Identification and Characterization of Specific Inhibitors Of

IDENTIFICATION AND CHARACTERIZATION OF SPECIFIC INHIBITORS OF

THE EYA2 PHOSPHATASE

AARON B. KRUEGER

B.S., North Dakota State University, 2006

A thesis submitted to the

Faculty of the Graduate School of the

University of Colorado in partial fulfillment

of the requirements for the degree of

Doctor of Philosophy

Structural Biology and Biochemistry

2013

This thesis for the Doctor of Philosophy degree by

Aaron B. Krueger

has been approved for the

Structural Biology and Biochemistry

Robert Hodges, Chair

Heide Ford

Changwei Liu

Steve Nordeen

Rui Zhao, Advisor

Date ___10/21/13______

Krueger, Aaron B. (Ph.D., Structural Biology and Biochemistry)

Identification and characterization of specific inhibitors of the Eya2 phosphatase

Thesis directed by Associate Professor Rui Zhao.

ABSTRACT

Eya proteins are transcriptional co-activators of the homeobox gene Six1 and contain a unique protein tyrosine phosphatase activity. Eya and Six1 are required for normal development and are down-regulated in most adult tissues.

However, Six1 and Eya are re-expressed in a large number of breast tumors and play a causal role in the initiation and development of these tumors. Eya’s unique phosphatase activity has been shown to be important for the transformation, migration, invasion, and metastasis of breast cancer cells. Because of its mechanistically unique phosphatase activity and its important role in the establishment and spread of breast cancer, I targeted Eya for anti-breast cancer therapy. To this end, I developed an HTS strategy to identify inhibitors of Eya2’s phosphatase activity. In collaboration with the NIH Chemical Genomics Center, I screened over 330,000 compounds and have identified a series of compounds effective at inhibiting Eya2’s phosphatase activity. Cell-based migration assays using MCF10A breast cells have demonstrated an import role for Eya2’s phosphatase activity and that these identified phosphatase inhibitors can effectively inhibit migration of these cells. Using biochemical, biophysical, and structural approaches, I further characterized these inhibitors as reversible,

iii

mixed mode inhibitors that likely bind at an allosteric site, with a KD of 2.0 µM. I

have demonstrated that these inhibitors are specific towards Eya2 over other

mechanistically similar phosphatases including its highly-conserved family member, Eya3. These characterizations are leading us to develop a model of how the inhibitor could specifically interact and inhibit Eya2 which will help us to further improve potency to make a viable anti-breast cancer drug.

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

Approved: Rui Zhao

DEDICATION

I dedicate this work to ‘the’ wife Katie who happily provided strength and support throughout the entire process, to my parents who kept me curious and never said no, and to my friends who kept me sane when I otherwise would not be.

ACKNOWLEDGMENTS

I would first like to thank the members of my committee for their many insightful questions and advice through the progression of the project. I would like to thank the prior and past members of the Zhao lab for the fun times, the assistance provided, and support given. I would like to especially thank Dr. Zhao for allowing me the opportunity to join her lab and for the years of teaching me to become a successful scientist. I would like to thank our collaborators whose contributions were instrumental in allowing the project to progress. Specifically, I would like to thank members of the NIH Chemical Genomics Center for their contributions to the high throughput screening, of the University of Colorado

BioPhysics CORE for their assistance and guidance, Philip Reigan and lab for assistance in modeling, Elan Eisenmesser and lab for assistance with NMR experiments, and Heide Ford and lab for essential contributions. Without these people, I would be nowhere. Finally, for the years of guidance and support, I would like to thank the fellow students, faculty, and staff of the Department of

Biochemistry and Molecular Genetics, and specifically those of the Structural

Biology and Biochemistry Program.

CONTENTS

CHAPTER

I. INTRODUCTION ...... 1

Transcription complexes as therapeutic targets ...... 1

The Six1/Eya transcriptional complex ...... 2

Six1 and Eya in cancer ...... 6

The transcription activator Eya is a HAD family phosphatase ...... 7

Functions of Eya’s protein tyrosine phosphatase activity ...... 15

Targeting protein tyrosine phosphatases for therapeutic drug design 17

Targeting the Eya protein tyrosine phosphatase activity ...... 21

II. THE IDENTIFICATION OF EYA2 PHOSPHATASE INHIBITORS ...... 23

Abstract ...... 23

Introduction ...... 24

Results ...... 27

Design of assays suitable for HTS ...... 27

Known phosphatase inhibitors do not significantly inhibit Eya2’s phosphatase activity ...... 31

The OMFP-based phosphatase assay is suitable for HTS ...... 32

A large scale primary screen identified a class of structurally related initial hits ...... 35

The best compounds from this series are active in a pH2AX-based secondary phosphatase assay ...... 38

The N-arylidenebenzohydrazide compounds do not inhibit other cellular phosphatases ...... 40

Discussion ...... 41

Methods ...... 45

vii

Protein expression and purification ...... 45

pNPP-based Eya phosphatase assay ...... 45

OMFP-based Eya phosphatase assay ...... 46

Miniaturized Eya phosphatase assay for HTS ...... 46

Compound library and instruments for liquid handling ...... 47

HTS data analysis ...... 48

pH2AX-based Eya phosphatase assay ...... 48

Phosphatase assays of PTP1B, PPM1A, and Scp1 ...... 49

III. DETERMINING THE MECHANISM OF ACTION OF EYA2 INHIBITORS..... 51

Abstract ...... 51

Introduction ...... 52

Results ...... 55

HTS identified a new chemical series that specifically inhibit Eya2’s phosphatase activity...... 55

Enzyme kinetic analyses suggest that the N- arylidenebenzohydrazide compounds are mixed mode inhibitors. 58

Attempt to determine the binding mechanism using X-ray crystallography...... 60

Attempt to determine the binding mechanism using NMR...... 61

Compounds likely do not bind in the active site and do not require Mg2+ for function...... 65

Compounds likely bind to an allosteric site as determined by mutational analysis...... 72

Evaluate the effect of inhibitors on cellular phenotypes...... 81

Discussion ...... 83

Methods ...... 90

Protein expression and purification ...... 90

viii

Eya2 phosphatase assays ...... 91

Eya2 kinetic experiments ...... 91

Isothermal titration calorimetry ...... 92

Circular dichroism ...... 93

NMR spectroscopy ...... 93

UV-Vis spectra analysis of selected hydrazides ...... 94

Molecular docking ...... 94

Motility assay...... 95

Cell lines ...... 95

IV. DEVELOPMENT OF ASSAYS TARGETING OTHER ASPECTS OF SIX1 ACTION ...... 96

Abstract ...... 96

Introduction ...... 97

Results ...... 98

Development and optimization of fluorescence anisotropy assay for Six1-DNA HTS...... 98

Development of competitive ELISA for identifying inhibitors of the Six1-Eya2 interaction...... 101

Discussion ...... 103

Methods ...... 108

Fluorescence polarization assay ...... 108

Optimized enzyme-linked immunosorbent assay ...... 108

V. DISCUSSION AND FUTURE DIRECTIONS ...... 109

Targeting the Six1 transcriptional complex ...... 109

Six1-Eya2 protein-protein interaction...... 110

Six1-DNA interaction ...... 112

Inhibitors of Eyas phosphatase activity ...... 114

Conclusion ...... 120

REFERENCES ...... 122

LIST OF TABLES

Table

1. Results of large scale HTS ...... 37

LIST OF FIGURES

Figure

1. Structure of the Six1-Eya2 complex...... 3

2. Eya is a HAD phosphatase...... 9

3. Eya contains signature HAD phosphatase motifs...... 11

4. Eya2 uses a unique two-step reaction mechanism...... 12

5. Histone H2AX is an Eya target...... 16

6. Purification of the Eya2 ED...... 28

7. Characterization of pNPP-based phosphatase assay...... 29

8. Characterization of the OMFP phosphatase assay ...... 30

9. Common phosphatase inhibitors do not significantly inhibit Eya2...... 31

10. HTS of the NIH Structural Diversity Set II...... 33

11. HTS assay optimization for the Eya2 phosphatase...... 34

12. A new series of Eya2 ED phosphatase inhibitors were identified in the HTS that inhibit Eya2’s phosphatase activity...... 36

13. Inhibitors function in secondary assay...... 39

14. Eya2 inhibtors are selective...... 40

15. A class of N-arylidenebenzohydrazide – containing compounds selectively inhibits Eya2...... 56

16. Identified inhibitors within the series binds Eya2 ED but does not induce significant structural changes...... 57

17. Kinetic experiments...... 59

18. Eya2 HSQC with inhibitor demonstrates binding...... 62

19. Solubility protocol for increasing solubility of NMR samples ...... 64

20. The interaction between Eya2 ED and the N-arylidenebenzohydrazide- containing compound does not require Mg2+...... 67

xii

21. Eya2 inhibitors likely do not bind in the active site...... 68

22. Binding of MLS000544460 does not require Mg2+...... 69

23. Representative docking models of inhibitors in the active site without Mg2+. 70

24. Dose response curves using BOR mutants...... 71

25. Sequence and surface differences between Eya2 and Eya3...... 73

26. Eya2 to Eya3 mutations cause a loss of inhibitor activity...... 74

27. ITC experiments show that MLS000544460 does not bind to mutants at allosteric site...... 75

28. Inhibitors likely bind to an allosteric site ...... 76

29. Role of Eya2 in sensitizing various cell lines to DNA-damaging agents and effect of Eya2 inhibitor on cellular migration...... 79

30. Possible mechanisms of allosteric inhibition by MLS000544460...... 85

31. Six1 binds DNA in FP assays but is not specific...... 99

32. Six1 binds Eya via ELISA...... 102

33. Drug-like properties of Eya2 inhibitors...... 117

xiii

LIST OF ABBREVIATIONS

*DNA fluorescein-labeled DNA ABTS 2,2’-azino-di(3-ethylbenzthiazoline-6-sulfonate) ATM ataxia-telangiectasia mutated ATR ataxia-telangiectasia and Rad-3-related BiFC bimolecular fluorescence complementation BOR brachio-oto-renal BSA bovine serum albumin DMSO dimethylsulfoxide ED Eya Domain EDTA ethylenediaminetetraacetic acid EGTA ethylene glycol tetraacetic acid ELISA enzyme-linked immunosorbent assay EMSA electrophoretic mobility shift assay EMT epithelial-mesenchymal transition EYA eyes absent FDP fluorescein diphosphate GFP green fluorescent protein GST glutathione S-transferase H2AX histone variant H2AX HAD haloacid dehalogenase HBM helix bundle motif HD homeodomain HDM2 Mouse double minute 2 homolog (also known as MDM2) HRP horseradish peroxidase HSQC heteronuclear single quantum coherence HTS High throughput screening IC50 half maximal inhibitory concentration ITC isothermal titration calorimetry JNK1 c-Jun N-terminal kinase 1 KO knockout MDCI mediator of DNA damage checkpoint protein 1 MRE11 meiotic recombination 11 NIH National Institutes of Health NMR nuclear magnetic resonance spectroscopy OMFP 3-O-methylfluorescein phosphate PEI polyethyleneimine pH2AX phosphorylated histone variant H2AX pNPP para-nitrophenylphosphate PPM1A Protein phosphatase 1A (also known as PP2CA) PTK protein tyrosine kinase PTP protein tyrosine phosphatase PTP1B protein tyrosine phosphatase 1B RDGN retinal gene determination network S/B signal-to-background xiv

SAR structure activity relationship Scp1 small C-terminal domain phosphatase 1 SD Six Domain SO Sine oculis TCGA The Cancer Genome Atlas TCPTP T-cell protein tyrosine phosphatase TIC tumor initiating cell UV ultraviolet WSTF Williams syndrome transcription factor

CHAPTER I

INTRODUCTION

Transcription complexes as therapeutic targets

It is becoming increasingly evident that cancer and normal development share many fundamental properties, such as changes in cell proliferation and differentiation, in cell death, in vascularization, in cell mobility, and in invasion of the surrounding tissues[1]. These processes depend on the activation of transcription factors at correct times during development. In cancer, improper regulation of these developmental transcription factors can lead to the induction of signaling pathways responsible for oncogenic properties[2], identifying them as critical regulators in tumorigenesis and metastasis. Despite this importance, few drugs have been developed that target transcription factors. The more common

(easier) strategy is to develop therapeutics against enzymes or receptors upstream of a transcription factor. Limited to a single pathway, they miss the central node that can respond to multiple oncogenic signals. As such, it is likely that developing therapeutics against transcription factors will lead to a more effective way to treat cancer. My thesis project is an effort to target one such transcription factor complex, the Six1/Eya complex.

The Six1/Eya transcriptional complex

The evolutionarily conserved Six1/Eya transcriptional complex was first identified in drosophila containing the Six1 homolog sine oculis (SO) and the Eya

homolog eyes absent (EYA). KO studies identified both the SO and EYA genes

as critical components of the conserved Retinal Determination Gene Network

(RDGN), required for proper drosophila eye formation. [3-7]. From the original

characterization of the RDGN network in drosophila, RDGN homologs have been

identified in numerous species including sponges, plants, worms, frogs, fish,

birds, and higher vertebrates where they have emerged as important regulators

of organogenesis[8-20]. Additionally, the mammalian RDGN is used reiteratively

in several developmental contexts. It is induced at early stages in development,

with roles in developing eye, nose, ear, and all sensory ganglia[14, 15, 19, 21-

23]. As part of the RDGN, the Six family of transcription factors function in

development through regulation of cell cycle related genes, such as CyclinA1,

CyclinD1, and c-Myc, leading to a tissue-specific expansion of precursor

populations[24-31]. In muscle and kidney development, Six family members

function in developmental epithelial-mesenchymal transition (EMT) [32-34], a

process involving the loss of cell polarity and adhesion, while gaining migratory

and invasive properties. Indeed, some members are required for maintenance of

a mesenchymal phenotype in renal precursor cells[33].

To mediate function, the Six family contains DNA-binding activity mediated

through its conserved homeodomain (HD), a structurally conserved domain that

binds DNA in a sequence-specific fashion[35, 36]. HD-containing genes function

as master regulators in development by controlling cellular processes such as

proliferation, differentiation, apoptosis, adhesion, and migration[37, 38]. A large

Figure 1. Structure of the Six1-Eya2 complex (A) Crystal structure of the Six1-Eya2 interaction. HD, SD, and ED are color- coordinated to panel 1B. MBP (red) was used as a fixed arm carrier protein for crystallization purposes. (B) Domain organization for Six1 and Eya2. Eya2 ED binds the Six1 SD. (C) Six1 binds Eya2 ED using a single helix at a site adjacent to the active site. Active site is represented by green sphere.

number of HDs recognize the same core DNA sequence TAAT[24] or a close variation, introducing the possibility of DNA-binding redundancy among HD- containing proteins. A prototypical HD consists of an unstructured N-terminal arm followed by three α-helices and is roughly 60 amino acids in length. The amino acid composition of the unstructured N-terminal arm and helix 3 contribute to the bulk of the DNA-binding specificity[35, 39]. The structure of the Six1 HD[40] solved by our lab reveals a signature HD fold consisting of 3 α-helices (Fig. 1A,

B)[41], yet supports a unique DNA-binding mechanism that requires regions outside the classical HD motif for binding DNA[42-44]. When overlaid with a known HD-DNA complex[45], the α6-helix of the Six1 Six Domain (SD) is positioned such that it may interact (with minor adjustments) with the DNA major groove and that mutations within the SD α6 helix abrogate DNA binding[46]. The

Six family N-terminal arms, crucial for prototypical HD DNA-binding and specificity, are not required for Six6 to bind DNA[47] and all family members contain acidic rather than basic amino acids found in canonical HDs. In addition, regions outside the HD are required for DNA binding by Six1[42], Six4[43], and

Six5[44], and the HDs of Six2 and Six6 alone are unable to bind its consensus sequence[47]. As such, the DNA-binding mechanism of the Six family of HDs differs from canonical HDs through significant changes in the protein sequences, electrostatics, and structural components involved in DNA-recognition[40, 47].

Indeed, Six1 does not recognize the classical HD core sequence but instead binds to unique sequences including the TCAGGTT core found in the MEF3

sequence element enriched in myoblasts within the promoter regions of

Myogenic Regulatory Factor target genes, such as myogenin[48].

Six1 requires the Eya co-activator to link the DNA-binding function of Six1 with the activation function of Eya, as Six1 does not contain intrinsic activation or repression domains. The Six1 and Eya interaction is mediated through the novel, highly conserved Six1SD and Eya2 Eya Domain (ED), a unique 271 amino acid domain found in all Eya members[15, 40] (Fig. 1A, B). The Six1-Eya2 crystal structure shows the SD-ED interface consists of a single α-helix from Six1 (α1) binding to a groove in the ED and the interaction is mediated by hydrophobic interactions, salt bridges, and hydrogen bonds (Fig. 1C) [40]. This interaction mode is highly similar to the protein-protein interaction observed in the p53-

MDM2[49] and Bak-BCL-XL[50] complexes. P53 and Bak each uses an amphipathic α-helix to bind to a hydrophobic pocket in MDM2 and Bcl-XL. Both

interactions have been successfully targeted using small molecules[51, 52].

Six1 and Eya1 knockdown studies in mice have shown that both are

necessary for proper development of muscle, the ears, kidney, sensory neurons,

and thymus. Indeed, knockdown of each gene phenocopies the other and results

in poor progenitor cell proliferation and survival[53-56] leading to programmed

cell death to cause a reduction in size or absence in organ development[57, 58].

Indeed, impairments of the Six transcriptional complex in normal development

often lead to diseases. Mutations in Six and Eya members that disrupt the Six-

Eya transcription complex (often by impairing Six1’s DNA binding ability or the

Six1/Eya interaction) cause branchio-oto-renal (BOR) syndrome, an autosomal

dominant disorder characterized by branchial, otic, and renal anomalies[40, 46,

59-61]. Eya1 mutations have been identified in patients with congenital cataracts

and ocular anterior segment anomalies[62] and have been linked to cardiofacial

syndrome[63]. Eya4 mutations cause late-onset autosomal dominant hearing

loss in the DFNA10 locus[64-66], and cardiomyopathy[64]. These data implicate

both individual members and interactions among members within the complex as

crucial for proper development and that disruption of the transcriptional complex

leads to disease.

Six1 and Eya in cancer

Six1 expression is lost in most adult tissues after organ development is

complete, however, it is re-expressed in a number of cancers including breast, brain, ovarian, cervical, lung, hepatocellular, colorectal, rhabdomyosarcoma,

Wilms’ tumor, and leukemias[25, 34, 67-76]. In breast cancer, Six1 is upregulated in 50% of primary breast tumors and 90% of metastatic lesions[25,

26, 77]. Overexpression of Six1 plays a causal role in breast tumorigenesis and

metastasis [26, 34, 75]. Over-expression of Six1 transforms immortalized, but

otherwise normal, mammary epithelial cells, forming highly aggressive tumors

when injected orthotopically into nude mice[78]. Furthermore, mammary specific,

inducible Six1 overexpression in a bitransgenic mouse model leads to

hyperplasia and highly aggressive and invasive tumor formation in vivo and

displays hallmarks of biologically aggressive carcinomas[75], including both

lymphatic and rare bone metastases[34] while displaying oncogenic EMT and 6

stem cell phenotypes[75]. Six1 may act as a global regulator of tumor

progression since it can induce cancer stem cell phenotypes, EMT, and

lymphangiogenesis through upregulation of the TGF-β[34, 79, 80] or VEGF-C[81] signaling pathways.

Since Six1 does not contain activation or repression domains, it relies on cofactors for function, including the Eya members. Eya proteins are known to function in Six1-mediated processes in both normal development[54, 82] and in

disease processes[59, 60, 83]. Wilms’ tumor, acute leukemia, and malignant

nerve-sheath tumors often have overexpression of both Six1 and Eya2[70, 76,

84], while each is independently implicated in ovarian cancer[67, 73]. Six1

expression in breast cancer patient samples[85, 86] significantly predicts

shortened time to relapse, shortened survival, and shortened time to metastasis

in tumors where relapse occurs within 5 years[34], but only with co-expression of

its co-activator Eya1 or Eya2 (the number of tumors overexpressing Eya3 and

Eya4 are too low to reach statistical significance)[87]. Six1 is unable to induce

EMT, TGF-β signaling, and stem cell phenotypes when Eya2 is knocked

down[87]. In fact, direct binding of Six1 to Eya2 is required for enhanced TGF-β

signaling, induction of pro-EMT characteristics, and Six1-mediated metastasis in

a mouse model[40].

The transcription activator Eya is a HAD family phosphatase

In 2003, three groups independently reported that the ED contains novel

protein tyrosine phosphatase (PTP) activity belonging to the phosphatase 7

subgroup of the diverse superfamily of haloacid dehalogenase (HAD) enzymes

(Fig. 2A)[54, 88, 89]. This exciting discovery defined Eya proteins as the first examples of transcription factors with inherent phosphatase activity and the first

HAD member with PTP activity. It prompted a renewed interest in the largelyuncharacterized HAD family[3, 90-92].

HAD proteins are an extremely large (at least 183 members in human[93]) and old superfamily with five HAD genes encoded in the last universal common ancestor of all evolutionary domains[94] and subsequently evolved into at least

23 protein families across all three superkingdoms of life to catalyze a diverse range of reactions, including phospho-transferase, epoxide hydrolase, and DNA kinase activity, among others[95-99]. Having undergone a large expansion during animal evolution, gene duplication events have introduced new domains and motifs to drive the diversity and specificity of HAD protein function, leading to multi-domain proteins that have extremely specified localization, functions, and activities[95]. As a result, the HAD phosphatases have evolved independently from canonical phosphatases[94] and use a unique phosphoaspartyl transferase mechanism. HAD phosphatases contain a conserved structural motif around the catalytic core unlike those found in traditional PTPs. This modified Rossmann fold consists of a three-stacked, α/β sandwich comprised of repeating β-α units

(Fig. 2B). A parallel β-sheet, typically containing five parallel strands ordered

‘54123’, orients four loops on which the catalytic groups are located[93, 95]. HAD phosphatases typically share less than 15% conservation, yet can be identified

Figure 2. Eya is a HAD phosphatase

Figure 2. Eya is a HAD phosphatase (A) Ribbon diagram of Eya2 ED. The unique HBM cap is shown in cyan and the conserved HAD Core is shown in orange. Mg2+ is shown as a green sphere and represents the location of the active site. (B) Structure of the Eya2 HAD catalytic core reveals Rossmann-like three-stacked, α/β sandwich comprised of repeating β-α units. The amino acids required for catalysis are located on loops extending off the β-strands and the HBM cap is situated over the active site.

through sequence analysis to identify four signature motifs[93, 94, 100, 101].

Motif I contains the catalytic aspartic acid in the hhDxDx(T/V)(L/V)h sequence, with h being a hydrophobic residue and x being any amino acid. Motif II has the conserved hhhhhh(S/T) that orientates the phosphate of the substrate for properly aligned nucleophilic attack. A poorly conserved Motif III centered on a

Lys or His residue 15-30 residues from Motif IV that serves to stabilize the negative charge of the reaction intermediate. Motif IV contains the conserved sequence (G/S)(D/S)x3-4(D/E)hhhh. An aspartic acid in Motif IV stabilizes the

Mg2+ along with the catalytic aspartic acids in Motif I. Eya phosphatases were initially identified from their VFVWDLDET sequence matching Motif I, a VLVT sequence matching Motif II, and HX15YVVIGDGNEE[88] comprising Motifs III and

IV. Figure 3 shows the signature HAD motifs and the conservation for the Eya family members.

Figure 3. Eya contains signature HAD phosphatase motifs Motif I is shown in red, Motif II is shown in green, and motifs III and IV are shown in blue. Figure adapted from Reference [88].

Figure 4. Eya2 uses a unique two-step reaction mechanism A water molecule between Asp267 and Glu277 facilitates the formation and breakdown of the phosphoenzyme complex by assisting the general acid/base, Asp276. Mg2+ is used for proper orientation of the substrate to initiate catalysis. Figure used from reference [102].

The unique, two-step reaction (Fig. 4) of HAD phosphatase uses an aspartic acid (the first D in Motif I) as the nucleophile to initiate a nucleophilic attack on the phosphoryl group of the substrate to form a phosphoaspartyl enzyme intermediate. A water molecule then initiates a nucleophilic attack on the phosphoaspartyl enzyme intermediate to regenerate the catalytic aspartic acid,

releasing free phosphate in the process. Located two residues C-term of the nucleophilic aspartic acid, the second aspartic acid serves as a general acid/base to protonate the leaving group in the first step and deprotonates the water nucleophile in the second step. A divalent Mg2+ cofactor is required for proper positioning of the nucleophile and substrate, while providing charge neutralization for the transition state[93]. In order to favor an efficient aspartic acid-based nucleophilic attack, solvent must be excluded during the first reaction step. The second step requires high solvent contact to hydrolyze the phosphoaspartyl intermediate. This mechanism requires the alternation of the active site cavity between a closed state (excludes solvent to favor the aspartic acid nucleophilic attack in the first step) and an open state (favors the hydrolysis of the phosphoaspartyl intermediate in the second step)[103]. This switching is mediated by a ‘squiggle and flap’ structural motif immediately downstream of the

β1-strand that alternates between a tightly or loosely wound helical conformation that dictates whether the ‘flap’ will partially cover the catalytic cavity[94].

By far, the greatest diversity in HAD enzymes are their cap structures.

Depending on the enzyme, these caps not only provide additional shielding for catalytic function, but host a number of possible, enzyme-specific functions such as mediating protein-protein interactions. Three types of HAD caps are known.

