LEE-DISSERTATION-2019.Pdf

Overcoming the Protective Effect of the Bone Marrow

Microenvironment by Dual FLT3 and BET Inhibition in FLT3-ITD

Acute Myeloid Leukemia

Lauren Y. Lee

A dissertation submitted to Johns Hopkins University in conformity with the

requirements for the degree of Doctor of Philosophy.

Baltimore, Maryland

March 2019

Abstract

Acute myeloid leukemia (AML) develops from a block in the terminal differentiation of myeloid progenitor cells. Internal tandem duplication mutations in the juxtamembrane region of the FLT3 receptor tyrosine kinase (FLT3-ITD) are found in about one-third of

AML cases and confer a poor prognosis, particularly in relapsed or refractory settings.

Although FLT3 tyrosine kinase inhibitors (TKIs) have been modestly successful as a monotherapy in FLT3-ITD AML, the disease almost always progresses within a few months’ time due to the emergence of resistant clones, including those harboring tyrosine kinase domain (TKD) mutations. We characterized gilteritinib as a novel FLT3

TKI inhibitor that targets FLT3/TKD mutants with high potency and less myelosuppressive effects, as compared to other clinical FLT3 inhibitors. Even so, gilteritinib fails to overcome the protective effects of the bone marrow environment. We thus explored the dual inhibition of converging FLT3 and BET pathways to target downstream effector proteins that promote leukemia cell survival and proliferation. This study is the first to investigate the combination of FLT3 and BET inhibitors in primary

AML cells modeled in a stromal microenvironment in vitro using clinical stage drugs.

Our pre-clinical suggests that this regimen would be translatable to a clinical study in

AML patients study with the novel combination of these two classes of drugs.

Mark Levis, M.D., Ph.D. (Thesis Advisor)

Gabriel Ghiaur, M.D., Ph.D. (Thesis Reader)

Acknowledgements

First and foremost, I cannot begin to express my gratitude to my thesis mentor, Mark

Levis, who has an open-door policy and has given me invaluable guidance on scientific troubleshooting, my non-traditional career path, and life in general. I genuinely cannot imagine completing my PhD studies under any other advisor and can only hope that I have absorbed a small part of his vast knowledge on drug development and world history over the past four years.

The former and current members of the Levis laboratory have also been instrumental in my thesis work. I am eternally grateful for Daniela Hernandez, who served as my immediate support line whenever I had experimental troubles or a bad day. I also had the pleasure of working with Triv Rajkhowa, who could always be counted on to manage laboratory needs and accomplish any given task.

I would next like to thank my committee members, Gabriel “Gabe” Ghiaur, Alan

Friedman, and John Isaacs, for providing insight and constructive feedback throughout my studies. Gabe and two members of his laboratory, Bogdan Paun and Laura Palau, were indispensable in developing my primary stroma models as they have always enthusiastically shared their bone marrow filters, reagents, and protocols with me.

I am deeply indebted to the CMM program and my classmates, who have continuously supported me in every aspect throughout my time here at Johns Hopkins.

iii

Special thanks to Justina Caushi, a close friend and CMM classmate, who shared her sonicator and baked goods with me.

Lastly, my academic career would not have made it this far if it were not for my family and Alex Wong, who have emotionally and financially supported me since the days of my adolescent dreams of becoming a Ph.D. scientist.

Contents

Abstract ...... ii Acknowledgements ...... iii Contents ...... v List of Tables ...... ix List of Figures ...... x Introduction ...... 1 1.1 Acute Myeloid Leukemia ...... 1

1.2 FLT3 Receptor Tyrosine Kinase ...... 3

1.3 FLT3 Tyrosine Kinase Inhibitors...... 5

Methodologies ...... 9 2.1 Drug Procurement ...... 10

2.1.1 FLT3 Inhibitors ...... 10

2.1.2 BET Inhibitors ...... 10

2.2 Cultured Cell Lines ...... 11

2.3 Patient Samples ...... 12

2.4 Primary Bone Marrow Stroma Culture ...... 12

2.5 Primary AML Blast and Stroma Co-Culture ...... 13

2.6 Cytotoxicity...... 13

2.7 Immunoblotting...... 14

2.8 Colony-forming Assays of Normal Human Bone Marrow ...... 15

2.9 FLT3 and Myc Plasma Inhibitory Activity (PIA) Assays ...... 15

2.10 Next-generation Sequencing ...... 16

2.11 RNA Sequencing ...... 17

2.12 Quantitative PCR ...... 18

Characterization of Gilteritinib ...... 19 3.1 Introduction to Gilteritinib ...... 19

3.2 FLT3 Inhibition in Cell Lines ...... 20

3.3 Estimation of In Vivo FLT3 Inhibition ...... 21

3.4 Cytotoxic Effects of Gilteritinib ...... 22

3.5 Effects of Gilteritinib on Hematopoiesis ...... 24

3.6 Gilteritinib is an Efficacious FLT3-ITD/TKD Inhibitor ...... 24

FLT3 TKI in the Clinic...... 45 4.1 FDA Approval of FLT3 TKI ...... 45

4.1.1 Midostaurin ...... 45

4.1.2 Gilteritinib ...... 46

4.2 Limitations of FLT3 Inhibition ...... 47

4.3 Overcoming FLT3 TKI Resistance ...... 48

4.4 Moving Forward with FLT3 TKI ...... 50

BET Regulates Myc Transcriptome ...... 55

5.1 Introduction to the BET Family ...... 55

5.2 Myc as a Therapeutic Target ...... 57

5.3 BET Inhibition in Overcoming AML Resistance ...... 58

Dual FLT3 and BET Inhibition ...... 59 6.1 A Prospective Synergistic Combination ...... 59

6.2 PLX51107 ...... 60

6.3 Quizartinib ...... 61

6.4 In vitro Characterization of PLX51107 ...... 62

6.5 Estimation of In Vivo BET Inhibition ...... 63

6.6 Screening for Synergy...... 64

6.7 The Intermittent Combination System ...... 65

6.8 Exploring Differential Gene Expression by RNA Expression Analysis ...... 66

6.8.1 Myc ...... 66

6.8.2 p63 and TNF ...... 67

6.9 Corresponding qPCR and Immunoblots ...... 68

6.10 Validation of the intermittent Combination System ...... 70

6.11 Clinical Implications of Intermittent BET Inhibition ...... 71

A Novel BET Inhibitor, PLX2853 ...... 92 7.1 Characterization of PLX2853 ...... 92

7.2 Cytotoxicity of PLX51107 vs PLX2853...... 93

vii

7.3 PLX2853 May Induce Myelosuppression ...... 94

7.4 Future Directions with PLX2853 ...... 95

Concluding Remarks ...... 102 References ...... 105 Curriculum Vitae ...... 112

viii

List of Tables

Table 1. Gilteritinib IC50 values by FLT3 receptor subtype...... 29 Table 2. Clinical characteristics of AML patients in gilteritinib in vitro studies...... 41 Table 3. Patient profiles in PLX51107 and quizartinib combination studies...... 82

List of Figures

Figure 1. Depiction of the FLT3 structure...... 6 Figure 3. Schematic diagram of the binding sites of type I and type II FLT3 TKIs. ..8 Figure 4. Gilteritinib Structure and Kinase Selectivity...... 27 Figure 5. P-FLT3 inhibition by gilteritinib in suspension cell lines in media...... 28 Figure 7. Phospho-AXL Inhibition by gilteritinib in Molm14 cells in media...... 31 Figure 8. Gilteritinib inhibits FLT3 phosphorylation in vivo...... 32 Figure 9. Comparison of the cytotoxic effects of gilteritinib and 4 other FLT3 TKIs against cell lines that express mutated or wild-type FLT3...... 33 Figure 10. Gilteritinib Exhibits Minimal Activity Against WT FLT3 AML...... 34 Figure 11. Comparison of cytotoxicity in diagnostic and relapse samples from the same patient (AML1)...... 35 Figure 12. FLT3 TKI cytotoxicity in a relapsed AML patient sample with high FLT3-ITD allelic burden (AML2)...... 36 Figure 13. FLT3 TKI cytotoxicity in a relapsed patient with high FLT3-ITD allelic burden (AML3)...... 37 Figure 14. (AML4) A patient with both FLT/ITD and TKD (D835Y) mutations, which confer resistance to type II inhibitors sorafenib and quizartinib when compared with counterparts lacking a FLT3/TKD mutation...... 38 Figure 15. (AML5) is a patient with a high FLT3-ITD allelic burden who developed a FLT3/TKD (D835I) mutation after being treated with quizartinib...... 39 Figure 16. Patient AML5 then enrolled on a crenolanib study and was refractory and was noted to have a new F691L gatekeeper mutation...... 40 Table 2. Clinical characteristics of AML patients in gilteritinib in vitro studies...... 41 Figure 17. Inhibition of c-Kit by quizartinib and gilteritinib...... 43 Figure 18. Effects on erythropoiesis by quizartinib and gilteritinib...... 44 Figure 19. Kaplan-Meier curve for OS for patients on CALGB10603...... 51 Figure 20. Kaplan-Meier curve for OS for patients on the salvage chemotherapy arm in the Cephalon 204 versus patients receiving gilteritinib on the Chrysalis trial...... 52 Figure 21. MTT data demonstrate that neither clinically used doses of quizartinib nor gilteritinib is able to eliminate blasts on stroma in vitro...... 53 Figure 22. Cartoon depicting the ways in which the stromal microenvironment maintains blasts on stroma...... 54 Figure 23. Properties of BET Inhibitor PLX51107...... 74 Figure 24. KinomeScan plots for gilteritinib and quizartinib, which exemplify how targeted each drug is for FLT3...... 75

Figure 25. Geometric mean trough plasma concentration of quizartinib and its metabolite AC886...... 76 Figure 26. Effect of quizartinib on Myc inhibition in Molm14 and OCI-AML3 cell lines...... 77 Figure 27. Effects on Myc expression by PLX51107 in media versus in plasma...... 78 Figure 28. Myc plasma inhibitory assays performed using plasma from patients who received 60 mg, 90 mg, 120 mg, or 160 mg PLX51107...... 79 Figure 29. Trough levels by time point from the Myc plasma inhibitory assays in Cohorts 4 through 6...... 80 Figure 30. The concentrations of PLX51107 were interpolated from the standard curve generated from the OCI-AML3 dose response in plasma. Each subject had nine plasma samples collected at different time points of cycle 1 day 1...... 81 Figure 31. Cell viability after 72 hours of exposure to continuous 50 nM quizartinib and 250 nM PLX51107...... 83 Figure 32. Schematic diagram providing the drug conditions used to study the effects of dual FLT3 and BET inhibition...... 84 Figure 33. Comparison of cell counts for AML6 and AML7 samples under continuous or intermittent combination treatment conditions...... 85 Figure 34. Heatmap generated from the RNA-seq study that looked at differentially expressed genes that had log-fold changes >2 as compared to the time-matched control with each treatment condition...... 86 Figure 35. Differentially expressed hallmark gene sets from the RNA-seq data...... 87 Figure 36. TP63 (p63) and TNF signaling are both predicted to be involved in the downstream effects of the combination of 50 nM quizartinib and 250 nM PLX51107 at the 6-hour time point...... 88 Figure 37. Myc and BCL6 RNA quantification from AML6 and AML7 FLT3-ITD relapse samples by qPCR...... 89 Figure 38. Myc, p63 and CD30 immunoblots from patient AML7's blasts on stroma.90 Figure 39. Total number of differentially affected genes by each condition...... 91 Figure 40. PLX2853 dose responses in media and plasma with OCI-AML3 cells. ....96 Figure 41. Myc plasma inhibitory assays were performed on cohorts 1 through 5 in an early phase clinical trial for PLX2853...... 97 Figure 41. Trough levels by time point and subjects in cohorts 1 through 5 for the PLX2853 study...... 98 Figure 43. PLX2853 concentrations from every patient sample interpolated from the standard curve and plotted...... 99 Figure 43. Comparison of the combination effects with quizartinib with either PLX51107 or PLX2853...... 100 Figure 44. Colony forming cell assays to evaluate the effect of PLX2583 and PLX51107 (alone or in combination with 50 nM quizartinib) on normal hematopoiesis...... 101

Introduction

1.1 Acute Myeloid Leukemia

Leukemia develops from impaired differentiation and abnormal proliferation of transformed hematopoietic stem cells or progenitors in the bone marrow or peripheral circulation. The type of leukemia is defined by the type of cell lineage affected (lymphoid or myeloid) and the timeline of the disease progression. The transformation can result in the production of mature, differentiated leukocytes in the blood. This is associated with a slower course of the disease, and the leukemia is classified as chronic; in contrast, an acute leukemia involves immature progenitor cells, or blasts, in the bone marrow and is

distinguished by a rapid disease trajectory. Acute myeloid leukemia arises from the progressive accumulation of mutations that occur within hematopoietic stem cells over time. The mutations are confined to a relatively narrow group of genes that have been identified using next-generation sequencing of large series of patient-derived AML samples. AML may arise de novo, or as a secondary leukemia following radiation exposure or chemotherapy. Presenting symptoms of anemia, neutropenia, and thrombocytopenia in patients are due to bone marrow failure and the infiltration of circulating leukemia cells.3

Acute myeloid leukemia (AML) comprises over 80% of acute leukemia cases in adults and there are over 20,000 new cases in the United States annually.4 AML is a clinically and molecularly heterogeneous disorder characterized by the dysregulated proliferation of hematopoietic myeloid cells, and subgroups are defined by cytogenetic profiles, such as mutations in FLT3-ITD, NPM1 and CEBPA.4 The mutational profiles help to stratify patients into favorable, intermediate, or adverse-risk groups that guide treatment and management, along with other prognostic factors such as age and chromosomal changes.

