within each group, IgG or PR).

32 identified by both mutagenesis studies and genome-wide analysis of PR binding in T47D cells 89,

90. To confirm that PR was binding to this PRE, we performed ChIP for PR followed by qPCR

using two independent sets of primers surrounding this region (Figure 3.1D). Indeed, there was

significant enrichment of PR recruited to the -1.1 kb PRE region in P4 compared to vehicle

treated T47D (Figure 3.1E). These data indicate that PR regulation of CK5 expression is via

direct transcriptional regulation.

Progesterone-expanded CK5+ breast cancer cells are more tumorigenic

Progesterone increases the CD44+ population in addition to the CK5+ population in

luminal breast cancer cell lines, and we have previously shown that these CD44+ cells are enriched in CK5 expression and are more tumor-initiating (Figure 3.2A)156. To more directly measure CK5 involvement in tumorigenicity, we used a system in which T47D breast cancer cells are integrated with a 6kb region of the CK5 promoter regulating a GFP reporter (Figure

3.1A)136. Cells were treated for 24 hours with P4 in vitro to induce a CK5+ cell population, then

CK5+ and CK5− cells were isolated by FACS using GFP as a surrogate (Figure 3.2B-D). Female nude mice supplemented with estrogen slow release pellets were bilaterally injected with sorted

CK5+ and CK5− cells subcutaneously in opposing fourth mammary fat pads at dilutions ranging from 102 to 105. Tumors were palpated through 6 weeks post-injection (Figure 3.2E). Extreme limiting dilution analysis (ELDA) revealed that CK5+ cells initiated tumors more efficiently than

CK5− cells (Table 3.1). These data provide direct confirmation that P4-induced CK5+ breast cancer cells have enhanced tumor initiation ability.

Progesterone-enhanced mammosphere formation of breast cancer cells requires CK5

P4 increases the mammosphere forming potential of breast cancer cells138, 161. To test

whether CK5 is necessary for the P4 effect on mammospheres, we utilized shRNA inhibition

33 A B % CD44+ / CD24- Cells % CD44+ Cells

* (%) 0.20 15 *

0.15 10 0.10 5 0.05 Time: 0 hours Time: 24 hours 0.00 0

Percent Sorted Population (%) Veh P4 Veh P4 E 6 weeks

C D 102 % GFP/CK5+ Cells

15 ** 102 10 CK5 − CK5+ GFP- GFP+ 5 104 0

Percent Sorted Population (%) Veh P4

104

CK5 CK5+ − Figure 3.2 CK5+ cells are more tumor initiating. (A) P4 treatment expands CSC populations. T47D cells treated with vehicle or 100 nM P4 for 24 hours were FACS sorted by expression of cell surface markers CD44 and CD24. Data represent mean ± s.e.m. *P < 0.05 (B) T47D breast cancer cells stably integrated with this CK5 promoter- driven GFP reporter and a constitutive nuclear mCherry vector were treated with P4 for 24 hours to induce CK5 reporter expression. (C-E) T47D cells expressing the CK5 promoter driven GFP reporter were treated with 100 nM P4 for 24 hours to induce expression, then CK5+ and CK5- cell populations were sorted based on GFP expression via FACS (C and D) and injected into the fourth mammary fat pads of nude mice (E). Tumor growth was assessed by palpation for 6 weeks. Data represent mean ± s.e.m. **P < 0.01.

34 Table 3.1 Tumor-initiating capacity of P4-induced CK5+ compared to CK5- T47D breast cancer cells. breast cancer cells. Number of cells injected per mammary fat Number of tumors per number of pad injected fat pads Week 4 after implantation CK5− CK5+ 1 X 104 6/10 10/10 1 X 103 3/10 8/10 1 X 102 1/10 6/10 Tumor-initiating frequency (95% CI) (1/7275) (1/360) Tumor-initiating range (95% CI) (1/14,665-1/3609) (1/735-1/176) P-value 1.32 X 10-10

Week 6 after implantation CK5− CK5+ 1 X 104 7/10 10/10 1 X 103 6/9a 7/9a 1 X 102 5/10 9/10 Tumor-initiating frequency (95% CI) (1/3071) (1/263) Tumor-initiating range (95% CI) (1/6425 - 1/1468) (1/581 - 1/119) P-value 6.05 X 10-9 Abbreviation: CI, confidence interval. Limiting dilution analysis of CK5+ vs. CK5− T47D cells. T47D cells stably harboring a CK5 promoter-GFP reporter were treated for 24 h with 100 nM P4 to induce a population of CK5+ cells for analysis and sorted by FACS for GFP+ and GFP− cells. aOne animal was euthanized for health reasons prior to week 6, and was excluded from the analysis. Data were analyzed using ELDA software 57.

35 coupled with an automated in vitro mammosphere assay. T47D cells constitutively expressing

Zs-green were transduced with lentiviral-packaged shRNAs to create cell lines with impaired P4 induction of CK5 (Figure 3.3A). Lines transduced with two independent shRNAs targeting CK5 or a non-targeting control (shCont) were seeded into mammosphere media with vehicle or P4; sphere number and size were assessed after two weeks (Figure 3.3C-E). While there were slight differences in adherent 2D growth with CK knockdown (one just reached significance, the other was not significantly different), there was no statistical difference in the basal size or absolute number of detectable spheres between treatment groups within each stable cell line (Figure 3.3B-

E). However, P4 increased the average mammosphere size by 1.6 fold in shCont cells; this increase was attenuated in the CK5 deficient cells (Figure 3.3E). These data imply that CK5 is necessary for P4 to produce larger mammospheres.

Discussion

In this chapter, we sought to determine both the mechanism of PR regulation over CK5 expression as well as the consequences of CK5 expression. Here we show that a region 1.1 kb upstream of the CK5 TSS is important in PR/P4 regulation of the gene and that, in fact, PR is recruited near this site with P4 stimulation. Additionally, we showed the first direct evidence that

P4 expanded CK5+ cells are more tumorigenic than their CK5- counterparts in vivo. This effect was also seen in vitro, whereby P4 treated mammospheres were larger than those treated with vehicle. Finally, we also showed that this increase in mammosphere potential was reliant on the expression of CK5, as knockdown of CK5 using stably expressed shRNA attenuated the ability of P4 to induce an increase in mammosphere size. These results were confirmed in another set of experiments performed by another graduate student, Olivia McGinn, using CK5 knockout cells developed using CRISPR technology (data not shown). Interestingly, Ms. McGinn also showed

36

B A 6 shCont shCK5 22 shCK5 78 shCont shCK5 22 Veh P4 Veh P4 P4 shCK5 78 4 CK5 * 2 α-tubulin 0

Confluence, fold change over time 0 0 50 100 Time Elasped (hours)

D C 200 shCont shCK5 22 shCK5 78 150

100

Veh 50

Mammosphere Number Mammosphere 0 1mm Veh P4 Veh P4 Veh P4 shCont shCK5 22 shCK5 78 E )

2 ** 8000 m µ P4 6000

4000

2000

Mammosphere Size ( Size Mammosphere 0 Veh P4 Veh P4 Veh P4 shCont shCK5 22 shCK5 78

Figure 3.3 The P4 dependent increase in mammosphere size requires expression of CK5. (A) T47D cells with constitutive ZsGreen expression were stably transduced with either a non- targeting shRNA (shCont) or one of two shRNAs targeting CK5 (shCK5). CK5 expression in response to 24 hour treatment with 100 nM P4 was analyzed by immunoblot, using α-tubulin as a loading control. Bar marks extra lanes removed from the same blot. (B) T47D shCont and shCK5 cells were plated at a density of 5000 cells per well in 96-well plates in full media, then imaged for 100 hours in the IncuCyte Zoom live cell imaging system. Cell confluence was graphed over time, normalized to density at time 0. Data represent mean ± s.e.m. *P < 0.05. (C) T47D-ZsGreen shCont and shCK5 cells were plated in Mammocult media at a density of 100 cells per well in quintuplicate in a 96-well plate and treated with either vehicle (EtOH) or 100 nM P4. After two weeks, mammospheres were imaged and analyzed using the IncuCyte Zoom live cell analysis system and software. Representative images of wells are shown. (D and E) Mammosphere number (D) and size (E) depicted for shCont and shCK5 cells treated with vehicle or P4. Data represent mean ± s.e.m. **P < 0.01.

37 that overexpression of CK5 in T47D cells leads to an increase in the basal number of

mammospheres (data not shown).

The use of mammospheres as a surrogate for CSC properties has been shown to faithfully

recapitulate repopulation experiments typically performed in vivo and test self-renewal

capacity162. Additionally these experiments have been useful as the cost and time to perform

these experiments is substantially less than traditional in vivo experiments. However, these experiments also have drawbacks. One factor to consider when analyzing these data is the interpretation of changes in sphere size and number. We propose that, in the conditions used in our experiments, the change in sphere size seen in response to P4 treatment is due to the expansion and maintained support of the minor, self-renewing CK5+ cell population in response to progesterone. This led to larger spheres and is it this relatively smaller number of larger spheres formed by the CK5+ cells that accounts for the increase in average sphere size. While it would be ideal to confirm this by plotting histograms of the individual sphere sizes from each well for each treatment group, we are unfortunately unable to download the data for each sphere from the Incuyte software (personal correspondence with Inctucyte technical support). As for the increase in sphere number seen in response to forced CK5 overexpression, we interpret this to be due to the increase in the number of cells expressing CK5 without the need for P4 stimulation.

Interestingly, Ms. McGinn also found that addition of P4 to the CK5 overexpressing cells leads to an increase in sphere size as seen in non-CK5 overexpressing cells, indicating that, while CK5 is directly contributing to the increase in mammosphere potential, P4 may be exerting additional effects. An additional tool to assess regenerative capacity is to perform secondary or even tertiary mammosphere experiments. While these experiments were attempted, they were unsuccessful due to the low yield of spheres in the format used (100 cells are plated in

38 quintuplicate in 96-well plates). Repeating these experiments using a larger format plate and

higher cell number in the initial sphere assay will allow for successful secondary formation

assays.

This series of experiments highlights the fact that CK5 is more than just a marker of the

CSC population. CK5 is itself an important mediator of P4-induced CSC properties, indicating that it or its downstream effectors may serve as potential therapeutic targets to suppress tumor recurrence.

39 CHAPTER IV

RETINOIC ACID BLOCKS EFFECTS OF PROGESTERONE ON LUMINAL BREAST

CANCER CELLS†

Introduction

Identifying a way to target the cancer stem cell (CSC) population in ER+ breast cancer may assist in developing treatments to prevent recurrence. In an attempt to identify compounds that suppress the progesterone (P4) expanded CK5+ population, we previously performed a small molecule screen using T47D cells transduced with the CK5 promoter-GFP reporter construct described earlier. With this screen, we discovered that retinoids, including all-trans retinoic acid (ATRA) and two synthetic retinoids, prevent the ability of P4 to produce CK5+ breast cancer cells163. Retinoids (e.g., ATRA, 9-cis RA, 13-cis RA) are ligands for nuclear receptors in the retinoid receptor subclass, which includes three retinoic acid receptors (RARα, -

β, and -γ) and three retinoid X receptors (RXRα, -β, and -γ). The canonical action of these receptors is to form RAR/RXR heterodimers that can occupy DNA in the absence of ligand and often repress transcription. Upon ligand binding they modulate gene transcription to regulate important cellular processes such as differentiation and cell death. However recent studies indicate that this may not be their only mechanism of transcriptional regulation110, 164, 165. In fact,

RARα has recently been identified as an important cofactor for ER and GR mediated transcription110, 166.

In the clinic, ATRA is an approved treatment for acute promyelocytic leukemia, which contains a PML-RAR fusion, as a differentiating agent 167. Additionally, retinoids have shown great efficacy in preclinical breast cancer models, including preventing carcinogen-induced

† Portions of this chapter were originally published in Fettig et al.37 and are republished with permission.

40 mammary tumorigenesis in rodent models and potently blocking growth of cancer cell lines168,

169. Treatment studies in breast cancer patients, however, have been mostly disappointing, with use of retinoids in combination treatment with Tam or chemotherapy failing to achieve study end points (reviewed in Garattini et al.170). One exception is the synthetic retinoid fenretinide (Fen), which has had some efficacy in prevention of recurrent premenopausal breast cancer171.

Understanding the interplay between RARs and steroid receptors is important in determining contexts under which they could be therapeutically useful in ER+ breast cancer.

Results

Retinoids block progestin-induced CK5+ expression and mammosphere growth

To validate our previous work that identified retinoids as potent inhibitors of P4-

dependent induction of CK5163, 172, we measured proliferation and CK5 reporter expression in

response to a gradient of 9-cis retinoic acid (RA) concentrations (1 nM – 1 µM) plus or minus

100 nM P4 (Figure 4.1A and B). P4 slightly inhibited proliferation but led to a drastic increase in

CK5 reporter expression, while RA further reduced proliferation and blocked the P4-induced increase in CK5 reporter expression in a concentration-dependent manner. Consistent with these findings, the synthetic progestin, medroxyprogesterone acetate (MPA), also slightly inhibited growth, increased CK5 reporter expression, and had its effects attenuated by RA (Figure 4.2A and B). To investigate endogenous CK5, we measured CK5 mRNA levels in response to vehicle,

P4, RA, or P4 plus RA in three ER+ breast cancer cell lines (T47D, MCF7, BT-474). MCF7 and

BT-474 cells were pre-treated with 17β-estradiol (E2) for 48 hours to induce PR levels. P4 increased CK5 mRNA levels 90-fold in T47D, 2-fold in MCF7 and 1120-fold in BT-474 while

RA alone had no effect on CK5 mRNA levels in two of the three cell lines. Nonetheless, RA imposed a significant attenuation of the P4-medated increase in CK5 transcript in all three cell

41 A B 6000 40000 Veh Veh P4 P4 P4 + RA 1nM P4 + RA 1nM P4 + RA 10 nM P4 + RA 10 nM P4 + RA 100 nM P4 + RA 100 nM P4 + RA 1000 nM 30000 P4 + RA 1000 nM 4000

20000

2000

10000 Cell number (red object count) CK5+ cell count (green object count)

0 0 0 20 40 60 80 100 0 20 40 60 80 100 Time (hours) Time (hours)

Figure 4.1 T47D proliferation and CK5 reporter respond to RA in a dose-dependent manner. 9-cis RA dose curves. To determine the lowest effective concentraions of 9-cis RA, T47D cells expressing constitutively expressed nuclear mCherry and CK5 promoter GFP were treated with the indicated increasing concentrations of RA with or without P4 for 24 hours. Cell number was assessed using red object count (A) and CK5 was assessed with GFP expression (B).

42 A B 400 2.5 Veh * P4 RA P+R 300 MPA 2.0 * M+R

200 1.5 Veh P4 RA 100 P+R 1.0 MPA M+R CK5+ cell count (green object count) 0

Cell number (red object count fold change) 0 20 40 60 0 20 40 60 Time elasped (h) Time elasped (h)

C D T47D MCF7 BT-474 ion ion

ion *** ** 2.5 *** 1500 *** * *** 2.0 100 *** 1000 1.5 * *** 500 50 1.0 0.5 5 0 0.0 0 VehicleVeh P4 P4 RA P4+RAP4+ VehVeh P4 RA P4+RAP4+ VehVeh P4 RA P4+RAP4+ Relative CK5 mRNA expression expression mRNA CK5 Relative RA RA RA +E2 +E2

E H ) 2 *** m 3000 µ ***

2000 **

1000

0 MammosphereSize ( Veh-1 P4 -1 RA -1 P4+RA -1

F Vehicle P4 RA P4 + RA

0.0% 5.2% 0.1% 1.2% G

Figure 4.2 RA blocks P4-mediated CK5 expression and P4 induction of large mammospheres.