C0 caps are small, consisting of loops or simple β-strands to form the structurally simplest motifs used predominantly with protein substrates in which the protein substrate provides the necessary solvent shielding upon binding[99, 104, 105].

The more common C1 modules consist of α-helical bundles of diverse

complexities and are large enough to seal the enzyme active site when in its closed state, making these caps predominantly used for small-molecule substrates. Likewise, C2 caps generally contain both α + β domains surrounding a core β-sheet motif and are highly diversified[95, 106]. The placement of the cap in relation to the active site is a distinguishing feature among HAD cap motifs. C0 and C1 caps are connected through the ‘squiggle and flap’ motif off β1 and C2 caps are connected after the core domain β3-strand. The crystal structure of the

Eya2 ED[102] revealed a C1 cap consisting of a helical bundle motif (HBM) containing seven elongated helices that is unique among all known HAD cap structures. Surprisingly, the ED cap does not participate in binding Six1 (as is often seen with HAD-interacting proteins[95]). Rather, Six1 binds to the core motif[40].

Of interest, Eya appears to be a dual-specificity phosphatase: the C- terminal ED contains PTP activity and the N-terminal region contains apparent

Ser/Thr phosphatase activity involved in the innate immune response to viruses[107], along with associating with members of the Immune Deficiency pathway to induce antimicrobial response genes[108]. The Eya Ser/Thr activity is poorly understood and no evidence has yet been found to correlate it with the transcription functions involving Six family members or to the PTP activity. Nor has a role in cancer or development been proposed.

Functions of Eya’s protein tyrosine phosphatase activity

The role of Eya’s PTP activity in Six1-mediated transcription remains unclear. In drosophila, EYA’s PTP activity was found to be required for a subset of SO-targeted genes[54], yet reduction of its phosphatase activity does not reduce global transcriptional output[109]. Although PTP targets involved in transcription have yet to be identified, there is extensive evidence that the PTP activity is required for both development and disease processes. In a developmental context, EYA phosphatase activity has a pro-angiogenic role in zebrafish development[110]. Although EYA is required for proper eye development in drosophila, it appears the PTP activity itself is not required for development or survival[109]. However, mutants of Eya1 lacking their intrinsic

PTP activity have been identified in Branchio-oto-renal (BOR) syndrome, highlighting an importance of PTP activity in normal human development[111].

Recent evidence shows Eya3 PTP activity promotes the migration, invasion, and transformation of breast tumor cells[112] and that Eya1 PTP activity is required for cell contact-independent growth, tumor colony formation, cellular proliferation, and mammosphere formation in breast cancer cells[113]. Most targets of Eya’s phosphatase activity remain elusive, while initial evidence suggests, yet not confirmed in vivo, that targets may include Eya itself[54], MAPK[114], and the C- terminal repeats of RNA polymerase[89].

Figure 5. Histone H2AX is an Eya target Upon dsDNA damage, H2AX is phosphorylated. Eya catalyzes the removal of Y142 to promote repair and survival. In the absence of Eya, pro-apoptotic complexes are recruited to lead to cell death. Figure from reference [115]

Eya’s PTP has an important role in DNA damage repair independent of

Six1/Eya mediated transcription through phosphorylating the histone variant

H2AX, the sole, in vivo confirmed target[115]. The phosphorylation status of

H2AX is a critical step in the double-strand DNA (dsDNA) repair pathway (Fig. 5).

H2AX contains two phosphorylation sites: Serine 139 and Tyrosine 142.

Phosphorylation states of these H2AX residues determine cell fate upon dsDNA damage. Ser139 is phosphorylated by the WSTF tyrosine kinase[116] and Eya3 is phosphorylated by the DNA-damage-response protein kinases ataxia- telangiectasia mutated (ATM) and ataxia-telangiectasia and Rad-3-related (ATR) upon dsDNA damage[115, 117, 118]. It has been shown that Eya3 binds the C- terminal tail of H2AX and catalyzes the removal of the phosphate at Tyr142, which is constitutively phosphorylated in the absence of dsDNA damage[115,

119]. If Tyr142 is dephosphorylated by Eya3 and Ser139 remains

phosphorylated, H2AX can be bound by DNA-repair factors, including factors that

promote cell survival and repair, such as mediator of DNA damage check point

protein 1 (MDCI), Rad50, and meiotic recombination 11 (MRE11). These factors

are unable to bind H2AX if Tyr142 remains phosphorylated (Eya3 is absent),

leading instead to the recruitment of pro-apoptotic factors including c-Jun N- terminal kinase 1 (JNK1) to promote cell death[115]. Although confirmed using

Eya3, it is likely that the other Eyas have similar roles (we know Eya2 ED can

dephosphorylate a H2AX peptide[120]), yet they remain to be tested in a cellular

environment. This role of Eya in DNA-damage repair may be unrelated to its role

in activating transcription.

Targeting protein tyrosine phosphatases for therapeutic drug design

The realization that Eya contains PTP activity introduces intriguing possibilities of targeting this phosphatase activity, which can potentially influence

Six1/Eya mediated transcription indirectly or modulate DNA damage response

[7]. Indeed, protein tyrosine phosphorylation is a major regulatory mechanism of cell signaling[121]. It is evident that aberrant protein tyrosine phosphorylation levels are associated with maintenance of the oncogenic state through promoting the abnormal survival, evasion of immune surveillance, and stress support of tumor cells along with being a primary cause of human cancer[122, 123]. The protein tyrosine phosphorylation state is maintained by the reciprocal functions of protein tyrosine kinases (PTKs) and PTPs. Early therapeutic focus has been in identifying small molecule inhibitors targeting PTKs, with successful molecules 17

targeting the Bcr-Abl kinase and epidermal growth factor receptor (EGFR) as examples. To date, at least 17 small molecule kinase inhibitors have been approved by the Food and Drug Administration for anti-cancer use[124-126]. A common belief is that because activation of PTKs are linked to various human cancers, PTPs, by catalyzing the reverse reactions of PTKs, function as negative regulators of cancer phenotypes[127-129] and that their inhibition would further enable oncogenesis rather than suppressing it. Luckily, this belief has been proven incorrect in that regulatory mechanisms involving protein tyrosine phosphorylation is more complex than the simply biochemical reaction. Indeed, dephosphorylation can serve to either propagate activation signals, or in fewer cases, terminate activation signals through loss of target enzyme activity or negative feedback mechanisms[130]. PTPs act in coordination with PTKs in promoting the growth and progression of cancer and changes in phosphorylation status is often a reflection of changes in the activities of one or both enzyme classes[125]. Traditionally, PTPs identified and targeted in cancer are those that act synergistically or antagonistically to known and validated kinases in oncogenic signaling pathways[131, 132] and that the PTP-dependent tumor suppressive or oncogenic activities within these pathways can be targeted for agonist or inhibitor development, respectively[127, 133-136]. Although traditionally viewed as less ideal targets for drug design compared to PTKs, the balance between phosphorylation states define PTPs as equally attractive therapeutic targets.

Traditional approaches used to identify PTP inhibitors are based on screening large chemical libraries against PTP activity using biochemical assays or structure using in silico docking[137]. Used most often, high throughput screening (HTS) methods are used to screen large chemical libraries against a validated target using enzymatic assays, fluorescence polarization[138], NMR- based approaches[139, 140], and other methods[141, 142]. After an assay is developed against a target, it is then miniaturized for robotic automation of liquid handling, incubation, and detection steps to screen the compound library[143].

Several PTP inhibitors have been identified using this chemical library screening approach[144-146] as well as additional approaches using in silico fragment- based screening[147] and combinations of methods[148, 149].

As one such example, HTS has successfully identified inhibitors of the

PTP Shp2. Shp2 functions as a positive regulator within the Ras-Erk1/2 mitogen- activated (MAP) kinase pathway involved in regulation of cell survival, proliferation, differentiation, adhesion, and migration[124]. The overexpression of

Shp2 is observed in infiltrating breast carcinoma[150] and the phosphatase activity is required for anchorage-independent growth of breast cancer cell lines[151]. Because of these roles in cancer, several inhibitors of Shp2 PTP activity have been identified by HTS. Among the successful inhibitors are NSC-

87877 (IC50 = 0.33 µM), which is able to inhibit Shp2 in epithelial carcinoma cells, endothelial cells, fibroblasts, muscle cells, and neuronal glioma cells[124, 152]. A virtual screen of over 2.7 million compounds identified two compounds with IC50s

of 0.63 µM and 2.1 µM that inhibit Shp2 in several cell-based assays and prevent

the transformation and proliferation of several cancer cells[137].

PTP1B is another highly targeted PTP. Similar to Eya, PTP1B plays a

causative role in several disease processes. PTP1B contributes to insulin

resistance in Type II diabetes and obesity through dephosphorylation of the

insulin receptor and inhibition of leptin signaling[144, 153]. PTP1B-dependent

dephosphorylation of Src is required for the transformation of MCF10A cells by

ErbB2[154], functions in tumor growth in SW48 colon cancer cells[155], and

promotes cell migration[156]. Overexpression of PTP1B in the mammary gland

results in the development of spontaneous mammary tumors and lung

metastasis[157]. Due to these roles, extensive efforts have been made to identify

PTP1B inhibitors. HTS using combinatorial libraries to simultaneously target the

catalytic pocket and adjacent substrate binding site has identified several potent

bidentate inhibitors[147, 158]. One such compound with high potency (IC50 = 120

nM) lowers blood glucose levels of mice in a dose dependent manner[159]. The

same compound also significantly delays the onset of ErbB2-induced mammary

tumors[157]. It is likely that HTS-identified PTP1B inhibitors will become the first

PTP inhibitors approved for clinical use[125].

Additional HTS programs against PTPs have identified both

activators[160] and inhibitors[161] of T-cell protein tyrosine phosphatase

(TCPTP) to modulate activity in a context-dependent manner as TCPTP has both

oncogenic and tumor suppressor roles. Additionally, high affinity inhibitors

targeting the PTPs CD45[162, 163], Leukocyte Common Antigen-Related[164], and PEP[165] have been identified and show effects in a cellular environment.

Targeting the Eya protein tyrosine phosphatase activity

Prior data suggest that targeting Eya PTP activity in cancer is a feasible anti-cancer strategy. The association of the PTP function of Eya with a cancer- associated cellular phenotype raises the distinct possibility that inhibition of Eya phosphatases could be an attractive new mode of targeted anti-cancer therapy.

Apart from inhibiting the functions in initiating tumors and promoting metastasis, selectively sensitizing tumor cells by engaging apoptotic programs of a cell is feasible and of great interest in the field of radiation oncology[166]. Eya PTP inhibitors may greatly increase the effectiveness of radiation therapy on cancers that are known to overexpress Eya. In addition, when combined with DNA- damage-inducing chemotherapies such as Adriamycin, mitomycin C, cicplatin, etoposide, and doxorubicin, inhibiting Eya phosphatase may increase overall effectiveness of current chemotherapies. Furthermore, targeting Eya for therapeutic development provides a distinct advantage: Eyas are mechanistically and structurally distinct from the large family of classical thiol-based PTPs, thus increasing the feasibility of selective inhibition.

Successes in developing potent inhibitors targeting several key phosphatases highlight the feasibility of targeting PTP activity in disease processes. Because of these successes, we believe that it is a feasible goal to obtain potent and selective inhibitors of Eya’s phosphatase activity that will result 21

in a therapeutic benefit. Inhibiting Eya PTP activity is advantageous for the following reasons: Eya PTP activity is important in the transformation and invasion of breast cancer cells (potentially independent of Six1-mediated functions), Eya PTP activity may be required for select Six1-mediated transcription necessary for its oncogenic roles, and Eya PTP activity is required in promoting DNA-damage repair upon DNA-damaging agents. In addition, targeting the Six1 transcription complex and the offers the chance to selectively target developmental programs initiated out of context with reduced adverse side effects. As presented in later chapters, this thesis work focuses predominantly on the development of HTS assays targeting Eya’s phosphatase activity, characterization of Eya inhibitors, and determining their binding mechanism.

CHAPTER II

THE IDENTIFICATION OF EYA2 PHOSPHATASE INHIBITORS

Abstract

Eya proteins are essential co-activators of the Six family of homeobox transcription factors and also contain a unique protein tyrosine phosphatase activity belonging to the haloacid dehalogenase family of phosphatases. The phosphatase activity of Eya is important for a subset of Six1-mediated transcription, making this a unique type of transcriptional control. Furthermore, the phosphatase activity of Eya is critical for transformation, migration, invasion, and metastasis of breast cancer cells. Upon DNA-damage, Eya phosphatase activity is also responsible for directing cells to a repair pathway to prevent apoptosis. Thus, inhibitors of the Eya phosphatase activity may function as anti- tumorigenic and anti-metastatic therapeutics, as well as sensitize cancer cells to

DNA damage inducing therapies. In this chapter, I adapted a high throughput screening assay that is used to identify a previously unknown chemical series that inhibits the Eya2 phosphatase activity with IC50s ranging from 1.8 to 79 µM.

Compound activity was confirmed using an alternative malachite green assay and H2AX, a known Eya substrate. Importantly, these Eya2 phosphatase inhibitors demonstrate specificity over several other cellular phosphatases. This research identifies the first selective Eya2 phosphatase inhibitors that can potentially be developed into chemical probes for functional studies of Eya phosphatase or into anti-cancer drugs in the future.

Introduction

The Eya proteins are mammalian homologues of the Drosophila Eyes

Absent genes and were first identified as essential co-activators of the Six family of transcription factors, including Six1. The Six1 homeoprotein is essential for the development of many organs, including the muscle, kidney, olfactory epithelium, and inner ear[167]. It is typically down-regulated after organ development is complete, and its expression level is low or absent in most adult tissues.

However, Six1 is over-expressed in numerous cancers, such as breast, ovarian, cervical, and hepatocellular carcinomas, as well as rhabdomyosarcomas, Wilms’ tumors, and leukemias[67, 167]. Six1 expression has been linked to transformation, tumor growth, and metastasis in multiple tumor types, including breast cancer[34, 75, 167, 168]. Experimentally lowering Six1 levels significantly decreases cancer cell proliferation[167] and metastasis[167, 168] in different cancer models.

Given that Six1 does not have an intrinsic activation or repression domain, it requires co-activators such as the Eya family of proteins to mediate its transcriptional activity, both in normal development[83, 167] and in various disease processes[60, 83, 167]. Eya proteins have been linked to many types of cancer in which Six1 is over-expressed[87, 112, 167]. Examination of the Wang and Van de Vijver public breast cancer microarray datasets[85, 86] demonstrated that over-expression of Six1 and Eya together significantly predict shortened time to relapse and metastasis and shortened survival, whereas each gene individually does not[87]. Furthermore, Eya2 knockdown in Six1-over-expressing

MCF7 cells inhibits the ability of Six1 to induce TGF-β signaling, epithelial- mesenchymal transition, and tumor initiating cell characteristics, properties that are associated with Six1-induced tumorigenesis and metastasis[87]. These data provide strong support that Six1 and Eya2 cooperate to induce tumorigenic and metastatic properties.

The C-terminal Eya Domain (ED)[88] contains signature motifs of the haloacid dehalogenase (HAD) hydrolases, a diverse collection of enzymes including phosphatases[88, 89, 167]. Eya proteins and other HAD family of phosphatases use an Asp as their catalytic active site residue instead of the more commonly used cysteine in cellular phosphatases[130]. Most HAD phosphatases target small molecule substrates, yet a few examples (such as

Scp1, Chronophin, Eya) target proteins[88]. Within this group, only Eya targets phosphorylated Tyr; all others are Ser/Thr phosphatases[119].

Much effort has been taken to understand the role of Eya’s phosphatase activity. Evidence from mouse models suggest that Eya proteins can utilize their intrinsic phosphatase activity to switch the Six1 transcriptional complex from a repressor to an activator complex for some Six1-induced genes[167], although the mechanism of this switch remains unclear. In Drosophila, while the phosphatase activity of Eya is not globally required for the ability of Six1 to induce transcription, it is required to induce transcription of a subset of Six1- targeted genes[169]. The Eya proteins therefore represent the first transcription factor with intrinsic phosphatase activity that can modulate transcriptional complexes[88, 89, 167]. Recently, Hegde and colleagues demonstrated that Eya

proteins, and in particular their Tyr phosphatase activity, are critical for the transformation, migration, invasion, and metastasis of breast cancer cells[112], although the specific mechanism by which Eya’s phosphatase activity induces the oncogenic phenotypes remains unknown.

In addition to its role in Six-mediated oncogenic functions, Eya proteins have recently been shown to function in DNA-damage repair. Eya proteins dephosphorylate phospho-Tyr142 on histone variant H2AX. This dephosphorylation is critical for the recruitment of DNA-repair complexes and pro-apoptotic factors are recruited if H2AX remains phosphorylated at

Tyr142[115, 119]. Knockdown of Eya proteins lead to a significant increase in apoptotic cells in response to hypoxia or ionizing radiation[115, 119]. Currently, about half of all people with cancer are treated with radiation therapy, either alone or with other cancer treatment, to kill cancer cells and reduce tumor burden. Selectively sensitizing tumor tissue by engaging the apoptotic program of a cell is of great interest to the field of radiation oncology[166]. It is foreseeable that inhibitors of Eya’s phosphatase activity may greatly increase the efficiency of radiation therapy, or of any DNA damaging related therapy (many cancer therapies use a combination of both), in cancers that are known to express Eya, including breast cancers, Wilms’ tumor, ovarian carcinomas[87, 167].

Although it has traditionally been difficult to identify specific phosphatase inhibitors, the fact that Eya proteins belong to the HAD family of protein phosphatases that use an aspartic acid instead of the more commonly used cysteine as their active site residue, provides a unique opportunity to potentially

identify specific Eya phosphatase inhibitors. In this chapter, I describe the identification of a previously unknown chemical series that specifically inhibits the

Eya2 phosphatase.

Results

Design of assays suitable for HTS

I chose the human Eya2 ED, the domain that contains the PTP activity, for identifying phosphatase inhibitors in this chapter. This is a practical choice as full- length human Eya2 is unstable and could not be expressed and purified from E. coli. In addition, Eya2 may contain an N-terminal Ser/Thr phosphatase domain with an unrelated and largely uncharacterized function[107]. Carrying out the

HTS using just the ED avoids complications from this additional phosphatase activity. I expressed the human Eya2 ED (Eya2 253-538) in E. coli as a GST- fusion protein. The protein was purified on a glutathione resin, cleaved off the

GST tag, and further purified using a Superdex-200 size exclusion column. Eya2

ED elutes as a monomer at 32 kDa (Fig. 6A). Using this protocol, I can obtain greater than 8 mg purified ED per liter of bacterial culture at a purity of >99% (Fig

6B).

To characterize the phosphatase activity of Eya2 ED, I first adapted a phosphatase assay using para-nitrophenylphosphate (pNPP) as a substrate. pNPP is a colorless substrate that produces a colored product, para-nitrophenyl,

Figure 6. Purification of the Eya2 ED (A) Superdex-200 elution profile of Eya2 ED. Eya2 ED elutes as a monomer. (B) SDS-PAGE of purified Eya2 ED.

upon dephosphorylation by Eya2, which can be measured by absorbance at 405

nm. I determined that the optimal Eya2 ED activity occurs at pH 6.5 with lesser

activity at other pHs between pH 5.5 and 7.7 (Fig. 7A). I optimized the Eya2 and

pNPP concentrations to produce an excellent (> 25) signal-to-background (S/B)

while maintaining a linear response over time. This is important since the HTS is

typically designed as an end-point analysis. Since Eyas are Mg2+-dependent

phosphatases[88, 89], I demonstrated that EDTA can be used as a positive

control for inhibition (Fig. 7B). I performed kinetic experiments for several Eya concentrations (Fig. 7C) that resulted in observing KM values between 9.7 ± 2.4

mM and 12.5 ± 1.2 mM. This reveals that pNPP is a rather poor substrate for

Eya2 and that the colorimetric-based assay is not especially a sensitive assay. I

therefore adopted a fluorescence-based phosphatase assay using human Eya2 28

Figure 7. Characterization of pNPP-based phosphatase assay (A) Eya2 ED has optimal activity at pH 6.5. (B) EDTA effectively inhibits the activity of Eya2 ED. Inset displays the same data plotted on a zoomed Y-axis. (C) Kinetic analysis of the pNPP-based phosphatase assay to determine VMAX and KM.

ED and the small molecule substrate 3-O-methylfluorescein phosphate (OMFP).

Upon hydrolysis, OMFP produces a fluorescent product (3-O-methylfluorescein)

that can be detected at 485/515 nm excitation/emission wavelengths. I

determined assay conditions that yield a linear response over time while

maintaining a good (> 10) S/B. As demonstrated above, Mg2+-chelators,such as

EDTA or EGTA, can be used to inhibit the phosphatase activity of Eya2 ED. Fig

8A demonstrates that EGTA has an IC50 value (the concentration of compound that produces 50% inhibition under assay conditions) of 65 µM and therefore

EGTA or EDTA can be used as positive controls for HTS assay development and screening. DMSO tolerance tests demonstrated that the phosphatase activity of

Eya2 ED is not affected by up to 10% DMSO concentration (Fig. 8B). I next attempted to determine the KM of the Eya2 ED in dephosphorylating OMFP.

Although OMFP cannot reach saturating levels in kinetic analyses, these

experiments reveal that the KM (463.0 ± 104.5 µM to 692.8 ± 83.5 µM) is higher than substrate concentration selected for the assay (100 µM) and will allow for easier identification of weakly competitive inhibitors (Fig. 8C)[170].

Figure 8. Characterization of the OMFP phosphatase assay (A) EGTA can be inhibits the Eya2 phosphatase activity. (B) DMSO (up to 10%) has no effect on the phosphatase activity of Eya2 ED. (C) Kinetic analysis of the OMFP-based phosphatase assay.

Figure 9. Common phosphatase inhibitors do not significantly inhibit Eya2 Compounds showing inhibition display high IC50 values: 11.3 mM for Na2MnO4, 8.2 mM for β-glycerophosphate, 6.6 mM for NaF, and 1.8 mM for Na3VO4. Okadaic acid, L-phenylalanine, cyclosporine A, 1,10-phenanthroline, and phenylarsine oxide do not show inhibition at any concentrations tested.

Known phosphatase inhibitors do not significantly inhibit Eya2’s

phosphatase activity

Before embarking on HTS, I first evaluated the effect of existing

phosphatase inhibitors on Eya using the OMFP-based phosphatase assay

described above. With assistance from Xueni Li, a PRA in our lab, I tested nine

common phosphatase inhibitors against the Eya2 ED including okadaic acid

(inhibitor of Ser/Thr phosphatase PP2A), L-phenylalanine (intestinal alkaline

phosphatase inhibitor), cyclosporine A (calcineurin inhibitor through binding to

cyclophilin), (1, 10)-phenanthroline, phenylarsine oxide (protein tyrosine phosphatase inhibitor), NaF (phospho-serine or threonine inhibitor), Na3VO4

(protein phosphotyrosyl phosphatase inhibitor), Na2MnO4 (protein phosphotyrosyl

phosphatase inhibitor), and β-glycerophosphate. Five inhibitors (okadaic acid, L- 31

phenylalanine, cyclosporine A, 1,10-phenanthroline, and phenylarsine oxide)

were inactive when tested against ED’s phosphatase activity at concentrations

that completely inhibit their cognitive phosphatases (data not shown). Four other

inhibitors only inhibited ED’s phosphatase activity at very high concentrations,

with IC50s of 11.3 mM for Na2MnO4, 8.2 mM for beta-glycerophosphate, 6.6 mM

for NaF, and 1.8 mM for Na3VO4 as seen in Figure. 9. These data suggest that

known phosphatase inhibitors do not significantly inhibit Eya’s phosphatase

activity, as predicted based upon Eya’s unique HAD phosphatase mechanism

and structure. This, in combination with the functional significance of Eya’s

phosphatase activity in tumorigenesis, metastasis, and DNA repair, confirmed

that novel potent and specific inhibitors of the Eya PTP activity are an achievable

goal.

The OMFP-based phosphatase assay is suitable for HTS

To evaluate whether the assay can be adapted for HTS, I first screened

the NIH Diversity Set II collection of compounds in a 96-well format. This

collection contains 480 compounds of structurally diverse chemical scaffolds. I

first determined assay suitability for HTS by calculating a Z-factor. This is a statistical parameter that describes the suitability of the assay for HTS by comparing the variance within positive (uninhibited) samples and the variance within negative (fully inhibited) samples and takes account of the signal window between positive and negative samples[171]. Possible values are between 0.0

and 1.0, where 1.0 is the perfect assay for HTS. The Eya2 ED phosphatase assay has a calculated Z-factor of 0.6, which is acceptable for HTS. I therefore screened the 480 compounds at a single 50 µM concentration in duplicate (Fig.

10A). The Z-factor remained at 0.50 and the S/B remained at 9.73 over the 12 total assay plates. Two compounds were identified in both screens that demonstrated potential inhibition; however, these were determined to be false positives since individual dose response curves covering a larger concentration range resulted in no observed inhibition (Fig. 10B). Even though this collection did not produce any viable lead compounds, it did validate the assay as suitable for screening even larger compound libraries.

Figure 10. HTS of the NIH Structural Diversity Set II (A) Screening results (in duplicate) of the 480 compounds in the Structural Diversity Set II. Two compounds (*) inhibited Eya2 ED in both trials. Z’-factor and S/B remained at 0.50 and 9.73, respectively, across all assay plates. (B) Compounds from (A) do not inhibit Eya2 and are false positives.