At the initial diagnosis, AML is a polyclonal disease and is treated with standard

“7+3” induction chemotherapy using continuous infusional cytarabine and an anthracycline, such as idarubicin.5 Approximately two-thirds of patients who receive the conventional induction therapy will go into remission, which is defined by <5% blasts in

the bone marrow cell population. If remission is achieved, the patients are then treated with consolidation therapy using high-dose cytarabine and possibly given the option of an allogeneic hematopoietic bone marrow transplant. Patients, however, often relapse quickly due to the clonal evolution of AML. At relapse, a subclone that survived the initial therapy that harbored or acquired resistant mutations will dominate and expand at relapse. 6

1.2 FLT3 Receptor Tyrosine Kinase

Approximately 30% of AML cases harbor activating mutations in the Fms-like tyrosine kinase receptor 3 (FLT3) gene on chromosome 13q12, which codes for a type III receptor tyrosine kinase that regulates differentiation of CD34+ hematopoietic stem cells and progenitors. 7-9 As shown in Figure 1, the FLT3 single-pass transmembrane receptor is characterized by five immunoglobulin-like repeats in the extracellular ligand-binding domain, a juxtamembrane domain, and a split cytoplasmic tyrosine kinase domain (TKD;

TKD1 and TKD2).10 It is structurally related to KIT, FMS, and PDGF receptors, all of which are highly involved in hematopoiesis.11 In wild-type FLT3 signaling, the binding of the FLT3 ligand (FL) initiates homodimerization of the receptor and induces the activation loop to assume an open conformation to allow the binding of ATP. This leads to the autophosphorylation of the tyrosine kinase domain, and subsequent activation of

downstream signaling pathways, including PI3K/AKT, STAT5, and Ras/Ras/MEK pathways that promote cell proliferation and survival, depicted in Figure 2.12-18

For more than two decades, activating mutations in the FLT3 receptor have been extensively studied. The most common mutations are in-tandem duplications in the juxtamembrane region (FLT3-ITD) and, depending on the size of the ITD and variable allele frequency, often confer a poor prognosis due a more aggressive blast phenotype, a lower chance or remission and an increased tendency to relapse.19 Clinically, higher mutant allele ratios and greater ITD lengths are associated with worse outcomes.20

Though less common, there are also activating mutations in exon 20 of the TKD region of the FLT3 receptor, usually at the activation loop with the substitution of aspartic acid by tyrosine at codon 835.21 Single-base mutations at the TKD constitute around 7% of de novo AML and presentation of these FLT3-TKD mutations have an unclear effect on the prognosis at diagnosis. At relapse, however, FLT-TKD represents a mechanism of resistance to FLT3 tyrosine kinase inhibitors (TKIs) in relapsed/relapsed settings since these mutations can cause conformational changes in the activation loop of the TKD.19,21-23

1.3 FLT3 Tyrosine Kinase Inhibitors

Over the past decade, a large number of small molecule FLT3 tyrosine kinase inhibitors (TKIs) have entered clinical development in order to block the constitutive activity of the mutant FLT3 receptor. There are two general classifications of FLT3 TKIs, depending on the drug’s binding sites on the FLT3 receptor.24 Type I FLT3 TKIs directly compete with ATP for the nucleotide-binding pocket and can bind regardless of whether the activation is in an “open” or “closed” conformation. This class of FLT3 TKIs, which includes midostaurin (PKC412), lestaurtinib (CEP-701), crenolanib (CP-868-596), and gilteritinib (ASP2215), therefore has the potential to block both ITD mutations and TKD activation loop mutations. Type II FLT3 TKIs, such as sorafenib (DB00398), ponatinib

(AP24534), and quizartinib (AC220), require both the ATP-binding pocket and an adjacent allosteric site that is only available with an inactive Asp-Phe-Gly in (“DFG- out”) in the closed activation loop. Since TKD mutations generally cause the activation loop to adopt a “DFG-in” conformation, type II TKIs are only effective against FLT3-

ITD phenotypes.25 Type II FLT3 TKIs are, however, more selective for FLT3 since they bind two sites, while type I only requires the ATP-binding pocket that is conserved within the type III RTK family, which may yield off-target activity against multiple different kinases as illustrated in Figure 3.

Figure 1. Depiction of the FLT3 structure. Printed from the ASH Education Book (January 1, 2006, vol. 2006 no. 1, 178-184) http://asheducationbook.hematologylibrary.org/content/2006/1/178/F1.expansion

Figure 2. Diagram of the FLT3 RTK and signaling pathways.

Activation Inactive form Active form loop form FLT3-ITD FLT3-TKD

Figure 3. Schematic diagram of the binding sites of type I and type II FLT3 TKIs.

Methodologies

In this section, we outline the methods used throughout the studies in the following chapters. Our studies involved investigating and comparing the activity of multiple drugs with cell lines in suspension, or primary AML samples in a bone marrow stromal co- cultural system.

2.1 Drug Procurement

2.1.1 FLT3 Inhibitors

Gilteritinib was provided by Astellas Pharma (Northbrook, IL) and from LC labs.

Crenolanib was supplied by Arog Pharmaceuticals (Dallas, TX). Quizartinib, sorafenib, and midostaurin were obtained from LC Laboratories (Woburn, MA). The compounds were dissolved in dimethyl sulfoxide (DMSO) at stock concentration of 1 mM for gilteritinib and 10 mM otherwise, and stored at -80° C. Dilutions of the inhibitors from the stock solutions were prepared in RPMI (Gibco, Waltham, MA) with 10% fetal bovine serum (FBS) (Gemini Bio Products, Sacramento, CA), penicillin/streptomycin, and 2M

L-glutamine (Gibco, Waltham, MA). The DMSO concentration in all working solutions was .1% or less.

2.1.2 BET Inhibitors

PLX51107 and PLX2853 were obtained from Plexxikon Inc. (Berkeley, CA), a

wholly owned subsidiary of Daiichi Sankyo.

2.2 Cultured Cell Lines

All cell lines and primary leukemia cells were cultured in RPMI with 10%FBS, penicillin/streptomycin, and 2mM L-glutamine at 37°C in 5% CO2, unless otherwise indicated. Molm14 cells (expressing FLT3-ITD mutations), OCI-AML3 cells (expressing a wild-type FLT3 gene, an NPM1 gene mutation (type A), and a DNMT3A R882C mutation), and SEMK2 cells (wild-type FLT3 gene) were purchased from DSMZ

(Deutsche Sammlung von Mikroorganismen and Zellkulturen, Braunschweig, Germany).

TF-1 cells (human AML M6, lacking expression of FLT3) and HL60 (lacking FLT3) cells were purchased from ATCC (Manassas, VA). TF/ITD cells, which express a FLT3-

ITD mutation in the wild-type TF-1 background, were generated as described.26 The

Ba/F3 mouse lymphoid line (DSMZ, Braunschweig, Germany) does not express endogenous FLT3 and was transfected with the FLT3-ITD DNA or the wild type FLT3 with the specified point mutation. The QuickChange Site-Directed Mutagenesis kit

(Stratagene, La Jolla, CA) and Primer 3 software were used to design primers and introduce point mutations into the wild-type FLT3 cDNA in the pBabe Neo vector. The

Nucleofector II kit (Amaxa Biosystems, Walkersville, MD) was used for transfection and the cells were selected for using G418 and analyzed for FLT3/CD135 expression using flow cytometry. The mutations in the transfected lines were confirmed by Sanger sequencing.

2.3 Patient Samples

Primary leukemia blasts were collected under an institutional review board- approved protocol from the Sidney Kimmel Cancer Center at Johns Hopkins Tumor and

Cell Procurement Bank. Plasma and bone marrow were also obtained from healthy donors under this procurement protocol. Any specimens used were from patients who gave informed consent according to the Declaration of Helsinki.

2.4 Primary Bone Marrow Stroma Culture

Cell culture flasks and plates for stromal culture were coated with .1% gelatin

(Sigma Aldrich) in PBS and incubated at 37ºC for 30 minutes prior to plating stromal cells. Excess gelatin was removed.

Total mononuclear cells were harvested from bone marrow aspirates from healthy donors and then plated onto gelatinized T75 flasks with FMBD1 medium (consisting of

Iscove’s modified Dulbecco’s medium (IMDM) (Sigma-Aldrich) supplemented with

10% horse serum (Sigma-Aldrich), 10% FBS, 10-5 M hydrocortisone 21-hemisuccinate

(Sigma-Aldrich), P/S, and 0.1 mM β-mercaptoethanol).27

After 72 hours of incubating the total mononuclear cells at 37ºC, non-stromal cells in suspension were removed and replaced with fresh FBMD1 medium. The attached

stromal cells in the flasks were incubated at 37ºC until the cells had expanded into a confluent monolayer. Stromal cells were passaged only when confluency was reached and re-plated at a 1:2 ratio onto gelatinized T175 flasks.

2.5 Primary AML Blast and Stroma Co-Culture

Stromal cells were attached to cell culture plates at least 24 hours prior to co-culture.

Stroma was detached from flasks using .25% trypsin-EDTA solution (Sigma Aldrich) and re-suspended in fresh FBMD1 medium. Stromal cells were then added onto gelatinized cell culture plates at specific concentrations depending on the type of plate. For 96 well plates, stroma was plated at a concentration of 3,000 cells/well. For 12 well plates, the starting stroma concentration was 30,000 cells/well. In 6 well plates, the starting stroma concentration was 100,000 cells/well. After 24 hours, stroma had attached and the

FBMD1 medium was removed. From this point forward, the co-cultures were incubated with RPMI with 10% FBS, penicillin/streptomycin, and L-glutamine.

2.6 Cytotoxicity

Cell lines cultured in 96-well plates were incubated with specified drug treatments for

48 hours prior to the assessment of cytotoxicity by a dimethyl-thiazole diphenol

tetrazolium bromide (MTT) assay (Roche Diagnostics, Indianapolis, IN) as described.28

Primary blast and bone marrow stroma co-cultures were similarly incubated with indicated drugs for 72 hours prior to MTT assessment. Cell duplicates were concurrently counted for cell viability by trypan blue, when indicated.

2.7 Immunoblotting

After the specified drug treatments with cells, immunoprecipitation and immunoblotting for FLT3 was performed as described.29 Similar immunoprecipitation and immunoblotting methods were used to evaluate c-Kit phosphorylation. The cell lysate was incubated overnight with anti-c-Kit antibody (C-19; Santa Cruz

Biotechnology, Santa Cruz, CA), mixed with Protein A-Agarose beads, and the phosphorylation status was probed with the antiphosphotyrosine (4G10) antibody. Total c-Kit levels were detected with the c-Kit antibody. The antibody used to detect P-Axl was from R & D Systems (Minneapolis, MN). Antibodies from Cell Signaling Technology

(Danvers, MA) were used to probe for Axl, Myc, TNFRSF8 (CD30) and beta-actin. The antibody for p63 (TP63) was purchased from Abcam (Cambridge, MA).

For calculation of IC50 (concentration that inhibits to 50% of untreated control) values, densitometric values of dose-response blots were expressed as a fractional effect and subjected to regression analysis after linear conversion using commercially-available

software (CalcuSyn; Biosoft, Cambridge, UK). Each experiment was performed at least

3 times and the results were averaged.

2.8 Colony-forming Assays of Normal Human Bone Marrow

Mononuclear cells were isolated from normal bone marrow using a Ficoll gradient. The mononuclear cells were plated on methylcellulose medium MethoCult

H4435 Enriched (Stem Cell Technologies, Vancouver, BC, Canada) containing the indicated drug at 100,000 cells/mL in 35-mm dishes. The dishes were incubated for 10 to

12 days and the quantification of colonies was visually evaluated using an Olympus

CKX31 inverted microscope.

2.9 FLT3 and Myc Plasma Inhibitory Activity (PIA) Assays

Plasma inhibitory activity assays for FLT3 and c-Myc represented an ex vivo

surrogate for the inhibitory activity of FLT3 and BET inhibitors in patients treated

with these drugs. Plasma for the PIA assays was collected from

relapsed/refractory AML patients enrolled in a phase 1/2 dose-

escalation/expansion study of gilteritinib (NCT02014558). Plasma samples were

collected at days 1, 8, and 15 of cycle 1, and day 1 of cycle 2 from patients that

took a once daily dose of 120 mg gilteritinib. The samples were stored at -80C

and used within 12 months of collection for the PIA assays.

FLT3 PIA assays were performed by incubating patient plasma taken from various time point with Molm14 cells for 1 hour. The Molm14 cells were then probed for pFLT3 and FLT3 by immunoprecipitation and immunoblotting.30

Myc PIA assays were done similarly with modifications. Patient plasma samples were incubated with OCI-AML3 cells for 3 hours, and OCI-AML3 cells were probed for Myc and β-actin.

2.10 Next-generation Sequencing

Genomic DNA was isolated from primary leukemia blasts with QIamp Mini Kit

(Qiagen, Valencia, CA) and 1 μg of DNA was fragmented into 250–300 bp lengths to construct the libraries. The libraries were then hybridized to a SureSelect custom clinical leukemia panel 2.8 M bait set from Agilent (Santa Clara, CA), which targeted 637 genes relevant to oncogenesis. This panel of genes is available upon request. Samples were clustered using a cBOT system and sequenced on a HiSeq2500 using a 2 x 150-bp phycoerythrin protocol (Illumina, San Diego, CA). Reads were aligned to the human

genome (GRCh37/hg19) with the Burrows-Wheeler alignment algorithm and variant calling was performed with a custom pipeline. The protocol covered >300 reads in >94% of all genomic regions.

2.11 RNA Sequencing

Sample Preparation – For each of the three patients, 5 million cells per condition were used. The cells were treated with the appropriate drug dose on stroma and, at each indicated time point, RNA was isolated from each sample using the RNeasy Plus Mini

Kit and stored at -80º C (Qiagen).

Library Preparation and Sequencing – mRNA is enriched using oligo(dT) beads.

Enriched mRNA is randomly fragmented by adding fragmentation buffer, then first strand cDNA is synthesized by using mRNA template and random hexamer primers followed by addition of a custom second-strand synthesis buffer (Illumina), dNTPs,

RNase H and DNA polymerase I to initiate the second-strand synthesis. Double-strand cDNA molecules are end repaired, followed by adenine ligation and then sequencing adaptor ligation. The double-stranded cDNA library is size selected and then PCR enriched. Libraries are sequenced on Illumina sequencers, generating an average of 30 million, cleaned, 150 base read pairs.

Analysis – RNA expression analyses were performed both iteratively and with the

Ingenuity Pathway Analysis.

2.12 Quantitative PCR

RNA samples were isolated from either 5 million cells per cell line, or 10 million blasts per primary AML sample. The Taqman GAPDH primer (Hs02758991_g) were purchased from Thermo Fisher Scientific (Waltham, MA). Myc and BCL6 were designed accordingly:

BCL6-F: 5′-AACCTGAAAACCCACACTCG-3′

BCL6-R: 5′-TTCGCATTTGTAGGGCTTCT-3′

Myc-F: 5’ CTGGTGCTCCATGAGGAGA-3’

Myc-R: 5’ CCTGCCTCTTTTCCACAGAA-3’

Characterization of Gilteritinib

3.1 Introduction to Gilteritinib

While there is some evidence that FLT3 TKIs have improved outcomes for AML patients, their development has been impeded by a number of obstacles.31,32 The first generation of FLT3 TKIs, such as lestaurtinib, midostaurin, and sorafenib, were “re- purposed” multi-kinase inhibitors that lacked potency and produced relatively limited clinical responses as monotherapy.26,30,33,34 Conversely, quizartinib is a highly potent

FLT3 TKI that also inhibits c-Kit, which can cause myelosuppression.35 Because FLT3 inhibition needs to be sustained in order to induce cytotoxicity, some inhibitors failed to

produce responses due to a short in vivo half-life.36 Finally, potent, sustained FLT3 inhibition can result in the emergence of resistance-conferring FLT3 point mutations, most often at residue D835. 37,38

Gilteritinib (ASP2215) is a pyrazinecarboxamide-derivative currently being studied in

AML clinical trials due to its potential selectivity, potency, and activity against all classes of FLT3 activating mutations (Figure 4). In the following, we investigated the activity of gilteritinib and compared it with four other FLT3 TKIs in clinical development- midostaurin, sorafenib, quizartinib, and crenolanib.29,39-41

3.2 FLT3 Inhibition in Cell Lines

We first tested the inhibitory activity of gilteritinib against different forms of

FLT3 in leukemia cells by immunoblotting (Figure 5 and Table 1). When tested in culture medium, the IC50 for inhibition of the wild type receptor was 5 nM and for the

ITD-mutated form ranged from 0.7 to 1.8 nM depending on the cell context. Using a panel of FLT3 point mutations known to confer resistance to type II inhibitors, such as sorafenib and quizartinib, we also found that gilteritinib had similar degrees of inhibitory activity against commonly identified TKD mutations (Table 1). It also had activity against the gatekeeper mutation at F691, although at a relatively higher IC50. In human

plasma, the IC50 for inhibition of the FLT3-ITD receptor ranged from 17 to 33 nM

(Figure 6).