43 Figure 4.2 RA blocks P4-mediated CK5 expression and P4 induction of large mammospheres. (A and B) T47D cells stably integrated with the CK5 promoter driven GFP reporter and constitutive nuclear mCherry were plated at a density of 5000 cells per well in a 96 well plate in hormone-depleted serum. Cells were treated at time 0 with vehicle (EtOH), 100nM P4, 100 nM 9-cis retinoic acid (RA), P4 plus RA, 100 nM MPA, or MPA plus RA for indicated time. Cell number was analyzed by red object count (A) and CK5 reporter levels indicated with GFP (B). (C) Treatment with RA blocks P4-induced CK5 transcription. T47D cells in C were treated with ethanol vehicle (Veh), 100 nM P4, 100 nM RA, or P4 plus RA for 10 hours. MCF7 and BT-474 cells were pre-treated with 10 nM E2 for 48 hours to induce PR expression, then treated in the same manner as T47D cells. Quantitative reverse-transcriptase PCR (qRT-PCR) was used to assess relative CK5 mRNA levels normalized to β-actin. Results are displayed as fold expression over vehicle control. Data represent mean ± s.e.m. *P< 0.05 **P< 0.01 ***P<0.001. (D and E) Immunoblot of CK5, PR, and RARα, in T47D cells treated with vehicle (EtOH), 100nM P4, RA (100nM in D or at indicated concentrations in E), or the combination. α-tubulin was used as a loading control. (F) CK5+ cells were measured in T47D cells via immunocytochemistry after 24 hours of the same treatments as in C. Fluorescent staining shows CK5 (red) and DAPI (blue). Percent CK5+ cells per field are indicated, calculated from five fields taken at 10x magnification. (G) Merged images of dual ICC for PR (red) and RARα (green) in T47D cells treated as in A for 24 hours. Nuclei are counterstained with DAPI (blue). (H) T47D-ZsGreen cells were plated at a density of 100 cells per well in quintuplicate in mammosphere media in 96-well plates plus indicated treatments. After two weeks, mammosphere size was analyzed via scanning on the IncuCyte Zoom. Data represent mean ± s.e.m. **P< 0.01 ***P<0.001.

44 lines (Figure 4.2C). CK5 protein expression was directly assessed in T47D cells by immunoblot

and immunocytochemistry (ICC). By immunoblot, the P4-mediated increase in CK5 protein was

blocked by RA (Figure 4.2D). By performing a dose response to RA plus or minus P4, we

determined via immunoblot that 100 nM was the most effective lowest concentration, and chose

that dose for remaining studies (Figure 4.2E). Similarly, ICC showed that P4 increased the

population of CK5+ cells to 5.2% compared to vehicle (0%), whereas co-treatment with RA

blocked this increase (Figure 4.2F). Thus, RA effectively blocks the progestin-mediated increase

in CK5 expression primarily through a reduction in mRNA transcripts.

We additionally assessed PR and RARα expression and co-localization in T47D cells under four different hormone conditions. By immunoblot PR (both PR-A and PR-B isoforms) protein decreased with RA treatment (0.76 fold versus vehicle), but to a lesser extent than with

P4-mediated downregulation (0.08 fold versus vehicle) (Figure 4.2D and E). RARα was present in T47D cells and its levels were unaffected by any treatments (Figure 4.2D). By dual ICC, PR and RARα were frequently co-localized in the nucleus of T47D cells, with occasional solely PR+ and RARα+, or double negative cells (Figure 4.2G). No differences in co-localization were observed under the different hormone conditions. Thus, PR levels decrease slightly with RA treatment but RARα expression was not affected by P4 treatment.

Since CK5 was required for the P4-mediated increase in mammosphere size, we next assessed if RA inhibition would also impede this process. T47D cells treated with vehicle, P4,

RA, or P4 plus RA were assessed for mammosphere formation (Figure 4.2H). Indeed, RA reduced baseline mammosphere size (0.4-fold) and also prevented the P4-mediated increase (1.7- fold, reduced to 0.9 fold) in mammosphere size; this was also determined to be dose dependent

(Figure 4.3A and B). This attenuation of P4-induced sphere size was also seen in response to

45 A B ) 150 ** 2 40000

m *** µ *** 30000 100 20000 50 10000 Mammosphere Number Mammosphere

0 ( Size Mammosphere 0 P4 P4 Veh Veh RA 1 RA 1 RA 10 RA 10 RA 100 RA 100 RA 1000P + RA 1 RA 1000P + RA 1 P + RA 10 P + RA 10 P + RA 100 P + RA 100 C P + RA 1000 D P + RA 1000 )

2 *** 150 40000 ** m *** µ 30000 100 20000 50 10000 Mammosphere Number Mammosphere 0 ( Size Mammosphere 0 P4 P4 Veh Veh ATRA 1 ATRA 1 ATRA 10 ATRA 10 ATRA 100 ATRA 100 ATRAP 1000+ ATRA 1 ATRAP 1000+ ATRA 1 P + ATRA 10 P + ATRA 10 P + ATRA 100 P + ATRA 100 P + ATRA 1000 P + ATRA 1000 E F

) *** 2 15000 150 m µ

100 10000

50 5000 Mammosphere Number Mammosphere 0 ( Size Mammosphere 0

VehMPA VehMPA M + RA 1 M + RA 1 M + RA 10 M + RA 10 M + RA 100M + ATRA 1 M + RA 100M + ATRA 1 M + RA 1000M + ATRA 10 M + RA 1000M + ATRA 10 M + ATRA 100 M + ATRA 100 M + ATRA 1000 M + ATRA 1000 Figure 4.3 Mammosphere formation is inhibited by retinoids in a concentration-dependent manner. T47D cells were plated at a density of 100 cells per well into ultra low attachment 96-well plates in Mammocult media and treated with vehicle, or increasing concentrations (1nM – 1µM) of 9- cis RA (A, B, E and F) or ATRA (C-F), with or without 100 nM P4 or 100 nM MPA. Spheres were allowed to grow for two weeks and then size and number analyzed on the InucCyte live cell imaging system. Data represent mean ± s.e.m. **P< 0.01 ***P<0.001.

46 A siNT siRARα siNT siRARγ P4: - + - + - + - + RARα RARγ

α-tubulin α-tubulin

RAR/α-tubulin: 1.0 1.0 0.1 0.1 1.0 0.9 0.3 0.2 (normalized to siNT)

B ns 4

3 *** * **

2

1

Fold Change Over Vehicle 0 P4 P4 P4 P4 RA RA RA RA Veh Veh Veh Veh P4+RA P4+RA P4+RA P4+RA siNT siRARα siRARγ siRARα + siRARγ

C

Figure 4.4 RA inhibition of P4-dependent expression of CK5 requires RARs. (A) T47D cells stably expressing a CK5 promoter-driven luciferase reporter were transfected with non-targeting siRNA (siNT), or siRNA to RARα or RARγ for 24 hours, then treated with either ethanol vehicle or 100 nM P4. Lysates were collected 24 hours post-treatment and analyzed by immunoblot. Relative RARα or RARγ levels are normalized to α-tubulin loading control and indicated relative to the vehicle treated siNT. (B) T47D cells were transfected as above and treated with vehicle, 100 nM P4, 100 nM 9-cis RA, or both P4 plus RA for 24 hours. Lysates were collected and luciferase assays performed. Luciferase was graphed as fold change over vehicle control for each group of siRNAs. Data represent mean ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.001. (C) T47D cells were treated with vehicle, 100 nM P4, 100 nM RA, 2 nM RARα specific agonist AM80, 2 nM RARγ specific agonist CD1520, or retinoid plus P4. CK5, RARα, and RARγ expression were measured by immunoblot with α-tubulin as a loading control.

47 ATRA (Figure 4.3C and D). Interestingly, MPA had no effect on mammosphere size or number

(Figure 4.3E and F). Therefore, RA antagonizes the effects of P4 on transcription and cell

phenotype.

Retinoic acid receptors are necessary for retinoid antagonism of progesterone-dependent CK5

expression and mammosphere formation

9-cis RA can bind and activate RXRs, which have several heterodimeric binding partners, in addition to RARs, therefore we next wanted to determine if its repressive action on CK5 expression was specifically occurring through RARs68. Luminal breast cancer cells primarily express RARα and RARγ173. Therefore, we used scrambled siRNA (siNT) or siRNA targeting

RARα (siRARα), RARγ (siRARγ), alone or simultaneously, in T47D cells stably expressing a

CK5 promoter-driven luciferase reporter (Figure 4.4A). Similar to non-targeting siRNA, when cells were transfected with either siRARα or siRARγ alone, RA still inhibited P4 induction of the

CK5 reporter (Figure 4.4B). When both RARα and RARγ were depleted, however, RA was

unable to attenuate P4 activation of the CK5 promoter, indicating that RA can act through either

RAR isoform to block CK5 induction. Interestingly, looking at endogenous protein, the RARα

specific agonist, AM80, was able to block P4-induced CK5 expression, however the RARγ

specific agonist CD1530 had no effect (Figure 4.4C).

We further investigated this specificity by using the RAR specific agonist, arotinoid acid

(TTNPB). In T47D cells, TTNPB blocked P4-induction of the CK5 promoter-GFP reporter and endogenous protein as efficiently as RA at doses as low as 1nM (Figure 4.5A-D). TTNPB also attenuated the P4 increase in mammosphere size similar to RA in MCF7 cells (Figure 4.5E).

These data indicate that agonist bound RARα or RARγ mediates repression of P4-induced CK5 expression and mammosphere size.

48 A

B C Cell Proliferation CK5+ Cell Population 30000 Veh Veh 4000 P4 P4 P4 + 1nM TTNPB P4 + 1nM TTNPB P4 + 10nM TTNPB P4 + 10nM TTNPB P4 + 50nM TTNPB P4 + 50nM TTNPB P4 + 100nM TTNPB P4 + 100nM TTNPB 3000 P4 + 1uM TTNPB 20000 P4 + 1uM TTNPB

2000

10000

1000 Cell number (red object count) CK5+ cell count (green object count)

0 0 0 24 48 72 96 0 24 48 Time (hours) Time (hours)

D E

) *** 2 10000 m µ P4+ TTNPB TTNPB TTNPB P4 Veh 8000 CK5 6000

B 4000 PR A 2000

Mammosphere Size ( Size Mammosphere 0 α-tubulin E2 Veh Veh P4+ E2+P4 E2+RA TTNPB TTNPB E2+P4+RA +E2 E2+TTNPB

E2+P4+TTNPB Figure 4.5 Retinoid blockade of P4-induced CK5 expression and mammosphere formation occurs through RARs. (A-C) To determine the lowest effective concentrations of the RAR-specific agonist TTNPB, T47D cells expressing constitutively expressed nuclear mCherry and CK5 promoter GFP reporter were treated with the indicated increasing concentrations of TTNPB with or without P4 for 24 hours. (A) Cell lysates were harvested using RIPA buffer and analyzed via immunoblot. α-tubulin was used as a loading control. (B and C) Cells were treated as in A and cell number was assessed using red object count (B) and CK5 was assessed with GFP expression (C). (D) T47D cells were treated with vehicle, 100 nM P4, 10 nM TTNPB or the combination for 24 hours. CK5 and PR expression were measured by immunoblot. (E) MCF7 cells stably expressing ZsGreen were plated at a density of 100 cells per well in quintuplicate in mammosphere media in 96-well plates and treated with 10 nM E2 (to induce PR levels) plus the following hormone combinations: ethanol vehicle, 100 nM P4, 100 nM 9-cis RA, P4 plus RA, 10 nM TTNPB, or P4 plus TTNPB. After two weeks, mammospheres were analyzed using the IncuCyte Zoom. Data represent mean mammosphere size ± s.e.m. *P < 0.05 **P < 0.01 ***P < 0.001 (ANOVA followed by Tukey post-hoc).

49 Retinoic acid antagonizes progesterone receptor regulation of CK5 by modulating RARα

recruitment near hormone response elements in the CK5 promoter

As we had previously determined that ligand-activated PR is recruited to the -1.1 kb PRE

in the CK5 promoter (Figure 3.1), we hypothesized that RA treatment may prevent this

recruitment. To test this, we pretreated cells with RA for 30 minutes followed by P4 treatment

for one hour. RA did not prevent PR from binding to the -1.1 kb PRE (Figure 4.6B), excluding

loss of PR DNA binding at this site as a mechanism of RA antagonism of PR-mediated CK5

transcription.

Previous studies in keratinocytes identified that RARα binds to the CK5 promoter ~130 bp upstream of the transcriptional start site (TSS) at what was termed a negative RA response element (nRARE) (AGGTGTgaccggtgAGCTCA)174, 175. We speculated a similar mechanism could occur in breast cancer cells. We therefore treated breast cancer cells with vehicle, P4, RA, or RA plus P4 and performed ChIP for RARα followed by qPCR with primers surrounding the nRARE (Figure 4.6A). Interestingly, RARα was only present at the nRARE with P4 treatment, and was absent with RA alone or RA plus P4 (Figure 4.6E). We therefore tested RARα

recruitment to the -1.1 kb PRE under the same conditions, and similarly found that RARα was associated with the PRE region only under P4 conditions, but not with either RA alone or RA plus P4 (Figure 4.6D). Furthermore, P4 also induced PR binding at the -130 bp region, which also contains PRE half sites176 but, unlike at the -1.1 kb PRE, addition of RA blocked this recruitment (Figure 4.6C). Similarly, transcriptional coactivators p300 and CBP are recruited to both sites with P4 treatment, but lost (p300) or significantly reduced (CBP) at both sites with the addition of RA (Figure 4.6B and C). Negative primers were designed for upstream in the promoter and downstream within the CK5 gene to determine signal specificity. All values were

50 A

-1.1 kb primers -130 bp primers

** *** B D 0.20 -1.1 kb primers ** *** 0.03 -1.1 kb primers *** ** 0.15 0.02 ns *** ***

% Input% 0.10 0.01 % Input

nd nd nd 0.05 0.00 RA RA RA RA P4+RA P4+RA P4+RA P4+RA Veh Veh Veh Veh P4 P4 P4 P4 0.00 P4 P4 RA RA Veh Veh P4+RA IP: PR IP: p300 IP: CBP IP: IgG P4+RA IP: RARα IP: IgG ** C *** 0.12 *** E 0.10 -130 bp primers *** 0.25 *** 0.08 ** 0.20 -130 bp primers 0.06 *** *** 0.04 0.15

% Input% * 0.02 % Input 0.10 * 0.020 0.015 0.05 0.010 0.00 0.005 nd nd P4 P4 RA 0.000 RA Veh Veh RA RA RA RA P4+RA P4+RA Veh Veh Veh Veh P4+RA P4+RA P4 P4 P4 P4 P4+RA P4+RA IP: RARα IP: IgG IP: PR IP: p300 IP: CBP IP: IgG

gene neg F promoter neg G * 0.06 * 0.08 ** 0.04 0.06

0.04 0.02 0.02 0.00 UT P4 RA P+R IgG UT IgG P4 IgG RA IgG P+R 0.00 UT P4 RA P+R IgG UT IgG P4 IgG RA IgG P+R

H Vehicle P4 RA P4 + RA

Figure 4.6 PR and RARα recruitment to the CK5 promoter at the distal PRE region and a proximal RARE-containing region is P4-dependent and is blocked by RA.

51 Figure 4.6 PR and RARα recruitment to the CK5 promoter at the distal PRE region and a proximal RARE-containing region is P4-dependent and is blocked by RA. (A) Diagram showing location of primer sets used in B-E. (B-E) ChIP for PR, p300, CBP, control IgG (B and C), and RARα (D and E) was performed on T47D cells treated with vehicle, 100 nM P4, P4 plus 100 nM 9-cis RA, or P4 plus RA as indicated for 1 hour. qPCR was performed for (B and C) a 200 bp region spanning the -1.1 kb PRE and (D and E) a 200 bp region spanning the -130 bp RARE. Data represent mean (percent input) ± s.e.m. **P < 0.01 ***P < 0.001. (F and G) CK5 negative control primers. To assess the specificity of binding in the CK5 promoter, negative control primer sets were designed to target 200 bp regions more than 500 bp upstream of the putative PRE in the CK5 promoter (F) or downstream of the TSS (G). Data represent mean ± s.e.m. *P<0.05 **P<0.01 (H) T47D cells were treated as described for B for 1 hour then fixed with PFA and incubated with PR and RARα antibodies for one hour at room temperature. Duolink protocol and reagents were used to complete the proximity ligation assay. Cells were imaged at 20x.

52 less than the average signal from IgG ChIP (Figure 4.6F and G). Taken together, these data suggest that the proximal binding region, while not sufficient for transcription itself, is a cis- acting element that enhances transcription primarily driven by the more distal site, and that RA through RARα redirects PR away from this region to impair CK5 transcription.

We next wanted to assess direct interaction between PR and RARα. To do so, we started by attempting a co-immunoprecipitation (IP) experiment by IPing for PR in T47D cells lysates followed by immunoblotting for RARα; however, attempts using three different lysis buffers did not uncover an interaction between the two proteins (data not shown). In parallel with these experiments, a proximity ligation assay was also performed. Unexpectedly, signal was only found in the cytoplasm (Figure 4.6H). Despite this, in both vehicle and P4 treated cells signal was observed, though interactions seemed to be slightly more prevalent in the P4 treated group.

With RA treatment alone, however, this interaction was completely lost. Interestingly, with the addition of both P4 and RA, the fluorescent signal was much more dim and had substantially different localization. However, additional studies must be performed to validate these data and to determine if this lack of nuclear signal was due to a technical issue with nuclear membrane permeabilization. These studies therefore did not identify interaction between PR and RARα.