An NIH R03 grant was next obtained and the OMFP-based phosphatase assay was optimized and miniaturized at the NIH Chemical Genomics Center

(NCGC) in a 1536-well format for large scale HTS. Eya2 and OMFP 33

concentrations were optimized to produce a linear response with good S/B of >

8-fold, which is reproduced in multiple experiments. To minimize consumption of enzyme in later, large scale screens, a 100 nM Eya2 concentration was selected.

As observed earlier, the solubility of the OMFP substrate in the assay buffer was limited and thus saturation was not reached due to insufficiently high substrate concentrations. The OMFP substrate dose response curve was still in the linear portion at the highest available substrate concentration of 1.25 mM (data not shown). Based upon these results, 25 µM OMFP concentration was selected for further experiments that yielded sufficient assay signal window.

Figure 11. HTS assay optimization for the Eya2 phosphatase (A) Scatter plot of a DMSO plate test of the Eya2 phosphatase assay in a 1536- well plate. The wells in column 2 contain 1 mM EGTA that was used as a positive control. All other wells contain DMSO as the negative (no inhibition) control. (B) Z’ and signal/background (S/B) values of a pilot screen using the LOPAC library. (C) The pilot screen revealed three compounds that inhibit Eya2 phosphatase activity. (D) Chemical structures of the three compounds in C. (E) Two pilot screen compounds inhibit Eya2 phosphate activity using the phospho-peptide substrate, pH2AX. 34

The OMFP-based Eya2 phosphatase assay was miniaturized at the NIH into a 3 µl/well assay in the 1536-well plate format for HTS. A DMSO plate was first used to assess the assay performance. The S/B was 7.75 and the Z factor[171] was 0.7 (Fig. 11A), indicating a robust assay that is suitable for HTS.

A pilot screen was run using the LOPAC library of 1,280 compounds with pharmacologically known activities (Sigma-Aldrich) as well as a ~2800 compound approved drug set (NCGC collection). Each compound was screened in a seven- concentration titration ranging from 0.12 to 76.6 µM as previously described[172].

The average S/B was 7-fold and the Z factor was 0.7 (Fig. 11B), similar to what was observed in the DMSO plate test. After eliminating obvious metal chelators, a total of three compounds were identified as primary hits with a hit selection criteria of IC50 <35 µM and maximal inhibition > 80 %, (Fig. 11C, D). The hit rate from this test screen was 0.55 %, which is an acceptable rate for HTS.

A large scale primary screen identified a class of structurally related initial hits

Due to the successes of the pilot screen, a large-scale screen of 329,717 compounds from the MLPCN (Molecular Libraries Probe Centers Network) was next run at four different concentrations ranging from 0.61 to 76.7 µM in a quantitative HTS format[172]. Dose responses were assigned to different curve classes to evaluate the results of qHTS[172]. Curve classes 1.1 and 2.1 both

Figure 12. A new series of Eya2 ED phosphatase inhibitors were identified in the HTS that inhibit Eya2’s phosphatase activity (A) Structures of compounds that demonstrate inhibitory activity in the large scale HTS. (B) Activity of this series was confirmed in 12-point dose response curves. IC50 values of each compound are listed. (C) Structures of four additional commercially available compounds that have the same N- arylidenebenzohydrazide core. (D) The four compounds tested display varying levels of inhibition towards Eya2 ED.

demonstrate high compound efficacy with either complete (class 1.1) or partial

(class 2.1) dose response curves. Compounds in these two curve classes have the highest chances of reproducing and are typically considered the most promising compounds from a primary screen, although no systematic studies have been reported that link these classes to successfully developed drugs. A few compounds increased the phosphatase activity in the primary screen and may warrant further investigation as possible activators of the Eya2 phosphatase.

Compounds that inhibit the Eya2 phosphatase activity were the primary focus, with the hope that these may be promising leads for anti-cancer drugs. As observed in Table 1, the overall hit rate of the primary screen was low and only 3 inhibitory compounds belonged to class 1.1 and 7 belonged to class 2.1.

However, all three class 1.1 compounds, one class 2.1 compound, and four compounds from other curve classes clearly belong to the same structural chemotype characterized by an N-arylidenebenzohydrazide core (Fig. 12A). This class of compounds was chosen for further characterization.

Table 1. Results of large scale HTS Curve classes are defined in Inglese et al.[172]. Compounds in class 1.1 and 2.1 have high efficacy with either complete (1.1) or partial (2.1) dose response curves and are typically considered the most promising compounds from the screen.

A 12-point dose response study was then carried out and validated all

eight compounds as inhibitors of Eya2 ED’s phosphatase activity in the OMFP-

based phosphatase assay (Fig. 12B). Inhibition assays were run under high Mg2+

(30 µM-1.25 mM) to exclude metal chelators, in the absence of DTT to remove

non-specific covalent modifiers, and by adding compounds immediately before

reading to rule out quenchers that interfere with the fluorescent readout of the

assay (data not shown). All eight compounds passed these false positive tests

and were thus considered highly promising primary screen hits.

Following the initial analysis, four other commercially available compounds

were identified that have the same chemotype as this class of initial hits (Fig.

12C). I tested these compounds using the OMFP-based phosphatase assay (Fig.

12D). NCGC00241225 was most active with an IC50 of 3.5 µM and

NCGC00241223 had an IC50 of 79.1 µM whereas compounds NCGC00241224

and NCGC00241226 were essentially inactive.

The best compounds from this series are active in a pH2AX-based secondary phosphatase assay

To further confirm that this class of compounds are authentic inhibitors of the Eya2 ED phosphatase, I adopted a malachite green-based secondary phosphatase assay using phosphorylated H2AX peptide (pH2AX), a known physiological substrate of the Eya PTP activity[115, 119]. In this assay, inorganic

phosphate released from the H2AX peptide upon dephosphorylation by Eya2 ED

forms a colored complex with malachite green and molybdate, of which the absorbance can be measured at 620 nm. I determined that two of the class 1.1 compounds (another class 1.1 compound was not tested due to limited availability of the compound) and the highly related compound NCGC00241225 were active in the pH2AX assay with IC50 values between 16.8 and 45.2 µM (Fig.

13). Several other compounds with higher IC50 values in the OMFP assay (Fig.

12B, D) do not significantly inhibit Eya2 ED’s PTP activity in the pH2AX assay

(Fig. 13), yet those that do maintain the rank order of IC50 values observed in the primary OMFP assay (Fig. 12).

Figure 13. Inhibitors function in secondary assay A secondary, malachite green based phosphatase assay using a phospho-H2AX peptide as the substrate confirmed the inhibition of Eya2’s phosphatase activity by the best compounds of this class.

In addition, I tested the three hits from the pilot screen and two of the

compounds, NCGC00181091 and NCGC00093729, are active in the pH2AX-

based assay with IC50 values of 32.0 and 18.3 µM, respectively (Fig. 11E). These

compounds therefore may provide an additional chemical scaffold for future

optimization.

The N-arylidenebenzohydrazide compounds do not inhibit other cellular

phosphatases

To evaluate specificity of the chemical series, I performed a counter

screen with the most potent compounds against a number of other cellular

phosphatases. These include: protein tyrosine phosphatase 1B (PTP1B), the

classical cysteine-based PTP that does not require Mg2+ for catalytic activity;

protein phosphatase Mg2+/Mn2+ dependent 1A (PPM1A), a Ser/Thr phosphatase

that, like Eya2 ED, requires Mg2+ for catalytic function[173]; and small C-terminal

Figure 14. Eya2 inhibitors are selective Identified lead compounds specifically inhibit the Eya2 Tyr phosphatase activity, and do not significantly inhibit other cellular phosphatases, including the Mg2+- dependent phosphatase PPM1A (A), protein tyrosine phosphatase 1B (PTP1B) (B), and HAD family member Scp1 (C) in an OMFP-based phosphatase assay.

domain phosphatase 1 (Scp1), a fellow HAD family protein phosphatase[174,

175]. Scp1 is a Mg2+-dependent phosphatase that, like Eya2 ED, uses a catalytic

Asp as the nucleophile in the dephosphorylation reaction. However, Scp1 targets phosphorylated Ser/Thr, as Eya ED is the only known HAD member that targets phosphorylated Tyr. I therefore adopted OMFP-based phosphatase assays for these phosphatases as modified from prior reports[173-175] and tested the activity of the three most potent Eya2 inhibitors (MLS000700592,

MLS000544460, and NCGC00241225) on these phosphatases. These Eya2 ED inhibitors do not significantly inhibit the activity of PTP1B, PPM1A, and Scp1 (Fig.

14A-C), demonstrated specificity to the Eya2 ED PTP activity.

Discussion

The Eya proteins, as novel HAD family PTPs, offer a unique opportunity to identify new therapeutics that can potentially inhibit breast tumorigenesis and metastasis and/or serve as sensitizers for therapeutics that induce DNA damage.

Although many phosphatases have been described as attractive targets for drug discovery[176, 177], the difficulty in obtaining small molecules that inhibit phosphatases in a selective manner is well known and is mostly due to the high degree of similarity between the catalytic domains of the enzymes. Targeting

Eyas provides a distinct advantage; they are mechanistically and structurally distinct from the classical thiol-based PTPs and thus increases the feasibility of obtaining selective inhibitors.

To that end, I adapted a fluorescent HTS phosphatase assay to the Eya2

ED using small molecule OMFP as a substrate. It is standard practice to use

artificial, small molecule substrates for HTS phosphatase assays since these

assays are sensitive, robust, and inexpensive compared to assays using

phosphorylated peptide or protein substrates[178]. Although full length Eya2

would be the most physiological relevant choice, the Eya2 ED was chosen due to

its ease of purification, the quantity of protein that can be obtained, while full

length Eya2 is not stable and cannot be purified. Using only the Eya2 ED also

avoids the complication of the purported Ser/Thr phosphatase domain at the N-

terminus of Eya2[107] that could interfere with the assay results.

After assay development, a quantitative HTS of over 330,000 compounds

was performed successfully and identified a series of small molecule inhibitors of

the Eya2 PTP activity that share a common chemotype. Analogs screened

containing this chemotype ranged from low micromolar IC50 values to no activity,

indicating inherent structure-activity-relationship (SAR) within the series. As evident from the primary screen, benzohydrazide substitutions show remarkable selectivity. For example, NCGC00241225 (no substitutions) has an IC50 value of

3.5 µM and MLS000700592 (-NO2 substitution at the meta position)

demonstrates a slight increase in IC50 value to 2.3 µM. However,

NCGC00241226, containing a -NO2 substitution at the ortho position, loses all

activity. meta-substituted halides (MLS000544460 and MLS000544490) are also

tolerated and show similar activity to non-substituted compound

(NCGC00241225). The furanyl-2-thio-2-pyridine substituent of the N-arylidene

functional group provided best activity although other benzylidene substituents

seemed to be tolerated and indicates larger modification will be tolerated with

these moieties in SAR optimization.

The qHTS results were confirmed using an orthogonal malachite green-

based phosphatase assay that monitors phosphate release from the pH2AX

peptide substrate and demonstrates that the compounds function against Eya2

ED using a physiologically relevant substrate. However, a reduction in IC50

values is observed, but can be attributed to the difference in assay conditions

and substrate selection, as IC50 values reflect inhibition under a defined set of

assay conditions. Of interest, the most potent compounds in the OMFP-based assay (MLS000700592, MLS000544460, and NCGC00241225) retain their inhibitory effects in the pH2AX assay, indicating that they are bona fide inhibitors of the Eya2 ED PTP activity. The importance of using a secondary assay is evident in that while two of the pilot screen identified compounds demonstrate inhibition, NCGC00181091 and NCGC00093729, the third compound,

NCGC00013130, failed to show activity and, as such, is a false positive through possibly interfering with the OMFP assay and not inhibiting the Eya2 ED PTP itself.

The goal of this research was to obtain selective inhibitors against the Eya

PTP activity by taking advantage of the unique mechanistic and structural differences between the HAD PTP Eya and classical phosphatases. My specificity assays demonstrated that these Eya2 inhibitors do not significantly inhibit several other cellular phosphatases. PTP1B was chosen as a classical

thiol-based PTP that has been intensely targeted by several groups and that specificity has been difficult to achieve[125]. Selectivity over the phosphatase

PPM1A demonstrated that these compounds do not simply chelate Mg2+ as a mechanism of action as PPM1A is an Mg2+-dependent phosphatase. Of particular interest, the compounds did not inhibit Scp1. Like Eya2 ED, Scp1 is a

HAD phosphatase and shares the signature HAD catalytic structural motif albeit targeting Ser/Thr phosphorylation. It is significant that this series does not simply target HAD phosphatase members as this would imply that they function against a similar catalytic or structural motif common to HAD members. However, this data indicates that not only is Eya2 ED selectively achieved over classical thiol- based phosphatases, but that selectivity over other mechanistically and structurally related HAD phosphatases is possible.

With this research, we have identified lead compounds that need to be characterized and improved before advancement as potential therapeutics.

Because these are relatively low affinity compounds, additional work is needed to improve their potency through intensive SAR studies. Understanding these compounds’ binding mechanism is critical for SAR studies. The next chapter of my thesis describes my effort in deciphering the binding mechanism of these compounds.

Methods

Protein expression and purification

Human Eya2 ED (residue 253-538) that contains the Eya phosphatase

activity was sub-cloned into the pGEX-6P1 (GE Healthcare) vector using the

BamHI and XhoI site and confirmed by DNA sequencing. Plasmids containing

these constructs were transformed into E. coli strain XA90. Cells were grown

until OD600 reached 0.8-1.0 and protein expression was induced at 20°C with 0.2

mM IPTG for 20 hours. Cell pellets were lysed by sonication in buffer L (50 mM

Tris, pH 7.5, 250 mM NaCl, 5% glycerol, 1 mM DTT) containing protease

inhibitors pepstatin A, leupeptin, and PMSF. Lysates were cleared via

centrifugation (2 x 45 minutes at 18,000 x g). The supernatant containing GST-

Eya2 ED proteins was loaded via gravity on glutathione-Sepharose 4B resin (GE

Healthcare) and thoroughly washed with buffer L. ED protein was cleaved from the glutathione resin with PreScission protease at 4°C for 16 hours, eluted, and concentrated. ED protein was further purified on a Superdex 200 size exclusion column (GE Healthcare) using buffer L. Purified protein was aliquoted and stored at -80°C.

pNPP-based Eya phosphatase assay

Eya activity was measured in clear, 96-well microtiter plates (Greiner Bio- one) with pNPP as the substrate (Sigma-Aldrich). Dephosphorylation of pNPP produces a colored product, pNP that can be observed at 405 nm. Assay conditions are 50 mM MES, pH 6.5, 50 mM NaCl, 5 µM MgCl2, 0.05% BSA, and

1 mM DTT. Reactions were stopped with addition of 75 mM EDTA and read at

405 nm using a Spectramax PLUS 384 plate reader (Molecular Devices).

OMFP-based Eya phosphatase assay

The activity of ED was measured in 50 µL reactions in black, 96-well, half-

volume microtiter plates (Greiner Bio-one) with OMFP (3-O-methylfluorescein

phosphate, Sigma-Aldrich) as the substrate. Upon dephosphorylation, OMFP is

converted to a fluorescent product OMF. Enzyme and substrate concentrations

were optimized to have a linear response during the assay, consume less than

15% of substrate after one hour, and substrate concentration below its Km. The final assay condition is 50 mM MES, pH 6.5, 50 mM NaCl, 5 µM MgCl2, 0.05%

BSA, 1 mM DTT, and the reaction contained 50 nM Eya2 ED and 100 µM OMFP.

Reactions were started by the addition of OMFP and were continued for 1 hour at

room temperature and terminated by the addition of 75 mM EDTA. ED is stable

and retains its activity after at least an hour at room temperature (data not

shown); therefore room temperature was chosen as the reaction temperature for

convenience. Fluorescence intensity was measured at 485/515 nm

excitation/emission on a Fluoromax-3 plate reader (Horiba Jobin Yvon).

Miniaturized Eya phosphatase assay for HTS

The OMFP-based Eya2 ED phosphatase assay were miniaturized and optimized in 1536-well black assay plates (Greiner Bio-one). 1.5 µL/well of 200 nM Eya2 ED was incubated with or without 23 nL compound (dissolved in DMSO

and the final concentration of DMSO in the well is 0.76%) or DMSO control for 10 minutes followed by an addition of 1.5 µL/well of 50 µM OMFP and incubated for

30 minutes. The resulting fluorescence intensity was measured on a Viewlux plate reader (PerkinElmer) with an excitation wavelength of 485 nm and emission of 525 nm. Since there were no Eya2 inhibitors available during the assay development, we used EGTA as a positive control to access the assay quality.

Compound library and instruments for liquid handling

The LOPAC library (Library of Pharmacologically Active Compounds,

Sigma-Aldrich) consisting of 1,280 compounds was used for the assay validation.

The collection of 331,609 compounds for the primary screen was provided by the

NIH’s Molecular Library Initiative (http://pubchem.ncbi.nlm.nih.gov/). All compounds were dissolved in DMSO as 10 mM stock solutions. All compounds were serially diluted at 1:5 ratio in DMSO in 384-well plates for 4 concentrations using a CyBi®-Well dispensing station with a 384-well head (Cybio) and then reformatted into 1536-well plates at 7 µL/well. An automated dispensing station

(BioRAPTR FRD, Beckman Coulter) was used to dispense reagents into 1536- well plates at volumes of 1-3 µL/well. Compounds were transferred to 1536-well assay plates in 23 nL/well using an automated pin-tool station (Kalypsys®). The purity of all compounds in the library is greater than 98% as analyzed by HPLC.

Further analyses by 1H NMR spectroscopy and mass spectrometry confirmed their structural identity. In addition, the bioactivity of 10 mM stock samples recapitulates in the primary assay after several months of storage. One of the

key compounds (MLS000544460) has been incubated with pH 4, 7 and 10 buffer and is stable over extended periods of time.

HTS data analysis

The primary screening data was analyzed as previously described[179].

IC50 values were calculated from the fluorescence signal intensity using the

Prism software (Graphpad Software, Inc.). The Z’ factor index of assay quality control[171] was defined as 1-(3*SSR/R), where SSR is the summation of the standard deviation of positive inhibition controls and negative inhibition controls and R is the mean of the positive controls minus the mean of negative controls.

All values were expressed as mean +/- SD.

pH2AX-based Eya phosphatase assay

pH2AX phosphatase assays (50 µL) were carried out in transparent, 96- well, half-area microplates (Greiner Bio-One). The assay was performed at pH

6.0 in 50 mM MES, 50 mM NaCl, 5 µM MgCl2, 0.05% BSA, 1 mM DTT and contained 3.9 µM ED and 50 µM pH2AX peptide (KATQASQEpY, Abgent).

Because of the lower sensitivity of the malachite green assay compared to the

OMFP-based phosphatase assay, a much higher enzyme concentration (3.9 µM) is used to achieve sufficient assay signal/background. Reactions were allowed to proceed for 40 minutes at 37°C and terminated by the addition of 100 µL malachite green solution (Millipore). The free phosphate released from dephosphorylation forms a green complex with malachite green and molybdate

that can be monitored using absorbance at 620 nm. After a 20-minute incubation

at room temperature, the absorbance was measured using a Spectramax PLUS

384 plate reader (Molecular Devices).

To evaluate the effect of a compound on ED’s phosphatase activity,

various concentrations of the compound were incubated with ED for 10 minutes

prior to the addition of pH2AX that starts the reaction. Dose response curves

were generated and IC50 values calculated using Prism.

Phosphatase assays of PTP1B, PPM1A, and Scp1

The phosphatase assay of PTP1B was conducted in 30 mM Tris, pH 7.5,

75 mM NaCl, 1.25 mM MgCl2, 1 mM EDTA, 0.033% BSA, 1 mM DTT using 20 nM PTP1B (R&D Systems) and 100 µM OMFP as the substrate. To evaluate the effect of a compound on PTP1B’s phosphatase activity, various concentrations of the compound were incubated with PTP1B for 10 minutes prior to the addition of

OMFP to start the reaction. Reactions were in 50 µL volumes and were set up in black, 96-well, half area microplates using an automated liquid handling system

(Janus) and bulk liquid dispenser (Biotek). Reactions proceeded for 60 minutes at room temperature followed by the addition of Na3VO4 to terminate the reaction.

Fluorescence intensity was measured using an EnVision plate reader (Perkin

Elmer).

The phosphatase assay of PPM1A (ProSpec) was conducted in 50 mM

Tris, pH 7.5, 0.1 mM EDTA, 0.5 mM DTT, 1.25 mM MgCl2 using 10.7 nM PPM1A

and 100 µM OMFP. The effect of compounds on PPM1A’s phosphatase activity

was evaluated similarly as for PTP1B, except that EDTA was used to terminate the reaction.

The phosphatase assay of Scp1 (a gift from Dr. Jessie Zhang, UT-Austin) was conducted in 50 mM MES, pH 5.5, 25 mM NaCl, 1.25 mM MgCl2 using 120 nM Scp1 and 100 µM OMFP. The effect of compounds on Scp1’s phosphatase activity was evaluated similarly as for PTP1B, except that EDTA was used to terminate the reaction.

CHAPTER III

DETERMINING THE MECHANISM OF ACTION OF EYA2 INHIBITORS

Abstract

Eya proteins are essential co-activators of the Six family of transcription factors and contain a unique tyrosine phosphatase activity belonging to the haloacid dehalogenase family of phosphatases. The phosphatase activity of Eya is important for the transcription of a subset of Six1-target genes, and also directs cells to a repair rather than apoptosis pathway upon DNA damage. Furthermore, the Eya phosphatase activity has been shown to mediate transformation, invasion, migration, and metastasis of breast cancer cells, making it a potential new drug target for breast cancer. We have identified a class of N- arylidenebenzohydrazide compounds that specifically inhibit the Eya2 phosphatase over other cellular phosphatases. Herein, I determined that these inhibitors selectively inhibit the phosphatase activity of Eya2, but not Eya3.

Mutagenesis results suggest that this class of compounds does not bind to the active site and the binding mechanism does not require Mg2+. Rather, these compounds likely bind to a site on the opposite face from the active site to function as allosteric inhibitors. This class of compounds inhibits Eya2 phosphatase-dependent cell migration, setting the foundation for these compounds to be developed into chemical probes to understand the specific function of the Eya2 phosphatase, and serving as a prototype for the development of Eya2 phosphatase-specific anti-cancer drugs.

Introduction

The Eyes Absent family of proteins (Eya1-4) were first discovered as essential coactivators[54, 88, 92, 169] of the homeobox transcription factor

Six1[32, 54, 59, 60, 83, 85, 180, 181]. Both Six1 and Eya proteins are necessary for cellular proliferation in a number of different cell types[26, 54, 85]. In support of a cooperative interaction between Six1 and Eya, the Eya1 knockout (KO) mouse phenocopies the Six1 KO mouse, and organ defects in both KO mice are due to a decrease in proliferation and an increase in apoptosis[53, 182]. Both

Six1 and Eya1 are linked to Branchio-Oto-Renal (BOR) syndrome, which is characterized by branchial defects, hearing loss, and renal abnormalities[59, 60].

These data suggest that Six1 and Eya cooperate in normal development of human and mouse tissues. In addition, Eya proteins have been linked to many cancers in which Six1 is overexpressed, such as Wilms’ tumor, ovarian cancer, and breast cancer [67, 70, 85, 87]. Recent evidence demonstrates that Eya2 knockdown inhibits the ability of Six1 to induce TGF-β signaling, epithelial-to- mesenchymal transition (EMT), and tumor initiating cell (TIC) characteristics in breast cancer, properties that are associated with Six1-induced breast tumorigenesis and metastasis[87]. Furthermore, analyses of microarray data from 295 breast tumors demonstrated that coordinated overexpression of Six1 and Eya1 or Eya2 together significantly predict decreased time to relapse and metastasis as well as shortened survival, whereas the overexpression of individual genes does not[87]. These clinical data strongly suggest that Eya

proteins are required for Six1-mediated breast tumorigenesis, and that Eya family members can cooperate with Six1 to confer poor prognosis in breast cancer.

Eya proteins also contain a unique protein tyrosine phosphatase (PTP) activity. In Drosophila, while the PTP activity of Eya is not globally required for the ability of Six1 to induce transcription, it is required to induce transcription for a subset of its target genes[169]. In fact, it is suggested that the Eya PTP activity can switch the Six1 transcriptional complex from a repressor to an activator for some Six1-induced genes in mice[54]. The Eya proteins therefore represent the first transcriptional co-activators with intrinsic phosphatase activity that can modulate transcription[54, 88, 89]. Eya proteins belong to the haloacid dehalogenase (HAD) family of enzymes and are the only HAD phosphatases that contain PTP activity. A few other HAD members target proteins, such as Scp1 and Chronophin, but most HAD phosphatases do not have protein substrates[88,

95] and those that do target protein substrates are serine/threonine phosphatases[95, 119]. Regardless of substrate, all HAD phosphatases share a highly conserved catalytic core that mechanistically uses an aspartic acid as the catalytic active-site residue instead of a cysteine used most often by classical phosphatases[95]. HAD phosphatases also contain a variable cap structure that may assist in the catalysis. The helical bundle cap observed in the Eya members is unique and not observed in any other HAD protein[95, 102].