Data from the kinase selectivity assay indicated that in addition to FLT3, gilteritinib has activity against the receptor tyrosine kinase Axl, which may modulate the activity of FLT3 in AML.42 We confirmed that the drug inhibits Axl (Figure 7), although the IC50 against this receptor is 41 nM, approximately 20-fold higher than what we observed for the ITD-mutated FLT3 receptor. We next tested the compound against a series of FLT3 point mutations that are known to confer resistance to type II inhibitors such as sorafenib and quizartinib.

3.3 Estimation of In Vivo FLT3 Inhibition

A primary obstacle in the development of FLT3 inhibitors has been the inability to achieve sustained FLT3 inhibition in vivo with a drug that appears to be potent in vitro.36,43,44 To confirm that gilteritinib could achieve high-level FLT3 inhibition in AML patients we used a surrogate assay for in vivo inhibition, the plasma inhibitory activity

(PIA) assay. Plasma was collected at trough time points from patients receiving gilteritinib while enrolled on a phase 1/2 study (NCT02014558) and used in PIA assays.

As can be seen in Figure 8, plasma from 3 separate patients receiving a daily dose of 120 mg gilteritinib fully inhibited the FLT3-ITD receptor expressed in Molm14 cells.

3.4 Cytotoxic Effects of Gilteritinib

We next compared the cytotoxic effects of gilteritinib versus other FLT3 TKIs in

FLT3-expressing leukemia cell lines and primary patient samples. Gilteritinib had a response profile typical of FLT3 inhibitors against the FLT3-ITD-expressing Molm14 and MV4-11 cell lines, and a modest effect against SEMK2 cells, which over-express wild type FLT3 (Figure 9). In HL60 cells, which express low levels of wild type FLT3, there was no cytotoxic effect. Accordingly, primary samples obtained from four AML patients lacking FLT3 activating mutations, no effect was observed, although this small sampling is admittedly not a representation of all AML subtypes (Figure 10).

We previously reported the difference in responsiveness of diagnostic versus relapsed

FLT3-ITD AML samples,28 likely due to the more polyclonal nature of the disease at diagnosis.45 Consistent with this finding, blasts from patient AML1 collected at diagnosis failed to respond to gilteritinib, while blasts collected at relapse from the same patient eight weeks later displayed a cytotoxic response (Figure 11). Therefore, we focused on specimens collected at relapse. Two representative samples out of the eight tested, patients AML2 and AML3, displayed the typical cytotoxic response to all five inhibitors

(Figure 12-13).

Of particular interest were samples that had been collected from two patients

(AML4 and AML5) following treatment with FLT3 TKIs. The cytotoxicity profiles of

these two samples exposed to the different inhibitors are consistent with previous reports on relative resistance conferred by TKD mutations to FLT3 TKIs (Figure 1C).46,47

Patient AML4 presented with a FLT3-ITD mutation at diagnosis and was refractory to induction and salvage chemotherapy. He then achieved a CRi (complete response with incomplete count recovery) with quizartinib and this was followed with an allogenic transplant. At relapse six months later, he initially responded to treatment with sorafenib, but his disease progressed with a FLT3-D835Y mutation. The results (Figure 14) show that only gilteritinib and crenolanib were effective at inducing a cytotoxic response in this relapse sample. Likewise, patient AML5 was refractory to conventional chemotherapy and responded initially to quizartinib before her disease developed a D835I mutation. As shown in Figure 15 (“AML5 post-quizartinib”), gilteritinib was the most effective TKI at this point and was the only TKI to inhibit FLT3 at a concentration of 20 nM by immunoblot. Patient AML5 was then referred to a clinical trial for crenolanib and progressed after several weeks of treatment. The blasts now harbored a mutation at

F691L and were resistant to all inhibitors, although gilteritinib produced a weak response

(Figure 16, “AML5 refractory”). The clinical characteristics of these AML patients are listed in Table 2.

3.5 Effects of Gilteritinib on Hematopoiesis

Inhibitors of FLT3 are often found to have activity against the structurally-related receptor c-Kit.29,35,48 This has important clinical implications because c-Kit is essential for normal hematopoiesis, and even partial inhibition of this receptor can result in marrow suppression.17,35 Three of the four FLT3 TKIs we compared gilteritinib to have activity against c-Kit- midostaurin, quizartinib, and crenolanib.29,35,49 We therefore tested the activity of gilteritinib against c-Kit in immunoblot assays using the erythroleukemia

50 line, TF-1. As shown in Figure 17, gilteritinib has an IC50 against wild type c-Kit of

102 nM, two orders of magnitude greater than its activity against mutant FLT3. As expected, therefore, it has minimal effects on normal hematopoiesis in comparison to a drug such as quizartinib, as shown in progenitor cell assays using bone marrow from healthy donors (Figure 18).

3.6 Gilteritinib is an Efficacious FLT3- ITD/TKD Inhibitor

At the time of this study in November 2016, FLT3 TKIs had been investigated for over a decade, but no such agent had yet received regulatory approval for use in AML.

Investigators seeking to develop clinically useful FLT3-TKIs have encountered a seemingly endless array of biologic obstacles- sub-optimal pharmacokinetics, poor selectivity and potency, myelosuppression, prolongation of the QT interval, and the emergence of resistance-conferring FLT3 point mutations. Based on our data, gilteritinib overcomes most of the obstacles thus far encountered. Its in vitro efficacy is equal to or greater than that of the other TKIs, it inhibits the FLT3/TKD mutations that are predominantly responsible for resistance, it is unlikely to be myelosuppressive, and it achieves sustained inhibition of FLT3 in vivo.

The issue of c-Kit inhibition by FLT3 TKIs is of some interest in this field. CD117 is very commonly expressed on AML blasts, and it is entirely possible that inhibition of c-

Kit may confer some anti-leukemic effect apart from FLT3 inhibition. Therefore, a TKI with dual activity may yield a better overall response rate, at the price of myelosuppression. It may be that gilteritinib’s ideal role would be as a very well- tolerated maintenance treatment after completion of consolidation therapy. An additional issue is that of the gatekeeper residue at F691. While gilteritinib has some activity against this mutation, it is unclear whether in vivo levels of the drug would be sufficient to achieve control of such a clone, and thus this remains a potential weakness. Finally, gilteritinib has inhibitory activity against Axl, although to a much less degree than it has against FLT3. Partial inhibition of Axl may confer some degree of anti-leukemic activity, or, alternately, such inhibition might counteract a mechanism of resistance.42 In

summary, gilteritinib is a potent, selective FLT3 TKI that lacks many of the weaknesses of previously developed inhibitors.

2-Pyrazinecarboxamide, 6-ethyl-3-((3-methoxy-4-(4-(4-methyl-1- piperazinyl)-1-piperidinyl)phenyl)amino)-5-((tetrahydro-2H-pyran- 4-yl)amino)-, (2E)-2-butenedioate (2:1)

Figure 4. Gilteritinib Structure and Kinase Selectivity (A) The structure of gilteritinib. (B) Inhibitory activity of gilteritinib against select kinases using a commercially-available kinase screen (information supplied by the industry sponsor).

Molm14 P-FLT3

FLT3

0 2 5 10 20 50 Gilteritinib (nM)

TF/ITD P-FLT3

FLT3 0 2 5 10 20 50 Gilteritinib (nM)

SEMK2 P-FLT3

FLT3

0 2 5 10 20 50 Gilteritinib (nM)

Figure 5. P-FLT3 inhibition by gilteritinib in suspension cell lines in media. FLT3-ITD cell lines (Molm14 and TF/ITD) and a wild-type FLT3 cell line (SEMK2) were treated with gilteritinib in RPMI/10% FBS for 1 hour, lysed, and probed for phospho-FLT3 and total FLT3.

FLT3 receptor subtype Gilteritinib IC 50 Wild type 5 nM Molm14 (ITD) 1.8 nM TF/ITD 1.4 nM Ba/F3 ITD 1.6 nM Ba/F3 D835Y 1.4 nM Ba/F3 D835H 2 nM Ba/F3 D835V 0.7 nM Ba/F3/ITD F691L 12.2 nM

Table 1. Gilteritinib IC50 values by FLT3 receptor subtype. Cells expressing the indicated subtype were incubated with gilteritinib for 1 hour at increasing concentrations in RPMI/10% fetal bovine serum (FBS), analyzed by immunoprecipitation and/ or immunoblotting for phospho- and total FLT3, and followed by densitometry. SEMK2 cells express wild-type FLT3, Molm14 cells express a 21-bp ITD mutation, and TF/ITD and Ba/F3 ITD cells express a transfected FLT3 construct containing an 18-bp ITD. Ba/F3 cell lines with the indicated mutant FLT3 isoforms were generated by transfecting the murine lymphocyte Ba/F3 line with constructs expressing the FLT3 receptor (non- ITD–containing, except for the F691L variant) containing the indicated single point mutation. Each IC50 value was calculated from multiple experiments.

Molm14 P-FLT3

FLT3

0 20 50 100 200 500 Gilteritinib (nM)

TF/ITD

P-FLT3

FLT3 0 20 50 100 200 500 Gilteritinib (nM)

Figure 6. P-FLT3 Inhibition by Gilteritinib in Cell Lines in Plasma. The above cell lines were also treated with gilteritinib in normal plasma and lysed to detect phospho-FLT3 and total FLT3.

Molm14

P-AXL

AXL

0 2 5 10 20 50 100 200 Gilteritinib (nM) Figure 7. Phospho-AXL Inhibition by gilteritinib in Molm14 cells in media. Molm14 cells, which express AXL, were treated with gilteritinib for 4 hours and analyzed by immunoblotting for phospho-AXL and AXL. MV411 cells were also tested and yielded a similar IC50 (data not shown).

C1 D1 C1 D8 C1 D15 C2 D1

P-FLT3

FLT3

20076

C1 D1 C1 D8 C1 D15 C2 D1

P-FLT3

FLT3

20134

C1 D1 C1 D8 C1 D15 C2 D1

P-FLT3

FLT3

20119

Figure 8. Gilteritinib inhibits FLT3 phosphorylation in vivo. Plasma samples collected at trough time points from 3 patients receiving 120 mg gilteritinib were incubated with Molm14 cells for 1 hour. Immunoprecipitation and immunoblotting were used to detect phospho-FLT3 and total FLT3 at these time points.

Molm14 MV4-11

100 Gliteritinib 100 Gliteritinib Crenolanib Crenolanib Quizartinib Quizartinib Midostaurin Midostaurin Sorafenib Sorafenib

80 80

l l

o o

r r t

t 60 60

n n

o o

C C

f f

o o

40 40

% %

20 20

0 0 0 20 40 60 80 100 0 20 40 60 80 100 [Drug nM] [Drug nM]

SEMK2 HL60

100 Gliteritinib 100 Crenolanib Quizartinib Midostaurin Sorafenib

80 80

r t

60 t 60

40 40

% %

Gliteritinib 20 20 Crenolanib Quizartinib Midostaurin Sorafenib 0 0 0 20 40 60 80 100 0 20 40 60 80 100 [Drug nM] [Drug nM]

Figure 9. Comparison of the cytotoxic effects of gilteritinib and 4 other FLT3 TKIs against cell lines that express mutated or wild-type FLT3. Both the Molm14 and MV4- 11 lines express FLT3-ITD and lack expression of wild type FLT3. Both cell lines show a dose-dependent response to FLT3 TKI treatment. SEMK2 cells express a wild-type FLT3 receptor and HL60 cells express minimal levels of wild-type FLT3. 33

Effect on Primary Wild-type FLT3 Blasts

100

)

l o

r 80

( 60

b a

i 40

e C 20 WT1AML6 WT2AML7 WT3AML8 WT4AML9

0 0 20 40 60 80 100 [Gilteritinib nM]

Figure 10. Gilteritinib Exhibits Minimal Activity Against WT FLT3 AML. Four primary AML samples lacking FLT3 activating mutations were incubated with increasing concentrations of gilteritinib. After 72 hours, an MTT assay was performed. WT1 and WT4 have complex cytogenetics; WT2 and WT3 have normal cytogenetics; WT2 has a G13R mutation in NRAS and a mutant NPM1, while WT3 also has a mutant NPM1 and an R172K mutation in IDH2.

AML1

100

) 80

o C

(

i l

i 40

i V

20 Diagnosis Relapse

0 0 20 40 60 80 100 [Gilteritinib nM]

Figure 11. Comparison of cytotoxicity in diagnostic and relapse samples from the same patient (AML1). One sample was collected at diagnosis, and the other at point of relapse.

AML2

100

) 80

o C

(

i l

i 40

i V Gilteritinib 20 Crenolanib Quizartinib Midostaurin Sorafenib 0 0 20 40 60 80 100 [Drug nM]

Figure 12. FLT3 TKI cytotoxicity in a relapsed AML patient sample with high FLT3-ITD allelic burden (AML2).

AML3

100 )

l 80

o C

(

i l

i 40

i V

20 Gilteritinib Crenolanib Quizartinib Midostaurin Sorafenib 0 0 20 40 60 80 100 [Drug nM]

Figure 13. FLT3 TKI cytotoxicity in a relapsed patient with high FLT3-ITD allelic burden (AML3).

AML4

100

% Control % of 40

Gilteritinib 20 Crenolanib Quizartinib Midostaurin Sorafenib 0 0 20 40 60 80 100 [Drug nM]

Figure 14. (AML4) A patient with both FLT/ITD and TKD (D835Y) mutations, which confer resistance to type II inhibitors sorafenib and quizartinib when compared with counterparts lacking a FLT3/TKD mutation.

AML5 (Pre-treatment)

100 FLT3 TKI (20nM)

pFLT3 80

% % Control of 40

Gilteritinib (G) 20 Crenolanib (C) Quizartinib (Q) Midostaurin (M) Sorafenib (S) 0 0 20 40 60 80 100 [Drug nM]

Figure 15. (AML5) is a patient with a high FLT3-ITD allelic burden who developed a FLT3/TKD (D835I) mutation after being treated with quizartinib. The MTT assay performed with the patient’s blasts after progression on quizartinib indicate that gilteritinib was the most potent of all 5 FLT3 TKIs against this patient’s blasts. The immunoblot shows the level of phospho-FLT3 expression in blasts after a 1-hour treatment with 20 nM of each FLT3 TKI in RPMI/10% FBS.

AML5 (Refractory)

100

r t

n 60

% 40

Gilteritinib 20 Crenolanib Quizartinib Midostaurin Sorafenib

0 0 20 40 60 80 100 [Drug nM]

Figure 16. Patient AML5 then enrolled on a crenolanib study and was refractory and was noted to have a new F691L gatekeeper mutation. The MTT assay was repeated with the patient’s blasts after treatment with crenolanib and shows the resistance conferred by the F691L mutation.

Table 2. Clinical characteristics of AML patients in gilteritinib in vitro studies.