Discussion

Increasing evidence shows that there is substantially more crosstalk between nuclear receptors than previously thought, indicating the need to consider the larger nuclear receptor context of the cell. Here we present the novel finding that RARα acts as a cofactor for PR in a

P4-dependent manner. However with the addition of its own ligand, RARα antagonizes the function of PR, leading to decreased CK5 gene expression and reduction in other P4/PR regulated functions such as mammosphere formation (Figure 4.7). This is not entirely

53 P4 +RA RA PR RAR

P4 coactivators p300 P4 CBP CBP PR PR PR RAR PR PR

PRE RARE/ KRT5 PRE RARE/ KRT5 (1.1kb) PRE half site (1.1kb) PRE half site (0.13kb) (0.13kb)

Figure 4.7 Diagram of proposed coactivator bridging between promoter elements. Under P4 conditions, PR and RARα occupy their respective response elements and coactivators form a functional bridge. RA removes RARα/p300 and reduces CPB occupancy, disrupting the bridge.

54 unexpected, as crosstalk between RARα and other steroid receptors has been recently identified.

In breast cancer cell lines, RARα can interact with ER binding sites, usually in an ER-dependent fashion110. Similar to what we observed with PR, loss of RARα at these sites lead to loss of transcription. In prostate cancer, which is often driven by androgen receptor (AR) in the same way many breast cancers are driven by ER, it was observed that RARγ co-occupies AR binding sites to modulate the AR transcriptome177. Interestingly, genetic knockdown of RARγ had an

even greater effect on reducing AR driven transcription than addition of the endogenous ligand

for RARγ. Together these data imply that RARs may be important cofactors for nuclear hormone receptors and act as important sensors for an additional layer of gene regulation.

These studies by our group and others show that in hormone driven cancer, utilizing nuclear receptor crosstalk may be another method to modulate hormone receptor function when eventual resistance to hormone therapies develops. In these studies we provide clear evidence of crosstalk between PR and RARα at the CK5 promoter, though the importance of these receptors in facilitating interaction at the two sites identified in the promoter region could be further delineated via site directed mutagenesis experiments. We were not able to identify direct interaction between PR and RARα through co-IP or PLA experiments, however additional studies need to be performed in order for these data to be conclusive. It is also feasible that there is no direct interaction between PR and RARα, and that are instead interacting indirectly as

components of a much larger complex. In this particular study we focused on the interplay of

these receptors at a specific locus, however this has allowed us to uncover a previously unknown

mechanism of crosstalk between PR and RARα, which we believe has important consequences

genome-wide that needs further evaluation.

55 CHAPTER V

RETINOIDS BLOCK ACCUMULATION OF CK5+ CELLS DURING ENDOCRINE

THERAPY TREATMENT‡

Introduction

In breast cancer, increasing prevalence of CK5+ cells has been correlated with poor prognosis and lower overall survival; this had been shown not only in the more aggressive basal- like subtype, but also in luminal breast cancers47, 48. Further, within luminal breast cancer, CK5 is also a marker of poor response to endocrine treatment38. Unfortunately, anti-estrogen agents such as tamoxifen (Tam), fulvestrant (ICI), or estrogen depletion lead to increased CK5 expression in breast cancer cell lines, and neoadjuvant aromatase inhibitor plus or minus Tam treatment enriched for CK5+ cells in patient biopsy samples36. These data indicate that the CK5+ population may be refractory to traditional treatments, particularly as they lack ER expression necessary for endocrine treatments, and therefore may need to be targeted by a different mechanism. As previously mentioned, our lab performed a small molecule screen to identify compounds that can suppress the P4-induced CK5+ population in breast cancer cells163. In this screen we identified three retinoids that were able to suppress the CK5+ cell population expansion in response to P4.

While there have been several studies looking at the effects of retinoids as a cancer treatment or preventative, results have been mixed (reviewed in Connolly et al.114). In breast cancer, the targeting of RXR has been unsuccessful, however treatments targeting RAR have shown promise120-122. Another factor that needs to be considered when evaluating these studies is timing of treatment. In the clinical trials using retinoids in breast cancer, retinoid treatment was

‡ Portions of this chapter were originally published in Fettig et al.37 and are republished with permission.

56 either given before cancer diagnosis or after chemo- or endocrine therapy was concluded. In our

in vitro studies, we have shown that co-treatment with retinoids during P4 treatment can prevent

the expansion of CK5+ cells (Figure 4.1); additionally, retinoids were very effective at

preventing tumorigenesis in carcinogen-induced rodent mammary tumor models when given

prophylactically168, 178. RAR agonists have also shown efficacy in preventing cancer development in high-risk women, in addition to preventing secondary tumor recurrence in premenopausal patients119, 120. We therefore reasoned that co-treatment with retinoids may reduce the number of CK5+ cells that accumulate, and thus could prevent recurrences by lessening residual tumor initiating cells.

Results

Co-treatment with retinoids reduces accumulation of therapy resistant CK5+ cells during

endocrine therapy

To test the hypothesis that co-treatment with retinoids can prevent the accumulation of

CK5+ cells seen in response to endocrine therapy, we injected female NSG mice with T47D cells supplemented with E2 and allowed tumors to establish to an average of 75 mm3. Tumors were then stratified into six groups: continued on E2 plus vehicle or the synthetic retinoid fenretinide

(Fen; 100 mg/kg), estrogen withdrawal (EWD) plus vehicle, EWD plus Fen, EWD plus ICI plus one dose of Fen, or EWD plus ICI plus continued Fen treatment, and treated for three weeks. The

EWD groups trended towards a decrease in tumor growth compared to E2 alone, achieving significance at the study endpoint for the EWD only and both EWD+ICI+Fen groups (Figure

5.1A). E2 plus Fen and EWD plus Fen groups also tended toward reduced growth but did not reach significance. At the end of the study, tumors were harvested, weighed, and evaluated for

CK5 expression via IHC. Overall, tumor mass mirrored trends seen via tumor measurements

57 A C

E2 E2+Fen

EWD EWD+Fen

EWD+ICI+Fen (pre) EWD+ICI+Fen

B 500

400 D

0.3 * 300 *

0.2 200 Tumor mass (mg) mass Tumor 0.1 100

Fraction CK5 positive cells positive CK5 Fraction 0.0 0 E2 E2 EWD EWD + Fen E2 + Fen + Fen + E2 + Fen EWD + ICI EWD + ICI EWD + Fen + Fen (pre) EWD + ICI ICI + EWD (pre)Fen + EWD + ICI ICI + EWD EWD + Fen EWD + ICI Fen

EWD + ICI Fen (pre) Figure 5.1 Co-treatment with retinoids during estrogen depletion reduces accumulation of CK5+ breast cancer cells. (A) A total of 1x106 T47D cells were implanted into the left and right mammary fat pads of female NOD/SCID mice. Mice were given E2 pellets at time of cell injection. When tumors reached 75 mm3 average volume they were stratified into six treatment groups, (n=12 tumors each group): continued on E2, E2 plus fenretinide (Fen), EWD, EWD plus Fen, EWD plus ICI plue one dose of Fen, EWD plus ICI plus continued Fen treatment. Change in tumor volumes relative to treatment start are plotted versus the number of days post-treatment. Data represent mean ± s.e.m. *P < 0.05. (B) Tumor masses measured at harvest. (C) Representative IHC for CK5 in tumor sections from all treatment groups. (D) The percent of CK5+ cells was analyzed using an Aperio digital pathology microscope for whole sections of tumors in each group (n=3) and plotted as percent positive cells ± s.e.m. *P < 0.05.

58 though none reached significance due to high variability (Figure 5.1B). When evaluated via IHC, tumors in the EWD group, which mimics the effects of aromatase inhibitors or depletion of estrogens, showed a robust increase in CK5+ cells compared to E2 treated tumors (18.2% versus

5.2%), while Fen prevented this increase during EWD with or without the presence of ICI

(EWD+Fen, 4%; EWD+ICI+Fen (pre), 4.3%; EWD+ICI+Fen, 4.4%) (Figure 5.1C and D).

Therefore, retinoids can decrease the accumulation of CK5+ cells associated with prolonged endocrine treatment.

Discussion

Many factors need to be considered when developing and testing new treatments for cancer patients, and in fact timing of treatment administration is a major player in the success or failure of treatments during clinical trials. While it was unexpected that co-treatment with Fen did not suppress tumor growth as retinoids have strong anti-proliferative effects in vitro, there was a trend toward reduced growth in the control E2 plus Fen group, indicating that over a longer time course differences may develop. Furthermore, it is possible that withdrawal of the pro-growth estrogen stimulus overrode the effects of Fen on tumor growth. We do know, however, that over time cells withdrawn from estrogen stimulus convert to a non-estrogen dependent state and begin to grow again, highlighting another reason that longer studies could be useful. As a surrogate to longer studies, tumors can also be removed, digested, and plated in mammosphere formation assays to evaluate further tumor initiating potential. While this type of study was attempted, there was high variability and low cell number making interpreting the data difficult (data not shown), and indicating that further assay optimization is necessary.

Nonetheless, the main purpose of our studies was to assess the effects of Fen on accumulation of

CK5+ cells. This study suggests that co-treatment with Fen during estrogen deprivation or

59 endocrine therapy suppresses the enrichment in CK5+ cells that has been observed, perhaps with just a single dose36. Further studies are still necessary to conclusively determine if Fen is able to prevent an increase in CK5 expression in response to ICI or even chemotherapies, however these data are promising. This study indicates that timing of retinoid administration could be what is driving the failure of retinoid agents in patients.

60 CHAPTER VI

RETINOID ANTAGONISM OF PROGESTERONE RECEPTOR ALTERS GENOMIC

RECRUITMENT AND TRANSCRIPTIONAL PROGRAMS

Introduction

Mounting evidence suggests that interplay between nuclear transcription factors is more prevalent than previously thought. Genome-wide studies have revealed many such interactions, both among steroid receptors, as well as between steroid receptors and other families of nuclear receptors110, 165, 166, 177. While studies in Chapter IV show a novel interaction between PR and

RARα at the CK5 promoter, these receptors have much farther-reaching effects throughout the genome. We therefore reasoned that RA/RARα would antagonize the effects of P4/PR on a large set of genes, particularly those involved in facilitating CSC characteristics.

Results

In order to assess gene regulation by P4 and RA on a global scale, we treated T47D cells in triplicate with vehicle, P4, RA, or P4 plus RA for 6 hours, harvested total RNA, performed poly-A enrichment to select mRNA, prepared libraries, and submitted samples for RNA- sequencing. We found that, in total, 3550 transcripts were significantly modulated with P4 treatment (Figure 6.1A; Adj p-value ≤ 0.05, Fold change ≥1.5, FDR < 0.1). Of those transcripts,

329 had the effects of P4 reversed by the addition of RA: 181 were increased by P4 then suppressed with addition of RA, and 148 were decreased by P4 then rescued by addition of RA

(Table 6.1). By using Ingenuity Pathway Analysis (IPA), we determined that several expected and unexpected pathways were regulated by this combination of hormones (Figure 6.1B).

Unsurprisingly, cancer related pathways such as MAPK signaling, HER-2 signaling, Death

61 A 3550 transcripts modulated with P4 65 2% Modulated with P4 329 9% Reversed with P4+RA Syngergistic P4+RA

3156 89%

Adj p-value ≤ 0.05 Fold change ≥1.5 FDR < 0.1

B

UVA—Induced MAPK Signaling

Antioxidant Action of Vitamin C

HER-2 Signaling in Breast Cancer

Acetate Conversion to Acetyl-CoA

Endothelin-1 Signaling

Phospolipases

Death Receptor Signaling

Sphingosine-1-phosphate Signaling

PI3K Signaling in B Lymphocytes

14-3-3-mediated Signaling

p70S6K Signaling

Oct4 in Embryonic Stem Cell Pluripotency

TNFR1 Signaling

Figure 6.1 RNA-seq data: P4 regulated transcripts and downstream pathways. (A) RNA-seq results identified 3550 transcripts modulated by P4 treatment based on an Adj p- value < 0.05, Fold change > 1.5, and FDR < 0.1. Of these transcripts, 329 had the effects of P4 reversed with addition of RA. For 65 transcripts, the effects of P4 plus RA were synergistic. (B) Ingenuity pathway analysis (IPA) was used to identify pathways enriched in the 329 gene set identified in A.

62 receptor signaling, p70S6K signaling, and PI3K signaling came up, as did the metabolism related pathway “Acetate conversion to acetyl-CoA”. Interestingly, the vast majority (30 out of 40) of the top 40 genes inversely regulated by P4 and RA were increased with P4 treatment and suppressed with RA. While investigating similarly regulated transcripts, some expected genes fall out; CK5 (KRT5) was the third highest affected in this manner (Table 6.1). Additional CSC associated genes regulated in this manner included KLF4, CD44, SOX4, ETS2, and RUNX1.

These data show that P4 supports gene expression and pathways that are important for sustained tumorigenesis, while RA potently opposes these effects.

When we look at the upstream regulator function in IPA, some interesting patterns emerge (Figure 6.2). In the RA alone treated group, the gene signature expectedly pulls RAR

(RARA) as one of the top 5 regulators and, in fact, ATRA (tretinoin) is the top upstream regulator. Unexpectedly, PR (PGR) is the second strongest predicted upstream regulator, despite the lack of progestin stimulation in this group. Similarly, in the P4 only treated group, RAR is found as a potential upstream regulator, again without the presence of its endogenous ligand.

Further, other nuclear receptors are also found on each of the potential upstream regulator lists including GR (NR3C1, dexamethosone) and ER (ESR1, beta-estradiol). These results highlight the importance of nuclear receptor crosstalk in gene regulation.

An important factor to consider from our analysis is that we were focused on genes that had differential expression across treatments. Many mRNA transcripts are represented in our complete gene list whose expression does not change among treatment groups. It is still possible, however, that PR and/or RAR may act as cofactors and assist in regulating their expression. In addition, previous ChIP-seq experiments performed in the lab identified a subset of genes with promoters occupied by PR regardless of the presence of ligand130. Interestingly, several of these