Recently, Eya proteins, and in particular their phosphatase activity, were shown to be critical for transformation, migration, invasion, and metastasis of breast cancer cells[183] and Eya1 PTP is essential for breast cancer

proliferation[184]. In addition, Eya’s phosphatase activity plays a critical role in the DNA damage response via its ability to dephosphorylate the histone variant

H2AX upon DNA damage, which directs cells to a repair pathway instead of inducing apoptosis[115, 119]. These data suggest that inhibiting Eya PTP activity may not only inhibit tumor onset and progression, but also sensitize cells to commonly used chemotherapeutics that act through inducing DNA damage.

Importantly, drugs that target the Eya phosphatases may confer limited side effects since Eya members are highly expressed in developing tissues, but are generally not expressed in most normal adult tissues[15].

We have identified a series of inhibitors targeting Eya2’s phosphatase activity with IC50 values of 2-73 µM (Chapter 2). Although these compounds demonstrate specificity towards Eya2 PTP activity, their potency is less than ideal. As such, these compounds serve as lead compounds and further optimization is needed to improve their potency and other pharmacokinetic properties. A first step is to understand how these inhibitors bind to Eya2.

Understanding the interactions of a compound within its binding site allows the medicinal chemistry efforts to be directed towards designing compounds that have higher complementarity to the binding site electrostatics and steric characteristics. This chapter details my efforts to understand the mechanism of action of the Eya2 inhibitors and to identify a potential binding site of these compounds.

Results

HTS identified a new chemical series that specifically inhibit Eya2’s

phosphatase activity

We have identified a series of structurally related compounds containing

an N-arylidenebenzohydrazide core that inhibit Eya2 PTP activity with varying

potency using an OMFP-based phosphatase assay and the Eya2 Eya Domain

(ED) (structures of selected members of this class are shown in Fig. 15A)[120].

These compounds did not inhibit three other cellular phosphatases (Scp1,

PPM1A, and PTP1B), demonstrating specificity toward the Eya2 ED[120]. Since

there are multiple Eya family members with high homology in the ED, I tested the

ability of this series of compounds to inhibit another Eya member. I expressed

and purified large quantities of human Eya3 ED but was unable to express and

purify sufficient quantities of human Eya1 ED or Eya4 ED. Using the OMFP-

based phosphatase assay, I demonstrated that this class of compounds has no

discernible inhibition against Eya3 ED (results with MLS000544460 is shown in

Fig. 15B). To further validate this selectivity, I measured the binding affinity of

compound MLS000544460 to Eya2 ED and Eya3 ED using ITC experiments.

MLS000544460 binds to the Eya2 ED with a KD of 2.0 ± 0.3 µM (Fig. 15C), but it

does not bind Eya3 (Fig. 15D), indicating that this class is selective towards

Eya2. I further used ITC to determine the affinities of several other identified

Eya2 inhibitors. MLS000700592 had a KD of 2.4 ± 0.5 (Fig 16A) and

NCGC00241225 had a KD of 3.2 ± 0.7 (Fig. 16B). In addition, the structurally related but inactive compound NCGC00241224 does not bind to Eya2 ED in ITC

Figure 15. A class of N-arylidenebenzohydrazide – containing compounds selectively inhibits Eya2 (A) Chemical structures of representative compounds identified in HTS (MLS000544460 and MLS000585814) and a structurally related but inactive compound (NCGC00241224). (B) MLS000544460 inhibits Eya2 but not Eya3 in an OMFP-based phosphatase assay. (C) MLS000544460 binds to Eya2 ED (KD = 2.0 ± 0.3 µM) (D) MLS000544460 does not bind Eya3 ED.

Figure 16. Identified inhibitors within the series binds Eya2 ED but does not induce significant structural changes (A) The active compound MLS000700592 binds Eya2 ED (KD = 2.4 ± 0.5 µM) (B) The active compound NCGC00241225 binds Eya2 ED (KD = 3.2 ± 0.7 µM) (E) The inactive compound NCGC00241224 does not bind to Eya2 ED. (D) NCGC00241225 does not induce significant changes to Eya2 ED structure upon binding using circular dichroism.

experiments (Fig. 16C), indicating that the binding of this chemical series is

driven by specific interactions of chemical moieties in the active compounds

rather than non-specific binding caused by the common chemotype within this

class. I next used circular dichroism to monitor secondary structural effects on

Eya2 ED upon compound binding. Selected as a representative inhibitor from

this series, NCGC00241225 does not produce significant changes to Eya2 ED

secondary structure upon binding as judging from the CD spectrum (Fig. 16D).

Enzyme kinetic analyses suggest that the N-arylidenebenzohydrazide

compounds are mixed mode inhibitors

To determine how this compound series functions, I set up a classical

kinetic experiment using OMFP. This was unsuccessful at generating useful data

due to limited solubility of OMFP. Our collaborators at NIH adapted the assay to

use the more soluble substrate fluorescein diphosphate (FDP) to generate high

stock concentrations (100 mM) to be used in the assays. Accurate Km and Vmax

values were obtained and a Lineweaver-Burke plot was generated for Eya2 and the lead compound MLS000544460 (Fig. 17). These data suggest a mixed mode behavior, between competitive and non-competitive inhibition, indicating that the enzyme is able to partially accommodate the inhibitor and the substrate at the same time.

Figure 17. Kinetic experiments Enzymatic kinetic experiments indicate that the compounds are mixed-mode inhibitors as evidence by the Lineweaver-Burke plot of MLS000544460 and Eya2

Attempt to determine the binding mechanism using X-ray crystallography

Since the X-ray crystal structure of Eya2 ED has been solved[102], I

decided to use crystallography for inhibitor-soaking or co-crystallization trials as a

method to determine how the compounds bind. I generated the exact construct

used, Eya2 268-538 in the pET-28a vector. I expressed and purified high

quantities of pure protein and produced crystals at 100 mM Bis-Tris pH 7.2, 2.4

M NaCl which is slightly different from the published condition (100 mM Bis-Tris

pH 6.4, 2.4 M NaCl). Crystals were soaked with various concentrations of

inhibitor (50 µM to 1 mM) for various amounts of time (several minutes to greater

than 24 hours) in the cryo-protectant (25% glycerol in crystallization buffer).

Elongated soaking produced poor diffraction patterns, whereas soaks of up to

several hours produced no discernible change in the diffraction pattern. I

harvested, screened, and selected several to collect data at the Advanced

Photon Source synchrotron at Argonne National Laboratory in Argonne, IL.

Fifteen inhibitor-soaked crystals were screened at the synchrotron and 7

complete data sets were collected on Eya2-soaked crystals containing the

inhibitors MLS000544460, MLS000700592, or NCGC00241225. Data were

processed using HKL2000 and the Eya2 structure was determined to 2.6 Å for

several inhibitor-soaked crystals using CCP4i software suite. Unfortunately, all

structures solved were of the apo-Eya2 ED structure and did not contain density

for the inhibitor bound at any position in Eya2. In addition to soaking trials, I

attempted to generate the co-crystal structure of the inhibitor-bound Eya2.

Various concentrations of Eya2 ED protein (both Eya2 253-538 and Eya2 268-

538) were incubated with compound at various concentrations up to 100 µM prior

to dispensing on the slide for hanging drop set-up. Additionally, compound was

directly added to protein drops and several crystal screens including the

Hampton Screen I/II, JMAC, and Silver Bullet were also used to try to generate

co-crystals. Unfortunately, no crystals were observed in co-crystallization

attempts.

Attempt to determine the binding mechanism using NMR

Since crystallography was not successful, I attempted to use Nuclear

Magnetic Resonance (NMR) spectroscopy to identify the compound binding site.

Since a 2H, 15N, 13C-triple-labeled sample would be needed for peak assignments

and triple-labeling requires a refolding protocol, I first used a His-tagged

construct (6xHis-Eya2 253-538) to generate proteins for NMR experiments.

However, this construct at 75-300 µM generated poor HSQC spectra with a high amount of aggregation even at low concentrations. To resolve this, I switched to a GST-fusion construct that is used for enzymatic assays (GST-Eya2 253-538).

Although not suitable for generating a triple-labeled sample (GST does not

refold), this construct allowed for initial binding studies as it produced a more

dispersed HSQC spectrum. Using concentrations less than 150 µM produced

HSQC spectra in which many unique peaks could be observed, although

significant peak overlap at the center of the spectrum is still present. Using this

construct, I incubated the inhibitor MLS000544460, the inactive compound

NCGC002241224, and vehicle (DMSO) with 15N-Eya2 ED and obtained HSQC

spectra for each condition. The inactive compound NCGC00241224 did not

produce significant changes in the NMR spectrum compared to vehicle control

(0.1% DMSO) (Fig. 18). However, when the active compound MLS000544460

was present, significant changes are observed in the HSQC spectrum of Eya2

ED including peak shifts, the appearance and disappearance of peaks, and a

slight overall improvement in spectra quality.

Figure 18. Eya2 HSQC with inhibitor demonstrates binding HSQC of 15N-Eya2 ED in the presence of vehicle (DMSO), inactive compound (NCGC00241224), and inhibitor (MLS000544460). Select regions are expanded to provide spectrum details. MLS000544460 but not NCGC00241224 generated significant changes in the Eya2 HSQC spectrum.

I attempted to optimize the Eya2 construct for NMR to further improve the

quality of the HSQC spectrum. It is evident in the X-ray crystal structure that a large, unstructured loop (amino acids 356-368) is present in the Eya2 ED (absent 62

in the crystal structure, indicating flexibility)[102]. Amino acids in flexible regions

of a protein often show up as large, intense peaks and cause peak overlaps in an

NMR spectrum, much like those observed in my Eya2 HSQC spectra. I therefore

used mutagenesis to remove this loop (Eya2 ED Δ356-368) and generated 15N-

Eya2 ED Δ356-368 to see if an improvement in spectra quality is observed.

Unfortunately, the removal of this loop provided no increase in spectra quality or the disappearance of the dominant and overlapping peaks observed in the

HSQC spectra.

I next attempted to optimize buffer conditions that would reduce the

amount of aggregation and peak overlap observed by screening for conditions

that promote the solubility Eya2 ED, following a published methodology, shown in

Figure 19[185]. I combined 1 µL of 8.5 mg/mL Eya2 ED 253-538 with 1 µL of

various buffer conditions and placed the protein-buffer mixture in a hanging drop

vapor diffusion setup to slowly concentrate the sample. Twenty-four buffer conditions were screened at pH values between 4.5 and 8.0. Levels of aggregation were determined daily by scoring the extent of precipitation on the surface of the drop, according to Figure 19. Concentrated Eya2 ED generally remained fully soluble (no precipitation observed) when the pH was greater than

7.5, although pH 7.0 was acceptable for three buffers (Tris, HEPES, and Bicine).

I chose to use Bicine buffer at pH 7.5 containing 50 mM NaCl for further optimization using additives. I next screened select additives including reductants, organic solvents, dissociating agents, chaotropes, polyols, detergents, osmolytes, polymers, and carbohydrates. A number of additives

Figure 19. Solubility protocol for increasing solubility of NMR samples Systematic, vapor-diffusion method to increase protein solubility for NMR. This protocol was adapted for Eya2 NMR experiments. Eya2 ED is mixed with buffer solutions and precipitation is observed and scored. This methodology screens numerous conditions to determine those that increase long term stability and preventing precipitation. Figure from reference [185].

prevented the precipitation of Eya2 over several days. DMSO was slightly more effective at preventing aggregation than sucrose and glycine which were slightly better than sacrosine and phenol which were slightly better than L-proline.

Unfortunately, I was unable to test these additives in NMR due to time

constraints and successes in other areas, so it remains to be seen whether they

alleviate Eya2 ED aggregation in the HSQC spectra.

In order to identify specific amino acids involved in the binding, I would

need to produce a triple-labeled protein using a purification method involving

denaturing and refolding Eya2 ED. I attempted this using unlabeled GST-Eya2

ED, but it was unable to refold after cleavage from GST. A refolding protocol

using a His-tag would be necessary since a His-tag can be refolded. Optimization

with this strategy may be difficult as the spectra produced using the His-cleaved

15N-Eya2 ED is of poor quality compared to the GST-cleaved construct.

Extensive effort will be needed if this method is to be used to identify the inhibitor binding site or solve the solution structure of inhibitor-bound Eya2 ED.

Compounds likely do not bind in the active site and do not require Mg2+ for

function

Since X-ray crystallography and NMR were unsuccessful in identifying the

compound binding site, I used in silico modeling to computationally dock the

compounds to Eya2 ED. With the assistance of Dr. Aaron Patrick, I modeled the

inhibitors into the Eya2 active site (PDBs: 3GEB and 3HBO) using Autodock 4.

Additionally, our collaborators at the NIH provided several models based upon

their own modeling. Together, these provided several potential models of how

the compound may bind the active site that can be tested experimentally using a

mutagenesis strategy. 65

Since these compounds contain a number of heteroatoms capable of

coordinating an Mg2+ ion, one model predicts that these inhibitors act as local

Mg2+-chelators in the active site. This model involves MLS000544460 bound in

the active site and orientated such that the N-arylidenebenzohydrazide oxygen

atom binds to the Mg2+ required for catalytic function while maintaining metal-

chelating geometry (Fig. 20A). The remainder of the compound is oriented within

the active site in such a manner that the pyridine ring inserts into the hydrophobic

active site pocket forming hydrogen bonds with R293, and the fluorobenzyl group

on the other side points to the region near H283 to generate pi-stacking

interactions (Fig. 21A). I used both biochemical methods and site-directed

mutagenesis to evaluate the Mg2+-binding ability of the compounds and to test additional contacts made within the active site.

In order to examine the Mg2+-chelating ability of the inhibitor in this model,

our collaborators at NIH obtained the UV spectra of compound MLS00054460 in

the presence of increasing concentrations of Mg2+ to determine whether this

inhibitor has intrinsic affinity towards Mg2+. There was no change in the UV

spectra of MLS00054460 in aqueous buffers (including the phosphatase assay

buffer) with increasing concentrations of Mg2+, indicating that this compound has very low, if any, affinity to Mg2+ in aqueous solution. However, in 100%

acetonitrile[186], a dose-dependent shift in the maximum absorbance wavelength

2+ 2+ (λmax) of MLS000544460 from 322 nm (0 mM Mg ) to 334 nm (100 mM Mg ) was observed (Fig 20B). Similar studies using the monovalent cation Na+

Figure 20. The interaction between Eya2 ED and the N- arylidenebenzohydrazide-containing compound does not require Mg2+ (A) Structures of two active compounds with heteroatoms that can potential coordinate Mg2+ shown in red. (B) UV absorption spectrum of compound 2+ MLS000544460 in the presence of Mg . λmax shifted from 322 nm to 334 nm with 2+ increasing concentrations of Mg . (C) The λmax shift observed in (B) is quantified as a function of Mg2+ concentration. (D) The UV absorption spectrum of + MLS000544460 in the presence of Na . λmax remained at 323 nm when titrated with Na+. (E) UV absorption spectrum of the low activity analog MLS000585814 2+ 2+ in the presence of Mg . λmax shifted from 316 nm to 327 nm with increasing Mg concentrations. (F) The λmax shift observed in (E) is quantified as a function of Mg2+ concentration. 67

Figure 21. Eya2 inhibitors likely do not bind in the active site (A) A representative docking model of MLS000544460 in the active site involving coordination of the Mg2+. (B) Representative dose response curves of active site mutations demonstrate no effect on the ability of MLS000544460 to inhibit the mutant Eya2.

produced no discernible shift (Fig. 20D), indicating that in acetonitrile these

compounds have intrinsic affinity for Mg2+ over monovalent cations such as Na+.

Subsequent analysis using the low activity inhibitor MLS000585814 (IC50 = 73

2+ µM) produced a similar Mg -induced, dose-dependent λmax shift from 316 nm (0

mM Mg2+) to 327 nm (50 mM Mg2+) (Fig. 20E). Although this compound has low

activity in the inhibition assay, it retained the ability to bind Mg2+ with similar affinity as the more potent compound MLS000544460 (Fig. 20C, F). These data suggest that this inhibitor series does not use its Mg2+-binding ability as the main

Figure 22. Binding of MLS000544460 does not require Mg2+ 2+ (A) MLS000544460 binds with lower affinity in the presence of 5 mM Mg (KD = 6.0 ± 1.2 µM). (B) MLS000544460 has higher affinity towards Eya2 when Mg2+ is removed with 10 mM EDTA (KD = 0.80 ± 0.04 µM).

mechanism for binding, but instead relies on direct and specific interactions to

Eya2 ED.

Concurrently using ITC, I examined the requirement of Mg2+ in the binding

mechanism. I determined the compound MLS000544460 binds Eya2 with a KD

value of 2.0 ± 0.3 µM containing only the Mg2+ carried through in the purification

2+ (Fig. 15C). The addition of 5 mM Mg shifted the KD to 6.0 ± 1.2 µM (Fig. 22A), implying that high Mg2+ concentrations potentially interferes with compound

binding. However, when Eya2 is dialyzed with 10 mM EDTA to remove all bound

2+ Mg , a higher affinity interaction was observed with a KD = 0.80 ± 0.04 µM (Fig.

22B). Based upon these results, these compounds likely do not rely on Mg2+- interactions as part of the binding mechanism.

In parallel with the Mg2+-dependency studies, I generated Eya2 ED active site mutants that abolish specific hydrogen bonding and pi-stacking interactions with this model (H283A, R293A, and H283A/R293A). These mutants were purified using conditions similar to wild type Eya2 ED and their phosphatase activity was tested. These mutants were inhibited by MLS000544460 with comparable IC50 values as wild type Eya2 ED (Fig. 21B). Together, the

mutagenesis and the Mg2+ ITC data suggest that a model involving coordination

of the active site Mg2+ ion is incorrect.

Figure 23. Representative docking models of inhibitors in the active site without Mg2+ Select Inhibitor docking models that do not require Mg2+ for binding. Amino acids colored yellow are the aspartic acids that coordinate the Mg2+ ion. Green amino acids have been mutated to abolish specific interaction or sterically block the inhibitor from binding.

Figure 24. Dose response curves using BOR mutants Dose response curves of BOR mutations demonstrate no effect on the ability of MLS000544460 to inhibit the mutant Eya2 proteins and do not reveal a potential binding site. BOR mutants are single amino acid changes that cause BOR syndrome.

I tested additional in silico derived, active site models (examples given in

Fig. 23) by generating Eya2 ED mutants that either abolish specific interactions

(T289A, T448A, T449A, Q450A, E505A, and E506A) or sterically block inhibitor binding (T448F and S521F). Additional Eya2 ED double-mutants were generated

(T289A/E505A and T448F/E505A) in case single amino acid changes were not sufficient to affect compound inhibition. Of these mutants, some (S521F, E505A)

displayed lower activity compared to wild type Eya2 ED, but all were inhibited by

MLS000544460. The Eya2 ED mutants E506A and T449A were inactive. These

additional results in combination with the observation that Mg2+ is not necessary

for compound binding suggest that the compound is likely not binding in the

active site.

Compounds likely bind to an allosteric site as determined by mutational

analysis

Since in silico modeling and site-directed mutagenesis suggests that the

compound does not bind in the active site, I first screened a collection of Eya2

BOR mutants (E309V, G372S, D375G, R386Q, V458F, V496E, M515T, and

ΔV499) for loss of inhibition by MLS000544460. If a mutant demonstrated a loss

of inhibition in the presence of the compounds, it would indicate that the mutated

amino acid may be involved in compound binding. All Eya2 ED BOR mutants

retained similar IC50 values as wild type (Fig. 24), indicating these residues are not involved in the compound binding site.

Since the compounds selectively inhibit Eya2 over Eya3, I next examined the amino acid sequence differences between Eya2 ED and Eya3 ED as a method to determine where the compounds bind. Eya2 ED and Eya3 ED are highly conserved with a 68% sequence identity (Fig. 25A) and when mapped to the Eya2 ED crystal structure (PDB: 3GEB) several regions containing multiple differences were observed (Fig. 25B). Four sets of Eya2 ED mutants were generated that replace Eya2 residues with corresponding Eya3 residues on the

Eya2 ED surface: 298P/T300V/S301V/R303G/I304S, L423S/T426G/H427T, 72

Figure 25. Sequence and surface differences between Eya2 and Eya3 (A) Sequence alignment of Eya2 ED and Eya3 ED. Residues in grey are not in the ED, but were in the purified Eya constructs used for phosphatase, ITC, and NMR experiments. Symbols beneath sequences represent conserved residues (*), strongly similar properties (:), weakly similar properties (;), or no similarity (no symbol). Eya domain begins at (|). Residues highlighted in red are surface amino acids that were mutated in Figure 11A,B. (B) A surface representation of the Eya2 ED structure with surface residues that differ between Eya2 and Eya3 highlighted in different colors.

Figure 26. Eya2 to Eya3 mutations cause a loss of inhibitor activity (A) Eya2 ED mutants generated to change corresponding Eya2 residues to those of Eya3. (B) Dose response curves of Eya2 mutants. One mutant, L423S/T426G/H427T, demonstrated a loss of inhibition by MLS000544460, indicating a potential binding site.

Figure 27. ITC experiments show that MLS000544460 does not bind to mutants at allosteric site (A) ITC results of Eya2 L423S/T426G/H427T, (B) ITC results of Eya2 L417W, and (C) ITC results of Eya2 L425N. All samples used MLS000544460 as a representative inhibitor.

T404P/P405Q/W410L/L411Q, and V504R/Q508I. Two additional Eya2 ED mutants, Δ359-368 and Δ356-368, were generated to delete various portions of the unstructured region present in the Eya2 ED crystal structure that contains little sequence conservation between Eya2 and Eya3 (Fig. 26A). The phosphatase activity of these mutants was tested and dose response curves with compound MLS000544460 were generated. One mutant,

T298P/T300V/S301V/R303G/I304S, was inactive. Eya2 ED mutants I402W and

M311W were generated from docking models at this site. These mutants retained their ability to be inhibited by MLS000544460, indicating that the compounds do not bind near T298/T300/S301/R303/I304. All but one of the remaining Eya2 ED mutants were inhibited by MLS000544460 with similar IC50 values as wild type Eya2 ED. Mutant L423S/T426G/H427T displayed an increase

c Eya2 E418W Eya2 E418A L417W :;< 1.2 () g 1.0 ~1.0 ~ 1.0 ~ 0.8 E ~ go.a 00.8 ~ ;o.6 ;o.6 ~0.6 ~ ~ 0.4 ~ 0.4 ~ 0.4 () () -=;; 0.2

Eya2 L425E :;< 1.4 Eya2 R414A Eya2 R414E ~ 1 . 2 g 1.2 :g-1.0 E 0 81.0 0 1.0 E0.8 E ;::::..0.8 ~ 0 00.8 ~.6 -fo.6 >- ~0. 6 ·5o.4 ~ 0.4 u ~0.4 ;;,R 0.2

Eya2 R414Y Eya2 R414W :g1.0 :g-1.0 0 0 E0.8 E0.8 0 0 ;:::..o,6 ~.6 >- >- ·50.4 ·5oA ti ~ u ~ .:t:0.2

Figure 28. Inhibitors likely bind to an allosteric site (A) The proposed allosteric compound binding site is on the opposite face of the active site. Compound is shown in green ball and stick model and the Mg2+ in the active site is represented by a green sphere. (B) A docking model demonstrating residues in the allosteric binding pocket surrounding the compound (MLS000544460). (C) Effect of mutation of residues in the allosteric binding pocket on the ability of MLS000544460 to inhibit the mutant Eya2 ED.

in the IC50 value, indicating that the compound binding site is affected these amino acid changes. ITC experiments confirm that MLS000544460 has a significantly reduced affinity to the Eya2 ED L423S/T426G/H427T mutant (Fig.

27A) and suggest that the compound may bind near the L423/T426/H427 region.

With assistance from the Computational Chemistry and Biology Core facility in the UC Denver Skaggs School of Pharmacy, I modeled the inhibitor

MLS000544460 into the shallow pocket adjacent to L423, T426, and H427 (Fig.

28A). Energy minimizations of this binding mode indicated that the hydrogen bond interactions involving E418, T426, and T421 at the helix-turn-helix motif between helices 7 and 8 of Eya2 ED may be disrupted upon compound binding to induce a pocket to allow insertion of the pyridine ring of MLS000544460 (Fig.

28B). To test this binding model, I designed multiple Eya2 ED mutants, including

L425E that will likely disrupt the hydrophobic surface and E418W and L417W that would physically block access to the induced pocket. All of these mutants were active and abolished the inhibition of Eya2 ED by MLS000544460 (Fig.

28C). ITC experiments using the Eya2 ED mutants L417W and L425N show no apparent binding (Fig. 27B and 27C), implicating this pocket as the potential allosteric binding site. The Eya2 ED mutant E418W could not be purified at sufficient quantities for ITC experiments. Of interest, the Eya2 ED mutants

R414E and R414A do not diminish the inhibition by MLS000544460, suggesting that the side chain of R414 is likely flexible and not directly involved in inhibitor binding. Substitutions of R414 with bulky amino acids, R414Y and R414W, cause an increase in the IC50 value, likely through steric hindrance.