Patient Age/Gender Karyotype Mutations Treatment received prior to collection of blasts AML1 62/F Diploid 1-FLT3-ITD 33 bp Induction with cytarabine, (Diagnostic) allelic ratio 0.55; idarubicin, etoposide; 2- ATM Refractory to induction

AML1 62/F Diploid 1-FLT3-ITD 33 bp Induction with cytarabine, (Refractory) allelic ratio 0.50 idarubicin, etoposide; 2- ATM Refractory to induction 3-IDH1 R132H

AML2 31/F Diploid 1-FLT3-ITD 57 bp, Induction with 7+3; allelic ratio 0.891; 4 cycles of HiDAc 2-NPM1mut Relapse, salvage MEC; 3-ASXL1 E865K Disease progression 4-CBL R585H AML3 27/F t(5;18) 1-FLT3-ITD 90 bp, Induction with 7+3; allelic ratio 2.73 Allogeneic transplant 2-WT1 Relapse post-transplant AML4 35M Diploid 1-FLT3-ITD 75 bp Induction with 7+3; allelic ratio 0.43; Refractory to induction 2-FLT3-D835Y Refractory to ida-FLAG allelic ratio 2.25; Response to quizartinib 3- WT1 P376fs Allogeneic transplant 4- IDH1 R132H Relapse Response to sorafenib, but disease progression AML5 62F Diploid 1-FLT3-ITD 42 bp Induction with 7+3; (Responsive) allelic ratio 0.43; Refractory to induction 2-FLT3-ITD 75 bp Refractory to ida-FLAG allelic ratio 0.083; Response to quizartinib 3- FLT3-D835I Disease progression allelic ratio 0.39 4- DNMT3A R882H 5- TET2 C1396R AML5 62F Diploid 1-FLT3-ITD 42 bp Induction with 7+3; (Resistant) allelic ratio 0.5; Refractory to induction 2- FLT3-D835I Refractory to ida-FLAG allelic ratio 0.38 Response to quizartinib 3- FLT3-F691L Disease progression 4- DNMT3A R882H Refractory to crenolanib 5-TET2 C1396R

1. Ratio relative to the wild type FLT3 allele. Chemotherapy regimens: 7+3 = intravenous cytarabine continuous infusion for 7 days; intravenous idarubicin or daunorubicin days 1-3; HiDAc = high dose cytarabine twice daily, days 1, 3, and 5; MEC = mitoxantrone, etoposide, cytarabine; ida-FLAG = idarubicin, fludarabine, cytarabine.

Quizartinib P-c-Kit

c-Kit

0 20 80 200 500

Gilteritinib P-c-Kit

c-Kit

0 20 80 200 500

Figure 17. Inhibition of c-Kit by quizartinib and gilteritinib. TF-1 cells were treated with increasing concentrations of either quizartinib or gilteritinib in RPMI/10% FBS for 1 hour. Immunoprecipitation and immunoblotting were performed to detect the phosphorylation status of c-Kit and total c-Kit.

Figure 18. Effects on erythropoiesis by quizartinib and gilteritinib. Mononuclear cells isolated from normal donor bone marrow were plated at 105 cells/mL in MethoCult. Increasing concentrations of quizartinib or gilteritinib were added. Counts for colony-forming unit-granulocyte, monocyte (CFU-GM) and burst- forming unit-erythroid (BFU-E) colonies were done after 10 to 12 days of incubation (n = 3). 44

FLT3 TKI in the Clinic

4.1 FDA Approval of FLT3 TKI

4.1.1 Midostaurin

In April 2017, midostaurin (Rydapt; Novartis Pharmaceuticals, Inc), a multi-kinase inhibitor that hits both FLT3-ITD and TKD mutants, was the first FLT3 TKI to be approved by the US Food and Drug Administration (FDA) for the use in newly diagnosed

FLT3-mutated AML. It had been in development for three decades and was the first

AML drug to be receive clearance since the year 2002. The FDA approval of Rydapt was based on the Phase III RATIFY (CALGB 10603 [Alliance]) clinical trial, which investigated whether the addition of midostaurin to induction and consolidation, followed

by 1 year of maintenance, would improve the OS of 717 patients with FLT3-mutated

AML. The current FDA-recommended dose is 50 mg twice daily on days 8 to 21 of each cycle of induction therapy with cytarabine and daunorubicin and days 8 to 21 of each cycle of consolidation with high-dose cytarabine. 51

Notably, the FDA label states that the drug is not intended to be used as a single- agent treatment. In the early phase clinical studies, midostaurin reduced blast burden in the blood and bone marrow, but was not effective in achieving remissions in patients as a monotherapy. This is most likely due to its multi-kinase targeting characteristic and low potency for the FLT3 mutations. However, since AML presents as a polyclonal disease at diagnosis, midostaurin may confer some level of benefit in initial induction and consolidation therapies, which was observed in the RATIFY trial and shown in Figure

19.1,51 There is therefore a need for more selective and potent FLT3 TKIs as midostaurin is not able to improve remission rates or clinical outcomes for FLT3-mutated AML patients.

4.1.2 Gilteritinib

Shortly after our pre-clinical studies on gilteritinib (Xospata; Astellas Pharma), the drug attained an FDA approval for use in relapsed/refractory FLT3-mutated AML as a single agent in November 2018. The approval was based on an interim analysis of the

ADMIRAL trial (NCT02421939), which was an open-label, multicenter, randomized study of gilteritinib versus salvage chemotherapy in adult patients with relapsed or

refractory AML having a FLT3 ITD, D835, or I836 TKD mutation. Patients were orally administered 120 mg daily. After a median follow-up of 4.6 months (range: 2.8 to 15.8), only 29 out of 126 patients with FLT3-ITD or FLT3-ITD/TKD mutations achieved complete remission (CR) or CR with partial hematologic recovery (CRh) (21%, 95% CI:

14.5, 28.8).

In the CHRYSALIS trial (NCT02014558), a phase 1/2 study of gilteritinib versus salvage chemotherapy in patients with FLT3-mutated relapsed or refractory AML, patients were given doses ranging from 20 mg to 450 mg daily. Similarly, only 37% of the 191 patients enrolled achieved a CR (complete remission), CR with incomplete hematological recovery (CRi), or CR with incomplete platelet recovery (CRp).52 Though the FLT3-mutated patients receiving ≥80 mg/days showed prolonged survival (median

OS, 31 weeks) compared with the reported survival data from salvage chemotherapy in the Cephalon 204 trial (median OS, 15 to 20 weeks), the difference appeared to be minor

(Figure 20).

Thus, gilteritinib as a monotherapy remains insufficient in overcoming bone marrow

(BM) stromal protection of the leukemic blasts and the resultant survival of resistant subclones.

4.2 Limitations of FLT3 Inhibition

Though quizartinib and gilteritinib are the two most potent FLT3 TKIs in late clinical stages, neither one is able to clear blasts on stroma in vitro (Figure 21). It has been

reported that while FLT3 TKIs induce apoptosis in circulating FLT3-mutated peripheral blasts in the blood, bone marrow blasts enter cell-cycle arrest and some will undergo terminal differentiation.53 The FLT3-mutated subclones that remain, however, are maintained by the protective bone marrow stroma in two ways: (1) stroma cells metabolize the drug via CYP3A4 enzyme, and (2) sustain AML blast signaling through activating stromal membrane proteins and secreted cytokines.54-58 Figure 22 summarizes the types of protection conferred by the stromal microenvironment. In the presence of prolonged FLT3 inhibition, these resistant subclones will transform to proliferate independently of FLT3 signaling by acquiring secondary FLT3-TKD mutations, such as the gatekeeper F691I/L in the TKD1 of the ATP binding pocket and the D835V/Y/F in the activation loop of TKD2, that hinder drug binding or activate and amplify the signaling of parallel pathways, particularly Ras/ERK.37,59-61 Consequently, responses to

TKI treatments are only months long before the clonal proliferation of resistant blasts predominates. We have previously attempted to identify the specific stromal-derived factors responsible for the persistent downstream signaling and detected a large array of secreted cytokines in a transwell assay, including CXCR4, TNF‐alpha, angiopoietins, interleukins, G‐CSF/GM‐CSF/TNF and VEGF/EGF/IGF.57

4.3 Overcoming FLT3 TKI Resistance

Given these findings, we proposed a novel strategy to undermine stromal protection and thereby hinder the growth of resistant blasts. Although we could increase the total

amount of any given drug to overcome the drug metabolizing activity of stromal cells, quizartinib is already the most potent FLT3 TKI available and the clinically achievable doses are not enough to prevent the rise of resistant FLT3-TKD mutants. As for the secretion of cytokines and stromal-related proteins that maintain signaling in blasts, there were too many identified to narrow down to the few that are most important.57

An alternate strategy is to prevent the activation of signaling pathways within the blasts by blocking both FLT3 signaling while also blocking any parallel signaling pathways maintain signaling and clonal evolution of FLT3 TKI-resistant subclones.

However, blocking key growth-regulating pathways, such as MAPK/ERK, would most likely be intolerable in humans as most wild-type cells in the body require them as well.

We therefore attempted to target the downstream kinase PIM in combination with FLT3 inhibition, but found that a low percentage of primary AML samples in vitro showed synergy in cytotoxicity in cell viability assays. The lack of synergy may have been due to the sustained signaling through another kinase pathway that also converged with FLT3 signaling downstream. Therefore, we proposed to look further downstream at the convergence of these pathways, which all drive transcription of genes directly linked to the survival and proliferation of TKI-resistant FLT3-TKD AML subclones. If the transcription systems of pro-survival proteins and transcription factors could be inhibited, signaling through the compensatory pathways that maintain the leukemic cells in the presence FLT3 TKI therapy could be rendered ineffective.62

4.4 Moving Forward with FLT3 TKI

While there is evidence that FLT3 TKI as a monotherapy confers some clinical benefits, these are often short-lived as the persistence resistant subclones eventually leads to relapsed or refractory AML. Consequently, there is a great need to address the mechanisms of resistance conferred by the protective stromal environment. We speculated that the disruption of proto-oncogenic activity of pro-survival proteins, such as

Myc or Bcl-2, might be necessary to prevent compensatory signaling of AML blasts in the presence of a FLT3 TKI on stroma. This hypothesis led us to explore the inhibition of bromodomain and extra-terminal domain (BET) proteins, which are master transcription regulators that influence the transcription of Myc.12

Figure 19. Kaplan-Meier curve for OS for patients on CALGB10603. This was presented at the 2015 annual meeting of the American Society of Hematology, Orlando, FL, 5-8 December 2015. Reprinted from Stone et al.1

MEC or HiDAc salvage (Cephalon 204 trial) Gilteritinib monotherapy (Chrysalis 2215-CL-0101 trial )

Figure 20. Kaplan-Meier curve for OS for patients on the salvage chemotherapy arm in the Cephalon 204 versus patients receiving gilteritinib on the Chrysalis trial.

100 nM Gilteritinib

50 nM Quizartinib

Figure 21. MTT data demonstrate that neither clinically used doses of quizartinib nor gilteritinib is able to eliminate blasts on stroma in vitro.

FLT3 TKI

Figure 22. Cartoon depicting the ways in which the stromal microenvironment maintains blasts on stroma.

BET Regulates Myc Transcriptome

5.1 Introduction to the BET Family

The BET family consists of Brd2, Brd3, Brd4, which are expressed in all tissues, and

Brdt, which is predominantly in the testes. These proteins are characterized by two conserved bromodomains (BD1 and BD2) and an extraterminal domain. The bromodomains function as chromatin “readers” by recognizing acetylated lysine residues on histone tails and other nuclear proteins.63 These epigenetic interactions play a crucial roles in regulating RNA polymerase II (Pol II)-mediated gene transcription during cellular proliferation and differentiation processes by recruiting regulatory protein complexes to the transcription start site.64 Each BET serves a different purpose in the activation of global transcription. The roles of BRD2 and BRD3 are less well-

characterized, though studies have demonstrated that BRD2 and BRD3 are involved in histone chaperoning to facilitate Pol II transcription through hyperacetylated nucleosomes and may share overlapping functions. 65 Brd4, on the other hand, has been found to interact with positive transcription elongation factor (P-TEFb), which phosphorylates and releases the promoter-proximally paused Pol II into active elongation during the transcription of target genes.62,66 Furthermore, functional studies of BRD4 have also demonstrated that BRD4 directly regulates the transcription of differentiation and growth-related genes by maintaining P-TEFb at the promoter sites throughout mitosis.67-69

A 2011 study by Zuber et al. was the first to show that BRD4 contributes to cancer maintenance using a negative selection RNAi screen performed in a mouse model of

MLL-rearranged AML and BET inhibition by the early generation small-molecule inhibitor JQ1.70 In another study, increasing BRD4 expression correlated positively with disease progression in multiple myeloma patients, but BRD2 and BRD3 did not.12 In

AML, there have been an abundance of data that suggests BET inhibition induces cells across a range of disease subtypes to undergo cell cycle arrest and apoptosis, presumably due to the downstream inhibition of genes promoting self-renewal and survival, such as

Myc, IRF8, and Bcl-2.70,71 72

5.2 Myc as a Therapeutic Target

Myc has been an attractive therapeutic target due to its modulation by multiple converging signaling cascades and functional studies that have shown the correlation between Myc overexpression and cancer pathogenesis, though strategies do not currently exist to directly inhibit the oncogenic activity of the protein. 73-75 A major obstacle in directly inhibiting Myc activity has been its structure and pleiotropic function as a transcription factor, and the lack of knowledge on the primary sequences of its active sites.76 With the advent of BET inhibitors, however, Myc has become a slightly more accessible therapeutic target. Across multiple studies, the Myc transcriptional program arrests in response to BET inhibition in various hematologic malignancies, including

AML.12,72,77

Similarly to BRD4, Myc has also been shown to recruit P-TEFb to release Pol II from the pause site and initiate elongation of a large population of actively transcribed genes that play key roles in balancing cell differentiation and cell self-renewal in hematopoietic stem cells.78,79 There is an overwhelming amount of data that support the notion that endogenous Myc promotes the entry of hematopoietic stem cells into a continuous cycle of cell division, thus maintaining the stem cell state.80,81 In hematologic malignancies,

Myc has been found to be overexpressed or amplified in many cases, and its activity has been shown to prevent myeloid cell differentiation. 76,81,82 In mouse models, transduction of bone marrow cells with Myc results in AML development83 and Myc knockdowns by short hairpin RNA (shRNA) resulted anti-leukemic lethality84. Likewise, inserting a

human Myc transgene into murine hematopoietic cells also induces AML, and inactivation of this transgene leads to leukemic regression.85

5.3 BET Inhibition in Overcoming AML Resistance

There is therefore overwhelming support the rationale for using BET inhibition to hamper Myc activity that is maintaining blast survival in AML through FLT3- independent signaling, particularly since Myc is the convergence point of many of the growth and proliferation signaling cascades, such as Ras/Raf/MEK, STAT5, and

PI3K/AKT pathways, in blasts. It is also important to clarify that although BRD4 has directly linked to leukemogenesis, none of the BET inhibitors currently being studied are specific for BRD4 yet and share similarly high affinities for BRD2/3/4, including JQ1,

OTX015, and PXL51107.

Dual FLT3 and BET Inhibition

6.1 A Prospective Synergistic Combination

Fiskus et al. previously reported that the combination of BET inhibition and FLT3 inhibition led to synergistic cytotoxicity in FLT3-ITD AML cells, including cell lines with acquired resistance to FLT3 inhibition.86 Although this was an important proof-of- principle, their studies were carried out using both cell lines and primary cells in suspension culture, which does not address the fundamental problem of stromal microenvironment-mediated resistance to FLT3 inhibition. Furthermore, their drug treatments were done with continuous exposure to JQ1, a clinically irrelevant BET inhibitor since it has only a half-life of 1 hour. This system would not be replicable in a human as continuous inhibition of BET proteins would likely lead to toxicity, since BET

protein functions are critical to global transcription. In order translate this in vitro combination into the clinic, a BET inhibitor is needed that can successfully suppress Myc levels in humans in a manner that parallels the in vitro studies.

Building on the previous findings, we reported that the combination of PLX51107,42 a structurally novel BET inhibitor currently in early phase clinical studies with clinically achievable pharmacokinetics, and the FLT3 inhibitor quizartinib results in synergistic cytotoxic effects in patient FLT3-ITD AML blasts on bone marrow stroma, and that this combination represents a clinically viable strategy to overcome microenvironment- mediated resistance to FLT3 inhibition.