63

pval padj 0.543 0.957 0.050 0.351 0.5700.081 0.964 0.462 0.304 0.835 0.087 0.478 0.7520.1190.709 0.988 0.567 0.984 0.561 0.962 0.165 0.666 0.006 0.080 0.053 0.360 0.734 0.987 0.140 0.616 0.341 0.863 0.013 0.143 0.282 0.813 0.659 0.979 0.763 0.989 0.037 0.290 0.013 0.144 RA vs. Veh -0.186 -0.360 -0.051 -0.310 -0.209 -0.242 -0.043 -0.240 -0.044 -0.064 -0.313 -0.395 -0.524 -0.098 -0.276 -0.144 -0.495 -0.391 -0.144 -0.075 -0.580 -0.689 log2(FC) pval padj on on 3.71E-18 1.17E-16 0.151 0.257 0.787 1.17E-09 1.76E-08 1.31E-25 6.19E-24 0.064 0.585 0.968 1.16E-11 2.12E-10 0.130 0.476 0.936 2.10E-06 2.01E-05 0.058 0.772 0.990 6.19E-17 1.80E-15 0.456 0.004 0.057 4.61E-36 3.56E-34 8.75E-57 1.44E-54 0.071 0.537 0.955 7.72E-17 2.23E-15 5.70E-22 2.21E-20 3.31E-13 7.22E-12 P4+RA vs. Veh -1.182 -1.124 -1.295 -1.303 -0.979 -1.355 -1.878 -1.907 -1.577 -1.556 -1.951 log2(FC) pval padj 1.34E-04 0.004113 1.013 1.80E-06 1.74E-05 0.346 0.111 0.548 1.32E-05 6.95E-04 1.213 2.05E-12 4.09E-11 5.65E-06 3.50E-04 1.015 2.77E-06 2.59E-05 0.100 0.647 0.977 1.89E-05 9.22E-04 0.928 0.001756 0.007949 0.459 0.127 0.584 4.58E-04 0.010312 1.273 7.67E-11 1.29E-09 2.30E-05 0.001063 1.979 1.93E-88 8.15E-86 0.0783.18E-04 0.444 0.007858 0.922 1.488 4.27E-27 2.19E-25 3.87E-042.91E-05 0.0091382.21E-04 0.001254 2.037 0.006028 1.850 2.004 3.02E-52 5.00E-34 3.88E-50 1.87E-69 3.62E-32 4.39E-67 5.34E-06 3.42E-04 1.622 1.21E-52 1.58E-50 2.74E-10 5.33E-08 0.781 4.00E-04 0.002221 2.61E-05 0.001156 1.757 3.12E-13 6.82E-12 4.13E-06 2.74E-04 1.745 5.40E-73 1.32E-70 0.258 0.008 0.102 5.90E-06 3.60E-04 2.840 9.80E-61 1.77E-58 0.418 0.023 0.213 3.91E-04 0.009166 2.801 3.30E-17 9.70E-16 3.53E-25 2.53E-22 1.319 1.46E-04 9.20E-04 0.032 0.930 0.998 2.03E-07 1.98E-05 4.159 1.08E-81 3.78E-79 1.21E-08 1.66E-06 4.467 6.52E-96 4.12E-93 0.001077 0.017675 1.298 2.18E-12 4.32E-11 0.362 0.056 0.373 0.001482 0.021396 0.892 0.002620 0.011124 0.001822 0.024063 1.201 4.30E-12 8.22E-11 0.071 0.688 0.981 0.001755 0.023594 1.614 7.64E-74 1.93E-71 0.003096 0.032387 1.285 7.22E-09 9.86E-08 0.360 0.106 0.535 0.005901 0.047826 1.448 3.78E-07 4.11E-06 0.002701 0.030462 2.193 2.19E-79 7.21E-77 0.082 0.499 0.945 0.001525 0.0217680.005794 2.706 0.0474350.004422 2.643 1.18E-17 0.040612 3.57E-16 2.623 7.71E-19 0.064 2.54E-17 2.29E-47 0.347 2.49E-45 0.851 0.282 0.997 0.813 0.001097 0.017852 1.936 8.17E-11 1.37E-09 P4+RA vs. P4 -0.572 -0.904 -0.523 -0.758 -0.276 -0.704 -0.977 -0.654 -1.199 -0.658 -0.405 -0.485 -0.464 -0.628 -0.404 -0.461 -0.770 -1.355 -0.331 -0.897 -0.439 -0.906 -0.703 -0.477 -0.699 -0.858 -1.046 -3.151 -0.798 -0.879 log2(FC) pval padj 7.68E-38 6.95E-36 0.591 2.58E-05 0.001156 5.73E-21 2.22E-19 0.677 6.90E-04 0.013303 3.32E-42 3.72E-40 0.428 0.001272 0.019660 1.40E-22 5.89E-21 0.638 0.002045 0.026018 1.96E-20 7.18E-19 1.031 3.75E-06 2.53E-04 4.06E-43 4.72E-41 0.953 2.94E-08 3.65E-06 3.67E-63 8.82E-61 0.721 8.29E-06 4.75E-04 3.92E-33 2.87E-31 0.717 2.18E-04 0.005985 3.17E-38 2.95E-36 0.619 5.13E-04 0.011074 6.79E-28 3.82E-26 1.270 4.19E-05 0.001685 2.02E-120 3.40E-117 1.051 1.57E-15 6.33E-13 P4 vs. Veh 1.870 1.33E-24 6.34E-23 1.796 7.23E-10 1.14E-08 1.724 1.26E-23 5.71E-22 1.772 1.45E-17 4.35E-16 1.890 1.10E-100 7.41E-98 1.917 1.27E-29 7.79E-28 1.992 2.99E-20 1.08E-18 1.939 2.32E-18 7.27E-17 2.127 2.02E-13 4.48E-12 1.931 2.10E-23 9.30E-22 1.973 1.02E-46 1.35E-44 2.384 2.27E-128 4.24E-125 2.5012.4782.408 3.29E-78 7.40E-60 1.18E-75 3.72E-100 1.54E-57 2.41E-97 2.082 9.59E-87 4.14E-84 2.218 5.51E-152.136 1.36E-13 9.34E-23 3.99E-21 2.524 2.76E-105 2.21E-102 2.654 2.28E-29 1.36E-27 2.184 1.30E-113 1.56E-110 3.6113.345 1.04E-30 6.76E-29 3.100 7.01E-30 4.40E-28 4.52E-66 1.12E-63 3.539 1.33E-94 6.78E-92 2.794 1.22E-21 4.92E-20 3.848 1.17E-31 7.82E-30 4.470 5.69E-42 6.35E-40 4.958 5.50E-117 7.12E-114 5.346 3.27E-138 1.10E-134 -1.773 -1.801 -1.723 -1.941 -2.010 -2.308 -2.599 -2.958 -2.294 -2.175 -3.221 PPL ETS2 TTC6 KLF4 GJB2 TCF4 SOX4 CD44 HES2 GLUL APLN KRT5 SGK1 RERG IKZF2 GRPR GINS2 KLF15 MAFB TAOK3 DUSP1 TTLL12 KCNG1 RUNX1 SAMD9 ARID5B MICAL2 SCUBE2 TRERF1 CHRM4 BTN3A1 SRGAP3 MMP15 IMPDH1 Top 40 genes inversely regulated by expressiregulated inversely absolute genesRA andTop P4 by by 40 SLCO4A1 ARHGEF4 ZC3H12A C16orf80 RHOBTB2 CAMSAP1 PPARGC1B 61.302 59.117 82.934 88.387 61.892 67.101 48.123 66.539 61.810 878.157 228.942 372.473 149.610 119.364 418.700 221.902 170.901 148.834 257.014 508.937 437.931 217.581 281.067 861.528 676.607 214.393 421.848 454.426 270.958 371.108 8566.548 1830.845 1219.497 1081.274 1812.458 2189.715 1426.177 2061.462 1074.827 1170.389 5078.983 baseMean gene_symbol log2(FC) Table Table 6.1

64 binding sites were near tRNA genes; motif analysis in these genes did not find canonical PRE

sequences but instead identified an RARE sequence. In our RNA-seq study, tRNAs were not

captured due to their lack of a poly(A) tail and the inability to sequence through their complex their secondary structure without additional processing. However, assessing these tRNAs by

ChIP-qPCR revealed that PR and RARα are both recruited to these genes (Figure 6.3). Due to these limitations of our study, our next focus was to investigate how RA through RARα modifies

the binding pattern of PR both with and without the present of P4. To do this, we treated T47D for one hour with vehicle, P4, RA, or P4 plus RA in triplicate. We then fixed chromatin with formalin, harvested, performed IP for PR, RARα, and PolR3a (a subunit of the polymerase used

to transcribe tRNAs), prepared libraries, and submitted samples for sequencing. Due to

antibodies being discontinued, additional antibody optimization was performed (Figure 6.4). We

are currently waiting for analysis of these data.

Discussion

It is clear from several recently published studies that interactions between nuclear

receptors are critical in regulating transcriptional programs. This is further highlighted in this

study as several identified potential upstream regulators include nuclear receptors beyond the

two under investigation, even in the absence of their endogenous ligands. It has also been

proposed that multiple nuclear receptors, including ER and RAR, have the ability to participate

in a “megatrans” complex in order to cross-regulate expression of each others target genes166.

Our data indicate that these complexes may be relatively abundant, and perhaps found at most or all nuclear receptor regulated genes. Additionally, these data highlight several pathways that are likely important in the P4-supported CSC phenotype that are suppressed with the addition of RA, indicating that they might be targetable as a way to prevent expansion of the CSC population.

65

Figure 6.2 Upstream regulators of P4 and RA transcriptomes. RNA-seq data were analyzed by Ingenuity pathway analysis (IPA) software and upstream regulators were identified for the denoted treatment group gene sets.

Figure 6.3 PR and RARα are recruited to tRNA promoters with P4 and RA cotreatment. PR and RARα ChIP was performed as previously described (Figure 4.6). Primers for listed tRNAs were used to assess recruitment to these regions.

66 However, further studies are needed to determine which of these pathways may be specific to this CK5+ CSC population.

As previously discussed, resistance to endocrine therapy is common in patients. Some have proposed that targeting PR in addition to ER may be more effective as PR is able to antagonize ER function. Our studies imply that treatment with PR agonists may in fact lead to the expansion of a more quiescent CSC population, which in turn could lead to increased recurrence later in life. However, it may be possible to identify a combination of nuclear receptor targets that may aid in preventing this. Studies utilizing technologies such as Rapid

Immunoprecipitation Mass spectrometry of Endogenous proteins (RIME), which combines both nuclear receptor recruitment to DNA and proteomics to identify binding partners may be useful in this endeavor. Further, if common downstream pathways can be identified using studies such as this, a more specific target may be identified, preventing unfavorable actions of these nuclear receptors such as enhanced tumorigenic potential, while still preserving their more favorable, anti-tumor action.

67 15

10

5

Fold change % Input % change Fold 0 Veh P4 Veh P4 Veh P4 Veh P4

PR F-4 PR 6A PR Ab8 IgG

Figure 6.4 PR antibody optimization for ChIP. Three PR antibodies were tested against IgG negative control with chromatin treated with either EtOH vehicle (Veh) or P4 treatment for one hour. All IPs used the same chromatin samples.

68 CHAPTER VII

DISCUSSION AND FUTURE DIRECTIONS

Cytokeratins (CKs) are intermediate filament proteins that form tetrameric complexes of two Type I and two Type II peptides important for cell structure and motility179. However, CKs can influence other cellular processes such as proliferation, migration, invasion, and stress related signaling180-183. For example, CK17 interaction with the scaffolding protein 14-3-3σ in

keratinocytes alters its location in the cytoplasm to activate Akt signaling and cell cycle

progression181. Likewise, in T4-2 basal-like breast cancer cells 14-3-3σ interacts with CK5,

CK17, and actin to facilitate cell migration and invasion184. Interestingly, CK17 (Type II) can be

a binding partner for CK5 (Type I), suggesting the CK5:CK17 tetramer could mediate these

interactions179. Similarly, CK14 and CK5 were found to be important for cancer cell invasion,

and are found in cells at the leading edge of breast cancer invadopodia 56, 184. Thus, CKs can

influence a wide variety of processes important for cancer cells. CK5 in particular is a signature

marker of poor prognosis in basal-like breast cancer54 and is also associated with worse

prognosis within luminal ER+ tumors55. In these studies we showed that P4/PR directly regulates

CK5 transcription via binding to the promoter, that these P4-induced CK5+ cells have high

tumorigenic potential, and that CK5 is necessary for the P4-mediated increase in mammosphere

size. P4 is a key hormone promoting expansion of murine mammary stem cells and human breast

progenitor cells185-187, an action that is maintained in some breast cancers 55, 136, 156. Therefore, PR

regulation of CK5 may be pivotal in dictating cancer cell phenotype.

Here we investigated the mechanism by which P4 and RA modulate expression of CK5

through their cognate receptors (PR and RAR, respectively). Both receptors are members of the

nuclear receptor family of ligand-activated transcription factors; PRs belong to the steroid

69 hormone receptor subclass (NR3), whereas RARs belong to the RXR heterodimer subclass

(NR1)188. Previous studies demonstrated that progestins and RA cross regulate expression of each other’s cognate receptors in breast cancer cells. RA decreased PR mRNA and protein levels, and conversely progestins decreased RARα and RARγ at the transcript level189-191.

However, little has been described concerning functional crosstalk between PR and RARs.

Here we define a unique convergence between RAR and PR signaling at the CK5 locus in breast cancer cells. We describe two core promoter regions that affect transcription. One at a more distal (-1.1 kb from the TSS) essential region containing a PRE and one in the proximal region centered at -130 bp that contains a previously described RARE 174, 175 and glucocorticoid receptor half sites 176; the latter could also function as PR binding sites. Somewhat surprisingly,

P4 recruits both PR and RARα to both of these regions. The distal PRE region is essential for robust P4 activation of the gene, while the proximal region is not sufficient for activation alone.

We therefore propose that the two PR binding regions act similarly to the MMTV promoter, which contains a strong distal PRE in addition to three more proximal half sites that act to further enhance transcription89. The fact that RA does not induce RARα binding at the RARE in the proximal region indicates that RAR cross regulation of cytokeratins occurs uniquely in breast cancer cells. In keratinocytes, RA induces RARα to bind to negative RAREs of multiple basal cytokeratin promoters including CK5 (-130 bp), CK6, CK14, and CK17, and prevents their expression, likely through recruitment of co-repressors, to maintain differentiation 174, 175. Our data suggest that RA perturbs PR binding at the proximal but not the distal PRE, while also reducing or preventing recruitment of essential cofactors p300 and CBP. We therefore propose that in breast cancer cells, RARα allows looping of the CK5 promoter to facilitate interaction between complexes at the distal site containing PR and the proximal site. Additionally, RARα

70 acts as a sensor for RA, which removes RAR from the proximal response element and reduces

CK5 transcription. However additional studies such as chromatin conformation capture (3C) are

needed to determine if promoter looping is in fact involved. That loss of RARs did not affect P4

transcription of CK5 suggests that RARα is not an essential cofactor, but is required for negative regulation at the proximal response element. This proposal is intriguing in light of a report by

Ross-Innes et al. that describes that RARα is co-localized at ER binding sites in breast cancer

cells, and acts as a positive cofactor for ER gene regulation110. PR can similarly redirect ER

away from a subset of its core binding sites in breast cancer cells, substantially influencing cell

growth129.

Thus far we have found no evidence that either estrogens or ER directly affect CK5

transcription in breast cancer cells. ER does, however, increase expression of CK19 in breast

cancer cells, one of three highly expressed CKs (CK8 and CK18) tied to a luminal breast cancer

phenotype192. The gradual increase in CK5+ cells seen in response to antiestrogen treatment36

suggests that this may occur through indirect transcriptional mechanisms. Collectively, our data

support that CK5 is an important target for hormonal and nuclear receptor regulation that affects

the downstream phenotype of breast cancer cells.

Preclinical studies on retinoids showing potent antiproliferative activity in breast cancer

cell lines and prevention of carcinogen induced mammary cancers in rodent models led to their

exploration for clinical use113. However, several trials using retinoids as single agents (usually

ATRA or RXR agonists) or in combination with Tam or chemotherapy for invasive breast cancer

failed to meet study objectives (reviewed in Garattini et al.170). One exception is the use of the

synthetic retinoid-like compound fenretinide, which is effective in prevention of secondary

breast cancer in young women171. Our unbiased screening results upheld that retinoids prevent

71 the P4 expansion of CK5+ breast cancer cells163. Another study surveyed gene expression

profiles of CSCs and used connectivity map analysis of FDA approved drugs to identify that

ATRA is negatively associated with a breast CSC phenotype193. RAs may therefore be more

efficacious in preventing initial transforming events or conversion to a CSC phenotype, but may

be less effective in rescuing cells that have already attained this state. Our data indicate that, in

the right contexts, retinoids may reduce the acquisition of cells prone to tumor recurrence, an

issue of particular relevance to luminal breast cancer, which can have long dormancy periods

prior to relapse149. In addition, the progestin-associated increase in breast cancer incidence

during hormone replacement therapy is hypothesized to occur through expansion of stem cells194;

RA co-treatment could prospectively prevent this.

Targeting CK5 itself has not been extensively explored since, as a structural protein,

compensatory expression of other CKs may be possible and there may be several off target

effects in normal cell types expressing CK5. As we have shown that CK5 itself is mediating at least some of the CSC characteristics induced by P4 treatment, studies are currently underway in our lab to identify binding partners of CK5 in order to more directly assess its effects (data not shown). These studies have identified β-catenin and 14-3-3 proteins as possible mediators of downstream CK5 action and therefore may lead to additional treatment strategies to suppress the

CSC population.

Another method to develop treatments targeting this CK5 population is to assess other genes and pathways that are regulated in a parallel fashion to CK5. Our RNA-seq studies have made progress toward this goal by identifying genes and pathways important in mediating the effects of P4 and RA beyond just expression of CK5. For example, pathway analysis showed that cellular metabolism is significantly modulated with treatment of both P4 and RA. Previous

72 studies have shown that, in contrast to normal pluripotent and progenitor cells that generally favor glycolysis, CSCs have heightened OXPHOS function195-198. However, this discrepancy may be due to differing metabolic requirements for specific CSC populations. Interestingly, knockdown of breast CSC markers CD44 and Oct1, which are increased by P4 treatment, induces a shift from glycolytic to mitochondrial respiration (RNA-seq data)199, 200; interestingly

RA partially or completely reverses this increase in expression indicating that targeting of these alterations in metabolism may in fact aid in suppression of CSC properties.

Taken together, our work describes crosstalk among nuclear receptors PR and RARα at a single gene that is tied to a breast CSC phenotype. However, we speculate this crosstalk occurs at a broader genome level in breast cancer cells as we have seen transcriptional alterations in pathways important for sustaining tumor survival with P4 and RA co-treatment. An emerging paradigm in nuclear receptor research is co-occupancy of multiple nuclear receptors at regulatory sites and co-dependency for transcriptional activation, or conversely transrepression110, 129, 155.

This has typically been restricted among Type I or Type II nuclear receptors, but it is becoming more evident that it occurs across the entire nuclear receptor family. Further understanding of these complex relationships at the genome-wide level and in whole tumor models may allow for better use of multiple-hormone treatments that target specific nuclear receptor relationships and co-dependencies to prevent or treat breast cancers.

73 REFERENCES

1. Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA: a cancer journal for clinicians. 2018;68(6):394-424. doi: 10.3322/caac.21492. PubMed PMID: 30207593.