A B --YFP (Control) 1.50 1.00 ----..- Eya2 1.25 >. ~ . 75 ~ 1.00 :.0 .0 ro 5 0.75 :>0.50 0 50 ~ · --Eya2 0/E 0.25 0.25 --Eya2 + Six1 0/E

0. 00 ...L.-___,=:.~...=::::....:;.::..:.....:...:....:....~:..:..:..:.....;:;.;..::...__.....,--E a2 D274N + Six1 0/E 0.1 1 10 0.1 1 10 Doxorubicin (IJM) Doxorubicin (IJM) c D 1.25 --YFP (Control) 1.50 ----..- Eya2 1.25 1.00 -+-Eya2 D274N -~ ~1.00 }5 0.75 :.0 ro . ~0.75 5 > ~0 . 50 0 ~ 0.50 --YFP (Control) 0.25 0.25 ----..- Eya2 -+-Eya2 D274N 0.00....__--r----"'T"'""------, 0.00 +--r--~-"""T'"-""T"""--r-----, 0.1 1 10 0.005 0.05 0.5 5 50 500 Doxorrubicin (IJM) Etoposide (IJM)

E * *** 1.75 *** 1.50

c~- E 1.25 .S::I. :6 ::; 1.00 Q)2 . ~ ~ 0.75 -0>ro ·- ~ E 0.50 - YFP 0.25 - Eya2 0.00 0D274N

Figure 29. Role of Eya2 in sensitizing various cell lines to DNA-damaging agents and effect of Eya2 inhibitor on cellular migration 79

Figure 29. Role of Eya2 in sensitizing various cell lines to DNA-damaging agents and effect of Eya2 inhibitor on cellular migration (A) Viability of MCF7 cells is not dependent on Eya2 in the presence of DNA- damage induced by Doxorubicin. (B) Viability of MCF10A cells are not dependent on Eya2 in the presence of DNA-damage induced by Doxorubicin. (C) MCF10A T1K cells are not dependent on Eya2 in the presence of DNA-damage induced by Doxorubicin or (D) Etoposide. (E) Compound MLS000544460 inhibits Eya2 phosphatase-dependent migration of MCF10A cells in a transwell migration assay whereas the structurally similar but inactive compound NCGC00241224 has no effect.

Evaluate the effect of inhibitors on cellular phenotypes

The effects of these inhibitors were tested in cell culture to determine whether they inhibit Eya2 PTP-mediated phenotypes in a cellular environment. I first used cell culture models to analyze the effects of Eya2 inhibitors on sensitization to DNA-damage in A673 Ewing sarcoma cells[187]. We chose this cell line because when treated with the DNA-damage inducing drug etoposide,

A673 cells with Eya3 knockdown have reduced viability compared to wild type cells[187]. I treated these A673 cell lines with a standard etoposide concentration of 1 µM and different doses of MLS000544460 (1 µM to 100 µM) and measured viability using a MTS assay. Treatment with increasing concentrations of inhibitor led to dose-dependent decrease of cell viability. However the decrease in viability is identical between cell lines with or without etoposide present, indicating that the effects observed are off-target and the inhibitor is not sensitizing cells to DNA damaging agent etoposide. In retrospect (these assays were performed prior to our specificity data), a reason for this is clear, as these inhibitors do not inhibit

Eya3 (A673 cells express high levels of Eya3 as opposed to Eya2) and therefore an Eya3-dependent effect is not expected.

I next examined additional cell lines to develop Eya2-dependent DNA- damage repair models. MCF7 breast cancer cells express Eya2 and are dependent on Eya2 for many of the cancer phenotypes observed[87]. I tested the effects of the DNA-damage inducing drug Doxorubicin on MCF7 cells expressing

Eya2, MCF7 cells expressing Eya2 and overexpressing Six1, and MCF7 cells expressing Eya2 D274N (phosphatase-dead Eya2 overexpression in the context

of knockdown of endogenous Eya2) and high Six1. All three cell lines respond almost identically upon treatment, indicating that there is no Eya2 phosphatase- dependent sensitivity to Doxorubicin (Fig 29A). Identical results were observed using MCF10A cells. MCF10A breast cancer cells do not express Eya2 endogenously, so I treated cells engineered to overexpress either Eya2 or YFP as a control with Doxorubicin. Eya2 does not increase viability upon DNA- damage (Fig. 29B). This was repeated using the MCF10A T1K variant cell lines overexpressing either Eya2, Eya2 D274N (phosphatase dead), or YFP (control).

Cells were treated with either doxorubicin (Fig. 29C) or etoposide (Fig. 29D) and demonstrated no Eya2 phosphatase dependent sensitization to DNA damaging agents.

David Drasin, a student in our collaborator’s lab examined Eya2-mediated cellular migration using a gap closure assay in which each of three MCF10A cell lines (Eya2, Eya2 phosphatase dead (D274N), and YFP expressing) were treated with vehicle control (DMSO), the inactive compound NCGC00241224, and the active compound MLS000544460 at 10 μM. When overexpressed, Eya2 induced an increase in migration over the YFP control while PTP-dead Eya2 did not, indicating that the PTP activity of Eya2 is important for the migration of

MCF10A cells, which is consistent with previous reports[110] (Fig. 29E, DMSO columns). The active compound MLS000544460 inhibits Eya2 PTP-mediated cell migration, whereas the inactive compound NCGC00241224 does not (Fig. 29E).

Taken together, these data indicate that the Eya2 PTP activity is required for

migration of MCF10A cells and that this migration is inhibited by our Eya2 PTP inhibitor.

Discussion

Phosphatases play important roles in multiple disease processes, including cancer, obesity, and Type II diabetes[188], yet they have typically been difficult to target due to non-specific effects of inhibitors. Indeed, difficulty in obtaining small molecules that inhibit phosphatases in a selective manner is well documented, and is highlighted by the fact that no phosphatase inhibitors have been approved for therapeutic use[125]. Because the active sites of phosphatases are generally highly conserved and positively charged, HTS approaches often lead to compounds lacking specificity or those that are highly negatively charged which limits their bioavailability and cell permeability[189].

Herein, we have identified a specific and allosteric inhibitor of Eya2 PTP activity that fails to inhibit even the highly conserved Eya family member Eya3. Inhibition of the Eya2 PTP activity may open a new avenue for the treatment of breast cancer, as the Eya2 PTP activity has been shown to mediate metastasis of breast cancers[112]. In addition, due to the known role of the Eya PTP activity in

DNA repair[115], these novel inhibitors may sensitize cells to DNA-damage inducing therapeutics.

Eya2 is a member of the unique HAD family of PTPs, that utilize an aspartic acid as the active site residue instead of the cysteine utilized by most cellular phosphatases. Aiming to exploit this mechanistic difference, we carried 83

out a high throughput screen (HTS) of over 330,000 compounds and identified a

series of compounds that demonstrated specificity to Eya2 over other cellular

phosphatases, including other HAD family members. Of particular interest, this

class of compounds specifically inhibited Eya2 and did not inhibit or interact with

Eya3. This result was surprising as the EDs of Eya2 and Eya3 have 68%

sequence identity and 81% similarity, and all the active site residues are identical

or highly conserved. Mutations of Eya2 ED active site amino acids had no effect

on inhibiting the Eya2 PTP activity by the compound, strongly suggesting that the

compound does not bind in the active site. Furthermore, we would expect

inhibition of both Eya2 and Eya3 if the compound binds to the active site due to

high conservation. Although the inhibitors contain weak, intrinsic Mg2+-binding activity, their interaction with Eya2 does not depend on Mg2+ and therefore supports the notion that these inhibitors do not bind to the active site. In fact, evidence supports a binding site on the opposite face from the active site on

Eya2 ED. Therefore, our inhibitor acts through a novel allosteric mechanism that has not been seen with other Eya or HAD inhibitors. The other Eya inhibitors, benzbromarone[110] and metal-chelating compounds[190], use a proposed

Mg2+-coordination mechanism in the active site[190] or bind in a pocket adjacent

to the Mg2+[110]. The HAD phosphatase Scp1 inhibitor Rabeprazole also binds in

a hydrophobic pocket adjacent to the Mg2+ in the Scp1 active site[110, 191].

As a member of the HAD family of enzymes, Eya contains a Rossmann-

like catalytic core common to all HAD family enzymes (Fig. 15A, Orange) and a

unique cap structure consisting of a helix-bundle-motif (HBM) only found in Eya

Figure 30. Possible mechanisms of allosteric inhibition by MLS000544460. (A) Eya2 is a HAD family member consisting of a catalytic core (orange) and a helix-bundle-motif (HBM) cap (cyan). The allosteric compound binding site is located at the interface between these two motifs (α7-α8 loop) (compound is shown in green ball and stick model). (B) Surface representation of the catalytic core (orange) and HBM (cyan) showing MLS000544460 at the interface. (C) A ribbon diagram demonstrating that the compound (green) binds directly behind the catalytic residues centered near the Mg2+ ion (green sphere). The compound binding at the allosteric site can potentially induces rigidity in the otherwise flexible loops where catalytic and Mg2+-coordinating residues D274, D276, and D502 are positioned. 85

(Fig. 30A, Cyan)[95]. A structural homology search to other HAD enzymes using the DALI server[192] shows that the catalytic core contains a highly conserved structure (Z = 9.7 – 12.3) whereas the HBM cap shows no structural homology with any other protein structures (Z = < 4.0)[102]. Indeed, a structural homology search of the Eya2 region consisting of the proposed binding site (helix 7, helix 8, and residues forming the base of the pocket (L463, G462, Y461)) displays no homology (Z = < 3.5) with any other protein structure. Of interest, the proposed allosteric inhibitor binding site is located at the interface between the catalytic core and the HBM cap (Fig. 30A, B). Helix 7 is part of the HBM cap whereas helix 8 is part of the catalytic core. Modeling inhibitors to this allosteric site indicates R414 may be involved in binding; however, R414E and R414A mutations have no effect on inhibition, suggesting that R414 is either flexible to accommodate inhibitor binding or is not involved in the binding mechanism.

Mutations to bulky amino acids (R414W and R414Y) decrease inhibition suggesting that while not involved in a binding mechanism, R414 may remain sterically close. Compound binding may lead to an induced fit in which the HBM adopts a conformation that moves helix 7 in the HBM (containing R414) sufficiently away from helix 8 in the catalytic core region to accommodate the inhibitors. Since the docking studies were carried out using the solved Eya2 ED

X-ray crystal structure, our inhibitors may also bind Eya2 in an Eya2 conformation not present in the crystal structure. Furthermore, this proposed binding surface is not accessible in the native Eya2 ED structures[102], which may explain why we cannot obtain inhibitor-bound Eya2 ED crystals in soaking

experiments. In the crystal structure of the Six1-Eya2 complex[40], the proposed allosteric binding site is solvent accessible; however, these crystals disintegrate upon soaking with inhibitors, which could be due to conformational changes of

Eya2 ED upon compound binding. Efforts in the lab are under way to generate new crystal forms of Eya2 that may assist in identifying multiple conformations of

Eya2 or generate appropriate unit cell organization to provide access to this binding site.

Binding at the interface between the catalytic core and the cap suggests a possible mechanism of inhibition by these compounds. Some HAD members

(phosphoserine phosphatase, phosphomannomutase, pyridoxal 5’-phosphatase) have been observed to utilize their cap motif for active site solvent occlusion in their catalytic mechanism[95]. The cap of HAD members is able to move from an

‘open’ conformation where the active site is exposed to a ‘closed’ conformation in which active site access is restricted after substrate binding[95]. Since a portion of the active site of Eya2 (on the opposite side of the allosteric compound binding site) is flanked by helices 1 and 2 of the HDM and that most HAD caps are flexible domains that require movement for catalysis[95], it is possible that these allosteric Eya2 inhibitors lock Eya2 into an ‘open’ conformation in which active site solvent occlusion cannot occur when substrate binds, consequently inhibiting

Eya2 activity. This model, in which the substrate can still bind but the HBM cap cannot close, is consistent with our observation that these compounds are mixed-mode inhibitors[120]. If the inhibitor functions with this mechanism, any changes in secondary structure must be limited since the CD binding data

indicate that minimal secondary structural changes occur upon inhibitor binding.

A similar mechanism has been identified for LTV-1, a mixed-mode inhibitor that targets the lymphoid tyrosine phosphatase Lyp, a canonical thiol-based, non-

HAD phosphatase involved in T cell antigen receptor signaling. LTV-1 selectively prevents the capping of the unstructured “WPD” loop and the inhibitor can only be modeled into an open conformation of Lyp[193]. This was only possible since crystal structures of Lyp were solved in both open and closed conformations. In the absence of structural information, analytical ultracentrifugation has been used in mechanistic studies to observe isomerization induced by the allosteric inhibitor phenylalanine when added to pyruvate kinase[194]. A similar method may be useful for Eya2 and these inhibitors if large movements occur without changing the secondary structure significantly.

Another possible allosteric mechanism may be that these inhibitors might disrupt the active site pocket geometry to prevent catalysis without disrupting the overall protein structure. Analysis of the crystal structure of Eya2 conjugated to

BeF3 that mimic the enzyme transition state demonstrates that the α4-α5 loop,

β2-α9 loop, and β3-α10 loop moves laterally compared to the native structure, suggesting intrinsic mobility may be required for catalysis[102]. Our inhibitor binds and inserts its pyridine ring into a pocket directly opposite the active site

(Fig. 30C) close to the loops that contain residues required for catalysis and coordination of the Mg2+ (D274, D276, and D502). It is possible that the positioning of the pyridine ring could either induce rigidity in the catalytic loops or interfere with the geometry of the Mg2+ ion to prevent catalysis from initiating, yet

still allow for substrate binding. This may explain why a mixed mode mechanism is observed.

The chemical series we identified specifically inhibit Eya2 but not Eya3.

However, we do not yet know if these compounds inhibit Eya1 or Eya4 since I have not been able to express and purify soluble human Eya1 or Eya4 protein.

Many of the amino acids in the allosteric binding site involved in binding

(according to mutagenesis studies) are conserved among all Eyas (R414, L417,

E418, L425) except for L423, T426, and H427. In Eya1 and Eya4, the T426 residue is conserved but not the L423 and H427 residues. It is at this point difficult to predict whether this class of compound will inhibit Eya1 and Eya4 and this further experimental evidence with purified Eya1 and 4 is likely required to address this specificity question.

Published data suggests that Eya2 is a dominant player in breast and ovarian cancers[67, 87], as this Eya correlates with poor prognosis in both these cancers, and is amplified in 14.8 percent of ovarian cancers[82]. Eya1 also correlates with adverse outcomes in breast cancer[87] and examination of the

TCGA breast cancer dataset[195] shows that Eya1 and Eya2 are amplified in

10.8% of breast cancers. Inhibiting Eya2 PTP activity is thus likely to have a significant therapeutic effect in these cancers. On the other hand, Eya4 may have both oncogenic[196, 197] and tumor suppressive roles[84], thereby limiting the therapeutic potential. In addition, recent data also indicate that some Eya family members may function in adult tissues. For example, the phosphatase activity of Eya1 regulates cell polarity in the lung epithelium[113]. Eya3 is

expressed in a tissue specific manner, with adult expression in the brain, skeletal

muscle, heart, kidneys, and weak expression in the lungs, but Eya3 knockout in

mice results in only mild phenotypic differences[198]. Thus, having an inhibitor

that is highly selective toward a specific Eya family member (i.e., Eya2) may be

advantageous in some situations, while having a pan-Eya inhibitor may be

needed in others in which multiple Eya members are involved.

In summary, I have investigated the mechanism of action of this inhibitor

series using several methods. I determined that these inhibitors likely function at

an allosteric site on the opposite face than the active site and do not require Mg2+

for a binding mechanism. Attempts to determine the inhibitor-bound structure

was unsuccessful yet in silico modeling was used to generate a potential binding

site based upon surface differences between Eya2 and Eya3. This hypothesized

model was tested through mutational analysis and key amino acids were

confirmed using ITC. Binding at an allosteric site and knowing the structural and

mechanistic properties of HAD phosphatases leads to a new model of how this

inhibitor series may function as allosteric inhibitors.

Methods

Protein expression and purification

Eya2 was sub-cloned on a pGEX-6p1 vector (GE Healthcare) and the protein was expressed as a GST-fusion protein from E. coli. The protein was first purified on a glutathione column. GST was then cleaved off using PreScission

protease and eluted Eya2 ED was further purified on a Superdex 200 (S200)

column (GE Healthcare).

Eya2 phosphatase assays

Eya activity was measured in 50 µL reactions using black, 96-well, half-

volume microtiter plates (Greiner Bio-one) with the substrate OMFP (3-O-

methylfluorescein phosphate, Sigma-Aldrich). OMFP is converted to a

fluorescent product OMF upon dephosphorylation. Enzyme and substrate

concentrations were optimized to consume less than 15% of substrate after one

hour, have a linear response during the assay, and have substrate concentration

below the Km. Final assay conditions are 50 mM MES, pH 6.5, 50 mM NaCl, 1.25

mM MgCl2, 0.05% BSA, 1 mM DTT, and the reaction contained 50 nM Eya2 ED

and 100 µM OMFP. Reactions were started by the addition of OMFP, preceded

for 1 hour at room temperature followed by terminating the reaction with the

addition of 75 mM EDTA. Fluorescence intensity was measured at 485/515 nm

excitation/emission on a Fluoromax-3 plate reader (Horiba Jobin Yvon).

Eya2 kinetic experiments

Due to the solubility limitations and relatively high Km values of the OMFP substrate, we developed a second fluorogenic Eya2 phosphatase assay using a different substrate, fluorescein diphosphate (FDP, Anaspec), to measure compounds’ effect on enzyme kinetics. All kinetic experiments were performed in

384 black medium binding plates (Greiner Bio-one). Using the FDP substrate we

were able to achieve an FDP stock solution of 100 mM in Eya2 ED buffer, therefore allowing us to test Eya2 kinetic experiments at the appropriate substrate versus compound concentrations to calculate an accurate km value and to generate a Lineweaver-Burke plot for Eya2 ED. Briefly, 2.5 µl of compound

(MLS000544460) at 100, 40, 20, 10, 5 or 0 µM was incubated with 2.5 µl of 2 µM

Eya2 ED for 10 minutes. Subsequently, 5 µl of FDP substrate at 20, 12, 7.2, 4.3,

2.6, 1.6, 1, or 0 mM was added and plates were incubated for an additional 30 minutes before measured on the Viewlux plate reader with excitation of 485 nm and emission of 525 nm. Kinetic analysis was performed using GraphPad Prism

4 (GraphPad Software, Inc.).

Isothermal titration calorimetry

Eya proteins were purified via S200 in ITC buffer (50 mM NaPO4 buffer, pH 6.5, 50 mM NaCl, 5% glycerol) followed by concentrating to appropriate concentrations. Inhibitor solutions were prepared by dilution from DMSO stock into ITC buffer. DMSO was added to protein solution to match concentration

(0.2%) and all solutions were degassed. ITC experiments were performed with a

VP-ITC microcalorimeter (MicroCal, Inc., GE Healthcare) at 25°C. Due to low solubility of inhibitors in aqueous buffer, the syringe contained Eya, the sample cell contained inhibitor, and the reference cell was filled with ITC buffer. After a 2

µL initial injection, 10 µL aliquots of 250 µM protein was added stepwise at 5 minute intervals into the sample cell containing 20 µM inhibitor until saturation.

Origin 7.0 was used to analyze data with embedded calorimetric fitting programs

to generate the binding curves.

Circular dichroism

Samples containing 0.33 mg/mL Eya2 were placed in a cuvette with 1 mm

path length. CD spectra were measured using a Jasco-810 instrument (Jasco,

Inc.) with reported spectra being the average of six scans at a temperature of

25°C. CD buffer was 5 mM phosphate pH 6.5, 25 mM NaCl, and 0.5% glycerol.

For samples containing inhibitor, 0.1 µL of 100 mM inhibitor in 100% DMSO was

added to sample to obtain 50 µM final inhibitor concentration. Final DMSO

concentration was 0.05% in the assay. Samples containing inhibitor were

compared to vehicle control (DMSO).

NMR spectroscopy

15N-enriched Eya2 was prepared under similar purification conditions as above. Optimal NMR buffer conditions were determined to be 50 mM Bicine pH

7.5, 50 mM NaCl, 0.5 % glycerol. Maximum Eya concentration used was 150 µM due to aggregation at higher concentrations, regardless of buffer. HSQC experiments were collected at 25°C on a Varian 900 Mhz at a concentration of

150 µM Eya concentration. Compound was added to saturating concentrations while keeping DMSO concentration below 0.1 % DMSO.

UV-Vis spectra analysis of selected hydrazides

Compounds were dissolved in acetonitrile at a final concentration of 25 µM

with final Mg2+ concentration at 0, 0.5, 1, 5, 10, 25, 50, 100 and 200 mM. UV- visible spectra were obtained using an Agilent 8453 UV-Visible

Spectrophotometer (Agilent Technologies) in 0.1 cm quartz cuvette at room temperature.

Molecular docking

The AutoDock 4.0 program was used for docking the compounds to the active site of the crystal structure of ED of Eya2 (pdb: 3HB1). The active site of the protein was defined by a grid of 70 x 70 x 70 points with a grid spacing of

0.375 Å centered at Mg2+ ion. The Lamarckian Genetic Algorithm (LGA) was

applied with 100 runs and the maximum number of energy evaluations was set to

2 x 106. Results differing by less than 1.5 Å in positional root-mean-square deviation (RMSD) of substrate were clustered and the final binding conformations were represented by the one with the most favorable free energy of binding. The optimal binding complexes were subjected a stepwise energy minimization and

MD simulations.

Docking at the allosteric site was performed with Accelrys Discovery

Studio 3.1 (Accelrys Software, http://accelrys.com). The crystal structure of Eya2

(PDB: 3GEB) was obtained from the protein data bank (http://www.pdb.org).

Predicted binding-induced changes in protein structure were determined by a

gradient minimization protocol using a CHARMM force field and the Generalized

Born implicit solvent model[199].

Motility assay

Motility was measured by gap closure assay, where a silicone u-well insert

(Ibidi, Verona, WI) in a 24-well plate was used to create an approximately 500 um gap between 40,000 cells/chamber that were plated overnight. Photos were taken of the gap immediately after removing the insert and adding media, and then again 4-8 hours later on a CKX41 microscope (Olympus, Tokyo, Japan).

Distance migrated was determined by subtracting the size of the gap at the end time point from the size of the gap at the initial time point, using DP2-BSW software (Version 2.2; Olympus).

Cell lines

Stable expression of human Eya2 or phosphatase dead Eya2 (D274N) in

MCF10A cells was achieved through retroviral transduction. Eya2 or D274N was cloned into pMSCV-IRES-YFP backbone, and BOSC cells were used to package viral particles. YFP-positive cells were sorted 1 week after infection.

CHAPTER IV

DEVELOPMENT OF ASSAYS TARGETING OTHER ASPECTS OF SIX1

ACTION

Abstract

The transcription factor Six1 is critical for numerous aspects of cancer

development and metastasis and as such is an ideal therapeutic target. Six1 and

Eya function as a bipartite transcription factor where Six1 provides the DNA-

binding ability, Eya2 provides the transactivation function, and the protein-protein

interaction between the two is required for function. Furthermore, Eya2 has

phosphatase activity that is required for transformation and metastasis of breast

cancer cells. Each of these functions provides an exciting target for developing

anti-cancer therapeutics. Chapters I and II outline the identification and

characterization of Eya2 phosphatase inhibitors, while this chapter details the

early stage development and optimization of assays targeting the Six1-DNA and

Six1-Eya2 interactions. To this end, I developed a fluorescent polarization assay

and an enzyme-linked immunosorbent assay (ELISA) for adaptation to HTS as a

strategy to obtain inhibitors. I was able to demonstrate Six1 binds DNA using a

5’-fluorescein-labeled DNA sequence and that recombinant Six1 binds Eya2 in the ELISA. Although the lack of specificity is an issue in both assays, further troubleshooting and fine tuning of conditions may enable these assays to be used for screening inhibitors targeting all aspects of Six1 action.

Introduction

The Six family of homeobox genes are members of the homeobox gene superfamily which has often been implicated in cancer[200-202]. During normal development, homeobox genes play a role in the proliferation of progenitor cells and stimulate migratory and invasive properties of cells within the embryo. As developmental genes, they are usually downregulated upon proper tissue and organ development. However, in neoplastic disease many of these genes are mutated, or overexpressed[203] as evident with the misexpression of HOX genes in colorectal, brain, lung, renal, and breast cancers[201, 204, 205] and the PAX genes with misexpression in rhabdomyosarcoma, Wilms’ tumors, glioblastoma multiforme, and thyroid and breast carcinomas[201, 202, 204, 206]. Likewise,

Six1 is upregulated in 50% of primary breast cancers and 90% of metastatic lesions[25, 26, 77] and this overexpression induces upregulation of the TGF-β signaling pathway to promote EMT leading to metastasis[34, 79]. Overexpression of Six1 is able to transform immortalized, but otherwise normal, mammary epithelial cells, forming highly aggressive tumors when injected orthotopically into nude mice and display oncogenic EMT and stem cell phenotypes[75]. A decrease in tumor size and metastasis is observed in animal models of breast cancer[81] and rhabdomyosarcoma[71] when Six1 is knocked down.

Six1 transcriptional control is mediated through recruitment of co-factors, including Eya, since it lacks any intrinsic activation or repression domains. The four Eyas (Eya1-4) contain a divergent N-terminus (containing a S/T phosphatase domain of unrelated function[107]), an internal proline-serine-

threonine rich activation domain, and a highly conserved C-terminal Eya Domain

(ED) containing protein tyrosine phosphatase (PTP) activity[7]. Six1 binds the

Eya ED to function as a bipartite transcription factor with the DNA-binding function provided by Six1 and the transactivation function provided by Eya. This

Eya mediated transcriptional activation plays a major role in both normal development processes[54, 82] and in a variety of diseases[59, 60, 83]. Over- expression of both Six1 and Eya1 or Six1 and Eya2 in breast cancer is correlated with a shortened time to relapse, shortened time to metastasis, and decreased survival[87]. Disruption of the Six1-Eya2 interaction leads to a loss of pro-EMT characteristics in cell lines and a decrease in metastasis in mouse models[40].