6.2 PLX51107

PLX51107 is a novel BET inhibitor (owned by Plexxikon, a subsidiary of Daiichi

Sankyo) under clinical development in Phase 1b (NCT02683395) for advanced hematologic malignancies. PLX51107 is a pioneering agent in a new class of BET inhibitors due to its structural deviations from preceding BET inhibitors (Figure 23A), such as JQ1 and OTX015.2 It preferentially binds to BD1 of BRD2, BRD3, and BRD4 domains (Figure 23B). In a preclinical CLL study, PLX51107 was 10-fold more potent than OTX015 in a Ba/F3-induced mouse splenomegaly model IC50 and transcriptional effects in HEXIM1, a pharmacodynamic marker BRD4 engagement with targets, substantially outlived the presence of PLX51107 in plasma in vivo (Figure 23C).2

6.3 Quizartinib

Although we previously characterized gilteritinib as a uniquely effective FLT3 inhibitor against both FLT-ITD and FLT3-TKD mutations, there are a few reasons we chose to move forward in our combination studies with quizartinib instead. First, looking back at the cell line and primary sample cytotoxicity data, quizartinib is more potent than gilteritinib in FLT3-ITD AML. This brings us to the second reason, which is that, based on the KinomeScan data (Figure 24), quizartinib is much more selective for FLT3 as compared to gilteritinib. Since we are trying to investigate the combination of FLT3 inhibition with a BET inhibitor, it is important that we use a FLT3 inhibitor that has a narrowly selects for FLT3 so that the effects of the combination observed can be attributed to the dual inhibition of only FLT3 and BET..

A phase 2 randomized, open-label study (NCT01565668) compared the overall survival of patients with relapsed or refractory FLT3-ITD AML receiving 30/60 mg qd of quizartinib as a monotherapy versus conventional salvage chemotherapy. Plasma obtained from patients throughout the first cycle showed that approximately 350 ng/ml is the highest concentration of quizartinib achieved in patients receiving 60 mg at steady- state (Figure 25). Given that quizartinib is 560 mg/mol and the concentration in plasma is about 18-fold less than that in media, the equivalent concentration in in vitro studies in media is 50 nM. We therefore only explored doses of quizartinib up to 50 nM in media to further explore the combination of FLT3 and BET inhibitors.

6.4 In vitro Characterization of PLX51107

PLX51107 is a small molecule BET inhibitor with a unique pharmacokinetic profile that makes it an ideal candidate as a combination reagent. In order to confirm the sustained inhibition of the surrogate protein Myc, which is known to be regulated by

BRD4 activity, and characterize the therapeutic window, we developed a modified PIA using the OCI-AML3 cell line rather than the Molm14 line we use in FLT3 PIA assays.30

Figure 26 shows the dose dependent effects of PLX51107 in either Molm14 or OCI-

AML3 cells cultured in media. Since Molm14 cells are derived from human FLT3-ITD

AML cells, quizartinib treated suppresses FLT3 signaling, which in turn suppresses Myc expression. In contrast, OCI-AML3 cells express wild-type FLT3 and Myc expression is therefore unaffected by quizartinib. The use of OCI-AML3 cells in this PIA consequently eliminates any unaccountable or supplementary Myc suppression caused by a patient is also on a FLT3 TKI, and maintains the integrity of assessing in vivo BET inhibition by using Myc as a surrogate marker.

In Figure 27, we investigated the protein binding and potential bioavailability of

PLX51107 by comparing the PLX51107 concentrations at which 50% of Myc expression is inhibited (IC50) in RPMI media/10% FBS versus control (non-drugged) human plasma.

The immunoblot data shown are representative of three independent experiments. The

IC50 in media and plasma were 80 nM and 1.1 M, respectively, which suggest that

PLX51107 is approximately 93% plasma protein-bound. The equivalent of any dose in media would thereby be 13.75-fold less than that of in plasma.

6.5 Estimation of In Vivo BET Inhibition

Plexxikon Inc, conducted a multi-center 3+3 dose escalation study in adults with solid tumors and AML to determine the recommended phase II dose (A Phase 1b/2a, Two-

Part, Dose Escalation and Expansion Study to Assess the Safety, Pharmacokinetics,

Pharmacodynamics, and Preliminary Efficacy of PLX51107 in Subjects with Advanced

Hematological Malignancies and Solid Tumors, NCT02683395). The patients were orally administered 20mg to 160mg PLX51107 qd in six cohorts and plasma was collected at 9 total time points throughout Day 1 of Cycle 1 (including pre-dose on Day 2).

We performed the PIA assays for patients across the six cohorts through our modified

PIA to find the optimal dosing regimen and therapeutic window for combining

PLX51107 with quizartinib. Figure 28 exhibits representative PIA Myc immunoblots from a patient in cohorts 3 through 6. The trough levels by time point for each cohort are also portrayed in Figure 29. We found that although cohort 6 was dosed 160 mg qd, the length of Myc suppression was similar to that of Cohort 5 at 120 mg dosing – both lasting for approximately 6 hours. From these data, we can infer that a 6-hour cell treatment with a Myc suppressive dose of PLX51107 in an in vitro model would roughly imitate the pharmacokinetics of dosing AML patients at 120 mg qd. After interpolating the dose concentrations of all samples from the standard curve (Figure 30), we concluded that maximal Myc suppression in the PIA assays was observed at 3500 nM PLX51107 in plasma, which roughly translates to 250 nM in media for our in vitro studies. These pharmacokinetics data corroborate that intermittent Myc inhibition with once-daily

dosing of PLX51107 were promising as continuous inhibition of epigenetic regulators as significant as BET proteins would most likely cause intolerable toxicity in patients.

6.6 Screening for Synergy

Continuous exposure to quizartinib or PLX51107 as a single-agent or combinations of both drugs were used to identify patient samples of interest that demonstrated synergistic cytotoxic activity. Samples were collected either at the time of AML diagnosis or at relapse. A total of 22 patient samples (8 diagnostic samples and 14 relapse samples) were co-cultured with bone marrow stroma and screened through a 72-hour incubation during which the cells were treated with various concentrations of quizartinib only, PLX51107 only, or a combination of both. At the end of 72 hours, cell viability in each condition was assessed by MTT. We then calculated for synergy by comparing the combination with the single agents using Median Effect analysis and strictly defined a sample as synergistic when the combination index (CI) value was ≤ 0.1. Of these samples, ten samples fit our criterion and the trend among them was that they were all

FLT3mut+ TKI-naïve relapse samples (relapsed patients who had no history of a FLT3

TKI). These patient samples had no acquired resistance-conferring TKD mutations from previous prolonged FLT3+ TKI exposure. The screening MTT data from two representative synergistic samples are shown in Figure 31. The combination condition in patient AML6 exhibited a CI value of 0.028 while the CI in patient AML7 was 0.015.

The patient characteristics and mutation profiles in the ten samples of interest that exhibited synergistic anti-leukemic activity can be found in Table 3.

6.7 The Intermittent Combination System

With these seven synergistic patient samples, we next modeled trough levels of Myc suppression in the 120 qd PLX51107 regimen with intermittent exposure to PLX51107 while maintaining quizartinib at a steady-state concentration of 50 nM. We achieved this by treating the primary AML blasts in vitro with 50 nM quizartinib + 250 nM PLX51107 for 6 hours, washed the cells thoroughly, and re-suspended them in fresh media with 50 nM quizartinib for the remainder of the 18 hours in the day. This protocol was repeated every 24 hours and the comparative conditions (quizartinib only, PLX51107, or continuous combination of the two drugs) underwent the same washing and given fresh drug in media, as illustrated in Figure 32. Cell viability was analyzed by hemocytometer cell counting with Trypan blue after 72 hours of drug treatment. Figure 33 exhibits the cell count data for Patient 4423878 and Patient 84216044, which are representative of the data for the other synergistic samples. Overall, the intermittent PLX51107 combination was equally, or more effective, than the continuous combination in reducing AML blast viability. These results indicate that 6 hours of Myc suppression is sufficient in producing the same inhibitory effect on BET-target genes as the continuous exposure to both drugs,

and supports the claim in Ozer et al. that the transcriptional effects of PLX51107 are sustained for a while after the drug has been metabolized.2

6.8 Exploring Differential Gene Expression by RNA Expression Analysis

6.8.1 Myc

To investigate the genes responsible for the observed synergistic anti-leukemic effects of the PLX51107 and quizartinib combination, we performed transcriptome analysis on three synergistic relapsed patient samples identified through our screening MTT assay – patients AML6 (4423878), AML7 (84216044), and AML8 (56509991). To prepare the

RNA samples, the cells were treated on stroma with either 50 nM quizartinib, 250 nM

PLX51107, or the combination of both, and collected at two time points: at 6 hours prior to washing and the divergence of the two combination models, and at 24 hours, as outlined in Figure 30. Each set of patient samples was prepared in duplicates.

By RNA expression analysis, an average of 30 million 150-base long reads were mapped to the human reference genome, assembled per gene, and condensed into expression values for each gene mapped against the transcriptome for each treated condition that was compared to its time-matched control. To visualize differential gene changes between each condition at 6h or 24h, we generated a heatmap (Figure 34) displaying log(fold-change) > 2 using the Ingenuity Pathway Analysis (IPA). The most

interesting result, which fits in neatly with the findings others had observed with BET inhibition,12,71,87was that Myc was found to be the most down-regulated gene at 24 hour time points with 50 nM quizartinib (single agent) and both of the combination treatments.

Similarly, when we mapped the differentially expressed genes (DEGSs) to hallmark gene sets (Figure 35), we saw that Myc-targeted pathways were dramatically down-regulated in the same three conditions.

6.8.2 p63 and TNF

While Myc appears to be a pronounced genetic component of the synergy, there is no doubt that many others are involved. Two genes of interest that emerged from performing an iterative analysis of the RNA-seq data were TP63 and the TNF family.

TP63 (tumor protein 63, better known as p63) is a p53 homologue that is often associated with poorer prognoses in certain cancers. Overexpression of p63, particularly the isoform ΔNp63, is associated with increased cell proliferation by overriding growth arrest signals and exerting a dominant negative effect on p53 and other homologues, thereby protecting the cells from p73-dependent apoptosis.88 89,90 91 92 Furthermore,

ΔNp63 is also known to contribute to tumorigenesis through the activation of the β- catenin signaling pathway and negative regulation of the AKT pathway, which promotes apoptosis. 93 94 In Figure 37, we looked at the expression profile of downstream DEGs from TP63 at the 6h time point. Myc is a target of the β-catenin pathway, so it is not surprisingly that the downregulation of Myc correlates with decreased TP63 signaling.

The downregulation in protein levels associated with TNF signaling at the 6 hour time point with the combination condition was an unexpected discovery (Figure 36).

AML is often associated with high production of pro-inflammatory cytokines, including

TNF-α, and thus, decreasing the level of TNF-α signaling may be crucial since TNF-α can auto-regulate through activation of NF-kB and MAPK, which also happen to be downstream of the FLT3 receptor.95

6.9 Corresponding qPCR and Immunoblots

We further validated our sequencing results by qPCR to quantify Myc mRNA expression in the same exact samples analyzed in RNA sequencing. These data shown in

Figure 37 are the average of AML6 and AML7 sets of duplicates. In concordance with

RNA-seq data, the combination treatment at 6 hours significantly reduced Myc mRNA.

At the 24 hour time point, however, the intermittent PLX51107 combination condition yields a partial rescue of Myc mRNA. Here, the 24h intermittent combination again looks similar to the 24h quizartinib (single agent) profile levels due to the washout of

PLX51107 at 6 hours.

We also investigated the mRNA levels of BCL6, since Ma et al. had previously reported that FLT3 inhibition increases BCL6 levels.96 BCL6 represents an oncogenic pro-survival factor in hematologic malignancies by suppressing p53 function.97,98 Since

BRD4 is associated with the superenhancers of the BCL6 gene,99 we had hoped that by inhibiting BRD4 by PLX51107, we would undermine the increased expression of BCL6

associated with FLT3 inhibition by quizartinib. This was, however, not the case – in

Figure 37, any condition that involved quizartinib increased the transcription level of

BCL6, whereas PLX51107 did not appear to influence BCL6 levels.

On the protein level, our immunoblots in Figure 38 confirmed that the transcriptional effects of PLX51107 outlast the drug. At 24 hours, the continuous combination condition saw an 84% reduction in Myc expression compared to the time-matched control, and 24h quizartinib and 24h intermittent combination both saw around a 65% decrease by densitometry. Although the Myc expression from the intermittent combination looked similarly to quizartinib (single agent) at 24 hours, the synergistic cytotoxic activity against blast on stroma were still apparent in the MTT assays.

p63 is a p53 homolog and has implications in cancer development. Since p63 was also identified as a gene of interest by RNA-seq (pictured in the heatmap in Figure 34), we also investigated p63 protein levels. The p63 blots were also intriguing as we saw an apparent decrease in the ΔNp63α isoform (66 kD), which is the most abundant p63 isoform and lacks the N-terminal transactivation domain. In Figure 38, both the 24h continuous combination and 24h intermittent combination conditions saw a 96% reduction in ΔNp63α expression compared to the control. The data thus far postulate that

Myc and ΔNp63α are involved in the mechanism of action in the synergistic activity with the combination treatments, though there are most certainly other genes involved as well.

From a separate iterative analysis, we also found evidence to suggest that TNF receptor superfamily 8 (TNFRSF8), also known as CD30, may also play a role in the synergistic activity against blasts on stroma, although there is little known about the

function of CD30 in the literature. In our protein assay, we discovered that both of the combination conditions induced a dramatic decrease in CD30 expression – the continuous combination decreased CD30 expression by 66% and the intermittent combination by 49%. Though we are unsure of the implications of these data, CD30 may be worth characterizing in future studies. The data shown in Figure 38 is from patient

AML7, but each blot is representative of the immunoblots performed with blasts from

AML6 and AML8 as well.

6.10 Validation of the intermittent Combination System

In addition, Figure 39 from our transcriptome data further confirmed that the intermittent combination system was in fact depleted of PLX51107 after the washing step at 6 hours. While the quizartinib and intermittent combination affects nearly 5000 DEGs at the 6 hour time point, the intermittent combination only alters around 1500 genes at 24 hours, compared to the time-matched controls. At 24 hours, the intermittent combination condition adopts a genetic profile more like that of quizartinib (single agent). Although the presence of PLX51107 is gone at the 24-hour time point with the intermittent combination, our MTT and cell count analyses with the intermittent condition prove that the transient genetic effects of PLX51107 are sufficient to produce a synergistic cytotoxic outcome in AML cells.

6.11 Clinical Implications of Intermittent BET Inhibition

We studied the combination of FLT3 and BET inhibition in FLT3-ITD AML to better understand if the approach could undermine the resistance conferred by the stromal microenvironment that AML blasts reside in. The signaling pathways downstream of

FLT3 and BET converge to promote the transcription of pro-survival factors, such as

Myc. The dual combination blocks these two these oncogenic modulatory points to reduce survival and growth signaling, and potentially hinder the sustained signaling that maintains the blasts on stroma. While 60 mg quizartinib is able reach a steady state in vivo based on clinical trial data, continuous BET inhibition is likely to intolerable in humans due to the importance of BET proteins in global transcription across all tissue types.

In our studies, we found that our intermittent combination model, which parallels the

6-hour length of Myc suppression seen in vivo with 120 mg once-daily dosing

PLX51107, is sufficient to produce the same level of synergistic cytotoxicity as continuous exposure to both drugs. This is due to the lasting transcriptional effects of

PLX5110,2 and the intermittent inhibition of Myc activity will likely be more tolerable in patients than the continuous protocol.