2. American Cancer Society. Cancer Facts & Figures 2018 Atlanta: American Cancer Society; 2018 [December 27, 2018]. Available from: http://www.cancer.org/content/dam/cancer- org/research/cancer-facts-and-statistics/annual-cancer-facts-and-figures/2018/cancer-facts-and- figures-2018.pdf.

3. Prat A, Perou CM. Deconstructing the molecular portraits of breast cancer. Molecular oncology. 2011;5(1):5-23. doi: 10.1016/j.molonc.2010.11.003. PubMed PMID: 21147047.

4. Prat A, Parker JS, Karginova O, Fan C, Livasy C, Herschkowitz JI, et al. Phenotypic and molecular characterization of the claudin-low intrinsic subtype of breast cancer. Breast cancer research : BCR. 2010;12(5):R68. doi: 10.1186/bcr2635. PubMed PMID: 20813035; PubMed Central PMCID: PMC3096954.

5. Dowsett M, Sestak I, Lopez-Knowles E, Sidhu K, Dunbier AK, Cowens JW, et al. Comparison of PAM50 risk of recurrence score with oncotype DX and IHC4 for predicting risk of distant recurrence after endocrine therapy. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2013;31(22):2783-90. doi: 10.1200/JCO.2012.46.1558. PubMed PMID: 23816962.

6. Brandao M, Ponde N, Piccart-Gebhart M. Mammaprint: a comprehensive review. Future oncology. 2018. doi: 10.2217/fon-2018-0221. PubMed PMID: 30156427.

7. Reeder-Hayes KE, Carey LA, Sikov WM. Clinical trials in triple negative breast cancer. Breast disease. 2010;32(1-2):123-36. doi: 10.3233/BD-2010-0310. PubMed PMID: 21778571.

8. Traina TA, Miller K, Yardley DA, Eakle J, Schwartzberg LS, O'Shaughnessy J, et al. Enzalutamide for the Treatment of Androgen Receptor-Expressing Triple-Negative Breast Cancer. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2018;36(9):884-90. doi: 10.1200/JCO.2016.71.3495. PubMed PMID: 29373071; PubMed Central PMCID: PMC5858523.

9. Emens LA. Breast Cancer Immunotherapy: Facts and Hopes. Clinical cancer research : an official journal of the American Association for Cancer Research. 2018;24(3):511-20. doi: 10.1158/1078-0432.CCR-16-3001. PubMed PMID: 28801472; PubMed Central PMCID: PMC5796849.

10. Papadimitriou M, Mountzios G, Papadimitriou CA. The role of PARP inhibition in triple- negative breast cancer: Unraveling the wide spectrum of synthetic lethality. Cancer treatment reviews. 2018;67:34-44. doi: 10.1016/j.ctrv.2018.04.010. PubMed PMID: 29753961.

74 11. Miller TW. Endocrine resistance: what do we know? American Society of Clinical Oncology educational book American Society of Clinical Oncology Annual Meeting. 2013. doi: 10.1200/EdBook_AM.2013.33.e37. PubMed PMID: 23714450.

12. Pan H, Gray R, Braybrooke J, Davies C, Taylor C, McGale P, et al. 20-Year Risks of Breast-Cancer Recurrence after Stopping Endocrine Therapy at 5 Years. The New England journal of medicine. 2017;377(19):1836-46. doi: 10.1056/NEJMoa1701830. PubMed PMID: 29117498; PubMed Central PMCID: PMC5734609.

13. Luque-Cabal M, Garcia-Teijido P, Fernandez-Perez Y, Sanchez-Lorenzo L, Palacio- Vazquez I. Mechanisms Behind the Resistance to Trastuzumab in HER2-Amplified Breast Cancer and Strategies to Overcome It. Clinical Medicine Insights Oncology. 2016;10(Suppl 1):21-30. doi: 10.4137/CMO.S34537. PubMed PMID: 27042153; PubMed Central PMCID: PMC4811269.

14. Marjanovic ND, Weinberg RA, Chaffer CL. Cell plasticity and heterogeneity in cancer. Clinical chemistry. 2013;59(1):168-79. doi: 10.1373/clinchem.2012.184655. PubMed PMID: 23220226; PubMed Central PMCID: PMC6220421.

15. Tan BT, Park CY, Ailles LE, Weissman IL. The cancer stem cell hypothesis: a work in progress. Laboratory investigation; a journal of technical methods and pathology. 2006;86(12):1203-7. doi: 10.1038/labinvest.3700488. PubMed PMID: 17075578.

16. Pozarowski P, Darzynkiewicz Z. Analysis of cell cycle by flow cytometry. Methods in molecular biology. 2004;281:301-11. doi: 10.1385/1-59259-811-0:301. PubMed PMID: 15220539.

17. Lapidot T, Sirard C, Vormoor J, Murdoch B, Hoang T, Caceres-Cortes J, et al. A cell initiating human acute myeloid leukaemia after transplantation into SCID mice. Nature. 1994;367(6464):645-8. doi: 10.1038/367645a0. PubMed PMID: 7509044.

18. Al-Hajj M, Wicha MS, Benito-Hernandez A, Morrison SJ, Clarke MF. Prospective identification of tumorigenic breast cancer cells. Proceedings of the National Academy of Sciences of the United States of America. 2003;100(7):3983-8. doi: 10.1073/pnas.0530291100. PubMed PMID: 12629218; PubMed Central PMCID: PMC153034.

19. Matsui W, Huff CA, Wang Q, Malehorn MT, Barber J, Tanhehco Y, et al. Characterization of clonogenic multiple myeloma cells. Blood. 2004;103(6):2332-6. doi: 10.1182/blood-2003-09-3064. PubMed PMID: 14630803; PubMed Central PMCID: PMC3311914.

20. Fang D, Nguyen TK, Leishear K, Finko R, Kulp AN, Hotz S, et al. A tumorigenic subpopulation with stem cell properties in melanomas. Cancer research. 2005;65(20):9328-37. doi: 10.1158/0008-5472.CAN-05-1343. PubMed PMID: 16230395.

21. Collins AT, Berry PA, Hyde C, Stower MJ, Maitland NJ. Prospective identification of tumorigenic prostate cancer stem cells. Cancer research. 2005;65(23):10946-51. doi: 10.1158/0008-5472.CAN-05-2018. PubMed PMID: 16322242.

75 22. Dalerba P, Dylla SJ, Park IK, Liu R, Wang X, Cho RW, et al. Phenotypic characterization of human colorectal cancer stem cells. Proceedings of the National Academy of Sciences of the United States of America. 2007;104(24):10158-63. doi: 10.1073/pnas.0703478104. PubMed PMID: 17548814; PubMed Central PMCID: PMC1891215.

23. Li C, Heidt DG, Dalerba P, Burant CF, Zhang L, Adsay V, et al. Identification of pancreatic cancer stem cells. Cancer research. 2007;67(3):1030-7. doi: 10.1158/0008- 5472.CAN-06-2030. PubMed PMID: 17283135.

24. Ginestier C, Hur MH, Charafe-Jauffret E, Monville F, Dutcher J, Brown M, et al. ALDH1 is a marker of normal and malignant human mammary stem cells and a predictor of poor clinical outcome. Cell stem cell. 2007;1(5):555-67. doi: 10.1016/j.stem.2007.08.014. PubMed PMID: 18371393; PubMed Central PMCID: PMC2423808.

25. Liu Y, Nenutil R, Appleyard MV, Murray K, Boylan M, Thompson AM, et al. Lack of correlation of stem cell markers in breast cancer stem cells. British journal of cancer. 2014;110(8):2063-71. doi: 10.1038/bjc.2014.105. PubMed PMID: 24577057; PubMed Central PMCID: PMC3992489.

26. Avilion AA, Nicolis SK, Pevny LH, Perez L, Vivian N, Lovell-Badge R. Multipotent cell lineages in early mouse development depend on SOX2 function. Genes & development. 2003;17(1):126-40. doi: 10.1101/gad.224503. PubMed PMID: 12514105; PubMed Central PMCID: PMC195970.

27. Leis O, Eguiara A, Lopez-Arribillaga E, Alberdi MJ, Hernandez-Garcia S, Elorriaga K, et al. Sox2 expression in breast tumours and activation in breast cancer stem cells. Oncogene. 2012;31(11):1354-65. doi: 10.1038/onc.2011.338. PubMed PMID: 21822303.

28. Scharenberg CW, Harkey MA, Torok-Storb B. The ABCG2 transporter is an efficient Hoechst 33342 efflux pump and is preferentially expressed by immature human hematopoietic progenitors. Blood. 2002;99(2):507-12. PubMed PMID: 11781231.

29. Alvi AJ, Clayton H, Joshi C, Enver T, Ashworth A, Vivanco M, et al. Functional and molecular characterisation of mammary side population cells. Breast cancer research : BCR. 2003;5(1):R1-8. PubMed PMID: 12559051; PubMed Central PMCID: PMC154129.

30. Engelmann K, Shen H, Finn OJ. MCF7 side population cells with characteristics of cancer stem/progenitor cells express the tumor antigen MUC1. Cancer research. 2008;68(7):2419-26. doi: 10.1158/0008-5472.CAN-07-2249. PubMed PMID: 18381450.

31. Christgen M, Ballmaier M, Bruchhardt H, von Wasielewski R, Kreipe H, Lehmann U. Identification of a distinct side population of cancer cells in the Cal-51 human breast carcinoma cell line. Molecular and cellular biochemistry. 2007;306(1-2):201-12. doi: 10.1007/s11010-007- 9570-y. PubMed PMID: 17660947.

76 32. Wright MH, Calcagno AM, Salcido CD, Carlson MD, Ambudkar SV, Varticovski L. Brca1 breast tumors contain distinct CD44+/CD24- and CD133+ cells with cancer stem cell characteristics. Breast cancer research : BCR. 2008;10(1):R10. doi: 10.1186/bcr1855. PubMed PMID: 18241344; PubMed Central PMCID: PMC2374965.

33. Cariati M, Naderi A, Brown JP, Smalley MJ, Pinder SE, Caldas C, et al. Alpha-6 integrin is necessary for the tumourigenicity of a stem cell-like subpopulation within the MCF7 breast cancer cell line. International journal of cancer. 2008;122(2):298-304. doi: 10.1002/ijc.23103. PubMed PMID: 17935134.

34. Vaillant F, Asselin-Labat ML, Shackleton M, Forrest NC, Lindeman GJ, Visvader JE. The mammary progenitor marker CD61/beta3 integrin identifies cancer stem cells in mouse models of mammary tumorigenesis. Cancer research. 2008;68(19):7711-7. doi: 10.1158/0008- 5472.CAN-08-1949. PubMed PMID: 18829523.

35. Bocker W, Moll R, Poremba C, Holland R, Van Diest PJ, Dervan P, et al. Common adult stem cells in the human breast give rise to glandular and myoepithelial cell lineages: a new cell biological concept. Laboratory investigation; a journal of technical methods and pathology. 2002;82(6):737-46. PubMed PMID: 12065684.

36. Kabos P, Haughian JM, Wang X, Dye WW, Finlayson C, Elias A, et al. Cytokeratin 5 positive cells represent a steroid receptor negative and therapy resistant subpopulation in luminal breast cancers. Breast cancer research and treatment. 2011;128(1):45-55. doi: 10.1007/s10549- 010-1078-6. PubMed PMID: 20665103; PubMed Central PMCID: PMC3851293.

37. Fettig LM, McGinn O, Finlay-Schultz J, LaBarbera DV, Nordeen SK, Sartorius CA. Cross talk between progesterone receptors and retinoic acid receptors in regulation of cytokeratin 5-positive breast cancer cells. Oncogene. 2017;36(44):6074-84. doi: 10.1038/onc.2017.204. PubMed PMID: 28692043; PubMed Central PMCID: PMC5668194.

38. Goodman CR, Sato T, Peck AR, Girondo MA, Yang N, Liu C, et al. Steroid induction of therapy-resistant cytokeratin-5-positive cells in estrogen receptor-positive breast cancer through a BCL6-dependent mechanism. Oncogene. 2015. doi: 10.1038/onc.2015.193. PubMed PMID: 26096934.

39. Snider NT, Omary MB. Post-translational modifications of intermediate filament proteins: mechanisms and functions. Nature reviews Molecular cell biology. 2014;15(3):163-77. doi: 10.1038/nrm3753. PubMed PMID: 24556839; PubMed Central PMCID: PMC4079540.

40. Thomas GH. The Cytoskeleton Penn State University2009 [cited 2019]. Available from: https://wikispaces.psu.edu/display/Biol230WFall09/The+Cytoskeleton.

41. Bragulla HH, Homberger DG. Structure and functions of keratin proteins in simple, stratified, keratinized and cornified epithelia. Journal of anatomy. 2009;214(4):516-59. doi: 10.1111/j.1469-7580.2009.01066.x. PubMed PMID: 19422428; PubMed Central PMCID: PMC2736122.

77 42. Moll R, Divo M, Langbein L. The human keratins: biology and pathology. Histochemistry and cell biology. 2008;129(6):705-33. doi: 10.1007/s00418-008-0435-6. PubMed PMID: 18461349; PubMed Central PMCID: PMC2386534.

43. Windoffer R, Beil M, Magin TM, Leube RE. Cytoskeleton in motion: the dynamics of keratin intermediate filaments in epithelia. The Journal of cell biology. 2011;194(5):669-78. doi: 10.1083/jcb.201008095. PubMed PMID: 21893596; PubMed Central PMCID: PMC3171125.

44. Snider NT, Park H, Omary MB. A conserved rod domain phosphotyrosine that is targeted by the phosphatase PTP1B promotes keratin 8 protein insolubility and filament organization. The Journal of biological chemistry. 2013;288(43):31329-37. doi: 10.1074/jbc.M113.502724. PubMed PMID: 24003221; PubMed Central PMCID: PMC3829444.

45. Woll S, Windoffer R, Leube RE. p38 MAPK-dependent shaping of the keratin cytoskeleton in cultured cells. The Journal of cell biology. 2007;177(5):795-807. doi: 10.1083/jcb.200703174. PubMed PMID: 17535969; PubMed Central PMCID: PMC2064280.

46. He T, Stepulak A, Holmstrom TH, Omary MB, Eriksson JE. The intermediate filament protein keratin 8 is a novel cytoplasmic substrate for c-Jun N-terminal kinase. The Journal of biological chemistry. 2002;277(13):10767-74. doi: 10.1074/jbc.M111436200. PubMed PMID: 11781324.

47. van de Rijn M, Perou CM, Tibshirani R, Haas P, Kallioniemi O, Kononen J, et al. Expression of cytokeratins 17 and 5 identifies a group of breast carcinomas with poor clinical outcome. The American journal of pathology. 2002;161(6):1991-6. doi: 10.1016/S0002- 9440(10)64476-8. PubMed PMID: 12466114; PubMed Central PMCID: PMC1850928.

48. Abd El-Rehim DM, Pinder SE, Paish CE, Bell J, Blamey RW, Robertson JF, et al. Expression of luminal and basal cytokeratins in human breast carcinoma. The Journal of pathology. 2004;203(2):661-71. doi: 10.1002/path.1559. PubMed PMID: 15141381.

49. Makino T, Yamasaki M, Takeno A, Shirakawa M, Miyata H, Takiguchi S, et al. Cytokeratins 18 and 8 are poor prognostic markers in patients with squamous cell carcinoma of the oesophagus. British journal of cancer. 2009;101(8):1298-306. doi: 10.1038/sj.bjc.6605313. PubMed PMID: 19755983; PubMed Central PMCID: PMC2768453.

50. Nhung NV, Mirejovsky P, Mirejovsky T, Melinova L. Cytokeratins and lung carcinomas. Ceskoslovenska patologie. 1999;35(3):80-4. PubMed PMID: 11038661.

51. Laakso M, Tanner M, Nilsson J, Wiklund T, Erikstein B, Kellokumpu-Lehtinen P, et al. Basoluminal carcinoma: a new biologically and prognostically distinct entity between basal and luminal breast cancer. Clinical cancer research : an official journal of the American Association for Cancer Research. 2006;12(14 Pt 1):4185-91. doi: 10.1158/1078-0432.CCR-06-0353. PubMed PMID: 16857790.

52. Malzahn K, Mitze M, Thoenes M, Moll R. Biological and prognostic significance of stratified epithelial cytokeratins in infiltrating ductal breast carcinomas. Virchows Archiv : an international journal of pathology. 1998;433(2):119-29. PubMed PMID: 9737789.

78 53. de Silva Rudland S, Platt-Higgins A, Winstanley JH, Jones NJ, Barraclough R, West C, et al. Statistical association of basal cell keratins with metastasis-inducing proteins in a prognostically unfavorable group of sporadic breast cancers. The American journal of pathology. 2011;179(2):1061-72. doi: 10.1016/j.ajpath.2011.04.022. PubMed PMID: 21801876; PubMed Central PMCID: PMC3157230.