These data implicate the Six1 transcription complex as an ideal therapeutic target and that multiple approaches can be used to target the complex, including the disruption of both the Six1-Eya interface and the Six1-DNA binding interaction. The recently published Six1-Eya2 co-crystal structure[40] provides invaluable molecular details about the Six1-Eya2 interaction and provides insights into the DNA-binding mechanism. Herein I discuss the efforts to target the Six1-DNA interaction and the Six1-Eya2 protein-protein interaction.

Results

Development and optimization of fluorescence anisotropy assay for Six1-

DNA HTS

To identify inhibitors of the Six1-DNA interaction using HTS, I decided to use a fluorescence polarization assay with a 5’fluorescein-labeled, double-strand

Figure 31. Six1 binds DNA in FP assays but is not specific (A) Six1 binds MEF3 promoter DNA with good signal. (B) Binding curve of Six1 to MEF3 promoter sequence. (C) Six1 but not the controls Eya2 or GST bind DNA. (D) Six1 non-specifically binds an SSrbC DNA sequence equally well as MEF3 promoter sequence.

DNA (*DNA). When excited by polarized light at 485 nm, the fluorophore will re- emit polarized light at 515 nm. Free *DNA in solution will tumble rapidly and emitted light has low polarization. When bound by Six1, the complex tumbles slower resulting in higher polarization of emitted light. Reduction in polarization will identify compounds that disrupt the Six1-DNA interaction.

This assay was designed using recombinant Six1 1-259 and *DNA containing a myoMEF3 sequence GGGCTCAGGTTTCTGT that Six1 binds with

specificity (the core binding sequence is underlined)[46]. Large amounts of

purified Six1 were obtained by expressing Six1 as a GST-fusion protein in E.

coli[46]. Addition of Six1 to *DNA increases polarization from 70 mA for free

*DNA to over 300 mA for Six1-bound *DNA (Fig. 31A). I therefore attempted to

develop this assay for HTS with the first task to determine the affinity of the

interaction. I set up a binding curve and determined the interaction has a KD =

287 ± 112 nM (Fig. 31B), which is consistent with prior published results using

electrophoretic mobility shift assays (EMSA)[46]. I therefore chose 300 nM Six1

for all subsequent assays in the HTS. As negative controls, GST and Eya2 do

not bind DNA (Fig. 31C).

I next assayed the binding specificity of Six1 using a fluorescein-labeled,

unrelated DNA sequence, ACACCATCGATTAATCTTCTGATGAAAC that Six1

should not bind. It is the recognition sequence for the Salmonella SSrbC protein.

Unfortunately, Six1 bound both DNA sequences and demonstrated no DNA- binding specificity (Fig. 31D). Extensive effort was made on resolving the binding specificity. Multiple buffer conditions were tested (HEPES pH 8.0, Tris pH 8.0,

Tris pH 8.0, Phosphate pH 7.5, and MES pH 6.5) in various ionic strengths (20-

500 mM NaCl or KCl), but did not resolve the specificity issues (data not shown).

I further screened additives such as metal ions (CaCl2, CdCH3CO2, CoCl2,

CrCO3, CsSO4, FeCl2, LiSO4, MgCl2, MnCl2, NiCl2, ZnCH3CO2, ZnCl2, and

ZnSO4) to stabilize Six1 and increase polarization signal, polyethyleneimine (PEI)

100

to neutralize charged surfaces, detergents such as Triton X-100 to neutralize hydrophobic surfaces, glycerol, bovine serum albumin (BSA), and polyIC, all of which has shown to promote specificity in various capacities. Every additive used did not resolve the specificity issues of the Six1-DNA interaction (data not shown). Because of this, the Six1-DNA polarization project was put on hold in favor of other methods to target the Six1 transcriptional complex.

Development of competitive ELISA for identifying inhibitors of the Six1-

Eya2 interaction

Concurrently, I began development of an ELISA-based assay for the Six1-

Eya2 protein-protein interaction. Eya2 ED was coated on the 96-well ELISA plate followed by a blocking step using BSA. Six1 was added, followed by a rabbit primary (1°) antibody recognizing human Six1. A horseradish peroxidase (HRP)- conjugated, anti-rabbit, goat secondary (2°) antibody was added. Wash steps were between all additions to remove any unbound sample. Addition of the HRP substrate ABTS produces a Six1-concentration-dependent color change observable at 420 nm.

I optimized the antibody additions and found that optimal dilutions to produce high signal-to-background were 1:1000 and 1:10,000 for 1° and 2°, respectively (Fig. 32A). I found that Six1 has high levels of non-specific binding as it binds to the BSA-blocked plate even when Eya is not coated, and I subsequently spent considerable effort to minimize this through optimizing buffer conditions. I screened a number of conditions with or without Eya2 bound to the 101

Figure 32. Six1 binds Eya via ELISA. (A) Primary and Secondary antibody optimization. (B) pH dependence on Six1 binding. (C) Conditions that alleviate a portion of Six1 nonspecific binding (high salt + 1% Triton X-100). (D) Binding curve of the Six1-Eya2 interaction. Non- specific binding data was subtracted. 102

plates as a way to assess non-specific binding of Six1. I performed a pH screen

to determine if there was a pH-dependency to the interaction. Six1 binds at all pH

values without resolving non-specific binding (Fig. 32B). I varied the ionic

strength of the buffer and screened select additives: Triton X-100 and PEI. These

were chosen to mitigate hydrophobic surfaces (Triton X-100) or to neutralize charged surfaces (PEI). Triton X-100 in combination with high ionic strength (500 mM NaCl) resolved a moderate amount of non-specific binding of Six1 (Fig.

32C). With acceptable conditions established, I generated a binding curve to investigate the affinity of the Six1-Eya2 interaction. Increasing concentrations of

Six1 were added to a constant amount of Eya2. After subtracting non-specific binding data, I determined that the Six1-Eya2 interaction has a KD = 140.9 ± 24.7

nM (Fig. 32D). Although able to calculate the affinity, the signal to background

(S/B) was under 5-fold and is not suitable for HTS. At this stage with other

projects progressing more favorably, the protein-protein HTS project was

transferred to a fellow graduate student in the lab, Melanie Blevins, who adapted

the Alphascreen®-based protein-protein interaction with success and further

optimized the ELISA for use as a secondary confirmation screen.

Discussion

The processes of cancer and normal development share many properties

including changes in cell proliferation, cell differentiation, neovascularization, cell

motility, cell death, and invasion into surrounding tissue[1]. These processes are

often controlled by ‘master regulators,’ transcription factors that function in 103

determining cell fate and identity[201]. Therapeutics that target master regulators have the potential to disrupt entire developmental programs initiated out of context in cancer, as is the case with the Six1 transcriptional complex. In fact, experimentally lowering Six1 levels in several different cancer models significantly lower cancer cell proliferation and metastasis[71, 81, 168].

Developing therapeutics targeting the Six1 transcription complex would be of great benefit since standard chemotherapeutic agents that target dividing cells often result in severe or lethal side effects. Because Six1 and Eya are regulators of embryonic development, inhibition of Six1 activity is proposed to have few side effects in adults where expression of each is low while preventing tumor initiation, progression, and metastasis. Although targeting transcription factors is traditionally difficult[2], we aim to overcome this by targeting multiple facets of

Six1’s biochemical mechanisms: the Six1-DNA binding, the Six1-Eya2 protein- protein interaction, and the Eya2-phosphatase activity.

Protein-DNA interactions are traditionally difficult to target for drug design.

There have been recent successes using HTS strategies in the identification of inhibitors of B-ZIP transcription factors binding to DNA[207], of a sub-micromolar inhibitor of Pot1 to telomeric DNA[208], and of inhibitors that bind specific DNA sequences to inhibit the hypoxia-inducible factor from binding its target, the hypoxia response element[209, 210]. These studies provide a blueprint to inhibit

DNA-binding of transcription factors.

Despite extensive troubleshooting of assay conditions, Six1 did not demonstrate specific DNA-binding activity. After this project was put on hold, a

104

broad study using protein-binding microarrays on sequence specificity of homeodomain-containing proteins[211] identified a novel Six1 consensus sequence, GGTATCA, that was determined to have a low nM KD[42]. In addition, a recently published study on optimizing the Six1 consensus sequence identified several higher affinity and previously unknown DNA sequences based on the

MEF3 DNA sequence[42]. In addition, Six1 may bind DNA using an atypical mechanism involving both the HD and SD. This opens the possibility that a longer DNA sequence involving regions outside of the core sequence may be necessary to confer full selectivity. A higher affinity DNA sequence for FP assays and using a longer sequence may resolve specificity issues and may warrant revisiting targeting the Six1-DNA interaction with HTS in the future.

It is emerging that protein-protein interactions, including Six1-Eya2, play critical roles in transcriptional activation, cell growth, DNA replication, translation, and transmembrane signal transduction[212]. As such, protein-protein interactions have emerged as valuable therapeutic targets[213-215] for novel therapeutics in cancer [52, 216, 217]. HTS is a preferred approach to discover selective hit compounds targeting protein-protein interactions[218-221] and was necessary for the Six1-Eya2 interaction as no structural information was known about the protein-protein interface when the project was initiated. To this end, I began development of an ELISA-based HTS assay. However, despite extensive troubleshooting of assay conditions, the non-specific binding of Six1 was not effectively resolved to levels suitable for HTS (S/B < 5). After transferring the project to Melanie Blevins, she successfully developed an Alphascreen®-based

105

assay. This solution-based assay does not require plating of components to a

surface and suggests that Six1 suffers from adsorption to the well in the ELISA

assay. Indeed, switching to a medium-binding ELISA plate from a high-binding plate resolved much of the observed non-specific Six1 binding. High binding plates have a net negative charge that attracts the highly positively charged Six1.

This interaction likely promotes the non-specific binding of Six1 when Eya2 is absent. Medium binding plates are hydrophobic and any binding is driven by hydrophobic interactions. This would limit non-specific binding to the plate surface since Six1 is highly charged.

After I had halted progress this HTS project, the Six1-Eya2 co-crystal was published[40] and provides several insights into the binding mechanism and suggests ways to successfully target the complex without a HTS approach. A single amphipathic α-helix binds to a hydrophobic cleft on the Eya2 surface.

Mutations of single amino acids within this helix disrupt Eya2 binding and cause inhibition of TGF-β signaling, EMT, and metastasis. Stable, synthetic α-helical peptides have been successfully used to interrupt with high potency the interactions between the HDM2-p53[222] and Bcl2-Bax[223] proteins to promote apoptotic activity in cancer models. A rationally designed peptide inhibitor is plausible for Six1-Eya2 since the binding mechanisms of HDM2-p53 and Bcl2-

Bax are highly similar in their use of a single α-helix for binding. Synthetic peptides are inherently disordered and extensive effort is needed to stabilize their helical conformations through a number of possible methods[224-226].

106

Another possible approach would be to use virtual screening to identify small molecules targeting protein ‘hot spots,’ small subsets of amino acids in the binding site where most of the binding energy is localized[227]. The binding

surface of Eya2 consists mainly of hydrophobic interactions with additional

contributions by hydrogen bonds and salt bridges[40]. This indicates that small

molecule inhibitors targeting hot spots are feasible since binding clefts that

recognize single α-helices are more amenable to small molecule inhibitors due to

a higher concentration of hot spots[228] and that a single turn of the α-helix

approximates the area bound by typical drug molecules[229]. Small molecule

inhibitors of the HDM2-p53 are mimics of residues that span one helical turn and

bind hotspots to prevent association of the complex. If targeted similarly, small

molecules would need to bind to Eya2 to prevent Six1 binding and subsequent

activation of the Six1 transcription complex.

New information about the Six1-Eya2 interaction and lessons learned from

successful endeavors at inhibition protein-protein protein-DNA interactions will be

crucial in the development of future Six1 therapeutics. The Six1 transcription

complex is an ideal therapeutic target for developing anti-cancer drugs due to its

roles in proliferation, invasion, and metastasis. Moreover, after normal

development is complete, expression of Six1 is turned off and is low or

undetectable in most tissues, but re-expressed locally in multiple stages of

cancer development. Targeting the Six1 activation complex will allow a tumor-

specific therapeutic with potentially limited side effects.

107

Methods

Fluorescence polarization assay

Fluorescence polarization measurements were made in 0.5 x 0.5 mM

quartz cuvette (Starna Scientific). Six1 (300 nM) was added to 5 nM fluorescein-

labeled DNA in 400 µL total volume. Following one hour incubation on ice, the

fluorophore was excited at 485 nm and polarization was recorded between 500-

550 nm using Fluoromax-3 spectrometer (Joriba Jobin Yvon).

Optimized enzyme-linked immunosorbent assay

Each well of a 96-well, high-binding assay plate (Costar 9018) was coated

with 0.5 µg/mL (15.5 nM) Eya2 ED in 50 mM bicarbonate buffer (pH 9.6) for one

hour at room temp. Plates were manually washed three times with TBST (10 mM

Tris pH 8.0, 150 mM NaCl, 1% Triton X-100) and BSA in TBST was used block

plates. 300 nM Six1 was added in TBST for one hour at room temperature,

followed by washing three times with TBST. Primary antibody (rabbit anti-Six1)

was added at a dilution of 1:1000 in TBST. After one hour incubation, plates were

washed three times with TBST. Secondary antibody (goat anti-rabbit, HRP-

conjugated) was added at 1:10,000 dilution in TBST and incubated for one hour

at room temperature. After washing three times with TBST, the HRP substrate

ABTS was added in citrate buffer (pH 5.0). After 30 minute incubation at room

temperature, the plates were read at 420 nM using a SpectraMax 384

spectrometer.

108

CHAPTER V

DISCUSSION AND FUTURE DIRECTIONS

Targeting the Six1 transcriptional complex

Transcription factors are appealing therapeutic targets in cancers and many other diseases. They are integral components of signaling pathways and, as oncogenes, can activate transcription targets, leading to unregulated cell growth through hijacking of pro-growth signaling pathways[230, 231]. In addition, numerous signaling pathways containing oncogenic proteins may converge to a single transcription factor. Therefore, inhibition of a single target, such as a transcription factor, would have the therapeutic benefit of indirectly targeting multiple downstream oncogenes.

The Six family of transcription factors is one such validated therapeutic target. Expression of Six1 outside of a developmental context is causative in many cancers including brain, lung, ovarian, hepatocellular, and breast, in which

Six1 is the most studied[25, 67-71]. Experimentally lowering Six1 levels significantly decreases cancer cell proliferation [26] and metastasis[15] in different cancer models. As such, therapeutics targeting the Six1 transcription complex could have a great impact on treating breast and other cancers in which it is overexpressed.

The Six1 transcription complex offers a number of opportunities to target therapeutically. As a bipartite transcription complex, Six1 provides the DNA- binding ability and its co-activator Eya provides the activation function. As a

109

complex, it can be targeted by disrupting the DNA-binding function of Six1 to prevent binding to target promoters and inhibiting the activation function provided by Eya through abolishing the protein-protein interaction. The fact that Eya has protein tyrosine phosphatase (PTP) activity adds additional complexity to transcriptional control and offers additional opportunities to inhibit the Six1 transcriptional complex. The PTP activity is required for the activation of a subset of Six1 target genes in drosophila[169], can switch the Six1 transcriptional

complex from a repressive to activating complex[54], is critical for the transformation, migration, invasion, and metastasis in breast cancer cells[112], and mediates cell survival decisions upon dsDNA damage[115]. I therefore attempted to target these three interactions to develop high throughput screening

(HTS) assays and will discuss them below.

Six1-Eya2 protein-protein interaction

To identify inhibitors of the Six1-Eya2 protein-protein interaction, I developed an enzyme linked immunosorbent assay (ELISA) assay to analyze the

Six1-Eya2 interaction. Unfortunately, the non-specific binding of Six1 could not be eliminated to improve the signal-to-background (S/B) to levels acceptable for

HTS, despite extensive optimization and troubleshooting. An alternative approach to HTS was needed and after transferring the project, Melanie Blevins successfully developed an Amplified Luminescence Proximity Homogenous

Assay (Alphascreen®)-based assay for HTS that has robust signal and high S/B.

This assay eliminates key issues present in the ELISA assay: it has high signal, 110

no wash steps, fewer component additions, is solution based, and demonstrates

no non-specific binding of Six1. After confirming suitability to HTS, a screen of

>330,000 compounds was performed with the results currently being analyzed.

An alternative method for inhibitor identification is possible using the

recently published Six1-Eya2 crystal structure from our labs[40] to allow a

rational, structure-based approach using computational methods to virtually screen large compound libraries for binding. This allows the rapid screening of compounds without the use of physical resources such as reagents and

recombinant proteins. However, as this method is virtual-based, all compounds will still need to be validated in the Alphascreen® assay to confirm the binding.

The binding mechanism of Six1 to Eya2 is highly similar to two successfully targeted protein-protein interactions in which helical mimetics

(peptide and small molecule) have been rationally designed that inhibit the interactions: HDM2-p53[222] and BH3-Bax[223]. A stapled p53 peptide that binds HDM2 with low nanomolar affinity causes an increase in pro-apoptotic caspase-3 activity in several HDM2-overexpressing cancer cell lines[222]. A stapled BH3 peptide binds directly to Bax, is cell permeable, interacts with endogenous Bax in cell culture to trigger a pro-apoptotic response[223]. A similar method may be used to generate a stable Six1 helical peptide (or mimetic) that can bind Eya2 and inhibit Six1 function. This remains a valid strategy if HTS approaches are unsuccessful.

Any identified compound would need to be characterized for its ability to transverse cell membranes and to disrupt the Six1-Eya2 interaction in a cellular

111

environment. One method to directly assess the function of identified compounds would be to develop a bimolecular fluorescence complementation (BiFC) assay[232-234]. This method requires dissecting a functional protein (such as

GFP[235] or Luciferase[234]) into inactive fragments that can be attached to protein binding partners (e.g. Six1 and Eya2). Binding of the partner proteins induces the assembly of the fragments into a functioning protein that can be visualized in cell-based systems. Such systems have been used identify and characterize inhibitors of the p53-HDM2 interaction in the cell [236]. In addition, established systems can be used to monitor effect of inhibitors in the cell[40] including monitoring TGFβ activation using phospho-SMAD3 levels or 3TP- luciferase signaling, monitoring changes in epithelial and mesenchymal markers, or measuring effects on Six1-induced metastasis in a mouse model[40].

Six1-DNA interaction

I used a fluorescence polarization assay with recombinant Six1 1-259 and the MEF3 promoter DNA sequence that Six1 binds[46]. I demonstrated that Six1 binds with a KD = 287 ± 112 nM. However, Six1 also binds non-specifically to any sequence presented. Despite extensive troubleshooting, this could not be resolved and the project was eventually put on hold.

Even so, the Six1-DNA interaction remains an attractive target that may be worth pursuing. Although the Six1-Eya2 structure is solved, it unfortunately cannot be used for a rational, structure-based approach to identify Six1-DNA inhibitors. Key regions of the HD involved in DNA-binding are flexible and are 112

lacking in the crystal structure. Furthermore, the Six1 DNA-binding mechanism

likely involves regions outside of the HD and these unique interactions are entirely unknown without a DNA molecule present in the crystal structure.

Therefore, a DNA-bound Six1 structure is necessary to identify key molecular

interactions before virtual screening methods can be used to identify inhibitors.

Regardless of the method used to identify inhibitors, the non-specific

binding of Six1 to DNA needs to be resolved. Recent studies on homeobox DNA- binding specificity[211] and Six1 homeodomain binding specificity[42] may offer

ways to resolve the non-specific binding of Six1. These studies identified novel,

higher affinity sequences that Six1 binds. These new Six core sequences have

higher affinities to Six1: KD = 10.8 nM and16.8 nM for the best sequences in Liu

et al.[42] and Berger et al.[211], respectively, compared to 343 nM[46] in our

studies. However, these studies used different Six1 constructs so the choice of

construct may be important. Full length mouse Six1was used to identify the 10.8

nM sequence[42] whereas a mouse Six1 fragment containing the HD + 15 flanking residues (N and C terminal) identified the 16.8 nM sequence[211]. Of

interest, the affinity of full length mouse Six1to the MEF3 promoter was 34.7

nM[42], roughly ten times higher affinity than what we observed with human Six1

1-259. This suggests that both the context of the DNA core sequence and the

Six1 construct chosen is likely important. Using a full length Six1 may be a better

choice than the Six1 1-259 construct I used in case regions outside of my

construct contribute to binding specificity.

113

Another approach could be to develop an Alphascreen®-based assay to serve as a primary screen for HTS or as a confirmatory secondary assay.

However, the non-specific binding of Six1 would still need to be resolved. A DNA- binding Alphascreen® assay has been successfully used to identify and characterize inhibitors of the TAR DNA binding protein 43 (TDP-43) with IC50

values between 100 nM and 10 µM[237]. If the Six1-DNA interaction is to be

pursued in the future, the development of an Alphascreen® assay, similar to that

of TDP-43, would be ideal as a primary or secondary assay.

Inhibitors of Eyas phosphatase activity

We identified specific inhibitors of the Eya2 phosphatase activity and have

demonstrated efficacy in a cellular context. The Eya phosphatase activity is

crucial to several disease processes, including promoting cell migration, invasion,

transformation, contact-independent growth, tumor colony formation, cellular

proliferation, and metastasis of breast cancer cells[112, 113]. As such, it is

hypothesized that a selective inhibitor of Eya will affect these implicated

processes within breast cancer. To this point, our inhibitors reduce the Eya2-

dependent migration of breast cells. Current plans are to expand these studies

using additional cell models to investigate effects of inhibitors on preventing Six1-

mediated cellular phenotypes and several assays are already developed that can

be used. Six1-overexpressing MCF7 cells require Eya expression to induce Six1-

mediated pro-growth and pro-metastatic phenotypes[34, 87]. Following treatment

with Eya PTP inhibitor, MTS assays, BrdU incorporation, and cell counts can be 114

used to observe effects on survival and growth with this system. In addition, the

dependence of Eya’s PTP activity on TGF-β signaling and EMT can be directly

determined using Eya PTP inhibitors. It is known that Eya is important for these

processes[34], but the role of the PTP activity has not been definitively

determined. Assays such as these will determine whether the Eya2 PTP

inhibitors affect the Six1-dependent pro-tumorigenic and pro-metastatic

properties, suggesting that compounds may inhibit tumorigenesis and metastasis

in vivo. It remains to be seen whether these Eya2 PTP inhibitors affect breast

cancer models in vivo. Testing potent inhibitors for activity against the migration,

invasion, transformation, and metastasis is of high priority and will emphasize the

therapeutic benefits of Eya inhibitors.

The compounds we identified likely need to be optimized for higher

potency to demonstrate efficacy in animal models. The inhibitor-bound Eya2 crystal structure will be invaluable for compound optimization since it will provide a more accurate and detailed binding interaction than modeling alone. It is likely that new crystal forms of Eya will be necessary as the proposed binding site is involved in crystal contact in the Eya2 ED alone crystals. Despite the proposed binding site being solvent exposed, the Six1-Eya2 protein co-crystal disintegrates upon inhibitor soaking. Indeed, extensive crystallography efforts are underway, including using a fixed-arm carrier protein with surface entropy reduction[238] that was instrumental in obtaining Six1-Eya2 co-crystals. This method uses an

ED and maltose binding protein fusion construct containing surface mutations designed to encourage crystal lattice formation through reduction of surface

115

entropy. Soaking any new crystal forms with inhibitors will be of high priority to assist in improving the potency of the inhibitors through chemistry optimization.

NMR remains a viable option as well. However, several difficulties remain. The aggregation of Eya in NMR conditions was improved, but not fully resolved.

Obtaining spectra capable of assigning peaks or generating a solution structure of Eya2 is not possible under current NMR conditions. A protein engineering approach to reduce aggregation and improve stability may be attempted. As one example, mutations of select amino acids of Pseudomonas mendocina lipase demonstrate a significant improvement in stability compared to wild type. These mutations do not induce significant changes to structure as assessed by circular dichroism, differential scanning calorimetry, and fluorescence studies[239].

However, to obtain suitable spectra for generating a structure, Eya2 needs to be denatured and refolded. Every attempt at refolding Eya2 has been unsuccessful.

If an NMR structure is required, both the NMR conditions and a new refolding method will need extensive optimization.

In addition to improving the potency, the drug-like quality of the compounds also needs to be considered in chemical optimization. Statistical analysis of drug and compound databases has determined that drug-like molecules share key physical properties that can often predict pharmacokinetic properties. Lipinski et al.[240] determined drug-like compounds generally share molecular weights (MW ≤ 500 Da), a lipophilicity range as computed by octanol- water partition coefficient (ClogP ≤ 5), hydrogen bond (HB) donors (OH’s + NH’s

≤ 5), and HB acceptors (N’s + O’s ≤ 10). Additional studies[241-243] constrains

116

Figure 33. Drug-like properties of Eya2 inhibitors. Identified Eya2 inhibitors have favorable drug-like properties as described by Lipinski et al[240].

the number of rotatable bonds (RB≤ 8), rings in the molecule (rings ≤ 4), molar refractivity, and polar surface area (PSA ≤ 120 Å 2). Each of these parameters

must be considered to produce a viable drug and conforming to these

parameters (ClogP, PSA, MW, HB donors/acceptors) is widely considered

necessary in developing early structure-activity relationships (SAR) around lead

compounds[244]. However, most hits identified through HTS have low affinity

with drug-like MW (300-500) and ClogP values of 3-5. Our active compounds fall

into this category: the MW = 323.7-402.3 and ClogP values are between 3.7 and

4.6 (Fig. 33), suggesting they are highly attractive lead compounds. Further

modification of lead compounds with optimized potency will also need to address

117

any issues in solubility, cell permeability, and metabolism. As such, medicinal

chemistry efforts to increase potency will have to run parallel with efforts to

improve drug-like properties before this compound series can be investigated in

vivo. Ultimately, any highly specific compound optimized for drug-like properties should have good ADME (absorption, distribution, metabolism, and excretion), in vivo efficacy, and favorable toxicology profiles. One obvious issue is that these compounds contain a hydrazide functional group. These are often used for crosslinking and may pose a problem for a potential inhibitor to function non- specifically. However, the chemical series identified has hydrazide-containing

compounds of varying potency (2.0 µM to inactive). This suggests that the

hydrazide appears to pose little problems in the biochemical assays. The

importance of the hydrazide has not been directly tested in a cellular context and

may still be a threat to specificity or produce off-target effects.