The use of Myc as a surrogate in the PIA assays to monitor BET-targeted inhibition in vivo was based on previous observations that it was the most down-regulated gene in response to JQ1 or OTX015. This was further validated by the fact that Myc was the

most affected gene in the RNA-seq analyses with the primary samples that responded synergistically to the combination of quizartinib and PLX51107. These studies have collectively identified the inhibition of Myc-dependent signaling as the key contributor to the anti-cancer activity from BET inhibition.

In addition, we also identified the ΔNp63α isoform as a likely component of the synergy resulting from dual BET and FLT3 inhibition. Though p63 has long been linked to limb formation and epidermal morphogenesis,100there is a debate on whether p63 is a tumor suppressor gene or oncogene.101 The locus at p63 produces multiple protein products from alternate mRNA splicing, so while the full-length transactivation domain

(TA) isoform of p63 has structural and functional similarity to wild-type p53 (a tumor suppressor), the ΔNp63α isoform confers antagonistic activity against p53 and TAp63 and is thus more of an oncogene. 101,102 Though no link between ΔNp63α expression and hematologic malignancies currently exists, it would be interesting to conduct functional studies of this protein since it is associated with others forms of cancer metastasis.

Given that the patient samples that responded synergistically to the intermittent combination in our studies were patients who had no previous FLT3 TKI history, any clinical protocol should move forward with this type of patient profile in the clinic with

60 mg quizartinib and 120 mg PLX51107 qd in vivo, which are the equivalent to 50 nM quizartinib and 250 nM PLX51107 in vitro. Though our primary blast and stroma co- culture system is not a perfect in vivo model, the synergistic lethality from the combination of FLT3 and BET inhibition nevertheless represents a prospective clinical strategy to overcome stromal-associated resistance and reduce the AML blast burden

more so than with a FLT3 TKI alone, so that patients can achieve a deeper and higher quality of remission for a successful bone marrow transplant.

Figure 23. Properties of BET Inhibitor PLX51107. (A) Structure of PLX51107. (B) PX51107 preferentially binds BD1 over BD2. (C) Transcriptional changes (represented by HEXIM1) in MV4-11 tumor xenograft substantially outlive the plasma drug levels of PLX51107. A single dose of PLX51107 was administered. AUC0–24h = 90,100 ng · h/mL (or 205 μmol/L · h), half-life T1/2 = 2.8 hours. Reprinted from Ozer et al.2

Gilteritinib

FLT3

Quizartinib

FLT3

Figure 24. KinomeScan plots for gilteritinib and quizartinib, which exemplify how targeted each drug is for FLT3.

Figure 25. Geometric mean trough plasma concentration of quizartinib and its metabolite AC886. Steady state is reached by cycle 1 day 15. At 60 mg qd, the combined concentration of quizartinib and AC886 is approximately 500 nM (in plasma), which corresponds to 50nM in media in vitro.

Figure 26. Effect of quizartinib on Myc inhibition in Molm14 and OCI-AML3 cell lines. Molm14 harbors a FLT3-ITD mutation, whereas OCI-AML3 cells possess wild-type FLT3 receptors (non-FLT-ITD). Since the objective is to isolate the contributions of FLT3 and BET inhibitors, Molm14 is not a suitable cell line for studying BET and FLT3 combinations.

OCI-AML3 in Media

Myc

Actin

PLX51107(nM)

Figure 27. Effects on Myc expression by PLX51107 in media versus in plasma. The IC50 in media is 80 nM and the IC50 in plasma is 1.1 M, which imply that PLX51107 is 93% protein bound in plasma. Standard curves were generated using average inhibition values from three separate dose response experiments.

Cohort 3 – 60mg Subject 11 C1D1 C1D2

Timepoint (h) Pre .5 1 2 3 5 7 9 Pre

Myc

Actin

Cohort 4 – 90mg Subject 15 C1D1 C1D2 Timepoint (h) Pre .5 1 2 3 5 7 9 Pre

Myc

Actin

Cohort 5 – 120mg

Subject 23 C1D1 C1D2 Timepoint (h) Pre .5 1 2 3 5 7 9 Pre

Myc

Actin

Cohort 6 – 160mg Subject 29 C1D1 C1D2 Timepoint (h) Pre .5 1 2 3 5 7 9 Pre

Myc

Actin

Figure 28. Myc plasma inhibitory assays performed using plasma from patients who received 60 mg, 90 mg, 120 mg, or 160 mg PLX51107. Cohorts 5 and 6 saw maximal Myc suppression that lasted for approximately 6 hours. 79

P L X 1 2 2 -0 1 C o h o r t 4 2 5 0

c 2 0 0 y

o 1 5 0

t n

o 1 0 0

5 0 %

e 5 1 2 3 5 7 9 e r . r p p 1 2 D D 1 1 C C T im e p o in ts (h )

PLX122-01 Cohort 5 PLX122-01 Cohort 6 250

300 c

200 c

l 200

o 150

o o

C 100

f 100

% 50 %

0 0 e 5 1 2 3 5 7 9 e r . r re .5 1 2 3 5 7 9 re p p p p 1 2 1 2 D D D D 1 1 1 1 C C C C Timepoints (h) Timepoints (h)

Figure 29. Trough levels by time point from the Myc plasma inhibitory assays in Cohorts 4 through 6.

Figure 30. The concentrations of PLX51107 were interpolated from the standard curve generated from the OCI-AML3 dose response in plasma. Each subject had nine plasma samples collected at different time points of cycle 1 day 1.

Patient ID Cytogenetic Profile FLT3-ITD Burden AML6 DNMT3a 93 bp, AR 0.41 AML7 NOTCH2, WT1, KMT2D, IDH1 93 bp, AR 2.73 AML8 DNMT3a, NPM1 45 bp, AR 2.6 AML9 NPM1, KMT2A, CBL, ASXL1 57 bp, AR 0.89 AML10 WT1 96 bp, AR 0.6 AML11 NPM1, WT1 36 bp, AR 2.54 AML12 NPM1, IDH1, DNMT3a, WT1 69 bp, AR 2.68 AML13 DNMT3a, NPM1 54 bp, AR 1.17 AML14 NPM1 45 bp, AR 6.41 AML15 NPM1 33 bp, AR 14.9 Table 3. Patient profiles in PLX51107 and quizartinib combination studies.

A M L 6 C I = .0 2 8

1 0 0

y t

c i

l 5 0

le 0 7 n c 2 0 o i 2 1 ti h C 1 a e A 5 n V X i M L b n P m 0 o 5 M C n 0 5 2

A M L 7 C I = .0 1 5

1 5 0

i t

c 1 0 0

b a

t 5 0

le 0 7 n c 2 0 o i 2 1 ti h C 1 a e A 5 n V X i M L b n P m 0 o 5 M C n 0 0 2

Figure 31. Cell viability after 72 hours of exposure to continuous 50 nM QUIZARTINIB and 250 nM PLX51107. Patients AML6 and AML7 are both FLT3-ITD relapse samples and the data shown are representative of what was defined as a “synergistic” sample. 83

6h 24h Vehicle

AC220

PLX51107

AC220+PLX51107 (continuous) AC220+PLX51107 (continuous)

wash AC220+PLX51107 (pulsatile) AC220 only

Figure 32. Schematic diagram providing the drug conditions used to study the effects of dual FLT3 and BET inhibition. Five conditions: vehicle (.1% DMSO in media), 50 nM quizartinib (single-agent), 250 nM PLX51107 (single-agent), continuous exposure to 50 nM quizartinib and 250 nM PX51107, and intermittent combination (6 hours of 50 nM quizartinib and 250 nM PX51107, washout, then 18 hours of 50 nM quizartinib only).

A M L 6

1 0 0

a i

V 5 0

le b 7 n 7 c i 0 o 0 i n 1 i 1 ti t b h r 1 a 1 i e 5 n 5 n V a X i ti iz b X r L L a u P m P z Q i o t u C n Q s te u it s o u u m o r u in e t t in n In t o n C o C

A M L 7

1 5 0

i l

i 1 0 0

l e

C 5 0

le b 7 n 7 c i 0 o 0 i n 1 i 1 ti t b h r 1 a 1 i e 5 n 5 n V a X i ti iz b X r L L a u P m P z Q i o t u C n Q s te u it s o u u m o r u in e t t in n In t o n C o C

Figure 33. Comparison of cell counts for AML6 and AML7 samples under continuous or intermittent combination treatment conditions. Both combination conditions produced similar levels of cytotoxicity against patient blasts on stroma. 85

Figure 34. Heatmap generated from the RNA-seq study that looked at differentially expressed genes that had log-fold changes >2 as compared to the time-matched control with each treatment condition.

Figure 35. Differentially expressed hallmark gene sets from the RNA expression analysis.

Upstream regulator Down stream DEGs

TP63↓

TNF↓

Figure 36. TP63 (p63) and TNF signaling are both predicted to be involved in the downstream effects of the combination of 50 nM quizartinib and 250 nM PLX51107 at the 6-hour time point.

B C L 6

T im e (h ): 0 6 2 4

) 4

n a

h 3

d l

o 2

(

R 1

e e 7 n e 0 7 l l b l 2 n 7 c c i 0 o c 0 o 0 i i n 1 ti i 2 1 i 1 ti t ib h h r 1 a h C 1 a 1 e e 5 n e A 5 n 5 n V V a X i V X i ti iz b b X r L L L a u P m P m P z Q o i o t u C C n Q s te u it s o u u m o r u in e t t in n In t o n C o n = 4 C

Figure 37. Myc and BCL6 RNA quantification from AML6 and AML7 FLT3-ITD relapse samples by qPCR.

AML7

c-Myc Densitometry (% of control): 100 50 34 14 100 36 54 16 35

p63 Densitometry (% of control): 100 70 68 78 100 34 16 7 6 CD30 Densitometry (% of control): 100 85 88 69 100 73 85 34 51

Actin

6h 24h

Figure 38. Myc, p63 and CD30 immunoblots from patient AML7's blasts on stroma. Myc protein expression correlates with the RNA-seq and qPCR data. p63 and CD30 are proteins of interest as both see dramatic decreases in protein expression in the combination treatment conditions.

Figure 39. Total number of differentially affected genes by each condition. The 24 hour time point provided distinctive data between the two combination models. While the continuous combination model appeared additively affect genes that were also affected by the single agents, the intermittent combination adopted a profile similar to that of QUIZARTINIB (single agent).

A Novel BET Inhibitor, PLX2853

7.1 Characterization of PLX2853

By the time our studies with PLX51107 and quizartinib had completed, Plexxikon had developed a second compound, PLX2853, that has high affinity for the BD2 of the

BET proteins. PLX51107 preferentially binds to BD1 of BET proteins.2

Plexxikon conducted a phase 1b/2a dose escalation study to assess PLX2853 in patients with advanced malignancies (NCT03297424). Similar to our PLX51107 studies, we first determined how protein-bound PLX2853 is by comparing the IC50 of the drug in media and the IC50 in plasma using OCI-AML3 cells (Figure 40). The IC50 was

14nM in media and 84nM in plasma, which translates to the drug being 83% protein- bound in plasma. The multiplier from a concentration in media to one in plasma is

therefore 6x. Compared to PLX51107, PLX2853 is 5x more potent in media and 13x more potent in plasma. This may be related to the drugs’ preferences for BD1 or BD2 of the BET proteins.

A representative Myc PIA blot from each cohort is shown in Figure 41 and the trough levels by time point in each cohort in Figure 42. Cohort 4 (40 mg) and cohort 5 (80 mg) suppress Myc in the PIA assays for ~6-7 hours before PLX2853 is metabolized, similar to the two highest dosed cohorts in the PLX51107 trial. The in vivo concentrations of

PLX2853 in all trial patients were interpolated from the plasma standard curve generated and plotted in Figure 43. Of note, the patient sample points formed a curve that falls below the expected plasma standard curve, which suggests that there is an active metabolite of PLX2853 present in the patient plasma suppressing BET more than what would be expected with PLX2853 alone.

7.2 Cytotoxicity of PLX51107 vs PLX2853

Using a previously screened synergistic FLT3-ITD patient sample, we compared the anti-leukemic activity against blasts on stroma with continuous exposure to FLT3 inhibition (quizartinib) and BET inhibition (PLX51107 or PLX2853) (Figure 43). 250 nM PLX2853 alone had no effect on blast viability on stroma, as expected since

PLX51107 as a single agent was equally ineffective. However, the combination of 50 nM quizartinib and 250 nM PLX2853 was slightly more cytotoxic than the comparative combination with 250 nM PLX51107. This was also unsurprising since we had noted that

PLX2853 is much more potent than PLX51107, which again may be due to the difference in preferential binding sites on the BET proteins.

7.3 PLX2853 May Induce Myelosuppression

In the final part of our comparative studies, we investigated each BET inhibitor’s capacity to hinder normal hematopoiesis using colony forming cell (CFC) assays. The equivalent of 250 nM PLX51107 is approximately 50 nM PLX2853, since our IC50 in

OCI-AML3 had suggested that PLX2853 is 5-fold more potent than PLX51107. The data in Figure 44 indicates that at 50 nM PLX2853, the division of BFU-E (erythroid progenitors) would be hindered by 50% and CFU-GM (granulocyte progenitors) by 35%.

PLX51107 would require >1000 nM to prevent hematopoiesis to the same degree. In combination with 50 nM quizartinib, PLX2853 also drastically reduced BFU-E colonies by >50% starting from 10 nM, whereas PLX51107 with 50 nM quizartinib did not have a comparative effect until >1000 nM. One caveat of this assay, however, is that we were not able to treat using a intermittent combination model due to the technical difficulty of washing the colonies. The colonies here were exposed to the specified drugs continuously for 10 days prior to colony counting.

7.4 Future Directions with PLX2853

In summary, while PLX2853 is a more potent and BD2-targeted drug than

PLX51107, the additional potency in Myc suppression does not appear to translate to a higher degree of synergistic cytotoxicity in combination with 50 nM quizartinib. In addition, our studies indicate that there is an unidentified active metabolite from

PLX2853 in vivo and that there is a high probability for myelosuppression in the clinic since PLX2853 has a pronounced negative impact on normal hematopoiesis at more than

10x lower doses than PLX51107. In the future, more studies need to be conducted to understand the molecular and clinical consequences of BD1 selectivity versus BD2.

Media

PLX2853 (nM ) 0 5 10 25 50 100 250 500

Myc

Actin

IC50 = 14nM

Plasma

PLX2853 (nM) 0 25 50 100 150 200 500 1000 2000

Myc

Actin

IC50 = 84nM

Figure 40. PLX2853 dose responses in media and plasma with OCI-AML3 cells.

Cohort 1 – 5 mg Subject 1 C1D1 C1D2 Timepoint (h) Pre .5 1 2 3 5 7 9 Pre

Myc

Actin

Cohort 2 – 10 mg Subject 7 C1D1 C1D2 Timepoint (h) Pre .5 1 2 3 5 7 9 Pre

Myc

Actin

Cohort 3 – 20 mg Subject 9 C1D1 C1D2 Timepoint (h) Pre .5 1 2 3 5 7 9 Pre

Myc

Actin

Cohort 4 – 40 mg Subject 13 C1D1 C1D2 Timepoint (h) Pre .5 1 2 3 5 7 9 Pre

Myc

Actin

Cohort 5 – 80 mg Subject 15 C1D1 C1D2 Timepoint (h) Pre .5 1 2 3 5 7 9 Pre

Myc

Actin

Figure 41. Myc plasma inhibitory assays were performed on cohorts 1 through 5 in an early phase clinical trial for PLX2853. Each blot is representative.