54. Cheang MC, Voduc D, Bajdik C, Leung S, McKinney S, Chia SK, et al. Basal-like breast cancer defined by five biomarkers has superior prognostic value than triple-negative phenotype. Clinical cancer research : an official journal of the American Association for Cancer Research. 2008;14(5):1368-76. doi: 10.1158/1078-0432.CCR-07-1658. PubMed PMID: 18316557.

55. Sato T, Tran TH, Peck AR, Girondo MA, Liu C, Goodman CR, et al. Prolactin suppresses a progestin-induced CK5-positive cell population in luminal breast cancer through inhibition of progestin-driven BCL6 expression. Oncogene. 2014;33(17):2215-24. doi: 10.1038/onc.2013.172. PubMed PMID: 23708665; PubMed Central PMCID: PMC3913798.

56. Cheung KJ, Gabrielson E, Werb Z, Ewald AJ. Collective invasion in breast cancer requires a conserved basal epithelial program. Cell. 2013;155(7):1639-51. doi: 10.1016/j.cell.2013.11.029. PubMed PMID: 24332913; PubMed Central PMCID: PMC3941206.

57. Bauman PA, Dalton WS, Anderson JM, Cress AE. Expression of cytokeratin confers multiple drug resistance. Proceedings of the National Academy of Sciences of the United States of America. 1994;91(12):5311-4. PubMed PMID: 7515497; PubMed Central PMCID: PMC43984.

58. Liu F, Fan D, Qi J, Zhu H, Zhou Y, Yang C, et al. Co-expression of cytokeratin 8 and breast cancer resistant protein indicates a multifactorial drug-resistant phenotype in human breast cancer cell line. Life sciences. 2008;83(13-14):496-501. doi: 10.1016/j.lfs.2008.07.017. PubMed PMID: 18725232.

59. Inada H, Izawa I, Nishizawa M, Fujita E, Kiyono T, Takahashi T, et al. Keratin attenuates tumor necrosis factor-induced cytotoxicity through association with TRADD. The Journal of cell biology. 2001;155(3):415-26. doi: 10.1083/jcb.200103078. PubMed PMID: 11684708; PubMed Central PMCID: PMC2150850.

60. Ku NO, Omary MB. A disease- and phosphorylation-related nonmechanical function for keratin 8. The Journal of cell biology. 2006;174(1):115-25. doi: 10.1083/jcb.200602146. PubMed PMID: 16818723; PubMed Central PMCID: PMC2064169.

61. Yang H, Wen YY, Zhao R, Lin YL, Fournier K, Yang HY, et al. DNA damage-induced protein 14-3-3 sigma inhibits protein kinase B/Akt activation and suppresses Akt-activated cancer. Cancer research. 2006;66(6):3096-105. doi: 10.1158/0008-5472.CAN-05-3620. PubMed PMID: 16540659.

62. Hermeking H, Lengauer C, Polyak K, He TC, Zhang L, Thiagalingam S, et al. 14-3-3 sigma is a p53-regulated inhibitor of G2/M progression. Molecular cell. 1997;1(1):3-11. PubMed PMID: 9659898.

79 63. Tata JR. Signalling through nuclear receptors. Nature reviews Molecular cell biology. 2002;3(9):702-10. doi: 10.1038/nrm914. PubMed PMID: 12209130.

64. Germain P, Staels B, Dacquet C, Spedding M, Laudet V. Overview of nomenclature of nuclear receptors. Pharmacological reviews. 2006;58(4):685-704. doi: 10.1124/pr.58.4.2. PubMed PMID: 17132848.

65. Sever R, Glass CK. Signaling by nuclear receptors. Cold Spring Harbor perspectives in biology. 2013;5(3):a016709. doi: 10.1101/cshperspect.a016709. PubMed PMID: 23457262; PubMed Central PMCID: PMC3578364.

66. Edwards DP. Regulation of signal transduction pathways by estrogen and progesterone. Annual review of physiology. 2005;67:335-76. doi: 10.1146/annurev.physiol.67.040403.120151. PubMed PMID: 15709962.

67. Nuclear Receptors Nomenclature C. A unified nomenclature system for the nuclear receptor superfamily. Cell. 1999;97(2):161-3. PubMed PMID: 10219237.

68. Lefebvre P, Benomar Y, Staels B. Retinoid X receptors: common heterodimerization partners with distinct functions. Trends in endocrinology and metabolism: TEM. 2010;21(11):676-83. doi: 10.1016/j.tem.2010.06.009. PubMed PMID: 20674387.

69. Kuiper GG, Enmark E, Pelto-Huikko M, Nilsson S, Gustafsson JA. Cloning of a novel receptor expressed in rat prostate and ovary. Proceedings of the National Academy of Sciences of the United States of America. 1996;93(12):5925-30. PubMed PMID: 8650195; PubMed Central PMCID: PMC39164.

70. Pawlak M, Lefebvre P, Staels B. General molecular biology and architecture of nuclear receptors. Current topics in medicinal chemistry. 2012;12(6):486-504. PubMed PMID: 22242852; PubMed Central PMCID: PMC3637177.

71. Cowley SM, Parker MG. A comparison of transcriptional activation by ER alpha and ER beta. The Journal of steroid biochemistry and molecular biology. 1999;69(1-6):165-75. PubMed PMID: 10418990.

72. McInerney EM, Weis KE, Sun J, Mosselman S, Katzenellenbogen BS. Transcription activation by the human estrogen receptor subtype beta (ER beta) studied with ER beta and ER alpha receptor chimeras. Endocrinology. 1998;139(11):4513-22. doi: 10.1210/endo.139.11.6298. PubMed PMID: 9794460.

73. Liang J, Shang Y. Estrogen and cancer. Annual review of physiology. 2013;75:225-40. doi: 10.1146/annurev-physiol-030212-183708. PubMed PMID: 23043248.

74. Berry NB, Fan M, Nephew KP. Estrogen receptor-alpha hinge-region lysines 302 and 303 regulate receptor degradation by the proteasome. Molecular endocrinology. 2008;22(7):1535-51. doi: 10.1210/me.2007-0449. PubMed PMID: 18388150; PubMed Central PMCID: PMC2453605.

80 75. Wang C, Fu M, Angeletti RH, Siconolfi-Baez L, Reutens AT, Albanese C, et al. Direct acetylation of the estrogen receptor alpha hinge region by p300 regulates transactivation and hormone sensitivity. The Journal of biological chemistry. 2001;276(21):18375-83. doi: 10.1074/jbc.M100800200. PubMed PMID: 11279135.

76. Acconcia F, Ascenzi P, Fabozzi G, Visca P, Marino M. S-palmitoylation modulates human estrogen receptor-alpha functions. Biochemical and biophysical research communications. 2004;316(3):878-83. doi: 10.1016/j.bbrc.2004.02.129. PubMed PMID: 15033483.

77. Le Romancer M, Treilleux I, Leconte N, Robin-Lespinasse Y, Sentis S, Bouchekioua- Bouzaghou K, et al. Regulation of estrogen rapid signaling through arginine methylation by PRMT1. Molecular cell. 2008;31(2):212-21. doi: 10.1016/j.molcel.2008.05.025. PubMed PMID: 18657504.

78. Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer discovery. 2012;2(5):401-4. doi: 10.1158/2159-8290.CD-12-0095. PubMed PMID: 22588877; PubMed Central PMCID: PMC3956037.

79. Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, et al. Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Science signaling. 2013;6(269):pl1. doi: 10.1126/scisignal.2004088. PubMed PMID: 23550210; PubMed Central PMCID: PMC4160307.

80. Caldon CE, Sutherland RL, Musgrove E. Cell cycle proteins in epithelial cell differentiation: implications for breast cancer. Cell cycle. 2010;9(10):1918-28. doi: 10.4161/cc.9.10.11474. PubMed PMID: 20473028.

81. Musgrove EA, Caldon CE, Barraclough J, Stone A, Sutherland RL. Cyclin D as a therapeutic target in cancer. Nature reviews Cancer. 2011;11(8):558-72. doi: 10.1038/nrc3090. PubMed PMID: 21734724.

82. Gompel A, Somai S, Chaouat M, Kazem A, Kloosterboer HJ, Beusman I, et al. Hormonal regulation of apoptosis in breast cells and tissues. Steroids. 2000;65(10-11):593-8. PubMed PMID: 11108864.

83. Leygue E, Murphy LC. A bi-faceted role of estrogen receptor beta in breast cancer. Endocrine-related cancer. 2013;20(3):R127-39. doi: 10.1530/ERC-12-0389. PubMed PMID: 23533249.

84. Fish EN. The X-files in immunity: sex-based differences predispose immune responses. Nature reviews Immunology. 2008;8(9):737-44. doi: 10.1038/nri2394. PubMed PMID: 18728636.

81 85. Kovats S. Estrogen receptors regulate an inflammatory pathway of dendritic cell differentiation: mechanisms and implications for immunity. Hormones and behavior. 2012;62(3):254-62. doi: 10.1016/j.yhbeh.2012.04.011. PubMed PMID: 22561458; PubMed Central PMCID: PMC3415586.

86. Kovats S. Estrogen receptors regulate innate immune cells and signaling pathways. Cellular immunology. 2015;294(2):63-9. doi: 10.1016/j.cellimm.2015.01.018. PubMed PMID: 25682174; PubMed Central PMCID: PMC4380804.

87. Miller VM, Duckles SP. Vascular actions of estrogens: functional implications. Pharmacological reviews. 2008;60(2):210-41. doi: 10.1124/pr.107.08002. PubMed PMID: 18579753; PubMed Central PMCID: PMC2637768.

88. Kastner P, Krust A, Turcotte B, Stropp U, Tora L, Gronemeyer H, et al. Two distinct estrogen-regulated promoters generate transcripts encoding the two functionally different human progesterone receptor forms A and B. The EMBO journal. 1990;9(5):1603-14. PubMed PMID: 2328727; PubMed Central PMCID: PMC551856.

89. Lieberman BA, Bona BJ, Edwards DP, Nordeen SK. The constitution of a progesterone response element. Molecular endocrinology. 1993;7(4):515-27. doi: 10.1210/mend.7.4.8388996. PubMed PMID: 8388996.

90. Clarke CL, Graham JD. Non-overlapping progesterone receptor cistromes contribute to cell-specific transcriptional outcomes. PloS one. 2012;7(4):e35859. doi: 10.1371/journal.pone.0035859. PubMed PMID: 22545144; PubMed Central PMCID: PMC3335806.

91. Jacobsen BM, Schittone SA, Richer JK, Horwitz KB. Progesterone-independent effects of human progesterone receptors (PRs) in estrogen receptor-positive breast cancer: PR isoform- specific gene regulation and tumor biology. Molecular endocrinology. 2005;19(3):574-87. doi: 10.1210/me.2004-0287. PubMed PMID: 15563544.

92. Mulac-Jericevic B, Lydon JP, DeMayo FJ, Conneely OM. Defective mammary gland morphogenesis in mice lacking the progesterone receptor B isoform. Proceedings of the National Academy of Sciences of the United States of America. 2003;100(17):9744-9. doi: 10.1073/pnas.1732707100. PubMed PMID: 12897242; PubMed Central PMCID: PMC187836.

93. Mulac-Jericevic B, Mullinax RA, DeMayo FJ, Lydon JP, Conneely OM. Subgroup of reproductive functions of progesterone mediated by progesterone receptor-B isoform. Science. 2000;289(5485):1751-4. PubMed PMID: 10976068.

94. Hagan CR, Daniel AR, Dressing GE, Lange CA. Role of phosphorylation in progesterone receptor signaling and specificity. Molecular and cellular endocrinology. 2012;357(1-2):43-9. doi: 10.1016/j.mce.2011.09.017. PubMed PMID: 21945472; PubMed Central PMCID: PMC3265648.

82 95. Hagan CR, Lange CA. Molecular determinants of context-dependent progesterone receptor action in breast cancer. BMC medicine. 2014;12:32. doi: 10.1186/1741-7015-12-32. PubMed PMID: 24552158; PubMed Central PMCID: PMC3929904.

96. Carroll JS, Hickey TE, Tarulli GA, Williams M, Tilley WD. Deciphering the divergent roles of progestogens in breast cancer. Nature reviews Cancer. 2017;17(1):54-64. doi: 10.1038/nrc.2016.116. PubMed PMID: 27885264.

97. Beral V, Million Women Study C. Breast cancer and hormone-replacement therapy in the Million Women Study. Lancet. 2003;362(9382):419-27. PubMed PMID: 12927427.

98. Chlebowski RT, Hendrix SL, Langer RD, Stefanick ML, Gass M, Lane D, et al. Influence of estrogen plus progestin on breast cancer and mammography in healthy postmenopausal women: the Women's Health Initiative Randomized Trial. Jama. 2003;289(24):3243-53. doi: 10.1001/jama.289.24.3243. PubMed PMID: 12824205.

99. Rossouw JE, Anderson GL, Prentice RL, LaCroix AZ, Kooperberg C, Stefanick ML, et al. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results From the Women's Health Initiative randomized controlled trial. Jama. 2002;288(3):321- 33. PubMed PMID: 12117397.

100. Grimm SL, Hartig SM, Edwards DP. Progesterone Receptor Signaling Mechanisms. Journal of molecular biology. 2016;428(19):3831-49. doi: 10.1016/j.jmb.2016.06.020. PubMed PMID: 27380738.

101. Hill KK, Roemer SC, Churchill ME, Edwards DP. Structural and functional analysis of domains of the progesterone receptor. Molecular and cellular endocrinology. 2012;348(2):418- 29. doi: 10.1016/j.mce.2011.07.017. PubMed PMID: 21803119; PubMed Central PMCID: PMC4437577.

102. Roemer SC, Donham DC, Sherman L, Pon VH, Edwards DP, Churchill ME. Structure of the progesterone receptor-deoxyribonucleic acid complex: novel interactions required for binding to half-site response elements. Molecular endocrinology. 2006;20(12):3042-52. doi: 10.1210/me.2005-0511. PubMed PMID: 16931575; PubMed Central PMCID: PMC2532839.

103. Bernstein L. Epidemiology of endocrine-related risk factors for breast cancer. Journal of mammary gland biology and neoplasia. 2002;7(1):3-15. PubMed PMID: 12160084.

104. Truong TH, Dwyer AR, Diep CH, Hu H, Hagen KM, Lange CA. Phosphorylated Progesterone Receptor Isoforms Mediate Opposing Stem Cell and Proliferative Breast Cancer Cell Fates. Endocrinology. 2019;160(2):430-46. doi: 10.1210/en.2018-00990. PubMed PMID: 30597041; PubMed Central PMCID: PMC6349004.

105. Escriva H, Bertrand S, Germain P, Robinson-Rechavi M, Umbhauer M, Cartry J, et al. Neofunctionalization in vertebrates: the example of retinoic acid receptors. PLoS genetics. 2006;2(7):e102. doi: 10.1371/journal.pgen.0020102. PubMed PMID: 16839186; PubMed Central PMCID: PMC1500811.

83 106. di Masi A, Leboffe L, De Marinis E, Pagano F, Cicconi L, Rochette-Egly C, et al. Retinoic acid receptors: from molecular mechanisms to cancer therapy. Molecular aspects of medicine. 2015;41:1-115. doi: 10.1016/j.mam.2014.12.003. PubMed PMID: 25543955.

107. Aranda A, Pascual A. Nuclear hormone receptors and gene expression. Physiological reviews. 2001;81(3):1269-304. doi: 10.1152/physrev.2001.81.3.1269. PubMed PMID: 11427696.

108. Glass CK, Rosenfeld MG. The coregulator exchange in transcriptional functions of nuclear receptors. Genes & development. 2000;14(2):121-41. PubMed PMID: 10652267.

109. Germain P, Iyer J, Zechel C, Gronemeyer H. Co-regulator recruitment and the mechanism of retinoic acid receptor synergy. Nature. 2002;415(6868):187-92. doi: 10.1038/415187a. PubMed PMID: 11805839.

110. Ross-Innes CS, Stark R, Holmes KA, Schmidt D, Spyrou C, Russell R, et al. Cooperative interaction between retinoic acid receptor-alpha and estrogen receptor in breast cancer. Genes & development. 2010;24(2):171-82. doi: 10.1101/gad.552910. PubMed PMID: 20080953; PubMed Central PMCID: PMC2807352.

111. Alsayed Y, Uddin S, Mahmud N, Lekmine F, Kalvakolanu DV, Minucci S, et al. Activation of Rac1 and the p38 mitogen-activated protein kinase pathway in response to all- trans-retinoic acid. The Journal of biological chemistry. 2001;276(6):4012-9. doi: 10.1074/jbc.M007431200. PubMed PMID: 11060298.