Of note, selective phosphatase inhibitors are inherently difficult to obtain.

PTP catalytic sites are relatively shallow and lack a defined pocket capable of

binding small molecules like ATP, such as those found in kinases. Screening of

compound libraries usually find negatively charged inhibitors with poor cell

permeability and bioavailability due to complementarities to the positive charged

PTP active site[125]. Although difficult, it is possible to develop inhibitors that

display >10-fold selectivity against similar PTPs and >100-fold selectivity against

other PTPs. PTP inhibitors displaying >100-fold selectivity against all PTPs,

including ones closely related, have yet to be seen. Given that our Eya2

inhibitors selectively inhibit Eya2 over Eya3 and verified by biochemical data, we

118

have demonstrated that high selectivity for PTP inhibitors is possible even for highly conserved family members.

In some cases, it may be advantageous to have a pan-Eya inhibitor. For example, Eya1 also correlates with adverse outcomes in breast cancer[87] and examination of the TCGA[195] breast cancer dataset shows that Eya1 and Eya2 are amplified in 10.8% of breast cancers, raising the possibility of multiple Eyas acting in concert. Published data suggests that Eya2 is a dominant player in breast and ovarian cancers[67, 87], as this Eya correlates with poor prognosis in both these cancers, and is amplified in 14.8 percent of ovarian cancers[82]. On the other hand, Eya4 may have both oncogenic[196, 197] and tumor suppressive roles[84], thereby limiting the therapeutic potential. In addition, recent data also indicate that some Eya family members may function in adult tissues. For example, the phosphatase activity of Eya1 regulates cell polarity in the lung epithelium[113]. Evidence from Eya3 knock-out mice suggest that Eya3 is expressed in a tissue specific manner, with adult expression in the brain, skeletal muscle, heart, kidneys, and weak expression in the lungs, resulting in only mild phenotypic differences[198]. Thus, having an inhibitor that is highly selective toward a specific Eya family member (i.e., Eya2) may be advantageous in some situations, while having a pan-Eya inhibitor may be needed in others in which multiple Eya members are involved.

Additional HTS efforts using a different library to identify additional Eya inhibitors discovered a second series of non-related inhibitors that target Eya2.

Having only been identified recently, this inhibitor series needs to be confirmed

119

for Eya2, has not been tested against other Eya family members, nor has been

characterized. In addition, early efforts are underway to develop covalent

inhibitors of Eya phosphatase activity. Such a strategy has been used to inhibit

the PTP CD45 using a specific, small tri-peptide containing an α-

bromobenzylphosphonate analog forms a covalent adduct with CD45[245].

Because the Eyas share a highly conserved active site, this strategy may be

successful in obtaining a pan-Eya inhibitor. Additional efforts may be to mirror the computational efforts of Park et al.[190, 246] to identify Eya inhibitors. They used

a virtual screening approach of >400, 000 compounds filtered for drug-like

qualities and identified inhibitors of Eya2 with low µM IC50s. Using a similar

approach with larger libraries or with virtual fragment libraries targeted towards

the Eya active site may provide novel inhibitors that function against all Eyas.

Conclusion

This thesis outlines the efforts made at identifying small molecules

targeting the Six1 transcription complex with the goal of developing them into

novel, tumor-specific chemotherapeutic agents able to treat multiple aspects of

breast tumorigenesis, including proliferation, survival, and metastasis.

Compounds have yet to be found that target the Six1-DNA interaction, but

promising compounds targeting the Six1-Eya2 interaction have been identified

and are currently being examined by others in the lab. Inhibitors against the Eya2

phosphatase have been successfully identified, function as expected in a cellular

120

context, and await an improvement in potency before they can be characterized in an in vivo system.

121

REFERENCES

1. Lewis, M.T., Homeobox genes in mammary gland development and neoplasia. Breast Cancer Res, 2000. 2(3): p. 158-69.

2. Darnell, J.E., Jr., Transcription factors as targets for cancer therapy. Nat Rev Cancer, 2002. 2(10): p. 740-9.

3. Bonini, N.M., W.M. Leiserson, and S. Benzer, The eyes absent gene: genetic control of cell survival and differentiation in the developing Drosophila eye. Cell, 1993. 72(3): p. 379-95.

4. Ohto, H., et al., Cooperation of six and eya in activation of their target genes through nuclear translocation of Eya. Mol Cell Biol, 1999. 19(10): p. 6815-24.

5. Silver, S.J., et al., Functional dissection of eyes absent reveals new modes of regulation within the retinal determination gene network. Mol Cell Biol, 2003. 23(17): p. 5989-99.

6. Xu, P.X., et al., Mouse Eya genes are expressed during limb tendon development and encode a transcriptional activation function. Proc Natl Acad Sci U S A, 1997. 94(22): p. 11974-9.

7. Jemc, J. and I. Rebay, The eyes absent family of phosphotyrosine phosphatases: properties and roles in developmental regulation of transcription. Annu Rev Biochem, 2007. 76: p. 513-38.

8. Mardon, G., N.M. Solomon, and G.M. Rubin, dachshund encodes a nuclear protein required for normal eye and leg development in Drosophila. Development, 1994. 120(12): p. 3473-86.

9. Hammond, K.L., et al., Mammalian and Drosophila dachshund genes are related to the Ski proto-oncogene and are expressed in eye and limb. Mech Dev, 1998. 74(1-2): p. 121-31.

10. Davis, R.J., et al., Mouse Dach2 mutants do not exhibit gross defects in eye development or brain function. Genesis, 2006. 44(2): p. 84-92.

11. Davis, R.J., et al., Mouse Dach, a homologue of Drosophila dachshund, is expressed in the developing retina, brain and limbs. Dev Genes Evol, 1999. 209(9): p. 526-36.

12. Seo, H.C., et al., Six class homeobox genes in drosophila belong to three distinct families and are involved in head development. Mech Dev, 1999. 83(1-2): p. 127-39.

122

13. Abdelhak, S., et al., A human homologue of the Drosophila eyes absent gene underlies branchio-oto-renal (BOR) syndrome and identifies a novel gene family. Nat Genet, 1997. 15(2): p. 157-64.

14. Xu, P.X., et al., Mouse Eya homologues of the Drosophila eyes absent gene require Pax6 for expression in lens and nasal placode. Development, 1997. 124(1): p. 219-31.

15. Zimmerman, J.E., et al., Cloning and characterization of two vertebrate homologs of the Drosophila eyes absent gene. Genome Res, 1997. 7(2): p. 128-41.

16. Duncan, M.K., et al., Eyes absent: a gene family found in several metazoan phyla. Mamm Genome, 1997. 8(7): p. 479-85.

17. Furuya, M., et al., The C. elegans eyes absent ortholog EYA-1 is required for tissue differentiation and plays partially redundant roles with PAX-6. Dev Biol, 2005. 286(2): p. 452-63.

18. Kozlowski, D.J., et al., The zebrafish dog-eared mutation disrupts eya1, a gene required for cell survival and differentiation in the inner ear and lateral line. Dev Biol, 2005. 277(1): p. 27-41.

19. Borsani, G., et al., EYA4, a novel vertebrate gene related to Drosophila eyes absent. Hum Mol Genet, 1999. 8(1): p. 11-23.

20. Xu, P.X., The EYA-SO/SIX complex in development and disease. Pediatr Nephrol, 2013. 28(6): p. 843-54.

21. Chen, B., E.H. Kim, and P.X. Xu, Initiation of olfactory placode development and neurogenesis is blocked in mice lacking both Six1 and Six4. Dev Biol, 2009. 326(1): p. 75-85.

22. Purcell, P., et al., Pax6-dependence of Six3, Eya1 and Dach1 expression during lens and nasal placode induction. Gene Expr Patterns, 2005. 6(1): p. 110-8.

23. Liu, W., et al., Six3 activation of Pax6 expression is essential for mammalian lens induction and specification. EMBO J, 2006. 25(22): p. 5383-95.

24. Oliver, G., et al., Homeobox genes and connective tissue patterning. Development, 1995. 121(3): p. 693-705.

25. Ford, H.L., et al., Abrogation of the G2 cell cycle checkpoint associated with overexpression of HSIX1: a possible mechanism of breast carcinogenesis. Proc Natl Acad Sci U S A, 1998. 95(21): p. 12608-13.

123

26. Coletta, R.D., et al., The Six1 homeoprotein stimulates tumorigenesis by reactivation of cyclin A1. Proc Natl Acad Sci U S A, 2004. 101(17): p. 6478-83.

27. Yu, Y., et al., The homeoprotein six1 transcriptionally activates multiple protumorigenic genes but requires ezrin to promote metastasis. Cancer Res, 2006. 66(4): p. 1982-9.

28. Kawakami, K., et al., Six family genes--structure and function as transcription factors and their roles in development. Bioessays, 2000. 22(7): p. 616-26.

29. Goudreau, G., et al., Mutually regulated expression of Pax6 and Six3 and its implications for the Pax6 haploinsufficient lens phenotype. Proc Natl Acad Sci U S A, 2002. 99(13): p. 8719-24.

30. Kobayashi, M., et al., Overexpression of the forebrain-specific homeobox gene six3 induces rostral forebrain enlargement in zebrafish. Development, 1998. 125(15): p. 2973-82.

31. Relaix, F. and M. Buckingham, From insect eye to vertebrate muscle: redeployment of a regulatory network. Genes Dev, 1999. 13(24): p. 3171- 8.

32. Grifone, R., et al., Six1 and Six4 homeoproteins are required for Pax3 and Mrf expression during myogenesis in the mouse embryo. Development, 2005. 132(9): p. 2235-49.

33. Xu, P.X., et al., Six1 is required for the early organogenesis of mammalian kidney. Development, 2003. 130(14): p. 3085-94.

34. Micalizzi, D.S., et al., The Six1 homeoprotein induces human mammary carcinoma cells to undergo epithelial-mesenchymal transition and metastasis in mice through increasing TGF-beta signaling. J Clin Invest, 2009. 119(9): p. 2678-90.

35. Gehring, W.J., et al., Homeodomain-DNA recognition. Cell, 1994. 78(2): p. 211-23.

36. McGinnis, W. and R. Krumlauf, Homeobox genes and axial patterning. Cell, 1992. 68(2): p. 283-302.

37. Levine, M. and T. Hoey, Homeobox proteins as sequence-specific transcription factors. Cell, 1988. 55(4): p. 537-40.

38. Pearson, J.C., D. Lemons, and W. McGinnis, Modulating Hox gene functions during animal body patterning. Nat Rev Genet, 2005. 6(12): p. 893-904. 124

39. Svingen, T. and K.F. Tonissen, Hox transcription factors and their elusive mammalian gene targets. Heredity (Edinb), 2006. 97(2): p. 88-96.

40. Patrick, A.N., et al., Structure-function analyses of the human SIX1-EYA2 complex reveal insights into metastasis and BOR syndrome. Nat Struct Mol Biol, 2013. 20(4): p. 447-53.

41. Ades, S.E. and R.T. Sauer, Specificity of minor-groove and major-groove interactions in a homeodomain-DNA complex. Biochemistry, 1995. 34(44): p. 14601-8.

42. Liu, Y., et al., Discovery, optimization and validation of an optimal DNA- binding sequence for the Six1 homeodomain transcription factor. Nucleic Acids Res, 2012. 40(17): p. 8227-39.

43. Kawakami, K., et al., Structure, function and expression of a murine homeobox protein AREC3, a homologue of Drosophila sine oculis gene product, and implication in development. Nucleic Acids Res, 1996. 24(2): p. 303-10.

44. Harris, S.E., C.L. Winchester, and K.J. Johnson, Functional analysis of the homeodomain protein SIX5. Nucleic Acids Res, 2000. 28(9): p. 1871-8.

45. Aishima, J., et al., A Hoogsteen base pair embedded in undistorted B- DNA. Nucleic Acids Res, 2002. 30(23): p. 5244-52.

46. Patrick, A.N., et al., Biochemical and functional characterization of six SIX1 Branchio-oto-renal syndrome mutations. J Biol Chem, 2009. 284(31): p. 20781-90.

47. Hu, S., A. Mamedova, and R.S. Hegde, DNA-binding and regulation mechanisms of the SIX family of retinal determination proteins. Biochemistry, 2008. 47(11): p. 3586-94.

48. Spitz, F., et al., Expression of myogenin during embryogenesis is controlled by Six/sine oculis homeoproteins through a conserved MEF3 binding site. Proc Natl Acad Sci U S A, 1998. 95(24): p. 14220-5.

49. Kussie, P.H., et al., Structure of the MDM2 oncoprotein bound to the p53 tumor suppressor transactivation domain. Science, 1996. 274(5289): p. 948-53.

50. Sattler, M., et al., Structure of Bcl-xL-Bak peptide complex: recognition between regulators of apoptosis. Science, 1997. 275(5302): p. 983-6.

51. Vassilev, L.T., et al., In vivo activation of the p53 pathway by small- molecule antagonists of MDM2. Science, 2004. 303(5659): p. 844-8.

125

52. Shangary, S. and S. Wang, Small-molecule inhibitors of the MDM2-p53 protein-protein interaction to reactivate p53 function: a novel approach for cancer therapy. Annu Rev Pharmacol Toxicol, 2009. 49: p. 223-41.

53. Xu, P.X., et al., Eya1 is required for the morphogenesis of mammalian thymus, parathyroid and thyroid. Development, 2002. 129(13): p. 3033-44.

54. Li, X., et al., Eya protein phosphatase activity regulates Six1-Dach-Eya transcriptional effects in mammalian organogenesis. Nature, 2003. 426(6964): p. 247-54.

55. Laclef, C., et al., Thymus, kidney and craniofacial abnormalities in Six 1 deficient mice. Mech Dev, 2003. 120(6): p. 669-79.

56. Ando, Z., et al., Slc12a2 is a direct target of two closely related homeobox proteins, Six1 and Six4. FEBS J, 2005. 272(12): p. 3026-41.

57. Zou, D., et al., Eya1 and Six1 are essential for early steps of sensory neurogenesis in mammalian cranial placodes. Development, 2004. 131(22): p. 5561-72.

58. Zou, D., et al., Eya1 regulates the growth of otic epithelium and interacts with Pax2 during the development of all sensory areas in the inner ear. Dev Biol, 2006. 298(2): p. 430-41.

59. Abdelhak, S., et al., Clustering of mutations responsible for branchio-oto- renal (BOR) syndrome in the eyes absent homologous region (eyaHR) of EYA1. Hum Mol Genet, 1997. 6(13): p. 2247-55.

60. Ruf, R.G., et al., SIX1 mutations cause branchio-oto-renal syndrome by disruption of EYA1-SIX1-DNA complexes. Proc Natl Acad Sci U S A, 2004. 101(21): p. 8090-5.

61. Hoskins, B.E., et al., Transcription factor SIX5 is mutated in patients with branchio-oto-renal syndrome. Am J Hum Genet, 2007. 80(4): p. 800-4.

62. Azuma, N., et al., Mutations of a human homologue of the Drosophila eyes absent gene (EYA1) detected in patients with congenital cataracts and ocular anterior segment anomalies. Hum Mol Genet, 2000. 9(3): p. 363-6.

63. Shimasaki, N., et al., EYA1 mutation in a newborn female presenting with cardiofacial syndrome. Pediatr Cardiol, 2004. 25(4): p. 411-3.

64. Schonberger, J., et al., Mutation in the transcriptional coactivator EYA4 causes dilated cardiomyopathy and sensorineural hearing loss. Nat Genet, 2005. 37(4): p. 418-22.

126

65. Wayne, S., et al., Mutations in the transcriptional activator EYA4 cause late-onset deafness at the DFNA10 locus. Hum Mol Genet, 2001. 10(3): p. 195-200.

66. Pfister, M., et al., A 4-bp insertion in the eya-homologous region (eyaHR) of EYA4 causes hearing impairment in a Hungarian family linked to DFNA10. Mol Med, 2002. 8(10): p. 607-11.

67. Behbakht, K., et al., Six1 overexpression in ovarian carcinoma causes resistance to TRAIL-mediated apoptosis and is associated with poor survival. Cancer Res, 2007. 67(7): p. 3036-42.

68. Ng, K.T., et al., Clinicopathological significance of homeoprotein Six1 in hepatocellular carcinoma. Br J Cancer, 2006. 95(8): p. 1050-5.

69. Wan, F., et al., Gene expression changes during HPV-mediated carcinogenesis: a comparison between an in vitro cell model and cervical cancer. Int J Cancer, 2008. 123(1): p. 32-40.

70. Li, C.M., et al., Gene expression in Wilms' tumor mimics the earliest committed stage in the metanephric mesenchymal-epithelial transition. Am J Pathol, 2002. 160(6): p. 2181-90.

71. Yu, Y., et al., Expression profiling identifies the cytoskeletal organizer ezrin and the developmental homeoprotein Six-1 as key metastatic regulators. Nat Med, 2004. 10(2): p. 175-81.

72. Mimae, T., et al., Upregulation of notch2 and six1 is associated with progression of early-stage lung adenocarcinoma and a more aggressive phenotype at advanced stages. Clin Cancer Res, 2012. 18(4): p. 945-55.

73. Zheng, X.H., et al., Expression and clinical implications of homeobox gene Six1 in cervical cancer cell lines and cervical epithelial tissues. Int J Gynecol Cancer, 2010. 20(9): p. 1587-92.

74. Ono, H., et al., SIX1 promotes epithelial-mesenchymal transition in colorectal cancer through ZEB1 activation. Oncogene, 2012. 31(47): p. 4923-34.

75. McCoy, E.L., et al., Six1 expands the mouse mammary epithelial stem/progenitor cell pool and induces mammary tumors that undergo epithelial-mesenchymal transition. J Clin Invest, 2009. 119(9): p. 2663-77.

76. Wang, Q.F., et al., MLL fusion proteins preferentially regulate a subset of wild-type MLL target genes in the leukemic genome. Blood, 2011. 117(25): p. 6895-905.

127

77. Reichenberger, K.J., et al., Gene amplification is a mechanism of Six1 overexpression in breast cancer. Cancer Res, 2005. 65(7): p. 2668-75.

78. Coletta, R.D., et al., Six1 overexpression in mammary cells induces genomic instability and is sufficient for malignant transformation. Cancer Res, 2008. 68(7): p. 2204-13.

79. Iwanaga, R., et al., Expression of Six1 in luminal breast cancers predicts poor prognosis and promotes increases in tumor initiating cells by activation of extracellular signal-regulated kinase and transforming growth factor-beta signaling pathways. Breast Cancer Res, 2012. 14(4): p. R100.

80. Smith, A.L., et al., The miR-106b-25 cluster targets Smad7, activates TGF-beta signaling, and induces EMT and tumor initiating cell characteristics downstream of Six1 in human breast cancer. Oncogene, 2012. 31(50): p. 5162-71.

81. Wang, C.A., et al., SIX1 induces lymphangiogenesis and metastasis via upregulation of VEGF-C in mouse models of breast cancer. J Clin Invest, 2012. 122(5): p. 1895-906.

82. Zhang, L., et al., Transcriptional coactivator Drosophila eyes absent homologue 2 is up-regulated in epithelial ovarian cancer and promotes tumor growth. Cancer Res, 2005. 65(3): p. 925-32.

83. Zhang, Y., et al., A comparative study of Eya1 and Eya4 protein function and its implication in branchio-oto-renal syndrome and DFNA10. J Assoc Res Otolaryngol, 2004. 5(3): p. 295-304.

84. Miller, S.J., et al., Inhibition of Eyes Absent Homolog 4 expression induces malignant peripheral nerve sheath tumor necrosis. Oncogene, 2010. 29(3): p. 368-79.

85. Wang, Y., et al., Gene-expression profiles to predict distant metastasis of lymph-node-negative primary breast cancer. Lancet, 2005. 365(9460): p. 671-9.

86. van de Vijver, M.J., et al., A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med, 2002. 347(25): p. 1999-2009.

87. Farabaugh, S.M., et al., Eya2 is required to mediate the pro-metastatic functions of Six1 via the induction of TGF-beta signaling, epithelial- mesenchymal transition, and cancer stem cell properties. Oncogene, 2012. 31(5): p. 552-62.

88. Rayapureddi, J.P., et al., Eyes absent represents a class of protein tyrosine phosphatases. Nature, 2003. 426(6964): p. 295-8.

128

89. Tootle, T.L., et al., The transcription factor Eyes absent is a protein tyrosine phosphatase. Nature, 2003. 426(6964): p. 299-302.

90. Silver, S.J. and I. Rebay, Signaling circuitries in development: insights from the retinal determination gene network. Development, 2005. 132(1): p. 3-13.

91. Donner, A.L. and R.L. Maas, Conservation and non-conservation of genetic pathways in eye specification. Int J Dev Biol, 2004. 48(8-9): p. 743-53.

92. Rebay, I., S.J. Silver, and T.L. Tootle, New vision from Eyes absent: transcription factors as enzymes. Trends Genet, 2005. 21(3): p. 163-71.

93. Allen, K.N. and D. Dunaway-Mariano, Markers of fitness in a successful enzyme superfamily. Curr Opin Struct Biol, 2009. 19(6): p. 658-65.

94. Burroughs, A.M., et al., Evolutionary genomics of the HAD superfamily: understanding the structural adaptations and catalytic diversity in a superfamily of phosphoesterases and allied enzymes. J Mol Biol, 2006. 361(5): p. 1003-34.

95. Seifried, A., J. Schultz, and A. Gohla, Human HAD phosphatases: structure, mechanism, and roles in health and disease. FEBS J, 2013. 280(2): p. 549-71.

96. Thaller, M.C., S. Schippa, and G.M. Rossolini, Conserved sequence motifs among bacterial, eukaryotic, and archaeal phosphatases that define a new phosphohydrolase superfamily. Protein Sci, 1998. 7(7): p. 1647-52.

97. Collet, J.F., V. Stroobant, and E. Van Schaftingen, Mechanistic studies of phosphoserine phosphatase, an enzyme related to P-type ATPases. J Biol Chem, 1999. 274(48): p. 33985-90.

98. Newman, J.W., et al., The soluble epoxide hydrolase encoded by EPXH2 is a bifunctional enzyme with novel lipid phosphate phosphatase activity. Proc Natl Acad Sci U S A, 2003. 100(4): p. 1558-63.

99. Bernstein, N.K., et al., The molecular architecture of the mammalian DNA repair enzyme, polynucleotide kinase. Mol Cell, 2005. 17(5): p. 657-70.

100. Koonin, E.V. and R.L. Tatusov, Computer analysis of bacterial haloacid dehalogenases defines a large superfamily of hydrolases with diverse specificity. Application of an iterative approach to database search. J Mol Biol, 1994. 244(1): p. 125-32.

129

101. Aravind, L., M.Y. Galperin, and E.V. Koonin, The catalytic domain of the P-type ATPase has the haloacid dehalogenase fold. Trends Biochem Sci, 1998. 23(4): p. 127-9.

102. Jung, S.K., et al., Crystal structure of ED-Eya2: insight into dual roles as a protein tyrosine phosphatase and a transcription factor. FASEB J, 2010. 24(2): p. 560-9.

103. Peisach, E., et al., X-ray crystal structure of the hypothetical phosphotyrosine phosphatase MDP-1 of the haloacid dehalogenase superfamily. Biochemistry, 2004. 43(40): p. 12770-9.

104. Zhang, Y., et al., Determinants for dephosphorylation of the RNA polymerase II C-terminal domain by Scp1. Mol Cell, 2006. 24(5): p. 759- 70.

105. Kamenski, T., et al., Structure and mechanism of RNA polymerase II CTD phosphatases. Mol Cell, 2004. 15(3): p. 399-407.

106. Almo, S.C., et al., Structural genomics of protein phosphatases. J Struct Funct Genomics, 2007. 8(2-3): p. 121-40.

107. Okabe, Y., T. Sano, and S. Nagata, Regulation of the innate immune response by threonine-phosphatase of Eyes absent. Nature, 2009. 460(7254): p. 520-4.

108. Liu, X., et al., Drosophila EYA regulates the immune response against DNA through an evolutionarily conserved threonine phosphatase motif. PLoS One, 2012. 7(8): p. e42725.

109. Jin, M., et al., Eyes absent tyrosine phosphatase activity is not required for Drosophila development or survival. PLoS One, 2013. 8(3): p. e58818.

110. Tadjuidje, E., et al., The EYA tyrosine phosphatase activity is pro- angiogenic and is inhibited by benzbromarone. PLoS One, 2012. 7(4): p. e34806.