Cohort 3 – Subject 9

C1D1 C1D2 Timepoint (h) Pre .5 1 2 3 5 7 9 Pre

Myc

Actin

Cohort 4 – Subject 13

C1D1 C1D2 Timepoint (h) Pre .5 1 2 3 5 7 9 Pre

Myc

Actin

Cohort 5 – Subject 15

C1D1 C1D2 Timepoint (h) Pre .5 1 2 3 5 7 9 Pre

Myc

Actin

P L X 1 2 4 -0 1 C o h o rt 1 P L X 1 2 4 -0 1 C o h o r t 2

4 0 0 3 0 0 c

c y

3 0 0 y

l 2 0 0

o o

r S u b 1

r S u b 4 t

t n

2 0 0 n

S u b 2 S u b 5

S u b 3 S u b 6 f

f 1 0 0

1 0 0

% %

0 0

e .5 1 2 3 5 7 9 e e 5 1 2 3 5 7 9 e r r r . r p p p p 1 2 1 2 D D D D 1 1 1 1 C C C C T im e p o in ts (h ) T im e p o in ts (h )

PLX124-01 Cohort 3 PLX124-01 Cohort 4

250 150 c c Sub 7 Sub 11

200 y y

M Sub 12 M

Sub 8 l 100

o o

150 r Sub 13 r

Sub 9 t

n n

Sub 10 o

C 100 f

% 50

0 0 re .5 1 2 3 5 7 9 re re .5 1 2 3 5 7 9 re p p p p 1 2 1 2 D D D D 1 1 1 1 C C C C Timepoints (h) Timepoints (h)

PLX124-01 Cohort 5

150

c y

M Sub 14

l 100

r Sub 15 t

n Sub 16

o C Sub 17

f 50

re .5 1 2 3 5 7 9 re p p 1 2 D D 1 1 C C Timepoints (h)

Figure 42. Trough levels by time point and subjects in cohorts 1 through 5 for the PLX2853 study.

S ta n d a rd (C trl P la s m a ) S u b 1 C o h o rt 1 P L X 1 2 4 -0 1 T ria l P IA S a m p le s S u b 2 Interpolated PLX2853 PIA S ta n d a rd (C trl P la s m a ) 5 m g (P L X 2 5 8 3 ) S u b 1 S u b 3 Samples C o h o rt 1 P L X 1 2 4 -0 1 T ria l P IA S a m p le s S u b 2 S u b 4 1 5 0 5 m g (P L X 2 5 8 3 ) S u b 3 C o h o rt 2 S u b 5 S u b 4 1 5 0 1 0 m g C o h o rt 2 S u b 5 S u b 6 1 0 m g S u b 6 S u b 7 S u b 7 S u b 8 C o h o rt 3 S u b 8 C o h o rt 3

c 2 0 m g y c S u b 9 1 0 0 2 0 m g y S u b 9

1 0 0 c

l S u b 1 0

r l t S u b 1 1

t S u b 1 1

n n

r C o h o rt 4 o

t S u b 1 2 C o h o rt 4 o

C S u b 1 2 n

4 0 m g f

C S u b 1 3

o 4 0 m g

o f S u b 1 3

5 0 C o

S u b 1 4

f 5 0

o C o h o rt 5 S u b 1 4

% S u b 1 5

% S u b 1 6 8 0 m g S u b 1 5 C o h o rt 5 S u b 1 7 S u b 1 6 8 0 m g

0 S u b 1 7 0 5 0 0 1 0 0 0 1 5 0 0 2 0 0 0 P L X 2 5 8 3 (n M ) 0 0 5 0 0 1 0 0 0 1 5 0 0 2 0 0 0 S ta n d a rd IC 5 0 : 8 1 .2 n M P L X 2 5 8 3 (n M ) [PLX2853 nM] Standard IC50: 81.2 nM S ta n d a rd c u rv e r = .9 7 Standard curve r = 0.97

S ta n d a rd IC 5 0 : 8 1 .2 n M FigureS ta n43d a. rPLX2853d c u rv e r =concentrations .9 7 from every patient sample interpolated from the standard curve and plotted. The actual ex vivo curve appears lower than the expected plasma standard curve, indicating that an active metabolite from PLX2853 may be present in the patient samples.

Figure 44. Comparison of the combination effects with quizartinib with either PLX51107 or PLX2853. The patient blasts were continuously exposed to the indicated drug conditions for 72 hours prior to the cell viability assessment by MTT.

100

1.6 PLX51107 Colony Forming Assay 1.6 PLX2853 Colony Forming Assay 1.4 1.4

1.2 1.2 1.0 1.0 BFU 0.8 BFU 0.8 CFU 0.6 CFU 0.6 0.4 0.4 0.2 0.2 0.0 0.0 0 nM 10 nM 25 nM 50 nM 100 nM 250 nM 500 nM 1000 0 nM 2.5 nM 5 nM 10 nM 25 nM 50 nM 75 nM 100 nM nM PLX51107 + 50 nM Quizartinib PLX2853 + 50 nM Quizartinib

1.6 1.6 1.4 1.4 1.2 1.2 1.0 1.0 0.8 BFU 0.8 BFU

0.6 CFU 0.6 CFU 0.4 0.4 0.2 0.2 0.0 0.0 0 nM 10 nM 25 nM 50 nM 100 nM 250 nM 500 nM 1000 0 nM 2.5 nM 5 nM 10 nM 25 nM 50 nM 75 nM 100 nM nM Figure 45. Colony forming cell assays to evaluate the effect of PLX2583 and PLX51107 (alone or in combination with 50 nM quizartinib) on normal hematopoiesis.

101

Concluding Remarks

In this report, we have built upon the proof-of-principle study by Fiskus et al., in which they investigated the combination of a FLT3 inhibition and a BET inhibitor, quizartinib and JQ1, respectively, in cells cultured in suspension.86 We have established a primary AML blast and healthy donor bone marrow stroma co-culture system that more closely parallels the in vivo microenvironment. [This system has been foundation in every study we conducted as we had mentioned in Chapters 3 and 4 that, although primary

AML blasts in suspension in vitro are completely dead with drug, a good portion of primary AML blasts cultured on stroma remain viable.]

In Chapter 5, we characterized a novel BET inhibitor, PLX51107, that has more clinically applicable pharmacokinetic properties than its predecessors, such as JQ1 and

OTX015. Though we had described gilteritinib as an efficacious FLT3 TKI that has

102

activity against resistance-conferring TKD mutations in Chapter 2, we selected the agent quizartinib to combine with PLX51107 due to its higher selectivity as a type II FLT3 inhibitor and its greater potency.

Because we postulated that continuous exposure to dual FLT3 and BET inhibition in vivo would most likely cause adverse events and consequential non-compliance in patients, we established an in vitro protocol to imitate the once-daily dosing of 120 mg

PLX51107 with 60 mg quizartinib. This “intermittent combination” protocol exposed primary AML cells to stroma to 6 hours of intermittent PLX51107 with continuous quizartinib for 24 hours. We discovered that although the intermittent combination adopted a more quizartinib-only RNA and protein profile at the end of the 24 hours, the transcriptional effects of targeted-BET inhibition combined with FLT3 inhibition produced synergistic cytotoxicity in the majority of AML samples from patients who had no prior FLT3 TKI usage. Though Myc and p63 were identified as likely players behind the synergy, there are likely more that will have to be explored in later studies. In addition to discerning other contributing components, additional studies on the epigenetic mechanisms and protein interactions will also need to be investigated.

Lastly, although PLX2853 may represent a more selective and potent BET inhibitor, it will be important to run more in vitro toxicity analyses prior to any clinical studies due to the possible presence of an unknown active metabolite and the potential negative impact on normal hematopoiesis.

103

In conclusion, the combination of PLX51107, a structurally novel BET inhibitor currently in early phase clinical studies, and the FLT3 inhibitor quizartinib results in synergistic cytotoxic effects in FLT3-ITD AML blasts on bone marrow stroma, and that this combination represents a clinically viable strategy to overcome microenvironment- mediated resistance to FLT3 inhibition.

104

References

1. Stone, R.M., et al. The Multi-Kinase Inhibitor Midostaurin (M) Prolongs Survival Compared with Placebo (P) in Combination with Daunorubicin (D)/Cytarabine (C) Induction (ind), High-Dose C Consolidation (consol), and As Maintenance (maint) Therapy in Newly Diagnosed Acute Myeloid Leukemia (AML) Patients (pts) Age 18-60 with FLT3 Mutations (muts): An International Prospective Randomized (rand) P-Controlled Double-Blind Trial (CALGB 10603/RATIFY [Alliance]). Blood 126, 6 (2015). 2. Ozer, H.G., et al. BRD4 Profiling Identifies Critical Chronic Lymphocytic Leukemia Oncogenic Circuits and Reveals Sensitivity to PLX51107, a Novel Structurally Distinct BET Inhibitor. Cancer discovery 8, 458-477 (2018). 3. DeVita, V.L., Theodore S.; Rosenberg, Steven A. . Lymphomas and Leukemias: Cancer: Principles & Practice of Oncology, (Wolters Kluwer, 2015). 4. De Kouchkovsky, I. & Abdul-Hay, M. ‘Acute myeloid leukemia: a comprehensive review and 2016 update’. Blood Cancer Journal 6, e441 (2016). 5. Pratz, K.W. & Levis, M. How I treat FLT3-mutated AML. Blood 129, 565-571 (2017). 6. Ding, L., et al. Clonal evolution in relapsed acute myeloid leukaemia revealed by whole-genome sequencing. Nature 481, 506 (2012). 7. Levis, M. & Small, D. FLT3: ITDoes matter in leukemia. Leukemia 17, 1738- 1752 (2003).

105

8. Gilliland, D.G. & Griffin, J.D. The roles of FLT3 in hematopoiesis and leukemia. Blood 100, 1532-1542 (2002). 9. Stirewalt, D.L. & Radich, J.P. The role of FLT3 in haematopoietic malignancies. Nature reviews. Cancer 3, 650-665 (2003). 10. Small, D. FLT3 mutations: biology and treatment. Hematology. American Society of Hematology. Education Program, 178-184 (2006). 11. Small, D., et al. STK-1, the human homolog of Flk-2/Flt-3, is selectively expressed in CD34+ human bone marrow cells and is involved in the proliferation of early progenitor/stem cells. Proceedings of the National Academy of Sciences of the United States of America 91, 459-463 (1994). 12. Delmore, J.E., et al. BET bromodomain inhibition as a therapeutic strategy to target c-Myc. Cell 146, 904-917 (2011). 13. Dosil, M., Wang, S. & Lemischka, I.R. Mitogenic signalling and substrate specificity of the Flk2/Flt3 receptor tyrosine kinase in fibroblasts and interleukin 3-dependent hematopoietic cells. Molecular and cellular biology 13, 6572-6585 (1993). 14. Takahashi, S. Downstream molecular pathways of FLT3 in the pathogenesis of acute myeloid leukemia: biology and therapeutic implications. Journal of hematology & oncology 4, 13 (2011). 15. Zhang, S., et al. Essential role of signal transducer and activator of transcription (Stat)5a but not Stat5b for Flt3-dependent signaling. The Journal of experimental medicine 192, 719-728 (2000). 16. Mizuki, M., et al. Suppression of myeloid transcription factors and induction of STAT response genes by AML-specific Flt3 mutations. Blood 101, 3164-3173 (2003). 17. Lyman, S.D. & Jacobsen, S.E. c-kit ligand and Flt3 ligand: stem/progenitor cell factors with overlapping yet distinct activities. Blood 91, 1101-1134 (1998). 18. Tse, K.F., Mukherjee, G. & Small, D. Constitutive activation of FLT3 stimulates multiple intracellular signal transducers and results in transformation. Leukemia 14, 1766 (2000). 19. Levis, M. FLT3 mutations in acute myeloid leukemia: what is the best approach in 2013? Hematology. American Society of Hematology. Education Program 2013, 220-226 (2013). 20. Stirewalt, D.L., et al. Size of FLT3 internal tandem duplication has prognostic significance in patients with acute myeloid leukemia. Blood 107, 3724-3726 (2006). 21. Leung, A.Y., Man, C.H. & Kwong, Y.L. FLT3 inhibition: a moving and evolving target in acute myeloid leukaemia. Leukemia 27, 260-268 (2013). 22. Smith, C.C., et al. Heterogeneous resistance to quizartinib in acute myeloid leukemia revealed by single-cell analysis. Blood 130, 48-58 (2017). 23. Daver, N., et al. Secondary mutations as mediators of resistance to targeted therapy in leukemia. Blood 125, 3236-3245 (2015).

106

24. Liu, Y. & Gray, N.S. Rational design of inhibitors that bind to inactive kinase conformations. Nature Chemical Biology 2, 358 (2006). 25. Smith, C.C., Lin, K., Stecula, A., Sali, A. & Shah, N.P. FLT3 D835 mutations confer differential resistance to type II FLT3 inhibitors. Leukemia 29, 2390 (2015). 26. Smith, B.D., et al. Single-agent CEP-701, a novel FLT3 inhibitor, shows biologic and clinical activity in patients with relapsed or refractory acute myeloid leukemia. Blood 103, 3669-3676 (2004). 27. Breems, D.A., Blokland, E.A., Neben, S. & Ploemacher, R.E. Frequency analysis of human primitive haematopoietic stem cell subsets using a cobblestone area forming cell assay. Leukemia 8, 1095-1104 (1994). 28. Pratz, K.W., et al. FLT3-mutant allelic burden and clinical status are predictive of response to FLT3 inhibitors in AML. Blood 115, 1425-1432 (2010). 29. Galanis, A., et al. Crenolanib is a potent inhibitor of FLT3 with activity against resistance-conferring point mutants. Blood 123, 94-100 (2014). 30. Levis, M., et al. Plasma inhibitory activity (PIA): a pharmacodynamic assay reveals insights into the basis for cytotoxic response to FLT3 inhibitors. Blood 108, 3477-3483 (2006). 31. Takahashi, K., et al. Salvage therapy using FLT3 inhibitors may improve long- term outcome of relapsed or refractory AML in patients with FLT3-ITD. British journal of haematology 161, 659-666 (2013). 32. Kayser, S. & Levis, M.J. FLT3 tyrosine kinase inhibitors in acute myeloid leukemia: clinical implications and limitations. Leukemia & lymphoma 55, 243- 255 (2014). 33. Fischer, T., et al. Phase IIB trial of oral Midostaurin (PKC412), the FMS-like tyrosine kinase 3 receptor (FLT3) and multi-targeted kinase inhibitor, in patients with acute myeloid leukemia and high-risk myelodysplastic syndrome with either wild-type or mutated FLT3. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 28, 4339-4345 (2010). 34. Borthakur, G., et al. Phase I study of sorafenib in patients with refractory or relapsed acute leukemias. Haematologica 96, 62-68 (2011). 35. Galanis, A. & Levis, M. Inhibition of c-Kit by tyrosine kinase inhibitors. Haematologica (2014). 36. Pratz, K.W., et al. A pharmacodynamic study of the FLT3 inhibitor KW-2449 yields insight into the basis for clinical response. Blood 113, 3938-3946 (2009). 37. Smith, C.C., et al. Validation of ITD mutations in FLT3 as a therapeutic target in human acute myeloid leukaemia. Nature 485, 260-263 (2012). 38. Man, C.H., et al. Sorafenib treatment of FLT3-ITD(+) acute myeloid leukemia: favorable initial outcome and mechanisms of subsequent nonresponsiveness associated with the emergence of a D835 mutation. Blood 119, 5133-5143 (2012). 39. Weisberg, E., et al. Inhibition of mutant FLT3 receptors in leukemia cells by the small molecule tyrosine kinase inhibitor PKC412. Cancer Cell 1, 433-443 (2002).