112. Masia S, Alvarez S, de Lera AR, Barettino D. Rapid, nongenomic actions of retinoic acid on phosphatidylinositol-3-kinase signaling pathway mediated by the retinoic acid receptor. Molecular endocrinology. 2007;21(10):2391-402. doi: 10.1210/me.2007-0062. PubMed PMID: 17595318.

113. Tang XH, Gudas LJ. Retinoids, retinoic acid receptors, and cancer. Annu Rev Pathol. 2011;6:345-64. Epub 2010/11/16. doi: 10.1146/annurev-pathol-011110-130303. PubMed PMID: 21073338.

114. Connolly RM, Nguyen NK, Sukumar S. Molecular pathways: current role and future directions of the retinoic acid pathway in cancer prevention and treatment. Clinical cancer research : an official journal of the American Association for Cancer Research. 2013;19(7):1651- 9. doi: 10.1158/1078-0432.CCR-12-3175. PubMed PMID: 23322901; PubMed Central PMCID: PMC3618522.

115. Khuri FR, Lee JJ, Lippman SM, Kim ES, Cooper JS, Benner SE, et al. Randomized phase III trial of low-dose isotretinoin for prevention of second primary tumors in stage I and II head and neck cancer patients. Journal of the National Cancer Institute. 2006;98(7):441-50. doi: 10.1093/jnci/djj091. PubMed PMID: 16595780.

116. Alpha-Tocopherol BCCPSG. The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers. The New England journal of medicine. 1994;330(15):1029-35. doi: 10.1056/NEJM199404143301501. PubMed PMID: 8127329.

84 117. Lee JS, Lippman SM, Benner SE, Lee JJ, Ro JY, Lukeman JM, et al. Randomized placebo-controlled trial of isotretinoin in chemoprevention of bronchial squamous metaplasia. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 1994;12(5):937-45. doi: 10.1200/JCO.1994.12.5.937. PubMed PMID: 8164045.

118. Lin LM, Li BX, Xiao JB, Lin DH, Yang BF. Synergistic effect of all-trans-retinoic acid and arsenic trioxide on growth inhibition and apoptosis in human hepatoma, breast cancer, and lung cancer cells in vitro. World journal of gastroenterology. 2005;11(36):5633-7. PubMed PMID: 16237756; PubMed Central PMCID: PMC4481479.

119. Decensi A, Robertson C, Guerrieri-Gonzaga A, Serrano D, Cazzaniga M, Mora S, et al. Randomized double-blind 2 x 2 trial of low-dose tamoxifen and fenretinide for breast cancer prevention in high-risk premenopausal women. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2009;27(23):3749-56. doi: 10.1200/JCO.2008.19.3797. PubMed PMID: 19597031; PubMed Central PMCID: PMC2799048.

120. Veronesi U, De Palo G, Marubini E, Costa A, Formelli F, Mariani L, et al. Randomized trial of fenretinide to prevent second breast malignancy in women with early breast cancer. Journal of the National Cancer Institute. 1999;91(21):1847-56. PubMed PMID: 10547391.

121. Cassidy J, Lippman M, Lacroix A, Peck G. Phase II trial of 13-cis-retinoic acid in metastatic breast cancer. European journal of cancer & clinical oncology. 1982;18(10):925-8. PubMed PMID: 6962067.

122. Lawrence JA, Adamson PC, Caruso R, Chow C, Kleiner D, Murphy RF, et al. Phase I clinical trial of alitretinoin and tamoxifen in breast cancer patients: toxicity, pharmacokinetic, and biomarker evaluations. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2001;19(10):2754-63. doi: 10.1200/JCO.2001.19.10.2754. PubMed PMID: 11352969.

123. Wu J, Yang R, Zhang L, Li Y, Liu B, Kang H, et al. Metabolomics research on potential role for 9-cis-retinoic acid in breast cancer progression. Cancer science. 2018;109(7):2315-26. doi: 10.1111/cas.13629. PubMed PMID: 29737597; PubMed Central PMCID: PMC6029828.

124. Allegra JC, Lippman ME, Thompson EB, Simon R, Barlock A, Green L, et al. Relationship between the progesterone, androgen, and glucocorticoid receptor and response rate to endocrine therapy in metastatic breast cancer. Cancer research. 1979;39(6 Pt 1):1973-9. PubMed PMID: 445396.

125. Horwitz KB, Freidenberg GR. Growth inhibition and increase of insulin receptors in antiestrogen-resistant T47DCO human breast cancer cells by progestins: implications for endocrine therapies. Cancer research. 1985;45(1):167-73. PubMed PMID: 3838085.

85 126. Hilton HN, Doan TB, Graham JD, Oakes SR, Silvestri A, Santucci N, et al. Acquired convergence of hormone signaling in breast cancer: ER and PR transition from functionally distinct in normal breast to predictors of metastatic disease. Oncotarget. 2014;5(18):8651-64. doi: 10.18632/oncotarget.2354. PubMed PMID: 25261374; PubMed Central PMCID: PMC4226711.

127. Daniel AR, Gaviglio AL, Knutson TP, Ostrander JH, D'Assoro AB, Ravindranathan P, et al. Progesterone receptor-B enhances estrogen responsiveness of breast cancer cells via scaffolding PELP1- and estrogen receptor-containing transcription complexes. Oncogene. 2015;34(4):506-15. doi: 10.1038/onc.2013.579. PubMed PMID: 24469035; PubMed Central PMCID: PMC4112172.

128. Giulianelli S, Vaque JP, Soldati R, Wargon V, Vanzulli SI, Martins R, et al. Estrogen receptor alpha mediates progestin-induced mammary tumor growth by interacting with progesterone receptors at the cyclin D1/MYC promoters. Cancer research. 2012;72(9):2416-27. doi: 10.1158/0008-5472.CAN-11-3290. PubMed PMID: 22396492.

129. Mohammed H, Russell IA, Stark R, Rueda OM, Hickey TE, Tarulli GA, et al. Progesterone receptor modulates ERalpha action in breast cancer. Nature. 2015;523(7560):313- 7. doi: 10.1038/nature14583. PubMed PMID: 26153859.

130. Finlay-Schultz J, Gillen AE, Brechbuhl HM, Ivie JJ, Matthews SB, Jacobsen BM, et al. Breast Cancer Suppression by Progesterone Receptors Is Mediated by Their Modulation of Estrogen Receptors and RNA Polymerase III. Cancer research. 2017;77(18):4934-46. doi: 10.1158/0008-5472.CAN-16-3541. PubMed PMID: 28729413; PubMed Central PMCID: PMC5600857.

131. Need EF, Selth LA, Harris TJ, Birrell SN, Tilley WD, Buchanan G. Research resource: interplay between the genomic and transcriptional networks of androgen receptor and estrogen receptor alpha in luminal breast cancer cells. Molecular endocrinology. 2012;26(11):1941-52. doi: 10.1210/me.2011-1314. PubMed PMID: 23023562; PubMed Central PMCID: PMC3487621.

132. Peters AA, Buchanan G, Ricciardelli C, Bianco-Miotto T, Centenera MM, Harris JM, et al. Androgen receptor inhibits estrogen receptor-alpha activity and is prognostic in breast cancer. Cancer research. 2009;69(15):6131-40. doi: 10.1158/0008-5472.CAN-09-0452. PubMed PMID: 19638585.

133. D'Amato NC, Gordon MA, Babbs B, Spoelstra NS, Carson Butterfield KT, Torkko KC, et al. Cooperative Dynamics of AR and ER Activity in Breast Cancer. Molecular cancer research : MCR. 2016;14(11):1054-67. doi: 10.1158/1541-7786.MCR-16-0167. PubMed PMID: 27565181; PubMed Central PMCID: PMC5107172.

134. Leo JC, Guo C, Woon CT, Aw SE, Lin VC. Glucocorticoid and mineralocorticoid cross- talk with progesterone receptor to induce focal adhesion and growth inhibition in breast cancer cells. Endocrinology. 2004;145(3):1314-21. doi: 10.1210/en.2003-0732. PubMed PMID: 14617569.

86 135. Frasor J, Danes JM, Komm B, Chang KC, Lyttle CR, Katzenellenbogen BS. Profiling of estrogen up- and down-regulated gene expression in human breast cancer cells: insights into gene networks and pathways underlying estrogenic control of proliferation and cell phenotype. Endocrinology. 2003;144(10):4562-74. doi: 10.1210/en.2003-0567. PubMed PMID: 12959972.

136. Axlund SD, Yoo BH, Rosen RB, Schaack J, Kabos P, Labarbera DV, et al. Progesterone- inducible cytokeratin 5-positive cells in luminal breast cancer exhibit progenitor properties. Hormones & cancer. 2013;4(1):36-49. doi: 10.1007/s12672-012-0127-5. PubMed PMID: 23184698; PubMed Central PMCID: PMC3549640.

137. Kabos P, Finlay-Schultz J, Li C, Kline E, Finlayson C, Wisell J, et al. Patient-derived luminal breast cancer xenografts retain hormone receptor heterogeneity and help define unique estrogen-dependent gene signatures. Breast cancer research and treatment. 2012;135(2):415-32. doi: 10.1007/s10549-012-2164-8. PubMed PMID: 22821401; PubMed Central PMCID: PMC3818141.

138. Finlay-Schultz J, Cittelly DM, Hendricks P, Patel P, Kabos P, Jacobsen BM, et al. Progesterone downregulation of miR-141 contributes to expansion of stem-like breast cancer cells through maintenance of progesterone receptor and Stat5a. Oncogene. 2015;34(28):3676-87. Epub 2014/09/23. doi: 10.1038/onc.2014.298. PubMed PMID: 25241899; PubMed Central PMCID: PMCPMC4369481.

139. Pfaffl MW. A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Res. 2001;29(9):e45. PubMed PMID: 11328886; PubMed Central PMCID: PMCPMC55695.

140. Martin M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnetjournal. 2011;17(1):10-2.

141. Kim D, Langmead B, Salzberg SL. HISAT: a fast spliced aligner with low memory requirements. Nature methods. 2015;12(4):357-60. doi: 10.1038/nmeth.3317. PubMed PMID: 25751142; PubMed Central PMCID: PMC4655817.

142. Liao Y, Smyth GK, Shi W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics. 2014;30(7):923-30. doi: 10.1093/bioinformatics/btt656. PubMed PMID: 24227677.

143. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic acids research. 2015;43(7):e47. doi: 10.1093/nar/gkv007. PubMed PMID: 25605792; PubMed Central PMCID: PMC4402510.

144. Law CW, Chen Y, Shi W, Smyth GK. voom: Precision weights unlock linear model analysis tools for RNA-seq read counts. Genome biology. 2014;15(2):R29. doi: 10.1186/gb- 2014-15-2-r29. PubMed PMID: 24485249; PubMed Central PMCID: PMC4053721.

87 145. Hu Y, Smyth GK. ELDA: extreme limiting dilution analysis for comparing depleted and enriched populations in stem cell and other assays. J Immunol Methods. 2009;347(1-2):70-8. Epub 2009/07/02. doi: 10.1016/j.jim.2009.06.008. PubMed PMID: 19567251.

146. Cancer Genome Atlas N. Comprehensive molecular portraits of human breast tumours. Nature. 2012;490(7418):61-70. doi: 10.1038/nature11412. PubMed PMID: 23000897; PubMed Central PMCID: PMCPMC3465532.

147. Osborne CK, Schiff R. Mechanisms of endocrine resistance in breast cancer. Annual review of medicine. 2011;62:233-47. doi: 10.1146/annurev-med-070909-182917. PubMed PMID: 20887199; PubMed Central PMCID: PMC3656649.

148. Allan AL, Vantyghem SA, Tuck AB, Chambers AF. Tumor dormancy and cancer stem cells: implications for the biology and treatment of breast cancer metastasis. Breast Dis. 2006;26:87-98. Epub 2007/05/03. PubMed PMID: 17473368.

149. Guedj M, Marisa L, de Reynies A, Orsetti B, Schiappa R, Bibeau F, et al. A refined molecular taxonomy of breast cancer. Oncogene. 2012;31(9):1196-206. Epub 2011/07/26. doi: 10.1038/onc.2011.301. PubMed PMID: 21785460.

150. Reya T, Morrison SJ, Clarke MF, Weissman IL. Stem cells, cancer, and cancer stem cells. Nature. 2001;414(6859):105-11. PubMed PMID: 11689955.

151. Chaffer CL, Brueckmann I, Scheel C, Kaestli AJ, Wiggins PA, Rodrigues LO, et al. Normal and neoplastic nonstem cells can spontaneously convert to a stem-like state. Proc Natl Acad Sci U S A. 2011;108(19):7950-5. Epub 2011/04/19. doi: 1102454108 [pii] 10.1073/pnas.1102454108. PubMed PMID: 21498687; PubMed Central PMCID: PMC3093533.

152. Iliopoulos D, Hirsch HA, Wang G, Struhl K. Inducible formation of breast cancer stem cells and their dynamic equilibrium with non-stem cancer cells via IL6 secretion. Proc Natl Acad Sci U S A. 2011;108(4):1397-402. Epub 2011/01/12. doi: 1018898108 [pii] 10.1073/pnas.1018898108. PubMed PMID: 21220315; PubMed Central PMCID: PMC3029760.

153. Osborne CK, Schiff R, Arpino G, Lee AS, Hilsenbeck VG. Endocrine responsiveness: understanding how progesterone receptor can be used to select endocrine therapy. Breast. 2005;14(6):458-65. Epub 2005/10/21. doi: S0960-9776(05)00190-6 [pii] 10.1016/j.breast.2005.08.024. PubMed PMID: 16236516.

154. Knutson TP, Daniel AR, Fan D, Silverstein KA, Covington KR, Fuqua SA, et al. Phosphorylated and sumoylation-deficient progesterone receptors drive proliferative gene signatures during breast cancer progression. Breast Cancer Res. 2012;14(3):R95. Epub 2012/06/16. doi: 10.1186/bcr3211. PubMed PMID: 22697792.

155. Singhal H, Greene ME, Tarulli G, Zarnke AL, Bourgo RJ, Laine M, et al. Genomic agonism and phenotypic antagonism between estrogen and progesterone receptors in breast cancer. Sci Adv. 2016;2(6):e1501924. Epub 2016/07/08. doi: 10.1126/sciadv.1501924. PubMed PMID: 27386569.

88 156. Horwitz KB, Dye WW, Harrell JC, Kabos P, Sartorius CA. Rare steroid receptor- negative basal-like tumorigenic cells in luminal subtype human breast cancer xenografts. Proc Natl Acad Sci U S A. 2008;105(15):5774-9. Epub 2008/04/09. doi: 10.1073/pnas.0706216105. PubMed PMID: 18391223.

157. Lim E, Vaillant F, Wu D, Forrest NC, Pal B, Hart AH, et al. Aberrant luminal progenitors as the candidate target population for basal tumor development in BRCA1 mutation carriers. Nat Med. 2009;15(8):907-13. Epub 2009/08/04. doi: 10.1038/nm.2000. PubMed PMID: 19648928.

158. Knox AJ, Scaling AL, Pinto MP, Bliesner BS, Haughian JM, Abdel-Hafiz HA, et al. Modeling luminal breast cancer heterogeneity: combination therapy to suppress a hormone receptor-negative, cytokeratin 5-positive subpopulation in luminal disease. Breast Cancer Res. 2014;16(4):418. Epub 2014/08/15. doi: 10.1186/s13058-014-0418-6. PubMed PMID: 25116921.

159. Cittelly DM, Finlay-Schultz J, Howe EN, Spoelstra NS, Axlund SD, Hendricks P, et al. Progestin suppression of miR-29 potentiates dedifferentiation of breast cancer cells via KLF4. Oncogene. 2013;32(20):2555-64. Epub 2012/07/04. doi: 10.1038/onc.2012.275. PubMed PMID: 22751119.

160. Lersch R, Stellmach V, Stocks C, Giudice G, Fuchs E. Isolation, sequence, and expression of a human keratin K5 gene: transcriptional regulation of keratins and insights into pairwise control. Molecular and cellular biology. 1989;9(9):3685-97. PubMed PMID: 2476664; PubMed Central PMCID: PMC362429.

161. Goodman CR, Sato T, Peck AR, Girondo MA, Yang N, Liu C, et al. Steroid induction of therapy-resistant cytokeratin-5-positive cells in estrogen receptor-positive breast cancer through a BCL6-dependent mechanism. Oncogene. 2016;35(11):1373-85. Epub 2015/06/23. doi: 10.1038/onc.2015.193. PubMed PMID: 26096934; PubMed Central PMCID: PMCPMC4800289.

162. Dontu G, Abdallah WM, Foley JM, Jackson KW, Clarke MF, Kawamura MJ, et al. In vitro propagation and transcriptional profiling of human mammary stem/progenitor cells. Genes & development. 2003;17(10):1253-70. doi: 10.1101/gad.1061803. PubMed PMID: 12756227; PubMed Central PMCID: PMC196056.