111. Rayapureddi, J.P. and R.S. Hegde, Branchio-oto-renal syndrome associated mutations in Eyes Absent 1 result in loss of phosphatase activity. FEBS Lett, 2006. 580(16): p. 3853-9.

112. Pandey, R.N., et al., The Eyes Absent phosphatase-transactivator proteins promote proliferation, transformation, migration, and invasion of tumor cells. Oncogene, 2010. 29(25): p. 3715-22.

113. El-Hashash, A.H., et al., Eya1 protein phosphatase regulates tight junction formation in lung distal epithelium. J Cell Sci, 2012. 125(Pt 17): p. 4036- 48. 130

114. Rayapureddi, J.P., et al., Characterization of a plant, tyrosine-specific phosphatase of the aspartyl class. Biochemistry, 2005. 44(2): p. 751-8.

115. Cook, P.J., et al., Tyrosine dephosphorylation of H2AX modulates apoptosis and survival decisions. Nature, 2009. 458(7238): p. 591-6.

116. Xiao, A., et al., WSTF regulates the H2A.X DNA damage response via a novel tyrosine kinase activity. Nature, 2009. 457(7225): p. 57-62.

117. Stokes, M.P., et al., Profiling of UV-induced ATM/ATR signaling pathways. Proc Natl Acad Sci U S A, 2007. 104(50): p. 19855-60.

118. Matsuoka, S., et al., ATM and ATR substrate analysis reveals extensive protein networks responsive to DNA damage. Science, 2007. 316(5828): p. 1160-6.

119. Krishnan, N., et al., Dephosphorylation of the C-terminal tyrosyl residue of the DNA damage-related histone H2A.X is mediated by the protein phosphatase eyes absent. J Biol Chem, 2009. 284(24): p. 16066-70.

120. Krueger, A.B., et al., Identification of a selective small-molecule inhibitor series targeting the eyes absent 2 (Eya2) phosphatase activity. J Biomol Screen, 2013. 18(1): p. 85-96.

121. Hunter, T., Tyrosine phosphorylation: thirty years and counting. Curr Opin Cell Biol, 2009. 21(2): p. 140-6.

122. Hanahan, D. and R.A. Weinberg, The hallmarks of cancer. Cell, 2000. 100(1): p. 57-70.

123. Luo, J., N.L. Solimini, and S.J. Elledge, Principles of cancer therapy: oncogene and non-oncogene addiction. Cell, 2009. 136(5): p. 823-37.

124. Scott, L.M., et al., Targeting protein tyrosine phosphatases for anticancer drug discovery. Curr Pharm Des, 2010. 16(16): p. 1843-62.

125. De Munter, S., M. Kohn, and M. Bollen, Challenges and opportunities in the development of protein phosphatase-directed therapeutics. ACS Chem Biol, 2013. 8(1): p. 36-45.

126. Gibbs, J.B., Mechanism-based target identification and drug discovery in cancer research. Science, 2000. 287(5460): p. 1969-73.

127. Ostman, A., C. Hellberg, and F.D. Bohmer, Protein-tyrosine phosphatases and cancer. Nat Rev Cancer, 2006. 6(4): p. 307-20.

128. Lazo, J.S. and P. Wipf, Is Cdc25 a druggable target? Anticancer Agents Med Chem, 2008. 8(8): p. 837-42. 131

129. Andersen, J.N., et al., A genomic perspective on protein tyrosine phosphatases: gene structure, pseudogenes, and genetic disease linkage. FASEB J, 2004. 18(1): p. 8-30.

130. Tonks, N.K., Protein tyrosine phosphatases: from genes, to function, to disease. Nat Rev Mol Cell Biol, 2006. 7(11): p. 833-46.

131. Knight, Z.A., H. Lin, and K.M. Shokat, Targeting the cancer kinome through polypharmacology. Nat Rev Cancer, 2010. 10(2): p. 130-7.

132. Rabiller, M., et al., Proteus in the world of proteins: conformational changes in protein kinases. Arch Pharm (Weinheim), 2010. 343(4): p. 193- 206.

133. O'Neill, A.K., M.J. Niederst, and A.C. Newton, Suppression of survival signalling pathways by the phosphatase PHLPP. FEBS J, 2013. 280(2): p. 572-83.

134. Julien, S.G., et al., Inside the human cancer tyrosine phosphatome. Nat Rev Cancer, 2011. 11(1): p. 35-49.

135. Le Guezennec, X. and D.V. Bulavin, WIP1 phosphatase at the crossroads of cancer and aging. Trends Biochem Sci, 2010. 35(2): p. 109-14.

136. Labbe, D.P., S. Hardy, and M.L. Tremblay, Protein tyrosine phosphatases in cancer: friends and foes! Prog Mol Biol Transl Sci, 2012. 106: p. 253- 306.

137. Hellmuth, K., et al., Specific inhibitors of the protein tyrosine phosphatase Shp2 identified by high-throughput docking. Proc Natl Acad Sci U S A, 2008. 105(20): p. 7275-80.

138. Parker, G.J., et al., Development of high throughput screening assays using fluorescence polarization: nuclear receptor-ligand-binding and kinase/phosphatase assays. J Biomol Screen, 2000. 5(2): p. 77-88.

139. Liu, G., et al., Selective protein tyrosine phosphatase 1B inhibitors: targeting the second phosphotyrosine binding site with non-carboxylic acid-containing ligands. J Med Chem, 2003. 46(16): p. 3437-40.

140. Liu, G., et al., Discovery and structure-activity relationship of oxalylarylaminobenzoic acids as inhibitors of protein tyrosine phosphatase 1B. J Med Chem, 2003. 46(11): p. 2093-103.

141. Skorey, K.I., et al., Development of a robust scintillation proximity assay for protein tyrosine phosphatase 1B using the catalytically inactive (C215S) mutant. Anal Biochem, 2001. 291(2): p. 269-78.

132

142. Montalibet, J. and B.P. Kennedy, Using yeast to screen for inhibitors of protein tyrosine phosphatase 1B. Biochem Pharmacol, 2004. 68(9): p. 1807-14.

143. Tautz, L. and T. Mustelin, Strategies for developing protein tyrosine phosphatase inhibitors. Methods, 2007. 42(3): p. 250-60.

144. Zhang, S. and Z.Y. Zhang, PTP1B as a drug target: recent developments in PTP1B inhibitor discovery. Drug Discov Today, 2007. 12(9-10): p. 373- 81.

145. Zhang, Z.Y., Protein tyrosine phosphatases: structure and function, substrate specificity, and inhibitor development. Annu Rev Pharmacol Toxicol, 2002. 42: p. 209-34.

146. Cheon, H.G., et al., Discovery of a novel protein tyrosine phosphatase-1B inhibitor, KR61639: potential development as an antihyperglycemic agent. Eur J Pharmacol, 2004. 485(1-3): p. 333-9.

147. Shen, K., et al., Acquisition of a specific and potent PTP1B inhibitor from a novel combinatorial library and screening procedure. J Biol Chem, 2001. 276(50): p. 47311-9.

148. Doman, T.N., et al., Molecular docking and high-throughput screening for novel inhibitors of protein tyrosine phosphatase-1B. J Med Chem, 2002. 45(11): p. 2213-21.

149. Shi, L., et al., Discovery of a novel competitive inhibitor of PTP1B by high- throughput screening. Acta Pharmacol Sin, 2008. 29(2): p. 278-84.

150. Zhou, X., et al., SHP2 is up-regulated in breast cancer cells and in infiltrating ductal carcinoma of the breast, implying its involvement in breast oncogenesis. Histopathology, 2008. 53(4): p. 389-402.

151. Zhou, X.D. and Y.M. Agazie, Inhibition of SHP2 leads to mesenchymal to epithelial transition in breast cancer cells. Cell Death Differ, 2008. 15(6): p. 988-96.

152. Chen, L., et al., Discovery of a novel shp2 protein tyrosine phosphatase inhibitor. Mol Pharmacol, 2006. 70(2): p. 562-70.

153. Bjorge, J.D., A. Pang, and D.J. Fujita, Identification of protein-tyrosine phosphatase 1B as the major tyrosine phosphatase activity capable of dephosphorylating and activating c-Src in several human breast cancer cell lines. J Biol Chem, 2000. 275(52): p. 41439-46.

133

154. Arias-Romero, L.E., et al., Activation of Src by protein tyrosine phosphatase 1B Is required for ErbB2 transformation of human breast epithelial cells. Cancer Res, 2009. 69(11): p. 4582-8.

155. Zhu, S., J.D. Bjorge, and D.J. Fujita, PTP1B contributes to the oncogenic properties of colon cancer cells through Src activation. Cancer Res, 2007. 67(21): p. 10129-37.

156. Zhang, Z., et al., Phosphorylated alpha-actinin and protein-tyrosine phosphatase 1B coregulate the disassembly of the focal adhesion kinase x Src complex and promote cell migration. J Biol Chem, 2006. 281(3): p. 1746-54.

157. Julien, S.G., et al., Protein tyrosine phosphatase 1B deficiency or inhibition delays ErbB2-induced mammary tumorigenesis and protects from lung metastasis. Nat Genet, 2007. 39(3): p. 338-46.

158. Bialy, L. and H. Waldmann, Inhibitors of protein tyrosine phosphatases: next-generation drugs? Angew Chem Int Ed Engl, 2005. 44(25): p. 3814- 39.

159. Han, Y., et al., Discovery of [(3-bromo-7-cyano-2-naphthyl) (difluoro) methyl]phosphonic acid, a potent and orally active small molecule PTP1B inhibitor. Bioorg Med Chem Lett, 2008. 18(11): p. 3200s-5.

160. Mattila, E., et al., Inhibition of receptor tyrosine kinase signalling by small molecule agonist of T-cell protein tyrosine phosphatase. BMC Cancer, 2010. 10: p. 7.

161. Zhang, S., et al., Acquisition of a potent and selective TC-PTP inhibitor via a stepwise fluorophore-tagged combinatorial synthesis and screening strategy. J Am Chem Soc, 2009. 131(36): p. 13072-9.

162. Waddleton, D., C. Ramachandran, and Q. Wang, Development of a method for evaluating protein tyrosine phosphatase CD45 inhibitors using Jurkat cell membrane. Anal Biochem, 2000. 285(1): p. 58-63.

163. Urbanek, R.A., et al., Potent reversible inhibitors of the protein tyrosine phosphatase CD45. J Med Chem, 2001. 44(11): p. 1777-93.

164. Yang, X., et al., Discovery of novel inhibitor of human leukocyte common antigen-related phosphatase. Biochim Biophys Acta, 2005. 1726(1): p. 34- 41.

165. Xie, Y., et al., Discovery of a novel submicromolar inhibitor of the lymphoid specific tyrosine phosphatase. Bioorg Med Chem Lett, 2008. 18(9): p. 2840-4.

134

166. Bernier, J., E.J. Hall, and A. Giaccia, Radiation oncology: a century of achievements. Nat Rev Cancer, 2004. 4(9): p. 737-47.

167. Christensen, K.L., et al., The six family of homeobox genes in development and cancer. Adv Cancer Res, 2008. 101: p. 93-126.

168. Ng, K.T., et al., Suppression of tumorigenesis and metastasis of hepatocellular carcinoma by shRNA interference targeting on homeoprotein Six1. Int J Cancer, 2010. 127(4): p. 859-72.

169. Jemc, J. and I. Rebay, Identification of transcriptional targets of the dual- function transcription factor/phosphatase eyes absent. Dev Biol, 2007. 310(2): p. 416-29.

170. Copeland, R.A., Mechanistic considerations in high-throughput screening. Anal Biochem, 2003. 320(1): p. 1-12.

171. Zhang, J.H., T.D. Chung, and K.R. Oldenburg, A Simple Statistical Parameter for Use in Evaluation and Validation of High Throughput Screening Assays. J Biomol Screen, 1999. 4(2): p. 67-73.

172. Inglese, J., et al., Quantitative high-throughput screening: a titration-based approach that efficiently identifies biological activities in large chemical libraries. Proc Natl Acad Sci U S A, 2006. 103(31): p. 11473-8.

173. Marley, A.E., et al., Biochemical characterization and deletion analysis of recombinant human protein phosphatase 2C alpha. Biochem J, 1996. 320 ( Pt 3): p. 801-6.

174. Ghosh, A., S. Shuman, and C.D. Lima, The structure of Fcp1, an essential RNA polymerase II CTD phosphatase. Mol Cell, 2008. 32(4): p. 478-90.

175. Zhang, M., et al., Structural and functional analysis of the phosphoryl transfer reaction mediated by the human small C-terminal domain phosphatase, Scp1. Protein Sci, 2010. 19(5): p. 974-86.

176. Jailkhani, N., V.K. Chaudhri, and K.V. Rao, Regulatory cascades of protein phosphatases: implications for cancer treatment. Anticancer Agents Med Chem, 2011. 11(1): p. 64-77.

177. Pereira, S.R., V.M. Vasconcelos, and A. Antunes, The phosphoprotein phosphatase family of Ser/Thr phosphatases as principal targets of naturally occurring toxins. Crit Rev Toxicol, 2011. 41(2): p. 83-110.

178. Tierno, M.B., et al., Development and optimization of high-throughput in vitro protein phosphatase screening assays. Nat Protoc, 2007. 2(5): p. 1134-44.

135

179. Southall, N.T., et al., Enabling the Large Scale Analysis of Quantitative High Throughput Screening Data, in Handbook of drug screening, R. Seethala and L. Zhang, Editors. 2009, Informa Healthcare: New York. p. 442-462.

180. Ikeda, K., et al., Six1 is essential for early neurogenesis in the development of olfactory epithelium. Dev Biol, 2007. 311(1): p. 53-68.

181. Zheng, W., et al., The role of Six1 in mammalian auditory system development. Development, 2003. 130(17): p. 3989-4000.

182. Xu, P.X., et al., Eya1-deficient mice lack ears and kidneys and show abnormal apoptosis of organ primordia. Nat Genet, 1999. 23(1): p. 113-7.

183. Pandey, R.N., et al., The Eyes Absent phosphatase-transactivator proteins promote proliferation, transformation, migration, and invasion of tumor cells. Oncogene, 2010.

184. Wu, K., et al., EYA1 Phosphatase Function Is Essential To Drive Breast Cancer Cell Proliferation Through Cyclin D1. Cancer Res, 2013.

185. Lepre, C.A. and J.M. Moore, Microdrop screening: a rapid method to optimize solvent conditions for NMR spectroscopy of proteins. J Biomol NMR, 1998. 12(4): p. 493-9.

186. Hojo, M., T. Ueda, and A. Inoue, UV-visible and H-1 NMR spectroscopic studies on direct chelate formation between alkaline earth metal ions and 1-(2-pyridylazo)-2-naphthol or 2-(5-bromo-2-pyridylazo)-5- diethylaminophenol in acetonitrile. Bulletin of the Chemical Society of Japan, 2002. 75(12): p. 2629-2636.

187. Robin, T.P., et al., EWS/FLI1 regulates EYA3 in Ewing sarcoma via modulation of miRNA-708, resulting in increased cell survival and chemoresistance. Mol Cancer Res, 2012. 10(8): p. 1098-108.

188. Mattila, E. and J. Ivaska, High-throughput methods in identification of protein tyrosine phosphatase inhibitors and activators. Anticancer Agents Med Chem, 2011. 11(1): p. 141-50.

189. Schneider, R., et al., Selective detection of allosteric phosphatase inhibitors. J Am Chem Soc, 2013. 135(18): p. 6838-41.

190. Park, H., et al., Structure-based virtual screening approach to the discovery of novel inhibitors of eyes absent 2 phosphatase with various metal chelating moieties. Chem Biol Drug Des, 2011. 78(4): p. 642-50.

136

191. Zhang, M., et al., Selective inactivation of a human neuronal silencing phosphatase by a small molecule inhibitor. ACS Chem Biol, 2011. 6(5): p. 511-9.

192. Holm, L. and C. Sander, Protein structure comparison by alignment of distance matrices. J Mol Biol, 1993. 233(1): p. 123-38.

193. Vang, T., et al., LYP inhibits T-cell activation when dissociated from CSK. Nat Chem Biol, 2012. 8(5): p. 437-46.

194. Harris, S.J. and D.J. Winzor, Thermodynamic nonideality as a probe of allosteric mechanisms: preexistence of the isomerization equilibrium for rabbit muscle pyruvate kinase. Arch Biochem Biophys, 1988. 265(2): p. 458-65.

195. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature, 2008. 455(7216): p. 1061-8.

196. Zou, H., et al., Frequent methylation of eyes absent 4 gene in Barrett's esophagus and esophageal adenocarcinoma. Cancer Epidemiol Biomarkers Prev, 2005. 14(4): p. 830-4.

197. Osborn, N.K., et al., Aberrant methylation of the eyes absent 4 gene in ulcerative colitis-associated dysplasia. Clin Gastroenterol Hepatol, 2006. 4(2): p. 212-8.

198. Soker, T., et al., Pleiotropic effects in Eya3 knockout mice. BMC Dev Biol, 2008. 8: p. 118.

199. Feig, M. and C.L. Brooks, 3rd, Recent advances in the development and application of implicit solvent models in biomolecule simulations. Curr Opin Struct Biol, 2004. 14(2): p. 217-24.

200. Cillo, C., et al., Homeobox genes in normal and malignant cells. J Cell Physiol, 2001. 188(2): p. 161-9.

201. Ford, H.L., Homeobox genes: a link between development, cell cycle, and cancer? Cell Biol Int, 1998. 22(6): p. 397-400.

202. Abate-Shen, C., Deregulated homeobox gene expression in cancer: cause or consequence? Nat Rev Cancer, 2002. 2(10): p. 777-85.

203. Gehring, W.J. and Y. Hiromi, Homeotic genes and the homeobox. Annu Rev Genet, 1986. 20: p. 147-73.

204. Cillo, C., et al., Homeobox genes and cancer. Exp Cell Res, 1999. 248(1): p. 1-9.

137

205. Tupler, R., G. Perini, and M.R. Green, Expressing the human genome. Nature, 2001. 409(6822): p. 832-3.

206. Stuart, E.T. and P. Gruss, PAX genes: what's new in developmental biology and cancer? Hum Mol Genet, 1995. 4 Spec No: p. 1717-20.

207. Rishi, V., et al., A high-throughput fluorescence-anisotropy screen that identifies small molecule inhibitors of the DNA binding of B-ZIP transcription factors. Anal Biochem, 2005. 340(2): p. 259-71.

208. Altschuler, S.E., J.E. Croy, and D.S. Wuttke, A small molecule inhibitor of Pot1 binding to telomeric DNA. Biochemistry, 2012. 51(40): p. 7833-45.

209. Vinson, C., A rationally designed small molecule that inhibits the HIF- 1alpha-ARNT heterodimer from binding to DNA in vivo. Sci STKE, 2005. 2005(284): p. pe23.

210. Olenyuk, B.Z., et al., Inhibition of vascular endothelial growth factor with a sequence-specific hypoxia response element antagonist. Proc Natl Acad Sci U S A, 2004. 101(48): p. 16768-73.

211. Berger, M.F., et al., Variation in homeodomain DNA binding revealed by high-resolution analysis of sequence preferences. Cell, 2008. 133(7): p. 1266-76.

212. Zinzalla, G. and D.E. Thurston, Targeting protein-protein interactions for therapeutic intervention: a challenge for the future. Future Med Chem, 2009. 1(1): p. 65-93.

213. Gerrard, J.A., C.A. Hutton, and M.A. Perugini, Inhibiting protein-protein interactions as an emerging paradigm for drug discovery. Mini Rev Med Chem, 2007. 7(2): p. 151-7.

214. Fischer, P.M., Protein-Protein Interactions in Drug Discovery. Drug Design Reviews - Online. 2(3): p. 179-207.

215. Chène, P., Drugs Targeting Protein–Protein Interactions. ChemMedChem, 2006. 1(4): p. 400-411.

216. Wells, J.A. and C.L. McClendon, Reaching for high-hanging fruit in drug discovery at protein-protein interfaces. Nature, 2007. 450(7172): p. 1001- 9.

217. Vogler, M., et al., Bcl-2 inhibitors: small molecules with a big impact on cancer therapy. Cell Death Differ, 2009. 16(3): p. 360-7.

138

218. Vicent, M.J., E. Perez-Paya, and M. Orzaez, Discovery of inhibitors of protein-protein interactions from combinatorial libraries. Curr Top Med Chem, 2007. 7(1): p. 83-95.

219. Berg, T., Modulation of protein-protein interactions with small organic molecules. Angew Chem Int Ed Engl, 2003. 42(22): p. 2462-81.

220. Fletcher, S. and A.D. Hamilton, Protein-protein interaction inhibitors: small molecules from screening techniques. Curr Top Med Chem, 2007. 7(10): p. 922-7.

221. Yin, H. and A.D. Hamilton, Strategies for targeting protein-protein interactions with synthetic agents. Angew Chem Int Ed Engl, 2005. 44(27): p. 4130-63.

222. Bernal, F., et al., Reactivation of the p53 tumor suppressor pathway by a stapled p53 peptide. J Am Chem Soc, 2007. 129(9): p. 2456-7.

223. Walensky, L.D., et al., A stapled BID BH3 helix directly binds and activates BAX. Mol Cell, 2006. 24(2): p. 199-210.

224. Schafmeister, C.E., J. Po, and G.L. Verdine, An All-Hydrocarbon Cross- Linking System for Enhancing the Helicity and Metabolic Stability of Peptides. J Am Chem Soc, 2000. 122(24): p. 5891-5892.

225. Henchey, L.K., A.L. Jochim, and P.S. Arora, Contemporary strategies for the stabilization of peptides in the alpha-helical conformation. Curr Opin Chem Biol, 2008. 12(6): p. 692-7.

226. Liu, J., et al., Atomic structure of a short alpha-helix stabilized by a main chain hydrogen-bond surrogate. J Am Chem Soc, 2008. 130(13): p. 4334- 7.

227. Clackson, T. and J.A. Wells, A hot spot of binding energy in a hormone- receptor interface. Science, 1995. 267(5196): p. 383-6.

228. Raj, M., B.N. Bullock, and P.S. Arora, Plucking the high hanging fruit: A systematic approach for targeting protein-protein interactions. Bioorg Med Chem, 2013. 21(14): p. 4051-7.

229. Davis, F.P. and A. Sali, The overlap of small molecule and protein binding sites within families of protein structures. PLoS Comput Biol, 2010. 6(2): p. e1000668.

230. Varmus, H.E., Oncogenes and transcriptional control. Science, 1987. 238(4832): p. 1337-9.

139

231. Blume-Jensen, P. and T. Hunter, Oncogenic kinase signalling. Nature, 2001. 411(6835): p. 355-65.

232. Lowder, M.A., et al., Visualizing protein partnerships in living cells and organisms. Curr Opin Chem Biol, 2011. 15(6): p. 781-8.

233. Zock, J.M., Applications of high content screening in life science research. Combinatorial chemistry & high throughput screening, 2009. 12(9): p. 870- 76.

234. Shekhawat, S.S. and I. Ghosh, Split-protein systems: beyond binary protein-protein interactions. Curr Opin Chem Biol, 2011. 15(6): p. 789-97.

235. Ghosh, I., A.D. Hamilton, and L. Regan, Antiparallel Leucine Zipper- Directed Protein Reassembly: Application to the Green Fluorescent Protein. J Am Chem Soc, 2000. 122(23): p. 5658-5659.

236. Porter, J.R., et al., Profiling small molecule inhibitors against helix-receptor interactions: the Bcl-2 family inhibitor BH3I-1 potently inhibits p53/hDM2. Chemical communications (Cambridge, England), 2010. 46(42): p. 8020- 2.

237. Cassel, J.A., et al., Development of a novel nonradiometric assay for nucleic acid binding to TDP-43 suitable for high-throughput screening using AlphaScreen technology. Journal of biomolecular screening, 2010. 15(9): p. 1099-106.

238. Moon, A.F., et al., A synergistic approach to protein crystallization: Combination of a fixed-arm carrier with surface entropy reduction. Protein Science, 2010. 19(5): p. 901-913.

239. Sibille, N., et al., Comparative NMR study on the impact of point mutations on protein stability of Pseudomonas mendocina lipase. Protein Sci, 2006. 15(8): p. 1915-27.

240. Lipinski, C.A., et al., Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Advanced Drug Delivery Reviews, 1997. 23(1–3): p. 3-25.

241. Kelder, J., et al., Polar molecular surface as a dominating determinant for oral absorption and brain penetration of drugs. Pharm Res, 1999. 16(10): p. 1514-9.

242. Oprea, T.I., Property distribution of drug-related chemical databases. J Comput Aided Mol Des, 2000. 14(3): p. 251-64.

140

243. Ghose, A.K., V.N. Viswanadhan, and J.J. Wendoloski, A knowledge- based approach in designing combinatorial or medicinal chemistry libraries for drug discovery. 1. A qualitative and quantitative characterization of known drug databases. J Comb Chem, 1999. 1(1): p. 55-68.

244. Muegge, I., Selection criteria for drug-like compounds. Med Res Rev, 2003. 23(3): p. 302-21.

245. Tulsi, N.S., A.M. Downey, and C.W. Cairo, A protected l- bromophosphonomethylphenylalanine amino acid derivative (BrPmp) for synthesis of irreversible protein tyrosine phosphatase inhibitors. Bioorg Med Chem, 2010. 18(24): p. 8679-8686.

246. Park, H., S.E. Ryu, and S.J. Kim, Structure-based de novo design of Eya2 phosphatase inhibitors. J Mol Graph Model, 2012. 38: p. 382-8.

141