107

40. Auclair, D., et al. Antitumor activity of sorafenib in FLT3-driven leukemic cells. Leukemia 21, 439-445 (2007). 41. Zarrinkar, P.P., et al. Quizartinib is a uniquely potent and selective inhibitor of FLT3 for the treatment of acute myeloid leukemia (AML). Blood 114, 2984-2992 (2009). 42. Park, I.K., et al. Inhibition of the receptor tyrosine kinase Axl impedes activation of the FLT3 internal tandem duplication in human acute myeloid leukemia: implications for Axl as a potential therapeutic target. Blood 121, 2064-2073 (2013). 43. Levis, M., et al. Results from a randomized trial of salvage chemotherapy followed by lestaurtinib for patients with FLT3 mutant AML in first relapse. Blood 117, 3294-3301 (2011). 44. Strati, P., et al. Phase I/II trial of the combination of midostaurin (PKC412) and 5- azacytidine for patients with acute myeloid leukemia and myelodysplastic syndrome. American journal of hematology 90, 276-281 (2015). 45. Ding, L., et al. Clonal evolution in relapsed acute myeloid leukaemia revealed by whole-genome sequencing. Nature 481, 506-510 (2012). 46. Williams, A.B., et al. Mutations of FLT3/ITD confer resistance to multiple tyrosine kinase inhibitors. Leukemia 27, 48-55 (2013). 47. Smith, C.C., Lin, K., Stecula, A., Sali, A. & Shah, N.P. FLT3 D835 mutations confer differential resistance to type II FLT3 inhibitors. Leukemia 29, 2390-2392 (2015). 48. Gotlib, J., et al. Efficacy and Safety of Midostaurin in Advanced Systemic Mastocytosis. The New England journal of medicine 374, 2530-2541 (2016). 49. Fabbro, D., et al. PKC412--a protein kinase inhibitor with a broad therapeutic potential. Anticancer Drug Des 15, 17-28 (2000). 50. Kitamura, T., et al. Establishment and characterization of a unique human cell line that proliferates dependently on GM-CSF, IL-3, or erythropoietin. Journal of cellular physiology 140, 323-334 (1989). 51. Levis, M. Midostaurin approved for FLT3-mutated AML. Blood 129, 3403 (2017). 52. Perl, A.E., et al. Selective inhibition of FLT3 by gilteritinib in relapsed or refractory acute myeloid leukaemia: a multicentre, first-in-human, open-label, phase 1-2 study. The Lancet. Oncology 18, 1061-1075 (2017). 53. Sexauer, A., et al. Terminal myeloid differentiation in vivo is induced by FLT3 inhibition in FLT3/ITD AML. Blood 120, 4205-4214 (2012). 54. Kojima, K., et al. p53 activation of mesenchymal stromal cells partially abrogates microenvironment-mediated resistance to FLT3 inhibition in AML through HIF- 1alpha-mediated down-regulation of CXCL12. Blood 118, 4431-4439 (2011). 55. Traer, E., et al. FGF2 from Marrow Microenvironment Promotes Resistance to FLT3 Inhibitors in Acute Myeloid Leukemia. Cancer research 76, 6471-6482 (2016).

108

56. Chang, Y.-T., Hernandez, D., Ghiaur, G., Levis, M.J. & Jones, R.J. Bone Marrow Stroma Protects FLT3 Acute Myeloid Leukemia (AML) through CYP3A4- Mediated Drug Metabolization of FLT3 Tyrosine Kinase Inhibitors (TKIs). Blood 130, 2519 (2017). 57. Yang, X., Sexauer, A. & Levis, M. Bone marrow stroma-mediated resistance to FLT3 inhibitors in FLT3-ITD AML is mediated by persistent activation of extracellular regulated kinase. British journal of haematology 164, 61-72 (2014). 58. Parmar, A., et al. Stromal Niche Cells Protect Early Leukemic FLT3- ITD+ Progenitor Cells against First-Generation FLT3 Tyrosine Kinase Inhibitors. Cancer research 71, 4696 (2011). 59. Piloto, O., et al. Prolonged exposure to FLT3 inhibitors leads to resistance via activation of parallel signaling pathways. Blood 109, 1643-1652 (2007). 60. Weisberg, E., Sattler, M., Ray, A. & Griffin, J.D. Drug resistance in mutant FLT3-positive AML. Oncogene 29, 5120-5134 (2010). 61. Smith, C.C. & Shah, N.P. The role of kinase inhibitors in the treatment of patients with acute myeloid leukemia. American Society of Clinical Oncology educational book. American Society of Clinical Oncology. Annual Meeting, 313-318 (2013). 62. Winter, G.E., et al. BET Bromodomain Proteins Function as Master Transcription Elongation Factors Independent of CDK9 Recruitment. Molecular cell 67, 5- 18.e19 (2017). 63. Dhalluin, C., et al. Structure and ligand of a histone acetyltransferase bromodomain. Nature 399, 491 (1999). 64. Taniguchi, Y. The Bromodomain and Extra-Terminal Domain (BET) Family: Functional Anatomy of BET Paralogous Proteins. International journal of molecular sciences 17, 1849 (2016). 65. LeRoy, G., Rickards, B. & Flint, S.J. The double bromodomain proteins Brd2 and Brd3 couple histone acetylation to transcription. Molecular cell 30, 51-60 (2008). 66. Jonkers, I. & Lis, J.T. Getting up to speed with transcription elongation by RNA polymerase II. Nature Reviews Molecular Cell Biology 16, 167 (2015). 67. Yang, Z., He, N. & Zhou, Q. Brd4 Recruits P-TEFb to Chromosomes at Late Mitosis To Promote G1 Gene Expression and Cell Cycle Progression. Molecular and cellular biology 28, 967 (2008). 68. Dey, A., Nishiyama, A., Karpova, T., McNally, J. & Ozato, K. Brd4 marks select genes on mitotic chromatin and directs postmitotic transcription. Molecular biology of the cell 20, 4899-4909 (2009). 69. Sakaguchi, T., et al. Bromodomain protein BRD4 inhibitor JQ1 regulates potential prognostic molecules in advanced renal cell carcinoma. Oncotarget 9, 23003-23017 (2018). 70. Zuber, J., et al. RNAi screen identifies Brd4 as a therapeutic target in acute myeloid leukaemia. Nature 478, 524 (2011). 71. Dawson, M.A., et al. Inhibition of BET recruitment to chromatin as an effective treatment for MLL-fusion leukaemia. Nature 478, 529 (2011).

109

72. Dawson, M.A., et al. Recurrent mutations, including NPM1c, activate a BRD4- dependent core transcriptional program in acute myeloid leukemia. Leukemia 28, 311-320 (2014). 73. Harris, A.W., et al. The E mu-myc transgenic mouse. A model for high-incidence spontaneous lymphoma and leukemia of early B cells. The Journal of experimental medicine 167, 353-371 (1988). 74. Leder, A., Pattengale, P.K., Kuo, A., Stewart, T.A. & Leder, P. Consequences of widespread deregulation of the c-myc gene in transgenic mice: multiple neoplasms and normal development. Cell 45, 485-495 (1986). 75. Stewart, T.A., Pattengale, P.K. & Leder, P. Spontaneous mammary adenocarcinomas in transgenic mice that carry and express MTV/myc fusion genes. Cell 38, 627-637 (1984). 76. Horiuchi, D., Anderton, B. & Goga, A. Taking on challenging targets: making MYC druggable. American Society of Clinical Oncology educational book. American Society of Clinical Oncology. Annual Meeting, e497-e502 (2014). 77. Shi, J., et al. Role of SWI/SNF in acute leukemia maintenance and enhancer- mediated Myc regulation. Genes & development 27, 2648-2662 (2013). 78. Rahl, P.B., et al. c-Myc regulates transcriptional pause release. Cell 141, 432-445 (2010). 79. Wilson, A., et al. c-Myc controls the balance between hematopoietic stem cell self-renewal and differentiation. Genes & development 18, 2747-2763 (2004). 80. Trumpp, A., et al. c-Myc regulates mammalian body size by controlling cell number but not cell size. Nature 414, 768 (2001). 81. Leon, J., Ferrandiz, N., Acosta, J.C. & Delgado, M.D. Inhibition of cell differentiation: a critical mechanism for MYC-mediated carcinogenesis? Cell cycle (Georgetown, Tex.) 8, 1148-1157 (2009). 82. Vita, M. & Henriksson, M. The Myc oncoprotein as a therapeutic target for human cancer. Seminars in cancer biology 16, 318-330 (2006). 83. Luo, H., et al. c-Myc rapidly induces acute myeloid leukemia in mice without evidence of lymphoma-associated antiapoptotic mutations. Blood 106, 2452-2461 (2005). 84. Zuber, J., et al. An integrated approach to dissecting oncogene addiction implicates a Myb-coordinated self-renewal program as essential for leukemia maintenance. Genes & development 25, 1628-1640 (2011). 85. Felsher, D.W. & Bishop, J.M. Reversible tumorigenesis by MYC in hematopoietic lineages. Molecular cell 4, 199-207 (1999). 86. Fiskus, W., et al. BET protein antagonist JQ1 is synergistically lethal with FLT3 tyrosine kinase inhibitor (TKI) and overcomes resistance to FLT3-TKI in AML cells expressing FLT-ITD. Molecular cancer therapeutics 13, 2315-2327 (2014). 87. Coudé, M.-M., et al. BET inhibitor OTX015 targets BRD2 and BRD4 and decreases c-MYC in acute leukemia cells. Oncotarget 6, 17698-17712 (2015). 88. King, K.E. & Weinberg, W.C. p63: Defining roles in morphogenesis, homeostasis, and neoplasia of the epidermis. 46, 716-724 (2007).

110

89. Chiang, C.-T., Chu, W.-K., Chow, S.-E. & Chen, J.-K. Overexpression of ΔNp63 in a human nasopharyngeal carcinoma cell line downregulates CKIs and enhances cell proliferation. 219, 117-122 (2009). 90. Rocco, J.W., Leong, C.-O., Kuperwasser, N., DeYoung, M.P. & Ellisen, L.W. p63 mediates survival in squamous cell carcinoma by suppression of p73- dependent apoptosis. Cancer Cell 9, 45-56 (2006). 91. Crook, T., Nicholls, J.M., Brooks, L., O'Nions, J. & Allday, M.J. High level expression of ΔN-p63: a mechanism for the inactivation of p53 in undifferentiated nasopharyngeal carcinoma (NPC)? Oncogene 19, 3439 (2000). 92. Mundt, H.M., et al. Dominant negative (ΔN) p63α induces drug resistance in hepatocellular carcinoma by interference with apoptosis signaling pathways. Biochemical and Biophysical Research Communications 396, 335-341 (2010). 93. Patturajan, M., et al. ΔNp63 induces β-catenin nuclear accumulation and signaling. Cancer Cell 1, 369-379 (2002). 94. Ogawa, E., et al. p51/p63 inhibits ultraviolet B-induced apoptosis via Akt activation. Oncogene 27, 848 (2007). 95. Kim, M.S., et al. IL-32θ gene expression in acute myeloid leukemia suppresses TNF-α production. Oncotarget 6, 40747-40761 (2015). 96. Ma, H.S., et al. All-trans retinoic acid synergizes with FLT3 inhibition to eliminate FLT3/ITD+ leukemia stem cells in vitro and in vivo. Blood 127, 2867- 2878 (2016). 97. Hurtz, C., et al. BCL6-mediated repression of p53 is critical for leukemia stem cell survival in chronic myeloid leukemia. The Journal of experimental medicine 208, 2163-2174 (2011). 98. Hurtz, C., et al. BCL6 Is Required for the Maintenance of Leukemia-Initiating Cells In Chronic Myeloid Leukemia. Blood 116, 202 (2010). 99. Devaiah, B.N., Gegonne, A. & Singer, D.S. Bromodomain 4: a cellular Swiss army knife. Journal of leukocyte biology 100, 679-686 (2016). 100. Moll, U.M. & Slade, N. p63 and p73: roles in development and tumor formation. Molecular cancer research : MCR 2, 371-386 (2004). 101. Inoue, K. & Fry, E.A. Alterations of p63 and p73 in human cancers. Sub-cellular biochemistry 85, 17-40 (2014). 102. Yang, A., et al. p63, a p53 homolog at 3q27-29, encodes multiple products with transactivating, death-inducing, and dominant-negative activities. Molecular cell 2, 305-316 (1998).

111

Curriculum Vitae

112

Curriculum Vitae Lauren Lee

Demographic Information Home Address Office Address 2602 Foster Ave. The Bunting-Blaustein Cancer Research Baltimore, MD 21224 Building 1650 Orleans St. Bldg 1, Rm 232 Baltimore, MD 21287

Date of Birth: June 26, 1991, Oakland, CA, USA

Education and Training

Doctoral/Graduate 2019 Ph.D. Cellular and Molecular Medicine

Undergraduate 2013 B.A. Molecular and Cell Biology, Immunology 2013 B.S. Environmental Economics and Policy

Professional Experience

March 2016 – April 2019 Ph.D. Candidate – Johns Hopkins University School of Medicine Advisor: Mark Levis, M.D., Ph.D. Project: Investigation of synergistic activity with dual FLT3 and BET inhibition in FLT3-ITD acute myeloid leukemia

January 2016 – March 2016 Graduate Student Rotation – Johns Hopkins University School of Medicine Advisor: Elizabeth Jaffee, M.D. Project: Characterization of tumor-associated macrophages

October 2015 – December 2015

113

Graduate Student Rotation – Johns Hopkins University School of Medicine Advisor: Katherine Whartenby, Ph.D. Project: Development of mouse multiple sclerosis models

June 2013 – August 2015 Research Associate, Translational Biology – Amgen Inc. (Onyx Pharmaceuticals) Project: Investigation of T cell plasticity and suppression of cytokine production in autoimmune diseases using ONX0914 and other in- house compounds

May 2011 – October 2012 Undergraduate Research – University of California, San Francisco Advisor: Jean-Marc Schwarz, Ph.D. Project: Studying the effects of high fructose diets on de novo lipogenesis

June 2012 – August 2012 June 2011 – August 2012 Drug Development Intern – Portola Pharmaceuticals Mentor: Lynn Kamen, Ph.D. Project: Development of SYK and JAK inhibitors

Publications

Lee, Lauren Y et al. “Preclinical studies of gilteritinib, a next-generation FLT3 inhibitor.” Blood 129.2 (2017): 257-260.

Leadership and Volunteer Experience

March 2018 – January 2019 Head of Strategy – Multisensor Diagnostics Role: Led the go-to-market strategy for the digital health product, MouthLab

June 2018 Bridge to BCG Participant, New York – Boston Consulting Group Role: Selected participant in a week-long workshop to work on consulting cases from Fortune 500 clients

August 2016 – Present Pro bono Consultant – Hopkins Graduate Consulting Club

114

2011 Co-founder of Phi Chi, a professional pre-health society – University of California, Berkeley

2011 – 2013 Director, Molecular and Cell Biology Undergraduate Student Association – University of California, Berkeley

2011 – 2013 Special Needs Swim Instructor – YMCA (Berkeley, CA)

Awards

Cal Alumni Association Leadership Scholar 2012

115