163. Yoo BH, Axlund SD, Kabos P, Reid BG, Schaack J, Sartorius CA, et al. A high-content assay to identify small-molecule modulators of a cancer stem cell population in luminal breast cancer. Journal of biomolecular screening. 2012;17(9):1211-20. doi: 10.1177/1087057112452138. PubMed PMID: 22751729; PubMed Central PMCID: PMC4158006.

164. le Maire A, Alvarez S, Shankaranarayanan P, Lera AR, Bourguet W, Gronemeyer H. Retinoid receptors and therapeutic applications of RAR/RXR modulators. Curr Top Med Chem. 2012;12(6):505-27. Epub 2012/01/17. PubMed PMID: 22242853.

165. Perissi V, Rosenfeld MG. Controlling nuclear receptors: the circular logic of cofactor cycles. Nature reviews Molecular cell biology. 2005;6(7):542-54. doi: 10.1038/nrm1682. PubMed PMID: 15957004.

89 166. Yang F, Ma Q, Liu Z, Li W, Tan Y, Jin C, et al. Glucocorticoid Receptor:MegaTrans Switching Mediates the Repression of an ERalpha-Regulated Transcriptional Program. Molecular cell. 2017;66(3):321-31 e6. doi: 10.1016/j.molcel.2017.03.019. PubMed PMID: 28475868; PubMed Central PMCID: PMC5510478.

167. Gianni M, Kalac Y, Ponzanelli I, Rambaldi A, Terao M, Garattini E. Tyrosine kinase inhibitor STI571 potentiates the pharmacologic activity of retinoic acid in acute promyelocytic leukemia cells: effects on the degradation of RARalpha and PML-RARalpha. Blood. 2001;97(10):3234-43. Epub 2001/05/09. PubMed PMID: 11342454.

168. Wu K, Kim HT, Rodriquez JL, Hilsenbeck SG, Mohsin SK, Xu XC, et al. Suppression of mammary tumorigenesis in transgenic mice by the RXR-selective retinoid, LGD1069. Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology. 2002;11(5):467-74. PubMed PMID: 12010861.

169. Rubin M, Fenig E, Rosenauer A, Menendez-Botet C, Achkar C, Bentel JM, et al. 9-Cis retinoic acid inhibits growth of breast cancer cells and down-regulates estrogen receptor RNA and protein. Cancer research. 1994;54(24):6549-56. PubMed PMID: 7987855.

170. Garattini E, Bolis M, Garattini SK, Fratelli M, Centritto F, Paroni G, et al. Retinoids and breast cancer: from basic studies to the clinic and back again. Cancer Treat Rev. 2014;40(6):739- 49. Epub 2014/02/01. doi: 10.1016/j.ctrv.2014.01.001. PubMed PMID: 24480385.

171. Veronesi U, Mariani L, Decensi A, Formelli F, Camerini T, Miceli R, et al. Fifteen-year results of a randomized phase III trial of fenretinide to prevent second breast cancer. Ann Oncol. 2006;17(7):1065-71. Epub 2006/05/06. doi: 10.1093/annonc/mdl047. PubMed PMID: 16675486.

172. Reid BG, Jerjian T, Patel P, Zhou Q, Yoo BH, Kabos P, et al. Live multicellular tumor spheroid models for high-content imaging and screening in cancer drug discovery. Curr Chem Genomics Transl Med. 2014;8(Suppl 1):27-35. Epub 2014/03/07. doi: 10.2174/2213988501408010027. PubMed PMID: 24596682.

173. Widschwendter M, Berger J, Daxenbichler G, Muller-Holzner E, Widschwendter A, Mayr A, et al. Loss of retinoic acid receptor beta expression in breast cancer and morphologically normal adjacent tissue but not in the normal breast tissue distant from the cancer. Cancer Res. 1997;57(19):4158-61. Epub 1997/10/23. PubMed PMID: 9331065.

174. Jho SH, Radoja N, Im MJ, Tomic-Canic M. Negative response elements in keratin genes mediate transcriptional repression and the cross-talk among nuclear receptors. J Biol Chem. 2001;276(49):45914-20. Epub 2001/10/10. doi: 10.1074/jbc.M103144200. PubMed PMID: 11591699.

175. Radoja N, Diaz DV, Minars TJ, Freedberg IM, Blumenberg M, Tomic-Canic M. Specific organization of the negative response elements for retinoic acid and thyroid hormone receptors in keratin gene family. The Journal of investigative dermatology. 1997;109(4):566-72. PubMed PMID: 9326392.

90 176. Radoja N, Komine M, Jho SH, Blumenberg M, Tomic-Canic M. Novel mechanism of steroid action in skin through glucocorticoid receptor monomers. Mol Cell Biol. 2000;20(12):4328-39. PubMed PMID: 10825196; PubMed Central PMCID: PMCPMC85800.

177. Long MD, Singh PK, Russell JR, Llimos G, Rosario S, Rizvi A, et al. The miR-96 and RARgamma signaling axis governs androgen signaling and prostate cancer progression. Oncogene. 2019;38(3):421-44. doi: 10.1038/s41388-018-0450-6. PubMed PMID: 30120411; PubMed Central PMCID: PMC6336686.

178. Anzano MA, Byers SW, Smith JM, Peer CW, Mullen LT, Brown CC, et al. Prevention of breast cancer in the rat with 9-cis-retinoic acid as a single agent and in combination with tamoxifen. Cancer research. 1994;54(17):4614-7. PubMed PMID: 8062253.

179. Moll R, Franke WW, Schiller DL, Geiger B, Krepler R. The catalog of human cytokeratins: patterns of expression in normal epithelia, tumors and cultured cells. Cell. 1982;31(1):11-24. PubMed PMID: 6186379.

180. Chung BM, Rotty JD, Coulombe PA. Networking galore: intermediate filaments and cell migration. Curr Opin Cell Biol. 2013;25(5):600-12. doi: 10.1016/j.ceb.2013.06.008. PubMed PMID: 23886476; PubMed Central PMCID: PMCPMC3780586.

181. Kim S, Wong P, Coulombe PA. A keratin cytoskeletal protein regulates protein synthesis and epithelial cell growth. Nature. 2006;441(7091):362-5. Epub 2006/05/20. doi: 10.1038/nature04659. PubMed PMID: 16710422.

182. Seltmann K, Fritsch AW, Kas JA, Magin TM. Keratins significantly contribute to cell stiffness and impact invasive behavior. Proc Natl Acad Sci U S A. 2013;110(46):18507-12. doi: 10.1073/pnas.1310493110. PubMed PMID: 24167274; PubMed Central PMCID: PMCPMC3832002.

183. Toivola DM, Strnad P, Habtezion A, Omary MB. Intermediate filaments take the heat as stress proteins. Trends Cell Biol. 2010;20(2):79-91. doi: 10.1016/j.tcb.2009.11.004. PubMed PMID: 20045331; PubMed Central PMCID: PMCPMC2843093.

184. Boudreau A, Tanner K, Wang D, Geyer FC, Reis-Filho JS, Bissell MJ. 14-3-3sigma stabilizes a complex of soluble actin and intermediate filament to enable breast tumor invasion. Proceedings of the National Academy of Sciences of the United States of America. 2013;110(41):E3937-44. doi: 10.1073/pnas.1315022110. PubMed PMID: 24067649; PubMed Central PMCID: PMC3799319.

185. Asselin-Labat ML, Vaillant F, Sheridan JM, Pal B, Wu D, Simpson ER, et al. Control of mammary stem cell function by steroid hormone signalling. Nature. 2010;465(7299):798-802. doi: 10.1038/nature09027. PubMed PMID: 20383121.

186. Graham JD, Mote PA, Salagame U, van Dijk JH, Balleine RL, Huschtscha LI, et al. DNA replication licensing and progenitor numbers are increased by progesterone in normal human breast. Endocrinology. 2009;150(7):3318-26. doi: 10.1210/en.2008-1630. PubMed PMID: 19342456; PubMed Central PMCID: PMC2703536.

91 187. Joshi PA, Jackson HW, Beristain AG, Di Grappa MA, Mote PA, Clarke CL, et al. Progesterone induces adult mammary stem cell expansion. Nature. 2010;465(7299):803-7. doi: 10.1038/nature09091. PubMed PMID: 20445538.

188. Mangelsdorf DJ, Thummel C, Beato M, Herrlich P, Schutz G, Umesono K, et al. The nuclear receptor superfamily: the second decade. Cell. 1995;83(6):835-9. Epub 1995/12/15. PubMed PMID: 8521507.

189. Clarke CL, Roman SD, Graham J, Koga M, Sutherland RL. Progesterone receptor regulation by retinoic acid in the human breast cancer cell line T-47D. J Biol Chem. 1990;265(21):12694-700. Epub 1990/07/25. PubMed PMID: 2373707.

190. Clarke CL, Graham J, Roman SD, Sutherland RL. Direct transcriptional regulation of the progesterone receptor by retinoic acid diminishes progestin responsiveness in the breast cancer cell line T-47D. J Biol Chem. 1991;266(28):18969-75. Epub 1991/10/05. PubMed PMID: 1918012.

191. Roman SD, Clarke CL, Hall RE, Alexander IE, Sutherland RL. Expression and regulation of retinoic acid receptors in human breast cancer cells. Cancer research. 1992;52(8):2236-42. PubMed PMID: 1313739.

192. Choi I, Gudas LJ, Katzenellenbogen BS. Regulation of keratin 19 gene expression by estrogen in human breast cancer cells and identification of the estrogen responsive gene region. Mol Cell Endocrinol. 2000;164(1-2):225-37. Epub 2000/10/12. PubMed PMID: 11026574.

193. Bhat-Nakshatri P, Goswami CP, Badve S, Sledge GW, Jr., Nakshatri H. Identification of FDA-approved drugs targeting breast cancer stem cells along with biomarkers of sensitivity. Sci Rep. 2013;3:2530. doi: 10.1038/srep02530. PubMed PMID: 23982413; PubMed Central PMCID: PMCPMC3965360.

194. Horwitz KB, Sartorius CA. Progestins in hormone replacement therapies reactivate cancer stem cells in women with preexisting breast cancers: a hypothesis. The Journal of clinical endocrinology and metabolism. 2008;93(9):3295-8. doi: 10.1210/jc.2008-0938. PubMed PMID: 18647813; PubMed Central PMCID: PMC2567860.

195. De Luca A, Fiorillo M, Peiris-Pages M, Ozsvari B, Smith DL, Sanchez-Alvarez R, et al. Mitochondrial biogenesis is required for the anchorage-independent survival and propagation of stem-like cancer cells. Oncotarget. 2015;6(17):14777-95. doi: 10.18632/oncotarget.4401. PubMed PMID: 26087310; PubMed Central PMCID: PMC4558115.

196. Schell JC, Olson KA, Jiang L, Hawkins AJ, Van Vranken JG, Xie J, et al. A role for the mitochondrial pyruvate carrier as a repressor of the Warburg effect and colon cancer cell growth. Molecular cell. 2014;56(3):400-13. doi: 10.1016/j.molcel.2014.09.026. PubMed PMID: 25458841; PubMed Central PMCID: PMC4268416.

92 197. Viale A, Pettazzoni P, Lyssiotis CA, Ying H, Sanchez N, Marchesini M, et al. Oncogene ablation-resistant pancreatic cancer cells depend on mitochondrial function. Nature. 2014;514(7524):628-32. doi: 10.1038/nature13611. PubMed PMID: 25119024; PubMed Central PMCID: PMC4376130.

198. Zhou Y, Zhou Y, Shingu T, Feng L, Chen Z, Ogasawara M, et al. Metabolic alterations in highly tumorigenic glioblastoma cells: preference for hypoxia and high dependency on glycolysis. The Journal of biological chemistry. 2011;286(37):32843-53. doi: 10.1074/jbc.M111.260935. PubMed PMID: 21795717; PubMed Central PMCID: PMC3173179.

199. Nam K, Oh S, Shin I. Ablation of CD44 induces glycolysis-to-oxidative phosphorylation transition via modulation of the c-Src-Akt-LKB1-AMPKalpha pathway. The Biochemical journal. 2016;473(19):3013-30. doi: 10.1042/BCJ20160613. PubMed PMID: 27458252.

200. Shakya A, Cooksey R, Cox JE, Wang V, McClain DA, Tantin D. Oct1 loss of function induces a coordinate metabolic shift that opposes tumorigenicity. Nature cell biology. 2009;11(3):320-7. doi: 10.1038/ncb1840. PubMed PMID: 19219035.

201. Kroenke CH, Michael YL, Poole EM, Kwan ML, Nechuta S, Leas E, et al. Postdiagnosis social networks and breast cancer mortality in the After Breast Cancer Pooling Project. Cancer. 2017;123(7):1228-37. doi: 10.1002/cncr.30440. PubMed PMID: 27943274; PubMed Central PMCID: PMC5360517.

93 APPENDIX A

Colorado Clinical and Translational Sciences Institute Shadowing Summary

As part of the requirements for grant funding received through the Colorado Clinical and

Translational Sciences Institute (CCTSI), I attended Breast Cancer Grand Rounds and shadowed

Dr. Peter Kabos in the clinic. The experience was enlightening both in how the various physicians interact and decide on the best course of treatment, as well as in regards to the issues that patients face during their treatments.

During Grand Rounds, pathologists, medical oncologist, surgeons, and radiation oncologists discuss patient cases that are not necessarily straightforward. By analyzing several factors such as tumor size, tumor grade, receptor status (ER, PR, HER2), lymph node or other metastatic involvement, and overall health, the physicians work to determine the best course of treatment for each patient. It is captivating to listen to the discussion that goes into these decisions and it is easy to see that, while they may not always be incomplete agreement on the course of action, all of the physicians are passionate about their jobs and doing what they think is best for the patient. One of the fascinating things about being at a research institution is that, during these discussions, it is clear that all of the physicians stay on top of the latest clinical trials, which are often thoroughly discussed during the course of the meetings. Many times strengths and weakness of the clinical trials are brought to light that may not have been easily evident.

As I shadowed in the clinic, I was impressed with the amount of communication among physicians, as each newly diagnosed patient sees a clinician from each discipline. While in the room with patients, I was able to observe how important it is to give the patients a safe platform to discuss their feelings and concerns. For example, patients will occasionally come in to get a

94 second opinion and during these appointments it is usually quite clear that trust has not been

developed with their original oncologist. Working to allow patients to feel heard and respected is a great way to make sure that they fully understand their options and are comfortable with treatment decisions. It also seems to drastically reduce the amount of anxiety the patients may feel.

Beyond the medical challenges that physicians and patients face while deciding on treatment, there are several other factors at play. Noncompliance due to not having reliable transport to receive treatment is, unfortunately, something that patients can face. One patient in particular lived an hour out of town and had a husband who traveled often for work. This made travel to receive treatment difficult and at times there was a need to deviate from an ideal treatment schedule. Another patient was very elderly and legally blind so, despite the moderately

aggressive tumor she had, she often was unable to travel to the clinic for her treatments. Other

issues can stem from concurrent medical issues, such as dementia, which led one patient to

ignore her palpable lump for four years before getting it examined. Some patients have trust

issues or a misconception of what their prognosis is, causing them to ignore extensive family

histories or their BRCA mutation carrier status and resist getting treatment for a tumor that,

without treatment, will be fatal. Additionally, while support from family members has been

shown to be beneficial to patient outcome201, patients can be inundated by several differing

opinions within their support group. It is also easy for patients to feel overwhelmed trying to

balance keeping everyone happy instead of following their own wishes and the advice of their

oncologist.

Despite the fact that I have experienced these processes from the perspective of a family

member - deciding on a treatment direction, completing treatment, and dealing with side effects -

95 when my mother was diagnosed with breast cancer six years ago, I gained a new appreciation for what oncologists experience during my time shadowing. It was abundantly clear that no two cases are alike, and that it takes an amazing amount of compassion and resilience to help patients through this difficult time in their lives. I also better understand the drive of physician scientists.

When you sit face to face with a patient while they discuss their symptoms from treatment such as problems with dry mouth, headaches, body aches, vomiting, and raw skin from radiation, it is abundantly clear that we need to do what we can to develop better, more effective treatments with fewer side effects.

96 APPENDIX B

Supplemental Table 1

Promoter construct Restriction enzyme(s)

4.6 kb ClaI, SnaBI

3.9 kb ClaI, PshAI

3.4 kb ClaI, PmlI

2.0 kb ClaI, SmaI

1.2 kb ClaI, NheI

1.0 kb ClaI, SbfI

0.8 kb ClaI, SmaI

0.2 kb ClaI, NheI

Δ0.2-1.2 kb NheI

Δ1.0-4.3 kb SbfI

Δ3.4-4.6 kb SnabI, PmlI

97